| { | |
| "best_metric": 42.9732, | |
| "best_model_checkpoint": "opus_base_wce_adaptified/checkpoint-80000", | |
| "epoch": 3.943217665615142, | |
| "global_step": 80000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9753918572555207e-05, | |
| "loss": 0.2051, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_bleu": 35.5597, | |
| "eval_gen_len": 39.8495, | |
| "eval_loss": 0.10906314104795456, | |
| "eval_runtime": 140.7392, | |
| "eval_samples_per_second": 7.411, | |
| "eval_steps_per_second": 0.234, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.950746746845426e-05, | |
| "loss": 0.1719, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "eval_bleu": 40.9319, | |
| "eval_gen_len": 34.6337, | |
| "eval_loss": 0.10657753050327301, | |
| "eval_runtime": 125.8016, | |
| "eval_samples_per_second": 8.291, | |
| "eval_steps_per_second": 0.262, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.9261016364353314e-05, | |
| "loss": 0.165, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "eval_bleu": 41.3844, | |
| "eval_gen_len": 34.6453, | |
| "eval_loss": 0.10533162206411362, | |
| "eval_runtime": 111.0041, | |
| "eval_samples_per_second": 9.396, | |
| "eval_steps_per_second": 0.297, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.9014565260252367e-05, | |
| "loss": 0.1596, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "eval_bleu": 41.3616, | |
| "eval_gen_len": 35.1601, | |
| "eval_loss": 0.10429085791110992, | |
| "eval_runtime": 141.5674, | |
| "eval_samples_per_second": 7.368, | |
| "eval_steps_per_second": 0.233, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.876811415615142e-05, | |
| "loss": 0.1573, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_bleu": 41.6835, | |
| "eval_gen_len": 34.9942, | |
| "eval_loss": 0.10373106598854065, | |
| "eval_runtime": 109.7842, | |
| "eval_samples_per_second": 9.5, | |
| "eval_steps_per_second": 0.301, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.8521786277602524e-05, | |
| "loss": 0.1457, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "eval_bleu": 41.1613, | |
| "eval_gen_len": 36.2349, | |
| "eval_loss": 0.10356967151165009, | |
| "eval_runtime": 116.302, | |
| "eval_samples_per_second": 8.968, | |
| "eval_steps_per_second": 0.284, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.827545839905363e-05, | |
| "loss": 0.1455, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "eval_bleu": 41.8822, | |
| "eval_gen_len": 34.4698, | |
| "eval_loss": 0.10275906324386597, | |
| "eval_runtime": 88.2743, | |
| "eval_samples_per_second": 11.815, | |
| "eval_steps_per_second": 0.374, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.8029130520504733e-05, | |
| "loss": 0.1451, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_bleu": 41.7092, | |
| "eval_gen_len": 35.2943, | |
| "eval_loss": 0.10221899300813675, | |
| "eval_runtime": 93.7664, | |
| "eval_samples_per_second": 11.123, | |
| "eval_steps_per_second": 0.352, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.7782802641955836e-05, | |
| "loss": 0.1437, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "eval_bleu": 40.1745, | |
| "eval_gen_len": 38.5484, | |
| "eval_loss": 0.10177244991064072, | |
| "eval_runtime": 92.869, | |
| "eval_samples_per_second": 11.231, | |
| "eval_steps_per_second": 0.355, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7536474763406942e-05, | |
| "loss": 0.1429, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "eval_bleu": 42.5078, | |
| "eval_gen_len": 34.3826, | |
| "eval_loss": 0.10191857814788818, | |
| "eval_runtime": 85.4817, | |
| "eval_samples_per_second": 12.201, | |
| "eval_steps_per_second": 0.386, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.7290146884858045e-05, | |
| "loss": 0.1354, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "eval_bleu": 42.0937, | |
| "eval_gen_len": 35.7114, | |
| "eval_loss": 0.10230503231287003, | |
| "eval_runtime": 96.9439, | |
| "eval_samples_per_second": 10.759, | |
| "eval_steps_per_second": 0.34, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.704381900630915e-05, | |
| "loss": 0.1345, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "eval_bleu": 42.3842, | |
| "eval_gen_len": 35.3586, | |
| "eval_loss": 0.10215254127979279, | |
| "eval_runtime": 94.9854, | |
| "eval_samples_per_second": 10.981, | |
| "eval_steps_per_second": 0.347, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.6797491127760255e-05, | |
| "loss": 0.1335, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "eval_bleu": 42.1993, | |
| "eval_gen_len": 34.0729, | |
| "eval_loss": 0.10184619575738907, | |
| "eval_runtime": 104.2027, | |
| "eval_samples_per_second": 10.009, | |
| "eval_steps_per_second": 0.317, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.6551101636435333e-05, | |
| "loss": 0.1338, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "eval_bleu": 42.3348, | |
| "eval_gen_len": 35.5158, | |
| "eval_loss": 0.10167574882507324, | |
| "eval_runtime": 95.3873, | |
| "eval_samples_per_second": 10.934, | |
| "eval_steps_per_second": 0.346, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1.6304773757886436e-05, | |
| "loss": 0.1341, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "eval_bleu": 42.6183, | |
| "eval_gen_len": 34.5225, | |
| "eval_loss": 0.10100951045751572, | |
| "eval_runtime": 105.9409, | |
| "eval_samples_per_second": 9.845, | |
| "eval_steps_per_second": 0.311, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 1.6058445879337542e-05, | |
| "loss": 0.1278, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "eval_bleu": 42.354, | |
| "eval_gen_len": 35.1323, | |
| "eval_loss": 0.10181604325771332, | |
| "eval_runtime": 90.7174, | |
| "eval_samples_per_second": 11.497, | |
| "eval_steps_per_second": 0.364, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.5812118000788645e-05, | |
| "loss": 0.1272, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "eval_bleu": 41.9485, | |
| "eval_gen_len": 35.4727, | |
| "eval_loss": 0.10201819986104965, | |
| "eval_runtime": 99.9879, | |
| "eval_samples_per_second": 10.431, | |
| "eval_steps_per_second": 0.33, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.5565728509463723e-05, | |
| "loss": 0.1261, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "eval_bleu": 42.3042, | |
| "eval_gen_len": 35.2704, | |
| "eval_loss": 0.10195478051900864, | |
| "eval_runtime": 70.0936, | |
| "eval_samples_per_second": 14.88, | |
| "eval_steps_per_second": 0.471, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 1.5319400630914826e-05, | |
| "loss": 0.1265, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "eval_bleu": 39.7039, | |
| "eval_gen_len": 37.767, | |
| "eval_loss": 0.10168451070785522, | |
| "eval_runtime": 112.9245, | |
| "eval_samples_per_second": 9.236, | |
| "eval_steps_per_second": 0.292, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.5073011139589906e-05, | |
| "loss": 0.1273, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "eval_bleu": 42.9732, | |
| "eval_gen_len": 34.5542, | |
| "eval_loss": 0.10154784470796585, | |
| "eval_runtime": 147.6247, | |
| "eval_samples_per_second": 7.065, | |
| "eval_steps_per_second": 0.224, | |
| "step": 80000 | |
| } | |
| ], | |
| "max_steps": 324608, | |
| "num_train_epochs": 16, | |
| "total_flos": 1.3796012702564352e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |