| { | |
| "best_metric": 36.4960604331718, | |
| "best_model_checkpoint": "./whisper-medium-ka/checkpoint-10000", | |
| "epoch": 2.5131942699170646, | |
| "eval_steps": 1000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02535496957403651, | |
| "grad_norm": 10.789830207824707, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 1.2488, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05070993914807302, | |
| "grad_norm": 6.402094841003418, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 0.9647, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07606490872210954, | |
| "grad_norm": 6.477363586425781, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.7572, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.10141987829614604, | |
| "grad_norm": 5.079893589019775, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.636, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12677484787018256, | |
| "grad_norm": 4.092740535736084, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.5668, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.15212981744421908, | |
| "grad_norm": 5.041500091552734, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.5045, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17748478701825557, | |
| "grad_norm": 5.795814514160156, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.4991, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.2028397565922921, | |
| "grad_norm": 5.117966651916504, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.4602, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2281947261663286, | |
| "grad_norm": 5.343120574951172, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.4458, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.2535496957403651, | |
| "grad_norm": 3.3774187564849854, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.4174, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.2789046653144016, | |
| "grad_norm": 3.576996326446533, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.4296, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.30425963488843816, | |
| "grad_norm": 3.852977991104126, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.3889, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.32961460446247465, | |
| "grad_norm": 4.133119583129883, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.3864, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.35496957403651114, | |
| "grad_norm": 4.041072368621826, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.3787, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.3803245436105477, | |
| "grad_norm": 4.2009806632995605, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.3722, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.4056795131845842, | |
| "grad_norm": 3.9817683696746826, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.366, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.43103448275862066, | |
| "grad_norm": 3.8101816177368164, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.3712, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.4563894523326572, | |
| "grad_norm": 3.7913081645965576, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.3743, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.4817444219066937, | |
| "grad_norm": 4.321682929992676, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.3467, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.5070993914807302, | |
| "grad_norm": 3.346144199371338, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.3429, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5324543610547667, | |
| "grad_norm": 4.033421516418457, | |
| "learning_rate": 9.975789473684211e-06, | |
| "loss": 0.3521, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.5578093306288032, | |
| "grad_norm": 4.106046676635742, | |
| "learning_rate": 9.949473684210526e-06, | |
| "loss": 0.3219, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.5831643002028397, | |
| "grad_norm": 3.8651669025421143, | |
| "learning_rate": 9.923157894736844e-06, | |
| "loss": 0.3323, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.6085192697768763, | |
| "grad_norm": 4.054211616516113, | |
| "learning_rate": 9.89684210526316e-06, | |
| "loss": 0.3443, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6338742393509128, | |
| "grad_norm": 3.2193450927734375, | |
| "learning_rate": 9.870526315789474e-06, | |
| "loss": 0.3159, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.6592292089249493, | |
| "grad_norm": 2.698641300201416, | |
| "learning_rate": 9.84421052631579e-06, | |
| "loss": 0.3197, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.6845841784989858, | |
| "grad_norm": 2.68990159034729, | |
| "learning_rate": 9.817894736842106e-06, | |
| "loss": 0.3151, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.7099391480730223, | |
| "grad_norm": 3.144531726837158, | |
| "learning_rate": 9.791578947368422e-06, | |
| "loss": 0.3094, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7352941176470589, | |
| "grad_norm": 3.1064047813415527, | |
| "learning_rate": 9.765263157894737e-06, | |
| "loss": 0.3026, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.7606490872210954, | |
| "grad_norm": 2.2833447456359863, | |
| "learning_rate": 9.738947368421054e-06, | |
| "loss": 0.3152, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.7860040567951319, | |
| "grad_norm": 2.9134676456451416, | |
| "learning_rate": 9.712631578947369e-06, | |
| "loss": 0.3018, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.8113590263691683, | |
| "grad_norm": 3.249375820159912, | |
| "learning_rate": 9.686315789473684e-06, | |
| "loss": 0.306, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.8367139959432048, | |
| "grad_norm": 3.1664018630981445, | |
| "learning_rate": 9.66e-06, | |
| "loss": 0.3053, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.8620689655172413, | |
| "grad_norm": 2.8798232078552246, | |
| "learning_rate": 9.633684210526316e-06, | |
| "loss": 0.2919, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.8874239350912779, | |
| "grad_norm": 3.306312322616577, | |
| "learning_rate": 9.607368421052632e-06, | |
| "loss": 0.3008, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.9127789046653144, | |
| "grad_norm": 3.241096258163452, | |
| "learning_rate": 9.581052631578947e-06, | |
| "loss": 0.3066, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9381338742393509, | |
| "grad_norm": 3.2305490970611572, | |
| "learning_rate": 9.554736842105264e-06, | |
| "loss": 0.293, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.9634888438133874, | |
| "grad_norm": 4.4702229499816895, | |
| "learning_rate": 9.52842105263158e-06, | |
| "loss": 0.2901, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.9888438133874239, | |
| "grad_norm": 3.1923646926879883, | |
| "learning_rate": 9.502105263157896e-06, | |
| "loss": 0.29, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.0141987829614605, | |
| "grad_norm": 2.5901284217834473, | |
| "learning_rate": 9.475789473684212e-06, | |
| "loss": 0.2611, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0141987829614605, | |
| "eval_loss": 0.2935636639595032, | |
| "eval_runtime": 7046.6021, | |
| "eval_samples_per_second": 0.69, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 62.766441678696374, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.039553752535497, | |
| "grad_norm": 2.9228506088256836, | |
| "learning_rate": 9.449473684210527e-06, | |
| "loss": 0.2488, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.0649087221095335, | |
| "grad_norm": 2.1438565254211426, | |
| "learning_rate": 9.423157894736842e-06, | |
| "loss": 0.2441, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.09026369168357, | |
| "grad_norm": 3.6437671184539795, | |
| "learning_rate": 9.396842105263159e-06, | |
| "loss": 0.2512, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.1156186612576064, | |
| "grad_norm": 2.413125991821289, | |
| "learning_rate": 9.370526315789474e-06, | |
| "loss": 0.2447, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.140973630831643, | |
| "grad_norm": 2.880699872970581, | |
| "learning_rate": 9.34421052631579e-06, | |
| "loss": 0.2463, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.1663286004056794, | |
| "grad_norm": 2.283259630203247, | |
| "learning_rate": 9.317894736842105e-06, | |
| "loss": 0.2488, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.1916835699797161, | |
| "grad_norm": 3.4118285179138184, | |
| "learning_rate": 9.291578947368422e-06, | |
| "loss": 0.2482, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.2170385395537526, | |
| "grad_norm": 2.1249563694000244, | |
| "learning_rate": 9.265263157894737e-06, | |
| "loss": 0.243, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.2423935091277891, | |
| "grad_norm": 2.6854898929595947, | |
| "learning_rate": 9.238947368421052e-06, | |
| "loss": 0.2563, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.2677484787018256, | |
| "grad_norm": 2.975567102432251, | |
| "learning_rate": 9.21263157894737e-06, | |
| "loss": 0.2529, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.293103448275862, | |
| "grad_norm": 2.6156153678894043, | |
| "learning_rate": 9.186315789473685e-06, | |
| "loss": 0.2378, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.3184584178498986, | |
| "grad_norm": 2.3254947662353516, | |
| "learning_rate": 9.16e-06, | |
| "loss": 0.2457, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.343813387423935, | |
| "grad_norm": 3.162607431411743, | |
| "learning_rate": 9.133684210526317e-06, | |
| "loss": 0.2452, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.3691683569979716, | |
| "grad_norm": 2.94197678565979, | |
| "learning_rate": 9.107368421052632e-06, | |
| "loss": 0.2411, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.394523326572008, | |
| "grad_norm": 2.53021502494812, | |
| "learning_rate": 9.081052631578949e-06, | |
| "loss": 0.2497, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.4198782961460445, | |
| "grad_norm": 2.740830421447754, | |
| "learning_rate": 9.054736842105264e-06, | |
| "loss": 0.228, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.445233265720081, | |
| "grad_norm": 3.1709723472595215, | |
| "learning_rate": 9.02842105263158e-06, | |
| "loss": 0.2598, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.4705882352941178, | |
| "grad_norm": 2.508091688156128, | |
| "learning_rate": 9.002105263157895e-06, | |
| "loss": 0.2508, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.495943204868154, | |
| "grad_norm": 2.7562551498413086, | |
| "learning_rate": 8.97578947368421e-06, | |
| "loss": 0.2462, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.5212981744421907, | |
| "grad_norm": 3.4288785457611084, | |
| "learning_rate": 8.949473684210527e-06, | |
| "loss": 0.2351, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.5466531440162272, | |
| "grad_norm": 2.4875781536102295, | |
| "learning_rate": 8.923157894736842e-06, | |
| "loss": 0.249, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.5720081135902637, | |
| "grad_norm": 2.5765998363494873, | |
| "learning_rate": 8.896842105263159e-06, | |
| "loss": 0.2354, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.5973630831643002, | |
| "grad_norm": 2.0511434078216553, | |
| "learning_rate": 8.870526315789474e-06, | |
| "loss": 0.248, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.6227180527383367, | |
| "grad_norm": 3.0492985248565674, | |
| "learning_rate": 8.84421052631579e-06, | |
| "loss": 0.2443, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.6480730223123732, | |
| "grad_norm": 3.98416805267334, | |
| "learning_rate": 8.817894736842107e-06, | |
| "loss": 0.2348, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.6734279918864097, | |
| "grad_norm": 2.4991466999053955, | |
| "learning_rate": 8.791578947368422e-06, | |
| "loss": 0.2514, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.6987829614604464, | |
| "grad_norm": 2.768167734146118, | |
| "learning_rate": 8.765263157894739e-06, | |
| "loss": 0.2517, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.7241379310344827, | |
| "grad_norm": 3.1816017627716064, | |
| "learning_rate": 8.738947368421053e-06, | |
| "loss": 0.2421, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.7494929006085194, | |
| "grad_norm": 3.7011590003967285, | |
| "learning_rate": 8.712631578947368e-06, | |
| "loss": 0.2376, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.7748478701825556, | |
| "grad_norm": 2.3581457138061523, | |
| "learning_rate": 8.686315789473685e-06, | |
| "loss": 0.2281, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.8002028397565923, | |
| "grad_norm": 2.812277317047119, | |
| "learning_rate": 8.66e-06, | |
| "loss": 0.2357, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.8255578093306288, | |
| "grad_norm": 3.1732356548309326, | |
| "learning_rate": 8.633684210526317e-06, | |
| "loss": 0.237, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.8509127789046653, | |
| "grad_norm": 1.9399900436401367, | |
| "learning_rate": 8.607368421052632e-06, | |
| "loss": 0.2395, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.8762677484787018, | |
| "grad_norm": 1.8843597173690796, | |
| "learning_rate": 8.581052631578948e-06, | |
| "loss": 0.2282, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.9016227180527383, | |
| "grad_norm": 2.314880132675171, | |
| "learning_rate": 8.554736842105263e-06, | |
| "loss": 0.2454, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.9269776876267748, | |
| "grad_norm": 4.113000392913818, | |
| "learning_rate": 8.528421052631578e-06, | |
| "loss": 0.2342, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.9523326572008113, | |
| "grad_norm": 2.076021432876587, | |
| "learning_rate": 8.502105263157897e-06, | |
| "loss": 0.2313, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.977687626774848, | |
| "grad_norm": 3.887740135192871, | |
| "learning_rate": 8.475789473684212e-06, | |
| "loss": 0.245, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.0030425963488843, | |
| "grad_norm": 2.359464406967163, | |
| "learning_rate": 8.449473684210527e-06, | |
| "loss": 0.2539, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.028397565922921, | |
| "grad_norm": 2.299802303314209, | |
| "learning_rate": 8.423157894736843e-06, | |
| "loss": 0.18, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.028397565922921, | |
| "eval_loss": 0.27181410789489746, | |
| "eval_runtime": 7074.1416, | |
| "eval_samples_per_second": 0.688, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 59.46705161718232, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.0537525354969572, | |
| "grad_norm": 4.299188613891602, | |
| "learning_rate": 8.396842105263158e-06, | |
| "loss": 0.1803, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.079107505070994, | |
| "grad_norm": 2.5189449787139893, | |
| "learning_rate": 8.370526315789475e-06, | |
| "loss": 0.1927, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.1044624746450302, | |
| "grad_norm": 2.5737998485565186, | |
| "learning_rate": 8.34421052631579e-06, | |
| "loss": 0.1818, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.129817444219067, | |
| "grad_norm": 3.098142623901367, | |
| "learning_rate": 8.317894736842107e-06, | |
| "loss": 0.1907, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.1551724137931036, | |
| "grad_norm": 1.9791940450668335, | |
| "learning_rate": 8.291578947368422e-06, | |
| "loss": 0.1809, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.18052738336714, | |
| "grad_norm": 1.9798074960708618, | |
| "learning_rate": 8.265263157894737e-06, | |
| "loss": 0.1842, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.2058823529411766, | |
| "grad_norm": 2.5827345848083496, | |
| "learning_rate": 8.238947368421053e-06, | |
| "loss": 0.1769, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.231237322515213, | |
| "grad_norm": 2.1781139373779297, | |
| "learning_rate": 8.212631578947368e-06, | |
| "loss": 0.1904, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.2565922920892496, | |
| "grad_norm": 2.2573533058166504, | |
| "learning_rate": 8.186315789473685e-06, | |
| "loss": 0.1892, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.281947261663286, | |
| "grad_norm": 2.1169838905334473, | |
| "learning_rate": 8.16e-06, | |
| "loss": 0.1868, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.3073022312373226, | |
| "grad_norm": 3.3498687744140625, | |
| "learning_rate": 8.133684210526316e-06, | |
| "loss": 0.197, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.332657200811359, | |
| "grad_norm": 2.25311541557312, | |
| "learning_rate": 8.107368421052633e-06, | |
| "loss": 0.1869, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.3580121703853956, | |
| "grad_norm": 2.396975040435791, | |
| "learning_rate": 8.081052631578948e-06, | |
| "loss": 0.189, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.3833671399594323, | |
| "grad_norm": 2.147547960281372, | |
| "learning_rate": 8.054736842105265e-06, | |
| "loss": 0.1893, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.4087221095334685, | |
| "grad_norm": 1.9963374137878418, | |
| "learning_rate": 8.02842105263158e-06, | |
| "loss": 0.2003, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.4340770791075053, | |
| "grad_norm": 3.571808099746704, | |
| "learning_rate": 8.002105263157895e-06, | |
| "loss": 0.1904, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.4594320486815415, | |
| "grad_norm": 2.6122543811798096, | |
| "learning_rate": 7.975789473684211e-06, | |
| "loss": 0.1855, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.4847870182555782, | |
| "grad_norm": 2.9478933811187744, | |
| "learning_rate": 7.949473684210526e-06, | |
| "loss": 0.1822, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.5101419878296145, | |
| "grad_norm": 2.368997097015381, | |
| "learning_rate": 7.923157894736843e-06, | |
| "loss": 0.186, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.535496957403651, | |
| "grad_norm": 2.529630661010742, | |
| "learning_rate": 7.896842105263158e-06, | |
| "loss": 0.191, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.5608519269776875, | |
| "grad_norm": 2.9214324951171875, | |
| "learning_rate": 7.870526315789475e-06, | |
| "loss": 0.1885, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.586206896551724, | |
| "grad_norm": 2.141688823699951, | |
| "learning_rate": 7.84421052631579e-06, | |
| "loss": 0.1847, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.6115618661257605, | |
| "grad_norm": 2.1608083248138428, | |
| "learning_rate": 7.817894736842105e-06, | |
| "loss": 0.1873, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.636916835699797, | |
| "grad_norm": 2.489492416381836, | |
| "learning_rate": 7.791578947368423e-06, | |
| "loss": 0.1959, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.662271805273834, | |
| "grad_norm": 2.31791353225708, | |
| "learning_rate": 7.765263157894738e-06, | |
| "loss": 0.1915, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.68762677484787, | |
| "grad_norm": 2.779777765274048, | |
| "learning_rate": 7.738947368421053e-06, | |
| "loss": 0.1878, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.7129817444219064, | |
| "grad_norm": 2.5656425952911377, | |
| "learning_rate": 7.71263157894737e-06, | |
| "loss": 0.1925, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.738336713995943, | |
| "grad_norm": 3.0531349182128906, | |
| "learning_rate": 7.686315789473685e-06, | |
| "loss": 0.189, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.76369168356998, | |
| "grad_norm": 2.912122964859009, | |
| "learning_rate": 7.660000000000001e-06, | |
| "loss": 0.1906, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.789046653144016, | |
| "grad_norm": 3.0555505752563477, | |
| "learning_rate": 7.633684210526316e-06, | |
| "loss": 0.1834, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.814401622718053, | |
| "grad_norm": 2.2603671550750732, | |
| "learning_rate": 7.607368421052632e-06, | |
| "loss": 0.1872, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.839756592292089, | |
| "grad_norm": 2.1887290477752686, | |
| "learning_rate": 7.581052631578948e-06, | |
| "loss": 0.1787, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.865111561866126, | |
| "grad_norm": 2.393935203552246, | |
| "learning_rate": 7.554736842105264e-06, | |
| "loss": 0.1883, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.890466531440162, | |
| "grad_norm": 2.3543920516967773, | |
| "learning_rate": 7.5284210526315794e-06, | |
| "loss": 0.1883, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.915821501014199, | |
| "grad_norm": 2.139833927154541, | |
| "learning_rate": 7.502105263157895e-06, | |
| "loss": 0.1997, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.9411764705882355, | |
| "grad_norm": 2.270552158355713, | |
| "learning_rate": 7.475789473684211e-06, | |
| "loss": 0.1932, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.9665314401622718, | |
| "grad_norm": 2.4720232486724854, | |
| "learning_rate": 7.449473684210526e-06, | |
| "loss": 0.1926, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.991886409736308, | |
| "grad_norm": 2.318767786026001, | |
| "learning_rate": 7.4231578947368436e-06, | |
| "loss": 0.1837, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.0172413793103448, | |
| "grad_norm": 1.8572547435760498, | |
| "learning_rate": 7.3968421052631585e-06, | |
| "loss": 0.1584, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 3.0425963488843815, | |
| "grad_norm": 2.6684017181396484, | |
| "learning_rate": 7.370526315789474e-06, | |
| "loss": 0.15, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.0425963488843815, | |
| "eval_loss": 0.2813817858695984, | |
| "eval_runtime": 7104.648, | |
| "eval_samples_per_second": 0.685, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 59.24206364688593, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.760241266649912, | |
| "grad_norm": 2.2316994667053223, | |
| "learning_rate": 8.26e-06, | |
| "loss": 0.1649, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.7665242523247047, | |
| "grad_norm": 1.9445712566375732, | |
| "learning_rate": 8.242758620689655e-06, | |
| "loss": 0.1373, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.7728072379994974, | |
| "grad_norm": 1.9695370197296143, | |
| "learning_rate": 8.22551724137931e-06, | |
| "loss": 0.1303, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.77909022367429, | |
| "grad_norm": 2.19242262840271, | |
| "learning_rate": 8.208275862068967e-06, | |
| "loss": 0.1204, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.7853732093490827, | |
| "grad_norm": 1.6107076406478882, | |
| "learning_rate": 8.191034482758622e-06, | |
| "loss": 0.1206, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.7916561950238753, | |
| "grad_norm": 1.8804924488067627, | |
| "learning_rate": 8.173793103448277e-06, | |
| "loss": 0.116, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.797939180698668, | |
| "grad_norm": 1.8132656812667847, | |
| "learning_rate": 8.156551724137931e-06, | |
| "loss": 0.1193, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.8042221663734607, | |
| "grad_norm": 1.9557982683181763, | |
| "learning_rate": 8.139310344827586e-06, | |
| "loss": 0.1139, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.8105051520482534, | |
| "grad_norm": 1.4428989887237549, | |
| "learning_rate": 8.122068965517243e-06, | |
| "loss": 0.1033, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.816788137723046, | |
| "grad_norm": 1.4548839330673218, | |
| "learning_rate": 8.104827586206898e-06, | |
| "loss": 0.0975, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.8230711233978386, | |
| "grad_norm": 1.8930418491363525, | |
| "learning_rate": 8.087586206896553e-06, | |
| "loss": 0.0997, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.8293541090726313, | |
| "grad_norm": 1.905731439590454, | |
| "learning_rate": 8.070344827586207e-06, | |
| "loss": 0.1069, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.835637094747424, | |
| "grad_norm": 1.7036992311477661, | |
| "learning_rate": 8.053103448275862e-06, | |
| "loss": 0.0921, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.8419200804222167, | |
| "grad_norm": 1.7952214479446411, | |
| "learning_rate": 8.035862068965517e-06, | |
| "loss": 0.1027, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.8482030660970094, | |
| "grad_norm": 2.0625522136688232, | |
| "learning_rate": 8.018620689655174e-06, | |
| "loss": 0.1051, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.8544860517718019, | |
| "grad_norm": 1.8595600128173828, | |
| "learning_rate": 8.001379310344829e-06, | |
| "loss": 0.0968, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.8607690374465946, | |
| "grad_norm": 1.5569204092025757, | |
| "learning_rate": 7.984137931034484e-06, | |
| "loss": 0.099, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.8670520231213873, | |
| "grad_norm": 1.6187876462936401, | |
| "learning_rate": 7.966896551724138e-06, | |
| "loss": 0.0942, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.87333500879618, | |
| "grad_norm": 1.494591474533081, | |
| "learning_rate": 7.949655172413793e-06, | |
| "loss": 0.096, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.8796179944709726, | |
| "grad_norm": 1.5578557252883911, | |
| "learning_rate": 7.932413793103448e-06, | |
| "loss": 0.0885, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.8859009801457652, | |
| "grad_norm": 2.047339677810669, | |
| "learning_rate": 7.915172413793105e-06, | |
| "loss": 0.098, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.8921839658205579, | |
| "grad_norm": 1.7731865644454956, | |
| "learning_rate": 7.89793103448276e-06, | |
| "loss": 0.0889, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.8984669514953506, | |
| "grad_norm": 2.1644279956817627, | |
| "learning_rate": 7.880689655172414e-06, | |
| "loss": 0.0864, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.9047499371701433, | |
| "grad_norm": 1.8717072010040283, | |
| "learning_rate": 7.86344827586207e-06, | |
| "loss": 0.0847, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.9110329228449359, | |
| "grad_norm": 1.481933832168579, | |
| "learning_rate": 7.846206896551724e-06, | |
| "loss": 0.0892, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.9173159085197286, | |
| "grad_norm": 2.3373663425445557, | |
| "learning_rate": 7.82896551724138e-06, | |
| "loss": 0.0904, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.9235988941945212, | |
| "grad_norm": 1.509282112121582, | |
| "learning_rate": 7.811724137931036e-06, | |
| "loss": 0.0973, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.9298818798693139, | |
| "grad_norm": 1.7900352478027344, | |
| "learning_rate": 7.79448275862069e-06, | |
| "loss": 0.0815, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.9361648655441066, | |
| "grad_norm": 1.6436471939086914, | |
| "learning_rate": 7.777241379310345e-06, | |
| "loss": 0.0853, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.9424478512188992, | |
| "grad_norm": 1.4675796031951904, | |
| "learning_rate": 7.76e-06, | |
| "loss": 0.0876, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.9487308368936919, | |
| "grad_norm": 1.8452798128128052, | |
| "learning_rate": 7.742758620689655e-06, | |
| "loss": 0.0808, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.9550138225684845, | |
| "grad_norm": 1.3618487119674683, | |
| "learning_rate": 7.725517241379312e-06, | |
| "loss": 0.086, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.9612968082432772, | |
| "grad_norm": 1.2715941667556763, | |
| "learning_rate": 7.708275862068967e-06, | |
| "loss": 0.088, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.9675797939180699, | |
| "grad_norm": 1.8105791807174683, | |
| "learning_rate": 7.691034482758621e-06, | |
| "loss": 0.0899, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.9738627795928625, | |
| "grad_norm": 2.1864423751831055, | |
| "learning_rate": 7.673793103448276e-06, | |
| "loss": 0.0857, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.9801457652676552, | |
| "grad_norm": 1.5500792264938354, | |
| "learning_rate": 7.656551724137931e-06, | |
| "loss": 0.0859, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.9864287509424479, | |
| "grad_norm": 1.8707691431045532, | |
| "learning_rate": 7.639310344827588e-06, | |
| "loss": 0.0882, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.9927117366172405, | |
| "grad_norm": 1.5884675979614258, | |
| "learning_rate": 7.622068965517242e-06, | |
| "loss": 0.0832, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.9989947222920331, | |
| "grad_norm": 1.444044828414917, | |
| "learning_rate": 7.6048275862068975e-06, | |
| "loss": 0.0786, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.0052777079668258, | |
| "grad_norm": 1.6613413095474243, | |
| "learning_rate": 7.588275862068966e-06, | |
| "loss": 0.0694, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.0052777079668258, | |
| "eval_loss": 0.13177034258842468, | |
| "eval_runtime": 28232.4806, | |
| "eval_samples_per_second": 0.692, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 40.84150014082778, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.0115606936416186, | |
| "grad_norm": 1.5078247785568237, | |
| "learning_rate": 7.571034482758622e-06, | |
| "loss": 0.0786, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.0178436793164112, | |
| "grad_norm": 2.1607069969177246, | |
| "learning_rate": 7.553793103448277e-06, | |
| "loss": 0.0748, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.0241266649912038, | |
| "grad_norm": 1.3016276359558105, | |
| "learning_rate": 7.5365517241379315e-06, | |
| "loss": 0.0782, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.0304096506659965, | |
| "grad_norm": 1.1447079181671143, | |
| "learning_rate": 7.519310344827587e-06, | |
| "loss": 0.0708, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.0366926363407891, | |
| "grad_norm": 1.6340376138687134, | |
| "learning_rate": 7.502068965517242e-06, | |
| "loss": 0.0705, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.042975622015582, | |
| "grad_norm": 1.5793544054031372, | |
| "learning_rate": 7.484827586206898e-06, | |
| "loss": 0.0723, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.0492586076903745, | |
| "grad_norm": 1.2882245779037476, | |
| "learning_rate": 7.467586206896552e-06, | |
| "loss": 0.0692, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.055541593365167, | |
| "grad_norm": 1.822688102722168, | |
| "learning_rate": 7.4503448275862075e-06, | |
| "loss": 0.071, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.0618245790399599, | |
| "grad_norm": 1.410294771194458, | |
| "learning_rate": 7.433103448275862e-06, | |
| "loss": 0.0666, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.0681075647147524, | |
| "grad_norm": 1.4143624305725098, | |
| "learning_rate": 7.415862068965518e-06, | |
| "loss": 0.0662, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.0743905503895452, | |
| "grad_norm": 1.5564229488372803, | |
| "learning_rate": 7.398620689655173e-06, | |
| "loss": 0.0752, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.0806735360643378, | |
| "grad_norm": 2.0174007415771484, | |
| "learning_rate": 7.381379310344829e-06, | |
| "loss": 0.0647, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.0869565217391304, | |
| "grad_norm": 1.2202295064926147, | |
| "learning_rate": 7.364137931034483e-06, | |
| "loss": 0.0728, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.0932395074139232, | |
| "grad_norm": 1.7252171039581299, | |
| "learning_rate": 7.346896551724138e-06, | |
| "loss": 0.0613, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.0995224930887157, | |
| "grad_norm": 1.6477744579315186, | |
| "learning_rate": 7.329655172413793e-06, | |
| "loss": 0.0762, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.1058054787635083, | |
| "grad_norm": 2.01273512840271, | |
| "learning_rate": 7.312413793103449e-06, | |
| "loss": 0.0723, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.1120884644383011, | |
| "grad_norm": 2.0177805423736572, | |
| "learning_rate": 7.295172413793105e-06, | |
| "loss": 0.0713, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.1183714501130937, | |
| "grad_norm": 1.0823686122894287, | |
| "learning_rate": 7.27793103448276e-06, | |
| "loss": 0.0699, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.1246544357878865, | |
| "grad_norm": 1.8306968212127686, | |
| "learning_rate": 7.2606896551724145e-06, | |
| "loss": 0.0692, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.130937421462679, | |
| "grad_norm": 1.9577009677886963, | |
| "learning_rate": 7.243448275862069e-06, | |
| "loss": 0.0707, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.1372204071374716, | |
| "grad_norm": 1.8456298112869263, | |
| "learning_rate": 7.226206896551725e-06, | |
| "loss": 0.0702, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.1435033928122644, | |
| "grad_norm": 1.369918942451477, | |
| "learning_rate": 7.20896551724138e-06, | |
| "loss": 0.0716, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.149786378487057, | |
| "grad_norm": 1.7100856304168701, | |
| "learning_rate": 7.191724137931036e-06, | |
| "loss": 0.0773, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.1560693641618498, | |
| "grad_norm": 1.2440359592437744, | |
| "learning_rate": 7.17448275862069e-06, | |
| "loss": 0.0736, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.1623523498366424, | |
| "grad_norm": 1.26316237449646, | |
| "learning_rate": 7.157241379310345e-06, | |
| "loss": 0.0683, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.168635335511435, | |
| "grad_norm": 1.6966075897216797, | |
| "learning_rate": 7.14e-06, | |
| "loss": 0.064, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.1749183211862277, | |
| "grad_norm": 1.7377158403396606, | |
| "learning_rate": 7.122758620689656e-06, | |
| "loss": 0.0685, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.1812013068610203, | |
| "grad_norm": 1.851913571357727, | |
| "learning_rate": 7.105517241379311e-06, | |
| "loss": 0.0586, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.1874842925358131, | |
| "grad_norm": 1.6973158121109009, | |
| "learning_rate": 7.0882758620689666e-06, | |
| "loss": 0.0696, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.1937672782106057, | |
| "grad_norm": 1.451185703277588, | |
| "learning_rate": 7.0710344827586206e-06, | |
| "loss": 0.0671, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.2000502638853983, | |
| "grad_norm": 1.7117061614990234, | |
| "learning_rate": 7.053793103448276e-06, | |
| "loss": 0.0563, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.206333249560191, | |
| "grad_norm": 1.4349240064620972, | |
| "learning_rate": 7.036551724137931e-06, | |
| "loss": 0.0696, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.2126162352349836, | |
| "grad_norm": 1.5705769062042236, | |
| "learning_rate": 7.019310344827587e-06, | |
| "loss": 0.0634, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.2188992209097762, | |
| "grad_norm": 1.5555791854858398, | |
| "learning_rate": 7.002068965517243e-06, | |
| "loss": 0.0665, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.225182206584569, | |
| "grad_norm": 1.4251762628555298, | |
| "learning_rate": 6.9848275862068975e-06, | |
| "loss": 0.0636, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.2314651922593616, | |
| "grad_norm": 1.7479225397109985, | |
| "learning_rate": 6.967586206896552e-06, | |
| "loss": 0.0736, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.2377481779341544, | |
| "grad_norm": 1.4017658233642578, | |
| "learning_rate": 6.950344827586207e-06, | |
| "loss": 0.0642, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.244031163608947, | |
| "grad_norm": 1.34666907787323, | |
| "learning_rate": 6.933103448275863e-06, | |
| "loss": 0.0647, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.2503141492837395, | |
| "grad_norm": 1.6432591676712036, | |
| "learning_rate": 6.915862068965518e-06, | |
| "loss": 0.069, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.2565971349585323, | |
| "grad_norm": 1.2850168943405151, | |
| "learning_rate": 6.8986206896551735e-06, | |
| "loss": 0.0638, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.2565971349585323, | |
| "eval_loss": 0.12251746654510498, | |
| "eval_runtime": 28262.6172, | |
| "eval_samples_per_second": 0.691, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 38.8171910996844, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.2628801206333249, | |
| "grad_norm": 1.6795750856399536, | |
| "learning_rate": 6.8813793103448275e-06, | |
| "loss": 0.069, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 1.2691631063081177, | |
| "grad_norm": 1.418845772743225, | |
| "learning_rate": 6.864137931034483e-06, | |
| "loss": 0.0609, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.2754460919829103, | |
| "grad_norm": 1.2165697813034058, | |
| "learning_rate": 6.846896551724138e-06, | |
| "loss": 0.0638, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 1.2817290776577028, | |
| "grad_norm": 2.068269729614258, | |
| "learning_rate": 6.829655172413794e-06, | |
| "loss": 0.0673, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.2880120633324956, | |
| "grad_norm": 1.2388827800750732, | |
| "learning_rate": 6.812413793103449e-06, | |
| "loss": 0.0701, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 1.2942950490072882, | |
| "grad_norm": 1.2347630262374878, | |
| "learning_rate": 6.7951724137931044e-06, | |
| "loss": 0.065, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.300578034682081, | |
| "grad_norm": 1.15070378780365, | |
| "learning_rate": 6.7779310344827585e-06, | |
| "loss": 0.0637, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 1.3068610203568736, | |
| "grad_norm": 1.2261390686035156, | |
| "learning_rate": 6.760689655172414e-06, | |
| "loss": 0.0676, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.3131440060316661, | |
| "grad_norm": 1.9254522323608398, | |
| "learning_rate": 6.74344827586207e-06, | |
| "loss": 0.0594, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 1.319426991706459, | |
| "grad_norm": 1.3794054985046387, | |
| "learning_rate": 6.726206896551725e-06, | |
| "loss": 0.0649, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.3257099773812515, | |
| "grad_norm": 1.6740128993988037, | |
| "learning_rate": 6.7089655172413805e-06, | |
| "loss": 0.0619, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 1.3319929630560443, | |
| "grad_norm": 1.2363388538360596, | |
| "learning_rate": 6.691724137931035e-06, | |
| "loss": 0.0646, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.3382759487308369, | |
| "grad_norm": 1.433228850364685, | |
| "learning_rate": 6.67448275862069e-06, | |
| "loss": 0.0663, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 1.3445589344056295, | |
| "grad_norm": 1.898812174797058, | |
| "learning_rate": 6.657241379310345e-06, | |
| "loss": 0.0574, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.3508419200804223, | |
| "grad_norm": 1.3165233135223389, | |
| "learning_rate": 6.640000000000001e-06, | |
| "loss": 0.0597, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 1.3571249057552148, | |
| "grad_norm": 1.8794306516647339, | |
| "learning_rate": 6.622758620689656e-06, | |
| "loss": 0.0697, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.3634078914300076, | |
| "grad_norm": 1.1872018575668335, | |
| "learning_rate": 6.605517241379311e-06, | |
| "loss": 0.0587, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 1.3696908771048002, | |
| "grad_norm": 1.639711856842041, | |
| "learning_rate": 6.588275862068965e-06, | |
| "loss": 0.0645, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.3759738627795928, | |
| "grad_norm": 1.2173725366592407, | |
| "learning_rate": 6.571034482758621e-06, | |
| "loss": 0.055, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 1.3822568484543856, | |
| "grad_norm": 1.2602193355560303, | |
| "learning_rate": 6.553793103448276e-06, | |
| "loss": 0.0543, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.3885398341291781, | |
| "grad_norm": 1.350376009941101, | |
| "learning_rate": 6.536551724137932e-06, | |
| "loss": 0.0675, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 1.394822819803971, | |
| "grad_norm": 1.314760446548462, | |
| "learning_rate": 6.519310344827587e-06, | |
| "loss": 0.0597, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.4011058054787635, | |
| "grad_norm": 1.0545654296875, | |
| "learning_rate": 6.502068965517242e-06, | |
| "loss": 0.0553, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 1.407388791153556, | |
| "grad_norm": 1.2332383394241333, | |
| "learning_rate": 6.484827586206896e-06, | |
| "loss": 0.0607, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.4136717768283489, | |
| "grad_norm": 1.3052699565887451, | |
| "learning_rate": 6.467586206896552e-06, | |
| "loss": 0.0689, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 1.4199547625031415, | |
| "grad_norm": 1.505159854888916, | |
| "learning_rate": 6.450344827586208e-06, | |
| "loss": 0.0585, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.4262377481779343, | |
| "grad_norm": 1.2076035737991333, | |
| "learning_rate": 6.433103448275863e-06, | |
| "loss": 0.0624, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 1.4325207338527268, | |
| "grad_norm": 0.716097354888916, | |
| "learning_rate": 6.415862068965518e-06, | |
| "loss": 0.056, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.4388037195275194, | |
| "grad_norm": 0.8870618939399719, | |
| "learning_rate": 6.398620689655173e-06, | |
| "loss": 0.0592, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 1.4450867052023122, | |
| "grad_norm": 2.086239814758301, | |
| "learning_rate": 6.381379310344828e-06, | |
| "loss": 0.0647, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.4513696908771048, | |
| "grad_norm": 1.2493882179260254, | |
| "learning_rate": 6.364137931034483e-06, | |
| "loss": 0.0664, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 1.4576526765518976, | |
| "grad_norm": 0.78863924741745, | |
| "learning_rate": 6.346896551724139e-06, | |
| "loss": 0.0545, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.4639356622266901, | |
| "grad_norm": 1.2974257469177246, | |
| "learning_rate": 6.3296551724137935e-06, | |
| "loss": 0.0643, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 1.4702186479014827, | |
| "grad_norm": 1.220800518989563, | |
| "learning_rate": 6.312413793103449e-06, | |
| "loss": 0.0664, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.4765016335762755, | |
| "grad_norm": 1.9717583656311035, | |
| "learning_rate": 6.295172413793103e-06, | |
| "loss": 0.0585, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 1.482784619251068, | |
| "grad_norm": 1.9224556684494019, | |
| "learning_rate": 6.277931034482759e-06, | |
| "loss": 0.0607, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.4890676049258609, | |
| "grad_norm": 1.3678847551345825, | |
| "learning_rate": 6.260689655172414e-06, | |
| "loss": 0.0619, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 1.4953505906006535, | |
| "grad_norm": 1.2710736989974976, | |
| "learning_rate": 6.24344827586207e-06, | |
| "loss": 0.0615, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.501633576275446, | |
| "grad_norm": 1.1889654397964478, | |
| "learning_rate": 6.2262068965517245e-06, | |
| "loss": 0.0595, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 1.5079165619502386, | |
| "grad_norm": 1.165711760520935, | |
| "learning_rate": 6.20896551724138e-06, | |
| "loss": 0.0529, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.5079165619502386, | |
| "eval_loss": 0.12242772430181503, | |
| "eval_runtime": 28299.0323, | |
| "eval_samples_per_second": 0.69, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 37.89097763366001, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.5141995476250314, | |
| "grad_norm": 1.0195108652114868, | |
| "learning_rate": 6.191724137931034e-06, | |
| "loss": 0.0586, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 1.5204825332998242, | |
| "grad_norm": 1.1056115627288818, | |
| "learning_rate": 6.17448275862069e-06, | |
| "loss": 0.0576, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.5267655189746168, | |
| "grad_norm": 1.2018024921417236, | |
| "learning_rate": 6.157241379310346e-06, | |
| "loss": 0.0641, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 1.5330485046494093, | |
| "grad_norm": 1.426589846611023, | |
| "learning_rate": 6.1400000000000005e-06, | |
| "loss": 0.0575, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.539331490324202, | |
| "grad_norm": 1.0339651107788086, | |
| "learning_rate": 6.122758620689656e-06, | |
| "loss": 0.0459, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 1.5456144759989947, | |
| "grad_norm": 1.3441721200942993, | |
| "learning_rate": 6.105517241379311e-06, | |
| "loss": 0.0588, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 1.5518974616737875, | |
| "grad_norm": 1.2048940658569336, | |
| "learning_rate": 6.088275862068966e-06, | |
| "loss": 0.0573, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 1.55818044734858, | |
| "grad_norm": 1.5876215696334839, | |
| "learning_rate": 6.071034482758621e-06, | |
| "loss": 0.0616, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.5644634330233727, | |
| "grad_norm": 1.1515843868255615, | |
| "learning_rate": 6.0537931034482766e-06, | |
| "loss": 0.0566, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 1.5707464186981652, | |
| "grad_norm": 1.2697322368621826, | |
| "learning_rate": 6.036551724137931e-06, | |
| "loss": 0.0654, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.577029404372958, | |
| "grad_norm": 0.8662827014923096, | |
| "learning_rate": 6.019310344827587e-06, | |
| "loss": 0.0568, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 1.5833123900477508, | |
| "grad_norm": 1.5702407360076904, | |
| "learning_rate": 6.002068965517241e-06, | |
| "loss": 0.0563, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.5895953757225434, | |
| "grad_norm": 1.2121763229370117, | |
| "learning_rate": 5.984827586206897e-06, | |
| "loss": 0.0638, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 1.595878361397336, | |
| "grad_norm": 1.257488489151001, | |
| "learning_rate": 5.967586206896552e-06, | |
| "loss": 0.0577, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 1.6021613470721285, | |
| "grad_norm": 1.1616463661193848, | |
| "learning_rate": 5.9503448275862075e-06, | |
| "loss": 0.0604, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 1.6084443327469213, | |
| "grad_norm": 1.3494690656661987, | |
| "learning_rate": 5.933103448275862e-06, | |
| "loss": 0.0628, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.6147273184217141, | |
| "grad_norm": 1.5194650888442993, | |
| "learning_rate": 5.915862068965518e-06, | |
| "loss": 0.0549, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 1.6210103040965067, | |
| "grad_norm": 1.3437527418136597, | |
| "learning_rate": 5.898620689655174e-06, | |
| "loss": 0.0579, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 1.6272932897712993, | |
| "grad_norm": 1.445145845413208, | |
| "learning_rate": 5.881379310344828e-06, | |
| "loss": 0.0619, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 1.6335762754460919, | |
| "grad_norm": 1.3654954433441162, | |
| "learning_rate": 5.8641379310344835e-06, | |
| "loss": 0.0572, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.6398592611208846, | |
| "grad_norm": 1.4988460540771484, | |
| "learning_rate": 5.846896551724138e-06, | |
| "loss": 0.0599, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 1.6461422467956774, | |
| "grad_norm": 1.3341230154037476, | |
| "learning_rate": 5.829655172413794e-06, | |
| "loss": 0.0555, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 1.65242523247047, | |
| "grad_norm": 1.2729185819625854, | |
| "learning_rate": 5.812413793103449e-06, | |
| "loss": 0.0542, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 1.6587082181452626, | |
| "grad_norm": 1.7550246715545654, | |
| "learning_rate": 5.795172413793104e-06, | |
| "loss": 0.0582, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.6649912038200552, | |
| "grad_norm": 1.5241115093231201, | |
| "learning_rate": 5.777931034482759e-06, | |
| "loss": 0.063, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 1.671274189494848, | |
| "grad_norm": 1.8420275449752808, | |
| "learning_rate": 5.7606896551724144e-06, | |
| "loss": 0.0656, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 1.6775571751696408, | |
| "grad_norm": 1.0507006645202637, | |
| "learning_rate": 5.743448275862069e-06, | |
| "loss": 0.0562, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 1.6838401608444333, | |
| "grad_norm": 2.0767569541931152, | |
| "learning_rate": 5.726206896551725e-06, | |
| "loss": 0.0582, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.690123146519226, | |
| "grad_norm": 0.8954183459281921, | |
| "learning_rate": 5.708965517241379e-06, | |
| "loss": 0.0602, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 1.6964061321940185, | |
| "grad_norm": 0.9078446626663208, | |
| "learning_rate": 5.691724137931035e-06, | |
| "loss": 0.0529, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.7026891178688113, | |
| "grad_norm": 1.581921935081482, | |
| "learning_rate": 5.67448275862069e-06, | |
| "loss": 0.0548, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 1.708972103543604, | |
| "grad_norm": 1.4554569721221924, | |
| "learning_rate": 5.657241379310345e-06, | |
| "loss": 0.0563, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.7152550892183966, | |
| "grad_norm": 0.9179530739784241, | |
| "learning_rate": 5.64e-06, | |
| "loss": 0.0544, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 1.7215380748931892, | |
| "grad_norm": 1.1374155282974243, | |
| "learning_rate": 5.622758620689656e-06, | |
| "loss": 0.057, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 1.7278210605679818, | |
| "grad_norm": 1.0349596738815308, | |
| "learning_rate": 5.605517241379312e-06, | |
| "loss": 0.0573, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 1.7341040462427746, | |
| "grad_norm": 1.4229092597961426, | |
| "learning_rate": 5.588275862068966e-06, | |
| "loss": 0.0487, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.7403870319175674, | |
| "grad_norm": 1.2893837690353394, | |
| "learning_rate": 5.571034482758621e-06, | |
| "loss": 0.0605, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 1.74667001759236, | |
| "grad_norm": 1.1475664377212524, | |
| "learning_rate": 5.553793103448276e-06, | |
| "loss": 0.0577, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 1.7529530032671525, | |
| "grad_norm": 1.1052597761154175, | |
| "learning_rate": 5.536551724137932e-06, | |
| "loss": 0.0531, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 1.759235988941945, | |
| "grad_norm": 1.0279254913330078, | |
| "learning_rate": 5.519310344827587e-06, | |
| "loss": 0.0624, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.759235988941945, | |
| "eval_loss": 0.12031704932451248, | |
| "eval_runtime": 28303.9606, | |
| "eval_samples_per_second": 0.69, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 37.24605862768746, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.765518974616738, | |
| "grad_norm": 1.0434989929199219, | |
| "learning_rate": 5.502068965517242e-06, | |
| "loss": 0.0488, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 1.7718019602915307, | |
| "grad_norm": 1.1990073919296265, | |
| "learning_rate": 5.484827586206897e-06, | |
| "loss": 0.0525, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 1.7780849459663233, | |
| "grad_norm": 1.026079535484314, | |
| "learning_rate": 5.467586206896552e-06, | |
| "loss": 0.0531, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 1.7843679316411158, | |
| "grad_norm": 0.9900615215301514, | |
| "learning_rate": 5.450344827586207e-06, | |
| "loss": 0.0545, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.7906509173159084, | |
| "grad_norm": 1.5279738903045654, | |
| "learning_rate": 5.433103448275863e-06, | |
| "loss": 0.0566, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 1.7969339029907012, | |
| "grad_norm": 0.8226191401481628, | |
| "learning_rate": 5.415862068965517e-06, | |
| "loss": 0.056, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 1.8032168886654938, | |
| "grad_norm": 1.3544007539749146, | |
| "learning_rate": 5.398620689655173e-06, | |
| "loss": 0.061, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 1.8094998743402866, | |
| "grad_norm": 1.2771939039230347, | |
| "learning_rate": 5.3813793103448275e-06, | |
| "loss": 0.0542, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.8157828600150792, | |
| "grad_norm": 1.200951099395752, | |
| "learning_rate": 5.364137931034483e-06, | |
| "loss": 0.0544, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 1.8220658456898717, | |
| "grad_norm": 0.9072504639625549, | |
| "learning_rate": 5.346896551724139e-06, | |
| "loss": 0.0538, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.8283488313646645, | |
| "grad_norm": 1.088958978652954, | |
| "learning_rate": 5.329655172413794e-06, | |
| "loss": 0.054, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 1.834631817039457, | |
| "grad_norm": 1.359937071800232, | |
| "learning_rate": 5.3124137931034495e-06, | |
| "loss": 0.0623, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.84091480271425, | |
| "grad_norm": 1.467264175415039, | |
| "learning_rate": 5.2951724137931035e-06, | |
| "loss": 0.0536, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 1.8471977883890425, | |
| "grad_norm": 1.4082632064819336, | |
| "learning_rate": 5.277931034482759e-06, | |
| "loss": 0.051, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 1.853480774063835, | |
| "grad_norm": 1.472396969795227, | |
| "learning_rate": 5.260689655172414e-06, | |
| "loss": 0.0662, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 1.8597637597386278, | |
| "grad_norm": 1.587661862373352, | |
| "learning_rate": 5.24344827586207e-06, | |
| "loss": 0.055, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.8660467454134204, | |
| "grad_norm": 1.0919044017791748, | |
| "learning_rate": 5.226206896551725e-06, | |
| "loss": 0.0509, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 1.8723297310882132, | |
| "grad_norm": 0.9456779956817627, | |
| "learning_rate": 5.20896551724138e-06, | |
| "loss": 0.0557, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 1.8786127167630058, | |
| "grad_norm": 1.6846345663070679, | |
| "learning_rate": 5.1917241379310345e-06, | |
| "loss": 0.0531, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 1.8848957024377984, | |
| "grad_norm": 0.5773513317108154, | |
| "learning_rate": 5.17448275862069e-06, | |
| "loss": 0.0554, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.8911786881125912, | |
| "grad_norm": 0.977917492389679, | |
| "learning_rate": 5.157241379310345e-06, | |
| "loss": 0.0561, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 1.8974616737873837, | |
| "grad_norm": 1.2408355474472046, | |
| "learning_rate": 5.140000000000001e-06, | |
| "loss": 0.0615, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 1.9037446594621765, | |
| "grad_norm": 1.273364543914795, | |
| "learning_rate": 5.122758620689655e-06, | |
| "loss": 0.0572, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 1.910027645136969, | |
| "grad_norm": 1.2105774879455566, | |
| "learning_rate": 5.1055172413793105e-06, | |
| "loss": 0.0469, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.9163106308117617, | |
| "grad_norm": 0.6686076521873474, | |
| "learning_rate": 5.088275862068965e-06, | |
| "loss": 0.0596, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 1.9225936164865542, | |
| "grad_norm": 1.473767876625061, | |
| "learning_rate": 5.071034482758621e-06, | |
| "loss": 0.0558, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 1.928876602161347, | |
| "grad_norm": 1.183693289756775, | |
| "learning_rate": 5.053793103448277e-06, | |
| "loss": 0.0578, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 1.9351595878361398, | |
| "grad_norm": 1.661081075668335, | |
| "learning_rate": 5.036551724137932e-06, | |
| "loss": 0.0577, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.9414425735109324, | |
| "grad_norm": 1.035583734512329, | |
| "learning_rate": 5.019310344827587e-06, | |
| "loss": 0.0504, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 1.947725559185725, | |
| "grad_norm": 1.2706879377365112, | |
| "learning_rate": 5.002068965517241e-06, | |
| "loss": 0.0523, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.9540085448605176, | |
| "grad_norm": 1.558969497680664, | |
| "learning_rate": 4.984827586206897e-06, | |
| "loss": 0.0527, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 1.9602915305353104, | |
| "grad_norm": 2.107837677001953, | |
| "learning_rate": 4.967586206896552e-06, | |
| "loss": 0.0496, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.9665745162101032, | |
| "grad_norm": 1.1281065940856934, | |
| "learning_rate": 4.950344827586207e-06, | |
| "loss": 0.0495, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 1.9728575018848957, | |
| "grad_norm": 0.92071133852005, | |
| "learning_rate": 4.933103448275863e-06, | |
| "loss": 0.0543, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 1.9791404875596883, | |
| "grad_norm": 1.5125892162322998, | |
| "learning_rate": 4.9158620689655175e-06, | |
| "loss": 0.0567, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 1.9854234732344809, | |
| "grad_norm": 1.4018179178237915, | |
| "learning_rate": 4.898620689655173e-06, | |
| "loss": 0.0577, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.9917064589092737, | |
| "grad_norm": 1.599665880203247, | |
| "learning_rate": 4.881379310344828e-06, | |
| "loss": 0.0511, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 1.9979894445840665, | |
| "grad_norm": 1.3747309446334839, | |
| "learning_rate": 4.864137931034483e-06, | |
| "loss": 0.0545, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 2.004272430258859, | |
| "grad_norm": 1.2372041940689087, | |
| "learning_rate": 4.846896551724139e-06, | |
| "loss": 0.0434, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 2.0105554159336516, | |
| "grad_norm": 1.0974595546722412, | |
| "learning_rate": 4.8296551724137935e-06, | |
| "loss": 0.0426, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.0105554159336516, | |
| "eval_loss": 0.1260567456483841, | |
| "eval_runtime": 28488.8565, | |
| "eval_samples_per_second": 0.686, | |
| "eval_steps_per_second": 0.086, | |
| "eval_wer": 36.9741526861996, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.016838401608444, | |
| "grad_norm": 0.5434245467185974, | |
| "learning_rate": 4.812413793103448e-06, | |
| "loss": 0.0372, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 2.023121387283237, | |
| "grad_norm": 1.4904873371124268, | |
| "learning_rate": 4.795172413793104e-06, | |
| "loss": 0.0322, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 2.02940437295803, | |
| "grad_norm": 0.8786129355430603, | |
| "learning_rate": 4.777931034482759e-06, | |
| "loss": 0.0378, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 2.0356873586328224, | |
| "grad_norm": 1.062193512916565, | |
| "learning_rate": 4.760689655172414e-06, | |
| "loss": 0.0335, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.041970344307615, | |
| "grad_norm": 1.1229702234268188, | |
| "learning_rate": 4.7434482758620696e-06, | |
| "loss": 0.0341, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 2.0482533299824075, | |
| "grad_norm": 1.5959960222244263, | |
| "learning_rate": 4.726206896551724e-06, | |
| "loss": 0.0326, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 2.0545363156572005, | |
| "grad_norm": 1.3407766819000244, | |
| "learning_rate": 4.708965517241379e-06, | |
| "loss": 0.0282, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 2.060819301331993, | |
| "grad_norm": 0.8177748322486877, | |
| "learning_rate": 4.691724137931035e-06, | |
| "loss": 0.0334, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.0671022870067857, | |
| "grad_norm": 0.5434231162071228, | |
| "learning_rate": 4.67448275862069e-06, | |
| "loss": 0.0332, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 2.0733852726815782, | |
| "grad_norm": 0.6810811758041382, | |
| "learning_rate": 4.657241379310346e-06, | |
| "loss": 0.0343, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 2.079668258356371, | |
| "grad_norm": 1.3621633052825928, | |
| "learning_rate": 4.6400000000000005e-06, | |
| "loss": 0.0358, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 2.085951244031164, | |
| "grad_norm": 0.8485309481620789, | |
| "learning_rate": 4.622758620689655e-06, | |
| "loss": 0.0283, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.0922342297059564, | |
| "grad_norm": 1.2133398056030273, | |
| "learning_rate": 4.605517241379311e-06, | |
| "loss": 0.0342, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 2.098517215380749, | |
| "grad_norm": 1.9074926376342773, | |
| "learning_rate": 4.588275862068966e-06, | |
| "loss": 0.0372, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 2.1048002010555416, | |
| "grad_norm": 1.3371448516845703, | |
| "learning_rate": 4.571034482758621e-06, | |
| "loss": 0.0356, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 2.111083186730334, | |
| "grad_norm": 1.3409150838851929, | |
| "learning_rate": 4.5537931034482765e-06, | |
| "loss": 0.0354, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.117366172405127, | |
| "grad_norm": 1.1407537460327148, | |
| "learning_rate": 4.536551724137931e-06, | |
| "loss": 0.0338, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 2.1236491580799197, | |
| "grad_norm": 1.0432764291763306, | |
| "learning_rate": 4.519310344827586e-06, | |
| "loss": 0.0325, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 2.1299321437547123, | |
| "grad_norm": 1.2592930793762207, | |
| "learning_rate": 4.502068965517242e-06, | |
| "loss": 0.0374, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 2.136215129429505, | |
| "grad_norm": 0.9935320019721985, | |
| "learning_rate": 4.484827586206897e-06, | |
| "loss": 0.0349, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.1424981151042974, | |
| "grad_norm": 1.1453524827957153, | |
| "learning_rate": 4.467586206896552e-06, | |
| "loss": 0.0357, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 2.1487811007790905, | |
| "grad_norm": 0.8577796816825867, | |
| "learning_rate": 4.4503448275862074e-06, | |
| "loss": 0.0349, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 2.155064086453883, | |
| "grad_norm": 1.0337741374969482, | |
| "learning_rate": 4.433103448275862e-06, | |
| "loss": 0.0321, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 2.1613470721286756, | |
| "grad_norm": 1.4780592918395996, | |
| "learning_rate": 4.415862068965517e-06, | |
| "loss": 0.0359, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.167630057803468, | |
| "grad_norm": 1.6528609991073608, | |
| "learning_rate": 4.398620689655173e-06, | |
| "loss": 0.0384, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 2.1739130434782608, | |
| "grad_norm": 0.7156565189361572, | |
| "learning_rate": 4.381379310344829e-06, | |
| "loss": 0.0353, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 2.1801960291530533, | |
| "grad_norm": 1.1753544807434082, | |
| "learning_rate": 4.3641379310344835e-06, | |
| "loss": 0.031, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 2.1864790148278463, | |
| "grad_norm": 0.7453944087028503, | |
| "learning_rate": 4.346896551724138e-06, | |
| "loss": 0.0348, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.192762000502639, | |
| "grad_norm": 1.872745156288147, | |
| "learning_rate": 4.329655172413793e-06, | |
| "loss": 0.0351, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 2.1990449861774315, | |
| "grad_norm": 0.6683670282363892, | |
| "learning_rate": 4.312413793103449e-06, | |
| "loss": 0.0351, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 2.205327971852224, | |
| "grad_norm": 1.3862112760543823, | |
| "learning_rate": 4.295172413793104e-06, | |
| "loss": 0.0333, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 2.2116109575270166, | |
| "grad_norm": 1.027766466140747, | |
| "learning_rate": 4.277931034482759e-06, | |
| "loss": 0.0287, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.2178939432018097, | |
| "grad_norm": 0.8764299154281616, | |
| "learning_rate": 4.260689655172414e-06, | |
| "loss": 0.0324, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 2.2241769288766022, | |
| "grad_norm": 0.8723062872886658, | |
| "learning_rate": 4.243448275862069e-06, | |
| "loss": 0.0283, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 2.230459914551395, | |
| "grad_norm": 1.1235435009002686, | |
| "learning_rate": 4.226206896551724e-06, | |
| "loss": 0.0376, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 2.2367429002261874, | |
| "grad_norm": 1.0615513324737549, | |
| "learning_rate": 4.20896551724138e-06, | |
| "loss": 0.0354, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.24302588590098, | |
| "grad_norm": 0.8142825365066528, | |
| "learning_rate": 4.191724137931035e-06, | |
| "loss": 0.0274, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 2.249308871575773, | |
| "grad_norm": 1.7816015481948853, | |
| "learning_rate": 4.17448275862069e-06, | |
| "loss": 0.0349, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 2.2555918572505655, | |
| "grad_norm": 1.0881839990615845, | |
| "learning_rate": 4.157241379310345e-06, | |
| "loss": 0.0344, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 2.261874842925358, | |
| "grad_norm": 0.554862916469574, | |
| "learning_rate": 4.14e-06, | |
| "loss": 0.0305, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.261874842925358, | |
| "eval_loss": 0.1346791386604309, | |
| "eval_runtime": 28846.9785, | |
| "eval_samples_per_second": 0.677, | |
| "eval_steps_per_second": 0.085, | |
| "eval_wer": 36.786382314919805, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.2681578286001507, | |
| "grad_norm": 1.2704734802246094, | |
| "learning_rate": 4.122758620689655e-06, | |
| "loss": 0.0334, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 2.2744408142749433, | |
| "grad_norm": 1.3331466913223267, | |
| "learning_rate": 4.105517241379311e-06, | |
| "loss": 0.0343, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 2.2807237999497363, | |
| "grad_norm": 0.3723588287830353, | |
| "learning_rate": 4.0882758620689665e-06, | |
| "loss": 0.0343, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 2.287006785624529, | |
| "grad_norm": 2.03139591217041, | |
| "learning_rate": 4.071034482758621e-06, | |
| "loss": 0.034, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.2932897712993214, | |
| "grad_norm": 1.230726718902588, | |
| "learning_rate": 4.053793103448276e-06, | |
| "loss": 0.0319, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 2.299572756974114, | |
| "grad_norm": 1.408710241317749, | |
| "learning_rate": 4.036551724137931e-06, | |
| "loss": 0.0267, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 2.3058557426489066, | |
| "grad_norm": 1.5532382726669312, | |
| "learning_rate": 4.019310344827587e-06, | |
| "loss": 0.0335, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 2.3121387283236996, | |
| "grad_norm": 1.1856962442398071, | |
| "learning_rate": 4.002068965517242e-06, | |
| "loss": 0.0346, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.318421713998492, | |
| "grad_norm": 1.0790759325027466, | |
| "learning_rate": 3.9848275862068965e-06, | |
| "loss": 0.0339, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 2.3247046996732847, | |
| "grad_norm": 0.7154790759086609, | |
| "learning_rate": 3.967586206896552e-06, | |
| "loss": 0.0353, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 2.3309876853480773, | |
| "grad_norm": 0.8203781843185425, | |
| "learning_rate": 3.950344827586207e-06, | |
| "loss": 0.0392, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 2.33727067102287, | |
| "grad_norm": 1.3644154071807861, | |
| "learning_rate": 3.933103448275862e-06, | |
| "loss": 0.0333, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.343553656697663, | |
| "grad_norm": 0.7084758877754211, | |
| "learning_rate": 3.915862068965518e-06, | |
| "loss": 0.0309, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 2.3498366423724555, | |
| "grad_norm": 0.7974054217338562, | |
| "learning_rate": 3.898620689655173e-06, | |
| "loss": 0.0337, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 2.356119628047248, | |
| "grad_norm": 0.9612919092178345, | |
| "learning_rate": 3.8813793103448275e-06, | |
| "loss": 0.0358, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 2.3624026137220406, | |
| "grad_norm": 1.3854459524154663, | |
| "learning_rate": 3.864137931034483e-06, | |
| "loss": 0.0299, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.368685599396833, | |
| "grad_norm": 1.1167787313461304, | |
| "learning_rate": 3.846896551724138e-06, | |
| "loss": 0.0337, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 2.3749685850716262, | |
| "grad_norm": 0.9742668271064758, | |
| "learning_rate": 3.829655172413793e-06, | |
| "loss": 0.0286, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 2.381251570746419, | |
| "grad_norm": 1.435309886932373, | |
| "learning_rate": 3.8124137931034486e-06, | |
| "loss": 0.0305, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 2.3875345564212114, | |
| "grad_norm": 1.1362321376800537, | |
| "learning_rate": 3.795172413793104e-06, | |
| "loss": 0.0313, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.393817542096004, | |
| "grad_norm": 1.0465309619903564, | |
| "learning_rate": 3.7779310344827592e-06, | |
| "loss": 0.0364, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 2.4001005277707965, | |
| "grad_norm": 1.4722024202346802, | |
| "learning_rate": 3.760689655172414e-06, | |
| "loss": 0.0451, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 2.406383513445589, | |
| "grad_norm": 1.1084930896759033, | |
| "learning_rate": 3.7434482758620694e-06, | |
| "loss": 0.0349, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 2.412666499120382, | |
| "grad_norm": 1.4382020235061646, | |
| "learning_rate": 3.7262068965517247e-06, | |
| "loss": 0.0338, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.4189494847951747, | |
| "grad_norm": 1.6590332984924316, | |
| "learning_rate": 3.7089655172413795e-06, | |
| "loss": 0.0335, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 2.4252324704699673, | |
| "grad_norm": 0.7975425720214844, | |
| "learning_rate": 3.691724137931035e-06, | |
| "loss": 0.0371, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 2.43151545614476, | |
| "grad_norm": 0.9135144352912903, | |
| "learning_rate": 3.67448275862069e-06, | |
| "loss": 0.0384, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 2.4377984418195524, | |
| "grad_norm": 1.74324369430542, | |
| "learning_rate": 3.657241379310345e-06, | |
| "loss": 0.029, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 2.4440814274943454, | |
| "grad_norm": 0.9986597299575806, | |
| "learning_rate": 3.6400000000000003e-06, | |
| "loss": 0.0363, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 2.450364413169138, | |
| "grad_norm": 0.8304340839385986, | |
| "learning_rate": 3.622758620689655e-06, | |
| "loss": 0.0337, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 2.4566473988439306, | |
| "grad_norm": 0.6401971578598022, | |
| "learning_rate": 3.6055172413793105e-06, | |
| "loss": 0.0309, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 2.462930384518723, | |
| "grad_norm": 1.2303663492202759, | |
| "learning_rate": 3.5882758620689658e-06, | |
| "loss": 0.0333, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 2.4692133701935157, | |
| "grad_norm": 1.2973604202270508, | |
| "learning_rate": 3.5710344827586206e-06, | |
| "loss": 0.0346, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 2.4754963558683087, | |
| "grad_norm": 1.0538204908370972, | |
| "learning_rate": 3.553793103448276e-06, | |
| "loss": 0.0335, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 2.4817793415431013, | |
| "grad_norm": 1.006469964981079, | |
| "learning_rate": 3.5365517241379316e-06, | |
| "loss": 0.0319, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 2.488062327217894, | |
| "grad_norm": 1.2951979637145996, | |
| "learning_rate": 3.5193103448275865e-06, | |
| "loss": 0.0309, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 2.4943453128926865, | |
| "grad_norm": 1.205333948135376, | |
| "learning_rate": 3.502068965517242e-06, | |
| "loss": 0.0322, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 2.500628298567479, | |
| "grad_norm": 1.051314115524292, | |
| "learning_rate": 3.484827586206897e-06, | |
| "loss": 0.0347, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 2.506911284242272, | |
| "grad_norm": 1.1379516124725342, | |
| "learning_rate": 3.467586206896552e-06, | |
| "loss": 0.0297, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 2.5131942699170646, | |
| "grad_norm": 1.4470369815826416, | |
| "learning_rate": 3.4503448275862073e-06, | |
| "loss": 0.0344, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.5131942699170646, | |
| "eval_loss": 0.13530105352401733, | |
| "eval_runtime": 28721.0802, | |
| "eval_samples_per_second": 0.68, | |
| "eval_steps_per_second": 0.085, | |
| "eval_wer": 36.4960604331718, | |
| "step": 10000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 15000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.6324881677254656e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |