{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.007605014556473174,
  "eval_steps": 500,
  "global_step": 48,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001584378032598578,
      "grad_norm": 4.8242447519442067e-05,
      "learning_rate": 0.0029998415465061004,
      "loss": 0.0,
      "loss/policy_avg": 7.299240678548813e-08,
      "objective/entropy": 67.43842315673828,
      "objective/kl": 0.0,
      "objective/rlhf_reward": 1.5219240188598633,
      "objective/scores": 1.52203369140625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 1,
      "timer/active_model_forward": 0.7594873309135437,
      "timer/calc_advantages": 31.33674430847168,
      "timer/calc_loss": 0.774461030960083,
      "timer/generate": 28.869869232177734,
      "timer/get_reward": 0.5197552442550659,
      "timer/model_forward": 0.7690438628196716,
      "timer/ref_model_forward": 0.7777600884437561,
      "timer/training_step": 34.545597076416016,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0003168756065197156,
      "grad_norm": 0.00026670348597690463,
      "learning_rate": 0.002999683093012201,
      "loss": -0.0,
      "loss/policy_avg": -1.7043203115463257e-07,
      "objective/entropy": 118.22715759277344,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -3.824268341064453,
      "objective/scores": -3.82421875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 2,
      "timer/active_model_forward": 0.6569761037826538,
      "timer/calc_advantages": 30.551788330078125,
      "timer/calc_loss": 0.6658673286437988,
      "timer/generate": 28.41728973388672,
      "timer/get_reward": 0.43157756328582764,
      "timer/model_forward": 0.6641299724578857,
      "timer/ref_model_forward": 0.67349773645401,
      "timer/training_step": 33.37031936645508,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0004753134097795734,
      "grad_norm": 0.00013700829003937542,
      "learning_rate": 0.0029995246395183015,
      "loss": -0.0,
      "loss/policy_avg": -2.9243528842926025e-07,
      "objective/entropy": 1081.7860107421875,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -5.476104736328125,
      "objective/scores": -5.476806640625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 3,
      "timer/active_model_forward": 0.9020705223083496,
      "timer/calc_advantages": 32.11817169189453,
      "timer/calc_loss": 0.9118615984916687,
      "timer/generate": 29.300296783447266,
      "timer/get_reward": 0.5959209203720093,
      "timer/model_forward": 0.9100804924964905,
      "timer/ref_model_forward": 0.9205952882766724,
      "timer/training_step": 35.83306121826172,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0006337512130394313,
      "grad_norm": 7.692693907301873e-05,
      "learning_rate": 0.002999366186024402,
      "loss": -0.0,
      "loss/policy_avg": -2.0489096641540527e-07,
      "objective/entropy": 1181.2274169921875,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -5.0478363037109375,
      "objective/scores": -5.04833984375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 4,
      "timer/active_model_forward": 0.8427364826202393,
      "timer/calc_advantages": 31.744182586669922,
      "timer/calc_loss": 0.8533791899681091,
      "timer/generate": 29.096416473388672,
      "timer/get_reward": 0.5531487464904785,
      "timer/model_forward": 0.851642906665802,
      "timer/ref_model_forward": 0.8617820143699646,
      "timer/training_step": 35.25868225097656,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.000792189016299289,
      "grad_norm": 3.911158273695037e-05,
      "learning_rate": 0.0029992077325305022,
      "loss": -0.0,
      "loss/policy_avg": -2.1513551473617554e-07,
      "objective/entropy": 1185.438232421875,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -4.198909759521484,
      "objective/scores": -4.198486328125,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 5,
      "timer/active_model_forward": 0.791987955570221,
      "timer/calc_advantages": 31.324127197265625,
      "timer/calc_loss": 0.8030194044113159,
      "timer/generate": 28.832427978515625,
      "timer/get_reward": 0.5202938914299011,
      "timer/model_forward": 0.8012962937355042,
      "timer/ref_model_forward": 0.8093616366386414,
      "timer/training_step": 34.65721130371094,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0009506268195591468,
      "grad_norm": 2.5278823159169406e-05,
      "learning_rate": 0.002999049279036603,
      "loss": -0.0,
      "loss/policy_avg": -1.2619420886039734e-07,
      "objective/entropy": 528.8932495117188,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -3.1182708740234375,
      "objective/scores": -3.1180419921875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 6,
      "timer/active_model_forward": 0.7996934652328491,
      "timer/calc_advantages": 31.30746078491211,
      "timer/calc_loss": 0.8107499480247498,
      "timer/generate": 28.788341522216797,
      "timer/get_reward": 0.5261460542678833,
      "timer/model_forward": 0.8090219497680664,
      "timer/ref_model_forward": 0.8171602487564087,
      "timer/training_step": 34.65579605102539,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0011090646228190046,
      "grad_norm": 1.0207173772869282e-06,
      "learning_rate": 0.0029988908255427034,
      "loss": -0.0,
      "loss/policy_avg": -3.1781382858753204e-08,
      "objective/entropy": 1.5396817922592163,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.2192144393920898,
      "objective/scores": -1.2193603515625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 7,
      "timer/active_model_forward": 1.0039308071136475,
      "timer/calc_advantages": 32.610069274902344,
      "timer/calc_loss": 1.0154139995574951,
      "timer/generate": 29.52051544189453,
      "timer/get_reward": 0.6604768633842468,
      "timer/model_forward": 1.0136762857437134,
      "timer/ref_model_forward": 1.0217972993850708,
      "timer/training_step": 36.71036911010742,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0012675024260788625,
      "grad_norm": 1.4357982536239433e-06,
      "learning_rate": 0.0029987323720488037,
      "loss": -0.0,
      "loss/policy_avg": -7.380731403827667e-08,
      "objective/entropy": 2.126983404159546,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.3588581085205078,
      "objective/scores": -1.35894775390625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 8,
      "timer/active_model_forward": 1.0782428979873657,
      "timer/calc_advantages": 33.10707092285156,
      "timer/calc_loss": 1.0924396514892578,
      "timer/generate": 29.794893264770508,
      "timer/get_reward": 0.7108417749404907,
      "timer/model_forward": 1.0906702280044556,
      "timer/ref_model_forward": 1.0941966772079468,
      "timer/training_step": 37.46529769897461,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0014259402293387202,
      "grad_norm": 1.58626687607466e-06,
      "learning_rate": 0.002998573918554904,
      "loss": -0.0,
      "loss/policy_avg": -5.851325113326311e-08,
      "objective/entropy": 5.8357086181640625,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.3656387329101562,
      "objective/scores": -1.36566162109375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 9,
      "timer/active_model_forward": 0.9783480167388916,
      "timer/calc_advantages": 32.70989990234375,
      "timer/calc_loss": 0.9892486333847046,
      "timer/generate": 29.658161163330078,
      "timer/get_reward": 0.6437245011329651,
      "timer/model_forward": 0.9874842166900635,
      "timer/ref_model_forward": 0.9964737892150879,
      "timer/training_step": 36.70344924926758,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.001584378032598578,
      "grad_norm": 1.8057213537758798e-06,
      "learning_rate": 0.002998415465061005,
      "loss": -0.0,
      "loss/policy_avg": -4.391768015921116e-08,
      "objective/entropy": 37.01044464111328,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.1269941329956055,
      "objective/scores": -1.126983642578125,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 10,
      "timer/active_model_forward": 0.7927595973014832,
      "timer/calc_advantages": 31.501224517822266,
      "timer/calc_loss": 0.8029042482376099,
      "timer/generate": 28.97046661376953,
      "timer/get_reward": 0.5210747718811035,
      "timer/model_forward": 0.8011468052864075,
      "timer/ref_model_forward": 0.8100461363792419,
      "timer/training_step": 34.83835983276367,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0017428158358584358,
      "grad_norm": 1.3606485254058498e-06,
      "learning_rate": 0.002998257011567105,
      "loss": -0.0,
      "loss/policy_avg": -6.2282197177410126e-09,
      "objective/entropy": 82.50379943847656,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.20352506637573242,
      "objective/scores": -0.20351409912109375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 11,
      "timer/active_model_forward": 0.8447611331939697,
      "timer/calc_advantages": 31.68113899230957,
      "timer/calc_loss": 0.8556842803955078,
      "timer/generate": 29.030838012695312,
      "timer/get_reward": 0.5556706786155701,
      "timer/model_forward": 0.853936493396759,
      "timer/ref_model_forward": 0.8623838424682617,
      "timer/training_step": 35.189205169677734,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0019012536391182935,
      "grad_norm": 1.5168055824688054e-06,
      "learning_rate": 0.0029980985580732055,
      "loss": -0.0,
      "loss/policy_avg": -5.324545782059431e-08,
      "objective/entropy": 43.37924575805664,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.273842453956604,
      "objective/scores": -1.2738037109375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 12,
      "timer/active_model_forward": 0.6771436929702759,
      "timer/calc_advantages": 30.566415786743164,
      "timer/calc_loss": 0.6872404217720032,
      "timer/generate": 28.381263732910156,
      "timer/get_reward": 0.4441335201263428,
      "timer/model_forward": 0.6854845881462097,
      "timer/ref_model_forward": 0.6946568489074707,
      "timer/training_step": 33.479454040527344,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0020596914423781513,
      "grad_norm": 9.032095817929076e-07,
      "learning_rate": 0.0029979401045793063,
      "loss": -0.0,
      "loss/policy_avg": -5.2401446737349033e-08,
      "objective/entropy": 43.8469123840332,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.2306889295578003,
      "objective/scores": -1.230804443359375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 13,
      "timer/active_model_forward": 0.8871802091598511,
      "timer/calc_advantages": 31.831775665283203,
      "timer/calc_loss": 0.8963775634765625,
      "timer/generate": 29.063201904296875,
      "timer/get_reward": 0.5812451839447021,
      "timer/model_forward": 0.8946273326873779,
      "timer/ref_model_forward": 0.9052593111991882,
      "timer/training_step": 35.49396514892578,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.002218129245638009,
      "grad_norm": 1.8163491404266097e-06,
      "learning_rate": 0.0029977816510854067,
      "loss": -0.0,
      "loss/policy_avg": -6.629852578043938e-08,
      "objective/entropy": 46.85642623901367,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.4318056106567383,
      "objective/scores": -1.431884765625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 14,
      "timer/active_model_forward": 0.7645549774169922,
      "timer/calc_advantages": 31.06265640258789,
      "timer/calc_loss": 0.775362491607666,
      "timer/generate": 28.632740020751953,
      "timer/get_reward": 0.5032914876937866,
      "timer/model_forward": 0.7736347317695618,
      "timer/ref_model_forward": 0.7821637988090515,
      "timer/training_step": 34.282867431640625,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.002376567048897867,
      "grad_norm": 1.1021771797459223e-06,
      "learning_rate": 0.002997623197591507,
      "loss": -0.0,
      "loss/policy_avg": -4.4819898903369904e-08,
      "objective/entropy": 83.53251647949219,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.6715226173400879,
      "objective/scores": -0.671539306640625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 15,
      "timer/active_model_forward": 1.1367318630218506,
      "timer/calc_advantages": 33.59859848022461,
      "timer/calc_loss": 1.2175476551055908,
      "timer/generate": 30.091690063476562,
      "timer/get_reward": 0.7473501563072205,
      "timer/model_forward": 1.2155828475952148,
      "timer/ref_model_forward": 1.1532230377197266,
      "timer/training_step": 38.26678466796875,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.002535004852157725,
      "grad_norm": 1.2331845482549397e-06,
      "learning_rate": 0.0029974647440976073,
      "loss": -0.0,
      "loss/policy_avg": -1.6909325495362282e-08,
      "objective/entropy": 119.47362518310547,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.5125551223754883,
      "objective/scores": -0.5126724243164062,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 16,
      "timer/active_model_forward": 0.9046459197998047,
      "timer/calc_advantages": 32.02373123168945,
      "timer/calc_loss": 0.914932906627655,
      "timer/generate": 29.190690994262695,
      "timer/get_reward": 0.5946121215820312,
      "timer/model_forward": 0.9131810665130615,
      "timer/ref_model_forward": 0.922136127948761,
      "timer/training_step": 35.75535202026367,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0026934426554175825,
      "grad_norm": 2.2155286387715023e-06,
      "learning_rate": 0.002997306290603708,
      "loss": -0.0,
      "loss/policy_avg": -4.8712536226958036e-08,
      "objective/entropy": 121.45540618896484,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.4430209398269653,
      "objective/scores": -1.443115234375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 17,
      "timer/active_model_forward": 0.7191606163978577,
      "timer/calc_advantages": 31.075271606445312,
      "timer/calc_loss": 0.7287229895591736,
      "timer/generate": 28.78363037109375,
      "timer/get_reward": 0.4706084430217743,
      "timer/model_forward": 0.7269375324249268,
      "timer/ref_model_forward": 0.737448513507843,
      "timer/training_step": 34.119632720947266,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0028518804586774404,
      "grad_norm": 8.308816177304834e-07,
      "learning_rate": 0.0029971478371098085,
      "loss": -0.0,
      "loss/policy_avg": -1.1129304766654968e-07,
      "objective/entropy": 96.97429656982422,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -2.8110389709472656,
      "objective/scores": -2.811279296875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 18,
      "timer/active_model_forward": 0.8409894704818726,
      "timer/calc_advantages": 31.73868179321289,
      "timer/calc_loss": 0.8524103164672852,
      "timer/generate": 29.05769920349121,
      "timer/get_reward": 0.5517525672912598,
      "timer/model_forward": 0.8506196737289429,
      "timer/ref_model_forward": 0.8585314154624939,
      "timer/training_step": 35.25176239013672,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0030103182619372983,
      "grad_norm": 3.799691228323354e-07,
      "learning_rate": 0.002996989383615909,
      "loss": -0.0,
      "loss/policy_avg": -9.569339454174042e-08,
      "objective/entropy": 85.54360961914062,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -2.591951370239258,
      "objective/scores": -2.591796875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 19,
      "timer/active_model_forward": 0.8125832080841064,
      "timer/calc_advantages": 31.4530029296875,
      "timer/calc_loss": 0.8229714632034302,
      "timer/generate": 28.860492706298828,
      "timer/get_reward": 0.5346251130104065,
      "timer/model_forward": 0.8212224245071411,
      "timer/ref_model_forward": 0.8305649757385254,
      "timer/training_step": 34.838218688964844,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.003168756065197156,
      "grad_norm": 3.7777222132717725e-06,
      "learning_rate": 0.002996830930122009,
      "loss": -0.0,
      "loss/policy_avg": -7.898779585957527e-08,
      "objective/entropy": 78.87588500976562,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -2.1325230598449707,
      "objective/scores": -2.13275146484375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 20,
      "timer/active_model_forward": 1.0649389028549194,
      "timer/calc_advantages": 33.12159729003906,
      "timer/calc_loss": 1.0808004140853882,
      "timer/generate": 29.827653884887695,
      "timer/get_reward": 0.7013706564903259,
      "timer/model_forward": 1.0790588855743408,
      "timer/ref_model_forward": 1.0825612545013428,
      "timer/training_step": 37.4581413269043,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0033271938684570138,
      "grad_norm": 2.958485652015952e-07,
      "learning_rate": 0.00299667247662811,
      "loss": -0.0,
      "loss/policy_avg": -6.577465683221817e-08,
      "objective/entropy": 26.52663803100586,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.5283641815185547,
      "objective/scores": -1.52838134765625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 21,
      "timer/active_model_forward": 0.8408429622650146,
      "timer/calc_advantages": 31.583477020263672,
      "timer/calc_loss": 0.8517074584960938,
      "timer/generate": 28.94709014892578,
      "timer/get_reward": 0.5524436235427856,
      "timer/model_forward": 0.8499797582626343,
      "timer/ref_model_forward": 0.859445333480835,
      "timer/training_step": 35.0816535949707,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0034856316717168717,
      "grad_norm": 2.1688055085178348e-07,
      "learning_rate": 0.0029965140231342103,
      "loss": -0.0,
      "loss/policy_avg": -7.275957614183426e-08,
      "objective/entropy": 5.097399711608887,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.3907923698425293,
      "objective/scores": -1.390869140625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 22,
      "timer/active_model_forward": 0.7821312546730042,
      "timer/calc_advantages": 31.17966079711914,
      "timer/calc_loss": 0.7932944297790527,
      "timer/generate": 28.714157104492188,
      "timer/get_reward": 0.5111904144287109,
      "timer/model_forward": 0.7915641069412231,
      "timer/ref_model_forward": 0.8004368543624878,
      "timer/training_step": 34.46521759033203,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0036440694749767296,
      "grad_norm": 1.0821660367810182e-07,
      "learning_rate": 0.0029963555696403106,
      "loss": -0.0,
      "loss/policy_avg": -5.2619725465774536e-08,
      "objective/entropy": 0.7932516932487488,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.3286209106445312,
      "objective/scores": -1.3287353515625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 23,
      "timer/active_model_forward": 0.7814183235168457,
      "timer/calc_advantages": 31.230453491210938,
      "timer/calc_loss": 0.7906038761138916,
      "timer/generate": 28.76888656616211,
      "timer/get_reward": 0.511726438999176,
      "timer/model_forward": 0.7888429164886475,
      "timer/ref_model_forward": 0.7980221509933472,
      "timer/training_step": 34.51930236816406,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.003802507278236587,
      "grad_norm": 7.358872267104744e-08,
      "learning_rate": 0.0029961971161464114,
      "loss": -0.0,
      "loss/policy_avg": -4.1211023926734924e-08,
      "objective/entropy": 0.42043036222457886,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.3438358306884766,
      "objective/scores": -1.34381103515625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 24,
      "timer/active_model_forward": 1.0947436094284058,
      "timer/calc_advantages": 33.27772903442383,
      "timer/calc_loss": 1.1070483922958374,
      "timer/generate": 29.915428161621094,
      "timer/get_reward": 0.7210176587104797,
      "timer/model_forward": 1.1053131818771362,
      "timer/ref_model_forward": 1.1129379272460938,
      "timer/training_step": 37.71331024169922,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0039609450814964454,
      "grad_norm": 2.472104299044986e-08,
      "learning_rate": 0.0029960386626525118,
      "loss": -0.0,
      "loss/policy_avg": -5.727633833885193e-08,
      "objective/entropy": 0.012148905545473099,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.9438667297363281,
      "objective/scores": -0.943817138671875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 25,
      "timer/active_model_forward": 0.8634592890739441,
      "timer/calc_advantages": 31.91621971130371,
      "timer/calc_loss": 0.8720701932907104,
      "timer/generate": 29.20632553100586,
      "timer/get_reward": 0.5676878094673157,
      "timer/model_forward": 0.8703302145004272,
      "timer/ref_model_forward": 0.8802404999732971,
      "timer/training_step": 35.50330352783203,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0041193828847563025,
      "grad_norm": 7.710570848473708e-09,
      "learning_rate": 0.002995880209158612,
      "loss": -0.0,
      "loss/policy_avg": -5.185938789509237e-08,
      "objective/entropy": 0.004173352383077145,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0292143821716309,
      "objective/scores": -1.02923583984375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 26,
      "timer/active_model_forward": 1.1470246315002441,
      "timer/calc_advantages": 33.62145233154297,
      "timer/calc_loss": 1.1597700119018555,
      "timer/generate": 30.103939056396484,
      "timer/get_reward": 0.7558026313781738,
      "timer/model_forward": 1.1580355167388916,
      "timer/ref_model_forward": 1.1651380062103271,
      "timer/training_step": 38.25925064086914,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.00427782068801616,
      "grad_norm": 3.3511178276768305e-09,
      "learning_rate": 0.0029957217556647125,
      "loss": -0.0,
      "loss/policy_avg": -5.727633833885193e-08,
      "objective/entropy": 0.0016420490574091673,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0198352336883545,
      "objective/scores": -1.0199356079101562,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 27,
      "timer/active_model_forward": 0.8805452585220337,
      "timer/calc_advantages": 31.86100196838379,
      "timer/calc_loss": 0.8916431665420532,
      "timer/generate": 29.115583419799805,
      "timer/get_reward": 0.5775084495544434,
      "timer/model_forward": 0.8899136185646057,
      "timer/ref_model_forward": 0.8984355926513672,
      "timer/training_step": 35.50636291503906,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.004436258491276018,
      "grad_norm": 1.413538508465706e-09,
      "learning_rate": 0.0029955633021708132,
      "loss": -0.0,
      "loss/policy_avg": -5.634501576423645e-08,
      "objective/entropy": 0.0007059703930281103,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.7993221282958984,
      "objective/scores": -0.79931640625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 28,
      "timer/active_model_forward": 1.1332266330718994,
      "timer/calc_advantages": 33.521759033203125,
      "timer/calc_loss": 1.1450132131576538,
      "timer/generate": 30.048032760620117,
      "timer/get_reward": 0.7485961318016052,
      "timer/model_forward": 1.1432639360427856,
      "timer/ref_model_forward": 1.1509253978729248,
      "timer/training_step": 38.09806442260742,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.004594696294535876,
      "grad_norm": 5.78209535806451e-10,
      "learning_rate": 0.0029954048486769136,
      "loss": -0.0,
      "loss/policy_avg": -4.7963112592697144e-08,
      "objective/entropy": 0.00033563331817276776,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.8303084373474121,
      "objective/scores": -0.8302459716796875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 29,
      "timer/active_model_forward": 0.5042444467544556,
      "timer/calc_advantages": 29.47503662109375,
      "timer/calc_loss": 0.5117393136024475,
      "timer/generate": 27.804088592529297,
      "timer/get_reward": 0.32876861095428467,
      "timer/model_forward": 0.5100182294845581,
      "timer/ref_model_forward": 0.5201169848442078,
      "timer/training_step": 31.75719451904297,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.004753134097795734,
      "grad_norm": 1.9897845870975317e-10,
      "learning_rate": 0.002995246395183014,
      "loss": -0.0,
      "loss/policy_avg": -3.341119736433029e-08,
      "objective/entropy": 0.00016836810391396284,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0440139770507812,
      "objective/scores": -1.0440673828125,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 30,
      "timer/active_model_forward": 0.8818363547325134,
      "timer/calc_advantages": 31.895645141601562,
      "timer/calc_loss": 0.8967288732528687,
      "timer/generate": 29.129695892333984,
      "timer/get_reward": 0.5798584222793579,
      "timer/model_forward": 0.8949993848800659,
      "timer/ref_model_forward": 0.8993439674377441,
      "timer/training_step": 35.555747985839844,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.004911571901055592,
      "grad_norm": 1.4891138322425945e-10,
      "learning_rate": 0.0029950879416891143,
      "loss": -0.0,
      "loss/policy_avg": -5.029141902923584e-08,
      "objective/entropy": 8.117404649965465e-05,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.2238502502441406,
      "objective/scores": -1.22381591796875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 31,
      "timer/active_model_forward": 0.9351563453674316,
      "timer/calc_advantages": 32.229801177978516,
      "timer/calc_loss": 0.946864128112793,
      "timer/generate": 29.3177490234375,
      "timer/get_reward": 0.6167161464691162,
      "timer/model_forward": 0.9451220035552979,
      "timer/ref_model_forward": 0.9523330330848694,
      "timer/training_step": 36.09205627441406,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.00507000970431545,
      "grad_norm": 8.760232816928948e-11,
      "learning_rate": 0.0029949294881952146,
      "loss": -0.0,
      "loss/policy_avg": -4.237517714500427e-08,
      "objective/entropy": 6.1035148974042386e-05,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0421943664550781,
      "objective/scores": -1.042266845703125,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 32,
      "timer/active_model_forward": 0.829505205154419,
      "timer/calc_advantages": 31.583553314208984,
      "timer/calc_loss": 0.8393364548683167,
      "timer/generate": 28.969213485717773,
      "timer/get_reward": 0.5460556745529175,
      "timer/model_forward": 0.8376072645187378,
      "timer/ref_model_forward": 0.8457875847816467,
      "timer/training_step": 35.03805160522461,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.005228447507575307,
      "grad_norm": 5.447125389634877e-11,
      "learning_rate": 0.002994771034701315,
      "loss": -0.0,
      "loss/policy_avg": -4.797766450792551e-08,
      "objective/entropy": 3.0517576306010596e-05,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0386168956756592,
      "objective/scores": -1.03875732421875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 33,
      "timer/active_model_forward": 0.7481950521469116,
      "timer/calc_advantages": 31.263355255126953,
      "timer/calc_loss": 0.7592823505401611,
      "timer/generate": 28.858570098876953,
      "timer/get_reward": 0.4890865087509155,
      "timer/model_forward": 0.7575281262397766,
      "timer/ref_model_forward": 0.7657535672187805,
      "timer/training_step": 34.43185806274414,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.005386885310835165,
      "grad_norm": 3.8278300668048715e-11,
      "learning_rate": 0.0029946125812074153,
      "loss": -0.0,
      "loss/policy_avg": -3.8475263863801956e-08,
      "objective/entropy": 3.0517576306010596e-05,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.1492676734924316,
      "objective/scores": -1.14923095703125,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 34,
      "timer/active_model_forward": 0.9886236190795898,
      "timer/calc_advantages": 32.65524673461914,
      "timer/calc_loss": 1.0020769834518433,
      "timer/generate": 29.594369888305664,
      "timer/get_reward": 0.6510753631591797,
      "timer/model_forward": 1.0003337860107422,
      "timer/ref_model_forward": 1.0066032409667969,
      "timer/training_step": 36.705928802490234,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.005545323114095023,
      "grad_norm": 3.6770659433971176e-11,
      "learning_rate": 0.002994454127713516,
      "loss": -0.0,
      "loss/policy_avg": -6.775371730327606e-08,
      "objective/entropy": 1.5258788153005298e-05,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0323657989501953,
      "objective/scores": -1.03228759765625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 35,
      "timer/active_model_forward": 0.8282456398010254,
      "timer/calc_advantages": 31.575471878051758,
      "timer/calc_loss": 0.8373382687568665,
      "timer/generate": 28.9643611907959,
      "timer/get_reward": 0.542504072189331,
      "timer/model_forward": 0.8355766534805298,
      "timer/ref_model_forward": 0.8458244204521179,
      "timer/training_step": 35.05115509033203,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.005703760917354881,
      "grad_norm": 1.9524635380685496e-11,
      "learning_rate": 0.0029942956742196165,
      "loss": -0.0,
      "loss/policy_avg": -3.958120942115784e-08,
      "objective/entropy": 1.5258788153005298e-05,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.149190902709961,
      "objective/scores": -1.1492919921875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 36,
      "timer/active_model_forward": 0.948405385017395,
      "timer/calc_advantages": 32.343502044677734,
      "timer/calc_loss": 0.9663164615631104,
      "timer/generate": 29.394954681396484,
      "timer/get_reward": 0.622231662273407,
