{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9893390191897655,
  "eval_steps": 60,
  "global_step": 174,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "clip_ratio": 0.0,
      "completion_length": 604.4832763671875,
      "epoch": 0.017057569296375266,
      "grad_norm": 0.4691731333732605,
      "kl": 0.0,
      "learning_rate": 1.6666666666666665e-07,
      "loss": 0.0291,
      "reward": 0.628348246216774,
      "reward_std": 0.34620763547718525,
      "rewards/accuracy_reward": 0.628348246216774,
      "step": 1
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 607.7860174179077,
      "epoch": 0.08528784648187633,
      "grad_norm": 0.9751039147377014,
      "kl": 0.0001576244831085205,
      "learning_rate": 8.333333333333334e-07,
      "loss": 0.0137,
      "reward": 0.6099330596625805,
      "reward_std": 0.35781326424330473,
      "rewards/accuracy_reward": 0.6099330596625805,
      "step": 5
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 591.4216735839843,
      "epoch": 0.17057569296375266,
      "grad_norm": 0.31423789262771606,
      "kl": 0.0007061958312988281,
      "learning_rate": 1.6666666666666669e-06,
      "loss": 0.0366,
      "reward": 0.6220982402563096,
      "reward_std": 0.34715419709682466,
      "rewards/accuracy_reward": 0.6220982402563096,
      "step": 10
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 615.2261436462402,
      "epoch": 0.255863539445629,
      "grad_norm": 0.427952378988266,
      "kl": 0.003865623474121094,
      "learning_rate": 2.5e-06,
      "loss": 0.0603,
      "reward": 0.6941964596509933,
      "reward_std": 0.30072550252079966,
      "rewards/accuracy_reward": 0.6941964596509933,
      "step": 15
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 613.013639831543,
      "epoch": 0.3411513859275053,
      "grad_norm": 0.2186485081911087,
      "kl": 0.05426101684570313,
      "learning_rate": 2.9987834972573546e-06,
      "loss": 0.0751,
      "reward": 0.742857176065445,
      "reward_std": 0.23966127894818784,
      "rewards/accuracy_reward": 0.742857176065445,
      "step": 20
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 607.6678848266602,
      "epoch": 0.42643923240938164,
      "grad_norm": 0.2839179039001465,
      "kl": 0.00532073974609375,
      "learning_rate": 2.9851204919417464e-06,
      "loss": 0.0546,
      "reward": 0.7526786044239998,
      "reward_std": 0.19872783049941062,
      "rewards/accuracy_reward": 0.7526786044239998,
      "step": 25
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 601.8375274658204,
      "epoch": 0.511727078891258,
      "grad_norm": 0.13062593340873718,
      "kl": 0.0034595489501953124,
      "learning_rate": 2.956412726139078e-06,
      "loss": 0.0385,
      "reward": 0.7564732477068901,
      "reward_std": 0.17959171254187822,
      "rewards/accuracy_reward": 0.7564732477068901,
      "step": 30
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 597.1578399658204,
      "epoch": 0.5970149253731343,
      "grad_norm": 0.35054460167884827,
      "kl": 0.005078506469726562,
      "learning_rate": 2.9129510189868974e-06,
      "loss": 0.0341,
      "reward": 0.7522321745753289,
      "reward_std": 0.16307315276935697,
      "rewards/accuracy_reward": 0.7522321745753289,
      "step": 35
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 581.5973480224609,
      "epoch": 0.6823027718550106,
      "grad_norm": 0.11960138380527496,
      "kl": 0.00443115234375,
      "learning_rate": 2.8551756519155732e-06,
      "loss": 0.0368,
      "reward": 0.7622768253087997,
      "reward_std": 0.17930736346170306,
      "rewards/accuracy_reward": 0.7622768253087997,
      "step": 40
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 596.7301620483398,
      "epoch": 0.767590618336887,
      "grad_norm": 0.11841974407434464,
      "kl": 0.005435562133789063,
      "learning_rate": 2.7836719084521715e-06,
      "loss": 0.0353,
      "reward": 0.7607143238186836,
      "reward_std": 0.18814852144569158,
      "rewards/accuracy_reward": 0.7607143238186836,
      "step": 45
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 585.3933280944824,
      "epoch": 0.8528784648187633,
      "grad_norm": 0.10040715336799622,
      "kl": 0.0057315826416015625,
      "learning_rate": 2.699164145105252e-06,
      "loss": 0.025,
      "reward": 0.7537946745753288,
      "reward_std": 0.16595598962157965,
      "rewards/accuracy_reward": 0.7537946745753288,
      "step": 50
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 577.9147621154785,
      "epoch": 0.9381663113006397,
      "grad_norm": 0.24622774124145508,
      "kl": 0.007873153686523438,
      "learning_rate": 2.602508453394493e-06,
      "loss": 0.0338,
      "reward": 0.787723246216774,
      "reward_std": 0.18768559023737907,
      "rewards/accuracy_reward": 0.787723246216774,
      "step": 55
    },
    {
      "epoch": 1.0341151385927505,
      "grad_norm": 0.15860386192798615,
      "learning_rate": 2.4946839873611927e-06,
      "loss": 0.0344,
      "step": 60
    },
    {
      "epoch": 1.0341151385927505,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 563.3263092041016,
      "eval_kl": 0.007824897766113281,
      "eval_loss": 0.012524000369012356,
      "eval_reward": 0.6738281548023224,
      "eval_reward_std": 0.2058188347145915,
      "eval_rewards/accuracy_reward": 0.6738281548023224,
      "eval_runtime": 420.7271,
      "eval_samples_per_second": 1.188,
      "eval_steps_per_second": 0.012,
      "step": 60
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 558.5659664154052,
      "epoch": 1.1194029850746268,
      "grad_norm": 15.58234977722168,
      "kl": 0.015855026245117188,
      "learning_rate": 2.3767830444148337e-06,
      "loss": 0.0307,
      "reward": 0.778013427555561,
      "reward_std": 0.17006983272731305,
      "rewards/accuracy_reward": 0.778013427555561,
      "step": 65
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 548.0241348266602,
      "epoch": 1.2046908315565032,
      "grad_norm": 0.21251261234283447,
      "kl": 0.00769195556640625,
      "learning_rate": 2.25e-06,
      "loss": 0.0296,
      "reward": 0.7854911044239998,
      "reward_std": 0.16244285507127643,
      "rewards/accuracy_reward": 0.7854911044239998,
      "step": 70
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 563.5969009399414,
      "epoch": 1.2899786780383795,
      "grad_norm": 0.2231518030166626,
      "kl": 0.01058197021484375,
      "learning_rate": 2.1156192081791355e-06,
      "loss": 0.0301,
      "reward": 0.7732143193483353,
      "reward_std": 0.19672715626657009,
      "rewards/accuracy_reward": 0.7732143193483353,
      "step": 75
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 550.5730125427247,
      "epoch": 1.375266524520256,
      "grad_norm": 0.20531561970710754,
      "kl": 0.010945892333984375,
      "learning_rate": 1.975001990702209e-06,
      "loss": 0.0303,
      "reward": 0.7774553924798966,
      "reward_std": 0.1880516828969121,
      "rewards/accuracy_reward": 0.7774553924798966,
      "step": 80
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 559.7966743469239,
      "epoch": 1.4605543710021323,
      "grad_norm": 0.33798426389694214,
      "kl": 0.01414337158203125,
      "learning_rate": 1.829572846368326e-06,
      "loss": 0.04,
      "reward": 0.7678571820259095,
      "reward_std": 0.2084697039797902,
      "rewards/accuracy_reward": 0.7678571820259095,
      "step": 85
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 579.8114044189454,
      "epoch": 1.5458422174840085,
      "grad_norm": 0.6812328696250916,
      "kl": 0.029425048828125,
      "learning_rate": 1.6808050203829845e-06,
      "loss": 0.074,
      "reward": 0.736160746216774,
      "reward_std": 0.23041144628077745,
      "rewards/accuracy_reward": 0.736160746216774,
      "step": 90
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 606.0049392700196,
      "epoch": 1.6311300639658848,
      "grad_norm": 2.7374346256256104,
      "kl": 0.090008544921875,
      "learning_rate": 1.5302055798981605e-06,
      "loss": 0.1251,
      "reward": 0.6779018133878708,
      "reward_std": 0.30644247531890867,
      "rewards/accuracy_reward": 0.6779018133878708,
      "step": 95
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 638.524806213379,
      "epoch": 1.716417910447761,
      "grad_norm": 5.840439319610596,
      "kl": 0.35400390625,
      "learning_rate": 1.3793001469249112e-06,
      "loss": 0.1864,
      "reward": 0.6270089536905289,
      "reward_std": 0.331193171441555,
      "rewards/accuracy_reward": 0.6270089536905289,
      "step": 100
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 650.1977951049805,
      "epoch": 1.8017057569296375,
      "grad_norm": 14.723860740661621,
      "kl": 0.818017578125,
      "learning_rate": 1.2296174432791415e-06,
      "loss": 0.1931,
      "reward": 0.5450893104076385,
      "reward_std": 0.33599049374461176,
      "rewards/accuracy_reward": 0.5450893104076385,
      "step": 105
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 552.9535926818847,
      "epoch": 1.886993603411514,
      "grad_norm": 26.323904037475586,
      "kl": 1.0900390625,
      "learning_rate": 1.0826738041253211e-06,
      "loss": 0.1206,
      "reward": 0.5149553798139095,
      "reward_std": 0.3531858079135418,
      "rewards/accuracy_reward": 0.5149553798139095,
      "step": 110
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 505.44801330566406,
      "epoch": 1.9722814498933903,
      "grad_norm": 15.791183471679688,
      "kl": 0.76748046875,
      "learning_rate": 9.399578170010685e-07,
      "loss": 0.0274,
      "reward": 0.48549109175801275,
      "reward_std": 0.3420914553105831,
      "rewards/accuracy_reward": 0.48549109175801275,
      "step": 115
    },
    {
      "epoch": 2.068230277185501,
      "grad_norm": 14.470078468322754,
      "learning_rate": 8.029152419343472e-07,
      "loss": -0.0002,
      "step": 120
    },
    {
      "epoch": 2.068230277185501,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 477.81743812561035,
      "eval_kl": 0.83087158203125,
      "eval_loss": 0.014172928407788277,
      "eval_reward": 0.4252232350409031,
      "eval_reward_std": 0.2951010586693883,
      "eval_rewards/accuracy_reward": 0.4252232350409031,
      "eval_runtime": 449.8696,
      "eval_samples_per_second": 1.111,
      "eval_steps_per_second": 0.011,
      "step": 120
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 477.33952713012695,
      "epoch": 2.1535181236673773,
      "grad_norm": 20.696521759033203,
      "kl": 0.8743408203125,
      "learning_rate": 6.729343654174626e-07,
      "loss": 0.0312,
      "reward": 0.4702009156346321,
      "reward_std": 0.3341362001374364,
      "rewards/accuracy_reward": 0.4702009156346321,
      "step": 125
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 519.6768089294434,
      "epoch": 2.2388059701492535,
      "grad_norm": 32.883033752441406,
      "kl": 1.52685546875,
      "learning_rate": 5.513319366069343e-07,
      "loss": 0.0803,
      "reward": 0.46651787906885145,
      "reward_std": 0.31465277783572676,
      "rewards/accuracy_reward": 0.46651787906885145,
      "step": 130
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 538.6719017028809,
      "epoch": 2.3240938166311302,
      "grad_norm": 10.882832527160645,
      "kl": 1.92373046875,
      "learning_rate": 4.3933982822017883e-07,
      "loss": 0.1152,
      "reward": 0.43080359026789666,
      "reward_std": 0.29358932226896284,
      "rewards/accuracy_reward": 0.43080359026789666,
      "step": 135
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 530.0337310791016,
      "epoch": 2.4093816631130065,
      "grad_norm": 17.36768341064453,
      "kl": 1.71376953125,
      "learning_rate": 3.380925572585183e-07,
      "loss": 0.1217,
      "reward": 0.44866073280572893,
      "reward_std": 0.31872403249144554,
      "rewards/accuracy_reward": 0.44866073280572893,
      "step": 140
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 523.4377464294433,
      "epoch": 2.4946695095948828,
      "grad_norm": 8.74123764038086,
      "kl": 1.97939453125,
      "learning_rate": 2.4861579197570804e-07,
      "loss": 0.1345,
      "reward": 0.5015625268220901,
      "reward_std": 0.323883581161499,
      "rewards/accuracy_reward": 0.5015625268220901,
      "step": 145
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 530.0359634399414,
      "epoch": 2.579957356076759,
      "grad_norm": 11.574440002441406,
      "kl": 1.9712890625,
      "learning_rate": 1.718159615201853e-07,
      "loss": 0.1274,
      "reward": 0.4540178768336773,
      "reward_std": 0.3195581167936325,
      "rewards/accuracy_reward": 0.4540178768336773,
      "step": 150
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 528.8886405944825,
      "epoch": 2.6652452025586353,
      "grad_norm": 16.876115798950195,
      "kl": 1.59755859375,
      "learning_rate": 1.0847107350878571e-07,
      "loss": 0.1166,
      "reward": 0.4720982313156128,
      "reward_std": 0.33180556632578373,
      "rewards/accuracy_reward": 0.4720982313156128,
      "step": 155
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 525.8712371826172,
      "epoch": 2.750533049040512,
      "grad_norm": 8.542109489440918,
      "kl": 2.288671875,
      "learning_rate": 5.922283255294164e-08,
      "loss": 0.1575,
      "reward": 0.4640625201165676,
      "reward_std": 0.32937051728367805,
      "rewards/accuracy_reward": 0.4640625201165676,
      "step": 160
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 515.5904273986816,
      "epoch": 2.835820895522388,
      "grad_norm": 59.300743103027344,
      "kl": 2.4236328125,
      "learning_rate": 2.4570139579284723e-08,
      "loss": 0.1451,
      "reward": 0.4558035910129547,
      "reward_std": 0.3251536630094051,
      "rewards/accuracy_reward": 0.4558035910129547,
      "step": 165
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 529.5910972595215,
      "epoch": 2.9211087420042645,
      "grad_norm": 11.339794158935547,
      "kl": 2.040234375,
      "learning_rate": 4.864037798685106e-09,
      "loss": 0.1336,
      "reward": 0.4558035910129547,
      "reward_std": 0.3135051976889372,
      "rewards/accuracy_reward": 0.4558035910129547,
      "step": 170
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 523.6137437820435,
      "epoch": 2.9893390191897655,
      "kl": 2.185546875,
      "reward": 0.43164064455777407,
      "reward_std": 0.3127579181455076,
      "rewards/accuracy_reward": 0.43164064455777407,
      "step": 174,
      "total_flos": 0.0,
      "train_loss": 0.07567721740583536,
      "train_runtime": 24349.6607,
      "train_samples_per_second": 0.924,
      "train_steps_per_second": 0.007
    }
  ],
  "logging_steps": 5,
  "max_steps": 174,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}