| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.034174125305126, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01017087062652563, | |
| "grad_norm": 16.184080123901367, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 1.6215, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02034174125305126, | |
| "grad_norm": 14.08956241607666, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 1.2721, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.030512611879576892, | |
| "grad_norm": 11.49959945678711, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.9884, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04068348250610252, | |
| "grad_norm": 8.895913124084473, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.8726, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.050854353132628156, | |
| "grad_norm": 9.52476692199707, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 0.8062, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.061025223759153785, | |
| "grad_norm": 9.508084297180176, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.7792, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07119609438567942, | |
| "grad_norm": 9.352806091308594, | |
| "learning_rate": 3.48e-06, | |
| "loss": 0.7541, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.08136696501220504, | |
| "grad_norm": 10.18264102935791, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 0.6983, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09153783563873068, | |
| "grad_norm": 10.840176582336426, | |
| "learning_rate": 4.48e-06, | |
| "loss": 0.7416, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.10170870626525631, | |
| "grad_norm": 9.403505325317383, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 0.7154, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.11187957689178193, | |
| "grad_norm": 8.170647621154785, | |
| "learning_rate": 5.480000000000001e-06, | |
| "loss": 0.6564, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.12205044751830757, | |
| "grad_norm": 9.974289894104004, | |
| "learning_rate": 5.98e-06, | |
| "loss": 0.6401, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1322213181448332, | |
| "grad_norm": 10.3145112991333, | |
| "learning_rate": 6.480000000000001e-06, | |
| "loss": 0.6564, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.14239218877135884, | |
| "grad_norm": 8.214014053344727, | |
| "learning_rate": 6.98e-06, | |
| "loss": 0.6191, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.15256305939788445, | |
| "grad_norm": 8.800423622131348, | |
| "learning_rate": 7.48e-06, | |
| "loss": 0.6492, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.16273393002441008, | |
| "grad_norm": 8.481072425842285, | |
| "learning_rate": 7.980000000000002e-06, | |
| "loss": 0.6295, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.17290480065093572, | |
| "grad_norm": 8.669546127319336, | |
| "learning_rate": 8.48e-06, | |
| "loss": 0.6167, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.18307567127746135, | |
| "grad_norm": 7.645444869995117, | |
| "learning_rate": 8.98e-06, | |
| "loss": 0.6021, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.193246541903987, | |
| "grad_norm": 11.719562530517578, | |
| "learning_rate": 9.48e-06, | |
| "loss": 0.6006, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.20341741253051263, | |
| "grad_norm": 9.42712116241455, | |
| "learning_rate": 9.980000000000001e-06, | |
| "loss": 0.5777, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.21358828315703823, | |
| "grad_norm": 7.9745988845825195, | |
| "learning_rate": 9.946666666666667e-06, | |
| "loss": 0.5928, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.22375915378356387, | |
| "grad_norm": 7.762495994567871, | |
| "learning_rate": 9.891111111111113e-06, | |
| "loss": 0.5735, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2339300244100895, | |
| "grad_norm": 8.57645320892334, | |
| "learning_rate": 9.835555555555556e-06, | |
| "loss": 0.5613, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.24410089503661514, | |
| "grad_norm": 7.481997489929199, | |
| "learning_rate": 9.780000000000001e-06, | |
| "loss": 0.5496, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.25427176566314075, | |
| "grad_norm": 6.589779376983643, | |
| "learning_rate": 9.724444444444445e-06, | |
| "loss": 0.5746, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.2644426362896664, | |
| "grad_norm": 8.590812683105469, | |
| "learning_rate": 9.66888888888889e-06, | |
| "loss": 0.5385, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.274613506916192, | |
| "grad_norm": 7.343937397003174, | |
| "learning_rate": 9.613333333333335e-06, | |
| "loss": 0.5432, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.2847843775427177, | |
| "grad_norm": 7.952889919281006, | |
| "learning_rate": 9.557777777777777e-06, | |
| "loss": 0.5213, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.2949552481692433, | |
| "grad_norm": 8.083431243896484, | |
| "learning_rate": 9.502222222222223e-06, | |
| "loss": 0.5367, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.3051261187957689, | |
| "grad_norm": 7.172378063201904, | |
| "learning_rate": 9.446666666666667e-06, | |
| "loss": 0.5317, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.31529698942229456, | |
| "grad_norm": 7.955025672912598, | |
| "learning_rate": 9.391111111111111e-06, | |
| "loss": 0.5208, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.32546786004882017, | |
| "grad_norm": 7.041996479034424, | |
| "learning_rate": 9.335555555555557e-06, | |
| "loss": 0.5123, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.33563873067534583, | |
| "grad_norm": 7.7593607902526855, | |
| "learning_rate": 9.280000000000001e-06, | |
| "loss": 0.5019, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.34580960130187144, | |
| "grad_norm": 9.114460945129395, | |
| "learning_rate": 9.224444444444445e-06, | |
| "loss": 0.5189, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.35598047192839705, | |
| "grad_norm": 8.199577331542969, | |
| "learning_rate": 9.171111111111112e-06, | |
| "loss": 0.4844, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.3661513425549227, | |
| "grad_norm": 7.192209243774414, | |
| "learning_rate": 9.115555555555556e-06, | |
| "loss": 0.4867, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.3763222131814483, | |
| "grad_norm": 7.112555980682373, | |
| "learning_rate": 9.060000000000001e-06, | |
| "loss": 0.4673, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.386493083807974, | |
| "grad_norm": 6.723269462585449, | |
| "learning_rate": 9.004444444444445e-06, | |
| "loss": 0.4886, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.3966639544344996, | |
| "grad_norm": 6.772889137268066, | |
| "learning_rate": 8.94888888888889e-06, | |
| "loss": 0.5116, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.40683482506102525, | |
| "grad_norm": 7.985610485076904, | |
| "learning_rate": 8.893333333333333e-06, | |
| "loss": 0.5072, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.40683482506102525, | |
| "eval_loss": 0.5020337700843811, | |
| "eval_runtime": 1789.7876, | |
| "eval_samples_per_second": 2.18, | |
| "eval_steps_per_second": 0.136, | |
| "eval_wer": 0.35670820052158797, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.41700569568755086, | |
| "grad_norm": 10.103032112121582, | |
| "learning_rate": 8.83777777777778e-06, | |
| "loss": 0.4951, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.42717656631407647, | |
| "grad_norm": 6.815390586853027, | |
| "learning_rate": 8.782222222222223e-06, | |
| "loss": 0.4682, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.43734743694060213, | |
| "grad_norm": 7.63904333114624, | |
| "learning_rate": 8.726666666666667e-06, | |
| "loss": 0.4842, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.44751830756712774, | |
| "grad_norm": 7.750236988067627, | |
| "learning_rate": 8.671111111111113e-06, | |
| "loss": 0.4823, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.4576891781936534, | |
| "grad_norm": 6.178125858306885, | |
| "learning_rate": 8.615555555555555e-06, | |
| "loss": 0.4764, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.467860048820179, | |
| "grad_norm": 6.7011284828186035, | |
| "learning_rate": 8.560000000000001e-06, | |
| "loss": 0.487, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.4780309194467046, | |
| "grad_norm": 6.630568504333496, | |
| "learning_rate": 8.504444444444445e-06, | |
| "loss": 0.444, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.4882017900732303, | |
| "grad_norm": 7.386712074279785, | |
| "learning_rate": 8.448888888888889e-06, | |
| "loss": 0.4372, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.4983726606997559, | |
| "grad_norm": 6.448077201843262, | |
| "learning_rate": 8.393333333333335e-06, | |
| "loss": 0.4474, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.5085435313262815, | |
| "grad_norm": 6.787522315979004, | |
| "learning_rate": 8.337777777777777e-06, | |
| "loss": 0.4692, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.5187144019528072, | |
| "grad_norm": 6.614894866943359, | |
| "learning_rate": 8.282222222222223e-06, | |
| "loss": 0.4401, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.5288852725793328, | |
| "grad_norm": 7.09251594543457, | |
| "learning_rate": 8.226666666666667e-06, | |
| "loss": 0.4419, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.5390561432058584, | |
| "grad_norm": 7.047774791717529, | |
| "learning_rate": 8.171111111111113e-06, | |
| "loss": 0.4323, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.549227013832384, | |
| "grad_norm": 6.616914749145508, | |
| "learning_rate": 8.115555555555557e-06, | |
| "loss": 0.4354, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.5593978844589097, | |
| "grad_norm": 8.16896915435791, | |
| "learning_rate": 8.06e-06, | |
| "loss": 0.4299, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.5695687550854354, | |
| "grad_norm": 7.04884672164917, | |
| "learning_rate": 8.004444444444445e-06, | |
| "loss": 0.4477, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.5797396257119609, | |
| "grad_norm": 6.052219867706299, | |
| "learning_rate": 7.948888888888889e-06, | |
| "loss": 0.4561, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.5899104963384866, | |
| "grad_norm": 6.257965087890625, | |
| "learning_rate": 7.893333333333335e-06, | |
| "loss": 0.4343, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.6000813669650122, | |
| "grad_norm": 7.182107925415039, | |
| "learning_rate": 7.837777777777779e-06, | |
| "loss": 0.4215, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.6102522375915378, | |
| "grad_norm": 6.276819705963135, | |
| "learning_rate": 7.782222222222223e-06, | |
| "loss": 0.4188, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.6204231082180635, | |
| "grad_norm": 6.533692359924316, | |
| "learning_rate": 7.726666666666667e-06, | |
| "loss": 0.4855, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.6305939788445891, | |
| "grad_norm": 6.108315467834473, | |
| "learning_rate": 7.67111111111111e-06, | |
| "loss": 0.4255, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.6407648494711147, | |
| "grad_norm": 6.741276741027832, | |
| "learning_rate": 7.6155555555555564e-06, | |
| "loss": 0.4529, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.6509357200976403, | |
| "grad_norm": 6.165073871612549, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 0.4541, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.661106590724166, | |
| "grad_norm": 7.246621608734131, | |
| "learning_rate": 7.504444444444445e-06, | |
| "loss": 0.4073, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.6712774613506917, | |
| "grad_norm": 7.50054407119751, | |
| "learning_rate": 7.44888888888889e-06, | |
| "loss": 0.4143, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.6814483319772172, | |
| "grad_norm": 6.225817680358887, | |
| "learning_rate": 7.393333333333333e-06, | |
| "loss": 0.438, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.6916192026037429, | |
| "grad_norm": 6.704942226409912, | |
| "learning_rate": 7.337777777777778e-06, | |
| "loss": 0.4209, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.7017900732302685, | |
| "grad_norm": 6.276001930236816, | |
| "learning_rate": 7.282222222222222e-06, | |
| "loss": 0.4033, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.7119609438567941, | |
| "grad_norm": 7.256411075592041, | |
| "learning_rate": 7.226666666666667e-06, | |
| "loss": 0.416, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.7221318144833198, | |
| "grad_norm": 6.347288608551025, | |
| "learning_rate": 7.171111111111112e-06, | |
| "loss": 0.4373, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.7323026851098454, | |
| "grad_norm": 6.241955757141113, | |
| "learning_rate": 7.115555555555557e-06, | |
| "loss": 0.403, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.7424735557363711, | |
| "grad_norm": 5.548867702484131, | |
| "learning_rate": 7.06e-06, | |
| "loss": 0.4034, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.7526444263628966, | |
| "grad_norm": 7.0312113761901855, | |
| "learning_rate": 7.004444444444445e-06, | |
| "loss": 0.4037, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.7628152969894223, | |
| "grad_norm": 7.7608256340026855, | |
| "learning_rate": 6.948888888888889e-06, | |
| "loss": 0.3939, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.772986167615948, | |
| "grad_norm": 6.072271823883057, | |
| "learning_rate": 6.893333333333334e-06, | |
| "loss": 0.4067, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.7831570382424735, | |
| "grad_norm": 6.9571123123168945, | |
| "learning_rate": 6.837777777777779e-06, | |
| "loss": 0.4107, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.7933279088689992, | |
| "grad_norm": 5.884681224822998, | |
| "learning_rate": 6.782222222222222e-06, | |
| "loss": 0.3995, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.8034987794955248, | |
| "grad_norm": 5.850441932678223, | |
| "learning_rate": 6.726666666666667e-06, | |
| "loss": 0.4017, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.8136696501220505, | |
| "grad_norm": 6.021271705627441, | |
| "learning_rate": 6.671111111111112e-06, | |
| "loss": 0.3708, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8136696501220505, | |
| "eval_loss": 0.42604002356529236, | |
| "eval_runtime": 1798.3084, | |
| "eval_samples_per_second": 2.169, | |
| "eval_steps_per_second": 0.136, | |
| "eval_wer": 0.32584758041147494, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.823840520748576, | |
| "grad_norm": 6.479699611663818, | |
| "learning_rate": 6.615555555555556e-06, | |
| "loss": 0.3914, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.8340113913751017, | |
| "grad_norm": 6.589993476867676, | |
| "learning_rate": 6.560000000000001e-06, | |
| "loss": 0.401, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.8441822620016274, | |
| "grad_norm": 5.026104927062988, | |
| "learning_rate": 6.504444444444446e-06, | |
| "loss": 0.38, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.8543531326281529, | |
| "grad_norm": 6.458845138549805, | |
| "learning_rate": 6.448888888888889e-06, | |
| "loss": 0.3885, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.8645240032546786, | |
| "grad_norm": 6.570550441741943, | |
| "learning_rate": 6.393333333333334e-06, | |
| "loss": 0.3992, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.8746948738812043, | |
| "grad_norm": 6.160183906555176, | |
| "learning_rate": 6.3377777777777786e-06, | |
| "loss": 0.385, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.8848657445077298, | |
| "grad_norm": 5.925890922546387, | |
| "learning_rate": 6.282222222222223e-06, | |
| "loss": 0.4082, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.8950366151342555, | |
| "grad_norm": 6.508133888244629, | |
| "learning_rate": 6.2266666666666675e-06, | |
| "loss": 0.4043, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.9052074857607811, | |
| "grad_norm": 6.300657272338867, | |
| "learning_rate": 6.171111111111112e-06, | |
| "loss": 0.3729, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.9153783563873068, | |
| "grad_norm": 5.284305572509766, | |
| "learning_rate": 6.1155555555555555e-06, | |
| "loss": 0.4046, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.9255492270138324, | |
| "grad_norm": 5.809286594390869, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 0.3864, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.935720097640358, | |
| "grad_norm": 6.667562007904053, | |
| "learning_rate": 6.004444444444445e-06, | |
| "loss": 0.3929, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.9458909682668837, | |
| "grad_norm": 6.4545392990112305, | |
| "learning_rate": 5.948888888888889e-06, | |
| "loss": 0.3975, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.9560618388934092, | |
| "grad_norm": 5.876199245452881, | |
| "learning_rate": 5.893333333333334e-06, | |
| "loss": 0.3769, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.9662327095199349, | |
| "grad_norm": 5.008429527282715, | |
| "learning_rate": 5.837777777777777e-06, | |
| "loss": 0.4031, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.9764035801464606, | |
| "grad_norm": 7.304080963134766, | |
| "learning_rate": 5.782222222222222e-06, | |
| "loss": 0.3804, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.9865744507729862, | |
| "grad_norm": 7.11293888092041, | |
| "learning_rate": 5.726666666666667e-06, | |
| "loss": 0.3849, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.9967453213995118, | |
| "grad_norm": 6.156993389129639, | |
| "learning_rate": 5.671111111111112e-06, | |
| "loss": 0.3804, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.0069161920260374, | |
| "grad_norm": 4.09175968170166, | |
| "learning_rate": 5.615555555555556e-06, | |
| "loss": 0.3138, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.017087062652563, | |
| "grad_norm": 5.104438304901123, | |
| "learning_rate": 5.560000000000001e-06, | |
| "loss": 0.2747, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0272579332790888, | |
| "grad_norm": 5.911495685577393, | |
| "learning_rate": 5.504444444444444e-06, | |
| "loss": 0.2825, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.0374288039056143, | |
| "grad_norm": 5.619592189788818, | |
| "learning_rate": 5.448888888888889e-06, | |
| "loss": 0.2991, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.0475996745321399, | |
| "grad_norm": 5.471043109893799, | |
| "learning_rate": 5.393333333333334e-06, | |
| "loss": 0.2766, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.0577705451586656, | |
| "grad_norm": 4.353448867797852, | |
| "learning_rate": 5.337777777777779e-06, | |
| "loss": 0.2883, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.0679414157851912, | |
| "grad_norm": 4.982877731323242, | |
| "learning_rate": 5.282222222222223e-06, | |
| "loss": 0.2888, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.0781122864117167, | |
| "grad_norm": 5.228238582611084, | |
| "learning_rate": 5.226666666666667e-06, | |
| "loss": 0.2802, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.0882831570382425, | |
| "grad_norm": 5.883535385131836, | |
| "learning_rate": 5.171111111111111e-06, | |
| "loss": 0.2779, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.098454027664768, | |
| "grad_norm": 4.599557876586914, | |
| "learning_rate": 5.115555555555556e-06, | |
| "loss": 0.2817, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.1086248982912936, | |
| "grad_norm": 5.081583023071289, | |
| "learning_rate": 5.060000000000001e-06, | |
| "loss": 0.2804, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.1187957689178194, | |
| "grad_norm": 5.387673854827881, | |
| "learning_rate": 5.004444444444445e-06, | |
| "loss": 0.2676, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.128966639544345, | |
| "grad_norm": 5.010275840759277, | |
| "learning_rate": 4.94888888888889e-06, | |
| "loss": 0.2752, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.1391375101708707, | |
| "grad_norm": 5.828437805175781, | |
| "learning_rate": 4.893333333333334e-06, | |
| "loss": 0.2795, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.1493083807973963, | |
| "grad_norm": 5.313021183013916, | |
| "learning_rate": 4.837777777777778e-06, | |
| "loss": 0.2876, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.1594792514239218, | |
| "grad_norm": 5.790019512176514, | |
| "learning_rate": 4.7822222222222226e-06, | |
| "loss": 0.283, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.1696501220504476, | |
| "grad_norm": 5.2214250564575195, | |
| "learning_rate": 4.7266666666666674e-06, | |
| "loss": 0.2762, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.1798209926769732, | |
| "grad_norm": 6.338456630706787, | |
| "learning_rate": 4.6711111111111115e-06, | |
| "loss": 0.2588, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.1899918633034987, | |
| "grad_norm": 6.858678340911865, | |
| "learning_rate": 4.6155555555555555e-06, | |
| "loss": 0.2724, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.2001627339300245, | |
| "grad_norm": 5.54526948928833, | |
| "learning_rate": 4.56e-06, | |
| "loss": 0.2893, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.21033360455655, | |
| "grad_norm": 5.165172100067139, | |
| "learning_rate": 4.504444444444444e-06, | |
| "loss": 0.2751, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.2205044751830756, | |
| "grad_norm": 6.359204292297363, | |
| "learning_rate": 4.448888888888889e-06, | |
| "loss": 0.2599, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2205044751830756, | |
| "eval_loss": 0.39731866121292114, | |
| "eval_runtime": 1796.0978, | |
| "eval_samples_per_second": 2.172, | |
| "eval_steps_per_second": 0.136, | |
| "eval_wer": 0.30034772529701537, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2306753458096014, | |
| "grad_norm": 4.958346366882324, | |
| "learning_rate": 4.393333333333334e-06, | |
| "loss": 0.2754, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.240846216436127, | |
| "grad_norm": 7.284126281738281, | |
| "learning_rate": 4.337777777777778e-06, | |
| "loss": 0.2764, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.2510170870626527, | |
| "grad_norm": 5.106224536895752, | |
| "learning_rate": 4.282222222222222e-06, | |
| "loss": 0.2679, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.2611879576891782, | |
| "grad_norm": 4.638324737548828, | |
| "learning_rate": 4.226666666666667e-06, | |
| "loss": 0.2678, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.2713588283157038, | |
| "grad_norm": 5.0809245109558105, | |
| "learning_rate": 4.171111111111111e-06, | |
| "loss": 0.2679, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.2815296989422293, | |
| "grad_norm": 5.177508354187012, | |
| "learning_rate": 4.115555555555556e-06, | |
| "loss": 0.2603, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.2917005695687551, | |
| "grad_norm": 4.664310932159424, | |
| "learning_rate": 4.060000000000001e-06, | |
| "loss": 0.2659, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.3018714401952807, | |
| "grad_norm": 5.230463981628418, | |
| "learning_rate": 4.004444444444445e-06, | |
| "loss": 0.271, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.3120423108218064, | |
| "grad_norm": 5.396975994110107, | |
| "learning_rate": 3.948888888888889e-06, | |
| "loss": 0.2653, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.322213181448332, | |
| "grad_norm": 4.997494220733643, | |
| "learning_rate": 3.893333333333333e-06, | |
| "loss": 0.2678, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.3323840520748575, | |
| "grad_norm": 6.467133045196533, | |
| "learning_rate": 3.837777777777778e-06, | |
| "loss": 0.2628, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.342554922701383, | |
| "grad_norm": 4.6464338302612305, | |
| "learning_rate": 3.782222222222223e-06, | |
| "loss": 0.2819, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.3527257933279089, | |
| "grad_norm": 5.477710723876953, | |
| "learning_rate": 3.726666666666667e-06, | |
| "loss": 0.2694, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.3628966639544344, | |
| "grad_norm": 5.745607376098633, | |
| "learning_rate": 3.6711111111111113e-06, | |
| "loss": 0.2731, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.3730675345809602, | |
| "grad_norm": 4.110312461853027, | |
| "learning_rate": 3.615555555555556e-06, | |
| "loss": 0.2726, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.3832384052074858, | |
| "grad_norm": 5.377216815948486, | |
| "learning_rate": 3.5600000000000002e-06, | |
| "loss": 0.2677, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.3934092758340113, | |
| "grad_norm": 5.4512763023376465, | |
| "learning_rate": 3.5044444444444447e-06, | |
| "loss": 0.2621, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.403580146460537, | |
| "grad_norm": 5.338203430175781, | |
| "learning_rate": 3.4488888888888896e-06, | |
| "loss": 0.2731, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.4137510170870626, | |
| "grad_norm": 5.551027774810791, | |
| "learning_rate": 3.3933333333333336e-06, | |
| "loss": 0.2821, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.4239218877135884, | |
| "grad_norm": 6.804419040679932, | |
| "learning_rate": 3.337777777777778e-06, | |
| "loss": 0.2652, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.434092758340114, | |
| "grad_norm": 5.15053653717041, | |
| "learning_rate": 3.282222222222223e-06, | |
| "loss": 0.2626, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.4442636289666395, | |
| "grad_norm": 4.987846851348877, | |
| "learning_rate": 3.226666666666667e-06, | |
| "loss": 0.2558, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.454434499593165, | |
| "grad_norm": 5.460192680358887, | |
| "learning_rate": 3.1711111111111114e-06, | |
| "loss": 0.2688, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.4646053702196908, | |
| "grad_norm": 5.20620059967041, | |
| "learning_rate": 3.1155555555555555e-06, | |
| "loss": 0.2541, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.4747762408462164, | |
| "grad_norm": 5.177892208099365, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 0.2546, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.4849471114727422, | |
| "grad_norm": 5.275447845458984, | |
| "learning_rate": 3.004444444444445e-06, | |
| "loss": 0.2487, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.4951179820992677, | |
| "grad_norm": 4.815367698669434, | |
| "learning_rate": 2.948888888888889e-06, | |
| "loss": 0.2505, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.5052888527257933, | |
| "grad_norm": 4.586023330688477, | |
| "learning_rate": 2.8933333333333337e-06, | |
| "loss": 0.2385, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.5154597233523188, | |
| "grad_norm": 5.920341491699219, | |
| "learning_rate": 2.837777777777778e-06, | |
| "loss": 0.2502, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.5256305939788446, | |
| "grad_norm": 5.697859764099121, | |
| "learning_rate": 2.7822222222222222e-06, | |
| "loss": 0.2574, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.5358014646053704, | |
| "grad_norm": 4.68679666519165, | |
| "learning_rate": 2.726666666666667e-06, | |
| "loss": 0.2543, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.545972335231896, | |
| "grad_norm": 5.064055442810059, | |
| "learning_rate": 2.6711111111111116e-06, | |
| "loss": 0.2639, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.5561432058584215, | |
| "grad_norm": 5.656905651092529, | |
| "learning_rate": 2.6155555555555556e-06, | |
| "loss": 0.2639, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.566314076484947, | |
| "grad_norm": 5.397951126098633, | |
| "learning_rate": 2.56e-06, | |
| "loss": 0.271, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.5764849471114726, | |
| "grad_norm": 6.378788948059082, | |
| "learning_rate": 2.504444444444445e-06, | |
| "loss": 0.2542, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.5866558177379984, | |
| "grad_norm": 6.6806960105896, | |
| "learning_rate": 2.448888888888889e-06, | |
| "loss": 0.2471, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.5968266883645241, | |
| "grad_norm": 4.837161064147949, | |
| "learning_rate": 2.3933333333333334e-06, | |
| "loss": 0.2497, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.6069975589910497, | |
| "grad_norm": 5.817267894744873, | |
| "learning_rate": 2.337777777777778e-06, | |
| "loss": 0.2583, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.6171684296175752, | |
| "grad_norm": 5.383573532104492, | |
| "learning_rate": 2.2822222222222223e-06, | |
| "loss": 0.2535, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.6273393002441008, | |
| "grad_norm": 4.556544780731201, | |
| "learning_rate": 2.226666666666667e-06, | |
| "loss": 0.2618, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.6273393002441008, | |
| "eval_loss": 0.37833017110824585, | |
| "eval_runtime": 1788.9834, | |
| "eval_samples_per_second": 2.181, | |
| "eval_steps_per_second": 0.136, | |
| "eval_wer": 0.2904713609581764, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.6375101708706266, | |
| "grad_norm": 5.119606018066406, | |
| "learning_rate": 2.1711111111111113e-06, | |
| "loss": 0.2613, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.647681041497152, | |
| "grad_norm": 5.435574054718018, | |
| "learning_rate": 2.1155555555555557e-06, | |
| "loss": 0.243, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.6578519121236779, | |
| "grad_norm": 5.2191643714904785, | |
| "learning_rate": 2.06e-06, | |
| "loss": 0.2563, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.6680227827502034, | |
| "grad_norm": 4.851512908935547, | |
| "learning_rate": 2.0044444444444446e-06, | |
| "loss": 0.2608, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.678193653376729, | |
| "grad_norm": 5.570041179656982, | |
| "learning_rate": 1.948888888888889e-06, | |
| "loss": 0.2572, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.6883645240032545, | |
| "grad_norm": 4.914021968841553, | |
| "learning_rate": 1.8933333333333333e-06, | |
| "loss": 0.2502, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.6985353946297803, | |
| "grad_norm": 4.91426944732666, | |
| "learning_rate": 1.837777777777778e-06, | |
| "loss": 0.2484, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.708706265256306, | |
| "grad_norm": 5.054934501647949, | |
| "learning_rate": 1.7822222222222225e-06, | |
| "loss": 0.2543, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.7188771358828316, | |
| "grad_norm": 5.295421600341797, | |
| "learning_rate": 1.7266666666666667e-06, | |
| "loss": 0.2479, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.7290480065093572, | |
| "grad_norm": 4.494225978851318, | |
| "learning_rate": 1.6711111111111112e-06, | |
| "loss": 0.2708, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.7392188771358827, | |
| "grad_norm": 5.637074947357178, | |
| "learning_rate": 1.6155555555555559e-06, | |
| "loss": 0.2536, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.7493897477624083, | |
| "grad_norm": 4.7007222175598145, | |
| "learning_rate": 1.56e-06, | |
| "loss": 0.2445, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.759560618388934, | |
| "grad_norm": 5.922358989715576, | |
| "learning_rate": 1.5044444444444446e-06, | |
| "loss": 0.2629, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.7697314890154598, | |
| "grad_norm": 5.818343162536621, | |
| "learning_rate": 1.4488888888888892e-06, | |
| "loss": 0.2462, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.7799023596419854, | |
| "grad_norm": 4.40110969543457, | |
| "learning_rate": 1.3933333333333335e-06, | |
| "loss": 0.2389, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.790073230268511, | |
| "grad_norm": 6.067351818084717, | |
| "learning_rate": 1.337777777777778e-06, | |
| "loss": 0.2662, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.8002441008950365, | |
| "grad_norm": 4.228072166442871, | |
| "learning_rate": 1.2822222222222222e-06, | |
| "loss": 0.2611, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.8104149715215623, | |
| "grad_norm": 5.135393142700195, | |
| "learning_rate": 1.2266666666666666e-06, | |
| "loss": 0.2607, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.8205858421480878, | |
| "grad_norm": 5.195592403411865, | |
| "learning_rate": 1.171111111111111e-06, | |
| "loss": 0.2534, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.8307567127746136, | |
| "grad_norm": 4.900667667388916, | |
| "learning_rate": 1.1155555555555558e-06, | |
| "loss": 0.2558, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.8409275834011392, | |
| "grad_norm": 4.837711811065674, | |
| "learning_rate": 1.06e-06, | |
| "loss": 0.2476, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.8510984540276647, | |
| "grad_norm": 5.072052955627441, | |
| "learning_rate": 1.0044444444444445e-06, | |
| "loss": 0.2333, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.8612693246541903, | |
| "grad_norm": 5.385496139526367, | |
| "learning_rate": 9.488888888888889e-07, | |
| "loss": 0.2453, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.871440195280716, | |
| "grad_norm": 5.36069393157959, | |
| "learning_rate": 8.933333333333334e-07, | |
| "loss": 0.2477, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.8816110659072418, | |
| "grad_norm": 4.781914710998535, | |
| "learning_rate": 8.37777777777778e-07, | |
| "loss": 0.2494, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.8917819365337674, | |
| "grad_norm": 5.366285800933838, | |
| "learning_rate": 7.822222222222223e-07, | |
| "loss": 0.2519, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.901952807160293, | |
| "grad_norm": 4.479983806610107, | |
| "learning_rate": 7.266666666666668e-07, | |
| "loss": 0.241, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.9121236777868185, | |
| "grad_norm": 5.053223133087158, | |
| "learning_rate": 6.711111111111111e-07, | |
| "loss": 0.2462, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.922294548413344, | |
| "grad_norm": 4.833332061767578, | |
| "learning_rate": 6.155555555555556e-07, | |
| "loss": 0.2698, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.9324654190398698, | |
| "grad_norm": 3.715808629989624, | |
| "learning_rate": 5.6e-07, | |
| "loss": 0.2338, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.9426362896663956, | |
| "grad_norm": 5.295703887939453, | |
| "learning_rate": 5.044444444444445e-07, | |
| "loss": 0.2508, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.9528071602929211, | |
| "grad_norm": 5.4455342292785645, | |
| "learning_rate": 4.488888888888889e-07, | |
| "loss": 0.2468, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.9629780309194467, | |
| "grad_norm": 4.946452617645264, | |
| "learning_rate": 3.9333333333333336e-07, | |
| "loss": 0.2459, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.9731489015459722, | |
| "grad_norm": 4.054887294769287, | |
| "learning_rate": 3.3777777777777777e-07, | |
| "loss": 0.2488, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.983319772172498, | |
| "grad_norm": 5.577613830566406, | |
| "learning_rate": 2.822222222222222e-07, | |
| "loss": 0.2435, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.9934906427990235, | |
| "grad_norm": 4.222945213317871, | |
| "learning_rate": 2.266666666666667e-07, | |
| "loss": 0.2399, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.0036615134255493, | |
| "grad_norm": 4.715478897094727, | |
| "learning_rate": 1.7111111111111114e-07, | |
| "loss": 0.2385, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.013832384052075, | |
| "grad_norm": 4.2399678230285645, | |
| "learning_rate": 1.1555555555555556e-07, | |
| "loss": 0.181, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.0240032546786004, | |
| "grad_norm": 4.25973653793335, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 0.1792, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.034174125305126, | |
| "grad_norm": 4.661466598510742, | |
| "learning_rate": 4.444444444444445e-09, | |
| "loss": 0.1846, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.034174125305126, | |
| "eval_loss": 0.36655664443969727, | |
| "eval_runtime": 1783.5562, | |
| "eval_samples_per_second": 2.187, | |
| "eval_steps_per_second": 0.137, | |
| "eval_wer": 0.27733507195981844, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.034174125305126, | |
| "step": 5000, | |
| "total_flos": 5.435045990970163e+20, | |
| "train_loss": 0.38478662071228026, | |
| "train_runtime": 58003.8059, | |
| "train_samples_per_second": 2.758, | |
| "train_steps_per_second": 0.086 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 5.435045990970163e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |