Invalid JSON:Unexpected token 'N', ..."ad_norm": NaN,
"... is not valid JSON
| { | |
| "best_global_step": 18532, | |
| "best_metric": 0.19350049787272564, | |
| "best_model_checkpoint": "/content/drive/MyDrive/asr_loracheckpoint2_backup/checkpoint-18532", | |
| "epoch": 44.0, | |
| "eval_steps": 500, | |
| "global_step": 19888, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11074197120708748, | |
| "grad_norm": 25.114500045776367, | |
| "learning_rate": 6.5044247787610615e-06, | |
| "loss": 24.7234, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22148394241417496, | |
| "grad_norm": 18.712068557739258, | |
| "learning_rate": 1.3141592920353982e-05, | |
| "loss": 24.4168, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.33222591362126247, | |
| "grad_norm": 27.625778198242188, | |
| "learning_rate": 1.9778761061946902e-05, | |
| "loss": 24.0671, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4429678848283499, | |
| "grad_norm": 21.45062828063965, | |
| "learning_rate": 2.641592920353982e-05, | |
| "loss": 23.263, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5537098560354374, | |
| "grad_norm": 38.891578674316406, | |
| "learning_rate": 3.305309734513274e-05, | |
| "loss": 21.274, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.6644518272425249, | |
| "grad_norm": 15.521417617797852, | |
| "learning_rate": 3.9690265486725656e-05, | |
| "loss": 9.7142, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7751937984496124, | |
| "grad_norm": 2.311408758163452, | |
| "learning_rate": 4.632743362831858e-05, | |
| "loss": 3.6891, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.8859357696566998, | |
| "grad_norm": 0.999423623085022, | |
| "learning_rate": 5.29646017699115e-05, | |
| "loss": 3.2125, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9966777408637874, | |
| "grad_norm": 0.38470658659935, | |
| "learning_rate": 5.960176991150442e-05, | |
| "loss": 3.1497, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_cer": 1.0, | |
| "eval_loss": 3.15576434135437, | |
| "eval_runtime": 483.1807, | |
| "eval_samples_per_second": 21.071, | |
| "eval_steps_per_second": 2.635, | |
| "eval_wer": 1.0, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.106312292358804, | |
| "grad_norm": 0.8518805503845215, | |
| "learning_rate": 6.623893805309734e-05, | |
| "loss": 3.1317, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.2170542635658914, | |
| "grad_norm": 0.38256514072418213, | |
| "learning_rate": 7.287610619469026e-05, | |
| "loss": 3.1353, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.327796234772979, | |
| "grad_norm": 1.8106805086135864, | |
| "learning_rate": 7.951327433628318e-05, | |
| "loss": 3.1262, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.4385382059800664, | |
| "grad_norm": 0.5565470457077026, | |
| "learning_rate": 8.61504424778761e-05, | |
| "loss": 3.1362, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.5492801771871538, | |
| "grad_norm": 0.8854584097862244, | |
| "learning_rate": 9.278761061946901e-05, | |
| "loss": 3.1189, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.6600221483942414, | |
| "grad_norm": 0.43876901268959045, | |
| "learning_rate": 9.942477876106195e-05, | |
| "loss": 3.1301, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.770764119601329, | |
| "grad_norm": 0.6520148515701294, | |
| "learning_rate": 0.00010606194690265486, | |
| "loss": 3.1249, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.8815060908084162, | |
| "grad_norm": 0.9588941931724548, | |
| "learning_rate": 0.00011269911504424777, | |
| "loss": 3.1179, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.9922480620155039, | |
| "grad_norm": 0.9581903219223022, | |
| "learning_rate": 0.0001193362831858407, | |
| "loss": 3.1061, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_cer": 1.0, | |
| "eval_loss": 3.11116886138916, | |
| "eval_runtime": 481.1502, | |
| "eval_samples_per_second": 21.16, | |
| "eval_steps_per_second": 2.646, | |
| "eval_wer": 1.0, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.1018826135105204, | |
| "grad_norm": 1.0075995922088623, | |
| "learning_rate": 0.0001259734513274336, | |
| "loss": 3.0901, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.212624584717608, | |
| "grad_norm": 1.4122037887573242, | |
| "learning_rate": 0.00013261061946902652, | |
| "loss": 3.0824, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.3233665559246957, | |
| "grad_norm": 0.37052252888679504, | |
| "learning_rate": 0.00013924778761061947, | |
| "loss": 3.0478, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.434108527131783, | |
| "grad_norm": 0.9149488210678101, | |
| "learning_rate": 0.00014588495575221239, | |
| "loss": 3.0426, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.5448504983388704, | |
| "grad_norm": 0.6695976853370667, | |
| "learning_rate": 0.0001525221238938053, | |
| "loss": 3.0352, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.655592469545958, | |
| "grad_norm": 0.28660523891448975, | |
| "learning_rate": 0.0001591592920353982, | |
| "loss": 3.0372, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.7663344407530452, | |
| "grad_norm": 2.475407838821411, | |
| "learning_rate": 0.00016579646017699114, | |
| "loss": 3.0092, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.877076411960133, | |
| "grad_norm": 0.40837517380714417, | |
| "learning_rate": 0.00017243362831858406, | |
| "loss": 2.9405, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.9878183831672205, | |
| "grad_norm": 0.7240709066390991, | |
| "learning_rate": 0.000179070796460177, | |
| "loss": 2.8017, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_cer": 1.0, | |
| "eval_loss": 2.5643131732940674, | |
| "eval_runtime": 484.9725, | |
| "eval_samples_per_second": 20.993, | |
| "eval_steps_per_second": 2.625, | |
| "eval_wer": 1.0, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 3.097452934662237, | |
| "grad_norm": 1.4658317565917969, | |
| "learning_rate": 0.0001857079646017699, | |
| "loss": 2.5261, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.2081949058693247, | |
| "grad_norm": 0.7111493945121765, | |
| "learning_rate": 0.0001923451327433628, | |
| "loss": 2.115, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.318936877076412, | |
| "grad_norm": 1.02842378616333, | |
| "learning_rate": 0.00019898230088495576, | |
| "loss": 1.7595, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.4296788482834994, | |
| "grad_norm": 1.2684359550476074, | |
| "learning_rate": 0.00020561946902654865, | |
| "loss": 1.4518, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.540420819490587, | |
| "grad_norm": 1.0996291637420654, | |
| "learning_rate": 0.00021225663716814157, | |
| "loss": 1.3586, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.6511627906976747, | |
| "grad_norm": 0.8183071613311768, | |
| "learning_rate": 0.0002188938053097345, | |
| "loss": 1.2308, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.761904761904762, | |
| "grad_norm": 0.8036144971847534, | |
| "learning_rate": 0.0002255309734513274, | |
| "loss": 1.1334, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.8726467331118495, | |
| "grad_norm": 1.1055711507797241, | |
| "learning_rate": 0.00023216814159292032, | |
| "loss": 1.0593, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.983388704318937, | |
| "grad_norm": 0.8789339065551758, | |
| "learning_rate": 0.00023880530973451327, | |
| "loss": 1.0541, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_cer": 0.1802003031379858, | |
| "eval_loss": 0.645916223526001, | |
| "eval_runtime": 487.2819, | |
| "eval_samples_per_second": 20.893, | |
| "eval_steps_per_second": 2.612, | |
| "eval_wer": 0.6615370688874808, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 4.093023255813954, | |
| "grad_norm": 1.180145025253296, | |
| "learning_rate": 0.00024544247787610616, | |
| "loss": 1.01, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.203765227021041, | |
| "grad_norm": 1.0302808284759521, | |
| "learning_rate": 0.0002520796460176991, | |
| "loss": 0.9623, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.314507198228128, | |
| "grad_norm": 0.7714611291885376, | |
| "learning_rate": 0.000258716814159292, | |
| "loss": 0.9025, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.425249169435216, | |
| "grad_norm": 0.6419712901115417, | |
| "learning_rate": 0.00026535398230088497, | |
| "loss": 0.8814, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.535991140642303, | |
| "grad_norm": 0.983061671257019, | |
| "learning_rate": 0.00027199115044247783, | |
| "loss": 0.8831, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.646733111849391, | |
| "grad_norm": 0.7041178345680237, | |
| "learning_rate": 0.0002786283185840708, | |
| "loss": 0.901, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.7574750830564785, | |
| "grad_norm": 0.8451335430145264, | |
| "learning_rate": 0.0002852654867256637, | |
| "loss": 0.8312, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.868217054263566, | |
| "grad_norm": 0.9010829329490662, | |
| "learning_rate": 0.0002919026548672566, | |
| "loss": 0.8204, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.978959025470654, | |
| "grad_norm": 0.7499646544456482, | |
| "learning_rate": 0.00029853982300884956, | |
| "loss": 0.8032, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_cer": 0.13106729228158473, | |
| "eval_loss": 0.4477018117904663, | |
| "eval_runtime": 486.3853, | |
| "eval_samples_per_second": 20.932, | |
| "eval_steps_per_second": 2.617, | |
| "eval_wer": 0.4958993391871096, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 5.08859357696567, | |
| "grad_norm": 0.6447732448577881, | |
| "learning_rate": 0.00029942477876106194, | |
| "loss": 0.8203, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.1993355481727574, | |
| "grad_norm": 0.6052475571632385, | |
| "learning_rate": 0.00029868731563421826, | |
| "loss": 0.7706, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 5.310077519379845, | |
| "grad_norm": 0.812167227268219, | |
| "learning_rate": 0.00029794985250737463, | |
| "loss": 0.7426, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 5.420819490586933, | |
| "grad_norm": 0.7104111909866333, | |
| "learning_rate": 0.00029721238938053095, | |
| "loss": 0.726, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 5.53156146179402, | |
| "grad_norm": 0.8796670436859131, | |
| "learning_rate": 0.0002964749262536873, | |
| "loss": 0.7275, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.642303433001107, | |
| "grad_norm": 0.7545107007026672, | |
| "learning_rate": 0.00029573746312684364, | |
| "loss": 0.737, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.753045404208195, | |
| "grad_norm": 0.6987699866294861, | |
| "learning_rate": 0.00029499999999999996, | |
| "loss": 0.7179, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.863787375415282, | |
| "grad_norm": 0.7108384966850281, | |
| "learning_rate": 0.00029426253687315633, | |
| "loss": 0.6831, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.97452934662237, | |
| "grad_norm": 0.5486265420913696, | |
| "learning_rate": 0.00029352507374631265, | |
| "loss": 0.6855, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_cer": 0.1090644059438252, | |
| "eval_loss": 0.3938077688217163, | |
| "eval_runtime": 486.997, | |
| "eval_samples_per_second": 20.906, | |
| "eval_steps_per_second": 2.614, | |
| "eval_wer": 0.41743459762831536, | |
| "step": 2712 | |
| }, | |
| { | |
| "epoch": 6.084163898117387, | |
| "grad_norm": 0.6461381316184998, | |
| "learning_rate": 0.000292787610619469, | |
| "loss": 0.6692, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 6.194905869324474, | |
| "grad_norm": 0.6610192060470581, | |
| "learning_rate": 0.00029205014749262534, | |
| "loss": 0.6749, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 6.305647840531561, | |
| "grad_norm": 0.6090611815452576, | |
| "learning_rate": 0.0002913126843657817, | |
| "loss": 0.6681, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 6.416389811738649, | |
| "grad_norm": 0.7427356243133545, | |
| "learning_rate": 0.00029057522123893803, | |
| "loss": 0.6565, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 6.5271317829457365, | |
| "grad_norm": 0.5656064748764038, | |
| "learning_rate": 0.00028983775811209435, | |
| "loss": 0.7006, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 6.637873754152824, | |
| "grad_norm": 0.661216676235199, | |
| "learning_rate": 0.0002891002949852507, | |
| "loss": 0.6375, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.748615725359912, | |
| "grad_norm": 0.6545315384864807, | |
| "learning_rate": 0.00028836283185840704, | |
| "loss": 0.6258, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 6.859357696566999, | |
| "grad_norm": 0.7606847882270813, | |
| "learning_rate": 0.0002876253687315634, | |
| "loss": 0.6252, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.970099667774086, | |
| "grad_norm": 0.7205169200897217, | |
| "learning_rate": 0.00028688790560471973, | |
| "loss": 0.6275, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_cer": 0.09786352988955044, | |
| "eval_loss": 0.34525957703590393, | |
| "eval_runtime": 489.2546, | |
| "eval_samples_per_second": 20.809, | |
| "eval_steps_per_second": 2.602, | |
| "eval_wer": 0.37516067710690687, | |
| "step": 3164 | |
| }, | |
| { | |
| "epoch": 7.079734219269103, | |
| "grad_norm": 0.7388074398040771, | |
| "learning_rate": 0.00028615044247787605, | |
| "loss": 0.6025, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 7.190476190476191, | |
| "grad_norm": 0.74979567527771, | |
| "learning_rate": 0.0002854129793510324, | |
| "loss": 0.6064, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 7.301218161683278, | |
| "grad_norm": 0.7727375626564026, | |
| "learning_rate": 0.00028467551622418874, | |
| "loss": 0.6019, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 7.411960132890365, | |
| "grad_norm": 0.6809210777282715, | |
| "learning_rate": 0.0002839380530973451, | |
| "loss": 0.6096, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 7.522702104097453, | |
| "grad_norm": 0.9523106813430786, | |
| "learning_rate": 0.00028320058997050143, | |
| "loss": 0.6002, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 7.63344407530454, | |
| "grad_norm": 0.7200210690498352, | |
| "learning_rate": 0.0002824631268436578, | |
| "loss": 0.577, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 7.7441860465116275, | |
| "grad_norm": 0.5474680662155151, | |
| "learning_rate": 0.0002817256637168141, | |
| "loss": 0.5672, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.8549280177187155, | |
| "grad_norm": 0.868034839630127, | |
| "learning_rate": 0.0002809882005899705, | |
| "loss": 0.571, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 7.965669988925803, | |
| "grad_norm": 0.5585504770278931, | |
| "learning_rate": 0.0002802507374631268, | |
| "loss": 0.6264, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_cer": 0.09037935760854587, | |
| "eval_loss": 0.3181062638759613, | |
| "eval_runtime": 489.6095, | |
| "eval_samples_per_second": 20.794, | |
| "eval_steps_per_second": 2.6, | |
| "eval_wer": 0.35024893636281346, | |
| "step": 3616 | |
| }, | |
| { | |
| "epoch": 8.07530454042082, | |
| "grad_norm": 0.6232703328132629, | |
| "learning_rate": 0.0002795132743362832, | |
| "loss": 0.5612, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 8.186046511627907, | |
| "grad_norm": 0.5712502002716064, | |
| "learning_rate": 0.0002787758112094395, | |
| "loss": 0.5539, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 8.296788482834994, | |
| "grad_norm": 0.6545235514640808, | |
| "learning_rate": 0.0002780383480825958, | |
| "loss": 0.6073, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 8.407530454042082, | |
| "grad_norm": 0.7191429138183594, | |
| "learning_rate": 0.0002773008849557522, | |
| "loss": 0.5501, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 8.518272425249169, | |
| "grad_norm": 0.6950603127479553, | |
| "learning_rate": 0.0002765634218289085, | |
| "loss": 0.5374, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 8.629014396456256, | |
| "grad_norm": 0.6837615966796875, | |
| "learning_rate": 0.0002758259587020649, | |
| "loss": 0.5679, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 8.739756367663345, | |
| "grad_norm": 0.6940374970436096, | |
| "learning_rate": 0.0002750884955752212, | |
| "loss": 0.558, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 8.850498338870432, | |
| "grad_norm": 0.643427848815918, | |
| "learning_rate": 0.0002743510324483776, | |
| "loss": 0.5241, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.96124031007752, | |
| "grad_norm": 0.6302089095115662, | |
| "learning_rate": 0.0002736135693215339, | |
| "loss": 0.5388, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_cer": 0.08582140966415502, | |
| "eval_loss": 0.30813154578208923, | |
| "eval_runtime": 488.7207, | |
| "eval_samples_per_second": 20.832, | |
| "eval_steps_per_second": 2.605, | |
| "eval_wer": 0.33156513080474337, | |
| "step": 4068 | |
| }, | |
| { | |
| "epoch": 9.070874861572536, | |
| "grad_norm": 0.6800464987754822, | |
| "learning_rate": 0.00027287610619469027, | |
| "loss": 0.5446, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 9.181616832779623, | |
| "grad_norm": 0.5239233374595642, | |
| "learning_rate": 0.0002721386430678466, | |
| "loss": 0.4957, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 9.29235880398671, | |
| "grad_norm": 0.6076834201812744, | |
| "learning_rate": 0.0002714011799410029, | |
| "loss": 0.5316, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 9.4031007751938, | |
| "grad_norm": 0.8120834827423096, | |
| "learning_rate": 0.0002706637168141593, | |
| "loss": 0.5766, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 9.513842746400886, | |
| "grad_norm": 0.6490704417228699, | |
| "learning_rate": 0.0002699262536873156, | |
| "loss": 0.5044, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 9.624584717607974, | |
| "grad_norm": 0.6406815648078918, | |
| "learning_rate": 0.00026918879056047197, | |
| "loss": 0.511, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 9.73532668881506, | |
| "grad_norm": 0.5792708992958069, | |
| "learning_rate": 0.0002684513274336283, | |
| "loss": 0.5195, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 9.846068660022148, | |
| "grad_norm": 0.7325103878974915, | |
| "learning_rate": 0.0002677138643067846, | |
| "loss": 0.5012, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 9.956810631229235, | |
| "grad_norm": 0.7060527205467224, | |
| "learning_rate": 0.000266976401179941, | |
| "loss": 0.5194, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_cer": 0.08035332255187068, | |
| "eval_loss": 0.28676968812942505, | |
| "eval_runtime": 487.2322, | |
| "eval_samples_per_second": 20.896, | |
| "eval_steps_per_second": 2.613, | |
| "eval_wer": 0.3129718475604236, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 10.066445182724253, | |
| "grad_norm": 0.6046482920646667, | |
| "learning_rate": 0.0002662389380530973, | |
| "loss": 0.489, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 10.17718715393134, | |
| "grad_norm": 0.7882891893386841, | |
| "learning_rate": 0.00026550147492625367, | |
| "loss": 0.4864, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 10.287929125138428, | |
| "grad_norm": 0.8176812529563904, | |
| "learning_rate": 0.00026476401179941, | |
| "loss": 0.4962, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 10.398671096345515, | |
| "grad_norm": 0.5068700313568115, | |
| "learning_rate": 0.00026402654867256636, | |
| "loss": 0.4777, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 10.509413067552602, | |
| "grad_norm": 0.5760357975959778, | |
| "learning_rate": 0.0002632890855457227, | |
| "loss": 0.4758, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 10.62015503875969, | |
| "grad_norm": 0.704738974571228, | |
| "learning_rate": 0.00026255162241887905, | |
| "loss": 0.4889, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 10.730897009966778, | |
| "grad_norm": 0.5809513926506042, | |
| "learning_rate": 0.00026181415929203537, | |
| "loss": 0.4811, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 10.841638981173865, | |
| "grad_norm": 0.5770090222358704, | |
| "learning_rate": 0.0002610766961651917, | |
| "loss": 0.515, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 10.952380952380953, | |
| "grad_norm": 0.5710371136665344, | |
| "learning_rate": 0.00026033923303834806, | |
| "loss": 0.4798, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_cer": 0.07714426612324227, | |
| "eval_loss": 0.27717578411102295, | |
| "eval_runtime": 489.1751, | |
| "eval_samples_per_second": 20.813, | |
| "eval_steps_per_second": 2.602, | |
| "eval_wer": 0.2999728433058749, | |
| "step": 4972 | |
| }, | |
| { | |
| "epoch": 11.062015503875969, | |
| "grad_norm": 0.6379249691963196, | |
| "learning_rate": 0.0002596017699115044, | |
| "loss": 0.5031, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 11.172757475083056, | |
| "grad_norm": 0.4557214677333832, | |
| "learning_rate": 0.00025886430678466075, | |
| "loss": 0.4546, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 11.283499446290143, | |
| "grad_norm": 0.9845902919769287, | |
| "learning_rate": 0.00025812684365781707, | |
| "loss": 0.4494, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 11.39424141749723, | |
| "grad_norm": 0.5986599922180176, | |
| "learning_rate": 0.00025738938053097344, | |
| "loss": 0.466, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 11.50498338870432, | |
| "grad_norm": 0.5532345771789551, | |
| "learning_rate": 0.00025665191740412976, | |
| "loss": 0.4696, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 11.615725359911407, | |
| "grad_norm": 0.5763828754425049, | |
| "learning_rate": 0.00025591445427728613, | |
| "loss": 0.4423, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 11.726467331118494, | |
| "grad_norm": 0.5517535209655762, | |
| "learning_rate": 0.00025517699115044245, | |
| "loss": 0.4344, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 11.837209302325581, | |
| "grad_norm": 0.6629925966262817, | |
| "learning_rate": 0.0002544395280235988, | |
| "loss": 0.452, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 11.947951273532668, | |
| "grad_norm": 0.5328862071037292, | |
| "learning_rate": 0.00025370206489675514, | |
| "loss": 0.4523, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_cer": 0.07210042714897999, | |
| "eval_loss": 0.26620689034461975, | |
| "eval_runtime": 489.7467, | |
| "eval_samples_per_second": 20.788, | |
| "eval_steps_per_second": 2.599, | |
| "eval_wer": 0.2804562324613017, | |
| "step": 5424 | |
| }, | |
| { | |
| "epoch": 12.057585825027685, | |
| "grad_norm": 0.5896037220954895, | |
| "learning_rate": 0.0002529646017699115, | |
| "loss": 0.4961, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 12.168327796234774, | |
| "grad_norm": 0.6215909123420715, | |
| "learning_rate": 0.00025222713864306783, | |
| "loss": 0.4503, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 12.279069767441861, | |
| "grad_norm": 0.7767740488052368, | |
| "learning_rate": 0.00025148967551622415, | |
| "loss": 0.4421, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 12.389811738648948, | |
| "grad_norm": 0.7437320351600647, | |
| "learning_rate": 0.0002507522123893805, | |
| "loss": 0.4316, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 12.500553709856035, | |
| "grad_norm": 0.8144622445106506, | |
| "learning_rate": 0.00025001474926253684, | |
| "loss": 0.4599, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 12.611295681063122, | |
| "grad_norm": 0.5992530584335327, | |
| "learning_rate": 0.00024927728613569316, | |
| "loss": 0.4339, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 12.72203765227021, | |
| "grad_norm": 0.7761603593826294, | |
| "learning_rate": 0.00024853982300884953, | |
| "loss": 0.4342, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 12.832779623477299, | |
| "grad_norm": 0.5413389205932617, | |
| "learning_rate": 0.00024780235988200585, | |
| "loss": 0.4442, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 12.943521594684386, | |
| "grad_norm": 0.5843682885169983, | |
| "learning_rate": 0.0002470648967551622, | |
| "loss": 0.4424, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_cer": 0.07174870006019247, | |
| "eval_loss": 0.26166489720344543, | |
| "eval_runtime": 492.0407, | |
| "eval_samples_per_second": 20.691, | |
| "eval_steps_per_second": 2.587, | |
| "eval_wer": 0.27929754684529734, | |
| "step": 5876 | |
| }, | |
| { | |
| "epoch": 13.053156146179402, | |
| "grad_norm": 0.9268521070480347, | |
| "learning_rate": 0.00024632743362831854, | |
| "loss": 0.4322, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 13.16389811738649, | |
| "grad_norm": 0.5855753421783447, | |
| "learning_rate": 0.0002455899705014749, | |
| "loss": 0.4314, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 13.274640088593577, | |
| "grad_norm": 0.710689127445221, | |
| "learning_rate": 0.00024485250737463123, | |
| "loss": 0.4407, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 13.385382059800664, | |
| "grad_norm": 0.6203360557556152, | |
| "learning_rate": 0.0002441150442477876, | |
| "loss": 0.4394, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 13.496124031007753, | |
| "grad_norm": 0.6015070676803589, | |
| "learning_rate": 0.00024337758112094392, | |
| "loss": 0.4359, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 13.60686600221484, | |
| "grad_norm": 0.7456609606742859, | |
| "learning_rate": 0.00024264011799410027, | |
| "loss": 0.477, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 13.717607973421927, | |
| "grad_norm": 0.6540862917900085, | |
| "learning_rate": 0.00024190265486725662, | |
| "loss": 0.4415, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 13.828349944629014, | |
| "grad_norm": 0.4988921880722046, | |
| "learning_rate": 0.00024116519174041293, | |
| "loss": 0.4131, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 13.939091915836102, | |
| "grad_norm": 0.7218478322029114, | |
| "learning_rate": 0.0002404277286135693, | |
| "loss": 0.4179, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_cer": 0.06874995467434424, | |
| "eval_loss": 0.2501542270183563, | |
| "eval_runtime": 495.114, | |
| "eval_samples_per_second": 20.563, | |
| "eval_steps_per_second": 2.571, | |
| "eval_wer": 0.2687245405992577, | |
| "step": 6328 | |
| }, | |
| { | |
| "epoch": 14.048726467331118, | |
| "grad_norm": 0.5830133557319641, | |
| "learning_rate": 0.00023969026548672563, | |
| "loss": 0.4341, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 14.159468438538205, | |
| "grad_norm": 0.762266993522644, | |
| "learning_rate": 0.000238952802359882, | |
| "loss": 0.4129, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 14.270210409745294, | |
| "grad_norm": 0.49856090545654297, | |
| "learning_rate": 0.00023821533923303832, | |
| "loss": 0.4356, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 14.380952380952381, | |
| "grad_norm": 0.6130903959274292, | |
| "learning_rate": 0.0002374778761061947, | |
| "loss": 0.4012, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 14.491694352159469, | |
| "grad_norm": 0.8215532898902893, | |
| "learning_rate": 0.000236740412979351, | |
| "loss": 0.4061, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 14.602436323366556, | |
| "grad_norm": 0.7206094861030579, | |
| "learning_rate": 0.00023600294985250735, | |
| "loss": 0.4021, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 14.713178294573643, | |
| "grad_norm": 0.6363425254821777, | |
| "learning_rate": 0.0002352654867256637, | |
| "loss": 0.412, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 14.82392026578073, | |
| "grad_norm": 0.5326490998268127, | |
| "learning_rate": 0.00023452802359882004, | |
| "loss": 0.3978, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 14.934662236987819, | |
| "grad_norm": 0.554918646812439, | |
| "learning_rate": 0.00023379056047197636, | |
| "loss": 0.415, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_cer": 0.06736842868642623, | |
| "eval_loss": 0.2500366270542145, | |
| "eval_runtime": 493.6452, | |
| "eval_samples_per_second": 20.624, | |
| "eval_steps_per_second": 2.579, | |
| "eval_wer": 0.2614465465737304, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 15.044296788482836, | |
| "grad_norm": 0.6449525356292725, | |
| "learning_rate": 0.00023305309734513273, | |
| "loss": 0.4243, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 15.155038759689923, | |
| "grad_norm": 1.446018099784851, | |
| "learning_rate": 0.00023231563421828905, | |
| "loss": 0.4122, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 15.26578073089701, | |
| "grad_norm": 0.5363207459449768, | |
| "learning_rate": 0.00023157817109144542, | |
| "loss": 0.4125, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 15.376522702104097, | |
| "grad_norm": 0.6152534484863281, | |
| "learning_rate": 0.00023084070796460174, | |
| "loss": 0.3998, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 15.487264673311184, | |
| "grad_norm": 0.592438817024231, | |
| "learning_rate": 0.00023010324483775812, | |
| "loss": 0.3853, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 15.598006644518273, | |
| "grad_norm": 0.6657503843307495, | |
| "learning_rate": 0.00022936578171091443, | |
| "loss": 0.3951, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 15.70874861572536, | |
| "grad_norm": 0.5211730599403381, | |
| "learning_rate": 0.00022862831858407078, | |
| "loss": 0.4023, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 15.819490586932448, | |
| "grad_norm": 0.5348416566848755, | |
| "learning_rate": 0.00022789085545722713, | |
| "loss": 0.4093, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 15.930232558139535, | |
| "grad_norm": 0.927044689655304, | |
| "learning_rate": 0.00022715339233038347, | |
| "loss": 0.4405, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_cer": 0.06604491953789587, | |
| "eval_loss": 0.2411191165447235, | |
| "eval_runtime": 494.0285, | |
| "eval_samples_per_second": 20.608, | |
| "eval_steps_per_second": 2.577, | |
| "eval_wer": 0.2560876255997103, | |
| "step": 7232 | |
| }, | |
| { | |
| "epoch": 16.039867109634553, | |
| "grad_norm": 0.5035960674285889, | |
| "learning_rate": 0.00022641592920353982, | |
| "loss": 0.4186, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 16.15060908084164, | |
| "grad_norm": 0.7938876748085022, | |
| "learning_rate": 0.00022567846607669613, | |
| "loss": 0.4016, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 16.261351052048727, | |
| "grad_norm": 0.6998234987258911, | |
| "learning_rate": 0.00022494100294985248, | |
| "loss": 0.3891, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 16.372093023255815, | |
| "grad_norm": 0.5634494423866272, | |
| "learning_rate": 0.00022420353982300883, | |
| "loss": 0.3768, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 16.482834994462902, | |
| "grad_norm": 0.6078976392745972, | |
| "learning_rate": 0.00022346607669616517, | |
| "loss": 0.3974, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 16.59357696566999, | |
| "grad_norm": 1.2107906341552734, | |
| "learning_rate": 0.0002227286135693215, | |
| "loss": 0.3791, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 16.704318936877076, | |
| "grad_norm": 0.7613789439201355, | |
| "learning_rate": 0.00022199115044247786, | |
| "loss": 0.3713, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 16.815060908084163, | |
| "grad_norm": 0.5816598534584045, | |
| "learning_rate": 0.00022125368731563418, | |
| "loss": 0.3817, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 16.92580287929125, | |
| "grad_norm": 0.8725369572639465, | |
| "learning_rate": 0.00022051622418879055, | |
| "loss": 0.3812, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_cer": 0.06427903198903481, | |
| "eval_loss": 0.23461875319480896, | |
| "eval_runtime": 492.7771, | |
| "eval_samples_per_second": 20.66, | |
| "eval_steps_per_second": 2.583, | |
| "eval_wer": 0.2501674662804381, | |
| "step": 7684 | |
| }, | |
| { | |
| "epoch": 17.035437430786267, | |
| "grad_norm": 0.5457949638366699, | |
| "learning_rate": 0.00021977876106194687, | |
| "loss": 0.389, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 17.146179401993354, | |
| "grad_norm": 0.6383467316627502, | |
| "learning_rate": 0.00021904129793510324, | |
| "loss": 0.3842, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 17.25692137320044, | |
| "grad_norm": 4.207304954528809, | |
| "learning_rate": 0.00021830383480825956, | |
| "loss": 0.3459, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 17.367663344407532, | |
| "grad_norm": 1.1675844192504883, | |
| "learning_rate": 0.0002175663716814159, | |
| "loss": 0.3825, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 17.47840531561462, | |
| "grad_norm": 0.44133588671684265, | |
| "learning_rate": 0.00021682890855457225, | |
| "loss": 0.3911, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 17.589147286821706, | |
| "grad_norm": 0.5262250304222107, | |
| "learning_rate": 0.0002160914454277286, | |
| "loss": 0.383, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 17.699889258028794, | |
| "grad_norm": 0.7032670378684998, | |
| "learning_rate": 0.00021535398230088494, | |
| "loss": 0.4, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 17.81063122923588, | |
| "grad_norm": 0.48254212737083435, | |
| "learning_rate": 0.0002146165191740413, | |
| "loss": 0.3697, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 17.921373200442968, | |
| "grad_norm": 0.629219651222229, | |
| "learning_rate": 0.0002138790560471976, | |
| "loss": 0.3752, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_cer": 0.06295189678804274, | |
| "eval_loss": 0.23475028574466705, | |
| "eval_runtime": 494.2681, | |
| "eval_samples_per_second": 20.598, | |
| "eval_steps_per_second": 2.576, | |
| "eval_wer": 0.2447180229926677, | |
| "step": 8136 | |
| }, | |
| { | |
| "epoch": 18.031007751937985, | |
| "grad_norm": 0.5522592663764954, | |
| "learning_rate": 0.00021314159292035398, | |
| "loss": 0.4118, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 18.14174972314507, | |
| "grad_norm": 0.6355589628219604, | |
| "learning_rate": 0.0002124041297935103, | |
| "loss": 0.3613, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 18.25249169435216, | |
| "grad_norm": 0.6213610172271729, | |
| "learning_rate": 0.00021166666666666667, | |
| "loss": 0.3561, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 18.363233665559246, | |
| "grad_norm": 0.6862598657608032, | |
| "learning_rate": 0.000210929203539823, | |
| "loss": 0.3822, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 18.473975636766333, | |
| "grad_norm": 0.4650695323944092, | |
| "learning_rate": 0.00021019174041297936, | |
| "loss": 0.3708, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 18.58471760797342, | |
| "grad_norm": 0.6250514388084412, | |
| "learning_rate": 0.00020945427728613568, | |
| "loss": 0.371, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 18.695459579180508, | |
| "grad_norm": 1.5113970041275024, | |
| "learning_rate": 0.000208716814159292, | |
| "loss": 0.3526, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 18.8062015503876, | |
| "grad_norm": 0.5471574068069458, | |
| "learning_rate": 0.00020797935103244837, | |
| "loss": 0.3789, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 18.916943521594686, | |
| "grad_norm": 0.5146489143371582, | |
| "learning_rate": 0.0002072418879056047, | |
| "loss": 0.3565, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_cer": 0.06287574968634646, | |
| "eval_loss": 0.23113583028316498, | |
| "eval_runtime": 494.568, | |
| "eval_samples_per_second": 20.586, | |
| "eval_steps_per_second": 2.574, | |
| "eval_wer": 0.24508011224766904, | |
| "step": 8588 | |
| }, | |
| { | |
| "epoch": 19.026578073089702, | |
| "grad_norm": 0.9078727960586548, | |
| "learning_rate": 0.00020650442477876103, | |
| "loss": 0.3678, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 19.13732004429679, | |
| "grad_norm": 0.6155833601951599, | |
| "learning_rate": 0.00020576696165191738, | |
| "loss": 0.3483, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 19.248062015503876, | |
| "grad_norm": 0.4708555042743683, | |
| "learning_rate": 0.00020502949852507373, | |
| "loss": 0.3504, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 19.358803986710964, | |
| "grad_norm": 0.5760027170181274, | |
| "learning_rate": 0.00020429203539823007, | |
| "loss": 0.3521, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 19.46954595791805, | |
| "grad_norm": 0.5396759510040283, | |
| "learning_rate": 0.00020355457227138642, | |
| "loss": 0.3583, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 19.580287929125138, | |
| "grad_norm": 0.4109143912792206, | |
| "learning_rate": 0.00020281710914454273, | |
| "loss": 0.3421, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 19.691029900332225, | |
| "grad_norm": 0.5602762699127197, | |
| "learning_rate": 0.0002020796460176991, | |
| "loss": 0.3616, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 19.801771871539312, | |
| "grad_norm": 0.4505002200603485, | |
| "learning_rate": 0.00020134218289085543, | |
| "loss": 0.3561, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 19.9125138427464, | |
| "grad_norm": 0.6884456872940063, | |
| "learning_rate": 0.0002006047197640118, | |
| "loss": 0.3675, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_cer": 0.06083065609793242, | |
| "eval_loss": 0.2296416014432907, | |
| "eval_runtime": 495.6675, | |
| "eval_samples_per_second": 20.54, | |
| "eval_steps_per_second": 2.568, | |
| "eval_wer": 0.23590114963338463, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 20.022148394241416, | |
| "grad_norm": 0.706529974937439, | |
| "learning_rate": 0.00019986725663716812, | |
| "loss": 0.3583, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 20.132890365448507, | |
| "grad_norm": 1.8911956548690796, | |
| "learning_rate": 0.0001991297935103245, | |
| "loss": 0.3548, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 20.243632336655594, | |
| "grad_norm": NaN, | |
| "learning_rate": 0.0001983923303834808, | |
| "loss": 0.3652, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 20.35437430786268, | |
| "grad_norm": 0.5242050290107727, | |
| "learning_rate": 0.00019765486725663715, | |
| "loss": 0.3481, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 20.46511627906977, | |
| "grad_norm": 0.4361858665943146, | |
| "learning_rate": 0.0001969174041297935, | |
| "loss": 0.346, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 20.575858250276855, | |
| "grad_norm": 0.4760480523109436, | |
| "learning_rate": 0.00019617994100294984, | |
| "loss": 0.3394, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 20.686600221483943, | |
| "grad_norm": 0.5500379800796509, | |
| "learning_rate": 0.00019544247787610616, | |
| "loss": 0.3631, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 20.79734219269103, | |
| "grad_norm": 0.5537223815917969, | |
| "learning_rate": 0.00019470501474926253, | |
| "loss": 0.3517, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 20.908084163898117, | |
| "grad_norm": 1.2052114009857178, | |
| "learning_rate": 0.00019396755162241885, | |
| "loss": 0.3318, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_cer": 0.06032663480575237, | |
| "eval_loss": 0.2251521646976471, | |
| "eval_runtime": 497.982, | |
| "eval_samples_per_second": 20.445, | |
| "eval_steps_per_second": 2.556, | |
| "eval_wer": 0.23347515162487553, | |
| "step": 9492 | |
| }, | |
| { | |
| "epoch": 21.017718715393134, | |
| "grad_norm": 0.4967649281024933, | |
| "learning_rate": 0.00019323008849557523, | |
| "loss": 0.3693, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 21.12846068660022, | |
| "grad_norm": 0.40202295780181885, | |
| "learning_rate": 0.00019249262536873154, | |
| "loss": 0.3454, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 21.239202657807308, | |
| "grad_norm": 0.4794892370700836, | |
| "learning_rate": 0.00019175516224188792, | |
| "loss": 0.3286, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 21.349944629014395, | |
| "grad_norm": 0.5660750865936279, | |
| "learning_rate": 0.00019101769911504423, | |
| "loss": 0.3384, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 21.460686600221482, | |
| "grad_norm": 0.5895997881889343, | |
| "learning_rate": 0.00019028023598820055, | |
| "loss": 0.3634, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 21.571428571428573, | |
| "grad_norm": 0.33758455514907837, | |
| "learning_rate": 0.00018954277286135693, | |
| "loss": 0.3376, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 21.68217054263566, | |
| "grad_norm": 0.44639402627944946, | |
| "learning_rate": 0.00018880530973451324, | |
| "loss": 0.3532, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 21.792912513842747, | |
| "grad_norm": 0.4666491746902466, | |
| "learning_rate": 0.00018806784660766962, | |
| "loss": 0.3467, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 21.903654485049834, | |
| "grad_norm": 0.36474844813346863, | |
| "learning_rate": 0.00018733038348082593, | |
| "loss": 0.3479, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_cer": 0.059242445119695994, | |
| "eval_loss": 0.2179824411869049, | |
| "eval_runtime": 496.9044, | |
| "eval_samples_per_second": 20.489, | |
| "eval_steps_per_second": 2.562, | |
| "eval_wer": 0.23155607857336832, | |
| "step": 9944 | |
| }, | |
| { | |
| "epoch": 22.01328903654485, | |
| "grad_norm": 0.42274388670921326, | |
| "learning_rate": 0.00018659292035398228, | |
| "loss": 0.3619, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 22.124031007751938, | |
| "grad_norm": 0.6178750395774841, | |
| "learning_rate": 0.00018585545722713863, | |
| "loss": 0.3358, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 22.234772978959025, | |
| "grad_norm": 0.39765465259552, | |
| "learning_rate": 0.00018511799410029497, | |
| "loss": 0.3354, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 22.345514950166113, | |
| "grad_norm": 1.0347713232040405, | |
| "learning_rate": 0.0001843805309734513, | |
| "loss": 0.3266, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 22.4562569213732, | |
| "grad_norm": 0.4053027927875519, | |
| "learning_rate": 0.00018364306784660766, | |
| "loss": 0.3445, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 22.566998892580287, | |
| "grad_norm": 0.5446871519088745, | |
| "learning_rate": 0.00018290560471976398, | |
| "loss": 0.3585, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 22.677740863787374, | |
| "grad_norm": 0.6107918620109558, | |
| "learning_rate": 0.00018216814159292035, | |
| "loss": 0.3418, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 22.78848283499446, | |
| "grad_norm": 0.5900046825408936, | |
| "learning_rate": 0.00018143067846607667, | |
| "loss": 0.3483, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 22.899224806201552, | |
| "grad_norm": 0.42756062746047974, | |
| "learning_rate": 0.00018069321533923304, | |
| "loss": 0.33, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_cer": 0.05697253627865488, | |
| "eval_loss": 0.21119394898414612, | |
| "eval_runtime": 496.1257, | |
| "eval_samples_per_second": 20.521, | |
| "eval_steps_per_second": 2.566, | |
| "eval_wer": 0.22308319000633656, | |
| "step": 10396 | |
| }, | |
| { | |
| "epoch": 23.00885935769657, | |
| "grad_norm": 0.6858476996421814, | |
| "learning_rate": 0.00017995575221238936, | |
| "loss": 0.3312, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 23.119601328903656, | |
| "grad_norm": 0.5675090551376343, | |
| "learning_rate": 0.0001792182890855457, | |
| "loss": 0.3219, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 23.230343300110743, | |
| "grad_norm": 0.36263322830200195, | |
| "learning_rate": 0.00017848082595870205, | |
| "loss": 0.3325, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 23.34108527131783, | |
| "grad_norm": 0.8479202389717102, | |
| "learning_rate": 0.0001777433628318584, | |
| "loss": 0.3265, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 23.451827242524917, | |
| "grad_norm": 0.6954662799835205, | |
| "learning_rate": 0.00017700589970501474, | |
| "loss": 0.3578, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 23.562569213732004, | |
| "grad_norm": 0.5398178100585938, | |
| "learning_rate": 0.0001762684365781711, | |
| "loss": 0.3234, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 23.67331118493909, | |
| "grad_norm": 0.6818826794624329, | |
| "learning_rate": 0.0001755309734513274, | |
| "loss": 0.3287, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 23.78405315614618, | |
| "grad_norm": 0.6283251047134399, | |
| "learning_rate": 0.00017479351032448378, | |
| "loss": 0.339, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 23.894795127353266, | |
| "grad_norm": 0.4861604869365692, | |
| "learning_rate": 0.0001740560471976401, | |
| "loss": 0.3288, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_cer": 0.05748018362329666, | |
| "eval_loss": 0.2127908617258072, | |
| "eval_runtime": 499.441, | |
| "eval_samples_per_second": 20.385, | |
| "eval_steps_per_second": 2.549, | |
| "eval_wer": 0.22348148818683805, | |
| "step": 10848 | |
| }, | |
| { | |
| "epoch": 24.004429678848282, | |
| "grad_norm": 0.2947724461555481, | |
| "learning_rate": 0.00017331858407079642, | |
| "loss": 0.3282, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 24.11517165005537, | |
| "grad_norm": 0.4310826063156128, | |
| "learning_rate": 0.0001725811209439528, | |
| "loss": 0.316, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 24.225913621262457, | |
| "grad_norm": 0.8667187094688416, | |
| "learning_rate": 0.0001718436578171091, | |
| "loss": 0.3381, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 24.336655592469548, | |
| "grad_norm": 0.5773921012878418, | |
| "learning_rate": 0.00017110619469026548, | |
| "loss": 0.3154, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 24.447397563676635, | |
| "grad_norm": 0.3089664578437805, | |
| "learning_rate": 0.0001703687315634218, | |
| "loss": 0.3151, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 24.558139534883722, | |
| "grad_norm": 0.5007433891296387, | |
| "learning_rate": 0.00016963126843657817, | |
| "loss": 0.3229, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 24.66888150609081, | |
| "grad_norm": 2.2603423595428467, | |
| "learning_rate": 0.0001688938053097345, | |
| "loss": 0.3313, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 24.779623477297896, | |
| "grad_norm": 0.34253740310668945, | |
| "learning_rate": 0.00016815634218289084, | |
| "loss": 0.3175, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 24.890365448504983, | |
| "grad_norm": 0.4452781081199646, | |
| "learning_rate": 0.00016741887905604718, | |
| "loss": 0.3382, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "grad_norm": 2.1675078868865967, | |
| "learning_rate": 0.00016668141592920353, | |
| "loss": 0.3192, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_cer": 0.056867380757264795, | |
| "eval_loss": 0.21508556604385376, | |
| "eval_runtime": 501.0298, | |
| "eval_samples_per_second": 20.32, | |
| "eval_steps_per_second": 2.541, | |
| "eval_wer": 0.22150810174708066, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 25.110741971207087, | |
| "grad_norm": 1.4607517719268799, | |
| "learning_rate": 0.00016594395280235987, | |
| "loss": 0.3148, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 25.221483942414174, | |
| "grad_norm": 1.1707969903945923, | |
| "learning_rate": 0.00016520648967551622, | |
| "loss": 0.3314, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 25.33222591362126, | |
| "grad_norm": 0.954971969127655, | |
| "learning_rate": 0.00016446902654867254, | |
| "loss": 0.3206, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 25.44296788482835, | |
| "grad_norm": 0.9992377758026123, | |
| "learning_rate": 0.0001637315634218289, | |
| "loss": 0.3116, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 25.553709856035436, | |
| "grad_norm": 0.6818801760673523, | |
| "learning_rate": 0.00016299410029498523, | |
| "loss": 0.3104, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 25.664451827242527, | |
| "grad_norm": 0.8313333988189697, | |
| "learning_rate": 0.0001622566371681416, | |
| "loss": 0.3047, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 25.775193798449614, | |
| "grad_norm": 2.883293390274048, | |
| "learning_rate": 0.00016151917404129792, | |
| "loss": 0.3036, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 25.8859357696567, | |
| "grad_norm": 1.2632169723510742, | |
| "learning_rate": 0.00016078171091445426, | |
| "loss": 0.3331, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 25.996677740863788, | |
| "grad_norm": 0.8442880511283875, | |
| "learning_rate": 0.0001600442477876106, | |
| "loss": 0.3158, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_cer": 0.05608415342553176, | |
| "eval_loss": 0.21561041474342346, | |
| "eval_runtime": 500.4961, | |
| "eval_samples_per_second": 20.342, | |
| "eval_steps_per_second": 2.543, | |
| "eval_wer": 0.21783289580881687, | |
| "step": 11752 | |
| }, | |
| { | |
| "epoch": 26.106312292358805, | |
| "grad_norm": 1.1655757427215576, | |
| "learning_rate": 0.00015930678466076695, | |
| "loss": 0.3046, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 26.217054263565892, | |
| "grad_norm": 1.1917389631271362, | |
| "learning_rate": 0.0001585693215339233, | |
| "loss": 0.3056, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 26.32779623477298, | |
| "grad_norm": 0.7422528862953186, | |
| "learning_rate": 0.00015783185840707964, | |
| "loss": 0.3326, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 26.438538205980066, | |
| "grad_norm": 0.8917913436889648, | |
| "learning_rate": 0.00015709439528023596, | |
| "loss": 0.3091, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 26.549280177187153, | |
| "grad_norm": 1.8631386756896973, | |
| "learning_rate": 0.0001563569321533923, | |
| "loss": 0.2933, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 26.66002214839424, | |
| "grad_norm": 1.4756150245666504, | |
| "learning_rate": 0.00015561946902654865, | |
| "loss": 0.2987, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 26.770764119601328, | |
| "grad_norm": 0.9095970988273621, | |
| "learning_rate": 0.000154882005899705, | |
| "loss": 0.308, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 26.881506090808415, | |
| "grad_norm": 0.768785834312439, | |
| "learning_rate": 0.00015414454277286134, | |
| "loss": 0.3219, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 26.992248062015506, | |
| "grad_norm": 1.0584475994110107, | |
| "learning_rate": 0.00015340707964601766, | |
| "loss": 0.323, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_cer": 0.05621831736661566, | |
| "eval_loss": 0.21734623610973358, | |
| "eval_runtime": 502.1183, | |
| "eval_samples_per_second": 20.276, | |
| "eval_steps_per_second": 2.535, | |
| "eval_wer": 0.21875622340907033, | |
| "step": 12204 | |
| }, | |
| { | |
| "epoch": 27.101882613510522, | |
| "grad_norm": 0.8777967691421509, | |
| "learning_rate": 0.00015266961651917404, | |
| "loss": 0.3043, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 27.21262458471761, | |
| "grad_norm": 0.6613890528678894, | |
| "learning_rate": 0.00015193215339233035, | |
| "loss": 0.3055, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 27.323366555924697, | |
| "grad_norm": 1.175444483757019, | |
| "learning_rate": 0.00015119469026548673, | |
| "loss": 0.3027, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 27.434108527131784, | |
| "grad_norm": 0.9431079626083374, | |
| "learning_rate": 0.00015045722713864304, | |
| "loss": 0.2968, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 27.54485049833887, | |
| "grad_norm": 0.8549213409423828, | |
| "learning_rate": 0.0001497197640117994, | |
| "loss": 0.2959, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 27.655592469545958, | |
| "grad_norm": 2.223446846008301, | |
| "learning_rate": 0.00014898230088495574, | |
| "loss": 0.3125, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 27.766334440753045, | |
| "grad_norm": 2.218383312225342, | |
| "learning_rate": 0.00014824483775811208, | |
| "loss": 0.307, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 27.877076411960132, | |
| "grad_norm": 12.813185691833496, | |
| "learning_rate": 0.00014750737463126843, | |
| "loss": 0.3076, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 27.98781838316722, | |
| "grad_norm": 0.6163749694824219, | |
| "learning_rate": 0.00014676991150442477, | |
| "loss": 0.3328, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_cer": 0.054448803765292875, | |
| "eval_loss": 0.2115621566772461, | |
| "eval_runtime": 500.9948, | |
| "eval_samples_per_second": 20.322, | |
| "eval_steps_per_second": 2.541, | |
| "eval_wer": 0.21100751335204126, | |
| "step": 12656 | |
| }, | |
| { | |
| "epoch": 28.097452934662236, | |
| "grad_norm": 1.2707273960113525, | |
| "learning_rate": 0.0001460324483775811, | |
| "loss": 0.3012, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 28.208194905869323, | |
| "grad_norm": 1.3631172180175781, | |
| "learning_rate": 0.00014529498525073744, | |
| "loss": 0.2988, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 28.31893687707641, | |
| "grad_norm": 1.3375684022903442, | |
| "learning_rate": 0.00014455752212389378, | |
| "loss": 0.307, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 28.4296788482835, | |
| "grad_norm": 1.4623414278030396, | |
| "learning_rate": 0.00014382005899705013, | |
| "loss": 0.3189, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 28.54042081949059, | |
| "grad_norm": 0.9495463967323303, | |
| "learning_rate": 0.00014308259587020647, | |
| "loss": 0.2902, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 28.651162790697676, | |
| "grad_norm": 0.6960974931716919, | |
| "learning_rate": 0.00014234513274336282, | |
| "loss": 0.3025, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 28.761904761904763, | |
| "grad_norm": 1.6524468660354614, | |
| "learning_rate": 0.00014160766961651916, | |
| "loss": 0.3054, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 28.87264673311185, | |
| "grad_norm": 0.7738968729972839, | |
| "learning_rate": 0.0001408702064896755, | |
| "loss": 0.3023, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 28.983388704318937, | |
| "grad_norm": 1.094317078590393, | |
| "learning_rate": 0.00014013274336283185, | |
| "loss": 0.3171, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_cer": 0.054949199005011204, | |
| "eval_loss": 0.21789278090000153, | |
| "eval_runtime": 503.5992, | |
| "eval_samples_per_second": 20.216, | |
| "eval_steps_per_second": 2.528, | |
| "eval_wer": 0.21321625780754955, | |
| "step": 13108 | |
| }, | |
| { | |
| "epoch": 29.093023255813954, | |
| "grad_norm": 0.723978579044342, | |
| "learning_rate": 0.0001393952802359882, | |
| "loss": 0.2999, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 29.20376522702104, | |
| "grad_norm": 0.9239691495895386, | |
| "learning_rate": 0.00013865781710914454, | |
| "loss": 0.2868, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 29.314507198228128, | |
| "grad_norm": 0.812727689743042, | |
| "learning_rate": 0.00013792035398230086, | |
| "loss": 0.2994, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 29.425249169435215, | |
| "grad_norm": 0.6128324866294861, | |
| "learning_rate": 0.0001371828908554572, | |
| "loss": 0.2986, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 29.535991140642302, | |
| "grad_norm": 0.9569301009178162, | |
| "learning_rate": 0.00013644542772861355, | |
| "loss": 0.309, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 29.64673311184939, | |
| "grad_norm": 1.0407278537750244, | |
| "learning_rate": 0.0001357079646017699, | |
| "loss": 0.2961, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 29.757475083056477, | |
| "grad_norm": 1.2411153316497803, | |
| "learning_rate": 0.00013497050147492624, | |
| "loss": 0.2863, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 29.868217054263567, | |
| "grad_norm": 1.976034164428711, | |
| "learning_rate": 0.0001342330383480826, | |
| "loss": 0.3, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 29.978959025470655, | |
| "grad_norm": 0.6932374238967896, | |
| "learning_rate": 0.00013349557522123894, | |
| "loss": 0.3038, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_cer": 0.0540898245715819, | |
| "eval_loss": 0.21122422814369202, | |
| "eval_runtime": 502.738, | |
| "eval_samples_per_second": 20.251, | |
| "eval_steps_per_second": 2.532, | |
| "eval_wer": 0.21066352855978998, | |
| "step": 13560 | |
| }, | |
| { | |
| "epoch": 30.08859357696567, | |
| "grad_norm": 0.7096422910690308, | |
| "learning_rate": 0.00013275811209439528, | |
| "loss": 0.2918, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 30.19933554817276, | |
| "grad_norm": 1.9430056810379028, | |
| "learning_rate": 0.00013202064896755163, | |
| "loss": 0.2878, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 30.310077519379846, | |
| "grad_norm": 2.081293821334839, | |
| "learning_rate": 0.00013128318584070797, | |
| "loss": 0.2791, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 30.420819490586933, | |
| "grad_norm": 1.2402158975601196, | |
| "learning_rate": 0.0001305457227138643, | |
| "loss": 0.2833, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 30.53156146179402, | |
| "grad_norm": 0.9645398855209351, | |
| "learning_rate": 0.00012980825958702064, | |
| "loss": 0.2828, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 30.642303433001107, | |
| "grad_norm": 0.6930665969848633, | |
| "learning_rate": 0.00012907079646017698, | |
| "loss": 0.3018, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 30.753045404208194, | |
| "grad_norm": 0.9082468748092651, | |
| "learning_rate": 0.00012833333333333333, | |
| "loss": 0.3037, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 30.86378737541528, | |
| "grad_norm": 0.7977900505065918, | |
| "learning_rate": 0.00012759587020648967, | |
| "loss": 0.3122, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 30.97452934662237, | |
| "grad_norm": 0.7910751104354858, | |
| "learning_rate": 0.000126858407079646, | |
| "loss": 0.2927, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_cer": 0.053854131161569646, | |
| "eval_loss": 0.21427859365940094, | |
| "eval_runtime": 504.444, | |
| "eval_samples_per_second": 20.183, | |
| "eval_steps_per_second": 2.524, | |
| "eval_wer": 0.2105549017832896, | |
| "step": 14012 | |
| }, | |
| { | |
| "epoch": 31.084163898117385, | |
| "grad_norm": 1.2776386737823486, | |
| "learning_rate": 0.00012612094395280234, | |
| "loss": 0.2853, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 31.194905869324472, | |
| "grad_norm": 1.0792760848999023, | |
| "learning_rate": 0.00012538348082595868, | |
| "loss": 0.2994, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 31.305647840531563, | |
| "grad_norm": 1.4567831754684448, | |
| "learning_rate": 0.00012464601769911503, | |
| "loss": 0.3088, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 31.41638981173865, | |
| "grad_norm": 1.4427411556243896, | |
| "learning_rate": 0.00012390855457227137, | |
| "loss": 0.3066, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 31.527131782945737, | |
| "grad_norm": 0.8337031006813049, | |
| "learning_rate": 0.00012317109144542772, | |
| "loss": 0.2906, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 31.637873754152825, | |
| "grad_norm": 0.5533416271209717, | |
| "learning_rate": 0.00012243362831858406, | |
| "loss": 0.2885, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 31.74861572535991, | |
| "grad_norm": 1.1606074571609497, | |
| "learning_rate": 0.00012169616519174041, | |
| "loss": 0.2802, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 31.859357696567, | |
| "grad_norm": 0.7705062627792358, | |
| "learning_rate": 0.00012095870206489674, | |
| "loss": 0.2766, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 31.970099667774086, | |
| "grad_norm": 1.1826943159103394, | |
| "learning_rate": 0.00012022123893805309, | |
| "loss": 0.2866, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_cer": 0.05413333720112263, | |
| "eval_loss": 0.2068941295146942, | |
| "eval_runtime": 502.2625, | |
| "eval_samples_per_second": 20.27, | |
| "eval_steps_per_second": 2.535, | |
| "eval_wer": 0.21153254277179326, | |
| "step": 14464 | |
| }, | |
| { | |
| "epoch": 32.079734219269106, | |
| "grad_norm": 0.8447093963623047, | |
| "learning_rate": 0.00011948377581120943, | |
| "loss": 0.2951, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 32.19047619047619, | |
| "grad_norm": 1.153529167175293, | |
| "learning_rate": 0.00011874631268436578, | |
| "loss": 0.2881, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 32.30121816168328, | |
| "grad_norm": 0.9775529503822327, | |
| "learning_rate": 0.00011800884955752212, | |
| "loss": 0.277, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 32.411960132890364, | |
| "grad_norm": 1.1522144079208374, | |
| "learning_rate": 0.00011727138643067845, | |
| "loss": 0.2783, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 32.522702104097455, | |
| "grad_norm": 0.7744702696800232, | |
| "learning_rate": 0.0001165339233038348, | |
| "loss": 0.2844, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 32.63344407530454, | |
| "grad_norm": 1.2563271522521973, | |
| "learning_rate": 0.00011579646017699114, | |
| "loss": 0.2768, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 32.74418604651163, | |
| "grad_norm": 0.5680158734321594, | |
| "learning_rate": 0.00011505899705014749, | |
| "loss": 0.2723, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 32.85492801771871, | |
| "grad_norm": 1.74644136428833, | |
| "learning_rate": 0.00011432153392330384, | |
| "loss": 0.2922, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 32.965669988925804, | |
| "grad_norm": 0.6420773863792419, | |
| "learning_rate": 0.00011358407079646018, | |
| "loss": 0.2988, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_cer": 0.05228767649810358, | |
| "eval_loss": 0.20571845769882202, | |
| "eval_runtime": 505.3192, | |
| "eval_samples_per_second": 20.148, | |
| "eval_steps_per_second": 2.519, | |
| "eval_wer": 0.2027699828007604, | |
| "step": 14916 | |
| }, | |
| { | |
| "epoch": 33.07530454042082, | |
| "grad_norm": 1.2276347875595093, | |
| "learning_rate": 0.0001128466076696165, | |
| "loss": 0.2973, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 33.18604651162791, | |
| "grad_norm": 1.1782604455947876, | |
| "learning_rate": 0.00011210914454277285, | |
| "loss": 0.2867, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 33.296788482835, | |
| "grad_norm": 1.052660346031189, | |
| "learning_rate": 0.00011137168141592919, | |
| "loss": 0.2911, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 33.40753045404208, | |
| "grad_norm": 0.8341677188873291, | |
| "learning_rate": 0.00011063421828908554, | |
| "loss": 0.2822, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 33.51827242524917, | |
| "grad_norm": 0.6574891209602356, | |
| "learning_rate": 0.00010989675516224187, | |
| "loss": 0.2734, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 33.629014396456256, | |
| "grad_norm": 0.5809639096260071, | |
| "learning_rate": 0.00010915929203539821, | |
| "loss": 0.2897, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 33.73975636766335, | |
| "grad_norm": 0.6616008281707764, | |
| "learning_rate": 0.00010842182890855456, | |
| "loss": 0.2716, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 33.85049833887043, | |
| "grad_norm": 1.0783253908157349, | |
| "learning_rate": 0.0001076843657817109, | |
| "loss": 0.3086, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 33.96124031007752, | |
| "grad_norm": 1.0111535787582397, | |
| "learning_rate": 0.00010694690265486725, | |
| "loss": 0.276, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_cer": 0.05319781566599706, | |
| "eval_loss": 0.20418909192085266, | |
| "eval_runtime": 502.8988, | |
| "eval_samples_per_second": 20.245, | |
| "eval_steps_per_second": 2.531, | |
| "eval_wer": 0.20802027699828007, | |
| "step": 15368 | |
| }, | |
| { | |
| "epoch": 34.070874861572534, | |
| "grad_norm": 0.7386752963066101, | |
| "learning_rate": 0.00010620943952802358, | |
| "loss": 0.2782, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 34.181616832779625, | |
| "grad_norm": 0.9310299754142761, | |
| "learning_rate": 0.00010547197640117993, | |
| "loss": 0.2745, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 34.29235880398671, | |
| "grad_norm": 1.1085954904556274, | |
| "learning_rate": 0.00010473451327433627, | |
| "loss": 0.265, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 34.4031007751938, | |
| "grad_norm": 0.7299510836601257, | |
| "learning_rate": 0.00010399705014749262, | |
| "loss": 0.2858, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 34.51384274640088, | |
| "grad_norm": 1.8487365245819092, | |
| "learning_rate": 0.00010325958702064896, | |
| "loss": 0.2941, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 34.62458471760797, | |
| "grad_norm": 1.3170086145401, | |
| "learning_rate": 0.00010252212389380531, | |
| "loss": 0.2873, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 34.735326688815064, | |
| "grad_norm": 0.7510431408882141, | |
| "learning_rate": 0.00010178466076696164, | |
| "loss": 0.2651, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 34.84606866002215, | |
| "grad_norm": 0.8078789710998535, | |
| "learning_rate": 0.00010104719764011799, | |
| "loss": 0.26, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 34.95681063122924, | |
| "grad_norm": 0.6926655769348145, | |
| "learning_rate": 0.00010030973451327433, | |
| "loss": 0.2757, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_cer": 0.051932323356854326, | |
| "eval_loss": 0.20506548881530762, | |
| "eval_runtime": 504.012, | |
| "eval_samples_per_second": 20.2, | |
| "eval_steps_per_second": 2.526, | |
| "eval_wer": 0.20217253553000814, | |
| "step": 15820 | |
| }, | |
| { | |
| "epoch": 35.06644518272425, | |
| "grad_norm": 0.6662414073944092, | |
| "learning_rate": 9.957227138643068e-05, | |
| "loss": 0.2745, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 35.17718715393134, | |
| "grad_norm": 0.7030307054519653, | |
| "learning_rate": 9.883480825958702e-05, | |
| "loss": 0.2773, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 35.287929125138426, | |
| "grad_norm": 1.443966031074524, | |
| "learning_rate": 9.809734513274335e-05, | |
| "loss": 0.2666, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 35.39867109634552, | |
| "grad_norm": 0.5570212602615356, | |
| "learning_rate": 9.73598820058997e-05, | |
| "loss": 0.2723, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 35.5094130675526, | |
| "grad_norm": 0.9807683229446411, | |
| "learning_rate": 9.662241887905605e-05, | |
| "loss": 0.2675, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 35.62015503875969, | |
| "grad_norm": 0.8136458992958069, | |
| "learning_rate": 9.588495575221239e-05, | |
| "loss": 0.2754, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 35.730897009966775, | |
| "grad_norm": 1.0001354217529297, | |
| "learning_rate": 9.514749262536871e-05, | |
| "loss": 0.2678, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 35.841638981173865, | |
| "grad_norm": 1.3988227844238281, | |
| "learning_rate": 9.441002949852505e-05, | |
| "loss": 0.2751, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 35.95238095238095, | |
| "grad_norm": 0.6191902756690979, | |
| "learning_rate": 9.36725663716814e-05, | |
| "loss": 0.2853, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_cer": 0.052490735435960285, | |
| "eval_loss": 0.20767323672771454, | |
| "eval_runtime": 504.3578, | |
| "eval_samples_per_second": 20.186, | |
| "eval_steps_per_second": 2.524, | |
| "eval_wer": 0.2030415497420114, | |
| "step": 16272 | |
| }, | |
| { | |
| "epoch": 36.06201550387597, | |
| "grad_norm": 0.6683135628700256, | |
| "learning_rate": 9.293510324483775e-05, | |
| "loss": 0.2632, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 36.17275747508306, | |
| "grad_norm": 0.541814923286438, | |
| "learning_rate": 9.219764011799409e-05, | |
| "loss": 0.2948, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 36.28349944629014, | |
| "grad_norm": 0.746822714805603, | |
| "learning_rate": 9.146017699115044e-05, | |
| "loss": 0.2708, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 36.394241417497234, | |
| "grad_norm": 0.9451239109039307, | |
| "learning_rate": 9.072271386430677e-05, | |
| "loss": 0.2697, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 36.50498338870432, | |
| "grad_norm": 2.489182233810425, | |
| "learning_rate": 8.998525073746311e-05, | |
| "loss": 0.2739, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 36.61572535991141, | |
| "grad_norm": 0.8976282477378845, | |
| "learning_rate": 8.924778761061946e-05, | |
| "loss": 0.2649, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 36.72646733111849, | |
| "grad_norm": 1.0089658498764038, | |
| "learning_rate": 8.85103244837758e-05, | |
| "loss": 0.2661, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 36.83720930232558, | |
| "grad_norm": 0.7189285755157471, | |
| "learning_rate": 8.777286135693215e-05, | |
| "loss": 0.2659, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 36.94795127353267, | |
| "grad_norm": 1.8182567358016968, | |
| "learning_rate": 8.703539823008848e-05, | |
| "loss": 0.2785, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_cer": 0.051094705238195384, | |
| "eval_loss": 0.20127856731414795, | |
| "eval_runtime": 506.5771, | |
| "eval_samples_per_second": 20.098, | |
| "eval_steps_per_second": 2.513, | |
| "eval_wer": 0.19960170181949852, | |
| "step": 16724 | |
| }, | |
| { | |
| "epoch": 37.05758582502769, | |
| "grad_norm": 0.747435450553894, | |
| "learning_rate": 8.629793510324483e-05, | |
| "loss": 0.2703, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 37.16832779623477, | |
| "grad_norm": 0.985905110836029, | |
| "learning_rate": 8.556047197640117e-05, | |
| "loss": 0.2638, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 37.27906976744186, | |
| "grad_norm": 0.4967564642429352, | |
| "learning_rate": 8.482300884955752e-05, | |
| "loss": 0.2574, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 37.389811738648945, | |
| "grad_norm": 3.6833720207214355, | |
| "learning_rate": 8.408554572271386e-05, | |
| "loss": 0.2663, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 37.500553709856035, | |
| "grad_norm": 0.45124557614326477, | |
| "learning_rate": 8.334808259587021e-05, | |
| "loss": 0.2625, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 37.611295681063126, | |
| "grad_norm": 0.6523784399032593, | |
| "learning_rate": 8.261061946902654e-05, | |
| "loss": 0.2652, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 37.72203765227021, | |
| "grad_norm": 0.9279195070266724, | |
| "learning_rate": 8.187315634218289e-05, | |
| "loss": 0.2632, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 37.8327796234773, | |
| "grad_norm": 2.810720205307007, | |
| "learning_rate": 8.113569321533923e-05, | |
| "loss": 0.2673, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 37.943521594684384, | |
| "grad_norm": 0.7658355832099915, | |
| "learning_rate": 8.039823008849558e-05, | |
| "loss": 0.2744, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_cer": 0.050583431841091875, | |
| "eval_loss": 0.19885742664337158, | |
| "eval_runtime": 502.4471, | |
| "eval_samples_per_second": 20.263, | |
| "eval_steps_per_second": 2.534, | |
| "eval_wer": 0.19665067439123743, | |
| "step": 17176 | |
| }, | |
| { | |
| "epoch": 38.053156146179404, | |
| "grad_norm": 0.9786189198493958, | |
| "learning_rate": 7.966076696165192e-05, | |
| "loss": 0.2677, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 38.16389811738649, | |
| "grad_norm": 0.7271732091903687, | |
| "learning_rate": 7.892330383480825e-05, | |
| "loss": 0.2758, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 38.27464008859358, | |
| "grad_norm": 1.0245264768600464, | |
| "learning_rate": 7.81858407079646e-05, | |
| "loss": 0.2769, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 38.38538205980066, | |
| "grad_norm": 0.5469414591789246, | |
| "learning_rate": 7.744837758112093e-05, | |
| "loss": 0.2946, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 38.49612403100775, | |
| "grad_norm": 1.3482476472854614, | |
| "learning_rate": 7.671091445427728e-05, | |
| "loss": 0.2756, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 38.606866002214836, | |
| "grad_norm": 1.6001390218734741, | |
| "learning_rate": 7.597345132743361e-05, | |
| "loss": 0.2797, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 38.71760797342193, | |
| "grad_norm": 0.7820600271224976, | |
| "learning_rate": 7.523598820058995e-05, | |
| "loss": 0.2655, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 38.82834994462901, | |
| "grad_norm": 1.1540759801864624, | |
| "learning_rate": 7.449852507374631e-05, | |
| "loss": 0.2672, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 38.9390919158361, | |
| "grad_norm": 0.48907431960105896, | |
| "learning_rate": 7.376106194690265e-05, | |
| "loss": 0.258, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_cer": 0.0508408815658745, | |
| "eval_loss": 0.2034897357225418, | |
| "eval_runtime": 505.9396, | |
| "eval_samples_per_second": 20.123, | |
| "eval_steps_per_second": 2.516, | |
| "eval_wer": 0.19721191273648955, | |
| "step": 17628 | |
| }, | |
| { | |
| "epoch": 39.04872646733112, | |
| "grad_norm": 0.5294991135597229, | |
| "learning_rate": 7.302359882005899e-05, | |
| "loss": 0.2743, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 39.159468438538205, | |
| "grad_norm": 0.49712470173835754, | |
| "learning_rate": 7.228613569321534e-05, | |
| "loss": 0.2649, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 39.270210409745296, | |
| "grad_norm": 1.055932641029358, | |
| "learning_rate": 7.154867256637167e-05, | |
| "loss": 0.2804, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 39.38095238095238, | |
| "grad_norm": 6.277849197387695, | |
| "learning_rate": 7.081120943952801e-05, | |
| "loss": 0.2576, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 39.49169435215947, | |
| "grad_norm": 0.6586483716964722, | |
| "learning_rate": 7.007374631268436e-05, | |
| "loss": 0.2789, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 39.602436323366554, | |
| "grad_norm": 0.6979815363883972, | |
| "learning_rate": 6.93362831858407e-05, | |
| "loss": 0.2625, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 39.713178294573645, | |
| "grad_norm": 0.6091492176055908, | |
| "learning_rate": 6.859882005899705e-05, | |
| "loss": 0.2645, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 39.82392026578073, | |
| "grad_norm": 0.5214503407478333, | |
| "learning_rate": 6.786135693215338e-05, | |
| "loss": 0.2578, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 39.93466223698782, | |
| "grad_norm": 2.7882440090179443, | |
| "learning_rate": 6.712389380530973e-05, | |
| "loss": 0.2709, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_cer": 0.051054818661116386, | |
| "eval_loss": 0.20327427983283997, | |
| "eval_runtime": 506.6892, | |
| "eval_samples_per_second": 20.093, | |
| "eval_steps_per_second": 2.512, | |
| "eval_wer": 0.19846112066624422, | |
| "step": 18080 | |
| }, | |
| { | |
| "epoch": 40.04429678848283, | |
| "grad_norm": 0.5243245363235474, | |
| "learning_rate": 6.638643067846607e-05, | |
| "loss": 0.2726, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 40.15503875968992, | |
| "grad_norm": 2.323392152786255, | |
| "learning_rate": 6.564896755162242e-05, | |
| "loss": 0.2977, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 40.26578073089701, | |
| "grad_norm": 0.7332302927970886, | |
| "learning_rate": 6.491150442477875e-05, | |
| "loss": 0.2459, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 40.3765227021041, | |
| "grad_norm": 0.7128420472145081, | |
| "learning_rate": 6.41740412979351e-05, | |
| "loss": 0.254, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 40.48726467331119, | |
| "grad_norm": 0.6172935366630554, | |
| "learning_rate": 6.343657817109144e-05, | |
| "loss": 0.271, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 40.59800664451827, | |
| "grad_norm": 0.8429868817329407, | |
| "learning_rate": 6.269911504424779e-05, | |
| "loss": 0.2801, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 40.70874861572536, | |
| "grad_norm": 1.258792519569397, | |
| "learning_rate": 6.196165191740412e-05, | |
| "loss": 0.2458, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 40.819490586932446, | |
| "grad_norm": 0.4864571690559387, | |
| "learning_rate": 6.122418879056046e-05, | |
| "loss": 0.2615, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 40.93023255813954, | |
| "grad_norm": 0.49407410621643066, | |
| "learning_rate": 6.048672566371681e-05, | |
| "loss": 0.2419, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_cer": 0.04970230109289221, | |
| "eval_loss": 0.20170935988426208, | |
| "eval_runtime": 504.9031, | |
| "eval_samples_per_second": 20.164, | |
| "eval_steps_per_second": 2.521, | |
| "eval_wer": 0.19350049787272564, | |
| "step": 18532 | |
| }, | |
| { | |
| "epoch": 41.03986710963455, | |
| "grad_norm": 28.640010833740234, | |
| "learning_rate": 5.9749262536873155e-05, | |
| "loss": 18.8906, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 41.15060908084164, | |
| "grad_norm": 0.40494856238365173, | |
| "learning_rate": 5.9011799410029494e-05, | |
| "loss": 6.7845, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 41.261351052048724, | |
| "grad_norm": 1.3764681816101074, | |
| "learning_rate": 5.827433628318584e-05, | |
| "loss": 3.1387, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 41.372093023255815, | |
| "grad_norm": 0.8231045007705688, | |
| "learning_rate": 5.753687315634218e-05, | |
| "loss": 3.1009, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 41.4828349944629, | |
| "grad_norm": 0.6420056223869324, | |
| "learning_rate": 5.679941002949852e-05, | |
| "loss": 3.0734, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 41.59357696566999, | |
| "grad_norm": 1.9376227855682373, | |
| "learning_rate": 5.606194690265486e-05, | |
| "loss": 2.8817, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 41.70431893687708, | |
| "grad_norm": 1.5035388469696045, | |
| "learning_rate": 5.53244837758112e-05, | |
| "loss": 1.8786, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 41.81506090808416, | |
| "grad_norm": 0.897224485874176, | |
| "learning_rate": 5.4587020648967546e-05, | |
| "loss": 1.1317, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 41.925802879291254, | |
| "grad_norm": 0.7507091760635376, | |
| "learning_rate": 5.3849557522123885e-05, | |
| "loss": 0.8011, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_cer": 0.062023627357840616, | |
| "eval_loss": 0.4412783980369568, | |
| "eval_runtime": 457.2135, | |
| "eval_samples_per_second": 22.267, | |
| "eval_steps_per_second": 2.784, | |
| "eval_wer": 0.24151353308590567, | |
| "step": 18984 | |
| }, | |
| { | |
| "epoch": 42.03543743078627, | |
| "grad_norm": 0.92021244764328, | |
| "learning_rate": 5.311209439528023e-05, | |
| "loss": 0.6342, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 42.14617940199336, | |
| "grad_norm": 1.2486605644226074, | |
| "learning_rate": 5.2374631268436575e-05, | |
| "loss": 0.5457, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 42.25692137320044, | |
| "grad_norm": 1.1830624341964722, | |
| "learning_rate": 5.1637168141592914e-05, | |
| "loss": 0.4583, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 42.36766334440753, | |
| "grad_norm": 0.9644584059715271, | |
| "learning_rate": 5.089970501474926e-05, | |
| "loss": 0.4325, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 42.478405315614616, | |
| "grad_norm": 0.5905857086181641, | |
| "learning_rate": 5.0162241887905605e-05, | |
| "loss": 0.4296, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 42.58914728682171, | |
| "grad_norm": 0.7383226752281189, | |
| "learning_rate": 4.9424778761061944e-05, | |
| "loss": 0.3936, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 42.69988925802879, | |
| "grad_norm": 1.9408543109893799, | |
| "learning_rate": 4.868731563421829e-05, | |
| "loss": 0.3912, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 42.81063122923588, | |
| "grad_norm": 0.9167234897613525, | |
| "learning_rate": 4.794985250737463e-05, | |
| "loss": 0.3729, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 42.921373200442964, | |
| "grad_norm": 0.4777889847755432, | |
| "learning_rate": 4.7212389380530966e-05, | |
| "loss": 0.3462, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_cer": 0.05363656801386602, | |
| "eval_loss": 0.23979651927947998, | |
| "eval_runtime": 447.7874, | |
| "eval_samples_per_second": 22.736, | |
| "eval_steps_per_second": 2.843, | |
| "eval_wer": 0.20780302344527926, | |
| "step": 19436 | |
| }, | |
| { | |
| "epoch": 43.031007751937985, | |
| "grad_norm": 0.6891963481903076, | |
| "learning_rate": 4.6474926253687305e-05, | |
| "loss": 0.3587, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 43.141749723145075, | |
| "grad_norm": 0.5241971611976624, | |
| "learning_rate": 4.573746312684365e-05, | |
| "loss": 0.3362, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 43.25249169435216, | |
| "grad_norm": 0.9411165714263916, | |
| "learning_rate": 4.4999999999999996e-05, | |
| "loss": 0.3442, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 43.36323366555925, | |
| "grad_norm": 0.5330546498298645, | |
| "learning_rate": 4.4262536873156335e-05, | |
| "loss": 0.3323, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 43.47397563676633, | |
| "grad_norm": 5.120829105377197, | |
| "learning_rate": 4.352507374631268e-05, | |
| "loss": 0.3192, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 43.584717607973424, | |
| "grad_norm": 0.6921789050102234, | |
| "learning_rate": 4.2787610619469026e-05, | |
| "loss": 0.3134, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 43.69545957918051, | |
| "grad_norm": 1.0152885913848877, | |
| "learning_rate": 4.2050147492625364e-05, | |
| "loss": 0.3127, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 43.8062015503876, | |
| "grad_norm": 0.5916356444358826, | |
| "learning_rate": 4.131268436578171e-05, | |
| "loss": 0.3069, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 43.91694352159468, | |
| "grad_norm": 1.337470531463623, | |
| "learning_rate": 4.0575221238938055e-05, | |
| "loss": 0.3384, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_cer": 0.050960541297111486, | |
| "eval_loss": 0.2203516960144043, | |
| "eval_runtime": 444.0494, | |
| "eval_samples_per_second": 22.928, | |
| "eval_steps_per_second": 2.867, | |
| "eval_wer": 0.19795419570924233, | |
| "step": 19888 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 22600, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 50, | |
| "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": 1.638225383074747e+20, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |