{
  "best_metric": 0.040502406656742096,
  "best_model_checkpoint": "./mistral/21-04-24-Weni-WeniGPT-Agents-Mistral-1.0.0-SFT-1.0.16-DPO_Experiment on DPO with other hyperparameters and best SFT model of WeniGPT-2_max_steps-180_batch_8_2024-04-21_ppid_9/checkpoint-180",
  "epoch": 5.806451612903226,
  "eval_steps": 30,
  "global_step": 180,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 23.757038116455078,
      "learning_rate": 4.971264367816092e-06,
      "logits/chosen": -1.8281418085098267,
      "logits/rejected": -1.8474775552749634,
      "logps/chosen": -128.32241821289062,
      "logps/rejected": -168.0380401611328,
      "loss": 0.6742,
      "rewards/accuracies": 0.5,
      "rewards/chosen": 0.024183962494134903,
      "rewards/margins": 0.03792798891663551,
      "rewards/rejected": -0.013744029216468334,
      "step": 10
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 14.753284454345703,
      "learning_rate": 4.683908045977012e-06,
      "logits/chosen": -1.7773618698120117,
      "logits/rejected": -1.780083417892456,
      "logps/chosen": -144.4004364013672,
      "logps/rejected": -171.8882598876953,
      "loss": 0.5367,
      "rewards/accuracies": 0.8500000238418579,
      "rewards/chosen": 0.282940536737442,
      "rewards/margins": 0.31420594453811646,
      "rewards/rejected": -0.03126540407538414,
      "step": 20
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 13.17676067352295,
      "learning_rate": 4.396551724137931e-06,
      "logits/chosen": -1.8475195169448853,
      "logits/rejected": -1.8598411083221436,
      "logps/chosen": -166.5201416015625,
      "logps/rejected": -198.87669372558594,
      "loss": 0.3994,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.7265206575393677,
      "rewards/margins": 1.12095308303833,
      "rewards/rejected": -0.3944324254989624,
      "step": 30
    },
    {
      "epoch": 0.967741935483871,
      "eval_logits/chosen": -1.7065293788909912,
      "eval_logits/rejected": -1.7336915731430054,
      "eval_logps/chosen": -112.41703033447266,
      "eval_logps/rejected": -210.2833709716797,
      "eval_loss": 0.339741975069046,
      "eval_rewards/accuracies": 0.7142857313156128,
      "eval_rewards/chosen": 0.7506826519966125,
      "eval_rewards/margins": 0.770815372467041,
      "eval_rewards/rejected": -0.020132770761847496,
      "eval_runtime": 10.1682,
      "eval_samples_per_second": 2.754,
      "eval_steps_per_second": 0.688,
      "step": 30
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 9.547523498535156,
      "learning_rate": 4.1091954022988515e-06,
      "logits/chosen": -1.7585010528564453,
      "logits/rejected": -1.7726383209228516,
      "logps/chosen": -123.3655014038086,
      "logps/rejected": -181.35037231445312,
      "loss": 0.2655,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.981993556022644,
      "rewards/margins": 1.3875294923782349,
      "rewards/rejected": -0.4055357575416565,
      "step": 40
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 6.552857398986816,
      "learning_rate": 3.82183908045977e-06,
      "logits/chosen": -1.7593097686767578,
      "logits/rejected": -1.7726491689682007,
      "logps/chosen": -79.5435562133789,
      "logps/rejected": -169.55801391601562,
      "loss": 0.2323,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 1.1555945873260498,
      "rewards/margins": 1.6152770519256592,
      "rewards/rejected": -0.45968255400657654,
      "step": 50
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 5.082254409790039,
      "learning_rate": 3.5344827586206898e-06,
      "logits/chosen": -1.7814455032348633,
      "logits/rejected": -1.8014984130859375,
      "logps/chosen": -135.05734252929688,
      "logps/rejected": -220.71621704101562,
      "loss": 0.1771,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.483880877494812,
      "rewards/margins": 2.265058994293213,
      "rewards/rejected": -0.7811781167984009,
      "step": 60
    },
    {
      "epoch": 1.935483870967742,
      "eval_logits/chosen": -1.7119187116622925,
      "eval_logits/rejected": -1.7429500818252563,
      "eval_logps/chosen": -106.78651428222656,
      "eval_logps/rejected": -215.82162475585938,
      "eval_loss": 0.17449094355106354,
      "eval_rewards/accuracies": 0.8571428656578064,
      "eval_rewards/chosen": 1.3137341737747192,
      "eval_rewards/margins": 1.8876904249191284,
      "eval_rewards/rejected": -0.5739563703536987,
      "eval_runtime": 10.1675,
      "eval_samples_per_second": 2.754,
      "eval_steps_per_second": 0.688,
      "step": 60
    },
    {
      "epoch": 2.258064516129032,
      "grad_norm": 3.0963690280914307,
      "learning_rate": 3.2758620689655175e-06,
      "logits/chosen": -1.8137136697769165,
      "logits/rejected": -1.831018090248108,
      "logps/chosen": -127.5184555053711,
      "logps/rejected": -193.3052215576172,
      "loss": 0.144,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 1.2200127840042114,
      "rewards/margins": 2.3993964195251465,
      "rewards/rejected": -1.179383635520935,
      "step": 70
    },
    {
      "epoch": 2.5806451612903225,
      "grad_norm": 4.23032808303833,
      "learning_rate": 2.988505747126437e-06,
      "logits/chosen": -1.8039875030517578,
      "logits/rejected": -1.8293946981430054,
      "logps/chosen": -109.42149353027344,
      "logps/rejected": -206.9613494873047,
      "loss": 0.1129,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.4697037935256958,
      "rewards/margins": 2.851051092147827,
      "rewards/rejected": -1.3813471794128418,
      "step": 80
    },
    {
      "epoch": 2.903225806451613,
      "grad_norm": 6.72817850112915,
      "learning_rate": 2.7011494252873567e-06,
      "logits/chosen": -1.814522385597229,
      "logits/rejected": -1.8382046222686768,
      "logps/chosen": -186.84329223632812,
      "logps/rejected": -260.76788330078125,
      "loss": 0.1314,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.1758986711502075,
      "rewards/margins": 2.6401941776275635,
      "rewards/rejected": -1.4642951488494873,
      "step": 90
    },
    {
      "epoch": 2.903225806451613,
      "eval_logits/chosen": -1.7031118869781494,
      "eval_logits/rejected": -1.7357984781265259,
      "eval_logps/chosen": -105.27674102783203,
      "eval_logps/rejected": -222.73390197753906,
      "eval_loss": 0.11280059069395065,
      "eval_rewards/accuracies": 0.8571428656578064,
      "eval_rewards/chosen": 1.4647114276885986,
      "eval_rewards/margins": 2.7298948764801025,
      "eval_rewards/rejected": -1.2651835680007935,
      "eval_runtime": 10.1661,
      "eval_samples_per_second": 2.754,
      "eval_steps_per_second": 0.689,
      "step": 90
    },
    {
      "epoch": 3.225806451612903,
      "grad_norm": 3.741824150085449,
      "learning_rate": 2.4137931034482762e-06,
      "logits/chosen": -1.9151725769042969,
      "logits/rejected": -1.929137945175171,
      "logps/chosen": -195.36695861816406,
      "logps/rejected": -271.27081298828125,
      "loss": 0.1269,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.3293390274047852,
      "rewards/margins": 3.3184096813201904,
      "rewards/rejected": -1.9890711307525635,
      "step": 100
    },
    {
      "epoch": 3.5483870967741935,
      "grad_norm": 3.6807944774627686,
      "learning_rate": 2.1264367816091954e-06,
      "logits/chosen": -1.7386070489883423,
      "logits/rejected": -1.766031265258789,
      "logps/chosen": -120.89439392089844,
      "logps/rejected": -234.75289916992188,
      "loss": 0.08,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.6623188257217407,
      "rewards/margins": 4.070118427276611,
      "rewards/rejected": -2.4078001976013184,
      "step": 110
    },
    {
      "epoch": 3.870967741935484,
      "grad_norm": 4.077445030212402,
      "learning_rate": 1.839080459770115e-06,
      "logits/chosen": -1.8381446599960327,
      "logits/rejected": -1.848647117614746,
      "logps/chosen": -162.987548828125,
      "logps/rejected": -245.94497680664062,
      "loss": 0.1016,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.698729157447815,
      "rewards/margins": 3.5379676818847656,
      "rewards/rejected": -1.8392385244369507,
      "step": 120
    },
    {
      "epoch": 3.870967741935484,
      "eval_logits/chosen": -1.6965323686599731,
      "eval_logits/rejected": -1.729313850402832,
      "eval_logps/chosen": -104.65281677246094,
      "eval_logps/rejected": -229.02908325195312,
      "eval_loss": 0.06442112475633621,
      "eval_rewards/accuracies": 1.0,
      "eval_rewards/chosen": 1.527103066444397,
      "eval_rewards/margins": 3.4218084812164307,
      "eval_rewards/rejected": -1.8947057723999023,
      "eval_runtime": 10.1682,
      "eval_samples_per_second": 2.754,
      "eval_steps_per_second": 0.688,
      "step": 120
    },
    {
      "epoch": 4.193548387096774,
      "grad_norm": 2.3105103969573975,
      "learning_rate": 1.5517241379310346e-06,
      "logits/chosen": -1.8013200759887695,
      "logits/rejected": -1.817307472229004,
      "logps/chosen": -132.51260375976562,
      "logps/rejected": -271.40521240234375,
      "loss": 0.0683,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.6829872131347656,
      "rewards/margins": 4.633041858673096,
      "rewards/rejected": -2.950054407119751,
      "step": 130
    },
    {
      "epoch": 4.516129032258064,
      "grad_norm": 2.9849331378936768,
      "learning_rate": 1.2643678160919542e-06,
      "logits/chosen": -1.8443206548690796,
      "logits/rejected": -1.8669426441192627,
      "logps/chosen": -173.345947265625,
      "logps/rejected": -228.7808380126953,
      "loss": 0.0546,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.7071704864501953,
      "rewards/margins": 3.6590206623077393,
      "rewards/rejected": -1.9518499374389648,
      "step": 140
    },
    {
      "epoch": 4.838709677419355,
      "grad_norm": 1.6962618827819824,
      "learning_rate": 9.770114942528738e-07,
      "logits/chosen": -1.8463163375854492,
      "logits/rejected": -1.8704414367675781,
      "logps/chosen": -127.43244934082031,
      "logps/rejected": -244.4454803466797,
      "loss": 0.0633,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.2895654439926147,
      "rewards/margins": 4.082577705383301,
      "rewards/rejected": -2.7930119037628174,
      "step": 150
    },
    {
      "epoch": 4.838709677419355,
      "eval_logits/chosen": -1.692850947380066,
      "eval_logits/rejected": -1.7251230478286743,
      "eval_logps/chosen": -105.38704681396484,
      "eval_logps/rejected": -233.88169860839844,
      "eval_loss": 0.04773273691534996,
      "eval_rewards/accuracies": 1.0,
      "eval_rewards/chosen": 1.4536798000335693,
      "eval_rewards/margins": 3.8336431980133057,
      "eval_rewards/rejected": -2.3799636363983154,
      "eval_runtime": 10.1649,
      "eval_samples_per_second": 2.755,
      "eval_steps_per_second": 0.689,
      "step": 150
    },
    {
      "epoch": 5.161290322580645,
      "grad_norm": 1.0977122783660889,
      "learning_rate": 6.896551724137931e-07,
      "logits/chosen": -1.854543924331665,
      "logits/rejected": -1.8709430694580078,
      "logps/chosen": -124.8465347290039,
      "logps/rejected": -214.95779418945312,
      "loss": 0.0374,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.6399686336517334,
      "rewards/margins": 4.3664116859436035,
      "rewards/rejected": -2.72644305229187,
      "step": 160
    },
    {
      "epoch": 5.483870967741936,
      "grad_norm": 0.5631425976753235,
      "learning_rate": 4.0229885057471266e-07,
      "logits/chosen": -1.882641077041626,
      "logits/rejected": -1.8940929174423218,
      "logps/chosen": -163.1198272705078,
      "logps/rejected": -238.10739135742188,
      "loss": 0.0451,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.1276298761367798,
      "rewards/margins": 3.9089503288269043,
      "rewards/rejected": -2.781320333480835,
      "step": 170
    },
    {
      "epoch": 5.806451612903226,
      "grad_norm": 1.9172598123550415,
      "learning_rate": 1.1494252873563219e-07,
      "logits/chosen": -1.7833986282348633,
      "logits/rejected": -1.8089481592178345,
      "logps/chosen": -177.43270874023438,
      "logps/rejected": -286.78851318359375,
      "loss": 0.0497,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 1.6389026641845703,
      "rewards/margins": 4.935039520263672,
      "rewards/rejected": -3.2961373329162598,
      "step": 180
    },
    {
      "epoch": 5.806451612903226,
      "eval_logits/chosen": -1.6861690282821655,
      "eval_logits/rejected": -1.7177609205245972,
      "eval_logps/chosen": -105.73729705810547,
      "eval_logps/rejected": -235.87796020507812,
      "eval_loss": 0.040502406656742096,
      "eval_rewards/accuracies": 1.0,
      "eval_rewards/chosen": 1.4186561107635498,
      "eval_rewards/margins": 3.9982476234436035,
      "eval_rewards/rejected": -2.579591751098633,
      "eval_runtime": 10.1675,
      "eval_samples_per_second": 2.754,
      "eval_steps_per_second": 0.688,
      "step": 180
    }
  ],
  "logging_steps": 10,
  "max_steps": 180,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 90,
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}