{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.986666666666667,
  "eval_steps": 500,
  "global_step": 168,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17777777777777778,
      "grad_norm": 1.5870693922042847,
      "learning_rate": 4.957230266673969e-06,
      "logits/chosen": -2.5848517417907715,
      "logits/rejected": -2.592822313308716,
      "logps/chosen": -1.2300995588302612,
      "logps/rejected": -1.2750391960144043,
      "loss": 1.3004,
      "odds_ratio_loss": 0.7031511068344116,
      "rewards/accuracies": 0.45625001192092896,
      "rewards/chosen": -0.12300996482372284,
      "rewards/margins": 0.004493957385420799,
      "rewards/rejected": -0.1275039166212082,
      "sft_loss": 1.2300995588302612,
      "step": 10
    },
    {
      "epoch": 0.35555555555555557,
      "grad_norm": 2.5478012561798096,
      "learning_rate": 4.84529614391025e-06,
      "logits/chosen": -2.568402051925659,
      "logits/rejected": -2.5743260383605957,
      "logps/chosen": -1.1148786544799805,
      "logps/rejected": -1.2647678852081299,
      "loss": 1.1815,
      "odds_ratio_loss": 0.6664617657661438,
      "rewards/accuracies": 0.48124998807907104,
      "rewards/chosen": -0.11148785054683685,
      "rewards/margins": 0.014988938346505165,
      "rewards/rejected": -0.12647680938243866,
      "sft_loss": 1.1148786544799805,
      "step": 20
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 1.738682746887207,
      "learning_rate": 4.667009949002349e-06,
      "logits/chosen": -2.5901498794555664,
      "logits/rejected": -2.6051435470581055,
      "logps/chosen": -1.2092516422271729,
      "logps/rejected": -1.3586838245391846,
      "loss": 1.2768,
      "odds_ratio_loss": 0.6759374737739563,
      "rewards/accuracies": 0.4375,
      "rewards/chosen": -0.120925173163414,
      "rewards/margins": 0.01494320947676897,
      "rewards/rejected": -0.1358683854341507,
      "sft_loss": 1.2092516422271729,
      "step": 30
    },
    {
      "epoch": 0.7111111111111111,
      "grad_norm": 10.248797416687012,
      "learning_rate": 4.397288409237892e-06,
      "logits/chosen": -2.4943628311157227,
      "logits/rejected": -2.5201773643493652,
      "logps/chosen": -1.1567463874816895,
      "logps/rejected": -1.3316763639450073,
      "loss": 1.2213,
      "odds_ratio_loss": 0.6450860500335693,
      "rewards/accuracies": 0.48124998807907104,
      "rewards/chosen": -0.11567463725805283,
      "rewards/margins": 0.017492998391389847,
      "rewards/rejected": -0.13316763937473297,
      "sft_loss": 1.1567463874816895,
      "step": 40
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 6.022939682006836,
      "learning_rate": 4.097593272948218e-06,
      "logits/chosen": -2.6041746139526367,
      "logits/rejected": -2.6204562187194824,
      "logps/chosen": -1.2715274095535278,
      "logps/rejected": -1.6485040187835693,
      "loss": 1.3385,
      "odds_ratio_loss": 0.6699766516685486,
      "rewards/accuracies": 0.48124998807907104,
      "rewards/chosen": -0.12715274095535278,
      "rewards/margins": 0.03769766539335251,
      "rewards/rejected": -0.1648503988981247,
      "sft_loss": 1.2715274095535278,
      "step": 50
    },
    {
      "epoch": 1.0666666666666667,
      "grad_norm": 1.5987056493759155,
      "learning_rate": 3.712001321256744e-06,
      "logits/chosen": -2.6027374267578125,
      "logits/rejected": -2.627061128616333,
      "logps/chosen": -1.0649330615997314,
      "logps/rejected": -1.3192559480667114,
      "loss": 1.1266,
      "odds_ratio_loss": 0.616315484046936,
      "rewards/accuracies": 0.550000011920929,
      "rewards/chosen": -0.10649331659078598,
      "rewards/margins": 0.025432268157601357,
      "rewards/rejected": -0.13192559778690338,
      "sft_loss": 1.0649330615997314,
      "step": 60
    },
    {
      "epoch": 1.2444444444444445,
      "grad_norm": 9.42609977722168,
      "learning_rate": 3.284100248421622e-06,
      "logits/chosen": -2.592658758163452,
      "logits/rejected": -2.612031936645508,
      "logps/chosen": -1.0701837539672852,
      "logps/rejected": -1.2456352710723877,
      "loss": 1.1365,
      "odds_ratio_loss": 0.6634659767150879,
      "rewards/accuracies": 0.512499988079071,
      "rewards/chosen": -0.10701838880777359,
      "rewards/margins": 0.017545154318213463,
      "rewards/rejected": -0.1245635375380516,
      "sft_loss": 1.0701837539672852,
      "step": 70
    },
    {
      "epoch": 1.4222222222222223,
      "grad_norm": 1.1247073411941528,
      "learning_rate": 2.8288274295187397e-06,
      "logits/chosen": -2.6690917015075684,
      "logits/rejected": -2.6884384155273438,
      "logps/chosen": -1.0463426113128662,
      "logps/rejected": -1.2134628295898438,
      "loss": 1.1088,
      "odds_ratio_loss": 0.6244794130325317,
      "rewards/accuracies": 0.4749999940395355,
      "rewards/chosen": -0.10463427007198334,
      "rewards/margins": 0.016712015494704247,
      "rewards/rejected": -0.12134629487991333,
      "sft_loss": 1.0463426113128662,
      "step": 80
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.079381942749023,
      "learning_rate": 2.3620757456122797e-06,
      "logits/chosen": -2.638796329498291,
      "logits/rejected": -2.6489083766937256,
      "logps/chosen": -1.0774247646331787,
      "logps/rejected": -1.465980887413025,
      "loss": 1.1411,
      "odds_ratio_loss": 0.6369243860244751,
      "rewards/accuracies": 0.512499988079071,
      "rewards/chosen": -0.10774248838424683,
      "rewards/margins": 0.03885561600327492,
      "rewards/rejected": -0.14659810066223145,
      "sft_loss": 1.0774247646331787,
      "step": 90
    },
    {
      "epoch": 1.7777777777777777,
      "grad_norm": 3.1488540172576904,
      "learning_rate": 1.9001387874698865e-06,
      "logits/chosen": -2.659306049346924,
      "logits/rejected": -2.67085599899292,
      "logps/chosen": -1.0682704448699951,
      "logps/rejected": -1.2112770080566406,
      "loss": 1.1326,
      "odds_ratio_loss": 0.6432015299797058,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.10682705789804459,
      "rewards/margins": 0.01430064719170332,
      "rewards/rejected": -0.12112770229578018,
      "sft_loss": 1.0682704448699951,
      "step": 100
    },
    {
      "epoch": 1.9555555555555557,
      "grad_norm": 2.191756010055542,
      "learning_rate": 1.4591420711120425e-06,
      "logits/chosen": -2.685157299041748,
      "logits/rejected": -2.695525646209717,
      "logps/chosen": -1.0902485847473145,
      "logps/rejected": -1.2659653425216675,
      "loss": 1.1523,
      "odds_ratio_loss": 0.6206409335136414,
      "rewards/accuracies": 0.4937500059604645,
      "rewards/chosen": -0.10902485996484756,
      "rewards/margins": 0.017571674659848213,
      "rewards/rejected": -0.12659654021263123,
      "sft_loss": 1.0902485847473145,
      "step": 110
    },
    {
      "epoch": 2.1333333333333333,
      "grad_norm": 1.3168927431106567,
      "learning_rate": 1.0544801205950327e-06,
      "logits/chosen": -2.6532204151153564,
      "logits/rejected": -2.6720328330993652,
      "logps/chosen": -1.0528920888900757,
      "logps/rejected": -1.2378777265548706,
      "loss": 1.1173,
      "odds_ratio_loss": 0.6442069411277771,
      "rewards/accuracies": 0.48750001192092896,
      "rewards/chosen": -0.10528920590877533,
      "rewards/margins": 0.018498573452234268,
      "rewards/rejected": -0.1237877756357193,
      "sft_loss": 1.0528920888900757,
      "step": 120
    },
    {
      "epoch": 2.311111111111111,
      "grad_norm": 0.9034324288368225,
      "learning_rate": 7.002790686106056e-07,
      "logits/chosen": -2.6588597297668457,
      "logits/rejected": -2.6591033935546875,
      "logps/chosen": -0.9490475654602051,
      "logps/rejected": -1.0968741178512573,
      "loss": 1.0107,
      "odds_ratio_loss": 0.616373598575592,
      "rewards/accuracies": 0.48750001192092896,
      "rewards/chosen": -0.09490474313497543,
      "rewards/margins": 0.01478266529738903,
      "rewards/rejected": -0.1096874251961708,
      "sft_loss": 0.9490475654602051,
      "step": 130
    },
    {
      "epoch": 2.488888888888889,
      "grad_norm": 1.7593573331832886,
      "learning_rate": 4.0890353469735717e-07,
      "logits/chosen": -2.686373472213745,
      "logits/rejected": -2.7087228298187256,
      "logps/chosen": -1.0427912473678589,
      "logps/rejected": -1.486315131187439,
      "loss": 1.1034,
      "odds_ratio_loss": 0.6059055328369141,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -0.10427911579608917,
      "rewards/margins": 0.04435240477323532,
      "rewards/rejected": -0.1486315280199051,
      "sft_loss": 1.0427912473678589,
      "step": 140
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 5.990176200866699,
      "learning_rate": 1.905249951948071e-07,
      "logits/chosen": -2.715557813644409,
      "logits/rejected": -2.731482982635498,
      "logps/chosen": -1.144046425819397,
      "logps/rejected": -1.2883504629135132,
      "loss": 1.2074,
      "odds_ratio_loss": 0.6331667304039001,
      "rewards/accuracies": 0.48124998807907104,
      "rewards/chosen": -0.11440464109182358,
      "rewards/margins": 0.014430420473217964,
      "rewards/rejected": -0.1288350522518158,
      "sft_loss": 1.144046425819397,
      "step": 150
    },
    {
      "epoch": 2.8444444444444446,
      "grad_norm": 2.793104887008667,
      "learning_rate": 5.2766712488021566e-08,
      "logits/chosen": -2.695949077606201,
      "logits/rejected": -2.7109360694885254,
      "logps/chosen": -0.9868586659431458,
      "logps/rejected": -1.2109596729278564,
      "loss": 1.0456,
      "odds_ratio_loss": 0.5878169536590576,
      "rewards/accuracies": 0.581250011920929,
      "rewards/chosen": -0.0986858680844307,
      "rewards/margins": 0.022410111501812935,
      "rewards/rejected": -0.12109597772359848,
      "sft_loss": 0.9868586659431458,
      "step": 160
    },
    {
      "epoch": 2.986666666666667,
      "step": 168,
      "total_flos": 2.303993975490478e+17,
      "train_loss": 1.1586682626179285,
      "train_runtime": 5352.1206,
      "train_samples_per_second": 0.504,
      "train_steps_per_second": 0.031
    }
  ],
  "logging_steps": 10,
  "max_steps": 168,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2.303993975490478e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}