trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.08237232289950576,
  "eval_steps": 1000,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.4658203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 88.3837890625,
      "completions/mean_terminated_length": 53.83729553222656,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0787938493303955,
      "epoch": 0.00041186161449752884,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.1026352643966675,
      "kl": 1.5408068257727336e-05,
      "learning_rate": 0.0,
      "loss": 0.0612,
      "num_tokens": 473618.0,
      "reward": -0.654300332069397,
      "reward_std": 1.2014957666397095,
      "rewards/reward_model/mean": -0.654300332069397,
      "rewards/reward_model/std": 1.4879947900772095,
      "step": 1,
      "step_time": 179.40438475832343
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.521484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 95.0380859375,
      "completions/mean_terminated_length": 59.11632537841797,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0889650019817054,
      "epoch": 0.0008237232289950577,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9445520043373108,
      "kl": 1.5487904489575044e-05,
      "learning_rate": 1.2345679012345681e-08,
      "loss": 0.0685,
      "num_tokens": 944384.0,
      "reward": -0.6944406032562256,
      "reward_std": 1.1158981323242188,
      "rewards/reward_model/mean": -0.6944406032562256,
      "rewards/reward_model/std": 1.4779117107391357,
      "step": 2,
      "step_time": 168.28568758117035
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.4921875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 93.1064453125,
      "completions/mean_terminated_length": 59.286537170410156,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.1078118681907654,
      "epoch": 0.0012355848434925864,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9809994101524353,
      "kl": 0.0009136445219155576,
      "learning_rate": 2.4691358024691362e-08,
      "loss": 0.061,
      "num_tokens": 1417434.0,
      "reward": -0.8067716956138611,
      "reward_std": 1.1805193424224854,
      "rewards/reward_model/mean": -0.8067716956138611,
      "rewards/reward_model/std": 1.5296157598495483,
      "step": 3,
      "step_time": 168.7894278760068
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.4912109375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 90.3203125,
      "completions/mean_terminated_length": 53.94241714477539,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0778324585407972,
      "epoch": 0.0016474464579901153,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9622601270675659,
      "kl": 0.0009502729969881329,
      "learning_rate": 3.7037037037037036e-08,
      "loss": 0.0854,
      "num_tokens": 1886250.0,
      "reward": -0.5533753037452698,
      "reward_std": 1.0693888664245605,
      "rewards/reward_model/mean": -0.5533753037452698,
      "rewards/reward_model/std": 1.3799840211868286,
      "step": 4,
      "step_time": 167.94514833204448
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.47802734375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 88.8896484375,
      "completions/mean_terminated_length": 53.07202911376953,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.164030898362398,
      "epoch": 0.002059308072487644,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0810271501541138,
      "kl": 0.0010240612220968615,
      "learning_rate": 4.9382716049382724e-08,
      "loss": 0.0979,
      "num_tokens": 2372616.0,
      "reward": -0.8290466070175171,
      "reward_std": 1.1383775472640991,
      "rewards/reward_model/mean": -0.8290466070175171,
      "rewards/reward_model/std": 1.4821057319641113,
      "step": 5,
      "step_time": 168.5208105482161
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.44775390625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 86.54541015625,
      "completions/mean_terminated_length": 52.934574127197266,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.121390470303595,
      "epoch": 0.002471169686985173,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.026131272315979,
      "kl": 0.000956788239591333,
      "learning_rate": 6.17283950617284e-08,
      "loss": 0.102,
      "num_tokens": 2856453.0,
      "reward": -0.5948619842529297,
      "reward_std": 1.0859686136245728,
      "rewards/reward_model/mean": -0.5948619842529297,
      "rewards/reward_model/std": 1.4433753490447998,
      "step": 6,
      "step_time": 169.30755526619032
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.47998046875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 88.40185546875,
      "completions/mean_terminated_length": 51.85258483886719,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.1293580746278167,
      "epoch": 0.002883031301482702,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9581882953643799,
      "kl": 0.0010509827170608332,
      "learning_rate": 7.407407407407407e-08,
      "loss": 0.0752,
      "num_tokens": 3349660.0,
      "reward": -0.8746315836906433,
      "reward_std": 1.1371493339538574,
      "rewards/reward_model/mean": -0.8746315836906433,
      "rewards/reward_model/std": 1.5432283878326416,
      "step": 7,
      "step_time": 170.4541406123899
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.462890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 89.59423828125,
      "completions/mean_terminated_length": 56.49545669555664,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.199626039713621,
      "epoch": 0.0032948929159802307,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.1233900785446167,
      "kl": 0.0011659049921490805,
      "learning_rate": 8.641975308641976e-08,
      "loss": 0.0835,
      "num_tokens": 3869181.0,
      "reward": -0.9943232536315918,
      "reward_std": 1.099515438079834,
      "rewards/reward_model/mean": -0.9943232536315918,
      "rewards/reward_model/std": 1.4042030572891235,
      "step": 8,
      "step_time": 168.8292339304462
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.50244140625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 92.26611328125,
      "completions/mean_terminated_length": 56.181549072265625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 2.103476638905704,
      "epoch": 0.0037067545304777594,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0071220397949219,
      "kl": 0.0010218678287401417,
      "learning_rate": 9.876543209876545e-08,
      "loss": 0.0786,
      "num_tokens": 4330526.0,
      "reward": -0.7287623286247253,
      "reward_std": 1.2205724716186523,
      "rewards/reward_model/mean": -0.7287623286247253,
      "rewards/reward_model/std": 1.5410621166229248,
      "step": 9,
      "step_time": 168.4490856071934
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.4384765625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 87.068359375,
      "completions/mean_terminated_length": 55.10608673095703,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 2.068316952791065,
      "epoch": 0.004118616144975288,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0786375999450684,
      "kl": 0.0011272716699295415,
      "learning_rate": 1.1111111111111111e-07,
      "loss": 0.0808,
      "num_tokens": 4813482.0,
      "reward": -0.8588310480117798,
      "reward_std": 1.1204930543899536,
      "rewards/reward_model/mean": -0.8588310480117798,
      "rewards/reward_model/std": 1.4020955562591553,
      "step": 10,
      "step_time": 169.3498973324895
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.486328125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 88.27197265625,
      "completions/mean_terminated_length": 50.65874481201172,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.085946503095329,
      "epoch": 0.004530477759472817,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.1249303817749023,
      "kl": 0.0010357791773003555,
      "learning_rate": 1.234567901234568e-07,
      "loss": 0.103,
      "num_tokens": 5276279.0,
      "reward": -0.7370425462722778,
      "reward_std": 1.1393404006958008,
      "rewards/reward_model/mean": -0.7370425462722778,
      "rewards/reward_model/std": 1.435203194618225,
      "step": 11,
      "step_time": 169.61693120608106
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.435546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 84.62060546875,
      "completions/mean_terminated_length": 51.147926330566406,
      "completions/min_length": 1.0,
      "completions/min_terminated_length": 1.0,
      "entropy": 2.0578739237971604,
      "epoch": 0.004942339373970346,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.1678587198257446,
      "kl": 0.0010205526389199804,
      "learning_rate": 1.3580246913580248e-07,
      "loss": 0.129,
      "num_tokens": 5750318.0,
      "reward": -0.6621623039245605,
      "reward_std": 1.1341545581817627,
      "rewards/reward_model/mean": -0.6621623039245605,
      "rewards/reward_model/std": 1.4956636428833008,
      "step": 12,
      "step_time": 170.6942683076486
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.48876953125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 89.955078125,
      "completions/mean_terminated_length": 53.581661224365234,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 2.1759862853214145,
      "epoch": 0.005354200988467875,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9353419542312622,
      "kl": 0.0009984489861381007,
      "learning_rate": 1.4814814814814815e-07,
      "loss": 0.0706,
      "num_tokens": 6237106.0,
      "reward": -0.71650230884552,
      "reward_std": 1.1081366539001465,
      "rewards/reward_model/mean": -0.71650230884552,
      "rewards/reward_model/std": 1.4882901906967163,
      "step": 13,
      "step_time": 168.60461562033743
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.47265625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 127.0,
      "completions/mean_length": 90.6298828125,
      "completions/mean_terminated_length": 57.13518524169922,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.1225685542449355,
      "epoch": 0.005766062602965404,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9926177263259888,
      "kl": 0.0010822901392657513,
      "learning_rate": 1.6049382716049383e-07,
      "loss": 0.0705,
      "num_tokens": 6768988.0,
      "reward": -0.8033103346824646,
      "reward_std": 1.1658474206924438,
      "rewards/reward_model/mean": -0.8033103346824646,
      "rewards/reward_model/std": 1.5343424081802368,
      "step": 14,
      "step_time": 169.76986178942025
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.43359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 87.02978515625,
      "completions/mean_terminated_length": 55.666378021240234,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.2208039346151054,
      "epoch": 0.006177924217462933,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.110655426979065,
      "kl": 0.00102766218378747,
      "learning_rate": 1.7283950617283952e-07,
      "loss": 0.1137,
      "num_tokens": 7264761.0,
      "reward": -0.8211149573326111,
      "reward_std": 1.1067304611206055,
      "rewards/reward_model/mean": -0.8211149573326111,
      "rewards/reward_model/std": 1.4263983964920044,
      "step": 15,
      "step_time": 169.37415388552472
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.447265625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 89.8525390625,
      "completions/mean_terminated_length": 58.984100341796875,
      "completions/min_length": 1.0,
      "completions/min_terminated_length": 1.0,
      "entropy": 2.036483039613813,
      "epoch": 0.006589785831960461,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0956333875656128,
      "kl": 0.0011385848249574337,
      "learning_rate": 1.8518518518518518e-07,
      "loss": 0.0679,
      "num_tokens": 7745675.0,
      "reward": -0.5313577651977539,
      "reward_std": 1.1804759502410889,
      "rewards/reward_model/mean": -0.5313577651977539,
      "rewards/reward_model/std": 1.5051146745681763,
      "step": 16,
      "step_time": 168.84541190741584
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.46142578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 87.02197265625,
      "completions/mean_terminated_length": 51.91387176513672,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 2.1997494087554514,
      "epoch": 0.00700164744645799,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0221517086029053,
      "kl": 0.0010186115512169636,
      "learning_rate": 1.975308641975309e-07,
      "loss": 0.0962,
      "num_tokens": 8243704.0,
      "reward": -0.89983731508255,
      "reward_std": 1.135831356048584,
      "rewards/reward_model/mean": -0.89983731508255,
      "rewards/reward_model/std": 1.4320958852767944,
      "step": 17,
      "step_time": 168.78324813907966
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.45947265625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 88.05908203125,
      "completions/mean_terminated_length": 54.10749816894531,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.176318216137588,
      "epoch": 0.007413509060955519,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0268303155899048,
      "kl": 0.0010050315393073106,
      "learning_rate": 2.0987654320987656e-07,
      "loss": 0.0845,
      "num_tokens": 8726801.0,
      "reward": -0.7434755563735962,
      "reward_std": 1.1786913871765137,
      "rewards/reward_model/mean": -0.7434755563735962,
      "rewards/reward_model/std": 1.4701310396194458,
      "step": 18,
      "step_time": 168.51915573468432
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.4658203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 88.00439453125,
      "completions/mean_terminated_length": 53.12705993652344,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0752989565953612,
      "epoch": 0.007825370675453048,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0904563665390015,
      "kl": 0.00123250130627639,
      "learning_rate": 2.2222222222222222e-07,
      "loss": 0.0885,
      "num_tokens": 9180858.0,
      "reward": -0.8568893074989319,
      "reward_std": 1.1963412761688232,
      "rewards/reward_model/mean": -0.8568893074989319,
      "rewards/reward_model/std": 1.5186042785644531,
      "step": 19,
      "step_time": 170.01141701499
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.4033203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 83.22021484375,
      "completions/mean_terminated_length": 52.95172119140625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 2.200795284938067,
      "epoch": 0.008237232289950576,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0073621273040771,
      "kl": 0.0012237756848207937,
      "learning_rate": 2.3456790123456793e-07,
      "loss": 0.0782,
      "num_tokens": 9702557.0,
      "reward": -0.9474191069602966,
      "reward_std": 1.101952314376831,
      "rewards/reward_model/mean": -0.9474191665649414,
      "rewards/reward_model/std": 1.514784336090088,
      "step": 20,
      "step_time": 168.7904914407991
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.44189453125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 127.0,
      "completions/mean_length": 85.07275390625,
      "completions/mean_terminated_length": 51.08399200439453,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.1343746068887413,
      "epoch": 0.008649093904448105,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.1044775247573853,
      "kl": 0.0011666002192214364,
      "learning_rate": 2.469135802469136e-07,
      "loss": 0.0659,
      "num_tokens": 10182002.0,
      "reward": -0.8981258869171143,
      "reward_std": 1.1897304058074951,
      "rewards/reward_model/mean": -0.8981258869171143,
      "rewards/reward_model/std": 1.4881244897842407,
      "step": 21,
      "step_time": 168.61277754418552
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.4580078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 88.984375,
      "completions/mean_terminated_length": 56.0144157409668,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.060287212021649,
      "epoch": 0.009060955518945634,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9674479961395264,
      "kl": 0.0012853052542141086,
      "learning_rate": 2.5925925925925923e-07,
      "loss": 0.0795,
      "num_tokens": 10599858.0,
      "reward": -0.7459607720375061,
      "reward_std": 1.18560791015625,
      "rewards/reward_model/mean": -0.7459607720375061,
      "rewards/reward_model/std": 1.4447804689407349,
      "step": 22,
      "step_time": 168.57235636515543
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.45751953125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 87.20947265625,
      "completions/mean_terminated_length": 52.80738067626953,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.9992840560153127,
      "epoch": 0.009472817133443162,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.2746193408966064,
      "kl": 0.001352380840899059,
      "learning_rate": 2.7160493827160497e-07,
      "loss": 0.0805,
      "num_tokens": 11135295.0,
      "reward": -0.9941644668579102,
      "reward_std": 1.2033442258834839,
      "rewards/reward_model/mean": -0.9941644668579102,
      "rewards/reward_model/std": 1.5118839740753174,
      "step": 23,
      "step_time": 168.93097670795396
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.458984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 88.12744140625,
      "completions/mean_terminated_length": 54.300540924072266,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 2.1036753226071596,
      "epoch": 0.009884678747940691,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9719477295875549,
      "kl": 0.0014479562626092957,
      "learning_rate": 2.839506172839506e-07,
      "loss": 0.0792,
      "num_tokens": 11647428.0,
      "reward": -0.7246302366256714,
      "reward_std": 1.1223700046539307,
      "rewards/reward_model/mean": -0.7246302366256714,
      "rewards/reward_model/std": 1.4486252069473267,
      "step": 24,
      "step_time": 168.1220847275108
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.48876953125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 92.43017578125,
      "completions/mean_terminated_length": 58.42311477661133,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0290252747945487,
      "epoch": 0.01029654036243822,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.036043643951416,
      "kl": 0.0014065650616430503,
      "learning_rate": 2.962962962962963e-07,
      "loss": 0.0464,
      "num_tokens": 12175925.0,
      "reward": -0.8139803409576416,
      "reward_std": 1.18918776512146,
      "rewards/reward_model/mean": -0.8139803409576416,
      "rewards/reward_model/std": 1.5184983015060425,
      "step": 25,
      "step_time": 169.08092289417982
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.48193359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 91.38720703125,
      "completions/mean_terminated_length": 57.3279914855957,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0865674833767116,
      "epoch": 0.01070840197693575,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9221011996269226,
      "kl": 0.0016496309149260924,
      "learning_rate": 3.08641975308642e-07,
      "loss": 0.0459,
      "num_tokens": 12671022.0,
      "reward": -0.6815944910049438,
      "reward_std": 1.1987043619155884,
      "rewards/reward_model/mean": -0.6815944910049438,
      "rewards/reward_model/std": 1.503211259841919,
      "step": 26,
      "step_time": 169.66068721655756
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.39208984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.84814453125,
      "completions/mean_terminated_length": 53.726104736328125,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.100129804573953,
      "epoch": 0.011120263591433279,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0829551219940186,
      "kl": 0.0020542718434626295,
      "learning_rate": 3.2098765432098767e-07,
      "loss": 0.1111,
      "num_tokens": 13159479.0,
      "reward": -0.7841147780418396,
      "reward_std": 1.1083781719207764,
      "rewards/reward_model/mean": -0.7841147780418396,
      "rewards/reward_model/std": 1.398116946220398,
      "step": 27,
      "step_time": 170.63890342088416
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.41845703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 84.29150390625,
      "completions/mean_terminated_length": 52.84046936035156,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0639419481158257,
      "epoch": 0.011532125205930808,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9612188935279846,
      "kl": 0.002871143702122936,
      "learning_rate": 3.333333333333333e-07,
      "loss": 0.0511,
      "num_tokens": 13624972.0,
      "reward": -0.41133514046669006,
      "reward_std": 1.0870225429534912,
      "rewards/reward_model/mean": -0.41133514046669006,
      "rewards/reward_model/std": 1.3928031921386719,
      "step": 28,
      "step_time": 169.01845826301724
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3896484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.62353515625,
      "completions/mean_terminated_length": 53.65519714355469,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0633218064904213,
      "epoch": 0.011943986820428337,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0220043659210205,
      "kl": 0.004001651409907936,
      "learning_rate": 3.4567901234567904e-07,
      "loss": 0.0608,
      "num_tokens": 14084265.0,
      "reward": -0.5280731916427612,
      "reward_std": 1.139591097831726,
      "rewards/reward_model/mean": -0.5280731916427612,
      "rewards/reward_model/std": 1.5284217596054077,
      "step": 29,
      "step_time": 169.68904952565208
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3447265625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.75439453125,
      "completions/mean_terminated_length": 51.321163177490234,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.053064794279635,
      "epoch": 0.012355848434925865,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0159528255462646,
      "kl": 0.004774259470650577,
      "learning_rate": 3.580246913580247e-07,
      "loss": 0.0826,
      "num_tokens": 14545778.0,
      "reward": -0.8308598399162292,
      "reward_std": 1.1439062356948853,
      "rewards/reward_model/mean": -0.8308598399162292,
      "rewards/reward_model/std": 1.4677071571350098,
      "step": 30,
      "step_time": 169.4162016301416
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.40478515625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 84.13525390625,
      "completions/mean_terminated_length": 54.30434799194336,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.9778149635531008,
      "epoch": 0.012767710049423394,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9643027186393738,
      "kl": 0.004632160428627685,
      "learning_rate": 3.7037037037037036e-07,
      "loss": 0.0324,
      "num_tokens": 15021479.0,
      "reward": -0.5928993225097656,
      "reward_std": 1.0915915966033936,
      "rewards/reward_model/mean": -0.5928993225097656,
      "rewards/reward_model/std": 1.4171936511993408,
      "step": 31,
      "step_time": 169.55369784962386
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.39404296875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.93603515625,
      "completions/mean_terminated_length": 51.981468200683594,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.9114997563883662,
      "epoch": 0.013179571663920923,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9297622442245483,
      "kl": 0.005185256256481807,
      "learning_rate": 3.8271604938271605e-07,
      "loss": 0.0641,
      "num_tokens": 15490468.0,
      "reward": -0.4294321537017822,
      "reward_std": 1.1095049381256104,
      "rewards/reward_model/mean": -0.4294321537017822,
      "rewards/reward_model/std": 1.4001518487930298,
      "step": 32,
      "step_time": 169.97963417787105
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.4033203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 84.73388671875,
      "completions/mean_terminated_length": 55.48854446411133,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.9920633286237717,
      "epoch": 0.013591433278418451,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9739150404930115,
      "kl": 0.0050489629365984,
      "learning_rate": 3.950617283950618e-07,
      "loss": 0.069,
      "num_tokens": 16033027.0,
      "reward": -0.5853164792060852,
      "reward_std": 1.1397128105163574,
      "rewards/reward_model/mean": -0.5853164792060852,
      "rewards/reward_model/std": 1.4342437982559204,
      "step": 33,
      "step_time": 170.0918092643842
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.33349609375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.16943359375,
      "completions/mean_terminated_length": 56.236629486083984,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.052212963812053,
      "epoch": 0.01400329489291598,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0108827352523804,
      "kl": 0.0070345894837373635,
      "learning_rate": 4.0740740740740737e-07,
      "loss": 0.0717,
      "num_tokens": 16443422.0,
      "reward": -0.40320760011672974,
      "reward_std": 1.023691177368164,
      "rewards/reward_model/mean": -0.40320760011672974,
      "rewards/reward_model/std": 1.3064631223678589,
      "step": 34,
      "step_time": 168.22575595136732
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3896484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.1884765625,
      "completions/mean_terminated_length": 52.94239807128906,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.8943170690909028,
      "epoch": 0.014415156507413509,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9463284015655518,
      "kl": 0.006503033908302314,
      "learning_rate": 4.197530864197531e-07,
      "loss": 0.0771,
      "num_tokens": 16958848.0,
      "reward": -0.46641844511032104,
      "reward_std": 1.1392958164215088,
      "rewards/reward_model/mean": -0.46641844511032104,
      "rewards/reward_model/std": 1.3904635906219482,
      "step": 35,
      "step_time": 169.57979472074658
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.2724609375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 69.291015625,
      "completions/mean_terminated_length": 47.3046989440918,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 2.0284548006020486,
      "epoch": 0.014827018121911038,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.202287197113037,
      "kl": 0.008711300118193321,
      "learning_rate": 4.320987654320988e-07,
      "loss": 0.0853,
      "num_tokens": 17445812.0,
      "reward": -0.5143425464630127,
      "reward_std": 1.080782175064087,
      "rewards/reward_model/mean": -0.5143425464630127,
      "rewards/reward_model/std": 1.3849540948867798,
      "step": 36,
      "step_time": 169.69654387421906
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3369140625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.77197265625,
      "completions/mean_terminated_length": 52.25110626220703,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.8757861303165555,
      "epoch": 0.015238879736408566,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8937506675720215,
      "kl": 0.010500569681425986,
      "learning_rate": 4.4444444444444444e-07,
      "loss": 0.0466,
      "num_tokens": 17886785.0,
      "reward": -0.2941930890083313,
      "reward_std": 1.089874267578125,
      "rewards/reward_model/mean": -0.2941930890083313,
      "rewards/reward_model/std": 1.3422448635101318,
      "step": 37,
      "step_time": 168.3831845112145
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.33203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.78515625,
      "completions/mean_terminated_length": 54.32163619995117,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.8339524874463677,
      "epoch": 0.015650741350906095,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8777432441711426,
      "kl": 0.014370948238138226,
      "learning_rate": 4.567901234567901e-07,
      "loss": 0.0431,
      "num_tokens": 18363593.0,
      "reward": -0.21549299359321594,
      "reward_std": 1.0654486417770386,
      "rewards/reward_model/mean": -0.21549299359321594,
      "rewards/reward_model/std": 1.286303997039795,
      "step": 38,
      "step_time": 168.6560257449746
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3623046875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.3076171875,
      "completions/mean_terminated_length": 53.211334228515625,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.809513804037124,
      "epoch": 0.016062602965403624,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.850346028804779,
      "kl": 0.019628787720648688,
      "learning_rate": 4.6913580246913586e-07,
      "loss": -0.0144,
      "num_tokens": 18782015.0,
      "reward": -0.19260446727275848,
      "reward_std": 1.0799050331115723,
      "rewards/reward_model/mean": -0.19260446727275848,
      "rewards/reward_model/std": 1.4198755025863647,
      "step": 39,
      "step_time": 169.06077374424785
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3212890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.080078125,
      "completions/mean_terminated_length": 54.44892120361328,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.856378594879061,
      "epoch": 0.016474464579901153,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8225836753845215,
      "kl": 0.023546985856228275,
      "learning_rate": 4.814814814814815e-07,
      "loss": 0.0268,
      "num_tokens": 19238691.0,
      "reward": -0.22110876441001892,
      "reward_std": 1.0441968441009521,
      "rewards/reward_model/mean": -0.22110876441001892,
      "rewards/reward_model/std": 1.3271934986114502,
      "step": 40,
      "step_time": 169.5594472438097
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.33251953125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.1416015625,
      "completions/mean_terminated_length": 51.805416107177734,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.764324402436614,
      "epoch": 0.01688632619439868,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7773052453994751,
      "kl": 0.029280464848852716,
      "learning_rate": 4.938271604938272e-07,
      "loss": 0.0194,
      "num_tokens": 19755301.0,
      "reward": -0.12512998282909393,
      "reward_std": 1.0090844631195068,
      "rewards/reward_model/mean": -0.12512998282909393,
      "rewards/reward_model/std": 1.2345008850097656,
      "step": 41,
      "step_time": 170.39410974271595
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.27490234375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 72.548828125,
      "completions/mean_terminated_length": 51.52592468261719,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.758555585052818,
      "epoch": 0.01729818780889621,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8362958431243896,
      "kl": 0.03495379232481355,
      "learning_rate": 5.061728395061729e-07,
      "loss": 0.0014,
      "num_tokens": 20199209.0,
      "reward": -0.034443896263837814,
      "reward_std": 1.0466477870941162,
      "rewards/reward_model/mean": -0.034443896263837814,
      "rewards/reward_model/std": 1.2755711078643799,
      "step": 42,
      "step_time": 170.17393092392012
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.25830078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 74.35302734375,
      "completions/mean_terminated_length": 55.6701774597168,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.6830066749826074,
      "epoch": 0.01771004942339374,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9207571148872375,
      "kl": 0.03826815243519377,
      "learning_rate": 5.185185185185185e-07,
      "loss": 0.0197,
      "num_tokens": 20667900.0,
      "reward": -0.03724297881126404,
      "reward_std": 0.9730924367904663,
      "rewards/reward_model/mean": -0.03724297881126404,
      "rewards/reward_model/std": 1.1648329496383667,
      "step": 43,
      "step_time": 168.8539799619466
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3427734375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.59326171875,
      "completions/mean_terminated_length": 58.911590576171875,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.722061135340482,
      "epoch": 0.018121911037891267,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.691871166229248,
      "kl": 0.035297417802212294,
      "learning_rate": 5.308641975308642e-07,
      "loss": 0.0252,
      "num_tokens": 21084443.0,
      "reward": 0.1364922821521759,
      "reward_std": 0.9992862939834595,
      "rewards/reward_model/mean": 0.1364922821521759,
      "rewards/reward_model/std": 1.338813066482544,
      "step": 44,
      "step_time": 168.30778062017635
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.2216796875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 70.23193359375,
      "completions/mean_terminated_length": 53.77854537963867,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.6960708745755255,
      "epoch": 0.018533772652388796,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8769946694374084,
      "kl": 0.04808991262325435,
      "learning_rate": 5.432098765432099e-07,
      "loss": -0.0016,
      "num_tokens": 21510710.0,
      "reward": 0.1898983120918274,
      "reward_std": 0.9757044911384583,
      "rewards/reward_model/mean": 0.1898983120918274,
      "rewards/reward_model/std": 1.1677379608154297,
      "step": 45,
      "step_time": 169.03864477854222
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.28125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.9912109375,
      "completions/mean_terminated_length": 58.42255401611328,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.6502066934481263,
      "epoch": 0.018945634266886325,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6699737906455994,
      "kl": 0.04364871226789546,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.0298,
      "num_tokens": 21968036.0,
      "reward": 0.14871619641780853,
      "reward_std": 0.8983126878738403,
      "rewards/reward_model/mean": 0.14871619641780853,
      "rewards/reward_model/std": 1.1425597667694092,
      "step": 46,
      "step_time": 169.17142802104354
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.27294921875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 73.318359375,
      "completions/mean_terminated_length": 52.789791107177734,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.806841244455427,
      "epoch": 0.019357495881383854,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7548915147781372,
      "kl": 0.050439021695638075,
      "learning_rate": 5.679012345679012e-07,
      "loss": 0.0261,
      "num_tokens": 22428752.0,
      "reward": 0.18833398818969727,
      "reward_std": 0.9490935802459717,
      "rewards/reward_model/mean": 0.18833398818969727,
      "rewards/reward_model/std": 1.218595027923584,
      "step": 47,
      "step_time": 169.8053262718022
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.23583984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 67.080078125,
      "completions/mean_terminated_length": 48.278594970703125,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.7256725700572133,
      "epoch": 0.019769357495881382,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8408772349357605,
      "kl": 0.05761563615669729,
      "learning_rate": 5.80246913580247e-07,
      "loss": -0.0234,
      "num_tokens": 22883444.0,
      "reward": 0.10819900035858154,
      "reward_std": 0.9136756062507629,
      "rewards/reward_model/mean": 0.10819900035858154,
      "rewards/reward_model/std": 1.13023042678833,
      "step": 48,
      "step_time": 170.28967663506046
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.25927734375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 71.54296875,
      "completions/mean_terminated_length": 51.78114700317383,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.676765794865787,
      "epoch": 0.02018121911037891,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6633880138397217,
      "kl": 0.05726497815339826,
      "learning_rate": 5.925925925925926e-07,
      "loss": 0.0077,
      "num_tokens": 23311980.0,
      "reward": 0.31039929389953613,
      "reward_std": 0.8826955556869507,
      "rewards/reward_model/mean": 0.31039929389953613,
      "rewards/reward_model/std": 1.1599924564361572,
      "step": 49,
      "step_time": 169.17084869695827
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.2392578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 73.46142578125,
      "completions/mean_terminated_length": 56.30873107910156,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.5921450154855847,
      "epoch": 0.02059308072487644,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7272253632545471,
      "kl": 0.05662718684470747,
      "learning_rate": 6.049382716049383e-07,
      "loss": -0.0038,
      "num_tokens": 23729245.0,
      "reward": 0.2335912585258484,
      "reward_std": 0.9175702929496765,
      "rewards/reward_model/mean": 0.2335912585258484,
      "rewards/reward_model/std": 1.1314274072647095,
      "step": 50,
      "step_time": 169.32695539435372
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.26220703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.47705078125,
      "completions/mean_terminated_length": 59.521507263183594,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.630979296285659,
      "epoch": 0.021004942339373972,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7209051847457886,
      "kl": 0.0596031873501488,
      "learning_rate": 6.17283950617284e-07,
      "loss": -0.019,
      "num_tokens": 24171054.0,
      "reward": 0.3881710171699524,
      "reward_std": 0.9779696464538574,
      "rewards/reward_model/mean": 0.3881710171699524,
      "rewards/reward_model/std": 1.2501736879348755,
      "step": 51,
      "step_time": 169.39474018104374
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3837890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 87.35888671875,
      "completions/mean_terminated_length": 62.0467529296875,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.5277671799995005,
      "epoch": 0.0214168039538715,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6657931208610535,
      "kl": 0.04347534721819102,
      "learning_rate": 6.296296296296296e-07,
      "loss": 0.0084,
      "num_tokens": 24640845.0,
      "reward": 0.40088099241256714,
      "reward_std": 0.8522671461105347,
      "rewards/reward_model/mean": 0.40088099241256714,
      "rewards/reward_model/std": 1.1760755777359009,
      "step": 52,
      "step_time": 169.99416326358914
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.26708984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.86572265625,
      "completions/mean_terminated_length": 58.23118209838867,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.5355965252965689,
      "epoch": 0.02182866556836903,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7246338725090027,
      "kl": 0.059198625254794024,
      "learning_rate": 6.419753086419753e-07,
      "loss": 0.0065,
      "num_tokens": 25146234.0,
      "reward": 0.32493141293525696,
      "reward_std": 0.8951080441474915,
      "rewards/reward_model/mean": 0.32493141293525696,
      "rewards/reward_model/std": 1.109892725944519,
      "step": 53,
      "step_time": 170.4134237067774
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.26220703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.32373046875,
      "completions/mean_terminated_length": 59.313697814941406,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.4214782847557217,
      "epoch": 0.022240527182866558,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.663506805896759,
      "kl": 0.07469483163731638,
      "learning_rate": 6.54320987654321e-07,
      "loss": -0.0106,
      "num_tokens": 25556273.0,
      "reward": 0.5579333305358887,
      "reward_std": 0.8257571458816528,
      "rewards/reward_model/mean": 0.5579333305358887,
      "rewards/reward_model/std": 1.0652962923049927,
      "step": 54,
      "step_time": 168.6097109238617
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.2158203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 67.8369140625,
      "completions/mean_terminated_length": 51.278953552246094,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.4409256265498698,
      "epoch": 0.022652388797364087,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6998105645179749,
      "kl": 0.09127819760760758,
      "learning_rate": 6.666666666666666e-07,
      "loss": -0.0145,
      "num_tokens": 26009315.0,
      "reward": 0.36190831661224365,
      "reward_std": 0.8500241637229919,
      "rewards/reward_model/mean": 0.36190831661224365,
      "rewards/reward_model/std": 1.0737853050231934,
      "step": 55,
      "step_time": 169.63903413154185
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.2265625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 71.69873046875,
      "completions/mean_terminated_length": 55.20643997192383,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.4477170635946095,
      "epoch": 0.023064250411861616,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6856215596199036,
      "kl": 0.10684883118665311,
      "learning_rate": 6.790123456790124e-07,
      "loss": -0.0154,
      "num_tokens": 26453082.0,
      "reward": 0.5452687740325928,
      "reward_std": 0.7654911875724792,
      "rewards/reward_model/mean": 0.5452687740325928,
      "rewards/reward_model/std": 0.9965056777000427,
      "step": 56,
      "step_time": 167.3982848683372
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.31689453125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.6259765625,
      "completions/mean_terminated_length": 58.6490364074707,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.3876500492915511,
      "epoch": 0.023476112026359144,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5802730321884155,
      "kl": 0.08226650860888185,
      "learning_rate": 6.913580246913581e-07,
      "loss": -0.0124,
      "num_tokens": 26859036.0,
      "reward": 0.4985049366950989,
      "reward_std": 0.8241320252418518,
      "rewards/reward_model/mean": 0.4985049366950989,
      "rewards/reward_model/std": 1.177066683769226,
      "step": 57,
      "step_time": 168.99145932588726
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.29150390625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.41162109375,
      "completions/mean_terminated_length": 62.24327850341797,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.3736186842434108,
      "epoch": 0.023887973640856673,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.672706127166748,
      "kl": 0.0943700206480571,
      "learning_rate": 7.037037037037037e-07,
      "loss": 0.0084,
      "num_tokens": 27318439.0,
      "reward": 0.7748833894729614,
      "reward_std": 0.8400471806526184,
      "rewards/reward_model/mean": 0.7748833894729614,
      "rewards/reward_model/std": 1.0984324216842651,
      "step": 58,
      "step_time": 168.8087218273431
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.30419921875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.111328125,
      "completions/mean_terminated_length": 57.737545013427734,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.3578353270422667,
      "epoch": 0.024299835255354202,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6414256691932678,
      "kl": 0.10844983752758708,
      "learning_rate": 7.160493827160494e-07,
      "loss": 0.0049,
      "num_tokens": 27845483.0,
      "reward": 0.6581840515136719,
      "reward_std": 0.7642059326171875,
      "rewards/reward_model/mean": 0.6581840515136719,
      "rewards/reward_model/std": 1.0230196714401245,
      "step": 59,
      "step_time": 170.09878712054342
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.20947265625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 68.9013671875,
      "completions/mean_terminated_length": 53.24150848388672,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.359937054105103,
      "epoch": 0.02471169686985173,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6327216029167175,
      "kl": 0.12389619748864789,
      "learning_rate": 7.283950617283951e-07,
      "loss": -0.0019,
      "num_tokens": 28360609.0,
      "reward": 0.5726691484451294,
      "reward_std": 0.7263065576553345,
      "rewards/reward_model/mean": 0.5726691484451294,
      "rewards/reward_model/std": 1.0532201528549194,
      "step": 60,
      "step_time": 169.033332105726
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.21728515625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 74.28515625,
      "completions/mean_terminated_length": 59.37367248535156,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.2873307822737843,
      "epoch": 0.02512355848434926,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6471663117408752,
      "kl": 0.11947597128164489,
      "learning_rate": 7.407407407407407e-07,
      "loss": -0.018,
      "num_tokens": 28796649.0,
      "reward": 0.8057171106338501,
      "reward_std": 0.6930927038192749,
      "rewards/reward_model/mean": 0.8057171106338501,
      "rewards/reward_model/std": 0.9504708647727966,
      "step": 61,
      "step_time": 170.3676045727916
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.27880859375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.5634765625,
      "completions/mean_terminated_length": 62.22477722167969,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.4397811936214566,
      "epoch": 0.025535420098846788,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5935730934143066,
      "kl": 0.11211827301303856,
      "learning_rate": 7.530864197530865e-07,
      "loss": 0.005,
      "num_tokens": 29272715.0,
      "reward": 0.5376583337783813,
      "reward_std": 0.7316970825195312,
      "rewards/reward_model/mean": 0.5376583337783813,
      "rewards/reward_model/std": 1.0817116498947144,
      "step": 62,
      "step_time": 170.07809142861515
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.27978515625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 83.00634765625,
      "completions/mean_terminated_length": 65.52745819091797,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.2591755213215947,
      "epoch": 0.025947281713344317,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6332337260246277,
      "kl": 0.09935169246455189,
      "learning_rate": 7.654320987654321e-07,
      "loss": -0.0047,
      "num_tokens": 29675416.0,
      "reward": 0.8634133338928223,
      "reward_std": 0.7280638217926025,
      "rewards/reward_model/mean": 0.8634133338928223,
      "rewards/reward_model/std": 1.0552853345870972,
      "step": 63,
      "step_time": 169.48511258373037
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.234375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.23291015625,
      "completions/mean_terminated_length": 64.30420684814453,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.3228048181626946,
      "epoch": 0.026359143327841845,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5896514058113098,
      "kl": 0.10629570209130179,
      "learning_rate": 7.777777777777778e-07,
      "loss": 0.0018,
      "num_tokens": 30103349.0,
      "reward": 0.7413797378540039,
      "reward_std": 0.6787456274032593,
      "rewards/reward_model/mean": 0.7413797378540039,
      "rewards/reward_model/std": 0.9844362735748291,
      "step": 64,
      "step_time": 168.49784950073808
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.2080078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.34521484375,
      "completions/mean_terminated_length": 64.04130554199219,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.3859816826879978,
      "epoch": 0.026771004942339374,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.569113552570343,
      "kl": 0.1355650291661732,
      "learning_rate": 7.901234567901236e-07,
      "loss": 0.0008,
      "num_tokens": 30598360.0,
      "reward": 0.8138879537582397,
      "reward_std": 0.6921124458312988,
      "rewards/reward_model/mean": 0.8138879537582397,
      "rewards/reward_model/std": 1.008180856704712,
      "step": 65,
      "step_time": 168.94378049625084
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.2333984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.275390625,
      "completions/mean_terminated_length": 60.52738952636719,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.276806804118678,
      "epoch": 0.027182866556836903,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.584193229675293,
      "kl": 0.12889757197990548,
      "learning_rate": 8.024691358024692e-07,
      "loss": 0.0109,
      "num_tokens": 31028524.0,
      "reward": 0.7695643901824951,
      "reward_std": 0.7420451641082764,
      "rewards/reward_model/mean": 0.7695643901824951,
      "rewards/reward_model/std": 1.1982769966125488,
      "step": 66,
      "step_time": 169.0462037078105
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.2314453125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.75390625,
      "completions/mean_terminated_length": 67.82718658447266,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.2658436398487538,
      "epoch": 0.02759472817133443,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5793888568878174,
      "kl": 0.10512553945591208,
      "learning_rate": 8.148148148148147e-07,
      "loss": 0.019,
      "num_tokens": 31459860.0,
      "reward": 0.9362199306488037,
      "reward_std": 0.6280190944671631,
      "rewards/reward_model/mean": 0.9362199306488037,
      "rewards/reward_model/std": 1.003322958946228,
      "step": 67,
      "step_time": 168.24778978247195
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.20068359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 75.294921875,
      "completions/mean_terminated_length": 62.06230926513672,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.3218755372799933,
      "epoch": 0.02800658978583196,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.619716465473175,
      "kl": 0.13950362912146375,
      "learning_rate": 8.271604938271605e-07,
      "loss": 0.0032,
      "num_tokens": 31900336.0,
      "reward": 0.7856715321540833,
      "reward_std": 0.6523309946060181,
      "rewards/reward_model/mean": 0.7856715321540833,
      "rewards/reward_model/std": 0.9243690371513367,
      "step": 68,
      "step_time": 168.62237379932776
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 72.62060546875,
      "completions/mean_terminated_length": 59.84074783325195,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.2817874399479479,
      "epoch": 0.02841845140032949,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6988398432731628,
      "kl": 0.1405531533237081,
      "learning_rate": 8.395061728395062e-07,
      "loss": 0.0001,
      "num_tokens": 32349991.0,
      "reward": 0.7539228200912476,
      "reward_std": 0.6927404403686523,
      "rewards/reward_model/mean": 0.7539228200912476,
      "rewards/reward_model/std": 1.1138005256652832,
      "step": 69,
      "step_time": 168.95463426411152
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.18310546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.68505859375,
      "completions/mean_terminated_length": 65.18290710449219,
      "completions/min_length": 1.0,
      "completions/min_terminated_length": 1.0,
      "entropy": 1.3052547052502632,
      "epoch": 0.028830313014827018,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5859233140945435,
      "kl": 0.14809596522536594,
      "learning_rate": 8.518518518518518e-07,
      "loss": 0.006,
      "num_tokens": 32799778.0,
      "reward": 0.899767279624939,
      "reward_std": 0.6600509881973267,
      "rewards/reward_model/mean": 0.899767279624939,
      "rewards/reward_model/std": 1.0378800630569458,
      "step": 70,
      "step_time": 168.27934673754498
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.21435546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.1396484375,
      "completions/mean_terminated_length": 63.26289749145508,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.2957828119397163,
      "epoch": 0.029242174629324547,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6063217520713806,
      "kl": 0.1499464159278432,
      "learning_rate": 8.641975308641976e-07,
      "loss": 0.0009,
      "num_tokens": 33258528.0,
      "reward": 0.9532963037490845,
      "reward_std": 0.5860557556152344,
      "rewards/reward_model/mean": 0.9532963037490845,
      "rewards/reward_model/std": 0.9753101468086243,
      "step": 71,
      "step_time": 168.12876597139984
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.123046875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 71.07666015625,
      "completions/mean_terminated_length": 63.08964538574219,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2831250824965537,
      "epoch": 0.029654036243822075,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.650992214679718,
      "kl": 0.14023015335260425,
      "learning_rate": 8.765432098765433e-07,
      "loss": 0.0053,
      "num_tokens": 33708125.0,
      "reward": 0.8464133739471436,
      "reward_std": 0.5981078147888184,
      "rewards/reward_model/mean": 0.8464133739471436,
      "rewards/reward_model/std": 0.9848034977912903,
      "step": 72,
      "step_time": 168.81821045372635
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1435546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 70.923828125,
      "completions/mean_terminated_length": 61.356895446777344,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.27306635864079,
      "epoch": 0.030065897858319604,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6549474000930786,
      "kl": 0.15989831037586555,
      "learning_rate": 8.888888888888889e-07,
      "loss": -0.0137,
      "num_tokens": 34163265.0,
      "reward": 0.9670735001564026,
      "reward_std": 0.590969979763031,
      "rewards/reward_model/mean": 0.9670735001564026,
      "rewards/reward_model/std": 0.9453141689300537,
      "step": 73,
      "step_time": 169.4882780299522
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.15234375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.10107421875,
      "completions/mean_terminated_length": 66.77362060546875,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2641989213880152,
      "epoch": 0.030477759472817133,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5717378854751587,
      "kl": 0.13075158167339396,
      "learning_rate": 9.012345679012347e-07,
      "loss": 0.0049,
      "num_tokens": 34635568.0,
      "reward": 1.0525561571121216,
      "reward_std": 0.5589165687561035,
      "rewards/reward_model/mean": 1.0525561571121216,
      "rewards/reward_model/std": 0.8849756121635437,
      "step": 74,
      "step_time": 169.85129849473014
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.18017578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.9130859375,
      "completions/mean_terminated_length": 65.6855239868164,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2528326134197414,
      "epoch": 0.03088962108731466,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5799550414085388,
      "kl": 0.13488844226230867,
      "learning_rate": 9.135802469135802e-07,
      "loss": 0.0233,
      "num_tokens": 35042110.0,
      "reward": 1.0121339559555054,
      "reward_std": 0.6093316078186035,
      "rewards/reward_model/mean": 1.0121339559555054,
      "rewards/reward_model/std": 0.9795147776603699,
      "step": 75,
      "step_time": 168.39488552790135
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.11328125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 72.35791015625,
      "completions/mean_terminated_length": 65.24944305419922,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.2186853648163378,
      "epoch": 0.03130148270181219,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6350732445716858,
      "kl": 0.13198584888596088,
      "learning_rate": 9.259259259259259e-07,
      "loss": 0.0245,
      "num_tokens": 35454619.0,
      "reward": 1.1278910636901855,
      "reward_std": 0.6185814738273621,
      "rewards/reward_model/mean": 1.1278910636901855,
      "rewards/reward_model/std": 0.9232901930809021,
      "step": 76,
      "step_time": 169.25788368703797
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.13134765625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 70.81689453125,
      "completions/mean_terminated_length": 62.170318603515625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.247056140564382,
      "epoch": 0.03171334431630972,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.759410560131073,
      "kl": 0.16601154587988276,
      "learning_rate": 9.382716049382717e-07,
      "loss": -0.0003,
      "num_tokens": 35845316.0,
      "reward": 1.0192276239395142,
      "reward_std": 0.5931369066238403,
      "rewards/reward_model/mean": 1.0192276239395142,
      "rewards/reward_model/std": 0.9772949814796448,
      "step": 77,
      "step_time": 168.0306376479566
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.09521484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 70.13916015625,
      "completions/mean_terminated_length": 64.05018615722656,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2540333659853786,
      "epoch": 0.03212520593080725,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6740376353263855,
      "kl": 0.1378997444990091,
      "learning_rate": 9.506172839506173e-07,
      "loss": -0.013,
      "num_tokens": 36287137.0,
      "reward": 0.9819941520690918,
      "reward_std": 0.604373574256897,
      "rewards/reward_model/mean": 0.9819941520690918,
      "rewards/reward_model/std": 0.9436709880828857,
      "step": 78,
      "step_time": 168.8950103893876
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.09130859375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 68.37744140625,
      "completions/mean_terminated_length": 62.3863525390625,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.1720305329654366,
      "epoch": 0.032537067545304776,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6466286182403564,
      "kl": 0.12133494461886585,
      "learning_rate": 9.62962962962963e-07,
      "loss": -0.0048,
      "num_tokens": 36719654.0,
      "reward": 1.1645737886428833,
      "reward_std": 0.5557790398597717,
      "rewards/reward_model/mean": 1.1645737886428833,
      "rewards/reward_model/std": 0.9391114711761475,
      "step": 79,
      "step_time": 169.75694013293833
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.13330078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 69.42236328125,
      "completions/mean_terminated_length": 60.412960052490234,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.2074508473742753,
      "epoch": 0.032948929159802305,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7178550362586975,
      "kl": 0.16269512700091582,
      "learning_rate": 9.753086419753088e-07,
      "loss": 0.0024,
      "num_tokens": 37198599.0,
      "reward": 1.0070809125900269,
      "reward_std": 0.6197090148925781,
      "rewards/reward_model/mean": 1.0070809125900269,
      "rewards/reward_model/std": 0.8695884943008423,
      "step": 80,
      "step_time": 170.65315298642963
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.10888671875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 69.23583984375,
      "completions/mean_terminated_length": 62.05534362792969,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.1959036465268582,
      "epoch": 0.033360790774299834,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6767128705978394,
      "kl": 0.1640322696766816,
      "learning_rate": 9.876543209876544e-07,
      "loss": -0.0102,
      "num_tokens": 37627498.0,
      "reward": 1.051206111907959,
      "reward_std": 0.6312122344970703,
      "rewards/reward_model/mean": 1.051206111907959,
      "rewards/reward_model/std": 1.006866455078125,
      "step": 81,
      "step_time": 169.01685216045007
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.15234375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.1484375,
      "completions/mean_terminated_length": 66.8294906616211,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2353556689340621,
      "epoch": 0.03377265238879736,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6121835708618164,
      "kl": 0.13194930272584315,
      "learning_rate": 1e-06,
      "loss": -0.0084,
      "num_tokens": 38077466.0,
      "reward": 1.1334960460662842,
      "reward_std": 0.5429809093475342,
      "rewards/reward_model/mean": 1.1334960460662842,
      "rewards/reward_model/std": 1.0062233209609985,
      "step": 82,
      "step_time": 169.77917499747127
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1455078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 74.54931640625,
      "completions/mean_terminated_length": 65.44742584228516,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.2487247881945223,
      "epoch": 0.03418451400329489,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8223838210105896,
      "kl": 0.1559760042873677,
      "learning_rate": 1.0123456790123457e-06,
      "loss": 0.0192,
      "num_tokens": 38560095.0,
      "reward": 1.0274322032928467,
      "reward_std": 0.6149877309799194,
      "rewards/reward_model/mean": 1.0274322032928467,
      "rewards/reward_model/std": 0.8848612308502197,
      "step": 83,
      "step_time": 168.7728981245309
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0693359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 68.9453125,
      "completions/mean_terminated_length": 64.54563903808594,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.203527741599828,
      "epoch": 0.03459637561779242,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6430616974830627,
      "kl": 0.16221579984994605,
      "learning_rate": 1.0246913580246913e-06,
      "loss": -0.0054,
      "num_tokens": 38989743.0,
      "reward": 1.1352043151855469,
      "reward_std": 0.5808489918708801,
      "rewards/reward_model/mean": 1.1352043151855469,
      "rewards/reward_model/std": 1.0034772157669067,
      "step": 84,
      "step_time": 167.36356884567067
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08544921875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 73.13720703125,
      "completions/mean_terminated_length": 68.01121520996094,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.1932638001162559,
      "epoch": 0.03500823723228995,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.618569552898407,
      "kl": 0.1509321930789156,
      "learning_rate": 1.037037037037037e-06,
      "loss": -0.0105,
      "num_tokens": 39413960.0,
      "reward": 1.1917307376861572,
      "reward_std": 0.6115972995758057,
      "rewards/reward_model/mean": 1.1917307376861572,
      "rewards/reward_model/std": 0.8577749729156494,
      "step": 85,
      "step_time": 169.4022615076974
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1328125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 74.16455078125,
      "completions/mean_terminated_length": 65.91948699951172,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.1511160423979163,
      "epoch": 0.03542009884678748,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6170347929000854,
      "kl": 0.16613518859958276,
      "learning_rate": 1.0493827160493827e-06,
      "loss": 0.0068,
      "num_tokens": 39897081.0,
      "reward": 1.2492460012435913,
      "reward_std": 0.6051790714263916,
      "rewards/reward_model/mean": 1.2492460012435913,
      "rewards/reward_model/std": 0.8991779685020447,
      "step": 86,
      "step_time": 168.7876625736244
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.091796875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 73.81884765625,
      "completions/mean_terminated_length": 68.34247589111328,
      "completions/min_length": 1.0,
      "completions/min_terminated_length": 1.0,
      "entropy": 1.216100089251995,
      "epoch": 0.035831960461285006,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6116214394569397,
      "kl": 0.17693783454888035,
      "learning_rate": 1.0617283950617285e-06,
      "loss": 0.014,
      "num_tokens": 40304870.0,
      "reward": 1.327715277671814,
      "reward_std": 0.5273313522338867,
      "rewards/reward_model/mean": 1.327715277671814,
      "rewards/reward_model/std": 0.8829416036605835,
      "step": 87,
      "step_time": 170.00215818034485
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.10498046875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.087890625,
      "completions/mean_terminated_length": 76.70267486572266,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2015210629906505,
      "epoch": 0.036243822075782535,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5836432576179504,
      "kl": 0.14079495580517687,
      "learning_rate": 1.074074074074074e-06,
      "loss": 0.0043,
      "num_tokens": 40738362.0,
      "reward": 1.2116522789001465,
      "reward_std": 0.5785905122756958,
      "rewards/reward_model/mean": 1.2116522789001465,
      "rewards/reward_model/std": 0.8599736094474792,
      "step": 88,
      "step_time": 170.04560359567404
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07080078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 71.1552734375,
      "completions/mean_terminated_length": 66.82395935058594,
      "completions/min_length": 1.0,
      "completions/min_terminated_length": 1.0,
      "entropy": 1.1739569688215852,
      "epoch": 0.036655683690280064,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5943244099617004,
      "kl": 0.16060514625860378,
      "learning_rate": 1.0864197530864199e-06,
      "loss": -0.0115,
      "num_tokens": 41174584.0,
      "reward": 1.079056739807129,
      "reward_std": 0.5492511987686157,
      "rewards/reward_model/mean": 1.079056739807129,
      "rewards/reward_model/std": 0.8875714540481567,
      "step": 89,
      "step_time": 170.3232544688508
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 73.50341796875,
      "completions/mean_terminated_length": 68.63350677490234,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2288584825582802,
      "epoch": 0.03706754530477759,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6955690979957581,
      "kl": 0.18743510471540503,
      "learning_rate": 1.0987654320987655e-06,
      "loss": 0.0392,
      "num_tokens": 41634879.0,
      "reward": 1.2760483026504517,
      "reward_std": 0.5275530219078064,
      "rewards/reward_model/mean": 1.2760483026504517,
      "rewards/reward_model/std": 0.8526185154914856,
      "step": 90,
      "step_time": 169.28955688048154
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.103515625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.4619140625,
      "completions/mean_terminated_length": 71.62635803222656,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.1926879836246371,
      "epoch": 0.03747940691927512,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6305184364318848,
      "kl": 0.16981972678331658,
      "learning_rate": 1.111111111111111e-06,
      "loss": -0.0032,
      "num_tokens": 42093777.0,
      "reward": 1.3892216682434082,
      "reward_std": 0.5210399627685547,
      "rewards/reward_model/mean": 1.3892216682434082,
      "rewards/reward_model/std": 0.8532023429870605,
      "step": 91,
      "step_time": 169.93193591805175
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1005859375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.02099609375,
      "completions/mean_terminated_length": 74.65526580810547,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.220891160191968,
      "epoch": 0.03789126853377265,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5791942477226257,
      "kl": 0.13990605663275346,
      "learning_rate": 1.1234567901234568e-06,
      "loss": 0.021,
      "num_tokens": 42533916.0,
      "reward": 1.3777389526367188,
      "reward_std": 0.5628249049186707,
      "rewards/reward_model/mean": 1.3777389526367188,
      "rewards/reward_model/std": 0.8695874214172363,
      "step": 92,
      "step_time": 168.67672005156055
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.13427734375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.66845703125,
      "completions/mean_terminated_length": 74.48223876953125,
      "completions/min_length": 8.0,
      "completions/min_terminated_length": 8.0,
      "entropy": 1.1679968070238829,
      "epoch": 0.03830313014827018,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.577627956867218,
      "kl": 0.14946075380430557,
      "learning_rate": 1.1358024691358024e-06,
      "loss": 0.0005,
      "num_tokens": 42977269.0,
      "reward": 1.2400810718536377,
      "reward_std": 0.5488580465316772,
      "rewards/reward_model/mean": 1.2400810718536377,
      "rewards/reward_model/std": 0.8818415999412537,
      "step": 93,
      "step_time": 169.11300712404773
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1591796875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 89.02783203125,
      "completions/mean_terminated_length": 81.64982604980469,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2256716336123645,
      "epoch": 0.03871499176276771,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.3020151853561401,
      "kl": 0.1575723221176304,
      "learning_rate": 1.1481481481481482e-06,
      "loss": 0.0078,
      "num_tokens": 43468782.0,
      "reward": 1.2878694534301758,
      "reward_std": 0.5034958124160767,
      "rewards/reward_model/mean": 1.2878694534301758,
      "rewards/reward_model/std": 0.8000524640083313,
      "step": 94,
      "step_time": 168.66197129152715
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.146484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.29296875,
      "completions/mean_terminated_length": 74.4485092163086,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2125889593735337,
      "epoch": 0.039126853377265236,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5775403380393982,
      "kl": 0.17348909995052963,
      "learning_rate": 1.160493827160494e-06,
      "loss": 0.0291,
      "num_tokens": 43925606.0,
      "reward": 1.3064830303192139,
      "reward_std": 0.5317621231079102,
      "rewards/reward_model/mean": 1.3064830303192139,
      "rewards/reward_model/std": 0.8767746090888977,
      "step": 95,
      "step_time": 168.66238435404375
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.11376953125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 83.8466796875,
      "completions/mean_terminated_length": 78.17851257324219,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2149158080574125,
      "epoch": 0.039538714991762765,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5917447209358215,
      "kl": 0.16629101379658096,
      "learning_rate": 1.1728395061728396e-06,
      "loss": 0.0372,
      "num_tokens": 44365228.0,
      "reward": 1.310151219367981,
      "reward_std": 0.5394536852836609,
      "rewards/reward_model/mean": 1.310151219367981,
      "rewards/reward_model/std": 0.8472654223442078,
      "step": 96,
      "step_time": 170.65063601452857
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.087890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.22119140625,
      "completions/mean_terminated_length": 76.7136001586914,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2331861625425518,
      "epoch": 0.039950576606260293,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6160045266151428,
      "kl": 0.18450818251585588,
      "learning_rate": 1.1851851851851852e-06,
      "loss": 0.0239,
      "num_tokens": 44809841.0,
      "reward": 1.388469934463501,
      "reward_std": 0.47841960191726685,
      "rewards/reward_model/mean": 1.388469934463501,
      "rewards/reward_model/std": 0.7604539394378662,
      "step": 97,
      "step_time": 168.5139070255682
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.14404296875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 86.03857421875,
      "completions/mean_terminated_length": 78.97718811035156,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2734931902959943,
      "epoch": 0.04036243822075782,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8496507406234741,
      "kl": 0.1486935554712545,
      "learning_rate": 1.197530864197531e-06,
      "loss": 0.024,
      "num_tokens": 45291392.0,
      "reward": 1.273500680923462,
      "reward_std": 0.5131819844245911,
      "rewards/reward_model/mean": 1.273500680923462,
      "rewards/reward_model/std": 0.8183842301368713,
      "step": 98,
      "step_time": 169.52931605745107
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07861328125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.30078125,
      "completions/mean_terminated_length": 71.8897705078125,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.210193380014971,
      "epoch": 0.04077429983525535,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6533998847007751,
      "kl": 0.16968276555417106,
      "learning_rate": 1.2098765432098765e-06,
      "loss": 0.0227,
      "num_tokens": 45769608.0,
      "reward": 1.374595046043396,
      "reward_std": 0.5025352835655212,
      "rewards/reward_model/mean": 1.374595046043396,
      "rewards/reward_model/std": 0.7355093359947205,
      "step": 99,
      "step_time": 170.76728575211018
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.10595703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.27490234375,
      "completions/mean_terminated_length": 74.61878204345703,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.1873215795494616,
      "epoch": 0.04118616144975288,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5959328413009644,
      "kl": 0.14610896681551822,
      "learning_rate": 1.2222222222222221e-06,
      "loss": 0.0218,
      "num_tokens": 46193339.0,
      "reward": 1.393322229385376,
      "reward_std": 0.5245035886764526,
      "rewards/reward_model/mean": 1.393322229385376,
      "rewards/reward_model/std": 0.8767962455749512,
      "step": 100,
      "step_time": 169.11898464756086
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1171875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.0029296875,
      "completions/mean_terminated_length": 74.7643814086914,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1970643543172628,
      "epoch": 0.04159802306425041,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6192975640296936,
      "kl": 0.17013628454878926,
      "learning_rate": 1.234567901234568e-06,
      "loss": 0.0146,
      "num_tokens": 46603041.0,
      "reward": 1.3850064277648926,
      "reward_std": 0.5491656064987183,
      "rewards/reward_model/mean": 1.3850064277648926,
      "rewards/reward_model/std": 0.8517816662788391,
      "step": 101,
      "step_time": 172.5300747868605
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.09130859375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.30126953125,
      "completions/mean_terminated_length": 74.40784454345703,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.2366106095723808,
      "epoch": 0.042009884678747944,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6139182448387146,
      "kl": 0.15838144707959145,
      "learning_rate": 1.2469135802469135e-06,
      "loss": 0.0044,
      "num_tokens": 47074410.0,
      "reward": 1.4079811573028564,
      "reward_std": 0.5216450691223145,
      "rewards/reward_model/mean": 1.4079811573028564,
      "rewards/reward_model/std": 0.7702781558036804,
      "step": 102,
      "step_time": 168.7905627740547
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 75.34375,
      "completions/mean_terminated_length": 70.88135528564453,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2070911666378379,
      "epoch": 0.04242174629324547,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6438104510307312,
      "kl": 0.18365478227497078,
      "learning_rate": 1.2592592592592593e-06,
      "loss": 0.0099,
      "num_tokens": 47573482.0,
      "reward": 1.379532814025879,
      "reward_std": 0.5358700156211853,
      "rewards/reward_model/mean": 1.379532814025879,
      "rewards/reward_model/std": 0.8822428584098816,
      "step": 103,
      "step_time": 169.78911154950038
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.12841796875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.322265625,
      "completions/mean_terminated_length": 73.2974853515625,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.2451989527326077,
      "epoch": 0.042833607907743,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5943405628204346,
      "kl": 0.15326940282830037,
      "learning_rate": 1.271604938271605e-06,
      "loss": 0.0054,
      "num_tokens": 48002462.0,
      "reward": 1.2383322715759277,
      "reward_std": 0.5271602869033813,
      "rewards/reward_model/mean": 1.2383322715759277,
      "rewards/reward_model/std": 1.0347553491592407,
      "step": 104,
      "step_time": 168.55086909374222
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06884765625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 72.16455078125,
      "completions/mean_terminated_length": 68.03618621826172,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1801623778883368,
      "epoch": 0.04324546952224053,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6378849148750305,
      "kl": 0.15821812994545326,
      "learning_rate": 1.2839506172839507e-06,
      "loss": 0.0035,
      "num_tokens": 48392175.0,
      "reward": 1.5423574447631836,
      "reward_std": 0.4824356138706207,
      "rewards/reward_model/mean": 1.5423574447631836,
      "rewards/reward_model/std": 0.8302909135818481,
      "step": 105,
      "step_time": 169.42736366018653
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0830078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.6953125,
      "completions/mean_terminated_length": 74.23216247558594,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.3169332989491522,
      "epoch": 0.04365733113673806,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5943061113357544,
      "kl": 0.17450666631339118,
      "learning_rate": 1.2962962962962962e-06,
      "loss": 0.0041,
      "num_tokens": 48840319.0,
      "reward": 1.3128201961517334,
      "reward_std": 0.5468074679374695,
      "rewards/reward_model/mean": 1.3128201961517334,
      "rewards/reward_model/std": 0.9597176909446716,
      "step": 106,
      "step_time": 169.50068523269147
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08642578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 73.5673828125,
      "completions/mean_terminated_length": 68.41796112060547,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1956646356265992,
      "epoch": 0.04406919275123559,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9529135823249817,
      "kl": 0.14988234487827867,
      "learning_rate": 1.308641975308642e-06,
      "loss": 0.0263,
      "num_tokens": 49281225.0,
      "reward": 1.289332628250122,
      "reward_std": 0.5048444867134094,
      "rewards/reward_model/mean": 1.289332628250122,
      "rewards/reward_model/std": 0.8929986953735352,
      "step": 107,
      "step_time": 169.95979618094862
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1025390625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.716796875,
      "completions/mean_terminated_length": 75.31446838378906,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2545830043964088,
      "epoch": 0.044481054365733116,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5836382508277893,
      "kl": 0.14862596249440685,
      "learning_rate": 1.3209876543209876e-06,
      "loss": 0.0258,
      "num_tokens": 49792933.0,
      "reward": 1.3871095180511475,
      "reward_std": 0.5084824562072754,
      "rewards/reward_model/mean": 1.3871095180511475,
      "rewards/reward_model/std": 0.8852909207344055,
      "step": 108,
      "step_time": 169.20244881836697
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.04638671875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 70.8515625,
      "completions/mean_terminated_length": 68.07168579101562,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2631548107601702,
      "epoch": 0.044892915980230645,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6485495567321777,
      "kl": 0.1719318167597521,
      "learning_rate": 1.3333333333333332e-06,
      "loss": -0.0057,
      "num_tokens": 50257109.0,
      "reward": 1.5582630634307861,
      "reward_std": 0.5006577968597412,
      "rewards/reward_model/mean": 1.5582630634307861,
      "rewards/reward_model/std": 0.8305342197418213,
      "step": 109,
      "step_time": 168.78190125897527
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0556640625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 71.658203125,
      "completions/mean_terminated_length": 68.33712768554688,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2364499233663082,
      "epoch": 0.045304777594728174,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.588637113571167,
      "kl": 0.16918645988334902,
      "learning_rate": 1.345679012345679e-06,
      "loss": 0.0182,
      "num_tokens": 50728345.0,
      "reward": 1.4181112051010132,
      "reward_std": 0.5059062242507935,
      "rewards/reward_model/mean": 1.4181112051010132,
      "rewards/reward_model/std": 0.9040238857269287,
      "step": 110,
      "step_time": 169.8461561407894
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0908203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.97265625,
      "completions/mean_terminated_length": 75.17507934570312,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.3138902625069022,
      "epoch": 0.0457166392092257,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5418975353240967,
      "kl": 0.16301694582216442,
      "learning_rate": 1.3580246913580248e-06,
      "loss": 0.0118,
      "num_tokens": 51243937.0,
      "reward": 1.2584481239318848,
      "reward_std": 0.4887618124485016,
      "rewards/reward_model/mean": 1.2584481239318848,
      "rewards/reward_model/std": 0.9598668813705444,
      "step": 111,
      "step_time": 169.55967817036435
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.10009765625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.0,
      "completions/mean_terminated_length": 71.32718658447266,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2457992839626968,
      "epoch": 0.04612850082372323,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7095631957054138,
      "kl": 0.18068589709582739,
      "learning_rate": 1.3703703703703704e-06,
      "loss": 0.0037,
      "num_tokens": 51729121.0,
      "reward": 1.4246959686279297,
      "reward_std": 0.44748643040657043,
      "rewards/reward_model/mean": 1.4246959686279297,
      "rewards/reward_model/std": 0.8918141722679138,
      "step": 112,
      "step_time": 169.78249773895368
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07568359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 73.623046875,
      "completions/mean_terminated_length": 69.17062377929688,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2187957542482764,
      "epoch": 0.04654036243822076,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9400354623794556,
      "kl": 0.22693650811561383,
      "learning_rate": 1.3827160493827162e-06,
      "loss": 0.0122,
      "num_tokens": 52150877.0,
      "reward": 1.5303287506103516,
      "reward_std": 0.4950566291809082,
      "rewards/reward_model/mean": 1.5303287506103516,
      "rewards/reward_model/std": 0.8175535798072815,
      "step": 113,
      "step_time": 169.6679522804916
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05517578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.8798828125,
      "completions/mean_terminated_length": 73.89456939697266,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.2598798810504377,
      "epoch": 0.04695222405271829,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5663942098617554,
      "kl": 0.18427117861574516,
      "learning_rate": 1.3950617283950617e-06,
      "loss": 0.0113,
      "num_tokens": 52610119.0,
      "reward": 1.4505566358566284,
      "reward_std": 0.47873079776763916,
      "rewards/reward_model/mean": 1.4505566358566284,
      "rewards/reward_model/std": 0.9025180339813232,
      "step": 114,
      "step_time": 168.86495931399986
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0615234375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.5009765625,
      "completions/mean_terminated_length": 73.12487030029297,
      "completions/min_length": 9.0,
      "completions/min_terminated_length": 9.0,
      "entropy": 1.2603700500912964,
      "epoch": 0.04736408566721582,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.594477653503418,
      "kl": 0.1864922004751861,
      "learning_rate": 1.4074074074074073e-06,
      "loss": 0.0007,
      "num_tokens": 53070697.0,
      "reward": 1.5104587078094482,
      "reward_std": 0.4391752779483795,
      "rewards/reward_model/mean": 1.5104587078094482,
      "rewards/reward_model/std": 0.7633559703826904,
      "step": 115,
      "step_time": 170.47059550089762
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08544921875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.56005859375,
      "completions/mean_terminated_length": 75.03417205810547,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2630685805343091,
      "epoch": 0.047775947281713346,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5976677536964417,
      "kl": 0.17307385959429666,
      "learning_rate": 1.4197530864197531e-06,
      "loss": 0.0185,
      "num_tokens": 53542148.0,
      "reward": 1.419188141822815,
      "reward_std": 0.4496381878852844,
      "rewards/reward_model/mean": 1.419188141822815,
      "rewards/reward_model/std": 0.8634824156761169,
      "step": 116,
      "step_time": 169.13070647930726
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.076171875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.9912109375,
      "completions/mean_terminated_length": 79.2801284790039,
      "completions/min_length": 10.0,
      "completions/min_terminated_length": 10.0,
      "entropy": 1.3025836027227342,
      "epoch": 0.048187808896210875,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5448313355445862,
      "kl": 0.1677903521922417,
      "learning_rate": 1.4320987654320987e-06,
      "loss": 0.0104,
      "num_tokens": 53986130.0,
      "reward": 1.4435430765151978,
      "reward_std": 0.4609750807285309,
      "rewards/reward_model/mean": 1.4435430765151978,
      "rewards/reward_model/std": 0.7846410870552063,
      "step": 117,
      "step_time": 169.9405871666968
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0830078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.50244140625,
      "completions/mean_terminated_length": 74.0218276977539,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.288459766190499,
      "epoch": 0.048599670510708404,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5833011865615845,
      "kl": 0.1983755056571681,
      "learning_rate": 1.4444444444444445e-06,
      "loss": 0.0079,
      "num_tokens": 54439799.0,
      "reward": 1.4345738887786865,
      "reward_std": 0.48493313789367676,
      "rewards/reward_model/mean": 1.4345738887786865,
      "rewards/reward_model/std": 0.8746789693832397,
      "step": 118,
      "step_time": 169.1852500126697
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07373046875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.51513671875,
      "completions/mean_terminated_length": 74.576171875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "entropy": 1.224245477002114,
      "epoch": 0.04901153212520593,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6017325520515442,
      "kl": 0.18918700981885195,
      "learning_rate": 1.4567901234567903e-06,
      "loss": 0.0202,
      "num_tokens": 54875606.0,
      "reward": 1.5676627159118652,
      "reward_std": 0.4177808165550232,
      "rewards/reward_model/mean": 1.5676627159118652,
      "rewards/reward_model/std": 0.854965329170227,
      "step": 119,
      "step_time": 169.29946460714564
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.10595703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.9853515625,
      "completions/mean_terminated_length": 72.05789184570312,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.209789669374004,
      "epoch": 0.04942339373970346,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8375149369239807,
      "kl": 0.20605222514132038,
      "learning_rate": 1.4691358024691359e-06,
      "loss": 0.016,
      "num_tokens": 55328024.0,
      "reward": 1.4581849575042725,
      "reward_std": 0.4875085949897766,
      "rewards/reward_model/mean": 1.4581849575042725,
      "rewards/reward_model/std": 0.9500890374183655,
      "step": 120,
      "step_time": 169.78462026640773
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.11181640625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 84.04736328125,
      "completions/mean_terminated_length": 78.51402282714844,
      "completions/min_length": 10.0,
      "completions/min_terminated_length": 10.0,
      "entropy": 1.232045111246407,
      "epoch": 0.04983525535420099,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5587739944458008,
      "kl": 0.18681989194010384,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.0177,
      "num_tokens": 55771033.0,
      "reward": 1.4740581512451172,
      "reward_std": 0.4508310556411743,
      "rewards/reward_model/mean": 1.4740581512451172,
      "rewards/reward_model/std": 0.8630524277687073,
      "step": 121,
      "step_time": 168.61212700419128
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08349609375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.2998046875,
      "completions/mean_terminated_length": 71.58977508544922,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2581788282841444,
      "epoch": 0.05024711696869852,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5757594704627991,
      "kl": 0.19532652158522978,
      "learning_rate": 1.4938271604938272e-06,
      "loss": 0.011,
      "num_tokens": 56215583.0,
      "reward": 1.4256713390350342,
      "reward_std": 0.4282228946685791,
      "rewards/reward_model/mean": 1.4256713390350342,
      "rewards/reward_model/std": 0.9199231863021851,
      "step": 122,
      "step_time": 168.55926717165858
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0947265625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.01416015625,
      "completions/mean_terminated_length": 72.78370666503906,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2156934614758939,
      "epoch": 0.05065897858319605,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5999395847320557,
      "kl": 0.19728930137353018,
      "learning_rate": 1.506172839506173e-06,
      "loss": 0.0113,
      "num_tokens": 56669980.0,
      "reward": 1.6116740703582764,
      "reward_std": 0.4245319962501526,
      "rewards/reward_model/mean": 1.6116740703582764,
      "rewards/reward_model/std": 0.7885070443153381,
      "step": 123,
      "step_time": 168.9781666644849
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07470703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.9921875,
      "completions/mean_terminated_length": 72.8738784790039,
      "completions/min_length": 8.0,
      "completions/min_terminated_length": 8.0,
      "entropy": 1.2410368812270463,
      "epoch": 0.051070840197693576,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5885641574859619,
      "kl": 0.22556541379890405,
      "learning_rate": 1.5185185185185186e-06,
      "loss": 0.0278,
      "num_tokens": 57086316.0,
      "reward": 1.7070322036743164,
      "reward_std": 0.4219028055667877,
      "rewards/reward_model/mean": 1.7070322036743164,
      "rewards/reward_model/std": 0.7904437184333801,
      "step": 124,
      "step_time": 168.52942496724427
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.14599609375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.66748046875,
      "completions/mean_terminated_length": 74.91766357421875,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2711665122769773,
      "epoch": 0.051482701812191105,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5652002096176147,
      "kl": 0.21976503301993944,
      "learning_rate": 1.5308641975308642e-06,
      "loss": 0.025,
      "num_tokens": 57550787.0,
      "reward": 1.5571563243865967,
      "reward_std": 0.43854257464408875,
      "rewards/reward_model/mean": 1.5571563243865967,
      "rewards/reward_model/std": 0.9019820690155029,
      "step": 125,
      "step_time": 169.8311795401387
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.083984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.79931640625,
      "completions/mean_terminated_length": 76.47174835205078,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2508959705010056,
      "epoch": 0.051894563426688634,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.587785542011261,
      "kl": 0.2117367441242095,
      "learning_rate": 1.5432098765432098e-06,
      "loss": 0.0343,
      "num_tokens": 57980552.0,
      "reward": 1.6625767946243286,
      "reward_std": 0.41725289821624756,
      "rewards/reward_model/mean": 1.6625767946243286,
      "rewards/reward_model/std": 0.781819760799408,
      "step": 126,
      "step_time": 170.32025544391945
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 74.45458984375,
      "completions/mean_terminated_length": 71.35691833496094,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2673575547523797,
      "epoch": 0.05230642504118616,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.663336992263794,
      "kl": 0.2514068058517296,
      "learning_rate": 1.5555555555555556e-06,
      "loss": 0.007,
      "num_tokens": 58421643.0,
      "reward": 1.6073158979415894,
      "reward_std": 0.431104838848114,
      "rewards/reward_model/mean": 1.6073158979415894,
      "rewards/reward_model/std": 0.7830482125282288,
      "step": 127,
      "step_time": 167.69548717467114
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05712890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.8486328125,
      "completions/mean_terminated_length": 76.93112182617188,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.3203896265476942,
      "epoch": 0.05271828665568369,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5664064288139343,
      "kl": 0.21401835474534892,
      "learning_rate": 1.5679012345679012e-06,
      "loss": 0.0218,
      "num_tokens": 58904437.0,
      "reward": 1.5991525650024414,
      "reward_std": 0.4234989285469055,
      "rewards/reward_model/mean": 1.5991525650024414,
      "rewards/reward_model/std": 0.8423987030982971,
      "step": 128,
      "step_time": 170.09795808279887
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0693359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.70556640625,
      "completions/mean_terminated_length": 72.88404846191406,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.2859586202539504,
      "epoch": 0.05313014827018122,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5938490033149719,
      "kl": 0.23812477025785483,
      "learning_rate": 1.5802469135802472e-06,
      "loss": 0.0198,
      "num_tokens": 59368826.0,
      "reward": 1.4986733198165894,
      "reward_std": 0.48006802797317505,
      "rewards/reward_model/mean": 1.4986733198165894,
      "rewards/reward_model/std": 0.9735957384109497,
      "step": 129,
      "step_time": 168.860689394176
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07666015625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.9814453125,
      "completions/mean_terminated_length": 74.91168975830078,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.210738726425916,
      "epoch": 0.05354200988467875,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5862839818000793,
      "kl": 0.21009712052182294,
      "learning_rate": 1.5925925925925927e-06,
      "loss": 0.0074,
      "num_tokens": 59805460.0,
      "reward": 1.6455633640289307,
      "reward_std": 0.4051080346107483,
      "rewards/reward_model/mean": 1.6455633640289307,
      "rewards/reward_model/std": 0.902286946773529,
      "step": 130,
      "step_time": 169.57672298140824
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0595703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 75.908203125,
      "completions/mean_terminated_length": 72.6085205078125,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2665375941433012,
      "epoch": 0.05395387149917628,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5997110605239868,
      "kl": 0.2307097713346593,
      "learning_rate": 1.6049382716049383e-06,
      "loss": 0.0136,
      "num_tokens": 60283128.0,
      "reward": 1.6560347080230713,
      "reward_std": 0.3901210129261017,
      "rewards/reward_model/mean": 1.6560347080230713,
      "rewards/reward_model/std": 0.7428885698318481,
      "step": 131,
      "step_time": 170.24099622154608
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05615234375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 75.75537109375,
      "completions/mean_terminated_length": 72.64718627929688,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "entropy": 1.280121157411486,
      "epoch": 0.054365733113673806,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6451330780982971,
      "kl": 0.24510033716796897,
      "learning_rate": 1.617283950617284e-06,
      "loss": 0.0219,
      "num_tokens": 60704675.0,
      "reward": 1.5663776397705078,
      "reward_std": 0.4158337414264679,
      "rewards/reward_model/mean": 1.5663776397705078,
      "rewards/reward_model/std": 0.8126929402351379,
      "step": 132,
      "step_time": 168.72953157825395
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03466796875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 72.27392578125,
      "completions/mean_terminated_length": 70.27263641357422,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.255036068148911,
      "epoch": 0.054777594728171335,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5956186056137085,
      "kl": 0.22500447847414762,
      "learning_rate": 1.6296296296296295e-06,
      "loss": 0.0051,
      "num_tokens": 61135252.0,
      "reward": 1.5705558061599731,
      "reward_std": 0.4293729066848755,
      "rewards/reward_model/mean": 1.5705558061599731,
      "rewards/reward_model/std": 0.8589349985122681,
      "step": 133,
      "step_time": 169.56890073092654
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0849609375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.0068359375,
      "completions/mean_terminated_length": 76.64353942871094,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.22501984029077,
      "epoch": 0.05518945634266886,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6271244883537292,
      "kl": 0.19278798706363887,
      "learning_rate": 1.6419753086419753e-06,
      "loss": 0.0213,
      "num_tokens": 61582082.0,
      "reward": 1.4659600257873535,
      "reward_std": 0.4655250310897827,
      "rewards/reward_model/mean": 1.4659600257873535,
      "rewards/reward_model/std": 0.8552236557006836,
      "step": 134,
      "step_time": 169.07396916579455
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06396484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.24267578125,
      "completions/mean_terminated_length": 73.77412414550781,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "entropy": 1.2544868639670312,
      "epoch": 0.05560131795716639,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.55135577917099,
      "kl": 0.21390190807869658,
      "learning_rate": 1.654320987654321e-06,
      "loss": 0.016,
      "num_tokens": 62051859.0,
      "reward": 1.7084484100341797,
      "reward_std": 0.42418497800827026,
      "rewards/reward_model/mean": 1.7084484100341797,
      "rewards/reward_model/std": 0.8365797996520996,
      "step": 135,
      "step_time": 168.55205999454483
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06396484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.744140625,
      "completions/mean_terminated_length": 73.24151611328125,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2688625399023294,
      "epoch": 0.05601317957166392,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6378998160362244,
      "kl": 0.23519006837159395,
      "learning_rate": 1.6666666666666669e-06,
      "loss": 0.0138,
      "num_tokens": 62528583.0,
      "reward": 1.6745352745056152,
      "reward_std": 0.4303410053253174,
      "rewards/reward_model/mean": 1.6745352745056152,
      "rewards/reward_model/std": 0.8760103583335876,
      "step": 136,
      "step_time": 167.9896323014982
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0771484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.08544921875,
      "completions/mean_terminated_length": 78.2470932006836,
      "completions/min_length": 7.0,
      "completions/min_terminated_length": 7.0,
      "entropy": 1.298474219162017,
      "epoch": 0.05642504118616145,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5870314240455627,
      "kl": 0.19887266453588381,
      "learning_rate": 1.6790123456790125e-06,
      "loss": 0.0049,
      "num_tokens": 63017686.0,
      "reward": 1.5558466911315918,
      "reward_std": 0.41865676641464233,
      "rewards/reward_model/mean": 1.5558466911315918,
      "rewards/reward_model/std": 0.8572432994842529,
      "step": 137,
      "step_time": 168.25818043760955
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.044921875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 74.47802734375,
      "completions/mean_terminated_length": 71.96063232421875,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2788015282712877,
      "epoch": 0.05683690280065898,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0465470552444458,
      "kl": 0.28733671543886885,
      "learning_rate": 1.691358024691358e-06,
      "loss": 0.0283,
      "num_tokens": 63458441.0,
      "reward": 1.8364841938018799,
      "reward_std": 0.3914712369441986,
      "rewards/reward_model/mean": 1.8364841938018799,
      "rewards/reward_model/std": 0.8168711066246033,
      "step": 138,
      "step_time": 167.18661137623712
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.74072265625,
      "completions/mean_terminated_length": 73.87821960449219,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2473736926913261,
      "epoch": 0.05724876441515651,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5875552892684937,
      "kl": 0.22631295106839389,
      "learning_rate": 1.7037037037037036e-06,
      "loss": 0.0339,
      "num_tokens": 63941110.0,
      "reward": 1.70628821849823,
      "reward_std": 0.4453471899032593,
      "rewards/reward_model/mean": 1.70628821849823,
      "rewards/reward_model/std": 0.8872243762016296,
      "step": 139,
      "step_time": 169.1617161957547
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08740234375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.11279296875,
      "completions/mean_terminated_length": 74.43070983886719,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2679406760726124,
      "epoch": 0.057660626029654036,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6072338819503784,
      "kl": 0.2250406270613894,
      "learning_rate": 1.7160493827160492e-06,
      "loss": 0.0179,
      "num_tokens": 64478525.0,
      "reward": 1.731923222541809,
      "reward_std": 0.3971579968929291,
      "rewards/reward_model/mean": 1.731923222541809,
      "rewards/reward_model/std": 0.8387468457221985,
      "step": 140,
      "step_time": 169.69765338627622
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06689453125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.95458984375,
      "completions/mean_terminated_length": 74.3668212890625,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.3288813619874418,
      "epoch": 0.058072487644151564,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9750301241874695,
      "kl": 0.22281288140220568,
      "learning_rate": 1.7283950617283952e-06,
      "loss": 0.0086,
      "num_tokens": 64943904.0,
      "reward": 1.5512080192565918,
      "reward_std": 0.4345375895500183,
      "rewards/reward_model/mean": 1.5512080192565918,
      "rewards/reward_model/std": 0.8531954288482666,
      "step": 141,
      "step_time": 169.35747446445748
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.09033203125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.90966796875,
      "completions/mean_terminated_length": 71.8362808227539,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2554950527846813,
      "epoch": 0.05848434925864909,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5846207141876221,
      "kl": 0.22971162866451778,
      "learning_rate": 1.7407407407407408e-06,
      "loss": 0.0123,
      "num_tokens": 65435015.0,
      "reward": 1.6992193460464478,
      "reward_std": 0.38921838998794556,
      "rewards/reward_model/mean": 1.6992193460464478,
      "rewards/reward_model/std": 0.8802086710929871,
      "step": 142,
      "step_time": 169.55797945754603
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0859375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.7138671875,
      "completions/mean_terminated_length": 74.08013153076172,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.241849954240024,
      "epoch": 0.05889621087314662,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5389864444732666,
      "kl": 0.19780789304059,
      "learning_rate": 1.7530864197530866e-06,
      "loss": 0.0162,
      "num_tokens": 65893181.0,
      "reward": 1.657914638519287,
      "reward_std": 0.43264538049697876,
      "rewards/reward_model/mean": 1.657914638519287,
      "rewards/reward_model/std": 0.9311876893043518,
      "step": 143,
      "step_time": 167.69665700104088
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.060546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.701171875,
      "completions/mean_terminated_length": 75.52391052246094,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.3291778746061027,
      "epoch": 0.05930807248764415,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5641889572143555,
      "kl": 0.23325793404364958,
      "learning_rate": 1.7654320987654322e-06,
      "loss": 0.0106,
      "num_tokens": 66306777.0,
      "reward": 1.6250760555267334,
      "reward_std": 0.45928484201431274,
      "rewards/reward_model/mean": 1.6250760555267334,
      "rewards/reward_model/std": 0.8440907001495361,
      "step": 144,
      "step_time": 167.09111699229106
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.5625,
      "completions/mean_terminated_length": 73.81069946289062,
      "completions/min_length": 10.0,
      "completions/min_terminated_length": 10.0,
      "entropy": 1.2597985244356096,
      "epoch": 0.05971993410214168,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6075507402420044,
      "kl": 0.2127083102823235,
      "learning_rate": 1.7777777777777777e-06,
      "loss": -0.0003,
      "num_tokens": 66685465.0,
      "reward": 1.637428641319275,
      "reward_std": 0.4271353781223297,
      "rewards/reward_model/mean": 1.637428641319275,
      "rewards/reward_model/std": 0.9868490099906921,
      "step": 145,
      "step_time": 168.623529009521
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08837890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.47509765625,
      "completions/mean_terminated_length": 75.8676986694336,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2390871203970164,
      "epoch": 0.06013179571663921,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.738397479057312,
      "kl": 0.29930633143521845,
      "learning_rate": 1.7901234567901233e-06,
      "loss": 0.0243,
      "num_tokens": 67148166.0,
      "reward": 1.8085708618164062,
      "reward_std": 0.4322406053543091,
      "rewards/reward_model/mean": 1.8085708618164062,
      "rewards/reward_model/std": 0.8529988527297974,
      "step": 146,
      "step_time": 168.20390673354268
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.037109375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.32373046875,
      "completions/mean_terminated_length": 74.33214569091797,
      "completions/min_length": 7.0,
      "completions/min_terminated_length": 7.0,
      "entropy": 1.2929235147312284,
      "epoch": 0.06054365733113674,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6655780673027039,
      "kl": 0.22947472316445783,
      "learning_rate": 1.8024691358024693e-06,
      "loss": 0.0168,
      "num_tokens": 67600669.0,
      "reward": 1.878852128982544,
      "reward_std": 0.38836777210235596,
      "rewards/reward_model/mean": 1.878852128982544,
      "rewards/reward_model/std": 0.9328538179397583,
      "step": 147,
      "step_time": 169.04876817949116
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07568359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.455078125,
      "completions/mean_terminated_length": 73.3164291381836,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.300009768921882,
      "epoch": 0.060955518945634266,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5745586156845093,
      "kl": 0.23608133487869054,
      "learning_rate": 1.814814814814815e-06,
      "loss": 0.0142,
      "num_tokens": 68108001.0,
      "reward": 1.627516269683838,
      "reward_std": 0.43401244282722473,
      "rewards/reward_model/mean": 1.627516269683838,
      "rewards/reward_model/std": 0.9561209082603455,
      "step": 148,
      "step_time": 167.96074909390882
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06884765625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.375,
      "completions/mean_terminated_length": 73.63188171386719,
      "completions/min_length": 8.0,
      "completions/min_terminated_length": 8.0,
      "entropy": 1.2789768348447978,
      "epoch": 0.061367380560131794,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6077526807785034,
      "kl": 0.2504324442707002,
      "learning_rate": 1.8271604938271605e-06,
      "loss": 0.0138,
      "num_tokens": 68588705.0,
      "reward": 1.7919840812683105,
      "reward_std": 0.40561193227767944,
      "rewards/reward_model/mean": 1.7919840812683105,
      "rewards/reward_model/std": 0.807359516620636,
      "step": 149,
      "step_time": 167.94775600917637
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0693359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.6728515625,
      "completions/mean_terminated_length": 77.14690399169922,
      "completions/min_length": 8.0,
      "completions/min_terminated_length": 8.0,
      "entropy": 1.286247756332159,
      "epoch": 0.06177924217462932,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5670334100723267,
      "kl": 0.22240044514182955,
      "learning_rate": 1.839506172839506e-06,
      "loss": 0.0155,
      "num_tokens": 69043619.0,
      "reward": 1.5120161771774292,
      "reward_std": 0.4330858290195465,
      "rewards/reward_model/mean": 1.5120161771774292,
      "rewards/reward_model/std": 0.8432108759880066,
      "step": 150,
      "step_time": 168.81714980350807
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.060546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.8125,
      "completions/mean_terminated_length": 76.70686340332031,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2792057455517352,
      "epoch": 0.06219110378912685,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5719607472419739,
      "kl": 0.25582361221313477,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.0076,
      "num_tokens": 69484995.0,
      "reward": 1.7927849292755127,
      "reward_std": 0.41597551107406616,
      "rewards/reward_model/mean": 1.7927849292755127,
      "rewards/reward_model/std": 0.9223642945289612,
      "step": 151,
      "step_time": 167.4020181344822
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0498046875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.515625,
      "completions/mean_terminated_length": 73.81706237792969,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.227274117525667,
      "epoch": 0.06260296540362438,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7039728164672852,
      "kl": 0.26282261524465866,
      "learning_rate": 1.8641975308641975e-06,
      "loss": 0.0056,
      "num_tokens": 69909667.0,
      "reward": 1.8002865314483643,
      "reward_std": 0.41272586584091187,
      "rewards/reward_model/mean": 1.8002865314483643,
      "rewards/reward_model/std": 0.8000503778457642,
      "step": 152,
      "step_time": 167.92274621222168
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06103515625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.96826171875,
      "completions/mean_terminated_length": 76.8460693359375,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.273845980875194,
      "epoch": 0.06301482701812192,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5925673246383667,
      "kl": 0.22481732474989258,
      "learning_rate": 1.8765432098765435e-06,
      "loss": 0.0193,
      "num_tokens": 70384674.0,
      "reward": 1.8214460611343384,
      "reward_std": 0.413318932056427,
      "rewards/reward_model/mean": 1.8214460611343384,
      "rewards/reward_model/std": 0.742709219455719,
      "step": 153,
      "step_time": 170.70164536684752
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.04638671875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 75.90673828125,
      "completions/mean_terminated_length": 73.37275695800781,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2647815430536866,
      "epoch": 0.06342668863261944,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6434057354927063,
      "kl": 0.30792183699668385,
      "learning_rate": 1.888888888888889e-06,
      "loss": 0.0344,
      "num_tokens": 70848451.0,
      "reward": 1.8151272535324097,
      "reward_std": 0.40342289209365845,
      "rewards/reward_model/mean": 1.8151272535324097,
      "rewards/reward_model/std": 0.8678472638130188,
      "step": 154,
      "step_time": 168.5453836512752
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.04736328125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.47216796875,
      "completions/mean_terminated_length": 74.96002197265625,
      "completions/min_length": 10.0,
      "completions/min_terminated_length": 10.0,
      "entropy": 1.269783977419138,
      "epoch": 0.06383855024711697,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6314256191253662,
      "kl": 0.2555277651408687,
      "learning_rate": 1.9012345679012346e-06,
      "loss": 0.0251,
      "num_tokens": 71328554.0,
      "reward": 1.7019236087799072,
      "reward_std": 0.4237982928752899,
      "rewards/reward_model/mean": 1.7019236087799072,
      "rewards/reward_model/std": 0.913361132144928,
      "step": 155,
      "step_time": 169.69989513559267
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.09521484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 87.9814453125,
      "completions/mean_terminated_length": 83.77010345458984,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.262259678915143,
      "epoch": 0.0642504118616145,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5448090434074402,
      "kl": 0.22080097059370019,
      "learning_rate": 1.9135802469135804e-06,
      "loss": 0.0095,
      "num_tokens": 71843364.0,
      "reward": 1.6485573053359985,
      "reward_std": 0.442721962928772,
      "rewards/reward_model/mean": 1.6485573053359985,
      "rewards/reward_model/std": 0.9059621691703796,
      "step": 156,
      "step_time": 168.91886990657076
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.076171875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 83.45654296875,
      "completions/mean_terminated_length": 79.78382873535156,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.2155877850018442,
      "epoch": 0.06466227347611203,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6289539933204651,
      "kl": 0.2453711698180996,
      "learning_rate": 1.925925925925926e-06,
      "loss": 0.026,
      "num_tokens": 72273771.0,
      "reward": 1.784334659576416,
      "reward_std": 0.4161246418952942,
      "rewards/reward_model/mean": 1.784334659576416,
      "rewards/reward_model/std": 0.757505476474762,
      "step": 157,
      "step_time": 167.44150482024997
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08154296875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.76904296875,
      "completions/mean_terminated_length": 77.66453552246094,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.235411421628669,
      "epoch": 0.06507413509060955,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5421144366264343,
      "kl": 0.22543006701744162,
      "learning_rate": 1.9382716049382716e-06,
      "loss": 0.0316,
      "num_tokens": 72800146.0,
      "reward": 1.7696501016616821,
      "reward_std": 0.3846498727798462,
      "rewards/reward_model/mean": 1.7696501016616821,
      "rewards/reward_model/std": 0.8624175786972046,
      "step": 158,
      "step_time": 169.24183974647895
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08642578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.5419921875,
      "completions/mean_terminated_length": 76.05238342285156,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.185378910973668,
      "epoch": 0.06548599670510709,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5627242922782898,
      "kl": 0.23744830160285346,
      "learning_rate": 1.9506172839506176e-06,
      "loss": 0.0326,
      "num_tokens": 73252072.0,
      "reward": 1.8059046268463135,
      "reward_std": 0.3923932611942291,
      "rewards/reward_model/mean": 1.8059046268463135,
      "rewards/reward_model/std": 0.8797404766082764,
      "step": 159,
      "step_time": 169.05729650333524
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.080078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.8115234375,
      "completions/mean_terminated_length": 77.79087829589844,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.23535214853473,
      "epoch": 0.06589785831960461,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5441705584526062,
      "kl": 0.24602960981428623,
      "learning_rate": 1.962962962962963e-06,
      "loss": 0.0213,
      "num_tokens": 73739494.0,
      "reward": 1.7673776149749756,
      "reward_std": 0.4512333571910858,
      "rewards/reward_model/mean": 1.7673776149749756,
      "rewards/reward_model/std": 0.9724159240722656,
      "step": 160,
      "step_time": 168.9636302953586
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06982421875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.20166015625,
      "completions/mean_terminated_length": 74.46351623535156,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2300576204434037,
      "epoch": 0.06630971993410215,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5943037271499634,
      "kl": 0.24898978223791346,
      "learning_rate": 1.9753086419753087e-06,
      "loss": 0.0222,
      "num_tokens": 74167715.0,
      "reward": 1.926592469215393,
      "reward_std": 0.41215306520462036,
      "rewards/reward_model/mean": 1.926592469215393,
      "rewards/reward_model/std": 1.001451015472412,
      "step": 161,
      "step_time": 167.6384666627273
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06884765625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 73.8603515625,
      "completions/mean_terminated_length": 69.85736846923828,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.198827504646033,
      "epoch": 0.06672158154859967,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5532698035240173,
      "kl": 0.27067929052282125,
      "learning_rate": 1.9876543209876543e-06,
      "loss": 0.0219,
      "num_tokens": 74647397.0,
      "reward": 1.9008184671401978,
      "reward_std": 0.38047462701797485,
      "rewards/reward_model/mean": 1.9008184671401978,
      "rewards/reward_model/std": 0.9279819130897522,
      "step": 162,
      "step_time": 167.92858559498563
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0458984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 75.6513671875,
      "completions/mean_terminated_length": 73.133056640625,
      "completions/min_length": 9.0,
      "completions/min_terminated_length": 9.0,
      "entropy": 1.2209027321077883,
      "epoch": 0.0671334431630972,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6697596907615662,
      "kl": 0.27582967511261813,
      "learning_rate": 2e-06,
      "loss": 0.0013,
      "num_tokens": 75100251.0,
      "reward": 1.778894305229187,
      "reward_std": 0.40786850452423096,
      "rewards/reward_model/mean": 1.778894305229187,
      "rewards/reward_model/std": 0.9986056089401245,
      "step": 163,
      "step_time": 168.67196059180424
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.04736328125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 76.04296875,
      "completions/mean_terminated_length": 73.45977020263672,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1774590322747827,
      "epoch": 0.06754530477759473,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5625696182250977,
      "kl": 0.24044052005046979,
      "learning_rate": 2.012345679012346e-06,
      "loss": 0.0079,
      "num_tokens": 75646451.0,
      "reward": 1.668849229812622,
      "reward_std": 0.40883883833885193,
      "rewards/reward_model/mean": 1.668849229812622,
      "rewards/reward_model/std": 0.9833309650421143,
      "step": 164,
      "step_time": 168.17798956111073
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0556640625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.30810546875,
      "completions/mean_terminated_length": 75.3790054321289,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2012779735960066,
      "epoch": 0.06795716639209226,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.647919774055481,
      "kl": 0.24133505855570547,
      "learning_rate": 2.0246913580246915e-06,
      "loss": -0.0057,
      "num_tokens": 76075114.0,
      "reward": 1.8032076358795166,
      "reward_std": 0.4206048548221588,
      "rewards/reward_model/mean": 1.8032076358795166,
      "rewards/reward_model/std": 0.8603482842445374,
      "step": 165,
      "step_time": 167.6303828060627
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0712890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 82.9677734375,
      "completions/mean_terminated_length": 79.51103973388672,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2237120219506323,
      "epoch": 0.06836902800658978,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5906450748443604,
      "kl": 0.23315264994744211,
      "learning_rate": 2.037037037037037e-06,
      "loss": 0.0018,
      "num_tokens": 76518888.0,
      "reward": 1.6204566955566406,
      "reward_std": 0.43912777304649353,
      "rewards/reward_model/mean": 1.6204566955566406,
      "rewards/reward_model/std": 0.8727616667747498,
      "step": 166,
      "step_time": 169.20850368216634
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.04345703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 78.6572265625,
      "completions/mean_terminated_length": 76.41551971435547,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1896002353169024,
      "epoch": 0.06878088962108732,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6026791930198669,
      "kl": 0.27406785456696525,
      "learning_rate": 2.0493827160493827e-06,
      "loss": 0.0206,
      "num_tokens": 77015562.0,
      "reward": 1.8125802278518677,
      "reward_std": 0.3493039608001709,
      "rewards/reward_model/mean": 1.8125802278518677,
      "rewards/reward_model/std": 0.8256182074546814,
      "step": 167,
      "step_time": 167.97593408357352
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0576171875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.49560546875,
      "completions/mean_terminated_length": 77.59119415283203,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "entropy": 1.2243428956717253,
      "epoch": 0.06919275123558484,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5614490509033203,
      "kl": 0.27035874902503565,
      "learning_rate": 2.0617283950617282e-06,
      "loss": 0.0263,
      "num_tokens": 77461185.0,
      "reward": 1.7474470138549805,
      "reward_std": 0.345928430557251,
      "rewards/reward_model/mean": 1.7474470138549805,
      "rewards/reward_model/std": 0.783444344997406,
      "step": 168,
      "step_time": 167.4764309921302
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1171875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 87.9755859375,
      "completions/mean_terminated_length": 82.66261291503906,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2144196771550924,
      "epoch": 0.06960461285008238,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7678112387657166,
      "kl": 0.2838464966916945,
      "learning_rate": 2.074074074074074e-06,
      "loss": 0.0225,
      "num_tokens": 77914863.0,
      "reward": 1.836325764656067,
      "reward_std": 0.39190569519996643,
      "rewards/reward_model/mean": 1.836325764656067,
      "rewards/reward_model/std": 0.8917631506919861,
      "step": 169,
      "step_time": 167.54137054365128
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1064453125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 83.44287109375,
      "completions/mean_terminated_length": 78.13497924804688,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.215761701343581,
      "epoch": 0.0700164744645799,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.551048219203949,
      "kl": 0.2747056516527664,
      "learning_rate": 2.08641975308642e-06,
      "loss": 0.0378,
      "num_tokens": 78387770.0,
      "reward": 1.7706103324890137,
      "reward_std": 0.3842095136642456,
      "rewards/reward_model/mean": 1.7706103324890137,
      "rewards/reward_model/std": 0.9329609274864197,
      "step": 170,
      "step_time": 168.65306692710146
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.11474609375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 86.2822265625,
      "completions/mean_terminated_length": 80.87478637695312,
      "completions/min_length": 8.0,
      "completions/min_terminated_length": 8.0,
      "entropy": 1.2296040374785662,
      "epoch": 0.07042833607907743,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6304351091384888,
      "kl": 0.2613049374194816,
      "learning_rate": 2.0987654320987654e-06,
      "loss": 0.0244,
      "num_tokens": 78848220.0,
      "reward": 1.8255634307861328,
      "reward_std": 0.3948723077774048,
      "rewards/reward_model/mean": 1.8255634307861328,
      "rewards/reward_model/std": 0.8326601982116699,
      "step": 171,
      "step_time": 168.58243151567876
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1396484375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 89.66064453125,
      "completions/mean_terminated_length": 83.43757629394531,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1852441588416696,
      "epoch": 0.07084019769357495,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5359068512916565,
      "kl": 0.25176908579305746,
      "learning_rate": 2.1111111111111114e-06,
      "loss": 0.0259,
      "num_tokens": 79300933.0,
      "reward": 1.806993007659912,
      "reward_std": 0.37374427914619446,
      "rewards/reward_model/mean": 1.806993007659912,
      "rewards/reward_model/std": 0.8301866054534912,
      "step": 172,
      "step_time": 168.35271108709276
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.13720703125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 85.8330078125,
      "completions/mean_terminated_length": 79.12733459472656,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2571615048218518,
      "epoch": 0.07125205930807249,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5572459101676941,
      "kl": 0.2960540017229505,
      "learning_rate": 2.123456790123457e-06,
      "loss": 0.0209,
      "num_tokens": 79797039.0,
      "reward": 1.8621840476989746,
      "reward_std": 0.40364766120910645,
      "rewards/reward_model/mean": 1.8621840476989746,
      "rewards/reward_model/std": 0.9432704448699951,
      "step": 173,
      "step_time": 167.40115015720949
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0693359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.65771484375,
      "completions/mean_terminated_length": 76.05613708496094,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.2446925514377654,
      "epoch": 0.07166392092257001,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5800526142120361,
      "kl": 0.2971960572176613,
      "learning_rate": 2.1358024691358026e-06,
      "loss": 0.0368,
      "num_tokens": 80245682.0,
      "reward": 1.9920142889022827,
      "reward_std": 0.3743855655193329,
      "rewards/reward_model/mean": 1.9920142889022827,
      "rewards/reward_model/std": 0.9964386820793152,
      "step": 174,
      "step_time": 168.70972929289564
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05517578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.859375,
      "completions/mean_terminated_length": 78.10645294189453,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2962157218717039,
      "epoch": 0.07207578253706755,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6903020143508911,
      "kl": 0.2973472154699266,
      "learning_rate": 2.148148148148148e-06,
      "loss": 0.0265,
      "num_tokens": 80705938.0,
      "reward": 1.7820935249328613,
      "reward_std": 0.37934696674346924,
      "rewards/reward_model/mean": 1.7820935249328613,
      "rewards/reward_model/std": 0.902191162109375,
      "step": 175,
      "step_time": 169.36274442402646
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.068359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.72021484375,
      "completions/mean_terminated_length": 78.32441711425781,
      "completions/min_length": 14.0,
      "completions/min_terminated_length": 14.0,
      "entropy": 1.27019348484464,
      "epoch": 0.07248764415156507,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5628058910369873,
      "kl": 0.2966971327550709,
      "learning_rate": 2.160493827160494e-06,
      "loss": 0.0268,
      "num_tokens": 81138997.0,
      "reward": 1.7612276077270508,
      "reward_std": 0.41708946228027344,
      "rewards/reward_model/mean": 1.7612276077270508,
      "rewards/reward_model/std": 0.9293939471244812,
      "step": 176,
      "step_time": 169.23447634931654
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0537109375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.7333984375,
      "completions/mean_terminated_length": 74.88029479980469,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "entropy": 1.2512029868084937,
      "epoch": 0.0728995057660626,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5666512250900269,
      "kl": 0.30354723357595503,
      "learning_rate": 2.1728395061728397e-06,
      "loss": 0.0181,
      "num_tokens": 81605171.0,
      "reward": 1.9841545820236206,
      "reward_std": 0.3431292474269867,
      "rewards/reward_model/mean": 1.9841545820236206,
      "rewards/reward_model/std": 0.8817236423492432,
      "step": 177,
      "step_time": 168.3823292935267
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07666015625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 75.712890625,
      "completions/mean_terminated_length": 71.37175750732422,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.263297796715051,
      "epoch": 0.07331136738056013,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6307470798492432,
      "kl": 0.3325087181874551,
      "learning_rate": 2.1851851851851853e-06,
      "loss": 0.0256,
      "num_tokens": 82068551.0,
      "reward": 1.8173794746398926,
      "reward_std": 0.3982235789299011,
      "rewards/reward_model/mean": 1.8173794746398926,
      "rewards/reward_model/std": 1.106950283050537,
      "step": 178,
      "step_time": 167.39255077391863
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06103515625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.169921875,
      "completions/mean_terminated_length": 75.99583435058594,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.271299744490534,
      "epoch": 0.07372322899505766,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6578369736671448,
      "kl": 0.33446174120763317,
      "learning_rate": 2.197530864197531e-06,
      "loss": 0.0284,
      "num_tokens": 82517795.0,
      "reward": 1.7375619411468506,
      "reward_std": 0.3846127390861511,
      "rewards/reward_model/mean": 1.7375619411468506,
      "rewards/reward_model/std": 0.8862913250923157,
      "step": 179,
      "step_time": 168.1992179742083
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0419921875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.02197265625,
      "completions/mean_terminated_length": 77.91896057128906,
      "completions/min_length": 16.0,
      "completions/min_terminated_length": 16.0,
      "entropy": 1.2483481515664607,
      "epoch": 0.07413509060955518,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5584097504615784,
      "kl": 0.30480305349919945,
      "learning_rate": 2.2098765432098765e-06,
      "loss": 0.0213,
      "num_tokens": 82938224.0,
      "reward": 1.9653817415237427,
      "reward_std": 0.37414827942848206,
      "rewards/reward_model/mean": 1.9653817415237427,
      "rewards/reward_model/std": 0.8811721205711365,
      "step": 180,
      "step_time": 167.49799311021343
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.02587890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 74.77294921875,
      "completions/mean_terminated_length": 73.35889434814453,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "entropy": 1.2098995437845588,
      "epoch": 0.07454695222405272,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7208864688873291,
      "kl": 0.3534813959267922,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0354,
      "num_tokens": 83381023.0,
      "reward": 2.026184320449829,
      "reward_std": 0.3664648234844208,
      "rewards/reward_model/mean": 2.026184320449829,
      "rewards/reward_model/std": 0.780437707901001,
      "step": 181,
      "step_time": 168.94889666279778
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.02685546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 74.79931640625,
      "completions/mean_terminated_length": 73.33116149902344,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.2554190349765122,
      "epoch": 0.07495881383855024,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6133463382720947,
      "kl": 0.3488931337487884,
      "learning_rate": 2.234567901234568e-06,
      "loss": -0.0,
      "num_tokens": 83815236.0,
      "reward": 2.015033006668091,
      "reward_std": 0.35350939631462097,
      "rewards/reward_model/mean": 2.015033006668091,
      "rewards/reward_model/std": 1.0514765977859497,
      "step": 182,
      "step_time": 168.09779750416055
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0869140625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.35546875,
      "completions/mean_terminated_length": 74.72513580322266,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.199432454770431,
      "epoch": 0.07537067545304778,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6157463192939758,
      "kl": 0.3277318494510837,
      "learning_rate": 2.2469135802469137e-06,
      "loss": 0.0357,
      "num_tokens": 84297404.0,
      "reward": 1.8543448448181152,
      "reward_std": 0.391369104385376,
      "rewards/reward_model/mean": 1.8543448448181152,
      "rewards/reward_model/std": 0.9162046909332275,
      "step": 183,
      "step_time": 167.68394365813583
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0869140625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 84.412109375,
      "completions/mean_terminated_length": 80.26309967041016,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2098595479037613,
      "epoch": 0.0757825370675453,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5533583760261536,
      "kl": 0.31157187622739,
      "learning_rate": 2.2592592592592592e-06,
      "loss": 0.0297,
      "num_tokens": 84797320.0,
      "reward": 1.829999327659607,
      "reward_std": 0.39016926288604736,
      "rewards/reward_model/mean": 1.829999327659607,
      "rewards/reward_model/std": 0.919818103313446,
      "step": 184,
      "step_time": 169.95830805273727
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08642578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.021484375,
      "completions/mean_terminated_length": 72.1988296508789,
      "completions/min_length": 1.0,
      "completions/min_terminated_length": 1.0,
      "entropy": 1.216996781527996,
      "epoch": 0.07619439868204284,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5928105115890503,
      "kl": 0.3597461999161169,
      "learning_rate": 2.271604938271605e-06,
      "loss": 0.0241,
      "num_tokens": 85345332.0,
      "reward": 1.7781141996383667,
      "reward_std": 0.40030625462532043,
      "rewards/reward_model/mean": 1.7781143188476562,
      "rewards/reward_model/std": 1.0846326351165771,
      "step": 185,
      "step_time": 168.48916833195835
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.12060546875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 86.26416015625,
      "completions/mean_terminated_length": 80.54025268554688,
      "completions/min_length": 17.0,
      "completions/min_terminated_length": 17.0,
      "entropy": 1.1919847468379885,
      "epoch": 0.07660626029654036,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5492737889289856,
      "kl": 0.30091724882368,
      "learning_rate": 2.2839506172839504e-06,
      "loss": 0.0307,
      "num_tokens": 85790705.0,
      "reward": 1.9125442504882812,
      "reward_std": 0.45980480313301086,
      "rewards/reward_model/mean": 1.9125442504882812,
      "rewards/reward_model/std": 0.884122371673584,
      "step": 186,
      "step_time": 168.23294077534229
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.083984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.7451171875,
      "completions/mean_terminated_length": 75.3208999633789,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1483043180778623,
      "epoch": 0.07701812191103789,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5400401949882507,
      "kl": 0.31244106526719406,
      "learning_rate": 2.2962962962962964e-06,
      "loss": 0.0363,
      "num_tokens": 86241287.0,
      "reward": 1.9626367092132568,
      "reward_std": 0.37234199047088623,
      "rewards/reward_model/mean": 1.9626367092132568,
      "rewards/reward_model/std": 0.9630360007286072,
      "step": 187,
      "step_time": 168.14725243346766
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0634765625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.87939453125,
      "completions/mean_terminated_length": 78.75338745117188,
      "completions/min_length": 13.0,
      "completions/min_terminated_length": 13.0,
      "entropy": 1.2079863131511956,
      "epoch": 0.07742998352553541,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.582582414150238,
      "kl": 0.3190855036955327,
      "learning_rate": 2.308641975308642e-06,
      "loss": 0.0175,
      "num_tokens": 86728272.0,
      "reward": 1.9138296842575073,
      "reward_std": 0.3768790364265442,
      "rewards/reward_model/mean": 1.9138296842575073,
      "rewards/reward_model/std": 1.0191558599472046,
      "step": 188,
      "step_time": 168.40571281081066
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.11376953125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 85.12548828125,
      "completions/mean_terminated_length": 79.62149047851562,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.169044690206647,
      "epoch": 0.07784184514003295,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5737549662590027,
      "kl": 0.31175463960971683,
      "learning_rate": 2.320987654320988e-06,
      "loss": 0.0287,
      "num_tokens": 87171473.0,
      "reward": 1.9407716989517212,
      "reward_std": 0.4355354905128479,
      "rewards/reward_model/mean": 1.9407716989517212,
      "rewards/reward_model/std": 1.0851298570632935,
      "step": 189,
      "step_time": 167.8595218854025
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08837890625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 83.955078125,
      "completions/mean_terminated_length": 79.68505859375,
      "completions/min_length": 11.0,
      "completions/min_terminated_length": 11.0,
      "entropy": 1.1897474566940218,
      "epoch": 0.07825370675453047,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5536565184593201,
      "kl": 0.32399573898874223,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 0.0254,
      "num_tokens": 87624149.0,
      "reward": 1.953005313873291,
      "reward_std": 0.39668840169906616,
      "rewards/reward_model/mean": 1.953005313873291,
      "rewards/reward_model/std": 0.9170160889625549,
      "step": 190,
      "step_time": 167.70779052050784
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07958984375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 87.31494140625,
      "completions/mean_terminated_length": 83.79681396484375,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1992020884063095,
      "epoch": 0.07866556836902801,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.534305214881897,
      "kl": 0.2823739905725233,
      "learning_rate": 2.345679012345679e-06,
      "loss": 0.0207,
      "num_tokens": 88065530.0,
      "reward": 1.8692455291748047,
      "reward_std": 0.3706667721271515,
      "rewards/reward_model/mean": 1.8692455291748047,
      "rewards/reward_model/std": 0.7108120918273926,
      "step": 191,
      "step_time": 168.080767756328
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06982421875,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.57177734375,
      "completions/mean_terminated_length": 77.01155090332031,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2336446167901158,
      "epoch": 0.07907742998352553,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7091127634048462,
      "kl": 0.3430620247963816,
      "learning_rate": 2.3580246913580247e-06,
      "loss": 0.0272,
      "num_tokens": 88569293.0,
      "reward": 1.8590911626815796,
      "reward_std": 0.3997899293899536,
      "rewards/reward_model/mean": 1.8590911626815796,
      "rewards/reward_model/std": 1.0301216840744019,
      "step": 192,
      "step_time": 168.8303936952725
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05224609375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.47802734375,
      "completions/mean_terminated_length": 74.69293975830078,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2118491139262915,
      "epoch": 0.07948929159802307,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6019130349159241,
      "kl": 0.35962342529091984,
      "learning_rate": 2.3703703703703703e-06,
      "loss": 0.0434,
      "num_tokens": 88987872.0,
      "reward": 2.0519325733184814,
      "reward_std": 0.383076012134552,
      "rewards/reward_model/mean": 2.0519325733184814,
      "rewards/reward_model/std": 0.8245954513549805,
      "step": 193,
      "step_time": 166.53060166956857
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03369140625,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 75.05859375,
      "completions/mean_terminated_length": 73.21273040771484,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "entropy": 1.186548251658678,
      "epoch": 0.07990115321252059,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7078003287315369,
      "kl": 0.3844323490629904,
      "learning_rate": 2.3827160493827163e-06,
      "loss": 0.0161,
      "num_tokens": 89383064.0,
      "reward": 2.240997076034546,
      "reward_std": 0.3178167939186096,
      "rewards/reward_model/mean": 2.240997076034546,
      "rewards/reward_model/std": 0.84669429063797,
      "step": 194,
      "step_time": 167.91490666009486
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0439453125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.45703125,
      "completions/mean_terminated_length": 75.1338119506836,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.1919686342589557,
      "epoch": 0.08031301482701812,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5621170401573181,
      "kl": 0.3380963269737549,
      "learning_rate": 2.395061728395062e-06,
      "loss": 0.0135,
      "num_tokens": 89828864.0,
      "reward": 1.9205503463745117,
      "reward_std": 0.37170130014419556,
      "rewards/reward_model/mean": 1.9205503463745117,
      "rewards/reward_model/std": 0.847759485244751,
      "step": 195,
      "step_time": 167.42564077628776
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05078125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 77.94970703125,
      "completions/mean_terminated_length": 75.2721176147461,
      "completions/min_length": 22.0,
      "completions/min_terminated_length": 22.0,
      "entropy": 1.1974371783435345,
      "epoch": 0.08072487644151564,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6827316880226135,
      "kl": 0.3753180764615536,
      "learning_rate": 2.4074074074074075e-06,
      "loss": 0.0271,
      "num_tokens": 90260377.0,
      "reward": 2.031513214111328,
      "reward_std": 0.3799358606338501,
      "rewards/reward_model/mean": 2.031513214111328,
      "rewards/reward_model/std": 1.0462548732757568,
      "step": 196,
      "step_time": 166.99446870852262
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07568359375,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.01416015625,
      "completions/mean_terminated_length": 76.08504486083984,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "entropy": 1.2283128183335066,
      "epoch": 0.08113673805601318,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.570841372013092,
      "kl": 0.3397949834470637,
      "learning_rate": 2.419753086419753e-06,
      "loss": 0.0326,
      "num_tokens": 90748310.0,
      "reward": 1.893890380859375,
      "reward_std": 0.3679511547088623,
      "rewards/reward_model/mean": 1.893890380859375,
      "rewards/reward_model/std": 0.8536400198936462,
      "step": 197,
      "step_time": 167.43677308317274
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.04736328125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 79.32080078125,
      "completions/mean_terminated_length": 76.90056610107422,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.2077249235007912,
      "epoch": 0.0815485996705107,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5746736526489258,
      "kl": 0.3641646616742946,
      "learning_rate": 2.4320987654320987e-06,
      "loss": 0.029,
      "num_tokens": 91213543.0,
      "reward": 2.129655361175537,
      "reward_std": 0.3387081027030945,
      "rewards/reward_model/mean": 2.129655361175537,
      "rewards/reward_model/std": 0.9924630522727966,
      "step": 198,
      "step_time": 168.12879333738238
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08642578125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 80.84912109375,
      "completions/mean_terminated_length": 76.38856506347656,
      "completions/min_length": 17.0,
      "completions/min_terminated_length": 17.0,
      "entropy": 1.106790901394561,
      "epoch": 0.08196046128500824,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6154218316078186,
      "kl": 0.3942809420404956,
      "learning_rate": 2.4444444444444442e-06,
      "loss": 0.0462,
      "num_tokens": 91686258.0,
      "reward": 2.0349197387695312,
      "reward_std": 0.38085871934890747,
      "rewards/reward_model/mean": 2.0349197387695312,
      "rewards/reward_model/std": 1.0781173706054688,
      "step": 199,
      "step_time": 166.76508231228217
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.064453125,
      "completions/max_length": 128.0,
      "completions/max_terminated_length": 128.0,
      "completions/mean_length": 81.89501953125,
      "completions/mean_terminated_length": 78.71868896484375,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.1731834115926176,
      "epoch": 0.08237232289950576,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5466514825820923,
      "kl": 0.34953623195178807,
      "learning_rate": 2.4567901234567902e-06,
      "loss": 0.0346,
      "num_tokens": 92113083.0,
      "reward": 2.178508758544922,
      "reward_std": 0.3616424798965454,
      "rewards/reward_model/mean": 2.178508758544922,
      "rewards/reward_model/std": 0.908944845199585,
      "step": 200,
      "step_time": 168.23503723321483
    }
  ],
  "logging_steps": 1,
  "max_steps": 2428,
  "num_input_tokens_seen": 92113083,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}