Add superseded/exploratory MetaMathQA VeLoRA results

MetaMathQA/lora--llama-3.2-3B-rank32-velora.json

@@ -0,0 +1,386 @@
+{
+  "run_info": {
+    "created_at": "2026-10-05T19:54:18+00:00",
+    "total_time": 1236.0781992480624,
+    "experiment_name": "lora/llama-3.2-3B-rank32-velora",
+    "peft_branch": "metamathqa-velora",
+    "peft_worktree_is_dirty": false,
+    "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": 0.0001,
+        "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": "CAUSAL_LM",
+      "peft_type": "LORA",
+      "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,
+      "r": 32,
+      "target_modules": [
+        "q_proj",
+        "v_proj"
+      ],
+      "exclude_modules": null,
+      "lora_alpha": 64,
+      "lora_dropout": 0.0,
+      "fan_in_fan_out": false,
+      "bias": "none",
+      "use_rslora": false,
+      "modules_to_save": null,
+      "init_lora_weights": true,
+      "layers_to_transform": null,
+      "layers_pattern": null,
+      "rank_pattern": {},
+      "alpha_pattern": {},
+      "megatron_config": null,
+      "megatron_core": "megatron.core",
+      "trainable_token_indices": null,
+      "loftq_config": {},
+      "eva_config": null,
+      "corda_config": null,
+      "astra_config": null,
+      "lora_ga_config": null,
+      "use_dora": false,
+      "velora_config": {
+        "num_groups": 32,
+        "scale": 1.0,
+        "init_type": "batch_average_once"
+      },
+      "alora_invocation_tokens": null,
+      "use_qalora": false,
+      "qalora_group_size": 16,
+      "monteclora_config": null,
+      "layer_replication": null,
+      "lora_bias": false,
+      "target_parameters": null,
+      "use_bdlora": null,
+      "arrow_config": null,
+      "kasa_config": null,
+      "ensure_weight_tying": false
+    },
+    "error_msg": ""
+  },
+  "train_info": {
+    "accelerator_memory_reserved_avg": 13472439572,
+    "accelerator_memory_max": 20285751296,
+    "accelerator_memory_reserved_99th": 18492686336,
+    "train_time": 636.5510473318864,
+    "file_size": 36744168,
+    "num_trainable_params": 9175040,
+    "num_total_params": 3221924864,
+    "status": "success",
+    "metrics": [
+      {
+        "step": 250,
+        "valid accuracy": 0.36,
+        "train loss": 1.0681839339733123,
+        "train samples": 1000,
+        "train time": 21.78048365120776,
+        "eval time": 6.390597966033965,
+        "tokens / sec": 9720.583040783848,
+        "mem allocated avg": 6976404051.968,
+        "mem reserved avg": 13557836021.76,
+        "elapsed time": 83.03203223505989
+      },
+      {
+        "step": 500,
+        "valid accuracy": 0.32,
+        "train loss": 0.7416704635620117,
+        "train samples": 2000,
+        "train time": 21.42817924497649,
+        "eval time": 6.158339032903314,
+        "tokens / sec": 9706.610982767528,
+        "mem allocated avg": 6969429551.104,
+        "mem reserved avg": 13233054285.824,
+        "elapsed time": 110.93252640403807
+      },
+      {
+        "step": 750,
+        "valid accuracy": 0.28,
+        "train loss": 0.6946747965812683,
+        "train samples": 3000,
+        "train time": 21.996751863858663,
+        "eval time": 5.121608830988407,
+        "tokens / sec": 9746.939062956264,
+        "mem allocated avg": 6979127867.392,
+        "mem reserved avg": 13454379319.296,
+        "elapsed time": 138.35988097602967
+      },
+      {
+        "step": 1000,
+        "valid accuracy": 0.38,
+        "train loss": 0.6761692506074906,
+        "train samples": 4000,
+        "train time": 21.540250341407955,
+        "eval time": 3.3909282711101696,
+        "tokens / sec": 9671.939587419964,
+        "mem allocated avg": 6970733434.88,
+        "mem reserved avg": 13472238665.728,
+        "elapsed time": 163.62149686401244
+      },
+      {
+        "step": 1250,
+        "valid accuracy": 0.46,
+        "train loss": 0.6729425090551376,
+        "train samples": 5000,
+        "train time": 21.533420955762267,
+        "eval time": 3.471343936980702,
+        "tokens / sec": 9684.387837325772,
+        "mem allocated avg": 6970701307.904,
+        "mem reserved avg": 13491230474.24,
+        "elapsed time": 188.93079189106356
+      },
+      {
+        "step": 1500,
+        "valid accuracy": 0.4,
+        "train loss": 0.6665173796415329,
+        "train samples": 6000,
+        "train time": 21.66879504127428,
+        "eval time": 4.656126852962188,
+        "tokens / sec": 9660.481794270081,
+        "mem allocated avg": 6972849438.72,
+        "mem reserved avg": 13442794651.648,
+        "elapsed time": 215.57841004501097
+      },
+      {
+        "step": 1750,
+        "valid accuracy": 0.44,
+        "train loss": 0.657946897149086,
+        "train samples": 7000,
+        "train time": 21.757563578314148,
+        "eval time": 3.890562706044875,
+        "tokens / sec": 9622.17112437465,
+        "mem allocated avg": 6974063777.792,
+        "mem reserved avg": 13771401592.832,
+        "elapsed time": 241.5479369009845
+      },
+      {
+        "step": 2000,
+        "valid accuracy": 0.4,
+        "train loss": 0.661081365942955,
+        "train samples": 8000,
+        "train time": 21.534604626824148,
+        "eval time": 4.96107355505228,
+        "tokens / sec": 9644.755666481457,
+        "mem allocated avg": 6970012760.064,
+        "mem reserved avg": 13468514123.776,
+        "elapsed time": 268.3700661120238
+      },
+      {
+        "step": 2250,
+        "valid accuracy": 0.36,
+        "train loss": 0.6539777635335923,
+        "train samples": 9000,
+        "train time": 22.16575576493051,
+        "eval time": 4.421835187007673,
+        "tokens / sec": 9697.300749838605,
+        "mem allocated avg": 6980709560.32,
+        "mem reserved avg": 13729752154.112,
+        "elapsed time": 295.2758137159981
+      },
+      {
+        "step": 2500,
+        "valid accuracy": 0.38,
+        "train loss": 0.6522042131423951,
+        "train samples": 10000,
+        "train time": 21.251954262377694,
+        "eval time": 4.000645584077574,
+        "tokens / sec": 9691.673408342644,
+        "mem allocated avg": 6966783256.576,
+        "mem reserved avg": 13264545120.256,
+        "elapsed time": 320.8456818240229
+      },
+      {
+        "step": 2750,
+        "valid accuracy": 0.42,
+        "train loss": 0.6438830876350403,
+        "train samples": 11000,
+        "train time": 21.968822030350566,
+        "eval time": 6.1554399110609666,
+        "tokens / sec": 9644.62271610559,
+        "mem allocated avg": 6976473348.096,
+        "mem reserved avg": 13598789206.016,
+        "elapsed time": 349.2858487960184
+      },
+      {
+        "step": 3000,
+        "valid accuracy": 0.42,
+        "train loss": 0.6366478145122528,
+        "train samples": 12000,
+        "train time": 21.68107317085378,
+        "eval time": 5.104307060013525,
+        "tokens / sec": 9627.337095130535,
+        "mem allocated avg": 6972363700.224,
+        "mem reserved avg": 13532116549.632,
+        "elapsed time": 376.39782021706924
+      },
+      {
+        "step": 3250,
+        "valid accuracy": 0.46,
+        "train loss": 0.6465079092979431,
+        "train samples": 13000,
+        "train time": 21.759236190933734,
+        "eval time": 3.756050892989151,
+        "tokens / sec": 9692.48176495619,
+        "mem allocated avg": 6975104358.4,
+        "mem reserved avg": 13401606586.368,
+        "elapsed time": 402.22271684603766
+      },
+      {
+        "step": 3500,
+        "valid accuracy": 0.42,
+        "train loss": 0.6306205905675888,
+        "train samples": 14000,
+        "train time": 21.757495325175114,
+        "eval time": 4.861972064943984,
+        "tokens / sec": 9640.355972284317,
+        "mem allocated avg": 6972123797.504,
+        "mem reserved avg": 13431537139.712,
+        "elapsed time": 429.1680319099687
+      },
+      {
+        "step": 3750,
+        "valid accuracy": 0.4,
+        "train loss": 0.6289126653671264,
+        "train samples": 15000,
+        "train time": 22.352520546875894,
+        "eval time": 4.041001550969668,
+        "tokens / sec": 9694.790327808805,
+        "mem allocated avg": 6983769554.944,
+        "mem reserved avg": 13723150319.616,
+        "elapsed time": 455.8837911869632
+      },
+      {
+        "step": 4000,
+        "valid accuracy": 0.44,
+        "train loss": 0.6428844587802887,
+        "train samples": 16000,
+        "train time": 21.264055045554414,
+        "eval time": 3.583834795979783,
+        "tokens / sec": 9611.195962490108,
+        "mem allocated avg": 6964402135.04,
+        "mem reserved avg": 13438323523.584,
+        "elapsed time": 481.0560817460064
+      },
+      {
+        "step": 4250,
+        "valid accuracy": 0.42,
+        "train loss": 0.6273980221748352,
+        "train samples": 17000,
+        "train time": 21.79138396505732,
+        "eval time": 6.160092841950245,
+        "tokens / sec": 9700.5770876675,
+        "mem allocated avg": 6975876909.056,
+        "mem reserved avg": 13418249584.64,
+        "elapsed time": 509.3064287250163
+      },
+      {
+        "step": 4500,
+        "valid accuracy": 0.38,
+        "train loss": 0.6342040377855301,
+        "train samples": 18000,
+        "train time": 21.543688778183423,
+        "eval time": 6.162608459009789,
+        "tokens / sec": 9646.351752465454,
+        "mem allocated avg": 6970482511.872,
+        "mem reserved avg": 13394266554.368,
+        "elapsed time": 537.3351886629825
+      },
+      {
+        "step": 4750,
+        "valid accuracy": 0.46,
+        "train loss": 0.6265834709405899,
+        "train samples": 19000,
+        "train time": 21.662689781631343,
+        "eval time": 4.292773816967383,
+        "tokens / sec": 9691.271126359185,
+        "mem allocated avg": 6972388280.32,
+        "mem reserved avg": 13463900389.376,
+        "elapsed time": 563.5941182300448
+      },
+      {
+        "step": 5000,
+        "valid accuracy": 0.42,
+        "train loss": 0.6329876257181167,
+        "train samples": 20000,
+        "train time": 21.435827036737464,
+        "eval time": 3.733421070035547,
+        "tokens / sec": 9716.443393718493,
+        "mem allocated avg": 6970227410.944,
+        "mem reserved avg": 13161105195.008,
+        "elapsed time": 589.0778772250051
+      },
+      {
+        "step": 5000,
+        "test accuracy": 0.44124336618650495,
+        "train loss": 0.6329876257181167,
+        "train samples": 20000,
+        "train total tokens": 4198051,
+        "forgetting": 0.10022342205047607
+      }
+    ]
+  },
+  "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": "5.18.0",
+      "transformers-commit-hash": null,
+      "peft-version": "0.21.3.dev0",
+      "peft-commit-hash": "e13de7e469d37e3163d93f2356ec013c7a7e1f0a",
+      "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"
+  }
+}