| { |
| "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": 0.016736401673640166, |
| "grad_norm": 12.245462384069063, |
| "learning_rate": 8.333333333333333e-08, |
| "logits/chosen": -2.8317770957946777, |
| "logits/rejected": -2.7326111793518066, |
| "logps/chosen": -190.3253173828125, |
| "logps/pi_response": -110.92029571533203, |
| "logps/ref_response": -110.92029571533203, |
| "logps/rejected": -223.7691192626953, |
| "loss": 0.6931, |
| "rewards/accuracies": 0.0, |
| "rewards/chosen": 0.0, |
| "rewards/margins": 0.0, |
| "rewards/rejected": 0.0, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.16736401673640167, |
| "grad_norm": 11.258294256737887, |
| "learning_rate": 4.930057285201027e-07, |
| "logits/chosen": -2.7557287216186523, |
| "logits/rejected": -2.7092771530151367, |
| "logps/chosen": -216.87799072265625, |
| "logps/pi_response": -114.42035675048828, |
| "logps/ref_response": -114.2578125, |
| "logps/rejected": -264.3336486816406, |
| "loss": 0.6848, |
| "rewards/accuracies": 0.5486111044883728, |
| "rewards/chosen": -0.02949061244726181, |
| "rewards/margins": 0.02440599910914898, |
| "rewards/rejected": -0.05389661341905594, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.33472803347280333, |
| "grad_norm": 26.158633449418044, |
| "learning_rate": 4.187457503795526e-07, |
| "logits/chosen": -2.7771294116973877, |
| "logits/rejected": -2.7362141609191895, |
| "logps/chosen": -254.28701782226562, |
| "logps/pi_response": -111.6153335571289, |
| "logps/ref_response": -113.4717788696289, |
| "logps/rejected": -308.6377868652344, |
| "loss": 0.6304, |
| "rewards/accuracies": 0.6312500238418579, |
| "rewards/chosen": -0.2588905096054077, |
| "rewards/margins": 0.27992284297943115, |
| "rewards/rejected": -0.5388133525848389, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.502092050209205, |
| "grad_norm": 15.587388407908863, |
| "learning_rate": 2.8691164100062034e-07, |
| "logits/chosen": -2.754197359085083, |
| "logits/rejected": -2.718698024749756, |
| "logps/chosen": -297.10418701171875, |
| "logps/pi_response": -142.0721893310547, |
| "logps/ref_response": -120.0687026977539, |
| "logps/rejected": -351.3653259277344, |
| "loss": 0.5839, |
| "rewards/accuracies": 0.690625011920929, |
| "rewards/chosen": -0.523945689201355, |
| "rewards/margins": 0.4522995352745056, |
| "rewards/rejected": -0.9762452840805054, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.6694560669456067, |
| "grad_norm": 16.430781172533628, |
| "learning_rate": 1.4248369943086995e-07, |
| "logits/chosen": -2.6763617992401123, |
| "logits/rejected": -2.6239538192749023, |
| "logps/chosen": -278.3578186035156, |
| "logps/pi_response": -144.41726684570312, |
| "logps/ref_response": -110.48355865478516, |
| "logps/rejected": -363.93450927734375, |
| "loss": 0.5503, |
| "rewards/accuracies": 0.699999988079071, |
| "rewards/chosen": -0.5653710961341858, |
| "rewards/margins": 0.5595625042915344, |
| "rewards/rejected": -1.1249334812164307, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.8368200836820083, |
| "grad_norm": 18.753898859495603, |
| "learning_rate": 3.473909705816111e-08, |
| "logits/chosen": -2.624819040298462, |
| "logits/rejected": -2.5662152767181396, |
| "logps/chosen": -287.62396240234375, |
| "logps/pi_response": -165.9267578125, |
| "logps/ref_response": -112.31349182128906, |
| "logps/rejected": -373.04034423828125, |
| "loss": 0.5435, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": -0.7290452122688293, |
| "rewards/margins": 0.625941276550293, |
| "rewards/rejected": -1.3549864292144775, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.9874476987447699, |
| "step": 59, |
| "total_flos": 0.0, |
| "train_loss": 0.587253408916926, |
| "train_runtime": 2659.1015, |
| "train_samples_per_second": 5.747, |
| "train_steps_per_second": 0.022 |
| } |
| ], |
| "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 |
| } |
|
|
