{ "run_info": { "created_at": "2025-07-31T14:52:50+00:00", "total_time": 2084.7194732099997, "experiment_name": "shira/llama-3.2-3B-lr_0.0003-random_seed_42", "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": 0.0003 }, "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": "SHIRA", "auto_mapping": null, "base_model_name_or_path": "meta-llama/Llama-3.2-3B", "revision": null, "inference_mode": false, "r": 32, "mask_type": "random", "random_seed": 42, "target_modules": [ "v_proj", "q_proj" ], "fan_in_fan_out": false, "init_weights": true, "modules_to_save": null }, "error_msg": "" }, "train_info": { "accelerator_memory_reserved_avg": 12240924809, "accelerator_memory_max": 21743271936, "accelerator_memory_reserved_99th": 17637383864, "train_time": 1867.0518525470034, "file_size": 110115520, "num_trainable_params": 9175040, "num_total_params": 3221924864, "status": "success", "metrics": [ { "step": 250, "valid accuracy": 0.38, "train loss": 0.9357188057899475, "train samples": 1000, "train time": 31.692333374005102, "eval time": 12.883808001000943, "tokens / sec": 6680.448470028325, "mem allocated avg": 6994551982.08, "mem reserved avg": 12283740684.288, "elapsed time": 96.08096698400004 }, { "step": 500, "valid accuracy": 0.36, "train loss": 0.7029980063438416, "train samples": 2000, "train time": 30.742395647013836, "eval time": 12.860161713999332, "tokens / sec": 6765.73818085656, "mem allocated avg": 6987138160.64, "mem reserved avg": 12187783397.376, "elapsed time": 185.3151257690006 }, { "step": 750, "valid accuracy": 0.46, "train loss": 0.6642508600950241, "train samples": 3000, "train time": 31.300478177990954, "eval time": 10.545964175000336, "tokens / sec": 6849.767558846972, "mem allocated avg": 6997475934.208, "mem reserved avg": 12327495663.616, "elapsed time": 273.0391829120017 }, { "step": 1000, "valid accuracy": 0.36, "train loss": 0.6439496507644653, "train samples": 4000, "train time": 31.05783393001184, "eval time": 12.813238828999602, "tokens / sec": 6708.00160981866, "mem allocated avg": 6989375735.808, "mem reserved avg": 12216002674.688, "elapsed time": 362.4678097830001 }, { "step": 1250, "valid accuracy": 0.32, "train loss": 0.6386180140972137, "train samples": 5000, "train time": 31.162159604989938, "eval time": 9.70755834100055, "tokens / sec": 6692.026568229476, "mem allocated avg": 6988624240.64, "mem reserved avg": 12208931078.144, "elapsed time": 449.31565678900006 }, { "step": 1500, "valid accuracy": 0.5, "train loss": 0.6296385749578476, "train samples": 6000, "train time": 31.37457448001078, "eval time": 12.791106022999884, "tokens / sec": 6671.994870667266, "mem allocated avg": 6990819698.688, "mem reserved avg": 12231488045.056, "elapsed time": 539.3741775020007 }, { "step": 1750, "valid accuracy": 0.52, "train loss": 0.6209055181741715, "train samples": 7000, "train time": 31.12063506498089, "eval time": 11.802694226998938, "tokens / sec": 6727.208476397092, "mem allocated avg": 6991199850.496, "mem reserved avg": 12244775600.128, "elapsed time": 628.1649310410012 }, { "step": 2000, "valid accuracy": 0.42, "train loss": 0.622630435705185, "train samples": 8000, "train time": 30.90304242698403, "eval time": 12.802878865999446, "tokens / sec": 6720.8916562416925, "mem allocated avg": 6988364103.68, "mem reserved avg": 12193303101.44, "elapsed time": 717.7866772650013 }, { "step": 2250, "valid accuracy": 0.5, "train loss": 0.6117782632112503, "train samples": 9000, "train time": 31.835284770990256, "eval time": 12.896296490000168, "tokens / sec": 6751.8792919945945, "mem allocated avg": 6999611363.328, "mem reserved avg": 12359498203.136, "elapsed time": 808.9164720260014 }, { "step": 2500, "valid accuracy": 0.48, "train loss": 0.6087703567743301, "train samples": 10000, "train time": 31.106087331994786, "eval time": 8.24222338599975, "tokens / sec": 6621.437077627842, "mem allocated avg": 6984549638.144, "mem reserved avg": 12142711406.592, "elapsed time": 894.0944889150014 }, { "step": 2750, "valid accuracy": 0.54, "train loss": 0.5988683942556381, "train samples": 11000, "train time": 31.549549333021787, "eval time": 10.239751285998864, "tokens / sec": 6715.817007827485, "mem allocated avg": 6995372679.168, "mem reserved avg": 12298949230.592, "elapsed time": 981.9011422450003 }, { "step": 3000, "valid accuracy": 0.5, "train loss": 0.5887085427045822, "train