| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.1595576619273302, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01579778830963665, | |
| "grad_norm": 39.78215789794922, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 1.9397, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0315955766192733, | |
| "grad_norm": 27.938426971435547, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 1.2937, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04739336492890995, | |
| "grad_norm": 25.550430297851562, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.9786, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0631911532385466, | |
| "grad_norm": 26.635549545288086, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.8818, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07898894154818326, | |
| "grad_norm": 25.53322410583496, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.8233, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0947867298578199, | |
| "grad_norm": 26.468778610229492, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.7945, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11058451816745656, | |
| "grad_norm": 22.71087074279785, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.752, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1263823064770932, | |
| "grad_norm": 21.494964599609375, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.7655, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14218009478672985, | |
| "grad_norm": 22.558645248413086, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.72, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1579778830963665, | |
| "grad_norm": 22.013551712036133, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.6961, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.17377567140600317, | |
| "grad_norm": 25.887876510620117, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.6897, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.1895734597156398, | |
| "grad_norm": 19.806230545043945, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.6839, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.20537124802527645, | |
| "grad_norm": 21.767576217651367, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.6767, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2211690363349131, | |
| "grad_norm": 19.838890075683594, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.6637, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.23696682464454977, | |
| "grad_norm": 22.119140625, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.6811, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2527646129541864, | |
| "grad_norm": 21.972688674926758, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.6843, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2685624012638231, | |
| "grad_norm": 21.99839973449707, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.6487, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.2843601895734597, | |
| "grad_norm": 18.968303680419922, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.6528, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.3001579778830964, | |
| "grad_norm": 20.77776336669922, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.6346, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.315955766192733, | |
| "grad_norm": 17.947195053100586, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.6105, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.33175355450236965, | |
| "grad_norm": 17.2167911529541, | |
| "learning_rate": 9.94888888888889e-06, | |
| "loss": 0.6399, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.34755134281200634, | |
| "grad_norm": 19.390045166015625, | |
| "learning_rate": 9.893333333333334e-06, | |
| "loss": 0.5825, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.36334913112164297, | |
| "grad_norm": 17.195106506347656, | |
| "learning_rate": 9.837777777777778e-06, | |
| "loss": 0.5895, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.3791469194312796, | |
| "grad_norm": 18.29102325439453, | |
| "learning_rate": 9.782222222222222e-06, | |
| "loss": 0.621, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3949447077409163, | |
| "grad_norm": 16.213546752929688, | |
| "learning_rate": 9.726666666666668e-06, | |
| "loss": 0.587, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.4107424960505529, | |
| "grad_norm": 15.86738109588623, | |
| "learning_rate": 9.671111111111112e-06, | |
| "loss": 0.5765, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4265402843601896, | |
| "grad_norm": 17.446897506713867, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.5914, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.4423380726698262, | |
| "grad_norm": 15.107172966003418, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.5517, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.45813586097946285, | |
| "grad_norm": 16.07261848449707, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.5523, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.47393364928909953, | |
| "grad_norm": 15.2976655960083, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.5419, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.48973143759873616, | |
| "grad_norm": 16.560869216918945, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.5763, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5055292259083728, | |
| "grad_norm": 18.58966064453125, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.5497, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5213270142180095, | |
| "grad_norm": 16.940420150756836, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.5177, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.5371248025276462, | |
| "grad_norm": 12.964641571044922, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.5293, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5529225908372828, | |
| "grad_norm": 16.428470611572266, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.5062, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.5687203791469194, | |
| "grad_norm": 17.921024322509766, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.5471, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.584518167456556, | |
| "grad_norm": 14.068251609802246, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.5072, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.6003159557661928, | |
| "grad_norm": 13.403594017028809, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.5051, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6161137440758294, | |
| "grad_norm": 14.557646751403809, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.5391, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.631911532385466, | |
| "grad_norm": 15.385436058044434, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.5037, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.631911532385466, | |
| "eval_loss": 0.551193118095398, | |
| "eval_runtime": 1363.4745, | |
| "eval_samples_per_second": 2.861, | |
| "eval_steps_per_second": 0.179, | |
| "eval_wer": 0.38095238095238093, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6477093206951027, | |
| "grad_norm": 13.815781593322754, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.4888, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.6635071090047393, | |
| "grad_norm": 16.213642120361328, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.4865, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.6793048973143759, | |
| "grad_norm": 14.779074668884277, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.4963, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.6951026856240127, | |
| "grad_norm": 13.794785499572754, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.482, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7109004739336493, | |
| "grad_norm": 14.701066970825195, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.5027, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.7266982622432859, | |
| "grad_norm": 14.537113189697266, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.4798, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.7424960505529226, | |
| "grad_norm": 14.886212348937988, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.4777, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.7582938388625592, | |
| "grad_norm": 13.01016616821289, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.4736, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.7740916271721959, | |
| "grad_norm": 14.213628768920898, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.4808, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.7898894154818326, | |
| "grad_norm": 14.541191101074219, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.5008, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.8056872037914692, | |
| "grad_norm": 14.092310905456543, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.4693, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.8214849921011058, | |
| "grad_norm": 14.289324760437012, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.4418, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.8372827804107424, | |
| "grad_norm": 13.257657051086426, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.4643, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.8530805687203792, | |
| "grad_norm": 15.05517864227295, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.4751, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.8688783570300158, | |
| "grad_norm": 13.352594375610352, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.442, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.8846761453396524, | |
| "grad_norm": 12.487988471984863, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.4496, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.9004739336492891, | |
| "grad_norm": 13.963912963867188, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.4415, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.9162717219589257, | |
| "grad_norm": 13.90829086303711, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.4349, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.9320695102685624, | |
| "grad_norm": 16.863481521606445, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.4722, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.9478672985781991, | |
| "grad_norm": 13.991209983825684, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.4397, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.9636650868878357, | |
| "grad_norm": 12.423737525939941, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.423, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.9794628751974723, | |
| "grad_norm": 13.574849128723145, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.4472, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.995260663507109, | |
| "grad_norm": 10.542879104614258, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.4319, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.0110584518167456, | |
| "grad_norm": 12.750446319580078, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.3401, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.0268562401263823, | |
| "grad_norm": 10.037585258483887, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.3236, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.042654028436019, | |
| "grad_norm": 12.519506454467773, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.3379, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.0584518167456556, | |
| "grad_norm": 11.666909217834473, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.3082, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.0742496050552923, | |
| "grad_norm": 14.614953994750977, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.3272, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.0900473933649288, | |
| "grad_norm": 11.243821144104004, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.3081, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.1058451816745656, | |
| "grad_norm": 10.896337509155273, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.3212, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.1216429699842023, | |
| "grad_norm": 11.159407615661621, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.3316, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.1374407582938388, | |
| "grad_norm": 10.72888469696045, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.3465, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.1532385466034756, | |
| "grad_norm": 11.136720657348633, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.3219, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.169036334913112, | |
| "grad_norm": 11.062825202941895, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.3234, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.1848341232227488, | |
| "grad_norm": 13.367472648620605, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.3215, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.2006319115323856, | |
| "grad_norm": 12.034098625183105, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.3195, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.216429699842022, | |
| "grad_norm": 12.74405574798584, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.3105, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.2322274881516588, | |
| "grad_norm": 14.234502792358398, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.3327, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.2480252764612954, | |
| "grad_norm": 12.721147537231445, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.3405, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.263823064770932, | |
| "grad_norm": 12.109272003173828, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.3264, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.263823064770932, | |
| "eval_loss": 0.4630681872367859, | |
| "eval_runtime": 1327.3738, | |
| "eval_samples_per_second": 2.939, | |
| "eval_steps_per_second": 0.184, | |
| "eval_wer": 0.331425673717763, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.2796208530805688, | |
| "grad_norm": 10.781163215637207, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.3152, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.2954186413902053, | |
| "grad_norm": 12.081149101257324, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.334, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.311216429699842, | |
| "grad_norm": 11.082480430603027, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.3182, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.3270142180094786, | |
| "grad_norm": 10.400703430175781, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.3209, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.3428120063191153, | |
| "grad_norm": 14.698405265808105, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.3358, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.358609794628752, | |
| "grad_norm": 11.301855087280273, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.3274, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.3744075829383886, | |
| "grad_norm": 13.15268611907959, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.3091, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.3902053712480253, | |
| "grad_norm": 10.712764739990234, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.3217, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.4060031595576619, | |
| "grad_norm": 9.865320205688477, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.3341, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.4218009478672986, | |
| "grad_norm": 11.386091232299805, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.308, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.4375987361769353, | |
| "grad_norm": 10.972213745117188, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.3184, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.4533965244865719, | |
| "grad_norm": 9.059267044067383, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.3285, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.4691943127962086, | |
| "grad_norm": 10.708878517150879, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.288, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.4849921011058451, | |
| "grad_norm": 9.580412864685059, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.3133, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.5007898894154819, | |
| "grad_norm": 13.611145973205566, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.3398, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.5165876777251186, | |
| "grad_norm": 11.563826560974121, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.3187, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.5323854660347551, | |
| "grad_norm": 11.612549781799316, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.3169, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.5481832543443916, | |
| "grad_norm": 9.463521003723145, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.3124, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.5639810426540284, | |
| "grad_norm": 14.442208290100098, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.3202, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.5797788309636651, | |
| "grad_norm": 11.377900123596191, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.317, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.5955766192733019, | |
| "grad_norm": 11.9774808883667, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.3188, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.6113744075829384, | |
| "grad_norm": 11.492107391357422, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.3109, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.627172195892575, | |
| "grad_norm": 12.987119674682617, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.3127, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.6429699842022116, | |
| "grad_norm": 11.403392791748047, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.3007, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.6587677725118484, | |
| "grad_norm": 10.354674339294434, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.3129, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.674565560821485, | |
| "grad_norm": 9.936201095581055, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.3226, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.6903633491311216, | |
| "grad_norm": 10.426243782043457, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.3021, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.7061611374407581, | |
| "grad_norm": 10.784858703613281, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.2923, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.7219589257503949, | |
| "grad_norm": 10.411828994750977, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.2893, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.7377567140600316, | |
| "grad_norm": 12.027934074401855, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.3136, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.7535545023696684, | |
| "grad_norm": 10.067774772644043, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.296, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.7693522906793049, | |
| "grad_norm": 10.396674156188965, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.2973, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.7851500789889414, | |
| "grad_norm": 9.719764709472656, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.2831, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.8009478672985781, | |
| "grad_norm": 11.552470207214355, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.2802, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.8167456556082149, | |
| "grad_norm": 10.932677268981934, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.3189, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.8325434439178516, | |
| "grad_norm": 12.281967163085938, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.3014, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.8483412322274881, | |
| "grad_norm": 12.78361988067627, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.3265, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.8641390205371247, | |
| "grad_norm": 11.523568153381348, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.3062, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.8799368088467614, | |
| "grad_norm": 12.009855270385742, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.294, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.8957345971563981, | |
| "grad_norm": 11.980591773986816, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.2997, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.8957345971563981, | |
| "eval_loss": 0.420254111289978, | |
| "eval_runtime": 1304.4467, | |
| "eval_samples_per_second": 2.991, | |
| "eval_steps_per_second": 0.187, | |
| "eval_wer": 0.30416304452815607, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.9115323854660349, | |
| "grad_norm": 9.556490898132324, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.3128, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.9273301737756714, | |
| "grad_norm": 12.445517539978027, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.2898, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.943127962085308, | |
| "grad_norm": 11.485321998596191, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.2922, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.9589257503949447, | |
| "grad_norm": 11.148347854614258, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.2941, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.9747235387045814, | |
| "grad_norm": 8.813661575317383, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.2738, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.9905213270142181, | |
| "grad_norm": 9.704155921936035, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.3064, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.006319115323855, | |
| "grad_norm": 8.704689979553223, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.2478, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.022116903633491, | |
| "grad_norm": 10.47323226928711, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.1972, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.037914691943128, | |
| "grad_norm": 9.401123046875, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.2048, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.0537124802527646, | |
| "grad_norm": 8.968587875366211, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.2047, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.0695102685624014, | |
| "grad_norm": 10.138334274291992, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.2041, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.085308056872038, | |
| "grad_norm": 9.157357215881348, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.1983, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.1011058451816744, | |
| "grad_norm": 11.541014671325684, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.2102, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.116903633491311, | |
| "grad_norm": 13.219682693481445, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.2032, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.132701421800948, | |
| "grad_norm": 9.11551284790039, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.183, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.1484992101105846, | |
| "grad_norm": 7.927098751068115, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.2022, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.1642969984202214, | |
| "grad_norm": 10.464715003967285, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.1877, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.1800947867298577, | |
| "grad_norm": 8.424384117126465, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.1879, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.1958925750394944, | |
| "grad_norm": 7.617179870605469, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.1973, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.211690363349131, | |
| "grad_norm": 9.299885749816895, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.1982, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.227488151658768, | |
| "grad_norm": 8.834092140197754, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.2003, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.2432859399684046, | |
| "grad_norm": 8.07299518585205, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.1971, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.259083728278041, | |
| "grad_norm": 10.275826454162598, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.1914, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.2748815165876777, | |
| "grad_norm": 9.910749435424805, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.2067, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.2906793048973144, | |
| "grad_norm": 10.053370475769043, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.2129, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.306477093206951, | |
| "grad_norm": 10.744956970214844, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.1984, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.322274881516588, | |
| "grad_norm": 9.880094528198242, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.1852, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.338072669826224, | |
| "grad_norm": 10.811684608459473, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.2019, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.353870458135861, | |
| "grad_norm": 12.169087409973145, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.1944, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.3696682464454977, | |
| "grad_norm": 10.1768217086792, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.2035, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.3854660347551344, | |
| "grad_norm": 8.389801979064941, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.2096, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.401263823064771, | |
| "grad_norm": 9.511481285095215, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.1868, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.4170616113744074, | |
| "grad_norm": 10.304895401000977, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.2023, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.432859399684044, | |
| "grad_norm": 11.822694778442383, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.1938, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.448657187993681, | |
| "grad_norm": 10.087789535522461, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.1987, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.4644549763033177, | |
| "grad_norm": 8.504409790039062, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.1845, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.4802527646129544, | |
| "grad_norm": 9.70301342010498, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.1832, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.4960505529225907, | |
| "grad_norm": 9.534614562988281, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.182, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.5118483412322274, | |
| "grad_norm": 8.997177124023438, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.205, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.527646129541864, | |
| "grad_norm": 9.527833938598633, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.1851, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.527646129541864, | |
| "eval_loss": 0.41017213463783264, | |
| "eval_runtime": 1287.8118, | |
| "eval_samples_per_second": 3.029, | |
| "eval_steps_per_second": 0.189, | |
| "eval_wer": 0.2870182555780933, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.543443917851501, | |
| "grad_norm": 8.116750717163086, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.1937, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.5592417061611377, | |
| "grad_norm": 11.475099563598633, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.1795, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.575039494470774, | |
| "grad_norm": 7.422430038452148, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.2029, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.5908372827804107, | |
| "grad_norm": 8.631953239440918, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.1883, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.6066350710900474, | |
| "grad_norm": 9.813713073730469, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.1901, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.622432859399684, | |
| "grad_norm": 10.284896850585938, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.183, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.638230647709321, | |
| "grad_norm": 9.403543472290039, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.1892, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.654028436018957, | |
| "grad_norm": 9.446948051452637, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.1886, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.669826224328594, | |
| "grad_norm": 10.583983421325684, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.2211, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.6856240126382307, | |
| "grad_norm": 10.528802871704102, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.1937, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.7014218009478674, | |
| "grad_norm": 8.71202278137207, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.1936, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.717219589257504, | |
| "grad_norm": 8.44046401977539, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.1994, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.7330173775671405, | |
| "grad_norm": 9.856565475463867, | |
| "learning_rate": 1.5088888888888889e-06, | |
| "loss": 0.1753, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.748815165876777, | |
| "grad_norm": 9.842414855957031, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.1815, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.764612954186414, | |
| "grad_norm": 9.333725929260254, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.1851, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.7804107424960507, | |
| "grad_norm": 9.97825813293457, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.1815, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.7962085308056874, | |
| "grad_norm": 9.474321365356445, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.1792, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.8120063191153237, | |
| "grad_norm": 8.71677303314209, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.1921, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.8278041074249605, | |
| "grad_norm": 9.69323444366455, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.1778, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.843601895734597, | |
| "grad_norm": 9.335270881652832, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.1783, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.859399684044234, | |
| "grad_norm": 8.661324501037598, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.1689, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.8751974723538707, | |
| "grad_norm": 9.32027530670166, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.1772, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.890995260663507, | |
| "grad_norm": 8.178330421447754, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.1806, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.9067930489731437, | |
| "grad_norm": 9.38011646270752, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.1753, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.9225908372827805, | |
| "grad_norm": 9.022958755493164, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.1813, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.938388625592417, | |
| "grad_norm": 9.93110466003418, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.1825, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.954186413902054, | |
| "grad_norm": 9.306452751159668, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.1943, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.9699842022116902, | |
| "grad_norm": 7.849850177764893, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.1696, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.985781990521327, | |
| "grad_norm": 10.920326232910156, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.1788, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 3.0015797788309637, | |
| "grad_norm": 7.569627285003662, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.1698, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.0173775671406005, | |
| "grad_norm": 8.557785987854004, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.1323, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 3.0331753554502368, | |
| "grad_norm": 7.7341132164001465, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.1288, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.0489731437598735, | |
| "grad_norm": 6.455957889556885, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.1253, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 3.0647709320695102, | |
| "grad_norm": 9.151886940002441, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.1369, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.080568720379147, | |
| "grad_norm": 8.11299991607666, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.1272, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 3.0963665086887837, | |
| "grad_norm": 6.720188617706299, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.1329, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.11216429699842, | |
| "grad_norm": 7.401381969451904, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.1257, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 3.1279620853080567, | |
| "grad_norm": 7.703917026519775, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.1224, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 3.1437598736176935, | |
| "grad_norm": 7.16009521484375, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.1296, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 3.1595576619273302, | |
| "grad_norm": 7.600194931030273, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.1313, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.1595576619273302, | |
| "eval_loss": 0.4071974754333496, | |
| "eval_runtime": 1289.8235, | |
| "eval_samples_per_second": 3.024, | |
| "eval_steps_per_second": 0.189, | |
| "eval_wer": 0.2801361924079977, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.1595576619273302, | |
| "step": 5000, | |
| "total_flos": 2.727921844224e+20, | |
| "train_loss": 0.3573408980369568, | |
| "train_runtime": 57234.8416, | |
| "train_samples_per_second": 2.796, | |
| "train_steps_per_second": 0.087 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.727921844224e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |