| { | |
| "best_metric": 0.9276220745449292, | |
| "best_model_checkpoint": "mobilenet_v2_1.0_224-finetuned-ISIC-dec2024test\\checkpoint-2430", | |
| "epoch": 4.998459167950694, | |
| "eval_steps": 500, | |
| "global_step": 2430, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02054442732408834, | |
| "grad_norm": 23.202468872070312, | |
| "learning_rate": 2.05761316872428e-06, | |
| "loss": 2.2149, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04108885464817668, | |
| "grad_norm": 19.670244216918945, | |
| "learning_rate": 4.11522633744856e-06, | |
| "loss": 2.2202, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.061633281972265024, | |
| "grad_norm": 18.64865493774414, | |
| "learning_rate": 6.172839506172839e-06, | |
| "loss": 2.1212, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08217770929635336, | |
| "grad_norm": 20.16490936279297, | |
| "learning_rate": 8.23045267489712e-06, | |
| "loss": 2.0225, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1027221366204417, | |
| "grad_norm": 17.26641845703125, | |
| "learning_rate": 1.02880658436214e-05, | |
| "loss": 1.9166, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12326656394453005, | |
| "grad_norm": 13.308320045471191, | |
| "learning_rate": 1.2345679012345678e-05, | |
| "loss": 1.7424, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14381099126861838, | |
| "grad_norm": 14.259194374084473, | |
| "learning_rate": 1.440329218106996e-05, | |
| "loss": 1.5742, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16435541859270672, | |
| "grad_norm": 9.573338508605957, | |
| "learning_rate": 1.646090534979424e-05, | |
| "loss": 1.4589, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18489984591679506, | |
| "grad_norm": 7.806981086730957, | |
| "learning_rate": 1.8518518518518518e-05, | |
| "loss": 1.2895, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2054442732408834, | |
| "grad_norm": 8.471112251281738, | |
| "learning_rate": 2.05761316872428e-05, | |
| "loss": 1.1936, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.22598870056497175, | |
| "grad_norm": 7.020042419433594, | |
| "learning_rate": 2.2633744855967078e-05, | |
| "loss": 0.9931, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2465331278890601, | |
| "grad_norm": 8.448640823364258, | |
| "learning_rate": 2.4691358024691357e-05, | |
| "loss": 1.0736, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2670775552131484, | |
| "grad_norm": 6.455481052398682, | |
| "learning_rate": 2.6748971193415638e-05, | |
| "loss": 1.0588, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.28762198253723675, | |
| "grad_norm": 7.331775665283203, | |
| "learning_rate": 2.880658436213992e-05, | |
| "loss": 1.0402, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3081664098613251, | |
| "grad_norm": 6.955947399139404, | |
| "learning_rate": 3.08641975308642e-05, | |
| "loss": 0.9195, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.32871083718541344, | |
| "grad_norm": 7.389803886413574, | |
| "learning_rate": 3.292181069958848e-05, | |
| "loss": 1.0719, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3492552645095018, | |
| "grad_norm": 7.89853572845459, | |
| "learning_rate": 3.497942386831276e-05, | |
| "loss": 1.0268, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3697996918335901, | |
| "grad_norm": 6.920297622680664, | |
| "learning_rate": 3.7037037037037037e-05, | |
| "loss": 0.8495, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.39034411915767847, | |
| "grad_norm": 6.326653480529785, | |
| "learning_rate": 3.909465020576132e-05, | |
| "loss": 0.8852, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.4108885464817668, | |
| "grad_norm": 6.971718788146973, | |
| "learning_rate": 4.11522633744856e-05, | |
| "loss": 0.8968, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.43143297380585516, | |
| "grad_norm": 5.848041534423828, | |
| "learning_rate": 4.3209876543209875e-05, | |
| "loss": 0.9205, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4519774011299435, | |
| "grad_norm": 8.545123100280762, | |
| "learning_rate": 4.5267489711934157e-05, | |
| "loss": 1.0261, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.47252182845403184, | |
| "grad_norm": 6.959014892578125, | |
| "learning_rate": 4.732510288065844e-05, | |
| "loss": 0.8751, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.4930662557781202, | |
| "grad_norm": 8.537935256958008, | |
| "learning_rate": 4.938271604938271e-05, | |
| "loss": 0.9604, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5136106831022085, | |
| "grad_norm": 7.854872226715088, | |
| "learning_rate": 4.983996342021033e-05, | |
| "loss": 0.9493, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5341551104262968, | |
| "grad_norm": 8.308143615722656, | |
| "learning_rate": 4.9611339734796525e-05, | |
| "loss": 0.9021, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5546995377503852, | |
| "grad_norm": 8.60777759552002, | |
| "learning_rate": 4.938271604938271e-05, | |
| "loss": 0.9398, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5752439650744735, | |
| "grad_norm": 6.065791606903076, | |
| "learning_rate": 4.9154092363968915e-05, | |
| "loss": 0.8961, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5957883923985619, | |
| "grad_norm": 7.975262641906738, | |
| "learning_rate": 4.89254686785551e-05, | |
| "loss": 0.9541, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6163328197226502, | |
| "grad_norm": 8.56069278717041, | |
| "learning_rate": 4.86968449931413e-05, | |
| "loss": 1.0714, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6368772470467385, | |
| "grad_norm": 7.618618488311768, | |
| "learning_rate": 4.8468221307727485e-05, | |
| "loss": 0.9913, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6574216743708269, | |
| "grad_norm": 5.433694839477539, | |
| "learning_rate": 4.823959762231367e-05, | |
| "loss": 0.8266, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6779661016949152, | |
| "grad_norm": 6.71955680847168, | |
| "learning_rate": 4.801097393689987e-05, | |
| "loss": 0.9065, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.6985105290190036, | |
| "grad_norm": 7.317810535430908, | |
| "learning_rate": 4.7782350251486056e-05, | |
| "loss": 0.8529, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7190549563430919, | |
| "grad_norm": 7.955277919769287, | |
| "learning_rate": 4.755372656607225e-05, | |
| "loss": 0.9454, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7395993836671803, | |
| "grad_norm": 8.274344444274902, | |
| "learning_rate": 4.732510288065844e-05, | |
| "loss": 0.9462, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7601438109912686, | |
| "grad_norm": 6.541558265686035, | |
| "learning_rate": 4.709647919524463e-05, | |
| "loss": 0.8823, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7806882383153569, | |
| "grad_norm": 7.624207019805908, | |
| "learning_rate": 4.686785550983082e-05, | |
| "loss": 0.9984, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8012326656394453, | |
| "grad_norm": 7.345012187957764, | |
| "learning_rate": 4.6639231824417016e-05, | |
| "loss": 0.9381, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8217770929635336, | |
| "grad_norm": 7.9643096923828125, | |
| "learning_rate": 4.6410608139003203e-05, | |
| "loss": 0.9472, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.842321520287622, | |
| "grad_norm": 6.939286231994629, | |
| "learning_rate": 4.618198445358939e-05, | |
| "loss": 0.9275, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8628659476117103, | |
| "grad_norm": 8.748644828796387, | |
| "learning_rate": 4.5953360768175586e-05, | |
| "loss": 1.0875, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.8834103749357987, | |
| "grad_norm": 6.044397354125977, | |
| "learning_rate": 4.5724737082761774e-05, | |
| "loss": 1.0232, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.903954802259887, | |
| "grad_norm": 10.523336410522461, | |
| "learning_rate": 4.549611339734797e-05, | |
| "loss": 0.9283, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9244992295839753, | |
| "grad_norm": 5.524984836578369, | |
| "learning_rate": 4.5267489711934157e-05, | |
| "loss": 0.9606, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9450436569080637, | |
| "grad_norm": 7.313499450683594, | |
| "learning_rate": 4.503886602652035e-05, | |
| "loss": 0.8652, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.965588084232152, | |
| "grad_norm": 6.494114398956299, | |
| "learning_rate": 4.481024234110654e-05, | |
| "loss": 0.9293, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.9861325115562404, | |
| "grad_norm": 6.250232219696045, | |
| "learning_rate": 4.4581618655692734e-05, | |
| "loss": 0.9055, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.9984591679506933, | |
| "eval_accuracy": 0.9195319271886738, | |
| "eval_loss": 0.19548115134239197, | |
| "eval_runtime": 1093.5686, | |
| "eval_samples_per_second": 6.33, | |
| "eval_steps_per_second": 0.198, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.0082177709296354, | |
| "grad_norm": 5.486711025238037, | |
| "learning_rate": 4.435299497027892e-05, | |
| "loss": 0.8596, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.0287621982537236, | |
| "grad_norm": 7.477694988250732, | |
| "learning_rate": 4.412437128486511e-05, | |
| "loss": 0.8668, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.049306625577812, | |
| "grad_norm": 5.820909023284912, | |
| "learning_rate": 4.3895747599451304e-05, | |
| "loss": 0.9094, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.0698510529019003, | |
| "grad_norm": 8.668384552001953, | |
| "learning_rate": 4.366712391403749e-05, | |
| "loss": 0.8712, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.0903954802259888, | |
| "grad_norm": 6.3633575439453125, | |
| "learning_rate": 4.343850022862369e-05, | |
| "loss": 0.8525, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.110939907550077, | |
| "grad_norm": 9.032384872436523, | |
| "learning_rate": 4.3209876543209875e-05, | |
| "loss": 0.7651, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.1314843348741654, | |
| "grad_norm": 8.188101768493652, | |
| "learning_rate": 4.298125285779607e-05, | |
| "loss": 0.9195, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.1520287621982537, | |
| "grad_norm": 6.771944046020508, | |
| "learning_rate": 4.2752629172382264e-05, | |
| "loss": 0.9012, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.1725731895223421, | |
| "grad_norm": 6.4685187339782715, | |
| "learning_rate": 4.252400548696845e-05, | |
| "loss": 0.9641, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.1931176168464304, | |
| "grad_norm": 6.535536289215088, | |
| "learning_rate": 4.229538180155465e-05, | |
| "loss": 0.8175, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.2136620441705188, | |
| "grad_norm": 7.542140960693359, | |
| "learning_rate": 4.2066758116140835e-05, | |
| "loss": 0.9377, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.234206471494607, | |
| "grad_norm": 5.927305221557617, | |
| "learning_rate": 4.183813443072703e-05, | |
| "loss": 0.9509, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.2547508988186955, | |
| "grad_norm": 7.140214920043945, | |
| "learning_rate": 4.160951074531322e-05, | |
| "loss": 0.8999, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.2752953261427837, | |
| "grad_norm": 7.2755231857299805, | |
| "learning_rate": 4.138088705989941e-05, | |
| "loss": 0.9626, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.2958397534668722, | |
| "grad_norm": 5.402399063110352, | |
| "learning_rate": 4.11522633744856e-05, | |
| "loss": 0.7817, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.3163841807909604, | |
| "grad_norm": 5.163234710693359, | |
| "learning_rate": 4.092363968907179e-05, | |
| "loss": 0.8986, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.3369286081150489, | |
| "grad_norm": 7.63501501083374, | |
| "learning_rate": 4.069501600365798e-05, | |
| "loss": 0.7566, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.357473035439137, | |
| "grad_norm": 7.454900741577148, | |
| "learning_rate": 4.046639231824417e-05, | |
| "loss": 0.8702, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.3780174627632253, | |
| "grad_norm": 6.798664569854736, | |
| "learning_rate": 4.0237768632830365e-05, | |
| "loss": 0.9139, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.3985618900873138, | |
| "grad_norm": 6.475697040557861, | |
| "learning_rate": 4.000914494741655e-05, | |
| "loss": 0.8935, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.4191063174114023, | |
| "grad_norm": 7.091508865356445, | |
| "learning_rate": 3.978052126200275e-05, | |
| "loss": 0.8321, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.4396507447354905, | |
| "grad_norm": 6.339083671569824, | |
| "learning_rate": 3.9551897576588936e-05, | |
| "loss": 0.803, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.4601951720595787, | |
| "grad_norm": 7.827945709228516, | |
| "learning_rate": 3.932327389117513e-05, | |
| "loss": 0.9727, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.4807395993836672, | |
| "grad_norm": 7.140174865722656, | |
| "learning_rate": 3.909465020576132e-05, | |
| "loss": 0.7649, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.5012840267077556, | |
| "grad_norm": 6.504294395446777, | |
| "learning_rate": 3.8866026520347506e-05, | |
| "loss": 0.8051, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.5218284540318439, | |
| "grad_norm": 7.512494087219238, | |
| "learning_rate": 3.86374028349337e-05, | |
| "loss": 0.9029, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.542372881355932, | |
| "grad_norm": 6.313861846923828, | |
| "learning_rate": 3.840877914951989e-05, | |
| "loss": 0.8271, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.5629173086800205, | |
| "grad_norm": 7.215080738067627, | |
| "learning_rate": 3.8180155464106083e-05, | |
| "loss": 0.9254, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.583461736004109, | |
| "grad_norm": 6.017473220825195, | |
| "learning_rate": 3.795153177869227e-05, | |
| "loss": 0.7945, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.6040061633281972, | |
| "grad_norm": 6.04453706741333, | |
| "learning_rate": 3.7722908093278466e-05, | |
| "loss": 0.8443, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.6245505906522855, | |
| "grad_norm": 7.159928798675537, | |
| "learning_rate": 3.7494284407864654e-05, | |
| "loss": 0.9719, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.645095017976374, | |
| "grad_norm": 6.73528528213501, | |
| "learning_rate": 3.726566072245085e-05, | |
| "loss": 0.8892, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.6656394453004624, | |
| "grad_norm": 7.667166709899902, | |
| "learning_rate": 3.7037037037037037e-05, | |
| "loss": 0.8208, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.6861838726245506, | |
| "grad_norm": 9.423199653625488, | |
| "learning_rate": 3.6808413351623224e-05, | |
| "loss": 0.8314, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.7067282999486388, | |
| "grad_norm": 8.343061447143555, | |
| "learning_rate": 3.657978966620942e-05, | |
| "loss": 0.8463, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.7272727272727273, | |
| "grad_norm": 5.084173679351807, | |
| "learning_rate": 3.635116598079561e-05, | |
| "loss": 0.8542, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.7478171545968157, | |
| "grad_norm": 6.383463382720947, | |
| "learning_rate": 3.612254229538181e-05, | |
| "loss": 0.8588, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.768361581920904, | |
| "grad_norm": 5.376101016998291, | |
| "learning_rate": 3.5893918609967996e-05, | |
| "loss": 0.752, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.7889060092449922, | |
| "grad_norm": 7.792232036590576, | |
| "learning_rate": 3.566529492455419e-05, | |
| "loss": 0.7482, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.8094504365690807, | |
| "grad_norm": 7.509520053863525, | |
| "learning_rate": 3.543667123914038e-05, | |
| "loss": 0.8904, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.8299948638931691, | |
| "grad_norm": 5.035109996795654, | |
| "learning_rate": 3.520804755372657e-05, | |
| "loss": 0.7929, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.8505392912172574, | |
| "grad_norm": 6.189474105834961, | |
| "learning_rate": 3.497942386831276e-05, | |
| "loss": 0.824, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.8710837185413456, | |
| "grad_norm": 5.561000347137451, | |
| "learning_rate": 3.475080018289895e-05, | |
| "loss": 0.8891, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.891628145865434, | |
| "grad_norm": 5.997035026550293, | |
| "learning_rate": 3.4522176497485144e-05, | |
| "loss": 0.8758, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.9121725731895225, | |
| "grad_norm": 8.139898300170898, | |
| "learning_rate": 3.429355281207133e-05, | |
| "loss": 0.8804, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.9327170005136107, | |
| "grad_norm": 5.594916820526123, | |
| "learning_rate": 3.406492912665753e-05, | |
| "loss": 0.8034, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.953261427837699, | |
| "grad_norm": 6.618235111236572, | |
| "learning_rate": 3.3836305441243715e-05, | |
| "loss": 0.8684, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.9738058551617874, | |
| "grad_norm": 5.163496017456055, | |
| "learning_rate": 3.360768175582991e-05, | |
| "loss": 0.8248, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.9943502824858759, | |
| "grad_norm": 6.3719916343688965, | |
| "learning_rate": 3.33790580704161e-05, | |
| "loss": 0.8797, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.9984591679506933, | |
| "eval_accuracy": 0.9137532505056342, | |
| "eval_loss": 0.20739668607711792, | |
| "eval_runtime": 1009.1763, | |
| "eval_samples_per_second": 6.859, | |
| "eval_steps_per_second": 0.215, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.0164355418592708, | |
| "grad_norm": 4.454899787902832, | |
| "learning_rate": 3.3150434385002285e-05, | |
| "loss": 0.7782, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.036979969183359, | |
| "grad_norm": 6.4835686683654785, | |
| "learning_rate": 3.292181069958848e-05, | |
| "loss": 0.9306, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.0575243965074472, | |
| "grad_norm": 7.925194263458252, | |
| "learning_rate": 3.269318701417467e-05, | |
| "loss": 0.8513, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.078068823831536, | |
| "grad_norm": 6.563453197479248, | |
| "learning_rate": 3.246456332876086e-05, | |
| "loss": 0.7011, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.098613251155624, | |
| "grad_norm": 7.153360366821289, | |
| "learning_rate": 3.223593964334705e-05, | |
| "loss": 0.8101, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.1191576784797124, | |
| "grad_norm": 6.468135356903076, | |
| "learning_rate": 3.2007315957933245e-05, | |
| "loss": 0.9004, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.1397021058038006, | |
| "grad_norm": 7.5966796875, | |
| "learning_rate": 3.177869227251943e-05, | |
| "loss": 0.7842, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.1602465331278893, | |
| "grad_norm": 6.29899787902832, | |
| "learning_rate": 3.155006858710563e-05, | |
| "loss": 0.7849, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.1807909604519775, | |
| "grad_norm": 7.601044654846191, | |
| "learning_rate": 3.1321444901691816e-05, | |
| "loss": 0.8189, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.2013353877760657, | |
| "grad_norm": 4.997586727142334, | |
| "learning_rate": 3.1092821216278004e-05, | |
| "loss": 0.9379, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.221879815100154, | |
| "grad_norm": 6.758198261260986, | |
| "learning_rate": 3.08641975308642e-05, | |
| "loss": 0.8441, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.242424242424242, | |
| "grad_norm": 8.121203422546387, | |
| "learning_rate": 3.0635573845450386e-05, | |
| "loss": 0.8064, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.262968669748331, | |
| "grad_norm": 6.168539047241211, | |
| "learning_rate": 3.0406950160036577e-05, | |
| "loss": 0.8648, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.283513097072419, | |
| "grad_norm": 8.698349952697754, | |
| "learning_rate": 3.017832647462277e-05, | |
| "loss": 0.9455, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.3040575243965074, | |
| "grad_norm": 6.176644802093506, | |
| "learning_rate": 2.994970278920896e-05, | |
| "loss": 0.7805, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.324601951720596, | |
| "grad_norm": 6.344507694244385, | |
| "learning_rate": 2.972107910379515e-05, | |
| "loss": 0.6886, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.3451463790446843, | |
| "grad_norm": 9.571560859680176, | |
| "learning_rate": 2.949245541838135e-05, | |
| "loss": 0.8937, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.3656908063687725, | |
| "grad_norm": 7.252879619598389, | |
| "learning_rate": 2.926383173296754e-05, | |
| "loss": 0.831, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.3862352336928607, | |
| "grad_norm": 8.151792526245117, | |
| "learning_rate": 2.903520804755373e-05, | |
| "loss": 0.8442, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.406779661016949, | |
| "grad_norm": 7.087300777435303, | |
| "learning_rate": 2.880658436213992e-05, | |
| "loss": 0.7779, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.4273240883410376, | |
| "grad_norm": 6.533515930175781, | |
| "learning_rate": 2.857796067672611e-05, | |
| "loss": 0.8222, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.447868515665126, | |
| "grad_norm": 7.180200576782227, | |
| "learning_rate": 2.8349336991312303e-05, | |
| "loss": 0.8127, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.468412942989214, | |
| "grad_norm": 6.345178604125977, | |
| "learning_rate": 2.8120713305898494e-05, | |
| "loss": 0.8132, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.4889573703133023, | |
| "grad_norm": 7.64429235458374, | |
| "learning_rate": 2.7892089620484685e-05, | |
| "loss": 0.9019, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.509501797637391, | |
| "grad_norm": 8.274768829345703, | |
| "learning_rate": 2.7663465935070876e-05, | |
| "loss": 0.8992, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.5300462249614792, | |
| "grad_norm": 6.977597713470459, | |
| "learning_rate": 2.7434842249657068e-05, | |
| "loss": 0.8527, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.5505906522855675, | |
| "grad_norm": 6.692368030548096, | |
| "learning_rate": 2.720621856424326e-05, | |
| "loss": 0.8021, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.5711350796096557, | |
| "grad_norm": 6.690873622894287, | |
| "learning_rate": 2.6977594878829447e-05, | |
| "loss": 0.8583, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.5916795069337444, | |
| "grad_norm": 6.140628337860107, | |
| "learning_rate": 2.6748971193415638e-05, | |
| "loss": 0.859, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.6122239342578326, | |
| "grad_norm": 6.804861545562744, | |
| "learning_rate": 2.652034750800183e-05, | |
| "loss": 0.7306, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.632768361581921, | |
| "grad_norm": 7.271435260772705, | |
| "learning_rate": 2.629172382258802e-05, | |
| "loss": 0.8166, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.653312788906009, | |
| "grad_norm": 6.456128120422363, | |
| "learning_rate": 2.6063100137174212e-05, | |
| "loss": 0.9386, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.6738572162300978, | |
| "grad_norm": 7.159631252288818, | |
| "learning_rate": 2.5834476451760403e-05, | |
| "loss": 0.8064, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.694401643554186, | |
| "grad_norm": 6.154369831085205, | |
| "learning_rate": 2.5605852766346595e-05, | |
| "loss": 0.8017, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.714946070878274, | |
| "grad_norm": 6.905427932739258, | |
| "learning_rate": 2.5377229080932786e-05, | |
| "loss": 0.8414, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.7354904982023625, | |
| "grad_norm": 7.776165962219238, | |
| "learning_rate": 2.5148605395518977e-05, | |
| "loss": 0.8297, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.7560349255264507, | |
| "grad_norm": 6.178536415100098, | |
| "learning_rate": 2.4919981710105165e-05, | |
| "loss": 0.7867, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.7765793528505394, | |
| "grad_norm": 5.276233196258545, | |
| "learning_rate": 2.4691358024691357e-05, | |
| "loss": 0.8698, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.7971237801746276, | |
| "grad_norm": 5.217291355133057, | |
| "learning_rate": 2.446273433927755e-05, | |
| "loss": 0.8277, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.817668207498716, | |
| "grad_norm": 5.9258856773376465, | |
| "learning_rate": 2.4234110653863743e-05, | |
| "loss": 0.8079, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.8382126348228045, | |
| "grad_norm": 5.345384120941162, | |
| "learning_rate": 2.4005486968449934e-05, | |
| "loss": 0.7356, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.8587570621468927, | |
| "grad_norm": 6.879024505615234, | |
| "learning_rate": 2.3776863283036125e-05, | |
| "loss": 0.8116, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.879301489470981, | |
| "grad_norm": 5.867737770080566, | |
| "learning_rate": 2.3548239597622316e-05, | |
| "loss": 0.7428, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.899845916795069, | |
| "grad_norm": 6.256878852844238, | |
| "learning_rate": 2.3319615912208508e-05, | |
| "loss": 0.8134, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.9203903441191574, | |
| "grad_norm": 6.346487045288086, | |
| "learning_rate": 2.3090992226794696e-05, | |
| "loss": 0.6877, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.940934771443246, | |
| "grad_norm": 5.782416820526123, | |
| "learning_rate": 2.2862368541380887e-05, | |
| "loss": 0.8478, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.9614791987673343, | |
| "grad_norm": 6.417751312255859, | |
| "learning_rate": 2.2633744855967078e-05, | |
| "loss": 0.7668, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.9820236260914226, | |
| "grad_norm": 6.743643760681152, | |
| "learning_rate": 2.240512117055327e-05, | |
| "loss": 0.8144, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.9984591679506933, | |
| "eval_accuracy": 0.9263218722912453, | |
| "eval_loss": 0.17973794043064117, | |
| "eval_runtime": 1009.1137, | |
| "eval_samples_per_second": 6.859, | |
| "eval_steps_per_second": 0.215, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 3.0041088854648175, | |
| "grad_norm": 6.051700592041016, | |
| "learning_rate": 2.217649748513946e-05, | |
| "loss": 0.9408, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 3.024653312788906, | |
| "grad_norm": 8.463972091674805, | |
| "learning_rate": 2.1947873799725652e-05, | |
| "loss": 0.7778, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 3.0451977401129944, | |
| "grad_norm": 6.033344268798828, | |
| "learning_rate": 2.1719250114311843e-05, | |
| "loss": 0.7648, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 3.0657421674370826, | |
| "grad_norm": 6.287738800048828, | |
| "learning_rate": 2.1490626428898035e-05, | |
| "loss": 0.7507, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 3.086286594761171, | |
| "grad_norm": 6.505873680114746, | |
| "learning_rate": 2.1262002743484226e-05, | |
| "loss": 0.7668, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.1068310220852595, | |
| "grad_norm": 5.928491115570068, | |
| "learning_rate": 2.1033379058070417e-05, | |
| "loss": 0.8387, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 3.1273754494093478, | |
| "grad_norm": 8.137348175048828, | |
| "learning_rate": 2.080475537265661e-05, | |
| "loss": 0.7546, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 3.147919876733436, | |
| "grad_norm": 6.729381561279297, | |
| "learning_rate": 2.05761316872428e-05, | |
| "loss": 0.7086, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 3.168464304057524, | |
| "grad_norm": 6.623689651489258, | |
| "learning_rate": 2.034750800182899e-05, | |
| "loss": 0.9436, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 3.189008731381613, | |
| "grad_norm": 7.943443298339844, | |
| "learning_rate": 2.0118884316415183e-05, | |
| "loss": 0.8645, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.209553158705701, | |
| "grad_norm": 5.353769302368164, | |
| "learning_rate": 1.9890260631001374e-05, | |
| "loss": 0.7909, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 3.2300975860297894, | |
| "grad_norm": 6.193889141082764, | |
| "learning_rate": 1.9661636945587565e-05, | |
| "loss": 0.7595, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 3.2506420133538776, | |
| "grad_norm": 8.73640251159668, | |
| "learning_rate": 1.9433013260173753e-05, | |
| "loss": 0.729, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 3.2711864406779663, | |
| "grad_norm": 6.345188617706299, | |
| "learning_rate": 1.9204389574759944e-05, | |
| "loss": 0.6933, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 3.2917308680020545, | |
| "grad_norm": 9.154464721679688, | |
| "learning_rate": 1.8975765889346136e-05, | |
| "loss": 0.8427, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.3122752953261427, | |
| "grad_norm": 7.497635841369629, | |
| "learning_rate": 1.8747142203932327e-05, | |
| "loss": 0.7922, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 3.332819722650231, | |
| "grad_norm": 6.4137468338012695, | |
| "learning_rate": 1.8518518518518518e-05, | |
| "loss": 0.7468, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 3.3533641499743196, | |
| "grad_norm": 8.157144546508789, | |
| "learning_rate": 1.828989483310471e-05, | |
| "loss": 0.7347, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 3.373908577298408, | |
| "grad_norm": 5.25002908706665, | |
| "learning_rate": 1.8061271147690904e-05, | |
| "loss": 0.7833, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 3.394453004622496, | |
| "grad_norm": 6.77322244644165, | |
| "learning_rate": 1.7832647462277096e-05, | |
| "loss": 0.7639, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.4149974319465843, | |
| "grad_norm": 6.466352462768555, | |
| "learning_rate": 1.7604023776863283e-05, | |
| "loss": 0.6913, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 3.435541859270673, | |
| "grad_norm": 6.149074077606201, | |
| "learning_rate": 1.7375400091449475e-05, | |
| "loss": 0.85, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 3.4560862865947612, | |
| "grad_norm": 8.840483665466309, | |
| "learning_rate": 1.7146776406035666e-05, | |
| "loss": 0.8483, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 3.4766307139188495, | |
| "grad_norm": 7.109951019287109, | |
| "learning_rate": 1.6918152720621857e-05, | |
| "loss": 0.8669, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 3.4971751412429377, | |
| "grad_norm": 7.431482315063477, | |
| "learning_rate": 1.668952903520805e-05, | |
| "loss": 0.7942, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.517719568567026, | |
| "grad_norm": 8.127689361572266, | |
| "learning_rate": 1.646090534979424e-05, | |
| "loss": 0.8801, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 3.5382639958911146, | |
| "grad_norm": 5.593295574188232, | |
| "learning_rate": 1.623228166438043e-05, | |
| "loss": 0.797, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 3.558808423215203, | |
| "grad_norm": 6.143307685852051, | |
| "learning_rate": 1.6003657978966623e-05, | |
| "loss": 0.8404, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.579352850539291, | |
| "grad_norm": 7.268124103546143, | |
| "learning_rate": 1.5775034293552814e-05, | |
| "loss": 0.7545, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.5998972778633798, | |
| "grad_norm": 6.966310501098633, | |
| "learning_rate": 1.5546410608139002e-05, | |
| "loss": 0.7656, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.620441705187468, | |
| "grad_norm": 7.398248672485352, | |
| "learning_rate": 1.5317786922725193e-05, | |
| "loss": 0.8126, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.6409861325115562, | |
| "grad_norm": 7.494919776916504, | |
| "learning_rate": 1.5089163237311384e-05, | |
| "loss": 0.908, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.6615305598356445, | |
| "grad_norm": 5.857889175415039, | |
| "learning_rate": 1.4860539551897576e-05, | |
| "loss": 0.7003, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.6820749871597327, | |
| "grad_norm": 8.693001747131348, | |
| "learning_rate": 1.463191586648377e-05, | |
| "loss": 0.8321, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 3.7026194144838214, | |
| "grad_norm": 8.051487922668457, | |
| "learning_rate": 1.440329218106996e-05, | |
| "loss": 0.8759, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.7231638418079096, | |
| "grad_norm": 5.1894612312316895, | |
| "learning_rate": 1.4174668495656151e-05, | |
| "loss": 0.7325, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 3.743708269131998, | |
| "grad_norm": 5.104062557220459, | |
| "learning_rate": 1.3946044810242343e-05, | |
| "loss": 0.7732, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 3.7642526964560865, | |
| "grad_norm": 7.708363056182861, | |
| "learning_rate": 1.3717421124828534e-05, | |
| "loss": 0.8328, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 3.7847971237801747, | |
| "grad_norm": 6.15858268737793, | |
| "learning_rate": 1.3488797439414723e-05, | |
| "loss": 0.738, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 3.805341551104263, | |
| "grad_norm": 7.959890365600586, | |
| "learning_rate": 1.3260173754000915e-05, | |
| "loss": 0.818, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.825885978428351, | |
| "grad_norm": 7.602783679962158, | |
| "learning_rate": 1.3031550068587106e-05, | |
| "loss": 0.9028, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 3.8464304057524394, | |
| "grad_norm": 7.417806625366211, | |
| "learning_rate": 1.2802926383173297e-05, | |
| "loss": 0.8158, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 3.866974833076528, | |
| "grad_norm": 6.925180435180664, | |
| "learning_rate": 1.2574302697759489e-05, | |
| "loss": 0.7867, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 3.8875192604006163, | |
| "grad_norm": 4.716423034667969, | |
| "learning_rate": 1.2345679012345678e-05, | |
| "loss": 0.6969, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 3.9080636877247046, | |
| "grad_norm": 5.976194381713867, | |
| "learning_rate": 1.2117055326931871e-05, | |
| "loss": 0.7292, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.9286081150487933, | |
| "grad_norm": 6.812644958496094, | |
| "learning_rate": 1.1888431641518063e-05, | |
| "loss": 0.7845, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 3.9491525423728815, | |
| "grad_norm": 7.749550819396973, | |
| "learning_rate": 1.1659807956104254e-05, | |
| "loss": 0.8008, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 3.9696969696969697, | |
| "grad_norm": 7.299574375152588, | |
| "learning_rate": 1.1431184270690443e-05, | |
| "loss": 0.7896, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 3.990241397021058, | |
| "grad_norm": 7.994142055511475, | |
| "learning_rate": 1.1202560585276635e-05, | |
| "loss": 0.9243, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 3.9984591679506933, | |
| "eval_accuracy": 0.9232880670326495, | |
| "eval_loss": 0.18616917729377747, | |
| "eval_runtime": 1074.4755, | |
| "eval_samples_per_second": 6.442, | |
| "eval_steps_per_second": 0.202, | |
| "step": 1944 | |
| }, | |
| { | |
| "epoch": 4.012326656394453, | |
| "grad_norm": 6.763792991638184, | |
| "learning_rate": 1.0973936899862826e-05, | |
| "loss": 0.7882, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.0328710837185415, | |
| "grad_norm": 7.702907085418701, | |
| "learning_rate": 1.0745313214449017e-05, | |
| "loss": 0.8705, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 4.05341551104263, | |
| "grad_norm": 6.545944690704346, | |
| "learning_rate": 1.0516689529035209e-05, | |
| "loss": 0.8295, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 4.073959938366718, | |
| "grad_norm": 7.43347692489624, | |
| "learning_rate": 1.02880658436214e-05, | |
| "loss": 0.7686, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 4.094504365690806, | |
| "grad_norm": 6.285999298095703, | |
| "learning_rate": 1.0059442158207591e-05, | |
| "loss": 0.7991, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 4.1150487930148945, | |
| "grad_norm": 8.586403846740723, | |
| "learning_rate": 9.830818472793783e-06, | |
| "loss": 0.7933, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.135593220338983, | |
| "grad_norm": 7.343191623687744, | |
| "learning_rate": 9.602194787379972e-06, | |
| "loss": 0.7762, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 4.156137647663072, | |
| "grad_norm": 5.3163933753967285, | |
| "learning_rate": 9.373571101966163e-06, | |
| "loss": 0.7508, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 4.17668207498716, | |
| "grad_norm": 6.618367671966553, | |
| "learning_rate": 9.144947416552355e-06, | |
| "loss": 0.671, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 4.197226502311248, | |
| "grad_norm": 4.874975681304932, | |
| "learning_rate": 8.916323731138548e-06, | |
| "loss": 0.713, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 4.2177709296353365, | |
| "grad_norm": 6.649152755737305, | |
| "learning_rate": 8.687700045724737e-06, | |
| "loss": 0.8196, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.238315356959425, | |
| "grad_norm": 6.106600284576416, | |
| "learning_rate": 8.459076360310929e-06, | |
| "loss": 0.744, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 4.258859784283513, | |
| "grad_norm": 7.9351630210876465, | |
| "learning_rate": 8.23045267489712e-06, | |
| "loss": 0.7915, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 4.279404211607601, | |
| "grad_norm": 8.719624519348145, | |
| "learning_rate": 8.001828989483311e-06, | |
| "loss": 0.8307, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 4.299948638931689, | |
| "grad_norm": 8.237951278686523, | |
| "learning_rate": 7.773205304069501e-06, | |
| "loss": 0.7875, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 4.320493066255779, | |
| "grad_norm": 7.945969581604004, | |
| "learning_rate": 7.544581618655692e-06, | |
| "loss": 0.7283, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.341037493579867, | |
| "grad_norm": 9.574934959411621, | |
| "learning_rate": 7.315957933241885e-06, | |
| "loss": 0.8088, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 4.361581920903955, | |
| "grad_norm": 7.645279884338379, | |
| "learning_rate": 7.087334247828076e-06, | |
| "loss": 0.77, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 4.382126348228043, | |
| "grad_norm": 7.663369178771973, | |
| "learning_rate": 6.858710562414267e-06, | |
| "loss": 0.7608, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 4.4026707755521315, | |
| "grad_norm": 7.651998996734619, | |
| "learning_rate": 6.630086877000457e-06, | |
| "loss": 0.7465, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 4.42321520287622, | |
| "grad_norm": 6.678437232971191, | |
| "learning_rate": 6.401463191586649e-06, | |
| "loss": 0.725, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.443759630200308, | |
| "grad_norm": 7.033961772918701, | |
| "learning_rate": 6.172839506172839e-06, | |
| "loss": 0.7678, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 4.464304057524396, | |
| "grad_norm": 7.408419609069824, | |
| "learning_rate": 5.944215820759031e-06, | |
| "loss": 0.7689, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 4.484848484848484, | |
| "grad_norm": 8.51754379272461, | |
| "learning_rate": 5.715592135345222e-06, | |
| "loss": 0.852, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 4.5053929121725735, | |
| "grad_norm": 5.929790019989014, | |
| "learning_rate": 5.486968449931413e-06, | |
| "loss": 0.7294, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 4.525937339496662, | |
| "grad_norm": 7.980464935302734, | |
| "learning_rate": 5.258344764517604e-06, | |
| "loss": 0.8615, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.54648176682075, | |
| "grad_norm": 9.191452026367188, | |
| "learning_rate": 5.029721079103796e-06, | |
| "loss": 0.9008, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 4.567026194144838, | |
| "grad_norm": 8.33234977722168, | |
| "learning_rate": 4.801097393689986e-06, | |
| "loss": 0.8745, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 4.5875706214689265, | |
| "grad_norm": 6.941461086273193, | |
| "learning_rate": 4.572473708276177e-06, | |
| "loss": 0.7385, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 4.608115048793015, | |
| "grad_norm": 4.740252494812012, | |
| "learning_rate": 4.343850022862369e-06, | |
| "loss": 0.7224, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 4.628659476117103, | |
| "grad_norm": 7.316972255706787, | |
| "learning_rate": 4.11522633744856e-06, | |
| "loss": 0.8351, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.649203903441192, | |
| "grad_norm": 6.795185565948486, | |
| "learning_rate": 3.8866026520347504e-06, | |
| "loss": 0.7554, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 4.66974833076528, | |
| "grad_norm": 8.930145263671875, | |
| "learning_rate": 3.6579789666209426e-06, | |
| "loss": 0.8177, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 4.6902927580893685, | |
| "grad_norm": 7.089832782745361, | |
| "learning_rate": 3.4293552812071335e-06, | |
| "loss": 0.7476, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 4.710837185413457, | |
| "grad_norm": 6.567149639129639, | |
| "learning_rate": 3.2007315957933243e-06, | |
| "loss": 0.841, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 4.731381612737545, | |
| "grad_norm": 6.558457851409912, | |
| "learning_rate": 2.9721079103795156e-06, | |
| "loss": 0.7575, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.751926040061633, | |
| "grad_norm": 8.95292854309082, | |
| "learning_rate": 2.7434842249657065e-06, | |
| "loss": 0.7493, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 4.7724704673857214, | |
| "grad_norm": 7.489271640777588, | |
| "learning_rate": 2.514860539551898e-06, | |
| "loss": 0.716, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 4.79301489470981, | |
| "grad_norm": 7.381133079528809, | |
| "learning_rate": 2.2862368541380887e-06, | |
| "loss": 0.8227, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 4.813559322033898, | |
| "grad_norm": 8.651927947998047, | |
| "learning_rate": 2.05761316872428e-06, | |
| "loss": 0.7768, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 4.834103749357987, | |
| "grad_norm": 7.568137168884277, | |
| "learning_rate": 1.8289894833104713e-06, | |
| "loss": 0.7768, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.854648176682075, | |
| "grad_norm": 8.011246681213379, | |
| "learning_rate": 1.6003657978966622e-06, | |
| "loss": 0.7113, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 4.8751926040061635, | |
| "grad_norm": 8.442760467529297, | |
| "learning_rate": 1.3717421124828533e-06, | |
| "loss": 0.7428, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 4.895737031330252, | |
| "grad_norm": 5.148819446563721, | |
| "learning_rate": 1.1431184270690443e-06, | |
| "loss": 0.7326, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 4.91628145865434, | |
| "grad_norm": 6.452014923095703, | |
| "learning_rate": 9.144947416552356e-07, | |
| "loss": 0.8117, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 4.936825885978428, | |
| "grad_norm": 5.830743789672852, | |
| "learning_rate": 6.858710562414266e-07, | |
| "loss": 0.7118, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.957370313302516, | |
| "grad_norm": 6.933480739593506, | |
| "learning_rate": 4.572473708276178e-07, | |
| "loss": 0.7462, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 4.977914740626605, | |
| "grad_norm": 6.512324810028076, | |
| "learning_rate": 2.286236854138089e-07, | |
| "loss": 0.7691, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 4.998459167950694, | |
| "grad_norm": 6.746326446533203, | |
| "learning_rate": 0.0, | |
| "loss": 0.8199, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 4.998459167950694, | |
| "eval_accuracy": 0.9276220745449292, | |
| "eval_loss": 0.17630332708358765, | |
| "eval_runtime": 1096.6155, | |
| "eval_samples_per_second": 6.312, | |
| "eval_steps_per_second": 0.198, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 4.998459167950694, | |
| "step": 2430, | |
| "total_flos": 8.175904340280607e+17, | |
| "train_loss": 0.8756545659445931, | |
| "train_runtime": 61216.2023, | |
| "train_samples_per_second": 5.088, | |
| "train_steps_per_second": 0.04 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2430, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8.175904340280607e+17, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |