| { | |
| "best_global_step": 1500, | |
| "best_metric": 0.10838954150676727, | |
| "best_model_checkpoint": "./whisper-small-splitted/checkpoint-1500", | |
| "epoch": 12.656151419558359, | |
| "eval_steps": 500, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.10515247108307045, | |
| "grad_norm": 37.98896408081055, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 4.0211, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.2103049421661409, | |
| "grad_norm": 28.073650360107422, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 3.1838, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.31545741324921134, | |
| "grad_norm": 12.10400676727295, | |
| "learning_rate": 1.48e-06, | |
| "loss": 2.0459, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.4206098843322818, | |
| "grad_norm": 8.05373477935791, | |
| "learning_rate": 1.98e-06, | |
| "loss": 1.3254, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5257623554153522, | |
| "grad_norm": 6.276533126831055, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 1.0124, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.6309148264984227, | |
| "grad_norm": 5.450071334838867, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.7972, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7360672975814931, | |
| "grad_norm": 6.463825225830078, | |
| "learning_rate": 3.48e-06, | |
| "loss": 0.6987, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.8412197686645636, | |
| "grad_norm": 6.113511085510254, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 0.6371, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.9463722397476341, | |
| "grad_norm": 5.203503131866455, | |
| "learning_rate": 4.48e-06, | |
| "loss": 0.5403, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.0546792849631967, | |
| "grad_norm": 3.6247618198394775, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 0.4597, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.159831756046267, | |
| "grad_norm": 4.7159552574157715, | |
| "learning_rate": 5.480000000000001e-06, | |
| "loss": 0.4074, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.2649842271293377, | |
| "grad_norm": 5.012338638305664, | |
| "learning_rate": 5.98e-06, | |
| "loss": 0.3846, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.370136698212408, | |
| "grad_norm": 4.738580703735352, | |
| "learning_rate": 6.480000000000001e-06, | |
| "loss": 0.3306, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.4752891692954784, | |
| "grad_norm": 4.1813578605651855, | |
| "learning_rate": 6.98e-06, | |
| "loss": 0.3145, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.5804416403785488, | |
| "grad_norm": 5.1083855628967285, | |
| "learning_rate": 7.48e-06, | |
| "loss": 0.2901, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.6855941114616193, | |
| "grad_norm": 3.654137372970581, | |
| "learning_rate": 7.980000000000002e-06, | |
| "loss": 0.2473, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.7907465825446898, | |
| "grad_norm": 5.576925754547119, | |
| "learning_rate": 8.48e-06, | |
| "loss": 0.2408, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.8958990536277602, | |
| "grad_norm": 5.050299167633057, | |
| "learning_rate": 8.98e-06, | |
| "loss": 0.222, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.004206098843323, | |
| "grad_norm": 4.0772705078125, | |
| "learning_rate": 9.48e-06, | |
| "loss": 0.2311, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.1093585699263935, | |
| "grad_norm": 3.189274311065674, | |
| "learning_rate": 9.980000000000001e-06, | |
| "loss": 0.1324, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.1093585699263935, | |
| "eval_loss": 0.1909431368112564, | |
| "eval_runtime": 1339.9903, | |
| "eval_samples_per_second": 2.432, | |
| "eval_steps_per_second": 0.608, | |
| "eval_wer": 0.3306828586574141, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.2145110410094637, | |
| "grad_norm": 3.6630983352661133, | |
| "learning_rate": 9.982507288629738e-06, | |
| "loss": 0.1425, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.319663512092534, | |
| "grad_norm": 3.793405532836914, | |
| "learning_rate": 9.964285714285714e-06, | |
| "loss": 0.1394, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.4248159831756047, | |
| "grad_norm": 3.442934036254883, | |
| "learning_rate": 9.94606413994169e-06, | |
| "loss": 0.1445, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.5299684542586753, | |
| "grad_norm": 2.4245970249176025, | |
| "learning_rate": 9.927842565597668e-06, | |
| "loss": 0.1247, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.6351209253417456, | |
| "grad_norm": 2.134147882461548, | |
| "learning_rate": 9.909620991253646e-06, | |
| "loss": 0.1246, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.740273396424816, | |
| "grad_norm": 3.135070562362671, | |
| "learning_rate": 9.891399416909622e-06, | |
| "loss": 0.1233, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.8454258675078865, | |
| "grad_norm": 2.723932981491089, | |
| "learning_rate": 9.873177842565598e-06, | |
| "loss": 0.119, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.9505783385909568, | |
| "grad_norm": 3.9131784439086914, | |
| "learning_rate": 9.854956268221574e-06, | |
| "loss": 0.1334, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.0588853838065195, | |
| "grad_norm": 4.044662952423096, | |
| "learning_rate": 9.836734693877552e-06, | |
| "loss": 0.0902, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.1640378548895898, | |
| "grad_norm": 2.1829943656921387, | |
| "learning_rate": 9.818513119533528e-06, | |
| "loss": 0.0702, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.2691903259726605, | |
| "grad_norm": 2.548558235168457, | |
| "learning_rate": 9.800291545189506e-06, | |
| "loss": 0.0712, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.3743427970557307, | |
| "grad_norm": 1.3480279445648193, | |
| "learning_rate": 9.782069970845482e-06, | |
| "loss": 0.0654, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.4794952681388014, | |
| "grad_norm": 2.091961622238159, | |
| "learning_rate": 9.763848396501458e-06, | |
| "loss": 0.0658, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.5846477392218716, | |
| "grad_norm": 2.2356009483337402, | |
| "learning_rate": 9.745626822157436e-06, | |
| "loss": 0.0682, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.6898002103049423, | |
| "grad_norm": 2.5908384323120117, | |
| "learning_rate": 9.727405247813412e-06, | |
| "loss": 0.056, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.7949526813880126, | |
| "grad_norm": 2.618177890777588, | |
| "learning_rate": 9.70918367346939e-06, | |
| "loss": 0.0623, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.900105152471083, | |
| "grad_norm": 2.1948764324188232, | |
| "learning_rate": 9.690962099125366e-06, | |
| "loss": 0.0752, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.008412197686646, | |
| "grad_norm": 1.5540448427200317, | |
| "learning_rate": 9.672740524781342e-06, | |
| "loss": 0.0638, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.113564668769716, | |
| "grad_norm": 1.905039668083191, | |
| "learning_rate": 9.654518950437318e-06, | |
| "loss": 0.0397, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.218717139852787, | |
| "grad_norm": 2.011355400085449, | |
| "learning_rate": 9.636297376093296e-06, | |
| "loss": 0.0383, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.218717139852787, | |
| "eval_loss": 0.1153440847992897, | |
| "eval_runtime": 1137.0265, | |
| "eval_samples_per_second": 2.866, | |
| "eval_steps_per_second": 0.717, | |
| "eval_wer": 0.11305479254011855, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.323869610935857, | |
| "grad_norm": 1.983497142791748, | |
| "learning_rate": 9.618075801749272e-06, | |
| "loss": 0.0376, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.429022082018927, | |
| "grad_norm": 2.7597391605377197, | |
| "learning_rate": 9.599854227405248e-06, | |
| "loss": 0.0384, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.534174553101998, | |
| "grad_norm": 2.3190319538116455, | |
| "learning_rate": 9.581632653061226e-06, | |
| "loss": 0.041, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.639327024185068, | |
| "grad_norm": 1.6487782001495361, | |
| "learning_rate": 9.563411078717202e-06, | |
| "loss": 0.0373, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.744479495268139, | |
| "grad_norm": 1.493482232093811, | |
| "learning_rate": 9.54518950437318e-06, | |
| "loss": 0.0453, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.849631966351209, | |
| "grad_norm": 1.9630647897720337, | |
| "learning_rate": 9.526967930029156e-06, | |
| "loss": 0.0447, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.95478443743428, | |
| "grad_norm": 2.1700949668884277, | |
| "learning_rate": 9.508746355685132e-06, | |
| "loss": 0.0416, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.063091482649842, | |
| "grad_norm": 2.0949151515960693, | |
| "learning_rate": 9.490524781341108e-06, | |
| "loss": 0.0296, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.168243953732913, | |
| "grad_norm": 1.232323169708252, | |
| "learning_rate": 9.472303206997086e-06, | |
| "loss": 0.0227, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 5.273396424815983, | |
| "grad_norm": 6.17132568359375, | |
| "learning_rate": 9.454081632653062e-06, | |
| "loss": 0.0284, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.378548895899054, | |
| "grad_norm": 0.9878859519958496, | |
| "learning_rate": 9.43586005830904e-06, | |
| "loss": 0.0211, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 5.483701366982124, | |
| "grad_norm": 2.1974642276763916, | |
| "learning_rate": 9.417638483965016e-06, | |
| "loss": 0.0203, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.588853838065194, | |
| "grad_norm": 1.2606751918792725, | |
| "learning_rate": 9.399416909620992e-06, | |
| "loss": 0.0271, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 5.694006309148265, | |
| "grad_norm": 1.851982593536377, | |
| "learning_rate": 9.38119533527697e-06, | |
| "loss": 0.0271, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.799158780231336, | |
| "grad_norm": 1.9509834051132202, | |
| "learning_rate": 9.362973760932945e-06, | |
| "loss": 0.0261, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.904311251314406, | |
| "grad_norm": 1.3138453960418701, | |
| "learning_rate": 9.344752186588922e-06, | |
| "loss": 0.0269, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.012618296529968, | |
| "grad_norm": 1.0709035396575928, | |
| "learning_rate": 9.326530612244898e-06, | |
| "loss": 0.0292, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 6.117770767613039, | |
| "grad_norm": 0.9309059381484985, | |
| "learning_rate": 9.308309037900875e-06, | |
| "loss": 0.0128, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 6.22292323869611, | |
| "grad_norm": 1.078444004058838, | |
| "learning_rate": 9.290087463556851e-06, | |
| "loss": 0.0189, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 6.3280757097791795, | |
| "grad_norm": 2.0890743732452393, | |
| "learning_rate": 9.27186588921283e-06, | |
| "loss": 0.0184, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.3280757097791795, | |
| "eval_loss": 0.10838954150676727, | |
| "eval_runtime": 1142.8117, | |
| "eval_samples_per_second": 2.852, | |
| "eval_steps_per_second": 0.713, | |
| "eval_wer": 0.11257288805358778, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.43322818086225, | |
| "grad_norm": 0.5473038554191589, | |
| "learning_rate": 9.253644314868805e-06, | |
| "loss": 0.0153, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 6.538380651945321, | |
| "grad_norm": 1.075249433517456, | |
| "learning_rate": 9.235422740524781e-06, | |
| "loss": 0.0172, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 6.643533123028391, | |
| "grad_norm": 1.237701177597046, | |
| "learning_rate": 9.21720116618076e-06, | |
| "loss": 0.0167, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 6.748685594111461, | |
| "grad_norm": 1.9220877885818481, | |
| "learning_rate": 9.198979591836735e-06, | |
| "loss": 0.0175, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 6.853838065194532, | |
| "grad_norm": 1.435571312904358, | |
| "learning_rate": 9.180758017492713e-06, | |
| "loss": 0.0182, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 6.958990536277603, | |
| "grad_norm": 1.9936102628707886, | |
| "learning_rate": 9.162536443148689e-06, | |
| "loss": 0.0158, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.067297581493165, | |
| "grad_norm": 1.101319670677185, | |
| "learning_rate": 9.144314868804665e-06, | |
| "loss": 0.0147, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 7.172450052576235, | |
| "grad_norm": 1.3670756816864014, | |
| "learning_rate": 9.126093294460641e-06, | |
| "loss": 0.0127, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 7.277602523659306, | |
| "grad_norm": 0.6474692225456238, | |
| "learning_rate": 9.107871720116619e-06, | |
| "loss": 0.0107, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 7.382754994742377, | |
| "grad_norm": 0.941483736038208, | |
| "learning_rate": 9.089650145772595e-06, | |
| "loss": 0.0129, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 7.4879074658254465, | |
| "grad_norm": 0.7835807204246521, | |
| "learning_rate": 9.071428571428573e-06, | |
| "loss": 0.015, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 7.593059936908517, | |
| "grad_norm": 0.7577527761459351, | |
| "learning_rate": 9.053206997084549e-06, | |
| "loss": 0.0119, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 7.698212407991588, | |
| "grad_norm": 1.8214268684387207, | |
| "learning_rate": 9.034985422740525e-06, | |
| "loss": 0.0116, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 7.803364879074659, | |
| "grad_norm": 0.8769547343254089, | |
| "learning_rate": 9.016763848396503e-06, | |
| "loss": 0.0131, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 7.908517350157728, | |
| "grad_norm": 1.227581262588501, | |
| "learning_rate": 8.998542274052479e-06, | |
| "loss": 0.012, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 8.016824395373291, | |
| "grad_norm": 1.28843355178833, | |
| "learning_rate": 8.980320699708455e-06, | |
| "loss": 0.0134, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 8.121976866456361, | |
| "grad_norm": 1.5814619064331055, | |
| "learning_rate": 8.962099125364431e-06, | |
| "loss": 0.0086, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.227129337539433, | |
| "grad_norm": 0.8933718800544739, | |
| "learning_rate": 8.943877551020409e-06, | |
| "loss": 0.0089, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 8.332281808622502, | |
| "grad_norm": 0.9096396565437317, | |
| "learning_rate": 8.925655976676385e-06, | |
| "loss": 0.0097, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 8.437434279705574, | |
| "grad_norm": 0.42109328508377075, | |
| "learning_rate": 8.907434402332363e-06, | |
| "loss": 0.0098, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.437434279705574, | |
| "eval_loss": 0.11219166219234467, | |
| "eval_runtime": 1135.5499, | |
| "eval_samples_per_second": 2.87, | |
| "eval_steps_per_second": 0.718, | |
| "eval_wer": 0.10139270396607393, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.542586750788644, | |
| "grad_norm": 1.2390159368515015, | |
| "learning_rate": 8.889212827988339e-06, | |
| "loss": 0.0106, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 8.647739221871714, | |
| "grad_norm": 0.9200573563575745, | |
| "learning_rate": 8.870991253644315e-06, | |
| "loss": 0.0082, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 8.752891692954785, | |
| "grad_norm": 0.814224362373352, | |
| "learning_rate": 8.852769679300293e-06, | |
| "loss": 0.0105, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 8.858044164037855, | |
| "grad_norm": 0.7412490248680115, | |
| "learning_rate": 8.834548104956269e-06, | |
| "loss": 0.0101, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 8.963196635120925, | |
| "grad_norm": 0.8766764402389526, | |
| "learning_rate": 8.816326530612247e-06, | |
| "loss": 0.0092, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 9.071503680336487, | |
| "grad_norm": 0.09623388946056366, | |
| "learning_rate": 8.798104956268223e-06, | |
| "loss": 0.0088, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 9.176656151419559, | |
| "grad_norm": 0.6915100812911987, | |
| "learning_rate": 8.779883381924199e-06, | |
| "loss": 0.0073, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 9.281808622502629, | |
| "grad_norm": 0.6448748707771301, | |
| "learning_rate": 8.761661807580175e-06, | |
| "loss": 0.0089, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.386961093585699, | |
| "grad_norm": 1.2592213153839111, | |
| "learning_rate": 8.743440233236153e-06, | |
| "loss": 0.0096, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 9.49211356466877, | |
| "grad_norm": 0.8664531111717224, | |
| "learning_rate": 8.72521865889213e-06, | |
| "loss": 0.0073, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 9.59726603575184, | |
| "grad_norm": 0.49285972118377686, | |
| "learning_rate": 8.706997084548106e-06, | |
| "loss": 0.0078, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 9.70241850683491, | |
| "grad_norm": 0.9507968425750732, | |
| "learning_rate": 8.688775510204082e-06, | |
| "loss": 0.0092, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 9.807570977917981, | |
| "grad_norm": 1.0036101341247559, | |
| "learning_rate": 8.670553935860059e-06, | |
| "loss": 0.0103, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 9.912723449001051, | |
| "grad_norm": 0.8855874538421631, | |
| "learning_rate": 8.653061224489798e-06, | |
| "loss": 0.0111, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 10.021030494216614, | |
| "grad_norm": 0.918958306312561, | |
| "learning_rate": 8.634839650145774e-06, | |
| "loss": 0.0097, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 10.126182965299684, | |
| "grad_norm": 1.442686915397644, | |
| "learning_rate": 8.61661807580175e-06, | |
| "loss": 0.0083, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 10.231335436382755, | |
| "grad_norm": 1.1829726696014404, | |
| "learning_rate": 8.598396501457726e-06, | |
| "loss": 0.0076, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 10.336487907465825, | |
| "grad_norm": 1.1316958665847778, | |
| "learning_rate": 8.580174927113704e-06, | |
| "loss": 0.0086, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 10.441640378548897, | |
| "grad_norm": 0.5993586778640747, | |
| "learning_rate": 8.56195335276968e-06, | |
| "loss": 0.0085, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 10.546792849631967, | |
| "grad_norm": 1.1639779806137085, | |
| "learning_rate": 8.543731778425656e-06, | |
| "loss": 0.0076, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.546792849631967, | |
| "eval_loss": 0.1101309061050415, | |
| "eval_runtime": 1133.8828, | |
| "eval_samples_per_second": 2.874, | |
| "eval_steps_per_second": 0.719, | |
| "eval_wer": 0.09657365910076623, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.651945320715036, | |
| "grad_norm": 1.1275087594985962, | |
| "learning_rate": 8.525510204081632e-06, | |
| "loss": 0.008, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 10.757097791798108, | |
| "grad_norm": 1.17903470993042, | |
| "learning_rate": 8.50728862973761e-06, | |
| "loss": 0.0082, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 10.862250262881178, | |
| "grad_norm": 0.35520169138908386, | |
| "learning_rate": 8.489067055393588e-06, | |
| "loss": 0.0085, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 10.967402733964247, | |
| "grad_norm": 1.5375522375106812, | |
| "learning_rate": 8.470845481049564e-06, | |
| "loss": 0.0062, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 11.07570977917981, | |
| "grad_norm": 0.49886855483055115, | |
| "learning_rate": 8.45262390670554e-06, | |
| "loss": 0.006, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 11.180862250262882, | |
| "grad_norm": 0.5618315935134888, | |
| "learning_rate": 8.434402332361516e-06, | |
| "loss": 0.0053, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 11.286014721345952, | |
| "grad_norm": 0.4988507032394409, | |
| "learning_rate": 8.416180758017494e-06, | |
| "loss": 0.006, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 11.391167192429021, | |
| "grad_norm": 0.4689825773239136, | |
| "learning_rate": 8.39795918367347e-06, | |
| "loss": 0.0059, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 11.496319663512093, | |
| "grad_norm": 0.637856662273407, | |
| "learning_rate": 8.379737609329447e-06, | |
| "loss": 0.0066, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 11.601472134595163, | |
| "grad_norm": 0.46745094656944275, | |
| "learning_rate": 8.361516034985424e-06, | |
| "loss": 0.007, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 11.706624605678233, | |
| "grad_norm": 1.153106689453125, | |
| "learning_rate": 8.3432944606414e-06, | |
| "loss": 0.0077, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 11.811777076761304, | |
| "grad_norm": 1.1093589067459106, | |
| "learning_rate": 8.325072886297377e-06, | |
| "loss": 0.0068, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 11.916929547844374, | |
| "grad_norm": 0.7349961400032043, | |
| "learning_rate": 8.306851311953353e-06, | |
| "loss": 0.0075, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 12.025236593059937, | |
| "grad_norm": 3.071120500564575, | |
| "learning_rate": 8.288629737609331e-06, | |
| "loss": 0.0067, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 12.130389064143007, | |
| "grad_norm": 1.33828604221344, | |
| "learning_rate": 8.270408163265307e-06, | |
| "loss": 0.0049, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 12.235541535226078, | |
| "grad_norm": 0.5961343050003052, | |
| "learning_rate": 8.252186588921283e-06, | |
| "loss": 0.0058, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 12.340694006309148, | |
| "grad_norm": 0.5771873593330383, | |
| "learning_rate": 8.23396501457726e-06, | |
| "loss": 0.0049, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 12.44584647739222, | |
| "grad_norm": 1.0507683753967285, | |
| "learning_rate": 8.215743440233237e-06, | |
| "loss": 0.0065, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 12.55099894847529, | |
| "grad_norm": 0.579768180847168, | |
| "learning_rate": 8.197521865889213e-06, | |
| "loss": 0.0058, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 12.656151419558359, | |
| "grad_norm": 0.8235365748405457, | |
| "learning_rate": 8.17930029154519e-06, | |
| "loss": 0.0075, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.656151419558359, | |
| "eval_loss": 0.11558376252651215, | |
| "eval_runtime": 1148.2367, | |
| "eval_samples_per_second": 2.838, | |
| "eval_steps_per_second": 0.71, | |
| "eval_wer": 0.09917594332803238, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.656151419558359, | |
| "step": 3000, | |
| "total_flos": 5.55469178535936e+19, | |
| "train_loss": 0.17092564110457897, | |
| "train_runtime": 92948.5763, | |
| "train_samples_per_second": 9.82, | |
| "train_steps_per_second": 0.153 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 14220, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 60, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 3 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.55469178535936e+19, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |