| { | |
| "best_metric": 0.21656855629922317, | |
| "best_model_checkpoint": "/cluster/home/torstefl/Master/saved_model/W2V/Combined/BB-NB-RUND-1b-ll/checkpoint-38778", | |
| "epoch": 23.0, | |
| "eval_steps": 500, | |
| "global_step": 38778, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 40.23420715332031, | |
| "learning_rate": 1.994419168941462e-05, | |
| "loss": 0.7914, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_bigbrother_loss": 1.3365620374679565, | |
| "eval_bigbrother_runtime": 55.0794, | |
| "eval_bigbrother_samples_per_second": 24.873, | |
| "eval_bigbrother_steps_per_second": 0.781, | |
| "eval_bigbrother_wer": 0.4995569161363087, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_BB_NB_RUND_loss": 0.7492167949676514, | |
| "eval_BB_NB_RUND_runtime": 219.5304, | |
| "eval_BB_NB_RUND_samples_per_second": 14.768, | |
| "eval_BB_NB_RUND_steps_per_second": 0.465, | |
| "eval_BB_NB_RUND_wer": 0.29118619466172085, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_rundkast_loss": 0.2676470875740051, | |
| "eval_rundkast_runtime": 39.1567, | |
| "eval_rundkast_samples_per_second": 34.196, | |
| "eval_rundkast_steps_per_second": 1.073, | |
| "eval_rundkast_wer": 0.1707726763717805, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_nb_samtale_loss": 0.44947025179862976, | |
| "eval_nb_samtale_runtime": 46.7998, | |
| "eval_nb_samtale_samples_per_second": 11.389, | |
| "eval_nb_samtale_steps_per_second": 0.363, | |
| "eval_nb_samtale_wer": 0.20320638020833334, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 34.448326110839844, | |
| "learning_rate": 1.943312101910828e-05, | |
| "loss": 0.6845, | |
| "step": 3372 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bigbrother_loss": 1.3286552429199219, | |
| "eval_bigbrother_runtime": 56.6035, | |
| "eval_bigbrother_samples_per_second": 24.203, | |
| "eval_bigbrother_steps_per_second": 0.76, | |
| "eval_bigbrother_wer": 0.4742608555546604, | |
| "step": 3372 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_BB_NB_RUND_loss": 0.7409276366233826, | |
| "eval_BB_NB_RUND_runtime": 208.4781, | |
| "eval_BB_NB_RUND_samples_per_second": 15.551, | |
| "eval_BB_NB_RUND_steps_per_second": 0.489, | |
| "eval_BB_NB_RUND_wer": 0.27532725855441764, | |
| "step": 3372 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_rundkast_loss": 0.25520867109298706, | |
| "eval_rundkast_runtime": 40.2745, | |
| "eval_rundkast_samples_per_second": 33.247, | |
| "eval_rundkast_steps_per_second": 1.043, | |
| "eval_rundkast_wer": 0.15853463445848665, | |
| "step": 3372 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_nb_samtale_loss": 0.45029357075691223, | |
| "eval_nb_samtale_runtime": 48.0222, | |
| "eval_nb_samtale_samples_per_second": 11.099, | |
| "eval_nb_samtale_steps_per_second": 0.354, | |
| "eval_nb_samtale_wer": 0.19246419270833334, | |
| "step": 3372 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 10.3843355178833, | |
| "learning_rate": 1.8921747042766152e-05, | |
| "loss": 0.6323, | |
| "step": 5058 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bigbrother_loss": 1.217179775238037, | |
| "eval_bigbrother_runtime": 56.073, | |
| "eval_bigbrother_samples_per_second": 24.432, | |
| "eval_bigbrother_steps_per_second": 0.767, | |
| "eval_bigbrother_wer": 0.44896479497301217, | |
| "step": 5058 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_BB_NB_RUND_loss": 0.6835288405418396, | |
| "eval_BB_NB_RUND_runtime": 193.4829, | |
| "eval_BB_NB_RUND_samples_per_second": 16.756, | |
| "eval_BB_NB_RUND_steps_per_second": 0.527, | |
| "eval_BB_NB_RUND_wer": 0.2632583393812327, | |
| "step": 5058 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_rundkast_loss": 0.2418593019247055, | |
| "eval_rundkast_runtime": 40.4751, | |
| "eval_rundkast_samples_per_second": 33.082, | |
| "eval_rundkast_steps_per_second": 1.038, | |
| "eval_rundkast_wer": 0.15301551751719725, | |
| "step": 5058 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_nb_samtale_loss": 0.42157283425331116, | |
