| { |
| "best_metric": 12.636168876478752, |
| "best_model_checkpoint": "/content/marianmt-santali-sanlish-to-bangla/checkpoint-3298", |
| "epoch": 22.0, |
| "eval_steps": 500, |
| "global_step": 4268, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.15463917525773196, |
| "grad_norm": 43.064876556396484, |
| "learning_rate": 8.93470790378007e-07, |
| "loss": 8.0531, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.30927835051546393, |
| "grad_norm": 38.152244567871094, |
| "learning_rate": 1.9243986254295534e-06, |
| "loss": 6.8732, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.4639175257731959, |
| "grad_norm": 20.582117080688477, |
| "learning_rate": 2.9553264604811e-06, |
| "loss": 4.9163, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.6185567010309279, |
| "grad_norm": 13.089325904846191, |
| "learning_rate": 3.986254295532647e-06, |
| "loss": 3.381, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.7731958762886598, |
| "grad_norm": 12.60307502746582, |
| "learning_rate": 5.017182130584193e-06, |
| "loss": 3.0247, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.9278350515463918, |
| "grad_norm": 12.468002319335938, |
| "learning_rate": 6.048109965635739e-06, |
| "loss": 2.942, |
| "step": 180 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_bertscore": 0.7088343220053559, |
| "eval_bleu": 0.3864012583347095, |
| "eval_chrf": 10.09476116429599, |
| "eval_loss": 2.6114141941070557, |
| "eval_meteor": 0.028841924396372502, |
| "eval_runtime": 38.0618, |
| "eval_samples_per_second": 5.071, |
| "eval_steps_per_second": 0.657, |
| "step": 194 |
| }, |
| { |
| "epoch": 1.0824742268041236, |
| "grad_norm": 10.807812690734863, |
| "learning_rate": 7.079037800687286e-06, |
| "loss": 2.6762, |
| "step": 210 |
| }, |
| { |
| "epoch": 1.2371134020618557, |
| "grad_norm": 10.950024604797363, |
| "learning_rate": 8.109965635738832e-06, |
| "loss": 2.6531, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.3917525773195876, |
| "grad_norm": 9.422192573547363, |
| "learning_rate": 9.140893470790379e-06, |
| "loss": 2.5974, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.5463917525773194, |
| "grad_norm": 14.578783988952637, |
| "learning_rate": 1.0171821305841924e-05, |
| "loss": 2.6236, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.7010309278350515, |
| "grad_norm": 13.789867401123047, |
| "learning_rate": 1.1202749140893473e-05, |
| "loss": 2.4943, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.8556701030927836, |
| "grad_norm": 12.978485107421875, |
| "learning_rate": 1.223367697594502e-05, |
| "loss": 2.5078, |
| "step": 360 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_bertscore": 0.7392782556578286, |
| "eval_bleu": 0.46239407700960067, |
| "eval_chrf": 12.818734803935335, |
| "eval_loss": 2.2853775024414062, |
| "eval_meteor": 0.048241173739640195, |
| "eval_runtime": 18.2718, |
| "eval_samples_per_second": 10.563, |
| "eval_steps_per_second": 1.368, |
| "step": 388 |
| }, |
| { |
| "epoch": 2.0103092783505154, |
| "grad_norm": 10.923307418823242, |
| "learning_rate": 1.3264604810996563e-05, |
| "loss": 2.4348, |
| "step": 390 |
| }, |
| { |
| "epoch": 2.1649484536082473, |
| "grad_norm": 11.970940589904785, |
| "learning_rate": 1.4295532646048112e-05, |
| "loss": 2.3111, |
| "step": 420 |
| }, |
| { |
| "epoch": 2.319587628865979, |
| "grad_norm": 13.831852912902832, |
| "learning_rate": 1.5326460481099657e-05, |
| "loss": 2.278, |
| "step": 450 |
| }, |
| { |
| "epoch": 2.4742268041237114, |
| "grad_norm": 13.530282974243164, |
| "learning_rate": 1.6357388316151203e-05, |
| "loss": 2.3752, |
| "step": 480 |
| }, |
| { |
| "epoch": 2.6288659793814433, |
| "grad_norm": 13.126559257507324, |
| "learning_rate": 1.738831615120275e-05, |
| "loss": 2.1868, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.783505154639175, |
| "grad_norm": 11.856134414672852, |
| "learning_rate": 1.8419243986254297e-05, |
| "loss": 2.1822, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.9381443298969074, |
| "grad_norm": 9.861915588378906, |
| "learning_rate": 1.9450171821305842e-05, |
| "loss": 2.1383, |
| "step": 570 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_bertscore": 0.7664902392446686, |
| "eval_bleu": 1.820820260915908, |
| "eval_chrf": 18.54167447294365, |
| "eval_loss": 2.055220603942871, |
| "eval_meteor": 0.11437707653982947, |
| "eval_runtime": 18.6431, |
| "eval_samples_per_second": 10.352, |
| "eval_steps_per_second": 1.341, |
| "step": 582 |
| }, |
| { |
| "epoch": 3.0927835051546393, |
| "grad_norm": 8.970413208007812, |
| "learning_rate": 1.994654448262696e-05, |
| "loss": 2.0298, |
| "step": 600 |
| }, |
| { |
| "epoch": 3.247422680412371, |
| "grad_norm": 11.038900375366211, |
| "learning_rate": 1.9831996945399007e-05, |
| "loss": 1.8987, |
| "step": 630 |
| }, |
| { |
| "epoch": 3.402061855670103, |
| "grad_norm": 9.999340057373047, |
| "learning_rate": 1.9717449408171058e-05, |
| "loss": 1.943, |
| "step": 660 |
| }, |
| { |
| "epoch": 3.556701030927835, |
| "grad_norm": 11.228028297424316, |
| "learning_rate": 1.960290187094311e-05, |
| "loss": 1.9018, |
| "step": 690 |
| }, |
| { |
| "epoch": 3.711340206185567, |
| "grad_norm": 14.46988296508789, |
| "learning_rate": 1.948835433371516e-05, |
| "loss": 1.916, |
| "step": 720 |
| }, |
| { |
| "epoch": 3.865979381443299, |
| "grad_norm": 13.187897682189941, |
| "learning_rate": 1.9373806796487212e-05, |
| "loss": 1.9096, |
| "step": 750 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_bertscore": 0.7867915055912393, |
| "eval_bleu": 3.1417647965851434, |
| "eval_chrf": 21.794103049706337, |
| "eval_loss": 1.8877290487289429, |
| "eval_meteor": 0.15566352755949234, |
| "eval_runtime": 17.9807, |
| "eval_samples_per_second": 10.734, |
| "eval_steps_per_second": 1.39, |
| "step": 776 |
| }, |
| { |
| "epoch": 4.020618556701031, |
| "grad_norm": 9.245710372924805, |
| "learning_rate": 1.925925925925926e-05, |
| "loss": 1.8022, |
| "step": 780 |
| }, |
| { |
| "epoch": 4.175257731958763, |
| "grad_norm": 12.707769393920898, |
| "learning_rate": 1.914471172203131e-05, |
| "loss": 1.7105, |
| "step": 810 |
| }, |
| { |
| "epoch": 4.329896907216495, |
| "grad_norm": 13.76497745513916, |
| "learning_rate": 1.9030164184803362e-05, |
| "loss": 1.7098, |
| "step": 840 |
| }, |
| { |
| "epoch": 4.484536082474227, |
| "grad_norm": 12.652901649475098, |
| "learning_rate": 1.8915616647575414e-05, |
| "loss": 1.6128, |
| "step": 870 |
| }, |
| { |
| "epoch": 4.639175257731958, |
| "grad_norm": 14.508373260498047, |
| "learning_rate": 1.880106911034746e-05, |
| "loss": 1.6844, |
| "step": 900 |
| }, |
| { |
| "epoch": 4.793814432989691, |
| "grad_norm": 10.372153282165527, |
| "learning_rate": 1.8686521573119513e-05, |
| "loss": 1.5839, |
| "step": 930 |
| }, |
| { |
| "epoch": 4.948453608247423, |
| "grad_norm": 10.381881713867188, |
| "learning_rate": 1.857197403589156e-05, |
| "loss": 1.5992, |
| "step": 960 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_bertscore": 0.8009231167753743, |
| "eval_bleu": 5.107916225528083, |
| "eval_chrf": 27.109939343296112, |
| "eval_loss": 1.7559988498687744, |
| "eval_meteor": 0.20502396164700096, |
| "eval_runtime": 19.6434, |
| "eval_samples_per_second": 9.825, |
| "eval_steps_per_second": 1.273, |
| "step": 970 |
| }, |
| { |
| "epoch": 5.103092783505154, |
| "grad_norm": 9.559462547302246, |
| "learning_rate": 1.8457426498663612e-05, |
| "loss": 1.4874, |
| "step": 990 |
| }, |
| { |
| "epoch": 5.257731958762887, |
| "grad_norm": 10.890524864196777, |
| "learning_rate": 1.8342878961435663e-05, |
| "loss": 1.4588, |
| "step": 1020 |
| }, |
| { |
| "epoch": 5.412371134020619, |
| "grad_norm": 10.118521690368652, |
| "learning_rate": 1.8228331424207715e-05, |
| "loss": 1.4945, |
| "step": 1050 |
| }, |
| { |
| "epoch": 5.56701030927835, |
| "grad_norm": 12.269123077392578, |
| "learning_rate": 1.8113783886979766e-05, |
| "loss": 1.4544, |
| "step": 1080 |
| }, |
| { |
| "epoch": 5.721649484536083, |
| "grad_norm": 9.705910682678223, |
| "learning_rate": 1.7999236349751814e-05, |
| "loss": 1.4013, |
| "step": 1110 |
| }, |
| { |
| "epoch": 5.876288659793815, |
| "grad_norm": 10.108024597167969, |
| "learning_rate": 1.7884688812523865e-05, |
| "loss": 1.3569, |
| "step": 1140 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_bertscore": 0.8189008788741314, |
| "eval_bleu": 7.80196732700183, |
| "eval_chrf": 31.07771783943801, |
| "eval_loss": 1.6978754997253418, |
| "eval_meteor": 0.2476381272119835, |
| "eval_runtime": 17.9742, |
| "eval_samples_per_second": 10.738, |
| "eval_steps_per_second": 1.391, |
| "step": 1164 |
| }, |
| { |
| "epoch": 6.030927835051546, |
| "grad_norm": 10.195992469787598, |
| "learning_rate": 1.7770141275295916e-05, |
| "loss": 1.4165, |
| "step": 1170 |
| }, |
| { |
| "epoch": 6.185567010309279, |
| "grad_norm": 10.022723197937012, |
| "learning_rate": 1.7655593738067968e-05, |
| "loss": 1.2383, |
| "step": 1200 |
| }, |
| { |
| "epoch": 6.34020618556701, |
| "grad_norm": 11.431436538696289, |
| "learning_rate": 1.754104620084002e-05, |
| "loss": 1.3013, |
| "step": 1230 |
| }, |
| { |
| "epoch": 6.494845360824742, |
| "grad_norm": 11.689397811889648, |
| "learning_rate": 1.7426498663612067e-05, |
| "loss": 1.2412, |
| "step": 1260 |
| }, |
| { |
| "epoch": 6.649484536082475, |
| "grad_norm": 11.549187660217285, |
| "learning_rate": 1.7311951126384118e-05, |
| "loss": 1.237, |
| "step": 1290 |
| }, |
| { |
| "epoch": 6.804123711340206, |
| "grad_norm": 11.665675163269043, |
| "learning_rate": 1.7197403589156166e-05, |
| "loss": 1.2955, |
| "step": 1320 |
| }, |
| { |
| "epoch": 6.958762886597938, |
| "grad_norm": 13.438718795776367, |
| "learning_rate": 1.7082856051928217e-05, |
| "loss": 1.195, |
| "step": 1350 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_bertscore": 0.8185995718358094, |
| "eval_bleu": 9.031051320914116, |
| "eval_chrf": 30.344158100165018, |
| "eval_loss": 1.6532801389694214, |
| "eval_meteor": 0.24965221252933717, |
| "eval_runtime": 17.6222, |
| "eval_samples_per_second": 10.952, |
| "eval_steps_per_second": 1.419, |
| "step": 1358 |
| }, |
| { |
| "epoch": 7.11340206185567, |
| "grad_norm": 9.628753662109375, |
| "learning_rate": 1.696830851470027e-05, |
| "loss": 1.1824, |
| "step": 1380 |
| }, |
| { |
| "epoch": 7.268041237113402, |
| "grad_norm": 15.119050025939941, |
| "learning_rate": 1.685376097747232e-05, |
| "loss": 1.0789, |
| "step": 1410 |
| }, |
| { |
| "epoch": 7.422680412371134, |
| "grad_norm": 10.973334312438965, |
| "learning_rate": 1.6739213440244368e-05, |
| "loss": 1.1213, |
| "step": 1440 |
| }, |
| { |
| "epoch": 7.577319587628866, |
| "grad_norm": 9.850281715393066, |
| "learning_rate": 1.662466590301642e-05, |
| "loss": 1.0941, |
| "step": 1470 |
| }, |
| { |
| "epoch": 7.731958762886598, |
| "grad_norm": 10.67463207244873, |
| "learning_rate": 1.651011836578847e-05, |
| "loss": 1.0986, |
| "step": 1500 |
| }, |
| { |
| "epoch": 7.88659793814433, |
| "grad_norm": 11.185319900512695, |
| "learning_rate": 1.639557082856052e-05, |
| "loss": 1.1345, |
| "step": 1530 |
| }, |
| { |
| "epoch": 8.0, |
| "eval_bertscore": 0.8248441731991546, |
| "eval_bleu": 9.326651170754813, |
| "eval_chrf": 32.19041027207769, |
| "eval_loss": 1.6205072402954102, |
| "eval_meteor": 0.2731252078182265, |
| "eval_runtime": 19.0403, |
| "eval_samples_per_second": 10.136, |
| "eval_steps_per_second": 1.313, |
| "step": 1552 |
| }, |
| { |
| "epoch": 8.041237113402062, |
| "grad_norm": 9.956164360046387, |
| "learning_rate": 1.6281023291332573e-05, |
| "loss": 1.0352, |
| "step": 1560 |
| }, |
| { |
| "epoch": 8.195876288659793, |
| "grad_norm": 13.017486572265625, |
| "learning_rate": 1.6166475754104624e-05, |
| "loss": 1.0962, |
| "step": 1590 |
| }, |
| { |
| "epoch": 8.350515463917526, |
| "grad_norm": 12.15532398223877, |
| "learning_rate": 1.6051928216876672e-05, |
| "loss": 0.9943, |
| "step": 1620 |
| }, |
| { |
| "epoch": 8.505154639175258, |
| "grad_norm": 11.221367835998535, |
| "learning_rate": 1.5937380679648723e-05, |
| "loss": 0.9336, |
| "step": 1650 |
| }, |
| { |
| "epoch": 8.65979381443299, |
| "grad_norm": 10.390768051147461, |
| "learning_rate": 1.5822833142420774e-05, |
| "loss": 0.9942, |
| "step": 1680 |
| }, |
| { |
| "epoch": 8.814432989690722, |
| "grad_norm": 10.396700859069824, |
| "learning_rate": 1.5708285605192822e-05, |
| "loss": 0.9616, |
| "step": 1710 |
| }, |
| { |
| "epoch": 8.969072164948454, |
| "grad_norm": 8.77428913116455, |
| "learning_rate": 1.5593738067964873e-05, |
| "loss": 0.9529, |
| "step": 1740 |
| }, |
| { |
| "epoch": 9.0, |
| "eval_bertscore": 0.8292578412461157, |
| "eval_bleu": 9.64249962743433, |
| "eval_chrf": 33.6951161128342, |
| "eval_loss": 1.6319836378097534, |
| "eval_meteor": 0.2765792548974407, |
| "eval_runtime": 17.0212, |
| "eval_samples_per_second": 11.339, |
| "eval_steps_per_second": 1.469, |
| "step": 1746 |
| }, |
| { |
| "epoch": 9.123711340206185, |
| "grad_norm": 8.413101196289062, |
| "learning_rate": 1.547919053073692e-05, |
| "loss": 0.9516, |
| "step": 1770 |
| }, |
| { |
| "epoch": 9.278350515463918, |
| "grad_norm": 10.946271896362305, |
| "learning_rate": 1.5364642993508973e-05, |
| "loss": 0.8993, |
| "step": 1800 |
| }, |
| { |
| "epoch": 9.43298969072165, |
| "grad_norm": 9.253824234008789, |
| "learning_rate": 1.5250095456281024e-05, |
| "loss": 0.8545, |
| "step": 1830 |
| }, |
| { |
| "epoch": 9.587628865979381, |
| "grad_norm": 9.479711532592773, |
| "learning_rate": 1.5135547919053075e-05, |
| "loss": 0.845, |
| "step": 1860 |
| }, |
| { |
| "epoch": 9.742268041237114, |
| "grad_norm": 12.220168113708496, |
| "learning_rate": 1.5021000381825126e-05, |
| "loss": 0.8806, |
| "step": 1890 |
| }, |
| { |
| "epoch": 9.896907216494846, |
| "grad_norm": 13.171895027160645, |
| "learning_rate": 1.4906452844597176e-05, |
| "loss": 0.9436, |
| "step": 1920 |
| }, |
| { |
| "epoch": 10.0, |
| "eval_bertscore": 0.8355941355537256, |
| "eval_bleu": 9.50918164746798, |
| "eval_chrf": 33.972073061006306, |
| "eval_loss": 1.590855598449707, |
| "eval_meteor": 0.29613551946552397, |
| "eval_runtime": 18.4791, |
| "eval_samples_per_second": 10.444, |
| "eval_steps_per_second": 1.353, |
| "step": 1940 |
| }, |
| { |
| "epoch": 10.051546391752577, |
| "grad_norm": 8.509225845336914, |
| "learning_rate": 1.4791905307369226e-05, |
| "loss": 0.8198, |
| "step": 1950 |
| }, |
| { |
| "epoch": 10.206185567010309, |
| "grad_norm": 11.032247543334961, |
| "learning_rate": 1.4677357770141277e-05, |
| "loss": 0.7803, |
| "step": 1980 |
| }, |
| { |
| "epoch": 10.360824742268042, |
| "grad_norm": 12.89780330657959, |
| "learning_rate": 1.4562810232913326e-05, |
| "loss": 0.7791, |
| "step": 2010 |
| }, |
| { |
| "epoch": 10.515463917525773, |
| "grad_norm": 11.072964668273926, |
| "learning_rate": 1.4448262695685378e-05, |
| "loss": 0.7733, |
| "step": 2040 |
| }, |
| { |
| "epoch": 10.670103092783505, |
| "grad_norm": 9.430943489074707, |
| "learning_rate": 1.4333715158457429e-05, |
| "loss": 0.8057, |
| "step": 2070 |
| }, |
| { |
| "epoch": 10.824742268041238, |
| "grad_norm": 10.810933113098145, |
| "learning_rate": 1.4219167621229477e-05, |
| "loss": 0.7735, |
| "step": 2100 |
| }, |
| { |
| "epoch": 10.97938144329897, |
| "grad_norm": 11.854232788085938, |
| "learning_rate": 1.4104620084001528e-05, |
| "loss": 0.7791, |
| "step": 2130 |
| }, |
| { |
| "epoch": 11.0, |
| "eval_bertscore": 0.8363540237431699, |
| "eval_bleu": 8.223251531908238, |
| "eval_chrf": 35.20734927484783, |
| "eval_loss": 1.6085046529769897, |
| "eval_meteor": 0.30757486063481376, |
| "eval_runtime": 17.4256, |
| "eval_samples_per_second": 11.076, |
| "eval_steps_per_second": 1.435, |
| "step": 2134 |
| }, |
| { |
| "epoch": 11.1340206185567, |
| "grad_norm": 11.463648796081543, |
| "learning_rate": 1.399007254677358e-05, |
| "loss": 0.7283, |
| "step": 2160 |
| }, |
| { |
| "epoch": 11.288659793814434, |
| "grad_norm": 11.482135772705078, |
| "learning_rate": 1.3875525009545629e-05, |
| "loss": 0.6976, |
| "step": 2190 |
| }, |
| { |
| "epoch": 11.443298969072165, |
| "grad_norm": 10.873443603515625, |
| "learning_rate": 1.376097747231768e-05, |
| "loss": 0.6903, |
| "step": 2220 |
| }, |
| { |
| "epoch": 11.597938144329897, |
| "grad_norm": 10.436446189880371, |
| "learning_rate": 1.3646429935089731e-05, |
| "loss": 0.7193, |
| "step": 2250 |
| }, |
| { |
| "epoch": 11.75257731958763, |
| "grad_norm": 11.538565635681152, |
| "learning_rate": 1.353188239786178e-05, |
| "loss": 0.758, |
| "step": 2280 |
| }, |
| { |
| "epoch": 11.907216494845361, |
| "grad_norm": 9.299684524536133, |
| "learning_rate": 1.341733486063383e-05, |
| "loss": 0.7079, |
| "step": 2310 |
| }, |
| { |
| "epoch": 12.0, |
| "eval_bertscore": 0.8390595162470724, |
| "eval_bleu": 10.159345704590937, |
| "eval_chrf": 35.84953170957151, |
| "eval_loss": 1.6416338682174683, |
| "eval_meteor": 0.2972449850354188, |
| "eval_runtime": 17.9183, |
| "eval_samples_per_second": 10.771, |
| "eval_steps_per_second": 1.395, |
| "step": 2328 |
| }, |
| { |
| "epoch": 12.061855670103093, |
| "grad_norm": 7.964834213256836, |
| "learning_rate": 1.3302787323405882e-05, |
| "loss": 0.6567, |
| "step": 2340 |
| }, |
| { |
| "epoch": 12.216494845360824, |
| "grad_norm": 9.399768829345703, |
| "learning_rate": 1.3188239786177931e-05, |
| "loss": 0.6519, |
| "step": 2370 |
| }, |
| { |
| "epoch": 12.371134020618557, |
| "grad_norm": 8.637943267822266, |
| "learning_rate": 1.3073692248949983e-05, |
| "loss": 0.6493, |
| "step": 2400 |
| }, |
| { |
| "epoch": 12.525773195876289, |
| "grad_norm": 9.09426498413086, |
| "learning_rate": 1.2959144711722032e-05, |
| "loss": 0.6101, |
| "step": 2430 |
| }, |
| { |
| "epoch": 12.68041237113402, |
| "grad_norm": 9.642086029052734, |
| "learning_rate": 1.2844597174494082e-05, |
| "loss": 0.6739, |
| "step": 2460 |
| }, |
| { |
| "epoch": 12.835051546391753, |
| "grad_norm": 10.607096672058105, |
| "learning_rate": 1.2730049637266133e-05, |
| "loss": 0.6254, |
| "step": 2490 |
| }, |
| { |
| "epoch": 12.989690721649485, |
| "grad_norm": 8.394698143005371, |
| "learning_rate": 1.2615502100038184e-05, |
| "loss": 0.6302, |
| "step": 2520 |
| }, |
| { |
| "epoch": 13.0, |
| "eval_bertscore": 0.8399155983035429, |
| "eval_bleu": 12.055074755933749, |
| "eval_chrf": 37.05697877497987, |
| "eval_loss": 1.5998330116271973, |
| "eval_meteor": 0.3044430922426012, |
| "eval_runtime": 16.9656, |
| "eval_samples_per_second": 11.376, |
| "eval_steps_per_second": 1.474, |
| "step": 2522 |
| }, |
| { |
| "epoch": 13.144329896907216, |
| "grad_norm": 5.606310844421387, |
| "learning_rate": 1.2500954562810234e-05, |
| "loss": 0.5787, |
| "step": 2550 |
| }, |
| { |
| "epoch": 13.29896907216495, |
| "grad_norm": 8.899696350097656, |
| "learning_rate": 1.2386407025582284e-05, |
| "loss": 0.5501, |
| "step": 2580 |
| }, |
| { |
| "epoch": 13.45360824742268, |
| "grad_norm": 8.362021446228027, |
| "learning_rate": 1.2271859488354335e-05, |
| "loss": 0.5424, |
| "step": 2610 |
| }, |
| { |
| "epoch": 13.608247422680412, |
| "grad_norm": 10.882037162780762, |
| "learning_rate": 1.2157311951126384e-05, |
| "loss": 0.5862, |
| "step": 2640 |
| }, |
| { |
| "epoch": 13.762886597938145, |
| "grad_norm": 12.276626586914062, |
| "learning_rate": 1.2042764413898436e-05, |
| "loss": 0.5826, |
| "step": 2670 |
| }, |
| { |
| "epoch": 13.917525773195877, |
| "grad_norm": 9.603856086730957, |
| "learning_rate": 1.1928216876670487e-05, |
| "loss": 0.5707, |
| "step": 2700 |
| }, |
| { |
| "epoch": 14.0, |
| "eval_bertscore": 0.8436598345405697, |
| "eval_bleu": 11.251579914916128, |
| "eval_chrf": 37.531349075403156, |
| "eval_loss": 1.6205471754074097, |
| "eval_meteor": 0.3182256654670704, |
| "eval_runtime": 17.6418, |
| "eval_samples_per_second": 10.94, |
| "eval_steps_per_second": 1.417, |
| "step": 2716 |
| }, |
| { |
| "epoch": 14.072164948453608, |
| "grad_norm": 9.971912384033203, |
| "learning_rate": 1.1813669339442537e-05, |
| "loss": 0.5769, |
| "step": 2730 |
| }, |
| { |
| "epoch": 14.22680412371134, |
| "grad_norm": 8.942472457885742, |
| "learning_rate": 1.1699121802214586e-05, |
| "loss": 0.4995, |
| "step": 2760 |
| }, |
| { |
| "epoch": 14.381443298969073, |
| "grad_norm": 9.037988662719727, |
| "learning_rate": 1.1584574264986637e-05, |
| "loss": 0.4999, |
| "step": 2790 |
| }, |
| { |
| "epoch": 14.536082474226804, |
| "grad_norm": 8.85892391204834, |
| "learning_rate": 1.1470026727758687e-05, |
| "loss": 0.5076, |
| "step": 2820 |
| }, |
| { |
| "epoch": 14.690721649484535, |
| "grad_norm": 9.50616455078125, |
| "learning_rate": 1.1355479190530738e-05, |
| "loss": 0.5365, |
| "step": 2850 |
| }, |
| { |
| "epoch": 14.845360824742269, |
| "grad_norm": 9.276251792907715, |
| "learning_rate": 1.124093165330279e-05, |
| "loss": 0.5066, |
| "step": 2880 |
| }, |
| { |
| "epoch": 15.0, |
| "grad_norm": 16.51072120666504, |
| "learning_rate": 1.1126384116074837e-05, |
| "loss": 0.5758, |
| "step": 2910 |
| }, |
| { |
| "epoch": 15.0, |
| "eval_bertscore": 0.8425244194855962, |
| "eval_bleu": 10.088393701362467, |
| "eval_chrf": 37.05532016982362, |
| "eval_loss": 1.6258227825164795, |
| "eval_meteor": 0.31680299225747693, |
| "eval_runtime": 17.8883, |
| "eval_samples_per_second": 10.789, |
| "eval_steps_per_second": 1.398, |
| "step": 2910 |
| }, |
| { |
| "epoch": 15.154639175257731, |
| "grad_norm": 9.54076099395752, |
| "learning_rate": 1.1011836578846889e-05, |
| "loss": 0.4469, |
| "step": 2940 |
| }, |
| { |
| "epoch": 15.309278350515465, |
| "grad_norm": 11.035090446472168, |
| "learning_rate": 1.089728904161894e-05, |
| "loss": 0.4619, |
| "step": 2970 |
| }, |
| { |
| "epoch": 15.463917525773196, |
| "grad_norm": 9.248506546020508, |
| "learning_rate": 1.078274150439099e-05, |
| "loss": 0.4447, |
| "step": 3000 |
| }, |
| { |
| "epoch": 15.618556701030927, |
| "grad_norm": 9.639225959777832, |
| "learning_rate": 1.066819396716304e-05, |
| "loss": 0.5252, |
| "step": 3030 |
| }, |
| { |
| "epoch": 15.77319587628866, |
| "grad_norm": 8.671944618225098, |
| "learning_rate": 1.0553646429935092e-05, |
| "loss": 0.5055, |
| "step": 3060 |
| }, |
| { |
| "epoch": 15.927835051546392, |
| "grad_norm": 9.651355743408203, |
| "learning_rate": 1.043909889270714e-05, |
| "loss": 0.4863, |
| "step": 3090 |
| }, |
| { |
| "epoch": 16.0, |
| "eval_bertscore": 0.8411734397547233, |
| "eval_bleu": 12.028464196218437, |
| "eval_chrf": 38.28723917033567, |
| "eval_loss": 1.6226447820663452, |
| "eval_meteor": 0.31954241450547227, |
| "eval_runtime": 18.5681, |
| "eval_samples_per_second": 10.394, |
| "eval_steps_per_second": 1.346, |
| "step": 3104 |
| }, |
| { |
| "epoch": 16.082474226804123, |
| "grad_norm": 8.140876770019531, |
| "learning_rate": 1.0324551355479191e-05, |
| "loss": 0.4979, |
| "step": 3120 |
| }, |
| { |
| "epoch": 16.237113402061855, |
| "grad_norm": 9.587776184082031, |
| "learning_rate": 1.0210003818251242e-05, |
| "loss": 0.4148, |
| "step": 3150 |
| }, |
| { |
| "epoch": 16.391752577319586, |
| "grad_norm": 10.633879661560059, |
| "learning_rate": 1.0095456281023292e-05, |
| "loss": 0.4426, |
| "step": 3180 |
| }, |
| { |
| "epoch": 16.54639175257732, |
| "grad_norm": 10.271921157836914, |
| "learning_rate": 9.980908743795342e-06, |
| "loss": 0.4568, |
| "step": 3210 |
| }, |
| { |
| "epoch": 16.701030927835053, |
| "grad_norm": 10.862954139709473, |
| "learning_rate": 9.866361206567393e-06, |
| "loss": 0.4211, |
| "step": 3240 |
| }, |
| { |
| "epoch": 16.855670103092784, |
| "grad_norm": 9.494423866271973, |
| "learning_rate": 9.751813669339444e-06, |
| "loss": 0.4588, |
| "step": 3270 |
| }, |
| { |
| "epoch": 17.0, |
| "eval_bertscore": 0.844244998662583, |
| "eval_bleu": 12.636168876478752, |
| "eval_chrf": 39.322891666892936, |
| "eval_loss": 1.6341180801391602, |
| "eval_meteor": 0.33544973163777175, |
| "eval_runtime": 17.7098, |
| "eval_samples_per_second": 10.898, |
| "eval_steps_per_second": 1.412, |
| "step": 3298 |
| }, |
| { |
| "epoch": 17.010309278350515, |
| "grad_norm": 8.34636116027832, |
| "learning_rate": 9.637266132111494e-06, |
| "loss": 0.4427, |
| "step": 3300 |
| }, |
| { |
| "epoch": 17.164948453608247, |
| "grad_norm": 9.896439552307129, |
| "learning_rate": 9.522718594883545e-06, |
| "loss": 0.393, |
| "step": 3330 |
| }, |
| { |
| "epoch": 17.31958762886598, |
| "grad_norm": 8.447748184204102, |
| "learning_rate": 9.408171057655595e-06, |
| "loss": 0.4072, |
| "step": 3360 |
| }, |
| { |
| "epoch": 17.47422680412371, |
| "grad_norm": 8.587246894836426, |
| "learning_rate": 9.293623520427644e-06, |
| "loss": 0.3783, |
| "step": 3390 |
| }, |
| { |
| "epoch": 17.628865979381445, |
| "grad_norm": 8.51424503326416, |
| "learning_rate": 9.179075983199695e-06, |
| "loss": 0.3923, |
| "step": 3420 |
| }, |
| { |
| "epoch": 17.783505154639176, |
| "grad_norm": 11.139188766479492, |
| "learning_rate": 9.064528445971745e-06, |
| "loss": 0.414, |
| "step": 3450 |
| }, |
| { |
| "epoch": 17.938144329896907, |
| "grad_norm": 10.412823677062988, |
| "learning_rate": 8.949980908743796e-06, |
| "loss": 0.4256, |
| "step": 3480 |
| }, |
| { |
| "epoch": 18.0, |
| "eval_bertscore": 0.8427423143016242, |
| "eval_bleu": 10.39081838124178, |
| "eval_chrf": 38.315665164919935, |
| "eval_loss": 1.651857852935791, |
| "eval_meteor": 0.3235056356864544, |
| "eval_runtime": 18.083, |
| "eval_samples_per_second": 10.673, |
| "eval_steps_per_second": 1.383, |
| "step": 3492 |
| }, |
| { |
| "epoch": 18.09278350515464, |
| "grad_norm": 7.963134288787842, |
| "learning_rate": 8.835433371515847e-06, |
| "loss": 0.3509, |
| "step": 3510 |
| }, |
| { |
| "epoch": 18.24742268041237, |
| "grad_norm": 6.555767059326172, |
| "learning_rate": 8.720885834287897e-06, |
| "loss": 0.3438, |
| "step": 3540 |
| }, |
| { |
| "epoch": 18.4020618556701, |
| "grad_norm": 9.116894721984863, |
| "learning_rate": 8.606338297059947e-06, |
| "loss": 0.3634, |
| "step": 3570 |
| }, |
| { |
| "epoch": 18.556701030927837, |
| "grad_norm": 8.772259712219238, |
| "learning_rate": 8.491790759831998e-06, |
| "loss": 0.3688, |
| "step": 3600 |
| }, |
| { |
| "epoch": 18.711340206185568, |
| "grad_norm": 6.532469272613525, |
| "learning_rate": 8.377243222604047e-06, |
| "loss": 0.3826, |
| "step": 3630 |
| }, |
| { |
| "epoch": 18.8659793814433, |
| "grad_norm": 8.094025611877441, |
| "learning_rate": 8.262695685376099e-06, |
| "loss": 0.4, |
| "step": 3660 |
| }, |
| { |
| "epoch": 19.0, |
| "eval_bertscore": 0.8423587293822531, |
| "eval_bleu": 12.487375649363784, |
| "eval_chrf": 39.54931349514667, |
| "eval_loss": 1.660538673400879, |
| "eval_meteor": 0.32780584893530307, |
| "eval_runtime": 19.1569, |
| "eval_samples_per_second": 10.075, |
| "eval_steps_per_second": 1.305, |
| "step": 3686 |
| }, |
| { |
| "epoch": 19.02061855670103, |
| "grad_norm": 6.79367733001709, |
| "learning_rate": 8.148148148148148e-06, |
| "loss": 0.3852, |
| "step": 3690 |
| }, |
| { |
| "epoch": 19.175257731958762, |
| "grad_norm": 9.372544288635254, |
| "learning_rate": 8.0336006109202e-06, |
| "loss": 0.3319, |
| "step": 3720 |
| }, |
| { |
| "epoch": 19.329896907216494, |
| "grad_norm": 10.440468788146973, |
| "learning_rate": 7.91905307369225e-06, |
| "loss": 0.3565, |
| "step": 3750 |
| }, |
| { |
| "epoch": 19.484536082474225, |
| "grad_norm": 8.199675559997559, |
| "learning_rate": 7.8045055364643e-06, |
| "loss": 0.3275, |
| "step": 3780 |
| }, |
| { |
| "epoch": 19.63917525773196, |
| "grad_norm": 9.151398658752441, |
| "learning_rate": 7.68995799923635e-06, |
| "loss": 0.3809, |
| "step": 3810 |
| }, |
| { |
| "epoch": 19.79381443298969, |
| "grad_norm": 8.963874816894531, |
| "learning_rate": 7.575410462008401e-06, |
| "loss": 0.3046, |
| "step": 3840 |
| }, |
| { |
| "epoch": 19.948453608247423, |
| "grad_norm": 7.874151229858398, |
| "learning_rate": 7.460862924780451e-06, |
| "loss": 0.3521, |
| "step": 3870 |
| }, |
| { |
| "epoch": 20.0, |
| "eval_bertscore": 0.8462644858681476, |
| "eval_bleu": 11.879411636317712, |
| "eval_chrf": 40.114348533135825, |
| "eval_loss": 1.6695095300674438, |
| "eval_meteor": 0.33837011315826326, |
| "eval_runtime": 17.5347, |
| "eval_samples_per_second": 11.007, |
| "eval_steps_per_second": 1.426, |
| "step": 3880 |
| }, |
| { |
| "epoch": 20.103092783505154, |
| "grad_norm": 6.499141216278076, |
| "learning_rate": 7.346315387552502e-06, |
| "loss": 0.3306, |
| "step": 3900 |
| }, |
| { |
| "epoch": 20.257731958762886, |
| "grad_norm": 6.233891487121582, |
| "learning_rate": 7.231767850324552e-06, |
| "loss": 0.3199, |
| "step": 3930 |
| }, |
| { |
| "epoch": 20.412371134020617, |
| "grad_norm": 9.419313430786133, |
| "learning_rate": 7.117220313096602e-06, |
| "loss": 0.3009, |
| "step": 3960 |
| }, |
| { |
| "epoch": 20.567010309278352, |
| "grad_norm": 9.547579765319824, |
| "learning_rate": 7.002672775868653e-06, |
| "loss": 0.3036, |
| "step": 3990 |
| }, |
| { |
| "epoch": 20.721649484536083, |
| "grad_norm": 6.532684326171875, |
| "learning_rate": 6.888125238640703e-06, |
| "loss": 0.322, |
| "step": 4020 |
| }, |
| { |
| "epoch": 20.876288659793815, |
| "grad_norm": 8.454936027526855, |
| "learning_rate": 6.773577701412753e-06, |
| "loss": 0.3298, |
| "step": 4050 |
| }, |
| { |
| "epoch": 21.0, |
| "eval_bertscore": 0.843647470128351, |
| "eval_bleu": 12.22598647677628, |
| "eval_chrf": 39.967488522101775, |
| "eval_loss": 1.6959513425827026, |
| "eval_meteor": 0.334625348793614, |
| "eval_runtime": 17.9122, |
| "eval_samples_per_second": 10.775, |
| "eval_steps_per_second": 1.396, |
| "step": 4074 |
| }, |
| { |
| "epoch": 21.030927835051546, |
| "grad_norm": 7.442008018493652, |
| "learning_rate": 6.659030164184805e-06, |
| "loss": 0.3098, |
| "step": 4080 |
| }, |
| { |
| "epoch": 21.185567010309278, |
| "grad_norm": 9.09302043914795, |
| "learning_rate": 6.544482626956854e-06, |
| "loss": 0.3093, |
| "step": 4110 |
| }, |
| { |
| "epoch": 21.34020618556701, |
| "grad_norm": 7.801342010498047, |
| "learning_rate": 6.429935089728905e-06, |
| "loss": 0.3066, |
| "step": 4140 |
| }, |
| { |
| "epoch": 21.49484536082474, |
| "grad_norm": 7.818614482879639, |
| "learning_rate": 6.315387552500956e-06, |
| "loss": 0.2984, |
| "step": 4170 |
| }, |
| { |
| "epoch": 21.649484536082475, |
| "grad_norm": 8.459864616394043, |
| "learning_rate": 6.2008400152730055e-06, |
| "loss": 0.2809, |
| "step": 4200 |
| }, |
| { |
| "epoch": 21.804123711340207, |
| "grad_norm": 8.437562942504883, |
| "learning_rate": 6.086292478045056e-06, |
| "loss": 0.2973, |
| "step": 4230 |
| }, |
| { |
| "epoch": 21.95876288659794, |
| "grad_norm": 10.7759428024292, |
| "learning_rate": 5.9717449408171055e-06, |
| "loss": 0.3029, |
| "step": 4260 |
| }, |
| { |
| "epoch": 22.0, |
| "eval_bertscore": 0.8437702862709915, |
| "eval_bleu": 11.328363034379192, |
| "eval_chrf": 40.31840295732892, |
| "eval_loss": 1.697576880455017, |
| "eval_meteor": 0.33887718936757705, |
| "eval_runtime": 18.6408, |
| "eval_samples_per_second": 10.354, |
| "eval_steps_per_second": 1.341, |
| "step": 4268 |
| }, |
| { |
| "epoch": 22.0, |
| "step": 4268, |
| "total_flos": 232218100629504.0, |
| "train_loss": 1.1187503173626128, |
| "train_runtime": 1520.2415, |
| "train_samples_per_second": 30.528, |
| "train_steps_per_second": 3.828 |
| } |
| ], |
| "logging_steps": 30, |
| "max_steps": 5820, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 30, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 5, |
| "early_stopping_threshold": 0.0 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 5 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 232218100629504.0, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
