{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.89588801399825,
  "eval_steps": 500,
  "global_step": 1536,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015764508928571397,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4069.0,
      "completions/mean_length": 603.188720703125,
      "completions/mean_terminated_length": 547.2442626953125,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "epoch": 0.004666083406240887,
      "grad_norm": 0.16825807094573975,
      "learning_rate": 1e-06,
      "loss": -0.0403,
      "num_tokens": 9160265.0,
      "reward": 0.4989537000656128,
      "reward_std": 0.2556326985359192,
      "rewards/simpleverify_reward/mean": 0.4989536702632904,
      "rewards/simpleverify_reward/std": 0.5000162720680237,
      "step": 1
    },
    {
      "clip_ratio/high_max": 0.0022431204488384537,
      "clip_ratio/high_mean": 0.0009628464467823505,
      "clip_ratio/low_mean": 0.0006316744875221048,
      "clip_ratio/low_min": 1.3116474292473868e-05,
      "clip_ratio/region_mean": 0.0015945209524943493,
      "epoch": 0.009332166812481774,
      "grad_norm": 0.15917454659938812,
      "learning_rate": 1e-06,
      "loss": 0.0233,
      "step": 2
    },
    {
      "clip_ratio/high_max": 0.0024326741622644477,
      "clip_ratio/high_mean": 0.0011160007470607525,
      "clip_ratio/low_mean": 0.0007444577131536789,
      "clip_ratio/low_min": 5.975654494250193e-05,
      "clip_ratio/region_mean": 0.0018604584111017175,
      "epoch": 0.01399825021872266,
      "grad_norm": 0.12686820328235626,
      "learning_rate": 1e-06,
      "loss": -0.0208,
      "step": 3
    },
    {
      "clip_ratio/high_max": 0.0024386195655097254,
      "clip_ratio/high_mean": 0.0011106257916253526,
      "clip_ratio/low_mean": 0.0009062907756742788,
      "clip_ratio/low_min": 8.275225536635844e-05,
      "clip_ratio/region_mean": 0.002016916587308515,
      "epoch": 0.018664333624963548,
      "grad_norm": 0.1381240040063858,
      "learning_rate": 1e-06,
      "loss": 0.06,
      "step": 4
    },
    {
      "clip_ratio/high_max": 0.002695557937840931,
      "clip_ratio/high_mean": 0.0011688486374623608,
      "clip_ratio/low_mean": 0.000957150463364087,
      "clip_ratio/low_min": 9.766997663973598e-05,
      "clip_ratio/region_mean": 0.0021259991044644266,
      "epoch": 0.023330417031204434,
      "grad_norm": 0.13304825127124786,
      "learning_rate": 1e-06,
      "loss": -0.0109,
      "step": 5
    },
    {
      "clip_ratio/high_max": 0.0027244676457485184,
      "clip_ratio/high_mean": 0.001221334961883258,
      "clip_ratio/low_mean": 0.0012220956159580965,
      "clip_ratio/low_min": 0.0001768647343851626,
      "clip_ratio/region_mean": 0.0024434305960312486,
      "epoch": 0.02799650043744532,
      "grad_norm": 0.13083291053771973,
      "learning_rate": 1e-06,
      "loss": -0.0095,
      "step": 6
    },
    {
      "clip_ratio/high_max": 0.0027735916155506857,
      "clip_ratio/high_mean": 0.0013042515602137428,
      "clip_ratio/low_mean": 0.0013594048214145005,
      "clip_ratio/low_min": 0.00015477509259653743,
      "clip_ratio/region_mean": 0.0026636563998181373,
      "epoch": 0.032662583843686206,
      "grad_norm": 0.12845644354820251,
      "learning_rate": 1e-06,
      "loss": -0.0029,
      "step": 7
    },
    {
      "clip_ratio/high_max": 0.003233741706935689,
      "clip_ratio/high_mean": 0.0015223221234919038,
      "clip_ratio/low_mean": 0.001188874084618874,
      "clip_ratio/low_min": 0.00014168537745717913,
      "clip_ratio/region_mean": 0.002711196182644926,
      "epoch": 0.037328667249927096,
      "grad_norm": 0.13185247778892517,
      "learning_rate": 1e-06,
      "loss": -0.0489,
      "step": 8
    },
    {
      "clip_ratio/high_max": 0.003264270315412432,
      "clip_ratio/high_mean": 0.0015198703003989067,
      "clip_ratio/low_mean": 0.0012674927584157558,
      "clip_ratio/low_min": 0.00025792716405703686,
      "clip_ratio/region_mean": 0.002787363038805779,
      "epoch": 0.04199475065616798,
      "grad_norm": 0.1370556801557541,
      "learning_rate": 1e-06,
      "loss": -0.0241,
      "step": 9
    },
    {
      "clip_ratio/high_max": 0.003038945986190811,
      "clip_ratio/high_mean": 0.0013063992992101703,
      "clip_ratio/low_mean": 0.0012288918114791159,
      "clip_ratio/low_min": 0.00022937654102861416,
      "clip_ratio/region_mean": 0.0025352911397931166,
      "epoch": 0.04666083406240887,
      "grad_norm": 0.13925476372241974,
      "learning_rate": 1e-06,
      "loss": 0.0092,
      "step": 10
    },
    {
      "clip_ratio/high_max": 0.002494553707947489,
      "clip_ratio/high_mean": 0.0012013574669254012,
      "clip_ratio/low_mean": 0.0013136883826518897,
      "clip_ratio/low_min": 0.0003180843195877969,
      "clip_ratio/region_mean": 0.0025150458968710154,
      "epoch": 0.05132691746864975,
      "grad_norm": 0.13018232583999634,
      "learning_rate": 1e-06,
      "loss": 0.0019,
      "step": 11
    },
    {
      "clip_ratio/high_max": 0.002584379901236389,
      "clip_ratio/high_mean": 0.0012154537653259467,
      "clip_ratio/low_mean": 0.0011643258076219354,
      "clip_ratio/low_min": 0.00020645100994443055,
      "clip_ratio/region_mean": 0.0023797796384315006,
      "epoch": 0.05599300087489064,
      "grad_norm": 0.12660029530525208,
      "learning_rate": 1e-06,
      "loss": 0.0087,
      "step": 12
    },
    {
      "clip_ratio/high_max": 0.002588285831734538,
      "clip_ratio/high_mean": 0.0012208971274958458,
      "clip_ratio/low_mean": 0.0010608917182253208,
      "clip_ratio/low_min": 7.698153967794497e-05,
      "clip_ratio/region_mean": 0.002281788816617336,
      "epoch": 0.06065908428113152,
      "grad_norm": 0.12888728082180023,
      "learning_rate": 1e-06,
      "loss": 0.0125,
      "step": 13
    },
    {
      "clip_ratio/high_max": 0.002168842482205946,
      "clip_ratio/high_mean": 0.0010724798994488083,
      "clip_ratio/low_mean": 0.0012188458531454671,
      "clip_ratio/low_min": 0.0003006507567988592,
      "clip_ratio/region_mean": 0.0022913257125765085,
      "epoch": 0.06532516768737241,
      "grad_norm": 0.12231365591287613,
      "learning_rate": 1e-06,
      "loss": 0.0022,
      "step": 14
    },
    {
      "clip_ratio/high_max": 0.002738081937422976,
      "clip_ratio/high_mean": 0.0012902747293992434,
      "clip_ratio/low_mean": 0.0010534348512010183,
      "clip_ratio/low_min": 0.00014821432250755606,
      "clip_ratio/region_mean": 0.0023437095442204736,
      "epoch": 0.0699912510936133,
      "grad_norm": 0.14451473951339722,
      "learning_rate": 1e-06,
      "loss": -0.0254,
      "step": 15
    },
    {
      "clip_ratio/high_max": 0.002559307074989192,
      "clip_ratio/high_mean": 0.0010771507968456717,
      "clip_ratio/low_mean": 0.0010660523912520148,
      "clip_ratio/low_min": 0.00013991356991027715,
      "clip_ratio/region_mean": 0.002143203171726782,
      "epoch": 0.07465733449985419,
      "grad_norm": 0.1317187398672104,
      "learning_rate": 1e-06,
      "loss": 0.0326,
      "step": 16
    },
    {
      "clip_ratio/high_max": 0.0035354366118554026,
      "clip_ratio/high_mean": 0.001526299245597329,
      "clip_ratio/low_mean": 0.001160885349236196,
      "clip_ratio/low_min": 9.711422944747028e-05,
      "clip_ratio/region_mean": 0.002687184649403207,
      "epoch": 0.07932341790609507,
      "grad_norm": 0.12933936715126038,
      "learning_rate": 1e-06,
      "loss": -0.0408,
      "step": 17
    },
    {
      "clip_ratio/high_max": 0.002537498548917938,
      "clip_ratio/high_mean": 0.0011709105401678244,
      "clip_ratio/low_mean": 0.0012086608294339385,
      "clip_ratio/low_min": 0.00017532482615933986,
      "clip_ratio/region_mean": 0.0023795713495928794,
      "epoch": 0.08398950131233596,
      "grad_norm": 0.12360536307096481,
      "learning_rate": 1e-06,
      "loss": 0.0229,
      "step": 18
    },
    {
      "clip_ratio/high_max": 0.0028355855683912523,
      "clip_ratio/high_mean": 0.0013620893587358296,
      "clip_ratio/low_mean": 0.0010958041602862068,
      "clip_ratio/low_min": 0.00014026590724824928,
      "clip_ratio/region_mean": 0.002457893548125867,
      "epoch": 0.08865558471857685,
      "grad_norm": 0.11669901013374329,
      "learning_rate": 1e-06,
      "loss": -0.0212,
      "step": 19
    },
    {
      "clip_ratio/high_max": 0.0028455474093789235,
      "clip_ratio/high_mean": 0.0012413982149155345,
      "clip_ratio/low_mean": 0.0013979854811623227,
      "clip_ratio/low_min": 0.00011787635321525158,
      "clip_ratio/region_mean": 0.0026393836669740267,
      "epoch": 0.09332166812481774,
      "grad_norm": 0.12858091294765472,
      "learning_rate": 1e-06,
      "loss": 0.0595,
      "step": 20
    },
    {
      "clip_ratio/high_max": 0.003188476213836111,
      "clip_ratio/high_mean": 0.0014160720565996598,
      "clip_ratio/low_mean": 0.0012823242032027338,
      "clip_ratio/low_min": 0.00020925595254084328,
      "clip_ratio/region_mean": 0.002698396216146648,
      "epoch": 0.09798775153105861,
      "grad_norm": 0.12983207404613495,
      "learning_rate": 1e-06,
      "loss": -0.0113,
      "step": 21
    },
    {
      "clip_ratio/high_max": 0.0028564574677147903,
      "clip_ratio/high_mean": 0.0013004683787585236,
      "clip_ratio/low_mean": 0.0011537495229276828,
      "clip_ratio/low_min": 0.0001961843572644284,
      "clip_ratio/region_mean": 0.0024542179380659945,
      "epoch": 0.1026538349372995,
      "grad_norm": 0.12031391263008118,
      "learning_rate": 1e-06,
      "loss": -0.0098,
      "step": 22
    },
    {
      "clip_ratio/high_max": 0.002843314148776699,
      "clip_ratio/high_mean": 0.00134471870114794,
      "clip_ratio/low_mean": 0.0012956337413925212,
      "clip_ratio/low_min": 5.7733907851797994e-05,
      "clip_ratio/region_mean": 0.0026403524607303552,
      "epoch": 0.10731991834354039,
      "grad_norm": 0.1164054349064827,
      "learning_rate": 1e-06,
      "loss": -0.0032,
      "step": 23
    },
    {
      "clip_ratio/high_max": 0.0032875677716219798,
      "clip_ratio/high_mean": 0.0014806390063313302,
      "clip_ratio/low_mean": 0.0011173126149515156,
      "clip_ratio/low_min": 7.565274154330837e-05,
      "clip_ratio/region_mean": 0.0025979516794905066,
      "epoch": 0.11198600174978128,
      "grad_norm": 0.13383948802947998,
      "learning_rate": 1e-06,
      "loss": -0.0492,
      "step": 24
    },
    {
      "clip_ratio/high_max": 0.0032452556770294905,
      "clip_ratio/high_mean": 0.0015567433474643622,
      "clip_ratio/low_mean": 0.0013572097213909728,
      "clip_ratio/low_min": 0.0002881341588363284,
      "clip_ratio/region_mean": 0.0029139530088286847,
      "epoch": 0.11665208515602217,
      "grad_norm": 0.12567749619483948,
      "learning_rate": 1e-06,
      "loss": -0.0245,
      "step": 25
    },
    {
      "clip_ratio/high_max": 0.002892524833441712,
      "clip_ratio/high_mean": 0.00136099345763796,
      "clip_ratio/low_mean": 0.0014545994272339158,
      "clip_ratio/low_min": 0.00016666963801981183,
      "clip_ratio/region_mean": 0.0028155929103377275,
      "epoch": 0.12131816856226305,
      "grad_norm": 0.12536926567554474,
      "learning_rate": 1e-06,
      "loss": 0.0088,
      "step": 26
    },
    {
      "clip_ratio/high_max": 0.00290807215787936,
      "clip_ratio/high_mean": 0.0013549137729569338,
      "clip_ratio/low_mean": 0.001522806011053035,
      "clip_ratio/low_min": 0.00029649123280250933,
      "clip_ratio/region_mean": 0.002877719874959439,
      "epoch": 0.12598425196850394,
      "grad_norm": 0.12423344701528549,
      "learning_rate": 1e-06,
      "loss": 0.0015,
      "step": 27
    },
    {
      "clip_ratio/high_max": 0.002823303402692545,
      "clip_ratio/high_mean": 0.0013742239534622058,
      "clip_ratio/low_mean": 0.0015396033850265667,
      "clip_ratio/low_min": 0.00020346302972029662,
      "clip_ratio/region_mean": 0.002913827273005154,
      "epoch": 0.13065033537474482,
      "grad_norm": 0.12013071775436401,
      "learning_rate": 1e-06,
      "loss": 0.0084,
      "step": 28
    },
    {
      "clip_ratio/high_max": 0.0029015862819505855,
      "clip_ratio/high_mean": 0.0014054770181246568,
      "clip_ratio/low_mean": 0.0014502845551760402,
      "clip_ratio/low_min": 0.00013019296420679893,
      "clip_ratio/region_mean": 0.0028557615587487817,
      "epoch": 0.13531641878098571,
      "grad_norm": 0.12109560519456863,
      "learning_rate": 1e-06,
      "loss": 0.0122,
      "step": 29
    },
    {
      "clip_ratio/high_max": 0.0030319598299684003,
      "clip_ratio/high_mean": 0.0013882502607884817,
      "clip_ratio/low_mean": 0.0016965941358648706,
      "clip_ratio/low_min": 0.000465552186142304,
      "clip_ratio/region_mean": 0.0030848444293951616,
      "epoch": 0.1399825021872266,
      "grad_norm": 0.1187513992190361,
      "learning_rate": 1e-06,
      "loss": 0.0018,
      "step": 30
    },
    {
      "clip_ratio/high_max": 0.003424561276915483,
      "clip_ratio/high_mean": 0.001583503617439419,
      "clip_ratio/low_mean": 0.0015595244040014222,
      "clip_ratio/low_min": 0.00025373905373271555,
      "clip_ratio/region_mean": 0.0031430279414053075,
      "epoch": 0.1446485855934675,
      "grad_norm": 0.14122727513313293,
      "learning_rate": 1e-06,
      "loss": -0.0257,
      "step": 31
    },
    {
      "clip_ratio/high_max": 0.0030605307983933017,
      "clip_ratio/high_mean": 0.001422762838046765,
      "clip_ratio/low_mean": 0.001624619500944391,
      "clip_ratio/low_min": 0.00013118834613123909,
      "clip_ratio/region_mean": 0.0030473822698695585,
      "epoch": 0.14931466899970838,
      "grad_norm": 0.12433193624019623,
      "learning_rate": 1e-06,
      "loss": 0.0322,
      "step": 32
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.013881138392857095,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4041.0,
      "completions/mean_length": 611.9447021484375,
      "completions/mean_terminated_length": 562.9012451171875,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.15398075240594924,
      "grad_norm": 0.13136796653270721,
      "learning_rate": 1e-06,
      "loss": -0.0096,
      "num_tokens": 18527584.0,
      "reward": 0.5294364094734192,
      "reward_std": 0.2430260330438614,
      "rewards/simpleverify_reward/mean": 0.5294364094734192,
      "rewards/simpleverify_reward/std": 0.4991501271724701,
      "step": 33
    },
    {
      "clip_ratio/high_max": 0.0023058708175085485,
      "clip_ratio/high_mean": 0.0010608333941490855,
      "clip_ratio/low_mean": 0.0006683274905299186,
      "clip_ratio/low_min": 2.6147899006900843e-05,
      "clip_ratio/region_mean": 0.0017291609110543504,
      "epoch": 0.15864683581219013,
      "grad_norm": 0.14291034638881683,
      "learning_rate": 1e-06,
      "loss": 0.0104,
      "step": 34
    },
    {
      "clip_ratio/high_max": 0.0024183506538975053,
      "clip_ratio/high_mean": 0.001051281866239151,
      "clip_ratio/low_mean": 0.0006239201175048947,
      "clip_ratio/low_min": 3.9680769987171516e-05,
      "clip_ratio/region_mean": 0.001675202016485855,
      "epoch": 0.16331291921843102,
      "grad_norm": 0.13595029711723328,
      "learning_rate": 1e-06,
      "loss": -0.0099,
      "step": 35
    },
    {
      "clip_ratio/high_max": 0.002563152498623822,
      "clip_ratio/high_mean": 0.0010042754547612276,
      "clip_ratio/low_mean": 0.000609123410868051,
      "clip_ratio/low_min": 4.9355025112163275e-05,
      "clip_ratio/region_mean": 0.0016133989120135084,
      "epoch": 0.1679790026246719,
      "grad_norm": 0.1255405843257904,
      "learning_rate": 1e-06,
      "loss": -0.0064,
      "step": 36
    },
    {
      "clip_ratio/high_max": 0.002357198747631628,
      "clip_ratio/high_mean": 0.0010236527887172997,
      "clip_ratio/low_mean": 0.0007413587427436141,
      "clip_ratio/low_min": 4.127016381971771e-05,
      "clip_ratio/region_mean": 0.0017650115405558608,
      "epoch": 0.1726450860309128,
      "grad_norm": 0.1268312633037567,
      "learning_rate": 1e-06,
      "loss": 0.0199,
      "step": 37
    },
    {
      "clip_ratio/high_max": 0.002156820279196836,
      "clip_ratio/high_mean": 0.0010005908807215746,
      "clip_ratio/low_mean": 0.0007532192448707065,
      "clip_ratio/low_min": 7.858727713028202e-05,
      "clip_ratio/region_mean": 0.0017538101237732917,
      "epoch": 0.1773111694371537,
      "grad_norm": 0.14367766678333282,
      "learning_rate": 1e-06,
      "loss": -0.0034,
      "step": 38
    },
    {
      "clip_ratio/high_max": 0.002738971532380674,
      "clip_ratio/high_mean": 0.0011742145397874992,
      "clip_ratio/low_mean": 0.000790875581515138,
      "clip_ratio/low_min": 6.495592515420867e-05,
      "clip_ratio/region_mean": 0.0019650901085697114,
      "epoch": 0.18197725284339458,
      "grad_norm": 0.13926085829734802,
      "learning_rate": 1e-06,
      "loss": -0.018,
      "step": 39
    },
    {
      "clip_ratio/high_max": 0.0023046898932079785,
      "clip_ratio/high_mean": 0.0011357082203176105,
      "clip_ratio/low_mean": 0.0009107508940360276,
      "clip_ratio/low_min": 0.00013045053947280394,
      "clip_ratio/region_mean": 0.002046459099801723,
      "epoch": 0.18664333624963547,
      "grad_norm": 0.1315917670726776,
      "learning_rate": 1e-06,
      "loss": -0.0079,
      "step": 40
    },
    {
      "clip_ratio/high_max": 0.0023842503360356204,
      "clip_ratio/high_mean": 0.00112566656753188,
      "clip_ratio/low_mean": 0.0010515346421016147,
      "clip_ratio/low_min": 0.00015354276274592848,
      "clip_ratio/region_mean": 0.002177201175072696,
      "epoch": 0.19130941965587636,
      "grad_norm": 0.13507364690303802,
      "learning_rate": 1e-06,
      "loss": -0.0215,
      "step": 41
    },
    {
      "clip_ratio/high_max": 0.0026018995922640897,
      "clip_ratio/high_mean": 0.00112384133899468,
      "clip_ratio/low_mean": 0.001048213771355222,
      "clip_ratio/low_min": 0.0001181974903374794,
      "clip_ratio/region_mean": 0.0021720550503232516,
      "epoch": 0.19597550306211722,
      "grad_norm": 0.11782612651586533,
      "learning_rate": 1e-06,
      "loss": 0.0084,
      "step": 42
    },
    {
      "clip_ratio/high_max": 0.0026973372441716492,
      "clip_ratio/high_mean": 0.001119166085118195,
      "clip_ratio/low_mean": 0.0011986727222392801,
      "clip_ratio/low_min": 0.00011101602831331547,
      "clip_ratio/region_mean": 0.0023178388437372632,
      "epoch": 0.2006415864683581,
      "grad_norm": 0.1235799640417099,
      "learning_rate": 1e-06,
      "loss": 0.0455,
      "step": 43
    },
    {
      "clip_ratio/high_max": 0.002128927102603484,
      "clip_ratio/high_mean": 0.0009653180422901642,
      "clip_ratio/low_mean": 0.0012455816759029403,
      "clip_ratio/low_min": 0.0001529217747702205,
      "clip_ratio/region_mean": 0.0022108996636234224,
      "epoch": 0.205307669874599,
      "grad_norm": 0.13647939264774323,
      "learning_rate": 1e-06,
      "loss": 0.056,
      "step": 44
    },
    {
      "clip_ratio/high_max": 0.0025862012844299898,
      "clip_ratio/high_mean": 0.001224781652126694,
      "clip_ratio/low_mean": 0.0011133782591059571,
      "clip_ratio/low_min": 0.00016104968563013244,
      "clip_ratio/region_mean": 0.002338159865757916,
      "epoch": 0.2099737532808399,
      "grad_norm": 0.12769809365272522,
      "learning_rate": 1e-06,
      "loss": 0.0086,
      "step": 45
    },
    {
      "clip_ratio/high_max": 0.0024678852132637985,
      "clip_ratio/high_mean": 0.0012310537495068274,
      "clip_ratio/low_mean": 0.0011564890082809143,
      "clip_ratio/low_min": 0.00017939623103302438,
      "clip_ratio/region_mean": 0.0023875428523751907,
      "epoch": 0.21463983668708078,
      "grad_norm": 0.12696535885334015,
      "learning_rate": 1e-06,
      "loss": -0.0188,
      "step": 46
    },
    {
      "clip_ratio/high_max": 0.002291153563419357,
      "clip_ratio/high_mean": 0.0011049769491364714,
      "clip_ratio/low_mean": 0.001051378101692535,
      "clip_ratio/low_min": 9.607033734937431e-05,
      "clip_ratio/region_mean": 0.002156354967155494,
      "epoch": 0.21930592009332167,
      "grad_norm": 0.12068193405866623,
      "learning_rate": 1e-06,
      "loss": -0.0154,
      "step": 47
    },
    {
      "clip_ratio/high_max": 0.0023848756100051105,
      "clip_ratio/high_mean": 0.0010859482645173557,
      "clip_ratio/low_mean": 0.0011789464060711907,
      "clip_ratio/low_min": 0.00021505491713469382,
      "clip_ratio/region_mean": 0.0022648946614935994,
      "epoch": 0.22397200349956256,
      "grad_norm": 0.11583662778139114,
      "learning_rate": 1e-06,
      "loss": 0.0199,
      "step": 48
    },
    {
      "clip_ratio/high_max": 0.003066568140638992,
      "clip_ratio/high_mean": 0.0013708297665289138,
      "clip_ratio/low_mean": 0.0013199262248235755,
      "clip_ratio/low_min": 0.00017432043205189984,
      "clip_ratio/region_mean": 0.0026907560022664256,
      "epoch": 0.22863808690580345,
      "grad_norm": 0.1144961416721344,
      "learning_rate": 1e-06,
      "loss": -0.0099,
      "step": 49
    },
    {
      "clip_ratio/high_max": 0.0027678100450430065,
      "clip_ratio/high_mean": 0.001266906912860577,
      "clip_ratio/low_mean": 0.0013194326420489233,
      "clip_ratio/low_min": 0.00020752859018102754,
      "clip_ratio/region_mean": 0.002586339571280405,
      "epoch": 0.23330417031204434,
      "grad_norm": 0.123209647834301,
      "learning_rate": 1e-06,
      "loss": 0.01,
      "step": 50
    },
    {
      "clip_ratio/high_max": 0.00342193063988816,
      "clip_ratio/high_mean": 0.0014160521241137758,
      "clip_ratio/low_mean": 0.0011307128879707307,
      "clip_ratio/low_min": 0.0001739427934808191,
      "clip_ratio/region_mean": 0.002546764982980676,
      "epoch": 0.2379702537182852,
      "grad_norm": 0.11714783310890198,
      "learning_rate": 1e-06,
      "loss": -0.0102,
      "step": 51
    },
    {
      "clip_ratio/high_max": 0.0029763600759906694,
      "clip_ratio/high_mean": 0.001263724894670304,
      "clip_ratio/low_mean": 0.0011015970521839336,
      "clip_ratio/low_min": 0.00015545579844911117,
      "clip_ratio/region_mean": 0.0023653219468542375,
      "epoch": 0.2426363371245261,
      "grad_norm": 0.12042421102523804,
      "learning_rate": 1e-06,
      "loss": -0.0067,
      "step": 52
    },
    {
      "clip_ratio/high_max": 0.0031162177911028266,
      "clip_ratio/high_mean": 0.001347083663858939,
      "clip_ratio/low_mean": 0.0012247143713466357,
      "clip_ratio/low_min": 0.00013263207893032813,
      "clip_ratio/region_mean": 0.0025717980024637654,
      "epoch": 0.24730242053076698,
      "grad_norm": 0.1240837574005127,
      "learning_rate": 1e-06,
      "loss": 0.0196,
      "step": 53
    },
    {
      "clip_ratio/high_max": 0.0027978752623312175,
      "clip_ratio/high_mean": 0.0012512018547568005,
      "clip_ratio/low_mean": 0.0011979926312051248,
      "clip_ratio/low_min": 0.00014231304703571368,
      "clip_ratio/region_mean": 0.0024491945368936285,
      "epoch": 0.25196850393700787,
      "grad_norm": 0.10426893085241318,
      "learning_rate": 1e-06,
      "loss": -0.0037,
      "step": 54
    },
    {
      "clip_ratio/high_max": 0.003130602119199466,
      "clip_ratio/high_mean": 0.0013899867408326827,
      "clip_ratio/low_mean": 0.0011404099532228429,
      "clip_ratio/low_min": 3.0508501367876306e-05,
      "clip_ratio/region_mean": 0.0025303966394858435,
      "epoch": 0.2566345873432488,
      "grad_norm": 0.12543822824954987,
      "learning_rate": 1e-06,
      "loss": -0.0185,
      "step": 55
    },
    {
      "clip_ratio/high_max": 0.0028997409463045187,
      "clip_ratio/high_mean": 0.0013613371520477813,
      "clip_ratio/low_mean": 0.0013036046475463081,
      "clip_ratio/low_min": 0.0003099424266110873,
      "clip_ratio/region_mean": 0.0026649418141460046,
      "epoch": 0.26130067074948965,
      "grad_norm": 0.1263367384672165,
      "learning_rate": 1e-06,
      "loss": -0.0083,
      "step": 56
    },
    {
      "clip_ratio/high_max": 0.0027787757862824947,
      "clip_ratio/high_mean": 0.0012391031195875257,
      "clip_ratio/low_mean": 0.0012746982702083187,
      "clip_ratio/low_min": 0.00015071545385580976,
      "clip_ratio/region_mean": 0.0025138013807008974,
      "epoch": 0.2659667541557305,
      "grad_norm": 0.12137658149003983,
      "learning_rate": 1e-06,
      "loss": -0.0219,
      "step": 57
    },
    {
      "clip_ratio/high_max": 0.002790854974591639,
      "clip_ratio/high_mean": 0.001271753182663815,
      "clip_ratio/low_mean": 0.00129299901163904,
      "clip_ratio/low_min": 0.0001493587233198923,
      "clip_ratio/region_mean": 0.0025647522052167915,
      "epoch": 0.27063283756197143,
      "grad_norm": 0.10908447206020355,
      "learning_rate": 1e-06,
      "loss": 0.0081,
      "step": 58
    },
    {
      "clip_ratio/high_max": 0.0030602671758970246,
      "clip_ratio/high_mean": 0.0011957619171880651,
      "clip_ratio/low_mean": 0.0014619633002439514,
      "clip_ratio/low_min": 0.00017814520015235757,
      "clip_ratio/region_mean": 0.0026577251992421225,
      "epoch": 0.2752989209682123,
      "grad_norm": 0.11459843069314957,
      "learning_rate": 1e-06,
      "loss": 0.0452,
      "step": 59
    },
    {
      "clip_ratio/high_max": 0.00275232477724785,
      "clip_ratio/high_mean": 0.0011349596425134223,
      "clip_ratio/low_mean": 0.0015506329109484795,
      "clip_ratio/low_min": 0.00018194458971265703,
      "clip_ratio/region_mean": 0.002685592553461902,
      "epoch": 0.2799650043744532,
      "grad_norm": 0.1261429339647293,
      "learning_rate": 1e-06,
      "loss": 0.0556,
      "step": 60
    },
    {
      "clip_ratio/high_max": 0.003160063140967395,
      "clip_ratio/high_mean": 0.0014107936622167472,
      "clip_ratio/low_mean": 0.0015674529240641277,
      "clip_ratio/low_min": 0.0002114010785589926,
      "clip_ratio/region_mean": 0.0029782466663164087,
      "epoch": 0.28463108778069407,
      "grad_norm": 0.11813414096832275,
      "learning_rate": 1e-06,
      "loss": 0.0082,
      "step": 61
    },
    {
      "clip_ratio/high_max": 0.0030250746640376747,
      "clip_ratio/high_mean": 0.0014572870677511673,
      "clip_ratio/low_mean": 0.0014617256565543357,
      "clip_ratio/low_min": 0.00023663033152843127,
      "clip_ratio/region_mean": 0.002919012746133376,
      "epoch": 0.289297171186935,
      "grad_norm": 0.1186414286494255,
      "learning_rate": 1e-06,
      "loss": -0.0192,
      "step": 62
    },
    {
      "clip_ratio/high_max": 0.003208254143828526,
      "clip_ratio/high_mean": 0.0014186432017595507,
      "clip_ratio/low_mean": 0.0014737911806150805,
      "clip_ratio/low_min": 0.00011337467549310531,
      "clip_ratio/region_mean": 0.002892434407840483,
      "epoch": 0.29396325459317585,
      "grad_norm": 0.11375010013580322,
      "learning_rate": 1e-06,
      "loss": -0.0158,
      "step": 63
    },
    {
      "clip_ratio/high_max": 0.002864666279492667,
      "clip_ratio/high_mean": 0.0012809197560272878,
      "clip_ratio/low_mean": 0.0015702278979006223,
      "clip_ratio/low_min": 0.00024213889810198452,
      "clip_ratio/region_mean": 0.002851147648470942,
      "epoch": 0.29862933799941677,
      "grad_norm": 0.11020326614379883,
      "learning_rate": 1e-06,
      "loss": 0.0196,
      "step": 64
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.014508928571428603,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4070.0,
      "completions/mean_length": 609.0919189453125,
      "completions/mean_terminated_length": 557.7557373046875,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.30329542140565763,
      "grad_norm": 0.1333010196685791,
      "learning_rate": 1e-06,
      "loss": 0.0193,
      "num_tokens": 27844941.0,
      "reward": 0.5291574001312256,
      "reward_std": 0.23031266033649445,
      "rewards/simpleverify_reward/mean": 0.5291573405265808,
      "rewards/simpleverify_reward/std": 0.4991665184497833,
      "step": 65
    },
    {
      "clip_ratio/high_max": 0.002095986492349766,
      "clip_ratio/high_mean": 0.0009549665555823594,
      "clip_ratio/low_mean": 0.0006022504849170218,
      "clip_ratio/low_min": 6.761079293937655e-05,
      "clip_ratio/region_mean": 0.0015572170159430243,
      "epoch": 0.3079615048118985,
      "grad_norm": 0.11942674964666367,
      "learning_rate": 1e-06,
      "loss": -0.0201,
      "step": 66
    },
    {
      "clip_ratio/high_max": 0.002184139368182514,
      "clip_ratio/high_mean": 0.0009601217097952031,
      "clip_ratio/low_mean": 0.0006109681016823743,
      "clip_ratio/low_min": 4.145236925978679e-05,
      "clip_ratio/region_mean": 0.0015710898514953442,
      "epoch": 0.3126275882181394,
      "grad_norm": 0.1350686401128769,
      "learning_rate": 1e-06,
      "loss": 0.0108,
      "step": 67
    },
    {
      "clip_ratio/high_max": 0.0021282443776726723,
      "clip_ratio/high_mean": 0.0010061018001579214,
      "clip_ratio/low_mean": 0.0006382428437063936,
      "clip_ratio/low_min": 1.2626263014681172e-05,
      "clip_ratio/region_mean": 0.0016443446293123998,
      "epoch": 0.31729367162438027,
      "grad_norm": 0.12547576427459717,
      "learning_rate": 1e-06,
      "loss": -0.0125,
      "step": 68
    },
    {
      "clip_ratio/high_max": 0.0022075731540098786,
      "clip_ratio/high_mean": 0.0009602974023437127,
      "clip_ratio/low_mean": 0.0007607722145621665,
      "clip_ratio/low_min": 8.331664412253303e-05,
      "clip_ratio/region_mean": 0.0017210696532856673,
      "epoch": 0.3219597550306212,
      "grad_norm": 0.13183489441871643,
      "learning_rate": 1e-06,
      "loss": 0.0434,
      "step": 69
    },
    {
      "clip_ratio/high_max": 0.002173729895730503,
      "clip_ratio/high_mean": 0.0009830275594140403,
      "clip_ratio/low_mean": 0.0008126407901727362,
      "clip_ratio/low_min": 9.501906424702611e-05,
      "clip_ratio/region_mean": 0.0017956683441298082,
      "epoch": 0.32662583843686205,
      "grad_norm": 0.1218208521604538,
      "learning_rate": 1e-06,
      "loss": -0.0121,
      "step": 70
    },
    {
      "clip_ratio/high_max": 0.002139653734047897,
      "clip_ratio/high_mean": 0.0009312951751780929,
      "clip_ratio/low_mean": 0.0009362947894260287,
      "clip_ratio/low_min": 0.00014902820930728922,
      "clip_ratio/region_mean": 0.0018675899700610898,
      "epoch": 0.33129192184310297,
      "grad_norm": 0.11886744946241379,
      "learning_rate": 1e-06,
      "loss": 0.0016,
      "step": 71
    },
    {
      "clip_ratio/high_max": 0.002137196312105516,
      "clip_ratio/high_mean": 0.0009359336963825626,
      "clip_ratio/low_mean": 0.0010420407306810375,
      "clip_ratio/low_min": 7.669547994737513e-05,
      "clip_ratio/region_mean": 0.0019779744325205684,
      "epoch": 0.3359580052493438,
      "grad_norm": 0.11734430491924286,
      "learning_rate": 1e-06,
      "loss": 0.0147,
      "step": 72
    },
    {
      "clip_ratio/high_max": 0.0020253926631994545,
      "clip_ratio/high_mean": 0.0009318325828644447,
      "clip_ratio/low_mean": 0.0011065478720411193,
      "clip_ratio/low_min": 9.858188423095271e-05,
      "clip_ratio/region_mean": 0.0020383804221637547,
      "epoch": 0.34062408865558474,
      "grad_norm": 0.12952370941638947,
      "learning_rate": 1e-06,
      "loss": 0.0019,
      "step": 73
    },
    {
      "clip_ratio/high_max": 0.0030815795544185676,
      "clip_ratio/high_mean": 0.0012394871810101904,
      "clip_ratio/low_mean": 0.0010314514183846768,
      "clip_ratio/low_min": 0.00010130217287951382,
      "clip_ratio/region_mean": 0.002270938639412634,
      "epoch": 0.3452901720618256,
      "grad_norm": 0.11882374435663223,
      "learning_rate": 1e-06,
      "loss": 0.0008,
      "step": 74
    },
    {
      "clip_ratio/high_max": 0.002301025109773036,
      "clip_ratio/high_mean": 0.0010060086569865234,
      "clip_ratio/low_mean": 0.001249170585651882,
      "clip_ratio/low_min": 0.00020350982686068164,
      "clip_ratio/region_mean": 0.002255179249914363,
      "epoch": 0.34995625546806647,
      "grad_norm": 0.118986114859581,
      "learning_rate": 1e-06,
      "loss": 0.0166,
      "step": 75
    },
    {
      "clip_ratio/high_max": 0.0022405715426430106,
      "clip_ratio/high_mean": 0.0010710069691413082,
      "clip_ratio/low_mean": 0.0012152361778134946,
      "clip_ratio/low_min": 0.00017451080293540144,
      "clip_ratio/region_mean": 0.0022862431142129935,
      "epoch": 0.3546223388743074,
      "grad_norm": 0.1328345239162445,
      "learning_rate": 1e-06,
      "loss": -0.0105,
      "step": 76
    },
    {
      "clip_ratio/high_max": 0.0024153872072929516,
      "clip_ratio/high_mean": 0.001028700189635856,
      "clip_ratio/low_mean": 0.001207942441396881,
      "clip_ratio/low_min": 2.5003879272844642e-05,
      "clip_ratio/region_mean": 0.0022366426419466734,
      "epoch": 0.35928842228054825,
      "grad_norm": 0.12570519745349884,
      "learning_rate": 1e-06,
      "loss": -0.0072,
      "step": 77
    },
    {
      "clip_ratio/high_max": 0.002154354573576711,
      "clip_ratio/high_mean": 0.0009097466227103723,
      "clip_ratio/low_mean": 0.0012079304033250082,
      "clip_ratio/low_min": 0.00019134853755531367,
      "clip_ratio/region_mean": 0.0021176769951125607,
      "epoch": 0.36395450568678916,
      "grad_norm": 0.12000148743391037,
      "learning_rate": 1e-06,
      "loss": 0.0441,
      "step": 78
    },
    {
      "clip_ratio/high_max": 0.002160135132726282,
      "clip_ratio/high_mean": 0.0009694766740722116,
      "clip_ratio/low_mean": 0.0011317591342958622,
      "clip_ratio/low_min": 0.00012337863336142618,
      "clip_ratio/region_mean": 0.0021012357537983917,
      "epoch": 0.36862058909303,
      "grad_norm": 0.13037844002246857,
      "learning_rate": 1e-06,
      "loss": 0.0074,
      "step": 79
    },
    {
      "clip_ratio/high_max": 0.0026232917880406603,
      "clip_ratio/high_mean": 0.0011170122197654564,
      "clip_ratio/low_mean": 0.0012029362114844844,
      "clip_ratio/low_min": 0.00013459848742058966,
      "clip_ratio/region_mean": 0.002319948427611962,
      "epoch": 0.37328667249927094,
      "grad_norm": 0.13097058236598969,
      "learning_rate": 1e-06,
      "loss": -0.0518,
      "step": 80
    },
    {
      "clip_ratio/high_max": 0.0029030659279669635,
      "clip_ratio/high_mean": 0.0013486304742400534,
      "clip_ratio/low_mean": 0.0015102603938430548,
      "clip_ratio/low_min": 0.00024855359515640885,
      "clip_ratio/region_mean": 0.0028588907880475745,
      "epoch": 0.3779527559055118,
      "grad_norm": 0.12466525286436081,
      "learning_rate": 1e-06,
      "loss": 0.0189,
      "step": 81
    },
    {
      "clip_ratio/high_max": 0.0029934592021163553,
      "clip_ratio/high_mean": 0.0012536262402136344,
      "clip_ratio/low_mean": 0.0013735974571318366,
      "clip_ratio/low_min": 0.0002149079400624032,
      "clip_ratio/region_mean": 0.0026272237082594074,
      "epoch": 0.3826188393117527,
      "grad_norm": 0.11244761943817139,
      "learning_rate": 1e-06,
      "loss": -0.0205,
      "step": 82
    },
    {
      "clip_ratio/high_max": 0.0031377016493934207,
      "clip_ratio/high_mean": 0.001302192387811374,
      "clip_ratio/low_mean": 0.0014633027858508285,
      "clip_ratio/low_min": 0.00017245216440642253,
      "clip_ratio/region_mean": 0.0027654952282318845,
      "epoch": 0.3872849227179936,
      "grad_norm": 0.12036320567131042,
      "learning_rate": 1e-06,
      "loss": 0.0103,
      "step": 83
    },
    {
      "clip_ratio/high_max": 0.0028718647590721957,
      "clip_ratio/high_mean": 0.0012610847988980822,
      "clip_ratio/low_mean": 0.0012295683700358495,
      "clip_ratio/low_min": 4.059804996359162e-05,
      "clip_ratio/region_mean": 0.0024906531543820165,
      "epoch": 0.39195100612423445,
      "grad_norm": 0.1192980483174324,
      "learning_rate": 1e-06,
      "loss": -0.0129,
      "step": 84
    },
    {
      "clip_ratio/high_max": 0.002950182279164437,
      "clip_ratio/high_mean": 0.0012620100169442594,
      "clip_ratio/low_mean": 0.0014633037244493607,
      "clip_ratio/low_min": 0.000233894261327805,
      "clip_ratio/region_mean": 0.0027253137523075566,
      "epoch": 0.39661708953047536,
      "grad_norm": 0.12224233150482178,
      "learning_rate": 1e-06,
      "loss": 0.0429,
      "step": 85
    },
    {
      "clip_ratio/high_max": 0.002965279338241089,
      "clip_ratio/high_mean": 0.0012996648183616344,
      "clip_ratio/low_mean": 0.0012757712611346506,
      "clip_ratio/low_min": 0.0001809010464057792,
      "clip_ratio/region_mean": 0.002575436039478518,
      "epoch": 0.4012831729367162,
      "grad_norm": 0.12633928656578064,
      "learning_rate": 1e-06,
      "loss": -0.0124,
      "step": 86
    },
    {
      "clip_ratio/high_max": 0.0028046031438861974,
      "clip_ratio/high_mean": 0.001274156995350495,
      "clip_ratio/low_mean": 0.0014248448969738092,
      "clip_ratio/low_min": 0.00025866729174595093,
      "clip_ratio/region_mean": 0.002699001932342071,
      "epoch": 0.40594925634295714,
      "grad_norm": 0.11714571714401245,
      "learning_rate": 1e-06,
      "loss": 0.0012,
      "step": 87
    },
    {
      "clip_ratio/high_max": 0.0030716592737007886,
      "clip_ratio/high_mean": 0.001227542825290584,
      "clip_ratio/low_mean": 0.001435277812561253,
      "clip_ratio/low_min": 0.0001626812754693674,
      "clip_ratio/region_mean": 0.0026628206978784874,
      "epoch": 0.410615339749198,
      "grad_norm": 0.11476991325616837,
      "learning_rate": 1e-06,
      "loss": 0.0143,
      "step": 88
    },
    {
      "clip_ratio/high_max": 0.002687366693862714,
      "clip_ratio/high_mean": 0.0011354204143572133,
      "clip_ratio/low_mean": 0.0014332951759570278,
      "clip_ratio/low_min": 0.00024562145063100616,
      "clip_ratio/region_mean": 0.0025687156958156265,
      "epoch": 0.4152814231554389,
      "grad_norm": 0.12055323272943497,
      "learning_rate": 1e-06,
      "loss": 0.0015,
      "step": 89
    },
    {
      "clip_ratio/high_max": 0.0032671166482032277,
      "clip_ratio/high_mean": 0.0014134002558421344,
      "clip_ratio/low_mean": 0.0013541813932533842,
      "clip_ratio/low_min": 0.00013320259586180327,
      "clip_ratio/region_mean": 0.002767581660009455,
      "epoch": 0.4199475065616798,
      "grad_norm": 0.11246298253536224,
      "learning_rate": 1e-06,
      "loss": 0.0003,
      "step": 90
    },
    {
      "clip_ratio/high_max": 0.0025632146789575927,
      "clip_ratio/high_mean": 0.0012043046335747931,
      "clip_ratio/low_mean": 0.001762463383784052,
      "clip_ratio/low_min": 0.00019503739258652786,
      "clip_ratio/region_mean": 0.002966768079204485,
      "epoch": 0.4246135899679207,
      "grad_norm": 0.11150757968425751,
      "learning_rate": 1e-06,
      "loss": 0.0162,
      "step": 91
    },
    {
      "clip_ratio/high_max": 0.0029760202378383838,
      "clip_ratio/high_mean": 0.001375091349473223,
      "clip_ratio/low_mean": 0.0016609986232651863,
      "clip_ratio/low_min": 0.00021439711963466834,
      "clip_ratio/region_mean": 0.0030360899545485154,
      "epoch": 0.42927967337416156,
      "grad_norm": 0.12024401128292084,
      "learning_rate": 1e-06,
      "loss": -0.011,
      "step": 92
    },
    {
      "clip_ratio/high_max": 0.003003647507284768,
      "clip_ratio/high_mean": 0.0012914321414427832,
      "clip_ratio/low_mean": 0.0017726602163747884,
      "clip_ratio/low_min": 0.00015167150559136644,
      "clip_ratio/region_mean": 0.0030640923359896988,
      "epoch": 0.4339457567804024,
      "grad_norm": 0.11853361129760742,
      "learning_rate": 1e-06,
      "loss": -0.0078,
      "step": 93
    },
    {
      "clip_ratio/high_max": 0.002971899950352963,
      "clip_ratio/high_mean": 0.0012267363817954902,
      "clip_ratio/low_mean": 0.0017872519238153473,
      "clip_ratio/low_min": 0.00029699699371121824,
      "clip_ratio/region_mean": 0.003013988316524774,
      "epoch": 0.43861184018664334,
      "grad_norm": 0.10903094708919525,
      "learning_rate": 1e-06,
      "loss": 0.0437,
      "step": 94
    },
    {
      "clip_ratio/high_max": 0.0027418189420131966,
      "clip_ratio/high_mean": 0.0012564454918901902,
      "clip_ratio/low_mean": 0.0016971448421827517,
      "clip_ratio/low_min": 0.00018493739662517328,
      "clip_ratio/region_mean": 0.002953590315883048,
      "epoch": 0.4432779235928842,
      "grad_norm": 0.12291538715362549,
      "learning_rate": 1e-06,
      "loss": 0.0068,
      "step": 95
    },
    {
      "clip_ratio/high_max": 0.003106837932136841,
      "clip_ratio/high_mean": 0.0014011914317961782,
      "clip_ratio/low_mean": 0.001698844149359502,
      "clip_ratio/low_min": 0.0002623428499646252,
      "clip_ratio/region_mean": 0.003100035566603765,
      "epoch": 0.4479440069991251,
      "grad_norm": 0.1240989938378334,
      "learning_rate": 1e-06,
      "loss": -0.0523,
      "step": 96
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0165318080357143,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4018.0,
      "completions/mean_length": 603.2432861328125,
      "completions/mean_terminated_length": 544.5310668945312,
      "completions/min_length": 16.0,
      "completions/min_terminated_length": 16.0,
      "epoch": 0.452610090405366,
      "grad_norm": 0.13344593346118927,
      "learning_rate": 1e-06,
      "loss": 0.0362,
      "num_tokens": 36922308.0,
      "reward": 0.5622209906578064,
      "reward_std": 0.21516825258731842,
      "rewards/simpleverify_reward/mean": 0.5622209906578064,
      "rewards/simpleverify_reward/std": 0.4961307644844055,
      "step": 97
    },
    {
      "clip_ratio/high_max": 0.003061515380977653,
      "clip_ratio/high_mean": 0.001179942266389844,
      "clip_ratio/low_mean": 0.000555939896912605,
      "clip_ratio/low_min": 4.299077409086749e-05,
      "clip_ratio/region_mean": 0.0017358822005917318,
      "epoch": 0.4572761738116069,
      "grad_norm": 0.13381463289260864,
      "learning_rate": 1e-06,
      "loss": -0.0098,
      "step": 98
    },
    {
      "clip_ratio/high_max": 0.002349179470911622,
      "clip_ratio/high_mean": 0.001020134466671152,
      "clip_ratio/low_mean": 0.0005440891563921468,
      "clip_ratio/low_min": 1.3507672520063352e-05,
      "clip_ratio/region_mean": 0.0015642236467101611,
      "epoch": 0.46194225721784776,
      "grad_norm": 0.1167745515704155,
      "learning_rate": 1e-06,
      "loss": -0.0149,
      "step": 99
    },
    {
      "clip_ratio/high_max": 0.0023128092507249676,
      "clip_ratio/high_mean": 0.0009480548451392679,
      "clip_ratio/low_mean": 0.0006903991043145652,
      "clip_ratio/low_min": 7.126853324734839e-05,
      "clip_ratio/region_mean": 0.0016384539339924231,
      "epoch": 0.4666083406240887,
      "grad_norm": 0.12403348088264465,
      "learning_rate": 1e-06,
      "loss": 0.0245,
      "step": 100
    },
    {
      "clip_ratio/high_max": 0.002502296381862834,
      "clip_ratio/high_mean": 0.0010109271388500929,
      "clip_ratio/low_mean": 0.0005228388636169257,
      "clip_ratio/low_min": 1.6200103345909156e-05,
      "clip_ratio/region_mean": 0.0015337660006480291,
      "epoch": 0.47127442403032954,
      "grad_norm": 0.13828738033771515,
      "learning_rate": 1e-06,
      "loss": -0.0542,
      "step": 101
    },
    {
      "clip_ratio/high_max": 0.0021421416313387454,
      "clip_ratio/high_mean": 0.0009780003547348315,
      "clip_ratio/low_mean": 0.0007443298763973871,
      "clip_ratio/low_min": 5.147692445461871e-05,
      "clip_ratio/region_mean": 0.0017223302493221126,
      "epoch": 0.4759405074365704,
      "grad_norm": 0.1175597757101059,
      "learning_rate": 1e-06,
      "loss": -0.0005,
      "step": 102
    },
    {
      "clip_ratio/high_max": 0.0020065509306732565,
      "clip_ratio/high_mean": 0.0009202863984683063,
      "clip_ratio/low_mean": 0.0006362996318785008,
      "clip_ratio/low_min": 3.835896950477036e-05,
      "clip_ratio/region_mean": 0.0015565860740025528,
      "epoch": 0.4806065908428113,
      "grad_norm": 0.126253142952919,
      "learning_rate": 1e-06,
      "loss": -0.0014,
      "step": 103
    },
    {
      "clip_ratio/high_max": 0.0021492191226570867,
      "clip_ratio/high_mean": 0.000973820177023299,
      "clip_ratio/low_mean": 0.0007368668557319324,
      "clip_ratio/low_min": 4.009028634754941e-05,
      "clip_ratio/region_mean": 0.001710687043669168,
      "epoch": 0.4852726742490522,
      "grad_norm": 0.12503156065940857,
      "learning_rate": 1e-06,
      "loss": 0.0279,
      "step": 104
    },
    {
      "clip_ratio/high_max": 0.0021064714310341515,
      "clip_ratio/high_mean": 0.001027931986754993,
      "clip_ratio/low_mean": 0.0008327936739078723,
      "clip_ratio/low_min": 5.584114660450723e-05,
      "clip_ratio/region_mean": 0.001860725664300844,
      "epoch": 0.4899387576552931,
      "grad_norm": 0.12067950516939163,
      "learning_rate": 1e-06,
      "loss": -0.0443,
      "step": 105
    },
    {
      "clip_ratio/high_max": 0.0021253314189380035,
      "clip_ratio/high_mean": 0.0008722530110389926,
      "clip_ratio/low_mean": 0.0008943315260694362,
      "clip_ratio/low_min": 0.00010063044737762539,
      "clip_ratio/region_mean": 0.001766584493452683,
      "epoch": 0.49460484106153396,
      "grad_norm": 0.1445721536874771,
      "learning_rate": 1e-06,
      "loss": 0.0318,
      "step": 106
    },
    {
      "clip_ratio/high_max": 0.002380074074608274,
      "clip_ratio/high_mean": 0.0010265141900163144,
      "clip_ratio/low_mean": 0.0009450614634261001,
      "clip_ratio/low_min": 0.00018350116079091094,
      "clip_ratio/region_mean": 0.001971575664356351,
      "epoch": 0.4992709244677749,
      "grad_norm": 0.126907080411911,
      "learning_rate": 1e-06,
      "loss": 0.0329,
      "step": 107
    },
    {
      "clip_ratio/high_max": 0.0025467996965744533,
      "clip_ratio/high_mean": 0.0010927866569545586,
      "clip_ratio/low_mean": 0.0008930186013458297,
      "clip_ratio/low_min": 6.541984748764662e-05,
      "clip_ratio/region_mean": 0.001985805298318155,
      "epoch": 0.5039370078740157,
      "grad_norm": 0.11740852892398834,
      "learning_rate": 1e-06,
      "loss": 0.0124,
      "step": 108
    },
    {
      "clip_ratio/high_max": 0.002364410935115302,
      "clip_ratio/high_mean": 0.0009945115180016728,
      "clip_ratio/low_mean": 0.0010006824450101703,
      "clip_ratio/low_min": 0.00011146509223181056,
      "clip_ratio/region_mean": 0.001995193975744769,
      "epoch": 0.5086030912802566,
      "grad_norm": 0.11474307626485825,
      "learning_rate": 1e-06,
      "loss": 0.027,
      "step": 109
    },
    {
      "clip_ratio/high_max": 0.0021619849212584086,
      "clip_ratio/high_mean": 0.0009646038633945864,
      "clip_ratio/low_mean": 0.00101947577422834,
      "clip_ratio/low_min": 0.00013564843538915738,
      "clip_ratio/region_mean": 0.001984079666726757,
      "epoch": 0.5132691746864976,
      "grad_norm": 0.12327814102172852,
      "learning_rate": 1e-06,
      "loss": 0.0551,
      "step": 110
    },
    {
      "clip_ratio/high_max": 0.002472227926773485,
      "clip_ratio/high_mean": 0.0011939582182094455,
      "clip_ratio/low_mean": 0.0009464238028158434,
      "clip_ratio/low_min": 0.00010620513876347104,
      "clip_ratio/region_mean": 0.0021403820792329498,
      "epoch": 0.5179352580927384,
      "grad_norm": 0.1299416869878769,
      "learning_rate": 1e-06,
      "loss": -0.0144,
      "step": 111
    },
    {
      "clip_ratio/high_max": 0.002420767195872031,
      "clip_ratio/high_mean": 0.0010285847820341587,
      "clip_ratio/low_mean": 0.000996541153654107,
      "clip_ratio/low_min": 0.00016540102842554916,
      "clip_ratio/region_mean": 0.002025125941145234,
      "epoch": 0.5226013414989793,
      "grad_norm": 0.1284252405166626,
      "learning_rate": 1e-06,
      "loss": 0.0276,
      "step": 112
    },
    {
      "clip_ratio/high_max": 0.0030303806852316484,
      "clip_ratio/high_mean": 0.0013427077501546592,
      "clip_ratio/low_mean": 0.001215070753460168,
      "clip_ratio/low_min": 5.0016868954116944e-05,
      "clip_ratio/region_mean": 0.002557778549089562,
      "epoch": 0.5272674249052202,
      "grad_norm": 0.11351756751537323,
      "learning_rate": 1e-06,
      "loss": 0.0357,
      "step": 113
    },
    {
      "clip_ratio/high_max": 0.004002993715403136,
      "clip_ratio/high_mean": 0.0015609030233463272,
      "clip_ratio/low_mean": 0.001315450201218482,
      "clip_ratio/low_min": 0.00018036401706922334,
      "clip_ratio/region_mean": 0.0028763532391167246,
      "epoch": 0.531933508311461,
      "grad_norm": 0.12259995937347412,
      "learning_rate": 1e-06,
      "loss": -0.0104,
      "step": 114
    },
    {
      "clip_ratio/high_max": 0.00345708731765626,
      "clip_ratio/high_mean": 0.0013046759195276536,
      "clip_ratio/low_mean": 0.0011637570751190651,
      "clip_ratio/low_min": 0.00011631832694547484,
      "clip_ratio/region_mean": 0.0024684329182491638,
      "epoch": 0.536599591717702,
      "grad_norm": 0.11007120460271835,
      "learning_rate": 1e-06,
      "loss": -0.0153,
      "step": 115
    },
    {
      "clip_ratio/high_max": 0.002803013594530057,
      "clip_ratio/high_mean": 0.0012423470107023604,
      "clip_ratio/low_mean": 0.0014616703265346587,
      "clip_ratio/low_min": 0.00013409452094492735,
      "clip_ratio/region_mean": 0.0027040172935812734,
      "epoch": 0.5412656751239429,
      "grad_norm": 0.11620628088712692,
      "learning_rate": 1e-06,
      "loss": 0.024,
      "step": 116
    },
    {
      "clip_ratio/high_max": 0.003403938375413418,
      "clip_ratio/high_mean": 0.0014087941308389418,
      "clip_ratio/low_mean": 0.0012075921622454189,
      "clip_ratio/low_min": 7.167585863498971e-05,
      "clip_ratio/region_mean": 0.0026163862421526574,
      "epoch": 0.5459317585301837,
      "grad_norm": 0.1281481236219406,
      "learning_rate": 1e-06,
      "loss": -0.0548,
      "step": 117
    },
    {
      "clip_ratio/high_max": 0.0035042481977143325,
      "clip_ratio/high_mean": 0.0014438756406889297,
      "clip_ratio/low_mean": 0.0013728199410252273,
      "clip_ratio/low_min": 0.00022288213858701056,
      "clip_ratio/region_mean": 0.0028166955744381994,
      "epoch": 0.5505978419364246,
      "grad_norm": 0.11159789562225342,
      "learning_rate": 1e-06,
      "loss": -0.001,
      "step": 118
    },
    {
      "clip_ratio/high_max": 0.002851492405170575,
      "clip_ratio/high_mean": 0.001359469439194072,
      "clip_ratio/low_mean": 0.0013244528090581298,
      "clip_ratio/low_min": 7.965722761582583e-05,
      "clip_ratio/region_mean": 0.0026839222700800747,
      "epoch": 0.5552639253426656,
      "grad_norm": 0.11402370780706406,
      "learning_rate": 1e-06,
      "loss": -0.0019,
      "step": 119
    },
    {
      "clip_ratio/high_max": 0.00288427992927609,
      "clip_ratio/high_mean": 0.0014102475142863113,
      "clip_ratio/low_mean": 0.0013363800098886713,
      "clip_ratio/low_min": 7.92927839938784e-05,
      "clip_ratio/region_mean": 0.00274662759329658,
      "epoch": 0.5599300087489064,
      "grad_norm": 0.1175575852394104,
      "learning_rate": 1e-06,
      "loss": 0.0274,
      "step": 120
    },
    {
      "clip_ratio/high_max": 0.0030200972105376422,
      "clip_ratio/high_mean": 0.0014857495116302744,
      "clip_ratio/low_mean": 0.0013425322867988143,
      "clip_ratio/low_min": 8.068737861322006e-05,
      "clip_ratio/region_mean": 0.0028282817220315337,
      "epoch": 0.5645960921551473,
      "grad_norm": 0.11188304424285889,
      "learning_rate": 1e-06,
      "loss": -0.0448,
      "step": 121
    },
    {
      "clip_ratio/high_max": 0.0032648017731844448,
      "clip_ratio/high_mean": 0.001307212864048779,
      "clip_ratio/low_mean": 0.0014480057070613839,
      "clip_ratio/low_min": 0.00015441594587173313,
      "clip_ratio/region_mean": 0.00275521854928229,
      "epoch": 0.5692621755613881,
      "grad_norm": 0.12039351463317871,
      "learning_rate": 1e-06,
      "loss": 0.0311,
      "step": 122
    },
    {
      "clip_ratio/high_max": 0.003350591388880275,
      "clip_ratio/high_mean": 0.0014234078807930928,
      "clip_ratio/low_mean": 0.0014780947640247177,
      "clip_ratio/low_min": 0.00024440218112431467,
      "clip_ratio/region_mean": 0.002901502579334192,
      "epoch": 0.573928258967629,
      "grad_norm": 0.11751885712146759,
      "learning_rate": 1e-06,
      "loss": 0.0323,
      "step": 123
    },
    {
      "clip_ratio/high_max": 0.0030487079711747356,
      "clip_ratio/high_mean": 0.0013791432247671764,
      "clip_ratio/low_mean": 0.0014340886154968757,
      "clip_ratio/low_min": 0.00015369544962595683,
      "clip_ratio/region_mean": 0.002813231825712137,
      "epoch": 0.57859434237387,
      "grad_norm": 0.1065361350774765,
      "learning_rate": 1e-06,
      "loss": 0.0118,
      "step": 124
    },
    {
      "clip_ratio/high_max": 0.0028889574023196474,
      "clip_ratio/high_mean": 0.0013628405977215152,
      "clip_ratio/low_mean": 0.0015056017655297183,
      "clip_ratio/low_min": 0.0002008831379498588,
      "clip_ratio/region_mean": 0.0028684423232334666,
      "epoch": 0.5832604257801108,
      "grad_norm": 0.10466253012418747,
      "learning_rate": 1e-06,
      "loss": 0.0266,
      "step": 125
    },
    {
      "clip_ratio/high_max": 0.002997315699758474,
      "clip_ratio/high_mean": 0.0013970356740173884,
      "clip_ratio/low_mean": 0.0016251877023023553,
      "clip_ratio/low_min": 0.00021143568756087916,
      "clip_ratio/region_mean": 0.0030222233763197437,
      "epoch": 0.5879265091863517,
      "grad_norm": 0.1156531274318695,
      "learning_rate": 1e-06,
      "loss": 0.0546,
      "step": 126
    },
    {
      "clip_ratio/high_max": 0.003504602354951203,
      "clip_ratio/high_mean": 0.0015556918988295365,
      "clip_ratio/low_mean": 0.0014803831145400181,
      "clip_ratio/low_min": 0.0001914514014060842,
      "clip_ratio/region_mean": 0.0030360749224200845,
      "epoch": 0.5925925925925926,
      "grad_norm": 0.12074324488639832,
      "learning_rate": 1e-06,
      "loss": -0.015,
      "step": 127
    },
    {
      "clip_ratio/high_max": 0.003545554558513686,
      "clip_ratio/high_mean": 0.0014461224054684862,
      "clip_ratio/low_mean": 0.0015403323814098258,
      "clip_ratio/low_min": 0.0003250622494306299,
      "clip_ratio/region_mean": 0.0029864547614124604,
      "epoch": 0.5972586759988335,
      "grad_norm": 0.11859126389026642,
      "learning_rate": 1e-06,
      "loss": 0.027,
      "step": 128
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01708984375,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4052.0,
      "completions/mean_length": 625.4926147460938,
      "completions/mean_terminated_length": 565.1509399414062,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.6019247594050744,
      "grad_norm": 0.12504246830940247,
      "learning_rate": 1e-06,
      "loss": -0.0211,
      "num_tokens": 46360250.0,
      "reward": 0.5486189126968384,
      "reward_std": 0.22039179503917694,
      "rewards/simpleverify_reward/mean": 0.5486188530921936,
      "rewards/simpleverify_reward/std": 0.4976479411125183,
      "step": 129
    },
    {
      "clip_ratio/high_max": 0.0019622973923105747,
      "clip_ratio/high_mean": 0.00086254951202136,
      "clip_ratio/low_mean": 0.000690543945893296,
      "clip_ratio/low_min": 4.503778018261073e-05,
      "clip_ratio/region_mean": 0.0015530934651906136,
      "epoch": 0.6065908428113153,
      "grad_norm": 0.12330756336450577,
      "learning_rate": 1e-06,
      "loss": -0.0256,
      "step": 130
    },
    {
      "clip_ratio/high_max": 0.002415656505036168,
      "clip_ratio/high_mean": 0.0010426417047710856,
      "clip_ratio/low_mean": 0.0007739345546724508,
      "clip_ratio/low_min": 0.0001219256500917254,
      "clip_ratio/region_mean": 0.0018165762594435364,
      "epoch": 0.6112569262175561,
      "grad_norm": 0.14477677643299103,
      "learning_rate": 1e-06,
      "loss": 0.0041,
      "step": 131
    },
    {
      "clip_ratio/high_max": 0.002318665814527776,
      "clip_ratio/high_mean": 0.0010169484485231806,
      "clip_ratio/low_mean": 0.0007320470376726007,
      "clip_ratio/low_min": 4.6667485548823606e-05,
      "clip_ratio/region_mean": 0.0017489955134806223,
      "epoch": 0.615923009623797,
      "grad_norm": 0.12318412214517593,
      "learning_rate": 1e-06,
      "loss": -0.035,
      "step": 132
    },
    {
      "clip_ratio/high_max": 0.002068623303784989,
      "clip_ratio/high_mean": 0.0009670791041571647,
      "clip_ratio/low_mean": 0.0007370205366896698,
      "clip_ratio/low_min": 8.613667068857467e-05,
      "clip_ratio/region_mean": 0.0017040996535797603,
      "epoch": 0.620589093030038,
      "grad_norm": 0.12779073417186737,
      "learning_rate": 1e-06,
      "loss": -0.0049,
      "step": 133
    },
    {
      "clip_ratio/high_max": 0.00234722915047314,
      "clip_ratio/high_mean": 0.000929608129808912,
      "clip_ratio/low_mean": 0.0007829385194781935,
      "clip_ratio/low_min": 8.551259543310152e-05,
      "clip_ratio/region_mean": 0.0017125465965364128,
      "epoch": 0.6252551764362788,
      "grad_norm": 0.1303846538066864,
      "learning_rate": 1e-06,
      "loss": 0.0277,
      "step": 134
    },
    {
      "clip_ratio/high_max": 0.002241820649942383,
      "clip_ratio/high_mean": 0.001005620923024253,
      "clip_ratio/low_mean": 0.0008878043463482754,
      "clip_ratio/low_min": 0.00010648371426213998,
      "clip_ratio/region_mean": 0.0018934252475446556,
      "epoch": 0.6299212598425197,
      "grad_norm": 0.12360312789678574,
      "learning_rate": 1e-06,
      "loss": 0.0062,
      "step": 135
    },
    {
      "clip_ratio/high_max": 0.0022526702232426032,
      "clip_ratio/high_mean": 0.0009839833837759215,
      "clip_ratio/low_mean": 0.0009566756398271536,
      "clip_ratio/low_min": 9.257162764697568e-05,
      "clip_ratio/region_mean": 0.001940659036336001,
      "epoch": 0.6345873432487605,
      "grad_norm": 0.13168051838874817,
      "learning_rate": 1e-06,
      "loss": 0.0231,
      "step": 136
    },
    {
      "clip_ratio/high_max": 0.0017736651498125866,
      "clip_ratio/high_mean": 0.0008472457593597937,
      "clip_ratio/low_mean": 0.0008509719627909362,
      "clip_ratio/low_min": 8.851047914504306e-05,
      "clip_ratio/region_mean": 0.0016982177185127512,
      "epoch": 0.6392534266550015,
      "grad_norm": 0.11745944619178772,
      "learning_rate": 1e-06,
      "loss": 0.0464,
      "step": 137
    },
    {
      "clip_ratio/high_max": 0.002357255965762306,
      "clip_ratio/high_mean": 0.0010218969673587708,
      "clip_ratio/low_mean": 0.0010448472276038956,
      "clip_ratio/low_min": 3.768465467146598e-05,
      "clip_ratio/region_mean": 0.0020667441858677194,
      "epoch": 0.6439195100612424,
      "grad_norm": 0.13640879094600677,
      "learning_rate": 1e-06,
      "loss": -0.0401,
      "step": 138
    },
    {
      "clip_ratio/high_max": 0.002176973233872559,
      "clip_ratio/high_mean": 0.0010496407194295898,
      "clip_ratio/low_mean": 0.0011276936093054246,
      "clip_ratio/low_min": 0.00010892456339206547,
      "clip_ratio/region_mean": 0.0021773343105451204,
      "epoch": 0.6485855934674832,
      "grad_norm": 0.12636250257492065,
      "learning_rate": 1e-06,
      "loss": 0.0194,
      "step": 139
    },
    {
      "clip_ratio/high_max": 0.0022459454557974823,
      "clip_ratio/high_mean": 0.0009187738578475546,
      "clip_ratio/low_mean": 0.0011709720110957278,
      "clip_ratio/low_min": 8.018929383979412e-05,
      "clip_ratio/region_mean": 0.0020897458816762082,
      "epoch": 0.6532516768737241,
      "grad_norm": 0.12697508931159973,
      "learning_rate": 1e-06,
      "loss": 0.061,
      "step": 140
    },
    {
      "clip_ratio/high_max": 0.0027129300142405555,
      "clip_ratio/high_mean": 0.0009579164870956447,
      "clip_ratio/low_mean": 0.0009603455837350339,
      "clip_ratio/low_min": 7.704151266807457e-05,
      "clip_ratio/region_mean": 0.0019182620744686574,
      "epoch": 0.657917760279965,
      "grad_norm": 0.12945355474948883,
      "learning_rate": 1e-06,
      "loss": 0.0254,
      "step": 141
    },
    {
      "clip_ratio/high_max": 0.002652611816301942,
      "clip_ratio/high_mean": 0.0011574287018447649,
      "clip_ratio/low_mean": 0.0009167208190774545,
      "clip_ratio/low_min": 0.0001130971431848593,
      "clip_ratio/region_mean": 0.0020741495391121134,
      "epoch": 0.6625838436862059,
      "grad_norm": 0.12411902099847794,
      "learning_rate": 1e-06,
      "loss": -0.0124,
      "step": 142
    },
    {
      "clip_ratio/high_max": 0.00243718452838948,
      "clip_ratio/high_mean": 0.001045304125000257,
      "clip_ratio/low_mean": 0.0010213898131041788,
      "clip_ratio/low_min": 0.00013535294237954076,
      "clip_ratio/region_mean": 0.0020666939672082663,
      "epoch": 0.6672499270924468,
      "grad_norm": 0.12218952178955078,
      "learning_rate": 1e-06,
      "loss": -0.0044,
      "step": 143
    },
    {
      "clip_ratio/high_max": 0.0023179636045824736,
      "clip_ratio/high_mean": 0.0009743330138007877,
      "clip_ratio/low_mean": 0.0011671989086607937,
      "clip_ratio/low_min": 0.00010179436776525108,
      "clip_ratio/region_mean": 0.0021415319279185496,
      "epoch": 0.6719160104986877,
      "grad_norm": 0.12230091542005539,
      "learning_rate": 1e-06,
      "loss": 0.0498,
      "step": 144
    },
    {
      "clip_ratio/high_max": 0.0030589635425712913,
      "clip_ratio/high_mean": 0.0014219260010577273,
      "clip_ratio/low_mean": 0.001473729033023119,
      "clip_ratio/low_min": 0.00016825882630655542,
      "clip_ratio/region_mean": 0.002895655117754359,
      "epoch": 0.6765820939049285,
      "grad_norm": 0.11068176478147507,
      "learning_rate": 1e-06,
      "loss": -0.0216,
      "step": 145
    },
    {
      "clip_ratio/high_max": 0.0030975044719525613,
      "clip_ratio/high_mean": 0.0012474586364987772,
      "clip_ratio/low_mean": 0.0014553499822795857,
      "clip_ratio/low_min": 0.0002019758057940635,
      "clip_ratio/region_mean": 0.0027028086478821933,
      "epoch": 0.6812481773111695,
      "grad_norm": 0.11317021399736404,
      "learning_rate": 1e-06,
      "loss": -0.0261,
      "step": 146
    },
    {
      "clip_ratio/high_max": 0.0031326866810559295,
      "clip_ratio/high_mean": 0.0014611303740821313,
      "clip_ratio/low_mean": 0.0016066228454292286,
      "clip_ratio/low_min": 0.0002908540764110512,
      "clip_ratio/region_mean": 0.0030677532049594447,
      "epoch": 0.6859142607174104,
      "grad_norm": 0.12762829661369324,
      "learning_rate": 1e-06,
      "loss": 0.0034,
      "step": 147
    },
    {
      "clip_ratio/high_max": 0.003193265845766291,
      "clip_ratio/high_mean": 0.0013287888978084084,
      "clip_ratio/low_mean": 0.0015411545864481013,
      "clip_ratio/low_min": 0.0001754615113895852,
      "clip_ratio/region_mean": 0.002869943462428637,
      "epoch": 0.6905803441236512,
      "grad_norm": 0.11296442151069641,
      "learning_rate": 1e-06,
      "loss": -0.0356,
      "step": 148
    },
    {
      "clip_ratio/high_max": 0.003124746464891359,
      "clip_ratio/high_mean": 0.001424162890543812,
      "clip_ratio/low_mean": 0.0016049580262915697,
      "clip_ratio/low_min": 0.00022178129347594222,
      "clip_ratio/region_mean": 0.0030291209332062863,
      "epoch": 0.6952464275298921,
      "grad_norm": 0.11597266048192978,
      "learning_rate": 1e-06,
      "loss": -0.0055,
      "step": 149
    },
    {
      "clip_ratio/high_max": 0.003220578819309594,
      "clip_ratio/high_mean": 0.0013671701453858986,
      "clip_ratio/low_mean": 0.0016575887439103099,
      "clip_ratio/low_min": 0.0002167654711229261,
      "clip_ratio/region_mean": 0.0030247588583733886,
      "epoch": 0.6999125109361329,
      "grad_norm": 0.12387190014123917,
      "learning_rate": 1e-06,
      "loss": 0.0271,
      "step": 150
    },
    {
      "clip_ratio/high_max": 0.0032617842007311992,
      "clip_ratio/high_mean": 0.0013756312982877716,
      "clip_ratio/low_mean": 0.0018368964811088517,
      "clip_ratio/low_min": 0.0001695835635473486,
      "clip_ratio/region_mean": 0.0032125277793966234,
      "epoch": 0.7045785943423739,
      "grad_norm": 0.11603932082653046,
      "learning_rate": 1e-06,
      "loss": 0.0056,
      "step": 151
    },
    {
      "clip_ratio/high_max": 0.0033011011182679795,
      "clip_ratio/high_mean": 0.0014124352564977016,
      "clip_ratio/low_mean": 0.0017514135652163532,
      "clip_ratio/low_min": 0.0002159983496312634,
      "clip_ratio/region_mean": 0.0031638488871976733,
      "epoch": 0.7092446777486148,
      "grad_norm": 0.1222923994064331,
      "learning_rate": 1e-06,
      "loss": 0.0225,
      "step": 152
    },
    {
      "clip_ratio/high_max": 0.0023524437201558612,
      "clip_ratio/high_mean": 0.001154064571892377,
      "clip_ratio/low_mean": 0.0015891659058979712,
      "clip_ratio/low_min": 0.0003527866556396475,
      "clip_ratio/region_mean": 0.0027432304341346025,
      "epoch": 0.7139107611548556,
      "grad_norm": 0.10908897966146469,
      "learning_rate": 1e-06,
      "loss": 0.0459,
      "step": 153
    },
    {
      "clip_ratio/high_max": 0.003170895215589553,
      "clip_ratio/high_mean": 0.001523188933788333,
      "clip_ratio/low_mean": 0.001526602020021528,
      "clip_ratio/low_min": 0.00013473693070409354,
      "clip_ratio/region_mean": 0.0030497910047415644,
      "epoch": 0.7185768445610965,
      "grad_norm": 0.12282628566026688,
      "learning_rate": 1e-06,
      "loss": -0.0408,
      "step": 154
    },
    {
      "clip_ratio/high_max": 0.0032785448056529276,
      "clip_ratio/high_mean": 0.0015166443372436333,
      "clip_ratio/low_mean": 0.0017930999092641287,
      "clip_ratio/low_min": 0.0002642217732500285,
      "clip_ratio/region_mean": 0.003309744242869783,
      "epoch": 0.7232429279673375,
      "grad_norm": 0.11680693924427032,
      "learning_rate": 1e-06,
      "loss": 0.0189,
      "step": 155
    },
    {
      "clip_ratio/high_max": 0.0029522183976951055,
      "clip_ratio/high_mean": 0.0012086244933016133,
      "clip_ratio/low_mean": 0.0019024655121029355,
      "clip_ratio/low_min": 0.00035396035673329607,
      "clip_ratio/region_mean": 0.0031110900454223156,
      "epoch": 0.7279090113735783,
      "grad_norm": 0.1127549484372139,
      "learning_rate": 1e-06,
      "loss": 0.0605,
      "step": 156
    },
    {
      "clip_ratio/high_max": 0.003389412297110539,
      "clip_ratio/high_mean": 0.0013614786530524725,
      "clip_ratio/low_mean": 0.001620078273845138,
      "clip_ratio/low_min": 9.173921262117801e-05,
      "clip_ratio/region_mean": 0.0029815569578204304,
      "epoch": 0.7325750947798192,
      "grad_norm": 0.11988050490617752,
      "learning_rate": 1e-06,
      "loss": 0.0248,
      "step": 157
    },
    {
      "clip_ratio/high_max": 0.0037293742134352215,
      "clip_ratio/high_mean": 0.0016134379984578118,
      "clip_ratio/low_mean": 0.0014948346579330973,
      "clip_ratio/low_min": 0.000297502941066341,
      "clip_ratio/region_mean": 0.0031082726345630363,
      "epoch": 0.73724117818606,
      "grad_norm": 0.11628247797489166,
      "learning_rate": 1e-06,
      "loss": -0.013,
      "step": 158
    },
    {
      "clip_ratio/high_max": 0.0031293387728510424,
      "clip_ratio/high_mean": 0.0014757179706066381,
      "clip_ratio/low_mean": 0.0015834566875128075,
      "clip_ratio/low_min": 0.00027981661150988657,
      "clip_ratio/region_mean": 0.003059174632653594,
      "epoch": 0.7419072615923009,
      "grad_norm": 0.11325806379318237,
      "learning_rate": 1e-06,
      "loss": -0.0049,
      "step": 159
    },
    {
      "clip_ratio/high_max": 0.0031949883195920847,
      "clip_ratio/high_mean": 0.0013877581077395007,
      "clip_ratio/low_mean": 0.0016084506023616996,
      "clip_ratio/low_min": 0.00015860439998505171,
      "clip_ratio/region_mean": 0.002996208677359391,
      "epoch": 0.7465733449985419,
      "grad_norm": 0.1116633489727974,
      "learning_rate": 1e-06,
      "loss": 0.0492,
      "step": 160
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.016392299107142905,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4057.0,
      "completions/mean_length": 618.296142578125,
      "completions/mean_terminated_length": 560.3385009765625,
      "completions/min_length": 20.0,
      "completions/min_terminated_length": 20.0,
      "epoch": 0.7512394284047827,
      "grad_norm": 0.1300407499074936,
      "learning_rate": 1e-06,
      "loss": -0.0128,
      "num_tokens": 55679407.0,
      "reward": 0.5765904188156128,
      "reward_std": 0.212571918964386,
      "rewards/simpleverify_reward/mean": 0.5765904188156128,
      "rewards/simpleverify_reward/std": 0.49411630630493164,
      "step": 161
    },
    {
      "clip_ratio/high_max": 0.0022728608528268524,
      "clip_ratio/high_mean": 0.0008833910596877104,
      "clip_ratio/low_mean": 0.0006150963381514885,
      "clip_ratio/low_min": 3.628986542025814e-05,
      "clip_ratio/region_mean": 0.0014984874069341458,
      "epoch": 0.7559055118110236,
      "grad_norm": 0.1208866685628891,
      "learning_rate": 1e-06,
      "loss": 0.0051,
      "step": 162
    },
    {
      "clip_ratio/high_max": 0.0024660796552780084,
      "clip_ratio/high_mean": 0.000990775044556358,
      "clip_ratio/low_mean": 0.000572200100577902,
      "clip_ratio/low_min": 2.037744525296148e-05,
      "clip_ratio/region_mean": 0.0015629751433152705,
      "epoch": 0.7605715952172645,
      "grad_norm": 0.1179087832570076,
      "learning_rate": 1e-06,
      "loss": 0.004,
      "step": 163
    },
    {
      "clip_ratio/high_max": 0.002314921235665679,
      "clip_ratio/high_mean": 0.0009412051113031339,
      "clip_ratio/low_mean": 0.000560982139177213,
      "clip_ratio/low_min": 3.2642863516230136e-05,
      "clip_ratio/region_mean": 0.0015021872750367038,
      "epoch": 0.7652376786235054,
      "grad_norm": 0.11520962417125702,
      "learning_rate": 1e-06,
      "loss": 0.0168,
      "step": 164
    },
    {
      "clip_ratio/high_max": 0.0022047006768843858,
      "clip_ratio/high_mean": 0.0009568603263687692,
      "clip_ratio/low_mean": 0.0006946019220777089,
      "clip_ratio/low_min": 3.6372794056660496e-05,
      "clip_ratio/region_mean": 0.001651462236623047,
      "epoch": 0.7699037620297463,
      "grad_norm": 0.12777192890644073,
      "learning_rate": 1e-06,
      "loss": 0.0274,
      "step": 165
    },
    {
      "clip_ratio/high_max": 0.002207837824244052,
      "clip_ratio/high_mean": 0.0009902433957904577,
      "clip_ratio/low_mean": 0.0007543263109255349,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017445697449147701,
      "epoch": 0.7745698454359872,
      "grad_norm": 0.12327396124601364,
      "learning_rate": 1e-06,
      "loss": 0.0022,
      "step": 166
    },
    {
      "clip_ratio/high_max": 0.002686720901692752,
      "clip_ratio/high_mean": 0.001066892669769004,
      "clip_ratio/low_mean": 0.0007163599202613113,
      "clip_ratio/low_min": 4.928340968035627e-05,
      "clip_ratio/region_mean": 0.001783252570021432,
      "epoch": 0.779235928842228,
      "grad_norm": 0.1391931027173996,
      "learning_rate": 1e-06,
      "loss": -0.0384,
      "step": 167
    },
    {
      "clip_ratio/high_max": 0.002355117438128218,
      "clip_ratio/high_mean": 0.0010019054825534113,
      "clip_ratio/low_mean": 0.0008151353649736848,
      "clip_ratio/low_min": 5.1902492486988194e-05,
      "clip_ratio/region_mean": 0.0018170408802689053,
      "epoch": 0.7839020122484689,
      "grad_norm": 0.1298067271709442,
      "learning_rate": 1e-06,
      "loss": 0.0076,
      "step": 168
    },
    {
      "clip_ratio/high_max": 0.0023464256228180602,
      "clip_ratio/high_mean": 0.0009794815869099693,
      "clip_ratio/low_mean": 0.0010000048805522965,
      "clip_ratio/low_min": 0.00012863148094766075,
      "clip_ratio/region_mean": 0.001979486441996414,
      "epoch": 0.7885680956547099,
      "grad_norm": 0.12335332483053207,
      "learning_rate": 1e-06,
      "loss": 0.0305,
      "step": 169
    },
    {
      "clip_ratio/high_max": 0.002366124994296115,
      "clip_ratio/high_mean": 0.001040869035932701,
      "clip_ratio/low_mean": 0.0009606007988622878,
      "clip_ratio/low_min": 7.547083623649087e-05,
      "clip_ratio/region_mean": 0.00200146983115701,
      "epoch": 0.7932341790609507,
      "grad_norm": 0.13561807572841644,
      "learning_rate": 1e-06,
      "loss": -0.0016,
      "step": 170
    },
    {
      "clip_ratio/high_max": 0.0028863431507488713,
      "clip_ratio/high_mean": 0.0011377375631127506,
      "clip_ratio/low_mean": 0.0008200303054763936,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001957767817657441,
      "epoch": 0.7979002624671916,
      "grad_norm": 0.131186380982399,
      "learning_rate": 1e-06,
      "loss": -0.0254,
      "step": 171
    },
    {
      "clip_ratio/high_max": 0.0023395269599859603,
      "clip_ratio/high_mean": 0.0010299486420990434,
      "clip_ratio/low_mean": 0.0010231237029074691,
      "clip_ratio/low_min": 0.000147094736348663,
      "clip_ratio/region_mean": 0.0020530723777483217,
      "epoch": 0.8025663458734325,
      "grad_norm": 0.1217319518327713,
      "learning_rate": 1e-06,
      "loss": 0.0084,
      "step": 172
    },
    {
      "clip_ratio/high_max": 0.0019382450191187672,
      "clip_ratio/high_mean": 0.0008459129167022184,
      "clip_ratio/low_mean": 0.0009874254337773891,
      "clip_ratio/low_min": 0.00013877356923330808,
      "clip_ratio/region_mean": 0.0018333383268327452,
      "epoch": 0.8072324292796734,
      "grad_norm": 0.11738027632236481,
      "learning_rate": 1e-06,
      "loss": 0.0268,
      "step": 173
    },
    {
      "clip_ratio/high_max": 0.0021850022894795984,
      "clip_ratio/high_mean": 0.0011033032988052582,
      "clip_ratio/low_mean": 0.001106167743273545,
      "clip_ratio/low_min": 0.00013648305775859626,
      "clip_ratio/region_mean": 0.002209471058449708,
      "epoch": 0.8118985126859143,
      "grad_norm": 0.12672440707683563,
      "learning_rate": 1e-06,
      "loss": -0.0125,
      "step": 174
    },
    {
      "clip_ratio/high_max": 0.0024052742883213796,
      "clip_ratio/high_mean": 0.0010238092763756867,
      "clip_ratio/low_mean": 0.0010055394959636033,
      "clip_ratio/low_min": 7.179893691500183e-05,
      "clip_ratio/region_mean": 0.002029348775977269,
      "epoch": 0.8165645960921551,
      "grad_norm": 0.12982292473316193,
      "learning_rate": 1e-06,
      "loss": 0.0364,
      "step": 175
    },
    {
      "clip_ratio/high_max": 0.002506316697690636,
      "clip_ratio/high_mean": 0.0010294234416505788,
      "clip_ratio/low_mean": 0.0009139028879872058,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019433263296377845,
      "epoch": 0.821230679498396,
      "grad_norm": 0.12429260462522507,
      "learning_rate": 1e-06,
      "loss": 0.0124,
      "step": 176
    },
    {
      "clip_ratio/high_max": 0.003091856327955611,
      "clip_ratio/high_mean": 0.001399352629960049,
      "clip_ratio/low_mean": 0.0013897682147216983,
      "clip_ratio/low_min": 0.0001232874856214039,
      "clip_ratio/region_mean": 0.002789120793750044,
      "epoch": 0.8258967629046369,
      "grad_norm": 0.11247675120830536,
      "learning_rate": 1e-06,
      "loss": -0.0133,
      "step": 177
    },
    {
      "clip_ratio/high_max": 0.0028565432658069767,
      "clip_ratio/high_mean": 0.0012777550236933166,
      "clip_ratio/low_mean": 0.001388784894515993,
      "clip_ratio/low_min": 0.0001572829769429518,
      "clip_ratio/region_mean": 0.002666539869096596,
      "epoch": 0.8305628463108778,
      "grad_norm": 0.10792485624551773,
      "learning_rate": 1e-06,
      "loss": 0.0045,
      "step": 178
    },
    {
      "clip_ratio/high_max": 0.003086590048042126,
      "clip_ratio/high_mean": 0.001362356255413033,
      "clip_ratio/low_mean": 0.0012803557874576654,
      "clip_ratio/low_min": 4.900712337985169e-05,
      "clip_ratio/region_mean": 0.0026427120246808045,
      "epoch": 0.8352289297171187,
      "grad_norm": 0.10708758980035782,
      "learning_rate": 1e-06,
      "loss": 0.0035,
      "step": 179
    },
    {
      "clip_ratio/high_max": 0.0027207307794014923,
      "clip_ratio/high_mean": 0.0012383792745822575,
      "clip_ratio/low_mean": 0.0011612797788984608,
      "clip_ratio/low_min": 6.0781432694057e-05,
      "clip_ratio/region_mean": 0.002399659075308591,
      "epoch": 0.8398950131233596,
      "grad_norm": 0.1075570285320282,
      "learning_rate": 1e-06,
      "loss": 0.0163,
      "step": 180
    },
    {
      "clip_ratio/high_max": 0.0028585142354131676,
      "clip_ratio/high_mean": 0.0013736776054429356,
      "clip_ratio/low_mean": 0.001392723228491377,
      "clip_ratio/low_min": 0.00010873119026655331,
      "clip_ratio/region_mean": 0.002766400830296334,
      "epoch": 0.8445610965296004,
      "grad_norm": 0.11319371312856674,
      "learning_rate": 1e-06,
      "loss": 0.0268,
      "step": 181
    },
    {
      "clip_ratio/high_max": 0.00307430540851783,
      "clip_ratio/high_mean": 0.0013480396373779513,
      "clip_ratio/low_mean": 0.0013493507303792285,
      "clip_ratio/low_min": 0.00010216513510385994,
      "clip_ratio/region_mean": 0.002697390373214148,
      "epoch": 0.8492271799358414,
      "grad_norm": 0.11054737865924835,
      "learning_rate": 1e-06,
      "loss": 0.0016,
      "step": 182
    },
    {
      "clip_ratio/high_max": 0.003473888740700204,
      "clip_ratio/high_mean": 0.0015409789848490618,
      "clip_ratio/low_mean": 0.0013487565811374225,
      "clip_ratio/low_min": 0.00019729429550352506,
      "clip_ratio/region_mean": 0.0028897355587105267,
      "epoch": 0.8538932633420823,
      "grad_norm": 0.11917661875486374,
      "learning_rate": 1e-06,
      "loss": -0.0392,
      "step": 183
    },
    {
      "clip_ratio/high_max": 0.0032659345015417784,
      "clip_ratio/high_mean": 0.0013825950372847728,
      "clip_ratio/low_mean": 0.0016036020388128236,
      "clip_ratio/low_min": 0.000264601141680032,
      "clip_ratio/region_mean": 0.0029861971124773845,
      "epoch": 0.8585593467483231,
      "grad_norm": 0.11535173654556274,
      "learning_rate": 1e-06,
      "loss": 0.0069,
      "step": 184
    },
    {
      "clip_ratio/high_max": 0.003165465190249961,
      "clip_ratio/high_mean": 0.0014062658046896104,
      "clip_ratio/low_mean": 0.0016643798517179675,
      "clip_ratio/low_min": 0.0002641809987835586,
      "clip_ratio/region_mean": 0.003070645601837896,
      "epoch": 0.863225430154564,
      "grad_norm": 0.11451252549886703,
      "learning_rate": 1e-06,
      "loss": 0.0299,
      "step": 185
    },
    {
      "clip_ratio/high_max": 0.003231565046007745,
      "clip_ratio/high_mean": 0.0015055474432301708,
      "clip_ratio/low_mean": 0.0017203881870955229,
      "clip_ratio/low_min": 0.00010887846838159021,
      "clip_ratio/region_mean": 0.0032259356885333546,
      "epoch": 0.8678915135608049,
      "grad_norm": 0.12903904914855957,
      "learning_rate": 1e-06,
      "loss": -0.0023,
      "step": 186
    },
    {
      "clip_ratio/high_max": 0.0038990563771221787,
      "clip_ratio/high_mean": 0.0016840885873534717,
      "clip_ratio/low_mean": 0.00150715892232256,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0031912474805722013,
      "epoch": 0.8725575969670458,
      "grad_norm": 0.12221875786781311,
      "learning_rate": 1e-06,
      "loss": -0.0261,
      "step": 187
    },
    {
      "clip_ratio/high_max": 0.0031884782074484974,
      "clip_ratio/high_mean": 0.0014102132518019062,
      "clip_ratio/low_mean": 0.0016071983118308708,
      "clip_ratio/low_min": 0.0002617763111629756,
      "clip_ratio/region_mean": 0.0030174115672707558,
      "epoch": 0.8772236803732867,
      "grad_norm": 0.11409705877304077,
      "learning_rate": 1e-06,
      "loss": 0.0079,
      "step": 188
    },
    {
      "clip_ratio/high_max": 0.003140487940981984,
      "clip_ratio/high_mean": 0.0013090760439808946,
      "clip_ratio/low_mean": 0.0016345995718438644,
      "clip_ratio/low_min": 0.0002664353369254968,
      "clip_ratio/region_mean": 0.0029436755867209285,
      "epoch": 0.8818897637795275,
      "grad_norm": 0.10862597823143005,
      "learning_rate": 1e-06,
      "loss": 0.0263,
      "step": 189
    },
    {
      "clip_ratio/high_max": 0.003256008174503222,
      "clip_ratio/high_mean": 0.0014727026427863166,
      "clip_ratio/low_mean": 0.0016130155454447959,
      "clip_ratio/low_min": 0.00019578502906369977,
      "clip_ratio/region_mean": 0.0030857182428007945,
      "epoch": 0.8865558471857684,
      "grad_norm": 0.11458273977041245,
      "learning_rate": 1e-06,
      "loss": -0.0131,
      "step": 190
    },
    {
      "clip_ratio/high_max": 0.0038847233226988465,
      "clip_ratio/high_mean": 0.0014843351746094413,
      "clip_ratio/low_mean": 0.0016515007009729743,
      "clip_ratio/low_min": 0.00015455462016689125,
      "clip_ratio/region_mean": 0.003135835941066034,
      "epoch": 0.8912219305920094,
      "grad_norm": 0.12143843621015549,
      "learning_rate": 1e-06,
      "loss": 0.0358,
      "step": 191
    },
    {
      "clip_ratio/high_max": 0.0037288971070665866,
      "clip_ratio/high_mean": 0.0016110568758449517,
      "clip_ratio/low_mean": 0.0014498142299999017,
      "clip_ratio/low_min": 5.2418817176658195e-05,
      "clip_ratio/region_mean": 0.0030608711240347475,
      "epoch": 0.8958880139982502,
      "grad_norm": 0.11275072395801544,
      "learning_rate": 1e-06,
      "loss": 0.0118,
      "step": 192
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.020298549107142905,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 3853.0,
      "completions/mean_length": 628.046875,
      "completions/mean_terminated_length": 556.1939086914062,
      "completions/min_length": 52.0,
      "completions/min_terminated_length": 52.0,
      "epoch": 1.0046660834062409,
      "grad_norm": 0.11700253933668137,
      "learning_rate": 1e-06,
      "loss": 0.0058,
      "num_tokens": 64943399.0,
      "reward": 0.5739397406578064,
      "reward_std": 0.20258471369743347,
      "rewards/simpleverify_reward/mean": 0.5739397406578064,
      "rewards/simpleverify_reward/std": 0.4945199489593506,
      "step": 193
    },
    {
      "clip_ratio/high_max": 0.001992187782889232,
      "clip_ratio/high_mean": 0.0008438729164481629,
      "clip_ratio/low_mean": 0.0006071827938285423,
      "clip_ratio/low_min": 1.5078408068802673e-05,
      "clip_ratio/region_mean": 0.0014510557084577158,
      "epoch": 1.0093321668124817,
      "grad_norm": 0.11676914244890213,
      "learning_rate": 1e-06,
      "loss": 0.0175,
      "step": 194
    },
    {
      "clip_ratio/high_max": 0.00222089902672451,
      "clip_ratio/high_mean": 0.0009028679196489975,
      "clip_ratio/low_mean": 0.0005745327225668007,
      "clip_ratio/low_min": 4.805685330211418e-05,
      "clip_ratio/region_mean": 0.0014774006413063034,
      "epoch": 1.0139982502187226,
      "grad_norm": 0.13004937767982483,
      "learning_rate": 1e-06,
      "loss": -0.0213,
      "step": 195
    },
    {
      "clip_ratio/high_max": 0.0019888310998794623,
      "clip_ratio/high_mean": 0.0008023303998925257,
      "clip_ratio/low_mean": 0.0005680742769982317,
      "clip_ratio/low_min": 3.028301398444455e-05,
      "clip_ratio/region_mean": 0.0013704047014471143,
      "epoch": 1.0186643336249634,
      "grad_norm": 0.12435458600521088,
      "learning_rate": 1e-06,
      "loss": 0.0099,
      "step": 196
    },
    {
      "clip_ratio/high_max": 0.002000114789552754,
      "clip_ratio/high_mean": 0.0009112575116887456,
      "clip_ratio/low_mean": 0.0005940267574260361,
      "clip_ratio/low_min": 4.376520973892184e-05,
      "clip_ratio/region_mean": 0.0015052842791192234,
      "epoch": 1.0233304170312045,
      "grad_norm": 0.13317565619945526,
      "learning_rate": 1e-06,
      "loss": -0.0231,
      "step": 197
    },
    {
      "clip_ratio/high_max": 0.0018851126296794973,
      "clip_ratio/high_mean": 0.0007911962311482057,
      "clip_ratio/low_mean": 0.0007478209172404604,
      "clip_ratio/low_min": 6.730256882292451e-05,
      "clip_ratio/region_mean": 0.001539017179311486,
      "epoch": 1.0279965004374454,
      "grad_norm": 0.12191758304834366,
      "learning_rate": 1e-06,
      "loss": 0.0544,
      "step": 198
    },
    {
      "clip_ratio/high_max": 0.002169060222513508,
      "clip_ratio/high_mean": 0.0010109215945703909,
      "clip_ratio/low_mean": 0.0007944734552438604,
      "clip_ratio/low_min": 5.708780827262672e-05,
      "clip_ratio/region_mean": 0.001805395026167389,
      "epoch": 1.0326625838436863,
      "grad_norm": 0.12367334961891174,
      "learning_rate": 1e-06,
      "loss": -0.0171,
      "step": 199
    },
    {
      "clip_ratio/high_max": 0.002637896985106636,
      "clip_ratio/high_mean": 0.001022678567096591,
      "clip_ratio/low_mean": 0.000851700087878271,
      "clip_ratio/low_min": 5.934360342507716e-05,
      "clip_ratio/region_mean": 0.0018743786931736395,
      "epoch": 1.0373286672499271,
      "grad_norm": 0.1231176033616066,
      "learning_rate": 1e-06,
      "loss": 0.0116,
      "step": 200
    },
    {
      "clip_ratio/high_max": 0.00252491715218639,
      "clip_ratio/high_mean": 0.001009621500998037,
      "clip_ratio/low_mean": 0.0008673445890963194,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018769660746329464,
      "epoch": 1.041994750656168,
      "grad_norm": 0.1276102066040039,
      "learning_rate": 1e-06,
      "loss": -0.0234,
      "step": 201
    },
    {
      "clip_ratio/high_max": 0.0021756897986051627,
      "clip_ratio/high_mean": 0.00102547588176094,
      "clip_ratio/low_mean": 0.0009182617504848167,
      "clip_ratio/low_min": 6.389906593540218e-05,
      "clip_ratio/region_mean": 0.00194373768317746,
      "epoch": 1.0466608340624088,
      "grad_norm": 0.12504766881465912,
      "learning_rate": 1e-06,
      "loss": -0.0107,
      "step": 202
    },
    {
      "clip_ratio/high_max": 0.002259956090711057,
      "clip_ratio/high_mean": 0.0010174774397455622,
      "clip_ratio/low_mean": 0.0010540294388192706,
      "clip_ratio/low_min": 6.897195817145985e-05,
      "clip_ratio/region_mean": 0.002071506853098981,
      "epoch": 1.0513269174686497,
      "grad_norm": 0.11805891990661621,
      "learning_rate": 1e-06,
      "loss": -0.0034,
      "step": 203
    },
    {
      "clip_ratio/high_max": 0.0021944052423350513,
      "clip_ratio/high_mean": 0.0009023564562085085,
      "clip_ratio/low_mean": 0.001000316085992381,
      "clip_ratio/low_min": 0.00016303732536471216,
      "clip_ratio/region_mean": 0.00190267257130472,
      "epoch": 1.0559930008748906,
      "grad_norm": 0.12581489980220795,
      "learning_rate": 1e-06,
      "loss": 0.0283,
      "step": 204
    },
    {
      "clip_ratio/high_max": 0.002338005848287139,
      "clip_ratio/high_mean": 0.0010561580311332364,
      "clip_ratio/low_mean": 0.0011566291759663727,
      "clip_ratio/low_min": 0.00016311892795783933,
      "clip_ratio/region_mean": 0.002212787207099609,
      "epoch": 1.0606590842811314,
      "grad_norm": 0.14355507493019104,
      "learning_rate": 1e-06,
      "loss": 0.0297,
      "step": 205
    },
    {
      "clip_ratio/high_max": 0.0021115864437888376,
      "clip_ratio/high_mean": 0.0009700281007098965,
      "clip_ratio/low_mean": 0.0009522005675535183,
      "clip_ratio/low_min": 7.406244185403921e-05,
      "clip_ratio/region_mean": 0.0019222286355216056,
      "epoch": 1.0653251676873725,
      "grad_norm": 0.12336955219507217,
      "learning_rate": 1e-06,
      "loss": 0.0089,
      "step": 206
    },
    {
      "clip_ratio/high_max": 0.002382258498982992,
      "clip_ratio/high_mean": 0.0010135391930816695,
      "clip_ratio/low_mean": 0.0010086361144203693,
      "clip_ratio/low_min": 3.1394937650475185e-05,
      "clip_ratio/region_mean": 0.0020221753511577845,
      "epoch": 1.0699912510936134,
      "grad_norm": 0.1214800477027893,
      "learning_rate": 1e-06,
      "loss": -0.0107,
      "step": 207
    },
    {
      "clip_ratio/high_max": 0.002250693134556059,
      "clip_ratio/high_mean": 0.0010060775239253417,
      "clip_ratio/low_mean": 0.000985951299298904,
      "clip_ratio/low_min": 7.19038780516712e-05,
      "clip_ratio/region_mean": 0.0019920288177672774,
      "epoch": 1.0746573344998542,
      "grad_norm": 0.11997034400701523,
      "learning_rate": 1e-06,
      "loss": -0.0186,
      "step": 208
    },
    {
      "clip_ratio/high_max": 0.0029026772317592986,
      "clip_ratio/high_mean": 0.0013477431457431521,
      "clip_ratio/low_mean": 0.0014825363869022112,
      "clip_ratio/low_min": 7.112406728992937e-05,
      "clip_ratio/region_mean": 0.002830279423505999,
      "epoch": 1.079323417906095,
      "grad_norm": 0.10266873240470886,
      "learning_rate": 1e-06,
      "loss": 0.0053,
      "step": 209
    },
    {
      "clip_ratio/high_max": 0.002884775363781955,
      "clip_ratio/high_mean": 0.0012554074091895018,
      "clip_ratio/low_mean": 0.0013781034140265547,
      "clip_ratio/low_min": 8.515732315572677e-05,
      "clip_ratio/region_mean": 0.0026335107686463743,
      "epoch": 1.083989501312336,
      "grad_norm": 0.10379394143819809,
      "learning_rate": 1e-06,
      "loss": 0.0169,
      "step": 210
    },
    {
      "clip_ratio/high_max": 0.0034519216424087062,
      "clip_ratio/high_mean": 0.0014074015489313751,
      "clip_ratio/low_mean": 0.0014037186228961218,
      "clip_ratio/low_min": 0.00019507929209794383,
      "clip_ratio/region_mean": 0.0028111201463616453,
      "epoch": 1.0886555847185768,
      "grad_norm": 0.1156601682305336,
      "learning_rate": 1e-06,
      "loss": -0.022,
      "step": 211
    },
    {
      "clip_ratio/high_max": 0.003109123594185803,
      "clip_ratio/high_mean": 0.0012554195054690354,
      "clip_ratio/low_mean": 0.0012679088504228275,
      "clip_ratio/low_min": 2.6870164219872095e-05,
      "clip_ratio/region_mean": 0.002523328315874096,
      "epoch": 1.0933216681248177,
      "grad_norm": 0.10980050265789032,
      "learning_rate": 1e-06,
      "loss": 0.0093,
      "step": 212
    },
    {
      "clip_ratio/high_max": 0.0031063580099726096,
      "clip_ratio/high_mean": 0.001376570260617882,
      "clip_ratio/low_mean": 0.0013982853379275184,
      "clip_ratio/low_min": 0.0001909848433570005,
      "clip_ratio/region_mean": 0.0027748556021833792,
      "epoch": 1.0979877515310585,
      "grad_norm": 0.12302590161561966,
      "learning_rate": 1e-06,
      "loss": -0.0238,
      "step": 213
    },
    {
      "clip_ratio/high_max": 0.0028415551787475124,
      "clip_ratio/high_mean": 0.0012036114148941124,
      "clip_ratio/low_mean": 0.0016315546963596717,
      "clip_ratio/low_min": 0.00012248194798303302,
      "clip_ratio/region_mean": 0.0028351660730550066,
      "epoch": 1.1026538349372994,
      "grad_norm": 0.11040081828832626,
      "learning_rate": 1e-06,
      "loss": 0.0538,
      "step": 214
    },
    {
      "clip_ratio/high_max": 0.003454074321780354,
      "clip_ratio/high_mean": 0.0014956598315620795,
      "clip_ratio/low_mean": 0.0014544298501277808,
      "clip_ratio/low_min": 0.00010792318607855123,
      "clip_ratio/region_mean": 0.002950089728983585,
      "epoch": 1.1073199183435405,
      "grad_norm": 0.11797984689474106,
      "learning_rate": 1e-06,
      "loss": -0.0177,
      "step": 215
    },
    {
      "clip_ratio/high_max": 0.003867048319079913,
      "clip_ratio/high_mean": 0.0014562339565600269,
      "clip_ratio/low_mean": 0.001637990069866646,
      "clip_ratio/low_min": 0.0001992618836084148,
      "clip_ratio/region_mean": 0.0030942239827709273,
      "epoch": 1.1119860017497813,
      "grad_norm": 0.11448762565851212,
      "learning_rate": 1e-06,
      "loss": 0.011,
      "step": 216
    },
    {
      "clip_ratio/high_max": 0.0038020240754121915,
      "clip_ratio/high_mean": 0.0014951990324334474,
      "clip_ratio/low_mean": 0.0014947549425414763,
      "clip_ratio/low_min": 1.8848009858629666e-05,
      "clip_ratio/region_mean": 0.002989953907672316,
      "epoch": 1.1166520851560222,
      "grad_norm": 0.11441394686698914,
      "learning_rate": 1e-06,
      "loss": -0.024,
      "step": 217
    },
    {
      "clip_ratio/high_max": 0.003183315522619523,
      "clip_ratio/high_mean": 0.0015579840255668387,
      "clip_ratio/low_mean": 0.001578374947712291,
      "clip_ratio/low_min": 8.262306801043451e-05,
      "clip_ratio/region_mean": 0.003136359024210833,
      "epoch": 1.121318168562263,
      "grad_norm": 0.10988708585500717,
      "learning_rate": 1e-06,
      "loss": -0.0113,
      "step": 218
    },
    {
      "clip_ratio/high_max": 0.003523297971696593,
      "clip_ratio/high_mean": 0.0015035541327961255,
      "clip_ratio/low_mean": 0.001617081888980465,
      "clip_ratio/low_min": 0.00021564741837210022,
      "clip_ratio/region_mean": 0.0031206360072246753,
      "epoch": 1.125984251968504,
      "grad_norm": 0.1102115735411644,
      "learning_rate": 1e-06,
      "loss": -0.0041,
      "step": 219
    },
    {
      "clip_ratio/high_max": 0.00301249188487418,
      "clip_ratio/high_mean": 0.0013081325560051482,
      "clip_ratio/low_mean": 0.0017725659417919815,
      "clip_ratio/low_min": 0.000280503664725984,
      "clip_ratio/region_mean": 0.0030806985159870237,
      "epoch": 1.1306503353747448,
      "grad_norm": 0.11126233637332916,
      "learning_rate": 1e-06,
      "loss": 0.0277,
      "step": 220
    },
    {
      "clip_ratio/high_max": 0.003518074328894727,
      "clip_ratio/high_mean": 0.001540239883979666,
      "clip_ratio/low_mean": 0.0019393683578527998,
      "clip_ratio/low_min": 0.0002625341985549312,
      "clip_ratio/region_mean": 0.003479608232737519,
      "epoch": 1.1353164187809857,
      "grad_norm": 0.1257893145084381,
      "learning_rate": 1e-06,
      "loss": 0.0288,
      "step": 221
    },
    {
      "clip_ratio/high_max": 0.00316535220190417,
      "clip_ratio/high_mean": 0.0014314170111902058,
      "clip_ratio/low_mean": 0.0015908013228909113,
      "clip_ratio/low_min": 0.0001344400443485938,
      "clip_ratio/region_mean": 0.003022218370460905,
      "epoch": 1.1399825021872265,
      "grad_norm": 0.11075523495674133,
      "learning_rate": 1e-06,
      "loss": 0.0084,
      "step": 222
    },
    {
      "clip_ratio/high_max": 0.003336296314955689,
      "clip_ratio/high_mean": 0.0014599908281525131,
      "clip_ratio/low_mean": 0.0015942298668960575,
      "clip_ratio/low_min": 4.5674279135710094e-05,
      "clip_ratio/region_mean": 0.0030542206950485706,
      "epoch": 1.1446485855934676,
      "grad_norm": 0.11315368115901947,
      "learning_rate": 1e-06,
      "loss": -0.0113,
      "step": 223
    },
    {
      "clip_ratio/high_max": 0.0034475962165743113,
      "clip_ratio/high_mean": 0.0014397078775800765,
      "clip_ratio/low_mean": 0.0014908503944752738,
      "clip_ratio/low_min": 8.294137023767689e-05,
      "clip_ratio/region_mean": 0.002930558235675562,
      "epoch": 1.1493146689997085,
      "grad_norm": 0.11203787475824356,
      "learning_rate": 1e-06,
      "loss": -0.0192,
      "step": 224
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.020856584821428603,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4096.0,
      "completions/mean_length": 630.7854614257812,
      "completions/mean_terminated_length": 556.973388671875,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "epoch": 1.1539807524059493,
      "grad_norm": 0.13044968247413635,
      "learning_rate": 1e-06,
      "loss": -0.0581,
      "num_tokens": 74183443.0,
      "reward": 0.5831473469734192,
      "reward_std": 0.20123086869716644,
      "rewards/simpleverify_reward/mean": 0.5831473469734192,
      "rewards/simpleverify_reward/std": 0.4930552542209625,
      "step": 225
    },
    {
      "clip_ratio/high_max": 0.0024373836640734226,
      "clip_ratio/high_mean": 0.0009342386965727201,
      "clip_ratio/low_mean": 0.0006380900795193156,
      "clip_ratio/low_min": 1.9373837858438492e-05,
      "clip_ratio/region_mean": 0.0015723287651780993,
      "epoch": 1.1586468358121902,
      "grad_norm": 0.1441110074520111,
      "learning_rate": 1e-06,
      "loss": -0.0016,
      "step": 226
    },
    {
      "clip_ratio/high_max": 0.0018679511122172698,
      "clip_ratio/high_mean": 0.0007531731098424643,
      "clip_ratio/low_mean": 0.0005514680660780868,
      "clip_ratio/low_min": 3.414710045035463e-05,
      "clip_ratio/region_mean": 0.0013046411768300459,
      "epoch": 1.163312919218431,
      "grad_norm": 0.11731994152069092,
      "learning_rate": 1e-06,
      "loss": -0.0113,
      "step": 227
    },
    {
      "clip_ratio/high_max": 0.0016853392771736253,
      "clip_ratio/high_mean": 0.0007521622155763907,
      "clip_ratio/low_mean": 0.0007140377692849142,
      "clip_ratio/low_min": 7.654259661649121e-05,
      "clip_ratio/region_mean": 0.001466200003051199,
      "epoch": 1.167979002624672,
      "grad_norm": 0.129197895526886,
      "learning_rate": 1e-06,
      "loss": 0.0722,
      "step": 228
    },
    {
      "clip_ratio/high_max": 0.001995598682697164,
      "clip_ratio/high_mean": 0.0008241361783802859,
      "clip_ratio/low_mean": 0.0006181895896588685,
      "clip_ratio/low_min": 6.0692511397064663e-05,
      "clip_ratio/region_mean": 0.0014423257962334901,
      "epoch": 1.1726450860309128,
      "grad_norm": 0.12741993367671967,
      "learning_rate": 1e-06,
      "loss": 0.0191,
      "step": 229
    },
    {
      "clip_ratio/high_max": 0.0019897380007023457,
      "clip_ratio/high_mean": 0.0007997162192623364,
      "clip_ratio/low_mean": 0.0007062193435558584,
      "clip_ratio/low_min": 0.00014848026694380678,
      "clip_ratio/region_mean": 0.001505935568275163,
      "epoch": 1.1773111694371536,
      "grad_norm": 0.12224078178405762,
      "learning_rate": 1e-06,
      "loss": 0.0392,
      "step": 230
    },
    {
      "clip_ratio/high_max": 0.0022919785260455683,
      "clip_ratio/high_mean": 0.0010403427004348487,
      "clip_ratio/low_mean": 0.0006667175257462077,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017070602261810564,
      "epoch": 1.1819772528433945,
      "grad_norm": 0.12054876983165741,
      "learning_rate": 1e-06,
      "loss": -0.0258,
      "step": 231
    },
    {
      "clip_ratio/high_max": 0.002352491552301217,
      "clip_ratio/high_mean": 0.0009984478128899354,
      "clip_ratio/low_mean": 0.0007823707128409296,
      "clip_ratio/low_min": 4.849775905313436e-05,
      "clip_ratio/region_mean": 0.0017808185220928863,
      "epoch": 1.1866433362496354,
      "grad_norm": 0.1264234185218811,
      "learning_rate": 1e-06,
      "loss": -0.0172,
      "step": 232
    },
    {
      "clip_ratio/high_max": 0.0024372233201575,
      "clip_ratio/high_mean": 0.0009753797385201324,
      "clip_ratio/low_mean": 0.0009172920572382282,
      "clip_ratio/low_min": 0.0001293046489081462,
      "clip_ratio/region_mean": 0.0018926717530121095,
      "epoch": 1.1913094196558764,
      "grad_norm": 0.13291189074516296,
      "learning_rate": 1e-06,
      "loss": 0.0267,
      "step": 233
    },
    {
      "clip_ratio/high_max": 0.002312093994987663,
      "clip_ratio/high_mean": 0.0010194439801125554,
      "clip_ratio/low_mean": 0.0009816762267291779,
      "clip_ratio/low_min": 5.794309436168987e-05,
      "clip_ratio/region_mean": 0.0020011202359455638,
      "epoch": 1.1959755030621173,
      "grad_norm": 0.13842757046222687,
      "learning_rate": 1e-06,
      "loss": -0.0025,
      "step": 234
    },
    {
      "clip_ratio/high_max": 0.0026617083785822615,
      "clip_ratio/high_mean": 0.0010231079741060967,
      "clip_ratio/low_mean": 0.0010454754701640923,
      "clip_ratio/low_min": 4.5284529733180534e-05,
      "clip_ratio/region_mean": 0.0020685834824689664,
      "epoch": 1.2006415864683582,
      "grad_norm": 0.1373901069164276,
      "learning_rate": 1e-06,
      "loss": 0.0178,
      "step": 235
    },
    {
      "clip_ratio/high_max": 0.0023979547040653415,
      "clip_ratio/high_mean": 0.0010478047333890572,
      "clip_ratio/low_mean": 0.0009897357194859069,
      "clip_ratio/low_min": 4.037210419483017e-05,
      "clip_ratio/region_mean": 0.0020375404419610277,
      "epoch": 1.205307669874599,
      "grad_norm": 0.10937754809856415,
      "learning_rate": 1e-06,
      "loss": -0.0111,
      "step": 236
    },
    {
      "clip_ratio/high_max": 0.0026427910634083673,
      "clip_ratio/high_mean": 0.0010993979885824956,
      "clip_ratio/low_mean": 0.0009976516230381094,
      "clip_ratio/low_min": 7.286989603016991e-05,
      "clip_ratio/region_mean": 0.002097049633448478,
      "epoch": 1.20997375328084,
      "grad_norm": 0.12430103123188019,
      "learning_rate": 1e-06,
      "loss": 0.0004,
      "step": 237
    },
    {
      "clip_ratio/high_max": 0.002406152125331573,
      "clip_ratio/high_mean": 0.0010517175251152366,
      "clip_ratio/low_mean": 0.0009459001103095943,
      "clip_ratio/low_min": 0.00012347814845270477,
      "clip_ratio/region_mean": 0.0019976176117779687,
      "epoch": 1.2146398366870808,
      "grad_norm": 0.11710339784622192,
      "learning_rate": 1e-06,
      "loss": -0.0412,
      "step": 238
    },
    {
      "clip_ratio/high_max": 0.002395392381004058,
      "clip_ratio/high_mean": 0.0010301377096766373,
      "clip_ratio/low_mean": 0.0010858089226530865,
      "clip_ratio/low_min": 7.406985696434276e-05,
      "clip_ratio/region_mean": 0.0021159466632525437,
      "epoch": 1.2193059200933216,
      "grad_norm": 0.12931126356124878,
      "learning_rate": 1e-06,
      "loss": 0.0054,
      "step": 239
    },
    {
      "clip_ratio/high_max": 0.0022705463561578654,
      "clip_ratio/high_mean": 0.0010105919864145108,
      "clip_ratio/low_mean": 0.0010050178279925603,
      "clip_ratio/low_min": 4.5728559598501306e-05,
      "clip_ratio/region_mean": 0.002015609796217177,
      "epoch": 1.2239720034995625,
      "grad_norm": 0.11710505932569504,
      "learning_rate": 1e-06,
      "loss": 0.0022,
      "step": 240
    },
    {
      "clip_ratio/high_max": 0.0037810509602422826,
      "clip_ratio/high_mean": 0.0016871471361810109,
      "clip_ratio/low_mean": 0.001232622227689717,
      "clip_ratio/low_min": 6.750053580617532e-05,
      "clip_ratio/region_mean": 0.0029197693656897172,
      "epoch": 1.2286380869058036,
      "grad_norm": 0.11273294687271118,
      "learning_rate": 1e-06,
      "loss": -0.0586,
      "step": 241
    },
    {
      "clip_ratio/high_max": 0.003505041175230872,
      "clip_ratio/high_mean": 0.0014232041576178744,
      "clip_ratio/low_mean": 0.0016376227940781973,
      "clip_ratio/low_min": 0.00010465596187714254,
      "clip_ratio/region_mean": 0.0030608269007643685,
      "epoch": 1.2333041703120444,
      "grad_norm": 0.12398556619882584,
      "learning_rate": 1e-06,
      "loss": -0.0024,
      "step": 242
    },
    {
      "clip_ratio/high_max": 0.003244882362196222,
      "clip_ratio/high_mean": 0.0012783937017957214,
      "clip_ratio/low_mean": 0.001400425418978557,
      "clip_ratio/low_min": 0.00013850336472387426,
      "clip_ratio/region_mean": 0.0026788191535160877,
      "epoch": 1.2379702537182853,
      "grad_norm": 0.10467694699764252,
      "learning_rate": 1e-06,
      "loss": -0.0119,
      "step": 243
    },
    {
      "clip_ratio/high_max": 0.0030949080755817704,
      "clip_ratio/high_mean": 0.0012627633659576531,
      "clip_ratio/low_mean": 0.0017948421445908025,
      "clip_ratio/low_min": 0.00030324369254230987,
      "clip_ratio/region_mean": 0.0030576054559787735,
      "epoch": 1.2426363371245261,
      "grad_norm": 0.11333815008401871,
      "learning_rate": 1e-06,
      "loss": 0.0715,
      "step": 244
    },
    {
      "clip_ratio/high_max": 0.0029974909775773995,
      "clip_ratio/high_mean": 0.0012981866748305038,
      "clip_ratio/low_mean": 0.0014917710614099633,
      "clip_ratio/low_min": 0.00015549595082120504,
      "clip_ratio/region_mean": 0.002789957681670785,
      "epoch": 1.247302420530767,
      "grad_norm": 0.1137121319770813,
      "learning_rate": 1e-06,
      "loss": 0.0185,
      "step": 245
    },
    {
      "clip_ratio/high_max": 0.0029772713023703545,
      "clip_ratio/high_mean": 0.0012945088965352625,
      "clip_ratio/low_mean": 0.0016020713519537821,
      "clip_ratio/low_min": 0.00036996435846958775,
      "clip_ratio/region_mean": 0.002896580277592875,
      "epoch": 1.2519685039370079,
      "grad_norm": 0.10847195982933044,
      "learning_rate": 1e-06,
      "loss": 0.0386,
      "step": 246
    },
    {
      "clip_ratio/high_max": 0.0037965277806506492,
      "clip_ratio/high_mean": 0.0016158244143298361,
      "clip_ratio/low_mean": 0.0014017973480804358,
      "clip_ratio/low_min": 1.8447461116011254e-05,
      "clip_ratio/region_mean": 0.0030176217696862295,
      "epoch": 1.2566345873432487,
      "grad_norm": 0.10925734043121338,
      "learning_rate": 1e-06,
      "loss": -0.0264,
      "step": 247
    },
    {
      "clip_ratio/high_max": 0.0032282832107739523,
      "clip_ratio/high_mean": 0.0014799068485444877,
      "clip_ratio/low_mean": 0.001553031939693028,
      "clip_ratio/low_min": 4.006410381407477e-05,
      "clip_ratio/region_mean": 0.0030329387518577278,
      "epoch": 1.2613006707494896,
      "grad_norm": 0.11067990958690643,
      "learning_rate": 1e-06,
      "loss": -0.0179,
      "step": 248
    },
    {
      "clip_ratio/high_max": 0.0037833037386008073,
      "clip_ratio/high_mean": 0.0015578158454445656,
      "clip_ratio/low_mean": 0.0017774082225514576,
      "clip_ratio/low_min": 0.00024992932321765693,
      "clip_ratio/region_mean": 0.0033352240279782563,
      "epoch": 1.2659667541557305,
      "grad_norm": 0.11833923310041428,
      "learning_rate": 1e-06,
      "loss": 0.0258,
      "step": 249
    },
    {
      "clip_ratio/high_max": 0.0030646495142718777,
      "clip_ratio/high_mean": 0.001412301578966435,
      "clip_ratio/low_mean": 0.0016754991374909878,
      "clip_ratio/low_min": 0.00018463748347130604,
      "clip_ratio/region_mean": 0.003087800709181465,
      "epoch": 1.2706328375619713,
      "grad_norm": 0.12140277028083801,
      "learning_rate": 1e-06,
      "loss": -0.0031,
      "step": 250
    },
    {
      "clip_ratio/high_max": 0.003906381083652377,
      "clip_ratio/high_mean": 0.001518819794000592,
      "clip_ratio/low_mean": 0.0017400763535988517,
      "clip_ratio/low_min": 0.00014918321903678589,
      "clip_ratio/region_mean": 0.0032588961912551895,
      "epoch": 1.2752989209682122,
      "grad_norm": 0.11883580684661865,
      "learning_rate": 1e-06,
      "loss": 0.017,
      "step": 251
    },
    {
      "clip_ratio/high_max": 0.0037435776393976994,
      "clip_ratio/high_mean": 0.0015063137616380118,
      "clip_ratio/low_mean": 0.0014569994200428482,
      "clip_ratio/low_min": 0.00014257947987061925,
      "clip_ratio/region_mean": 0.0029633131925947964,
      "epoch": 1.2799650043744533,
      "grad_norm": 0.10138453543186188,
      "learning_rate": 1e-06,
      "loss": -0.0117,
      "step": 252
    },
    {
      "clip_ratio/high_max": 0.0032856406323844567,
      "clip_ratio/high_mean": 0.0014929415192455053,
      "clip_ratio/low_mean": 0.0015651887915737461,
      "clip_ratio/low_min": 6.695614501950331e-05,
      "clip_ratio/region_mean": 0.0030581303435610607,
      "epoch": 1.2846310877806941,
      "grad_norm": 0.11209844052791595,
      "learning_rate": 1e-06,
      "loss": -0.0003,
      "step": 253
    },
    {
      "clip_ratio/high_max": 0.003584752899769228,
      "clip_ratio/high_mean": 0.0015977217735780869,
      "clip_ratio/low_mean": 0.0014488598208117764,
      "clip_ratio/low_min": 0.00017075835057767108,
      "clip_ratio/region_mean": 0.0030465816234936938,
      "epoch": 1.289297171186935,
      "grad_norm": 0.10436536371707916,
      "learning_rate": 1e-06,
      "loss": -0.0418,
      "step": 254
    },
    {
      "clip_ratio/high_max": 0.0037288929233909585,
      "clip_ratio/high_mean": 0.0015298034595616627,
      "clip_ratio/low_mean": 0.0015732653882878367,
      "clip_ratio/low_min": 0.0001267130828637164,
      "clip_ratio/region_mean": 0.0031030688260216266,
      "epoch": 1.2939632545931758,
      "grad_norm": 0.11299216747283936,
      "learning_rate": 1e-06,
      "loss": 0.0048,
      "step": 255
    },
    {
      "clip_ratio/high_max": 0.003397808046429418,
      "clip_ratio/high_mean": 0.0015400144511659164,
      "clip_ratio/low_mean": 0.0016235503462667111,
      "clip_ratio/low_min": 7.976430606504437e-05,
      "clip_ratio/region_mean": 0.0031635647173970938,
      "epoch": 1.2986293379994167,
      "grad_norm": 0.10357529670000076,
      "learning_rate": 1e-06,
      "loss": 0.0016,
      "step": 256
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.023716517857142905,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4040.0,
      "completions/mean_length": 650.9552001953125,
      "completions/mean_terminated_length": 567.265869140625,
      "completions/min_length": 77.0,
      "completions/min_terminated_length": 77.0,
      "epoch": 1.3032954214056576,
      "grad_norm": 0.13808058202266693,
      "learning_rate": 1e-06,
      "loss": 0.0098,
      "num_tokens": 83570808.0,
      "reward": 0.5787528157234192,
      "reward_std": 0.1950569897890091,
      "rewards/simpleverify_reward/mean": 0.5787528157234192,
      "rewards/simpleverify_reward/std": 0.4937762916088104,
      "step": 257
    },
    {
      "clip_ratio/high_max": 0.0021518862813536543,
      "clip_ratio/high_mean": 0.0008497496492054779,
      "clip_ratio/low_mean": 0.0005704966224584496,
      "clip_ratio/low_min": 3.926059616787825e-05,
      "clip_ratio/region_mean": 0.0014202462589310016,
      "epoch": 1.3079615048118984,
      "grad_norm": 0.1268431395292282,
      "learning_rate": 1e-06,
      "loss": 0.0151,
      "step": 258
    },
    {
      "clip_ratio/high_max": 0.001882116968772607,
      "clip_ratio/high_mean": 0.0008119259218801744,
      "clip_ratio/low_mean": 0.0005783470151072834,
      "clip_ratio/low_min": 2.575498729129322e-05,
      "clip_ratio/region_mean": 0.0013902729187975638,
      "epoch": 1.3126275882181395,
      "grad_norm": 0.11980373412370682,
      "learning_rate": 1e-06,
      "loss": -0.0265,
      "step": 259
    },
    {
      "clip_ratio/high_max": 0.0019301171923871152,
      "clip_ratio/high_mean": 0.0008276496791950194,
      "clip_ratio/low_mean": 0.0006175059224915458,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014451555980485864,
      "epoch": 1.3172936716243804,
      "grad_norm": 0.12828215956687927,
      "learning_rate": 1e-06,
      "loss": 0.0095,
      "step": 260
    },
    {
      "clip_ratio/high_max": 0.0023053980694385245,
      "clip_ratio/high_mean": 0.0009798652172321454,
      "clip_ratio/low_mean": 0.0006183355890243547,
      "clip_ratio/low_min": 1.200076803797856e-05,
      "clip_ratio/region_mean": 0.001598200818989426,
      "epoch": 1.3219597550306212,
      "grad_norm": 0.11807168275117874,
      "learning_rate": 1e-06,
      "loss": -0.006,
      "step": 261
    },
    {
      "clip_ratio/high_max": 0.0023316209772019647,
      "clip_ratio/high_mean": 0.0009486436429142486,
      "clip_ratio/low_mean": 0.0006555089648827561,
      "clip_ratio/low_min": 6.467975981649943e-05,
      "clip_ratio/region_mean": 0.0016041525814216584,
      "epoch": 1.326625838436862,
      "grad_norm": 0.12681686878204346,
      "learning_rate": 1e-06,
      "loss": 0.0073,
      "step": 262
    },
    {
      "clip_ratio/high_max": 0.0022550920439243782,
      "clip_ratio/high_mean": 0.0009245580640708795,
      "clip_ratio/low_mean": 0.0006850120917079039,
      "clip_ratio/low_min": 3.9433081838069484e-05,
      "clip_ratio/region_mean": 0.0016095701248559635,
      "epoch": 1.331291921843103,
      "grad_norm": 0.1202269196510315,
      "learning_rate": 1e-06,
      "loss": -0.0027,
      "step": 263
    },
    {
      "clip_ratio/high_max": 0.002182445976359304,
      "clip_ratio/high_mean": 0.000812479091109708,
      "clip_ratio/low_mean": 0.0007228911181300646,
      "clip_ratio/low_min": 6.683496303594438e-05,
      "clip_ratio/region_mean": 0.0015353702328866348,
      "epoch": 1.3359580052493438,
      "grad_norm": 0.12395063787698746,
      "learning_rate": 1e-06,
      "loss": 0.0459,
      "step": 264
    },
    {
      "clip_ratio/high_max": 0.0023663923275307752,
      "clip_ratio/high_mean": 0.0009941451462509576,
      "clip_ratio/low_mean": 0.0008251246872532647,
      "clip_ratio/low_min": 4.236000131641049e-05,
      "clip_ratio/region_mean": 0.0018192698335042223,
      "epoch": 1.3406240886555847,
      "grad_norm": 0.12286343425512314,
      "learning_rate": 1e-06,
      "loss": -0.0336,
      "step": 265
    },
    {
      "clip_ratio/high_max": 0.001909790084027918,
      "clip_ratio/high_mean": 0.0008609516589785926,
      "clip_ratio/low_mean": 0.0010515750764170662,
      "clip_ratio/low_min": 0.00010901141831709538,
      "clip_ratio/region_mean": 0.0019125267208437435,
      "epoch": 1.3452901720618256,
      "grad_norm": 0.12255705893039703,
      "learning_rate": 1e-06,
      "loss": 0.0392,
      "step": 266
    },
    {
      "clip_ratio/high_max": 0.002582270171842538,
      "clip_ratio/high_mean": 0.0009708337383926846,
      "clip_ratio/low_mean": 0.0007798731166985817,
      "clip_ratio/low_min": 3.863987512886524e-05,
      "clip_ratio/region_mean": 0.0017507069132989272,
      "epoch": 1.3499562554680664,
      "grad_norm": 0.11288341134786606,
      "learning_rate": 1e-06,
      "loss": -0.0433,
      "step": 267
    },
    {
      "clip_ratio/high_max": 0.0022060873016016558,
      "clip_ratio/high_mean": 0.0008841539329296211,
      "clip_ratio/low_mean": 0.000865750517732522,
      "clip_ratio/low_min": 4.153243025939446e-05,
      "clip_ratio/region_mean": 0.0017499044406577013,
      "epoch": 1.3546223388743073,
      "grad_norm": 0.1235344186425209,
      "learning_rate": 1e-06,
      "loss": -0.0102,
      "step": 268
    },
    {
      "clip_ratio/high_max": 0.0021967909269733354,
      "clip_ratio/high_mean": 0.0009288905166613404,
      "clip_ratio/low_mean": 0.0009439619279874023,
      "clip_ratio/low_min": 9.255539043806493e-05,
      "clip_ratio/region_mean": 0.001872852459200658,
      "epoch": 1.3592884222805481,
      "grad_norm": 0.11238186806440353,
      "learning_rate": 1e-06,
      "loss": -0.0242,
      "step": 269
    },
    {
      "clip_ratio/high_max": 0.0021392481721704826,
      "clip_ratio/high_mean": 0.0009035512339323759,
      "clip_ratio/low_mean": 0.0011324123552185483,
      "clip_ratio/low_min": 5.951017010374926e-05,
      "clip_ratio/region_mean": 0.00203596363280667,
      "epoch": 1.3639545056867892,
      "grad_norm": 0.12755899131298065,
      "learning_rate": 1e-06,
      "loss": 0.0242,
      "step": 270
    },
    {
      "clip_ratio/high_max": 0.0020534530194709077,
      "clip_ratio/high_mean": 0.0008245141434599645,
      "clip_ratio/low_mean": 0.0009495596787019167,
      "clip_ratio/low_min": 2.5625256967032328e-05,
      "clip_ratio/region_mean": 0.0017740738039719872,
      "epoch": 1.36862058909303,
      "grad_norm": 0.12751242518424988,
      "learning_rate": 1e-06,
      "loss": -0.0092,
      "step": 271
    },
    {
      "clip_ratio/high_max": 0.002213423460489139,
      "clip_ratio/high_mean": 0.0009677669040684123,
      "clip_ratio/low_mean": 0.0010348346986575052,
      "clip_ratio/low_min": 9.139573558059055e-05,
      "clip_ratio/region_mean": 0.0020026016354677267,
      "epoch": 1.373286672499271,
      "grad_norm": 0.1233547031879425,
      "learning_rate": 1e-06,
      "loss": -0.0087,
      "step": 272
    },
    {
      "clip_ratio/high_max": 0.004143118596402928,
      "clip_ratio/high_mean": 0.0016997970378724858,
      "clip_ratio/low_mean": 0.0014228126583475387,
      "clip_ratio/low_min": 0.00011823543354694266,
      "clip_ratio/region_mean": 0.0031226096907630563,
      "epoch": 1.3779527559055118,
      "grad_norm": 0.11155719310045242,
      "learning_rate": 1e-06,
      "loss": 0.0091,
      "step": 273
    },
    {
      "clip_ratio/high_max": 0.003690662757435348,
      "clip_ratio/high_mean": 0.0013826434142174548,
      "clip_ratio/low_mean": 0.0013987527127028443,
      "clip_ratio/low_min": 0.00010228711289528292,
      "clip_ratio/region_mean": 0.0027813961132778786,
      "epoch": 1.3826188393117527,
      "grad_norm": 0.11417661607265472,
      "learning_rate": 1e-06,
      "loss": 0.0144,
      "step": 274
    },
    {
      "clip_ratio/high_max": 0.0029901251473347656,
      "clip_ratio/high_mean": 0.001377975564537337,
      "clip_ratio/low_mean": 0.0013107542326906696,
      "clip_ratio/low_min": 7.288961751328316e-05,
      "clip_ratio/region_mean": 0.0026887298372457735,
      "epoch": 1.3872849227179935,
      "grad_norm": 0.10374260693788528,
      "learning_rate": 1e-06,
      "loss": -0.0271,
      "step": 275
    },
    {
      "clip_ratio/high_max": 0.0033425524743506685,
      "clip_ratio/high_mean": 0.0013792945865134243,
      "clip_ratio/low_mean": 0.001423518766387133,
      "clip_ratio/low_min": 8.59089705045335e-05,
      "clip_ratio/region_mean": 0.002802813396556303,
      "epoch": 1.3919510061242344,
      "grad_norm": 0.11074087768793106,
      "learning_rate": 1e-06,
      "loss": 0.0088,
      "step": 276
    },
    {
      "clip_ratio/high_max": 0.0035491459566401318,
      "clip_ratio/high_mean": 0.0015569902971037664,
      "clip_ratio/low_mean": 0.0013285467903187964,
      "clip_ratio/low_min": 8.863146467774641e-05,
      "clip_ratio/region_mean": 0.0028855371347162873,
      "epoch": 1.3966170895304755,
      "grad_norm": 0.1087472066283226,
      "learning_rate": 1e-06,
      "loss": -0.0066,
      "step": 277
    },
    {
      "clip_ratio/high_max": 0.0035486265187500976,
      "clip_ratio/high_mean": 0.0014977077116782311,
      "clip_ratio/low_mean": 0.0015511686997342622,
      "clip_ratio/low_min": 0.0001948686931427801,
      "clip_ratio/region_mean": 0.0030488764023175463,
      "epoch": 1.4012831729367163,
      "grad_norm": 0.11436290293931961,
      "learning_rate": 1e-06,
      "loss": 0.0067,
      "step": 278
    },
    {
      "clip_ratio/high_max": 0.003244845043809619,
      "clip_ratio/high_mean": 0.001384616411087336,
      "clip_ratio/low_mean": 0.0014929241151548922,
      "clip_ratio/low_min": 0.00020924706313962815,
      "clip_ratio/region_mean": 0.0028775404862244613,
      "epoch": 1.4059492563429572,
      "grad_norm": 0.10665514320135117,
      "learning_rate": 1e-06,
      "loss": -0.0034,
      "step": 279
    },
    {
      "clip_ratio/high_max": 0.00314182773581706,
      "clip_ratio/high_mean": 0.0012556378642329946,
      "clip_ratio/low_mean": 0.0014716223122377414,
      "clip_ratio/low_min": 0.00011792662735388149,
      "clip_ratio/region_mean": 0.002727260216488503,
      "epoch": 1.410615339749198,
      "grad_norm": 0.11049500852823257,
      "learning_rate": 1e-06,
      "loss": 0.0452,
      "step": 280
    },
    {
      "clip_ratio/high_max": 0.003479857674392406,
      "clip_ratio/high_mean": 0.0015052516137075145,
      "clip_ratio/low_mean": 0.0014905521202308591,
      "clip_ratio/low_min": 8.315634659084026e-05,
      "clip_ratio/region_mean": 0.002995803763042204,
      "epoch": 1.415281423155439,
      "grad_norm": 0.11019416898488998,
      "learning_rate": 1e-06,
      "loss": -0.0343,
      "step": 281
    },
    {
      "clip_ratio/high_max": 0.002758573042228818,
      "clip_ratio/high_mean": 0.0012585490912897512,
      "clip_ratio/low_mean": 0.0018928376084659249,
      "clip_ratio/low_min": 0.00024258907433249988,
      "clip_ratio/region_mean": 0.0031513867143075913,
      "epoch": 1.4199475065616798,
      "grad_norm": 0.10782289505004883,
      "learning_rate": 1e-06,
      "loss": 0.0386,
      "step": 282
    },
    {
      "clip_ratio/high_max": 0.0035305990313645452,
      "clip_ratio/high_mean": 0.0013958701529190876,
      "clip_ratio/low_mean": 0.0015106644386833068,
      "clip_ratio/low_min": 0.00010343916801502928,
      "clip_ratio/region_mean": 0.0029065345879644156,
      "epoch": 1.4246135899679206,
      "grad_norm": 0.10185535997152328,
      "learning_rate": 1e-06,
      "loss": -0.0439,
      "step": 283
    },
    {
      "clip_ratio/high_max": 0.0036924673840985633,
      "clip_ratio/high_mean": 0.001454448327422142,
      "clip_ratio/low_mean": 0.0015676819421059918,
      "clip_ratio/low_min": 6.317899715213571e-05,
      "clip_ratio/region_mean": 0.003022130193130579,
      "epoch": 1.4292796733741615,
      "grad_norm": 0.1135973334312439,
      "learning_rate": 1e-06,
      "loss": -0.011,
      "step": 284
    },
    {
      "clip_ratio/high_max": 0.003138421798212221,
      "clip_ratio/high_mean": 0.001457390211726306,
      "clip_ratio/low_mean": 0.0014976706479501445,
      "clip_ratio/low_min": 7.617904338985682e-05,
      "clip_ratio/region_mean": 0.002955060816020705,
      "epoch": 1.4339457567804024,
      "grad_norm": 0.10585653781890869,
      "learning_rate": 1e-06,
      "loss": -0.0247,
      "step": 285
    },
    {
      "clip_ratio/high_max": 0.0034529809272498824,
      "clip_ratio/high_mean": 0.0014299384092737455,
      "clip_ratio/low_mean": 0.0019528168813849334,
      "clip_ratio/low_min": 0.00019594030891312286,
      "clip_ratio/region_mean": 0.0033827553124865517,
      "epoch": 1.4386118401866432,
      "grad_norm": 0.11233805119991302,
      "learning_rate": 1e-06,
      "loss": 0.0234,
      "step": 286
    },
    {
      "clip_ratio/high_max": 0.003996423431090079,
      "clip_ratio/high_mean": 0.0014924178685760126,
      "clip_ratio/low_mean": 0.0016111696968437172,
      "clip_ratio/low_min": 9.366350059281103e-05,
      "clip_ratio/region_mean": 0.0031035875290399417,
      "epoch": 1.443277923592884,
      "grad_norm": 0.10819321870803833,
      "learning_rate": 1e-06,
      "loss": -0.0099,
      "step": 287
    },
    {
      "clip_ratio/high_max": 0.003536757591064088,
      "clip_ratio/high_mean": 0.0015316535609599669,
      "clip_ratio/low_mean": 0.00167769544350449,
      "clip_ratio/low_min": 0.0001826616799007752,
      "clip_ratio/region_mean": 0.0032093490153783932,
      "epoch": 1.4479440069991252,
      "grad_norm": 0.10967979580163956,
      "learning_rate": 1e-06,
      "loss": -0.0095,
      "step": 288
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.026576450892857095,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4063.0,
      "completions/mean_length": 666.6900634765625,
      "completions/mean_terminated_length": 573.0628662109375,
      "completions/min_length": 38.0,
      "completions/min_terminated_length": 38.0,
      "epoch": 1.452610090405366,
      "grad_norm": 0.12555056810379028,
      "learning_rate": 1e-06,
      "loss": -0.0117,
      "num_tokens": 93001900.0,
      "reward": 0.5763811469078064,
      "reward_std": 0.19148065149784088,
      "rewards/simpleverify_reward/mean": 0.5763811469078064,
      "rewards/simpleverify_reward/std": 0.49414870142936707,
      "step": 289
    },
    {
      "clip_ratio/high_max": 0.002160648495191708,
      "clip_ratio/high_mean": 0.0008633962243038695,
      "clip_ratio/low_mean": 0.0005861948720848886,
      "clip_ratio/low_min": 1.3397641851042863e-05,
      "clip_ratio/region_mean": 0.0014495910872938111,
      "epoch": 1.457276173811607,
      "grad_norm": 0.116352379322052,
      "learning_rate": 1e-06,
      "loss": 0.0111,
      "step": 290
    },
    {
      "clip_ratio/high_max": 0.0022229286696529016,
      "clip_ratio/high_mean": 0.0009015193463710602,
      "clip_ratio/low_mean": 0.00047310482113971375,
      "clip_ratio/low_min": 3.0259212508099154e-05,
      "clip_ratio/region_mean": 0.0013746241456829011,
      "epoch": 1.4619422572178478,
      "grad_norm": 0.119874507188797,
      "learning_rate": 1e-06,
      "loss": 0.0023,
      "step": 291
    },
    {
      "clip_ratio/high_max": 0.002048235764959827,
      "clip_ratio/high_mean": 0.000786712755143526,
      "clip_ratio/low_mean": 0.0006537581693919492,
      "clip_ratio/low_min": 4.4827045712736435e-05,
      "clip_ratio/region_mean": 0.0014404709399968851,
      "epoch": 1.4666083406240886,
      "grad_norm": 0.12702429294586182,
      "learning_rate": 1e-06,
      "loss": 0.0025,
      "step": 292
    },
    {
      "clip_ratio/high_max": 0.0021407743806776125,
      "clip_ratio/high_mean": 0.0008375630550290225,
      "clip_ratio/low_mean": 0.0006024478880135575,
      "clip_ratio/low_min": 1.315789450018201e-05,
      "clip_ratio/region_mean": 0.0014400109103007708,
      "epoch": 1.4712744240303295,
      "grad_norm": 0.12075016647577286,
      "learning_rate": 1e-06,
      "loss": 0.0081,
      "step": 293
    },
    {
      "clip_ratio/high_max": 0.002063258543785196,
      "clip_ratio/high_mean": 0.0008076634185272269,
      "clip_ratio/low_mean": 0.0005785654211649671,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013862288360542152,
      "epoch": 1.4759405074365703,
      "grad_norm": 0.12994447350502014,
      "learning_rate": 1e-06,
      "loss": -0.0123,
      "step": 294
    },
    {
      "clip_ratio/high_max": 0.001978936474188231,
      "clip_ratio/high_mean": 0.0008879185297701042,
      "clip_ratio/low_mean": 0.0006442111534852302,
      "clip_ratio/low_min": 5.831062844663393e-05,
      "clip_ratio/region_mean": 0.0015321296777983662,
      "epoch": 1.4806065908428114,
      "grad_norm": 0.1255219429731369,
      "learning_rate": 1e-06,
      "loss": -0.0137,
      "step": 295
    },
    {
      "clip_ratio/high_max": 0.0021538263717957307,
      "clip_ratio/high_mean": 0.0009058354498847621,
      "clip_ratio/low_mean": 0.0007116124215826858,
      "clip_ratio/low_min": 3.09917340928223e-05,
      "clip_ratio/region_mean": 0.0016174478550965432,
      "epoch": 1.4852726742490523,
      "grad_norm": 0.13035698235034943,
      "learning_rate": 1e-06,
      "loss": -0.0243,
      "step": 296
    },
    {
      "clip_ratio/high_max": 0.0019168691142112948,
      "clip_ratio/high_mean": 0.0008599165921623353,
      "clip_ratio/low_mean": 0.0007308219292099238,
      "clip_ratio/low_min": 5.262999366095755e-05,
      "clip_ratio/region_mean": 0.0015907385241007432,
      "epoch": 1.4899387576552932,
      "grad_norm": 0.1258716583251953,
      "learning_rate": 1e-06,
      "loss": -0.015,
      "step": 297
    },
    {
      "clip_ratio/high_max": 0.002627965142892208,
      "clip_ratio/high_mean": 0.0010332556594221387,
      "clip_ratio/low_mean": 0.0008474831356579671,
      "clip_ratio/low_min": 4.2870276956819e-05,
      "clip_ratio/region_mean": 0.0018807388260029256,
      "epoch": 1.494604841061534,
      "grad_norm": 0.1218964010477066,
      "learning_rate": 1e-06,
      "loss": -0.0256,
      "step": 298
    },
    {
      "clip_ratio/high_max": 0.0021075557669973932,
      "clip_ratio/high_mean": 0.000935405092604924,
      "clip_ratio/low_mean": 0.0008960193354141666,
      "clip_ratio/low_min": 6.579255659744376e-05,
      "clip_ratio/region_mean": 0.0018314243934582919,
      "epoch": 1.4992709244677749,
      "grad_norm": 0.13343334197998047,
      "learning_rate": 1e-06,
      "loss": -0.0053,
      "step": 299
    },
    {
      "clip_ratio/high_max": 0.002547697236877866,
      "clip_ratio/high_mean": 0.0009431301823497051,
      "clip_ratio/low_mean": 0.0009233878772647586,
      "clip_ratio/low_min": 6.67174444970442e-05,
      "clip_ratio/region_mean": 0.0018665180905372836,
      "epoch": 1.5039370078740157,
      "grad_norm": 0.12178276479244232,
      "learning_rate": 1e-06,
      "loss": 0.0207,
      "step": 300
    },
    {
      "clip_ratio/high_max": 0.002082353639707435,
      "clip_ratio/high_mean": 0.000841930006572511,
      "clip_ratio/low_mean": 0.0009710317517601652,
      "clip_ratio/low_min": 7.222014937724452e-05,
      "clip_ratio/region_mean": 0.0018129617819795385,
      "epoch": 1.5086030912802566,
      "grad_norm": 0.11422529816627502,
      "learning_rate": 1e-06,
      "loss": 0.0267,
      "step": 301
    },
    {
      "clip_ratio/high_max": 0.0023214292523334734,
      "clip_ratio/high_mean": 0.0009400309099873994,
      "clip_ratio/low_mean": 0.0008903673624445219,
      "clip_ratio/low_min": 3.713310979946982e-05,
      "clip_ratio/region_mean": 0.0018303982651559636,
      "epoch": 1.5132691746864975,
      "grad_norm": 0.1316445916891098,
      "learning_rate": 1e-06,
      "loss": -0.0185,
      "step": 302
    },
    {
      "clip_ratio/high_max": 0.0022857034055050462,
      "clip_ratio/high_mean": 0.0009954730066965567,
      "clip_ratio/low_mean": 0.0009809096591197886,
      "clip_ratio/low_min": 2.8668434424616862e-05,
      "clip_ratio/region_mean": 0.0019763826931011863,
      "epoch": 1.5179352580927383,
      "grad_norm": 0.13324519991874695,
      "learning_rate": 1e-06,
      "loss": 0.0231,
      "step": 303
    },
    {
      "clip_ratio/high_max": 0.002196057641413063,
      "clip_ratio/high_mean": 0.0010242104908684269,
      "clip_ratio/low_mean": 0.0010243193200949463,
      "clip_ratio/low_min": 0.0001708466952550225,
      "clip_ratio/region_mean": 0.002048529793682974,
      "epoch": 1.5226013414989792,
      "grad_norm": 0.11277655512094498,
      "learning_rate": 1e-06,
      "loss": 0.02,
      "step": 304
    },
    {
      "clip_ratio/high_max": 0.003713684949616436,
      "clip_ratio/high_mean": 0.0015837133105378598,
      "clip_ratio/low_mean": 0.0014507636769849341,
      "clip_ratio/low_min": 0.00016174334996321704,
      "clip_ratio/region_mean": 0.003034477005712688,
      "epoch": 1.52726742490522,
      "grad_norm": 0.10689415782690048,
      "learning_rate": 1e-06,
      "loss": -0.0123,
      "step": 305
    },
    {
      "clip_ratio/high_max": 0.00324887359602144,
      "clip_ratio/high_mean": 0.0013520644970412832,
      "clip_ratio/low_mean": 0.0014389029602170922,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.002790967482724227,
      "epoch": 1.531933508311461,
      "grad_norm": 0.10016504675149918,
      "learning_rate": 1e-06,
      "loss": 0.0105,
      "step": 306
    },
    {
      "clip_ratio/high_max": 0.0031683656634413637,
      "clip_ratio/high_mean": 0.001272190758754732,
      "clip_ratio/low_mean": 0.0013830785392201506,
      "clip_ratio/low_min": 0.00011225256639590953,
      "clip_ratio/region_mean": 0.0026552692579571158,
      "epoch": 1.536599591717702,
      "grad_norm": 0.10379119217395782,
      "learning_rate": 1e-06,
      "loss": 0.0017,
      "step": 307
    },
    {
      "clip_ratio/high_max": 0.003237110322515946,
      "clip_ratio/high_mean": 0.0012915346887893975,
      "clip_ratio/low_mean": 0.00157935261449893,
      "clip_ratio/low_min": 0.0002178343311243225,
      "clip_ratio/region_mean": 0.002870887314202264,
      "epoch": 1.5412656751239429,
      "grad_norm": 0.11277302354574203,
      "learning_rate": 1e-06,
      "loss": 0.0018,
      "step": 308
    },
    {
      "clip_ratio/high_max": 0.0034774578671203926,
      "clip_ratio/high_mean": 0.0013901542624807917,
      "clip_ratio/low_mean": 0.0014009257902216632,
      "clip_ratio/low_min": 0.00012018540292046964,
      "clip_ratio/region_mean": 0.002791080085444264,
      "epoch": 1.5459317585301837,
      "grad_norm": 0.10168322175741196,
      "learning_rate": 1e-06,
      "loss": 0.0074,
      "step": 309
    },
    {
      "clip_ratio/high_max": 0.003713514954142738,
      "clip_ratio/high_mean": 0.0014563254699169192,
      "clip_ratio/low_mean": 0.0015838820581848267,
      "clip_ratio/low_min": 0.00016683366720826598,
      "clip_ratio/region_mean": 0.0030402075717574917,
      "epoch": 1.5505978419364246,
      "grad_norm": 0.10710260272026062,
      "learning_rate": 1e-06,
      "loss": -0.013,
      "step": 310
    },
    {
      "clip_ratio/high_max": 0.003839484168565832,
      "clip_ratio/high_mean": 0.0015137174414121546,
      "clip_ratio/low_mean": 0.0014364117378136143,
      "clip_ratio/low_min": 0.00019619084559963085,
      "clip_ratio/region_mean": 0.002950129215605557,
      "epoch": 1.5552639253426657,
      "grad_norm": 0.11161097139120102,
      "learning_rate": 1e-06,
      "loss": -0.0144,
      "step": 311
    },
    {
      "clip_ratio/high_max": 0.0034504338836995885,
      "clip_ratio/high_mean": 0.0015048963214212563,
      "clip_ratio/low_mean": 0.0013769024881185032,
      "clip_ratio/low_min": 7.843084495107178e-05,
      "clip_ratio/region_mean": 0.002881798856833484,
      "epoch": 1.5599300087489065,
      "grad_norm": 0.11442887783050537,
      "learning_rate": 1e-06,
      "loss": -0.025,
      "step": 312
    },
    {
      "clip_ratio/high_max": 0.003165156696923077,
      "clip_ratio/high_mean": 0.0015039196150610223,
      "clip_ratio/low_mean": 0.0015210767869575648,
      "clip_ratio/low_min": 0.00011679028193611884,
      "clip_ratio/region_mean": 0.003024996360181831,
      "epoch": 1.5645960921551474,
      "grad_norm": 0.11068805307149887,
      "learning_rate": 1e-06,
      "loss": -0.0157,
      "step": 313
    },
    {
      "clip_ratio/high_max": 0.004089017282240093,
      "clip_ratio/high_mean": 0.0017181080402224325,
      "clip_ratio/low_mean": 0.0015445030985574704,
      "clip_ratio/low_min": 8.78369937709067e-05,
      "clip_ratio/region_mean": 0.0032626111496938393,
      "epoch": 1.5692621755613883,
      "grad_norm": 0.10936661064624786,
      "learning_rate": 1e-06,
      "loss": -0.0263,
      "step": 314
    },
    {
      "clip_ratio/high_max": 0.003602085984311998,
      "clip_ratio/high_mean": 0.00157911660062382,
      "clip_ratio/low_mean": 0.0016133072713273577,
      "clip_ratio/low_min": 0.00020800574202439748,
      "clip_ratio/region_mean": 0.0031924238719511777,
      "epoch": 1.5739282589676291,
      "grad_norm": 0.11808856576681137,
      "learning_rate": 1e-06,
      "loss": -0.006,
      "step": 315
    },
    {
      "clip_ratio/high_max": 0.0038861082066432573,
      "clip_ratio/high_mean": 0.001470939714636188,
      "clip_ratio/low_mean": 0.0016469136389787309,
      "clip_ratio/low_min": 0.0001798182347556576,
      "clip_ratio/region_mean": 0.00311785341182258,
      "epoch": 1.57859434237387,
      "grad_norm": 0.10651401430368423,
      "learning_rate": 1e-06,
      "loss": 0.0199,
      "step": 316
    },
    {
      "clip_ratio/high_max": 0.002934512034698855,
      "clip_ratio/high_mean": 0.0013017165329074487,
      "clip_ratio/low_mean": 0.0015416514142998494,
      "clip_ratio/low_min": 0.0001656166659813607,
      "clip_ratio/region_mean": 0.002843367896275595,
      "epoch": 1.5832604257801108,
      "grad_norm": 0.10243922472000122,
      "learning_rate": 1e-06,
      "loss": 0.0261,
      "step": 317
    },
    {
      "clip_ratio/high_max": 0.003942028328310698,
      "clip_ratio/high_mean": 0.0016009552891773637,
      "clip_ratio/low_mean": 0.0016188754889299162,
      "clip_ratio/low_min": 7.892062967584934e-05,
      "clip_ratio/region_mean": 0.003219830774469301,
      "epoch": 1.5879265091863517,
      "grad_norm": 0.11441457271575928,
      "learning_rate": 1e-06,
      "loss": -0.0193,
      "step": 318
    },
    {
      "clip_ratio/high_max": 0.0030020347621757537,
      "clip_ratio/high_mean": 0.0014899942070769612,
      "clip_ratio/low_mean": 0.0018149302995880134,
      "clip_ratio/low_min": 7.857345190132037e-05,
      "clip_ratio/region_mean": 0.003304924597614445,
      "epoch": 1.5925925925925926,
      "grad_norm": 0.11282162368297577,
      "learning_rate": 1e-06,
      "loss": 0.0223,
      "step": 319
    },
    {
      "clip_ratio/high_max": 0.0033832769258879125,
      "clip_ratio/high_mean": 0.001501935545093147,
      "clip_ratio/low_mean": 0.0015683064757467946,
      "clip_ratio/low_min": 0.0002546941541368142,
      "clip_ratio/region_mean": 0.003070241968089249,
      "epoch": 1.5972586759988334,
      "grad_norm": 0.10051309317350388,
      "learning_rate": 1e-06,
      "loss": 0.0194,
      "step": 320
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0239955357142857,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4094.0,
      "completions/mean_length": 651.27587890625,
      "completions/mean_terminated_length": 566.585693359375,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 1.6019247594050743,
      "grad_norm": 0.13548938930034637,
      "learning_rate": 1e-06,
      "loss": -0.053,
      "num_tokens": 102375263.0,
      "reward": 0.59521484375,
      "reward_std": 0.18863287568092346,
      "rewards/simpleverify_reward/mean": 0.59521484375,
      "rewards/simpleverify_reward/std": 0.4908675253391266,
      "step": 321
    },
    {
      "clip_ratio/high_max": 0.0019266981180408038,
      "clip_ratio/high_mean": 0.0008372876018256648,
      "clip_ratio/low_mean": 0.0005748561143263942,
      "clip_ratio/low_min": 2.7478713491291273e-05,
      "clip_ratio/region_mean": 0.0014121436652203556,
      "epoch": 1.6065908428113151,
      "grad_norm": 0.11016207933425903,
      "learning_rate": 1e-06,
      "loss": 0.0208,
      "step": 322
    },
    {
      "clip_ratio/high_max": 0.001917662113555707,
      "clip_ratio/high_mean": 0.0008604406539234333,
      "clip_ratio/low_mean": 0.0006482774142568815,
      "clip_ratio/low_min": 2.8280543119763024e-05,
      "clip_ratio/region_mean": 0.0015087180436239578,
      "epoch": 1.611256926217556,
      "grad_norm": 0.12021241337060928,
      "learning_rate": 1e-06,
      "loss": -0.0052,
      "step": 323
    },
    {
      "clip_ratio/high_max": 0.002288687217514962,
      "clip_ratio/high_mean": 0.0009081361768039642,
      "clip_ratio/low_mean": 0.0006733211685059359,
      "clip_ratio/low_min": 1.5439723938470706e-05,
      "clip_ratio/region_mean": 0.0015814573634997942,
      "epoch": 1.6159230096237969,
      "grad_norm": 0.1373334527015686,
      "learning_rate": 1e-06,
      "loss": 0.0179,
      "step": 324
    },
    {
      "clip_ratio/high_max": 0.0020306098740547895,
      "clip_ratio/high_mean": 0.000874390661920188,
      "clip_ratio/low_mean": 0.0006464388952736044,
      "clip_ratio/low_min": 2.6329883439757396e-05,
      "clip_ratio/region_mean": 0.0015208295590127818,
      "epoch": 1.620589093030038,
      "grad_norm": 0.12998969852924347,
      "learning_rate": 1e-06,
      "loss": 0.0039,
      "step": 325
    },
    {
      "clip_ratio/high_max": 0.0019063415165874176,
      "clip_ratio/high_mean": 0.000806250416644616,
      "clip_ratio/low_mean": 0.0007089508962963009,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015152013183978852,
      "epoch": 1.6252551764362788,
      "grad_norm": 0.11522424966096878,
      "learning_rate": 1e-06,
      "loss": -0.0083,
      "step": 326
    },
    {
      "clip_ratio/high_max": 0.002445299323881045,
      "clip_ratio/high_mean": 0.0009924290789058432,
      "clip_ratio/low_mean": 0.0007170979333750438,
      "clip_ratio/low_min": 1.678979242569767e-05,
      "clip_ratio/region_mean": 0.0017095269868150353,
      "epoch": 1.6299212598425197,
      "grad_norm": 0.12448902428150177,
      "learning_rate": 1e-06,
      "loss": -0.0258,
      "step": 327
    },
    {
      "clip_ratio/high_max": 0.0019845134956995025,
      "clip_ratio/high_mean": 0.0008472294539387804,
      "clip_ratio/low_mean": 0.0008545283617422683,
      "clip_ratio/low_min": 4.0416341107629705e-05,
      "clip_ratio/region_mean": 0.001701757857517805,
      "epoch": 1.6345873432487605,
      "grad_norm": 0.1182471439242363,
      "learning_rate": 1e-06,
      "loss": 0.0429,
      "step": 328
    },
    {
      "clip_ratio/high_max": 0.002270363645948237,
      "clip_ratio/high_mean": 0.001017747759760823,
      "clip_ratio/low_mean": 0.0008699028749106219,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001887650738353841,
      "epoch": 1.6392534266550016,
      "grad_norm": 0.1319698691368103,
      "learning_rate": 1e-06,
      "loss": 0.0047,
      "step": 329
    },
    {
      "clip_ratio/high_max": 0.0022671481056022458,
      "clip_ratio/high_mean": 0.0009886706284305546,
      "clip_ratio/low_mean": 0.0010321784575353377,
      "clip_ratio/low_min": 4.962312050338369e-05,
      "clip_ratio/region_mean": 0.0020208490968798287,
      "epoch": 1.6439195100612425,
      "grad_norm": 0.12157025188207626,
      "learning_rate": 1e-06,
      "loss": 0.0393,
      "step": 330
    },
    {
      "clip_ratio/high_max": 0.0026783751309267245,
      "clip_ratio/high_mean": 0.0010065884653158719,
      "clip_ratio/low_mean": 0.0008635749181848951,
      "clip_ratio/low_min": 6.690670670650434e-05,
      "clip_ratio/region_mean": 0.0018701634035096504,
      "epoch": 1.6485855934674833,
      "grad_norm": 0.11779691278934479,
      "learning_rate": 1e-06,
      "loss": -0.0096,
      "step": 331
    },
    {
      "clip_ratio/high_max": 0.002075159442028962,
      "clip_ratio/high_mean": 0.0009364311827084748,
      "clip_ratio/low_mean": 0.0010303963372280123,
      "clip_ratio/low_min": 5.960333510302007e-05,
      "clip_ratio/region_mean": 0.0019668275272124447,
      "epoch": 1.6532516768737242,
      "grad_norm": 0.1396941840648651,
      "learning_rate": 1e-06,
      "loss": -0.0202,
      "step": 332
    },
    {
      "clip_ratio/high_max": 0.0020541640915325843,
      "clip_ratio/high_mean": 0.0009431538928765804,
      "clip_ratio/low_mean": 0.0010075270402012393,
      "clip_ratio/low_min": 6.621025659114821e-05,
      "clip_ratio/region_mean": 0.001950680925801862,
      "epoch": 1.657917760279965,
      "grad_norm": 0.12700136005878448,
      "learning_rate": 1e-06,
      "loss": 0.0427,
      "step": 333
    },
    {
      "clip_ratio/high_max": 0.0023619765124749392,
      "clip_ratio/high_mean": 0.0010166736119572306,
      "clip_ratio/low_mean": 0.0008600363744335482,
      "clip_ratio/low_min": 6.924876288394444e-05,
      "clip_ratio/region_mean": 0.0018767099754768424,
      "epoch": 1.662583843686206,
      "grad_norm": 0.11084874719381332,
      "learning_rate": 1e-06,
      "loss": -0.0348,
      "step": 334
    },
    {
      "clip_ratio/high_max": 0.0026726905489340425,
      "clip_ratio/high_mean": 0.0010358676845498849,
      "clip_ratio/low_mean": 0.001002803946903441,
      "clip_ratio/low_min": 6.115995711297728e-05,
      "clip_ratio/region_mean": 0.0020386716059874743,
      "epoch": 1.6672499270924468,
      "grad_norm": 0.13239501416683197,
      "learning_rate": 1e-06,
      "loss": 0.0058,
      "step": 335
    },
    {
      "clip_ratio/high_max": 0.0021409119690360967,
      "clip_ratio/high_mean": 0.0009473096197325503,
      "clip_ratio/low_mean": 0.0007952069536258932,
      "clip_ratio/low_min": 6.538748129969463e-05,
      "clip_ratio/region_mean": 0.0017425165788154118,
      "epoch": 1.6719160104986877,
      "grad_norm": 0.11610430479049683,
      "learning_rate": 1e-06,
      "loss": -0.0289,
      "step": 336
    },
    {
      "clip_ratio/high_max": 0.0042418408338562585,
      "clip_ratio/high_mean": 0.0017963969439733773,
      "clip_ratio/low_mean": 0.0012557707959786057,
      "clip_ratio/low_min": 4.86381322843954e-05,
      "clip_ratio/region_mean": 0.0030521677326760255,
      "epoch": 1.6765820939049285,
      "grad_norm": 0.1103362962603569,
      "learning_rate": 1e-06,
      "loss": -0.0538,
      "step": 337
    },
    {
      "clip_ratio/high_max": 0.0030311374212033115,
      "clip_ratio/high_mean": 0.0012728021392831579,
      "clip_ratio/low_mean": 0.0012046853316860506,
      "clip_ratio/low_min": 5.457049519463908e-05,
      "clip_ratio/region_mean": 0.0024774875419097953,
      "epoch": 1.6812481773111694,
      "grad_norm": 0.10016685724258423,
      "learning_rate": 1e-06,
      "loss": 0.0202,
      "step": 338
    },
    {
      "clip_ratio/high_max": 0.003286378865595907,
      "clip_ratio/high_mean": 0.0014490114263026044,
      "clip_ratio/low_mean": 0.0013372733064898057,
      "clip_ratio/low_min": 2.015478821704164e-05,
      "clip_ratio/region_mean": 0.00278628476371523,
      "epoch": 1.6859142607174102,
      "grad_norm": 0.10625132918357849,
      "learning_rate": 1e-06,
      "loss": -0.0059,
      "step": 339
    },
    {
      "clip_ratio/high_max": 0.003658809815533459,
      "clip_ratio/high_mean": 0.0015765216412546579,
      "clip_ratio/low_mean": 0.0015578031161567196,
      "clip_ratio/low_min": 9.62185840762686e-05,
      "clip_ratio/region_mean": 0.003134324782877229,
      "epoch": 1.690580344123651,
      "grad_norm": 0.11801841109991074,
      "learning_rate": 1e-06,
      "loss": 0.0171,
      "step": 340
    },
    {
      "clip_ratio/high_max": 0.0037345462405937724,
      "clip_ratio/high_mean": 0.0014937598061806057,
      "clip_ratio/low_mean": 0.0013748585370194633,
      "clip_ratio/low_min": 0.0001134939411713276,
      "clip_ratio/region_mean": 0.0028686183504760265,
      "epoch": 1.695246427529892,
      "grad_norm": 0.11255044490098953,
      "learning_rate": 1e-06,
      "loss": 0.0032,
      "step": 341
    },
    {
      "clip_ratio/high_max": 0.0032759173554950394,
      "clip_ratio/high_mean": 0.001367050999760977,
      "clip_ratio/low_mean": 0.0015151435254665557,
      "clip_ratio/low_min": 0.00010566521359578473,
      "clip_ratio/region_mean": 0.002882194494304713,
      "epoch": 1.6999125109361328,
      "grad_norm": 0.10701259225606918,
      "learning_rate": 1e-06,
      "loss": -0.0089,
      "step": 342
    },
    {
      "clip_ratio/high_max": 0.0035532793699530885,
      "clip_ratio/high_mean": 0.001531706984678749,
      "clip_ratio/low_mean": 0.0013342510792426765,
      "clip_ratio/low_min": 7.329217260121368e-05,
      "clip_ratio/region_mean": 0.002865957976609934,
      "epoch": 1.704578594342374,
      "grad_norm": 0.1093229353427887,
      "learning_rate": 1e-06,
      "loss": -0.0264,
      "step": 343
    },
    {
      "clip_ratio/high_max": 0.0030399296010727994,
      "clip_ratio/high_mean": 0.0013797879546473268,
      "clip_ratio/low_mean": 0.0016490490961587057,
      "clip_ratio/low_min": 0.00018255741997563746,
      "clip_ratio/region_mean": 0.0030288370471680537,
      "epoch": 1.7092446777486148,
      "grad_norm": 0.10501661896705627,
      "learning_rate": 1e-06,
      "loss": 0.0422,
      "step": 344
    },
    {
      "clip_ratio/high_max": 0.004481605261389632,
      "clip_ratio/high_mean": 0.0016860480900504626,
      "clip_ratio/low_mean": 0.0015173658430285286,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00320341393671697,
      "epoch": 1.7139107611548556,
      "grad_norm": 0.113813616335392,
      "learning_rate": 1e-06,
      "loss": 0.0039,
      "step": 345
    },
    {
      "clip_ratio/high_max": 0.0035676207771757618,
      "clip_ratio/high_mean": 0.0014478334014711436,
      "clip_ratio/low_mean": 0.0016892831117729656,
      "clip_ratio/low_min": 2.860411950678099e-05,
      "clip_ratio/region_mean": 0.003137116495054215,
      "epoch": 1.7185768445610965,
      "grad_norm": 0.10499369353055954,
      "learning_rate": 1e-06,
      "loss": 0.0386,
      "step": 346
    },
    {
      "clip_ratio/high_max": 0.004160481141298078,
      "clip_ratio/high_mean": 0.0015307986832340248,
      "clip_ratio/low_mean": 0.0013833694138156716,
      "clip_ratio/low_min": 0.0001095561747206375,
      "clip_ratio/region_mean": 0.002914168122515548,
      "epoch": 1.7232429279673376,
      "grad_norm": 0.10324528068304062,
      "learning_rate": 1e-06,
      "loss": -0.0102,
      "step": 347
    },
    {
      "clip_ratio/high_max": 0.0032277135105687194,
      "clip_ratio/high_mean": 0.0014614682186220307,
      "clip_ratio/low_mean": 0.001711125503788935,
      "clip_ratio/low_min": 0.0001377127618980012,
      "clip_ratio/region_mean": 0.0031725938024464995,
      "epoch": 1.7279090113735784,
      "grad_norm": 0.12369127571582794,
      "learning_rate": 1e-06,
      "loss": -0.0211,
      "step": 348
    },
    {
      "clip_ratio/high_max": 0.003204410008038394,
      "clip_ratio/high_mean": 0.001322816104220692,
      "clip_ratio/low_mean": 0.001885768029751489,
      "clip_ratio/low_min": 0.00017562640277901664,
      "clip_ratio/region_mean": 0.003208584093954414,
      "epoch": 1.7325750947798193,
      "grad_norm": 0.11398528516292572,
      "learning_rate": 1e-06,
      "loss": 0.042,
      "step": 349
    },
    {
      "clip_ratio/high_max": 0.0035515979543561116,
      "clip_ratio/high_mean": 0.0014988111070124432,
      "clip_ratio/low_mean": 0.0013470172561937943,
      "clip_ratio/low_min": 0.0001781159771780949,
      "clip_ratio/region_mean": 0.002845828392310068,
      "epoch": 1.7372411781860602,
      "grad_norm": 0.09687419980764389,
      "learning_rate": 1e-06,
      "loss": -0.0354,
      "step": 350
    },
    {
      "clip_ratio/high_max": 0.003829060879070312,
      "clip_ratio/high_mean": 0.0015959006650518859,
      "clip_ratio/low_mean": 0.0016343101306119934,
      "clip_ratio/low_min": 0.0002144947229680838,
      "clip_ratio/region_mean": 0.0032302107429131866,
      "epoch": 1.741907261592301,
      "grad_norm": 0.11488457024097443,
      "learning_rate": 1e-06,
      "loss": 0.005,
      "step": 351
    },
    {
      "clip_ratio/high_max": 0.003590352505852934,
      "clip_ratio/high_mean": 0.0015548110677627847,
      "clip_ratio/low_mean": 0.0013329005905688973,
      "clip_ratio/low_min": 0.0001250026216439437,
      "clip_ratio/region_mean": 0.002887711612856947,
      "epoch": 1.7465733449985419,
      "grad_norm": 0.10290053486824036,
      "learning_rate": 1e-06,
      "loss": -0.0296,
      "step": 352
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.030622209821428603,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4075.0,
      "completions/mean_length": 682.173583984375,
      "completions/mean_terminated_length": 574.332275390625,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "epoch": 1.7512394284047827,
      "grad_norm": 0.12827764451503754,
      "learning_rate": 1e-06,
      "loss": 0.003,
      "num_tokens": 111769247.0,
      "reward": 0.5813337564468384,
      "reward_std": 0.19285298883914948,
      "rewards/simpleverify_reward/mean": 0.5813336968421936,
      "rewards/simpleverify_reward/std": 0.49335768818855286,
      "step": 353
    },
    {
      "clip_ratio/high_max": 0.0020785781234735623,
      "clip_ratio/high_mean": 0.0008503277485942817,
      "clip_ratio/low_mean": 0.000629880663836957,
      "clip_ratio/low_min": 3.462603854131885e-05,
      "clip_ratio/region_mean": 0.0014802084151597228,
      "epoch": 1.7559055118110236,
      "grad_norm": 0.1269812434911728,
      "learning_rate": 1e-06,
      "loss": -0.0128,
      "step": 354
    },
    {
      "clip_ratio/high_max": 0.0023443043100996874,
      "clip_ratio/high_mean": 0.0009015752002596855,
      "clip_ratio/low_mean": 0.000669885932438774,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015714611363364384,
      "epoch": 1.7605715952172645,
      "grad_norm": 0.12718863785266876,
      "learning_rate": 1e-06,
      "loss": -0.0011,
      "step": 355
    },
    {
      "clip_ratio/high_max": 0.0019307877255414496,
      "clip_ratio/high_mean": 0.0008263811923825415,
      "clip_ratio/low_mean": 0.000640767248114571,
      "clip_ratio/low_min": 8.741689634916838e-05,
      "clip_ratio/region_mean": 0.0014671484423161019,
      "epoch": 1.7652376786235053,
      "grad_norm": 0.15040023624897003,
      "learning_rate": 1e-06,
      "loss": 0.0313,
      "step": 356
    },
    {
      "clip_ratio/high_max": 0.0021755788366135675,
      "clip_ratio/high_mean": 0.0008778421488386812,
      "clip_ratio/low_mean": 0.0006345813553707558,
      "clip_ratio/low_min": 2.4923860109993257e-05,
      "clip_ratio/region_mean": 0.001512423506937921,
      "epoch": 1.7699037620297462,
      "grad_norm": 0.13119052350521088,
      "learning_rate": 1e-06,
      "loss": -0.0003,
      "step": 357
    },
    {
      "clip_ratio/high_max": 0.0018839223594113719,
      "clip_ratio/high_mean": 0.0007673159125261009,
      "clip_ratio/low_mean": 0.000752088275476126,
      "clip_ratio/low_min": 8.605176117271185e-05,
      "clip_ratio/region_mean": 0.001519404184364248,
      "epoch": 1.774569845435987,
      "grad_norm": 0.11358306556940079,
      "learning_rate": 1e-06,
      "loss": 0.0126,
      "step": 358
    },
    {
      "clip_ratio/high_max": 0.002341660609090468,
      "clip_ratio/high_mean": 0.0010169257893721806,
      "clip_ratio/low_mean": 0.0007849286957934964,
      "clip_ratio/low_min": 1.6715699530323036e-05,
      "clip_ratio/region_mean": 0.0018018544724327512,
      "epoch": 1.779235928842228,
      "grad_norm": 0.127577543258667,
      "learning_rate": 1e-06,
      "loss": -0.0076,
      "step": 359
    },
    {
      "clip_ratio/high_max": 0.002158691844670102,
      "clip_ratio/high_mean": 0.0010219817959296051,
      "clip_ratio/low_mean": 0.0007234496533783386,
      "clip_ratio/low_min": 7.750570421194425e-05,
      "clip_ratio/region_mean": 0.0017454314074711874,
      "epoch": 1.7839020122484688,
      "grad_norm": 0.13786529004573822,
      "learning_rate": 1e-06,
      "loss": -0.0454,
      "step": 360
    },
    {
      "clip_ratio/high_max": 0.0020796127355424687,
      "clip_ratio/high_mean": 0.0008258370071416721,
      "clip_ratio/low_mean": 0.00102595693715557,
      "clip_ratio/low_min": 5.177436196390772e-05,
      "clip_ratio/region_mean": 0.0018517939024604857,
      "epoch": 1.7885680956547099,
      "grad_norm": 0.12394144386053085,
      "learning_rate": 1e-06,
      "loss": 0.0436,
      "step": 361
    },
    {
      "clip_ratio/high_max": 0.002249546305392869,
      "clip_ratio/high_mean": 0.0010305317664460745,
      "clip_ratio/low_mean": 0.00095282848997158,
      "clip_ratio/low_min": 0.00011355648985045264,
      "clip_ratio/region_mean": 0.001983360263693612,
      "epoch": 1.7932341790609507,
      "grad_norm": 0.13357269763946533,
      "learning_rate": 1e-06,
      "loss": -0.006,
      "step": 362
    },
    {
      "clip_ratio/high_max": 0.0024230120252468623,
      "clip_ratio/high_mean": 0.0009460509118071059,
      "clip_ratio/low_mean": 0.0008383164094993845,
      "clip_ratio/low_min": 4.249613903084537e-05,
      "clip_ratio/region_mean": 0.0017843673049355857,
      "epoch": 1.7979002624671916,
      "grad_norm": 0.12783877551555634,
      "learning_rate": 1e-06,
      "loss": -0.0056,
      "step": 363
    },
    {
      "clip_ratio/high_max": 0.0024057246118900366,
      "clip_ratio/high_mean": 0.0010915157181443647,
      "clip_ratio/low_mean": 0.0009038983935170108,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001995414146222174,
      "epoch": 1.8025663458734325,
      "grad_norm": 0.12972626090049744,
      "learning_rate": 1e-06,
      "loss": 0.0025,
      "step": 364
    },
    {
      "clip_ratio/high_max": 0.002393270769971423,
      "clip_ratio/high_mean": 0.0011562036888790317,
      "clip_ratio/low_mean": 0.0008253280907410954,
      "clip_ratio/low_min": 7.776207621645881e-05,
      "clip_ratio/region_mean": 0.0019815317864413373,
      "epoch": 1.8072324292796735,
      "grad_norm": 0.12222950160503387,
      "learning_rate": 1e-06,
      "loss": -0.0397,
      "step": 365
    },
    {
      "clip_ratio/high_max": 0.0026781432788993698,
      "clip_ratio/high_mean": 0.0011825664332718588,
      "clip_ratio/low_mean": 0.0006742381956428289,
      "clip_ratio/low_min": 1.6887328456505202e-05,
      "clip_ratio/region_mean": 0.0018568046471045818,
      "epoch": 1.8118985126859144,
      "grad_norm": 0.12620043754577637,
      "learning_rate": 1e-06,
      "loss": -0.063,
      "step": 366
    },
    {
      "clip_ratio/high_max": 0.002211867700680159,
      "clip_ratio/high_mean": 0.0008838869034661911,
      "clip_ratio/low_mean": 0.0009511059870419558,
      "clip_ratio/low_min": 6.131609370640945e-05,
      "clip_ratio/region_mean": 0.0018349928941461258,
      "epoch": 1.8165645960921553,
      "grad_norm": 0.11981648206710815,
      "learning_rate": 1e-06,
      "loss": 0.0162,
      "step": 367
    },
    {
      "clip_ratio/high_max": 0.0022732884463039227,
      "clip_ratio/high_mean": 0.000955751740548294,
      "clip_ratio/low_mean": 0.0010068793890241068,
      "clip_ratio/low_min": 6.87444771756418e-05,
      "clip_ratio/region_mean": 0.001962631147762295,
      "epoch": 1.8212306794983961,
      "grad_norm": 0.13691291213035583,
      "learning_rate": 1e-06,
      "loss": 0.0307,
      "step": 368
    },
    {
      "clip_ratio/high_max": 0.0033569572915439494,
      "clip_ratio/high_mean": 0.0015392535206046887,
      "clip_ratio/low_mean": 0.0015115887472347822,
      "clip_ratio/low_min": 8.52354169182945e-05,
      "clip_ratio/region_mean": 0.003050842191441916,
      "epoch": 1.825896762904637,
      "grad_norm": 0.10022418200969696,
      "learning_rate": 1e-06,
      "loss": 0.0023,
      "step": 369
    },
    {
      "clip_ratio/high_max": 0.004083825915586203,
      "clip_ratio/high_mean": 0.0015683019264542963,
      "clip_ratio/low_mean": 0.001630250411835732,
      "clip_ratio/low_min": 0.00016472469724249095,
      "clip_ratio/region_mean": 0.0031985523310140707,
      "epoch": 1.8305628463108778,
      "grad_norm": 0.11440886557102203,
      "learning_rate": 1e-06,
      "loss": -0.0136,
      "step": 370
    },
    {
      "clip_ratio/high_max": 0.0034999630006495863,
      "clip_ratio/high_mean": 0.0014020929484104272,
      "clip_ratio/low_mean": 0.0016902677052712534,
      "clip_ratio/low_min": 0.00014178428682498634,
      "clip_ratio/region_mean": 0.0030923606973374262,
      "epoch": 1.8352289297171187,
      "grad_norm": 0.11437682807445526,
      "learning_rate": 1e-06,
      "loss": -0.0018,
      "step": 371
    },
    {
      "clip_ratio/high_max": 0.0035921438175137155,
      "clip_ratio/high_mean": 0.0014920344583515543,
      "clip_ratio/low_mean": 0.001933707964781206,
      "clip_ratio/low_min": 0.0002751626125245821,
      "clip_ratio/region_mean": 0.0034257424267707393,
      "epoch": 1.8398950131233596,
      "grad_norm": 0.11441878974437714,
      "learning_rate": 1e-06,
      "loss": 0.0305,
      "step": 372
    },
    {
      "clip_ratio/high_max": 0.004091057162440848,
      "clip_ratio/high_mean": 0.0015690306317992508,
      "clip_ratio/low_mean": 0.0017073304352379637,
      "clip_ratio/low_min": 0.00025986087166529614,
      "clip_ratio/region_mean": 0.0032763610943220556,
      "epoch": 1.8445610965296004,
      "grad_norm": 0.11429788917303085,
      "learning_rate": 1e-06,
      "loss": -0.0012,
      "step": 373
    },
    {
      "clip_ratio/high_max": 0.003304840764030814,
      "clip_ratio/high_mean": 0.001242177508174791,
      "clip_ratio/low_mean": 0.001681275425653439,
      "clip_ratio/low_min": 0.00023983259234228171,
      "clip_ratio/region_mean": 0.0029234528919914737,
      "epoch": 1.8492271799358413,
      "grad_norm": 0.10144300013780594,
      "learning_rate": 1e-06,
      "loss": 0.012,
      "step": 374
    },
    {
      "clip_ratio/high_max": 0.003399322580662556,
      "clip_ratio/high_mean": 0.0015267052403942216,
      "clip_ratio/low_mean": 0.0016492387585458346,
      "clip_ratio/low_min": 7.255605669342913e-05,
      "clip_ratio/region_mean": 0.003175943944370374,
      "epoch": 1.8538932633420822,
      "grad_norm": 0.11459530889987946,
      "learning_rate": 1e-06,
      "loss": -0.0083,
      "step": 375
    },
    {
      "clip_ratio/high_max": 0.004202735843136907,
      "clip_ratio/high_mean": 0.0017986674974963535,
      "clip_ratio/low_mean": 0.0014887752040522173,
      "clip_ratio/low_min": 0.00024920992291299626,
      "clip_ratio/region_mean": 0.003287442756118253,
      "epoch": 1.858559346748323,
      "grad_norm": 0.11719462275505066,
      "learning_rate": 1e-06,
      "loss": -0.0462,
      "step": 376
    },
    {
      "clip_ratio/high_max": 0.003653102699900046,
      "clip_ratio/high_mean": 0.0014300389811978675,
      "clip_ratio/low_mean": 0.0019336772202223074,
      "clip_ratio/low_min": 0.00012109909857827006,
      "clip_ratio/region_mean": 0.0033637161686783656,
      "epoch": 1.8632254301545639,
      "grad_norm": 0.11154165863990784,
      "learning_rate": 1e-06,
      "loss": 0.0428,
      "step": 377
    },
    {
      "clip_ratio/high_max": 0.0038883319502929226,
      "clip_ratio/high_mean": 0.0016257368370133918,
      "clip_ratio/low_mean": 0.0017939439967449289,
      "clip_ratio/low_min": 0.0002152343495254172,
      "clip_ratio/region_mean": 0.003419680826482363,
      "epoch": 1.8678915135608047,
      "grad_norm": 0.10971345752477646,
      "learning_rate": 1e-06,
      "loss": -0.0068,
      "step": 378
    },
    {
      "clip_ratio/high_max": 0.0038878767081769183,
      "clip_ratio/high_mean": 0.0016080887471616734,
      "clip_ratio/low_mean": 0.0015735847664473113,
      "clip_ratio/low_min": 5.1741701099672355e-05,
      "clip_ratio/region_mean": 0.003181673411745578,
      "epoch": 1.8725575969670458,
      "grad_norm": 0.11119736731052399,
      "learning_rate": 1e-06,
      "loss": -0.0064,
      "step": 379
    },
    {
      "clip_ratio/high_max": 0.003721888417203445,
      "clip_ratio/high_mean": 0.001680263831076445,
      "clip_ratio/low_mean": 0.0016717045691621024,
      "clip_ratio/low_min": 6.486692564067198e-05,
      "clip_ratio/region_mean": 0.003351968312927056,
      "epoch": 1.8772236803732867,
      "grad_norm": 0.11024277657270432,
      "learning_rate": 1e-06,
      "loss": 0.0017,
      "step": 380
    },
    {
      "clip_ratio/high_max": 0.003542009784723632,
      "clip_ratio/high_mean": 0.0017832112353062257,
      "clip_ratio/low_mean": 0.0013568868071160978,
      "clip_ratio/low_min": 0.0001835613584262319,
      "clip_ratio/region_mean": 0.003140098044241313,
      "epoch": 1.8818897637795275,
      "grad_norm": 0.10931335389614105,
      "learning_rate": 1e-06,
      "loss": -0.0404,
      "step": 381
    },
    {
      "clip_ratio/high_max": 0.004412916503497399,
      "clip_ratio/high_mean": 0.001886008605652023,
      "clip_ratio/low_mean": 0.0012965114874532446,
      "clip_ratio/low_min": 1.6887328456505202e-05,
      "clip_ratio/region_mean": 0.0031825201731408015,
      "epoch": 1.8865558471857684,
      "grad_norm": 0.11360908299684525,
      "learning_rate": 1e-06,
      "loss": -0.0637,
      "step": 382
    },
    {
      "clip_ratio/high_max": 0.003960326328524388,
      "clip_ratio/high_mean": 0.0015011436080385465,
      "clip_ratio/low_mean": 0.0014402225497178733,
      "clip_ratio/low_min": 0.00015322612216550624,
      "clip_ratio/region_mean": 0.002941366139566526,
      "epoch": 1.8912219305920095,
      "grad_norm": 0.10976927727460861,
      "learning_rate": 1e-06,
      "loss": 0.0155,
      "step": 383
    },
    {
      "clip_ratio/high_max": 0.003886464415700175,
      "clip_ratio/high_mean": 0.0016368966280424502,
      "clip_ratio/low_mean": 0.0016564458092034329,
      "clip_ratio/low_min": 0.00012087353024980985,
      "clip_ratio/region_mean": 0.0032933423717622645,
      "epoch": 1.8958880139982504,
      "grad_norm": 0.11942066252231598,
      "learning_rate": 1e-06,
      "loss": 0.0299,
      "step": 384
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.030552455357142905,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4095.0,
      "completions/mean_length": 685.3372192382812,
      "completions/mean_terminated_length": 577.8489990234375,
      "completions/min_length": 78.0,
      "completions/min_terminated_length": 78.0,
      "epoch": 2.004666083406241,
      "grad_norm": 0.1330665796995163,
      "learning_rate": 1e-06,
      "loss": -0.0066,
      "num_tokens": 121253577.0,
      "reward": 0.6083984375,
      "reward_std": 0.18795929849147797,
      "rewards/simpleverify_reward/mean": 0.6083984375,
      "rewards/simpleverify_reward/std": 0.48812538385391235,
      "step": 385
    },
    {
      "clip_ratio/high_max": 0.002078529607388191,
      "clip_ratio/high_mean": 0.000927934241190087,
      "clip_ratio/low_mean": 0.0004505121960391989,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001378446446324233,
      "epoch": 2.0093321668124817,
      "grad_norm": 0.12242374569177628,
      "learning_rate": 1e-06,
      "loss": -0.0558,
      "step": 386
    },
    {
      "clip_ratio/high_max": 0.002174368783016689,
      "clip_ratio/high_mean": 0.0009199488122249022,
      "clip_ratio/low_mean": 0.0005688992951036198,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014888480873196386,
      "epoch": 2.0139982502187226,
      "grad_norm": 0.13828489184379578,
      "learning_rate": 1e-06,
      "loss": 0.0362,
      "step": 387
    },
    {
      "clip_ratio/high_max": 0.002613817858218681,
      "clip_ratio/high_mean": 0.0010798559596878476,
      "clip_ratio/low_mean": 0.0005303921925587929,
      "clip_ratio/low_min": 3.116131301794667e-05,
      "clip_ratio/region_mean": 0.0016102481422421988,
      "epoch": 2.0186643336249634,
      "grad_norm": 0.12653018534183502,
      "learning_rate": 1e-06,
      "loss": -0.0728,
      "step": 388
    },
    {
      "clip_ratio/high_max": 0.0018697531195357442,
      "clip_ratio/high_mean": 0.0007625891303177923,
      "clip_ratio/low_mean": 0.0006168238396639936,
      "clip_ratio/low_min": 3.9324662793660536e-05,
      "clip_ratio/region_mean": 0.0013794129736197647,
      "epoch": 2.0233304170312043,
      "grad_norm": 0.11558361351490021,
      "learning_rate": 1e-06,
      "loss": -0.025,
      "step": 389
    },
    {
      "clip_ratio/high_max": 0.0020666335185524076,
      "clip_ratio/high_mean": 0.0008618993560958188,
      "clip_ratio/low_mean": 0.0007046818591334159,
      "clip_ratio/low_min": 2.6307861844543368e-05,
      "clip_ratio/region_mean": 0.001566581235238118,
      "epoch": 2.027996500437445,
      "grad_norm": 0.13028889894485474,
      "learning_rate": 1e-06,
      "loss": 0.012,
      "step": 390
    },
    {
      "clip_ratio/high_max": 0.002014424455410335,
      "clip_ratio/high_mean": 0.0008683326450409368,
      "clip_ratio/low_mean": 0.000848259442136623,
      "clip_ratio/low_min": 6.993009083089419e-05,
      "clip_ratio/region_mean": 0.0017165920871775597,
      "epoch": 2.032662583843686,
      "grad_norm": 0.17612363398075104,
      "learning_rate": 1e-06,
      "loss": 0.0136,
      "step": 391
    },
    {
      "clip_ratio/high_max": 0.0024020767777983565,
      "clip_ratio/high_mean": 0.0009875263131107204,
      "clip_ratio/low_mean": 0.0008998762186820386,
      "clip_ratio/low_min": 6.384871176123852e-05,
      "clip_ratio/region_mean": 0.0018874025117838755,
      "epoch": 2.037328667249927,
      "grad_norm": 0.12755659222602844,
      "learning_rate": 1e-06,
      "loss": -0.0212,
      "step": 392
    },
    {
      "clip_ratio/high_max": 0.0024498723360011354,
      "clip_ratio/high_mean": 0.0010138337456737645,
      "clip_ratio/low_mean": 0.0008478861273033544,
      "clip_ratio/low_min": 1.234689807461109e-05,
      "clip_ratio/region_mean": 0.0018617198584252037,
      "epoch": 2.041994750656168,
      "grad_norm": 0.11342108249664307,
      "learning_rate": 1e-06,
      "loss": 0.0033,
      "step": 393
    },
    {
      "clip_ratio/high_max": 0.0023018126012175344,
      "clip_ratio/high_mean": 0.0009311229514423758,
      "clip_ratio/low_mean": 0.000948539696764783,
      "clip_ratio/low_min": 8.847866683936445e-05,
      "clip_ratio/region_mean": 0.001879662690043915,
      "epoch": 2.046660834062409,
      "grad_norm": 0.12213778495788574,
      "learning_rate": 1e-06,
      "loss": -0.0179,
      "step": 394
    },
    {
      "clip_ratio/high_max": 0.002534204766561743,
      "clip_ratio/high_mean": 0.0010039549215434818,
      "clip_ratio/low_mean": 0.0009770175674930215,
      "clip_ratio/low_min": 2.0394843886606395e-05,
      "clip_ratio/region_mean": 0.0019809725126833655,
      "epoch": 2.05132691746865,
      "grad_norm": 0.13308873772621155,
      "learning_rate": 1e-06,
      "loss": 0.0124,
      "step": 395
    },
    {
      "clip_ratio/high_max": 0.002023231289058458,
      "clip_ratio/high_mean": 0.0008096847777778748,
      "clip_ratio/low_mean": 0.0011474332623038208,
      "clip_ratio/low_min": 0.00014910923164279666,
      "clip_ratio/region_mean": 0.0019571179946069606,
      "epoch": 2.055993000874891,
      "grad_norm": 0.13124310970306396,
      "learning_rate": 1e-06,
      "loss": 0.0595,
      "step": 396
    },
    {
      "clip_ratio/high_max": 0.002735275571467355,
      "clip_ratio/high_mean": 0.001185059987619752,
      "clip_ratio/low_mean": 0.0009641533688409254,
      "clip_ratio/low_min": 7.641969750693534e-05,
      "clip_ratio/region_mean": 0.002149213381926529,
      "epoch": 2.0606590842811316,
      "grad_norm": 0.12203901261091232,
      "learning_rate": 1e-06,
      "loss": -0.0026,
      "step": 397
    },
    {
      "clip_ratio/high_max": 0.0024362768526771106,
      "clip_ratio/high_mean": 0.0009537299538351363,
      "clip_ratio/low_mean": 0.0010708052013796987,
      "clip_ratio/low_min": 0.0001378891374770319,
      "clip_ratio/region_mean": 0.002024535191594623,
      "epoch": 2.0653251676873725,
      "grad_norm": 0.1257878690958023,
      "learning_rate": 1e-06,
      "loss": 0.0063,
      "step": 398
    },
    {
      "clip_ratio/high_max": 0.002449552484904416,
      "clip_ratio/high_mean": 0.0011017015058314428,
      "clip_ratio/low_mean": 0.0010120234292116947,
      "clip_ratio/low_min": 4.369905400380958e-05,
      "clip_ratio/region_mean": 0.0021137249132152647,
      "epoch": 2.0699912510936134,
      "grad_norm": 0.1251668483018875,
      "learning_rate": 1e-06,
      "loss": -0.0198,
      "step": 399
    },
    {
      "clip_ratio/high_max": 0.0021176970039959997,
      "clip_ratio/high_mean": 0.0010002863800764317,
      "clip_ratio/low_mean": 0.001083603136066813,
      "clip_ratio/low_min": 0.0001312460190092679,
      "clip_ratio/region_mean": 0.002083889441564679,
      "epoch": 2.0746573344998542,
      "grad_norm": 0.12752240896224976,
      "learning_rate": 1e-06,
      "loss": 0.0162,
      "step": 400
    },
    {
      "clip_ratio/high_max": 0.003637856731074862,
      "clip_ratio/high_mean": 0.0014870851700834464,
      "clip_ratio/low_mean": 0.0015769202000228688,
      "clip_ratio/low_min": 8.633852485218085e-05,
      "clip_ratio/region_mean": 0.003064005504711531,
      "epoch": 2.079323417906095,
      "grad_norm": 0.10936375707387924,
      "learning_rate": 1e-06,
      "loss": -0.0073,
      "step": 401
    },
    {
      "clip_ratio/high_max": 0.0035837859904859215,
      "clip_ratio/high_mean": 0.0015862231666687876,
      "clip_ratio/low_mean": 0.001181528055894887,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0027677513062371872,
      "epoch": 2.083989501312336,
      "grad_norm": 0.10769211500883102,
      "learning_rate": 1e-06,
      "loss": -0.0565,
      "step": 402
    },
    {
      "clip_ratio/high_max": 0.003620268478698563,
      "clip_ratio/high_mean": 0.0015115561182028614,
      "clip_ratio/low_mean": 0.001711774202703964,
      "clip_ratio/low_min": 5.112231519888155e-05,
      "clip_ratio/region_mean": 0.0032233303136308677,
      "epoch": 2.088655584718577,
      "grad_norm": 0.11903084069490433,
      "learning_rate": 1e-06,
      "loss": 0.0354,
      "step": 403
    },
    {
      "clip_ratio/high_max": 0.004328246366640087,
      "clip_ratio/high_mean": 0.001920813763717888,
      "clip_ratio/low_mean": 0.0013620666031783912,
      "clip_ratio/low_min": 0.0002128134437953122,
      "clip_ratio/region_mean": 0.003282880366896279,
      "epoch": 2.0933216681248177,
      "grad_norm": 0.1078646183013916,
      "learning_rate": 1e-06,
      "loss": -0.0736,
      "step": 404
    },
    {
      "clip_ratio/high_max": 0.003231616494304035,
      "clip_ratio/high_mean": 0.0014701861873618327,
      "clip_ratio/low_mean": 0.001407937397743808,
      "clip_ratio/low_min": 5.243288615019992e-05,
      "clip_ratio/region_mean": 0.0028781236105714925,
      "epoch": 2.0979877515310585,
      "grad_norm": 0.10333307832479477,
      "learning_rate": 1e-06,
      "loss": -0.0256,
      "step": 405
    },
    {
      "clip_ratio/high_max": 0.003635908695287071,
      "clip_ratio/high_mean": 0.0015776671170897316,
      "clip_ratio/low_mean": 0.001695445302175358,
      "clip_ratio/low_min": 0.0001422545810783049,
      "clip_ratio/region_mean": 0.0032731124229030684,
      "epoch": 2.1026538349372994,
      "grad_norm": 0.11389170587062836,
      "learning_rate": 1e-06,
      "loss": 0.0112,
      "step": 406
    },
    {
      "clip_ratio/high_max": 0.0031745763481012546,
      "clip_ratio/high_mean": 0.0013398703995335381,
      "clip_ratio/low_mean": 0.0016521694087714422,
      "clip_ratio/low_min": 0.00010802012184285559,
      "clip_ratio/region_mean": 0.0029920397282694466,
      "epoch": 2.1073199183435403,
      "grad_norm": 0.10970176011323929,
      "learning_rate": 1e-06,
      "loss": 0.0129,
      "step": 407
    },
    {
      "clip_ratio/high_max": 0.004052613636304159,
      "clip_ratio/high_mean": 0.0017313632743025664,
      "clip_ratio/low_mean": 0.0016591520652582403,
      "clip_ratio/low_min": 0.00019382750360819045,
      "clip_ratio/region_mean": 0.0033905152376974,
      "epoch": 2.111986001749781,
      "grad_norm": 0.11538665741682053,
      "learning_rate": 1e-06,
      "loss": -0.0219,
      "step": 408
    },
    {
      "clip_ratio/high_max": 0.0037640930895577185,
      "clip_ratio/high_mean": 0.0015785114192112815,
      "clip_ratio/low_mean": 0.0015149501123232767,
      "clip_ratio/low_min": 6.529494748974685e-05,
      "clip_ratio/region_mean": 0.00309346155700041,
      "epoch": 2.116652085156022,
      "grad_norm": 0.09933169931173325,
      "learning_rate": 1e-06,
      "loss": 0.0026,
      "step": 409
    },
    {
      "clip_ratio/high_max": 0.0036888261674903333,
      "clip_ratio/high_mean": 0.0015189411169558298,
      "clip_ratio/low_mean": 0.0016641882066323888,
      "clip_ratio/low_min": 0.0001844062535383273,
      "clip_ratio/region_mean": 0.003183129330864176,
      "epoch": 2.121318168562263,
      "grad_norm": 0.10626459866762161,
      "learning_rate": 1e-06,
      "loss": -0.0186,
      "step": 410
    },
    {
      "clip_ratio/high_max": 0.0037286237566149794,
      "clip_ratio/high_mean": 0.0016477864810440224,
      "clip_ratio/low_mean": 0.0017910263013618533,
      "clip_ratio/low_min": 0.00012334531857050024,
      "clip_ratio/region_mean": 0.0034388128260616213,
      "epoch": 2.1259842519685037,
      "grad_norm": 0.12006929516792297,
      "learning_rate": 1e-06,
      "loss": 0.0116,
      "step": 411
    },
    {
      "clip_ratio/high_max": 0.003761015868803952,
      "clip_ratio/high_mean": 0.0014651818901256775,
      "clip_ratio/low_mean": 0.002208865902503021,
      "clip_ratio/low_min": 0.00035455115539662074,
      "clip_ratio/region_mean": 0.003674047715321649,
      "epoch": 2.130650335374745,
      "grad_norm": 0.11688859760761261,
      "learning_rate": 1e-06,
      "loss": 0.0587,
      "step": 412
    },
    {
      "clip_ratio/high_max": 0.0043728131568059325,
      "clip_ratio/high_mean": 0.0017421028860553633,
      "clip_ratio/low_mean": 0.0016818589210743085,
      "clip_ratio/low_min": 0.00010463003673066851,
      "clip_ratio/region_mean": 0.0034239617962157354,
      "epoch": 2.135316418780986,
      "grad_norm": 0.10648229718208313,
      "learning_rate": 1e-06,
      "loss": -0.0034,
      "step": 413
    },
    {
      "clip_ratio/high_max": 0.003689129662234336,
      "clip_ratio/high_mean": 0.0015235339596983977,
      "clip_ratio/low_mean": 0.0017985088379646186,
      "clip_ratio/low_min": 0.00022967340191826224,
      "clip_ratio/region_mean": 0.003322042743093334,
      "epoch": 2.1399825021872267,
      "grad_norm": 0.11225876957178116,
      "learning_rate": 1e-06,
      "loss": 0.0055,
      "step": 414
    },
    {
      "clip_ratio/high_max": 0.003691285601234995,
      "clip_ratio/high_mean": 0.0017477754154242575,
      "clip_ratio/low_mean": 0.0017051547329174355,
      "clip_ratio/low_min": 0.00014654442566097714,
      "clip_ratio/region_mean": 0.003452930206549354,
      "epoch": 2.1446485855934676,
      "grad_norm": 0.11444263905286789,
      "learning_rate": 1e-06,
      "loss": -0.0206,
      "step": 415
    },
    {
      "clip_ratio/high_max": 0.003370915168488864,
      "clip_ratio/high_mean": 0.0015085931081557646,
      "clip_ratio/low_mean": 0.001830224398872815,
      "clip_ratio/low_min": 0.00018205205469712382,
      "clip_ratio/region_mean": 0.003338817463372834,
      "epoch": 2.1493146689997085,
      "grad_norm": 0.10421435534954071,
      "learning_rate": 1e-06,
      "loss": 0.0154,
      "step": 416
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.032017299107142905,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4065.0,
      "completions/mean_length": 671.1072387695312,
      "completions/mean_terminated_length": 557.8244018554688,
      "completions/min_length": 90.0,
      "completions/min_terminated_length": 90.0,
      "epoch": 2.1539807524059493,
      "grad_norm": 0.12873737514019012,
      "learning_rate": 1e-06,
      "loss": 0.0034,
      "num_tokens": 130388770.0,
      "reward": 0.6077009439468384,
      "reward_std": 0.18060724437236786,
      "rewards/simpleverify_reward/mean": 0.6077008843421936,
      "rewards/simpleverify_reward/std": 0.488279789686203,
      "step": 417
    },
    {
      "clip_ratio/high_max": 0.0021688674460165203,
      "clip_ratio/high_mean": 0.0008151452038873686,
      "clip_ratio/low_mean": 0.0005665232865794678,
      "clip_ratio/low_min": 3.8096014577604365e-05,
      "clip_ratio/region_mean": 0.0013816685095662251,
      "epoch": 2.15864683581219,
      "grad_norm": 0.12198550999164581,
      "learning_rate": 1e-06,
      "loss": 0.0073,
      "step": 418
    },
    {
      "clip_ratio/high_max": 0.0020757316742674448,
      "clip_ratio/high_mean": 0.0009019742974487599,
      "clip_ratio/low_mean": 0.0005356578240025556,
      "clip_ratio/low_min": 1.7472742911195382e-05,
      "clip_ratio/region_mean": 0.0014376321269082837,
      "epoch": 2.163312919218431,
      "grad_norm": 0.13432927429676056,
      "learning_rate": 1e-06,
      "loss": 0.0151,
      "step": 419
    },
    {
      "clip_ratio/high_max": 0.002020560459641274,
      "clip_ratio/high_mean": 0.0008356838707186398,
      "clip_ratio/low_mean": 0.0005019247291784268,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013376086026255507,
      "epoch": 2.167979002624672,
      "grad_norm": 0.12839467823505402,
      "learning_rate": 1e-06,
      "loss": -0.0178,
      "step": 420
    },
    {
      "clip_ratio/high_max": 0.002041657640802441,
      "clip_ratio/high_mean": 0.0008013724054762861,
      "clip_ratio/low_mean": 0.0005724491579712776,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013738215748162474,
      "epoch": 2.1726450860309128,
      "grad_norm": 0.12452315539121628,
      "learning_rate": 1e-06,
      "loss": 0.0114,
      "step": 421
    },
    {
      "clip_ratio/high_max": 0.0018366385593253653,
      "clip_ratio/high_mean": 0.0008400008809985593,
      "clip_ratio/low_mean": 0.0007201819389592856,
      "clip_ratio/low_min": 3.400094374228502e-05,
      "clip_ratio/region_mean": 0.0015601828599756118,
      "epoch": 2.1773111694371536,
      "grad_norm": 0.13958384096622467,
      "learning_rate": 1e-06,
      "loss": 0.0153,
      "step": 422
    },
    {
      "clip_ratio/high_max": 0.002015688398387283,
      "clip_ratio/high_mean": 0.000833709207654465,
      "clip_ratio/low_mean": 0.0006359328890539473,
      "clip_ratio/low_min": 2.9350784643611405e-05,
      "clip_ratio/region_mean": 0.0014696421058033593,
      "epoch": 2.1819772528433945,
      "grad_norm": 0.14534762501716614,
      "learning_rate": 1e-06,
      "loss": 0.0098,
      "step": 423
    },
    {
      "clip_ratio/high_max": 0.0017657078024058137,
      "clip_ratio/high_mean": 0.0007911612119642086,
      "clip_ratio/low_mean": 0.000757917701776023,
      "clip_ratio/low_min": 0.00010911672507063486,
      "clip_ratio/region_mean": 0.0015490789191971999,
      "epoch": 2.1866433362496354,
      "grad_norm": 0.13252082467079163,
      "learning_rate": 1e-06,
      "loss": 0.0262,
      "step": 424
    },
    {
      "clip_ratio/high_max": 0.0022240313519432675,
      "clip_ratio/high_mean": 0.0009690386777947424,
      "clip_ratio/low_mean": 0.0006769015526515432,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001645940250455169,
      "epoch": 2.1913094196558762,
      "grad_norm": 0.1220291256904602,
      "learning_rate": 1e-06,
      "loss": -0.0257,
      "step": 425
    },
    {
      "clip_ratio/high_max": 0.002485303550201934,
      "clip_ratio/high_mean": 0.0009092185991903534,
      "clip_ratio/low_mean": 0.0007724371753283776,
      "clip_ratio/low_min": 1.3691128515347373e-05,
      "clip_ratio/region_mean": 0.0016816557617858052,
      "epoch": 2.195975503062117,
      "grad_norm": 0.12133394181728363,
      "learning_rate": 1e-06,
      "loss": -0.0181,
      "step": 426
    },
    {
      "clip_ratio/high_max": 0.0024866541862138547,
      "clip_ratio/high_mean": 0.00095881270863174,
      "clip_ratio/low_mean": 0.000666898839881469,
      "clip_ratio/low_min": 3.0958125535107683e-05,
      "clip_ratio/region_mean": 0.0016257115348707885,
      "epoch": 2.200641586468358,
      "grad_norm": 0.12553566694259644,
      "learning_rate": 1e-06,
      "loss": -0.0343,
      "step": 427
    },
    {
      "clip_ratio/high_max": 0.002327847312699305,
      "clip_ratio/high_mean": 0.0009998264904425014,
      "clip_ratio/low_mean": 0.0006324681171463453,
      "clip_ratio/low_min": 2.616725214465987e-05,
      "clip_ratio/region_mean": 0.001632294624869246,
      "epoch": 2.205307669874599,
      "grad_norm": 0.13535764813423157,
      "learning_rate": 1e-06,
      "loss": -0.0511,
      "step": 428
    },
    {
      "clip_ratio/high_max": 0.0025342936569359154,
      "clip_ratio/high_mean": 0.0010408837733848486,
      "clip_ratio/low_mean": 0.0008872068847267656,
      "clip_ratio/low_min": 0.00011486966286611278,
      "clip_ratio/region_mean": 0.0019280906490166672,
      "epoch": 2.20997375328084,
      "grad_norm": 0.13542316854000092,
      "learning_rate": 1e-06,
      "loss": -0.0035,
      "step": 429
    },
    {
      "clip_ratio/high_max": 0.002540831257647369,
      "clip_ratio/high_mean": 0.00099457749092835,
      "clip_ratio/low_mean": 0.000833474343380658,
      "clip_ratio/low_min": 8.961288676800905e-05,
      "clip_ratio/region_mean": 0.0018280518415849656,
      "epoch": 2.214639836687081,
      "grad_norm": 0.1258094757795334,
      "learning_rate": 1e-06,
      "loss": -0.0245,
      "step": 430
    },
    {
      "clip_ratio/high_max": 0.0022217575897229835,
      "clip_ratio/high_mean": 0.0008792716944299173,
      "clip_ratio/low_mean": 0.00093852799909655,
      "clip_ratio/low_min": 6.621166448894655e-05,
      "clip_ratio/region_mean": 0.0018177997117163613,
      "epoch": 2.219305920093322,
      "grad_norm": 0.1328386515378952,
      "learning_rate": 1e-06,
      "loss": -0.0048,
      "step": 431
    },
    {
      "clip_ratio/high_max": 0.0024301523517351598,
      "clip_ratio/high_mean": 0.00101914665356162,
      "clip_ratio/low_mean": 0.0010499246054678224,
      "clip_ratio/low_min": 9.598919223208213e-05,
      "clip_ratio/region_mean": 0.002069071306323167,
      "epoch": 2.2239720034995627,
      "grad_norm": 0.14714941382408142,
      "learning_rate": 1e-06,
      "loss": -0.0078,
      "step": 432
    },
    {
      "clip_ratio/high_max": 0.004045515117468312,
      "clip_ratio/high_mean": 0.0014495478826574981,
      "clip_ratio/low_mean": 0.0014891990686010104,
      "clip_ratio/low_min": 4.9229449359700084e-05,
      "clip_ratio/region_mean": 0.0029387469112407416,
      "epoch": 2.2286380869058036,
      "grad_norm": 0.1016538143157959,
      "learning_rate": 1e-06,
      "loss": 0.0027,
      "step": 433
    },
    {
      "clip_ratio/high_max": 0.004073299947776832,
      "clip_ratio/high_mean": 0.0015483728384424467,
      "clip_ratio/low_mean": 0.0016881891424418427,
      "clip_ratio/low_min": 0.0001257203439308796,
      "clip_ratio/region_mean": 0.0032365619990741834,
      "epoch": 2.2333041703120444,
      "grad_norm": 0.10631687194108963,
      "learning_rate": 1e-06,
      "loss": 0.0066,
      "step": 434
    },
    {
      "clip_ratio/high_max": 0.00381794516579248,
      "clip_ratio/high_mean": 0.0015347961962106638,
      "clip_ratio/low_mean": 0.0018125336719094776,
      "clip_ratio/low_min": 0.00010464232400408946,
      "clip_ratio/region_mean": 0.0033473298681201413,
      "epoch": 2.2379702537182853,
      "grad_norm": 0.1090458333492279,
      "learning_rate": 1e-06,
      "loss": 0.0142,
      "step": 435
    },
    {
      "clip_ratio/high_max": 0.0034262624685652554,
      "clip_ratio/high_mean": 0.001429756406650995,
      "clip_ratio/low_mean": 0.0015347723965533078,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0029645288159372285,
      "epoch": 2.242636337124526,
      "grad_norm": 0.10741777718067169,
      "learning_rate": 1e-06,
      "loss": -0.0186,
      "step": 436
    },
    {
      "clip_ratio/high_max": 0.0035252406960353255,
      "clip_ratio/high_mean": 0.0013253975375846494,
      "clip_ratio/low_mean": 0.0015984931960701942,
      "clip_ratio/low_min": 4.451566928764805e-05,
      "clip_ratio/region_mean": 0.002923890635429416,
      "epoch": 2.247302420530767,
      "grad_norm": 0.10843174159526825,
      "learning_rate": 1e-06,
      "loss": 0.0107,
      "step": 437
    },
    {
      "clip_ratio/high_max": 0.003338203954626806,
      "clip_ratio/high_mean": 0.0015138691378524527,
      "clip_ratio/low_mean": 0.0018626106248120777,
      "clip_ratio/low_min": 5.695778600056656e-05,
      "clip_ratio/region_mean": 0.0033764796826289967,
      "epoch": 2.251968503937008,
      "grad_norm": 0.11338115483522415,
      "learning_rate": 1e-06,
      "loss": 0.0144,
      "step": 438
    },
    {
      "clip_ratio/high_max": 0.0035849112027790397,
      "clip_ratio/high_mean": 0.0014993317763583036,
      "clip_ratio/low_mean": 0.0015981576616468374,
      "clip_ratio/low_min": 0.0002949760055344086,
      "clip_ratio/region_mean": 0.0030974893816164695,
      "epoch": 2.2566345873432487,
      "grad_norm": 0.1093079000711441,
      "learning_rate": 1e-06,
      "loss": 0.009,
      "step": 439
    },
    {
      "clip_ratio/high_max": 0.0032063937760540284,
      "clip_ratio/high_mean": 0.001391851870721439,
      "clip_ratio/low_mean": 0.001739134757372085,
      "clip_ratio/low_min": 0.000219652940359083,
      "clip_ratio/region_mean": 0.003130986624455545,
      "epoch": 2.2613006707494896,
      "grad_norm": 0.1108342856168747,
      "learning_rate": 1e-06,
      "loss": 0.0253,
      "step": 440
    },
    {
      "clip_ratio/high_max": 0.0036253995494917035,
      "clip_ratio/high_mean": 0.0015476660410058685,
      "clip_ratio/low_mean": 0.0014662232515547657,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0030138892616378143,
      "epoch": 2.2659667541557305,
      "grad_norm": 0.10288877040147781,
      "learning_rate": 1e-06,
      "loss": -0.0264,
      "step": 441
    },
    {
      "clip_ratio/high_max": 0.0037480466125998646,
      "clip_ratio/high_mean": 0.0016365422488888726,
      "clip_ratio/low_mean": 0.00156330502795754,
      "clip_ratio/low_min": 8.596474799560383e-05,
      "clip_ratio/region_mean": 0.003199847327778116,
      "epoch": 2.2706328375619713,
      "grad_norm": 0.10725315660238266,
      "learning_rate": 1e-06,
      "loss": -0.0188,
      "step": 442
    },
    {
      "clip_ratio/high_max": 0.004500964569160715,
      "clip_ratio/high_mean": 0.001705740451143356,
      "clip_ratio/low_mean": 0.0012819714902434498,
      "clip_ratio/low_min": 8.288236676889937e-05,
      "clip_ratio/region_mean": 0.0029877119523007423,
      "epoch": 2.275298920968212,
      "grad_norm": 0.10594787448644638,
      "learning_rate": 1e-06,
      "loss": -0.0351,
      "step": 443
    },
    {
      "clip_ratio/high_max": 0.004100797508726828,
      "clip_ratio/high_mean": 0.001740014151437208,
      "clip_ratio/low_mean": 0.001421985107299406,
      "clip_ratio/low_min": 6.356195262924302e-05,
      "clip_ratio/region_mean": 0.0031619991932529956,
      "epoch": 2.279965004374453,
      "grad_norm": 0.10533588379621506,
      "learning_rate": 1e-06,
      "loss": -0.0518,
      "step": 444
    },
    {
      "clip_ratio/high_max": 0.004400987891131081,
      "clip_ratio/high_mean": 0.0017435043409932405,
      "clip_ratio/low_mean": 0.0017768207253539003,
      "clip_ratio/low_min": 0.0002692307716642972,
      "clip_ratio/region_mean": 0.003520325044519268,
      "epoch": 2.284631087780694,
      "grad_norm": 0.11279461532831192,
      "learning_rate": 1e-06,
      "loss": -0.0043,
      "step": 445
    },
    {
      "clip_ratio/high_max": 0.004288588956114836,
      "clip_ratio/high_mean": 0.001777282464900054,
      "clip_ratio/low_mean": 0.001472322539484594,
      "clip_ratio/low_min": 0.00015669946697016712,
      "clip_ratio/region_mean": 0.0032496050553163514,
      "epoch": 2.289297171186935,
      "grad_norm": 0.10840941220521927,
      "learning_rate": 1e-06,
      "loss": -0.0253,
      "step": 446
    },
    {
      "clip_ratio/high_max": 0.004287127871066332,
      "clip_ratio/high_mean": 0.0015566591318929568,
      "clip_ratio/low_mean": 0.0017069350724341348,
      "clip_ratio/low_min": 9.460927503823768e-05,
      "clip_ratio/region_mean": 0.0032635941461194307,
      "epoch": 2.2939632545931756,
      "grad_norm": 0.10871279239654541,
      "learning_rate": 1e-06,
      "loss": -0.0057,
      "step": 447
    },
    {
      "clip_ratio/high_max": 0.0036741375806741416,
      "clip_ratio/high_mean": 0.001712800654786406,
      "clip_ratio/low_mean": 0.0018228745502710808,
      "clip_ratio/low_min": 0.00013380825657804962,
      "clip_ratio/region_mean": 0.003535675161401741,
      "epoch": 2.298629337999417,
      "grad_norm": 0.1089814305305481,
      "learning_rate": 1e-06,
      "loss": -0.0087,
      "step": 448
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.031947544642857095,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4000.0,
      "completions/mean_length": 675.1322631835938,
      "completions/mean_terminated_length": 562.2372436523438,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 2.303295421405658,
      "grad_norm": 0.13560453057289124,
      "learning_rate": 1e-06,
      "loss": -0.0022,
      "num_tokens": 139616098.0,
      "reward": 0.6134207844734192,
      "reward_std": 0.18269461393356323,
      "rewards/simpleverify_reward/mean": 0.6134207844734192,
      "rewards/simpleverify_reward/std": 0.4869828224182129,
      "step": 449
    },
    {
      "clip_ratio/high_max": 0.002662798229721375,
      "clip_ratio/high_mean": 0.0010066003178508254,
      "clip_ratio/low_mean": 0.0005860599876541528,
      "clip_ratio/low_min": 1.711391087155789e-05,
      "clip_ratio/region_mean": 0.0015926602864055894,
      "epoch": 2.3079615048118987,
      "grad_norm": 0.1373986303806305,
      "learning_rate": 1e-06,
      "loss": 0.0097,
      "step": 450
    },
    {
      "clip_ratio/high_max": 0.0020532387716230005,
      "clip_ratio/high_mean": 0.0008312873451359337,
      "clip_ratio/low_mean": 0.0006628893934248481,
      "clip_ratio/low_min": 6.048646628187271e-05,
      "clip_ratio/region_mean": 0.0014941767585696653,
      "epoch": 2.3126275882181395,
      "grad_norm": 0.14679957926273346,
      "learning_rate": 1e-06,
      "loss": 0.0176,
      "step": 451
    },
    {
      "clip_ratio/high_max": 0.002346332825254649,
      "clip_ratio/high_mean": 0.0008792419430392329,
      "clip_ratio/low_mean": 0.0006438411310227821,
      "clip_ratio/low_min": 5.002196212444687e-05,
      "clip_ratio/region_mean": 0.0015230830758810043,
      "epoch": 2.3172936716243804,
      "grad_norm": 0.15684324502944946,
      "learning_rate": 1e-06,
      "loss": -0.0145,
      "step": 452
    },
    {
      "clip_ratio/high_max": 0.0017245245508092921,
      "clip_ratio/high_mean": 0.0007756291270197835,
      "clip_ratio/low_mean": 0.000635628761301632,
      "clip_ratio/low_min": 3.6554008147504646e-05,
      "clip_ratio/region_mean": 0.001411257868312532,
      "epoch": 2.3219597550306212,
      "grad_norm": 0.12472525238990784,
      "learning_rate": 1e-06,
      "loss": -0.0057,
      "step": 453
    },
    {
      "clip_ratio/high_max": 0.001852979195973603,
      "clip_ratio/high_mean": 0.0008053446053963853,
      "clip_ratio/low_mean": 0.0007463751599061652,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015517197680310346,
      "epoch": 2.326625838436862,
      "grad_norm": 0.1205039843916893,
      "learning_rate": 1e-06,
      "loss": 0.0192,
      "step": 454
    },
    {
      "clip_ratio/high_max": 0.0019482991265249439,
      "clip_ratio/high_mean": 0.0007998351611604448,
      "clip_ratio/low_mean": 0.0007874399507272756,
      "clip_ratio/low_min": 3.118762469966896e-05,
      "clip_ratio/region_mean": 0.001587275073688943,
      "epoch": 2.331291921843103,
      "grad_norm": 0.1338139921426773,
      "learning_rate": 1e-06,
      "loss": 0.007,
      "step": 455
    },
    {
      "clip_ratio/high_max": 0.002057045930996537,
      "clip_ratio/high_mean": 0.0008334159028891008,
      "clip_ratio/low_mean": 0.0008740925368329044,
      "clip_ratio/low_min": 9.247234447684605e-05,
      "clip_ratio/region_mean": 0.0017075084324460477,
      "epoch": 2.335958005249344,
      "grad_norm": 0.12611453235149384,
      "learning_rate": 1e-06,
      "loss": -0.0007,
      "step": 456
    },
    {
      "clip_ratio/high_max": 0.002171009764424525,
      "clip_ratio/high_mean": 0.0008809429327811813,
      "clip_ratio/low_mean": 0.0008542825289623579,
      "clip_ratio/low_min": 8.190461539925309e-05,
      "clip_ratio/region_mean": 0.0017352254217257723,
      "epoch": 2.3406240886555847,
      "grad_norm": 0.1341463327407837,
      "learning_rate": 1e-06,
      "loss": -0.0094,
      "step": 457
    },
    {
      "clip_ratio/high_max": 0.0023732053887215443,
      "clip_ratio/high_mean": 0.000965045075645321,
      "clip_ratio/low_mean": 0.000761050605433411,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017260956919926684,
      "epoch": 2.3452901720618256,
      "grad_norm": 0.15129999816417694,
      "learning_rate": 1e-06,
      "loss": -0.0069,
      "step": 458
    },
    {
      "clip_ratio/high_max": 0.0028331815701676533,
      "clip_ratio/high_mean": 0.0011413144566176925,
      "clip_ratio/low_mean": 0.0009408555924892426,
      "clip_ratio/low_min": 3.997015664936043e-05,
      "clip_ratio/region_mean": 0.0020821700454689562,
      "epoch": 2.3499562554680664,
      "grad_norm": 0.1275661438703537,
      "learning_rate": 1e-06,
      "loss": -0.0079,
      "step": 459
    },
    {
      "clip_ratio/high_max": 0.0024318585856235586,
      "clip_ratio/high_mean": 0.0009451883579458809,
      "clip_ratio/low_mean": 0.0007190724145402783,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016642607661196962,
      "epoch": 2.3546223388743073,
      "grad_norm": 0.11729730665683746,
      "learning_rate": 1e-06,
      "loss": -0.0198,
      "step": 460
    },
    {
      "clip_ratio/high_max": 0.002341119004995562,
      "clip_ratio/high_mean": 0.0009724948085931828,
      "clip_ratio/low_mean": 0.0007580633891848265,
      "clip_ratio/low_min": 2.6749410608317703e-05,
      "clip_ratio/region_mean": 0.001730558229610324,
      "epoch": 2.359288422280548,
      "grad_norm": 0.13335779309272766,
      "learning_rate": 1e-06,
      "loss": 0.0096,
      "step": 461
    },
    {
      "clip_ratio/high_max": 0.0022573262394871563,
      "clip_ratio/high_mean": 0.0008976729377536685,
      "clip_ratio/low_mean": 0.0009814268378249835,
      "clip_ratio/low_min": 3.0986700039647985e-05,
      "clip_ratio/region_mean": 0.0018790997346513905,
      "epoch": 2.363954505686789,
      "grad_norm": 0.12357683479785919,
      "learning_rate": 1e-06,
      "loss": 0.0343,
      "step": 462
    },
    {
      "clip_ratio/high_max": 0.0033517487318022177,
      "clip_ratio/high_mean": 0.0013386930950218812,
      "clip_ratio/low_mean": 0.0008741458204895025,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0022128388955025002,
      "epoch": 2.36862058909303,
      "grad_norm": 0.13469386100769043,
      "learning_rate": 1e-06,
      "loss": -0.0441,
      "step": 463
    },
    {
      "clip_ratio/high_max": 0.0025744532686076127,
      "clip_ratio/high_mean": 0.001144655772804981,
      "clip_ratio/low_mean": 0.0009500453143118648,
      "clip_ratio/low_min": 4.818975867237896e-05,
      "clip_ratio/region_mean": 0.002094701078021899,
      "epoch": 2.3732866724992707,
      "grad_norm": 0.12226388603448868,
      "learning_rate": 1e-06,
      "loss": -0.0329,
      "step": 464
    },
    {
      "clip_ratio/high_max": 0.004671756469178945,
      "clip_ratio/high_mean": 0.001799834550183732,
      "clip_ratio/low_mean": 0.001565938462590566,
      "clip_ratio/low_min": 2.939793012046721e-05,
      "clip_ratio/region_mean": 0.003365772936376743,
      "epoch": 2.377952755905512,
      "grad_norm": 0.10790296643972397,
      "learning_rate": 1e-06,
      "loss": -0.003,
      "step": 465
    },
    {
      "clip_ratio/high_max": 0.00415375752345426,
      "clip_ratio/high_mean": 0.0016365406627301127,
      "clip_ratio/low_mean": 0.0017844852081907447,
      "clip_ratio/low_min": 6.052293065295089e-05,
      "clip_ratio/region_mean": 0.003421025918214582,
      "epoch": 2.382618839311753,
      "grad_norm": 0.11333378404378891,
      "learning_rate": 1e-06,
      "loss": 0.0088,
      "step": 466
    },
    {
      "clip_ratio/high_max": 0.0034713881759671494,
      "clip_ratio/high_mean": 0.0015430484563694336,
      "clip_ratio/low_mean": 0.0017231767596967984,
      "clip_ratio/low_min": 9.146076445176732e-05,
      "clip_ratio/region_mean": 0.003266225176048465,
      "epoch": 2.3872849227179938,
      "grad_norm": 0.12146488577127457,
      "learning_rate": 1e-06,
      "loss": 0.0167,
      "step": 467
    },
    {
      "clip_ratio/high_max": 0.0036573573743226007,
      "clip_ratio/high_mean": 0.0015316211720346473,
      "clip_ratio/low_mean": 0.001459345188777661,
      "clip_ratio/low_min": 8.820857601676835e-05,
      "clip_ratio/region_mean": 0.002990966459037736,
      "epoch": 2.3919510061242346,
      "grad_norm": 0.10335852950811386,
      "learning_rate": 1e-06,
      "loss": -0.0153,
      "step": 468
    },
    {
      "clip_ratio/high_max": 0.0034409155559842475,
      "clip_ratio/high_mean": 0.0014419471262954175,
      "clip_ratio/low_mean": 0.0014170799549901858,
      "clip_ratio/low_min": 0.00013716963712795405,
      "clip_ratio/region_mean": 0.002859027103113476,
      "epoch": 2.3966170895304755,
      "grad_norm": 0.10907674580812454,
      "learning_rate": 1e-06,
      "loss": -0.0065,
      "step": 469
    },
    {
      "clip_ratio/high_max": 0.003288904183136765,
      "clip_ratio/high_mean": 0.0013959249918116257,
      "clip_ratio/low_mean": 0.0015320188313125982,
      "clip_ratio/low_min": 1.6855447029229254e-05,
      "clip_ratio/region_mean": 0.002927943831309676,
      "epoch": 2.4012831729367163,
      "grad_norm": 0.10341474413871765,
      "learning_rate": 1e-06,
      "loss": 0.0184,
      "step": 470
    },
    {
      "clip_ratio/high_max": 0.0037991502176737413,
      "clip_ratio/high_mean": 0.0015618669840478105,
      "clip_ratio/low_mean": 0.0015101758908713236,
      "clip_ratio/low_min": 0.00010493427816982148,
      "clip_ratio/region_mean": 0.0030720428985659964,
      "epoch": 2.405949256342957,
      "grad_norm": 0.11208900064229965,
      "learning_rate": 1e-06,
      "loss": 0.0062,
      "step": 471
    },
    {
      "clip_ratio/high_max": 0.003399367029487621,
      "clip_ratio/high_mean": 0.0014939034008421004,
      "clip_ratio/low_mean": 0.0016324436874128878,
      "clip_ratio/low_min": 0.00018419170555716846,
      "clip_ratio/region_mean": 0.003126347146462649,
      "epoch": 2.410615339749198,
      "grad_norm": 0.10840333253145218,
      "learning_rate": 1e-06,
      "loss": -0.0015,
      "step": 472
    },
    {
      "clip_ratio/high_max": 0.003826958723948337,
      "clip_ratio/high_mean": 0.0016156801648321562,
      "clip_ratio/low_mean": 0.0017929284913407173,
      "clip_ratio/low_min": 0.00014283613563748077,
      "clip_ratio/region_mean": 0.0034086086525348946,
      "epoch": 2.415281423155439,
      "grad_norm": 0.11251238733530045,
      "learning_rate": 1e-06,
      "loss": -0.0103,
      "step": 473
    },
    {
      "clip_ratio/high_max": 0.00387181145924842,
      "clip_ratio/high_mean": 0.0015990150968718808,
      "clip_ratio/low_mean": 0.0015509948789258488,
      "clip_ratio/low_min": 1.8876471585826948e-05,
      "clip_ratio/region_mean": 0.0031500099430559203,
      "epoch": 2.41994750656168,
      "grad_norm": 0.10769617557525635,
      "learning_rate": 1e-06,
      "loss": -0.0077,
      "step": 474
    },
    {
      "clip_ratio/high_max": 0.004455184505786747,
      "clip_ratio/high_mean": 0.001851320979767479,
      "clip_ratio/low_mean": 0.0019092342117801309,
      "clip_ratio/low_min": 5.718206739402376e-05,
      "clip_ratio/region_mean": 0.003760555264307186,
      "epoch": 2.4246135899679206,
      "grad_norm": 0.11215966194868088,
      "learning_rate": 1e-06,
      "loss": -0.0088,
      "step": 475
    },
    {
      "clip_ratio/high_max": 0.003923103271517903,
      "clip_ratio/high_mean": 0.0016443180502392352,
      "clip_ratio/low_mean": 0.0014249700361688156,
      "clip_ratio/low_min": 0.00012167021486675367,
      "clip_ratio/region_mean": 0.003069288097321987,
      "epoch": 2.4292796733741615,
      "grad_norm": 0.10074469447135925,
      "learning_rate": 1e-06,
      "loss": -0.0205,
      "step": 476
    },
    {
      "clip_ratio/high_max": 0.004374944735900499,
      "clip_ratio/high_mean": 0.0017573763761902228,
      "clip_ratio/low_mean": 0.0016272440589091275,
      "clip_ratio/low_min": 6.687353015877306e-05,
      "clip_ratio/region_mean": 0.0033846204460132867,
      "epoch": 2.4339457567804024,
      "grad_norm": 0.1102832704782486,
      "learning_rate": 1e-06,
      "loss": 0.0088,
      "step": 477
    },
    {
      "clip_ratio/high_max": 0.00414867246945505,
      "clip_ratio/high_mean": 0.0014914937546564033,
      "clip_ratio/low_mean": 0.0018508052453398705,
      "clip_ratio/low_min": 8.579262976127211e-05,
      "clip_ratio/region_mean": 0.0033422989363316447,
      "epoch": 2.4386118401866432,
      "grad_norm": 0.10344887524843216,
      "learning_rate": 1e-06,
      "loss": 0.0335,
      "step": 478
    },
    {
      "clip_ratio/high_max": 0.005057324873632751,
      "clip_ratio/high_mean": 0.002148227460565977,
      "clip_ratio/low_mean": 0.0016917338070925325,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038399612676585093,
      "epoch": 2.443277923592884,
      "grad_norm": 0.11644785106182098,
      "learning_rate": 1e-06,
      "loss": -0.0449,
      "step": 479
    },
    {
      "clip_ratio/high_max": 0.004269847355317324,
      "clip_ratio/high_mean": 0.0018749643277260475,
      "clip_ratio/low_mean": 0.0016782127677288372,
      "clip_ratio/low_min": 0.0001598740491317585,
      "clip_ratio/region_mean": 0.003553177055437118,
      "epoch": 2.447944006999125,
      "grad_norm": 0.10135716199874878,
      "learning_rate": 1e-06,
      "loss": -0.0337,
      "step": 480
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0374581473214286,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4092.0,
      "completions/mean_length": 705.0792846679688,
      "completions/mean_terminated_length": 573.1185913085938,
      "completions/min_length": 31.0,
      "completions/min_terminated_length": 31.0,
      "epoch": 2.452610090405366,
      "grad_norm": 0.12852303683757782,
      "learning_rate": 1e-06,
      "loss": 0.0317,
      "num_tokens": 148976402.0,
      "reward": 0.607979953289032,
      "reward_std": 0.18433348834514618,
      "rewards/simpleverify_reward/mean": 0.6079798936843872,
      "rewards/simpleverify_reward/std": 0.48821812868118286,
      "step": 481
    },
    {
      "clip_ratio/high_max": 0.0021084692125441507,
      "clip_ratio/high_mean": 0.0008005121344467625,
      "clip_ratio/low_mean": 0.000607643460170948,
      "clip_ratio/low_min": 2.724874775594799e-05,
      "clip_ratio/region_mean": 0.0014081555455049966,
      "epoch": 2.457276173811607,
      "grad_norm": 0.1232134997844696,
      "learning_rate": 1e-06,
      "loss": -0.0181,
      "step": 482
    },
    {
      "clip_ratio/high_max": 0.0019168363505741581,
      "clip_ratio/high_mean": 0.0008189441050490132,
      "clip_ratio/low_mean": 0.0006217357731657103,
      "clip_ratio/low_min": 3.1742001738166437e-05,
      "clip_ratio/region_mean": 0.001440679894585628,
      "epoch": 2.4619422572178475,
      "grad_norm": 0.1317141354084015,
      "learning_rate": 1e-06,
      "loss": -0.0012,
      "step": 483
    },
    {
      "clip_ratio/high_max": 0.0020278677548049018,
      "clip_ratio/high_mean": 0.0008180289787560469,
      "clip_ratio/low_mean": 0.0006180010841490002,
      "clip_ratio/low_min": 5.0158341764472425e-05,
      "clip_ratio/region_mean": 0.0014360300338012166,
      "epoch": 2.466608340624089,
      "grad_norm": 0.12922532856464386,
      "learning_rate": 1e-06,
      "loss": -0.0125,
      "step": 484
    },
    {
      "clip_ratio/high_max": 0.0019680234327097423,
      "clip_ratio/high_mean": 0.0007916820950413239,
      "clip_ratio/low_mean": 0.0007306614625122165,
      "clip_ratio/low_min": 0.00012786738079739735,
      "clip_ratio/region_mean": 0.0015223435693769716,
      "epoch": 2.4712744240303297,
      "grad_norm": 0.14038164913654327,
      "learning_rate": 1e-06,
      "loss": 0.0214,
      "step": 485
    },
    {
      "clip_ratio/high_max": 0.0025270534606534056,
      "clip_ratio/high_mean": 0.0010848390593309887,
      "clip_ratio/low_mean": 0.0007770573247398715,
      "clip_ratio/low_min": 6.573472091986332e-05,
      "clip_ratio/region_mean": 0.0018618964386405423,
      "epoch": 2.4759405074365706,
      "grad_norm": 0.13785941898822784,
      "learning_rate": 1e-06,
      "loss": -0.0327,
      "step": 486
    },
    {
      "clip_ratio/high_max": 0.0022735283491783775,
      "clip_ratio/high_mean": 0.0009832481373450719,
      "clip_ratio/low_mean": 0.0008201137879950693,
      "clip_ratio/low_min": 1.4599392670788802e-05,
      "clip_ratio/region_mean": 0.001803361956262961,
      "epoch": 2.4806065908428114,
      "grad_norm": 0.13082294166088104,
      "learning_rate": 1e-06,
      "loss": -0.0369,
      "step": 487
    },
    {
      "clip_ratio/high_max": 0.0020144440277363174,
      "clip_ratio/high_mean": 0.0008189416512323078,
      "clip_ratio/low_mean": 0.0007431305002683075,
      "clip_ratio/low_min": 1.236399566550972e-05,
      "clip_ratio/region_mean": 0.0015620721314917319,
      "epoch": 2.4852726742490523,
      "grad_norm": 0.11946125328540802,
      "learning_rate": 1e-06,
      "loss": -0.025,
      "step": 488
    },
    {
      "clip_ratio/high_max": 0.0023797392132109962,
      "clip_ratio/high_mean": 0.0009919476651703008,
      "clip_ratio/low_mean": 0.0008940879906731425,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018860356649383903,
      "epoch": 2.489938757655293,
      "grad_norm": 0.13558228313922882,
      "learning_rate": 1e-06,
      "loss": -0.0014,
      "step": 489
    },
    {
      "clip_ratio/high_max": 0.0021597225568257272,
      "clip_ratio/high_mean": 0.0009490915035712533,
      "clip_ratio/low_mean": 0.0008577941516705323,
      "clip_ratio/low_min": 4.657228055293672e-05,
      "clip_ratio/region_mean": 0.0018068856588797644,
      "epoch": 2.494604841061534,
      "grad_norm": 0.1347866654396057,
      "learning_rate": 1e-06,
      "loss": -0.0099,
      "step": 490
    },
    {
      "clip_ratio/high_max": 0.002171225831261836,
      "clip_ratio/high_mean": 0.0008730764202482533,
      "clip_ratio/low_mean": 0.0010358804611314554,
      "clip_ratio/low_min": 6.224335811566561e-05,
      "clip_ratio/region_mean": 0.0019089569250354543,
      "epoch": 2.499270924467775,
      "grad_norm": 0.14114463329315186,
      "learning_rate": 1e-06,
      "loss": 0.022,
      "step": 491
    },
    {
      "clip_ratio/high_max": 0.0021842748901690356,
      "clip_ratio/high_mean": 0.0008533395957783796,
      "clip_ratio/low_mean": 0.0008676701436343137,
      "clip_ratio/low_min": 3.713897876878036e-05,
      "clip_ratio/region_mean": 0.0017210097139468417,
      "epoch": 2.5039370078740157,
      "grad_norm": 0.12093949317932129,
      "learning_rate": 1e-06,
      "loss": -0.0087,
      "step": 492
    },
    {
      "clip_ratio/high_max": 0.0028033791968482547,
      "clip_ratio/high_mean": 0.0011462533238955075,
      "clip_ratio/low_mean": 0.000855478972880519,
      "clip_ratio/low_min": 9.531225532555254e-05,
      "clip_ratio/region_mean": 0.002001732304051984,
      "epoch": 2.5086030912802566,
      "grad_norm": 0.12668496370315552,
      "learning_rate": 1e-06,
      "loss": -0.0114,
      "step": 493
    },
    {
      "clip_ratio/high_max": 0.0021759943701908924,
      "clip_ratio/high_mean": 0.0009915331083902856,
      "clip_ratio/low_mean": 0.0009056482849700842,
      "clip_ratio/low_min": 4.5057275201543234e-05,
      "clip_ratio/region_mean": 0.001897181380627444,
      "epoch": 2.5132691746864975,
      "grad_norm": 0.13443271815776825,
      "learning_rate": 1e-06,
      "loss": 0.0273,
      "step": 494
    },
    {
      "clip_ratio/high_max": 0.002347228000871837,
      "clip_ratio/high_mean": 0.000997881365037756,
      "clip_ratio/low_mean": 0.0008956998663052218,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018935812258860096,
      "epoch": 2.5179352580927383,
      "grad_norm": 0.12433648854494095,
      "learning_rate": 1e-06,
      "loss": -0.0449,
      "step": 495
    },
    {
      "clip_ratio/high_max": 0.0024005414088605903,
      "clip_ratio/high_mean": 0.0009374722117172496,
      "clip_ratio/low_mean": 0.0009103637257794617,
      "clip_ratio/low_min": 3.594019653974101e-05,
      "clip_ratio/region_mean": 0.00184783587246784,
      "epoch": 2.522601341498979,
      "grad_norm": 0.1413072794675827,
      "learning_rate": 1e-06,
      "loss": 0.0159,
      "step": 496
    },
    {
      "clip_ratio/high_max": 0.003674121995572932,
      "clip_ratio/high_mean": 0.0014186388725647703,
      "clip_ratio/low_mean": 0.0017598874073883053,
      "clip_ratio/low_min": 8.525672819814645e-05,
      "clip_ratio/region_mean": 0.0031785261671757326,
      "epoch": 2.52726742490522,
      "grad_norm": 0.09868941456079483,
      "learning_rate": 1e-06,
      "loss": 0.0311,
      "step": 497
    },
    {
      "clip_ratio/high_max": 0.003550434521457646,
      "clip_ratio/high_mean": 0.0014151476934785023,
      "clip_ratio/low_mean": 0.0015810614640940912,
      "clip_ratio/low_min": 0.00021174636276555248,
      "clip_ratio/region_mean": 0.0029962091794004664,
      "epoch": 2.531933508311461,
      "grad_norm": 0.10350857675075531,
      "learning_rate": 1e-06,
      "loss": -0.0189,
      "step": 498
    },
    {
      "clip_ratio/high_max": 0.003506194698275067,
      "clip_ratio/high_mean": 0.0014947397576179355,
      "clip_ratio/low_mean": 0.0016258524665317964,
      "clip_ratio/low_min": 0.00010182085861742962,
      "clip_ratio/region_mean": 0.0031205921550281346,
      "epoch": 2.536599591717702,
      "grad_norm": 0.11277589946985245,
      "learning_rate": 1e-06,
      "loss": -0.002,
      "step": 499
    },
    {
      "clip_ratio/high_max": 0.003773927062866278,
      "clip_ratio/high_mean": 0.001571871300257044,
      "clip_ratio/low_mean": 0.0017267957155127078,
      "clip_ratio/low_min": 7.919558538560523e-05,
      "clip_ratio/region_mean": 0.003298667012131773,
      "epoch": 2.5412656751239426,
      "grad_norm": 0.11263630539178848,
      "learning_rate": 1e-06,
      "loss": -0.0133,
      "step": 500
    },
    {
      "clip_ratio/high_max": 0.0039356502165901475,
      "clip_ratio/high_mean": 0.0015397098522953456,
      "clip_ratio/low_mean": 0.0017598910235392395,
      "clip_ratio/low_min": 0.0002002167202590499,
      "clip_ratio/region_mean": 0.0032996008230838925,
      "epoch": 2.545931758530184,
      "grad_norm": 0.11384733021259308,
      "learning_rate": 1e-06,
      "loss": 0.0205,
      "step": 501
    },
    {
      "clip_ratio/high_max": 0.004131684509047773,
      "clip_ratio/high_mean": 0.0018422975626890548,
      "clip_ratio/low_mean": 0.0018192712050222326,
      "clip_ratio/low_min": 0.00019052830793953035,
      "clip_ratio/region_mean": 0.003661568815005012,
      "epoch": 2.5505978419364244,
      "grad_norm": 0.11411489546298981,
      "learning_rate": 1e-06,
      "loss": -0.0337,
      "step": 502
    },
    {
      "clip_ratio/high_max": 0.00433524209074676,
      "clip_ratio/high_mean": 0.0018117061408702284,
      "clip_ratio/low_mean": 0.0017565114831086248,
      "clip_ratio/low_min": 4.3798176193377e-05,
      "clip_ratio/region_mean": 0.0035682175657711923,
      "epoch": 2.5552639253426657,
      "grad_norm": 0.11059143394231796,
      "learning_rate": 1e-06,
      "loss": -0.0377,
      "step": 503
    },
    {
      "clip_ratio/high_max": 0.003451166870945599,
      "clip_ratio/high_mean": 0.0014772715949220583,
      "clip_ratio/low_mean": 0.0016109920143208,
      "clip_ratio/low_min": 3.9388687582686543e-05,
      "clip_ratio/region_mean": 0.0030882636492606252,
      "epoch": 2.5599300087489065,
      "grad_norm": 0.10452582687139511,
      "learning_rate": 1e-06,
      "loss": -0.0258,
      "step": 504
    },
    {
      "clip_ratio/high_max": 0.004215137523715384,
      "clip_ratio/high_mean": 0.00173685846675653,
      "clip_ratio/low_mean": 0.0019297957078379113,
      "clip_ratio/low_min": 0.00016768705609138124,
      "clip_ratio/region_mean": 0.003666654141852632,
      "epoch": 2.5645960921551474,
      "grad_norm": 0.11398331820964813,
      "learning_rate": 1e-06,
      "loss": -0.0022,
      "step": 505
    },
    {
      "clip_ratio/high_max": 0.0038633923686575145,
      "clip_ratio/high_mean": 0.0016690843985998072,
      "clip_ratio/low_mean": 0.0017599783823243342,
      "clip_ratio/low_min": 7.389907659671735e-05,
      "clip_ratio/region_mean": 0.003429062824579887,
      "epoch": 2.5692621755613883,
      "grad_norm": 0.11204434931278229,
      "learning_rate": 1e-06,
      "loss": -0.0108,
      "step": 506
    },
    {
      "clip_ratio/high_max": 0.0037272492627380416,
      "clip_ratio/high_mean": 0.0015997838127077557,
      "clip_ratio/low_mean": 0.001997105748159811,
      "clip_ratio/low_min": 0.00018258004638482817,
      "clip_ratio/region_mean": 0.0035968895390396938,
      "epoch": 2.573928258967629,
      "grad_norm": 0.11457454413175583,
      "learning_rate": 1e-06,
      "loss": 0.0211,
      "step": 507
    },
    {
      "clip_ratio/high_max": 0.0034826722694560885,
      "clip_ratio/high_mean": 0.0014327403987408616,
      "clip_ratio/low_mean": 0.0015533028235950042,
      "clip_ratio/low_min": 0.00014523957179335412,
      "clip_ratio/region_mean": 0.0029860432114219293,
      "epoch": 2.57859434237387,
      "grad_norm": 0.1013416275382042,
      "learning_rate": 1e-06,
      "loss": -0.0094,
      "step": 508
    },
    {
      "clip_ratio/high_max": 0.003945011740142945,
      "clip_ratio/high_mean": 0.0017254103004233912,
      "clip_ratio/low_mean": 0.0017066509208234493,
      "clip_ratio/low_min": 0.0002684723103811848,
      "clip_ratio/region_mean": 0.0034320610866416246,
      "epoch": 2.583260425780111,
      "grad_norm": 0.10854203253984451,
      "learning_rate": 1e-06,
      "loss": -0.0122,
      "step": 509
    },
    {
      "clip_ratio/high_max": 0.003653533356555272,
      "clip_ratio/high_mean": 0.001625843327929033,
      "clip_ratio/low_mean": 0.0018112894831574522,
      "clip_ratio/low_min": 0.00012891844380646944,
      "clip_ratio/region_mean": 0.003437132829276379,
      "epoch": 2.5879265091863517,
      "grad_norm": 0.10890080779790878,
      "learning_rate": 1e-06,
      "loss": 0.0265,
      "step": 510
    },
    {
      "clip_ratio/high_max": 0.004187806960544549,
      "clip_ratio/high_mean": 0.0017740552866598591,
      "clip_ratio/low_mean": 0.0014635746265412308,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0032376299204770476,
      "epoch": 2.5925925925925926,
      "grad_norm": 0.1078527644276619,
      "learning_rate": 1e-06,
      "loss": -0.0456,
      "step": 511
    },
    {
      "clip_ratio/high_max": 0.003899440009263344,
      "clip_ratio/high_mean": 0.0015392374043585733,
      "clip_ratio/low_mean": 0.0017809617420425639,
      "clip_ratio/low_min": 8.25409370008856e-05,
      "clip_ratio/region_mean": 0.0033201991391251795,
      "epoch": 2.5972586759988334,
      "grad_norm": 0.10906153172254562,
      "learning_rate": 1e-06,
      "loss": 0.0151,
      "step": 512
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0350864955357143,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4094.0,
      "completions/mean_length": 693.477294921875,
      "completions/mean_terminated_length": 569.7536010742188,
      "completions/min_length": 4.0,
      "completions/min_terminated_length": 4.0,
      "epoch": 2.6019247594050743,
      "grad_norm": 0.1475260853767395,
      "learning_rate": 1e-06,
      "loss": -0.0097,
      "num_tokens": 158284748.0,
      "reward": 0.6111189126968384,
      "reward_std": 0.18402212858200073,
      "rewards/simpleverify_reward/mean": 0.6111188530921936,
      "rewards/simpleverify_reward/std": 0.4875132739543915,
      "step": 513
    },
    {
      "clip_ratio/high_max": 0.0019652151349873748,
      "clip_ratio/high_mean": 0.0008320316374010872,
      "clip_ratio/low_mean": 0.0005833359300595475,
      "clip_ratio/low_min": 1.072593113349285e-05,
      "clip_ratio/region_mean": 0.0014153675183479208,
      "epoch": 2.606590842811315,
      "grad_norm": 0.11962283402681351,
      "learning_rate": 1e-06,
      "loss": 0.0187,
      "step": 514
    },
    {
      "clip_ratio/high_max": 0.002113870516041061,
      "clip_ratio/high_mean": 0.0009243134263670072,
      "clip_ratio/low_mean": 0.0005054532375652343,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014297666857601143,
      "epoch": 2.611256926217556,
      "grad_norm": 0.12318944185972214,
      "learning_rate": 1e-06,
      "loss": -0.0669,
      "step": 515
    },
    {
      "clip_ratio/high_max": 0.0018610406150401104,
      "clip_ratio/high_mean": 0.0007785976849845611,
      "clip_ratio/low_mean": 0.0006265361207624665,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001405133807566017,
      "epoch": 2.615923009623797,
      "grad_norm": 0.12739652395248413,
      "learning_rate": 1e-06,
      "loss": 0.0371,
      "step": 516
    },
    {
      "clip_ratio/high_max": 0.00208322043181397,
      "clip_ratio/high_mean": 0.0009511413118161727,
      "clip_ratio/low_mean": 0.0006188477600517217,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015699891009717248,
      "epoch": 2.6205890930300377,
      "grad_norm": 0.13826248049736023,
      "learning_rate": 1e-06,
      "loss": -0.0026,
      "step": 517
    },
    {
      "clip_ratio/high_max": 0.0019689827895490453,
      "clip_ratio/high_mean": 0.000817127724076272,
      "clip_ratio/low_mean": 0.0005973431498205173,
      "clip_ratio/low_min": 1.781641913112253e-05,
      "clip_ratio/region_mean": 0.0014144708584353793,
      "epoch": 2.625255176436279,
      "grad_norm": 0.1283932477235794,
      "learning_rate": 1e-06,
      "loss": -0.0237,
      "step": 518
    },
    {
      "clip_ratio/high_max": 0.0023613442754140124,
      "clip_ratio/high_mean": 0.0009085352576221339,
      "clip_ratio/low_mean": 0.0007759288200759329,
      "clip_ratio/low_min": 4.847200125368545e-05,
      "clip_ratio/region_mean": 0.0016844640485942364,
      "epoch": 2.6299212598425195,
      "grad_norm": 0.13844038546085358,
      "learning_rate": 1e-06,
      "loss": 0.0214,
      "step": 519
    },
    {
      "clip_ratio/high_max": 0.002220612019300461,
      "clip_ratio/high_mean": 0.00092753521312261,
      "clip_ratio/low_mean": 0.0007337414181165514,
      "clip_ratio/low_min": 5.5997400522755925e-05,
      "clip_ratio/region_mean": 0.0016612766194157302,
      "epoch": 2.6345873432487608,
      "grad_norm": 0.13712015748023987,
      "learning_rate": 1e-06,
      "loss": 0.0096,
      "step": 520
    },
    {
      "clip_ratio/high_max": 0.0021743183315265924,
      "clip_ratio/high_mean": 0.0008042732570174849,
      "clip_ratio/low_mean": 0.0007825538923498243,
      "clip_ratio/low_min": 3.0854040232952684e-05,
      "clip_ratio/region_mean": 0.0015868271730141714,
      "epoch": 2.6392534266550016,
      "grad_norm": 0.126194030046463,
      "learning_rate": 1e-06,
      "loss": -0.0213,
      "step": 521
    },
    {
      "clip_ratio/high_max": 0.002162649812817108,
      "clip_ratio/high_mean": 0.0008732019650778966,
      "clip_ratio/low_mean": 0.0007293393955478678,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016025413715397008,
      "epoch": 2.6439195100612425,
      "grad_norm": 0.11985906213521957,
      "learning_rate": 1e-06,
      "loss": -0.0225,
      "step": 522
    },
    {
      "clip_ratio/high_max": 0.0025311114732176065,
      "clip_ratio/high_mean": 0.0009471539833612042,
      "clip_ratio/low_mean": 0.0009047878102137474,
      "clip_ratio/low_min": 1.5031265320430975e-05,
      "clip_ratio/region_mean": 0.0018519418081268668,
      "epoch": 2.6485855934674833,
      "grad_norm": 0.12939739227294922,
      "learning_rate": 1e-06,
      "loss": 0.007,
      "step": 523
    },
    {
      "clip_ratio/high_max": 0.0025607263451092876,
      "clip_ratio/high_mean": 0.0010655549303919543,
      "clip_ratio/low_mean": 0.0008945987319748383,
      "clip_ratio/low_min": 1.4501159967039712e-05,
      "clip_ratio/region_mean": 0.0019601537205744535,
      "epoch": 2.653251676873724,
      "grad_norm": 0.12821923196315765,
      "learning_rate": 1e-06,
      "loss": -0.0148,
      "step": 524
    },
    {
      "clip_ratio/high_max": 0.0021615474252030253,
      "clip_ratio/high_mean": 0.0010168604494538158,
      "clip_ratio/low_mean": 0.0007930749343358912,
      "clip_ratio/low_min": 4.620494382834295e-05,
      "clip_ratio/region_mean": 0.001809935401979601,
      "epoch": 2.657917760279965,
      "grad_norm": 0.13569261133670807,
      "learning_rate": 1e-06,
      "loss": -0.0196,
      "step": 525
    },
    {
      "clip_ratio/high_max": 0.0020904801494907588,
      "clip_ratio/high_mean": 0.0008589704539190279,
      "clip_ratio/low_mean": 0.000916332444830914,
      "clip_ratio/low_min": 3.3943462767638266e-05,
      "clip_ratio/region_mean": 0.0017753029314917512,
      "epoch": 2.662583843686206,
      "grad_norm": 0.11915679275989532,
      "learning_rate": 1e-06,
      "loss": 0.0151,
      "step": 526
    },
    {
      "clip_ratio/high_max": 0.002438196010189131,
      "clip_ratio/high_mean": 0.0009277621356886812,
      "clip_ratio/low_mean": 0.0010768672018457437,
      "clip_ratio/low_min": 0.00012085863363608951,
      "clip_ratio/region_mean": 0.002004629379371181,
      "epoch": 2.667249927092447,
      "grad_norm": 0.1261160969734192,
      "learning_rate": 1e-06,
      "loss": 0.013,
      "step": 527
    },
    {
      "clip_ratio/high_max": 0.002577171726443339,
      "clip_ratio/high_mean": 0.0011147954428452067,
      "clip_ratio/low_mean": 0.0009820488085097168,
      "clip_ratio/low_min": 4.5479733671527356e-05,
      "clip_ratio/region_mean": 0.0020968442840967327,
      "epoch": 2.6719160104986877,
      "grad_norm": 0.14079050719738007,
      "learning_rate": 1e-06,
      "loss": -0.0395,
      "step": 528
    },
    {
      "clip_ratio/high_max": 0.004909978495561518,
      "clip_ratio/high_mean": 0.0018630345657584257,
      "clip_ratio/low_mean": 0.0016284975754388142,
      "clip_ratio/low_min": 8.20120440039318e-05,
      "clip_ratio/region_mean": 0.0034915321666630916,
      "epoch": 2.6765820939049285,
      "grad_norm": 0.10812269896268845,
      "learning_rate": 1e-06,
      "loss": -0.0106,
      "step": 529
    },
    {
      "clip_ratio/high_max": 0.003492072835797444,
      "clip_ratio/high_mean": 0.0015238677005982026,
      "clip_ratio/low_mean": 0.0017204745818162337,
      "clip_ratio/low_min": 9.878124183160253e-05,
      "clip_ratio/region_mean": 0.0032443422969663516,
      "epoch": 2.6812481773111694,
      "grad_norm": 0.10290117561817169,
      "learning_rate": 1e-06,
      "loss": 0.018,
      "step": 530
    },
    {
      "clip_ratio/high_max": 0.004130755361984484,
      "clip_ratio/high_mean": 0.0017278304003411904,
      "clip_ratio/low_mean": 0.0013872321324015502,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0031150625582085922,
      "epoch": 2.6859142607174102,
      "grad_norm": 0.10638885945081711,
      "learning_rate": 1e-06,
      "loss": -0.0677,
      "step": 531
    },
    {
      "clip_ratio/high_max": 0.0036738228154717945,
      "clip_ratio/high_mean": 0.0014542073986376636,
      "clip_ratio/low_mean": 0.0018049408572551329,
      "clip_ratio/low_min": 1.9177661670255475e-05,
      "clip_ratio/region_mean": 0.0032591483686701395,
      "epoch": 2.690580344123651,
      "grad_norm": 0.10722865909337997,
      "learning_rate": 1e-06,
      "loss": 0.0363,
      "step": 532
    },
    {
      "clip_ratio/high_max": 0.004179177354671992,
      "clip_ratio/high_mean": 0.001712921879516216,
      "clip_ratio/low_mean": 0.001874919304100331,
      "clip_ratio/low_min": 0.00018479634309187531,
      "clip_ratio/region_mean": 0.0035878411144949496,
      "epoch": 2.695246427529892,
      "grad_norm": 0.1133003756403923,
      "learning_rate": 1e-06,
      "loss": -0.0036,
      "step": 533
    },
    {
      "clip_ratio/high_max": 0.003596190952521283,
      "clip_ratio/high_mean": 0.0015738239271740895,
      "clip_ratio/low_mean": 0.0015485044787055813,
      "clip_ratio/low_min": 0.0001844694052124396,
      "clip_ratio/region_mean": 0.0031223283949657343,
      "epoch": 2.699912510936133,
      "grad_norm": 0.1069144681096077,
      "learning_rate": 1e-06,
      "loss": -0.0245,
      "step": 534
    },
    {
      "clip_ratio/high_max": 0.0038797700035502203,
      "clip_ratio/high_mean": 0.001496370736276731,
      "clip_ratio/low_mean": 0.0019469347898848355,
      "clip_ratio/low_min": 0.00012884446368843783,
      "clip_ratio/region_mean": 0.0034433055989211425,
      "epoch": 2.704578594342374,
      "grad_norm": 0.11485174298286438,
      "learning_rate": 1e-06,
      "loss": 0.0205,
      "step": 535
    },
    {
      "clip_ratio/high_max": 0.003940434180549346,
      "clip_ratio/high_mean": 0.0017118309042416513,
      "clip_ratio/low_mean": 0.0019194923661416396,
      "clip_ratio/low_min": 9.81588191280025e-05,
      "clip_ratio/region_mean": 0.0036313232994871214,
      "epoch": 2.7092446777486145,
      "grad_norm": 0.10777334868907928,
      "learning_rate": 1e-06,
      "loss": 0.0088,
      "step": 536
    },
    {
      "clip_ratio/high_max": 0.004144428370636888,
      "clip_ratio/high_mean": 0.001616441943042446,
      "clip_ratio/low_mean": 0.0017348838337056804,
      "clip_ratio/low_min": 8.14066079328768e-05,
      "clip_ratio/region_mean": 0.0033513257221784443,
      "epoch": 2.713910761154856,
      "grad_norm": 0.10382639616727829,
      "learning_rate": 1e-06,
      "loss": -0.022,
      "step": 537
    },
    {
      "clip_ratio/high_max": 0.004053154756547883,
      "clip_ratio/high_mean": 0.0015738232650619466,
      "clip_ratio/low_mean": 0.0016331381993950345,
      "clip_ratio/low_min": 2.6686591809266247e-05,
      "clip_ratio/region_mean": 0.0032069614389911294,
      "epoch": 2.7185768445610963,
      "grad_norm": 0.09988241642713547,
      "learning_rate": 1e-06,
      "loss": -0.0233,
      "step": 538
    },
    {
      "clip_ratio/high_max": 0.00402184094855329,
      "clip_ratio/high_mean": 0.0016831040848046541,
      "clip_ratio/low_mean": 0.001825773790187668,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003508877707645297,
      "epoch": 2.7232429279673376,
      "grad_norm": 0.10986864566802979,
      "learning_rate": 1e-06,
      "loss": 0.0061,
      "step": 539
    },
    {
      "clip_ratio/high_max": 0.004511283041210845,
      "clip_ratio/high_mean": 0.001832957030273974,
      "clip_ratio/low_mean": 0.0017417254821339156,
      "clip_ratio/low_min": 5.5598064136574976e-05,
      "clip_ratio/region_mean": 0.0035746825742535293,
      "epoch": 2.7279090113735784,
      "grad_norm": 0.10535531491041183,
      "learning_rate": 1e-06,
      "loss": -0.0156,
      "step": 540
    },
    {
      "clip_ratio/high_max": 0.00435531378025189,
      "clip_ratio/high_mean": 0.0017407431168976473,
      "clip_ratio/low_mean": 0.0016569660874665715,
      "clip_ratio/low_min": 0.00021277108680806123,
      "clip_ratio/region_mean": 0.0033977091588894837,
      "epoch": 2.7325750947798193,
      "grad_norm": 0.105378657579422,
      "learning_rate": 1e-06,
      "loss": -0.0205,
      "step": 541
    },
    {
      "clip_ratio/high_max": 0.003990085984696634,
      "clip_ratio/high_mean": 0.0015394604524772149,
      "clip_ratio/low_mean": 0.001715981215966167,
      "clip_ratio/low_min": 7.436196028720587e-05,
      "clip_ratio/region_mean": 0.0032554416538914666,
      "epoch": 2.73724117818606,
      "grad_norm": 0.09573499858379364,
      "learning_rate": 1e-06,
      "loss": 0.0144,
      "step": 542
    },
    {
      "clip_ratio/high_max": 0.004024940775707364,
      "clip_ratio/high_mean": 0.0016183969419216737,
      "clip_ratio/low_mean": 0.0019635228272818495,
      "clip_ratio/low_min": 0.00028928117535542697,
      "clip_ratio/region_mean": 0.0035819198164972477,
      "epoch": 2.741907261592301,
      "grad_norm": 0.10836490988731384,
      "learning_rate": 1e-06,
      "loss": 0.0122,
      "step": 543
    },
    {
      "clip_ratio/high_max": 0.004349221708253026,
      "clip_ratio/high_mean": 0.001780041384336073,
      "clip_ratio/low_mean": 0.001862231467384845,
      "clip_ratio/low_min": 9.7070508672914e-05,
      "clip_ratio/region_mean": 0.0036422728589968756,
      "epoch": 2.746573344998542,
      "grad_norm": 0.11187615245580673,
      "learning_rate": 1e-06,
      "loss": -0.0404,
      "step": 544
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0426199776785714,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4089.0,
      "completions/mean_length": 717.9034423828125,
      "completions/mean_terminated_length": 567.5196533203125,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 2.7512394284047827,
      "grad_norm": 0.14116381108760834,
      "learning_rate": 1e-06,
      "loss": -0.0136,
      "num_tokens": 167471355.0,
      "reward": 0.6053990125656128,
      "reward_std": 0.18358783423900604,
      "rewards/simpleverify_reward/mean": 0.6053990125656128,
      "rewards/simpleverify_reward/std": 0.4887818992137909,
      "step": 545
    },
    {
      "clip_ratio/high_max": 0.0020968794779037125,
      "clip_ratio/high_mean": 0.0008818594124022638,
      "clip_ratio/low_mean": 0.0005725626460844069,
      "clip_ratio/low_min": 2.2285612431005575e-05,
      "clip_ratio/region_mean": 0.0014544220321113244,
      "epoch": 2.7559055118110236,
      "grad_norm": 0.13967737555503845,
      "learning_rate": 1e-06,
      "loss": -0.0144,
      "step": 546
    },
    {
      "clip_ratio/high_max": 0.0021594412537524477,
      "clip_ratio/high_mean": 0.0008386442477785749,
      "clip_ratio/low_mean": 0.0006798493777750991,
      "clip_ratio/low_min": 3.586220555007458e-05,
      "clip_ratio/region_mean": 0.0015184936492005363,
      "epoch": 2.7605715952172645,
      "grad_norm": 0.1388244926929474,
      "learning_rate": 1e-06,
      "loss": 0.0146,
      "step": 547
    },
    {
      "clip_ratio/high_max": 0.0021169140527490526,
      "clip_ratio/high_mean": 0.0009170684188575251,
      "clip_ratio/low_mean": 0.0006611868520849384,
      "clip_ratio/low_min": 6.811607181589352e-05,
      "clip_ratio/region_mean": 0.00157825525093358,
      "epoch": 2.7652376786235053,
      "grad_norm": 0.12524600327014923,
      "learning_rate": 1e-06,
      "loss": -0.0265,
      "step": 548
    },
    {
      "clip_ratio/high_max": 0.001675629973760806,
      "clip_ratio/high_mean": 0.0006925517864146968,
      "clip_ratio/low_mean": 0.0007028551681287354,
      "clip_ratio/low_min": 2.647488236107165e-05,
      "clip_ratio/region_mean": 0.0013954069509054534,
      "epoch": 2.769903762029746,
      "grad_norm": 0.13385963439941406,
      "learning_rate": 1e-06,
      "loss": 0.0006,
      "step": 549
    },
    {
      "clip_ratio/high_max": 0.0024384894568356685,
      "clip_ratio/high_mean": 0.0009183941892842995,
      "clip_ratio/low_mean": 0.000669166764964757,
      "clip_ratio/low_min": 1.547412648505997e-05,
      "clip_ratio/region_mean": 0.0015875609278737102,
      "epoch": 2.774569845435987,
      "grad_norm": 0.1454942524433136,
      "learning_rate": 1e-06,
      "loss": -0.0139,
      "step": 550
    },
    {
      "clip_ratio/high_max": 0.0026716771826613694,
      "clip_ratio/high_mean": 0.0010955473135254579,
      "clip_ratio/low_mean": 0.0009681672636361327,
      "clip_ratio/low_min": 8.554321993869962e-05,
      "clip_ratio/region_mean": 0.002063714615360368,
      "epoch": 2.779235928842228,
      "grad_norm": 0.13865989446640015,
      "learning_rate": 1e-06,
      "loss": 0.0056,
      "step": 551
    },
    {
      "clip_ratio/high_max": 0.0021705419858335517,
      "clip_ratio/high_mean": 0.0008571414418838685,
      "clip_ratio/low_mean": 0.0007973707952260156,
      "clip_ratio/low_min": 4.931644525640877e-05,
      "clip_ratio/region_mean": 0.0016545122125535272,
      "epoch": 2.783902012248469,
      "grad_norm": 0.12419628351926804,
      "learning_rate": 1e-06,
      "loss": -0.0065,
      "step": 552
    },
    {
      "clip_ratio/high_max": 0.002367948087339755,
      "clip_ratio/high_mean": 0.0009834785960265435,
      "clip_ratio/low_mean": 0.0009141218979493715,
      "clip_ratio/low_min": 9.0511259259074e-05,
      "clip_ratio/region_mean": 0.0018976005376316607,
      "epoch": 2.7885680956547096,
      "grad_norm": 0.13322414457798004,
      "learning_rate": 1e-06,
      "loss": 0.0248,
      "step": 553
    },
    {
      "clip_ratio/high_max": 0.0021278387284837663,
      "clip_ratio/high_mean": 0.0009322490532213124,
      "clip_ratio/low_mean": 0.0008498225506627932,
      "clip_ratio/low_min": 6.091130489949137e-05,
      "clip_ratio/region_mean": 0.0017820715947891586,
      "epoch": 2.793234179060951,
      "grad_norm": 0.13163484632968903,
      "learning_rate": 1e-06,
      "loss": 0.0048,
      "step": 554
    },
    {
      "clip_ratio/high_max": 0.0026034091933979653,
      "clip_ratio/high_mean": 0.001054678532455,
      "clip_ratio/low_mean": 0.0007656920643057674,
      "clip_ratio/low_min": 2.7412281269789673e-05,
      "clip_ratio/region_mean": 0.0018203706094936933,
      "epoch": 2.7979002624671914,
      "grad_norm": 0.1273442655801773,
      "learning_rate": 1e-06,
      "loss": -0.022,
      "step": 555
    },
    {
      "clip_ratio/high_max": 0.002598668186692521,
      "clip_ratio/high_mean": 0.001138136452937033,
      "clip_ratio/low_mean": 0.0007196783690233133,
      "clip_ratio/low_min": 2.2202486434252933e-05,
      "clip_ratio/region_mean": 0.0018578148301457986,
      "epoch": 2.8025663458734327,
      "grad_norm": 0.1400269716978073,
      "learning_rate": 1e-06,
      "loss": -0.0288,
      "step": 556
    },
    {
      "clip_ratio/high_max": 0.0021816730441059917,
      "clip_ratio/high_mean": 0.0008486886363243684,
      "clip_ratio/low_mean": 0.0009521437132207211,
      "clip_ratio/low_min": 6.399961603165139e-05,
      "clip_ratio/region_mean": 0.0018008323313551955,
      "epoch": 2.8072324292796735,
      "grad_norm": 0.12354227900505066,
      "learning_rate": 1e-06,
      "loss": 0.0202,
      "step": 557
    },
    {
      "clip_ratio/high_max": 0.0021771731480839662,
      "clip_ratio/high_mean": 0.001063119969330728,
      "clip_ratio/low_mean": 0.0008848187626426807,
      "clip_ratio/low_min": 4.150066524744034e-05,
      "clip_ratio/region_mean": 0.0019479387410683557,
      "epoch": 2.8118985126859144,
      "grad_norm": 0.1312863975763321,
      "learning_rate": 1e-06,
      "loss": -0.0616,
      "step": 558
    },
    {
      "clip_ratio/high_max": 0.0024527662608306855,
      "clip_ratio/high_mean": 0.001107616431909264,
      "clip_ratio/low_mean": 0.0009406776443938725,
      "clip_ratio/low_min": 5.859469729330158e-05,
      "clip_ratio/region_mean": 0.002048294038104359,
      "epoch": 2.8165645960921553,
      "grad_norm": 0.13248229026794434,
      "learning_rate": 1e-06,
      "loss": -0.0344,
      "step": 559
    },
    {
      "clip_ratio/high_max": 0.002234244191640755,
      "clip_ratio/high_mean": 0.0009262936237064423,
      "clip_ratio/low_mean": 0.0007471953795175068,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016734890195948537,
      "epoch": 2.821230679498396,
      "grad_norm": 0.13744542002677917,
      "learning_rate": 1e-06,
      "loss": -0.0018,
      "step": 560
    },
    {
      "clip_ratio/high_max": 0.004274173661542591,
      "clip_ratio/high_mean": 0.0018019320341409184,
      "clip_ratio/low_mean": 0.0017173339365399443,
      "clip_ratio/low_min": 0.00020110520017624367,
      "clip_ratio/region_mean": 0.003519265985232778,
      "epoch": 2.825896762904637,
      "grad_norm": 0.1059526652097702,
      "learning_rate": 1e-06,
      "loss": -0.0145,
      "step": 561
    },
    {
      "clip_ratio/high_max": 0.004140589953749441,
      "clip_ratio/high_mean": 0.0016246105260506738,
      "clip_ratio/low_mean": 0.0016780312598712044,
      "clip_ratio/low_min": 0.00012833722303184913,
      "clip_ratio/region_mean": 0.0033026418095687404,
      "epoch": 2.830562846310878,
      "grad_norm": 0.10314527153968811,
      "learning_rate": 1e-06,
      "loss": -0.0153,
      "step": 562
    },
    {
      "clip_ratio/high_max": 0.003760451581911184,
      "clip_ratio/high_mean": 0.0016546626247873064,
      "clip_ratio/low_mean": 0.0018367113661952317,
      "clip_ratio/low_min": 0.0002478587775840424,
      "clip_ratio/region_mean": 0.003491373936412856,
      "epoch": 2.8352289297171187,
      "grad_norm": 0.11692745983600616,
      "learning_rate": 1e-06,
      "loss": 0.0136,
      "step": 563
    },
    {
      "clip_ratio/high_max": 0.003624541684985161,
      "clip_ratio/high_mean": 0.001824067108827876,
      "clip_ratio/low_mean": 0.001636762019188609,
      "clip_ratio/low_min": 0.00021367341287259478,
      "clip_ratio/region_mean": 0.003460829073446803,
      "epoch": 2.8398950131233596,
      "grad_norm": 0.10747841000556946,
      "learning_rate": 1e-06,
      "loss": -0.0273,
      "step": 564
    },
    {
      "clip_ratio/high_max": 0.0031877466826699674,
      "clip_ratio/high_mean": 0.0013683579563803505,
      "clip_ratio/low_mean": 0.0017441871532355435,
      "clip_ratio/low_min": 0.00013237440725788474,
      "clip_ratio/region_mean": 0.0031125450477702543,
      "epoch": 2.8445610965296004,
      "grad_norm": 0.09903515130281448,
      "learning_rate": 1e-06,
      "loss": -0.0002,
      "step": 565
    },
    {
      "clip_ratio/high_max": 0.00380195549223572,
      "clip_ratio/high_mean": 0.0015198203218460549,
      "clip_ratio/low_mean": 0.0015914174073259346,
      "clip_ratio/low_min": 2.9460286896210164e-05,
      "clip_ratio/region_mean": 0.003111237718258053,
      "epoch": 2.8492271799358413,
      "grad_norm": 0.10769470781087875,
      "learning_rate": 1e-06,
      "loss": -0.0147,
      "step": 566
    },
    {
      "clip_ratio/high_max": 0.004457996037672274,
      "clip_ratio/high_mean": 0.0017717695991450455,
      "clip_ratio/low_mean": 0.00208011174981948,
      "clip_ratio/low_min": 0.00032777949672890827,
      "clip_ratio/region_mean": 0.003851881338050589,
      "epoch": 2.853893263342082,
      "grad_norm": 0.11015890538692474,
      "learning_rate": 1e-06,
      "loss": 0.0048,
      "step": 567
    },
    {
      "clip_ratio/high_max": 0.004142918303841725,
      "clip_ratio/high_mean": 0.0017271442848141305,
      "clip_ratio/low_mean": 0.0015439957460330334,
      "clip_ratio/low_min": 0.00016872864944161847,
      "clip_ratio/region_mean": 0.00327114001265727,
      "epoch": 2.858559346748323,
      "grad_norm": 0.10232267528772354,
      "learning_rate": 1e-06,
      "loss": -0.0072,
      "step": 568
    },
    {
      "clip_ratio/high_max": 0.004255149920936674,
      "clip_ratio/high_mean": 0.0017199331705342047,
      "clip_ratio/low_mean": 0.001891954270831775,
      "clip_ratio/low_min": 0.0002598880819277838,
      "clip_ratio/region_mean": 0.0036118874413659796,
      "epoch": 2.863225430154564,
      "grad_norm": 0.10648967325687408,
      "learning_rate": 1e-06,
      "loss": 0.0239,
      "step": 569
    },
    {
      "clip_ratio/high_max": 0.0038735465932404622,
      "clip_ratio/high_mean": 0.001664111128775403,
      "clip_ratio/low_mean": 0.0018075093867082614,
      "clip_ratio/low_min": 0.0001073148669092916,
      "clip_ratio/region_mean": 0.0034716205555014312,
      "epoch": 2.8678915135608047,
      "grad_norm": 0.11051440984010696,
      "learning_rate": 1e-06,
      "loss": 0.0039,
      "step": 570
    },
    {
      "clip_ratio/high_max": 0.0039921955758472905,
      "clip_ratio/high_mean": 0.001765259828971466,
      "clip_ratio/low_mean": 0.0016162668107426725,
      "clip_ratio/low_min": 2.7412281269789673e-05,
      "clip_ratio/region_mean": 0.003381526650628075,
      "epoch": 2.872557596967046,
      "grad_norm": 0.10644158720970154,
      "learning_rate": 1e-06,
      "loss": -0.0228,
      "step": 571
    },
    {
      "clip_ratio/high_max": 0.004905263805994764,
      "clip_ratio/high_mean": 0.0019660901816678233,
      "clip_ratio/low_mean": 0.0016763871171860956,
      "clip_ratio/low_min": 0.00010509254934731871,
      "clip_ratio/region_mean": 0.0036424772697500885,
      "epoch": 2.8772236803732865,
      "grad_norm": 0.11353842914104462,
      "learning_rate": 1e-06,
      "loss": -0.0297,
      "step": 572
    },
    {
      "clip_ratio/high_max": 0.0035181549537810497,
      "clip_ratio/high_mean": 0.0014502295816782862,
      "clip_ratio/low_mean": 0.0019052155621466227,
      "clip_ratio/low_min": 0.00018920008369605057,
      "clip_ratio/region_mean": 0.003355445194756612,
      "epoch": 2.8818897637795278,
      "grad_norm": 0.1028515174984932,
      "learning_rate": 1e-06,
      "loss": 0.0194,
      "step": 573
    },
    {
      "clip_ratio/high_max": 0.004392321279738098,
      "clip_ratio/high_mean": 0.002010108612012118,
      "clip_ratio/low_mean": 0.0015710492880316451,
      "clip_ratio/low_min": 5.187583155930042e-05,
      "clip_ratio/region_mean": 0.0035811579728033394,
      "epoch": 2.886555847185768,
      "grad_norm": 0.1130344569683075,
      "learning_rate": 1e-06,
      "loss": -0.0624,
      "step": 574
    },
    {
      "clip_ratio/high_max": 0.00486399355577305,
      "clip_ratio/high_mean": 0.002024112342041917,
      "clip_ratio/low_mean": 0.0018441505380906165,
      "clip_ratio/low_min": 0.000149856223288225,
      "clip_ratio/region_mean": 0.003868263040203601,
      "epoch": 2.8912219305920095,
      "grad_norm": 0.11336081475019455,
      "learning_rate": 1e-06,
      "loss": -0.0352,
      "step": 575
    },
    {
      "clip_ratio/high_max": 0.003587546780181583,
      "clip_ratio/high_mean": 0.0014841021857137093,
      "clip_ratio/low_mean": 0.0015968015486578224,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0030809037198196165,
      "epoch": 2.8958880139982504,
      "grad_norm": 0.10892689228057861,
      "learning_rate": 1e-06,
      "loss": -0.0027,
      "step": 576
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0394810267857143,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4086.0,
      "completions/mean_length": 705.7078247070312,
      "completions/mean_terminated_length": 566.3536987304688,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 3.004666083406241,
      "grad_norm": 0.1297464817762375,
      "learning_rate": 1e-06,
      "loss": -0.029,
      "num_tokens": 176704134.0,
      "reward": 0.618722140789032,
      "reward_std": 0.16918261349201202,
      "rewards/simpleverify_reward/mean": 0.6187220811843872,
      "rewards/simpleverify_reward/std": 0.4857175350189209,
      "step": 577
    },
    {
      "clip_ratio/high_max": 0.0018989417585544288,
      "clip_ratio/high_mean": 0.000783735473305569,
      "clip_ratio/low_mean": 0.0005315109619914438,
      "clip_ratio/low_min": 4.201998035568977e-05,
      "clip_ratio/region_mean": 0.0013152464307495393,
      "epoch": 3.0093321668124817,
      "grad_norm": 0.13070416450500488,
      "learning_rate": 1e-06,
      "loss": 0.0218,
      "step": 578
    },
    {
      "clip_ratio/high_max": 0.0021109459921717644,
      "clip_ratio/high_mean": 0.0008489151005051099,
      "clip_ratio/low_mean": 0.0006225379947863985,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014714530880155507,
      "epoch": 3.0139982502187226,
      "grad_norm": 0.12053510546684265,
      "learning_rate": 1e-06,
      "loss": -0.0079,
      "step": 579
    },
    {
      "clip_ratio/high_max": 0.0019092993134108838,
      "clip_ratio/high_mean": 0.000706828361217049,
      "clip_ratio/low_mean": 0.0006537435820064275,
      "clip_ratio/low_min": 2.4892449800972827e-05,
      "clip_ratio/region_mean": 0.0013605719541374128,
      "epoch": 3.0186643336249634,
      "grad_norm": 0.13016866147518158,
      "learning_rate": 1e-06,
      "loss": 0.005,
      "step": 580
    },
    {
      "clip_ratio/high_max": 0.0017881428502732888,
      "clip_ratio/high_mean": 0.000659828618154279,
      "clip_ratio/low_mean": 0.000642074968709494,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013019035795878153,
      "epoch": 3.0233304170312043,
      "grad_norm": 0.136167511343956,
      "learning_rate": 1e-06,
      "loss": 0.0218,
      "step": 581
    },
    {
      "clip_ratio/high_max": 0.0018641004899109248,
      "clip_ratio/high_mean": 0.0008369101287826197,
      "clip_ratio/low_mean": 0.000613246332250128,
      "clip_ratio/low_min": 2.5461544282734394e-05,
      "clip_ratio/region_mean": 0.0014501564692182,
      "epoch": 3.027996500437445,
      "grad_norm": 0.13248851895332336,
      "learning_rate": 1e-06,
      "loss": -0.014,
      "step": 582
    },
    {
      "clip_ratio/high_max": 0.002190739138313802,
      "clip_ratio/high_mean": 0.0010033369308075635,
      "clip_ratio/low_mean": 0.0006626780996157322,
      "clip_ratio/low_min": 3.0676131245854776e-05,
      "clip_ratio/region_mean": 0.001666015014052391,
      "epoch": 3.032662583843686,
      "grad_norm": 0.13862930238246918,
      "learning_rate": 1e-06,
      "loss": -0.0561,
      "step": 583
    },
    {
      "clip_ratio/high_max": 0.002345991673792014,
      "clip_ratio/high_mean": 0.0008923112932279764,
      "clip_ratio/low_mean": 0.0008589484114054358,
      "clip_ratio/low_min": 8.714826981304213e-05,
      "clip_ratio/region_mean": 0.00175125971873058,
      "epoch": 3.037328667249927,
      "grad_norm": 0.15334631502628326,
      "learning_rate": 1e-06,
      "loss": -0.0111,
      "step": 584
    },
    {
      "clip_ratio/high_max": 0.0027672940341290087,
      "clip_ratio/high_mean": 0.0009380742576468037,
      "clip_ratio/low_mean": 0.0007088610309438081,
      "clip_ratio/low_min": 1.2254901776032057e-05,
      "clip_ratio/region_mean": 0.0016469352704007179,
      "epoch": 3.041994750656168,
      "grad_norm": 0.12957873940467834,
      "learning_rate": 1e-06,
      "loss": -0.0108,
      "step": 585
    },
    {
      "clip_ratio/high_max": 0.002369338530115783,
      "clip_ratio/high_mean": 0.000981785746262176,
      "clip_ratio/low_mean": 0.000874948367709294,
      "clip_ratio/low_min": 6.245668737392407e-05,
      "clip_ratio/region_mean": 0.0018567340957815759,
      "epoch": 3.046660834062409,
      "grad_norm": 0.1311190277338028,
      "learning_rate": 1e-06,
      "loss": 0.0056,
      "step": 586
    },
    {
      "clip_ratio/high_max": 0.0024873573711374775,
      "clip_ratio/high_mean": 0.0008692332012287807,
      "clip_ratio/low_mean": 0.0007939138668007217,
      "clip_ratio/low_min": 3.510250098770484e-05,
      "clip_ratio/region_mean": 0.0016631471080472693,
      "epoch": 3.05132691746865,
      "grad_norm": 0.13392719626426697,
      "learning_rate": 1e-06,
      "loss": -0.0227,
      "step": 587
    },
    {
      "clip_ratio/high_max": 0.0024968598809209652,
      "clip_ratio/high_mean": 0.000883779732248513,
      "clip_ratio/low_mean": 0.0008742020499994396,
      "clip_ratio/low_min": 0.0001033503267535707,
      "clip_ratio/region_mean": 0.0017579817795194685,
      "epoch": 3.055993000874891,
      "grad_norm": 0.14168329536914825,
      "learning_rate": 1e-06,
      "loss": 0.0495,
      "step": 588
    },
    {
      "clip_ratio/high_max": 0.0021550678720814176,
      "clip_ratio/high_mean": 0.000981088429398369,
      "clip_ratio/low_mean": 0.0007827279569028178,
      "clip_ratio/low_min": 3.882812688971171e-05,
      "clip_ratio/region_mean": 0.0017638164063100703,
      "epoch": 3.0606590842811316,
      "grad_norm": 0.12165799736976624,
      "learning_rate": 1e-06,
      "loss": -0.046,
      "step": 589
    },
    {
      "clip_ratio/high_max": 0.0021453929803101346,
      "clip_ratio/high_mean": 0.0009457574033149285,
      "clip_ratio/low_mean": 0.0008421002603427041,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001787857705494389,
      "epoch": 3.0653251676873725,
      "grad_norm": 0.1449015587568283,
      "learning_rate": 1e-06,
      "loss": -0.0336,
      "step": 590
    },
    {
      "clip_ratio/high_max": 0.0020800196180061903,
      "clip_ratio/high_mean": 0.0009002483166113961,
      "clip_ratio/low_mean": 0.0010150725538551342,
      "clip_ratio/low_min": 8.30836297609494e-05,
      "clip_ratio/region_mean": 0.0019153208268107846,
      "epoch": 3.0699912510936134,
      "grad_norm": 0.1379612535238266,
      "learning_rate": 1e-06,
      "loss": -0.0001,
      "step": 591
    },
    {
      "clip_ratio/high_max": 0.0024064477256615646,
      "clip_ratio/high_mean": 0.0008938748833315913,
      "clip_ratio/low_mean": 0.000986194250799599,
      "clip_ratio/low_min": 1.27577059174655e-05,
      "clip_ratio/region_mean": 0.0018800691541400738,
      "epoch": 3.0746573344998542,
      "grad_norm": 0.12321005761623383,
      "learning_rate": 1e-06,
      "loss": -0.0018,
      "step": 592
    },
    {
      "clip_ratio/high_max": 0.003995479084551334,
      "clip_ratio/high_mean": 0.0016497129763592966,
      "clip_ratio/low_mean": 0.0013652391789946705,
      "clip_ratio/low_min": 4.703963713836856e-05,
      "clip_ratio/region_mean": 0.0030149521626299247,
      "epoch": 3.079323417906095,
      "grad_norm": 0.09836620837450027,
      "learning_rate": 1e-06,
      "loss": -0.0296,
      "step": 593
    },
    {
      "clip_ratio/high_max": 0.003406795542105101,
      "clip_ratio/high_mean": 0.0014184886276780162,
      "clip_ratio/low_mean": 0.0017059965448424919,
      "clip_ratio/low_min": 0.00016521043380635092,
      "clip_ratio/region_mean": 0.003124485163425561,
      "epoch": 3.083989501312336,
      "grad_norm": 0.10543031245470047,
      "learning_rate": 1e-06,
      "loss": 0.0209,
      "step": 594
    },
    {
      "clip_ratio/high_max": 0.0038427935942308977,
      "clip_ratio/high_mean": 0.001502671653724974,
      "clip_ratio/low_mean": 0.0016468072208226658,
      "clip_ratio/low_min": 3.391209975234233e-05,
      "clip_ratio/region_mean": 0.0031494788199779578,
      "epoch": 3.088655584718577,
      "grad_norm": 0.100646011531353,
      "learning_rate": 1e-06,
      "loss": -0.0087,
      "step": 595
    },
    {
      "clip_ratio/high_max": 0.0034824540598492604,
      "clip_ratio/high_mean": 0.0013763015685981372,
      "clip_ratio/low_mean": 0.0017292783086304553,
      "clip_ratio/low_min": 0.00014441743041970767,
      "clip_ratio/region_mean": 0.003105579802650027,
      "epoch": 3.0933216681248177,
      "grad_norm": 0.09727215021848679,
      "learning_rate": 1e-06,
      "loss": 0.0042,
      "step": 596
    },
    {
      "clip_ratio/high_max": 0.0036293155135354027,
      "clip_ratio/high_mean": 0.0013377864597714506,
      "clip_ratio/low_mean": 0.001752502575982362,
      "clip_ratio/low_min": 6.800896517233923e-05,
      "clip_ratio/region_mean": 0.0030902890430297703,
      "epoch": 3.0979877515310585,
      "grad_norm": 0.10720931738615036,
      "learning_rate": 1e-06,
      "loss": 0.0209,
      "step": 597
    },
    {
      "clip_ratio/high_max": 0.00363532191113336,
      "clip_ratio/high_mean": 0.0016436097466794308,
      "clip_ratio/low_mean": 0.0015543411936960183,
      "clip_ratio/low_min": 6.810110426158644e-05,
      "clip_ratio/region_mean": 0.003197950849425979,
      "epoch": 3.1026538349372994,
      "grad_norm": 0.1042325496673584,
      "learning_rate": 1e-06,
      "loss": -0.0149,
      "step": 598
    },
    {
      "clip_ratio/high_max": 0.004205332188575994,
      "clip_ratio/high_mean": 0.001850219203333836,
      "clip_ratio/low_mean": 0.0016532176814507693,
      "clip_ratio/low_min": 0.00016409991803811863,
      "clip_ratio/region_mean": 0.0035034368629567325,
      "epoch": 3.1073199183435403,
      "grad_norm": 0.1130886971950531,
      "learning_rate": 1e-06,
      "loss": -0.057,
      "step": 599
    },
    {
      "clip_ratio/high_max": 0.004173622306552716,
      "clip_ratio/high_mean": 0.00169884116621688,
      "clip_ratio/low_mean": 0.0019548626769392285,
      "clip_ratio/low_min": 0.00018240689678350464,
      "clip_ratio/region_mean": 0.003653703795862384,
      "epoch": 3.111986001749781,
      "grad_norm": 0.11387147754430771,
      "learning_rate": 1e-06,
      "loss": -0.0122,
      "step": 600
    },
    {
      "clip_ratio/high_max": 0.005090958613436669,
      "clip_ratio/high_mean": 0.001800794587325072,
      "clip_ratio/low_mean": 0.0015632430186087731,
      "clip_ratio/low_min": 6.127451342763379e-05,
      "clip_ratio/region_mean": 0.0033640376204857603,
      "epoch": 3.116652085156022,
      "grad_norm": 0.10405631363391876,
      "learning_rate": 1e-06,
      "loss": -0.0116,
      "step": 601
    },
    {
      "clip_ratio/high_max": 0.004095646385394502,
      "clip_ratio/high_mean": 0.001665675237745745,
      "clip_ratio/low_mean": 0.0021039583953097463,
      "clip_ratio/low_min": 0.00010477626165084075,
      "clip_ratio/region_mean": 0.0037696336294175126,
      "epoch": 3.121318168562263,
      "grad_norm": 0.10751927644014359,
      "learning_rate": 1e-06,
      "loss": 0.0048,
      "step": 602
    },
    {
      "clip_ratio/high_max": 0.003909106089849956,
      "clip_ratio/high_mean": 0.0016336360713467002,
      "clip_ratio/low_mean": 0.001738358405418694,
      "clip_ratio/low_min": 7.020500197540969e-05,
      "clip_ratio/region_mean": 0.003371994462213479,
      "epoch": 3.1259842519685037,
      "grad_norm": 0.1027611568570137,
      "learning_rate": 1e-06,
      "loss": -0.0236,
      "step": 603
    },
    {
      "clip_ratio/high_max": 0.00411385262850672,
      "clip_ratio/high_mean": 0.0016044451585912611,
      "clip_ratio/low_mean": 0.002101535770634655,
      "clip_ratio/low_min": 0.00020444583424250595,
      "clip_ratio/region_mean": 0.003705980983795598,
      "epoch": 3.130650335374745,
      "grad_norm": 0.1110607162117958,
      "learning_rate": 1e-06,
      "loss": 0.0486,
      "step": 604
    },
    {
      "clip_ratio/high_max": 0.004037739403429441,
      "clip_ratio/high_mean": 0.0017334331569145434,
      "clip_ratio/low_mean": 0.0014916374548192834,
      "clip_ratio/low_min": 0.00012129297465435229,
      "clip_ratio/region_mean": 0.003225070620828774,
      "epoch": 3.135316418780986,
      "grad_norm": 0.10643091797828674,
      "learning_rate": 1e-06,
      "loss": -0.0468,
      "step": 605
    },
    {
      "clip_ratio/high_max": 0.004383464198326692,
      "clip_ratio/high_mean": 0.001803305080102291,
      "clip_ratio/low_mean": 0.0015659096425224561,
      "clip_ratio/low_min": 1.4835034562565852e-05,
      "clip_ratio/region_mean": 0.0033692147189867683,
      "epoch": 3.1399825021872267,
      "grad_norm": 0.11684251576662064,
      "learning_rate": 1e-06,
      "loss": -0.0345,
      "step": 606
    },
    {
      "clip_ratio/high_max": 0.004195768618956208,
      "clip_ratio/high_mean": 0.0017371565772918984,
      "clip_ratio/low_mean": 0.001987223055039067,
      "clip_ratio/low_min": 0.0002798161658574827,
      "clip_ratio/region_mean": 0.003724379712366499,
      "epoch": 3.1446485855934676,
      "grad_norm": 0.11397506296634674,
      "learning_rate": 1e-06,
      "loss": -0.0011,
      "step": 607
    },
    {
      "clip_ratio/high_max": 0.004217505091219209,
      "clip_ratio/high_mean": 0.0016205803403863683,
      "clip_ratio/low_mean": 0.0016421983236796223,
      "clip_ratio/low_min": 6.37885314063169e-05,
      "clip_ratio/region_mean": 0.003262778678617906,
      "epoch": 3.1493146689997085,
      "grad_norm": 0.10415603220462799,
      "learning_rate": 1e-06,
      "loss": -0.0026,
      "step": 608
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0494559151785714,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4051.0,
      "completions/mean_length": 762.7545776367188,
      "completions/mean_terminated_length": 589.3289184570312,
      "completions/min_length": 90.0,
      "completions/min_terminated_length": 90.0,
      "epoch": 3.1539807524059493,
      "grad_norm": 0.13022832572460175,
      "learning_rate": 1e-06,
      "loss": -0.0051,
      "num_tokens": 186201039.0,
      "reward": 0.5954939126968384,
      "reward_std": 0.1724170297384262,
      "rewards/simpleverify_reward/mean": 0.5954938530921936,
      "rewards/simpleverify_reward/std": 0.49081334471702576,
      "step": 609
    },
    {
      "clip_ratio/high_max": 0.0019000004795088898,
      "clip_ratio/high_mean": 0.0007194360860012239,
      "clip_ratio/low_mean": 0.0005228891932347324,
      "clip_ratio/low_min": 1.2891914593637921e-05,
      "clip_ratio/region_mean": 0.001242325291968882,
      "epoch": 3.15864683581219,
      "grad_norm": 0.11680217832326889,
      "learning_rate": 1e-06,
      "loss": -0.0179,
      "step": 610
    },
    {
      "clip_ratio/high_max": 0.0022567846390302293,
      "clip_ratio/high_mean": 0.0009244008051609853,
      "clip_ratio/low_mean": 0.0004673724870372098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001391773283103248,
      "epoch": 3.163312919218431,
      "grad_norm": 0.1458456665277481,
      "learning_rate": 1e-06,
      "loss": -0.0447,
      "step": 611
    },
    {
      "clip_ratio/high_max": 0.0016577017377130687,
      "clip_ratio/high_mean": 0.0008093183132587001,
      "clip_ratio/low_mean": 0.0005865902840014314,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013959085808892269,
      "epoch": 3.167979002624672,
      "grad_norm": 0.1305447667837143,
      "learning_rate": 1e-06,
      "loss": -0.0085,
      "step": 612
    },
    {
      "clip_ratio/high_max": 0.0016875612163858023,
      "clip_ratio/high_mean": 0.0007692748695262708,
      "clip_ratio/low_mean": 0.0005504302171175368,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013197050830058288,
      "epoch": 3.1726450860309128,
      "grad_norm": 0.13962143659591675,
      "learning_rate": 1e-06,
      "loss": -0.0421,
      "step": 613
    },
    {
      "clip_ratio/high_max": 0.002256116677017417,
      "clip_ratio/high_mean": 0.0007989624200490653,
      "clip_ratio/low_mean": 0.0006242389827093575,
      "clip_ratio/low_min": 3.251244652346941e-05,
      "clip_ratio/region_mean": 0.001423201410943875,
      "epoch": 3.1773111694371536,
      "grad_norm": 0.13130025565624237,
      "learning_rate": 1e-06,
      "loss": -0.0506,
      "step": 614
    },
    {
      "clip_ratio/high_max": 0.001926960190758109,
      "clip_ratio/high_mean": 0.0007885514232839341,
      "clip_ratio/low_mean": 0.0006558587447216269,
      "clip_ratio/low_min": 1.0154346455237828e-05,
      "clip_ratio/region_mean": 0.001444410183466971,
      "epoch": 3.1819772528433945,
      "grad_norm": 0.12116069346666336,
      "learning_rate": 1e-06,
      "loss": -0.0044,
      "step": 615
    },
    {
      "clip_ratio/high_max": 0.0018085795381921344,
      "clip_ratio/high_mean": 0.0007075285757309757,
      "clip_ratio/low_mean": 0.0007380584811471635,
      "clip_ratio/low_min": 5.704748764401302e-05,
      "clip_ratio/region_mean": 0.0014455870550591499,
      "epoch": 3.1866433362496354,
      "grad_norm": 0.12010941654443741,
      "learning_rate": 1e-06,
      "loss": 0.0089,
      "step": 616
    },
    {
      "clip_ratio/high_max": 0.002363851781410631,
      "clip_ratio/high_mean": 0.0008664549350214656,
      "clip_ratio/low_mean": 0.0009849494599620812,
      "clip_ratio/low_min": 8.205053927667905e-05,
      "clip_ratio/region_mean": 0.0018514044058974832,
      "epoch": 3.1913094196558762,
      "grad_norm": 0.1326657384634018,
      "learning_rate": 1e-06,
      "loss": 0.0028,
      "step": 617
    },
    {
      "clip_ratio/high_max": 0.002366474516747985,
      "clip_ratio/high_mean": 0.000962843279921799,
      "clip_ratio/low_mean": 0.0008817527968858485,
      "clip_ratio/low_min": 3.0019211408216506e-05,
      "clip_ratio/region_mean": 0.001844596081355121,
      "epoch": 3.195975503062117,
      "grad_norm": 0.141387939453125,
      "learning_rate": 1e-06,
      "loss": -0.0082,
      "step": 618
    },
    {
      "clip_ratio/high_max": 0.002450492946081795,
      "clip_ratio/high_mean": 0.0008913672099879477,
      "clip_ratio/low_mean": 0.0009695914650365012,
      "clip_ratio/low_min": 3.617877428041538e-05,
      "clip_ratio/region_mean": 0.0018609586695674807,
      "epoch": 3.200641586468358,
      "grad_norm": 0.12486181408166885,
      "learning_rate": 1e-06,
      "loss": -0.0077,
      "step": 619
    },
    {
      "clip_ratio/high_max": 0.002591315336758271,
      "clip_ratio/high_mean": 0.001013109886116581,
      "clip_ratio/low_mean": 0.0009311078229075065,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019442177144810557,
      "epoch": 3.205307669874599,
      "grad_norm": 0.12602552771568298,
      "learning_rate": 1e-06,
      "loss": -0.0221,
      "step": 620
    },
    {
      "clip_ratio/high_max": 0.002283390895172488,
      "clip_ratio/high_mean": 0.000866814221808454,
      "clip_ratio/low_mean": 0.0008843828763929196,
      "clip_ratio/low_min": 6.866859803267289e-05,
      "clip_ratio/region_mean": 0.0017511970654595643,
      "epoch": 3.20997375328084,
      "grad_norm": 0.12433513253927231,
      "learning_rate": 1e-06,
      "loss": 0.0062,
      "step": 621
    },
    {
      "clip_ratio/high_max": 0.002088221055601025,
      "clip_ratio/high_mean": 0.0008685759203217458,
      "clip_ratio/low_mean": 0.0009978582966141403,
      "clip_ratio/low_min": 0.00010648265742929652,
      "clip_ratio/region_mean": 0.0018664341587282252,
      "epoch": 3.214639836687081,
      "grad_norm": 0.1304394006729126,
      "learning_rate": 1e-06,
      "loss": -0.0128,
      "step": 622
    },
    {
      "clip_ratio/high_max": 0.002259496206534095,
      "clip_ratio/high_mean": 0.000953817056142725,
      "clip_ratio/low_mean": 0.0008947519145294791,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018485689797671512,
      "epoch": 3.219305920093322,
      "grad_norm": 0.12420185655355453,
      "learning_rate": 1e-06,
      "loss": 0.0366,
      "step": 623
    },
    {
      "clip_ratio/high_max": 0.002494355743692722,
      "clip_ratio/high_mean": 0.0009324685634055641,
      "clip_ratio/low_mean": 0.0010561314247752307,
      "clip_ratio/low_min": 3.200204810127616e-05,
      "clip_ratio/region_mean": 0.001988599957257975,
      "epoch": 3.2239720034995627,
      "grad_norm": 0.14329595863819122,
      "learning_rate": 1e-06,
      "loss": 0.0012,
      "step": 624
    },
    {
      "clip_ratio/high_max": 0.004006963696156163,
      "clip_ratio/high_mean": 0.001710728040052345,
      "clip_ratio/low_mean": 0.0016071704230853356,
      "clip_ratio/low_min": 0.0001188762835226953,
      "clip_ratio/region_mean": 0.0033178984813275747,
      "epoch": 3.2286380869058036,
      "grad_norm": 0.10261663049459457,
      "learning_rate": 1e-06,
      "loss": -0.0059,
      "step": 625
    },
    {
      "clip_ratio/high_max": 0.003251519039622508,
      "clip_ratio/high_mean": 0.001407215306244325,
      "clip_ratio/low_mean": 0.001471388230129378,
      "clip_ratio/low_min": 6.833876796008553e-05,
      "clip_ratio/region_mean": 0.0028786034672521055,
      "epoch": 3.2333041703120444,
      "grad_norm": 0.10264348238706589,
      "learning_rate": 1e-06,
      "loss": -0.0186,
      "step": 626
    },
    {
      "clip_ratio/high_max": 0.004369752772618085,
      "clip_ratio/high_mean": 0.001973792695935117,
      "clip_ratio/low_mean": 0.0014771976602787618,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034509903198340908,
      "epoch": 3.2379702537182853,
      "grad_norm": 0.12080986052751541,
      "learning_rate": 1e-06,
      "loss": -0.0457,
      "step": 627
    },
    {
      "clip_ratio/high_max": 0.0036728047343785875,
      "clip_ratio/high_mean": 0.0016646357980789617,
      "clip_ratio/low_mean": 0.0016309115417243447,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00329554732888937,
      "epoch": 3.242636337124526,
      "grad_norm": 0.10374809056520462,
      "learning_rate": 1e-06,
      "loss": -0.0093,
      "step": 628
    },
    {
      "clip_ratio/high_max": 0.004053356067743152,
      "clip_ratio/high_mean": 0.0017904860578710213,
      "clip_ratio/low_mean": 0.0015496700107178185,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033401561086066067,
      "epoch": 3.247302420530767,
      "grad_norm": 0.11321540921926498,
      "learning_rate": 1e-06,
      "loss": -0.0431,
      "step": 629
    },
    {
      "clip_ratio/high_max": 0.004582806330290623,
      "clip_ratio/high_mean": 0.0016558061324758455,
      "clip_ratio/low_mean": 0.0014693157536385115,
      "clip_ratio/low_min": 0.00010000446127378382,
      "clip_ratio/region_mean": 0.003125121831544675,
      "epoch": 3.251968503937008,
      "grad_norm": 0.10362175107002258,
      "learning_rate": 1e-06,
      "loss": -0.0514,
      "step": 630
    },
    {
      "clip_ratio/high_max": 0.003786268789554015,
      "clip_ratio/high_mean": 0.0016751951952755917,
      "clip_ratio/low_mean": 0.001524067236459814,
      "clip_ratio/low_min": 3.046303754672408e-05,
      "clip_ratio/region_mean": 0.0031992623989935964,
      "epoch": 3.2566345873432487,
      "grad_norm": 0.09969361126422882,
      "learning_rate": 1e-06,
      "loss": -0.0051,
      "step": 631
    },
    {
      "clip_ratio/high_max": 0.0034105085287592374,
      "clip_ratio/high_mean": 0.0014426092493522447,
      "clip_ratio/low_mean": 0.0016897187815629877,
      "clip_ratio/low_min": 0.00019282766515971161,
      "clip_ratio/region_mean": 0.003132328056381084,
      "epoch": 3.2613006707494896,
      "grad_norm": 0.09795522689819336,
      "learning_rate": 1e-06,
      "loss": 0.0082,
      "step": 632
    },
    {
      "clip_ratio/high_max": 0.004529601872491185,
      "clip_ratio/high_mean": 0.0015576269252051134,
      "clip_ratio/low_mean": 0.0020816438482142985,
      "clip_ratio/low_min": 0.00024160535394912586,
      "clip_ratio/region_mean": 0.0036392707261256874,
      "epoch": 3.2659667541557305,
      "grad_norm": 0.10518714785575867,
      "learning_rate": 1e-06,
      "loss": 0.002,
      "step": 633
    },
    {
      "clip_ratio/high_max": 0.004199200586299412,
      "clip_ratio/high_mean": 0.0017250252967642155,
      "clip_ratio/low_mean": 0.0018690196302486584,
      "clip_ratio/low_min": 7.511227704526391e-05,
      "clip_ratio/region_mean": 0.003594044828787446,
      "epoch": 3.2706328375619713,
      "grad_norm": 0.10764548182487488,
      "learning_rate": 1e-06,
      "loss": -0.0091,
      "step": 634
    },
    {
      "clip_ratio/high_max": 0.00408559955394594,
      "clip_ratio/high_mean": 0.001564082154800417,
      "clip_ratio/low_mean": 0.0018179331382270902,
      "clip_ratio/low_min": 8.483331384923076e-05,
      "clip_ratio/region_mean": 0.003382015318493359,
      "epoch": 3.275298920968212,
      "grad_norm": 0.10013151168823242,
      "learning_rate": 1e-06,
      "loss": -0.0085,
      "step": 635
    },
    {
      "clip_ratio/high_max": 0.004208264799672179,
      "clip_ratio/high_mean": 0.0016985590955300722,
      "clip_ratio/low_mean": 0.0018666411215235712,
      "clip_ratio/low_min": 8.387854177271947e-05,
      "clip_ratio/region_mean": 0.00356520019704476,
      "epoch": 3.279965004374453,
      "grad_norm": 0.1021655797958374,
      "learning_rate": 1e-06,
      "loss": -0.0229,
      "step": 636
    },
    {
      "clip_ratio/high_max": 0.0040453202309436165,
      "clip_ratio/high_mean": 0.001501298051152844,
      "clip_ratio/low_mean": 0.0018029722741630394,
      "clip_ratio/low_min": 8.979059930425137e-05,
      "clip_ratio/region_mean": 0.003304270372609608,
      "epoch": 3.284631087780694,
      "grad_norm": 0.09687436372041702,
      "learning_rate": 1e-06,
      "loss": 0.0054,
      "step": 637
    },
    {
      "clip_ratio/high_max": 0.004028237104648724,
      "clip_ratio/high_mean": 0.0016927702417888213,
      "clip_ratio/low_mean": 0.0019515422736731125,
      "clip_ratio/low_min": 0.0002449123130645603,
      "clip_ratio/region_mean": 0.003644312593678478,
      "epoch": 3.289297171186935,
      "grad_norm": 0.10302521288394928,
      "learning_rate": 1e-06,
      "loss": -0.0136,
      "step": 638
    },
    {
      "clip_ratio/high_max": 0.0036676239251391962,
      "clip_ratio/high_mean": 0.0015250520373228937,
      "clip_ratio/low_mean": 0.0016797626303741708,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0032048146094894037,
      "epoch": 3.2939632545931756,
      "grad_norm": 0.09853555262088776,
      "learning_rate": 1e-06,
      "loss": 0.0359,
      "step": 639
    },
    {
      "clip_ratio/high_max": 0.004594740836182609,
      "clip_ratio/high_mean": 0.0017249968768737745,
      "clip_ratio/low_mean": 0.0019127596024191007,
      "clip_ratio/low_min": 3.200204810127616e-05,
      "clip_ratio/region_mean": 0.00363775640289532,
      "epoch": 3.298629337999417,
      "grad_norm": 0.11317408084869385,
      "learning_rate": 1e-06,
      "loss": 0.0003,
      "step": 640
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0474330357142857,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4082.0,
      "completions/mean_length": 726.4277954101562,
      "completions/mean_terminated_length": 558.6400146484375,
      "completions/min_length": 103.0,
      "completions/min_terminated_length": 103.0,
      "epoch": 3.303295421405658,
      "grad_norm": 0.14014358818531036,
      "learning_rate": 1e-06,
      "loss": -0.0393,
      "num_tokens": 195299795.0,
      "reward": 0.61767578125,
      "reward_std": 0.17741146683692932,
      "rewards/simpleverify_reward/mean": 0.61767578125,
      "rewards/simpleverify_reward/std": 0.4859721064567566,
      "step": 641
    },
    {
      "clip_ratio/high_max": 0.002066903663944686,
      "clip_ratio/high_mean": 0.000848774168844102,
      "clip_ratio/low_mean": 0.0005408977258412051,
      "clip_ratio/low_min": 1.4602804185415152e-05,
      "clip_ratio/region_mean": 0.0013896718883188441,
      "epoch": 3.3079615048118987,
      "grad_norm": 0.13495449721813202,
      "learning_rate": 1e-06,
      "loss": -0.0119,
      "step": 642
    },
    {
      "clip_ratio/high_max": 0.002042150324996328,
      "clip_ratio/high_mean": 0.0008760671298659872,
      "clip_ratio/low_mean": 0.0005293992844599416,
      "clip_ratio/low_min": 1.200076803797856e-05,
      "clip_ratio/region_mean": 0.0014054664097784553,
      "epoch": 3.3126275882181395,
      "grad_norm": 0.1311677098274231,
      "learning_rate": 1e-06,
      "loss": 0.0224,
      "step": 643
    },
    {
      "clip_ratio/high_max": 0.0022365580007317476,
      "clip_ratio/high_mean": 0.00092781108651252,
      "clip_ratio/low_mean": 0.0004887322784270509,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014165434149617795,
      "epoch": 3.3172936716243804,
      "grad_norm": 0.15040592849254608,
      "learning_rate": 1e-06,
      "loss": -0.0069,
      "step": 644
    },
    {
      "clip_ratio/high_max": 0.002318853665201459,
      "clip_ratio/high_mean": 0.0009354324811283732,
      "clip_ratio/low_mean": 0.0006117621805969975,
      "clip_ratio/low_min": 1.4272664884629194e-05,
      "clip_ratio/region_mean": 0.0015471946680918336,
      "epoch": 3.3219597550306212,
      "grad_norm": 0.15013548731803894,
      "learning_rate": 1e-06,
      "loss": -0.0075,
      "step": 645
    },
    {
      "clip_ratio/high_max": 0.0018589166793390177,
      "clip_ratio/high_mean": 0.0008079504314082442,
      "clip_ratio/low_mean": 0.0007287500648089917,
      "clip_ratio/low_min": 8.046938546613092e-05,
      "clip_ratio/region_mean": 0.0015367004998552147,
      "epoch": 3.326625838436862,
      "grad_norm": 0.13903145492076874,
      "learning_rate": 1e-06,
      "loss": 0.0002,
      "step": 646
    },
    {
      "clip_ratio/high_max": 0.0021955265001452062,
      "clip_ratio/high_mean": 0.0008579009081586264,
      "clip_ratio/low_mean": 0.0008154346633091336,
      "clip_ratio/low_min": 7.345266931224614e-05,
      "clip_ratio/region_mean": 0.0016733355587348342,
      "epoch": 3.331291921843103,
      "grad_norm": 0.1499493271112442,
      "learning_rate": 1e-06,
      "loss": 0.0529,
      "step": 647
    },
    {
      "clip_ratio/high_max": 0.0022960445130593143,
      "clip_ratio/high_mean": 0.000984575308393687,
      "clip_ratio/low_mean": 0.0006591099700017367,
      "clip_ratio/low_min": 1.3141295312379953e-05,
      "clip_ratio/region_mean": 0.00164368528930936,
      "epoch": 3.335958005249344,
      "grad_norm": 0.13286639750003815,
      "learning_rate": 1e-06,
      "loss": -0.0664,
      "step": 648
    },
    {
      "clip_ratio/high_max": 0.0024783850894891657,
      "clip_ratio/high_mean": 0.0010142752071260475,
      "clip_ratio/low_mean": 0.0009281232760258717,
      "clip_ratio/low_min": 6.058152575860731e-05,
      "clip_ratio/region_mean": 0.0019423984558670782,
      "epoch": 3.3406240886555847,
      "grad_norm": 0.1306142956018448,
      "learning_rate": 1e-06,
      "loss": 0.0138,
      "step": 649
    },
    {
      "clip_ratio/high_max": 0.0021204129516263492,
      "clip_ratio/high_mean": 0.0010155941927223466,
      "clip_ratio/low_mean": 0.0008665206842124462,
      "clip_ratio/low_min": 0.00011700736376951681,
      "clip_ratio/region_mean": 0.0018821148696588352,
      "epoch": 3.3452901720618256,
      "grad_norm": 0.13773073256015778,
      "learning_rate": 1e-06,
      "loss": -0.0292,
      "step": 650
    },
    {
      "clip_ratio/high_max": 0.002631053372169845,
      "clip_ratio/high_mean": 0.0011039887067454401,
      "clip_ratio/low_mean": 0.0009044942235050257,
      "clip_ratio/low_min": 8.548718324163929e-05,
      "clip_ratio/region_mean": 0.0020084829593542963,
      "epoch": 3.3499562554680664,
      "grad_norm": 0.1380700320005417,
      "learning_rate": 1e-06,
      "loss": -0.0206,
      "step": 651
    },
    {
      "clip_ratio/high_max": 0.0021688921624445356,
      "clip_ratio/high_mean": 0.000924098196264822,
      "clip_ratio/low_mean": 0.0010092142838402651,
      "clip_ratio/low_min": 4.021879067295231e-05,
      "clip_ratio/region_mean": 0.0019333124801050872,
      "epoch": 3.3546223388743073,
      "grad_norm": 0.1590724140405655,
      "learning_rate": 1e-06,
      "loss": 0.004,
      "step": 652
    },
    {
      "clip_ratio/high_max": 0.0027883377697435208,
      "clip_ratio/high_mean": 0.0011420621449360624,
      "clip_ratio/low_mean": 0.0009596440140740015,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.002101706115354318,
      "epoch": 3.359288422280548,
      "grad_norm": 0.146029531955719,
      "learning_rate": 1e-06,
      "loss": 0.0183,
      "step": 653
    },
    {
      "clip_ratio/high_max": 0.002291004937433172,
      "clip_ratio/high_mean": 0.0009532289805065375,
      "clip_ratio/low_mean": 0.0009386378169438103,
      "clip_ratio/low_min": 4.9195223255082965e-05,
      "clip_ratio/region_mean": 0.0018918667992693372,
      "epoch": 3.363954505686789,
      "grad_norm": 0.147115558385849,
      "learning_rate": 1e-06,
      "loss": -0.0084,
      "step": 654
    },
    {
      "clip_ratio/high_max": 0.00254554464481771,
      "clip_ratio/high_mean": 0.0010649877858668333,
      "clip_ratio/low_mean": 0.0009034327740664594,
      "clip_ratio/low_min": 1.7780939742806368e-05,
      "clip_ratio/region_mean": 0.0019684206054080278,
      "epoch": 3.36862058909303,
      "grad_norm": 0.14575709402561188,
      "learning_rate": 1e-06,
      "loss": -0.0317,
      "step": 655
    },
    {
      "clip_ratio/high_max": 0.0022353297972586006,
      "clip_ratio/high_mean": 0.0009694115087768296,
      "clip_ratio/low_mean": 0.0010076715643663192,
      "clip_ratio/low_min": 4.455798625713214e-05,
      "clip_ratio/region_mean": 0.0019770830549532548,
      "epoch": 3.3732866724992707,
      "grad_norm": 0.1381767988204956,
      "learning_rate": 1e-06,
      "loss": 0.0039,
      "step": 656
    },
    {
      "clip_ratio/high_max": 0.005044901394285262,
      "clip_ratio/high_mean": 0.0020702473557321355,
      "clip_ratio/low_mean": 0.0015968887137205456,
      "clip_ratio/low_min": 5.5203670854098164e-05,
      "clip_ratio/region_mean": 0.0036671360285254195,
      "epoch": 3.377952755905512,
      "grad_norm": 0.10805666446685791,
      "learning_rate": 1e-06,
      "loss": -0.0401,
      "step": 657
    },
    {
      "clip_ratio/high_max": 0.0041514738331898116,
      "clip_ratio/high_mean": 0.001632462299312465,
      "clip_ratio/low_mean": 0.0019125303733744659,
      "clip_ratio/low_min": 0.00019378033903194591,
      "clip_ratio/region_mean": 0.003544992723618634,
      "epoch": 3.382618839311753,
      "grad_norm": 0.11008191853761673,
      "learning_rate": 1e-06,
      "loss": -0.0127,
      "step": 658
    },
    {
      "clip_ratio/high_max": 0.003644034164608456,
      "clip_ratio/high_mean": 0.0014865095872664824,
      "clip_ratio/low_mean": 0.001745428398862714,
      "clip_ratio/low_min": 6.600791130040307e-05,
      "clip_ratio/region_mean": 0.0032319379679393023,
      "epoch": 3.3872849227179938,
      "grad_norm": 0.10289491713047028,
      "learning_rate": 1e-06,
      "loss": 0.0216,
      "step": 659
    },
    {
      "clip_ratio/high_max": 0.004477594993659295,
      "clip_ratio/high_mean": 0.0016738822268962394,
      "clip_ratio/low_mean": 0.001609667502634693,
      "clip_ratio/low_min": 0.00015565632565994747,
      "clip_ratio/region_mean": 0.0032835496531333774,
      "epoch": 3.3919510061242346,
      "grad_norm": 0.10740148276090622,
      "learning_rate": 1e-06,
      "loss": -0.0078,
      "step": 660
    },
    {
      "clip_ratio/high_max": 0.004334258119342849,
      "clip_ratio/high_mean": 0.0018854044683394022,
      "clip_ratio/low_mean": 0.0018751603001874173,
      "clip_ratio/low_min": 0.0002083036470139632,
      "clip_ratio/region_mean": 0.0037605647230520844,
      "epoch": 3.3966170895304755,
      "grad_norm": 0.11443742364645004,
      "learning_rate": 1e-06,
      "loss": -0.0084,
      "step": 661
    },
    {
      "clip_ratio/high_max": 0.00394822473754175,
      "clip_ratio/high_mean": 0.001523849739896832,
      "clip_ratio/low_mean": 0.002100512880133465,
      "clip_ratio/low_min": 0.00021312579519872088,
      "clip_ratio/region_mean": 0.0036243626673240215,
      "epoch": 3.4012831729367163,
      "grad_norm": 0.11889758706092834,
      "learning_rate": 1e-06,
      "loss": -0.0008,
      "step": 662
    },
    {
      "clip_ratio/high_max": 0.004420377197675407,
      "clip_ratio/high_mean": 0.0016236180381383747,
      "clip_ratio/low_mean": 0.0022603662764595356,
      "clip_ratio/low_min": 0.0002510506074031582,
      "clip_ratio/region_mean": 0.0038839843473397195,
      "epoch": 3.405949256342957,
      "grad_norm": 0.12119041383266449,
      "learning_rate": 1e-06,
      "loss": 0.0518,
      "step": 663
    },
    {
      "clip_ratio/high_max": 0.0045833216136088595,
      "clip_ratio/high_mean": 0.0019681698868225794,
      "clip_ratio/low_mean": 0.0016429882125521544,
      "clip_ratio/low_min": 3.5765546272159554e-05,
      "clip_ratio/region_mean": 0.0036111581575823948,
      "epoch": 3.410615339749198,
      "grad_norm": 0.1092284768819809,
      "learning_rate": 1e-06,
      "loss": -0.0673,
      "step": 664
    },
    {
      "clip_ratio/high_max": 0.003909898172423709,
      "clip_ratio/high_mean": 0.0016008672064344864,
      "clip_ratio/low_mean": 0.001996095677895937,
      "clip_ratio/low_min": 0.00024429691620753147,
      "clip_ratio/region_mean": 0.0035969629389001057,
      "epoch": 3.415281423155439,
      "grad_norm": 0.11005459725856781,
      "learning_rate": 1e-06,
      "loss": 0.013,
      "step": 665
    },
    {
      "clip_ratio/high_max": 0.003609065810451284,
      "clip_ratio/high_mean": 0.0018118260086339433,
      "clip_ratio/low_mean": 0.001869842908490682,
      "clip_ratio/low_min": 0.00023208119455375709,
      "clip_ratio/region_mean": 0.003681668938952498,
      "epoch": 3.41994750656168,
      "grad_norm": 0.11277961730957031,
      "learning_rate": 1e-06,
      "loss": -0.0301,
      "step": 666
    },
    {
      "clip_ratio/high_max": 0.00498825368413236,
      "clip_ratio/high_mean": 0.0020805437670787796,
      "clip_ratio/low_mean": 0.001959514291229425,
      "clip_ratio/low_min": 0.00035709681105799973,
      "clip_ratio/region_mean": 0.0040400580619461834,
      "epoch": 3.4246135899679206,
      "grad_norm": 0.11262824386358261,
      "learning_rate": 1e-06,
      "loss": -0.0216,
      "step": 667
    },
    {
      "clip_ratio/high_max": 0.004037264792714268,
      "clip_ratio/high_mean": 0.0016997149759845342,
      "clip_ratio/low_mean": 0.0018412630597595125,
      "clip_ratio/low_min": 0.00029850217470084317,
      "clip_ratio/region_mean": 0.0035409779811743647,
      "epoch": 3.4292796733741615,
      "grad_norm": 0.10545282810926437,
      "learning_rate": 1e-06,
      "loss": 0.0031,
      "step": 668
    },
    {
      "clip_ratio/high_max": 0.0048037907254183665,
      "clip_ratio/high_mean": 0.0020303584351495374,
      "clip_ratio/low_mean": 0.0019416738614381757,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003972032252931967,
      "epoch": 3.4339457567804024,
      "grad_norm": 0.10566996037960052,
      "learning_rate": 1e-06,
      "loss": 0.0174,
      "step": 669
    },
    {
      "clip_ratio/high_max": 0.004213619977235794,
      "clip_ratio/high_mean": 0.001741656604281161,
      "clip_ratio/low_mean": 0.0019449365645414218,
      "clip_ratio/low_min": 0.00013560805746237747,
      "clip_ratio/region_mean": 0.003686593263410032,
      "epoch": 3.4386118401866432,
      "grad_norm": 0.12119334936141968,
      "learning_rate": 1e-06,
      "loss": -0.0095,
      "step": 670
    },
    {
      "clip_ratio/high_max": 0.004940413469739724,
      "clip_ratio/high_mean": 0.0019349591239006259,
      "clip_ratio/low_mean": 0.001774188542185584,
      "clip_ratio/low_min": 0.0001243501501448918,
      "clip_ratio/region_mean": 0.003709147684276104,
      "epoch": 3.443277923592884,
      "grad_norm": 0.11867986619472504,
      "learning_rate": 1e-06,
      "loss": -0.0327,
      "step": 671
    },
    {
      "clip_ratio/high_max": 0.004274389721103944,
      "clip_ratio/high_mean": 0.0015988001032383181,
      "clip_ratio/low_mean": 0.0018123163608834147,
      "clip_ratio/low_min": 0.0001417928287992254,
      "clip_ratio/region_mean": 0.003411116515053436,
      "epoch": 3.447944006999125,
      "grad_norm": 0.1104457825422287,
      "learning_rate": 1e-06,
      "loss": 0.003,
      "step": 672
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.051967075892857095,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 3944.0,
      "completions/mean_length": 735.8509521484375,
      "completions/mean_terminated_length": 551.6620483398438,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 3.452610090405366,
      "grad_norm": 0.17222283780574799,
      "learning_rate": 1e-06,
      "loss": -0.0115,
      "num_tokens": 204250578.0,
      "reward": 0.6261160969734192,
      "reward_std": 0.18874570727348328,
      "rewards/simpleverify_reward/mean": 0.6261160969734192,
      "rewards/simpleverify_reward/std": 0.48385027050971985,
      "step": 673
    },
    {
      "clip_ratio/high_max": 0.001980857821763493,
      "clip_ratio/high_mean": 0.0008756244969845284,
      "clip_ratio/low_mean": 0.0006200690604600823,
      "clip_ratio/low_min": 4.0948902096715756e-05,
      "clip_ratio/region_mean": 0.0014956935483496636,
      "epoch": 3.457276173811607,
      "grad_norm": 0.16398774087429047,
      "learning_rate": 1e-06,
      "loss": -0.0162,
      "step": 674
    },
    {
      "clip_ratio/high_max": 0.0025196804563165642,
      "clip_ratio/high_mean": 0.0009564410756865982,
      "clip_ratio/low_mean": 0.0005882041787117487,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015446452744072303,
      "epoch": 3.4619422572178475,
      "grad_norm": 0.1514146476984024,
      "learning_rate": 1e-06,
      "loss": -0.0373,
      "step": 675
    },
    {
      "clip_ratio/high_max": 0.0019539371314749587,
      "clip_ratio/high_mean": 0.0008573754857934546,
      "clip_ratio/low_mean": 0.0007221938067232259,
      "clip_ratio/low_min": 1.3278096048452426e-05,
      "clip_ratio/region_mean": 0.0015795693107065745,
      "epoch": 3.466608340624089,
      "grad_norm": 0.13135387003421783,
      "learning_rate": 1e-06,
      "loss": -0.0114,
      "step": 676
    },
    {
      "clip_ratio/high_max": 0.0029063174151815474,
      "clip_ratio/high_mean": 0.001123106456361711,
      "clip_ratio/low_mean": 0.0007890469278208911,
      "clip_ratio/low_min": 7.141552760003833e-05,
      "clip_ratio/region_mean": 0.0019121534278383479,
      "epoch": 3.4712744240303297,
      "grad_norm": 0.16005004942417145,
      "learning_rate": 1e-06,
      "loss": -0.005,
      "step": 677
    },
    {
      "clip_ratio/high_max": 0.002180146570026409,
      "clip_ratio/high_mean": 0.0009474923335801577,
      "clip_ratio/low_mean": 0.0008806268087937497,
      "clip_ratio/low_min": 4.369026646600105e-05,
      "clip_ratio/region_mean": 0.0018281191150890663,
      "epoch": 3.4759405074365706,
      "grad_norm": 0.14573457837104797,
      "learning_rate": 1e-06,
      "loss": -0.0089,
      "step": 678
    },
    {
      "clip_ratio/high_max": 0.002764007069345098,
      "clip_ratio/high_mean": 0.0010274022351950407,
      "clip_ratio/low_mean": 0.0008278212535515195,
      "clip_ratio/low_min": 9.704004332888871e-05,
      "clip_ratio/region_mean": 0.0018552234905655496,
      "epoch": 3.4806065908428114,
      "grad_norm": 0.1608944535255432,
      "learning_rate": 1e-06,
      "loss": -0.0167,
      "step": 679
    },
    {
      "clip_ratio/high_max": 0.0024787839938653633,
      "clip_ratio/high_mean": 0.0011237841135880444,
      "clip_ratio/low_mean": 0.0006414969884644961,
      "clip_ratio/low_min": 1.306438116444042e-05,
      "clip_ratio/region_mean": 0.0017652811075095087,
      "epoch": 3.4852726742490523,
      "grad_norm": 0.14125318825244904,
      "learning_rate": 1e-06,
      "loss": -0.0583,
      "step": 680
    },
    {
      "clip_ratio/high_max": 0.0020170112620689906,
      "clip_ratio/high_mean": 0.0008679190195834963,
      "clip_ratio/low_mean": 0.0009032525049406104,
      "clip_ratio/low_min": 3.626025045377901e-05,
      "clip_ratio/region_mean": 0.001771171526343096,
      "epoch": 3.489938757655293,
      "grad_norm": 0.15285366773605347,
      "learning_rate": 1e-06,
      "loss": 0.0259,
      "step": 681
    },
    {
      "clip_ratio/high_max": 0.0027065124959335662,
      "clip_ratio/high_mean": 0.001050167626090115,
      "clip_ratio/low_mean": 0.0007910495078249369,
      "clip_ratio/low_min": 3.331556581542827e-05,
      "clip_ratio/region_mean": 0.0018412171375530306,
      "epoch": 3.494604841061534,
      "grad_norm": 0.14905956387519836,
      "learning_rate": 1e-06,
      "loss": -0.0171,
      "step": 682
    },
    {
      "clip_ratio/high_max": 0.0028013944829581305,
      "clip_ratio/high_mean": 0.0011154164167237468,
      "clip_ratio/low_mean": 0.0009514720532024512,
      "clip_ratio/low_min": 6.775842302886304e-05,
      "clip_ratio/region_mean": 0.0020668884317274205,
      "epoch": 3.499270924467775,
      "grad_norm": 0.1458147019147873,
      "learning_rate": 1e-06,
      "loss": -0.0258,
      "step": 683
    },
    {
      "clip_ratio/high_max": 0.0025980886930483393,
      "clip_ratio/high_mean": 0.0012200003511679824,
      "clip_ratio/low_mean": 0.0010374091580160893,
      "clip_ratio/low_min": 5.84668205192429e-05,
      "clip_ratio/region_mean": 0.0022574094764422625,
      "epoch": 3.5039370078740157,
      "grad_norm": 0.13864310085773468,
      "learning_rate": 1e-06,
      "loss": -0.0204,
      "step": 684
    },
    {
      "clip_ratio/high_max": 0.002224741372629069,
      "clip_ratio/high_mean": 0.000967970525380224,
      "clip_ratio/low_mean": 0.0009146253260041703,
      "clip_ratio/low_min": 5.269441317068413e-05,
      "clip_ratio/region_mean": 0.0018825958541128784,
      "epoch": 3.5086030912802566,
      "grad_norm": 0.14988622069358826,
      "learning_rate": 1e-06,
      "loss": 0.0141,
      "step": 685
    },
    {
      "clip_ratio/high_max": 0.002817836204485502,
      "clip_ratio/high_mean": 0.0011961388008785434,
      "clip_ratio/low_mean": 0.0011763399579649558,
      "clip_ratio/low_min": 0.00013196072632126743,
      "clip_ratio/region_mean": 0.0023724787679384463,
      "epoch": 3.5132691746864975,
      "grad_norm": 0.14847752451896667,
      "learning_rate": 1e-06,
      "loss": -0.0221,
      "step": 686
    },
    {
      "clip_ratio/high_max": 0.0022620249874307774,
      "clip_ratio/high_mean": 0.00096858550386969,
      "clip_ratio/low_mean": 0.0007860835721658077,
      "clip_ratio/low_min": 2.2119978893897496e-05,
      "clip_ratio/region_mean": 0.0017546690651215613,
      "epoch": 3.5179352580927383,
      "grad_norm": 0.1235848218202591,
      "learning_rate": 1e-06,
      "loss": -0.0494,
      "step": 687
    },
    {
      "clip_ratio/high_max": 0.0031135671088122763,
      "clip_ratio/high_mean": 0.0011754685547202826,
      "clip_ratio/low_mean": 0.0009287907469115453,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0021042592998128384,
      "epoch": 3.522601341498979,
      "grad_norm": 0.13883580267429352,
      "learning_rate": 1e-06,
      "loss": -0.0316,
      "step": 688
    },
    {
      "clip_ratio/high_max": 0.005382393283070996,
      "clip_ratio/high_mean": 0.0020826052932534367,
      "clip_ratio/low_mean": 0.0018338317822781391,
      "clip_ratio/low_min": 0.00018301716954738367,
      "clip_ratio/region_mean": 0.0039164370828075334,
      "epoch": 3.52726742490522,
      "grad_norm": 0.11869249492883682,
      "learning_rate": 1e-06,
      "loss": -0.0125,
      "step": 689
    },
    {
      "clip_ratio/high_max": 0.0048578141577309,
      "clip_ratio/high_mean": 0.002021719170443248,
      "clip_ratio/low_mean": 0.002312681608600542,
      "clip_ratio/low_min": 6.89756761858007e-05,
      "clip_ratio/region_mean": 0.004334400757215917,
      "epoch": 3.531933508311461,
      "grad_norm": 0.1230340451002121,
      "learning_rate": 1e-06,
      "loss": -0.0174,
      "step": 690
    },
    {
      "clip_ratio/high_max": 0.005167898998479359,
      "clip_ratio/high_mean": 0.0019759668575716205,
      "clip_ratio/low_mean": 0.0020079830937902443,
      "clip_ratio/low_min": 5.4811773225083016e-05,
      "clip_ratio/region_mean": 0.003983949951361865,
      "epoch": 3.536599591717702,
      "grad_norm": 0.11668127030134201,
      "learning_rate": 1e-06,
      "loss": -0.0384,
      "step": 691
    },
    {
      "clip_ratio/high_max": 0.004001097360742278,
      "clip_ratio/high_mean": 0.0017204966279678047,
      "clip_ratio/low_mean": 0.002034656143223401,
      "clip_ratio/low_min": 0.00013382639008341357,
      "clip_ratio/region_mean": 0.0037551527348114178,
      "epoch": 3.5412656751239426,
      "grad_norm": 0.10663443803787231,
      "learning_rate": 1e-06,
      "loss": -0.0123,
      "step": 692
    },
    {
      "clip_ratio/high_max": 0.004988473068806343,
      "clip_ratio/high_mean": 0.002163104945793748,
      "clip_ratio/low_mean": 0.0022495699595310725,
      "clip_ratio/low_min": 0.00022599931253353134,
      "clip_ratio/region_mean": 0.004412674781633541,
      "epoch": 3.545931758530184,
      "grad_norm": 14.693273544311523,
      "learning_rate": 1e-06,
      "loss": -0.0056,
      "step": 693
    },
    {
      "clip_ratio/high_max": 0.00434461300028488,
      "clip_ratio/high_mean": 0.0017873449978651479,
      "clip_ratio/low_mean": 0.002183643708121963,
      "clip_ratio/low_min": 0.0001980489178095013,
      "clip_ratio/region_mean": 0.003970988735090941,
      "epoch": 3.5505978419364244,
      "grad_norm": 0.11578813195228577,
      "learning_rate": 1e-06,
      "loss": -0.0098,
      "step": 694
    },
    {
      "clip_ratio/high_max": 0.005354479304514825,
      "clip_ratio/high_mean": 0.0020086369186174124,
      "clip_ratio/low_mean": 0.002013766898016911,
      "clip_ratio/low_min": 0.00017393503731000237,
      "clip_ratio/region_mean": 0.004022403809358366,
      "epoch": 3.5552639253426657,
      "grad_norm": 0.1182750016450882,
      "learning_rate": 1e-06,
      "loss": -0.0178,
      "step": 695
    },
    {
      "clip_ratio/high_max": 0.005097582499729469,
      "clip_ratio/high_mean": 0.0022468328097602352,
      "clip_ratio/low_mean": 0.001675467807217501,
      "clip_ratio/low_min": 1.628028076083865e-05,
      "clip_ratio/region_mean": 0.003922300529666245,
      "epoch": 3.5599300087489065,
      "grad_norm": 0.161624476313591,
      "learning_rate": 1e-06,
      "loss": -0.059,
      "step": 696
    },
    {
      "clip_ratio/high_max": 0.003914291868568398,
      "clip_ratio/high_mean": 0.0017340208778477972,
      "clip_ratio/low_mean": 0.002251035850349581,
      "clip_ratio/low_min": 0.00014183842358761467,
      "clip_ratio/region_mean": 0.00398505678458605,
      "epoch": 3.5645960921551474,
      "grad_norm": 0.11544261127710342,
      "learning_rate": 1e-06,
      "loss": 0.0249,
      "step": 697
    },
    {
      "clip_ratio/high_max": 0.00518498920428101,
      "clip_ratio/high_mean": 0.0020181063227937557,
      "clip_ratio/low_mean": 0.0020307432969275396,
      "clip_ratio/low_min": 0.00014992004435043782,
      "clip_ratio/region_mean": 0.004048849550599698,
      "epoch": 3.5692621755613883,
      "grad_norm": 0.1165173128247261,
      "learning_rate": 1e-06,
      "loss": -0.0181,
      "step": 698
    },
    {
      "clip_ratio/high_max": 0.004927886839141138,
      "clip_ratio/high_mean": 0.00218375600525178,
      "clip_ratio/low_mean": 0.0022608401050092652,
      "clip_ratio/low_min": 0.00014047945114725735,
      "clip_ratio/region_mean": 0.004444596197572537,
      "epoch": 3.573928258967629,
      "grad_norm": 0.11463657766580582,
      "learning_rate": 1e-06,
      "loss": -0.0268,
      "step": 699
    },
    {
      "clip_ratio/high_max": 0.005064605502411723,
      "clip_ratio/high_mean": 0.0022345846955431625,
      "clip_ratio/low_mean": 0.002261064939375501,
      "clip_ratio/low_min": 0.00012846695790358353,
      "clip_ratio/region_mean": 0.00449564955488313,
      "epoch": 3.57859434237387,
      "grad_norm": 0.12003596127033234,
      "learning_rate": 1e-06,
      "loss": -0.0212,
      "step": 700
    },
    {
      "clip_ratio/high_max": 0.00493295572232455,
      "clip_ratio/high_mean": 0.002002900408115238,
      "clip_ratio/low_mean": 0.0022291371333267307,
      "clip_ratio/low_min": 5.5010803407640196e-05,
      "clip_ratio/region_mean": 0.004232037594192661,
      "epoch": 3.583260425780111,
      "grad_norm": 0.1145007386803627,
      "learning_rate": 1e-06,
      "loss": 0.013,
      "step": 701
    },
    {
      "clip_ratio/high_max": 0.005295188268064521,
      "clip_ratio/high_mean": 0.0022089182639319915,
      "clip_ratio/low_mean": 0.0024476660655636806,
      "clip_ratio/low_min": 0.00035798074532067403,
      "clip_ratio/region_mean": 0.004656584424083121,
      "epoch": 3.5879265091863517,
      "grad_norm": 0.11989594250917435,
      "learning_rate": 1e-06,
      "loss": -0.0231,
      "step": 702
    },
    {
      "clip_ratio/high_max": 0.004397689423058182,
      "clip_ratio/high_mean": 0.0018595462242956273,
      "clip_ratio/low_mean": 0.0016044462136051152,
      "clip_ratio/low_min": 7.313091009564232e-05,
      "clip_ratio/region_mean": 0.003463992426986806,
      "epoch": 3.5925925925925926,
      "grad_norm": 0.11015208065509796,
      "learning_rate": 1e-06,
      "loss": -0.0501,
      "step": 703
    },
    {
      "clip_ratio/high_max": 0.00532678839226719,
      "clip_ratio/high_mean": 0.002088845882099122,
      "clip_ratio/low_mean": 0.002012509139603935,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.004101354934391566,
      "epoch": 3.5972586759988334,
      "grad_norm": 0.11353772133588791,
      "learning_rate": 1e-06,
      "loss": -0.0325,
      "step": 704
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.049037388392857095,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4058.0,
      "completions/mean_length": 732.408935546875,
      "completions/mean_terminated_length": 558.9617919921875,
      "completions/min_length": 109.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 3.6019247594050743,
      "grad_norm": 0.13796812295913696,
      "learning_rate": 1e-06,
      "loss": -0.0226,
      "num_tokens": 213268688.0,
      "reward": 0.6263253688812256,
      "reward_std": 0.17682276666164398,
      "rewards/simpleverify_reward/mean": 0.6263253092765808,
      "rewards/simpleverify_reward/std": 0.4837956726551056,
      "step": 705
    },
    {
      "clip_ratio/high_max": 0.001707718682155246,
      "clip_ratio/high_mean": 0.0007608414143760456,
      "clip_ratio/low_mean": 0.0005688443379767705,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013296857287059538,
      "epoch": 3.606590842811315,
      "grad_norm": 0.13988927006721497,
      "learning_rate": 1e-06,
      "loss": -0.043,
      "step": 706
    },
    {
      "clip_ratio/high_max": 0.0017335513039142825,
      "clip_ratio/high_mean": 0.0006721417103108251,
      "clip_ratio/low_mean": 0.0006290203837124864,
      "clip_ratio/low_min": 2.7062134904554114e-05,
      "clip_ratio/region_mean": 0.0013011620903853327,
      "epoch": 3.611256926217556,
      "grad_norm": 0.1380269080400467,
      "learning_rate": 1e-06,
      "loss": 0.0052,
      "step": 707
    },
    {
      "clip_ratio/high_max": 0.0017275933714699931,
      "clip_ratio/high_mean": 0.0008268210658570752,
      "clip_ratio/low_mean": 0.0006764011086488608,
      "clip_ratio/low_min": 1.4912908227415755e-05,
      "clip_ratio/region_mean": 0.0015032221417641267,
      "epoch": 3.615923009623797,
      "grad_norm": 0.13663262128829956,
      "learning_rate": 1e-06,
      "loss": -0.0011,
      "step": 708
    },
    {
      "clip_ratio/high_max": 0.0021695724135497585,
      "clip_ratio/high_mean": 0.000924284162465483,
      "clip_ratio/low_mean": 0.000647056125671952,
      "clip_ratio/low_min": 1.771792994986754e-05,
      "clip_ratio/region_mean": 0.0015713403045083396,
      "epoch": 3.6205890930300377,
      "grad_norm": 0.1439342051744461,
      "learning_rate": 1e-06,
      "loss": 0.0254,
      "step": 709
    },
    {
      "clip_ratio/high_max": 0.001976675044716103,
      "clip_ratio/high_mean": 0.0008985688309621764,
      "clip_ratio/low_mean": 0.0006537300268973922,
      "clip_ratio/low_min": 2.824326656991616e-05,
      "clip_ratio/region_mean": 0.0015522988360316958,
      "epoch": 3.625255176436279,
      "grad_norm": 0.1437269002199173,
      "learning_rate": 1e-06,
      "loss": -0.0648,
      "step": 710
    },
    {
      "clip_ratio/high_max": 0.0021579306994681247,
      "clip_ratio/high_mean": 0.0008526164310751483,
      "clip_ratio/low_mean": 0.0006334783101920038,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001486094726715237,
      "epoch": 3.6299212598425195,
      "grad_norm": 0.1405710130929947,
      "learning_rate": 1e-06,
      "loss": 0.0011,
      "step": 711
    },
    {
      "clip_ratio/high_max": 0.001778824458597228,
      "clip_ratio/high_mean": 0.0007698830795561662,
      "clip_ratio/low_mean": 0.0007373526987066725,
      "clip_ratio/low_min": 3.947056393371895e-05,
      "clip_ratio/region_mean": 0.0015072357964527328,
      "epoch": 3.6345873432487608,
      "grad_norm": 0.1362738460302353,
      "learning_rate": 1e-06,
      "loss": 0.0148,
      "step": 712
    },
    {
      "clip_ratio/high_max": 0.0019354814721737057,
      "clip_ratio/high_mean": 0.000855296657391591,
      "clip_ratio/low_mean": 0.0007056312952045118,
      "clip_ratio/low_min": 3.554165596142411e-05,
      "clip_ratio/region_mean": 0.001560927979880944,
      "epoch": 3.6392534266550016,
      "grad_norm": 0.13625776767730713,
      "learning_rate": 1e-06,
      "loss": -0.0527,
      "step": 713
    },
    {
      "clip_ratio/high_max": 0.0020584117446560413,
      "clip_ratio/high_mean": 0.0008291475460282527,
      "clip_ratio/low_mean": 0.0008382462865483831,
      "clip_ratio/low_min": 2.7179821699974127e-05,
      "clip_ratio/region_mean": 0.0016673938189342152,
      "epoch": 3.6439195100612425,
      "grad_norm": 0.1302632838487625,
      "learning_rate": 1e-06,
      "loss": 0.0012,
      "step": 714
    },
    {
      "clip_ratio/high_max": 0.0022871950786793604,
      "clip_ratio/high_mean": 0.0009641977485443931,
      "clip_ratio/low_mean": 0.0009316934483649675,
      "clip_ratio/low_min": 3.747002483578399e-05,
      "clip_ratio/region_mean": 0.0018958911459776573,
      "epoch": 3.6485855934674833,
      "grad_norm": 0.1516473889350891,
      "learning_rate": 1e-06,
      "loss": -0.0079,
      "step": 715
    },
    {
      "clip_ratio/high_max": 0.002037832535279449,
      "clip_ratio/high_mean": 0.0008557994042348582,
      "clip_ratio/low_mean": 0.0008380950330320047,
      "clip_ratio/low_min": 1.6267569662886672e-05,
      "clip_ratio/region_mean": 0.00169389444636181,
      "epoch": 3.653251676873724,
      "grad_norm": 0.14155632257461548,
      "learning_rate": 1e-06,
      "loss": -0.0375,
      "step": 716
    },
    {
      "clip_ratio/high_max": 0.0021008499279560056,
      "clip_ratio/high_mean": 0.0009644745805417188,
      "clip_ratio/low_mean": 0.0008633770030428423,
      "clip_ratio/low_min": 2.870044409064576e-05,
      "clip_ratio/region_mean": 0.0018278515781275928,
      "epoch": 3.657917760279965,
      "grad_norm": 0.14588196575641632,
      "learning_rate": 1e-06,
      "loss": -0.0756,
      "step": 717
    },
    {
      "clip_ratio/high_max": 0.0020923038537148386,
      "clip_ratio/high_mean": 0.0009377236256113974,
      "clip_ratio/low_mean": 0.0008512311378581217,
      "clip_ratio/low_min": 7.207609451143071e-05,
      "clip_ratio/region_mean": 0.0017889547816594131,
      "epoch": 3.662583843686206,
      "grad_norm": 0.12750868499279022,
      "learning_rate": 1e-06,
      "loss": -0.0528,
      "step": 718
    },
    {
      "clip_ratio/high_max": 0.0023973630159161985,
      "clip_ratio/high_mean": 0.0010299165951437317,
      "clip_ratio/low_mean": 0.0011046579948015278,
      "clip_ratio/low_min": 6.793599095544778e-05,
      "clip_ratio/region_mean": 0.0021345745408325456,
      "epoch": 3.667249927092447,
      "grad_norm": 0.15035173296928406,
      "learning_rate": 1e-06,
      "loss": 0.0367,
      "step": 719
    },
    {
      "clip_ratio/high_max": 0.002176510948629584,
      "clip_ratio/high_mean": 0.0009333652178611374,
      "clip_ratio/low_mean": 0.0010895014584093587,
      "clip_ratio/low_min": 4.154180714976974e-05,
      "clip_ratio/region_mean": 0.002022866661718581,
      "epoch": 3.6719160104986877,
      "grad_norm": 0.14976878464221954,
      "learning_rate": 1e-06,
      "loss": 0.0137,
      "step": 720
    },
    {
      "clip_ratio/high_max": 0.0043633645691443235,
      "clip_ratio/high_mean": 0.0016300785264320439,
      "clip_ratio/low_mean": 0.0014974916593928356,
      "clip_ratio/low_min": 7.270823698490858e-05,
      "clip_ratio/region_mean": 0.003127570220385678,
      "epoch": 3.6765820939049285,
      "grad_norm": 0.09291525930166245,
      "learning_rate": 1e-06,
      "loss": -0.0233,
      "step": 721
    },
    {
      "clip_ratio/high_max": 0.004633081814972684,
      "clip_ratio/high_mean": 0.0018503991814213805,
      "clip_ratio/low_mean": 0.0015906114531389903,
      "clip_ratio/low_min": 0.00016880440580280265,
      "clip_ratio/region_mean": 0.003441010703681968,
      "epoch": 3.6812481773111694,
      "grad_norm": 0.10746114701032639,
      "learning_rate": 1e-06,
      "loss": -0.0439,
      "step": 722
    },
    {
      "clip_ratio/high_max": 0.0035965180722996593,
      "clip_ratio/high_mean": 0.0014306397024483886,
      "clip_ratio/low_mean": 0.0018339150337851606,
      "clip_ratio/low_min": 8.446964966424275e-05,
      "clip_ratio/region_mean": 0.003264554761699401,
      "epoch": 3.6859142607174102,
      "grad_norm": 0.1101069375872612,
      "learning_rate": 1e-06,
      "loss": 0.0043,
      "step": 723
    },
    {
      "clip_ratio/high_max": 0.003934079228201881,
      "clip_ratio/high_mean": 0.001699183452728903,
      "clip_ratio/low_mean": 0.0018036453984677792,
      "clip_ratio/low_min": 0.0001386386156809749,
      "clip_ratio/region_mean": 0.0035028288402827457,
      "epoch": 3.690580344123651,
      "grad_norm": 0.10536739230155945,
      "learning_rate": 1e-06,
      "loss": -0.002,
      "step": 724
    },
    {
      "clip_ratio/high_max": 0.004065997738507576,
      "clip_ratio/high_mean": 0.001666085132455919,
      "clip_ratio/low_mean": 0.0019013668497791514,
      "clip_ratio/low_min": 0.0002288434116053395,
      "clip_ratio/region_mean": 0.0035674519749591127,
      "epoch": 3.695246427529892,
      "grad_norm": 0.11310018599033356,
      "learning_rate": 1e-06,
      "loss": 0.0245,
      "step": 725
    },
    {
      "clip_ratio/high_max": 0.0047646881866967306,
      "clip_ratio/high_mean": 0.002020320054725744,
      "clip_ratio/low_mean": 0.0017728058555803727,
      "clip_ratio/low_min": 9.492444587522186e-05,
      "clip_ratio/region_mean": 0.0037931259139440954,
      "epoch": 3.699912510936133,
      "grad_norm": 0.11736725270748138,
      "learning_rate": 1e-06,
      "loss": -0.0658,
      "step": 726
    },
    {
      "clip_ratio/high_max": 0.004748114384710789,
      "clip_ratio/high_mean": 0.001793054208974354,
      "clip_ratio/low_mean": 0.0017506387230241671,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003543693048413843,
      "epoch": 3.704578594342374,
      "grad_norm": 0.11295180022716522,
      "learning_rate": 1e-06,
      "loss": 0.0002,
      "step": 727
    },
    {
      "clip_ratio/high_max": 0.003740929299965501,
      "clip_ratio/high_mean": 0.0014832897886662977,
      "clip_ratio/low_mean": 0.001872063810878899,
      "clip_ratio/low_min": 0.00022408315999200568,
      "clip_ratio/region_mean": 0.0033553535904502496,
      "epoch": 3.7092446777486145,
      "grad_norm": 0.10654005408287048,
      "learning_rate": 1e-06,
      "loss": 0.0139,
      "step": 728
    },
    {
      "clip_ratio/high_max": 0.004244787895004265,
      "clip_ratio/high_mean": 0.001821457306505181,
      "clip_ratio/low_mean": 0.0015946164749038871,
      "clip_ratio/low_min": 6.794709042878821e-05,
      "clip_ratio/region_mean": 0.0034160738141508773,
      "epoch": 3.713910761154856,
      "grad_norm": 0.11029591411352158,
      "learning_rate": 1e-06,
      "loss": -0.0535,
      "step": 729
    },
    {
      "clip_ratio/high_max": 0.003686222968099173,
      "clip_ratio/high_mean": 0.0015619132755091414,
      "clip_ratio/low_mean": 0.001800961108529009,
      "clip_ratio/low_min": 6.676781413261779e-05,
      "clip_ratio/region_mean": 0.003362874369486235,
      "epoch": 3.7185768445610963,
      "grad_norm": 0.10085614770650864,
      "learning_rate": 1e-06,
      "loss": 0.0003,
      "step": 730
    },
    {
      "clip_ratio/high_max": 0.004310803866246715,
      "clip_ratio/high_mean": 0.0018141360524168704,
      "clip_ratio/low_mean": 0.0019254810213169549,
      "clip_ratio/low_min": 7.618234667461365e-05,
      "clip_ratio/region_mean": 0.0037396170082502067,
      "epoch": 3.7232429279673376,
      "grad_norm": 0.11083915084600449,
      "learning_rate": 1e-06,
      "loss": -0.0089,
      "step": 731
    },
    {
      "clip_ratio/high_max": 0.004351902585767675,
      "clip_ratio/high_mean": 0.001663338793150615,
      "clip_ratio/low_mean": 0.0016601150891801808,
      "clip_ratio/low_min": 2.6349072868470103e-05,
      "clip_ratio/region_mean": 0.003323453784105368,
      "epoch": 3.7279090113735784,
      "grad_norm": 0.10564175993204117,
      "learning_rate": 1e-06,
      "loss": -0.0384,
      "step": 732
    },
    {
      "clip_ratio/high_max": 0.004283978501916863,
      "clip_ratio/high_mean": 0.0019504584197420627,
      "clip_ratio/low_mean": 0.001666402011323953,
      "clip_ratio/low_min": 6.592826684936881e-05,
      "clip_ratio/region_mean": 0.0036168604128761217,
      "epoch": 3.7325750947798193,
      "grad_norm": 0.11122928559780121,
      "learning_rate": 1e-06,
      "loss": -0.0765,
      "step": 733
    },
    {
      "clip_ratio/high_max": 0.004652982868719846,
      "clip_ratio/high_mean": 0.001839220931287855,
      "clip_ratio/low_mean": 0.00168227751419181,
      "clip_ratio/low_min": 0.00013934968228568323,
      "clip_ratio/region_mean": 0.0035214984964113683,
      "epoch": 3.73724117818606,
      "grad_norm": 0.10405309498310089,
      "learning_rate": 1e-06,
      "loss": -0.0536,
      "step": 734
    },
    {
      "clip_ratio/high_max": 0.004151810164330527,
      "clip_ratio/high_mean": 0.0017772249411791563,
      "clip_ratio/low_mean": 0.002403065747785149,
      "clip_ratio/low_min": 0.0002291443834110396,
      "clip_ratio/region_mean": 0.004180290721706115,
      "epoch": 3.741907261592301,
      "grad_norm": 0.11087476462125778,
      "learning_rate": 1e-06,
      "loss": 0.0357,
      "step": 735
    },
    {
      "clip_ratio/high_max": 0.003940795955713838,
      "clip_ratio/high_mean": 0.0016709093397366814,
      "clip_ratio/low_mean": 0.0021931754090473987,
      "clip_ratio/low_min": 0.00013855856741429307,
      "clip_ratio/region_mean": 0.0038640848215436563,
      "epoch": 3.746573344998542,
      "grad_norm": 0.17536933720111847,
      "learning_rate": 1e-06,
      "loss": 0.0127,
      "step": 736
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0526646205357143,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4070.0,
      "completions/mean_length": 750.3714599609375,
      "completions/mean_terminated_length": 564.3800659179688,
      "completions/min_length": 98.0,
      "completions/min_terminated_length": 98.0,
      "epoch": 3.7512394284047827,
      "grad_norm": 0.17477920651435852,
      "learning_rate": 1e-06,
      "loss": -0.0647,
      "num_tokens": 222321741.0,
      "reward": 0.6285575032234192,
      "reward_std": 0.17536064982414246,
      "rewards/simpleverify_reward/mean": 0.6285575032234192,
      "rewards/simpleverify_reward/std": 0.48320725560188293,
      "step": 737
    },
    {
      "clip_ratio/high_max": 0.002359038648137357,
      "clip_ratio/high_mean": 0.0008411856597376755,
      "clip_ratio/low_mean": 0.0006273077415244188,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00146849341399502,
      "epoch": 3.7559055118110236,
      "grad_norm": 2.43017578125,
      "learning_rate": 1e-06,
      "loss": -0.0092,
      "step": 738
    },
    {
      "clip_ratio/high_max": 0.00226938795458409,
      "clip_ratio/high_mean": 0.0008588581622461788,
      "clip_ratio/low_mean": 0.0005710975647161831,
      "clip_ratio/low_min": 1.537893695058301e-05,
      "clip_ratio/region_mean": 0.0014299557333288249,
      "epoch": 3.7605715952172645,
      "grad_norm": 0.14628635346889496,
      "learning_rate": 1e-06,
      "loss": -0.0438,
      "step": 739
    },
    {
      "clip_ratio/high_max": 0.002076726992527256,
      "clip_ratio/high_mean": 0.0008326972883878625,
      "clip_ratio/low_mean": 0.000561808639758965,
      "clip_ratio/low_min": 1.3464024050335865e-05,
      "clip_ratio/region_mean": 0.0013945059217803646,
      "epoch": 3.7652376786235053,
      "grad_norm": 0.14165563881397247,
      "learning_rate": 1e-06,
      "loss": -0.0115,
      "step": 740
    },
    {
      "clip_ratio/high_max": 0.001913825712108519,
      "clip_ratio/high_mean": 0.0007372847721853759,
      "clip_ratio/low_mean": 0.0006441420955525246,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013814268313581124,
      "epoch": 3.769903762029746,
      "grad_norm": 0.13239187002182007,
      "learning_rate": 1e-06,
      "loss": -0.0009,
      "step": 741
    },
    {
      "clip_ratio/high_max": 0.002100650694046635,
      "clip_ratio/high_mean": 0.0007973564133862965,
      "clip_ratio/low_mean": 0.000842191906485823,
      "clip_ratio/low_min": 1.2804753168893512e-05,
      "clip_ratio/region_mean": 0.0016395483107771724,
      "epoch": 3.774569845435987,
      "grad_norm": 0.12319058179855347,
      "learning_rate": 1e-06,
      "loss": -0.0042,
      "step": 742
    },
    {
      "clip_ratio/high_max": 0.0018320144154131413,
      "clip_ratio/high_mean": 0.0008133291703416035,
      "clip_ratio/low_mean": 0.0009640983225835953,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017774275038391352,
      "epoch": 3.779235928842228,
      "grad_norm": 0.140071302652359,
      "learning_rate": 1e-06,
      "loss": 0.0448,
      "step": 743
    },
    {
      "clip_ratio/high_max": 0.002301309519680217,
      "clip_ratio/high_mean": 0.0010237305596092483,
      "clip_ratio/low_mean": 0.0007424470386467874,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017661775927990675,
      "epoch": 3.783902012248469,
      "grad_norm": 0.16917061805725098,
      "learning_rate": 1e-06,
      "loss": -0.0396,
      "step": 744
    },
    {
      "clip_ratio/high_max": 0.0018802807844622293,
      "clip_ratio/high_mean": 0.0007775138351462374,
      "clip_ratio/low_mean": 0.0008676103370817145,
      "clip_ratio/low_min": 4.0618530874780845e-05,
      "clip_ratio/region_mean": 0.0016451241608592682,
      "epoch": 3.7885680956547096,
      "grad_norm": 0.13434794545173645,
      "learning_rate": 1e-06,
      "loss": 0.0187,
      "step": 745
    },
    {
      "clip_ratio/high_max": 0.0022772336305934004,
      "clip_ratio/high_mean": 0.000921388669667067,
      "clip_ratio/low_mean": 0.0008603563092037803,
      "clip_ratio/low_min": 1.625487675482873e-05,
      "clip_ratio/region_mean": 0.0017817449843278155,
      "epoch": 3.793234179060951,
      "grad_norm": 0.13847526907920837,
      "learning_rate": 1e-06,
      "loss": -0.0228,
      "step": 746
    },
    {
      "clip_ratio/high_max": 0.002215390544733964,
      "clip_ratio/high_mean": 0.0009477393650740851,
      "clip_ratio/low_mean": 0.0010245876692351885,
      "clip_ratio/low_min": 1.4060742614674382e-05,
      "clip_ratio/region_mean": 0.0019723270306712948,
      "epoch": 3.7979002624671914,
      "grad_norm": 0.15017037093639374,
      "learning_rate": 1e-06,
      "loss": 0.0196,
      "step": 747
    },
    {
      "clip_ratio/high_max": 0.002312032134796027,
      "clip_ratio/high_mean": 0.0009924015939759556,
      "clip_ratio/low_mean": 0.0009686341563792666,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001961035763088148,
      "epoch": 3.8025663458734327,
      "grad_norm": 0.13074450194835663,
      "learning_rate": 1e-06,
      "loss": -0.0257,
      "step": 748
    },
    {
      "clip_ratio/high_max": 0.0021165127254789695,
      "clip_ratio/high_mean": 0.0008259641408585594,
      "clip_ratio/low_mean": 0.0009743193113536108,
      "clip_ratio/low_min": 3.6923986044712365e-05,
      "clip_ratio/region_mean": 0.0018002834476646967,
      "epoch": 3.8072324292796735,
      "grad_norm": 0.13358743488788605,
      "learning_rate": 1e-06,
      "loss": 0.0086,
      "step": 749
    },
    {
      "clip_ratio/high_max": 0.0030714675958734006,
      "clip_ratio/high_mean": 0.0011656206497718813,
      "clip_ratio/low_mean": 0.0007731245386821683,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019387452557566576,
      "epoch": 3.8118985126859144,
      "grad_norm": 0.12120359390974045,
      "learning_rate": 1e-06,
      "loss": -0.0747,
      "step": 750
    },
    {
      "clip_ratio/high_max": 0.002324391760339495,
      "clip_ratio/high_mean": 0.0009623090481909458,
      "clip_ratio/low_mean": 0.0009358588449686067,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018981679022544995,
      "epoch": 3.8165645960921553,
      "grad_norm": 0.13132305443286896,
      "learning_rate": 1e-06,
      "loss": -0.0451,
      "step": 751
    },
    {
      "clip_ratio/high_max": 0.002670203997695353,
      "clip_ratio/high_mean": 0.001091261779947672,
      "clip_ratio/low_mean": 0.0009214749770762865,
      "clip_ratio/low_min": 3.470266346994322e-05,
      "clip_ratio/region_mean": 0.002012736767937895,
      "epoch": 3.821230679498396,
      "grad_norm": 0.12459760904312134,
      "learning_rate": 1e-06,
      "loss": -0.0283,
      "step": 752
    },
    {
      "clip_ratio/high_max": 0.005750593772972934,
      "clip_ratio/high_mean": 0.002223786694230512,
      "clip_ratio/low_mean": 0.0017182409756060224,
      "clip_ratio/low_min": 9.561421393300407e-05,
      "clip_ratio/region_mean": 0.003942027629818767,
      "epoch": 3.825896762904637,
      "grad_norm": 0.11602294445037842,
      "learning_rate": 1e-06,
      "loss": -0.0657,
      "step": 753
    },
    {
      "clip_ratio/high_max": 0.003535622003255412,
      "clip_ratio/high_mean": 0.0013698389630008023,
      "clip_ratio/low_mean": 0.0014545201884175185,
      "clip_ratio/low_min": 4.64082168036839e-05,
      "clip_ratio/region_mean": 0.0028243591805221513,
      "epoch": 3.830562846310878,
      "grad_norm": 0.13256803154945374,
      "learning_rate": 1e-06,
      "loss": -0.0097,
      "step": 754
    },
    {
      "clip_ratio/high_max": 0.004794902604771778,
      "clip_ratio/high_mean": 0.0018382197958999313,
      "clip_ratio/low_mean": 0.0016228220520133618,
      "clip_ratio/low_min": 4.613681085174903e-05,
      "clip_ratio/region_mean": 0.0034610418369993567,
      "epoch": 3.8352289297171187,
      "grad_norm": 0.11848633736371994,
      "learning_rate": 1e-06,
      "loss": -0.0447,
      "step": 755
    },
    {
      "clip_ratio/high_max": 0.004067458568897564,
      "clip_ratio/high_mean": 0.0017436861216992838,
      "clip_ratio/low_mean": 0.0018153616292693187,
      "clip_ratio/low_min": 0.00011876183998538181,
      "clip_ratio/region_mean": 0.0035590476327342913,
      "epoch": 3.8398950131233596,
      "grad_norm": 0.11235970258712769,
      "learning_rate": 1e-06,
      "loss": -0.0123,
      "step": 756
    },
    {
      "clip_ratio/high_max": 0.0035849043051712215,
      "clip_ratio/high_mean": 0.0015424811754201073,
      "clip_ratio/low_mean": 0.0017083663333323784,
      "clip_ratio/low_min": 0.00012292051178519614,
      "clip_ratio/region_mean": 0.0032508474978385493,
      "epoch": 3.8445610965296004,
      "grad_norm": 0.11216001212596893,
      "learning_rate": 1e-06,
      "loss": -0.0017,
      "step": 757
    },
    {
      "clip_ratio/high_max": 0.0038531647078343667,
      "clip_ratio/high_mean": 0.0014799288255744614,
      "clip_ratio/low_mean": 0.001976608797122026,
      "clip_ratio/low_min": 5.913903260079678e-05,
      "clip_ratio/region_mean": 0.0034565376408863813,
      "epoch": 3.8492271799358413,
      "grad_norm": 0.10212117433547974,
      "learning_rate": 1e-06,
      "loss": -0.0048,
      "step": 758
    },
    {
      "clip_ratio/high_max": 0.004134741029702127,
      "clip_ratio/high_mean": 0.0014983863366069272,
      "clip_ratio/low_mean": 0.0022383331670425832,
      "clip_ratio/low_min": 9.652051085140556e-05,
      "clip_ratio/region_mean": 0.0037367193726822734,
      "epoch": 3.853893263342082,
      "grad_norm": 0.1095103770494461,
      "learning_rate": 1e-06,
      "loss": 0.0439,
      "step": 759
    },
    {
      "clip_ratio/high_max": 0.004754321649670601,
      "clip_ratio/high_mean": 0.001988091695238836,
      "clip_ratio/low_mean": 0.0017386630606779363,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037267547741066664,
      "epoch": 3.858559346748323,
      "grad_norm": 0.11243551969528198,
      "learning_rate": 1e-06,
      "loss": -0.0406,
      "step": 760
    },
    {
      "clip_ratio/high_max": 0.004078614703757921,
      "clip_ratio/high_mean": 0.0015099863321665907,
      "clip_ratio/low_mean": 0.0018661021022126079,
      "clip_ratio/low_min": 5.624297045869753e-05,
      "clip_ratio/region_mean": 0.0033760884834919125,
      "epoch": 3.863225430154564,
      "grad_norm": 0.10717935115098953,
      "learning_rate": 1e-06,
      "loss": 0.0179,
      "step": 761
    },
    {
      "clip_ratio/high_max": 0.004397791883093305,
      "clip_ratio/high_mean": 0.0017535864826641046,
      "clip_ratio/low_mean": 0.0017320508231932763,
      "clip_ratio/low_min": 0.00013003901403862983,
      "clip_ratio/region_mean": 0.003485637265839614,
      "epoch": 3.8678915135608047,
      "grad_norm": 0.1085558608174324,
      "learning_rate": 1e-06,
      "loss": -0.0237,
      "step": 762
    },
    {
      "clip_ratio/high_max": 0.004115596319024917,
      "clip_ratio/high_mean": 0.0016645620198687539,
      "clip_ratio/low_mean": 0.002103723309119232,
      "clip_ratio/low_min": 1.449107367079705e-05,
      "clip_ratio/region_mean": 0.003768285343539901,
      "epoch": 3.872557596967046,
      "grad_norm": 0.11942525953054428,
      "learning_rate": 1e-06,
      "loss": 0.0187,
      "step": 763
    },
    {
      "clip_ratio/high_max": 0.00424246299371589,
      "clip_ratio/high_mean": 0.0017170476203318685,
      "clip_ratio/low_mean": 0.001973378617549315,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036904262378811836,
      "epoch": 3.8772236803732865,
      "grad_norm": 0.10570453852415085,
      "learning_rate": 1e-06,
      "loss": -0.0265,
      "step": 764
    },
    {
      "clip_ratio/high_max": 0.0040037437283899635,
      "clip_ratio/high_mean": 0.0015547644688922446,
      "clip_ratio/low_mean": 0.001978310403501382,
      "clip_ratio/low_min": 7.850350084481761e-05,
      "clip_ratio/region_mean": 0.003533074908773415,
      "epoch": 3.8818897637795278,
      "grad_norm": 0.1083567664027214,
      "learning_rate": 1e-06,
      "loss": 0.0077,
      "step": 765
    },
    {
      "clip_ratio/high_max": 0.0046975701552582905,
      "clip_ratio/high_mean": 0.001908313533931505,
      "clip_ratio/low_mean": 0.0014650588782387786,
      "clip_ratio/low_min": 3.4473247069399804e-05,
      "clip_ratio/region_mean": 0.003373372441274114,
      "epoch": 3.886555847185768,
      "grad_norm": 0.09803938865661621,
      "learning_rate": 1e-06,
      "loss": -0.0754,
      "step": 766
    },
    {
      "clip_ratio/high_max": 0.004197909656795673,
      "clip_ratio/high_mean": 0.0016492087124788668,
      "clip_ratio/low_mean": 0.0016601090392214246,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003309317777166143,
      "epoch": 3.8912219305920095,
      "grad_norm": 0.10391347110271454,
      "learning_rate": 1e-06,
      "loss": -0.0459,
      "step": 767
    },
    {
      "clip_ratio/high_max": 0.004114445735467598,
      "clip_ratio/high_mean": 0.0017251825993298553,
      "clip_ratio/low_mean": 0.0017964808685064781,
      "clip_ratio/low_min": 4.848132266488392e-05,
      "clip_ratio/region_mean": 0.003521663456922397,
      "epoch": 3.8958880139982504,
      "grad_norm": 0.10053832828998566,
      "learning_rate": 1e-06,
      "loss": -0.029,
      "step": 768
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05224609375,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4076.0,
      "completions/mean_length": 751.1327514648438,
      "completions/mean_terminated_length": 566.7428588867188,
      "completions/min_length": 93.0,
      "completions/min_terminated_length": 93.0,
      "epoch": 4.004666083406241,
      "grad_norm": 0.15310677886009216,
      "learning_rate": 1e-06,
      "loss": -0.0267,
      "num_tokens": 231437548.0,
      "reward": 0.6320452094078064,
      "reward_std": 0.1672099530696869,
      "rewards/simpleverify_reward/mean": 0.6320452094078064,
      "rewards/simpleverify_reward/std": 0.48226580023765564,
      "step": 769
    },
    {
      "clip_ratio/high_max": 0.0019281686836620793,
      "clip_ratio/high_mean": 0.000810307843494229,
      "clip_ratio/low_mean": 0.0005624724508379586,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013727803016081452,
      "epoch": 4.009332166812482,
      "grad_norm": 0.14683762192726135,
      "learning_rate": 1e-06,
      "loss": -0.0043,
      "step": 770
    },
    {
      "clip_ratio/high_max": 0.002079153142403811,
      "clip_ratio/high_mean": 0.0009315488823631313,
      "clip_ratio/low_mean": 0.000509585847339622,
      "clip_ratio/low_min": 1.4357913642015774e-05,
      "clip_ratio/region_mean": 0.0014411347001441754,
      "epoch": 4.013998250218723,
      "grad_norm": 0.166998952627182,
      "learning_rate": 1e-06,
      "loss": -0.0529,
      "step": 771
    },
    {
      "clip_ratio/high_max": 0.0019399192751734518,
      "clip_ratio/high_mean": 0.0008342950713995378,
      "clip_ratio/low_mean": 0.0004934847083859495,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001327779762505088,
      "epoch": 4.0186643336249634,
      "grad_norm": 0.13920444250106812,
      "learning_rate": 1e-06,
      "loss": -0.0777,
      "step": 772
    },
    {
      "clip_ratio/high_max": 0.002094589752232423,
      "clip_ratio/high_mean": 0.0007418531313305721,
      "clip_ratio/low_mean": 0.0005927519014221616,
      "clip_ratio/low_min": 1.3443751413433347e-05,
      "clip_ratio/region_mean": 0.0013346050182008184,
      "epoch": 4.023330417031205,
      "grad_norm": 0.14391952753067017,
      "learning_rate": 1e-06,
      "loss": 0.0313,
      "step": 773
    },
    {
      "clip_ratio/high_max": 0.0016304237433359958,
      "clip_ratio/high_mean": 0.0006564975301444065,
      "clip_ratio/low_mean": 0.0009182446938211797,
      "clip_ratio/low_min": 9.740125460666604e-05,
      "clip_ratio/region_mean": 0.0015747422658023424,
      "epoch": 4.027996500437445,
      "grad_norm": 0.1473110467195511,
      "learning_rate": 1e-06,
      "loss": 0.0725,
      "step": 774
    },
    {
      "clip_ratio/high_max": 0.001807895336241927,
      "clip_ratio/high_mean": 0.0007525314977101516,
      "clip_ratio/low_mean": 0.0006893953623148263,
      "clip_ratio/low_min": 1.4282450138125569e-05,
      "clip_ratio/region_mean": 0.0014419268190977164,
      "epoch": 4.0326625838436865,
      "grad_norm": 0.2618180215358734,
      "learning_rate": 1e-06,
      "loss": 0.0087,
      "step": 775
    },
    {
      "clip_ratio/high_max": 0.0019271043056505732,
      "clip_ratio/high_mean": 0.0007989125369931571,
      "clip_ratio/low_mean": 0.0007507249192713061,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015496374471695162,
      "epoch": 4.037328667249927,
      "grad_norm": 0.1659613400697708,
      "learning_rate": 1e-06,
      "loss": 0.0537,
      "step": 776
    },
    {
      "clip_ratio/high_max": 0.002457275360939093,
      "clip_ratio/high_mean": 0.0010113995485880878,
      "clip_ratio/low_mean": 0.0006561719774254016,
      "clip_ratio/low_min": 6.624775960517582e-05,
      "clip_ratio/region_mean": 0.0016675715560268145,
      "epoch": 4.041994750656168,
      "grad_norm": 0.1470591127872467,
      "learning_rate": 1e-06,
      "loss": -0.0696,
      "step": 777
    },
    {
      "clip_ratio/high_max": 0.0023316239748965017,
      "clip_ratio/high_mean": 0.0008853714643919375,
      "clip_ratio/low_mean": 0.0007322361934711807,
      "clip_ratio/low_min": 1.6684463844285347e-05,
      "clip_ratio/region_mean": 0.0016176076023839414,
      "epoch": 4.046660834062409,
      "grad_norm": 0.1329738199710846,
      "learning_rate": 1e-06,
      "loss": -0.0272,
      "step": 778
    },
    {
      "clip_ratio/high_max": 0.0023746980587020516,
      "clip_ratio/high_mean": 0.000955700097620138,
      "clip_ratio/low_mean": 0.0007923496677904041,
      "clip_ratio/low_min": 0.00011225816160731483,
      "clip_ratio/region_mean": 0.0017480497917858884,
      "epoch": 4.05132691746865,
      "grad_norm": 0.1371714174747467,
      "learning_rate": 1e-06,
      "loss": -0.0351,
      "step": 779
    },
    {
      "clip_ratio/high_max": 0.002152440862118965,
      "clip_ratio/high_mean": 0.000914997246582061,
      "clip_ratio/low_mean": 0.0007267953774316993,
      "clip_ratio/low_min": 2.956480602733791e-05,
      "clip_ratio/region_mean": 0.001641792623559013,
      "epoch": 4.05599300087489,
      "grad_norm": 0.15127252042293549,
      "learning_rate": 1e-06,
      "loss": -0.0008,
      "step": 780
    },
    {
      "clip_ratio/high_max": 0.0026071308384416625,
      "clip_ratio/high_mean": 0.0010602735819702502,
      "clip_ratio/low_mean": 0.0008386076415263233,
      "clip_ratio/low_min": 4.0919271668826696e-05,
      "clip_ratio/region_mean": 0.0018988811898452695,
      "epoch": 4.060659084281132,
      "grad_norm": 0.14045506715774536,
      "learning_rate": 1e-06,
      "loss": -0.0268,
      "step": 781
    },
    {
      "clip_ratio/high_max": 0.0023875536026025657,
      "clip_ratio/high_mean": 0.0009982529518310912,
      "clip_ratio/low_mean": 0.0007220503011922119,
      "clip_ratio/low_min": 1.4511260815197602e-05,
      "clip_ratio/region_mean": 0.0017203032366523985,
      "epoch": 4.065325167687372,
      "grad_norm": 0.1363794058561325,
      "learning_rate": 1e-06,
      "loss": -0.061,
      "step": 782
    },
    {
      "clip_ratio/high_max": 0.002237800872535445,
      "clip_ratio/high_mean": 0.0009411189221282257,
      "clip_ratio/low_mean": 0.0007834278085283586,
      "clip_ratio/low_min": 1.3058921467745677e-05,
      "clip_ratio/region_mean": 0.001724546767945867,
      "epoch": 4.069991251093613,
      "grad_norm": 0.14006073772907257,
      "learning_rate": 1e-06,
      "loss": -0.0118,
      "step": 783
    },
    {
      "clip_ratio/high_max": 0.0023545168733107857,
      "clip_ratio/high_mean": 0.0009650763731769985,
      "clip_ratio/low_mean": 0.0008394488713747705,
      "clip_ratio/low_min": 1.2945319213031325e-05,
      "clip_ratio/region_mean": 0.0018045252436422743,
      "epoch": 4.074657334499854,
      "grad_norm": 0.12399918586015701,
      "learning_rate": 1e-06,
      "loss": -0.0149,
      "step": 784
    },
    {
      "clip_ratio/high_max": 0.004768627492012456,
      "clip_ratio/high_mean": 0.0019168325743521564,
      "clip_ratio/low_mean": 0.0019984095233667176,
      "clip_ratio/low_min": 0.00013257576210889965,
      "clip_ratio/region_mean": 0.003915242094080895,
      "epoch": 4.079323417906095,
      "grad_norm": 0.10813162475824356,
      "learning_rate": 1e-06,
      "loss": -0.0277,
      "step": 785
    },
    {
      "clip_ratio/high_max": 0.0038928389622014947,
      "clip_ratio/high_mean": 0.001720463815217954,
      "clip_ratio/low_mean": 0.0017948549648281187,
      "clip_ratio/low_min": 9.112090629059821e-05,
      "clip_ratio/region_mean": 0.0035153187927789986,
      "epoch": 4.083989501312336,
      "grad_norm": 0.11050586402416229,
      "learning_rate": 1e-06,
      "loss": -0.0053,
      "step": 786
    },
    {
      "clip_ratio/high_max": 0.0042497809627093375,
      "clip_ratio/high_mean": 0.0018293542743776925,
      "clip_ratio/low_mean": 0.0015810461409273557,
      "clip_ratio/low_min": 4.887903924100101e-05,
      "clip_ratio/region_mean": 0.0034104005026165396,
      "epoch": 4.088655584718577,
      "grad_norm": 0.10018651187419891,
      "learning_rate": 1e-06,
      "loss": -0.0537,
      "step": 787
    },
    {
      "clip_ratio/high_max": 0.004470041996682994,
      "clip_ratio/high_mean": 0.0018017599068116397,
      "clip_ratio/low_mean": 0.0014861097724860883,
      "clip_ratio/low_min": 6.933658733032644e-05,
      "clip_ratio/region_mean": 0.003287869672931265,
      "epoch": 4.093321668124818,
      "grad_norm": 0.10895606875419617,
      "learning_rate": 1e-06,
      "loss": -0.0786,
      "step": 788
    },
    {
      "clip_ratio/high_max": 0.004060956765897572,
      "clip_ratio/high_mean": 0.0014696106518385932,
      "clip_ratio/low_mean": 0.0016836730610521045,
      "clip_ratio/low_min": 0.00017380488225171575,
      "clip_ratio/region_mean": 0.003153283687424846,
      "epoch": 4.0979877515310585,
      "grad_norm": 0.1082402765750885,
      "learning_rate": 1e-06,
      "loss": 0.0304,
      "step": 789
    },
    {
      "clip_ratio/high_max": 0.0034149702478316613,
      "clip_ratio/high_mean": 0.001354354008071823,
      "clip_ratio/low_mean": 0.002163726749131456,
      "clip_ratio/low_min": 0.00010136599303223193,
      "clip_ratio/region_mean": 0.0035180808044970036,
      "epoch": 4.1026538349373,
      "grad_norm": 0.11549611389636993,
      "learning_rate": 1e-06,
      "loss": 0.0716,
      "step": 790
    },
    {
      "clip_ratio/high_max": 0.003594898749724962,
      "clip_ratio/high_mean": 0.0014823518176854122,
      "clip_ratio/low_mean": 0.0019362849016033579,
      "clip_ratio/low_min": 4.2087540350621566e-05,
      "clip_ratio/region_mean": 0.0034186366538051516,
      "epoch": 4.10731991834354,
      "grad_norm": 0.10483682155609131,
      "learning_rate": 1e-06,
      "loss": 0.0078,
      "step": 791
    },
    {
      "clip_ratio/high_max": 0.0035983753041364253,
      "clip_ratio/high_mean": 0.0015340608879341744,
      "clip_ratio/low_mean": 0.0019362895254744217,
      "clip_ratio/low_min": 4.0577830077381805e-05,
      "clip_ratio/region_mean": 0.003470350435236469,
      "epoch": 4.111986001749782,
      "grad_norm": 0.10962541401386261,
      "learning_rate": 1e-06,
      "loss": 0.0528,
      "step": 792
    },
    {
      "clip_ratio/high_max": 0.004705524479504675,
      "clip_ratio/high_mean": 0.0018390097429801244,
      "clip_ratio/low_mean": 0.0015835553567740135,
      "clip_ratio/low_min": 0.00020804732594115194,
      "clip_ratio/region_mean": 0.003422565132495947,
      "epoch": 4.116652085156022,
      "grad_norm": 0.11367785930633545,
      "learning_rate": 1e-06,
      "loss": -0.0705,
      "step": 793
    },
    {
      "clip_ratio/high_max": 0.004744290665257722,
      "clip_ratio/high_mean": 0.0018269469437655061,
      "clip_ratio/low_mean": 0.0016876386980584357,
      "clip_ratio/low_min": 7.148984877858311e-05,
      "clip_ratio/region_mean": 0.003514585696393624,
      "epoch": 4.121318168562263,
      "grad_norm": 0.11523880809545517,
      "learning_rate": 1e-06,
      "loss": -0.0281,
      "step": 794
    },
    {
      "clip_ratio/high_max": 0.004780321134603582,
      "clip_ratio/high_mean": 0.0017885620854940498,
      "clip_ratio/low_mean": 0.0017926065120263956,
      "clip_ratio/low_min": 0.00018352220467932057,
      "clip_ratio/region_mean": 0.0035811686975648627,
      "epoch": 4.125984251968504,
      "grad_norm": 0.10960574448108673,
      "learning_rate": 1e-06,
      "loss": -0.0361,
      "step": 795
    },
    {
      "clip_ratio/high_max": 0.004405864514410496,
      "clip_ratio/high_mean": 0.00181055924622342,
      "clip_ratio/low_mean": 0.0017280713764193933,
      "clip_ratio/low_min": 4.628486931324005e-05,
      "clip_ratio/region_mean": 0.003538630604452919,
      "epoch": 4.130650335374745,
      "grad_norm": 0.11767042428255081,
      "learning_rate": 1e-06,
      "loss": -0.0018,
      "step": 796
    },
    {
      "clip_ratio/high_max": 0.005056446942035109,
      "clip_ratio/high_mean": 0.001986493145523127,
      "clip_ratio/low_mean": 0.0016335325453837868,
      "clip_ratio/low_min": 0.00010130920418305323,
      "clip_ratio/region_mean": 0.0036200256072334014,
      "epoch": 4.135316418780985,
      "grad_norm": 0.10663873702287674,
      "learning_rate": 1e-06,
      "loss": -0.0276,
      "step": 797
    },
    {
      "clip_ratio/high_max": 0.004029067713418044,
      "clip_ratio/high_mean": 0.0018686227303987835,
      "clip_ratio/low_mean": 0.0015110203494259622,
      "clip_ratio/low_min": 8.672525291331112e-05,
      "clip_ratio/region_mean": 0.003379643130756449,
      "epoch": 4.139982502187227,
      "grad_norm": 0.10719020664691925,
      "learning_rate": 1e-06,
      "loss": -0.0618,
      "step": 798
    },
    {
      "clip_ratio/high_max": 0.004220316142891534,
      "clip_ratio/high_mean": 0.0018387219788564835,
      "clip_ratio/low_mean": 0.0017396884795743972,
      "clip_ratio/low_min": 2.89821473415941e-05,
      "clip_ratio/region_mean": 0.0035784105712082237,
      "epoch": 4.144648585593467,
      "grad_norm": 0.1055833175778389,
      "learning_rate": 1e-06,
      "loss": -0.0127,
      "step": 799
    },
    {
      "clip_ratio/high_max": 0.004139769880566746,
      "clip_ratio/high_mean": 0.0017569779702171218,
      "clip_ratio/low_mean": 0.001630351831408916,
      "clip_ratio/low_min": 6.589509575860575e-05,
      "clip_ratio/region_mean": 0.0033873297506943345,
      "epoch": 4.1493146689997085,
      "grad_norm": 0.09660372138023376,
      "learning_rate": 1e-06,
      "loss": -0.0156,
      "step": 800
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.054966517857142905,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 3951.0,
      "completions/mean_length": 757.5895385742188,
      "completions/mean_terminated_length": 563.4156494140625,
      "completions/min_length": 87.0,
      "completions/min_terminated_length": 87.0,
      "epoch": 4.153980752405949,
      "grad_norm": 0.1303890198469162,
      "learning_rate": 1e-06,
      "loss": 0.0233,
      "num_tokens": 240510959.0,
      "reward": 0.6400669813156128,
      "reward_std": 0.16997414827346802,
      "rewards/simpleverify_reward/mean": 0.6400669813156128,
      "rewards/simpleverify_reward/std": 0.4799971878528595,
      "step": 801
    },
    {
      "clip_ratio/high_max": 0.0022182005195645615,
      "clip_ratio/high_mean": 0.0009035354141815333,
      "clip_ratio/low_mean": 0.00040883870860852767,
      "clip_ratio/low_min": 2.0414829123183154e-05,
      "clip_ratio/region_mean": 0.0013123741155141033,
      "epoch": 4.15864683581219,
      "grad_norm": 0.1353575736284256,
      "learning_rate": 1e-06,
      "loss": -0.0451,
      "step": 802
    },
    {
      "clip_ratio/high_max": 0.0019282172725070268,
      "clip_ratio/high_mean": 0.0006982145287111052,
      "clip_ratio/low_mean": 0.0004980692733624892,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0011962838179897517,
      "epoch": 4.163312919218431,
      "grad_norm": 0.1320056915283203,
      "learning_rate": 1e-06,
      "loss": 0.0104,
      "step": 803
    },
    {
      "clip_ratio/high_max": 0.002186690879170783,
      "clip_ratio/high_mean": 0.0008233120861405041,
      "clip_ratio/low_mean": 0.000502015504935116,
      "clip_ratio/low_min": 1.4980824744270649e-05,
      "clip_ratio/region_mean": 0.0013253275719762314,
      "epoch": 4.167979002624672,
      "grad_norm": 0.14056254923343658,
      "learning_rate": 1e-06,
      "loss": -0.0136,
      "step": 804
    },
    {
      "clip_ratio/high_max": 0.0018644992742338218,
      "clip_ratio/high_mean": 0.0007338922723647556,
      "clip_ratio/low_mean": 0.0006387510229615145,
      "clip_ratio/low_min": 1.2457643606467173e-05,
      "clip_ratio/region_mean": 0.0013726432880503125,
      "epoch": 4.172645086030913,
      "grad_norm": 0.142314150929451,
      "learning_rate": 1e-06,
      "loss": 0.0426,
      "step": 805
    },
    {
      "clip_ratio/high_max": 0.0022365255281329155,
      "clip_ratio/high_mean": 0.0009228459985024529,
      "clip_ratio/low_mean": 0.0005956899294687901,
      "clip_ratio/low_min": 4.7568952140863985e-05,
      "clip_ratio/region_mean": 0.001518535915238317,
      "epoch": 4.177311169437154,
      "grad_norm": 0.15085534751415253,
      "learning_rate": 1e-06,
      "loss": -0.0449,
      "step": 806
    },
    {
      "clip_ratio/high_max": 0.002015844114794163,
      "clip_ratio/high_mean": 0.0007960460789036006,
      "clip_ratio/low_mean": 0.0007721027604929986,
      "clip_ratio/low_min": 6.227522681001574e-05,
      "clip_ratio/region_mean": 0.001568148840306094,
      "epoch": 4.181977252843395,
      "grad_norm": 0.1615273654460907,
      "learning_rate": 1e-06,
      "loss": -0.0028,
      "step": 807
    },
    {
      "clip_ratio/high_max": 0.002182118456403259,
      "clip_ratio/high_mean": 0.000867377075337572,
      "clip_ratio/low_mean": 0.0005352647349354811,
      "clip_ratio/low_min": 1.6391293684137054e-05,
      "clip_ratio/region_mean": 0.0014026417811692227,
      "epoch": 4.186643336249635,
      "grad_norm": 0.14414745569229126,
      "learning_rate": 1e-06,
      "loss": -0.0459,
      "step": 808
    },
    {
      "clip_ratio/high_max": 0.002051880284852814,
      "clip_ratio/high_mean": 0.0009167857915599598,
      "clip_ratio/low_mean": 0.0007480407002731226,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016648264499963261,
      "epoch": 4.191309419655877,
      "grad_norm": 0.14732277393341064,
      "learning_rate": 1e-06,
      "loss": -0.0004,
      "step": 809
    },
    {
      "clip_ratio/high_max": 0.0022003626872901805,
      "clip_ratio/high_mean": 0.0008670665047247894,
      "clip_ratio/low_mean": 0.0006310636917987722,
      "clip_ratio/low_min": 1.4955730875954032e-05,
      "clip_ratio/region_mean": 0.0014981302010710351,
      "epoch": 4.195975503062117,
      "grad_norm": 0.14946165680885315,
      "learning_rate": 1e-06,
      "loss": -0.0403,
      "step": 810
    },
    {
      "clip_ratio/high_max": 0.0025730982815730385,
      "clip_ratio/high_mean": 0.0010119561811734457,
      "clip_ratio/low_mean": 0.000793170767792617,
      "clip_ratio/low_min": 3.861623099510325e-05,
      "clip_ratio/region_mean": 0.0018051269580610096,
      "epoch": 4.200641586468358,
      "grad_norm": 0.20283643901348114,
      "learning_rate": 1e-06,
      "loss": -0.0117,
      "step": 811
    },
    {
      "clip_ratio/high_max": 0.0021455318419612013,
      "clip_ratio/high_mean": 0.0008771239990892354,
      "clip_ratio/low_mean": 0.0007881807368903537,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001665304764173925,
      "epoch": 4.205307669874599,
      "grad_norm": 0.13793624937534332,
      "learning_rate": 1e-06,
      "loss": -0.049,
      "step": 812
    },
    {
      "clip_ratio/high_max": 0.002361922037380282,
      "clip_ratio/high_mean": 0.0009592125097697135,
      "clip_ratio/low_mean": 0.0007195807465905091,
      "clip_ratio/low_min": 4.42750497313682e-05,
      "clip_ratio/region_mean": 0.001678793239989318,
      "epoch": 4.20997375328084,
      "grad_norm": 0.14665831625461578,
      "learning_rate": 1e-06,
      "loss": -0.033,
      "step": 813
    },
    {
      "clip_ratio/high_max": 0.0025589886427042075,
      "clip_ratio/high_mean": 0.001097928319722996,
      "clip_ratio/low_mean": 0.0007390546379610896,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018369829485891387,
      "epoch": 4.2146398366870805,
      "grad_norm": 0.142076775431633,
      "learning_rate": 1e-06,
      "loss": -0.0695,
      "step": 814
    },
    {
      "clip_ratio/high_max": 0.0027473546069813892,
      "clip_ratio/high_mean": 0.0010548358768573962,
      "clip_ratio/low_mean": 0.0009258007758035092,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019806366763077676,
      "epoch": 4.219305920093322,
      "grad_norm": 0.14724980294704437,
      "learning_rate": 1e-06,
      "loss": -0.0079,
      "step": 815
    },
    {
      "clip_ratio/high_max": 0.0032117781302076764,
      "clip_ratio/high_mean": 0.0012206568790134043,
      "clip_ratio/low_mean": 0.0009935115704138298,
      "clip_ratio/low_min": 3.166160240652971e-05,
      "clip_ratio/region_mean": 0.0022141684603411704,
      "epoch": 4.223972003499562,
      "grad_norm": 0.146607905626297,
      "learning_rate": 1e-06,
      "loss": -0.0,
      "step": 816
    },
    {
      "clip_ratio/high_max": 0.004683214487158693,
      "clip_ratio/high_mean": 0.0016601214156253263,
      "clip_ratio/low_mean": 0.0015739032824058086,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003234024756238796,
      "epoch": 4.228638086905804,
      "grad_norm": 0.0962480679154396,
      "learning_rate": 1e-06,
      "loss": 0.0226,
      "step": 817
    },
    {
      "clip_ratio/high_max": 0.0051162459421902895,
      "clip_ratio/high_mean": 0.00207439846417401,
      "clip_ratio/low_mean": 0.00145659243571572,
      "clip_ratio/low_min": 7.145190465962514e-05,
      "clip_ratio/region_mean": 0.003530990972649306,
      "epoch": 4.233304170312044,
      "grad_norm": 0.11082332581281662,
      "learning_rate": 1e-06,
      "loss": -0.0459,
      "step": 818
    },
    {
      "clip_ratio/high_max": 0.00396244965668302,
      "clip_ratio/high_mean": 0.0015430440325872041,
      "clip_ratio/low_mean": 0.001758814545610221,
      "clip_ratio/low_min": 3.710667078848928e-05,
      "clip_ratio/region_mean": 0.003301858509075828,
      "epoch": 4.237970253718285,
      "grad_norm": 0.13086576759815216,
      "learning_rate": 1e-06,
      "loss": 0.0097,
      "step": 819
    },
    {
      "clip_ratio/high_max": 0.004030764815979637,
      "clip_ratio/high_mean": 0.0017902548061101697,
      "clip_ratio/low_mean": 0.001622117943043122,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034123726582038216,
      "epoch": 4.242636337124526,
      "grad_norm": 0.10510283708572388,
      "learning_rate": 1e-06,
      "loss": -0.0145,
      "step": 820
    },
    {
      "clip_ratio/high_max": 0.0037007724822615273,
      "clip_ratio/high_mean": 0.0014824314021097962,
      "clip_ratio/low_mean": 0.0019096084688499104,
      "clip_ratio/low_min": 6.586123345186934e-05,
      "clip_ratio/region_mean": 0.003392039885511622,
      "epoch": 4.247302420530767,
      "grad_norm": 0.1070796474814415,
      "learning_rate": 1e-06,
      "loss": 0.0417,
      "step": 821
    },
    {
      "clip_ratio/high_max": 0.004464153011213057,
      "clip_ratio/high_mean": 0.0018111991194018628,
      "clip_ratio/low_mean": 0.0017414308404113399,
      "clip_ratio/low_min": 9.955874702427536e-05,
      "clip_ratio/region_mean": 0.0035526299761841074,
      "epoch": 4.251968503937007,
      "grad_norm": 0.10536172986030579,
      "learning_rate": 1e-06,
      "loss": -0.0458,
      "step": 822
    },
    {
      "clip_ratio/high_max": 0.004796643399458844,
      "clip_ratio/high_mean": 0.0020294683636166155,
      "clip_ratio/low_mean": 0.002219650499682757,
      "clip_ratio/low_min": 0.00022132278900244273,
      "clip_ratio/region_mean": 0.004249118763254955,
      "epoch": 4.256634587343249,
      "grad_norm": 0.11842436343431473,
      "learning_rate": 1e-06,
      "loss": -0.004,
      "step": 823
    },
    {
      "clip_ratio/high_max": 0.004542186914477497,
      "clip_ratio/high_mean": 0.0018836865274352022,
      "clip_ratio/low_mean": 0.0015959499269229127,
      "clip_ratio/low_min": 3.278258736827411e-05,
      "clip_ratio/region_mean": 0.0034796364489011467,
      "epoch": 4.26130067074949,
      "grad_norm": 0.10479302704334259,
      "learning_rate": 1e-06,
      "loss": -0.0469,
      "step": 824
    },
    {
      "clip_ratio/high_max": 0.004993004273273982,
      "clip_ratio/high_mean": 0.0019716628521564417,
      "clip_ratio/low_mean": 0.0019497534976835595,
      "clip_ratio/low_min": 5.7192533859051764e-05,
      "clip_ratio/region_mean": 0.003921416304365266,
      "epoch": 4.2659667541557305,
      "grad_norm": 0.11368942260742188,
      "learning_rate": 1e-06,
      "loss": -0.0014,
      "step": 825
    },
    {
      "clip_ratio/high_max": 0.00470474392932374,
      "clip_ratio/high_mean": 0.0018914956162916496,
      "clip_ratio/low_mean": 0.0015624966254108585,
      "clip_ratio/low_min": 4.2671825212892145e-05,
      "clip_ratio/region_mean": 0.003453992190770805,
      "epoch": 4.270632837561972,
      "grad_norm": 0.10735386610031128,
      "learning_rate": 1e-06,
      "loss": -0.0411,
      "step": 826
    },
    {
      "clip_ratio/high_max": 0.004756406488013454,
      "clip_ratio/high_mean": 0.001861303004261572,
      "clip_ratio/low_mean": 0.0018831373272405472,
      "clip_ratio/low_min": 0.00012971574687981047,
      "clip_ratio/region_mean": 0.0037444403569679707,
      "epoch": 4.275298920968212,
      "grad_norm": 0.10722675174474716,
      "learning_rate": 1e-06,
      "loss": -0.0127,
      "step": 827
    },
    {
      "clip_ratio/high_max": 0.0039394246123265475,
      "clip_ratio/high_mean": 0.0017110314838646445,
      "clip_ratio/low_mean": 0.0016145914778462611,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033256230381084606,
      "epoch": 4.2799650043744535,
      "grad_norm": 0.11127717047929764,
      "learning_rate": 1e-06,
      "loss": -0.0498,
      "step": 828
    },
    {
      "clip_ratio/high_max": 0.004596872029651422,
      "clip_ratio/high_mean": 0.0018493219176889397,
      "clip_ratio/low_mean": 0.0016120491272886284,
      "clip_ratio/low_min": 0.0001126769020629581,
      "clip_ratio/region_mean": 0.0034613710013218224,
      "epoch": 4.284631087780694,
      "grad_norm": 0.10674639791250229,
      "learning_rate": 1e-06,
      "loss": -0.0339,
      "step": 829
    },
    {
      "clip_ratio/high_max": 0.004938447622407693,
      "clip_ratio/high_mean": 0.0021008079893363174,
      "clip_ratio/low_mean": 0.001592003183759516,
      "clip_ratio/low_min": 5.755064557888545e-05,
      "clip_ratio/region_mean": 0.0036928112531313673,
      "epoch": 4.289297171186935,
      "grad_norm": 0.11336338520050049,
      "learning_rate": 1e-06,
      "loss": -0.0704,
      "step": 830
    },
    {
      "clip_ratio/high_max": 0.004813632971490733,
      "clip_ratio/high_mean": 0.001890468458441319,
      "clip_ratio/low_mean": 0.0019733882982109208,
      "clip_ratio/low_min": 5.0423557695467025e-05,
      "clip_ratio/region_mean": 0.0038638567930320278,
      "epoch": 4.293963254593176,
      "grad_norm": 0.11263824254274368,
      "learning_rate": 1e-06,
      "loss": -0.0089,
      "step": 831
    },
    {
      "clip_ratio/high_max": 0.005107821853016503,
      "clip_ratio/high_mean": 0.0019648220113595016,
      "clip_ratio/low_mean": 0.0020307529703131877,
      "clip_ratio/low_min": 5.191029777051881e-05,
      "clip_ratio/region_mean": 0.003995574952568859,
      "epoch": 4.298629337999417,
      "grad_norm": 0.11687400192022324,
      "learning_rate": 1e-06,
      "loss": -0.001,
      "step": 832
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0552455357142857,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4050.0,
      "completions/mean_length": 746.4529418945312,
      "completions/mean_terminated_length": 550.5845336914062,
      "completions/min_length": 107.0,
      "completions/min_terminated_length": 107.0,
      "epoch": 4.303295421405657,
      "grad_norm": 0.15614362061023712,
      "learning_rate": 1e-06,
      "loss": -0.0253,
      "num_tokens": 249417172.0,
      "reward": 0.6404157876968384,
      "reward_std": 0.16133955121040344,
      "rewards/simpleverify_reward/mean": 0.6404157280921936,
      "rewards/simpleverify_reward/std": 0.47989529371261597,
      "step": 833
    },
    {
      "clip_ratio/high_max": 0.0019733297231141478,
      "clip_ratio/high_mean": 0.0007616561470058514,
      "clip_ratio/low_mean": 0.0003771799752030347,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0011388361308490857,
      "epoch": 4.307961504811899,
      "grad_norm": 0.11703892052173615,
      "learning_rate": 1e-06,
      "loss": -0.0354,
      "step": 834
    },
    {
      "clip_ratio/high_max": 0.0017955149014596827,
      "clip_ratio/high_mean": 0.00075140794251638,
      "clip_ratio/low_mean": 0.0005205176976232906,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012719256410491653,
      "epoch": 4.312627588218139,
      "grad_norm": 0.13232852518558502,
      "learning_rate": 1e-06,
      "loss": -0.0301,
      "step": 835
    },
    {
      "clip_ratio/high_max": 0.0022504101289086975,
      "clip_ratio/high_mean": 0.0007580840783703024,
      "clip_ratio/low_mean": 0.0006088165519031463,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001366900651191827,
      "epoch": 4.31729367162438,
      "grad_norm": 0.1509833186864853,
      "learning_rate": 1e-06,
      "loss": -0.004,
      "step": 836
    },
    {
      "clip_ratio/high_max": 0.0018517387979954947,
      "clip_ratio/high_mean": 0.0006814233456680086,
      "clip_ratio/low_mean": 0.0005646014424200985,
      "clip_ratio/low_min": 4.8543690354563296e-05,
      "clip_ratio/region_mean": 0.0012460247962735593,
      "epoch": 4.321959755030621,
      "grad_norm": 0.14457760751247406,
      "learning_rate": 1e-06,
      "loss": -0.0033,
      "step": 837
    },
    {
      "clip_ratio/high_max": 0.0020943353047186974,
      "clip_ratio/high_mean": 0.0009089495742955478,
      "clip_ratio/low_mean": 0.000545835501725378,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014547850587405264,
      "epoch": 4.326625838436862,
      "grad_norm": 0.16517773270606995,
      "learning_rate": 1e-06,
      "loss": -0.0521,
      "step": 838
    },
    {
      "clip_ratio/high_max": 0.0020902541946270503,
      "clip_ratio/high_mean": 0.000818183794763172,
      "clip_ratio/low_mean": 0.0005284679818942095,
      "clip_ratio/low_min": 1.6451698684250005e-05,
      "clip_ratio/region_mean": 0.001346651781204855,
      "epoch": 4.331291921843103,
      "grad_norm": 0.1452416032552719,
      "learning_rate": 1e-06,
      "loss": -0.0028,
      "step": 839
    },
    {
      "clip_ratio/high_max": 0.0022448662639362738,
      "clip_ratio/high_mean": 0.0009470668483118061,
      "clip_ratio/low_mean": 0.0007544977806901443,
      "clip_ratio/low_min": 2.0051331375725567e-05,
      "clip_ratio/region_mean": 0.0017015646080835722,
      "epoch": 4.335958005249344,
      "grad_norm": 0.154227152466774,
      "learning_rate": 1e-06,
      "loss": -0.0499,
      "step": 840
    },
    {
      "clip_ratio/high_max": 0.002260458692035172,
      "clip_ratio/high_mean": 0.0009264874788641464,
      "clip_ratio/low_mean": 0.0007122709248505998,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016387584328185767,
      "epoch": 4.340624088655585,
      "grad_norm": 0.15906360745429993,
      "learning_rate": 1e-06,
      "loss": -0.0753,
      "step": 841
    },
    {
      "clip_ratio/high_max": 0.001790504771634005,
      "clip_ratio/high_mean": 0.0007345174362853868,
      "clip_ratio/low_mean": 0.0008167722917278297,
      "clip_ratio/low_min": 3.391209975234233e-05,
      "clip_ratio/region_mean": 0.0015512897371081635,
      "epoch": 4.3452901720618256,
      "grad_norm": 0.14767055213451385,
      "learning_rate": 1e-06,
      "loss": -0.0089,
      "step": 842
    },
    {
      "clip_ratio/high_max": 0.002371604168729391,
      "clip_ratio/high_mean": 0.0009641461310820887,
      "clip_ratio/low_mean": 0.0008687915960763348,
      "clip_ratio/low_min": 4.038337283418514e-05,
      "clip_ratio/region_mean": 0.0018329377344343811,
      "epoch": 4.349956255468067,
      "grad_norm": 0.14764977991580963,
      "learning_rate": 1e-06,
      "loss": -0.0236,
      "step": 843
    },
    {
      "clip_ratio/high_max": 0.002153606779756956,
      "clip_ratio/high_mean": 0.000962589367190958,
      "clip_ratio/low_mean": 0.0008469747272101813,
      "clip_ratio/low_min": 1.8200349586550146e-05,
      "clip_ratio/region_mean": 0.0018095641207764857,
      "epoch": 4.354622338874307,
      "grad_norm": 0.15038618445396423,
      "learning_rate": 1e-06,
      "loss": -0.0243,
      "step": 844
    },
    {
      "clip_ratio/high_max": 0.0024876581810531206,
      "clip_ratio/high_mean": 0.0008999401725304779,
      "clip_ratio/low_mean": 0.0008811707884888165,
      "clip_ratio/low_min": 3.200204810127616e-05,
      "clip_ratio/region_mean": 0.0017811109064496122,
      "epoch": 4.359288422280549,
      "grad_norm": 0.14355981349945068,
      "learning_rate": 1e-06,
      "loss": -0.0087,
      "step": 845
    },
    {
      "clip_ratio/high_max": 0.0027049496093241032,
      "clip_ratio/high_mean": 0.0010384314173279563,
      "clip_ratio/low_mean": 0.0009056093240360497,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019440407340880483,
      "epoch": 4.363954505686789,
      "grad_norm": 0.13249464333057404,
      "learning_rate": 1e-06,
      "loss": -0.0204,
      "step": 846
    },
    {
      "clip_ratio/high_max": 0.0026887355488725007,
      "clip_ratio/high_mean": 0.0009471602461417206,
      "clip_ratio/low_mean": 0.0011728872632374987,
      "clip_ratio/low_min": 1.4820962860540021e-05,
      "clip_ratio/region_mean": 0.0021200475603109226,
      "epoch": 4.36862058909303,
      "grad_norm": 0.16055814921855927,
      "learning_rate": 1e-06,
      "loss": 0.0564,
      "step": 847
    },
    {
      "clip_ratio/high_max": 0.002603636574349366,
      "clip_ratio/high_mean": 0.0009957883503375342,
      "clip_ratio/low_mean": 0.0009716837648738874,
      "clip_ratio/low_min": 3.131262565148063e-05,
      "clip_ratio/region_mean": 0.0019674721334013157,
      "epoch": 4.373286672499271,
      "grad_norm": 0.4194774925708771,
      "learning_rate": 1e-06,
      "loss": 0.0092,
      "step": 848
    },
    {
      "clip_ratio/high_max": 0.005164644433534704,
      "clip_ratio/high_mean": 0.0017719701754685957,
      "clip_ratio/low_mean": 0.0017693263198452769,
      "clip_ratio/low_min": 0.00012767993757734075,
      "clip_ratio/region_mean": 0.0035412964716670103,
      "epoch": 4.377952755905512,
      "grad_norm": 0.10239679366350174,
      "learning_rate": 1e-06,
      "loss": -0.0261,
      "step": 849
    },
    {
      "clip_ratio/high_max": 0.004233103740261868,
      "clip_ratio/high_mean": 0.00167551855702186,
      "clip_ratio/low_mean": 0.0012020007525279652,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.002877519349567592,
      "epoch": 4.3826188393117524,
      "grad_norm": 0.09712014347314835,
      "learning_rate": 1e-06,
      "loss": -0.036,
      "step": 850
    },
    {
      "clip_ratio/high_max": 0.004521616967394948,
      "clip_ratio/high_mean": 0.001693811413133517,
      "clip_ratio/low_mean": 0.0015810784134373534,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003274889815656934,
      "epoch": 4.387284922717994,
      "grad_norm": 0.10434824973344803,
      "learning_rate": 1e-06,
      "loss": -0.0309,
      "step": 851
    },
    {
      "clip_ratio/high_max": 0.00442529335123254,
      "clip_ratio/high_mean": 0.0015839472398511134,
      "clip_ratio/low_mean": 0.0017927135959325824,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003376660868525505,
      "epoch": 4.391951006124234,
      "grad_norm": 0.11312887817621231,
      "learning_rate": 1e-06,
      "loss": -0.0048,
      "step": 852
    },
    {
      "clip_ratio/high_max": 0.003923805696103955,
      "clip_ratio/high_mean": 0.0015902004452073015,
      "clip_ratio/low_mean": 0.0017551334349263925,
      "clip_ratio/low_min": 0.0002065648695861455,
      "clip_ratio/region_mean": 0.003345333934703376,
      "epoch": 4.3966170895304755,
      "grad_norm": 0.10837895423173904,
      "learning_rate": 1e-06,
      "loss": -0.0042,
      "step": 853
    },
    {
      "clip_ratio/high_max": 0.004731163033284247,
      "clip_ratio/high_mean": 0.0019692518435476813,
      "clip_ratio/low_mean": 0.0015142955780902412,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034835473634302616,
      "epoch": 4.401283172936716,
      "grad_norm": 0.11120729148387909,
      "learning_rate": 1e-06,
      "loss": -0.053,
      "step": 854
    },
    {
      "clip_ratio/high_max": 0.004113192786462605,
      "clip_ratio/high_mean": 0.0017632048693485558,
      "clip_ratio/low_mean": 0.0017256184910365846,
      "clip_ratio/low_min": 4.935509423376061e-05,
      "clip_ratio/region_mean": 0.0034888233203673735,
      "epoch": 4.405949256342957,
      "grad_norm": 0.10507329553365707,
      "learning_rate": 1e-06,
      "loss": -0.0037,
      "step": 855
    },
    {
      "clip_ratio/high_max": 0.005143274262081832,
      "clip_ratio/high_mean": 0.0020832320569752483,
      "clip_ratio/low_mean": 0.0019940908168791793,
      "clip_ratio/low_min": 3.105590076302178e-05,
      "clip_ratio/region_mean": 0.004077322853845544,
      "epoch": 4.410615339749198,
      "grad_norm": 0.11597028374671936,
      "learning_rate": 1e-06,
      "loss": -0.0509,
      "step": 856
    },
    {
      "clip_ratio/high_max": 0.0049623505474301055,
      "clip_ratio/high_mean": 0.002057961784885265,
      "clip_ratio/low_mean": 0.0015668369633203838,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036247987300157547,
      "epoch": 4.415281423155439,
      "grad_norm": 0.10977782309055328,
      "learning_rate": 1e-06,
      "loss": -0.0762,
      "step": 857
    },
    {
      "clip_ratio/high_max": 0.004179463256150484,
      "clip_ratio/high_mean": 0.0016374404294765554,
      "clip_ratio/low_mean": 0.001921766066516284,
      "clip_ratio/low_min": 6.827616016380489e-05,
      "clip_ratio/region_mean": 0.003559206481440924,
      "epoch": 4.41994750656168,
      "grad_norm": 0.09874669462442398,
      "learning_rate": 1e-06,
      "loss": -0.0097,
      "step": 858
    },
    {
      "clip_ratio/high_max": 0.005240919272182509,
      "clip_ratio/high_mean": 0.002115168419550173,
      "clip_ratio/low_mean": 0.001900317489344161,
      "clip_ratio/low_min": 6.975446740398183e-05,
      "clip_ratio/region_mean": 0.00401548579975497,
      "epoch": 4.424613589967921,
      "grad_norm": 0.12094337493181229,
      "learning_rate": 1e-06,
      "loss": -0.0245,
      "step": 859
    },
    {
      "clip_ratio/high_max": 0.004078470228705555,
      "clip_ratio/high_mean": 0.0019232929043937474,
      "clip_ratio/low_mean": 0.001969290155102499,
      "clip_ratio/low_min": 0.00014264184937928803,
      "clip_ratio/region_mean": 0.0038925828994251788,
      "epoch": 4.429279673374162,
      "grad_norm": 0.11632602661848068,
      "learning_rate": 1e-06,
      "loss": -0.0252,
      "step": 860
    },
    {
      "clip_ratio/high_max": 0.004177798458840698,
      "clip_ratio/high_mean": 0.0016859750394360162,
      "clip_ratio/low_mean": 0.0020808482622669544,
      "clip_ratio/low_min": 6.092290277592838e-05,
      "clip_ratio/region_mean": 0.0037668233271688223,
      "epoch": 4.433945756780402,
      "grad_norm": 0.10976847261190414,
      "learning_rate": 1e-06,
      "loss": -0.0096,
      "step": 861
    },
    {
      "clip_ratio/high_max": 0.004763852419273462,
      "clip_ratio/high_mean": 0.0018648332043085247,
      "clip_ratio/low_mean": 0.0018139741932827746,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003678807450341992,
      "epoch": 4.438611840186644,
      "grad_norm": 0.10736335068941116,
      "learning_rate": 1e-06,
      "loss": -0.0211,
      "step": 862
    },
    {
      "clip_ratio/high_max": 0.005099577596411109,
      "clip_ratio/high_mean": 0.0018054244974337053,
      "clip_ratio/low_mean": 0.002453087792673614,
      "clip_ratio/low_min": 5.2112285629846156e-05,
      "clip_ratio/region_mean": 0.004258512228261679,
      "epoch": 4.443277923592884,
      "grad_norm": 0.12012824416160583,
      "learning_rate": 1e-06,
      "loss": 0.0553,
      "step": 863
    },
    {
      "clip_ratio/high_max": 0.0041862075158860534,
      "clip_ratio/high_mean": 0.001806170283089159,
      "clip_ratio/low_mean": 0.0021111234709678683,
      "clip_ratio/low_min": 0.00012395874364301562,
      "clip_ratio/region_mean": 0.003917293754057027,
      "epoch": 4.447944006999125,
      "grad_norm": 0.11286681890487671,
      "learning_rate": 1e-06,
      "loss": 0.0082,
      "step": 864
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.059849330357142905,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4087.0,
      "completions/mean_length": 760.3589477539062,
      "completions/mean_terminated_length": 548.0143432617188,
      "completions/min_length": 85.0,
      "completions/min_terminated_length": 85.0,
      "epoch": 4.452610090405366,
      "grad_norm": 0.14467744529247284,
      "learning_rate": 1e-06,
      "loss": -0.0089,
      "num_tokens": 258211005.0,
      "reward": 0.6513671875,
      "reward_std": 0.16271862387657166,
      "rewards/simpleverify_reward/mean": 0.6513671875,
      "rewards/simpleverify_reward/std": 0.47655409574508667,
      "step": 865
    },
    {
      "clip_ratio/high_max": 0.002316336896910798,
      "clip_ratio/high_mean": 0.0009091276860999642,
      "clip_ratio/low_mean": 0.0005119927582200035,
      "clip_ratio/low_min": 1.4341440873977263e-05,
      "clip_ratio/region_mean": 0.001421120443410473,
      "epoch": 4.457276173811607,
      "grad_norm": 0.1792125105857849,
      "learning_rate": 1e-06,
      "loss": 0.0152,
      "step": 866
    },
    {
      "clip_ratio/high_max": 0.0019441972399363294,
      "clip_ratio/high_mean": 0.0008390541006519925,
      "clip_ratio/low_mean": 0.0005526151089725317,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013916692259954289,
      "epoch": 4.4619422572178475,
      "grad_norm": 0.14807169139385223,
      "learning_rate": 1e-06,
      "loss": 0.004,
      "step": 867
    },
    {
      "clip_ratio/high_max": 0.002378000226599397,
      "clip_ratio/high_mean": 0.0008174100212272606,
      "clip_ratio/low_mean": 0.0005076649413240375,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013250749288999941,
      "epoch": 4.466608340624089,
      "grad_norm": 0.14011111855506897,
      "learning_rate": 1e-06,
      "loss": -0.0136,
      "step": 868
    },
    {
      "clip_ratio/high_max": 0.0019595753110479563,
      "clip_ratio/high_mean": 0.0008377066769753583,
      "clip_ratio/low_mean": 0.0006263971245061839,
      "clip_ratio/low_min": 2.6795283702085726e-05,
      "clip_ratio/region_mean": 0.001464103796024574,
      "epoch": 4.471274424030329,
      "grad_norm": 0.16807053983211517,
      "learning_rate": 1e-06,
      "loss": -0.0146,
      "step": 869
    },
    {
      "clip_ratio/high_max": 0.0023144043152569793,
      "clip_ratio/high_mean": 0.0009316980140283704,
      "clip_ratio/low_mean": 0.00048414221146231284,
      "clip_ratio/low_min": 2.0038474758621305e-05,
      "clip_ratio/region_mean": 0.001415840244590072,
      "epoch": 4.475940507436571,
      "grad_norm": 0.1402735710144043,
      "learning_rate": 1e-06,
      "loss": -0.043,
      "step": 870
    },
    {
      "clip_ratio/high_max": 0.0023137359239626676,
      "clip_ratio/high_mean": 0.0008994987783808028,
      "clip_ratio/low_mean": 0.000516015511493606,
      "clip_ratio/low_min": 2.61451586993644e-05,
      "clip_ratio/region_mean": 0.0014155143035168294,
      "epoch": 4.480606590842811,
      "grad_norm": 0.1389661282300949,
      "learning_rate": 1e-06,
      "loss": -0.0388,
      "step": 871
    },
    {
      "clip_ratio/high_max": 0.0020878209252259694,
      "clip_ratio/high_mean": 0.0008631467462691944,
      "clip_ratio/low_mean": 0.000618815704001463,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014819624411757104,
      "epoch": 4.485272674249052,
      "grad_norm": 0.1456226110458374,
      "learning_rate": 1e-06,
      "loss": -0.0067,
      "step": 872
    },
    {
      "clip_ratio/high_max": 0.0027428164539742284,
      "clip_ratio/high_mean": 0.0010457027237862349,
      "clip_ratio/low_mean": 0.0006245064837457903,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016702092325431295,
      "epoch": 4.489938757655293,
      "grad_norm": 0.14806191623210907,
      "learning_rate": 1e-06,
      "loss": -0.0566,
      "step": 873
    },
    {
      "clip_ratio/high_max": 0.0025542981675243936,
      "clip_ratio/high_mean": 0.0009610265551600605,
      "clip_ratio/low_mean": 0.0008237967913373723,
      "clip_ratio/low_min": 3.612716682255268e-05,
      "clip_ratio/region_mean": 0.001784823376510758,
      "epoch": 4.494604841061534,
      "grad_norm": 0.13969510793685913,
      "learning_rate": 1e-06,
      "loss": -0.0007,
      "step": 874
    },
    {
      "clip_ratio/high_max": 0.002498247482435545,
      "clip_ratio/high_mean": 0.0008596607276558643,
      "clip_ratio/low_mean": 0.0007649064564247965,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016245672304648906,
      "epoch": 4.499270924467774,
      "grad_norm": 0.14741335809230804,
      "learning_rate": 1e-06,
      "loss": -0.008,
      "step": 875
    },
    {
      "clip_ratio/high_max": 0.002230259844509419,
      "clip_ratio/high_mean": 0.0009351415774290217,
      "clip_ratio/low_mean": 0.0006591346937057097,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015942762547638267,
      "epoch": 4.503937007874016,
      "grad_norm": 0.14288948476314545,
      "learning_rate": 1e-06,
      "loss": -0.0744,
      "step": 876
    },
    {
      "clip_ratio/high_max": 0.002228765268228017,
      "clip_ratio/high_mean": 0.0009118919861066388,
      "clip_ratio/low_mean": 0.0008585191244492307,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001770411130564753,
      "epoch": 4.508603091280257,
      "grad_norm": 0.13459570705890656,
      "learning_rate": 1e-06,
      "loss": -0.0504,
      "step": 877
    },
    {
      "clip_ratio/high_max": 0.0025594074395485222,
      "clip_ratio/high_mean": 0.0010769996661110781,
      "clip_ratio/low_mean": 0.0007173201038312982,
      "clip_ratio/low_min": 4.5391858293442056e-05,
      "clip_ratio/region_mean": 0.0017943197817658074,
      "epoch": 4.5132691746864975,
      "grad_norm": 0.14680016040802002,
      "learning_rate": 1e-06,
      "loss": -0.0328,
      "step": 878
    },
    {
      "clip_ratio/high_max": 0.002306700167537201,
      "clip_ratio/high_mean": 0.0008955223729572026,
      "clip_ratio/low_mean": 0.0007326979175559245,
      "clip_ratio/low_min": 5.7451481552561745e-05,
      "clip_ratio/region_mean": 0.0016282202959700953,
      "epoch": 4.517935258092739,
      "grad_norm": 0.13933761417865753,
      "learning_rate": 1e-06,
      "loss": -0.0047,
      "step": 879
    },
    {
      "clip_ratio/high_max": 0.0028688012826023623,
      "clip_ratio/high_mean": 0.0011098889663117006,
      "clip_ratio/low_mean": 0.000850883605380659,
      "clip_ratio/low_min": 3.868592921207892e-05,
      "clip_ratio/region_mean": 0.001960772558959434,
      "epoch": 4.522601341498979,
      "grad_norm": 0.15788087248802185,
      "learning_rate": 1e-06,
      "loss": -0.0223,
      "step": 880
    },
    {
      "clip_ratio/high_max": 0.00535259504977148,
      "clip_ratio/high_mean": 0.0019520611167536117,
      "clip_ratio/low_mean": 0.001639450252696406,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035915113548981026,
      "epoch": 4.5272674249052205,
      "grad_norm": 0.09956730902194977,
      "learning_rate": 1e-06,
      "loss": -0.0097,
      "step": 881
    },
    {
      "clip_ratio/high_max": 0.004989013526937924,
      "clip_ratio/high_mean": 0.0018150042742490768,
      "clip_ratio/low_mean": 0.0019589620096667204,
      "clip_ratio/low_min": 5.9637404774548486e-05,
      "clip_ratio/region_mean": 0.0037739662366220728,
      "epoch": 4.531933508311461,
      "grad_norm": 0.11524045467376709,
      "learning_rate": 1e-06,
      "loss": 0.0141,
      "step": 882
    },
    {
      "clip_ratio/high_max": 0.003978137872763909,
      "clip_ratio/high_mean": 0.0017849262621894013,
      "clip_ratio/low_mean": 0.0019400392247916898,
      "clip_ratio/low_min": 0.00016200301979552023,
      "clip_ratio/region_mean": 0.00372496536874678,
      "epoch": 4.536599591717702,
      "grad_norm": 0.10538440197706223,
      "learning_rate": 1e-06,
      "loss": 0.0031,
      "step": 883
    },
    {
      "clip_ratio/high_max": 0.0054128130213939585,
      "clip_ratio/high_mean": 0.0017640075366216479,
      "clip_ratio/low_mean": 0.001633014808248845,
      "clip_ratio/low_min": 4.260395508026704e-05,
      "clip_ratio/region_mean": 0.0033970223848882597,
      "epoch": 4.541265675123943,
      "grad_norm": 0.1047298014163971,
      "learning_rate": 1e-06,
      "loss": -0.0144,
      "step": 884
    },
    {
      "clip_ratio/high_max": 0.004757896866067313,
      "clip_ratio/high_mean": 0.0019963004961027764,
      "clip_ratio/low_mean": 0.001823432648961898,
      "clip_ratio/low_min": 9.751690413395409e-05,
      "clip_ratio/region_mean": 0.003819733072305098,
      "epoch": 4.545931758530184,
      "grad_norm": 0.11066095530986786,
      "learning_rate": 1e-06,
      "loss": -0.0155,
      "step": 885
    },
    {
      "clip_ratio/high_max": 0.004701394456787966,
      "clip_ratio/high_mean": 0.001973842332517961,
      "clip_ratio/low_mean": 0.0014590519895136822,
      "clip_ratio/low_min": 8.305855590151623e-05,
      "clip_ratio/region_mean": 0.003432894285651855,
      "epoch": 4.550597841936424,
      "grad_norm": 0.10153724998235703,
      "learning_rate": 1e-06,
      "loss": -0.0438,
      "step": 886
    },
    {
      "clip_ratio/high_max": 0.00507149487384595,
      "clip_ratio/high_mean": 0.0019511599730321905,
      "clip_ratio/low_mean": 0.0015741557072033174,
      "clip_ratio/low_min": 3.0361914468812756e-05,
      "clip_ratio/region_mean": 0.0035253157111583278,
      "epoch": 4.555263925342666,
      "grad_norm": 0.10322874039411545,
      "learning_rate": 1e-06,
      "loss": -0.0397,
      "step": 887
    },
    {
      "clip_ratio/high_max": 0.004596612459863536,
      "clip_ratio/high_mean": 0.0017719544666761067,
      "clip_ratio/low_mean": 0.0016361703092115931,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034081247285939753,
      "epoch": 4.559930008748906,
      "grad_norm": 0.10048892349004745,
      "learning_rate": 1e-06,
      "loss": -0.0076,
      "step": 888
    },
    {
      "clip_ratio/high_max": 0.005299713113345206,
      "clip_ratio/high_mean": 0.002177820435463218,
      "clip_ratio/low_mean": 0.00166122873997665,
      "clip_ratio/low_min": 3.2075957278721035e-05,
      "clip_ratio/region_mean": 0.0038390491536119953,
      "epoch": 4.564596092155147,
      "grad_norm": 0.11002933233976364,
      "learning_rate": 1e-06,
      "loss": -0.0576,
      "step": 889
    },
    {
      "clip_ratio/high_max": 0.005094123160233721,
      "clip_ratio/high_mean": 0.0020380398200359195,
      "clip_ratio/low_mean": 0.0019353531279193703,
      "clip_ratio/low_min": 8.312551653943956e-05,
      "clip_ratio/region_mean": 0.003973393046180718,
      "epoch": 4.569262175561388,
      "grad_norm": 0.11039287596940994,
      "learning_rate": 1e-06,
      "loss": -0.0016,
      "step": 890
    },
    {
      "clip_ratio/high_max": 0.004597242412273772,
      "clip_ratio/high_mean": 0.0017055854477803223,
      "clip_ratio/low_mean": 0.0019164369950885884,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003622022377385292,
      "epoch": 4.573928258967629,
      "grad_norm": 0.11102067679166794,
      "learning_rate": 1e-06,
      "loss": -0.0091,
      "step": 891
    },
    {
      "clip_ratio/high_max": 0.005018611598643474,
      "clip_ratio/high_mean": 0.0019941198697779328,
      "clip_ratio/low_mean": 0.0016786272935860325,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036727471742779016,
      "epoch": 4.57859434237387,
      "grad_norm": 0.12340915203094482,
      "learning_rate": 1e-06,
      "loss": -0.0753,
      "step": 892
    },
    {
      "clip_ratio/high_max": 0.004183171215117909,
      "clip_ratio/high_mean": 0.0018596687259559985,
      "clip_ratio/low_mean": 0.0019171984677086584,
      "clip_ratio/low_min": 6.156119343359023e-05,
      "clip_ratio/region_mean": 0.003776867190026678,
      "epoch": 4.583260425780111,
      "grad_norm": 0.10531151294708252,
      "learning_rate": 1e-06,
      "loss": -0.0512,
      "step": 893
    },
    {
      "clip_ratio/high_max": 0.005291488443617709,
      "clip_ratio/high_mean": 0.0020982137757528108,
      "clip_ratio/low_mean": 0.0016920844827836845,
      "clip_ratio/low_min": 0.00013617557851830497,
      "clip_ratio/region_mean": 0.003790298127569258,
      "epoch": 4.587926509186351,
      "grad_norm": 0.11162614077329636,
      "learning_rate": 1e-06,
      "loss": -0.0336,
      "step": 894
    },
    {
      "clip_ratio/high_max": 0.004409324668813497,
      "clip_ratio/high_mean": 0.0018550814274931327,
      "clip_ratio/low_mean": 0.0019025323053938337,
      "clip_ratio/low_min": 0.00019946683823945932,
      "clip_ratio/region_mean": 0.0037576137256110087,
      "epoch": 4.592592592592593,
      "grad_norm": 0.10397019237279892,
      "learning_rate": 1e-06,
      "loss": -0.0056,
      "step": 895
    },
    {
      "clip_ratio/high_max": 0.005536343996936921,
      "clip_ratio/high_mean": 0.0021662823201040737,
      "clip_ratio/low_mean": 0.0020387462864164263,
      "clip_ratio/low_min": 0.000132779701743857,
      "clip_ratio/region_mean": 0.004205028613796458,
      "epoch": 4.597258675998834,
      "grad_norm": 0.12072121351957321,
      "learning_rate": 1e-06,
      "loss": -0.0234,
      "step": 896
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0640345982142857,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4092.0,
      "completions/mean_length": 779.6520385742188,
      "completions/mean_terminated_length": 552.7621459960938,
      "completions/min_length": 104.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 4.601924759405074,
      "grad_norm": 0.1514691263437271,
      "learning_rate": 1e-06,
      "loss": -0.0414,
      "num_tokens": 267081832.0,
      "reward": 0.6273019313812256,
      "reward_std": 0.17339251935482025,
      "rewards/simpleverify_reward/mean": 0.6273018717765808,
      "rewards/simpleverify_reward/std": 0.4835395812988281,
      "step": 897
    },
    {
      "clip_ratio/high_max": 0.0019966550171375275,
      "clip_ratio/high_mean": 0.0008502315777150216,
      "clip_ratio/low_mean": 0.0005893587976970593,
      "clip_ratio/low_min": 1.0537852176639717e-05,
      "clip_ratio/region_mean": 0.001439590396330459,
      "epoch": 4.606590842811316,
      "grad_norm": 0.14296402037143707,
      "learning_rate": 1e-06,
      "loss": -0.0248,
      "step": 898
    },
    {
      "clip_ratio/high_max": 0.002014832331042271,
      "clip_ratio/high_mean": 0.0008722093480173498,
      "clip_ratio/low_mean": 0.0005629072593364981,
      "clip_ratio/low_min": 2.0458264771150425e-05,
      "clip_ratio/region_mean": 0.001435116610082332,
      "epoch": 4.611256926217556,
      "grad_norm": 0.14364726841449738,
      "learning_rate": 1e-06,
      "loss": -0.0404,
      "step": 899
    },
    {
      "clip_ratio/high_max": 0.001931311859152629,
      "clip_ratio/high_mean": 0.0007064443070703419,
      "clip_ratio/low_mean": 0.0005196976344450377,
      "clip_ratio/low_min": 1.7857142665889114e-05,
      "clip_ratio/region_mean": 0.001226141952429316,
      "epoch": 4.615923009623797,
      "grad_norm": 0.15297484397888184,
      "learning_rate": 1e-06,
      "loss": 0.0051,
      "step": 900
    },
    {
      "clip_ratio/high_max": 0.0021477193004102446,
      "clip_ratio/high_mean": 0.0008941083542595152,
      "clip_ratio/low_mean": 0.0006637904762101243,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001557898816827219,
      "epoch": 4.620589093030038,
      "grad_norm": 0.16430549323558807,
      "learning_rate": 1e-06,
      "loss": -0.0043,
      "step": 901
    },
    {
      "clip_ratio/high_max": 0.0021034933233750053,
      "clip_ratio/high_mean": 0.0008832715120661305,
      "clip_ratio/low_mean": 0.0005339268254829221,
      "clip_ratio/low_min": 1.535249248263426e-05,
      "clip_ratio/region_mean": 0.001417198305716738,
      "epoch": 4.625255176436279,
      "grad_norm": 0.12906159460544586,
      "learning_rate": 1e-06,
      "loss": -0.0428,
      "step": 902
    },
    {
      "clip_ratio/high_max": 0.0017007193819154054,
      "clip_ratio/high_mean": 0.000825791841634782,
      "clip_ratio/low_mean": 0.0006772770047973609,
      "clip_ratio/low_min": 2.825090632541105e-05,
      "clip_ratio/region_mean": 0.0015030688191473018,
      "epoch": 4.6299212598425195,
      "grad_norm": 0.13809776306152344,
      "learning_rate": 1e-06,
      "loss": -0.0136,
      "step": 903
    },
    {
      "clip_ratio/high_max": 0.0018239171695313416,
      "clip_ratio/high_mean": 0.0008989349298644811,
      "clip_ratio/low_mean": 0.0007857135205995291,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016846484140842222,
      "epoch": 4.634587343248761,
      "grad_norm": 0.1569642871618271,
      "learning_rate": 1e-06,
      "loss": -0.0476,
      "step": 904
    },
    {
      "clip_ratio/high_max": 0.0022080958005972207,
      "clip_ratio/high_mean": 0.0009907887379085878,
      "clip_ratio/low_mean": 0.0007729441331321141,
      "clip_ratio/low_min": 2.535790736146737e-05,
      "clip_ratio/region_mean": 0.0017637328746786807,
      "epoch": 4.639253426655001,
      "grad_norm": 0.13189558684825897,
      "learning_rate": 1e-06,
      "loss": -0.0164,
      "step": 905
    },
    {
      "clip_ratio/high_max": 0.002198292910179589,
      "clip_ratio/high_mean": 0.0009344591671833768,
      "clip_ratio/low_mean": 0.0008431569713138742,
      "clip_ratio/low_min": 1.546264138596598e-05,
      "clip_ratio/region_mean": 0.00177761611121241,
      "epoch": 4.6439195100612425,
      "grad_norm": 1.2006735801696777,
      "learning_rate": 1e-06,
      "loss": -0.0119,
      "step": 906
    },
    {
      "clip_ratio/high_max": 0.0023442394885933027,
      "clip_ratio/high_mean": 0.0008907200317480601,
      "clip_ratio/low_mean": 0.0008983572843135335,
      "clip_ratio/low_min": 1.826417246775236e-05,
      "clip_ratio/region_mean": 0.001789077268767869,
      "epoch": 4.648585593467483,
      "grad_norm": 0.17841781675815582,
      "learning_rate": 1e-06,
      "loss": -0.0247,
      "step": 907
    },
    {
      "clip_ratio/high_max": 0.002398787924903445,
      "clip_ratio/high_mean": 0.0009394316330144648,
      "clip_ratio/low_mean": 0.0009981245384551585,
      "clip_ratio/low_min": 8.655553392600268e-05,
      "clip_ratio/region_mean": 0.00193755621148739,
      "epoch": 4.653251676873724,
      "grad_norm": 0.13568075001239777,
      "learning_rate": 1e-06,
      "loss": -0.0113,
      "step": 908
    },
    {
      "clip_ratio/high_max": 0.0024681854047230445,
      "clip_ratio/high_mean": 0.0011434089319664054,
      "clip_ratio/low_mean": 0.0009468560929235537,
      "clip_ratio/low_min": 4.797081237484235e-05,
      "clip_ratio/region_mean": 0.002090265064907726,
      "epoch": 4.657917760279965,
      "grad_norm": 0.15703392028808594,
      "learning_rate": 1e-06,
      "loss": -0.06,
      "step": 909
    },
    {
      "clip_ratio/high_max": 0.002165435485949274,
      "clip_ratio/high_mean": 0.0008746803232497768,
      "clip_ratio/low_mean": 0.0009741417288751109,
      "clip_ratio/low_min": 5.2604104894271586e-05,
      "clip_ratio/region_mean": 0.001848822066676803,
      "epoch": 4.662583843686206,
      "grad_norm": 0.14709343016147614,
      "learning_rate": 1e-06,
      "loss": 0.0095,
      "step": 910
    },
    {
      "clip_ratio/high_max": 0.002896125013648998,
      "clip_ratio/high_mean": 0.001228104272740893,
      "clip_ratio/low_mean": 0.0009200112190228538,
      "clip_ratio/low_min": 4.656675082514994e-05,
      "clip_ratio/region_mean": 0.00214811552723404,
      "epoch": 4.667249927092447,
      "grad_norm": 0.1403370052576065,
      "learning_rate": 1e-06,
      "loss": -0.0347,
      "step": 911
    },
    {
      "clip_ratio/high_max": 0.0025390020018676296,
      "clip_ratio/high_mean": 0.0012303942494327202,
      "clip_ratio/low_mean": 0.0011628588417806895,
      "clip_ratio/low_min": 5.816253542434424e-05,
      "clip_ratio/region_mean": 0.0023932530712045263,
      "epoch": 4.671916010498688,
      "grad_norm": 0.22160589694976807,
      "learning_rate": 1e-06,
      "loss": -0.0357,
      "step": 912
    },
    {
      "clip_ratio/high_max": 0.004892050696071237,
      "clip_ratio/high_mean": 0.001986209419555962,
      "clip_ratio/low_mean": 0.0016969894204521552,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036831988181802444,
      "epoch": 4.676582093904928,
      "grad_norm": 0.104917012155056,
      "learning_rate": 1e-06,
      "loss": -0.0422,
      "step": 913
    },
    {
      "clip_ratio/high_max": 0.004811933271412272,
      "clip_ratio/high_mean": 0.001927438690472627,
      "clip_ratio/low_mean": 0.002063471223664237,
      "clip_ratio/low_min": 0.00014265083882492036,
      "clip_ratio/region_mean": 0.003990909899584949,
      "epoch": 4.681248177311169,
      "grad_norm": 0.10945775359869003,
      "learning_rate": 1e-06,
      "loss": -0.0256,
      "step": 914
    },
    {
      "clip_ratio/high_max": 0.0042466986560611986,
      "clip_ratio/high_mean": 0.0018081294510920998,
      "clip_ratio/low_mean": 0.0019243499027652433,
      "clip_ratio/low_min": 0.0002875029167626053,
      "clip_ratio/region_mean": 0.003732479381142184,
      "epoch": 4.685914260717411,
      "grad_norm": 0.11086112260818481,
      "learning_rate": 1e-06,
      "loss": -0.0412,
      "step": 915
    },
    {
      "clip_ratio/high_max": 0.0041995878927991726,
      "clip_ratio/high_mean": 0.0017002133972709998,
      "clip_ratio/low_mean": 0.002053764088486787,
      "clip_ratio/low_min": 3.422782174311578e-05,
      "clip_ratio/region_mean": 0.003753977500309702,
      "epoch": 4.690580344123651,
      "grad_norm": 0.10955814272165298,
      "learning_rate": 1e-06,
      "loss": 0.0042,
      "step": 916
    },
    {
      "clip_ratio/high_max": 0.005060057876107749,
      "clip_ratio/high_mean": 0.0021903947490500286,
      "clip_ratio/low_mean": 0.0021353217744035646,
      "clip_ratio/low_min": 4.075644028489478e-05,
      "clip_ratio/region_mean": 0.004325716552557424,
      "epoch": 4.695246427529892,
      "grad_norm": 0.13480030000209808,
      "learning_rate": 1e-06,
      "loss": -0.0054,
      "step": 917
    },
    {
      "clip_ratio/high_max": 0.004877847895841114,
      "clip_ratio/high_mean": 0.002042485401034355,
      "clip_ratio/low_mean": 0.0016304570890497416,
      "clip_ratio/low_min": 3.070498496526852e-05,
      "clip_ratio/region_mean": 0.0036729424609802663,
      "epoch": 4.699912510936133,
      "grad_norm": 0.10425883531570435,
      "learning_rate": 1e-06,
      "loss": -0.0435,
      "step": 918
    },
    {
      "clip_ratio/high_max": 0.003946854681998957,
      "clip_ratio/high_mean": 0.0018309454935661051,
      "clip_ratio/low_mean": 0.0018375474792264868,
      "clip_ratio/low_min": 0.0002036843252426479,
      "clip_ratio/region_mean": 0.0036684929218608886,
      "epoch": 4.704578594342374,
      "grad_norm": 0.1084047481417656,
      "learning_rate": 1e-06,
      "loss": -0.0144,
      "step": 919
    },
    {
      "clip_ratio/high_max": 0.004446893581189215,
      "clip_ratio/high_mean": 0.0019434659261605702,
      "clip_ratio/low_mean": 0.002068542980850907,
      "clip_ratio/low_min": 4.321272353990935e-05,
      "clip_ratio/region_mean": 0.004012008881545626,
      "epoch": 4.7092446777486145,
      "grad_norm": 0.1510782539844513,
      "learning_rate": 1e-06,
      "loss": -0.0485,
      "step": 920
    },
    {
      "clip_ratio/high_max": 0.004283720336388797,
      "clip_ratio/high_mean": 0.0018741002641036175,
      "clip_ratio/low_mean": 0.0018227697546535637,
      "clip_ratio/low_min": 7.896795341366669e-05,
      "clip_ratio/region_mean": 0.0036968700151192024,
      "epoch": 4.713910761154856,
      "grad_norm": 0.10664052516222,
      "learning_rate": 1e-06,
      "loss": -0.0171,
      "step": 921
    },
    {
      "clip_ratio/high_max": 0.004387929628137499,
      "clip_ratio/high_mean": 0.001822286933020223,
      "clip_ratio/low_mean": 0.002020908665144816,
      "clip_ratio/low_min": 0.00010714295603975188,
      "clip_ratio/region_mean": 0.0038431956636486575,
      "epoch": 4.718576844561096,
      "grad_norm": 0.1320187747478485,
      "learning_rate": 1e-06,
      "loss": -0.0135,
      "step": 922
    },
    {
      "clip_ratio/high_max": 0.004836086663999595,
      "clip_ratio/high_mean": 0.0018862230645027012,
      "clip_ratio/low_mean": 0.0020955869695171714,
      "clip_ratio/low_min": 0.00019287233226350509,
      "clip_ratio/region_mean": 0.003981809961260296,
      "epoch": 4.723242927967338,
      "grad_norm": 0.1082654595375061,
      "learning_rate": 1e-06,
      "loss": -0.0256,
      "step": 923
    },
    {
      "clip_ratio/high_max": 0.004527326280367561,
      "clip_ratio/high_mean": 0.0018808628992701415,
      "clip_ratio/low_mean": 0.0021697771371691488,
      "clip_ratio/low_min": 0.0001757780737534631,
      "clip_ratio/region_mean": 0.0040506400691810995,
      "epoch": 4.727909011373578,
      "grad_norm": 0.10881368815898895,
      "learning_rate": 1e-06,
      "loss": -0.0121,
      "step": 924
    },
    {
      "clip_ratio/high_max": 0.00433174853969831,
      "clip_ratio/high_mean": 0.0021152673398319166,
      "clip_ratio/low_mean": 0.001989982614759356,
      "clip_ratio/low_min": 7.648932842130307e-05,
      "clip_ratio/region_mean": 0.00410524990002159,
      "epoch": 4.732575094779819,
      "grad_norm": 0.115203358232975,
      "learning_rate": 1e-06,
      "loss": -0.0609,
      "step": 925
    },
    {
      "clip_ratio/high_max": 0.004231540784530807,
      "clip_ratio/high_mean": 0.0017437462593079545,
      "clip_ratio/low_mean": 0.002007946306548547,
      "clip_ratio/low_min": 0.0001181820043711923,
      "clip_ratio/region_mean": 0.0037516925513045862,
      "epoch": 4.73724117818606,
      "grad_norm": 0.11197567731142044,
      "learning_rate": 1e-06,
      "loss": 0.0085,
      "step": 926
    },
    {
      "clip_ratio/high_max": 0.005584392165474128,
      "clip_ratio/high_mean": 0.0022223177875275724,
      "clip_ratio/low_mean": 0.0017691505308903288,
      "clip_ratio/low_min": 6.256367851165123e-05,
      "clip_ratio/region_mean": 0.003991468358435668,
      "epoch": 4.741907261592301,
      "grad_norm": 0.12189286947250366,
      "learning_rate": 1e-06,
      "loss": -0.0355,
      "step": 927
    },
    {
      "clip_ratio/high_max": 0.0050614687643246725,
      "clip_ratio/high_mean": 0.0023151986715674866,
      "clip_ratio/low_mean": 0.0022876011753396597,
      "clip_ratio/low_min": 0.00024112741084536538,
      "clip_ratio/region_mean": 0.004602799934218638,
      "epoch": 4.746573344998541,
      "grad_norm": 0.1318483203649521,
      "learning_rate": 1e-06,
      "loss": -0.0371,
      "step": 928
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0656389508928571,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 3892.0,
      "completions/mean_length": 783.4395141601562,
      "completions/mean_terminated_length": 550.7317504882812,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 4.751239428404783,
      "grad_norm": 0.159715935587883,
      "learning_rate": 1e-06,
      "loss": -0.0382,
      "num_tokens": 275896892.0,
      "reward": 0.638253390789032,
      "reward_std": 0.15481005609035492,
      "rewards/simpleverify_reward/mean": 0.6382533311843872,
      "rewards/simpleverify_reward/std": 0.4805227816104889,
      "step": 929
    },
    {
      "clip_ratio/high_max": 0.0020554618749883957,
      "clip_ratio/high_mean": 0.0008090114497463219,
      "clip_ratio/low_mean": 0.0005496207622854854,
      "clip_ratio/low_min": 2.48065098276129e-05,
      "clip_ratio/region_mean": 0.0013586322456831113,
      "epoch": 4.755905511811024,
      "grad_norm": 0.1484919786453247,
      "learning_rate": 1e-06,
      "loss": -0.0431,
      "step": 930
    },
    {
      "clip_ratio/high_max": 0.0018220637721242383,
      "clip_ratio/high_mean": 0.0006127040596766165,
      "clip_ratio/low_mean": 0.00044348830215312773,
      "clip_ratio/low_min": 1.35398613565485e-05,
      "clip_ratio/region_mean": 0.0010561923627392389,
      "epoch": 4.7605715952172645,
      "grad_norm": 0.14512133598327637,
      "learning_rate": 1e-06,
      "loss": 0.0141,
      "step": 931
    },
    {
      "clip_ratio/high_max": 0.0020306450605858117,
      "clip_ratio/high_mean": 0.0008300516037706984,
      "clip_ratio/low_mean": 0.0006354021197694237,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014654537335445639,
      "epoch": 4.765237678623506,
      "grad_norm": 0.1508951038122177,
      "learning_rate": 1e-06,
      "loss": -0.0547,
      "step": 932
    },
    {
      "clip_ratio/high_max": 0.002150512529624393,
      "clip_ratio/high_mean": 0.000912185809283983,
      "clip_ratio/low_mean": 0.0005712286874768324,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014834145003987942,
      "epoch": 4.769903762029746,
      "grad_norm": 0.13094373047351837,
      "learning_rate": 1e-06,
      "loss": -0.0429,
      "step": 933
    },
    {
      "clip_ratio/high_max": 0.0020429766009328887,
      "clip_ratio/high_mean": 0.0007692655763094081,
      "clip_ratio/low_mean": 0.0006541620823554695,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001423427642293973,
      "epoch": 4.7745698454359875,
      "grad_norm": 0.13760513067245483,
      "learning_rate": 1e-06,
      "loss": -0.032,
      "step": 934
    },
    {
      "clip_ratio/high_max": 0.0021114697156008333,
      "clip_ratio/high_mean": 0.0007809801372786751,
      "clip_ratio/low_mean": 0.00069660631561419,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014775864328839816,
      "epoch": 4.779235928842228,
      "grad_norm": 0.15440140664577484,
      "learning_rate": 1e-06,
      "loss": 0.0005,
      "step": 935
    },
    {
      "clip_ratio/high_max": 0.002296822720381897,
      "clip_ratio/high_mean": 0.0008933182980399579,
      "clip_ratio/low_mean": 0.0006223407153811422,
      "clip_ratio/low_min": 1.8768769223242998e-05,
      "clip_ratio/region_mean": 0.0015156590270635206,
      "epoch": 4.783902012248469,
      "grad_norm": 0.1917096972465515,
      "learning_rate": 1e-06,
      "loss": -0.019,
      "step": 936
    },
    {
      "clip_ratio/high_max": 0.001814560037018964,
      "clip_ratio/high_mean": 0.0008024966682569357,
      "clip_ratio/low_mean": 0.0008624395850347355,
      "clip_ratio/low_min": 4.576175888360012e-05,
      "clip_ratio/region_mean": 0.0016649362369207665,
      "epoch": 4.78856809565471,
      "grad_norm": 0.4116879403591156,
      "learning_rate": 1e-06,
      "loss": 0.0076,
      "step": 937
    },
    {
      "clip_ratio/high_max": 0.002389597320870962,
      "clip_ratio/high_mean": 0.000893748148882878,
      "clip_ratio/low_mean": 0.0008153154285537312,
      "clip_ratio/low_min": 7.735586041235365e-05,
      "clip_ratio/region_mean": 0.0017090635774366092,
      "epoch": 4.793234179060951,
      "grad_norm": 0.16660092771053314,
      "learning_rate": 1e-06,
      "loss": 0.0136,
      "step": 938
    },
    {
      "clip_ratio/high_max": 0.002578572675702162,
      "clip_ratio/high_mean": 0.0010321015943191014,
      "clip_ratio/low_mean": 0.0007902187990111997,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001822320387873333,
      "epoch": 4.797900262467191,
      "grad_norm": 0.13554871082305908,
      "learning_rate": 1e-06,
      "loss": -0.0276,
      "step": 939
    },
    {
      "clip_ratio/high_max": 0.0020533044953481294,
      "clip_ratio/high_mean": 0.0007841348906367784,
      "clip_ratio/low_mean": 0.0006604758827961632,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014446107634285,
      "epoch": 4.802566345873433,
      "grad_norm": 0.13301295042037964,
      "learning_rate": 1e-06,
      "loss": -0.0157,
      "step": 940
    },
    {
      "clip_ratio/high_max": 0.002186465288104955,
      "clip_ratio/high_mean": 0.000920623836464074,
      "clip_ratio/low_mean": 0.0007418143559334567,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001662438200582983,
      "epoch": 4.807232429279673,
      "grad_norm": 0.12652911245822906,
      "learning_rate": 1e-06,
      "loss": -0.0029,
      "step": 941
    },
    {
      "clip_ratio/high_max": 0.002254081457067514,
      "clip_ratio/high_mean": 0.0009264382424589712,
      "clip_ratio/low_mean": 0.0008024652561289258,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001728903480398003,
      "epoch": 4.811898512685914,
      "grad_norm": 0.13703419268131256,
      "learning_rate": 1e-06,
      "loss": -0.0249,
      "step": 942
    },
    {
      "clip_ratio/high_max": 0.002292283003043849,
      "clip_ratio/high_mean": 0.0009604744955140632,
      "clip_ratio/low_mean": 0.0008194325037038652,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017799070119508542,
      "epoch": 4.816564596092155,
      "grad_norm": 0.14402545988559723,
      "learning_rate": 1e-06,
      "loss": -0.0348,
      "step": 943
    },
    {
      "clip_ratio/high_max": 0.0023272093385457993,
      "clip_ratio/high_mean": 0.0009783410023374017,
      "clip_ratio/low_mean": 0.0007571067553726607,
      "clip_ratio/low_min": 1.5006002286099829e-05,
      "clip_ratio/region_mean": 0.0017354477822664194,
      "epoch": 4.821230679498396,
      "grad_norm": 0.14965122938156128,
      "learning_rate": 1e-06,
      "loss": 0.0011,
      "step": 944
    },
    {
      "clip_ratio/high_max": 0.0049721550167305395,
      "clip_ratio/high_mean": 0.0020464786102820653,
      "clip_ratio/low_mean": 0.0018046193799818866,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003851098066661507,
      "epoch": 4.8258967629046365,
      "grad_norm": 0.10149649530649185,
      "learning_rate": 1e-06,
      "loss": -0.0391,
      "step": 945
    },
    {
      "clip_ratio/high_max": 0.005088968478958122,
      "clip_ratio/high_mean": 0.0019277975079603493,
      "clip_ratio/low_mean": 0.0016199627498281188,
      "clip_ratio/low_min": 6.703131657559425e-05,
      "clip_ratio/region_mean": 0.0035477601923048496,
      "epoch": 4.830562846310878,
      "grad_norm": 0.10678434371948242,
      "learning_rate": 1e-06,
      "loss": -0.044,
      "step": 946
    },
    {
      "clip_ratio/high_max": 0.00415263109607622,
      "clip_ratio/high_mean": 0.0015010385832283646,
      "clip_ratio/low_mean": 0.0016842340974108083,
      "clip_ratio/low_min": 5.060728653916158e-05,
      "clip_ratio/region_mean": 0.00318527268973412,
      "epoch": 4.835228929717118,
      "grad_norm": 0.10137312859296799,
      "learning_rate": 1e-06,
      "loss": 0.0132,
      "step": 947
    },
    {
      "clip_ratio/high_max": 0.004087245135451667,
      "clip_ratio/high_mean": 0.0017850204712885898,
      "clip_ratio/low_mean": 0.0016947056246863212,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034797260814229958,
      "epoch": 4.83989501312336,
      "grad_norm": 0.11224978417158127,
      "learning_rate": 1e-06,
      "loss": -0.0556,
      "step": 948
    },
    {
      "clip_ratio/high_max": 0.004553834689431824,
      "clip_ratio/high_mean": 0.001844798811362125,
      "clip_ratio/low_mean": 0.0015333011797338258,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033781000674935058,
      "epoch": 4.844561096529601,
      "grad_norm": 0.09770405292510986,
      "learning_rate": 1e-06,
      "loss": -0.0436,
      "step": 949
    },
    {
      "clip_ratio/high_max": 0.004226598524837755,
      "clip_ratio/high_mean": 0.0016337919259967748,
      "clip_ratio/low_mean": 0.001581754527251178,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0032155465087271295,
      "epoch": 4.849227179935841,
      "grad_norm": 18.484195709228516,
      "learning_rate": 1e-06,
      "loss": -0.0285,
      "step": 950
    },
    {
      "clip_ratio/high_max": 0.004315058395150118,
      "clip_ratio/high_mean": 0.0017018895014189184,
      "clip_ratio/low_mean": 0.0019697889147209935,
      "clip_ratio/low_min": 2.8280543119763024e-05,
      "clip_ratio/region_mean": 0.0036716784379677847,
      "epoch": 4.853893263342083,
      "grad_norm": 0.10661263018846512,
      "learning_rate": 1e-06,
      "loss": -0.0005,
      "step": 951
    },
    {
      "clip_ratio/high_max": 0.004197582638880704,
      "clip_ratio/high_mean": 0.0016967068913800176,
      "clip_ratio/low_mean": 0.0016142671956913546,
      "clip_ratio/low_min": 9.007618064060807e-05,
      "clip_ratio/region_mean": 0.0033109740979853086,
      "epoch": 4.858559346748323,
      "grad_norm": 0.11126602441072464,
      "learning_rate": 1e-06,
      "loss": -0.0199,
      "step": 952
    },
    {
      "clip_ratio/high_max": 0.004582642883178778,
      "clip_ratio/high_mean": 0.0018451784017088357,
      "clip_ratio/low_mean": 0.0023961000188137405,
      "clip_ratio/low_min": 0.00010348653813707642,
      "clip_ratio/region_mean": 0.0042412783950567245,
      "epoch": 4.863225430154564,
      "grad_norm": 0.1211385428905487,
      "learning_rate": 1e-06,
      "loss": 0.0063,
      "step": 953
    },
    {
      "clip_ratio/high_max": 0.0047500235014013015,
      "clip_ratio/high_mean": 0.0017259489686693996,
      "clip_ratio/low_mean": 0.002125600480212597,
      "clip_ratio/low_min": 0.0001804160656320164,
      "clip_ratio/region_mean": 0.003851549423416145,
      "epoch": 4.867891513560805,
      "grad_norm": 0.11494329571723938,
      "learning_rate": 1e-06,
      "loss": 0.0126,
      "step": 954
    },
    {
      "clip_ratio/high_max": 0.0045973342639626935,
      "clip_ratio/high_mean": 0.0018686926341615617,
      "clip_ratio/low_mean": 0.001895394267194206,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003764086897717789,
      "epoch": 4.872557596967046,
      "grad_norm": 0.10718845576047897,
      "learning_rate": 1e-06,
      "loss": -0.0283,
      "step": 955
    },
    {
      "clip_ratio/high_max": 0.004053079588629771,
      "clip_ratio/high_mean": 0.0014844440484012011,
      "clip_ratio/low_mean": 0.0017039077829394955,
      "clip_ratio/low_min": 8.639415318612009e-05,
      "clip_ratio/region_mean": 0.003188351824064739,
      "epoch": 4.8772236803732865,
      "grad_norm": 0.10361669957637787,
      "learning_rate": 1e-06,
      "loss": -0.0164,
      "step": 956
    },
    {
      "clip_ratio/high_max": 0.004060480059706606,
      "clip_ratio/high_mean": 0.0017420621479686815,
      "clip_ratio/low_mean": 0.0017068008855858352,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003448863048106432,
      "epoch": 4.881889763779528,
      "grad_norm": 0.09397058188915253,
      "learning_rate": 1e-06,
      "loss": -0.0035,
      "step": 957
    },
    {
      "clip_ratio/high_max": 0.004708353015303146,
      "clip_ratio/high_mean": 0.0019292090946692042,
      "clip_ratio/low_mean": 0.0017499009336461313,
      "clip_ratio/low_min": 3.354204091010615e-05,
      "clip_ratio/region_mean": 0.0036791100428672507,
      "epoch": 4.886555847185768,
      "grad_norm": 0.837611198425293,
      "learning_rate": 1e-06,
      "loss": -0.0253,
      "step": 958
    },
    {
      "clip_ratio/high_max": 0.004463962875888683,
      "clip_ratio/high_mean": 0.0019200675742467865,
      "clip_ratio/low_mean": 0.0018050010803563055,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003725068731000647,
      "epoch": 4.8912219305920095,
      "grad_norm": 0.10971732437610626,
      "learning_rate": 1e-06,
      "loss": -0.0356,
      "step": 959
    },
    {
      "clip_ratio/high_max": 0.004368073656223714,
      "clip_ratio/high_mean": 0.0017574460252944846,
      "clip_ratio/low_mean": 0.001854871938121505,
      "clip_ratio/low_min": 3.480924715404399e-05,
      "clip_ratio/region_mean": 0.0036123179525020532,
      "epoch": 4.89588801399825,
      "grad_norm": 0.10876951366662979,
      "learning_rate": 1e-06,
      "loss": 0.0003,
      "step": 960
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0673130580357143,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4094.0,
      "completions/mean_length": 787.8792114257812,
      "completions/mean_terminated_length": 549.12841796875,
      "completions/min_length": 75.0,
      "completions/min_terminated_length": 75.0,
      "epoch": 5.004666083406241,
      "grad_norm": 0.15265098214149475,
      "learning_rate": 1e-06,
      "loss": -0.0605,
      "num_tokens": 284643888.0,
      "reward": 0.6535993814468384,
      "reward_std": 0.15969029068946838,
      "rewards/simpleverify_reward/mean": 0.6535993218421936,
      "rewards/simpleverify_reward/std": 0.4758393168449402,
      "step": 961
    },
    {
      "clip_ratio/high_max": 0.0015449587008333765,
      "clip_ratio/high_mean": 0.0006328750387183391,
      "clip_ratio/low_mean": 0.00038651525574096013,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001019390307192225,
      "epoch": 5.009332166812482,
      "grad_norm": 0.1369791328907013,
      "learning_rate": 1e-06,
      "loss": -0.0003,
      "step": 962
    },
    {
      "clip_ratio/high_max": 0.002080726913845865,
      "clip_ratio/high_mean": 0.0007893530382716563,
      "clip_ratio/low_mean": 0.00048584852174826665,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012752015827572905,
      "epoch": 5.013998250218723,
      "grad_norm": 0.16443632543087006,
      "learning_rate": 1e-06,
      "loss": -0.0112,
      "step": 963
    },
    {
      "clip_ratio/high_max": 0.002666403310286114,
      "clip_ratio/high_mean": 0.00095054303528741,
      "clip_ratio/low_mean": 0.0005697498290828662,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015202929062070325,
      "epoch": 5.0186643336249634,
      "grad_norm": 0.15811331570148468,
      "learning_rate": 1e-06,
      "loss": -0.0324,
      "step": 964
    },
    {
      "clip_ratio/high_max": 0.0020964365467079915,
      "clip_ratio/high_mean": 0.0008281324171548476,
      "clip_ratio/low_mean": 0.0006010557517583948,
      "clip_ratio/low_min": 1.5439723938470706e-05,
      "clip_ratio/region_mean": 0.001429188203474041,
      "epoch": 5.023330417031205,
      "grad_norm": 0.15350686013698578,
      "learning_rate": 1e-06,
      "loss": -0.0444,
      "step": 965
    },
    {
      "clip_ratio/high_max": 0.0019612005635281093,
      "clip_ratio/high_mean": 0.0007759465188428294,
      "clip_ratio/low_mean": 0.0006645556522926199,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014405021938728169,
      "epoch": 5.027996500437445,
      "grad_norm": 0.16173428297042847,
      "learning_rate": 1e-06,
      "loss": -0.0141,
      "step": 966
    },
    {
      "clip_ratio/high_max": 0.0020538765675155446,
      "clip_ratio/high_mean": 0.0008916555689211236,
      "clip_ratio/low_mean": 0.0006211086874827743,
      "clip_ratio/low_min": 2.3773298380547203e-05,
      "clip_ratio/region_mean": 0.0015127642800507601,
      "epoch": 5.0326625838436865,
      "grad_norm": 0.3055776059627533,
      "learning_rate": 1e-06,
      "loss": -0.0427,
      "step": 967
    },
    {
      "clip_ratio/high_max": 0.002021764033997897,
      "clip_ratio/high_mean": 0.0008895534883777145,
      "clip_ratio/low_mean": 0.0007763721532683121,
      "clip_ratio/low_min": 1.604621320439037e-05,
      "clip_ratio/region_mean": 0.0016659256580169313,
      "epoch": 5.037328667249927,
      "grad_norm": 0.15162530541419983,
      "learning_rate": 1e-06,
      "loss": -0.0229,
      "step": 968
    },
    {
      "clip_ratio/high_max": 0.0023166202881839126,
      "clip_ratio/high_mean": 0.0010773405392683344,
      "clip_ratio/low_mean": 0.0006841995609647711,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017615400975046214,
      "epoch": 5.041994750656168,
      "grad_norm": 0.15083321928977966,
      "learning_rate": 1e-06,
      "loss": -0.0519,
      "step": 969
    },
    {
      "clip_ratio/high_max": 0.002525533207517583,
      "clip_ratio/high_mean": 0.0009593003469490213,
      "clip_ratio/low_mean": 0.0007786151818436338,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017379155178787187,
      "epoch": 5.046660834062409,
      "grad_norm": 0.14743049442768097,
      "learning_rate": 1e-06,
      "loss": 0.0032,
      "step": 970
    },
    {
      "clip_ratio/high_max": 0.0021769272279925644,
      "clip_ratio/high_mean": 0.0008751045024837367,
      "clip_ratio/low_mean": 0.000754498791138758,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016296032808895689,
      "epoch": 5.05132691746865,
      "grad_norm": 0.15316613018512726,
      "learning_rate": 1e-06,
      "loss": 0.0007,
      "step": 971
    },
    {
      "clip_ratio/high_max": 0.002614977056509815,
      "clip_ratio/high_mean": 0.0010195441718678921,
      "clip_ratio/low_mean": 0.0007461757559212856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017657199496170506,
      "epoch": 5.05599300087489,
      "grad_norm": 0.1445103883743286,
      "learning_rate": 1e-06,
      "loss": -0.0281,
      "step": 972
    },
    {
      "clip_ratio/high_max": 0.002598426493932493,
      "clip_ratio/high_mean": 0.0011274473254161421,
      "clip_ratio/low_mean": 0.0007086557998263743,
      "clip_ratio/low_min": 3.5561879485612735e-05,
      "clip_ratio/region_mean": 0.0018361031798121985,
      "epoch": 5.060659084281132,
      "grad_norm": 0.16397222876548767,
      "learning_rate": 1e-06,
      "loss": -0.065,
      "step": 973
    },
    {
      "clip_ratio/high_max": 0.0025484296202193946,
      "clip_ratio/high_mean": 0.0009887233918561833,
      "clip_ratio/low_mean": 0.0010092802231156384,
      "clip_ratio/low_min": 1.4702422959089745e-05,
      "clip_ratio/region_mean": 0.001998003637709189,
      "epoch": 5.065325167687372,
      "grad_norm": 0.15653736889362335,
      "learning_rate": 1e-06,
      "loss": 0.0044,
      "step": 974
    },
    {
      "clip_ratio/high_max": 0.0026231064257444814,
      "clip_ratio/high_mean": 0.0012304056035645772,
      "clip_ratio/low_mean": 0.0009291319292970002,
      "clip_ratio/low_min": 2.3561542548122816e-05,
      "clip_ratio/region_mean": 0.002159537500119768,
      "epoch": 5.069991251093613,
      "grad_norm": 0.1562759131193161,
      "learning_rate": 1e-06,
      "loss": -0.0305,
      "step": 975
    },
    {
      "clip_ratio/high_max": 0.00226770499284612,
      "clip_ratio/high_mean": 0.0009163102477032226,
      "clip_ratio/low_mean": 0.0009694444834167371,
      "clip_ratio/low_min": 3.327051854284946e-05,
      "clip_ratio/region_mean": 0.0018857547365769278,
      "epoch": 5.074657334499854,
      "grad_norm": 0.1452087163925171,
      "learning_rate": 1e-06,
      "loss": 0.009,
      "step": 976
    },
    {
      "clip_ratio/high_max": 0.005008928033930715,
      "clip_ratio/high_mean": 0.0020603489101631567,
      "clip_ratio/low_mean": 0.0016285567744489526,
      "clip_ratio/low_min": 4.883575456915423e-05,
      "clip_ratio/region_mean": 0.003688905722810887,
      "epoch": 5.079323417906095,
      "grad_norm": 0.10370133817195892,
      "learning_rate": 1e-06,
      "loss": -0.0613,
      "step": 977
    },
    {
      "clip_ratio/high_max": 0.0036495929598459043,
      "clip_ratio/high_mean": 0.0015194915413303534,
      "clip_ratio/low_mean": 0.0015842710636206903,
      "clip_ratio/low_min": 0.00014450841081270482,
      "clip_ratio/region_mean": 0.003103762515820563,
      "epoch": 5.083989501312336,
      "grad_norm": 0.1010274812579155,
      "learning_rate": 1e-06,
      "loss": -0.001,
      "step": 978
    },
    {
      "clip_ratio/high_max": 0.0047291311420849524,
      "clip_ratio/high_mean": 0.0018774631898850203,
      "clip_ratio/low_mean": 0.0018876308276958298,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037650940648745745,
      "epoch": 5.088655584718577,
      "grad_norm": 0.10905813425779343,
      "learning_rate": 1e-06,
      "loss": -0.0121,
      "step": 979
    },
    {
      "clip_ratio/high_max": 0.0047671905340394005,
      "clip_ratio/high_mean": 0.0019036019148188643,
      "clip_ratio/low_mean": 0.0018992231380252633,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038028250564821064,
      "epoch": 5.093321668124818,
      "grad_norm": 0.11359624564647675,
      "learning_rate": 1e-06,
      "loss": -0.0333,
      "step": 980
    },
    {
      "clip_ratio/high_max": 0.00482658707187511,
      "clip_ratio/high_mean": 0.0017627755369176157,
      "clip_ratio/low_mean": 0.0017244030241272412,
      "clip_ratio/low_min": 7.458234176738188e-05,
      "clip_ratio/region_mean": 0.0034871784882852808,
      "epoch": 5.0979877515310585,
      "grad_norm": 0.11160120368003845,
      "learning_rate": 1e-06,
      "loss": -0.0453,
      "step": 981
    },
    {
      "clip_ratio/high_max": 0.004728345738840289,
      "clip_ratio/high_mean": 0.0018364019997534342,
      "clip_ratio/low_mean": 0.0017556422608322464,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035920441732741892,
      "epoch": 5.1026538349373,
      "grad_norm": 0.10930287092924118,
      "learning_rate": 1e-06,
      "loss": -0.015,
      "step": 982
    },
    {
      "clip_ratio/high_max": 0.004754893736389931,
      "clip_ratio/high_mean": 0.0018541152639954817,
      "clip_ratio/low_mean": 0.0016714120574761182,
      "clip_ratio/low_min": 8.838607027428225e-05,
      "clip_ratio/region_mean": 0.0035255272887297906,
      "epoch": 5.10731991834354,
      "grad_norm": 0.11352820694446564,
      "learning_rate": 1e-06,
      "loss": -0.0436,
      "step": 983
    },
    {
      "clip_ratio/high_max": 0.004237971021211706,
      "clip_ratio/high_mean": 0.0018223492006654851,
      "clip_ratio/low_mean": 0.0018080762565659825,
      "clip_ratio/low_min": 0.00011208964497200213,
      "clip_ratio/region_mean": 0.0036304254026617855,
      "epoch": 5.111986001749782,
      "grad_norm": 0.10900255292654037,
      "learning_rate": 1e-06,
      "loss": -0.0237,
      "step": 984
    },
    {
      "clip_ratio/high_max": 0.004388932880829088,
      "clip_ratio/high_mean": 0.0020051638675795402,
      "clip_ratio/low_mean": 0.0015720104820502456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035771743714576587,
      "epoch": 5.116652085156022,
      "grad_norm": 0.1130501925945282,
      "learning_rate": 1e-06,
      "loss": -0.0528,
      "step": 985
    },
    {
      "clip_ratio/high_max": 0.004771657433593646,
      "clip_ratio/high_mean": 0.0018092952741426416,
      "clip_ratio/low_mean": 0.0019510745405568741,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003760369843803346,
      "epoch": 5.121318168562263,
      "grad_norm": 0.10710525512695312,
      "learning_rate": 1e-06,
      "loss": 0.0023,
      "step": 986
    },
    {
      "clip_ratio/high_max": 0.004259601439116523,
      "clip_ratio/high_mean": 0.0016855404355737846,
      "clip_ratio/low_mean": 0.0017088917084038258,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033944321985472925,
      "epoch": 5.125984251968504,
      "grad_norm": 0.10002481192350388,
      "learning_rate": 1e-06,
      "loss": -0.0001,
      "step": 987
    },
    {
      "clip_ratio/high_max": 0.00473655374662485,
      "clip_ratio/high_mean": 0.0018687648516788613,
      "clip_ratio/low_mean": 0.001634989010199206,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003503753847326152,
      "epoch": 5.130650335374745,
      "grad_norm": 0.10728838294744492,
      "learning_rate": 1e-06,
      "loss": -0.0289,
      "step": 988
    },
    {
      "clip_ratio/high_max": 0.005118130226037465,
      "clip_ratio/high_mean": 0.002142934983567102,
      "clip_ratio/low_mean": 0.0016328284436895046,
      "clip_ratio/low_min": 7.112375897122547e-05,
      "clip_ratio/region_mean": 0.003775763470912352,
      "epoch": 5.135316418780985,
      "grad_norm": 0.11420293897390366,
      "learning_rate": 1e-06,
      "loss": -0.066,
      "step": 989
    },
    {
      "clip_ratio/high_max": 0.0044603657006518915,
      "clip_ratio/high_mean": 0.0018084128423652146,
      "clip_ratio/low_mean": 0.0020731912227347493,
      "clip_ratio/low_min": 3.426535113248974e-05,
      "clip_ratio/region_mean": 0.0038816040760139003,
      "epoch": 5.139982502187227,
      "grad_norm": 0.11369369924068451,
      "learning_rate": 1e-06,
      "loss": 0.0034,
      "step": 990
    },
    {
      "clip_ratio/high_max": 0.004287930874852464,
      "clip_ratio/high_mean": 0.0019494573280098848,
      "clip_ratio/low_mean": 0.002007729548495263,
      "clip_ratio/low_min": 6.232415398699231e-05,
      "clip_ratio/region_mean": 0.003957186854677275,
      "epoch": 5.144648585593467,
      "grad_norm": 0.11844845861196518,
      "learning_rate": 1e-06,
      "loss": -0.0315,
      "step": 991
    },
    {
      "clip_ratio/high_max": 0.003953901003114879,
      "clip_ratio/high_mean": 0.0016898707726795692,
      "clip_ratio/low_mean": 0.0020398725173436105,
      "clip_ratio/low_min": 0.00010927770199486986,
      "clip_ratio/region_mean": 0.0037297434464562684,
      "epoch": 5.1493146689997085,
      "grad_norm": 0.11048706620931625,
      "learning_rate": 1e-06,
      "loss": 0.0082,
      "step": 992
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0729631696428571,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 3814.0,
      "completions/mean_length": 819.5020141601562,
      "completions/mean_terminated_length": 561.62255859375,
      "completions/min_length": 93.0,
      "completions/min_terminated_length": 93.0,
      "epoch": 5.153980752405949,
      "grad_norm": 0.17419984936714172,
      "learning_rate": 1e-06,
      "loss": -0.0407,
      "num_tokens": 293529060.0,
      "reward": 0.638741672039032,
      "reward_std": 0.1680195927619934,
      "rewards/simpleverify_reward/mean": 0.6387416124343872,
      "rewards/simpleverify_reward/std": 0.4803819954395294,
      "step": 993
    },
    {
      "clip_ratio/high_max": 0.002075881082419073,
      "clip_ratio/high_mean": 0.0008263601012004074,
      "clip_ratio/low_mean": 0.0005146906523805228,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001341050759947393,
      "epoch": 5.15864683581219,
      "grad_norm": 0.15556688606739044,
      "learning_rate": 1e-06,
      "loss": -0.0246,
      "step": 994
    },
    {
      "clip_ratio/high_max": 0.002147895091184182,
      "clip_ratio/high_mean": 0.0009027577998494962,
      "clip_ratio/low_mean": 0.0004864289367105812,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013891867456550244,
      "epoch": 5.163312919218431,
      "grad_norm": 0.1463095247745514,
      "learning_rate": 1e-06,
      "loss": -0.0561,
      "step": 995
    },
    {
      "clip_ratio/high_max": 0.0020475842684390955,
      "clip_ratio/high_mean": 0.0007753284135105787,
      "clip_ratio/low_mean": 0.0006058864819351584,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001381214904540684,
      "epoch": 5.167979002624672,
      "grad_norm": 0.16062968969345093,
      "learning_rate": 1e-06,
      "loss": -0.0059,
      "step": 996
    },
    {
      "clip_ratio/high_max": 0.0023307400806515943,
      "clip_ratio/high_mean": 0.0008343964818777749,
      "clip_ratio/low_mean": 0.0005391193506056879,
      "clip_ratio/low_min": 1.2852148756792303e-05,
      "clip_ratio/region_mean": 0.0013735158354393207,
      "epoch": 5.172645086030913,
      "grad_norm": 0.1553787887096405,
      "learning_rate": 1e-06,
      "loss": -0.046,
      "step": 997
    },
    {
      "clip_ratio/high_max": 0.0022699056935380213,
      "clip_ratio/high_mean": 0.0009314027211075881,
      "clip_ratio/low_mean": 0.0005514473541552434,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014828500752628315,
      "epoch": 5.177311169437154,
      "grad_norm": 0.1535441279411316,
      "learning_rate": 1e-06,
      "loss": -0.0533,
      "step": 998
    },
    {
      "clip_ratio/high_max": 0.0021797854351461865,
      "clip_ratio/high_mean": 0.000865523954416858,
      "clip_ratio/low_mean": 0.0005951901521257241,
      "clip_ratio/low_min": 1.8882175936596468e-05,
      "clip_ratio/region_mean": 0.001460714076529257,
      "epoch": 5.181977252843395,
      "grad_norm": 0.1486525535583496,
      "learning_rate": 1e-06,
      "loss": -0.0368,
      "step": 999
    },
    {
      "clip_ratio/high_max": 0.001936640423082281,
      "clip_ratio/high_mean": 0.0007841346305212937,
      "clip_ratio/low_mean": 0.0006156760136946104,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013998106733197346,
      "epoch": 5.186643336249635,
      "grad_norm": 0.12503276765346527,
      "learning_rate": 1e-06,
      "loss": -0.0421,
      "step": 1000
    },
    {
      "clip_ratio/high_max": 0.0020422697707545012,
      "clip_ratio/high_mean": 0.0008638777890155325,
      "clip_ratio/low_mean": 0.0008741974161239341,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00173807522514835,
      "epoch": 5.191309419655877,
      "grad_norm": 0.16644065082073212,
      "learning_rate": 1e-06,
      "loss": 0.0336,
      "step": 1001
    },
    {
      "clip_ratio/high_max": 0.0027344133777660318,
      "clip_ratio/high_mean": 0.0011106047677458264,
      "clip_ratio/low_mean": 0.0006984382453083526,
      "clip_ratio/low_min": 4.552203245111741e-05,
      "clip_ratio/region_mean": 0.001809042994864285,
      "epoch": 5.195975503062117,
      "grad_norm": 0.1430450826883316,
      "learning_rate": 1e-06,
      "loss": -0.0494,
      "step": 1002
    },
    {
      "clip_ratio/high_max": 0.002473121763614472,
      "clip_ratio/high_mean": 0.0010891216006712057,
      "clip_ratio/low_mean": 0.0008091115796560189,
      "clip_ratio/low_min": 2.7789889827545267e-05,
      "clip_ratio/region_mean": 0.001898233160318341,
      "epoch": 5.200641586468358,
      "grad_norm": 0.15866203606128693,
      "learning_rate": 1e-06,
      "loss": -0.0535,
      "step": 1003
    },
    {
      "clip_ratio/high_max": 0.002856684586731717,
      "clip_ratio/high_mean": 0.0009380237552250037,
      "clip_ratio/low_mean": 0.0008077421898633474,
      "clip_ratio/low_min": 2.752552245510742e-05,
      "clip_ratio/region_mean": 0.0017457659414503723,
      "epoch": 5.205307669874599,
      "grad_norm": 0.15228678286075592,
      "learning_rate": 1e-06,
      "loss": -0.0054,
      "step": 1004
    },
    {
      "clip_ratio/high_max": 0.002359831109060906,
      "clip_ratio/high_mean": 0.0010075241279992042,
      "clip_ratio/low_mean": 0.0008300671506731305,
      "clip_ratio/low_min": 1.8662287402548827e-05,
      "clip_ratio/region_mean": 0.0018375912695773877,
      "epoch": 5.20997375328084,
      "grad_norm": 0.19849438965320587,
      "learning_rate": 1e-06,
      "loss": 0.0139,
      "step": 1005
    },
    {
      "clip_ratio/high_max": 0.002412451616692124,
      "clip_ratio/high_mean": 0.0009395330635015853,
      "clip_ratio/low_mean": 0.0010428279365441995,
      "clip_ratio/low_min": 1.710454307612963e-05,
      "clip_ratio/region_mean": 0.0019823610127787106,
      "epoch": 5.2146398366870805,
      "grad_norm": 0.17778748273849487,
      "learning_rate": 1e-06,
      "loss": -0.0024,
      "step": 1006
    },
    {
      "clip_ratio/high_max": 0.002688764285267098,
      "clip_ratio/high_mean": 0.0011151650487590814,
      "clip_ratio/low_mean": 0.0008966112636699108,
      "clip_ratio/low_min": 4.430529224919155e-05,
      "clip_ratio/region_mean": 0.0020117762323934585,
      "epoch": 5.219305920093322,
      "grad_norm": 0.1618351936340332,
      "learning_rate": 1e-06,
      "loss": -0.0314,
      "step": 1007
    },
    {
      "clip_ratio/high_max": 0.002520723966881633,
      "clip_ratio/high_mean": 0.0010178906759392703,
      "clip_ratio/low_mean": 0.0008467458392260596,
      "clip_ratio/low_min": 4.185410580248572e-05,
      "clip_ratio/region_mean": 0.0018646365206222981,
      "epoch": 5.223972003499562,
      "grad_norm": 0.14499059319496155,
      "learning_rate": 1e-06,
      "loss": -0.0371,
      "step": 1008
    },
    {
      "clip_ratio/high_max": 0.0055825999297667295,
      "clip_ratio/high_mean": 0.002216798107838258,
      "clip_ratio/low_mean": 0.0020892682441626675,
      "clip_ratio/low_min": 0.0002328925475012511,
      "clip_ratio/region_mean": 0.004306066344724968,
      "epoch": 5.228638086905804,
      "grad_norm": 0.11005309969186783,
      "learning_rate": 1e-06,
      "loss": -0.0417,
      "step": 1009
    },
    {
      "clip_ratio/high_max": 0.004347499430878088,
      "clip_ratio/high_mean": 0.0017518723761895671,
      "clip_ratio/low_mean": 0.0017631687187531497,
      "clip_ratio/low_min": 1.0877131899178494e-05,
      "clip_ratio/region_mean": 0.003515041171340272,
      "epoch": 5.233304170312044,
      "grad_norm": 0.1101030558347702,
      "learning_rate": 1e-06,
      "loss": -0.0255,
      "step": 1010
    },
    {
      "clip_ratio/high_max": 0.0043478384031914175,
      "clip_ratio/high_mean": 0.0018308321341464762,
      "clip_ratio/low_mean": 0.0015104327831068076,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033412650373065844,
      "epoch": 5.237970253718285,
      "grad_norm": 0.10577068477869034,
      "learning_rate": 1e-06,
      "loss": -0.0568,
      "step": 1011
    },
    {
      "clip_ratio/high_max": 0.004228475634590723,
      "clip_ratio/high_mean": 0.001662406968534924,
      "clip_ratio/low_mean": 0.001927349188918015,
      "clip_ratio/low_min": 5.152514495421201e-05,
      "clip_ratio/region_mean": 0.0035897560883313417,
      "epoch": 5.242636337124526,
      "grad_norm": 0.11040983349084854,
      "learning_rate": 1e-06,
      "loss": -0.0068,
      "step": 1012
    },
    {
      "clip_ratio/high_max": 0.004759146053402219,
      "clip_ratio/high_mean": 0.0018668192642508075,
      "clip_ratio/low_mean": 0.001728527121940715,
      "clip_ratio/low_min": 9.341413169750012e-05,
      "clip_ratio/region_mean": 0.003595346410293132,
      "epoch": 5.247302420530767,
      "grad_norm": 0.11569134891033173,
      "learning_rate": 1e-06,
      "loss": -0.047,
      "step": 1013
    },
    {
      "clip_ratio/high_max": 0.004707593601779081,
      "clip_ratio/high_mean": 0.001989219083043281,
      "clip_ratio/low_mean": 0.001605293608008651,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003594512731069699,
      "epoch": 5.251968503937007,
      "grad_norm": 0.11057382822036743,
      "learning_rate": 1e-06,
      "loss": -0.0542,
      "step": 1014
    },
    {
      "clip_ratio/high_max": 0.004848193901125342,
      "clip_ratio/high_mean": 0.001878933770058211,
      "clip_ratio/low_mean": 0.0016232934976869728,
      "clip_ratio/low_min": 9.441087604500353e-05,
      "clip_ratio/region_mean": 0.003502227205899544,
      "epoch": 5.256634587343249,
      "grad_norm": 0.10582774132490158,
      "learning_rate": 1e-06,
      "loss": -0.0375,
      "step": 1015
    },
    {
      "clip_ratio/high_max": 0.004101810191059485,
      "clip_ratio/high_mean": 0.001611021340067964,
      "clip_ratio/low_mean": 0.0015275996047421359,
      "clip_ratio/low_min": 3.7821482692379504e-05,
      "clip_ratio/region_mean": 0.0031386208938783966,
      "epoch": 5.26130067074949,
      "grad_norm": 0.09692647308111191,
      "learning_rate": 1e-06,
      "loss": -0.0428,
      "step": 1016
    },
    {
      "clip_ratio/high_max": 0.004056684148963541,
      "clip_ratio/high_mean": 0.0016346950651495717,
      "clip_ratio/low_mean": 0.002188072379794903,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038227673794608563,
      "epoch": 5.2659667541557305,
      "grad_norm": 0.12245281040668488,
      "learning_rate": 1e-06,
      "loss": 0.0326,
      "step": 1017
    },
    {
      "clip_ratio/high_max": 0.004675952921388671,
      "clip_ratio/high_mean": 0.0020794166302948724,
      "clip_ratio/low_mean": 0.0016500580404681386,
      "clip_ratio/low_min": 0.00014556574751622975,
      "clip_ratio/region_mean": 0.003729474628926255,
      "epoch": 5.270632837561972,
      "grad_norm": 0.10504581034183502,
      "learning_rate": 1e-06,
      "loss": -0.0502,
      "step": 1018
    },
    {
      "clip_ratio/high_max": 0.00457784635364078,
      "clip_ratio/high_mean": 0.002114764240104705,
      "clip_ratio/low_mean": 0.001816475563828135,
      "clip_ratio/low_min": 0.0001811290203477256,
      "clip_ratio/region_mean": 0.003931239814846776,
      "epoch": 5.275298920968212,
      "grad_norm": 0.11754226684570312,
      "learning_rate": 1e-06,
      "loss": -0.0544,
      "step": 1019
    },
    {
      "clip_ratio/high_max": 0.005019220378017053,
      "clip_ratio/high_mean": 0.0017417570961697493,
      "clip_ratio/low_mean": 0.0017731974403432105,
      "clip_ratio/low_min": 5.505104491021484e-05,
      "clip_ratio/region_mean": 0.0035149546019965783,
      "epoch": 5.2799650043744535,
      "grad_norm": 0.10802837461233139,
      "learning_rate": 1e-06,
      "loss": -0.0062,
      "step": 1020
    },
    {
      "clip_ratio/high_max": 0.0049049765220843256,
      "clip_ratio/high_mean": 0.0018938581051770598,
      "clip_ratio/low_mean": 0.002084886218653992,
      "clip_ratio/low_min": 4.9710111852618866e-05,
      "clip_ratio/region_mean": 0.003978744367486797,
      "epoch": 5.284631087780694,
      "grad_norm": 0.12198702245950699,
      "learning_rate": 1e-06,
      "loss": 0.0127,
      "step": 1021
    },
    {
      "clip_ratio/high_max": 0.004180621843261179,
      "clip_ratio/high_mean": 0.0016994495381368324,
      "clip_ratio/low_mean": 0.002120194199960679,
      "clip_ratio/low_min": 0.00022694712606607936,
      "clip_ratio/region_mean": 0.003819643723545596,
      "epoch": 5.289297171186935,
      "grad_norm": 0.12182199209928513,
      "learning_rate": 1e-06,
      "loss": -0.0035,
      "step": 1022
    },
    {
      "clip_ratio/high_max": 0.005215120690991171,
      "clip_ratio/high_mean": 0.002139085489034187,
      "clip_ratio/low_mean": 0.0019649441965157166,
      "clip_ratio/low_min": 4.937783887726255e-05,
      "clip_ratio/region_mean": 0.004104029692825861,
      "epoch": 5.293963254593176,
      "grad_norm": 0.12089893966913223,
      "learning_rate": 1e-06,
      "loss": -0.0325,
      "step": 1023
    },
    {
      "clip_ratio/high_max": 0.00402004765055608,
      "clip_ratio/high_mean": 0.001815882667870028,
      "clip_ratio/low_mean": 0.0017658336619206239,
      "clip_ratio/low_min": 0.00011170033394591883,
      "clip_ratio/region_mean": 0.003581716286134906,
      "epoch": 5.298629337999417,
      "grad_norm": 0.1064288318157196,
      "learning_rate": 1e-06,
      "loss": -0.038,
      "step": 1024
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0703822544642857,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4088.0,
      "completions/mean_length": 796.1515502929688,
      "completions/mean_terminated_length": 546.3167724609375,
      "completions/min_length": 102.0,
      "completions/min_terminated_length": 102.0,
      "epoch": 5.303295421405657,
      "grad_norm": 0.16900043189525604,
      "learning_rate": 1e-06,
      "loss": -0.0407,
      "num_tokens": 302235488.0,
      "reward": 0.64453125,
      "reward_std": 0.15173809230327606,
      "rewards/simpleverify_reward/mean": 0.64453125,
      "rewards/simpleverify_reward/std": 0.4786717891693115,
      "step": 1025
    },
    {
      "clip_ratio/high_max": 0.0019901058039977215,
      "clip_ratio/high_mean": 0.0007580732817586977,
      "clip_ratio/low_mean": 0.0004407145856930583,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0011987878679065034,
      "epoch": 5.307961504811899,
      "grad_norm": 0.30315497517585754,
      "learning_rate": 1e-06,
      "loss": -0.0187,
      "step": 1026
    },
    {
      "clip_ratio/high_max": 0.0017779778499971144,
      "clip_ratio/high_mean": 0.0007188380077423062,
      "clip_ratio/low_mean": 0.00045006515028944705,
      "clip_ratio/low_min": 1.700217580946628e-05,
      "clip_ratio/region_mean": 0.0011689031598507427,
      "epoch": 5.312627588218139,
      "grad_norm": 0.13995777070522308,
      "learning_rate": 1e-06,
      "loss": -0.0235,
      "step": 1027
    },
    {
      "clip_ratio/high_max": 0.0018832266214303672,
      "clip_ratio/high_mean": 0.0007336670914810384,
      "clip_ratio/low_mean": 0.0004941793376929127,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012278464309929404,
      "epoch": 5.31729367162438,
      "grad_norm": 0.14066806435585022,
      "learning_rate": 1e-06,
      "loss": -0.0408,
      "step": 1028
    },
    {
      "clip_ratio/high_max": 0.0020801038554054685,
      "clip_ratio/high_mean": 0.0008948110171331791,
      "clip_ratio/low_mean": 0.0005780501096523949,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014728611458849628,
      "epoch": 5.321959755030621,
      "grad_norm": 0.14191918075084686,
      "learning_rate": 1e-06,
      "loss": -0.017,
      "step": 1029
    },
    {
      "clip_ratio/high_max": 0.0019493888903525658,
      "clip_ratio/high_mean": 0.0008005459312698804,
      "clip_ratio/low_mean": 0.0005298624591887346,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013304083731782157,
      "epoch": 5.326625838436862,
      "grad_norm": 0.1472458392381668,
      "learning_rate": 1e-06,
      "loss": -0.0328,
      "step": 1030
    },
    {
      "clip_ratio/high_max": 0.0022974872917984612,
      "clip_ratio/high_mean": 0.0008544014235667419,
      "clip_ratio/low_mean": 0.0005862041334694368,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014406055270228535,
      "epoch": 5.331291921843103,
      "grad_norm": 0.14370617270469666,
      "learning_rate": 1e-06,
      "loss": -0.0494,
      "step": 1031
    },
    {
      "clip_ratio/high_max": 0.00222120506077772,
      "clip_ratio/high_mean": 0.0008841853414196521,
      "clip_ratio/low_mean": 0.0007317423262520606,
      "clip_ratio/low_min": 5.127830445417203e-05,
      "clip_ratio/region_mean": 0.001615927649254445,
      "epoch": 5.335958005249344,
      "grad_norm": 0.164176344871521,
      "learning_rate": 1e-06,
      "loss": 0.0062,
      "step": 1032
    },
    {
      "clip_ratio/high_max": 0.001983374317205744,
      "clip_ratio/high_mean": 0.0008687130357429851,
      "clip_ratio/low_mean": 0.0006822798013672582,
      "clip_ratio/low_min": 2.1208008547546342e-05,
      "clip_ratio/region_mean": 0.001550992819829844,
      "epoch": 5.340624088655585,
      "grad_norm": 0.15979741513729095,
      "learning_rate": 1e-06,
      "loss": -0.0045,
      "step": 1033
    },
    {
      "clip_ratio/high_max": 0.002303304754605051,
      "clip_ratio/high_mean": 0.0009059564308699919,
      "clip_ratio/low_mean": 0.0005827184099871374,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014886748540448025,
      "epoch": 5.3452901720618256,
      "grad_norm": 0.17692916095256805,
      "learning_rate": 1e-06,
      "loss": -0.0234,
      "step": 1034
    },
    {
      "clip_ratio/high_max": 0.0019705890881596133,
      "clip_ratio/high_mean": 0.0007961713872646214,
      "clip_ratio/low_mean": 0.0007159205924835987,
      "clip_ratio/low_min": 4.466944665182382e-05,
      "clip_ratio/region_mean": 0.0015120919997571036,
      "epoch": 5.349956255468067,
      "grad_norm": 0.12925322353839874,
      "learning_rate": 1e-06,
      "loss": -0.0375,
      "step": 1035
    },
    {
      "clip_ratio/high_max": 0.002061121122096665,
      "clip_ratio/high_mean": 0.000853650688441121,
      "clip_ratio/low_mean": 0.0007101527780832839,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015638034747098573,
      "epoch": 5.354622338874307,
      "grad_norm": 0.1644068956375122,
      "learning_rate": 1e-06,
      "loss": -0.0292,
      "step": 1036
    },
    {
      "clip_ratio/high_max": 0.002102839069266338,
      "clip_ratio/high_mean": 0.0009394999797223136,
      "clip_ratio/low_mean": 0.0007577474707431975,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016972474550129846,
      "epoch": 5.359288422280549,
      "grad_norm": 0.18274541199207306,
      "learning_rate": 1e-06,
      "loss": -0.0273,
      "step": 1037
    },
    {
      "clip_ratio/high_max": 0.0029189602355472744,
      "clip_ratio/high_mean": 0.0010747261840151623,
      "clip_ratio/low_mean": 0.0007616193743160693,
      "clip_ratio/low_min": 4.585473288898356e-05,
      "clip_ratio/region_mean": 0.0018363455383223481,
      "epoch": 5.363954505686789,
      "grad_norm": 0.13675089180469513,
      "learning_rate": 1e-06,
      "loss": -0.0172,
      "step": 1038
    },
    {
      "clip_ratio/high_max": 0.002146710750821512,
      "clip_ratio/high_mean": 0.0009420935148227727,
      "clip_ratio/low_mean": 0.0008759860538702924,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001818079577788012,
      "epoch": 5.36862058909303,
      "grad_norm": 0.1734006553888321,
      "learning_rate": 1e-06,
      "loss": -0.0138,
      "step": 1039
    },
    {
      "clip_ratio/high_max": 0.0025304565497208387,
      "clip_ratio/high_mean": 0.0008993535375338979,
      "clip_ratio/low_mean": 0.0008891449360817205,
      "clip_ratio/low_min": 1.2685203728324268e-05,
      "clip_ratio/region_mean": 0.0017884984699776396,
      "epoch": 5.373286672499271,
      "grad_norm": 0.15388691425323486,
      "learning_rate": 1e-06,
      "loss": -0.0048,
      "step": 1040
    },
    {
      "clip_ratio/high_max": 0.004619701052433811,
      "clip_ratio/high_mean": 0.0019604874978540465,
      "clip_ratio/low_mean": 0.0015945161358104087,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003555003640940413,
      "epoch": 5.377952755905512,
      "grad_norm": 0.09656365215778351,
      "learning_rate": 1e-06,
      "loss": -0.0415,
      "step": 1041
    },
    {
      "clip_ratio/high_max": 0.004308526469685603,
      "clip_ratio/high_mean": 0.0017046349639713299,
      "clip_ratio/low_mean": 0.0015700769690738525,
      "clip_ratio/low_min": 0.00014946940791560337,
      "clip_ratio/region_mean": 0.0032747119257692248,
      "epoch": 5.3826188393117524,
      "grad_norm": 0.10604587197303772,
      "learning_rate": 1e-06,
      "loss": -0.0196,
      "step": 1042
    },
    {
      "clip_ratio/high_max": 0.004013026918983087,
      "clip_ratio/high_mean": 0.001673447659413796,
      "clip_ratio/low_mean": 0.0017098265161621384,
      "clip_ratio/low_min": 8.573388186050579e-05,
      "clip_ratio/region_mean": 0.0033832742337835953,
      "epoch": 5.387284922717994,
      "grad_norm": 0.10498189926147461,
      "learning_rate": 1e-06,
      "loss": -0.0243,
      "step": 1043
    },
    {
      "clip_ratio/high_max": 0.004044265740958508,
      "clip_ratio/high_mean": 0.0016807025749585591,
      "clip_ratio/low_mean": 0.0016275965026579797,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003308299033960793,
      "epoch": 5.391951006124234,
      "grad_norm": 0.0970829576253891,
      "learning_rate": 1e-06,
      "loss": -0.0416,
      "step": 1044
    },
    {
      "clip_ratio/high_max": 0.004686776621383615,
      "clip_ratio/high_mean": 0.0018879604613175616,
      "clip_ratio/low_mean": 0.0016483919480378972,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035363523493288085,
      "epoch": 5.3966170895304755,
      "grad_norm": 0.10536900907754898,
      "learning_rate": 1e-06,
      "loss": -0.0177,
      "step": 1045
    },
    {
      "clip_ratio/high_max": 0.004191796950181015,
      "clip_ratio/high_mean": 0.0017523608075862285,
      "clip_ratio/low_mean": 0.0016774851937952917,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003429846008657478,
      "epoch": 5.401283172936716,
      "grad_norm": 0.09806672483682632,
      "learning_rate": 1e-06,
      "loss": -0.0336,
      "step": 1046
    },
    {
      "clip_ratio/high_max": 0.005034724992583506,
      "clip_ratio/high_mean": 0.0018981661414727569,
      "clip_ratio/low_mean": 0.0014747300283488585,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00337289618619252,
      "epoch": 5.405949256342957,
      "grad_norm": 0.1024920791387558,
      "learning_rate": 1e-06,
      "loss": -0.0502,
      "step": 1047
    },
    {
      "clip_ratio/high_max": 0.004791682600625791,
      "clip_ratio/high_mean": 0.0018399969194433652,
      "clip_ratio/low_mean": 0.0019361706745257834,
      "clip_ratio/low_min": 0.00012376033919281326,
      "clip_ratio/region_mean": 0.00377616765035782,
      "epoch": 5.410615339749198,
      "grad_norm": 0.115408755838871,
      "learning_rate": 1e-06,
      "loss": 0.0052,
      "step": 1048
    },
    {
      "clip_ratio/high_max": 0.004598794512276072,
      "clip_ratio/high_mean": 0.0017598953018023167,
      "clip_ratio/low_mean": 0.0018190212904301006,
      "clip_ratio/low_min": 2.1208008547546342e-05,
      "clip_ratio/region_mean": 0.003578916519472841,
      "epoch": 5.415281423155439,
      "grad_norm": 0.10779279470443726,
      "learning_rate": 1e-06,
      "loss": -0.0054,
      "step": 1049
    },
    {
      "clip_ratio/high_max": 0.004387183973449282,
      "clip_ratio/high_mean": 0.0018157201557187364,
      "clip_ratio/low_mean": 0.0016432555639767088,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003458975683315657,
      "epoch": 5.41994750656168,
      "grad_norm": 0.11029481142759323,
      "learning_rate": 1e-06,
      "loss": -0.0245,
      "step": 1050
    },
    {
      "clip_ratio/high_max": 0.004310808617447037,
      "clip_ratio/high_mean": 0.0016411887845606543,
      "clip_ratio/low_mean": 0.0015889382902969373,
      "clip_ratio/low_min": 6.700416997773573e-05,
      "clip_ratio/region_mean": 0.00323012717126403,
      "epoch": 5.424613589967921,
      "grad_norm": 0.1090804859995842,
      "learning_rate": 1e-06,
      "loss": -0.0382,
      "step": 1051
    },
    {
      "clip_ratio/high_max": 0.004544772586086765,
      "clip_ratio/high_mean": 0.001863996196334483,
      "clip_ratio/low_mean": 0.001818540484237019,
      "clip_ratio/low_min": 5.2065977797610685e-05,
      "clip_ratio/region_mean": 0.003682536698761396,
      "epoch": 5.429279673374162,
      "grad_norm": 0.10998934507369995,
      "learning_rate": 1e-06,
      "loss": -0.0302,
      "step": 1052
    },
    {
      "clip_ratio/high_max": 0.004446835315320641,
      "clip_ratio/high_mean": 0.0018928991667053197,
      "clip_ratio/low_mean": 0.0019031375068152556,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037960366753395647,
      "epoch": 5.433945756780402,
      "grad_norm": 0.11447273939847946,
      "learning_rate": 1e-06,
      "loss": -0.0283,
      "step": 1053
    },
    {
      "clip_ratio/high_max": 0.004872256686212495,
      "clip_ratio/high_mean": 0.0018683382477320265,
      "clip_ratio/low_mean": 0.0017378241464029998,
      "clip_ratio/low_min": 8.043758134590462e-05,
      "clip_ratio/region_mean": 0.0036061624414287508,
      "epoch": 5.438611840186644,
      "grad_norm": 0.10243591666221619,
      "learning_rate": 1e-06,
      "loss": -0.018,
      "step": 1054
    },
    {
      "clip_ratio/high_max": 0.0048987017798935995,
      "clip_ratio/high_mean": 0.0018398495521978475,
      "clip_ratio/low_mean": 0.0019785495933319908,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003818399185547605,
      "epoch": 5.443277923592884,
      "grad_norm": 0.11500147730112076,
      "learning_rate": 1e-06,
      "loss": -0.0148,
      "step": 1055
    },
    {
      "clip_ratio/high_max": 0.004209175094729289,
      "clip_ratio/high_mean": 0.0016091649413283449,
      "clip_ratio/low_mean": 0.001991480603464879,
      "clip_ratio/low_min": 3.8055612094467506e-05,
      "clip_ratio/region_mean": 0.0036006455484312028,
      "epoch": 5.447944006999125,
      "grad_norm": 0.09771430492401123,
      "learning_rate": 1e-06,
      "loss": -0.0055,
      "step": 1056
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08203125,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4028.0,
      "completions/mean_length": 844.0969848632812,
      "completions/mean_terminated_length": 553.5013427734375,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 5.452610090405366,
      "grad_norm": 0.18503591418266296,
      "learning_rate": 1e-06,
      "loss": -0.0423,
      "num_tokens": 310969310.0,
      "reward": 0.6193498969078064,
      "reward_std": 0.16954688727855682,
      "rewards/simpleverify_reward/mean": 0.6193498969078064,
      "rewards/simpleverify_reward/std": 0.48556363582611084,
      "step": 1057
    },
    {
      "clip_ratio/high_max": 0.0021010026830481365,
      "clip_ratio/high_mean": 0.0008167638879967853,
      "clip_ratio/low_mean": 0.0005597656472673407,
      "clip_ratio/low_min": 3.10698133034748e-05,
      "clip_ratio/region_mean": 0.001376529544359073,
      "epoch": 5.457276173811607,
      "grad_norm": 0.15392066538333893,
      "learning_rate": 1e-06,
      "loss": -0.0204,
      "step": 1058
    },
    {
      "clip_ratio/high_max": 0.002218676880147541,
      "clip_ratio/high_mean": 0.0008808757229417097,
      "clip_ratio/low_mean": 0.00044940638440493785,
      "clip_ratio/low_min": 1.230557154485723e-05,
      "clip_ratio/region_mean": 0.0013302821207616944,
      "epoch": 5.4619422572178475,
      "grad_norm": 0.14625273644924164,
      "learning_rate": 1e-06,
      "loss": -0.0445,
      "step": 1059
    },
    {
      "clip_ratio/high_max": 0.0019228427627240308,
      "clip_ratio/high_mean": 0.0008116791414067848,
      "clip_ratio/low_mean": 0.0004936997529512155,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013053789080004208,
      "epoch": 5.466608340624089,
      "grad_norm": 0.18353533744812012,
      "learning_rate": 1e-06,
      "loss": -0.0338,
      "step": 1060
    },
    {
      "clip_ratio/high_max": 0.00246318471909035,
      "clip_ratio/high_mean": 0.0009007052176457364,
      "clip_ratio/low_mean": 0.0005558047769227414,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001456509988202015,
      "epoch": 5.471274424030329,
      "grad_norm": 0.15294893085956573,
      "learning_rate": 1e-06,
      "loss": -0.038,
      "step": 1061
    },
    {
      "clip_ratio/high_max": 0.002179118077037856,
      "clip_ratio/high_mean": 0.0009593284976290306,
      "clip_ratio/low_mean": 0.0006045227364666061,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001563851212267764,
      "epoch": 5.475940507436571,
      "grad_norm": 0.16398212313652039,
      "learning_rate": 1e-06,
      "loss": -0.0505,
      "step": 1062
    },
    {
      "clip_ratio/high_max": 0.0023751916523906402,
      "clip_ratio/high_mean": 0.0009796846388780978,
      "clip_ratio/low_mean": 0.0007294433135029976,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017091279478336219,
      "epoch": 5.480606590842811,
      "grad_norm": 0.17016813158988953,
      "learning_rate": 1e-06,
      "loss": -0.019,
      "step": 1063
    },
    {
      "clip_ratio/high_max": 0.002340564598853234,
      "clip_ratio/high_mean": 0.0009789278356038267,
      "clip_ratio/low_mean": 0.0008047365790844196,
      "clip_ratio/low_min": 2.706799568841234e-05,
      "clip_ratio/region_mean": 0.0017836644183262251,
      "epoch": 5.485272674249052,
      "grad_norm": 0.1788434386253357,
      "learning_rate": 1e-06,
      "loss": -0.0424,
      "step": 1064
    },
    {
      "clip_ratio/high_max": 0.0021957421995466575,
      "clip_ratio/high_mean": 0.0009893622482195497,
      "clip_ratio/low_mean": 0.0007605226783198304,
      "clip_ratio/low_min": 1.6037978639360517e-05,
      "clip_ratio/region_mean": 0.0017498849192634225,
      "epoch": 5.489938757655293,
      "grad_norm": 0.1769745945930481,
      "learning_rate": 1e-06,
      "loss": -0.0257,
      "step": 1065
    },
    {
      "clip_ratio/high_max": 0.002496462398994481,
      "clip_ratio/high_mean": 0.0009307483560405672,
      "clip_ratio/low_mean": 0.0006463490763053414,
      "clip_ratio/low_min": 2.1570318494923413e-05,
      "clip_ratio/region_mean": 0.0015770974387123715,
      "epoch": 5.494604841061534,
      "grad_norm": 0.14289164543151855,
      "learning_rate": 1e-06,
      "loss": -0.0413,
      "step": 1066
    },
    {
      "clip_ratio/high_max": 0.0019630641108960845,
      "clip_ratio/high_mean": 0.0008781963006185833,
      "clip_ratio/low_mean": 0.0009712367609608918,
      "clip_ratio/low_min": 3.799853766395245e-05,
      "clip_ratio/region_mean": 0.0018494330797693692,
      "epoch": 5.499270924467774,
      "grad_norm": 0.15272675454616547,
      "learning_rate": 1e-06,
      "loss": 0.022,
      "step": 1067
    },
    {
      "clip_ratio/high_max": 0.002803758798108902,
      "clip_ratio/high_mean": 0.0011310019144730177,
      "clip_ratio/low_mean": 0.0007846928119761287,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019156947018927895,
      "epoch": 5.503937007874016,
      "grad_norm": 0.14283253252506256,
      "learning_rate": 1e-06,
      "loss": -0.0268,
      "step": 1068
    },
    {
      "clip_ratio/high_max": 0.002550644741859287,
      "clip_ratio/high_mean": 0.0011323581566102803,
      "clip_ratio/low_mean": 0.000811888476164313,
      "clip_ratio/low_min": 2.7162102924194187e-05,
      "clip_ratio/region_mean": 0.0019442466073087417,
      "epoch": 5.508603091280257,
      "grad_norm": 0.18627505004405975,
      "learning_rate": 1e-06,
      "loss": -0.0231,
      "step": 1069
    },
    {
      "clip_ratio/high_max": 0.0024651432540849783,
      "clip_ratio/high_mean": 0.0010573405561444815,
      "clip_ratio/low_mean": 0.0007259911490109516,
      "clip_ratio/low_min": 1.2636473911697976e-05,
      "clip_ratio/region_mean": 0.0017833317106124014,
      "epoch": 5.5132691746864975,
      "grad_norm": 0.15494415163993835,
      "learning_rate": 1e-06,
      "loss": -0.0277,
      "step": 1070
    },
    {
      "clip_ratio/high_max": 0.0023478621369577013,
      "clip_ratio/high_mean": 0.0010715455155150266,
      "clip_ratio/low_mean": 0.0006710978486808017,
      "clip_ratio/low_min": 3.34582437062636e-05,
      "clip_ratio/region_mean": 0.0017426433369109873,
      "epoch": 5.517935258092739,
      "grad_norm": 0.14109767973423004,
      "learning_rate": 1e-06,
      "loss": -0.037,
      "step": 1071
    },
    {
      "clip_ratio/high_max": 0.0022095215899753384,
      "clip_ratio/high_mean": 0.0010863233892450808,
      "clip_ratio/low_mean": 0.0008011010177142452,
      "clip_ratio/low_min": 3.37837846018374e-05,
      "clip_ratio/region_mean": 0.0018874244196922518,
      "epoch": 5.522601341498979,
      "grad_norm": 0.17250967025756836,
      "learning_rate": 1e-06,
      "loss": -0.0699,
      "step": 1072
    },
    {
      "clip_ratio/high_max": 0.005617723465547897,
      "clip_ratio/high_mean": 0.002320662810234353,
      "clip_ratio/low_mean": 0.0019079897974734195,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.004228652687743306,
      "epoch": 5.5272674249052205,
      "grad_norm": 0.11128078401088715,
      "learning_rate": 1e-06,
      "loss": -0.0433,
      "step": 1073
    },
    {
      "clip_ratio/high_max": 0.00512938011524966,
      "clip_ratio/high_mean": 0.001960689442057628,
      "clip_ratio/low_mean": 0.0019387240965897945,
      "clip_ratio/low_min": 0.00017297096201218665,
      "clip_ratio/region_mean": 0.0038994134665699676,
      "epoch": 5.531933508311461,
      "grad_norm": 0.11491167545318604,
      "learning_rate": 1e-06,
      "loss": -0.0213,
      "step": 1074
    },
    {
      "clip_ratio/high_max": 0.0048753045266494155,
      "clip_ratio/high_mean": 0.0019276366438134573,
      "clip_ratio/low_mean": 0.0016700985834177118,
      "clip_ratio/low_min": 0.0001384274655720219,
      "clip_ratio/region_mean": 0.00359773519448936,
      "epoch": 5.536599591717702,
      "grad_norm": 0.12996047735214233,
      "learning_rate": 1e-06,
      "loss": -0.0454,
      "step": 1075
    },
    {
      "clip_ratio/high_max": 0.004236992390360683,
      "clip_ratio/high_mean": 0.0019387972351978533,
      "clip_ratio/low_mean": 0.001695863469649339,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036346606648294255,
      "epoch": 5.541265675123943,
      "grad_norm": 0.10617971420288086,
      "learning_rate": 1e-06,
      "loss": -0.0346,
      "step": 1076
    },
    {
      "clip_ratio/high_max": 0.005691662343451753,
      "clip_ratio/high_mean": 0.0020724655187223107,
      "clip_ratio/low_mean": 0.0018812367343343794,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003953702165745199,
      "epoch": 5.545931758530184,
      "grad_norm": 0.1152462512254715,
      "learning_rate": 1e-06,
      "loss": -0.0389,
      "step": 1077
    },
    {
      "clip_ratio/high_max": 0.004724372047348879,
      "clip_ratio/high_mean": 0.002063470528810285,
      "clip_ratio/low_mean": 0.001929070131154731,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00399254058720544,
      "epoch": 5.550597841936424,
      "grad_norm": 0.11365631967782974,
      "learning_rate": 1e-06,
      "loss": -0.0516,
      "step": 1078
    },
    {
      "clip_ratio/high_max": 0.005498058104421943,
      "clip_ratio/high_mean": 0.002033653436228633,
      "clip_ratio/low_mean": 0.002255057930597104,
      "clip_ratio/low_min": 5.693463754141703e-05,
      "clip_ratio/region_mean": 0.004288711308618076,
      "epoch": 5.555263925342666,
      "grad_norm": 0.11926291882991791,
      "learning_rate": 1e-06,
      "loss": -0.02,
      "step": 1079
    },
    {
      "clip_ratio/high_max": 0.0046259300579549745,
      "clip_ratio/high_mean": 0.0021313317483873107,
      "clip_ratio/low_mean": 0.001989418040466262,
      "clip_ratio/low_min": 7.97193861217238e-05,
      "clip_ratio/region_mean": 0.004120749741559848,
      "epoch": 5.559930008748906,
      "grad_norm": 0.1176396906375885,
      "learning_rate": 1e-06,
      "loss": -0.0434,
      "step": 1080
    },
    {
      "clip_ratio/high_max": 0.005511932453373447,
      "clip_ratio/high_mean": 0.002204282194725238,
      "clip_ratio/low_mean": 0.0021201263734837994,
      "clip_ratio/low_min": 0.00016949895234574797,
      "clip_ratio/region_mean": 0.004324408466345631,
      "epoch": 5.564596092155147,
      "grad_norm": 0.13098770380020142,
      "learning_rate": 1e-06,
      "loss": -0.0269,
      "step": 1081
    },
    {
      "clip_ratio/high_max": 0.004959311118000187,
      "clip_ratio/high_mean": 0.001934655385412043,
      "clip_ratio/low_mean": 0.0018828015745384619,
      "clip_ratio/low_min": 4.3140636989846826e-05,
      "clip_ratio/region_mean": 0.003817457014520187,
      "epoch": 5.569262175561388,
      "grad_norm": 0.11082416027784348,
      "learning_rate": 1e-06,
      "loss": -0.0421,
      "step": 1082
    },
    {
      "clip_ratio/high_max": 0.0037324036748032086,
      "clip_ratio/high_mean": 0.0017099572214647196,
      "clip_ratio/low_mean": 0.002334678851184435,
      "clip_ratio/low_min": 5.456360486277845e-05,
      "clip_ratio/region_mean": 0.004044636138132773,
      "epoch": 5.573928258967629,
      "grad_norm": 0.1414852887392044,
      "learning_rate": 1e-06,
      "loss": 0.0211,
      "step": 1083
    },
    {
      "clip_ratio/high_max": 0.005095094718853943,
      "clip_ratio/high_mean": 0.0019395994531805627,
      "clip_ratio/low_mean": 0.0018185253939009272,
      "clip_ratio/low_min": 3.838280463241972e-05,
      "clip_ratio/region_mean": 0.0037581248761853203,
      "epoch": 5.57859434237387,
      "grad_norm": 0.1120387390255928,
      "learning_rate": 1e-06,
      "loss": -0.0276,
      "step": 1084
    },
    {
      "clip_ratio/high_max": 0.004994406510377303,
      "clip_ratio/high_mean": 0.002108082568156533,
      "clip_ratio/low_mean": 0.0021776220492029097,
      "clip_ratio/low_min": 4.074315438629128e-05,
      "clip_ratio/region_mean": 0.00428570453368593,
      "epoch": 5.583260425780111,
      "grad_norm": 0.18482013046741486,
      "learning_rate": 1e-06,
      "loss": -0.0243,
      "step": 1085
    },
    {
      "clip_ratio/high_max": 0.004674766169046052,
      "clip_ratio/high_mean": 0.0019227692828280851,
      "clip_ratio/low_mean": 0.001774288590240758,
      "clip_ratio/low_min": 4.752851600642316e-05,
      "clip_ratio/region_mean": 0.0036970578075852245,
      "epoch": 5.587926509186351,
      "grad_norm": 0.1114470362663269,
      "learning_rate": 1e-06,
      "loss": -0.0287,
      "step": 1086
    },
    {
      "clip_ratio/high_max": 0.0048480612895218655,
      "clip_ratio/high_mean": 0.0020140057094977237,
      "clip_ratio/low_mean": 0.001531969082861906,
      "clip_ratio/low_min": 3.806333916145377e-05,
      "clip_ratio/region_mean": 0.003545974803273566,
      "epoch": 5.592592592592593,
      "grad_norm": 0.10625629127025604,
      "learning_rate": 1e-06,
      "loss": -0.0378,
      "step": 1087
    },
    {
      "clip_ratio/high_max": 0.0047394736466230825,
      "clip_ratio/high_mean": 0.002096644733683206,
      "clip_ratio/low_mean": 0.001747896931192372,
      "clip_ratio/low_min": 0.00010532051237532869,
      "clip_ratio/region_mean": 0.0038445415993919596,
      "epoch": 5.597258675998834,
      "grad_norm": 0.09825687855482101,
      "learning_rate": 1e-06,
      "loss": -0.0707,
      "step": 1088
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0721261160714286,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4073.0,
      "completions/mean_length": 801.2921752929688,
      "completions/mean_terminated_length": 545.1856689453125,
      "completions/min_length": 89.0,
      "completions/min_terminated_length": 89.0,
      "epoch": 5.601924759405074,
      "grad_norm": 0.17363311350345612,
      "learning_rate": 1e-06,
      "loss": -0.0254,
      "num_tokens": 319663274.0,
      "reward": 0.66015625,
      "reward_std": 0.16099147498607635,
      "rewards/simpleverify_reward/mean": 0.66015625,
      "rewards/simpleverify_reward/std": 0.473672479391098,
      "step": 1089
    },
    {
      "clip_ratio/high_max": 0.00196332907944452,
      "clip_ratio/high_mean": 0.0007311119425139623,
      "clip_ratio/low_mean": 0.0004850490440730937,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012161609702161513,
      "epoch": 5.606590842811316,
      "grad_norm": 0.16732467710971832,
      "learning_rate": 1e-06,
      "loss": -0.0208,
      "step": 1090
    },
    {
      "clip_ratio/high_max": 0.001616110388567904,
      "clip_ratio/high_mean": 0.0006934280008863425,
      "clip_ratio/low_mean": 0.0005145723725945572,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001208000390761299,
      "epoch": 5.611256926217556,
      "grad_norm": 0.1546524167060852,
      "learning_rate": 1e-06,
      "loss": -0.0027,
      "step": 1091
    },
    {
      "clip_ratio/high_max": 0.0017909990892803762,
      "clip_ratio/high_mean": 0.0007520955605286872,
      "clip_ratio/low_mean": 0.0004986000549251912,
      "clip_ratio/low_min": 2.0627063349820673e-05,
      "clip_ratio/region_mean": 0.0012506956118158996,
      "epoch": 5.615923009623797,
      "grad_norm": 0.15939919650554657,
      "learning_rate": 1e-06,
      "loss": -0.036,
      "step": 1092
    },
    {
      "clip_ratio/high_max": 0.002326019777683541,
      "clip_ratio/high_mean": 0.0008496173995808931,
      "clip_ratio/low_mean": 0.0005462897142933798,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013959071038698312,
      "epoch": 5.620589093030038,
      "grad_norm": 0.34524571895599365,
      "learning_rate": 1e-06,
      "loss": -0.0367,
      "step": 1093
    },
    {
      "clip_ratio/high_max": 0.0025907351737259887,
      "clip_ratio/high_mean": 0.0010017334243457299,
      "clip_ratio/low_mean": 0.0006092508074289071,
      "clip_ratio/low_min": 3.235642634535907e-05,
      "clip_ratio/region_mean": 0.001610984189028386,
      "epoch": 5.625255176436279,
      "grad_norm": 0.2278021275997162,
      "learning_rate": 1e-06,
      "loss": -0.0663,
      "step": 1094
    },
    {
      "clip_ratio/high_max": 0.0023175524911493994,
      "clip_ratio/high_mean": 0.0008981495684565743,
      "clip_ratio/low_mean": 0.0007825216453056782,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016806712010293268,
      "epoch": 5.6299212598425195,
      "grad_norm": 0.16181299090385437,
      "learning_rate": 1e-06,
      "loss": -0.0676,
      "step": 1095
    },
    {
      "clip_ratio/high_max": 0.002330497867660597,
      "clip_ratio/high_mean": 0.0007984306157595711,
      "clip_ratio/low_mean": 0.000825373417683295,
      "clip_ratio/low_min": 4.587155854096636e-05,
      "clip_ratio/region_mean": 0.001623804015252972,
      "epoch": 5.634587343248761,
      "grad_norm": 0.15510323643684387,
      "learning_rate": 1e-06,
      "loss": 0.0053,
      "step": 1096
    },
    {
      "clip_ratio/high_max": 0.00197228736942634,
      "clip_ratio/high_mean": 0.0008431544520135503,
      "clip_ratio/low_mean": 0.0007782273205521051,
      "clip_ratio/low_min": 1.766534842317924e-05,
      "clip_ratio/region_mean": 0.0016213818089454435,
      "epoch": 5.639253426655001,
      "grad_norm": 0.1508847326040268,
      "learning_rate": 1e-06,
      "loss": -0.0176,
      "step": 1097
    },
    {
      "clip_ratio/high_max": 0.002444359219225589,
      "clip_ratio/high_mean": 0.0009905890728987288,
      "clip_ratio/low_mean": 0.0008047451269703743,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017953341521206312,
      "epoch": 5.6439195100612425,
      "grad_norm": 0.15566974878311157,
      "learning_rate": 1e-06,
      "loss": -0.0634,
      "step": 1098
    },
    {
      "clip_ratio/high_max": 0.0026409386919112876,
      "clip_ratio/high_mean": 0.0010656040212779772,
      "clip_ratio/low_mean": 0.0008403113897657022,
      "clip_ratio/low_min": 5.142429654370062e-05,
      "clip_ratio/region_mean": 0.0019059153928537853,
      "epoch": 5.648585593467483,
      "grad_norm": 0.14686118066310883,
      "learning_rate": 1e-06,
      "loss": -0.0326,
      "step": 1099
    },
    {
      "clip_ratio/high_max": 0.0022543475060956553,
      "clip_ratio/high_mean": 0.000969434300714056,
      "clip_ratio/low_mean": 0.0008012929883989273,
      "clip_ratio/low_min": 8.432835875282763e-05,
      "clip_ratio/region_mean": 0.0017707272854750045,
      "epoch": 5.653251676873724,
      "grad_norm": 0.1577766388654709,
      "learning_rate": 1e-06,
      "loss": -0.0164,
      "step": 1100
    },
    {
      "clip_ratio/high_max": 0.0026114291977137327,
      "clip_ratio/high_mean": 0.0011314384209981654,
      "clip_ratio/low_mean": 0.0007933738379506394,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019248122625867836,
      "epoch": 5.657917760279965,
      "grad_norm": 0.16805461049079895,
      "learning_rate": 1e-06,
      "loss": -0.048,
      "step": 1101
    },
    {
      "clip_ratio/high_max": 0.002807301214488689,
      "clip_ratio/high_mean": 0.0010178441880270839,
      "clip_ratio/low_mean": 0.0007994123307071277,
      "clip_ratio/low_min": 3.0259017876232974e-05,
      "clip_ratio/region_mean": 0.0018172565614804626,
      "epoch": 5.662583843686206,
      "grad_norm": 0.16753637790679932,
      "learning_rate": 1e-06,
      "loss": -0.0254,
      "step": 1102
    },
    {
      "clip_ratio/high_max": 0.0028521964559331536,
      "clip_ratio/high_mean": 0.0010553336069278885,
      "clip_ratio/low_mean": 0.0007586399824504042,
      "clip_ratio/low_min": 1.26237127915374e-05,
      "clip_ratio/region_mean": 0.001813973613025155,
      "epoch": 5.667249927092447,
      "grad_norm": 0.15705883502960205,
      "learning_rate": 1e-06,
      "loss": -0.0512,
      "step": 1103
    },
    {
      "clip_ratio/high_max": 0.0023743039564578794,
      "clip_ratio/high_mean": 0.0010277639485138934,
      "clip_ratio/low_mean": 0.0009079600440600188,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019357239725650288,
      "epoch": 5.671916010498688,
      "grad_norm": 0.3744487762451172,
      "learning_rate": 1e-06,
      "loss": -0.0063,
      "step": 1104
    },
    {
      "clip_ratio/high_max": 0.004711620123998728,
      "clip_ratio/high_mean": 0.0021453926419781055,
      "clip_ratio/low_mean": 0.0020286341932660434,
      "clip_ratio/low_min": 5.2936193242203444e-05,
      "clip_ratio/region_mean": 0.004174026864347979,
      "epoch": 5.676582093904928,
      "grad_norm": 0.10706911236047745,
      "learning_rate": 1e-06,
      "loss": -0.0264,
      "step": 1105
    },
    {
      "clip_ratio/high_max": 0.004823029492399655,
      "clip_ratio/high_mean": 0.0018834305519703776,
      "clip_ratio/low_mean": 0.0019089536344836233,
      "clip_ratio/low_min": 5.804504326079041e-05,
      "clip_ratio/region_mean": 0.0037923841300653294,
      "epoch": 5.681248177311169,
      "grad_norm": 0.1027434915304184,
      "learning_rate": 1e-06,
      "loss": -0.0219,
      "step": 1106
    },
    {
      "clip_ratio/high_max": 0.004103656210645568,
      "clip_ratio/high_mean": 0.0016622874500171747,
      "clip_ratio/low_mean": 0.001836797686337377,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034990852000191808,
      "epoch": 5.685914260717411,
      "grad_norm": 0.10081746429204941,
      "learning_rate": 1e-06,
      "loss": -0.0035,
      "step": 1107
    },
    {
      "clip_ratio/high_max": 0.004733275738544762,
      "clip_ratio/high_mean": 0.0018238598349853419,
      "clip_ratio/low_mean": 0.0015205755989882164,
      "clip_ratio/low_min": 9.480412791162962e-05,
      "clip_ratio/region_mean": 0.0033444353539380245,
      "epoch": 5.690580344123651,
      "grad_norm": 0.10527554154396057,
      "learning_rate": 1e-06,
      "loss": -0.0369,
      "step": 1108
    },
    {
      "clip_ratio/high_max": 0.004575478262268007,
      "clip_ratio/high_mean": 0.0018464363274688367,
      "clip_ratio/low_mean": 0.0017243847651116084,
      "clip_ratio/low_min": 4.298486965126358e-05,
      "clip_ratio/region_mean": 0.0035708210634766147,
      "epoch": 5.695246427529892,
      "grad_norm": 0.11438775062561035,
      "learning_rate": 1e-06,
      "loss": -0.0376,
      "step": 1109
    },
    {
      "clip_ratio/high_max": 0.005584066020674072,
      "clip_ratio/high_mean": 0.0021269731296342798,
      "clip_ratio/low_mean": 0.0017782876293495065,
      "clip_ratio/low_min": 0.0001413922946085222,
      "clip_ratio/region_mean": 0.0039052607608027756,
      "epoch": 5.699912510936133,
      "grad_norm": 0.12058404833078384,
      "learning_rate": 1e-06,
      "loss": -0.0673,
      "step": 1110
    },
    {
      "clip_ratio/high_max": 0.005755571139161475,
      "clip_ratio/high_mean": 0.002112916648911778,
      "clip_ratio/low_mean": 0.001908920079586096,
      "clip_ratio/low_min": 6.401502469088882e-05,
      "clip_ratio/region_mean": 0.004021836764877662,
      "epoch": 5.704578594342374,
      "grad_norm": 0.1170424222946167,
      "learning_rate": 1e-06,
      "loss": -0.0686,
      "step": 1111
    },
    {
      "clip_ratio/high_max": 0.004587032104609534,
      "clip_ratio/high_mean": 0.001741022126225289,
      "clip_ratio/low_mean": 0.0020003209792776033,
      "clip_ratio/low_min": 7.352940883720294e-05,
      "clip_ratio/region_mean": 0.003741343149158638,
      "epoch": 5.7092446777486145,
      "grad_norm": 0.11023940145969391,
      "learning_rate": 1e-06,
      "loss": 0.0045,
      "step": 1112
    },
    {
      "clip_ratio/high_max": 0.004333077071350999,
      "clip_ratio/high_mean": 0.0018212781506008469,
      "clip_ratio/low_mean": 0.0019266284434706904,
      "clip_ratio/low_min": 6.54450268484652e-05,
      "clip_ratio/region_mean": 0.0037479064922081307,
      "epoch": 5.713910761154856,
      "grad_norm": 0.1100669801235199,
      "learning_rate": 1e-06,
      "loss": -0.0185,
      "step": 1113
    },
    {
      "clip_ratio/high_max": 0.005343921991880052,
      "clip_ratio/high_mean": 0.0021120144956512377,
      "clip_ratio/low_mean": 0.0018746332407317823,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003986647701822221,
      "epoch": 5.718576844561096,
      "grad_norm": 0.12329089641571045,
      "learning_rate": 1e-06,
      "loss": -0.0643,
      "step": 1114
    },
    {
      "clip_ratio/high_max": 0.005325925361830741,
      "clip_ratio/high_mean": 0.0019957017175329383,
      "clip_ratio/low_mean": 0.0018841441451513674,
      "clip_ratio/low_min": 5.041570693720132e-05,
      "clip_ratio/region_mean": 0.0038798459281679243,
      "epoch": 5.723242927967338,
      "grad_norm": 0.1092992052435875,
      "learning_rate": 1e-06,
      "loss": -0.0333,
      "step": 1115
    },
    {
      "clip_ratio/high_max": 0.005351192216039635,
      "clip_ratio/high_mean": 0.0020601542782969773,
      "clip_ratio/low_mean": 0.0018666402938833926,
      "clip_ratio/low_min": 0.00019100204735877924,
      "clip_ratio/region_mean": 0.0039267945976462215,
      "epoch": 5.727909011373578,
      "grad_norm": 0.11630088090896606,
      "learning_rate": 1e-06,
      "loss": -0.0174,
      "step": 1116
    },
    {
      "clip_ratio/high_max": 0.0045871324255131185,
      "clip_ratio/high_mean": 0.0019645351421786472,
      "clip_ratio/low_mean": 0.0019200983188056853,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038846334064146504,
      "epoch": 5.732575094779819,
      "grad_norm": 0.11486152559518814,
      "learning_rate": 1e-06,
      "loss": -0.0489,
      "step": 1117
    },
    {
      "clip_ratio/high_max": 0.004793749481905252,
      "clip_ratio/high_mean": 0.0018304229961358942,
      "clip_ratio/low_mean": 0.001937795370395179,
      "clip_ratio/low_min": 4.538852590485476e-05,
      "clip_ratio/region_mean": 0.003768218281038571,
      "epoch": 5.73724117818606,
      "grad_norm": 0.10911096632480621,
      "learning_rate": 1e-06,
      "loss": -0.0263,
      "step": 1118
    },
    {
      "clip_ratio/high_max": 0.005130968100274913,
      "clip_ratio/high_mean": 0.0019012518459931016,
      "clip_ratio/low_mean": 0.001879583091067616,
      "clip_ratio/low_min": 8.83659886312671e-05,
      "clip_ratio/region_mean": 0.0037808349152328447,
      "epoch": 5.741907261592301,
      "grad_norm": 0.11238105595111847,
      "learning_rate": 1e-06,
      "loss": -0.0521,
      "step": 1119
    },
    {
      "clip_ratio/high_max": 0.005286748561047716,
      "clip_ratio/high_mean": 0.0019838979951600777,
      "clip_ratio/low_mean": 0.002030055969953537,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.004013953875983134,
      "epoch": 5.746573344998541,
      "grad_norm": 0.11621847748756409,
      "learning_rate": 1e-06,
      "loss": -0.0073,
      "step": 1120
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0686383928571429,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4058.0,
      "completions/mean_length": 790.1351318359375,
      "completions/mean_terminated_length": 546.50341796875,
      "completions/min_length": 52.0,
      "completions/min_terminated_length": 52.0,
      "epoch": 5.751239428404783,
      "grad_norm": 0.18430668115615845,
      "learning_rate": 1e-06,
      "loss": -0.0391,
      "num_tokens": 328396203.0,
      "reward": 0.6566685438156128,
      "reward_std": 0.15478026866912842,
      "rewards/simpleverify_reward/mean": 0.6566685438156128,
      "rewards/simpleverify_reward/std": 0.4748375415802002,
      "step": 1121
    },
    {
      "clip_ratio/high_max": 0.0018891220570367295,
      "clip_ratio/high_mean": 0.000707031127603841,
      "clip_ratio/low_mean": 0.00039948274570633657,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00110651386603422,
      "epoch": 5.755905511811024,
      "grad_norm": 0.15498477220535278,
      "learning_rate": 1e-06,
      "loss": -0.0244,
      "step": 1122
    },
    {
      "clip_ratio/high_max": 0.002032228709140327,
      "clip_ratio/high_mean": 0.0008019923952815589,
      "clip_ratio/low_mean": 0.000571771780414565,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013737642038904596,
      "epoch": 5.7605715952172645,
      "grad_norm": 0.1581478714942932,
      "learning_rate": 1e-06,
      "loss": -0.0245,
      "step": 1123
    },
    {
      "clip_ratio/high_max": 0.002226620723376982,
      "clip_ratio/high_mean": 0.0009234493281837786,
      "clip_ratio/low_mean": 0.0006573238442797447,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015807732015673537,
      "epoch": 5.765237678623506,
      "grad_norm": 0.17381872236728668,
      "learning_rate": 1e-06,
      "loss": -0.0238,
      "step": 1124
    },
    {
      "clip_ratio/high_max": 0.0021960548911010846,
      "clip_ratio/high_mean": 0.0008710607471584808,
      "clip_ratio/low_mean": 0.0004864364336754079,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013574971635534894,
      "epoch": 5.769903762029746,
      "grad_norm": 0.1541343480348587,
      "learning_rate": 1e-06,
      "loss": -0.0236,
      "step": 1125
    },
    {
      "clip_ratio/high_max": 0.002459991505020298,
      "clip_ratio/high_mean": 0.0010374924004281638,
      "clip_ratio/low_mean": 0.000506134047100204,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015436263965966646,
      "epoch": 5.7745698454359875,
      "grad_norm": 0.2002820521593094,
      "learning_rate": 1e-06,
      "loss": -0.0382,
      "step": 1126
    },
    {
      "clip_ratio/high_max": 0.0021695071482099593,
      "clip_ratio/high_mean": 0.0008159552180586616,
      "clip_ratio/low_mean": 0.0006894877496961271,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015054429532028735,
      "epoch": 5.779235928842228,
      "grad_norm": 0.1626857966184616,
      "learning_rate": 1e-06,
      "loss": 0.0064,
      "step": 1127
    },
    {
      "clip_ratio/high_max": 0.0027781318640336394,
      "clip_ratio/high_mean": 0.0009724751816975186,
      "clip_ratio/low_mean": 0.0006995455764808867,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016720208077458665,
      "epoch": 5.783902012248469,
      "grad_norm": 0.1575443595647812,
      "learning_rate": 1e-06,
      "loss": -0.0353,
      "step": 1128
    },
    {
      "clip_ratio/high_max": 0.0024203429420595057,
      "clip_ratio/high_mean": 0.0009116146902670152,
      "clip_ratio/low_mean": 0.0007138412656786386,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016254559122899082,
      "epoch": 5.78856809565471,
      "grad_norm": 0.1576351821422577,
      "learning_rate": 1e-06,
      "loss": -0.0088,
      "step": 1129
    },
    {
      "clip_ratio/high_max": 0.002319489431101829,
      "clip_ratio/high_mean": 0.001001957340122317,
      "clip_ratio/low_mean": 0.0005997584121359978,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016017158013710286,
      "epoch": 5.793234179060951,
      "grad_norm": 0.2136499285697937,
      "learning_rate": 1e-06,
      "loss": -0.009,
      "step": 1130
    },
    {
      "clip_ratio/high_max": 0.002192228399508167,
      "clip_ratio/high_mean": 0.0008696250970388064,
      "clip_ratio/low_mean": 0.0006664772299700417,
      "clip_ratio/low_min": 1.3631406545755453e-05,
      "clip_ratio/region_mean": 0.001536102336103795,
      "epoch": 5.797900262467191,
      "grad_norm": 0.21299482882022858,
      "learning_rate": 1e-06,
      "loss": -0.0376,
      "step": 1131
    },
    {
      "clip_ratio/high_max": 0.0023163586229202338,
      "clip_ratio/high_mean": 0.000869337822223315,
      "clip_ratio/low_mean": 0.0008491279440931976,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001718465719022788,
      "epoch": 5.802566345873433,
      "grad_norm": 0.14687684178352356,
      "learning_rate": 1e-06,
      "loss": -0.0274,
      "step": 1132
    },
    {
      "clip_ratio/high_max": 0.002249891826068051,
      "clip_ratio/high_mean": 0.00098193742451258,
      "clip_ratio/low_mean": 0.0007998976843737182,
      "clip_ratio/low_min": 4.605039794114418e-05,
      "clip_ratio/region_mean": 0.0017818351443565916,
      "epoch": 5.807232429279673,
      "grad_norm": 0.15908323228359222,
      "learning_rate": 1e-06,
      "loss": -0.0248,
      "step": 1133
    },
    {
      "clip_ratio/high_max": 0.002293161494890228,
      "clip_ratio/high_mean": 0.0009758439846336842,
      "clip_ratio/low_mean": 0.0008544246284145629,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018302686075912789,
      "epoch": 5.811898512685914,
      "grad_norm": 0.14060647785663605,
      "learning_rate": 1e-06,
      "loss": -0.0879,
      "step": 1134
    },
    {
      "clip_ratio/high_max": 0.0026730988538474776,
      "clip_ratio/high_mean": 0.0011894568560819607,
      "clip_ratio/low_mean": 0.0007574568626296241,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019469136677798815,
      "epoch": 5.816564596092155,
      "grad_norm": 0.19878049194812775,
      "learning_rate": 1e-06,
      "loss": -0.074,
      "step": 1135
    },
    {
      "clip_ratio/high_max": 0.002440112686599605,
      "clip_ratio/high_mean": 0.0009724094215926016,
      "clip_ratio/low_mean": 0.0008294244266835449,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018018338305409998,
      "epoch": 5.821230679498396,
      "grad_norm": 0.15591241419315338,
      "learning_rate": 1e-06,
      "loss": 0.0256,
      "step": 1136
    },
    {
      "clip_ratio/high_max": 0.005504639950231649,
      "clip_ratio/high_mean": 0.0023303269626921974,
      "clip_ratio/low_mean": 0.001662674556428101,
      "clip_ratio/low_min": 5.6715063692536205e-05,
      "clip_ratio/region_mean": 0.0039930015336722136,
      "epoch": 5.8258967629046365,
      "grad_norm": 0.10306641459465027,
      "learning_rate": 1e-06,
      "loss": -0.0401,
      "step": 1137
    },
    {
      "clip_ratio/high_max": 0.005226370238233358,
      "clip_ratio/high_mean": 0.0016609989979770035,
      "clip_ratio/low_mean": 0.0015529863662777643,
      "clip_ratio/low_min": 3.5734705306822434e-05,
      "clip_ratio/region_mean": 0.0032139852773980238,
      "epoch": 5.830562846310878,
      "grad_norm": 0.10660423338413239,
      "learning_rate": 1e-06,
      "loss": -0.0253,
      "step": 1138
    },
    {
      "clip_ratio/high_max": 0.004313852696213871,
      "clip_ratio/high_mean": 0.0017088600434362888,
      "clip_ratio/low_mean": 0.0017686615319689736,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034775214953697287,
      "epoch": 5.835228929717118,
      "grad_norm": 0.11858677864074707,
      "learning_rate": 1e-06,
      "loss": -0.0253,
      "step": 1139
    },
    {
      "clip_ratio/high_max": 0.004938046971801668,
      "clip_ratio/high_mean": 0.001908992780954577,
      "clip_ratio/low_mean": 0.001893316577479709,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003802309380262159,
      "epoch": 5.83989501312336,
      "grad_norm": 0.11748365312814713,
      "learning_rate": 1e-06,
      "loss": -0.0246,
      "step": 1140
    },
    {
      "clip_ratio/high_max": 0.0049477749853394926,
      "clip_ratio/high_mean": 0.0018091693309543189,
      "clip_ratio/low_mean": 0.0016056250087785884,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034147944024880417,
      "epoch": 5.844561096529601,
      "grad_norm": 0.1080893725156784,
      "learning_rate": 1e-06,
      "loss": -0.0244,
      "step": 1141
    },
    {
      "clip_ratio/high_max": 0.004769791499711573,
      "clip_ratio/high_mean": 0.002059310289041605,
      "clip_ratio/low_mean": 0.001641244802158326,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003700555083923973,
      "epoch": 5.849227179935841,
      "grad_norm": 0.11255943030118942,
      "learning_rate": 1e-06,
      "loss": -0.0391,
      "step": 1142
    },
    {
      "clip_ratio/high_max": 0.0047813797573326156,
      "clip_ratio/high_mean": 0.0017254416125069838,
      "clip_ratio/low_mean": 0.0018506154738133773,
      "clip_ratio/low_min": 7.246376480907202e-05,
      "clip_ratio/region_mean": 0.0035760571772698313,
      "epoch": 5.853893263342083,
      "grad_norm": 0.11120809614658356,
      "learning_rate": 1e-06,
      "loss": 0.0055,
      "step": 1143
    },
    {
      "clip_ratio/high_max": 0.004502664844039828,
      "clip_ratio/high_mean": 0.0017670512679615058,
      "clip_ratio/low_mean": 0.0017467857906012796,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003513837087666616,
      "epoch": 5.858559346748323,
      "grad_norm": 0.11343419551849365,
      "learning_rate": 1e-06,
      "loss": -0.0363,
      "step": 1144
    },
    {
      "clip_ratio/high_max": 0.004410411180288065,
      "clip_ratio/high_mean": 0.0018015401183220092,
      "clip_ratio/low_mean": 0.0019400491728447378,
      "clip_ratio/low_min": 8.901093497115653e-05,
      "clip_ratio/region_mean": 0.0037415893893921748,
      "epoch": 5.863225430154564,
      "grad_norm": 0.11715330928564072,
      "learning_rate": 1e-06,
      "loss": -0.0097,
      "step": 1145
    },
    {
      "clip_ratio/high_max": 0.004683272010879591,
      "clip_ratio/high_mean": 0.001918780766573036,
      "clip_ratio/low_mean": 0.0016194566269405186,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003538237448083237,
      "epoch": 5.867891513560805,
      "grad_norm": 0.10791654884815216,
      "learning_rate": 1e-06,
      "loss": -0.0099,
      "step": 1146
    },
    {
      "clip_ratio/high_max": 0.003957893750339281,
      "clip_ratio/high_mean": 0.0017250058299396187,
      "clip_ratio/low_mean": 0.001734820663841674,
      "clip_ratio/low_min": 4.003843787359074e-05,
      "clip_ratio/region_mean": 0.0034598266502143815,
      "epoch": 5.872557596967046,
      "grad_norm": 0.10622411966323853,
      "learning_rate": 1e-06,
      "loss": -0.0384,
      "step": 1147
    },
    {
      "clip_ratio/high_max": 0.004854585014982149,
      "clip_ratio/high_mean": 0.001847105897468282,
      "clip_ratio/low_mean": 0.0017949430330190808,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036420489923330024,
      "epoch": 5.8772236803732865,
      "grad_norm": 0.10256685316562653,
      "learning_rate": 1e-06,
      "loss": -0.0283,
      "step": 1148
    },
    {
      "clip_ratio/high_max": 0.004042023836518638,
      "clip_ratio/high_mean": 0.0018867390535888262,
      "clip_ratio/low_mean": 0.0019108781634713523,
      "clip_ratio/low_min": 0.0001165127287094947,
      "clip_ratio/region_mean": 0.0037976172316120937,
      "epoch": 5.881889763779528,
      "grad_norm": 0.11155485361814499,
      "learning_rate": 1e-06,
      "loss": -0.0256,
      "step": 1149
    },
    {
      "clip_ratio/high_max": 0.00501972998608835,
      "clip_ratio/high_mean": 0.002004125024541281,
      "clip_ratio/low_mean": 0.0017382519363309257,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037423769827000797,
      "epoch": 5.886555847185768,
      "grad_norm": 0.10853773355484009,
      "learning_rate": 1e-06,
      "loss": 0.109,
      "step": 1150
    },
    {
      "clip_ratio/high_max": 0.004434913855220657,
      "clip_ratio/high_mean": 0.0020659262700064573,
      "clip_ratio/low_mean": 0.0017229504010174423,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037888767546974123,
      "epoch": 5.8912219305920095,
      "grad_norm": 0.11831418424844742,
      "learning_rate": 1e-06,
      "loss": -0.075,
      "step": 1151
    },
    {
      "clip_ratio/high_max": 0.004260365531081334,
      "clip_ratio/high_mean": 0.0017086459702113643,
      "clip_ratio/low_mean": 0.0019094327826678636,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003618078728322871,
      "epoch": 5.89588801399825,
      "grad_norm": 0.1175091415643692,
      "learning_rate": 1e-06,
      "loss": 0.0247,
      "step": 1152
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0709402901785714,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 3824.0,
      "completions/mean_length": 796.6719360351562,
      "completions/mean_terminated_length": 544.7447509765625,
      "completions/min_length": 103.0,
      "completions/min_terminated_length": 103.0,
      "epoch": 6.004666083406241,
      "grad_norm": 0.18242177367210388,
      "learning_rate": 1e-06,
      "loss": -0.0283,
      "num_tokens": 337082867.0,
      "reward": 0.6605747938156128,
      "reward_std": 0.15867824852466583,
      "rewards/simpleverify_reward/mean": 0.6605747938156128,
      "rewards/simpleverify_reward/std": 0.47353073954582214,
      "step": 1153
    },
    {
      "clip_ratio/high_max": 0.002072514696919825,
      "clip_ratio/high_mean": 0.0008343372883246047,
      "clip_ratio/low_mean": 0.0004541730272649147,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012885103205917403,
      "epoch": 6.009332166812482,
      "grad_norm": 0.17426519095897675,
      "learning_rate": 1e-06,
      "loss": -0.0393,
      "step": 1154
    },
    {
      "clip_ratio/high_max": 0.0020436446357052773,
      "clip_ratio/high_mean": 0.0008077742604655214,
      "clip_ratio/low_mean": 0.0004613940982380882,
      "clip_ratio/low_min": 2.2381378585123457e-05,
      "clip_ratio/region_mean": 0.0012691683587036096,
      "epoch": 6.013998250218723,
      "grad_norm": 0.16550327837467194,
      "learning_rate": 1e-06,
      "loss": -0.0506,
      "step": 1155
    },
    {
      "clip_ratio/high_max": 0.0022926710626052227,
      "clip_ratio/high_mean": 0.0008705636064405553,
      "clip_ratio/low_mean": 0.0005356902420317056,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001406253835739335,
      "epoch": 6.0186643336249634,
      "grad_norm": 0.17232991755008698,
      "learning_rate": 1e-06,
      "loss": -0.0157,
      "step": 1156
    },
    {
      "clip_ratio/high_max": 0.0022572078814846464,
      "clip_ratio/high_mean": 0.0008629720759927295,
      "clip_ratio/low_mean": 0.0006348547349261935,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014978268227423541,
      "epoch": 6.023330417031205,
      "grad_norm": 0.18174593150615692,
      "learning_rate": 1e-06,
      "loss": -0.0085,
      "step": 1157
    },
    {
      "clip_ratio/high_max": 0.0021461659562191926,
      "clip_ratio/high_mean": 0.0009425906382602989,
      "clip_ratio/low_mean": 0.0006370857663569041,
      "clip_ratio/low_min": 1.1180679393874016e-05,
      "clip_ratio/region_mean": 0.0015796764273545705,
      "epoch": 6.027996500437445,
      "grad_norm": 0.15307855606079102,
      "learning_rate": 1e-06,
      "loss": -0.0491,
      "step": 1158
    },
    {
      "clip_ratio/high_max": 0.0023902650355012156,
      "clip_ratio/high_mean": 0.0009994628053391352,
      "clip_ratio/low_mean": 0.0005698486793335178,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015693114764872007,
      "epoch": 6.0326625838436865,
      "grad_norm": 0.33632391691207886,
      "learning_rate": 1e-06,
      "loss": -0.0511,
      "step": 1159
    },
    {
      "clip_ratio/high_max": 0.0023203793170978315,
      "clip_ratio/high_mean": 0.0008954471250035567,
      "clip_ratio/low_mean": 0.0005489798259077361,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014444269399973564,
      "epoch": 6.037328667249927,
      "grad_norm": 0.16559644043445587,
      "learning_rate": 1e-06,
      "loss": -0.0009,
      "step": 1160
    },
    {
      "clip_ratio/high_max": 0.002605904774100054,
      "clip_ratio/high_mean": 0.0010145978849323,
      "clip_ratio/low_mean": 0.0006279068811636535,
      "clip_ratio/low_min": 2.8908418244100176e-05,
      "clip_ratio/region_mean": 0.0016425047942902893,
      "epoch": 6.041994750656168,
      "grad_norm": 0.16781704127788544,
      "learning_rate": 1e-06,
      "loss": -0.0392,
      "step": 1161
    },
    {
      "clip_ratio/high_max": 0.002401622485194821,
      "clip_ratio/high_mean": 0.0010046773895737715,
      "clip_ratio/low_mean": 0.0006829894409747794,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016876667868928052,
      "epoch": 6.046660834062409,
      "grad_norm": 0.16037827730178833,
      "learning_rate": 1e-06,
      "loss": -0.0432,
      "step": 1162
    },
    {
      "clip_ratio/high_max": 0.002678324828593759,
      "clip_ratio/high_mean": 0.001070161441930395,
      "clip_ratio/low_mean": 0.0005487863222697342,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016189477719308343,
      "epoch": 6.05132691746865,
      "grad_norm": 0.33192017674446106,
      "learning_rate": 1e-06,
      "loss": -0.0528,
      "step": 1163
    },
    {
      "clip_ratio/high_max": 0.0022977693952270783,
      "clip_ratio/high_mean": 0.0010240583214908838,
      "clip_ratio/low_mean": 0.0007470167329302058,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017710750544210896,
      "epoch": 6.05599300087489,
      "grad_norm": 0.16474823653697968,
      "learning_rate": 1e-06,
      "loss": -0.0411,
      "step": 1164
    },
    {
      "clip_ratio/high_max": 0.002620228857267648,
      "clip_ratio/high_mean": 0.0010338915508327773,
      "clip_ratio/low_mean": 0.0007665776374778943,
      "clip_ratio/low_min": 4.7866147724562325e-05,
      "clip_ratio/region_mean": 0.0018004691773967352,
      "epoch": 6.060659084281132,
      "grad_norm": 0.14066080749034882,
      "learning_rate": 1e-06,
      "loss": -0.0195,
      "step": 1165
    },
    {
      "clip_ratio/high_max": 0.0023299558597500436,
      "clip_ratio/high_mean": 0.0009172357749775983,
      "clip_ratio/low_mean": 0.0007585945604660083,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016758303900132887,
      "epoch": 6.065325167687372,
      "grad_norm": 19.465309143066406,
      "learning_rate": 1e-06,
      "loss": -0.015,
      "step": 1166
    },
    {
      "clip_ratio/high_max": 0.002731207772740163,
      "clip_ratio/high_mean": 0.001106960484321462,
      "clip_ratio/low_mean": 0.0008519178991264198,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001958878361620009,
      "epoch": 6.069991251093613,
      "grad_norm": 0.1689734160900116,
      "learning_rate": 1e-06,
      "loss": -0.0215,
      "step": 1167
    },
    {
      "clip_ratio/high_max": 0.003216896264348179,
      "clip_ratio/high_mean": 0.001274391886909143,
      "clip_ratio/low_mean": 0.0009468493699387182,
      "clip_ratio/low_min": 2.4347487851628102e-05,
      "clip_ratio/region_mean": 0.002221241287770681,
      "epoch": 6.074657334499854,
      "grad_norm": 0.17878632247447968,
      "learning_rate": 1e-06,
      "loss": -0.0317,
      "step": 1168
    },
    {
      "clip_ratio/high_max": 0.006398129946319386,
      "clip_ratio/high_mean": 0.0023172747787612025,
      "clip_ratio/low_mean": 0.001806093407139997,
      "clip_ratio/low_min": 0.00010767232834041351,
      "clip_ratio/region_mean": 0.00412336815497838,
      "epoch": 6.079323417906095,
      "grad_norm": 0.10401533544063568,
      "learning_rate": 1e-06,
      "loss": -0.0291,
      "step": 1169
    },
    {
      "clip_ratio/high_max": 0.005229033122304827,
      "clip_ratio/high_mean": 0.0019465270415821578,
      "clip_ratio/low_mean": 0.001863125438831048,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003809652422205545,
      "epoch": 6.083989501312336,
      "grad_norm": 0.11464226990938187,
      "learning_rate": 1e-06,
      "loss": -0.0403,
      "step": 1170
    },
    {
      "clip_ratio/high_max": 0.004947922789142467,
      "clip_ratio/high_mean": 0.0019519247216521762,
      "clip_ratio/low_mean": 0.001802089642296778,
      "clip_ratio/low_min": 6.003842281643301e-05,
      "clip_ratio/region_mean": 0.0037540144112426788,
      "epoch": 6.088655584718577,
      "grad_norm": 0.11119744181632996,
      "learning_rate": 1e-06,
      "loss": -0.0514,
      "step": 1171
    },
    {
      "clip_ratio/high_max": 0.004886896247626282,
      "clip_ratio/high_mean": 0.0020213665266055614,
      "clip_ratio/low_mean": 0.0019833549740724266,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.004004721544333734,
      "epoch": 6.093321668124818,
      "grad_norm": 0.11953351646661758,
      "learning_rate": 1e-06,
      "loss": -0.0167,
      "step": 1172
    },
    {
      "clip_ratio/high_max": 0.005028811312513426,
      "clip_ratio/high_mean": 0.0019269078438810539,
      "clip_ratio/low_mean": 0.0020151378375885542,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003942045572330244,
      "epoch": 6.0979877515310585,
      "grad_norm": 0.11792458593845367,
      "learning_rate": 1e-06,
      "loss": -0.0095,
      "step": 1173
    },
    {
      "clip_ratio/high_max": 0.005057659742305987,
      "clip_ratio/high_mean": 0.0020642356648750138,
      "clip_ratio/low_mean": 0.0017972054811252747,
      "clip_ratio/low_min": 2.466131809342187e-05,
      "clip_ratio/region_mean": 0.0038614411605522037,
      "epoch": 6.1026538349373,
      "grad_norm": 0.11919765174388885,
      "learning_rate": 1e-06,
      "loss": -0.0499,
      "step": 1174
    },
    {
      "clip_ratio/high_max": 0.005015512273530476,
      "clip_ratio/high_mean": 0.0019917262616218068,
      "clip_ratio/low_mean": 0.0016201725011342205,
      "clip_ratio/low_min": 3.2929398003034294e-05,
      "clip_ratio/region_mean": 0.0036118987191002816,
      "epoch": 6.10731991834354,
      "grad_norm": 0.10362087935209274,
      "learning_rate": 1e-06,
      "loss": -0.0518,
      "step": 1175
    },
    {
      "clip_ratio/high_max": 0.005355516172130592,
      "clip_ratio/high_mean": 0.0019264529473730363,
      "clip_ratio/low_mean": 0.0016400171480199788,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035664700117195025,
      "epoch": 6.111986001749782,
      "grad_norm": 0.10846006870269775,
      "learning_rate": 1e-06,
      "loss": -0.0019,
      "step": 1176
    },
    {
      "clip_ratio/high_max": 0.005369786260416731,
      "clip_ratio/high_mean": 0.001958933575224364,
      "clip_ratio/low_mean": 0.0017523655042168684,
      "clip_ratio/low_min": 5.529749978450127e-05,
      "clip_ratio/region_mean": 0.0037112990394234657,
      "epoch": 6.116652085156022,
      "grad_norm": 0.20648466050624847,
      "learning_rate": 1e-06,
      "loss": -0.04,
      "step": 1177
    },
    {
      "clip_ratio/high_max": 0.0046888306678738445,
      "clip_ratio/high_mean": 0.0020841070472670253,
      "clip_ratio/low_mean": 0.00181154056917876,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003895647605531849,
      "epoch": 6.121318168562263,
      "grad_norm": 0.11252682656049728,
      "learning_rate": 1e-06,
      "loss": -0.044,
      "step": 1178
    },
    {
      "clip_ratio/high_max": 0.00471595144335879,
      "clip_ratio/high_mean": 0.0020018911563965958,
      "clip_ratio/low_mean": 0.0016060811203715275,
      "clip_ratio/low_min": 3.790942355408333e-05,
      "clip_ratio/region_mean": 0.0036079723286093213,
      "epoch": 6.125984251968504,
      "grad_norm": 0.17672204971313477,
      "learning_rate": 1e-06,
      "loss": -0.0536,
      "step": 1179
    },
    {
      "clip_ratio/high_max": 0.0044734130933647975,
      "clip_ratio/high_mean": 0.0020178506383672357,
      "clip_ratio/low_mean": 0.0019025384244741872,
      "clip_ratio/low_min": 2.3165308448369615e-05,
      "clip_ratio/region_mean": 0.003920389121049084,
      "epoch": 6.130650335374745,
      "grad_norm": 0.329829603433609,
      "learning_rate": 1e-06,
      "loss": -0.042,
      "step": 1180
    },
    {
      "clip_ratio/high_max": 0.004576621045998763,
      "clip_ratio/high_mean": 0.0018703969071793836,
      "clip_ratio/low_mean": 0.0017712166081764735,
      "clip_ratio/low_min": 7.933452070574276e-05,
      "clip_ratio/region_mean": 0.0036416134535102174,
      "epoch": 6.135316418780985,
      "grad_norm": 0.10979607701301575,
      "learning_rate": 1e-06,
      "loss": -0.0203,
      "step": 1181
    },
    {
      "clip_ratio/high_max": 0.002820237939886283,
      "clip_ratio/high_mean": 0.001266337185370503,
      "clip_ratio/low_mean": 0.001247524796781363,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0025138619530480355,
      "epoch": 6.139982502187227,
      "grad_norm": 0.16531234979629517,
      "learning_rate": 1e-06,
      "loss": -0.0028,
      "step": 1182
    },
    {
      "clip_ratio/high_max": 0.00467645660683047,
      "clip_ratio/high_mean": 0.0019299036539450753,
      "clip_ratio/low_mean": 0.0019747214591916418,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003904625147697516,
      "epoch": 6.144648585593467,
      "grad_norm": 0.11468078941106796,
      "learning_rate": 1e-06,
      "loss": -0.0224,
      "step": 1183
    },
    {
      "clip_ratio/high_max": 0.00507454094622517,
      "clip_ratio/high_mean": 0.002216409702668898,
      "clip_ratio/low_mean": 0.0019296281898277812,
      "clip_ratio/low_min": 0.00010956369078485295,
      "clip_ratio/region_mean": 0.004146037841564976,
      "epoch": 6.1493146689997085,
      "grad_norm": 0.12081310153007507,
      "learning_rate": 1e-06,
      "loss": -0.0325,
      "step": 1184
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0782645089285714,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4091.0,
      "completions/mean_length": 818.815673828125,
      "completions/mean_terminated_length": 540.5501098632812,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 6.153980752405949,
      "grad_norm": 0.18249818682670593,
      "learning_rate": 1e-06,
      "loss": -0.051,
      "num_tokens": 345663720.0,
      "reward": 0.6590402126312256,
      "reward_std": 0.1498425006866455,
      "rewards/simpleverify_reward/mean": 0.6590401530265808,
      "rewards/simpleverify_reward/std": 0.4740484058856964,
      "step": 1185
    },
    {
      "clip_ratio/high_max": 0.0019249927281634882,
      "clip_ratio/high_mean": 0.0007629003885085694,
      "clip_ratio/low_mean": 0.0005314342215569923,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012943346118845511,
      "epoch": 6.15864683581219,
      "grad_norm": 0.14904159307479858,
      "learning_rate": 1e-06,
      "loss": -0.0101,
      "step": 1186
    },
    {
      "clip_ratio/high_max": 0.002117677671776619,
      "clip_ratio/high_mean": 0.0007956616900628433,
      "clip_ratio/low_mean": 0.0005051299444858159,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013007916350034066,
      "epoch": 6.163312919218431,
      "grad_norm": 0.18026763200759888,
      "learning_rate": 1e-06,
      "loss": -0.0017,
      "step": 1187
    },
    {
      "clip_ratio/high_max": 0.001876542019090266,
      "clip_ratio/high_mean": 0.0006761473482583824,
      "clip_ratio/low_mean": 0.00043008777993236436,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0011062351295549888,
      "epoch": 6.167979002624672,
      "grad_norm": 0.15997542440891266,
      "learning_rate": 1e-06,
      "loss": -0.0118,
      "step": 1188
    },
    {
      "clip_ratio/high_max": 0.002008318668231368,
      "clip_ratio/high_mean": 0.0009143218667304609,
      "clip_ratio/low_mean": 0.00048609018358547473,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014004120457684621,
      "epoch": 6.172645086030913,
      "grad_norm": 0.17452406883239746,
      "learning_rate": 1e-06,
      "loss": -0.0824,
      "step": 1189
    },
    {
      "clip_ratio/high_max": 0.001978471605980303,
      "clip_ratio/high_mean": 0.0008168477779690875,
      "clip_ratio/low_mean": 0.0006283910734055098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00144523890412529,
      "epoch": 6.177311169437154,
      "grad_norm": 0.2948774993419647,
      "learning_rate": 1e-06,
      "loss": -0.0065,
      "step": 1190
    },
    {
      "clip_ratio/high_max": 0.0021932979361736216,
      "clip_ratio/high_mean": 0.0009638114697736455,
      "clip_ratio/low_mean": 0.0006231919232959626,
      "clip_ratio/low_min": 1.5723269825684838e-05,
      "clip_ratio/region_mean": 0.0015870034258114174,
      "epoch": 6.181977252843395,
      "grad_norm": 0.16570262610912323,
      "learning_rate": 1e-06,
      "loss": -0.035,
      "step": 1191
    },
    {
      "clip_ratio/high_max": 0.002613428106997162,
      "clip_ratio/high_mean": 0.000963521282756119,
      "clip_ratio/low_mean": 0.0005710237746825442,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001534545106551377,
      "epoch": 6.186643336249635,
      "grad_norm": 0.15736079216003418,
      "learning_rate": 1e-06,
      "loss": -0.0483,
      "step": 1192
    },
    {
      "clip_ratio/high_max": 0.0024491959520673845,
      "clip_ratio/high_mean": 0.0009756617546372581,
      "clip_ratio/low_mean": 0.0007420453548547812,
      "clip_ratio/low_min": 5.886508006369695e-05,
      "clip_ratio/region_mean": 0.0017177070694742724,
      "epoch": 6.191309419655877,
      "grad_norm": 0.34508633613586426,
      "learning_rate": 1e-06,
      "loss": -0.0271,
      "step": 1193
    },
    {
      "clip_ratio/high_max": 0.0022464419234893285,
      "clip_ratio/high_mean": 0.0009135547807090916,
      "clip_ratio/low_mean": 0.0005562316064242623,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001469786380766891,
      "epoch": 6.195975503062117,
      "grad_norm": 0.1551639586687088,
      "learning_rate": 1e-06,
      "loss": -0.0163,
      "step": 1194
    },
    {
      "clip_ratio/high_max": 0.002401197954895906,
      "clip_ratio/high_mean": 0.0009552566370985005,
      "clip_ratio/low_mean": 0.0008104696398731903,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017657262942520902,
      "epoch": 6.200641586468358,
      "grad_norm": 0.16015034914016724,
      "learning_rate": 1e-06,
      "loss": -0.0157,
      "step": 1195
    },
    {
      "clip_ratio/high_max": 0.0027758564829127863,
      "clip_ratio/high_mean": 0.0010753981696325354,
      "clip_ratio/low_mean": 0.0007095154569469742,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017849136056611314,
      "epoch": 6.205307669874599,
      "grad_norm": 0.13522781431674957,
      "learning_rate": 1e-06,
      "loss": -0.0569,
      "step": 1196
    },
    {
      "clip_ratio/high_max": 0.002522499817132484,
      "clip_ratio/high_mean": 0.0010829261827893788,
      "clip_ratio/low_mean": 0.0006819726731919218,
      "clip_ratio/low_min": 1.3718174159293994e-05,
      "clip_ratio/region_mean": 0.0017648988723522052,
      "epoch": 6.20997375328084,
      "grad_norm": 0.21462500095367432,
      "learning_rate": 1e-06,
      "loss": -0.05,
      "step": 1197
    },
    {
      "clip_ratio/high_max": 0.0022522393410326913,
      "clip_ratio/high_mean": 0.0010421756705909502,
      "clip_ratio/low_mean": 0.000800628297838557,
      "clip_ratio/low_min": 1.6212710761465132e-05,
      "clip_ratio/region_mean": 0.0018428039984428324,
      "epoch": 6.2146398366870805,
      "grad_norm": 0.14750517904758453,
      "learning_rate": 1e-06,
      "loss": -0.0119,
      "step": 1198
    },
    {
      "clip_ratio/high_max": 0.0026642066295607947,
      "clip_ratio/high_mean": 0.001035273053275887,
      "clip_ratio/low_mean": 0.0007639464683961705,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017992195062106475,
      "epoch": 6.219305920093322,
      "grad_norm": 0.16001133620738983,
      "learning_rate": 1e-06,
      "loss": -0.0313,
      "step": 1199
    },
    {
      "clip_ratio/high_max": 0.002767722893622704,
      "clip_ratio/high_mean": 0.0010996754263032926,
      "clip_ratio/low_mean": 0.0007883840153226629,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018880594252550509,
      "epoch": 6.223972003499562,
      "grad_norm": 0.16357660293579102,
      "learning_rate": 1e-06,
      "loss": -0.0246,
      "step": 1200
    },
    {
      "clip_ratio/high_max": 0.006018579952069558,
      "clip_ratio/high_mean": 0.0023861274639784824,
      "clip_ratio/low_mean": 0.001599101200554287,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003985228526289575,
      "epoch": 6.228638086905804,
      "grad_norm": 0.10670291632413864,
      "learning_rate": 1e-06,
      "loss": -0.052,
      "step": 1201
    },
    {
      "clip_ratio/high_max": 0.004198655595246237,
      "clip_ratio/high_mean": 0.001708283551124623,
      "clip_ratio/low_mean": 0.0017101457924582064,
      "clip_ratio/low_min": 7.571168680442497e-05,
      "clip_ratio/region_mean": 0.003418429318116978,
      "epoch": 6.233304170312044,
      "grad_norm": 0.11198136955499649,
      "learning_rate": 1e-06,
      "loss": -0.011,
      "step": 1202
    },
    {
      "clip_ratio/high_max": 0.005343893571989611,
      "clip_ratio/high_mean": 0.0018912964369519614,
      "clip_ratio/low_mean": 0.0018836276194633683,
      "clip_ratio/low_min": 8.408155190409161e-05,
      "clip_ratio/region_mean": 0.0037749240873381495,
      "epoch": 6.237970253718285,
      "grad_norm": 0.12194002419710159,
      "learning_rate": 1e-06,
      "loss": -0.0028,
      "step": 1203
    },
    {
      "clip_ratio/high_max": 0.004177896309556672,
      "clip_ratio/high_mean": 0.0015575670386169804,
      "clip_ratio/low_mean": 0.001511940914497245,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0030695079622091725,
      "epoch": 6.242636337124526,
      "grad_norm": 0.11400716006755829,
      "learning_rate": 1e-06,
      "loss": -0.0126,
      "step": 1204
    },
    {
      "clip_ratio/high_max": 0.004740503485663794,
      "clip_ratio/high_mean": 0.002137157742254203,
      "clip_ratio/low_mean": 0.0014760709636902902,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036132286622887477,
      "epoch": 6.247302420530767,
      "grad_norm": 0.1198422759771347,
      "learning_rate": 1e-06,
      "loss": -0.0833,
      "step": 1205
    },
    {
      "clip_ratio/high_max": 0.0046564669464714825,
      "clip_ratio/high_mean": 0.0018952506725327112,
      "clip_ratio/low_mean": 0.0019569300529838074,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003852180772810243,
      "epoch": 6.251968503937007,
      "grad_norm": 0.11283595114946365,
      "learning_rate": 1e-06,
      "loss": -0.0075,
      "step": 1206
    },
    {
      "clip_ratio/high_max": 0.0046794385416433215,
      "clip_ratio/high_mean": 0.002063173582428135,
      "clip_ratio/low_mean": 0.0018245583032694412,
      "clip_ratio/low_min": 0.00010994718832080252,
      "clip_ratio/region_mean": 0.003887731916620396,
      "epoch": 6.256634587343249,
      "grad_norm": 0.11396816372871399,
      "learning_rate": 1e-06,
      "loss": -0.0359,
      "step": 1207
    },
    {
      "clip_ratio/high_max": 0.005474982172017917,
      "clip_ratio/high_mean": 0.0019591440504882485,
      "clip_ratio/low_mean": 0.001593008233612636,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035521522513590753,
      "epoch": 6.26130067074949,
      "grad_norm": 0.10996533930301666,
      "learning_rate": 1e-06,
      "loss": -0.0491,
      "step": 1208
    },
    {
      "clip_ratio/high_max": 0.004971664020558819,
      "clip_ratio/high_mean": 0.0018457627957104705,
      "clip_ratio/low_mean": 0.0020515508913376834,
      "clip_ratio/low_min": 0.00012034660176141188,
      "clip_ratio/region_mean": 0.003897313741617836,
      "epoch": 6.2659667541557305,
      "grad_norm": 0.10976897925138474,
      "learning_rate": 1e-06,
      "loss": -0.0281,
      "step": 1209
    },
    {
      "clip_ratio/high_max": 0.003971547328546876,
      "clip_ratio/high_mean": 0.001673905384450336,
      "clip_ratio/low_mean": 0.0016067895248852437,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003280694865679834,
      "epoch": 6.270632837561972,
      "grad_norm": 0.09842522442340851,
      "learning_rate": 1e-06,
      "loss": -0.0171,
      "step": 1210
    },
    {
      "clip_ratio/high_max": 0.0043607683328446,
      "clip_ratio/high_mean": 0.0018856956739909947,
      "clip_ratio/low_mean": 0.0018964723094541114,
      "clip_ratio/low_min": 6.259389192564413e-05,
      "clip_ratio/region_mean": 0.003782167928875424,
      "epoch": 6.275298920968212,
      "grad_norm": 0.11316193640232086,
      "learning_rate": 1e-06,
      "loss": -0.0165,
      "step": 1211
    },
    {
      "clip_ratio/high_max": 0.0042263482173439115,
      "clip_ratio/high_mean": 0.0018156182231905404,
      "clip_ratio/low_mean": 0.0015529227275692392,
      "clip_ratio/low_min": 1.8729397197603248e-05,
      "clip_ratio/region_mean": 0.003368540928931907,
      "epoch": 6.2799650043744535,
      "grad_norm": 0.09869983792304993,
      "learning_rate": 1e-06,
      "loss": -0.0576,
      "step": 1212
    },
    {
      "clip_ratio/high_max": 0.004855348859564401,
      "clip_ratio/high_mean": 0.002084437852317933,
      "clip_ratio/low_mean": 0.001737388603942236,
      "clip_ratio/low_min": 1.3718174159293994e-05,
      "clip_ratio/region_mean": 0.003821826510829851,
      "epoch": 6.284631087780694,
      "grad_norm": 0.11986733227968216,
      "learning_rate": 1e-06,
      "loss": -0.0511,
      "step": 1213
    },
    {
      "clip_ratio/high_max": 0.004412332120409701,
      "clip_ratio/high_mean": 0.0018371955775364768,
      "clip_ratio/low_mean": 0.0018056739954772638,
      "clip_ratio/low_min": 0.00014779370030737482,
      "clip_ratio/region_mean": 0.0036428695602808148,
      "epoch": 6.289297171186935,
      "grad_norm": 0.10239291936159134,
      "learning_rate": 1e-06,
      "loss": -0.0127,
      "step": 1214
    },
    {
      "clip_ratio/high_max": 0.004548331198748201,
      "clip_ratio/high_mean": 0.0017318788741249591,
      "clip_ratio/low_mean": 0.0016407292569056153,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003372608101926744,
      "epoch": 6.293963254593176,
      "grad_norm": 0.10203015804290771,
      "learning_rate": 1e-06,
      "loss": -0.0321,
      "step": 1215
    },
    {
      "clip_ratio/high_max": 0.005308132007485256,
      "clip_ratio/high_mean": 0.001979669748834567,
      "clip_ratio/low_mean": 0.001633669660805026,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036133393805357628,
      "epoch": 6.298629337999417,
      "grad_norm": 0.10270557552576065,
      "learning_rate": 1e-06,
      "loss": -0.0255,
      "step": 1216
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0863560267857143,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4007.0,
      "completions/mean_length": 852.9284057617188,
      "completions/mean_terminated_length": 546.39892578125,
      "completions/min_length": 102.0,
      "completions/min_terminated_length": 102.0,
      "epoch": 6.303295421405657,
      "grad_norm": 0.1704927235841751,
      "learning_rate": 1e-06,
      "loss": -0.0395,
      "num_tokens": 354247965.0,
      "reward": 0.6564592719078064,
      "reward_std": 0.16587857902050018,
      "rewards/simpleverify_reward/mean": 0.6564592719078064,
      "rewards/simpleverify_reward/std": 0.4749065339565277,
      "step": 1217
    },
    {
      "clip_ratio/high_max": 0.002146811945749505,
      "clip_ratio/high_mean": 0.0008467401848974987,
      "clip_ratio/low_mean": 0.0005049297005825792,
      "clip_ratio/low_min": 1.367315690004034e-05,
      "clip_ratio/region_mean": 0.0013516699109459296,
      "epoch": 6.307961504811899,
      "grad_norm": 0.15601028501987457,
      "learning_rate": 1e-06,
      "loss": -0.0368,
      "step": 1218
    },
    {
      "clip_ratio/high_max": 0.001707838546280982,
      "clip_ratio/high_mean": 0.0007874560560594546,
      "clip_ratio/low_mean": 0.0005103927223899518,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001297848779358901,
      "epoch": 6.312627588218139,
      "grad_norm": 0.1513843685388565,
      "learning_rate": 1e-06,
      "loss": -0.0558,
      "step": 1219
    },
    {
      "clip_ratio/high_max": 0.002039637894995394,
      "clip_ratio/high_mean": 0.0008051618733588839,
      "clip_ratio/low_mean": 0.0006329370862658834,
      "clip_ratio/low_min": 1.1438506589911412e-05,
      "clip_ratio/region_mean": 0.0014380989778146613,
      "epoch": 6.31729367162438,
      "grad_norm": 0.18281784653663635,
      "learning_rate": 1e-06,
      "loss": -0.0055,
      "step": 1220
    },
    {
      "clip_ratio/high_max": 0.002099801327858586,
      "clip_ratio/high_mean": 0.0007978338417160558,
      "clip_ratio/low_mean": 0.0006451449980886537,
      "clip_ratio/low_min": 3.295685201010201e-05,
      "clip_ratio/region_mean": 0.001442978831619257,
      "epoch": 6.321959755030621,
      "grad_norm": 0.19137556850910187,
      "learning_rate": 1e-06,
      "loss": 0.0367,
      "step": 1221
    },
    {
      "clip_ratio/high_max": 0.0022526787251990754,
      "clip_ratio/high_mean": 0.0009003073064377531,
      "clip_ratio/low_mean": 0.0004778951874868653,
      "clip_ratio/low_min": 1.9672646885737777e-05,
      "clip_ratio/region_mean": 0.0013782024871034082,
      "epoch": 6.326625838436862,
      "grad_norm": 0.1912624090909958,
      "learning_rate": 1e-06,
      "loss": -0.0486,
      "step": 1222
    },
    {
      "clip_ratio/high_max": 0.0025964861852116883,
      "clip_ratio/high_mean": 0.0010127169734914787,
      "clip_ratio/low_mean": 0.0006189005343912868,
      "clip_ratio/low_min": 2.3710166715318337e-05,
      "clip_ratio/region_mean": 0.0016316174805979244,
      "epoch": 6.331291921843103,
      "grad_norm": 0.17751407623291016,
      "learning_rate": 1e-06,
      "loss": -0.0433,
      "step": 1223
    },
    {
      "clip_ratio/high_max": 0.0022765305984648876,
      "clip_ratio/high_mean": 0.0008940868356148712,
      "clip_ratio/low_mean": 0.0007412671930069337,
      "clip_ratio/low_min": 2.9157919925637543e-05,
      "clip_ratio/region_mean": 0.0016353540486306883,
      "epoch": 6.335958005249344,
      "grad_norm": 0.17658056318759918,
      "learning_rate": 1e-06,
      "loss": -0.0183,
      "step": 1224
    },
    {
      "clip_ratio/high_max": 0.002015412970649777,
      "clip_ratio/high_mean": 0.0008641426793474238,
      "clip_ratio/low_mean": 0.0006890414470035466,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015531841272604652,
      "epoch": 6.340624088655585,
      "grad_norm": 0.17431703209877014,
      "learning_rate": 1e-06,
      "loss": -0.0496,
      "step": 1225
    },
    {
      "clip_ratio/high_max": 0.0024399706453550607,
      "clip_ratio/high_mean": 0.0009905127262754831,
      "clip_ratio/low_mean": 0.0006987517472225591,
      "clip_ratio/low_min": 6.957571713428479e-05,
      "clip_ratio/region_mean": 0.0016892644816834945,
      "epoch": 6.3452901720618256,
      "grad_norm": 0.14510475099086761,
      "learning_rate": 1e-06,
      "loss": -0.0308,
      "step": 1226
    },
    {
      "clip_ratio/high_max": 0.002814829713315703,
      "clip_ratio/high_mean": 0.0012531741922430228,
      "clip_ratio/low_mean": 0.0008226851459767204,
      "clip_ratio/low_min": 1.344086012977641e-05,
      "clip_ratio/region_mean": 0.002075859301839955,
      "epoch": 6.349956255468067,
      "grad_norm": 0.17179343104362488,
      "learning_rate": 1e-06,
      "loss": -0.0631,
      "step": 1227
    },
    {
      "clip_ratio/high_max": 0.002603218672447838,
      "clip_ratio/high_mean": 0.0010548194804869127,
      "clip_ratio/low_mean": 0.0009420140968359192,
      "clip_ratio/low_min": 3.553155329427682e-05,
      "clip_ratio/region_mean": 0.001996833576413337,
      "epoch": 6.354622338874307,
      "grad_norm": 0.14865462481975555,
      "learning_rate": 1e-06,
      "loss": -0.0198,
      "step": 1228
    },
    {
      "clip_ratio/high_max": 0.0028972109430469573,
      "clip_ratio/high_mean": 0.0013142562929715496,
      "clip_ratio/low_mean": 0.000696472498930234,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.002010728800087236,
      "epoch": 6.359288422280549,
      "grad_norm": 0.1715545803308487,
      "learning_rate": 1e-06,
      "loss": -0.0856,
      "step": 1229
    },
    {
      "clip_ratio/high_max": 0.0021622092863253783,
      "clip_ratio/high_mean": 0.0009393882737640524,
      "clip_ratio/low_mean": 0.0009024466435221257,
      "clip_ratio/low_min": 0.00012624126884475118,
      "clip_ratio/region_mean": 0.0018418349136481993,
      "epoch": 6.363954505686789,
      "grad_norm": 0.14434809982776642,
      "learning_rate": 1e-06,
      "loss": -0.0654,
      "step": 1230
    },
    {
      "clip_ratio/high_max": 0.002288113973918371,
      "clip_ratio/high_mean": 0.0010011587855842663,
      "clip_ratio/low_mean": 0.0009305912826675922,
      "clip_ratio/low_min": 3.062224277528003e-05,
      "clip_ratio/region_mean": 0.0019317500773468055,
      "epoch": 6.36862058909303,
      "grad_norm": 0.18370015919208527,
      "learning_rate": 1e-06,
      "loss": -0.0274,
      "step": 1231
    },
    {
      "clip_ratio/high_max": 0.0027717963748727925,
      "clip_ratio/high_mean": 0.001122956851759227,
      "clip_ratio/low_mean": 0.0008680083974468289,
      "clip_ratio/low_min": 3.7004145269747823e-05,
      "clip_ratio/region_mean": 0.0019909652255591936,
      "epoch": 6.373286672499271,
      "grad_norm": 0.16649724543094635,
      "learning_rate": 1e-06,
      "loss": -0.0261,
      "step": 1232
    },
    {
      "clip_ratio/high_max": 0.004594765974616166,
      "clip_ratio/high_mean": 0.0019482305069686845,
      "clip_ratio/low_mean": 0.0017197363849845715,
      "clip_ratio/low_min": 7.365940109593794e-05,
      "clip_ratio/region_mean": 0.003667966928333044,
      "epoch": 6.377952755905512,
      "grad_norm": 0.09771525859832764,
      "learning_rate": 1e-06,
      "loss": -0.0403,
      "step": 1233
    },
    {
      "clip_ratio/high_max": 0.004898667881207075,
      "clip_ratio/high_mean": 0.0018849668031180045,
      "clip_ratio/low_mean": 0.0018895881294156425,
      "clip_ratio/low_min": 0.0001461853680666536,
      "clip_ratio/region_mean": 0.003774554963456467,
      "epoch": 6.3826188393117524,
      "grad_norm": 0.11019729822874069,
      "learning_rate": 1e-06,
      "loss": -0.0376,
      "step": 1234
    },
    {
      "clip_ratio/high_max": 0.0046329854521900415,
      "clip_ratio/high_mean": 0.001956934458576143,
      "clip_ratio/low_mean": 0.001620308859855868,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003577243347535841,
      "epoch": 6.387284922717994,
      "grad_norm": 0.11509384214878082,
      "learning_rate": 1e-06,
      "loss": -0.0565,
      "step": 1235
    },
    {
      "clip_ratio/high_max": 0.005114857107400894,
      "clip_ratio/high_mean": 0.0019909675138478633,
      "clip_ratio/low_mean": 0.002072450406558346,
      "clip_ratio/low_min": 0.00013840092287864536,
      "clip_ratio/region_mean": 0.004063417931320146,
      "epoch": 6.391951006124234,
      "grad_norm": 0.11662361770868301,
      "learning_rate": 1e-06,
      "loss": -0.0065,
      "step": 1236
    },
    {
      "clip_ratio/high_max": 0.004130422894377261,
      "clip_ratio/high_mean": 0.0016557166736674844,
      "clip_ratio/low_mean": 0.0022906041849637404,
      "clip_ratio/low_min": 0.00019724996309378184,
      "clip_ratio/region_mean": 0.003946320852264762,
      "epoch": 6.3966170895304755,
      "grad_norm": 0.11197635531425476,
      "learning_rate": 1e-06,
      "loss": 0.0357,
      "step": 1237
    },
    {
      "clip_ratio/high_max": 0.005132495876750909,
      "clip_ratio/high_mean": 0.0020246076237526722,
      "clip_ratio/low_mean": 0.001689822347543668,
      "clip_ratio/low_min": 0.00017310166731476784,
      "clip_ratio/region_mean": 0.003714430014952086,
      "epoch": 6.401283172936716,
      "grad_norm": 0.11977191269397736,
      "learning_rate": 1e-06,
      "loss": -0.0497,
      "step": 1238
    },
    {
      "clip_ratio/high_max": 0.005149161588633433,
      "clip_ratio/high_mean": 0.0020841576406382956,
      "clip_ratio/low_mean": 0.0017784126466722228,
      "clip_ratio/low_min": 1.5078408068802673e-05,
      "clip_ratio/region_mean": 0.003862570272758603,
      "epoch": 6.405949256342957,
      "grad_norm": 0.12726153433322906,
      "learning_rate": 1e-06,
      "loss": -0.0443,
      "step": 1239
    },
    {
      "clip_ratio/high_max": 0.005396588516305201,
      "clip_ratio/high_mean": 0.00198792675655568,
      "clip_ratio/low_mean": 0.0019857112056342885,
      "clip_ratio/low_min": 0.00023187206261354731,
      "clip_ratio/region_mean": 0.003973637925810181,
      "epoch": 6.410615339749198,
      "grad_norm": 0.11807332187891006,
      "learning_rate": 1e-06,
      "loss": -0.0193,
      "step": 1240
    },
    {
      "clip_ratio/high_max": 0.005263083818135783,
      "clip_ratio/high_mean": 0.001937594061018899,
      "clip_ratio/low_mean": 0.001664674262428889,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003602268305257894,
      "epoch": 6.415281423155439,
      "grad_norm": 0.10560297220945358,
      "learning_rate": 1e-06,
      "loss": -0.0504,
      "step": 1241
    },
    {
      "clip_ratio/high_max": 0.005251878552371636,
      "clip_ratio/high_mean": 0.0019670318288262933,
      "clip_ratio/low_mean": 0.00175074038997991,
      "clip_ratio/low_min": 0.0002072963397949934,
      "clip_ratio/region_mean": 0.0037177721678745,
      "epoch": 6.41994750656168,
      "grad_norm": 0.10284769535064697,
      "learning_rate": 1e-06,
      "loss": -0.0315,
      "step": 1242
    },
    {
      "clip_ratio/high_max": 0.005487411763169803,
      "clip_ratio/high_mean": 0.002174908047891222,
      "clip_ratio/low_mean": 0.0018319984519621357,
      "clip_ratio/low_min": 9.408601908944547e-05,
      "clip_ratio/region_mean": 0.004006906572612934,
      "epoch": 6.424613589967921,
      "grad_norm": 0.12225215137004852,
      "learning_rate": 1e-06,
      "loss": -0.0641,
      "step": 1243
    },
    {
      "clip_ratio/high_max": 0.00431483234569896,
      "clip_ratio/high_mean": 0.0018213755975011736,
      "clip_ratio/low_mean": 0.0020509240312094335,
      "clip_ratio/low_min": 0.00010236504022032022,
      "clip_ratio/region_mean": 0.0038722997705917805,
      "epoch": 6.429279673374162,
      "grad_norm": 0.1257714480161667,
      "learning_rate": 1e-06,
      "loss": -0.0206,
      "step": 1244
    },
    {
      "clip_ratio/high_max": 0.006187643710291013,
      "clip_ratio/high_mean": 0.002480797928001266,
      "clip_ratio/low_mean": 0.0016642445298202801,
      "clip_ratio/low_min": 7.527852721977979e-05,
      "clip_ratio/region_mean": 0.004145042417803779,
      "epoch": 6.433945756780402,
      "grad_norm": 0.13330836594104767,
      "learning_rate": 1e-06,
      "loss": -0.0866,
      "step": 1245
    },
    {
      "clip_ratio/high_max": 0.004366725261206739,
      "clip_ratio/high_mean": 0.0018188512294727843,
      "clip_ratio/low_mean": 0.0018376738680672133,
      "clip_ratio/low_min": 0.0002837841821019538,
      "clip_ratio/region_mean": 0.0036565251066349447,
      "epoch": 6.438611840186644,
      "grad_norm": 0.10906616598367691,
      "learning_rate": 1e-06,
      "loss": -0.0661,
      "step": 1246
    },
    {
      "clip_ratio/high_max": 0.0054893064516363665,
      "clip_ratio/high_mean": 0.002120609839039389,
      "clip_ratio/low_mean": 0.0023535639520559926,
      "clip_ratio/low_min": 9.18667356017977e-05,
      "clip_ratio/region_mean": 0.00447417379473336,
      "epoch": 6.443277923592884,
      "grad_norm": 0.1393863707780838,
      "learning_rate": 1e-06,
      "loss": -0.0286,
      "step": 1247
    },
    {
      "clip_ratio/high_max": 0.004559741108096205,
      "clip_ratio/high_mean": 0.0020982206879125442,
      "clip_ratio/low_mean": 0.0020661255875893403,
      "clip_ratio/low_min": 7.400829053949565e-05,
      "clip_ratio/region_mean": 0.0041643462900538,
      "epoch": 6.447944006999125,
      "grad_norm": 0.11689954251050949,
      "learning_rate": 1e-06,
      "loss": -0.0271,
      "step": 1248
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0832868303571429,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4090.0,
      "completions/mean_length": 832.9703369140625,
      "completions/mean_terminated_length": 536.5117797851562,
      "completions/min_length": 58.0,
      "completions/min_terminated_length": 58.0,
      "epoch": 6.452610090405366,
      "grad_norm": 0.18437311053276062,
      "learning_rate": 1e-06,
      "loss": -0.0279,
      "num_tokens": 362704363.0,
      "reward": 0.6664341688156128,
      "reward_std": 0.1563601940870285,
      "rewards/simpleverify_reward/mean": 0.6664341688156128,
      "rewards/simpleverify_reward/std": 0.4715031087398529,
      "step": 1249
    },
    {
      "clip_ratio/high_max": 0.0020737703671329655,
      "clip_ratio/high_mean": 0.0007545195212514955,
      "clip_ratio/low_mean": 0.0005650765860991669,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013195961037126835,
      "epoch": 6.457276173811607,
      "grad_norm": 0.18921905755996704,
      "learning_rate": 1e-06,
      "loss": -0.0131,
      "step": 1250
    },
    {
      "clip_ratio/high_max": 0.002053243668342475,
      "clip_ratio/high_mean": 0.0007782194943501963,
      "clip_ratio/low_mean": 0.0004931878283969127,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012714073091046885,
      "epoch": 6.4619422572178475,
      "grad_norm": 0.15873084962368011,
      "learning_rate": 1e-06,
      "loss": -0.0096,
      "step": 1251
    },
    {
      "clip_ratio/high_max": 0.0022888294042786583,
      "clip_ratio/high_mean": 0.0008206631173379719,
      "clip_ratio/low_mean": 0.00042945601444444037,
      "clip_ratio/low_min": 2.7268761186860502e-05,
      "clip_ratio/region_mean": 0.0012501191413321067,
      "epoch": 6.466608340624089,
      "grad_norm": 0.26511502265930176,
      "learning_rate": 1e-06,
      "loss": -0.0045,
      "step": 1252
    },
    {
      "clip_ratio/high_max": 0.002409052576695103,
      "clip_ratio/high_mean": 0.0008763158675719751,
      "clip_ratio/low_mean": 0.0005598667421509163,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001436182632460259,
      "epoch": 6.471274424030329,
      "grad_norm": 0.1777404248714447,
      "learning_rate": 1e-06,
      "loss": -0.0395,
      "step": 1253
    },
    {
      "clip_ratio/high_max": 0.002106476815242786,
      "clip_ratio/high_mean": 0.000838238598589669,
      "clip_ratio/low_mean": 0.0005465806000302109,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001384819217491895,
      "epoch": 6.475940507436571,
      "grad_norm": 0.20361950993537903,
      "learning_rate": 1e-06,
      "loss": -0.067,
      "step": 1254
    },
    {
      "clip_ratio/high_max": 0.002371962858887855,
      "clip_ratio/high_mean": 0.0010376533973612823,
      "clip_ratio/low_mean": 0.0005475880329868232,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015852413853281178,
      "epoch": 6.480606590842811,
      "grad_norm": 0.1775011420249939,
      "learning_rate": 1e-06,
      "loss": -0.0581,
      "step": 1255
    },
    {
      "clip_ratio/high_max": 0.002263595502881799,
      "clip_ratio/high_mean": 0.0010149803056265227,
      "clip_ratio/low_mean": 0.0006556017633556621,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016705820708011743,
      "epoch": 6.485272674249052,
      "grad_norm": 0.16694331169128418,
      "learning_rate": 1e-06,
      "loss": -0.0429,
      "step": 1256
    },
    {
      "clip_ratio/high_max": 0.0021845341616426595,
      "clip_ratio/high_mean": 0.0009564846168359509,
      "clip_ratio/low_mean": 0.0007550688642368186,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00171155347925378,
      "epoch": 6.489938757655293,
      "grad_norm": 0.18367934226989746,
      "learning_rate": 1e-06,
      "loss": 0.0047,
      "step": 1257
    },
    {
      "clip_ratio/high_max": 0.0020965269177395385,
      "clip_ratio/high_mean": 0.0008520956016582204,
      "clip_ratio/low_mean": 0.0006599311291211052,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015120267344173044,
      "epoch": 6.494604841061534,
      "grad_norm": 0.1605372279882431,
      "learning_rate": 1e-06,
      "loss": -0.0199,
      "step": 1258
    },
    {
      "clip_ratio/high_max": 0.002465185083565302,
      "clip_ratio/high_mean": 0.0009113260202866513,
      "clip_ratio/low_mean": 0.0007340760166698601,
      "clip_ratio/low_min": 5.369415885070339e-05,
      "clip_ratio/region_mean": 0.0016454020369565114,
      "epoch": 6.499270924467774,
      "grad_norm": 0.18609274923801422,
      "learning_rate": 1e-06,
      "loss": -0.029,
      "step": 1259
    },
    {
      "clip_ratio/high_max": 0.002885288467950886,
      "clip_ratio/high_mean": 0.001226793581736274,
      "clip_ratio/low_mean": 0.0005141270394233288,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001740920626616571,
      "epoch": 6.503937007874016,
      "grad_norm": 0.17888261377811432,
      "learning_rate": 1e-06,
      "loss": -0.0992,
      "step": 1260
    },
    {
      "clip_ratio/high_max": 0.0023663549291086383,
      "clip_ratio/high_mean": 0.0010804844005178893,
      "clip_ratio/low_mean": 0.0007774941750540165,
      "clip_ratio/low_min": 1.991714452742599e-05,
      "clip_ratio/region_mean": 0.001857978517364245,
      "epoch": 6.508603091280257,
      "grad_norm": 0.24670886993408203,
      "learning_rate": 1e-06,
      "loss": 0.0081,
      "step": 1261
    },
    {
      "clip_ratio/high_max": 0.0024268515917356126,
      "clip_ratio/high_mean": 0.0009823406107898336,
      "clip_ratio/low_mean": 0.0008366631766421051,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018190037553722505,
      "epoch": 6.5132691746864975,
      "grad_norm": 0.15875528752803802,
      "learning_rate": 1e-06,
      "loss": -0.0198,
      "step": 1262
    },
    {
      "clip_ratio/high_max": 0.0030257625112426467,
      "clip_ratio/high_mean": 0.0012849957784055732,
      "clip_ratio/low_mean": 0.0007378506988970912,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0020228464709362015,
      "epoch": 6.517935258092739,
      "grad_norm": 0.16234177350997925,
      "learning_rate": 1e-06,
      "loss": -0.0786,
      "step": 1263
    },
    {
      "clip_ratio/high_max": 0.0027808162121800706,
      "clip_ratio/high_mean": 0.0010067533930850914,
      "clip_ratio/low_mean": 0.00081771037002909,
      "clip_ratio/low_min": 2.89754279947374e-05,
      "clip_ratio/region_mean": 0.0018244637685711496,
      "epoch": 6.522601341498979,
      "grad_norm": 0.18712608516216278,
      "learning_rate": 1e-06,
      "loss": -0.0535,
      "step": 1264
    },
    {
      "clip_ratio/high_max": 0.005155796650797129,
      "clip_ratio/high_mean": 0.0021140287244634237,
      "clip_ratio/low_mean": 0.0021710636428906582,
      "clip_ratio/low_min": 7.776395068503916e-05,
      "clip_ratio/region_mean": 0.004285092378268018,
      "epoch": 6.5272674249052205,
      "grad_norm": 0.10155811905860901,
      "learning_rate": 1e-06,
      "loss": -0.0288,
      "step": 1265
    },
    {
      "clip_ratio/high_max": 0.004997641284717247,
      "clip_ratio/high_mean": 0.0019011046751984395,
      "clip_ratio/low_mean": 0.0018498868375900201,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003750991469132714,
      "epoch": 6.531933508311461,
      "grad_norm": 0.11834058165550232,
      "learning_rate": 1e-06,
      "loss": -0.014,
      "step": 1266
    },
    {
      "clip_ratio/high_max": 0.004403603772516362,
      "clip_ratio/high_mean": 0.001696750034170691,
      "clip_ratio/low_mean": 0.001828711505368119,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003525461463141255,
      "epoch": 6.536599591717702,
      "grad_norm": 0.10455948114395142,
      "learning_rate": 1e-06,
      "loss": -0.0104,
      "step": 1267
    },
    {
      "clip_ratio/high_max": 0.00482768785150256,
      "clip_ratio/high_mean": 0.0018547420841059648,
      "clip_ratio/low_mean": 0.0019660325760924025,
      "clip_ratio/low_min": 5.827505810884759e-05,
      "clip_ratio/region_mean": 0.0038207746256375685,
      "epoch": 6.541265675123943,
      "grad_norm": 0.11500366777181625,
      "learning_rate": 1e-06,
      "loss": -0.0056,
      "step": 1268
    },
    {
      "clip_ratio/high_max": 0.005339832496247254,
      "clip_ratio/high_mean": 0.0020964208597433753,
      "clip_ratio/low_mean": 0.0015764073596074013,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003672828242997639,
      "epoch": 6.545931758530184,
      "grad_norm": 0.16839399933815002,
      "learning_rate": 1e-06,
      "loss": -0.0403,
      "step": 1269
    },
    {
      "clip_ratio/high_max": 0.004849508506595157,
      "clip_ratio/high_mean": 0.002016175138123799,
      "clip_ratio/low_mean": 0.0018071316762870993,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038233067170949653,
      "epoch": 6.550597841936424,
      "grad_norm": 0.12244345247745514,
      "learning_rate": 1e-06,
      "loss": -0.068,
      "step": 1270
    },
    {
      "clip_ratio/high_max": 0.005067257021437399,
      "clip_ratio/high_mean": 0.0022382496936188545,
      "clip_ratio/low_mean": 0.0017724037643347401,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.004010653487057425,
      "epoch": 6.555263925342666,
      "grad_norm": 0.11108452826738358,
      "learning_rate": 1e-06,
      "loss": -0.0591,
      "step": 1271
    },
    {
      "clip_ratio/high_max": 0.00570140773197636,
      "clip_ratio/high_mean": 0.0021544756782532204,
      "clip_ratio/low_mean": 0.001804309693397954,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003958785404392984,
      "epoch": 6.559930008748906,
      "grad_norm": 0.11429981887340546,
      "learning_rate": 1e-06,
      "loss": -0.0438,
      "step": 1272
    },
    {
      "clip_ratio/high_max": 0.00486232855473645,
      "clip_ratio/high_mean": 0.0020633556341635995,
      "clip_ratio/low_mean": 0.0021252667502267286,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0041886223771143705,
      "epoch": 6.564596092155147,
      "grad_norm": 0.171039417386055,
      "learning_rate": 1e-06,
      "loss": 0.0037,
      "step": 1273
    },
    {
      "clip_ratio/high_max": 0.004511142098635901,
      "clip_ratio/high_mean": 0.001732706052280264,
      "clip_ratio/low_mean": 0.0017383242447976954,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003471030227956362,
      "epoch": 6.569262175561388,
      "grad_norm": 0.10713984817266464,
      "learning_rate": 1e-06,
      "loss": -0.0207,
      "step": 1274
    },
    {
      "clip_ratio/high_max": 0.004778518385137431,
      "clip_ratio/high_mean": 0.0017622845261939801,
      "clip_ratio/low_mean": 0.0017426028571208008,
      "clip_ratio/low_min": 0.00010738831770140678,
      "clip_ratio/region_mean": 0.0035048873542109504,
      "epoch": 6.573928258967629,
      "grad_norm": 0.10906531661748886,
      "learning_rate": 1e-06,
      "loss": -0.0297,
      "step": 1275
    },
    {
      "clip_ratio/high_max": 0.005706370429834351,
      "clip_ratio/high_mean": 0.002383837396337185,
      "clip_ratio/low_mean": 0.001539257813419681,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003923095064237714,
      "epoch": 6.57859434237387,
      "grad_norm": 0.10869808495044708,
      "learning_rate": 1e-06,
      "loss": -0.1001,
      "step": 1276
    },
    {
      "clip_ratio/high_max": 0.004778533839271404,
      "clip_ratio/high_mean": 0.001964031733223237,
      "clip_ratio/low_mean": 0.001914957661938388,
      "clip_ratio/low_min": 1.991714452742599e-05,
      "clip_ratio/region_mean": 0.0038789894024375826,
      "epoch": 6.583260425780111,
      "grad_norm": 0.11235576868057251,
      "learning_rate": 1e-06,
      "loss": 0.0071,
      "step": 1277
    },
    {
      "clip_ratio/high_max": 0.004232354753185064,
      "clip_ratio/high_mean": 0.001736542191792978,
      "clip_ratio/low_mean": 0.001948829783941619,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003685372030304279,
      "epoch": 6.587926509186351,
      "grad_norm": 0.1074599176645279,
      "learning_rate": 1e-06,
      "loss": -0.0206,
      "step": 1278
    },
    {
      "clip_ratio/high_max": 0.005401410424383357,
      "clip_ratio/high_mean": 0.002200720671680756,
      "clip_ratio/low_mean": 0.001666042415308766,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003866763086989522,
      "epoch": 6.592592592592593,
      "grad_norm": 0.115765780210495,
      "learning_rate": 1e-06,
      "loss": -0.0796,
      "step": 1279
    },
    {
      "clip_ratio/high_max": 0.005033837980590761,
      "clip_ratio/high_mean": 0.0017954665381694213,
      "clip_ratio/low_mean": 0.0016746928176871734,
      "clip_ratio/low_min": 9.289076115237549e-05,
      "clip_ratio/region_mean": 0.0034701594413490966,
      "epoch": 6.597258675998834,
      "grad_norm": 0.11097413301467896,
      "learning_rate": 1e-06,
      "loss": -0.0543,
      "step": 1280
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0845424107142857,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4087.0,
      "completions/mean_length": 834.0260009765625,
      "completions/mean_terminated_length": 532.782958984375,
      "completions/min_length": 85.0,
      "completions/min_terminated_length": 85.0,
      "epoch": 6.601924759405074,
      "grad_norm": 0.20398442447185516,
      "learning_rate": 1e-06,
      "loss": -0.0454,
      "num_tokens": 371163879.0,
      "reward": 0.6556919813156128,
      "reward_std": 0.15002141892910004,
      "rewards/simpleverify_reward/mean": 0.6556919813156128,
      "rewards/simpleverify_reward/std": 0.4751586616039276,
      "step": 1281
    },
    {
      "clip_ratio/high_max": 0.002215487271314487,
      "clip_ratio/high_mean": 0.0007864549188525416,
      "clip_ratio/low_mean": 0.0004259348663708806,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001212389764987165,
      "epoch": 6.606590842811316,
      "grad_norm": 0.15625602006912231,
      "learning_rate": 1e-06,
      "loss": -0.0263,
      "step": 1282
    },
    {
      "clip_ratio/high_max": 0.0023638328893866856,
      "clip_ratio/high_mean": 0.0008452327674604021,
      "clip_ratio/low_mean": 0.00041162919524140307,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012568619931698777,
      "epoch": 6.611256926217556,
      "grad_norm": 0.16662487387657166,
      "learning_rate": 1e-06,
      "loss": -0.0606,
      "step": 1283
    },
    {
      "clip_ratio/high_max": 0.002677121628948953,
      "clip_ratio/high_mean": 0.000907784080482088,
      "clip_ratio/low_mean": 0.0004303069626985234,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013380910531850532,
      "epoch": 6.615923009623797,
      "grad_norm": 0.1663932353258133,
      "learning_rate": 1e-06,
      "loss": -0.0634,
      "step": 1284
    },
    {
      "clip_ratio/high_max": 0.002282319521327736,
      "clip_ratio/high_mean": 0.0008145811334543396,
      "clip_ratio/low_mean": 0.0005402464821600006,
      "clip_ratio/low_min": 1.6204303392441943e-05,
      "clip_ratio/region_mean": 0.001354827585601015,
      "epoch": 6.620589093030038,
      "grad_norm": 0.17662297189235687,
      "learning_rate": 1e-06,
      "loss": -0.0471,
      "step": 1285
    },
    {
      "clip_ratio/high_max": 0.002419477932562586,
      "clip_ratio/high_mean": 0.0008632155258965213,
      "clip_ratio/low_mean": 0.0006132813959993655,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014764969346288126,
      "epoch": 6.625255176436279,
      "grad_norm": 0.16501308977603912,
      "learning_rate": 1e-06,
      "loss": -0.0431,
      "step": 1286
    },
    {
      "clip_ratio/high_max": 0.002202585805207491,
      "clip_ratio/high_mean": 0.0008467680581816239,
      "clip_ratio/low_mean": 0.0006372148459377058,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001483982883655699,
      "epoch": 6.6299212598425195,
      "grad_norm": 0.19370634853839874,
      "learning_rate": 1e-06,
      "loss": -0.0171,
      "step": 1287
    },
    {
      "clip_ratio/high_max": 0.0025696512675494887,
      "clip_ratio/high_mean": 0.0009489596814091783,
      "clip_ratio/low_mean": 0.000713155426637968,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001662115115323104,
      "epoch": 6.634587343248761,
      "grad_norm": 0.16496732831001282,
      "learning_rate": 1e-06,
      "loss": -0.0263,
      "step": 1288
    },
    {
      "clip_ratio/high_max": 0.0023115937801776454,
      "clip_ratio/high_mean": 0.0009358461966257892,
      "clip_ratio/low_mean": 0.0006549650615852443,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001590811265486991,
      "epoch": 6.639253426655001,
      "grad_norm": 0.16607387363910675,
      "learning_rate": 1e-06,
      "loss": -0.0148,
      "step": 1289
    },
    {
      "clip_ratio/high_max": 0.0020615349858417176,
      "clip_ratio/high_mean": 0.0008180037402780727,
      "clip_ratio/low_mean": 0.0008169778120645788,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016349815632565878,
      "epoch": 6.6439195100612425,
      "grad_norm": 0.1518193483352661,
      "learning_rate": 1e-06,
      "loss": -0.0078,
      "step": 1290
    },
    {
      "clip_ratio/high_max": 0.0027319323489791714,
      "clip_ratio/high_mean": 0.0011226893511775415,
      "clip_ratio/low_mean": 0.0007116372271411819,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001834326598327607,
      "epoch": 6.648585593467483,
      "grad_norm": 0.1648380607366562,
      "learning_rate": 1e-06,
      "loss": -0.0338,
      "step": 1291
    },
    {
      "clip_ratio/high_max": 0.002710820783249801,
      "clip_ratio/high_mean": 0.0009817145200941013,
      "clip_ratio/low_mean": 0.0009063977213372709,
      "clip_ratio/low_min": 1.6037978639360517e-05,
      "clip_ratio/region_mean": 0.0018881122014136054,
      "epoch": 6.653251676873724,
      "grad_norm": 0.16792446374893188,
      "learning_rate": 1e-06,
      "loss": -0.0176,
      "step": 1292
    },
    {
      "clip_ratio/high_max": 0.002489826343662571,
      "clip_ratio/high_mean": 0.0009659428797021974,
      "clip_ratio/low_mean": 0.0007873951799410861,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017533380560053047,
      "epoch": 6.657917760279965,
      "grad_norm": 0.16345618665218353,
      "learning_rate": 1e-06,
      "loss": -0.0651,
      "step": 1293
    },
    {
      "clip_ratio/high_max": 0.0024271590664284304,
      "clip_ratio/high_mean": 0.000883765984326601,
      "clip_ratio/low_mean": 0.0010309950848750304,
      "clip_ratio/low_min": 7.438704778905958e-05,
      "clip_ratio/region_mean": 0.0019147610437357798,
      "epoch": 6.662583843686206,
      "grad_norm": 0.18774068355560303,
      "learning_rate": 1e-06,
      "loss": -0.006,
      "step": 1294
    },
    {
      "clip_ratio/high_max": 0.0030691239226143807,
      "clip_ratio/high_mean": 0.0011352186629665084,
      "clip_ratio/low_mean": 0.0008994620147859678,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0020346806923043914,
      "epoch": 6.667249927092447,
      "grad_norm": 0.1552181988954544,
      "learning_rate": 1e-06,
      "loss": -0.0288,
      "step": 1295
    },
    {
      "clip_ratio/high_max": 0.00267529708071379,
      "clip_ratio/high_mean": 0.0010187566240347223,
      "clip_ratio/low_mean": 0.0009265052731279866,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001945261872606352,
      "epoch": 6.671916010498688,
      "grad_norm": 0.18326842784881592,
      "learning_rate": 1e-06,
      "loss": -0.0284,
      "step": 1296
    },
    {
      "clip_ratio/high_max": 0.005656683992128819,
      "clip_ratio/high_mean": 0.0024527592868253123,
      "clip_ratio/low_mean": 0.0019266367798991269,
      "clip_ratio/low_min": 0.00011856878518301528,
      "clip_ratio/region_mean": 0.004379396152216941,
      "epoch": 6.676582093904928,
      "grad_norm": 0.10701899230480194,
      "learning_rate": 1e-06,
      "loss": -0.0464,
      "step": 1297
    },
    {
      "clip_ratio/high_max": 0.005006497594877146,
      "clip_ratio/high_mean": 0.0017055287426046561,
      "clip_ratio/low_mean": 0.001643291503569344,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003348820246174,
      "epoch": 6.681248177311169,
      "grad_norm": 0.11276061832904816,
      "learning_rate": 1e-06,
      "loss": -0.027,
      "step": 1298
    },
    {
      "clip_ratio/high_max": 0.004742298056953587,
      "clip_ratio/high_mean": 0.0018760728344204836,
      "clip_ratio/low_mean": 0.0014991972202551551,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033752700401237234,
      "epoch": 6.685914260717411,
      "grad_norm": 0.1114305928349495,
      "learning_rate": 1e-06,
      "loss": -0.0615,
      "step": 1299
    },
    {
      "clip_ratio/high_max": 0.004631867494026665,
      "clip_ratio/high_mean": 0.0018172040581703186,
      "clip_ratio/low_mean": 0.001295792108066962,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003112996091658715,
      "epoch": 6.690580344123651,
      "grad_norm": 0.11501974612474442,
      "learning_rate": 1e-06,
      "loss": -0.0641,
      "step": 1300
    },
    {
      "clip_ratio/high_max": 0.005480855506903026,
      "clip_ratio/high_mean": 0.0018107604737451766,
      "clip_ratio/low_mean": 0.0016421642758359667,
      "clip_ratio/low_min": 7.706534961471334e-05,
      "clip_ratio/region_mean": 0.0034529247423051856,
      "epoch": 6.695246427529892,
      "grad_norm": 0.11503326892852783,
      "learning_rate": 1e-06,
      "loss": -0.048,
      "step": 1301
    },
    {
      "clip_ratio/high_max": 0.004888847732217982,
      "clip_ratio/high_mean": 0.0018378643617324997,
      "clip_ratio/low_mean": 0.0017399261523678433,
      "clip_ratio/low_min": 1.7313019270659424e-05,
      "clip_ratio/region_mean": 0.003577790455892682,
      "epoch": 6.699912510936133,
      "grad_norm": 0.11644850671291351,
      "learning_rate": 1e-06,
      "loss": -0.044,
      "step": 1302
    },
    {
      "clip_ratio/high_max": 0.005241201331955381,
      "clip_ratio/high_mean": 0.001978694406716386,
      "clip_ratio/low_mean": 0.0018992362784047145,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003877930750604719,
      "epoch": 6.704578594342374,
      "grad_norm": 0.11634170264005661,
      "learning_rate": 1e-06,
      "loss": -0.0183,
      "step": 1303
    },
    {
      "clip_ratio/high_max": 0.005474980731378309,
      "clip_ratio/high_mean": 0.0019617874713731,
      "clip_ratio/low_mean": 0.0017697162311378634,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003731503675226122,
      "epoch": 6.7092446777486145,
      "grad_norm": 0.10746832937002182,
      "learning_rate": 1e-06,
      "loss": -0.0271,
      "step": 1304
    },
    {
      "clip_ratio/high_max": 0.00486699357861653,
      "clip_ratio/high_mean": 0.001828067619499052,
      "clip_ratio/low_mean": 0.0017811469497246435,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003609214625612367,
      "epoch": 6.713910761154856,
      "grad_norm": 0.10577815771102905,
      "learning_rate": 1e-06,
      "loss": -0.0157,
      "step": 1305
    },
    {
      "clip_ratio/high_max": 0.0037362330112955533,
      "clip_ratio/high_mean": 0.0014327140743262134,
      "clip_ratio/low_mean": 0.0019113114540232345,
      "clip_ratio/low_min": 6.948304508114234e-05,
      "clip_ratio/region_mean": 0.003344025506521575,
      "epoch": 6.718576844561096,
      "grad_norm": 0.09911538660526276,
      "learning_rate": 1e-06,
      "loss": -0.0085,
      "step": 1306
    },
    {
      "clip_ratio/high_max": 0.005705976247554645,
      "clip_ratio/high_mean": 0.0021988089283695444,
      "clip_ratio/low_mean": 0.0016976686129055452,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038964774721534923,
      "epoch": 6.723242927967338,
      "grad_norm": 0.10936553031206131,
      "learning_rate": 1e-06,
      "loss": -0.0347,
      "step": 1307
    },
    {
      "clip_ratio/high_max": 0.004975000294507481,
      "clip_ratio/high_mean": 0.0018125532697013114,
      "clip_ratio/low_mean": 0.002116371295414865,
      "clip_ratio/low_min": 9.434979438083246e-05,
      "clip_ratio/region_mean": 0.003928924517822452,
      "epoch": 6.727909011373578,
      "grad_norm": 0.11729011684656143,
      "learning_rate": 1e-06,
      "loss": -0.0185,
      "step": 1308
    },
    {
      "clip_ratio/high_max": 0.00514453426876571,
      "clip_ratio/high_mean": 0.0019941599275625776,
      "clip_ratio/low_mean": 0.001806404394301353,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003800564241828397,
      "epoch": 6.732575094779819,
      "grad_norm": 0.11275199055671692,
      "learning_rate": 1e-06,
      "loss": -0.0661,
      "step": 1309
    },
    {
      "clip_ratio/high_max": 0.004617457329004537,
      "clip_ratio/high_mean": 0.0016860247269505635,
      "clip_ratio/low_mean": 0.002243429706140887,
      "clip_ratio/low_min": 0.0002863877161871642,
      "clip_ratio/region_mean": 0.003929454382159747,
      "epoch": 6.73724117818606,
      "grad_norm": 0.11298539489507675,
      "learning_rate": 1e-06,
      "loss": -0.007,
      "step": 1310
    },
    {
      "clip_ratio/high_max": 0.004866485498496331,
      "clip_ratio/high_mean": 0.0019647844710561913,
      "clip_ratio/low_mean": 0.001932533371473255,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038973178016021848,
      "epoch": 6.741907261592301,
      "grad_norm": 0.10804520547389984,
      "learning_rate": 1e-06,
      "loss": -0.0297,
      "step": 1311
    },
    {
      "clip_ratio/high_max": 0.005027879757108167,
      "clip_ratio/high_mean": 0.0018853389883588534,
      "clip_ratio/low_mean": 0.002026876427407842,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003912215426680632,
      "epoch": 6.746573344998541,
      "grad_norm": 0.11780452728271484,
      "learning_rate": 1e-06,
      "loss": -0.0294,
      "step": 1312
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0920061383928571,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4090.0,
      "completions/mean_length": 875.3718872070312,
      "completions/mean_terminated_length": 549.0287475585938,
      "completions/min_length": 100.0,
      "completions/min_terminated_length": 100.0,
      "epoch": 6.751239428404783,
      "grad_norm": 0.17691953480243683,
      "learning_rate": 1e-06,
      "loss": -0.0489,
      "num_tokens": 379747162.0,
      "reward": 0.6428571939468384,
      "reward_std": 0.1478021889925003,
      "rewards/simpleverify_reward/mean": 0.6428571343421936,
      "rewards/simpleverify_reward/std": 0.4791741371154785,
      "step": 1313
    },
    {
      "clip_ratio/high_max": 0.001906260171381291,
      "clip_ratio/high_mean": 0.0007849453104427084,
      "clip_ratio/low_mean": 0.00043045746815550956,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012154028008808382,
      "epoch": 6.755905511811024,
      "grad_norm": 0.15455153584480286,
      "learning_rate": 1e-06,
      "loss": -0.0425,
      "step": 1314
    },
    {
      "clip_ratio/high_max": 0.001524897052149754,
      "clip_ratio/high_mean": 0.0006239451231522253,
      "clip_ratio/low_mean": 0.00047506844839517726,
      "clip_ratio/low_min": 1.5574383724015206e-05,
      "clip_ratio/region_mean": 0.0010990135742758866,
      "epoch": 6.7605715952172645,
      "grad_norm": 0.1451701521873474,
      "learning_rate": 1e-06,
      "loss": -0.0424,
      "step": 1315
    },
    {
      "clip_ratio/high_max": 0.0017683661353657953,
      "clip_ratio/high_mean": 0.0006558477889484493,
      "clip_ratio/low_mean": 0.00047970484774850775,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0011355526330589782,
      "epoch": 6.765237678623506,
      "grad_norm": 0.17063352465629578,
      "learning_rate": 1e-06,
      "loss": -0.0305,
      "step": 1316
    },
    {
      "clip_ratio/high_max": 0.0021927087000221945,
      "clip_ratio/high_mean": 0.0008543516833015019,
      "clip_ratio/low_mean": 0.00045433903778757667,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001308690738369478,
      "epoch": 6.769903762029746,
      "grad_norm": 0.15368495881557465,
      "learning_rate": 1e-06,
      "loss": -0.0733,
      "step": 1317
    },
    {
      "clip_ratio/high_max": 0.002204972661274951,
      "clip_ratio/high_mean": 0.0008919720303310896,
      "clip_ratio/low_mean": 0.0006047590723028407,
      "clip_ratio/low_min": 5.2741667786904145e-05,
      "clip_ratio/region_mean": 0.0014967311253712978,
      "epoch": 6.7745698454359875,
      "grad_norm": 0.18146827816963196,
      "learning_rate": 1e-06,
      "loss": -0.0481,
      "step": 1318
    },
    {
      "clip_ratio/high_max": 0.001990661730815191,
      "clip_ratio/high_mean": 0.000726216256225598,
      "clip_ratio/low_mean": 0.0005277826821838971,
      "clip_ratio/low_min": 1.663561306486372e-05,
      "clip_ratio/region_mean": 0.001253998969332315,
      "epoch": 6.779235928842228,
      "grad_norm": 0.1759994775056839,
      "learning_rate": 1e-06,
      "loss": -0.0289,
      "step": 1319
    },
    {
      "clip_ratio/high_max": 0.0016953256672422867,
      "clip_ratio/high_mean": 0.0007611136843479471,
      "clip_ratio/low_mean": 0.0005802895593660651,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013414032146101817,
      "epoch": 6.783902012248469,
      "grad_norm": 0.14383743703365326,
      "learning_rate": 1e-06,
      "loss": -0.0263,
      "step": 1320
    },
    {
      "clip_ratio/high_max": 0.0019516379761626013,
      "clip_ratio/high_mean": 0.0007484894540539244,
      "clip_ratio/low_mean": 0.0006153910271677887,
      "clip_ratio/low_min": 1.2173743925814051e-05,
      "clip_ratio/region_mean": 0.0013638804775837343,
