Add superseded/exploratory LoHa results for MetaMathQA

MetaMathQA/loha--llama-3.2-3B-rank32-1e-3-wd-1e-4.json

@@ -0,0 +1,360 @@
+{
+  "run_info": {
+    "created_at": "2026-10-14T17:13:54+00:00",
+    "total_time": 970.7966464890633,
+    "experiment_name": "loha/llama-3.2-3B-rank32",
+    "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": "LOHA",
+      "auto_mapping": null,
+      "peft_version": "0.21.3.dev0@UNKNOWN",
+      "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
+      "revision": null,
+      "inference_mode": false,
+      "rank_pattern": {},
+      "alpha_pattern": {},
+      "r": 32,
+      "alpha": 64,
+      "rank_dropout": 0.0,
+      "module_dropout": 0.0,
+      "use_effective_conv2d": false,
+      "target_modules": [
+        "q_proj",
+        "v_proj"
+      ],
+      "exclude_modules": null,
+      "init_weights": true,
+      "layers_to_transform": null,
+      "layers_pattern": null,
+      "modules_to_save": null
+    },
+    "error_msg": ""
+  },
+  "train_info": {
+    "accelerator_memory_reserved_avg": 16046795574,
+    "accelerator_memory_max": 23924310016,
+    "accelerator_memory_reserved_99th": 21848129536,
+    "train_time": 855.561030873796,
+    "file_size": 73429560,
+    "num_trainable_params": 18350080,
+    "num_total_params": 3231099904,
+    "status": "success",
+    "metrics": [
+      {
+        "step": 250,
+        "valid accuracy": 0.42,
+        "train loss": 0.9902734475135804,
+        "train samples": 1000,
+        "train time": 32.485731806606054,
+        "eval time": 7.629349319031462,
+        "tokens / sec": 6517.2919994662525,
+        "mem allocated avg": 7124894255.104,
+        "mem reserved avg": 16177229201.408,
+        "elapsed time": 56.41213567892555
+      },
+      {
+        "step": 500,
+        "valid accuracy": 0.46,
+        "train loss": 0.7017386666536332,
+        "train samples": 2000,
+        "train time": 32.04551741608884,
+        "eval time": 7.505304517108016,
+        "tokens / sec": 6490.611379411636,
+        "mem allocated avg": 7118103789.568,
+        "mem reserved avg": 15777235206.144,
+        "elapsed time": 96.42229366395622
+      },
+      {
+        "step": 750,
+        "valid accuracy": 0.4,
+        "train loss": 0.6680633959770202,
+        "train samples": 3000,
+        "train time": 32.81265900353901,
+        "eval time": 5.488881752011366,
+        "tokens / sec": 6534.094051228088,
+        "mem allocated avg": 7127977418.752,
+        "mem reserved avg": 16003702456.32,
+        "elapsed time": 135.16874703299254
+      },
+      {
+        "step": 1000,
+        "valid accuracy": 0.44,
+        "train loss": 0.647118297457695,
+        "train samples": 4000,
+        "train time": 32.19494153058622,
+        "eval time": 5.517084015067667,
+        "tokens / sec": 6471.078688000541,
+        "mem allocated avg": 7119628976.128,
+        "mem reserved avg": 16061768400.896,
+        "elapsed time": 173.3509741409216
+      },
+      {
+        "step": 1250,
+        "valid accuracy": 0.44,
+        "train loss": 0.6420405964851379,
+        "train samples": 5000,
+        "train time": 32.173693171236664,
+        "eval time": 5.378987038973719,
+        "tokens / sec": 6481.630781088984,
+        "mem allocated avg": 7119241891.84,
+        "mem reserved avg": 16092168716.288,
+        "elapsed time": 211.3480433699442
+      },
+      {
+        "step": 1500,
+        "valid accuracy": 0.5,
+        "train loss": 0.632697790503502,
+        "train samples": 6000,
+        "train time": 32.35018158936873,
+        "eval time": 4.927239834913053,
+        "tokens / sec": 6470.782843110613,
+        "mem allocated avg": 7121132320.768,
+        "mem reserved avg": 16066952560.64,
+        "elapsed time": 249.0960811630357
+      },
+      {
+        "step": 1750,
+        "valid accuracy": 0.44,
+        "train loss": 0.6238378701210022,
+        "train samples": 7000,
+        "train time": 32.4416706985794,
+        "eval time": 4.53242261603009,
+        "tokens / sec": 6453.274307144962,
+        "mem allocated avg": 7122440562.688,
+        "mem reserved avg": 16377867927.552,
+        "elapsed time": 286.5335081689991
+      },
+      {
+        "step": 2000,
+        "valid accuracy": 0.42,
+        "train loss": 0.6243440184593201,
+        "train samples": 8000,
+        "train time": 32.145614706445485,
+        "eval time": 5.144568339921534,
+        "tokens / sec": 6461.099030044526,
+        "mem allocated avg": 7118683537.408,
+        "mem reserved avg": 16049957240.832,
+        "elapsed time": 324.29806715995073
+      },
+      {
+        "step": 2250,
+        "valid accuracy": 0.5,
+        "train loss": 0.6115626794099808,
+        "train samples": 9000,
+        "train time": 33.01589849439915,
+        "eval time": 4.707950338022783,
+        "tokens / sec": 6510.439206628406,
+        "mem allocated avg": 7130056898.56,
+        "mem reserved avg": 16311253991.424,
+        "elapsed time": 362.4861203249311
+      },
+      {
+        "step": 2500,
+        "valid accuracy": 0.5,
+        "train loss": 0.6064757812023163,
+        "train samples": 10000,
+        "train time": 31.803604864166118,
+        "eval time": 3.93870457704179,
+        "tokens / sec": 6476.215538448849,
+        "mem allocated avg": 7116140187.648,
+        "mem reserved avg": 15806930878.464,
+        "elapsed time": 398.6874922210118
+      },
+      {
+        "step": 2750,
+        "valid accuracy": 0.44,
+        "train loss": 0.5948366488218307,
+        "train samples": 11000,
+        "train time": 32.70738513453398,
+        "eval time": 7.513557799044065,
+        "tokens / sec": 6478.0782422220045,
+        "mem allocated avg": 7125151420.416,
+        "mem reserved avg": 16212520075.264,
+        "elapsed time": 439.3632753599668
+      },
+      {
+        "step": 3000,
+        "valid accuracy": 0.5,
+        "train loss": 0.5827998027801514,
+        "train samples": 12000,
+        "train time": 32.335960104479454,
+        "eval time": 4.482683449983597,
+        "tokens / sec": 6455.073525745871,
+        "mem allocated avg": 7120734087.168,
+        "mem reserved avg": 16107251433.472,
+        "elapsed time": 476.654414204997
+      },
+      {
+        "step": 3250,
+        "valid accuracy": 0.46,
+        "train loss": 0.5898663109540939,
+        "train samples": 13000,
+        "train time": 32.456866090535186,
+        "eval time": 4.7175954460399225,
+        "tokens / sec": 6497.8855140145915,
+        "mem allocated avg": 7123690192.896,
+        "mem reserved avg": 15982194065.408,
+        "elapsed time": 514.2771410649875
+      },
+      {
+        "step": 3500,
+        "valid accuracy": 0.56,
+        "train loss": 0.5707491111755371,
+        "train samples": 14000,
+        "train time": 32.44301219272893,
+        "eval time": 5.8224255529930815,
+        "tokens / sec": 6465.182664111836,
+        "mem allocated avg": 7120448550.912,
+        "mem reserved avg": 15947347787.776,
+        "elapsed time": 553.0051284109941
+      },
+      {
+        "step": 3750,
+        "valid accuracy": 0.5,
+        "train loss": 0.5663204175233841,
+        "train samples": 15000,
+        "train time": 33.229280695319176,
+        "eval time": 7.150615982944146,
+        "tokens / sec": 6521.447213587315,
+        "mem allocated avg": 7132748165.12,
+        "mem reserved avg": 16321999798.272,
+        "elapsed time": 593.8430619590217
+      },
+      {
+        "step": 4000,
+        "valid accuracy": 0.52,
+        "train loss": 0.5740908039808273,
+        "train samples": 16000,
+        "train time": 31.761840207502246,
+        "eval time": 4.683991283993237,
+        "tokens / sec": 6434.545311758305,
+        "mem allocated avg": 7113135618.048,
+        "mem reserved avg": 16033154859.008,
+        "elapsed time": 630.7567312360043
+      },
+      {
+        "step": 4250,
+        "valid accuracy": 0.56,
+        "train loss": 0.5613958271741867,
+        "train samples": 17000,
+        "train time": 32.49048618308734,
+        "eval time": 4.667790389969014,
+        "tokens / sec": 6506.181495986257,
+        "mem allocated avg": 7124313315.328,
+        "mem reserved avg": 15974761758.72,
+        "elapsed time": 668.352332490962
+      },
+      {
+        "step": 4500,
+        "valid accuracy": 0.54,
+        "train loss": 0.5672679399251938,
+        "train samples": 18000,
+        "train time": 32.15747860132251,
+        "eval time": 7.097362707951106,
+        "tokens / sec": 6462.5091592676445,
+        "mem allocated avg": 7119053879.296,
+        "mem reserved avg": 15951567257.6,
+        "elapsed time": 708.0739437940065
+      },
+      {
+        "step": 4750,
+        "valid accuracy": 0.54,
+        "train loss": 0.5567854583263397,
+        "train samples": 19000,
+        "train time": 32.329061916098,
+        "eval time": 6.977537356084213,
+        "tokens / sec": 6493.816632998638,
+        "mem allocated avg": 7121662765.056,
+        "mem reserved avg": 16008442019.84,
+        "elapsed time": 747.8250464119483
+      },
+      {
+        "step": 5000,
+        "valid accuracy": 0.58,
+        "train loss": 0.5642280827760696,
+        "train samples": 20000,
+        "train time": 32.03754023369402,
+        "eval time": 4.823180010076612,
+        "tokens / sec": 6501.123322225314,
+        "mem allocated avg": 7118618368.0,
+        "mem reserved avg": 15671605854.208,
+        "elapsed time": 785.1359469059389
+      },
+      {
+        "step": 5000,
+        "test accuracy": 0.535253980288097,
+        "train loss": 0.5642280827760696,
+        "train samples": 20000,
+        "train total tokens": 4198051,
+        "forgetting": 0.0697629451751709
+      }
+    ]
+  },
+  "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": "98d4849e3ecc30142782ec0c418a1f86f78fb562",
+      "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/loha--llama-3.2-3B-rank32-5e-4-wd-1e-4.json

@@ -0,0 +1,360 @@
+{
+  "run_info": {
+    "created_at": "2026-10-14T16:56:15+00:00",
+    "total_time": 954.2629669710295,
+    "experiment_name": "loha/llama-3.2-3B-rank32",
+    "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": 0.0005,
+        "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": "LOHA",
+      "auto_mapping": null,
+      "peft_version": "0.21.3.dev0@UNKNOWN",
+      "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
+      "revision": null,
+      "inference_mode": false,
+      "rank_pattern": {},
+      "alpha_pattern": {},
+      "r": 32,
+      "alpha": 64,
+      "rank_dropout": 0.0,
+      "module_dropout": 0.0,
+      "use_effective_conv2d": false,
+      "target_modules": [
+        "q_proj",
+        "v_proj"
+      ],
+      "exclude_modules": null,
+      "init_weights": true,
+      "layers_to_transform": null,
+      "layers_pattern": null,
+      "modules_to_save": null
+    },
+    "error_msg": ""
+  },
+  "train_info": {
+    "accelerator_memory_reserved_avg": 16046795574,
+    "accelerator_memory_max": 23924310016,
+    "accelerator_memory_reserved_99th": 21848129536,
+    "train_time": 833.3988380349474,
+    "file_size": 73429560,
+    "num_trainable_params": 18350080,
+    "num_total_params": 3231099904,
+    "status": "success",
+    "metrics": [
+      {
+        "step": 250,
+        "valid accuracy": 0.34,
+        "train loss": 1.079560376405716,
+        "train samples": 1000,
+        "train time": 32.45247818226926,
+        "eval time": 7.630643949029036,
+        "tokens / sec": 6523.970182211688,
+        "mem allocated avg": 7124894255.104,
+        "mem reserved avg": 16177229201.408,
+        "elapsed time": 56.13032239302993
+      },
+      {
+        "step": 500,
+        "valid accuracy": 0.34,
+        "train loss": 0.7237268801927567,
+        "train samples": 2000,
+        "train time": 32.01787415763829,
+        "eval time": 7.168605238897726,
+        "tokens / sec": 6496.21517580923,
+        "mem allocated avg": 7118103789.568,
+        "mem reserved avg": 15777235206.144,
+        "elapsed time": 95.77683904604055
+      },
+      {
+        "step": 750,
+        "valid accuracy": 0.42,
+        "train loss": 0.675168325304985,
+        "train samples": 3000,
+        "train time": 32.78650948917493,
+        "eval time": 5.598999081994407,
+        "tokens / sec": 6539.3054442342645,
+        "mem allocated avg": 7127977418.752,
+        "mem reserved avg": 16003702456.32,
+        "elapsed time": 134.60598525498062
+      },
+      {
+        "step": 1000,
+        "valid accuracy": 0.4,
+        "train loss": 0.6531686210632324,
+        "train samples": 4000,
+        "train time": 32.1652398359729,
+        "eval time": 7.511855768039823,
+        "tokens / sec": 6477.054144859868,
+        "mem allocated avg": 7119628976.128,
+        "mem reserved avg": 16061768400.896,
+        "elapsed time": 174.7542857020162
+      },
+      {
+        "step": 1250,
+        "valid accuracy": 0.46,
+        "train loss": 0.6476688065528869,
+        "train samples": 5000,
+        "train time": 32.14541555568576,
+        "eval time": 7.27379516500514,
+        "tokens / sec": 6487.332529229494,
+        "mem allocated avg": 7119241891.84,
+        "mem reserved avg": 16092168716.288,
+        "elapsed time": 214.62127449002583
+      },
+      {
+        "step": 1500,
+        "valid accuracy": 0.44,
+        "train loss": 0.6379475784301758,
+        "train samples": 6000,
+        "train time": 32.336382034700364,
+        "eval time": 5.115500686923042,
+        "tokens / sec": 6473.544250416317,
+        "mem allocated avg": 7121132320.768,
+        "mem reserved avg": 16066952560.64,
+        "elapsed time": 252.54313600901514
+      },
+      {
+        "step": 1750,
+        "valid accuracy": 0.48,
+        "train loss": 0.629106404542923,
+        "train samples": 7000,
+        "train time": 32.432342515094206,
+        "eval time": 6.911258863052353,
+        "tokens / sec": 6455.130396534414,
+        "mem allocated avg": 7122440562.688,
+        "mem reserved avg": 16377867927.552,
+        "elapsed time": 292.34993721498176
+      },
+      {
+        "step": 2000,
+        "valid accuracy": 0.54,
+        "train loss": 0.6309418694972992,
+        "train samples": 8000,
+        "train time": 32.12539123080205,
+        "eval time": 7.336019244045019,
+        "tokens / sec": 6465.1664008642365,
+        "mem allocated avg": 7118683537.408,
+        "mem reserved avg": 16049957240.832,
+        "elapsed time": 332.2858091519447
+      },
+      {
+        "step": 2250,
+        "valid accuracy": 0.4,
+        "train loss": 0.620747056722641,
+        "train samples": 9000,
+        "train time": 33.009332997608,
+        "eval time": 7.506129134097137,
+        "tokens / sec": 6511.7341212431065,
+        "mem allocated avg": 7130056898.56,
+        "mem reserved avg": 16311253991.424,
+        "elapsed time": 373.26916755596176
+      },
+      {
+        "step": 2500,
+        "valid accuracy": 0.52,
+        "train loss": 0.6168318339586258,
+        "train samples": 10000,
+        "train time": 31.805065547465347,
+        "eval time": 5.380960662965663,
+        "tokens / sec": 6475.918110988273,
+        "mem allocated avg": 7116140187.648,
+        "mem reserved avg": 15806930878.464,
+        "elapsed time": 410.9146865049843
+      },
+      {
+        "step": 2750,
+        "valid accuracy": 0.46,
+        "train loss": 0.608428325176239,
+        "train samples": 11000,
+        "train time": 32.713718246202916,
+        "eval time": 5.9023718340322375,
+        "tokens / sec": 6476.824138588803,
+        "mem allocated avg": 7125151420.416,
+        "mem reserved avg": 16212520075.264,
+        "elapsed time": 449.9857141500106
+      },
+      {
+        "step": 3000,
+        "valid accuracy": 0.42,
+        "train loss": 0.5987516103982925,
+        "train samples": 12000,
+        "train time": 32.34677883586846,
+        "eval time": 5.252711561974138,
+        "tokens / sec": 6452.914556318785,
+        "mem allocated avg": 7120734087.168,
+        "mem reserved avg": 16107251433.472,
+        "elapsed time": 488.0572223030031
+      },
+      {
+        "step": 3250,
+        "valid accuracy": 0.46,
+        "train loss": 0.607488405585289,
+        "train samples": 13000,
+        "train time": 32.457327542011626,
+        "eval time": 5.465676949941553,
+        "tokens / sec": 6497.793132444966,
+        "mem allocated avg": 7123690192.896,
+        "mem reserved avg": 15982194065.408,
+        "elapsed time": 526.4285796340555
+      },
+      {
+        "step": 3500,
+        "valid accuracy": 0.54,
+        "train loss": 0.5906782056093216,
+        "train samples": 14000,
+        "train time": 32.442000140435994,
+        "eval time": 4.847414086922072,
+        "tokens / sec": 6465.384350287508,
+        "mem allocated avg": 7120448550.912,
+        "mem reserved avg": 15947347787.776,
+        "elapsed time": 564.1799046599772
+      },
+      {
+        "step": 3750,
+        "valid accuracy": 0.52,
+        "train loss": 0.5881383534669876,
+        "train samples": 15000,
+        "train time": 33.23058606637642,
+        "eval time": 4.371005772030912,
+        "tokens / sec": 6521.191036689714,
+        "mem allocated avg": 7132748165.12,
+        "mem reserved avg": 16321999798.272,
+        "elapsed time": 602.2411791590275
+      },
+      {
+        "step": 4000,
+        "valid accuracy": 0.46,
+        "train loss": 0.5990954364538192,
+        "train samples": 16000,
+        "train time": 31.758256679633632,
+        "eval time": 5.704399707028642,
+        "tokens / sec": 6435.271370895592,
+        "mem allocated avg": 7113135618.048,
+        "mem reserved avg": 16033154859.008,
+        "elapsed time": 640.1719372379594
+      },
+      {
+        "step": 4250,
+        "valid accuracy": 0.48,
+        "train loss": 0.5854876210689545,
+        "train samples": 17000,
+        "train time": 32.489558273460716,
+        "eval time": 4.558923620032147,
+        "tokens / sec": 6506.3673141002455,
+        "mem allocated avg": 7124313315.328,
+        "mem reserved avg": 15974761758.72,
+        "elapsed time": 677.6578216060298
+      },
+      {
+        "step": 4500,
+        "valid accuracy": 0.46,
+        "train loss": 0.5922646141052246,
+        "train samples": 18000,
+        "train time": 32.16165658435784,
+        "eval time": 4.171202325029299,
+        "tokens / sec": 6461.669642386346,
+        "mem allocated avg": 7119053879.296,
+        "mem reserved avg": 15951567257.6,
+        "elapsed time": 714.457467147964
+      },
+      {
+        "step": 4750,
+        "valid accuracy": 0.54,
+        "train loss": 0.5838238140344619,
+        "train samples": 19000,
+        "train time": 32.32759334752336,
+        "eval time": 5.960213870042935,
+        "tokens / sec": 6494.111632225279,
+        "mem allocated avg": 7121662765.056,
+        "mem reserved avg": 16008442019.84,
+        "elapsed time": 753.1879774139961
+      },
+      {
+        "step": 5000,
+        "valid accuracy": 0.5,
+        "train loss": 0.5905282303094864,
+        "train samples": 20000,
+        "train time": 32.046189901418984,
+        "eval time": 4.635392899974249,
+        "tokens / sec": 6499.368587676549,
+        "mem allocated avg": 7118618368.0,
+        "mem reserved avg": 15671605854.208,
+        "elapsed time": 790.3203212779481
+      },
+      {
+        "step": 5000,
+        "test accuracy": 0.4874905231235785,
+        "train loss": 0.5905282303094864,
+        "train samples": 20000,
+        "train total tokens": 4198051,
+        "forgetting": 0.051932573318481445
+      }
+    ]
+  },
+  "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": "98d4849e3ecc30142782ec0c418a1f86f78fb562",
+      "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"
+  }
+}