{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9991024490319271,
  "eval_steps": 400,
  "global_step": 974,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "count/fg_chosen": 14.079999923706055,
      "count/fg_rejected": 3.7826087474823,
      "epoch": 0.01025772534940377,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.78272247314453,
      "fg_logps/policy_chosen": -15.784832000732422,
      "fg_logps/policy_rejected": -17.75761604309082,
      "fg_logps/reference_KL": -21.78750991821289,
      "fg_logps/reference_chosen": -15.790050506591797,
      "fg_logps/reference_rejected": -17.756908416748047,
      "fg_loss": 0.808033287525177,
      "fg_rewards/chosen_sum": 0.006448890082538128,
      "fg_rewards/rejected_sum": -2.7408803362050094e-05,
      "grad_norm": 18.746104238140656,
      "kl": 0.03848903253674507,
      "learning_rate": 2.1220544726955587e-07,
      "logps/chosen": -305.5322529560811,
      "logps/rejected": -327.7266306322674,
      "loss": 0.6362,
      "rewards/chosen": -0.001747507501292873,
      "rewards/margins": -0.002081668250248688,
      "rewards/rejected": 0.00033416074895581535,
      "step": 10
    },
    {
      "count/fg_chosen": 13.947368621826172,
      "count/fg_rejected": 3.5714285373687744,
      "epoch": 0.02051545069880754,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.42728614807129,
      "fg_logps/policy_chosen": -15.204089164733887,
      "fg_logps/policy_rejected": -20.7266902923584,
      "fg_logps/reference_KL": -20.41632652282715,
      "fg_logps/reference_chosen": -15.188560485839844,
      "fg_logps/reference_rejected": -20.721162796020508,
      "fg_loss": 0.7661333084106445,
      "fg_rewards/chosen_sum": -0.025545625016093254,
      "fg_rewards/rejected_sum": -0.00522112101316452,
      "grad_norm": 19.071229786810708,
      "kl": 0.006695461459457874,
      "learning_rate": 3.0896765977473373e-07,
      "logps/chosen": -313.60243626644734,
      "logps/rejected": -306.7811569940476,
      "loss": 0.5877,
      "rewards/chosen": -0.012838767547356454,
      "rewards/margins": 0.004141214908215037,
      "rewards/rejected": -0.01697998245557149,
      "step": 20
    },
    {
      "count/fg_chosen": 15.894737243652344,
      "count/fg_rejected": 3.5,
      "epoch": 0.03077317604821131,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.23096466064453,
      "fg_logps/policy_chosen": -14.642033576965332,
      "fg_logps/policy_rejected": -16.705078125,
      "fg_logps/reference_KL": -19.1680850982666,
      "fg_logps/reference_chosen": -14.53127384185791,
      "fg_logps/reference_rejected": -16.613784790039062,
      "fg_loss": 0.7854437232017517,
      "fg_rewards/chosen_sum": -0.16305610537528992,
      "fg_rewards/rejected_sum": -0.03273481875658035,
      "grad_norm": 27.51124230932652,
      "kl": 0.0037935494910925627,
      "learning_rate": 3.5530203356711474e-07,
      "logps/chosen": -328.168653612013,
      "logps/rejected": -307.9456419427711,
      "loss": 0.5906,
      "rewards/chosen": -0.04671859121941901,
      "rewards/margins": 0.03360887939913921,
      "rewards/rejected": -0.08032747061855822,
      "step": 30
    },
    {
      "count/fg_chosen": 16.538461685180664,
      "count/fg_rejected": 3.4761905670166016,
      "epoch": 0.04103090139761508,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.668180465698242,
      "fg_logps/policy_chosen": -13.914729118347168,
      "fg_logps/policy_rejected": -17.876625061035156,
      "fg_logps/reference_KL": -19.54073143005371,
      "fg_logps/reference_chosen": -13.71956729888916,
      "fg_logps/reference_rejected": -17.639841079711914,
      "fg_loss": 0.7382856011390686,
      "fg_rewards/chosen_sum": -0.32776451110839844,
      "fg_rewards/rejected_sum": -0.08581020683050156,
      "grad_norm": 18.904689541215447,
      "kl": 0.0,
      "learning_rate": 3.907900061589056e-07,
      "logps/chosen": -324.57538132440476,
      "logps/rejected": -341.6456363075658,
      "loss": 0.6186,
      "rewards/chosen": -0.12434386071704683,
      "rewards/margins": 0.07653144666724336,
      "rewards/rejected": -0.20087530738429019,
      "step": 40
    },
    {
      "count/fg_chosen": 13.052631378173828,
      "count/fg_rejected": 3.5999999046325684,
      "epoch": 0.051288626747018846,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.696481704711914,
      "fg_logps/policy_chosen": -14.908551216125488,
      "fg_logps/policy_rejected": -17.680747985839844,
      "fg_logps/reference_KL": -22.48174476623535,
      "fg_logps/reference_chosen": -14.603896141052246,
      "fg_logps/reference_rejected": -17.34904670715332,
      "fg_loss": 0.7404539585113525,
      "fg_rewards/chosen_sum": -0.35476553440093994,
      "fg_rewards/rejected_sum": -0.14453592896461487,
      "grad_norm": 14.07450204568209,
      "kl": 0.014550399966537952,
      "learning_rate": 4.175211072161436e-07,
      "logps/chosen": -291.7388392857143,
      "logps/rejected": -314.5031061746988,
      "loss": 0.5781,
      "rewards/chosen": -0.10712227264007965,
      "rewards/margins": 0.12518125482426995,
      "rewards/rejected": -0.2323035274643496,
      "step": 50
    },
    {
      "count/fg_chosen": 14.5,
      "count/fg_rejected": 4.166666507720947,
      "epoch": 0.06154635209642262,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.49053955078125,
      "fg_logps/policy_chosen": -15.30705738067627,
      "fg_logps/policy_rejected": -18.280315399169922,
      "fg_logps/reference_KL": -20.184072494506836,
      "fg_logps/reference_chosen": -14.819132804870605,
      "fg_logps/reference_rejected": -17.81497573852539,
      "fg_loss": 0.7277103066444397,
      "fg_rewards/chosen_sum": -0.6670733690261841,
      "fg_rewards/rejected_sum": -0.22372370958328247,
      "grad_norm": 17.53900194639282,
      "kl": 0.0,
      "learning_rate": 4.389727636522206e-07,
      "logps/chosen": -306.5228568412162,
      "logps/rejected": -308.6946538880814,
      "loss": 0.5582,
      "rewards/chosen": -0.19411424688390783,
      "rewards/margins": 0.22253594857053438,
      "rewards/rejected": -0.4166501954544422,
      "step": 60
    },
    {
      "count/fg_chosen": 15.692307472229004,
      "count/fg_rejected": 4.458333492279053,
      "epoch": 0.07180407744582638,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.50032615661621,
      "fg_logps/policy_chosen": -15.366405487060547,
      "fg_logps/policy_rejected": -20.129003524780273,
      "fg_logps/reference_KL": -20.127044677734375,
      "fg_logps/reference_chosen": -14.826613426208496,
      "fg_logps/reference_rejected": -19.553565979003906,
      "fg_loss": 0.7948136925697327,
      "fg_rewards/chosen_sum": -0.8405532836914062,
      "fg_rewards/rejected_sum": -0.24530810117721558,
      "grad_norm": 17.707988179125024,
      "kl": 0.016248703002929688,
      "learning_rate": 4.568903445353587e-07,
      "logps/chosen": -266.67440580985914,
      "logps/rejected": -392.16322858146066,
      "loss": 0.5804,
      "rewards/chosen": -0.3093531433965119,
      "rewards/margins": 0.4345650290326051,
      "rewards/rejected": -0.743918172429117,
      "step": 70
    },
    {
      "count/fg_chosen": 13.82608699798584,
      "count/fg_rejected": 4.736842155456543,
      "epoch": 0.08206180279523016,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.310537338256836,
      "fg_logps/policy_chosen": -15.233660697937012,
      "fg_logps/policy_rejected": -19.051939010620117,
      "fg_logps/reference_KL": -19.766571044921875,
      "fg_logps/reference_chosen": -14.48674201965332,
      "fg_logps/reference_rejected": -18.337350845336914,
      "fg_loss": 0.840079665184021,
      "fg_rewards/chosen_sum": -1.008780837059021,
      "fg_rewards/rejected_sum": -0.38555219769477844,
      "grad_norm": 16.529452607844135,
      "kl": 0.0,
      "learning_rate": 4.7227525172567226e-07,
      "logps/chosen": -258.29019579475306,
      "logps/rejected": -339.7275761471519,
      "loss": 0.5753,
      "rewards/chosen": -0.38390738875777636,
      "rewards/margins": 0.5281775528648007,
      "rewards/rejected": -0.9120849416225771,
      "step": 80
    },
    {
      "count/fg_chosen": 16.94444465637207,
      "count/fg_rejected": 4.5625,
      "epoch": 0.09231952814463393,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.532676696777344,
      "fg_logps/policy_chosen": -14.168781280517578,
      "fg_logps/policy_rejected": -15.193059921264648,
      "fg_logps/reference_KL": -17.916645050048828,
      "fg_logps/reference_chosen": -13.280251502990723,
      "fg_logps/reference_rejected": -14.359517097473145,
      "fg_loss": 0.7089979648590088,
      "fg_rewards/chosen_sum": -1.4005531072616577,
      "fg_rewards/rejected_sum": -0.3471396267414093,
      "grad_norm": 22.36999453602448,
      "kl": 0.013136100955307484,
      "learning_rate": 4.857556042627658e-07,
      "logps/chosen": -354.51475432981925,
      "logps/rejected": -378.41106939935065,
      "loss": 0.55,
      "rewards/chosen": -0.5630912321159639,
      "rewards/margins": 0.6156807405402861,
      "rewards/rejected": -1.17877197265625,
      "step": 90
    },
    {
      "count/fg_chosen": 17.559999465942383,
      "count/fg_rejected": 3.4583332538604736,
      "epoch": 0.10257725349403769,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.625595092773438,
      "fg_logps/policy_chosen": -14.884417533874512,
      "fg_logps/policy_rejected": -19.52889060974121,
      "fg_logps/reference_KL": -18.972139358520508,
      "fg_logps/reference_chosen": -13.96281623840332,
      "fg_logps/reference_rejected": -18.573102951049805,
      "fg_loss": 0.7564048171043396,
      "fg_rewards/chosen_sum": -1.531907081604004,
      "fg_rewards/rejected_sum": -0.3621842861175537,
      "grad_norm": 16.9566179157687,
      "kl": 0.0,
      "learning_rate": 4.977514249230057e-07,
      "logps/chosen": -296.1236049107143,
      "logps/rejected": -352.51732113486844,
      "loss": 0.6056,
      "rewards/chosen": -0.8061084747314453,
      "rewards/margins": 0.3998990309865851,
      "rewards/rejected": -1.2060075057180304,
      "step": 100
    },
    {
      "count/fg_chosen": 17.423076629638672,
      "count/fg_rejected": 3.277777671813965,
      "epoch": 0.11283497884344147,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.713420867919922,
      "fg_logps/policy_chosen": -14.724455833435059,
      "fg_logps/policy_rejected": -17.52227020263672,
      "fg_logps/reference_KL": -19.24376678466797,
      "fg_logps/reference_chosen": -14.040182113647461,
      "fg_logps/reference_rejected": -16.54549789428711,
      "fg_loss": 0.6636641621589661,
      "fg_rewards/chosen_sum": -0.9770669341087341,
      "fg_rewards/rejected_sum": -0.3046890199184418,
      "grad_norm": 15.8810865349395,
      "kl": 0.013418341055512428,
      "learning_rate": 4.960045662100457e-07,
      "logps/chosen": -319.42288602941176,
      "logps/rejected": -328.1127604166667,
      "loss": 0.5794,
      "rewards/chosen": -0.6423200719496783,
      "rewards/margins": 0.5826459518133426,
      "rewards/rejected": -1.2249660237630209,
      "step": 110
    },
    {
      "count/fg_chosen": 17.038461685180664,
      "count/fg_rejected": 3.549999952316284,
      "epoch": 0.12309270419284524,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.378173828125,
      "fg_logps/policy_chosen": -15.218153953552246,
      "fg_logps/policy_rejected": -17.869609832763672,
      "fg_logps/reference_KL": -20.997371673583984,
      "fg_logps/reference_chosen": -14.897927284240723,
      "fg_logps/reference_rejected": -17.389179229736328,
      "fg_loss": 0.7092593908309937,
      "fg_rewards/chosen_sum": -0.43586668372154236,
      "fg_rewards/rejected_sum": -0.22385962307453156,
      "grad_norm": 13.44458367896507,
      "kl": 0.11229915916919708,
      "learning_rate": 4.90296803652968e-07,
      "logps/chosen": -279.3057077891791,
      "logps/rejected": -315.11082409274195,
      "loss": 0.5637,
      "rewards/chosen": -0.42403762020281893,
      "rewards/margins": 0.6563830071860841,
      "rewards/rejected": -1.080420627388903,
      "step": 120
    },
    {
      "count/fg_chosen": 18.882352828979492,
      "count/fg_rejected": 4.5333333015441895,
      "epoch": 0.13335042954224902,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.839466094970703,
