diff --git a/.gitignore b/.gitignore index ea1d42787cd63e066d5ebd2486921002f13b7035..6989ea9f725aa2bd5445927a9a3e275f403f55e1 100644 --- a/.gitignore +++ b/.gitignore @@ -62,3 +62,6 @@ tmp/ temp/ .specstory/ smollm3_robust/ mlruns/ +smollm3_robust/ +mlruns/ +.specstory/ diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/meta.yaml b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/meta.yaml deleted file mode 100644 index 856b1cefb142e9ff8d32f6818eb6cf6812537246..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/0d212b72b30d42f784c5fba529d33c38/artifacts -end_time: 1753092408955 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: 0d212b72b30d42f784c5fba529d33c38 -run_name: ./smollm_tool_adapter -run_uuid: 0d212b72b30d42f784c5fba529d33c38 -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753092389985 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/epoch b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/epoch deleted file mode 100644 index 7b87aead9be3739148ff60ed8a2ad7aa49ebd4ba..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/epoch +++ /dev/null @@ -1,7 +0,0 @@ -1753092397035 0.5 1 -1753092399791 1.0 2 -1753092401876 1.5 3 -1753092403857 2.0 4 -1753092405888 2.5 5 -1753092408205 3.0 6 -1753092408953 3.0 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/grad_norm b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/grad_norm deleted file mode 100644 index e099238253b2ed6dd52c23eb856adf192c48deb5..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/grad_norm +++ /dev/null @@ -1,6 +0,0 @@ -1753092397035 1.475852131843567 1 -1753092399791 1.4370522499084473 2 -1753092401876 1.3117226362228394 3 -1753092403857 1.602066993713379 4 -1753092405888 1.452284812927246 5 -1753092408205 1.3940032720565796 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/learning_rate b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/learning_rate deleted file mode 100644 index 15d8363d76717b493a7642233add80826c896441..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/learning_rate +++ /dev/null @@ -1,6 +0,0 @@ -1753092397035 0.0 1 -1753092399791 5e-06 2 -1753092401876 1e-05 3 -1753092403857 1.5e-05 4 -1753092405888 2e-05 5 -1753092408205 2.5e-05 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/loss b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/loss deleted file mode 100644 index 0ac4cfdcf98f2984a2c422e320fd4fbdfafebd06..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/loss +++ /dev/null @@ -1,6 +0,0 @@ -1753092397035 2.3957 1 -1753092399791 2.41 2 -1753092401876 2.2712 3 -1753092403857 2.5251 4 -1753092405888 2.4042 5 -1753092408205 2.288 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/total_flos b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/total_flos deleted file mode 100644 index dccdfd1a06e4c236d55e69587eef25aecdcf0b60..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/total_flos +++ /dev/null @@ -1 +0,0 @@ -1753092408953 43237794852864.0 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_loss b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_loss deleted file mode 100644 index 088c08dd9fba51141324b8a9a638d746133a8b3a..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_loss +++ /dev/null @@ -1 +0,0 @@ -1753092408953 2.3823566834131875 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_runtime b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_runtime deleted file mode 100644 index aea40ef92198860609494ce0fa4b57141012b08d..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_runtime +++ /dev/null @@ -1 +0,0 @@ -1753092408953 19.2905 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_samples_per_second b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_samples_per_second deleted file mode 100644 index 10108ab667836379fbbf5fa25446a6002783ae91..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_samples_per_second +++ /dev/null @@ -1 +0,0 @@ -1753092408953 1.244 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_steps_per_second b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_steps_per_second deleted file mode 100644 index 34a0ffadc03593132b4fc031a06489bc972acaa8..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/metrics/train_steps_per_second +++ /dev/null @@ -1 +0,0 @@ -1753092408953 0.311 6 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/_name_or_path b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/_name_or_path deleted file mode 100644 index c96b622fd969ae1f54c75f69eaa9ca73b7d94a55..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM2-1.7B-Instruct \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/accelerator_config b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adafactor b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_beta1 b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_beta2 b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_epsilon b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/add_cross_attention b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/architectures b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/architectures deleted file mode 100644 index fe06827e8a90199228b9e7009f79062405f3d52f..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['LlamaForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/attention_bias b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/attention_dropout b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/auto_find_batch_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/average_tokens_across_devices b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bad_words_ids b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/batch_eval_metrics b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/begin_suppress_tokens b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bf16 b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bf16_full_eval b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bos_token_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bos_token_id deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/chunk_size_feed_forward b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/cross_attention_hidden_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/data_seed b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_drop_last b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_num_workers b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_persistent_workers b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_pin_memory b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_prefetch_factor b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_backend b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_broadcast_buffers b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_bucket_cap_mb b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_find_unused_parameters b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_timeout b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/debug b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/decoder_start_token_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/deepspeed b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/disable_tqdm b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/diversity_penalty b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_eval b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_predict b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_sample b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_train b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/early_stopping b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/encoder_no_repeat_ngram_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eos_token_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eos_token_id deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_accumulation_steps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_delay b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_do_concat_batches b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_on_start b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_steps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_strategy b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_use_gather_object b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/exponential_decay_length_penalty b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/finetuning_task b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/forced_bos_token_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/forced_eos_token_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16 b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_backend b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_full_eval b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_opt_level b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_config b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_min_num_params b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/full_determinism b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_accumulation_steps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_accumulation_steps deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_checkpointing b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_checkpointing deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_checkpointing_kwargs b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/greater_is_better b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/group_by_length b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/half_precision_backend b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/head_dim b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/head_dim deleted file mode 100644 index 4b6f9c39e5c757bf387d465c53026b336dd8b96c..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/head_dim +++ /dev/null @@ -1 +0,0 @@ -64 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hidden_act b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hidden_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_always_push b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_model_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_private_repo b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_revision b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_strategy b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_token b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/id2label b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ignore_data_skip b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_for_metrics b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_inputs_for_metrics b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_num_input_tokens_seen b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_tokens_per_second b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/initializer_range b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/intermediate_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/intermediate_size deleted file mode 100644 index e0c3f84a6747696c58b1a32f81129c66498e094a..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -8192 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/is_decoder b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/is_encoder_decoder b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/jit_mode_eval b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label2id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label_names b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label_smoothing_factor b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/learning_rate b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/learning_rate deleted file mode 100644 index 335f2feb826f13cdd89c6b5234de9a8c96396bb7..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -5e-05 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/length_column_name b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/length_penalty b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/liger_kernel_config b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/load_best_model_at_end b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/local_rank b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_level b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_level_replica b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_on_each_node b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_dir b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_dir deleted file mode 100644 index 6bc8522feb83b127cc7e1e76a426e8d1a973d69c..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./logs \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_first_step b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_nan_inf_filter b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_steps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_steps deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_strategy b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/lr_scheduler_kwargs b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/lr_scheduler_type b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/lr_scheduler_type deleted file mode 100644 index 9c01abbc1e45086237d1a59d6ec6bdea2400f1a7..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -linear \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_grad_norm b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_length b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_position_embeddings b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_position_embeddings deleted file mode 100644 index e0c3f84a6747696c58b1a32f81129c66498e094a..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -8192 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_steps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/metric_for_best_model b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/min_length b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/mlp_bias b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/model_type b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/model_type deleted file mode 100644 index 056bf100b8c479097594926c1ab454fe1ee0f39c..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/model_type +++ /dev/null @@ -1 +0,0 @@ -llama \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/mp_parameters b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/neftune_noise_alpha b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/no_cuda b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/no_repeat_ngram_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_attention_heads b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_attention_heads deleted file mode 100644 index 1758dddccea2b3b02d21228a0d06a45a35c0d861..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -32 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_beam_groups b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_beams b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_hidden_layers b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_hidden_layers deleted file mode 100644 index cabf43b5ddf813cbe89697372a21373f14921884..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -24 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_key_value_heads b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_key_value_heads deleted file mode 100644 index 1758dddccea2b3b02d21228a0d06a45a35c0d861..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -32 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_return_sequences b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_train_epochs b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_train_epochs deleted file mode 100644 index e440e5c842586965a7fb77deda2eca68612b1f53..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -3 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim_args b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim_target_modules b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_attentions b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_dir b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_dir deleted file mode 100644 index 742802ce651cc0a4611e995cf12c1f3626cbe7e4..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm_tool_adapter \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_hidden_states b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_scores b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/overwrite_output_dir b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pad_token_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pad_token_id deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/past_index b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_device_eval_batch_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_device_train_batch_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_device_train_batch_size deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_gpu_eval_batch_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_gpu_train_batch_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/prediction_loss_only b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/prefix b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pretraining_tp b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/problem_type b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pruned_heads b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_model_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_organization b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_token b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ray_scope b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/remove_invalid_values b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/remove_unused_columns b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/remove_unused_columns deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/repetition_penalty b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/report_to b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/restore_callback_states_from_checkpoint b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/resume_from_checkpoint b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/return_dict b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/return_dict_in_generate b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rms_norm_eps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rms_norm_eps deleted file mode 100644 index 5868ff147459cee04c24f2de58e75969024870b8..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-05 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rope_scaling b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rope_theta b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rope_theta deleted file mode 100644 index b21fe1ccb7f2c4f71b891115b7fdbce0fa1e9382..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -130000 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/run_name b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/run_name deleted file mode 100644 index 742802ce651cc0a4611e995cf12c1f3626cbe7e4..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm_tool_adapter \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_on_each_node b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_only_model b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_safetensors b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_steps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_steps deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_strategy b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_total_limit b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_total_limit deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/seed b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/sep_token_id b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/skip_memory_metrics b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/suppress_tokens b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/task_specific_params b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/temperature b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tf32 b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tf_legacy_loss b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tie_encoder_decoder b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tie_word_embeddings b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tokenizer_class b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/top_k b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/top_p b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile_backend b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile_mode b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_dtype b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_empty_cache_steps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torchdynamo b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torchscript b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tpu_metrics_debug b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tpu_num_cores b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/transformers.js_config b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/transformers.js_config deleted file mode 100644 index 80d856d83717f7c7e67534fc3b6999564fa14ed9..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/transformers.js_config +++ /dev/null @@ -1 +0,0 @@ -{'dtype': 'q4', 'kv_cache_dtype': {'q4f16': 'float16', 'fp16': 'float16'}, 'use_external_data_format': {'model.onnx': True, 'model_fp16.onnx': True}} \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/transformers_version b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/typical_p b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_bfloat16 b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_cache b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_cache deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_cpu b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_ipex b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_legacy_prediction_loop b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_liger_kernel b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_mps_device b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/vocab_size b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/vocab_size deleted file mode 100644 index 02ce579f3e592b1449b2491efd0c978e6478c61b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -49152 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/warmup_ratio b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/warmup_steps b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/warmup_steps deleted file mode 100644 index 9a037142aa3c1b4c490e1a38251620f113465330..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -10 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/weight_decay b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/weight_decay deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.runName b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.runName deleted file mode 100644 index 742802ce651cc0a4611e995cf12c1f3626cbe7e4..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm_tool_adapter \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.source.name b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.source.name deleted file mode 100644 index 2f76a1ec5d70457f4d7b446a127c75d53b488fee..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_simple.py \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.source.type b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.user b/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/0d212b72b30d42f784c5fba529d33c38/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/meta.yaml b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/meta.yaml deleted file mode 100644 index 1b57fc2767e49069c7f968a50bd9d1866999d475..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/44b7fee647244a65b313e31b2391d0dd/artifacts -end_time: 1753099050050 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: 44b7fee647244a65b313e31b2391d0dd -run_name: ./smollm3_intensive -run_uuid: 44b7fee647244a65b313e31b2391d0dd -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753099048038 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/_name_or_path b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/_name_or_path deleted file mode 100644 index 3a8ba38a16a767019a426ed9ae2b3cd2746d622d..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM3-3B \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/accelerator_config b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adafactor b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_beta1 b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_beta2 b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_epsilon b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/add_cross_attention b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/architectures b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/architectures deleted file mode 100644 index 55febbc88b5c08e00d44e95e7ed72640c6d034b6..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['SmolLM3ForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/attention_bias b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/attention_dropout b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/auto_find_batch_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/average_tokens_across_devices b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bad_words_ids b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/batch_eval_metrics b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/begin_suppress_tokens b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bf16 b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bf16_full_eval b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bos_token_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bos_token_id deleted file mode 100644 index 5499007cbac38bc897e3c2766b82a647ac28c735..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -128000 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/chunk_size_feed_forward b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/cross_attention_hidden_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/data_seed b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_drop_last b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_num_workers b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_persistent_workers b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_pin_memory b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_prefetch_factor b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_backend b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_broadcast_buffers b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_bucket_cap_mb b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_find_unused_parameters b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_timeout b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/debug b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/decoder_start_token_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/deepspeed b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/disable_tqdm b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/diversity_penalty b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_eval b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_predict b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_sample b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_train b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/early_stopping b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/encoder_no_repeat_ngram_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eos_token_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eos_token_id deleted file mode 100644 index d78b6326c2a8961e1672862523846efebe7ffdd6..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -128012 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_accumulation_steps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_delay b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_do_concat_batches b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_on_start b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_steps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_strategy b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_use_gather_object b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/exponential_decay_length_penalty b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/finetuning_task b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/forced_bos_token_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/forced_eos_token_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16 b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_backend b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_full_eval b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_opt_level b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_config b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_min_num_params b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/full_determinism b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_accumulation_steps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_accumulation_steps deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_checkpointing b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_checkpointing deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_checkpointing_kwargs b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/greater_is_better b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/group_by_length b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/half_precision_backend b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hidden_act b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hidden_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_always_push b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_model_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_private_repo b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_revision b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_strategy b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_token b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/id2label b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ignore_data_skip b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_for_metrics b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_inputs_for_metrics b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_num_input_tokens_seen b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_tokens_per_second b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/initializer_range b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/intermediate_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/intermediate_size deleted file mode 100644 index 9b7babb1160a44a2cf9ae0dbfd43efd0fd70838b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -11008 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/is_decoder b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/is_encoder_decoder b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/jit_mode_eval b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label2id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label_names b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label_smoothing_factor b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/layer_types b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/layer_types deleted file mode 100644 index bc44845b7022ebf6e7ac1a8dac64d5ba0daa15b1..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/layer_types +++ /dev/null @@ -1 +0,0 @@ -['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'] \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/learning_rate b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/learning_rate deleted file mode 100644 index 9b726543243d3b3128c0d6d2835b1af1ea5db211..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -3e-05 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/length_column_name b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/length_penalty b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/liger_kernel_config b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/load_best_model_at_end b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/local_rank b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_level b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_level_replica b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_on_each_node b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_dir b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_dir deleted file mode 100644 index 9e6efd316034ed66cbdd06db7326925d4b811992..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_intensive/runs/Jul21_12-57-26_Jasons-MacBook-Pro.local \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_first_step b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_nan_inf_filter b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_steps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_steps deleted file mode 100644 index 9a037142aa3c1b4c490e1a38251620f113465330..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -10 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_strategy b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/lr_scheduler_kwargs b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/lr_scheduler_type b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/lr_scheduler_type deleted file mode 100644 index 84aa3999b5b7cae7f78b1f77e04d182643005a92..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -cosine \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_grad_norm b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_length b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_position_embeddings b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_position_embeddings deleted file mode 100644 index 83c406c4fd41a6ec8a25290b17cd3894ca57ac4b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -65536 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_steps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_window_layers b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_window_layers deleted file mode 100644 index 368f89ceef179cc546403ac0d5ef1d0e4b340447..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/max_window_layers +++ /dev/null @@ -1 +0,0 @@ -28 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/metric_for_best_model b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/min_length b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/mlp_bias b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/model_type b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/model_type deleted file mode 100644 index 34bda861a9301e0dab58899ec4828a0ee9f464b1..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/model_type +++ /dev/null @@ -1 +0,0 @@ -smollm3 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/mp_parameters b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/neftune_noise_alpha b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_cuda b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_repeat_ngram_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_rope_layer_interval b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_rope_layer_interval deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_rope_layer_interval +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_rope_layers b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_rope_layers deleted file mode 100644 index 2cdfac1b0d51ce92d6d6469bd544bc76a3521cec..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/no_rope_layers +++ /dev/null @@ -1 +0,0 @@ -[1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0] \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_attention_heads b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_attention_heads deleted file mode 100644 index 19c7bdba7b1e9bfe80365a50420a6d538ca503c3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -16 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_beam_groups b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_beams b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_hidden_layers b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_hidden_layers deleted file mode 100644 index dce6588ca1420a69eea5699f2ec5d666daf16a6a..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -36 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_key_value_heads b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_key_value_heads deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_return_sequences b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_train_epochs b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_train_epochs deleted file mode 100644 index 3cacc0b93c9c9c03a72da624ca28a09ba5c1336f..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -12 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim_args b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim_target_modules b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_attentions b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_dir b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_dir deleted file mode 100644 index 22fce0d759d1c276e98c2a86334d63a4ac2c0a19..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_intensive \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_hidden_states b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_scores b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/overwrite_output_dir b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pad_token_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pad_token_id deleted file mode 100644 index 8d9491efc0fb90c90645db18bae256c4f9daff70..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -128004 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/past_index b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_device_eval_batch_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_device_train_batch_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_device_train_batch_size deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_gpu_eval_batch_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_gpu_train_batch_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/prediction_loss_only b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/prefix b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pretraining_tp b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/problem_type b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pruned_heads b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_model_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_organization b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_token b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ray_scope b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/remove_invalid_values b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/remove_unused_columns b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/remove_unused_columns deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/repetition_penalty b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/report_to b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/restore_callback_states_from_checkpoint b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/resume_from_checkpoint b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/return_dict b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/return_dict_in_generate b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rms_norm_eps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rms_norm_eps deleted file mode 100644 index d6bc26a15d7729bbaf8f0e522bea2d0a8781886d..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-06 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rope_scaling b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rope_theta b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rope_theta deleted file mode 100644 index 445ba6793f521c521c5da54f26f04a56ecc83ee2..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -5000000.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/run_name b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/run_name deleted file mode 100644 index 22fce0d759d1c276e98c2a86334d63a4ac2c0a19..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm3_intensive \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_on_each_node b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_only_model b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_safetensors b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_steps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_steps deleted file mode 100644 index 105d7d9ad3afc7bb78a0dec4d829880831605dfb..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -100 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_strategy b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_total_limit b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_total_limit deleted file mode 100644 index e440e5c842586965a7fb77deda2eca68612b1f53..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -3 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/seed b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/sep_token_id b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/skip_memory_metrics b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/sliding_window b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/sliding_window deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/sliding_window +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/suppress_tokens b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/task_specific_params b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/temperature b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tf32 b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tf_legacy_loss b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tie_encoder_decoder b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tie_word_embeddings b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tokenizer_class b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/top_k b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/top_p b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile_backend b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile_mode b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_dtype b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_empty_cache_steps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torchdynamo b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torchscript b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tpu_metrics_debug b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tpu_num_cores b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/transformers_version b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/typical_p b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_bfloat16 b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_cache b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_cache deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_cpu b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_ipex b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_legacy_prediction_loop b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_liger_kernel b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_mps_device b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_sliding_window b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_sliding_window deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/use_sliding_window +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/vocab_size b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/vocab_size deleted file mode 100644 index 34c44b19378193a3b4fa853df4426fc426c15535..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -128256 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/warmup_ratio b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/warmup_steps b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/warmup_steps deleted file mode 100644 index 105d7d9ad3afc7bb78a0dec4d829880831605dfb..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -100 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/weight_decay b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/weight_decay deleted file mode 100644 index d1c6331b3109accd73f01907062e6c174e28200a..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.01 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.runName b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.runName deleted file mode 100644 index 22fce0d759d1c276e98c2a86334d63a4ac2c0a19..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm3_intensive \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.git.commit b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.git.commit deleted file mode 100644 index 64393076c1626bfaa09d72c60ec8c9a9769145c8..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.git.commit +++ /dev/null @@ -1 +0,0 @@ -e2610e8c39108b089782f99d0af9c3a8939adc61 \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.name b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.name deleted file mode 100644 index 0587e3141c06fc9376abcc9e4bf3263c4fdc7cfc..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_intensive.py \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.type b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.user b/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/44b7fee647244a65b313e31b2391d0dd/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/meta.yaml b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/meta.yaml deleted file mode 100644 index 45bf68c9ee6d90d94170caa270ac6fe59f75cf79..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/5f3eb1695be74072935c7183724e6a7f/artifacts -end_time: 1753096735528 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: 5f3eb1695be74072935c7183724e6a7f -run_name: ./smollm3_robust -run_uuid: 5f3eb1695be74072935c7183724e6a7f -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753096456316 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/epoch b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/epoch deleted file mode 100644 index e612379477918eb65de4028b343e11b6eb7d2ac2..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/epoch +++ /dev/null @@ -1,22 +0,0 @@ -1753096473389 0.14285714285714285 2 -1753096489407 0.2857142857142857 4 -1753096505014 0.42857142857142855 6 -1753096520525 0.5714285714285714 8 -1753096532693 0.7142857142857143 10 -1753096544914 0.8571428571428571 12 -1753096557067 1.0 14 -1753096569426 1.1428571428571428 16 -1753096581852 1.2857142857142856 18 -1753096594098 1.4285714285714286 20 -1753096608461 1.5714285714285714 22 -1753096620872 1.7142857142857144 24 -1753096633538 1.8571428571428572 26 -1753096646183 2.0 28 -1753096658815 2.142857142857143 30 -1753096671336 2.2857142857142856 32 -1753096683818 2.4285714285714284 34 -1753096696249 2.571428571428571 36 -1753096708741 2.7142857142857144 38 -1753096721260 2.857142857142857 40 -1753096734548 3.0 42 -1753096735527 3.0 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/grad_norm b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/grad_norm deleted file mode 100644 index 91a2feba0bd07218cc0bdd2e921f9f58e9192d2f..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/grad_norm +++ /dev/null @@ -1,21 +0,0 @@ -1753096473389 3.7707648277282715 2 -1753096489407 3.1055235862731934 4 -1753096505014 2.5664002895355225 6 -1753096520525 2.0128557682037354 8 -1753096532693 1.8992419242858887 10 -1753096544914 1.767743706703186 12 -1753096557067 1.8402271270751953 14 -1753096569426 1.5837992429733276 16 -1753096581852 1.2706984281539917 18 -1753096594098 1.0353035926818848 20 -1753096608461 0.6108320355415344 22 -1753096620872 0.5223175287246704 24 -1753096633538 0.42949551343917847 26 -1753096646183 0.44767066836357117 28 -1753096658815 0.4750105142593384 30 -1753096671336 0.3679599165916443 32 -1753096683818 0.43462860584259033 34 -1753096696249 0.4953954219818115 36 -1753096708741 0.4565838873386383 38 -1753096721260 0.46194586157798767 40 -1753096734548 0.3986656963825226 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/learning_rate b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/learning_rate deleted file mode 100644 index 86d580dd0374668f911f9ad7ca5d02f23b238a68..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/learning_rate +++ /dev/null @@ -1,21 +0,0 @@ -1753096473389 5e-06 2 -1753096489407 1.5e-05 4 -1753096505014 2.5e-05 6 -1753096520525 3.5e-05 8 -1753096532693 4.5e-05 10 -1753096544914 4.8437500000000005e-05 12 -1753096557067 4.5312500000000004e-05 14 -1753096569426 4.21875e-05 16 -1753096581852 3.90625e-05 18 -1753096594098 3.59375e-05 20 -1753096608461 3.2812500000000005e-05 22 -1753096620872 2.96875e-05 24 -1753096633538 2.6562500000000002e-05 26 -1753096646183 2.34375e-05 28 -1753096658815 2.0312500000000002e-05 30 -1753096671336 1.71875e-05 32 -1753096683818 1.4062500000000001e-05 34 -1753096696249 1.09375e-05 36 -1753096708741 7.8125e-06 38 -1753096721260 4.6875000000000004e-06 40 -1753096734548 1.5625e-06 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/loss b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/loss deleted file mode 100644 index d4852c53108a975ebe7329c3cad634110bd21d17..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/loss +++ /dev/null @@ -1,21 +0,0 @@ -1753096473389 1.6966 2 -1753096489407 1.5885 4 -1753096505014 1.495 6 -1753096520525 1.3152 8 -1753096532693 1.173 10 -1753096544914 0.9337 12 -1753096557067 0.8148 14 -1753096569426 0.6651 16 -1753096581852 0.4811 18 -1753096594098 0.3789 20 -1753096608461 0.3672 22 -1753096620872 0.3119 24 -1753096633538 0.3104 26 -1753096646183 0.2639 28 -1753096658815 0.2866 30 -1753096671336 0.2706 32 -1753096683818 0.2492 34 -1753096696249 0.262 36 -1753096708741 0.2056 38 -1753096721260 0.2481 40 -1753096734548 0.232 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/total_flos b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/total_flos deleted file mode 100644 index 33a61be98d2875abdb6390d1c56d99a80e912277..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/total_flos +++ /dev/null @@ -1 +0,0 @@ -1753096735527 1150194237714432.0 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_loss b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_loss deleted file mode 100644 index f0b5d707c5dc78dffe9b7b350d6d9faf533ceedf..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_loss +++ /dev/null @@ -1 +0,0 @@ -1753096735527 0.6452025870482127 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_runtime b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_runtime deleted file mode 100644 index 4c1b5849ba7cca8cb70aeff92f576d564b65c898..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_runtime +++ /dev/null @@ -1 +0,0 @@ -1753096735527 279.6181 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_samples_per_second b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_samples_per_second deleted file mode 100644 index c90b2aa4fe9f884be90d178fe7238ffccd242f17..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_samples_per_second +++ /dev/null @@ -1 +0,0 @@ -1753096735527 1.202 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_steps_per_second b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_steps_per_second deleted file mode 100644 index 37fd6da89a1be10e39e6afd9f1e8c7ce9544fd86..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/metrics/train_steps_per_second +++ /dev/null @@ -1 +0,0 @@ -1753096735527 0.15 42 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/_name_or_path b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/_name_or_path deleted file mode 100644 index 3a8ba38a16a767019a426ed9ae2b3cd2746d622d..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM3-3B \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/accelerator_config b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adafactor b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_beta1 b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_beta2 b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_epsilon b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/add_cross_attention b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/architectures b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/architectures deleted file mode 100644 index 55febbc88b5c08e00d44e95e7ed72640c6d034b6..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['SmolLM3ForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/attention_bias b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/attention_dropout b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/auto_find_batch_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/average_tokens_across_devices b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bad_words_ids b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/batch_eval_metrics b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/begin_suppress_tokens b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bf16 b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bf16_full_eval b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bos_token_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bos_token_id deleted file mode 100644 index 5499007cbac38bc897e3c2766b82a647ac28c735..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -128000 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/chunk_size_feed_forward b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/cross_attention_hidden_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/data_seed b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_drop_last b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_num_workers b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_persistent_workers b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_pin_memory b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_prefetch_factor b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_backend b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_broadcast_buffers b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_bucket_cap_mb b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_find_unused_parameters b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_timeout b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/debug b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/decoder_start_token_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/deepspeed b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/disable_tqdm b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/diversity_penalty b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_eval b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_predict b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_sample b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_train b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/early_stopping b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/encoder_no_repeat_ngram_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eos_token_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eos_token_id deleted file mode 100644 index d78b6326c2a8961e1672862523846efebe7ffdd6..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -128012 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_accumulation_steps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_delay b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_do_concat_batches b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_on_start b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_steps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_strategy b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_use_gather_object b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/exponential_decay_length_penalty b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/finetuning_task b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/forced_bos_token_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/forced_eos_token_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16 b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_backend b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_full_eval b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_opt_level b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_config b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_min_num_params b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/full_determinism b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_accumulation_steps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_accumulation_steps deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_checkpointing b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_checkpointing deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_checkpointing_kwargs b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/greater_is_better b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/group_by_length b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/half_precision_backend b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hidden_act b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hidden_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_always_push b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_model_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_private_repo b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_revision b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_strategy b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_token b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/id2label b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ignore_data_skip b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_for_metrics b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_inputs_for_metrics b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_num_input_tokens_seen b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_tokens_per_second b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/initializer_range b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/intermediate_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/intermediate_size deleted file mode 100644 index 9b7babb1160a44a2cf9ae0dbfd43efd0fd70838b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -11008 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/is_decoder b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/is_encoder_decoder b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/jit_mode_eval b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label2id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label_names b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label_smoothing_factor b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/layer_types b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/layer_types deleted file mode 100644 index bc44845b7022ebf6e7ac1a8dac64d5ba0daa15b1..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/layer_types +++ /dev/null @@ -1 +0,0 @@ -['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'] \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/learning_rate b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/learning_rate deleted file mode 100644 index 335f2feb826f13cdd89c6b5234de9a8c96396bb7..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -5e-05 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/length_column_name b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/length_penalty b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/liger_kernel_config b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/load_best_model_at_end b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/local_rank b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_level b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_level_replica b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_on_each_node b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_dir b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_dir deleted file mode 100644 index 159cb53a946c26a54f884bdff3fca2533d1de066..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust/runs/Jul21_12-14-15_Jasons-MacBook-Pro.local \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_first_step b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_nan_inf_filter b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_steps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_steps deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_strategy b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/lr_scheduler_kwargs b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/lr_scheduler_type b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/lr_scheduler_type deleted file mode 100644 index 9c01abbc1e45086237d1a59d6ec6bdea2400f1a7..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -linear \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_grad_norm b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_length b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_position_embeddings b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_position_embeddings deleted file mode 100644 index 83c406c4fd41a6ec8a25290b17cd3894ca57ac4b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -65536 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_steps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_window_layers b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_window_layers deleted file mode 100644 index 368f89ceef179cc546403ac0d5ef1d0e4b340447..