      "timer/model_forward": 0.9645813703536987,
      "timer/ref_model_forward": 0.9657047390937805,
      "timer/training_step": 36.252403259277344,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.005862198720614739,
      "grad_norm": 3.229160688844601e-11,
      "learning_rate": 0.002994137220725717,
      "loss": -0.0,
      "loss/policy_avg": -6.612390279769897e-08,
      "objective/entropy": 1.5258788153005298e-05,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.2540512084960938,
      "objective/scores": -1.25396728515625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 37,
      "timer/active_model_forward": 1.062816858291626,
      "timer/calc_advantages": 33.06248474121094,
      "timer/calc_loss": 1.074930191040039,
      "timer/generate": 29.784706115722656,
      "timer/get_reward": 0.7008613348007202,
      "timer/model_forward": 1.0731862783432007,
      "timer/ref_model_forward": 1.0801432132720947,
      "timer/training_step": 37.3824348449707,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.006020636523874597,
      "grad_norm": 7.194660232162642e-12,
      "learning_rate": 0.0029939787672318176,
      "loss": -0.0,
      "loss/policy_avg": -3.9639417082071304e-08,
      "objective/entropy": 3.0249354949773988e-06,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0819144248962402,
      "objective/scores": -1.082000732421875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 38,
      "timer/active_model_forward": 0.9271302223205566,
      "timer/calc_advantages": 32.216827392578125,
      "timer/calc_loss": 0.9388477802276611,
      "timer/generate": 29.325376510620117,
      "timer/get_reward": 0.6095032095909119,
      "timer/model_forward": 0.9370960593223572,
      "timer/ref_model_forward": 0.9444317817687988,
      "timer/training_step": 36.04484558105469,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.006179074327134455,
      "grad_norm": 7.418058355079804e-12,
      "learning_rate": 0.002993820313737918,
      "loss": -0.0,
      "loss/policy_avg": -4.330695446697064e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.1705617904663086,
      "objective/scores": -1.170654296875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 39,
      "timer/active_model_forward": 0.7922002077102661,
      "timer/calc_advantages": 31.311412811279297,
      "timer/calc_loss": 0.8026509284973145,
      "timer/generate": 28.80554962158203,
      "timer/get_reward": 0.5210146903991699,
      "timer/model_forward": 0.800923764705658,
      "timer/ref_model_forward": 0.8090575933456421,
      "timer/training_step": 34.634342193603516,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.006337512130394312,
      "grad_norm": 4.4267285150001445e-12,
      "learning_rate": 0.0029936618602440183,
      "loss": -0.0,
      "loss/policy_avg": -3.7136487662792206e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.7537899017333984,
      "objective/scores": -0.7537307739257812,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 40,
      "timer/active_model_forward": 0.7792460322380066,
      "timer/calc_advantages": 31.258018493652344,
      "timer/calc_loss": 0.7891390323638916,
      "timer/generate": 28.808034896850586,
      "timer/get_reward": 0.51292884349823,
      "timer/model_forward": 0.7873988151550293,
      "timer/ref_model_forward": 0.795951247215271,
      "timer/training_step": 34.5465087890625,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.00649594993365417,
      "grad_norm": 7.40804379645299e-12,
      "learning_rate": 0.0029935034067501186,
      "loss": -0.0,
      "loss/policy_avg": -6.239861249923706e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0780410766601562,
      "objective/scores": -1.0780029296875,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 41,
      "timer/active_model_forward": 0.9782811999320984,
      "timer/calc_advantages": 32.72919845581055,
      "timer/calc_loss": 0.9898173809051514,
      "timer/generate": 29.66815185546875,
      "timer/get_reward": 0.6456098556518555,
      "timer/model_forward": 0.9880228042602539,
      "timer/ref_model_forward": 0.9960311651229858,
      "timer/training_step": 36.747779846191406,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0066543877369140275,
      "grad_norm": 4.3237414516783534e-12,
      "learning_rate": 0.0029933449532562194,
      "loss": -0.0,
      "loss/policy_avg": -5.203764885663986e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0865364074707031,
      "objective/scores": -1.0865478515625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 42,
      "timer/active_model_forward": 0.9354751110076904,
      "timer/calc_advantages": 32.381378173828125,
      "timer/calc_loss": 0.9460563659667969,
      "timer/generate": 29.42566680908203,
      "timer/get_reward": 0.6166284680366516,
      "timer/model_forward": 0.9442933797836304,
      "timer/ref_model_forward": 0.9522690773010254,
      "timer/training_step": 36.222293853759766,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.0068128255401738854,
      "grad_norm": 3.2976431914993887e-12,
      "learning_rate": 0.0029931864997623197,
      "loss": -0.0,
      "loss/policy_avg": -4.0046870708465576e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.9958648681640625,
      "objective/scores": -0.99603271484375,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 43,
      "timer/active_model_forward": 0.7636290192604065,
      "timer/calc_advantages": 31.25667381286621,
      "timer/calc_loss": 0.7747111916542053,
      "timer/generate": 28.780200958251953,
      "timer/get_reward": 0.5010501146316528,
      "timer/model_forward": 0.7729100584983826,
      "timer/ref_model_forward": 0.7797021269798279,
      "timer/training_step": 34.46833038330078,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.006971263343433743,
      "grad_norm": 5.412355025963267e-12,
      "learning_rate": 0.00299302804626842,
      "loss": -0.0,
      "loss/policy_avg": -7.310882210731506e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.306192398071289,
      "objective/scores": -1.30621337890625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 44,
      "timer/active_model_forward": 0.8105420470237732,
      "timer/calc_advantages": 31.56263542175293,
      "timer/calc_loss": 0.8194728493690491,
      "timer/generate": 28.94143295288086,
      "timer/get_reward": 0.5309967994689941,
      "timer/model_forward": 0.8177112340927124,
      "timer/ref_model_forward": 0.8259291052818298,
      "timer/training_step": 34.96255111694336,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.007129701146693601,
      "grad_norm": 3.038823749926256e-12,
      "learning_rate": 0.0029928695927745204,
      "loss": -0.0,
      "loss/policy_avg": -5.2619725465774536e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.0578193664550781,
      "objective/scores": -1.05767822265625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 45,
      "timer/active_model_forward": 1.0397440195083618,
      "timer/calc_advantages": 33.09510803222656,
      "timer/calc_loss": 1.0498113632202148,
      "timer/generate": 29.858869552612305,
      "timer/get_reward": 0.6861503720283508,
      "timer/model_forward": 1.0480440855026245,
      "timer/ref_model_forward": 1.055608868598938,
      "timer/training_step": 37.31883239746094,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.007288138949953459,
      "grad_norm": 1.18787076529292e-12,
      "learning_rate": 0.0029927111392806212,
      "loss": -0.0,
      "loss/policy_avg": -2.1420419216156006e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.8209381103515625,
      "objective/scores": -0.82086181640625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 46,
      "timer/active_model_forward": 1.0535329580307007,
      "timer/calc_advantages": 33.19418716430664,
      "timer/calc_loss": 1.0623672008514404,
      "timer/generate": 29.923227310180664,
      "timer/get_reward": 0.6941207647323608,
      "timer/model_forward": 1.0605970621109009,
      "timer/ref_model_forward": 1.070003628730774,
      "timer/training_step": 37.47746276855469,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.007446576753213317,
      "grad_norm": 2.7435091958655233e-12,
      "learning_rate": 0.0029925526857867216,
      "loss": -0.0,
      "loss/policy_avg": -5.3085386753082275e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -0.9966888427734375,
      "objective/scores": -0.9967041015625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 47,
      "timer/active_model_forward": 0.7242672443389893,
      "timer/calc_advantages": 30.85483741760254,
      "timer/calc_loss": 0.7333927154541016,
      "timer/generate": 28.548473358154297,
      "timer/get_reward": 0.4773656129837036,
      "timer/model_forward": 0.7316561341285706,
      "timer/ref_model_forward": 0.7414883971214294,
      "timer/training_step": 33.913536071777344,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    },
    {
      "epoch": 0.007605014556473174,
      "grad_norm": 2.9446627428098004e-12,
      "learning_rate": 0.002992394232292822,
      "loss": -0.0,
      "loss/policy_avg": -5.8673322200775146e-08,
      "objective/entropy": 0.0,
      "objective/kl": 0.0,
      "objective/rlhf_reward": -1.2014045715332031,
      "objective/scores": -1.20123291015625,
      "policy/approxkl_avg": 0.0,
      "policy/clipfrac_avg": 0.0,
      "step": 48,
      "timer/active_model_forward": 0.7452950477600098,
      "timer/calc_advantages": 30.971996307373047,
      "timer/calc_loss": 0.7579642534255981,
      "timer/generate": 28.605379104614258,
      "timer/get_reward": 0.48896369338035583,
      "timer/model_forward": 0.7562232613563538,
      "timer/ref_model_forward": 0.7629098892211914,
      "timer/training_step": 34.12104797363281,
      "val/num_eos_tokens": 0.0,
      "val/ratio": 1.0
    }
  ],
  "logging_steps": 1,
  "max_steps": 18933,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 8,
  "total_flos": 2.519438650707149e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}