|
{ |
|
"best_metric": 0.15877817571163177, |
|
"best_model_checkpoint": "./vit-finetune-scrap/checkpoint-300", |
|
"epoch": 4.0, |
|
"eval_steps": 100, |
|
"global_step": 624, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 2.869983673095703, |
|
"learning_rate": 0.00019679487179487178, |
|
"loss": 1.9747, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"grad_norm": 2.9758315086364746, |
|
"learning_rate": 0.0001935897435897436, |
|
"loss": 1.209, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"grad_norm": 3.3387157917022705, |
|
"learning_rate": 0.00019038461538461538, |
|
"loss": 0.7205, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"grad_norm": 2.921093463897705, |
|
"learning_rate": 0.0001871794871794872, |
|
"loss": 0.4159, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"grad_norm": 2.4197134971618652, |
|
"learning_rate": 0.00018397435897435897, |
|
"loss": 0.3879, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"grad_norm": 2.498006582260132, |
|
"learning_rate": 0.00018076923076923077, |
|
"loss": 0.2695, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"grad_norm": 0.9913655519485474, |
|
"learning_rate": 0.00017756410256410257, |
|
"loss": 0.33, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"grad_norm": 5.167428493499756, |
|
"learning_rate": 0.00017435897435897436, |
|
"loss": 0.2374, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"grad_norm": 5.158258438110352, |
|
"learning_rate": 0.00017115384615384616, |
|
"loss": 0.2531, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"grad_norm": 2.1700403690338135, |
|
"learning_rate": 0.00016794871794871796, |
|
"loss": 0.1672, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"eval_accuracy": 0.9485530546623794, |
|
"eval_loss": 0.22496841847896576, |
|
"eval_runtime": 11.2945, |
|
"eval_samples_per_second": 55.071, |
|
"eval_steps_per_second": 6.906, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"grad_norm": 4.994268894195557, |
|
"learning_rate": 0.00016474358974358976, |
|
"loss": 0.1319, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"grad_norm": 2.2666022777557373, |
|
"learning_rate": 0.00016153846153846155, |
|
"loss": 0.283, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"grad_norm": 3.1912319660186768, |
|
"learning_rate": 0.00015833333333333332, |
|
"loss": 0.1666, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"grad_norm": 3.1358578205108643, |
|
"learning_rate": 0.00015512820512820515, |
|
"loss": 0.2819, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"grad_norm": 2.050241470336914, |
|
"learning_rate": 0.00015192307692307692, |
|
"loss": 0.2874, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"grad_norm": 5.193918704986572, |
|
"learning_rate": 0.00014871794871794872, |
|
"loss": 0.2103, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"grad_norm": 0.405056893825531, |
|
"learning_rate": 0.00014551282051282051, |
|
"loss": 0.1818, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"grad_norm": 11.979884147644043, |
|
"learning_rate": 0.0001423076923076923, |
|
"loss": 0.1626, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"grad_norm": 0.3927968442440033, |
|
"learning_rate": 0.0001391025641025641, |
|
"loss": 0.0824, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"grad_norm": 0.11850783228874207, |
|
"learning_rate": 0.0001358974358974359, |
|
"loss": 0.1277, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"eval_accuracy": 0.9372990353697749, |
|
"eval_loss": 0.24667073786258698, |
|
"eval_runtime": 76.2101, |
|
"eval_samples_per_second": 8.162, |
|
"eval_steps_per_second": 1.023, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"grad_norm": 0.8870755434036255, |
|
"learning_rate": 0.0001326923076923077, |
|
"loss": 0.1308, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"grad_norm": 0.12857934832572937, |
|
"learning_rate": 0.0001294871794871795, |
|
"loss": 0.1047, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"grad_norm": 0.11967829614877701, |
|
"learning_rate": 0.00012628205128205127, |
|
"loss": 0.0523, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"grad_norm": 1.8435248136520386, |
|
"learning_rate": 0.0001230769230769231, |
|
"loss": 0.089, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"grad_norm": 0.07049544900655746, |
|
"learning_rate": 0.00011987179487179487, |
|
"loss": 0.0651, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"grad_norm": 7.795147895812988, |
|
"learning_rate": 0.00011666666666666668, |
|
"loss": 0.0827, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"grad_norm": 0.06726662069559097, |
|
"learning_rate": 0.00011346153846153846, |
|
"loss": 0.1727, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"grad_norm": 4.7732672691345215, |
|
"learning_rate": 0.00011025641025641027, |
|
"loss": 0.0867, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"grad_norm": 0.08257576823234558, |
|
"learning_rate": 0.00010705128205128206, |
|
"loss": 0.0349, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"grad_norm": 0.15855157375335693, |
|
"learning_rate": 0.00010384615384615386, |
|
"loss": 0.0253, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"eval_accuracy": 0.954983922829582, |
|
"eval_loss": 0.15877817571163177, |
|
"eval_runtime": 9.849, |
|
"eval_samples_per_second": 63.153, |
|
"eval_steps_per_second": 7.92, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"grad_norm": 11.21109676361084, |
|
"learning_rate": 0.00010064102564102564, |
|
"loss": 0.0988, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"grad_norm": 0.06449388712644577, |
|
"learning_rate": 9.743589743589744e-05, |
|
"loss": 0.0666, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"grad_norm": 0.988405168056488, |
|
"learning_rate": 9.423076923076924e-05, |
|
"loss": 0.0295, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"grad_norm": 0.06675518304109573, |
|
"learning_rate": 9.102564102564103e-05, |
|
"loss": 0.018, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"grad_norm": 0.08486536890268326, |
|
"learning_rate": 8.782051282051283e-05, |
|
"loss": 0.0714, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"grad_norm": 0.05260853096842766, |
|
"learning_rate": 8.461538461538461e-05, |
|
"loss": 0.0354, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 2.37, |
|
"grad_norm": 0.053938668221235275, |
|
"learning_rate": 8.141025641025641e-05, |
|
"loss": 0.0548, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"grad_norm": 0.06470278650522232, |
|
"learning_rate": 7.820512820512821e-05, |
|
"loss": 0.0162, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 2.5, |
|
"grad_norm": 0.0850602388381958, |
|
"learning_rate": 7.500000000000001e-05, |
|
"loss": 0.033, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"grad_norm": 0.04342366382479668, |
|
"learning_rate": 7.17948717948718e-05, |
|
"loss": 0.0224, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"eval_accuracy": 0.9533762057877814, |
|
"eval_loss": 0.16905710101127625, |
|
"eval_runtime": 9.8491, |
|
"eval_samples_per_second": 63.153, |
|
"eval_steps_per_second": 7.92, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 2.63, |
|
"grad_norm": 0.05912560597062111, |
|
"learning_rate": 6.858974358974359e-05, |
|
"loss": 0.0503, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"grad_norm": 0.04359501227736473, |
|
"learning_rate": 6.538461538461539e-05, |
|
"loss": 0.0537, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 2.76, |
|
"grad_norm": 0.0935799852013588, |
|
"learning_rate": 6.217948717948718e-05, |
|
"loss": 0.0145, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 2.82, |
|
"grad_norm": 0.05057013779878616, |
|
"learning_rate": 5.897435897435898e-05, |
|
"loss": 0.0132, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 2.88, |
|
"grad_norm": 0.4015754461288452, |
|
"learning_rate": 5.576923076923077e-05, |
|
"loss": 0.0382, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 2.95, |
|
"grad_norm": 0.03421848267316818, |
|
"learning_rate": 5.256410256410257e-05, |
|
"loss": 0.0123, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 3.01, |
|
"grad_norm": 0.043205052614212036, |
|
"learning_rate": 4.935897435897436e-05, |
|
"loss": 0.0417, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 3.08, |
|
"grad_norm": 0.03611929342150688, |
|
"learning_rate": 4.615384615384616e-05, |
|
"loss": 0.0125, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 3.14, |
|
"grad_norm": 0.04078483581542969, |
|
"learning_rate": 4.294871794871795e-05, |
|
"loss": 0.0193, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 3.21, |
|
"grad_norm": 0.043482307344675064, |
|
"learning_rate": 3.974358974358974e-05, |
|
"loss": 0.0321, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 3.21, |
|
"eval_accuracy": 0.9565916398713826, |
|
"eval_loss": 0.17511475086212158, |
|
"eval_runtime": 9.8922, |
|
"eval_samples_per_second": 62.878, |
|
"eval_steps_per_second": 7.885, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 3.27, |
|
"grad_norm": 0.03763527050614357, |
|
"learning_rate": 3.653846153846154e-05, |
|
"loss": 0.0118, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 3.33, |
|
"grad_norm": 0.05929262936115265, |
|
"learning_rate": 3.3333333333333335e-05, |
|
"loss": 0.0636, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 3.4, |
|
"grad_norm": 0.039751436561346054, |
|
"learning_rate": 3.012820512820513e-05, |
|
"loss": 0.0105, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 3.46, |
|
"grad_norm": 0.03735564276576042, |
|
"learning_rate": 2.6923076923076923e-05, |
|
"loss": 0.0495, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 3.53, |
|
"grad_norm": 0.051983293145895004, |
|
"learning_rate": 2.3717948717948718e-05, |
|
"loss": 0.011, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 3.59, |
|
"grad_norm": 0.034572675824165344, |
|
"learning_rate": 2.0512820512820512e-05, |
|
"loss": 0.0109, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 3.65, |
|
"grad_norm": 0.04169879108667374, |
|
"learning_rate": 1.730769230769231e-05, |
|
"loss": 0.0108, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 3.72, |
|
"grad_norm": 0.032876156270504, |
|
"learning_rate": 1.4102564102564104e-05, |
|
"loss": 0.0104, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 3.78, |
|
"grad_norm": 0.03522384166717529, |
|
"learning_rate": 1.0897435897435898e-05, |
|
"loss": 0.0109, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 3.85, |
|
"grad_norm": 0.11409874260425568, |
|
"learning_rate": 7.692307692307694e-06, |
|
"loss": 0.0112, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 3.85, |
|
"eval_accuracy": 0.954983922829582, |
|
"eval_loss": 0.18050101399421692, |
|
"eval_runtime": 9.8888, |
|
"eval_samples_per_second": 62.899, |
|
"eval_steps_per_second": 7.888, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 3.91, |
|
"grad_norm": 0.04359051212668419, |
|
"learning_rate": 4.487179487179488e-06, |
|
"loss": 0.0109, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 3.97, |
|
"grad_norm": 0.03071708232164383, |
|
"learning_rate": 1.282051282051282e-06, |
|
"loss": 0.0429, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"step": 624, |
|
"total_flos": 7.703325099767808e+17, |
|
"train_loss": 0.15572628828410345, |
|
"train_runtime": 552.98, |
|
"train_samples_per_second": 17.975, |
|
"train_steps_per_second": 1.128 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 624, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 4, |
|
"save_steps": 100, |
|
"total_flos": 7.703325099767808e+17, |
|
"train_batch_size": 16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|