trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5263157894736842,
  "eval_steps": 50,
  "global_step": 250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 451.125,
      "completions/max_terminated_length": 451.125,
      "completions/mean_length": 241.560546875,
      "completions/mean_terminated_length": 241.560546875,
      "completions/min_length": 111.75,
      "completions/min_terminated_length": 111.75,
      "epoch": 0.002105263157894737,
      "grad_norm": 0.0383942686021328,
      "learning_rate": 0.0,
      "loss": -0.0025,
      "num_tokens": 484639.0,
      "reward": 0.9500823765993118,
      "reward_std": 0.6353622525930405,
      "rewards/format_reward_embodied/mean": 0.501953125,
      "rewards/format_reward_embodied/std": 0.4904305227100849,
      "rewards/stop_prediction_reward/mean": 0.439453125,
      "rewards/stop_prediction_reward/std": 0.3919360339641571,
      "rewards/waypoint_pred_accuracy/mean": 0.004338064874811504,
      "rewards/waypoint_pred_accuracy/std": 0.012508220294698669,
      "step": 1
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 985.25,
      "completions/max_terminated_length": 985.25,
      "completions/mean_length": 266.818359375,
      "completions/mean_terminated_length": 267.813928604126,
      "completions/min_length": 94.0,
      "completions/min_terminated_length": 119.375,
      "epoch": 0.004210526315789474,
      "grad_norm": 0.041216954588890076,
      "learning_rate": 2.083333333333333e-08,
      "loss": 0.0025,
      "num_tokens": 982274.0,
      "reward": 0.9463644102215767,
      "reward_std": 0.6123590245842934,
      "rewards/format_reward_embodied/mean": 0.50390625,
      "rewards/format_reward_embodied/std": 0.49139947816729546,
      "rewards/stop_prediction_reward/mean": 0.42578125,
      "rewards/stop_prediction_reward/std": 0.3781757093966007,
      "rewards/waypoint_pred_accuracy/mean": 0.008338454590998856,
      "rewards/waypoint_pred_accuracy/std": 0.02023129865005227,
      "step": 2
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 744.0,
      "completions/max_terminated_length": 744.0,
      "completions/mean_length": 247.1015625,
      "completions/mean_terminated_length": 247.1015625,
      "completions/min_length": 117.875,
      "completions/min_terminated_length": 117.875,
      "epoch": 0.00631578947368421,
      "grad_norm": 0.035625893622636795,
      "learning_rate": 4.166666666666666e-08,
      "loss": -0.001,
      "num_tokens": 1467318.0,
      "reward": 0.9623514339327812,
      "reward_std": 0.6189808771014214,
      "rewards/format_reward_embodied/mean": 0.498046875,
      "rewards/format_reward_embodied/std": 0.4842093959450722,
      "rewards/stop_prediction_reward/mean": 0.443359375,
      "rewards/stop_prediction_reward/std": 0.44101808220148087,
      "rewards/waypoint_pred_accuracy/mean": 0.010472597823421942,
      "rewards/waypoint_pred_accuracy/std": 0.03422736286900763,
      "step": 3
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 509.5,
      "completions/max_terminated_length": 509.5,
      "completions/mean_length": 237.654296875,
      "completions/mean_terminated_length": 237.654296875,
      "completions/min_length": 113.375,
      "completions/min_terminated_length": 113.375,
      "epoch": 0.008421052631578947,
      "grad_norm": 0.03751353174448013,
      "learning_rate": 6.25e-08,
      "loss": -0.0035,
      "num_tokens": 1950021.0,
      "reward": 0.9695519432425499,
      "reward_std": 0.6773256361484528,
      "rewards/format_reward_embodied/mean": 0.4765625,
      "rewards/format_reward_embodied/std": 0.4892418272793293,
      "rewards/stop_prediction_reward/mean": 0.421875,
      "rewards/stop_prediction_reward/std": 0.3986336216330528,
      "rewards/waypoint_pred_accuracy/mean": 0.03555722400778907,
      "rewards/waypoint_pred_accuracy/std": 0.08721220167353771,
      "step": 4
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 813.375,
      "completions/max_terminated_length": 813.375,
      "completions/mean_length": 265.58984375,
      "completions/mean_terminated_length": 265.58984375,
      "completions/min_length": 116.625,
      "completions/min_terminated_length": 116.625,
      "epoch": 0.010526315789473684,
      "grad_norm": 0.0453697107732296,
      "learning_rate": 8.333333333333333e-08,
      "loss": 0.0003,
      "num_tokens": 2446643.0,
      "reward": 0.9188483878970146,
      "reward_std": 0.610739640891552,
      "rewards/format_reward_embodied/mean": 0.5625,
      "rewards/format_reward_embodied/std": 0.4938563257455826,
      "rewards/stop_prediction_reward/mean": 0.345703125,
      "rewards/stop_prediction_reward/std": 0.4086693823337555,
      "rewards/waypoint_pred_accuracy/mean": 0.005322630658819445,
      "rewards/waypoint_pred_accuracy/std": 0.016403043182279513,
      "step": 5
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 812.0,
      "completions/max_terminated_length": 812.0,
      "completions/mean_length": 252.169921875,
      "completions/mean_terminated_length": 252.66043663024902,
      "completions/min_length": 102.375,
      "completions/min_terminated_length": 118.25,
      "epoch": 0.01263157894736842,
      "grad_norm": 0.049299150705337524,
      "learning_rate": 1.0416666666666667e-07,
      "loss": 0.0011,
      "num_tokens": 2935818.0,
      "reward": 0.877131775021553,
      "reward_std": 0.6145607680082321,
      "rewards/format_reward_embodied/mean": 0.509765625,
      "rewards/format_reward_embodied/std": 0.4843035563826561,
      "rewards/stop_prediction_reward/mean": 0.3671875,
      "rewards/stop_prediction_reward/std": 0.38137195259332657,
      "rewards/waypoint_pred_accuracy/mean": 8.932340400540724e-05,
      "rewards/waypoint_pred_accuracy/std": 0.0005101569792639827,
      "step": 6
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 1163.75,
      "completions/max_terminated_length": 1163.75,
      "completions/mean_length": 255.529296875,
      "completions/mean_terminated_length": 255.529296875,
      "completions/min_length": 105.0,
      "completions/min_terminated_length": 105.0,
      "epoch": 0.014736842105263158,
      "grad_norm": 0.036646511405706406,
      "learning_rate": 1.25e-07,
      "loss": 0.001,
      "num_tokens": 3427097.0,
      "reward": 0.6614178493618965,
      "reward_std": 0.6105708554387093,
      "rewards/format_reward_embodied/mean": 0.46484375,
      "rewards/format_reward_embodied/std": 0.48722705617547035,
      "rewards/stop_prediction_reward/mean": 0.173828125,
      "rewards/stop_prediction_reward/std": 0.3640986457467079,
      "rewards/waypoint_pred_accuracy/mean": 0.011372994726074323,
      "rewards/waypoint_pred_accuracy/std": 0.04579899070052374,
      "step": 7
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 467.25,
      "completions/max_terminated_length": 467.25,
      "completions/mean_length": 246.98828125,
      "completions/mean_terminated_length": 246.98828125,
      "completions/min_length": 119.5,
      "completions/min_terminated_length": 119.5,
      "epoch": 0.016842105263157894,
      "grad_norm": 0.047037359327077866,
      "learning_rate": 1.4583333333333335e-07,
      "loss": -0.0023,
      "num_tokens": 3915411.0,
      "reward": 0.9769175350666046,
      "reward_std": 0.6314118355512619,
      "rewards/format_reward_embodied/mean": 0.568359375,
      "rewards/format_reward_embodied/std": 0.4869700260460377,
      "rewards/stop_prediction_reward/mean": 0.376953125,
      "rewards/stop_prediction_reward/std": 0.4093479886651039,
      "rewards/waypoint_pred_accuracy/mean": 0.015802525533167768,
      "rewards/waypoint_pred_accuracy/std": 0.032816135895782333,
      "step": 8
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 481.125,
      "completions/max_terminated_length": 481.125,
      "completions/mean_length": 241.935546875,
      "completions/mean_terminated_length": 241.935546875,
      "completions/min_length": 114.5,
      "completions/min_terminated_length": 114.5,
      "epoch": 0.018947368421052633,
      "grad_norm": 0.04171831160783768,
      "learning_rate": 1.6666666666666665e-07,
      "loss": -0.0015,
      "num_tokens": 4397042.0,
      "reward": 0.9027341902256012,
      "reward_std": 0.6853612437844276,
      "rewards/format_reward_embodied/mean": 0.48828125,
      "rewards/format_reward_embodied/std": 0.4951612576842308,
      "rewards/stop_prediction_reward/mean": 0.322265625,
      "rewards/stop_prediction_reward/std": 0.4115743637084961,
      "rewards/waypoint_pred_accuracy/mean": 0.04609366483055356,
      "rewards/waypoint_pred_accuracy/std": 0.09995413944127579,
      "step": 9
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 695.625,
      "completions/max_terminated_length": 695.625,
      "completions/mean_length": 261.609375,
      "completions/mean_terminated_length": 261.609375,
      "completions/min_length": 110.0,
      "completions/min_terminated_length": 110.0,
      "epoch": 0.021052631578947368,
      "grad_norm": 0.03897935897111893,
      "learning_rate": 1.875e-07,
      "loss": 0.0002,
      "num_tokens": 4893354.0,
      "reward": 1.0423217862844467,
      "reward_std": 0.6282145008444786,
      "rewards/format_reward_embodied/mean": 0.548828125,
      "rewards/format_reward_embodied/std": 0.48774589598178864,
      "rewards/stop_prediction_reward/mean": 0.478515625,
      "rewards/stop_prediction_reward/std": 0.4023555275052786,
      "rewards/waypoint_pred_accuracy/mean": 0.007489029231998282,
      "rewards/waypoint_pred_accuracy/std": 0.0251983865261218,
      "step": 10
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 470.375,
      "completions/max_terminated_length": 470.375,
      "completions/mean_length": 253.248046875,
      "completions/mean_terminated_length": 253.248046875,
      "completions/min_length": 125.0,
      "completions/min_terminated_length": 125.0,
      "epoch": 0.023157894736842106,
      "grad_norm": 0.039429888129234314,
      "learning_rate": 2.0833333333333333e-07,
      "loss": -0.0022,
      "num_tokens": 5384873.0,
      "reward": 0.7660543769598007,
      "reward_std": 0.5604145936667919,
      "rewards/format_reward_embodied/mean": 0.607421875,
      "rewards/format_reward_embodied/std": 0.479397177696228,
      "rewards/stop_prediction_reward/mean": 0.15625,
      "rewards/stop_prediction_reward/std": 0.3427934180945158,
      "rewards/waypoint_pred_accuracy/mean": 0.0011912494257570604,
      "rewards/waypoint_pred_accuracy/std": 0.00802605507872417,
      "step": 11
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 1082.0,
      "completions/max_terminated_length": 1082.0,
      "completions/mean_length": 263.59375,
      "completions/mean_terminated_length": 264.0594940185547,
      "completions/min_length": 99.75,
      "completions/min_terminated_length": 114.625,
      "epoch": 0.02526315789473684,
      "grad_norm": 0.030105428770184517,
      "learning_rate": 2.2916666666666663e-07,
      "loss": 0.0033,
      "num_tokens": 5878681.0,
      "reward": 1.0901148244738579,
      "reward_std": 0.6814222931861877,
      "rewards/format_reward_embodied/mean": 0.6171875,
      "rewards/format_reward_embodied/std": 0.46372338756918907,
      "rewards/stop_prediction_reward/mean": 0.400390625,
      "rewards/stop_prediction_reward/std": 0.4396743141114712,
      "rewards/waypoint_pred_accuracy/mean": 0.036268358699724924,
      "rewards/waypoint_pred_accuracy/std": 0.07070713029423034,
      "step": 12
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 476.875,
      "completions/max_terminated_length": 476.875,
      "completions/mean_length": 257.3125,
      "completions/mean_terminated_length": 257.3125,
      "completions/min_length": 118.625,
      "completions/min_terminated_length": 118.625,
      "epoch": 0.02736842105263158,
      "grad_norm": 0.03954648971557617,
      "learning_rate": 2.5e-07,
      "loss": -0.0013,
      "num_tokens": 6370745.0,
      "reward": 1.1011288091540337,
      "reward_std": 0.6459922045469284,
      "rewards/format_reward_embodied/mean": 0.634765625,
      "rewards/format_reward_embodied/std": 0.47677353397011757,
      "rewards/stop_prediction_reward/mean": 0.44140625,
      "rewards/stop_prediction_reward/std": 0.42084217444062233,
      "rewards/waypoint_pred_accuracy/mean": 0.012478479564244083,
      "rewards/waypoint_pred_accuracy/std": 0.04567733465950141,
      "step": 13
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 1179.5,
      "completions/max_terminated_length": 1179.5,
      "completions/mean_length": 272.15625,
      "completions/mean_terminated_length": 272.15625,
      "completions/min_length": 120.375,
      "completions/min_terminated_length": 120.375,
      "epoch": 0.029473684210526315,
      "grad_norm": 0.03879372030496597,
      "learning_rate": 2.708333333333333e-07,
      "loss": 0.0013,
      "num_tokens": 6872585.0,
      "reward": 1.069977581501007,
      "reward_std": 0.6529600322246552,
      "rewards/format_reward_embodied/mean": 0.626953125,
      "rewards/format_reward_embodied/std": 0.46956589445471764,
      "rewards/stop_prediction_reward/mean": 0.412109375,
      "rewards/stop_prediction_reward/std": 0.42556022480130196,
      "rewards/waypoint_pred_accuracy/mean": 0.015457541714965616,
      "rewards/waypoint_pred_accuracy/std": 0.05261327166544845,
      "step": 14
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 456.75,
      "completions/max_terminated_length": 456.75,
      "completions/mean_length": 247.5078125,
      "completions/mean_terminated_length": 247.5078125,
      "completions/min_length": 116.0,
      "completions/min_terminated_length": 116.0,
      "epoch": 0.031578947368421054,
      "grad_norm": 0.03746689483523369,
      "learning_rate": 2.916666666666667e-07,
      "loss": -0.004,
      "num_tokens": 7358669.0,
      "reward": 0.816399596631527,
      "reward_std": 0.6439896002411842,
      "rewards/format_reward_embodied/mean": 0.5546875,
      "rewards/format_reward_embodied/std": 0.49549105390906334,
      "rewards/stop_prediction_reward/mean": 0.216796875,
      "rewards/stop_prediction_reward/std": 0.3949273619800806,
      "rewards/waypoint_pred_accuracy/mean": 0.022457610069225337,
      "rewards/waypoint_pred_accuracy/std": 0.04548973154789149,
      "step": 15
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 501.25,
      "completions/max_terminated_length": 501.25,
      "completions/mean_length": 254.716796875,
      "completions/mean_terminated_length": 254.716796875,
      "completions/min_length": 120.375,
      "completions/min_terminated_length": 120.375,
      "epoch": 0.03368421052631579,
      "grad_norm": 0.053112324327230453,
      "learning_rate": 3.1249999999999997e-07,
      "loss": 0.0002,
      "num_tokens": 7852156.0,
      "reward": 1.2096271365880966,
      "reward_std": 0.6100753545761108,
      "rewards/format_reward_embodied/mean": 0.76171875,
      "rewards/format_reward_embodied/std": 0.4093044362962246,
      "rewards/stop_prediction_reward/mean": 0.4453125,
      "rewards/stop_prediction_reward/std": 0.4225916638970375,
      "rewards/waypoint_pred_accuracy/mean": 0.0012979521083353873,
      "rewards/waypoint_pred_accuracy/std": 0.009194809671299708,
      "step": 16
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 472.375,
      "completions/max_terminated_length": 472.375,
      "completions/mean_length": 247.376953125,
      "completions/mean_terminated_length": 247.376953125,
      "completions/min_length": 116.25,
      "completions/min_terminated_length": 116.25,
      "epoch": 0.035789473684210524,
      "grad_norm": 0.03221385180950165,
      "learning_rate": 3.333333333333333e-07,
      "loss": -0.0006,
      "num_tokens": 8339517.0,
      "reward": 1.1803481727838516,
      "reward_std": 0.5297410599887371,
      "rewards/format_reward_embodied/mean": 0.814453125,
      "rewards/format_reward_embodied/std": 0.3874172270298004,
      "rewards/stop_prediction_reward/mean": 0.345703125,
      "rewards/stop_prediction_reward/std": 0.3674583863466978,
      "rewards/waypoint_pred_accuracy/mean": 0.010095963222276419,
      "rewards/waypoint_pred_accuracy/std": 0.0323235778856652,
      "step": 17
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 811.875,
      "completions/max_terminated_length": 811.875,
      "completions/mean_length": 254.984375,
      "completions/mean_terminated_length": 254.984375,
      "completions/min_length": 112.625,
      "completions/min_terminated_length": 112.625,
      "epoch": 0.037894736842105266,
      "grad_norm": 0.034303538501262665,
      "learning_rate": 3.541666666666667e-07,
      "loss": 0.0065,
      "num_tokens": 8828533.0,
      "reward": 1.0827482342720032,
      "reward_std": 0.49626101925969124,
      "rewards/format_reward_embodied/mean": 0.84765625,
      "rewards/format_reward_embodied/std": 0.35438157618045807,
      "rewards/stop_prediction_reward/mean": 0.234375,
      "rewards/stop_prediction_reward/std": 0.360489659011364,
      "rewards/waypoint_pred_accuracy/mean": 0.00035850519751079446,
      "rewards/waypoint_pred_accuracy/std": 0.000990356254078506,
      "step": 18
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 469.5,
      "completions/max_terminated_length": 469.5,
      "completions/mean_length": 242.603515625,
      "completions/mean_terminated_length": 243.05292129516602,
      "completions/min_length": 99.5,
      "completions/min_terminated_length": 107.75,
      "epoch": 0.04,
      "grad_norm": 0.03210434690117836,
      "learning_rate": 3.75e-07,
      "loss": 0.0002,
      "num_tokens": 9313706.0,
      "reward": 1.1656895354390144,
      "reward_std": 0.5405256152153015,
      "rewards/format_reward_embodied/mean": 0.841796875,
      "rewards/format_reward_embodied/std": 0.35758682526648045,
      "rewards/stop_prediction_reward/mean": 0.322265625,
      "rewards/stop_prediction_reward/std": 0.4072440378367901,
      "rewards/waypoint_pred_accuracy/mean": 0.0008135107927961789,
      "rewards/waypoint_pred_accuracy/std": 0.004946927132555481,
      "step": 19
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 887.5,
      "completions/max_terminated_length": 887.5,
      "completions/mean_length": 257.341796875,
      "completions/mean_terminated_length": 257.341796875,
      "completions/min_length": 117.75,
      "completions/min_terminated_length": 117.75,
      "epoch": 0.042105263157894736,
      "grad_norm": 0.023722035810351372,
      "learning_rate": 3.958333333333333e-07,
      "loss": 0.0063,
      "num_tokens": 9807001.0,
      "reward": 1.2326279431581497,
      "reward_std": 0.5118205770850182,
      "rewards/format_reward_embodied/mean": 0.880859375,
      "rewards/format_reward_embodied/std": 0.31777896732091904,
      "rewards/stop_prediction_reward/mean": 0.29296875,
      "rewards/stop_prediction_reward/std": 0.3428589329123497,
      "rewards/waypoint_pred_accuracy/mean": 0.029399914224090686,
      "rewards/waypoint_pred_accuracy/std": 0.06371456215110564,
      "step": 20
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 1251.25,
      "completions/max_terminated_length": 1251.25,
      "completions/mean_length": 258.982421875,
      "completions/mean_terminated_length": 258.982421875,
      "completions/min_length": 107.75,
      "completions/min_terminated_length": 107.75,
      "epoch": 0.04421052631578947,
      "grad_norm": 0.027127819135785103,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 0.0107,
      "num_tokens": 10300304.0,
      "reward": 1.3775597661733627,
      "reward_std": 0.5649962350726128,
      "rewards/format_reward_embodied/mean": 0.880859375,
      "rewards/format_reward_embodied/std": 0.3204925637692213,
      "rewards/stop_prediction_reward/mean": 0.451171875,
      "rewards/stop_prediction_reward/std": 0.4135790057480335,
      "rewards/waypoint_pred_accuracy/mean": 0.02276425497597198,
      "rewards/waypoint_pred_accuracy/std": 0.053010769921375774,
      "step": 21
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 764.5,
      "completions/max_terminated_length": 764.5,
      "completions/mean_length": 247.626953125,
      "completions/mean_terminated_length": 247.626953125,
      "completions/min_length": 120.625,
      "completions/min_terminated_length": 120.625,
      "epoch": 0.04631578947368421,
      "grad_norm": 0.026827372610569,
      "learning_rate": 4.375e-07,
      "loss": 0.0046,
      "num_tokens": 10787473.0,
      "reward": 1.2097989320755005,
      "reward_std": 0.4106667507439852,
      "rewards/format_reward_embodied/mean": 0.9609375,
      "rewards/format_reward_embodied/std": 0.1782014612108469,
      "rewards/stop_prediction_reward/mean": 0.248046875,
      "rewards/stop_prediction_reward/std": 0.3576600421220064,
      "rewards/waypoint_pred_accuracy/mean": 0.0004072752802812829,
      "rewards/waypoint_pred_accuracy/std": 0.001209557721267629,
      "step": 22
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 699.125,
      "completions/max_terminated_length": 699.125,
      "completions/mean_length": 247.458984375,
      "completions/mean_terminated_length": 247.458984375,
      "completions/min_length": 121.5,
      "completions/min_terminated_length": 121.5,
      "epoch": 0.04842105263157895,
      "grad_norm": 0.024113576859235764,
      "learning_rate": 4.5833333333333327e-07,
      "loss": 0.0032,
      "num_tokens": 11273276.0,
      "reward": 1.3881124705076218,
      "reward_std": 0.5006838031113148,
      "rewards/format_reward_embodied/mean": 0.943359375,
      "rewards/format_reward_embodied/std": 0.22258390858769417,
      "rewards/stop_prediction_reward/mean": 0.42578125,
      "rewards/stop_prediction_reward/std": 0.4117406941950321,
      "rewards/waypoint_pred_accuracy/mean": 0.00948592593158537,
      "rewards/waypoint_pred_accuracy/std": 0.03859481842846435,
      "step": 23
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 1426.0,
      "completions/max_terminated_length": 1426.0,
      "completions/mean_length": 270.671875,
      "completions/mean_terminated_length": 271.2647590637207,
      "completions/min_length": 100.0,
      "completions/min_terminated_length": 115.5,
      "epoch": 0.05052631578947368,
      "grad_norm": 0.021521741524338722,
      "learning_rate": 4.791666666666667e-07,
      "loss": 0.0138,
      "num_tokens": 11772628.0,
      "reward": 1.412862166762352,
      "reward_std": 0.4836365692317486,
      "rewards/format_reward_embodied/mean": 0.94140625,
      "rewards/format_reward_embodied/std": 0.22402114421129227,
      "rewards/stop_prediction_reward/mean": 0.46875,
      "rewards/stop_prediction_reward/std": 0.4050610587000847,
      "rewards/waypoint_pred_accuracy/mean": 0.0013529594958612285,
      "rewards/waypoint_pred_accuracy/std": 0.008442860726859186,
      "step": 24
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 503.375,
      "completions/max_terminated_length": 503.375,
      "completions/mean_length": 252.7265625,
      "completions/mean_terminated_length": 252.7265625,
      "completions/min_length": 115.75,
      "completions/min_terminated_length": 115.75,
      "epoch": 0.05263157894736842,
      "grad_norm": 0.023303357884287834,
      "learning_rate": 5e-07,
      "loss": 0.0022,
      "num_tokens": 12260936.0,
      "reward": 1.3870358616113663,
      "reward_std": 0.4634270928800106,
      "rewards/format_reward_embodied/mean": 0.97265625,
      "rewards/format_reward_embodied/std": 0.14094455912709236,
      "rewards/stop_prediction_reward/mean": 0.404296875,
      "rewards/stop_prediction_reward/std": 0.41460882127285004,
      "rewards/waypoint_pred_accuracy/mean": 0.005041355174832356,
      "rewards/waypoint_pred_accuracy/std": 0.030215839982022222,
      "step": 25
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 802.125,
      "completions/max_terminated_length": 802.125,
      "completions/mean_length": 252.96484375,
      "completions/mean_terminated_length": 252.96484375,
      "completions/min_length": 117.625,
      "completions/min_terminated_length": 117.625,
      "epoch": 0.05473684210526316,
      "grad_norm": 0.022509992122650146,
      "learning_rate": 5.208333333333334e-07,
      "loss": 0.0072,
      "num_tokens": 12752118.0,
      "reward": 1.3777707070112228,
      "reward_std": 0.4540855921804905,
      "rewards/format_reward_embodied/mean": 0.962890625,
      "rewards/format_reward_embodied/std": 0.1879090555012226,
      "rewards/stop_prediction_reward/mean": 0.388671875,
      "rewards/stop_prediction_reward/std": 0.3826281502842903,
      "rewards/waypoint_pred_accuracy/mean": 0.013104110299219339,
      "rewards/waypoint_pred_accuracy/std": 0.023219284697190704,
      "step": 26
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 1812.5,
      "completions/max_terminated_length": 1812.5,
      "completions/mean_length": 276.763671875,
      "completions/mean_terminated_length": 277.41750717163086,
      "completions/min_length": 101.75,
      "completions/min_terminated_length": 118.5,
      "epoch": 0.056842105263157895,
      "grad_norm": 0.025979243218898773,
      "learning_rate": 5.416666666666666e-07,
      "loss": 0.0201,
      "num_tokens": 13257853.0,
      "reward": 1.195212036371231,
      "reward_std": 0.4585261270403862,
      "rewards/format_reward_embodied/mean": 0.9453125,
      "rewards/format_reward_embodied/std": 0.21002393402159214,
      "rewards/stop_prediction_reward/mean": 0.248046875,
      "rewards/stop_prediction_reward/std": 0.3809744007885456,
      "rewards/waypoint_pred_accuracy/mean": 0.0009263323242719891,
      "rewards/waypoint_pred_accuracy/std": 0.0071097194065939074,
      "step": 27
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 1423.875,
      "completions/max_terminated_length": 1423.875,
      "completions/mean_length": 266.958984375,
      "completions/mean_terminated_length": 266.958984375,
      "completions/min_length": 121.5,
      "completions/min_terminated_length": 121.5,
      "epoch": 0.05894736842105263,
      "grad_norm": 0.02617989294230938,
      "learning_rate": 5.625e-07,
      "loss": 0.0123,
      "num_tokens": 13756648.0,
      "reward": 1.4108145833015442,
      "reward_std": 0.4676571935415268,
      "rewards/format_reward_embodied/mean": 0.96484375,
      "rewards/format_reward_embodied/std": 0.1660303734242916,
      "rewards/stop_prediction_reward/mean": 0.427734375,
      "rewards/stop_prediction_reward/std": 0.40000360645353794,
      "rewards/waypoint_pred_accuracy/mean": 0.009118233890500024,
      "rewards/waypoint_pred_accuracy/std": 0.03189600147387662,
      "step": 28
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 790.375,
      "completions/max_terminated_length": 790.375,
      "completions/mean_length": 253.384765625,
      "completions/mean_terminated_length": 253.384765625,
      "completions/min_length": 113.125,
      "completions/min_terminated_length": 113.125,
      "epoch": 0.061052631578947365,
      "grad_norm": 0.0284319706261158,
      "learning_rate": 5.833333333333334e-07,
      "loss": 0.004,
      "num_tokens": 14246829.0,
      "reward": 1.3669871091842651,
      "reward_std": 0.476172287017107,
      "rewards/format_reward_embodied/mean": 0.9609375,
      "rewards/format_reward_embodied/std": 0.19018890894949436,
      "rewards/stop_prediction_reward/mean": 0.376953125,
      "rewards/stop_prediction_reward/std": 0.38546351715922356,
      "rewards/waypoint_pred_accuracy/mean": 0.014548240964149528,
      "rewards/waypoint_pred_accuracy/std": 0.048405178813538896,
      "step": 29
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 946.125,
      "completions/max_terminated_length": 946.125,
      "completions/mean_length": 267.513671875,
      "completions/mean_terminated_length": 267.513671875,
      "completions/min_length": 118.25,
      "completions/min_terminated_length": 118.25,
      "epoch": 0.06315789473684211,
      "grad_norm": 0.02758762799203396,
      "learning_rate": 6.041666666666666e-07,
      "loss": 0.0087,
      "num_tokens": 14742708.0,
      "reward": 1.3311925828456879,
      "reward_std": 0.5349335558712482,
      "rewards/format_reward_embodied/mean": 0.94921875,
      "rewards/format_reward_embodied/std": 0.21471346728503704,
      "rewards/stop_prediction_reward/mean": 0.375,
      "rewards/stop_prediction_reward/std": 0.4617812857031822,
      "rewards/waypoint_pred_accuracy/mean": 0.003486919093547547,
      "rewards/waypoint_pred_accuracy/std": 0.022435040602790033,
      "step": 30
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 468.875,
      "completions/max_terminated_length": 468.875,
      "completions/mean_length": 247.80859375,
      "completions/mean_terminated_length": 247.80859375,
      "completions/min_length": 116.125,
      "completions/min_terminated_length": 116.125,
      "epoch": 0.06526315789473684,
      "grad_norm": 0.024238400161266327,
      "learning_rate": 6.249999999999999e-07,
      "loss": 0.0006,
      "num_tokens": 15230930.0,
      "reward": 1.4348655045032501,
      "reward_std": 0.4678279310464859,
      "rewards/format_reward_embodied/mean": 0.982421875,
      "rewards/format_reward_embodied/std": 0.09929289110004902,
      "rewards/stop_prediction_reward/mean": 0.439453125,
      "rewards/stop_prediction_reward/std": 0.4299692139029503,
      "rewards/waypoint_pred_accuracy/mean": 0.006495264507384302,
      "rewards/waypoint_pred_accuracy/std": 0.03605599632994938,
      "step": 31
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 701.5,
      "completions/max_terminated_length": 701.5,
      "completions/mean_length": 247.302734375,
      "completions/mean_terminated_length": 247.302734375,
      "completions/min_length": 102.625,
      "completions/min_terminated_length": 102.625,
      "epoch": 0.06736842105263158,
      "grad_norm": 0.03117675893008709,
      "learning_rate": 6.458333333333333e-07,
      "loss": 0.0048,
      "num_tokens": 15717165.0,
      "reward": 1.276307299733162,
      "reward_std": 0.4745783172547817,
      "rewards/format_reward_embodied/mean": 0.9765625,
      "rewards/format_reward_embodied/std": 0.11361248232424259,
      "rewards/stop_prediction_reward/mean": 0.287109375,
      "rewards/stop_prediction_reward/std": 0.4378196634352207,
      "rewards/waypoint_pred_accuracy/mean": 0.006317712715827028,
      "rewards/waypoint_pred_accuracy/std": 0.02758992835879357,
      "step": 32
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 736.625,
      "completions/max_terminated_length": 736.625,
      "completions/mean_length": 259.27734375,
      "completions/mean_terminated_length": 259.27734375,
      "completions/min_length": 132.125,
      "completions/min_terminated_length": 132.125,
      "epoch": 0.06947368421052631,
      "grad_norm": 0.0244793351739645,
      "learning_rate": 6.666666666666666e-07,
      "loss": 0.0053,
      "num_tokens": 16211835.0,
      "reward": 1.540584921836853,
      "reward_std": 0.49051226675510406,
      "rewards/format_reward_embodied/mean": 0.96484375,
      "rewards/format_reward_embodied/std": 0.18235719576478004,
      "rewards/stop_prediction_reward/mean": 0.57421875,
      "rewards/stop_prediction_reward/std": 0.43413203582167625,
      "rewards/waypoint_pred_accuracy/mean": 0.0007612065199357397,
      "rewards/waypoint_pred_accuracy/std": 0.005327658200312494,
      "step": 33
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 881.375,
      "completions/max_terminated_length": 881.375,
      "completions/mean_length": 264.103515625,
      "completions/mean_terminated_length": 264.58401679992676,
      "completions/min_length": 101.375,
      "completions/min_terminated_length": 113.25,
      "epoch": 0.07157894736842105,
      "grad_norm": 0.0207452904433012,
      "learning_rate": 6.875e-07,
      "loss": 0.0068,
      "num_tokens": 16706928.0,
      "reward": 1.4342780411243439,
      "reward_std": 0.5214410163462162,
      "rewards/format_reward_embodied/mean": 0.9765625,
      "rewards/format_reward_embodied/std": 0.12835253402590752,
      "rewards/stop_prediction_reward/mean": 0.453125,
      "rewards/stop_prediction_reward/std": 0.47532549500465393,
      "rewards/waypoint_pred_accuracy/mean": 0.0022952774760399775,
      "rewards/waypoint_pred_accuracy/std": 0.016314118760955294,
      "step": 34
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 544.75,
      "completions/max_terminated_length": 544.75,
      "completions/mean_length": 247.744140625,
      "completions/mean_terminated_length": 247.744140625,
      "completions/min_length": 115.375,
      "completions/min_terminated_length": 115.375,
      "epoch": 0.07368421052631578,
      "grad_norm": 0.02019825391471386,
      "learning_rate": 7.083333333333334e-07,
      "loss": 0.0026,
      "num_tokens": 17196205.0,
      "reward": 1.4062748402357101,
      "reward_std": 0.5094473846256733,
      "rewards/format_reward_embodied/mean": 0.97265625,
      "rewards/format_reward_embodied/std": 0.15960253402590752,
      "rewards/stop_prediction_reward/mean": 0.43359375,
      "rewards/stop_prediction_reward/std": 0.4651285596191883,
      "rewards/waypoint_pred_accuracy/mean": 1.2417779022436235e-05,
      "rewards/waypoint_pred_accuracy/std": 2.711040125229227e-05,
      "step": 35
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 1784.875,
      "completions/max_terminated_length": 1784.875,
      "completions/mean_length": 291.40625,
      "completions/mean_terminated_length": 291.40625,
      "completions/min_length": 128.75,
      "completions/min_terminated_length": 128.75,
      "epoch": 0.07578947368421053,
      "grad_norm": 0.038221534341573715,
      "learning_rate": 7.291666666666666e-07,
      "loss": 0.0175,
      "num_tokens": 17710461.0,
      "reward": 1.4911664873361588,
      "reward_std": 0.5095744393765926,
      "rewards/format_reward_embodied/mean": 0.978515625,
      "rewards/format_reward_embodied/std": 0.11029814556241035,
      "rewards/stop_prediction_reward/mean": 0.498046875,
      "rewards/stop_prediction_reward/std": 0.46525831148028374,
      "rewards/waypoint_pred_accuracy/mean": 0.007301997149683504,
      "rewards/waypoint_pred_accuracy/std": 0.024017843105910822,
      "step": 36
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 472.75,
      "completions/max_terminated_length": 472.75,
      "completions/mean_length": 248.03125,
      "completions/mean_terminated_length": 248.03125,
      "completions/min_length": 117.375,
      "completions/min_terminated_length": 117.375,
      "epoch": 0.07789473684210527,
      "grad_norm": 0.02083674818277359,
      "learning_rate": 7.5e-07,
      "loss": 0.0023,
      "num_tokens": 18197965.0,
      "reward": 1.4861425906419754,
      "reward_std": 0.5738211683928967,
      "rewards/format_reward_embodied/mean": 0.9921875,
      "rewards/format_reward_embodied/std": 0.05317101255059242,
      "rewards/stop_prediction_reward/mean": 0.42578125,
      "rewards/stop_prediction_reward/std": 0.4866231083869934,
      "rewards/waypoint_pred_accuracy/mean": 0.03408691443473799,
      "rewards/waypoint_pred_accuracy/std": 0.08652095211436972,
      "step": 37
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 884.625,
      "completions/max_terminated_length": 884.625,
      "completions/mean_length": 258.75390625,
      "completions/mean_terminated_length": 258.75390625,
      "completions/min_length": 125.5,
      "completions/min_terminated_length": 125.5,
      "epoch": 0.08,
      "grad_norm": 0.021931249648332596,
      "learning_rate": 7.708333333333333e-07,
      "loss": 0.0048,
      "num_tokens": 18692239.0,
      "reward": 1.6138971894979477,
      "reward_std": 0.5961987935006618,
      "rewards/format_reward_embodied/mean": 0.98828125,
      "rewards/format_reward_embodied/std": 0.07509202510118484,
      "rewards/stop_prediction_reward/mean": 0.513671875,
      "rewards/stop_prediction_reward/std": 0.488056443631649,
      "rewards/waypoint_pred_accuracy/mean": 0.05597203067737698,
      "rewards/waypoint_pred_accuracy/std": 0.09776196270839843,
      "step": 38
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 755.875,
      "completions/max_terminated_length": 755.875,
      "completions/mean_length": 250.787109375,
      "completions/mean_terminated_length": 250.787109375,
      "completions/min_length": 119.125,
      "completions/min_terminated_length": 119.125,
      "epoch": 0.08210526315789474,
      "grad_norm": 0.15428805351257324,
      "learning_rate": 7.916666666666666e-07,
      "loss": 0.0043,
      "num_tokens": 19183074.0,
      "reward": 1.4990187734365463,
      "reward_std": 0.49899255111813545,
      "rewards/format_reward_embodied/mean": 0.98046875,
      "rewards/format_reward_embodied/std": 0.1064315214753151,
      "rewards/stop_prediction_reward/mean": 0.517578125,
      "rewards/stop_prediction_reward/std": 0.4714191108942032,
      "rewards/waypoint_pred_accuracy/mean": 0.00048596067600919097,
      "rewards/waypoint_pred_accuracy/std": 0.0015497554879284957,
      "step": 39
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 443.0,
      "completions/max_terminated_length": 443.0,
      "completions/mean_length": 246.05078125,
      "completions/mean_terminated_length": 246.05078125,
      "completions/min_length": 117.125,
      "completions/min_terminated_length": 117.125,
      "epoch": 0.08421052631578947,
      "grad_norm": 0.033689290285110474,
      "learning_rate": 8.125e-07,
      "loss": 0.0002,
      "num_tokens": 19674556.0,
      "reward": 1.52012699842453,
      "reward_std": 0.5320924893021584,
      "rewards/format_reward_embodied/mean": 0.98828125,
      "rewards/format_reward_embodied/std": 0.09375,
      "rewards/stop_prediction_reward/mean": 0.494140625,
      "rewards/stop_prediction_reward/std": 0.4897289089858532,
      "rewards/waypoint_pred_accuracy/mean": 0.018852562323445468,
      "rewards/waypoint_pred_accuracy/std": 0.04221876614610676,
      "step": 40
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 451.625,
      "completions/max_terminated_length": 451.625,
      "completions/mean_length": 233.865234375,
      "completions/mean_terminated_length": 233.865234375,
      "completions/min_length": 106.0,
      "completions/min_terminated_length": 106.0,
      "epoch": 0.0863157894736842,
      "grad_norm": 0.025640549138188362,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.0011,
      "num_tokens": 20156151.0,
      "reward": 1.591551125049591,
      "reward_std": 0.5175898559391499,
      "rewards/format_reward_embodied/mean": 0.98828125,
      "rewards/format_reward_embodied/std": 0.08442101255059242,
      "rewards/stop_prediction_reward/mean": 0.5703125,
      "rewards/stop_prediction_reward/std": 0.4659374840557575,
      "rewards/waypoint_pred_accuracy/mean": 0.016478684779628456,
      "rewards/waypoint_pred_accuracy/std": 0.05339623770077609,
      "step": 41
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 500.0,
      "completions/max_terminated_length": 500.0,
      "completions/mean_length": 242.478515625,
      "completions/mean_terminated_length": 242.478515625,
      "completions/min_length": 112.25,
      "completions/min_terminated_length": 112.25,
      "epoch": 0.08842105263157894,
      "grad_norm": 0.026281312108039856,
      "learning_rate": 8.541666666666666e-07,
      "loss": -0.0,
      "num_tokens": 20639788.0,
      "reward": 1.5611481666564941,
      "reward_std": 0.4784001186490059,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.560546875,
      "rewards/stop_prediction_reward/std": 0.4726823903620243,
      "rewards/waypoint_pred_accuracy/mean": 0.002253779932873278,
      "rewards/waypoint_pred_accuracy/std": 0.015716358982028673,
      "step": 42
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 810.875,
      "completions/max_terminated_length": 810.875,
      "completions/mean_length": 249.939453125,
      "completions/mean_terminated_length": 249.939453125,
      "completions/min_length": 122.875,
      "completions/min_terminated_length": 122.875,
      "epoch": 0.09052631578947369,
      "grad_norm": 0.021544892340898514,
      "learning_rate": 8.75e-07,
      "loss": 0.0029,
      "num_tokens": 21128013.0,
      "reward": 1.4784268736839294,
      "reward_std": 0.48740382865071297,
      "rewards/format_reward_embodied/mean": 0.990234375,
      "rewards/format_reward_embodied/std": 0.06879601255059242,
      "rewards/stop_prediction_reward/mean": 0.484375,
      "rewards/stop_prediction_reward/std": 0.4655333496630192,
      "rewards/waypoint_pred_accuracy/mean": 0.0019087545888396566,
      "rewards/waypoint_pred_accuracy/std": 0.010897019660660593,
      "step": 43
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 643.625,
      "completions/max_terminated_length": 643.625,
      "completions/mean_length": 245.134765625,
      "completions/mean_terminated_length": 245.66865158081055,
      "completions/min_length": 101.25,
      "completions/min_terminated_length": 117.75,
      "epoch": 0.09263157894736843,
      "grad_norm": 0.025143882259726524,
      "learning_rate": 8.958333333333334e-07,
      "loss": 0.0038,
      "num_tokens": 21614866.0,
      "reward": 1.6192794144153595,
      "reward_std": 0.4844088666141033,
      "rewards/format_reward_embodied/mean": 0.982421875,
      "rewards/format_reward_embodied/std": 0.11263803765177727,
      "rewards/stop_prediction_reward/mean": 0.63671875,
      "rewards/stop_prediction_reward/std": 0.4543136991560459,
      "rewards/waypoint_pred_accuracy/mean": 6.939472140954406e-05,
      "rewards/waypoint_pred_accuracy/std": 0.00043998260686706566,
      "step": 44
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 479.375,
      "completions/max_terminated_length": 479.375,
      "completions/mean_length": 238.673828125,
      "completions/mean_terminated_length": 238.673828125,
      "completions/min_length": 115.625,
      "completions/min_terminated_length": 115.625,
      "epoch": 0.09473684210526316,
      "grad_norm": 0.023476244881749153,
      "learning_rate": 9.166666666666665e-07,
      "loss": 0.0007,
      "num_tokens": 22098731.0,
      "reward": 1.512737661600113,
      "reward_std": 0.5365940853953362,
      "rewards/format_reward_embodied/mean": 0.994140625,
      "rewards/format_reward_embodied/std": 0.026630254462361336,
      "rewards/stop_prediction_reward/mean": 0.45703125,
      "rewards/stop_prediction_reward/std": 0.48733755201101303,
      "rewards/waypoint_pred_accuracy/mean": 0.03078289819380149,
      "rewards/waypoint_pred_accuracy/std": 0.043302708805491946,
      "step": 45
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 641.875,
      "completions/max_terminated_length": 641.875,
      "completions/mean_length": 240.724609375,
      "completions/mean_terminated_length": 240.724609375,
      "completions/min_length": 118.625,
      "completions/min_terminated_length": 118.625,
      "epoch": 0.0968421052631579,
      "grad_norm": 0.020364033058285713,
      "learning_rate": 9.374999999999999e-07,
      "loss": 0.0029,
      "num_tokens": 22583070.0,
      "reward": 1.6677764654159546,
      "reward_std": 0.5154935717582703,
      "rewards/format_reward_embodied/mean": 0.98828125,
      "rewards/format_reward_embodied/std": 0.07350525446236134,
      "rewards/stop_prediction_reward/mean": 0.626953125,
      "rewards/stop_prediction_reward/std": 0.44941750913858414,
      "rewards/waypoint_pred_accuracy/mean": 0.02627104918529134,
      "rewards/waypoint_pred_accuracy/std": 0.0397095277659446,
      "step": 46
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 657.375,
      "completions/max_terminated_length": 657.375,
      "completions/mean_length": 239.5546875,
      "completions/mean_terminated_length": 239.5546875,
      "completions/min_length": 118.875,
      "completions/min_terminated_length": 118.875,
      "epoch": 0.09894736842105263,
      "grad_norm": 0.023410305380821228,
      "learning_rate": 9.583333333333334e-07,
      "loss": 0.0039,
      "num_tokens": 23067066.0,
      "reward": 1.7648355215787888,
      "reward_std": 0.5304676033556461,
      "rewards/format_reward_embodied/mean": 0.994140625,
      "rewards/format_reward_embodied/std": 0.046875,
      "rewards/stop_prediction_reward/mean": 0.685546875,
      "rewards/stop_prediction_reward/std": 0.45773619785904884,
      "rewards/waypoint_pred_accuracy/mean": 0.042574012356195935,
      "rewards/waypoint_pred_accuracy/std": 0.07306883804197034,
      "step": 47
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 423.625,
      "completions/max_terminated_length": 423.625,
      "completions/mean_length": 236.12890625,
      "completions/mean_terminated_length": 236.12890625,
      "completions/min_length": 117.0,
      "completions/min_terminated_length": 117.0,
      "epoch": 0.10105263157894737,
      "grad_norm": 0.023688504472374916,
      "learning_rate": 9.791666666666667e-07,
      "loss": -0.0006,
      "num_tokens": 23549436.0,
      "reward": 1.70817232131958,
      "reward_std": 0.5296522080898285,
      "rewards/format_reward_embodied/mean": 0.990234375,
      "rewards/format_reward_embodied/std": 0.06879601255059242,
      "rewards/stop_prediction_reward/mean": 0.63671875,
      "rewards/stop_prediction_reward/std": 0.45957546308636665,
      "rewards/waypoint_pred_accuracy/mean": 0.04060960025526583,
      "rewards/waypoint_pred_accuracy/std": 0.07721152482554317,
      "step": 48
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 439.5,
      "completions/max_terminated_length": 439.5,
      "completions/mean_length": 239.724609375,
      "completions/mean_terminated_length": 239.724609375,
      "completions/min_length": 112.625,
      "completions/min_terminated_length": 112.625,
      "epoch": 0.1031578947368421,
      "grad_norm": 0.04075814038515091,
      "learning_rate": 1e-06,
      "loss": 0.0012,
      "num_tokens": 24036719.0,
      "reward": 1.5996688604354858,
      "reward_std": 0.43832388520240784,
      "rewards/format_reward_embodied/mean": 0.9921875,
      "rewards/format_reward_embodied/std": 0.05317101255059242,
      "rewards/stop_prediction_reward/mean": 0.607421875,
      "rewards/stop_prediction_reward/std": 0.4225967414677143,
      "rewards/waypoint_pred_accuracy/mean": 2.974685131453066e-05,
      "rewards/waypoint_pred_accuracy/std": 0.00014841147029464143,
      "step": 49
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 0.02866051159799099,
      "learning_rate": 9.999878206375666e-07,
      "loss": 0.0015,
      "step": 50
    },
    {
      "epoch": 0.10526315789473684,
      "eval_clip_ratio": 0.0,
      "eval_completions/clipped_ratio": 0.00015625,
      "eval_completions/max_length": 683.96,
      "eval_completions/max_terminated_length": 683.96,
      "eval_completions/mean_length": 238.32913192749024,
      "eval_completions/mean_terminated_length": 238.36774322509766,
      "eval_completions/min_length": 112.41,
      "eval_completions/min_terminated_length": 113.78,
      "eval_loss": 0.003631497733294964,
      "eval_num_tokens": 24520656.0,
      "eval_reward": 1.6925481045246125,
      "eval_reward_std": 0.4437328398227692,
      "eval_rewards/format_reward_embodied/mean": 0.99328125,
      "eval_rewards/format_reward_embodied/std": 0.04765250638127327,
      "eval_rewards/stop_prediction_reward/mean": 0.66328125,
      "eval_rewards/stop_prediction_reward/std": 0.40405205205082895,
      "eval_rewards/waypoint_pred_accuracy/mean": 0.017992802756573712,
      "eval_rewards/waypoint_pred_accuracy/std": 0.04001634344926445,
      "eval_runtime": 1355.6164,
      "eval_samples_per_second": 0.074,
      "eval_steps_per_second": 0.001,
      "step": 50
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 450.375,
      "completions/max_terminated_length": 450.375,
      "completions/mean_length": 239.6259765625,
      "completions/mean_terminated_length": 239.6259765625,
      "completions/min_length": 116.5,
      "completions/min_terminated_length": 116.5,
      "epoch": 0.10736842105263159,
      "grad_norm": 0.02182384580373764,
      "learning_rate": 9.999512832095417e-07,
      "loss": 0.0005,
      "num_tokens": 25006064.0,
      "reward": 1.68679628521204,
      "reward_std": 0.4470259975641966,
      "rewards/format_reward_embodied/mean": 0.9931640625,
      "rewards/format_reward_embodied/std": 0.05002300627529621,
      "rewards/stop_prediction_reward/mean": 0.65234375,
      "rewards/stop_prediction_reward/std": 0.4022445324808359,
      "rewards/waypoint_pred_accuracy/mean": 0.02064424328141934,
      "rewards/waypoint_pred_accuracy/std": 0.05010251636930185,
      "step": 51
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 427.5,
      "completions/max_terminated_length": 427.5,
      "completions/mean_length": 230.73828125,
      "completions/mean_terminated_length": 231.21041297912598,
      "completions/min_length": 99.875,
      "completions/min_terminated_length": 114.375,
      "epoch": 0.10947368421052632,
      "grad_norm": 0.02381049655377865,
      "learning_rate": 9.998903896937148e-07,
      "loss": 0.0012,
      "num_tokens": 25485546.0,
      "reward": 1.7381224930286407,
      "reward_std": 0.4321938529610634,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.716796875,
      "rewards/stop_prediction_reward/std": 0.3914393372833729,
      "rewards/waypoint_pred_accuracy/mean": 0.01261594578698389,
      "rewards/waypoint_pred_accuracy/std": 0.03588760504630355,
      "step": 52
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 450.5,
      "completions/max_terminated_length": 450.5,
      "completions/mean_length": 234.287109375,
      "completions/mean_terminated_length": 234.287109375,
      "completions/min_length": 113.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.11157894736842106,
      "grad_norm": 0.05357426404953003,
      "learning_rate": 9.998051433862818e-07,
      "loss": -0.0001,
      "num_tokens": 25966589.0,
      "reward": 1.6857829988002777,
      "reward_std": 0.3670726828277111,
      "rewards/format_reward_embodied/mean": 0.994140625,
      "rewards/format_reward_embodied/std": 0.046875,
      "rewards/stop_prediction_reward/mean": 0.69140625,
      "rewards/stop_prediction_reward/std": 0.3526647798717022,
      "rewards/waypoint_pred_accuracy/mean": 0.00011806207834793561,
      "rewards/waypoint_pred_accuracy/std": 0.0009230129067357363,
      "step": 53
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 413.75,
      "completions/max_terminated_length": 413.75,
      "completions/mean_length": 226.93359375,
      "completions/mean_terminated_length": 226.93359375,
      "completions/min_length": 108.125,
      "completions/min_terminated_length": 108.125,
      "epoch": 0.11368421052631579,
      "grad_norm": 0.024311864748597145,
      "learning_rate": 9.996955489016681e-07,
      "loss": 0.0006,
      "num_tokens": 26444507.0,
      "reward": 1.5553173422813416,
      "reward_std": 0.38553351908922195,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.5234375,
      "rewards/stop_prediction_reward/std": 0.3467428870499134,
      "rewards/waypoint_pred_accuracy/mean": 0.015939914255370412,
      "rewards/waypoint_pred_accuracy/std": 0.039412272800977984,
      "step": 54
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 667.5,
      "completions/max_terminated_length": 667.5,
      "completions/mean_length": 246.060546875,
      "completions/mean_terminated_length": 246.060546875,
      "completions/min_length": 111.125,
      "completions/min_terminated_length": 111.125,
      "epoch": 0.11578947368421053,
      "grad_norm": 0.020382562652230263,
      "learning_rate": 9.995616121722783e-07,
      "loss": 0.0024,
      "num_tokens": 26931770.0,
      "reward": 1.543064832687378,
      "reward_std": 0.35424431413412094,
      "rewards/format_reward_embodied/mean": 0.990234375,
      "rewards/format_reward_embodied/std": 0.06879601255059242,
      "rewards/stop_prediction_reward/mean": 0.548828125,
      "rewards/stop_prediction_reward/std": 0.3284119311720133,
      "rewards/waypoint_pred_accuracy/mean": 0.002001162469869708,
      "rewards/waypoint_pred_accuracy/std": 0.01592240231514961,
      "step": 55
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 462.75,
      "completions/max_terminated_length": 462.75,
      "completions/mean_length": 225.984375,
      "completions/mean_terminated_length": 225.984375,
      "completions/min_length": 113.875,
      "completions/min_terminated_length": 113.875,
      "epoch": 0.11789473684210526,
      "grad_norm": 0.020005574449896812,
      "learning_rate": 9.994033404481736e-07,
      "loss": 0.0004,
      "num_tokens": 27406770.0,
      "reward": 1.6594459414482117,
      "reward_std": 0.3725608382374048,
      "rewards/format_reward_embodied/mean": 0.986328125,
      "rewards/format_reward_embodied/std": 0.07980126701295376,
      "rewards/stop_prediction_reward/mean": 0.6640625,
      "rewards/stop_prediction_reward/std": 0.33192422799766064,
      "rewards/waypoint_pred_accuracy/mean": 0.004527658324403952,
      "rewards/waypoint_pred_accuracy/std": 0.017977507960307844,
      "step": 56
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 450.5,
      "completions/max_terminated_length": 450.5,
      "completions/mean_length": 234.10546875,
      "completions/mean_terminated_length": 234.10546875,
      "completions/min_length": 115.0,
      "completions/min_terminated_length": 115.0,
      "epoch": 0.12,
      "grad_norm": 0.02229822427034378,
      "learning_rate": 9.992207422966824e-07,
      "loss": -0.0001,
      "num_tokens": 27889640.0,
      "reward": 1.4391246140003204,
      "reward_std": 0.35483869537711143,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.4375,
      "rewards/stop_prediction_reward/std": 0.34185592643916607,
      "rewards/waypoint_pred_accuracy/mean": 0.002765428128381,
      "rewards/waypoint_pred_accuracy/std": 0.014523094108374813,
      "step": 57
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 747.625,
      "completions/max_terminated_length": 747.625,
      "completions/mean_length": 233.369140625,
      "completions/mean_terminated_length": 233.90501022338867,
      "completions/min_length": 99.75,
      "completions/min_terminated_length": 113.5,
      "epoch": 0.12210526315789473,
      "grad_norm": 0.025426389649510384,
      "learning_rate": 9.990138276019335e-07,
      "loss": 0.0038,
      "num_tokens": 28370213.0,
      "reward": 1.6101858913898468,
      "reward_std": 0.3749941308051348,
      "rewards/format_reward_embodied/mean": 0.9921875,
      "rewards/format_reward_embodied/std": 0.05317101255059242,
      "rewards/stop_prediction_reward/mean": 0.59375,
      "rewards/stop_prediction_reward/std": 0.32783540338277817,
      "rewards/waypoint_pred_accuracy/mean": 0.012124196402905074,
      "rewards/waypoint_pred_accuracy/std": 0.041881935120967384,
      "step": 58
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 1494.75,
      "completions/max_terminated_length": 1494.75,
      "completions/mean_length": 264.875,
      "completions/mean_terminated_length": 264.875,
      "completions/min_length": 119.125,
      "completions/min_terminated_length": 119.125,
      "epoch": 0.12421052631578948,
      "grad_norm": 0.029843533411622047,
      "learning_rate": 9.987826075643228e-07,
      "loss": 0.0225,
      "num_tokens": 28866405.0,
      "reward": 1.608642503619194,
      "reward_std": 0.37335721030831337,
      "rewards/format_reward_embodied/mean": 0.98828125,
      "rewards/format_reward_embodied/std": 0.07509202510118484,
      "rewards/stop_prediction_reward/mean": 0.619140625,
      "rewards/stop_prediction_reward/std": 0.33958676643669605,
      "rewards/waypoint_pred_accuracy/mean": 0.0006103267239083258,
      "rewards/waypoint_pred_accuracy/std": 0.003143552353714926,
      "step": 59
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 460.875,
      "completions/max_terminated_length": 460.875,
      "completions/mean_length": 250.77734375,
      "completions/mean_terminated_length": 250.77734375,
      "completions/min_length": 122.25,
      "completions/min_terminated_length": 122.25,
      "epoch": 0.12631578947368421,
      "grad_norm": 0.02350872941315174,
      "learning_rate": 9.985270946999066e-07,
      "loss": -0.0006,
      "num_tokens": 29357939.0,
      "reward": 1.5302962958812714,
      "reward_std": 0.39041563123464584,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.51171875,
      "rewards/stop_prediction_reward/std": 0.3654658328741789,
      "rewards/waypoint_pred_accuracy/mean": 0.01026534708216827,
      "rewards/waypoint_pred_accuracy/std": 0.03301238967106072,
      "step": 60
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 740.0,
      "completions/max_terminated_length": 740.0,
      "completions/mean_length": 254.34765625,
      "completions/mean_terminated_length": 254.9265899658203,
      "completions/min_length": 107.25,
      "completions/min_terminated_length": 128.25,
      "epoch": 0.12842105263157894,
      "grad_norm": 0.028141073882579803,
      "learning_rate": 9.982473028397236e-07,
      "loss": 0.004,
      "num_tokens": 29850341.0,
      "reward": 1.908710554242134,
      "reward_std": 0.4512513056397438,
      "rewards/format_reward_embodied/mean": 0.98828125,
      "rewards/format_reward_embodied/std": 0.07509202510118484,
      "rewards/stop_prediction_reward/mean": 0.8515625,
      "rewards/stop_prediction_reward/std": 0.34670688211917877,
      "rewards/waypoint_pred_accuracy/mean": 0.034433397710529334,
      "rewards/waypoint_pred_accuracy/std": 0.06185820607151982,
      "step": 61
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 888.0,
      "completions/max_terminated_length": 888.0,
      "completions/mean_length": 255.419921875,
      "completions/mean_terminated_length": 255.419921875,
      "completions/min_length": 119.125,
      "completions/min_terminated_length": 119.125,
      "epoch": 0.13052631578947368,
      "grad_norm": 0.0289030522108078,
      "learning_rate": 9.979432471290472e-07,
      "loss": 0.0055,
      "num_tokens": 30342588.0,
      "reward": 1.9491963237524033,
      "reward_std": 0.4529041275382042,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.802734375,
      "rewards/stop_prediction_reward/std": 0.3396564405411482,
      "rewards/waypoint_pred_accuracy/mean": 0.07518408738549646,
      "rewards/waypoint_pred_accuracy/std": 0.09874280986073203,
      "step": 62
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 485.625,
      "completions/max_terminated_length": 485.625,
      "completions/mean_length": 243.8125,
      "completions/mean_terminated_length": 243.8125,
      "completions/min_length": 104.25,
      "completions/min_terminated_length": 104.25,
      "epoch": 0.13263157894736843,
      "grad_norm": 0.02499276027083397,
      "learning_rate": 9.97614944026565e-07,
      "loss": -0.0005,
      "num_tokens": 30829532.0,
      "reward": 1.6152346730232239,
      "reward_std": 0.3560887239873409,
      "rewards/format_reward_embodied/mean": 0.9921875,
      "rewards/format_reward_embodied/std": 0.043842025101184845,
      "rewards/stop_prediction_reward/mean": 0.623046875,
      "rewards/stop_prediction_reward/std": 0.3392041679471731,
      "rewards/waypoint_pred_accuracy/mean": 1.5068036063656496e-07,
      "rewards/waypoint_pred_accuracy/std": 1.1848213716913133e-06,
      "step": 63
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 487.75,
      "completions/max_terminated_length": 487.75,
      "completions/mean_length": 242.8671875,
      "completions/mean_terminated_length": 242.8671875,
      "completions/min_length": 116.875,
      "completions/min_terminated_length": 116.875,
      "epoch": 0.13473684210526315,
      "grad_norm": 0.08438508957624435,
      "learning_rate": 9.97262411303488e-07,
      "loss": -0.0008,
      "num_tokens": 31316376.0,
      "reward": 1.676452487707138,
      "reward_std": 0.38882749900221825,
      "rewards/format_reward_embodied/mean": 0.994140625,
      "rewards/format_reward_embodied/std": 0.046875,
      "rewards/stop_prediction_reward/mean": 0.654296875,
      "rewards/stop_prediction_reward/std": 0.3573997803032398,
      "rewards/waypoint_pred_accuracy/mean": 0.014007493944973248,
      "rewards/waypoint_pred_accuracy/std": 0.033498459750262555,
      "step": 64
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 715.75,
      "completions/max_terminated_length": 715.75,
      "completions/mean_length": 239.49609375,
      "completions/mean_terminated_length": 239.49609375,
      "completions/min_length": 118.25,
      "completions/min_terminated_length": 118.25,
      "epoch": 0.1368421052631579,
      "grad_norm": 0.027747681364417076,
      "learning_rate": 9.968856680425886e-07,
      "loss": 0.0053,
      "num_tokens": 31798614.0,
      "reward": 1.7896013855934143,
      "reward_std": 0.33628559671342373,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.7890625,
      "rewards/stop_prediction_reward/std": 0.3223333489149809,
      "rewards/waypoint_pred_accuracy/mean": 0.0022225715887884694,
      "rewards/waypoint_pred_accuracy/std": 0.00479067146991952,
      "step": 65
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 639.75,
      "completions/max_terminated_length": 639.75,
      "completions/mean_length": 251.96875,
      "completions/mean_terminated_length": 251.96875,
      "completions/min_length": 115.625,
      "completions/min_terminated_length": 115.625,
      "epoch": 0.13894736842105262,
      "grad_norm": 0.028032371774315834,
      "learning_rate": 9.964847346371676e-07,
      "loss": 0.004,
      "num_tokens": 32286790.0,
      "reward": 1.8506833761930466,
      "reward_std": 0.3642146345227957,
      "rewards/format_reward_embodied/mean": 0.994140625,
      "rewards/format_reward_embodied/std": 0.046875,
      "rewards/stop_prediction_reward/mean": 0.830078125,
      "rewards/stop_prediction_reward/std": 0.3009116370230913,
      "rewards/waypoint_pred_accuracy/mean": 0.013232328761660028,
      "rewards/waypoint_pred_accuracy/std": 0.03860976189025678,
      "step": 66
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 481.25,
      "completions/max_terminated_length": 481.25,
      "completions/mean_length": 247.291015625,
      "completions/mean_terminated_length": 247.7331371307373,
      "completions/min_length": 100.25,
      "completions/min_terminated_length": 115.125,
      "epoch": 0.14105263157894737,
      "grad_norm": 0.021038714796304703,
      "learning_rate": 9.96059632789951e-07,
      "loss": -0.0006,
      "num_tokens": 32774747.0,
      "reward": 1.5779267400503159,
      "reward_std": 0.42914118245244026,
      "rewards/format_reward_embodied/mean": 0.9921875,
      "rewards/format_reward_embodied/std": 0.05317101255059242,
      "rewards/stop_prediction_reward/mean": 0.505859375,
      "rewards/stop_prediction_reward/std": 0.3645100612193346,
      "rewards/waypoint_pred_accuracy/mean": 0.039939936966421924,
      "rewards/waypoint_pred_accuracy/std": 0.058027153559010024,
      "step": 67
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 748.875,
      "completions/max_terminated_length": 748.875,
      "completions/mean_length": 253.40234375,
      "completions/mean_terminated_length": 253.40234375,
      "completions/min_length": 119.875,
      "completions/min_terminated_length": 119.875,
      "epoch": 0.1431578947368421,
      "grad_norm": 0.024472616612911224,
      "learning_rate": 9.956103855119138e-07,
      "loss": 0.0036,
      "num_tokens": 33265961.0,
      "reward": 1.6555908024311066,
      "reward_std": 0.385839419439435,
      "rewards/format_reward_embodied/mean": 0.990234375,
      "rewards/format_reward_embodied/std": 0.078125,
      "rewards/stop_prediction_reward/mean": 0.6640625,
      "rewards/stop_prediction_reward/std": 0.36797660402953625,
      "rewards/waypoint_pred_accuracy/mean": 0.0006469660570671656,
      "rewards/waypoint_pred_accuracy/std": 0.003268853310080662,
      "step": 68
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 496.375,
      "completions/max_terminated_length": 496.375,
      "completions/mean_length": 253.37109375,
      "completions/mean_terminated_length": 253.37109375,
      "completions/min_length": 122.0,
      "completions/min_terminated_length": 122.0,
      "epoch": 0.14526315789473684,
      "grad_norm": 0.038161348551511765,
      "learning_rate": 9.951370171210359e-07,
      "loss": 0.002,
      "num_tokens": 33757543.0,
      "reward": 1.755501314997673,
      "reward_std": 0.34114088118076324,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.32307766377925873,
      "rewards/waypoint_pred_accuracy/mean": 0.004703786010358479,
      "rewards/waypoint_pred_accuracy/std": 0.016449308837348298,
      "step": 69
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 833.25,
      "completions/max_terminated_length": 833.25,
      "completions/mean_length": 262.56640625,
      "completions/mean_terminated_length": 262.56640625,
      "completions/min_length": 109.875,
      "completions/min_terminated_length": 109.875,
      "epoch": 0.14736842105263157,
      "grad_norm": 0.024708310142159462,
      "learning_rate": 9.946395532409847e-07,
      "loss": 0.0085,
      "num_tokens": 34253513.0,
      "reward": 1.7724248170852661,
      "reward_std": 0.354646734893322,
      "rewards/format_reward_embodied/mean": 0.98828125,
      "rewards/format_reward_embodied/std": 0.06417626701295376,
      "rewards/stop_prediction_reward/mean": 0.7734375,
      "rewards/stop_prediction_reward/std": 0.3255934212356806,
      "rewards/waypoint_pred_accuracy/mean": 0.005353036525642015,
      "rewards/waypoint_pred_accuracy/std": 0.013709596910865714,
      "step": 70
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 497.25,
      "completions/max_terminated_length": 497.25,
      "completions/mean_length": 261.169921875,
      "completions/mean_terminated_length": 261.169921875,
      "completions/min_length": 127.25,
      "completions/min_terminated_length": 127.25,
      "epoch": 0.14947368421052631,
      "grad_norm": 0.02306721918284893,
      "learning_rate": 9.941180207997288e-07,
      "loss": -0.0001,
      "num_tokens": 34747616.0,
      "reward": 1.8229519128799438,
      "reward_std": 0.33919697254896164,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.81640625,
      "rewards/stop_prediction_reward/std": 0.3285626471042633,
      "rewards/waypoint_pred_accuracy/mean": 0.003272829200625438,
      "rewards/waypoint_pred_accuracy/std": 0.01800350467010503,
      "step": 71
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 454.0,
      "completions/max_terminated_length": 454.0,
      "completions/mean_length": 244.701171875,
      "completions/mean_terminated_length": 244.701171875,
      "completions/min_length": 120.0,
      "completions/min_terminated_length": 120.0,
      "epoch": 0.15157894736842106,
      "grad_norm": 0.03545621410012245,
      "learning_rate": 9.935724480280795e-07,
      "loss": 0.0,
      "num_tokens": 35235719.0,
      "reward": 1.7902133017778397,
      "reward_std": 0.45963282138109207,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.6796875,
      "rewards/stop_prediction_reward/std": 0.38505756109952927,
      "rewards/waypoint_pred_accuracy/mean": 0.056239478069983306,
      "rewards/waypoint_pred_accuracy/std": 0.08321574779984076,
      "step": 72
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 423.125,
      "completions/max_terminated_length": 423.125,
      "completions/mean_length": 241.056640625,
      "completions/mean_terminated_length": 241.056640625,
      "completions/min_length": 115.875,
      "completions/min_terminated_length": 115.875,
      "epoch": 0.15368421052631578,
      "grad_norm": 0.025743646547198296,
      "learning_rate": 9.93002864458164e-07,
      "loss": 0.0017,
      "num_tokens": 35717284.0,
      "reward": 1.7117034643888474,
      "reward_std": 0.34931979328393936,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.7109375,
      "rewards/stop_prediction_reward/std": 0.3488571159541607,
      "rewards/waypoint_pred_accuracy/mean": 0.00038298743026394556,
      "rewards/waypoint_pred_accuracy/std": 0.0022788381146801144,
      "step": 73
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 484.375,
      "completions/max_terminated_length": 484.375,
      "completions/mean_length": 236.037109375,
      "completions/mean_terminated_length": 236.037109375,
      "completions/min_length": 110.375,
      "completions/min_terminated_length": 110.375,
      "epoch": 0.15578947368421053,
      "grad_norm": 0.031387291848659515,
      "learning_rate": 9.924093009218252e-07,
      "loss": -0.0001,
      "num_tokens": 36196791.0,
      "reward": 1.53607939183712,
      "reward_std": 0.3938233330845833,
      "rewards/format_reward_embodied/mean": 0.990234375,
      "rewards/format_reward_embodied/std": 0.04855126701295376,
      "rewards/stop_prediction_reward/mean": 0.515625,
      "rewards/stop_prediction_reward/std": 0.35425616055727005,
      "rewards/waypoint_pred_accuracy/mean": 0.01511000881408156,
      "rewards/waypoint_pred_accuracy/std": 0.024702310350773107,
      "step": 74
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 682.875,
      "completions/max_terminated_length": 682.875,
      "completions/mean_length": 246.58203125,
      "completions/mean_terminated_length": 247.04151344299316,
      "completions/min_length": 102.375,
      "completions/min_terminated_length": 120.375,
      "epoch": 0.15789473684210525,
      "grad_norm": 0.026448730379343033,
      "learning_rate": 9.917917895489542e-07,
      "loss": 0.0036,
      "num_tokens": 36683937.0,
      "reward": 1.7273263335227966,
      "reward_std": 0.3566751927137375,
      "rewards/format_reward_embodied/mean": 0.98046875,
      "rewards/format_reward_embodied/std": 0.12826303765177727,
      "rewards/stop_prediction_reward/mean": 0.7109375,
      "rewards/stop_prediction_reward/std": 0.2865038700401783,
      "rewards/waypoint_pred_accuracy/mean": 0.01796005329746431,
      "rewards/waypoint_pred_accuracy/std": 0.03300400403918848,
      "step": 75
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 466.5,
      "completions/max_terminated_length": 466.5,
      "completions/mean_length": 246.384765625,
      "completions/mean_terminated_length": 246.384765625,
      "completions/min_length": 122.625,
      "completions/min_terminated_length": 122.625,
      "epoch": 0.16,
      "grad_norm": 0.026203418150544167,
      "learning_rate": 9.9115036376575e-07,
      "loss": 0.0007,
      "num_tokens": 37170854.0,
      "reward": 1.4314483553171158,
      "reward_std": 0.3206907380372286,
      "rewards/format_reward_embodied/mean": 0.994140625,
      "rewards/format_reward_embodied/std": 0.046875,
      "rewards/stop_prediction_reward/mean": 0.435546875,
      "rewards/stop_prediction_reward/std": 0.31111637130379677,
      "rewards/waypoint_pred_accuracy/mean": 0.0008804261656223389,
      "rewards/waypoint_pred_accuracy/std": 0.007043409327252448,
      "step": 76
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 461.375,
      "completions/max_terminated_length": 461.375,
      "completions/mean_length": 245.017578125,
      "completions/mean_terminated_length": 245.51959609985352,
      "completions/min_length": 97.0,
      "completions/min_terminated_length": 110.875,
      "epoch": 0.16210526315789472,
      "grad_norm": 0.021599190309643745,
      "learning_rate": 9.904850582929109e-07,
      "loss": -0.001,
      "num_tokens": 37659375.0,
      "reward": 1.7587501555681229,
      "reward_std": 0.45513000525534153,
      "rewards/format_reward_embodied/mean": 0.9765625,
      "rewards/format_reward_embodied/std": 0.12835253402590752,
      "rewards/stop_prediction_reward/mean": 0.6328125,
      "rewards/stop_prediction_reward/std": 0.3041498549282551,
      "rewards/waypoint_pred_accuracy/mean": 0.07468759280809832,
      "rewards/waypoint_pred_accuracy/std": 0.08736424083447313,
      "step": 77
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 431.375,
      "completions/max_terminated_length": 431.375,
      "completions/mean_length": 251.69921875,
      "completions/mean_terminated_length": 251.69921875,
      "completions/min_length": 118.125,
      "completions/min_terminated_length": 118.125,
      "epoch": 0.16421052631578947,
      "grad_norm": 0.023476749658584595,
      "learning_rate": 9.897959091437545e-07,
      "loss": -0.0017,
      "num_tokens": 38147157.0,
      "reward": 1.7629946172237396,
      "reward_std": 0.3216873835772276,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.72265625,
      "rewards/stop_prediction_reward/std": 0.2610730957239866,
      "rewards/waypoint_pred_accuracy/mean": 0.02212230950329659,
      "rewards/waypoint_pred_accuracy/std": 0.03588474957179061,
      "step": 78
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 464.875,
      "completions/max_terminated_length": 464.875,
      "completions/mean_length": 238.640625,
      "completions/mean_terminated_length": 238.640625,
      "completions/min_length": 118.375,
      "completions/min_terminated_length": 118.375,
      "epoch": 0.16631578947368422,
      "grad_norm": 0.03286973387002945,
      "learning_rate": 9.890829536222686e-07,
      "loss": -0.0025,
      "num_tokens": 38629981.0,
      "reward": 1.7297946512699127,
      "reward_std": 0.2887336425483227,
      "rewards/format_reward_embodied/mean": 0.974609375,
      "rewards/format_reward_embodied/std": 0.12744012847542763,
      "rewards/stop_prediction_reward/mean": 0.74609375,
      "rewards/stop_prediction_reward/std": 0.23198767006397247,
      "rewards/waypoint_pred_accuracy/mean": 0.004545772711879333,
      "rewards/waypoint_pred_accuracy/std": 0.008970102488097233,
      "step": 79
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 476.125,
      "completions/max_terminated_length": 476.125,
      "completions/mean_length": 233.671875,
      "completions/mean_terminated_length": 233.671875,
      "completions/min_length": 97.875,
      "completions/min_terminated_length": 97.875,
      "epoch": 0.16842105263157894,
      "grad_norm": 0.022502202540636063,
      "learning_rate": 9.88346230321092e-07,
      "loss": 0.0003,
      "num_tokens": 39111733.0,
      "reward": 1.6976664066314697,
      "reward_std": 0.36001696437597275,
      "rewards/format_reward_embodied/mean": 0.986328125,
      "rewards/format_reward_embodied/std": 0.09071702510118484,
      "rewards/stop_prediction_reward/mean": 0.68359375,
      "rewards/stop_prediction_reward/std": 0.3152890168130398,
      "rewards/waypoint_pred_accuracy/mean": 0.013872268769774525,
      "rewards/waypoint_pred_accuracy/std": 0.03006945856032351,
      "step": 80
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 428.875,
      "completions/max_terminated_length": 428.875,
      "completions/mean_length": 236.8515625,
      "completions/mean_terminated_length": 236.8515625,
      "completions/min_length": 110.125,
      "completions/min_terminated_length": 110.125,
      "epoch": 0.1705263157894737,
      "grad_norm": 0.03461969271302223,
      "learning_rate": 9.875857791194251e-07,
      "loss": 0.0001,
      "num_tokens": 39593449.0,
      "reward": 1.8557344675064087,
      "reward_std": 0.3654259257018566,
      "rewards/format_reward_embodied/mean": 0.9921875,
      "rewards/format_reward_embodied/std": 0.05317101255059242,
      "rewards/stop_prediction_reward/mean": 0.798828125,
      "rewards/stop_prediction_reward/std": 0.2945959325879812,
      "rewards/waypoint_pred_accuracy/mean": 0.032359407392959616,
      "rewards/waypoint_pred_accuracy/std": 0.051165802276623415,
      "step": 81
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 773.75,
      "completions/max_terminated_length": 773.75,
      "completions/mean_length": 242.849609375,
      "completions/mean_terminated_length": 242.849609375,
      "completions/min_length": 111.625,
      "completions/min_terminated_length": 111.625,
      "epoch": 0.1726315789473684,
      "grad_norm": 0.025375094264745712,
      "learning_rate": 9.868016411808711e-07,
      "loss": 0.008,
      "num_tokens": 40080732.0,
      "reward": 1.8185276985168457,
      "reward_std": 0.2778010666370392,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.822265625,
      "rewards/stop_prediction_reward/std": 0.26144965551793575,
      "rewards/waypoint_pred_accuracy/mean": 8.415357677159232e-05,
      "rewards/waypoint_pred_accuracy/std": 0.0006628100762255675,
      "step": 82
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 466.0,
      "completions/max_terminated_length": 466.0,
      "completions/mean_length": 239.771484375,
      "completions/mean_terminated_length": 239.771484375,
      "completions/min_length": 118.125,
      "completions/min_terminated_length": 118.125,
      "epoch": 0.17473684210526316,
      "grad_norm": 0.02282722294330597,
      "learning_rate": 9.85993858951209e-07,
      "loss": -0.0,
      "num_tokens": 40564327.0,
      "reward": 1.8071236461400986,
      "reward_std": 0.3150872718542814,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.7109375,
      "rewards/stop_prediction_reward/std": 0.2674474287778139,
      "rewards/waypoint_pred_accuracy/mean": 0.04809306015794699,
      "rewards/waypoint_pred_accuracy/std": 0.0513845629028903,
      "step": 83
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 428.875,
      "completions/max_terminated_length": 428.875,
      "completions/mean_length": 237.19140625,
      "completions/mean_terminated_length": 237.19140625,
      "completions/min_length": 117.75,
      "completions/min_terminated_length": 117.75,
      "epoch": 0.17684210526315788,
      "grad_norm": 0.024264369159936905,
      "learning_rate": 9.851624761560941e-07,
      "loss": 0.0003,
      "num_tokens": 41047305.0,
      "reward": 1.6365228593349457,
      "reward_std": 0.2356659732758999,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.603515625,
      "rewards/stop_prediction_reward/std": 0.21942270919680595,
      "rewards/waypoint_pred_accuracy/mean": 0.016503638941257396,
      "rewards/waypoint_pred_accuracy/std": 0.019282840457430492,
      "step": 84
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 443.125,
      "completions/max_terminated_length": 443.125,
      "completions/mean_length": 241.103515625,
      "completions/mean_terminated_length": 241.103515625,
      "completions/min_length": 119.125,
      "completions/min_terminated_length": 119.125,
      "epoch": 0.17894736842105263,
      "grad_norm": 0.020238297060132027,
      "learning_rate": 9.843075377986927e-07,
      "loss": 0.0003,
      "num_tokens": 41531134.0,
      "reward": 1.74451445043087,
      "reward_std": 0.27907660976052284,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.693359375,
      "rewards/stop_prediction_reward/std": 0.22875236719846725,
      "rewards/waypoint_pred_accuracy/mean": 0.025577549161396262,
      "rewards/waypoint_pred_accuracy/std": 0.03110517306985683,
      "step": 85
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 419.5,
      "completions/max_terminated_length": 419.5,
      "completions/mean_length": 231.287109375,
      "completions/mean_terminated_length": 231.287109375,
      "completions/min_length": 118.875,
      "completions/min_terminated_length": 118.875,
      "epoch": 0.18105263157894738,
      "grad_norm": 0.02773391455411911,
      "learning_rate": 9.834290901572454e-07,
      "loss": 0.0005,
      "num_tokens": 42010385.0,
      "reward": 1.8274007737636566,
      "reward_std": 0.2518170941621065,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.822265625,
      "rewards/stop_prediction_reward/std": 0.2328737936913967,
      "rewards/waypoint_pred_accuracy/mean": 0.0035441384432527657,
      "rewards/waypoint_pred_accuracy/std": 0.011852368814229316,
      "step": 86
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 486.25,
      "completions/max_terminated_length": 486.25,
      "completions/mean_length": 242.701171875,
      "completions/mean_terminated_length": 242.701171875,
      "completions/min_length": 112.125,
      "completions/min_terminated_length": 112.125,
      "epoch": 0.1831578947368421,
      "grad_norm": 0.017116429284214973,
      "learning_rate": 9.82527180782562e-07,
      "loss": 0.0007,
      "num_tokens": 42495544.0,
      "reward": 1.7215514183044434,
      "reward_std": 0.24272998422384262,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.71875,
      "rewards/stop_prediction_reward/std": 0.22877886332571507,
      "rewards/waypoint_pred_accuracy/mean": 0.0033538464465563867,
      "rewards/waypoint_pred_accuracy/std": 0.008977477041369256,
      "step": 87
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 407.625,
      "completions/max_terminated_length": 407.625,
      "completions/mean_length": 227.666015625,
      "completions/mean_terminated_length": 227.666015625,
      "completions/min_length": 124.125,
      "completions/min_terminated_length": 124.125,
      "epoch": 0.18526315789473685,
      "grad_norm": 0.025679390877485275,
      "learning_rate": 9.816018584954474e-07,
      "loss": -0.0001,
      "num_tokens": 42972109.0,
      "reward": 1.543754830956459,
      "reward_std": 0.3515052441507578,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.423828125,
      "rewards/stop_prediction_reward/std": 0.25312468968331814,
      "rewards/waypoint_pred_accuracy/mean": 0.06093992558778956,
      "rewards/waypoint_pred_accuracy/std": 0.090891926485829,
      "step": 88
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 424.0,
      "completions/max_terminated_length": 424.0,
      "completions/mean_length": 229.5078125,
      "completions/mean_terminated_length": 229.5078125,
      "completions/min_length": 114.375,
      "completions/min_terminated_length": 114.375,
      "epoch": 0.18736842105263157,
      "grad_norm": 0.025896085426211357,
      "learning_rate": 9.806531733840594e-07,
      "loss": 0.0009,
      "num_tokens": 43453905.0,
      "reward": 1.9336253255605698,
      "reward_std": 0.3347325623035431,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.82421875,
      "rewards/stop_prediction_reward/std": 0.24601666443049908,
      "rewards/waypoint_pred_accuracy/mean": 0.05470328652882017,
      "rewards/waypoint_pred_accuracy/std": 0.07181658712215722,
      "step": 89
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 432.375,
      "completions/max_terminated_length": 432.375,
      "completions/mean_length": 235.38671875,
      "completions/mean_terminated_length": 235.38671875,
      "completions/min_length": 115.875,
      "completions/min_terminated_length": 115.875,
      "epoch": 0.18947368421052632,
      "grad_norm": 0.03356796130537987,
      "learning_rate": 9.796811768011975e-07,
      "loss": -0.0006,
      "num_tokens": 43934935.0,
      "reward": 1.7414738535881042,
      "reward_std": 0.30686922930181026,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.607421875,
      "rewards/stop_prediction_reward/std": 0.2297498807311058,
      "rewards/waypoint_pred_accuracy/mean": 0.06702600460093983,
      "rewards/waypoint_pred_accuracy/std": 0.05166525034704193,
      "step": 90
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 406.25,
      "completions/max_terminated_length": 406.25,
      "completions/mean_length": 222.359375,
      "completions/mean_terminated_length": 222.359375,
      "completions/min_length": 118.875,
      "completions/min_terminated_length": 118.875,
      "epoch": 0.19157894736842104,
      "grad_norm": 0.01943252608180046,
      "learning_rate": 9.78685921361522e-07,
      "loss": -0.0005,
      "num_tokens": 44409551.0,
      "reward": 1.6487830728292465,
      "reward_std": 0.30546887032687664,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.58203125,
      "rewards/stop_prediction_reward/std": 0.2547878734767437,
      "rewards/waypoint_pred_accuracy/mean": 0.03337590532140797,
      "rewards/waypoint_pred_accuracy/std": 0.052433368229147945,
      "step": 91
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 385.375,
      "completions/max_terminated_length": 385.375,
      "completions/mean_length": 217.119140625,
      "completions/mean_terminated_length": 217.119140625,
      "completions/min_length": 113.5,
      "completions/min_terminated_length": 113.5,
      "epoch": 0.1936842105263158,
      "grad_norm": 0.021023401990532875,
      "learning_rate": 9.776674609387076e-07,
      "loss": -0.0006,
      "num_tokens": 44880844.0,
      "reward": 1.756135731935501,
      "reward_std": 0.264364130795002,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.7109375,
      "rewards/stop_prediction_reward/std": 0.23147857002913952,
      "rewards/waypoint_pred_accuracy/mean": 0.022599116048866108,
      "rewards/waypoint_pred_accuracy/std": 0.03211659606313333,
      "step": 92
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 400.125,
      "completions/max_terminated_length": 400.125,
      "completions/mean_length": 218.37109375,
      "completions/mean_terminated_length": 218.37109375,
      "completions/min_length": 107.0,
      "completions/min_terminated_length": 107.0,
      "epoch": 0.1957894736842105,
      "grad_norm": 0.020804792642593384,
      "learning_rate": 9.766258506625257e-07,
      "loss": 0.0009,
      "num_tokens": 45354762.0,
      "reward": 1.703179121017456,
      "reward_std": 0.31406174413859844,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.623046875,
      "rewards/stop_prediction_reward/std": 0.2401380892843008,
      "rewards/waypoint_pred_accuracy/mean": 0.04104270155312406,
      "rewards/waypoint_pred_accuracy/std": 0.047771165754966205,
      "step": 93
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 385.125,
      "completions/max_terminated_length": 385.125,
      "completions/mean_length": 220.767578125,
      "completions/mean_terminated_length": 220.767578125,
      "completions/min_length": 108.25,
      "completions/min_terminated_length": 108.25,
      "epoch": 0.19789473684210526,
      "grad_norm": 0.05812095105648041,
      "learning_rate": 9.75561146915861e-07,
      "loss": 0.0007,
      "num_tokens": 45828691.0,
      "reward": 1.7804509848356247,
      "reward_std": 0.24969635531306267,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.728515625,
      "rewards/stop_prediction_reward/std": 0.2045249417424202,
      "rewards/waypoint_pred_accuracy/mean": 0.02596768177553266,
      "rewards/waypoint_pred_accuracy/std": 0.03161624588427525,
      "step": 94
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 406.125,
      "completions/max_terminated_length": 406.125,
      "completions/mean_length": 221.861328125,
      "completions/mean_terminated_length": 221.861328125,
      "completions/min_length": 118.75,
      "completions/min_terminated_length": 118.75,
      "epoch": 0.2,
      "grad_norm": 0.019957855343818665,
      "learning_rate": 9.744734073316595e-07,
      "loss": -0.0003,
      "num_tokens": 46303244.0,
      "reward": 1.6937783360481262,
      "reward_std": 0.18066157400608063,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.6171875,
      "rewards/stop_prediction_reward/std": 0.12835253402590752,
      "rewards/waypoint_pred_accuracy/mean": 0.03829541802406311,
      "rewards/waypoint_pred_accuracy/std": 0.038537144660949735,
      "step": 95
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 407.5,
      "completions/max_terminated_length": 407.5,
      "completions/mean_length": 213.76953125,
      "completions/mean_terminated_length": 213.76953125,
      "completions/min_length": 101.625,
      "completions/min_terminated_length": 101.625,
      "epoch": 0.20210526315789473,
      "grad_norm": 0.02102278172969818,
      "learning_rate": 9.73362690789808e-07,
      "loss": -0.001,
      "num_tokens": 46770710.0,
      "reward": 1.7461132854223251,
      "reward_std": 0.28678105026483536,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.640625,
      "rewards/stop_prediction_reward/std": 0.21690494194626808,
      "rewards/waypoint_pred_accuracy/mean": 0.05372073073522188,
      "rewards/waypoint_pred_accuracy/std": 0.054212798771914095,
      "step": 96
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 457.0,
      "completions/max_terminated_length": 457.0,
      "completions/mean_length": 219.494140625,
      "completions/mean_terminated_length": 219.494140625,
      "completions/min_length": 107.75,
      "completions/min_terminated_length": 107.75,
      "epoch": 0.20421052631578948,
      "grad_norm": 0.023958692327141762,
      "learning_rate": 9.722290574139486e-07,
      "loss": -0.0001,
      "num_tokens": 47243155.0,
      "reward": 1.747300535440445,
      "reward_std": 0.17760824598371983,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.7421875,
      "rewards/stop_prediction_reward/std": 0.15745450742542744,
      "rewards/waypoint_pred_accuracy/mean": 0.0035330769751453772,
      "rewards/waypoint_pred_accuracy/std": 0.017783919582143426,
      "step": 97
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 380.375,
      "completions/max_terminated_length": 380.375,
      "completions/mean_length": 215.083984375,
      "completions/mean_terminated_length": 215.083984375,
      "completions/min_length": 109.125,
      "completions/min_terminated_length": 109.125,
      "epoch": 0.2063157894736842,
      "grad_norm": 0.016943305730819702,
      "learning_rate": 9.71072568568222e-07,
      "loss": -0.0005,
      "num_tokens": 47713406.0,
      "reward": 1.851816326379776,
      "reward_std": 0.2104954868555069,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.849609375,
      "rewards/stop_prediction_reward/std": 0.1879090555012226,
      "rewards/waypoint_pred_accuracy/mean": 0.0020800431666430333,
      "rewards/waypoint_pred_accuracy/std": 0.011219009378692333,
      "step": 98
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 392.625,
      "completions/max_terminated_length": 392.625,
      "completions/mean_length": 211.724609375,
      "completions/mean_terminated_length": 211.724609375,
      "completions/min_length": 104.125,
      "completions/min_terminated_length": 104.125,
      "epoch": 0.20842105263157895,
      "grad_norm": 0.014324544928967953,
      "learning_rate": 9.698932868539475e-07,
      "loss": -0.0006,
      "num_tokens": 48180849.0,
      "reward": 1.6212971061468124,
      "reward_std": 0.16710768891550742,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.498046875,
      "rewards/stop_prediction_reward/std": 0.07980126701295376,
      "rewards/waypoint_pred_accuracy/mean": 0.06260167788853366,
      "rewards/waypoint_pred_accuracy/std": 0.04448781939231594,
      "step": 99
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 0.016340678557753563,
      "learning_rate": 9.686912761062337e-07,
      "loss": -0.001,
      "step": 100
    },
    {
      "epoch": 0.21052631578947367,
      "eval_clip_ratio": 0.0,
      "eval_completions/clipped_ratio": 0.00015625,
      "eval_completions/max_length": 457.89,
      "eval_completions/max_terminated_length": 457.89,
      "eval_completions/mean_length": 219.56675354003906,
      "eval_completions/mean_terminated_length": 219.59981658935547,
      "eval_completions/min_length": 112.23,
      "eval_completions/min_terminated_length": 113.11,
      "eval_loss": 0.0013147207209840417,
      "eval_num_tokens": 48647110.0,
      "eval_reward": 1.8446430933475495,
      "eval_reward_std": 0.20415274247365858,
      "eval_rewards/format_reward_embodied/mean": 0.99921875,
      "eval_rewards/format_reward_embodied/std": 0.00625,
      "eval_rewards/stop_prediction_reward/mean": 0.75734375,
      "eval_rewards/stop_prediction_reward/std": 0.1349847713112831,
      "eval_rewards/waypoint_pred_accuracy/mean": 0.04404030321765386,
      "eval_rewards/waypoint_pred_accuracy/std": 0.04408785154897764,
      "eval_runtime": 1143.7465,
      "eval_samples_per_second": 0.087,
      "eval_steps_per_second": 0.002,
      "step": 100
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 387.1875,
      "completions/max_terminated_length": 387.1875,
      "completions/mean_length": 211.564453125,
      "completions/mean_terminated_length": 211.564453125,
      "completions/min_length": 109.25,
      "completions/min_terminated_length": 109.25,
      "epoch": 0.21263157894736842,
      "grad_norm": 7.573169568786398e-05,
      "learning_rate": 9.674666013905223e-07,
      "loss": 0.0,
      "num_tokens": 49119859.0,
      "reward": 1.8795468658208847,
      "reward_std": 0.13644913337626896,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.8642578125,
      "rewards/stop_prediction_reward/std": 0.10762263275682926,
      "rewards/waypoint_pred_accuracy/mean": 0.007644527649440409,
      "rewards/waypoint_pred_accuracy/std": 0.02138127497060599,
      "step": 101
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 423.625,
      "completions/max_terminated_length": 423.625,
      "completions/mean_length": 216.150390625,
      "completions/mean_terminated_length": 216.150390625,
      "completions/min_length": 112.875,
      "completions/min_terminated_length": 112.875,
      "epoch": 0.21473684210526317,
      "grad_norm": 0.02024409919977188,
      "learning_rate": 9.662193289990683e-07,
      "loss": 0.0001,
      "num_tokens": 49589696.0,
      "reward": 1.9582752585411072,
      "reward_std": 0.170343800484261,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.873046875,
      "rewards/stop_prediction_reward/std": 0.059467025101184845,
      "rewards/waypoint_pred_accuracy/mean": 0.04359079086862039,
      "rewards/waypoint_pred_accuracy/std": 0.05229037126991898,
      "step": 102
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 419.875,
      "completions/max_terminated_length": 419.875,
      "completions/mean_length": 220.193359375,
      "completions/mean_terminated_length": 220.193359375,
      "completions/min_length": 110.5,
      "completions/min_terminated_length": 110.5,
      "epoch": 0.2168421052631579,
      "grad_norm": 0.020008524879813194,
      "learning_rate": 9.649495264473496e-07,
      "loss": -0.0008,
      "num_tokens": 50062179.0,
      "reward": 1.4985756427049637,
      "reward_std": 0.15706685557961464,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.5,
      "rewards/stop_prediction_reward/std": 0.15027354657649994,
      "rewards/waypoint_pred_accuracy/mean": 0.00026439113654092976,
      "rewards/waypoint_pred_accuracy/std": 0.0021146056694039514,
      "step": 103
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 389.375,
      "completions/max_terminated_length": 389.375,
      "completions/mean_length": 224.572265625,
      "completions/mean_terminated_length": 224.572265625,
      "completions/min_length": 113.5,
      "completions/min_terminated_length": 113.5,
      "epoch": 0.21894736842105264,
      "grad_norm": 0.010618757456541061,
      "learning_rate": 9.636572624704126e-07,
      "loss": 0.0008,
      "num_tokens": 50537928.0,
      "reward": 2.029994383454323,
      "reward_std": 0.16062275879085064,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.98046875,
      "rewards/stop_prediction_reward/std": 0.11734727956354618,
      "rewards/waypoint_pred_accuracy/mean": 0.02476281741601838,
      "rewards/waypoint_pred_accuracy/std": 0.035420115480633285,
      "step": 104
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 375.75,
      "completions/max_terminated_length": 375.75,
      "completions/mean_length": 208.478515625,
      "completions/mean_terminated_length": 208.478515625,
      "completions/min_length": 112.875,
      "completions/min_terminated_length": 112.875,
      "epoch": 0.22105263157894736,
      "grad_norm": 0.0,
      "learning_rate": 9.62342607019152e-07,
      "loss": 0.0003,
      "num_tokens": 51004477.0,
      "reward": 1.7948767840862274,
      "reward_std": 0.0999540267221164,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.734375,
      "rewards/stop_prediction_reward/std": 0.0729166679084301,
      "rewards/waypoint_pred_accuracy/mean": 0.03025089303362355,
      "rewards/waypoint_pred_accuracy/std": 0.024821761748978988,
      "step": 105
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 391.25,
      "completions/max_terminated_length": 391.25,
      "completions/mean_length": 218.388671875,
      "completions/mean_terminated_length": 218.388671875,
      "completions/min_length": 113.375,
      "completions/min_terminated_length": 113.375,
      "epoch": 0.2231578947368421,
      "grad_norm": 0.024067817255854607,
      "learning_rate": 9.610056312565245e-07,
      "loss": 0.0004,
      "num_tokens": 51477636.0,
      "reward": 2.026669681072235,
      "reward_std": 0.1696237076412217,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.8671875,
      "rewards/stop_prediction_reward/std": 0.09542626701295376,
      "rewards/waypoint_pred_accuracy/mean": 0.07974108902908483,
      "rewards/waypoint_pred_accuracy/std": 0.04533190939855558,
      "step": 106
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 420.75,
      "completions/max_terminated_length": 420.75,
      "completions/mean_length": 217.77734375,
      "completions/mean_terminated_length": 217.77734375,
      "completions/min_length": 119.0,
      "completions/min_terminated_length": 119.0,
      "epoch": 0.22526315789473683,
      "grad_norm": 0.013704587705433369,
      "learning_rate": 9.596464075536963e-07,
      "loss": 0.0003,
      "num_tokens": 51951186.0,
      "reward": 1.8652345538139343,
      "reward_std": 0.09080793828513833,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.865234375,
      "rewards/stop_prediction_reward/std": 0.0908065214753151,
      "rewards/waypoint_pred_accuracy/mean": 9.960914946921718e-08,
      "rewards/waypoint_pred_accuracy/std": 7.087302705995463e-07,
      "step": 107
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 403.875,
      "completions/max_terminated_length": 403.875,
      "completions/mean_length": 217.728515625,
      "completions/mean_terminated_length": 218.15721321105957,
      "completions/min_length": 100.25,
      "completions/min_terminated_length": 116.625,
      "epoch": 0.22736842105263158,
      "grad_norm": 0.02499217353761196,
      "learning_rate": 9.582650094861256e-07,
      "loss": -0.0001,
      "num_tokens": 52426055.0,
      "reward": 1.8535159230232239,
      "reward_std": 0.1513118724361675,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.857421875,
      "rewards/stop_prediction_reward/std": 0.1338059287518263,
      "rewards/waypoint_pred_accuracy/mean": 1.5656563736666032e-07,
      "rewards/waypoint_pred_accuracy/std": 8.466679199507776e-07,
      "step": 108
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 437.0,
      "completions/max_terminated_length": 437.0,
      "completions/mean_length": 217.58984375,
      "completions/mean_terminated_length": 217.58984375,
      "completions/min_length": 108.5,
      "completions/min_terminated_length": 108.5,
      "epoch": 0.2294736842105263,
      "grad_norm": 0.018018925562500954,
      "learning_rate": 9.568615118295798e-07,
      "loss": 0.0002,
      "num_tokens": 52898613.0,
      "reward": 1.5074883997440338,
      "reward_std": 0.08622953689905444,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.505859375,
      "rewards/stop_prediction_reward/std": 0.07980126701295376,
      "rewards/waypoint_pred_accuracy/mean": 0.0008145241825951426,
      "rewards/waypoint_pred_accuracy/std": 0.003929519949304638,
      "step": 109
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 410.625,
      "completions/max_terminated_length": 410.625,
      "completions/mean_length": 220.09375,
      "completions/mean_terminated_length": 220.09375,
      "completions/min_length": 112.5,
      "completions/min_terminated_length": 112.5,
      "epoch": 0.23157894736842105,
      "grad_norm": 0.026054121553897858,
      "learning_rate": 9.554359905560885e-07,
      "loss": 0.0009,
      "num_tokens": 53371493.0,
      "reward": 1.6272786408662796,
      "reward_std": 0.1345605030655861,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.607421875,
      "rewards/stop_prediction_reward/std": 0.10731646977365017,
      "rewards/waypoint_pred_accuracy/mean": 0.009928377814540968,
      "rewards/waypoint_pred_accuracy/std": 0.021277805810678764,
      "step": 110
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 451.5,
      "completions/max_terminated_length": 451.5,
      "completions/mean_length": 226.404296875,
      "completions/mean_terminated_length": 226.404296875,
      "completions/min_length": 104.75,
      "completions/min_terminated_length": 104.75,
      "epoch": 0.2336842105263158,
      "grad_norm": 0.019862722605466843,
      "learning_rate": 9.53988522829831e-07,
      "loss": -0.0002,
      "num_tokens": 53849844.0,
      "reward": 1.7356369495391846,
      "reward_std": 0.17475299397483468,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.6328125,
      "rewards/stop_prediction_reward/std": 0.0929968785494566,
      "rewards/waypoint_pred_accuracy/mean": 0.05238880167820055,
      "rewards/waypoint_pred_accuracy/std": 0.03885504556402302,
      "step": 111
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 429.125,
      "completions/max_terminated_length": 429.125,
      "completions/mean_length": 228.794921875,
      "completions/mean_terminated_length": 228.794921875,
      "completions/min_length": 102.625,
      "completions/min_terminated_length": 102.625,
      "epoch": 0.23578947368421052,
      "grad_norm": 0.013491553254425526,
      "learning_rate": 9.52519187002958e-07,
      "loss": -0.0013,
      "num_tokens": 54326667.0,
      "reward": 2.0176089107990265,
      "reward_std": 0.09997917944565415,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.990234375,
      "rewards/stop_prediction_reward/std": 0.06879601255059242,
      "rewards/waypoint_pred_accuracy/mean": 0.013687264542047125,
      "rewards/waypoint_pred_accuracy/std": 0.02499246856624292,
      "step": 112
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 419.25,
      "completions/max_terminated_length": 419.25,
      "completions/mean_length": 223.84765625,
      "completions/mean_terminated_length": 223.84765625,
      "completions/min_length": 116.125,
      "completions/min_terminated_length": 116.125,
      "epoch": 0.23789473684210527,
      "grad_norm": 0.0,
      "learning_rate": 9.510280626113524e-07,
      "loss": 0.0002,
      "num_tokens": 54804477.0,
      "reward": 2.0416500568389893,
      "reward_std": 0.15113097801804543,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.87109375,
      "rewards/stop_prediction_reward/std": 0.03125,
      "rewards/waypoint_pred_accuracy/mean": 0.08527813665590386,
      "rewards/waypoint_pred_accuracy/std": 0.06661079078944249,
      "step": 113
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 418.375,
      "completions/max_terminated_length": 418.375,
      "completions/mean_length": 230.408203125,
      "completions/mean_terminated_length": 230.9030590057373,
      "completions/min_length": 102.375,
      "completions/min_terminated_length": 116.25,
      "epoch": 0.24,
      "grad_norm": 0.027175500988960266,
      "learning_rate": 9.495152303703225e-07,
      "loss": -0.0,
      "num_tokens": 55282254.0,
      "reward": 1.9152479320764542,
      "reward_std": 0.15890819625928998,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.869140625,
      "rewards/stop_prediction_reward/std": 0.08138803765177727,
      "rewards/waypoint_pred_accuracy/mean": 0.024030234897509217,
      "rewards/waypoint_pred_accuracy/std": 0.038782567949965596,
      "step": 114
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 379.5,
      "completions/max_terminated_length": 379.5,
      "completions/mean_length": 218.421875,
      "completions/mean_terminated_length": 218.421875,
      "completions/min_length": 108.625,
      "completions/min_terminated_length": 108.625,
      "epoch": 0.24210526315789474,
      "grad_norm": 0.0,
      "learning_rate": 9.479807721702337e-07,
      "loss": 0.0009,
      "num_tokens": 55753702.0,
      "reward": 1.964790791273117,
      "reward_std": 0.14700112864375114,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.869140625,
      "rewards/stop_prediction_reward/std": 0.046875,
      "rewards/waypoint_pred_accuracy/mean": 0.04880167031660676,
      "rewards/waypoint_pred_accuracy/std": 0.04630230367183685,
      "step": 115
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 418.75,
      "completions/max_terminated_length": 418.75,
      "completions/mean_length": 234.615234375,
      "completions/mean_terminated_length": 234.615234375,
      "completions/min_length": 107.25,
      "completions/min_terminated_length": 107.25,
      "epoch": 0.24421052631578946,
      "grad_norm": 0.01868380233645439,
      "learning_rate": 9.46424771072075e-07,
      "loss": -0.0016,
      "num_tokens": 56236001.0,
      "reward": 1.7634397000074387,
      "reward_std": 0.11784930247813463,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.744140625,
      "rewards/stop_prediction_reward/std": 0.08138803765177727,
      "rewards/waypoint_pred_accuracy/mean": 0.009649543033447117,
      "rewards/waypoint_pred_accuracy/std": 0.01907464297255501,
      "step": 116
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 393.5,
      "completions/max_terminated_length": 393.5,
      "completions/mean_length": 218.193359375,
      "completions/mean_terminated_length": 218.193359375,
      "completions/min_length": 111.75,
      "completions/min_terminated_length": 111.75,
      "epoch": 0.2463157894736842,
      "grad_norm": 0.013114881701767445,
      "learning_rate": 9.448473113029633e-07,
      "loss": 0.0009,
      "num_tokens": 56707844.0,
      "reward": 1.6348197907209396,
      "reward_std": 0.15209808605868602,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.513671875,
      "rewards/stop_prediction_reward/std": 0.08837713301181793,
      "rewards/waypoint_pred_accuracy/mean": 0.0605739434017778,
      "rewards/waypoint_pred_accuracy/std": 0.03603998847574985,
      "step": 117
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 385.625,
      "completions/max_terminated_length": 385.625,
      "completions/mean_length": 217.125,
      "completions/mean_terminated_length": 217.125,
      "completions/min_length": 115.125,
      "completions/min_terminated_length": 115.125,
      "epoch": 0.24842105263157896,
      "grad_norm": 0.021751945838332176,
      "learning_rate": 9.432484782515842e-07,
      "loss": 0.0011,
      "num_tokens": 57177540.0,
      "reward": 1.8951680064201355,
      "reward_std": 0.09731905919034034,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.87109375,
      "rewards/stop_prediction_reward/std": 0.05317101255059242,
      "rewards/waypoint_pred_accuracy/mean": 0.012037134467476562,
      "rewards/waypoint_pred_accuracy/std": 0.022074746749979113,
      "step": 118
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 733.875,
      "completions/max_terminated_length": 733.875,
      "completions/mean_length": 230.4921875,
      "completions/mean_terminated_length": 230.4921875,
      "completions/min_length": 114.125,
      "completions/min_terminated_length": 114.125,
      "epoch": 0.2505263157894737,
      "grad_norm": 0.017523573711514473,
      "learning_rate": 9.416283584635699e-07,
      "loss": 0.0073,
      "num_tokens": 57653632.0,
      "reward": 1.7819086909294128,
      "reward_std": 0.13167815032647923,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.74609375,
      "rewards/stop_prediction_reward/std": 0.0625,
      "rewards/waypoint_pred_accuracy/mean": 0.018884045333834365,
      "rewards/waypoint_pred_accuracy/std": 0.04270522284787148,
      "step": 119
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 383.75,
      "completions/max_terminated_length": 383.75,
      "completions/mean_length": 230.603515625,
      "completions/mean_terminated_length": 230.603515625,
      "completions/min_length": 109.25,
      "completions/min_terminated_length": 109.25,
      "epoch": 0.25263157894736843,
      "grad_norm": 0.020878760144114494,
      "learning_rate": 9.399870396368137e-07,
      "loss": -0.0003,
      "num_tokens": 58132533.0,
      "reward": 1.75615593791008,
      "reward_std": 0.08499195147633398,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.744140625,
      "rewards/stop_prediction_reward/std": 0.03754601255059242,
      "rewards/waypoint_pred_accuracy/mean": 0.006007667677934543,
      "rewards/waypoint_pred_accuracy/std": 0.023722964530922208,
      "step": 120
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 393.75,
      "completions/max_terminated_length": 393.75,
      "completions/mean_length": 228.326171875,
      "completions/mean_terminated_length": 228.326171875,
      "completions/min_length": 110.875,
      "completions/min_terminated_length": 110.875,
      "epoch": 0.25473684210526315,
      "grad_norm": 0.030050212517380714,
      "learning_rate": 9.383246106167244e-07,
      "loss": -0.0007,
      "num_tokens": 58610652.0,
      "reward": 2.005710780620575,
      "reward_std": 0.1616469284220443,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.8671875,
      "rewards/stop_prediction_reward/std": 0.05317101255059242,
      "rewards/waypoint_pred_accuracy/mean": 0.06926164017690084,
      "rewards/waypoint_pred_accuracy/std": 0.06174082592511354,
      "step": 121
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 411.75,
      "completions/max_terminated_length": 411.75,
      "completions/mean_length": 235.259765625,
      "completions/mean_terminated_length": 235.259765625,
      "completions/min_length": 119.75,
      "completions/min_terminated_length": 119.75,
      "epoch": 0.25684210526315787,
      "grad_norm": 0.05214720964431763,
      "learning_rate": 9.366411613914151e-07,
      "loss": 0.0,
      "num_tokens": 59091681.0,
      "reward": 2.0128368139266968,
      "reward_std": 0.041966003568632004,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.998046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.007394973285158812,
      "rewards/waypoint_pred_accuracy/std": 0.013188560594699084,
      "step": 122
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 738.375,
      "completions/max_terminated_length": 738.375,
      "completions/mean_length": 229.73828125,
      "completions/mean_terminated_length": 229.73828125,
      "completions/min_length": 113.125,
      "completions/min_terminated_length": 113.125,
      "epoch": 0.25894736842105265,
      "grad_norm": 0.020252572372555733,
      "learning_rate": 9.349367830868338e-07,
      "loss": 0.0065,
      "num_tokens": 59571867.0,
      "reward": 1.7482015490531921,
      "reward_std": 0.07813655398786068,
      "rewards/format_reward_embodied/mean": 0.994140625,
      "rewards/format_reward_embodied/std": 0.046875,
      "rewards/stop_prediction_reward/mean": 0.75390625,
      "rewards/stop_prediction_reward/std": 0.042255254462361336,
      "rewards/waypoint_pred_accuracy/mean": 7.734416431048885e-05,
      "rewards/waypoint_pred_accuracy/std": 0.00014155825192574412,
      "step": 123
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 421.5,
      "completions/max_terminated_length": 421.5,
      "completions/mean_length": 232.515625,
      "completions/mean_terminated_length": 232.515625,
      "completions/min_length": 116.625,
      "completions/min_terminated_length": 116.625,
      "epoch": 0.26105263157894737,
      "grad_norm": 0.012004735879600048,
      "learning_rate": 9.332115679618299e-07,
      "loss": -0.0002,
      "num_tokens": 60051875.0,
      "reward": 1.87109375,
      "reward_std": 0.06475212238729,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.873046875,
      "rewards/stop_prediction_reward/std": 0.059467025101184845,
      "rewards/waypoint_pred_accuracy/mean": 3.7687339174058405e-25,
      "rewards/waypoint_pred_accuracy/std": 0.0,
      "step": 124
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 373.375,
      "completions/max_terminated_length": 373.375,
      "completions/mean_length": 221.771484375,
      "completions/mean_terminated_length": 221.771484375,
      "completions/min_length": 113.625,
      "completions/min_terminated_length": 113.625,
      "epoch": 0.2631578947368421,
      "grad_norm": 0.01381534244865179,
      "learning_rate": 9.3146560940316e-07,
      "loss": -0.0001,
      "num_tokens": 60525038.0,
      "reward": 1.8031584918498993,
      "reward_std": 0.08486801406252198,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.74609375,
      "rewards/stop_prediction_reward/std": 0.021921012550592422,
      "rewards/waypoint_pred_accuracy/mean": 0.02853236788253366,
      "rewards/waypoint_pred_accuracy/std": 0.03149077049183591,
      "step": 125
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 379.125,
      "completions/max_terminated_length": 379.125,
      "completions/mean_length": 225.11328125,
      "completions/mean_terminated_length": 225.11328125,
      "completions/min_length": 106.375,
      "completions/min_terminated_length": 106.375,
      "epoch": 0.26526315789473687,
      "grad_norm": 0.0,
      "learning_rate": 9.296990019204335e-07,
      "loss": 0.0003,
      "num_tokens": 61001576.0,
      "reward": 1.838148683309555,
      "reward_std": 0.13138162437826395,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.03125,
      "rewards/waypoint_pred_accuracy/mean": 0.1065743277722504,
      "rewards/waypoint_pred_accuracy/std": 0.058355900342576206,
      "step": 126
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 596.5,
      "completions/max_terminated_length": 596.5,
      "completions/mean_length": 223.703125,
      "completions/mean_terminated_length": 223.703125,
      "completions/min_length": 122.375,
      "completions/min_terminated_length": 122.375,
      "epoch": 0.2673684210526316,
      "grad_norm": 0.015371584333479404,
      "learning_rate": 9.279118411409962e-07,
      "loss": 0.0076,
      "num_tokens": 61475152.0,
      "reward": 2.0289103388786316,
      "reward_std": 0.14714059105608612,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.986328125,
      "rewards/stop_prediction_reward/std": 0.05234810337424278,
      "rewards/waypoint_pred_accuracy/mean": 0.02226770008679324,
      "rewards/waypoint_pred_accuracy/std": 0.03958419876477137,
      "step": 127
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 389.0,
      "completions/max_terminated_length": 389.0,
      "completions/mean_length": 225.896484375,
      "completions/mean_terminated_length": 225.896484375,
      "completions/min_length": 109.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 0.2694736842105263,
      "grad_norm": 0.0,
      "learning_rate": 9.261042238047539e-07,
      "loss": 0.0003,
      "num_tokens": 61948507.0,
      "reward": 1.7672365009784698,
      "reward_std": 0.08540481339514372,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.07111826360536146,
      "rewards/waypoint_pred_accuracy/std": 0.04270240558859675,
      "step": 128
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 364.0,
      "completions/max_terminated_length": 364.0,
      "completions/mean_length": 229.087890625,
      "completions/mean_terminated_length": 229.087890625,
      "completions/min_length": 122.25,
      "completions/min_terminated_length": 122.25,
      "epoch": 0.27157894736842103,
      "grad_norm": 0.0,
      "learning_rate": 9.242762477589369e-07,
      "loss": -0.0008,
      "num_tokens": 62425864.0,
      "reward": 1.7555950731039047,
      "reward_std": 0.150370123796165,
      "rewards/format_reward_embodied/mean": 0.99609375,
      "rewards/format_reward_embodied/std": 0.03125,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.03125,
      "rewards/waypoint_pred_accuracy/mean": 0.06725065042865234,
      "rewards/waypoint_pred_accuracy/std": 0.060643474211165796,
      "step": 129
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 377.25,
      "completions/max_terminated_length": 377.25,
      "completions/mean_length": 218.873046875,
      "completions/mean_terminated_length": 219.36554527282715,
      "completions/min_length": 95.875,
      "completions/min_terminated_length": 110.5,
      "epoch": 0.2736842105263158,
      "grad_norm": 0.015500541776418686,
      "learning_rate": 9.224280119528013e-07,
      "loss": -0.0006,
      "num_tokens": 62900679.0,
      "reward": 1.86328125,
      "reward_std": 0.08442101255059242,
      "rewards/format_reward_embodied/mean": 0.994140625,
      "rewards/format_reward_embodied/std": 0.046875,
      "rewards/stop_prediction_reward/mean": 0.869140625,
      "rewards/stop_prediction_reward/std": 0.046875,
      "rewards/waypoint_pred_accuracy/mean": 5.195248799303594e-14,
      "rewards/waypoint_pred_accuracy/std": 3.9519006556770764e-13,
      "step": 130
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 389.375,
      "completions/max_terminated_length": 389.375,
      "completions/mean_length": 223.375,
      "completions/mean_terminated_length": 223.375,
      "completions/min_length": 116.375,
      "completions/min_terminated_length": 116.375,
      "epoch": 0.27578947368421053,
      "grad_norm": 0.0,
      "learning_rate": 9.205596164322753e-07,
      "loss": 0.0005,
      "num_tokens": 63375047.0,
      "reward": 1.8198472261428833,
      "reward_std": 0.06600932776927948,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.74609375,
      "rewards/stop_prediction_reward/std": 0.03125,
      "rewards/waypoint_pred_accuracy/mean": 0.03687673434615135,
      "rewards/waypoint_pred_accuracy/std": 0.023463066667318344,
      "step": 131
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 386.875,
      "completions/max_terminated_length": 386.875,
      "completions/mean_length": 223.994140625,
      "completions/mean_terminated_length": 223.994140625,
      "completions/min_length": 106.125,
      "completions/min_terminated_length": 106.125,
      "epoch": 0.27789473684210525,
      "grad_norm": 5.267659071250819e-05,
      "learning_rate": 9.186711623345419e-07,
      "loss": 0.0007,
      "num_tokens": 63848068.0,
      "reward": 1.8460404872894287,
      "reward_std": 0.07753154253146377,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.04802026903570478,
      "rewards/waypoint_pred_accuracy/std": 0.03876577728947339,
      "step": 132
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 370.625,
      "completions/max_terminated_length": 370.625,
      "completions/mean_length": 216.8359375,
      "completions/mean_terminated_length": 216.8359375,
      "completions/min_length": 113.25,
      "completions/min_terminated_length": 113.25,
      "epoch": 0.28,
      "grad_norm": 0.019873064011335373,
      "learning_rate": 9.167627518825651e-07,
      "loss": -0.0003,
      "num_tokens": 64319408.0,
      "reward": 1.66434708237648,
      "reward_std": 0.10717772238422185,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.619140625,
      "rewards/stop_prediction_reward/std": 0.046875,
      "rewards/waypoint_pred_accuracy/mean": 0.022603242181673977,
      "rewards/waypoint_pred_accuracy/std": 0.030156509747939708,
      "step": 133
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 394.75,
      "completions/max_terminated_length": 394.75,
      "completions/mean_length": 221.185546875,
      "completions/mean_terminated_length": 221.185546875,
      "completions/min_length": 105.75,
      "completions/min_terminated_length": 105.75,
      "epoch": 0.28210526315789475,
      "grad_norm": 0.016748478636145592,
      "learning_rate": 9.148344883795563e-07,
      "loss": 0.0002,
      "num_tokens": 64797519.0,
      "reward": 1.6543543934822083,
      "reward_std": 0.09143044333904982,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0625,
      "rewards/waypoint_pred_accuracy/mean": 0.014677208887757254,
      "rewards/waypoint_pred_accuracy/std": 0.02070214717129737,
      "step": 134
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 347.125,
      "completions/max_terminated_length": 347.125,
      "completions/mean_length": 207.6796875,
      "completions/mean_terminated_length": 207.6796875,
      "completions/min_length": 100.75,
      "completions/min_terminated_length": 100.75,
      "epoch": 0.28421052631578947,
      "grad_norm": 0.018281355500221252,
      "learning_rate": 9.128864762033824e-07,
      "loss": 0.0009,
      "num_tokens": 65264811.0,
      "reward": 1.8661604225635529,
      "reward_std": 0.13888043258339167,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.03754601255059242,
      "rewards/waypoint_pred_accuracy/mean": 0.05905676480875123,
      "rewards/waypoint_pred_accuracy/std": 0.053127349918608825,
      "step": 135
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 354.625,
      "completions/max_terminated_length": 354.625,
      "completions/mean_length": 213.609375,
      "completions/mean_terminated_length": 213.609375,
      "completions/min_length": 109.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 0.2863157894736842,
      "grad_norm": 0.01513614784926176,
      "learning_rate": 9.10918820800916e-07,
      "loss": -0.0,
      "num_tokens": 65734563.0,
      "reward": 1.7799546718597412,
      "reward_std": 0.05648380851107504,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.751953125,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.014000790968564693,
      "rewards/waypoint_pred_accuracy/std": 0.0204294033423741,
      "step": 136
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 359.5,
      "completions/max_terminated_length": 359.5,
      "completions/mean_length": 209.94140625,
      "completions/mean_terminated_length": 209.94140625,
      "completions/min_length": 112.625,
      "completions/min_terminated_length": 112.625,
      "epoch": 0.28842105263157897,
      "grad_norm": 0.014532121829688549,
      "learning_rate": 9.089316286823274e-07,
      "loss": -0.0006,
      "num_tokens": 66202821.0,
      "reward": 1.8316712975502014,
      "reward_std": 0.14127142806610715,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.042788804041257456,
      "rewards/waypoint_pred_accuracy/std": 0.05501071766701504,
      "step": 137
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 384.75,
      "completions/max_terminated_length": 384.75,
      "completions/mean_length": 226.625,
      "completions/mean_terminated_length": 226.625,
      "completions/min_length": 112.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.2905263157894737,
      "grad_norm": 0.013312633149325848,
      "learning_rate": 9.069250074153191e-07,
      "loss": -0.0001,
      "num_tokens": 66681989.0,
      "reward": 1.9901870042085648,
      "reward_std": 0.08008617826271802,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.98828125,
      "rewards/stop_prediction_reward/std": 0.07509202510118484,
      "rewards/waypoint_pred_accuracy/mean": 0.0009528863083687822,
      "rewards/waypoint_pred_accuracy/std": 0.007327620231080863,
      "step": 138
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 373.375,
      "completions/max_terminated_length": 373.375,
      "completions/mean_length": 206.775390625,
      "completions/mean_terminated_length": 206.775390625,
      "completions/min_length": 109.5,
      "completions/min_terminated_length": 109.5,
      "epoch": 0.2926315789473684,
      "grad_norm": 0.0,
      "learning_rate": 9.048990656193024e-07,
      "loss": 0.0,
      "num_tokens": 67151378.0,
      "reward": 1.8710939586162567,
      "reward_std": 0.03125098500925105,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.87109375,
      "rewards/stop_prediction_reward/std": 0.03125,
      "rewards/waypoint_pred_accuracy/mean": 1.0124033455018937e-07,
      "rewards/waypoint_pred_accuracy/std": 4.912120291946227e-07,
      "step": 139
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 365.25,
      "completions/max_terminated_length": 365.25,
      "completions/mean_length": 223.37109375,
      "completions/mean_terminated_length": 223.37109375,
      "completions/min_length": 115.875,
      "completions/min_terminated_length": 115.875,
      "epoch": 0.29473684210526313,
      "grad_norm": 0.020622270181775093,
      "learning_rate": 9.028539129595197e-07,
      "loss": -0.0001,
      "num_tokens": 67625936.0,
      "reward": 1.9263398349285126,
      "reward_std": 0.07740109786391258,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.03125,
      "rewards/waypoint_pred_accuracy/mean": 0.025669913738965988,
      "rewards/waypoint_pred_accuracy/std": 0.028465650044381622,
      "step": 140
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 378.75,
      "completions/max_terminated_length": 378.75,
      "completions/mean_length": 213.451171875,
      "completions/mean_terminated_length": 213.451171875,
      "completions/min_length": 107.875,
      "completions/min_terminated_length": 107.875,
      "epoch": 0.2968421052631579,
      "grad_norm": 0.01621430739760399,
      "learning_rate": 9.00789660141106e-07,
      "loss": -0.0004,
      "num_tokens": 68094391.0,
      "reward": 1.8975248336791992,
      "reward_std": 0.056784010463506895,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.873046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.012238975709265076,
      "rewards/waypoint_pred_accuracy/std": 0.02289922521287524,
      "step": 141
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 366.125,
      "completions/max_terminated_length": 366.125,
      "completions/mean_length": 207.638671875,
      "completions/mean_terminated_length": 207.638671875,
      "completions/min_length": 111.25,
      "completions/min_terminated_length": 111.25,
      "epoch": 0.29894736842105263,
      "grad_norm": 0.0012398953549563885,
      "learning_rate": 8.987064189030983e-07,
      "loss": -0.0,
      "num_tokens": 68561854.0,
      "reward": 1.809123456478119,
      "reward_std": 0.05541337472914165,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.029561737023570345,
      "rewards/waypoint_pred_accuracy/std": 0.027706685019552424,
      "step": 142
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 336.5,
      "completions/max_terminated_length": 336.5,
      "completions/mean_length": 201.529296875,
      "completions/mean_terminated_length": 201.529296875,
      "completions/min_length": 111.875,
      "completions/min_terminated_length": 111.875,
      "epoch": 0.30105263157894735,
      "grad_norm": 0.0,
      "learning_rate": 8.966043020123855e-07,
      "loss": 0.0001,
      "num_tokens": 69026509.0,
      "reward": 2.1186273992061615,
      "reward_std": 0.09686689289469541,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.994140625,
      "rewards/stop_prediction_reward/std": 0.03754601255059242,
      "rewards/waypoint_pred_accuracy/mean": 0.06224340945057809,
      "rewards/waypoint_pred_accuracy/std": 0.029665624278586974,
      "step": 143
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 371.5,
      "completions/max_terminated_length": 371.5,
      "completions/mean_length": 212.501953125,
      "completions/mean_terminated_length": 212.501953125,
      "completions/min_length": 111.5,
      "completions/min_terminated_length": 111.5,
      "epoch": 0.3031578947368421,
      "grad_norm": 0.0,
      "learning_rate": 8.944834232576054e-07,
      "loss": 0.0002,
      "num_tokens": 69494414.0,
      "reward": 2.0669292509555817,
      "reward_std": 0.09019226813688874,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.99609375,
      "rewards/stop_prediction_reward/std": 0.021921012550592422,
      "rewards/waypoint_pred_accuracy/mean": 0.03541775744817477,
      "rewards/waypoint_pred_accuracy/std": 0.03677688956680679,
      "step": 144
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 377.875,
      "completions/max_terminated_length": 377.875,
      "completions/mean_length": 214.53125,
      "completions/mean_terminated_length": 214.53125,
      "completions/min_length": 111.875,
      "completions/min_terminated_length": 111.875,
      "epoch": 0.30526315789473685,
      "grad_norm": 0.0,
      "learning_rate": 8.923438974429849e-07,
      "loss": -0.0,
      "num_tokens": 69966238.0,
      "reward": 1.500933289527893,
      "reward_std": 0.022255118004977703,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.498046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.0014432001626119018,
      "rewards/waypoint_pred_accuracy/std": 0.0033150608651340008,
      "step": 145
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 341.375,
      "completions/max_terminated_length": 341.375,
      "completions/mean_length": 207.609375,
      "completions/mean_terminated_length": 207.609375,
      "completions/min_length": 108.625,
      "completions/min_terminated_length": 108.625,
      "epoch": 0.30736842105263157,
      "grad_norm": 0.01606505736708641,
      "learning_rate": 8.901858403821253e-07,
      "loss": -0.0003,
      "num_tokens": 70430934.0,
      "reward": 1.9067281186580658,
      "reward_std": 0.11152400076389313,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.03754601255059242,
      "rewards/waypoint_pred_accuracy/mean": 0.08031718447636607,
      "rewards/waypoint_pred_accuracy/std": 0.041061242358370054,
      "step": 146
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 339.125,
      "completions/max_terminated_length": 339.125,
      "completions/mean_length": 200.755859375,
      "completions/mean_terminated_length": 200.755859375,
      "completions/min_length": 108.125,
      "completions/min_terminated_length": 108.125,
      "epoch": 0.3094736842105263,
      "grad_norm": 0.022619424387812614,
      "learning_rate": 8.880093688917338e-07,
      "loss": 0.0006,
      "num_tokens": 70895897.0,
      "reward": 1.979694738984108,
      "reward_std": 0.10517071333015338,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.052347380133141996,
      "rewards/waypoint_pred_accuracy/std": 0.05258536203473341,
      "step": 147
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 351.25,
      "completions/max_terminated_length": 351.25,
      "completions/mean_length": 209.3203125,
      "completions/mean_terminated_length": 209.3203125,
      "completions/min_length": 106.375,
      "completions/min_terminated_length": 106.375,
      "epoch": 0.31157894736842107,
      "grad_norm": 0.015080302953720093,
      "learning_rate": 8.858146007853e-07,
      "loss": 0.0007,
      "num_tokens": 71362301.0,
      "reward": 1.7541356086730957,
      "reward_std": 0.0744232046417892,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.046875,
      "rewards/waypoint_pred_accuracy/mean": 0.003044351096450817,
      "rewards/waypoint_pred_accuracy/std": 0.014287834603145175,
      "step": 148
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 339.25,
      "completions/max_terminated_length": 339.25,
      "completions/mean_length": 204.564453125,
      "completions/mean_terminated_length": 204.564453125,
      "completions/min_length": 109.5,
      "completions/min_terminated_length": 109.5,
      "epoch": 0.3136842105263158,
      "grad_norm": 0.014518975280225277,
      "learning_rate": 8.836016548667178e-07,
      "loss": 0.0,
      "num_tokens": 71831198.0,
      "reward": 1.7602195739746094,
      "reward_std": 0.04601290519349277,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.751953125,
      "rewards/stop_prediction_reward/std": 0.03754601255059242,
      "rewards/waypoint_pred_accuracy/mean": 0.004133254632145133,
      "rewards/waypoint_pred_accuracy/std": 0.004251232765162753,
      "step": 149
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.0,
      "learning_rate": 8.813706509238558e-07,
      "loss": -0.0002,
      "step": 150
    },
    {
      "epoch": 0.3157894736842105,
      "eval_clip_ratio": 0.0,
      "eval_completions/clipped_ratio": 0.0,
      "eval_completions/max_length": 410.41,
      "eval_completions/max_terminated_length": 410.41,
      "eval_completions/mean_length": 207.72815979003906,
      "eval_completions/mean_terminated_length": 207.72815979003906,
      "eval_completions/min_length": 111.67,
      "eval_completions/min_terminated_length": 111.67,
      "eval_loss": 0.0016332893865182996,
      "eval_num_tokens": 72298062.0,
      "eval_reward": 1.8651788556575775,
      "eval_reward_std": 0.10482742591684201,
      "eval_rewards/format_reward_embodied/mean": 0.9990625,
      "eval_rewards/format_reward_embodied/std": 0.0075,
      "eval_rewards/stop_prediction_reward/mean": 0.76859375,
      "eval_rewards/stop_prediction_reward/std": 0.020503681004047394,
      "eval_rewards/waypoint_pred_accuracy/mean": 0.04876130852479674,
      "eval_rewards/waypoint_pred_accuracy/std": 0.0403332443083783,
      "eval_runtime": 1073.7169,
      "eval_samples_per_second": 0.093,
      "eval_steps_per_second": 0.002,
      "step": 150
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 359.75,
      "completions/max_terminated_length": 359.75,
      "completions/mean_length": 208.6748046875,
      "completions/mean_terminated_length": 208.6748046875,
      "completions/min_length": 112.5,
      "completions/min_terminated_length": 112.5,
      "epoch": 0.3178947368421053,
      "grad_norm": 0.013742033392190933,
      "learning_rate": 8.791217097220724e-07,
      "loss": -0.0001,
      "num_tokens": 72761681.0,
      "reward": 1.875585325062275,
      "reward_std": 0.023173605810638875,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.8720703125,
      "rewards/stop_prediction_reward/std": 0.01877300627529621,
      "rewards/waypoint_pred_accuracy/mean": 0.0017575172029685837,
      "rewards/waypoint_pred_accuracy/std": 0.0022003025424544072,
      "step": 151
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 350.375,
      "completions/max_terminated_length": 350.375,
      "completions/mean_length": 210.005859375,
      "completions/mean_terminated_length": 210.005859375,
      "completions/min_length": 111.0,
      "completions/min_terminated_length": 111.0,
      "epoch": 0.32,
      "grad_norm": 0.010572736151516438,
      "learning_rate": 8.768549529976783e-07,
      "loss": 0.0006,
      "num_tokens": 73228180.0,
      "reward": 1.9192677438259125,
      "reward_std": 0.049104438461654354,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.022133867223146808,
      "rewards/waypoint_pred_accuracy/std": 0.02455222301614413,
      "step": 152
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 368.0,
      "completions/max_terminated_length": 368.0,
      "completions/mean_length": 197.775390625,
      "completions/mean_terminated_length": 197.775390625,
      "completions/min_length": 107.625,
      "completions/min_terminated_length": 107.625,
      "epoch": 0.32210526315789473,
      "grad_norm": 0.01287839561700821,
      "learning_rate": 8.74570503451348e-07,
      "loss": 0.0003,
      "num_tokens": 73691105.0,
      "reward": 1.9361660480499268,
      "reward_std": 0.10891422609623902,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.87109375,
      "rewards/stop_prediction_reward/std": 0.03125,
      "rewards/waypoint_pred_accuracy/mean": 0.03253614324701938,
      "rewards/waypoint_pred_accuracy/std": 0.03883211300195525,
      "step": 153
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 326.0,
      "completions/max_terminated_length": 326.0,
      "completions/mean_length": 202.033203125,
      "completions/mean_terminated_length": 202.033203125,
      "completions/min_length": 115.75,
      "completions/min_terminated_length": 115.75,
      "epoch": 0.32421052631578945,
      "grad_norm": 0.018562331795692444,
      "learning_rate": 8.72268484741477e-07,
      "loss": -0.0005,
      "num_tokens": 74156146.0,
      "reward": 1.8251054883003235,
      "reward_std": 0.1710243321698499,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.626953125,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.1000527671312958,
      "rewards/waypoint_pred_accuracy/std": 0.07437628054339515,
      "step": 154
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 372.0,
      "completions/max_terminated_length": 372.0,
      "completions/mean_length": 204.990234375,
      "completions/mean_terminated_length": 204.990234375,
      "completions/min_length": 115.75,
      "completions/min_terminated_length": 115.75,
      "epoch": 0.3263157894736842,
      "grad_norm": 0.018523743376135826,
      "learning_rate": 8.699490214774881e-07,
      "loss": 0.0002,
      "num_tokens": 74622701.0,
      "reward": 1.7455661296844482,
      "reward_std": 0.05205453363100787,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.744140625,
      "rewards/stop_prediction_reward/std": 0.046875,
      "rewards/waypoint_pred_accuracy/mean": 0.0007127649293749982,
      "rewards/waypoint_pred_accuracy/std": 0.0025897676093791233,
      "step": 155
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 333.75,
      "completions/max_terminated_length": 333.75,
      "completions/mean_length": 193.486328125,
      "completions/mean_terminated_length": 193.486328125,
      "completions/min_length": 102.625,
      "completions/min_terminated_length": 102.625,
      "epoch": 0.32842105263157895,
      "grad_norm": 0.0,
      "learning_rate": 8.676122392130872e-07,
      "loss": 0.0,
      "num_tokens": 75081254.0,
      "reward": 1.623046875,
      "reward_std": 0.015625,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.623046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 7.546804675536426e-25,
      "rewards/waypoint_pred_accuracy/std": 4.67924008313652e-24,
      "step": 156
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 352.25,
      "completions/max_terminated_length": 352.25,
      "completions/mean_length": 199.650390625,
      "completions/mean_terminated_length": 199.650390625,
      "completions/min_length": 109.75,
      "completions/min_terminated_length": 109.75,
      "epoch": 0.33052631578947367,
      "grad_norm": 0.0,
      "learning_rate": 8.652582644394657e-07,
      "loss": 0.001,
      "num_tokens": 75543923.0,
      "reward": 1.763193666934967,
      "reward_std": 0.043326430561137386,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.007573404463160793,
      "rewards/waypoint_pred_accuracy/std": 0.013850717227414266,
      "step": 157
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 351.375,
      "completions/max_terminated_length": 351.375,
      "completions/mean_length": 211.6484375,
      "completions/mean_terminated_length": 211.6484375,
      "completions/min_length": 105.125,
      "completions/min_terminated_length": 105.125,
      "epoch": 0.33263157894736844,
      "grad_norm": 0.015552366152405739,
      "learning_rate": 8.628872245784545e-07,
      "loss": 0.0007,
      "num_tokens": 76014975.0,
      "reward": 1.7972655892372131,
      "reward_std": 0.11691518849693239,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.046875,
      "rewards/waypoint_pred_accuracy/mean": 0.025585949169908424,
      "rewards/waypoint_pred_accuracy/std": 0.027767218511144182,
      "step": 158
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 335.0,
      "completions/max_terminated_length": 335.0,
      "completions/mean_length": 200.451171875,
      "completions/mean_terminated_length": 200.451171875,
      "completions/min_length": 113.625,
      "completions/min_terminated_length": 113.625,
      "epoch": 0.33473684210526317,
      "grad_norm": 0.012406791560351849,
      "learning_rate": 8.60499247975626e-07,
      "loss": 0.0001,
      "num_tokens": 76479526.0,
      "reward": 1.692464992403984,
      "reward_std": 0.10753844678401947,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.626953125,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.0327559362485772,
      "rewards/waypoint_pred_accuracy/std": 0.0465778008219786,
      "step": 159
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 320.875,
      "completions/max_terminated_length": 320.875,
      "completions/mean_length": 193.51171875,
      "completions/mean_terminated_length": 193.51171875,
      "completions/min_length": 102.875,
      "completions/min_terminated_length": 102.875,
      "epoch": 0.3368421052631579,
      "grad_norm": 0.030549675226211548,
      "learning_rate": 8.58094463893347e-07,
      "loss": 0.0002,
      "num_tokens": 76938732.0,
      "reward": 1.9054777026176453,
      "reward_std": 0.05647589443033718,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.873046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.016215412090739154,
      "rewards/waypoint_pred_accuracy/std": 0.02042544638196753,
      "step": 160
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 334.125,
      "completions/max_terminated_length": 334.125,
      "completions/mean_length": 199.755859375,
      "completions/mean_terminated_length": 199.755859375,
      "completions/min_length": 113.875,
      "completions/min_terminated_length": 113.875,
      "epoch": 0.3389473684210526,
      "grad_norm": 0.014368158765137196,
      "learning_rate": 8.556730025037819e-07,
      "loss": -0.0002,
      "num_tokens": 77399855.0,
      "reward": 1.9117814898490906,
      "reward_std": 0.06853678584000633,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.873046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.020343898673879646,
      "rewards/waypoint_pred_accuracy/std": 0.018643394690372794,
      "step": 161
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 311.125,
      "completions/max_terminated_length": 311.125,
      "completions/mean_length": 194.9609375,
      "completions/mean_terminated_length": 194.9609375,
      "completions/min_length": 111.875,
      "completions/min_terminated_length": 111.875,
      "epoch": 0.3410526315789474,
      "grad_norm": 0.0,
      "learning_rate": 8.532349948818453e-07,
      "loss": 0.0001,
      "num_tokens": 77858715.0,
      "reward": 1.5617362409830093,
      "reward_std": 0.09266455079254143,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.5,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.030868121356284917,
      "rewards/waypoint_pred_accuracy/std": 0.04633227763988046,
      "step": 162
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 359.125,
      "completions/max_terminated_length": 359.125,
      "completions/mean_length": 195.875,
      "completions/mean_terminated_length": 195.875,
      "completions/min_length": 114.25,
      "completions/min_terminated_length": 114.25,
      "epoch": 0.3431578947368421,
      "grad_norm": 0.0,
      "learning_rate": 8.507805729981081e-07,
      "loss": -0.0003,
      "num_tokens": 78320411.0,
      "reward": 1.5313882529735565,
      "reward_std": 0.13795647164806724,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.375,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.07819415128325216,
      "rewards/waypoint_pred_accuracy/std": 0.06897824443884724,
      "step": 163
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 356.125,
      "completions/max_terminated_length": 356.125,
      "completions/mean_length": 200.798828125,
      "completions/mean_terminated_length": 200.798828125,
      "completions/min_length": 112.75,
      "completions/min_terminated_length": 112.75,
      "epoch": 0.3452631578947368,
      "grad_norm": 0.0,
      "learning_rate": 8.483098697116535e-07,
      "loss": -0.0002,
      "num_tokens": 78785204.0,
      "reward": 1.8663674592971802,
      "reward_std": 0.0740682063976692,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.05818372880702327,
      "rewards/waypoint_pred_accuracy/std": 0.03703410336356683,
      "step": 164
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 323.0,
      "completions/max_terminated_length": 323.0,
      "completions/mean_length": 184.767578125,
      "completions/mean_terminated_length": 184.767578125,
      "completions/min_length": 108.125,
      "completions/min_terminated_length": 108.125,
      "epoch": 0.3473684210526316,
      "grad_norm": 0.014639639295637608,
      "learning_rate": 8.45823018762885e-07,
      "loss": 0.0004,
      "num_tokens": 79241469.0,
      "reward": 1.5147821009159088,
      "reward_std": 0.028108830246765137,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.5,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.007391049890770773,
      "rewards/waypoint_pred_accuracy/std": 0.01405441654196693,
      "step": 165
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 327.125,
      "completions/max_terminated_length": 327.125,
      "completions/mean_length": 188.45703125,
      "completions/mean_terminated_length": 188.45703125,
      "completions/min_length": 108.75,
      "completions/min_terminated_length": 108.75,
      "epoch": 0.3494736842105263,
      "grad_norm": 0.01364043541252613,
      "learning_rate": 8.43320154766287e-07,
      "loss": 0.0,
      "num_tokens": 79696807.0,
      "reward": 1.9586426615715027,
      "reward_std": 0.09136595235713685,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.04182136098216205,
      "rewards/waypoint_pred_accuracy/std": 0.04568298065798615,
      "step": 166
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 332.375,
      "completions/max_terminated_length": 332.375,
      "completions/mean_length": 192.078125,
      "completions/mean_terminated_length": 192.078125,
      "completions/min_length": 112.125,
      "completions/min_terminated_length": 112.125,
      "epoch": 0.35157894736842105,
      "grad_norm": 0.0016441630432382226,
      "learning_rate": 8.408014132031385e-07,
      "loss": -0.0002,
      "num_tokens": 80154511.0,
      "reward": 1.753662645816803,
      "reward_std": 0.020446277248106215,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0018313333685417293,
      "rewards/waypoint_pred_accuracy/std": 0.010223136260532173,
      "step": 167
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 350.625,
      "completions/max_terminated_length": 350.625,
      "completions/mean_length": 195.697265625,
      "completions/mean_terminated_length": 195.697265625,
      "completions/min_length": 111.25,
      "completions/min_terminated_length": 111.25,
      "epoch": 0.35368421052631577,
      "grad_norm": 0.0191575326025486,
      "learning_rate": 8.382669304141789e-07,
      "loss": 0.0002,
      "num_tokens": 80616180.0,
      "reward": 1.7313858270645142,
      "reward_std": 0.07452307981657214,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.05319291944942961,
      "rewards/waypoint_pred_accuracy/std": 0.03726154523974404,
      "step": 168
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 339.125,
      "completions/max_terminated_length": 339.125,
      "completions/mean_length": 199.1796875,
      "completions/mean_terminated_length": 199.1796875,
      "completions/min_length": 113.125,
      "completions/min_terminated_length": 113.125,
      "epoch": 0.35578947368421054,
      "grad_norm": 0.0011054413625970483,
      "learning_rate": 8.35716843592228e-07,
      "loss": -0.0002,
      "num_tokens": 81079184.0,
      "reward": 1.877403125166893,
      "reward_std": 0.014616520323585291,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0012015849925077877,
      "rewards/waypoint_pred_accuracy/std": 0.007308262612468781,
      "step": 169
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 345.625,
      "completions/max_terminated_length": 345.625,
      "completions/mean_length": 194.3984375,
      "completions/mean_terminated_length": 194.3984375,
      "completions/min_length": 118.875,
      "completions/min_terminated_length": 118.875,
      "epoch": 0.35789473684210527,
      "grad_norm": 0.04289071634411812,
      "learning_rate": 8.331512907747596e-07,
      "loss": -0.0002,
      "num_tokens": 81539356.0,
      "reward": 1.8431425243616104,
      "reward_std": 0.17017098766780236,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.10907126411490253,
      "rewards/waypoint_pred_accuracy/std": 0.0850854907983205,
      "step": 170
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 312.125,
      "completions/max_terminated_length": 312.125,
      "completions/mean_length": 189.50390625,
      "completions/mean_terminated_length": 189.50390625,
      "completions/min_length": 114.375,
      "completions/min_terminated_length": 114.375,
      "epoch": 0.36,
      "grad_norm": 0.0,
      "learning_rate": 8.305704108364301e-07,
      "loss": 0.0003,
      "num_tokens": 81995934.0,
      "reward": 1.8737435936927795,
      "reward_std": 0.02003212797418996,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.873046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.0003483741428915721,
      "rewards/waypoint_pred_accuracy/std": 0.0022035639689050868,
      "step": 171
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 353.625,
      "completions/max_terminated_length": 353.625,
      "completions/mean_length": 196.31640625,
      "completions/mean_terminated_length": 196.31640625,
      "completions/min_length": 111.5,
      "completions/min_terminated_length": 111.5,
      "epoch": 0.36210526315789476,
      "grad_norm": 0.014004958793520927,
      "learning_rate": 8.279743434815599e-07,
      "loss": 0.0,
      "num_tokens": 82457920.0,
      "reward": 1.6893496811389923,
      "reward_std": 0.07697248342446983,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.03217484994092959,
      "rewards/waypoint_pred_accuracy/std": 0.038486242177896396,
      "step": 172
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 310.375,
      "completions/max_terminated_length": 310.375,
      "completions/mean_length": 184.642578125,
      "completions/mean_terminated_length": 184.642578125,
      "completions/min_length": 110.875,
      "completions/min_terminated_length": 110.875,
      "epoch": 0.3642105263157895,
      "grad_norm": 0.013178675435483456,
      "learning_rate": 8.253632292365726e-07,
      "loss": 0.0,
      "num_tokens": 82915145.0,
      "reward": 2.0115868896245956,
      "reward_std": 0.08305720053613186,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.06829345226288144,
      "rewards/waypoint_pred_accuracy/std": 0.041528596542799806,
      "step": 173
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 318.125,
      "completions/max_terminated_length": 318.125,
      "completions/mean_length": 190.236328125,
      "completions/mean_terminated_length": 190.236328125,
      "completions/min_length": 112.125,
      "completions/min_terminated_length": 112.125,
      "epoch": 0.3663157894736842,
      "grad_norm": 0.02363615669310093,
      "learning_rate": 8.227372094423864e-07,
      "loss": -0.0001,
      "num_tokens": 83374914.0,
      "reward": 2.0038606971502304,
      "reward_std": 0.14384562149643898,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0644303746521473,
      "rewards/waypoint_pred_accuracy/std": 0.07192281540483236,
      "step": 174
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 332.125,
      "completions/max_terminated_length": 332.125,
      "completions/mean_length": 180.150390625,
      "completions/mean_terminated_length": 180.150390625,
      "completions/min_length": 106.875,
      "completions/min_terminated_length": 106.875,
      "epoch": 0.3684210526315789,
      "grad_norm": 0.0,
      "learning_rate": 8.200964262467656e-07,
      "loss": 0.0001,
      "num_tokens": 83825615.0,
      "reward": 1.4097924530506134,
      "reward_std": 0.04181510955095291,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.375,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0173962339758873,
      "rewards/waypoint_pred_accuracy/std": 0.020907556638121605,
      "step": 175
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 307.625,
      "completions/max_terminated_length": 307.625,
      "completions/mean_length": 184.662109375,
      "completions/mean_terminated_length": 184.662109375,
      "completions/min_length": 108.25,
      "completions/min_terminated_length": 108.25,
      "epoch": 0.3705263157894737,
      "grad_norm": 0.01097350474447012,
      "learning_rate": 8.174410225966239e-07,
      "loss": 0.0001,
      "num_tokens": 84280098.0,
      "reward": 2.0639708340168,
      "reward_std": 0.17662093978196225,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.09448541931482657,
      "rewards/waypoint_pred_accuracy/std": 0.0883104762174689,
      "step": 176
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 311.375,
      "completions/max_terminated_length": 311.375,
      "completions/mean_length": 179.3046875,
      "completions/mean_terminated_length": 179.3046875,
      "completions/min_length": 104.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.3726315789473684,
      "grad_norm": 0.0,
      "learning_rate": 8.147711422302881e-07,
      "loss": 0.0,
      "num_tokens": 84732926.0,
      "reward": 1.5250985324382782,
      "reward_std": 0.028973333232215737,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.5,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.01254925754006564,
      "rewards/waypoint_pred_accuracy/std": 0.014486670532335214,
      "step": 177
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 315.625,
      "completions/max_terminated_length": 315.625,
      "completions/mean_length": 182.26171875,
      "completions/mean_terminated_length": 182.26171875,
      "completions/min_length": 112.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.37473684210526315,
      "grad_norm": 0.010854351334273815,
      "learning_rate": 8.120869296697162e-07,
      "loss": -0.0,
      "num_tokens": 85187204.0,
      "reward": 1.7528592348098755,
      "reward_std": 0.01418565196615873,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0014296227405005812,
      "rewards/waypoint_pred_accuracy/std": 0.007092827672981208,
      "step": 178
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 529.0,
      "completions/max_terminated_length": 529.0,
      "completions/mean_length": 185.01953125,
      "completions/mean_terminated_length": 185.01953125,
      "completions/min_length": 105.25,
      "completions/min_terminated_length": 105.25,
      "epoch": 0.37684210526315787,
      "grad_norm": 0.022924024611711502,
      "learning_rate": 8.093885302126754e-07,
      "loss": 0.0073,
      "num_tokens": 85641038.0,
      "reward": 1.7433076351881027,
      "reward_std": 0.06334595192311099,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.744140625,
      "rewards/stop_prediction_reward/std": 0.046875,
      "rewards/waypoint_pred_accuracy/mean": 0.0005600605727522634,
      "rewards/waypoint_pred_accuracy/std": 0.003230394551792415,
      "step": 179
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 310.0,
      "completions/max_terminated_length": 310.0,
      "completions/mean_length": 179.8359375,
      "completions/mean_terminated_length": 179.8359375,
      "completions/min_length": 108.625,
      "completions/min_terminated_length": 108.625,
      "epoch": 0.37894736842105264,
      "grad_norm": 0.0,
      "learning_rate": 8.06676089924877e-07,
      "loss": 0.0004,
      "num_tokens": 86092666.0,
      "reward": 1.8267612159252167,
      "reward_std": 0.1466955652579145,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.10088061677343774,
      "rewards/waypoint_pred_accuracy/std": 0.0733477777656617,
      "step": 180
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 298.75,
      "completions/max_terminated_length": 298.75,
      "completions/mean_length": 184.97265625,
      "completions/mean_terminated_length": 184.97265625,
      "completions/min_length": 111.625,
      "completions/min_terminated_length": 111.625,
      "epoch": 0.38105263157894737,
      "grad_norm": 0.0,
      "learning_rate": 8.03949755632069e-07,
      "loss": -0.0004,
      "num_tokens": 86546476.0,
      "reward": 1.7814702987670898,
      "reward_std": 0.08937373897060752,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.07823515147902071,
      "rewards/waypoint_pred_accuracy/std": 0.04468687262851745,
      "step": 181
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 309.625,
      "completions/max_terminated_length": 309.625,
      "completions/mean_length": 179.16015625,
      "completions/mean_terminated_length": 179.16015625,
      "completions/min_length": 105.625,
      "completions/min_terminated_length": 105.625,
      "epoch": 0.3831578947368421,
      "grad_norm": 0.02093261480331421,
      "learning_rate": 8.01209674912089e-07,
      "loss": 0.0,
      "num_tokens": 86998398.0,
      "reward": 1.9096409678459167,
      "reward_std": 0.0670090508647263,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.07982050045393407,
      "rewards/waypoint_pred_accuracy/std": 0.033504527527838945,
      "step": 182
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 286.375,
      "completions/max_terminated_length": 286.375,
      "completions/mean_length": 171.54296875,
      "completions/mean_terminated_length": 171.54296875,
      "completions/min_length": 102.875,
      "completions/min_terminated_length": 102.875,
      "epoch": 0.38526315789473686,
      "grad_norm": 0.01432411465793848,
      "learning_rate": 7.984559960868759e-07,
      "loss": -0.0004,
      "num_tokens": 87445908.0,
      "reward": 1.6769181191921234,
      "reward_std": 0.10013374220579863,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.02595905796625718,
      "rewards/waypoint_pred_accuracy/std": 0.05006687459543879,
      "step": 183
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 317.375,
      "completions/max_terminated_length": 317.375,
      "completions/mean_length": 181.97265625,
      "completions/mean_terminated_length": 181.97265625,
      "completions/min_length": 106.875,
      "completions/min_terminated_length": 106.875,
      "epoch": 0.3873684210526316,
      "grad_norm": 0.0,
      "learning_rate": 7.956888682144403e-07,
      "loss": 0.0001,
      "num_tokens": 87901126.0,
      "reward": 2.113648146390915,
      "reward_std": 0.08789801027160138,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 1.0,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.05682408035500002,
      "rewards/waypoint_pred_accuracy/std": 0.04394900894840466,
      "step": 184
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 297.375,
      "completions/max_terminated_length": 297.375,
      "completions/mean_length": 170.03125,
      "completions/mean_terminated_length": 170.03125,
      "completions/min_length": 107.25,
      "completions/min_terminated_length": 107.25,
      "epoch": 0.3894736842105263,
      "grad_norm": 0.00041478071943856776,
      "learning_rate": 7.929084410807964e-07,
      "loss": -0.0,
      "num_tokens": 88348630.0,
      "reward": 1.8102026730775833,
      "reward_std": 0.08258083421748097,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.032054444891858935,
      "rewards/waypoint_pred_accuracy/std": 0.025665418093367975,
      "step": 185
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 285.125,
      "completions/max_terminated_length": 285.125,
      "completions/mean_length": 167.068359375,
      "completions/mean_terminated_length": 167.068359375,
      "completions/min_length": 102.0,
      "completions/min_terminated_length": 102.0,
      "epoch": 0.391578947368421,
      "grad_norm": 0.010830595158040524,
      "learning_rate": 7.90114865191855e-07,
      "loss": -0.0002,
      "num_tokens": 88793017.0,
      "reward": 1.899033010005951,
      "reward_std": 0.05657581372270215,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.012016504072278167,
      "rewards/waypoint_pred_accuracy/std": 0.02828790664943881,
      "step": 186
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 270.625,
      "completions/max_terminated_length": 270.625,
      "completions/mean_length": 170.416015625,
      "completions/mean_terminated_length": 170.416015625,
      "completions/min_length": 110.5,
      "completions/min_terminated_length": 110.5,
      "epoch": 0.3936842105263158,
      "grad_norm": 0.02063567005097866,
      "learning_rate": 7.873082917652743e-07,
      "loss": -0.0001,
      "num_tokens": 89241230.0,
      "reward": 1.6706158965826035,
      "reward_std": 0.083322549238801,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.023784521038484087,
      "rewards/waypoint_pred_accuracy/std": 0.03384877370171229,
      "step": 187
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 305.25,
      "completions/max_terminated_length": 305.25,
      "completions/mean_length": 181.23046875,
      "completions/mean_terminated_length": 181.23046875,
      "completions/min_length": 114.5,
      "completions/min_terminated_length": 114.5,
      "epoch": 0.3957894736842105,
      "grad_norm": 0.0,
      "learning_rate": 7.844888727222768e-07,
      "loss": 0.0004,
      "num_tokens": 89692484.0,
      "reward": 1.630469560623169,
      "reward_std": 0.031483914237469435,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0037113359924784914,
      "rewards/waypoint_pred_accuracy/std": 0.00797604240300253,
      "step": 188
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 331.875,
      "completions/max_terminated_length": 331.875,
      "completions/mean_length": 173.87109375,
      "completions/mean_terminated_length": 173.87109375,
      "completions/min_length": 113.75,
      "completions/min_terminated_length": 113.75,
      "epoch": 0.39789473684210525,
      "grad_norm": 0.042730070650577545,
      "learning_rate": 7.816567606794239e-07,
      "loss": -0.0002,
      "num_tokens": 90142082.0,
      "reward": 1.7876133099198341,
      "reward_std": 0.06541969033423811,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0197832117555663,
      "rewards/waypoint_pred_accuracy/std": 0.02541783277411014,
      "step": 189
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 302.125,
      "completions/max_terminated_length": 302.125,
      "completions/mean_length": 182.103515625,
      "completions/mean_terminated_length": 182.103515625,
      "completions/min_length": 115.25,
      "completions/min_terminated_length": 115.25,
      "epoch": 0.4,
      "grad_norm": 0.018826643005013466,
      "learning_rate": 7.788121089403557e-07,
      "loss": 0.0001,
      "num_tokens": 90596087.0,
      "reward": 1.805692046880722,
      "reward_std": 0.06880141280907992,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.027846033899431428,
      "rewards/waypoint_pred_accuracy/std": 0.03440070046775373,
      "step": 190
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 267.875,
      "completions/max_terminated_length": 267.875,
      "completions/mean_length": 175.751953125,
      "completions/mean_terminated_length": 175.751953125,
      "completions/min_length": 108.0,
      "completions/min_terminated_length": 108.0,
      "epoch": 0.40210526315789474,
      "grad_norm": 0.013081498444080353,
      "learning_rate": 7.759550714874924e-07,
      "loss": -0.0004,
      "num_tokens": 91046072.0,
      "reward": 2.0215499103069305,
      "reward_std": 0.13955920189619064,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.873046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.07522808883824439,
      "rewards/waypoint_pred_accuracy/std": 0.05415468077226393,
      "step": 191
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 278.5,
      "completions/max_terminated_length": 278.5,
      "completions/mean_length": 169.3125,
      "completions/mean_terminated_length": 169.3125,
      "completions/min_length": 114.75,
      "completions/min_terminated_length": 114.75,
      "epoch": 0.40421052631578946,
      "grad_norm": 0.0,
      "learning_rate": 7.730858029736989e-07,
      "loss": 0.0,
      "num_tokens": 91491928.0,
      "reward": 1.75,
      "reward_std": 0.0,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 1.474326771777612e-10,
      "rewards/waypoint_pred_accuracy/std": 1.1786529530155576e-09,
      "step": 192
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 306.875,
      "completions/max_terminated_length": 306.875,
      "completions/mean_length": 177.765625,
      "completions/mean_terminated_length": 177.765625,
      "completions/min_length": 114.875,
      "completions/min_terminated_length": 114.875,
      "epoch": 0.4063157894736842,
      "grad_norm": 0.0,
      "learning_rate": 7.702044587139137e-07,
      "loss": 0.0002,
      "num_tokens": 91941856.0,
      "reward": 1.7487657219171524,
      "reward_std": 0.01981517393141985,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.0003594244599299801,
      "rewards/waypoint_pred_accuracy/std": 0.002140052256436602,
      "step": 193
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 303.5,
      "completions/max_terminated_length": 303.5,
      "completions/mean_length": 171.814453125,
      "completions/mean_terminated_length": 172.19029235839844,
      "completions/min_length": 92.375,
      "completions/min_terminated_length": 106.75,
      "epoch": 0.40842105263157896,
      "grad_norm": 0.016274407505989075,
      "learning_rate": 7.673111946767413e-07,
      "loss": -0.0001,
      "num_tokens": 92389185.0,
      "reward": 2.0959380865097046,
      "reward_std": 0.12629193731117994,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.11046904484828701,
      "rewards/waypoint_pred_accuracy/std": 0.06314596923766658,
      "step": 194
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 290.125,
      "completions/max_terminated_length": 290.125,
      "completions/mean_length": 180.162109375,
      "completions/mean_terminated_length": 180.162109375,
      "completions/min_length": 107.5,
      "completions/min_terminated_length": 107.5,
      "epoch": 0.4105263157894737,
      "grad_norm": 0.020258145406842232,
      "learning_rate": 7.644061674760101e-07,
      "loss": 0.0001,
      "num_tokens": 92840532.0,
      "reward": 2.294894278049469,
      "reward_std": 0.23648597935971338,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.998046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.1494002838226665,
      "rewards/waypoint_pred_accuracy/std": 0.11227664479537411,
      "step": 195
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 299.75,
      "completions/max_terminated_length": 299.75,
      "completions/mean_length": 169.263671875,
      "completions/mean_terminated_length": 169.263671875,
      "completions/min_length": 113.25,
      "completions/min_terminated_length": 113.25,
      "epoch": 0.4126315789473684,
      "grad_norm": 0.04483957961201668,
      "learning_rate": 7.61489534362294e-07,
      "loss": -0.0002,
      "num_tokens": 93288091.0,
      "reward": 1.5267165899276733,
      "reward_std": 0.05776224182045553,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.501953125,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.012381742581055732,
      "rewards/waypoint_pred_accuracy/std": 0.02106862150685629,
      "step": 196
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 264.125,
      "completions/max_terminated_length": 264.125,
      "completions/mean_length": 165.439453125,
      "completions/mean_terminated_length": 165.439453125,
      "completions/min_length": 107.625,
      "completions/min_terminated_length": 107.625,
      "epoch": 0.4147368421052632,
      "grad_norm": 0.0,
      "learning_rate": 7.585614532144007e-07,
      "loss": 0.0002,
      "num_tokens": 93733372.0,
      "reward": 1.80105559527874,
      "reward_std": 0.17875106693827547,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.626953125,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.08705123437539442,
      "rewards/waypoint_pred_accuracy/std": 0.08156303651776398,
      "step": 197
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 328.5,
      "completions/max_terminated_length": 328.5,
      "completions/mean_length": 170.4609375,
      "completions/mean_terminated_length": 170.4609375,
      "completions/min_length": 103.125,
      "completions/min_terminated_length": 103.125,
      "epoch": 0.4168421052631579,
      "grad_norm": 0.009551014751195908,
      "learning_rate": 7.556220825308261e-07,
      "loss": 0.0002,
      "num_tokens": 94180968.0,
      "reward": 1.8914762139320374,
      "reward_std": 0.05954795209981967,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.009214683991414209,
      "rewards/waypoint_pred_accuracy/std": 0.021961479235898466,
      "step": 198
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 315.125,
      "completions/max_terminated_length": 315.125,
      "completions/mean_length": 176.51171875,
      "completions/mean_terminated_length": 176.51171875,
      "completions/min_length": 119.25,
      "completions/min_terminated_length": 119.25,
      "epoch": 0.4189473684210526,
      "grad_norm": 0.0,
      "learning_rate": 7.526715814211739e-07,
      "loss": 0.0001,
      "num_tokens": 94629294.0,
      "reward": 2.0064243376255035,
      "reward_std": 0.03297104453667998,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 1.0,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.004188739636447281,
      "rewards/waypoint_pred_accuracy/std": 0.008869364886777475,
      "step": 199
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.0,
      "learning_rate": 7.49710109597544e-07,
      "loss": 0.0002,
      "step": 200
    },
    {
      "epoch": 0.42105263157894735,
      "eval_clip_ratio": 0.0,
      "eval_completions/clipped_ratio": 0.0,
      "eval_completions/max_length": 290.91,
      "eval_completions/max_terminated_length": 290.91,
      "eval_completions/mean_length": 172.36274307250977,
      "eval_completions/mean_terminated_length": 172.36274307250977,
      "eval_completions/min_length": 109.78,
      "eval_completions/min_terminated_length": 109.78,
      "eval_loss": -4.892464494332671e-05,
      "eval_num_tokens": 95080220.0,
      "eval_reward": 1.8596287977695465,
      "eval_reward_std": 0.08604613540126138,
      "eval_rewards/format_reward_embodied/mean": 0.9996875,
      "eval_rewards/format_reward_embodied/std": 0.0025,
      "eval_rewards/stop_prediction_reward/mean": 0.7696875,
      "eval_rewards/stop_prediction_reward/std": 0.005,
      "eval_rewards/waypoint_pred_accuracy/mean": 0.045126906880960875,
      "eval_rewards/waypoint_pred_accuracy/std": 0.039966236149646756,
      "eval_runtime": 941.9095,
      "eval_samples_per_second": 0.106,
      "eval_steps_per_second": 0.002,
      "step": 200
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 292.9375,
      "completions/max_terminated_length": 292.9375,
      "completions/mean_length": 174.2548828125,
      "completions/mean_terminated_length": 174.2548828125,
      "completions/min_length": 110.5625,
      "completions/min_terminated_length": 110.5625,
      "epoch": 0.4231578947368421,
      "grad_norm": 0.0,
      "learning_rate": 7.467378273658856e-07,
      "loss": 0.0005,
      "num_tokens": 95528819.0,
      "reward": 1.9443908333778381,
      "reward_std": 0.04222029652737547,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.8740234375,
      "rewards/stop_prediction_reward/std": 0.0078125,
      "rewards/waypoint_pred_accuracy/mean": 0.03518370707206486,
      "rewards/waypoint_pred_accuracy/std": 0.017203900250024166,
      "step": 201
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 305.5,
      "completions/max_terminated_length": 305.5,
      "completions/mean_length": 170.724609375,
      "completions/mean_terminated_length": 170.724609375,
      "completions/min_length": 112.375,
      "completions/min_terminated_length": 112.375,
      "epoch": 0.42526315789473684,
      "grad_norm": 0.012682443484663963,
      "learning_rate": 7.437548956173213e-07,
      "loss": -0.0004,
      "num_tokens": 95979302.0,
      "reward": 1.9156423211097717,
      "reward_std": 0.1263027695240453,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.751953125,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.0818446125079697,
      "rewards/waypoint_pred_accuracy/std": 0.055338893387193444,
      "step": 202
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 293.875,
      "completions/max_terminated_length": 293.875,
      "completions/mean_length": 176.544921875,
      "completions/mean_terminated_length": 176.8624153137207,
      "completions/min_length": 101.625,
      "completions/min_terminated_length": 111.625,
      "epoch": 0.42736842105263156,
      "grad_norm": 0.0,
      "learning_rate": 7.407614758194373e-07,
      "loss": -0.0006,
      "num_tokens": 96431805.0,
      "reward": 1.707748532295227,
      "reward_std": 0.10709417768262597,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.623046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.04332738941334985,
      "rewards/waypoint_pred_accuracy/std": 0.03793920004429459,
      "step": 203
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 321.5,
      "completions/max_terminated_length": 321.5,
      "completions/mean_length": 176.828125,
      "completions/mean_terminated_length": 176.828125,
      "completions/min_length": 115.625,
      "completions/min_terminated_length": 115.625,
      "epoch": 0.42947368421052634,
      "grad_norm": 0.0,
      "learning_rate": 7.377577300075431e-07,
      "loss": 0.0,
      "num_tokens": 96881189.0,
      "reward": 1.760355144739151,
      "reward_std": 0.02672452749078502,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.005177572603687428,
      "rewards/waypoint_pred_accuracy/std": 0.013362265083738705,
      "step": 204
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 304.75,
      "completions/max_terminated_length": 304.75,
      "completions/mean_length": 175.419921875,
      "completions/mean_terminated_length": 175.419921875,
      "completions/min_length": 117.75,
      "completions/min_terminated_length": 117.75,
      "epoch": 0.43157894736842106,
      "grad_norm": 0.015611842274665833,
      "learning_rate": 7.347438207759002e-07,
      "loss": -0.0002,
      "num_tokens": 97333116.0,
      "reward": 1.9850184619426727,
      "reward_std": 0.06815559589631448,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.055009242740135744,
      "rewards/waypoint_pred_accuracy/std": 0.0340777950465283,
      "step": 205
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 306.875,
      "completions/max_terminated_length": 306.875,
      "completions/mean_length": 174.68359375,
      "completions/mean_terminated_length": 174.68359375,
      "completions/min_length": 107.125,
      "completions/min_terminated_length": 107.125,
      "epoch": 0.4336842105263158,
      "grad_norm": 0.0,
      "learning_rate": 7.317199112689219e-07,
      "loss": -0.0003,
      "num_tokens": 97780314.0,
      "reward": 1.6253042817115784,
      "reward_std": 0.0009640372365589123,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0001521414744787989,
      "rewards/waypoint_pred_accuracy/std": 0.00048202241833421994,
      "step": 206
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 294.625,
      "completions/max_terminated_length": 294.625,
      "completions/mean_length": 178.39453125,
      "completions/mean_terminated_length": 178.39453125,
      "completions/min_length": 117.125,
      "completions/min_terminated_length": 117.125,
      "epoch": 0.4357894736842105,
      "grad_norm": 0.0,
      "learning_rate": 7.286861651723403e-07,
      "loss": 0.0,
      "num_tokens": 98230564.0,
      "reward": 1.8962776064872742,
      "reward_std": 0.06431814459938323,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.07313878076765164,
      "rewards/waypoint_pred_accuracy/std": 0.03215907396928667,
      "step": 207
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 297.125,
      "completions/max_terminated_length": 297.125,
      "completions/mean_length": 170.755859375,
      "completions/mean_terminated_length": 170.755859375,
      "completions/min_length": 108.125,
      "completions/min_terminated_length": 108.125,
      "epoch": 0.4378947368421053,
      "grad_norm": 0.02003273367881775,
      "learning_rate": 7.256427467043479e-07,
      "loss": 0.0001,
      "num_tokens": 98680935.0,
      "reward": 1.7773285955190659,
      "reward_std": 0.045251342578694675,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.013664303890479346,
      "rewards/waypoint_pred_accuracy/std": 0.022625670864954373,
      "step": 208
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 307.625,
      "completions/max_terminated_length": 307.625,
      "completions/mean_length": 175.458984375,
      "completions/mean_terminated_length": 175.458984375,
      "completions/min_length": 109.75,
      "completions/min_terminated_length": 109.75,
      "epoch": 0.44,
      "grad_norm": 0.0,
      "learning_rate": 7.225898206067071e-07,
      "loss": 0.0,
      "num_tokens": 99131986.0,
      "reward": 2.000874102115631,
      "reward_std": 0.07641031977254897,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.06293705673306249,
      "rewards/waypoint_pred_accuracy/std": 0.038205162913072854,
      "step": 209
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 295.25,
      "completions/max_terminated_length": 295.25,
      "completions/mean_length": 177.837890625,
      "completions/mean_terminated_length": 177.837890625,
      "completions/min_length": 118.75,
      "completions/min_terminated_length": 118.75,
      "epoch": 0.4421052631578947,
      "grad_norm": 0.01405387930572033,
      "learning_rate": 7.195275521358332e-07,
      "loss": -0.0003,
      "num_tokens": 99580031.0,
      "reward": 2.028854936361313,
      "reward_std": 0.11646178726022072,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.07692747169583214,
      "rewards/waypoint_pred_accuracy/std": 0.05823090146415666,
      "step": 210
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 307.5,
      "completions/max_terminated_length": 307.5,
      "completions/mean_length": 177.76171875,
      "completions/mean_terminated_length": 177.76171875,
      "completions/min_length": 115.5,
      "completions/min_terminated_length": 115.5,
      "epoch": 0.4442105263157895,
      "grad_norm": 0.0,
      "learning_rate": 7.164561070538488e-07,
      "loss": 0.0003,
      "num_tokens": 100033989.0,
      "reward": 1.7999018132686615,
      "reward_std": 0.21613861247897148,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.5,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.14995091408491135,
      "rewards/waypoint_pred_accuracy/std": 0.10806930996477604,
      "step": 211
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 431.875,
      "completions/max_terminated_length": 431.875,
      "completions/mean_length": 174.076171875,
      "completions/mean_terminated_length": 174.076171875,
      "completions/min_length": 112.375,
      "completions/min_terminated_length": 112.375,
      "epoch": 0.4463157894736842,
      "grad_norm": 0.04999241605401039,
      "learning_rate": 7.133756516196107e-07,
      "loss": -0.0003,
      "num_tokens": 100487340.0,
      "reward": 1.7891514897346497,
      "reward_std": 0.0568979331983428,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.01957578670394547,
      "rewards/waypoint_pred_accuracy/std": 0.028448972130328657,
      "step": 212
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 284.0,
      "completions/max_terminated_length": 284.0,
      "completions/mean_length": 171.359375,
      "completions/mean_terminated_length": 171.359375,
      "completions/min_length": 109.375,
      "completions/min_terminated_length": 109.375,
      "epoch": 0.44842105263157894,
      "grad_norm": 0.0,
      "learning_rate": 7.102863525797112e-07,
      "loss": -0.0001,
      "num_tokens": 100937124.0,
      "reward": 1.9840654134750366,
      "reward_std": 0.17244431003928185,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.17953270860016346,
      "rewards/waypoint_pred_accuracy/std": 0.08622215129435062,
      "step": 213
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 291.375,
      "completions/max_terminated_length": 291.375,
      "completions/mean_length": 175.158203125,
      "completions/mean_terminated_length": 175.158203125,
      "completions/min_length": 113.25,
      "completions/min_terminated_length": 113.25,
      "epoch": 0.45052631578947366,
      "grad_norm": 0.01138628926128149,
      "learning_rate": 7.071883771594509e-07,
      "loss": 0.0,
      "num_tokens": 101387957.0,
      "reward": 1.8491481095552444,
      "reward_std": 0.1147658722824616,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.049574058981537675,
      "rewards/waypoint_pred_accuracy/std": 0.05738293592632728,
      "step": 214
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 327.125,
      "completions/max_terminated_length": 327.125,
      "completions/mean_length": 187.966796875,
      "completions/mean_terminated_length": 187.966796875,
      "completions/min_length": 114.625,
      "completions/min_terminated_length": 114.625,
      "epoch": 0.45263157894736844,
      "grad_norm": 0.011563337408006191,
      "learning_rate": 7.040818930537874e-07,
      "loss": -0.0003,
      "num_tokens": 101845412.0,
      "reward": 1.6611975878477097,
      "reward_std": 0.0494131935941482,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.018098798598909838,
      "rewards/waypoint_pred_accuracy/std": 0.02470659996008351,
      "step": 215
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 287.25,
      "completions/max_terminated_length": 287.25,
      "completions/mean_length": 171.841796875,
      "completions/mean_terminated_length": 171.841796875,
      "completions/min_length": 107.125,
      "completions/min_terminated_length": 107.125,
      "epoch": 0.45473684210526316,
      "grad_norm": 2.7086246063845465e-06,
      "learning_rate": 7.009670684182576e-07,
      "loss": -0.0,
      "num_tokens": 102293587.0,
      "reward": 1.5,
      "reward_std": 3.251700020356907e-09,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.5,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 4.380867890674267e-10,
      "rewards/waypoint_pred_accuracy/std": 1.9065319947775272e-09,
      "step": 216
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 286.75,
      "completions/max_terminated_length": 286.75,
      "completions/mean_length": 176.490234375,
      "completions/mean_terminated_length": 176.490234375,
      "completions/min_length": 114.5,
      "completions/min_terminated_length": 114.5,
      "epoch": 0.4568421052631579,
      "grad_norm": 0.0,
      "learning_rate": 6.978440718598756e-07,
      "loss": 0.0002,
      "num_tokens": 102744270.0,
      "reward": 1.7497325837612152,
      "reward_std": 0.06904890944133513,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.06236628795954857,
      "rewards/waypoint_pred_accuracy/std": 0.03452445384209568,
      "step": 217
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 314.0,
      "completions/max_terminated_length": 314.0,
      "completions/mean_length": 182.732421875,
      "completions/mean_terminated_length": 182.732421875,
      "completions/min_length": 111.375,
      "completions/min_terminated_length": 111.375,
      "epoch": 0.4589473684210526,
      "grad_norm": 0.029011964797973633,
      "learning_rate": 6.947130724280057e-07,
      "loss": 0.0,
      "num_tokens": 103198789.0,
      "reward": 1.7562294006347656,
      "reward_std": 0.013579967227997258,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0031146957662715976,
      "rewards/waypoint_pred_accuracy/std": 0.006789985702200646,
      "step": 218
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 270.375,
      "completions/max_terminated_length": 270.375,
      "completions/mean_length": 172.20703125,
      "completions/mean_terminated_length": 172.20703125,
      "completions/min_length": 111.875,
      "completions/min_terminated_length": 111.875,
      "epoch": 0.4610526315789474,
      "grad_norm": 0.013365295715630054,
      "learning_rate": 6.915742396052115e-07,
      "loss": -0.0001,
      "num_tokens": 103649519.0,
      "reward": 1.5070368647575378,
      "reward_std": 0.023722524622826313,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.501953125,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.002541873606073473,
      "rewards/waypoint_pred_accuracy/std": 0.004048763120422751,
      "step": 219
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 291.375,
      "completions/max_terminated_length": 291.375,
      "completions/mean_length": 176.5546875,
      "completions/mean_terminated_length": 176.5546875,
      "completions/min_length": 110.875,
      "completions/min_terminated_length": 110.875,
      "epoch": 0.4631578947368421,
      "grad_norm": 0.0,
      "learning_rate": 6.884277432980825e-07,
      "loss": 0.0001,
      "num_tokens": 104099915.0,
      "reward": 1.8983599245548248,
      "reward_std": 0.022330745094222948,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.011679970655677607,
      "rewards/waypoint_pred_accuracy/std": 0.01116537469351897,
      "step": 220
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 272.875,
      "completions/max_terminated_length": 272.875,
      "completions/mean_length": 176.126953125,
      "completions/mean_terminated_length": 176.126953125,
      "completions/min_length": 118.875,
      "completions/min_terminated_length": 118.875,
      "epoch": 0.4652631578947368,
      "grad_norm": 0.017878547310829163,
      "learning_rate": 6.852737538280359e-07,
      "loss": -0.0001,
      "num_tokens": 104550732.0,
      "reward": 1.7993512451648712,
      "reward_std": 0.045344060357820126,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.024675624659721507,
      "rewards/waypoint_pred_accuracy/std": 0.022672027718726895,
      "step": 221
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 305.5,
      "completions/max_terminated_length": 305.5,
      "completions/mean_length": 181.193359375,
      "completions/mean_terminated_length": 181.193359375,
      "completions/min_length": 113.875,
      "completions/min_terminated_length": 113.875,
      "epoch": 0.4673684210526316,
      "grad_norm": 0.01165629643946886,
      "learning_rate": 6.821124419220978e-07,
      "loss": 0.0002,
      "num_tokens": 105002991.0,
      "reward": 1.8998810648918152,
      "reward_std": 0.040346091078029334,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.012440536318531065,
      "rewards/waypoint_pred_accuracy/std": 0.020173047916574705,
      "step": 222
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 288.375,
      "completions/max_terminated_length": 288.375,
      "completions/mean_length": 175.021484375,
      "completions/mean_terminated_length": 175.021484375,
      "completions/min_length": 109.75,
      "completions/min_terminated_length": 109.75,
      "epoch": 0.4694736842105263,
      "grad_norm": 0.0,
      "learning_rate": 6.789439787036614e-07,
      "loss": -0.0001,
      "num_tokens": 105453626.0,
      "reward": 1.9986501336097717,
      "reward_std": 0.019883400294929743,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.998046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.0003016188566107303,
      "rewards/waypoint_pred_accuracy/std": 0.0021291994489729404,
      "step": 223
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 309.375,
      "completions/max_terminated_length": 309.375,
      "completions/mean_length": 179.6640625,
      "completions/mean_terminated_length": 179.6640625,
      "completions/min_length": 112.75,
      "completions/min_terminated_length": 112.75,
      "epoch": 0.47157894736842104,
      "grad_norm": 0.015561181120574474,
      "learning_rate": 6.757685356832242e-07,
      "loss": -0.0,
      "num_tokens": 105904398.0,
      "reward": 2.101376533508301,
      "reward_std": 0.06816481053829193,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 1.0,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.05068826675858098,
      "rewards/waypoint_pred_accuracy/std": 0.0340824015760622,
      "step": 224
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 294.0,
      "completions/max_terminated_length": 294.0,
      "completions/mean_length": 175.58203125,
      "completions/mean_terminated_length": 175.58203125,
      "completions/min_length": 109.625,
      "completions/min_terminated_length": 109.625,
      "epoch": 0.47368421052631576,
      "grad_norm": 0.013554830104112625,
      "learning_rate": 6.725862847491034e-07,
      "loss": 0.0002,
      "num_tokens": 106353592.0,
      "reward": 1.7516418248414993,
      "reward_std": 0.0034361608559265733,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0008209160487240297,
      "rewards/waypoint_pred_accuracy/std": 0.0017180802678922191,
      "step": 225
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 306.5,
      "completions/max_terminated_length": 306.5,
      "completions/mean_length": 183.095703125,
      "completions/mean_terminated_length": 183.095703125,
      "completions/min_length": 108.125,
      "completions/min_terminated_length": 108.125,
      "epoch": 0.47578947368421054,
      "grad_norm": 0.00029330080724321306,
      "learning_rate": 6.693973981581324e-07,
      "loss": 0.0,
      "num_tokens": 106808553.0,
      "reward": 1.6250907480716705,
      "reward_std": 0.00024353076181782285,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 4.53742529797907e-05,
      "rewards/waypoint_pred_accuracy/std": 0.00012176844195366245,
      "step": 226
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 292.5,
      "completions/max_terminated_length": 292.5,
      "completions/mean_length": 181.337890625,
      "completions/mean_terminated_length": 181.337890625,
      "completions/min_length": 119.625,
      "completions/min_terminated_length": 119.625,
      "epoch": 0.47789473684210526,
      "grad_norm": 0.0,
      "learning_rate": 6.662020485263358e-07,
      "loss": -0.0001,
      "num_tokens": 107258774.0,
      "reward": 1.625608280301094,
      "reward_std": 0.0014176478143781424,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.00030413969053576185,
      "rewards/waypoint_pred_accuracy/std": 0.00070882499138026,
      "step": 227
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 291.25,
      "completions/max_terminated_length": 291.25,
      "completions/mean_length": 173.318359375,
      "completions/mean_terminated_length": 173.318359375,
      "completions/min_length": 107.875,
      "completions/min_terminated_length": 107.875,
      "epoch": 0.48,
      "grad_norm": 0.0,
      "learning_rate": 6.630004088195858e-07,
      "loss": 0.0002,
      "num_tokens": 107708793.0,
      "reward": 1.876689851284027,
      "reward_std": 0.0026773642748594284,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0008449186572583362,
      "rewards/waypoint_pred_accuracy/std": 0.0013386833028957005,
      "step": 228
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 308.5,
      "completions/max_terminated_length": 308.5,
      "completions/mean_length": 176.880859375,
      "completions/mean_terminated_length": 176.880859375,
      "completions/min_length": 117.25,
      "completions/min_terminated_length": 117.25,
      "epoch": 0.48210526315789476,
      "grad_norm": 0.0,
      "learning_rate": 6.597926523442398e-07,
      "loss": 0.0,
      "num_tokens": 108161148.0,
      "reward": 1.509762555360794,
      "reward_std": 0.021483093870038772,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.5,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0048812878156252685,
      "rewards/waypoint_pred_accuracy/std": 0.01074154757443697,
      "step": 229
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 312.125,
      "completions/max_terminated_length": 312.125,
      "completions/mean_length": 173.00390625,
      "completions/mean_terminated_length": 173.00390625,
      "completions/min_length": 110.25,
      "completions/min_terminated_length": 110.25,
      "epoch": 0.4842105263157895,
      "grad_norm": 0.013146106153726578,
      "learning_rate": 6.565789527377587e-07,
      "loss": -0.0005,
      "num_tokens": 108611454.0,
      "reward": 1.7406707108020782,
      "reward_std": 0.09716045763343573,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.05783534119836986,
      "rewards/waypoint_pred_accuracy/std": 0.048580223228782415,
      "step": 230
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 286.625,
      "completions/max_terminated_length": 286.625,
      "completions/mean_length": 172.7578125,
      "completions/mean_terminated_length": 172.7578125,
      "completions/min_length": 111.75,
      "completions/min_terminated_length": 111.75,
      "epoch": 0.4863157894736842,
      "grad_norm": 0.00952562689781189,
      "learning_rate": 6.533594839593081e-07,
      "loss": 0.0001,
      "num_tokens": 109059522.0,
      "reward": 1.7838719189167023,
      "reward_std": 0.09348210319876671,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.07943596761145057,
      "rewards/waypoint_pred_accuracy/std": 0.04674104697497585,
      "step": 231
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 293.25,
      "completions/max_terminated_length": 293.25,
      "completions/mean_length": 175.3828125,
      "completions/mean_terminated_length": 175.3828125,
      "completions/min_length": 114.75,
      "completions/min_terminated_length": 114.75,
      "epoch": 0.4884210526315789,
      "grad_norm": 0.0,
      "learning_rate": 6.501344202803414e-07,
      "loss": 0.0001,
      "num_tokens": 109511942.0,
      "reward": 1.5068429559469223,
      "reward_std": 0.023580931854667142,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.5,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.003421477313622745,
      "rewards/waypoint_pred_accuracy/std": 0.011790467111495673,
      "step": 232
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 316.0,
      "completions/max_terminated_length": 316.0,
      "completions/mean_length": 178.611328125,
      "completions/mean_terminated_length": 178.9438877105713,
      "completions/min_length": 97.375,
      "completions/min_terminated_length": 110.125,
      "epoch": 0.4905263157894737,
      "grad_norm": 0.010548449121415615,
      "learning_rate": 6.469039362751677e-07,
      "loss": -0.0002,
      "num_tokens": 109963455.0,
      "reward": 1.7827188670635223,
      "reward_std": 0.06010658299783245,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.018312573189177783,
      "rewards/waypoint_pred_accuracy/std": 0.024115337153489236,
      "step": 233
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 301.125,
      "completions/max_terminated_length": 301.125,
      "completions/mean_length": 179.953125,
      "completions/mean_terminated_length": 180.33076095581055,
      "completions/min_length": 101.375,
      "completions/min_terminated_length": 115.75,
      "epoch": 0.4926315789473684,
      "grad_norm": 0.0,
      "learning_rate": 6.436682068115002e-07,
      "loss": -0.0007,
      "num_tokens": 110416615.0,
      "reward": 1.99609375,
      "reward_std": 0.03125,
      "rewards/format_reward_embodied/mean": 0.998046875,
      "rewards/format_reward_embodied/std": 0.015625,
      "rewards/stop_prediction_reward/mean": 0.998046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 2.0173116242793235e-14,
      "rewards/waypoint_pred_accuracy/std": 1.1195781985957193e-13,
      "step": 234
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 269.0,
      "completions/max_terminated_length": 269.0,
      "completions/mean_length": 173.30859375,
      "completions/mean_terminated_length": 173.30859375,
      "completions/min_length": 112.75,
      "completions/min_terminated_length": 112.75,
      "epoch": 0.49473684210526314,
      "grad_norm": 0.023477498441934586,
      "learning_rate": 6.404274070409915e-07,
      "loss": 0.0002,
      "num_tokens": 110865797.0,
      "reward": 1.8803559094667435,
      "reward_std": 0.11074852512797406,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.06517796947535714,
      "rewards/waypoint_pred_accuracy/std": 0.05537425884915592,
      "step": 235
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 299.875,
      "completions/max_terminated_length": 299.875,
      "completions/mean_length": 176.0390625,
      "completions/mean_terminated_length": 176.0390625,
      "completions/min_length": 114.375,
      "completions/min_terminated_length": 114.375,
      "epoch": 0.4968421052631579,
      "grad_norm": 0.008372652344405651,
      "learning_rate": 6.371817123897528e-07,
      "loss": 0.0006,
      "num_tokens": 111319513.0,
      "reward": 1.8294343054294586,
      "reward_std": 0.08038223959738389,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.748046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.040693737070411296,
      "rewards/waypoint_pred_accuracy/std": 0.03237862857612228,
      "step": 236
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 251.25,
      "completions/max_terminated_length": 251.25,
      "completions/mean_length": 166.775390625,
      "completions/mean_terminated_length": 166.775390625,
      "completions/min_length": 109.5,
      "completions/min_terminated_length": 109.5,
      "epoch": 0.49894736842105264,
      "grad_norm": 0.0,
      "learning_rate": 6.339312985488576e-07,
      "loss": -0.0001,
      "num_tokens": 111764710.0,
      "reward": 1.8851255774497986,
      "reward_std": 0.04906696546822786,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.005062782554890241,
      "rewards/waypoint_pred_accuracy/std": 0.024533490184718572,
      "step": 237
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 296.5,
      "completions/max_terminated_length": 296.5,
      "completions/mean_length": 172.90625,
      "completions/mean_terminated_length": 172.90625,
      "completions/min_length": 111.125,
      "completions/min_terminated_length": 111.125,
      "epoch": 0.5010526315789474,
      "grad_norm": 0.01637883298099041,
      "learning_rate": 6.30676341464831e-07,
      "loss": -0.0003,
      "num_tokens": 112215734.0,
      "reward": 1.6509484648704529,
      "reward_std": 0.06736253991999908,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.03125,
      "rewards/waypoint_pred_accuracy/mean": 0.012974242886534648,
      "rewards/waypoint_pred_accuracy/std": 0.01805627301899171,
      "step": 238
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 289.625,
      "completions/max_terminated_length": 289.625,
      "completions/mean_length": 169.388671875,
      "completions/mean_terminated_length": 169.388671875,
      "completions/min_length": 109.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 0.5031578947368421,
      "grad_norm": 0.0,
      "learning_rate": 6.274170173301268e-07,
      "loss": 0.0001,
      "num_tokens": 112666173.0,
      "reward": 1.9938509166240692,
      "reward_std": 0.11555472994223237,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.05942547577433288,
      "rewards/waypoint_pred_accuracy/std": 0.05777736520394683,
      "step": 239
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 289.5,
      "completions/max_terminated_length": 289.5,
      "completions/mean_length": 172.37890625,
      "completions/mean_terminated_length": 172.37890625,
      "completions/min_length": 108.875,
      "completions/min_terminated_length": 108.875,
      "epoch": 0.5052631578947369,
      "grad_norm": 0.0,
      "learning_rate": 6.24153502573589e-07,
      "loss": -0.0001,
      "num_tokens": 113115903.0,
      "reward": 1.7577707767486572,
      "reward_std": 0.016411395743489265,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.0038853867445141077,
      "rewards/waypoint_pred_accuracy/std": 0.008205699268728495,
      "step": 240
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 286.0,
      "completions/max_terminated_length": 286.0,
      "completions/mean_length": 175.09375,
      "completions/mean_terminated_length": 175.09375,
      "completions/min_length": 116.375,
      "completions/min_terminated_length": 116.375,
      "epoch": 0.5073684210526316,
      "grad_norm": 0.03419042006134987,
      "learning_rate": 6.208859738509021e-07,
      "loss": 0.0007,
      "num_tokens": 113568495.0,
      "reward": 1.7237209975719452,
      "reward_std": 0.14016160182654858,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.049360513221472516,
      "rewards/waypoint_pred_accuracy/std": 0.07008079765364533,
      "step": 241
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 289.875,
      "completions/max_terminated_length": 289.875,
      "completions/mean_length": 173.462890625,
      "completions/mean_terminated_length": 173.462890625,
      "completions/min_length": 112.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.5094736842105263,
      "grad_norm": 0.0,
      "learning_rate": 6.176146080350286e-07,
      "loss": 0.0,
      "num_tokens": 114018972.0,
      "reward": 1.8751783967018127,
      "reward_std": 0.00036079369601793587,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 8.91994423000142e-05,
      "rewards/waypoint_pred_accuracy/std": 0.00018039710994344205,
      "step": 242
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 298.0,
      "completions/max_terminated_length": 298.0,
      "completions/mean_length": 173.265625,
      "completions/mean_terminated_length": 173.265625,
      "completions/min_length": 114.25,
      "completions/min_terminated_length": 114.25,
      "epoch": 0.511578947368421,
      "grad_norm": 0.013551232405006886,
      "learning_rate": 6.14339582206635e-07,
      "loss": 0.0001,
      "num_tokens": 114468132.0,
      "reward": 1.9178512692451477,
      "reward_std": 0.06416846066713333,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.021425632760024103,
      "rewards/waypoint_pred_accuracy/std": 0.03208423405926866,
      "step": 243
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 312.125,
      "completions/max_terminated_length": 312.125,
      "completions/mean_length": 175.140625,
      "completions/mean_terminated_length": 175.140625,
      "completions/min_length": 110.75,
      "completions/min_terminated_length": 110.75,
      "epoch": 0.5136842105263157,
      "grad_norm": 0.02614566497504711,
      "learning_rate": 6.110610736445058e-07,
      "loss": 0.0003,
      "num_tokens": 114913708.0,
      "reward": 1.8622263967990875,
      "reward_std": 0.18642316292971373,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.11861319048330188,
      "rewards/waypoint_pred_accuracy/std": 0.09321157418889925,
      "step": 244
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 301.125,
      "completions/max_terminated_length": 301.125,
      "completions/mean_length": 172.37890625,
      "completions/mean_terminated_length": 172.37890625,
      "completions/min_length": 110.75,
      "completions/min_terminated_length": 110.75,
      "epoch": 0.5157894736842106,
      "grad_norm": 0.013545077294111252,
      "learning_rate": 6.077792598159479e-07,
      "loss": -0.0,
      "num_tokens": 115361006.0,
      "reward": 1.9425796866416931,
      "reward_std": 0.10113994629730882,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.03378988173824164,
      "rewards/waypoint_pred_accuracy/std": 0.05056997878441809,
      "step": 245
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 276.5,
      "completions/max_terminated_length": 276.5,
      "completions/mean_length": 173.775390625,
      "completions/mean_terminated_length": 173.775390625,
      "completions/min_length": 116.5,
      "completions/min_terminated_length": 116.5,
      "epoch": 0.5178947368421053,
      "grad_norm": 0.02074316143989563,
      "learning_rate": 6.044943183671836e-07,
      "loss": 0.0004,
      "num_tokens": 115809723.0,
      "reward": 1.7704734951257706,
      "reward_std": 0.09013272261904604,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.623046875,
      "rewards/stop_prediction_reward/std": 0.015625,
      "rewards/waypoint_pred_accuracy/mean": 0.07371331164245536,
      "rewards/waypoint_pred_accuracy/std": 0.03725385823372892,
      "step": 246
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 258.375,
      "completions/max_terminated_length": 258.375,
      "completions/mean_length": 170.69140625,
      "completions/mean_terminated_length": 170.69140625,
      "completions/min_length": 119.125,
      "completions/min_terminated_length": 119.125,
      "epoch": 0.52,
      "grad_norm": 0.021677250042557716,
      "learning_rate": 6.01206427113735e-07,
      "loss": -0.0001,
      "num_tokens": 116257053.0,
      "reward": 1.8927133083343506,
      "reward_std": 0.0853966644051809,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.75,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.07135666430194423,
      "rewards/waypoint_pred_accuracy/std": 0.04269833582034271,
      "step": 247
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 272.125,
      "completions/max_terminated_length": 272.125,
      "completions/mean_length": 169.240234375,
      "completions/mean_terminated_length": 169.240234375,
      "completions/min_length": 112.375,
      "completions/min_terminated_length": 112.375,
      "epoch": 0.5221052631578947,
      "grad_norm": 0.000306050234939903,
      "learning_rate": 5.97915764030799e-07,
      "loss": -0.0,
      "num_tokens": 116703576.0,
      "reward": 1.6304270327091217,
      "reward_std": 0.011630857972136255,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.625,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.002713506846724556,
      "rewards/waypoint_pred_accuracy/std": 0.005815430337122507,
      "step": 248
    },
    {
      "clip_ratio": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 282.375,
      "completions/max_terminated_length": 282.375,
      "completions/mean_length": 166.033203125,
      "completions/mean_terminated_length": 166.033203125,
      "completions/min_length": 113.625,
      "completions/min_terminated_length": 113.625,
      "epoch": 0.5242105263157895,
      "grad_norm": 0.0,
      "learning_rate": 5.946225072436121e-07,
      "loss": 0.0001,
      "num_tokens": 117151145.0,
      "reward": 1.8943032920360565,
      "reward_std": 0.03212926587002585,
      "rewards/format_reward_embodied/mean": 1.0,
      "rewards/format_reward_embodied/std": 0.0,
      "rewards/stop_prediction_reward/mean": 0.875,
      "rewards/stop_prediction_reward/std": 0.0,
      "rewards/waypoint_pred_accuracy/mean": 0.009651642787417297,
      "rewards/waypoint_pred_accuracy/std": 0.016064628072191757,
      "step": 249
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.015220913104712963,
      "learning_rate": 5.913268350178101e-07,
      "loss": 0.0001,
      "step": 250
    },
    {
      "epoch": 0.5263157894736842,
      "eval_clip_ratio": 0.0,
      "eval_completions/clipped_ratio": 0.00015625,
      "eval_completions/max_length": 278.13,
      "eval_completions/max_terminated_length": 278.13,
      "eval_completions/mean_length": 169.34763885498046,
      "eval_completions/mean_terminated_length": 169.375066986084,
      "eval_completions/min_length": 110.76,
      "eval_completions/min_terminated_length": 111.92,
      "eval_loss": -2.504486656107474e-05,
      "eval_num_tokens": 117598301.0,
      "eval_reward": 1.865585025548935,
      "eval_reward_std": 0.0873453421616474,
      "eval_rewards/format_reward_embodied/mean": 0.9996875,
      "eval_rewards/format_reward_embodied/std": 0.0025,
      "eval_rewards/stop_prediction_reward/mean": 0.76984375,
      "eval_rewards/stop_prediction_reward/std": 0.008003681004047393,
      "eval_rewards/waypoint_pred_accuracy/mean": 0.048026895228197336,
      "eval_rewards/waypoint_pred_accuracy/std": 0.038523975913848735,
      "eval_runtime": 963.8699,
      "eval_samples_per_second": 0.104,
      "eval_steps_per_second": 0.002,
      "step": 250
    }
  ],
  "logging_steps": 1,
  "max_steps": 475,
  "num_input_tokens_seen": 117598301,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}