samples": 12000, "train time": 30.952114122006606, "eval time": 6.788169478999407, "tokens / sec": 6743.6750580986845, "mem allocated avg": 6990271899.648, "mem reserved avg": 12226782035.968, "elapsed time": 1065.3513825200007 }, { "step": 3250, "valid accuracy": 0.5, "train loss": 0.5973232421875, "train samples": 13000, "train time": 31.107180875995255, "eval time": 8.763070525999865, "tokens / sec": 6779.81720171717, "mem allocated avg": 6991821465.6, "mem reserved avg": 12247829053.44, "elapsed time": 1151.3527770540004 }, { "step": 3500, "valid accuracy": 0.58, "train loss": 0.5808243087530136, "train samples": 14000, "train time": 31.351620633020502, "eval time": 12.858672252999895, "tokens / sec": 6690.244260581693, "mem allocated avg": 6991037339.648, "mem reserved avg": 12237720780.8, "elapsed time": 1241.8340592570003 }, { "step": 3750, "valid accuracy": 0.54, "train loss": 0.5781804740428924, "train samples": 15000, "train time": 31.949568995005393, "eval time": 10.286192837000272, "tokens / sec": 6782.658008121384, "mem allocated avg": 7002176088.064, "mem reserved avg": 12393589506.048, "elapsed time": 1330.699673422001 }, { "step": 4000, "valid accuracy": 0.52, "train loss": 0.5873791750669479, "train samples": 16000, "train time": 30.883606043998952, "eval time": 12.840774923999561, "tokens / sec": 6617.523863917831, "mem allocated avg": 6983247890.432, "mem reserved avg": 12128861814.784, "elapsed time": 1420.291728133001 }, { "step": 4250, "valid accuracy": 0.54, "train loss": 0.5750357346534729, "train samples": 17000, "train time": 31.41448626901547, "eval time": 7.603731496999899, "tokens / sec": 6729.029346199935, "mem allocated avg": 6993511036.928, "mem reserved avg": 12266485317.632, "elapsed time": 1505.506280988 }, { "step": 4500, "valid accuracy": 0.54, "train loss": 0.5816998761892319, "train samples": 18000, "train time": 30.81339496200053, "eval time": 6.927671567998914, "tokens / sec": 6744.404511618528, "mem allocated avg": 6988376672.256, "mem reserved avg": 12194603335.68, "elapsed time": 1588.958452021001 }, { "step": 4750, "valid accuracy": 0.58, "train loss": 0.5723758825063705, "train samples": 19000, "train time": 31.16517290099182, "eval time": 7.508142915999997, "tokens / sec": 6736.333556272963, "mem allocated avg": 6991473983.488, "mem reserved avg": 12240195420.16, "elapsed time": 1673.9547475400013 }, { "step": 5000, "valid accuracy": 0.54, "train loss": 0.5789329997301101, "train samples": 20000, "train time": 31.123193277984683, "eval time": 7.151714429001004, "tokens / sec": 6692.115366816459, "mem allocated avg": 6987954006.016, "mem reserved avg": 12187749842.944, "elapsed time": 1758.2295099830008 }, { "step": 5000, "test accuracy": 0.5072024260803639, "train loss": 0.5789329997301101, "train samples": 20000, "train total tokens": 4198051 } ] }, "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": "e53f048856ff4f594e959d75785d2c2d37b678ee", "created_at": "2022-04-12T10:22:10+00:00" } }, "package_info": { "transformers-version": "4.52.4", "transformers-commit-hash": null, "peft-version": "0.16.1.dev0", "peft-commit-hash": "25e5c6b25c4589eb2683484ede1ba3d985d8a760", "datasets-version": "3.6.0", "datasets-commit-hash": null, "bitsandbytes-version": "0.46.0", "bitsandbytes-commit-hash": null, "torch-version": "2.7.1+cu126", "torch-commit-hash": null }, "system_info": { "system": "Linux", "release": "6.8.0-1031-aws", "version": "#33-Ubuntu SMP Fri Jun 20 18:11:07 UTC 2025", "machine": "x86_64", "processor": "x86_64", "accelerator": "NVIDIA L40S" }, "pytorch_info": "PyTorch built with:\n - GCC 11.2\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 12.6\n - NVCC architecture flags: -gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-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\n - CuDNN 90.7.1 (built against CUDA 12.8)\n - Built with CuDNN 90.5.1\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=e2d141dbde55c2a4370fac5165b0561b6af4798b, CUDA_VERSION=12.6, CUDNN_VERSION=9.5.1, CXX_COMPILER=/opt/rh/gcc-toolset-11/root/usr/bin/c++, CXX_FLAGS= -D_GLIBCXX_USE_CXX11_ABI=1 -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 -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 -fdiagnostics-color=always -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.7.1, 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, \n" } }