beit-base-patch16-224 / trainer_state.json
HorcruxNo13's picture
beit-mass-secondstep
e851381 verified
raw
history blame
15.3 kB
{
"best_metric": 0.8627450980392157,
"best_model_checkpoint": "beit-base-patch16-224/checkpoint-47",
"epoch": 36.0,
"eval_steps": 500,
"global_step": 90,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.8,
"eval_accuracy": 0.5882352941176471,
"eval_f1_score": 0.554074074074074,
"eval_loss": 0.6992508172988892,
"eval_precision": 0.5390243902439025,
"eval_recall": 0.5882352941176471,
"eval_runtime": 0.8655,
"eval_samples_per_second": 58.922,
"eval_steps_per_second": 2.311,
"step": 2
},
{
"epoch": 2.0,
"eval_accuracy": 0.6862745098039216,
"eval_f1_score": 0.6032520325203252,
"eval_loss": 0.5970537662506104,
"eval_precision": 0.6805555555555555,
"eval_recall": 0.6862745098039216,
"eval_runtime": 0.8959,
"eval_samples_per_second": 56.925,
"eval_steps_per_second": 2.232,
"step": 5
},
{
"epoch": 2.8,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.800595238095238,
"eval_loss": 0.5305531024932861,
"eval_precision": 0.7999999999999999,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9046,
"eval_samples_per_second": 56.379,
"eval_steps_per_second": 2.211,
"step": 7
},
{
"epoch": 4.0,
"eval_accuracy": 0.7254901960784313,
"eval_f1_score": 0.6858974358974359,
"eval_loss": 0.48283636569976807,
"eval_precision": 0.722943722943723,
"eval_recall": 0.7254901960784313,
"eval_runtime": 0.9029,
"eval_samples_per_second": 56.482,
"eval_steps_per_second": 2.215,
"step": 10
},
{
"epoch": 4.8,
"eval_accuracy": 0.7843137254901961,
"eval_f1_score": 0.7784340451310011,
"eval_loss": 0.3811856508255005,
"eval_precision": 0.7786357786357786,
"eval_recall": 0.7843137254901961,
"eval_runtime": 0.92,
"eval_samples_per_second": 55.436,
"eval_steps_per_second": 2.174,
"step": 12
},
{
"epoch": 6.0,
"grad_norm": 4.578022480010986,
"learning_rate": 4.62962962962963e-05,
"loss": 0.5413,
"step": 15
},
{
"epoch": 6.0,
"eval_accuracy": 0.7450980392156863,
"eval_f1_score": 0.7141125541125543,
"eval_loss": 0.5268120765686035,
"eval_precision": 0.7461240310077519,
"eval_recall": 0.7450980392156863,
"eval_runtime": 0.9096,
"eval_samples_per_second": 56.066,
"eval_steps_per_second": 2.199,
"step": 15
},
{
"epoch": 6.8,
"eval_accuracy": 0.7450980392156863,
"eval_f1_score": 0.7502256608639587,
"eval_loss": 0.5349109768867493,
"eval_precision": 0.8555555555555555,
"eval_recall": 0.7450980392156863,
"eval_runtime": 0.9137,
"eval_samples_per_second": 55.818,
"eval_steps_per_second": 2.189,
"step": 17
},
{
"epoch": 8.0,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.7756410256410257,
"eval_loss": 0.4119790494441986,
"eval_precision": 0.8484848484848485,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9237,
"eval_samples_per_second": 55.215,
"eval_steps_per_second": 2.165,
"step": 20
},
{
"epoch": 8.8,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.7962962962962962,
"eval_loss": 0.3156317472457886,
"eval_precision": 0.8002699055330634,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9335,
"eval_samples_per_second": 54.63,
"eval_steps_per_second": 2.142,
"step": 22
},
{
"epoch": 10.0,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.7908622908622909,
"eval_loss": 0.3216821253299713,
"eval_precision": 0.806060606060606,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9256,
"eval_samples_per_second": 55.1,
"eval_steps_per_second": 2.161,
"step": 25
},
{
"epoch": 10.8,
"eval_accuracy": 0.7843137254901961,
"eval_f1_score": 0.7664197530864199,
"eval_loss": 0.5160595774650574,
"eval_precision": 0.7869918699186993,
"eval_recall": 0.7843137254901961,
"eval_runtime": 0.9267,
"eval_samples_per_second": 55.031,
"eval_steps_per_second": 2.158,
"step": 27
},
{
"epoch": 12.0,
"grad_norm": 3.5482540130615234,
"learning_rate": 3.7037037037037037e-05,
"loss": 0.0919,
"step": 30
},
{
"epoch": 12.0,
"eval_accuracy": 0.8431372549019608,
"eval_f1_score": 0.845117845117845,
"eval_loss": 0.36771491169929504,
"eval_precision": 0.849780701754386,
"eval_recall": 0.8431372549019608,
"eval_runtime": 0.942,
"eval_samples_per_second": 54.142,
"eval_steps_per_second": 2.123,
"step": 30
},
{
"epoch": 12.8,
"eval_accuracy": 0.8431372549019608,
"eval_f1_score": 0.8404761904761906,
"eval_loss": 0.46310773491859436,
"eval_precision": 0.8407407407407408,
"eval_recall": 0.8431372549019608,
"eval_runtime": 0.9403,
"eval_samples_per_second": 54.24,
"eval_steps_per_second": 2.127,
"step": 32
},
{
"epoch": 14.0,
"eval_accuracy": 0.8235294117647058,
"eval_f1_score": 0.8221343873517787,
"eval_loss": 0.5000560879707336,
"eval_precision": 0.8214285714285714,
"eval_recall": 0.8235294117647058,
"eval_runtime": 0.9615,
"eval_samples_per_second": 53.039,
"eval_steps_per_second": 2.08,
"step": 35
},
{
"epoch": 14.8,
"eval_accuracy": 0.8431372549019608,
"eval_f1_score": 0.8431372549019608,
"eval_loss": 0.4489041268825531,
"eval_precision": 0.8431372549019608,
"eval_recall": 0.8431372549019608,
"eval_runtime": 0.9337,
"eval_samples_per_second": 54.621,
"eval_steps_per_second": 2.142,
"step": 37
},
{
"epoch": 16.0,
"eval_accuracy": 0.7843137254901961,
"eval_f1_score": 0.7731065973862385,
"eval_loss": 0.5892294049263,
"eval_precision": 0.7799145299145298,
"eval_recall": 0.7843137254901961,
"eval_runtime": 1.0872,
"eval_samples_per_second": 46.909,
"eval_steps_per_second": 1.84,
"step": 40
},
{
"epoch": 16.8,
"eval_accuracy": 0.7843137254901961,
"eval_f1_score": 0.7731065973862385,
"eval_loss": 0.6578794717788696,
"eval_precision": 0.7799145299145298,
"eval_recall": 0.7843137254901961,
"eval_runtime": 0.9215,
"eval_samples_per_second": 55.345,
"eval_steps_per_second": 2.17,
"step": 42
},
{
"epoch": 18.0,
"grad_norm": 3.25277042388916,
"learning_rate": 2.777777777777778e-05,
"loss": 0.006,
"step": 45
},
{
"epoch": 18.0,
"eval_accuracy": 0.7843137254901961,
"eval_f1_score": 0.7731065973862385,
"eval_loss": 0.703818678855896,
"eval_precision": 0.7799145299145298,
"eval_recall": 0.7843137254901961,
"eval_runtime": 1.0077,
"eval_samples_per_second": 50.61,
"eval_steps_per_second": 1.985,
"step": 45
},
{
"epoch": 18.8,
"eval_accuracy": 0.8627450980392157,
"eval_f1_score": 0.865142065142065,
"eval_loss": 0.5864243507385254,
"eval_precision": 0.8736559139784946,
"eval_recall": 0.8627450980392157,
"eval_runtime": 0.9259,
"eval_samples_per_second": 55.08,
"eval_steps_per_second": 2.16,
"step": 47
},
{
"epoch": 20.0,
"eval_accuracy": 0.8627450980392157,
"eval_f1_score": 0.865142065142065,
"eval_loss": 0.5488199591636658,
"eval_precision": 0.8736559139784946,
"eval_recall": 0.8627450980392157,
"eval_runtime": 0.9318,
"eval_samples_per_second": 54.735,
"eval_steps_per_second": 2.146,
"step": 50
},
{
"epoch": 20.8,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.7962962962962962,
"eval_loss": 0.6650967597961426,
"eval_precision": 0.8002699055330634,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9328,
"eval_samples_per_second": 54.677,
"eval_steps_per_second": 2.144,
"step": 52
},
{
"epoch": 22.0,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.800595238095238,
"eval_loss": 0.6264931559562683,
"eval_precision": 0.7999999999999999,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9317,
"eval_samples_per_second": 54.741,
"eval_steps_per_second": 2.147,
"step": 55
},
{
"epoch": 22.8,
"eval_accuracy": 0.8627450980392157,
"eval_f1_score": 0.8636815920398009,
"eval_loss": 0.5228903889656067,
"eval_precision": 0.8653198653198653,
"eval_recall": 0.8627450980392157,
"eval_runtime": 0.9295,
"eval_samples_per_second": 54.868,
"eval_steps_per_second": 2.152,
"step": 57
},
{
"epoch": 24.0,
"grad_norm": 0.0452270582318306,
"learning_rate": 1.8518518518518518e-05,
"loss": 0.0048,
"step": 60
},
{
"epoch": 24.0,
"eval_accuracy": 0.8627450980392157,
"eval_f1_score": 0.8636815920398009,
"eval_loss": 0.542142927646637,
"eval_precision": 0.8653198653198653,
"eval_recall": 0.8627450980392157,
"eval_runtime": 0.9409,
"eval_samples_per_second": 54.206,
"eval_steps_per_second": 2.126,
"step": 60
},
{
"epoch": 24.8,
"eval_accuracy": 0.8235294117647058,
"eval_f1_score": 0.8187187641980918,
"eval_loss": 0.6334545016288757,
"eval_precision": 0.8204633204633205,
"eval_recall": 0.8235294117647058,
"eval_runtime": 0.9368,
"eval_samples_per_second": 54.438,
"eval_steps_per_second": 2.135,
"step": 62
},
{
"epoch": 26.0,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.7840755735492576,
"eval_loss": 1.0379055738449097,
"eval_precision": 0.82010582010582,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.927,
"eval_samples_per_second": 55.015,
"eval_steps_per_second": 2.157,
"step": 65
},
{
"epoch": 26.8,
"eval_accuracy": 0.8235294117647058,
"eval_f1_score": 0.808888888888889,
"eval_loss": 0.9758451581001282,
"eval_precision": 0.8365853658536586,
"eval_recall": 0.8235294117647058,
"eval_runtime": 0.927,
"eval_samples_per_second": 55.017,
"eval_steps_per_second": 2.158,
"step": 67
},
{
"epoch": 28.0,
"eval_accuracy": 0.8235294117647058,
"eval_f1_score": 0.8187187641980918,
"eval_loss": 0.6116669774055481,
"eval_precision": 0.8204633204633205,
"eval_recall": 0.8235294117647058,
"eval_runtime": 0.9261,
"eval_samples_per_second": 55.07,
"eval_steps_per_second": 2.16,
"step": 70
},
{
"epoch": 28.8,
"eval_accuracy": 0.8627450980392157,
"eval_f1_score": 0.8616600790513834,
"eval_loss": 0.540273904800415,
"eval_precision": 0.8613095238095237,
"eval_recall": 0.8627450980392157,
"eval_runtime": 0.9247,
"eval_samples_per_second": 55.15,
"eval_steps_per_second": 2.163,
"step": 72
},
{
"epoch": 30.0,
"grad_norm": 0.026938632130622864,
"learning_rate": 9.259259259259259e-06,
"loss": 0.0063,
"step": 75
},
{
"epoch": 30.0,
"eval_accuracy": 0.8431372549019608,
"eval_f1_score": 0.8404761904761906,
"eval_loss": 0.6468568444252014,
"eval_precision": 0.8407407407407408,
"eval_recall": 0.8431372549019608,
"eval_runtime": 0.9235,
"eval_samples_per_second": 55.223,
"eval_steps_per_second": 2.166,
"step": 75
},
{
"epoch": 30.8,
"eval_accuracy": 0.8235294117647058,
"eval_f1_score": 0.8187187641980918,
"eval_loss": 0.7013790607452393,
"eval_precision": 0.8204633204633205,
"eval_recall": 0.8235294117647058,
"eval_runtime": 1.1388,
"eval_samples_per_second": 44.785,
"eval_steps_per_second": 1.756,
"step": 77
},
{
"epoch": 32.0,
"eval_accuracy": 0.8235294117647058,
"eval_f1_score": 0.8187187641980918,
"eval_loss": 0.7514360547065735,
"eval_precision": 0.8204633204633205,
"eval_recall": 0.8235294117647058,
"eval_runtime": 0.9424,
"eval_samples_per_second": 54.118,
"eval_steps_per_second": 2.122,
"step": 80
},
{
"epoch": 32.8,
"eval_accuracy": 0.8235294117647058,
"eval_f1_score": 0.8143599433160132,
"eval_loss": 0.7771488428115845,
"eval_precision": 0.8247863247863249,
"eval_recall": 0.8235294117647058,
"eval_runtime": 0.9338,
"eval_samples_per_second": 54.616,
"eval_steps_per_second": 2.142,
"step": 82
},
{
"epoch": 34.0,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.7962962962962962,
"eval_loss": 0.7598747611045837,
"eval_precision": 0.8002699055330634,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9331,
"eval_samples_per_second": 54.655,
"eval_steps_per_second": 2.143,
"step": 85
},
{
"epoch": 34.8,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.7962962962962962,
"eval_loss": 0.7554459571838379,
"eval_precision": 0.8002699055330634,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9307,
"eval_samples_per_second": 54.796,
"eval_steps_per_second": 2.149,
"step": 87
},
{
"epoch": 36.0,
"grad_norm": 0.014645076356828213,
"learning_rate": 0.0,
"loss": 0.0045,
"step": 90
},
{
"epoch": 36.0,
"eval_accuracy": 0.803921568627451,
"eval_f1_score": 0.7962962962962962,
"eval_loss": 0.7308478951454163,
"eval_precision": 0.8002699055330634,
"eval_recall": 0.803921568627451,
"eval_runtime": 0.9231,
"eval_samples_per_second": 55.246,
"eval_steps_per_second": 2.167,
"step": 90
},
{
"epoch": 36.0,
"step": 90,
"total_flos": 1.2659877490145034e+18,
"train_loss": 0.10912525819407569,
"train_runtime": 949.2365,
"train_samples_per_second": 21.523,
"train_steps_per_second": 0.095
},
{
"epoch": 36.0,
"eval_accuracy": 0.8267716535433071,
"eval_f1_score": 0.8283048858023182,
"eval_loss": 0.8527529239654541,
"eval_precision": 0.8302904444636728,
"eval_recall": 0.8267716535433071,
"eval_runtime": 2.5545,
"eval_samples_per_second": 49.716,
"eval_steps_per_second": 1.174,
"step": 90
}
],
"logging_steps": 15,
"max_steps": 90,
"num_input_tokens_seen": 0,
"num_train_epochs": 45,
"save_steps": 500,
"total_flos": 1.2659877490145034e+18,
"train_batch_size": 48,
"trial_name": null,
"trial_params": null
}