      "fg_logps/policy_chosen": -14.21582317352295,
      "fg_logps/policy_rejected": -15.989607810974121,
      "fg_logps/reference_KL": -18.35956382751465,
      "fg_logps/reference_chosen": -13.778548240661621,
      "fg_logps/reference_rejected": -15.061809539794922,
      "fg_loss": 0.752821147441864,
      "fg_rewards/chosen_sum": -0.353499174118042,
      "fg_rewards/rejected_sum": -0.5261321663856506,
      "grad_norm": 18.943079661475036,
      "kl": 0.0,
      "learning_rate": 4.845890410958904e-07,
      "logps/chosen": -399.24955100574715,
      "logps/rejected": -355.0519852311644,
      "loss": 0.5522,
      "rewards/chosen": -0.5169386589664152,
      "rewards/margins": 0.7560970959297749,
      "rewards/rejected": -1.2730357548961901,
      "step": 130
    },
    {
      "count/fg_chosen": 14.600000381469727,
      "count/fg_rejected": 3.2941176891326904,
      "epoch": 0.14360815489165277,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.31153106689453,
      "fg_logps/policy_chosen": -15.935779571533203,
      "fg_logps/policy_rejected": -19.569046020507812,
      "fg_logps/reference_KL": -21.87625503540039,
      "fg_logps/reference_chosen": -15.854820251464844,
      "fg_logps/reference_rejected": -19.18140983581543,
      "fg_loss": 0.6795336604118347,
      "fg_rewards/chosen_sum": 0.07530271261930466,
      "fg_rewards/rejected_sum": -0.0755414366722107,
      "grad_norm": 14.329744807366827,
      "kl": 0.010508442297577858,
      "learning_rate": 4.788812785388127e-07,
      "logps/chosen": -288.93102254746833,
      "logps/rejected": -339.4910059799383,
      "loss": 0.5424,
      "rewards/chosen": -0.38591092749486994,
      "rewards/margins": 0.9102081973062006,
      "rewards/rejected": -1.2961191248010706,
      "step": 140
    },
    {
      "count/fg_chosen": 17.049999237060547,
      "count/fg_rejected": 3.2352941036224365,
      "epoch": 0.15386588024105655,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.11859703063965,
      "fg_logps/policy_chosen": -15.353170394897461,
      "fg_logps/policy_rejected": -19.041038513183594,
      "fg_logps/reference_KL": -20.675270080566406,
      "fg_logps/reference_chosen": -15.19269847869873,
      "fg_logps/reference_rejected": -18.65860366821289,
      "fg_loss": 0.6720137596130371,
      "fg_rewards/chosen_sum": -0.040867485105991364,
      "fg_rewards/rejected_sum": -0.07588446885347366,
      "grad_norm": 15.943835226437113,
      "kl": 0.0,
      "learning_rate": 4.7317351598173515e-07,
      "logps/chosen": -372.51135706018516,
      "logps/rejected": -344.8247132120253,
      "loss": 0.5326,
      "rewards/chosen": -0.4752860363618827,
      "rewards/margins": 0.8493531029342507,
      "rewards/rejected": -1.3246391392961334,
      "step": 150
    },
    {
      "count/fg_chosen": 14.833333015441895,
      "count/fg_rejected": 2.642857074737549,
      "epoch": 0.16412360559046033,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.43903350830078,
      "fg_logps/policy_chosen": -14.500401496887207,
      "fg_logps/policy_rejected": -17.249717712402344,
      "fg_logps/reference_KL": -21.081066131591797,
      "fg_logps/reference_chosen": -14.807108879089355,
      "fg_logps/reference_rejected": -17.323772430419922,
      "fg_loss": 0.6379384994506836,
      "fg_rewards/chosen_sum": 0.5964276194572449,
      "fg_rewards/rejected_sum": -0.0015845863381400704,
      "grad_norm": 13.675438870382441,
      "kl": 0.0012632369762286544,
      "learning_rate": 4.6746575342465747e-07,
      "logps/chosen": -327.4292566636029,
      "logps/rejected": -389.55515455163044,
      "loss": 0.4826,
      "rewards/chosen": -0.45451857061947093,
      "rewards/margins": 0.7520349300121103,
      "rewards/rejected": -1.2065535006315813,
      "step": 160
    },
    {
      "count/fg_chosen": 18.764705657958984,
      "count/fg_rejected": 3.538461446762085,
      "epoch": 0.17438133093986408,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.866031646728516,
      "fg_logps/policy_chosen": -14.610063552856445,
      "fg_logps/policy_rejected": -15.764613151550293,
      "fg_logps/reference_KL": -19.758529663085938,
      "fg_logps/reference_chosen": -15.149678230285645,
      "fg_logps/reference_rejected": -15.872199058532715,
      "fg_loss": 0.7423752546310425,
      "fg_rewards/chosen_sum": 1.0474270582199097,
      "fg_rewards/rejected_sum": -0.00038578076055273414,
      "grad_norm": 14.85184229504073,
      "kl": 0.05892143398523331,
      "learning_rate": 4.617579908675799e-07,
      "logps/chosen": -289.5171331091772,
      "logps/rejected": -364.5406298225309,
      "loss": 0.5105,
      "rewards/chosen": -0.3850521618806863,
      "rewards/margins": 0.9919037888209918,
      "rewards/rejected": -1.3769559507016782,
      "step": 170
    },
    {
      "count/fg_chosen": 14.399999618530273,
      "count/fg_rejected": 4.5,
      "epoch": 0.18463905628926786,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.385639190673828,
      "fg_logps/policy_chosen": -15.719804763793945,
      "fg_logps/policy_rejected": -21.617740631103516,
      "fg_logps/reference_KL": -21.118305206298828,
      "fg_logps/reference_chosen": -15.717147827148438,
      "fg_logps/reference_rejected": -21.633119583129883,
      "fg_loss": 0.875866711139679,
      "fg_rewards/chosen_sum": 0.0969744473695755,
      "fg_rewards/rejected_sum": -0.13854043185710907,
      "grad_norm": 14.688764944309868,
      "kl": 0.0,
      "learning_rate": 4.5605022831050226e-07,
      "logps/chosen": -251.9897216796875,
      "logps/rejected": -383.8666259765625,
      "loss": 0.5364,
      "rewards/chosen": -0.6086512088775635,
      "rewards/margins": 1.0637765407562256,
      "rewards/rejected": -1.6724277496337892,
      "step": 180
    },
    {
      "count/fg_chosen": 19.217391967773438,
      "count/fg_rejected": 4.523809432983398,
      "epoch": 0.19489678163867163,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.718454360961914,
      "fg_logps/policy_chosen": -13.899271965026855,
      "fg_logps/policy_rejected": -18.827178955078125,
      "fg_logps/reference_KL": -18.324739456176758,
      "fg_logps/reference_chosen": -13.679468154907227,
      "fg_logps/reference_rejected": -18.46164321899414,
      "fg_loss": 0.7202942967414856,
      "fg_rewards/chosen_sum": -0.006031676661223173,
      "fg_rewards/rejected_sum": -0.15737372636795044,
      "grad_norm": 14.404560894432748,
      "kl": 0.05560264736413956,
      "learning_rate": 4.503424657534247e-07,
      "logps/chosen": -338.58002804487177,
      "logps/rejected": -338.0175543064024,
      "loss": 0.5344,
      "rewards/chosen": -0.542063737526918,
      "rewards/margins": 1.0375677887688135,
      "rewards/rejected": -1.5796315262957317,
      "step": 190
    },
    {
      "count/fg_chosen": 17.399999618530273,
      "count/fg_rejected": 4.789473533630371,
      "epoch": 0.20515450698807539,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.66611099243164,
      "fg_logps/policy_chosen": -14.653355598449707,
      "fg_logps/policy_rejected": -17.5184383392334,
      "fg_logps/reference_KL": -19.25302505493164,
      "fg_logps/reference_chosen": -14.066553115844727,
      "fg_logps/reference_rejected": -17.042163848876953,
      "fg_loss": 0.9406682848930359,
      "fg_rewards/chosen_sum": -0.7670488357543945,
      "fg_rewards/rejected_sum": -0.16450081765651703,
      "grad_norm": 13.472817295849191,
      "kl": 0.0,
      "learning_rate": 4.44634703196347e-07,
      "logps/chosen": -328.9538845486111,
      "logps/rejected": -382.8152622767857,
      "loss": 0.5533,
      "rewards/chosen": -0.6610004001193577,
      "rewards/margins": 1.229798332093254,
      "rewards/rejected": -1.8907987322126116,
      "step": 200
    },
    {
      "count/fg_chosen": 17.6875,
      "count/fg_rejected": 5.384615421295166,
      "epoch": 0.21541223233747916,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.451927185058594,
      "fg_logps/policy_chosen": -14.831332206726074,
      "fg_logps/policy_rejected": -17.854537963867188,
      "fg_logps/reference_KL": -20.010757446289062,
      "fg_logps/reference_chosen": -14.1827392578125,
      "fg_logps/reference_rejected": -16.983478546142578,
      "fg_loss": 0.7702024579048157,
      "fg_rewards/chosen_sum": -1.0590859651565552,
      "fg_rewards/rejected_sum": -0.4240546226501465,
      "grad_norm": 16.379366840236408,
      "kl": 0.0,
      "learning_rate": 4.3892694063926936e-07,
      "logps/chosen": -336.93881138392857,
      "logps/rejected": -410.259375,
      "loss": 0.4634,
      "rewards/chosen": -0.4613551548549107,
      "rewards/margins": 1.4628642733134922,
      "rewards/rejected": -1.9242194281684029,
      "step": 210
    },
    {
      "count/fg_chosen": 14.0,
      "count/fg_rejected": 4.181818008422852,
      "epoch": 0.22566995768688294,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.021493911743164,
      "fg_logps/policy_chosen": -16.28066062927246,
      "fg_logps/policy_rejected": -20.48377799987793,
      "fg_logps/reference_KL": -20.17574691772461,
      "fg_logps/reference_chosen": -15.166167259216309,
      "fg_logps/reference_rejected": -19.03717803955078,
      "fg_loss": 0.6626110672950745,
      "fg_rewards/chosen_sum": -1.1887147426605225,
      "fg_rewards/rejected_sum": -0.47799989581108093,
      "grad_norm": 13.99674523050283,
      "kl": 0.0,
      "learning_rate": 4.332191780821918e-07,
      "logps/chosen": -288.56021278782896,
      "logps/rejected": -361.5675455729167,
      "loss": 0.4491,
      "rewards/chosen": -0.5620980011789423,
      "rewards/margins": 1.510199412964938,
      "rewards/rejected": -2.0722974141438804,
      "step": 220
    },
    {
      "count/fg_chosen": 14.4399995803833,
      "count/fg_rejected": 4.888888835906982,
      "epoch": 0.2359276830362867,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.916282653808594,
      "fg_logps/policy_chosen": -15.752463340759277,
      "fg_logps/policy_rejected": -19.13568878173828,
      "fg_logps/reference_KL": -20.100528717041016,
      "fg_logps/reference_chosen": -14.519099235534668,
      "fg_logps/reference_rejected": -17.75797462463379,
      "fg_loss": 0.8086566925048828,
      "fg_rewards/chosen_sum": -1.4387943744659424,
      "fg_rewards/rejected_sum": -0.7463889122009277,
      "grad_norm": 9.096647980350944,
      "kl": 0.0,
      "learning_rate": 4.2751141552511415e-07,
      "logps/chosen": -325.9561071908602,
      "logps/rejected": -342.84818097014926,
      "loss": 0.5797,
      "rewards/chosen": -0.93502930671938,
      "rewards/margins": 1.3941938706068793,
      "rewards/rejected": -2.3292231773262593,
      "step": 230
    },
    {
      "count/fg_chosen": 19.450000762939453,
      "count/fg_rejected": 3.882352828979492,
      "epoch": 0.24618540838569047,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.968975067138672,
      "fg_logps/policy_chosen": -15.759946823120117,
      "fg_logps/policy_rejected": -18.90660858154297,
      "fg_logps/reference_KL": -19.11962890625,
      "fg_logps/reference_chosen": -14.694534301757812,
      "fg_logps/reference_rejected": -17.608003616333008,
      "fg_loss": 0.7771142721176147,
      "fg_rewards/chosen_sum": -2.0326945781707764,
      "fg_rewards/rejected_sum": -0.5817491412162781,
      "grad_norm": 13.085165433618942,
      "kl": 0.0,
      "learning_rate": 4.218036529680365e-07,
      "logps/chosen": -324.1948939732143,
      "logps/rejected": -355.4172092013889,
      "loss": 0.4984,
      "rewards/chosen": -0.8741681780133929,
      "rewards/margins": 0.9406820145864335,
      "rewards/rejected": -1.8148501925998264,
      "step": 240
    },
    {
      "count/fg_chosen": 15.941176414489746,
      "count/fg_rejected": 3.5625,
      "epoch": 0.25644313373509425,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.16470718383789,
      "fg_logps/policy_chosen": -15.544363975524902,
      "fg_logps/policy_rejected": -22.452640533447266,
      "fg_logps/reference_KL": -20.34503746032715,
      "fg_logps/reference_chosen": -14.364212989807129,
      "fg_logps/reference_rejected": -20.861942291259766,
      "fg_loss": 0.8010571599006653,
      "fg_rewards/chosen_sum": -1.5489312410354614,
      "fg_rewards/rejected_sum": -0.5739867687225342,
      "grad_norm": 20.269756145960844,
      "kl": 0.0004318237188272178,
      "learning_rate": 4.160958904109589e-07,
      "logps/chosen": -284.01116415895063,
      "logps/rejected": -337.00729331487344,
      "loss": 0.4967,
      "rewards/chosen": -0.7318777390468267,
      "rewards/margins": 1.3799085359235055,
      "rewards/rejected": -2.1117862749703322,
      "step": 250
    },
    {
      "count/fg_chosen": 14.647058486938477,
      "count/fg_rejected": 3.8333332538604736,
      "epoch": 0.26670085908449803,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.48432159423828,
      "fg_logps/policy_chosen": -15.442632675170898,
      "fg_logps/policy_rejected": -21.998336791992188,
      "fg_logps/reference_KL": -20.76389503479004,
      "fg_logps/reference_chosen": -14.410895347595215,
      "fg_logps/reference_rejected": -20.8482723236084,
      "fg_loss": 0.7149848937988281,
      "fg_rewards/chosen_sum": -0.9253503680229187,
      "fg_rewards/rejected_sum": -0.5536249279975891,
      "grad_norm": 13.117998935505918,
      "kl": 0.0,
      "learning_rate": 4.1038812785388125e-07,
      "logps/chosen": -288.13616943359375,
      "logps/rejected": -351.7694905598958,
      "loss": 0.4464,
      "rewards/chosen": -0.422292560338974,
      "rewards/margins": 1.4216915667057037,
      "rewards/rejected": -1.8439841270446777,
      "step": 260
    },
    {
      "count/fg_chosen": 14.785714149475098,
      "count/fg_rejected": 3.642857074737549,
      "epoch": 0.27695858443390176,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.614913940429688,
      "fg_logps/policy_chosen": -16.907217025756836,
      "fg_logps/policy_rejected": -19.078645706176758,
      "fg_logps/reference_KL": -21.76515007019043,
      "fg_logps/reference_chosen": -16.108179092407227,
      "fg_logps/reference_rejected": -18.381593704223633,
      "fg_loss": 0.819545567035675,
      "fg_rewards/chosen_sum": -0.7028852105140686,
      "fg_rewards/rejected_sum": -0.34045127034187317,
      "grad_norm": 18.77208905467823,
      "kl": 0.0,
      "learning_rate": 4.046803652968037e-07,
      "logps/chosen": -345.2209114583333,
      "logps/rejected": -358.41994485294117,
      "loss": 0.475,
      "rewards/chosen": -0.8343807983398438,
      "rewards/margins": 1.4960064517750458,
      "rewards/rejected": -2.3303872501148897,
      "step": 270
    },
    {
      "count/fg_chosen": 17.962963104248047,
      "count/fg_rejected": 5.884615421295166,
      "epoch": 0.28721630978330553,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.402196884155273,
      "fg_logps/policy_chosen": -15.454611778259277,
      "fg_logps/policy_rejected": -17.580909729003906,
      "fg_logps/reference_KL": -19.303863525390625,
      "fg_logps/reference_chosen": -14.586993217468262,
      "fg_logps/reference_rejected": -16.358680725097656,
      "fg_loss": 0.8306005597114563,
      "fg_rewards/chosen_sum": -1.1611218452453613,
      "fg_rewards/rejected_sum": -0.7079381942749023,
      "grad_norm": 13.80952507861355,
      "kl": 0.0,
      "learning_rate": 3.98972602739726e-07,
      "logps/chosen": -283.1960693359375,
      "logps/rejected": -412.07763671875,
      "loss": 0.5757,
      "rewards/chosen": -0.4877366542816162,
      "rewards/margins": 1.6564238071441653,
      "rewards/rejected": -2.1441604614257814,
      "step": 280
    },
    {
      "count/fg_chosen": 16.549999237060547,
      "count/fg_rejected": 5.277777671813965,
      "epoch": 0.2974740351327093,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.114633560180664,
      "fg_logps/policy_chosen": -15.825154304504395,
      "fg_logps/policy_rejected": -20.11035919189453,
      "fg_logps/reference_KL": -21.259445190429688,
      "fg_logps/reference_chosen": -15.046856880187988,
      "fg_logps/reference_rejected": -19.045265197753906,
      "fg_loss": 0.8168869614601135,
      "fg_rewards/chosen_sum": -0.7166315913200378,
      "fg_rewards/rejected_sum": -0.4433734118938446,
      "grad_norm": 14.177325128931766,
      "kl": 0.007422161288559437,
      "learning_rate": 3.9326484018264836e-07,
      "logps/chosen": -319.059375,
      "logps/rejected": -395.9953125,
      "loss": 0.5099,
      "rewards/chosen": -0.6542438761393229,
      "rewards/margins": 1.4790777288698682,
      "rewards/rejected": -2.133321605009191,
      "step": 290
    },
    {
      "count/fg_chosen": 13.82608699798584,
      "count/fg_rejected": 4.7727274894714355,
      "epoch": 0.3077317604821131,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.241731643676758,
      "fg_logps/policy_chosen": -16.656099319458008,
      "fg_logps/policy_rejected": -18.52686882019043,
      "fg_logps/reference_KL": -20.161357879638672,
      "fg_logps/reference_chosen": -15.559574127197266,
      "fg_logps/reference_rejected": -17.188322067260742,
      "fg_loss": 0.8407912850379944,
      "fg_rewards/chosen_sum": -1.1080502271652222,
      "fg_rewards/rejected_sum": -0.7194547057151794,
      "grad_norm": 11.479402820786039,
      "kl": 0.0,
      "learning_rate": 3.875570776255708e-07,
      "logps/chosen": -260.6196986607143,
      "logps/rejected": -403.26040097891564,
      "loss": 0.547,
      "rewards/chosen": -1.1976228193803267,
      "rewards/margins": 1.5034547452728173,
      "rewards/rejected": -2.701077564653144,
      "step": 300
    },
    {
      "count/fg_chosen": 17.0,
      "count/fg_rejected": 5.5333333015441895,
      "epoch": 0.31798948583151687,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.364051818847656,
      "fg_logps/policy_chosen": -15.91810417175293,
      "fg_logps/policy_rejected": -18.95937728881836,
      "fg_logps/reference_KL": -20.408035278320312,
      "fg_logps/reference_chosen": -14.995894432067871,
      "fg_logps/reference_rejected": -17.87520408630371,
      "fg_loss": 0.8350895047187805,
      "fg_rewards/chosen_sum": -1.1522879600524902,
      "fg_rewards/rejected_sum": -0.6541081070899963,
      "grad_norm": 12.005685770721653,
      "kl": 0.05910682678222656,
      "learning_rate": 3.8184931506849315e-07,
      "logps/chosen": -314.4315476190476,
      "logps/rejected": -347.2052580180921,
      "loss": 0.5311,
      "rewards/chosen": -0.8637207576206752,
      "rewards/margins": 1.5000286962752951,
      "rewards/rejected": -2.3637494538959705,
      "step": 310
    },
    {
      "count/fg_chosen": 14.047618865966797,
      "count/fg_rejected": 5.5,
      "epoch": 0.32824721118092065,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.606046676635742,
      "fg_logps/policy_chosen": -15.302529335021973,
      "fg_logps/policy_rejected": -20.181062698364258,
      "fg_logps/reference_KL": -20.86503791809082,
      "fg_logps/reference_chosen": -14.479813575744629,
      "fg_logps/reference_rejected": -19.489940643310547,
      "fg_loss": 0.7766517996788025,
      "fg_rewards/chosen_sum": -0.8992184400558472,
      "fg_rewards/rejected_sum": -0.4495273232460022,
      "grad_norm": 16.284183960507036,
      "kl": 0.0,
      "learning_rate": 3.761415525114155e-07,
      "logps/chosen": -260.18896484375,
      "logps/rejected": -417.2262073863636,
      "loss": 0.4704,
      "rewards/chosen": -0.6337332195705838,
      "rewards/margins": 1.9360473035561916,
      "rewards/rejected": -2.5697805231267754,
      "step": 320
    },
    {
      "count/fg_chosen": 16.22222137451172,
      "count/fg_rejected": 3.3333332538604736,
      "epoch": 0.3385049365303244,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.270763397216797,
      "fg_logps/policy_chosen": -16.191619873046875,
      "fg_logps/policy_rejected": -19.492420196533203,
      "fg_logps/reference_KL": -19.561973571777344,
      "fg_logps/reference_chosen": -15.197806358337402,
      "fg_logps/reference_rejected": -18.365306854248047,
      "fg_loss": 0.7132466435432434,
      "fg_rewards/chosen_sum": -1.5309462547302246,
      "fg_rewards/rejected_sum": -0.2937175929546356,
      "grad_norm": 7.818574791129848,
      "kl": 0.0,
      "learning_rate": 3.704337899543379e-07,
      "logps/chosen": -285.40846946022725,
      "logps/rejected": -327.64527925531917,
      "loss": 0.4679,
      "rewards/chosen": -1.0337238889752012,
      "rewards/margins": 1.5292926372673188,
      "rewards/rejected": -2.56301652624252,
      "step": 330
    },
    {
      "count/fg_chosen": 16.36842155456543,
      "count/fg_rejected": 8.75,
      "epoch": 0.34876266187972815,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.425670623779297,
      "fg_logps/policy_chosen": -13.722508430480957,
      "fg_logps/policy_rejected": -16.77787971496582,
      "fg_logps/reference_KL": -17.679426193237305,
      "fg_logps/reference_chosen": -13.155657768249512,
      "fg_logps/reference_rejected": -15.926578521728516,
      "fg_loss": 0.7438153624534607,
      "fg_rewards/chosen_sum": -0.5837015509605408,
      "fg_rewards/rejected_sum": -0.5333263874053955,
      "grad_norm": 9.802323506414373,
      "kl": 0.0,
      "learning_rate": 3.6472602739726025e-07,
      "logps/chosen": -300.561328125,
      "logps/rejected": -314.146240234375,
      "loss": 0.5005,
      "rewards/chosen": -0.9040058135986329,
      "rewards/margins": 1.4882362365722657,
      "rewards/rejected": -2.3922420501708985,
      "step": 340
    },
    {
      "count/fg_chosen": 16.14285659790039,
      "count/fg_rejected": 5.315789699554443,
      "epoch": 0.35902038722913193,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.929489135742188,
      "fg_logps/policy_chosen": -15.678800582885742,
      "fg_logps/policy_rejected": -19.949668884277344,
      "fg_logps/reference_KL": -19.97319984436035,
      "fg_logps/reference_chosen": -14.735365867614746,
      "fg_logps/reference_rejected": -18.699697494506836,
      "fg_loss": 0.9030398726463318,
      "fg_rewards/chosen_sum": -1.2398407459259033,
      "fg_rewards/rejected_sum": -0.8212930560112,
      "grad_norm": 8.643717223258967,
      "kl": 0.0,
      "learning_rate": 3.590182648401826e-07,
      "logps/chosen": -351.5288783482143,
      "logps/rejected": -359.00362356085526,
      "loss": 0.5488,
      "rewards/chosen": -1.1266803741455078,
      "rewards/margins": 1.3310967495566919,
      "rewards/rejected": -2.4577771237021997,
      "step": 350
    },
    {
      "count/fg_chosen": 14.214285850524902,
      "count/fg_rejected": 2.6666667461395264,
      "epoch": 0.3692781125785357,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -23.768299102783203,
      "fg_logps/policy_chosen": -16.969907760620117,
      "fg_logps/policy_rejected": -20.13111114501953,
      "fg_logps/reference_KL": -22.658279418945312,
      "fg_logps/reference_chosen": -15.871198654174805,
      "fg_logps/reference_rejected": -18.806407928466797,
      "fg_loss": 0.7307645678520203,
      "fg_rewards/chosen_sum": -0.9898194670677185,
      "fg_rewards/rejected_sum": -0.20603026449680328,
      "grad_norm": 8.142135660669938,
      "kl": 0.0,
      "learning_rate": 3.53310502283105e-07,
      "logps/chosen": -282.40988869863014,
      "logps/rejected": -349.8757632902299,
      "loss": 0.464,
      "rewards/chosen": -1.211091551062179,
      "rewards/margins": 1.277019747597573,
      "rewards/rejected": -2.488111298659752,
      "step": 360
    },
    {
      "count/fg_chosen": 15.15999984741211,
      "count/fg_rejected": 3.4166667461395264,
      "epoch": 0.3795358379279395,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.647724151611328,
      "fg_logps/policy_chosen": -15.575891494750977,
      "fg_logps/policy_rejected": -18.71537971496582,
      "fg_logps/reference_KL": -20.737442016601562,
      "fg_logps/reference_chosen": -14.95889663696289,
      "fg_logps/reference_rejected": -18.047760009765625,
      "fg_loss": 0.7305776476860046,
      "fg_rewards/chosen_sum": -0.11250638961791992,
      "fg_rewards/rejected_sum": -0.22363699972629547,
      "grad_norm": 10.677432473272551,
      "kl": 0.09735889732837677,
      "learning_rate": 3.476027397260274e-07,
      "logps/chosen": -302.0405040922619,
      "logps/rejected": -364.09403268914474,
      "loss": 0.5405,
      "rewards/chosen": -0.9889777047293526,
      "rewards/margins": 1.4359847907733199,
      "rewards/rejected": -2.4249624955026725,
      "step": 370
    },
    {
      "count/fg_chosen": 16.35714340209961,
      "count/fg_rejected": 4.461538314819336,
      "epoch": 0.38979356327734327,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -23.035646438598633,
      "fg_logps/policy_chosen": -16.462678909301758,
      "fg_logps/policy_rejected": -19.290132522583008,
      "fg_logps/reference_KL": -22.22718620300293,
      "fg_logps/reference_chosen": -16.05594825744629,
      "fg_logps/reference_rejected": -18.451143264770508,
      "fg_loss": 0.772994875907898,
      "fg_rewards/chosen_sum": -0.47333768010139465,
      "fg_rewards/rejected_sum": -0.43138498067855835,
      "grad_norm": 12.589055564062804,
      "kl": 0.0,
      "learning_rate": 3.418949771689498e-07,
      "logps/chosen": -303.3091415777439,
      "logps/rejected": -422.3029346955128,
      "loss": 0.4738,
      "rewards/chosen": -1.0245074760623094,
      "rewards/margins": 1.969195730317899,
      "rewards/rejected": -2.9937032063802085,
      "step": 380
    },
    {
      "count/fg_chosen": 15.047618865966797,
      "count/fg_rejected": 3.6666667461395264,
      "epoch": 0.400051288626747,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.204946517944336,
      "fg_logps/policy_chosen": -16.388538360595703,
      "fg_logps/policy_rejected": -19.46855926513672,
      "fg_logps/reference_KL": -21.356998443603516,
      "fg_logps/reference_chosen": -15.460061073303223,
      "fg_logps/reference_rejected": -17.982017517089844,
      "fg_loss": 0.7031415700912476,
      "fg_rewards/chosen_sum": -1.0803974866867065,
      "fg_rewards/rejected_sum": -0.587272047996521,
      "grad_norm": 13.352594752925569,
      "kl": 0.013134384527802467,
      "learning_rate": 3.361872146118721e-07,
      "logps/chosen": -344.073916153169,
      "logps/rejected": -408.09998244382024,
      "loss": 0.5212,
      "rewards/chosen": -1.473697286256602,
      "rewards/margins": 1.0633102491849358,
      "rewards/rejected": -2.537007535441538,
      "step": 390
    },
    {
      "count/fg_chosen": 14.714285850524902,
      "count/fg_rejected": 3.444444417953491,
      "epoch": 0.41030901397615077,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.212703704833984,
      "fg_logps/policy_chosen": -15.624214172363281,
      "fg_logps/policy_rejected": -18.56574249267578,
      "fg_logps/reference_KL": -20.04758071899414,
      "fg_logps/reference_chosen": -14.2003173828125,
      "fg_logps/reference_rejected": -17.362810134887695,
      "fg_loss": 0.7282993197441101,
      "fg_rewards/chosen_sum": -1.263887643814087,
      "fg_rewards/rejected_sum": -0.6069123148918152,
      "grad_norm": 10.430038786087746,
      "kl": 0.0,
      "learning_rate": 3.304794520547945e-07,
      "logps/chosen": -331.27660778985506,
      "logps/rejected": -399.446042239011,
      "loss": 0.4495,
      "rewards/chosen": -0.9196471615114074,
      "rewards/margins": 1.7936450546952378,
      "rewards/rejected": -2.713292216206645,
      "step": 400
    },
    {
      "epoch": 0.41030901397615077,
      "eval_count/fg_chosen": 16.01369857788086,
      "eval_count/fg_rejected": 4.0824174880981445,
      "eval_fg_kl": NaN,
      "eval_fg_logps/policy_KL": -21.182559967041016,
      "eval_fg_logps/policy_chosen": -16.106983184814453,
      "eval_fg_logps/policy_rejected": -20.267147064208984,
      "eval_fg_logps/reference_KL": -20.207021713256836,
      "eval_fg_logps/reference_chosen": -14.929487228393555,
      "eval_fg_logps/reference_rejected": -18.78682518005371,
      "eval_fg_loss": 0.744895875453949,
      "eval_fg_rewards/chosen_sum": -1.3183764219284058,
      "eval_fg_rewards/rejected_sum": -0.5732491612434387,
      "eval_kl": 0.005370728671550751,
      "eval_logps/chosen": -303.5075557511737,
      "eval_logps/rejected": -365.12120890022675,
      "eval_loss": 0.4977550208568573,
      "eval_rewards/chosen": -1.0396734246625587,
      "eval_rewards/margins": 1.678511342733983,
      "eval_rewards/rejected": -2.718184767396542,
      "eval_runtime": 806.8942,
      "eval_samples_per_second": 2.148,
      "eval_steps_per_second": 1.074,
      "step": 400
    },
    {
      "count/fg_chosen": 13.666666984558105,
      "count/fg_rejected": 3.3181817531585693,
      "epoch": 0.42056673932555455,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -26.42474937438965,
      "fg_logps/policy_chosen": -20.587556838989258,
      "fg_logps/policy_rejected": -19.216453552246094,
      "fg_logps/reference_KL": -25.272415161132812,
      "fg_logps/reference_chosen": -18.695642471313477,
      "fg_logps/reference_rejected": -16.948997497558594,
      "fg_loss": 0.8068851232528687,
      "fg_rewards/chosen_sum": -1.4713115692138672,
      "fg_rewards/rejected_sum": -0.7158498167991638,
      "grad_norm": 7.617410625793688,
      "kl": 0.0,
      "learning_rate": 3.247716894977169e-07,
      "logps/chosen": -285.80265184859155,
      "logps/rejected": -366.92332338483146,
      "loss": 0.4728,
      "rewards/chosen": -0.8456243058325539,
      "rewards/margins": 1.8824731597683062,
      "rewards/rejected": -2.72809746560086,
      "step": 410
    },
    {
      "count/fg_chosen": 15.4375,
      "count/fg_rejected": 3.799999952316284,
      "epoch": 0.43082446467495833,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.31452178955078,
      "fg_logps/policy_chosen": -16.454280853271484,
      "fg_logps/policy_rejected": -22.019365310668945,
      "fg_logps/reference_KL": -20.437578201293945,
      "fg_logps/reference_chosen": -15.224520683288574,
      "fg_logps/reference_rejected": -20.551807403564453,
      "fg_loss": 0.8755480647087097,
      "fg_rewards/chosen_sum": -1.7564505338668823,
      "fg_rewards/rejected_sum": -0.6421544551849365,
      "grad_norm": 14.132450428785305,
      "kl": 0.0,
      "learning_rate": 3.1906392694063925e-07,
      "logps/chosen": -319.19919463734567,
      "logps/rejected": -400.0357001582278,
      "loss": 0.494,
      "rewards/chosen": -1.2088648478190105,
      "rewards/margins": 1.8420809653237902,
      "rewards/rejected": -3.0509458131428007,
      "step": 420
    },
    {
      "count/fg_chosen": 15.55555534362793,
      "count/fg_rejected": 3.8235294818878174,
      "epoch": 0.4410821900243621,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.4989013671875,
      "fg_logps/policy_chosen": -16.78192138671875,
      "fg_logps/policy_rejected": -19.125896453857422,
      "fg_logps/reference_KL": -20.556718826293945,
      "fg_logps/reference_chosen": -15.185633659362793,
      "fg_logps/reference_rejected": -17.72509002685547,
      "fg_loss": 0.8559948205947876,
      "fg_rewards/chosen_sum": -1.8476042747497559,
      "fg_rewards/rejected_sum": -0.710954487323761,
      "grad_norm": 11.780117349913754,
      "kl": 0.0,
      "learning_rate": 3.133561643835616e-07,
      "logps/chosen": -301.3424876143293,
      "logps/rejected": -394.38636818910254,
      "loss": 0.4855,
      "rewards/chosen": -0.9122947134622713,
      "rewards/margins": 2.028106999591114,
      "rewards/rejected": -2.9404017130533853,
      "step": 430
    },
    {
      "count/fg_chosen": 16.0,
      "count/fg_rejected": 4.75,
      "epoch": 0.4513399153737659,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.407926559448242,
      "fg_logps/policy_chosen": -14.23660945892334,
      "fg_logps/policy_rejected": -19.2972354888916,
      "fg_logps/reference_KL": -17.730350494384766,
      "fg_logps/reference_chosen": -13.197031021118164,
      "fg_logps/reference_rejected": -17.424985885620117,
      "fg_loss": 0.9003034234046936,
      "fg_rewards/chosen_sum": -1.1896860599517822,
      "fg_rewards/rejected_sum": -0.8951053023338318,
      "grad_norm": 13.19151786370417,
      "kl": 0.0,
      "learning_rate": 3.07648401826484e-07,
      "logps/chosen": -381.12164834104937,
      "logps/rejected": -368.0486797863924,
      "loss": 0.4868,
      "rewards/chosen": -1.618426099235629,
      "rewards/margins": 0.9258255479409927,
      "rewards/rejected": -2.5442516471766217,
      "step": 440
    },
    {
      "count/fg_chosen": 14.588234901428223,
      "count/fg_rejected": 4.25,
      "epoch": 0.4615976407231696,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.48902130126953,
      "fg_logps/policy_chosen": -16.183223724365234,
      "fg_logps/policy_rejected": -20.021495819091797,
      "fg_logps/reference_KL": -20.65131950378418,
      "fg_logps/reference_chosen": -15.696285247802734,
      "fg_logps/reference_rejected": -18.851573944091797,
      "fg_loss": 0.8272340297698975,
      "fg_rewards/chosen_sum": -0.4996866285800934,
      "fg_rewards/rejected_sum": -0.5080854296684265,
      "grad_norm": 11.534503591617172,
      "kl": 0.038506411015987396,
      "learning_rate": 3.019406392694064e-07,
      "logps/chosen": -257.8545692845395,
      "logps/rejected": -389.2383277529762,
      "loss": 0.4618,
      "rewards/chosen": -1.0474451968544407,
      "rewards/margins": 1.9993883864323896,
      "rewards/rejected": -3.0468335832868303,
      "step": 450
    },
    {
      "count/fg_chosen": 13.0,
      "count/fg_rejected": 3.5999999046325684,
      "epoch": 0.4718553660725734,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -25.045570373535156,
      "fg_logps/policy_chosen": -18.018939971923828,
      "fg_logps/policy_rejected": -24.219295501708984,
      "fg_logps/reference_KL": -23.617990493774414,
      "fg_logps/reference_chosen": -16.746540069580078,
      "fg_logps/reference_rejected": -22.393386840820312,
      "fg_loss": 0.7469481825828552,
      "fg_rewards/chosen_sum": -0.8979846239089966,
      "fg_rewards/rejected_sum": -0.5428987145423889,
      "grad_norm": 9.60552438901234,
      "kl": 0.045855142176151276,
      "learning_rate": 2.9623287671232877e-07,
      "logps/chosen": -297.8484563253012,
      "logps/rejected": -350.15962357954544,
      "loss": 0.4823,
      "rewards/chosen": -1.071040233933782,
      "rewards/margins": 1.3700017867896352,
      "rewards/rejected": -2.441042020723417,
      "step": 460
    },
    {
      "count/fg_chosen": 18.479999542236328,
      "count/fg_rejected": 2.7727272510528564,
      "epoch": 0.48211309142197717,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.86458969116211,
      "fg_logps/policy_chosen": -13.49354362487793,
      "fg_logps/policy_rejected": -17.55750274658203,
      "fg_logps/reference_KL": -18.0754337310791,
      "fg_logps/reference_chosen": -13.031634330749512,
      "fg_logps/reference_rejected": -17.194589614868164,
      "fg_loss": 0.6403311491012573,
      "fg_rewards/chosen_sum": -0.32538533210754395,
      "fg_rewards/rejected_sum": -0.19457949697971344,
      "grad_norm": 10.665362097968403,
      "kl": 0.0,
      "learning_rate": 2.905251141552511e-07,
      "logps/chosen": -284.00316540948273,
      "logps/rejected": -342.78692208904107,
      "loss": 0.5004,
      "rewards/chosen": -0.6770254990150189,
      "rewards/margins": 2.199749119055401,
      "rewards/rejected": -2.8767746180704195,
      "step": 470
    },
    {
      "count/fg_chosen": 15.523809432983398,
      "count/fg_rejected": 4.5714287757873535,
      "epoch": 0.49237081677138095,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.194419860839844,
      "fg_logps/policy_chosen": -15.528361320495605,
      "fg_logps/policy_rejected": -17.91610336303711,
      "fg_logps/reference_KL": -19.58913803100586,
      "fg_logps/reference_chosen": -15.298972129821777,
      "fg_logps/reference_rejected": -17.643596649169922,
      "fg_loss": 0.7117694616317749,
      "fg_rewards/chosen_sum": 0.15593282878398895,
      "fg_rewards/rejected_sum": -0.3202955722808838,
      "grad_norm": 8.89018776362184,
      "kl": 0.0,
      "learning_rate": 2.848173515981735e-07,
      "logps/chosen": -366.11363389756946,
      "logps/rejected": -336.78042879971593,
      "loss": 0.5079,
      "rewards/chosen": -1.2623120413886175,
      "rewards/margins": 1.1003893746270075,
      "rewards/rejected": -2.362701416015625,
      "step": 480
    },
    {
      "count/fg_chosen": 18.47058868408203,
      "count/fg_rejected": 2.3333332538604736,
      "epoch": 0.5026285421207847,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.13791275024414,
      "fg_logps/policy_chosen": -15.788673400878906,
      "fg_logps/policy_rejected": -16.09772491455078,
      "fg_logps/reference_KL": -21.38779640197754,
      "fg_logps/reference_chosen": -15.412402153015137,
      "fg_logps/reference_rejected": -15.27255916595459,
      "fg_loss": 0.5969161987304688,
      "fg_rewards/chosen_sum": -0.09672492742538452,
