| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.390835579514825, | |
| "eval_steps": 30, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08086253369272237, | |
| "eval_accuracy": 0.34545454545454546, | |
| "eval_f1": 0.08213880790411401, | |
| "eval_loss": 1.7152690887451172, | |
| "eval_precision": 0.12064459930313588, | |
| "eval_recall": 0.14761904761904762, | |
| "eval_runtime": 13.6508, | |
| "eval_samples_per_second": 24.174, | |
| "eval_steps_per_second": 3.077, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.16172506738544473, | |
| "eval_accuracy": 0.4393939393939394, | |
| "eval_f1": 0.21690145794050794, | |
| "eval_loss": 1.5937703847885132, | |
| "eval_precision": 0.32583090861779385, | |
| "eval_recall": 0.25099255283137484, | |
| "eval_runtime": 13.1278, | |
| "eval_samples_per_second": 25.137, | |
| "eval_steps_per_second": 3.199, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24258760107816713, | |
| "eval_accuracy": 0.38181818181818183, | |
| "eval_f1": 0.13698374256214413, | |
| "eval_loss": 1.5304551124572754, | |
| "eval_precision": 0.23018433179723502, | |
| "eval_recall": 0.17921977437427894, | |
| "eval_runtime": 13.0828, | |
| "eval_samples_per_second": 25.224, | |
| "eval_steps_per_second": 3.21, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.32345013477088946, | |
| "eval_accuracy": 0.37575757575757573, | |
| "eval_f1": 0.13802749957562382, | |
| "eval_loss": 1.6096415519714355, | |
| "eval_precision": 0.1890046674322801, | |
| "eval_recall": 0.18045360027080612, | |
| "eval_runtime": 13.3288, | |
| "eval_samples_per_second": 24.758, | |
| "eval_steps_per_second": 3.151, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.40431266846361186, | |
| "eval_accuracy": 0.45151515151515154, | |
| "eval_f1": 0.23452986300450943, | |
| "eval_loss": 1.4274822473526, | |
| "eval_precision": 0.38950040524911433, | |
| "eval_recall": 0.27641592488584005, | |
| "eval_runtime": 13.4187, | |
| "eval_samples_per_second": 24.593, | |
| "eval_steps_per_second": 3.13, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.48517520215633425, | |
| "eval_accuracy": 0.5424242424242425, | |
| "eval_f1": 0.327924405270561, | |
| "eval_loss": 1.2386934757232666, | |
| "eval_precision": 0.35548957287833166, | |
| "eval_recall": 0.35121955898274004, | |
| "eval_runtime": 14.6075, | |
| "eval_samples_per_second": 22.591, | |
| "eval_steps_per_second": 2.875, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5660377358490566, | |
| "eval_accuracy": 0.41818181818181815, | |
| "eval_f1": 0.2642022704662224, | |
| "eval_loss": 1.7232695817947388, | |
| "eval_precision": 0.440507406905074, | |
| "eval_recall": 0.29508974215683664, | |
| "eval_runtime": 14.2568, | |
| "eval_samples_per_second": 23.147, | |
| "eval_steps_per_second": 2.946, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.6469002695417789, | |
| "eval_accuracy": 0.5575757575757576, | |
| "eval_f1": 0.36732006348386254, | |
| "eval_loss": 1.1753942966461182, | |
| "eval_precision": 0.43188775510204075, | |
| "eval_recall": 0.36647921681366585, | |
| "eval_runtime": 13.3212, | |
| "eval_samples_per_second": 24.772, | |
| "eval_steps_per_second": 3.153, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7277628032345014, | |
| "eval_accuracy": 0.5393939393939394, | |
| "eval_f1": 0.32937378728076405, | |
| "eval_loss": 1.2770304679870605, | |
| "eval_precision": 0.39731428747888, | |
| "eval_recall": 0.32973768555006505, | |
| "eval_runtime": 13.2685, | |
| "eval_samples_per_second": 24.871, | |
| "eval_steps_per_second": 3.165, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.8086253369272237, | |
| "eval_accuracy": 0.5606060606060606, | |
| "eval_f1": 0.3439473243062972, | |
| "eval_loss": 1.2049232721328735, | |
| "eval_precision": 0.4835807771246264, | |
| "eval_recall": 0.33995508230465793, | |
| "eval_runtime": 15.2882, | |
| "eval_samples_per_second": 21.585, | |
| "eval_steps_per_second": 2.747, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.889487870619946, | |
| "eval_accuracy": 0.5666666666666667, | |
| "eval_f1": 0.394947488354592, | |
| "eval_loss": 1.108344554901123, | |
| "eval_precision": 0.5166598073288214, | |
| "eval_recall": 0.3978135018724203, | |
| "eval_runtime": 14.5889, | |
| "eval_samples_per_second": 22.62, | |
| "eval_steps_per_second": 2.879, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.9703504043126685, | |
| "eval_accuracy": 0.6121212121212121, | |
| "eval_f1": 0.428135738050604, | |
| "eval_loss": 1.141391634941101, | |
| "eval_precision": 0.6312669683257919, | |
| "eval_recall": 0.40262976955895236, | |
| "eval_runtime": 12.9679, | |
| "eval_samples_per_second": 25.447, | |
| "eval_steps_per_second": 3.239, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.0512129380053907, | |
| "eval_accuracy": 0.603030303030303, | |
| "eval_f1": 0.4038251382972519, | |
| "eval_loss": 1.116855263710022, | |
| "eval_precision": 0.4923091038697975, | |
| "eval_recall": 0.41548484912495504, | |
| "eval_runtime": 13.6567, | |
| "eval_samples_per_second": 24.164, | |
| "eval_steps_per_second": 3.075, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.1320754716981132, | |
| "eval_accuracy": 0.5757575757575758, | |
| "eval_f1": 0.41737552463332295, | |
| "eval_loss": 1.13466477394104, | |
| "eval_precision": 0.6076030412164866, | |
| "eval_recall": 0.4208784549569585, | |
| "eval_runtime": 12.847, | |
| "eval_samples_per_second": 25.687, | |
| "eval_steps_per_second": 3.269, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.2129380053908356, | |
| "eval_accuracy": 0.603030303030303, | |
| "eval_f1": 0.43365687052323004, | |
| "eval_loss": 1.1254369020462036, | |
| "eval_precision": 0.5694631128061721, | |
| "eval_recall": 0.41092089239472196, | |
| "eval_runtime": 12.8187, | |
| "eval_samples_per_second": 25.744, | |
| "eval_steps_per_second": 3.276, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.2938005390835579, | |
| "eval_accuracy": 0.5696969696969697, | |
| "eval_f1": 0.3965291918438572, | |
| "eval_loss": 1.2160786390304565, | |
| "eval_precision": 0.4828996768066082, | |
| "eval_recall": 0.38449281792865353, | |
| "eval_runtime": 14.7332, | |
| "eval_samples_per_second": 22.398, | |
| "eval_steps_per_second": 2.851, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.3477088948787062, | |
| "grad_norm": 18.155202865600586, | |
| "learning_rate": 3.876909254267745e-05, | |
| "loss": 1.3924, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.3746630727762803, | |
| "eval_accuracy": 0.5424242424242425, | |
| "eval_f1": 0.4319260406224048, | |
| "eval_loss": 1.1475260257720947, | |
| "eval_precision": 0.45103525254505755, | |
| "eval_recall": 0.4340482151274285, | |
| "eval_runtime": 32.076, | |
| "eval_samples_per_second": 10.288, | |
| "eval_steps_per_second": 1.309, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.4555256064690028, | |
| "eval_accuracy": 0.5878787878787879, | |
| "eval_f1": 0.3939948328709428, | |
| "eval_loss": 1.1194807291030884, | |
| "eval_precision": 0.608180480837931, | |
| "eval_recall": 0.392683155449795, | |
| "eval_runtime": 110.8563, | |
| "eval_samples_per_second": 2.977, | |
| "eval_steps_per_second": 0.379, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.536388140161725, | |
| "eval_accuracy": 0.5818181818181818, | |
| "eval_f1": 0.4490235853265899, | |
| "eval_loss": 1.1963447332382202, | |
| "eval_precision": 0.5116256986573146, | |
| "eval_recall": 0.4377877348363712, | |
| "eval_runtime": 64.5493, | |
| "eval_samples_per_second": 5.112, | |
| "eval_steps_per_second": 0.651, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.6172506738544474, | |
| "eval_accuracy": 0.6060606060606061, | |
| "eval_f1": 0.44688978798681084, | |
| "eval_loss": 1.1052302122116089, | |
| "eval_precision": 0.5367127769479213, | |
| "eval_recall": 0.4296217856765177, | |
| "eval_runtime": 71.5496, | |
| "eval_samples_per_second": 4.612, | |
| "eval_steps_per_second": 0.587, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.6981132075471699, | |
| "eval_accuracy": 0.5848484848484848, | |
| "eval_f1": 0.46957050415643353, | |
| "eval_loss": 1.108028769493103, | |
| "eval_precision": 0.572241929381981, | |
| "eval_recall": 0.4537304361700488, | |
| "eval_runtime": 12.7244, | |
| "eval_samples_per_second": 25.935, | |
| "eval_steps_per_second": 3.301, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.778975741239892, | |
| "eval_accuracy": 0.6212121212121212, | |
| "eval_f1": 0.47748925036798145, | |
| "eval_loss": 1.033095359802246, | |
| "eval_precision": 0.5650004027150538, | |
| "eval_recall": 0.4613483358977623, | |
| "eval_runtime": 13.0381, | |
| "eval_samples_per_second": 25.31, | |
| "eval_steps_per_second": 3.221, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.8598382749326146, | |
| "eval_accuracy": 0.6242424242424243, | |
| "eval_f1": 0.4612218625033871, | |
| "eval_loss": 1.0287344455718994, | |
| "eval_precision": 0.5787994888332474, | |
| "eval_recall": 0.4401355279945231, | |
| "eval_runtime": 14.0033, | |
| "eval_samples_per_second": 23.566, | |
| "eval_steps_per_second": 2.999, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.940700808625337, | |
| "eval_accuracy": 0.6, | |
| "eval_f1": 0.43355058251543566, | |
| "eval_loss": 1.0855339765548706, | |
| "eval_precision": 0.5482738467345629, | |
| "eval_recall": 0.407901742703353, | |
| "eval_runtime": 14.8086, | |
| "eval_samples_per_second": 22.284, | |
| "eval_steps_per_second": 2.836, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.0215633423180592, | |
| "eval_accuracy": 0.6060606060606061, | |
| "eval_f1": 0.4278903186584704, | |
| "eval_loss": 1.0721008777618408, | |
| "eval_precision": 0.6261473906210748, | |
| "eval_recall": 0.400766872593428, | |
| "eval_runtime": 13.33, | |
| "eval_samples_per_second": 24.756, | |
| "eval_steps_per_second": 3.151, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.1024258760107815, | |
| "eval_accuracy": 0.5818181818181818, | |
| "eval_f1": 0.46279487414189074, | |
| "eval_loss": 1.0543911457061768, | |
| "eval_precision": 0.4927064392103887, | |
| "eval_recall": 0.4551381887414431, | |
| "eval_runtime": 13.3061, | |
| "eval_samples_per_second": 24.801, | |
| "eval_steps_per_second": 3.156, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.183288409703504, | |
| "eval_accuracy": 0.6, | |
| "eval_f1": 0.44787598241263715, | |
| "eval_loss": 1.1306463479995728, | |
| "eval_precision": 0.5029518040278441, | |
| "eval_recall": 0.42696476597535643, | |
| "eval_runtime": 14.7754, | |
| "eval_samples_per_second": 22.334, | |
| "eval_steps_per_second": 2.843, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.2641509433962264, | |
| "eval_accuracy": 0.6151515151515151, | |
| "eval_f1": 0.47831287112215076, | |
| "eval_loss": 1.1025209426879883, | |
| "eval_precision": 0.5683183097103888, | |
| "eval_recall": 0.45416730293429747, | |
| "eval_runtime": 14.712, | |
| "eval_samples_per_second": 22.431, | |
| "eval_steps_per_second": 2.855, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.3450134770889486, | |
| "eval_accuracy": 0.5848484848484848, | |
| "eval_f1": 0.4612687851809403, | |
| "eval_loss": 1.1565966606140137, | |
| "eval_precision": 0.5073562417803803, | |
| "eval_recall": 0.4593010520365005, | |
| "eval_runtime": 13.517, | |
| "eval_samples_per_second": 24.414, | |
| "eval_steps_per_second": 3.107, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.4258760107816713, | |
| "eval_accuracy": 0.5636363636363636, | |
| "eval_f1": 0.45972084178362305, | |
| "eval_loss": 1.2057921886444092, | |
| "eval_precision": 0.5695289983467323, | |
| "eval_recall": 0.461413466501444, | |
| "eval_runtime": 13.0432, | |
| "eval_samples_per_second": 25.301, | |
| "eval_steps_per_second": 3.22, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.5067385444743935, | |
| "eval_accuracy": 0.6090909090909091, | |
| "eval_f1": 0.4628685133833875, | |
| "eval_loss": 1.1381313800811768, | |
| "eval_precision": 0.5632985257985258, | |
| "eval_recall": 0.4583315282026837, | |
| "eval_runtime": 14.1243, | |
| "eval_samples_per_second": 23.364, | |
| "eval_steps_per_second": 2.974, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.5876010781671157, | |
| "eval_accuracy": 0.6242424242424243, | |
| "eval_f1": 0.4615624143566702, | |
| "eval_loss": 1.0724554061889648, | |
| "eval_precision": 0.5741660836493663, | |
| "eval_recall": 0.4438527652883962, | |
| "eval_runtime": 14.1621, | |
| "eval_samples_per_second": 23.302, | |
| "eval_steps_per_second": 2.966, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.6684636118598384, | |
| "eval_accuracy": 0.6060606060606061, | |
| "eval_f1": 0.47087755462465813, | |
| "eval_loss": 1.135140299797058, | |
| "eval_precision": 0.5209651599354117, | |
| "eval_recall": 0.45427047745523014, | |
| "eval_runtime": 14.0682, | |
| "eval_samples_per_second": 23.457, | |
| "eval_steps_per_second": 2.985, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.6954177897574123, | |
| "grad_norm": 7.412640571594238, | |
| "learning_rate": 2.75381850853549e-05, | |
| "loss": 0.9741, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.7493261455525606, | |
| "eval_accuracy": 0.6242424242424243, | |
| "eval_f1": 0.4549752385442883, | |
| "eval_loss": 0.9898973703384399, | |
| "eval_precision": 0.5431213833871716, | |
| "eval_recall": 0.44127615017259786, | |
| "eval_runtime": 113.2331, | |
| "eval_samples_per_second": 2.914, | |
| "eval_steps_per_second": 0.371, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.830188679245283, | |
| "eval_accuracy": 0.6212121212121212, | |
| "eval_f1": 0.477029569791836, | |
| "eval_loss": 1.1227794885635376, | |
| "eval_precision": 0.5355122892068706, | |
| "eval_recall": 0.4702582406289971, | |
| "eval_runtime": 22.5487, | |
| "eval_samples_per_second": 14.635, | |
| "eval_steps_per_second": 1.863, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.9110512129380055, | |
| "eval_accuracy": 0.6272727272727273, | |
| "eval_f1": 0.4661588293248819, | |
| "eval_loss": 1.0796787738800049, | |
| "eval_precision": 0.5698680297271846, | |
| "eval_recall": 0.4402755664195842, | |
| "eval_runtime": 142.0116, | |
| "eval_samples_per_second": 2.324, | |
| "eval_steps_per_second": 0.296, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.9919137466307277, | |
| "eval_accuracy": 0.6424242424242425, | |
| "eval_f1": 0.49676860579479715, | |
| "eval_loss": 1.0576601028442383, | |
| "eval_precision": 0.5889560125125319, | |
| "eval_recall": 0.48000300606800994, | |
| "eval_runtime": 25.0835, | |
| "eval_samples_per_second": 13.156, | |
| "eval_steps_per_second": 1.674, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 3.07277628032345, | |
| "eval_accuracy": 0.6333333333333333, | |
| "eval_f1": 0.5076311071541079, | |
| "eval_loss": 1.1107242107391357, | |
| "eval_precision": 0.5412034459392385, | |
| "eval_recall": 0.4934487547962528, | |
| "eval_runtime": 13.978, | |
| "eval_samples_per_second": 23.608, | |
| "eval_steps_per_second": 3.005, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 3.1536388140161726, | |
| "eval_accuracy": 0.6515151515151515, | |
| "eval_f1": 0.5186595057636214, | |
| "eval_loss": 1.1433833837509155, | |
| "eval_precision": 0.5322955893403752, | |
| "eval_recall": 0.5121777726730862, | |
| "eval_runtime": 14.8138, | |
| "eval_samples_per_second": 22.277, | |
| "eval_steps_per_second": 2.835, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 3.234501347708895, | |
| "eval_accuracy": 0.6303030303030303, | |
| "eval_f1": 0.4818021563155342, | |
| "eval_loss": 1.2040390968322754, | |
| "eval_precision": 0.5245503034019318, | |
| "eval_recall": 0.4790353988658817, | |
| "eval_runtime": 14.5347, | |
| "eval_samples_per_second": 22.704, | |
| "eval_steps_per_second": 2.89, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.315363881401617, | |
| "eval_accuracy": 0.6303030303030303, | |
| "eval_f1": 0.5005894502956657, | |
| "eval_loss": 1.106711983680725, | |
| "eval_precision": 0.5347839031818471, | |
| "eval_recall": 0.49606350872380334, | |
| "eval_runtime": 14.5181, | |
| "eval_samples_per_second": 22.73, | |
| "eval_steps_per_second": 2.893, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 3.3962264150943398, | |
| "eval_accuracy": 0.6363636363636364, | |
| "eval_f1": 0.49230222929692463, | |
| "eval_loss": 1.100297451019287, | |
| "eval_precision": 0.5613737992397406, | |
| "eval_recall": 0.48222645824949406, | |
| "eval_runtime": 14.6402, | |
| "eval_samples_per_second": 22.541, | |
| "eval_steps_per_second": 2.869, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 3.477088948787062, | |
| "eval_accuracy": 0.603030303030303, | |
| "eval_f1": 0.48380416653626124, | |
| "eval_loss": 1.1756219863891602, | |
| "eval_precision": 0.5033237551313431, | |
| "eval_recall": 0.4908017943866942, | |
| "eval_runtime": 13.454, | |
| "eval_samples_per_second": 24.528, | |
| "eval_steps_per_second": 3.122, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 3.557951482479784, | |
| "eval_accuracy": 0.6060606060606061, | |
| "eval_f1": 0.4764702057860691, | |
| "eval_loss": 1.1003729104995728, | |
| "eval_precision": 0.5288839860268432, | |
| "eval_recall": 0.4590755558938861, | |
| "eval_runtime": 12.5627, | |
| "eval_samples_per_second": 26.268, | |
| "eval_steps_per_second": 3.343, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 3.638814016172507, | |
| "eval_accuracy": 0.593939393939394, | |
| "eval_f1": 0.487375493221817, | |
| "eval_loss": 1.086908221244812, | |
| "eval_precision": 0.48992444445192695, | |
| "eval_recall": 0.5073454405336776, | |
| "eval_runtime": 12.6831, | |
| "eval_samples_per_second": 26.019, | |
| "eval_steps_per_second": 3.311, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.719676549865229, | |
| "eval_accuracy": 0.5818181818181818, | |
| "eval_f1": 0.4671306646522057, | |
| "eval_loss": 1.1851921081542969, | |
| "eval_precision": 0.5193033472658716, | |
| "eval_recall": 0.46147135451955557, | |
| "eval_runtime": 13.4674, | |
| "eval_samples_per_second": 24.504, | |
| "eval_steps_per_second": 3.119, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 3.8005390835579513, | |
| "eval_accuracy": 0.6272727272727273, | |
| "eval_f1": 0.4738897906354585, | |
| "eval_loss": 1.1233046054840088, | |
| "eval_precision": 0.5308286308286309, | |
| "eval_recall": 0.45352857445191025, | |
| "eval_runtime": 14.4652, | |
| "eval_samples_per_second": 22.813, | |
| "eval_steps_per_second": 2.904, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 3.881401617250674, | |
| "eval_accuracy": 0.6333333333333333, | |
| "eval_f1": 0.49797095491412685, | |
| "eval_loss": 1.0570012331008911, | |
| "eval_precision": 0.5228267459278104, | |
| "eval_recall": 0.48949827535886625, | |
| "eval_runtime": 13.7881, | |
| "eval_samples_per_second": 23.934, | |
| "eval_steps_per_second": 3.046, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 3.9622641509433962, | |
| "eval_accuracy": 0.5696969696969697, | |
| "eval_f1": 0.436851400958813, | |
| "eval_loss": 1.196516513824463, | |
| "eval_precision": 0.4607919084994466, | |
| "eval_recall": 0.44320607800419104, | |
| "eval_runtime": 13.9502, | |
| "eval_samples_per_second": 23.656, | |
| "eval_steps_per_second": 3.011, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 4.0431266846361185, | |
| "grad_norm": 4.396773815155029, | |
| "learning_rate": 1.6307277628032348e-05, | |
| "loss": 0.6945, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.0431266846361185, | |
| "eval_accuracy": 0.6242424242424243, | |
| "eval_f1": 0.48252518710343445, | |
| "eval_loss": 1.0697778463363647, | |
| "eval_precision": 0.529925883396837, | |
| "eval_recall": 0.4736448771051681, | |
| "eval_runtime": 14.005, | |
| "eval_samples_per_second": 23.563, | |
| "eval_steps_per_second": 2.999, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.123989218328841, | |
| "eval_accuracy": 0.6242424242424243, | |
| "eval_f1": 0.49720491179937104, | |
| "eval_loss": 1.2521579265594482, | |
| "eval_precision": 0.5386641378880501, | |
| "eval_recall": 0.49817853705778303, | |
| "eval_runtime": 115.3902, | |
| "eval_samples_per_second": 2.86, | |
| "eval_steps_per_second": 0.364, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 4.204851752021563, | |
| "eval_accuracy": 0.6181818181818182, | |
| "eval_f1": 0.47697868890738926, | |
| "eval_loss": 1.4063215255737305, | |
| "eval_precision": 0.4929803350172045, | |
| "eval_recall": 0.4728290050243222, | |
| "eval_runtime": 101.5928, | |
| "eval_samples_per_second": 3.248, | |
| "eval_steps_per_second": 0.413, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 4.285714285714286, | |
| "eval_accuracy": 0.6212121212121212, | |
| "eval_f1": 0.4993387370778744, | |
| "eval_loss": 1.349918007850647, | |
| "eval_precision": 0.49880570409982167, | |
| "eval_recall": 0.5059772713055806, | |
| "eval_runtime": 115.6893, | |
| "eval_samples_per_second": 2.852, | |
| "eval_steps_per_second": 0.363, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 4.366576819407008, | |
| "eval_accuracy": 0.6, | |
| "eval_f1": 0.46437393231058216, | |
| "eval_loss": 1.340740442276001, | |
| "eval_precision": 0.47078319079488345, | |
| "eval_recall": 0.46173072420332834, | |
| "eval_runtime": 13.0095, | |
| "eval_samples_per_second": 25.366, | |
| "eval_steps_per_second": 3.228, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 4.44743935309973, | |
| "eval_accuracy": 0.6393939393939394, | |
| "eval_f1": 0.4956975851278201, | |
| "eval_loss": 1.4181333780288696, | |
| "eval_precision": 0.5323631437686738, | |
| "eval_recall": 0.48370483391830105, | |
| "eval_runtime": 15.6022, | |
| "eval_samples_per_second": 21.151, | |
| "eval_steps_per_second": 2.692, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 4.528301886792453, | |
| "eval_accuracy": 0.6212121212121212, | |
| "eval_f1": 0.48744288293160476, | |
| "eval_loss": 1.4648175239562988, | |
| "eval_precision": 0.5288562135045842, | |
| "eval_recall": 0.484239463132841, | |
| "eval_runtime": 14.8896, | |
| "eval_samples_per_second": 22.163, | |
| "eval_steps_per_second": 2.821, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 4.609164420485175, | |
| "eval_accuracy": 0.6363636363636364, | |
| "eval_f1": 0.49874561310412674, | |
| "eval_loss": 1.3795430660247803, | |
| "eval_precision": 0.515957965289801, | |
| "eval_recall": 0.49555231275739653, | |
| "eval_runtime": 15.2135, | |
| "eval_samples_per_second": 21.691, | |
| "eval_steps_per_second": 2.761, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 4.690026954177897, | |
| "eval_accuracy": 0.6545454545454545, | |
| "eval_f1": 0.5072945393065025, | |
| "eval_loss": 1.255421757698059, | |
| "eval_precision": 0.527637085699966, | |
| "eval_recall": 0.5017196060314337, | |
| "eval_runtime": 13.3432, | |
| "eval_samples_per_second": 24.732, | |
| "eval_steps_per_second": 3.148, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 4.77088948787062, | |
| "eval_accuracy": 0.6181818181818182, | |
| "eval_f1": 0.47364322391678987, | |
| "eval_loss": 1.2744801044464111, | |
| "eval_precision": 0.5261603756546867, | |
| "eval_recall": 0.4762025977768268, | |
| "eval_runtime": 13.3766, | |
| "eval_samples_per_second": 24.67, | |
| "eval_steps_per_second": 3.14, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 4.8517520215633425, | |
| "eval_accuracy": 0.6272727272727273, | |
| "eval_f1": 0.5006190896494088, | |
| "eval_loss": 1.2532024383544922, | |
| "eval_precision": 0.5347134093222614, | |
| "eval_recall": 0.4860070789020448, | |
| "eval_runtime": 13.4318, | |
| "eval_samples_per_second": 24.569, | |
| "eval_steps_per_second": 3.127, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.932614555256064, | |
| "eval_accuracy": 0.6060606060606061, | |
| "eval_f1": 0.47951387194587414, | |
| "eval_loss": 1.2477487325668335, | |
| "eval_precision": 0.4841348163818714, | |
| "eval_recall": 0.4813308565204252, | |
| "eval_runtime": 13.4442, | |
| "eval_samples_per_second": 24.546, | |
| "eval_steps_per_second": 3.124, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 5.013477088948787, | |
| "eval_accuracy": 0.6393939393939394, | |
| "eval_f1": 0.4989217275259948, | |
| "eval_loss": 1.2193477153778076, | |
| "eval_precision": 0.5258493064306811, | |
| "eval_recall": 0.4869117013024005, | |
| "eval_runtime": 13.4387, | |
| "eval_samples_per_second": 24.556, | |
| "eval_steps_per_second": 3.125, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 5.09433962264151, | |
| "eval_accuracy": 0.6424242424242425, | |
| "eval_f1": 0.5115200267534169, | |
| "eval_loss": 1.381670355796814, | |
| "eval_precision": 0.5336989009157052, | |
| "eval_recall": 0.5001792056559704, | |
| "eval_runtime": 14.4622, | |
| "eval_samples_per_second": 22.818, | |
| "eval_steps_per_second": 2.904, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 5.175202156334231, | |
| "eval_accuracy": 0.6272727272727273, | |
| "eval_f1": 0.49956294294308473, | |
| "eval_loss": 1.4055507183074951, | |
| "eval_precision": 0.510424264019138, | |
| "eval_recall": 0.49196977915696183, | |
| "eval_runtime": 13.4719, | |
| "eval_samples_per_second": 24.496, | |
| "eval_steps_per_second": 3.118, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 5.256064690026954, | |
| "eval_accuracy": 0.6303030303030303, | |
| "eval_f1": 0.49657096716171206, | |
| "eval_loss": 1.4433475732803345, | |
| "eval_precision": 0.5023020702533887, | |
| "eval_recall": 0.49618528944682805, | |
| "eval_runtime": 12.5628, | |
| "eval_samples_per_second": 26.268, | |
| "eval_steps_per_second": 3.343, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 5.336927223719677, | |
| "eval_accuracy": 0.6212121212121212, | |
| "eval_f1": 0.4846771824827281, | |
| "eval_loss": 1.4632567167282104, | |
| "eval_precision": 0.49600215736518566, | |
| "eval_recall": 0.4812536923450428, | |
| "eval_runtime": 12.7703, | |
| "eval_samples_per_second": 25.841, | |
| "eval_steps_per_second": 3.289, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 5.390835579514825, | |
| "grad_norm": 10.319731712341309, | |
| "learning_rate": 5.0763701707097935e-06, | |
| "loss": 0.4025, | |
| "step": 2000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 2226, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 6, | |
| "save_steps": 500, | |
| "total_flos": 9271286495496432.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |