Add superseded/exploratory BEFT results for MetaMathQA

MetaMathQA/beft--llama-3.2-3B-target-v_proj-lr-1e-3-wd-1e-4.json

@@ -0,0 +1,349 @@
+{
+  "run_info": {
+    "created_at": "2026-10-13T17:14:28+00:00",
+    "total_time": 659.3155662360368,
+    "experiment_name": "beft/llama-3.2-3B-target-v_proj",
+    "peft_branch": "main",
+    "peft_worktree_is_dirty": true,
+    "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,
+      "use_gc": false,
+      "query_template": "Question: {query} Think step by step.\nAnswer:",
+      "seed": 0,
+      "grad_norm_clip": 1.0,
+      "optimizer_type": "AdamW",
+      "optimizer_kwargs": {
+        "lr": 0.001,
+        "weight_decay": 0.0001
+      },
+      "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": "BEFT",
+      "auto_mapping": null,
+      "peft_version": "0.19.2.dev0@UNKNOWN",
+      "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
+      "revision": null,
+      "inference_mode": false,
+      "target_modules": [
+        "v_proj"
+      ],
+      "modules_to_save": null,
+      "init_weights": true
+    },
+    "error_msg": ""
+  },
+  "train_info": {
+    "accelerator_memory_reserved_avg": 13328198716,
+    "accelerator_memory_max": 20182990848,
+    "accelerator_memory_reserved_99th": 18387828736,
+    "train_time": 557.7726578359725,
+    "file_size": 118192,
+    "num_trainable_params": 28672,
+    "num_total_params": 3212778496,
+    "status": "success",
+    "metrics": [
+      {
+        "step": 250,
+        "valid accuracy": 0.36,
+        "train loss": 1.0245312263965607,
+        "train samples": 1000,
+        "train time": 20.326527078985237,
+        "eval time": 5.963012895022985,
+        "tokens / sec": 10415.896388856687,
+        "mem allocated avg": 6830068822.016,
+        "mem reserved avg": 13434909360.128,
+        "elapsed time": 42.52748516597785
+      },
+      {
+        "step": 500,
+        "valid accuracy": 0.34,
+        "train loss": 0.7735863168239594,
+        "train samples": 2000,
+        "train time": 19.99406438571168,
+        "eval time": 5.8460714189568534,
+        "tokens / sec": 10402.8373615041,
+        "mem allocated avg": 6822936879.104,
+        "mem reserved avg": 13089852358.656,
+        "elapsed time": 68.68402589001926
+      },
+      {
+        "step": 750,
+        "valid accuracy": 0.28,
+        "train loss": 0.7335375688076019,
+        "train samples": 3000,
+        "train time": 20.508623699250165,
+        "eval time": 5.842787707981188,
+        "tokens / sec": 10454.18762097814,
+        "mem allocated avg": 6832603828.224,
+        "mem reserved avg": 13295113207.808,
+        "elapsed time": 95.34244975302136
+      },
+      {
+        "step": 1000,
+        "valid accuracy": 0.34,
+        "train loss": 0.7169527032375336,
+        "train samples": 4000,
+        "train time": 20.072288797353394,
+        "eval time": 5.058653557964135,
+        "tokens / sec": 10379.28469958393,
+        "mem allocated avg": 6824452143.104,
+        "mem reserved avg": 13320673296.384,
+        "elapsed time": 120.80394215800334
+      },
+      {
+        "step": 1250,
+        "valid accuracy": 0.34,
+        "train loss": 0.716132005572319,
+        "train samples": 5000,
+        "train time": 20.062871788803022,
+        "eval time": 5.344415274041239,
+        "tokens / sec": 10394.224824602821,
+        "mem allocated avg": 6824035008.512,
+        "mem reserved avg": 13342953439.232,
+        "elapsed time": 146.51645337202353
+      },
+      {
+        "step": 1500,
+        "valid accuracy": 0.36,
+        "train loss": 0.7126122795343399,
+        "train samples": 6000,
+        "train time": 20.19250567513518,
+        "eval time": 5.84832541400101,
+        "tokens / sec": 10366.76692669042,
+        "mem allocated avg": 6826312718.336,
+        "mem reserved avg": 13286531661.824,
+        "elapsed time": 172.87925322598312
+      },
+      {
+        "step": 1750,
+        "valid accuracy": 0.38,
+        "train loss": 0.7054878860712052,
+        "train samples": 7000,
+        "train time": 20.27162997412961,
+        "eval time": 5.838491978007369,
+        "tokens / sec": 10327.487245336271,
+        "mem allocated avg": 6827604639.744,
+        "mem reserved avg": 13636487610.368,
+        "elapsed time": 199.31679203402018
+      },
+      {
+        "step": 2000,
+        "valid accuracy": 0.32,
+        "train loss": 0.7112382639646531,
+        "train samples": 8000,
+        "train time": 20.055665995110758,
+        "eval time": 5.8460150539758615,
+        "tokens / sec": 10355.976213935397,
+        "mem allocated avg": 6823309834.24,
+        "mem reserved avg": 13322443292.672,
+        "elapsed time": 225.54922061500838
+      },
+      {
+        "step": 2250,
+        "valid accuracy": 0.28,
+        "train loss": 0.7040931502580643,
+        "train samples": 9000,
+        "train time": 20.668595944764093,
+        "eval time": 4.443191917031072,
+        "tokens / sec": 10399.738839272828,
+        "mem allocated avg": 6834229719.04,
+        "mem reserved avg": 13572960681.984,
+        "elapsed time": 250.98201276699547
+      },
+      {
+        "step": 2500,
+        "valid accuracy": 0.4,
+        "train loss": 0.704752881526947,
+        "train samples": 10000,
+        "train time": 19.796692247036844,
+        "eval time": 2.954163371003233,
+        "tokens / sec": 10404.11182988557,
+        "mem allocated avg": 6820292937.728,
+        "mem reserved avg": 13120546275.328,
+        "elapsed time": 274.05205344199203
+      },
+      {
+        "step": 2750,
+        "valid accuracy": 0.46,
+        "train loss": 0.6977959246635437,
+        "train samples": 11000,
+        "train time": 20.478589773527347,
+        "eval time": 3.356092380010523,
+        "tokens / sec": 10346.464397362868,
+        "mem allocated avg": 6830585915.392,
+        "mem reserved avg": 13469445259.264,
+        "elapsed time": 298.2019372640061
+      },
+      {
+        "step": 3000,
+        "valid accuracy": 0.4,
+        "train loss": 0.6920568525791169,
+        "train samples": 12000,
+        "train time": 20.20500026352238,
+        "eval time": 5.850723210023716,
+        "tokens / sec": 10330.660592805729,
+        "mem allocated avg": 6825755551.744,
+        "mem reserved avg": 13388813959.168,
+        "elapsed time": 324.59037534799427
+      },
+      {
+        "step": 3250,
+        "valid accuracy": 0.44,
+        "train loss": 0.7001098393201828,
+        "train samples": 13000,
+        "train time": 20.280712964886334,
+        "eval time": 4.446415318991058,
+        "tokens / sec": 10399.092002591342,
+        "mem allocated avg": 6828777338.88,
+        "mem reserved avg": 13261692993.536,
+        "elapsed time": 349.62496270798147
+      },
+      {
+        "step": 3500,
+        "valid accuracy": 0.36,
+        "train loss": 0.6871621907949448,
+        "train samples": 14000,
+        "train time": 20.276403566880617,
+        "eval time": 4.379504684009589,
+        "tokens / sec": 10344.536658493258,
+        "mem allocated avg": 6825605904.384,
+        "mem reserved avg": 13284451287.04,
+        "elapsed time": 374.60806634800974
+      },
+      {
+        "step": 3750,
+        "valid accuracy": 0.44,
+        "train loss": 0.6828345060348511,
+        "train samples": 15000,
+        "train time": 20.846530717564747,
+        "eval time": 5.842629740945995,
+        "tokens / sec": 10395.158932484226,
+        "mem allocated avg": 6837498036.224,
+        "mem reserved avg": 13575712145.408,
+        "elapsed time": 401.61404635902727
+      },
+      {
+        "step": 4000,
+        "valid accuracy": 0.32,
+        "train loss": 0.7008326334953308,
+        "train samples": 16000,
+        "train time": 19.806991704390384,
+        "eval time": 5.851346745039336,
+        "tokens / sec": 10318.22515252021,
+        "mem allocated avg": 6817916000.256,
+        "mem reserved avg": 13297638178.816,
+        "elapsed time": 427.602031411021
+      },
+      {
+        "step": 4250,
+        "valid accuracy": 0.28,
+        "train loss": 0.6811694331169128,
+        "train samples": 17000,
+        "train time": 20.30570751806954,
+        "eval time": 3.334930925979279,
+        "tokens / sec": 10410.324280101555,
+        "mem allocated avg": 6829602656.256,
+        "mem reserved avg": 13281045512.192,
+        "elapsed time": 451.54076346597867
+      },
+      {
+        "step": 4500,
+        "valid accuracy": 0.44,
+        "train loss": 0.6895727490186692,
+        "train samples": 18000,
+        "train time": 20.077282178215683,
+        "eval time": 5.431811109010596,
+        "tokens / sec": 10350.902983546615,
+        "mem allocated avg": 6824009670.656,
+        "mem reserved avg": 13248850034.688,
+        "elapsed time": 477.3743845350109
+      },
+      {
+        "step": 4750,
+        "valid accuracy": 0.34,
+        "train loss": 0.6827702203989029,
+        "train samples": 19000,
+        "train time": 20.181565830600448,
+        "eval time": 4.181478453043383,
+        "tokens / sec": 10402.512954751928,
+        "mem allocated avg": 6826501228.544,
+        "mem reserved avg": 13321327607.808,
+        "elapsed time": 502.0437595670228
+      },
+      {
+        "step": 5000,
+        "valid accuracy": 0.4,
+        "train loss": 0.6894405043125152,
+        "train samples": 20000,
+        "train time": 19.969160319655202,
+        "eval time": 3.5645567569881678,
+        "tokens / sec": 10430.083021317356,
+        "mem allocated avg": 6823599671.296,
+        "mem reserved avg": 13012526170.112,
+        "elapsed time": 525.8902350760181
+      },
+      {
+        "step": 5000,
+        "test accuracy": 0.3692191053828658,
+        "train loss": 0.6894405043125152,
+        "train samples": 20000,
+        "train total tokens": 4198051,
+        "forgetting": 0.32077717781066895
+      }
+    ]
+  },
+  "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": "5.19.0",
+      "transformers-commit-hash": null,
+      "peft-version": "0.21.3.dev0",
+      "peft-commit-hash": "eac46efeef6e5f63ecc38de5c9148513df602b29",
+      "datasets-version": "5.1.0",
+      "datasets-commit-hash": null,
+      "bitsandbytes-version": "0.50.2",
+      "bitsandbytes-commit-hash": null,
+      "torch-version": "2.14.1+cu130",
+      "torch-commit-hash": null
+    },
+    "system_info": {
+      "system": "Linux",
+      "release": "7.0.0-34-generic",
+      "version": "#34-Ubuntu SMP PREEMPT_DYNAMIC Wed Sep  2 14:29:37 UTC 2026",
+      "machine": "x86_64",
+      "processor": "desconocido",
+      "accelerator": "NVIDIA GeForce RTX 5090"
+    },
+    "pytorch_info": "PyTorch built with:\n  - GCC 13.3\n  - C++ Version: 202002\n  - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n  - Intel(R) MKL-DNN v3.12.0 (Git Hash 80afa71049cd69a3df32adcccb623b12cd7baa22)\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 13.0\n  - NVCC architecture flags: -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 92.4  (built against CUDA 13.3)\n  - Magma 2.6.1\n  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=5c4886908584029761b579af026dcfb627c84070, CUDA_FLAGS= -DLIBCUDACXX_ENABLE_SIMPLIFIED_COMPLEX_OPERATIONS -Xfatbin -compress-all -DONNX_NAMESPACE=onnx_torch -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 -Xcudafe --diag_suppress=cc_clobber_ignored,--diag_suppress=field_without_dll_interface,--diag_suppress=base_class_has_different_dll_interface,--diag_suppress=dll_interface_conflict_none_assumed,--diag_suppress=dll_interface_conflict_dllexport_assumed,--diag_suppress=bad_friend_decl --expt-relaxed-constexpr --expt-extended-lambda -Xfatbin -compress-all --threads 2 -compress-mode=size -Wno-deprecated-gpu-targets --expt-extended-lambda -DCUB_WRAPPED_NAMESPACE=at_cuda_detail -DDISABLE_CUSPARSE_DEPRECATED -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -D__CUDA_NO_BFLOAT16_CONVERSIONS__ -DC10_NODEPRECATED, CUDA_VERSION=13.0, CUDNN_VERSION=9.24.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DHAS_CUPTI -DUSE_FBGEMM -DUSE_MSLK -DUSE_PYTORCH_QNNPACK -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=2.14.1, USE_CUDA=1, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_HIPSPARSELT=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
+  }
+}

MetaMathQA/beft--llama-3.2-3B-target-v_proj-lr-3e-3-wd-1e-4.json

@@ -0,0 +1,349 @@
+{
+  "run_info": {
+    "created_at": "2026-10-13T17:27:08+00:00",
+    "total_time": 666.437604762963,
+    "experiment_name": "beft/llama-3.2-3B-target-v_proj",
+    "peft_branch": "main",
+    "peft_worktree_is_dirty": true,
+    "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,
+      "use_gc": false,
+      "query_template": "Question: {query} Think step by step.\nAnswer:",
+      "seed": 0,
+      "grad_norm_clip": 1.0,
+      "optimizer_type": "AdamW",
+      "optimizer_kwargs": {
+        "lr": 0.003,
+        "weight_decay": 0.0001
+      },
+      "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": "BEFT",
+      "auto_mapping": null,
+      "peft_version": "0.19.2.dev0@UNKNOWN",
+      "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
+      "revision": null,
+      "inference_mode": false,
+      "target_modules": [
+        "v_proj"
+      ],
+      "modules_to_save": null,
+      "init_weights": true
+    },
+    "error_msg": ""
+  },
+  "train_info": {
+    "accelerator_memory_reserved_avg": 13328198716,
+    "accelerator_memory_max": 20182990848,
+    "accelerator_memory_reserved_99th": 18387828736,
+    "train_time": 564.3411493170424,
+    "file_size": 118192,
+    "num_trainable_params": 28672,
+    "num_total_params": 3212778496,
+    "status": "success",
+    "metrics": [
+      {
+        "step": 250,
+        "valid accuracy": 0.3,
+        "train loss": 0.9300625491142273,
+        "train samples": 1000,
+        "train time": 20.3385904164752,
+        "eval time": 5.96221298299497,
+        "tokens / sec": 10409.718454652482,
+        "mem allocated avg": 6830068822.016,
+        "mem reserved avg": 13434909360.128,
+        "elapsed time": 42.145745913963765
+      },
+      {
+        "step": 500,
+        "valid accuracy": 0.24,
+        "train loss": 0.7540297241210937,
+        "train samples": 2000,
+        "train time": 20.004022866953164,
+        "eval time": 5.846885908977129,
+        "tokens / sec": 10397.658580145382,
+        "mem allocated avg": 6822936879.104,
+        "mem reserved avg": 13089852358.656,
+        "elapsed time": 68.31273779697949
+      },
+      {
+        "step": 750,
+        "valid accuracy": 0.38,
+        "train loss": 0.7402237559556961,
+        "train samples": 3000,
+        "train time": 20.517465984623414,
+        "eval time": 4.533389645977877,
+        "tokens / sec": 10449.682244419484,
+        "mem allocated avg": 6832603828.224,
+        "mem reserved avg": 13295113207.808,
+        "elapsed time": 93.67016269895248
+      },
+      {
+        "step": 1000,
+        "valid accuracy": 0.34,
+        "train loss": 0.7276687721014022,
+        "train samples": 4000,
+        "train time": 20.08168734848732,
+        "eval time": 5.8457320829620585,
+        "tokens / sec": 10374.42702819956,
+        "mem allocated avg": 6824452143.104,
+        "mem reserved avg": 13320673296.384,
+        "elapsed time": 119.92784123896854
+      },
+      {
+        "step": 1250,
+        "valid accuracy": 0.32,
+        "train loss": 0.7296145635843277,
+        "train samples": 5000,
+        "train time": 20.074126126943156,
+        "eval time": 5.843177411996294,
+        "tokens / sec": 10388.397416717622,
+        "mem allocated avg": 6824035008.512,
+        "mem reserved avg": 13342953439.232,
+        "elapsed time": 146.15179624495795
+      },
+      {
+        "step": 1500,
+        "valid accuracy": 0.32,
+        "train loss": 0.7264302505254745,
+        "train samples": 6000,
+        "train time": 20.201129232998937,
+        "eval time": 3.505315147980582,
+        "tokens / sec": 10362.341510001022,
+        "mem allocated avg": 6826312718.336,
+        "mem reserved avg": 13286531661.824,
+        "elapsed time": 170.17988358595176
+      },
+      {
+        "step": 1750,
+        "valid accuracy": 0.3,
+        "train loss": 0.7198800925016403,
+        "train samples": 7000,
+        "train time": 20.285379203327466,
+        "eval time": 4.755016795010306,
+        "tokens / sec": 10320.487376723968,
+        "mem allocated avg": 6827604639.744,
+        "mem reserved avg": 13636487610.368,
+        "elapsed time": 195.54621281696018
+      },
+      {
+        "step": 2000,
+        "valid accuracy": 0.34,
+        "train loss": 0.726202164530754,
+        "train samples": 8000,
+        "train time": 20.069145588960964,
+        "eval time": 4.423285711964127,
+        "tokens / sec": 10349.020543965918,
+        "mem allocated avg": 6823309834.24,
+        "mem reserved avg": 13322443292.672,
+        "elapsed time": 220.36964575597085
+      },
+      {
+        "step": 2250,
+        "valid accuracy": 0.2,
+        "train loss": 0.7168622585535049,
+        "train samples": 9000,
+        "train time": 20.668943284312263,
+        "eval time": 3.7040826269658282,
+        "tokens / sec": 10399.56407268995,
+        "mem allocated avg": 6834229719.04,
+        "mem reserved avg": 13572960681.984,
+        "elapsed time": 245.0626095989719
+      },
+      {
+        "step": 2500,
+        "valid accuracy": 0.3,
+        "train loss": 0.7168004482984542,
+        "train samples": 10000,
+        "train time": 19.796269813959952,
+        "eval time": 5.156895759981126,
+        "tokens / sec": 10404.333843477723,
+        "mem allocated avg": 6820292937.728,
+        "mem reserved avg": 13120546275.328,
+        "elapsed time": 270.3352643069811
+      },
+      {
+        "step": 2750,
+        "valid accuracy": 0.34,
+        "train loss": 0.707111300945282,
+        "train samples": 11000,
+        "train time": 20.475248530099634,
+        "eval time": 5.492366418009624,
+        "tokens / sec": 10348.15278010054,
+        "mem allocated avg": 6830585915.392,
+        "mem reserved avg": 13469445259.264,
+        "elapsed time": 296.62060505896807
+      },
+      {
+        "step": 3000,
+        "valid accuracy": 0.28,
+        "train loss": 0.6994408316612244,
+        "train samples": 12000,
+        "train time": 20.203367022098973,
+        "eval time": 5.851800836040638,
+        "tokens / sec": 10331.4957240387,
+        "mem allocated avg": 6825755551.744,
+        "mem reserved avg": 13388813959.168,
+        "elapsed time": 323.0093792899861
+      },
+      {
+        "step": 3250,
+        "valid accuracy": 0.34,
+        "train loss": 0.7055730600357055,
+        "train samples": 13000,
+        "train time": 20.280480366549455,
+        "eval time": 5.846150314027909,
+        "tokens / sec": 10399.211270550539,
+        "mem allocated avg": 6828777338.88,
+        "mem reserved avg": 13261692993.536,
+        "elapsed time": 349.4439625289524
+      },
+      {
+        "step": 3500,
+        "valid accuracy": 0.3,
+        "train loss": 0.6899930350780487,
+        "train samples": 14000,
+        "train time": 20.281909691926558,
+        "eval time": 4.625385458988603,
+        "tokens / sec": 10341.728327658087,
+        "mem allocated avg": 6825605904.384,
+        "mem reserved avg": 13284451287.04,
+        "elapsed time": 374.6788030509488
+      },
+      {
+        "step": 3750,
+        "valid accuracy": 0.36,
+        "train loss": 0.6853874101638794,
+        "train samples": 15000,
+        "train time": 20.842822803067975,
+        "eval time": 5.842294892005157,
+        "tokens / sec": 10397.008219448195,
+        "mem allocated avg": 6837498036.224,
+        "mem reserved avg": 13575712145.408,
+        "elapsed time": 401.680529352976
+      },
+      {
+        "step": 4000,
+        "valid accuracy": 0.38,
+        "train loss": 0.7004596869945526,
+        "train samples": 16000,
+        "train time": 19.79744659521384,
+        "eval time": 3.7759735580184497,
+        "tokens / sec": 10323.199965060569,
+        "mem allocated avg": 6817916000.256,
+        "mem reserved avg": 13297638178.816,
+        "elapsed time": 425.5818783219438
+      },
+      {
+        "step": 4250,
+        "valid accuracy": 0.36,
+        "train loss": 0.6792559452056884,
+        "train samples": 17000,
+        "train time": 20.296205644495785,
+        "eval time": 5.838855997018982,
+        "tokens / sec": 10415.197978511194,
+        "mem allocated avg": 6829602656.256,
+        "mem reserved avg": 13281045512.192,
+        "elapsed time": 452.0135783869773
+      },
+      {
+        "step": 4500,
+        "valid accuracy": 0.36,
+        "train loss": 0.6860878685712815,
+        "train samples": 18000,
+        "train time": 20.065862707037013,
+        "eval time": 3.6924940180033445,
+        "tokens / sec": 10356.793676611727,
+        "mem allocated avg": 6824009670.656,
+        "mem reserved avg": 13248850034.688,
+        "elapsed time": 476.09644777997164
+      },
+      {
+        "step": 4750,
+        "valid accuracy": 0.32,
+        "train loss": 0.6793690195083618,
+        "train samples": 19000,
+        "train time": 20.174648426065687,
+        "eval time": 3.414061944989953,
+        "tokens / sec": 10406.07972770214,
+        "mem allocated avg": 6826501228.544,
+        "mem reserved avg": 13321327607.808,
+        "elapsed time": 499.99087693099864
+      },
+      {
+        "step": 5000,
+        "valid accuracy": 0.38,
+        "train loss": 0.684873697757721,
+        "train samples": 20000,
+        "train time": 19.966303129680455,
+        "eval time": 5.851652743993327,
+        "tokens / sec": 10431.575572464693,
+        "mem allocated avg": 6823599671.296,
+        "mem reserved avg": 13012526170.112,
+        "elapsed time": 526.1213023129967
+      },
+      {
+        "step": 5000,
+        "test accuracy": 0.35329795299469297,
+        "train loss": 0.684873697757721,
+        "train samples": 20000,
+        "train total tokens": 4198051,
+        "forgetting": 0.43311285972595215
+      }
+    ]
+  },
+  "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": "5.19.0",
+      "transformers-commit-hash": null,
+      "peft-version": "0.21.3.dev0",
+      "peft-commit-hash": "eac46efeef6e5f63ecc38de5c9148513df602b29",
+      "datasets-version": "5.1.0",
+      "datasets-commit-hash": null,
+      "bitsandbytes-version": "0.50.2",
+      "bitsandbytes-commit-hash": null,
+      "torch-version": "2.14.1+cu130",
+      "torch-commit-hash": null
+    },
+    "system_info": {
+      "system": "Linux",
+      "release": "7.0.0-34-generic",
+      "version": "#34-Ubuntu SMP PREEMPT_DYNAMIC Wed Sep  2 14:29:37 UTC 2026",
+      "machine": "x86_64",
+      "processor": "desconocido",
+      "accelerator": "NVIDIA GeForce RTX 5090"
+    },
+    "pytorch_info": "PyTorch built with:\n  - GCC 13.3\n  - C++ Version: 202002\n  - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n  - Intel(R) MKL-DNN v3.12.0 (Git Hash 80afa71049cd69a3df32adcccb623b12cd7baa22)\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 13.0\n  - NVCC architecture flags: -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 92.4  (built against CUDA 13.3)\n  - Magma 2.6.1\n  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=5c4886908584029761b579af026dcfb627c84070, CUDA_FLAGS= -DLIBCUDACXX_ENABLE_SIMPLIFIED_COMPLEX_OPERATIONS -Xfatbin -compress-all -DONNX_NAMESPACE=onnx_torch -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 -Xcudafe --diag_suppress=cc_clobber_ignored,--diag_suppress=field_without_dll_interface,--diag_suppress=base_class_has_different_dll_interface,--diag_suppress=dll_interface_conflict_none_assumed,--diag_suppress=dll_interface_conflict_dllexport_assumed,--diag_suppress=bad_friend_decl --expt-relaxed-constexpr --expt-extended-lambda -Xfatbin -compress-all --threads 2 -compress-mode=size -Wno-deprecated-gpu-targets --expt-extended-lambda -DCUB_WRAPPED_NAMESPACE=at_cuda_detail -DDISABLE_CUSPARSE_DEPRECATED -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -D__CUDA_NO_BFLOAT16_CONVERSIONS__ -DC10_NODEPRECATED, CUDA_VERSION=13.0, CUDNN_VERSION=9.24.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DHAS_CUPTI -DUSE_FBGEMM -DUSE_MSLK -DUSE_PYTORCH_QNNPACK -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=2.14.1, USE_CUDA=1, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_HIPSPARSELT=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
+  }
+}