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/max_window_layers +++ /dev/null @@ -1 +0,0 @@ -28 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/metric_for_best_model b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/min_length b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/mlp_bias b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/model_type b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/model_type deleted file mode 100644 index 34bda861a9301e0dab58899ec4828a0ee9f464b1..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/model_type +++ /dev/null @@ -1 +0,0 @@ -smollm3 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/mp_parameters b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/neftune_noise_alpha b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_cuda b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_repeat_ngram_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_rope_layer_interval b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_rope_layer_interval deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_rope_layer_interval +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_rope_layers b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_rope_layers deleted file mode 100644 index 2cdfac1b0d51ce92d6d6469bd544bc76a3521cec..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/no_rope_layers +++ /dev/null @@ -1 +0,0 @@ -[1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0] \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_attention_heads b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_attention_heads deleted file mode 100644 index 19c7bdba7b1e9bfe80365a50420a6d538ca503c3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -16 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_beam_groups b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_beams b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_hidden_layers b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_hidden_layers deleted file mode 100644 index dce6588ca1420a69eea5699f2ec5d666daf16a6a..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -36 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_key_value_heads b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_key_value_heads deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_return_sequences b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_train_epochs b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_train_epochs deleted file mode 100644 index e440e5c842586965a7fb77deda2eca68612b1f53..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -3 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim_args b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim_target_modules b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_attentions b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_dir b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_dir deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_hidden_states b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_scores b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/overwrite_output_dir b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pad_token_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pad_token_id deleted file mode 100644 index 8d9491efc0fb90c90645db18bae256c4f9daff70..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -128004 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/past_index b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_device_eval_batch_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_device_train_batch_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_device_train_batch_size deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_gpu_eval_batch_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_gpu_train_batch_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/prediction_loss_only b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/prefix b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pretraining_tp b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/problem_type b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pruned_heads b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_model_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_organization b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_token b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ray_scope b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/remove_invalid_values b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/remove_unused_columns b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/remove_unused_columns deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/repetition_penalty b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/report_to b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/restore_callback_states_from_checkpoint b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/resume_from_checkpoint b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/return_dict b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/return_dict_in_generate b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rms_norm_eps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rms_norm_eps deleted file mode 100644 index d6bc26a15d7729bbaf8f0e522bea2d0a8781886d..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-06 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rope_scaling b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rope_theta b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rope_theta deleted file mode 100644 index 445ba6793f521c521c5da54f26f04a56ecc83ee2..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -5000000.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/run_name b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/run_name deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_on_each_node b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_only_model b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_safetensors b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_steps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_steps deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_strategy b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_total_limit b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_total_limit deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/seed b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/sep_token_id b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/skip_memory_metrics b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/sliding_window b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/sliding_window deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/sliding_window +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/suppress_tokens b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/task_specific_params b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/temperature b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tf32 b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tf_legacy_loss b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tie_encoder_decoder b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tie_word_embeddings b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tokenizer_class b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/top_k b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/top_p b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile_backend b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile_mode b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_dtype b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_empty_cache_steps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torchdynamo b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torchscript b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tpu_metrics_debug b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tpu_num_cores b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/transformers_version b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/typical_p b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_bfloat16 b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_cache b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_cache deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_cpu b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_ipex b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_legacy_prediction_loop b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_liger_kernel b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_mps_device b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_sliding_window b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_sliding_window deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/use_sliding_window +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/vocab_size b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/vocab_size deleted file mode 100644 index 34c44b19378193a3b4fa853df4426fc426c15535..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -128256 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/warmup_ratio b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/warmup_steps b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/warmup_steps deleted file mode 100644 index 9a037142aa3c1b4c490e1a38251620f113465330..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -10 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/weight_decay b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/weight_decay deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.runName b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.runName deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.git.commit b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.git.commit deleted file mode 100644 index 64393076c1626bfaa09d72c60ec8c9a9769145c8..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.git.commit +++ /dev/null @@ -1 +0,0 @@ -e2610e8c39108b089782f99d0af9c3a8939adc61 \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.name b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.name deleted file mode 100644 index 124d840089086dc6280e8eb4e84f59ad6b0b015f..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_simple_robust.py \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.type b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.user b/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/5f3eb1695be74072935c7183724e6a7f/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/meta.yaml b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/meta.yaml deleted file mode 100644 index 2553f9434e37ce625924c32a7b938eb5a7f147ad..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/67a82e40de7245e68717d18c2cb8a858/artifacts -end_time: 1753092896402 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: 67a82e40de7245e68717d18c2cb8a858 -run_name: ./smollm3_tool_adapter -run_uuid: 67a82e40de7245e68717d18c2cb8a858 -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753092895341 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/_name_or_path b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/_name_or_path deleted file mode 100644 index 3a8ba38a16a767019a426ed9ae2b3cd2746d622d..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM3-3B \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/accelerator_config b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adafactor b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_beta1 b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_beta2 b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_epsilon b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/add_cross_attention b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/architectures b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/architectures deleted file mode 100644 index 55febbc88b5c08e00d44e95e7ed72640c6d034b6..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['SmolLM3ForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/attention_bias b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/attention_dropout b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/auto_find_batch_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/average_tokens_across_devices b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bad_words_ids b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/batch_eval_metrics b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/begin_suppress_tokens b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bf16 b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bf16_full_eval b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bos_token_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bos_token_id deleted file mode 100644 index 5499007cbac38bc897e3c2766b82a647ac28c735..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -128000 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/chunk_size_feed_forward b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/cross_attention_hidden_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/data_seed b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_drop_last b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_num_workers b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_persistent_workers b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_pin_memory b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_prefetch_factor b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_backend b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_broadcast_buffers b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_bucket_cap_mb b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_find_unused_parameters b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_timeout b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/debug b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/decoder_start_token_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/deepspeed b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/disable_tqdm b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/diversity_penalty b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_eval b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_predict b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_sample b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_train b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/early_stopping b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/encoder_no_repeat_ngram_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eos_token_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eos_token_id deleted file mode 100644 index d78b6326c2a8961e1672862523846efebe7ffdd6..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -128012 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_accumulation_steps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_delay b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_do_concat_batches b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_on_start b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_steps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_strategy b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_use_gather_object b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/exponential_decay_length_penalty b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/finetuning_task b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/forced_bos_token_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/forced_eos_token_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16 b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_backend b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_full_eval b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_opt_level b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_config b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_min_num_params b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/full_determinism b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_accumulation_steps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_accumulation_steps deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_checkpointing b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_checkpointing deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_checkpointing_kwargs b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/greater_is_better b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/group_by_length b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/half_precision_backend b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hidden_act b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hidden_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_always_push b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_model_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_private_repo b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_revision b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_strategy b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_token b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/id2label b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ignore_data_skip b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_for_metrics b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_inputs_for_metrics b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_num_input_tokens_seen b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_tokens_per_second b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/initializer_range b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/intermediate_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/intermediate_size deleted file mode 100644 index 9b7babb1160a44a2cf9ae0dbfd43efd0fd70838b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -11008 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/is_decoder b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/is_encoder_decoder b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/jit_mode_eval b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label2id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label_names b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label_smoothing_factor b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/layer_types b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/layer_types deleted file mode 100644 index bc44845b7022ebf6e7ac1a8dac64d5ba0daa15b1..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/layer_types +++ /dev/null @@ -1 +0,0 @@ -['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'] \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/learning_rate b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/learning_rate deleted file mode 100644 index 83f55466e252f8ed340b201b70d8470f0cd54c4b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -0.0003 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/length_column_name b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/length_penalty b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/liger_kernel_config b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/load_best_model_at_end b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/local_rank b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_level b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_level_replica b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_on_each_node b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_dir b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_dir deleted file mode 100644 index 6bc8522feb83b127cc7e1e76a426e8d1a973d69c..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./logs \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_first_step b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_nan_inf_filter b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_steps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_steps deleted file mode 100644 index 7813681f5b41c028345ca62a2be376bae70b7f61..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -5 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_strategy b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/lr_scheduler_kwargs b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/lr_scheduler_type b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/lr_scheduler_type deleted file mode 100644 index 84aa3999b5b7cae7f78b1f77e04d182643005a92..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -cosine \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_grad_norm b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_length b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_position_embeddings b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_position_embeddings deleted file mode 100644 index 83c406c4fd41a6ec8a25290b17cd3894ca57ac4b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -65536 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_steps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_window_layers b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_window_layers deleted file mode 100644 index 368f89ceef179cc546403ac0d5ef1d0e4b340447..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/max_window_layers +++ /dev/null @@ -1 +0,0 @@ -28 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/metric_for_best_model b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/min_length b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/mlp_bias b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/model_type b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/model_type deleted file mode 100644 index 34bda861a9301e0dab58899ec4828a0ee9f464b1..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/model_type +++ /dev/null @@ -1 +0,0 @@ -smollm3 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/mp_parameters b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/neftune_noise_alpha b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_cuda b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_repeat_ngram_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_rope_layer_interval b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_rope_layer_interval deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_rope_layer_interval +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_rope_layers b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_rope_layers deleted file mode 100644 index 2cdfac1b0d51ce92d6d6469bd544bc76a3521cec..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/no_rope_layers +++ /dev/null @@ -1 +0,0 @@ -[1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0] \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_attention_heads b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_attention_heads deleted file mode 100644 index 19c7bdba7b1e9bfe80365a50420a6d538ca503c3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -16 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_beam_groups b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_beams b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_hidden_layers b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_hidden_layers deleted file mode 100644 index dce6588ca1420a69eea5699f2ec5d666daf16a6a..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -36 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_key_value_heads b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_key_value_heads deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_return_sequences b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_train_epochs b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_train_epochs deleted file mode 100644 index 7813681f5b41c028345ca62a2be376bae70b7f61..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -5 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim_args b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim_target_modules b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_attentions b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_dir b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_dir deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_hidden_states b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_scores b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/overwrite_output_dir b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pad_token_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pad_token_id deleted file mode 100644 index 8d9491efc0fb90c90645db18bae256c4f9daff70..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -128004 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/past_index b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_device_eval_batch_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_device_train_batch_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_device_train_batch_size deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_gpu_eval_batch_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_gpu_train_batch_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/prediction_loss_only b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/prefix b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pretraining_tp b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/problem_type b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pruned_heads b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_model_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_organization b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_token b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ray_scope b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/remove_invalid_values b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/remove_unused_columns b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/remove_unused_columns deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/repetition_penalty b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/report_to b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/restore_callback_states_from_checkpoint b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/resume_from_checkpoint b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/return_dict b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/return_dict_in_generate b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rms_norm_eps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rms_norm_eps deleted file mode 100644 index d6bc26a15d7729bbaf8f0e522bea2d0a8781886d..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-06 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rope_scaling b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rope_theta b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rope_theta deleted file mode 100644 index 445ba6793f521c521c5da54f26f04a56ecc83ee2..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -5000000.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/run_name b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/run_name deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_on_each_node b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_only_model b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_safetensors b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_steps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_steps deleted file mode 100644 index 410b14d2ce6f958c13adcb30807e8673074c49d9..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -25 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_strategy b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_total_limit b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_total_limit deleted file mode 100644 index e440e5c842586965a7fb77deda2eca68612b1f53..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -3 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/seed b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/sep_token_id b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/skip_memory_metrics b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/sliding_window b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/sliding_window deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/sliding_window +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/suppress_tokens b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/task_specific_params b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/temperature b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tf32 b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tf_legacy_loss b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tie_encoder_decoder b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tie_word_embeddings b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tokenizer_class b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/top_k b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/top_p b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile_backend b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile_mode b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_dtype b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_empty_cache_steps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torchdynamo b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torchscript b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tpu_metrics_debug b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tpu_num_cores b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/transformers_version b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/typical_p b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_bfloat16 b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_cache b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_cache deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_cpu b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_ipex b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_legacy_prediction_loop b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_liger_kernel b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_mps_device b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_sliding_window b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_sliding_window deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/use_sliding_window +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/vocab_size b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/vocab_size deleted file mode 100644 index 34c44b19378193a3b4fa853df4426fc426c15535..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -128256 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/warmup_ratio b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/warmup_steps b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/warmup_steps deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/weight_decay b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/weight_decay deleted file mode 100644 index d1c6331b3109accd73f01907062e6c174e28200a..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.01 \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.runName b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.runName deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.source.name b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.source.name deleted file mode 100644 index 0a36aad02c6a55095bdc0d02c86ab40ec03251d6..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_m4_max.py \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.source.type b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.user b/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/67a82e40de7245e68717d18c2cb8a858/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/meta.yaml b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/meta.yaml deleted file mode 100644 index f0946f9808e8cfdd7c5c0f2939b3c59da2d70b06..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/742bc5abc90846e89dbeca259a5dd939/artifacts -end_time: 1753092947050 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: 742bc5abc90846e89dbeca259a5dd939 -run_name: ./smollm3_tool_adapter -run_uuid: 742bc5abc90846e89dbeca259a5dd939 -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753092946038 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/_name_or_path b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/_name_or_path deleted file mode 100644 index 3a8ba38a16a767019a426ed9ae2b3cd2746d622d..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM3-3B \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/accelerator_config b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adafactor b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_beta1 b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_beta2 b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_epsilon b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/add_cross_attention b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/architectures b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/architectures deleted file mode 100644 index 55febbc88b5c08e00d44e95e7ed72640c6d034b6..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['SmolLM3ForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/attention_bias b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/attention_dropout b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/auto_find_batch_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/average_tokens_across_devices b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bad_words_ids b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/batch_eval_metrics b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/begin_suppress_tokens b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bf16 b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bf16_full_eval b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bos_token_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bos_token_id deleted file mode 100644 index 5499007cbac38bc897e3c2766b82a647ac28c735..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -128000 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/chunk_size_feed_forward b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/cross_attention_hidden_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/data_seed b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_drop_last b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_num_workers b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_persistent_workers b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_pin_memory b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_prefetch_factor b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_backend b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_broadcast_buffers b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_bucket_cap_mb b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_find_unused_parameters b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_timeout b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/debug b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/decoder_start_token_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/deepspeed b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/disable_tqdm b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/diversity_penalty b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_eval b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_predict b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_sample b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_train b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/early_stopping b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/encoder_no_repeat_ngram_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eos_token_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eos_token_id deleted file mode 100644 index d78b6326c2a8961e1672862523846efebe7ffdd6..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -128012 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_accumulation_steps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_delay b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_do_concat_batches b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_on_start b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_steps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_strategy b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_use_gather_object b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/exponential_decay_length_penalty b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/finetuning_task b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/forced_bos_token_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/forced_eos_token_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16 b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_backend b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_full_eval b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_opt_level b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_config b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_min_num_params b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/full_determinism b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_accumulation_steps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_accumulation_steps deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_checkpointing b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_checkpointing deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_checkpointing_kwargs b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/greater_is_better b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/group_by_length b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/half_precision_backend b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hidden_act b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hidden_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_always_push b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_model_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_private_repo b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_revision b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_strategy b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_token b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/id2label b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ignore_data_skip b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_for_metrics b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_inputs_for_metrics b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_num_input_tokens_seen b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_tokens_per_second b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/initializer_range b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/intermediate_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/intermediate_size deleted file mode 100644 index 9b7babb1160a44a2cf9ae0dbfd43efd0fd70838b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -11008 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/is_decoder b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/is_encoder_decoder b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/jit_mode_eval b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label2id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label_names b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label_smoothing_factor b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/layer_types b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/layer_types deleted file mode 100644 index bc44845b7022ebf6e7ac1a8dac64d5ba0daa15b1..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/layer_types +++ /dev/null @@ -1 +0,0 @@ -['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'] \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/learning_rate b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/learning_rate deleted file mode 100644 index 83f55466e252f8ed340b201b70d8470f0cd54c4b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -0.0003 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/length_column_name b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/length_penalty b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/liger_kernel_config b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/load_best_model_at_end b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/local_rank b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_level b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_level_replica b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_on_each_node b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_dir b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_dir deleted file mode 100644 index 6bc8522feb83b127cc7e1e76a426e8d1a973d69c..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./logs \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_first_step b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_nan_inf_filter b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_steps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_steps deleted file mode 100644 index 7813681f5b41c028345ca62a2be376bae70b7f61..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -5 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_strategy b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/lr_scheduler_kwargs b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/lr_scheduler_type b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/lr_scheduler_type deleted file mode 100644 index 84aa3999b5b7cae7f78b1f77e04d182643005a92..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -cosine \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_grad_norm b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_length b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_position_embeddings b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_position_embeddings deleted file mode 100644 index 83c406c4fd41a6ec8a25290b17cd3894ca57ac4b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -65536 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_steps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_window_layers b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_window_layers deleted file mode 100644 index 368f89ceef179cc546403ac0d5ef1d0e4b340447..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/max_window_layers +++ /dev/null @@ -1 +0,0 @@ -28 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/metric_for_best_model b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/min_length b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/mlp_bias b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/model_type b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/model_type deleted file mode 100644 index 34bda861a9301e0dab58899ec4828a0ee9f464b1..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/model_type +++ /dev/null @@ -1 +0,0 @@ -smollm3 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/mp_parameters b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/neftune_noise_alpha b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_cuda b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_repeat_ngram_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_rope_layer_interval b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_rope_layer_interval deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_rope_layer_interval +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_rope_layers b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_rope_layers deleted file mode 100644 index 2cdfac1b0d51ce92d6d6469bd544bc76a3521cec..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/no_rope_layers +++ /dev/null @@ -1 +0,0 @@ -[1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0] \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_attention_heads b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_attention_heads deleted file mode 100644 index 19c7bdba7b1e9bfe80365a50420a6d538ca503c3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -16 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_beam_groups b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_beams b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_hidden_layers b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_hidden_layers deleted file mode 100644 index dce6588ca1420a69eea5699f2ec5d666daf16a6a..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -36 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_key_value_heads b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_key_value_heads deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_return_sequences b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_train_epochs b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_train_epochs deleted file mode 100644 index 7813681f5b41c028345ca62a2be376bae70b7f61..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -5 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim_args b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim_target_modules b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_attentions b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_dir b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_dir deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_hidden_states b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_scores b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/overwrite_output_dir b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pad_token_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pad_token_id deleted file mode 100644 index 8d9491efc0fb90c90645db18bae256c4f9daff70..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -128004 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/past_index b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_device_eval_batch_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_device_train_batch_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_device_train_batch_size deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_gpu_eval_batch_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_gpu_train_batch_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/prediction_loss_only b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/prefix b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pretraining_tp b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/problem_type b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pruned_heads b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_model_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_organization b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_token b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ray_scope b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/remove_invalid_values b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/remove_unused_columns b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/remove_unused_columns deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/repetition_penalty b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/report_to b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/restore_callback_states_from_checkpoint b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/resume_from_checkpoint b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/return_dict b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/return_dict_in_generate b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rms_norm_eps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rms_norm_eps deleted file mode 100644 index d6bc26a15d7729bbaf8f0e522bea2d0a8781886d..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-06 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rope_scaling b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rope_theta b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rope_theta deleted file mode 100644 index 445ba6793f521c521c5da54f26f04a56ecc83ee2..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -5000000.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/run_name b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/run_name deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_on_each_node b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_only_model b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_safetensors b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_steps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_steps deleted file mode 100644 index 410b14d2ce6f958c13adcb30807e8673074c49d9..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -25 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_strategy b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_total_limit b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_total_limit deleted file mode 100644 index e440e5c842586965a7fb77deda2eca68612b1f53..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -3 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/seed b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/sep_token_id b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/skip_memory_metrics b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/sliding_window b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/sliding_window deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/sliding_window +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/suppress_tokens b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/task_specific_params b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/temperature b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tf32 b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tf_legacy_loss b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tie_encoder_decoder b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tie_word_embeddings b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tokenizer_class b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/top_k b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/top_p b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile_backend b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile_mode b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_dtype b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_empty_cache_steps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torchdynamo b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torchscript b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tpu_metrics_debug b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tpu_num_cores b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/transformers_version b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/typical_p b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_bfloat16 b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_cache b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_cache deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_cpu b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_ipex b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_legacy_prediction_loop b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_liger_kernel b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_mps_device b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_sliding_window b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_sliding_window deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/use_sliding_window +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/vocab_size b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/vocab_size deleted file mode 100644 index 34c44b19378193a3b4fa853df4426fc426c15535..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -128256 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/warmup_ratio b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/warmup_steps b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/warmup_steps deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/weight_decay b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/weight_decay deleted file mode 100644 index d1c6331b3109accd73f01907062e6c174e28200a..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.01 \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.runName b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.runName deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.source.name b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.source.name deleted file mode 100644 index 0a36aad02c6a55095bdc0d02c86ab40ec03251d6..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_m4_max.py \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.source.type b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.user b/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/742bc5abc90846e89dbeca259a5dd939/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/meta.yaml b/mlruns/0/7f40f91efa194516b776d3ec655303de/meta.yaml deleted file mode 100644 index 911688a6d9cc667bd3638d0329dc4b86dabdee56..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/7f40f91efa194516b776d3ec655303de/artifacts -end_time: 1753093092717 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: 7f40f91efa194516b776d3ec655303de -run_name: ./smollm3_robust -run_uuid: 7f40f91efa194516b776d3ec655303de -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753093003269 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/epoch b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/epoch deleted file mode 100644 index 474faaa00b78aa0972737400051ab2cae2c95c8e..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/epoch +++ /dev/null @@ -1,7 +0,0 @@ -1753093021254 0.5517241379310345 2 -1753093034931 1.0 4 -1753093050414 1.5517241379310345 6 -1753093063220 2.0 8 -1753093078799 2.5517241379310347 10 -1753093091599 3.0 12 -1753093092715 3.0 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/grad_norm b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/grad_norm deleted file mode 100644 index 795f6c7e96b8c40a90c567488f5413fa702b6931..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/grad_norm +++ /dev/null @@ -1,6 +0,0 @@ -1753093021254 3.880420446395874 2 -1753093034931 3.3943324089050293 4 -1753093050414 2.610947847366333 6 -1753093063220 2.4453938007354736 8 -1753093078799 2.0043485164642334 10 -1753093091599 2.077148199081421 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/learning_rate b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/learning_rate deleted file mode 100644 index 58917eb26d01b512af95320392a6dd56fc7b4629..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/learning_rate +++ /dev/null @@ -1,6 +0,0 @@ -1753093021254 5e-06 2 -1753093034931 1.5e-05 4 -1753093050414 2.5e-05 6 -1753093063220 3.5e-05 8 -1753093078799 4.5e-05 10 -1753093091599 2.5e-05 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/loss b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/loss deleted file mode 100644 index 7c8e832ebe1ce14260a5b95ac9270cbb06536c39..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/loss +++ /dev/null @@ -1,6 +0,0 @@ -1753093021254 1.7874 2 -1753093034931 1.8274 4 -1753093050414 1.6015 6 -1753093063220 1.4931 8 -1753093078799 1.2396 10 -1753093091599 1.0348 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/total_flos b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/total_flos deleted file mode 100644 index 4f2bafad8bd7228973c214335138b3a2702ef305..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/total_flos +++ /dev/null @@ -1 +0,0 @@ -1753093092715 277789289324544.0 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_loss b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_loss deleted file mode 100644 index 54aa07d96113183c89987c6084f19f17bb4de1b4..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_loss +++ /dev/null @@ -1 +0,0 @@ -1753093092715 1.497321605682373 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_runtime b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_runtime deleted file mode 100644 index f7066f150b66f3c61a3a019e43fb3c02749c32aa..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_runtime +++ /dev/null @@ -1 +0,0 @@ -1753093092715 89.7576 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_samples_per_second b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_samples_per_second deleted file mode 100644 index a826ffa7eba750cbdaa8046eee6b2a4fa5e82536..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_samples_per_second +++ /dev/null @@ -1 +0,0 @@ -1753093092715 0.969 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_steps_per_second b/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_steps_per_second deleted file mode 100644 index 5dc0dcc4123e3e4a34a2dd5406033a915790e977..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/metrics/train_steps_per_second +++ /dev/null @@ -1 +0,0 @@ -1753093092715 0.134 12 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/_name_or_path b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/_name_or_path deleted file mode 100644 index 3a8ba38a16a767019a426ed9ae2b3cd2746d622d..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM3-3B \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/accelerator_config b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adafactor b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_beta1 b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_beta2 b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_epsilon b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/add_cross_attention b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/architectures b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/architectures deleted file mode 100644 index 55febbc88b5c08e00d44e95e7ed72640c6d034b6..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['SmolLM3ForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/attention_bias b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/attention_dropout b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/auto_find_batch_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/average_tokens_across_devices b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bad_words_ids b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/batch_eval_metrics b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/begin_suppress_tokens b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bf16 b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bf16_full_eval b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bos_token_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bos_token_id deleted file mode 100644 index 5499007cbac38bc897e3c2766b82a647ac28c735..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -128000 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/chunk_size_feed_forward b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/cross_attention_hidden_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/data_seed b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_drop_last b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_num_workers b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_persistent_workers b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_pin_memory b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_prefetch_factor b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_backend b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_broadcast_buffers b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_bucket_cap_mb b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_find_unused_parameters b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_timeout b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/debug b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/decoder_start_token_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/deepspeed b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/disable_tqdm b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/diversity_penalty b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_eval b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_predict b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_sample b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_train b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/early_stopping b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/encoder_no_repeat_ngram_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eos_token_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eos_token_id deleted file mode 100644 index d78b6326c2a8961e1672862523846efebe7ffdd6..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -128012 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_accumulation_steps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_delay b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_do_concat_batches b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_on_start b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_steps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_strategy b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_use_gather_object b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/exponential_decay_length_penalty b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/finetuning_task b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/forced_bos_token_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/forced_eos_token_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16 b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_backend b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_full_eval b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_opt_level b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_config b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_min_num_params b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/full_determinism b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_accumulation_steps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_accumulation_steps deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_checkpointing b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_checkpointing deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_checkpointing_kwargs b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/greater_is_better b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/group_by_length b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/half_precision_backend b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hidden_act b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hidden_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_always_push b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_model_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_private_repo b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_revision b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_strategy b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_token b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/id2label b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ignore_data_skip b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_for_metrics b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_inputs_for_metrics b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_num_input_tokens_seen b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_tokens_per_second b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/initializer_range b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/intermediate_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/intermediate_size deleted file mode 100644 index 9b7babb1160a44a2cf9ae0dbfd43efd0fd70838b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -11008 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/is_decoder b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/is_encoder_decoder b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/jit_mode_eval b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label2id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label_names b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label_smoothing_factor b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/layer_types b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/layer_types deleted file mode 100644 index bc44845b7022ebf6e7ac1a8dac64d5ba0daa15b1..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/layer_types +++ /dev/null @@ -1 +0,0 @@ -['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'] \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/learning_rate b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/learning_rate deleted file mode 100644 index 335f2feb826f13cdd89c6b5234de9a8c96396bb7..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -5e-05 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/length_column_name b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/length_penalty b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/liger_kernel_config b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/load_best_model_at_end b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/local_rank b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_level b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_level_replica b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_on_each_node b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_dir b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_dir deleted file mode 100644 index 165db1a557506709f320d4a2ba0b9df0de71bd36..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust/runs/Jul21_11-16-42_Jasons-MacBook-Pro.local \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_first_step b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_nan_inf_filter b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_steps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_steps deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_strategy b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/lr_scheduler_kwargs b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/lr_scheduler_type b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/lr_scheduler_type deleted file mode 100644 index 9c01abbc1e45086237d1a59d6ec6bdea2400f1a7..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -linear \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_grad_norm b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_length b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_position_embeddings b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_position_embeddings deleted file mode 100644 index 83c406c4fd41a6ec8a25290b17cd3894ca57ac4b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -65536 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_steps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_window_layers b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_window_layers deleted file mode 100644 index 368f89ceef179cc546403ac0d5ef1d0e4b340447..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/max_window_layers +++ /dev/null @@ -1 +0,0 @@ -28 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/metric_for_best_model b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/min_length b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/mlp_bias b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/model_type b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/model_type deleted file mode 100644 index 34bda861a9301e0dab58899ec4828a0ee9f464b1..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/model_type +++ /dev/null @@ -1 +0,0 @@ -smollm3 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/mp_parameters b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/neftune_noise_alpha b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_cuda b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_repeat_ngram_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_rope_layer_interval b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_rope_layer_interval deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_rope_layer_interval +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_rope_layers b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_rope_layers deleted file mode 100644 index 2cdfac1b0d51ce92d6d6469bd544bc76a3521cec..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/no_rope_layers +++ /dev/null @@ -1 +0,0 @@ -[1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0] \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_attention_heads b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_attention_heads deleted file mode 100644 index 19c7bdba7b1e9bfe80365a50420a6d538ca503c3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -16 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_beam_groups b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_beams b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_hidden_layers b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_hidden_layers deleted file mode 100644 index dce6588ca1420a69eea5699f2ec5d666daf16a6a..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -36 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_key_value_heads b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_key_value_heads deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_return_sequences b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_train_epochs b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_train_epochs deleted file mode 100644 index e440e5c842586965a7fb77deda2eca68612b1f53..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -3 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim_args b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim_target_modules b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_attentions b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_dir b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_dir deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_hidden_states b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_scores b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/overwrite_output_dir b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pad_token_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pad_token_id deleted file mode 100644 index 8d9491efc0fb90c90645db18bae256c4f9daff70..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -128004 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/past_index b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_device_eval_batch_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_device_train_batch_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_device_train_batch_size deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_gpu_eval_batch_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_gpu_train_batch_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/prediction_loss_only b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/prefix b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pretraining_tp b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/problem_type b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pruned_heads b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_model_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_organization b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_token b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ray_scope b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/remove_invalid_values b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/remove_unused_columns b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/remove_unused_columns deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/repetition_penalty b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/report_to b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/restore_callback_states_from_checkpoint b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/resume_from_checkpoint b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/return_dict b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/return_dict_in_generate b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rms_norm_eps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rms_norm_eps deleted file mode 100644 index d6bc26a15d7729bbaf8f0e522bea2d0a8781886d..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-06 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rope_scaling b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rope_theta b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rope_theta deleted file mode 100644 index 445ba6793f521c521c5da54f26f04a56ecc83ee2..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -5000000.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/run_name b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/run_name deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_on_each_node b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_only_model b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_safetensors b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_steps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_steps deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_strategy b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_total_limit b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_total_limit deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/seed b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/sep_token_id b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/skip_memory_metrics b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/sliding_window b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/sliding_window deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/sliding_window +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/suppress_tokens b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/task_specific_params b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/temperature b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tf32 b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tf_legacy_loss b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tie_encoder_decoder b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tie_word_embeddings b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tokenizer_class b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/top_k b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/top_p b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile_backend b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile_mode b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_dtype b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_empty_cache_steps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torchdynamo b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torchscript b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tpu_metrics_debug b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tpu_num_cores b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/transformers_version b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/typical_p b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_bfloat16 b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_cache b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_cache deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_cpu b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_ipex b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_legacy_prediction_loop b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_liger_kernel b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_mps_device b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_sliding_window b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_sliding_window deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/use_sliding_window +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/vocab_size b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/vocab_size deleted file mode 100644 index 34c44b19378193a3b4fa853df4426fc426c15535..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -128256 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/warmup_ratio b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/warmup_steps b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/warmup_steps deleted file mode 100644 index 9a037142aa3c1b4c490e1a38251620f113465330..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -10 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/weight_decay b/mlruns/0/7f40f91efa194516b776d3ec655303de/params/weight_decay deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.runName b/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.runName deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.source.name b/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.source.name deleted file mode 100644 index 124d840089086dc6280e8eb4e84f59ad6b0b015f..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_simple_robust.py \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.source.type b/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.user b/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/7f40f91efa194516b776d3ec655303de/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/meta.yaml b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/meta.yaml deleted file mode 100644 index 61bd3d950c76737fdc771eda1d7a8284f9ab54ab..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/a9ce4acd189c4737ba94346689f2416b/artifacts -end_time: 1753103996877 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: a9ce4acd189c4737ba94346689f2416b -run_name: ./smollm3_robust -run_uuid: a9ce4acd189c4737ba94346689f2416b -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753099081712 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/epoch b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/epoch deleted file mode 100644 index 47c2a0f1a2d75cbbf233a5d69ab11114a44d14b6..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/epoch +++ /dev/null @@ -1,336 +0,0 @@ -1753099099197 0.0299625468164794 2 -1753099117338 0.0599250936329588 4 -1753099133987 0.0898876404494382 6 -1753099150841 0.1198501872659176 8 -1753099167353 0.149812734082397 10 -1753099183522 0.1797752808988764 12 -1753099199672 0.20973782771535582 14 -1753099216002 0.2397003745318352 16 -1753099232389 0.2696629213483146 18 -1753099248975 0.299625468164794 20 -1753099267528 0.3295880149812734 22 -1753099287703 0.3595505617977528 24 -1753099306825 0.3895131086142322 26 -1753099326820 0.41947565543071164 28 -1753099345585 0.449438202247191 30 -1753099363772 0.4794007490636704 32 -1753099380259 0.5093632958801498 34 -1753099396729 0.5393258426966292 36 -1753099413982 0.5692883895131086 38 -1753099431269 0.599250936329588 40 -1753099447383 0.6292134831460674 42 -1753099464471 0.6591760299625468 44 -1753099481040 0.6891385767790262 46 -1753099497069 0.7191011235955056 48 -1753099515752 0.7490636704119851 50 -1753099534493 0.7790262172284644 52 -1753099551081 0.8089887640449438 54 -1753099570403 0.8389513108614233 56 -1753099586817 0.8689138576779026 58 -1753099602727 0.898876404494382 60 -1753099620077 0.9288389513108615 62 -1753099638583 0.9588014981273408 64 -1753099654694 0.9887640449438202 66 -1753099668722 1.0149812734082397 68 -1753099684879 1.0449438202247192 70 -1753099700891 1.0749063670411985 72 -1753099716854 1.104868913857678 74 -1753099733096 1.1348314606741572 76 -1753099749599 1.1647940074906367 78 -1753099765909 1.1947565543071161 80 -1753099783263 1.2247191011235956 82 -1753099799395 1.2546816479400749 84 -1753099815643 1.2846441947565543 86 -1753099831792 1.3146067415730336 88 -1753099847949 1.344569288389513 90 -1753099863985 1.3745318352059925 92 -1753099880315 1.404494382022472 94 -1753099896452 1.4344569288389513 96 -1753099912502 1.4644194756554307 98 -1753099928706 1.49438202247191 100 -1753099946155 1.5243445692883895 102 -1753099962388 1.554307116104869 104 -1753099979182 1.5842696629213484 106 -1753099997221 1.6142322097378277 108 -1753100015117 1.6441947565543071 110 -1753100031324 1.6741573033707864 112 -1753100047565 1.7041198501872659 114 -1753100064664 1.7340823970037453 116 -1753100080943 1.7640449438202248 118 -1753100097331 1.7940074906367043 120 -1753100114685 1.8239700374531835 122 -1753100131933 1.8539325842696628 124 -1753100148753 1.8838951310861423 126 -1753100165684 1.9138576779026217 128 -1753100181802 1.9438202247191012 130 -1753100197990 1.9737827715355807 132 -1753100212648 2.0 134 -1753100228619 2.0299625468164795 136 -1753100244985 2.059925093632959 138 -1753100263859 2.0898876404494384 140 -1753100281198 2.1198501872659175 142 -1753100297284 2.149812734082397 144 -1753100313650 2.1797752808988764 146 -1753100329826 2.209737827715356 148 -1753100346032 2.2397003745318353 150 -1753100365008 2.2696629213483144 152 -1753100381122 2.299625468164794 154 -1753100397392 2.3295880149812733 156 -1753100413896 2.359550561797753 158 -1753100430772 2.3895131086142323 160 -1753100448101 2.4194756554307117 162 -1753100464233 2.449438202247191 164 -1753100480637 2.4794007490636703 166 -1753100496812 2.5093632958801497 168 -1753100512763 2.539325842696629 170 -1753100529101 2.5692883895131087 172 -1753100545662 2.599250936329588 174 -1753100561842 2.629213483146067 176 -1753100577915 2.6591760299625467 178 -1753100597827 2.689138576779026 180 -1753100617625 2.7191011235955056 182 -1753100633805 2.749063670411985 184 -1753100649953 2.7790262172284645 186 -1753100666118 2.808988764044944 188 -1753100682558 2.8389513108614235 190 -1753100698868 2.8689138576779025 192 -1753100714949 2.898876404494382 194 -1753100731373 2.9288389513108615 196 -1753100747827 2.958801498127341 198 -1753100764846 2.98876404494382 200 -1753100781020 3.0149812734082397 202 -1753100797597 3.044943820224719 204 -1753100813317 3.0749063670411987 206 -1753100830432 3.1048689138576777 208 -1753100850133 3.134831460674157 210 -1753100866690 3.1647940074906367 212 -1753100882925 3.194756554307116 214 -1753100899245 3.2247191011235956 216 -1753100915684 3.254681647940075 218 -1753100932178 3.284644194756554 220 -1753100949229 3.3146067415730336 222 -1753100966051 3.344569288389513 224 -1753100982253 3.3745318352059925 226 -1753100999028 3.404494382022472 228 -1753101015198 3.4344569288389515 230 -1753101031769 3.464419475655431 232 -1753101048085 3.49438202247191 234 -1753101064616 3.5243445692883895 236 -1753101080911 3.554307116104869 238 -1753101097275 3.5842696629213484 240 -1753101114344 3.6142322097378274 242 -1753101130342 3.644194756554307 244 -1753101146579 3.6741573033707864 246 -1753101162626 3.704119850187266 248 -1753101181901 3.7340823970037453 250 -1753101198100 3.764044943820225 252 -1753101214528 3.7940074906367043 254 -1753101231597 3.8239700374531838 256 -1753101248995 3.853932584269663 258 -1753101266907 3.8838951310861423 260 -1753101284453 3.9138576779026217 262 -1753101300751 3.943820224719101 264 -1753101319044 3.9737827715355807 266 -1753101333739 4.0 268 -1753101351938 4.0299625468164795 270 -1753101370058 4.059925093632959 272 -1753101387398 4.089887640449438 274 -1753101404642 4.119850187265918 276 -1753101422160 4.149812734082397 278 -1753101439833 4.179775280898877 280 -1753101457342 4.209737827715355 282 -1753101473364 4.239700374531835 284 -1753101489441 4.269662921348314 286 -1753101506484 4.299625468164794 288 -1753101522875 4.329588014981273 290 -1753101539258 4.359550561797753 292 -1753101551955 4.389513108614232 294 -1753101567021 4.419475655430712 296 -1753101579449 4.449438202247191 298 -1753101591811 4.479400749063671 300 -1753101605116 4.50936329588015 302 -1753101617605 4.539325842696629 304 -1753101630107 4.569288389513108 306 -1753101642603 4.599250936329588 308 -1753101654987 4.629213483146067 310 -1753101667355 4.659176029962547 312 -1753101679931 4.689138576779026 314 -1753101692230 4.719101123595506 316 -1753101704746 4.749063670411985 318 -1753101717711 4.7790262172284645 320 -1753101731491 4.808988764044944 322 -1753101744045 4.8389513108614235 324 -1753101756502 4.868913857677903 326 -1753101768918 4.898876404494382 328 -1753101781177 4.928838951310862 330 -1753101794186 4.9588014981273405 332 -1753101806829 4.98876404494382 334 -1753101817841 5.01498127340824 336 -1753101830273 5.044943820224719 338 -1753101842727 5.074906367041199 340 -1753101856035 5.104868913857678 342 -1753101868461 5.134831460674158 344 -1753101881035 5.164794007490637 346 -1753101893425 5.194756554307116 348 -1753101905823 5.224719101123595 350 -1753101918118 5.254681647940075 352 -1753101930581 5.284644194756554 354 -1753101942951 5.314606741573034 356 -1753101955464 5.344569288389513 358 -1753101967892 5.3745318352059925 360 -1753101981213 5.404494382022472 362 -1753101993491 5.4344569288389515 364 -1753102005939 5.464419475655431 366 -1753102018511 5.49438202247191 368 -1753102030741 5.52434456928839 370 -1753102043029 5.554307116104869 372 -1753102055570 5.584269662921348 374 -1753102068468 5.614232209737827 376 -1753102081353 5.644194756554307 378 -1753102093722 5.674157303370786 380 -1753102107017 5.704119850187266 382 -1753102119465 5.734082397003745 384 -1753102132053 5.764044943820225 386 -1753102144419 5.794007490636704 388 -1753102156818 5.823970037453184 390 -1753102169259 5.853932584269663 392 -1753102181603 5.883895131086143 394 -1753102194057 5.913857677902621 396 -1753102206442 5.943820224719101 398 -1753102218668 5.97378277153558 400 -1753102230435 6.0 402 -1753102242794 6.0299625468164795 404 -1753102255032 6.059925093632959 406 -1753102267198 6.089887640449438 408 -1753102279979 6.119850187265918 410 -1753102292549 6.149812734082397 412 -1753102305253 6.179775280898877 414 -1753102322016 6.209737827715355 416 -1753102339290 6.239700374531835 418 -1753102356703 6.269662921348314 420 -1753102378067 6.299625468164794 422 -1753102396288 6.329588014981273 424 -1753102417476 6.359550561797753 426 -1753102435595 6.389513108614232 428 -1753102452231 6.419475655430712 430 -1753102469008 6.449438202247191 432 -1753102486414 6.479400749063671 434 -1753102504814 6.50936329588015 436 -1753102522833 6.539325842696629 438 -1753102541953 6.569288389513108 440 -1753102562836 6.599250936329588 442 -1753102579387 6.629213483146067 444 -1753102596188 6.659176029962547 446 -1753102613449 6.689138576779026 448 -1753102625983 6.719101123595506 450 -1753102638547 6.749063670411985 452 -1753102651016 6.7790262172284645 454 -1753102663839 6.808988764044944 456 -1753102676468 6.8389513108614235 458 -1753102689469 6.868913857677903 460 -1753102703211 6.898876404494382 462 -1753102715962 6.928838951310862 464 -1753102728637 6.9588014981273405 466 -1753102740978 6.98876404494382 468 -1753102751737 7.01498127340824 470 -1753102763889 7.044943820224719 472 -1753102776221 7.074906367041199 474 -1753102788524 7.104868913857678 476 -1753102801040 7.134831460674158 478 -1753102813226 7.164794007490637 480 -1753102826667 7.194756554307116 482 -1753102838780 7.224719101123595 484 -1753102850896 7.254681647940075 486 -1753102863487 7.284644194756554 488 -1753102876122 7.314606741573034 490 -1753102888516 7.344569288389513 492 -1753102900844 7.3745318352059925 494 -1753102913114 7.404494382022472 496 -1753102925750 7.4344569288389515 498 -1753102938142 7.464419475655431 500 -1753102951176 7.49438202247191 502 -1753102963259 7.52434456928839 504 -1753102975580 7.554307116104869 506 -1753102987954 7.584269662921348 508 -1753103000438 7.614232209737827 510 -1753103012937 7.644194756554307 512 -1753103025293 7.674157303370786 514 -1753103037822 7.704119850187266 516 -1753103050099 7.734082397003745 518 -1753103062194 7.764044943820225 520 -1753103075646 7.794007490636704 522 -1753103087907 7.823970037453184 524 -1753103100239 7.853932584269663 526 -1753103112507 7.883895131086143 528 -1753103124952 7.913857677902621 530 -1753103137352 7.943820224719101 532 -1753103149521 7.97378277153558 534 -1753103160281 8.0 536 -1753103172737 8.02996254681648 538 -1753103185085 8.059925093632959 540 -1753103198775 8.089887640449438 542 -1753103211421 8.119850187265918 544 -1753103223847 8.149812734082397 546 -1753103236348 8.179775280898877 548 -1753103248804 8.209737827715356 550 -1753103261193 8.239700374531836 552 -1753103273576 8.269662921348315 554 -1753103285828 8.299625468164795 556 -1753103298280 8.329588014981274 558 -1753103310727 8.359550561797754 560 -1753103324545 8.389513108614231 562 -1753103336895 8.41947565543071 564 -1753103349333 8.44943820224719 566 -1753103361816 8.47940074906367 568 -1753103374215 8.50936329588015 570 -1753103386774 8.539325842696629 572 -1753103399224 8.569288389513108 574 -1753103411594 8.599250936329588 576 -1753103423999 8.629213483146067 578 -1753103436440 8.659176029962547 580 -1753103449807 8.689138576779026 582 -1753103462230 8.719101123595506 584 -1753103474665 8.749063670411985 586 -1753103487149 8.779026217228465 588 -1753103499704 8.808988764044944 590 -1753103512138 8.838951310861423 592 -1753103524731 8.868913857677903 594 -1753103537109 8.898876404494382 596 -1753103549566 8.928838951310862 598 -1753103561938 8.958801498127341 600 -1753103575590 8.98876404494382 602 -1753103586472 9.014981273408239 604 -1753103598901 9.044943820224718 606 -1753103611487 9.074906367041198 608 -1753103623784 9.104868913857677 610 -1753103636042 9.134831460674157 612 -1753103648314 9.164794007490636 614 -1753103660702 9.194756554307116 616 -1753103673071 9.224719101123595 618 -1753103685477 9.254681647940075 620 -1753103698696 9.284644194756554 622 -1753103710924 9.314606741573034 624 -1753103723356 9.344569288389513 626 -1753103736037 9.374531835205993 628 -1753103748360 9.404494382022472 630 -1753103760610 9.434456928838951 632 -1753103772762 9.464419475655431 634 -1753103785207 9.49438202247191 636 -1753103797805 9.52434456928839 638 -1753103810295 9.55430711610487 640 -1753103823660 9.584269662921349 642 -1753103836107 9.614232209737828 644 -1753103848492 9.644194756554308 646 -1753103861064 9.674157303370787 648 -1753103873523 9.704119850187267 650 -1753103885869 9.734082397003746 652 -1753103897958 9.764044943820224 654 -1753103910220 9.794007490636703 656 -1753103922705 9.823970037453183 658 -1753103935001 9.853932584269662 660 -1753103948168 9.883895131086142 662 -1753103960368 9.913857677902621 664 -1753103972703 9.9438202247191 666 -1753103985078 9.97378277153558 668 -1753103995846 10.0 670 -1753103996876 10.0 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/grad_norm b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/grad_norm deleted file mode 100644 index 42d96e3485a38c148099b778b1f2bc8688d563e8..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/grad_norm +++ /dev/null @@ -1,335 +0,0 @@ -1753099099197 3.722816228866577 2 -1753099117338 3.4061238765716553 4 -1753099133987 2.510838270187378 6 -1753099150841 2.095601797103882 8 -1753099167353 2.001142978668213 10 -1753099183522 1.8859666585922241 12 -1753099199672 2.0142483711242676 14 -1753099216002 1.6467182636260986 16 -1753099232389 1.2968051433563232 18 -1753099248975 1.001763105392456 20 -1753099267528 0.47088268399238586 22 -1753099287703 0.6152827739715576 24 -1753099306825 0.5034743547439575 26 -1753099326820 0.5429691076278687 28 -1753099345585 0.6162884831428528 30 -1753099363772 0.7645140886306763 32 -1753099380259 0.600817084312439 34 -1753099396729 0.4553565979003906 36 -1753099413982 0.7166281342506409 38 -1753099431269 0.42647111415863037 40 -1753099447383 0.4419521689414978 42 -1753099464471 0.49058037996292114 44 -1753099481040 0.3956356644630432 46 -1753099497069 0.4841479957103729 48 -1753099515752 0.5453945398330688 50 -1753099534493 0.44312047958374023 52 -1753099551081 0.5744581818580627 54 -1753099570403 0.9026234745979309 56 -1753099586817 0.4250333905220032 58 -1753099602727 0.6075786352157593 60 -1753099620077 0.6797979474067688 62 -1753099638583 0.296935498714447 64 -1753099654694 0.28508949279785156 66 -1753099668722 0.44689077138900757 68 -1753099684879 0.36559513211250305 70 -1753099700891 0.3891605734825134 72 -1753099716854 0.36933985352516174 74 -1753099733096 0.41140684485435486 76 -1753099749599 0.5980117321014404 78 -1753099765909 0.23804394900798798 80 -1753099783263 0.23067264258861542 82 -1753099799395 0.36974087357521057 84 -1753099815643 0.4070376753807068 86 -1753099831792 0.2764546275138855 88 -1753099847949 0.20103320479393005 90 -1753099863985 0.26625537872314453 92 -1753099880315 0.24225257337093353 94 -1753099896452 0.3456275463104248 96 -1753099912502 0.21905088424682617 98 -1753099928706 0.5236210227012634 100 -1753099946155 0.35533836483955383 102 -1753099962388 0.3094329833984375 104 -1753099979182 0.24415704607963562 106 -1753099997221 0.44377341866493225 108 -1753100015117 0.4970695972442627 110 -1753100031324 0.20200444757938385 112 -1753100047565 0.21261580288410187 114 -1753100064664 0.2608441412448883 116 -1753100080943 0.20836834609508514 118 -1753100097331 0.29434454441070557 120 -1753100114685 0.2228062003850937 122 -1753100131933 0.3206649124622345 124 -1753100148753 0.18132628500461578 126 -1753100165684 0.2030370533466339 128 -1753100181802 0.2079804688692093 130 -1753100197990 0.23763149976730347 132 -1753100212648 0.21091270446777344 134 -1753100228619 0.21226680278778076 136 -1753100244985 0.2453354448080063 138 -1753100263859 0.17463107407093048 140 -1753100281198 0.307685524225235 142 -1753100297284 0.23119647800922394 144 -1753100313650 0.15694215893745422 146 -1753100329826 0.26551949977874756 148 -1753100346032 0.21518471837043762 150 -1753100365008 0.1955074667930603 152 -1753100381122 0.24419444799423218 154 -1753100397392 0.29677078127861023 156 -1753100413896 0.19572412967681885 158 -1753100430772 0.20857453346252441 160 -1753100448101 0.14987444877624512 162 -1753100464233 0.22435317933559418 164 -1753100480637 0.29637566208839417 166 -1753100496812 0.24401557445526123 168 -1753100512763 0.19472604990005493 170 -1753100529101 0.21067409217357635 172 -1753100545662 0.157924622297287 174 -1753100561842 0.22833997011184692 176 -1753100577915 0.1746760606765747 178 -1753100597827 0.25366461277008057 180 -1753100617625 0.1569383144378662 182 -1753100633805 0.10542655736207962 184 -1753100649953 0.11274619400501251 186 -1753100666118 0.2910693287849426 188 -1753100682558 0.15630793571472168 190 -1753100698868 0.1520637422800064 192 -1753100714949 0.1700475662946701 194 -1753100731373 0.2383895367383957 196 -1753100747827 0.2498546838760376 198 -1753100764846 0.24929316341876984 200 -1753100781020 0.1542021781206131 202 -1753100797597 0.15488319098949432 204 -1753100813317 0.22363406419754028 206 -1753100830432 0.23685605823993683 208 -1753100850133 0.18075552582740784 210 -1753100866690 0.18518349528312683 212 -1753100882925 0.29419267177581787 214 -1753100899245 0.13987213373184204 216 -1753100915684 0.2520444989204407 218 -1753100932178 0.12775233387947083 220 -1753100949229 0.15709002315998077 222 -1753100966051 0.17532695829868317 224 -1753100982253 0.21435686945915222 226 -1753100999028 0.23152689635753632 228 -1753101015198 0.20635737478733063 230 -1753101031769 0.18343976140022278 232 -1753101048085 0.14534398913383484 234 -1753101064616 0.2103314846754074 236 -1753101080911 0.13487893342971802 238 -1753101097275 0.13789679110050201 240 -1753101114344 0.21411816775798798 242 -1753101130342 0.22373120486736298 244 -1753101146579 0.17626555263996124 246 -1753101162626 0.15838968753814697 248 -1753101181901 0.15824346244335175 250 -1753101198100 0.1346617490053177 252 -1753101214528 0.15259407460689545 254 -1753101231597 0.10529737174510956 256 -1753101248995 0.21343955397605896 258 -1753101266907 0.19346514344215393 260 -1753101284453 0.11122281849384308 262 -1753101300751 0.1087551936507225 264 -1753101319044 0.10952229797840118 266 -1753101333739 0.21926254034042358 268 -1753101351938 0.11138854920864105 270 -1753101370058 0.18572665750980377 272 -1753101387398 0.11445632576942444 274 -1753101404642 0.13404381275177002 276 -1753101422160 0.22214095294475555 278 -1753101439833 0.16340972483158112 280 -1753101457342 0.1711144745349884 282 -1753101473364 0.13778892159461975 284 -1753101489441 0.15408281981945038 286 -1753101506484 0.12467946112155914 288 -1753101522875 0.1988392323255539 290 -1753101539258 0.1968161165714264 292 -1753101551955 0.16299834847450256 294 -1753101567021 0.20078356564044952 296 -1753101579449 0.10105714946985245 298 -1753101591811 0.14914706349372864 300 -1753101605116 0.12265647202730179 302 -1753101617605 0.11637593805789948 304 -1753101630107 0.11158251017332077 306 -1753101642603 0.14547735452651978 308 -1753101654987 0.15811274945735931 310 -1753101667355 0.1495320349931717 312 -1753101679931 0.14297890663146973 314 -1753101692230 0.1328546106815338 316 -1753101704746 0.15500609576702118 318 -1753101717711 0.1334412395954132 320 -1753101731491 0.11698229610919952 322 -1753101744045 0.10686776041984558 324 -1753101756502 0.13976678252220154 326 -1753101768918 0.1366475522518158 328 -1753101781177 0.16024161875247955 330 -1753101794186 0.11987770348787308 332 -1753101806829 0.12122765928506851 334 -1753101817841 0.14996588230133057 336 -1753101830273 0.1114385575056076 338 -1753101842727 0.11765974760055542 340 -1753101856035 0.18139192461967468 342 -1753101868461 0.12986940145492554 344 -1753101881035 0.12934552133083344 346 -1753101893425 0.13571321964263916 348 -1753101905823 0.13134482502937317 350 -1753101918118 0.1282089650630951 352 -1753101930581 0.12280598282814026 354 -1753101942951 0.1045297384262085 356 -1753101955464 0.10783874243497849 358 -1753101967892 0.28152090311050415 360 -1753101981213 0.12164758890867233 362 -1753101993491 0.16581693291664124 364 -1753102005939 0.10845212638378143 366 -1753102018511 0.11847496032714844 368 -1753102030741 0.1990269422531128 370 -1753102043029 0.13564549386501312 372 -1753102055570 0.12416112422943115 374 -1753102068468 0.21298328042030334 376 -1753102081353 0.12686102092266083 378 -1753102093722 0.11951331794261932 380 -1753102107017 0.15928389132022858 382 -1753102119465 0.11539971828460693 384 -1753102132053 0.1547948122024536 386 -1753102144419 0.15363723039627075 388 -1753102156818 0.13726523518562317 390 -1753102169259 0.11726450175046921 392 -1753102181603 0.09192411601543427 394 -1753102194057 0.12325743585824966 396 -1753102206442 0.1755753755569458 398 -1753102218668 0.1385994851589203 400 -1753102230435 0.12166863679885864 402 -1753102242794 0.135303795337677 404 -1753102255032 0.14891491830348969 406 -1753102267198 0.13520725071430206 408 -1753102279979 0.1810332089662552 410 -1753102292549 0.12100538611412048 412 -1753102305253 0.11748135834932327 414 -1753102322016 0.1747063398361206 416 -1753102339290 0.15986011922359467 418 -1753102356703 0.130351260304451 420 -1753102378067 0.09828981012105942 422 -1753102396288 0.11312004178762436 424 -1753102417476 0.1399003267288208 426 -1753102435595 0.19829927384853363 428 -1753102452231 0.12163085490465164 430 -1753102469008 0.1325439214706421 432 -1753102486414 0.10004521906375885 434 -1753102504814 0.22187596559524536 436 -1753102522833 0.12271460145711899 438 -1753102541953 0.16459853947162628 440 -1753102562836 0.11979979276657104 442 -1753102579387 0.24495816230773926 444 -1753102596188 0.1692766696214676 446 -1753102613449 0.1197497546672821 448 -1753102625983 0.09570708870887756 450 -1753102638547 0.10233695805072784 452 -1753102651016 0.13967226445674896 454 -1753102663839 0.14382530748844147 456 -1753102676468 0.18934357166290283 458 -1753102689469 0.09580834209918976 460 -1753102703211 0.16412261128425598 462 -1753102715962 0.16064710915088654 464 -1753102728637 0.17852792143821716 466 -1753102740978 0.17345724999904633 468 -1753102751737 0.1276731640100479 470 -1753102763889 0.1590038686990738 472 -1753102776221 0.12557406723499298 474 -1753102788524 0.21979106962680817 476 -1753102801040 0.19031833112239838 478 -1753102813226 0.23122136294841766 480 -1753102826667 0.12799808382987976 482 -1753102838780 0.09180168062448502 484 -1753102850896 0.13481584191322327 486 -1753102863487 0.1296118050813675 488 -1753102876122 0.12827853858470917 490 -1753102888516 0.12804710865020752 492 -1753102900844 0.14896075427532196 494 -1753102913114 0.11220405250787735 496 -1753102925750 0.10928992182016373 498 -1753102938142 0.12223732471466064 500 -1753102951176 0.12335951626300812 502 -1753102963259 0.12804357707500458 504 -1753102975580 0.11515405774116516 506 -1753102987954 0.1302037239074707 508 -1753103000438 0.1083846315741539 510 -1753103012937 0.11610821634531021 512 -1753103025293 0.08522523194551468 514 -1753103037822 0.0896814838051796 516 -1753103050099 0.15006373822689056 518 -1753103062194 0.1151801198720932 520 -1753103075646 0.10852045565843582 522 -1753103087907 0.20251013338565826 524 -1753103100239 0.1366216391324997 526 -1753103112507 0.12531571090221405 528 -1753103124952 0.14330235123634338 530 -1753103137352 0.1384236365556717 532 -1753103149521 0.15142616629600525 534 -1753103160281 0.11959421634674072 536 -1753103172737 0.14299067854881287 538 -1753103185085 0.15421266853809357 540 -1753103198775 0.11853787302970886 542 -1753103211421 0.11856868863105774 544 -1753103223847 0.10615833103656769 546 -1753103236348 0.14076802134513855 548 -1753103248804 0.12974779307842255 550 -1753103261193 0.11985109746456146 552 -1753103273576 0.21702255308628082 554 -1753103285828 0.14472782611846924 556 -1753103298280 0.1377476155757904 558 -1753103310727 0.15174493193626404 560 -1753103324545 0.1872493326663971 562 -1753103336895 0.1931612491607666 564 -1753103349333 0.18126638233661652 566 -1753103361816 0.16266301274299622 568 -1753103374215 0.1196930930018425 570 -1753103386774 0.14626623690128326 572 -1753103399224 0.12978863716125488 574 -1753103411594 0.16399815678596497 576 -1753103423999 0.13139177858829498 578 -1753103436440 0.10869726538658142 580 -1753103449807 0.11966531723737717 582 -1753103462230 0.15545986592769623 584 -1753103474665 0.1065947413444519 586 -1753103487149 0.17026154696941376 588 -1753103499704 0.1124507337808609 590 -1753103512138 0.11526783555746078 592 -1753103524731 0.11815094202756882 594 -1753103537109 0.115007683634758 596 -1753103549566 0.12640294432640076 598 -1753103561938 0.18849849700927734 600 -1753103575590 0.09228724241256714 602 -1753103586472 0.10718464851379395 604 -1753103598901 0.15939927101135254 606 -1753103611487 0.10619861632585526 608 -1753103623784 0.1796608418226242 610 -1753103636042 0.11252462863922119 612 -1753103648314 0.12761755287647247 614 -1753103660702 0.11963596194982529 616 -1753103673071 0.14573565125465393 618 -1753103685477 0.1269391030073166 620 -1753103698696 0.14515313506126404 622 -1753103710924 0.1334671527147293 624 -1753103723356 0.12002371996641159 626 -1753103736037 0.14389854669570923 628 -1753103748360 0.11606308072805405 630 -1753103760610 0.18154092133045197 632 -1753103772762 0.2114187330007553 634 -1753103785207 0.12829148769378662 636 -1753103797805 0.14554138481616974 638 -1753103810295 0.11862059682607651 640 -1753103823660 0.11568762362003326 642 -1753103836107 0.10297347605228424 644 -1753103848492 0.14339599013328552 646 -1753103861064 0.15518030524253845 648 -1753103873523 0.13086441159248352 650 -1753103885869 0.1343931257724762 652 -1753103897958 0.16945284605026245 654 -1753103910220 0.14710381627082825 656 -1753103922705 0.15412309765815735 658 -1753103935001 0.1276121288537979 660 -1753103948168 0.17896993458271027 662 -1753103960368 0.18712849915027618 664 -1753103972703 0.14763027429580688 666 -1753103985078 0.13407444953918457 668 -1753103995846 0.18560239672660828 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/learning_rate b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/learning_rate deleted file mode 100644 index df81d79363b924786f24fe9b612eb70ac23358ab..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/learning_rate +++ /dev/null @@ -1,335 +0,0 @@ -1753099099197 5e-06 2 -1753099117338 1.5e-05 4 -1753099133987 2.5e-05 6 -1753099150841 3.5e-05 8 -1753099167353 4.5e-05 10 -1753099183522 4.992424242424243e-05 12 -1753099199672 4.9772727272727275e-05 14 -1753099216002 4.962121212121213e-05 16 -1753099232389 4.946969696969697e-05 18 -1753099248975 4.931818181818182e-05 20 -1753099267528 4.9166666666666665e-05 22 -1753099287703 4.901515151515152e-05 24 -1753099306825 4.886363636363637e-05 26 -1753099326820 4.8712121212121216e-05 28 -1753099345585 4.856060606060606e-05 30 -1753099363772 4.840909090909091e-05 32 -1753099380259 4.825757575757576e-05 34 -1753099396729 4.810606060606061e-05 36 -1753099413982 4.795454545454546e-05 38 -1753099431269 4.7803030303030304e-05 40 -1753099447383 4.765151515151515e-05 42 -1753099464471 4.75e-05 44 -1753099481040 4.7348484848484855e-05 46 -1753099497069 4.71969696969697e-05 48 -1753099515752 4.704545454545455e-05 50 -1753099534493 4.689393939393939e-05 52 -1753099551081 4.6742424242424245e-05 54 -1753099570403 4.659090909090909e-05 56 -1753099586817 4.6439393939393944e-05 58 -1753099602727 4.628787878787879e-05 60 -1753099620077 4.6136363636363635e-05 62 -1753099638583 4.598484848484849e-05 64 -1753099654694 4.5833333333333334e-05 66 -1753099668722 4.5681818181818186e-05 68 -1753099684879 4.553030303030303e-05 70 -1753099700891 4.5378787878787885e-05 72 -1753099716854 4.522727272727273e-05 74 -1753099733096 4.5075757575757577e-05 76 -1753099749599 4.492424242424242e-05 78 -1753099765909 4.4772727272727275e-05 80 -1753099783263 4.462121212121213e-05 82 -1753099799395 4.4469696969696973e-05 84 -1753099815643 4.431818181818182e-05 86 -1753099831792 4.4166666666666665e-05 88 -1753099847949 4.401515151515152e-05 90 -1753099863985 4.386363636363637e-05 92 -1753099880315 4.3712121212121216e-05 94 -1753099896452 4.356060606060606e-05 96 -1753099912502 4.340909090909091e-05 98 -1753099928706 4.325757575757576e-05 100 -1753099946155 4.3106060606060606e-05 102 -1753099962388 4.295454545454546e-05 104 -1753099979182 4.2803030303030305e-05 106 -1753099997221 4.265151515151515e-05 108 -1753100015117 4.25e-05 110 -1753100031324 4.234848484848485e-05 112 -1753100047565 4.21969696969697e-05 114 -1753100064664 4.204545454545455e-05 116 -1753100080943 4.189393939393939e-05 118 -1753100097331 4.1742424242424246e-05 120 -1753100114685 4.159090909090909e-05 122 -1753100131933 4.143939393939394e-05 124 -1753100148753 4.128787878787879e-05 126 -1753100165684 4.113636363636364e-05 128 -1753100181802 4.098484848484849e-05 130 -1753100197990 4.0833333333333334e-05 132 -1753100212648 4.068181818181818e-05 134 -1753100228619 4.053030303030303e-05 136 -1753100244985 4.0378787878787885e-05 138 -1753100263859 4.022727272727273e-05 140 -1753100281198 4.007575757575758e-05 142 -1753100297284 3.992424242424242e-05 144 -1753100313650 3.9772727272727275e-05 146 -1753100329826 3.962121212121213e-05 148 -1753100346032 3.9469696969696974e-05 150 -1753100365008 3.931818181818182e-05 152 -1753100381122 3.9166666666666665e-05 154 -1753100397392 3.901515151515152e-05 156 -1753100413896 3.8863636363636364e-05 158 -1753100430772 3.8712121212121217e-05 160 -1753100448101 3.856060606060606e-05 162 -1753100464233 3.840909090909091e-05 164 -1753100480637 3.825757575757576e-05 166 -1753100496812 3.810606060606061e-05 168 -1753100512763 3.795454545454545e-05 170 -1753100529101 3.7803030303030305e-05 172 -1753100545662 3.765151515151516e-05 174 -1753100561842 3.7500000000000003e-05 176 -1753100577915 3.734848484848485e-05 178 -1753100597827 3.7196969696969695e-05 180 -1753100617625 3.704545454545455e-05 182 -1753100633805 3.68939393939394e-05 184 -1753100649953 3.6742424242424246e-05 186 -1753100666118 3.659090909090909e-05 188 -1753100682558 3.643939393939394e-05 190 -1753100698868 3.628787878787879e-05 192 -1753100714949 3.613636363636364e-05 194 -1753100731373 3.598484848484849e-05 196 -1753100747827 3.5833333333333335e-05 198 -1753100764846 3.568181818181818e-05 200 -1753100781020 3.553030303030303e-05 202 -1753100797597 3.537878787878788e-05 204 -1753100813317 3.522727272727273e-05 206 -1753100830432 3.507575757575758e-05 208 -1753100850133 3.492424242424242e-05 210 -1753100866690 3.4772727272727276e-05 212 -1753100882925 3.462121212121212e-05 214 -1753100899245 3.4469696969696974e-05 216 -1753100915684 3.431818181818182e-05 218 -1753100932178 3.4166666666666666e-05 220 -1753100949229 3.401515151515152e-05 222 -1753100966051 3.3863636363636364e-05 224 -1753100982253 3.371212121212121e-05 226 -1753100999028 3.356060606060606e-05 228 -1753101015198 3.3409090909090915e-05 230 -1753101031769 3.325757575757576e-05 232 -1753101048085 3.310606060606061e-05 234 -1753101064616 3.295454545454545e-05 236 -1753101080911 3.2803030303030305e-05 238 -1753101097275 3.265151515151516e-05 240 -1753101114344 3.2500000000000004e-05 242 -1753101130342 3.234848484848485e-05 244 -1753101146579 3.2196969696969696e-05 246 -1753101162626 3.204545454545455e-05 248 -1753101181901 3.18939393939394e-05 250 -1753101198100 3.174242424242425e-05 252 -1753101214528 3.159090909090909e-05 254 -1753101231597 3.143939393939394e-05 256 -1753101248995 3.128787878787879e-05 258 -1753101266907 3.113636363636364e-05 260 -1753101284453 3.098484848484849e-05 262 -1753101300751 3.0833333333333335e-05 264 -1753101319044 3.068181818181818e-05 266 -1753101333739 3.0530303030303034e-05 268 -1753101351938 3.037878787878788e-05 270 -1753101370058 3.0227272727272725e-05 272 -1753101387398 3.0075757575757578e-05 274 -1753101404642 2.9924242424242427e-05 276 -1753101422160 2.9772727272727273e-05 278 -1753101439833 2.9621212121212122e-05 280 -1753101457342 2.9469696969696968e-05 282 -1753101473364 2.9318181818181817e-05 284 -1753101489441 2.916666666666667e-05 286 -1753101506484 2.901515151515152e-05 288 -1753101522875 2.8863636363636365e-05 290 -1753101539258 2.8712121212121214e-05 292 -1753101551955 2.856060606060606e-05 294 -1753101567021 2.8409090909090912e-05 296 -1753101579449 2.825757575757576e-05 298 -1753101591811 2.8106060606060607e-05 300 -1753101605116 2.7954545454545457e-05 302 -1753101617605 2.7803030303030303e-05 304 -1753101630107 2.7651515151515152e-05 306 -1753101642603 2.7500000000000004e-05 308 -1753101654987 2.734848484848485e-05 310 -1753101667355 2.71969696969697e-05 312 -1753101679931 2.7045454545454545e-05 314 -1753101692230 2.6893939393939394e-05 316 -1753101704746 2.674242424242424e-05 318 -1753101717711 2.6590909090909093e-05 320 -1753101731491 2.6439393939393942e-05 322 -1753101744045 2.6287878787878788e-05 324 -1753101756502 2.6136363636363637e-05 326 -1753101768918 2.5984848484848483e-05 328 -1753101781177 2.5833333333333336e-05 330 -1753101794186 2.5681818181818185e-05 332 -1753101806829 2.553030303030303e-05 334 -1753101817841 2.537878787878788e-05 336 -1753101830273 2.5227272727272726e-05 338 -1753101842727 2.5075757575757575e-05 340 -1753101856035 2.4924242424242424e-05 342 -1753101868461 2.4772727272727277e-05 344 -1753101881035 2.4621212121212123e-05 346 -1753101893425 2.4469696969696972e-05 348 -1753101905823 2.431818181818182e-05 350 -1753101918118 2.4166666666666667e-05 352 -1753101930581 2.4015151515151516e-05 354 -1753101942951 2.3863636363636365e-05 356 -1753101955464 2.3712121212121214e-05 358 -1753101967892 2.356060606060606e-05 360 -1753101981213 2.340909090909091e-05 362 -1753101993491 2.325757575757576e-05 364 -1753102005939 2.3106060606060605e-05 366 -1753102018511 2.2954545454545457e-05 368 -1753102030741 2.2803030303030303e-05 370 -1753102043029 2.2651515151515152e-05 372 -1753102055570 2.25e-05 374 -1753102068468 2.234848484848485e-05 376 -1753102081353 2.21969696969697e-05 378 -1753102093722 2.2045454545454546e-05 380 -1753102107017 2.1893939393939395e-05 382 -1753102119465 2.1742424242424244e-05 384 -1753102132053 2.1590909090909093e-05 386 -1753102144419 2.143939393939394e-05 388 -1753102156818 2.128787878787879e-05 390 -1753102169259 2.1136363636363638e-05 392 -1753102181603 2.0984848484848483e-05 394 -1753102194057 2.0833333333333336e-05 396 -1753102206442 2.0681818181818182e-05 398 -1753102218668 2.053030303030303e-05 400 -1753102230435 2.037878787878788e-05 402 -1753102242794 2.022727272727273e-05 404 -1753102255032 2.0075757575757575e-05 406 -1753102267198 1.9924242424242425e-05 408 -1753102279979 1.9772727272727274e-05 410 -1753102292549 1.962121212121212e-05 412 -1753102305253 1.9469696969696972e-05 414 -1753102322016 1.9318181818181818e-05 416 -1753102339290 1.9166666666666667e-05 418 -1753102356703 1.9015151515151516e-05 420 -1753102378067 1.8863636363636362e-05 422 -1753102396288 1.8712121212121215e-05 424 -1753102417476 1.856060606060606e-05 426 -1753102435595 1.840909090909091e-05 428 -1753102452231 1.825757575757576e-05 430 -1753102469008 1.810606060606061e-05 432 -1753102486414 1.7954545454545454e-05 434 -1753102504814 1.7803030303030303e-05 436 -1753102522833 1.7651515151515153e-05 438 -1753102541953 1.75e-05 440 -1753102562836 1.734848484848485e-05 442 -1753102579387 1.7196969696969697e-05 444 -1753102596188 1.7045454545454546e-05 446 -1753102613449 1.6893939393939395e-05 448 -1753102625983 1.674242424242424e-05 450 -1753102638547 1.6590909090909094e-05 452 -1753102651016 1.643939393939394e-05 454 -1753102663839 1.628787878787879e-05 456 -1753102676468 1.6136363636363638e-05 458 -1753102689469 1.5984848484848487e-05 460 -1753102703211 1.5833333333333333e-05 462 -1753102715962 1.5681818181818182e-05 464 -1753102728637 1.553030303030303e-05 466 -1753102740978 1.5378787878787877e-05 468 -1753102751737 1.5227272727272728e-05 470 -1753102763889 1.5075757575757576e-05 472 -1753102776221 1.4924242424242423e-05 474 -1753102788524 1.4772727272727274e-05 476 -1753102801040 1.4621212121212122e-05 478 -1753102813226 1.446969696969697e-05 480 -1753102826667 1.431818181818182e-05 482 -1753102838780 1.4166666666666668e-05 484 -1753102850896 1.4015151515151515e-05 486 -1753102863487 1.3863636363636364e-05 488 -1753102876122 1.3712121212121212e-05 490 -1753102888516 1.3560606060606063e-05 492 -1753102900844 1.340909090909091e-05 494 -1753102913114 1.3257575757575758e-05 496 -1753102925750 1.3106060606060607e-05 498 -1753102938142 1.2954545454545455e-05 500 -1753102951176 1.2803030303030302e-05 502 -1753102963259 1.2651515151515153e-05 504 -1753102975580 1.25e-05 506 -1753102987954 1.234848484848485e-05 508 -1753103000438 1.2196969696969697e-05 510 -1753103012937 1.2045454545454547e-05 512 -1753103025293 1.1893939393939394e-05 514 -1753103037822 1.1742424242424243e-05 516 -1753103050099 1.159090909090909e-05 518 -1753103062194 1.143939393939394e-05 520 -1753103075646 1.128787878787879e-05 522 -1753103087907 1.1136363636363637e-05 524 -1753103100239 1.0984848484848486e-05 526 -1753103112507 1.0833333333333334e-05 528 -1753103124952 1.0681818181818181e-05 530 -1753103137352 1.053030303030303e-05 532 -1753103149521 1.037878787878788e-05 534 -1753103160281 1.0227272727272729e-05 536 -1753103172737 1.0075757575757576e-05 538 -1753103185085 9.924242424242425e-06 540 -1753103198775 9.772727272727273e-06 542 -1753103211421 9.62121212121212e-06 544 -1753103223847 9.46969696969697e-06 546 -1753103236348 9.318181818181819e-06 548 -1753103248804 9.166666666666666e-06 550 -1753103261193 9.015151515151516e-06 552 -1753103273576 8.863636363636365e-06 554 -1753103285828 8.712121212121212e-06 556 -1753103298280 8.56060606060606e-06 558 -1753103310727 8.409090909090909e-06 560 -1753103324545 8.257575757575758e-06 562 -1753103336895 8.106060606060606e-06 564 -1753103349333 7.954545454545455e-06 566 -1753103361816 7.803030303030304e-06 568 -1753103374215 7.651515151515152e-06 570 -1753103386774 7.5e-06 572 -1753103399224 7.3484848484848486e-06 574 -1753103411594 7.196969696969698e-06 576 -1753103423999 7.045454545454545e-06 578 -1753103436440 6.8939393939393945e-06 580 -1753103449807 6.742424242424243e-06 582 -1753103462230 6.59090909090909e-06 584 -1753103474665 6.43939393939394e-06 586 -1753103487149 6.287878787878789e-06 588 -1753103499704 6.136363636363636e-06 590 -1753103512138 5.984848484848485e-06 592 -1753103524731 5.833333333333334e-06 594 -1753103537109 5.681818181818182e-06 596 -1753103549566 5.530303030303031e-06 598 -1753103561938 5.378787878787879e-06 600 -1753103575590 5.2272727272727274e-06 602 -1753103586472 5.075757575757576e-06 604 -1753103598901 4.924242424242424e-06 606 -1753103611487 4.772727272727273e-06 608 -1753103623784 4.621212121212122e-06 610 -1753103636042 4.46969696969697e-06 612 -1753103648314 4.3181818181818185e-06 614 -1753103660702 4.166666666666667e-06 616 -1753103673071 4.015151515151515e-06 618 -1753103685477 3.863636363636364e-06 620 -1753103698696 3.7121212121212124e-06 622 -1753103710924 3.5606060606060608e-06 624 -1753103723356 3.409090909090909e-06 626 -1753103736037 3.257575757575758e-06 628 -1753103748360 3.106060606060606e-06 630 -1753103760610 2.9545454545454547e-06 632 -1753103772762 2.803030303030303e-06 634 -1753103785207 2.651515151515152e-06 636 -1753103797805 2.5e-06 638 -1753103810295 2.3484848484848486e-06 640 -1753103823660 2.196969696969697e-06 642 -1753103836107 2.0454545454545457e-06 644 -1753103848492 1.8939393939393941e-06 646 -1753103861064 1.7424242424242427e-06 648 -1753103873523 1.5909090909090908e-06 650 -1753103885869 1.4393939393939396e-06 652 -1753103897958 1.287878787878788e-06 654 -1753103910220 1.1363636363636364e-06 656 -1753103922705 9.848484848484847e-07 658 -1753103935001 8.333333333333333e-07 660 -1753103948168 6.818181818181818e-07 662 -1753103960368 5.303030303030304e-07 664 -1753103972703 3.787878787878788e-07 666 -1753103985078 2.2727272727272726e-07 668 -1753103995846 7.575757575757576e-08 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/loss b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/loss deleted file mode 100644 index 07973ebf67238ed9e0863236249869b112a7a38e..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/loss +++ /dev/null @@ -1,335 +0,0 @@ -1753099099197 1.6973 2 -1753099117338 1.6822 4 -1753099133987 1.4687 6 -1753099150841 1.4093 8 -1753099167353 1.1657 10 -1753099183522 0.9447 12 -1753099199672 0.8334 14 -1753099216002 0.6336 16 -1753099232389 0.4959 18 -1753099248975 0.3829 20 -1753099267528 0.3361 22 -1753099287703 0.3142 24 -1753099306825 0.2581 26 -1753099326820 0.247 28 -1753099345585 0.2266 30 -1753099363772 0.1985 32 -1753099380259 0.1634 34 -1753099396729 0.1626 36 -1753099413982 0.1796 38 -1753099431269 0.1384 40 -1753099447383 0.1498 42 -1753099464471 0.0965 44 -1753099481040 0.0873 46 -1753099497069 0.0909 48 -1753099515752 0.0719 50 -1753099534493 0.0553 52 -1753099551081 0.0744 54 -1753099570403 0.0608 56 -1753099586817 0.0364 58 -1753099602727 0.0495 60 -1753099620077 0.0471 62 -1753099638583 0.0299 64 -1753099654694 0.0309 66 -1753099668722 0.0487 68 -1753099684879 0.0304 70 -1753099700891 0.0254 72 -1753099716854 0.0363 74 -1753099733096 0.0386 76 -1753099749599 0.0201 78 -1753099765909 0.0187 80 -1753099783263 0.0175 82 -1753099799395 0.0303 84 -1753099815643 0.0268 86 -1753099831792 0.0224 88 -1753099847949 0.0202 90 -1753099863985 0.0216 92 -1753099880315 0.0168 94 -1753099896452 0.0207 96 -1753099912502 0.0145 98 -1753099928706 0.025 100 -1753099946155 0.0183 102 -1753099962388 0.022 104 -1753099979182 0.0149 106 -1753099997221 0.0152 108 -1753100015117 0.0196 110 -1753100031324 0.0165 112 -1753100047565 0.0168 114 -1753100064664 0.0146 116 -1753100080943 0.0195 118 -1753100097331 0.0187 120 -1753100114685 0.0171 122 -1753100131933 0.0204 124 -1753100148753 0.0203 126 -1753100165684 0.0179 128 -1753100181802 0.0198 130 -1753100197990 0.0182 132 -1753100212648 0.0186 134 -1753100228619 0.0166 136 -1753100244985 0.0162 138 -1753100263859 0.0156 140 -1753100281198 0.0211 142 -1753100297284 0.0184 144 -1753100313650 0.0165 146 -1753100329826 0.0184 148 -1753100346032 0.0168 150 -1753100365008 0.016 152 -1753100381122 0.0151 154 -1753100397392 0.0167 156 -1753100413896 0.0162 158 -1753100430772 0.0166 160 -1753100448101 0.015 162 -1753100464233 0.0171 164 -1753100480637 0.0189 166 -1753100496812 0.0164 168 -1753100512763 0.0172 170 -1753100529101 0.0147 172 -1753100545662 0.0146 174 -1753100561842 0.0158 176 -1753100577915 0.0173 178 -1753100597827 0.0162 180 -1753100617625 0.0163 182 -1753100633805 0.014 184 -1753100649953 0.0141 186 -1753100666118 0.0199 188 -1753100682558 0.0142 190 -1753100698868 0.0166 192 -1753100714949 0.0136 194 -1753100731373 0.0146 196 -1753100747827 0.0167 198 -1753100764846 0.0195 200 -1753100781020 0.0126 202 -1753100797597 0.0143 204 -1753100813317 0.0172 206 -1753100830432 0.0152 208 -1753100850133 0.016 210 -1753100866690 0.0147 212 -1753100882925 0.0159 214 -1753100899245 0.0136 216 -1753100915684 0.0172 218 -1753100932178 0.0149 220 -1753100949229 0.0153 222 -1753100966051 0.0149 224 -1753100982253 0.0158 226 -1753100999028 0.0169 228 -1753101015198 0.0167 230 -1753101031769 0.0144 232 -1753101048085 0.0138 234 -1753101064616 0.0164 236 -1753101080911 0.0149 238 -1753101097275 0.0137 240 -1753101114344 0.0181 242 -1753101130342 0.0158 244 -1753101146579 0.0158 246 -1753101162626 0.0149 248 -1753101181901 0.0151 250 -1753101198100 0.0155 252 -1753101214528 0.0153 254 -1753101231597 0.0131 256 -1753101248995 0.0191 258 -1753101266907 0.0166 260 -1753101284453 0.0146 262 -1753101300751 0.0152 264 -1753101319044 0.0131 266 -1753101333739 0.0183 268 -1753101351938 0.0157 270 -1753101370058 0.0143 272 -1753101387398 0.014 274 -1753101404642 0.016 276 -1753101422160 0.0158 278 -1753101439833 0.0168 280 -1753101457342 0.0152 282 -1753101473364 0.0164 284 -1753101489441 0.0174 286 -1753101506484 0.0152 288 -1753101522875 0.0159 290 -1753101539258 0.0164 292 -1753101551955 0.0126 294 -1753101567021 0.0167 296 -1753101579449 0.0161 298 -1753101591811 0.0172 300 -1753101605116 0.0158 302 -1753101617605 0.0151 304 -1753101630107 0.0166 306 -1753101642603 0.0157 308 -1753101654987 0.0145 310 -1753101667355 0.0165 312 -1753101679931 0.016 314 -1753101692230 0.0159 316 -1753101704746 0.0141 318 -1753101717711 0.014 320 -1753101731491 0.0128 322 -1753101744045 0.0115 324 -1753101756502 0.0154 326 -1753101768918 0.0155 328 -1753101781177 0.0186 330 -1753101794186 0.0151 332 -1753101806829 0.0129 334 -1753101817841 0.0134 336 -1753101830273 0.0138 338 -1753101842727 0.013 340 -1753101856035 0.0151 342 -1753101868461 0.0149 344 -1753101881035 0.0145 346 -1753101893425 0.0175 348 -1753101905823 0.0144 350 -1753101918118 0.0168 352 -1753101930581 0.0133 354 -1753101942951 0.0135 356 -1753101955464 0.0147 358 -1753101967892 0.0155 360 -1753101981213 0.0138 362 -1753101993491 0.0179 364 -1753102005939 0.0144 366 -1753102018511 0.0132 368 -1753102030741 0.0179 370 -1753102043029 0.0155 372 -1753102055570 0.0142 374 -1753102068468 0.0179 376 -1753102081353 0.0142 378 -1753102093722 0.0158 380 -1753102107017 0.0153 382 -1753102119465 0.0132 384 -1753102132053 0.0132 386 -1753102144419 0.0188 388 -1753102156818 0.0155 390 -1753102169259 0.0154 392 -1753102181603 0.0144 394 -1753102194057 0.0138 396 -1753102206442 0.0159 398 -1753102218668 0.016 400 -1753102230435 0.0137 402 -1753102242794 0.0146 404 -1753102255032 0.016 406 -1753102267198 0.017 408 -1753102279979 0.0161 410 -1753102292549 0.0143 412 -1753102305253 0.0148 414 -1753102322016 0.0153 416 -1753102339290 0.0153 418 -1753102356703 0.0131 420 -1753102378067 0.0132 422 -1753102396288 0.0137 424 -1753102417476 0.0155 426 -1753102435595 0.0181 428 -1753102452231 0.0143 430 -1753102469008 0.0147 432 -1753102486414 0.0126 434 -1753102504814 0.0162 436 -1753102522833 0.013 438 -1753102541953 0.0145 440 -1753102562836 0.0135 442 -1753102579387 0.0164 444 -1753102596188 0.0152 446 -1753102613449 0.0153 448 -1753102625983 0.0134 450 -1753102638547 0.0135 452 -1753102651016 0.0159 454 -1753102663839 0.0141 456 -1753102676468 0.015 458 -1753102689469 0.0141 460 -1753102703211 0.0157 462 -1753102715962 0.0141 464 -1753102728637 0.0156 466 -1753102740978 0.0169 468 -1753102751737 0.0143 470 -1753102763889 0.0158 472 -1753102776221 0.0136 474 -1753102788524 0.0159 476 -1753102801040 0.0155 478 -1753102813226 0.0171 480 -1753102826667 0.0142 482 -1753102838780 0.0133 484 -1753102850896 0.014 486 -1753102863487 0.015 488 -1753102876122 0.0143 490 -1753102888516 0.0144 492 -1753102900844 0.0145 494 -1753102913114 0.0156 496 -1753102925750 0.0141 498 -1753102938142 0.0146 500 -1753102951176 0.0141 502 -1753102963259 0.0128 504 -1753102975580 0.0128 506 -1753102987954 0.0161 508 -1753103000438 0.0149 510 -1753103012937 0.0138 512 -1753103025293 0.0132 514 -1753103037822 0.0131 516 -1753103050099 0.0162 518 -1753103062194 0.015 520 -1753103075646 0.0132 522 -1753103087907 0.0144 524 -1753103100239 0.0157 526 -1753103112507 0.015 528 -1753103124952 0.0125 530 -1753103137352 0.014 532 -1753103149521 0.0175 534 -1753103160281 0.0144 536 -1753103172737 0.0121 538 -1753103185085 0.0157 540 -1753103198775 0.0141 542 -1753103211421 0.0145 544 -1753103223847 0.0146 546 -1753103236348 0.014 548 -1753103248804 0.0139 550 -1753103261193 0.0148 552 -1753103273576 0.0134 554 -1753103285828 0.0151 556 -1753103298280 0.0139 558 -1753103310727 0.0129 560 -1753103324545 0.0119 562 -1753103336895 0.0175 564 -1753103349333 0.0153 566 -1753103361816 0.0143 568 -1753103374215 0.015 570 -1753103386774 0.0154 572 -1753103399224 0.0141 574 -1753103411594 0.0151 576 -1753103423999 0.0132 578 -1753103436440 0.0143 580 -1753103449807 0.0153 582 -1753103462230 0.0133 584 -1753103474665 0.0146 586 -1753103487149 0.0152 588 -1753103499704 0.0143 590 -1753103512138 0.0143 592 -1753103524731 0.0142 594 -1753103537109 0.0148 596 -1753103549566 0.0156 598 -1753103561938 0.0146 600 -1753103575590 0.0132 602 -1753103586472 0.0142 604 -1753103598901 0.015 606 -1753103611487 0.0139 608 -1753103623784 0.0157 610 -1753103636042 0.0152 612 -1753103648314 0.0143 614 -1753103660702 0.0128 616 -1753103673071 0.0156 618 -1753103685477 0.0135 620 -1753103698696 0.0151 622 -1753103710924 0.015 624 -1753103723356 0.0141 626 -1753103736037 0.0136 628 -1753103748360 0.0152 630 -1753103760610 0.0137 632 -1753103772762 0.0145 634 -1753103785207 0.0137 636 -1753103797805 0.0122 638 -1753103810295 0.0137 640 -1753103823660 0.0147 642 -1753103836107 0.0139 644 -1753103848492 0.0138 646 -1753103861064 0.0127 648 -1753103873523 0.013 650 -1753103885869 0.014 652 -1753103897958 0.0153 654 -1753103910220 0.0145 656 -1753103922705 0.0147 658 -1753103935001 0.0157 660 -1753103948168 0.0154 662 -1753103960368 0.0133 664 -1753103972703 0.0142 666 -1753103985078 0.0144 668 -1753103995846 0.0139 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/total_flos b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/total_flos deleted file mode 100644 index f7ed21833d2fab5e69149acdecf1869e92838bba..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/total_flos +++ /dev/null @@ -1 +0,0 @@ -1753103996876 1.804222263410688e+16 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_loss b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_loss deleted file mode 100644 index 7ac374d4695fce863e50141d4ef4f8d5b0c241c1..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_loss +++ /dev/null @@ -1 +0,0 @@ -1753103996876 0.05549748351436053 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_runtime b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_runtime deleted file mode 100644 index 4fb68c5cdcd4e75bc7b393486cac924a5c7c1753..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_runtime +++ /dev/null @@ -1 +0,0 @@ -1753103996876 4915.5746 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_samples_per_second b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_samples_per_second deleted file mode 100644 index 27f6d7b8e81b402f5dc9775fecb2a148d9e32637..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_samples_per_second +++ /dev/null @@ -1 +0,0 @@ -1753103996876 1.086 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_steps_per_second b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_steps_per_second deleted file mode 100644 index c284b86626510f772172e31a34b07d5b7fe0d5f5..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/metrics/train_steps_per_second +++ /dev/null @@ -1 +0,0 @@ -1753103996876 0.136 670 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/_name_or_path b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/_name_or_path deleted file mode 100644 index 3a8ba38a16a767019a426ed9ae2b3cd2746d622d..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM3-3B \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/accelerator_config b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adafactor b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_beta1 b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_beta2 b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_epsilon b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/add_cross_attention b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/architectures b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/architectures deleted file mode 100644 index 55febbc88b5c08e00d44e95e7ed72640c6d034b6..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['SmolLM3ForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/attention_bias b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/attention_dropout b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/auto_find_batch_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/average_tokens_across_devices b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bad_words_ids b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/batch_eval_metrics b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/begin_suppress_tokens b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bf16 b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bf16_full_eval b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bos_token_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bos_token_id deleted file mode 100644 index 5499007cbac38bc897e3c2766b82a647ac28c735..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -128000 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/chunk_size_feed_forward b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/cross_attention_hidden_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/data_seed b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_drop_last b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_num_workers b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_persistent_workers b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_pin_memory b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_prefetch_factor b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_backend b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_broadcast_buffers b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_bucket_cap_mb b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_find_unused_parameters b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_timeout b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/debug b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/decoder_start_token_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/deepspeed b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/disable_tqdm b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/diversity_penalty b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_eval b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_predict b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_sample b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_train b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/early_stopping b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/encoder_no_repeat_ngram_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eos_token_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eos_token_id deleted file mode 100644 index d78b6326c2a8961e1672862523846efebe7ffdd6..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -128012 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_accumulation_steps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_delay b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_do_concat_batches b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_on_start b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_steps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_strategy b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_use_gather_object b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/exponential_decay_length_penalty b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/finetuning_task b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/forced_bos_token_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/forced_eos_token_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16 b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_backend b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_full_eval b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_opt_level b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_config b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_min_num_params b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/full_determinism b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_accumulation_steps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_accumulation_steps deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_checkpointing b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_checkpointing deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_checkpointing_kwargs b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/greater_is_better b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/group_by_length b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/half_precision_backend b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hidden_act b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hidden_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_always_push b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_model_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_private_repo b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_revision b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_strategy b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_token b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/id2label b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ignore_data_skip b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_for_metrics b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_inputs_for_metrics b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_num_input_tokens_seen b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_tokens_per_second b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/initializer_range b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/intermediate_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/intermediate_size deleted file mode 100644 index 9b7babb1160a44a2cf9ae0dbfd43efd0fd70838b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -11008 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/is_decoder b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/is_encoder_decoder b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/jit_mode_eval b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label2id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label_names b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label_smoothing_factor b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/layer_types b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/layer_types deleted file mode 100644 index bc44845b7022ebf6e7ac1a8dac64d5ba0daa15b1..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/layer_types +++ /dev/null @@ -1 +0,0 @@ -['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'] \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/learning_rate b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/learning_rate deleted file mode 100644 index 335f2feb826f13cdd89c6b5234de9a8c96396bb7..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -5e-05 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/length_column_name b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/length_penalty b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/liger_kernel_config b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/load_best_model_at_end b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/local_rank b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_level b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_level_replica b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_on_each_node b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_dir b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_dir deleted file mode 100644 index a1d6221059c99dfc4fb740aedff7d2695df7e244..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust/runs/Jul21_12-58-00_Jasons-MacBook-Pro.local \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_first_step b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_nan_inf_filter b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_steps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_steps deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_strategy b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/lr_scheduler_kwargs b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/lr_scheduler_type b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/lr_scheduler_type deleted file mode 100644 index 9c01abbc1e45086237d1a59d6ec6bdea2400f1a7..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -linear \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_grad_norm b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_length b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_position_embeddings b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_position_embeddings deleted file mode 100644 index 83c406c4fd41a6ec8a25290b17cd3894ca57ac4b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -65536 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_steps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_window_layers b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_window_layers deleted file mode 100644 index 368f89ceef179cc546403ac0d5ef1d0e4b340447..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/max_window_layers +++ /dev/null @@ -1 +0,0 @@ -28 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/metric_for_best_model b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/min_length b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/mlp_bias b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/model_type b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/model_type deleted file mode 100644 index 34bda861a9301e0dab58899ec4828a0ee9f464b1..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/model_type +++ /dev/null @@ -1 +0,0 @@ -smollm3 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/mp_parameters b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/neftune_noise_alpha b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_cuda b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_repeat_ngram_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_rope_layer_interval b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_rope_layer_interval deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_rope_layer_interval +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_rope_layers b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_rope_layers deleted file mode 100644 index 2cdfac1b0d51ce92d6d6469bd544bc76a3521cec..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/no_rope_layers +++ /dev/null @@ -1 +0,0 @@ -[1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0] \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_attention_heads b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_attention_heads deleted file mode 100644 index 19c7bdba7b1e9bfe80365a50420a6d538ca503c3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -16 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_beam_groups b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_beams b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_hidden_layers b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_hidden_layers deleted file mode 100644 index dce6588ca1420a69eea5699f2ec5d666daf16a6a..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -36 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_key_value_heads b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_key_value_heads deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_return_sequences b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_train_epochs b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_train_epochs deleted file mode 100644 index 9a037142aa3c1b4c490e1a38251620f113465330..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -10 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim_args b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim_target_modules b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_attentions b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_dir b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_dir deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_hidden_states b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_scores b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/overwrite_output_dir b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pad_token_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pad_token_id deleted file mode 100644 index 8d9491efc0fb90c90645db18bae256c4f9daff70..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -128004 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/past_index b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_device_eval_batch_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_device_train_batch_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_device_train_batch_size deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_gpu_eval_batch_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_gpu_train_batch_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/prediction_loss_only b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/prefix b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pretraining_tp b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/problem_type b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pruned_heads b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_model_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_organization b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_token b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ray_scope b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/remove_invalid_values b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/remove_unused_columns b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/remove_unused_columns deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/repetition_penalty b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/report_to b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/restore_callback_states_from_checkpoint b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/resume_from_checkpoint b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/return_dict b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/return_dict_in_generate b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rms_norm_eps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rms_norm_eps deleted file mode 100644 index d6bc26a15d7729bbaf8f0e522bea2d0a8781886d..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-06 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rope_scaling b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rope_theta b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rope_theta deleted file mode 100644 index 445ba6793f521c521c5da54f26f04a56ecc83ee2..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -5000000.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/run_name b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/run_name deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_on_each_node b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_only_model b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_safetensors b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_steps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_steps deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_strategy b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_total_limit b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_total_limit deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/seed b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/sep_token_id b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/skip_memory_metrics b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/sliding_window b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/sliding_window deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/sliding_window +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/suppress_tokens b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/task_specific_params b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/temperature b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tf32 b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tf_legacy_loss b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tie_encoder_decoder b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tie_word_embeddings b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tokenizer_class b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/top_k b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/top_p b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile_backend b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile_mode b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_dtype b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_empty_cache_steps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torchdynamo b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torchscript b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tpu_metrics_debug b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tpu_num_cores b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/transformers_version b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/typical_p b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_bfloat16 b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_cache b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_cache deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_cpu b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_ipex b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_legacy_prediction_loop b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_liger_kernel b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_mps_device b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_sliding_window b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_sliding_window deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/use_sliding_window +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/vocab_size b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/vocab_size deleted file mode 100644 index 34c44b19378193a3b4fa853df4426fc426c15535..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -128256 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/warmup_ratio b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/warmup_steps b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/warmup_steps deleted file mode 100644 index 9a037142aa3c1b4c490e1a38251620f113465330..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -10 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/weight_decay b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/weight_decay deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.runName b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.runName deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.git.commit b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.git.commit deleted file mode 100644 index 64393076c1626bfaa09d72c60ec8c9a9769145c8..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.git.commit +++ /dev/null @@ -1 +0,0 @@ -e2610e8c39108b089782f99d0af9c3a8939adc61 \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.name b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.name deleted file mode 100644 index 124d840089086dc6280e8eb4e84f59ad6b0b015f..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_simple_robust.py \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.type b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.user b/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/a9ce4acd189c4737ba94346689f2416b/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/meta.yaml b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/meta.yaml deleted file mode 100644 index 7dfec86cecc01a25f2dc10d1eb17ed7c1eafb4b6..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/c9ad7410dfa04abcbffef241e323303d/artifacts -end_time: 1753096025423 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: c9ad7410dfa04abcbffef241e323303d -run_name: ./smollm3_tool_adapter -run_uuid: c9ad7410dfa04abcbffef241e323303d -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753096024408 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/_name_or_path b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/_name_or_path deleted file mode 100644 index 3a8ba38a16a767019a426ed9ae2b3cd2746d622d..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM3-3B \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/accelerator_config b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adafactor b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_beta1 b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_beta2 b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_epsilon b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/add_cross_attention b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/architectures b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/architectures deleted file mode 100644 index 55febbc88b5c08e00d44e95e7ed72640c6d034b6..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['SmolLM3ForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/attention_bias b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/attention_dropout b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/auto_find_batch_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/average_tokens_across_devices b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bad_words_ids b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/batch_eval_metrics b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/begin_suppress_tokens b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bf16 b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bf16_full_eval b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bos_token_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bos_token_id deleted file mode 100644 index 5499007cbac38bc897e3c2766b82a647ac28c735..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -128000 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/chunk_size_feed_forward b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/cross_attention_hidden_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/data_seed b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_drop_last b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_num_workers b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_persistent_workers b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_pin_memory b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_prefetch_factor b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_backend b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_broadcast_buffers b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_bucket_cap_mb b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_find_unused_parameters b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_timeout b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/debug b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/decoder_start_token_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/deepspeed b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/disable_tqdm b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/diversity_penalty b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_eval b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_predict b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_sample b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_train b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/early_stopping b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/encoder_no_repeat_ngram_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eos_token_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eos_token_id deleted file mode 100644 index d78b6326c2a8961e1672862523846efebe7ffdd6..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -128012 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_accumulation_steps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_delay b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_do_concat_batches b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_on_start b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_steps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_strategy b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_use_gather_object b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/exponential_decay_length_penalty b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/finetuning_task b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/forced_bos_token_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/forced_eos_token_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16 b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_backend b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_full_eval b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_opt_level b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_config b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_min_num_params b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/full_determinism b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_accumulation_steps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_accumulation_steps deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_checkpointing b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_checkpointing deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_checkpointing_kwargs b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/greater_is_better b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/group_by_length b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/half_precision_backend b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hidden_act b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hidden_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_always_push b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_model_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_private_repo b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_revision b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_strategy b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_token b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/id2label b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ignore_data_skip b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_for_metrics b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_inputs_for_metrics b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_num_input_tokens_seen b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_tokens_per_second b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/initializer_range b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/intermediate_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/intermediate_size deleted file mode 100644 index 9b7babb1160a44a2cf9ae0dbfd43efd0fd70838b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -11008 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/is_decoder b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/is_encoder_decoder b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/jit_mode_eval b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label2id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label_names b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label_smoothing_factor b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/layer_types b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/layer_types deleted file mode 100644 index bc44845b7022ebf6e7ac1a8dac64d5ba0daa15b1..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/layer_types +++ /dev/null @@ -1 +0,0 @@ -['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'] \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/learning_rate b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/learning_rate deleted file mode 100644 index 83f55466e252f8ed340b201b70d8470f0cd54c4b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -0.0003 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/length_column_name b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/length_penalty b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/liger_kernel_config b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/load_best_model_at_end b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/local_rank b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_level b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_level_replica b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_on_each_node b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_dir b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_dir deleted file mode 100644 index 6bc8522feb83b127cc7e1e76a426e8d1a973d69c..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./logs \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_first_step b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_nan_inf_filter b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_steps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_steps deleted file mode 100644 index 7813681f5b41c028345ca62a2be376bae70b7f61..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -5 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_strategy b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/lr_scheduler_kwargs b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/lr_scheduler_type b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/lr_scheduler_type deleted file mode 100644 index 84aa3999b5b7cae7f78b1f77e04d182643005a92..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -cosine \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_grad_norm b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_length b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_position_embeddings b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_position_embeddings deleted file mode 100644 index 83c406c4fd41a6ec8a25290b17cd3894ca57ac4b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -65536 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_steps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_window_layers b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_window_layers deleted file mode 100644 index 368f89ceef179cc546403ac0d5ef1d0e4b340447..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/max_window_layers +++ /dev/null @@ -1 +0,0 @@ -28 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/metric_for_best_model b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/min_length b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/mlp_bias b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/model_type b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/model_type deleted file mode 100644 index 34bda861a9301e0dab58899ec4828a0ee9f464b1..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/model_type +++ /dev/null @@ -1 +0,0 @@ -smollm3 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/mp_parameters b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/neftune_noise_alpha b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_cuda b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_repeat_ngram_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_rope_layer_interval b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_rope_layer_interval deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_rope_layer_interval +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_rope_layers b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_rope_layers deleted file mode 100644 index 2cdfac1b0d51ce92d6d6469bd544bc76a3521cec..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/no_rope_layers +++ /dev/null @@ -1 +0,0 @@ -[1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0] \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_attention_heads b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_attention_heads deleted file mode 100644 index 19c7bdba7b1e9bfe80365a50420a6d538ca503c3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -16 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_beam_groups b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_beams b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_hidden_layers b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_hidden_layers deleted file mode 100644 index dce6588ca1420a69eea5699f2ec5d666daf16a6a..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -36 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_key_value_heads b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_key_value_heads deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_return_sequences b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_train_epochs b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_train_epochs deleted file mode 100644 index 7813681f5b41c028345ca62a2be376bae70b7f61..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -5 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim_args b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim_target_modules b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_attentions b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_dir b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_dir deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_hidden_states b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_scores b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/overwrite_output_dir b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pad_token_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pad_token_id deleted file mode 100644 index 8d9491efc0fb90c90645db18bae256c4f9daff70..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -128004 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/past_index b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_device_eval_batch_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_device_train_batch_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_device_train_batch_size deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_gpu_eval_batch_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_gpu_train_batch_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/prediction_loss_only b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/prefix b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pretraining_tp b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/problem_type b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pruned_heads b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_model_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_organization b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_token b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ray_scope b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/remove_invalid_values b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/remove_unused_columns b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/remove_unused_columns deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/repetition_penalty b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/report_to b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/restore_callback_states_from_checkpoint b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/resume_from_checkpoint b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/return_dict b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/return_dict_in_generate b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rms_norm_eps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rms_norm_eps deleted file mode 100644 index d6bc26a15d7729bbaf8f0e522bea2d0a8781886d..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-06 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rope_scaling b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rope_theta b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rope_theta deleted file mode 100644 index 445ba6793f521c521c5da54f26f04a56ecc83ee2..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -5000000.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/run_name b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/run_name deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_on_each_node b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_only_model b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_safetensors b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_steps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_steps deleted file mode 100644 index 410b14d2ce6f958c13adcb30807e8673074c49d9..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -25 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_strategy b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_total_limit b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_total_limit deleted file mode 100644 index e440e5c842586965a7fb77deda2eca68612b1f53..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -3 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/seed b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/sep_token_id b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/skip_memory_metrics b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/sliding_window b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/sliding_window deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/sliding_window +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/suppress_tokens b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/task_specific_params b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/temperature b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tf32 b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tf_legacy_loss b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tie_encoder_decoder b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tie_word_embeddings b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tokenizer_class b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/top_k b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/top_p b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile_backend b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile_mode b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_dtype b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_empty_cache_steps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torchdynamo b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torchscript b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tpu_metrics_debug b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tpu_num_cores b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/transformers_version b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/typical_p b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_bfloat16 b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_cache b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_cache deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_cpu b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_ipex b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_legacy_prediction_loop b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_liger_kernel b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_mps_device b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_sliding_window b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_sliding_window deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/use_sliding_window +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/vocab_size b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/vocab_size deleted file mode 100644 index 34c44b19378193a3b4fa853df4426fc426c15535..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -128256 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/warmup_ratio b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/warmup_steps b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/warmup_steps deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/weight_decay b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/weight_decay deleted file mode 100644 index d1c6331b3109accd73f01907062e6c174e28200a..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.01 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.runName b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.runName deleted file mode 100644 index 636eb8e925f90a0881c57147993c775b4aee209d..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm3_tool_adapter \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.git.commit b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.git.commit deleted file mode 100644 index 64393076c1626bfaa09d72c60ec8c9a9769145c8..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.git.commit +++ /dev/null @@ -1 +0,0 @@ -e2610e8c39108b089782f99d0af9c3a8939adc61 \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.name b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.name deleted file mode 100644 index 0a36aad02c6a55095bdc0d02c86ab40ec03251d6..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_m4_max.py \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.type b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.user b/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/c9ad7410dfa04abcbffef241e323303d/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/meta.yaml b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/meta.yaml deleted file mode 100644 index 64deb241a49cf391db1265a2629521c067cf668b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/meta.yaml +++ /dev/null @@ -1,15 +0,0 @@ -artifact_uri: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/artifacts -end_time: 1753096288939 -entry_point_name: '' -experiment_id: '0' -lifecycle_stage: active -run_id: cb7ad079da3f477cae56d62ec8ee80a9 -run_name: ./smollm3_robust -run_uuid: cb7ad079da3f477cae56d62ec8ee80a9 -source_name: '' -source_type: 4 -source_version: '' -start_time: 1753096125827 -status: 3 -tags: [] -user_id: jasonlovell diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/epoch b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/epoch deleted file mode 100644 index 40c39f921323f15630bac21d79c7021662e26f34..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/epoch +++ /dev/null @@ -1,10 +0,0 @@ -1753096146747 0.37209302325581395 2 -1753096165647 0.7441860465116279 4 -1753096178946 1.0 6 -1753096200635 1.372093023255814 8 -1753096222422 1.744186046511628 10 -1753096235685 2.0 12 -1753096255021 2.3720930232558137 14 -1753096273689 2.744186046511628 16 -1753096287681 3.0 18 -1753096288937 3.0 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/grad_norm b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/grad_norm deleted file mode 100644 index 2f81dbeb5b7b31a50cd168e1282fc67010a48739..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/grad_norm +++ /dev/null @@ -1,9 +0,0 @@ -1753096146747 2.2244515419006348 2 -1753096165647 2.0368847846984863 4 -1753096178946 1.6745638847351074 6 -1753096200635 1.316573977470398 8 -1753096222422 1.3743547201156616 10 -1753096235685 1.2856214046478271 12 -1753096255021 1.2929272651672363 14 -1753096273689 1.1765117645263672 16 -1753096287681 1.26664137840271 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/learning_rate b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/learning_rate deleted file mode 100644 index 70a07ebdaea770633039900e267edc0aa69f237b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/learning_rate +++ /dev/null @@ -1,9 +0,0 @@ -1753096146747 5e-06 2 -1753096165647 1.5e-05 4 -1753096178946 2.5e-05 6 -1753096200635 3.5e-05 8 -1753096222422 4.5e-05 10 -1753096235685 4.375e-05 12 -1753096255021 3.125e-05 14 -1753096273689 1.8750000000000002e-05 16 -1753096287681 6.25e-06 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/loss b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/loss deleted file mode 100644 index 5667fba7b7c9edd4bce05262500e35f257568964..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/loss +++ /dev/null @@ -1,9 +0,0 @@ -1753096146747 1.4107 2 -1753096165647 1.3657 4 -1753096178946 1.2511 6 -1753096200635 1.138 8 -1753096222422 1.0354 10 -1753096235685 0.9303 12 -1753096255021 0.7941 14 -1753096273689 0.702 16 -1753096287681 0.6598 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/total_flos b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/total_flos deleted file mode 100644 index a2849b87c315ab2a2b000a69085f9b54277a7bef..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/total_flos +++ /dev/null @@ -1 +0,0 @@ -1753096288937 661340055969792.0 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_loss b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_loss deleted file mode 100644 index 34481631ebd6c9a77360bcc3815bc15117eea6bb..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_loss +++ /dev/null @@ -1 +0,0 @@ -1753096288937 1.0318962070677016 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_runtime b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_runtime deleted file mode 100644 index 8016cfd7a006afab94fe0a593a8156dc15f1c14a..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_runtime +++ /dev/null @@ -1 +0,0 @@ -1753096288937 163.4242 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_samples_per_second b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_samples_per_second deleted file mode 100644 index de32f71686ceac7fd8ae3696ee765c355ff03e91..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_samples_per_second +++ /dev/null @@ -1 +0,0 @@ -1753096288937 0.789 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_steps_per_second b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_steps_per_second deleted file mode 100644 index 07b9bc92fa3e32a93a022c8ac86c634b7319fc94..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/metrics/train_steps_per_second +++ /dev/null @@ -1 +0,0 @@ -1753096288937 0.11 18 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/_name_or_path b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/_name_or_path deleted file mode 100644 index 3a8ba38a16a767019a426ed9ae2b3cd2746d622d..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/_name_or_path +++ /dev/null @@ -1 +0,0 @@ -HuggingFaceTB/SmolLM3-3B \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/accelerator_config b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/accelerator_config deleted file mode 100644 index b1e7502bd7285c87efc171b09bce733551c8d179..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/accelerator_config +++ /dev/null @@ -1 +0,0 @@ -{'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None} \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adafactor b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adafactor deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adafactor +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_beta1 b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_beta1 deleted file mode 100644 index 9a7d84f2a96bb56f53bfc3a42ac10d06459e55c3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_beta1 +++ /dev/null @@ -1 +0,0 @@ -0.9 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_beta2 b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_beta2 deleted file mode 100644 index 79cbfdf0652c46b13ed8946e54aa94ff7bdd44ab..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_beta2 +++ /dev/null @@ -1 +0,0 @@ -0.999 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_epsilon b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_epsilon deleted file mode 100644 index 851199be9c9a0b8c721d7f305f5af1759637102d..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/adam_epsilon +++ /dev/null @@ -1 +0,0 @@ -1e-08 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/add_cross_attention b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/add_cross_attention deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/add_cross_attention +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/architectures b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/architectures deleted file mode 100644 index 55febbc88b5c08e00d44e95e7ed72640c6d034b6..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/architectures +++ /dev/null @@ -1 +0,0 @@ -['SmolLM3ForCausalLM'] \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/attention_bias b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/attention_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/attention_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/attention_dropout b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/attention_dropout deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/attention_dropout +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/auto_find_batch_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/auto_find_batch_size deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/auto_find_batch_size +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/average_tokens_across_devices b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/average_tokens_across_devices deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/average_tokens_across_devices +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bad_words_ids b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bad_words_ids deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bad_words_ids +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/batch_eval_metrics b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/batch_eval_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/batch_eval_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/begin_suppress_tokens b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/begin_suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/begin_suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bf16 b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bf16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bf16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bf16_full_eval b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bf16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bf16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bos_token_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bos_token_id deleted file mode 100644 index 5499007cbac38bc897e3c2766b82a647ac28c735..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/bos_token_id +++ /dev/null @@ -1 +0,0 @@ -128000 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/chunk_size_feed_forward b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/chunk_size_feed_forward deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/chunk_size_feed_forward +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/cross_attention_hidden_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/cross_attention_hidden_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/cross_attention_hidden_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/data_seed b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/data_seed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/data_seed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_drop_last b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_drop_last deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_drop_last +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_num_workers b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_num_workers deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_num_workers +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_persistent_workers b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_persistent_workers deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_persistent_workers +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_pin_memory b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_pin_memory deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_pin_memory +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_prefetch_factor b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_prefetch_factor deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/dataloader_prefetch_factor +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_backend b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_broadcast_buffers b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_broadcast_buffers deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_broadcast_buffers +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_bucket_cap_mb b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_bucket_cap_mb deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_bucket_cap_mb +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_find_unused_parameters b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_find_unused_parameters deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_find_unused_parameters +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_timeout b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_timeout deleted file mode 100644 index 2974009fd3610ead1e61c1cd20f510a934dd6f91..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ddp_timeout +++ /dev/null @@ -1 +0,0 @@ -1800 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/debug b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/debug deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/debug +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/decoder_start_token_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/decoder_start_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/decoder_start_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/deepspeed b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/deepspeed deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/deepspeed +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/disable_tqdm b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/disable_tqdm deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/disable_tqdm +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/diversity_penalty b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/diversity_penalty deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/diversity_penalty +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_eval b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_predict b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_predict deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_predict +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_sample b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_sample deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_sample +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_train b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_train deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/do_train +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/early_stopping b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/early_stopping deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/early_stopping +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/encoder_no_repeat_ngram_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/encoder_no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/encoder_no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eos_token_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eos_token_id deleted file mode 100644 index d78b6326c2a8961e1672862523846efebe7ffdd6..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eos_token_id +++ /dev/null @@ -1 +0,0 @@ -128012 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_accumulation_steps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_accumulation_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_delay b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_delay deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_delay +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_do_concat_batches b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_do_concat_batches deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_do_concat_batches +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_on_start b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_on_start deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_on_start +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_steps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_strategy b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_strategy deleted file mode 100644 index 54299a48fb3ae76c848b3acc12248574d05d81b8..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_strategy +++ /dev/null @@ -1 +0,0 @@ -no \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_use_gather_object b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_use_gather_object deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/eval_use_gather_object +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/exponential_decay_length_penalty b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/exponential_decay_length_penalty deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/exponential_decay_length_penalty +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/finetuning_task b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/finetuning_task deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/finetuning_task +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/forced_bos_token_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/forced_bos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/forced_bos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/forced_eos_token_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/forced_eos_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/forced_eos_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16 b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_backend b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_full_eval b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_full_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_full_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_opt_level b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_opt_level deleted file mode 100644 index a9ada426ac8819467c6dc392dcbea40183a3e16e..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fp16_opt_level +++ /dev/null @@ -1 +0,0 @@ -O1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_config b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_config deleted file mode 100644 index 9d33480169a14dfac929530aefc3cd1f5776a983..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_config +++ /dev/null @@ -1 +0,0 @@ -{'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False} \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_min_num_params b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_min_num_params deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_min_num_params +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_transformer_layer_cls_to_wrap b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_transformer_layer_cls_to_wrap deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/fsdp_transformer_layer_cls_to_wrap +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/full_determinism b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/full_determinism deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/full_determinism +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_accumulation_steps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_accumulation_steps deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_accumulation_steps +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_checkpointing b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_checkpointing deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_checkpointing +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_checkpointing_kwargs b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_checkpointing_kwargs deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/gradient_checkpointing_kwargs +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/greater_is_better b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/greater_is_better deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/greater_is_better +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/group_by_length b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/group_by_length deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/group_by_length +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/half_precision_backend b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/half_precision_backend deleted file mode 100644 index 4d18c3e59ecf5c28b46b06ce26f2406b2d449870..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/half_precision_backend +++ /dev/null @@ -1 +0,0 @@ -auto \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hidden_act b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hidden_act deleted file mode 100644 index 84972cd9564e61cac416981cb71bb1e176046f68..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hidden_act +++ /dev/null @@ -1 +0,0 @@ -silu \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hidden_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hidden_size deleted file mode 100644 index f3e53ee118f90809468f69873ccb9d675089cd74..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hidden_size +++ /dev/null @@ -1 +0,0 @@ -2048 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_always_push b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_always_push deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_always_push +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_model_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_private_repo b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_private_repo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_private_repo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_revision b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_revision deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_revision +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_strategy b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_strategy deleted file mode 100644 index 8532b12ca8add8fe61b84623fab9d559a366ce3c..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_strategy +++ /dev/null @@ -1 +0,0 @@ -every_save \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_token b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_token deleted file mode 100644 index 0a574a354979ef783f5f4fe08c3595f79596ff41..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/id2label b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/id2label deleted file mode 100644 index 74c276dcae370126a18f5657c0e1ed72e72325e9..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/id2label +++ /dev/null @@ -1 +0,0 @@ -{0: 'LABEL_0', 1: 'LABEL_1'} \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ignore_data_skip b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ignore_data_skip deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ignore_data_skip +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_for_metrics b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_for_metrics deleted file mode 100644 index 0637a088a01e8ddab3bf3fa98dbe804cbde1a0dc..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_for_metrics +++ /dev/null @@ -1 +0,0 @@ -[] \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_inputs_for_metrics b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_inputs_for_metrics deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_inputs_for_metrics +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_num_input_tokens_seen b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_num_input_tokens_seen deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_num_input_tokens_seen +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_tokens_per_second b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_tokens_per_second deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/include_tokens_per_second +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/initializer_range b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/initializer_range deleted file mode 100644 index 79dd775c1e90ab736c362ede2f2332678eccf47e..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/initializer_range +++ /dev/null @@ -1 +0,0 @@ -0.02 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/intermediate_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/intermediate_size deleted file mode 100644 index 9b7babb1160a44a2cf9ae0dbfd43efd0fd70838b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/intermediate_size +++ /dev/null @@ -1 +0,0 @@ -11008 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/is_decoder b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/is_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/is_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/is_encoder_decoder b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/is_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/is_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/jit_mode_eval b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/jit_mode_eval deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/jit_mode_eval +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label2id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label2id deleted file mode 100644 index 0589857be5c3ad7b568bf7c79a4172a5aa887693..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label2id +++ /dev/null @@ -1 +0,0 @@ -{'LABEL_0': 0, 'LABEL_1': 1} \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label_names b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label_names deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label_names +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label_smoothing_factor b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label_smoothing_factor deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/label_smoothing_factor +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/layer_types b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/layer_types deleted file mode 100644 index bc44845b7022ebf6e7ac1a8dac64d5ba0daa15b1..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/layer_types +++ /dev/null @@ -1 +0,0 @@ -['full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention', 'full_attention'] \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/learning_rate b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/learning_rate deleted file mode 100644 index 335f2feb826f13cdd89c6b5234de9a8c96396bb7..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/learning_rate +++ /dev/null @@ -1 +0,0 @@ -5e-05 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/length_column_name b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/length_column_name deleted file mode 100644 index c2e7ec839dabf14d5d59f187c6b8fdb3460872aa..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/length_column_name +++ /dev/null @@ -1 +0,0 @@ -length \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/length_penalty b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/length_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/length_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/liger_kernel_config b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/liger_kernel_config deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/liger_kernel_config +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/load_best_model_at_end b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/load_best_model_at_end deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/load_best_model_at_end +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/local_rank b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/local_rank deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/local_rank +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_level b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_level deleted file mode 100644 index ecf328558d66d304c19bdd373f647085a3f0880d..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_level +++ /dev/null @@ -1 +0,0 @@ -passive \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_level_replica b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_level_replica deleted file mode 100644 index 14b472df8d4481c6fea79c066ae4650980f02b7c..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_level_replica +++ /dev/null @@ -1 +0,0 @@ -warning \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_on_each_node b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_on_each_node deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/log_on_each_node +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_dir b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_dir deleted file mode 100644 index 549d3e4f6076465f53d3d8895d082ab2cf9b6e8f..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust/runs/Jul21_12-08-44_Jasons-MacBook-Pro.local \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_first_step b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_first_step deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_first_step +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_nan_inf_filter b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_nan_inf_filter deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_nan_inf_filter +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_steps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_steps deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_steps +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_strategy b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/logging_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/lr_scheduler_kwargs b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/lr_scheduler_kwargs deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/lr_scheduler_kwargs +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/lr_scheduler_type b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/lr_scheduler_type deleted file mode 100644 index 9c01abbc1e45086237d1a59d6ec6bdea2400f1a7..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/lr_scheduler_type +++ /dev/null @@ -1 +0,0 @@ -linear \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_grad_norm b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_grad_norm deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_grad_norm +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_length b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_length deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_length +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_position_embeddings b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_position_embeddings deleted file mode 100644 index 83c406c4fd41a6ec8a25290b17cd3894ca57ac4b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_position_embeddings +++ /dev/null @@ -1 +0,0 @@ -65536 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_steps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_steps deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_steps +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_window_layers b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_window_layers deleted file mode 100644 index 368f89ceef179cc546403ac0d5ef1d0e4b340447..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/max_window_layers +++ /dev/null @@ -1 +0,0 @@ -28 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/metric_for_best_model b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/metric_for_best_model deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/metric_for_best_model +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/min_length b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/min_length deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/min_length +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/mlp_bias b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/mlp_bias deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/mlp_bias +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/model_type b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/model_type deleted file mode 100644 index 34bda861a9301e0dab58899ec4828a0ee9f464b1..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/model_type +++ /dev/null @@ -1 +0,0 @@ -smollm3 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/mp_parameters b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/mp_parameters deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/neftune_noise_alpha b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/neftune_noise_alpha deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/neftune_noise_alpha +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_cuda b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_cuda deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_cuda +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_repeat_ngram_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_repeat_ngram_size deleted file mode 100644 index c227083464fb9af8955c90d2924774ee50abb547..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_repeat_ngram_size +++ /dev/null @@ -1 +0,0 @@ -0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_rope_layer_interval b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_rope_layer_interval deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_rope_layer_interval +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_rope_layers b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_rope_layers deleted file mode 100644 index 2cdfac1b0d51ce92d6d6469bd544bc76a3521cec..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/no_rope_layers +++ /dev/null @@ -1 +0,0 @@ -[1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0] \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_attention_heads b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_attention_heads deleted file mode 100644 index 19c7bdba7b1e9bfe80365a50420a6d538ca503c3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_attention_heads +++ /dev/null @@ -1 +0,0 @@ -16 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_beam_groups b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_beam_groups deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_beam_groups +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_beams b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_beams deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_beams +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_hidden_layers b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_hidden_layers deleted file mode 100644 index dce6588ca1420a69eea5699f2ec5d666daf16a6a..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_hidden_layers +++ /dev/null @@ -1 +0,0 @@ -36 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_key_value_heads b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_key_value_heads deleted file mode 100644 index bf0d87ab1b2b0ec1a11a3973d2845b42413d9767..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_key_value_heads +++ /dev/null @@ -1 +0,0 @@ -4 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_return_sequences b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_return_sequences deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_return_sequences +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_train_epochs b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_train_epochs deleted file mode 100644 index e440e5c842586965a7fb77deda2eca68612b1f53..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/num_train_epochs +++ /dev/null @@ -1 +0,0 @@ -3 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim deleted file mode 100644 index 2fd30f30cf2a0413799ab7959d66333f63162f20..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim +++ /dev/null @@ -1 +0,0 @@ -adamw_torch \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim_args b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim_args deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim_args +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim_target_modules b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim_target_modules deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/optim_target_modules +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_attentions b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_attentions deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_attentions +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_dir b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_dir deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_dir +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_hidden_states b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_hidden_states deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_hidden_states +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_scores b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_scores deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/output_scores +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/overwrite_output_dir b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/overwrite_output_dir deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/overwrite_output_dir +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pad_token_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pad_token_id deleted file mode 100644 index 8d9491efc0fb90c90645db18bae256c4f9daff70..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pad_token_id +++ /dev/null @@ -1 +0,0 @@ -128004 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/past_index b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/past_index deleted file mode 100644 index d7d17fcbef95ca19081c4cc5e97cbc592cc7081f..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/past_index +++ /dev/null @@ -1 +0,0 @@ --1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_device_eval_batch_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_device_eval_batch_size deleted file mode 100644 index 301160a93062df23030a69f4b5e4d9bf71866ee9..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_device_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -8 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_device_train_batch_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_device_train_batch_size deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_device_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_gpu_eval_batch_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_gpu_eval_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_gpu_eval_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_gpu_train_batch_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_gpu_train_batch_size deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/per_gpu_train_batch_size +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/prediction_loss_only b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/prediction_loss_only deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/prediction_loss_only +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/prefix b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/prefix deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/prefix +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pretraining_tp b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pretraining_tp deleted file mode 100644 index 56a6051ca2b02b04ef92d5150c9ef600403cb1de..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pretraining_tp +++ /dev/null @@ -1 +0,0 @@ -1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/problem_type b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/problem_type deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/problem_type +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pruned_heads b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pruned_heads deleted file mode 100644 index 9e26dfeeb6e641a33dae4961196235bdb965b21b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/pruned_heads +++ /dev/null @@ -1 +0,0 @@ -{} \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_model_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_model_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_model_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_organization b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_organization deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_organization +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_token b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_token deleted file mode 100644 index 36e61093756f7c43b24cd50fc63164c08bcf50f1..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/push_to_hub_token +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ray_scope b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ray_scope deleted file mode 100644 index 1c1206e8bf4337e96dad9a6d139628852077558d..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/ray_scope +++ /dev/null @@ -1 +0,0 @@ -last \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/remove_invalid_values b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/remove_invalid_values deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/remove_invalid_values +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/remove_unused_columns b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/remove_unused_columns deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/remove_unused_columns +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/repetition_penalty b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/repetition_penalty deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/repetition_penalty +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/report_to b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/report_to deleted file mode 100644 index b4c01973f975b264ce9a4952bbeaa1f1b8bdb018..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/report_to +++ /dev/null @@ -1 +0,0 @@ -['mlflow', 'tensorboard'] \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/restore_callback_states_from_checkpoint b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/restore_callback_states_from_checkpoint deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/restore_callback_states_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/resume_from_checkpoint b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/resume_from_checkpoint deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/resume_from_checkpoint +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/return_dict b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/return_dict deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/return_dict +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/return_dict_in_generate b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/return_dict_in_generate deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/return_dict_in_generate +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rms_norm_eps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rms_norm_eps deleted file mode 100644 index d6bc26a15d7729bbaf8f0e522bea2d0a8781886d..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rms_norm_eps +++ /dev/null @@ -1 +0,0 @@ -1e-06 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rope_scaling b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rope_scaling deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rope_scaling +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rope_theta b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rope_theta deleted file mode 100644 index 445ba6793f521c521c5da54f26f04a56ecc83ee2..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/rope_theta +++ /dev/null @@ -1 +0,0 @@ -5000000.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/run_name b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/run_name deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/run_name +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_on_each_node b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_on_each_node deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_on_each_node +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_only_model b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_only_model deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_only_model +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_safetensors b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_safetensors deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_safetensors +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_steps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_steps deleted file mode 100644 index 2edeafb09db0093bae6ff060e2dcd2166f5c9387..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_steps +++ /dev/null @@ -1 +0,0 @@ -20 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_strategy b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_strategy deleted file mode 100644 index 17f15e19cf5e8064aff8d528657b70e9611eb59e..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_strategy +++ /dev/null @@ -1 +0,0 @@ -steps \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_total_limit b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_total_limit deleted file mode 100644 index d8263ee9860594d2806b0dfd1bfd17528b0ba2a4..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/save_total_limit +++ /dev/null @@ -1 +0,0 @@ -2 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/seed b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/seed deleted file mode 100644 index f70d7bba4ae1f07682e0358bd7a2068094fc023b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/seed +++ /dev/null @@ -1 +0,0 @@ -42 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/sep_token_id b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/sep_token_id deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/sep_token_id +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/skip_memory_metrics b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/skip_memory_metrics deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/skip_memory_metrics +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/sliding_window b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/sliding_window deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/sliding_window +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/suppress_tokens b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/suppress_tokens deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/suppress_tokens +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/task_specific_params b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/task_specific_params deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/task_specific_params +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/temperature b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/temperature deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/temperature +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tf32 b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tf32 deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tf32 +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tf_legacy_loss b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tf_legacy_loss deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tf_legacy_loss +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tie_encoder_decoder b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tie_encoder_decoder deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tie_encoder_decoder +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tie_word_embeddings b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tie_word_embeddings deleted file mode 100644 index 4791ed5559bd77f54e1520025768e2b368705876..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tie_word_embeddings +++ /dev/null @@ -1 +0,0 @@ -True \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tokenizer_class b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tokenizer_class deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tokenizer_class +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/top_k b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/top_k deleted file mode 100644 index c5b431b6cba29540b4b284840ff229bce0460886..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/top_k +++ /dev/null @@ -1 +0,0 @@ -50 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/top_p b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/top_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/top_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile_backend b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile_backend deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile_backend +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile_mode b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile_mode deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_compile_mode +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_dtype b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_dtype deleted file mode 100644 index 0bbd74cbf1ceabb0ecd8bc8edd8b32f9c6145df9..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_dtype +++ /dev/null @@ -1 +0,0 @@ -float32 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_empty_cache_steps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_empty_cache_steps deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torch_empty_cache_steps +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torchdynamo b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torchdynamo deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torchdynamo +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torchscript b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torchscript deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/torchscript +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tpu_metrics_debug b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tpu_metrics_debug deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tpu_metrics_debug +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tpu_num_cores b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tpu_num_cores deleted file mode 100644 index 4af18322e32f3dd19579c80e26e4a306ad11e049..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/tpu_num_cores +++ /dev/null @@ -1 +0,0 @@ -None \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/transformers_version b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/transformers_version deleted file mode 100644 index ac845d9f6e1bec07745cc011b3f0c5a46371931b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/transformers_version +++ /dev/null @@ -1 +0,0 @@ -4.53.1 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/typical_p b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/typical_p deleted file mode 100644 index 9f8e9b69a33f4e8067d5b21661a35d8856758aba..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/typical_p +++ /dev/null @@ -1 +0,0 @@ -1.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_bfloat16 b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_bfloat16 deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_bfloat16 +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_cache b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_cache deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_cache +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_cpu b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_cpu deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_cpu +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_ipex b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_ipex deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_ipex +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_legacy_prediction_loop b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_legacy_prediction_loop deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_legacy_prediction_loop +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_liger_kernel b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_liger_kernel deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_liger_kernel +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_mps_device b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_mps_device deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_mps_device +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_sliding_window b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_sliding_window deleted file mode 100644 index c1f22fbc23bb6ee67824843d6685826db10313d3..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/use_sliding_window +++ /dev/null @@ -1 +0,0 @@ -False \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/vocab_size b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/vocab_size deleted file mode 100644 index 34c44b19378193a3b4fa853df4426fc426c15535..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/vocab_size +++ /dev/null @@ -1 +0,0 @@ -128256 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/warmup_ratio b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/warmup_ratio deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/warmup_ratio +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/warmup_steps b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/warmup_steps deleted file mode 100644 index 9a037142aa3c1b4c490e1a38251620f113465330..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/warmup_steps +++ /dev/null @@ -1 +0,0 @@ -10 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/weight_decay b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/weight_decay deleted file mode 100644 index 171538eb0b00f4eddffa17929796de55b838f34b..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/params/weight_decay +++ /dev/null @@ -1 +0,0 @@ -0.0 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.runName b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.runName deleted file mode 100644 index 24c59d2f037f809b72f92a9fc774206667220976..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.runName +++ /dev/null @@ -1 +0,0 @@ -./smollm3_robust \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.git.commit b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.git.commit deleted file mode 100644 index 64393076c1626bfaa09d72c60ec8c9a9769145c8..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.git.commit +++ /dev/null @@ -1 +0,0 @@ -e2610e8c39108b089782f99d0af9c3a8939adc61 \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.name b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.name deleted file mode 100644 index 124d840089086dc6280e8eb4e84f59ad6b0b015f..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.name +++ /dev/null @@ -1 +0,0 @@ -tool_trainer_simple_robust.py \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.type b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.type deleted file mode 100644 index 0c2c1fe9dc63b7040bb81006635e50fd528f056f..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.source.type +++ /dev/null @@ -1 +0,0 @@ -LOCAL \ No newline at end of file diff --git a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.user b/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.user deleted file mode 100644 index 177a54f531f5330375a61bd19a0fe32029ca991c..0000000000000000000000000000000000000000 --- a/mlruns/0/cb7ad079da3f477cae56d62ec8ee80a9/tags/mlflow.user +++ /dev/null @@ -1 +0,0 @@ -jasonlovell \ No newline at end of file diff --git a/mlruns/0/meta.yaml b/mlruns/0/meta.yaml deleted file mode 100644 index df8a8c144556ceca35583fe20347d17af0150be0..0000000000000000000000000000000000000000 --- a/mlruns/0/meta.yaml +++ /dev/null @@ -1,6 +0,0 @@ -artifact_location: file:///Users/jasonlovell/AI/Learning%20Projects/Dynamic%20Function-Calling%20Agent/mlruns/0 -creation_time: 1753092389691 -experiment_id: '0' -last_update_time: 1753092389691 -lifecycle_stage: active -name: Default diff --git a/smollm3_robust/adapter_config.json b/smollm3_robust/adapter_config.json deleted file mode 100644 index 0be8c070aaf97a2debdeb19f27452147952b65b5..0000000000000000000000000000000000000000 --- a/smollm3_robust/adapter_config.json +++ /dev/null @@ -1,41 +0,0 @@ -{ - "alpha_pattern": {}, - "auto_mapping": null, - "base_model_name_or_path": "HuggingFaceTB/SmolLM3-3B", - "bias": "none", - "corda_config": null, - "eva_config": null, - "exclude_modules": null, - "fan_in_fan_out": false, - "inference_mode": true, - "init_lora_weights": true, - "layer_replication": null, - "layers_pattern": null, - "layers_to_transform": null, - "loftq_config": {}, - "lora_alpha": 16, - "lora_bias": false, - "lora_dropout": 0.1, - "megatron_config": null, - "megatron_core": "megatron.core", - "modules_to_save": null, - "peft_type": "LORA", - "qalora_group_size": 16, - "r": 8, - "rank_pattern": {}, - "revision": null, - "target_modules": [ - "k_proj", - "down_proj", - "gate_proj", - "o_proj", - "q_proj", - "up_proj", - "v_proj" - ], - "task_type": "CAUSAL_LM", - "trainable_token_indices": null, - "use_dora": false, - "use_qalora": false, - "use_rslora": false -} \ No newline at end of file diff --git a/smollm3_robust/adapter_model.safetensors.REMOVED.git-id b/smollm3_robust/adapter_model.safetensors.REMOVED.git-id deleted file mode 100644 index cee193d4bd399f8d3b3b25a3e4aac7e10747f658..0000000000000000000000000000000000000000 --- a/smollm3_robust/adapter_model.safetensors.REMOVED.git-id +++ /dev/null @@ -1 +0,0 @@ -e1ab487a3063f2ab75b74748b76635cee4549110 \ No newline at end of file diff --git a/smollm3_robust/chat_template.jinja b/smollm3_robust/chat_template.jinja deleted file mode 100644 index e01e3a1bca00ae47bca8326b38cc397729f87481..0000000000000000000000000000000000000000 --- a/smollm3_robust/chat_template.jinja +++ /dev/null @@ -1,94 +0,0 @@ -{# ───── defaults ───── #} -{%- if enable_thinking is not defined -%} -{%- set enable_thinking = true -%} -{%- endif -%} - -{# ───── reasoning mode ───── #} -{%- if enable_thinking -%} - {%- set reasoning_mode = "/think" -%} -{%- else -%} - {%- set reasoning_mode = "/no_think" -%} -{%- endif -%} - -{# ───── header (system message) ───── #} -{{- "<|im_start|>system\n" -}} - -{%- if messages[0].role == "system" -%} - {%- set system_message = messages[0].content -%} - {%- if "/no_think" in system_message -%} - {%- set reasoning_mode = "/no_think" -%} - {%- elif "/think" in system_message -%} - {%- set reasoning_mode = "/think" -%} - {%- endif -%} - {%- set custom_instructions = system_message.replace("/no_think", "").replace("/think", "").rstrip() -%} -{%- endif -%} - -{%- if "/system_override" in system_message -%} - {{- custom_instructions.replace("/system_override", "").rstrip() -}} - {{- "<|im_end|>\n" -}} -{%- else -%} - {{- "## Metadata\n\n" -}} - {{- "Knowledge Cutoff Date: June 2025\n" -}} - {%- set today = strftime_now("%d %B %Y") -%} - {{- "Today Date: " ~ today ~ "\n" -}} - {{- "Reasoning Mode: " + reasoning_mode + "\n\n" -}} - - {{- "## Custom Instructions\n\n" -}} - {%- if custom_instructions -%} - {{- custom_instructions + "\n\n" -}} - {%- elif reasoning_mode == "/think" -%} - {{- "You are a helpful AI assistant named SmolLM, trained by Hugging Face. Your role as an assistant involves thoroughly exploring questions through a systematic thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracking, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution using the specified format: Thought section Solution section. In the Thought section, detail your reasoning process in steps. Each step should include detailed considerations such as analysing questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The Solution section should be logical, accurate, and concise and detail necessary steps needed to reach the conclusion.\n\n" -}} - {%- else -%} - {{- "You are a helpful AI assistant named SmolLM, trained by Hugging Face.\n\n" -}} - {%- endif -%} - - {%- if xml_tools or python_tools or tools -%} - {{- "### Tools\n\n" -}} - {%- if xml_tools or tools -%} - {%- if tools -%} - {%- set xml_tools = tools -%} - {%- endif -%} - {%- set ns = namespace(xml_tool_string="You may call one or more functions to assist with the user query.\nYou are provided with function signatures within XML tags:\n\n\n") -%} - {%- for tool in xml_tools[:] -%} {# The slicing makes sure that xml_tools is a list #} - {%- set ns.xml_tool_string = ns.xml_tool_string ~ (tool | string) ~ "\n" -%} - {%- endfor -%} - {%- set xml_tool_string = ns.xml_tool_string + "\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n" -%} - {{- xml_tool_string -}} - {%- endif -%} - {%- if python_tools -%} - {%- set ns = namespace(python_tool_string="When you send a message containing Python code between '' and '' tags, it will be executed in a stateful Jupyter notebook environment, and you will then be given the output to continued reasoning in an agentic loop.\n\nYou can use the following tools in your python code like regular functions:\n\n") -%} - {%- for tool in python_tools[:] -%} {# The slicing makes sure that python_tools is a list #} - {%- set ns.python_tool_string = ns.python_tool_string ~ (tool | string) ~ "\n" -%} - {%- endfor -%} - {%- set python_tool_string = ns.python_tool_string + "\n\nThe state persists between code executions: so variables that you define in one step are still available thereafter." -%} - {{- python_tool_string -}} - {%- endif -%} - {{- "\n\n" -}} - {{- "<|im_end|>\n" -}} - {%- endif -%} -{%- endif -%} -{# ───── main loop ───── #} -{%- for message in messages -%} - {%- set content = message.content if message.content is string else "" -%} - {%- if message.role == "user" -%} - {{ "<|im_start|>" + message.role + "\n" + content + "<|im_end|>\n" }} - {%- elif message.role == "assistant" -%} - {% generation %} - {%- if reasoning_mode == "/think" -%} - {{ "<|im_start|>assistant\n" + content.lstrip("\n") + "<|im_end|>\n" }} - {%- else -%} - {{ "<|im_start|>assistant\n" + "\n\n\n" + content.lstrip("\n") + "<|im_end|>\n" }} - {%- endif -%} - {% endgeneration %} - {%- elif message.role == "tool" -%} - {{ "<|im_start|>" + "user\n" + content + "<|im_end|>\n" }} - {%- endif -%} -{%- endfor -%} -{# ───── generation prompt ───── #} -{%- if add_generation_prompt -%} - {%- if reasoning_mode == "/think" -%} - {{ "<|im_start|>assistant\n" }} - {%- else -%} - {{ "<|im_start|>assistant\n" + "\n\n\n" }} - {%- endif -%} -{%- endif -%} \ No newline at end of file diff --git a/smollm3_robust/checkpoint-660/adapter_config.json b/smollm3_robust/checkpoint-660/adapter_config.json deleted file mode 100644 index 0be8c070aaf97a2debdeb19f27452147952b65b5..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-660/adapter_config.json +++ /dev/null @@ -1,41 +0,0 @@ -{ - "alpha_pattern": {}, - "auto_mapping": null, - "base_model_name_or_path": "HuggingFaceTB/SmolLM3-3B", - "bias": "none", - "corda_config": null, - "eva_config": null, - "exclude_modules": null, - "fan_in_fan_out": false, - "inference_mode": true, - "init_lora_weights": true, - "layer_replication": null, - "layers_pattern": null, - "layers_to_transform": null, - "loftq_config": {}, - "lora_alpha": 16, - "lora_bias": false, - "lora_dropout": 0.1, - "megatron_config": null, - "megatron_core": "megatron.core", - "modules_to_save": null, - "peft_type": "LORA", - "qalora_group_size": 16, - "r": 8, - "rank_pattern": {}, - "revision": null, - "target_modules": [ - "k_proj", - "down_proj", - "gate_proj", - "o_proj", - "q_proj", - "up_proj", - "v_proj" - ], - "task_type": "CAUSAL_LM", - "trainable_token_indices": null, - "use_dora": false, - "use_qalora": false, - "use_rslora": false -} \ No newline at end of file diff --git a/smollm3_robust/checkpoint-660/adapter_model.safetensors.REMOVED.git-id b/smollm3_robust/checkpoint-660/adapter_model.safetensors.REMOVED.git-id deleted file mode 100644 index e194680c388ba8f448dff8a80347ea0a6709201d..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-660/adapter_model.safetensors.REMOVED.git-id +++ /dev/null @@ -1 +0,0 @@ -01b966d4241f8853f02bdbc30bb6340f7bbc0b73 \ No newline at end of file diff --git a/smollm3_robust/checkpoint-660/chat_template.jinja b/smollm3_robust/checkpoint-660/chat_template.jinja deleted file mode 100644 index e01e3a1bca00ae47bca8326b38cc397729f87481..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-660/chat_template.jinja +++ /dev/null @@ -1,94 +0,0 @@ -{# ───── defaults ───── #} -{%- if enable_thinking is not defined -%} -{%- set enable_thinking = true -%} -{%- endif -%} - -{# ───── reasoning mode ───── #} -{%- if enable_thinking -%} - {%- set reasoning_mode = "/think" -%} -{%- else -%} - {%- set reasoning_mode = "/no_think" -%} -{%- endif -%} - -{# ───── header (system message) ───── #} -{{- "<|im_start|>system\n" -}} - -{%- if messages[0].role == "system" -%} - {%- set system_message = messages[0].content -%} - {%- if "/no_think" in system_message -%} - {%- set reasoning_mode = "/no_think" -%} - {%- elif "/think" in system_message -%} - {%- set reasoning_mode = "/think" -%} - {%- endif -%} - {%- set custom_instructions = system_message.replace("/no_think", "").replace("/think", "").rstrip() -%} -{%- endif -%} - -{%- if "/system_override" in system_message -%} - {{- custom_instructions.replace("/system_override", "").rstrip() -}} - {{- "<|im_end|>\n" -}} -{%- else -%} - {{- "## Metadata\n\n" -}} - {{- "Knowledge Cutoff Date: June 2025\n" -}} - {%- set today = strftime_now("%d %B %Y") -%} - {{- "Today Date: " ~ today ~ "\n" -}} - {{- "Reasoning Mode: " + reasoning_mode + "\n\n" -}} - - {{- "## Custom Instructions\n\n" -}} - {%- if custom_instructions -%} - {{- custom_instructions + "\n\n" -}} - {%- elif reasoning_mode == "/think" -%} - {{- "You are a helpful AI assistant named SmolLM, trained by Hugging Face. Your role as an assistant involves thoroughly exploring questions through a systematic thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracking, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution using the specified format: Thought section Solution section. In the Thought section, detail your reasoning process in steps. Each step should include detailed considerations such as analysing questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The Solution section should be logical, accurate, and concise and detail necessary steps needed to reach the conclusion.\n\n" -}} - {%- else -%} - {{- "You are a helpful AI assistant named SmolLM, trained by Hugging Face.\n\n" -}} - {%- endif -%} - - {%- if xml_tools or python_tools or tools -%} - {{- "### Tools\n\n" -}} - {%- if xml_tools or tools -%} - {%- if tools -%} - {%- set xml_tools = tools -%} - {%- endif -%} - {%- set ns = namespace(xml_tool_string="You may call one or more functions to assist with the user query.\nYou are provided with function signatures within XML tags:\n\n\n") -%} - {%- for tool in xml_tools[:] -%} {# The slicing makes sure that xml_tools is a list #} - {%- set ns.xml_tool_string = ns.xml_tool_string ~ (tool | string) ~ "\n" -%} - {%- endfor -%} - {%- set xml_tool_string = ns.xml_tool_string + "\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n" -%} - {{- xml_tool_string -}} - {%- endif -%} - {%- if python_tools -%} - {%- set ns = namespace(python_tool_string="When you send a message containing Python code between '' and '' tags, it will be executed in a stateful Jupyter notebook environment, and you will then be given the output to continued reasoning in an agentic loop.\n\nYou can use the following tools in your python code like regular functions:\n\n") -%} - {%- for tool in python_tools[:] -%} {# The slicing makes sure that python_tools is a list #} - {%- set ns.python_tool_string = ns.python_tool_string ~ (tool | string) ~ "\n" -%} - {%- endfor -%} - {%- set python_tool_string = ns.python_tool_string + "\n\nThe state persists between code executions: so variables that you define in one step are still available thereafter." -%} - {{- python_tool_string -}} - {%- endif -%} - {{- "\n\n" -}} - {{- "<|im_end|>\n" -}} - {%- endif -%} -{%- endif -%} -{# ───── main loop ───── #} -{%- for message in messages -%} - {%- set content = message.content if message.content is string else "" -%} - {%- if message.role == "user" -%} - {{ "<|im_start|>" + message.role + "\n" + content + "<|im_end|>\n" }} - {%- elif message.role == "assistant" -%} - {% generation %} - {%- if reasoning_mode == "/think" -%} - {{ "<|im_start|>assistant\n" + content.lstrip("\n") + "<|im_end|>\n" }} - {%- else -%} - {{ "<|im_start|>assistant\n" + "\n\n\n" + content.lstrip("\n") + "<|im_end|>\n" }} - {%- endif -%} - {% endgeneration %} - {%- elif message.role == "tool" -%} - {{ "<|im_start|>" + "user\n" + content + "<|im_end|>\n" }} - {%- endif -%} -{%- endfor -%} -{# ───── generation prompt ───── #} -{%- if add_generation_prompt -%} - {%- if reasoning_mode == "/think" -%} - {{ "<|im_start|>assistant\n" }} - {%- else -%} - {{ "<|im_start|>assistant\n" + "\n\n\n" }} - {%- endif -%} -{%- endif -%} \ No newline at end of file diff --git a/smollm3_robust/checkpoint-660/optimizer.pt.REMOVED.git-id b/smollm3_robust/checkpoint-660/optimizer.pt.REMOVED.git-id deleted file mode 100644 index f72f83f15b2401bddaf97ef8ff52d470709bae02..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-660/optimizer.pt.REMOVED.git-id +++ /dev/null @@ -1 +0,0 @@ -061b7ca68a78c9f8a0f9bb1897c4f09ef44cdcc4 \ No newline at end of file diff --git a/smollm3_robust/checkpoint-660/rng_state.pth b/smollm3_robust/checkpoint-660/rng_state.pth deleted file mode 100644 index 0676050097953d03d17d3687760f20bc5609de6f..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/checkpoint-660/rng_state.pth and /dev/null differ diff --git a/smollm3_robust/checkpoint-660/scheduler.pt b/smollm3_robust/checkpoint-660/scheduler.pt deleted file mode 100644 index 74e0d0bc34d207dae966aa1afd6f4e5d632fbf32..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/checkpoint-660/scheduler.pt and /dev/null differ diff --git a/smollm3_robust/checkpoint-660/special_tokens_map.json b/smollm3_robust/checkpoint-660/special_tokens_map.json deleted file mode 100644 index 190d5624dbbc1ad56f2f34c9d58e03fef7e5328b..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-660/special_tokens_map.json +++ /dev/null @@ -1,16 +0,0 @@ -{ - "eos_token": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false - }, - "pad_token": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false - } -} diff --git a/smollm3_robust/checkpoint-660/tokenizer.json.REMOVED.git-id b/smollm3_robust/checkpoint-660/tokenizer.json.REMOVED.git-id deleted file mode 100644 index 3081f40cf643c6511c69d3e37e6f8238a5cfd53e..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-660/tokenizer.json.REMOVED.git-id +++ /dev/null @@ -1 +0,0 @@ -b70d58b80eb0efbdad29dcf2df4ea60dca527d44 \ No newline at end of file diff --git a/smollm3_robust/checkpoint-660/tokenizer_config.json b/smollm3_robust/checkpoint-660/tokenizer_config.json deleted file mode 100644 index 61910c2db5cbdc9e6a6f37e14aaf00584cc6ad47..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-660/tokenizer_config.json +++ /dev/null @@ -1,2064 +0,0 @@ -{ - "added_tokens_decoder": { - "128000": { - "content": "<|begin_of_text|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128001": { - "content": "<|end_of_text|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128002": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128003": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128004": { - "content": "<|finetune_right_pad_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128005": { - "content": "<|reserved_special_token_2|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128006": { - "content": "<|start_header_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128007": { - "content": "<|end_header_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128008": { - "content": "<|eom_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128009": { - "content": "<|eot_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128010": { - "content": "<|python_tag|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128011": { - "content": "<|im_start|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128012": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128013": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128014": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128015": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128016": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128017": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128018": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128019": { - "content": "<|reserved_special_token_11|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128020": { - "content": "<|reserved_special_token_12|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128021": { - "content": "<|reserved_special_token_13|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128022": { - "content": "<|reserved_special_token_14|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128023": { - "content": "<|reserved_special_token_15|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128024": { - "content": "<|reserved_special_token_16|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128025": { - "content": "<|reserved_special_token_17|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128026": { - "content": "<|reserved_special_token_18|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128027": { - "content": "<|reserved_special_token_19|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128028": { - "content": "<|reserved_special_token_20|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128029": { - "content": "<|reserved_special_token_21|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128030": { - "content": "<|reserved_special_token_22|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128031": { - "content": "<|reserved_special_token_23|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128032": { - "content": "<|reserved_special_token_24|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128033": { - "content": "<|reserved_special_token_25|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128034": { - "content": "<|reserved_special_token_26|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128035": { - "content": "<|reserved_special_token_27|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128036": { - "content": "<|reserved_special_token_28|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128037": { - "content": "<|reserved_special_token_29|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128038": { - "content": "<|reserved_special_token_30|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128039": { - "content": "<|reserved_special_token_31|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128040": { - "content": "<|reserved_special_token_32|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128041": { - "content": "<|reserved_special_token_33|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128042": { - "content": "<|reserved_special_token_34|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128043": { - "content": "<|reserved_special_token_35|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128044": { - "content": "<|reserved_special_token_36|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128045": { - "content": "<|reserved_special_token_37|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128046": { - "content": "<|reserved_special_token_38|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128047": { - "content": "<|reserved_special_token_39|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128048": { - "content": "<|reserved_special_token_40|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128049": { - "content": "<|reserved_special_token_41|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128050": { - "content": "<|reserved_special_token_42|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128051": { - "content": "<|reserved_special_token_43|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128052": { - "content": "<|reserved_special_token_44|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128053": { - "content": "<|reserved_special_token_45|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128054": { - "content": "<|reserved_special_token_46|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128055": { - "content": "<|reserved_special_token_47|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128056": { - "content": "<|reserved_special_token_48|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128057": { - "content": "<|reserved_special_token_49|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128058": { - "content": "<|reserved_special_token_50|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128059": { - "content": "<|reserved_special_token_51|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128060": { - "content": "<|reserved_special_token_52|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128061": { - "content": "<|reserved_special_token_53|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128062": { - "content": "<|reserved_special_token_54|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128063": { - "content": "<|reserved_special_token_55|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128064": { - "content": "<|reserved_special_token_56|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128065": { - "content": "<|reserved_special_token_57|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128066": { - "content": "<|reserved_special_token_58|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128067": { - "content": "<|reserved_special_token_59|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128068": { - "content": "<|reserved_special_token_60|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128069": { - "content": "<|reserved_special_token_61|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128070": { - "content": "<|reserved_special_token_62|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128071": { - "content": "<|reserved_special_token_63|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128072": { - "content": "<|reserved_special_token_64|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128073": { - "content": "<|reserved_special_token_65|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128074": { - "content": "<|reserved_special_token_66|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128075": { - "content": "<|reserved_special_token_67|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128076": { - "content": "<|reserved_special_token_68|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128077": { - "content": "<|reserved_special_token_69|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128078": { - "content": "<|reserved_special_token_70|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128079": { - "content": "<|reserved_special_token_71|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128080": { - "content": "<|reserved_special_token_72|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128081": { - "content": "<|reserved_special_token_73|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128082": { - "content": "<|reserved_special_token_74|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128083": { - "content": "<|reserved_special_token_75|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128084": { - "content": "<|reserved_special_token_76|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128085": { - "content": "<|reserved_special_token_77|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128086": { - "content": "<|reserved_special_token_78|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128087": { - "content": "<|reserved_special_token_79|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128088": { - "content": "<|reserved_special_token_80|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128089": { - "content": "<|reserved_special_token_81|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128090": { - "content": "<|reserved_special_token_82|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128091": { - "content": "<|reserved_special_token_83|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128092": { - "content": "<|reserved_special_token_84|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128093": { - "content": "<|reserved_special_token_85|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128094": { - "content": "<|reserved_special_token_86|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128095": { - "content": "<|reserved_special_token_87|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128096": { - "content": "<|reserved_special_token_88|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128097": { - "content": "<|reserved_special_token_89|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128098": { - "content": "<|reserved_special_token_90|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128099": { - "content": "<|reserved_special_token_91|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128100": { - "content": "<|reserved_special_token_92|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128101": { - "content": "<|reserved_special_token_93|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128102": { - "content": "<|reserved_special_token_94|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128103": { - "content": "<|reserved_special_token_95|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128104": { - "content": "<|reserved_special_token_96|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128105": { - "content": "<|reserved_special_token_97|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128106": { - "content": "<|reserved_special_token_98|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128107": { - "content": "<|reserved_special_token_99|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128108": { - "content": "<|reserved_special_token_100|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128109": { - "content": "<|reserved_special_token_101|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128110": { - "content": "<|reserved_special_token_102|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128111": { - "content": "<|reserved_special_token_103|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128112": { - "content": "<|reserved_special_token_104|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128113": { - "content": "<|reserved_special_token_105|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128114": { - "content": "<|reserved_special_token_106|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128115": { - "content": "<|reserved_special_token_107|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128116": { - "content": "<|reserved_special_token_108|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128117": { - "content": "<|reserved_special_token_109|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128118": { - "content": "<|reserved_special_token_110|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128119": { - "content": "<|reserved_special_token_111|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128120": { - "content": "<|reserved_special_token_112|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128121": { - "content": "<|reserved_special_token_113|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128122": { - "content": "<|reserved_special_token_114|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128123": { - "content": "<|reserved_special_token_115|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128124": { - "content": "<|reserved_special_token_116|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128125": { - "content": "<|reserved_special_token_117|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128126": { - "content": "<|reserved_special_token_118|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128127": { - "content": "<|reserved_special_token_119|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128128": { - "content": "<|reserved_special_token_120|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128129": { - "content": "<|reserved_special_token_121|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128130": { - "content": "<|reserved_special_token_122|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128131": { - "content": "<|reserved_special_token_123|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128132": { - "content": "<|reserved_special_token_124|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128133": { - "content": "<|reserved_special_token_125|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128134": { - "content": "<|reserved_special_token_126|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128135": { - "content": "<|reserved_special_token_127|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128136": { - "content": "<|reserved_special_token_128|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128137": { - "content": "<|reserved_special_token_129|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128138": { - "content": "<|reserved_special_token_130|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128139": { - "content": "<|reserved_special_token_131|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128140": { - "content": "<|reserved_special_token_132|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128141": { - "content": "<|reserved_special_token_133|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128142": { - "content": "<|reserved_special_token_134|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128143": { - "content": "<|reserved_special_token_135|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128144": { - "content": "<|reserved_special_token_136|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128145": { - "content": "<|reserved_special_token_137|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128146": { - "content": "<|reserved_special_token_138|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128147": { - "content": "<|reserved_special_token_139|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128148": { - "content": "<|reserved_special_token_140|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128149": { - "content": "<|reserved_special_token_141|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128150": { - "content": "<|reserved_special_token_142|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128151": { - "content": "<|reserved_special_token_143|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128152": { - "content": "<|reserved_special_token_144|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128153": { - "content": "<|reserved_special_token_145|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128154": { - "content": "<|reserved_special_token_146|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128155": { - "content": "<|reserved_special_token_147|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128156": { - "content": "<|reserved_special_token_148|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128157": { - "content": "<|reserved_special_token_149|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128158": { - "content": "<|reserved_special_token_150|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128159": { - "content": "<|reserved_special_token_151|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128160": { - "content": "<|reserved_special_token_152|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128161": { - "content": "<|reserved_special_token_153|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128162": { - "content": "<|reserved_special_token_154|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128163": { - "content": "<|reserved_special_token_155|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128164": { - "content": "<|reserved_special_token_156|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128165": { - "content": "<|reserved_special_token_157|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128166": { - "content": "<|reserved_special_token_158|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128167": { - "content": "<|reserved_special_token_159|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128168": { - "content": "<|reserved_special_token_160|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128169": { - "content": "<|reserved_special_token_161|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128170": { - "content": "<|reserved_special_token_162|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128171": { - "content": "<|reserved_special_token_163|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128172": { - "content": "<|reserved_special_token_164|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128173": { - "content": "<|reserved_special_token_165|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128174": { - "content": "<|reserved_special_token_166|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128175": { - "content": "<|reserved_special_token_167|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128176": { - "content": "<|reserved_special_token_168|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128177": { - "content": "<|reserved_special_token_169|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128178": { - "content": "<|reserved_special_token_170|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128179": { - "content": "<|reserved_special_token_171|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128180": { - "content": "<|reserved_special_token_172|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128181": { - "content": "<|reserved_special_token_173|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128182": { - "content": "<|reserved_special_token_174|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128183": { - "content": "<|reserved_special_token_175|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128184": { - "content": "<|reserved_special_token_176|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128185": { - "content": "<|reserved_special_token_177|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128186": { - "content": "<|reserved_special_token_178|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128187": { - "content": "<|reserved_special_token_179|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128188": { - "content": "<|reserved_special_token_180|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128189": { - "content": "<|reserved_special_token_181|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128190": { - "content": "<|reserved_special_token_182|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128191": { - "content": "<|reserved_special_token_183|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128192": { - "content": "<|reserved_special_token_184|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128193": { - "content": "<|reserved_special_token_185|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128194": { - "content": "<|reserved_special_token_186|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128195": { - "content": "<|reserved_special_token_187|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128196": { - "content": "<|reserved_special_token_188|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128197": { - "content": "<|reserved_special_token_189|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128198": { - "content": "<|reserved_special_token_190|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128199": { - "content": "<|reserved_special_token_191|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128200": { - "content": "<|reserved_special_token_192|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128201": { - "content": "<|reserved_special_token_193|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128202": { - "content": "<|reserved_special_token_194|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128203": { - "content": "<|reserved_special_token_195|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128204": { - "content": "<|reserved_special_token_196|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128205": { - "content": "<|reserved_special_token_197|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128206": { - "content": "<|reserved_special_token_198|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128207": { - "content": "<|reserved_special_token_199|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128208": { - "content": "<|reserved_special_token_200|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128209": { - "content": "<|reserved_special_token_201|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128210": { - "content": "<|reserved_special_token_202|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128211": { - "content": "<|reserved_special_token_203|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128212": { - "content": "<|reserved_special_token_204|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128213": { - "content": "<|reserved_special_token_205|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128214": { - "content": "<|reserved_special_token_206|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128215": { - "content": "<|reserved_special_token_207|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128216": { - "content": "<|reserved_special_token_208|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128217": { - "content": "<|reserved_special_token_209|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128218": { - "content": "<|reserved_special_token_210|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128219": { - "content": "<|reserved_special_token_211|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128220": { - "content": "<|reserved_special_token_212|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128221": { - "content": "<|reserved_special_token_213|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128222": { - "content": "<|reserved_special_token_214|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128223": { - "content": "<|reserved_special_token_215|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128224": { - "content": "<|reserved_special_token_216|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128225": { - "content": "<|reserved_special_token_217|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128226": { - "content": "<|reserved_special_token_218|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128227": { - "content": "<|reserved_special_token_219|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128228": { - "content": "<|reserved_special_token_220|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128229": { - "content": "<|reserved_special_token_221|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128230": { - "content": "<|reserved_special_token_222|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128231": { - "content": "<|reserved_special_token_223|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128232": { - "content": "<|reserved_special_token_224|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128233": { - "content": "<|reserved_special_token_225|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128234": { - "content": "<|reserved_special_token_226|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128235": { - "content": "<|reserved_special_token_227|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128236": { - "content": "<|reserved_special_token_228|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128237": { - "content": "<|reserved_special_token_229|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128238": { - "content": "<|reserved_special_token_230|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128239": { - "content": "<|reserved_special_token_231|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128240": { - "content": "<|reserved_special_token_232|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128241": { - "content": "<|reserved_special_token_233|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128242": { - "content": "<|reserved_special_token_234|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128243": { - "content": "<|reserved_special_token_235|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128244": { - "content": "<|reserved_special_token_236|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128245": { - "content": "<|reserved_special_token_237|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128246": { - "content": "<|reserved_special_token_238|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128247": { - "content": "<|reserved_special_token_239|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128248": { - "content": "<|reserved_special_token_240|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128249": { - "content": "<|reserved_special_token_241|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128250": { - "content": "<|reserved_special_token_242|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128251": { - "content": "<|reserved_special_token_243|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128252": { - "content": "<|reserved_special_token_244|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128253": { - "content": "<|reserved_special_token_245|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128254": { - "content": "<|reserved_special_token_246|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128255": { - "content": "<|reserved_special_token_247|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - } - }, - "bos_token": null, - "clean_up_tokenization_spaces": true, - "eos_token": "<|im_end|>", - "extra_special_tokens": {}, - "fast": false, - "model_input_names": [ - "input_ids", - "attention_mask" - ], - "model_max_length": 131072, - "pad_token": "<|im_end|>", - "tokenizer_class": "PreTrainedTokenizerFast" -} diff --git a/smollm3_robust/checkpoint-660/trainer_state.json b/smollm3_robust/checkpoint-660/trainer_state.json deleted file mode 100644 index a39bc019af85b2acfab1cd3fb0c2920267f9ac41..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-660/trainer_state.json +++ /dev/null @@ -1,2344 +0,0 @@ -{ - "best_global_step": null, - "best_metric": null, - "best_model_checkpoint": null, - "epoch": 9.853932584269662, - "eval_steps": 500, - "global_step": 660, - "is_hyper_param_search": false, - "is_local_process_zero": true, - "is_world_process_zero": true, - "log_history": [ - { - "epoch": 0.0299625468164794, - "grad_norm": 3.722816228866577, - "learning_rate": 5e-06, - "loss": 1.6973, - "step": 2 - }, - { - "epoch": 0.0599250936329588, - "grad_norm": 3.4061238765716553, - "learning_rate": 1.5e-05, - "loss": 1.6822, - "step": 4 - }, - { - "epoch": 0.0898876404494382, - "grad_norm": 2.510838270187378, - "learning_rate": 2.5e-05, - "loss": 1.4687, - "step": 6 - }, - { - "epoch": 0.1198501872659176, - "grad_norm": 2.095601797103882, - "learning_rate": 3.5e-05, - "loss": 1.4093, - "step": 8 - }, - { - "epoch": 0.149812734082397, - "grad_norm": 2.001142978668213, - "learning_rate": 4.5e-05, - "loss": 1.1657, - "step": 10 - }, - { - "epoch": 0.1797752808988764, - "grad_norm": 1.8859666585922241, - "learning_rate": 4.992424242424243e-05, - "loss": 0.9447, - "step": 12 - }, - { - "epoch": 0.20973782771535582, - "grad_norm": 2.0142483711242676, - "learning_rate": 4.9772727272727275e-05, - "loss": 0.8334, - "step": 14 - }, - { - "epoch": 0.2397003745318352, - "grad_norm": 1.6467182636260986, - "learning_rate": 4.962121212121213e-05, - "loss": 0.6336, - "step": 16 - }, - { - "epoch": 0.2696629213483146, - "grad_norm": 1.2968051433563232, - "learning_rate": 4.946969696969697e-05, - "loss": 0.4959, - "step": 18 - }, - { - "epoch": 0.299625468164794, - "grad_norm": 1.001763105392456, - "learning_rate": 4.931818181818182e-05, - "loss": 0.3829, - "step": 20 - }, - { - "epoch": 0.3295880149812734, - "grad_norm": 0.47088268399238586, - "learning_rate": 4.9166666666666665e-05, - "loss": 0.3361, - "step": 22 - }, - { - "epoch": 0.3595505617977528, - "grad_norm": 0.6152827739715576, - "learning_rate": 4.901515151515152e-05, - "loss": 0.3142, - "step": 24 - }, - { - "epoch": 0.3895131086142322, - "grad_norm": 0.5034743547439575, - "learning_rate": 4.886363636363637e-05, - "loss": 0.2581, - "step": 26 - }, - { - "epoch": 0.41947565543071164, - "grad_norm": 0.5429691076278687, - "learning_rate": 4.8712121212121216e-05, - "loss": 0.247, - "step": 28 - }, - { - "epoch": 0.449438202247191, - "grad_norm": 0.6162884831428528, - "learning_rate": 4.856060606060606e-05, - "loss": 0.2266, - "step": 30 - }, - { - "epoch": 0.4794007490636704, - "grad_norm": 0.7645140886306763, - "learning_rate": 4.840909090909091e-05, - "loss": 0.1985, - "step": 32 - }, - { - "epoch": 0.5093632958801498, - "grad_norm": 0.600817084312439, - "learning_rate": 4.825757575757576e-05, - "loss": 0.1634, - "step": 34 - }, - { - "epoch": 0.5393258426966292, - "grad_norm": 0.4553565979003906, - "learning_rate": 4.810606060606061e-05, - "loss": 0.1626, - "step": 36 - }, - { - "epoch": 0.5692883895131086, - "grad_norm": 0.7166281342506409, - "learning_rate": 4.795454545454546e-05, - "loss": 0.1796, - "step": 38 - }, - { - "epoch": 0.599250936329588, - "grad_norm": 0.42647111415863037, - "learning_rate": 4.7803030303030304e-05, - "loss": 0.1384, - "step": 40 - }, - { - "epoch": 0.6292134831460674, - "grad_norm": 0.4419521689414978, - "learning_rate": 4.765151515151515e-05, - "loss": 0.1498, - "step": 42 - }, - { - "epoch": 0.6591760299625468, - "grad_norm": 0.49058037996292114, - "learning_rate": 4.75e-05, - "loss": 0.0965, - "step": 44 - }, - { - "epoch": 0.6891385767790262, - "grad_norm": 0.3956356644630432, - "learning_rate": 4.7348484848484855e-05, - "loss": 0.0873, - "step": 46 - }, - { - "epoch": 0.7191011235955056, - "grad_norm": 0.4841479957103729, - "learning_rate": 4.71969696969697e-05, - "loss": 0.0909, - "step": 48 - }, - { - "epoch": 0.7490636704119851, - "grad_norm": 0.5453945398330688, - "learning_rate": 4.704545454545455e-05, - "loss": 0.0719, - "step": 50 - }, - { - "epoch": 0.7790262172284644, - "grad_norm": 0.44312047958374023, - "learning_rate": 4.689393939393939e-05, - "loss": 0.0553, - "step": 52 - }, - { - "epoch": 0.8089887640449438, - "grad_norm": 0.5744581818580627, - "learning_rate": 4.6742424242424245e-05, - "loss": 0.0744, - "step": 54 - }, - { - "epoch": 0.8389513108614233, - "grad_norm": 0.9026234745979309, - "learning_rate": 4.659090909090909e-05, - "loss": 0.0608, - "step": 56 - }, - { - "epoch": 0.8689138576779026, - "grad_norm": 0.4250333905220032, - "learning_rate": 4.6439393939393944e-05, - "loss": 0.0364, - "step": 58 - }, - { - "epoch": 0.898876404494382, - "grad_norm": 0.6075786352157593, - "learning_rate": 4.628787878787879e-05, - "loss": 0.0495, - "step": 60 - }, - { - "epoch": 0.9288389513108615, - "grad_norm": 0.6797979474067688, - "learning_rate": 4.6136363636363635e-05, - "loss": 0.0471, - "step": 62 - }, - { - "epoch": 0.9588014981273408, - "grad_norm": 0.296935498714447, - "learning_rate": 4.598484848484849e-05, - "loss": 0.0299, - "step": 64 - }, - { - "epoch": 0.9887640449438202, - "grad_norm": 0.28508949279785156, - "learning_rate": 4.5833333333333334e-05, - "loss": 0.0309, - "step": 66 - }, - { - "epoch": 1.0149812734082397, - "grad_norm": 0.44689077138900757, - "learning_rate": 4.5681818181818186e-05, - "loss": 0.0487, - "step": 68 - }, - { - "epoch": 1.0449438202247192, - "grad_norm": 0.36559513211250305, - "learning_rate": 4.553030303030303e-05, - "loss": 0.0304, - "step": 70 - }, - { - "epoch": 1.0749063670411985, - "grad_norm": 0.3891605734825134, - "learning_rate": 4.5378787878787885e-05, - "loss": 0.0254, - "step": 72 - }, - { - "epoch": 1.104868913857678, - "grad_norm": 0.36933985352516174, - "learning_rate": 4.522727272727273e-05, - "loss": 0.0363, - "step": 74 - }, - { - "epoch": 1.1348314606741572, - "grad_norm": 0.41140684485435486, - "learning_rate": 4.5075757575757577e-05, - "loss": 0.0386, - "step": 76 - }, - { - "epoch": 1.1647940074906367, - "grad_norm": 0.5980117321014404, - "learning_rate": 4.492424242424242e-05, - "loss": 0.0201, - "step": 78 - }, - { - "epoch": 1.1947565543071161, - "grad_norm": 0.23804394900798798, - "learning_rate": 4.4772727272727275e-05, - "loss": 0.0187, - "step": 80 - }, - { - "epoch": 1.2247191011235956, - "grad_norm": 0.23067264258861542, - "learning_rate": 4.462121212121213e-05, - "loss": 0.0175, - "step": 82 - }, - { - "epoch": 1.2546816479400749, - "grad_norm": 0.36974087357521057, - "learning_rate": 4.4469696969696973e-05, - "loss": 0.0303, - "step": 84 - }, - { - "epoch": 1.2846441947565543, - "grad_norm": 0.4070376753807068, - "learning_rate": 4.431818181818182e-05, - "loss": 0.0268, - "step": 86 - }, - { - "epoch": 1.3146067415730336, - "grad_norm": 0.2764546275138855, - "learning_rate": 4.4166666666666665e-05, - "loss": 0.0224, - "step": 88 - }, - { - "epoch": 1.344569288389513, - "grad_norm": 0.20103320479393005, - "learning_rate": 4.401515151515152e-05, - "loss": 0.0202, - "step": 90 - }, - { - "epoch": 1.3745318352059925, - "grad_norm": 0.26625537872314453, - "learning_rate": 4.386363636363637e-05, - "loss": 0.0216, - "step": 92 - }, - { - "epoch": 1.404494382022472, - "grad_norm": 0.24225257337093353, - "learning_rate": 4.3712121212121216e-05, - "loss": 0.0168, - "step": 94 - }, - { - "epoch": 1.4344569288389513, - "grad_norm": 0.3456275463104248, - "learning_rate": 4.356060606060606e-05, - "loss": 0.0207, - "step": 96 - }, - { - "epoch": 1.4644194756554307, - "grad_norm": 0.21905088424682617, - "learning_rate": 4.340909090909091e-05, - "loss": 0.0145, - "step": 98 - }, - { - "epoch": 1.49438202247191, - "grad_norm": 0.5236210227012634, - "learning_rate": 4.325757575757576e-05, - "loss": 0.025, - "step": 100 - }, - { - "epoch": 1.5243445692883895, - "grad_norm": 0.35533836483955383, - "learning_rate": 4.3106060606060606e-05, - "loss": 0.0183, - "step": 102 - }, - { - "epoch": 1.554307116104869, - "grad_norm": 0.3094329833984375, - "learning_rate": 4.295454545454546e-05, - "loss": 0.022, - "step": 104 - }, - { - "epoch": 1.5842696629213484, - "grad_norm": 0.24415704607963562, - "learning_rate": 4.2803030303030305e-05, - "loss": 0.0149, - "step": 106 - }, - { - "epoch": 1.6142322097378277, - "grad_norm": 0.44377341866493225, - "learning_rate": 4.265151515151515e-05, - "loss": 0.0152, - "step": 108 - }, - { - "epoch": 1.6441947565543071, - "grad_norm": 0.4970695972442627, - "learning_rate": 4.25e-05, - "loss": 0.0196, - "step": 110 - }, - { - "epoch": 1.6741573033707864, - "grad_norm": 0.20200444757938385, - "learning_rate": 4.234848484848485e-05, - "loss": 0.0165, - "step": 112 - }, - { - "epoch": 1.7041198501872659, - "grad_norm": 0.21261580288410187, - "learning_rate": 4.21969696969697e-05, - "loss": 0.0168, - "step": 114 - }, - { - "epoch": 1.7340823970037453, - "grad_norm": 0.2608441412448883, - "learning_rate": 4.204545454545455e-05, - "loss": 0.0146, - "step": 116 - }, - { - "epoch": 1.7640449438202248, - "grad_norm": 0.20836834609508514, - "learning_rate": 4.189393939393939e-05, - "loss": 0.0195, - "step": 118 - }, - { - "epoch": 1.7940074906367043, - "grad_norm": 0.29434454441070557, - "learning_rate": 4.1742424242424246e-05, - "loss": 0.0187, - "step": 120 - }, - { - "epoch": 1.8239700374531835, - "grad_norm": 0.2228062003850937, - "learning_rate": 4.159090909090909e-05, - "loss": 0.0171, - "step": 122 - }, - { - "epoch": 1.8539325842696628, - "grad_norm": 0.3206649124622345, - "learning_rate": 4.143939393939394e-05, - "loss": 0.0204, - "step": 124 - }, - { - "epoch": 1.8838951310861423, - "grad_norm": 0.18132628500461578, - "learning_rate": 4.128787878787879e-05, - "loss": 0.0203, - "step": 126 - }, - { - "epoch": 1.9138576779026217, - "grad_norm": 0.2030370533466339, - "learning_rate": 4.113636363636364e-05, - "loss": 0.0179, - "step": 128 - }, - { - "epoch": 1.9438202247191012, - "grad_norm": 0.2079804688692093, - "learning_rate": 4.098484848484849e-05, - "loss": 0.0198, - "step": 130 - }, - { - "epoch": 1.9737827715355807, - "grad_norm": 0.23763149976730347, - "learning_rate": 4.0833333333333334e-05, - "loss": 0.0182, - "step": 132 - }, - { - "epoch": 2.0, - "grad_norm": 0.21091270446777344, - "learning_rate": 4.068181818181818e-05, - "loss": 0.0186, - "step": 134 - }, - { - "epoch": 2.0299625468164795, - "grad_norm": 0.21226680278778076, - "learning_rate": 4.053030303030303e-05, - "loss": 0.0166, - "step": 136 - }, - { - "epoch": 2.059925093632959, - "grad_norm": 0.2453354448080063, - "learning_rate": 4.0378787878787885e-05, - "loss": 0.0162, - "step": 138 - }, - { - "epoch": 2.0898876404494384, - "grad_norm": 0.17463107407093048, - "learning_rate": 4.022727272727273e-05, - "loss": 0.0156, - "step": 140 - }, - { - "epoch": 2.1198501872659175, - "grad_norm": 0.307685524225235, - "learning_rate": 4.007575757575758e-05, - "loss": 0.0211, - "step": 142 - }, - { - "epoch": 2.149812734082397, - "grad_norm": 0.23119647800922394, - "learning_rate": 3.992424242424242e-05, - "loss": 0.0184, - "step": 144 - }, - { - "epoch": 2.1797752808988764, - "grad_norm": 0.15694215893745422, - "learning_rate": 3.9772727272727275e-05, - "loss": 0.0165, - "step": 146 - }, - { - "epoch": 2.209737827715356, - "grad_norm": 0.26551949977874756, - "learning_rate": 3.962121212121213e-05, - "loss": 0.0184, - "step": 148 - }, - { - "epoch": 2.2397003745318353, - "grad_norm": 0.21518471837043762, - "learning_rate": 3.9469696969696974e-05, - "loss": 0.0168, - "step": 150 - }, - { - "epoch": 2.2696629213483144, - "grad_norm": 0.1955074667930603, - "learning_rate": 3.931818181818182e-05, - "loss": 0.016, - "step": 152 - }, - { - "epoch": 2.299625468164794, - "grad_norm": 0.24419444799423218, - "learning_rate": 3.9166666666666665e-05, - "loss": 0.0151, - "step": 154 - }, - { - "epoch": 2.3295880149812733, - "grad_norm": 0.29677078127861023, - "learning_rate": 3.901515151515152e-05, - "loss": 0.0167, - "step": 156 - }, - { - "epoch": 2.359550561797753, - "grad_norm": 0.19572412967681885, - "learning_rate": 3.8863636363636364e-05, - "loss": 0.0162, - "step": 158 - }, - { - "epoch": 2.3895131086142323, - "grad_norm": 0.20857453346252441, - "learning_rate": 3.8712121212121217e-05, - "loss": 0.0166, - "step": 160 - }, - { - "epoch": 2.4194756554307117, - "grad_norm": 0.14987444877624512, - "learning_rate": 3.856060606060606e-05, - "loss": 0.015, - "step": 162 - }, - { - "epoch": 2.449438202247191, - "grad_norm": 0.22435317933559418, - "learning_rate": 3.840909090909091e-05, - "loss": 0.0171, - "step": 164 - }, - { - "epoch": 2.4794007490636703, - "grad_norm": 0.29637566208839417, - "learning_rate": 3.825757575757576e-05, - "loss": 0.0189, - "step": 166 - }, - { - "epoch": 2.5093632958801497, - "grad_norm": 0.24401557445526123, - "learning_rate": 3.810606060606061e-05, - "loss": 0.0164, - "step": 168 - }, - { - "epoch": 2.539325842696629, - "grad_norm": 0.19472604990005493, - "learning_rate": 3.795454545454545e-05, - "loss": 0.0172, - "step": 170 - }, - { - "epoch": 2.5692883895131087, - "grad_norm": 0.21067409217357635, - "learning_rate": 3.7803030303030305e-05, - "loss": 0.0147, - "step": 172 - }, - { - "epoch": 2.599250936329588, - "grad_norm": 0.157924622297287, - "learning_rate": 3.765151515151516e-05, - "loss": 0.0146, - "step": 174 - }, - { - "epoch": 2.629213483146067, - "grad_norm": 0.22833997011184692, - "learning_rate": 3.7500000000000003e-05, - "loss": 0.0158, - "step": 176 - }, - { - "epoch": 2.6591760299625467, - "grad_norm": 0.1746760606765747, - "learning_rate": 3.734848484848485e-05, - "loss": 0.0173, - "step": 178 - }, - { - "epoch": 2.689138576779026, - "grad_norm": 0.25366461277008057, - "learning_rate": 3.7196969696969695e-05, - "loss": 0.0162, - "step": 180 - }, - { - "epoch": 2.7191011235955056, - "grad_norm": 0.1569383144378662, - "learning_rate": 3.704545454545455e-05, - "loss": 0.0163, - "step": 182 - }, - { - "epoch": 2.749063670411985, - "grad_norm": 0.10542655736207962, - "learning_rate": 3.68939393939394e-05, - "loss": 0.014, - "step": 184 - }, - { - "epoch": 2.7790262172284645, - "grad_norm": 0.11274619400501251, - "learning_rate": 3.6742424242424246e-05, - "loss": 0.0141, - "step": 186 - }, - { - "epoch": 2.808988764044944, - "grad_norm": 0.2910693287849426, - "learning_rate": 3.659090909090909e-05, - "loss": 0.0199, - "step": 188 - }, - { - "epoch": 2.8389513108614235, - "grad_norm": 0.15630793571472168, - "learning_rate": 3.643939393939394e-05, - "loss": 0.0142, - "step": 190 - }, - { - "epoch": 2.8689138576779025, - "grad_norm": 0.1520637422800064, - "learning_rate": 3.628787878787879e-05, - "loss": 0.0166, - "step": 192 - }, - { - "epoch": 2.898876404494382, - "grad_norm": 0.1700475662946701, - "learning_rate": 3.613636363636364e-05, - "loss": 0.0136, - "step": 194 - }, - { - "epoch": 2.9288389513108615, - "grad_norm": 0.2383895367383957, - "learning_rate": 3.598484848484849e-05, - "loss": 0.0146, - "step": 196 - }, - { - "epoch": 2.958801498127341, - "grad_norm": 0.2498546838760376, - "learning_rate": 3.5833333333333335e-05, - "loss": 0.0167, - "step": 198 - }, - { - "epoch": 2.98876404494382, - "grad_norm": 0.24929316341876984, - "learning_rate": 3.568181818181818e-05, - "loss": 0.0195, - "step": 200 - }, - { - "epoch": 3.0149812734082397, - "grad_norm": 0.1542021781206131, - "learning_rate": 3.553030303030303e-05, - "loss": 0.0126, - "step": 202 - }, - { - "epoch": 3.044943820224719, - "grad_norm": 0.15488319098949432, - "learning_rate": 3.537878787878788e-05, - "loss": 0.0143, - "step": 204 - }, - { - "epoch": 3.0749063670411987, - "grad_norm": 0.22363406419754028, - "learning_rate": 3.522727272727273e-05, - "loss": 0.0172, - "step": 206 - }, - { - "epoch": 3.1048689138576777, - "grad_norm": 0.23685605823993683, - "learning_rate": 3.507575757575758e-05, - "loss": 0.0152, - "step": 208 - }, - { - "epoch": 3.134831460674157, - "grad_norm": 0.18075552582740784, - "learning_rate": 3.492424242424242e-05, - "loss": 0.016, - "step": 210 - }, - { - "epoch": 3.1647940074906367, - "grad_norm": 0.18518349528312683, - "learning_rate": 3.4772727272727276e-05, - "loss": 0.0147, - "step": 212 - }, - { - "epoch": 3.194756554307116, - "grad_norm": 0.29419267177581787, - "learning_rate": 3.462121212121212e-05, - "loss": 0.0159, - "step": 214 - }, - { - "epoch": 3.2247191011235956, - "grad_norm": 0.13987213373184204, - "learning_rate": 3.4469696969696974e-05, - "loss": 0.0136, - "step": 216 - }, - { - "epoch": 3.254681647940075, - "grad_norm": 0.2520444989204407, - "learning_rate": 3.431818181818182e-05, - "loss": 0.0172, - "step": 218 - }, - { - "epoch": 3.284644194756554, - "grad_norm": 0.12775233387947083, - "learning_rate": 3.4166666666666666e-05, - "loss": 0.0149, - "step": 220 - }, - { - "epoch": 3.3146067415730336, - "grad_norm": 0.15709002315998077, - "learning_rate": 3.401515151515152e-05, - "loss": 0.0153, - "step": 222 - }, - { - "epoch": 3.344569288389513, - "grad_norm": 0.17532695829868317, - "learning_rate": 3.3863636363636364e-05, - "loss": 0.0149, - "step": 224 - }, - { - "epoch": 3.3745318352059925, - "grad_norm": 0.21435686945915222, - "learning_rate": 3.371212121212121e-05, - "loss": 0.0158, - "step": 226 - }, - { - "epoch": 3.404494382022472, - "grad_norm": 0.23152689635753632, - "learning_rate": 3.356060606060606e-05, - "loss": 0.0169, - "step": 228 - }, - { - "epoch": 3.4344569288389515, - "grad_norm": 0.20635737478733063, - "learning_rate": 3.3409090909090915e-05, - "loss": 0.0167, - "step": 230 - }, - { - "epoch": 3.464419475655431, - "grad_norm": 0.18343976140022278, - "learning_rate": 3.325757575757576e-05, - "loss": 0.0144, - "step": 232 - }, - { - "epoch": 3.49438202247191, - "grad_norm": 0.14534398913383484, - "learning_rate": 3.310606060606061e-05, - "loss": 0.0138, - "step": 234 - }, - { - "epoch": 3.5243445692883895, - "grad_norm": 0.2103314846754074, - "learning_rate": 3.295454545454545e-05, - "loss": 0.0164, - "step": 236 - }, - { - "epoch": 3.554307116104869, - "grad_norm": 0.13487893342971802, - "learning_rate": 3.2803030303030305e-05, - "loss": 0.0149, - "step": 238 - }, - { - "epoch": 3.5842696629213484, - "grad_norm": 0.13789679110050201, - "learning_rate": 3.265151515151516e-05, - "loss": 0.0137, - "step": 240 - }, - { - "epoch": 3.6142322097378274, - "grad_norm": 0.21411816775798798, - "learning_rate": 3.2500000000000004e-05, - "loss": 0.0181, - "step": 242 - }, - { - "epoch": 3.644194756554307, - "grad_norm": 0.22373120486736298, - "learning_rate": 3.234848484848485e-05, - "loss": 0.0158, - "step": 244 - }, - { - "epoch": 3.6741573033707864, - "grad_norm": 0.17626555263996124, - "learning_rate": 3.2196969696969696e-05, - "loss": 0.0158, - "step": 246 - }, - { - "epoch": 3.704119850187266, - "grad_norm": 0.15838968753814697, - "learning_rate": 3.204545454545455e-05, - "loss": 0.0149, - "step": 248 - }, - { - "epoch": 3.7340823970037453, - "grad_norm": 0.15824346244335175, - "learning_rate": 3.18939393939394e-05, - "loss": 0.0151, - "step": 250 - }, - { - "epoch": 3.764044943820225, - "grad_norm": 0.1346617490053177, - "learning_rate": 3.174242424242425e-05, - "loss": 0.0155, - "step": 252 - }, - { - "epoch": 3.7940074906367043, - "grad_norm": 0.15259407460689545, - "learning_rate": 3.159090909090909e-05, - "loss": 0.0153, - "step": 254 - }, - { - "epoch": 3.8239700374531838, - "grad_norm": 0.10529737174510956, - "learning_rate": 3.143939393939394e-05, - "loss": 0.0131, - "step": 256 - }, - { - "epoch": 3.853932584269663, - "grad_norm": 0.21343955397605896, - "learning_rate": 3.128787878787879e-05, - "loss": 0.0191, - "step": 258 - }, - { - "epoch": 3.8838951310861423, - "grad_norm": 0.19346514344215393, - "learning_rate": 3.113636363636364e-05, - "loss": 0.0166, - "step": 260 - }, - { - "epoch": 3.9138576779026217, - "grad_norm": 0.11122281849384308, - "learning_rate": 3.098484848484849e-05, - "loss": 0.0146, - "step": 262 - }, - { - "epoch": 3.943820224719101, - "grad_norm": 0.1087551936507225, - "learning_rate": 3.0833333333333335e-05, - "loss": 0.0152, - "step": 264 - }, - { - "epoch": 3.9737827715355807, - "grad_norm": 0.10952229797840118, - "learning_rate": 3.068181818181818e-05, - "loss": 0.0131, - "step": 266 - }, - { - "epoch": 4.0, - "grad_norm": 0.21926254034042358, - "learning_rate": 3.0530303030303034e-05, - "loss": 0.0183, - "step": 268 - }, - { - "epoch": 4.0299625468164795, - "grad_norm": 0.11138854920864105, - "learning_rate": 3.037878787878788e-05, - "loss": 0.0157, - "step": 270 - }, - { - "epoch": 4.059925093632959, - "grad_norm": 0.18572665750980377, - "learning_rate": 3.0227272727272725e-05, - "loss": 0.0143, - "step": 272 - }, - { - "epoch": 4.089887640449438, - "grad_norm": 0.11445632576942444, - "learning_rate": 3.0075757575757578e-05, - "loss": 0.014, - "step": 274 - }, - { - "epoch": 4.119850187265918, - "grad_norm": 0.13404381275177002, - "learning_rate": 2.9924242424242427e-05, - "loss": 0.016, - "step": 276 - }, - { - "epoch": 4.149812734082397, - "grad_norm": 0.22214095294475555, - "learning_rate": 2.9772727272727273e-05, - "loss": 0.0158, - "step": 278 - }, - { - "epoch": 4.179775280898877, - "grad_norm": 0.16340972483158112, - "learning_rate": 2.9621212121212122e-05, - "loss": 0.0168, - "step": 280 - }, - { - "epoch": 4.209737827715355, - "grad_norm": 0.1711144745349884, - "learning_rate": 2.9469696969696968e-05, - "loss": 0.0152, - "step": 282 - }, - { - "epoch": 4.239700374531835, - "grad_norm": 0.13778892159461975, - "learning_rate": 2.9318181818181817e-05, - "loss": 0.0164, - "step": 284 - }, - { - "epoch": 4.269662921348314, - "grad_norm": 0.15408281981945038, - "learning_rate": 2.916666666666667e-05, - "loss": 0.0174, - "step": 286 - }, - { - "epoch": 4.299625468164794, - "grad_norm": 0.12467946112155914, - "learning_rate": 2.901515151515152e-05, - "loss": 0.0152, - "step": 288 - }, - { - "epoch": 4.329588014981273, - "grad_norm": 0.1988392323255539, - "learning_rate": 2.8863636363636365e-05, - "loss": 0.0159, - "step": 290 - }, - { - "epoch": 4.359550561797753, - "grad_norm": 0.1968161165714264, - "learning_rate": 2.8712121212121214e-05, - "loss": 0.0164, - "step": 292 - }, - { - "epoch": 4.389513108614232, - "grad_norm": 0.16299834847450256, - "learning_rate": 2.856060606060606e-05, - "loss": 0.0126, - "step": 294 - }, - { - "epoch": 4.419475655430712, - "grad_norm": 0.20078356564044952, - "learning_rate": 2.8409090909090912e-05, - "loss": 0.0167, - "step": 296 - }, - { - "epoch": 4.449438202247191, - "grad_norm": 0.10105714946985245, - "learning_rate": 2.825757575757576e-05, - "loss": 0.0161, - "step": 298 - }, - { - "epoch": 4.479400749063671, - "grad_norm": 0.14914706349372864, - "learning_rate": 2.8106060606060607e-05, - "loss": 0.0172, - "step": 300 - }, - { - "epoch": 4.50936329588015, - "grad_norm": 0.12265647202730179, - "learning_rate": 2.7954545454545457e-05, - "loss": 0.0158, - "step": 302 - }, - { - "epoch": 4.539325842696629, - "grad_norm": 0.11637593805789948, - "learning_rate": 2.7803030303030303e-05, - "loss": 0.0151, - "step": 304 - }, - { - "epoch": 4.569288389513108, - "grad_norm": 0.11158251017332077, - "learning_rate": 2.7651515151515152e-05, - "loss": 0.0166, - "step": 306 - }, - { - "epoch": 4.599250936329588, - "grad_norm": 0.14547735452651978, - "learning_rate": 2.7500000000000004e-05, - "loss": 0.0157, - "step": 308 - }, - { - "epoch": 4.629213483146067, - "grad_norm": 0.15811274945735931, - "learning_rate": 2.734848484848485e-05, - "loss": 0.0145, - "step": 310 - }, - { - "epoch": 4.659176029962547, - "grad_norm": 0.1495320349931717, - "learning_rate": 2.71969696969697e-05, - "loss": 0.0165, - "step": 312 - }, - { - "epoch": 4.689138576779026, - "grad_norm": 0.14297890663146973, - "learning_rate": 2.7045454545454545e-05, - "loss": 0.016, - "step": 314 - }, - { - "epoch": 4.719101123595506, - "grad_norm": 0.1328546106815338, - "learning_rate": 2.6893939393939394e-05, - "loss": 0.0159, - "step": 316 - }, - { - "epoch": 4.749063670411985, - "grad_norm": 0.15500609576702118, - "learning_rate": 2.674242424242424e-05, - "loss": 0.0141, - "step": 318 - }, - { - "epoch": 4.7790262172284645, - "grad_norm": 0.1334412395954132, - "learning_rate": 2.6590909090909093e-05, - "loss": 0.014, - "step": 320 - }, - { - "epoch": 4.808988764044944, - "grad_norm": 0.11698229610919952, - "learning_rate": 2.6439393939393942e-05, - "loss": 0.0128, - "step": 322 - }, - { - "epoch": 4.8389513108614235, - "grad_norm": 0.10686776041984558, - "learning_rate": 2.6287878787878788e-05, - "loss": 0.0115, - "step": 324 - }, - { - "epoch": 4.868913857677903, - "grad_norm": 0.13976678252220154, - "learning_rate": 2.6136363636363637e-05, - "loss": 0.0154, - "step": 326 - }, - { - "epoch": 4.898876404494382, - "grad_norm": 0.1366475522518158, - "learning_rate": 2.5984848484848483e-05, - "loss": 0.0155, - "step": 328 - }, - { - "epoch": 4.928838951310862, - "grad_norm": 0.16024161875247955, - "learning_rate": 2.5833333333333336e-05, - "loss": 0.0186, - "step": 330 - }, - { - "epoch": 4.9588014981273405, - "grad_norm": 0.11987770348787308, - "learning_rate": 2.5681818181818185e-05, - "loss": 0.0151, - "step": 332 - }, - { - "epoch": 4.98876404494382, - "grad_norm": 0.12122765928506851, - "learning_rate": 2.553030303030303e-05, - "loss": 0.0129, - "step": 334 - }, - { - "epoch": 5.01498127340824, - "grad_norm": 0.14996588230133057, - "learning_rate": 2.537878787878788e-05, - "loss": 0.0134, - "step": 336 - }, - { - "epoch": 5.044943820224719, - "grad_norm": 0.1114385575056076, - "learning_rate": 2.5227272727272726e-05, - "loss": 0.0138, - "step": 338 - }, - { - "epoch": 5.074906367041199, - "grad_norm": 0.11765974760055542, - "learning_rate": 2.5075757575757575e-05, - "loss": 0.013, - "step": 340 - }, - { - "epoch": 5.104868913857678, - "grad_norm": 0.18139192461967468, - "learning_rate": 2.4924242424242424e-05, - "loss": 0.0151, - "step": 342 - }, - { - "epoch": 5.134831460674158, - "grad_norm": 0.12986940145492554, - "learning_rate": 2.4772727272727277e-05, - "loss": 0.0149, - "step": 344 - }, - { - "epoch": 5.164794007490637, - "grad_norm": 0.12934552133083344, - "learning_rate": 2.4621212121212123e-05, - "loss": 0.0145, - "step": 346 - }, - { - "epoch": 5.194756554307116, - "grad_norm": 0.13571321964263916, - "learning_rate": 2.4469696969696972e-05, - "loss": 0.0175, - "step": 348 - }, - { - "epoch": 5.224719101123595, - "grad_norm": 0.13134482502937317, - "learning_rate": 2.431818181818182e-05, - "loss": 0.0144, - "step": 350 - }, - { - "epoch": 5.254681647940075, - "grad_norm": 0.1282089650630951, - "learning_rate": 2.4166666666666667e-05, - "loss": 0.0168, - "step": 352 - }, - { - "epoch": 5.284644194756554, - "grad_norm": 0.12280598282814026, - "learning_rate": 2.4015151515151516e-05, - "loss": 0.0133, - "step": 354 - }, - { - "epoch": 5.314606741573034, - "grad_norm": 0.1045297384262085, - "learning_rate": 2.3863636363636365e-05, - "loss": 0.0135, - "step": 356 - }, - { - "epoch": 5.344569288389513, - "grad_norm": 0.10783874243497849, - "learning_rate": 2.3712121212121214e-05, - "loss": 0.0147, - "step": 358 - }, - { - "epoch": 5.3745318352059925, - "grad_norm": 0.28152090311050415, - "learning_rate": 2.356060606060606e-05, - "loss": 0.0155, - "step": 360 - }, - { - "epoch": 5.404494382022472, - "grad_norm": 0.12164758890867233, - "learning_rate": 2.340909090909091e-05, - "loss": 0.0138, - "step": 362 - }, - { - "epoch": 5.4344569288389515, - "grad_norm": 0.16581693291664124, - "learning_rate": 2.325757575757576e-05, - "loss": 0.0179, - "step": 364 - }, - { - "epoch": 5.464419475655431, - "grad_norm": 0.10845212638378143, - "learning_rate": 2.3106060606060605e-05, - "loss": 0.0144, - "step": 366 - }, - { - "epoch": 5.49438202247191, - "grad_norm": 0.11847496032714844, - "learning_rate": 2.2954545454545457e-05, - "loss": 0.0132, - "step": 368 - }, - { - "epoch": 5.52434456928839, - "grad_norm": 0.1990269422531128, - "learning_rate": 2.2803030303030303e-05, - "loss": 0.0179, - "step": 370 - }, - { - "epoch": 5.554307116104869, - "grad_norm": 0.13564549386501312, - "learning_rate": 2.2651515151515152e-05, - "loss": 0.0155, - "step": 372 - }, - { - "epoch": 5.584269662921348, - "grad_norm": 0.12416112422943115, - "learning_rate": 2.25e-05, - "loss": 0.0142, - "step": 374 - }, - { - "epoch": 5.614232209737827, - "grad_norm": 0.21298328042030334, - "learning_rate": 2.234848484848485e-05, - "loss": 0.0179, - "step": 376 - }, - { - "epoch": 5.644194756554307, - "grad_norm": 0.12686102092266083, - "learning_rate": 2.21969696969697e-05, - "loss": 0.0142, - "step": 378 - }, - { - "epoch": 5.674157303370786, - "grad_norm": 0.11951331794261932, - "learning_rate": 2.2045454545454546e-05, - "loss": 0.0158, - "step": 380 - }, - { - "epoch": 5.704119850187266, - "grad_norm": 0.15928389132022858, - "learning_rate": 2.1893939393939395e-05, - "loss": 0.0153, - "step": 382 - }, - { - "epoch": 5.734082397003745, - "grad_norm": 0.11539971828460693, - "learning_rate": 2.1742424242424244e-05, - "loss": 0.0132, - "step": 384 - }, - { - "epoch": 5.764044943820225, - "grad_norm": 0.1547948122024536, - "learning_rate": 2.1590909090909093e-05, - "loss": 0.0132, - "step": 386 - }, - { - "epoch": 5.794007490636704, - "grad_norm": 0.15363723039627075, - "learning_rate": 2.143939393939394e-05, - "loss": 0.0188, - "step": 388 - }, - { - "epoch": 5.823970037453184, - "grad_norm": 0.13726523518562317, - "learning_rate": 2.128787878787879e-05, - "loss": 0.0155, - "step": 390 - }, - { - "epoch": 5.853932584269663, - "grad_norm": 0.11726450175046921, - "learning_rate": 2.1136363636363638e-05, - "loss": 0.0154, - "step": 392 - }, - { - "epoch": 5.883895131086143, - "grad_norm": 0.09192411601543427, - "learning_rate": 2.0984848484848483e-05, - "loss": 0.0144, - "step": 394 - }, - { - "epoch": 5.913857677902621, - "grad_norm": 0.12325743585824966, - "learning_rate": 2.0833333333333336e-05, - "loss": 0.0138, - "step": 396 - }, - { - "epoch": 5.943820224719101, - "grad_norm": 0.1755753755569458, - "learning_rate": 2.0681818181818182e-05, - "loss": 0.0159, - "step": 398 - }, - { - "epoch": 5.97378277153558, - "grad_norm": 0.1385994851589203, - "learning_rate": 2.053030303030303e-05, - "loss": 0.016, - "step": 400 - }, - { - "epoch": 6.0, - "grad_norm": 0.12166863679885864, - "learning_rate": 2.037878787878788e-05, - "loss": 0.0137, - "step": 402 - }, - { - "epoch": 6.0299625468164795, - "grad_norm": 0.135303795337677, - "learning_rate": 2.022727272727273e-05, - "loss": 0.0146, - "step": 404 - }, - { - "epoch": 6.059925093632959, - "grad_norm": 0.14891491830348969, - "learning_rate": 2.0075757575757575e-05, - "loss": 0.016, - "step": 406 - }, - { - "epoch": 6.089887640449438, - "grad_norm": 0.13520725071430206, - "learning_rate": 1.9924242424242425e-05, - "loss": 0.017, - "step": 408 - }, - { - "epoch": 6.119850187265918, - "grad_norm": 0.1810332089662552, - "learning_rate": 1.9772727272727274e-05, - "loss": 0.0161, - "step": 410 - }, - { - "epoch": 6.149812734082397, - "grad_norm": 0.12100538611412048, - "learning_rate": 1.962121212121212e-05, - "loss": 0.0143, - "step": 412 - }, - { - "epoch": 6.179775280898877, - "grad_norm": 0.11748135834932327, - "learning_rate": 1.9469696969696972e-05, - "loss": 0.0148, - "step": 414 - }, - { - "epoch": 6.209737827715355, - "grad_norm": 0.1747063398361206, - "learning_rate": 1.9318181818181818e-05, - "loss": 0.0153, - "step": 416 - }, - { - "epoch": 6.239700374531835, - "grad_norm": 0.15986011922359467, - "learning_rate": 1.9166666666666667e-05, - "loss": 0.0153, - "step": 418 - }, - { - "epoch": 6.269662921348314, - "grad_norm": 0.130351260304451, - "learning_rate": 1.9015151515151516e-05, - "loss": 0.0131, - "step": 420 - }, - { - "epoch": 6.299625468164794, - "grad_norm": 0.09828981012105942, - "learning_rate": 1.8863636363636362e-05, - "loss": 0.0132, - "step": 422 - }, - { - "epoch": 6.329588014981273, - "grad_norm": 0.11312004178762436, - "learning_rate": 1.8712121212121215e-05, - "loss": 0.0137, - "step": 424 - }, - { - "epoch": 6.359550561797753, - "grad_norm": 0.1399003267288208, - "learning_rate": 1.856060606060606e-05, - "loss": 0.0155, - "step": 426 - }, - { - "epoch": 6.389513108614232, - "grad_norm": 0.19829927384853363, - "learning_rate": 1.840909090909091e-05, - "loss": 0.0181, - "step": 428 - }, - { - "epoch": 6.419475655430712, - "grad_norm": 0.12163085490465164, - "learning_rate": 1.825757575757576e-05, - "loss": 0.0143, - "step": 430 - }, - { - "epoch": 6.449438202247191, - "grad_norm": 0.1325439214706421, - "learning_rate": 1.810606060606061e-05, - "loss": 0.0147, - "step": 432 - }, - { - "epoch": 6.479400749063671, - "grad_norm": 0.10004521906375885, - "learning_rate": 1.7954545454545454e-05, - "loss": 0.0126, - "step": 434 - }, - { - "epoch": 6.50936329588015, - "grad_norm": 0.22187596559524536, - "learning_rate": 1.7803030303030303e-05, - "loss": 0.0162, - "step": 436 - }, - { - "epoch": 6.539325842696629, - "grad_norm": 0.12271460145711899, - "learning_rate": 1.7651515151515153e-05, - "loss": 0.013, - "step": 438 - }, - { - "epoch": 6.569288389513108, - "grad_norm": 0.16459853947162628, - "learning_rate": 1.75e-05, - "loss": 0.0145, - "step": 440 - }, - { - "epoch": 6.599250936329588, - "grad_norm": 0.11979979276657104, - "learning_rate": 1.734848484848485e-05, - "loss": 0.0135, - "step": 442 - }, - { - "epoch": 6.629213483146067, - "grad_norm": 0.24495816230773926, - "learning_rate": 1.7196969696969697e-05, - "loss": 0.0164, - "step": 444 - }, - { - "epoch": 6.659176029962547, - "grad_norm": 0.1692766696214676, - "learning_rate": 1.7045454545454546e-05, - "loss": 0.0152, - "step": 446 - }, - { - "epoch": 6.689138576779026, - "grad_norm": 0.1197497546672821, - "learning_rate": 1.6893939393939395e-05, - "loss": 0.0153, - "step": 448 - }, - { - "epoch": 6.719101123595506, - "grad_norm": 0.09570708870887756, - "learning_rate": 1.674242424242424e-05, - "loss": 0.0134, - "step": 450 - }, - { - "epoch": 6.749063670411985, - "grad_norm": 0.10233695805072784, - "learning_rate": 1.6590909090909094e-05, - "loss": 0.0135, - "step": 452 - }, - { - "epoch": 6.7790262172284645, - "grad_norm": 0.13967226445674896, - "learning_rate": 1.643939393939394e-05, - "loss": 0.0159, - "step": 454 - }, - { - "epoch": 6.808988764044944, - "grad_norm": 0.14382530748844147, - "learning_rate": 1.628787878787879e-05, - "loss": 0.0141, - "step": 456 - }, - { - "epoch": 6.8389513108614235, - "grad_norm": 0.18934357166290283, - "learning_rate": 1.6136363636363638e-05, - "loss": 0.015, - "step": 458 - }, - { - "epoch": 6.868913857677903, - "grad_norm": 0.09580834209918976, - "learning_rate": 1.5984848484848487e-05, - "loss": 0.0141, - "step": 460 - }, - { - "epoch": 6.898876404494382, - "grad_norm": 0.16412261128425598, - "learning_rate": 1.5833333333333333e-05, - "loss": 0.0157, - "step": 462 - }, - { - "epoch": 6.928838951310862, - "grad_norm": 0.16064710915088654, - "learning_rate": 1.5681818181818182e-05, - "loss": 0.0141, - "step": 464 - }, - { - "epoch": 6.9588014981273405, - "grad_norm": 0.17852792143821716, - "learning_rate": 1.553030303030303e-05, - "loss": 0.0156, - "step": 466 - }, - { - "epoch": 6.98876404494382, - "grad_norm": 0.17345724999904633, - "learning_rate": 1.5378787878787877e-05, - "loss": 0.0169, - "step": 468 - }, - { - "epoch": 7.01498127340824, - "grad_norm": 0.1276731640100479, - "learning_rate": 1.5227272727272728e-05, - "loss": 0.0143, - "step": 470 - }, - { - "epoch": 7.044943820224719, - "grad_norm": 0.1590038686990738, - "learning_rate": 1.5075757575757576e-05, - "loss": 0.0158, - "step": 472 - }, - { - "epoch": 7.074906367041199, - "grad_norm": 0.12557406723499298, - "learning_rate": 1.4924242424242423e-05, - "loss": 0.0136, - "step": 474 - }, - { - "epoch": 7.104868913857678, - "grad_norm": 0.21979106962680817, - "learning_rate": 1.4772727272727274e-05, - "loss": 0.0159, - "step": 476 - }, - { - "epoch": 7.134831460674158, - "grad_norm": 0.19031833112239838, - "learning_rate": 1.4621212121212122e-05, - "loss": 0.0155, - "step": 478 - }, - { - "epoch": 7.164794007490637, - "grad_norm": 0.23122136294841766, - "learning_rate": 1.446969696969697e-05, - "loss": 0.0171, - "step": 480 - }, - { - "epoch": 7.194756554307116, - "grad_norm": 0.12799808382987976, - "learning_rate": 1.431818181818182e-05, - "loss": 0.0142, - "step": 482 - }, - { - "epoch": 7.224719101123595, - "grad_norm": 0.09180168062448502, - "learning_rate": 1.4166666666666668e-05, - "loss": 0.0133, - "step": 484 - }, - { - "epoch": 7.254681647940075, - "grad_norm": 0.13481584191322327, - "learning_rate": 1.4015151515151515e-05, - "loss": 0.014, - "step": 486 - }, - { - "epoch": 7.284644194756554, - "grad_norm": 0.1296118050813675, - "learning_rate": 1.3863636363636364e-05, - "loss": 0.015, - "step": 488 - }, - { - "epoch": 7.314606741573034, - "grad_norm": 0.12827853858470917, - "learning_rate": 1.3712121212121212e-05, - "loss": 0.0143, - "step": 490 - }, - { - "epoch": 7.344569288389513, - "grad_norm": 0.12804710865020752, - "learning_rate": 1.3560606060606063e-05, - "loss": 0.0144, - "step": 492 - }, - { - "epoch": 7.3745318352059925, - "grad_norm": 0.14896075427532196, - "learning_rate": 1.340909090909091e-05, - "loss": 0.0145, - "step": 494 - }, - { - "epoch": 7.404494382022472, - "grad_norm": 0.11220405250787735, - "learning_rate": 1.3257575757575758e-05, - "loss": 0.0156, - "step": 496 - }, - { - "epoch": 7.4344569288389515, - "grad_norm": 0.10928992182016373, - "learning_rate": 1.3106060606060607e-05, - "loss": 0.0141, - "step": 498 - }, - { - "epoch": 7.464419475655431, - "grad_norm": 0.12223732471466064, - "learning_rate": 1.2954545454545455e-05, - "loss": 0.0146, - "step": 500 - }, - { - "epoch": 7.49438202247191, - "grad_norm": 0.12335951626300812, - "learning_rate": 1.2803030303030302e-05, - "loss": 0.0141, - "step": 502 - }, - { - "epoch": 7.52434456928839, - "grad_norm": 0.12804357707500458, - "learning_rate": 1.2651515151515153e-05, - "loss": 0.0128, - "step": 504 - }, - { - "epoch": 7.554307116104869, - "grad_norm": 0.11515405774116516, - "learning_rate": 1.25e-05, - "loss": 0.0128, - "step": 506 - }, - { - "epoch": 7.584269662921348, - "grad_norm": 0.1302037239074707, - "learning_rate": 1.234848484848485e-05, - "loss": 0.0161, - "step": 508 - }, - { - "epoch": 7.614232209737827, - "grad_norm": 0.1083846315741539, - "learning_rate": 1.2196969696969697e-05, - "loss": 0.0149, - "step": 510 - }, - { - "epoch": 7.644194756554307, - "grad_norm": 0.11610821634531021, - "learning_rate": 1.2045454545454547e-05, - "loss": 0.0138, - "step": 512 - }, - { - "epoch": 7.674157303370786, - "grad_norm": 0.08522523194551468, - "learning_rate": 1.1893939393939394e-05, - "loss": 0.0132, - "step": 514 - }, - { - "epoch": 7.704119850187266, - "grad_norm": 0.0896814838051796, - "learning_rate": 1.1742424242424243e-05, - "loss": 0.0131, - "step": 516 - }, - { - "epoch": 7.734082397003745, - "grad_norm": 0.15006373822689056, - "learning_rate": 1.159090909090909e-05, - "loss": 0.0162, - "step": 518 - }, - { - "epoch": 7.764044943820225, - "grad_norm": 0.1151801198720932, - "learning_rate": 1.143939393939394e-05, - "loss": 0.015, - "step": 520 - }, - { - "epoch": 7.794007490636704, - "grad_norm": 0.10852045565843582, - "learning_rate": 1.128787878787879e-05, - "loss": 0.0132, - "step": 522 - }, - { - "epoch": 7.823970037453184, - "grad_norm": 0.20251013338565826, - "learning_rate": 1.1136363636363637e-05, - "loss": 0.0144, - "step": 524 - }, - { - "epoch": 7.853932584269663, - "grad_norm": 0.1366216391324997, - "learning_rate": 1.0984848484848486e-05, - "loss": 0.0157, - "step": 526 - }, - { - "epoch": 7.883895131086143, - "grad_norm": 0.12531571090221405, - "learning_rate": 1.0833333333333334e-05, - "loss": 0.015, - "step": 528 - }, - { - "epoch": 7.913857677902621, - "grad_norm": 0.14330235123634338, - "learning_rate": 1.0681818181818181e-05, - "loss": 0.0125, - "step": 530 - }, - { - "epoch": 7.943820224719101, - "grad_norm": 0.1384236365556717, - "learning_rate": 1.053030303030303e-05, - "loss": 0.014, - "step": 532 - }, - { - "epoch": 7.97378277153558, - "grad_norm": 0.15142616629600525, - "learning_rate": 1.037878787878788e-05, - "loss": 0.0175, - "step": 534 - }, - { - "epoch": 8.0, - "grad_norm": 0.11959421634674072, - "learning_rate": 1.0227272727272729e-05, - "loss": 0.0144, - "step": 536 - }, - { - "epoch": 8.02996254681648, - "grad_norm": 0.14299067854881287, - "learning_rate": 1.0075757575757576e-05, - "loss": 0.0121, - "step": 538 - }, - { - "epoch": 8.059925093632959, - "grad_norm": 0.15421266853809357, - "learning_rate": 9.924242424242425e-06, - "loss": 0.0157, - "step": 540 - }, - { - "epoch": 8.089887640449438, - "grad_norm": 0.11853787302970886, - "learning_rate": 9.772727272727273e-06, - "loss": 0.0141, - "step": 542 - }, - { - "epoch": 8.119850187265918, - "grad_norm": 0.11856868863105774, - "learning_rate": 9.62121212121212e-06, - "loss": 0.0145, - "step": 544 - }, - { - "epoch": 8.149812734082397, - "grad_norm": 0.10615833103656769, - "learning_rate": 9.46969696969697e-06, - "loss": 0.0146, - "step": 546 - }, - { - "epoch": 8.179775280898877, - "grad_norm": 0.14076802134513855, - "learning_rate": 9.318181818181819e-06, - "loss": 0.014, - "step": 548 - }, - { - "epoch": 8.209737827715356, - "grad_norm": 0.12974779307842255, - "learning_rate": 9.166666666666666e-06, - "loss": 0.0139, - "step": 550 - }, - { - "epoch": 8.239700374531836, - "grad_norm": 0.11985109746456146, - "learning_rate": 9.015151515151516e-06, - "loss": 0.0148, - "step": 552 - }, - { - "epoch": 8.269662921348315, - "grad_norm": 0.21702255308628082, - "learning_rate": 8.863636363636365e-06, - "loss": 0.0134, - "step": 554 - }, - { - "epoch": 8.299625468164795, - "grad_norm": 0.14472782611846924, - "learning_rate": 8.712121212121212e-06, - "loss": 0.0151, - "step": 556 - }, - { - "epoch": 8.329588014981274, - "grad_norm": 0.1377476155757904, - "learning_rate": 8.56060606060606e-06, - "loss": 0.0139, - "step": 558 - }, - { - "epoch": 8.359550561797754, - "grad_norm": 0.15174493193626404, - "learning_rate": 8.409090909090909e-06, - "loss": 0.0129, - "step": 560 - }, - { - "epoch": 8.389513108614231, - "grad_norm": 0.1872493326663971, - "learning_rate": 8.257575757575758e-06, - "loss": 0.0119, - "step": 562 - }, - { - "epoch": 8.41947565543071, - "grad_norm": 0.1931612491607666, - "learning_rate": 8.106060606060606e-06, - "loss": 0.0175, - "step": 564 - }, - { - "epoch": 8.44943820224719, - "grad_norm": 0.18126638233661652, - "learning_rate": 7.954545454545455e-06, - "loss": 0.0153, - "step": 566 - }, - { - "epoch": 8.47940074906367, - "grad_norm": 0.16266301274299622, - "learning_rate": 7.803030303030304e-06, - "loss": 0.0143, - "step": 568 - }, - { - "epoch": 8.50936329588015, - "grad_norm": 0.1196930930018425, - "learning_rate": 7.651515151515152e-06, - "loss": 0.015, - "step": 570 - }, - { - "epoch": 8.539325842696629, - "grad_norm": 0.14626623690128326, - "learning_rate": 7.5e-06, - "loss": 0.0154, - "step": 572 - }, - { - "epoch": 8.569288389513108, - "grad_norm": 0.12978863716125488, - "learning_rate": 7.3484848484848486e-06, - "loss": 0.0141, - "step": 574 - }, - { - "epoch": 8.599250936329588, - "grad_norm": 0.16399815678596497, - "learning_rate": 7.196969696969698e-06, - "loss": 0.0151, - "step": 576 - }, - { - "epoch": 8.629213483146067, - "grad_norm": 0.13139177858829498, - "learning_rate": 7.045454545454545e-06, - "loss": 0.0132, - "step": 578 - }, - { - "epoch": 8.659176029962547, - "grad_norm": 0.10869726538658142, - "learning_rate": 6.8939393939393945e-06, - "loss": 0.0143, - "step": 580 - }, - { - "epoch": 8.689138576779026, - "grad_norm": 0.11966531723737717, - "learning_rate": 6.742424242424243e-06, - "loss": 0.0153, - "step": 582 - }, - { - "epoch": 8.719101123595506, - "grad_norm": 0.15545986592769623, - "learning_rate": 6.59090909090909e-06, - "loss": 0.0133, - "step": 584 - }, - { - "epoch": 8.749063670411985, - "grad_norm": 0.1065947413444519, - "learning_rate": 6.43939393939394e-06, - "loss": 0.0146, - "step": 586 - }, - { - "epoch": 8.779026217228465, - "grad_norm": 0.17026154696941376, - "learning_rate": 6.287878787878789e-06, - "loss": 0.0152, - "step": 588 - }, - { - "epoch": 8.808988764044944, - "grad_norm": 0.1124507337808609, - "learning_rate": 6.136363636363636e-06, - "loss": 0.0143, - "step": 590 - }, - { - "epoch": 8.838951310861423, - "grad_norm": 0.11526783555746078, - "learning_rate": 5.984848484848485e-06, - "loss": 0.0143, - "step": 592 - }, - { - "epoch": 8.868913857677903, - "grad_norm": 0.11815094202756882, - "learning_rate": 5.833333333333334e-06, - "loss": 0.0142, - "step": 594 - }, - { - "epoch": 8.898876404494382, - "grad_norm": 0.115007683634758, - "learning_rate": 5.681818181818182e-06, - "loss": 0.0148, - "step": 596 - }, - { - "epoch": 8.928838951310862, - "grad_norm": 0.12640294432640076, - "learning_rate": 5.530303030303031e-06, - "loss": 0.0156, - "step": 598 - }, - { - "epoch": 8.958801498127341, - "grad_norm": 0.18849849700927734, - "learning_rate": 5.378787878787879e-06, - "loss": 0.0146, - "step": 600 - }, - { - "epoch": 8.98876404494382, - "grad_norm": 0.09228724241256714, - "learning_rate": 5.2272727272727274e-06, - "loss": 0.0132, - "step": 602 - }, - { - "epoch": 9.014981273408239, - "grad_norm": 0.10718464851379395, - "learning_rate": 5.075757575757576e-06, - "loss": 0.0142, - "step": 604 - }, - { - "epoch": 9.044943820224718, - "grad_norm": 0.15939927101135254, - "learning_rate": 4.924242424242424e-06, - "loss": 0.015, - "step": 606 - }, - { - "epoch": 9.074906367041198, - "grad_norm": 0.10619861632585526, - "learning_rate": 4.772727272727273e-06, - "loss": 0.0139, - "step": 608 - }, - { - "epoch": 9.104868913857677, - "grad_norm": 0.1796608418226242, - "learning_rate": 4.621212121212122e-06, - "loss": 0.0157, - "step": 610 - }, - { - "epoch": 9.134831460674157, - "grad_norm": 0.11252462863922119, - "learning_rate": 4.46969696969697e-06, - "loss": 0.0152, - "step": 612 - }, - { - "epoch": 9.164794007490636, - "grad_norm": 0.12761755287647247, - "learning_rate": 4.3181818181818185e-06, - "loss": 0.0143, - "step": 614 - }, - { - "epoch": 9.194756554307116, - "grad_norm": 0.11963596194982529, - "learning_rate": 4.166666666666667e-06, - "loss": 0.0128, - "step": 616 - }, - { - "epoch": 9.224719101123595, - "grad_norm": 0.14573565125465393, - "learning_rate": 4.015151515151515e-06, - "loss": 0.0156, - "step": 618 - }, - { - "epoch": 9.254681647940075, - "grad_norm": 0.1269391030073166, - "learning_rate": 3.863636363636364e-06, - "loss": 0.0135, - "step": 620 - }, - { - "epoch": 9.284644194756554, - "grad_norm": 0.14515313506126404, - "learning_rate": 3.7121212121212124e-06, - "loss": 0.0151, - "step": 622 - }, - { - "epoch": 9.314606741573034, - "grad_norm": 0.1334671527147293, - "learning_rate": 3.5606060606060608e-06, - "loss": 0.015, - "step": 624 - }, - { - "epoch": 9.344569288389513, - "grad_norm": 0.12002371996641159, - "learning_rate": 3.409090909090909e-06, - "loss": 0.0141, - "step": 626 - }, - { - "epoch": 9.374531835205993, - "grad_norm": 0.14389854669570923, - "learning_rate": 3.257575757575758e-06, - "loss": 0.0136, - "step": 628 - }, - { - "epoch": 9.404494382022472, - "grad_norm": 0.11606308072805405, - "learning_rate": 3.106060606060606e-06, - "loss": 0.0152, - "step": 630 - }, - { - "epoch": 9.434456928838951, - "grad_norm": 0.18154092133045197, - "learning_rate": 2.9545454545454547e-06, - "loss": 0.0137, - "step": 632 - }, - { - "epoch": 9.464419475655431, - "grad_norm": 0.2114187330007553, - "learning_rate": 2.803030303030303e-06, - "loss": 0.0145, - "step": 634 - }, - { - "epoch": 9.49438202247191, - "grad_norm": 0.12829148769378662, - "learning_rate": 2.651515151515152e-06, - "loss": 0.0137, - "step": 636 - }, - { - "epoch": 9.52434456928839, - "grad_norm": 0.14554138481616974, - "learning_rate": 2.5e-06, - "loss": 0.0122, - "step": 638 - }, - { - "epoch": 9.55430711610487, - "grad_norm": 0.11862059682607651, - "learning_rate": 2.3484848484848486e-06, - "loss": 0.0137, - "step": 640 - }, - { - "epoch": 9.584269662921349, - "grad_norm": 0.11568762362003326, - "learning_rate": 2.196969696969697e-06, - "loss": 0.0147, - "step": 642 - }, - { - "epoch": 9.614232209737828, - "grad_norm": 0.10297347605228424, - "learning_rate": 2.0454545454545457e-06, - "loss": 0.0139, - "step": 644 - }, - { - "epoch": 9.644194756554308, - "grad_norm": 0.14339599013328552, - "learning_rate": 1.8939393939393941e-06, - "loss": 0.0138, - "step": 646 - }, - { - "epoch": 9.674157303370787, - "grad_norm": 0.15518030524253845, - "learning_rate": 1.7424242424242427e-06, - "loss": 0.0127, - "step": 648 - }, - { - "epoch": 9.704119850187267, - "grad_norm": 0.13086441159248352, - "learning_rate": 1.5909090909090908e-06, - "loss": 0.013, - "step": 650 - }, - { - "epoch": 9.734082397003746, - "grad_norm": 0.1343931257724762, - "learning_rate": 1.4393939393939396e-06, - "loss": 0.014, - "step": 652 - }, - { - "epoch": 9.764044943820224, - "grad_norm": 0.16945284605026245, - "learning_rate": 1.287878787878788e-06, - "loss": 0.0153, - "step": 654 - }, - { - "epoch": 9.794007490636703, - "grad_norm": 0.14710381627082825, - "learning_rate": 1.1363636363636364e-06, - "loss": 0.0145, - "step": 656 - }, - { - "epoch": 9.823970037453183, - "grad_norm": 0.15412309765815735, - "learning_rate": 9.848484848484847e-07, - "loss": 0.0147, - "step": 658 - }, - { - "epoch": 9.853932584269662, - "grad_norm": 0.1276121288537979, - "learning_rate": 8.333333333333333e-07, - "loss": 0.0157, - "step": 660 - } - ], - "logging_steps": 2, - "max_steps": 670, - "num_input_tokens_seen": 0, - "num_train_epochs": 10, - "save_steps": 20, - "stateful_callbacks": { - "TrainerControl": { - "args": { - "should_epoch_stop": false, - "should_evaluate": false, - "should_log": false, - "should_save": true, - "should_training_stop": false - }, - "attributes": {} - } - }, - "total_flos": 1.778197543133184e+16, - "train_batch_size": 1, - "trial_name": null, - "trial_params": null -} diff --git a/smollm3_robust/checkpoint-660/training_args.bin b/smollm3_robust/checkpoint-660/training_args.bin deleted file mode 100644 index 288236fd628e342cba9dba026dc49e62bfa02807..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/checkpoint-660/training_args.bin and /dev/null differ diff --git a/smollm3_robust/checkpoint-670/adapter_config.json b/smollm3_robust/checkpoint-670/adapter_config.json deleted file mode 100644 index 0be8c070aaf97a2debdeb19f27452147952b65b5..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-670/adapter_config.json +++ /dev/null @@ -1,41 +0,0 @@ -{ - "alpha_pattern": {}, - "auto_mapping": null, - "base_model_name_or_path": "HuggingFaceTB/SmolLM3-3B", - "bias": "none", - "corda_config": null, - "eva_config": null, - "exclude_modules": null, - "fan_in_fan_out": false, - "inference_mode": true, - "init_lora_weights": true, - "layer_replication": null, - "layers_pattern": null, - "layers_to_transform": null, - "loftq_config": {}, - "lora_alpha": 16, - "lora_bias": false, - "lora_dropout": 0.1, - "megatron_config": null, - "megatron_core": "megatron.core", - "modules_to_save": null, - "peft_type": "LORA", - "qalora_group_size": 16, - "r": 8, - "rank_pattern": {}, - "revision": null, - "target_modules": [ - "k_proj", - "down_proj", - "gate_proj", - "o_proj", - "q_proj", - "up_proj", - "v_proj" - ], - "task_type": "CAUSAL_LM", - "trainable_token_indices": null, - "use_dora": false, - "use_qalora": false, - "use_rslora": false -} \ No newline at end of file diff --git a/smollm3_robust/checkpoint-670/adapter_model.safetensors.REMOVED.git-id b/smollm3_robust/checkpoint-670/adapter_model.safetensors.REMOVED.git-id deleted file mode 100644 index cee193d4bd399f8d3b3b25a3e4aac7e10747f658..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-670/adapter_model.safetensors.REMOVED.git-id +++ /dev/null @@ -1 +0,0 @@ -e1ab487a3063f2ab75b74748b76635cee4549110 \ No newline at end of file diff --git a/smollm3_robust/checkpoint-670/chat_template.jinja b/smollm3_robust/checkpoint-670/chat_template.jinja deleted file mode 100644 index e01e3a1bca00ae47bca8326b38cc397729f87481..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-670/chat_template.jinja +++ /dev/null @@ -1,94 +0,0 @@ -{# ───── defaults ───── #} -{%- if enable_thinking is not defined -%} -{%- set enable_thinking = true -%} -{%- endif -%} - -{# ───── reasoning mode ───── #} -{%- if enable_thinking -%} - {%- set reasoning_mode = "/think" -%} -{%- else -%} - {%- set reasoning_mode = "/no_think" -%} -{%- endif -%} - -{# ───── header (system message) ───── #} -{{- "<|im_start|>system\n" -}} - -{%- if messages[0].role == "system" -%} - {%- set system_message = messages[0].content -%} - {%- if "/no_think" in system_message -%} - {%- set reasoning_mode = "/no_think" -%} - {%- elif "/think" in system_message -%} - {%- set reasoning_mode = "/think" -%} - {%- endif -%} - {%- set custom_instructions = system_message.replace("/no_think", "").replace("/think", "").rstrip() -%} -{%- endif -%} - -{%- if "/system_override" in system_message -%} - {{- custom_instructions.replace("/system_override", "").rstrip() -}} - {{- "<|im_end|>\n" -}} -{%- else -%} - {{- "## Metadata\n\n" -}} - {{- "Knowledge Cutoff Date: June 2025\n" -}} - {%- set today = strftime_now("%d %B %Y") -%} - {{- "Today Date: " ~ today ~ "\n" -}} - {{- "Reasoning Mode: " + reasoning_mode + "\n\n" -}} - - {{- "## Custom Instructions\n\n" -}} - {%- if custom_instructions -%} - {{- custom_instructions + "\n\n" -}} - {%- elif reasoning_mode == "/think" -%} - {{- "You are a helpful AI assistant named SmolLM, trained by Hugging Face. Your role as an assistant involves thoroughly exploring questions through a systematic thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracking, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution using the specified format: Thought section Solution section. In the Thought section, detail your reasoning process in steps. Each step should include detailed considerations such as analysing questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The Solution section should be logical, accurate, and concise and detail necessary steps needed to reach the conclusion.\n\n" -}} - {%- else -%} - {{- "You are a helpful AI assistant named SmolLM, trained by Hugging Face.\n\n" -}} - {%- endif -%} - - {%- if xml_tools or python_tools or tools -%} - {{- "### Tools\n\n" -}} - {%- if xml_tools or tools -%} - {%- if tools -%} - {%- set xml_tools = tools -%} - {%- endif -%} - {%- set ns = namespace(xml_tool_string="You may call one or more functions to assist with the user query.\nYou are provided with function signatures within XML tags:\n\n\n") -%} - {%- for tool in xml_tools[:] -%} {# The slicing makes sure that xml_tools is a list #} - {%- set ns.xml_tool_string = ns.xml_tool_string ~ (tool | string) ~ "\n" -%} - {%- endfor -%} - {%- set xml_tool_string = ns.xml_tool_string + "\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n" -%} - {{- xml_tool_string -}} - {%- endif -%} - {%- if python_tools -%} - {%- set ns = namespace(python_tool_string="When you send a message containing Python code between '' and '' tags, it will be executed in a stateful Jupyter notebook environment, and you will then be given the output to continued reasoning in an agentic loop.\n\nYou can use the following tools in your python code like regular functions:\n\n") -%} - {%- for tool in python_tools[:] -%} {# The slicing makes sure that python_tools is a list #} - {%- set ns.python_tool_string = ns.python_tool_string ~ (tool | string) ~ "\n" -%} - {%- endfor -%} - {%- set python_tool_string = ns.python_tool_string + "\n\nThe state persists between code executions: so variables that you define in one step are still available thereafter." -%} - {{- python_tool_string -}} - {%- endif -%} - {{- "\n\n" -}} - {{- "<|im_end|>\n" -}} - {%- endif -%} -{%- endif -%} -{# ───── main loop ───── #} -{%- for message in messages -%} - {%- set content = message.content if message.content is string else "" -%} - {%- if message.role == "user" -%} - {{ "<|im_start|>" + message.role + "\n" + content + "<|im_end|>\n" }} - {%- elif message.role == "assistant" -%} - {% generation %} - {%- if reasoning_mode == "/think" -%} - {{ "<|im_start|>assistant\n" + content.lstrip("\n") + "<|im_end|>\n" }} - {%- else -%} - {{ "<|im_start|>assistant\n" + "\n\n\n" + content.lstrip("\n") + "<|im_end|>\n" }} - {%- endif -%} - {% endgeneration %} - {%- elif message.role == "tool" -%} - {{ "<|im_start|>" + "user\n" + content + "<|im_end|>\n" }} - {%- endif -%} -{%- endfor -%} -{# ───── generation prompt ───── #} -{%- if add_generation_prompt -%} - {%- if reasoning_mode == "/think" -%} - {{ "<|im_start|>assistant\n" }} - {%- else -%} - {{ "<|im_start|>assistant\n" + "\n\n\n" }} - {%- endif -%} -{%- endif -%} \ No newline at end of file diff --git a/smollm3_robust/checkpoint-670/optimizer.pt.REMOVED.git-id b/smollm3_robust/checkpoint-670/optimizer.pt.REMOVED.git-id deleted file mode 100644 index 43a4cbbfac38020f29e45f6c18d60de22b3b4dc6..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-670/optimizer.pt.REMOVED.git-id +++ /dev/null @@ -1 +0,0 @@ -22e820dcbb30f4dca126abdfa89d5d3ba67a1427 \ No newline at end of file diff --git a/smollm3_robust/checkpoint-670/rng_state.pth b/smollm3_robust/checkpoint-670/rng_state.pth deleted file mode 100644 index 0676050097953d03d17d3687760f20bc5609de6f..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/checkpoint-670/rng_state.pth and /dev/null differ diff --git a/smollm3_robust/checkpoint-670/scheduler.pt b/smollm3_robust/checkpoint-670/scheduler.pt deleted file mode 100644 index 889acf1b30aaf4b747043b2144ea040b33d22e30..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/checkpoint-670/scheduler.pt and /dev/null differ diff --git a/smollm3_robust/checkpoint-670/special_tokens_map.json b/smollm3_robust/checkpoint-670/special_tokens_map.json deleted file mode 100644 index 190d5624dbbc1ad56f2f34c9d58e03fef7e5328b..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-670/special_tokens_map.json +++ /dev/null @@ -1,16 +0,0 @@ -{ - "eos_token": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false - }, - "pad_token": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false - } -} diff --git a/smollm3_robust/checkpoint-670/tokenizer.json.REMOVED.git-id b/smollm3_robust/checkpoint-670/tokenizer.json.REMOVED.git-id deleted file mode 100644 index 3081f40cf643c6511c69d3e37e6f8238a5cfd53e..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-670/tokenizer.json.REMOVED.git-id +++ /dev/null @@ -1 +0,0 @@ -b70d58b80eb0efbdad29dcf2df4ea60dca527d44 \ No newline at end of file diff --git a/smollm3_robust/checkpoint-670/tokenizer_config.json b/smollm3_robust/checkpoint-670/tokenizer_config.json deleted file mode 100644 index 61910c2db5cbdc9e6a6f37e14aaf00584cc6ad47..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-670/tokenizer_config.json +++ /dev/null @@ -1,2064 +0,0 @@ -{ - "added_tokens_decoder": { - "128000": { - "content": "<|begin_of_text|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128001": { - "content": "<|end_of_text|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128002": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128003": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128004": { - "content": "<|finetune_right_pad_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128005": { - "content": "<|reserved_special_token_2|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128006": { - "content": "<|start_header_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128007": { - "content": "<|end_header_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128008": { - "content": "<|eom_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128009": { - "content": "<|eot_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128010": { - "content": "<|python_tag|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128011": { - "content": "<|im_start|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128012": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128013": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128014": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128015": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128016": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128017": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128018": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128019": { - "content": "<|reserved_special_token_11|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128020": { - "content": "<|reserved_special_token_12|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128021": { - "content": "<|reserved_special_token_13|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128022": { - "content": "<|reserved_special_token_14|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128023": { - "content": "<|reserved_special_token_15|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128024": { - "content": "<|reserved_special_token_16|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128025": { - "content": "<|reserved_special_token_17|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128026": { - "content": "<|reserved_special_token_18|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128027": { - "content": "<|reserved_special_token_19|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128028": { - "content": "<|reserved_special_token_20|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128029": { - "content": "<|reserved_special_token_21|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128030": { - "content": "<|reserved_special_token_22|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128031": { - "content": "<|reserved_special_token_23|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128032": { - "content": "<|reserved_special_token_24|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128033": { - "content": "<|reserved_special_token_25|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128034": { - "content": "<|reserved_special_token_26|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128035": { - "content": "<|reserved_special_token_27|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128036": { - "content": "<|reserved_special_token_28|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128037": { - "content": "<|reserved_special_token_29|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128038": { - "content": "<|reserved_special_token_30|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128039": { - "content": "<|reserved_special_token_31|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128040": { - "content": "<|reserved_special_token_32|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128041": { - "content": "<|reserved_special_token_33|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128042": { - "content": "<|reserved_special_token_34|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128043": { - "content": "<|reserved_special_token_35|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128044": { - "content": "<|reserved_special_token_36|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128045": { - "content": "<|reserved_special_token_37|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128046": { - "content": "<|reserved_special_token_38|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128047": { - "content": "<|reserved_special_token_39|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128048": { - "content": "<|reserved_special_token_40|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128049": { - "content": "<|reserved_special_token_41|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128050": { - "content": "<|reserved_special_token_42|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128051": { - "content": "<|reserved_special_token_43|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128052": { - "content": "<|reserved_special_token_44|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128053": { - "content": "<|reserved_special_token_45|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128054": { - "content": "<|reserved_special_token_46|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128055": { - "content": "<|reserved_special_token_47|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128056": { - "content": "<|reserved_special_token_48|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128057": { - "content": "<|reserved_special_token_49|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128058": { - "content": "<|reserved_special_token_50|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128059": { - "content": "<|reserved_special_token_51|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128060": { - "content": "<|reserved_special_token_52|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128061": { - "content": "<|reserved_special_token_53|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128062": { - "content": "<|reserved_special_token_54|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128063": { - "content": "<|reserved_special_token_55|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128064": { - "content": "<|reserved_special_token_56|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128065": { - "content": "<|reserved_special_token_57|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128066": { - "content": "<|reserved_special_token_58|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128067": { - "content": "<|reserved_special_token_59|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128068": { - "content": "<|reserved_special_token_60|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128069": { - "content": "<|reserved_special_token_61|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128070": { - "content": "<|reserved_special_token_62|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128071": { - "content": "<|reserved_special_token_63|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128072": { - "content": "<|reserved_special_token_64|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128073": { - "content": "<|reserved_special_token_65|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128074": { - "content": "<|reserved_special_token_66|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128075": { - "content": "<|reserved_special_token_67|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128076": { - "content": "<|reserved_special_token_68|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128077": { - "content": "<|reserved_special_token_69|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128078": { - "content": "<|reserved_special_token_70|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128079": { - "content": "<|reserved_special_token_71|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128080": { - "content": "<|reserved_special_token_72|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128081": { - "content": "<|reserved_special_token_73|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128082": { - "content": "<|reserved_special_token_74|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128083": { - "content": "<|reserved_special_token_75|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128084": { - "content": "<|reserved_special_token_76|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128085": { - "content": "<|reserved_special_token_77|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128086": { - "content": "<|reserved_special_token_78|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128087": { - "content": "<|reserved_special_token_79|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128088": { - "content": "<|reserved_special_token_80|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128089": { - "content": "<|reserved_special_token_81|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128090": { - "content": "<|reserved_special_token_82|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128091": { - "content": "<|reserved_special_token_83|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128092": { - "content": "<|reserved_special_token_84|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128093": { - "content": "<|reserved_special_token_85|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128094": { - "content": "<|reserved_special_token_86|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128095": { - "content": "<|reserved_special_token_87|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128096": { - "content": "<|reserved_special_token_88|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128097": { - "content": "<|reserved_special_token_89|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128098": { - "content": "<|reserved_special_token_90|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128099": { - "content": "<|reserved_special_token_91|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128100": { - "content": "<|reserved_special_token_92|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128101": { - "content": "<|reserved_special_token_93|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128102": { - "content": "<|reserved_special_token_94|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128103": { - "content": "<|reserved_special_token_95|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128104": { - "content": "<|reserved_special_token_96|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128105": { - "content": "<|reserved_special_token_97|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128106": { - "content": "<|reserved_special_token_98|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128107": { - "content": "<|reserved_special_token_99|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128108": { - "content": "<|reserved_special_token_100|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128109": { - "content": "<|reserved_special_token_101|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128110": { - "content": "<|reserved_special_token_102|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128111": { - "content": "<|reserved_special_token_103|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128112": { - "content": "<|reserved_special_token_104|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128113": { - "content": "<|reserved_special_token_105|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128114": { - "content": "<|reserved_special_token_106|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128115": { - "content": "<|reserved_special_token_107|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128116": { - "content": "<|reserved_special_token_108|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128117": { - "content": "<|reserved_special_token_109|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128118": { - "content": "<|reserved_special_token_110|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128119": { - "content": "<|reserved_special_token_111|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128120": { - "content": "<|reserved_special_token_112|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128121": { - "content": "<|reserved_special_token_113|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128122": { - "content": "<|reserved_special_token_114|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128123": { - "content": "<|reserved_special_token_115|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128124": { - "content": "<|reserved_special_token_116|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128125": { - "content": "<|reserved_special_token_117|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128126": { - "content": "<|reserved_special_token_118|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128127": { - "content": "<|reserved_special_token_119|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128128": { - "content": "<|reserved_special_token_120|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128129": { - "content": "<|reserved_special_token_121|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128130": { - "content": "<|reserved_special_token_122|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128131": { - "content": "<|reserved_special_token_123|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128132": { - "content": "<|reserved_special_token_124|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128133": { - "content": "<|reserved_special_token_125|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128134": { - "content": "<|reserved_special_token_126|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128135": { - "content": "<|reserved_special_token_127|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128136": { - "content": "<|reserved_special_token_128|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128137": { - "content": "<|reserved_special_token_129|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128138": { - "content": "<|reserved_special_token_130|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128139": { - "content": "<|reserved_special_token_131|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128140": { - "content": "<|reserved_special_token_132|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128141": { - "content": "<|reserved_special_token_133|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128142": { - "content": "<|reserved_special_token_134|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128143": { - "content": "<|reserved_special_token_135|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128144": { - "content": "<|reserved_special_token_136|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128145": { - "content": "<|reserved_special_token_137|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128146": { - "content": "<|reserved_special_token_138|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128147": { - "content": "<|reserved_special_token_139|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128148": { - "content": "<|reserved_special_token_140|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128149": { - "content": "<|reserved_special_token_141|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128150": { - "content": "<|reserved_special_token_142|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128151": { - "content": "<|reserved_special_token_143|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128152": { - "content": "<|reserved_special_token_144|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128153": { - "content": "<|reserved_special_token_145|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128154": { - "content": "<|reserved_special_token_146|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128155": { - "content": "<|reserved_special_token_147|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128156": { - "content": "<|reserved_special_token_148|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128157": { - "content": "<|reserved_special_token_149|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128158": { - "content": "<|reserved_special_token_150|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128159": { - "content": "<|reserved_special_token_151|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128160": { - "content": "<|reserved_special_token_152|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128161": { - "content": "<|reserved_special_token_153|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128162": { - "content": "<|reserved_special_token_154|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128163": { - "content": "<|reserved_special_token_155|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128164": { - "content": "<|reserved_special_token_156|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128165": { - "content": "<|reserved_special_token_157|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128166": { - "content": "<|reserved_special_token_158|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128167": { - "content": "<|reserved_special_token_159|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128168": { - "content": "<|reserved_special_token_160|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128169": { - "content": "<|reserved_special_token_161|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128170": { - "content": "<|reserved_special_token_162|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128171": { - "content": "<|reserved_special_token_163|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128172": { - "content": "<|reserved_special_token_164|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128173": { - "content": "<|reserved_special_token_165|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128174": { - "content": "<|reserved_special_token_166|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128175": { - "content": "<|reserved_special_token_167|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128176": { - "content": "<|reserved_special_token_168|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128177": { - "content": "<|reserved_special_token_169|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128178": { - "content": "<|reserved_special_token_170|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128179": { - "content": "<|reserved_special_token_171|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128180": { - "content": "<|reserved_special_token_172|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128181": { - "content": "<|reserved_special_token_173|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128182": { - "content": "<|reserved_special_token_174|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128183": { - "content": "<|reserved_special_token_175|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128184": { - "content": "<|reserved_special_token_176|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128185": { - "content": "<|reserved_special_token_177|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128186": { - "content": "<|reserved_special_token_178|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128187": { - "content": "<|reserved_special_token_179|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128188": { - "content": "<|reserved_special_token_180|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128189": { - "content": "<|reserved_special_token_181|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128190": { - "content": "<|reserved_special_token_182|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128191": { - "content": "<|reserved_special_token_183|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128192": { - "content": "<|reserved_special_token_184|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128193": { - "content": "<|reserved_special_token_185|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128194": { - "content": "<|reserved_special_token_186|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128195": { - "content": "<|reserved_special_token_187|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128196": { - "content": "<|reserved_special_token_188|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128197": { - "content": "<|reserved_special_token_189|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128198": { - "content": "<|reserved_special_token_190|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128199": { - "content": "<|reserved_special_token_191|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128200": { - "content": "<|reserved_special_token_192|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128201": { - "content": "<|reserved_special_token_193|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128202": { - "content": "<|reserved_special_token_194|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128203": { - "content": "<|reserved_special_token_195|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128204": { - "content": "<|reserved_special_token_196|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128205": { - "content": "<|reserved_special_token_197|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128206": { - "content": "<|reserved_special_token_198|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128207": { - "content": "<|reserved_special_token_199|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128208": { - "content": "<|reserved_special_token_200|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128209": { - "content": "<|reserved_special_token_201|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128210": { - "content": "<|reserved_special_token_202|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128211": { - "content": "<|reserved_special_token_203|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128212": { - "content": "<|reserved_special_token_204|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128213": { - "content": "<|reserved_special_token_205|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128214": { - "content": "<|reserved_special_token_206|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128215": { - "content": "<|reserved_special_token_207|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128216": { - "content": "<|reserved_special_token_208|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128217": { - "content": "<|reserved_special_token_209|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128218": { - "content": "<|reserved_special_token_210|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128219": { - "content": "<|reserved_special_token_211|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128220": { - "content": "<|reserved_special_token_212|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128221": { - "content": "<|reserved_special_token_213|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128222": { - "content": "<|reserved_special_token_214|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128223": { - "content": "<|reserved_special_token_215|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128224": { - "content": "<|reserved_special_token_216|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128225": { - "content": "<|reserved_special_token_217|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128226": { - "content": "<|reserved_special_token_218|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128227": { - "content": "<|reserved_special_token_219|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128228": { - "content": "<|reserved_special_token_220|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128229": { - "content": "<|reserved_special_token_221|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128230": { - "content": "<|reserved_special_token_222|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128231": { - "content": "<|reserved_special_token_223|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128232": { - "content": "<|reserved_special_token_224|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128233": { - "content": "<|reserved_special_token_225|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128234": { - "content": "<|reserved_special_token_226|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128235": { - "content": "<|reserved_special_token_227|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128236": { - "content": "<|reserved_special_token_228|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128237": { - "content": "<|reserved_special_token_229|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128238": { - "content": "<|reserved_special_token_230|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128239": { - "content": "<|reserved_special_token_231|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128240": { - "content": "<|reserved_special_token_232|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128241": { - "content": "<|reserved_special_token_233|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128242": { - "content": "<|reserved_special_token_234|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128243": { - "content": "<|reserved_special_token_235|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128244": { - "content": "<|reserved_special_token_236|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128245": { - "content": "<|reserved_special_token_237|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128246": { - "content": "<|reserved_special_token_238|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128247": { - "content": "<|reserved_special_token_239|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128248": { - "content": "<|reserved_special_token_240|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128249": { - "content": "<|reserved_special_token_241|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128250": { - "content": "<|reserved_special_token_242|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128251": { - "content": "<|reserved_special_token_243|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128252": { - "content": "<|reserved_special_token_244|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128253": { - "content": "<|reserved_special_token_245|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128254": { - "content": "<|reserved_special_token_246|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128255": { - "content": "<|reserved_special_token_247|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - } - }, - "bos_token": null, - "clean_up_tokenization_spaces": true, - "eos_token": "<|im_end|>", - "extra_special_tokens": {}, - "fast": false, - "model_input_names": [ - "input_ids", - "attention_mask" - ], - "model_max_length": 131072, - "pad_token": "<|im_end|>", - "tokenizer_class": "PreTrainedTokenizerFast" -} diff --git a/smollm3_robust/checkpoint-670/trainer_state.json b/smollm3_robust/checkpoint-670/trainer_state.json deleted file mode 100644 index 345faa538a19ea6bb853b0d26453829092c20dd4..0000000000000000000000000000000000000000 --- a/smollm3_robust/checkpoint-670/trainer_state.json +++ /dev/null @@ -1,2379 +0,0 @@ -{ - "best_global_step": null, - "best_metric": null, - "best_model_checkpoint": null, - "epoch": 10.0, - "eval_steps": 500, - "global_step": 670, - "is_hyper_param_search": false, - "is_local_process_zero": true, - "is_world_process_zero": true, - "log_history": [ - { - "epoch": 0.0299625468164794, - "grad_norm": 3.722816228866577, - "learning_rate": 5e-06, - "loss": 1.6973, - "step": 2 - }, - { - "epoch": 0.0599250936329588, - "grad_norm": 3.4061238765716553, - "learning_rate": 1.5e-05, - "loss": 1.6822, - "step": 4 - }, - { - "epoch": 0.0898876404494382, - "grad_norm": 2.510838270187378, - "learning_rate": 2.5e-05, - "loss": 1.4687, - "step": 6 - }, - { - "epoch": 0.1198501872659176, - "grad_norm": 2.095601797103882, - "learning_rate": 3.5e-05, - "loss": 1.4093, - "step": 8 - }, - { - "epoch": 0.149812734082397, - "grad_norm": 2.001142978668213, - "learning_rate": 4.5e-05, - "loss": 1.1657, - "step": 10 - }, - { - "epoch": 0.1797752808988764, - "grad_norm": 1.8859666585922241, - "learning_rate": 4.992424242424243e-05, - "loss": 0.9447, - "step": 12 - }, - { - "epoch": 0.20973782771535582, - "grad_norm": 2.0142483711242676, - "learning_rate": 4.9772727272727275e-05, - "loss": 0.8334, - "step": 14 - }, - { - "epoch": 0.2397003745318352, - "grad_norm": 1.6467182636260986, - "learning_rate": 4.962121212121213e-05, - "loss": 0.6336, - "step": 16 - }, - { - "epoch": 0.2696629213483146, - "grad_norm": 1.2968051433563232, - "learning_rate": 4.946969696969697e-05, - "loss": 0.4959, - "step": 18 - }, - { - "epoch": 0.299625468164794, - "grad_norm": 1.001763105392456, - "learning_rate": 4.931818181818182e-05, - "loss": 0.3829, - "step": 20 - }, - { - "epoch": 0.3295880149812734, - "grad_norm": 0.47088268399238586, - "learning_rate": 4.9166666666666665e-05, - "loss": 0.3361, - "step": 22 - }, - { - "epoch": 0.3595505617977528, - "grad_norm": 0.6152827739715576, - "learning_rate": 4.901515151515152e-05, - "loss": 0.3142, - "step": 24 - }, - { - "epoch": 0.3895131086142322, - "grad_norm": 0.5034743547439575, - "learning_rate": 4.886363636363637e-05, - "loss": 0.2581, - "step": 26 - }, - { - "epoch": 0.41947565543071164, - "grad_norm": 0.5429691076278687, - "learning_rate": 4.8712121212121216e-05, - "loss": 0.247, - "step": 28 - }, - { - "epoch": 0.449438202247191, - "grad_norm": 0.6162884831428528, - "learning_rate": 4.856060606060606e-05, - "loss": 0.2266, - "step": 30 - }, - { - "epoch": 0.4794007490636704, - "grad_norm": 0.7645140886306763, - "learning_rate": 4.840909090909091e-05, - "loss": 0.1985, - "step": 32 - }, - { - "epoch": 0.5093632958801498, - "grad_norm": 0.600817084312439, - "learning_rate": 4.825757575757576e-05, - "loss": 0.1634, - "step": 34 - }, - { - "epoch": 0.5393258426966292, - "grad_norm": 0.4553565979003906, - "learning_rate": 4.810606060606061e-05, - "loss": 0.1626, - "step": 36 - }, - { - "epoch": 0.5692883895131086, - "grad_norm": 0.7166281342506409, - "learning_rate": 4.795454545454546e-05, - "loss": 0.1796, - "step": 38 - }, - { - "epoch": 0.599250936329588, - "grad_norm": 0.42647111415863037, - "learning_rate": 4.7803030303030304e-05, - "loss": 0.1384, - "step": 40 - }, - { - "epoch": 0.6292134831460674, - "grad_norm": 0.4419521689414978, - "learning_rate": 4.765151515151515e-05, - "loss": 0.1498, - "step": 42 - }, - { - "epoch": 0.6591760299625468, - "grad_norm": 0.49058037996292114, - "learning_rate": 4.75e-05, - "loss": 0.0965, - "step": 44 - }, - { - "epoch": 0.6891385767790262, - "grad_norm": 0.3956356644630432, - "learning_rate": 4.7348484848484855e-05, - "loss": 0.0873, - "step": 46 - }, - { - "epoch": 0.7191011235955056, - "grad_norm": 0.4841479957103729, - "learning_rate": 4.71969696969697e-05, - "loss": 0.0909, - "step": 48 - }, - { - "epoch": 0.7490636704119851, - "grad_norm": 0.5453945398330688, - "learning_rate": 4.704545454545455e-05, - "loss": 0.0719, - "step": 50 - }, - { - "epoch": 0.7790262172284644, - "grad_norm": 0.44312047958374023, - "learning_rate": 4.689393939393939e-05, - "loss": 0.0553, - "step": 52 - }, - { - "epoch": 0.8089887640449438, - "grad_norm": 0.5744581818580627, - "learning_rate": 4.6742424242424245e-05, - "loss": 0.0744, - "step": 54 - }, - { - "epoch": 0.8389513108614233, - "grad_norm": 0.9026234745979309, - "learning_rate": 4.659090909090909e-05, - "loss": 0.0608, - "step": 56 - }, - { - "epoch": 0.8689138576779026, - "grad_norm": 0.4250333905220032, - "learning_rate": 4.6439393939393944e-05, - "loss": 0.0364, - "step": 58 - }, - { - "epoch": 0.898876404494382, - "grad_norm": 0.6075786352157593, - "learning_rate": 4.628787878787879e-05, - "loss": 0.0495, - "step": 60 - }, - { - "epoch": 0.9288389513108615, - "grad_norm": 0.6797979474067688, - "learning_rate": 4.6136363636363635e-05, - "loss": 0.0471, - "step": 62 - }, - { - "epoch": 0.9588014981273408, - "grad_norm": 0.296935498714447, - "learning_rate": 4.598484848484849e-05, - "loss": 0.0299, - "step": 64 - }, - { - "epoch": 0.9887640449438202, - "grad_norm": 0.28508949279785156, - "learning_rate": 4.5833333333333334e-05, - "loss": 0.0309, - "step": 66 - }, - { - "epoch": 1.0149812734082397, - "grad_norm": 0.44689077138900757, - "learning_rate": 4.5681818181818186e-05, - "loss": 0.0487, - "step": 68 - }, - { - "epoch": 1.0449438202247192, - "grad_norm": 0.36559513211250305, - "learning_rate": 4.553030303030303e-05, - "loss": 0.0304, - "step": 70 - }, - { - "epoch": 1.0749063670411985, - "grad_norm": 0.3891605734825134, - "learning_rate": 4.5378787878787885e-05, - "loss": 0.0254, - "step": 72 - }, - { - "epoch": 1.104868913857678, - "grad_norm": 0.36933985352516174, - "learning_rate": 4.522727272727273e-05, - "loss": 0.0363, - "step": 74 - }, - { - "epoch": 1.1348314606741572, - "grad_norm": 0.41140684485435486, - "learning_rate": 4.5075757575757577e-05, - "loss": 0.0386, - "step": 76 - }, - { - "epoch": 1.1647940074906367, - "grad_norm": 0.5980117321014404, - "learning_rate": 4.492424242424242e-05, - "loss": 0.0201, - "step": 78 - }, - { - "epoch": 1.1947565543071161, - "grad_norm": 0.23804394900798798, - "learning_rate": 4.4772727272727275e-05, - "loss": 0.0187, - "step": 80 - }, - { - "epoch": 1.2247191011235956, - "grad_norm": 0.23067264258861542, - "learning_rate": 4.462121212121213e-05, - "loss": 0.0175, - "step": 82 - }, - { - "epoch": 1.2546816479400749, - "grad_norm": 0.36974087357521057, - "learning_rate": 4.4469696969696973e-05, - "loss": 0.0303, - "step": 84 - }, - { - "epoch": 1.2846441947565543, - "grad_norm": 0.4070376753807068, - "learning_rate": 4.431818181818182e-05, - "loss": 0.0268, - "step": 86 - }, - { - "epoch": 1.3146067415730336, - "grad_norm": 0.2764546275138855, - "learning_rate": 4.4166666666666665e-05, - "loss": 0.0224, - "step": 88 - }, - { - "epoch": 1.344569288389513, - "grad_norm": 0.20103320479393005, - "learning_rate": 4.401515151515152e-05, - "loss": 0.0202, - "step": 90 - }, - { - "epoch": 1.3745318352059925, - "grad_norm": 0.26625537872314453, - "learning_rate": 4.386363636363637e-05, - "loss": 0.0216, - "step": 92 - }, - { - "epoch": 1.404494382022472, - "grad_norm": 0.24225257337093353, - "learning_rate": 4.3712121212121216e-05, - "loss": 0.0168, - "step": 94 - }, - { - "epoch": 1.4344569288389513, - "grad_norm": 0.3456275463104248, - "learning_rate": 4.356060606060606e-05, - "loss": 0.0207, - "step": 96 - }, - { - "epoch": 1.4644194756554307, - "grad_norm": 0.21905088424682617, - "learning_rate": 4.340909090909091e-05, - "loss": 0.0145, - "step": 98 - }, - { - "epoch": 1.49438202247191, - "grad_norm": 0.5236210227012634, - "learning_rate": 4.325757575757576e-05, - "loss": 0.025, - "step": 100 - }, - { - "epoch": 1.5243445692883895, - "grad_norm": 0.35533836483955383, - "learning_rate": 4.3106060606060606e-05, - "loss": 0.0183, - "step": 102 - }, - { - "epoch": 1.554307116104869, - "grad_norm": 0.3094329833984375, - "learning_rate": 4.295454545454546e-05, - "loss": 0.022, - "step": 104 - }, - { - "epoch": 1.5842696629213484, - "grad_norm": 0.24415704607963562, - "learning_rate": 4.2803030303030305e-05, - "loss": 0.0149, - "step": 106 - }, - { - "epoch": 1.6142322097378277, - "grad_norm": 0.44377341866493225, - "learning_rate": 4.265151515151515e-05, - "loss": 0.0152, - "step": 108 - }, - { - "epoch": 1.6441947565543071, - "grad_norm": 0.4970695972442627, - "learning_rate": 4.25e-05, - "loss": 0.0196, - "step": 110 - }, - { - "epoch": 1.6741573033707864, - "grad_norm": 0.20200444757938385, - "learning_rate": 4.234848484848485e-05, - "loss": 0.0165, - "step": 112 - }, - { - "epoch": 1.7041198501872659, - "grad_norm": 0.21261580288410187, - "learning_rate": 4.21969696969697e-05, - "loss": 0.0168, - "step": 114 - }, - { - "epoch": 1.7340823970037453, - "grad_norm": 0.2608441412448883, - "learning_rate": 4.204545454545455e-05, - "loss": 0.0146, - "step": 116 - }, - { - "epoch": 1.7640449438202248, - "grad_norm": 0.20836834609508514, - "learning_rate": 4.189393939393939e-05, - "loss": 0.0195, - "step": 118 - }, - { - "epoch": 1.7940074906367043, - "grad_norm": 0.29434454441070557, - "learning_rate": 4.1742424242424246e-05, - "loss": 0.0187, - "step": 120 - }, - { - "epoch": 1.8239700374531835, - "grad_norm": 0.2228062003850937, - "learning_rate": 4.159090909090909e-05, - "loss": 0.0171, - "step": 122 - }, - { - "epoch": 1.8539325842696628, - "grad_norm": 0.3206649124622345, - "learning_rate": 4.143939393939394e-05, - "loss": 0.0204, - "step": 124 - }, - { - "epoch": 1.8838951310861423, - "grad_norm": 0.18132628500461578, - "learning_rate": 4.128787878787879e-05, - "loss": 0.0203, - "step": 126 - }, - { - "epoch": 1.9138576779026217, - "grad_norm": 0.2030370533466339, - "learning_rate": 4.113636363636364e-05, - "loss": 0.0179, - "step": 128 - }, - { - "epoch": 1.9438202247191012, - "grad_norm": 0.2079804688692093, - "learning_rate": 4.098484848484849e-05, - "loss": 0.0198, - "step": 130 - }, - { - "epoch": 1.9737827715355807, - "grad_norm": 0.23763149976730347, - "learning_rate": 4.0833333333333334e-05, - "loss": 0.0182, - "step": 132 - }, - { - "epoch": 2.0, - "grad_norm": 0.21091270446777344, - "learning_rate": 4.068181818181818e-05, - "loss": 0.0186, - "step": 134 - }, - { - "epoch": 2.0299625468164795, - "grad_norm": 0.21226680278778076, - "learning_rate": 4.053030303030303e-05, - "loss": 0.0166, - "step": 136 - }, - { - "epoch": 2.059925093632959, - "grad_norm": 0.2453354448080063, - "learning_rate": 4.0378787878787885e-05, - "loss": 0.0162, - "step": 138 - }, - { - "epoch": 2.0898876404494384, - "grad_norm": 0.17463107407093048, - "learning_rate": 4.022727272727273e-05, - "loss": 0.0156, - "step": 140 - }, - { - "epoch": 2.1198501872659175, - "grad_norm": 0.307685524225235, - "learning_rate": 4.007575757575758e-05, - "loss": 0.0211, - "step": 142 - }, - { - "epoch": 2.149812734082397, - "grad_norm": 0.23119647800922394, - "learning_rate": 3.992424242424242e-05, - "loss": 0.0184, - "step": 144 - }, - { - "epoch": 2.1797752808988764, - "grad_norm": 0.15694215893745422, - "learning_rate": 3.9772727272727275e-05, - "loss": 0.0165, - "step": 146 - }, - { - "epoch": 2.209737827715356, - "grad_norm": 0.26551949977874756, - "learning_rate": 3.962121212121213e-05, - "loss": 0.0184, - "step": 148 - }, - { - "epoch": 2.2397003745318353, - "grad_norm": 0.21518471837043762, - "learning_rate": 3.9469696969696974e-05, - "loss": 0.0168, - "step": 150 - }, - { - "epoch": 2.2696629213483144, - "grad_norm": 0.1955074667930603, - "learning_rate": 3.931818181818182e-05, - "loss": 0.016, - "step": 152 - }, - { - "epoch": 2.299625468164794, - "grad_norm": 0.24419444799423218, - "learning_rate": 3.9166666666666665e-05, - "loss": 0.0151, - "step": 154 - }, - { - "epoch": 2.3295880149812733, - "grad_norm": 0.29677078127861023, - "learning_rate": 3.901515151515152e-05, - "loss": 0.0167, - "step": 156 - }, - { - "epoch": 2.359550561797753, - "grad_norm": 0.19572412967681885, - "learning_rate": 3.8863636363636364e-05, - "loss": 0.0162, - "step": 158 - }, - { - "epoch": 2.3895131086142323, - "grad_norm": 0.20857453346252441, - "learning_rate": 3.8712121212121217e-05, - "loss": 0.0166, - "step": 160 - }, - { - "epoch": 2.4194756554307117, - "grad_norm": 0.14987444877624512, - "learning_rate": 3.856060606060606e-05, - "loss": 0.015, - "step": 162 - }, - { - "epoch": 2.449438202247191, - "grad_norm": 0.22435317933559418, - "learning_rate": 3.840909090909091e-05, - "loss": 0.0171, - "step": 164 - }, - { - "epoch": 2.4794007490636703, - "grad_norm": 0.29637566208839417, - "learning_rate": 3.825757575757576e-05, - "loss": 0.0189, - "step": 166 - }, - { - "epoch": 2.5093632958801497, - "grad_norm": 0.24401557445526123, - "learning_rate": 3.810606060606061e-05, - "loss": 0.0164, - "step": 168 - }, - { - "epoch": 2.539325842696629, - "grad_norm": 0.19472604990005493, - "learning_rate": 3.795454545454545e-05, - "loss": 0.0172, - "step": 170 - }, - { - "epoch": 2.5692883895131087, - "grad_norm": 0.21067409217357635, - "learning_rate": 3.7803030303030305e-05, - "loss": 0.0147, - "step": 172 - }, - { - "epoch": 2.599250936329588, - "grad_norm": 0.157924622297287, - "learning_rate": 3.765151515151516e-05, - "loss": 0.0146, - "step": 174 - }, - { - "epoch": 2.629213483146067, - "grad_norm": 0.22833997011184692, - "learning_rate": 3.7500000000000003e-05, - "loss": 0.0158, - "step": 176 - }, - { - "epoch": 2.6591760299625467, - "grad_norm": 0.1746760606765747, - "learning_rate": 3.734848484848485e-05, - "loss": 0.0173, - "step": 178 - }, - { - "epoch": 2.689138576779026, - "grad_norm": 0.25366461277008057, - "learning_rate": 3.7196969696969695e-05, - "loss": 0.0162, - "step": 180 - }, - { - "epoch": 2.7191011235955056, - "grad_norm": 0.1569383144378662, - "learning_rate": 3.704545454545455e-05, - "loss": 0.0163, - "step": 182 - }, - { - "epoch": 2.749063670411985, - "grad_norm": 0.10542655736207962, - "learning_rate": 3.68939393939394e-05, - "loss": 0.014, - "step": 184 - }, - { - "epoch": 2.7790262172284645, - "grad_norm": 0.11274619400501251, - "learning_rate": 3.6742424242424246e-05, - "loss": 0.0141, - "step": 186 - }, - { - "epoch": 2.808988764044944, - "grad_norm": 0.2910693287849426, - "learning_rate": 3.659090909090909e-05, - "loss": 0.0199, - "step": 188 - }, - { - "epoch": 2.8389513108614235, - "grad_norm": 0.15630793571472168, - "learning_rate": 3.643939393939394e-05, - "loss": 0.0142, - "step": 190 - }, - { - "epoch": 2.8689138576779025, - "grad_norm": 0.1520637422800064, - "learning_rate": 3.628787878787879e-05, - "loss": 0.0166, - "step": 192 - }, - { - "epoch": 2.898876404494382, - "grad_norm": 0.1700475662946701, - "learning_rate": 3.613636363636364e-05, - "loss": 0.0136, - "step": 194 - }, - { - "epoch": 2.9288389513108615, - "grad_norm": 0.2383895367383957, - "learning_rate": 3.598484848484849e-05, - "loss": 0.0146, - "step": 196 - }, - { - "epoch": 2.958801498127341, - "grad_norm": 0.2498546838760376, - "learning_rate": 3.5833333333333335e-05, - "loss": 0.0167, - "step": 198 - }, - { - "epoch": 2.98876404494382, - "grad_norm": 0.24929316341876984, - "learning_rate": 3.568181818181818e-05, - "loss": 0.0195, - "step": 200 - }, - { - "epoch": 3.0149812734082397, - "grad_norm": 0.1542021781206131, - "learning_rate": 3.553030303030303e-05, - "loss": 0.0126, - "step": 202 - }, - { - "epoch": 3.044943820224719, - "grad_norm": 0.15488319098949432, - "learning_rate": 3.537878787878788e-05, - "loss": 0.0143, - "step": 204 - }, - { - "epoch": 3.0749063670411987, - "grad_norm": 0.22363406419754028, - "learning_rate": 3.522727272727273e-05, - "loss": 0.0172, - "step": 206 - }, - { - "epoch": 3.1048689138576777, - "grad_norm": 0.23685605823993683, - "learning_rate": 3.507575757575758e-05, - "loss": 0.0152, - "step": 208 - }, - { - "epoch": 3.134831460674157, - "grad_norm": 0.18075552582740784, - "learning_rate": 3.492424242424242e-05, - "loss": 0.016, - "step": 210 - }, - { - "epoch": 3.1647940074906367, - "grad_norm": 0.18518349528312683, - "learning_rate": 3.4772727272727276e-05, - "loss": 0.0147, - "step": 212 - }, - { - "epoch": 3.194756554307116, - "grad_norm": 0.29419267177581787, - "learning_rate": 3.462121212121212e-05, - "loss": 0.0159, - "step": 214 - }, - { - "epoch": 3.2247191011235956, - "grad_norm": 0.13987213373184204, - "learning_rate": 3.4469696969696974e-05, - "loss": 0.0136, - "step": 216 - }, - { - "epoch": 3.254681647940075, - "grad_norm": 0.2520444989204407, - "learning_rate": 3.431818181818182e-05, - "loss": 0.0172, - "step": 218 - }, - { - "epoch": 3.284644194756554, - "grad_norm": 0.12775233387947083, - "learning_rate": 3.4166666666666666e-05, - "loss": 0.0149, - "step": 220 - }, - { - "epoch": 3.3146067415730336, - "grad_norm": 0.15709002315998077, - "learning_rate": 3.401515151515152e-05, - "loss": 0.0153, - "step": 222 - }, - { - "epoch": 3.344569288389513, - "grad_norm": 0.17532695829868317, - "learning_rate": 3.3863636363636364e-05, - "loss": 0.0149, - "step": 224 - }, - { - "epoch": 3.3745318352059925, - "grad_norm": 0.21435686945915222, - "learning_rate": 3.371212121212121e-05, - "loss": 0.0158, - "step": 226 - }, - { - "epoch": 3.404494382022472, - "grad_norm": 0.23152689635753632, - "learning_rate": 3.356060606060606e-05, - "loss": 0.0169, - "step": 228 - }, - { - "epoch": 3.4344569288389515, - "grad_norm": 0.20635737478733063, - "learning_rate": 3.3409090909090915e-05, - "loss": 0.0167, - "step": 230 - }, - { - "epoch": 3.464419475655431, - "grad_norm": 0.18343976140022278, - "learning_rate": 3.325757575757576e-05, - "loss": 0.0144, - "step": 232 - }, - { - "epoch": 3.49438202247191, - "grad_norm": 0.14534398913383484, - "learning_rate": 3.310606060606061e-05, - "loss": 0.0138, - "step": 234 - }, - { - "epoch": 3.5243445692883895, - "grad_norm": 0.2103314846754074, - "learning_rate": 3.295454545454545e-05, - "loss": 0.0164, - "step": 236 - }, - { - "epoch": 3.554307116104869, - "grad_norm": 0.13487893342971802, - "learning_rate": 3.2803030303030305e-05, - "loss": 0.0149, - "step": 238 - }, - { - "epoch": 3.5842696629213484, - "grad_norm": 0.13789679110050201, - "learning_rate": 3.265151515151516e-05, - "loss": 0.0137, - "step": 240 - }, - { - "epoch": 3.6142322097378274, - "grad_norm": 0.21411816775798798, - "learning_rate": 3.2500000000000004e-05, - "loss": 0.0181, - "step": 242 - }, - { - "epoch": 3.644194756554307, - "grad_norm": 0.22373120486736298, - "learning_rate": 3.234848484848485e-05, - "loss": 0.0158, - "step": 244 - }, - { - "epoch": 3.6741573033707864, - "grad_norm": 0.17626555263996124, - "learning_rate": 3.2196969696969696e-05, - "loss": 0.0158, - "step": 246 - }, - { - "epoch": 3.704119850187266, - "grad_norm": 0.15838968753814697, - "learning_rate": 3.204545454545455e-05, - "loss": 0.0149, - "step": 248 - }, - { - "epoch": 3.7340823970037453, - "grad_norm": 0.15824346244335175, - "learning_rate": 3.18939393939394e-05, - "loss": 0.0151, - "step": 250 - }, - { - "epoch": 3.764044943820225, - "grad_norm": 0.1346617490053177, - "learning_rate": 3.174242424242425e-05, - "loss": 0.0155, - "step": 252 - }, - { - "epoch": 3.7940074906367043, - "grad_norm": 0.15259407460689545, - "learning_rate": 3.159090909090909e-05, - "loss": 0.0153, - "step": 254 - }, - { - "epoch": 3.8239700374531838, - "grad_norm": 0.10529737174510956, - "learning_rate": 3.143939393939394e-05, - "loss": 0.0131, - "step": 256 - }, - { - "epoch": 3.853932584269663, - "grad_norm": 0.21343955397605896, - "learning_rate": 3.128787878787879e-05, - "loss": 0.0191, - "step": 258 - }, - { - "epoch": 3.8838951310861423, - "grad_norm": 0.19346514344215393, - "learning_rate": 3.113636363636364e-05, - "loss": 0.0166, - "step": 260 - }, - { - "epoch": 3.9138576779026217, - "grad_norm": 0.11122281849384308, - "learning_rate": 3.098484848484849e-05, - "loss": 0.0146, - "step": 262 - }, - { - "epoch": 3.943820224719101, - "grad_norm": 0.1087551936507225, - "learning_rate": 3.0833333333333335e-05, - "loss": 0.0152, - "step": 264 - }, - { - "epoch": 3.9737827715355807, - "grad_norm": 0.10952229797840118, - "learning_rate": 3.068181818181818e-05, - "loss": 0.0131, - "step": 266 - }, - { - "epoch": 4.0, - "grad_norm": 0.21926254034042358, - "learning_rate": 3.0530303030303034e-05, - "loss": 0.0183, - "step": 268 - }, - { - "epoch": 4.0299625468164795, - "grad_norm": 0.11138854920864105, - "learning_rate": 3.037878787878788e-05, - "loss": 0.0157, - "step": 270 - }, - { - "epoch": 4.059925093632959, - "grad_norm": 0.18572665750980377, - "learning_rate": 3.0227272727272725e-05, - "loss": 0.0143, - "step": 272 - }, - { - "epoch": 4.089887640449438, - "grad_norm": 0.11445632576942444, - "learning_rate": 3.0075757575757578e-05, - "loss": 0.014, - "step": 274 - }, - { - "epoch": 4.119850187265918, - "grad_norm": 0.13404381275177002, - "learning_rate": 2.9924242424242427e-05, - "loss": 0.016, - "step": 276 - }, - { - "epoch": 4.149812734082397, - "grad_norm": 0.22214095294475555, - "learning_rate": 2.9772727272727273e-05, - "loss": 0.0158, - "step": 278 - }, - { - "epoch": 4.179775280898877, - "grad_norm": 0.16340972483158112, - "learning_rate": 2.9621212121212122e-05, - "loss": 0.0168, - "step": 280 - }, - { - "epoch": 4.209737827715355, - "grad_norm": 0.1711144745349884, - "learning_rate": 2.9469696969696968e-05, - "loss": 0.0152, - "step": 282 - }, - { - "epoch": 4.239700374531835, - "grad_norm": 0.13778892159461975, - "learning_rate": 2.9318181818181817e-05, - "loss": 0.0164, - "step": 284 - }, - { - "epoch": 4.269662921348314, - "grad_norm": 0.15408281981945038, - "learning_rate": 2.916666666666667e-05, - "loss": 0.0174, - "step": 286 - }, - { - "epoch": 4.299625468164794, - "grad_norm": 0.12467946112155914, - "learning_rate": 2.901515151515152e-05, - "loss": 0.0152, - "step": 288 - }, - { - "epoch": 4.329588014981273, - "grad_norm": 0.1988392323255539, - "learning_rate": 2.8863636363636365e-05, - "loss": 0.0159, - "step": 290 - }, - { - "epoch": 4.359550561797753, - "grad_norm": 0.1968161165714264, - "learning_rate": 2.8712121212121214e-05, - "loss": 0.0164, - "step": 292 - }, - { - "epoch": 4.389513108614232, - "grad_norm": 0.16299834847450256, - "learning_rate": 2.856060606060606e-05, - "loss": 0.0126, - "step": 294 - }, - { - "epoch": 4.419475655430712, - "grad_norm": 0.20078356564044952, - "learning_rate": 2.8409090909090912e-05, - "loss": 0.0167, - "step": 296 - }, - { - "epoch": 4.449438202247191, - "grad_norm": 0.10105714946985245, - "learning_rate": 2.825757575757576e-05, - "loss": 0.0161, - "step": 298 - }, - { - "epoch": 4.479400749063671, - "grad_norm": 0.14914706349372864, - "learning_rate": 2.8106060606060607e-05, - "loss": 0.0172, - "step": 300 - }, - { - "epoch": 4.50936329588015, - "grad_norm": 0.12265647202730179, - "learning_rate": 2.7954545454545457e-05, - "loss": 0.0158, - "step": 302 - }, - { - "epoch": 4.539325842696629, - "grad_norm": 0.11637593805789948, - "learning_rate": 2.7803030303030303e-05, - "loss": 0.0151, - "step": 304 - }, - { - "epoch": 4.569288389513108, - "grad_norm": 0.11158251017332077, - "learning_rate": 2.7651515151515152e-05, - "loss": 0.0166, - "step": 306 - }, - { - "epoch": 4.599250936329588, - "grad_norm": 0.14547735452651978, - "learning_rate": 2.7500000000000004e-05, - "loss": 0.0157, - "step": 308 - }, - { - "epoch": 4.629213483146067, - "grad_norm": 0.15811274945735931, - "learning_rate": 2.734848484848485e-05, - "loss": 0.0145, - "step": 310 - }, - { - "epoch": 4.659176029962547, - "grad_norm": 0.1495320349931717, - "learning_rate": 2.71969696969697e-05, - "loss": 0.0165, - "step": 312 - }, - { - "epoch": 4.689138576779026, - "grad_norm": 0.14297890663146973, - "learning_rate": 2.7045454545454545e-05, - "loss": 0.016, - "step": 314 - }, - { - "epoch": 4.719101123595506, - "grad_norm": 0.1328546106815338, - "learning_rate": 2.6893939393939394e-05, - "loss": 0.0159, - "step": 316 - }, - { - "epoch": 4.749063670411985, - "grad_norm": 0.15500609576702118, - "learning_rate": 2.674242424242424e-05, - "loss": 0.0141, - "step": 318 - }, - { - "epoch": 4.7790262172284645, - "grad_norm": 0.1334412395954132, - "learning_rate": 2.6590909090909093e-05, - "loss": 0.014, - "step": 320 - }, - { - "epoch": 4.808988764044944, - "grad_norm": 0.11698229610919952, - "learning_rate": 2.6439393939393942e-05, - "loss": 0.0128, - "step": 322 - }, - { - "epoch": 4.8389513108614235, - "grad_norm": 0.10686776041984558, - "learning_rate": 2.6287878787878788e-05, - "loss": 0.0115, - "step": 324 - }, - { - "epoch": 4.868913857677903, - "grad_norm": 0.13976678252220154, - "learning_rate": 2.6136363636363637e-05, - "loss": 0.0154, - "step": 326 - }, - { - "epoch": 4.898876404494382, - "grad_norm": 0.1366475522518158, - "learning_rate": 2.5984848484848483e-05, - "loss": 0.0155, - "step": 328 - }, - { - "epoch": 4.928838951310862, - "grad_norm": 0.16024161875247955, - "learning_rate": 2.5833333333333336e-05, - "loss": 0.0186, - "step": 330 - }, - { - "epoch": 4.9588014981273405, - "grad_norm": 0.11987770348787308, - "learning_rate": 2.5681818181818185e-05, - "loss": 0.0151, - "step": 332 - }, - { - "epoch": 4.98876404494382, - "grad_norm": 0.12122765928506851, - "learning_rate": 2.553030303030303e-05, - "loss": 0.0129, - "step": 334 - }, - { - "epoch": 5.01498127340824, - "grad_norm": 0.14996588230133057, - "learning_rate": 2.537878787878788e-05, - "loss": 0.0134, - "step": 336 - }, - { - "epoch": 5.044943820224719, - "grad_norm": 0.1114385575056076, - "learning_rate": 2.5227272727272726e-05, - "loss": 0.0138, - "step": 338 - }, - { - "epoch": 5.074906367041199, - "grad_norm": 0.11765974760055542, - "learning_rate": 2.5075757575757575e-05, - "loss": 0.013, - "step": 340 - }, - { - "epoch": 5.104868913857678, - "grad_norm": 0.18139192461967468, - "learning_rate": 2.4924242424242424e-05, - "loss": 0.0151, - "step": 342 - }, - { - "epoch": 5.134831460674158, - "grad_norm": 0.12986940145492554, - "learning_rate": 2.4772727272727277e-05, - "loss": 0.0149, - "step": 344 - }, - { - "epoch": 5.164794007490637, - "grad_norm": 0.12934552133083344, - "learning_rate": 2.4621212121212123e-05, - "loss": 0.0145, - "step": 346 - }, - { - "epoch": 5.194756554307116, - "grad_norm": 0.13571321964263916, - "learning_rate": 2.4469696969696972e-05, - "loss": 0.0175, - "step": 348 - }, - { - "epoch": 5.224719101123595, - "grad_norm": 0.13134482502937317, - "learning_rate": 2.431818181818182e-05, - "loss": 0.0144, - "step": 350 - }, - { - "epoch": 5.254681647940075, - "grad_norm": 0.1282089650630951, - "learning_rate": 2.4166666666666667e-05, - "loss": 0.0168, - "step": 352 - }, - { - "epoch": 5.284644194756554, - "grad_norm": 0.12280598282814026, - "learning_rate": 2.4015151515151516e-05, - "loss": 0.0133, - "step": 354 - }, - { - "epoch": 5.314606741573034, - "grad_norm": 0.1045297384262085, - "learning_rate": 2.3863636363636365e-05, - "loss": 0.0135, - "step": 356 - }, - { - "epoch": 5.344569288389513, - "grad_norm": 0.10783874243497849, - "learning_rate": 2.3712121212121214e-05, - "loss": 0.0147, - "step": 358 - }, - { - "epoch": 5.3745318352059925, - "grad_norm": 0.28152090311050415, - "learning_rate": 2.356060606060606e-05, - "loss": 0.0155, - "step": 360 - }, - { - "epoch": 5.404494382022472, - "grad_norm": 0.12164758890867233, - "learning_rate": 2.340909090909091e-05, - "loss": 0.0138, - "step": 362 - }, - { - "epoch": 5.4344569288389515, - "grad_norm": 0.16581693291664124, - "learning_rate": 2.325757575757576e-05, - "loss": 0.0179, - "step": 364 - }, - { - "epoch": 5.464419475655431, - "grad_norm": 0.10845212638378143, - "learning_rate": 2.3106060606060605e-05, - "loss": 0.0144, - "step": 366 - }, - { - "epoch": 5.49438202247191, - "grad_norm": 0.11847496032714844, - "learning_rate": 2.2954545454545457e-05, - "loss": 0.0132, - "step": 368 - }, - { - "epoch": 5.52434456928839, - "grad_norm": 0.1990269422531128, - "learning_rate": 2.2803030303030303e-05, - "loss": 0.0179, - "step": 370 - }, - { - "epoch": 5.554307116104869, - "grad_norm": 0.13564549386501312, - "learning_rate": 2.2651515151515152e-05, - "loss": 0.0155, - "step": 372 - }, - { - "epoch": 5.584269662921348, - "grad_norm": 0.12416112422943115, - "learning_rate": 2.25e-05, - "loss": 0.0142, - "step": 374 - }, - { - "epoch": 5.614232209737827, - "grad_norm": 0.21298328042030334, - "learning_rate": 2.234848484848485e-05, - "loss": 0.0179, - "step": 376 - }, - { - "epoch": 5.644194756554307, - "grad_norm": 0.12686102092266083, - "learning_rate": 2.21969696969697e-05, - "loss": 0.0142, - "step": 378 - }, - { - "epoch": 5.674157303370786, - "grad_norm": 0.11951331794261932, - "learning_rate": 2.2045454545454546e-05, - "loss": 0.0158, - "step": 380 - }, - { - "epoch": 5.704119850187266, - "grad_norm": 0.15928389132022858, - "learning_rate": 2.1893939393939395e-05, - "loss": 0.0153, - "step": 382 - }, - { - "epoch": 5.734082397003745, - "grad_norm": 0.11539971828460693, - "learning_rate": 2.1742424242424244e-05, - "loss": 0.0132, - "step": 384 - }, - { - "epoch": 5.764044943820225, - "grad_norm": 0.1547948122024536, - "learning_rate": 2.1590909090909093e-05, - "loss": 0.0132, - "step": 386 - }, - { - "epoch": 5.794007490636704, - "grad_norm": 0.15363723039627075, - "learning_rate": 2.143939393939394e-05, - "loss": 0.0188, - "step": 388 - }, - { - "epoch": 5.823970037453184, - "grad_norm": 0.13726523518562317, - "learning_rate": 2.128787878787879e-05, - "loss": 0.0155, - "step": 390 - }, - { - "epoch": 5.853932584269663, - "grad_norm": 0.11726450175046921, - "learning_rate": 2.1136363636363638e-05, - "loss": 0.0154, - "step": 392 - }, - { - "epoch": 5.883895131086143, - "grad_norm": 0.09192411601543427, - "learning_rate": 2.0984848484848483e-05, - "loss": 0.0144, - "step": 394 - }, - { - "epoch": 5.913857677902621, - "grad_norm": 0.12325743585824966, - "learning_rate": 2.0833333333333336e-05, - "loss": 0.0138, - "step": 396 - }, - { - "epoch": 5.943820224719101, - "grad_norm": 0.1755753755569458, - "learning_rate": 2.0681818181818182e-05, - "loss": 0.0159, - "step": 398 - }, - { - "epoch": 5.97378277153558, - "grad_norm": 0.1385994851589203, - "learning_rate": 2.053030303030303e-05, - "loss": 0.016, - "step": 400 - }, - { - "epoch": 6.0, - "grad_norm": 0.12166863679885864, - "learning_rate": 2.037878787878788e-05, - "loss": 0.0137, - "step": 402 - }, - { - "epoch": 6.0299625468164795, - "grad_norm": 0.135303795337677, - "learning_rate": 2.022727272727273e-05, - "loss": 0.0146, - "step": 404 - }, - { - "epoch": 6.059925093632959, - "grad_norm": 0.14891491830348969, - "learning_rate": 2.0075757575757575e-05, - "loss": 0.016, - "step": 406 - }, - { - "epoch": 6.089887640449438, - "grad_norm": 0.13520725071430206, - "learning_rate": 1.9924242424242425e-05, - "loss": 0.017, - "step": 408 - }, - { - "epoch": 6.119850187265918, - "grad_norm": 0.1810332089662552, - "learning_rate": 1.9772727272727274e-05, - "loss": 0.0161, - "step": 410 - }, - { - "epoch": 6.149812734082397, - "grad_norm": 0.12100538611412048, - "learning_rate": 1.962121212121212e-05, - "loss": 0.0143, - "step": 412 - }, - { - "epoch": 6.179775280898877, - "grad_norm": 0.11748135834932327, - "learning_rate": 1.9469696969696972e-05, - "loss": 0.0148, - "step": 414 - }, - { - "epoch": 6.209737827715355, - "grad_norm": 0.1747063398361206, - "learning_rate": 1.9318181818181818e-05, - "loss": 0.0153, - "step": 416 - }, - { - "epoch": 6.239700374531835, - "grad_norm": 0.15986011922359467, - "learning_rate": 1.9166666666666667e-05, - "loss": 0.0153, - "step": 418 - }, - { - "epoch": 6.269662921348314, - "grad_norm": 0.130351260304451, - "learning_rate": 1.9015151515151516e-05, - "loss": 0.0131, - "step": 420 - }, - { - "epoch": 6.299625468164794, - "grad_norm": 0.09828981012105942, - "learning_rate": 1.8863636363636362e-05, - "loss": 0.0132, - "step": 422 - }, - { - "epoch": 6.329588014981273, - "grad_norm": 0.11312004178762436, - "learning_rate": 1.8712121212121215e-05, - "loss": 0.0137, - "step": 424 - }, - { - "epoch": 6.359550561797753, - "grad_norm": 0.1399003267288208, - "learning_rate": 1.856060606060606e-05, - "loss": 0.0155, - "step": 426 - }, - { - "epoch": 6.389513108614232, - "grad_norm": 0.19829927384853363, - "learning_rate": 1.840909090909091e-05, - "loss": 0.0181, - "step": 428 - }, - { - "epoch": 6.419475655430712, - "grad_norm": 0.12163085490465164, - "learning_rate": 1.825757575757576e-05, - "loss": 0.0143, - "step": 430 - }, - { - "epoch": 6.449438202247191, - "grad_norm": 0.1325439214706421, - "learning_rate": 1.810606060606061e-05, - "loss": 0.0147, - "step": 432 - }, - { - "epoch": 6.479400749063671, - "grad_norm": 0.10004521906375885, - "learning_rate": 1.7954545454545454e-05, - "loss": 0.0126, - "step": 434 - }, - { - "epoch": 6.50936329588015, - "grad_norm": 0.22187596559524536, - "learning_rate": 1.7803030303030303e-05, - "loss": 0.0162, - "step": 436 - }, - { - "epoch": 6.539325842696629, - "grad_norm": 0.12271460145711899, - "learning_rate": 1.7651515151515153e-05, - "loss": 0.013, - "step": 438 - }, - { - "epoch": 6.569288389513108, - "grad_norm": 0.16459853947162628, - "learning_rate": 1.75e-05, - "loss": 0.0145, - "step": 440 - }, - { - "epoch": 6.599250936329588, - "grad_norm": 0.11979979276657104, - "learning_rate": 1.734848484848485e-05, - "loss": 0.0135, - "step": 442 - }, - { - "epoch": 6.629213483146067, - "grad_norm": 0.24495816230773926, - "learning_rate": 1.7196969696969697e-05, - "loss": 0.0164, - "step": 444 - }, - { - "epoch": 6.659176029962547, - "grad_norm": 0.1692766696214676, - "learning_rate": 1.7045454545454546e-05, - "loss": 0.0152, - "step": 446 - }, - { - "epoch": 6.689138576779026, - "grad_norm": 0.1197497546672821, - "learning_rate": 1.6893939393939395e-05, - "loss": 0.0153, - "step": 448 - }, - { - "epoch": 6.719101123595506, - "grad_norm": 0.09570708870887756, - "learning_rate": 1.674242424242424e-05, - "loss": 0.0134, - "step": 450 - }, - { - "epoch": 6.749063670411985, - "grad_norm": 0.10233695805072784, - "learning_rate": 1.6590909090909094e-05, - "loss": 0.0135, - "step": 452 - }, - { - "epoch": 6.7790262172284645, - "grad_norm": 0.13967226445674896, - "learning_rate": 1.643939393939394e-05, - "loss": 0.0159, - "step": 454 - }, - { - "epoch": 6.808988764044944, - "grad_norm": 0.14382530748844147, - "learning_rate": 1.628787878787879e-05, - "loss": 0.0141, - "step": 456 - }, - { - "epoch": 6.8389513108614235, - "grad_norm": 0.18934357166290283, - "learning_rate": 1.6136363636363638e-05, - "loss": 0.015, - "step": 458 - }, - { - "epoch": 6.868913857677903, - "grad_norm": 0.09580834209918976, - "learning_rate": 1.5984848484848487e-05, - "loss": 0.0141, - "step": 460 - }, - { - "epoch": 6.898876404494382, - "grad_norm": 0.16412261128425598, - "learning_rate": 1.5833333333333333e-05, - "loss": 0.0157, - "step": 462 - }, - { - "epoch": 6.928838951310862, - "grad_norm": 0.16064710915088654, - "learning_rate": 1.5681818181818182e-05, - "loss": 0.0141, - "step": 464 - }, - { - "epoch": 6.9588014981273405, - "grad_norm": 0.17852792143821716, - "learning_rate": 1.553030303030303e-05, - "loss": 0.0156, - "step": 466 - }, - { - "epoch": 6.98876404494382, - "grad_norm": 0.17345724999904633, - "learning_rate": 1.5378787878787877e-05, - "loss": 0.0169, - "step": 468 - }, - { - "epoch": 7.01498127340824, - "grad_norm": 0.1276731640100479, - "learning_rate": 1.5227272727272728e-05, - "loss": 0.0143, - "step": 470 - }, - { - "epoch": 7.044943820224719, - "grad_norm": 0.1590038686990738, - "learning_rate": 1.5075757575757576e-05, - "loss": 0.0158, - "step": 472 - }, - { - "epoch": 7.074906367041199, - "grad_norm": 0.12557406723499298, - "learning_rate": 1.4924242424242423e-05, - "loss": 0.0136, - "step": 474 - }, - { - "epoch": 7.104868913857678, - "grad_norm": 0.21979106962680817, - "learning_rate": 1.4772727272727274e-05, - "loss": 0.0159, - "step": 476 - }, - { - "epoch": 7.134831460674158, - "grad_norm": 0.19031833112239838, - "learning_rate": 1.4621212121212122e-05, - "loss": 0.0155, - "step": 478 - }, - { - "epoch": 7.164794007490637, - "grad_norm": 0.23122136294841766, - "learning_rate": 1.446969696969697e-05, - "loss": 0.0171, - "step": 480 - }, - { - "epoch": 7.194756554307116, - "grad_norm": 0.12799808382987976, - "learning_rate": 1.431818181818182e-05, - "loss": 0.0142, - "step": 482 - }, - { - "epoch": 7.224719101123595, - "grad_norm": 0.09180168062448502, - "learning_rate": 1.4166666666666668e-05, - "loss": 0.0133, - "step": 484 - }, - { - "epoch": 7.254681647940075, - "grad_norm": 0.13481584191322327, - "learning_rate": 1.4015151515151515e-05, - "loss": 0.014, - "step": 486 - }, - { - "epoch": 7.284644194756554, - "grad_norm": 0.1296118050813675, - "learning_rate": 1.3863636363636364e-05, - "loss": 0.015, - "step": 488 - }, - { - "epoch": 7.314606741573034, - "grad_norm": 0.12827853858470917, - "learning_rate": 1.3712121212121212e-05, - "loss": 0.0143, - "step": 490 - }, - { - "epoch": 7.344569288389513, - "grad_norm": 0.12804710865020752, - "learning_rate": 1.3560606060606063e-05, - "loss": 0.0144, - "step": 492 - }, - { - "epoch": 7.3745318352059925, - "grad_norm": 0.14896075427532196, - "learning_rate": 1.340909090909091e-05, - "loss": 0.0145, - "step": 494 - }, - { - "epoch": 7.404494382022472, - "grad_norm": 0.11220405250787735, - "learning_rate": 1.3257575757575758e-05, - "loss": 0.0156, - "step": 496 - }, - { - "epoch": 7.4344569288389515, - "grad_norm": 0.10928992182016373, - "learning_rate": 1.3106060606060607e-05, - "loss": 0.0141, - "step": 498 - }, - { - "epoch": 7.464419475655431, - "grad_norm": 0.12223732471466064, - "learning_rate": 1.2954545454545455e-05, - "loss": 0.0146, - "step": 500 - }, - { - "epoch": 7.49438202247191, - "grad_norm": 0.12335951626300812, - "learning_rate": 1.2803030303030302e-05, - "loss": 0.0141, - "step": 502 - }, - { - "epoch": 7.52434456928839, - "grad_norm": 0.12804357707500458, - "learning_rate": 1.2651515151515153e-05, - "loss": 0.0128, - "step": 504 - }, - { - "epoch": 7.554307116104869, - "grad_norm": 0.11515405774116516, - "learning_rate": 1.25e-05, - "loss": 0.0128, - "step": 506 - }, - { - "epoch": 7.584269662921348, - "grad_norm": 0.1302037239074707, - "learning_rate": 1.234848484848485e-05, - "loss": 0.0161, - "step": 508 - }, - { - "epoch": 7.614232209737827, - "grad_norm": 0.1083846315741539, - "learning_rate": 1.2196969696969697e-05, - "loss": 0.0149, - "step": 510 - }, - { - "epoch": 7.644194756554307, - "grad_norm": 0.11610821634531021, - "learning_rate": 1.2045454545454547e-05, - "loss": 0.0138, - "step": 512 - }, - { - "epoch": 7.674157303370786, - "grad_norm": 0.08522523194551468, - "learning_rate": 1.1893939393939394e-05, - "loss": 0.0132, - "step": 514 - }, - { - "epoch": 7.704119850187266, - "grad_norm": 0.0896814838051796, - "learning_rate": 1.1742424242424243e-05, - "loss": 0.0131, - "step": 516 - }, - { - "epoch": 7.734082397003745, - "grad_norm": 0.15006373822689056, - "learning_rate": 1.159090909090909e-05, - "loss": 0.0162, - "step": 518 - }, - { - "epoch": 7.764044943820225, - "grad_norm": 0.1151801198720932, - "learning_rate": 1.143939393939394e-05, - "loss": 0.015, - "step": 520 - }, - { - "epoch": 7.794007490636704, - "grad_norm": 0.10852045565843582, - "learning_rate": 1.128787878787879e-05, - "loss": 0.0132, - "step": 522 - }, - { - "epoch": 7.823970037453184, - "grad_norm": 0.20251013338565826, - "learning_rate": 1.1136363636363637e-05, - "loss": 0.0144, - "step": 524 - }, - { - "epoch": 7.853932584269663, - "grad_norm": 0.1366216391324997, - "learning_rate": 1.0984848484848486e-05, - "loss": 0.0157, - "step": 526 - }, - { - "epoch": 7.883895131086143, - "grad_norm": 0.12531571090221405, - "learning_rate": 1.0833333333333334e-05, - "loss": 0.015, - "step": 528 - }, - { - "epoch": 7.913857677902621, - "grad_norm": 0.14330235123634338, - "learning_rate": 1.0681818181818181e-05, - "loss": 0.0125, - "step": 530 - }, - { - "epoch": 7.943820224719101, - "grad_norm": 0.1384236365556717, - "learning_rate": 1.053030303030303e-05, - "loss": 0.014, - "step": 532 - }, - { - "epoch": 7.97378277153558, - "grad_norm": 0.15142616629600525, - "learning_rate": 1.037878787878788e-05, - "loss": 0.0175, - "step": 534 - }, - { - "epoch": 8.0, - "grad_norm": 0.11959421634674072, - "learning_rate": 1.0227272727272729e-05, - "loss": 0.0144, - "step": 536 - }, - { - "epoch": 8.02996254681648, - "grad_norm": 0.14299067854881287, - "learning_rate": 1.0075757575757576e-05, - "loss": 0.0121, - "step": 538 - }, - { - "epoch": 8.059925093632959, - "grad_norm": 0.15421266853809357, - "learning_rate": 9.924242424242425e-06, - "loss": 0.0157, - "step": 540 - }, - { - "epoch": 8.089887640449438, - "grad_norm": 0.11853787302970886, - "learning_rate": 9.772727272727273e-06, - "loss": 0.0141, - "step": 542 - }, - { - "epoch": 8.119850187265918, - "grad_norm": 0.11856868863105774, - "learning_rate": 9.62121212121212e-06, - "loss": 0.0145, - "step": 544 - }, - { - "epoch": 8.149812734082397, - "grad_norm": 0.10615833103656769, - "learning_rate": 9.46969696969697e-06, - "loss": 0.0146, - "step": 546 - }, - { - "epoch": 8.179775280898877, - "grad_norm": 0.14076802134513855, - "learning_rate": 9.318181818181819e-06, - "loss": 0.014, - "step": 548 - }, - { - "epoch": 8.209737827715356, - "grad_norm": 0.12974779307842255, - "learning_rate": 9.166666666666666e-06, - "loss": 0.0139, - "step": 550 - }, - { - "epoch": 8.239700374531836, - "grad_norm": 0.11985109746456146, - "learning_rate": 9.015151515151516e-06, - "loss": 0.0148, - "step": 552 - }, - { - "epoch": 8.269662921348315, - "grad_norm": 0.21702255308628082, - "learning_rate": 8.863636363636365e-06, - "loss": 0.0134, - "step": 554 - }, - { - "epoch": 8.299625468164795, - "grad_norm": 0.14472782611846924, - "learning_rate": 8.712121212121212e-06, - "loss": 0.0151, - "step": 556 - }, - { - "epoch": 8.329588014981274, - "grad_norm": 0.1377476155757904, - "learning_rate": 8.56060606060606e-06, - "loss": 0.0139, - "step": 558 - }, - { - "epoch": 8.359550561797754, - "grad_norm": 0.15174493193626404, - "learning_rate": 8.409090909090909e-06, - "loss": 0.0129, - "step": 560 - }, - { - "epoch": 8.389513108614231, - "grad_norm": 0.1872493326663971, - "learning_rate": 8.257575757575758e-06, - "loss": 0.0119, - "step": 562 - }, - { - "epoch": 8.41947565543071, - "grad_norm": 0.1931612491607666, - "learning_rate": 8.106060606060606e-06, - "loss": 0.0175, - "step": 564 - }, - { - "epoch": 8.44943820224719, - "grad_norm": 0.18126638233661652, - "learning_rate": 7.954545454545455e-06, - "loss": 0.0153, - "step": 566 - }, - { - "epoch": 8.47940074906367, - "grad_norm": 0.16266301274299622, - "learning_rate": 7.803030303030304e-06, - "loss": 0.0143, - "step": 568 - }, - { - "epoch": 8.50936329588015, - "grad_norm": 0.1196930930018425, - "learning_rate": 7.651515151515152e-06, - "loss": 0.015, - "step": 570 - }, - { - "epoch": 8.539325842696629, - "grad_norm": 0.14626623690128326, - "learning_rate": 7.5e-06, - "loss": 0.0154, - "step": 572 - }, - { - "epoch": 8.569288389513108, - "grad_norm": 0.12978863716125488, - "learning_rate": 7.3484848484848486e-06, - "loss": 0.0141, - "step": 574 - }, - { - "epoch": 8.599250936329588, - "grad_norm": 0.16399815678596497, - "learning_rate": 7.196969696969698e-06, - "loss": 0.0151, - "step": 576 - }, - { - "epoch": 8.629213483146067, - "grad_norm": 0.13139177858829498, - "learning_rate": 7.045454545454545e-06, - "loss": 0.0132, - "step": 578 - }, - { - "epoch": 8.659176029962547, - "grad_norm": 0.10869726538658142, - "learning_rate": 6.8939393939393945e-06, - "loss": 0.0143, - "step": 580 - }, - { - "epoch": 8.689138576779026, - "grad_norm": 0.11966531723737717, - "learning_rate": 6.742424242424243e-06, - "loss": 0.0153, - "step": 582 - }, - { - "epoch": 8.719101123595506, - "grad_norm": 0.15545986592769623, - "learning_rate": 6.59090909090909e-06, - "loss": 0.0133, - "step": 584 - }, - { - "epoch": 8.749063670411985, - "grad_norm": 0.1065947413444519, - "learning_rate": 6.43939393939394e-06, - "loss": 0.0146, - "step": 586 - }, - { - "epoch": 8.779026217228465, - "grad_norm": 0.17026154696941376, - "learning_rate": 6.287878787878789e-06, - "loss": 0.0152, - "step": 588 - }, - { - "epoch": 8.808988764044944, - "grad_norm": 0.1124507337808609, - "learning_rate": 6.136363636363636e-06, - "loss": 0.0143, - "step": 590 - }, - { - "epoch": 8.838951310861423, - "grad_norm": 0.11526783555746078, - "learning_rate": 5.984848484848485e-06, - "loss": 0.0143, - "step": 592 - }, - { - "epoch": 8.868913857677903, - "grad_norm": 0.11815094202756882, - "learning_rate": 5.833333333333334e-06, - "loss": 0.0142, - "step": 594 - }, - { - "epoch": 8.898876404494382, - "grad_norm": 0.115007683634758, - "learning_rate": 5.681818181818182e-06, - "loss": 0.0148, - "step": 596 - }, - { - "epoch": 8.928838951310862, - "grad_norm": 0.12640294432640076, - "learning_rate": 5.530303030303031e-06, - "loss": 0.0156, - "step": 598 - }, - { - "epoch": 8.958801498127341, - "grad_norm": 0.18849849700927734, - "learning_rate": 5.378787878787879e-06, - "loss": 0.0146, - "step": 600 - }, - { - "epoch": 8.98876404494382, - "grad_norm": 0.09228724241256714, - "learning_rate": 5.2272727272727274e-06, - "loss": 0.0132, - "step": 602 - }, - { - "epoch": 9.014981273408239, - "grad_norm": 0.10718464851379395, - "learning_rate": 5.075757575757576e-06, - "loss": 0.0142, - "step": 604 - }, - { - "epoch": 9.044943820224718, - "grad_norm": 0.15939927101135254, - "learning_rate": 4.924242424242424e-06, - "loss": 0.015, - "step": 606 - }, - { - "epoch": 9.074906367041198, - "grad_norm": 0.10619861632585526, - "learning_rate": 4.772727272727273e-06, - "loss": 0.0139, - "step": 608 - }, - { - "epoch": 9.104868913857677, - "grad_norm": 0.1796608418226242, - "learning_rate": 4.621212121212122e-06, - "loss": 0.0157, - "step": 610 - }, - { - "epoch": 9.134831460674157, - "grad_norm": 0.11252462863922119, - "learning_rate": 4.46969696969697e-06, - "loss": 0.0152, - "step": 612 - }, - { - "epoch": 9.164794007490636, - "grad_norm": 0.12761755287647247, - "learning_rate": 4.3181818181818185e-06, - "loss": 0.0143, - "step": 614 - }, - { - "epoch": 9.194756554307116, - "grad_norm": 0.11963596194982529, - "learning_rate": 4.166666666666667e-06, - "loss": 0.0128, - "step": 616 - }, - { - "epoch": 9.224719101123595, - "grad_norm": 0.14573565125465393, - "learning_rate": 4.015151515151515e-06, - "loss": 0.0156, - "step": 618 - }, - { - "epoch": 9.254681647940075, - "grad_norm": 0.1269391030073166, - "learning_rate": 3.863636363636364e-06, - "loss": 0.0135, - "step": 620 - }, - { - "epoch": 9.284644194756554, - "grad_norm": 0.14515313506126404, - "learning_rate": 3.7121212121212124e-06, - "loss": 0.0151, - "step": 622 - }, - { - "epoch": 9.314606741573034, - "grad_norm": 0.1334671527147293, - "learning_rate": 3.5606060606060608e-06, - "loss": 0.015, - "step": 624 - }, - { - "epoch": 9.344569288389513, - "grad_norm": 0.12002371996641159, - "learning_rate": 3.409090909090909e-06, - "loss": 0.0141, - "step": 626 - }, - { - "epoch": 9.374531835205993, - "grad_norm": 0.14389854669570923, - "learning_rate": 3.257575757575758e-06, - "loss": 0.0136, - "step": 628 - }, - { - "epoch": 9.404494382022472, - "grad_norm": 0.11606308072805405, - "learning_rate": 3.106060606060606e-06, - "loss": 0.0152, - "step": 630 - }, - { - "epoch": 9.434456928838951, - "grad_norm": 0.18154092133045197, - "learning_rate": 2.9545454545454547e-06, - "loss": 0.0137, - "step": 632 - }, - { - "epoch": 9.464419475655431, - "grad_norm": 0.2114187330007553, - "learning_rate": 2.803030303030303e-06, - "loss": 0.0145, - "step": 634 - }, - { - "epoch": 9.49438202247191, - "grad_norm": 0.12829148769378662, - "learning_rate": 2.651515151515152e-06, - "loss": 0.0137, - "step": 636 - }, - { - "epoch": 9.52434456928839, - "grad_norm": 0.14554138481616974, - "learning_rate": 2.5e-06, - "loss": 0.0122, - "step": 638 - }, - { - "epoch": 9.55430711610487, - "grad_norm": 0.11862059682607651, - "learning_rate": 2.3484848484848486e-06, - "loss": 0.0137, - "step": 640 - }, - { - "epoch": 9.584269662921349, - "grad_norm": 0.11568762362003326, - "learning_rate": 2.196969696969697e-06, - "loss": 0.0147, - "step": 642 - }, - { - "epoch": 9.614232209737828, - "grad_norm": 0.10297347605228424, - "learning_rate": 2.0454545454545457e-06, - "loss": 0.0139, - "step": 644 - }, - { - "epoch": 9.644194756554308, - "grad_norm": 0.14339599013328552, - "learning_rate": 1.8939393939393941e-06, - "loss": 0.0138, - "step": 646 - }, - { - "epoch": 9.674157303370787, - "grad_norm": 0.15518030524253845, - "learning_rate": 1.7424242424242427e-06, - "loss": 0.0127, - "step": 648 - }, - { - "epoch": 9.704119850187267, - "grad_norm": 0.13086441159248352, - "learning_rate": 1.5909090909090908e-06, - "loss": 0.013, - "step": 650 - }, - { - "epoch": 9.734082397003746, - "grad_norm": 0.1343931257724762, - "learning_rate": 1.4393939393939396e-06, - "loss": 0.014, - "step": 652 - }, - { - "epoch": 9.764044943820224, - "grad_norm": 0.16945284605026245, - "learning_rate": 1.287878787878788e-06, - "loss": 0.0153, - "step": 654 - }, - { - "epoch": 9.794007490636703, - "grad_norm": 0.14710381627082825, - "learning_rate": 1.1363636363636364e-06, - "loss": 0.0145, - "step": 656 - }, - { - "epoch": 9.823970037453183, - "grad_norm": 0.15412309765815735, - "learning_rate": 9.848484848484847e-07, - "loss": 0.0147, - "step": 658 - }, - { - "epoch": 9.853932584269662, - "grad_norm": 0.1276121288537979, - "learning_rate": 8.333333333333333e-07, - "loss": 0.0157, - "step": 660 - }, - { - "epoch": 9.883895131086142, - "grad_norm": 0.17896993458271027, - "learning_rate": 6.818181818181818e-07, - "loss": 0.0154, - "step": 662 - }, - { - "epoch": 9.913857677902621, - "grad_norm": 0.18712849915027618, - "learning_rate": 5.303030303030304e-07, - "loss": 0.0133, - "step": 664 - }, - { - "epoch": 9.9438202247191, - "grad_norm": 0.14763027429580688, - "learning_rate": 3.787878787878788e-07, - "loss": 0.0142, - "step": 666 - }, - { - "epoch": 9.97378277153558, - "grad_norm": 0.13407444953918457, - "learning_rate": 2.2727272727272726e-07, - "loss": 0.0144, - "step": 668 - }, - { - "epoch": 10.0, - "grad_norm": 0.18560239672660828, - "learning_rate": 7.575757575757576e-08, - "loss": 0.0139, - "step": 670 - } - ], - "logging_steps": 2, - "max_steps": 670, - "num_input_tokens_seen": 0, - "num_train_epochs": 10, - "save_steps": 20, - "stateful_callbacks": { - "TrainerControl": { - "args": { - "should_epoch_stop": false, - "should_evaluate": false, - "should_log": false, - "should_save": true, - "should_training_stop": true - }, - "attributes": {} - } - }, - "total_flos": 1.804222263410688e+16, - "train_batch_size": 1, - "trial_name": null, - "trial_params": null -} diff --git a/smollm3_robust/checkpoint-670/training_args.bin b/smollm3_robust/checkpoint-670/training_args.bin deleted file mode 100644 index 288236fd628e342cba9dba026dc49e62bfa02807..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/checkpoint-670/training_args.bin and /dev/null differ diff --git a/smollm3_robust/runs/Jul21_11-16-42_Jasons-MacBook-Pro.local/events.out.tfevents.1753093003.Jasons-MacBook-Pro.local.20775.0 b/smollm3_robust/runs/Jul21_11-16-42_Jasons-MacBook-Pro.local/events.out.tfevents.1753093003.Jasons-MacBook-Pro.local.20775.0 deleted file mode 100644 index 23fe94a8c9c238bdcf7a98a8b2decc275c413507..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/runs/Jul21_11-16-42_Jasons-MacBook-Pro.local/events.out.tfevents.1753093003.Jasons-MacBook-Pro.local.20775.0 and /dev/null differ diff --git a/smollm3_robust/runs/Jul21_12-08-44_Jasons-MacBook-Pro.local/events.out.tfevents.1753096125.Jasons-MacBook-Pro.local.28161.0 b/smollm3_robust/runs/Jul21_12-08-44_Jasons-MacBook-Pro.local/events.out.tfevents.1753096125.Jasons-MacBook-Pro.local.28161.0 deleted file mode 100644 index bd08f08a59af6b9fb53da969066ecde4eec1a569..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/runs/Jul21_12-08-44_Jasons-MacBook-Pro.local/events.out.tfevents.1753096125.Jasons-MacBook-Pro.local.28161.0 and /dev/null differ diff --git a/smollm3_robust/runs/Jul21_12-14-15_Jasons-MacBook-Pro.local/events.out.tfevents.1753096456.Jasons-MacBook-Pro.local.29348.0 b/smollm3_robust/runs/Jul21_12-14-15_Jasons-MacBook-Pro.local/events.out.tfevents.1753096456.Jasons-MacBook-Pro.local.29348.0 deleted file mode 100644 index 4a54deeab93e78c1be30e7189ef391b04dc8aa61..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/runs/Jul21_12-14-15_Jasons-MacBook-Pro.local/events.out.tfevents.1753096456.Jasons-MacBook-Pro.local.29348.0 and /dev/null differ diff --git a/smollm3_robust/runs/Jul21_12-58-00_Jasons-MacBook-Pro.local/events.out.tfevents.1753099081.Jasons-MacBook-Pro.local.31977.0 b/smollm3_robust/runs/Jul21_12-58-00_Jasons-MacBook-Pro.local/events.out.tfevents.1753099081.Jasons-MacBook-Pro.local.31977.0 deleted file mode 100644 index c01227254e819311b6e4be24a90e4ac7db59ca4b..0000000000000000000000000000000000000000 Binary files a/smollm3_robust/runs/Jul21_12-58-00_Jasons-MacBook-Pro.local/events.out.tfevents.1753099081.Jasons-MacBook-Pro.local.31977.0 and /dev/null differ diff --git a/smollm3_robust/special_tokens_map.json b/smollm3_robust/special_tokens_map.json deleted file mode 100644 index 190d5624dbbc1ad56f2f34c9d58e03fef7e5328b..0000000000000000000000000000000000000000 --- a/smollm3_robust/special_tokens_map.json +++ /dev/null @@ -1,16 +0,0 @@ -{ - "eos_token": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false - }, - "pad_token": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false - } -} diff --git a/smollm3_robust/tokenizer.json.REMOVED.git-id b/smollm3_robust/tokenizer.json.REMOVED.git-id deleted file mode 100644 index 3081f40cf643c6511c69d3e37e6f8238a5cfd53e..0000000000000000000000000000000000000000 --- a/smollm3_robust/tokenizer.json.REMOVED.git-id +++ /dev/null @@ -1 +0,0 @@ -b70d58b80eb0efbdad29dcf2df4ea60dca527d44 \ No newline at end of file diff --git a/smollm3_robust/tokenizer_config.json b/smollm3_robust/tokenizer_config.json deleted file mode 100644 index 61910c2db5cbdc9e6a6f37e14aaf00584cc6ad47..0000000000000000000000000000000000000000 --- a/smollm3_robust/tokenizer_config.json +++ /dev/null @@ -1,2064 +0,0 @@ -{ - "added_tokens_decoder": { - "128000": { - "content": "<|begin_of_text|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128001": { - "content": "<|end_of_text|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128002": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128003": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128004": { - "content": "<|finetune_right_pad_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128005": { - "content": "<|reserved_special_token_2|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128006": { - "content": "<|start_header_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128007": { - "content": "<|end_header_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128008": { - "content": "<|eom_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128009": { - "content": "<|eot_id|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128010": { - "content": "<|python_tag|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128011": { - "content": "<|im_start|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128012": { - "content": "<|im_end|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128013": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128014": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128015": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128016": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128017": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128018": { - "content": "", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": false - }, - "128019": { - "content": "<|reserved_special_token_11|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128020": { - "content": "<|reserved_special_token_12|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128021": { - "content": "<|reserved_special_token_13|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128022": { - "content": "<|reserved_special_token_14|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128023": { - "content": "<|reserved_special_token_15|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128024": { - "content": "<|reserved_special_token_16|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128025": { - "content": "<|reserved_special_token_17|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128026": { - "content": "<|reserved_special_token_18|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128027": { - "content": "<|reserved_special_token_19|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128028": { - "content": "<|reserved_special_token_20|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128029": { - "content": "<|reserved_special_token_21|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128030": { - "content": "<|reserved_special_token_22|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128031": { - "content": "<|reserved_special_token_23|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128032": { - "content": "<|reserved_special_token_24|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128033": { - "content": "<|reserved_special_token_25|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128034": { - "content": "<|reserved_special_token_26|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128035": { - "content": "<|reserved_special_token_27|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128036": { - "content": "<|reserved_special_token_28|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128037": { - "content": "<|reserved_special_token_29|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128038": { - "content": "<|reserved_special_token_30|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128039": { - "content": "<|reserved_special_token_31|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128040": { - "content": "<|reserved_special_token_32|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128041": { - "content": "<|reserved_special_token_33|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128042": { - "content": "<|reserved_special_token_34|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128043": { - "content": "<|reserved_special_token_35|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128044": { - "content": "<|reserved_special_token_36|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128045": { - "content": "<|reserved_special_token_37|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128046": { - "content": "<|reserved_special_token_38|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128047": { - "content": "<|reserved_special_token_39|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128048": { - "content": "<|reserved_special_token_40|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128049": { - "content": "<|reserved_special_token_41|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128050": { - "content": "<|reserved_special_token_42|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128051": { - "content": "<|reserved_special_token_43|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128052": { - "content": "<|reserved_special_token_44|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128053": { - "content": "<|reserved_special_token_45|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128054": { - "content": "<|reserved_special_token_46|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128055": { - "content": "<|reserved_special_token_47|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128056": { - "content": "<|reserved_special_token_48|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128057": { - "content": "<|reserved_special_token_49|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128058": { - "content": "<|reserved_special_token_50|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128059": { - "content": "<|reserved_special_token_51|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128060": { - "content": "<|reserved_special_token_52|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128061": { - "content": "<|reserved_special_token_53|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128062": { - "content": "<|reserved_special_token_54|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128063": { - "content": "<|reserved_special_token_55|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128064": { - "content": "<|reserved_special_token_56|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128065": { - "content": "<|reserved_special_token_57|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128066": { - "content": "<|reserved_special_token_58|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128067": { - "content": "<|reserved_special_token_59|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128068": { - "content": "<|reserved_special_token_60|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128069": { - "content": "<|reserved_special_token_61|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128070": { - "content": "<|reserved_special_token_62|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128071": { - "content": "<|reserved_special_token_63|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128072": { - "content": "<|reserved_special_token_64|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128073": { - "content": "<|reserved_special_token_65|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128074": { - "content": "<|reserved_special_token_66|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128075": { - "content": "<|reserved_special_token_67|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128076": { - "content": "<|reserved_special_token_68|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128077": { - "content": "<|reserved_special_token_69|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128078": { - "content": "<|reserved_special_token_70|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128079": { - "content": "<|reserved_special_token_71|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128080": { - "content": "<|reserved_special_token_72|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128081": { - "content": "<|reserved_special_token_73|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128082": { - "content": "<|reserved_special_token_74|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128083": { - "content": "<|reserved_special_token_75|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128084": { - "content": "<|reserved_special_token_76|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128085": { - "content": "<|reserved_special_token_77|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128086": { - "content": "<|reserved_special_token_78|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128087": { - "content": "<|reserved_special_token_79|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128088": { - "content": "<|reserved_special_token_80|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128089": { - "content": "<|reserved_special_token_81|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128090": { - "content": "<|reserved_special_token_82|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128091": { - "content": "<|reserved_special_token_83|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128092": { - "content": "<|reserved_special_token_84|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128093": { - "content": "<|reserved_special_token_85|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128094": { - "content": "<|reserved_special_token_86|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128095": { - "content": "<|reserved_special_token_87|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128096": { - "content": "<|reserved_special_token_88|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128097": { - "content": "<|reserved_special_token_89|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128098": { - "content": "<|reserved_special_token_90|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128099": { - "content": "<|reserved_special_token_91|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128100": { - "content": "<|reserved_special_token_92|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128101": { - "content": "<|reserved_special_token_93|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128102": { - "content": "<|reserved_special_token_94|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128103": { - "content": "<|reserved_special_token_95|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128104": { - "content": "<|reserved_special_token_96|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128105": { - "content": "<|reserved_special_token_97|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128106": { - "content": "<|reserved_special_token_98|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128107": { - "content": "<|reserved_special_token_99|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128108": { - "content": "<|reserved_special_token_100|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128109": { - "content": "<|reserved_special_token_101|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128110": { - "content": "<|reserved_special_token_102|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128111": { - "content": "<|reserved_special_token_103|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128112": { - "content": "<|reserved_special_token_104|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128113": { - "content": "<|reserved_special_token_105|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128114": { - "content": "<|reserved_special_token_106|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128115": { - "content": "<|reserved_special_token_107|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128116": { - "content": "<|reserved_special_token_108|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128117": { - "content": "<|reserved_special_token_109|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128118": { - "content": "<|reserved_special_token_110|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128119": { - "content": "<|reserved_special_token_111|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128120": { - "content": "<|reserved_special_token_112|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128121": { - "content": "<|reserved_special_token_113|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128122": { - "content": "<|reserved_special_token_114|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128123": { - "content": "<|reserved_special_token_115|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128124": { - "content": "<|reserved_special_token_116|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128125": { - "content": "<|reserved_special_token_117|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128126": { - "content": "<|reserved_special_token_118|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128127": { - "content": "<|reserved_special_token_119|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128128": { - "content": "<|reserved_special_token_120|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128129": { - "content": "<|reserved_special_token_121|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128130": { - "content": "<|reserved_special_token_122|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128131": { - "content": "<|reserved_special_token_123|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128132": { - "content": "<|reserved_special_token_124|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128133": { - "content": "<|reserved_special_token_125|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128134": { - "content": "<|reserved_special_token_126|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128135": { - "content": "<|reserved_special_token_127|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128136": { - "content": "<|reserved_special_token_128|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128137": { - "content": "<|reserved_special_token_129|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128138": { - "content": "<|reserved_special_token_130|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128139": { - "content": "<|reserved_special_token_131|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128140": { - "content": "<|reserved_special_token_132|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128141": { - "content": "<|reserved_special_token_133|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128142": { - "content": "<|reserved_special_token_134|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128143": { - "content": "<|reserved_special_token_135|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128144": { - "content": "<|reserved_special_token_136|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128145": { - "content": "<|reserved_special_token_137|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128146": { - "content": "<|reserved_special_token_138|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128147": { - "content": "<|reserved_special_token_139|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128148": { - "content": "<|reserved_special_token_140|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128149": { - "content": "<|reserved_special_token_141|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128150": { - "content": "<|reserved_special_token_142|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128151": { - "content": "<|reserved_special_token_143|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128152": { - "content": "<|reserved_special_token_144|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128153": { - "content": "<|reserved_special_token_145|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128154": { - "content": "<|reserved_special_token_146|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128155": { - "content": "<|reserved_special_token_147|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128156": { - "content": "<|reserved_special_token_148|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128157": { - "content": "<|reserved_special_token_149|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128158": { - "content": "<|reserved_special_token_150|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128159": { - "content": "<|reserved_special_token_151|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128160": { - "content": "<|reserved_special_token_152|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128161": { - "content": "<|reserved_special_token_153|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128162": { - "content": "<|reserved_special_token_154|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128163": { - "content": "<|reserved_special_token_155|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128164": { - "content": "<|reserved_special_token_156|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128165": { - "content": "<|reserved_special_token_157|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128166": { - "content": "<|reserved_special_token_158|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128167": { - "content": "<|reserved_special_token_159|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128168": { - "content": "<|reserved_special_token_160|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128169": { - "content": "<|reserved_special_token_161|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128170": { - "content": "<|reserved_special_token_162|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128171": { - "content": "<|reserved_special_token_163|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128172": { - "content": "<|reserved_special_token_164|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128173": { - "content": "<|reserved_special_token_165|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128174": { - "content": "<|reserved_special_token_166|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128175": { - "content": "<|reserved_special_token_167|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128176": { - "content": "<|reserved_special_token_168|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128177": { - "content": "<|reserved_special_token_169|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128178": { - "content": "<|reserved_special_token_170|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128179": { - "content": "<|reserved_special_token_171|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128180": { - "content": "<|reserved_special_token_172|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128181": { - "content": "<|reserved_special_token_173|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128182": { - "content": "<|reserved_special_token_174|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128183": { - "content": "<|reserved_special_token_175|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128184": { - "content": "<|reserved_special_token_176|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128185": { - "content": "<|reserved_special_token_177|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128186": { - "content": "<|reserved_special_token_178|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128187": { - "content": "<|reserved_special_token_179|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128188": { - "content": "<|reserved_special_token_180|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128189": { - "content": "<|reserved_special_token_181|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128190": { - "content": "<|reserved_special_token_182|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128191": { - "content": "<|reserved_special_token_183|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128192": { - "content": "<|reserved_special_token_184|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128193": { - "content": "<|reserved_special_token_185|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128194": { - "content": "<|reserved_special_token_186|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128195": { - "content": "<|reserved_special_token_187|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128196": { - "content": "<|reserved_special_token_188|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128197": { - "content": "<|reserved_special_token_189|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128198": { - "content": "<|reserved_special_token_190|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128199": { - "content": "<|reserved_special_token_191|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128200": { - "content": "<|reserved_special_token_192|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128201": { - "content": "<|reserved_special_token_193|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128202": { - "content": "<|reserved_special_token_194|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128203": { - "content": "<|reserved_special_token_195|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128204": { - "content": "<|reserved_special_token_196|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128205": { - "content": "<|reserved_special_token_197|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128206": { - "content": "<|reserved_special_token_198|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128207": { - "content": "<|reserved_special_token_199|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128208": { - "content": "<|reserved_special_token_200|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128209": { - "content": "<|reserved_special_token_201|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128210": { - "content": "<|reserved_special_token_202|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128211": { - "content": "<|reserved_special_token_203|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128212": { - "content": "<|reserved_special_token_204|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128213": { - "content": "<|reserved_special_token_205|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128214": { - "content": "<|reserved_special_token_206|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128215": { - "content": "<|reserved_special_token_207|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128216": { - "content": "<|reserved_special_token_208|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128217": { - "content": "<|reserved_special_token_209|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128218": { - "content": "<|reserved_special_token_210|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128219": { - "content": "<|reserved_special_token_211|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128220": { - "content": "<|reserved_special_token_212|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128221": { - "content": "<|reserved_special_token_213|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128222": { - "content": "<|reserved_special_token_214|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128223": { - "content": "<|reserved_special_token_215|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128224": { - "content": "<|reserved_special_token_216|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128225": { - "content": "<|reserved_special_token_217|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128226": { - "content": "<|reserved_special_token_218|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128227": { - "content": "<|reserved_special_token_219|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128228": { - "content": "<|reserved_special_token_220|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128229": { - "content": "<|reserved_special_token_221|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128230": { - "content": "<|reserved_special_token_222|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128231": { - "content": "<|reserved_special_token_223|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128232": { - "content": "<|reserved_special_token_224|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128233": { - "content": "<|reserved_special_token_225|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128234": { - "content": "<|reserved_special_token_226|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128235": { - "content": "<|reserved_special_token_227|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128236": { - "content": "<|reserved_special_token_228|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128237": { - "content": "<|reserved_special_token_229|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128238": { - "content": "<|reserved_special_token_230|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128239": { - "content": "<|reserved_special_token_231|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128240": { - "content": "<|reserved_special_token_232|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128241": { - "content": "<|reserved_special_token_233|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128242": { - "content": "<|reserved_special_token_234|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128243": { - "content": "<|reserved_special_token_235|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128244": { - "content": "<|reserved_special_token_236|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128245": { - "content": "<|reserved_special_token_237|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128246": { - "content": "<|reserved_special_token_238|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128247": { - "content": "<|reserved_special_token_239|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128248": { - "content": "<|reserved_special_token_240|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128249": { - "content": "<|reserved_special_token_241|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128250": { - "content": "<|reserved_special_token_242|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128251": { - "content": "<|reserved_special_token_243|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128252": { - "content": "<|reserved_special_token_244|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128253": { - "content": "<|reserved_special_token_245|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128254": { - "content": "<|reserved_special_token_246|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - }, - "128255": { - "content": "<|reserved_special_token_247|>", - "lstrip": false, - "normalized": false, - "rstrip": false, - "single_word": false, - "special": true - } - }, - "bos_token": null, - "clean_up_tokenization_spaces": true, - "eos_token": "<|im_end|>", - "extra_special_tokens": {}, - "fast": false, - "model_input_names": [ - "input_ids", - "attention_mask" - ], - "model_max_length": 131072, - "pad_token": "<|im_end|>", - "tokenizer_class": "PreTrainedTokenizerFast" -}