random_500_gold / epoch_2 /trainer_state.json
PranavJosh's picture
Upload epoch_2 checkpoint
5658d1c verified
Raw
History Blame Contribute Delete
20.5 kB
{
"best_global_step": 2596,
"best_metric": 21.05581734399,
"best_model_checkpoint": "./whisper_s_punjabi_ddp_4gpu/checkpoint-2596",
"epoch": 2.0,
"eval_steps": 500,
"global_step": 2596,
"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
}
],
"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": 0
}
},
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 4.45114379815346e+20,
"train_batch_size": 24,
"trial_name": null,
"trial_params": null
}