| { | |
| "best_metric": 1.1881097555160522, | |
| "best_model_checkpoint": "../facebook/nllb-200-3.3B-finetuned/checkpoint-1000", | |
| "epoch": 5000.0, | |
| "eval_steps": 500, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 500.0, | |
| "grad_norm": 13.488256454467773, | |
| "learning_rate": 9.758186548415274e-06, | |
| "loss": 4.34, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 500.0, | |
| "eval_gen_len": 26.0, | |
| "eval_loss": 2.901174306869507, | |
| "eval_rouge": 0.05, | |
| "eval_runtime": 7.0444, | |
| "eval_samples_per_second": 0.284, | |
| "eval_steps_per_second": 0.142, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1000.0, | |
| "grad_norm": 0.9894864559173584, | |
| "learning_rate": 9.050617527877911e-06, | |
| "loss": 0.609, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1000.0, | |
| "eval_gen_len": 26.0, | |
| "eval_loss": 1.1881097555160522, | |
| "eval_rouge": 0.098, | |
| "eval_runtime": 1.1556, | |
| "eval_samples_per_second": 1.731, | |
| "eval_steps_per_second": 0.865, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1500.0, | |
| "grad_norm": 0.1642373949289322, | |
| "learning_rate": 7.946545841400035e-06, | |
| "loss": 0.0196, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1500.0, | |
| "eval_gen_len": 30.5, | |
| "eval_loss": 1.5324629545211792, | |
| "eval_rouge": 0.1493, | |
| "eval_runtime": 6.9442, | |
| "eval_samples_per_second": 0.288, | |
| "eval_steps_per_second": 0.144, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2000.0, | |
| "grad_norm": 0.08364015817642212, | |
| "learning_rate": 6.554045718173867e-06, | |
| "loss": 0.0061, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2000.0, | |
| "eval_gen_len": 33.5, | |
| "eval_loss": 1.5447587966918945, | |
| "eval_rouge": 0.125, | |
| "eval_runtime": 6.9455, | |
| "eval_samples_per_second": 0.288, | |
| "eval_steps_per_second": 0.144, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2500.0, | |
| "grad_norm": 0.05517512932419777, | |
| "learning_rate": 5.0094247723796405e-06, | |
| "loss": 0.0036, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2500.0, | |
| "eval_gen_len": 33.5, | |
| "eval_loss": 1.5625241994857788, | |
| "eval_rouge": 0.125, | |
| "eval_runtime": 6.9319, | |
| "eval_samples_per_second": 0.289, | |
| "eval_steps_per_second": 0.144, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3000.0, | |
| "grad_norm": 0.04180261120200157, | |
| "learning_rate": 3.463881264198645e-06, | |
| "loss": 0.0025, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3000.0, | |
| "eval_gen_len": 33.5, | |
| "eval_loss": 1.5641270875930786, | |
| "eval_rouge": 0.125, | |
| "eval_runtime": 6.9474, | |
| "eval_samples_per_second": 0.288, | |
| "eval_steps_per_second": 0.144, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3500.0, | |
| "grad_norm": 0.03480682149529457, | |
| "learning_rate": 2.0687037606464554e-06, | |
| "loss": 0.002, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3500.0, | |
| "eval_gen_len": 33.5, | |
| "eval_loss": 1.5625765323638916, | |
| "eval_rouge": 0.125, | |
| "eval_runtime": 6.9229, | |
| "eval_samples_per_second": 0.289, | |
| "eval_steps_per_second": 0.144, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 4000.0, | |
| "grad_norm": 0.031249279156327248, | |
| "learning_rate": 9.60461956544021e-07, | |
| "loss": 0.0017, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4000.0, | |
| "eval_gen_len": 24.5, | |
| "eval_loss": 1.4340063333511353, | |
| "eval_rouge": 0.0357, | |
| "eval_runtime": 6.9464, | |
| "eval_samples_per_second": 0.288, | |
| "eval_steps_per_second": 0.144, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4500.0, | |
| "grad_norm": 0.029812365770339966, | |
| "learning_rate": 2.4763828125157654e-07, | |
| "loss": 0.0016, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 4500.0, | |
| "eval_gen_len": 24.5, | |
| "eval_loss": 1.4486483335494995, | |
| "eval_rouge": 0.0357, | |
| "eval_runtime": 6.9642, | |
| "eval_samples_per_second": 0.287, | |
| "eval_steps_per_second": 0.144, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 5000.0, | |
| "grad_norm": 0.029535507783293724, | |
| "learning_rate": 8.882641330809627e-12, | |
| "loss": 0.0016, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 5000.0, | |
| "eval_gen_len": 24.5, | |
| "eval_loss": 1.4524964094161987, | |
| "eval_rouge": 0.0357, | |
| "eval_runtime": 6.9555, | |
| "eval_samples_per_second": 0.288, | |
| "eval_steps_per_second": 0.144, | |
| "step": 5000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5000, | |
| "save_steps": 500, | |
| "total_flos": 3.181132972032e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |