| { | |
| "best_metric": 9.547895591357562, | |
| "best_model_checkpoint": "./checkpoint-10000", | |
| "epoch": 1.0, | |
| "eval_steps": 1000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0025, | |
| "grad_norm": 17.154964447021484, | |
| "learning_rate": 5.5e-07, | |
| "loss": 2.3215, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.005, | |
| "grad_norm": 9.312045097351074, | |
| "learning_rate": 1.175e-06, | |
| "loss": 1.864, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0075, | |
| "grad_norm": 9.26079273223877, | |
| "learning_rate": 1.8e-06, | |
| "loss": 1.2223, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 6.739448070526123, | |
| "learning_rate": 2.425e-06, | |
| "loss": 0.9215, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0125, | |
| "grad_norm": 5.92506217956543, | |
| "learning_rate": 3.05e-06, | |
| "loss": 0.7773, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.015, | |
| "grad_norm": 5.521960735321045, | |
| "learning_rate": 3.675e-06, | |
| "loss": 0.6549, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0175, | |
| "grad_norm": 5.140420913696289, | |
| "learning_rate": 4.2999999999999995e-06, | |
| "loss": 0.6097, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 5.54941987991333, | |
| "learning_rate": 4.925e-06, | |
| "loss": 0.5707, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0225, | |
| "grad_norm": 5.2704644203186035, | |
| "learning_rate": 5.55e-06, | |
| "loss": 0.4844, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "grad_norm": 6.43791389465332, | |
| "learning_rate": 6.175e-06, | |
| "loss": 0.5029, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.0275, | |
| "grad_norm": 5.382338523864746, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 0.5223, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.741837024688721, | |
| "learning_rate": 7.425e-06, | |
| "loss": 0.5109, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.0325, | |
| "grad_norm": 4.826135635375977, | |
| "learning_rate": 8.050000000000001e-06, | |
| "loss": 0.447, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.035, | |
| "grad_norm": 4.86759090423584, | |
| "learning_rate": 8.674999999999999e-06, | |
| "loss": 0.3823, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.0375, | |
| "grad_norm": 5.171136379241943, | |
| "learning_rate": 9.3e-06, | |
| "loss": 0.3243, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 3.679882287979126, | |
| "learning_rate": 9.925e-06, | |
| "loss": 0.3157, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.0425, | |
| "grad_norm": 3.856377124786377, | |
| "learning_rate": 1.055e-05, | |
| "loss": 0.2852, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.045, | |
| "grad_norm": 3.587339401245117, | |
| "learning_rate": 1.1175e-05, | |
| "loss": 0.2729, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.0475, | |
| "grad_norm": 3.875290870666504, | |
| "learning_rate": 1.18e-05, | |
| "loss": 0.253, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 3.3402156829833984, | |
| "learning_rate": 1.2425e-05, | |
| "loss": 0.2432, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.0525, | |
| "grad_norm": 4.392176151275635, | |
| "learning_rate": 1.2471052631578948e-05, | |
| "loss": 0.2231, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.055, | |
| "grad_norm": 3.0165233612060547, | |
| "learning_rate": 1.2438157894736843e-05, | |
| "loss": 0.221, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.0575, | |
| "grad_norm": 3.5857527256011963, | |
| "learning_rate": 1.2405263157894738e-05, | |
| "loss": 0.2065, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 4.286252021789551, | |
| "learning_rate": 1.2372368421052632e-05, | |
| "loss": 0.2102, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.0625, | |
| "grad_norm": 3.300917148590088, | |
| "learning_rate": 1.2339473684210527e-05, | |
| "loss": 0.1838, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.065, | |
| "grad_norm": 2.5947461128234863, | |
| "learning_rate": 1.2306578947368421e-05, | |
| "loss": 0.2015, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.0675, | |
| "grad_norm": 2.932685375213623, | |
| "learning_rate": 1.2273684210526316e-05, | |
| "loss": 0.1983, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 3.6400482654571533, | |
| "learning_rate": 1.224078947368421e-05, | |
| "loss": 0.1977, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.0725, | |
| "grad_norm": 4.641479015350342, | |
| "learning_rate": 1.2207894736842105e-05, | |
| "loss": 0.2214, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "grad_norm": 4.521893501281738, | |
| "learning_rate": 1.2175e-05, | |
| "loss": 0.3166, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.0775, | |
| "grad_norm": 4.501023292541504, | |
| "learning_rate": 1.2142105263157896e-05, | |
| "loss": 0.3019, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 4.930638790130615, | |
| "learning_rate": 1.210921052631579e-05, | |
| "loss": 0.3381, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.0825, | |
| "grad_norm": 3.6409413814544678, | |
| "learning_rate": 1.2076315789473685e-05, | |
| "loss": 0.2239, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.085, | |
| "grad_norm": 2.9868454933166504, | |
| "learning_rate": 1.204342105263158e-05, | |
| "loss": 0.1808, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.0875, | |
| "grad_norm": 2.4896903038024902, | |
| "learning_rate": 1.2010526315789476e-05, | |
| "loss": 0.1638, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.112133502960205, | |
| "learning_rate": 1.197763157894737e-05, | |
| "loss": 0.2276, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.0925, | |
| "grad_norm": 4.977311134338379, | |
| "learning_rate": 1.1944736842105265e-05, | |
| "loss": 0.2745, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.095, | |
| "grad_norm": 4.248500347137451, | |
| "learning_rate": 1.191184210526316e-05, | |
| "loss": 0.2651, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.0975, | |
| "grad_norm": 5.643575191497803, | |
| "learning_rate": 1.1878947368421054e-05, | |
| "loss": 0.4676, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.4507365226745605, | |
| "learning_rate": 1.1846052631578948e-05, | |
| "loss": 0.3863, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 0.4089725911617279, | |
| "eval_runtime": 110.8228, | |
| "eval_samples_per_second": 15.259, | |
| "eval_steps_per_second": 0.956, | |
| "eval_wer": 21.21893343324591, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1025, | |
| "grad_norm": 4.655358791351318, | |
| "learning_rate": 1.1813157894736843e-05, | |
| "loss": 0.3148, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.105, | |
| "grad_norm": 3.134038209915161, | |
| "learning_rate": 1.1780263157894738e-05, | |
| "loss": 0.2218, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.1075, | |
| "grad_norm": 2.7970399856567383, | |
| "learning_rate": 1.1747368421052632e-05, | |
| "loss": 0.163, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 2.5460939407348633, | |
| "learning_rate": 1.1714473684210527e-05, | |
| "loss": 0.1353, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1125, | |
| "grad_norm": 2.4967522621154785, | |
| "learning_rate": 1.1681578947368421e-05, | |
| "loss": 0.1318, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.115, | |
| "grad_norm": 3.2984485626220703, | |
| "learning_rate": 1.1648684210526316e-05, | |
| "loss": 0.1417, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.1175, | |
| "grad_norm": 2.9972639083862305, | |
| "learning_rate": 1.161578947368421e-05, | |
| "loss": 0.1429, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 2.817171335220337, | |
| "learning_rate": 1.1582894736842107e-05, | |
| "loss": 0.1407, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.1225, | |
| "grad_norm": 4.157276153564453, | |
| "learning_rate": 1.1550000000000001e-05, | |
| "loss": 0.2104, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 4.1519012451171875, | |
| "learning_rate": 1.1517105263157896e-05, | |
| "loss": 0.228, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.1275, | |
| "grad_norm": 4.5924811363220215, | |
| "learning_rate": 1.148421052631579e-05, | |
| "loss": 0.2494, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 4.487509727478027, | |
| "learning_rate": 1.1451315789473685e-05, | |
| "loss": 0.2569, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.1325, | |
| "grad_norm": 4.964361667633057, | |
| "learning_rate": 1.141842105263158e-05, | |
| "loss": 0.2242, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.135, | |
| "grad_norm": 4.045988082885742, | |
| "learning_rate": 1.1385526315789474e-05, | |
| "loss": 0.2291, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.1375, | |
| "grad_norm": 5.821484088897705, | |
| "learning_rate": 1.1352631578947369e-05, | |
| "loss": 0.2133, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 4.365011692047119, | |
| "learning_rate": 1.1319736842105263e-05, | |
| "loss": 0.2065, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.1425, | |
| "grad_norm": 3.424377202987671, | |
| "learning_rate": 1.1286842105263158e-05, | |
| "loss": 0.2126, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.145, | |
| "grad_norm": 3.383793830871582, | |
| "learning_rate": 1.1253947368421052e-05, | |
| "loss": 0.2013, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.1475, | |
| "grad_norm": 2.5089175701141357, | |
| "learning_rate": 1.1221052631578947e-05, | |
| "loss": 0.1484, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 2.754504442214966, | |
| "learning_rate": 1.1188157894736841e-05, | |
| "loss": 0.1276, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.1525, | |
| "grad_norm": 2.481276512145996, | |
| "learning_rate": 1.1155263157894736e-05, | |
| "loss": 0.1363, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.155, | |
| "grad_norm": 3.6863863468170166, | |
| "learning_rate": 1.1122368421052632e-05, | |
| "loss": 0.1297, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.1575, | |
| "grad_norm": 2.3154377937316895, | |
| "learning_rate": 1.1089473684210527e-05, | |
| "loss": 0.1122, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 2.529965877532959, | |
| "learning_rate": 1.1056578947368421e-05, | |
| "loss": 0.1206, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.1625, | |
| "grad_norm": 4.41377592086792, | |
| "learning_rate": 1.1023684210526316e-05, | |
| "loss": 0.1778, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.165, | |
| "grad_norm": 3.306654214859009, | |
| "learning_rate": 1.099078947368421e-05, | |
| "loss": 0.1909, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.1675, | |
| "grad_norm": 4.408228874206543, | |
| "learning_rate": 1.0957894736842107e-05, | |
| "loss": 0.1935, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 2.594836950302124, | |
| "learning_rate": 1.0925000000000001e-05, | |
| "loss": 0.1859, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.1725, | |
| "grad_norm": 2.800992250442505, | |
| "learning_rate": 1.0892105263157896e-05, | |
| "loss": 0.1377, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "grad_norm": 2.8291330337524414, | |
| "learning_rate": 1.085921052631579e-05, | |
| "loss": 0.1344, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.1775, | |
| "grad_norm": 2.815859317779541, | |
| "learning_rate": 1.0826315789473685e-05, | |
| "loss": 0.1157, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 2.485546827316284, | |
| "learning_rate": 1.079342105263158e-05, | |
| "loss": 0.131, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.1825, | |
| "grad_norm": 2.3912546634674072, | |
| "learning_rate": 1.0760526315789474e-05, | |
| "loss": 0.1143, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.185, | |
| "grad_norm": 2.0489096641540527, | |
| "learning_rate": 1.0727631578947369e-05, | |
| "loss": 0.1112, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.1875, | |
| "grad_norm": 2.72310209274292, | |
| "learning_rate": 1.0694736842105265e-05, | |
| "loss": 0.1128, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 2.5979485511779785, | |
| "learning_rate": 1.066184210526316e-05, | |
| "loss": 0.1145, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.1925, | |
| "grad_norm": 2.1391546726226807, | |
| "learning_rate": 1.0628947368421054e-05, | |
| "loss": 0.1083, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.195, | |
| "grad_norm": 3.562283515930176, | |
| "learning_rate": 1.0596052631578949e-05, | |
| "loss": 0.1456, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.1975, | |
| "grad_norm": 3.5763707160949707, | |
| "learning_rate": 1.0563157894736843e-05, | |
| "loss": 0.1929, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 3.659284830093384, | |
| "learning_rate": 1.0530263157894738e-05, | |
| "loss": 0.1897, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.3456783592700958, | |
| "eval_runtime": 109.2542, | |
| "eval_samples_per_second": 15.478, | |
| "eval_steps_per_second": 0.97, | |
| "eval_wer": 15.448982140626951, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2025, | |
| "grad_norm": 2.3411598205566406, | |
| "learning_rate": 1.0497368421052632e-05, | |
| "loss": 0.1575, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.205, | |
| "grad_norm": 2.3811259269714355, | |
| "learning_rate": 1.0464473684210527e-05, | |
| "loss": 0.1096, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.2075, | |
| "grad_norm": 2.5568268299102783, | |
| "learning_rate": 1.0431578947368421e-05, | |
| "loss": 0.1079, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 2.217919111251831, | |
| "learning_rate": 1.0398684210526316e-05, | |
| "loss": 0.109, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.2125, | |
| "grad_norm": 2.830374002456665, | |
| "learning_rate": 1.036578947368421e-05, | |
| "loss": 0.096, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.215, | |
| "grad_norm": 2.265371084213257, | |
| "learning_rate": 1.0332894736842105e-05, | |
| "loss": 0.0942, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.2175, | |
| "grad_norm": 2.7084972858428955, | |
| "learning_rate": 1.03e-05, | |
| "loss": 0.0966, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 3.6277427673339844, | |
| "learning_rate": 1.0267105263157894e-05, | |
| "loss": 0.168, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.2225, | |
| "grad_norm": 4.01047945022583, | |
| "learning_rate": 1.023421052631579e-05, | |
| "loss": 0.1743, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "grad_norm": 3.7068288326263428, | |
| "learning_rate": 1.0201315789473685e-05, | |
| "loss": 0.192, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2275, | |
| "grad_norm": 2.932413339614868, | |
| "learning_rate": 1.016842105263158e-05, | |
| "loss": 0.1261, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 2.187391757965088, | |
| "learning_rate": 1.0135526315789474e-05, | |
| "loss": 0.0932, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.2325, | |
| "grad_norm": 2.829944372177124, | |
| "learning_rate": 1.0102631578947369e-05, | |
| "loss": 0.0906, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.235, | |
| "grad_norm": 3.1608800888061523, | |
| "learning_rate": 1.0069736842105263e-05, | |
| "loss": 0.0892, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2375, | |
| "grad_norm": 1.8967108726501465, | |
| "learning_rate": 1.0036842105263158e-05, | |
| "loss": 0.0976, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 2.088034152984619, | |
| "learning_rate": 1.0003947368421052e-05, | |
| "loss": 0.0926, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.2425, | |
| "grad_norm": 1.8807507753372192, | |
| "learning_rate": 9.971052631578947e-06, | |
| "loss": 0.1022, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.245, | |
| "grad_norm": 1.8424572944641113, | |
| "learning_rate": 9.938157894736842e-06, | |
| "loss": 0.0945, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.2475, | |
| "grad_norm": 2.1998050212860107, | |
| "learning_rate": 9.905263157894736e-06, | |
| "loss": 0.1008, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 3.174199104309082, | |
| "learning_rate": 9.872368421052632e-06, | |
| "loss": 0.1111, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.2525, | |
| "grad_norm": 3.0761404037475586, | |
| "learning_rate": 9.839473684210527e-06, | |
| "loss": 0.1474, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.255, | |
| "grad_norm": 3.375256299972534, | |
| "learning_rate": 9.806578947368421e-06, | |
| "loss": 0.1427, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.2575, | |
| "grad_norm": 3.164018154144287, | |
| "learning_rate": 9.773684210526316e-06, | |
| "loss": 0.1707, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 2.6457083225250244, | |
| "learning_rate": 9.740789473684212e-06, | |
| "loss": 0.1277, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.2625, | |
| "grad_norm": 2.7457237243652344, | |
| "learning_rate": 9.707894736842107e-06, | |
| "loss": 0.1052, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.265, | |
| "grad_norm": 1.7891429662704468, | |
| "learning_rate": 9.675000000000001e-06, | |
| "loss": 0.0954, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.2675, | |
| "grad_norm": 2.1033453941345215, | |
| "learning_rate": 9.642105263157896e-06, | |
| "loss": 0.0897, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 2.163673162460327, | |
| "learning_rate": 9.60921052631579e-06, | |
| "loss": 0.0914, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.2725, | |
| "grad_norm": 1.7669174671173096, | |
| "learning_rate": 9.576315789473685e-06, | |
| "loss": 0.1058, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "grad_norm": 2.96284818649292, | |
| "learning_rate": 9.54342105263158e-06, | |
| "loss": 0.1092, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.2775, | |
| "grad_norm": 3.970527410507202, | |
| "learning_rate": 9.510526315789474e-06, | |
| "loss": 0.1937, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 3.5204854011535645, | |
| "learning_rate": 9.477631578947369e-06, | |
| "loss": 0.183, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.2825, | |
| "grad_norm": 3.5082786083221436, | |
| "learning_rate": 9.444736842105263e-06, | |
| "loss": 0.152, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.285, | |
| "grad_norm": 3.042353630065918, | |
| "learning_rate": 9.411842105263158e-06, | |
| "loss": 0.1505, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.2875, | |
| "grad_norm": 3.401519536972046, | |
| "learning_rate": 9.378947368421052e-06, | |
| "loss": 0.1568, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 2.9089572429656982, | |
| "learning_rate": 9.346052631578949e-06, | |
| "loss": 0.1482, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.2925, | |
| "grad_norm": 2.2120282649993896, | |
| "learning_rate": 9.313157894736843e-06, | |
| "loss": 0.1219, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.295, | |
| "grad_norm": 2.358677625656128, | |
| "learning_rate": 9.280263157894738e-06, | |
| "loss": 0.0925, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.2975, | |
| "grad_norm": 2.1122074127197266, | |
| "learning_rate": 9.247368421052632e-06, | |
| "loss": 0.0993, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 2.7415077686309814, | |
| "learning_rate": 9.214473684210527e-06, | |
| "loss": 0.1379, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_loss": 0.3283405303955078, | |
| "eval_runtime": 108.699, | |
| "eval_samples_per_second": 15.557, | |
| "eval_steps_per_second": 0.975, | |
| "eval_wer": 13.57562133133508, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3025, | |
| "grad_norm": 3.9535341262817383, | |
| "learning_rate": 9.181578947368422e-06, | |
| "loss": 0.1678, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.305, | |
| "grad_norm": 4.159903526306152, | |
| "learning_rate": 9.148684210526316e-06, | |
| "loss": 0.1903, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.3075, | |
| "grad_norm": 2.060307264328003, | |
| "learning_rate": 9.11578947368421e-06, | |
| "loss": 0.1233, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 2.7883658409118652, | |
| "learning_rate": 9.082894736842105e-06, | |
| "loss": 0.0932, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.3125, | |
| "grad_norm": 2.121624231338501, | |
| "learning_rate": 9.05e-06, | |
| "loss": 0.0903, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.315, | |
| "grad_norm": 2.487900972366333, | |
| "learning_rate": 9.017105263157894e-06, | |
| "loss": 0.1073, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.3175, | |
| "grad_norm": 3.3630971908569336, | |
| "learning_rate": 8.984210526315789e-06, | |
| "loss": 0.1667, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 3.9546568393707275, | |
| "learning_rate": 8.951315789473683e-06, | |
| "loss": 0.152, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.3225, | |
| "grad_norm": 3.8344779014587402, | |
| "learning_rate": 8.91842105263158e-06, | |
| "loss": 0.1644, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "grad_norm": 3.418964147567749, | |
| "learning_rate": 8.885526315789474e-06, | |
| "loss": 0.1287, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.3275, | |
| "grad_norm": 3.6898674964904785, | |
| "learning_rate": 8.852631578947369e-06, | |
| "loss": 0.1363, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 2.754051685333252, | |
| "learning_rate": 8.819736842105263e-06, | |
| "loss": 0.1568, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.3325, | |
| "grad_norm": 2.718393087387085, | |
| "learning_rate": 8.786842105263158e-06, | |
| "loss": 0.1177, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.335, | |
| "grad_norm": 2.952477216720581, | |
| "learning_rate": 8.753947368421052e-06, | |
| "loss": 0.0932, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.3375, | |
| "grad_norm": 2.300719976425171, | |
| "learning_rate": 8.721052631578947e-06, | |
| "loss": 0.0814, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 2.1434884071350098, | |
| "learning_rate": 8.688157894736843e-06, | |
| "loss": 0.0882, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.3425, | |
| "grad_norm": 2.9985721111297607, | |
| "learning_rate": 8.655263157894738e-06, | |
| "loss": 0.0921, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.345, | |
| "grad_norm": 2.030388116836548, | |
| "learning_rate": 8.622368421052632e-06, | |
| "loss": 0.0839, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3475, | |
| "grad_norm": 1.9179843664169312, | |
| "learning_rate": 8.589473684210527e-06, | |
| "loss": 0.0789, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 2.291533946990967, | |
| "learning_rate": 8.556578947368422e-06, | |
| "loss": 0.0928, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.3525, | |
| "grad_norm": 2.0841875076293945, | |
| "learning_rate": 8.523684210526316e-06, | |
| "loss": 0.0871, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.355, | |
| "grad_norm": 2.898160457611084, | |
| "learning_rate": 8.49078947368421e-06, | |
| "loss": 0.0872, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.3575, | |
| "grad_norm": 3.860246419906616, | |
| "learning_rate": 8.457894736842107e-06, | |
| "loss": 0.1361, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 3.1953701972961426, | |
| "learning_rate": 8.425000000000001e-06, | |
| "loss": 0.1536, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.3625, | |
| "grad_norm": 2.907952308654785, | |
| "learning_rate": 8.392105263157896e-06, | |
| "loss": 0.1502, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.365, | |
| "grad_norm": 1.8253568410873413, | |
| "learning_rate": 8.35921052631579e-06, | |
| "loss": 0.0884, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.3675, | |
| "grad_norm": 1.9003596305847168, | |
| "learning_rate": 8.326315789473685e-06, | |
| "loss": 0.0815, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 2.44956111907959, | |
| "learning_rate": 8.29342105263158e-06, | |
| "loss": 0.0868, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.3725, | |
| "grad_norm": 2.3651952743530273, | |
| "learning_rate": 8.260526315789474e-06, | |
| "loss": 0.0966, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 3.287203550338745, | |
| "learning_rate": 8.227631578947369e-06, | |
| "loss": 0.1213, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.3775, | |
| "grad_norm": 3.2496094703674316, | |
| "learning_rate": 8.194736842105263e-06, | |
| "loss": 0.1525, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 4.660492420196533, | |
| "learning_rate": 8.161842105263158e-06, | |
| "loss": 0.1604, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.3825, | |
| "grad_norm": 3.705150842666626, | |
| "learning_rate": 8.128947368421053e-06, | |
| "loss": 0.1473, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.385, | |
| "grad_norm": 3.797199010848999, | |
| "learning_rate": 8.096052631578947e-06, | |
| "loss": 0.1544, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.3875, | |
| "grad_norm": 2.2780754566192627, | |
| "learning_rate": 8.063157894736842e-06, | |
| "loss": 0.1452, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 2.3213751316070557, | |
| "learning_rate": 8.030263157894736e-06, | |
| "loss": 0.1036, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.3925, | |
| "grad_norm": 2.1721794605255127, | |
| "learning_rate": 7.997368421052632e-06, | |
| "loss": 0.0883, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.395, | |
| "grad_norm": 2.4722604751586914, | |
| "learning_rate": 7.964473684210527e-06, | |
| "loss": 0.0825, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.3975, | |
| "grad_norm": 3.6939098834991455, | |
| "learning_rate": 7.931578947368422e-06, | |
| "loss": 0.1424, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 3.4190080165863037, | |
| "learning_rate": 7.898684210526316e-06, | |
| "loss": 0.1825, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.3023500144481659, | |
| "eval_runtime": 108.6852, | |
| "eval_samples_per_second": 15.559, | |
| "eval_steps_per_second": 0.975, | |
| "eval_wer": 12.395404021481205, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4025, | |
| "grad_norm": 2.4307820796966553, | |
| "learning_rate": 7.86578947368421e-06, | |
| "loss": 0.1358, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.405, | |
| "grad_norm": 4.717021942138672, | |
| "learning_rate": 7.832894736842105e-06, | |
| "loss": 0.1963, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.4075, | |
| "grad_norm": 3.309920310974121, | |
| "learning_rate": 7.8e-06, | |
| "loss": 0.1378, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 2.411762237548828, | |
| "learning_rate": 7.767105263157894e-06, | |
| "loss": 0.1299, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.4125, | |
| "grad_norm": 2.3351423740386963, | |
| "learning_rate": 7.734210526315789e-06, | |
| "loss": 0.0917, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.415, | |
| "grad_norm": 2.250337839126587, | |
| "learning_rate": 7.701315789473684e-06, | |
| "loss": 0.0798, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.4175, | |
| "grad_norm": 1.9465301036834717, | |
| "learning_rate": 7.668421052631578e-06, | |
| "loss": 0.082, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1.652019739151001, | |
| "learning_rate": 7.635526315789474e-06, | |
| "loss": 0.076, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.4225, | |
| "grad_norm": 1.7151554822921753, | |
| "learning_rate": 7.602631578947368e-06, | |
| "loss": 0.0823, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.425, | |
| "grad_norm": 2.1287248134613037, | |
| "learning_rate": 7.569736842105264e-06, | |
| "loss": 0.0786, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.4275, | |
| "grad_norm": 3.07344388961792, | |
| "learning_rate": 7.536842105263159e-06, | |
| "loss": 0.088, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 3.5005462169647217, | |
| "learning_rate": 7.5039473684210535e-06, | |
| "loss": 0.1402, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.4325, | |
| "grad_norm": 2.705017328262329, | |
| "learning_rate": 7.471052631578948e-06, | |
| "loss": 0.1201, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.435, | |
| "grad_norm": 3.2306909561157227, | |
| "learning_rate": 7.4381578947368426e-06, | |
| "loss": 0.1263, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.4375, | |
| "grad_norm": 1.8453395366668701, | |
| "learning_rate": 7.405263157894737e-06, | |
| "loss": 0.0749, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 1.7151731252670288, | |
| "learning_rate": 7.372368421052632e-06, | |
| "loss": 0.067, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.4425, | |
| "grad_norm": 2.109825372695923, | |
| "learning_rate": 7.339473684210526e-06, | |
| "loss": 0.0773, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.445, | |
| "grad_norm": 2.3926897048950195, | |
| "learning_rate": 7.306578947368422e-06, | |
| "loss": 0.0827, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.4475, | |
| "grad_norm": 2.0515196323394775, | |
| "learning_rate": 7.273684210526316e-06, | |
| "loss": 0.0916, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 2.471918821334839, | |
| "learning_rate": 7.240789473684211e-06, | |
| "loss": 0.0857, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.4525, | |
| "grad_norm": 2.649667263031006, | |
| "learning_rate": 7.207894736842105e-06, | |
| "loss": 0.1089, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.455, | |
| "grad_norm": 3.062638998031616, | |
| "learning_rate": 7.175e-06, | |
| "loss": 0.1376, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.4575, | |
| "grad_norm": 3.592533826828003, | |
| "learning_rate": 7.1421052631578945e-06, | |
| "loss": 0.1378, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 3.7652299404144287, | |
| "learning_rate": 7.109210526315791e-06, | |
| "loss": 0.1408, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.4625, | |
| "grad_norm": 3.15897798538208, | |
| "learning_rate": 7.076315789473685e-06, | |
| "loss": 0.1446, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.465, | |
| "grad_norm": 3.27856707572937, | |
| "learning_rate": 7.04342105263158e-06, | |
| "loss": 0.1358, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.4675, | |
| "grad_norm": 3.6503591537475586, | |
| "learning_rate": 7.010526315789474e-06, | |
| "loss": 0.1236, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 2.3351187705993652, | |
| "learning_rate": 6.977631578947369e-06, | |
| "loss": 0.109, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.4725, | |
| "grad_norm": 3.16107439994812, | |
| "learning_rate": 6.9447368421052635e-06, | |
| "loss": 0.1162, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.475, | |
| "grad_norm": 2.666114091873169, | |
| "learning_rate": 6.911842105263158e-06, | |
| "loss": 0.1169, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.4775, | |
| "grad_norm": 1.9524102210998535, | |
| "learning_rate": 6.878947368421053e-06, | |
| "loss": 0.0837, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.8562246561050415, | |
| "learning_rate": 6.846052631578947e-06, | |
| "loss": 0.0907, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.4825, | |
| "grad_norm": 2.4069790840148926, | |
| "learning_rate": 6.813157894736842e-06, | |
| "loss": 0.075, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.485, | |
| "grad_norm": 2.5987889766693115, | |
| "learning_rate": 6.780263157894736e-06, | |
| "loss": 0.0944, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.4875, | |
| "grad_norm": 2.7949743270874023, | |
| "learning_rate": 6.747368421052632e-06, | |
| "loss": 0.1001, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 3.0958948135375977, | |
| "learning_rate": 6.714473684210526e-06, | |
| "loss": 0.112, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.4925, | |
| "grad_norm": 2.0432355403900146, | |
| "learning_rate": 6.681578947368422e-06, | |
| "loss": 0.1006, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.495, | |
| "grad_norm": 2.425602912902832, | |
| "learning_rate": 6.648684210526316e-06, | |
| "loss": 0.0853, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.4975, | |
| "grad_norm": 2.167152166366577, | |
| "learning_rate": 6.615789473684212e-06, | |
| "loss": 0.0809, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 2.699022054672241, | |
| "learning_rate": 6.582894736842106e-06, | |
| "loss": 0.0775, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_loss": 0.3198271691799164, | |
| "eval_runtime": 108.5474, | |
| "eval_samples_per_second": 15.578, | |
| "eval_steps_per_second": 0.977, | |
| "eval_wer": 11.877107530910454, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5025, | |
| "grad_norm": 2.366492748260498, | |
| "learning_rate": 6.550000000000001e-06, | |
| "loss": 0.1144, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.505, | |
| "grad_norm": 3.174971342086792, | |
| "learning_rate": 6.517105263157895e-06, | |
| "loss": 0.104, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.5075, | |
| "grad_norm": 3.391756296157837, | |
| "learning_rate": 6.48421052631579e-06, | |
| "loss": 0.1545, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 3.718075752258301, | |
| "learning_rate": 6.4513157894736845e-06, | |
| "loss": 0.1432, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.5125, | |
| "grad_norm": 3.2013394832611084, | |
| "learning_rate": 6.418421052631579e-06, | |
| "loss": 0.1258, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.515, | |
| "grad_norm": 3.259122610092163, | |
| "learning_rate": 6.385526315789474e-06, | |
| "loss": 0.1265, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.5175, | |
| "grad_norm": 2.2417266368865967, | |
| "learning_rate": 6.352631578947368e-06, | |
| "loss": 0.1043, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 2.150083541870117, | |
| "learning_rate": 6.319736842105263e-06, | |
| "loss": 0.0687, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.5225, | |
| "grad_norm": 1.597461462020874, | |
| "learning_rate": 6.286842105263157e-06, | |
| "loss": 0.0719, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.525, | |
| "grad_norm": 1.995166540145874, | |
| "learning_rate": 6.253947368421052e-06, | |
| "loss": 0.0761, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.5275, | |
| "grad_norm": 1.8248108625411987, | |
| "learning_rate": 6.221052631578947e-06, | |
| "loss": 0.0777, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 2.475926637649536, | |
| "learning_rate": 6.188157894736842e-06, | |
| "loss": 0.08, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.5325, | |
| "grad_norm": 2.4988951683044434, | |
| "learning_rate": 6.155263157894737e-06, | |
| "loss": 0.0781, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.535, | |
| "grad_norm": 3.3215720653533936, | |
| "learning_rate": 6.122368421052632e-06, | |
| "loss": 0.1015, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.5375, | |
| "grad_norm": 3.0651330947875977, | |
| "learning_rate": 6.089473684210527e-06, | |
| "loss": 0.1196, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 2.73099684715271, | |
| "learning_rate": 6.056578947368422e-06, | |
| "loss": 0.1132, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.5425, | |
| "grad_norm": 3.1001064777374268, | |
| "learning_rate": 6.023684210526316e-06, | |
| "loss": 0.1152, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.545, | |
| "grad_norm": 2.7154345512390137, | |
| "learning_rate": 5.990789473684211e-06, | |
| "loss": 0.1034, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.5475, | |
| "grad_norm": 2.9686317443847656, | |
| "learning_rate": 5.9578947368421055e-06, | |
| "loss": 0.0817, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 2.079238176345825, | |
| "learning_rate": 5.925e-06, | |
| "loss": 0.0863, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.5525, | |
| "grad_norm": 3.6043894290924072, | |
| "learning_rate": 5.8921052631578954e-06, | |
| "loss": 0.109, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.555, | |
| "grad_norm": 3.189924716949463, | |
| "learning_rate": 5.85921052631579e-06, | |
| "loss": 0.1268, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.5575, | |
| "grad_norm": 2.064168691635132, | |
| "learning_rate": 5.8263157894736846e-06, | |
| "loss": 0.0964, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 1.738265872001648, | |
| "learning_rate": 5.793421052631579e-06, | |
| "loss": 0.0625, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.5625, | |
| "grad_norm": 2.038506269454956, | |
| "learning_rate": 5.760526315789474e-06, | |
| "loss": 0.0497, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.565, | |
| "grad_norm": 2.134082078933716, | |
| "learning_rate": 5.727631578947368e-06, | |
| "loss": 0.051, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.5675, | |
| "grad_norm": 2.1143264770507812, | |
| "learning_rate": 5.694736842105263e-06, | |
| "loss": 0.0672, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 2.5272130966186523, | |
| "learning_rate": 5.661842105263158e-06, | |
| "loss": 0.0733, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.5725, | |
| "grad_norm": 1.941775918006897, | |
| "learning_rate": 5.628947368421053e-06, | |
| "loss": 0.0721, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.575, | |
| "grad_norm": 1.680059790611267, | |
| "learning_rate": 5.596052631578947e-06, | |
| "loss": 0.0727, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.5775, | |
| "grad_norm": 1.909376859664917, | |
| "learning_rate": 5.563157894736843e-06, | |
| "loss": 0.0708, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1.746079444885254, | |
| "learning_rate": 5.530263157894737e-06, | |
| "loss": 0.0668, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.5825, | |
| "grad_norm": 1.6772959232330322, | |
| "learning_rate": 5.497368421052632e-06, | |
| "loss": 0.061, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.585, | |
| "grad_norm": 2.339190721511841, | |
| "learning_rate": 5.4644736842105264e-06, | |
| "loss": 0.0665, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.5875, | |
| "grad_norm": 1.8088990449905396, | |
| "learning_rate": 5.431578947368422e-06, | |
| "loss": 0.07, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 2.6846957206726074, | |
| "learning_rate": 5.398684210526316e-06, | |
| "loss": 0.0866, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.5925, | |
| "grad_norm": 3.235233783721924, | |
| "learning_rate": 5.365789473684211e-06, | |
| "loss": 0.1282, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.595, | |
| "grad_norm": 2.8694686889648438, | |
| "learning_rate": 5.3328947368421055e-06, | |
| "loss": 0.1188, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.5975, | |
| "grad_norm": 2.7638120651245117, | |
| "learning_rate": 5.3e-06, | |
| "loss": 0.0998, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 2.570558786392212, | |
| "learning_rate": 5.267105263157895e-06, | |
| "loss": 0.0975, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.2923527956008911, | |
| "eval_runtime": 108.3631, | |
| "eval_samples_per_second": 15.605, | |
| "eval_steps_per_second": 0.978, | |
| "eval_wer": 11.258898463844137, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6025, | |
| "grad_norm": 2.571687698364258, | |
| "learning_rate": 5.23421052631579e-06, | |
| "loss": 0.0904, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.605, | |
| "grad_norm": 2.654676914215088, | |
| "learning_rate": 5.201315789473685e-06, | |
| "loss": 0.0793, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.6075, | |
| "grad_norm": 1.8399786949157715, | |
| "learning_rate": 5.168421052631579e-06, | |
| "loss": 0.0708, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 1.8271868228912354, | |
| "learning_rate": 5.135526315789474e-06, | |
| "loss": 0.0672, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.6125, | |
| "grad_norm": 2.6549460887908936, | |
| "learning_rate": 5.102631578947368e-06, | |
| "loss": 0.0714, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.615, | |
| "grad_norm": 2.3854265213012695, | |
| "learning_rate": 5.069736842105263e-06, | |
| "loss": 0.0772, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.6175, | |
| "grad_norm": 2.193765163421631, | |
| "learning_rate": 5.036842105263158e-06, | |
| "loss": 0.0769, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 2.371039390563965, | |
| "learning_rate": 5.003947368421053e-06, | |
| "loss": 0.0763, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.6225, | |
| "grad_norm": 2.19585919380188, | |
| "learning_rate": 4.971052631578948e-06, | |
| "loss": 0.0718, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 2.88238787651062, | |
| "learning_rate": 4.938157894736843e-06, | |
| "loss": 0.0736, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.6275, | |
| "grad_norm": 2.133556604385376, | |
| "learning_rate": 4.905263157894737e-06, | |
| "loss": 0.0835, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 2.584014654159546, | |
| "learning_rate": 4.872368421052632e-06, | |
| "loss": 0.0899, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.6325, | |
| "grad_norm": 2.235517740249634, | |
