SLM_vs_LLM_experiments
/
max_seq_length_128_experiments
/Qwen
/Qwen1.5_1.8B_scotus
/trainer_state.json
| { | |
| "best_metric": 1.6402901411056519, | |
| "best_model_checkpoint": "../../experiments_checkpoints/MAdAiLab/Qwen/Qwen1.5_1.8B_scotus/checkpoint-300", | |
| "epoch": 3.0, | |
| "eval_steps": 50, | |
| "global_step": 471, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 206.6490020751953, | |
| "learning_rate": 4.893842887473461e-06, | |
| "loss": 5.8906, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 142.15260314941406, | |
| "learning_rate": 4.787685774946922e-06, | |
| "loss": 4.4914, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 47.5128173828125, | |
| "learning_rate": 4.6815286624203824e-06, | |
| "loss": 3.1742, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 90.23841094970703, | |
| "learning_rate": 4.575371549893844e-06, | |
| "loss": 2.3301, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 106.74323272705078, | |
| "learning_rate": 4.469214437367304e-06, | |
| "loss": 2.1746, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_accuracy": 0.2692857142857143, | |
| "eval_f1_macro": 0.11482650807382964, | |
| "eval_f1_micro": 0.2692857142857143, | |
| "eval_loss": 2.2791965007781982, | |
| "eval_runtime": 3.6276, | |
| "eval_samples_per_second": 385.929, | |
| "eval_steps_per_second": 12.129, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 87.4228515625, | |
| "learning_rate": 4.3630573248407645e-06, | |
| "loss": 1.9594, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 61.52133560180664, | |
| "learning_rate": 4.256900212314226e-06, | |
| "loss": 1.8109, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 55.21061706542969, | |
| "learning_rate": 4.150743099787686e-06, | |
| "loss": 1.7418, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 71.5464096069336, | |
| "learning_rate": 4.044585987261147e-06, | |
| "loss": 1.6996, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 50.51081848144531, | |
| "learning_rate": 3.938428874734608e-06, | |
| "loss": 1.5758, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "eval_accuracy": 0.41, | |
| "eval_f1_macro": 0.22906680413417876, | |
| "eval_f1_micro": 0.41, | |
| "eval_loss": 1.930892825126648, | |
| "eval_runtime": 3.6465, | |
| "eval_samples_per_second": 383.927, | |
| "eval_steps_per_second": 12.066, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 108.21426391601562, | |
| "learning_rate": 3.832271762208068e-06, | |
| "loss": 1.6871, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 62.29310607910156, | |
| "learning_rate": 3.7261146496815285e-06, | |
| "loss": 1.6055, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 79.33343505859375, | |
| "learning_rate": 3.6199575371549893e-06, | |
| "loss": 1.6043, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 77.69229888916016, | |
| "learning_rate": 3.51380042462845e-06, | |
| "loss": 1.5793, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 67.4636001586914, | |
| "learning_rate": 3.407643312101911e-06, | |
| "loss": 1.7312, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "eval_accuracy": 0.4714285714285714, | |
| "eval_f1_macro": 0.2546511715546434, | |
| "eval_f1_micro": 0.4714285714285714, | |
| "eval_loss": 1.7475446462631226, | |
| "eval_runtime": 3.6415, | |
| "eval_samples_per_second": 384.454, | |
| "eval_steps_per_second": 12.083, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 72.03321838378906, | |
| "learning_rate": 3.3014861995753718e-06, | |
| "loss": 1.5695, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 59.265869140625, | |
| "learning_rate": 3.195329087048832e-06, | |
| "loss": 1.2879, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 78.44482421875, | |
| "learning_rate": 3.089171974522293e-06, | |
| "loss": 1.3846, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "grad_norm": 46.30598068237305, | |
| "learning_rate": 2.983014861995754e-06, | |
| "loss": 1.365, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 62.2653923034668, | |
| "learning_rate": 2.8768577494692146e-06, | |
| "loss": 1.2859, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "eval_accuracy": 0.4685714285714286, | |
| "eval_f1_macro": 0.30187587586686543, | |
| "eval_f1_micro": 0.4685714285714286, | |
| "eval_loss": 1.793928623199463, | |
| "eval_runtime": 3.6406, | |
| "eval_samples_per_second": 384.55, | |
| "eval_steps_per_second": 12.086, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 38.229000091552734, | |
| "learning_rate": 2.7707006369426754e-06, | |
| "loss": 1.3562, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 64.51058197021484, | |
| "learning_rate": 2.6645435244161363e-06, | |
| "loss": 1.2486, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "grad_norm": 45.04010009765625, | |
| "learning_rate": 2.5583864118895966e-06, | |
| "loss": 1.3775, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 36.0023307800293, | |
| "learning_rate": 2.4522292993630575e-06, | |
| "loss": 1.2434, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 42.964847564697266, | |
| "learning_rate": 2.3460721868365183e-06, | |
| "loss": 1.1971, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_accuracy": 0.4957142857142857, | |
| "eval_f1_macro": 0.33073042661457275, | |
| "eval_f1_micro": 0.4957142857142857, | |
| "eval_loss": 1.7145981788635254, | |
| "eval_runtime": 3.8826, | |
| "eval_samples_per_second": 360.584, | |
| "eval_steps_per_second": 11.333, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "grad_norm": 43.22111511230469, | |
| "learning_rate": 2.239915074309979e-06, | |
| "loss": 1.2471, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 49.360374450683594, | |
| "learning_rate": 2.13375796178344e-06, | |
| "loss": 1.2869, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 54.94723129272461, | |
| "learning_rate": 2.0276008492569003e-06, | |
| "loss": 1.1953, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "grad_norm": 58.562225341796875, | |
| "learning_rate": 1.921443736730361e-06, | |
| "loss": 1.1973, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "grad_norm": 47.98759460449219, | |
| "learning_rate": 1.8152866242038217e-06, | |
| "loss": 1.1271, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "eval_accuracy": 0.515, | |
| "eval_f1_macro": 0.35688242427831446, | |
| "eval_f1_micro": 0.515, | |
| "eval_loss": 1.6402901411056519, | |
| "eval_runtime": 3.6453, | |
| "eval_samples_per_second": 384.061, | |
| "eval_steps_per_second": 12.071, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "grad_norm": 49.362281799316406, | |
| "learning_rate": 1.7091295116772823e-06, | |
| "loss": 1.2777, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 39.05293273925781, | |
| "learning_rate": 1.6029723991507432e-06, | |
| "loss": 0.9527, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 44.27582550048828, | |
| "learning_rate": 1.496815286624204e-06, | |
| "loss": 0.8315, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "grad_norm": 37.02938461303711, | |
| "learning_rate": 1.3906581740976646e-06, | |
| "loss": 0.8729, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "grad_norm": 45.579402923583984, | |
| "learning_rate": 1.2845010615711254e-06, | |
| "loss": 0.8436, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "eval_accuracy": 0.51, | |
| "eval_f1_macro": 0.3717426255688204, | |
| "eval_f1_micro": 0.51, | |
| "eval_loss": 1.7994866371154785, | |
| "eval_runtime": 3.644, | |
| "eval_samples_per_second": 384.193, | |
| "eval_steps_per_second": 12.075, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "grad_norm": 28.59382438659668, | |
| "learning_rate": 1.178343949044586e-06, | |
| "loss": 0.7411, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 57.68552780151367, | |
| "learning_rate": 1.0721868365180468e-06, | |
| "loss": 0.8667, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 60.32491683959961, | |
| "learning_rate": 9.660297239915076e-07, | |
| "loss": 0.7807, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 52.709407806396484, | |
| "learning_rate": 8.598726114649681e-07, | |
| "loss": 0.7298, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 61.55146408081055, | |
| "learning_rate": 7.537154989384289e-07, | |
| "loss": 0.7723, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "eval_accuracy": 0.5028571428571429, | |
| "eval_f1_macro": 0.37231960236652384, | |
| "eval_f1_micro": 0.5028571428571429, | |
| "eval_loss": 1.8401116132736206, | |
| "eval_runtime": 3.6434, | |
| "eval_samples_per_second": 384.253, | |
| "eval_steps_per_second": 12.077, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "grad_norm": 46.969879150390625, | |
| "learning_rate": 6.475583864118897e-07, | |
| "loss": 0.8337, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 52.0188102722168, | |
| "learning_rate": 5.414012738853504e-07, | |
| "loss": 0.8848, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 53.10190963745117, | |
| "learning_rate": 4.35244161358811e-07, | |
| "loss": 0.718, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 58.61328125, | |
| "learning_rate": 3.290870488322718e-07, | |
| "loss": 0.7417, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "grad_norm": 63.344459533691406, | |
| "learning_rate": 2.229299363057325e-07, | |
| "loss": 0.7211, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "eval_accuracy": 0.5178571428571429, | |
| "eval_f1_macro": 0.375345717505455, | |
| "eval_f1_micro": 0.5178571428571429, | |
| "eval_loss": 1.8231920003890991, | |
| "eval_runtime": 3.6452, | |
| "eval_samples_per_second": 384.067, | |
| "eval_steps_per_second": 12.071, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "grad_norm": 42.954627990722656, | |
| "learning_rate": 1.1677282377919321e-07, | |
| "loss": 0.7757, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "grad_norm": 49.02348327636719, | |
| "learning_rate": 1.0615711252653928e-08, | |
| "loss": 0.8159, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 471, | |
| "total_flos": 1.405849146753024e+16, | |
| "train_loss": 1.457581732683121, | |
| "train_runtime": 520.0867, | |
| "train_samples_per_second": 28.841, | |
| "train_steps_per_second": 0.906 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 471, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 50, | |
| "total_flos": 1.405849146753024e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |