random_500_gold / epoch_7 /trainer_state.json
PranavJosh's picture
Upload epoch_7 checkpoint
8b1d751 verified
Raw
History Blame Contribute Delete
69.6 kB
{
"best_global_step": 5192,
"best_metric": 20.828885822637613,
"best_model_checkpoint": "./whisper_s_punjabi_ddp_4gpu/checkpoint-5192",
"epoch": 7.0,
"eval_steps": 500,
"global_step": 9086,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01926040061633282,
"grad_norm": 58.414207458496094,
"learning_rate": 8.000000000000001e-07,
"loss": 7.913150634765625,
"step": 25
},
{
"epoch": 0.03852080123266564,
"grad_norm": 18.150222778320312,
"learning_rate": 1.6333333333333335e-06,
"loss": 5.986227416992188,
"step": 50
},
{
"epoch": 0.05778120184899846,
"grad_norm": 14.175592422485352,
"learning_rate": 2.466666666666667e-06,
"loss": 4.43172119140625,
"step": 75
},
{
"epoch": 0.07704160246533127,
"grad_norm": 16.95926284790039,
"learning_rate": 3.3000000000000006e-06,
"loss": 3.4311566162109375,
"step": 100
},
{
"epoch": 0.0963020030816641,
"grad_norm": 9.69564151763916,
"learning_rate": 4.133333333333333e-06,
"loss": 2.0594117736816404,
"step": 125
},
{
"epoch": 0.11556240369799692,
"grad_norm": 10.076292991638184,
"learning_rate": 4.966666666666667e-06,
"loss": 1.8318702697753906,
"step": 150
},
{
"epoch": 0.13482280431432975,
"grad_norm": 9.035318374633789,
"learning_rate": 5.8e-06,
"loss": 1.7396725463867186,
"step": 175
},
{
"epoch": 0.15408320493066255,
"grad_norm": 10.635228157043457,
"learning_rate": 6.633333333333334e-06,
"loss": 1.6932928466796875,
"step": 200
},
{
"epoch": 0.17334360554699538,
"grad_norm": 10.217312812805176,
"learning_rate": 7.4666666666666675e-06,
"loss": 1.6321292114257813,
"step": 225
},
{
"epoch": 0.1926040061633282,
"grad_norm": 11.088924407958984,
"learning_rate": 8.3e-06,
"loss": 1.5770054626464844,
"step": 250
},
{
"epoch": 0.211864406779661,
"grad_norm": 10.08351993560791,
"learning_rate": 9.133333333333335e-06,
"loss": 1.5158828735351562,
"step": 275
},
{
"epoch": 0.23112480739599384,
"grad_norm": 9.66342544555664,
"learning_rate": 9.966666666666667e-06,
"loss": 1.5335569763183594,
"step": 300
},
{
"epoch": 0.25038520801232667,
"grad_norm": 10.034809112548828,
"learning_rate": 9.981072555205049e-06,
"loss": 1.4476705932617187,
"step": 325
},
{
"epoch": 0.2696456086286595,
"grad_norm": 7.876269817352295,
"learning_rate": 9.961356466876972e-06,
"loss": 1.4681803894042968,
"step": 350
},
{
"epoch": 0.2889060092449923,
"grad_norm": 10.250846862792969,
"learning_rate": 9.941640378548896e-06,
"loss": 1.437877197265625,
"step": 375
},
{
"epoch": 0.3081664098613251,
"grad_norm": 9.018561363220215,
"learning_rate": 9.921924290220821e-06,
"loss": 1.3937382507324219,
"step": 400
},
{
"epoch": 0.3274268104776579,
"grad_norm": 8.793310165405273,
"learning_rate": 9.902208201892746e-06,
"loss": 1.3915924072265624,
"step": 425
},
{
"epoch": 0.34668721109399075,
"grad_norm": 9.753164291381836,
"learning_rate": 9.88249211356467e-06,
"loss": 1.3350201416015626,
"step": 450
},
{
"epoch": 0.3659476117103236,
"grad_norm": 9.90793514251709,
"learning_rate": 9.862776025236594e-06,
"loss": 1.3452024841308594,
"step": 475
},
{
"epoch": 0.3852080123266564,
"grad_norm": 7.425978660583496,
"learning_rate": 9.843059936908518e-06,
"loss": 1.3636004638671875,
"step": 500
},
{
"epoch": 0.40446841294298924,
"grad_norm": 7.5032501220703125,
"learning_rate": 9.823343848580441e-06,
"loss": 1.337528076171875,
"step": 525
},
{
"epoch": 0.423728813559322,
"grad_norm": 8.89260482788086,
"learning_rate": 9.803627760252366e-06,
"loss": 1.3338491821289062,
"step": 550
},
{
"epoch": 0.44298921417565484,
"grad_norm": 8.321249008178711,
"learning_rate": 9.783911671924291e-06,
"loss": 1.333787384033203,
"step": 575
},
{
"epoch": 0.4622496147919877,
"grad_norm": 8.888858795166016,
"learning_rate": 9.764195583596216e-06,
"loss": 1.3222378540039061,
"step": 600
},
{
"epoch": 0.4815100154083205,
"grad_norm": 7.648203372955322,
"learning_rate": 9.74447949526814e-06,
"loss": 1.3167262268066406,
"step": 625
},
{
"epoch": 0.5007704160246533,
"grad_norm": 6.553903102874756,
"learning_rate": 9.724763406940063e-06,
"loss": 1.298903045654297,
"step": 650
},
{
"epoch": 0.5200308166409862,
"grad_norm": 7.545315265655518,
"learning_rate": 9.705047318611988e-06,
"loss": 1.266801223754883,
"step": 675
},
{
"epoch": 0.539291217257319,
"grad_norm": 7.492407321929932,
"learning_rate": 9.685331230283911e-06,
"loss": 1.2619186401367188,
"step": 700
},
{
"epoch": 0.5585516178736518,
"grad_norm": 6.489583969116211,
"learning_rate": 9.665615141955836e-06,
"loss": 1.2714739990234376,
"step": 725
},
{
"epoch": 0.5778120184899846,
"grad_norm": 8.90630054473877,
"learning_rate": 9.64589905362776e-06,
"loss": 1.2946762084960937,
"step": 750
},
{
"epoch": 0.5970724191063174,
"grad_norm": 7.635409832000732,
"learning_rate": 9.626182965299686e-06,
"loss": 1.269822463989258,
"step": 775
},
{
"epoch": 0.6163328197226502,
"grad_norm": 7.459207534790039,
"learning_rate": 9.60646687697161e-06,
"loss": 1.2825979614257812,
"step": 800
},
{
"epoch": 0.635593220338983,
"grad_norm": 8.159686088562012,
"learning_rate": 9.586750788643533e-06,
"loss": 1.2717716217041015,
"step": 825
},
{
"epoch": 0.6548536209553159,
"grad_norm": 6.630347728729248,
"learning_rate": 9.567034700315458e-06,
"loss": 1.252648162841797,
"step": 850
},
{
"epoch": 0.6741140215716487,
"grad_norm": 7.215192794799805,
"learning_rate": 9.547318611987383e-06,
"loss": 1.252330093383789,
"step": 875
},
{
"epoch": 0.6933744221879815,
"grad_norm": 8.178110122680664,
"learning_rate": 9.527602523659308e-06,
"loss": 1.2153298187255859,
"step": 900
},
{
"epoch": 0.7126348228043143,
"grad_norm": 6.375010013580322,
"learning_rate": 9.50788643533123e-06,
"loss": 1.2400372314453125,
"step": 925
},
{
"epoch": 0.7318952234206472,
"grad_norm": 8.464449882507324,
"learning_rate": 9.488170347003155e-06,
"loss": 1.2374497222900391,
"step": 950
},
{
"epoch": 0.75115562403698,
"grad_norm": 6.5914764404296875,
"learning_rate": 9.46845425867508e-06,
"loss": 1.2367049407958985,
"step": 975
},
{
"epoch": 0.7704160246533128,
"grad_norm": 6.147579193115234,
"learning_rate": 9.448738170347003e-06,
"loss": 1.2031460571289063,
"step": 1000
},
{
"epoch": 0.7896764252696457,
"grad_norm": 6.786445617675781,
"learning_rate": 9.429022082018928e-06,
"loss": 1.1893964385986329,
"step": 1025
},
{
"epoch": 0.8089368258859785,
"grad_norm": 5.8521504402160645,
"learning_rate": 9.409305993690853e-06,
"loss": 1.1855683898925782,
"step": 1050
},
{
"epoch": 0.8281972265023112,
"grad_norm": 5.927206993103027,
"learning_rate": 9.389589905362777e-06,
"loss": 1.2393891906738281,
"step": 1075
},
{
"epoch": 0.847457627118644,
"grad_norm": 7.765108108520508,
"learning_rate": 9.369873817034702e-06,
"loss": 1.2078982543945314,
"step": 1100
},
{
"epoch": 0.8667180277349769,
"grad_norm": 6.578537464141846,
"learning_rate": 9.350157728706625e-06,
"loss": 1.2135992431640625,
