| { | |
| "best_metric": 3.4803478717803955, | |
| "best_model_checkpoint": "opus-mt-zh-en-finetuned-zh-to-en/checkpoint-2", | |
| "epoch": 100.0, | |
| "eval_steps": 500, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_bleu": 5.6918, | |
| "eval_gen_len": 57.3333, | |
| "eval_loss": 3.5217063426971436, | |
| "eval_runtime": 10.9763, | |
| "eval_samples_per_second": 0.273, | |
| "eval_steps_per_second": 0.091, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bleu": 7.6304, | |
| "eval_gen_len": 55.3333, | |
| "eval_loss": 3.4803478717803955, | |
| "eval_runtime": 8.3555, | |
| "eval_samples_per_second": 0.359, | |
| "eval_steps_per_second": 0.12, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bleu": 7.9453, | |
| "eval_gen_len": 56.0, | |
| "eval_loss": 3.491027593612671, | |
| "eval_runtime": 9.8597, | |
| "eval_samples_per_second": 0.304, | |
| "eval_steps_per_second": 0.101, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bleu": 6.5495, | |
| "eval_gen_len": 79.0, | |
| "eval_loss": 3.5166053771972656, | |
| "eval_runtime": 13.007, | |
| "eval_samples_per_second": 0.231, | |
| "eval_steps_per_second": 0.077, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bleu": 6.6932, | |
| "eval_gen_len": 80.0, | |
| "eval_loss": 3.5517735481262207, | |
| "eval_runtime": 14.9524, | |
| "eval_samples_per_second": 0.201, | |
| "eval_steps_per_second": 0.067, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bleu": 5.9333, | |
| "eval_gen_len": 137.0, | |
| "eval_loss": 3.5805959701538086, | |
| "eval_runtime": 61.8901, | |
| "eval_samples_per_second": 0.048, | |
| "eval_steps_per_second": 0.016, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_bleu": 5.6605, | |
| "eval_gen_len": 133.6667, | |
| "eval_loss": 3.6102888584136963, | |
| "eval_runtime": 55.757, | |
| "eval_samples_per_second": 0.054, | |
| "eval_steps_per_second": 0.018, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_bleu": 7.1686, | |
| "eval_gen_len": 135.3333, | |
| "eval_loss": 3.645458936691284, | |
| "eval_runtime": 52.3901, | |
| "eval_samples_per_second": 0.057, | |
| "eval_steps_per_second": 0.019, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_bleu": 9.8037, | |
| "eval_gen_len": 146.0, | |
| "eval_loss": 3.6737775802612305, | |
| "eval_runtime": 50.7208, | |
| "eval_samples_per_second": 0.059, | |
| "eval_steps_per_second": 0.02, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_bleu": 9.9552, | |
| "eval_gen_len": 139.0, | |
| "eval_loss": 3.707125186920166, | |
| "eval_runtime": 49.5408, | |
| "eval_samples_per_second": 0.061, | |
| "eval_steps_per_second": 0.02, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_bleu": 10.311, | |
| "eval_gen_len": 134.0, | |
| "eval_loss": 3.7424490451812744, | |
| "eval_runtime": 45.4229, | |
| "eval_samples_per_second": 0.066, | |
| "eval_steps_per_second": 0.022, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_bleu": 10.1929, | |
| "eval_gen_len": 131.6667, | |
| "eval_loss": 3.786301612854004, | |
| "eval_runtime": 45.6583, | |
| "eval_samples_per_second": 0.066, | |
| "eval_steps_per_second": 0.022, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_bleu": 9.8671, | |
| "eval_gen_len": 119.3333, | |
| "eval_loss": 3.825416088104248, | |
| "eval_runtime": 44.3396, | |
| "eval_samples_per_second": 0.068, | |
| "eval_steps_per_second": 0.023, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_bleu": 11.4259, | |
| "eval_gen_len": 107.3333, | |
| "eval_loss": 3.866556167602539, | |
| "eval_runtime": 60.118, | |
| "eval_samples_per_second": 0.05, | |
| "eval_steps_per_second": 0.017, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_bleu": 11.4172, | |
| "eval_gen_len": 100.6667, | |
| "eval_loss": 3.900843620300293, | |
| "eval_runtime": 56.7929, | |
| "eval_samples_per_second": 0.053, | |
| "eval_steps_per_second": 0.018, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_bleu": 10.8738, | |
| "eval_gen_len": 117.6667, | |
| "eval_loss": 3.935436964035034, | |
| "eval_runtime": 38.5973, | |
| "eval_samples_per_second": 0.078, | |
| "eval_steps_per_second": 0.026, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_bleu": 10.7717, | |
| "eval_gen_len": 117.0, | |
| "eval_loss": 3.9768829345703125, | |
| "eval_runtime": 36.4507, | |
| "eval_samples_per_second": 0.082, | |
| "eval_steps_per_second": 0.027, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_bleu": 9.9034, | |
| "eval_gen_len": 118.6667, | |
| "eval_loss": 4.017246246337891, | |
| "eval_runtime": 36.6426, | |
| "eval_samples_per_second": 0.082, | |
| "eval_steps_per_second": 0.027, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_bleu": 10.0376, | |
| "eval_gen_len": 121.6667, | |
| "eval_loss": 4.05275821685791, | |
| "eval_runtime": 34.9881, | |
| "eval_samples_per_second": 0.086, | |
| "eval_steps_per_second": 0.029, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_bleu": 8.6054, | |
| "eval_gen_len": 61.0, | |
| "eval_loss": 4.080726623535156, | |
| "eval_runtime": 9.9523, | |
| "eval_samples_per_second": 0.301, | |
| "eval_steps_per_second": 0.1, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_bleu": 10.9496, | |
| "eval_gen_len": 67.0, | |
| "eval_loss": 4.109140396118164, | |
| "eval_runtime": 11.0517, | |
| "eval_samples_per_second": 0.271, | |
| "eval_steps_per_second": 0.09, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_bleu": 10.8268, | |
| "eval_gen_len": 65.6667, | |
| "eval_loss": 4.133188247680664, | |
| "eval_runtime": 10.5163, | |
| "eval_samples_per_second": 0.285, | |
| "eval_steps_per_second": 0.095, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_bleu": 10.5556, | |
| "eval_gen_len": 63.6667, | |
| "eval_loss": 4.158883571624756, | |
| "eval_runtime": 9.4741, | |
| "eval_samples_per_second": 0.317, | |
| "eval_steps_per_second": 0.106, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_bleu": 10.8678, | |
| "eval_gen_len": 62.0, | |
| "eval_loss": 4.181970119476318, | |
| "eval_runtime": 10.5606, | |
| "eval_samples_per_second": 0.284, | |
| "eval_steps_per_second": 0.095, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_bleu": 10.8678, | |
| "eval_gen_len": 61.6667, | |
| "eval_loss": 4.196631908416748, | |
| "eval_runtime": 9.7369, | |
| "eval_samples_per_second": 0.308, | |
| "eval_steps_per_second": 0.103, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_bleu": 10.27, | |
| "eval_gen_len": 61.6667, | |
| "eval_loss": 4.209479808807373, | |
| "eval_runtime": 9.9007, | |
| "eval_samples_per_second": 0.303, | |
| "eval_steps_per_second": 0.101, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_bleu": 10.6577, | |
| "eval_gen_len": 60.0, | |
| "eval_loss": 4.220566272735596, | |
| "eval_runtime": 9.9798, | |
| "eval_samples_per_second": 0.301, | |
| "eval_steps_per_second": 0.1, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_bleu": 10.2137, | |
| "eval_gen_len": 60.6667, | |
| "eval_loss": 4.2330474853515625, | |
| "eval_runtime": 11.0027, | |
| "eval_samples_per_second": 0.273, | |
| "eval_steps_per_second": 0.091, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_bleu": 10.0108, | |
| "eval_gen_len": 62.0, | |
| "eval_loss": 4.245401382446289, | |
| "eval_runtime": 10.7321, | |
| "eval_samples_per_second": 0.28, | |
| "eval_steps_per_second": 0.093, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_bleu": 10.6984, | |
| "eval_gen_len": 60.0, | |
| "eval_loss": 4.257299423217773, | |
| "eval_runtime": 9.8765, | |
| "eval_samples_per_second": 0.304, | |
| "eval_steps_per_second": 0.101, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_bleu": 9.6494, | |
| "eval_gen_len": 83.0, | |
| "eval_loss": 4.2664995193481445, | |
| "eval_runtime": 13.2171, | |
| "eval_samples_per_second": 0.227, | |
| "eval_steps_per_second": 0.076, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_bleu": 9.7503, | |
| "eval_gen_len": 84.3333, | |
| "eval_loss": 4.276342868804932, | |
| "eval_runtime": 12.9373, | |
| "eval_samples_per_second": 0.232, | |
| "eval_steps_per_second": 0.077, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_bleu": 9.7503, | |
| "eval_gen_len": 84.3333, | |
| "eval_loss": 4.287381649017334, | |
| "eval_runtime": 12.9473, | |
| "eval_samples_per_second": 0.232, | |
| "eval_steps_per_second": 0.077, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_bleu": 9.7503, | |
| "eval_gen_len": 84.3333, | |
| "eval_loss": 4.295342445373535, | |
| "eval_runtime": 13.0533, | |
| "eval_samples_per_second": 0.23, | |
| "eval_steps_per_second": 0.077, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_bleu": 9.7503, | |
| "eval_gen_len": 84.3333, | |
| "eval_loss": 4.301455020904541, | |
| "eval_runtime": 13.2102, | |
| "eval_samples_per_second": 0.227, | |
| "eval_steps_per_second": 0.076, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_bleu": 9.4923, | |
| "eval_gen_len": 70.0, | |
| "eval_loss": 4.303869247436523, | |
| "eval_runtime": 105.6574, | |
| "eval_samples_per_second": 0.028, | |
| "eval_steps_per_second": 0.009, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_bleu": 9.7956, | |
| "eval_gen_len": 64.6667, | |
| "eval_loss": 4.308076858520508, | |
| "eval_runtime": 86.8335, | |
| "eval_samples_per_second": 0.035, | |
| "eval_steps_per_second": 0.012, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_bleu": 10.3592, | |
| "eval_gen_len": 70.6667, | |
| "eval_loss": 4.311928749084473, | |
| "eval_runtime": 10.4025, | |
| "eval_samples_per_second": 0.288, | |
| "eval_steps_per_second": 0.096, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_bleu": 11.5296, | |
| "eval_gen_len": 86.0, | |
| "eval_loss": 4.315915584564209, | |
| "eval_runtime": 13.3622, | |
| "eval_samples_per_second": 0.225, | |
| "eval_steps_per_second": 0.075, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_bleu": 11.1681, | |
| "eval_gen_len": 78.0, | |
| "eval_loss": 4.319802284240723, | |
| "eval_runtime": 13.5231, | |
| "eval_samples_per_second": 0.222, | |
| "eval_steps_per_second": 0.074, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_bleu": 10.2925, | |
| "eval_gen_len": 82.6667, | |
| "eval_loss": 4.323220252990723, | |
| "eval_runtime": 13.3207, | |
| "eval_samples_per_second": 0.225, | |
| "eval_steps_per_second": 0.075, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_bleu": 11.2059, | |
| "eval_gen_len": 82.3333, | |
| "eval_loss": 4.325761795043945, | |
| "eval_runtime": 16.8934, | |
| "eval_samples_per_second": 0.178, | |
| "eval_steps_per_second": 0.059, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_bleu": 11.1664, | |
| "eval_gen_len": 85.3333, | |
| "eval_loss": 4.327906131744385, | |
| "eval_runtime": 14.4307, | |
| "eval_samples_per_second": 0.208, | |
| "eval_steps_per_second": 0.069, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_bleu": 11.2609, | |
| "eval_gen_len": 81.6667, | |
| "eval_loss": 4.329981803894043, | |
| "eval_runtime": 14.847, | |
| "eval_samples_per_second": 0.202, | |
| "eval_steps_per_second": 0.067, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_bleu": 11.2609, | |
| "eval_gen_len": 81.6667, | |
| "eval_loss": 4.332195281982422, | |
| "eval_runtime": 15.2747, | |
| "eval_samples_per_second": 0.196, | |
| "eval_steps_per_second": 0.065, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_bleu": 11.0203, | |
| "eval_gen_len": 90.3333, | |
| "eval_loss": 4.3332905769348145, | |
| "eval_runtime": 15.7496, | |
| "eval_samples_per_second": 0.19, | |
| "eval_steps_per_second": 0.063, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_bleu": 11.1664, | |
| "eval_gen_len": 85.3333, | |
| "eval_loss": 4.3369364738464355, | |
| "eval_runtime": 14.8004, | |
| "eval_samples_per_second": 0.203, | |
| "eval_steps_per_second": 0.068, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_bleu": 11.3295, | |
| "eval_gen_len": 82.0, | |
| "eval_loss": 4.340444564819336, | |
| "eval_runtime": 13.1824, | |
| "eval_samples_per_second": 0.228, | |
| "eval_steps_per_second": 0.076, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_bleu": 11.1364, | |
| "eval_gen_len": 68.3333, | |
| "eval_loss": 4.343842506408691, | |
| "eval_runtime": 12.8795, | |
| "eval_samples_per_second": 0.233, | |
| "eval_steps_per_second": 0.078, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_bleu": 11.1364, | |
| "eval_gen_len": 68.3333, | |
| "eval_loss": 4.347080230712891, | |
| "eval_runtime": 13.1248, | |
| "eval_samples_per_second": 0.229, | |
| "eval_steps_per_second": 0.076, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_bleu": 11.1208, | |
| "eval_gen_len": 65.6667, | |
| "eval_loss": 4.349867820739746, | |
| "eval_runtime": 11.6147, | |
| "eval_samples_per_second": 0.258, | |
| "eval_steps_per_second": 0.086, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_bleu": 11.1208, | |
| "eval_gen_len": 65.6667, | |
| "eval_loss": 4.352280616760254, | |
| "eval_runtime": 11.3187, | |
| "eval_samples_per_second": 0.265, | |
| "eval_steps_per_second": 0.088, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_bleu": 11.1208, | |
| "eval_gen_len": 65.6667, | |
| "eval_loss": 4.353754997253418, | |
| "eval_runtime": 12.1023, | |
| "eval_samples_per_second": 0.248, | |
| "eval_steps_per_second": 0.083, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_bleu": 10.9215, | |
| "eval_gen_len": 66.0, | |
| "eval_loss": 4.355079174041748, | |
| "eval_runtime": 10.6524, | |
| "eval_samples_per_second": 0.282, | |
| "eval_steps_per_second": 0.094, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_bleu": 11.0309, | |
| "eval_gen_len": 72.0, | |
| "eval_loss": 4.356247901916504, | |
| "eval_runtime": 14.1128, | |
| "eval_samples_per_second": 0.213, | |
| "eval_steps_per_second": 0.071, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_bleu": 11.0309, | |
| "eval_gen_len": 72.0, | |
| "eval_loss": 4.357237815856934, | |
| "eval_runtime": 14.0892, | |
| "eval_samples_per_second": 0.213, | |
| "eval_steps_per_second": 0.071, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_bleu": 10.678, | |
| "eval_gen_len": 71.3333, | |
| "eval_loss": 4.358083248138428, | |
| "eval_runtime": 12.7806, | |
| "eval_samples_per_second": 0.235, | |
| "eval_steps_per_second": 0.078, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_bleu": 10.678, | |
| "eval_gen_len": 71.3333, | |
| "eval_loss": 4.358962059020996, | |
| "eval_runtime": 12.8538, | |
| "eval_samples_per_second": 0.233, | |
| "eval_steps_per_second": 0.078, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_bleu": 10.678, | |
| "eval_gen_len": 71.3333, | |
| "eval_loss": 4.359888076782227, | |
| "eval_runtime": 12.3355, | |
| "eval_samples_per_second": 0.243, | |
| "eval_steps_per_second": 0.081, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_bleu": 10.678, | |
| "eval_gen_len": 71.3333, | |
| "eval_loss": 4.360787391662598, | |
| "eval_runtime": 12.7009, | |
| "eval_samples_per_second": 0.236, | |
| "eval_steps_per_second": 0.079, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_bleu": 10.4416, | |
| "eval_gen_len": 86.3333, | |
| "eval_loss": 4.36198091506958, | |
| "eval_runtime": 12.4279, | |
| "eval_samples_per_second": 0.241, | |
| "eval_steps_per_second": 0.08, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_bleu": 10.678, | |
| "eval_gen_len": 71.3333, | |
| "eval_loss": 4.362994194030762, | |
| "eval_runtime": 12.6907, | |
| "eval_samples_per_second": 0.236, | |
| "eval_steps_per_second": 0.079, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_bleu": 10.678, | |
| "eval_gen_len": 71.3333, | |
| "eval_loss": 4.366813659667969, | |
| "eval_runtime": 12.6482, | |
| "eval_samples_per_second": 0.237, | |
| "eval_steps_per_second": 0.079, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_bleu": 11.1601, | |
| "eval_gen_len": 74.6667, | |
| "eval_loss": 4.370214462280273, | |
| "eval_runtime": 11.7414, | |
| "eval_samples_per_second": 0.256, | |
| "eval_steps_per_second": 0.085, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_bleu": 11.137, | |
| "eval_gen_len": 78.0, | |
| "eval_loss": 4.37328577041626, | |
| "eval_runtime": 13.9564, | |
| "eval_samples_per_second": 0.215, | |
| "eval_steps_per_second": 0.072, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_bleu": 11.137, | |
| "eval_gen_len": 78.0, | |
| "eval_loss": 4.3749165534973145, | |
| "eval_runtime": 13.8294, | |
| "eval_samples_per_second": 0.217, | |
| "eval_steps_per_second": 0.072, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_bleu": 11.137, | |
| "eval_gen_len": 78.0, | |
| "eval_loss": 4.3767194747924805, | |
| "eval_runtime": 13.7066, | |
| "eval_samples_per_second": 0.219, | |
| "eval_steps_per_second": 0.073, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_bleu": 10.8916, | |
| "eval_gen_len": 88.0, | |
| "eval_loss": 4.3782854080200195, | |
| "eval_runtime": 16.3165, | |
| "eval_samples_per_second": 0.184, | |
| "eval_steps_per_second": 0.061, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_bleu": 10.7733, | |
| "eval_gen_len": 67.3333, | |
| "eval_loss": 4.379751682281494, | |
| "eval_runtime": 11.7272, | |
| "eval_samples_per_second": 0.256, | |
| "eval_steps_per_second": 0.085, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_bleu": 11.5509, | |
| "eval_gen_len": 66.3333, | |
| "eval_loss": 4.381042957305908, | |
| "eval_runtime": 11.8594, | |
| "eval_samples_per_second": 0.253, | |
| "eval_steps_per_second": 0.084, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_bleu": 11.4262, | |
| "eval_gen_len": 70.0, | |
| "eval_loss": 4.382328510284424, | |
| "eval_runtime": 13.0297, | |
| "eval_samples_per_second": 0.23, | |
| "eval_steps_per_second": 0.077, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_bleu": 11.4262, | |
| "eval_gen_len": 70.0, | |
| "eval_loss": 4.383648872375488, | |
| "eval_runtime": 14.0598, | |
| "eval_samples_per_second": 0.213, | |
| "eval_steps_per_second": 0.071, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_bleu": 11.195, | |
| "eval_gen_len": 66.3333, | |
| "eval_loss": 4.384763717651367, | |
| "eval_runtime": 11.4527, | |
| "eval_samples_per_second": 0.262, | |
| "eval_steps_per_second": 0.087, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_bleu": 11.1115, | |
| "eval_gen_len": 67.0, | |
| "eval_loss": 4.384941577911377, | |
| "eval_runtime": 12.3777, | |
| "eval_samples_per_second": 0.242, | |
| "eval_steps_per_second": 0.081, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_bleu": 11.2741, | |
| "eval_gen_len": 67.6667, | |
| "eval_loss": 4.385134696960449, | |
| "eval_runtime": 12.3325, | |
| "eval_samples_per_second": 0.243, | |
| "eval_steps_per_second": 0.081, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_bleu": 11.6137, | |
| "eval_gen_len": 71.0, | |
| "eval_loss": 4.38496208190918, | |
| "eval_runtime": 12.6988, | |
| "eval_samples_per_second": 0.236, | |
| "eval_steps_per_second": 0.079, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_bleu": 11.2741, | |
| "eval_gen_len": 67.6667, | |
| "eval_loss": 4.3848419189453125, | |
| "eval_runtime": 11.4524, | |
| "eval_samples_per_second": 0.262, | |
| "eval_steps_per_second": 0.087, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_bleu": 11.1454, | |
| "eval_gen_len": 62.0, | |
| "eval_loss": 4.38457727432251, | |
| "eval_runtime": 10.4375, | |
| "eval_samples_per_second": 0.287, | |
| "eval_steps_per_second": 0.096, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_bleu": 11.1454, | |
| "eval_gen_len": 62.0, | |
| "eval_loss": 4.3842973709106445, | |
| "eval_runtime": 10.4146, | |
| "eval_samples_per_second": 0.288, | |
| "eval_steps_per_second": 0.096, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_bleu": 11.267, | |
| "eval_gen_len": 68.6667, | |
| "eval_loss": 4.383995532989502, | |
| "eval_runtime": 12.6896, | |
| "eval_samples_per_second": 0.236, | |
| "eval_steps_per_second": 0.079, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 81.0, | |
| "eval_bleu": 11.267, | |
| "eval_gen_len": 68.6667, | |
| "eval_loss": 4.383782386779785, | |
| "eval_runtime": 12.4982, | |
| "eval_samples_per_second": 0.24, | |
| "eval_steps_per_second": 0.08, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 82.0, | |
| "eval_bleu": 11.267, | |
| "eval_gen_len": 68.6667, | |
| "eval_loss": 4.383605480194092, | |
| "eval_runtime": 12.2559, | |
| "eval_samples_per_second": 0.245, | |
| "eval_steps_per_second": 0.082, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 83.0, | |
| "eval_bleu": 11.267, | |
| "eval_gen_len": 68.6667, | |
| "eval_loss": 4.383455753326416, | |
| "eval_runtime": 12.7951, | |
| "eval_samples_per_second": 0.234, | |
| "eval_steps_per_second": 0.078, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 84.0, | |
| "eval_bleu": 11.267, | |
| "eval_gen_len": 68.6667, | |
| "eval_loss": 4.38338565826416, | |
| "eval_runtime": 12.5193, | |
| "eval_samples_per_second": 0.24, | |
| "eval_steps_per_second": 0.08, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 85.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.383331298828125, | |
| "eval_runtime": 12.8027, | |
| "eval_samples_per_second": 0.234, | |
| "eval_steps_per_second": 0.078, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 86.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.38341760635376, | |
| "eval_runtime": 12.849, | |
| "eval_samples_per_second": 0.233, | |
| "eval_steps_per_second": 0.078, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 87.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.383518695831299, | |
| "eval_runtime": 12.3575, | |
| "eval_samples_per_second": 0.243, | |
| "eval_steps_per_second": 0.081, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 88.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.3836445808410645, | |
| "eval_runtime": 12.8174, | |
| "eval_samples_per_second": 0.234, | |
| "eval_steps_per_second": 0.078, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 89.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.383749961853027, | |
| "eval_runtime": 12.8695, | |
| "eval_samples_per_second": 0.233, | |
| "eval_steps_per_second": 0.078, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 90.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.383831977844238, | |
| "eval_runtime": 12.8527, | |
| "eval_samples_per_second": 0.233, | |
| "eval_steps_per_second": 0.078, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 91.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.383894920349121, | |
| "eval_runtime": 12.7272, | |
| "eval_samples_per_second": 0.236, | |
| "eval_steps_per_second": 0.079, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 92.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.383920669555664, | |
| "eval_runtime": 14.3034, | |
| "eval_samples_per_second": 0.21, | |
| "eval_steps_per_second": 0.07, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 93.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.38393497467041, | |
| "eval_runtime": 12.4454, | |
| "eval_samples_per_second": 0.241, | |
| "eval_steps_per_second": 0.08, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 94.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.384005546569824, | |
| "eval_runtime": 12.8192, | |
| "eval_samples_per_second": 0.234, | |
| "eval_steps_per_second": 0.078, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 95.0, | |
| "eval_bleu": 11.4092, | |
| "eval_gen_len": 69.0, | |
| "eval_loss": 4.3840460777282715, | |
| "eval_runtime": 12.4402, | |
| "eval_samples_per_second": 0.241, | |
| "eval_steps_per_second": 0.08, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 96.0, | |
| "eval_bleu": 10.7891, | |
| "eval_gen_len": 70.6667, | |
| "eval_loss": 4.384109020233154, | |
| "eval_runtime": 12.8195, | |
| "eval_samples_per_second": 0.234, | |
| "eval_steps_per_second": 0.078, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 97.0, | |
| "eval_bleu": 10.7891, | |
| "eval_gen_len": 70.6667, | |
| "eval_loss": 4.384162902832031, | |
| "eval_runtime": 12.3694, | |
| "eval_samples_per_second": 0.243, | |
| "eval_steps_per_second": 0.081, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 98.0, | |
| "eval_bleu": 10.7891, | |
| "eval_gen_len": 70.6667, | |
| "eval_loss": 4.384200096130371, | |
| "eval_runtime": 12.8132, | |
| "eval_samples_per_second": 0.234, | |
| "eval_steps_per_second": 0.078, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 99.0, | |
| "eval_bleu": 10.7891, | |
| "eval_gen_len": 70.6667, | |
| "eval_loss": 4.384230613708496, | |
| "eval_runtime": 12.6103, | |
| "eval_samples_per_second": 0.238, | |
| "eval_steps_per_second": 0.079, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 100.0, | |
| "eval_bleu": 10.7891, | |
| "eval_gen_len": 70.6667, | |
| "eval_loss": 4.384246349334717, | |
| "eval_runtime": 12.8394, | |
| "eval_samples_per_second": 0.234, | |
| "eval_steps_per_second": 0.078, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 100, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 7786030694400.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |