| { | |
| "best_metric": 26.439685364873398, | |
| "best_model_checkpoint": "./whisper-ft-2/checkpoint-5000", | |
| "epoch": 1.0, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 26.441381454467773, | |
| "learning_rate": 6.875e-06, | |
| "loss": 5.6669, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 16.109895706176758, | |
| "learning_rate": 9.969806763285025e-06, | |
| "loss": 3.5576, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 17.341266632080078, | |
| "learning_rate": 9.919484702093398e-06, | |
| "loss": 2.3173, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 14.84139347076416, | |
| "learning_rate": 9.869162640901772e-06, | |
| "loss": 1.732, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 13.36408519744873, | |
| "learning_rate": 9.818840579710146e-06, | |
| "loss": 1.4347, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 15.8803071975708, | |
| "learning_rate": 9.768518518518519e-06, | |
| "loss": 1.2313, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 17.588510513305664, | |
| "learning_rate": 9.718196457326893e-06, | |
| "loss": 1.0921, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 14.752327919006348, | |
| "learning_rate": 9.667874396135266e-06, | |
| "loss": 0.9493, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 13.243648529052734, | |
| "learning_rate": 9.61755233494364e-06, | |
| "loss": 0.8546, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 15.082148551940918, | |
| "learning_rate": 9.567230273752013e-06, | |
| "loss": 0.8029, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 7.932609558105469, | |
| "learning_rate": 9.516908212560388e-06, | |
| "loss": 0.7795, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 14.98747444152832, | |
| "learning_rate": 9.46658615136876e-06, | |
| "loss": 0.7094, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 10.299836158752441, | |
| "learning_rate": 9.416264090177135e-06, | |
| "loss": 0.6991, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 9.431194305419922, | |
| "learning_rate": 9.36594202898551e-06, | |
| "loss": 0.6727, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 12.089103698730469, | |
| "learning_rate": 9.315619967793882e-06, | |
| "loss": 0.6613, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 11.374141693115234, | |
| "learning_rate": 9.265297906602255e-06, | |
| "loss": 0.6281, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 10.616567611694336, | |
| "learning_rate": 9.214975845410629e-06, | |
| "loss": 0.5975, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 12.249051094055176, | |
| "learning_rate": 9.164653784219002e-06, | |
| "loss": 0.5734, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 12.755301475524902, | |
| "learning_rate": 9.114331723027376e-06, | |
| "loss": 0.525, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 8.402365684509277, | |
| "learning_rate": 9.06400966183575e-06, | |
| "loss": 0.5021, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 8.66573429107666, | |
| "learning_rate": 9.013687600644123e-06, | |
| "loss": 0.5239, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 11.849873542785645, | |
| "learning_rate": 8.963365539452496e-06, | |
| "loss": 0.4548, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 8.986794471740723, | |
| "learning_rate": 8.91304347826087e-06, | |
| "loss": 0.4634, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 18.423568725585938, | |
| "learning_rate": 8.862721417069245e-06, | |
| "loss": 0.4796, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 7.8966217041015625, | |
| "learning_rate": 8.812399355877618e-06, | |
| "loss": 0.4142, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 8.645583152770996, | |
| "learning_rate": 8.76207729468599e-06, | |
| "loss": 0.438, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 8.235858917236328, | |
| "learning_rate": 8.711755233494365e-06, | |
| "loss": 0.448, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 10.634264945983887, | |
| "learning_rate": 8.661433172302737e-06, | |
| "loss": 0.454, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 7.304662704467773, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.441, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 7.09723424911499, | |
| "learning_rate": 8.560789049919486e-06, | |
| "loss": 0.4172, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 11.544793128967285, | |
| "learning_rate": 8.510466988727859e-06, | |
| "loss": 0.4305, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 8.155645370483398, | |
| "learning_rate": 8.460144927536232e-06, | |
| "loss": 0.4273, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 7.164479732513428, | |
| "learning_rate": 8.409822866344606e-06, | |
| "loss": 0.4339, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 8.852354049682617, | |
| "learning_rate": 8.35950080515298e-06, | |
| "loss": 0.4595, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 9.475948333740234, | |
| "learning_rate": 8.309178743961353e-06, | |
| "loss": 0.4104, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 9.111871719360352, | |
| "learning_rate": 8.258856682769728e-06, | |
| "loss": 0.4573, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 8.254072189331055, | |
| "learning_rate": 8.2085346215781e-06, | |
| "loss": 0.5042, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 9.376124382019043, | |
| "learning_rate": 8.158212560386473e-06, | |
| "loss": 0.488, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 8.461128234863281, | |
| "learning_rate": 8.107890499194848e-06, | |
| "loss": 0.4621, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 8.678207397460938, | |
| "learning_rate": 8.057568438003222e-06, | |
| "loss": 0.4632, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.5091106295585632, | |
| "eval_runtime": 245.5352, | |
| "eval_samples_per_second": 4.073, | |
| "eval_steps_per_second": 0.509, | |
| "eval_wer": 38.22701957857266, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 8.345236778259277, | |
| "learning_rate": 8.007246376811595e-06, | |
| "loss": 0.4577, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 8.78304386138916, | |
| "learning_rate": 7.956924315619969e-06, | |
| "loss": 0.4378, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 9.15285873413086, | |
| "learning_rate": 7.906602254428342e-06, | |
| "loss": 0.4534, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 10.154984474182129, | |
| "learning_rate": 7.856280193236716e-06, | |
| "loss": 0.4185, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 8.451964378356934, | |
| "learning_rate": 7.805958132045089e-06, | |
| "loss": 0.4098, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 7.336400508880615, | |
| "learning_rate": 7.755636070853463e-06, | |
| "loss": 0.3994, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 8.46390151977539, | |
| "learning_rate": 7.705314009661836e-06, | |
| "loss": 0.4028, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 10.172577857971191, | |
| "learning_rate": 7.654991948470209e-06, | |
| "loss": 0.3936, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 10.022027015686035, | |
| "learning_rate": 7.604669887278584e-06, | |
| "loss": 0.4025, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 10.936142921447754, | |
| "learning_rate": 7.5543478260869576e-06, | |
| "loss": 0.3701, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 8.6853609085083, | |
| "learning_rate": 7.504025764895331e-06, | |
| "loss": 0.3529, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 8.691036224365234, | |
| "learning_rate": 7.453703703703704e-06, | |
| "loss": 0.3525, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 7.8596978187561035, | |
| "learning_rate": 7.403381642512077e-06, | |
| "loss": 0.3481, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 6.271942615509033, | |
| "learning_rate": 7.353059581320452e-06, | |
| "loss": 0.3225, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 7.836112022399902, | |
| "learning_rate": 7.302737520128825e-06, | |
| "loss": 0.3821, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 9.443171501159668, | |
| "learning_rate": 7.252415458937199e-06, | |
| "loss": 0.4297, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 8.32925033569336, | |
| "learning_rate": 7.202093397745572e-06, | |
| "loss": 0.4633, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 10.218457221984863, | |
| "learning_rate": 7.151771336553945e-06, | |
| "loss": 0.4489, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 8.612972259521484, | |
| "learning_rate": 7.10144927536232e-06, | |
| "loss": 0.4022, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 9.206517219543457, | |
| "learning_rate": 7.051127214170693e-06, | |
| "loss": 0.3668, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 10.358244895935059, | |
| "learning_rate": 7.000805152979067e-06, | |
| "loss": 0.4162, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 12.02837085723877, | |
| "learning_rate": 6.95048309178744e-06, | |
| "loss": 0.436, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 10.476570129394531, | |
| "learning_rate": 6.900161030595813e-06, | |
| "loss": 0.3599, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 8.006072044372559, | |
| "learning_rate": 6.8498389694041875e-06, | |
| "loss": 0.3669, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 16.798221588134766, | |
| "learning_rate": 6.799516908212561e-06, | |
| "loss": 0.3632, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 8.768495559692383, | |
| "learning_rate": 6.749194847020935e-06, | |
| "loss": 0.3225, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 10.018780708312988, | |
| "learning_rate": 6.698872785829308e-06, | |
| "loss": 0.325, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 10.418109893798828, | |
| "learning_rate": 6.648550724637681e-06, | |
| "loss": 0.3321, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 6.952179908752441, | |
| "learning_rate": 6.598228663446056e-06, | |
| "loss": 0.3242, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 10.024521827697754, | |
| "learning_rate": 6.547906602254429e-06, | |
| "loss": 0.3291, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 7.440296173095703, | |
| "learning_rate": 6.497584541062802e-06, | |
| "loss": 0.329, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 8.184727668762207, | |
| "learning_rate": 6.447262479871176e-06, | |
| "loss": 0.3228, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 8.101869583129883, | |
| "learning_rate": 6.3969404186795495e-06, | |
| "loss": 0.3222, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 12.450462341308594, | |
| "learning_rate": 6.346618357487924e-06, | |
| "loss": 0.35, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 13.891828536987305, | |
| "learning_rate": 6.296296296296297e-06, | |
| "loss": 0.3659, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 9.327531814575195, | |
| "learning_rate": 6.24597423510467e-06, | |
| "loss": 0.3609, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 7.067734241485596, | |
| "learning_rate": 6.195652173913044e-06, | |
| "loss": 0.3558, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 11.90934944152832, | |
| "learning_rate": 6.145330112721417e-06, | |
| "loss": 0.3312, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 8.565199851989746, | |
| "learning_rate": 6.095008051529792e-06, | |
| "loss": 0.3347, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 6.947161674499512, | |
| "learning_rate": 6.044685990338165e-06, | |
| "loss": 0.2988, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.42369821667671204, | |
| "eval_runtime": 243.0792, | |
| "eval_samples_per_second": 4.114, | |
| "eval_steps_per_second": 0.514, | |
| "eval_wer": 32.68071424470345, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 10.25965404510498, | |
| "learning_rate": 5.994363929146538e-06, | |
| "loss": 0.3133, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 12.943211555480957, | |
| "learning_rate": 5.944041867954912e-06, | |
| "loss": 0.3662, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 8.808210372924805, | |
| "learning_rate": 5.893719806763285e-06, | |
| "loss": 0.3933, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 8.858198165893555, | |
| "learning_rate": 5.84339774557166e-06, | |
| "loss": 0.4115, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 8.931889533996582, | |
| "learning_rate": 5.793075684380033e-06, | |
| "loss": 0.3585, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 9.787882804870605, | |
| "learning_rate": 5.742753623188406e-06, | |
| "loss": 0.351, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 8.339137077331543, | |
| "learning_rate": 5.692431561996779e-06, | |
| "loss": 0.3425, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 7.447598457336426, | |
| "learning_rate": 5.642109500805153e-06, | |
| "loss": 0.3567, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 7.393462181091309, | |
| "learning_rate": 5.591787439613527e-06, | |
| "loss": 0.3852, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 9.472132682800293, | |
| "learning_rate": 5.541465378421901e-06, | |
| "loss": 0.3978, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 9.781373023986816, | |
| "learning_rate": 5.4911433172302745e-06, | |
| "loss": 0.3917, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 10.88880443572998, | |
| "learning_rate": 5.440821256038647e-06, | |
| "loss": 0.4059, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 10.626157760620117, | |
| "learning_rate": 5.390499194847021e-06, | |
| "loss": 0.3995, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 10.283743858337402, | |
| "learning_rate": 5.340177133655395e-06, | |
| "loss": 0.4168, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 9.085090637207031, | |
| "learning_rate": 5.289855072463769e-06, | |
| "loss": 0.3932, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 9.050111770629883, | |
| "learning_rate": 5.239533011272142e-06, | |
| "loss": 0.386, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 7.001188278198242, | |
| "learning_rate": 5.189210950080515e-06, | |
| "loss": 0.3305, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 8.82314395904541, | |
| "learning_rate": 5.138888888888889e-06, | |
| "loss": 0.318, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 8.965357780456543, | |
| "learning_rate": 5.088566827697263e-06, | |
| "loss": 0.3568, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 8.673226356506348, | |
| "learning_rate": 5.038244766505637e-06, | |
| "loss": 0.3693, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 9.710456848144531, | |
| "learning_rate": 4.98792270531401e-06, | |
| "loss": 0.4114, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 11.537069320678711, | |
| "learning_rate": 4.937600644122384e-06, | |
| "loss": 0.4652, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 11.736916542053223, | |
| "learning_rate": 4.887278582930757e-06, | |
| "loss": 0.4348, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 8.418719291687012, | |
| "learning_rate": 4.836956521739131e-06, | |
| "loss": 0.3498, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 8.387438774108887, | |
| "learning_rate": 4.786634460547504e-06, | |
| "loss": 0.33, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 7.15954065322876, | |
| "learning_rate": 4.736312399355878e-06, | |
| "loss": 0.3454, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 7.480624675750732, | |
| "learning_rate": 4.6859903381642516e-06, | |
| "loss": 0.3317, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 7.614017963409424, | |
| "learning_rate": 4.635668276972625e-06, | |
| "loss": 0.3202, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 9.156942367553711, | |
| "learning_rate": 4.585346215780999e-06, | |
| "loss": 0.2789, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 10.527360916137695, | |
| "learning_rate": 4.535024154589372e-06, | |
| "loss": 0.2546, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 7.232975959777832, | |
| "learning_rate": 4.484702093397746e-06, | |
| "loss": 0.2724, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 6.251858711242676, | |
| "learning_rate": 4.434380032206119e-06, | |
| "loss": 0.2791, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 8.357397079467773, | |
| "learning_rate": 4.384057971014493e-06, | |
| "loss": 0.307, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 8.629735946655273, | |
| "learning_rate": 4.3337359098228665e-06, | |
| "loss": 0.3721, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 8.04019832611084, | |
| "learning_rate": 4.28341384863124e-06, | |
| "loss": 0.3302, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 10.451481819152832, | |
| "learning_rate": 4.233091787439614e-06, | |
| "loss": 0.3338, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 7.338428020477295, | |
| "learning_rate": 4.182769726247988e-06, | |
| "loss": 0.3381, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 6.68162727355957, | |
| "learning_rate": 4.132447665056361e-06, | |
| "loss": 0.2993, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 7.178064823150635, | |
| "learning_rate": 4.082125603864734e-06, | |
| "loss": 0.2728, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 6.970361709594727, | |
| "learning_rate": 4.031803542673109e-06, | |
| "loss": 0.2715, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.3627614378929138, | |
| "eval_runtime": 242.2219, | |
| "eval_samples_per_second": 4.128, | |
| "eval_steps_per_second": 0.516, | |
| "eval_wer": 28.190848022047426, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 7.680355072021484, | |
| "learning_rate": 3.9814814814814814e-06, | |
| "loss": 0.2641, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 7.352539539337158, | |
| "learning_rate": 3.931159420289856e-06, | |
| "loss": 0.2947, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 7.088489055633545, | |
| "learning_rate": 3.880837359098229e-06, | |
| "loss": 0.3204, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 9.058192253112793, | |
| "learning_rate": 3.830515297906602e-06, | |
| "loss": 0.3085, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 10.828897476196289, | |
| "learning_rate": 3.780193236714976e-06, | |
| "loss": 0.2772, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 7.811698913574219, | |
| "learning_rate": 3.7298711755233497e-06, | |
| "loss": 0.2449, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 7.802531719207764, | |
| "learning_rate": 3.6795491143317237e-06, | |
| "loss": 0.2515, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 7.107741832733154, | |
| "learning_rate": 3.629227053140097e-06, | |
| "loss": 0.2964, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 7.613478183746338, | |
| "learning_rate": 3.5789049919484704e-06, | |
| "loss": 0.2901, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 6.808854103088379, | |
| "learning_rate": 3.5285829307568444e-06, | |
| "loss": 0.2764, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 8.42264461517334, | |
| "learning_rate": 3.4782608695652175e-06, | |
| "loss": 0.2439, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 6.853939533233643, | |
| "learning_rate": 3.4279388083735915e-06, | |
| "loss": 0.263, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 7.323888301849365, | |
| "learning_rate": 3.377616747181965e-06, | |
| "loss": 0.3071, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 8.018033027648926, | |
| "learning_rate": 3.327294685990338e-06, | |
| "loss": 0.3161, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 6.292661666870117, | |
| "learning_rate": 3.276972624798712e-06, | |
| "loss": 0.2889, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 7.282548904418945, | |
| "learning_rate": 3.2266505636070853e-06, | |
| "loss": 0.2909, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 9.771224975585938, | |
| "learning_rate": 3.1763285024154593e-06, | |
| "loss": 0.2804, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 6.97499418258667, | |
| "learning_rate": 3.126006441223833e-06, | |
| "loss": 0.3079, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 6.536733627319336, | |
| "learning_rate": 3.075684380032206e-06, | |
| "loss": 0.2727, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 9.328429222106934, | |
| "learning_rate": 3.02536231884058e-06, | |
| "loss": 0.2927, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 10.615299224853516, | |
| "learning_rate": 2.9750402576489536e-06, | |
| "loss": 0.2957, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 8.456421852111816, | |
| "learning_rate": 2.9247181964573276e-06, | |
| "loss": 0.3089, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 7.571265697479248, | |
| "learning_rate": 2.8743961352657007e-06, | |
| "loss": 0.3166, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 7.515317440032959, | |
| "learning_rate": 2.8240740740740743e-06, | |
| "loss": 0.3027, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 9.137214660644531, | |
| "learning_rate": 2.773752012882448e-06, | |
| "loss": 0.2972, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 8.118337631225586, | |
| "learning_rate": 2.7234299516908214e-06, | |
| "loss": 0.3027, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 11.383882522583008, | |
| "learning_rate": 2.6731078904991954e-06, | |
| "loss": 0.3401, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 8.050200462341309, | |
| "learning_rate": 2.6227858293075685e-06, | |
| "loss": 0.3462, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 8.983549118041992, | |
| "learning_rate": 2.572463768115942e-06, | |
| "loss": 0.3346, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 5.890308380126953, | |
| "learning_rate": 2.522141706924316e-06, | |
| "loss": 0.3202, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 7.097532272338867, | |
| "learning_rate": 2.4718196457326892e-06, | |
| "loss": 0.2893, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 8.50672721862793, | |
| "learning_rate": 2.4214975845410628e-06, | |
| "loss": 0.2999, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 6.510353088378906, | |
| "learning_rate": 2.3711755233494368e-06, | |
| "loss": 0.3054, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 7.819746971130371, | |
| "learning_rate": 2.3208534621578103e-06, | |
| "loss": 0.2801, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 9.897961616516113, | |
| "learning_rate": 2.270531400966184e-06, | |
| "loss": 0.2668, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 6.692742347717285, | |
| "learning_rate": 2.220209339774557e-06, | |
| "loss": 0.2632, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 6.9810404777526855, | |
| "learning_rate": 2.169887278582931e-06, | |
| "loss": 0.24, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 6.765547275543213, | |
| "learning_rate": 2.1195652173913046e-06, | |
| "loss": 0.2831, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 8.033112525939941, | |
| "learning_rate": 2.069243156199678e-06, | |
| "loss": 0.316, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 9.17114543914795, | |
| "learning_rate": 2.0189210950080517e-06, | |
| "loss": 0.3315, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.3504044711589813, | |
| "eval_runtime": 243.0184, | |
| "eval_samples_per_second": 4.115, | |
| "eval_steps_per_second": 0.514, | |
| "eval_wer": 28.288453809496467, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 8.472304344177246, | |
| "learning_rate": 1.9685990338164253e-06, | |
| "loss": 0.3486, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 8.651259422302246, | |
| "learning_rate": 1.918276972624799e-06, | |
| "loss": 0.3579, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 7.6784772872924805, | |
| "learning_rate": 1.8679549114331724e-06, | |
| "loss": 0.34, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 8.391368865966797, | |
| "learning_rate": 1.817632850241546e-06, | |
| "loss": 0.3065, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 6.728703022003174, | |
| "learning_rate": 1.7673107890499197e-06, | |
| "loss": 0.2958, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 6.842586517333984, | |
| "learning_rate": 1.7169887278582933e-06, | |
| "loss": 0.2566, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 6.050861835479736, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.2687, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 9.184779167175293, | |
| "learning_rate": 1.6163446054750404e-06, | |
| "loss": 0.2652, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 7.263152122497559, | |
| "learning_rate": 1.566022544283414e-06, | |
| "loss": 0.2604, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 7.312118053436279, | |
| "learning_rate": 1.5157004830917876e-06, | |
| "loss": 0.2352, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 10.993212699890137, | |
| "learning_rate": 1.4653784219001613e-06, | |
| "loss": 0.253, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 5.131811618804932, | |
| "learning_rate": 1.4150563607085347e-06, | |
| "loss": 0.2723, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 7.325843334197998, | |
| "learning_rate": 1.3647342995169083e-06, | |
| "loss": 0.2579, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 7.593961715698242, | |
| "learning_rate": 1.3144122383252818e-06, | |
| "loss": 0.2666, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 8.811367988586426, | |
| "learning_rate": 1.2640901771336556e-06, | |
| "loss": 0.2799, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 10.823134422302246, | |
| "learning_rate": 1.2137681159420292e-06, | |
| "loss": 0.2999, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 7.269660949707031, | |
| "learning_rate": 1.1634460547504027e-06, | |
| "loss": 0.3029, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 9.235569953918457, | |
| "learning_rate": 1.1131239935587763e-06, | |
| "loss": 0.3232, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 8.534402847290039, | |
| "learning_rate": 1.0628019323671499e-06, | |
| "loss": 0.3101, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 9.410504341125488, | |
| "learning_rate": 1.0124798711755234e-06, | |
| "loss": 0.3178, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 8.4865083694458, | |
| "learning_rate": 9.62157809983897e-07, | |
| "loss": 0.2748, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 7.169732093811035, | |
| "learning_rate": 9.118357487922707e-07, | |
| "loss": 0.281, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 7.270644187927246, | |
| "learning_rate": 8.615136876006441e-07, | |
| "loss": 0.2714, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 6.573908805847168, | |
| "learning_rate": 8.111916264090178e-07, | |
| "loss": 0.278, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 9.489112854003906, | |
| "learning_rate": 7.608695652173914e-07, | |
| "loss": 0.2941, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 6.89200496673584, | |
| "learning_rate": 7.105475040257649e-07, | |
| "loss": 0.2792, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 8.81523323059082, | |
| "learning_rate": 6.602254428341386e-07, | |
| "loss": 0.2586, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 6.394697666168213, | |
| "learning_rate": 6.099033816425121e-07, | |
| "loss": 0.3147, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 7.134545803070068, | |
| "learning_rate": 5.595813204508857e-07, | |
| "loss": 0.3337, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 8.418858528137207, | |
| "learning_rate": 5.092592592592593e-07, | |
| "loss": 0.3001, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 7.931731224060059, | |
| "learning_rate": 4.5893719806763294e-07, | |
| "loss": 0.31, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 8.287907600402832, | |
| "learning_rate": 4.086151368760065e-07, | |
| "loss": 0.2927, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 6.868620872497559, | |
| "learning_rate": 3.5829307568438007e-07, | |
| "loss": 0.2962, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 8.824197769165039, | |
| "learning_rate": 3.079710144927537e-07, | |
| "loss": 0.2618, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 7.290480136871338, | |
| "learning_rate": 2.5764895330112725e-07, | |
| "loss": 0.248, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 5.285947799682617, | |
| "learning_rate": 2.073268921095008e-07, | |
| "loss": 0.2223, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 6.8819169998168945, | |
| "learning_rate": 1.570048309178744e-07, | |
| "loss": 0.2055, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 6.588069915771484, | |
| "learning_rate": 1.0668276972624801e-07, | |
| "loss": 0.2279, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 11.339116096496582, | |
| "learning_rate": 5.6360708534621584e-08, | |
| "loss": 0.2176, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 6.7692036628723145, | |
| "learning_rate": 6.0386473429951695e-09, | |
| "loss": 0.2065, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.3357778787612915, | |
| "eval_runtime": 244.8142, | |
| "eval_samples_per_second": 4.085, | |
| "eval_steps_per_second": 0.511, | |
| "eval_wer": 26.439685364873398, | |
| "step": 5000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "total_flos": 1.9695108096e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |