{
  "best_metric": 0.0067643518559634686,
  "best_model_checkpoint": "./Zephyr/28-03-24-Weni-WeniGPT-QA-Zephyr-7B-5.0.0-KTO_WeniGPT Experiment using KTO trainer with no collator, Mixstral model and random system prompt.-2_max_steps-786_batch_32_2024-03-28_ppid_9/checkpoint-600",
  "epoch": 4.562737642585551,
  "eval_steps": 50,
  "global_step": 600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.15,
      "grad_norm": 3.2390341758728027,
      "kl": 1.4365953207015991,
      "learning_rate": 0.0001666666666666667,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 1.2323,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 20
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6226775050163269,
      "kl": 0.1346512734889984,
      "learning_rate": 0.00019580052493438322,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.1437,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 40
    },
    {
      "epoch": 0.38,
      "eval_kl": 0.0960722267627716,
      "eval_logps/chosen": -158.278564453125,
      "eval_logps/rejected": -400.1436767578125,
      "eval_loss": 0.028233768418431282,
      "eval_rewards/chosen": 5.2842488288879395,
      "eval_rewards/margins": 25.394336700439453,
      "eval_rewards/rejected": -20.11008644104004,
      "eval_runtime": 214.1148,
      "eval_samples_per_second": 2.232,
      "eval_steps_per_second": 0.56,
      "step": 50
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9644193649291992,
      "kl": 0.026479745283722878,
      "learning_rate": 0.0001905511811023622,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.1405,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 60
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28626948595046997,
      "kl": 0.018679404631257057,
      "learning_rate": 0.00018530183727034121,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.1041,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 80
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.0979766920208931,
      "kl": 0.0,
      "learning_rate": 0.00018005249343832022,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0615,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 100
    },
    {
      "epoch": 0.76,
      "eval_kl": 0.5886130928993225,
      "eval_logps/chosen": -153.61863708496094,
      "eval_logps/rejected": -383.54296875,
      "eval_loss": 0.022206587716937065,
      "eval_rewards/chosen": 5.750241756439209,
      "eval_rewards/margins": 24.20025062561035,
      "eval_rewards/rejected": -18.450010299682617,
      "eval_runtime": 213.9917,
      "eval_samples_per_second": 2.234,
      "eval_steps_per_second": 0.561,
      "step": 100
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3924414813518524,
      "kl": 0.0,
      "learning_rate": 0.00017480314960629923,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0396,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 120
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.0917697548866272,
      "kl": 0.2697001099586487,
      "learning_rate": 0.00016955380577427821,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0346,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 140
    },
    {
      "epoch": 1.14,
      "eval_kl": 0.3035987615585327,
      "eval_logps/chosen": -162.28248596191406,
      "eval_logps/rejected": -612.7335205078125,
      "eval_loss": 0.03983871266245842,
      "eval_rewards/chosen": 4.883855819702148,
      "eval_rewards/margins": 46.25292205810547,
      "eval_rewards/rejected": -41.36906814575195,
      "eval_runtime": 214.0091,
      "eval_samples_per_second": 2.234,
      "eval_steps_per_second": 0.561,
      "step": 150
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.16112379729747772,
      "kl": 0.2598298192024231,
      "learning_rate": 0.00016430446194225722,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0437,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 160
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.04856234788894653,
      "kl": 0.09709737449884415,
      "learning_rate": 0.00015958005249343832,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0288,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 180
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.5220792889595032,
      "kl": 0.055810533463954926,
      "learning_rate": 0.00015433070866141733,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0563,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 200
    },
    {
      "epoch": 1.52,
      "eval_kl": 0.15839433670043945,
      "eval_logps/chosen": -149.37533569335938,
      "eval_logps/rejected": -463.8907165527344,
      "eval_loss": 0.021165071055293083,
      "eval_rewards/chosen": 6.174570083618164,
      "eval_rewards/margins": 32.65935134887695,
      "eval_rewards/rejected": -26.48478126525879,
      "eval_runtime": 214.0155,
      "eval_samples_per_second": 2.233,
      "eval_steps_per_second": 0.561,
      "step": 200
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.05050266534090042,
      "kl": 0.0,
      "learning_rate": 0.00014908136482939634,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0298,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 220
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.530806064605713,
      "kl": 0.0,
      "learning_rate": 0.00014383202099737535,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0533,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 240
    },
    {
      "epoch": 1.9,
      "eval_kl": 0.4594874680042267,
      "eval_logps/chosen": -149.20762634277344,
      "eval_logps/rejected": -489.6085205078125,
      "eval_loss": 0.013410959392786026,
      "eval_rewards/chosen": 6.191342353820801,
      "eval_rewards/margins": 35.24790573120117,
      "eval_rewards/rejected": -29.056562423706055,
      "eval_runtime": 214.0202,
      "eval_samples_per_second": 2.233,
      "eval_steps_per_second": 0.561,
      "step": 250
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.017549920827150345,
      "kl": 0.0,
      "learning_rate": 0.00013858267716535433,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0461,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 260
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.021922318264842033,
      "kl": 0.4756055772304535,
      "learning_rate": 0.00013333333333333334,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.013,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 280
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.02235027775168419,
      "kl": 0.16024529933929443,
      "learning_rate": 0.00012808398950131235,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0076,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 300
    },
    {
      "epoch": 2.28,
      "eval_kl": 0.6192654967308044,
      "eval_logps/chosen": -147.9675750732422,
      "eval_logps/rejected": -453.9040222167969,
      "eval_loss": 0.016069181263446808,
      "eval_rewards/chosen": 6.315345764160156,
      "eval_rewards/margins": 31.80146026611328,
      "eval_rewards/rejected": -25.486116409301758,
      "eval_runtime": 214.0005,
      "eval_samples_per_second": 2.234,
      "eval_steps_per_second": 0.561,
      "step": 300
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.6879442930221558,
      "kl": 0.0,
      "learning_rate": 0.00012283464566929136,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0068,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 320
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.011550226248800755,
      "kl": 0.0,
      "learning_rate": 0.00011758530183727034,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.011,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 340
    },
    {
      "epoch": 2.66,
      "eval_kl": 0.4912835657596588,
      "eval_logps/chosen": -147.81866455078125,
      "eval_logps/rejected": -575.878662109375,
      "eval_loss": 0.011971179395914078,
      "eval_rewards/chosen": 6.3302388191223145,
      "eval_rewards/margins": 44.013816833496094,
      "eval_rewards/rejected": -37.68357849121094,
      "eval_runtime": 214.4054,
      "eval_samples_per_second": 2.229,
      "eval_steps_per_second": 0.56,
      "step": 350
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.019854838028550148,
      "kl": 0.05510709434747696,
      "learning_rate": 0.00011233595800524934,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0422,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 360
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.03696136921644211,
      "kl": 0.0,
      "learning_rate": 0.00010708661417322836,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0192,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 380
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.016547098755836487,
      "kl": 0.17841024696826935,
      "learning_rate": 0.00010183727034120735,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0049,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 400
    },
    {
      "epoch": 3.04,
      "eval_kl": 0.46487128734588623,
      "eval_logps/chosen": -147.84844970703125,
      "eval_logps/rejected": -498.3661804199219,
      "eval_loss": 0.010233537293970585,
      "eval_rewards/chosen": 6.327260494232178,
      "eval_rewards/margins": 36.25959014892578,
      "eval_rewards/rejected": -29.93233299255371,
      "eval_runtime": 214.0009,
      "eval_samples_per_second": 2.234,
      "eval_steps_per_second": 0.561,
      "step": 400
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.029680678620934486,
      "kl": 0.40129226446151733,
      "learning_rate": 9.658792650918635e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0096,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 420
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.01727895997464657,
      "kl": 0.006236672401428223,
      "learning_rate": 9.133858267716536e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0028,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 440
    },
    {
      "epoch": 3.42,
      "eval_kl": 0.29486456513404846,
      "eval_logps/chosen": -145.90557861328125,
      "eval_logps/rejected": -540.0706787109375,
      "eval_loss": 0.008336723782122135,
      "eval_rewards/chosen": 6.521546840667725,
      "eval_rewards/margins": 40.62433624267578,
      "eval_rewards/rejected": -34.10279083251953,
      "eval_runtime": 214.0142,
      "eval_samples_per_second": 2.233,
      "eval_steps_per_second": 0.561,
      "step": 450
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.7774410843849182,
      "kl": 0.0,
      "learning_rate": 8.608923884514435e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0052,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 460
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.009982769377529621,
      "kl": 0.0,
      "learning_rate": 8.083989501312336e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0075,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 480
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.04710818827152252,
      "kl": 0.0,
      "learning_rate": 7.559055118110236e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0087,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 500
    },
    {
      "epoch": 3.8,
      "eval_kl": 0.09233824163675308,
      "eval_logps/chosen": -147.00436401367188,
      "eval_logps/rejected": -551.1768798828125,
      "eval_loss": 0.009622834622859955,
      "eval_rewards/chosen": 6.411670207977295,
      "eval_rewards/margins": 41.62507247924805,
      "eval_rewards/rejected": -35.213401794433594,
      "eval_runtime": 213.9724,
      "eval_samples_per_second": 2.234,
      "eval_steps_per_second": 0.561,
      "step": 500
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.015628578141331673,
      "kl": 0.0,
      "learning_rate": 7.034120734908137e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.004,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 520
    },
    {
      "epoch": 4.11,
      "grad_norm": 0.006123641971498728,
      "kl": 0.14367903769016266,
      "learning_rate": 6.509186351706036e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.004,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 540
    },
    {
      "epoch": 4.18,
      "eval_kl": 0.15743787586688995,
      "eval_logps/chosen": -145.41305541992188,
      "eval_logps/rejected": -575.3412475585938,
      "eval_loss": 0.007457592524588108,
      "eval_rewards/chosen": 6.57080078125,
      "eval_rewards/margins": 44.20064163208008,
      "eval_rewards/rejected": -37.62984085083008,
      "eval_runtime": 214.0334,
      "eval_samples_per_second": 2.233,
      "eval_steps_per_second": 0.561,
      "step": 550
    },
    {
      "epoch": 4.26,
      "grad_norm": 0.0043973783031105995,
      "kl": 0.09317927062511444,
      "learning_rate": 5.984251968503938e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0027,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 560
    },
    {
      "epoch": 4.41,
      "grad_norm": 0.009126279503107071,
      "kl": 0.0,
      "learning_rate": 5.4593175853018376e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0031,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 580
    },
    {
      "epoch": 4.56,
      "grad_norm": 0.00867411121726036,
      "kl": 0.0,
      "learning_rate": 4.934383202099738e-05,
      "logps/chosen": NaN,
      "logps/rejected": NaN,
      "loss": 0.0036,
      "rewards/chosen": NaN,
      "rewards/margins": NaN,
      "rewards/rejected": NaN,
      "step": 600
    },
    {
      "epoch": 4.56,
      "eval_kl": 0.16292737424373627,
      "eval_logps/chosen": -144.68930053710938,
      "eval_logps/rejected": -565.9076538085938,
      "eval_loss": 0.0067643518559634686,
      "eval_rewards/chosen": 6.643174648284912,
      "eval_rewards/margins": 43.329654693603516,
      "eval_rewards/rejected": -36.68648147583008,
      "eval_runtime": 213.9275,
      "eval_samples_per_second": 2.234,
      "eval_steps_per_second": 0.561,
      "step": 600
    }
  ],
  "logging_steps": 20,
  "max_steps": 786,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}