{ "run_info": { "created_at": "2026-01-10T05:55:49+00:00", "total_time": 1348.0078762479825, "experiment_name": "oft/llama-3.2-3B-rank32", "peft_branch": "main", "train_config": { "model_id": "meta-llama/Llama-3.2-3B", "dtype": "bfloat16", "max_seq_length": 768, "batch_size": 4, "batch_size_eval": 50, "max_steps": 5000, "eval_steps": 250, "compile": false, "query_template": "Question: {query} Think step by step.\nAnswer:", "seed": 0, "grad_norm_clip": 1.0, "optimizer_type": "AdamW", "optimizer_kwargs": { "lr": 0.0001, "weight_decay": 0.1 }, "lr_scheduler": "cosine", "use_amp": false, "autocast_adapter_dtype": true, "generation_kwargs": { "max_length": 800, "max_new_tokens": 300 }, "attn_implementation": null }, "peft_config": { "task_type": null, "peft_type": "OFT", "auto_mapping": null, "peft_version": "0.18.1.dev0@UNKNOWN", "base_model_name_or_path": "meta-llama/Llama-3.2-3B", "revision": null, "inference_mode": false, "r": 32, "oft_block_size": 0, "module_dropout": 0.0, "target_modules": [ "v_proj", "q_proj" ], "fan_in_fan_out": false, "bias": "none", "exclude_modules": null, "init_weights": true, "layers_to_transform": null, "layers_pattern": null, "modules_to_save": null, "coft": false, "eps": 6e-05, "block_share": false, "use_cayley_neumann": true, "num_cayley_neumann_terms": 5 }, "error_msg": "" }, "train_info": { "accelerator_memory_reserved_avg": 14570148069, "accelerator_memory_max": 22315794432, "accelerator_memory_reserved_99th": 20145242112, "train_time": 1118.0740923878038, "file_size": 32693568, "num_trainable_params": 8171520, "num_total_params": 3220921344, "status": "success", "metrics": [ { "step": 250, "valid accuracy": 0.36, "train loss": 0.9629777348041535, "train samples": 1000, "train time": 34.88295750680845, "eval time": 14.028832973039243, "tokens / sec": 6069.410827871367, "mem allocated avg": 6910183786.496, "mem reserved avg": 14692730798.08, "elapsed time": 71.68712687701918 }, { "step": 500, "valid accuracy": 0.32, "train loss": 0.71439142203331, "train samples": 2000, "train time": 34.68627706926782, "eval time": 8.79689448804129, "tokens / sec": 5996.463661540789, "mem allocated avg": 6902787000.32, "mem reserved avg": 14280673984.512, "elapsed time": 118.46906502003549 }, { "step": 750, "valid accuracy": 0.48, "train loss": 0.6710897196531296, "train samples": 3000, "train time": 35.380497081787325, "eval time": 12.356959308963269, "tokens / sec": 6059.863983945164, "mem allocated avg": 6913287290.88, "mem reserved avg": 14527609438.208, "elapsed time": 169.46450086700497 }, { "step": 1000, "valid accuracy": 0.46, "train loss": 0.6509144109487534, "train samples": 4000, "train time": 35.38656886102399, "eval time": 14.169413066993002, "tokens / sec": 5887.431494650181, "mem allocated avg": 6904497004.544, "mem reserved avg": 14534924304.384, "elapsed time": 222.43686137301847 }, { "step": 1250, "valid accuracy": 0.36, "train loss": 0.645386314868927, "train samples": 5000, "train time": 34.42120103741763, "eval time": 14.01571749499999, "tokens / sec": 6058.417304303483, "mem allocated avg": 6905021065.216, "mem reserved avg": 14595632660.48, "elapsed time": 274.1503414300387 }, { "step": 1500, "valid accuracy": 0.46, "train loss": 0.6368236445188522, "train samples": 6000, "train time": 34.666979860921856, "eval time": 9.192593308980577, "tokens / sec": 6038.339677693329, "mem allocated avg": 6905748031.488, "mem reserved avg": 14529454931.968, "elapsed time": 321.3953521400108 }, { "step": 1750, "valid accuracy": 0.5, "train loss": 0.6279961000680924, "train samples": 7000, "train time": 34.723764617112465, "eval time": 11.785299010982271, "tokens / sec": 6029.15617901713, "mem allocated avg": 6907708614.656, "mem reserved avg": 14900634058.752, "elapsed time": 371.2769475780078 }, { "step": 2000, "valid accuracy": 0.4, "train loss": 0.6302493048906326, "train samples": 8000, "train time": 34.55560137151042, "eval time": 14.030101582000498, "tokens / sec": 6010.487207762393, "mem allocated avg": 6903469649.92, "mem reserved avg": 14548564180.992, "elapsed time": 423.27879621402826 }, { "step": 2250, "valid accuracy": 0.44, "train loss": 0.6211440742015839, "train samples": 9000, "train time": 35.52928709477419, "eval time": 14.005440895038191, "tokens / sec": 6049.882155716418, "mem allocated avg": 6915304634.368, "mem reserved avg": 14843717353.472, "elapsed time": 476.20078854600433 }, { "step": 2500, "valid accuracy": 0.42, "train loss": 0.6180904031991958, "train samples": 10000, "train time": 35.02202940842835, "eval time": 14.004316847014707, "tokens / sec": 5881.069814601671, "mem allocated avg": 6900778860.544, "mem reserved avg": 14355215155.2, "elapsed time": 528.5787644560332 }, { "step": 2750, "valid accuracy": 0.46, "train loss": 0.6097299265861511, "train samples": 11000, "train time": 35.6146717875381, "eval time": 9.144496340013575, "tokens / sec": 5949.261620716075, "mem allocated avg": 6911029626.88, "mem reserved avg": 14734220853.248, "elapsed time": 576.7091627839836 }, { "step": 3000, "valid accuracy": 0.42, "train loss": 0.600579794883728, "train samples": 12000, "train time": 35.16766768670641, "eval time": 14.045691164967138, "tokens / sec": 5935.309724247127, "mem allocated avg": 6906340362.24, "mem reserved avg": 14634706796.544, "elapsed time": 629.347200824006 }, { "step": 3250, "valid accuracy": 0.56, "train loss": 0.6099310213327408, "train samples": 13000, "train time": 35.43659597542137, "eval time": 11.369511878991034, "tokens / sec": 5951.502795197366, "mem allocated avg": 6908234280.96, "mem reserved avg": 14499599876.096, "elapsed time": 679.4317268640152 }, { "step": 3500, "valid accuracy": 0.5, "train loss": 0.5943480662107468, "train samples": 14000, "train time": 34.85756022395799, "eval time": 8.693275505036581, "tokens / sec": 6017.345983263524, "mem allocated avg": 6906331211.776, "mem reserved avg": 14555694497.792, "elapsed time": 726.3931989329867 }, { "step": 3750, "valid accuracy": 0.52, "train loss": 0.5926959769725799, "train samples": 15000, "train time": 35.249166297202464, "eval time": 11.4140949450084, "tokens / sec": 6147.748238153325, "mem allocated avg": 6917726124.032, "mem reserved avg": 14819759489.024, "elapsed time": 776.4263785420335 }, { "step": 4000, "valid accuracy": 0.52, "train loss": 0.603757186293602, "train samples": 16000, "train time": 34.53601357748266, "eval time": 8.989645537047181, "tokens / sec": 5917.677775446856, "mem allocated avg": 6898480144.384, "mem reserved avg": 14544931913.728, "elapsed time": 823.3236909009865 }, { "step": 4250, "valid accuracy": 0.52, "train loss": 0.5904598969221115, "train samples": 17000, "train time": 34.97911310993368, "eval time": 9.157656271010637, "tokens / sec": 6043.292159399201, "mem allocated avg": 6909286631.424, "mem reserved avg": 14509322272.768, "elapsed time": 870.700260128011 }, { "step": 4500, "valid accuracy": 0.54, "train loss": 0.5976542398929596, "train samples": 18000, "train time": 35.34081694687484, "eval time": 9.293714377039578, "tokens / sec": 5880.396039299176, "mem allocated avg": 6904009969.664, "mem reserved avg": 14498240921.6, "elapsed time": 918.7880094650318 }, { "step": 4750, "valid accuracy": 0.52, "train loss": 0.5886972019672394, "train samples": 19000, "train time": 35.23478025314398, "eval time": 9.220501197036356, "tokens / sec": 5958.288897836031, "mem allocated avg": 6907173064.704, "mem reserved avg": 14554260045.824, "elapsed time": 966.5027436820092 }, { "step": 5000, "valid accuracy": 0.54, "train loss": 0.5947221873998642, "train samples": 20000, "train time": 35.268570395885035, "eval time": 9.204281533020549, "tokens / sec": 5905.541326514927, "mem allocated avg": 6903379494.912, "mem reserved avg": 14243067854.848, "elapsed time": 1014.2827233370044 }, { "step": 5000, "test accuracy": 0.5034116755117514, "train loss": 0.5947221873998642, "train samples": 20000, "train total tokens": 4198051, "forgetting": 0.5187549591064453 } ] }, "meta_info": { "model_info": { "sha": "13afe5124825b4f3751f836b40dafda64c1ed062", "created_at": "2024-09-18T15:23:48+00:00" }, "dataset_info": { "metamath": { "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18", "created_at": "2023-09-21T17:22:46+00:00" }, "gsm8k": { "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c", "created_at": "2022-04-12T10:22:10+00:00" } }, "package_info": { "transformers-version": "4.57.1", "transformers-commit-hash": null, "peft-version": "0.18.1.dev0", "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26", "datasets-version": "4.2.0", "datasets-commit-hash": null, "bitsandbytes-version": "0.46.0", "bitsandbytes-commit-hash": null, "torch-version": "2.9.0+cu128", "torch-commit-hash": null }, "system_info": { "system": "Linux", "release": "6.14.0-1016-aws", "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025", "machine": "x86_64", "processor": "x86_64", "accelerator": "NVIDIA L40S" }, "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n" } }