      "fg_rewards/rejected_sum": -0.2052246332168579,
      "grad_norm": 9.190876919128005,
      "kl": 0.0,
      "learning_rate": 2.791095890410959e-07,
      "logps/chosen": -293.7143950591216,
      "logps/rejected": -421.82026707848837,
      "loss": 0.454,
      "rewards/chosen": -0.9402487471296981,
      "rewards/margins": 2.0355014884943934,
      "rewards/rejected": -2.9757502356240915,
      "step": 490
    },
    {
      "count/fg_chosen": 19.263158798217773,
      "count/fg_rejected": 3.1764705181121826,
      "epoch": 0.5128862674701885,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.44475746154785,
      "fg_logps/policy_chosen": -13.578866958618164,
      "fg_logps/policy_rejected": -16.77104949951172,
      "fg_logps/reference_KL": -18.07316780090332,
      "fg_logps/reference_chosen": -13.79211139678955,
      "fg_logps/reference_rejected": -16.677942276000977,
      "fg_loss": 0.8001216650009155,
      "fg_rewards/chosen_sum": 0.6667798161506653,
      "fg_rewards/rejected_sum": -0.12318305671215057,
      "grad_norm": 11.958302083795623,
      "kl": 0.06755809485912323,
      "learning_rate": 2.734018264840183e-07,
      "logps/chosen": -306.4125142911585,
      "logps/rejected": -373.8045873397436,
      "loss": 0.4904,
      "rewards/chosen": -0.8678309742997332,
      "rewards/margins": 1.8353534958525104,
      "rewards/rejected": -2.7031844701522436,
      "step": 500
    },
    {
      "count/fg_chosen": 17.157894134521484,
      "count/fg_rejected": 3.8888888359069824,
      "epoch": 0.5231439928195922,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.518945693969727,
      "fg_logps/policy_chosen": -14.510261535644531,
      "fg_logps/policy_rejected": -19.071426391601562,
      "fg_logps/reference_KL": -19.78034782409668,
      "fg_logps/reference_chosen": -15.087010383605957,
      "fg_logps/reference_rejected": -18.8708553314209,
      "fg_loss": 0.7365949749946594,
      "fg_rewards/chosen_sum": 1.3324851989746094,
      "fg_rewards/rejected_sum": -0.13402292132377625,
      "grad_norm": 11.840579393816673,
      "kl": 0.0,
      "learning_rate": 2.676940639269406e-07,
      "logps/chosen": -276.75473257211536,
      "logps/rejected": -376.3443216463415,
      "loss": 0.4727,
      "rewards/chosen": -0.42943524091671675,
      "rewards/margins": 2.161930108085284,
      "rewards/rejected": -2.5913653490020008,
      "step": 510
    },
    {
      "count/fg_chosen": 14.416666984558105,
      "count/fg_rejected": 4.473684310913086,
      "epoch": 0.5334017181689961,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -23.022239685058594,
      "fg_logps/policy_chosen": -15.866630554199219,
      "fg_logps/policy_rejected": -18.78508949279785,
      "fg_logps/reference_KL": -21.714487075805664,
      "fg_logps/reference_chosen": -15.187236785888672,
      "fg_logps/reference_rejected": -18.08281898498535,
      "fg_loss": 0.7238275408744812,
      "fg_rewards/chosen_sum": -0.16382475197315216,
      "fg_rewards/rejected_sum": -0.5171225070953369,
      "grad_norm": 12.176540624619156,
      "kl": 0.04069461673498154,
      "learning_rate": 2.61986301369863e-07,
      "logps/chosen": -308.16112012987014,
      "logps/rejected": -321.8852362575301,
      "loss": 0.5229,
      "rewards/chosen": -0.8690564044110187,
      "rewards/margins": 1.6282350686099714,
      "rewards/rejected": -2.49729147302099,
      "step": 520
    },
    {
      "count/fg_chosen": 16.81818199157715,
      "count/fg_rejected": 3.588235378265381,
      "epoch": 0.5436594435183998,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.363672256469727,
      "fg_logps/policy_chosen": -14.961847305297852,
      "fg_logps/policy_rejected": -21.47898292541504,
      "fg_logps/reference_KL": -19.768022537231445,
      "fg_logps/reference_chosen": -14.601174354553223,
      "fg_logps/reference_rejected": -20.11042594909668,
      "fg_loss": 0.6309139728546143,
      "fg_rewards/chosen_sum": 0.3970043957233429,
      "fg_rewards/rejected_sum": -0.6216399669647217,
      "grad_norm": 9.438462165579672,
      "kl": 0.017117690294981003,
      "learning_rate": 2.562785388127854e-07,
      "logps/chosen": -286.2251333841463,
      "logps/rejected": -331.3464793669872,
      "loss": 0.512,
      "rewards/chosen": -1.0500634356242855,
      "rewards/margins": 1.6451627935298612,
      "rewards/rejected": -2.6952262291541467,
      "step": 530
    },
    {
      "count/fg_chosen": 15.850000381469727,
      "count/fg_rejected": 2.6875,
      "epoch": 0.5539171688678035,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.712533950805664,
      "fg_logps/policy_chosen": -14.07727336883545,
      "fg_logps/policy_rejected": -18.43448829650879,
      "fg_logps/reference_KL": -18.728797912597656,
      "fg_logps/reference_chosen": -13.999521255493164,
      "fg_logps/reference_rejected": -17.450742721557617,
      "fg_loss": 0.6835896372795105,
      "fg_rewards/chosen_sum": 0.26210981607437134,
      "fg_rewards/rejected_sum": -0.32241931557655334,
      "grad_norm": 12.291319609660407,
      "kl": 0.0,
      "learning_rate": 2.5057077625570777e-07,
      "logps/chosen": -306.774658203125,
      "logps/rejected": -356.60145399305554,
      "loss": 0.5117,
      "rewards/chosen": -0.8661181709983132,
      "rewards/margins": 1.7041260883061573,
      "rewards/rejected": -2.5702442593044705,
      "step": 540
    },
    {
      "count/fg_chosen": 18.714284896850586,
      "count/fg_rejected": 4.722222328186035,
      "epoch": 0.5641748942172073,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.954633712768555,
      "fg_logps/policy_chosen": -13.810935974121094,
      "fg_logps/policy_rejected": -17.9754638671875,
      "fg_logps/reference_KL": -19.242063522338867,
      "fg_logps/reference_chosen": -13.879505157470703,
      "fg_logps/reference_rejected": -17.572124481201172,
      "fg_loss": 0.7275994420051575,
      "fg_rewards/chosen_sum": 0.31742504239082336,
      "fg_rewards/rejected_sum": -0.11265398561954498,
      "grad_norm": 8.68692731946405,
      "kl": 0.03491802141070366,
      "learning_rate": 2.4486301369863014e-07,
      "logps/chosen": -307.91368272569446,
      "logps/rejected": -344.2369939630682,
      "loss": 0.4965,
      "rewards/chosen": -0.7141922314961752,
      "rewards/margins": 1.454848968621456,
      "rewards/rejected": -2.1690412001176314,
      "step": 550
    },
    {
      "count/fg_chosen": 17.190475463867188,
      "count/fg_rejected": 3.5,
      "epoch": 0.5744326195666111,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.567123413085938,
      "fg_logps/policy_chosen": -14.335834503173828,
      "fg_logps/policy_rejected": -19.222152709960938,
      "fg_logps/reference_KL": -19.11014747619629,
      "fg_logps/reference_chosen": -14.28973388671875,
      "fg_logps/reference_rejected": -18.501300811767578,
      "fg_loss": 0.7595401406288147,
      "fg_rewards/chosen_sum": 0.21668264269828796,
      "fg_rewards/rejected_sum": -0.24018199741840363,
      "grad_norm": 13.295487655176196,
      "kl": 0.01426544226706028,
      "learning_rate": 2.391552511415525e-07,
      "logps/chosen": -340.173639871988,
      "logps/rejected": -346.4476207386364,
      "loss": 0.5126,
      "rewards/chosen": -0.6746720463396555,
      "rewards/margins": 1.3085595353111157,
      "rewards/rejected": -1.983231581650771,
      "step": 560
    },
    {
      "count/fg_chosen": 15.61111068725586,
      "count/fg_rejected": 3.230769157409668,
      "epoch": 0.5846903449160149,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.59079933166504,
      "fg_logps/policy_chosen": -15.83668327331543,
      "fg_logps/policy_rejected": -20.33367919921875,
      "fg_logps/reference_KL": -19.622900009155273,
      "fg_logps/reference_chosen": -14.985246658325195,
      "fg_logps/reference_rejected": -19.151033401489258,
      "fg_loss": 0.7142822742462158,
      "fg_rewards/chosen_sum": -1.0015321969985962,
      "fg_rewards/rejected_sum": -0.4118236005306244,
      "grad_norm": 14.880021607580103,
      "kl": 0.0,
      "learning_rate": 2.3344748858447487e-07,
      "logps/chosen": -321.2857349537037,
      "logps/rejected": -375.5545638844937,
      "loss": 0.4752,
      "rewards/chosen": -1.0196603845666956,
      "rewards/margins": 1.8235435825639412,
      "rewards/rejected": -2.843203967130637,
      "step": 570
    },
    {
      "count/fg_chosen": 14.615385055541992,
      "count/fg_rejected": 3.2727272510528564,
      "epoch": 0.5949480702654186,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.73748779296875,
      "fg_logps/policy_chosen": -16.519458770751953,
      "fg_logps/policy_rejected": -19.338417053222656,
      "fg_logps/reference_KL": -21.60148048400879,
      "fg_logps/reference_chosen": -15.273176193237305,
      "fg_logps/reference_rejected": -18.030052185058594,
      "fg_loss": 0.7020931243896484,
      "fg_rewards/chosen_sum": -1.3088314533233643,
      "fg_rewards/rejected_sum": -0.47078707814216614,
      "grad_norm": 16.46811452693527,
      "kl": 0.010451125912368298,
      "learning_rate": 2.2773972602739724e-07,
      "logps/chosen": -263.2815152391975,
      "logps/rejected": -350.35057357594934,
      "loss": 0.531,
      "rewards/chosen": -1.0503113358109086,
      "rewards/margins": 1.5224891194031933,
      "rewards/rejected": -2.572800455214102,
      "step": 580
    },
    {
      "count/fg_chosen": 15.391304016113281,
      "count/fg_rejected": 3.941176414489746,
      "epoch": 0.6052057956148225,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.143524169921875,
      "fg_logps/policy_chosen": -16.308269500732422,
      "fg_logps/policy_rejected": -21.572189331054688,
      "fg_logps/reference_KL": -20.941722869873047,
      "fg_logps/reference_chosen": -15.216353416442871,
      "fg_logps/reference_rejected": -19.638399124145508,
      "fg_loss": 0.7051105499267578,
      "fg_rewards/chosen_sum": -1.149840235710144,
      "fg_rewards/rejected_sum": -0.7558993101119995,
      "grad_norm": 10.114785462372456,
      "kl": 0.0,
      "learning_rate": 2.2203196347031963e-07,
      "logps/chosen": -397.5863940746753,
      "logps/rejected": -363.87947100903614,
      "loss": 0.4982,
      "rewards/chosen": -1.2836351270799513,
      "rewards/margins": 1.5669618408699733,
      "rewards/rejected": -2.8505969679499246,
      "step": 590
    },
    {
      "count/fg_chosen": 13.0,
      "count/fg_rejected": 3.461538553237915,
      "epoch": 0.6154635209642262,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.54435920715332,
      "fg_logps/policy_chosen": -16.62588119506836,
      "fg_logps/policy_rejected": -24.728574752807617,
      "fg_logps/reference_KL": -20.335765838623047,
      "fg_logps/reference_chosen": -15.187963485717773,
      "fg_logps/reference_rejected": -22.167306900024414,
      "fg_loss": 0.6712305545806885,
      "fg_rewards/chosen_sum": -1.496664047241211,
      "fg_rewards/rejected_sum": -0.9014762043952942,
      "grad_norm": 13.343339163568743,
      "kl": 0.0,
      "learning_rate": 2.16324200913242e-07,
      "logps/chosen": -308.1182432432432,
      "logps/rejected": -434.2431413517442,
      "loss": 0.4397,
      "rewards/chosen": -0.7798370155128272,
      "rewards/margins": 2.0589501520584848,
      "rewards/rejected": -2.838787167571312,
      "step": 600
    },
    {
      "count/fg_chosen": 14.222222328186035,
      "count/fg_rejected": 4.470588207244873,
      "epoch": 0.6257212463136299,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.16152572631836,
      "fg_logps/policy_chosen": -16.751249313354492,
      "fg_logps/policy_rejected": -19.572917938232422,
      "fg_logps/reference_KL": -20.091609954833984,
      "fg_logps/reference_chosen": -15.278923988342285,
      "fg_logps/reference_rejected": -17.713302612304688,
      "fg_loss": 0.7949765920639038,
      "fg_rewards/chosen_sum": -1.593385100364685,
      "fg_rewards/rejected_sum": -0.8538177609443665,
      "grad_norm": 13.466395129856458,
      "kl": 0.0,
      "learning_rate": 2.106164383561644e-07,
      "logps/chosen": -293.43453125,
      "logps/rejected": -324.9224264705882,
      "loss": 0.4617,
      "rewards/chosen": -1.0275954182942708,
      "rewards/margins": 1.7896552710439646,
      "rewards/rejected": -2.8172506893382354,
      "step": 610
    },
    {
