{ "run_info": { "created_at": "2026-04-14T18:18:17+00:00", "total_time": 1468.6276556560042, "experiment_name": "psoft/llama-3.2-3B-fast", "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, "init_kv_cache_prefix": null }, "peft_config": { "task_type": null, "peft_type": "PSOFT", "auto_mapping": null, "peft_version": "0.19.0", "base_model_name_or_path": "meta-llama/Llama-3.2-3B", "revision": null, "inference_mode": false, "r": 256, "target_modules": [ "v_proj", "q_proj" ], "exclude_modules": null, "psoft_alpha": 256, "psoft_dropout": 0.0, "fan_in_fan_out": false, "ab_svd_init": "psoft_init", "psoft_svd": "lowrank", "psoft_svd_lowrank_niter": 10, "psoft_orth": true, "psoft_mag_b": true, "psoft_mag_a": true, "use_cayley_neumann": true, "num_cayley_neumann_terms": 5, "cayley_neumann_eps": null, "modules_to_save": null, "init_weights": true, "layers_to_transform": null, "layers_pattern": null }, "error_msg": "" }, "train_info": { "accelerator_memory_reserved_avg": 13830374345, "accelerator_memory_max": 20770193408, "accelerator_memory_reserved_99th": 18964545536, "train_time": 1200.1861080279923, "file_size": 7448144, "num_trainable_params": 1856512, "num_total_params": 3214606336, "status": "success", "metrics": [ { "step": 250, "valid accuracy": 0.38, "train loss": 1.0145607323646546, "train samples": 1000, "train time": 36.71533618000103, "eval time": 15.666215841993107, "tokens / sec": 5766.5003790793035, "mem allocated avg": 7133637742.592, "mem reserved avg": 13938871762.944, "elapsed time": 74.23814530199161 }, { "step": 500, "valid accuracy": 0.36, "train loss": 0.7338115975856782, "train samples": 2000, "train time": 37.135233334905934, "eval time": 15.582825934994617, "tokens / sec": 5601.015028616808, "mem allocated avg": 7126403880.96, "mem reserved avg": 13578312613.888, "elapsed time": 130.02693150399136 }, { "step": 750, "valid accuracy": 0.4, "train loss": 0.688268399477005, "train samples": 3000, "train time": 37.55874616494111, "eval time": 11.907032759001595, "tokens / sec": 5708.417396535211, "mem allocated avg": 7136393115.648, "mem reserved avg": 13794109554.688, "elapsed time": 182.50203327499912 }, { "step": 1000, "valid accuracy": 0.44, "train loss": 0.6681705112457276, "train samples": 4000, "train time": 37.44897099403897, "eval time": 15.572288518989808, "tokens / sec": 5563.196917564501, "mem allocated avg": 7127546568.704, "mem reserved avg": 13830113460.224, "elapsed time": 238.6246955839888 }, { "step": 1250, "valid accuracy": 0.5, "train loss": 0.66438540828228, "train samples": 5000, "train time": 37.498516489053145, "eval time": 11.428519556997344, "tokens / sec": 5561.233337347572, "mem allocated avg": 7128205649.92, "mem reserved avg": 13868633948.16, "elapsed time": 290.5487072549877 }, { "step": 1500, "valid accuracy": 0.34, "train loss": 0.6563739287853241, "train samples": 6000, "train time": 37.84935744001996, "eval time": 11.965642277995357, "tokens / sec": 5530.635502378812, "mem allocated avg": 7129830410.24, "mem reserved avg": 13790418567.168, "elapsed time": 343.42470807499194 }, { "step": 1750, "valid accuracy": 0.46, "train loss": 0.6491832717657089, "train samples": 7000, "train time": 37.42066152404004, "eval time": 9.395735846002935, "tokens / sec": 5594.6365316258425, "mem allocated avg": 7131694694.4, "mem reserved avg": 14133361639.424, "elapsed time": 393.339392444992 }, { "step": 2000, "valid accuracy": 0.42, "train loss": 0.6512416273355484, "train samples": 8000, "train time": 37.458074163863785, "eval time": 15.324161733005894, "tokens / sec": 5544.759164377078, "mem allocated avg": 7127220852.736, "mem reserved avg": 13816918179.84, "elapsed time": 449.26280711599975 }, { "step": 2250, "valid accuracy": 0.42, "train loss": 0.6431628746986389, "train samples": 9000, "train time": 38.11420117891976, "eval time": 15.267252121004276, "tokens / sec": 5639.57772566105, "mem allocated avg": 7137530970.112, "mem reserved avg": 14086007947.264, "elapsed time": 505.74102219998895 }, { "step": 2500, "valid accuracy": 0.36, "train loss": 0.6422915599346161, "train samples": 10000, "train time": 37.665808972960804, "eval time": 15.502652708993992, "tokens / sec": 5468.274958540192, "mem allocated avg": 7124247066.624, "mem reserved avg": 13621077737.472, "elapsed time": 561.9800479659898 }, { "step": 2750, "valid accuracy": 0.48, "train loss": 0.6344620395898819, "train samples": 11000, "train time": 37.729417049049516, "eval time": 14.667384405009216, "tokens / sec": 5615.803703633893, "mem allocated avg": 7134092083.2, "mem reserved avg": 13956924047.36, "elapsed time": 617.4243246029946 }, { "step": 3000, "valid accuracy": 0.44, "train loss": 0.6265726834535599, "train samples": 12000, "train time": 37.67032070188725, "eval time": 10.909270905001904, "tokens / sec": 5540.993442870869, "mem allocated avg": 7129951842.304, "mem reserved avg": 13895787872.256, "elapsed time": 669.1392444079975 }, { "step": 3250, "valid accuracy": 0.4, "train loss": 0.6364836758375167, "train samples": 13000, "train time": 37.44520631492196, "eval time": 15.535097809988656, "tokens / sec": 5632.256322111803, "mem allocated avg": 7132112439.296, "mem reserved avg": 13748290977.792, "elapsed time": 725.1316548469913 }, { "step": 3500, "valid accuracy": 0.42, "train loss": 0.6213824005126953, "train samples": 14000, "train time": 37.835670114000095, "eval time": 15.68471787100134, "tokens / sec": 5543.710455451601, "mem allocated avg": 7129427118.08, "mem reserved avg": 13794428321.792, "elapsed time": 781.7749934619933 }, { "step": 3750, "valid accuracy": 0.44, "train loss": 0.6196332731246948, "train samples": 15000, "train time": 38.61936162307393, "eval time": 10.862610769996536, "tokens / sec": 5611.252773026842, "mem allocated avg": 7140212428.8, "mem reserved avg": 14072191909.888, "elapsed time": 834.3267648119945 }, { "step": 4000, "valid accuracy": 0.46, "train loss": 0.6326277133226395, "train samples": 16000, "train time": 37.171097063837806, "eval time": 10.443736899003852, "tokens / sec": 5498.169710972181, "mem allocated avg": 7121660831.744, "mem reserved avg": 13814359654.4, "elapsed time": 885.054980269997 }, { "step": 4250, "valid accuracy": 0.38, "train loss": 0.618290886759758, "train samples": 17000, "train time": 37.33161400198878, "eval time": 14.996235694008647, "tokens / sec": 5662.4661336297595, "mem allocated avg": 7132966852.608, "mem reserved avg": 13792633159.68, "elapsed time": 940.3594205099944 }, { "step": 4500, "valid accuracy": 0.42, "train loss": 0.6252515635490418, "train samples": 18000, "train time": 37.50360515809734, "eval time": 11.315083836001577, "tokens / sec": 5541.280608195885, "mem allocated avg": 7127867189.248, "mem reserved avg": 13762828435.456, "elapsed time": 992.3431118549925 }, { "step": 4750, "valid accuracy": 0.4, "train loss": 0.6172080825567245, "train samples": 19000, "train time": 37.360017932980554, "eval time": 9.651658456001314, "tokens / sec": 5619.349551079063, "mem allocated avg": 7130523303.936, "mem reserved avg": 13808235970.56, "elapsed time": 1042.3337350419897 }, { "step": 5000, "valid accuracy": 0.42, "train loss": 0.6243040920495987, "train samples": 20000, "train time": 37.632196523059974, "eval time": 10.714210777005064, "tokens / sec": 5534.622457457984, "mem allocated avg": 7126828070.912, "mem reserved avg": 13503981158.4, "elapsed time": 1093.72050657199 }, { "step": 5000, "test accuracy": 0.44351781652767247, "train loss": 0.6243040920495987, "train samples": 20000, "train total tokens": 4198051, "forgetting": 0.5018253326416016 } ] }, "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": "740312add88f781978c0658806c59bc2815b9866", "created_at": "2022-04-12T10:22:10+00:00" } }, "package_info": { "transformers-version": "5.5.4", "transformers-commit-hash": null, "peft-version": "0.19.0", "peft-commit-hash": "6d5a6f4f2f902dbf13d21d2661d57c3c05df1dae", "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": "6.17.0-1009-aws", "version": "#9~24.04.2-Ubuntu SMP Fri Mar 6 23:50:29 UTC 2026", "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" } }