| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.9874476987447699, |
| "eval_steps": 500, |
| "global_step": 59, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.02, |
| "learning_rate": 8.333333333333333e-08, |
| "logits/chosen": -2.940945625305176, |
| "logits/rejected": -2.8081953525543213, |
| "logps/chosen": -373.83038330078125, |
| "logps/pi_response": -367.2508544921875, |
| "logps/ref_response": -367.2508544921875, |
| "logps/rejected": -354.8029479980469, |
| "loss": 0.6931, |
| "rewards/accuracies": 0.0, |
| "rewards/chosen": 0.0, |
| "rewards/margins": 0.0, |
| "rewards/rejected": 0.0, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.930057285201027e-07, |
| "logits/chosen": -2.8032095432281494, |
| "logits/rejected": -2.7498764991760254, |
| "logps/chosen": -334.4217529296875, |
| "logps/pi_response": -352.21014404296875, |
| "logps/ref_response": -351.01263427734375, |
| "logps/rejected": -357.6781311035156, |
| "loss": 0.6883, |
| "rewards/accuracies": 0.5763888955116272, |
| "rewards/chosen": -0.020477451384067535, |
| "rewards/margins": 0.022380363196134567, |
| "rewards/rejected": -0.0428578145802021, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 4.187457503795526e-07, |
| "logits/chosen": -2.7823891639709473, |
| "logits/rejected": -2.7310681343078613, |
| "logps/chosen": -357.344482421875, |
| "logps/pi_response": -359.2858581542969, |
| "logps/ref_response": -349.52899169921875, |
| "logps/rejected": -366.3606872558594, |
| "loss": 0.6344, |
| "rewards/accuracies": 0.690625011920929, |
| "rewards/chosen": -0.1733517050743103, |
| "rewards/margins": 0.18930229544639587, |
| "rewards/rejected": -0.3626540005207062, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 2.8691164100062034e-07, |
| "logits/chosen": -2.6963818073272705, |
| "logits/rejected": -2.6740260124206543, |
| "logps/chosen": -387.28399658203125, |
| "logps/pi_response": -385.0074157714844, |
| "logps/ref_response": -349.4725341796875, |
| "logps/rejected": -426.54205322265625, |
| "loss": 0.588, |
| "rewards/accuracies": 0.671875, |
| "rewards/chosen": -0.5384770035743713, |
| "rewards/margins": 0.38316935300827026, |
| "rewards/rejected": -0.9216462969779968, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 1.4248369943086995e-07, |
| "logits/chosen": -2.576090097427368, |
| "logits/rejected": -2.5352632999420166, |
| "logps/chosen": -388.36114501953125, |
| "logps/pi_response": -392.815185546875, |
| "logps/ref_response": -334.27154541015625, |
| "logps/rejected": -480.95147705078125, |
| "loss": 0.5299, |
| "rewards/accuracies": 0.6968749761581421, |
| "rewards/chosen": -0.759750485420227, |
| "rewards/margins": 0.683925211429596, |
| "rewards/rejected": -1.4436757564544678, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.473909705816111e-08, |
| "logits/chosen": -2.5585336685180664, |
| "logits/rejected": -2.5140857696533203, |
| "logps/chosen": -400.4565734863281, |
| "logps/pi_response": -407.38433837890625, |
| "logps/ref_response": -343.3310546875, |
| "logps/rejected": -472.02996826171875, |
| "loss": 0.5298, |
| "rewards/accuracies": 0.731249988079071, |
| "rewards/chosen": -0.7769991159439087, |
| "rewards/margins": 0.6114187240600586, |
| "rewards/rejected": -1.3884178400039673, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.99, |
| "step": 59, |
| "total_flos": 0.0, |
| "train_loss": 0.5893503851809744, |
| "train_runtime": 3552.0479, |
| "train_samples_per_second": 4.303, |
| "train_steps_per_second": 0.017 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 59, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 100, |
| "total_flos": 0.0, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
