| { | |
| "best_metric": 44.0582094924667, | |
| "best_model_checkpoint": "./whisper-large-v3-Marathi-Version1/checkpoint-8000", | |
| "epoch": 17.24137931034483, | |
| "eval_steps": 2000, | |
| "global_step": 8000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.10775862068965517, | |
| "grad_norm": 0.3231394290924072, | |
| "learning_rate": 1.5000000000000002e-07, | |
| "loss": 0.5889, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.21551724137931033, | |
| "grad_norm": 0.3586913049221039, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 0.5863, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3232758620689655, | |
| "grad_norm": 0.415686696767807, | |
| "learning_rate": 4.5e-07, | |
| "loss": 0.5768, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.43103448275862066, | |
| "grad_norm": 0.4167291224002838, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 0.5821, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5387931034482759, | |
| "grad_norm": 0.34522315859794617, | |
| "learning_rate": 7.5e-07, | |
| "loss": 0.5665, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.646551724137931, | |
| "grad_norm": 0.3691834807395935, | |
| "learning_rate": 9e-07, | |
| "loss": 0.5628, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7543103448275862, | |
| "grad_norm": 0.344501793384552, | |
| "learning_rate": 1.05e-06, | |
| "loss": 0.5705, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.8620689655172413, | |
| "grad_norm": 0.2788464426994324, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 0.5616, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9698275862068966, | |
| "grad_norm": 0.27220970392227173, | |
| "learning_rate": 1.35e-06, | |
| "loss": 0.5344, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.0775862068965518, | |
| "grad_norm": 0.2518285810947418, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.5566, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.1853448275862069, | |
| "grad_norm": 0.29989466071128845, | |
| "learning_rate": 1.65e-06, | |
| "loss": 0.5213, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.293103448275862, | |
| "grad_norm": 0.26115870475769043, | |
| "learning_rate": 1.8e-06, | |
| "loss": 0.5062, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.4008620689655173, | |
| "grad_norm": 0.2464788407087326, | |
| "learning_rate": 1.95e-06, | |
| "loss": 0.5022, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.5086206896551724, | |
| "grad_norm": 0.2869219183921814, | |
| "learning_rate": 2.1e-06, | |
| "loss": 0.4828, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.6163793103448276, | |
| "grad_norm": 0.2619161307811737, | |
| "learning_rate": 2.25e-06, | |
| "loss": 0.4896, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.7241379310344827, | |
| "grad_norm": 0.33810245990753174, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.4639, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.831896551724138, | |
| "grad_norm": 0.2645559012889862, | |
| "learning_rate": 2.55e-06, | |
| "loss": 0.4464, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.9396551724137931, | |
| "grad_norm": 0.2778465151786804, | |
| "learning_rate": 2.7e-06, | |
| "loss": 0.4273, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.0474137931034484, | |
| "grad_norm": 0.2977587580680847, | |
| "learning_rate": 2.85e-06, | |
| "loss": 0.401, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.1551724137931036, | |
| "grad_norm": 0.3463541567325592, | |
| "learning_rate": 3e-06, | |
| "loss": 0.4094, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.2629310344827585, | |
| "grad_norm": 0.26817262172698975, | |
| "learning_rate": 2.992105263157895e-06, | |
| "loss": 0.3967, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.3706896551724137, | |
| "grad_norm": 0.34868425130844116, | |
| "learning_rate": 2.9842105263157896e-06, | |
| "loss": 0.3987, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.478448275862069, | |
| "grad_norm": 0.33089494705200195, | |
| "learning_rate": 2.9763157894736843e-06, | |
| "loss": 0.3953, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.586206896551724, | |
| "grad_norm": 0.3616925776004791, | |
| "learning_rate": 2.968421052631579e-06, | |
| "loss": 0.3668, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.6939655172413794, | |
| "grad_norm": 0.33365535736083984, | |
| "learning_rate": 2.960526315789474e-06, | |
| "loss": 0.3735, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.8017241379310347, | |
| "grad_norm": 0.3913170099258423, | |
| "learning_rate": 2.9526315789473685e-06, | |
| "loss": 0.3632, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.9094827586206895, | |
| "grad_norm": 0.2940693497657776, | |
| "learning_rate": 2.9447368421052633e-06, | |
| "loss": 0.359, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.0172413793103448, | |
| "grad_norm": 0.3293132781982422, | |
| "learning_rate": 2.936842105263158e-06, | |
| "loss": 0.3824, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.125, | |
| "grad_norm": 0.34137511253356934, | |
| "learning_rate": 2.9289473684210528e-06, | |
| "loss": 0.3616, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.2327586206896552, | |
| "grad_norm": 0.34217217564582825, | |
| "learning_rate": 2.9210526315789475e-06, | |
| "loss": 0.3525, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.3405172413793105, | |
| "grad_norm": 0.39114320278167725, | |
| "learning_rate": 2.9131578947368423e-06, | |
| "loss": 0.3413, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.4482758620689653, | |
| "grad_norm": 0.3646846115589142, | |
| "learning_rate": 2.905263157894737e-06, | |
| "loss": 0.3385, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.5560344827586206, | |
| "grad_norm": 0.4652530550956726, | |
| "learning_rate": 2.8973684210526318e-06, | |
| "loss": 0.3534, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.663793103448276, | |
| "grad_norm": 0.3728253245353699, | |
| "learning_rate": 2.8894736842105265e-06, | |
| "loss": 0.3392, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.771551724137931, | |
| "grad_norm": 0.4309278428554535, | |
| "learning_rate": 2.8815789473684213e-06, | |
| "loss": 0.3249, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.8793103448275863, | |
| "grad_norm": 0.4407905042171478, | |
| "learning_rate": 2.873684210526316e-06, | |
| "loss": 0.3218, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.987068965517241, | |
| "grad_norm": 0.41783958673477173, | |
| "learning_rate": 2.8657894736842103e-06, | |
| "loss": 0.3285, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.094827586206897, | |
| "grad_norm": 0.38858509063720703, | |
| "learning_rate": 2.857894736842105e-06, | |
| "loss": 0.3292, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.202586206896552, | |
| "grad_norm": 0.37383902072906494, | |
| "learning_rate": 2.85e-06, | |
| "loss": 0.3252, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.310344827586207, | |
| "grad_norm": 0.5148123502731323, | |
| "learning_rate": 2.8421052631578946e-06, | |
| "loss": 0.315, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.310344827586207, | |
| "eval_loss": 0.3307957649230957, | |
| "eval_runtime": 5462.6499, | |
| "eval_samples_per_second": 0.186, | |
| "eval_steps_per_second": 0.023, | |
| "eval_wer": 52.398827582660566, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.418103448275862, | |
| "grad_norm": 0.4609113931655884, | |
| "learning_rate": 2.8342105263157897e-06, | |
| "loss": 0.318, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.525862068965517, | |
| "grad_norm": 0.502840518951416, | |
| "learning_rate": 2.8263157894736845e-06, | |
| "loss": 0.3232, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.633620689655173, | |
| "grad_norm": 0.4469599425792694, | |
| "learning_rate": 2.8184210526315792e-06, | |
| "loss": 0.3118, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.741379310344827, | |
| "grad_norm": 0.4370408356189728, | |
| "learning_rate": 2.810526315789474e-06, | |
| "loss": 0.3068, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.849137931034483, | |
| "grad_norm": 0.46128618717193604, | |
| "learning_rate": 2.8026315789473687e-06, | |
| "loss": 0.3154, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.956896551724138, | |
| "grad_norm": 0.5146879553794861, | |
| "learning_rate": 2.7947368421052635e-06, | |
| "loss": 0.3102, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.064655172413793, | |
| "grad_norm": 0.5018391609191895, | |
| "learning_rate": 2.7868421052631578e-06, | |
| "loss": 0.3113, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 5.172413793103448, | |
| "grad_norm": 0.42680951952934265, | |
| "learning_rate": 2.7789473684210525e-06, | |
| "loss": 0.3143, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 5.280172413793103, | |
| "grad_norm": 0.5996211767196655, | |
| "learning_rate": 2.7710526315789473e-06, | |
| "loss": 0.2932, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 5.387931034482759, | |
| "grad_norm": 0.48398110270500183, | |
| "learning_rate": 2.763157894736842e-06, | |
| "loss": 0.287, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.495689655172414, | |
| "grad_norm": 0.6341568231582642, | |
| "learning_rate": 2.7552631578947368e-06, | |
| "loss": 0.3083, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.603448275862069, | |
| "grad_norm": 0.42023149132728577, | |
| "learning_rate": 2.7473684210526315e-06, | |
| "loss": 0.3064, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.711206896551724, | |
| "grad_norm": 0.5808025598526001, | |
| "learning_rate": 2.7394736842105263e-06, | |
| "loss": 0.2893, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.818965517241379, | |
| "grad_norm": 0.5769814252853394, | |
| "learning_rate": 2.7315789473684214e-06, | |
| "loss": 0.2837, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.926724137931035, | |
| "grad_norm": 0.5669300556182861, | |
| "learning_rate": 2.723684210526316e-06, | |
| "loss": 0.2951, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 6.0344827586206895, | |
| "grad_norm": 0.4516044855117798, | |
| "learning_rate": 2.715789473684211e-06, | |
| "loss": 0.3043, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 6.142241379310345, | |
| "grad_norm": 0.4493812322616577, | |
| "learning_rate": 2.7078947368421052e-06, | |
| "loss": 0.2882, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "grad_norm": 0.5748199820518494, | |
| "learning_rate": 2.7e-06, | |
| "loss": 0.2935, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 6.357758620689655, | |
| "grad_norm": 0.556123673915863, | |
| "learning_rate": 2.6921052631578947e-06, | |
| "loss": 0.2876, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 6.4655172413793105, | |
| "grad_norm": 0.5574390292167664, | |
| "learning_rate": 2.6842105263157895e-06, | |
| "loss": 0.289, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.573275862068965, | |
| "grad_norm": 0.6384408473968506, | |
| "learning_rate": 2.6763157894736842e-06, | |
| "loss": 0.2916, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 6.681034482758621, | |
| "grad_norm": 0.5367217659950256, | |
| "learning_rate": 2.668421052631579e-06, | |
| "loss": 0.2747, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.788793103448276, | |
| "grad_norm": 0.7160293459892273, | |
| "learning_rate": 2.6605263157894737e-06, | |
| "loss": 0.2938, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.896551724137931, | |
| "grad_norm": 0.623241662979126, | |
| "learning_rate": 2.6526315789473685e-06, | |
| "loss": 0.2826, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 7.004310344827586, | |
| "grad_norm": 0.5312695503234863, | |
| "learning_rate": 2.644736842105263e-06, | |
| "loss": 0.2774, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 7.112068965517241, | |
| "grad_norm": 0.5798662900924683, | |
| "learning_rate": 2.636842105263158e-06, | |
| "loss": 0.2906, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 7.219827586206897, | |
| "grad_norm": 0.5495738387107849, | |
| "learning_rate": 2.6289473684210527e-06, | |
| "loss": 0.2853, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 7.327586206896552, | |
| "grad_norm": 0.6797431111335754, | |
| "learning_rate": 2.6210526315789474e-06, | |
| "loss": 0.2723, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 7.435344827586207, | |
| "grad_norm": 0.6902515292167664, | |
| "learning_rate": 2.613157894736842e-06, | |
| "loss": 0.2785, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 7.543103448275862, | |
| "grad_norm": 0.6549153327941895, | |
| "learning_rate": 2.605263157894737e-06, | |
| "loss": 0.2826, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.650862068965517, | |
| "grad_norm": 0.6934303641319275, | |
| "learning_rate": 2.5973684210526317e-06, | |
| "loss": 0.279, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 7.758620689655173, | |
| "grad_norm": 0.5600142478942871, | |
| "learning_rate": 2.5894736842105264e-06, | |
| "loss": 0.2755, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 7.866379310344827, | |
| "grad_norm": 0.6655827164649963, | |
| "learning_rate": 2.581578947368421e-06, | |
| "loss": 0.2512, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 7.974137931034483, | |
| "grad_norm": 0.6687513589859009, | |
| "learning_rate": 2.573684210526316e-06, | |
| "loss": 0.2817, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 8.081896551724139, | |
| "grad_norm": 0.6291229724884033, | |
| "learning_rate": 2.5657894736842107e-06, | |
| "loss": 0.2637, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 8.189655172413794, | |
| "grad_norm": 0.5600979328155518, | |
| "learning_rate": 2.5578947368421054e-06, | |
| "loss": 0.2808, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 8.297413793103448, | |
| "grad_norm": 0.7537655830383301, | |
| "learning_rate": 2.55e-06, | |
| "loss": 0.2757, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 8.405172413793103, | |
| "grad_norm": 0.6032606363296509, | |
| "learning_rate": 2.542105263157895e-06, | |
| "loss": 0.2719, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 8.512931034482758, | |
| "grad_norm": 0.7251238822937012, | |
| "learning_rate": 2.5342105263157892e-06, | |
| "loss": 0.2828, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 8.620689655172415, | |
| "grad_norm": 0.5968174338340759, | |
| "learning_rate": 2.526315789473684e-06, | |
| "loss": 0.2584, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.620689655172415, | |
| "eval_loss": 0.2882911264896393, | |
| "eval_runtime": 12265.8085, | |
| "eval_samples_per_second": 0.083, | |
| "eval_steps_per_second": 0.01, | |
| "eval_wer": 47.524039697639736, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.72844827586207, | |
| "grad_norm": 0.5381304621696472, | |
| "learning_rate": 2.5184210526315787e-06, | |
| "loss": 0.2712, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 8.836206896551724, | |
| "grad_norm": 0.5945307612419128, | |
| "learning_rate": 2.510526315789474e-06, | |
| "loss": 0.2562, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 8.943965517241379, | |
| "grad_norm": 0.6147953867912292, | |
| "learning_rate": 2.5026315789473686e-06, | |
| "loss": 0.2604, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 9.051724137931034, | |
| "grad_norm": 0.6225630640983582, | |
| "learning_rate": 2.4947368421052634e-06, | |
| "loss": 0.2712, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 9.15948275862069, | |
| "grad_norm": 0.6314706206321716, | |
| "learning_rate": 2.486842105263158e-06, | |
| "loss": 0.2522, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 9.267241379310345, | |
| "grad_norm": 0.7159410119056702, | |
| "learning_rate": 2.478947368421053e-06, | |
| "loss": 0.2554, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 9.375, | |
| "grad_norm": 0.7329779267311096, | |
| "learning_rate": 2.4710526315789476e-06, | |
| "loss": 0.2648, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 9.482758620689655, | |
| "grad_norm": 0.6220468878746033, | |
| "learning_rate": 2.4631578947368424e-06, | |
| "loss": 0.2753, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 9.59051724137931, | |
| "grad_norm": 0.697035551071167, | |
| "learning_rate": 2.4552631578947367e-06, | |
| "loss": 0.2636, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 9.698275862068966, | |
| "grad_norm": 0.6941649913787842, | |
| "learning_rate": 2.4473684210526314e-06, | |
| "loss": 0.2547, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 9.806034482758621, | |
| "grad_norm": 0.525019109249115, | |
| "learning_rate": 2.439473684210526e-06, | |
| "loss": 0.2574, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 9.913793103448276, | |
| "grad_norm": 0.652886688709259, | |
| "learning_rate": 2.431578947368421e-06, | |
| "loss": 0.2629, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 10.02155172413793, | |
| "grad_norm": 0.8766529560089111, | |
| "learning_rate": 2.4236842105263157e-06, | |
| "loss": 0.2546, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 10.129310344827585, | |
| "grad_norm": 0.5778140425682068, | |
| "learning_rate": 2.4157894736842104e-06, | |
| "loss": 0.2675, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 10.237068965517242, | |
| "grad_norm": 0.6798537373542786, | |
| "learning_rate": 2.4078947368421056e-06, | |
| "loss": 0.259, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 10.344827586206897, | |
| "grad_norm": 0.5612985491752625, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.2539, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 10.452586206896552, | |
| "grad_norm": 0.7032902240753174, | |
| "learning_rate": 2.392105263157895e-06, | |
| "loss": 0.2553, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 10.560344827586206, | |
| "grad_norm": 0.7675493359565735, | |
| "learning_rate": 2.38421052631579e-06, | |
| "loss": 0.2558, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 10.668103448275861, | |
| "grad_norm": 0.6644128561019897, | |
| "learning_rate": 2.376315789473684e-06, | |
| "loss": 0.2611, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 10.775862068965518, | |
| "grad_norm": 0.7273836135864258, | |
| "learning_rate": 2.368421052631579e-06, | |
| "loss": 0.2448, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.883620689655173, | |
| "grad_norm": 0.7355319857597351, | |
| "learning_rate": 2.3605263157894736e-06, | |
| "loss": 0.2516, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 10.991379310344827, | |
| "grad_norm": 0.7023867964744568, | |
| "learning_rate": 2.3526315789473684e-06, | |
| "loss": 0.2486, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 11.099137931034482, | |
| "grad_norm": 0.7921417355537415, | |
| "learning_rate": 2.344736842105263e-06, | |
| "loss": 0.2379, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 11.206896551724139, | |
| "grad_norm": 0.7202710509300232, | |
| "learning_rate": 2.336842105263158e-06, | |
| "loss": 0.2474, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 11.314655172413794, | |
| "grad_norm": 0.7898572087287903, | |
| "learning_rate": 2.3289473684210526e-06, | |
| "loss": 0.2724, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 11.422413793103448, | |
| "grad_norm": 0.6319573521614075, | |
| "learning_rate": 2.3210526315789473e-06, | |
| "loss": 0.2532, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 11.530172413793103, | |
| "grad_norm": 0.717001736164093, | |
| "learning_rate": 2.313157894736842e-06, | |
| "loss": 0.2448, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 11.637931034482758, | |
| "grad_norm": 0.8957849740982056, | |
| "learning_rate": 2.305263157894737e-06, | |
| "loss": 0.2584, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 11.745689655172415, | |
| "grad_norm": 0.6746819019317627, | |
| "learning_rate": 2.2973684210526316e-06, | |
| "loss": 0.2425, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 11.85344827586207, | |
| "grad_norm": 0.7924005389213562, | |
| "learning_rate": 2.2894736842105263e-06, | |
| "loss": 0.245, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 11.961206896551724, | |
| "grad_norm": 0.8761442303657532, | |
| "learning_rate": 2.281578947368421e-06, | |
| "loss": 0.253, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 12.068965517241379, | |
| "grad_norm": 0.6118531823158264, | |
| "learning_rate": 2.273684210526316e-06, | |
| "loss": 0.2492, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 12.176724137931034, | |
| "grad_norm": 0.698029637336731, | |
| "learning_rate": 2.2657894736842106e-06, | |
| "loss": 0.2538, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 12.28448275862069, | |
| "grad_norm": 0.6611655354499817, | |
| "learning_rate": 2.2578947368421053e-06, | |
| "loss": 0.2442, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 12.392241379310345, | |
| "grad_norm": 0.7785851955413818, | |
| "learning_rate": 2.25e-06, | |
| "loss": 0.2592, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "grad_norm": 0.7249414920806885, | |
| "learning_rate": 2.242105263157895e-06, | |
| "loss": 0.2394, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 12.607758620689655, | |
| "grad_norm": 0.8529052734375, | |
| "learning_rate": 2.2342105263157895e-06, | |
| "loss": 0.2353, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 12.71551724137931, | |
| "grad_norm": 0.825858473777771, | |
| "learning_rate": 2.2263157894736843e-06, | |
| "loss": 0.2391, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 12.823275862068966, | |
| "grad_norm": 0.9452285170555115, | |
| "learning_rate": 2.218421052631579e-06, | |
| "loss": 0.2522, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 12.931034482758621, | |
| "grad_norm": 0.744655430316925, | |
| "learning_rate": 2.2105263157894738e-06, | |
| "loss": 0.2474, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 12.931034482758621, | |
| "eval_loss": 0.270245224237442, | |
| "eval_runtime": 12887.0574, | |
| "eval_samples_per_second": 0.079, | |
| "eval_steps_per_second": 0.01, | |
| "eval_wer": 45.25119555715534, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 13.038793103448276, | |
| "grad_norm": 0.6413202285766602, | |
| "learning_rate": 2.2026315789473685e-06, | |
| "loss": 0.2392, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 13.14655172413793, | |
| "grad_norm": 1.1068989038467407, | |
| "learning_rate": 2.1947368421052633e-06, | |
| "loss": 0.2394, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 13.254310344827585, | |
| "grad_norm": 0.7797178030014038, | |
| "learning_rate": 2.186842105263158e-06, | |
| "loss": 0.2501, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 13.362068965517242, | |
| "grad_norm": 0.7657078504562378, | |
| "learning_rate": 2.1789473684210528e-06, | |
| "loss": 0.2466, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 13.469827586206897, | |
| "grad_norm": 0.6744815111160278, | |
| "learning_rate": 2.1710526315789475e-06, | |
| "loss": 0.2302, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 13.577586206896552, | |
| "grad_norm": 0.662436842918396, | |
| "learning_rate": 2.1631578947368423e-06, | |
| "loss": 0.2485, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 13.685344827586206, | |
| "grad_norm": 0.5807084441184998, | |
| "learning_rate": 2.155263157894737e-06, | |
| "loss": 0.2434, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 13.793103448275861, | |
| "grad_norm": 0.7120242118835449, | |
| "learning_rate": 2.1473684210526317e-06, | |
| "loss": 0.2397, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 13.900862068965518, | |
| "grad_norm": 0.7638838291168213, | |
| "learning_rate": 2.1394736842105265e-06, | |
| "loss": 0.235, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 14.008620689655173, | |
| "grad_norm": 0.6631995439529419, | |
| "learning_rate": 2.1315789473684212e-06, | |
| "loss": 0.2431, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 14.116379310344827, | |
| "grad_norm": 0.7738016247749329, | |
| "learning_rate": 2.123684210526316e-06, | |
| "loss": 0.2463, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 14.224137931034482, | |
| "grad_norm": 0.6762473583221436, | |
| "learning_rate": 2.1157894736842103e-06, | |
| "loss": 0.2333, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 14.331896551724139, | |
| "grad_norm": 0.5386962890625, | |
| "learning_rate": 2.107894736842105e-06, | |
| "loss": 0.2224, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 14.439655172413794, | |
| "grad_norm": 0.7853811383247375, | |
| "learning_rate": 2.1e-06, | |
| "loss": 0.2522, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 14.547413793103448, | |
| "grad_norm": 0.849454402923584, | |
| "learning_rate": 2.0921052631578945e-06, | |
| "loss": 0.2288, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 14.655172413793103, | |
| "grad_norm": 0.8840324878692627, | |
| "learning_rate": 2.0842105263157897e-06, | |
| "loss": 0.2347, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 14.762931034482758, | |
| "grad_norm": 0.753038227558136, | |
| "learning_rate": 2.0763157894736845e-06, | |
| "loss": 0.2346, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 14.870689655172415, | |
| "grad_norm": 0.5537305474281311, | |
| "learning_rate": 2.068421052631579e-06, | |
| "loss": 0.2377, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 14.97844827586207, | |
| "grad_norm": 0.7764011025428772, | |
| "learning_rate": 2.060526315789474e-06, | |
| "loss": 0.2396, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 15.086206896551724, | |
| "grad_norm": 0.8520766496658325, | |
| "learning_rate": 2.0526315789473687e-06, | |
| "loss": 0.2345, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 15.193965517241379, | |
| "grad_norm": 0.6806856989860535, | |
| "learning_rate": 2.0447368421052634e-06, | |
| "loss": 0.2323, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 15.301724137931034, | |
| "grad_norm": 0.8488845825195312, | |
| "learning_rate": 2.0368421052631578e-06, | |
| "loss": 0.2315, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 15.40948275862069, | |
| "grad_norm": 0.7247233390808105, | |
| "learning_rate": 2.0289473684210525e-06, | |
| "loss": 0.226, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 15.517241379310345, | |
| "grad_norm": 0.9292821288108826, | |
| "learning_rate": 2.0210526315789473e-06, | |
| "loss": 0.2517, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 15.625, | |
| "grad_norm": 0.7515670657157898, | |
| "learning_rate": 2.013157894736842e-06, | |
| "loss": 0.2407, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 15.732758620689655, | |
| "grad_norm": 0.6184987425804138, | |
| "learning_rate": 2.0052631578947367e-06, | |
| "loss": 0.2267, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 15.84051724137931, | |
| "grad_norm": 0.9107720255851746, | |
| "learning_rate": 1.9973684210526315e-06, | |
| "loss": 0.2251, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 15.948275862068966, | |
| "grad_norm": 0.6925145387649536, | |
| "learning_rate": 1.9894736842105262e-06, | |
| "loss": 0.2433, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 16.05603448275862, | |
| "grad_norm": 0.8509221076965332, | |
| "learning_rate": 1.9815789473684214e-06, | |
| "loss": 0.2255, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 16.163793103448278, | |
| "grad_norm": 0.835840106010437, | |
| "learning_rate": 1.973684210526316e-06, | |
| "loss": 0.24, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 16.271551724137932, | |
| "grad_norm": 0.8754560351371765, | |
| "learning_rate": 1.965789473684211e-06, | |
| "loss": 0.2315, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 16.379310344827587, | |
| "grad_norm": 0.8880627751350403, | |
| "learning_rate": 1.9578947368421052e-06, | |
| "loss": 0.2281, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 16.487068965517242, | |
| "grad_norm": 0.8073711395263672, | |
| "learning_rate": 1.95e-06, | |
| "loss": 0.2347, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 16.594827586206897, | |
| "grad_norm": 0.839345395565033, | |
| "learning_rate": 1.9421052631578947e-06, | |
| "loss": 0.2295, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 16.70258620689655, | |
| "grad_norm": 0.8276099562644958, | |
| "learning_rate": 1.9342105263157895e-06, | |
| "loss": 0.2331, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 16.810344827586206, | |
| "grad_norm": 0.9326604008674622, | |
| "learning_rate": 1.926315789473684e-06, | |
| "loss": 0.2257, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 16.91810344827586, | |
| "grad_norm": 0.751990020275116, | |
| "learning_rate": 1.918421052631579e-06, | |
| "loss": 0.2355, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 17.025862068965516, | |
| "grad_norm": 0.9207496643066406, | |
| "learning_rate": 1.9105263157894737e-06, | |
| "loss": 0.2289, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 17.13362068965517, | |
| "grad_norm": 0.8190683126449585, | |
| "learning_rate": 1.9026315789473684e-06, | |
| "loss": 0.2304, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 17.24137931034483, | |
| "grad_norm": 0.8016071319580078, | |
| "learning_rate": 1.8947368421052632e-06, | |
| "loss": 0.2292, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 17.24137931034483, | |
| "eval_loss": 0.26053452491760254, | |
| "eval_runtime": 12992.487, | |
| "eval_samples_per_second": 0.078, | |
| "eval_steps_per_second": 0.01, | |
| "eval_wer": 44.0582094924667, | |
| "step": 8000 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 20000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 44, | |
| "save_steps": 2000, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.1975917395968e+20, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |