| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 189, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.015873015873015872, | |
| "grad_norm": 15.294358962776673, | |
| "learning_rate": 2.6315789473684208e-08, | |
| "logits/chosen": -1.6171875, | |
| "logits/rejected": -1.4296875, | |
| "logps/chosen": -139.0, | |
| "logps/rejected": -104.5, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.15873015873015872, | |
| "grad_norm": 15.094161248794443, | |
| "learning_rate": 2.631578947368421e-07, | |
| "logits/chosen": -1.5859375, | |
| "logits/rejected": -1.53125, | |
| "logps/chosen": -149.0, | |
| "logps/rejected": -129.0, | |
| "loss": 0.6923, | |
| "rewards/accuracies": 0.2222222238779068, | |
| "rewards/chosen": -0.0013885498046875, | |
| "rewards/margins": 0.0045166015625, | |
| "rewards/rejected": -0.005889892578125, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.31746031746031744, | |
| "grad_norm": 15.606155891832502, | |
| "learning_rate": 4.970588235294118e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.546875, | |
| "logps/chosen": -152.0, | |
| "logps/rejected": -133.0, | |
| "loss": 0.6871, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.031005859375, | |
| "rewards/margins": 0.00125885009765625, | |
| "rewards/rejected": -0.0322265625, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.47619047619047616, | |
| "grad_norm": 14.846625318284318, | |
| "learning_rate": 4.676470588235294e-07, | |
| "logits/chosen": -1.6015625, | |
| "logits/rejected": -1.609375, | |
| "logps/chosen": -159.0, | |
| "logps/rejected": -149.0, | |
| "loss": 0.6824, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.06298828125, | |
| "rewards/margins": 0.0184326171875, | |
| "rewards/rejected": -0.08154296875, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.6349206349206349, | |
| "grad_norm": 12.673127513530932, | |
| "learning_rate": 4.38235294117647e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.578125, | |
| "logps/chosen": -170.0, | |
| "logps/rejected": -155.0, | |
| "loss": 0.6793, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.0751953125, | |
| "rewards/margins": 0.042236328125, | |
| "rewards/rejected": -0.11767578125, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.7936507936507936, | |
| "grad_norm": 14.318265772228457, | |
| "learning_rate": 4.0882352941176465e-07, | |
| "logits/chosen": -1.59375, | |
| "logits/rejected": -1.546875, | |
| "logps/chosen": -146.0, | |
| "logps/rejected": -126.5, | |
| "loss": 0.6741, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.1357421875, | |
| "rewards/margins": 0.0634765625, | |
| "rewards/rejected": -0.19921875, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 14.53000414550404, | |
| "learning_rate": 3.7941176470588235e-07, | |
| "logits/chosen": -1.5234375, | |
| "logits/rejected": -1.5, | |
| "logps/chosen": -138.0, | |
| "logps/rejected": -138.0, | |
| "loss": 0.6696, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -0.1845703125, | |
| "rewards/margins": 0.024169921875, | |
| "rewards/rejected": -0.208984375, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -1.5, | |
| "eval_logits/rejected": -1.4921875, | |
| "eval_logps/chosen": -140.0, | |
| "eval_logps/rejected": -127.0, | |
| "eval_loss": 0.6568750143051147, | |
| "eval_rewards/accuracies": 0.5714285969734192, | |
| "eval_rewards/chosen": -0.1796875, | |
| "eval_rewards/margins": 0.0966796875, | |
| "eval_rewards/rejected": -0.27734375, | |
| "eval_runtime": 14.7253, | |
| "eval_samples_per_second": 13.582, | |
| "eval_steps_per_second": 0.475, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 13.811509421726951, | |
| "learning_rate": 3.5e-07, | |
| "logits/chosen": -1.53125, | |
| "logits/rejected": -1.546875, | |
| "logps/chosen": -142.0, | |
| "logps/rejected": -147.0, | |
| "loss": 0.5957, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.1337890625, | |
| "rewards/margins": 0.2177734375, | |
| "rewards/rejected": -0.3515625, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.2698412698412698, | |
| "grad_norm": 11.993229269242237, | |
| "learning_rate": 3.205882352941177e-07, | |
| "logits/chosen": -1.5546875, | |
| "logits/rejected": -1.5390625, | |
| "logps/chosen": -150.0, | |
| "logps/rejected": -135.0, | |
| "loss": 0.5614, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.1328125, | |
| "rewards/margins": 0.28515625, | |
| "rewards/rejected": -0.41796875, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.4285714285714286, | |
| "grad_norm": 11.581245327928885, | |
| "learning_rate": 2.911764705882353e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.5078125, | |
| "logps/chosen": -144.0, | |
| "logps/rejected": -133.0, | |
| "loss": 0.5406, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.2333984375, | |
| "rewards/margins": 0.341796875, | |
| "rewards/rejected": -0.57421875, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.5873015873015874, | |
| "grad_norm": 12.825913209383442, | |
| "learning_rate": 2.6176470588235295e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.53125, | |
| "logps/chosen": -152.0, | |
| "logps/rejected": -138.0, | |
| "loss": 0.5382, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.26953125, | |
| "rewards/margins": 0.390625, | |
| "rewards/rejected": -0.66015625, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.746031746031746, | |
| "grad_norm": 12.686457389394935, | |
| "learning_rate": 2.323529411764706e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.5625, | |
| "logps/chosen": -160.0, | |
| "logps/rejected": -154.0, | |
| "loss": 0.5295, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.23046875, | |
| "rewards/margins": 0.46484375, | |
| "rewards/rejected": -0.6953125, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.9047619047619047, | |
| "grad_norm": 11.815278327307368, | |
| "learning_rate": 2.0294117647058823e-07, | |
| "logits/chosen": -1.5546875, | |
| "logits/rejected": -1.5, | |
| "logps/chosen": -138.0, | |
| "logps/rejected": -137.0, | |
| "loss": 0.5284, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.33984375, | |
| "rewards/margins": 0.365234375, | |
| "rewards/rejected": -0.70703125, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -1.484375, | |
| "eval_logits/rejected": -1.4765625, | |
| "eval_logps/chosen": -143.0, | |
| "eval_logps/rejected": -130.0, | |
| "eval_loss": 0.6310937404632568, | |
| "eval_rewards/accuracies": 0.625, | |
| "eval_rewards/chosen": -0.388671875, | |
| "eval_rewards/margins": 0.185546875, | |
| "eval_rewards/rejected": -0.57421875, | |
| "eval_runtime": 17.8485, | |
| "eval_samples_per_second": 11.205, | |
| "eval_steps_per_second": 0.392, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.0634920634920633, | |
| "grad_norm": 10.190155022883303, | |
| "learning_rate": 1.7352941176470587e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.515625, | |
| "logps/chosen": -154.0, | |
| "logps/rejected": -146.0, | |
| "loss": 0.4902, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.298828125, | |
| "rewards/margins": 0.51953125, | |
| "rewards/rejected": -0.8203125, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 11.482317016645608, | |
| "learning_rate": 1.441176470588235e-07, | |
| "logits/chosen": -1.5703125, | |
| "logits/rejected": -1.5390625, | |
| "logps/chosen": -146.0, | |
| "logps/rejected": -147.0, | |
| "loss": 0.4618, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.298828125, | |
| "rewards/margins": 0.57421875, | |
| "rewards/rejected": -0.875, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.380952380952381, | |
| "grad_norm": 10.664901408549657, | |
| "learning_rate": 1.1470588235294116e-07, | |
| "logits/chosen": -1.484375, | |
| "logits/rejected": -1.484375, | |
| "logps/chosen": -139.0, | |
| "logps/rejected": -135.0, | |
| "loss": 0.4529, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.392578125, | |
| "rewards/margins": 0.56640625, | |
| "rewards/rejected": -0.95703125, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.5396825396825395, | |
| "grad_norm": 11.135694541597557, | |
| "learning_rate": 8.529411764705883e-08, | |
| "logits/chosen": -1.546875, | |
| "logits/rejected": -1.46875, | |
| "logps/chosen": -138.0, | |
| "logps/rejected": -125.5, | |
| "loss": 0.4585, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.34375, | |
| "rewards/margins": 0.62109375, | |
| "rewards/rejected": -0.96484375, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.6984126984126986, | |
| "grad_norm": 9.969592838659052, | |
| "learning_rate": 5.5882352941176474e-08, | |
| "logits/chosen": -1.515625, | |
| "logits/rejected": -1.515625, | |
| "logps/chosen": -144.0, | |
| "logps/rejected": -147.0, | |
| "loss": 0.4698, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.359375, | |
| "rewards/margins": 0.60546875, | |
| "rewards/rejected": -0.96484375, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 10.412087107819227, | |
| "learning_rate": 2.6470588235294116e-08, | |
| "logits/chosen": -1.5234375, | |
| "logits/rejected": -1.4921875, | |
| "logps/chosen": -147.0, | |
| "logps/rejected": -137.0, | |
| "loss": 0.4529, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.267578125, | |
| "rewards/margins": 0.63671875, | |
| "rewards/rejected": -0.90234375, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_logits/chosen": -1.484375, | |
| "eval_logits/rejected": -1.46875, | |
| "eval_logps/chosen": -143.0, | |
| "eval_logps/rejected": -131.0, | |
| "eval_loss": 0.6217187643051147, | |
| "eval_rewards/accuracies": 0.6785714030265808, | |
| "eval_rewards/chosen": -0.462890625, | |
| "eval_rewards/margins": 0.220703125, | |
| "eval_rewards/rejected": -0.68359375, | |
| "eval_runtime": 17.9909, | |
| "eval_samples_per_second": 11.117, | |
| "eval_steps_per_second": 0.389, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 189, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5609137318121693, | |
| "train_runtime": 1965.8465, | |
| "train_samples_per_second": 3.051, | |
| "train_steps_per_second": 0.096 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 189, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |