Augusto777's picture
End of training
2b5835a verified
{
"best_metric": 0.7666666666666667,
"best_model_checkpoint": "swinv2-tiny-patch4-window8-256-dmae-va-U5-42C/checkpoint-232",
"epoch": 37.935483870967744,
"eval_steps": 500,
"global_step": 294,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.9032258064516129,
"eval_accuracy": 0.35,
"eval_loss": 1.3925895690917969,
"eval_runtime": 1.9549,
"eval_samples_per_second": 30.693,
"eval_steps_per_second": 1.023,
"step": 7
},
{
"epoch": 1.2903225806451613,
"grad_norm": 3.2569632530212402,
"learning_rate": 3.3333333333333333e-06,
"loss": 1.4087,
"step": 10
},
{
"epoch": 1.935483870967742,
"eval_accuracy": 0.4166666666666667,
"eval_loss": 1.3364887237548828,
"eval_runtime": 1.315,
"eval_samples_per_second": 45.628,
"eval_steps_per_second": 1.521,
"step": 15
},
{
"epoch": 2.5806451612903225,
"grad_norm": 2.8684847354888916,
"learning_rate": 6.666666666666667e-06,
"loss": 1.3807,
"step": 20
},
{
"epoch": 2.967741935483871,
"eval_accuracy": 0.4166666666666667,
"eval_loss": 1.2813018560409546,
"eval_runtime": 1.3067,
"eval_samples_per_second": 45.917,
"eval_steps_per_second": 1.531,
"step": 23
},
{
"epoch": 3.870967741935484,
"grad_norm": 3.6313109397888184,
"learning_rate": 1e-05,
"loss": 1.35,
"step": 30
},
{
"epoch": 4.0,
"eval_accuracy": 0.4,
"eval_loss": 1.2406634092330933,
"eval_runtime": 1.3069,
"eval_samples_per_second": 45.909,
"eval_steps_per_second": 1.53,
"step": 31
},
{
"epoch": 4.903225806451613,
"eval_accuracy": 0.48333333333333334,
"eval_loss": 1.2116148471832275,
"eval_runtime": 1.3014,
"eval_samples_per_second": 46.103,
"eval_steps_per_second": 1.537,
"step": 38
},
{
"epoch": 5.161290322580645,
"grad_norm": 3.7117483615875244,
"learning_rate": 9.621212121212122e-06,
"loss": 1.2933,
"step": 40
},
{
"epoch": 5.935483870967742,
"eval_accuracy": 0.48333333333333334,
"eval_loss": 1.165272831916809,
"eval_runtime": 1.3337,
"eval_samples_per_second": 44.987,
"eval_steps_per_second": 1.5,
"step": 46
},
{
"epoch": 6.451612903225806,
"grad_norm": 4.500431060791016,
"learning_rate": 9.242424242424244e-06,
"loss": 1.2426,
"step": 50
},
{
"epoch": 6.967741935483871,
"eval_accuracy": 0.5166666666666667,
"eval_loss": 1.115120768547058,
"eval_runtime": 1.3716,
"eval_samples_per_second": 43.745,
"eval_steps_per_second": 1.458,
"step": 54
},
{
"epoch": 7.741935483870968,
"grad_norm": 4.069227695465088,
"learning_rate": 8.863636363636365e-06,
"loss": 1.1771,
"step": 60
},
{
"epoch": 8.0,
"eval_accuracy": 0.6,
"eval_loss": 1.0440610647201538,
"eval_runtime": 1.9614,
"eval_samples_per_second": 30.59,
"eval_steps_per_second": 1.02,
"step": 62
},
{
"epoch": 8.903225806451612,
"eval_accuracy": 0.5666666666666667,
"eval_loss": 0.9990090727806091,
"eval_runtime": 1.6529,
"eval_samples_per_second": 36.299,
"eval_steps_per_second": 1.21,
"step": 69
},
{
"epoch": 9.03225806451613,
"grad_norm": 6.175364971160889,
"learning_rate": 8.484848484848486e-06,
"loss": 1.0983,
"step": 70
},
{
"epoch": 9.935483870967742,
"eval_accuracy": 0.6333333333333333,
"eval_loss": 0.9455800652503967,
"eval_runtime": 1.336,
"eval_samples_per_second": 44.911,
"eval_steps_per_second": 1.497,
"step": 77
},
{
"epoch": 10.32258064516129,
"grad_norm": 5.833083152770996,
"learning_rate": 8.106060606060606e-06,
"loss": 1.0338,
"step": 80
},
{
"epoch": 10.967741935483872,
"eval_accuracy": 0.6833333333333333,
"eval_loss": 0.9160192012786865,
"eval_runtime": 1.3605,
"eval_samples_per_second": 44.101,
"eval_steps_per_second": 1.47,
"step": 85
},
{
"epoch": 11.612903225806452,
"grad_norm": 7.081052303314209,
"learning_rate": 7.727272727272727e-06,
"loss": 0.9665,
"step": 90
},
{
"epoch": 12.0,
"eval_accuracy": 0.6833333333333333,
"eval_loss": 0.8940223455429077,
"eval_runtime": 1.6293,
"eval_samples_per_second": 36.825,
"eval_steps_per_second": 1.228,
"step": 93
},
{
"epoch": 12.903225806451612,
"grad_norm": 10.133336067199707,
"learning_rate": 7.348484848484849e-06,
"loss": 0.9133,
"step": 100
},
{
"epoch": 12.903225806451612,
"eval_accuracy": 0.6,
"eval_loss": 0.8753000497817993,
"eval_runtime": 1.851,
"eval_samples_per_second": 32.415,
"eval_steps_per_second": 1.081,
"step": 100
},
{
"epoch": 13.935483870967742,
"eval_accuracy": 0.6666666666666666,
"eval_loss": 0.8517557978630066,
"eval_runtime": 1.7325,
"eval_samples_per_second": 34.633,
"eval_steps_per_second": 1.154,
"step": 108
},
{
"epoch": 14.193548387096774,
"grad_norm": 7.324928283691406,
"learning_rate": 6.969696969696971e-06,
"loss": 0.8521,
"step": 110
},
{
"epoch": 14.967741935483872,
"eval_accuracy": 0.65,
"eval_loss": 0.8514594435691833,
"eval_runtime": 1.3473,
"eval_samples_per_second": 44.535,
"eval_steps_per_second": 1.484,
"step": 116
},
{
"epoch": 15.483870967741936,
"grad_norm": 8.610685348510742,
"learning_rate": 6.590909090909091e-06,
"loss": 0.8461,
"step": 120
},
{
"epoch": 16.0,
"eval_accuracy": 0.65,
"eval_loss": 0.8406927585601807,
"eval_runtime": 1.3128,
"eval_samples_per_second": 45.703,
"eval_steps_per_second": 1.523,
"step": 124
},
{
"epoch": 16.774193548387096,
"grad_norm": 8.986543655395508,
"learning_rate": 6.212121212121213e-06,
"loss": 0.808,
"step": 130
},
{
"epoch": 16.903225806451612,
"eval_accuracy": 0.65,
"eval_loss": 0.8217867016792297,
"eval_runtime": 1.3566,
"eval_samples_per_second": 44.227,
"eval_steps_per_second": 1.474,
"step": 131
},
{
"epoch": 17.93548387096774,
"eval_accuracy": 0.6833333333333333,
"eval_loss": 0.817019522190094,
"eval_runtime": 1.3417,
"eval_samples_per_second": 44.72,
"eval_steps_per_second": 1.491,
"step": 139
},
{
"epoch": 18.06451612903226,
"grad_norm": 14.52387523651123,
"learning_rate": 5.833333333333334e-06,
"loss": 0.7779,
"step": 140
},
{
"epoch": 18.967741935483872,
"eval_accuracy": 0.7166666666666667,
"eval_loss": 0.7972093224525452,
"eval_runtime": 1.333,
"eval_samples_per_second": 45.011,
"eval_steps_per_second": 1.5,
"step": 147
},
{
"epoch": 19.35483870967742,
"grad_norm": 9.242515563964844,
"learning_rate": 5.4545454545454545e-06,
"loss": 0.758,
"step": 150
},
{
"epoch": 20.0,
"eval_accuracy": 0.7333333333333333,
"eval_loss": 0.7816523313522339,
"eval_runtime": 1.3563,
"eval_samples_per_second": 44.24,
"eval_steps_per_second": 1.475,
"step": 155
},
{
"epoch": 20.64516129032258,
"grad_norm": 10.172256469726562,
"learning_rate": 5.075757575757576e-06,
"loss": 0.7416,
"step": 160
},
{
"epoch": 20.903225806451612,
"eval_accuracy": 0.7166666666666667,
"eval_loss": 0.7677698135375977,
"eval_runtime": 1.3666,
"eval_samples_per_second": 43.906,
"eval_steps_per_second": 1.464,
"step": 162
},
{
"epoch": 21.93548387096774,
"grad_norm": 9.232048988342285,
"learning_rate": 4.696969696969698e-06,
"loss": 0.7344,
"step": 170
},
{
"epoch": 21.93548387096774,
"eval_accuracy": 0.7166666666666667,
"eval_loss": 0.7650207281112671,
"eval_runtime": 1.3874,
"eval_samples_per_second": 43.248,
"eval_steps_per_second": 1.442,
"step": 170
},
{
"epoch": 22.967741935483872,
"eval_accuracy": 0.7333333333333333,
"eval_loss": 0.7427585124969482,
"eval_runtime": 1.3402,
"eval_samples_per_second": 44.768,
"eval_steps_per_second": 1.492,
"step": 178
},
{
"epoch": 23.225806451612904,
"grad_norm": 8.738112449645996,
"learning_rate": 4.3181818181818185e-06,
"loss": 0.7091,
"step": 180
},
{
"epoch": 24.0,
"eval_accuracy": 0.75,
"eval_loss": 0.7280309200286865,
"eval_runtime": 1.3549,
"eval_samples_per_second": 44.284,
"eval_steps_per_second": 1.476,
"step": 186
},
{
"epoch": 24.516129032258064,
"grad_norm": 9.839088439941406,
"learning_rate": 3.93939393939394e-06,
"loss": 0.6876,
"step": 190
},
{
"epoch": 24.903225806451612,
"eval_accuracy": 0.75,
"eval_loss": 0.7235161066055298,
"eval_runtime": 1.346,
"eval_samples_per_second": 44.577,
"eval_steps_per_second": 1.486,
"step": 193
},
{
"epoch": 25.806451612903224,
"grad_norm": 9.253539085388184,
"learning_rate": 3.560606060606061e-06,
"loss": 0.6887,
"step": 200
},
{
"epoch": 25.93548387096774,
"eval_accuracy": 0.75,
"eval_loss": 0.7277832627296448,
"eval_runtime": 1.4422,
"eval_samples_per_second": 41.604,
"eval_steps_per_second": 1.387,
"step": 201
},
{
"epoch": 26.967741935483872,
"eval_accuracy": 0.75,
"eval_loss": 0.7264487147331238,
"eval_runtime": 1.3795,
"eval_samples_per_second": 43.494,
"eval_steps_per_second": 1.45,
"step": 209
},
{
"epoch": 27.096774193548388,
"grad_norm": 9.031746864318848,
"learning_rate": 3.181818181818182e-06,
"loss": 0.6897,
"step": 210
},
{
"epoch": 28.0,
"eval_accuracy": 0.75,
"eval_loss": 0.7228216528892517,
"eval_runtime": 1.5018,
"eval_samples_per_second": 39.952,
"eval_steps_per_second": 1.332,
"step": 217
},
{
"epoch": 28.387096774193548,
"grad_norm": 9.238479614257812,
"learning_rate": 2.803030303030303e-06,
"loss": 0.6637,
"step": 220
},
{
"epoch": 28.903225806451612,
"eval_accuracy": 0.75,
"eval_loss": 0.7162961363792419,
"eval_runtime": 1.8314,
"eval_samples_per_second": 32.762,
"eval_steps_per_second": 1.092,
"step": 224
},
{
"epoch": 29.677419354838708,
"grad_norm": 11.410670280456543,
"learning_rate": 2.4242424242424244e-06,
"loss": 0.6924,
"step": 230
},
{
"epoch": 29.93548387096774,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7072709798812866,
"eval_runtime": 1.7495,
"eval_samples_per_second": 34.295,
"eval_steps_per_second": 1.143,
"step": 232
},
{
"epoch": 30.967741935483872,
"grad_norm": 9.04150676727295,
"learning_rate": 2.0454545454545457e-06,
"loss": 0.6234,
"step": 240
},
{
"epoch": 30.967741935483872,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7057181596755981,
"eval_runtime": 1.7336,
"eval_samples_per_second": 34.609,
"eval_steps_per_second": 1.154,
"step": 240
},
{
"epoch": 32.0,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7090018391609192,
"eval_runtime": 1.9304,
"eval_samples_per_second": 31.081,
"eval_steps_per_second": 1.036,
"step": 248
},
{
"epoch": 32.25806451612903,
"grad_norm": 10.956854820251465,
"learning_rate": 1.6666666666666667e-06,
"loss": 0.6652,
"step": 250
},
{
"epoch": 32.903225806451616,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7051603198051453,
"eval_runtime": 2.0298,
"eval_samples_per_second": 29.559,
"eval_steps_per_second": 0.985,
"step": 255
},
{
"epoch": 33.54838709677419,
"grad_norm": 11.52700424194336,
"learning_rate": 1.287878787878788e-06,
"loss": 0.6343,
"step": 260
},
{
"epoch": 33.935483870967744,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7009490132331848,
"eval_runtime": 1.5868,
"eval_samples_per_second": 37.811,
"eval_steps_per_second": 1.26,
"step": 263
},
{
"epoch": 34.83870967741935,
"grad_norm": 10.979878425598145,
"learning_rate": 9.090909090909091e-07,
"loss": 0.6327,
"step": 270
},
{
"epoch": 34.96774193548387,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7016925811767578,
"eval_runtime": 1.3715,
"eval_samples_per_second": 43.748,
"eval_steps_per_second": 1.458,
"step": 271
},
{
"epoch": 36.0,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7023322582244873,
"eval_runtime": 1.3676,
"eval_samples_per_second": 43.872,
"eval_steps_per_second": 1.462,
"step": 279
},
{
"epoch": 36.12903225806452,
"grad_norm": 8.543764114379883,
"learning_rate": 5.303030303030304e-07,
"loss": 0.6339,
"step": 280
},
{
"epoch": 36.903225806451616,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7027052044868469,
"eval_runtime": 1.3882,
"eval_samples_per_second": 43.221,
"eval_steps_per_second": 1.441,
"step": 286
},
{
"epoch": 37.41935483870968,
"grad_norm": 10.100813865661621,
"learning_rate": 1.5151515151515152e-07,
"loss": 0.6275,
"step": 290
},
{
"epoch": 37.935483870967744,
"eval_accuracy": 0.7666666666666667,
"eval_loss": 0.7030532956123352,
"eval_runtime": 1.3783,
"eval_samples_per_second": 43.53,
"eval_steps_per_second": 1.451,
"step": 294
},
{
"epoch": 37.935483870967744,
"step": 294,
"total_flos": 1.2027310550050406e+18,
"train_loss": 0.8700437156521544,
"train_runtime": 913.4545,
"train_samples_per_second": 44.784,
"train_steps_per_second": 0.322
}
],
"logging_steps": 10,
"max_steps": 294,
"num_input_tokens_seen": 0,
"num_train_epochs": 42,
"save_steps": 500,
"total_flos": 1.2027310550050406e+18,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}