      "count/fg_chosen": 18.760000228881836,
      "count/fg_rejected": 4.304347991943359,
      "epoch": 0.6359789716630337,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.953258514404297,
      "fg_logps/policy_chosen": -14.654019355773926,
      "fg_logps/policy_rejected": -18.075929641723633,
      "fg_logps/reference_KL": -18.280025482177734,
      "fg_logps/reference_chosen": -14.031706809997559,
      "fg_logps/reference_rejected": -16.499387741088867,
      "fg_loss": 0.753377377986908,
      "fg_rewards/chosen_sum": -0.787641704082489,
      "fg_rewards/rejected_sum": -0.6412522792816162,
      "grad_norm": 11.259123987500764,
      "kl": 0.0,
      "learning_rate": 2.0490867579908674e-07,
      "logps/chosen": -303.9673755787037,
      "logps/rejected": -421.6932357594937,
      "loss": 0.4916,
      "rewards/chosen": -0.5601168502995997,
      "rewards/margins": 2.57166944784864,
      "rewards/rejected": -3.13178629814824,
      "step": 620
    },
    {
      "count/fg_chosen": 19.100000381469727,
      "count/fg_rejected": 3.9375,
      "epoch": 0.6462366970124375,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -17.8834171295166,
      "fg_logps/policy_chosen": -12.904516220092773,
      "fg_logps/policy_rejected": -18.650463104248047,
      "fg_logps/reference_KL": -17.336223602294922,
      "fg_logps/reference_chosen": -12.578201293945312,
      "fg_logps/reference_rejected": -18.051301956176758,
      "fg_loss": 0.8024751543998718,
      "fg_rewards/chosen_sum": -0.19778959453105927,
      "fg_rewards/rejected_sum": -0.45491790771484375,
      "grad_norm": 11.633599458809858,
      "kl": 0.0,
      "learning_rate": 1.9920091324200913e-07,
      "logps/chosen": -319.94747740963857,
      "logps/rejected": -482.38976258116884,
      "loss": 0.4819,
      "rewards/chosen": -0.8397377657603069,
      "rewards/margins": 2.030176697827863,
      "rewards/rejected": -2.8699144635881697,
      "step": 630
    },
    {
      "count/fg_chosen": 14.666666984558105,
      "count/fg_rejected": 4.599999904632568,
      "epoch": 0.6564944223618413,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.435789108276367,
      "fg_logps/policy_chosen": -14.770197868347168,
      "fg_logps/policy_rejected": -21.445520401000977,
      "fg_logps/reference_KL": -20.57337760925293,
      "fg_logps/reference_chosen": -14.659423828125,
      "fg_logps/reference_rejected": -20.503273010253906,
      "fg_loss": 0.7505216598510742,
      "fg_rewards/chosen_sum": 0.48458048701286316,
      "fg_rewards/rejected_sum": -0.5640232563018799,
      "grad_norm": 10.675237142753284,
      "kl": 0.0,
      "learning_rate": 1.934931506849315e-07,
      "logps/chosen": -296.4902083333333,
      "logps/rejected": -395.98147977941176,
      "loss": 0.4865,
      "rewards/chosen": -0.7284186808268229,
      "rewards/margins": 1.9219920498717067,
      "rewards/rejected": -2.6504107306985296,
      "step": 640
    },
    {
      "count/fg_chosen": 13.600000381469727,
      "count/fg_rejected": 3.642857074737549,
      "epoch": 0.666752147711245,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.876583099365234,
      "fg_logps/policy_chosen": -15.880990982055664,
      "fg_logps/policy_rejected": -18.033313751220703,
      "fg_logps/reference_KL": -20.89072608947754,
      "fg_logps/reference_chosen": -15.644922256469727,
      "fg_logps/reference_rejected": -17.162748336791992,
      "fg_loss": 0.8513435125350952,
      "fg_rewards/chosen_sum": 0.21025246381759644,
      "fg_rewards/rejected_sum": -0.4860546588897705,
      "grad_norm": 11.01420885518033,
      "kl": 0.0,
      "learning_rate": 1.877853881278539e-07,
      "logps/chosen": -283.9858993902439,
      "logps/rejected": -359.59252303685895,
      "loss": 0.4902,
      "rewards/chosen": -0.6547578951207603,
      "rewards/margins": 1.8769426304075854,
      "rewards/rejected": -2.5317005255283456,
      "step": 650
    },
    {
      "count/fg_chosen": 15.045454978942871,
      "count/fg_rejected": 4.222222328186035,
      "epoch": 0.6770098730606487,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.071636199951172,
      "fg_logps/policy_chosen": -15.461947441101074,
      "fg_logps/policy_rejected": -16.1773738861084,
      "fg_logps/reference_KL": -20.327823638916016,
      "fg_logps/reference_chosen": -14.584894180297852,
      "fg_logps/reference_rejected": -15.22111988067627,
      "fg_loss": 0.7847881317138672,
      "fg_rewards/chosen_sum": -0.37347137928009033,
      "fg_rewards/rejected_sum": -0.5809666514396667,
      "grad_norm": 10.971292603882963,
      "kl": 0.0,
      "learning_rate": 1.8207762557077624e-07,
      "logps/chosen": -292.3511575838415,
      "logps/rejected": -411.4802684294872,
      "loss": 0.4927,
      "rewards/chosen": -0.615756523318407,
      "rewards/margins": 2.001151623466449,
      "rewards/rejected": -2.616908146784856,
      "step": 660
    },
    {
      "count/fg_chosen": 17.521739959716797,
      "count/fg_rejected": 3.8947367668151855,
      "epoch": 0.6872675984100526,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.02415657043457,
      "fg_logps/policy_chosen": -16.135929107666016,
      "fg_logps/policy_rejected": -17.662372589111328,
      "fg_logps/reference_KL": -20.120943069458008,
      "fg_logps/reference_chosen": -15.36439037322998,
      "fg_logps/reference_rejected": -16.619319915771484,
      "fg_loss": 0.6095057725906372,
      "fg_rewards/chosen_sum": -0.9091306924819946,
      "fg_rewards/rejected_sum": -0.6099987030029297,
      "grad_norm": 11.277192107560657,
      "kl": 0.0,
      "learning_rate": 1.7636986301369863e-07,
      "logps/chosen": -329.480522260274,
      "logps/rejected": -394.565149066092,
      "loss": 0.5064,
      "rewards/chosen": -1.2582495702456122,
      "rewards/margins": 1.6947859816805733,
      "rewards/rejected": -2.9530355519261855,
      "step": 670
    },
    {
      "count/fg_chosen": 17.882352828979492,
      "count/fg_rejected": 2.9166667461395264,
      "epoch": 0.6975253237594563,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.715160369873047,
      "fg_logps/policy_chosen": -13.929475784301758,
      "fg_logps/policy_rejected": -18.618078231811523,
      "fg_logps/reference_KL": -18.27341651916504,
      "fg_logps/reference_chosen": -13.522153854370117,
      "fg_logps/reference_rejected": -17.502342224121094,
      "fg_loss": 0.6171839237213135,
      "fg_rewards/chosen_sum": -0.3216637372970581,
      "fg_rewards/rejected_sum": -0.12122553586959839,
      "grad_norm": 10.871199583632286,
      "kl": 0.07263422012329102,
      "learning_rate": 1.70662100456621e-07,
      "logps/chosen": -290.58342978395063,
      "logps/rejected": -358.15182456487344,
      "loss": 0.473,
      "rewards/chosen": -0.5404636712721836,
      "rewards/margins": 2.1079875423230496,
      "rewards/rejected": -2.648451213595233,
      "step": 680
    },
    {
      "count/fg_chosen": 16.10714340209961,
      "count/fg_rejected": 3.5714285373687744,
      "epoch": 0.7077830491088601,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.88849449157715,
      "fg_logps/policy_chosen": -15.352334022521973,
      "fg_logps/policy_rejected": -19.385269165039062,
      "fg_logps/reference_KL": -19.914424896240234,
      "fg_logps/reference_chosen": -14.209139823913574,
      "fg_logps/reference_rejected": -18.167560577392578,
      "fg_loss": 0.6616089344024658,
      "fg_rewards/chosen_sum": -0.9578856229782104,
      "fg_rewards/rejected_sum": -0.687567949295044,
      "grad_norm": 7.140501610514079,
      "kl": 0.0,
      "learning_rate": 1.649543378995434e-07,
      "logps/chosen": -293.50786458333334,
      "logps/rejected": -386.10523897058823,
      "loss": 0.4551,
      "rewards/chosen": -0.3227819315592448,
      "rewards/margins": 2.913227015476601,
      "rewards/rejected": -3.2360089470358457,
      "step": 690
    },
    {
      "count/fg_chosen": 18.0,
      "count/fg_rejected": 4.599999904632568,
      "epoch": 0.7180407744582639,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.225204467773438,
      "fg_logps/policy_chosen": -15.224019050598145,
      "fg_logps/policy_rejected": -17.033180236816406,
      "fg_logps/reference_KL": -19.995820999145508,
      "fg_logps/reference_chosen": -13.913763046264648,
      "fg_logps/reference_rejected": -15.521989822387695,
      "fg_loss": 0.799098014831543,
      "fg_rewards/chosen_sum": -1.0094504356384277,
      "fg_rewards/rejected_sum": -0.5853248238563538,
      "grad_norm": 12.12648011854002,
      "kl": 0.0,
      "learning_rate": 1.5924657534246573e-07,
      "logps/chosen": -344.3683810763889,
      "logps/rejected": -375.64266183035716,
      "loss": 0.4812,
      "rewards/chosen": -0.6706111907958985,
      "rewards/margins": 2.2407723018101287,
      "rewards/rejected": -2.911383492606027,
      "step": 700
    },
    {
      "count/fg_chosen": 17.1200008392334,
      "count/fg_rejected": 3.095238208770752,
      "epoch": 0.7282984998076677,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.293447494506836,
      "fg_logps/policy_chosen": -13.494970321655273,
      "fg_logps/policy_rejected": -16.938003540039062,
      "fg_logps/reference_KL": -18.482240676879883,
      "fg_logps/reference_chosen": -13.644604682922363,
      "fg_logps/reference_rejected": -16.76604652404785,
      "fg_loss": 0.7007278800010681,
      "fg_rewards/chosen_sum": 0.5598518252372742,
      "fg_rewards/rejected_sum": -0.21593458950519562,
      "grad_norm": 14.686077759756456,
      "kl": 0.005865669343620539,
      "learning_rate": 1.5353881278538813e-07,
      "logps/chosen": -292.7752586570946,
      "logps/rejected": -331.10535519622096,
      "loss": 0.4337,
      "rewards/chosen": -0.09871104601267222,
      "rewards/margins": 2.7478330866635634,
      "rewards/rejected": -2.8465441326762355,
      "step": 710
    },
    {
      "count/fg_chosen": 16.35714340209961,
      "count/fg_rejected": 3.0,
      "epoch": 0.7385562251570714,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.972545623779297,
      "fg_logps/policy_chosen": -14.150968551635742,
      "fg_logps/policy_rejected": -22.872663497924805,
      "fg_logps/reference_KL": -20.97467041015625,
      "fg_logps/reference_chosen": -14.420245170593262,
      "fg_logps/reference_rejected": -23.34649658203125,
      "fg_loss": 0.6548932194709778,
      "fg_rewards/chosen_sum": 1.0662188529968262,
      "fg_rewards/rejected_sum": 0.059594761580228806,
      "grad_norm": 10.578828543316053,
      "kl": 0.0,
      "learning_rate": 1.478310502283105e-07,
      "logps/chosen": -315.2850294237013,
      "logps/rejected": -390.93107586596386,
      "loss": 0.4322,
      "rewards/chosen": -0.855069940740412,
      "rewards/margins": 1.7850939864527353,
      "rewards/rejected": -2.6401639271931474,
      "step": 720
    },
    {
      "count/fg_chosen": 18.545454025268555,
      "count/fg_rejected": 4.666666507720947,
      "epoch": 0.7488139505064751,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.168006896972656,
      "fg_logps/policy_chosen": -15.429848670959473,
      "fg_logps/policy_rejected": -20.59855079650879,
      "fg_logps/reference_KL": -21.434024810791016,
      "fg_logps/reference_chosen": -15.218304634094238,
      "fg_logps/reference_rejected": -20.059797286987305,
      "fg_loss": 0.7126922607421875,
      "fg_rewards/chosen_sum": 0.7979265451431274,
      "fg_rewards/rejected_sum": -0.11843083798885345,
      "grad_norm": 14.792127676836103,
      "kl": 0.06886587291955948,
      "learning_rate": 1.421232876712329e-07,
      "logps/chosen": -287.5471157962329,
      "logps/rejected": -396.7852460488506,
      "loss": 0.437,
      "rewards/chosen": 0.011234314474340987,
      "rewards/margins": 3.0592356586508855,
      "rewards/rejected": -3.0480013441765443,
      "step": 730
    },
    {
      "count/fg_chosen": 14.793103218078613,
      "count/fg_rejected": 4.148148059844971,
      "epoch": 0.759071675855879,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.16313934326172,
      "fg_logps/policy_chosen": -15.684663772583008,
      "fg_logps/policy_rejected": -19.98002052307129,
      "fg_logps/reference_KL": -20.154132843017578,
      "fg_logps/reference_chosen": -14.688092231750488,
      "fg_logps/reference_rejected": -18.67272186279297,
      "fg_loss": 0.8155136704444885,
      "fg_rewards/chosen_sum": -0.6768214106559753,
      "fg_rewards/rejected_sum": -0.5260854959487915,
      "grad_norm": 15.891406104752123,
      "kl": 0.0,
      "learning_rate": 1.3641552511415523e-07,
      "logps/chosen": -287.50723958333333,
      "logps/rejected": -408.88382352941176,
      "loss": 0.5193,
      "rewards/chosen": -0.7413622029622395,
      "rewards/margins": 2.1912876562978707,
      "rewards/rejected": -2.9326498592601102,
      "step": 740
    },
    {
      "count/fg_chosen": 17.5,
      "count/fg_rejected": 4.222222328186035,
      "epoch": 0.7693294012052827,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.54020881652832,
      "fg_logps/policy_chosen": -14.070555686950684,
      "fg_logps/policy_rejected": -19.403593063354492,
      "fg_logps/reference_KL": -17.662343978881836,
      "fg_logps/reference_chosen": -13.439438819885254,
      "fg_logps/reference_rejected": -18.562532424926758,
      "fg_loss": 0.7397133111953735,
      "fg_rewards/chosen_sum": -0.7740154266357422,
      "fg_rewards/rejected_sum": -0.5391371846199036,
      "grad_norm": 10.009694005800096,
      "kl": 0.0,
      "learning_rate": 1.3070776255707763e-07,
      "logps/chosen": -312.1899646577381,
      "logps/rejected": -437.6450966282895,
      "loss": 0.5177,
      "rewards/chosen": -1.1885592142740886,
      "rewards/margins": 1.7126708448978893,
      "rewards/rejected": -2.901230059171978,
      "step": 750
    },
    {
      "count/fg_chosen": 15.52173900604248,
      "count/fg_rejected": 3.25,
      "epoch": 0.7795871265546865,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.690628051757812,
      "fg_logps/policy_chosen": -15.920393943786621,
      "fg_logps/policy_rejected": -17.173805236816406,
      "fg_logps/reference_KL": -20.706315994262695,
      "fg_logps/reference_chosen": -15.248817443847656,
      "fg_logps/reference_rejected": -16.315380096435547,
      "fg_loss": 0.7596006393432617,
      "fg_rewards/chosen_sum": -0.31831929087638855,
      "fg_rewards/rejected_sum": -0.46696019172668457,
      "grad_norm": 11.001152687682147,
      "kl": 0.1602586805820465,
      "learning_rate": 1.25e-07,
      "logps/chosen": -315.33207514044943,
      "logps/rejected": -346.9461652728873,
      "loss": 0.5177,
      "rewards/chosen": -0.877791586886631,
      "rewards/margins": 1.9533982805897114,
      "rewards/rejected": -2.8311898674763425,
      "step": 760
    },
    {
      "count/fg_chosen": 13.954545021057129,
      "count/fg_rejected": 5.0625,
      "epoch": 0.7898448519040903,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.25006866455078,
      "fg_logps/policy_chosen": -15.683282852172852,
      "fg_logps/policy_rejected": -20.789840698242188,
      "fg_logps/reference_KL": -20.35251808166504,
      "fg_logps/reference_chosen": -15.138045310974121,
      "fg_logps/reference_rejected": -19.545236587524414,
      "fg_loss": 0.7877352833747864,
      "fg_rewards/chosen_sum": -0.0805804431438446,
      "fg_rewards/rejected_sum": -0.919762909412384,
      "grad_norm": 8.914441470208311,
      "kl": 0.0056018829345703125,
      "learning_rate": 1.1929223744292236e-07,
      "logps/chosen": -287.3155048076923,
      "logps/rejected": -411.3341749237805,
      "loss": 0.4771,
      "rewards/chosen": -0.5340637304844,
      "rewards/margins": 2.177349211052852,
      "rewards/rejected": -2.7114129415372523,
      "step": 770
    },
    {
      "count/fg_chosen": 16.94444465637207,
      "count/fg_rejected": 3.8125,
      "epoch": 0.800102577253494,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.90068244934082,
      "fg_logps/policy_chosen": -14.70145034790039,
      "fg_logps/policy_rejected": -18.811277389526367,
      "fg_logps/reference_KL": -19.578754425048828,
      "fg_logps/reference_chosen": -14.712576866149902,
      "fg_logps/reference_rejected": -18.912837982177734,
      "fg_loss": 0.7731603980064392,
      "fg_rewards/chosen_sum": 0.47520291805267334,
      "fg_rewards/rejected_sum": -0.34054040908813477,
      "grad_norm": 8.587072014013287,
      "kl": 0.09591908752918243,
      "learning_rate": 1.1358447488584474e-07,
      "logps/chosen": -269.6305419921875,
      "logps/rejected": -346.35361328125,
      "loss": 0.4703,
      "rewards/chosen": -0.6143415451049805,
      "rewards/margins": 1.9741443634033204,
      "rewards/rejected": -2.588485908508301,
      "step": 780
    },
    {
      "count/fg_chosen": 16.72222137451172,
      "count/fg_rejected": 4.916666507720947,
      "epoch": 0.8103603026028978,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.655256271362305,
      "fg_logps/policy_chosen": -14.51056957244873,
      "fg_logps/policy_rejected": -17.415449142456055,
      "fg_logps/reference_KL": -19.09449005126953,
      "fg_logps/reference_chosen": -14.248104095458984,
      "fg_logps/reference_rejected": -16.654375076293945,
      "fg_loss": 0.6147817969322205,
      "fg_rewards/chosen_sum": -0.14415740966796875,
      "fg_rewards/rejected_sum": -0.5222800970077515,
      "grad_norm": 15.457265468907643,
      "kl": 0.08856544643640518,
      "learning_rate": 1.0787671232876712e-07,
      "logps/chosen": -288.4888599537037,
      "logps/rejected": -336.5465783227848,
      "loss": 0.4963,
      "rewards/chosen": -0.8326729668511285,
      "rewards/margins": 1.523808464554627,
      "rewards/rejected": -2.3564814314057556,
      "step": 790
    },
    {
      "count/fg_chosen": 18.117647171020508,
      "count/fg_rejected": 5.333333492279053,
      "epoch": 0.8206180279523015,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.511550903320312,
      "fg_logps/policy_chosen": -13.713569641113281,
      "fg_logps/policy_rejected": -20.020904541015625,
      "fg_logps/reference_KL": -17.690017700195312,
      "fg_logps/reference_chosen": -13.43185806274414,
      "fg_logps/reference_rejected": -19.19187355041504,
      "fg_loss": 0.7455827593803406,
      "fg_rewards/chosen_sum": 0.051257286220788956,
      "fg_rewards/rejected_sum": -0.5338117480278015,
      "grad_norm": 15.017949800951369,
      "kl": 0.0,
      "learning_rate": 1.0216894977168949e-07,
      "logps/chosen": -324.4031684027778,
      "logps/rejected": -468.073828125,
      "loss": 0.5189,
      "rewards/chosen": -0.744783443874783,
      "rewards/margins": 1.6178517053997705,
      "rewards/rejected": -2.3626351492745536,
      "step": 800
    },
    {
      "epoch": 0.8206180279523015,
      "eval_count/fg_chosen": 16.01369857788086,
      "eval_count/fg_rejected": 4.0824174880981445,
      "eval_fg_kl": NaN,
      "eval_fg_logps/policy_KL": -21.125959396362305,
      "eval_fg_logps/policy_chosen": -15.278054237365723,
      "eval_fg_logps/policy_rejected": -19.65519905090332,
      "eval_fg_logps/reference_KL": -20.207021713256836,
      "eval_fg_logps/reference_chosen": -14.929487228393555,
      "eval_fg_logps/reference_rejected": -18.78682518005371,
      "eval_fg_loss": 0.736534595489502,
      "eval_fg_rewards/chosen_sum": 0.0694078877568245,
      "eval_fg_rewards/rejected_sum": -0.3622537851333618,
      "eval_kl": 0.008128926157951355,
      "eval_logps/chosen": -299.7121112089202,
      "eval_logps/rejected": -364.37436224489795,
      "eval_loss": 0.4815324544906616,
      "eval_rewards/chosen": -0.6601278815470951,
      "eval_rewards/margins": 1.9833717220703937,
      "eval_rewards/rejected": -2.6434996036174887,
      "eval_runtime": 808.1531,
      "eval_samples_per_second": 2.144,
      "eval_steps_per_second": 1.073,
      "step": 800
    },
    {
      "count/fg_chosen": 16.33333396911621,
      "count/fg_rejected": 2.7857143878936768,
      "epoch": 0.8308757533017054,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -23.438562393188477,
      "fg_logps/policy_chosen": -16.027084350585938,
      "fg_logps/policy_rejected": -18.6650390625,
      "fg_logps/reference_KL": -22.405675888061523,
      "fg_logps/reference_chosen": -15.913044929504395,
      "fg_logps/reference_rejected": -18.558101654052734,
      "fg_loss": 0.6452624201774597,
      "fg_rewards/chosen_sum": 0.40020328760147095,
      "fg_rewards/rejected_sum": -0.3264440894126892,
      "grad_norm": 7.011400584597267,
      "kl": 0.0,
      "learning_rate": 9.646118721461187e-08,
      "logps/chosen": -305.2624952936747,
      "logps/rejected": -375.5898944805195,
      "loss": 0.4673,
      "rewards/chosen": -0.7119015613234186,
      "rewards/margins": 2.1217946801792196,
      "rewards/rejected": -2.833696241502638,
      "step": 810
    },
    {
      "count/fg_chosen": 17.05555534362793,
      "count/fg_rejected": 4.8125,
      "epoch": 0.8411334786511091,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.090803146362305,
      "fg_logps/policy_chosen": -13.760422706604004,
      "fg_logps/policy_rejected": -19.657625198364258,
      "fg_logps/reference_KL": -19.27540397644043,
      "fg_logps/reference_chosen": -13.429532051086426,
      "fg_logps/reference_rejected": -18.345483779907227,
      "fg_loss": 0.7558559775352478,
      "fg_rewards/chosen_sum": 0.17663703858852386,
      "fg_rewards/rejected_sum": -0.7809099555015564,
      "grad_norm": 12.43361776512671,
      "kl": 0.0,
      "learning_rate": 9.075342465753424e-08,
      "logps/chosen": -316.12268350290697,
      "logps/rejected": -377.34243032094594,
      "loss": 0.4807,
      "rewards/chosen": -0.8466514764830123,
      "rewards/margins": 2.1515591211996314,
      "rewards/rejected": -2.9982105976826436,
      "step": 820
    },
    {
      "count/fg_chosen": 18.95833396911621,
      "count/fg_rejected": 3.5999999046325684,
      "epoch": 0.8513912040005129,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.093595504760742,
      "fg_logps/policy_chosen": -14.394129753112793,
      "fg_logps/policy_rejected": -17.22884178161621,
      "fg_logps/reference_KL": -19.071619033813477,
      "fg_logps/reference_chosen": -13.635478019714355,
      "fg_logps/reference_rejected": -16.50658416748047,
      "fg_loss": 0.7184759974479675,
      "fg_rewards/chosen_sum": 0.43327081203460693,
      "fg_rewards/rejected_sum": -0.22198212146759033,
      "grad_norm": 13.25277133030139,
      "kl": 0.04477081447839737,
      "learning_rate": 8.504566210045662e-08,
      "logps/chosen": -338.75426793981484,
      "logps/rejected": -406.7663419699367,
      "loss": 0.5598,
      "rewards/chosen": -0.7480648653006848,
      "rewards/margins": 1.1518268552417399,
      "rewards/rejected": -1.8998917205424248,
      "step": 830
    },
    {
      "count/fg_chosen": 16.428571701049805,
      "count/fg_rejected": 4.277777671813965,
      "epoch": 0.8616489293499167,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.675304412841797,
      "fg_logps/policy_chosen": -14.73193073272705,
      "fg_logps/policy_rejected": -17.31828498840332,
      "fg_logps/reference_KL": -19.20144271850586,
      "fg_logps/reference_chosen": -14.565011024475098,
      "fg_logps/reference_rejected": -17.02730941772461,
      "fg_loss": 0.7666863799095154,
      "fg_rewards/chosen_sum": 0.5958693623542786,
      "fg_rewards/rejected_sum": -0.01988927833735943,
      "grad_norm": 11.745424230577617,
      "kl": 0.0641721710562706,
      "learning_rate": 7.933789954337899e-08,
      "logps/chosen": -295.41467126623377,
      "logps/rejected": -367.2850856551205,
      "loss": 0.5065,
      "rewards/chosen": -0.7968547375171215,
      "rewards/margins": 1.916948473373315,
      "rewards/rejected": -2.7138032108904366,
      "step": 840
    },
    {
      "count/fg_chosen": 16.1200008392334,
      "count/fg_rejected": 4.5714287757873535,
      "epoch": 0.8719066546993204,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.93068504333496,
      "fg_logps/policy_chosen": -14.386673927307129,
      "fg_logps/policy_rejected": -18.88810157775879,
      "fg_logps/reference_KL": -19.163719177246094,
      "fg_logps/reference_chosen": -13.809516906738281,
      "fg_logps/reference_rejected": -17.88907241821289,
      "fg_loss": 0.7888970375061035,
      "fg_rewards/chosen_sum": -0.009096489287912846,
      "fg_rewards/rejected_sum": -0.5189210772514343,
      "grad_norm": 10.115452996310198,
      "kl": 0.027889441698789597,
      "learning_rate": 7.363013698630137e-08,
      "logps/chosen": -345.9682221283784,
      "logps/rejected": -356.40765806686045,
      "loss": 0.5075,
      "rewards/chosen": -0.9859014975058066,
      "rewards/margins": 1.6604730496984996,
      "rewards/rejected": -2.646374547204306,
      "step": 850
    },
    {
      "count/fg_chosen": 14.9375,
      "count/fg_rejected": 4.214285850524902,
      "epoch": 0.8821643800487242,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.199317932128906,
      "fg_logps/policy_chosen": -14.535076141357422,
      "fg_logps/policy_rejected": -18.46771240234375,
      "fg_logps/reference_KL": -20.52707862854004,
      "fg_logps/reference_chosen": -14.20512866973877,
      "fg_logps/reference_rejected": -17.470090866088867,
      "fg_loss": 0.8201218247413635,
      "fg_rewards/chosen_sum": -0.14333221316337585,
      "fg_rewards/rejected_sum": -0.3655402958393097,
      "grad_norm": 10.49340621720449,
      "kl": 0.0,
      "learning_rate": 6.792237442922374e-08,
      "logps/chosen": -259.49934050324674,
      "logps/rejected": -336.4366999246988,
      "loss": 0.4497,
      "rewards/chosen": -0.43501192563539975,
      "rewards/margins": 2.218185581188772,
      "rewards/rejected": -2.6531975068241715,
      "step": 860
    },
    {
      "count/fg_chosen": 15.117647171020508,
      "count/fg_rejected": 3.076923131942749,
      "epoch": 0.8924221053981279,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -22.28684425354004,
      "fg_logps/policy_chosen": -14.350448608398438,
      "fg_logps/policy_rejected": -20.323551177978516,
      "fg_logps/reference_KL": -21.138818740844727,
      "fg_logps/reference_chosen": -14.148969650268555,
      "fg_logps/reference_rejected": -19.21809959411621,
      "fg_loss": 0.7609926462173462,
      "fg_rewards/chosen_sum": 0.2540108561515808,
      "fg_rewards/rejected_sum": -0.4159541428089142,
      "grad_norm": 14.052655899212164,
      "kl": 0.0,
      "learning_rate": 6.221461187214611e-08,
      "logps/chosen": -313.48386548913044,
      "logps/rejected": -408.48944024725273,
      "loss": 0.4252,
      "rewards/chosen": -0.7578884622325068,
      "rewards/margins": 2.286565588355577,
      "rewards/rejected": -3.044454050588084,
      "step": 870
    },
    {
      "count/fg_chosen": 13.785714149475098,
      "count/fg_rejected": 4.363636493682861,
      "epoch": 0.9026798307475318,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -23.224103927612305,
      "fg_logps/policy_chosen": -17.700504302978516,
      "fg_logps/policy_rejected": -18.443458557128906,
      "fg_logps/reference_KL": -21.812795639038086,
      "fg_logps/reference_chosen": -16.529930114746094,
      "fg_logps/reference_rejected": -16.90056800842285,
      "fg_loss": 0.861905038356781,
      "fg_rewards/chosen_sum": -1.0540215969085693,
      "fg_rewards/rejected_sum": -0.6782628297805786,
      "grad_norm": 10.169312742642402,
      "kl": 0.0,
      "learning_rate": 5.650684931506849e-08,
      "logps/chosen": -271.11566840277777,
      "logps/rejected": -324.0809215198864,
      "loss": 0.4212,
      "rewards/chosen": -0.5925836563110352,
      "rewards/margins": 1.9108110774647105,
      "rewards/rejected": -2.5033947337757456,
      "step": 880
    },
    {
      "count/fg_chosen": 17.526315689086914,
      "count/fg_rejected": 6.0,
      "epoch": 0.9129375560969355,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.294906616210938,
      "fg_logps/policy_chosen": -13.935691833496094,
      "fg_logps/policy_rejected": -16.94110107421875,
      "fg_logps/reference_KL": -18.384138107299805,
      "fg_logps/reference_chosen": -13.67742919921875,
      "fg_logps/reference_rejected": -15.758844375610352,
      "fg_loss": 0.6794268488883972,
      "fg_rewards/chosen_sum": 0.0076361955143511295,
      "fg_rewards/rejected_sum": -0.4533938765525818,
      "grad_norm": 9.707055272049164,
      "kl": 0.0,
      "learning_rate": 5.0799086757990863e-08,
      "logps/chosen": -275.1984604779412,
      "logps/rejected": -400.91375,
      "loss": 0.4898,
      "rewards/chosen": -0.607273774988511,
      "rewards/margins": 2.4971288942823224,
      "rewards/rejected": -3.1044026692708333,
      "step": 890
    },
    {
      "count/fg_chosen": 16.18181800842285,
      "count/fg_rejected": 5.333333492279053,
      "epoch": 0.9231952814463392,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -19.024534225463867,
      "fg_logps/policy_chosen": -14.489638328552246,
      "fg_logps/policy_rejected": -17.007856369018555,
      "fg_logps/reference_KL": -18.624956130981445,
      "fg_logps/reference_chosen": -14.467068672180176,
      "fg_logps/reference_rejected": -16.36551284790039,
      "fg_loss": 0.7552723288536072,
      "fg_rewards/chosen_sum": 0.37475159764289856,
      "fg_rewards/rejected_sum": -0.4549465775489807,
      "grad_norm": 12.023866951403578,
      "kl": 0.0394529327750206,
      "learning_rate": 4.509132420091324e-08,
      "logps/chosen": -344.09639537183546,
      "logps/rejected": -425.6873553240741,
      "loss": 0.4817,
      "rewards/chosen": -0.8015480524376978,
      "rewards/margins": 2.245464465043977,
      "rewards/rejected": -3.0470125174816745,
      "step": 900
    },
    {
      "count/fg_chosen": 16.54166603088379,
      "count/fg_rejected": 4.318181991577148,
      "epoch": 0.9334530067957431,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.088489532470703,
      "fg_logps/policy_chosen": -15.405600547790527,
      "fg_logps/policy_rejected": -16.747190475463867,
      "fg_logps/reference_KL": -19.137371063232422,
      "fg_logps/reference_chosen": -14.657370567321777,
      "fg_logps/reference_rejected": -15.770890235900879,
      "fg_loss": 0.6983441114425659,
      "fg_rewards/chosen_sum": -0.3915860652923584,
      "fg_rewards/rejected_sum": -0.3197772204875946,
      "grad_norm": 11.088479553468312,
      "kl": 0.0,
      "learning_rate": 3.938356164383561e-08,
      "logps/chosen": -306.00605715981015,
      "logps/rejected": -396.05049189814815,
      "loss": 0.4692,
      "rewards/chosen": -0.3026618957519531,
      "rewards/margins": 2.5836239567509405,
      "rewards/rejected": -2.8862858525028936,
      "step": 910
    },
    {
      "count/fg_chosen": 17.454545974731445,
      "count/fg_rejected": 3.4000000953674316,
      "epoch": 0.9437107321451468,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -24.18083953857422,
      "fg_logps/policy_chosen": -15.12738037109375,
      "fg_logps/policy_rejected": -22.837459564208984,
      "fg_logps/reference_KL": -22.198238372802734,
      "fg_logps/reference_chosen": -13.988216400146484,
      "fg_logps/reference_rejected": -20.653087615966797,
      "fg_loss": 0.7821993231773376,
      "fg_rewards/chosen_sum": -1.5599383115768433,
      "fg_rewards/rejected_sum": -0.6531942486763,
      "grad_norm": 13.573243756082668,
      "kl": 0.0,
      "learning_rate": 3.367579908675799e-08,
      "logps/chosen": -334.25313527960526,
      "logps/rejected": -345.7750651041667,
      "loss": 0.5174,
      "rewards/chosen": -1.169078224583676,
      "rewards/margins": 1.8482150696871575,
      "rewards/rejected": -3.0172932942708335,
      "step": 920
    },
    {
      "count/fg_chosen": 14.75,
      "count/fg_rejected": 3.5,
      "epoch": 0.9539684574945506,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.988378524780273,
      "fg_logps/policy_chosen": -15.057655334472656,
      "fg_logps/policy_rejected": -21.019062042236328,
      "fg_logps/reference_KL": -20.459566116333008,
      "fg_logps/reference_chosen": -14.091443061828613,
      "fg_logps/reference_rejected": -20.186948776245117,
      "fg_loss": 0.7185892462730408,
      "fg_rewards/chosen_sum": -0.44822633266448975,
      "fg_rewards/rejected_sum": -0.26188215613365173,
      "grad_norm": 16.393627175860047,
      "kl": 0.0,
      "learning_rate": 2.796803652968036e-08,
      "logps/chosen": -283.63548677884614,
      "logps/rejected": -336.15740131578946,
      "loss": 0.4039,
      "rewards/chosen": -0.3187708928034856,
      "rewards/margins": 2.802559930592896,
      "rewards/rejected": -3.1213308233963817,
      "step": 930
    },
    {
      "count/fg_chosen": 18.83333396911621,
      "count/fg_rejected": 5.933333396911621,
      "epoch": 0.9642261828439543,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -20.48985481262207,
      "fg_logps/policy_chosen": -15.466116905212402,
      "fg_logps/policy_rejected": -21.363821029663086,
      "fg_logps/reference_KL": -19.345491409301758,
      "fg_logps/reference_chosen": -14.387165069580078,
      "fg_logps/reference_rejected": -19.8946533203125,
      "fg_loss": 0.7254918217658997,
      "fg_rewards/chosen_sum": -1.1295708417892456,
      "fg_rewards/rejected_sum": -0.7322808504104614,
      "grad_norm": 12.073302615788874,
      "kl": 0.0,
      "learning_rate": 2.2260273972602736e-08,
      "logps/chosen": -311.50599500868054,
      "logps/rejected": -380.01899857954544,
      "loss": 0.4599,
      "rewards/chosen": -0.8199851247999403,
      "rewards/margins": 2.1154369055622757,
      "rewards/rejected": -2.935422030362216,
      "step": 940
    },
    {
      "count/fg_chosen": 15.636363983154297,
      "count/fg_rejected": 3.4000000953674316,
      "epoch": 0.9744839081933582,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -18.018638610839844,
      "fg_logps/policy_chosen": -14.090137481689453,
      "fg_logps/policy_rejected": -13.102763175964355,
      "fg_logps/reference_KL": -16.83769416809082,
      "fg_logps/reference_chosen": -13.336016654968262,
      "fg_logps/reference_rejected": -12.24677848815918,
      "fg_loss": 0.8313923478126526,
      "fg_rewards/chosen_sum": -0.822435736656189,
      "fg_rewards/rejected_sum": -0.4055534303188324,
      "grad_norm": 8.942456920046205,
      "kl": 0.027703475207090378,
      "learning_rate": 1.6552511415525114e-08,
      "logps/chosen": -319.5060622970779,
      "logps/rejected": -353.5703830948795,
      "loss": 0.4562,
      "rewards/chosen": -1.2102871188869724,
      "rewards/margins": 1.2892790167493229,
      "rewards/rejected": -2.4995661356362953,
      "step": 950
    },
    {
      "count/fg_chosen": 14.909090995788574,
      "count/fg_rejected": 3.6315789222717285,
      "epoch": 0.9847416335427619,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -21.546527862548828,
      "fg_logps/policy_chosen": -15.540050506591797,
      "fg_logps/policy_rejected": -20.36016273498535,
      "fg_logps/reference_KL": -20.43901252746582,
      "fg_logps/reference_chosen": -14.649651527404785,
      "fg_logps/reference_rejected": -18.264400482177734,
      "fg_loss": 0.7249688506126404,
      "fg_rewards/chosen_sum": -0.600669801235199,
      "fg_rewards/rejected_sum": -0.7846862077713013,
      "grad_norm": 13.343937741670935,
      "kl": 0.0,
      "learning_rate": 1.0844748858447487e-08,
      "logps/chosen": -326.93775699013156,
      "logps/rejected": -363.2979445684524,
      "loss": 0.5157,
      "rewards/chosen": -1.1762562801963405,
      "rewards/margins": 1.8776036121492694,
      "rewards/rejected": -3.05385989234561,
      "step": 960
    },
    {
      "count/fg_chosen": 14.61111068725586,
      "count/fg_rejected": 3.3333332538604736,
      "epoch": 0.9949993588921656,
      "fg_kl": NaN,
      "fg_logps/policy_KL": -24.816984176635742,
      "fg_logps/policy_chosen": -17.12736701965332,
      "fg_logps/policy_rejected": -22.242828369140625,
      "fg_logps/reference_KL": -22.654382705688477,
      "fg_logps/reference_chosen": -15.033406257629395,
      "fg_logps/reference_rejected": -21.171716690063477,
      "fg_loss": 0.7910918593406677,
      "fg_rewards/chosen_sum": -1.5589678287506104,
      "fg_rewards/rejected_sum": -0.5098855495452881,
      "grad_norm": 12.686683889493278,
      "kl": 0.0,
      "learning_rate": 5.136986301369862e-09,
      "logps/chosen": -375.69940350506755,
      "logps/rejected": -364.8770212572674,
      "loss": 0.481,
      "rewards/chosen": -1.2619872222075592,
      "rewards/margins": 1.2282116672364813,
      "rewards/rejected": -2.4901988894440406,
      "step": 970
    },
    {
      "epoch": 0.9991024490319271,
      "step": 974,
      "total_flos": 0.0,
      "train_loss": 0.5029905087159644,
      "train_runtime": 13638.2416,
      "train_samples_per_second": 1.144,
      "train_steps_per_second": 0.071
    }
  ],
  "logging_steps": 10,
  "max_steps": 974,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}