| "learning_rate": 4.8394736842105265e-06, | |
| "loss": 0.1156, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.635, | |
| "grad_norm": 3.3249387741088867, | |
| "learning_rate": 4.806578947368421e-06, | |
| "loss": 0.1104, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.6375, | |
| "grad_norm": 2.0971710681915283, | |
| "learning_rate": 4.7736842105263164e-06, | |
| "loss": 0.0994, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 1.9291256666183472, | |
| "learning_rate": 4.740789473684211e-06, | |
| "loss": 0.0823, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.6425, | |
| "grad_norm": 2.2015464305877686, | |
| "learning_rate": 4.7078947368421056e-06, | |
| "loss": 0.0701, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.645, | |
| "grad_norm": 2.357741355895996, | |
| "learning_rate": 4.675e-06, | |
| "loss": 0.0684, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.6475, | |
| "grad_norm": 3.1389389038085938, | |
| "learning_rate": 4.642105263157895e-06, | |
| "loss": 0.0993, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 2.419999122619629, | |
| "learning_rate": 4.609210526315789e-06, | |
| "loss": 0.0944, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.6525, | |
| "grad_norm": 3.8924739360809326, | |
| "learning_rate": 4.576315789473684e-06, | |
| "loss": 0.1216, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.655, | |
| "grad_norm": 2.7731151580810547, | |
| "learning_rate": 4.543421052631579e-06, | |
| "loss": 0.1094, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.6575, | |
| "grad_norm": 2.8039772510528564, | |
| "learning_rate": 4.510526315789474e-06, | |
| "loss": 0.1107, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 2.598423719406128, | |
| "learning_rate": 4.477631578947368e-06, | |
| "loss": 0.1028, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.6625, | |
| "grad_norm": 2.9384443759918213, | |
| "learning_rate": 4.444736842105264e-06, | |
| "loss": 0.1046, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.665, | |
| "grad_norm": 2.54249906539917, | |
| "learning_rate": 4.411842105263158e-06, | |
| "loss": 0.12, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.6675, | |
| "grad_norm": 2.804525136947632, | |
| "learning_rate": 4.378947368421053e-06, | |
| "loss": 0.1024, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 1.912115454673767, | |
| "learning_rate": 4.3460526315789474e-06, | |
| "loss": 0.0956, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.6725, | |
| "grad_norm": 1.8333637714385986, | |
| "learning_rate": 4.313157894736843e-06, | |
| "loss": 0.0724, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.675, | |
| "grad_norm": 2.1709723472595215, | |
| "learning_rate": 4.280263157894737e-06, | |
| "loss": 0.068, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.6775, | |
| "grad_norm": 2.6692330837249756, | |
| "learning_rate": 4.247368421052632e-06, | |
| "loss": 0.0809, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 2.787583589553833, | |
| "learning_rate": 4.2144736842105265e-06, | |
| "loss": 0.1336, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.6825, | |
| "grad_norm": 3.7850124835968018, | |
| "learning_rate": 4.181578947368421e-06, | |
| "loss": 0.1178, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.685, | |
| "grad_norm": 3.4124090671539307, | |
| "learning_rate": 4.148684210526316e-06, | |
| "loss": 0.1211, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.6875, | |
| "grad_norm": 2.470259189605713, | |
| "learning_rate": 4.115789473684211e-06, | |
| "loss": 0.1133, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 3.100062847137451, | |
| "learning_rate": 4.082894736842106e-06, | |
| "loss": 0.1083, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.6925, | |
| "grad_norm": 2.644653558731079, | |
| "learning_rate": 4.05e-06, | |
| "loss": 0.1019, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.695, | |
| "grad_norm": 2.8097853660583496, | |
| "learning_rate": 4.018421052631579e-06, | |
| "loss": 0.1068, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.6975, | |
| "grad_norm": 3.0686898231506348, | |
| "learning_rate": 3.985526315789474e-06, | |
| "loss": 0.1187, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 2.7191102504730225, | |
| "learning_rate": 3.952631578947369e-06, | |
| "loss": 0.1132, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 0.2969239056110382, | |
| "eval_runtime": 108.3888, | |
| "eval_samples_per_second": 15.601, | |
| "eval_steps_per_second": 0.978, | |
| "eval_wer": 10.846759085799926, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7025, | |
| "grad_norm": 4.9354567527771, | |
| "learning_rate": 3.922368421052632e-06, | |
| "loss": 0.2171, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.705, | |
| "grad_norm": 4.480522632598877, | |
| "learning_rate": 3.889473684210526e-06, | |
| "loss": 0.3333, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.7075, | |
| "grad_norm": 4.159786701202393, | |
| "learning_rate": 3.856578947368422e-06, | |
| "loss": 0.3131, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 3.261507987976074, | |
| "learning_rate": 3.823684210526316e-06, | |
| "loss": 0.2489, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.7125, | |
| "grad_norm": 2.8575055599212646, | |
| "learning_rate": 3.7907894736842108e-06, | |
| "loss": 0.147, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.715, | |
| "grad_norm": 3.800828218460083, | |
| "learning_rate": 3.7578947368421053e-06, | |
| "loss": 0.1173, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.7175, | |
| "grad_norm": 2.675981283187866, | |
| "learning_rate": 3.725e-06, | |
| "loss": 0.1039, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 2.7404372692108154, | |
| "learning_rate": 3.6921052631578945e-06, | |
| "loss": 0.0817, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.7225, | |
| "grad_norm": 2.429400682449341, | |
| "learning_rate": 3.65921052631579e-06, | |
| "loss": 0.0816, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.725, | |
| "grad_norm": 1.525938630104065, | |
| "learning_rate": 3.6263157894736844e-06, | |
| "loss": 0.0825, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.7275, | |
| "grad_norm": 1.7473913431167603, | |
| "learning_rate": 3.5934210526315794e-06, | |
| "loss": 0.071, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 1.690190315246582, | |
| "learning_rate": 3.560526315789474e-06, | |
| "loss": 0.0582, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.7325, | |
| "grad_norm": 1.7410528659820557, | |
| "learning_rate": 3.5276315789473685e-06, | |
| "loss": 0.0628, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.735, | |
| "grad_norm": 2.584449052810669, | |
| "learning_rate": 3.494736842105263e-06, | |
| "loss": 0.081, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.7375, | |
| "grad_norm": 2.2174675464630127, | |
| "learning_rate": 3.4618421052631577e-06, | |
| "loss": 0.1095, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 2.5199625492095947, | |
| "learning_rate": 3.428947368421053e-06, | |
| "loss": 0.0859, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.7425, | |
| "grad_norm": 1.9601330757141113, | |
| "learning_rate": 3.3960526315789476e-06, | |
| "loss": 0.0973, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.745, | |
| "grad_norm": 1.9718409776687622, | |
| "learning_rate": 3.363157894736842e-06, | |
| "loss": 0.069, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.7475, | |
| "grad_norm": 2.3192267417907715, | |
| "learning_rate": 3.330263157894737e-06, | |
| "loss": 0.0662, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 2.357396364212036, | |
| "learning_rate": 3.2973684210526318e-06, | |
| "loss": 0.0625, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.7525, | |
| "grad_norm": 2.821035623550415, | |
| "learning_rate": 3.2644736842105263e-06, | |
| "loss": 0.1253, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.755, | |
| "grad_norm": 3.693251848220825, | |
| "learning_rate": 3.231578947368421e-06, | |
| "loss": 0.165, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.7575, | |
| "grad_norm": 3.8312151432037354, | |
| "learning_rate": 3.1986842105263163e-06, | |
| "loss": 0.1901, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 1.4963386058807373, | |
| "learning_rate": 3.165789473684211e-06, | |
| "loss": 0.1203, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.7625, | |
| "grad_norm": 2.3876097202301025, | |
| "learning_rate": 3.1328947368421054e-06, | |
| "loss": 0.08, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.765, | |
| "grad_norm": 1.5950793027877808, | |
| "learning_rate": 3.1e-06, | |
| "loss": 0.066, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.7675, | |
| "grad_norm": 2.980989694595337, | |
| "learning_rate": 3.067105263157895e-06, | |
| "loss": 0.0912, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 4.786663055419922, | |
| "learning_rate": 3.03421052631579e-06, | |
| "loss": 0.1057, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.7725, | |
| "grad_norm": 3.2149651050567627, | |
| "learning_rate": 3.0013157894736845e-06, | |
| "loss": 0.1162, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 0.775, | |
| "grad_norm": 2.1451265811920166, | |
| "learning_rate": 2.968421052631579e-06, | |
| "loss": 0.1024, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.7775, | |
| "grad_norm": 2.4378089904785156, | |
| "learning_rate": 2.935526315789474e-06, | |
| "loss": 0.0731, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 2.8874340057373047, | |
| "learning_rate": 2.9026315789473686e-06, | |
| "loss": 0.0885, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.7825, | |
| "grad_norm": 2.4116005897521973, | |
| "learning_rate": 2.869736842105263e-06, | |
| "loss": 0.183, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 0.785, | |
| "grad_norm": 1.9551997184753418, | |
| "learning_rate": 2.8368421052631577e-06, | |
| "loss": 0.0911, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.7875, | |
| "grad_norm": 1.4724156856536865, | |
| "learning_rate": 2.8039473684210527e-06, | |
| "loss": 0.073, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1.8941609859466553, | |
| "learning_rate": 2.7710526315789477e-06, | |
| "loss": 0.0697, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.7925, | |
| "grad_norm": 1.880815863609314, | |
| "learning_rate": 2.7381578947368423e-06, | |
| "loss": 0.0722, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 0.795, | |
| "grad_norm": 1.5334910154342651, | |
| "learning_rate": 2.7052631578947372e-06, | |
| "loss": 0.0631, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.7975, | |
| "grad_norm": 1.5681651830673218, | |
| "learning_rate": 2.672368421052632e-06, | |
| "loss": 0.0718, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 2.644123077392578, | |
| "learning_rate": 2.6394736842105264e-06, | |
| "loss": 0.0852, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.22370582818984985, | |
| "eval_runtime": 108.5849, | |
| "eval_samples_per_second": 15.573, | |
| "eval_steps_per_second": 0.976, | |
| "eval_wer": 9.772698888472586, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8025, | |
| "grad_norm": 2.5814223289489746, | |
| "learning_rate": 2.606578947368421e-06, | |
| "loss": 0.0991, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 0.805, | |
| "grad_norm": 2.196101188659668, | |
| "learning_rate": 2.573684210526316e-06, | |
| "loss": 0.1063, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.8075, | |
| "grad_norm": 2.1308398246765137, | |
| "learning_rate": 2.5407894736842105e-06, | |
| "loss": 0.082, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 1.9144341945648193, | |
| "learning_rate": 2.5078947368421055e-06, | |
| "loss": 0.0754, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.8125, | |
| "grad_norm": 1.7134053707122803, | |
| "learning_rate": 2.4750000000000004e-06, | |
| "loss": 0.0608, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 0.815, | |
| "grad_norm": 2.3351683616638184, | |
| "learning_rate": 2.442105263157895e-06, | |
| "loss": 0.0626, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.8175, | |
| "grad_norm": 1.5327272415161133, | |
| "learning_rate": 2.4092105263157896e-06, | |
| "loss": 0.0587, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 2.235727071762085, | |
| "learning_rate": 2.3763157894736846e-06, | |
| "loss": 0.0474, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.8225, | |
| "grad_norm": 1.5293081998825073, | |
| "learning_rate": 2.343421052631579e-06, | |
| "loss": 0.0619, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 0.825, | |
| "grad_norm": 2.498485565185547, | |
| "learning_rate": 2.3105263157894737e-06, | |
| "loss": 0.0542, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.8275, | |
| "grad_norm": 1.7372788190841675, | |
| "learning_rate": 2.2776315789473682e-06, | |
| "loss": 0.0581, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 1.3887490034103394, | |
| "learning_rate": 2.2447368421052632e-06, | |
| "loss": 0.0592, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.8325, | |
| "grad_norm": 2.6547152996063232, | |
| "learning_rate": 2.2118421052631578e-06, | |
| "loss": 0.0818, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 0.835, | |
| "grad_norm": 2.8091819286346436, | |
| "learning_rate": 2.1789473684210528e-06, | |
| "loss": 0.1006, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.8375, | |
| "grad_norm": 2.7749269008636475, | |
| "learning_rate": 2.1460526315789478e-06, | |
| "loss": 0.1182, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 2.2278335094451904, | |
| "learning_rate": 2.1131578947368423e-06, | |
| "loss": 0.0957, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.8425, | |
| "grad_norm": 2.1413509845733643, | |
| "learning_rate": 2.080263157894737e-06, | |
| "loss": 0.0707, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 0.845, | |
| "grad_norm": 1.9786673784255981, | |
| "learning_rate": 2.0473684210526314e-06, | |
| "loss": 0.0649, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.8475, | |
| "grad_norm": 2.812910318374634, | |
| "learning_rate": 2.0144736842105264e-06, | |
| "loss": 0.074, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 2.2044918537139893, | |
| "learning_rate": 1.981578947368421e-06, | |
| "loss": 0.0828, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.8525, | |
| "grad_norm": 2.3589258193969727, | |
| "learning_rate": 1.9486842105263155e-06, | |
| "loss": 0.087, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 0.855, | |
| "grad_norm": 4.131094455718994, | |
| "learning_rate": 1.9157894736842105e-06, | |
| "loss": 0.1186, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.8575, | |
| "grad_norm": 3.9992008209228516, | |
| "learning_rate": 1.8828947368421053e-06, | |
| "loss": 0.2169, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 5.6015849113464355, | |
| "learning_rate": 1.85e-06, | |
| "loss": 0.303, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.8625, | |
| "grad_norm": 5.061160087585449, | |
| "learning_rate": 1.817105263157895e-06, | |
| "loss": 0.3653, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.865, | |
| "grad_norm": 2.206181287765503, | |
| "learning_rate": 1.7842105263157896e-06, | |
| "loss": 0.2101, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.8675, | |
| "grad_norm": 3.2147305011749268, | |
| "learning_rate": 1.7513157894736842e-06, | |
| "loss": 0.1292, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 3.8435158729553223, | |
| "learning_rate": 1.718421052631579e-06, | |
| "loss": 0.1566, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.8725, | |
| "grad_norm": 2.289605140686035, | |
| "learning_rate": 1.685526315789474e-06, | |
| "loss": 0.1001, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 1.7797821760177612, | |
| "learning_rate": 1.6526315789473685e-06, | |
| "loss": 0.0631, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.8775, | |
| "grad_norm": 1.8683706521987915, | |
| "learning_rate": 1.619736842105263e-06, | |
| "loss": 0.0684, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 1.9965925216674805, | |
| "learning_rate": 1.586842105263158e-06, | |
| "loss": 0.0761, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.8825, | |
| "grad_norm": 2.9362261295318604, | |
| "learning_rate": 1.5539473684210528e-06, | |
| "loss": 0.0933, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 0.885, | |
| "grad_norm": 2.6258389949798584, | |
| "learning_rate": 1.5210526315789474e-06, | |
| "loss": 0.0951, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.8875, | |
| "grad_norm": 2.6913561820983887, | |
| "learning_rate": 1.4881578947368422e-06, | |
| "loss": 0.0894, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 2.097146511077881, | |
| "learning_rate": 1.455263157894737e-06, | |
| "loss": 0.0642, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.8925, | |
| "grad_norm": 2.0721096992492676, | |
| "learning_rate": 1.4223684210526317e-06, | |
| "loss": 0.0653, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 0.895, | |
| "grad_norm": 2.383824586868286, | |
| "learning_rate": 1.3894736842105265e-06, | |
| "loss": 0.0673, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.8975, | |
| "grad_norm": 2.066480875015259, | |
| "learning_rate": 1.356578947368421e-06, | |
| "loss": 0.0567, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 1.9253851175308228, | |
| "learning_rate": 1.3236842105263158e-06, | |
| "loss": 0.0585, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_loss": 0.23174172639846802, | |
| "eval_runtime": 108.0471, | |
| "eval_samples_per_second": 15.651, | |
| "eval_steps_per_second": 0.981, | |
| "eval_wer": 9.62907455976021, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9025, | |
| "grad_norm": 2.2203736305236816, | |
| "learning_rate": 1.2907894736842106e-06, | |
| "loss": 0.0696, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 0.905, | |
| "grad_norm": 2.5881643295288086, | |
| "learning_rate": 1.2578947368421054e-06, | |
| "loss": 0.0802, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.9075, | |
| "grad_norm": 3.149264097213745, | |
| "learning_rate": 1.2250000000000001e-06, | |
| "loss": 0.1068, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 2.7941627502441406, | |
| "learning_rate": 1.1921052631578947e-06, | |
| "loss": 0.1036, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.9125, | |
| "grad_norm": 1.6882115602493286, | |
| "learning_rate": 1.1592105263157895e-06, | |
| "loss": 0.0818, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 0.915, | |
| "grad_norm": 1.7995328903198242, | |
| "learning_rate": 1.1263157894736842e-06, | |
| "loss": 0.0817, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.9175, | |
| "grad_norm": 2.3154258728027344, | |
| "learning_rate": 1.093421052631579e-06, | |
| "loss": 0.0752, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 2.428879737854004, | |
| "learning_rate": 1.0605263157894738e-06, | |
| "loss": 0.0805, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.9225, | |
| "grad_norm": 3.1283998489379883, | |
| "learning_rate": 1.0276315789473683e-06, | |
| "loss": 0.0957, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 0.925, | |
| "grad_norm": 2.8848423957824707, | |
| "learning_rate": 9.947368421052633e-07, | |
| "loss": 0.1028, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.9275, | |
| "grad_norm": 2.741641044616699, | |
| "learning_rate": 9.618421052631579e-07, | |
| "loss": 0.1203, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 1.3423601388931274, | |
| "learning_rate": 9.289473684210527e-07, | |
| "loss": 0.0712, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.9325, | |
| "grad_norm": 1.9067927598953247, | |
| "learning_rate": 8.960526315789474e-07, | |
| "loss": 0.076, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 0.935, | |
| "grad_norm": 1.605100154876709, | |
| "learning_rate": 8.631578947368421e-07, | |
| "loss": 0.0624, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.9375, | |
| "grad_norm": 2.283745050430298, | |
| "learning_rate": 8.302631578947369e-07, | |
| "loss": 0.0722, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 2.803049087524414, | |
| "learning_rate": 7.973684210526315e-07, | |
| "loss": 0.0947, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.9425, | |
| "grad_norm": 2.6914260387420654, | |
| "learning_rate": 7.644736842105263e-07, | |
| "loss": 0.1081, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 0.945, | |
| "grad_norm": 1.8262979984283447, | |
| "learning_rate": 7.315789473684211e-07, | |
| "loss": 0.2328, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.9475, | |
| "grad_norm": 1.848897099494934, | |
| "learning_rate": 6.986842105263158e-07, | |
| "loss": 0.1125, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 2.561948299407959, | |
| "learning_rate": 6.657894736842105e-07, | |
| "loss": 0.0851, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.9525, | |
| "grad_norm": 3.199554204940796, | |
| "learning_rate": 6.328947368421053e-07, | |
| "loss": 0.1023, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 0.955, | |
| "grad_norm": 2.5224554538726807, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 0.1093, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.9575, | |
| "grad_norm": 3.5115911960601807, | |
| "learning_rate": 5.671052631578947e-07, | |
| "loss": 0.1171, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 2.1708128452301025, | |
| "learning_rate": 5.342105263157895e-07, | |
| "loss": 0.0901, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.9625, | |
| "grad_norm": 2.0158002376556396, | |
| "learning_rate": 5.013157894736842e-07, | |
| "loss": 0.0714, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 0.965, | |
| "grad_norm": 1.924566388130188, | |
| "learning_rate": 4.68421052631579e-07, | |
| "loss": 0.0555, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.9675, | |
| "grad_norm": 1.6960805654525757, | |
| "learning_rate": 4.3552631578947373e-07, | |
| "loss": 0.0598, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 1.8255702257156372, | |
| "learning_rate": 4.0263157894736845e-07, | |
| "loss": 0.0606, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.9725, | |
| "grad_norm": 1.2460139989852905, | |
| "learning_rate": 3.6973684210526317e-07, | |
| "loss": 0.0581, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 0.975, | |
| "grad_norm": 1.5349912643432617, | |
| "learning_rate": 3.3684210526315795e-07, | |
| "loss": 0.0628, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.9775, | |
| "grad_norm": 1.9929566383361816, | |
| "learning_rate": 3.0394736842105267e-07, | |
| "loss": 0.0689, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 2.2622244358062744, | |
| "learning_rate": 2.710526315789474e-07, | |
| "loss": 0.0945, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.9825, | |
| "grad_norm": 3.0095293521881104, | |
| "learning_rate": 2.3815789473684213e-07, | |
| "loss": 0.1028, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 0.985, | |
| "grad_norm": 2.2643580436706543, | |
| "learning_rate": 2.0526315789473685e-07, | |
| "loss": 0.0892, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.9875, | |
| "grad_norm": 2.1976678371429443, | |
| "learning_rate": 1.723684210526316e-07, | |
| "loss": 0.0904, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 4.183482646942139, | |
| "learning_rate": 1.3947368421052632e-07, | |
| "loss": 0.1343, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.9925, | |
| "grad_norm": 1.8398394584655762, | |
| "learning_rate": 1.0657894736842105e-07, | |
| "loss": 0.1207, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 0.995, | |
| "grad_norm": 2.1588056087493896, | |
| "learning_rate": 7.36842105263158e-08, | |
| "loss": 0.0644, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 0.9975, | |
| "grad_norm": 1.9245156049728394, | |
| "learning_rate": 4.078947368421053e-08, | |
| "loss": 0.0659, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1.6552762985229492, | |
| "learning_rate": 7.894736842105264e-09, | |
| "loss": 0.0654, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.23531682789325714, | |
| "eval_runtime": 109.2289, | |
| "eval_samples_per_second": 15.481, | |
| "eval_steps_per_second": 0.97, | |
| "eval_wer": 9.547895591357562, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 10000, | |
| "total_flos": 9.23473281024e+19, | |
| "train_loss": 0.14711211260557175, | |
| "train_runtime": 12741.5974, | |
| "train_samples_per_second": 25.115, | |
| "train_steps_per_second": 0.785 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 9.23473281024e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |