{
  "best_metric": 1.9277071952819824,
  "best_model_checkpoint": "./Zephyr/08-03-24-Weni-WeniGPT-2.10.1-Zephyr-7B-DPO-prompt-binarized-GPTQ_DPO tests with binarized dataset GPTQ-2_max_steps-896_batch_16_2024-03-08_ppid_7990/checkpoint-100",
  "epoch": 5.333333333333333,
  "eval_steps": 100,
  "global_step": 600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.18,
      "grad_norm": 119.0400161743164,
      "learning_rate": 3.111111111111111e-05,
      "logits/chosen": -2.6437883377075195,
      "logits/rejected": -2.640676498413086,
      "logps/chosen": -346.0354309082031,
      "logps/rejected": -315.1640319824219,
      "loss": 0.6706,
      "rewards/accuracies": 0.30000001192092896,
      "rewards/chosen": 0.2039356678724289,
      "rewards/margins": 0.1839628666639328,
      "rewards/rejected": 0.019972801208496094,
      "step": 20
    },
    {
      "epoch": 0.36,
      "grad_norm": 55.844058990478516,
      "learning_rate": 7.555555555555556e-05,
      "logits/chosen": -2.7053043842315674,
      "logits/rejected": -2.706561326980591,
      "logps/chosen": -348.75,
      "logps/rejected": -336.07830810546875,
      "loss": 0.7255,
      "rewards/accuracies": 0.5062500238418579,
      "rewards/chosen": -0.3999950885772705,
      "rewards/margins": 0.9534912109375,
      "rewards/rejected": -1.3534862995147705,
      "step": 40
    },
    {
      "epoch": 0.53,
      "grad_norm": 156.59803771972656,
      "learning_rate": 0.00011777777777777779,
      "logits/chosen": -2.644763946533203,
      "logits/rejected": -2.6682467460632324,
      "logps/chosen": -352.20233154296875,
      "logps/rejected": -343.33941650390625,
      "loss": 0.8382,
      "rewards/accuracies": 0.5718749761581421,
      "rewards/chosen": 2.873349189758301,
      "rewards/margins": 1.9845993518829346,
      "rewards/rejected": 0.8887494802474976,
      "step": 60
    },
    {
      "epoch": 0.71,
      "grad_norm": 113.42001342773438,
      "learning_rate": 0.00016222222222222224,
      "logits/chosen": -2.652919292449951,
      "logits/rejected": -2.663282871246338,
      "logps/chosen": -350.60443115234375,
      "logps/rejected": -333.491455078125,
      "loss": 1.6409,
      "rewards/accuracies": 0.512499988079071,
      "rewards/chosen": 4.428155899047852,
      "rewards/margins": 2.1724541187286377,
      "rewards/rejected": 2.255702018737793,
      "step": 80
    },
    {
      "epoch": 0.89,
      "grad_norm": 77.21934509277344,
      "learning_rate": 0.0001992555831265509,
      "logits/chosen": -2.6474595069885254,
      "logits/rejected": -2.6597702503204346,
      "logps/chosen": -318.25653076171875,
      "logps/rejected": -302.04241943359375,
      "loss": 1.9207,
      "rewards/accuracies": 0.4937500059604645,
      "rewards/chosen": 14.573875427246094,
      "rewards/margins": 3.523595094680786,
      "rewards/rejected": 11.050280570983887,
      "step": 100
    },
    {
      "epoch": 0.89,
      "eval_logits/chosen": -2.7368459701538086,
      "eval_logits/rejected": -2.746166467666626,
      "eval_logps/chosen": -334.51287841796875,
      "eval_logps/rejected": -329.4024658203125,
      "eval_loss": 1.9277071952819824,
      "eval_rewards/accuracies": 0.5350000262260437,
      "eval_rewards/chosen": 18.78862190246582,
      "eval_rewards/margins": 5.63677978515625,
      "eval_rewards/rejected": 13.151841163635254,
      "eval_runtime": 94.4139,
      "eval_samples_per_second": 2.118,
      "eval_steps_per_second": 0.53,
      "step": 100
    },
    {
      "epoch": 1.07,
      "grad_norm": 83.98187255859375,
      "learning_rate": 0.00019429280397022334,
      "logits/chosen": -2.639448642730713,
      "logits/rejected": -2.6476964950561523,
      "logps/chosen": -316.87982177734375,
      "logps/rejected": -308.38568115234375,
      "loss": 1.7601,
      "rewards/accuracies": 0.5562499761581421,
      "rewards/chosen": 27.463403701782227,
      "rewards/margins": 10.681607246398926,
      "rewards/rejected": 16.781795501708984,
      "step": 120
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.0,
      "learning_rate": 0.0001893300248138958,
      "logits/chosen": -2.6575229167938232,
      "logits/rejected": -2.6546902656555176,
      "logps/chosen": -350.1793212890625,
      "logps/rejected": -348.28985595703125,
      "loss": 1.8357,
      "rewards/accuracies": 0.6156250238418579,
      "rewards/chosen": -23.570585250854492,
      "rewards/margins": 16.757661819458008,
      "rewards/rejected": -40.32825469970703,
      "step": 140
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0552074909210205,
      "learning_rate": 0.00018436724565756824,
      "logits/chosen": -2.690781831741333,
      "logits/rejected": -2.691371440887451,
      "logps/chosen": -403.81378173828125,
      "logps/rejected": -407.43475341796875,
      "loss": 1.4241,
      "rewards/accuracies": 0.6343749761581421,
      "rewards/chosen": -46.7193489074707,
      "rewards/margins": 20.419803619384766,
      "rewards/rejected": -67.13915252685547,
      "step": 160
    },
    {
      "epoch": 1.6,
      "grad_norm": 88.9915771484375,
      "learning_rate": 0.0001794044665012407,
      "logits/chosen": -2.6750125885009766,
      "logits/rejected": -2.682309627532959,
      "logps/chosen": -399.62066650390625,
      "logps/rejected": -409.8529357910156,
      "loss": 1.9108,
      "rewards/accuracies": 0.621874988079071,
      "rewards/chosen": -49.57743453979492,
      "rewards/margins": 20.643938064575195,
      "rewards/rejected": -70.22136688232422,
      "step": 180
    },
    {
      "epoch": 1.78,
      "grad_norm": 138.35623168945312,
      "learning_rate": 0.00017444168734491314,
      "logits/chosen": -2.7050137519836426,
      "logits/rejected": -2.712310552597046,
      "logps/chosen": -397.1112365722656,
      "logps/rejected": -406.0185852050781,
      "loss": 2.0578,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -53.7899284362793,
      "rewards/margins": 17.770038604736328,
      "rewards/rejected": -71.55997467041016,
      "step": 200
    },
    {
      "epoch": 1.78,
      "eval_logits/chosen": -2.6775686740875244,
      "eval_logits/rejected": -2.6850674152374268,
      "eval_logps/chosen": -419.0586242675781,
      "eval_logps/rejected": -420.5866394042969,
      "eval_loss": 2.5765089988708496,
      "eval_rewards/accuracies": 0.550000011920929,
      "eval_rewards/chosen": -65.75714874267578,
      "eval_rewards/margins": 12.275188446044922,
      "eval_rewards/rejected": -78.03234100341797,
      "eval_runtime": 94.1202,
      "eval_samples_per_second": 2.125,
      "eval_steps_per_second": 0.531,
      "step": 200
    },
    {
      "epoch": 1.96,
      "grad_norm": 30.972963333129883,
      "learning_rate": 0.0001694789081885856,
      "logits/chosen": -2.631683349609375,
      "logits/rejected": -2.6248221397399902,
      "logps/chosen": -404.0619201660156,
      "logps/rejected": -399.0223083496094,
      "loss": 1.8819,
      "rewards/accuracies": 0.596875011920929,
      "rewards/chosen": -62.639190673828125,
      "rewards/margins": 19.827577590942383,
      "rewards/rejected": -82.46675872802734,
      "step": 220
    },
    {
      "epoch": 2.13,
      "grad_norm": 65.85337829589844,
      "learning_rate": 0.00016451612903225807,
      "logits/chosen": -2.711599826812744,
      "logits/rejected": -2.7432861328125,
      "logps/chosen": -392.7776794433594,
      "logps/rejected": -422.9435119628906,
      "loss": 1.381,
      "rewards/accuracies": 0.643750011920929,
      "rewards/chosen": -58.72906494140625,
      "rewards/margins": 29.807327270507812,
      "rewards/rejected": -88.536376953125,
      "step": 240
    },
    {
      "epoch": 2.31,
      "grad_norm": 86.84841918945312,
      "learning_rate": 0.00015955334987593052,
      "logits/chosen": -2.8453807830810547,
      "logits/rejected": -2.849914073944092,
      "logps/chosen": -400.3071594238281,
      "logps/rejected": -420.60968017578125,
      "loss": 0.9731,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -69.37321472167969,
      "rewards/margins": 32.82529830932617,
      "rewards/rejected": -102.19852447509766,
      "step": 260
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.010256004519760609,
      "learning_rate": 0.000154590570719603,
      "logits/chosen": -2.7864909172058105,
      "logits/rejected": -2.812525749206543,
      "logps/chosen": -376.9371643066406,
      "logps/rejected": -395.2316589355469,
      "loss": 0.8817,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -35.111305236816406,
      "rewards/margins": 33.674293518066406,
      "rewards/rejected": -68.78559875488281,
      "step": 280
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.26802292466163635,
      "learning_rate": 0.00014962779156327545,
      "logits/chosen": -2.7736704349517822,
      "logits/rejected": -2.792346477508545,
      "logps/chosen": -372.4452209472656,
      "logps/rejected": -392.68878173828125,
      "loss": 0.3389,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": -25.840368270874023,
      "rewards/margins": 39.69981002807617,
      "rewards/rejected": -65.54017639160156,
      "step": 300
    },
    {
      "epoch": 2.67,
      "eval_logits/chosen": -2.8219432830810547,
      "eval_logits/rejected": -2.8389434814453125,
      "eval_logps/chosen": -374.2080078125,
      "eval_logps/rejected": -374.9266052246094,
      "eval_loss": 3.5374395847320557,
      "eval_rewards/accuracies": 0.5400000214576721,
      "eval_rewards/chosen": -20.90648651123047,
      "eval_rewards/margins": 11.465815544128418,
      "eval_rewards/rejected": -32.37229919433594,
      "eval_runtime": 94.0956,
      "eval_samples_per_second": 2.125,
      "eval_steps_per_second": 0.531,
      "step": 300
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.30229315161705017,
      "learning_rate": 0.0001446650124069479,
      "logits/chosen": -2.718658447265625,
      "logits/rejected": -2.7346837520599365,
      "logps/chosen": -379.4544677734375,
      "logps/rejected": -398.12213134765625,
      "loss": 1.0938,
      "rewards/accuracies": 0.6781250238418579,
      "rewards/chosen": -28.148609161376953,
      "rewards/margins": 33.832645416259766,
      "rewards/rejected": -61.98125076293945,
      "step": 320
    },
    {
      "epoch": 3.02,
      "grad_norm": 43.86916732788086,
      "learning_rate": 0.00013970223325062035,
      "logits/chosen": -2.66723370552063,
      "logits/rejected": -2.6703853607177734,
      "logps/chosen": -419.7493591308594,
      "logps/rejected": -418.89227294921875,
      "loss": 0.9074,
      "rewards/accuracies": 0.7093750238418579,
      "rewards/chosen": -54.372032165527344,
      "rewards/margins": 28.145233154296875,
      "rewards/rejected": -82.51727294921875,
      "step": 340
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.2495471239089966,
      "learning_rate": 0.0001347394540942928,
      "logits/chosen": -2.741353988647461,
      "logits/rejected": -2.7415261268615723,
      "logps/chosen": -394.04681396484375,
      "logps/rejected": -423.58538818359375,
      "loss": 0.3777,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -41.517425537109375,
      "rewards/margins": 49.255271911621094,
      "rewards/rejected": -90.77268981933594,
      "step": 360
    },
    {
      "epoch": 3.38,
      "grad_norm": 7.586065292358398,
      "learning_rate": 0.00012977667493796526,
      "logits/chosen": -2.683011054992676,
      "logits/rejected": -2.697779893875122,
      "logps/chosen": -374.07745361328125,
      "logps/rejected": -395.69390869140625,
      "loss": 0.2784,
      "rewards/accuracies": 0.731249988079071,
      "rewards/chosen": -44.96805953979492,
      "rewards/margins": 42.65815734863281,
      "rewards/rejected": -87.626220703125,
      "step": 380
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.3450467586517334,
      "learning_rate": 0.0001248138957816377,
      "logits/chosen": -2.78257417678833,
      "logits/rejected": -2.797196865081787,
      "logps/chosen": -436.71185302734375,
      "logps/rejected": -443.575927734375,
      "loss": 0.6031,
      "rewards/accuracies": 0.7593749761581421,
      "rewards/chosen": -59.74363327026367,
      "rewards/margins": 44.275550842285156,
      "rewards/rejected": -104.01918029785156,
      "step": 400
    },
    {
      "epoch": 3.56,
      "eval_logits/chosen": -2.8664298057556152,
      "eval_logits/rejected": -2.879199266433716,
      "eval_logps/chosen": -433.5849914550781,
      "eval_logps/rejected": -436.34722900390625,
      "eval_loss": 2.886720657348633,
      "eval_rewards/accuracies": 0.5899999737739563,
      "eval_rewards/chosen": -80.2834701538086,
      "eval_rewards/margins": 13.509483337402344,
      "eval_rewards/rejected": -93.7929458618164,
      "eval_runtime": 94.1029,
      "eval_samples_per_second": 2.125,
      "eval_steps_per_second": 0.531,
      "step": 400
    },
    {
      "epoch": 3.73,
      "grad_norm": 54.701847076416016,
      "learning_rate": 0.00011985111662531019,
      "logits/chosen": -2.803107738494873,
      "logits/rejected": -2.8200833797454834,
      "logps/chosen": -420.0624084472656,
      "logps/rejected": -450.6493225097656,
      "loss": 0.4575,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -65.81521606445312,
      "rewards/margins": 47.31630325317383,
      "rewards/rejected": -113.13151550292969,
      "step": 420
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.8251222372055054,
      "learning_rate": 0.00011488833746898264,
      "logits/chosen": -2.81154203414917,
      "logits/rejected": -2.823908805847168,
      "logps/chosen": -400.0002746582031,
      "logps/rejected": -430.2955017089844,
      "loss": 0.4602,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -70.34940338134766,
      "rewards/margins": 37.30230712890625,
      "rewards/rejected": -107.6517105102539,
      "step": 440
    },
    {
      "epoch": 4.09,
      "grad_norm": 0.010179584845900536,
      "learning_rate": 0.0001099255583126551,
      "logits/chosen": -2.829603672027588,
      "logits/rejected": -2.845191478729248,
      "logps/chosen": -398.1234436035156,
      "logps/rejected": -432.2434997558594,
      "loss": 0.2573,
      "rewards/accuracies": 0.715624988079071,
      "rewards/chosen": -60.11717987060547,
      "rewards/margins": 44.31230926513672,
      "rewards/rejected": -104.42948150634766,
      "step": 460
    },
    {
      "epoch": 4.27,
      "grad_norm": 0.6254100203514099,
      "learning_rate": 0.00010496277915632755,
      "logits/chosen": -2.826686143875122,
      "logits/rejected": -2.8347971439361572,
      "logps/chosen": -401.7522888183594,
      "logps/rejected": -440.53424072265625,
      "loss": 0.3482,
      "rewards/accuracies": 0.721875011920929,
      "rewards/chosen": -58.31787109375,
      "rewards/margins": 41.26551818847656,
      "rewards/rejected": -99.58338928222656,
      "step": 480
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.0359001656468081e-08,
      "learning_rate": 0.0001,
      "logits/chosen": -2.858133554458618,
      "logits/rejected": -2.8563179969787598,
      "logps/chosen": -429.6060485839844,
      "logps/rejected": -461.37158203125,
      "loss": 0.2601,
      "rewards/accuracies": 0.778124988079071,
      "rewards/chosen": -60.158851623535156,
      "rewards/margins": 52.268470764160156,
      "rewards/rejected": -112.42732238769531,
      "step": 500
    },
    {
      "epoch": 4.44,
      "eval_logits/chosen": -2.846097946166992,
      "eval_logits/rejected": -2.8565187454223633,
      "eval_logps/chosen": -423.2083740234375,
      "eval_logps/rejected": -426.1557922363281,
      "eval_loss": 2.561472177505493,
      "eval_rewards/accuracies": 0.5799999833106995,
      "eval_rewards/chosen": -69.90685272216797,
      "eval_rewards/margins": 13.694626808166504,
      "eval_rewards/rejected": -83.60147094726562,
      "eval_runtime": 94.096,
      "eval_samples_per_second": 2.125,
      "eval_steps_per_second": 0.531,
      "step": 500
    },
    {
      "epoch": 4.62,
      "grad_norm": 0.0,
      "learning_rate": 9.503722084367246e-05,
      "logits/chosen": -2.799070358276367,
      "logits/rejected": -2.8123724460601807,
      "logps/chosen": -379.1603088378906,
      "logps/rejected": -415.482666015625,
      "loss": 0.2794,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": -56.228111267089844,
      "rewards/margins": 44.66722869873047,
      "rewards/rejected": -100.89533996582031,
      "step": 520
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.0027911989018321037,
      "learning_rate": 9.007444168734491e-05,
      "logits/chosen": -2.7665505409240723,
      "logits/rejected": -2.745327949523926,
      "logps/chosen": -395.28546142578125,
      "logps/rejected": -414.8994140625,
      "loss": 0.2362,
      "rewards/accuracies": 0.746874988079071,
      "rewards/chosen": -53.981964111328125,
      "rewards/margins": 49.71052932739258,
      "rewards/rejected": -103.69248962402344,
      "step": 540
    },
    {
      "epoch": 4.98,
      "grad_norm": 0.0,
      "learning_rate": 8.511166253101737e-05,
      "logits/chosen": -2.7394261360168457,
      "logits/rejected": -2.7414393424987793,
      "logps/chosen": -405.83502197265625,
      "logps/rejected": -441.4840393066406,
      "loss": 0.179,
      "rewards/accuracies": 0.746874988079071,
      "rewards/chosen": -58.963035583496094,
      "rewards/margins": 53.1982421875,
      "rewards/rejected": -112.1612777709961,
      "step": 560
    },
    {
      "epoch": 5.16,
      "grad_norm": 7.180324956834738e-08,
      "learning_rate": 8.014888337468984e-05,
      "logits/chosen": -2.72349214553833,
      "logits/rejected": -2.735581159591675,
      "logps/chosen": -380.3550720214844,
      "logps/rejected": -412.8421936035156,
      "loss": 0.2018,
      "rewards/accuracies": 0.7406250238418579,
      "rewards/chosen": -46.29833984375,
      "rewards/margins": 56.88873291015625,
      "rewards/rejected": -103.18707275390625,
      "step": 580
    },
    {
      "epoch": 5.33,
      "grad_norm": 0.0,
      "learning_rate": 7.518610421836229e-05,
      "logits/chosen": -2.8052825927734375,
      "logits/rejected": -2.8020119667053223,
      "logps/chosen": -397.83770751953125,
      "logps/rejected": -443.18511962890625,
      "loss": 0.1889,
      "rewards/accuracies": 0.753125011920929,
      "rewards/chosen": -45.621360778808594,
      "rewards/margins": 56.839752197265625,
      "rewards/rejected": -102.46110534667969,
      "step": 600
    },
    {
      "epoch": 5.33,
      "eval_logits/chosen": -2.81626033782959,
      "eval_logits/rejected": -2.824737548828125,
      "eval_logps/chosen": -413.4210510253906,
      "eval_logps/rejected": -419.48870849609375,
      "eval_loss": 3.0569708347320557,
      "eval_rewards/accuracies": 0.5950000286102295,
      "eval_rewards/chosen": -60.11954116821289,
      "eval_rewards/margins": 16.814842224121094,
      "eval_rewards/rejected": -76.93437194824219,
      "eval_runtime": 94.0927,
      "eval_samples_per_second": 2.126,
      "eval_steps_per_second": 0.531,
      "step": 600
    }
  ],
  "logging_steps": 20,
  "max_steps": 896,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}