"step": 1125
},
{
"epoch": 0.8859784283513097,
"grad_norm": 6.826841831207275,
"learning_rate": 9.33044164037855e-06,
"loss": 1.2246592712402344,
"step": 1150
},
{
"epoch": 0.9052388289676425,
"grad_norm": 5.9232611656188965,
"learning_rate": 9.310725552050473e-06,
"loss": 1.2366727447509767,
"step": 1175
},
{
"epoch": 0.9244992295839753,
"grad_norm": 6.405348777770996,
"learning_rate": 9.291009463722398e-06,
"loss": 1.1822604370117187,
"step": 1200
},
{
"epoch": 0.9437596302003082,
"grad_norm": 7.846672534942627,
"learning_rate": 9.271293375394322e-06,
"loss": 1.1902696228027343,
"step": 1225
},
{
"epoch": 0.963020030816641,
"grad_norm": 6.556903839111328,
"learning_rate": 9.251577287066247e-06,
"loss": 1.2147691345214844,
"step": 1250
},
{
"epoch": 0.9822804314329738,
"grad_norm": 6.414576530456543,
"learning_rate": 9.231861198738172e-06,
"loss": 1.1895433044433594,
"step": 1275
},
{
"epoch": 1.0,
"eval_loss": 0.1692090779542923,
"eval_runtime": 1293.6351,
"eval_samples_per_second": 18.724,
"eval_steps_per_second": 0.084,
"eval_wer": 23.27414889831906,
"step": 1298
},
{
"epoch": 1.0015408320493067,
"grad_norm": 5.701938152313232,
"learning_rate": 9.212145110410095e-06,
"loss": 1.1426918029785156,
"step": 1300
},
{
"epoch": 1.0208012326656395,
"grad_norm": 7.372975826263428,
"learning_rate": 9.19242902208202e-06,
"loss": 1.0540249633789063,
"step": 1325
},
{
"epoch": 1.0400616332819723,
"grad_norm": 6.552262783050537,
"learning_rate": 9.172712933753945e-06,
"loss": 1.0437205505371094,
"step": 1350
},
{
"epoch": 1.0593220338983051,
"grad_norm": 7.7785444259643555,
"learning_rate": 9.152996845425868e-06,
"loss": 1.063336639404297,
"step": 1375
},
{
"epoch": 1.078582434514638,
"grad_norm": 6.104959011077881,
"learning_rate": 9.133280757097792e-06,
"loss": 1.0619095611572265,
"step": 1400
},
{
"epoch": 1.0978428351309708,
"grad_norm": 7.427105903625488,
"learning_rate": 9.113564668769717e-06,
"loss": 1.0477345275878907,
"step": 1425
},
{
"epoch": 1.1171032357473036,
"grad_norm": 4.97503137588501,
"learning_rate": 9.093848580441642e-06,
"loss": 1.0352927398681642,
"step": 1450
},
{
"epoch": 1.1363636363636362,
"grad_norm": 6.956381320953369,
"learning_rate": 9.074132492113565e-06,
"loss": 1.0552413940429688,
"step": 1475
},
{
"epoch": 1.1556240369799693,
"grad_norm": 6.349787712097168,
"learning_rate": 9.05441640378549e-06,
"loss": 1.0734415435791016,
"step": 1500
},
{
"epoch": 1.174884437596302,
"grad_norm": 6.110345363616943,
"learning_rate": 9.034700315457414e-06,
"loss": 1.0556398773193358,
"step": 1525
},
{
"epoch": 1.1941448382126347,
"grad_norm": 5.747076511383057,
"learning_rate": 9.014984227129339e-06,
"loss": 1.0406858825683594,
"step": 1550
},
{
"epoch": 1.2134052388289676,
"grad_norm": 5.9346113204956055,
"learning_rate": 8.995268138801264e-06,
"loss": 1.0497844696044922,
"step": 1575
},
{
"epoch": 1.2326656394453004,
"grad_norm": 5.690814971923828,
"learning_rate": 8.975552050473187e-06,
"loss": 1.0486802673339843,
"step": 1600
},
{
"epoch": 1.2519260400616332,
"grad_norm": 5.829971790313721,
"learning_rate": 8.95583596214511e-06,
"loss": 1.0496593475341798,
"step": 1625
},
{
"epoch": 1.271186440677966,
"grad_norm": 5.7554144859313965,
"learning_rate": 8.936119873817035e-06,
"loss": 1.031878204345703,
"step": 1650
},
{
"epoch": 1.2904468412942989,
"grad_norm": 5.836212158203125,
"learning_rate": 8.91640378548896e-06,
"loss": 1.0397360229492187,
"step": 1675
},
{
"epoch": 1.3097072419106317,
"grad_norm": 5.595889091491699,
"learning_rate": 8.896687697160884e-06,
"loss": 1.0481429290771485,
"step": 1700
},
{
"epoch": 1.3289676425269645,
"grad_norm": 6.345583915710449,
"learning_rate": 8.876971608832809e-06,
"loss": 1.0513285064697266,
"step": 1725
},
{
"epoch": 1.3482280431432974,
"grad_norm": 6.793821334838867,
"learning_rate": 8.857255520504732e-06,
"loss": 1.0579839324951172,
"step": 1750
},
{
"epoch": 1.3674884437596302,
"grad_norm": 6.589428901672363,
"learning_rate": 8.837539432176657e-06,
"loss": 1.0568167114257812,
"step": 1775
},
{
"epoch": 1.386748844375963,
"grad_norm": 6.022590637207031,
"learning_rate": 8.817823343848581e-06,
"loss": 1.0284172821044921,
"step": 1800
},
{
"epoch": 1.4060092449922958,
"grad_norm": 6.013957500457764,
"learning_rate": 8.798107255520505e-06,
"loss": 1.0445652770996094,
"step": 1825
},
{
"epoch": 1.4252696456086287,
"grad_norm": 6.47821044921875,
"learning_rate": 8.77839116719243e-06,
"loss": 1.0797194671630859,
"step": 1850
},
{
"epoch": 1.4445300462249615,
"grad_norm": 5.593505382537842,
"learning_rate": 8.758675078864354e-06,
"loss": 1.052033920288086,
"step": 1875
},
{
"epoch": 1.4637904468412943,
"grad_norm": 6.251946926116943,
"learning_rate": 8.738958990536279e-06,
"loss": 1.070428695678711,
"step": 1900
},
{
"epoch": 1.4830508474576272,
"grad_norm": 6.3281755447387695,
"learning_rate": 8.719242902208202e-06,
"loss": 1.0281287384033204,
"step": 1925
},
{
"epoch": 1.50231124807396,
"grad_norm": 5.988689422607422,
"learning_rate": 8.699526813880127e-06,
"loss": 1.0324861907958984,
"step": 1950
},
{
"epoch": 1.5215716486902928,
"grad_norm": 6.062340259552002,
"learning_rate": 8.679810725552051e-06,
"loss": 1.0550352478027343,
"step": 1975
},
{
"epoch": 1.5408320493066254,
"grad_norm": 4.881484031677246,
"learning_rate": 8.660094637223976e-06,
"loss": 1.0413783264160157,
"step": 2000
},
{
"epoch": 1.5600924499229585,
"grad_norm": 5.4565324783325195,
"learning_rate": 8.6403785488959e-06,
"loss": 1.0413751983642578,
"step": 2025
},
{
"epoch": 1.579352850539291,
"grad_norm": 5.8962483406066895,
"learning_rate": 8.620662460567824e-06,
"loss": 1.0537427520751954,
"step": 2050
},
{
"epoch": 1.5986132511556241,
"grad_norm": 5.976631164550781,
"learning_rate": 8.600946372239749e-06,
"loss": 1.0484622955322265,
"step": 2075
},
{
"epoch": 1.6178736517719567,
"grad_norm": 5.480655670166016,
"learning_rate": 8.581230283911672e-06,
"loss": 1.009260482788086,
"step": 2100
},
{
"epoch": 1.6371340523882898,
"grad_norm": 5.977313041687012,
"learning_rate": 8.561514195583596e-06,
"loss": 1.0521608734130858,
"step": 2125
},
{
"epoch": 1.6563944530046224,
"grad_norm": 5.078767776489258,
"learning_rate": 8.541798107255521e-06,
"loss": 1.0576594543457032,
"step": 2150
},
{
"epoch": 1.6756548536209555,
"grad_norm": 5.783039093017578,
"learning_rate": 8.522082018927446e-06,
"loss": 1.029164810180664,
"step": 2175
},
{
"epoch": 1.694915254237288,
"grad_norm": 6.353992938995361,
"learning_rate": 8.50236593059937e-06,
"loss": 1.0603709411621094,
"step": 2200
},
{
"epoch": 1.7141756548536211,
"grad_norm": 5.704638957977295,
"learning_rate": 8.482649842271294e-06,
"loss": 1.0352437591552734,
"step": 2225
},
{
"epoch": 1.7334360554699537,
"grad_norm": 6.092396259307861,
"learning_rate": 8.462933753943218e-06,
"loss": 1.0399227142333984,
"step": 2250
},
{
"epoch": 1.7526964560862865,
"grad_norm": 7.1011223793029785,
"learning_rate": 8.443217665615142e-06,
"loss": 1.0415158081054687,
"step": 2275
},
{
"epoch": 1.7719568567026194,
"grad_norm": 4.735724449157715,
"learning_rate": 8.423501577287066e-06,
"loss": 1.0339994049072265,
"step": 2300
},
{
"epoch": 1.7912172573189522,
"grad_norm": 6.994238376617432,
"learning_rate": 8.403785488958991e-06,
"loss": 1.0387593841552734,
"step": 2325
},
{
"epoch": 1.810477657935285,
"grad_norm": 6.1796956062316895,
"learning_rate": 8.384069400630916e-06,
"loss": 1.0579666900634765,
"step": 2350
},
{
"epoch": 1.8297380585516179,
"grad_norm": 5.0321197509765625,
"learning_rate": 8.36435331230284e-06,
"loss": 1.0294721984863282,
"step": 2375
},
{
"epoch": 1.8489984591679507,
"grad_norm": 6.216546535491943,
"learning_rate": 8.344637223974764e-06,
"loss": 1.0312883758544922,
"step": 2400
},
{
"epoch": 1.8682588597842835,
"grad_norm": 5.854124546051025,
"learning_rate": 8.324921135646688e-06,
"loss": 1.0346050262451172,
"step": 2425
},
{
"epoch": 1.8875192604006163,
"grad_norm": 7.55112361907959,
"learning_rate": 8.305205047318613e-06,
"loss": 1.0578290557861327,
"step": 2450
},
{
"epoch": 1.9067796610169492,
"grad_norm": 5.812201499938965,
"learning_rate": 8.285488958990538e-06,
"loss": 1.0345248413085937,
"step": 2475
},
{
"epoch": 1.926040061633282,
"grad_norm": 7.083204746246338,
"learning_rate": 8.26577287066246e-06,
"loss": 1.0337892913818358,
"step": 2500
},
{
"epoch": 1.9453004622496148,
"grad_norm": 5.301337242126465,
"learning_rate": 8.246056782334386e-06,
"loss": 1.0415627288818359,
"step": 2525
},
{
"epoch": 1.9645608628659477,
"grad_norm": 5.613387107849121,
"learning_rate": 8.22634069400631e-06,
"loss": 1.040993194580078,
"step": 2550
},
{
"epoch": 1.9838212634822803,
"grad_norm": 5.031157493591309,
"learning_rate": 8.206624605678233e-06,
"loss": 1.0369261932373046,
"step": 2575
},
{
"epoch": 2.0,
"eval_loss": 0.16153943538665771,
"eval_runtime": 1177.5263,
"eval_samples_per_second": 20.57,
"eval_steps_per_second": 0.093,
"eval_wer": 21.05581734399,
"step": 2596
},
{
"epoch": 2.0030816640986133,
"grad_norm": 5.883306503295898,
"learning_rate": 8.186908517350158e-06,
"loss": 0.9993476867675781,
"step": 2600
},
{
"epoch": 2.022342064714946,
"grad_norm": 5.52811336517334,
"learning_rate": 8.167192429022083e-06,
"loss": 0.8988736724853515,
"step": 2625
},
{
"epoch": 2.041602465331279,
"grad_norm": 5.482069969177246,
"learning_rate": 8.147476340694008e-06,
"loss": 0.8860294342041015,
"step": 2650
},
{
"epoch": 2.0608628659476116,
"grad_norm": 6.871766090393066,
"learning_rate": 8.127760252365932e-06,
"loss": 0.8951603698730469,
"step": 2675
},
{
"epoch": 2.0801232665639446,
"grad_norm": 5.174409866333008,
"learning_rate": 8.108044164037855e-06,
"loss": 0.8793936920166016,
"step": 2700
},
{
"epoch": 2.0993836671802772,
"grad_norm": 7.379999160766602,
"learning_rate": 8.088328075709778e-06,
"loss": 0.8864609527587891,
"step": 2725
},
{
"epoch": 2.1186440677966103,
"grad_norm": 6.127875328063965,
"learning_rate": 8.068611987381703e-06,
"loss": 0.8880971527099609,
"step": 2750
},
{
"epoch": 2.137904468412943,
"grad_norm": 5.28157901763916,
"learning_rate": 8.048895899053628e-06,
"loss": 0.8864345550537109,
"step": 2775
},
{
"epoch": 2.157164869029276,
"grad_norm": 5.634916305541992,
"learning_rate": 8.029179810725553e-06,
"loss": 0.8789139556884765,
"step": 2800
},
{
"epoch": 2.1764252696456086,
"grad_norm": 6.002861022949219,
"learning_rate": 8.009463722397477e-06,
"loss": 0.8731598663330078,
"step": 2825
},
{
"epoch": 2.1956856702619416,
"grad_norm": 4.893068313598633,
"learning_rate": 7.9897476340694e-06,
"loss": 0.8869886779785157,
"step": 2850
},
{
"epoch": 2.214946070878274,
"grad_norm": 6.026525974273682,
"learning_rate": 7.970031545741325e-06,
"loss": 0.8979975128173828,
"step": 2875
},
{
"epoch": 2.2342064714946073,
"grad_norm": 6.521470546722412,
"learning_rate": 7.95031545741325e-06,
"loss": 0.8889907073974609,
"step": 2900
},
{
"epoch": 2.25346687211094,
"grad_norm": 5.973586082458496,
"learning_rate": 7.930599369085175e-06,
"loss": 0.9008167266845704,
"step": 2925
},
{
"epoch": 2.2727272727272725,
"grad_norm": 4.965330123901367,
"learning_rate": 7.9108832807571e-06,
"loss": 0.9247480773925781,
"step": 2950
},
{
"epoch": 2.2919876733436055,
"grad_norm": 6.2271223068237305,
"learning_rate": 7.891167192429023e-06,
"loss": 0.8841385650634765,
"step": 2975
},
{
"epoch": 2.3112480739599386,
"grad_norm": 5.103269100189209,
"learning_rate": 7.871451104100947e-06,
"loss": 0.9012676239013672,
"step": 3000
},
{
"epoch": 2.330508474576271,
"grad_norm": 6.160482883453369,
"learning_rate": 7.85173501577287e-06,
"loss": 0.8942781066894532,
"step": 3025
},
{
"epoch": 2.349768875192604,
"grad_norm": 5.235211372375488,
"learning_rate": 7.832018927444795e-06,
"loss": 0.8858016967773438,
"step": 3050
},
{
"epoch": 2.369029275808937,
"grad_norm": 5.827929496765137,
"learning_rate": 7.81230283911672e-06,
"loss": 0.9006576538085938,
"step": 3075
},
{
"epoch": 2.3882896764252695,
"grad_norm": 5.158037185668945,
"learning_rate": 7.792586750788645e-06,
"loss": 0.8841426086425781,
"step": 3100
},
{
"epoch": 2.4075500770416025,
"grad_norm": 5.732335090637207,
"learning_rate": 7.77287066246057e-06,
"loss": 0.8895353698730468,
"step": 3125
},
{
"epoch": 2.426810477657935,
"grad_norm": 5.733401298522949,
"learning_rate": 7.753154574132492e-06,
"loss": 0.8874980926513671,
"step": 3150
},
{
"epoch": 2.446070878274268,
"grad_norm": 5.1814985275268555,
"learning_rate": 7.733438485804417e-06,
"loss": 0.8906858825683593,
"step": 3175
},
{
"epoch": 2.4653312788906008,
"grad_norm": 5.38557243347168,
"learning_rate": 7.71372239747634e-06,
"loss": 0.9022510528564454,
"step": 3200
},
{
"epoch": 2.484591679506934,
"grad_norm": 4.910186290740967,
"learning_rate": 7.694006309148265e-06,
"loss": 0.8867702484130859,
"step": 3225
},
{
"epoch": 2.5038520801232664,
"grad_norm": 4.856821537017822,
"learning_rate": 7.67429022082019e-06,
"loss": 0.8708806610107422,
"step": 3250
},
{
"epoch": 2.5231124807395995,
"grad_norm": 5.405826091766357,
"learning_rate": 7.654574132492114e-06,
"loss": 0.8903508758544922,
"step": 3275
},
{
"epoch": 2.542372881355932,
"grad_norm": 5.789135932922363,
"learning_rate": 7.63485804416404e-06,
"loss": 0.9085517883300781,
"step": 3300
},
{
"epoch": 2.561633281972265,
"grad_norm": 5.805831432342529,
"learning_rate": 7.615141955835963e-06,
"loss": 0.8931381988525391,
"step": 3325
},
{
"epoch": 2.5808936825885977,
"grad_norm": 5.687262535095215,
"learning_rate": 7.595425867507887e-06,
"loss": 0.881778335571289,
"step": 3350
},
{
"epoch": 2.600154083204931,
"grad_norm": 5.325355529785156,
"learning_rate": 7.575709779179812e-06,
"loss": 0.9197476196289063,
"step": 3375
},
{
"epoch": 2.6194144838212634,
"grad_norm": 5.787827968597412,
"learning_rate": 7.5559936908517365e-06,
"loss": 0.8949344635009766,
"step": 3400
},
{
"epoch": 2.6386748844375965,
"grad_norm": 5.829502105712891,
"learning_rate": 7.5362776025236595e-06,
"loss": 0.886966323852539,
"step": 3425
},
{
"epoch": 2.657935285053929,
"grad_norm": 5.501461505889893,
"learning_rate": 7.516561514195584e-06,
"loss": 0.909044418334961,
"step": 3450
},
{
"epoch": 2.677195685670262,
"grad_norm": 5.974215984344482,
"learning_rate": 7.496845425867508e-06,
"loss": 0.8642605590820313,
"step": 3475
},
{
"epoch": 2.6964560862865947,
"grad_norm": 6.091645240783691,
"learning_rate": 7.477129337539433e-06,
"loss": 0.9022230529785156,
"step": 3500
},
{
"epoch": 2.7157164869029273,
"grad_norm": 6.491245746612549,
"learning_rate": 7.457413249211357e-06,
"loss": 0.8908538055419922,
"step": 3525
},
{
"epoch": 2.7349768875192604,
"grad_norm": 6.0474958419799805,
"learning_rate": 7.4376971608832815e-06,
"loss": 0.8709038543701172,
"step": 3550
},
{
"epoch": 2.7542372881355934,
"grad_norm": 5.613045692443848,
"learning_rate": 7.417981072555206e-06,
"loss": 0.907492904663086,
"step": 3575
},
{
"epoch": 2.773497688751926,
"grad_norm": 5.941285610198975,
"learning_rate": 7.39826498422713e-06,
"loss": 0.8690555572509766,
"step": 3600
},
{
"epoch": 2.7927580893682586,
"grad_norm": 6.043056011199951,
"learning_rate": 7.378548895899055e-06,
"loss": 0.9102870941162109,
"step": 3625
},
{
"epoch": 2.8120184899845917,
"grad_norm": 5.732340335845947,
"learning_rate": 7.358832807570978e-06,
"loss": 0.8737263488769531,
"step": 3650
},
{
"epoch": 2.8312788906009247,
"grad_norm": 5.132423400878906,
"learning_rate": 7.339116719242903e-06,
"loss": 0.9003482818603515,
"step": 3675
},
{
"epoch": 2.8505392912172574,
"grad_norm": 5.89089298248291,
"learning_rate": 7.319400630914827e-06,
"loss": 0.8797879028320312,
"step": 3700
},
{
"epoch": 2.86979969183359,
"grad_norm": 5.688767910003662,
"learning_rate": 7.299684542586751e-06,
"loss": 0.9052462768554688,
"step": 3725
},
{
"epoch": 2.889060092449923,
"grad_norm": 4.86261510848999,
"learning_rate": 7.279968454258676e-06,
"loss": 0.907494125366211,
"step": 3750
},
{
"epoch": 2.9083204930662556,
"grad_norm": 5.176732540130615,
"learning_rate": 7.2602523659306e-06,
"loss": 0.8940404510498047,
"step": 3775
},
{
"epoch": 2.9275808936825887,
"grad_norm": 5.177511692047119,
"learning_rate": 7.240536277602525e-06,
"loss": 0.8763419342041016,
"step": 3800
},
{
"epoch": 2.9468412942989213,
"grad_norm": 5.829214572906494,
"learning_rate": 7.220820189274449e-06,
"loss": 0.8920821380615235,
"step": 3825
},
{
"epoch": 2.9661016949152543,
"grad_norm": 5.4694037437438965,
"learning_rate": 7.201104100946373e-06,
"loss": 0.8830232238769531,
"step": 3850
},
{
"epoch": 2.985362095531587,
"grad_norm": 5.96112585067749,
"learning_rate": 7.1813880126182965e-06,
"loss": 0.9158976745605468,
"step": 3875
},
{
"epoch": 3.0,
"eval_loss": 0.1625322848558426,
"eval_runtime": 1136.4451,
"eval_samples_per_second": 21.314,
"eval_steps_per_second": 0.096,
"eval_wer": 21.166896635512398,
"step": 3894
},
{
"epoch": 3.00462249614792,
"grad_norm": 5.552287578582764,
"learning_rate": 7.161671924290221e-06,
"loss": 0.8661737823486328,
"step": 3900
},
{
"epoch": 3.0238828967642526,
"grad_norm": 6.5318603515625,
"learning_rate": 7.141955835962145e-06,
"loss": 0.7396669006347656,
"step": 3925
},
{
"epoch": 3.0431432973805856,
"grad_norm": 4.689014911651611,
"learning_rate": 7.12223974763407e-06,
"loss": 0.7659902191162109,
"step": 3950
},
{
"epoch": 3.0624036979969183,
"grad_norm": 5.672070026397705,
"learning_rate": 7.102523659305995e-06,
"loss": 0.7574844360351562,
"step": 3975
},
{
"epoch": 3.0816640986132513,
"grad_norm": 4.803565502166748,
"learning_rate": 7.0828075709779185e-06,
"loss": 0.7242268371582031,
"step": 4000
},
{
"epoch": 3.100924499229584,
"grad_norm": 5.713749408721924,
"learning_rate": 7.063091482649843e-06,
"loss": 0.732081298828125,
"step": 4025
},
{
"epoch": 3.120184899845917,
"grad_norm": 5.710560321807861,
"learning_rate": 7.043375394321767e-06,
"loss": 0.7410992431640625,
"step": 4050
},
{
"epoch": 3.1394453004622496,
"grad_norm": 4.877614974975586,
"learning_rate": 7.023659305993692e-06,
"loss": 0.7561956024169922,
"step": 4075
},
{
"epoch": 3.1587057010785826,
"grad_norm": 5.199946403503418,
"learning_rate": 7.003943217665615e-06,
"loss": 0.7609020233154297,
"step": 4100
},
{
"epoch": 3.1779661016949152,
"grad_norm": 5.588535785675049,
"learning_rate": 6.98422712933754e-06,
"loss": 0.7510771179199218,
"step": 4125
},
{
"epoch": 3.1972265023112483,
"grad_norm": 4.650348663330078,
"learning_rate": 6.9645110410094644e-06,
"loss": 0.7404030609130859,
"step": 4150
},
{
"epoch": 3.216486902927581,
"grad_norm": 5.466901779174805,
"learning_rate": 6.944794952681388e-06,
"loss": 0.7516416931152343,
"step": 4175
},
{
"epoch": 3.2357473035439135,
"grad_norm": 5.119050025939941,
"learning_rate": 6.925078864353313e-06,
"loss": 0.7514974212646485,
"step": 4200
},
{
"epoch": 3.2550077041602465,
"grad_norm": 4.731340408325195,
"learning_rate": 6.905362776025237e-06,
"loss": 0.7492866516113281,
"step": 4225
},
{
"epoch": 3.2742681047765796,
"grad_norm": 4.669198036193848,
"learning_rate": 6.885646687697162e-06,
"loss": 0.7451268768310547,
"step": 4250
},
{
"epoch": 3.293528505392912,
"grad_norm": 5.735082626342773,
"learning_rate": 6.8659305993690865e-06,
"loss": 0.7610366058349609,
"step": 4275
},
{
"epoch": 3.312788906009245,
"grad_norm": 5.074744701385498,
"learning_rate": 6.84621451104101e-06,
"loss": 0.7417109680175781,
"step": 4300
},
{
"epoch": 3.332049306625578,
"grad_norm": 6.84859561920166,
"learning_rate": 6.8264984227129334e-06,
"loss": 0.7541703033447266,
"step": 4325
},
{
"epoch": 3.3513097072419105,
"grad_norm": 5.154798984527588,
"learning_rate": 6.806782334384858e-06,
"loss": 0.7489517974853516,
"step": 4350
},
{
"epoch": 3.3705701078582435,
"grad_norm": 5.426772117614746,
"learning_rate": 6.787066246056783e-06,
"loss": 0.7446839904785156,
"step": 4375
},
{
"epoch": 3.389830508474576,
"grad_norm": 5.358837604522705,
"learning_rate": 6.767350157728707e-06,
"loss": 0.7538381958007813,
"step": 4400
},
{
"epoch": 3.409090909090909,
"grad_norm": 5.3394904136657715,
"learning_rate": 6.7476340694006316e-06,
"loss": 0.7434349060058594,
"step": 4425
},
{
"epoch": 3.428351309707242,
"grad_norm": 6.098830223083496,
"learning_rate": 6.727917981072556e-06,
"loss": 0.7466286468505859,
"step": 4450
},
{
"epoch": 3.447611710323575,
"grad_norm": 5.0356831550598145,
"learning_rate": 6.70820189274448e-06,
"loss": 0.7625394439697266,
"step": 4475
},
{
"epoch": 3.4668721109399074,
"grad_norm": 5.831299781799316,
"learning_rate": 6.688485804416405e-06,
"loss": 0.7527046203613281,
"step": 4500
},
{
"epoch": 3.4861325115562405,
"grad_norm": 5.351603984832764,
"learning_rate": 6.668769716088329e-06,
"loss": 0.7737387084960937,
"step": 4525
},
{
"epoch": 3.505392912172573,
"grad_norm": 5.928788185119629,
"learning_rate": 6.649053627760253e-06,
"loss": 0.743348388671875,
"step": 4550
},
{
"epoch": 3.524653312788906,
"grad_norm": 5.242292404174805,
"learning_rate": 6.629337539432177e-06,
"loss": 0.7569735717773437,
"step": 4575
},
{
"epoch": 3.5439137134052388,
"grad_norm": 5.537062168121338,
"learning_rate": 6.609621451104101e-06,
"loss": 0.7640843200683594,
"step": 4600
},
{
"epoch": 3.563174114021572,
"grad_norm": 4.72006893157959,
"learning_rate": 6.589905362776025e-06,
"loss": 0.7594134521484375,
"step": 4625
},
{
"epoch": 3.5824345146379044,
"grad_norm": 5.570235252380371,
"learning_rate": 6.57018927444795e-06,
"loss": 0.7546492004394532,
"step": 4650
},
{
"epoch": 3.601694915254237,
"grad_norm": 5.033544063568115,
"learning_rate": 6.550473186119875e-06,
"loss": 0.7491130065917969,
"step": 4675
},
{
"epoch": 3.62095531587057,
"grad_norm": 5.854029178619385,
"learning_rate": 6.530757097791799e-06,
"loss": 0.7524951171875,
"step": 4700
},
{
"epoch": 3.640215716486903,
"grad_norm": 4.8701043128967285,
"learning_rate": 6.5110410094637234e-06,
"loss": 0.7602452087402344,
"step": 4725
},
{
"epoch": 3.6594761171032357,
"grad_norm": 5.200259685516357,
"learning_rate": 6.491324921135647e-06,
"loss": 0.7586729431152344,
"step": 4750
},
{
"epoch": 3.6787365177195683,
"grad_norm": 5.2361297607421875,
"learning_rate": 6.471608832807571e-06,
"loss": 0.7725418853759766,
"step": 4775
},
{
"epoch": 3.6979969183359014,
"grad_norm": 5.228827476501465,
"learning_rate": 6.451892744479495e-06,
"loss": 0.7784210205078125,
"step": 4800
},
{
"epoch": 3.7172573189522344,
"grad_norm": 4.758535385131836,
"learning_rate": 6.43217665615142e-06,
"loss": 0.7555014038085938,
"step": 4825
},
{
"epoch": 3.736517719568567,
"grad_norm": 5.373162269592285,
"learning_rate": 6.412460567823345e-06,
"loss": 0.7722117614746093,
"step": 4850
},
{
"epoch": 3.7557781201848996,
"grad_norm": 5.506257057189941,
"learning_rate": 6.3927444794952685e-06,
"loss": 0.7604951477050781,
"step": 4875
},
{
"epoch": 3.7750385208012327,
"grad_norm": 4.666244983673096,
"learning_rate": 6.373028391167193e-06,
"loss": 0.7671040344238281,
"step": 4900
},
{
"epoch": 3.7942989214175658,
"grad_norm": 6.130381107330322,
"learning_rate": 6.353312302839117e-06,
"loss": 0.7677198791503906,
"step": 4925
},
{
"epoch": 3.8135593220338984,
"grad_norm": 5.654706001281738,
"learning_rate": 6.333596214511042e-06,
"loss": 0.7743594360351562,
"step": 4950
},
{
"epoch": 3.832819722650231,
"grad_norm": 5.019384384155273,
"learning_rate": 6.313880126182967e-06,
"loss": 0.7450533294677735,
"step": 4975
},
{
"epoch": 3.852080123266564,
"grad_norm": 6.141390323638916,
"learning_rate": 6.29416403785489e-06,
"loss": 0.7698895263671875,
"step": 5000
},
{
"epoch": 3.8713405238828966,
"grad_norm": 6.0247883796691895,
"learning_rate": 6.274447949526814e-06,
"loss": 0.7623857879638671,
"step": 5025
},
{
"epoch": 3.8906009244992297,
"grad_norm": 4.987785339355469,
"learning_rate": 6.254731861198738e-06,
"loss": 0.7761954498291016,
"step": 5050
},
{
"epoch": 3.9098613251155623,
"grad_norm": 6.38145637512207,
"learning_rate": 6.235015772870663e-06,
"loss": 0.758653793334961,
"step": 5075
},
{
"epoch": 3.9291217257318953,
"grad_norm": 4.845369338989258,
"learning_rate": 6.215299684542587e-06,
"loss": 0.7622804260253906,
"step": 5100
},
{
"epoch": 3.948382126348228,
"grad_norm": 4.98997163772583,
"learning_rate": 6.195583596214512e-06,
"loss": 0.7737734985351562,
"step": 5125
},
{
"epoch": 3.967642526964561,
"grad_norm": 5.214897632598877,
"learning_rate": 6.1758675078864365e-06,
"loss": 0.7438212585449219,
"step": 5150
},
{
"epoch": 3.9869029275808936,
"grad_norm": 6.027677059173584,
"learning_rate": 6.15615141955836e-06,
"loss": 0.7527304077148438,
"step": 5175
},
{
"epoch": 4.0,
"eval_loss": 0.17093561589717865,
"eval_runtime": 1096.3535,
"eval_samples_per_second": 22.093,
"eval_steps_per_second": 0.099,
"eval_wer": 20.828885822637613,
"step": 5192
},
{
"epoch": 4.006163328197227,
"grad_norm": 5.08741569519043,
"learning_rate": 6.136435331230285e-06,
"loss": 0.7012213134765625,
"step": 5200
},
{
"epoch": 4.02542372881356,
"grad_norm": 5.682927131652832,
"learning_rate": 6.116719242902208e-06,
"loss": 0.6227888870239258,
"step": 5225
},
{
"epoch": 4.044684129429892,
"grad_norm": 4.897796630859375,
"learning_rate": 6.097003154574133e-06,
"loss": 0.6233341979980469,
"step": 5250
},
{
"epoch": 4.063944530046225,
"grad_norm": 6.243351459503174,
"learning_rate": 6.077287066246057e-06,
"loss": 0.6161975479125976,
"step": 5275
},
{
"epoch": 4.083204930662558,
"grad_norm": 5.645864009857178,
"learning_rate": 6.057570977917982e-06,
"loss": 0.595079460144043,
"step": 5300
},
{
"epoch": 4.102465331278891,
"grad_norm": 5.0616278648376465,
"learning_rate": 6.0378548895899055e-06,
"loss": 0.6014439392089844,
"step": 5325
},
{
"epoch": 4.121725731895223,
"grad_norm": 5.87223482131958,
"learning_rate": 6.01813880126183e-06,
"loss": 0.6178970718383789,
"step": 5350
},
{
"epoch": 4.140986132511556,
"grad_norm": 5.19507360458374,
"learning_rate": 5.998422712933755e-06,
"loss": 0.6022650527954102,
"step": 5375
},
{
"epoch": 4.160246533127889,
"grad_norm": 5.169468879699707,
"learning_rate": 5.978706624605679e-06,
"loss": 0.6180243301391601,
"step": 5400
},
{
"epoch": 4.179506933744221,
"grad_norm": 5.511230945587158,
"learning_rate": 5.958990536277604e-06,
"loss": 0.5995005798339844,
"step": 5425
},
{
"epoch": 4.1987673343605545,
"grad_norm": 5.659884452819824,
"learning_rate": 5.939274447949527e-06,
"loss": 0.6076582717895508,
"step": 5450
},
{
"epoch": 4.2180277349768875,
"grad_norm": 5.777623176574707,
"learning_rate": 5.919558359621451e-06,
"loss": 0.6149098587036133,
"step": 5475
},
{
"epoch": 4.237288135593221,
"grad_norm": 5.309556007385254,
"learning_rate": 5.899842271293375e-06,
"loss": 0.6108769607543946,
"step": 5500
},
{
"epoch": 4.256548536209553,
"grad_norm": 5.455935955047607,
"learning_rate": 5.8801261829653e-06,
"loss": 0.6126166534423828,
"step": 5525
},
{
"epoch": 4.275808936825886,
"grad_norm": 5.241320610046387,
"learning_rate": 5.860410094637225e-06,
"loss": 0.6169355010986328,
"step": 5550
},
{
"epoch": 4.295069337442219,
"grad_norm": 5.369098663330078,
"learning_rate": 5.840694006309149e-06,
"loss": 0.5984256744384766,
"step": 5575
},
{
"epoch": 4.314329738058552,
"grad_norm": 4.623167037963867,
"learning_rate": 5.8209779179810735e-06,
"loss": 0.6112961959838867,
"step": 5600
},
{
"epoch": 4.333590138674884,
"grad_norm": 5.853650093078613,
"learning_rate": 5.801261829652997e-06,
"loss": 0.6166348266601562,
"step": 5625
},
{
"epoch": 4.352850539291217,
"grad_norm": 4.812239646911621,
"learning_rate": 5.781545741324922e-06,
"loss": 0.597599983215332,
"step": 5650
},
{
"epoch": 4.37211093990755,
"grad_norm": 4.949807167053223,
"learning_rate": 5.761829652996845e-06,
"loss": 0.6153563690185547,
"step": 5675
},
{
"epoch": 4.391371340523883,
"grad_norm": 4.839872360229492,
"learning_rate": 5.74211356466877e-06,
"loss": 0.6126491165161133,
"step": 5700
},
{
"epoch": 4.410631741140215,
"grad_norm": 5.813593864440918,
"learning_rate": 5.722397476340694e-06,
"loss": 0.6261103820800781,
"step": 5725
},
{
"epoch": 4.429892141756548,
"grad_norm": 5.7325239181518555,
"learning_rate": 5.7026813880126185e-06,
"loss": 0.6063761901855469,
"step": 5750
},
{
"epoch": 4.4491525423728815,
"grad_norm": 5.425716876983643,
"learning_rate": 5.682965299684543e-06,
"loss": 0.6294102096557617,
"step": 5775
},
{
"epoch": 4.4684129429892145,
"grad_norm": 5.379292964935303,
"learning_rate": 5.663249211356467e-06,
"loss": 0.6135920333862305,
"step": 5800
},
{
"epoch": 4.487673343605547,
"grad_norm": 5.157162189483643,
"learning_rate": 5.643533123028392e-06,
"loss": 0.6252361297607422,
"step": 5825
},
{
"epoch": 4.50693374422188,
"grad_norm": 5.1321702003479,
"learning_rate": 5.623817034700316e-06,
"loss": 0.6251173400878907,
"step": 5850
},
{
"epoch": 4.526194144838213,
"grad_norm": 5.372640609741211,
"learning_rate": 5.604100946372241e-06,
"loss": 0.6194912719726563,
"step": 5875
},
{
"epoch": 4.545454545454545,
"grad_norm": 5.1659626960754395,
"learning_rate": 5.584384858044164e-06,
"loss": 0.6289920806884766,
"step": 5900
},
{
"epoch": 4.564714946070878,
"grad_norm": 5.255802631378174,
"learning_rate": 5.564668769716088e-06,
"loss": 0.6240493011474609,
"step": 5925
},
{
"epoch": 4.583975346687211,
"grad_norm": 5.6852803230285645,
"learning_rate": 5.544952681388013e-06,
"loss": 0.6095463562011719,
"step": 5950
},
{
"epoch": 4.603235747303544,
"grad_norm": 5.392953872680664,
"learning_rate": 5.525236593059937e-06,
"loss": 0.6142960357666015,
"step": 5975
},
{
"epoch": 4.622496147919877,
"grad_norm": 5.90056037902832,
"learning_rate": 5.505520504731862e-06,
"loss": 0.635225944519043,
"step": 6000
},
{
"epoch": 4.641756548536209,
"grad_norm": 5.827224254608154,
"learning_rate": 5.485804416403786e-06,
"loss": 0.6327444458007813,
"step": 6025
},
{
"epoch": 4.661016949152542,
"grad_norm": 5.974770545959473,
"learning_rate": 5.46608832807571e-06,
"loss": 0.60680419921875,
"step": 6050
},
{
"epoch": 4.680277349768875,
"grad_norm": 5.715795993804932,
"learning_rate": 5.446372239747635e-06,
"loss": 0.6298756027221679,
"step": 6075
},
{
"epoch": 4.699537750385208,
"grad_norm": 5.36166524887085,
"learning_rate": 5.426656151419559e-06,
"loss": 0.6100086975097656,
"step": 6100
},
{
"epoch": 4.718798151001541,
"grad_norm": 5.193503379821777,
"learning_rate": 5.406940063091483e-06,
"loss": 0.6205142211914062,
"step": 6125
},
{
"epoch": 4.738058551617874,
"grad_norm": 5.747341632843018,
"learning_rate": 5.387223974763407e-06,
"loss": 0.6245947647094726,
"step": 6150
},
{
"epoch": 4.757318952234207,
"grad_norm": 5.33910608291626,
"learning_rate": 5.367507886435332e-06,
"loss": 0.6322902297973633,
"step": 6175
},
{
"epoch": 4.776579352850539,
"grad_norm": 5.16536283493042,
"learning_rate": 5.3477917981072555e-06,
"loss": 0.6207213592529297,
"step": 6200
},
{
"epoch": 4.795839753466872,
"grad_norm": 5.795975208282471,
"learning_rate": 5.32807570977918e-06,
"loss": 0.624023551940918,
"step": 6225
},
{
"epoch": 4.815100154083205,
"grad_norm": 5.030208110809326,
"learning_rate": 5.308359621451105e-06,
"loss": 0.6217356109619141,
"step": 6250
},
{
"epoch": 4.834360554699538,
"grad_norm": 6.521645545959473,
"learning_rate": 5.288643533123029e-06,
"loss": 0.6168215942382812,
"step": 6275
},
{
"epoch": 4.85362095531587,
"grad_norm": 5.482723712921143,
"learning_rate": 5.268927444794954e-06,
"loss": 0.6234802627563476,
"step": 6300
},
{
"epoch": 4.872881355932203,
"grad_norm": 5.789270877838135,
"learning_rate": 5.2492113564668775e-06,
"loss": 0.6294479370117188,
"step": 6325
},
{
"epoch": 4.892141756548536,
"grad_norm": 5.041879653930664,
"learning_rate": 5.2294952681388014e-06,
"loss": 0.6222351837158203,
"step": 6350
},
{
"epoch": 4.911402157164869,
"grad_norm": 6.19069766998291,
"learning_rate": 5.209779179810725e-06,
"loss": 0.6256502151489258,
"step": 6375
},
{
"epoch": 4.9306625577812015,
"grad_norm": 5.716239929199219,
"learning_rate": 5.19006309148265e-06,
"loss": 0.6432769012451172,
"step": 6400
},
{
"epoch": 4.949922958397535,
"grad_norm": 5.757542133331299,
"learning_rate": 5.170347003154574e-06,
"loss": 0.6266519927978516,
"step": 6425
},
{
"epoch": 4.969183359013868,
"grad_norm": 5.742589950561523,
"learning_rate": 5.150630914826499e-06,
"loss": 0.6203784561157226,
"step": 6450
},
{
"epoch": 4.988443759630201,
"grad_norm": 5.062778949737549,
"learning_rate": 5.1309148264984235e-06,
"loss": 0.6260740661621094,
"step": 6475
},
{
"epoch": 5.0,
"eval_loss": 0.18512727320194244,
"eval_runtime": 1120.4935,
"eval_samples_per_second": 21.617,
"eval_steps_per_second": 0.097,
"eval_wer": 21.49688023396076,
"step": 6490
},
{
"epoch": 5.007704160246533,
"grad_norm": 5.034289360046387,
"learning_rate": 5.111198738170347e-06,
"loss": 0.5662796783447266,
"step": 6500
},
{
"epoch": 5.026964560862866,
"grad_norm": 5.292657375335693,
"learning_rate": 5.091482649842272e-06,
"loss": 0.47537033081054686,
"step": 6525
},
{
"epoch": 5.046224961479199,
"grad_norm": 5.090963363647461,
"learning_rate": 5.071766561514196e-06,
"loss": 0.47929203033447265,
"step": 6550
},
{
"epoch": 5.065485362095532,
"grad_norm": 5.760522842407227,
"learning_rate": 5.05205047318612e-06,
"loss": 0.4834568786621094,
"step": 6575
},
{
"epoch": 5.084745762711864,
"grad_norm": 4.214590549468994,
"learning_rate": 5.032334384858044e-06,
"loss": 0.47515312194824216,
"step": 6600
},
{
"epoch": 5.104006163328197,
"grad_norm": 5.125479221343994,
"learning_rate": 5.0126182965299686e-06,
"loss": 0.47720046997070314,
"step": 6625
},
{
"epoch": 5.12326656394453,
"grad_norm": 5.678186416625977,
"learning_rate": 4.992902208201893e-06,
"loss": 0.48319450378417966,
"step": 6650
},
{
"epoch": 5.1425269645608624,
"grad_norm": 5.061611652374268,
"learning_rate": 4.973186119873817e-06,
"loss": 0.47992664337158203,
"step": 6675
},
{
"epoch": 5.1617873651771955,
"grad_norm": 5.910008907318115,
"learning_rate": 4.953470031545742e-06,
"loss": 0.48137985229492186,
"step": 6700
},
{
"epoch": 5.1810477657935285,
"grad_norm": 6.257612228393555,
"learning_rate": 4.933753943217666e-06,
"loss": 0.4876537322998047,
"step": 6725
},
{
"epoch": 5.200308166409862,
"grad_norm": 5.265006065368652,
"learning_rate": 4.91403785488959e-06,
"loss": 0.47759193420410156,
"step": 6750
},
{
"epoch": 5.219568567026194,
"grad_norm": 5.452073097229004,
"learning_rate": 4.8943217665615145e-06,
"loss": 0.4787986373901367,
"step": 6775
},
{
"epoch": 5.238828967642527,
"grad_norm": 5.545107364654541,
"learning_rate": 4.874605678233439e-06,
"loss": 0.47380306243896486,
"step": 6800
},
{
"epoch": 5.25808936825886,
"grad_norm": 5.7491865158081055,
"learning_rate": 4.854889589905363e-06,
"loss": 0.4716271209716797,
"step": 6825
},
{
"epoch": 5.277349768875193,
"grad_norm": 5.415458679199219,
"learning_rate": 4.835173501577288e-06,
"loss": 0.4782722091674805,
"step": 6850
},
{
"epoch": 5.296610169491525,
"grad_norm": 5.625531196594238,
"learning_rate": 4.815457413249212e-06,
"loss": 0.4818852996826172,
"step": 6875
},
{
"epoch": 5.315870570107858,
"grad_norm": 5.539705276489258,
"learning_rate": 4.795741324921136e-06,
"loss": 0.47021194458007814,
"step": 6900
},
{
"epoch": 5.335130970724191,
"grad_norm": 6.839880466461182,
"learning_rate": 4.7760252365930604e-06,
"loss": 0.4833596420288086,
"step": 6925
},
{
"epoch": 5.354391371340524,
"grad_norm": 5.86253023147583,
"learning_rate": 4.756309148264985e-06,
"loss": 0.495352668762207,
"step": 6950
},
{
"epoch": 5.373651771956856,
"grad_norm": 5.184374809265137,
"learning_rate": 4.736593059936908e-06,
"loss": 0.4801821517944336,
"step": 6975
},
{
"epoch": 5.3929121725731894,
"grad_norm": 5.105012893676758,
"learning_rate": 4.716876971608833e-06,
"loss": 0.47543926239013673,
"step": 7000
},
{
"epoch": 5.4121725731895225,
"grad_norm": 5.638117790222168,
"learning_rate": 4.697160883280758e-06,
"loss": 0.4772523498535156,
"step": 7025
},
{
"epoch": 5.4314329738058555,
"grad_norm": 5.81176233291626,
"learning_rate": 4.677444794952682e-06,
"loss": 0.4763611221313477,
"step": 7050
},
{
"epoch": 5.450693374422188,
"grad_norm": 6.53574275970459,
"learning_rate": 4.657728706624606e-06,
"loss": 0.4876799774169922,
"step": 7075
},
{
"epoch": 5.469953775038521,
"grad_norm": 5.385960578918457,
"learning_rate": 4.63801261829653e-06,
"loss": 0.4872008514404297,
"step": 7100
},
{
"epoch": 5.489214175654854,
"grad_norm": 5.694986343383789,
"learning_rate": 4.618296529968454e-06,
"loss": 0.48962074279785156,
"step": 7125
},
{
"epoch": 5.508474576271187,
"grad_norm": 4.94402551651001,
"learning_rate": 4.598580441640379e-06,
"loss": 0.4754409027099609,
"step": 7150
},
{
"epoch": 5.527734976887519,
"grad_norm": 5.079399108886719,
"learning_rate": 4.578864353312304e-06,
"loss": 0.4800136184692383,
"step": 7175
},
{
"epoch": 5.546995377503852,
"grad_norm": 6.455406188964844,
"learning_rate": 4.5591482649842276e-06,
"loss": 0.4975376892089844,
"step": 7200
},
{
"epoch": 5.566255778120185,
"grad_norm": 5.424092769622803,
"learning_rate": 4.5394321766561515e-06,
"loss": 0.4917482376098633,
"step": 7225
},
{
"epoch": 5.585516178736517,
"grad_norm": 6.038609981536865,
"learning_rate": 4.519716088328076e-06,
"loss": 0.48865249633789065,
"step": 7250
},
{
"epoch": 5.60477657935285,
"grad_norm": 5.937739372253418,
"learning_rate": 4.5e-06,
"loss": 0.4839106369018555,
"step": 7275
},
{
"epoch": 5.624036979969183,
"grad_norm": 5.836992263793945,
"learning_rate": 4.480283911671925e-06,
"loss": 0.4853042221069336,
"step": 7300
},
{
"epoch": 5.643297380585516,
"grad_norm": 6.14811897277832,
"learning_rate": 4.460567823343849e-06,
"loss": 0.4945429992675781,
"step": 7325
},
{
"epoch": 5.6625577812018495,
"grad_norm": 5.823431968688965,
"learning_rate": 4.4408517350157735e-06,
"loss": 0.48616283416748046,
"step": 7350
},
{
"epoch": 5.681818181818182,
"grad_norm": 6.761408805847168,
"learning_rate": 4.421135646687697e-06,
"loss": 0.4741324234008789,
"step": 7375
},
{
"epoch": 5.701078582434515,
"grad_norm": 5.743547439575195,
"learning_rate": 4.401419558359622e-06,
"loss": 0.48080249786376955,
"step": 7400
},
{
"epoch": 5.720338983050848,
"grad_norm": 6.16947078704834,
"learning_rate": 4.381703470031546e-06,
"loss": 0.47442409515380857,
"step": 7425
},
{
"epoch": 5.73959938366718,
"grad_norm": 5.7636613845825195,
"learning_rate": 4.36198738170347e-06,
"loss": 0.4784553909301758,
"step": 7450
},
{
"epoch": 5.758859784283513,
"grad_norm": 5.6138505935668945,
"learning_rate": 4.342271293375395e-06,
"loss": 0.47864688873291017,
"step": 7475
},
{
"epoch": 5.778120184899846,
"grad_norm": 5.92537260055542,
"learning_rate": 4.3225552050473194e-06,
"loss": 0.4892653274536133,
"step": 7500
},
{
"epoch": 5.797380585516179,
"grad_norm": 5.650518894195557,
"learning_rate": 4.302839116719243e-06,
"loss": 0.4778163146972656,
"step": 7525
},
{
"epoch": 5.816640986132511,
"grad_norm": 6.602982521057129,
"learning_rate": 4.283123028391167e-06,
"loss": 0.503909683227539,
"step": 7550
},
{
"epoch": 5.835901386748844,
"grad_norm": 5.893745422363281,
"learning_rate": 4.263406940063092e-06,
"loss": 0.4868201446533203,
"step": 7575
},
{
"epoch": 5.855161787365177,
"grad_norm": 6.460921287536621,
"learning_rate": 4.243690851735016e-06,
"loss": 0.48137042999267576,
"step": 7600
},
{
"epoch": 5.87442218798151,
"grad_norm": 5.695516109466553,
"learning_rate": 4.223974763406941e-06,
"loss": 0.4823105621337891,
"step": 7625
},
{
"epoch": 5.8936825885978426,
"grad_norm": 5.476104259490967,
"learning_rate": 4.2042586750788645e-06,
"loss": 0.47455204010009766,
"step": 7650
},
{
"epoch": 5.912942989214176,
"grad_norm": 5.6681671142578125,
"learning_rate": 4.1845425867507884e-06,
"loss": 0.49075904846191404,
"step": 7675
},
{
"epoch": 5.932203389830509,
"grad_norm": 6.429839611053467,
"learning_rate": 4.164826498422713e-06,
"loss": 0.47588310241699217,
"step": 7700
},
{
"epoch": 5.951463790446841,
"grad_norm": 5.590009689331055,
"learning_rate": 4.145110410094638e-06,
"loss": 0.48280776977539064,
"step": 7725
},
{
"epoch": 5.970724191063174,
"grad_norm": 5.808004379272461,
"learning_rate": 4.125394321766562e-06,
"loss": 0.4881787490844727,
"step": 7750
},
{
"epoch": 5.989984591679507,
"grad_norm": 5.411890983581543,
"learning_rate": 4.105678233438486e-06,
"loss": 0.4787063980102539,
"step": 7775
},
{
"epoch": 6.0,
"eval_loss": 0.20830762386322021,
"eval_runtime": 1076.2889,
"eval_samples_per_second": 22.505,
"eval_steps_per_second": 0.101,
"eval_wer": 21.849860716982114,
"step": 7788
},
{
"epoch": 6.00924499229584,
"grad_norm": 4.889241695404053,
"learning_rate": 4.0859621451104105e-06,
"loss": 0.42858932495117186,
"step": 7800
},
{
"epoch": 6.028505392912172,
"grad_norm": 5.174338340759277,
"learning_rate": 4.066246056782334e-06,
"loss": 0.3610303497314453,
"step": 7825
},
{
"epoch": 6.047765793528505,
"grad_norm": 5.670193195343018,
"learning_rate": 4.046529968454259e-06,
"loss": 0.351124153137207,
"step": 7850
},
{
"epoch": 6.067026194144838,
"grad_norm": 5.768033981323242,
"learning_rate": 4.026813880126184e-06,
"loss": 0.3566075134277344,
"step": 7875
},
{
"epoch": 6.086286594761171,
"grad_norm": 5.415530681610107,
"learning_rate": 4.007097791798108e-06,
"loss": 0.35931068420410156,
"step": 7900
},
{
"epoch": 6.1055469953775034,
"grad_norm": 4.7216796875,
"learning_rate": 3.987381703470032e-06,
"loss": 0.3501945114135742,
"step": 7925
},
{
"epoch": 6.1248073959938365,
"grad_norm": 5.062224864959717,
"learning_rate": 3.967665615141956e-06,
"loss": 0.3557395935058594,
"step": 7950
},
{
"epoch": 6.1440677966101696,
"grad_norm": 5.604755878448486,
"learning_rate": 3.94794952681388e-06,
"loss": 0.3643395614624023,
"step": 7975
},
{
"epoch": 6.163328197226503,
"grad_norm": 5.9803009033203125,
"learning_rate": 3.928233438485804e-06,
"loss": 0.36169349670410156,
"step": 8000
},
{
"epoch": 6.182588597842835,
"grad_norm": 5.072482585906982,
"learning_rate": 3.908517350157729e-06,
"loss": 0.35094886779785156,
"step": 8025
},
{
"epoch": 6.201848998459168,
"grad_norm": 5.030255317687988,
"learning_rate": 3.888801261829654e-06,
"loss": 0.35401191711425783,
"step": 8050
},
{
"epoch": 6.221109399075501,
"grad_norm": 5.7693023681640625,
"learning_rate": 3.869085173501578e-06,
"loss": 0.35537750244140626,
"step": 8075
},
{
"epoch": 6.240369799691834,
"grad_norm": 5.407600402832031,
"learning_rate": 3.849369085173502e-06,
"loss": 0.3509661865234375,
"step": 8100
},
{
"epoch": 6.259630200308166,
"grad_norm": 5.739716529846191,
"learning_rate": 3.829652996845426e-06,
"loss": 0.35854976654052734,
"step": 8125
},
{
"epoch": 6.278890600924499,
"grad_norm": 5.548368453979492,
"learning_rate": 3.8099369085173506e-06,
"loss": 0.3608596420288086,
"step": 8150
},
{
"epoch": 6.298151001540832,
"grad_norm": 16.185115814208984,
"learning_rate": 3.790220820189275e-06,
"loss": 0.3938618850708008,
"step": 8175
},
{
"epoch": 6.317411402157165,
"grad_norm": 5.405980110168457,
"learning_rate": 3.770504731861199e-06,
"loss": 0.3615694046020508,
"step": 8200
},
{
"epoch": 6.336671802773497,
"grad_norm": 6.4496660232543945,
"learning_rate": 3.750788643533123e-06,
"loss": 0.36363319396972654,
"step": 8225
},
{
"epoch": 6.3559322033898304,
"grad_norm": 5.091163635253906,
"learning_rate": 3.7310725552050474e-06,
"loss": 0.35484718322753905,
"step": 8250
},
{
"epoch": 6.3751926040061635,
"grad_norm": 5.882087707519531,
"learning_rate": 3.7113564668769718e-06,
"loss": 0.36338321685791014,
"step": 8275
},
{
"epoch": 6.3944530046224966,
"grad_norm": 5.807699680328369,
"learning_rate": 3.6916403785488965e-06,
"loss": 0.3659554672241211,
"step": 8300
},
{
"epoch": 6.413713405238829,
"grad_norm": 5.627602577209473,
"learning_rate": 3.671924290220821e-06,
"loss": 0.36366878509521483,
"step": 8325
},
{
"epoch": 6.432973805855162,
"grad_norm": 6.055703163146973,
"learning_rate": 3.6522082018927447e-06,
"loss": 0.3526373291015625,
"step": 8350
},
{
"epoch": 6.452234206471495,
"grad_norm": 5.582004547119141,
"learning_rate": 3.632492113564669e-06,
"loss": 0.3500963592529297,
"step": 8375
},
{
"epoch": 6.471494607087827,
"grad_norm": 5.392638206481934,
"learning_rate": 3.6127760252365934e-06,
"loss": 0.35544357299804685,
"step": 8400
},
{
"epoch": 6.49075500770416,
"grad_norm": 6.201164722442627,
"learning_rate": 3.5930599369085177e-06,
"loss": 0.36704177856445314,
"step": 8425
},
{
"epoch": 6.510015408320493,
"grad_norm": 6.226408958435059,
"learning_rate": 3.5733438485804416e-06,
"loss": 0.3502212905883789,
"step": 8450
},
{
"epoch": 6.529275808936826,
"grad_norm": 5.938023090362549,
"learning_rate": 3.553627760252366e-06,
"loss": 0.3535276412963867,
"step": 8475
},
{
"epoch": 6.548536209553159,
"grad_norm": 5.055141448974609,
"learning_rate": 3.5339116719242907e-06,
"loss": 0.36611476898193357,
"step": 8500
},
{
"epoch": 6.567796610169491,
"grad_norm": 5.726068019866943,
"learning_rate": 3.514195583596215e-06,
"loss": 0.36194461822509766,
"step": 8525
},
{
"epoch": 6.587057010785824,
"grad_norm": 5.854666233062744,
"learning_rate": 3.4944794952681393e-06,
"loss": 0.35246604919433594,
"step": 8550
},
{
"epoch": 6.6063174114021574,
"grad_norm": 5.347609996795654,
"learning_rate": 3.474763406940063e-06,
"loss": 0.3571212387084961,
"step": 8575
},
{
"epoch": 6.62557781201849,
"grad_norm": 6.217226028442383,
"learning_rate": 3.4550473186119875e-06,
"loss": 0.3514929962158203,
"step": 8600
},
{
"epoch": 6.644838212634823,
"grad_norm": 5.599356174468994,
"learning_rate": 3.435331230283912e-06,
"loss": 0.3598286056518555,
"step": 8625
},
{
"epoch": 6.664098613251156,
"grad_norm": 5.66002082824707,
"learning_rate": 3.4156151419558366e-06,
"loss": 0.3629193115234375,
"step": 8650
},
{
"epoch": 6.683359013867489,
"grad_norm": 5.900570392608643,
"learning_rate": 3.3958990536277605e-06,
"loss": 0.36112438201904296,
"step": 8675
},
{
"epoch": 6.702619414483821,
"grad_norm": 5.6205058097839355,
"learning_rate": 3.376182965299685e-06,
"loss": 0.35814327239990235,
"step": 8700
},
{
"epoch": 6.721879815100154,
"grad_norm": 5.015433311462402,
"learning_rate": 3.356466876971609e-06,
"loss": 0.3551091384887695,
"step": 8725
},
{
"epoch": 6.741140215716487,
"grad_norm": 5.684232234954834,
"learning_rate": 3.3367507886435335e-06,
"loss": 0.36389984130859376,
"step": 8750
},
{
"epoch": 6.76040061633282,
"grad_norm": 6.055419921875,
"learning_rate": 3.3170347003154578e-06,
"loss": 0.36103096008300783,
"step": 8775
},
{
"epoch": 6.779661016949152,
"grad_norm": 5.762523174285889,
"learning_rate": 3.2973186119873817e-06,
"loss": 0.3544279479980469,
"step": 8800
},
{
"epoch": 6.798921417565485,
"grad_norm": 5.5188093185424805,
"learning_rate": 3.277602523659306e-06,
"loss": 0.3438130950927734,
"step": 8825
},
{
"epoch": 6.818181818181818,
"grad_norm": 5.758239269256592,
"learning_rate": 3.2578864353312307e-06,
"loss": 0.3498713684082031,
"step": 8850
},
{
"epoch": 6.8374422187981505,
"grad_norm": 5.552741050720215,
"learning_rate": 3.238170347003155e-06,
"loss": 0.35503765106201174,
"step": 8875
},
{
"epoch": 6.856702619414484,
"grad_norm": 5.898087501525879,
"learning_rate": 3.218454258675079e-06,
"loss": 0.3598700714111328,
"step": 8900
},
{
"epoch": 6.875963020030817,
"grad_norm": 6.149432182312012,
"learning_rate": 3.1987381703470033e-06,
"loss": 0.35951732635498046,
"step": 8925
},
{
"epoch": 6.89522342064715,
"grad_norm": 6.110939025878906,
"learning_rate": 3.1790220820189276e-06,
"loss": 0.3535256576538086,
"step": 8950
},
{
"epoch": 6.914483821263483,
"grad_norm": 5.947456359863281,
"learning_rate": 3.159305993690852e-06,
"loss": 0.3574999237060547,
"step": 8975
},
{
"epoch": 6.933744221879815,
"grad_norm": 5.893382549285889,
"learning_rate": 3.1395899053627767e-06,
"loss": 0.3584942626953125,
"step": 9000
},
{
"epoch": 6.953004622496148,
"grad_norm": 6.020825386047363,
"learning_rate": 3.1198738170347006e-06,
"loss": 0.3466789245605469,
"step": 9025
},
{
"epoch": 6.972265023112481,
"grad_norm": 6.020845890045166,
"learning_rate": 3.100157728706625e-06,
"loss": 0.36809425354003905,
"step": 9050
},
{
"epoch": 6.991525423728813,
"grad_norm": 5.895872116088867,
"learning_rate": 3.0804416403785492e-06,
"loss": 0.3656966018676758,
"step": 9075
},
{
"epoch": 7.0,
"eval_loss": 0.23961010575294495,
"eval_runtime": 1087.013,
"eval_samples_per_second": 22.283,
"eval_steps_per_second": 0.1,
"eval_wer": 22.459929013390262,
"step": 9086
}
],
"logging_steps": 25,
"max_steps": 12980,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 500,
"stateful_callbacks": {
"EarlyStoppingCallback": {
"args": {
"early_stopping_patience": 3,
"early_stopping_threshold": 0.0
},
"attributes": {
"early_stopping_patience_counter": 3
}
},
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.5579003297660278e+21,
"train_batch_size": 24,
"trial_name": null,
"trial_params": null
}