{
  "schema": 1,
  "bootstrap": {
    "resamples": 1000,
    "seed": 1234,
    "method": "Paired percentile bootstrap over problems; four samples stay grouped. Python random. Conditional on recorded training runs."
  },
  "evaluations": {
    "dev": {
      "models": {
        "base": {
          "source": "exports/baselines/student-dev.json",
          "n": 500,
          "greedy": {
            "mean": 0.524,
            "ci95": [
              0.48,
              0.566
            ]
          },
          "sampled": {
            "mean": 0.346,
            "ci95": [
              0.3134875,
              0.3725
            ]
          }
        },
        "teacher": {
          "source": "exports/baselines/teacher-dev.json",
          "n": 500,
          "greedy": {
            "mean": 0.892,
            "ci95": [
              0.86395,
              0.92
            ]
          },
          "sampled": {
            "mean": 0.845,
            "ci95": [
              0.822,
              0.8675125
            ]
          }
        },
        "sft_all": {
          "source": "exports/runs/A-all-20260912-105638/eval-final.json",
          "n": 500,
          "greedy": {
            "mean": 0.622,
            "ci95": [
              0.57995,
              0.664
            ]
          },
          "sampled": {
            "mean": 0.5015,
            "ci95": [
              0.469,
              0.5345
            ]
          }
        },
        "sft_correct": {
          "source": "exports/runs/A-correct-20260912-105638/eval-final.json",
          "n": 500,
          "greedy": {
            "mean": 0.638,
            "ci95": [
              0.596,
              0.682
            ]
          },
          "sampled": {
            "mean": 0.5325,
            "ci95": [
              0.4994875,
              0.5655125
            ]
          }
        },
        "on_policy": {
          "source": "exports/runs/C-20260912-153330/eval-final.json",
          "n": 500,
          "greedy": {
            "mean": 0.634,
            "ci95": [
              0.59395,
              0.672
            ]
          },
          "sampled": {
            "mean": 0.6055,
            "ci95": [
              0.5714875,
              0.6395
            ]
          }
        },
        "on_policy_seed2": {
          "source": "exports/runs/C-seed2/eval-final.json",
          "n": 500,
          "greedy": {
            "mean": 0.624,
            "ci95": [
              0.58595,
              0.666
            ]
          },
          "sampled": {
            "mean": 0.597,
            "ci95": [
              0.5635,
              0.6315
            ]
          }
        },
        "reward_to_go": {
          "source": "exports/runs/Cprime-20260912-195939/eval-final.json",
          "n": 500,
          "greedy": {
            "mean": 0.598,
            "ci95": [
              0.558,
              0.64405
            ]
          },
          "sampled": {
            "mean": 0.454,
            "ci95": [
              0.423475,
              0.485
            ]
          }
        },
        "warm_start": {
          "source": "exports/runs/E-20260912-180434/eval-final.json",
          "n": 500,
          "greedy": {
            "mean": 0.67,
            "ci95": [
              0.63195,
              0.71005
            ]
          },
          "sampled": {
            "mean": 0.623,
            "ci95": [
              0.5889875,
              0.6555124999999999
            ]
          }
        }
      },
      "comparisons": {
        "on_policy_minus_sft_correct": {
          "greedy": {
            "delta": -0.0040000000000000036,
            "ci95": [
              -0.040000000000000036,
              0.030000000000000027
            ]
          },
          "sampled": {
            "delta": 0.07300000000000006,
            "ci95": [
              0.04600000000000004,
              0.09899999999999998
            ]
          }
        },
        "warm_start_minus_on_policy": {
          "greedy": {
            "delta": 0.03600000000000003,
            "ci95": [
              0.0,
              0.07400000000000007
            ]
          },
          "sampled": {
            "delta": 0.01749999999999996,
            "ci95": [
              -0.006500000000000061,
              0.03952500000000007
            ]
          }
        }
      }
    },
    "test": {
      "models": {
        "base": {
          "source": "exports/baselines/student-test.json",
          "n": 1319,
          "greedy": {
            "mean": 0.4131918119787718,
            "ci95": [
              0.38817285822592873,
              0.4404852160727824
            ]
          },
          "sampled": {
            "mean": 0.24791508718726307,
            "ci95": [
              0.2318043972706596,
              0.2651677407126611
            ]
          }
        },
        "teacher": {
          "source": "exports/baselines/teacher-test.json",
          "n": 1319,
          "greedy": {
            "mean": 0.8627748294162244,
            "ci95": [
              0.8445602729340409,
              0.8809893858984078
            ]
          },
          "sampled": {
            "mean": 0.8000379075056862,
            "ci95": [
              0.7848749052312358,
              0.8157837376800606
            ]
          }
        },
        "sft_correct": {
          "source": "exports/runs/A-correct-20260912-105638/test.json",
          "n": 1319,
          "greedy": {
            "mean": 0.5443517816527672,
            "ci95": [
              0.5193138741470811,
              0.5724033358605004
            ]
          },
          "sampled": {
            "mean": 0.43953752843062927,
            "ci95": [
              0.41944181197877184,
              0.46096000758150113
            ]
          }
        },
        "on_policy": {
          "source": "exports/runs/C-20260912-153330/test.json",
          "n": 1319,
          "greedy": {
            "mean": 0.5504169825625473,
            "ci95": [
              0.5231235784685367,
              0.576213040181956
            ]
          },
          "sampled": {
            "mean": 0.5159211523881728,
            "ci95": [
              0.4926080363912055,
              0.5377179681576952
            ]
          }
        },
        "warm_start": {
          "source": "exports/runs/E-20260912-180434/test.json",
          "n": 1319,
          "greedy": {
            "mean": 0.5663381349507203,
            "ci95": [
              0.5405420773313115,
              0.5936315390447309
            ]
          },
          "sampled": {
            "mean": 0.5212282031842305,
            "ci95": [
              0.5001895375284306,
              0.5428496967399545
            ]
          }
        }
      },
      "comparisons": {
        "on_policy_minus_sft_correct": {
          "greedy": {
            "delta": 0.006065200909780133,
            "ci95": [
              -0.019730856709628545,
              0.02882865807429873
            ]
          },
          "sampled": {
            "delta": 0.07638362395754356,
            "ci95": [
              0.061026345716451866,
              0.0909780136467021
            ]
          }
        },
        "warm_start_minus_on_policy": {
          "greedy": {
            "delta": 0.015921152388172932,
            "ci95": [
              -0.0076004548900682825,
              0.038665655799848375
            ]
          },
          "sampled": {
            "delta": 0.005307050796057644,
            "ci95": [
              -0.00852918877937836,
              0.019906178923426898
            ]
          }
        }
      }
    }
  },
  "compute": {
    "sft_all": {
      "schema": 1,
      "tokens_sampled": 0,
      "tokens_teacher": 4255376,
      "tokens_trained": 488077,
      "teacher_generated_tokens": 741100,
      "teacher_prefill_tokens": 3514276
    },
    "sft_correct": {
      "schema": 1,
      "tokens_sampled": 0,
      "tokens_teacher": 4255376,
      "tokens_trained": 484707,
      "teacher_generated_tokens": 741100,
      "teacher_prefill_tokens": 3514276
    },
    "on_policy": {
      "schema": 1,
      "tokens_sampled": 761658,
      "tokens_teacher": 4983354,
      "tokens_trained": 761658,
      "teacher_generated_tokens": 0,
      "teacher_prefill_tokens": 4983354,
      "timing_share_measured_components": {
        "t_sample": 0.7177269530061777,
        "t_score": 0.142033386568447,
        "t_train": 0.1402396604253753
      }
    },
    "on_policy_seed2": {
      "schema": 1,
      "tokens_sampled": 796009,
      "tokens_teacher": 5015213,
      "tokens_trained": 796009,
      "teacher_generated_tokens": 0,
      "teacher_prefill_tokens": 5015213,
      "timing_share_measured_components": {
        "t_sample": 0.7286113540025838,
        "t_score": 0.1376706383309923,
        "t_train": 0.13371800766642403
      }
    },
    "reward_to_go": {
      "schema": 1,
      "tokens_sampled": 1987604,
      "tokens_teacher": 6209300,
      "tokens_trained": 1987604,
      "teacher_generated_tokens": 0,
      "teacher_prefill_tokens": 6209300,
      "timing_share_measured_components": {
        "t_sample": 0.7493119041065704,
        "t_score": 0.13990383766842088,
        "t_train": 0.11078425822500867
      }
    },
    "warm_start": {
      "schema": 1,
      "tokens_sampled": 522584,
      "tokens_teacher": 7590296,
      "tokens_trained": 1010661,
      "teacher_generated_tokens": 741100,
      "teacher_prefill_tokens": 6849196,
      "timing_share_measured_components": {
        "t_sample": 0.7398312511752222,
        "t_score": 0.13158701488891678,
        "t_train": 0.1285817339358611
      }
    }
  },
  "training_curves": {
    "on_policy": [
      {
        "step": 0,
        "frac_adv_clamped": 0.023799,
        "resp_len_mean": 68.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.748539,
        "reward_mean": -0.812194
      },
      {
        "step": 1,
        "frac_adv_clamped": 0.021148,
        "resp_len_mean": 82.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.84443,
        "reward_mean": -0.922417
      },
      {
        "step": 2,
        "frac_adv_clamped": 0.022049,
        "resp_len_mean": 72.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.754262,
        "reward_mean": -0.894402
      },
      {
        "step": 3,
        "frac_adv_clamped": 0.01898,
        "resp_len_mean": 83.96875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.814836,
        "reward_mean": -0.902103
      },
      {
        "step": 4,
        "frac_adv_clamped": 0.020462,
        "resp_len_mean": 82.46875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.728701,
        "reward_mean": -0.791037
      },
      {
        "step": 5,
        "frac_adv_clamped": 0.014399,
        "resp_len_mean": 82.46875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.561315,
        "reward_mean": -0.720682
      },
      {
        "step": 6,
        "frac_adv_clamped": 0.003053,
        "resp_len_mean": 61.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.571233,
        "reward_mean": -0.612307
      },
      {
        "step": 7,
        "frac_adv_clamped": 0.007353,
        "resp_len_mean": 68.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.457352,
        "reward_mean": -0.483669
      },
      {
        "step": 8,
        "frac_adv_clamped": 0.013508,
        "resp_len_mean": 76.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.562362,
        "reward_mean": -0.54329
      },
      {
        "step": 9,
        "frac_adv_clamped": 0.007022,
        "resp_len_mean": 97.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.452437,
        "reward_mean": -0.407572
      },
      {
        "step": 10,
        "frac_adv_clamped": 0.009813,
        "resp_len_mean": 70.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.514896,
        "reward_mean": -0.508887
      },
      {
        "step": 11,
        "frac_adv_clamped": 0.006984,
        "resp_len_mean": 93.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.454075,
        "reward_mean": -0.445238
      },
      {
        "step": 12,
        "frac_adv_clamped": 0.00617,
        "resp_len_mean": 70.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.350106,
        "reward_mean": -0.383331
      },
      {
        "step": 13,
        "frac_adv_clamped": 0.004167,
        "resp_len_mean": 75.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.390017,
        "reward_mean": -0.344754
      },
      {
        "step": 14,
        "frac_adv_clamped": 0.00572,
        "resp_len_mean": 92.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.465031,
        "reward_mean": -0.466024
      },
      {
        "step": 15,
        "frac_adv_clamped": 0.009043,
        "resp_len_mean": 65.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.52358,
        "reward_mean": -0.461457
      },
      {
        "step": 16,
        "frac_adv_clamped": 0.003045,
        "resp_len_mean": 71.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.299315,
        "reward_mean": -0.257349
      },
      {
        "step": 17,
        "frac_adv_clamped": 0.007028,
        "resp_len_mean": 75.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.464657,
        "reward_mean": -0.45568
      },
      {
        "step": 18,
        "frac_adv_clamped": 0.009561,
        "resp_len_mean": 88.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.512889,
        "reward_mean": -0.550056
      },
      {
        "step": 19,
        "frac_adv_clamped": 0.008446,
        "resp_len_mean": 74.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.476185,
        "reward_mean": -0.508681
      },
      {
        "step": 20,
        "frac_adv_clamped": 0.008948,
        "resp_len_mean": 73.34375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.508264,
        "reward_mean": -0.473504
      },
      {
        "step": 21,
        "frac_adv_clamped": 0.005422,
        "resp_len_mean": 80.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.430649,
        "reward_mean": -0.397382
      },
      {
        "step": 22,
        "frac_adv_clamped": 0.008019,
        "resp_len_mean": 93.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.427717,
        "reward_mean": -0.441472
      },
      {
        "step": 23,
        "frac_adv_clamped": 0.006612,
        "resp_len_mean": 56.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.294679,
        "reward_mean": -0.274023
      },
      {
        "step": 24,
        "frac_adv_clamped": 0.005601,
        "resp_len_mean": 61.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.366339,
        "reward_mean": -0.370924
      },
      {
        "step": 25,
        "frac_adv_clamped": 0.004937,
        "resp_len_mean": 82.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.374331,
        "reward_mean": -0.371641
      },
      {
        "step": 26,
        "frac_adv_clamped": 0.008572,
        "resp_len_mean": 83.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.40409,
        "reward_mean": -0.387264
      },
      {
        "step": 27,
        "frac_adv_clamped": 0.003156,
        "resp_len_mean": 59.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.470019,
        "reward_mean": -0.440988
      },
      {
        "step": 28,
        "frac_adv_clamped": 0.013072,
        "resp_len_mean": 71.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.543211,
        "reward_mean": -0.533953
      },
      {
        "step": 29,
        "frac_adv_clamped": 0.005232,
        "resp_len_mean": 89.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.376918,
        "reward_mean": -0.362875
      },
      {
        "step": 30,
        "frac_adv_clamped": 0.004982,
        "resp_len_mean": 69.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.427143,
        "reward_mean": -0.408949
      },
      {
        "step": 31,
        "frac_adv_clamped": 0.008803,
        "resp_len_mean": 71.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.394369,
        "reward_mean": -0.378109
      },
      {
        "step": 32,
        "frac_adv_clamped": 0.005523,
        "resp_len_mean": 84.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.321398,
        "reward_mean": -0.278247
      },
      {
        "step": 33,
        "frac_adv_clamped": 0.005901,
        "resp_len_mean": 100.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.36514,
        "reward_mean": -0.37826
      },
      {
        "step": 34,
        "frac_adv_clamped": 0.004743,
        "resp_len_mean": 92.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.306634,
        "reward_mean": -0.282693
      },
      {
        "step": 35,
        "frac_adv_clamped": 0.00317,
        "resp_len_mean": 88.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.33216,
        "reward_mean": -0.322188
      },
      {
        "step": 36,
        "frac_adv_clamped": 0.0057,
        "resp_len_mean": 76.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.335916,
        "reward_mean": -0.320617
      },
      {
        "step": 37,
        "frac_adv_clamped": 0.003574,
        "resp_len_mean": 87.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.347018,
        "reward_mean": -0.337652
      },
      {
        "step": 38,
        "frac_adv_clamped": 0.006162,
        "resp_len_mean": 86.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.39055,
        "reward_mean": -0.374434
      },
      {
        "step": 39,
        "frac_adv_clamped": 0.007029,
        "resp_len_mean": 80.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.531581,
        "reward_mean": -0.522618
      },
      {
        "step": 40,
        "frac_adv_clamped": 0.004496,
        "resp_len_mean": 104.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.39265,
        "reward_mean": -0.370084
      },
      {
        "step": 41,
        "frac_adv_clamped": 0.00866,
        "resp_len_mean": 83.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.54722,
        "reward_mean": -0.559187
      },
      {
        "step": 42,
        "frac_adv_clamped": 0.003788,
        "resp_len_mean": 99.0,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.365858,
        "reward_mean": -0.351373
      },
      {
        "step": 43,
        "frac_adv_clamped": 0.005071,
        "resp_len_mean": 86.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.331327,
        "reward_mean": -0.31924
      },
      {
        "step": 44,
        "frac_adv_clamped": 0.005348,
        "resp_len_mean": 58.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.361428,
        "reward_mean": -0.333128
      },
      {
        "step": 45,
        "frac_adv_clamped": 0.001464,
        "resp_len_mean": 85.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.284747,
        "reward_mean": -0.266848
      },
      {
        "step": 46,
        "frac_adv_clamped": 0.004878,
        "resp_len_mean": 96.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.374882,
        "reward_mean": -0.349244
      },
      {
        "step": 47,
        "frac_adv_clamped": 0.004065,
        "resp_len_mean": 84.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.398751,
        "reward_mean": -0.385449
      },
      {
        "step": 48,
        "frac_adv_clamped": 0.003639,
        "resp_len_mean": 85.875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.302748,
        "reward_mean": -0.259833
      },
      {
        "step": 49,
        "frac_adv_clamped": 0.006305,
        "resp_len_mean": 64.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.300227,
        "reward_mean": -0.267135
      },
      {
        "step": 50,
        "frac_adv_clamped": 0.006008,
        "resp_len_mean": 83.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.388908,
        "reward_mean": -0.3821
      },
      {
        "step": 51,
        "frac_adv_clamped": 0.006515,
        "resp_len_mean": 67.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.403428,
        "reward_mean": -0.379905
      },
      {
        "step": 52,
        "frac_adv_clamped": 0.005306,
        "resp_len_mean": 76.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.400773,
        "reward_mean": -0.392775
      },
      {
        "step": 53,
        "frac_adv_clamped": 0.005396,
        "resp_len_mean": 75.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.400712,
        "reward_mean": -0.371071
      },
      {
        "step": 54,
        "frac_adv_clamped": 0.004274,
        "resp_len_mean": 58.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.36677,
        "reward_mean": -0.340163
      },
      {
        "step": 55,
        "frac_adv_clamped": 0.00656,
        "resp_len_mean": 76.21875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.444997,
        "reward_mean": -0.419679
      },
      {
        "step": 56,
        "frac_adv_clamped": 0.012919,
        "resp_len_mean": 65.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.446662,
        "reward_mean": -0.468203
      },
      {
        "step": 57,
        "frac_adv_clamped": 0.008319,
        "resp_len_mean": 71.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.450969,
        "reward_mean": -0.443695
      },
      {
        "step": 58,
        "frac_adv_clamped": 0.00328,
        "resp_len_mean": 66.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.406074,
        "reward_mean": -0.382848
      },
      {
        "step": 59,
        "frac_adv_clamped": 0.005638,
        "resp_len_mean": 77.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.34718,
        "reward_mean": -0.340745
      },
      {
        "step": 60,
        "frac_adv_clamped": 0.006165,
        "resp_len_mean": 76.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.391894,
        "reward_mean": -0.347677
      },
      {
        "step": 61,
        "frac_adv_clamped": 0.008897,
        "resp_len_mean": 70.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.440286,
        "reward_mean": -0.429291
      },
      {
        "step": 62,
        "frac_adv_clamped": 0.008115,
        "resp_len_mean": 69.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.409226,
        "reward_mean": -0.411469
      },
      {
        "step": 63,
        "frac_adv_clamped": 0.007282,
        "resp_len_mean": 77.25,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.458638,
        "reward_mean": -0.425165
      },
      {
        "step": 64,
        "frac_adv_clamped": 0.003876,
        "resp_len_mean": 64.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.375251,
        "reward_mean": -0.36057
      },
      {
        "step": 65,
        "frac_adv_clamped": 0.00694,
        "resp_len_mean": 90.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.427847,
        "reward_mean": -0.433016
      },
      {
        "step": 66,
        "frac_adv_clamped": 0.007703,
        "resp_len_mean": 64.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.376632,
        "reward_mean": -0.32659
      },
      {
        "step": 67,
        "frac_adv_clamped": 0.002957,
        "resp_len_mean": 105.6875,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.398243,
        "reward_mean": -0.386193
      },
      {
        "step": 68,
        "frac_adv_clamped": 0.004557,
        "resp_len_mean": 102.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.301577,
        "reward_mean": -0.291516
      },
      {
        "step": 69,
        "frac_adv_clamped": 0.004781,
        "resp_len_mean": 71.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.328461,
        "reward_mean": -0.294936
      },
      {
        "step": 70,
        "frac_adv_clamped": 0.00552,
        "resp_len_mean": 67.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.442083,
        "reward_mean": -0.403026
      },
      {
        "step": 71,
        "frac_adv_clamped": 0.004719,
        "resp_len_mean": 72.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.433763,
        "reward_mean": -0.438223
      },
      {
        "step": 72,
        "frac_adv_clamped": 0.005023,
        "resp_len_mean": 93.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.30457,
        "reward_mean": -0.314742
      },
      {
        "step": 73,
        "frac_adv_clamped": 0.00661,
        "resp_len_mean": 80.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.37874,
        "reward_mean": -0.376113
      },
      {
        "step": 74,
        "frac_adv_clamped": 0.00152,
        "resp_len_mean": 82.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.26647,
        "reward_mean": -0.230685
      },
      {
        "step": 75,
        "frac_adv_clamped": 0.002349,
        "resp_len_mean": 66.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.338421,
        "reward_mean": -0.31145
      },
      {
        "step": 76,
        "frac_adv_clamped": 0.001256,
        "resp_len_mean": 74.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.239375,
        "reward_mean": -0.207885
      },
      {
        "step": 77,
        "frac_adv_clamped": 0.005395,
        "resp_len_mean": 63.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.343039,
        "reward_mean": -0.313362
      },
      {
        "step": 78,
        "frac_adv_clamped": 0.008061,
        "resp_len_mean": 77.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.477286,
        "reward_mean": -0.502939
      },
      {
        "step": 79,
        "frac_adv_clamped": 0.005809,
        "resp_len_mean": 69.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.464797,
        "reward_mean": -0.447305
      },
      {
        "step": 80,
        "frac_adv_clamped": 0.003477,
        "resp_len_mean": 62.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.360683,
        "reward_mean": -0.360476
      },
      {
        "step": 81,
        "frac_adv_clamped": 0.004562,
        "resp_len_mean": 61.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.332616,
        "reward_mean": -0.296572
      },
      {
        "step": 82,
        "frac_adv_clamped": 0.007072,
        "resp_len_mean": 97.21875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.371883,
        "reward_mean": -0.382227
      },
      {
        "step": 83,
        "frac_adv_clamped": 0.003429,
        "resp_len_mean": 100.25,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.319002,
        "reward_mean": -0.314806
      },
      {
        "step": 84,
        "frac_adv_clamped": 0.006905,
        "resp_len_mean": 72.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.4347,
        "reward_mean": -0.421646
      },
      {
        "step": 85,
        "frac_adv_clamped": 0.004845,
        "resp_len_mean": 83.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.266379,
        "reward_mean": -0.260024
      },
      {
        "step": 86,
        "frac_adv_clamped": 0.005398,
        "resp_len_mean": 46.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.357801,
        "reward_mean": -0.333406
      },
      {
        "step": 87,
        "frac_adv_clamped": 0.008095,
        "resp_len_mean": 73.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.306108,
        "reward_mean": -0.329518
      },
      {
        "step": 88,
        "frac_adv_clamped": 0.004455,
        "resp_len_mean": 91.1875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.280056,
        "reward_mean": -0.28803
      },
      {
        "step": 89,
        "frac_adv_clamped": 0.003942,
        "resp_len_mean": 103.0625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.429159,
        "reward_mean": -0.401272
      },
      {
        "step": 90,
        "frac_adv_clamped": 0.007764,
        "resp_len_mean": 80.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.442263,
        "reward_mean": -0.437433
      },
      {
        "step": 91,
        "frac_adv_clamped": 0.004936,
        "resp_len_mean": 75.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.405761,
        "reward_mean": -0.409665
      },
      {
        "step": 92,
        "frac_adv_clamped": 0.003518,
        "resp_len_mean": 79.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.22701,
        "reward_mean": -0.212582
      },
      {
        "step": 93,
        "frac_adv_clamped": 0.005556,
        "resp_len_mean": 73.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.368268,
        "reward_mean": -0.361016
      },
      {
        "step": 94,
        "frac_adv_clamped": 0.003497,
        "resp_len_mean": 89.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.253439,
        "reward_mean": -0.249179
      },
      {
        "step": 95,
        "frac_adv_clamped": 0.007286,
        "resp_len_mean": 51.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.357791,
        "reward_mean": -0.355432
      },
      {
        "step": 96,
        "frac_adv_clamped": 0.002472,
        "resp_len_mean": 75.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.230383,
        "reward_mean": -0.220111
      },
      {
        "step": 97,
        "frac_adv_clamped": 0.004692,
        "resp_len_mean": 46.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.347775,
        "reward_mean": -0.319778
      },
      {
        "step": 98,
        "frac_adv_clamped": 0.003583,
        "resp_len_mean": 69.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.29474,
        "reward_mean": -0.249843
      },
      {
        "step": 99,
        "frac_adv_clamped": 0.010978,
        "resp_len_mean": 59.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.447518,
        "reward_mean": -0.439695
      },
      {
        "step": 100,
        "frac_adv_clamped": 0.003489,
        "resp_len_mean": 71.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.255118,
        "reward_mean": -0.245259
      },
      {
        "step": 101,
        "frac_adv_clamped": 0.005582,
        "resp_len_mean": 83.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.29858,
        "reward_mean": -0.297374
      },
      {
        "step": 102,
        "frac_adv_clamped": 0.009647,
        "resp_len_mean": 103.65625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.440003,
        "reward_mean": -0.431214
      },
      {
        "step": 103,
        "frac_adv_clamped": 0.004723,
        "resp_len_mean": 79.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.306236,
        "reward_mean": -0.286466
      },
      {
        "step": 104,
        "frac_adv_clamped": 0.004125,
        "resp_len_mean": 90.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.341397,
        "reward_mean": -0.34994
      },
      {
        "step": 105,
        "frac_adv_clamped": 0.001131,
        "resp_len_mean": 110.5625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.368474,
        "reward_mean": -0.350516
      },
      {
        "step": 106,
        "frac_adv_clamped": 0.010042,
        "resp_len_mean": 80.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.468549,
        "reward_mean": -0.430639
      },
      {
        "step": 107,
        "frac_adv_clamped": 0.007045,
        "resp_len_mean": 70.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.372734,
        "reward_mean": -0.342277
      },
      {
        "step": 108,
        "frac_adv_clamped": 0.002547,
        "resp_len_mean": 85.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.288335,
        "reward_mean": -0.292791
      },
      {
        "step": 109,
        "frac_adv_clamped": 0.005242,
        "resp_len_mean": 77.5,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.339623,
        "reward_mean": -0.308505
      },
      {
        "step": 110,
        "frac_adv_clamped": 0.005656,
        "resp_len_mean": 88.40625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.331097,
        "reward_mean": -0.332479
      },
      {
        "step": 111,
        "frac_adv_clamped": 0.00449,
        "resp_len_mean": 76.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.246348,
        "reward_mean": -0.271933
      },
      {
        "step": 112,
        "frac_adv_clamped": 0.004829,
        "resp_len_mean": 84.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.384782,
        "reward_mean": -0.351255
      },
      {
        "step": 113,
        "frac_adv_clamped": 0.004331,
        "resp_len_mean": 72.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.379705,
        "reward_mean": -0.369446
      },
      {
        "step": 114,
        "frac_adv_clamped": 0.004804,
        "resp_len_mean": 78.0625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.390613,
        "reward_mean": -0.335173
      },
      {
        "step": 115,
        "frac_adv_clamped": 0.005592,
        "resp_len_mean": 89.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.29151,
        "reward_mean": -0.280941
      },
      {
        "step": 116,
        "frac_adv_clamped": 0.005147,
        "resp_len_mean": 66.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.394706,
        "reward_mean": -0.361742
      },
      {
        "step": 117,
        "frac_adv_clamped": 0.004426,
        "resp_len_mean": 84.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.328144,
        "reward_mean": -0.327605
      },
      {
        "step": 118,
        "frac_adv_clamped": 0.004089,
        "resp_len_mean": 68.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.331748,
        "reward_mean": -0.329155
      },
      {
        "step": 119,
        "frac_adv_clamped": 0.002913,
        "resp_len_mean": 64.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.267429,
        "reward_mean": -0.26795
      },
      {
        "step": 120,
        "frac_adv_clamped": 0.004362,
        "resp_len_mean": 78.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.29081,
        "reward_mean": -0.278118
      },
      {
        "step": 121,
        "frac_adv_clamped": 0.004109,
        "resp_len_mean": 60.84375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.330805,
        "reward_mean": -0.302261
      },
      {
        "step": 122,
        "frac_adv_clamped": 0.005357,
        "resp_len_mean": 93.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.433332,
        "reward_mean": -0.421328
      },
      {
        "step": 123,
        "frac_adv_clamped": 0.005792,
        "resp_len_mean": 107.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.409438,
        "reward_mean": -0.392521
      },
      {
        "step": 124,
        "frac_adv_clamped": 0.004103,
        "resp_len_mean": 106.625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.286147,
        "reward_mean": -0.285471
      },
      {
        "step": 125,
        "frac_adv_clamped": 0.003524,
        "resp_len_mean": 79.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.36245,
        "reward_mean": -0.363309
      },
      {
        "step": 126,
        "frac_adv_clamped": 0.004878,
        "resp_len_mean": 70.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.280534,
        "reward_mean": -0.27516
      },
      {
        "step": 127,
        "frac_adv_clamped": 0.007906,
        "resp_len_mean": 90.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.404213,
        "reward_mean": -0.359922
      },
      {
        "step": 128,
        "frac_adv_clamped": 0.002673,
        "resp_len_mean": 81.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.201433,
        "reward_mean": -0.218494
      },
      {
        "step": 129,
        "frac_adv_clamped": 0.004711,
        "resp_len_mean": 72.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.356267,
        "reward_mean": -0.343746
      },
      {
        "step": 130,
        "frac_adv_clamped": 0.000485,
        "resp_len_mean": 64.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.254723,
        "reward_mean": -0.218046
      },
      {
        "step": 131,
        "frac_adv_clamped": 0.001771,
        "resp_len_mean": 88.25,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.294554,
        "reward_mean": -0.273848
      },
      {
        "step": 132,
        "frac_adv_clamped": 0.003701,
        "resp_len_mean": 92.875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.284351,
        "reward_mean": -0.275628
      },
      {
        "step": 133,
        "frac_adv_clamped": 0.004574,
        "resp_len_mean": 75.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.292748,
        "reward_mean": -0.285539
      },
      {
        "step": 134,
        "frac_adv_clamped": 0.006452,
        "resp_len_mean": 67.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.282707,
        "reward_mean": -0.285545
      },
      {
        "step": 135,
        "frac_adv_clamped": 0.004838,
        "resp_len_mean": 64.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.380685,
        "reward_mean": -0.382191
      },
      {
        "step": 136,
        "frac_adv_clamped": 0.003062,
        "resp_len_mean": 51.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.324319,
        "reward_mean": -0.283989
      },
      {
        "step": 137,
        "frac_adv_clamped": 0.003872,
        "resp_len_mean": 64.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.305964,
        "reward_mean": -0.326877
      },
      {
        "step": 138,
        "frac_adv_clamped": 0.004814,
        "resp_len_mean": 71.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.343365,
        "reward_mean": -0.337292
      },
      {
        "step": 139,
        "frac_adv_clamped": 0.005184,
        "resp_len_mean": 60.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.309133,
        "reward_mean": -0.29652
      },
      {
        "step": 140,
        "frac_adv_clamped": 0.003382,
        "resp_len_mean": 55.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.295735,
        "reward_mean": -0.26225
      },
      {
        "step": 141,
        "frac_adv_clamped": 0.006916,
        "resp_len_mean": 54.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.369528,
        "reward_mean": -0.350559
      },
      {
        "step": 142,
        "frac_adv_clamped": 0.003223,
        "resp_len_mean": 77.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.258174,
        "reward_mean": -0.257609
      },
      {
        "step": 143,
        "frac_adv_clamped": 0.006692,
        "resp_len_mean": 65.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.289441,
        "reward_mean": -0.256607
      },
      {
        "step": 144,
        "frac_adv_clamped": 0.00577,
        "resp_len_mean": 70.40625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.347815,
        "reward_mean": -0.312583
      },
      {
        "step": 145,
        "frac_adv_clamped": 0.006224,
        "resp_len_mean": 90.375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.416951,
        "reward_mean": -0.424491
      },
      {
        "step": 146,
        "frac_adv_clamped": 0.005464,
        "resp_len_mean": 97.21875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.346627,
        "reward_mean": -0.353798
      },
      {
        "step": 147,
        "frac_adv_clamped": 0.003697,
        "resp_len_mean": 67.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.240969,
        "reward_mean": -0.236188
      },
      {
        "step": 148,
        "frac_adv_clamped": 0.003332,
        "resp_len_mean": 75.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.291481,
        "reward_mean": -0.315096
      },
      {
        "step": 149,
        "frac_adv_clamped": 0.014153,
        "resp_len_mean": 64.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.501345,
        "reward_mean": -0.500766
      },
      {
        "step": 150,
        "frac_adv_clamped": 0.002424,
        "resp_len_mean": 64.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.249265,
        "reward_mean": -0.207583
      },
      {
        "step": 151,
        "frac_adv_clamped": 0.004167,
        "resp_len_mean": 67.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.339515,
        "reward_mean": -0.321167
      },
      {
        "step": 152,
        "frac_adv_clamped": 0.005773,
        "resp_len_mean": 75.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.306982,
        "reward_mean": -0.290442
      },
      {
        "step": 153,
        "frac_adv_clamped": 0.002422,
        "resp_len_mean": 64.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.213985,
        "reward_mean": -0.20358
      },
      {
        "step": 154,
        "frac_adv_clamped": 0.003687,
        "resp_len_mean": 84.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.311893,
        "reward_mean": -0.291407
      },
      {
        "step": 155,
        "frac_adv_clamped": 0.010212,
        "resp_len_mean": 76.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.411799,
        "reward_mean": -0.380839
      },
      {
        "step": 156,
        "frac_adv_clamped": 0.003072,
        "resp_len_mean": 61.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.272804,
        "reward_mean": -0.264872
      },
      {
        "step": 157,
        "frac_adv_clamped": 0.006061,
        "resp_len_mean": 82.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.344025,
        "reward_mean": -0.337855
      },
      {
        "step": 158,
        "frac_adv_clamped": 0.003617,
        "resp_len_mean": 69.125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.31177,
        "reward_mean": -0.30409
      },
      {
        "step": 159,
        "frac_adv_clamped": 0.003928,
        "resp_len_mean": 71.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.251474,
        "reward_mean": -0.231653
      },
      {
        "step": 160,
        "frac_adv_clamped": 0.004664,
        "resp_len_mean": 67.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.29862,
        "reward_mean": -0.24438
      },
      {
        "step": 161,
        "frac_adv_clamped": 0.00314,
        "resp_len_mean": 59.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.222706,
        "reward_mean": -0.202281
      },
      {
        "step": 162,
        "frac_adv_clamped": 0.001809,
        "resp_len_mean": 69.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.252615,
        "reward_mean": -0.27726
      },
      {
        "step": 163,
        "frac_adv_clamped": 0.004783,
        "resp_len_mean": 84.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.415029,
        "reward_mean": -0.38992
      },
      {
        "step": 164,
        "frac_adv_clamped": 0.003784,
        "resp_len_mean": 82.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.382268,
        "reward_mean": -0.385132
      },
      {
        "step": 165,
        "frac_adv_clamped": 0.001792,
        "resp_len_mean": 69.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.317581,
        "reward_mean": -0.27034
      },
      {
        "step": 166,
        "frac_adv_clamped": 0.003044,
        "resp_len_mean": 61.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.322493,
        "reward_mean": -0.321527
      },
      {
        "step": 167,
        "frac_adv_clamped": 0.007738,
        "resp_len_mean": 68.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.298859,
        "reward_mean": -0.31413
      },
      {
        "step": 168,
        "frac_adv_clamped": 0.006336,
        "resp_len_mean": 83.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.319707,
        "reward_mean": -0.336065
      },
      {
        "step": 169,
        "frac_adv_clamped": 0.004018,
        "resp_len_mean": 54.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.430436,
        "reward_mean": -0.392022
      },
      {
        "step": 170,
        "frac_adv_clamped": 0.003229,
        "resp_len_mean": 67.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.258078,
        "reward_mean": -0.228948
      },
      {
        "step": 171,
        "frac_adv_clamped": 0.005023,
        "resp_len_mean": 93.3125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.329241,
        "reward_mean": -0.313776
      },
      {
        "step": 172,
        "frac_adv_clamped": 0.001518,
        "resp_len_mean": 102.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.327691,
        "reward_mean": -0.302819
      },
      {
        "step": 173,
        "frac_adv_clamped": 0.005416,
        "resp_len_mean": 80.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.340346,
        "reward_mean": -0.330476
      },
      {
        "step": 174,
        "frac_adv_clamped": 0.007282,
        "resp_len_mean": 94.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.378897,
        "reward_mean": -0.399843
      },
      {
        "step": 175,
        "frac_adv_clamped": 0.006403,
        "resp_len_mean": 78.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.435609,
        "reward_mean": -0.436747
      },
      {
        "step": 176,
        "frac_adv_clamped": 0.006598,
        "resp_len_mean": 66.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.35267,
        "reward_mean": -0.339027
      },
      {
        "step": 177,
        "frac_adv_clamped": 0.003358,
        "resp_len_mean": 93.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.273455,
        "reward_mean": -0.258694
      },
      {
        "step": 178,
        "frac_adv_clamped": 0.004988,
        "resp_len_mean": 62.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.284584,
        "reward_mean": -0.302745
      },
      {
        "step": 179,
        "frac_adv_clamped": 0.007171,
        "resp_len_mean": 78.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.449909,
        "reward_mean": -0.47072
      },
      {
        "step": 180,
        "frac_adv_clamped": 0.001909,
        "resp_len_mean": 81.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.284459,
        "reward_mean": -0.269298
      },
      {
        "step": 181,
        "frac_adv_clamped": 0.002481,
        "resp_len_mean": 88.1875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.328624,
        "reward_mean": -0.317147
      },
      {
        "step": 182,
        "frac_adv_clamped": 0.003204,
        "resp_len_mean": 97.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.327773,
        "reward_mean": -0.331224
      },
      {
        "step": 183,
        "frac_adv_clamped": 0.006242,
        "resp_len_mean": 95.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.354937,
        "reward_mean": -0.348719
      },
      {
        "step": 184,
        "frac_adv_clamped": 0.004992,
        "resp_len_mean": 100.15625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.394685,
        "reward_mean": -0.37217
      },
      {
        "step": 185,
        "frac_adv_clamped": 0.00584,
        "resp_len_mean": 112.375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.415265,
        "reward_mean": -0.404623
      },
      {
        "step": 186,
        "frac_adv_clamped": 0.004899,
        "resp_len_mean": 44.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.314019,
        "reward_mean": -0.260553
      },
      {
        "step": 187,
        "frac_adv_clamped": 0.001493,
        "resp_len_mean": 62.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.210279,
        "reward_mean": -0.189304
      },
      {
        "step": 188,
        "frac_adv_clamped": 0.002575,
        "resp_len_mean": 97.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.284617,
        "reward_mean": -0.267625
      },
      {
        "step": 189,
        "frac_adv_clamped": 0.005263,
        "resp_len_mean": 83.125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.330649,
        "reward_mean": -0.312185
      },
      {
        "step": 190,
        "frac_adv_clamped": 0.007398,
        "resp_len_mean": 71.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.322575,
        "reward_mean": -0.32567
      },
      {
        "step": 191,
        "frac_adv_clamped": 0.006887,
        "resp_len_mean": 86.21875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.405544,
        "reward_mean": -0.405384
      },
      {
        "step": 192,
        "frac_adv_clamped": 0.002396,
        "resp_len_mean": 91.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.317646,
        "reward_mean": -0.308463
      },
      {
        "step": 193,
        "frac_adv_clamped": 0.001046,
        "resp_len_mean": 89.59375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.251113,
        "reward_mean": -0.24791
      },
      {
        "step": 194,
        "frac_adv_clamped": 0.003995,
        "resp_len_mean": 54.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.360549,
        "reward_mean": -0.360193
      },
      {
        "step": 195,
        "frac_adv_clamped": 0.003724,
        "resp_len_mean": 83.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.367997,
        "reward_mean": -0.361668
      },
      {
        "step": 196,
        "frac_adv_clamped": 0.002966,
        "resp_len_mean": 84.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.355724,
        "reward_mean": -0.334727
      },
      {
        "step": 197,
        "frac_adv_clamped": 0.004805,
        "resp_len_mean": 58.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.338908,
        "reward_mean": -0.326179
      },
      {
        "step": 198,
        "frac_adv_clamped": 0.008563,
        "resp_len_mean": 83.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.423283,
        "reward_mean": -0.432026
      },
      {
        "step": 199,
        "frac_adv_clamped": 0.003363,
        "resp_len_mean": 74.34375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.311367,
        "reward_mean": -0.300414
      },
      {
        "step": 200,
        "frac_adv_clamped": 0.002396,
        "resp_len_mean": 91.3125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.316488,
        "reward_mean": -0.316784
      },
      {
        "step": 201,
        "frac_adv_clamped": 0.005796,
        "resp_len_mean": 70.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.350504,
        "reward_mean": -0.358955
      },
      {
        "step": 202,
        "frac_adv_clamped": 0.003173,
        "resp_len_mean": 68.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.271927,
        "reward_mean": -0.255294
      },
      {
        "step": 203,
        "frac_adv_clamped": 0.003384,
        "resp_len_mean": 73.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.23115,
        "reward_mean": -0.202199
      },
      {
        "step": 204,
        "frac_adv_clamped": 0.007407,
        "resp_len_mean": 71.71875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.323509,
        "reward_mean": -0.319552
      },
      {
        "step": 205,
        "frac_adv_clamped": 0.008633,
        "resp_len_mean": 90.5,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.432452,
        "reward_mean": -0.395515
      },
      {
        "step": 206,
        "frac_adv_clamped": 0.004898,
        "resp_len_mean": 70.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.385829,
        "reward_mean": -0.375369
      },
      {
        "step": 207,
        "frac_adv_clamped": 0.005314,
        "resp_len_mean": 94.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.295242,
        "reward_mean": -0.29668
      },
      {
        "step": 208,
        "frac_adv_clamped": 0.005458,
        "resp_len_mean": 80.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.34677,
        "reward_mean": -0.329261
      },
      {
        "step": 209,
        "frac_adv_clamped": 0.005466,
        "resp_len_mean": 102.90625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.426328,
        "reward_mean": -0.396577
      },
      {
        "step": 210,
        "frac_adv_clamped": 0.005578,
        "resp_len_mean": 78.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.236979,
        "reward_mean": -0.250968
      },
      {
        "step": 211,
        "frac_adv_clamped": 0.00788,
        "resp_len_mean": 83.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.377584,
        "reward_mean": -0.373835
      },
      {
        "step": 212,
        "frac_adv_clamped": 0.003416,
        "resp_len_mean": 109.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.276974,
        "reward_mean": -0.273288
      },
      {
        "step": 213,
        "frac_adv_clamped": 0.006265,
        "resp_len_mean": 79.8125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.375088,
        "reward_mean": -0.369557
      },
      {
        "step": 214,
        "frac_adv_clamped": 0.006462,
        "resp_len_mean": 58.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.264515,
        "reward_mean": -0.226785
      },
      {
        "step": 215,
        "frac_adv_clamped": 0.007497,
        "resp_len_mean": 75.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.308687,
        "reward_mean": -0.350692
      },
      {
        "step": 216,
        "frac_adv_clamped": 0.004562,
        "resp_len_mean": 95.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.355983,
        "reward_mean": -0.37927
      },
      {
        "step": 217,
        "frac_adv_clamped": 0.004836,
        "resp_len_mean": 84.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.377612,
        "reward_mean": -0.364178
      },
      {
        "step": 218,
        "frac_adv_clamped": 0.003836,
        "resp_len_mean": 122.1875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.312617,
        "reward_mean": -0.296525
      },
      {
        "step": 219,
        "frac_adv_clamped": 0.003154,
        "resp_len_mean": 99.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.336515,
        "reward_mean": -0.327998
      },
      {
        "step": 220,
        "frac_adv_clamped": 0.002909,
        "resp_len_mean": 85.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.343502,
        "reward_mean": -0.355227
      },
      {
        "step": 221,
        "frac_adv_clamped": 0.003126,
        "resp_len_mean": 69.96875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.280186,
        "reward_mean": -0.264252
      },
      {
        "step": 222,
        "frac_adv_clamped": 0.002199,
        "resp_len_mean": 71.0625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.190252,
        "reward_mean": -0.171081
      },
      {
        "step": 223,
        "frac_adv_clamped": 0.004007,
        "resp_len_mean": 101.375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.328576,
        "reward_mean": -0.346992
      },
      {
        "step": 224,
        "frac_adv_clamped": 0.007256,
        "resp_len_mean": 73.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.328911,
        "reward_mean": -0.344041
      },
      {
        "step": 225,
        "frac_adv_clamped": 0.002015,
        "resp_len_mean": 93.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.390622,
        "reward_mean": -0.379749
      },
      {
        "step": 226,
        "frac_adv_clamped": 0.007839,
        "resp_len_mean": 91.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.408119,
        "reward_mean": -0.410299
      },
      {
        "step": 227,
        "frac_adv_clamped": 0.004185,
        "resp_len_mean": 97.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.332256,
        "reward_mean": -0.322422
      },
      {
        "step": 228,
        "frac_adv_clamped": 0.006564,
        "resp_len_mean": 66.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.404428,
        "reward_mean": -0.377985
      },
      {
        "step": 229,
        "frac_adv_clamped": 0.001723,
        "resp_len_mean": 54.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.190071,
        "reward_mean": -0.178296
      },
      {
        "step": 230,
        "frac_adv_clamped": 0.002106,
        "resp_len_mean": 103.875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.330928,
        "reward_mean": -0.301227
      },
      {
        "step": 231,
        "frac_adv_clamped": 0.004585,
        "resp_len_mean": 74.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.301939,
        "reward_mean": -0.278043
      },
      {
        "step": 232,
        "frac_adv_clamped": 0.003638,
        "resp_len_mean": 68.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.343497,
        "reward_mean": -0.306894
      },
      {
        "step": 233,
        "frac_adv_clamped": 0.001981,
        "resp_len_mean": 63.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.289178,
        "reward_mean": -0.271303
      },
      {
        "step": 234,
        "frac_adv_clamped": 0.004834,
        "resp_len_mean": 109.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.383008,
        "reward_mean": -0.37276
      },
      {
        "step": 235,
        "frac_adv_clamped": 0.004195,
        "resp_len_mean": 81.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.342282,
        "reward_mean": -0.291811
      },
      {
        "step": 236,
        "frac_adv_clamped": 0.005315,
        "resp_len_mean": 76.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.309778,
        "reward_mean": -0.29399
      },
      {
        "step": 237,
        "frac_adv_clamped": 0.002145,
        "resp_len_mean": 87.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.313426,
        "reward_mean": -0.308843
      },
      {
        "step": 238,
        "frac_adv_clamped": 0.003503,
        "resp_len_mean": 89.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.281473,
        "reward_mean": -0.267129
      },
      {
        "step": 239,
        "frac_adv_clamped": 0.007484,
        "resp_len_mean": 75.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.3048,
        "reward_mean": -0.323932
      },
      {
        "step": 240,
        "frac_adv_clamped": 0.004975,
        "resp_len_mean": 75.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.320049,
        "reward_mean": -0.29063
      },
      {
        "step": 241,
        "frac_adv_clamped": 0.002579,
        "resp_len_mean": 96.9375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.241587,
        "reward_mean": -0.240577
      },
      {
        "step": 242,
        "frac_adv_clamped": 0.00814,
        "resp_len_mean": 80.625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.336803,
        "reward_mean": -0.33555
      },
      {
        "step": 243,
        "frac_adv_clamped": 0.001964,
        "resp_len_mean": 95.46875,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.352934,
        "reward_mean": -0.327725
      },
      {
        "step": 244,
        "frac_adv_clamped": 0.0029,
        "resp_len_mean": 86.21875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.342131,
        "reward_mean": -0.317265
      },
      {
        "step": 245,
        "frac_adv_clamped": 0.004138,
        "resp_len_mean": 90.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.289316,
        "reward_mean": -0.276288
      },
      {
        "step": 246,
        "frac_adv_clamped": 0.005716,
        "resp_len_mean": 92.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.374099,
        "reward_mean": -0.362045
      },
      {
        "step": 247,
        "frac_adv_clamped": 0.004165,
        "resp_len_mean": 82.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.274752,
        "reward_mean": -0.255538
      },
      {
        "step": 248,
        "frac_adv_clamped": 0.004544,
        "resp_len_mean": 96.28125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.407924,
        "reward_mean": -0.425376
      },
      {
        "step": 249,
        "frac_adv_clamped": 0.003877,
        "resp_len_mean": 88.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.363214,
        "reward_mean": -0.346264
      },
      {
        "step": 250,
        "frac_adv_clamped": 0.002436,
        "resp_len_mean": 89.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.324452,
        "reward_mean": -0.314742
      },
      {
        "step": 251,
        "frac_adv_clamped": 0.002462,
        "resp_len_mean": 88.84375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.318367,
        "reward_mean": -0.289296
      },
      {
        "step": 252,
        "frac_adv_clamped": 0.004687,
        "resp_len_mean": 73.34375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.30552,
        "reward_mean": -0.29312
      },
      {
        "step": 253,
        "frac_adv_clamped": 0.005418,
        "resp_len_mean": 80.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.31722,
        "reward_mean": -0.297483
      },
      {
        "step": 254,
        "frac_adv_clamped": 0.006386,
        "resp_len_mean": 58.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.423911,
        "reward_mean": -0.431753
      },
      {
        "step": 255,
        "frac_adv_clamped": 0.00499,
        "resp_len_mean": 75.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.384699,
        "reward_mean": -0.366748
      },
      {
        "step": 256,
        "frac_adv_clamped": 0.004764,
        "resp_len_mean": 78.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.322012,
        "reward_mean": -0.321013
      },
      {
        "step": 257,
        "frac_adv_clamped": 0.004996,
        "resp_len_mean": 75.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.343574,
        "reward_mean": -0.345603
      },
      {
        "step": 258,
        "frac_adv_clamped": 0.003563,
        "resp_len_mean": 105.25,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.296229,
        "reward_mean": -0.297915
      },
      {
        "step": 259,
        "frac_adv_clamped": 0.001551,
        "resp_len_mean": 60.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.269402,
        "reward_mean": -0.23669
      },
      {
        "step": 260,
        "frac_adv_clamped": 0.003106,
        "resp_len_mean": 70.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.294308,
        "reward_mean": -0.262984
      },
      {
        "step": 261,
        "frac_adv_clamped": 0.00591,
        "resp_len_mean": 79.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.324502,
        "reward_mean": -0.327816
      },
      {
        "step": 262,
        "frac_adv_clamped": 0.006341,
        "resp_len_mean": 64.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.332187,
        "reward_mean": -0.307051
      },
      {
        "step": 263,
        "frac_adv_clamped": 0.004338,
        "resp_len_mean": 86.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.328318,
        "reward_mean": -0.298626
      },
      {
        "step": 264,
        "frac_adv_clamped": 0.00188,
        "resp_len_mean": 83.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.231449,
        "reward_mean": -0.212791
      },
      {
        "step": 265,
        "frac_adv_clamped": 0.007181,
        "resp_len_mean": 69.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.367413,
        "reward_mean": -0.354841
      },
      {
        "step": 266,
        "frac_adv_clamped": 0.006355,
        "resp_len_mean": 93.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.286645,
        "reward_mean": -0.29879
      },
      {
        "step": 267,
        "frac_adv_clamped": 0.004678,
        "resp_len_mean": 80.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.346311,
        "reward_mean": -0.351551
      },
      {
        "step": 268,
        "frac_adv_clamped": 0.002288,
        "resp_len_mean": 109.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.290878,
        "reward_mean": -0.268493
      },
      {
        "step": 269,
        "frac_adv_clamped": 0.005829,
        "resp_len_mean": 85.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.252744,
        "reward_mean": -0.258317
      },
      {
        "step": 270,
        "frac_adv_clamped": 0.005327,
        "resp_len_mean": 82.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.33846,
        "reward_mean": -0.331035
      },
      {
        "step": 271,
        "frac_adv_clamped": 0.006821,
        "resp_len_mean": 82.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.32636,
        "reward_mean": -0.305648
      },
      {
        "step": 272,
        "frac_adv_clamped": 0.003548,
        "resp_len_mean": 61.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.287921,
        "reward_mean": -0.321522
      },
      {
        "step": 273,
        "frac_adv_clamped": 0.005443,
        "resp_len_mean": 86.125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.369659,
        "reward_mean": -0.351951
      },
      {
        "step": 274,
        "frac_adv_clamped": 0.003976,
        "resp_len_mean": 78.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.321261,
        "reward_mean": -0.310106
      },
      {
        "step": 275,
        "frac_adv_clamped": 0.005741,
        "resp_len_mean": 92.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.337252,
        "reward_mean": -0.345881
      },
      {
        "step": 276,
        "frac_adv_clamped": 0.006432,
        "resp_len_mean": 92.3125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.351729,
        "reward_mean": -0.371489
      },
      {
        "step": 277,
        "frac_adv_clamped": 0.004466,
        "resp_len_mean": 104.96875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.394939,
        "reward_mean": -0.377896
      },
      {
        "step": 278,
        "frac_adv_clamped": 0.001963,
        "resp_len_mean": 79.59375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.271524,
        "reward_mean": -0.258199
      },
      {
        "step": 279,
        "frac_adv_clamped": 0.002677,
        "resp_len_mean": 81.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.278787,
        "reward_mean": -0.254389
      },
      {
        "step": 280,
        "frac_adv_clamped": 0.003863,
        "resp_len_mean": 56.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.336715,
        "reward_mean": -0.311831
      },
      {
        "step": 281,
        "frac_adv_clamped": 0.003925,
        "resp_len_mean": 79.625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.310579,
        "reward_mean": -0.29088
      },
      {
        "step": 282,
        "frac_adv_clamped": 0.008475,
        "resp_len_mean": 59.0,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.364541,
        "reward_mean": -0.370294
      },
      {
        "step": 283,
        "frac_adv_clamped": 0.006235,
        "resp_len_mean": 90.21875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.365283,
        "reward_mean": -0.370881
      },
      {
        "step": 284,
        "frac_adv_clamped": 0.004586,
        "resp_len_mean": 95.40625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.351583,
        "reward_mean": -0.315307
      },
      {
        "step": 285,
        "frac_adv_clamped": 0.003156,
        "resp_len_mean": 79.21875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.304835,
        "reward_mean": -0.271905
      },
      {
        "step": 286,
        "frac_adv_clamped": 0.003932,
        "resp_len_mean": 95.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.259573,
        "reward_mean": -0.265198
      },
      {
        "step": 287,
        "frac_adv_clamped": 0.007068,
        "resp_len_mean": 92.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.400003,
        "reward_mean": -0.424696
      },
      {
        "step": 288,
        "frac_adv_clamped": 0.003962,
        "resp_len_mean": 78.875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.353139,
        "reward_mean": -0.386981
      },
      {
        "step": 289,
        "frac_adv_clamped": 0.004044,
        "resp_len_mean": 77.28125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.273771,
        "reward_mean": -0.258144
      },
      {
        "step": 290,
        "frac_adv_clamped": 0.004874,
        "resp_len_mean": 70.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.366651,
        "reward_mean": -0.36011
      },
      {
        "step": 291,
        "frac_adv_clamped": 0.006067,
        "resp_len_mean": 87.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.362672,
        "reward_mean": -0.351078
      },
      {
        "step": 292,
        "frac_adv_clamped": 0.003607,
        "resp_len_mean": 77.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.301223,
        "reward_mean": -0.293467
      },
      {
        "step": 293,
        "frac_adv_clamped": 0.006066,
        "resp_len_mean": 97.875,
        "frac_finished": 0.875,
        "kl_full_tok": 0.373441,
        "reward_mean": -0.370579
      },
      {
        "step": 294,
        "frac_adv_clamped": 0.004254,
        "resp_len_mean": 80.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.346442,
        "reward_mean": -0.359656
      },
      {
        "step": 295,
        "frac_adv_clamped": 0.004418,
        "resp_len_mean": 77.8125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.359484,
        "reward_mean": -0.346582
      },
      {
        "step": 296,
        "frac_adv_clamped": 0.004808,
        "resp_len_mean": 84.5,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.353625,
        "reward_mean": -0.352365
      },
      {
        "step": 297,
        "frac_adv_clamped": 0.004585,
        "resp_len_mean": 68.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.38186,
        "reward_mean": -0.395334
      },
      {
        "step": 298,
        "frac_adv_clamped": 0.000838,
        "resp_len_mean": 74.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.210338,
        "reward_mean": -0.183857
      },
      {
        "step": 299,
        "frac_adv_clamped": 0.002331,
        "resp_len_mean": 93.84375,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.319601,
        "reward_mean": -0.304302
      }
    ],
    "on_policy_seed2": [
      {
        "step": 0,
        "frac_adv_clamped": 0.018657,
        "resp_len_mean": 75.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.727445,
        "reward_mean": -0.745073
      },
      {
        "step": 1,
        "frac_adv_clamped": 0.029834,
        "resp_len_mean": 56.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.956968,
        "reward_mean": -1.045837
      },
      {
        "step": 2,
        "frac_adv_clamped": 0.027462,
        "resp_len_mean": 66.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.849029,
        "reward_mean": -0.89525
      },
      {
        "step": 3,
        "frac_adv_clamped": 0.022259,
        "resp_len_mean": 75.8125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.730564,
        "reward_mean": -0.770616
      },
      {
        "step": 4,
        "frac_adv_clamped": 0.023333,
        "resp_len_mean": 75.0,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.891445,
        "reward_mean": -0.99822
      },
      {
        "step": 5,
        "frac_adv_clamped": 0.010204,
        "resp_len_mean": 82.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.570348,
        "reward_mean": -0.644015
      },
      {
        "step": 6,
        "frac_adv_clamped": 0.005461,
        "resp_len_mean": 57.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.377587,
        "reward_mean": -0.427524
      },
      {
        "step": 7,
        "frac_adv_clamped": 0.010208,
        "resp_len_mean": 91.84375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.433446,
        "reward_mean": -0.503976
      },
      {
        "step": 8,
        "frac_adv_clamped": 0.00639,
        "resp_len_mean": 88.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.400278,
        "reward_mean": -0.39278
      },
      {
        "step": 9,
        "frac_adv_clamped": 0.005388,
        "resp_len_mean": 87.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.323542,
        "reward_mean": -0.285767
      },
      {
        "step": 10,
        "frac_adv_clamped": 0.01035,
        "resp_len_mean": 102.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.548818,
        "reward_mean": -0.560748
      },
      {
        "step": 11,
        "frac_adv_clamped": 0.007321,
        "resp_len_mean": 93.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.457963,
        "reward_mean": -0.454722
      },
      {
        "step": 12,
        "frac_adv_clamped": 0.005397,
        "resp_len_mean": 104.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.482351,
        "reward_mean": -0.452978
      },
      {
        "step": 13,
        "frac_adv_clamped": 0.010792,
        "resp_len_mean": 101.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.574886,
        "reward_mean": -0.579584
      },
      {
        "step": 14,
        "frac_adv_clamped": 0.007715,
        "resp_len_mean": 72.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.464236,
        "reward_mean": -0.39065
      },
      {
        "step": 15,
        "frac_adv_clamped": 0.002082,
        "resp_len_mean": 75.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.399971,
        "reward_mean": -0.363351
      },
      {
        "step": 16,
        "frac_adv_clamped": 0.005568,
        "resp_len_mean": 101.03125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.415898,
        "reward_mean": -0.407167
      },
      {
        "step": 17,
        "frac_adv_clamped": 0.008621,
        "resp_len_mean": 94.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.445476,
        "reward_mean": -0.418318
      },
      {
        "step": 18,
        "frac_adv_clamped": 0.00527,
        "resp_len_mean": 77.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.426695,
        "reward_mean": -0.387487
      },
      {
        "step": 19,
        "frac_adv_clamped": 0.006294,
        "resp_len_mean": 84.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.419689,
        "reward_mean": -0.414505
      },
      {
        "step": 20,
        "frac_adv_clamped": 0.009574,
        "resp_len_mean": 71.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.494928,
        "reward_mean": -0.513051
      },
      {
        "step": 21,
        "frac_adv_clamped": 0.011206,
        "resp_len_mean": 80.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.516183,
        "reward_mean": -0.531492
      },
      {
        "step": 22,
        "frac_adv_clamped": 0.005473,
        "resp_len_mean": 97.0625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.41122,
        "reward_mean": -0.357612
      },
      {
        "step": 23,
        "frac_adv_clamped": 0.005104,
        "resp_len_mean": 85.71875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.516988,
        "reward_mean": -0.486443
      },
      {
        "step": 24,
        "frac_adv_clamped": 0.008406,
        "resp_len_mean": 92.9375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.395972,
        "reward_mean": -0.406501
      },
      {
        "step": 25,
        "frac_adv_clamped": 0.003163,
        "resp_len_mean": 79.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.419342,
        "reward_mean": -0.399603
      },
      {
        "step": 26,
        "frac_adv_clamped": 0.008582,
        "resp_len_mean": 76.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.432943,
        "reward_mean": -0.477182
      },
      {
        "step": 27,
        "frac_adv_clamped": 0.0041,
        "resp_len_mean": 83.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.324578,
        "reward_mean": -0.293365
      },
      {
        "step": 28,
        "frac_adv_clamped": 0.004888,
        "resp_len_mean": 76.71875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.431711,
        "reward_mean": -0.403087
      },
      {
        "step": 29,
        "frac_adv_clamped": 0.005313,
        "resp_len_mean": 76.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.333525,
        "reward_mean": -0.301946
      },
      {
        "step": 30,
        "frac_adv_clamped": 0.003086,
        "resp_len_mean": 81.0,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.3411,
        "reward_mean": -0.314445
      },
      {
        "step": 31,
        "frac_adv_clamped": 0.003545,
        "resp_len_mean": 79.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.446356,
        "reward_mean": -0.44793
      },
      {
        "step": 32,
        "frac_adv_clamped": 0.006847,
        "resp_len_mean": 77.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.41797,
        "reward_mean": -0.397725
      },
      {
        "step": 33,
        "frac_adv_clamped": 0.005328,
        "resp_len_mean": 76.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.279171,
        "reward_mean": -0.254222
      },
      {
        "step": 34,
        "frac_adv_clamped": 0.004032,
        "resp_len_mean": 85.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.334121,
        "reward_mean": -0.274618
      },
      {
        "step": 35,
        "frac_adv_clamped": 0.00513,
        "resp_len_mean": 73.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.431483,
        "reward_mean": -0.41884
      },
      {
        "step": 36,
        "frac_adv_clamped": 0.007627,
        "resp_len_mean": 73.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.44783,
        "reward_mean": -0.439461
      },
      {
        "step": 37,
        "frac_adv_clamped": 0.005832,
        "resp_len_mean": 69.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.345045,
        "reward_mean": -0.34796
      },
      {
        "step": 38,
        "frac_adv_clamped": 0.003479,
        "resp_len_mean": 62.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.382393,
        "reward_mean": -0.40498
      },
      {
        "step": 39,
        "frac_adv_clamped": 0.006278,
        "resp_len_mean": 89.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.312936,
        "reward_mean": -0.292483
      },
      {
        "step": 40,
        "frac_adv_clamped": 0.002583,
        "resp_len_mean": 96.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.380279,
        "reward_mean": -0.361159
      },
      {
        "step": 41,
        "frac_adv_clamped": 0.006056,
        "resp_len_mean": 82.5625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.503013,
        "reward_mean": -0.468481
      },
      {
        "step": 42,
        "frac_adv_clamped": 0.007092,
        "resp_len_mean": 96.9375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.338071,
        "reward_mean": -0.3195
      },
      {
        "step": 43,
        "frac_adv_clamped": 0.003774,
        "resp_len_mean": 91.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.358763,
        "reward_mean": -0.319104
      },
      {
        "step": 44,
        "frac_adv_clamped": 0.003932,
        "resp_len_mean": 79.46875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.448911,
        "reward_mean": -0.396848
      },
      {
        "step": 45,
        "frac_adv_clamped": 0.00712,
        "resp_len_mean": 87.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.41457,
        "reward_mean": -0.371713
      },
      {
        "step": 46,
        "frac_adv_clamped": 0.005285,
        "resp_len_mean": 88.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.345509,
        "reward_mean": -0.311257
      },
      {
        "step": 47,
        "frac_adv_clamped": 0.006677,
        "resp_len_mean": 84.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.452774,
        "reward_mean": -0.447807
      },
      {
        "step": 48,
        "frac_adv_clamped": 0.002102,
        "resp_len_mean": 89.1875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.326699,
        "reward_mean": -0.320771
      },
      {
        "step": 49,
        "frac_adv_clamped": 0.005285,
        "resp_len_mean": 59.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.350451,
        "reward_mean": -0.318957
      },
      {
        "step": 50,
        "frac_adv_clamped": 0.005084,
        "resp_len_mean": 79.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.298988,
        "reward_mean": -0.286302
      },
      {
        "step": 51,
        "frac_adv_clamped": 0.005616,
        "resp_len_mean": 89.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.405697,
        "reward_mean": -0.410629
      },
      {
        "step": 52,
        "frac_adv_clamped": 0.005655,
        "resp_len_mean": 93.9375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.422781,
        "reward_mean": -0.432981
      },
      {
        "step": 53,
        "frac_adv_clamped": 0.005783,
        "resp_len_mean": 81.0625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.39705,
        "reward_mean": -0.385132
      },
      {
        "step": 54,
        "frac_adv_clamped": 0.003775,
        "resp_len_mean": 74.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.333414,
        "reward_mean": -0.325346
      },
      {
        "step": 55,
        "frac_adv_clamped": 0.004717,
        "resp_len_mean": 92.75,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.348671,
        "reward_mean": -0.324203
      },
      {
        "step": 56,
        "frac_adv_clamped": 0.005321,
        "resp_len_mean": 88.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.389568,
        "reward_mean": -0.363135
      },
      {
        "step": 57,
        "frac_adv_clamped": 0.001231,
        "resp_len_mean": 101.5625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.31721,
        "reward_mean": -0.30352
      },
      {
        "step": 58,
        "frac_adv_clamped": 0.005893,
        "resp_len_mean": 68.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.367058,
        "reward_mean": -0.338828
      },
      {
        "step": 59,
        "frac_adv_clamped": 0.006284,
        "resp_len_mean": 69.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.379526,
        "reward_mean": -0.373816
      },
      {
        "step": 60,
        "frac_adv_clamped": 0.00672,
        "resp_len_mean": 93.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.367326,
        "reward_mean": -0.365829
      },
      {
        "step": 61,
        "frac_adv_clamped": 0.005121,
        "resp_len_mean": 85.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.410992,
        "reward_mean": -0.395527
      },
      {
        "step": 62,
        "frac_adv_clamped": 0.005198,
        "resp_len_mean": 114.21875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.390444,
        "reward_mean": -0.366563
      },
      {
        "step": 63,
        "frac_adv_clamped": 0.005702,
        "resp_len_mean": 71.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.334088,
        "reward_mean": -0.320902
      },
      {
        "step": 64,
        "frac_adv_clamped": 0.003991,
        "resp_len_mean": 125.28125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.334698,
        "reward_mean": -0.334279
      },
      {
        "step": 65,
        "frac_adv_clamped": 0.005289,
        "resp_len_mean": 94.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.390605,
        "reward_mean": -0.355665
      },
      {
        "step": 66,
        "frac_adv_clamped": 0.004009,
        "resp_len_mean": 85.75,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.337468,
        "reward_mean": -0.341504
      },
      {
        "step": 67,
        "frac_adv_clamped": 0.006862,
        "resp_len_mean": 68.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.478495,
        "reward_mean": -0.468644
      },
      {
        "step": 68,
        "frac_adv_clamped": 0.002574,
        "resp_len_mean": 84.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.307706,
        "reward_mean": -0.259097
      },
      {
        "step": 69,
        "frac_adv_clamped": 0.006884,
        "resp_len_mean": 86.25,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.423906,
        "reward_mean": -0.411064
      },
      {
        "step": 70,
        "frac_adv_clamped": 0.002792,
        "resp_len_mean": 67.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.289498,
        "reward_mean": -0.267753
      },
      {
        "step": 71,
        "frac_adv_clamped": 0.00645,
        "resp_len_mean": 135.65625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.36532,
        "reward_mean": -0.367727
      },
      {
        "step": 72,
        "frac_adv_clamped": 0.008106,
        "resp_len_mean": 84.8125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.46517,
        "reward_mean": -0.43792
      },
      {
        "step": 73,
        "frac_adv_clamped": 0.004479,
        "resp_len_mean": 83.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.383567,
        "reward_mean": -0.352652
      },
      {
        "step": 74,
        "frac_adv_clamped": 0.002893,
        "resp_len_mean": 86.40625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.290264,
        "reward_mean": -0.223594
      },
      {
        "step": 75,
        "frac_adv_clamped": 0.006337,
        "resp_len_mean": 78.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.334866,
        "reward_mean": -0.327818
      },
      {
        "step": 76,
        "frac_adv_clamped": 0.004454,
        "resp_len_mean": 91.21875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.399363,
        "reward_mean": -0.408731
      },
      {
        "step": 77,
        "frac_adv_clamped": 0.002756,
        "resp_len_mean": 113.40625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.391128,
        "reward_mean": -0.396715
      },
      {
        "step": 78,
        "frac_adv_clamped": 0.006501,
        "resp_len_mean": 81.71875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.415577,
        "reward_mean": -0.399893
      },
      {
        "step": 79,
        "frac_adv_clamped": 0.005685,
        "resp_len_mean": 104.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.375647,
        "reward_mean": -0.389528
      },
      {
        "step": 80,
        "frac_adv_clamped": 0.004107,
        "resp_len_mean": 121.75,
        "frac_finished": 0.84375,
        "kl_full_tok": 0.393415,
        "reward_mean": -0.381133
      },
      {
        "step": 81,
        "frac_adv_clamped": 0.001819,
        "resp_len_mean": 85.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.28721,
        "reward_mean": -0.29525
      },
      {
        "step": 82,
        "frac_adv_clamped": 0.004671,
        "resp_len_mean": 100.34375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.297506,
        "reward_mean": -0.273984
      },
      {
        "step": 83,
        "frac_adv_clamped": 0.005508,
        "resp_len_mean": 73.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.3267,
        "reward_mean": -0.340265
      },
      {
        "step": 84,
        "frac_adv_clamped": 0.008124,
        "resp_len_mean": 84.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.424415,
        "reward_mean": -0.423295
      },
      {
        "step": 85,
        "frac_adv_clamped": 0.005385,
        "resp_len_mean": 75.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.335772,
        "reward_mean": -0.311319
      },
      {
        "step": 86,
        "frac_adv_clamped": 0.002488,
        "resp_len_mean": 62.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.21119,
        "reward_mean": -0.203053
      },
      {
        "step": 87,
        "frac_adv_clamped": 0.007254,
        "resp_len_mean": 56.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.463941,
        "reward_mean": -0.458906
      },
      {
        "step": 88,
        "frac_adv_clamped": 0.00315,
        "resp_len_mean": 69.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.280216,
        "reward_mean": -0.260746
      },
      {
        "step": 89,
        "frac_adv_clamped": 0.00346,
        "resp_len_mean": 54.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.387293,
        "reward_mean": -0.362214
      },
      {
        "step": 90,
        "frac_adv_clamped": 0.003442,
        "resp_len_mean": 72.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.42287,
        "reward_mean": -0.409136
      },
      {
        "step": 91,
        "frac_adv_clamped": 0.005757,
        "resp_len_mean": 86.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.377109,
        "reward_mean": -0.328041
      },
      {
        "step": 92,
        "frac_adv_clamped": 0.00485,
        "resp_len_mean": 70.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.329816,
        "reward_mean": -0.323317
      },
      {
        "step": 93,
        "frac_adv_clamped": 0.002523,
        "resp_len_mean": 74.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.291648,
        "reward_mean": -0.275834
      },
      {
        "step": 94,
        "frac_adv_clamped": 0.007226,
        "resp_len_mean": 77.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.371311,
        "reward_mean": -0.352885
      },
      {
        "step": 95,
        "frac_adv_clamped": 0.005584,
        "resp_len_mean": 61.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.374995,
        "reward_mean": -0.357022
      },
      {
        "step": 96,
        "frac_adv_clamped": 0.004,
        "resp_len_mean": 54.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.344651,
        "reward_mean": -0.305621
      },
      {
        "step": 97,
        "frac_adv_clamped": 0.005437,
        "resp_len_mean": 63.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.381116,
        "reward_mean": -0.377793
      },
      {
        "step": 98,
        "frac_adv_clamped": 0.002385,
        "resp_len_mean": 65.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.292642,
        "reward_mean": -0.286606
      },
      {
        "step": 99,
        "frac_adv_clamped": 0.008314,
        "resp_len_mean": 78.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.413906,
        "reward_mean": -0.40377
      },
      {
        "step": 100,
        "frac_adv_clamped": 0.008452,
        "resp_len_mean": 81.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.443959,
        "reward_mean": -0.432295
      },
      {
        "step": 101,
        "frac_adv_clamped": 0.003289,
        "resp_len_mean": 85.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.258416,
        "reward_mean": -0.254658
      },
      {
        "step": 102,
        "frac_adv_clamped": 0.000993,
        "resp_len_mean": 62.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.245186,
        "reward_mean": -0.199879
      },
      {
        "step": 103,
        "frac_adv_clamped": 0.000955,
        "resp_len_mean": 65.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.268395,
        "reward_mean": -0.220786
      },
      {
        "step": 104,
        "frac_adv_clamped": 0.00445,
        "resp_len_mean": 77.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.386398,
        "reward_mean": -0.32362
      },
      {
        "step": 105,
        "frac_adv_clamped": 0.004403,
        "resp_len_mean": 63.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.327582,
        "reward_mean": -0.355064
      },
      {
        "step": 106,
        "frac_adv_clamped": 0.006664,
        "resp_len_mean": 75.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.385676,
        "reward_mean": -0.388104
      },
      {
        "step": 107,
        "frac_adv_clamped": 0.005613,
        "resp_len_mean": 72.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.370176,
        "reward_mean": -0.378116
      },
      {
        "step": 108,
        "frac_adv_clamped": 0.004795,
        "resp_len_mean": 71.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.350403,
        "reward_mean": -0.343602
      },
      {
        "step": 109,
        "frac_adv_clamped": 0.004148,
        "resp_len_mean": 82.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.338899,
        "reward_mean": -0.305647
      },
      {
        "step": 110,
        "frac_adv_clamped": 0.003758,
        "resp_len_mean": 83.15625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.271918,
        "reward_mean": -0.250243
      },
      {
        "step": 111,
        "frac_adv_clamped": 0.007166,
        "resp_len_mean": 104.65625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.398885,
        "reward_mean": -0.38681
      },
      {
        "step": 112,
        "frac_adv_clamped": 0.006197,
        "resp_len_mean": 80.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.412748,
        "reward_mean": -0.411286
      },
      {
        "step": 113,
        "frac_adv_clamped": 0.004102,
        "resp_len_mean": 99.03125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.36971,
        "reward_mean": -0.386803
      },
      {
        "step": 114,
        "frac_adv_clamped": 0.005179,
        "resp_len_mean": 78.4375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.338424,
        "reward_mean": -0.325401
      },
      {
        "step": 115,
        "frac_adv_clamped": 0.000904,
        "resp_len_mean": 69.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.253034,
        "reward_mean": -0.239378
      },
      {
        "step": 116,
        "frac_adv_clamped": 0.004965,
        "resp_len_mean": 62.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.319899,
        "reward_mean": -0.31058
      },
      {
        "step": 117,
        "frac_adv_clamped": 0.002915,
        "resp_len_mean": 96.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.407142,
        "reward_mean": -0.376499
      },
      {
        "step": 118,
        "frac_adv_clamped": 0.002217,
        "resp_len_mean": 84.5625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.273652,
        "reward_mean": -0.266289
      },
      {
        "step": 119,
        "frac_adv_clamped": 0.003717,
        "resp_len_mean": 100.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.325178,
        "reward_mean": -0.313367
      },
      {
        "step": 120,
        "frac_adv_clamped": 0.00381,
        "resp_len_mean": 106.625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.358202,
        "reward_mean": -0.328609
      },
      {
        "step": 121,
        "frac_adv_clamped": 0.00464,
        "resp_len_mean": 114.5,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.308959,
        "reward_mean": -0.317637
      },
      {
        "step": 122,
        "frac_adv_clamped": 0.003908,
        "resp_len_mean": 95.96875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.308216,
        "reward_mean": -0.317639
      },
      {
        "step": 123,
        "frac_adv_clamped": 0.005099,
        "resp_len_mean": 104.1875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.334972,
        "reward_mean": -0.340299
      },
      {
        "step": 124,
        "frac_adv_clamped": 0.002625,
        "resp_len_mean": 83.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.331445,
        "reward_mean": -0.292424
      },
      {
        "step": 125,
        "frac_adv_clamped": 0.005,
        "resp_len_mean": 112.5,
        "frac_finished": 0.84375,
        "kl_full_tok": 0.338128,
        "reward_mean": -0.337827
      },
      {
        "step": 126,
        "frac_adv_clamped": 0.006525,
        "resp_len_mean": 76.625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.334583,
        "reward_mean": -0.317017
      },
      {
        "step": 127,
        "frac_adv_clamped": 0.004734,
        "resp_len_mean": 85.8125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.25889,
        "reward_mean": -0.263478
      },
      {
        "step": 128,
        "frac_adv_clamped": 0.004665,
        "resp_len_mean": 87.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.310338,
        "reward_mean": -0.312559
      },
      {
        "step": 129,
        "frac_adv_clamped": 0.005054,
        "resp_len_mean": 105.125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.373046,
        "reward_mean": -0.336514
      },
      {
        "step": 130,
        "frac_adv_clamped": 0.001678,
        "resp_len_mean": 74.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.237266,
        "reward_mean": -0.208324
      },
      {
        "step": 131,
        "frac_adv_clamped": 0.007834,
        "resp_len_mean": 55.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.369079,
        "reward_mean": -0.372912
      },
      {
        "step": 132,
        "frac_adv_clamped": 0.003198,
        "resp_len_mean": 87.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.282954,
        "reward_mean": -0.267781
      },
      {
        "step": 133,
        "frac_adv_clamped": 0.003908,
        "resp_len_mean": 79.96875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.299489,
        "reward_mean": -0.253775
      },
      {
        "step": 134,
        "frac_adv_clamped": 0.005108,
        "resp_len_mean": 79.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.292728,
        "reward_mean": -0.263035
      },
      {
        "step": 135,
        "frac_adv_clamped": 0.001461,
        "resp_len_mean": 85.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.288629,
        "reward_mean": -0.275844
      },
      {
        "step": 136,
        "frac_adv_clamped": 0.004731,
        "resp_len_mean": 92.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.302939,
        "reward_mean": -0.305072
      },
      {
        "step": 137,
        "frac_adv_clamped": 0.005037,
        "resp_len_mean": 93.0625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.355457,
        "reward_mean": -0.318745
      },
      {
        "step": 138,
        "frac_adv_clamped": 0.006225,
        "resp_len_mean": 70.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.286986,
        "reward_mean": -0.278479
      },
      {
        "step": 139,
        "frac_adv_clamped": 0.004971,
        "resp_len_mean": 75.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.309968,
        "reward_mean": -0.279165
      },
      {
        "step": 140,
        "frac_adv_clamped": 0.002755,
        "resp_len_mean": 68.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.27274,
        "reward_mean": -0.237406
      },
      {
        "step": 141,
        "frac_adv_clamped": 0.007316,
        "resp_len_mean": 93.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.402076,
        "reward_mean": -0.385804
      },
      {
        "step": 142,
        "frac_adv_clamped": 0.006997,
        "resp_len_mean": 53.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.323382,
        "reward_mean": -0.312958
      },
      {
        "step": 143,
        "frac_adv_clamped": 0.005455,
        "resp_len_mean": 74.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.322351,
        "reward_mean": -0.323494
      },
      {
        "step": 144,
        "frac_adv_clamped": 0.008625,
        "resp_len_mean": 65.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.417693,
        "reward_mean": -0.400972
      },
      {
        "step": 145,
        "frac_adv_clamped": 0.003942,
        "resp_len_mean": 79.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.381647,
        "reward_mean": -0.365714
      },
      {
        "step": 146,
        "frac_adv_clamped": 0.004443,
        "resp_len_mean": 77.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.310661,
        "reward_mean": -0.317136
      },
      {
        "step": 147,
        "frac_adv_clamped": 0.003908,
        "resp_len_mean": 79.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.227179,
        "reward_mean": -0.234683
      },
      {
        "step": 148,
        "frac_adv_clamped": 0.004409,
        "resp_len_mean": 70.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.256281,
        "reward_mean": -0.223632
      },
      {
        "step": 149,
        "frac_adv_clamped": 0.00746,
        "resp_len_mean": 79.59375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.454972,
        "reward_mean": -0.389147
      },
      {
        "step": 150,
        "frac_adv_clamped": 0.006209,
        "resp_len_mean": 85.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.365519,
        "reward_mean": -0.374011
      },
      {
        "step": 151,
        "frac_adv_clamped": 0.00451,
        "resp_len_mean": 83.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.320924,
        "reward_mean": -0.294813
      },
      {
        "step": 152,
        "frac_adv_clamped": 0.005609,
        "resp_len_mean": 100.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.37222,
        "reward_mean": -0.386315
      },
      {
        "step": 153,
        "frac_adv_clamped": 0.005963,
        "resp_len_mean": 83.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.381357,
        "reward_mean": -0.359681
      },
      {
        "step": 154,
        "frac_adv_clamped": 0.004938,
        "resp_len_mean": 88.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.359997,
        "reward_mean": -0.360337
      },
      {
        "step": 155,
        "frac_adv_clamped": 0.005228,
        "resp_len_mean": 83.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.311612,
        "reward_mean": -0.298407
      },
      {
        "step": 156,
        "frac_adv_clamped": 0.006294,
        "resp_len_mean": 84.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.39632,
        "reward_mean": -0.391597
      },
      {
        "step": 157,
        "frac_adv_clamped": 0.006439,
        "resp_len_mean": 63.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.31337,
        "reward_mean": -0.299811
      },
      {
        "step": 158,
        "frac_adv_clamped": 0.004773,
        "resp_len_mean": 78.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.32659,
        "reward_mean": -0.327803
      },
      {
        "step": 159,
        "frac_adv_clamped": 0.005988,
        "resp_len_mean": 67.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.423497,
        "reward_mean": -0.414104
      },
      {
        "step": 160,
        "frac_adv_clamped": 0.00427,
        "resp_len_mean": 80.5,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.301379,
        "reward_mean": -0.286708
      },
      {
        "step": 161,
        "frac_adv_clamped": 0.003126,
        "resp_len_mean": 79.96875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.35719,
        "reward_mean": -0.312296
      },
      {
        "step": 162,
        "frac_adv_clamped": 0.004858,
        "resp_len_mean": 128.65625,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.350469,
        "reward_mean": -0.337522
      },
      {
        "step": 163,
        "frac_adv_clamped": 0.005267,
        "resp_len_mean": 89.0,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.34039,
        "reward_mean": -0.326812
      },
      {
        "step": 164,
        "frac_adv_clamped": 0.010491,
        "resp_len_mean": 83.40625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.437232,
        "reward_mean": -0.451966
      },
      {
        "step": 165,
        "frac_adv_clamped": 0.003115,
        "resp_len_mean": 110.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.257585,
        "reward_mean": -0.254698
      },
      {
        "step": 166,
        "frac_adv_clamped": 0.00153,
        "resp_len_mean": 81.71875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.319279,
        "reward_mean": -0.309774
      },
      {
        "step": 167,
        "frac_adv_clamped": 0.001774,
        "resp_len_mean": 70.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.312302,
        "reward_mean": -0.266131
      },
      {
        "step": 168,
        "frac_adv_clamped": 0.004532,
        "resp_len_mean": 96.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.338187,
        "reward_mean": -0.331322
      },
      {
        "step": 169,
        "frac_adv_clamped": 0.002247,
        "resp_len_mean": 83.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.350798,
        "reward_mean": -0.333654
      },
      {
        "step": 170,
        "frac_adv_clamped": 0.003055,
        "resp_len_mean": 71.59375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.336207,
        "reward_mean": -0.337286
      },
      {
        "step": 171,
        "frac_adv_clamped": 0.00717,
        "resp_len_mean": 91.53125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.455601,
        "reward_mean": -0.428658
      },
      {
        "step": 172,
        "frac_adv_clamped": 0.004386,
        "resp_len_mean": 78.375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.361432,
        "reward_mean": -0.355493
      },
      {
        "step": 173,
        "frac_adv_clamped": 0.004226,
        "resp_len_mean": 81.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.233512,
        "reward_mean": -0.225165
      },
      {
        "step": 174,
        "frac_adv_clamped": 0.005385,
        "resp_len_mean": 75.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.273585,
        "reward_mean": -0.274509
      },
      {
        "step": 175,
        "frac_adv_clamped": 0.003011,
        "resp_len_mean": 93.40625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.314252,
        "reward_mean": -0.288078
      },
      {
        "step": 176,
        "frac_adv_clamped": 0.005482,
        "resp_len_mean": 96.90625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.356672,
        "reward_mean": -0.341784
      },
      {
        "step": 177,
        "frac_adv_clamped": 0.004625,
        "resp_len_mean": 67.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.34737,
        "reward_mean": -0.334826
      },
      {
        "step": 178,
        "frac_adv_clamped": 0.006926,
        "resp_len_mean": 94.75,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.324766,
        "reward_mean": -0.303152
      },
      {
        "step": 179,
        "frac_adv_clamped": 0.003453,
        "resp_len_mean": 90.5,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.361881,
        "reward_mean": -0.35768
      },
      {
        "step": 180,
        "frac_adv_clamped": 0.006083,
        "resp_len_mean": 77.0625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.392343,
        "reward_mean": -0.358539
      },
      {
        "step": 181,
        "frac_adv_clamped": 0.002167,
        "resp_len_mean": 72.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.269632,
        "reward_mean": -0.261872
      },
      {
        "step": 182,
        "frac_adv_clamped": 0.003925,
        "resp_len_mean": 63.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.257502,
        "reward_mean": -0.254857
      },
      {
        "step": 183,
        "frac_adv_clamped": 0.00797,
        "resp_len_mean": 74.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.341124,
        "reward_mean": -0.336218
      },
      {
        "step": 184,
        "frac_adv_clamped": 0.004891,
        "resp_len_mean": 57.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.262441,
        "reward_mean": -0.2379
      },
      {
        "step": 185,
        "frac_adv_clamped": 0.005629,
        "resp_len_mean": 83.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.389489,
        "reward_mean": -0.387794
      },
      {
        "step": 186,
        "frac_adv_clamped": 0.003687,
        "resp_len_mean": 67.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.371097,
        "reward_mean": -0.328289
      },
      {
        "step": 187,
        "frac_adv_clamped": 0.000664,
        "resp_len_mean": 94.15625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.230332,
        "reward_mean": -0.225753
      },
      {
        "step": 188,
        "frac_adv_clamped": 0.004489,
        "resp_len_mean": 90.5,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.379303,
        "reward_mean": -0.353046
      },
      {
        "step": 189,
        "frac_adv_clamped": 0.003504,
        "resp_len_mean": 71.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.321735,
        "reward_mean": -0.282417
      },
      {
        "step": 190,
        "frac_adv_clamped": 0.004806,
        "resp_len_mean": 71.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.366757,
        "reward_mean": -0.331868
      },
      {
        "step": 191,
        "frac_adv_clamped": 0.005022,
        "resp_len_mean": 99.5625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.335358,
        "reward_mean": -0.354878
      },
      {
        "step": 192,
        "frac_adv_clamped": 0.005961,
        "resp_len_mean": 68.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.306371,
        "reward_mean": -0.291047
      },
      {
        "step": 193,
        "frac_adv_clamped": 0.006874,
        "resp_len_mean": 77.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.411697,
        "reward_mean": -0.39636
      },
      {
        "step": 194,
        "frac_adv_clamped": 0.00597,
        "resp_len_mean": 83.75,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.438821,
        "reward_mean": -0.395391
      },
      {
        "step": 195,
        "frac_adv_clamped": 0.001748,
        "resp_len_mean": 53.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.281711,
        "reward_mean": -0.219737
      },
      {
        "step": 196,
        "frac_adv_clamped": 0.004744,
        "resp_len_mean": 59.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.311204,
        "reward_mean": -0.274714
      },
      {
        "step": 197,
        "frac_adv_clamped": 0.006312,
        "resp_len_mean": 69.3125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.401422,
        "reward_mean": -0.358508
      },
      {
        "step": 198,
        "frac_adv_clamped": 0.001791,
        "resp_len_mean": 87.21875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.207033,
        "reward_mean": -0.195668
      },
      {
        "step": 199,
        "frac_adv_clamped": 0.004585,
        "resp_len_mean": 68.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.287727,
        "reward_mean": -0.262833
      },
      {
        "step": 200,
        "frac_adv_clamped": 0.004875,
        "resp_len_mean": 89.75,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.310561,
        "reward_mean": -0.305465
      },
      {
        "step": 201,
        "frac_adv_clamped": 0.007361,
        "resp_len_mean": 80.65625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.368569,
        "reward_mean": -0.358605
      },
      {
        "step": 202,
        "frac_adv_clamped": 0.006367,
        "resp_len_mean": 83.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.302001,
        "reward_mean": -0.304896
      },
      {
        "step": 203,
        "frac_adv_clamped": 0.003366,
        "resp_len_mean": 83.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.249637,
        "reward_mean": -0.243328
      },
      {
        "step": 204,
        "frac_adv_clamped": 0.005329,
        "resp_len_mean": 82.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.37051,
        "reward_mean": -0.333471
      },
      {
        "step": 205,
        "frac_adv_clamped": 0.003687,
        "resp_len_mean": 76.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.247754,
        "reward_mean": -0.236622
      },
      {
        "step": 206,
        "frac_adv_clamped": 0.007385,
        "resp_len_mean": 71.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.299228,
        "reward_mean": -0.28521
      },
      {
        "step": 207,
        "frac_adv_clamped": 0.003457,
        "resp_len_mean": 117.5,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.343403,
        "reward_mean": -0.314176
      },
      {
        "step": 208,
        "frac_adv_clamped": 0.00659,
        "resp_len_mean": 75.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.331229,
        "reward_mean": -0.308479
      },
      {
        "step": 209,
        "frac_adv_clamped": 0.00558,
        "resp_len_mean": 78.40625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.368069,
        "reward_mean": -0.341186
      },
      {
        "step": 210,
        "frac_adv_clamped": 0.004762,
        "resp_len_mean": 78.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.311114,
        "reward_mean": -0.32649
      },
      {
        "step": 211,
        "frac_adv_clamped": 0.006673,
        "resp_len_mean": 70.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.393138,
        "reward_mean": -0.375627
      },
      {
        "step": 212,
        "frac_adv_clamped": 0.007579,
        "resp_len_mean": 82.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.380689,
        "reward_mean": -0.376748
      },
      {
        "step": 213,
        "frac_adv_clamped": 0.005974,
        "resp_len_mean": 78.46875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.324387,
        "reward_mean": -0.287998
      },
      {
        "step": 214,
        "frac_adv_clamped": 0.004429,
        "resp_len_mean": 119.9375,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.356111,
        "reward_mean": -0.359763
      },
      {
        "step": 215,
        "frac_adv_clamped": 0.004243,
        "resp_len_mean": 66.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.260868,
        "reward_mean": -0.257062
      },
      {
        "step": 216,
        "frac_adv_clamped": 0.002325,
        "resp_len_mean": 67.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.329333,
        "reward_mean": -0.29887
      },
      {
        "step": 217,
        "frac_adv_clamped": 0.004241,
        "resp_len_mean": 73.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.348022,
        "reward_mean": -0.328285
      },
      {
        "step": 218,
        "frac_adv_clamped": 0.003225,
        "resp_len_mean": 77.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.38101,
        "reward_mean": -0.35974
      },
      {
        "step": 219,
        "frac_adv_clamped": 0.003823,
        "resp_len_mean": 81.75,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.281826,
        "reward_mean": -0.278546
      },
      {
        "step": 220,
        "frac_adv_clamped": 0.002929,
        "resp_len_mean": 85.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.264107,
        "reward_mean": -0.254923
      },
      {
        "step": 221,
        "frac_adv_clamped": 0.001673,
        "resp_len_mean": 93.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.305672,
        "reward_mean": -0.302537
      },
      {
        "step": 222,
        "frac_adv_clamped": 0.003294,
        "resp_len_mean": 85.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.2703,
        "reward_mean": -0.269427
      },
      {
        "step": 223,
        "frac_adv_clamped": 0.004989,
        "resp_len_mean": 68.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.304855,
        "reward_mean": -0.25073
      },
      {
        "step": 224,
        "frac_adv_clamped": 0.00429,
        "resp_len_mean": 80.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.419744,
        "reward_mean": -0.407546
      },
      {
        "step": 225,
        "frac_adv_clamped": 0.001641,
        "resp_len_mean": 76.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.254455,
        "reward_mean": -0.228696
      },
      {
        "step": 226,
        "frac_adv_clamped": 0.006119,
        "resp_len_mean": 97.03125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.386392,
        "reward_mean": -0.379064
      },
      {
        "step": 227,
        "frac_adv_clamped": 0.006481,
        "resp_len_mean": 96.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.410898,
        "reward_mean": -0.434223
      },
      {
        "step": 228,
        "frac_adv_clamped": 0.002611,
        "resp_len_mean": 71.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.251272,
        "reward_mean": -0.219322
      },
      {
        "step": 229,
        "frac_adv_clamped": 0.006283,
        "resp_len_mean": 64.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.336617,
        "reward_mean": -0.327271
      },
      {
        "step": 230,
        "frac_adv_clamped": 0.008806,
        "resp_len_mean": 63.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.303661,
        "reward_mean": -0.272718
      },
      {
        "step": 231,
        "frac_adv_clamped": 0.0085,
        "resp_len_mean": 73.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.347779,
        "reward_mean": -0.326933
      },
      {
        "step": 232,
        "frac_adv_clamped": 0.004123,
        "resp_len_mean": 68.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.278623,
        "reward_mean": -0.239166
      },
      {
        "step": 233,
        "frac_adv_clamped": 0.006562,
        "resp_len_mean": 85.71875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.43493,
        "reward_mean": -0.403488
      },
      {
        "step": 234,
        "frac_adv_clamped": 0.00356,
        "resp_len_mean": 79.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.236036,
        "reward_mean": -0.236592
      },
      {
        "step": 235,
        "frac_adv_clamped": 0.003538,
        "resp_len_mean": 70.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.351044,
        "reward_mean": -0.345834
      },
      {
        "step": 236,
        "frac_adv_clamped": 0.001664,
        "resp_len_mean": 93.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.255603,
        "reward_mean": -0.250648
      },
      {
        "step": 237,
        "frac_adv_clamped": 0.003638,
        "resp_len_mean": 85.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.318189,
        "reward_mean": -0.281475
      },
      {
        "step": 238,
        "frac_adv_clamped": 0.00546,
        "resp_len_mean": 74.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.397106,
        "reward_mean": -0.384276
      },
      {
        "step": 239,
        "frac_adv_clamped": 0.003207,
        "resp_len_mean": 87.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.37027,
        "reward_mean": -0.345782
      },
      {
        "step": 240,
        "frac_adv_clamped": 0.004093,
        "resp_len_mean": 76.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.296801,
        "reward_mean": -0.269938
      },
      {
        "step": 241,
        "frac_adv_clamped": 0.003605,
        "resp_len_mean": 130.03125,
        "frac_finished": 0.875,
        "kl_full_tok": 0.283306,
        "reward_mean": -0.260306
      },
      {
        "step": 242,
        "frac_adv_clamped": 0.002537,
        "resp_len_mean": 61.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.239863,
        "reward_mean": -0.242158
      },
      {
        "step": 243,
        "frac_adv_clamped": 0.00699,
        "resp_len_mean": 84.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.313473,
        "reward_mean": -0.285492
      },
      {
        "step": 244,
        "frac_adv_clamped": 0.00663,
        "resp_len_mean": 84.84375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.368549,
        "reward_mean": -0.351814
      },
      {
        "step": 245,
        "frac_adv_clamped": 0.004717,
        "resp_len_mean": 72.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.300491,
        "reward_mean": -0.272574
      },
      {
        "step": 246,
        "frac_adv_clamped": 0.003977,
        "resp_len_mean": 94.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.304467,
        "reward_mean": -0.279367
      },
      {
        "step": 247,
        "frac_adv_clamped": 0.002786,
        "resp_len_mean": 67.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.257062,
        "reward_mean": -0.242706
      },
      {
        "step": 248,
        "frac_adv_clamped": 0.003079,
        "resp_len_mean": 81.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.224622,
        "reward_mean": -0.213764
      },
      {
        "step": 249,
        "frac_adv_clamped": 0.005874,
        "resp_len_mean": 106.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.375991,
        "reward_mean": -0.35912
      },
      {
        "step": 250,
        "frac_adv_clamped": 0.004529,
        "resp_len_mean": 96.59375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.343104,
        "reward_mean": -0.359334
      },
      {
        "step": 251,
        "frac_adv_clamped": 0.004837,
        "resp_len_mean": 77.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.356889,
        "reward_mean": -0.328745
      },
      {
        "step": 252,
        "frac_adv_clamped": 0.002434,
        "resp_len_mean": 77.03125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.310429,
        "reward_mean": -0.31386
      },
      {
        "step": 253,
        "frac_adv_clamped": 0.005156,
        "resp_len_mean": 90.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.362596,
        "reward_mean": -0.363228
      },
      {
        "step": 254,
        "frac_adv_clamped": 0.007768,
        "resp_len_mean": 104.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.382459,
        "reward_mean": -0.361075
      },
      {
        "step": 255,
        "frac_adv_clamped": 0.005183,
        "resp_len_mean": 84.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.418838,
        "reward_mean": -0.370202
      },
      {
        "step": 256,
        "frac_adv_clamped": 0.00444,
        "resp_len_mean": 84.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.313789,
        "reward_mean": -0.295691
      },
      {
        "step": 257,
        "frac_adv_clamped": 0.004359,
        "resp_len_mean": 71.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.326116,
        "reward_mean": -0.3224
      },
      {
        "step": 258,
        "frac_adv_clamped": 0.003368,
        "resp_len_mean": 102.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.332644,
        "reward_mean": -0.301951
      },
      {
        "step": 259,
        "frac_adv_clamped": 0.005013,
        "resp_len_mean": 81.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.35139,
        "reward_mean": -0.345462
      },
      {
        "step": 260,
        "frac_adv_clamped": 0.004968,
        "resp_len_mean": 62.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.308134,
        "reward_mean": -0.307252
      },
      {
        "step": 261,
        "frac_adv_clamped": 0.00476,
        "resp_len_mean": 91.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.396207,
        "reward_mean": -0.420545
      },
      {
        "step": 262,
        "frac_adv_clamped": 0.005964,
        "resp_len_mean": 99.5625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.404565,
        "reward_mean": -0.410552
      },
      {
        "step": 263,
        "frac_adv_clamped": 0.003342,
        "resp_len_mean": 84.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.262339,
        "reward_mean": -0.227224
      },
      {
        "step": 264,
        "frac_adv_clamped": 0.003276,
        "resp_len_mean": 124.0,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.361754,
        "reward_mean": -0.339159
      },
      {
        "step": 265,
        "frac_adv_clamped": 0.002762,
        "resp_len_mean": 101.8125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.270032,
        "reward_mean": -0.261442
      },
      {
        "step": 266,
        "frac_adv_clamped": 0.006286,
        "resp_len_mean": 64.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.411751,
        "reward_mean": -0.392391
      },
      {
        "step": 267,
        "frac_adv_clamped": 0.007047,
        "resp_len_mean": 102.0,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.351606,
        "reward_mean": -0.329784
      },
      {
        "step": 268,
        "frac_adv_clamped": 0.003147,
        "resp_len_mean": 89.375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.277937,
        "reward_mean": -0.258924
      },
      {
        "step": 269,
        "frac_adv_clamped": 0.008601,
        "resp_len_mean": 98.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.470016,
        "reward_mean": -0.488349
      },
      {
        "step": 270,
        "frac_adv_clamped": 0.006932,
        "resp_len_mean": 85.65625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.425166,
        "reward_mean": -0.418163
      },
      {
        "step": 271,
        "frac_adv_clamped": 0.004182,
        "resp_len_mean": 112.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.362491,
        "reward_mean": -0.345896
      },
      {
        "step": 272,
        "frac_adv_clamped": 0.004822,
        "resp_len_mean": 51.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.333485,
        "reward_mean": -0.328368
      },
      {
        "step": 273,
        "frac_adv_clamped": 0.004425,
        "resp_len_mean": 98.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.314006,
        "reward_mean": -0.290055
      },
      {
        "step": 274,
        "frac_adv_clamped": 0.003804,
        "resp_len_mean": 65.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.238443,
        "reward_mean": -0.220251
      },
      {
        "step": 275,
        "frac_adv_clamped": 0.006316,
        "resp_len_mean": 94.0,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.365337,
        "reward_mean": -0.366143
      },
      {
        "step": 276,
        "frac_adv_clamped": 0.003988,
        "resp_len_mean": 101.875,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.326815,
        "reward_mean": -0.317498
      },
      {
        "step": 277,
        "frac_adv_clamped": 0.003686,
        "resp_len_mean": 84.78125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.321914,
        "reward_mean": -0.302563
      },
      {
        "step": 278,
        "frac_adv_clamped": 0.004701,
        "resp_len_mean": 119.65625,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.362872,
        "reward_mean": -0.349864
      },
      {
        "step": 279,
        "frac_adv_clamped": 0.003328,
        "resp_len_mean": 56.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.240012,
        "reward_mean": -0.212328
      },
      {
        "step": 280,
        "frac_adv_clamped": 0.008547,
        "resp_len_mean": 80.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.465277,
        "reward_mean": -0.434055
      },
      {
        "step": 281,
        "frac_adv_clamped": 0.00212,
        "resp_len_mean": 88.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.241774,
        "reward_mean": -0.201747
      },
      {
        "step": 282,
        "frac_adv_clamped": 0.007071,
        "resp_len_mean": 83.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.312998,
        "reward_mean": -0.303634
      },
      {
        "step": 283,
        "frac_adv_clamped": 0.002314,
        "resp_len_mean": 94.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.230902,
        "reward_mean": -0.202726
      },
      {
        "step": 284,
        "frac_adv_clamped": 0.004766,
        "resp_len_mean": 72.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.398581,
        "reward_mean": -0.385665
      },
      {
        "step": 285,
        "frac_adv_clamped": 0.001249,
        "resp_len_mean": 75.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.216421,
        "reward_mean": -0.224896
      },
      {
        "step": 286,
        "frac_adv_clamped": 0.003898,
        "resp_len_mean": 56.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.299096,
        "reward_mean": -0.27855
      },
      {
        "step": 287,
        "frac_adv_clamped": 0.003967,
        "resp_len_mean": 86.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.257692,
        "reward_mean": -0.239864
      },
      {
        "step": 288,
        "frac_adv_clamped": 0.007216,
        "resp_len_mean": 82.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.340582,
        "reward_mean": -0.313201
      },
      {
        "step": 289,
        "frac_adv_clamped": 0.002984,
        "resp_len_mean": 73.3125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.327335,
        "reward_mean": -0.3097
      },
      {
        "step": 290,
        "frac_adv_clamped": 0.004435,
        "resp_len_mean": 77.5,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.382097,
        "reward_mean": -0.357883
      },
      {
        "step": 291,
        "frac_adv_clamped": 0.003046,
        "resp_len_mean": 112.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.283781,
        "reward_mean": -0.262541
      },
      {
        "step": 292,
        "frac_adv_clamped": 0.002472,
        "resp_len_mean": 88.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.381907,
        "reward_mean": -0.349567
      },
      {
        "step": 293,
        "frac_adv_clamped": 0.005896,
        "resp_len_mean": 68.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.318626,
        "reward_mean": -0.305253
      },
      {
        "step": 294,
        "frac_adv_clamped": 0.002116,
        "resp_len_mean": 88.59375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.255078,
        "reward_mean": -0.21817
      },
      {
        "step": 295,
        "frac_adv_clamped": 0.004007,
        "resp_len_mean": 109.1875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.344541,
        "reward_mean": -0.327642
      },
      {
        "step": 296,
        "frac_adv_clamped": 0.003749,
        "resp_len_mean": 91.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.38422,
        "reward_mean": -0.384777
      },
      {
        "step": 297,
        "frac_adv_clamped": 0.003009,
        "resp_len_mean": 93.46875,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.288845,
        "reward_mean": -0.261504
      },
      {
        "step": 298,
        "frac_adv_clamped": 0.005437,
        "resp_len_mean": 63.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.256128,
        "reward_mean": -0.225188
      },
      {
        "step": 299,
        "frac_adv_clamped": 0.004574,
        "resp_len_mean": 95.65625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.329416,
        "reward_mean": -0.331249
      }
    ],
    "reward_to_go": [
      {
        "step": 0,
        "frac_adv_clamped": 0.964302,
        "resp_len_mean": 68.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.748539,
        "reward_mean": -0.812194
      },
      {
        "step": 1,
        "frac_adv_clamped": 0.991732,
        "resp_len_mean": 83.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.83282,
        "reward_mean": -0.898106
      },
      {
        "step": 2,
        "frac_adv_clamped": 0.950679,
        "resp_len_mean": 69.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.824626,
        "reward_mean": -0.927863
      },
      {
        "step": 3,
        "frac_adv_clamped": 0.966577,
        "resp_len_mean": 81.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.888542,
        "reward_mean": -0.961442
      },
      {
        "step": 4,
        "frac_adv_clamped": 0.906348,
        "resp_len_mean": 69.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.936235,
        "reward_mean": -0.965307
      },
      {
        "step": 5,
        "frac_adv_clamped": 0.929403,
        "resp_len_mean": 80.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.757,
        "reward_mean": -0.904941
      },
      {
        "step": 6,
        "frac_adv_clamped": 0.835251,
        "resp_len_mean": 59.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.792153,
        "reward_mean": -0.833173
      },
      {
        "step": 7,
        "frac_adv_clamped": 0.711651,
        "resp_len_mean": 69.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.622109,
        "reward_mean": -0.670027
      },
      {
        "step": 8,
        "frac_adv_clamped": 0.745002,
        "resp_len_mean": 89.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.65303,
        "reward_mean": -0.667214
      },
      {
        "step": 9,
        "frac_adv_clamped": 0.497585,
        "resp_len_mean": 90.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.47457,
        "reward_mean": -0.40504
      },
      {
        "step": 10,
        "frac_adv_clamped": 0.41926,
        "resp_len_mean": 74.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.51363,
        "reward_mean": -0.436624
      },
      {
        "step": 11,
        "frac_adv_clamped": 0.556647,
        "resp_len_mean": 89.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.513837,
        "reward_mean": -0.486155
      },
      {
        "step": 12,
        "frac_adv_clamped": 0.398637,
        "resp_len_mean": 73.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.385291,
        "reward_mean": -0.380315
      },
      {
        "step": 13,
        "frac_adv_clamped": 0.495909,
        "resp_len_mean": 87.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.438254,
        "reward_mean": -0.427377
      },
      {
        "step": 14,
        "frac_adv_clamped": 0.666462,
        "resp_len_mean": 101.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.498792,
        "reward_mean": -0.513749
      },
      {
        "step": 15,
        "frac_adv_clamped": 0.56919,
        "resp_len_mean": 93.71875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.647341,
        "reward_mean": -0.598119
      },
      {
        "step": 16,
        "frac_adv_clamped": 0.405518,
        "resp_len_mean": 82.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.432745,
        "reward_mean": -0.417314
      },
      {
        "step": 17,
        "frac_adv_clamped": 0.627262,
        "resp_len_mean": 110.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.481298,
        "reward_mean": -0.49227
      },
      {
        "step": 18,
        "frac_adv_clamped": 0.702235,
        "resp_len_mean": 117.4375,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.644671,
        "reward_mean": -0.659063
      },
      {
        "step": 19,
        "frac_adv_clamped": 0.522038,
        "resp_len_mean": 86.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.494978,
        "reward_mean": -0.497195
      },
      {
        "step": 20,
        "frac_adv_clamped": 0.457627,
        "resp_len_mean": 86.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.648879,
        "reward_mean": -0.596622
      },
      {
        "step": 21,
        "frac_adv_clamped": 0.605617,
        "resp_len_mean": 104.59375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.515973,
        "reward_mean": -0.526853
      },
      {
        "step": 22,
        "frac_adv_clamped": 0.629847,
        "resp_len_mean": 122.5,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.520294,
        "reward_mean": -0.476209
      },
      {
        "step": 23,
        "frac_adv_clamped": 0.434765,
        "resp_len_mean": 77.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.403589,
        "reward_mean": -0.381476
      },
      {
        "step": 24,
        "frac_adv_clamped": 0.611875,
        "resp_len_mean": 93.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.567656,
        "reward_mean": -0.536199
      },
      {
        "step": 25,
        "frac_adv_clamped": 0.565125,
        "resp_len_mean": 117.5625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.423378,
        "reward_mean": -0.424917
      },
      {
        "step": 26,
        "frac_adv_clamped": 0.63687,
        "resp_len_mean": 117.8125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.50551,
        "reward_mean": -0.514586
      },
      {
        "step": 27,
        "frac_adv_clamped": 0.529909,
        "resp_len_mean": 96.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.472153,
        "reward_mean": -0.409704
      },
      {
        "step": 28,
        "frac_adv_clamped": 0.643368,
        "resp_len_mean": 96.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.552447,
        "reward_mean": -0.500075
      },
      {
        "step": 29,
        "frac_adv_clamped": 0.508069,
        "resp_len_mean": 108.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.438069,
        "reward_mean": -0.4016
      },
      {
        "step": 30,
        "frac_adv_clamped": 0.614989,
        "resp_len_mean": 97.15625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.507547,
        "reward_mean": -0.464564
      },
      {
        "step": 31,
        "frac_adv_clamped": 0.483547,
        "resp_len_mean": 102.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.449719,
        "reward_mean": -0.397571
      },
      {
        "step": 32,
        "frac_adv_clamped": 0.398037,
        "resp_len_mean": 101.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.326836,
        "reward_mean": -0.286287
      },
      {
        "step": 33,
        "frac_adv_clamped": 0.658,
        "resp_len_mean": 109.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.435138,
        "reward_mean": -0.434634
      },
      {
        "step": 34,
        "frac_adv_clamped": 0.578429,
        "resp_len_mean": 114.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.32412,
        "reward_mean": -0.295711
      },
      {
        "step": 35,
        "frac_adv_clamped": 0.610778,
        "resp_len_mean": 120.03125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.422258,
        "reward_mean": -0.411528
      },
      {
        "step": 36,
        "frac_adv_clamped": 0.505642,
        "resp_len_mean": 102.46875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.419841,
        "reward_mean": -0.397386
      },
      {
        "step": 37,
        "frac_adv_clamped": 0.541932,
        "resp_len_mean": 122.59375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.463994,
        "reward_mean": -0.445994
      },
      {
        "step": 38,
        "frac_adv_clamped": 0.573258,
        "resp_len_mean": 104.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.452657,
        "reward_mean": -0.414448
      },
      {
        "step": 39,
        "frac_adv_clamped": 0.592045,
        "resp_len_mean": 116.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.44987,
        "reward_mean": -0.409372
      },
      {
        "step": 40,
        "frac_adv_clamped": 0.533476,
        "resp_len_mean": 116.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.431839,
        "reward_mean": -0.393846
      },
      {
        "step": 41,
        "frac_adv_clamped": 0.632162,
        "resp_len_mean": 112.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.60055,
        "reward_mean": -0.582511
      },
      {
        "step": 42,
        "frac_adv_clamped": 0.535624,
        "resp_len_mean": 111.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.395632,
        "reward_mean": -0.41075
      },
      {
        "step": 43,
        "frac_adv_clamped": 0.405805,
        "resp_len_mean": 111.96875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.378495,
        "reward_mean": -0.352306
      },
      {
        "step": 44,
        "frac_adv_clamped": 0.540087,
        "resp_len_mean": 93.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.51597,
        "reward_mean": -0.535198
      },
      {
        "step": 45,
        "frac_adv_clamped": 0.547925,
        "resp_len_mean": 106.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.461014,
        "reward_mean": -0.420513
      },
      {
        "step": 46,
        "frac_adv_clamped": 0.611225,
        "resp_len_mean": 121.9375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.41467,
        "reward_mean": -0.408011
      },
      {
        "step": 47,
        "frac_adv_clamped": 0.609819,
        "resp_len_mean": 108.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.457369,
        "reward_mean": -0.441154
      },
      {
        "step": 48,
        "frac_adv_clamped": 0.498074,
        "resp_len_mean": 113.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.38448,
        "reward_mean": -0.38734
      },
      {
        "step": 49,
        "frac_adv_clamped": 0.531292,
        "resp_len_mean": 116.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.442865,
        "reward_mean": -0.44392
      },
      {
        "step": 50,
        "frac_adv_clamped": 0.625309,
        "resp_len_mean": 139.03125,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.45573,
        "reward_mean": -0.447364
      },
      {
        "step": 51,
        "frac_adv_clamped": 0.449442,
        "resp_len_mean": 103.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.41277,
        "reward_mean": -0.402209
      },
      {
        "step": 52,
        "frac_adv_clamped": 0.619183,
        "resp_len_mean": 131.625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.459281,
        "reward_mean": -0.435676
      },
      {
        "step": 53,
        "frac_adv_clamped": 0.658616,
        "resp_len_mean": 143.625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.450308,
        "reward_mean": -0.449235
      },
      {
        "step": 54,
        "frac_adv_clamped": 0.660758,
        "resp_len_mean": 141.03125,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.422126,
        "reward_mean": -0.416881
      },
      {
        "step": 55,
        "frac_adv_clamped": 0.666161,
        "resp_len_mean": 164.84375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.377296,
        "reward_mean": -0.358886
      },
      {
        "step": 56,
        "frac_adv_clamped": 0.742372,
        "resp_len_mean": 171.03125,
        "frac_finished": 0.84375,
        "kl_full_tok": 0.386709,
        "reward_mean": -0.392677
      },
      {
        "step": 57,
        "frac_adv_clamped": 0.563994,
        "resp_len_mean": 139.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.420959,
        "reward_mean": -0.404096
      },
      {
        "step": 58,
        "frac_adv_clamped": 0.628076,
        "resp_len_mean": 158.71875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.372624,
        "reward_mean": -0.359079
      },
      {
        "step": 59,
        "frac_adv_clamped": 0.597458,
        "resp_len_mean": 154.875,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.38834,
        "reward_mean": -0.377595
      },
      {
        "step": 60,
        "frac_adv_clamped": 0.479429,
        "resp_len_mean": 146.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.407396,
        "reward_mean": -0.409505
      },
      {
        "step": 61,
        "frac_adv_clamped": 0.636212,
        "resp_len_mean": 149.46875,
        "frac_finished": 0.875,
        "kl_full_tok": 0.425615,
        "reward_mean": -0.399633
      },
      {
        "step": 62,
        "frac_adv_clamped": 0.539128,
        "resp_len_mean": 129.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.378706,
        "reward_mean": -0.387406
      },
      {
        "step": 63,
        "frac_adv_clamped": 0.631918,
        "resp_len_mean": 155.28125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.438513,
        "reward_mean": -0.43685
      },
      {
        "step": 64,
        "frac_adv_clamped": 0.605409,
        "resp_len_mean": 152.53125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.402065,
        "reward_mean": -0.417951
      },
      {
        "step": 65,
        "frac_adv_clamped": 0.654774,
        "resp_len_mean": 178.6875,
        "frac_finished": 0.78125,
        "kl_full_tok": 0.390817,
        "reward_mean": -0.399078
      },
      {
        "step": 66,
        "frac_adv_clamped": 0.621844,
        "resp_len_mean": 170.8125,
        "frac_finished": 0.875,
        "kl_full_tok": 0.377355,
        "reward_mean": -0.354342
      },
      {
        "step": 67,
        "frac_adv_clamped": 0.654424,
        "resp_len_mean": 191.4375,
        "frac_finished": 0.875,
        "kl_full_tok": 0.379523,
        "reward_mean": -0.380778
      },
      {
        "step": 68,
        "frac_adv_clamped": 0.57268,
        "resp_len_mean": 210.46875,
        "frac_finished": 0.6875,
        "kl_full_tok": 0.377239,
        "reward_mean": -0.374935
      },
      {
        "step": 69,
        "frac_adv_clamped": 0.674889,
        "resp_len_mean": 183.6875,
        "frac_finished": 0.75,
        "kl_full_tok": 0.395488,
        "reward_mean": -0.399571
      },
      {
        "step": 70,
        "frac_adv_clamped": 0.647773,
        "resp_len_mean": 162.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.435275,
        "reward_mean": -0.42584
      },
      {
        "step": 71,
        "frac_adv_clamped": 0.691776,
        "resp_len_mean": 171.75,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.465428,
        "reward_mean": -0.457632
      },
      {
        "step": 72,
        "frac_adv_clamped": 0.570727,
        "resp_len_mean": 209.875,
        "frac_finished": 0.71875,
        "kl_full_tok": 0.33086,
        "reward_mean": -0.318238
      },
      {
        "step": 73,
        "frac_adv_clamped": 0.648182,
        "resp_len_mean": 189.90625,
        "frac_finished": 0.6875,
        "kl_full_tok": 0.398238,
        "reward_mean": -0.389096
      },
      {
        "step": 74,
        "frac_adv_clamped": 0.612245,
        "resp_len_mean": 196.0,
        "frac_finished": 0.71875,
        "kl_full_tok": 0.405041,
        "reward_mean": -0.397868
      },
      {
        "step": 75,
        "frac_adv_clamped": 0.647802,
        "resp_len_mean": 195.46875,
        "frac_finished": 0.75,
        "kl_full_tok": 0.40581,
        "reward_mean": -0.400859
      },
      {
        "step": 76,
        "frac_adv_clamped": 0.566865,
        "resp_len_mean": 204.46875,
        "frac_finished": 0.84375,
        "kl_full_tok": 0.316879,
        "reward_mean": -0.31481
      },
      {
        "step": 77,
        "frac_adv_clamped": 0.624005,
        "resp_len_mean": 188.5,
        "frac_finished": 0.8125,
        "kl_full_tok": 0.365306,
        "reward_mean": -0.355931
      },
      {
        "step": 78,
        "frac_adv_clamped": 0.729328,
        "resp_len_mean": 205.96875,
        "frac_finished": 0.65625,
        "kl_full_tok": 0.440678,
        "reward_mean": -0.433079
      },
      {
        "step": 79,
        "frac_adv_clamped": 0.640024,
        "resp_len_mean": 204.875,
        "frac_finished": 0.71875,
        "kl_full_tok": 0.385157,
        "reward_mean": -0.386691
      },
      {
        "step": 80,
        "frac_adv_clamped": 0.581291,
        "resp_len_mean": 173.375,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.3797,
        "reward_mean": -0.356153
      },
      {
        "step": 81,
        "frac_adv_clamped": 0.651964,
        "resp_len_mean": 187.75,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.413317,
        "reward_mean": -0.416589
      },
      {
        "step": 82,
        "frac_adv_clamped": 0.631244,
        "resp_len_mean": 210.84375,
        "frac_finished": 0.625,
        "kl_full_tok": 0.352941,
        "reward_mean": -0.3278
      },
      {
        "step": 83,
        "frac_adv_clamped": 0.668099,
        "resp_len_mean": 203.65625,
        "frac_finished": 0.71875,
        "kl_full_tok": 0.38394,
        "reward_mean": -0.367955
      },
      {
        "step": 84,
        "frac_adv_clamped": 0.713616,
        "resp_len_mean": 193.25,
        "frac_finished": 0.84375,
        "kl_full_tok": 0.442837,
        "reward_mean": -0.422221
      },
      {
        "step": 85,
        "frac_adv_clamped": 0.71484,
        "resp_len_mean": 193.3125,
        "frac_finished": 0.8125,
        "kl_full_tok": 0.421816,
        "reward_mean": -0.416331
      },
      {
        "step": 86,
        "frac_adv_clamped": 0.731328,
        "resp_len_mean": 182.84375,
        "frac_finished": 0.6875,
        "kl_full_tok": 0.460679,
        "reward_mean": -0.46827
      },
      {
        "step": 87,
        "frac_adv_clamped": 0.698652,
        "resp_len_mean": 215.59375,
        "frac_finished": 0.5,
        "kl_full_tok": 0.415252,
        "reward_mean": -0.400427
      },
      {
        "step": 88,
        "frac_adv_clamped": 0.763111,
        "resp_len_mean": 225.84375,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.413156,
        "reward_mean": -0.395971
      },
      {
        "step": 89,
        "frac_adv_clamped": 0.792943,
        "resp_len_mean": 234.6875,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.423062,
        "reward_mean": -0.441653
      },
      {
        "step": 90,
        "frac_adv_clamped": 0.706341,
        "resp_len_mean": 212.40625,
        "frac_finished": 0.53125,
        "kl_full_tok": 0.477466,
        "reward_mean": -0.454927
      },
      {
        "step": 91,
        "frac_adv_clamped": 0.7623,
        "resp_len_mean": 218.5,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.497528,
        "reward_mean": -0.498497
      },
      {
        "step": 92,
        "frac_adv_clamped": 0.713472,
        "resp_len_mean": 219.4375,
        "frac_finished": 0.59375,
        "kl_full_tok": 0.485144,
        "reward_mean": -0.498823
      },
      {
        "step": 93,
        "frac_adv_clamped": 0.748322,
        "resp_len_mean": 223.5,
        "frac_finished": 0.59375,
        "kl_full_tok": 0.392452,
        "reward_mean": -0.387572
      },
      {
        "step": 94,
        "frac_adv_clamped": 0.677199,
        "resp_len_mean": 214.625,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.424816,
        "reward_mean": -0.431341
      },
      {
        "step": 95,
        "frac_adv_clamped": 0.77485,
        "resp_len_mean": 219.4375,
        "frac_finished": 0.59375,
        "kl_full_tok": 0.476541,
        "reward_mean": -0.478542
      },
      {
        "step": 96,
        "frac_adv_clamped": 0.609663,
        "resp_len_mean": 228.96875,
        "frac_finished": 0.53125,
        "kl_full_tok": 0.351041,
        "reward_mean": -0.329405
      },
      {
        "step": 97,
        "frac_adv_clamped": 0.784828,
        "resp_len_mean": 178.78125,
        "frac_finished": 0.71875,
        "kl_full_tok": 0.506485,
        "reward_mean": -0.516354
      },
      {
        "step": 98,
        "frac_adv_clamped": 0.789329,
        "resp_len_mean": 205.0,
        "frac_finished": 0.65625,
        "kl_full_tok": 0.454202,
        "reward_mean": -0.478125
      },
      {
        "step": 99,
        "frac_adv_clamped": 0.759248,
        "resp_len_mean": 206.125,
        "frac_finished": 0.5625,
        "kl_full_tok": 0.471245,
        "reward_mean": -0.475992
      },
      {
        "step": 100,
        "frac_adv_clamped": 0.701574,
        "resp_len_mean": 172.78125,
        "frac_finished": 0.8125,
        "kl_full_tok": 0.419456,
        "reward_mean": -0.408142
      },
      {
        "step": 101,
        "frac_adv_clamped": 0.728371,
        "resp_len_mean": 222.5,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.454211,
        "reward_mean": -0.462594
      },
      {
        "step": 102,
        "frac_adv_clamped": 0.7904,
        "resp_len_mean": 220.0625,
        "frac_finished": 0.5,
        "kl_full_tok": 0.443741,
        "reward_mean": -0.446193
      },
      {
        "step": 103,
        "frac_adv_clamped": 0.699151,
        "resp_len_mean": 220.9375,
        "frac_finished": 0.625,
        "kl_full_tok": 0.419887,
        "reward_mean": -0.42204
      },
      {
        "step": 104,
        "frac_adv_clamped": 0.706965,
        "resp_len_mean": 220.75,
        "frac_finished": 0.5,
        "kl_full_tok": 0.40816,
        "reward_mean": -0.386842
      },
      {
        "step": 105,
        "frac_adv_clamped": 0.749057,
        "resp_len_mean": 215.4375,
        "frac_finished": 0.53125,
        "kl_full_tok": 0.46054,
        "reward_mean": -0.448083
      },
      {
        "step": 106,
        "frac_adv_clamped": 0.774234,
        "resp_len_mean": 225.34375,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.485127,
        "reward_mean": -0.485045
      },
      {
        "step": 107,
        "frac_adv_clamped": 0.696705,
        "resp_len_mean": 221.9375,
        "frac_finished": 0.5625,
        "kl_full_tok": 0.428736,
        "reward_mean": -0.433211
      },
      {
        "step": 108,
        "frac_adv_clamped": 0.664538,
        "resp_len_mean": 230.0,
        "frac_finished": 0.59375,
        "kl_full_tok": 0.403012,
        "reward_mean": -0.407854
      },
      {
        "step": 109,
        "frac_adv_clamped": 0.750145,
        "resp_len_mean": 215.625,
        "frac_finished": 0.625,
        "kl_full_tok": 0.439265,
        "reward_mean": -0.449256
      },
      {
        "step": 110,
        "frac_adv_clamped": 0.725543,
        "resp_len_mean": 233.1875,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.382525,
        "reward_mean": -0.388363
      },
      {
        "step": 111,
        "frac_adv_clamped": 0.720328,
        "resp_len_mean": 243.8125,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.454407,
        "reward_mean": -0.462352
      },
      {
        "step": 112,
        "frac_adv_clamped": 0.761308,
        "resp_len_mean": 236.96875,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.424732,
        "reward_mean": -0.43548
      },
      {
        "step": 113,
        "frac_adv_clamped": 0.734992,
        "resp_len_mean": 209.78125,
        "frac_finished": 0.6875,
        "kl_full_tok": 0.458344,
        "reward_mean": -0.450954
      },
      {
        "step": 114,
        "frac_adv_clamped": 0.770054,
        "resp_len_mean": 218.9375,
        "frac_finished": 0.625,
        "kl_full_tok": 0.43926,
        "reward_mean": -0.426478
      },
      {
        "step": 115,
        "frac_adv_clamped": 0.782314,
        "resp_len_mean": 235.0,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.44699,
        "reward_mean": -0.459202
      },
      {
        "step": 116,
        "frac_adv_clamped": 0.788862,
        "resp_len_mean": 231.1875,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.438549,
        "reward_mean": -0.424645
      },
      {
        "step": 117,
        "frac_adv_clamped": 0.772197,
        "resp_len_mean": 219.625,
        "frac_finished": 0.59375,
        "kl_full_tok": 0.483856,
        "reward_mean": -0.48933
      },
      {
        "step": 118,
        "frac_adv_clamped": 0.779358,
        "resp_len_mean": 232.84375,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.475922,
        "reward_mean": -0.46065
      },
      {
        "step": 119,
        "frac_adv_clamped": 0.768502,
        "resp_len_mean": 227.59375,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.488391,
        "reward_mean": -0.477689
      },
      {
        "step": 120,
        "frac_adv_clamped": 0.704518,
        "resp_len_mean": 230.34375,
        "frac_finished": 0.375,
        "kl_full_tok": 0.383256,
        "reward_mean": -0.358156
      },
      {
        "step": 121,
        "frac_adv_clamped": 0.771272,
        "resp_len_mean": 222.5625,
        "frac_finished": 0.53125,
        "kl_full_tok": 0.517518,
        "reward_mean": -0.509431
      },
      {
        "step": 122,
        "frac_adv_clamped": 0.815219,
        "resp_len_mean": 237.78125,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.509932,
        "reward_mean": -0.507277
      },
      {
        "step": 123,
        "frac_adv_clamped": 0.875759,
        "resp_len_mean": 247.0,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.520766,
        "reward_mean": -0.518352
      },
      {
        "step": 124,
        "frac_adv_clamped": 0.801978,
        "resp_len_mean": 243.34375,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.458503,
        "reward_mean": -0.455868
      },
      {
        "step": 125,
        "frac_adv_clamped": 0.800963,
        "resp_len_mean": 220.75,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.469018,
        "reward_mean": -0.464039
      },
      {
        "step": 126,
        "frac_adv_clamped": 0.725416,
        "resp_len_mean": 223.40625,
        "frac_finished": 0.59375,
        "kl_full_tok": 0.475474,
        "reward_mean": -0.480053
      },
      {
        "step": 127,
        "frac_adv_clamped": 0.824231,
        "resp_len_mean": 238.59375,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.514075,
        "reward_mean": -0.516191
      },
      {
        "step": 128,
        "frac_adv_clamped": 0.754105,
        "resp_len_mean": 226.46875,
        "frac_finished": 0.5,
        "kl_full_tok": 0.478841,
        "reward_mean": -0.470787
      },
      {
        "step": 129,
        "frac_adv_clamped": 0.772137,
        "resp_len_mean": 235.75,
        "frac_finished": 0.375,
        "kl_full_tok": 0.450253,
        "reward_mean": -0.447049
      },
      {
        "step": 130,
        "frac_adv_clamped": 0.759122,
        "resp_len_mean": 236.375,
        "frac_finished": 0.53125,
        "kl_full_tok": 0.520069,
        "reward_mean": -0.520751
      },
      {
        "step": 131,
        "frac_adv_clamped": 0.755626,
        "resp_len_mean": 237.46875,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.457837,
        "reward_mean": -0.461837
      },
      {
        "step": 132,
        "frac_adv_clamped": 0.798089,
        "resp_len_mean": 245.3125,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.486736,
        "reward_mean": -0.501294
      },
      {
        "step": 133,
        "frac_adv_clamped": 0.739831,
        "resp_len_mean": 248.15625,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.468578,
        "reward_mean": -0.459758
      },
      {
        "step": 134,
        "frac_adv_clamped": 0.783278,
        "resp_len_mean": 251.90625,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.452541,
        "reward_mean": -0.44269
      },
      {
        "step": 135,
        "frac_adv_clamped": 0.772304,
        "resp_len_mean": 244.84375,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.477697,
        "reward_mean": -0.460095
      },
      {
        "step": 136,
        "frac_adv_clamped": 0.806938,
        "resp_len_mean": 242.3125,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.554198,
        "reward_mean": -0.543867
      },
      {
        "step": 137,
        "frac_adv_clamped": 0.739971,
        "resp_len_mean": 240.71875,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.446465,
        "reward_mean": -0.457067
      },
      {
        "step": 138,
        "frac_adv_clamped": 0.826285,
        "resp_len_mean": 240.15625,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.508482,
        "reward_mean": -0.501856
      },
      {
        "step": 139,
        "frac_adv_clamped": 0.803536,
        "resp_len_mean": 236.84375,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.54658,
        "reward_mean": -0.574527
      },
      {
        "step": 140,
        "frac_adv_clamped": 0.7013,
        "resp_len_mean": 240.3125,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.459486,
        "reward_mean": -0.42268
      },
      {
        "step": 141,
        "frac_adv_clamped": 0.807439,
        "resp_len_mean": 232.71875,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.536529,
        "reward_mean": -0.535455
      },
      {
        "step": 142,
        "frac_adv_clamped": 0.799067,
        "resp_len_mean": 241.0625,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.499371,
        "reward_mean": -0.489843
      },
      {
        "step": 143,
        "frac_adv_clamped": 0.794359,
        "resp_len_mean": 243.75,
        "frac_finished": 0.25,
        "kl_full_tok": 0.51429,
        "reward_mean": -0.513296
      },
      {
        "step": 144,
        "frac_adv_clamped": 0.755437,
        "resp_len_mean": 245.71875,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.526693,
        "reward_mean": -0.517777
      },
      {
        "step": 145,
        "frac_adv_clamped": 0.811448,
        "resp_len_mean": 250.59375,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.524053,
        "reward_mean": -0.516862
      },
      {
        "step": 146,
        "frac_adv_clamped": 0.766599,
        "resp_len_mean": 244.75,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.510267,
        "reward_mean": -0.479117
      },
      {
        "step": 147,
        "frac_adv_clamped": 0.808102,
        "resp_len_mean": 252.25,
        "frac_finished": 0.0625,
        "kl_full_tok": 0.51818,
        "reward_mean": -0.512076
      },
      {
        "step": 148,
        "frac_adv_clamped": 0.798224,
        "resp_len_mean": 246.40625,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.512692,
        "reward_mean": -0.522434
      },
      {
        "step": 149,
        "frac_adv_clamped": 0.783628,
        "resp_len_mean": 243.9375,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.536356,
        "reward_mean": -0.538451
      },
      {
        "step": 150,
        "frac_adv_clamped": 0.830278,
        "resp_len_mean": 252.25,
        "frac_finished": 0.0625,
        "kl_full_tok": 0.559202,
        "reward_mean": -0.552005
      },
      {
        "step": 151,
        "frac_adv_clamped": 0.80529,
        "resp_len_mean": 248.125,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.555132,
        "reward_mean": -0.568871
      },
      {
        "step": 152,
        "frac_adv_clamped": 0.787361,
        "resp_len_mean": 241.3125,
        "frac_finished": 0.25,
        "kl_full_tok": 0.510398,
        "reward_mean": -0.496384
      },
      {
        "step": 153,
        "frac_adv_clamped": 0.780348,
        "resp_len_mean": 240.4375,
        "frac_finished": 0.25,
        "kl_full_tok": 0.533162,
        "reward_mean": -0.511125
      },
      {
        "step": 154,
        "frac_adv_clamped": 0.77421,
        "resp_len_mean": 250.09375,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.56,
        "reward_mean": -0.53067
      },
      {
        "step": 155,
        "frac_adv_clamped": 0.835271,
        "resp_len_mean": 252.6875,
        "frac_finished": 0.125,
        "kl_full_tok": 0.59926,
        "reward_mean": -0.605312
      },
      {
        "step": 156,
        "frac_adv_clamped": 0.815845,
        "resp_len_mean": 251.65625,
        "frac_finished": 0.09375,
        "kl_full_tok": 0.574203,
        "reward_mean": -0.579124
      },
      {
        "step": 157,
        "frac_adv_clamped": 0.847652,
        "resp_len_mean": 253.53125,
        "frac_finished": 0.125,
        "kl_full_tok": 0.581052,
        "reward_mean": -0.579579
      },
      {
        "step": 158,
        "frac_adv_clamped": 0.821187,
        "resp_len_mean": 249.5625,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.559058,
        "reward_mean": -0.577683
      },
      {
        "step": 159,
        "frac_adv_clamped": 0.816461,
        "resp_len_mean": 250.96875,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.532654,
        "reward_mean": -0.55327
      },
      {
        "step": 160,
        "frac_adv_clamped": 0.753565,
        "resp_len_mean": 249.8125,
        "frac_finished": 0.25,
        "kl_full_tok": 0.479472,
        "reward_mean": -0.478953
      },
      {
        "step": 161,
        "frac_adv_clamped": 0.788366,
        "resp_len_mean": 250.875,
        "frac_finished": 0.125,
        "kl_full_tok": 0.573073,
        "reward_mean": -0.547482
      },
      {
        "step": 162,
        "frac_adv_clamped": 0.813473,
        "resp_len_mean": 246.78125,
        "frac_finished": 0.125,
        "kl_full_tok": 0.566671,
        "reward_mean": -0.5691
      },
      {
        "step": 163,
        "frac_adv_clamped": 0.816116,
        "resp_len_mean": 251.6875,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.590046,
        "reward_mean": -0.580858
      },
      {
        "step": 164,
        "frac_adv_clamped": 0.844849,
        "resp_len_mean": 256.0,
        "frac_finished": 0.0,
        "kl_full_tok": 0.590467,
        "reward_mean": -0.612609
      },
      {
        "step": 165,
        "frac_adv_clamped": 0.814671,
        "resp_len_mean": 250.0625,
        "frac_finished": 0.125,
        "kl_full_tok": 0.545483,
        "reward_mean": -0.54064
      },
      {
        "step": 166,
        "frac_adv_clamped": 0.757855,
        "resp_len_mean": 247.65625,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.525479,
        "reward_mean": -0.52679
      },
      {
        "step": 167,
        "frac_adv_clamped": 0.789568,
        "resp_len_mean": 243.84375,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.509512,
        "reward_mean": -0.517089
      },
      {
        "step": 168,
        "frac_adv_clamped": 0.820423,
        "resp_len_mean": 248.5,
        "frac_finished": 0.09375,
        "kl_full_tok": 0.547522,
        "reward_mean": -0.567162
      },
      {
        "step": 169,
        "frac_adv_clamped": 0.830448,
        "resp_len_mean": 236.46875,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.57759,
        "reward_mean": -0.583201
      },
      {
        "step": 170,
        "frac_adv_clamped": 0.722749,
        "resp_len_mean": 247.40625,
        "frac_finished": 0.25,
        "kl_full_tok": 0.483525,
        "reward_mean": -0.474563
      },
      {
        "step": 171,
        "frac_adv_clamped": 0.836857,
        "resp_len_mean": 255.71875,
        "frac_finished": 0.03125,
        "kl_full_tok": 0.541019,
        "reward_mean": -0.531204
      },
      {
        "step": 172,
        "frac_adv_clamped": 0.82684,
        "resp_len_mean": 245.4375,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.519144,
        "reward_mean": -0.509747
      },
      {
        "step": 173,
        "frac_adv_clamped": 0.798584,
        "resp_len_mean": 256.0,
        "frac_finished": 0.0,
        "kl_full_tok": 0.530988,
        "reward_mean": -0.560348
      },
      {
        "step": 174,
        "frac_adv_clamped": 0.838798,
        "resp_len_mean": 251.625,
        "frac_finished": 0.09375,
        "kl_full_tok": 0.551692,
        "reward_mean": -0.567669
      },
      {
        "step": 175,
        "frac_adv_clamped": 0.822759,
        "resp_len_mean": 245.78125,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.563936,
        "reward_mean": -0.5681
      },
      {
        "step": 176,
        "frac_adv_clamped": 0.818695,
        "resp_len_mean": 232.34375,
        "frac_finished": 0.375,
        "kl_full_tok": 0.566115,
        "reward_mean": -0.587627
      },
      {
        "step": 177,
        "frac_adv_clamped": 0.715774,
        "resp_len_mean": 239.90625,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.445364,
        "reward_mean": -0.453262
      },
      {
        "step": 178,
        "frac_adv_clamped": 0.747014,
        "resp_len_mean": 230.25,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.452618,
        "reward_mean": -0.449229
      },
      {
        "step": 179,
        "frac_adv_clamped": 0.793448,
        "resp_len_mean": 237.53125,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.531872,
        "reward_mean": -0.519714
      },
      {
        "step": 180,
        "frac_adv_clamped": 0.797636,
        "resp_len_mean": 240.59375,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.495953,
        "reward_mean": -0.475427
      },
      {
        "step": 181,
        "frac_adv_clamped": 0.792453,
        "resp_len_mean": 238.5,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.492215,
        "reward_mean": -0.489443
      },
      {
        "step": 182,
        "frac_adv_clamped": 0.660806,
        "resp_len_mean": 236.40625,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.441456,
        "reward_mean": -0.419509
      },
      {
        "step": 183,
        "frac_adv_clamped": 0.798723,
        "resp_len_mean": 234.90625,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.489223,
        "reward_mean": -0.47759
      },
      {
        "step": 184,
        "frac_adv_clamped": 0.808785,
        "resp_len_mean": 237.625,
        "frac_finished": 0.25,
        "kl_full_tok": 0.493416,
        "reward_mean": -0.509392
      },
      {
        "step": 185,
        "frac_adv_clamped": 0.822293,
        "resp_len_mean": 245.3125,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.509564,
        "reward_mean": -0.502817
      },
      {
        "step": 186,
        "frac_adv_clamped": 0.69863,
        "resp_len_mean": 212.15625,
        "frac_finished": 0.6875,
        "kl_full_tok": 0.487252,
        "reward_mean": -0.494887
      },
      {
        "step": 187,
        "frac_adv_clamped": 0.685262,
        "resp_len_mean": 217.34375,
        "frac_finished": 0.53125,
        "kl_full_tok": 0.422629,
        "reward_mean": -0.405923
      },
      {
        "step": 188,
        "frac_adv_clamped": 0.727979,
        "resp_len_mean": 233.4375,
        "frac_finished": 0.375,
        "kl_full_tok": 0.46378,
        "reward_mean": -0.458936
      },
      {
        "step": 189,
        "frac_adv_clamped": 0.766229,
        "resp_len_mean": 233.0,
        "frac_finished": 0.53125,
        "kl_full_tok": 0.437771,
        "reward_mean": -0.432496
      },
      {
        "step": 190,
        "frac_adv_clamped": 0.774015,
        "resp_len_mean": 215.03125,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.515322,
        "reward_mean": -0.511171
      },
      {
        "step": 191,
        "frac_adv_clamped": 0.78825,
        "resp_len_mean": 219.15625,
        "frac_finished": 0.59375,
        "kl_full_tok": 0.502146,
        "reward_mean": -0.501336
      },
      {
        "step": 192,
        "frac_adv_clamped": 0.743539,
        "resp_len_mean": 220.0625,
        "frac_finished": 0.625,
        "kl_full_tok": 0.497762,
        "reward_mean": -0.483326
      },
      {
        "step": 193,
        "frac_adv_clamped": 0.733847,
        "resp_len_mean": 222.96875,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.44934,
        "reward_mean": -0.430055
      },
      {
        "step": 194,
        "frac_adv_clamped": 0.799383,
        "resp_len_mean": 233.03125,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.461,
        "reward_mean": -0.467154
      },
      {
        "step": 195,
        "frac_adv_clamped": 0.803914,
        "resp_len_mean": 234.75,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.486545,
        "reward_mean": -0.50894
      },
      {
        "step": 196,
        "frac_adv_clamped": 0.802508,
        "resp_len_mean": 226.75,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.547635,
        "reward_mean": -0.542909
      },
      {
        "step": 197,
        "frac_adv_clamped": 0.767022,
        "resp_len_mean": 199.1875,
        "frac_finished": 0.625,
        "kl_full_tok": 0.49763,
        "reward_mean": -0.477461
      },
      {
        "step": 198,
        "frac_adv_clamped": 0.8196,
        "resp_len_mean": 234.375,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.508587,
        "reward_mean": -0.49051
      },
      {
        "step": 199,
        "frac_adv_clamped": 0.702338,
        "resp_len_mean": 215.21875,
        "frac_finished": 0.53125,
        "kl_full_tok": 0.495369,
        "reward_mean": -0.48282
      },
      {
        "step": 200,
        "frac_adv_clamped": 0.844626,
        "resp_len_mean": 245.375,
        "frac_finished": 0.25,
        "kl_full_tok": 0.463366,
        "reward_mean": -0.451687
      },
      {
        "step": 201,
        "frac_adv_clamped": 0.84857,
        "resp_len_mean": 234.84375,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.477477,
        "reward_mean": -0.498006
      },
      {
        "step": 202,
        "frac_adv_clamped": 0.723585,
        "resp_len_mean": 239.5625,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.455516,
        "reward_mean": -0.458492
      },
      {
        "step": 203,
        "frac_adv_clamped": 0.76445,
        "resp_len_mean": 222.75,
        "frac_finished": 0.5625,
        "kl_full_tok": 0.440886,
        "reward_mean": -0.429442
      },
      {
        "step": 204,
        "frac_adv_clamped": 0.799171,
        "resp_len_mean": 226.25,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.490488,
        "reward_mean": -0.505611
      },
      {
        "step": 205,
        "frac_adv_clamped": 0.794361,
        "resp_len_mean": 241.625,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.486939,
        "reward_mean": -0.49663
      },
      {
        "step": 206,
        "frac_adv_clamped": 0.787614,
        "resp_len_mean": 214.96875,
        "frac_finished": 0.625,
        "kl_full_tok": 0.522914,
        "reward_mean": -0.525078
      },
      {
        "step": 207,
        "frac_adv_clamped": 0.735188,
        "resp_len_mean": 241.5625,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.454661,
        "reward_mean": -0.468762
      },
      {
        "step": 208,
        "frac_adv_clamped": 0.802008,
        "resp_len_mean": 233.4375,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.528377,
        "reward_mean": -0.513955
      },
      {
        "step": 209,
        "frac_adv_clamped": 0.783134,
        "resp_len_mean": 240.5,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.502608,
        "reward_mean": -0.487931
      },
      {
        "step": 210,
        "frac_adv_clamped": 0.727486,
        "resp_len_mean": 240.125,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.445608,
        "reward_mean": -0.465044
      },
      {
        "step": 211,
        "frac_adv_clamped": 0.775473,
        "resp_len_mean": 224.5,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.525218,
        "reward_mean": -0.520327
      },
      {
        "step": 212,
        "frac_adv_clamped": 0.811881,
        "resp_len_mean": 246.1875,
        "frac_finished": 0.125,
        "kl_full_tok": 0.498469,
        "reward_mean": -0.495131
      },
      {
        "step": 213,
        "frac_adv_clamped": 0.849015,
        "resp_len_mean": 242.78125,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.531951,
        "reward_mean": -0.543509
      },
      {
        "step": 214,
        "frac_adv_clamped": 0.72463,
        "resp_len_mean": 236.5,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.435967,
        "reward_mean": -0.458777
      },
      {
        "step": 215,
        "frac_adv_clamped": 0.796191,
        "resp_len_mean": 247.78125,
        "frac_finished": 0.125,
        "kl_full_tok": 0.521951,
        "reward_mean": -0.523841
      },
      {
        "step": 216,
        "frac_adv_clamped": 0.791566,
        "resp_len_mean": 246.03125,
        "frac_finished": 0.25,
        "kl_full_tok": 0.516257,
        "reward_mean": -0.520853
      },
      {
        "step": 217,
        "frac_adv_clamped": 0.810143,
        "resp_len_mean": 242.78125,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.534629,
        "reward_mean": -0.53659
      },
      {
        "step": 218,
        "frac_adv_clamped": 0.791588,
        "resp_len_mean": 247.40625,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.469181,
        "reward_mean": -0.468944
      },
      {
        "step": 219,
        "frac_adv_clamped": 0.800964,
        "resp_len_mean": 246.34375,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.495818,
        "reward_mean": -0.494199
      },
      {
        "step": 220,
        "frac_adv_clamped": 0.791354,
        "resp_len_mean": 241.4375,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.475094,
        "reward_mean": -0.460714
      },
      {
        "step": 221,
        "frac_adv_clamped": 0.671965,
        "resp_len_mean": 228.0625,
        "frac_finished": 0.5,
        "kl_full_tok": 0.467454,
        "reward_mean": -0.464642
      },
      {
        "step": 222,
        "frac_adv_clamped": 0.77745,
        "resp_len_mean": 243.90625,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.503252,
        "reward_mean": -0.492293
      },
      {
        "step": 223,
        "frac_adv_clamped": 0.7859,
        "resp_len_mean": 240.6875,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.510943,
        "reward_mean": -0.526618
      },
      {
        "step": 224,
        "frac_adv_clamped": 0.838196,
        "resp_len_mean": 247.40625,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.529387,
        "reward_mean": -0.525759
      },
      {
        "step": 225,
        "frac_adv_clamped": 0.806922,
        "resp_len_mean": 241.96875,
        "frac_finished": 0.25,
        "kl_full_tok": 0.519512,
        "reward_mean": -0.490542
      },
      {
        "step": 226,
        "frac_adv_clamped": 0.749676,
        "resp_len_mean": 241.3125,
        "frac_finished": 0.25,
        "kl_full_tok": 0.475565,
        "reward_mean": -0.48718
      },
      {
        "step": 227,
        "frac_adv_clamped": 0.748366,
        "resp_len_mean": 243.90625,
        "frac_finished": 0.125,
        "kl_full_tok": 0.441508,
        "reward_mean": -0.432599
      },
      {
        "step": 228,
        "frac_adv_clamped": 0.763165,
        "resp_len_mean": 239.75,
        "frac_finished": 0.34375,
        "kl_full_tok": 0.516166,
        "reward_mean": -0.520601
      },
      {
        "step": 229,
        "frac_adv_clamped": 0.69709,
        "resp_len_mean": 228.71875,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.50518,
        "reward_mean": -0.479226
      },
      {
        "step": 230,
        "frac_adv_clamped": 0.761773,
        "resp_len_mean": 237.5625,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.513474,
        "reward_mean": -0.493383
      },
      {
        "step": 231,
        "frac_adv_clamped": 0.746179,
        "resp_len_mean": 239.21875,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.490728,
        "reward_mean": -0.495949
      },
      {
        "step": 232,
        "frac_adv_clamped": 0.773841,
        "resp_len_mean": 223.15625,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.475471,
        "reward_mean": -0.462384
      },
      {
        "step": 233,
        "frac_adv_clamped": 0.710724,
        "resp_len_mean": 236.90625,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.481838,
        "reward_mean": -0.476597
      },
      {
        "step": 234,
        "frac_adv_clamped": 0.842231,
        "resp_len_mean": 248.78125,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.519103,
        "reward_mean": -0.544019
      },
      {
        "step": 235,
        "frac_adv_clamped": 0.768302,
        "resp_len_mean": 248.4375,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.460339,
        "reward_mean": -0.454813
      },
      {
        "step": 236,
        "frac_adv_clamped": 0.779844,
        "resp_len_mean": 240.3125,
        "frac_finished": 0.3125,
        "kl_full_tok": 0.476386,
        "reward_mean": -0.45423
      },
      {
        "step": 237,
        "frac_adv_clamped": 0.763082,
        "resp_len_mean": 250.21875,
        "frac_finished": 0.09375,
        "kl_full_tok": 0.471403,
        "reward_mean": -0.467908
      },
      {
        "step": 238,
        "frac_adv_clamped": 0.787726,
        "resp_len_mean": 247.46875,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.495338,
        "reward_mean": -0.508705
      },
      {
        "step": 239,
        "frac_adv_clamped": 0.804295,
        "resp_len_mean": 244.46875,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.494905,
        "reward_mean": -0.490099
      },
      {
        "step": 240,
        "frac_adv_clamped": 0.745985,
        "resp_len_mean": 245.1875,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.470308,
        "reward_mean": -0.47899
      },
      {
        "step": 241,
        "frac_adv_clamped": 0.720254,
        "resp_len_mean": 250.5625,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.458787,
        "reward_mean": -0.434248
      },
      {
        "step": 242,
        "frac_adv_clamped": 0.77926,
        "resp_len_mean": 245.90625,
        "frac_finished": 0.25,
        "kl_full_tok": 0.474237,
        "reward_mean": -0.509874
      },
      {
        "step": 243,
        "frac_adv_clamped": 0.760656,
        "resp_len_mean": 247.8125,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.468127,
        "reward_mean": -0.456956
      },
      {
        "step": 244,
        "frac_adv_clamped": 0.776552,
        "resp_len_mean": 249.21875,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.509783,
        "reward_mean": -0.500526
      },
      {
        "step": 245,
        "frac_adv_clamped": 0.773674,
        "resp_len_mean": 246.1875,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.485377,
        "reward_mean": -0.496472
      },
      {
        "step": 246,
        "frac_adv_clamped": 0.814676,
        "resp_len_mean": 233.375,
        "frac_finished": 0.40625,
        "kl_full_tok": 0.507648,
        "reward_mean": -0.49482
      },
      {
        "step": 247,
        "frac_adv_clamped": 0.74513,
        "resp_len_mean": 247.0625,
        "frac_finished": 0.25,
        "kl_full_tok": 0.472258,
        "reward_mean": -0.488511
      },
      {
        "step": 248,
        "frac_adv_clamped": 0.744204,
        "resp_len_mean": 245.3125,
        "frac_finished": 0.25,
        "kl_full_tok": 0.500744,
        "reward_mean": -0.513976
      },
      {
        "step": 249,
        "frac_adv_clamped": 0.792397,
        "resp_len_mean": 253.1875,
        "frac_finished": 0.125,
        "kl_full_tok": 0.502723,
        "reward_mean": -0.502803
      },
      {
        "step": 250,
        "frac_adv_clamped": 0.765756,
        "resp_len_mean": 238.0,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.491514,
        "reward_mean": -0.468527
      },
      {
        "step": 251,
        "frac_adv_clamped": 0.739952,
        "resp_len_mean": 249.59375,
        "frac_finished": 0.125,
        "kl_full_tok": 0.494476,
        "reward_mean": -0.49941
      },
      {
        "step": 252,
        "frac_adv_clamped": 0.785872,
        "resp_len_mean": 240.21875,
        "frac_finished": 0.4375,
        "kl_full_tok": 0.560714,
        "reward_mean": -0.571271
      },
      {
        "step": 253,
        "frac_adv_clamped": 0.773396,
        "resp_len_mean": 252.78125,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.485656,
        "reward_mean": -0.473595
      },
      {
        "step": 254,
        "frac_adv_clamped": 0.786481,
        "resp_len_mean": 247.34375,
        "frac_finished": 0.125,
        "kl_full_tok": 0.514227,
        "reward_mean": -0.521593
      },
      {
        "step": 255,
        "frac_adv_clamped": 0.813011,
        "resp_len_mean": 248.34375,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.491635,
        "reward_mean": -0.511012
      },
      {
        "step": 256,
        "frac_adv_clamped": 0.810705,
        "resp_len_mean": 248.125,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.52768,
        "reward_mean": -0.55418
      },
      {
        "step": 257,
        "frac_adv_clamped": 0.80503,
        "resp_len_mean": 241.0625,
        "frac_finished": 0.25,
        "kl_full_tok": 0.514293,
        "reward_mean": -0.514061
      },
      {
        "step": 258,
        "frac_adv_clamped": 0.813327,
        "resp_len_mean": 250.4375,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.527118,
        "reward_mean": -0.507122
      },
      {
        "step": 259,
        "frac_adv_clamped": 0.744946,
        "resp_len_mean": 238.0625,
        "frac_finished": 0.46875,
        "kl_full_tok": 0.497524,
        "reward_mean": -0.495613
      },
      {
        "step": 260,
        "frac_adv_clamped": 0.782534,
        "resp_len_mean": 237.25,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.492748,
        "reward_mean": -0.491762
      },
      {
        "step": 261,
        "frac_adv_clamped": 0.775624,
        "resp_len_mean": 251.53125,
        "frac_finished": 0.125,
        "kl_full_tok": 0.49623,
        "reward_mean": -0.500777
      },
      {
        "step": 262,
        "frac_adv_clamped": 0.759891,
        "resp_len_mean": 241.6875,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.475929,
        "reward_mean": -0.479306
      },
      {
        "step": 263,
        "frac_adv_clamped": 0.782935,
        "resp_len_mean": 245.75,
        "frac_finished": 0.25,
        "kl_full_tok": 0.521468,
        "reward_mean": -0.502055
      },
      {
        "step": 264,
        "frac_adv_clamped": 0.720532,
        "resp_len_mean": 246.5625,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.450354,
        "reward_mean": -0.444834
      },
      {
        "step": 265,
        "frac_adv_clamped": 0.795077,
        "resp_len_mean": 250.09375,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.526383,
        "reward_mean": -0.529707
      },
      {
        "step": 266,
        "frac_adv_clamped": 0.769613,
        "resp_len_mean": 244.96875,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.454135,
        "reward_mean": -0.451896
      },
      {
        "step": 267,
        "frac_adv_clamped": 0.828837,
        "resp_len_mean": 252.5,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.524153,
        "reward_mean": -0.534702
      },
      {
        "step": 268,
        "frac_adv_clamped": 0.720152,
        "resp_len_mean": 246.5625,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.46692,
        "reward_mean": -0.45464
      },
      {
        "step": 269,
        "frac_adv_clamped": 0.748859,
        "resp_len_mean": 246.5,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.477247,
        "reward_mean": -0.479292
      },
      {
        "step": 270,
        "frac_adv_clamped": 0.795796,
        "resp_len_mean": 249.75,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.501892,
        "reward_mean": -0.496808
      },
      {
        "step": 271,
        "frac_adv_clamped": 0.765083,
        "resp_len_mean": 253.28125,
        "frac_finished": 0.0625,
        "kl_full_tok": 0.507865,
        "reward_mean": -0.508925
      },
      {
        "step": 272,
        "frac_adv_clamped": 0.79815,
        "resp_len_mean": 246.625,
        "frac_finished": 0.25,
        "kl_full_tok": 0.532093,
        "reward_mean": -0.532258
      },
      {
        "step": 273,
        "frac_adv_clamped": 0.807187,
        "resp_len_mean": 249.59375,
        "frac_finished": 0.09375,
        "kl_full_tok": 0.537827,
        "reward_mean": -0.5468
      },
      {
        "step": 274,
        "frac_adv_clamped": 0.779727,
        "resp_len_mean": 242.3125,
        "frac_finished": 0.25,
        "kl_full_tok": 0.50754,
        "reward_mean": -0.519346
      },
      {
        "step": 275,
        "frac_adv_clamped": 0.785937,
        "resp_len_mean": 251.09375,
        "frac_finished": 0.125,
        "kl_full_tok": 0.464944,
        "reward_mean": -0.440415
      },
      {
        "step": 276,
        "frac_adv_clamped": 0.803547,
        "resp_len_mean": 253.71875,
        "frac_finished": 0.0625,
        "kl_full_tok": 0.517425,
        "reward_mean": -0.515505
      },
      {
        "step": 277,
        "frac_adv_clamped": 0.820025,
        "resp_len_mean": 248.125,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.51199,
        "reward_mean": -0.502133
      },
      {
        "step": 278,
        "frac_adv_clamped": 0.736789,
        "resp_len_mean": 247.1875,
        "frac_finished": 0.25,
        "kl_full_tok": 0.484275,
        "reward_mean": -0.472404
      },
      {
        "step": 279,
        "frac_adv_clamped": 0.798972,
        "resp_len_mean": 255.25,
        "frac_finished": 0.03125,
        "kl_full_tok": 0.50688,
        "reward_mean": -0.510213
      },
      {
        "step": 280,
        "frac_adv_clamped": 0.763712,
        "resp_len_mean": 247.84375,
        "frac_finished": 0.25,
        "kl_full_tok": 0.511705,
        "reward_mean": -0.502187
      },
      {
        "step": 281,
        "frac_adv_clamped": 0.765136,
        "resp_len_mean": 243.625,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.528801,
        "reward_mean": -0.506981
      },
      {
        "step": 282,
        "frac_adv_clamped": 0.804579,
        "resp_len_mean": 244.34375,
        "frac_finished": 0.25,
        "kl_full_tok": 0.563187,
        "reward_mean": -0.552101
      },
      {
        "step": 283,
        "frac_adv_clamped": 0.76168,
        "resp_len_mean": 245.46875,
        "frac_finished": 0.25,
        "kl_full_tok": 0.530914,
        "reward_mean": -0.52408
      },
      {
        "step": 284,
        "frac_adv_clamped": 0.738151,
        "resp_len_mean": 240.0,
        "frac_finished": 0.25,
        "kl_full_tok": 0.488947,
        "reward_mean": -0.482363
      },
      {
        "step": 285,
        "frac_adv_clamped": 0.80055,
        "resp_len_mean": 250.21875,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.483582,
        "reward_mean": -0.488225
      },
      {
        "step": 286,
        "frac_adv_clamped": 0.703951,
        "resp_len_mean": 248.375,
        "frac_finished": 0.125,
        "kl_full_tok": 0.457053,
        "reward_mean": -0.452216
      },
      {
        "step": 287,
        "frac_adv_clamped": 0.80318,
        "resp_len_mean": 255.46875,
        "frac_finished": 0.03125,
        "kl_full_tok": 0.501373,
        "reward_mean": -0.502101
      },
      {
        "step": 288,
        "frac_adv_clamped": 0.767542,
        "resp_len_mean": 246.28125,
        "frac_finished": 0.25,
        "kl_full_tok": 0.491297,
        "reward_mean": -0.466298
      },
      {
        "step": 289,
        "frac_adv_clamped": 0.779388,
        "resp_len_mean": 250.15625,
        "frac_finished": 0.125,
        "kl_full_tok": 0.516155,
        "reward_mean": -0.494303
      },
      {
        "step": 290,
        "frac_adv_clamped": 0.802511,
        "resp_len_mean": 248.90625,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.514416,
        "reward_mean": -0.529389
      },
      {
        "step": 291,
        "frac_adv_clamped": 0.796936,
        "resp_len_mean": 255.0,
        "frac_finished": 0.03125,
        "kl_full_tok": 0.51233,
        "reward_mean": -0.508694
      },
      {
        "step": 292,
        "frac_adv_clamped": 0.813773,
        "resp_len_mean": 250.03125,
        "frac_finished": 0.125,
        "kl_full_tok": 0.518493,
        "reward_mean": -0.534566
      },
      {
        "step": 293,
        "frac_adv_clamped": 0.736743,
        "resp_len_mean": 249.28125,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.472468,
        "reward_mean": -0.461975
      },
      {
        "step": 294,
        "frac_adv_clamped": 0.789467,
        "resp_len_mean": 245.65625,
        "frac_finished": 0.21875,
        "kl_full_tok": 0.524021,
        "reward_mean": -0.533626
      },
      {
        "step": 295,
        "frac_adv_clamped": 0.801907,
        "resp_len_mean": 239.3125,
        "frac_finished": 0.375,
        "kl_full_tok": 0.523628,
        "reward_mean": -0.52883
      },
      {
        "step": 296,
        "frac_adv_clamped": 0.765952,
        "resp_len_mean": 252.21875,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.4752,
        "reward_mean": -0.473668
      },
      {
        "step": 297,
        "frac_adv_clamped": 0.797504,
        "resp_len_mean": 245.375,
        "frac_finished": 0.1875,
        "kl_full_tok": 0.519965,
        "reward_mean": -0.5148
      },
      {
        "step": 298,
        "frac_adv_clamped": 0.730018,
        "resp_len_mean": 243.1875,
        "frac_finished": 0.28125,
        "kl_full_tok": 0.453752,
        "reward_mean": -0.461101
      },
      {
        "step": 299,
        "frac_adv_clamped": 0.79965,
        "resp_len_mean": 250.34375,
        "frac_finished": 0.15625,
        "kl_full_tok": 0.515592,
        "reward_mean": -0.515706
      }
    ],
    "warm_start": [
      {
        "step": 0,
        "frac_adv_clamped": 0.012574,
        "resp_len_mean": 84.5,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.585491,
        "reward_mean": -0.586111
      },
      {
        "step": 1,
        "frac_adv_clamped": 0.014227,
        "resp_len_mean": 83.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.720213,
        "reward_mean": -0.719192
      },
      {
        "step": 2,
        "frac_adv_clamped": 0.017065,
        "resp_len_mean": 82.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.653975,
        "reward_mean": -0.638003
      },
      {
        "step": 3,
        "frac_adv_clamped": 0.02054,
        "resp_len_mean": 77.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.715095,
        "reward_mean": -0.730572
      },
      {
        "step": 4,
        "frac_adv_clamped": 0.014655,
        "resp_len_mean": 89.5625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.677511,
        "reward_mean": -0.664503
      },
      {
        "step": 5,
        "frac_adv_clamped": 0.017516,
        "resp_len_mean": 98.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.630204,
        "reward_mean": -0.628684
      },
      {
        "step": 6,
        "frac_adv_clamped": 0.016834,
        "resp_len_mean": 77.96875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.637048,
        "reward_mean": -0.620884
      },
      {
        "step": 7,
        "frac_adv_clamped": 0.017892,
        "resp_len_mean": 64.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.706367,
        "reward_mean": -0.670906
      },
      {
        "step": 8,
        "frac_adv_clamped": 0.026655,
        "resp_len_mean": 70.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.805336,
        "reward_mean": -0.801996
      },
      {
        "step": 9,
        "frac_adv_clamped": 0.015439,
        "resp_len_mean": 78.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.640471,
        "reward_mean": -0.651886
      },
      {
        "step": 10,
        "frac_adv_clamped": 0.02215,
        "resp_len_mean": 57.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.79385,
        "reward_mean": -0.797699
      },
      {
        "step": 11,
        "frac_adv_clamped": 0.019947,
        "resp_len_mean": 70.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.760226,
        "reward_mean": -0.735752
      },
      {
        "step": 12,
        "frac_adv_clamped": 0.019807,
        "resp_len_mean": 61.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.64659,
        "reward_mean": -0.620604
      },
      {
        "step": 13,
        "frac_adv_clamped": 0.01567,
        "resp_len_mean": 75.78125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.541622,
        "reward_mean": -0.528942
      },
      {
        "step": 14,
        "frac_adv_clamped": 0.014575,
        "resp_len_mean": 98.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.588997,
        "reward_mean": -0.590312
      },
      {
        "step": 15,
        "frac_adv_clamped": 0.01532,
        "resp_len_mean": 89.75,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.639467,
        "reward_mean": -0.607713
      },
      {
        "step": 16,
        "frac_adv_clamped": 0.01727,
        "resp_len_mean": 74.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.60527,
        "reward_mean": -0.600552
      },
      {
        "step": 17,
        "frac_adv_clamped": 0.015521,
        "resp_len_mean": 84.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.655393,
        "reward_mean": -0.647162
      },
      {
        "step": 18,
        "frac_adv_clamped": 0.020273,
        "resp_len_mean": 75.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.707575,
        "reward_mean": -0.708124
      },
      {
        "step": 19,
        "frac_adv_clamped": 0.018065,
        "resp_len_mean": 77.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.672844,
        "reward_mean": -0.677527
      },
      {
        "step": 20,
        "frac_adv_clamped": 0.016963,
        "resp_len_mean": 79.21875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.675684,
        "reward_mean": -0.657433
      },
      {
        "step": 21,
        "frac_adv_clamped": 0.018769,
        "resp_len_mean": 94.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.610059,
        "reward_mean": -0.628507
      },
      {
        "step": 22,
        "frac_adv_clamped": 0.015854,
        "resp_len_mean": 104.46875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.598739,
        "reward_mean": -0.619348
      },
      {
        "step": 23,
        "frac_adv_clamped": 0.01892,
        "resp_len_mean": 56.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.628942,
        "reward_mean": -0.607847
      },
      {
        "step": 24,
        "frac_adv_clamped": 0.01698,
        "resp_len_mean": 68.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.608712,
        "reward_mean": -0.584247
      },
      {
        "step": 25,
        "frac_adv_clamped": 0.015312,
        "resp_len_mean": 100.0,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.528452,
        "reward_mean": -0.541628
      },
      {
        "step": 26,
        "frac_adv_clamped": 0.014808,
        "resp_len_mean": 99.1875,
        "frac_finished": 0.875,
        "kl_full_tok": 0.598058,
        "reward_mean": -0.596731
      },
      {
        "step": 27,
        "frac_adv_clamped": 0.015396,
        "resp_len_mean": 83.21875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.685612,
        "reward_mean": -0.659678
      },
      {
        "step": 28,
        "frac_adv_clamped": 0.019699,
        "resp_len_mean": 80.90625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.677081,
        "reward_mean": -0.649119
      },
      {
        "step": 29,
        "frac_adv_clamped": 0.019166,
        "resp_len_mean": 83.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.605151,
        "reward_mean": -0.630477
      },
      {
        "step": 30,
        "frac_adv_clamped": 0.018836,
        "resp_len_mean": 74.65625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.74121,
        "reward_mean": -0.721095
      },
      {
        "step": 31,
        "frac_adv_clamped": 0.015798,
        "resp_len_mean": 79.125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.547644,
        "reward_mean": -0.568303
      },
      {
        "step": 32,
        "frac_adv_clamped": 0.017577,
        "resp_len_mean": 83.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.537861,
        "reward_mean": -0.570435
      },
      {
        "step": 33,
        "frac_adv_clamped": 0.017062,
        "resp_len_mean": 84.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.639822,
        "reward_mean": -0.639407
      },
      {
        "step": 34,
        "frac_adv_clamped": 0.014036,
        "resp_len_mean": 82.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.488911,
        "reward_mean": -0.492119
      },
      {
        "step": 35,
        "frac_adv_clamped": 0.015254,
        "resp_len_mean": 88.09375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.557686,
        "reward_mean": -0.568635
      },
      {
        "step": 36,
        "frac_adv_clamped": 0.019144,
        "resp_len_mean": 83.25,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.576195,
        "reward_mean": -0.59637
      },
      {
        "step": 37,
        "frac_adv_clamped": 0.012821,
        "resp_len_mean": 97.5,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.574852,
        "reward_mean": -0.565251
      },
      {
        "step": 38,
        "frac_adv_clamped": 0.020375,
        "resp_len_mean": 76.6875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.599744,
        "reward_mean": -0.598547
      },
      {
        "step": 39,
        "frac_adv_clamped": 0.017762,
        "resp_len_mean": 87.96875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.690104,
        "reward_mean": -0.712727
      },
      {
        "step": 40,
        "frac_adv_clamped": 0.013986,
        "resp_len_mean": 120.65625,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.529504,
        "reward_mean": -0.551161
      },
      {
        "step": 41,
        "frac_adv_clamped": 0.019747,
        "resp_len_mean": 79.125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.807027,
        "reward_mean": -0.810812
      },
      {
        "step": 42,
        "frac_adv_clamped": 0.015707,
        "resp_len_mean": 83.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.554455,
        "reward_mean": -0.560274
      },
      {
        "step": 43,
        "frac_adv_clamped": 0.019145,
        "resp_len_mean": 76.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.55817,
        "reward_mean": -0.580809
      },
      {
        "step": 44,
        "frac_adv_clamped": 0.017961,
        "resp_len_mean": 59.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.664229,
        "reward_mean": -0.642789
      },
      {
        "step": 45,
        "frac_adv_clamped": 0.013846,
        "resp_len_mean": 81.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.485775,
        "reward_mean": -0.504953
      },
      {
        "step": 46,
        "frac_adv_clamped": 0.01368,
        "resp_len_mean": 91.375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.523905,
        "reward_mean": -0.513177
      },
      {
        "step": 47,
        "frac_adv_clamped": 0.014065,
        "resp_len_mean": 91.09375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.550647,
        "reward_mean": -0.555465
      },
      {
        "step": 48,
        "frac_adv_clamped": 0.018619,
        "resp_len_mean": 68.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.60737,
        "reward_mean": -0.595277
      },
      {
        "step": 49,
        "frac_adv_clamped": 0.016928,
        "resp_len_mean": 73.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.558372,
        "reward_mean": -0.553457
      },
      {
        "step": 50,
        "frac_adv_clamped": 0.013121,
        "resp_len_mean": 88.125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.543506,
        "reward_mean": -0.533347
      },
      {
        "step": 51,
        "frac_adv_clamped": 0.019144,
        "resp_len_mean": 62.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.62695,
        "reward_mean": -0.637892
      },
      {
        "step": 52,
        "frac_adv_clamped": 0.012944,
        "resp_len_mean": 94.15625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.544047,
        "reward_mean": -0.533226
      },
      {
        "step": 53,
        "frac_adv_clamped": 0.012406,
        "resp_len_mean": 83.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.567302,
        "reward_mean": -0.55391
      },
      {
        "step": 54,
        "frac_adv_clamped": 0.014433,
        "resp_len_mean": 75.78125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.577154,
        "reward_mean": -0.580621
      },
      {
        "step": 55,
        "frac_adv_clamped": 0.010315,
        "resp_len_mean": 106.03125,
        "frac_finished": 0.8125,
        "kl_full_tok": 0.488492,
        "reward_mean": -0.482027
      },
      {
        "step": 56,
        "frac_adv_clamped": 0.018223,
        "resp_len_mean": 82.3125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.603907,
        "reward_mean": -0.594873
      },
      {
        "step": 57,
        "frac_adv_clamped": 0.016906,
        "resp_len_mean": 73.9375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.603981,
        "reward_mean": -0.584459
      },
      {
        "step": 58,
        "frac_adv_clamped": 0.019299,
        "resp_len_mean": 61.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.695534,
        "reward_mean": -0.728318
      },
      {
        "step": 59,
        "frac_adv_clamped": 0.015,
        "resp_len_mean": 93.75,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.493899,
        "reward_mean": -0.532542
      },
      {
        "step": 60,
        "frac_adv_clamped": 0.017591,
        "resp_len_mean": 81.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.604796,
        "reward_mean": -0.612816
      },
      {
        "step": 61,
        "frac_adv_clamped": 0.021626,
        "resp_len_mean": 72.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.693528,
        "reward_mean": -0.703968
      },
      {
        "step": 62,
        "frac_adv_clamped": 0.018777,
        "resp_len_mean": 64.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.614929,
        "reward_mean": -0.598273
      },
      {
        "step": 63,
        "frac_adv_clamped": 0.014633,
        "resp_len_mean": 100.375,
        "frac_finished": 0.875,
        "kl_full_tok": 0.56793,
        "reward_mean": -0.570328
      },
      {
        "step": 64,
        "frac_adv_clamped": 0.015397,
        "resp_len_mean": 75.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.625063,
        "reward_mean": -0.616029
      },
      {
        "step": 65,
        "frac_adv_clamped": 0.014187,
        "resp_len_mean": 99.125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.594039,
        "reward_mean": -0.607133
      },
      {
        "step": 66,
        "frac_adv_clamped": 0.017873,
        "resp_len_mean": 69.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.546127,
        "reward_mean": -0.52917
      },
      {
        "step": 67,
        "frac_adv_clamped": 0.015742,
        "resp_len_mean": 83.375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.599313,
        "reward_mean": -0.626308
      },
      {
        "step": 68,
        "frac_adv_clamped": 0.013243,
        "resp_len_mean": 92.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.460822,
        "reward_mean": -0.431949
      },
      {
        "step": 69,
        "frac_adv_clamped": 0.017734,
        "resp_len_mean": 72.25,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.550908,
        "reward_mean": -0.557829
      },
      {
        "step": 70,
        "frac_adv_clamped": 0.016713,
        "resp_len_mean": 78.53125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.661541,
        "reward_mean": -0.669065
      },
      {
        "step": 71,
        "frac_adv_clamped": 0.019635,
        "resp_len_mean": 66.84375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.726174,
        "reward_mean": -0.762157
      },
      {
        "step": 72,
        "frac_adv_clamped": 0.013804,
        "resp_len_mean": 97.34375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.460573,
        "reward_mean": -0.440713
      },
      {
        "step": 73,
        "frac_adv_clamped": 0.014645,
        "resp_len_mean": 83.21875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.54474,
        "reward_mean": -0.542128
      },
      {
        "step": 74,
        "frac_adv_clamped": 0.014291,
        "resp_len_mean": 85.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.519312,
        "reward_mean": -0.515221
      },
      {
        "step": 75,
        "frac_adv_clamped": 0.017536,
        "resp_len_mean": 65.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.624911,
        "reward_mean": -0.630288
      },
      {
        "step": 76,
        "frac_adv_clamped": 0.018215,
        "resp_len_mean": 68.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.578102,
        "reward_mean": -0.573944
      },
      {
        "step": 77,
        "frac_adv_clamped": 0.018207,
        "resp_len_mean": 66.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.615447,
        "reward_mean": -0.603487
      },
      {
        "step": 78,
        "frac_adv_clamped": 0.019239,
        "resp_len_mean": 69.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.772011,
        "reward_mean": -0.825068
      },
      {
        "step": 79,
        "frac_adv_clamped": 0.01372,
        "resp_len_mean": 97.9375,
        "frac_finished": 0.875,
        "kl_full_tok": 0.535312,
        "reward_mean": -0.546409
      },
      {
        "step": 80,
        "frac_adv_clamped": 0.016675,
        "resp_len_mean": 63.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.61819,
        "reward_mean": -0.578746
      },
      {
        "step": 81,
        "frac_adv_clamped": 0.01676,
        "resp_len_mean": 72.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.538589,
        "reward_mean": -0.572347
      },
      {
        "step": 82,
        "frac_adv_clamped": 0.014049,
        "resp_len_mean": 100.09375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.493548,
        "reward_mean": -0.502043
      },
      {
        "step": 83,
        "frac_adv_clamped": 0.017633,
        "resp_len_mean": 92.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.620603,
        "reward_mean": -0.640254
      },
      {
        "step": 84,
        "frac_adv_clamped": 0.015113,
        "resp_len_mean": 86.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.664325,
        "reward_mean": -0.659922
      },
      {
        "step": 85,
        "frac_adv_clamped": 0.016241,
        "resp_len_mean": 80.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.553468,
        "reward_mean": -0.577832
      },
      {
        "step": 86,
        "frac_adv_clamped": 0.023451,
        "resp_len_mean": 55.96875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.721044,
        "reward_mean": -0.743166
      },
      {
        "step": 87,
        "frac_adv_clamped": 0.015707,
        "resp_len_mean": 77.59375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.545526,
        "reward_mean": -0.544831
      },
      {
        "step": 88,
        "frac_adv_clamped": 0.014314,
        "resp_len_mean": 93.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.487918,
        "reward_mean": -0.4839
      },
      {
        "step": 89,
        "frac_adv_clamped": 0.01326,
        "resp_len_mean": 113.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.549778,
        "reward_mean": -0.530891
      },
      {
        "step": 90,
        "frac_adv_clamped": 0.01914,
        "resp_len_mean": 86.53125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.633415,
        "reward_mean": -0.647794
      },
      {
        "step": 91,
        "frac_adv_clamped": 0.013884,
        "resp_len_mean": 83.28125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.600508,
        "reward_mean": -0.585929
      },
      {
        "step": 92,
        "frac_adv_clamped": 0.013745,
        "resp_len_mean": 102.3125,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.499962,
        "reward_mean": -0.523863
      },
      {
        "step": 93,
        "frac_adv_clamped": 0.012914,
        "resp_len_mean": 89.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.489244,
        "reward_mean": -0.476621
      },
      {
        "step": 94,
        "frac_adv_clamped": 0.012436,
        "resp_len_mean": 98.0,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.507155,
        "reward_mean": -0.49431
      },
      {
        "step": 95,
        "frac_adv_clamped": 0.016257,
        "resp_len_mean": 71.125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.648138,
        "reward_mean": -0.636112
      },
      {
        "step": 96,
        "frac_adv_clamped": 0.013837,
        "resp_len_mean": 83.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.49865,
        "reward_mean": -0.500348
      },
      {
        "step": 97,
        "frac_adv_clamped": 0.020225,
        "resp_len_mean": 55.625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.765871,
        "reward_mean": -0.779931
      },
      {
        "step": 98,
        "frac_adv_clamped": 0.016673,
        "resp_len_mean": 80.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.572414,
        "reward_mean": -0.573538
      },
      {
        "step": 99,
        "frac_adv_clamped": 0.021891,
        "resp_len_mean": 62.8125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.690555,
        "reward_mean": -0.699681
      },
      {
        "step": 100,
        "frac_adv_clamped": 0.017998,
        "resp_len_mean": 71.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.569475,
        "reward_mean": -0.583997
      },
      {
        "step": 101,
        "frac_adv_clamped": 0.01287,
        "resp_len_mean": 97.125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.517716,
        "reward_mean": -0.519486
      },
      {
        "step": 102,
        "frac_adv_clamped": 0.015327,
        "resp_len_mean": 91.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.600847,
        "reward_mean": -0.607314
      },
      {
        "step": 103,
        "frac_adv_clamped": 0.01436,
        "resp_len_mean": 78.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.513334,
        "reward_mean": -0.478347
      },
      {
        "step": 104,
        "frac_adv_clamped": 0.014749,
        "resp_len_mean": 84.75,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.560086,
        "reward_mean": -0.574849
      },
      {
        "step": 105,
        "frac_adv_clamped": 0.010992,
        "resp_len_mean": 102.34375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.548085,
        "reward_mean": -0.580815
      },
      {
        "step": 106,
        "frac_adv_clamped": 0.020692,
        "resp_len_mean": 87.59375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.619818,
        "reward_mean": -0.591836
      },
      {
        "step": 107,
        "frac_adv_clamped": 0.014371,
        "resp_len_mean": 89.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.557493,
        "reward_mean": -0.534852
      },
      {
        "step": 108,
        "frac_adv_clamped": 0.015745,
        "resp_len_mean": 85.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.585858,
        "reward_mean": -0.587769
      },
      {
        "step": 109,
        "frac_adv_clamped": 0.015107,
        "resp_len_mean": 74.46875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.544223,
        "reward_mean": -0.527991
      },
      {
        "step": 110,
        "frac_adv_clamped": 0.015414,
        "resp_len_mean": 99.34375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.494849,
        "reward_mean": -0.499485
      },
      {
        "step": 111,
        "frac_adv_clamped": 0.016281,
        "resp_len_mean": 72.9375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.522263,
        "reward_mean": -0.542026
      },
      {
        "step": 112,
        "frac_adv_clamped": 0.015806,
        "resp_len_mean": 81.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.592368,
        "reward_mean": -0.592472
      },
      {
        "step": 113,
        "frac_adv_clamped": 0.015674,
        "resp_len_mean": 89.71875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.601301,
        "reward_mean": -0.581426
      },
      {
        "step": 114,
        "frac_adv_clamped": 0.018041,
        "resp_len_mean": 72.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.66292,
        "reward_mean": -0.657696
      },
      {
        "step": 115,
        "frac_adv_clamped": 0.016629,
        "resp_len_mean": 97.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.552539,
        "reward_mean": -0.530011
      },
      {
        "step": 116,
        "frac_adv_clamped": 0.014589,
        "resp_len_mean": 94.25,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.552068,
        "reward_mean": -0.540752
      },
      {
        "step": 117,
        "frac_adv_clamped": 0.015228,
        "resp_len_mean": 80.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.549268,
        "reward_mean": -0.55215
      },
      {
        "step": 118,
        "frac_adv_clamped": 0.014741,
        "resp_len_mean": 78.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.595337,
        "reward_mean": -0.58644
      },
      {
        "step": 119,
        "frac_adv_clamped": 0.018447,
        "resp_len_mean": 72.84375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.581888,
        "reward_mean": -0.582663
      },
      {
        "step": 120,
        "frac_adv_clamped": 0.013082,
        "resp_len_mean": 81.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.455528,
        "reward_mean": -0.468505
      },
      {
        "step": 121,
        "frac_adv_clamped": 0.021763,
        "resp_len_mean": 56.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.699995,
        "reward_mean": -0.716129
      },
      {
        "step": 122,
        "frac_adv_clamped": 0.014256,
        "resp_len_mean": 89.875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.618158,
        "reward_mean": -0.615924
      },
      {
        "step": 123,
        "frac_adv_clamped": 0.014802,
        "resp_len_mean": 109.78125,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.554766,
        "reward_mean": -0.561055
      },
      {
        "step": 124,
        "frac_adv_clamped": 0.011681,
        "resp_len_mean": 109.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.453662,
        "reward_mean": -0.438266
      },
      {
        "step": 125,
        "frac_adv_clamped": 0.01495,
        "resp_len_mean": 75.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.612732,
        "reward_mean": -0.609865
      },
      {
        "step": 126,
        "frac_adv_clamped": 0.018902,
        "resp_len_mean": 69.4375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.575626,
        "reward_mean": -0.571085
      },
      {
        "step": 127,
        "frac_adv_clamped": 0.015791,
        "resp_len_mean": 96.96875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.573337,
        "reward_mean": -0.570549
      },
      {
        "step": 128,
        "frac_adv_clamped": 0.013077,
        "resp_len_mean": 81.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.451289,
        "reward_mean": -0.440198
      },
      {
        "step": 129,
        "frac_adv_clamped": 0.018717,
        "resp_len_mean": 70.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.640061,
        "reward_mean": -0.668283
      },
      {
        "step": 130,
        "frac_adv_clamped": 0.017944,
        "resp_len_mean": 64.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.603654,
        "reward_mean": -0.596221
      },
      {
        "step": 131,
        "frac_adv_clamped": 0.017764,
        "resp_len_mean": 72.125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.603627,
        "reward_mean": -0.647615
      },
      {
        "step": 132,
        "frac_adv_clamped": 0.013898,
        "resp_len_mean": 96.6875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.522055,
        "reward_mean": -0.526562
      },
      {
        "step": 133,
        "frac_adv_clamped": 0.017467,
        "resp_len_mean": 71.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.60325,
        "reward_mean": -0.610309
      },
      {
        "step": 134,
        "frac_adv_clamped": 0.018859,
        "resp_len_mean": 66.28125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.607299,
        "reward_mean": -0.60301
      },
      {
        "step": 135,
        "frac_adv_clamped": 0.019089,
        "resp_len_mean": 72.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.673956,
        "reward_mean": -0.676133
      },
      {
        "step": 136,
        "frac_adv_clamped": 0.017512,
        "resp_len_mean": 67.8125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.642119,
        "reward_mean": -0.622096
      },
      {
        "step": 137,
        "frac_adv_clamped": 0.016646,
        "resp_len_mean": 75.09375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.635619,
        "reward_mean": -0.653873
      },
      {
        "step": 138,
        "frac_adv_clamped": 0.015458,
        "resp_len_mean": 78.84375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.546182,
        "reward_mean": -0.568054
      },
      {
        "step": 139,
        "frac_adv_clamped": 0.016474,
        "resp_len_mean": 70.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.629251,
        "reward_mean": -0.616064
      },
      {
        "step": 140,
        "frac_adv_clamped": 0.015121,
        "resp_len_mean": 76.46875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.552585,
        "reward_mean": -0.558941
      },
      {
        "step": 141,
        "frac_adv_clamped": 0.017975,
        "resp_len_mean": 52.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.659303,
        "reward_mean": -0.63659
      },
      {
        "step": 142,
        "frac_adv_clamped": 0.013324,
        "resp_len_mean": 93.8125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.522595,
        "reward_mean": -0.524398
      },
      {
        "step": 143,
        "frac_adv_clamped": 0.018695,
        "resp_len_mean": 78.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.633346,
        "reward_mean": -0.608788
      },
      {
        "step": 144,
        "frac_adv_clamped": 0.015721,
        "resp_len_mean": 71.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.51658,
        "reward_mean": -0.518254
      },
      {
        "step": 145,
        "frac_adv_clamped": 0.012202,
        "resp_len_mean": 112.6875,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.553827,
        "reward_mean": -0.535363
      },
      {
        "step": 146,
        "frac_adv_clamped": 0.01369,
        "resp_len_mean": 105.0,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.555863,
        "reward_mean": -0.57246
      },
      {
        "step": 147,
        "frac_adv_clamped": 0.01502,
        "resp_len_mean": 79.0625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.545332,
        "reward_mean": -0.514292
      },
      {
        "step": 148,
        "frac_adv_clamped": 0.012492,
        "resp_len_mean": 97.5625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.475921,
        "reward_mean": -0.477039
      },
      {
        "step": 149,
        "frac_adv_clamped": 0.014724,
        "resp_len_mean": 101.875,
        "frac_finished": 0.875,
        "kl_full_tok": 0.60042,
        "reward_mean": -0.573475
      },
      {
        "step": 150,
        "frac_adv_clamped": 0.015067,
        "resp_len_mean": 89.1875,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.558697,
        "reward_mean": -0.577083
      },
      {
        "step": 151,
        "frac_adv_clamped": 0.019726,
        "resp_len_mean": 72.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.60021,
        "reward_mean": -0.62335
      },
      {
        "step": 152,
        "frac_adv_clamped": 0.014438,
        "resp_len_mean": 90.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.505018,
        "reward_mean": -0.495559
      },
      {
        "step": 153,
        "frac_adv_clamped": 0.016532,
        "resp_len_mean": 73.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.600813,
        "reward_mean": -0.620602
      },
      {
        "step": 154,
        "frac_adv_clamped": 0.01597,
        "resp_len_mean": 82.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.566242,
        "reward_mean": -0.548318
      },
      {
        "step": 155,
        "frac_adv_clamped": 0.016364,
        "resp_len_mean": 87.84375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.591471,
        "reward_mean": -0.5879
      },
      {
        "step": 156,
        "frac_adv_clamped": 0.012667,
        "resp_len_mean": 91.28125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.528915,
        "reward_mean": -0.508313
      },
      {
        "step": 157,
        "frac_adv_clamped": 0.011862,
        "resp_len_mean": 105.375,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.518147,
        "reward_mean": -0.514836
      },
      {
        "step": 158,
        "frac_adv_clamped": 0.015216,
        "resp_len_mean": 73.9375,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.534727,
        "reward_mean": -0.529576
      },
      {
        "step": 159,
        "frac_adv_clamped": 0.015341,
        "resp_len_mean": 77.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.532163,
        "reward_mean": -0.523366
      },
      {
        "step": 160,
        "frac_adv_clamped": 0.017929,
        "resp_len_mean": 69.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.563601,
        "reward_mean": -0.530239
      },
      {
        "step": 161,
        "frac_adv_clamped": 0.015928,
        "resp_len_mean": 72.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.518699,
        "reward_mean": -0.53206
      },
      {
        "step": 162,
        "frac_adv_clamped": 0.017122,
        "resp_len_mean": 76.65625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.59884,
        "reward_mean": -0.600321
      },
      {
        "step": 163,
        "frac_adv_clamped": 0.014418,
        "resp_len_mean": 97.53125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.578035,
        "reward_mean": -0.551582
      },
      {
        "step": 164,
        "frac_adv_clamped": 0.014296,
        "resp_len_mean": 89.625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.576831,
        "reward_mean": -0.614844
      },
      {
        "step": 165,
        "frac_adv_clamped": 0.015645,
        "resp_len_mean": 71.90625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.565189,
        "reward_mean": -0.589121
      },
      {
        "step": 166,
        "frac_adv_clamped": 0.021097,
        "resp_len_mean": 59.25,
        "frac_finished": 1.0,
        "kl_full_tok": 0.699676,
        "reward_mean": -0.72239
      },
      {
        "step": 167,
        "frac_adv_clamped": 0.018945,
        "resp_len_mean": 61.03125,
        "frac_finished": 1.0,
        "kl_full_tok": 0.63652,
        "reward_mean": -0.61672
      },
      {
        "step": 168,
        "frac_adv_clamped": 0.018405,
        "resp_len_mean": 86.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.59059,
        "reward_mean": -0.60018
      },
      {
        "step": 169,
        "frac_adv_clamped": 0.017943,
        "resp_len_mean": 71.40625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.653413,
        "reward_mean": -0.652739
      },
      {
        "step": 170,
        "frac_adv_clamped": 0.015587,
        "resp_len_mean": 64.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.578702,
        "reward_mean": -0.589703
      },
      {
        "step": 171,
        "frac_adv_clamped": 0.01187,
        "resp_len_mean": 107.9375,
        "frac_finished": 0.90625,
        "kl_full_tok": 0.493498,
        "reward_mean": -0.491305
      },
      {
        "step": 172,
        "frac_adv_clamped": 0.010768,
        "resp_len_mean": 110.28125,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.514636,
        "reward_mean": -0.524024
      },
      {
        "step": 173,
        "frac_adv_clamped": 0.01845,
        "resp_len_mean": 84.6875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.614849,
        "reward_mean": -0.619648
      },
      {
        "step": 174,
        "frac_adv_clamped": 0.014675,
        "resp_len_mean": 89.4375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.571997,
        "reward_mean": -0.578636
      },
      {
        "step": 175,
        "frac_adv_clamped": 0.022719,
        "resp_len_mean": 81.15625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.738067,
        "reward_mean": -0.751249
      },
      {
        "step": 176,
        "frac_adv_clamped": 0.019645,
        "resp_len_mean": 65.21875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.670362,
        "reward_mean": -0.628018
      },
      {
        "step": 177,
        "frac_adv_clamped": 0.016428,
        "resp_len_mean": 74.1875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.563233,
        "reward_mean": -0.543704
      },
      {
        "step": 178,
        "frac_adv_clamped": 0.020812,
        "resp_len_mean": 60.0625,
        "frac_finished": 1.0,
        "kl_full_tok": 0.61712,
        "reward_mean": -0.631709
      },
      {
        "step": 179,
        "frac_adv_clamped": 0.016151,
        "resp_len_mean": 90.9375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.633,
        "reward_mean": -0.630379
      },
      {
        "step": 180,
        "frac_adv_clamped": 0.016122,
        "resp_len_mean": 75.59375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.583076,
        "reward_mean": -0.593704
      },
      {
        "step": 181,
        "frac_adv_clamped": 0.013919,
        "resp_len_mean": 87.5625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.516088,
        "reward_mean": -0.517112
      },
      {
        "step": 182,
        "frac_adv_clamped": 0.012341,
        "resp_len_mean": 91.15625,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.536941,
        "reward_mean": -0.541225
      },
      {
        "step": 183,
        "frac_adv_clamped": 0.014404,
        "resp_len_mean": 101.96875,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.531632,
        "reward_mean": -0.544675
      },
      {
        "step": 184,
        "frac_adv_clamped": 0.015721,
        "resp_len_mean": 81.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.606884,
        "reward_mean": -0.592318
      },
      {
        "step": 185,
        "frac_adv_clamped": 0.015502,
        "resp_len_mean": 110.875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.601941,
        "reward_mean": -0.628577
      },
      {
        "step": 186,
        "frac_adv_clamped": 0.02658,
        "resp_len_mean": 43.5,
        "frac_finished": 1.0,
        "kl_full_tok": 0.816371,
        "reward_mean": -0.806629
      },
      {
        "step": 187,
        "frac_adv_clamped": 0.015683,
        "resp_len_mean": 67.75,
        "frac_finished": 1.0,
        "kl_full_tok": 0.545651,
        "reward_mean": -0.559077
      },
      {
        "step": 188,
        "frac_adv_clamped": 0.013536,
        "resp_len_mean": 92.34375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.462033,
        "reward_mean": -0.45268
      },
      {
        "step": 189,
        "frac_adv_clamped": 0.01727,
        "resp_len_mean": 76.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.609926,
        "reward_mean": -0.610699
      },
      {
        "step": 190,
        "frac_adv_clamped": 0.018835,
        "resp_len_mean": 71.34375,
        "frac_finished": 1.0,
        "kl_full_tok": 0.673891,
        "reward_mean": -0.709495
      },
      {
        "step": 191,
        "frac_adv_clamped": 0.020262,
        "resp_len_mean": 78.65625,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.660112,
        "reward_mean": -0.686633
      },
      {
        "step": 192,
        "frac_adv_clamped": 0.014414,
        "resp_len_mean": 86.71875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.566377,
        "reward_mean": -0.536727
      },
      {
        "step": 193,
        "frac_adv_clamped": 0.01199,
        "resp_len_mean": 91.21875,
        "frac_finished": 0.875,
        "kl_full_tok": 0.525483,
        "reward_mean": -0.508699
      },
      {
        "step": 194,
        "frac_adv_clamped": 0.018789,
        "resp_len_mean": 74.84375,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.598872,
        "reward_mean": -0.631455
      },
      {
        "step": 195,
        "frac_adv_clamped": 0.019928,
        "resp_len_mean": 69.0,
        "frac_finished": 1.0,
        "kl_full_tok": 0.690691,
        "reward_mean": -0.678766
      },
      {
        "step": 196,
        "frac_adv_clamped": 0.016061,
        "resp_len_mean": 81.71875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.627569,
        "reward_mean": -0.625022
      },
      {
        "step": 197,
        "frac_adv_clamped": 0.018591,
        "resp_len_mean": 63.875,
        "frac_finished": 1.0,
        "kl_full_tok": 0.607193,
        "reward_mean": -0.61441
      },
      {
        "step": 198,
        "frac_adv_clamped": 0.018361,
        "resp_len_mean": 95.3125,
        "frac_finished": 0.9375,
        "kl_full_tok": 0.638989,
        "reward_mean": -0.645417
      },
      {
        "step": 199,
        "frac_adv_clamped": 0.02038,
        "resp_len_mean": 67.46875,
        "frac_finished": 0.96875,
        "kl_full_tok": 0.712558,
        "reward_mean": -0.7098
      }
    ]
  },
  "heatmap": {
    "prompt_id": "train:410",
    "sample_idx": 105,
    "prompt_text": "Mark wants to set the record for most consecutive ropes jumped.  The record is 54,000.  He can jump 3 times a second.  How many hours would he need to jump rope?",
    "text": "54,000 / 3 = 18,000 jumps.\n18,000 / 60 = 300 seconds.\n300 / 60 = 5 hours.\n#### 5\n\nQuestion:",
    "token_strs": [
      "5",
      "4",
      ",",
      "0",
      "0",
      "0",
      " /",
      " ",
      "3",
      " =",
      " ",
      "1",
      "8",
      ",",
      "0",
      "0",
      "0",
      " jumps",
      ".\n",
      "1",
      "8",
      ",",
      "0",
      "0",
      "0",
      " /",
      " ",
      "6",
      "0",
      " =",
      " ",
      "3",
      "0",
      "0",
      " seconds",
      ".\n",
      "3",
      "0",
      "0",
      " /",
      " ",
      "6",
      "0",
      " =",
      " ",
      "5",
      " hours",
      ".\n",
      "####",
      " ",
      "5",
      "\n\n",
      "Question",
      ":",
      "<|endoftext|>"
    ],
    "student_logp": [
      -0.034707,
      -1.4e-05,
      -0.001334,
      -7.5e-05,
      -0.0,
      0.0,
      -0.13405,
      -0.386872,
      -0.000127,
      -0.000409,
      -0.0,
      -1.2e-05,
      -1e-05,
      -5.2e-05,
      -4e-06,
      0.0,
      0.0,
      -0.021482,
      -0.106393,
      -0.20885,
      -7e-06,
      -4e-06,
      -3e-06,
      0.0,
      0.0,
      -0.429556,
      -0.00049,
      -0.024236,
      -1e-06,
      -0.001411,
      -1e-06,
      -0.000164,
      -1e-06,
      0.0,
      -0.047452,
      -0.032358,
      -0.001109,
      -0.0,
      0.0,
      -0.043157,
      -1.1e-05,
      -2e-06,
      -0.0,
      -2e-06,
      -1e-06,
      -7e-06,
      -0.201774,
      -0.000505,
      -5.2e-05,
      0.0,
      -5e-06,
      -9.1e-05,
      -5.9e-05,
      -3e-06,
      -25.127838
    ],
    "teacher_logp": [
      -0.070521,
      -1.2e-05,
      -0.201447,
      -1.2e-05,
      0.0,
      -7e-06,
      -0.092895,
      -0.002815,
      -0.000214,
      -2.2e-05,
      -3.7e-05,
      -1.5e-05,
      -5e-06,
      -8.5e-05,
      -0.0,
      0.0,
      0.0,
      -23.25,
      -6.13446,
      -0.499552,
      -0.000442,
      -0.000162,
      -0.0,
      0.0,
      0.0,
      -0.014943,
      -0.009758,
      -2.6e-05,
      -0.0,
      -0.012914,
      -7e-06,
      -0.0,
      -1e-06,
      -6e-06,
      -9.250271,
      -0.007452,
      -0.000239,
      -2e-06,
      0.0,
      -0.003182,
      -2e-06,
      -3.9e-05,
      -0.0,
      -3.5e-05,
      -1e-06,
      0.0,
      -0.000105,
      -0.001747,
      -0.002919,
      -1e-06,
      0.0,
      -0.281512,
      -0.000113,
      -0.000527,
      -24.805735
    ],
    "kl_rev": [
      0.032015,
      2.4e-05,
      0.193324,
      7.7e-05,
      6e-06,
      7e-06,
      0.260639,
      1.260151,
      0.00019,
      0.004027,
      3.7e-05,
      2e-05,
      1.7e-05,
      9e-06,
      1.6e-05,
      0.0,
      0.0,
      22.725309,
      5.298337,
      0.200302,
      0.000419,
      0.000148,
      4e-06,
      0.0,
      0.0,
      0.893258,
      0.007809,
      0.152627,
      1e-06,
      0.017873,
      9e-06,
      0.000978,
      2e-06,
      6e-06,
      8.698518,
      0.040522,
      0.004665,
      2e-06,
      -0.0,
      0.071422,
      2.5e-05,
      3.3e-05,
      0.0,
      2.9e-05,
      3e-06,
      4.9e-05,
      1.510126,
      0.000827,
      0.002764,
      1e-06,
      2.5e-05,
      0.281721,
      0.000311,
      0.000556,
      0.939636
    ],
    "gold": "5",
    "extracted": "5",
    "correct": true,
    "selection": "Largest single-position reverse KL among correct final-C probe responses of 30\u201360 tokens; illustrative, deliberately selected.",
    "reward": [
      -0.035814,
      1.9999999999999995e-06,
      -0.20011299999999999,
      6.299999999999999e-05,
      0.0,
      -7e-06,
      0.041155,
      0.384057,
      -8.7e-05,
      0.00038700000000000003,
      -3.7e-05,
      -3e-06,
      5e-06,
      -3.300000000000001e-05,
      4e-06,
      0.0,
      0.0,
      -23.228518,
      -6.028067,
      -0.290702,
      -0.000435,
      -0.00015800000000000002,
      3e-06,
      0.0,
      0.0,
      0.414613,
      -0.009267999999999998,
      0.02421,
      1e-06,
      -0.011503,
      -6e-06,
      0.000164,
      0.0,
      -6e-06,
      -9.202819,
      0.024905999999999998,
      0.0008699999999999999,
      -2e-06,
      0.0,
      0.039975000000000004,
      9e-06,
      -3.7e-05,
      0.0,
      -3.2999999999999996e-05,
      0.0,
      7e-06,
      0.20166900000000001,
      -0.001242,
      -0.002867,
      -1e-06,
      5e-06,
      -0.281421,
      -5.4e-05,
      -0.000524,
      0.322103000000002
    ]
  },
  "heldout": {
    "on_policy": {
      "step": 300,
      "responses": 256,
      "tokens": 19498,
      "reverse_kl_per_token": 0.3166057724382066
    },
    "on_policy_seed2": {
      "step": 300,
      "responses": 256,
      "tokens": 19197,
      "reverse_kl_per_token": 0.31948736031672625
    },
    "reward_to_go": {
      "step": 300,
      "responses": 256,
      "tokens": 63330,
      "reverse_kl_per_token": 0.48208037937786685
    },
    "warm_start": {
      "step": 200,
      "responses": 256,
      "tokens": 19971,
      "reverse_kl_per_token": 0.6017797949026198
    }
  },
  "corrected_capability": {
    "base": 0.523333,
    "teacher": 0.673333,
    "sft_all": 0.496667,
    "sft_correct": 0.523333,
    "on_policy": 0.5,
    "warm_start": 0.513333
  },
  "illustrative_kl": {
    "teacher": [
      0.7,
      0.2,
      0.1
    ],
    "student": [
      0.4,
      0.4,
      0.2
    ],
    "forward_kl": 0.1837868973868122,
    "reverse_kl": 0.19204199316179812,
    "forward_contributions": [
      0.39173105155479576,
      -0.13862943611198905,
      -0.06931471805599453
    ]
  }
}
