| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.537318712415989, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.017686593562079942, | |
| "grad_norm": 11.730201721191406, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 1.906, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.035373187124159884, | |
| "grad_norm": 9.43655014038086, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 1.5891, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05305978068623983, | |
| "grad_norm": 8.554606437683105, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.9582, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07074637424831977, | |
| "grad_norm": 7.075723171234131, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.8164, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08843296781039972, | |
| "grad_norm": 7.806982517242432, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.7283, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.10611956137247966, | |
| "grad_norm": 6.871402740478516, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.721, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1238061549345596, | |
| "grad_norm": 7.813017845153809, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.68, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.14149274849663954, | |
| "grad_norm": 7.54969596862793, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.7248, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1591793420587195, | |
| "grad_norm": 7.196774482727051, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.6417, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.17686593562079944, | |
| "grad_norm": 8.30762004852295, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.6594, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.19455252918287938, | |
| "grad_norm": 6.641123294830322, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.6359, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.21223912274495932, | |
| "grad_norm": 8.47127914428711, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.6455, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.22992571630703926, | |
| "grad_norm": 8.078540802001953, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.6563, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2476123098691192, | |
| "grad_norm": 7.481531620025635, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.6099, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.26529890343119916, | |
| "grad_norm": 8.171069145202637, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.5922, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2829854969932791, | |
| "grad_norm": 7.401958465576172, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.6232, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.30067209055535904, | |
| "grad_norm": 6.448988437652588, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.6328, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.318358684117439, | |
| "grad_norm": 6.956504821777344, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.6026, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.3360452776795189, | |
| "grad_norm": 8.774309158325195, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.6318, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.3537318712415989, | |
| "grad_norm": 6.220521450042725, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.6186, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3714184648036788, | |
| "grad_norm": 6.165685176849365, | |
| "learning_rate": 9.94888888888889e-06, | |
| "loss": 0.5509, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.38910505836575876, | |
| "grad_norm": 6.3863348960876465, | |
| "learning_rate": 9.893333333333334e-06, | |
| "loss": 0.63, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.40679165192783867, | |
| "grad_norm": 6.65989351272583, | |
| "learning_rate": 9.837777777777778e-06, | |
| "loss": 0.5981, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.42447824548991864, | |
| "grad_norm": 5.605571746826172, | |
| "learning_rate": 9.782222222222222e-06, | |
| "loss": 0.5611, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.4421648390519986, | |
| "grad_norm": 7.080026149749756, | |
| "learning_rate": 9.726666666666668e-06, | |
| "loss": 0.5498, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.4598514326140785, | |
| "grad_norm": 5.857707977294922, | |
| "learning_rate": 9.671111111111112e-06, | |
| "loss": 0.5942, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4775380261761585, | |
| "grad_norm": 6.002816200256348, | |
| "learning_rate": 9.615555555555558e-06, | |
| "loss": 0.5488, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.4952246197382384, | |
| "grad_norm": 6.393026828765869, | |
| "learning_rate": 9.56e-06, | |
| "loss": 0.5464, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5129112133003184, | |
| "grad_norm": 6.212553024291992, | |
| "learning_rate": 9.504444444444446e-06, | |
| "loss": 0.5416, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.5305978068623983, | |
| "grad_norm": 6.353979110717773, | |
| "learning_rate": 9.44888888888889e-06, | |
| "loss": 0.5493, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5482844004244782, | |
| "grad_norm": 5.198184013366699, | |
| "learning_rate": 9.393333333333334e-06, | |
| "loss": 0.562, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5659709939865581, | |
| "grad_norm": 5.739233493804932, | |
| "learning_rate": 9.33777777777778e-06, | |
| "loss": 0.5245, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5836575875486382, | |
| "grad_norm": 5.637094974517822, | |
| "learning_rate": 9.282222222222222e-06, | |
| "loss": 0.5056, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.6013441811107181, | |
| "grad_norm": 5.2869181632995605, | |
| "learning_rate": 9.226666666666668e-06, | |
| "loss": 0.495, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.619030774672798, | |
| "grad_norm": 6.895484447479248, | |
| "learning_rate": 9.171111111111112e-06, | |
| "loss": 0.5311, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.636717368234878, | |
| "grad_norm": 5.642014026641846, | |
| "learning_rate": 9.115555555555556e-06, | |
| "loss": 0.4611, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6544039617969579, | |
| "grad_norm": 6.39215087890625, | |
| "learning_rate": 9.060000000000001e-06, | |
| "loss": 0.5128, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.6720905553590378, | |
| "grad_norm": 4.879695892333984, | |
| "learning_rate": 9.004444444444445e-06, | |
| "loss": 0.502, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6897771489211177, | |
| "grad_norm": 6.497096538543701, | |
| "learning_rate": 8.94888888888889e-06, | |
| "loss": 0.4927, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.7074637424831978, | |
| "grad_norm": 6.05295991897583, | |
| "learning_rate": 8.893333333333333e-06, | |
| "loss": 0.4825, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7074637424831978, | |
| "eval_loss": 0.2707957625389099, | |
| "eval_runtime": 4792.033, | |
| "eval_samples_per_second": 2.246, | |
| "eval_steps_per_second": 0.14, | |
| "eval_wer": 0.18103144801689744, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7251503360452777, | |
| "grad_norm": 6.068356990814209, | |
| "learning_rate": 8.83777777777778e-06, | |
| "loss": 0.4787, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.7428369296073576, | |
| "grad_norm": 6.832317352294922, | |
| "learning_rate": 8.782222222222223e-06, | |
| "loss": 0.4583, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7605235231694376, | |
| "grad_norm": 5.528912544250488, | |
| "learning_rate": 8.726666666666667e-06, | |
| "loss": 0.4733, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.7782101167315175, | |
| "grad_norm": 6.220973968505859, | |
| "learning_rate": 8.671111111111113e-06, | |
| "loss": 0.4994, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7958967102935974, | |
| "grad_norm": 6.543335914611816, | |
| "learning_rate": 8.615555555555555e-06, | |
| "loss": 0.4694, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.8135833038556773, | |
| "grad_norm": 4.369144439697266, | |
| "learning_rate": 8.560000000000001e-06, | |
| "loss": 0.4525, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8312698974177574, | |
| "grad_norm": 4.928637981414795, | |
| "learning_rate": 8.504444444444445e-06, | |
| "loss": 0.4802, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.8489564909798373, | |
| "grad_norm": 5.016990661621094, | |
| "learning_rate": 8.448888888888889e-06, | |
| "loss": 0.462, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8666430845419172, | |
| "grad_norm": 5.185001373291016, | |
| "learning_rate": 8.393333333333335e-06, | |
| "loss": 0.4189, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.8843296781039972, | |
| "grad_norm": 5.9500627517700195, | |
| "learning_rate": 8.337777777777777e-06, | |
| "loss": 0.4862, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.9020162716660771, | |
| "grad_norm": 4.7840166091918945, | |
| "learning_rate": 8.282222222222223e-06, | |
| "loss": 0.4201, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.919702865228157, | |
| "grad_norm": 5.590398788452148, | |
| "learning_rate": 8.226666666666667e-06, | |
| "loss": 0.438, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.937389458790237, | |
| "grad_norm": 5.356804370880127, | |
| "learning_rate": 8.171111111111113e-06, | |
| "loss": 0.4343, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.955076052352317, | |
| "grad_norm": 5.994735240936279, | |
| "learning_rate": 8.115555555555557e-06, | |
| "loss": 0.4281, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.9727626459143969, | |
| "grad_norm": 4.92245626449585, | |
| "learning_rate": 8.06e-06, | |
| "loss": 0.4711, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.9904492394764768, | |
| "grad_norm": 6.673391342163086, | |
| "learning_rate": 8.004444444444445e-06, | |
| "loss": 0.4396, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.0081358330385568, | |
| "grad_norm": 4.157800197601318, | |
| "learning_rate": 7.948888888888889e-06, | |
| "loss": 0.3655, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.0258224266006368, | |
| "grad_norm": 4.4156060218811035, | |
| "learning_rate": 7.893333333333335e-06, | |
| "loss": 0.3032, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.0435090201627166, | |
| "grad_norm": 3.762485980987549, | |
| "learning_rate": 7.837777777777779e-06, | |
| "loss": 0.2597, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.0611956137247966, | |
| "grad_norm": 3.0248515605926514, | |
| "learning_rate": 7.782222222222223e-06, | |
| "loss": 0.2829, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.0788822072868764, | |
| "grad_norm": 5.492848873138428, | |
| "learning_rate": 7.726666666666667e-06, | |
| "loss": 0.229, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.0965688008489565, | |
| "grad_norm": 5.225216388702393, | |
| "learning_rate": 7.67111111111111e-06, | |
| "loss": 0.2677, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.1142553944110365, | |
| "grad_norm": 3.8239035606384277, | |
| "learning_rate": 7.6155555555555564e-06, | |
| "loss": 0.2517, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.1319419879731163, | |
| "grad_norm": 5.077475070953369, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 0.2489, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.1496285815351963, | |
| "grad_norm": 4.364721298217773, | |
| "learning_rate": 7.504444444444445e-06, | |
| "loss": 0.2598, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.1673151750972763, | |
| "grad_norm": 3.3700525760650635, | |
| "learning_rate": 7.44888888888889e-06, | |
| "loss": 0.2472, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.1850017686593561, | |
| "grad_norm": 4.548206329345703, | |
| "learning_rate": 7.393333333333333e-06, | |
| "loss": 0.2635, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.2026883622214362, | |
| "grad_norm": 4.475757122039795, | |
| "learning_rate": 7.337777777777778e-06, | |
| "loss": 0.2345, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.2203749557835162, | |
| "grad_norm": 4.205091953277588, | |
| "learning_rate": 7.282222222222222e-06, | |
| "loss": 0.2464, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.238061549345596, | |
| "grad_norm": 4.778344631195068, | |
| "learning_rate": 7.226666666666667e-06, | |
| "loss": 0.238, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.255748142907676, | |
| "grad_norm": 3.172917604446411, | |
| "learning_rate": 7.171111111111112e-06, | |
| "loss": 0.2364, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.2734347364697558, | |
| "grad_norm": 4.275852203369141, | |
| "learning_rate": 7.115555555555557e-06, | |
| "loss": 0.2573, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.2911213300318358, | |
| "grad_norm": 4.308451175689697, | |
| "learning_rate": 7.06e-06, | |
| "loss": 0.2569, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.3088079235939158, | |
| "grad_norm": 4.079460144042969, | |
| "learning_rate": 7.004444444444445e-06, | |
| "loss": 0.2261, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.3264945171559956, | |
| "grad_norm": 6.641946792602539, | |
| "learning_rate": 6.948888888888889e-06, | |
| "loss": 0.2334, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.3441811107180757, | |
| "grad_norm": 4.23126745223999, | |
| "learning_rate": 6.893333333333334e-06, | |
| "loss": 0.2307, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.3618677042801557, | |
| "grad_norm": 3.1530086994171143, | |
| "learning_rate": 6.837777777777779e-06, | |
| "loss": 0.2154, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.3795542978422355, | |
| "grad_norm": 3.8190791606903076, | |
| "learning_rate": 6.782222222222222e-06, | |
| "loss": 0.2274, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.3972408914043155, | |
| "grad_norm": 3.953749179840088, | |
| "learning_rate": 6.726666666666667e-06, | |
| "loss": 0.2426, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.4149274849663955, | |
| "grad_norm": 3.527085304260254, | |
| "learning_rate": 6.671111111111112e-06, | |
| "loss": 0.2262, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.4149274849663955, | |
| "eval_loss": 0.2485629916191101, | |
| "eval_runtime": 4723.0929, | |
| "eval_samples_per_second": 2.279, | |
| "eval_steps_per_second": 0.142, | |
| "eval_wer": 0.15938160056324807, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.4326140785284753, | |
| "grad_norm": 4.03377628326416, | |
| "learning_rate": 6.615555555555556e-06, | |
| "loss": 0.2367, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.4503006720905554, | |
| "grad_norm": 3.128779888153076, | |
| "learning_rate": 6.560000000000001e-06, | |
| "loss": 0.2372, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.4679872656526354, | |
| "grad_norm": 3.7657673358917236, | |
| "learning_rate": 6.504444444444446e-06, | |
| "loss": 0.2394, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.4856738592147152, | |
| "grad_norm": 3.009976387023926, | |
| "learning_rate": 6.448888888888889e-06, | |
| "loss": 0.2285, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.5033604527767952, | |
| "grad_norm": 3.916525363922119, | |
| "learning_rate": 6.393333333333334e-06, | |
| "loss": 0.2359, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.5210470463388752, | |
| "grad_norm": 4.418327808380127, | |
| "learning_rate": 6.3377777777777786e-06, | |
| "loss": 0.2499, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.538733639900955, | |
| "grad_norm": 4.2388105392456055, | |
| "learning_rate": 6.282222222222223e-06, | |
| "loss": 0.226, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.556420233463035, | |
| "grad_norm": 3.664215564727783, | |
| "learning_rate": 6.2266666666666675e-06, | |
| "loss": 0.2316, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.574106827025115, | |
| "grad_norm": 4.053125858306885, | |
| "learning_rate": 6.171111111111112e-06, | |
| "loss": 0.228, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.5917934205871949, | |
| "grad_norm": 4.452697277069092, | |
| "learning_rate": 6.1155555555555555e-06, | |
| "loss": 0.2196, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.6094800141492749, | |
| "grad_norm": 3.919962167739868, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 0.2495, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.627166607711355, | |
| "grad_norm": 4.33357572555542, | |
| "learning_rate": 6.004444444444445e-06, | |
| "loss": 0.2245, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.6448532012734347, | |
| "grad_norm": 4.634097099304199, | |
| "learning_rate": 5.948888888888889e-06, | |
| "loss": 0.2438, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.6625397948355147, | |
| "grad_norm": 4.564420223236084, | |
| "learning_rate": 5.893333333333334e-06, | |
| "loss": 0.2047, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.6802263883975948, | |
| "grad_norm": 4.051385879516602, | |
| "learning_rate": 5.837777777777777e-06, | |
| "loss": 0.2392, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.6979129819596745, | |
| "grad_norm": 3.313216209411621, | |
| "learning_rate": 5.782222222222222e-06, | |
| "loss": 0.2286, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.7155995755217543, | |
| "grad_norm": 3.334672451019287, | |
| "learning_rate": 5.726666666666667e-06, | |
| "loss": 0.2296, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.7332861690838346, | |
| "grad_norm": 3.519235134124756, | |
| "learning_rate": 5.671111111111112e-06, | |
| "loss": 0.2107, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.7509727626459144, | |
| "grad_norm": 4.1753339767456055, | |
| "learning_rate": 5.615555555555556e-06, | |
| "loss": 0.2464, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.7686593562079942, | |
| "grad_norm": 3.395427703857422, | |
| "learning_rate": 5.560000000000001e-06, | |
| "loss": 0.2323, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.7863459497700744, | |
| "grad_norm": 3.480686902999878, | |
| "learning_rate": 5.504444444444444e-06, | |
| "loss": 0.2205, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.8040325433321542, | |
| "grad_norm": 4.275322914123535, | |
| "learning_rate": 5.448888888888889e-06, | |
| "loss": 0.2147, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.821719136894234, | |
| "grad_norm": 3.6306569576263428, | |
| "learning_rate": 5.393333333333334e-06, | |
| "loss": 0.2404, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.839405730456314, | |
| "grad_norm": 3.426995038986206, | |
| "learning_rate": 5.337777777777779e-06, | |
| "loss": 0.2259, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.857092324018394, | |
| "grad_norm": 4.106579303741455, | |
| "learning_rate": 5.282222222222223e-06, | |
| "loss": 0.2043, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.8747789175804739, | |
| "grad_norm": 3.365471601486206, | |
| "learning_rate": 5.226666666666667e-06, | |
| "loss": 0.2179, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.892465511142554, | |
| "grad_norm": 3.4525575637817383, | |
| "learning_rate": 5.171111111111111e-06, | |
| "loss": 0.2185, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.910152104704634, | |
| "grad_norm": 3.5960958003997803, | |
| "learning_rate": 5.115555555555556e-06, | |
| "loss": 0.2007, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.9278386982667137, | |
| "grad_norm": 3.91766619682312, | |
| "learning_rate": 5.060000000000001e-06, | |
| "loss": 0.2284, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.9455252918287937, | |
| "grad_norm": 4.662227630615234, | |
| "learning_rate": 5.004444444444445e-06, | |
| "loss": 0.2238, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.9632118853908738, | |
| "grad_norm": 3.227084159851074, | |
| "learning_rate": 4.94888888888889e-06, | |
| "loss": 0.2068, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.9808984789529536, | |
| "grad_norm": 3.7078394889831543, | |
| "learning_rate": 4.893333333333334e-06, | |
| "loss": 0.225, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.9985850725150336, | |
| "grad_norm": 3.540781259536743, | |
| "learning_rate": 4.837777777777778e-06, | |
| "loss": 0.2115, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.0162716660771136, | |
| "grad_norm": 2.6315805912017822, | |
| "learning_rate": 4.7822222222222226e-06, | |
| "loss": 0.1164, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.0339582596391934, | |
| "grad_norm": 3.7077784538269043, | |
| "learning_rate": 4.7266666666666674e-06, | |
| "loss": 0.1052, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.0516448532012737, | |
| "grad_norm": 2.5548624992370605, | |
| "learning_rate": 4.6711111111111115e-06, | |
| "loss": 0.1033, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.0693314467633535, | |
| "grad_norm": 3.648667097091675, | |
| "learning_rate": 4.6155555555555555e-06, | |
| "loss": 0.0954, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.0870180403254333, | |
| "grad_norm": 1.9462636709213257, | |
| "learning_rate": 4.56e-06, | |
| "loss": 0.0907, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.104704633887513, | |
| "grad_norm": 2.353156089782715, | |
| "learning_rate": 4.504444444444444e-06, | |
| "loss": 0.0962, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.1223912274495933, | |
| "grad_norm": 1.8443711996078491, | |
| "learning_rate": 4.448888888888889e-06, | |
| "loss": 0.0867, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.1223912274495933, | |
| "eval_loss": 0.25060394406318665, | |
| "eval_runtime": 4656.2877, | |
| "eval_samples_per_second": 2.311, | |
| "eval_steps_per_second": 0.145, | |
| "eval_wer": 0.15110889462567473, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.140077821011673, | |
| "grad_norm": 2.377577781677246, | |
| "learning_rate": 4.393333333333334e-06, | |
| "loss": 0.0825, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.157764414573753, | |
| "grad_norm": 2.523731231689453, | |
| "learning_rate": 4.337777777777778e-06, | |
| "loss": 0.1005, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.175451008135833, | |
| "grad_norm": 3.5590755939483643, | |
| "learning_rate": 4.282222222222222e-06, | |
| "loss": 0.1085, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.193137601697913, | |
| "grad_norm": 2.7134900093078613, | |
| "learning_rate": 4.226666666666667e-06, | |
| "loss": 0.0957, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.2108241952599927, | |
| "grad_norm": 3.198664903640747, | |
| "learning_rate": 4.171111111111111e-06, | |
| "loss": 0.0898, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.228510788822073, | |
| "grad_norm": 2.207460403442383, | |
| "learning_rate": 4.115555555555556e-06, | |
| "loss": 0.0922, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.246197382384153, | |
| "grad_norm": 3.6293723583221436, | |
| "learning_rate": 4.060000000000001e-06, | |
| "loss": 0.0847, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.2638839759462326, | |
| "grad_norm": 2.8405864238739014, | |
| "learning_rate": 4.004444444444445e-06, | |
| "loss": 0.097, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.281570569508313, | |
| "grad_norm": 3.193572521209717, | |
| "learning_rate": 3.948888888888889e-06, | |
| "loss": 0.0997, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.2992571630703926, | |
| "grad_norm": 2.4953181743621826, | |
| "learning_rate": 3.893333333333333e-06, | |
| "loss": 0.0947, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.3169437566324724, | |
| "grad_norm": 2.287153959274292, | |
| "learning_rate": 3.837777777777778e-06, | |
| "loss": 0.0829, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.3346303501945527, | |
| "grad_norm": 3.0835964679718018, | |
| "learning_rate": 3.782222222222223e-06, | |
| "loss": 0.0986, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.3523169437566325, | |
| "grad_norm": 3.700576066970825, | |
| "learning_rate": 3.726666666666667e-06, | |
| "loss": 0.0938, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.3700035373187123, | |
| "grad_norm": 2.685814380645752, | |
| "learning_rate": 3.6711111111111113e-06, | |
| "loss": 0.0912, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.3876901308807925, | |
| "grad_norm": 2.80168080329895, | |
| "learning_rate": 3.615555555555556e-06, | |
| "loss": 0.1046, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.4053767244428723, | |
| "grad_norm": 1.993931531906128, | |
| "learning_rate": 3.5600000000000002e-06, | |
| "loss": 0.0956, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.423063318004952, | |
| "grad_norm": 2.2069132328033447, | |
| "learning_rate": 3.5044444444444447e-06, | |
| "loss": 0.0897, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.4407499115670324, | |
| "grad_norm": 1.9554574489593506, | |
| "learning_rate": 3.4488888888888896e-06, | |
| "loss": 0.0916, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.458436505129112, | |
| "grad_norm": 2.17236590385437, | |
| "learning_rate": 3.3933333333333336e-06, | |
| "loss": 0.0914, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.476123098691192, | |
| "grad_norm": 1.3875476121902466, | |
| "learning_rate": 3.337777777777778e-06, | |
| "loss": 0.0841, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.493809692253272, | |
| "grad_norm": 2.773275852203369, | |
| "learning_rate": 3.282222222222223e-06, | |
| "loss": 0.0776, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.511496285815352, | |
| "grad_norm": 2.6522185802459717, | |
| "learning_rate": 3.226666666666667e-06, | |
| "loss": 0.1017, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.529182879377432, | |
| "grad_norm": 2.5875155925750732, | |
| "learning_rate": 3.1711111111111114e-06, | |
| "loss": 0.0873, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.5468694729395116, | |
| "grad_norm": 2.877279281616211, | |
| "learning_rate": 3.1155555555555555e-06, | |
| "loss": 0.1071, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.564556066501592, | |
| "grad_norm": 2.6924633979797363, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 0.0885, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.5822426600636716, | |
| "grad_norm": 3.1496784687042236, | |
| "learning_rate": 3.004444444444445e-06, | |
| "loss": 0.0875, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.599929253625752, | |
| "grad_norm": 2.8723342418670654, | |
| "learning_rate": 2.948888888888889e-06, | |
| "loss": 0.0849, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.6176158471878317, | |
| "grad_norm": 2.9840595722198486, | |
| "learning_rate": 2.8933333333333337e-06, | |
| "loss": 0.085, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.6353024407499115, | |
| "grad_norm": 2.006192684173584, | |
| "learning_rate": 2.837777777777778e-06, | |
| "loss": 0.082, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.6529890343119913, | |
| "grad_norm": 2.020374298095703, | |
| "learning_rate": 2.7822222222222222e-06, | |
| "loss": 0.083, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.6706756278740715, | |
| "grad_norm": 2.8968160152435303, | |
| "learning_rate": 2.726666666666667e-06, | |
| "loss": 0.0896, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.6883622214361513, | |
| "grad_norm": 2.3445401191711426, | |
| "learning_rate": 2.6711111111111116e-06, | |
| "loss": 0.0956, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.7060488149982316, | |
| "grad_norm": 2.18906307220459, | |
| "learning_rate": 2.6155555555555556e-06, | |
| "loss": 0.0968, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.7237354085603114, | |
| "grad_norm": 3.4939792156219482, | |
| "learning_rate": 2.56e-06, | |
| "loss": 0.0807, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.741422002122391, | |
| "grad_norm": 2.769930839538574, | |
| "learning_rate": 2.504444444444445e-06, | |
| "loss": 0.0807, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.759108595684471, | |
| "grad_norm": 2.675159215927124, | |
| "learning_rate": 2.448888888888889e-06, | |
| "loss": 0.0852, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.776795189246551, | |
| "grad_norm": 1.9144662618637085, | |
| "learning_rate": 2.3933333333333334e-06, | |
| "loss": 0.0827, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.794481782808631, | |
| "grad_norm": 2.1294009685516357, | |
| "learning_rate": 2.337777777777778e-06, | |
| "loss": 0.0723, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.8121683763707113, | |
| "grad_norm": 2.5433287620544434, | |
| "learning_rate": 2.2822222222222223e-06, | |
| "loss": 0.0895, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.829854969932791, | |
| "grad_norm": 2.70267391204834, | |
| "learning_rate": 2.226666666666667e-06, | |
| "loss": 0.0973, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.829854969932791, | |
| "eval_loss": 0.24439197778701782, | |
| "eval_runtime": 4587.2812, | |
| "eval_samples_per_second": 2.346, | |
| "eval_steps_per_second": 0.147, | |
| "eval_wer": 0.14902605022295237, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.847541563494871, | |
| "grad_norm": 2.82698392868042, | |
| "learning_rate": 2.1711111111111113e-06, | |
| "loss": 0.087, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.8652281570569507, | |
| "grad_norm": 2.4205260276794434, | |
| "learning_rate": 2.1155555555555557e-06, | |
| "loss": 0.073, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.882914750619031, | |
| "grad_norm": 2.8398914337158203, | |
| "learning_rate": 2.06e-06, | |
| "loss": 0.0829, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.9006013441811107, | |
| "grad_norm": 2.544459581375122, | |
| "learning_rate": 2.0044444444444446e-06, | |
| "loss": 0.0944, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.9182879377431905, | |
| "grad_norm": 3.273761034011841, | |
| "learning_rate": 1.948888888888889e-06, | |
| "loss": 0.0894, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.9359745313052708, | |
| "grad_norm": 2.9628796577453613, | |
| "learning_rate": 1.8933333333333333e-06, | |
| "loss": 0.0779, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.9536611248673506, | |
| "grad_norm": 2.375798225402832, | |
| "learning_rate": 1.837777777777778e-06, | |
| "loss": 0.0887, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.9713477184294304, | |
| "grad_norm": 3.536142587661743, | |
| "learning_rate": 1.7822222222222225e-06, | |
| "loss": 0.0802, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.9890343119915106, | |
| "grad_norm": 2.959638833999634, | |
| "learning_rate": 1.7266666666666667e-06, | |
| "loss": 0.0736, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 3.0067209055535904, | |
| "grad_norm": 1.6278612613677979, | |
| "learning_rate": 1.6711111111111112e-06, | |
| "loss": 0.0726, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 3.02440749911567, | |
| "grad_norm": 2.070136308670044, | |
| "learning_rate": 1.6155555555555559e-06, | |
| "loss": 0.0342, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 3.0420940926777504, | |
| "grad_norm": 1.3914729356765747, | |
| "learning_rate": 1.56e-06, | |
| "loss": 0.0358, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.0597806862398302, | |
| "grad_norm": 2.264988422393799, | |
| "learning_rate": 1.5044444444444446e-06, | |
| "loss": 0.0333, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 3.07746727980191, | |
| "grad_norm": 2.115565538406372, | |
| "learning_rate": 1.4488888888888892e-06, | |
| "loss": 0.0441, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 3.0951538733639903, | |
| "grad_norm": 1.555413007736206, | |
| "learning_rate": 1.3933333333333335e-06, | |
| "loss": 0.0333, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 3.11284046692607, | |
| "grad_norm": 1.6975215673446655, | |
| "learning_rate": 1.337777777777778e-06, | |
| "loss": 0.0347, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.13052706048815, | |
| "grad_norm": 1.0206265449523926, | |
| "learning_rate": 1.2822222222222222e-06, | |
| "loss": 0.0288, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 3.14821365405023, | |
| "grad_norm": 0.8475884199142456, | |
| "learning_rate": 1.2266666666666666e-06, | |
| "loss": 0.0282, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 3.16590024761231, | |
| "grad_norm": 1.2225841283798218, | |
| "learning_rate": 1.171111111111111e-06, | |
| "loss": 0.0419, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 3.1835868411743897, | |
| "grad_norm": 1.5350502729415894, | |
| "learning_rate": 1.1155555555555558e-06, | |
| "loss": 0.033, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.2012734347364695, | |
| "grad_norm": 1.7746856212615967, | |
| "learning_rate": 1.06e-06, | |
| "loss": 0.0314, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 3.2189600282985498, | |
| "grad_norm": 0.9352214932441711, | |
| "learning_rate": 1.0044444444444445e-06, | |
| "loss": 0.0309, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.2366466218606296, | |
| "grad_norm": 1.1960421800613403, | |
| "learning_rate": 9.488888888888889e-07, | |
| "loss": 0.0432, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 3.25433321542271, | |
| "grad_norm": 2.057264566421509, | |
| "learning_rate": 8.933333333333334e-07, | |
| "loss": 0.0361, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.2720198089847896, | |
| "grad_norm": 1.3137741088867188, | |
| "learning_rate": 8.37777777777778e-07, | |
| "loss": 0.0274, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 3.2897064025468694, | |
| "grad_norm": 2.9971253871917725, | |
| "learning_rate": 7.822222222222223e-07, | |
| "loss": 0.0334, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 3.307392996108949, | |
| "grad_norm": 1.1959739923477173, | |
| "learning_rate": 7.266666666666668e-07, | |
| "loss": 0.0321, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 3.3250795896710295, | |
| "grad_norm": 1.4771987199783325, | |
| "learning_rate": 6.711111111111111e-07, | |
| "loss": 0.0304, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.3427661832331093, | |
| "grad_norm": 1.407965898513794, | |
| "learning_rate": 6.155555555555556e-07, | |
| "loss": 0.0358, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 3.360452776795189, | |
| "grad_norm": 1.603348970413208, | |
| "learning_rate": 5.6e-07, | |
| "loss": 0.0253, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.3781393703572693, | |
| "grad_norm": 0.6914874911308289, | |
| "learning_rate": 5.044444444444445e-07, | |
| "loss": 0.0314, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 3.395825963919349, | |
| "grad_norm": 0.9917079210281372, | |
| "learning_rate": 4.488888888888889e-07, | |
| "loss": 0.0366, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.413512557481429, | |
| "grad_norm": 1.6925675868988037, | |
| "learning_rate": 3.9333333333333336e-07, | |
| "loss": 0.0325, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 3.431199151043509, | |
| "grad_norm": 1.986444115638733, | |
| "learning_rate": 3.3777777777777777e-07, | |
| "loss": 0.0281, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.448885744605589, | |
| "grad_norm": 0.704007089138031, | |
| "learning_rate": 2.822222222222222e-07, | |
| "loss": 0.0313, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 3.4665723381676687, | |
| "grad_norm": 1.0404950380325317, | |
| "learning_rate": 2.266666666666667e-07, | |
| "loss": 0.0311, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.484258931729749, | |
| "grad_norm": 1.7548632621765137, | |
| "learning_rate": 1.7111111111111114e-07, | |
| "loss": 0.029, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 3.501945525291829, | |
| "grad_norm": 2.3233606815338135, | |
| "learning_rate": 1.1555555555555556e-07, | |
| "loss": 0.0364, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 3.5196321188539086, | |
| "grad_norm": 1.7240501642227173, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 0.035, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 3.537318712415989, | |
| "grad_norm": 0.9116590023040771, | |
| "learning_rate": 4.444444444444445e-09, | |
| "loss": 0.0303, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.537318712415989, | |
| "eval_loss": 0.27435705065727234, | |
| "eval_runtime": 4637.7857, | |
| "eval_samples_per_second": 2.321, | |
| "eval_steps_per_second": 0.145, | |
| "eval_wer": 0.14744191504341703, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.537318712415989, | |
| "step": 5000, | |
| "total_flos": 5.435079965953229e+20, | |
| "train_loss": 0.26693406739234926, | |
| "train_runtime": 74036.7422, | |
| "train_samples_per_second": 2.161, | |
| "train_steps_per_second": 0.068 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.435079965953229e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |