vit-base-renovation / trainer_state.json
rshrott's picture
🍻 cheers
455f9b9
raw
history blame
11.1 kB
{
"best_metric": 0.7634429931640625,
"best_model_checkpoint": "./vit-base-renovation/checkpoint-175",
"epoch": 4.0,
"global_step": 496,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.08,
"learning_rate": 0.00019596774193548388,
"loss": 0.9826,
"step": 10
},
{
"epoch": 0.16,
"learning_rate": 0.00019193548387096775,
"loss": 0.9741,
"step": 20
},
{
"epoch": 0.2,
"eval_accuracy": 0.4818181818181818,
"eval_loss": 0.9574973583221436,
"eval_runtime": 7.5527,
"eval_samples_per_second": 29.129,
"eval_steps_per_second": 3.707,
"step": 25
},
{
"epoch": 0.24,
"learning_rate": 0.00018790322580645164,
"loss": 1.0046,
"step": 30
},
{
"epoch": 0.32,
"learning_rate": 0.00018387096774193548,
"loss": 0.9326,
"step": 40
},
{
"epoch": 0.4,
"learning_rate": 0.00017983870967741935,
"loss": 0.9827,
"step": 50
},
{
"epoch": 0.4,
"eval_accuracy": 0.5181818181818182,
"eval_loss": 0.9343744516372681,
"eval_runtime": 7.6495,
"eval_samples_per_second": 28.76,
"eval_steps_per_second": 3.66,
"step": 50
},
{
"epoch": 0.48,
"learning_rate": 0.00017580645161290325,
"loss": 0.91,
"step": 60
},
{
"epoch": 0.56,
"learning_rate": 0.00017177419354838711,
"loss": 0.8578,
"step": 70
},
{
"epoch": 0.6,
"eval_accuracy": 0.6181818181818182,
"eval_loss": 0.8342534899711609,
"eval_runtime": 7.5132,
"eval_samples_per_second": 29.282,
"eval_steps_per_second": 3.727,
"step": 75
},
{
"epoch": 0.65,
"learning_rate": 0.00016774193548387098,
"loss": 0.9354,
"step": 80
},
{
"epoch": 0.73,
"learning_rate": 0.00016370967741935485,
"loss": 0.8993,
"step": 90
},
{
"epoch": 0.81,
"learning_rate": 0.00015967741935483872,
"loss": 0.9373,
"step": 100
},
{
"epoch": 0.81,
"eval_accuracy": 0.5909090909090909,
"eval_loss": 0.8896223306655884,
"eval_runtime": 8.4017,
"eval_samples_per_second": 26.185,
"eval_steps_per_second": 3.333,
"step": 100
},
{
"epoch": 0.89,
"learning_rate": 0.0001556451612903226,
"loss": 0.8373,
"step": 110
},
{
"epoch": 0.97,
"learning_rate": 0.00015161290322580646,
"loss": 0.7462,
"step": 120
},
{
"epoch": 1.01,
"eval_accuracy": 0.6363636363636364,
"eval_loss": 0.7968847155570984,
"eval_runtime": 7.5692,
"eval_samples_per_second": 29.065,
"eval_steps_per_second": 3.699,
"step": 125
},
{
"epoch": 1.05,
"learning_rate": 0.00014758064516129032,
"loss": 0.8443,
"step": 130
},
{
"epoch": 1.13,
"learning_rate": 0.00014354838709677422,
"loss": 0.5794,
"step": 140
},
{
"epoch": 1.21,
"learning_rate": 0.0001395161290322581,
"loss": 0.6953,
"step": 150
},
{
"epoch": 1.21,
"eval_accuracy": 0.6363636363636364,
"eval_loss": 0.8157215714454651,
"eval_runtime": 7.5224,
"eval_samples_per_second": 29.246,
"eval_steps_per_second": 3.722,
"step": 150
},
{
"epoch": 1.29,
"learning_rate": 0.00013548387096774193,
"loss": 0.6258,
"step": 160
},
{
"epoch": 1.37,
"learning_rate": 0.0001314516129032258,
"loss": 0.5461,
"step": 170
},
{
"epoch": 1.41,
"eval_accuracy": 0.6772727272727272,
"eval_loss": 0.7634429931640625,
"eval_runtime": 7.4134,
"eval_samples_per_second": 29.676,
"eval_steps_per_second": 3.777,
"step": 175
},
{
"epoch": 1.45,
"learning_rate": 0.0001274193548387097,
"loss": 0.6182,
"step": 180
},
{
"epoch": 1.53,
"learning_rate": 0.00012338709677419356,
"loss": 0.6352,
"step": 190
},
{
"epoch": 1.61,
"learning_rate": 0.00011935483870967743,
"loss": 0.6445,
"step": 200
},
{
"epoch": 1.61,
"eval_accuracy": 0.6545454545454545,
"eval_loss": 0.7743316888809204,
"eval_runtime": 7.6503,
"eval_samples_per_second": 28.757,
"eval_steps_per_second": 3.66,
"step": 200
},
{
"epoch": 1.69,
"learning_rate": 0.00011532258064516131,
"loss": 0.5757,
"step": 210
},
{
"epoch": 1.77,
"learning_rate": 0.00011129032258064515,
"loss": 0.5437,
"step": 220
},
{
"epoch": 1.81,
"eval_accuracy": 0.65,
"eval_loss": 0.7716971635818481,
"eval_runtime": 7.5883,
"eval_samples_per_second": 28.992,
"eval_steps_per_second": 3.69,
"step": 225
},
{
"epoch": 1.85,
"learning_rate": 0.00010725806451612903,
"loss": 0.4878,
"step": 230
},
{
"epoch": 1.94,
"learning_rate": 0.0001032258064516129,
"loss": 0.5527,
"step": 240
},
{
"epoch": 2.02,
"learning_rate": 9.919354838709678e-05,
"loss": 0.5911,
"step": 250
},
{
"epoch": 2.02,
"eval_accuracy": 0.6363636363636364,
"eval_loss": 0.8338578939437866,
"eval_runtime": 8.4178,
"eval_samples_per_second": 26.135,
"eval_steps_per_second": 3.326,
"step": 250
},
{
"epoch": 2.1,
"learning_rate": 9.516129032258065e-05,
"loss": 0.3191,
"step": 260
},
{
"epoch": 2.18,
"learning_rate": 9.112903225806452e-05,
"loss": 0.2483,
"step": 270
},
{
"epoch": 2.22,
"eval_accuracy": 0.6318181818181818,
"eval_loss": 0.8596317768096924,
"eval_runtime": 7.3913,
"eval_samples_per_second": 29.765,
"eval_steps_per_second": 3.788,
"step": 275
},
{
"epoch": 2.26,
"learning_rate": 8.709677419354839e-05,
"loss": 0.3026,
"step": 280
},
{
"epoch": 2.34,
"learning_rate": 8.306451612903227e-05,
"loss": 0.229,
"step": 290
},
{
"epoch": 2.42,
"learning_rate": 7.903225806451613e-05,
"loss": 0.378,
"step": 300
},
{
"epoch": 2.42,
"eval_accuracy": 0.6181818181818182,
"eval_loss": 0.9896994829177856,
"eval_runtime": 7.5826,
"eval_samples_per_second": 29.014,
"eval_steps_per_second": 3.693,
"step": 300
},
{
"epoch": 2.5,
"learning_rate": 7.500000000000001e-05,
"loss": 0.2432,
"step": 310
},
{
"epoch": 2.58,
"learning_rate": 7.096774193548388e-05,
"loss": 0.2742,
"step": 320
},
{
"epoch": 2.62,
"eval_accuracy": 0.6909090909090909,
"eval_loss": 0.8965399861335754,
"eval_runtime": 7.3219,
"eval_samples_per_second": 30.047,
"eval_steps_per_second": 3.824,
"step": 325
},
{
"epoch": 2.66,
"learning_rate": 6.693548387096774e-05,
"loss": 0.1942,
"step": 330
},
{
"epoch": 2.74,
"learning_rate": 6.290322580645161e-05,
"loss": 0.179,
"step": 340
},
{
"epoch": 2.82,
"learning_rate": 5.887096774193549e-05,
"loss": 0.1898,
"step": 350
},
{
"epoch": 2.82,
"eval_accuracy": 0.6681818181818182,
"eval_loss": 1.026204228401184,
"eval_runtime": 7.7003,
"eval_samples_per_second": 28.57,
"eval_steps_per_second": 3.636,
"step": 350
},
{
"epoch": 2.9,
"learning_rate": 5.4838709677419355e-05,
"loss": 0.3122,
"step": 360
},
{
"epoch": 2.98,
"learning_rate": 5.080645161290323e-05,
"loss": 0.2116,
"step": 370
},
{
"epoch": 3.02,
"eval_accuracy": 0.6409090909090909,
"eval_loss": 1.1058470010757446,
"eval_runtime": 7.4881,
"eval_samples_per_second": 29.38,
"eval_steps_per_second": 3.739,
"step": 375
},
{
"epoch": 3.06,
"learning_rate": 4.67741935483871e-05,
"loss": 0.2225,
"step": 380
},
{
"epoch": 3.15,
"learning_rate": 4.2741935483870973e-05,
"loss": 0.076,
"step": 390
},
{
"epoch": 3.23,
"learning_rate": 3.870967741935484e-05,
"loss": 0.0702,
"step": 400
},
{
"epoch": 3.23,
"eval_accuracy": 0.6545454545454545,
"eval_loss": 1.0472766160964966,
"eval_runtime": 7.6075,
"eval_samples_per_second": 28.919,
"eval_steps_per_second": 3.681,
"step": 400
},
{
"epoch": 3.31,
"learning_rate": 3.467741935483872e-05,
"loss": 0.0637,
"step": 410
},
{
"epoch": 3.39,
"learning_rate": 3.0645161290322585e-05,
"loss": 0.0566,
"step": 420
},
{
"epoch": 3.43,
"eval_accuracy": 0.6681818181818182,
"eval_loss": 1.0961891412734985,
"eval_runtime": 8.2248,
"eval_samples_per_second": 26.748,
"eval_steps_per_second": 3.404,
"step": 425
},
{
"epoch": 3.47,
"learning_rate": 2.661290322580645e-05,
"loss": 0.0421,
"step": 430
},
{
"epoch": 3.55,
"learning_rate": 2.258064516129032e-05,
"loss": 0.1011,
"step": 440
},
{
"epoch": 3.63,
"learning_rate": 1.8951612903225807e-05,
"loss": 0.0775,
"step": 450
},
{
"epoch": 3.63,
"eval_accuracy": 0.65,
"eval_loss": 1.1501661539077759,
"eval_runtime": 7.3146,
"eval_samples_per_second": 30.077,
"eval_steps_per_second": 3.828,
"step": 450
},
{
"epoch": 3.71,
"learning_rate": 1.4919354838709679e-05,
"loss": 0.0519,
"step": 460
},
{
"epoch": 3.79,
"learning_rate": 1.0887096774193549e-05,
"loss": 0.0485,
"step": 470
},
{
"epoch": 3.83,
"eval_accuracy": 0.6454545454545455,
"eval_loss": 1.1838457584381104,
"eval_runtime": 7.191,
"eval_samples_per_second": 30.594,
"eval_steps_per_second": 3.894,
"step": 475
},
{
"epoch": 3.87,
"learning_rate": 6.854838709677419e-06,
"loss": 0.0686,
"step": 480
},
{
"epoch": 3.95,
"learning_rate": 2.82258064516129e-06,
"loss": 0.0616,
"step": 490
},
{
"epoch": 4.0,
"step": 496,
"total_flos": 6.137420584742093e+17,
"train_loss": 0.4632607804190728,
"train_runtime": 749.353,
"train_samples_per_second": 10.569,
"train_steps_per_second": 0.662
}
],
"max_steps": 496,
"num_train_epochs": 4,
"total_flos": 6.137420584742093e+17,
"trial_name": null,
"trial_params": null
}