| "eval_nb_samtale_runtime": 47.3972, | |
| "eval_nb_samtale_samples_per_second": 11.245, | |
| "eval_nb_samtale_steps_per_second": 0.359, | |
| "eval_nb_samtale_wer": 0.18790690104166666, | |
| "step": 5058 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 17.625219345092773, | |
| "learning_rate": 1.8410979678495603e-05, | |
| "loss": 0.5949, | |
| "step": 6744 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bigbrother_loss": 1.384012222290039, | |
| "eval_bigbrother_runtime": 55.0197, | |
| "eval_bigbrother_samples_per_second": 24.9, | |
| "eval_bigbrother_steps_per_second": 0.782, | |
| "eval_bigbrother_wer": 0.44872311286554417, | |
| "step": 6744 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_BB_NB_RUND_loss": 0.7864356637001038, | |
| "eval_BB_NB_RUND_runtime": 181.8859, | |
| "eval_BB_NB_RUND_samples_per_second": 17.824, | |
| "eval_BB_NB_RUND_steps_per_second": 0.561, | |
| "eval_BB_NB_RUND_wer": 0.25925328602532055, | |
| "step": 6744 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_rundkast_loss": 0.3103683590888977, | |
| "eval_rundkast_runtime": 40.029, | |
| "eval_rundkast_samples_per_second": 33.451, | |
| "eval_rundkast_steps_per_second": 1.049, | |
| "eval_rundkast_wer": 0.14637657974724044, | |
| "step": 6744 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_nb_samtale_loss": 0.4464314579963684, | |
| "eval_nb_samtale_runtime": 47.5761, | |
| "eval_nb_samtale_samples_per_second": 11.203, | |
| "eval_nb_samtale_steps_per_second": 0.357, | |
| "eval_nb_samtale_wer": 0.182861328125, | |
| "step": 6744 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 14.08320426940918, | |
| "learning_rate": 1.7899605702153475e-05, | |
| "loss": 0.5659, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bigbrother_loss": 1.3207027912139893, | |
| "eval_bigbrother_runtime": 55.6797, | |
| "eval_bigbrother_samples_per_second": 24.605, | |
| "eval_bigbrother_steps_per_second": 0.772, | |
| "eval_bigbrother_wer": 0.4434061065012487, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_BB_NB_RUND_loss": 0.7573540210723877, | |
| "eval_BB_NB_RUND_runtime": 179.8263, | |
| "eval_BB_NB_RUND_samples_per_second": 18.029, | |
| "eval_BB_NB_RUND_steps_per_second": 0.567, | |
| "eval_BB_NB_RUND_wer": 0.2573448377818993, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_rundkast_loss": 0.31218844652175903, | |
| "eval_rundkast_runtime": 40.0113, | |
| "eval_rundkast_samples_per_second": 33.466, | |
| "eval_rundkast_steps_per_second": 1.05, | |
| "eval_rundkast_wer": 0.1435770276755719, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_nb_samtale_loss": 0.4283309280872345, | |
| "eval_nb_samtale_runtime": 47.4461, | |
| "eval_nb_samtale_samples_per_second": 11.234, | |
| "eval_nb_samtale_steps_per_second": 0.358, | |
| "eval_nb_samtale_wer": 0.18497721354166666, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 20.013416290283203, | |
| "learning_rate": 1.7388231725811344e-05, | |
| "loss": 1.6171144283853038e+30, | |
| "step": 10116 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bigbrother_loss": 1.1751562356948853, | |
| "eval_bigbrother_runtime": 55.6865, | |
| "eval_bigbrother_samples_per_second": 24.602, | |
| "eval_bigbrother_steps_per_second": 0.772, | |
| "eval_bigbrother_wer": 0.4262466768710223, | |
| "step": 10116 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_BB_NB_RUND_loss": 0.6712942719459534, | |
| "eval_BB_NB_RUND_runtime": 186.6214, | |
| "eval_BB_NB_RUND_samples_per_second": 17.372, | |
| "eval_BB_NB_RUND_steps_per_second": 0.547, | |
| "eval_BB_NB_RUND_wer": 0.25070558825901135, | |
| "step": 10116 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_rundkast_loss": 0.26327571272850037, | |
| "eval_rundkast_runtime": 40.3853, | |
| "eval_rundkast_samples_per_second": 33.156, | |
| "eval_rundkast_steps_per_second": 1.04, | |
| "eval_rundkast_wer": 0.14549672052471604, | |
| "step": 10116 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_nb_samtale_loss": 0.4015786945819855, | |
| "eval_nb_samtale_runtime": 48.2988, | |
| "eval_nb_samtale_samples_per_second": 11.035, | |
| "eval_nb_samtale_steps_per_second": 0.352, | |
| "eval_nb_samtale_wer": 0.18123372395833334, | |
| "step": 10116 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 11.362643241882324, | |
| "learning_rate": 1.6877161055505007e-05, | |
| "loss": 0.5188, | |
| "step": 11802 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_bigbrother_loss": 1.2031224966049194, | |
| "eval_bigbrother_runtime": 56.0574, | |
| "eval_bigbrother_samples_per_second": 24.439, | |
| "eval_bigbrother_steps_per_second": 0.767, | |
| "eval_bigbrother_wer": 0.4243937807137678, | |
| "step": 11802 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_BB_NB_RUND_loss": 0.6787930130958557, | |
| "eval_BB_NB_RUND_runtime": 188.339, | |
| "eval_BB_NB_RUND_samples_per_second": 17.214, | |
| "eval_BB_NB_RUND_steps_per_second": 0.542, | |
| "eval_BB_NB_RUND_wer": 0.24680805311399617, | |
| "step": 11802 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_rundkast_loss": 0.2649306356906891, | |
| "eval_rundkast_runtime": 40.4917, | |
| "eval_rundkast_samples_per_second": 33.069, | |
| "eval_rundkast_steps_per_second": 1.037, | |
| "eval_rundkast_wer": 0.14077747560390338, | |
| "step": 11802 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_nb_samtale_loss": 0.37048718333244324, | |
| "eval_nb_samtale_runtime": 48.165, | |
| "eval_nb_samtale_samples_per_second": 11.066, | |
| "eval_nb_samtale_steps_per_second": 0.353, | |
| "eval_nb_samtale_wer": 0.17569986979166666, | |
| "step": 11802 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 14.940473556518555, | |
| "learning_rate": 1.6366090385198667e-05, | |
| "loss": 0.4948, | |
| "step": 13488 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_bigbrother_loss": 1.2190874814987183, | |
| "eval_bigbrother_runtime": 56.6486, | |
| "eval_bigbrother_samples_per_second": 24.184, | |
| "eval_bigbrother_steps_per_second": 0.759, | |
| "eval_bigbrother_wer": 0.41134294691049705, | |
| "step": 13488 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_BB_NB_RUND_loss": 0.7082846164703369, | |
| "eval_BB_NB_RUND_runtime": 187.8309, | |
| "eval_BB_NB_RUND_samples_per_second": 17.26, | |
| "eval_BB_NB_RUND_steps_per_second": 0.543, | |
| "eval_BB_NB_RUND_wer": 0.24280299975808403, | |
| "step": 13488 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_rundkast_loss": 0.3050954043865204, | |
| "eval_rundkast_runtime": 40.703, | |
| "eval_rundkast_samples_per_second": 32.897, | |
| "eval_rundkast_steps_per_second": 1.032, | |
| "eval_rundkast_wer": 0.1406974884018557, | |
| "step": 13488 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_nb_samtale_loss": 0.4080573618412018, | |
| "eval_nb_samtale_runtime": 48.4037, | |
| "eval_nb_samtale_samples_per_second": 11.012, | |
| "eval_nb_samtale_steps_per_second": 0.351, | |
| "eval_nb_samtale_wer": 0.176513671875, | |
| "step": 13488 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 14.726886749267578, | |
| "learning_rate": 1.585471640885654e-05, | |
| "loss": 0.4819, | |
| "step": 15174 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_bigbrother_loss": 1.1392048597335815, | |
| "eval_bigbrother_runtime": 56.0533, | |
| "eval_bigbrother_samples_per_second": 24.441, | |
| "eval_bigbrother_steps_per_second": 0.767, | |
| "eval_bigbrother_wer": 0.4111012648030291, | |
| "step": 15174 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_BB_NB_RUND_loss": 0.6559087634086609, | |
| "eval_BB_NB_RUND_runtime": 178.4738, | |
| "eval_BB_NB_RUND_samples_per_second": 18.165, | |
| "eval_BB_NB_RUND_steps_per_second": 0.572, | |
| "eval_BB_NB_RUND_wer": 0.24307179528532646, | |
| "step": 15174 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_rundkast_loss": 0.2665271759033203, | |
| "eval_rundkast_runtime": 40.869, | |
| "eval_rundkast_samples_per_second": 32.763, | |
| "eval_rundkast_steps_per_second": 1.028, | |
| "eval_rundkast_wer": 0.13973764197728364, | |
| "step": 15174 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_nb_samtale_loss": 0.39354923367500305, | |
| "eval_nb_samtale_runtime": 48.6692, | |
| "eval_nb_samtale_samples_per_second": 10.951, | |
| "eval_nb_samtale_steps_per_second": 0.349, | |
| "eval_nb_samtale_wer": 0.178466796875, | |
| "step": 15174 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 28.40872573852539, | |
| "learning_rate": 1.53436457385502e-05, | |
| "loss": 0.4676, | |
| "step": 16860 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_bigbrother_loss": 1.2027722597122192, | |
| "eval_bigbrother_runtime": 56.4265, | |
| "eval_bigbrother_samples_per_second": 24.279, | |
| "eval_bigbrother_steps_per_second": 0.762, | |
| "eval_bigbrother_wer": 0.3986143559171836, | |
| "step": 16860 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_BB_NB_RUND_loss": 0.7014443278312683, | |
| "eval_BB_NB_RUND_runtime": 184.3191, | |
| "eval_BB_NB_RUND_samples_per_second": 17.589, | |
| "eval_BB_NB_RUND_steps_per_second": 0.553, | |
| "eval_BB_NB_RUND_wer": 0.2354648818643658, | |
| "step": 16860 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_rundkast_loss": 0.3014816343784332, | |
| "eval_rundkast_runtime": 40.6222, | |
| "eval_rundkast_samples_per_second": 32.962, | |
| "eval_rundkast_steps_per_second": 1.034, | |
| "eval_rundkast_wer": 0.13389857622780355, | |
| "step": 16860 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_nb_samtale_loss": 0.4174867272377014, | |
| "eval_nb_samtale_runtime": 50.8259, | |
| "eval_nb_samtale_samples_per_second": 10.487, | |
| "eval_nb_samtale_steps_per_second": 0.334, | |
| "eval_nb_samtale_wer": 0.174072265625, | |
| "step": 16860 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "grad_norm": 17.167695999145508, | |
| "learning_rate": 1.483257506824386e-05, | |
| "loss": 0.4529, | |
| "step": 18546 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_bigbrother_loss": 1.2520931959152222, | |
| "eval_bigbrother_runtime": 56.0621, | |
| "eval_bigbrother_samples_per_second": 24.437, | |
| "eval_bigbrother_steps_per_second": 0.767, | |
| "eval_bigbrother_wer": 0.4011117376943527, | |
| "step": 18546 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_BB_NB_RUND_loss": 0.7142683863639832, | |
| "eval_BB_NB_RUND_runtime": 188.0039, | |
| "eval_BB_NB_RUND_samples_per_second": 17.244, | |
| "eval_BB_NB_RUND_steps_per_second": 0.543, | |
| "eval_BB_NB_RUND_wer": 0.2372389323441658, | |
| "step": 18546 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_rundkast_loss": 0.2879612147808075, | |
| "eval_rundkast_runtime": 41.0453, | |
| "eval_rundkast_samples_per_second": 32.622, | |
| "eval_rundkast_steps_per_second": 1.023, | |
| "eval_rundkast_wer": 0.1372580387138058, | |
| "step": 18546 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_nb_samtale_loss": 0.4028375446796417, | |
| "eval_nb_samtale_runtime": 48.6427, | |
| "eval_nb_samtale_samples_per_second": 10.957, | |
| "eval_nb_samtale_steps_per_second": 0.349, | |
| "eval_nb_samtale_wer": 0.17342122395833334, | |
| "step": 18546 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "grad_norm": 10.578996658325195, | |
| "learning_rate": 1.432120109190173e-05, | |
| "loss": 0.4393, | |
| "step": 20232 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_bigbrother_loss": 1.1568479537963867, | |
| "eval_bigbrother_runtime": 56.7267, | |
| "eval_bigbrother_samples_per_second": 24.151, | |
| "eval_bigbrother_steps_per_second": 0.758, | |
| "eval_bigbrother_wer": 0.3946668814952066, | |
| "step": 20232 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_BB_NB_RUND_loss": 0.6711928844451904, | |
| "eval_BB_NB_RUND_runtime": 189.16, | |
| "eval_BB_NB_RUND_samples_per_second": 17.139, | |
| "eval_BB_NB_RUND_steps_per_second": 0.539, | |
| "eval_BB_NB_RUND_wer": 0.23304572211918392, | |
| "step": 20232 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_rundkast_loss": 0.281593918800354, | |
| "eval_rundkast_runtime": 40.9816, | |
| "eval_rundkast_samples_per_second": 32.673, | |
| "eval_rundkast_steps_per_second": 1.025, | |
| "eval_rundkast_wer": 0.13509838425851864, | |
| "step": 20232 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_nb_samtale_loss": 0.40170300006866455, | |
| "eval_nb_samtale_runtime": 48.5059, | |
| "eval_nb_samtale_samples_per_second": 10.988, | |
| "eval_nb_samtale_steps_per_second": 0.35, | |
| "eval_nb_samtale_wer": 0.17008463541666666, | |
| "step": 20232 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "grad_norm": 26.488574981689453, | |
| "learning_rate": 1.3810130421595392e-05, | |
| "loss": 0.4286, | |
| "step": 21918 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_bigbrother_loss": 1.211380958557129, | |
| "eval_bigbrother_runtime": 56.1397, | |
| "eval_bigbrother_samples_per_second": 24.403, | |
| "eval_bigbrother_steps_per_second": 0.766, | |
| "eval_bigbrother_wer": 0.39313622814790944, | |
| "step": 21918 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_BB_NB_RUND_loss": 0.7042691707611084, | |
| "eval_BB_NB_RUND_runtime": 192.1708, | |
| "eval_BB_NB_RUND_samples_per_second": 16.87, | |
| "eval_BB_NB_RUND_steps_per_second": 0.531, | |
| "eval_BB_NB_RUND_wer": 0.23051904416310512, | |
| "step": 21918 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_rundkast_loss": 0.2999264895915985, | |
| "eval_rundkast_runtime": 40.6906, | |
| "eval_rundkast_samples_per_second": 32.907, | |
| "eval_rundkast_steps_per_second": 1.032, | |
| "eval_rundkast_wer": 0.1297392417213246, | |
| "step": 21918 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_nb_samtale_loss": 0.41735681891441345, | |
| "eval_nb_samtale_runtime": 48.1494, | |
| "eval_nb_samtale_samples_per_second": 11.07, | |
| "eval_nb_samtale_steps_per_second": 0.353, | |
| "eval_nb_samtale_wer": 0.169189453125, | |
| "step": 21918 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "grad_norm": 8.799247741699219, | |
| "learning_rate": 1.3298756445253262e-05, | |
| "loss": 0.4158, | |
| "step": 23604 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_bigbrother_loss": 1.2186354398727417, | |
| "eval_bigbrother_runtime": 55.8345, | |
| "eval_bigbrother_samples_per_second": 24.537, | |
| "eval_bigbrother_steps_per_second": 0.77, | |
| "eval_bigbrother_wer": 0.39225006042052685, | |
| "step": 23604 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_BB_NB_RUND_loss": 0.7078685760498047, | |
| "eval_BB_NB_RUND_runtime": 188.634, | |
| "eval_BB_NB_RUND_samples_per_second": 17.187, | |
| "eval_BB_NB_RUND_steps_per_second": 0.541, | |
| "eval_BB_NB_RUND_wer": 0.23111039432303845, | |
| "step": 23604 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_rundkast_loss": 0.3022992014884949, | |
| "eval_rundkast_runtime": 40.9858, | |
| "eval_rundkast_samples_per_second": 32.67, | |
| "eval_rundkast_steps_per_second": 1.025, | |
| "eval_rundkast_wer": 0.1311790113581827, | |
| "step": 23604 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_nb_samtale_loss": 0.4138488471508026, | |
| "eval_nb_samtale_runtime": 48.2354, | |
| "eval_nb_samtale_samples_per_second": 11.05, | |
| "eval_nb_samtale_steps_per_second": 0.352, | |
| "eval_nb_samtale_wer": 0.16943359375, | |
| "step": 23604 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "grad_norm": 8.913851737976074, | |
| "learning_rate": 1.2787989080982713e-05, | |
| "loss": 0.4061, | |
| "step": 25290 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_bigbrother_loss": 1.2142188549041748, | |
| "eval_bigbrother_runtime": 56.7489, | |
| "eval_bigbrother_samples_per_second": 24.141, | |
| "eval_bigbrother_steps_per_second": 0.758, | |
| "eval_bigbrother_wer": 0.3879803431885926, | |
| "step": 25290 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_BB_NB_RUND_loss": 0.7142959833145142, | |
| "eval_BB_NB_RUND_runtime": 176.8357, | |
| "eval_BB_NB_RUND_samples_per_second": 18.333, | |
| "eval_BB_NB_RUND_steps_per_second": 0.577, | |
| "eval_BB_NB_RUND_wer": 0.2274816547052657, | |
| "step": 25290 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_rundkast_loss": 0.325125128030777, | |
| "eval_rundkast_runtime": 41.4419, | |
| "eval_rundkast_samples_per_second": 32.31, | |
| "eval_rundkast_steps_per_second": 1.013, | |
| "eval_rundkast_wer": 0.13093904975203968, | |
| "step": 25290 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_nb_samtale_loss": 0.4068756699562073, | |
| "eval_nb_samtale_runtime": 48.9343, | |
| "eval_nb_samtale_samples_per_second": 10.892, | |
| "eval_nb_samtale_steps_per_second": 0.347, | |
| "eval_nb_samtale_wer": 0.16357421875, | |
| "step": 25290 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "grad_norm": 16.82915496826172, | |
| "learning_rate": 1.2276615104640583e-05, | |
| "loss": 0.3964, | |
| "step": 26976 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_bigbrother_loss": 1.1660621166229248, | |
| "eval_bigbrother_runtime": 56.8428, | |
| "eval_bigbrother_samples_per_second": 24.102, | |
| "eval_bigbrother_steps_per_second": 0.756, | |
| "eval_bigbrother_wer": 0.38072987996455326, | |
| "step": 26976 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_BB_NB_RUND_loss": 0.6708107590675354, | |
| "eval_BB_NB_RUND_runtime": 211.8358, | |
| "eval_BB_NB_RUND_samples_per_second": 15.304, | |
| "eval_BB_NB_RUND_steps_per_second": 0.482, | |
| "eval_BB_NB_RUND_wer": 0.22318092626938688, | |
| "step": 26976 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_rundkast_loss": 0.2847665250301361, | |
| "eval_rundkast_runtime": 40.762, | |
| "eval_rundkast_samples_per_second": 32.849, | |
| "eval_rundkast_steps_per_second": 1.03, | |
| "eval_rundkast_wer": 0.126219804831227, | |
| "step": 26976 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_nb_samtale_loss": 0.3676682114601135, | |
| "eval_nb_samtale_runtime": 48.3802, | |
| "eval_nb_samtale_samples_per_second": 11.017, | |
| "eval_nb_samtale_steps_per_second": 0.351, | |
| "eval_nb_samtale_wer": 0.162353515625, | |
| "step": 26976 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "grad_norm": 12.924592018127441, | |
| "learning_rate": 1.1765544434334245e-05, | |
| "loss": 0.3839, | |
| "step": 28662 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_bigbrother_loss": 1.2347418069839478, | |
| "eval_bigbrother_runtime": 56.4815, | |
| "eval_bigbrother_samples_per_second": 24.256, | |
| "eval_bigbrother_steps_per_second": 0.761, | |
| "eval_bigbrother_wer": 0.3837911866591477, | |
| "step": 28662 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_BB_NB_RUND_loss": 0.7269824147224426, | |
| "eval_BB_NB_RUND_runtime": 198.7713, | |
| "eval_BB_NB_RUND_samples_per_second": 16.31, | |
| "eval_BB_NB_RUND_steps_per_second": 0.513, | |
| "eval_BB_NB_RUND_wer": 0.2254656882509475, | |
| "step": 28662 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_rundkast_loss": 0.33907437324523926, | |
| "eval_rundkast_runtime": 40.907, | |
| "eval_rundkast_samples_per_second": 32.733, | |
| "eval_rundkast_steps_per_second": 1.027, | |
| "eval_rundkast_wer": 0.12925931850903855, | |
| "step": 28662 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_nb_samtale_loss": 0.39654919505119324, | |
| "eval_nb_samtale_runtime": 48.4458, | |
| "eval_nb_samtale_samples_per_second": 11.002, | |
| "eval_nb_samtale_steps_per_second": 0.351, | |
| "eval_nb_samtale_wer": 0.163818359375, | |
| "step": 28662 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "grad_norm": 56.129154205322266, | |
| "learning_rate": 1.1254170457992115e-05, | |
| "loss": 0.3754, | |
| "step": 30348 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_bigbrother_loss": 1.2182375192642212, | |
| "eval_bigbrother_runtime": 56.0084, | |
| "eval_bigbrother_samples_per_second": 24.461, | |
| "eval_bigbrother_steps_per_second": 0.768, | |
| "eval_bigbrother_wer": 0.3763796020301297, | |
| "step": 30348 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_BB_NB_RUND_loss": 0.6986347436904907, | |
| "eval_BB_NB_RUND_runtime": 210.8249, | |
| "eval_BB_NB_RUND_samples_per_second": 15.378, | |
| "eval_BB_NB_RUND_steps_per_second": 0.484, | |
| "eval_BB_NB_RUND_wer": 0.22167567131682928, | |
| "step": 30348 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_rundkast_loss": 0.2828354239463806, | |
| "eval_rundkast_runtime": 40.4262, | |
| "eval_rundkast_samples_per_second": 33.122, | |
| "eval_rundkast_steps_per_second": 1.039, | |
| "eval_rundkast_wer": 0.12525995840665494, | |
| "step": 30348 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_nb_samtale_loss": 0.40772777795791626, | |
| "eval_nb_samtale_runtime": 47.7849, | |
| "eval_nb_samtale_samples_per_second": 11.154, | |
| "eval_nb_samtale_steps_per_second": 0.356, | |
| "eval_nb_samtale_wer": 0.16300455729166666, | |
| "step": 30348 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "grad_norm": 25.304649353027344, | |
| "learning_rate": 1.0743099787685776e-05, | |
| "loss": 0.366, | |
| "step": 32034 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_bigbrother_loss": 1.1887106895446777, | |
| "eval_bigbrother_runtime": 56.2869, | |
| "eval_bigbrother_samples_per_second": 24.34, | |
| "eval_bigbrother_steps_per_second": 0.764, | |
| "eval_bigbrother_wer": 0.3763796020301297, | |
| "step": 32034 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_BB_NB_RUND_loss": 0.7024869918823242, | |
| "eval_BB_NB_RUND_runtime": 194.4983, | |
| "eval_BB_NB_RUND_samples_per_second": 16.669, | |
| "eval_BB_NB_RUND_steps_per_second": 0.524, | |
| "eval_BB_NB_RUND_wer": 0.22148751444775958, | |
| "step": 32034 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_rundkast_loss": 0.3243892788887024, | |
| "eval_rundkast_runtime": 42.0254, | |
| "eval_rundkast_samples_per_second": 31.862, | |
| "eval_rundkast_steps_per_second": 0.999, | |
| "eval_rundkast_wer": 0.125979843225084, | |
| "step": 32034 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_nb_samtale_loss": 0.4026784598827362, | |
| "eval_nb_samtale_runtime": 47.8774, | |
| "eval_nb_samtale_samples_per_second": 11.133, | |
| "eval_nb_samtale_steps_per_second": 0.355, | |
| "eval_nb_samtale_wer": 0.162353515625, | |
| "step": 32034 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "grad_norm": 7.739073753356934, | |
| "learning_rate": 1.0231725811343647e-05, | |
| "loss": 0.3577, | |
| "step": 33720 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_bigbrother_loss": 1.2659293413162231, | |
| "eval_bigbrother_runtime": 55.6963, | |
| "eval_bigbrother_samples_per_second": 24.598, | |
| "eval_bigbrother_steps_per_second": 0.772, | |
| "eval_bigbrother_wer": 0.37339885603802464, | |
| "step": 33720 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_BB_NB_RUND_loss": 0.7312197685241699, | |
| "eval_BB_NB_RUND_runtime": 175.0676, | |
| "eval_BB_NB_RUND_samples_per_second": 18.519, | |
| "eval_BB_NB_RUND_steps_per_second": 0.583, | |
| "eval_BB_NB_RUND_wer": 0.2187189205171626, | |
| "step": 33720 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_rundkast_loss": 0.3108943998813629, | |
| "eval_rundkast_runtime": 40.3883, | |
| "eval_rundkast_samples_per_second": 33.153, | |
| "eval_rundkast_steps_per_second": 1.04, | |
| "eval_rundkast_wer": 0.12270036794112942, | |
| "step": 33720 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_nb_samtale_loss": 0.41297149658203125, | |
| "eval_nb_samtale_runtime": 48.2786, | |
| "eval_nb_samtale_samples_per_second": 11.04, | |
| "eval_nb_samtale_steps_per_second": 0.352, | |
| "eval_nb_samtale_wer": 0.16015625, | |
| "step": 33720 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "grad_norm": 11.251120567321777, | |
| "learning_rate": 9.720655141037308e-06, | |
| "loss": 0.3504, | |
| "step": 35406 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_bigbrother_loss": 1.274532437324524, | |
| "eval_bigbrother_runtime": 57.087, | |
| "eval_bigbrother_samples_per_second": 23.998, | |
| "eval_bigbrother_steps_per_second": 0.753, | |
| "eval_bigbrother_wer": 0.37275437041811005, | |
| "step": 35406 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_BB_NB_RUND_loss": 0.7424564957618713, | |
| "eval_BB_NB_RUND_runtime": 190.1044, | |
| "eval_BB_NB_RUND_samples_per_second": 17.054, | |
| "eval_BB_NB_RUND_steps_per_second": 0.537, | |
| "eval_BB_NB_RUND_wer": 0.22003601860065047, | |
| "step": 35406 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_rundkast_loss": 0.3286913335323334, | |
| "eval_rundkast_runtime": 40.9067, | |
| "eval_rundkast_samples_per_second": 32.733, | |
| "eval_rundkast_steps_per_second": 1.027, | |
| "eval_rundkast_wer": 0.12494000959846424, | |
| "step": 35406 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_nb_samtale_loss": 0.4145500361919403, | |
| "eval_nb_samtale_runtime": 48.4768, | |
| "eval_nb_samtale_samples_per_second": 10.995, | |
| "eval_nb_samtale_steps_per_second": 0.351, | |
| "eval_nb_samtale_wer": 0.16243489583333334, | |
| "step": 35406 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "grad_norm": 9.809260368347168, | |
| "learning_rate": 9.209584470730968e-06, | |
| "loss": 0.3439, | |
| "step": 37092 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_bigbrother_loss": 1.3220369815826416, | |
| "eval_bigbrother_runtime": 57.3394, | |
| "eval_bigbrother_samples_per_second": 23.893, | |
| "eval_bigbrother_steps_per_second": 0.75, | |
| "eval_bigbrother_wer": 0.3744461451703859, | |
| "step": 37092 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_BB_NB_RUND_loss": 0.7814194560050964, | |
| "eval_BB_NB_RUND_runtime": 216.0386, | |
| "eval_BB_NB_RUND_samples_per_second": 15.007, | |
| "eval_BB_NB_RUND_steps_per_second": 0.472, | |
| "eval_BB_NB_RUND_wer": 0.22033169368061714, | |
| "step": 37092 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_rundkast_loss": 0.36451050639152527, | |
| "eval_rundkast_runtime": 41.6479, | |
| "eval_rundkast_samples_per_second": 32.15, | |
| "eval_rundkast_steps_per_second": 1.008, | |
| "eval_rundkast_wer": 0.12589985602303633, | |
| "step": 37092 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_nb_samtale_loss": 0.4385349452495575, | |
| "eval_nb_samtale_runtime": 48.7109, | |
| "eval_nb_samtale_samples_per_second": 10.942, | |
| "eval_nb_samtale_steps_per_second": 0.349, | |
| "eval_nb_samtale_wer": 0.16015625, | |
| "step": 37092 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "grad_norm": 12.92618179321289, | |
| "learning_rate": 8.69851380042463e-06, | |
| "loss": 0.3378, | |
| "step": 38778 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_bigbrother_loss": 1.227617859840393, | |
| "eval_bigbrother_runtime": 57.2357, | |
| "eval_bigbrother_samples_per_second": 23.936, | |
| "eval_bigbrother_steps_per_second": 0.751, | |
| "eval_bigbrother_wer": 0.3678401675662612, | |
| "step": 38778 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_BB_NB_RUND_loss": 0.7239949107170105, | |
| "eval_BB_NB_RUND_runtime": 206.829, | |
| "eval_BB_NB_RUND_samples_per_second": 15.675, | |
| "eval_BB_NB_RUND_steps_per_second": 0.493, | |
| "eval_BB_NB_RUND_wer": 0.21656855629922317, | |
| "step": 38778 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_rundkast_loss": 0.3332752287387848, | |
| "eval_rundkast_runtime": 41.5376, | |
| "eval_rundkast_samples_per_second": 32.236, | |
| "eval_rundkast_steps_per_second": 1.011, | |
| "eval_rundkast_wer": 0.12286034234522476, | |
| "step": 38778 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_nb_samtale_loss": 0.4117254614830017, | |
| "eval_nb_samtale_runtime": 48.8538, | |
| "eval_nb_samtale_samples_per_second": 10.91, | |
| "eval_nb_samtale_steps_per_second": 0.348, | |
| "eval_nb_samtale_wer": 0.1591796875, | |
| "step": 38778 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 67440, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 40, | |
| "save_steps": 500, | |
| "total_flos": 6.56874003739803e+20, | |
| "train_batch_size": 48, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |