| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.002535004852157725, |
| "eval_steps": 500, |
| "global_step": 8, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0003168756065197156, |
| "grad_norm": 6.832624876551563e-06, |
| "learning_rate": 2.9996830427892237e-06, |
| "loss": 0.0, |
| "loss/policy_avg": 2.0139850676059723e-08, |
| "objective/entropy": 56.029441833496094, |
| "objective/kl": 0.0, |
| "objective/rlhf_reward": -0.5591201782226562, |
| "objective/scores": -0.559173583984375, |
| "policy/approxkl_avg": 0.0, |
| "policy/clipfrac_avg": 0.0, |
| "policy/entropy_avg": 0.46982458233833313, |
| "step": 1, |
| "timer/training_step": 4.71628999710083, |
| "val/num_eos_tokens": 0.75, |
| "val/ratio": 1.0 |
| }, |
| { |
| "epoch": 0.0006337512130394313, |
| "grad_norm": 7.318521966226399e-06, |
| "learning_rate": 2.999366085578447e-06, |
| "loss": 0.0, |
| "loss/policy_avg": 1.0943040251731873e-08, |
| "objective/entropy": 49.566184997558594, |
| "objective/kl": 0.0, |
| "objective/rlhf_reward": -0.7360076904296875, |
| "objective/scores": -0.73590087890625, |
| "policy/approxkl_avg": 0.0, |
| "policy/clipfrac_avg": 0.0, |
| "policy/entropy_avg": 0.47101089358329773, |
| "step": 2, |
| "timer/training_step": 5.188745975494385, |
| "val/num_eos_tokens": 1.21875, |
| "val/ratio": 1.0 |
| }, |
| { |
| "epoch": 0.0009506268195591468, |
| "grad_norm": 6.249741545616416e-06, |
| "learning_rate": 2.9990491283676707e-06, |
| "loss": 0.0, |
| "loss/policy_avg": 8.149072527885437e-09, |
| "objective/entropy": 57.22004699707031, |
| "objective/kl": 0.0, |
| "objective/rlhf_reward": -0.70709228515625, |
| "objective/scores": -0.70703125, |
| "policy/approxkl_avg": 0.0, |
| "policy/clipfrac_avg": 0.0, |
| "policy/entropy_avg": 0.4886399507522583, |
| "step": 3, |
| "timer/training_step": 4.847959518432617, |
| "val/num_eos_tokens": 0.53125, |
| "val/ratio": 1.0 |
| }, |
| { |
| "epoch": 0.0012675024260788625, |
| "grad_norm": 7.052934051898774e-06, |
| "learning_rate": 2.998732171156894e-06, |
| "loss": 0.0, |
| "loss/policy_avg": 1.0710209608078003e-08, |
| "objective/entropy": 59.56321716308594, |
| "objective/kl": 0.0, |
| "objective/rlhf_reward": -0.716552734375, |
| "objective/scores": -0.71685791015625, |
| "policy/approxkl_avg": 0.0, |
| "policy/clipfrac_avg": 0.0, |
| "policy/entropy_avg": 0.4839285612106323, |
| "step": 4, |
| "timer/training_step": 5.577988624572754, |
| "val/num_eos_tokens": 0.4375, |
| "val/ratio": 1.0 |
| }, |
| { |
| "epoch": 0.001584378032598578, |
| "grad_norm": 8.835036169330124e-06, |
| "learning_rate": 2.9984152139461176e-06, |
| "loss": 0.0, |
| "loss/policy_avg": 2.3748725652694702e-08, |
| "objective/entropy": 45.10405731201172, |
| "objective/kl": 0.0, |
| "objective/rlhf_reward": -0.444000244140625, |
| "objective/scores": -0.44439697265625, |
| "policy/approxkl_avg": 0.0, |
| "policy/clipfrac_avg": 0.0, |
| "policy/entropy_avg": 0.4351224899291992, |
| "step": 5, |
| "timer/training_step": 5.305084705352783, |
| "val/num_eos_tokens": 1.0, |
| "val/ratio": 1.0 |
| }, |
| { |
| "epoch": 0.0019012536391182935, |
| "grad_norm": 6.025600214343285e-06, |
| "learning_rate": 2.998098256735341e-06, |
| "loss": 0.0, |
| "loss/policy_avg": 7.450580596923828e-09, |
| "objective/entropy": 60.55895233154297, |
| "objective/kl": 0.0, |
| "objective/rlhf_reward": -0.746917724609375, |
| "objective/scores": -0.7470703125, |
| "policy/approxkl_avg": 0.0, |
| "policy/clipfrac_avg": 0.0, |
| "policy/entropy_avg": 0.4965381920337677, |
| "step": 6, |
| "timer/training_step": 4.756265640258789, |
| "val/num_eos_tokens": 0.53125, |
| "val/ratio": 1.0 |
| }, |
| { |
| "epoch": 0.002218129245638009, |
| "grad_norm": 9.883336133498233e-06, |
| "learning_rate": 2.997781299524564e-06, |
| "loss": 0.0, |
| "loss/policy_avg": 1.6298145055770874e-08, |
| "objective/entropy": 56.12989807128906, |
| "objective/kl": 0.0, |
| "objective/rlhf_reward": -0.58819580078125, |
| "objective/scores": -0.58782958984375, |
| "policy/approxkl_avg": 0.0, |
| "policy/clipfrac_avg": 0.0, |
| "policy/entropy_avg": 0.4676559567451477, |
| "step": 7, |
| "timer/training_step": 5.001364707946777, |
| "val/num_eos_tokens": 0.65625, |
| "val/ratio": 1.0 |
| }, |
| { |
| "epoch": 0.002535004852157725, |
| "grad_norm": 3.0753049031773116e-06, |
| "learning_rate": 2.9974643423137877e-06, |
| "loss": 0.0, |
| "loss/policy_avg": 3.259629011154175e-09, |
| "objective/entropy": 52.772064208984375, |
| "objective/kl": 0.0, |
| "objective/rlhf_reward": -0.8326416015625, |
| "objective/scores": -0.83258056640625, |
| "policy/approxkl_avg": 0.0, |
| "policy/clipfrac_avg": 0.0, |
| "policy/entropy_avg": 0.4242881238460541, |
| "step": 8, |
| "timer/training_step": 5.542971134185791, |
| "val/num_eos_tokens": 0.40625, |
| "val/ratio": 1.0 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 9465, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 8, |
| "total_flos": 5677993863069696.0, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
