Add superseded/exploratory BEFT results for MetaMathQA (2)

MetaMathQA/beft--llama-3.2-3B-target-all-linear-lr_0.001.json

@@ -0,0 +1,355 @@
+{
+  "run_info": {
+    "created_at": "2026-10-25T18:25:17+00:00",
+    "total_time": 2351.752165269998,
+    "experiment_name": "beft/llama-3.2-3B-target-v_proj",
+    "peft_branch": "update-tinylora-image-gen-experiment",
+    "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": true,
+      "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": 0.1
+      },
+      "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": "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": [
+        "gate_proj",
+        "q_proj",
+        "down_proj",
+        "o_proj",
+        "v_proj",
+        "k_proj",
+        "up_proj"
+      ],
+      "modules_to_save": null,
+      "init_weights": true
+    },
+    "error_msg": ""
+  },
+  "train_info": {
+    "accelerator_memory_reserved_avg": 9740771917,
+    "accelerator_memory_max": 13304332288,
+    "accelerator_memory_reserved_99th": 12129927168,
+    "train_time": 2079.222630835,
+    "file_size": 3121184,
+    "num_trainable_params": 774144,
+    "num_total_params": 3213523968,
+    "status": "success",
+    "metrics": [
+      {
+        "step": 250,
+        "valid accuracy": 0.34,
+        "train loss": 0.8553074641227723,
+        "train samples": 1000,
+        "train time": 80.40527667801507,
+        "eval time": 16.738490661999094,
+        "tokens / sec": 2633.148081161812,
+        "mem allocated avg": 6738612701.184,
+        "mem reserved avg": 9814201597.952,
+        "elapsed time": 121.00934449399938
+      },
+      {
+        "step": 500,
+        "valid accuracy": 0.24,
+        "train loss": 0.752122419476509,
+        "train samples": 2000,
+        "train time": 80.95715767595902,
+        "eval time": 16.304893870998058,
+        "tokens / sec": 2569.19840037524,
+        "mem allocated avg": 6733570070.528,
+        "mem reserved avg": 9632856670.208,
+        "elapsed time": 218.5353863679993
+      },
+      {
+        "step": 750,
+        "valid accuracy": 0.34,
+        "train loss": 0.7467145235538483,
+        "train samples": 3000,
+        "train time": 85.3085584900291,
+        "eval time": 16.651178225998592,
+        "tokens / sec": 2513.2413886123663,
+        "mem allocated avg": 6740894121.984,
+        "mem reserved avg": 9752536940.544,
+        "elapsed time": 320.7634998909998
+      },
+      {
+        "step": 1000,
+        "valid accuracy": 0.26,
+        "train loss": 0.7276310513019562,
+        "train samples": 4000,
+        "train time": 83.75018924500182,
+        "eval time": 8.149916423997638,
+        "tokens / sec": 2487.588408791965,
+        "mem allocated avg": 6735134597.12,
+        "mem reserved avg": 9668181098.496,
+        "elapsed time": 412.91634084900215
+      },
+      {
+        "step": 1250,
+        "valid accuracy": 0.26,
+        "train loss": 0.7247908207178115,
+        "train samples": 5000,
+        "train time": 82.59961786699932,
+        "eval time": 16.324929948001227,
+        "tokens / sec": 2524.6848034549603,
+        "mem allocated avg": 6734710927.36,
+        "mem reserved avg": 9730944663.552,
+        "elapsed time": 512.0997955800012
+      },
+      {
+        "step": 1500,
+        "valid accuracy": 0.36,
+        "train loss": 0.7197483174800873,
+        "train samples": 6000,
+        "train time": 83.59951431699301,
+        "eval time": 10.140888480000285,
+        "tokens / sec": 2503.973877243566,
+        "mem allocated avg": 6735837075.456,
+        "mem reserved avg": 9728948174.848,
+        "elapsed time": 606.1017242310008
+      },
+      {
+        "step": 1750,
+        "valid accuracy": 0.38,
+        "train loss": 0.7081621831655502,
+        "train samples": 7000,
+        "train time": 83.0456547679787,
+        "eval time": 16.527016435000405,
+        "tokens / sec": 2520.9627232745293,
+        "mem allocated avg": 6736629872.64,
+        "mem reserved avg": 9868903710.72,
+        "elapsed time": 705.9425626740012
+      },
+      {
+        "step": 2000,
+        "valid accuracy": 0.44,
+        "train loss": 0.7083892493247986,
+        "train samples": 8000,
+        "train time": 80.63616915501188,
+        "eval time": 9.313062053999602,
+        "tokens / sec": 2575.7175988945255,
+        "mem allocated avg": 6734107213.824,
+        "mem reserved avg": 9703438417.92,
+        "elapsed time": 796.153989022001
+      },
+      {
+        "step": 2250,
+        "valid accuracy": 0.34,
+        "train loss": 0.6948926150798798,
+        "train samples": 9000,
+        "train time": 85.13438451103502,
+        "eval time": 15.885713808002038,
+        "tokens / sec": 2524.8082926134116,
+        "mem allocated avg": 6742452533.248,
+        "mem reserved avg": 9850977255.424,
+        "elapsed time": 897.4274087580015
+      },
+      {
+        "step": 2500,
+        "valid accuracy": 0.32,
+        "train loss": 0.6885149501562119,
+        "train samples": 10000,
+        "train time": 81.97940732295683,
+        "eval time": 15.732348456000182,
+        "tokens / sec": 2512.423628394819,
+        "mem allocated avg": 6731558375.424,
+        "mem reserved avg": 9665899397.12,
+        "elapsed time": 995.4055037040016
+      },
+      {
+        "step": 2750,
+        "valid accuracy": 0.4,
+        "train loss": 0.6741755467653274,
+        "train samples": 11000,
+        "train time": 84.20481224698597,
+        "eval time": 9.697093232996849,
+        "tokens / sec": 2516.257614571002,
+        "mem allocated avg": 6739382056.96,
+        "mem reserved avg": 9734383992.832,
+        "elapsed time": 1089.5657239630018
+      },
+      {
+        "step": 3000,
+        "valid accuracy": 0.44,
+        "train loss": 0.6621904965639115,
+        "train samples": 12000,
+        "train time": 82.08975169201585,
+        "eval time": 12.160228757998993,
+        "tokens / sec": 2542.7169128628443,
+        "mem allocated avg": 6735558879.232,
+        "mem reserved avg": 9766118096.896,
+        "elapsed time": 1184.0932429550012
+      },
+      {
+        "step": 3250,
+        "valid accuracy": 0.38,
+        "train loss": 0.6654124256372452,
+        "train samples": 13000,
+        "train time": 82.60495989599804,
+        "eval time": 9.422949998002878,
+        "tokens / sec": 2553.1275635934003,
+        "mem allocated avg": 6736861765.632,
+        "mem reserved avg": 9693153984.512,
+        "elapsed time": 1276.3797298749996
+      },
+      {
+        "step": 3500,
+        "valid accuracy": 0.48,
+        "train loss": 0.6440585649013519,
+        "train samples": 14000,
+        "train time": 83.49933236895231,
+        "eval time": 16.45183570000154,
+        "tokens / sec": 2511.9961327737715,
+        "mem allocated avg": 6736224780.288,
+        "mem reserved avg": 9762519384.064,
+        "elapsed time": 1376.5911240860005
+      },
+      {
+        "step": 3750,
+        "valid accuracy": 0.52,
+        "train loss": 0.6372588503360749,
+        "train samples": 15000,
+        "train time": 88.35799511100049,
+        "eval time": 16.617606013001932,
+        "tokens / sec": 2452.5567802638006,
+        "mem allocated avg": 6744234921.984,
+        "mem reserved avg": 9867569922.048,
+        "elapsed time": 1481.8319968249998
+      },
+      {
+        "step": 4000,
+        "valid accuracy": 0.42,
+        "train loss": 0.6461446195840835,
+        "train samples": 16000,
+        "train time": 80.48123654198207,
+        "eval time": 9.247131195999827,
+        "tokens / sec": 2539.386927701978,
+        "mem allocated avg": 6730679050.24,
+        "mem reserved avg": 9747973537.792,
+        "elapsed time": 1571.8246489320009
+      },
+      {
+        "step": 4250,
+        "valid accuracy": 0.46,
+        "train loss": 0.6245053981542588,
+        "train samples": 17000,
+        "train time": 80.87224496699855,
+        "eval time": 16.23763043100189,
+        "tokens / sec": 2613.8633852227213,
+        "mem allocated avg": 6737841104.896,
+        "mem reserved avg": 9717875212.288,
+        "elapsed time": 1669.188425274002
+      },
+      {
+        "step": 4500,
+        "valid accuracy": 0.52,
+        "train loss": 0.629442654967308,
+        "train samples": 18000,
+        "train time": 81.4218764899997,
+        "eval time": 16.688479027998255,
+        "tokens / sec": 2552.3607285754997,
+        "mem allocated avg": 6734041757.696,
+        "mem reserved avg": 9734392381.44,
+        "elapsed time": 1767.555608701001
+      },
+      {
+        "step": 4750,
+        "valid accuracy": 0.5,
+        "train loss": 0.6200414321422577,
+        "train samples": 19000,
+        "train time": 82.67123844201706,
+        "eval time": 11.222833117000846,
+        "tokens / sec": 2539.4442366705857,
+        "mem allocated avg": 6736360562.688,
+        "mem reserved avg": 9768634679.296,
+        "elapsed time": 1861.7091022060013
+      },
+      {
+        "step": 5000,
+        "valid accuracy": 0.5,
+        "train loss": 0.6259837001562119,
+        "train samples": 20000,
+        "train time": 81.25220811798499,
+        "eval time": 9.313751439000043,
+        "tokens / sec": 2563.376489381803,
+        "mem allocated avg": 6733851441.152,
+        "mem reserved avg": 9605929238.528,
+        "elapsed time": 1952.5241099100022
+      },
+      {
+        "step": 5000,
+        "test accuracy": 0.43290371493555724,
+        "train loss": 0.6259837001562119,
+        "train samples": 20000,
+        "train total tokens": 4198051,
+        "forgetting": 0.45430731773376465
+      }
+    ]
+  },
+  "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": "5.16.0",
+      "transformers-commit-hash": null,
+      "peft-version": "0.21.3.dev0",
+      "peft-commit-hash": "c2c2b15a6e30e72fe6c7a8e43ac2be1428003978",
+      "datasets-version": "5.0.1",
+      "datasets-commit-hash": null,
+      "bitsandbytes-version": "0.50.2",
+      "bitsandbytes-commit-hash": null,
+      "torch-version": "2.13.0+cu130",
+      "torch-commit-hash": null
+    },
+    "system_info": {
+      "system": "Linux",
+      "release": "7.0.0-38-generic",
+      "version": "#38~24.04.4-Ubuntu SMP PREEMPT_DYNAMIC Mon Sep 14 16:37:11 UTC 2",
+      "machine": "x86_64",
+      "processor": "x86_64",
+      "accelerator": "NVIDIA GeForce RTX 5070 Ti"
+    },
+    "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 92.0  (built against CUDA 13.2)\n  - Magma 2.6.1\n  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=cf30153c4c131c8164ee7798e5022d810682e2cb, 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=9.20.0, 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=2.13.0, USE_CUDA=1, 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=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
+  }
+}