matthieulel's picture
End of training
ae16f21 verified
{
"best_metric": 0.8370913190529876,
"best_model_checkpoint": "vit-large-patch32-384-finetuned-galaxy10-decals/checkpoint-843",
"epoch": 29.76,
"eval_steps": 500,
"global_step": 930,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.32,
"learning_rate": 1.0752688172043012e-05,
"loss": 2.2838,
"step": 10
},
{
"epoch": 0.64,
"learning_rate": 2.1505376344086024e-05,
"loss": 1.8298,
"step": 20
},
{
"epoch": 0.96,
"learning_rate": 3.2258064516129034e-05,
"loss": 1.3342,
"step": 30
},
{
"epoch": 0.99,
"eval_accuracy": 0.6313416009019166,
"eval_f1": 0.6051741872345295,
"eval_loss": 1.0491372346878052,
"eval_precision": 0.6077378601373529,
"eval_recall": 0.6313416009019166,
"eval_runtime": 21.763,
"eval_samples_per_second": 81.515,
"eval_steps_per_second": 0.643,
"step": 31
},
{
"epoch": 1.28,
"learning_rate": 4.301075268817205e-05,
"loss": 1.0557,
"step": 40
},
{
"epoch": 1.6,
"learning_rate": 5.3763440860215054e-05,
"loss": 0.8471,
"step": 50
},
{
"epoch": 1.92,
"learning_rate": 6.451612903225807e-05,
"loss": 0.7979,
"step": 60
},
{
"epoch": 1.98,
"eval_accuracy": 0.7671927846674182,
"eval_f1": 0.7652430186918779,
"eval_loss": 0.6901012659072876,
"eval_precision": 0.7716921452381017,
"eval_recall": 0.7671927846674182,
"eval_runtime": 21.1847,
"eval_samples_per_second": 83.74,
"eval_steps_per_second": 0.661,
"step": 62
},
{
"epoch": 2.24,
"learning_rate": 7.526881720430108e-05,
"loss": 0.741,
"step": 70
},
{
"epoch": 2.56,
"learning_rate": 8.60215053763441e-05,
"loss": 0.6928,
"step": 80
},
{
"epoch": 2.88,
"learning_rate": 9.677419354838711e-05,
"loss": 0.7197,
"step": 90
},
{
"epoch": 2.98,
"eval_accuracy": 0.7784667418263811,
"eval_f1": 0.7704940588846539,
"eval_loss": 0.6200318336486816,
"eval_precision": 0.7715960207478135,
"eval_recall": 0.7784667418263811,
"eval_runtime": 21.4187,
"eval_samples_per_second": 82.825,
"eval_steps_per_second": 0.654,
"step": 93
},
{
"epoch": 3.2,
"learning_rate": 9.916367980884111e-05,
"loss": 0.691,
"step": 100
},
{
"epoch": 3.52,
"learning_rate": 9.79689366786141e-05,
"loss": 0.6486,
"step": 110
},
{
"epoch": 3.84,
"learning_rate": 9.677419354838711e-05,
"loss": 0.6321,
"step": 120
},
{
"epoch": 4.0,
"eval_accuracy": 0.8060879368658399,
"eval_f1": 0.7957042518679185,
"eval_loss": 0.5692886710166931,
"eval_precision": 0.8035246330156649,
"eval_recall": 0.8060879368658399,
"eval_runtime": 21.6457,
"eval_samples_per_second": 81.956,
"eval_steps_per_second": 0.647,
"step": 125
},
{
"epoch": 4.16,
"learning_rate": 9.55794504181601e-05,
"loss": 0.5927,
"step": 130
},
{
"epoch": 4.48,
"learning_rate": 9.438470728793309e-05,
"loss": 0.5869,
"step": 140
},
{
"epoch": 4.8,
"learning_rate": 9.31899641577061e-05,
"loss": 0.5768,
"step": 150
},
{
"epoch": 4.99,
"eval_accuracy": 0.8111612175873731,
"eval_f1": 0.8134180707239236,
"eval_loss": 0.5500875115394592,
"eval_precision": 0.8213061319477135,
"eval_recall": 0.8111612175873731,
"eval_runtime": 21.6943,
"eval_samples_per_second": 81.773,
"eval_steps_per_second": 0.645,
"step": 156
},
{
"epoch": 5.12,
"learning_rate": 9.199522102747909e-05,
"loss": 0.5286,
"step": 160
},
{
"epoch": 5.44,
"learning_rate": 9.080047789725208e-05,
"loss": 0.5513,
"step": 170
},
{
"epoch": 5.76,
"learning_rate": 8.960573476702509e-05,
"loss": 0.5173,
"step": 180
},
{
"epoch": 5.98,
"eval_accuracy": 0.8213077790304397,
"eval_f1": 0.8201874264260287,
"eval_loss": 0.5164684057235718,
"eval_precision": 0.8305619852770596,
"eval_recall": 0.8213077790304397,
"eval_runtime": 21.6677,
"eval_samples_per_second": 81.873,
"eval_steps_per_second": 0.646,
"step": 187
},
{
"epoch": 6.08,
"learning_rate": 8.84109916367981e-05,
"loss": 0.5143,
"step": 190
},
{
"epoch": 6.4,
"learning_rate": 8.72162485065711e-05,
"loss": 0.498,
"step": 200
},
{
"epoch": 6.72,
"learning_rate": 8.60215053763441e-05,
"loss": 0.4781,
"step": 210
},
{
"epoch": 6.98,
"eval_accuracy": 0.8105975197294251,
"eval_f1": 0.8089843496723177,
"eval_loss": 0.5220049023628235,
"eval_precision": 0.8161270722410587,
"eval_recall": 0.8105975197294251,
"eval_runtime": 21.7247,
"eval_samples_per_second": 81.658,
"eval_steps_per_second": 0.644,
"step": 218
},
{
"epoch": 7.04,
"learning_rate": 8.482676224611709e-05,
"loss": 0.4999,
"step": 220
},
{
"epoch": 7.36,
"learning_rate": 8.363201911589009e-05,
"loss": 0.4604,
"step": 230
},
{
"epoch": 7.68,
"learning_rate": 8.243727598566309e-05,
"loss": 0.4807,
"step": 240
},
{
"epoch": 8.0,
"learning_rate": 8.124253285543608e-05,
"loss": 0.451,
"step": 250
},
{
"epoch": 8.0,
"eval_accuracy": 0.818489289740699,
"eval_f1": 0.8152695783103332,
"eval_loss": 0.5133233666419983,
"eval_precision": 0.8227391031387269,
"eval_recall": 0.818489289740699,
"eval_runtime": 21.9192,
"eval_samples_per_second": 80.934,
"eval_steps_per_second": 0.639,
"step": 250
},
{
"epoch": 8.32,
"learning_rate": 8.004778972520908e-05,
"loss": 0.4252,
"step": 260
},
{
"epoch": 8.64,
"learning_rate": 7.885304659498209e-05,
"loss": 0.4359,
"step": 270
},
{
"epoch": 8.96,
"learning_rate": 7.765830346475508e-05,
"loss": 0.4373,
"step": 280
},
{
"epoch": 8.99,
"eval_accuracy": 0.8303269447576099,
"eval_f1": 0.8288008582358363,
"eval_loss": 0.511769711971283,
"eval_precision": 0.8325406030306683,
"eval_recall": 0.8303269447576099,
"eval_runtime": 21.4401,
"eval_samples_per_second": 82.742,
"eval_steps_per_second": 0.653,
"step": 281
},
{
"epoch": 9.28,
"learning_rate": 7.646356033452809e-05,
"loss": 0.4023,
"step": 290
},
{
"epoch": 9.6,
"learning_rate": 7.526881720430108e-05,
"loss": 0.396,
"step": 300
},
{
"epoch": 9.92,
"learning_rate": 7.407407407407407e-05,
"loss": 0.3826,
"step": 310
},
{
"epoch": 9.98,
"eval_accuracy": 0.8258173618940248,
"eval_f1": 0.8243103846150182,
"eval_loss": 0.5279704928398132,
"eval_precision": 0.8268898344898333,
"eval_recall": 0.8258173618940248,
"eval_runtime": 21.4879,
"eval_samples_per_second": 82.558,
"eval_steps_per_second": 0.652,
"step": 312
},
{
"epoch": 10.24,
"learning_rate": 7.287933094384708e-05,
"loss": 0.363,
"step": 320
},
{
"epoch": 10.56,
"learning_rate": 7.168458781362007e-05,
"loss": 0.376,
"step": 330
},
{
"epoch": 10.88,
"learning_rate": 7.048984468339306e-05,
"loss": 0.378,
"step": 340
},
{
"epoch": 10.98,
"eval_accuracy": 0.8173618940248027,
"eval_f1": 0.8141777832297877,
"eval_loss": 0.5476882457733154,
"eval_precision": 0.8156045012406116,
"eval_recall": 0.8173618940248027,
"eval_runtime": 21.5324,
"eval_samples_per_second": 82.387,
"eval_steps_per_second": 0.65,
"step": 343
},
{
"epoch": 11.2,
"learning_rate": 6.929510155316607e-05,
"loss": 0.3416,
"step": 350
},
{
"epoch": 11.52,
"learning_rate": 6.810035842293908e-05,
"loss": 0.337,
"step": 360
},
{
"epoch": 11.84,
"learning_rate": 6.690561529271207e-05,
"loss": 0.3509,
"step": 370
},
{
"epoch": 12.0,
"eval_accuracy": 0.8280721533258174,
"eval_f1": 0.8244469418718917,
"eval_loss": 0.5437383055686951,
"eval_precision": 0.8291545990959505,
"eval_recall": 0.8280721533258174,
"eval_runtime": 21.4443,
"eval_samples_per_second": 82.726,
"eval_steps_per_second": 0.653,
"step": 375
},
{
"epoch": 12.16,
"learning_rate": 6.571087216248507e-05,
"loss": 0.3258,
"step": 380
},
{
"epoch": 12.48,
"learning_rate": 6.451612903225807e-05,
"loss": 0.3173,
"step": 390
},
{
"epoch": 12.8,
"learning_rate": 6.332138590203107e-05,
"loss": 0.3358,
"step": 400
},
{
"epoch": 12.99,
"eval_accuracy": 0.8258173618940248,
"eval_f1": 0.8240960752396663,
"eval_loss": 0.5626731514930725,
"eval_precision": 0.8267626478179836,
"eval_recall": 0.8258173618940248,
"eval_runtime": 21.2051,
"eval_samples_per_second": 83.659,
"eval_steps_per_second": 0.66,
"step": 406
},
{
"epoch": 13.12,
"learning_rate": 6.212664277180407e-05,
"loss": 0.3247,
"step": 410
},
{
"epoch": 13.44,
"learning_rate": 6.093189964157706e-05,
"loss": 0.3109,
"step": 420
},
{
"epoch": 13.76,
"learning_rate": 5.9737156511350064e-05,
"loss": 0.3027,
"step": 430
},
{
"epoch": 13.98,
"eval_accuracy": 0.8325817361894025,
"eval_f1": 0.8309989322501777,
"eval_loss": 0.5557559728622437,
"eval_precision": 0.8341182995121916,
"eval_recall": 0.8325817361894025,
"eval_runtime": 21.7522,
"eval_samples_per_second": 81.555,
"eval_steps_per_second": 0.644,
"step": 437
},
{
"epoch": 14.08,
"learning_rate": 5.8542413381123063e-05,
"loss": 0.308,
"step": 440
},
{
"epoch": 14.4,
"learning_rate": 5.7347670250896056e-05,
"loss": 0.2783,
"step": 450
},
{
"epoch": 14.72,
"learning_rate": 5.615292712066906e-05,
"loss": 0.3027,
"step": 460
},
{
"epoch": 14.98,
"eval_accuracy": 0.8325817361894025,
"eval_f1": 0.8294913013624391,
"eval_loss": 0.5702593922615051,
"eval_precision": 0.835774810800051,
"eval_recall": 0.8325817361894025,
"eval_runtime": 21.7146,
"eval_samples_per_second": 81.696,
"eval_steps_per_second": 0.645,
"step": 468
},
{
"epoch": 15.04,
"learning_rate": 5.495818399044206e-05,
"loss": 0.2816,
"step": 470
},
{
"epoch": 15.36,
"learning_rate": 5.3763440860215054e-05,
"loss": 0.2662,
"step": 480
},
{
"epoch": 15.68,
"learning_rate": 5.256869772998806e-05,
"loss": 0.2708,
"step": 490
},
{
"epoch": 16.0,
"learning_rate": 5.137395459976105e-05,
"loss": 0.2786,
"step": 500
},
{
"epoch": 16.0,
"eval_accuracy": 0.8280721533258174,
"eval_f1": 0.8248780707496759,
"eval_loss": 0.5790648460388184,
"eval_precision": 0.8268104951659752,
"eval_recall": 0.8280721533258174,
"eval_runtime": 21.3703,
"eval_samples_per_second": 83.012,
"eval_steps_per_second": 0.655,
"step": 500
},
{
"epoch": 16.32,
"learning_rate": 5.017921146953405e-05,
"loss": 0.2498,
"step": 510
},
{
"epoch": 16.64,
"learning_rate": 4.898446833930705e-05,
"loss": 0.2711,
"step": 520
},
{
"epoch": 16.96,
"learning_rate": 4.778972520908005e-05,
"loss": 0.2379,
"step": 530
},
{
"epoch": 16.99,
"eval_accuracy": 0.8275084554678692,
"eval_f1": 0.825106664588863,
"eval_loss": 0.5864401459693909,
"eval_precision": 0.8264428043886305,
"eval_recall": 0.8275084554678692,
"eval_runtime": 21.5135,
"eval_samples_per_second": 82.46,
"eval_steps_per_second": 0.651,
"step": 531
},
{
"epoch": 17.28,
"learning_rate": 4.659498207885305e-05,
"loss": 0.2566,
"step": 540
},
{
"epoch": 17.6,
"learning_rate": 4.540023894862604e-05,
"loss": 0.2602,
"step": 550
},
{
"epoch": 17.92,
"learning_rate": 4.420549581839905e-05,
"loss": 0.2426,
"step": 560
},
{
"epoch": 17.98,
"eval_accuracy": 0.8320180383314544,
"eval_f1": 0.8305311136335276,
"eval_loss": 0.5984331369400024,
"eval_precision": 0.8319552266233834,
"eval_recall": 0.8320180383314544,
"eval_runtime": 21.366,
"eval_samples_per_second": 83.029,
"eval_steps_per_second": 0.655,
"step": 562
},
{
"epoch": 18.24,
"learning_rate": 4.301075268817205e-05,
"loss": 0.241,
"step": 570
},
{
"epoch": 18.56,
"learning_rate": 4.1816009557945046e-05,
"loss": 0.2346,
"step": 580
},
{
"epoch": 18.88,
"learning_rate": 4.062126642771804e-05,
"loss": 0.2325,
"step": 590
},
{
"epoch": 18.98,
"eval_accuracy": 0.826381059751973,
"eval_f1": 0.8251810645598993,
"eval_loss": 0.6216564774513245,
"eval_precision": 0.8280960122554618,
"eval_recall": 0.826381059751973,
"eval_runtime": 21.6911,
"eval_samples_per_second": 81.785,
"eval_steps_per_second": 0.645,
"step": 593
},
{
"epoch": 19.2,
"learning_rate": 3.9426523297491045e-05,
"loss": 0.2296,
"step": 600
},
{
"epoch": 19.52,
"learning_rate": 3.8231780167264044e-05,
"loss": 0.2296,
"step": 610
},
{
"epoch": 19.84,
"learning_rate": 3.7037037037037037e-05,
"loss": 0.2208,
"step": 620
},
{
"epoch": 20.0,
"eval_accuracy": 0.8258173618940248,
"eval_f1": 0.8235775028778175,
"eval_loss": 0.6165610551834106,
"eval_precision": 0.8229737232245232,
"eval_recall": 0.8258173618940248,
"eval_runtime": 22.0389,
"eval_samples_per_second": 80.494,
"eval_steps_per_second": 0.635,
"step": 625
},
{
"epoch": 20.16,
"learning_rate": 3.5842293906810036e-05,
"loss": 0.2187,
"step": 630
},
{
"epoch": 20.48,
"learning_rate": 3.4647550776583035e-05,
"loss": 0.2067,
"step": 640
},
{
"epoch": 20.8,
"learning_rate": 3.3452807646356034e-05,
"loss": 0.2196,
"step": 650
},
{
"epoch": 20.99,
"eval_accuracy": 0.8286358511837655,
"eval_f1": 0.825876975856749,
"eval_loss": 0.6308055520057678,
"eval_precision": 0.8279920279498516,
"eval_recall": 0.8286358511837655,
"eval_runtime": 21.4064,
"eval_samples_per_second": 82.872,
"eval_steps_per_second": 0.654,
"step": 656
},
{
"epoch": 21.12,
"learning_rate": 3.2258064516129034e-05,
"loss": 0.2034,
"step": 660
},
{
"epoch": 21.44,
"learning_rate": 3.106332138590203e-05,
"loss": 0.2098,
"step": 670
},
{
"epoch": 21.76,
"learning_rate": 2.9868578255675032e-05,
"loss": 0.2077,
"step": 680
},
{
"epoch": 21.98,
"eval_accuracy": 0.8325817361894025,
"eval_f1": 0.8305408181674179,
"eval_loss": 0.624207079410553,
"eval_precision": 0.8307265402722898,
"eval_recall": 0.8325817361894025,
"eval_runtime": 21.4301,
"eval_samples_per_second": 82.781,
"eval_steps_per_second": 0.653,
"step": 687
},
{
"epoch": 22.08,
"learning_rate": 2.8673835125448028e-05,
"loss": 0.2177,
"step": 690
},
{
"epoch": 22.4,
"learning_rate": 2.747909199522103e-05,
"loss": 0.2023,
"step": 700
},
{
"epoch": 22.72,
"learning_rate": 2.628434886499403e-05,
"loss": 0.2048,
"step": 710
},
{
"epoch": 22.98,
"eval_accuracy": 0.8275084554678692,
"eval_f1": 0.8263134782090512,
"eval_loss": 0.6801280379295349,
"eval_precision": 0.8303435791241508,
"eval_recall": 0.8275084554678692,
"eval_runtime": 21.9211,
"eval_samples_per_second": 80.927,
"eval_steps_per_second": 0.639,
"step": 718
},
{
"epoch": 23.04,
"learning_rate": 2.5089605734767026e-05,
"loss": 0.2072,
"step": 720
},
{
"epoch": 23.36,
"learning_rate": 2.3894862604540025e-05,
"loss": 0.2009,
"step": 730
},
{
"epoch": 23.68,
"learning_rate": 2.270011947431302e-05,
"loss": 0.2083,
"step": 740
},
{
"epoch": 24.0,
"learning_rate": 2.1505376344086024e-05,
"loss": 0.1886,
"step": 750
},
{
"epoch": 24.0,
"eval_accuracy": 0.826381059751973,
"eval_f1": 0.8255542142311617,
"eval_loss": 0.6615379452705383,
"eval_precision": 0.8279561568090393,
"eval_recall": 0.826381059751973,
"eval_runtime": 21.4625,
"eval_samples_per_second": 82.656,
"eval_steps_per_second": 0.652,
"step": 750
},
{
"epoch": 24.32,
"learning_rate": 2.031063321385902e-05,
"loss": 0.2118,
"step": 760
},
{
"epoch": 24.64,
"learning_rate": 1.9115890083632022e-05,
"loss": 0.1872,
"step": 770
},
{
"epoch": 24.96,
"learning_rate": 1.7921146953405018e-05,
"loss": 0.2007,
"step": 780
},
{
"epoch": 24.99,
"eval_accuracy": 0.8275084554678692,
"eval_f1": 0.8267436914020984,
"eval_loss": 0.6847040057182312,
"eval_precision": 0.8279552479088658,
"eval_recall": 0.8275084554678692,
"eval_runtime": 21.4978,
"eval_samples_per_second": 82.52,
"eval_steps_per_second": 0.651,
"step": 781
},
{
"epoch": 25.28,
"learning_rate": 1.6726403823178017e-05,
"loss": 0.1844,
"step": 790
},
{
"epoch": 25.6,
"learning_rate": 1.5531660692951016e-05,
"loss": 0.1852,
"step": 800
},
{
"epoch": 25.92,
"learning_rate": 1.4336917562724014e-05,
"loss": 0.1815,
"step": 810
},
{
"epoch": 25.98,
"eval_accuracy": 0.8325817361894025,
"eval_f1": 0.830537122348923,
"eval_loss": 0.6668907999992371,
"eval_precision": 0.8311252959744954,
"eval_recall": 0.8325817361894025,
"eval_runtime": 21.8156,
"eval_samples_per_second": 81.318,
"eval_steps_per_second": 0.642,
"step": 812
},
{
"epoch": 26.24,
"learning_rate": 1.3142174432497015e-05,
"loss": 0.197,
"step": 820
},
{
"epoch": 26.56,
"learning_rate": 1.1947431302270013e-05,
"loss": 0.1913,
"step": 830
},
{
"epoch": 26.88,
"learning_rate": 1.0752688172043012e-05,
"loss": 0.1958,
"step": 840
},
{
"epoch": 26.98,
"eval_accuracy": 0.8370913190529876,
"eval_f1": 0.8357334072148467,
"eval_loss": 0.6766165494918823,
"eval_precision": 0.8373589976039267,
"eval_recall": 0.8370913190529876,
"eval_runtime": 21.0261,
"eval_samples_per_second": 84.371,
"eval_steps_per_second": 0.666,
"step": 843
},
{
"epoch": 27.2,
"learning_rate": 9.557945041816011e-06,
"loss": 0.1884,
"step": 850
},
{
"epoch": 27.52,
"learning_rate": 8.363201911589009e-06,
"loss": 0.1969,
"step": 860
},
{
"epoch": 27.84,
"learning_rate": 7.168458781362007e-06,
"loss": 0.1806,
"step": 870
},
{
"epoch": 28.0,
"eval_accuracy": 0.8359639233370914,
"eval_f1": 0.8341906674178488,
"eval_loss": 0.6678570508956909,
"eval_precision": 0.8353079467237093,
"eval_recall": 0.8359639233370914,
"eval_runtime": 21.3886,
"eval_samples_per_second": 82.941,
"eval_steps_per_second": 0.655,
"step": 875
},
{
"epoch": 28.16,
"learning_rate": 5.973715651135006e-06,
"loss": 0.1958,
"step": 880
},
{
"epoch": 28.48,
"learning_rate": 4.7789725209080055e-06,
"loss": 0.1751,
"step": 890
},
{
"epoch": 28.8,
"learning_rate": 3.5842293906810035e-06,
"loss": 0.1835,
"step": 900
},
{
"epoch": 28.99,
"eval_accuracy": 0.8348365276211951,
"eval_f1": 0.8327921869627218,
"eval_loss": 0.6767047047615051,
"eval_precision": 0.8333572524880747,
"eval_recall": 0.8348365276211951,
"eval_runtime": 21.4548,
"eval_samples_per_second": 82.685,
"eval_steps_per_second": 0.653,
"step": 906
},
{
"epoch": 29.12,
"learning_rate": 2.3894862604540028e-06,
"loss": 0.1893,
"step": 910
},
{
"epoch": 29.44,
"learning_rate": 1.1947431302270014e-06,
"loss": 0.1696,
"step": 920
},
{
"epoch": 29.76,
"learning_rate": 0.0,
"loss": 0.1796,
"step": 930
},
{
"epoch": 29.76,
"eval_accuracy": 0.8342728297632469,
"eval_f1": 0.8325855632553308,
"eval_loss": 0.6786999702453613,
"eval_precision": 0.8335690065497708,
"eval_recall": 0.8342728297632469,
"eval_runtime": 21.4373,
"eval_samples_per_second": 82.753,
"eval_steps_per_second": 0.653,
"step": 930
},
{
"epoch": 29.76,
"step": 930,
"total_flos": 3.853544865721595e+20,
"train_loss": 0.3928444991829575,
"train_runtime": 10565.0301,
"train_samples_per_second": 45.325,
"train_steps_per_second": 0.088
}
],
"logging_steps": 10,
"max_steps": 930,
"num_input_tokens_seen": 0,
"num_train_epochs": 30,
"save_steps": 500,
"total_flos": 3.853544865721595e+20,
"train_batch_size": 128,
"trial_name": null,
"trial_params": null
}