{
  "best_metric": 0.26913971,
  "best_model_checkpoint": "/home/ec2-user/output/qwen2-vl-2b-instruct/v0-20241209-111947/checkpoint-20",
  "epoch": 4.0,
  "eval_steps": 50,
  "global_step": 20,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 47.92485809326172,
      "learning_rate": 5e-05,
      "logits/chosen": -1.8237018585205078,
      "logits/rejected": -1.9020615816116333,
      "logps/chosen": -191.52467346191406,
      "logps/rejected": -212.57785034179688,
      "loss": 5.88650322,
      "memory(GiB)": 30.84,
      "nll_loss": 0.42665722966194153,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1,
      "train_speed(iter/s)": 0.018393
    },
    {
      "epoch": 1.0,
      "grad_norm": 18.606019973754883,
      "learning_rate": 4.4728512734909844e-05,
      "logits/chosen": -1.8926528692245483,
      "logits/rejected": -1.9645055532455444,
      "logps/chosen": -195.32687377929688,
      "logps/rejected": -217.2638702392578,
      "loss": 4.4658556,
      "memory(GiB)": 31.66,
      "nll_loss": 0.43571263551712036,
      "rewards/accuracies": 0.6666666865348816,
      "rewards/chosen": 0.321090966463089,
      "rewards/margins": 0.17073126137256622,
      "rewards/rejected": 0.15035970509052277,
      "step": 5,
      "train_speed(iter/s)": 0.020714
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.947022914886475,
      "learning_rate": 2.7064483636808313e-05,
      "logits/chosen": -1.8044353723526,
      "logits/rejected": -1.8731192350387573,
      "logps/chosen": -175.5177764892578,
      "logps/rejected": -212.43545532226562,
      "loss": 1.8374548,
      "memory(GiB)": 26.6,
      "nll_loss": 0.3899259567260742,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 2.1417019367218018,
      "rewards/margins": 1.649977684020996,
      "rewards/rejected": 0.49172407388687134,
      "step": 10,
      "train_speed(iter/s)": 0.021154
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7851582765579224,
      "learning_rate": 8.067960709356478e-06,
      "logits/chosen": -1.728115200996399,
      "logits/rejected": -1.7883909940719604,
      "logps/chosen": -163.91143798828125,
      "logps/rejected": -218.22042846679688,
      "loss": 0.79830809,
      "memory(GiB)": 26.6,
      "nll_loss": 0.36455366015434265,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 3.3023338317871094,
      "rewards/margins": 3.389106035232544,
      "rewards/rejected": -0.08677242696285248,
      "step": 15,
      "train_speed(iter/s)": 0.021297
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.076888084411621,
      "learning_rate": 0.0,
      "logits/chosen": -1.695635199546814,
      "logits/rejected": -1.7527247667312622,
      "logps/chosen": -159.94375610351562,
      "logps/rejected": -223.05526733398438,
      "loss": 0.61259127,
      "memory(GiB)": 26.6,
      "nll_loss": 0.3550187349319458,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 3.699101448059082,
      "rewards/margins": 4.269360065460205,
      "rewards/rejected": -0.5702579617500305,
      "step": 20,
      "train_speed(iter/s)": 0.021369
    },
    {
      "epoch": 4.0,
      "eval_logits/chosen": -1.8032017946243286,
      "eval_logits/rejected": -1.8270058631896973,
      "eval_logps/chosen": -159.05787658691406,
      "eval_logps/rejected": -185.4047393798828,
      "eval_loss": 0.26913970708847046,
      "eval_nll_loss": 0.4101408123970032,
      "eval_rewards/accuracies": 1.0,
      "eval_rewards/chosen": 3.333481788635254,
      "eval_rewards/margins": 1.5688062906265259,
      "eval_rewards/rejected": 1.764675498008728,
      "eval_runtime": 80.7016,
      "eval_samples_per_second": 0.186,
      "eval_steps_per_second": 0.186,
      "step": 20
    }
  ],
  "logging_steps": 5,
  "max_steps": 20,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4272401148051456.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}