Upload folder using huggingface_hub
Browse files- train_config.json +76 -1
train_config.json
CHANGED
|
@@ -1 +1,76 @@
|
|
| 1 |
-
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data": {
|
| 3 |
+
"train_files": "data/fineweb_10B_gpt2/fineweb_train_*.bin",
|
| 4 |
+
"train_tokens": null,
|
| 5 |
+
"val_files": "data/fineweb_10B_gpt2/fineweb_val_*.bin",
|
| 6 |
+
"val_tokens": null,
|
| 7 |
+
"batch_size": 512,
|
| 8 |
+
"device_batch_size": 32
|
| 9 |
+
},
|
| 10 |
+
"model": {
|
| 11 |
+
"dim": 768,
|
| 12 |
+
"n_layers": 12,
|
| 13 |
+
"n_heads": 12,
|
| 14 |
+
"n_kv_heads": 12,
|
| 15 |
+
"vocab_size": 50257,
|
| 16 |
+
"multiple_of": 256,
|
| 17 |
+
"ffn_dim_multiplier": 4,
|
| 18 |
+
"norm_eps": 1e-05,
|
| 19 |
+
"rope_theta": 10000,
|
| 20 |
+
"use_scaled_rope": false,
|
| 21 |
+
"max_seq_len": 1024,
|
| 22 |
+
"initializer_range": 0.02,
|
| 23 |
+
"zero_init_masks": false
|
| 24 |
+
},
|
| 25 |
+
"optimizer": {
|
| 26 |
+
"default": {
|
| 27 |
+
"lr": 0.001,
|
| 28 |
+
"beta1": 0.8,
|
| 29 |
+
"beta2": 0.95,
|
| 30 |
+
"eps": 1e-10,
|
| 31 |
+
"weight_decay": 0
|
| 32 |
+
},
|
| 33 |
+
"masks": {
|
| 34 |
+
"lr": 0.001,
|
| 35 |
+
"beta1": 0.8,
|
| 36 |
+
"beta2": 0.95,
|
| 37 |
+
"eps": 1e-10,
|
| 38 |
+
"weight_decay": 0
|
| 39 |
+
},
|
| 40 |
+
"norms": {
|
| 41 |
+
"lr": 0.001,
|
| 42 |
+
"beta1": 0.8,
|
| 43 |
+
"beta2": 0.95,
|
| 44 |
+
"eps": 1e-10,
|
| 45 |
+
"weight_decay": 0
|
| 46 |
+
}
|
| 47 |
+
},
|
| 48 |
+
"scheduler": {
|
| 49 |
+
"warmup_steps": 0.1,
|
| 50 |
+
"start_factor": 0.1
|
| 51 |
+
},
|
| 52 |
+
"gates": {
|
| 53 |
+
"mean_targets": "auto",
|
| 54 |
+
"mean_target_start": 1.0,
|
| 55 |
+
"mean_target_end": 0.7,
|
| 56 |
+
"var_target_delta": 0,
|
| 57 |
+
"mean_coef_init": 0,
|
| 58 |
+
"var_coef_init": 0,
|
| 59 |
+
"coef_max": "+inf",
|
| 60 |
+
"coef_min": "-inf",
|
| 61 |
+
"ema_steps_short": 1,
|
| 62 |
+
"ema_steps_long": 10,
|
| 63 |
+
"delta_min": 0.01,
|
| 64 |
+
"coef_update_multiplier": 0.001
|
| 65 |
+
},
|
| 66 |
+
"gates_zero_eps": 1e-08,
|
| 67 |
+
"seed": 0,
|
| 68 |
+
"project": "fineweb-gated",
|
| 69 |
+
"run_id": null,
|
| 70 |
+
"logdir": "logs/fineweb-gated",
|
| 71 |
+
"log_gradients": false,
|
| 72 |
+
"log_params": false,
|
| 73 |
+
"log_every_steps": 10,
|
| 74 |
+
"val_every_steps": -1,
|
| 75 |
+
"save_every_steps": -1
|
| 76 |
+
}
|