      "epoch": 6.78856809565471,
      "grad_norm": 0.1594594120979309,
      "learning_rate": 1e-06,
      "loss": -0.0515,
      "step": 1321
    },
    {
      "clip_ratio/high_max": 0.001825968316552462,
      "clip_ratio/high_mean": 0.0007441487150572357,
      "clip_ratio/low_mean": 0.0005924970091655268,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013366457205847837,
      "epoch": 6.793234179060951,
      "grad_norm": 0.16743826866149902,
      "learning_rate": 1e-06,
      "loss": -0.0423,
      "step": 1322
    },
    {
      "clip_ratio/high_max": 0.0022548476554220542,
      "clip_ratio/high_mean": 0.0009680027742433595,
      "clip_ratio/low_mean": 0.0006425620831578271,
      "clip_ratio/low_min": 2.6310250177630223e-05,
      "clip_ratio/region_mean": 0.0016105648646771442,
      "epoch": 6.797900262467191,
      "grad_norm": 0.16004908084869385,
      "learning_rate": 1e-06,
      "loss": -0.0746,
      "step": 1323
    },
    {
      "clip_ratio/high_max": 0.0022688432363793254,
      "clip_ratio/high_mean": 0.0009288006385759218,
      "clip_ratio/low_mean": 0.0006257376398934866,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001554538252094062,
      "epoch": 6.802566345873433,
      "grad_norm": 0.18003647029399872,
      "learning_rate": 1e-06,
      "loss": -0.0203,
      "step": 1324
    },
    {
      "clip_ratio/high_max": 0.002170684554585023,
      "clip_ratio/high_mean": 0.0008112687664834084,
      "clip_ratio/low_mean": 0.0007583111546409782,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015695799447712488,
      "epoch": 6.807232429279673,
      "grad_norm": 0.16193418204784393,
      "learning_rate": 1e-06,
      "loss": -0.0116,
      "step": 1325
    },
    {
      "clip_ratio/high_max": 0.0020727996015921235,
      "clip_ratio/high_mean": 0.0008878374073901796,
      "clip_ratio/low_mean": 0.0006438633945435868,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015317008037527557,
      "epoch": 6.811898512685914,
      "grad_norm": 0.16347189247608185,
      "learning_rate": 1e-06,
      "loss": -0.0395,
      "step": 1326
    },
    {
      "clip_ratio/high_max": 0.002393558643234428,
      "clip_ratio/high_mean": 0.0009384025979670696,
      "clip_ratio/low_mean": 0.0007500589053961448,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016884614888112992,
      "epoch": 6.816564596092155,
      "grad_norm": 0.15207186341285706,
      "learning_rate": 1e-06,
      "loss": -0.043,
      "step": 1327
    },
    {
      "clip_ratio/high_max": 0.0025760254211490974,
      "clip_ratio/high_mean": 0.0010946907386824023,
      "clip_ratio/low_mean": 0.0008016890255930775,
      "clip_ratio/low_min": 6.64626459183637e-05,
      "clip_ratio/region_mean": 0.0018963798211188987,
      "epoch": 6.821230679498396,
      "grad_norm": 0.16872480511665344,
      "learning_rate": 1e-06,
      "loss": -0.0435,
      "step": 1328
    },
    {
      "clip_ratio/high_max": 0.006034898586221971,
      "clip_ratio/high_mean": 0.0022732440993422642,
      "clip_ratio/low_mean": 0.001452861906727776,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037261059042066336,
      "epoch": 6.8258967629046365,
      "grad_norm": 0.09486385434865952,
      "learning_rate": 1e-06,
      "loss": -0.0497,
      "step": 1329
    },
    {
      "clip_ratio/high_max": 0.0044036493709427305,
      "clip_ratio/high_mean": 0.001887764796265401,
      "clip_ratio/low_mean": 0.0015684372774558142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003456202073721215,
      "epoch": 6.830562846310878,
      "grad_norm": 0.11497434973716736,
      "learning_rate": 1e-06,
      "loss": -0.0433,
      "step": 1330
    },
    {
      "clip_ratio/high_max": 0.004667037501349114,
      "clip_ratio/high_mean": 0.001613572116184514,
      "clip_ratio/low_mean": 0.001647188651986653,
      "clip_ratio/low_min": 0.00018259330681758001,
      "clip_ratio/region_mean": 0.0032607608009129763,
      "epoch": 6.835228929717118,
      "grad_norm": 0.1100429967045784,
      "learning_rate": 1e-06,
      "loss": -0.0431,
      "step": 1331
    },
    {
      "clip_ratio/high_max": 0.004490446168347262,
      "clip_ratio/high_mean": 0.0017626873268454801,
      "clip_ratio/low_mean": 0.0017411917870049365,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003503879110212438,
      "epoch": 6.83989501312336,
      "grad_norm": 0.11178852617740631,
      "learning_rate": 1e-06,
      "loss": -0.0314,
      "step": 1332
    },
    {
      "clip_ratio/high_max": 0.0048140449725906365,
      "clip_ratio/high_mean": 0.0019473693137115333,
      "clip_ratio/low_mean": 0.0013984386205265764,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033458078833064064,
      "epoch": 6.844561096529601,
      "grad_norm": 0.09910988062620163,
      "learning_rate": 1e-06,
      "loss": -0.074,
      "step": 1333
    },
    {
      "clip_ratio/high_max": 0.004908949893433601,
      "clip_ratio/high_mean": 0.0018512294846004806,
      "clip_ratio/low_mean": 0.0017365465391776524,
      "clip_ratio/low_min": 0.00014344750525197014,
      "clip_ratio/region_mean": 0.0035877760383300483,
      "epoch": 6.849227179935841,
      "grad_norm": 0.10603483766317368,
      "learning_rate": 1e-06,
      "loss": -0.0491,
      "step": 1334
    },
    {
      "clip_ratio/high_max": 0.0044283151073614135,
      "clip_ratio/high_mean": 0.0017259447886317503,
      "clip_ratio/low_mean": 0.0018376313601038419,
      "clip_ratio/low_min": 0.00011309658293612301,
      "clip_ratio/region_mean": 0.0035635762324091047,
      "epoch": 6.853893263342083,
      "grad_norm": 0.10247107595205307,
      "learning_rate": 1e-06,
      "loss": -0.0298,
      "step": 1335
    },
    {
      "clip_ratio/high_max": 0.004174747125944123,
      "clip_ratio/high_mean": 0.0016834230918902904,
      "clip_ratio/low_mean": 0.0015430074236064684,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0032264305191347376,
      "epoch": 6.858559346748323,
      "grad_norm": 0.10308176279067993,
      "learning_rate": 1e-06,
      "loss": -0.027,
      "step": 1336
    },
    {
      "clip_ratio/high_max": 0.004295174905564636,
      "clip_ratio/high_mean": 0.0016767433189670555,
      "clip_ratio/low_mean": 0.0016643594535707962,
      "clip_ratio/low_min": 7.304246537387371e-05,
      "clip_ratio/region_mean": 0.0033411026961402968,
      "epoch": 6.863225430154564,
      "grad_norm": 0.09953628480434418,
      "learning_rate": 1e-06,
      "loss": -0.0523,
      "step": 1337
    },
    {
      "clip_ratio/high_max": 0.004616234728018753,
      "clip_ratio/high_mean": 0.0016249923919531284,
      "clip_ratio/low_mean": 0.0014194940486049745,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0030444864605669864,
      "epoch": 6.867891513560805,
      "grad_norm": 0.09776140004396439,
      "learning_rate": 1e-06,
      "loss": -0.0431,
      "step": 1338
    },
    {
      "clip_ratio/high_max": 0.00528674409724772,
      "clip_ratio/high_mean": 0.0019236216685385443,
      "clip_ratio/low_mean": 0.0017211059221153846,
      "clip_ratio/low_min": 4.8744346713647246e-05,
      "clip_ratio/region_mean": 0.0036447275488171726,
      "epoch": 6.872557596967046,
      "grad_norm": 0.11920148879289627,
      "learning_rate": 1e-06,
      "loss": -0.0756,
      "step": 1339
    },
    {
      "clip_ratio/high_max": 0.004785435143276118,
      "clip_ratio/high_mean": 0.0019439743300608825,
      "clip_ratio/low_mean": 0.0019317464757477865,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003875720751238987,
      "epoch": 6.8772236803732865,
      "grad_norm": 0.12807448208332062,
      "learning_rate": 1e-06,
      "loss": -0.0214,
      "step": 1340
    },
    {
      "clip_ratio/high_max": 0.0038489385187858716,
      "clip_ratio/high_mean": 0.0015965677812346257,
      "clip_ratio/low_mean": 0.001979966225917451,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035765339416684583,
      "epoch": 6.881889763779528,
      "grad_norm": 0.10300271958112717,
      "learning_rate": 1e-06,
      "loss": -0.0124,
      "step": 1341
    },
    {
      "clip_ratio/high_max": 0.0043094366701552644,
      "clip_ratio/high_mean": 0.001805086252716137,
      "clip_ratio/low_mean": 0.0016290790117636789,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034341652935836464,
      "epoch": 6.886555847185768,
      "grad_norm": 0.10691041499376297,
      "learning_rate": 1e-06,
      "loss": -0.0404,
      "step": 1342
    },
    {
      "clip_ratio/high_max": 0.004835788655327633,
      "clip_ratio/high_mean": 0.0018454608316460508,
      "clip_ratio/low_mean": 0.0016319052192557137,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034773660299833864,
      "epoch": 6.8912219305920095,
      "grad_norm": 0.1092439815402031,
      "learning_rate": 1e-06,
      "loss": -0.0438,
      "step": 1343
    },
    {
      "clip_ratio/high_max": 0.0047408887767232955,
      "clip_ratio/high_mean": 0.00193258942454122,
      "clip_ratio/low_mean": 0.0018805811305355746,
      "clip_ratio/low_min": 9.941027383320034e-05,
      "clip_ratio/region_mean": 0.0038131705223349854,
      "epoch": 6.89588801399825,
      "grad_norm": 0.11647233366966248,
      "learning_rate": 1e-06,
      "loss": -0.0445,
      "step": 1344
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0899832589285714,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4053.0,
      "completions/mean_length": 863.3488159179688,
      "completions/mean_terminated_length": 543.7013549804688,
      "completions/min_length": 62.0,
      "completions/min_terminated_length": 62.0,
      "epoch": 7.004666083406241,
      "grad_norm": 0.17813444137573242,
      "learning_rate": 1e-06,
      "loss": -0.0041,
      "num_tokens": 388280834.0,
      "reward": 0.6662248969078064,
      "reward_std": 0.14652320742607117,
      "rewards/simpleverify_reward/mean": 0.6662248969078064,
      "rewards/simpleverify_reward/std": 0.4715769290924072,
      "step": 1345
    },
    {
      "clip_ratio/high_max": 0.00196512307957164,
      "clip_ratio/high_mean": 0.0007401004359053331,
      "clip_ratio/low_mean": 0.0003890584271175612,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0011291588671156205,
      "epoch": 7.009332166812482,
      "grad_norm": 0.1573495864868164,
      "learning_rate": 1e-06,
      "loss": -0.057,
      "step": 1346
    },
    {
      "clip_ratio/high_max": 0.0022226737237360794,
      "clip_ratio/high_mean": 0.0008395700351684354,
      "clip_ratio/low_mean": 0.00041905999614755274,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012586300326802302,
      "epoch": 7.013998250218723,
      "grad_norm": 0.16541199386119843,
      "learning_rate": 1e-06,
      "loss": -0.063,
      "step": 1347
    },
    {
      "clip_ratio/high_max": 0.002010684343986213,
      "clip_ratio/high_mean": 0.0008371070343855536,
      "clip_ratio/low_mean": 0.00048016077971624327,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00131726781182806,
      "epoch": 7.0186643336249634,
      "grad_norm": 0.15292195975780487,
      "learning_rate": 1e-06,
      "loss": -0.05,
      "step": 1348
    },
    {
      "clip_ratio/high_max": 0.0018231259600725025,
      "clip_ratio/high_mean": 0.0007352385819103802,
      "clip_ratio/low_mean": 0.0005110916599733173,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001246330237336224,
      "epoch": 7.023330417031205,
      "grad_norm": 0.1643696278333664,
      "learning_rate": 1e-06,
      "loss": -0.0207,
      "step": 1349
    },
    {
      "clip_ratio/high_max": 0.002042754211288411,
      "clip_ratio/high_mean": 0.0008114329157251632,
      "clip_ratio/low_mean": 0.0005021154647693038,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013135483568476047,
      "epoch": 7.027996500437445,
      "grad_norm": 0.16176468133926392,
      "learning_rate": 1e-06,
      "loss": -0.0233,
      "step": 1350
    },
    {
      "clip_ratio/high_max": 0.0025353020973852836,
      "clip_ratio/high_mean": 0.0009854450399870984,
      "clip_ratio/low_mean": 0.0005678753614120069,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015533203804807272,
      "epoch": 7.0326625838436865,
      "grad_norm": 0.1666657030582428,
      "learning_rate": 1e-06,
      "loss": -0.0372,
      "step": 1351
    },
    {
      "clip_ratio/high_max": 0.0022654654894722626,
      "clip_ratio/high_mean": 0.0009013194921863033,
      "clip_ratio/low_mean": 0.000599218793468026,
      "clip_ratio/low_min": 2.0620256691472605e-05,
      "clip_ratio/region_mean": 0.0015005382920207921,
      "epoch": 7.037328667249927,
      "grad_norm": 0.15766793489456177,
      "learning_rate": 1e-06,
      "loss": -0.0508,
      "step": 1352
    },
    {
      "clip_ratio/high_max": 0.0025432959155295976,
      "clip_ratio/high_mean": 0.0009783390960365068,
      "clip_ratio/low_mean": 0.000559994508876116,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015383336030936334,
      "epoch": 7.041994750656168,
      "grad_norm": 0.32127246260643005,
      "learning_rate": 1e-06,
      "loss": -0.048,
      "step": 1353
    },
    {
      "clip_ratio/high_max": 0.0019405158891458996,
      "clip_ratio/high_mean": 0.0008728833381610457,
      "clip_ratio/low_mean": 0.0006260184018174186,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014989017327025067,
      "epoch": 7.046660834062409,
      "grad_norm": 0.15023750066757202,
      "learning_rate": 1e-06,
      "loss": -0.0504,
      "step": 1354
    },
    {
      "clip_ratio/high_max": 0.0020431004886631854,
      "clip_ratio/high_mean": 0.0008276506696347496,
      "clip_ratio/low_mean": 0.0006937142552487785,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001521364953077864,
      "epoch": 7.05132691746865,
      "grad_norm": 0.1613745391368866,
      "learning_rate": 1e-06,
      "loss": 0.0219,
      "step": 1355
    },
    {
      "clip_ratio/high_max": 0.0024921642507251818,
      "clip_ratio/high_mean": 0.0009526120193186216,
      "clip_ratio/low_mean": 0.0005702993935301492,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001522911414213013,
      "epoch": 7.05599300087489,
      "grad_norm": 0.15816907584667206,
      "learning_rate": 1e-06,
      "loss": -0.0929,
      "step": 1356
    },
    {
      "clip_ratio/high_max": 0.0027314904873492196,
      "clip_ratio/high_mean": 0.0010463655180501519,
      "clip_ratio/low_mean": 0.0007975598318807897,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001843925376306288,
      "epoch": 7.060659084281132,
      "grad_norm": 0.1678362488746643,
      "learning_rate": 1e-06,
      "loss": -0.0132,
      "step": 1357
    },
    {
      "clip_ratio/high_max": 0.0022472815253422596,
      "clip_ratio/high_mean": 0.0009136543012573384,
      "clip_ratio/low_mean": 0.0006535810296099953,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015672353511035908,
      "epoch": 7.065325167687372,
      "grad_norm": 0.15273858606815338,
      "learning_rate": 1e-06,
      "loss": -0.0507,
      "step": 1358
    },
    {
      "clip_ratio/high_max": 0.002006586561037693,
      "clip_ratio/high_mean": 0.0008285951644211309,
      "clip_ratio/low_mean": 0.000700228871210129,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015288240610971116,
      "epoch": 7.069991251093613,
      "grad_norm": 0.15688160061836243,
      "learning_rate": 1e-06,
      "loss": -0.0464,
      "step": 1359
    },
    {
      "clip_ratio/high_max": 0.0027816422007163055,
      "clip_ratio/high_mean": 0.0010781456767290365,
      "clip_ratio/low_mean": 0.0008850695403452846,
      "clip_ratio/low_min": 9.020498328027315e-05,
      "clip_ratio/region_mean": 0.0019632151888799854,
      "epoch": 7.074657334499854,
      "grad_norm": 0.16717086732387543,
      "learning_rate": 1e-06,
      "loss": -0.0585,
      "step": 1360
    },
    {
      "clip_ratio/high_max": 0.005641895477310754,
      "clip_ratio/high_mean": 0.002011078944633482,
      "clip_ratio/low_mean": 0.002089990914100781,
      "clip_ratio/low_min": 3.89650886063464e-05,
      "clip_ratio/region_mean": 0.004101069891476072,
      "epoch": 7.079323417906095,
      "grad_norm": 0.09761442989110947,
      "learning_rate": 1e-06,
      "loss": -0.0049,
      "step": 1361
    },
    {
      "clip_ratio/high_max": 0.004146211882471107,
      "clip_ratio/high_mean": 0.0018958972541440744,
      "clip_ratio/low_mean": 0.0015333277688114322,
      "clip_ratio/low_min": 1.711859840725083e-05,
      "clip_ratio/region_mean": 0.0034292250784346834,
      "epoch": 7.083989501312336,
      "grad_norm": 0.10227840393781662,
      "learning_rate": 1e-06,
      "loss": -0.0578,
      "step": 1362
    },
    {
      "clip_ratio/high_max": 0.0043850333750015125,
      "clip_ratio/high_mean": 0.001878627903352026,
      "clip_ratio/low_mean": 0.001450759133149404,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00332938703650143,
      "epoch": 7.088655584718577,
      "grad_norm": 0.11053407192230225,
      "learning_rate": 1e-06,
      "loss": -0.0638,
      "step": 1363
    },
    {
      "clip_ratio/high_max": 0.004361015395261347,
      "clip_ratio/high_mean": 0.0019230335019528866,
      "clip_ratio/low_mean": 0.0016113154924823903,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035343490162631497,
      "epoch": 7.093321668124818,
      "grad_norm": 0.10813720524311066,
      "learning_rate": 1e-06,
      "loss": -0.0507,
      "step": 1364
    },
    {
      "clip_ratio/high_max": 0.004132738540647551,
      "clip_ratio/high_mean": 0.0015850717500143219,
      "clip_ratio/low_mean": 0.0017067740809579846,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0032918459037318826,
      "epoch": 7.0979877515310585,
      "grad_norm": 0.11003698408603668,
      "learning_rate": 1e-06,
      "loss": -0.0216,
      "step": 1365
    },
    {
      "clip_ratio/high_max": 0.004409364344610367,
      "clip_ratio/high_mean": 0.0016792884925962426,
      "clip_ratio/low_mean": 0.001629454465728486,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033087429619627073,
      "epoch": 7.1026538349373,
      "grad_norm": 0.10226931422948837,
      "learning_rate": 1e-06,
      "loss": -0.0241,
      "step": 1366
    },
    {
      "clip_ratio/high_max": 0.005147559306351468,
      "clip_ratio/high_mean": 0.001902114148833789,
      "clip_ratio/low_mean": 0.0015262373508448945,
      "clip_ratio/low_min": 5.9222995332675055e-05,
      "clip_ratio/region_mean": 0.00342835157061927,
      "epoch": 7.10731991834354,
      "grad_norm": 0.11141891777515411,
      "learning_rate": 1e-06,
      "loss": -0.038,
      "step": 1367
    },
    {
      "clip_ratio/high_max": 0.004734716727398336,
      "clip_ratio/high_mean": 0.0019135097281832714,
      "clip_ratio/low_mean": 0.0016106326511362568,
      "clip_ratio/low_min": 0.00010110577932209708,
      "clip_ratio/region_mean": 0.0035241423174738884,
      "epoch": 7.111986001749782,
      "grad_norm": 0.1046186313033104,
      "learning_rate": 1e-06,
      "loss": -0.0517,
      "step": 1368
    },
    {
      "clip_ratio/high_max": 0.005908478982746601,
      "clip_ratio/high_mean": 0.002036659760051407,
      "clip_ratio/low_mean": 0.0017616262812225614,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037982860521879047,
      "epoch": 7.116652085156022,
      "grad_norm": 0.13145965337753296,
      "learning_rate": 1e-06,
      "loss": -0.0492,
      "step": 1369
    },
    {
      "clip_ratio/high_max": 0.004432545145391487,
      "clip_ratio/high_mean": 0.0017289940151385963,
      "clip_ratio/low_mean": 0.0013501349712896626,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003079129040997941,
      "epoch": 7.121318168562263,
      "grad_norm": 0.10737859457731247,
      "learning_rate": 1e-06,
      "loss": -0.0511,
      "step": 1370
    },
    {
      "clip_ratio/high_max": 0.003907241902197711,
      "clip_ratio/high_mean": 0.0017393593007000163,
      "clip_ratio/low_mean": 0.0018372789018030744,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00357663808244979,
      "epoch": 7.125984251968504,
      "grad_norm": 0.12362054735422134,
      "learning_rate": 1e-06,
      "loss": 0.021,
      "step": 1371
    },
    {
      "clip_ratio/high_max": 0.0046059713495196775,
      "clip_ratio/high_mean": 0.0018688447016756982,
      "clip_ratio/low_mean": 0.0014330796243484656,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003301924305560533,
      "epoch": 7.130650335374745,
      "grad_norm": 0.11188887804746628,
      "learning_rate": 1e-06,
      "loss": -0.0938,
      "step": 1372
    },
    {
      "clip_ratio/high_max": 0.0047178358800010756,
      "clip_ratio/high_mean": 0.001794469528249465,
      "clip_ratio/low_mean": 0.0019175440938852262,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037120135966688395,
      "epoch": 7.135316418780985,
      "grad_norm": 0.1485195755958557,
      "learning_rate": 1e-06,
      "loss": -0.0141,
      "step": 1373
    },
    {
      "clip_ratio/high_max": 0.004903913533780724,
      "clip_ratio/high_mean": 0.0019435995709500276,
      "clip_ratio/low_mean": 0.001473207804338017,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034168073907494545,
      "epoch": 7.139982502187227,
      "grad_norm": 0.10329054296016693,
      "learning_rate": 1e-06,
      "loss": -0.0515,
      "step": 1374
    },
    {
      "clip_ratio/high_max": 0.003859941331029404,
      "clip_ratio/high_mean": 0.0016578102586208843,
      "clip_ratio/low_mean": 0.001723356243019225,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033811665780376643,
      "epoch": 7.144648585593467,
      "grad_norm": 0.11751967668533325,
      "learning_rate": 1e-06,
      "loss": -0.0473,
      "step": 1375
    },
    {
      "clip_ratio/high_max": 0.005230279944953509,
      "clip_ratio/high_mean": 0.0018768908958008979,
      "clip_ratio/low_mean": 0.001889971495984355,
      "clip_ratio/low_min": 0.00016869320097612217,
      "clip_ratio/region_mean": 0.003766862311749719,
      "epoch": 7.1493146689997085,
      "grad_norm": 0.10344547033309937,
      "learning_rate": 1e-06,
      "loss": -0.0594,
      "step": 1376
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0876116071428571,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4080.0,
      "completions/mean_length": 855.8145751953125,
      "completions/mean_terminated_length": 544.6774291992188,
      "completions/min_length": 103.0,
      "completions/min_terminated_length": 103.0,
      "epoch": 7.153980752405949,
      "grad_norm": 0.1809016764163971,
      "learning_rate": 1e-06,
      "loss": -0.0312,
      "num_tokens": 396821783.0,
      "reward": 0.6712472438812256,
      "reward_std": 0.15039551258087158,
      "rewards/simpleverify_reward/mean": 0.6712471842765808,
      "rewards/simpleverify_reward/std": 0.46977630257606506,
      "step": 1377
    },
    {
      "clip_ratio/high_max": 0.0018236126634292305,
      "clip_ratio/high_mean": 0.0007716597483522492,
      "clip_ratio/low_mean": 0.0004701726845723897,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001241832444065949,
      "epoch": 7.15864683581219,
      "grad_norm": 0.17265312373638153,
      "learning_rate": 1e-06,
      "loss": -0.0118,
      "step": 1378
    },
    {
      "clip_ratio/high_max": 0.0018742871543508954,
      "clip_ratio/high_mean": 0.0008148845290634199,
      "clip_ratio/low_mean": 0.0004923002179566538,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013071847533865366,
      "epoch": 7.163312919218431,
      "grad_norm": 0.18306361138820648,
      "learning_rate": 1e-06,
      "loss": -0.0475,
      "step": 1379
    },
    {
      "clip_ratio/high_max": 0.0020929436868755147,
      "clip_ratio/high_mean": 0.000829510898256558,
      "clip_ratio/low_mean": 0.0005188782524783164,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013483892034855671,
      "epoch": 7.167979002624672,
      "grad_norm": 0.16039815545082092,
      "learning_rate": 1e-06,
      "loss": -0.0348,
      "step": 1380
    },
    {
      "clip_ratio/high_max": 0.002377553861151682,
      "clip_ratio/high_mean": 0.0009278282250306802,
      "clip_ratio/low_mean": 0.0005433923106465954,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014712205666000955,
      "epoch": 7.172645086030913,
      "grad_norm": 0.1640148013830185,
      "learning_rate": 1e-06,
      "loss": -0.0759,
      "step": 1381
    },
    {
      "clip_ratio/high_max": 0.0019394484988879412,
      "clip_ratio/high_mean": 0.0008216792430175701,
      "clip_ratio/low_mean": 0.0005633455002680421,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013850247487425804,
      "epoch": 7.177311169437154,
      "grad_norm": 0.18771474063396454,
      "learning_rate": 1e-06,
      "loss": -0.0171,
      "step": 1382
    },
    {
      "clip_ratio/high_max": 0.002581395805464126,
      "clip_ratio/high_mean": 0.0010066006580018438,
      "clip_ratio/low_mean": 0.0005797643070764025,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015863649787206668,
      "epoch": 7.181977252843395,
      "grad_norm": 0.15857377648353577,
      "learning_rate": 1e-06,
      "loss": -0.0421,
      "step": 1383
    },
    {
      "clip_ratio/high_max": 0.002351253788219765,
      "clip_ratio/high_mean": 0.000866480566401151,
      "clip_ratio/low_mean": 0.0005698717595805647,
      "clip_ratio/low_min": 1.1390559848223347e-05,
      "clip_ratio/region_mean": 0.0014363523296196945,
      "epoch": 7.186643336249635,
      "grad_norm": 0.18740224838256836,
      "learning_rate": 1e-06,
      "loss": -0.0322,
      "step": 1384
    },
    {
      "clip_ratio/high_max": 0.0022295348317129537,
      "clip_ratio/high_mean": 0.0007747080489934888,
      "clip_ratio/low_mean": 0.0006551564933943155,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014298645219241735,
      "epoch": 7.191309419655877,
      "grad_norm": 0.5735350251197815,
      "learning_rate": 1e-06,
      "loss": 0.0,
      "step": 1385
    },
    {
      "clip_ratio/high_max": 0.002230768157460261,
      "clip_ratio/high_mean": 0.0008987396358861588,
      "clip_ratio/low_mean": 0.0006252862276596716,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015240258289850317,
      "epoch": 7.195975503062117,
      "grad_norm": 0.17041385173797607,
      "learning_rate": 1e-06,
      "loss": -0.0216,
      "step": 1386
    },
    {
      "clip_ratio/high_max": 0.0024237356992671266,
      "clip_ratio/high_mean": 0.001005453639663756,
      "clip_ratio/low_mean": 0.0006922833144926699,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016977369741653092,
      "epoch": 7.200641586468358,
      "grad_norm": 0.30392420291900635,
      "learning_rate": 1e-06,
      "loss": -0.0452,
      "step": 1387
    },
    {
      "clip_ratio/high_max": 0.002598405546450522,
      "clip_ratio/high_mean": 0.0011434978987381328,
      "clip_ratio/low_mean": 0.0006337152381092892,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017772130813682452,
      "epoch": 7.205307669874599,
      "grad_norm": 0.16137278079986572,
      "learning_rate": 1e-06,
      "loss": -0.0726,
      "step": 1388
    },
    {
      "clip_ratio/high_max": 0.003019725263584405,
      "clip_ratio/high_mean": 0.0011345838793204166,
      "clip_ratio/low_mean": 0.0006780875637559802,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018126713839592412,
      "epoch": 7.20997375328084,
      "grad_norm": 0.18081513047218323,
      "learning_rate": 1e-06,
      "loss": -0.0831,
      "step": 1389
    },
    {
      "clip_ratio/high_max": 0.0027243441800237633,
      "clip_ratio/high_mean": 0.0011600549551076256,
      "clip_ratio/low_mean": 0.0008558843055652687,
      "clip_ratio/low_min": 1.6212710761465132e-05,
      "clip_ratio/region_mean": 0.0020159392443019897,
      "epoch": 7.2146398366870805,
      "grad_norm": 0.18026256561279297,
      "learning_rate": 1e-06,
      "loss": -0.0514,
      "step": 1390
    },
    {
      "clip_ratio/high_max": 0.002497679044608958,
      "clip_ratio/high_mean": 0.0010388221890025306,
      "clip_ratio/low_mean": 0.0007422344824590255,
      "clip_ratio/low_min": 2.137117371603381e-05,
      "clip_ratio/region_mean": 0.001781056635081768,
      "epoch": 7.219305920093322,
      "grad_norm": 0.18348857760429382,
      "learning_rate": 1e-06,
      "loss": -0.0227,
      "step": 1391
    },
    {
      "clip_ratio/high_max": 0.002736147085670382,
      "clip_ratio/high_mean": 0.0010727581902756356,
      "clip_ratio/low_mean": 0.000830135806609178,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019028940150747076,
      "epoch": 7.223972003499562,
      "grad_norm": 0.16022197902202606,
      "learning_rate": 1e-06,
      "loss": -0.0713,
      "step": 1392
    },
    {
      "clip_ratio/high_max": 0.005690870952093974,
      "clip_ratio/high_mean": 0.002291249758854974,
      "clip_ratio/low_mean": 0.0018303840442968067,
      "clip_ratio/low_min": 6.855576066300273e-05,
      "clip_ratio/region_mean": 0.004121633814065717,
      "epoch": 7.228638086905804,
      "grad_norm": 0.10240629315376282,
      "learning_rate": 1e-06,
      "loss": -0.032,
      "step": 1393
    },
    {
      "clip_ratio/high_max": 0.004066449255333282,
      "clip_ratio/high_mean": 0.0016682984787621535,
      "clip_ratio/low_mean": 0.0017844524718384491,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034527509487816133,
      "epoch": 7.233304170312044,
      "grad_norm": 0.11568844318389893,
      "learning_rate": 1e-06,
      "loss": -0.0125,
      "step": 1394
    },
    {
      "clip_ratio/high_max": 0.004259329340129625,
      "clip_ratio/high_mean": 0.0019790427177213132,
      "clip_ratio/low_mean": 0.001971847035747487,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003950889775296673,
      "epoch": 7.237970253718285,
      "grad_norm": 0.10941144078969955,
      "learning_rate": 1e-06,
      "loss": -0.0484,
      "step": 1395
    },
    {
      "clip_ratio/high_max": 0.005220945196924731,
      "clip_ratio/high_mean": 0.002107031708874274,
      "clip_ratio/low_mean": 0.001557528074044967,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036645597574533895,
      "epoch": 7.242636337124526,
      "grad_norm": 0.10521205514669418,
      "learning_rate": 1e-06,
      "loss": -0.0357,
      "step": 1396
    },
    {
      "clip_ratio/high_max": 0.005285412407829426,
      "clip_ratio/high_mean": 0.001960551064257743,
      "clip_ratio/low_mean": 0.0016429114912170917,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003603462566388771,
      "epoch": 7.247302420530767,
      "grad_norm": 0.2321300506591797,
      "learning_rate": 1e-06,
      "loss": -0.0766,
      "step": 1397
    },
    {
      "clip_ratio/high_max": 0.0041945805423893034,
      "clip_ratio/high_mean": 0.0018376325715507846,
      "clip_ratio/low_mean": 0.0019463643220660742,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003783996871788986,
      "epoch": 7.251968503937007,
      "grad_norm": 0.12059333920478821,
      "learning_rate": 1e-06,
      "loss": -0.018,
      "step": 1398
    },
    {
      "clip_ratio/high_max": 0.005382211311371066,
      "clip_ratio/high_mean": 0.0021551075151364785,
      "clip_ratio/low_mean": 0.001576917595230043,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003732025215867907,
      "epoch": 7.256634587343249,
      "grad_norm": 0.11205251514911652,
      "learning_rate": 1e-06,
      "loss": -0.0429,
      "step": 1399
    },
    {
      "clip_ratio/high_max": 0.0046652302698930725,
      "clip_ratio/high_mean": 0.0018036202418443281,
      "clip_ratio/low_mean": 0.001764358294167323,
      "clip_ratio/low_min": 6.834335363237187e-05,
      "clip_ratio/region_mean": 0.0035679785505635664,
      "epoch": 7.26130067074949,
      "grad_norm": 0.11788871884346008,
      "learning_rate": 1e-06,
      "loss": -0.033,
      "step": 1400
    },
    {
      "clip_ratio/high_max": 0.004217626075842418,
      "clip_ratio/high_mean": 0.0015111187949514715,
      "clip_ratio/low_mean": 0.0020126082345086616,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035237269403296523,
      "epoch": 7.2659667541557305,
      "grad_norm": 0.10562445968389511,
      "learning_rate": 1e-06,
      "loss": -0.0009,
      "step": 1401
    },
    {
      "clip_ratio/high_max": 0.0043130563863087445,
      "clip_ratio/high_mean": 0.0017501031688880175,
      "clip_ratio/low_mean": 0.0015789838580531068,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033290870196651667,
      "epoch": 7.270632837561972,
      "grad_norm": 0.10943048447370529,
      "learning_rate": 1e-06,
      "loss": -0.0223,
      "step": 1402
    },
    {
      "clip_ratio/high_max": 0.0045612145331688225,
      "clip_ratio/high_mean": 0.0020862354067503475,
      "clip_ratio/low_mean": 0.0016740255960030481,
      "clip_ratio/low_min": 6.851192301837727e-05,
      "clip_ratio/region_mean": 0.0037602610245812684,
      "epoch": 7.275298920968212,
      "grad_norm": 0.1013014167547226,
      "learning_rate": 1e-06,
      "loss": -0.046,
      "step": 1403
    },
    {
      "clip_ratio/high_max": 0.005731341414502822,
      "clip_ratio/high_mean": 0.002243163253297098,
      "clip_ratio/low_mean": 0.0015785994473844767,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038217626279219985,
      "epoch": 7.2799650043744535,
      "grad_norm": 0.10732930898666382,
      "learning_rate": 1e-06,
      "loss": -0.0734,
      "step": 1404
    },
    {
      "clip_ratio/high_max": 0.005860163291799836,
      "clip_ratio/high_mean": 0.0021694279712392017,
      "clip_ratio/low_mean": 0.0015291275412891991,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003698555432492867,
      "epoch": 7.284631087780694,
      "grad_norm": 0.1119331642985344,
      "learning_rate": 1e-06,
      "loss": -0.084,
      "step": 1405
    },
    {
      "clip_ratio/high_max": 0.00473516571946675,
      "clip_ratio/high_mean": 0.0020533799288386945,
      "clip_ratio/low_mean": 0.001929699592437828,
      "clip_ratio/low_min": 8.106355380732566e-05,
      "clip_ratio/region_mean": 0.003983079477620777,
      "epoch": 7.289297171186935,
      "grad_norm": 0.11101459711790085,
      "learning_rate": 1e-06,
      "loss": -0.0523,
      "step": 1406
    },
    {
      "clip_ratio/high_max": 0.004839467161218636,
      "clip_ratio/high_mean": 0.0018198908364865929,
      "clip_ratio/low_mean": 0.0017049133493856061,
      "clip_ratio/low_min": 5.105166565044783e-05,
      "clip_ratio/region_mean": 0.0035248041385784745,
      "epoch": 7.293963254593176,
      "grad_norm": 0.10066674649715424,
      "learning_rate": 1e-06,
      "loss": -0.0236,
      "step": 1407
    },
    {
      "clip_ratio/high_max": 0.004732530520414002,
      "clip_ratio/high_mean": 0.0018126526301784907,
      "clip_ratio/low_mean": 0.0016282599026453681,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00344091262013535,
      "epoch": 7.298629337999417,
      "grad_norm": 0.10134289413690567,
      "learning_rate": 1e-06,
      "loss": -0.072,
      "step": 1408
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0848214285714286,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4090.0,
      "completions/mean_length": 836.7695922851562,
      "completions/mean_terminated_length": 534.6945190429688,
      "completions/min_length": 95.0,
      "completions/min_terminated_length": 95.0,
      "epoch": 7.303295421405657,
      "grad_norm": 0.1717565357685089,
      "learning_rate": 1e-06,
      "loss": -0.0632,
      "num_tokens": 405258927.0,
      "reward": 0.6694336533546448,
      "reward_std": 0.15305370092391968,
      "rewards/simpleverify_reward/mean": 0.66943359375,
      "rewards/simpleverify_reward/std": 0.4704335033893585,
      "step": 1409
    },
    {
      "clip_ratio/high_max": 0.001910406703245826,
      "clip_ratio/high_mean": 0.0007401852672046516,
      "clip_ratio/low_mean": 0.00033430713938287226,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0010744924220489338,
      "epoch": 7.307961504811899,
      "grad_norm": 0.1613350659608841,
      "learning_rate": 1e-06,
      "loss": -0.0806,
      "step": 1410
    },
    {
      "clip_ratio/high_max": 0.002128610689396737,
      "clip_ratio/high_mean": 0.0007921959731902461,
      "clip_ratio/low_mean": 0.0004581749753924669,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012503709513111971,
      "epoch": 7.312627588218139,
      "grad_norm": 0.2004249095916748,
      "learning_rate": 1e-06,
      "loss": -0.0193,
      "step": 1411
    },
    {
      "clip_ratio/high_max": 0.002432449735351838,
      "clip_ratio/high_mean": 0.0008601702575106174,
      "clip_ratio/low_mean": 0.0005078526764918934,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013680229094461538,
      "epoch": 7.31729367162438,
      "grad_norm": 0.20410509407520294,
      "learning_rate": 1e-06,
      "loss": -0.0328,
      "step": 1412
    },
    {
      "clip_ratio/high_max": 0.0021632494463119656,
      "clip_ratio/high_mean": 0.000842975479827146,
      "clip_ratio/low_mean": 0.0005120429727867304,
      "clip_ratio/low_min": 1.560549389978405e-05,
      "clip_ratio/region_mean": 0.0013550184412451927,
      "epoch": 7.321959755030621,
      "grad_norm": 0.1816713511943817,
      "learning_rate": 1e-06,
      "loss": -0.0283,
      "step": 1413
    },
    {
      "clip_ratio/high_max": 0.002122090674674837,
      "clip_ratio/high_mean": 0.0007969508715177653,
      "clip_ratio/low_mean": 0.0005472665761772078,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001344217460427899,
      "epoch": 7.326625838436862,
      "grad_norm": 0.18468163907527924,
      "learning_rate": 1e-06,
      "loss": -0.0135,
      "step": 1414
    },
    {
      "clip_ratio/high_max": 0.0022095729073043913,
      "clip_ratio/high_mean": 0.0009760061520864838,
      "clip_ratio/low_mean": 0.000473567757580895,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001449573908757884,
      "epoch": 7.331291921843103,
      "grad_norm": 0.17401708662509918,
      "learning_rate": 1e-06,
      "loss": -0.0524,
      "step": 1415
    },
    {
      "clip_ratio/high_max": 0.0019754380937229143,
      "clip_ratio/high_mean": 0.0008297085032609175,
      "clip_ratio/low_mean": 0.0005366010082070716,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013663095305673778,
      "epoch": 7.335958005249344,
      "grad_norm": 0.22758713364601135,
      "learning_rate": 1e-06,
      "loss": -0.034,
      "step": 1416
    },
    {
      "clip_ratio/high_max": 0.0021523628747672774,
      "clip_ratio/high_mean": 0.0008522993321093963,
      "clip_ratio/low_mean": 0.0006713470738759497,
      "clip_ratio/low_min": 3.659964932012372e-05,
      "clip_ratio/region_mean": 0.0015236463805194944,
      "epoch": 7.340624088655585,
      "grad_norm": 0.16731558740139008,
      "learning_rate": 1e-06,
      "loss": -0.0573,
      "step": 1417
    },
    {
      "clip_ratio/high_max": 0.00222210639185505,
      "clip_ratio/high_mean": 0.0009079246665351093,
      "clip_ratio/low_mean": 0.0005921211709392082,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001500045840657549,
      "epoch": 7.3452901720618256,
      "grad_norm": 0.1762041300535202,
      "learning_rate": 1e-06,
      "loss": -0.0157,
      "step": 1418
    },
    {
      "clip_ratio/high_max": 0.0024547977081965655,
      "clip_ratio/high_mean": 0.0010494962698430754,
      "clip_ratio/low_mean": 0.0007614346050104359,
      "clip_ratio/low_min": 1.907814475998748e-05,
      "clip_ratio/region_mean": 0.001810930851206649,
      "epoch": 7.349956255468067,
      "grad_norm": 0.17196029424667358,
      "learning_rate": 1e-06,
      "loss": -0.0291,
      "step": 1419
    },
    {
      "clip_ratio/high_max": 0.0025236308138119057,
      "clip_ratio/high_mean": 0.0009809689472604077,
      "clip_ratio/low_mean": 0.0007225120289149345,
      "clip_ratio/low_min": 3.866378028760664e-05,
      "clip_ratio/region_mean": 0.001703480946162017,
      "epoch": 7.354622338874307,
      "grad_norm": 0.17653904855251312,
      "learning_rate": 1e-06,
      "loss": -0.0409,
      "step": 1420
    },
    {
      "clip_ratio/high_max": 0.0026429281570017338,
      "clip_ratio/high_mean": 0.001069426047251909,
      "clip_ratio/low_mean": 0.0007676737559449975,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00183709980774438,
      "epoch": 7.359288422280549,
      "grad_norm": 0.18160517513751984,
      "learning_rate": 1e-06,
      "loss": -0.0753,
      "step": 1421
    },
    {
      "clip_ratio/high_max": 0.002748325336142443,
      "clip_ratio/high_mean": 0.0011163898634549696,
      "clip_ratio/low_mean": 0.0008297794315694773,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019461693009361625,
      "epoch": 7.363954505686789,
      "grad_norm": 0.13417622447013855,
      "learning_rate": 1e-06,
      "loss": -0.0394,
      "step": 1422
    },
    {
      "clip_ratio/high_max": 0.002782740990369348,
      "clip_ratio/high_mean": 0.0010608749453240307,
      "clip_ratio/low_mean": 0.0007423377719533164,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018032126827165484,
      "epoch": 7.36862058909303,
      "grad_norm": 0.19278652966022491,
      "learning_rate": 1e-06,
      "loss": -0.0573,
      "step": 1423
    },
    {
      "clip_ratio/high_max": 0.0024179421307053417,
      "clip_ratio/high_mean": 0.0009885039999062428,
      "clip_ratio/low_mean": 0.0007979971933309571,
      "clip_ratio/low_min": 2.891510484914761e-05,
      "clip_ratio/region_mean": 0.001786501205060631,
      "epoch": 7.373286672499271,
      "grad_norm": 0.1465693861246109,
      "learning_rate": 1e-06,
      "loss": -0.0383,
      "step": 1424
    },
    {
      "clip_ratio/high_max": 0.005212131407461129,
      "clip_ratio/high_mean": 0.002197588426497532,
      "clip_ratio/low_mean": 0.0015867582442297135,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037843466852791607,
      "epoch": 7.377952755905512,
      "grad_norm": 0.09423007071018219,
      "learning_rate": 1e-06,
      "loss": -0.064,
      "step": 1425
    },
    {
      "clip_ratio/high_max": 0.004595384525600821,
      "clip_ratio/high_mean": 0.0018918486566690262,
      "clip_ratio/low_mean": 0.0012599759438671754,
      "clip_ratio/low_min": 1.4361213288793806e-05,
      "clip_ratio/region_mean": 0.003151824545057025,
      "epoch": 7.3826188393117524,
      "grad_norm": 0.09894728660583496,
      "learning_rate": 1e-06,
      "loss": -0.0813,
      "step": 1426
    },
    {
      "clip_ratio/high_max": 0.0055534994608024135,
      "clip_ratio/high_mean": 0.0020802060316782445,
      "clip_ratio/low_mean": 0.0017840438267739955,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038642498911940493,
      "epoch": 7.387284922717994,
      "grad_norm": 0.10407675802707672,
      "learning_rate": 1e-06,
      "loss": -0.0202,
      "step": 1427
    },
    {
      "clip_ratio/high_max": 0.0059713546797866,
      "clip_ratio/high_mean": 0.0021649666523444466,
      "clip_ratio/low_mean": 0.0020009116851724684,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0041658783302409574,
      "epoch": 7.391951006124234,
      "grad_norm": 0.12078286707401276,
      "learning_rate": 1e-06,
      "loss": -0.0338,
      "step": 1428
    },
    {
      "clip_ratio/high_max": 0.004623753986379597,
      "clip_ratio/high_mean": 0.001861581979028415,
      "clip_ratio/low_mean": 0.0018612840467540082,
      "clip_ratio/low_min": 3.12109877995681e-05,
      "clip_ratio/region_mean": 0.0037228660221444443,
      "epoch": 7.3966170895304755,
      "grad_norm": 0.11224246770143509,
      "learning_rate": 1e-06,
      "loss": -0.0292,
      "step": 1429
    },
    {
      "clip_ratio/high_max": 0.005018732277676463,
      "clip_ratio/high_mean": 0.002022135609877296,
      "clip_ratio/low_mean": 0.0019154238689225167,
      "clip_ratio/low_min": 6.55411058687605e-05,
      "clip_ratio/region_mean": 0.003937559537007473,
      "epoch": 7.401283172936716,
      "grad_norm": 0.11177490651607513,
      "learning_rate": 1e-06,
      "loss": -0.0145,
      "step": 1430
    },
    {
      "clip_ratio/high_max": 0.0052157543250359595,
      "clip_ratio/high_mean": 0.0021048767048341688,
      "clip_ratio/low_mean": 0.0016186998545890674,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037235765485092998,
      "epoch": 7.405949256342957,
      "grad_norm": 0.1137535572052002,
      "learning_rate": 1e-06,
      "loss": -0.0533,
      "step": 1431
    },
    {
      "clip_ratio/high_max": 0.005313912908604834,
      "clip_ratio/high_mean": 0.002002594046643935,
      "clip_ratio/low_mean": 0.0016767497618275229,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003679343812109437,
      "epoch": 7.410615339749198,
      "grad_norm": 0.11435745656490326,
      "learning_rate": 1e-06,
      "loss": -0.035,
      "step": 1432
    },
    {
      "clip_ratio/high_max": 0.005562998943787534,
      "clip_ratio/high_mean": 0.0019858788618876133,
      "clip_ratio/low_mean": 0.0016467466184622026,
      "clip_ratio/low_min": 7.319929864024743e-05,
      "clip_ratio/region_mean": 0.003632625564932823,
      "epoch": 7.415281423155439,
      "grad_norm": 0.11291353404521942,
      "learning_rate": 1e-06,
      "loss": -0.0581,
      "step": 1433
    },
    {
      "clip_ratio/high_max": 0.004402678270707838,
      "clip_ratio/high_mean": 0.0017808313277782872,
      "clip_ratio/low_mean": 0.0016999414292513393,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003480772633338347,
      "epoch": 7.41994750656168,
      "grad_norm": 0.1485971361398697,
      "learning_rate": 1e-06,
      "loss": -0.0165,
      "step": 1434
    },
    {
      "clip_ratio/high_max": 0.004717145566246472,
      "clip_ratio/high_mean": 0.0019051778508583084,
      "clip_ratio/low_mean": 0.0019269497170171235,
      "clip_ratio/low_min": 5.344925739336759e-05,
      "clip_ratio/region_mean": 0.003832127636997029,
      "epoch": 7.424613589967921,
      "grad_norm": 0.1016089916229248,
      "learning_rate": 1e-06,
      "loss": -0.0299,
      "step": 1435
    },
    {
      "clip_ratio/high_max": 0.005057233145635109,
      "clip_ratio/high_mean": 0.0019317501974001061,
      "clip_ratio/low_mean": 0.0019208869780413806,
      "clip_ratio/low_min": 1.5085686754900962e-05,
      "clip_ratio/region_mean": 0.0038526371645275503,
      "epoch": 7.429279673374162,
      "grad_norm": 0.1129857674241066,
      "learning_rate": 1e-06,
      "loss": -0.042,
      "step": 1436
    },
    {
      "clip_ratio/high_max": 0.005249914574960712,
      "clip_ratio/high_mean": 0.002101309542922536,
      "clip_ratio/low_mean": 0.0016488298042531824,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037501393526326865,
      "epoch": 7.433945756780402,
      "grad_norm": 0.1125134825706482,
      "learning_rate": 1e-06,
      "loss": -0.0762,
      "step": 1437
    },
    {
      "clip_ratio/high_max": 0.004501464907662012,
      "clip_ratio/high_mean": 0.0018051889055641368,
      "clip_ratio/low_mean": 0.0016548458297620527,
      "clip_ratio/low_min": 0.00010890343401115388,
      "clip_ratio/region_mean": 0.0034600348444655538,
      "epoch": 7.438611840186644,
      "grad_norm": 0.10212256759405136,
      "learning_rate": 1e-06,
      "loss": -0.04,
      "step": 1438
    },
    {
      "clip_ratio/high_max": 0.0056072988954838365,
      "clip_ratio/high_mean": 0.0020926353899994865,
      "clip_ratio/low_mean": 0.0017589937488082796,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003851629138807766,
      "epoch": 7.443277923592884,
      "grad_norm": 0.10990682244300842,
      "learning_rate": 1e-06,
      "loss": -0.0582,
      "step": 1439
    },
    {
      "clip_ratio/high_max": 0.0038470451763714664,
      "clip_ratio/high_mean": 0.001627318106329767,
      "clip_ratio/low_mean": 0.0016522667692697723,
      "clip_ratio/low_min": 5.818012505187653e-05,
      "clip_ratio/region_mean": 0.0032795848819660023,
      "epoch": 7.447944006999125,
      "grad_norm": 0.10437417775392532,
      "learning_rate": 1e-06,
      "loss": -0.039,
      "step": 1440
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0936802455357143,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4040.0,
      "completions/mean_length": 889.7994995117188,
      "completions/mean_terminated_length": 558.3958129882812,
      "completions/min_length": 5.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 7.452610090405366,
      "grad_norm": 0.19270630180835724,
      "learning_rate": 1e-06,
      "loss": -0.066,
      "num_tokens": 413934732.0,
      "reward": 0.65869140625,
      "reward_std": 0.1502629965543747,
      "rewards/simpleverify_reward/mean": 0.65869140625,
      "rewards/simpleverify_reward/std": 0.47416529059410095,
      "step": 1441
    },
    {
      "clip_ratio/high_max": 0.0018305212288396433,
      "clip_ratio/high_mean": 0.0007223765041999286,
      "clip_ratio/low_mean": 0.00042316487360949395,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001145541391451843,
      "epoch": 7.457276173811607,
      "grad_norm": 0.15830476582050323,
      "learning_rate": 1e-06,
      "loss": -0.0616,
      "step": 1442
    },
    {
      "clip_ratio/high_max": 0.0017543510439281818,
      "clip_ratio/high_mean": 0.0007199926803878043,
      "clip_ratio/low_mean": 0.00048615474452162744,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012061474117217585,
      "epoch": 7.4619422572178475,
      "grad_norm": 0.15458270907402039,
      "learning_rate": 1e-06,
      "loss": -0.0129,
      "step": 1443
    },
    {
      "clip_ratio/high_max": 0.002132483608875191,
      "clip_ratio/high_mean": 0.000872970713317045,
      "clip_ratio/low_mean": 0.00045244008833833504,
      "clip_ratio/low_min": 3.901373202097602e-05,
      "clip_ratio/region_mean": 0.0013254108162072953,
      "epoch": 7.466608340624089,
      "grad_norm": 0.25120580196380615,
      "learning_rate": 1e-06,
      "loss": -0.0481,
      "step": 1444
    },
    {
      "clip_ratio/high_max": 0.0016460290462418925,
      "clip_ratio/high_mean": 0.0007130425437935628,
      "clip_ratio/low_mean": 0.0005152259427632089,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012282684638194041,
      "epoch": 7.471274424030329,
      "grad_norm": 0.15117956697940826,
      "learning_rate": 1e-06,
      "loss": -0.0018,
      "step": 1445
    },
    {
      "clip_ratio/high_max": 0.002032519470958505,
      "clip_ratio/high_mean": 0.0008346390459337272,
      "clip_ratio/low_mean": 0.00047760332381585613,
      "clip_ratio/low_min": 1.4484357052424457e-05,
      "clip_ratio/region_mean": 0.0013122423624736257,
      "epoch": 7.475940507436571,
      "grad_norm": 0.16944393515586853,
      "learning_rate": 1e-06,
      "loss": -0.0183,
      "step": 1446
    },
    {
      "clip_ratio/high_max": 0.00216623317101039,
      "clip_ratio/high_mean": 0.0008510964617016725,
      "clip_ratio/low_mean": 0.0006067982267268235,
      "clip_ratio/low_min": 2.5863851988106035e-05,
      "clip_ratio/region_mean": 0.0014578947011614218,
      "epoch": 7.480606590842811,
      "grad_norm": 0.2070213109254837,
      "learning_rate": 1e-06,
      "loss": -0.0629,
      "step": 1447
    },
    {
      "clip_ratio/high_max": 0.0022003844460414257,
      "clip_ratio/high_mean": 0.0008437011820205953,
      "clip_ratio/low_mean": 0.000610818784480216,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014545199446729384,
      "epoch": 7.485272674249052,
      "grad_norm": 0.16186128556728363,
      "learning_rate": 1e-06,
      "loss": -0.0606,
      "step": 1448
    },
    {
      "clip_ratio/high_max": 0.0022420187924581114,
      "clip_ratio/high_mean": 0.0008143928571371362,
      "clip_ratio/low_mean": 0.0005092860565127921,
      "clip_ratio/low_min": 1.367016648146091e-05,
      "clip_ratio/region_mean": 0.0013236789345683064,
      "epoch": 7.489938757655293,
      "grad_norm": 0.18141378462314606,
      "learning_rate": 1e-06,
      "loss": -0.0621,
      "step": 1449
    },
    {
      "clip_ratio/high_max": 0.002240421952592442,
      "clip_ratio/high_mean": 0.0008935572950576898,
      "clip_ratio/low_mean": 0.0006871131508887629,
      "clip_ratio/low_min": 3.1871495593804866e-05,
      "clip_ratio/region_mean": 0.0015806704614078626,
      "epoch": 7.494604841061534,
      "grad_norm": 0.3602309226989746,
      "learning_rate": 1e-06,
      "loss": -0.0056,
      "step": 1450
    },
    {
      "clip_ratio/high_max": 0.0023575757222715765,
      "clip_ratio/high_mean": 0.0008182661877071951,
      "clip_ratio/low_mean": 0.0006290435480877932,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014473097580776084,
      "epoch": 7.499270924467774,
      "grad_norm": 0.17450061440467834,
      "learning_rate": 1e-06,
      "loss": -0.0214,
      "step": 1451
    },
    {
      "clip_ratio/high_max": 0.002445107926178025,
      "clip_ratio/high_mean": 0.0009358380939374911,
      "clip_ratio/low_mean": 0.000705860047673923,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001641698177991202,
      "epoch": 7.503937007874016,
      "grad_norm": 0.15424740314483643,
      "learning_rate": 1e-06,
      "loss": -0.0787,
      "step": 1452
    },
    {
      "clip_ratio/high_max": 0.00229677321476629,
      "clip_ratio/high_mean": 0.0009761892433743924,
      "clip_ratio/low_mean": 0.0005903874534851639,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015665766877646092,
      "epoch": 7.508603091280257,
      "grad_norm": 0.15647411346435547,
      "learning_rate": 1e-06,
      "loss": -0.0829,
      "step": 1453
    },
    {
      "clip_ratio/high_max": 0.0025270449550589547,
      "clip_ratio/high_mean": 0.000984163023531437,
      "clip_ratio/low_mean": 0.0006354441438816139,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00161960718469345,
      "epoch": 7.5132691746864975,
      "grad_norm": 0.7884874939918518,
      "learning_rate": 1e-06,
      "loss": -0.0567,
      "step": 1454
    },
    {
      "clip_ratio/high_max": 0.0021180681869736873,
      "clip_ratio/high_mean": 0.0008684027852723375,
      "clip_ratio/low_mean": 0.0008248721624113386,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0016932749676925596,
      "epoch": 7.517935258092739,
      "grad_norm": 0.18797625601291656,
      "learning_rate": 1e-06,
      "loss": -0.0367,
      "step": 1455
    },
    {
      "clip_ratio/high_max": 0.0024011031237023417,
      "clip_ratio/high_mean": 0.001046918996507884,
      "clip_ratio/low_mean": 0.0007792649557814002,
      "clip_ratio/low_min": 5.631966450891923e-05,
      "clip_ratio/region_mean": 0.0018261839577462524,
      "epoch": 7.522601341498979,
      "grad_norm": 0.183904767036438,
      "learning_rate": 1e-06,
      "loss": -0.0603,
      "step": 1456
    },
    {
      "clip_ratio/high_max": 0.005467097056680359,
      "clip_ratio/high_mean": 0.0022446903967647813,
      "clip_ratio/low_mean": 0.0017680774253676645,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00401276774937287,
      "epoch": 7.5272674249052205,
      "grad_norm": 0.10196895897388458,
      "learning_rate": 1e-06,
      "loss": -0.0669,
      "step": 1457
    },
    {
      "clip_ratio/high_max": 0.004395519688841887,
      "clip_ratio/high_mean": 0.0017328943613392767,
      "clip_ratio/low_mean": 0.0015983417761162855,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033312361483694986,
      "epoch": 7.531933508311461,
      "grad_norm": 0.11472579091787338,
      "learning_rate": 1e-06,
      "loss": -0.0623,
      "step": 1458
    },
    {
      "clip_ratio/high_max": 0.003650425722298678,
      "clip_ratio/high_mean": 0.0017068459255824564,
      "clip_ratio/low_mean": 0.001875266996648861,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035821128549287096,
      "epoch": 7.536599591717702,
      "grad_norm": 0.1033015325665474,
      "learning_rate": 1e-06,
      "loss": -0.0137,
      "step": 1459
    },
    {
      "clip_ratio/high_max": 0.0046282951516332105,
      "clip_ratio/high_mean": 0.0018897248010034673,
      "clip_ratio/low_mean": 0.001587980423209956,
      "clip_ratio/low_min": 9.103204502025619e-05,
      "clip_ratio/region_mean": 0.0034777051987475716,
      "epoch": 7.541265675123943,
      "grad_norm": 0.11019978672266006,
      "learning_rate": 1e-06,
      "loss": -0.0488,
      "step": 1460
    },
    {
      "clip_ratio/high_max": 0.0038333841803250834,
      "clip_ratio/high_mean": 0.0015590045550197829,
      "clip_ratio/low_mean": 0.0017600630017113872,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033190676476806402,
      "epoch": 7.545931758530184,
      "grad_norm": 0.10194586962461472,
      "learning_rate": 1e-06,
      "loss": -0.0025,
      "step": 1461
    },
    {
      "clip_ratio/high_max": 0.004471045824175235,
      "clip_ratio/high_mean": 0.001729344734485494,
      "clip_ratio/low_mean": 0.0015799501961737406,
      "clip_ratio/low_min": 8.690614049555734e-05,
      "clip_ratio/region_mean": 0.0033092949452111498,
      "epoch": 7.550597841936424,
      "grad_norm": 0.10734774172306061,
      "learning_rate": 1e-06,
      "loss": -0.0191,
      "step": 1462
    },
    {
      "clip_ratio/high_max": 0.00559540864196606,
      "clip_ratio/high_mean": 0.0020423575733730104,
      "clip_ratio/low_mean": 0.001704288773908047,
      "clip_ratio/low_min": 0.00011638733849395066,
      "clip_ratio/region_mean": 0.0037466463109012693,
      "epoch": 7.555263925342666,
      "grad_norm": 0.11830799281597137,
      "learning_rate": 1e-06,
      "loss": -0.0637,
      "step": 1463
    },
    {
      "clip_ratio/high_max": 0.004802870040293783,
      "clip_ratio/high_mean": 0.0018571846376289614,
      "clip_ratio/low_mean": 0.0015380685372292646,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003395253195776604,
      "epoch": 7.559930008748906,
      "grad_norm": 0.10258246213197708,
      "learning_rate": 1e-06,
      "loss": -0.0612,
      "step": 1464
    },
    {
      "clip_ratio/high_max": 0.004036631864437368,
      "clip_ratio/high_mean": 0.0016082869442470837,
      "clip_ratio/low_mean": 0.0015946135245030746,
      "clip_ratio/low_min": 5.468066592584364e-05,
      "clip_ratio/region_mean": 0.0032029004796640947,
      "epoch": 7.564596092155147,
      "grad_norm": 0.10809513926506042,
      "learning_rate": 1e-06,
      "loss": -0.0628,
      "step": 1465
    },
    {
      "clip_ratio/high_max": 0.004903237408143468,
      "clip_ratio/high_mean": 0.0018993035228049848,
      "clip_ratio/low_mean": 0.0019947527835029177,
      "clip_ratio/low_min": 0.00018593157255963888,
      "clip_ratio/region_mean": 0.0038940563244977966,
      "epoch": 7.569262175561388,
      "grad_norm": 0.135222390294075,
      "learning_rate": 1e-06,
      "loss": -0.0065,
      "step": 1466
    },
    {
      "clip_ratio/high_max": 0.004870341552305035,
      "clip_ratio/high_mean": 0.0017010760784614831,
      "clip_ratio/low_mean": 0.001801039088604739,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003502115185256116,
      "epoch": 7.573928258967629,
      "grad_norm": 0.10393046587705612,
      "learning_rate": 1e-06,
      "loss": -0.0222,
      "step": 1467
    },
    {
      "clip_ratio/high_max": 0.00513146941375453,
      "clip_ratio/high_mean": 0.0019443690944171976,
      "clip_ratio/low_mean": 0.0016804766783025116,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003624845776357688,
      "epoch": 7.57859434237387,
      "grad_norm": 0.11001542210578918,
      "learning_rate": 1e-06,
      "loss": -0.0794,
      "step": 1468
    },
    {
      "clip_ratio/high_max": 0.004238697511027567,
      "clip_ratio/high_mean": 0.0018393221835140139,
      "clip_ratio/low_mean": 0.0015272068740159739,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033665290102362633,
      "epoch": 7.583260425780111,
      "grad_norm": 0.11938067525625229,
      "learning_rate": 1e-06,
      "loss": -0.0836,
      "step": 1469
    },
    {
      "clip_ratio/high_max": 0.004776272995513864,
      "clip_ratio/high_mean": 0.001788714580470696,
      "clip_ratio/low_mean": 0.0019292604447400663,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037179750652285293,
      "epoch": 7.587926509186351,
      "grad_norm": 0.11862795799970627,
      "learning_rate": 1e-06,
      "loss": -0.0576,
      "step": 1470
    },
    {
      "clip_ratio/high_max": 0.00465654103027191,
      "clip_ratio/high_mean": 0.0017225914998562075,
      "clip_ratio/low_mean": 0.0018139689054805785,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0035365603398531675,
      "epoch": 7.592592592592593,
      "grad_norm": 0.1361895054578781,
      "learning_rate": 1e-06,
      "loss": -0.0374,
      "step": 1471
    },
    {
      "clip_ratio/high_max": 0.004124632258026395,
      "clip_ratio/high_mean": 0.0018400444132566918,
      "clip_ratio/low_mean": 0.0017699192176223733,
      "clip_ratio/low_min": 0.00014766353706363589,
      "clip_ratio/region_mean": 0.0036099636199651286,
      "epoch": 7.597258675998834,
      "grad_norm": 0.11981182545423508,
      "learning_rate": 1e-06,
      "loss": -0.0611,
      "step": 1472
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0891462053571429,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4012.0,
      "completions/mean_length": 854.3715209960938,
      "completions/mean_terminated_length": 537.1099853515625,
      "completions/min_length": 78.0,
      "completions/min_terminated_length": 78.0,
      "epoch": 7.601924759405074,
      "grad_norm": 0.20368652045726776,
      "learning_rate": 1e-06,
      "loss": -0.0432,
      "num_tokens": 422382450.0,
      "reward": 0.6672014594078064,
      "reward_std": 0.14123475551605225,
      "rewards/simpleverify_reward/mean": 0.6672014594078064,
      "rewards/simpleverify_reward/std": 0.4712315499782562,
      "step": 1473
    },
    {
      "clip_ratio/high_max": 0.0019321077816130128,
      "clip_ratio/high_mean": 0.0007545481576016755,
      "clip_ratio/low_mean": 0.0003618924652073474,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0011164406496391166,
      "epoch": 7.606590842811316,
      "grad_norm": 0.162889301776886,
      "learning_rate": 1e-06,
      "loss": -0.0523,
      "step": 1474
    },
    {
      "clip_ratio/high_max": 0.00210403493110789,
      "clip_ratio/high_mean": 0.0008370339382963721,
      "clip_ratio/low_mean": 0.00038616514302702853,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012231991240696516,
      "epoch": 7.611256926217556,
      "grad_norm": 0.1682155430316925,
      "learning_rate": 1e-06,
      "loss": -0.0286,
      "step": 1475
    },
    {
      "clip_ratio/high_max": 0.002167194747016765,
      "clip_ratio/high_mean": 0.0008144522307702573,
      "clip_ratio/low_mean": 0.0004282597008113953,
      "clip_ratio/low_min": 1.713267556624487e-05,
      "clip_ratio/region_mean": 0.001242711914528627,
      "epoch": 7.615923009623797,
      "grad_norm": 0.1678251177072525,
      "learning_rate": 1e-06,
      "loss": -0.0633,
      "step": 1476
    },
    {
      "clip_ratio/high_max": 0.0023542965172964614,
      "clip_ratio/high_mean": 0.0008686422679602401,
      "clip_ratio/low_mean": 0.00048729862601248897,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013559409089793917,
      "epoch": 7.620589093030038,
      "grad_norm": 0.20742952823638916,
      "learning_rate": 1e-06,
      "loss": -0.0664,
      "step": 1477
    },
    {
      "clip_ratio/high_max": 0.0020516479780781083,
      "clip_ratio/high_mean": 0.0009298053355450975,
      "clip_ratio/low_mean": 0.00047200236349453917,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014018077126820572,
      "epoch": 7.625255176436279,
      "grad_norm": 0.168976292014122,
      "learning_rate": 1e-06,
      "loss": -0.0611,
      "step": 1478
    },
    {
      "clip_ratio/high_max": 0.001802879025490256,
      "clip_ratio/high_mean": 0.0006674254759673204,
      "clip_ratio/low_mean": 0.0005177582534088288,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00118518371891696,
      "epoch": 7.6299212598425195,
      "grad_norm": 0.1982998251914978,
      "learning_rate": 1e-06,
      "loss": -0.0097,
      "step": 1479
    },
    {
      "clip_ratio/high_max": 0.002118727847118862,
      "clip_ratio/high_mean": 0.000792240407463396,
      "clip_ratio/low_mean": 0.0004534460072136426,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012456864060368389,
      "epoch": 7.634587343248761,
      "grad_norm": 0.14808295667171478,
      "learning_rate": 1e-06,
      "loss": -0.0327,
      "step": 1480
    },
    {
      "clip_ratio/high_max": 0.002626515328302048,
      "clip_ratio/high_mean": 0.0008949331077019451,
      "clip_ratio/low_mean": 0.0005744934651374933,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014694265883008484,
      "epoch": 7.639253426655001,
      "grad_norm": 0.18225401639938354,
      "learning_rate": 1e-06,
      "loss": -0.0274,
      "step": 1481
    },
    {
      "clip_ratio/high_max": 0.0023778744362061843,
      "clip_ratio/high_mean": 0.0009791627198865172,
      "clip_ratio/low_mean": 0.000501624456319405,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014807871957600582,
      "epoch": 7.6439195100612425,
      "grad_norm": 0.16968855261802673,
      "learning_rate": 1e-06,
      "loss": -0.0572,
      "step": 1482
    },
    {
      "clip_ratio/high_max": 0.0021586116636171937,
      "clip_ratio/high_mean": 0.0008928717506933026,
      "clip_ratio/low_mean": 0.000576390482365241,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014692622353322804,
      "epoch": 7.648585593467483,
      "grad_norm": 0.18596337735652924,
      "learning_rate": 1e-06,
      "loss": -0.0606,
      "step": 1483
    },
    {
      "clip_ratio/high_max": 0.002512024228053633,
      "clip_ratio/high_mean": 0.0009815312405407894,
      "clip_ratio/low_mean": 0.0006831329665146768,
      "clip_ratio/low_min": 3.410641147638671e-05,
      "clip_ratio/region_mean": 0.0016646642143314239,
      "epoch": 7.653251676873724,
      "grad_norm": 0.21784955263137817,
      "learning_rate": 1e-06,
      "loss": -0.0123,
      "step": 1484
    },
    {
      "clip_ratio/high_max": 0.002348741385503672,
      "clip_ratio/high_mean": 0.0008950869196269196,
      "clip_ratio/low_mean": 0.0006804961440138868,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0015755830681882799,
      "epoch": 7.657917760279965,
      "grad_norm": 0.2041342705488205,
      "learning_rate": 1e-06,
      "loss": -0.0552,
      "step": 1485
    },
    {
      "clip_ratio/high_max": 0.002357837547606323,
      "clip_ratio/high_mean": 0.0009750519384397194,
      "clip_ratio/low_mean": 0.0007263015977514442,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001701353583484888,
      "epoch": 7.662583843686206,
      "grad_norm": 0.2158804088830948,
      "learning_rate": 1e-06,
      "loss": -0.0309,
      "step": 1486
    },
    {
      "clip_ratio/high_max": 0.003039932722458616,
      "clip_ratio/high_mean": 0.0010895789946516743,
      "clip_ratio/low_mean": 0.0007536582807006198,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018432372962706722,
      "epoch": 7.667249927092447,
      "grad_norm": 0.18324705958366394,
      "learning_rate": 1e-06,
      "loss": -0.0167,
      "step": 1487
    },
    {
      "clip_ratio/high_max": 0.002248109351057792,
      "clip_ratio/high_mean": 0.0009024063429023954,
      "clip_ratio/low_mean": 0.0008267970879387576,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017292034681304358,
      "epoch": 7.671916010498688,
      "grad_norm": 0.2000238001346588,
      "learning_rate": 1e-06,
      "loss": -0.0186,
      "step": 1488
    },
    {
      "clip_ratio/high_max": 0.005637190231936984,
      "clip_ratio/high_mean": 0.0023293374470085837,
      "clip_ratio/low_mean": 0.0019038573627767619,
      "clip_ratio/low_min": 5.559509008890018e-05,
      "clip_ratio/region_mean": 0.004233194937114604,
      "epoch": 7.676582093904928,
      "grad_norm": 0.09963242709636688,
      "learning_rate": 1e-06,
      "loss": -0.0441,
      "step": 1489
    },
    {
      "clip_ratio/high_max": 0.00421796010050457,
      "clip_ratio/high_mean": 0.0016571489177295007,
      "clip_ratio/low_mean": 0.0014588923477276694,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0031160412763711065,
      "epoch": 7.681248177311169,
      "grad_norm": 0.10052011162042618,
      "learning_rate": 1e-06,
      "loss": -0.053,
      "step": 1490
    },
    {
      "clip_ratio/high_max": 0.0038538493317901157,
      "clip_ratio/high_mean": 0.0017124608166341204,
      "clip_ratio/low_mean": 0.001361603297482361,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0030740640722797252,
      "epoch": 7.685914260717411,
      "grad_norm": 0.10273779928684235,
      "learning_rate": 1e-06,
      "loss": -0.0292,
      "step": 1491
    },
    {
      "clip_ratio/high_max": 0.004771810876263771,
      "clip_ratio/high_mean": 0.0019117758638458326,
      "clip_ratio/low_mean": 0.0014474881827482022,
      "clip_ratio/low_min": 6.952168769203126e-05,
      "clip_ratio/region_mean": 0.003359264104801696,
      "epoch": 7.690580344123651,
      "grad_norm": 0.10847639292478561,
      "learning_rate": 1e-06,
      "loss": -0.0641,
      "step": 1492
    },
    {
      "clip_ratio/high_max": 0.00603368591691833,
      "clip_ratio/high_mean": 0.0020579002193699125,
      "clip_ratio/low_mean": 0.0018534946320869494,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003911394829628989,
      "epoch": 7.695246427529892,
      "grad_norm": 0.11956518888473511,
      "learning_rate": 1e-06,
      "loss": -0.0675,
      "step": 1493
    },
    {
      "clip_ratio/high_max": 0.003999712091172114,
      "clip_ratio/high_mean": 0.0017442529133404605,
      "clip_ratio/low_mean": 0.0016306047491525533,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0033748576388461515,
      "epoch": 7.699912510936133,
      "grad_norm": 0.10121136903762817,
      "learning_rate": 1e-06,
      "loss": -0.0619,
      "step": 1494
    },
    {
      "clip_ratio/high_max": 0.004087908600922674,
      "clip_ratio/high_mean": 0.0013940994795120787,
      "clip_ratio/low_mean": 0.001587731054314645,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0029818304828950204,
      "epoch": 7.704578594342374,
      "grad_norm": 0.09256784617900848,
      "learning_rate": 1e-06,
      "loss": -0.0104,
      "step": 1495
    },
    {
      "clip_ratio/high_max": 0.00432162905781297,
      "clip_ratio/high_mean": 0.001621355659153778,
      "clip_ratio/low_mean": 0.001472005038522184,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0030933607486076653,
      "epoch": 7.7092446777486145,
      "grad_norm": 0.09782762080430984,
      "learning_rate": 1e-06,
      "loss": -0.0333,
      "step": 1496
    },
    {
      "clip_ratio/high_max": 0.004736357841466088,
      "clip_ratio/high_mean": 0.0017792898579500616,
      "clip_ratio/low_mean": 0.0016899218298931373,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0034692116896621883,
      "epoch": 7.713910761154856,
      "grad_norm": 0.10189592093229294,
      "learning_rate": 1e-06,
      "loss": -0.0282,
      "step": 1497
    },
    {
      "clip_ratio/high_max": 0.005149058604729362,
      "clip_ratio/high_mean": 0.0020658233697758988,
      "clip_ratio/low_mean": 0.0013104771314829122,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003376300497620832,
      "epoch": 7.718576844561096,
      "grad_norm": 0.10006771236658096,
      "learning_rate": 1e-06,
      "loss": -0.0579,
      "step": 1498
    },
    {
      "clip_ratio/high_max": 0.004630413815903012,
      "clip_ratio/high_mean": 0.0018425955859129317,
      "clip_ratio/low_mean": 0.0015567682085020351,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003399363878997974,
      "epoch": 7.723242927967338,
      "grad_norm": 0.09941881150007248,
      "learning_rate": 1e-06,
      "loss": -0.0614,
      "step": 1499
    },
    {
      "clip_ratio/high_max": 0.004295340389944613,
      "clip_ratio/high_mean": 0.0017776333370420616,
      "clip_ratio/low_mean": 0.0018908265737991314,
      "clip_ratio/low_min": 8.526603050995618e-05,
      "clip_ratio/region_mean": 0.0036684599763248116,
      "epoch": 7.727909011373578,
      "grad_norm": 0.11763487011194229,
      "learning_rate": 1e-06,
      "loss": -0.0134,
      "step": 1500
    },
    {
      "clip_ratio/high_max": 0.005803777487017214,
      "clip_ratio/high_mean": 0.001966384254046716,
      "clip_ratio/low_mean": 0.001671139914833475,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036375242489157245,
      "epoch": 7.732575094779819,
      "grad_norm": 0.12014557421207428,
      "learning_rate": 1e-06,
      "loss": -0.0562,
      "step": 1501
    },
    {
      "clip_ratio/high_max": 0.0042572036472847685,
      "clip_ratio/high_mean": 0.001770846010913374,
      "clip_ratio/low_mean": 0.0018354734675085638,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003606319412938319,
      "epoch": 7.73724117818606,
      "grad_norm": 0.10845109075307846,
      "learning_rate": 1e-06,
      "loss": -0.0318,
      "step": 1502
    },
    {
      "clip_ratio/high_max": 0.005232460724073462,
      "clip_ratio/high_mean": 0.0018847653409466147,
      "clip_ratio/low_mean": 0.0018729867369984277,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003757752027013339,
      "epoch": 7.741907261592301,
      "grad_norm": 0.107645682990551,
      "learning_rate": 1e-06,
      "loss": -0.0176,
      "step": 1503
    },
    {
      "clip_ratio/high_max": 0.0044113863332313485,
      "clip_ratio/high_mean": 0.0016779145080363378,
      "clip_ratio/low_mean": 0.0019482728639559355,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003626187266490888,
      "epoch": 7.746573344998541,
      "grad_norm": 0.11277163028717041,
      "learning_rate": 1e-06,
      "loss": -0.0197,
      "step": 1504
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0836356026785714,
      "completions/max_length": 4096.0,
      "completions/max_terminated_length": 4012.0,
      "completions/mean_length": 823.0955810546875,
      "completions/mean_terminated_length": 524.3810424804688,
      "completions/min_length": 104.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 7.751239428404783,
      "grad_norm": 0.18448418378829956,
      "learning_rate": 1e-06,
      "loss": -0.0322,
      "num_tokens": 430715788.0,
      "reward": 0.6712472438812256,
      "reward_std": 0.13961604237556458,
      "rewards/simpleverify_reward/mean": 0.6712471842765808,
      "rewards/simpleverify_reward/std": 0.46977630257606506,
      "step": 1505
    },
    {
      "clip_ratio/high_max": 0.0016724604211049154,
      "clip_ratio/high_mean": 0.0006774661960662343,
      "clip_ratio/low_mean": 0.0004023361912004475,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001079802394087892,
      "epoch": 7.755905511811024,
      "grad_norm": 0.19419632852077484,
      "learning_rate": 1e-06,
      "loss": -0.0247,
      "step": 1506
    },
    {
      "clip_ratio/high_max": 0.002159832467441447,
      "clip_ratio/high_mean": 0.0008151285328494851,
      "clip_ratio/low_mean": 0.00047424965123354923,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0012893781731690979,
      "epoch": 7.7605715952172645,
      "grad_norm": 0.19834081828594208,
      "learning_rate": 1e-06,
      "loss": -0.0706,
      "step": 1507
    },
    {
      "clip_ratio/high_max": 0.002334334538318217,
      "clip_ratio/high_mean": 0.000795241558989801,
      "clip_ratio/low_mean": 0.0005752848264819477,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013705263681913493,
      "epoch": 7.765237678623506,
      "grad_norm": 0.20305268466472626,
      "learning_rate": 1e-06,
      "loss": 0.0017,
      "step": 1508
    },
    {
      "clip_ratio/high_max": 0.0021882962573727127,
      "clip_ratio/high_mean": 0.0008869058274285635,
      "clip_ratio/low_mean": 0.0005346108612229727,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.001421516699338099,
      "epoch": 7.769903762029746,
      "grad_norm": 0.17426122725009918,
      "learning_rate": 1e-06,
      "loss": -0.0906,
      "step": 1509
    },
    {
      "clip_ratio/high_max": 0.002131869288859889,
      "clip_ratio/high_mean": 0.0008643377805128694,
      "clip_ratio/low_mean": 0.0005741007180404267,
      "clip_ratio/low_min": 1.7448352082283236e-05,
      "clip_ratio/region_mean": 0.001438438455807045,
      "epoch": 7.7745698454359875,
      "grad_norm": 0.20878808200359344,
      "learning_rate": 1e-06,
      "loss": -0.0588,
      "step": 1510
    },
    {
      "clip_ratio/high_max": 0.001977486965188291,
      "clip_ratio/high_mean": 0.0007866337273298996,
      "clip_ratio/low_mean": 0.000651401966024423,
      "clip_ratio/low_min": 1.414347116224235e-05,
      "clip_ratio/region_mean": 0.001438035658793524,
      "epoch": 7.779235928842228,
      "grad_norm": 0.1959061324596405,
      "learning_rate": 1e-06,
      "loss": 0.01,
      "step": 1511
    },
    {
      "clip_ratio/high_max": 0.002065342774585588,
      "clip_ratio/high_mean": 0.000752827079850249,
      "clip_ratio/low_mean": 0.0005871303965250263,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013399574636423495,
      "epoch": 7.783902012248469,
      "grad_norm": 0.17067955434322357,
      "learning_rate": 1e-06,
      "loss": -0.0212,
      "step": 1512
    },
    {
      "clip_ratio/high_max": 0.002073963805742096,
      "clip_ratio/high_mean": 0.0007970221668074373,
      "clip_ratio/low_mean": 0.0006472511922765989,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0014442733954638243,
      "epoch": 7.78856809565471,
      "grad_norm": 0.1597839593887329,
      "learning_rate": 1e-06,
      "loss": -0.0176,
      "step": 1513
    },
    {
      "clip_ratio/high_max": 0.0021184734687267337,
      "clip_ratio/high_mean": 0.0008468701234960463,
      "clip_ratio/low_mean": 0.0006247014016480534,
      "clip_ratio/low_min": 1.6543144738534465e-05,
      "clip_ratio/region_mean": 0.001471571515139658,
      "epoch": 7.793234179060951,
      "grad_norm": 0.1552352011203766,
      "learning_rate": 1e-06,
      "loss": -0.0401,
      "step": 1514
    },
    {
      "clip_ratio/high_max": 0.001943274319273769,
      "clip_ratio/high_mean": 0.0007048133215903363,
      "clip_ratio/low_mean": 0.0006485898538812762,
      "clip_ratio/low_min": 3.170631953253178e-05,
      "clip_ratio/region_mean": 0.001353403187749791,
      "epoch": 7.797900262467191,
      "grad_norm": 0.1676725596189499,
      "learning_rate": 1e-06,
      "loss": 0.016,
      "step": 1515
    },
    {
      "clip_ratio/high_max": 0.002463613374857232,
      "clip_ratio/high_mean": 0.0009564288939145627,
      "clip_ratio/low_mean": 0.0007616740522280452,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017181029325001873,
      "epoch": 7.802566345873433,
      "grad_norm": 0.1631651669740677,
      "learning_rate": 1e-06,
      "loss": -0.0618,
      "step": 1516
    },
    {
      "clip_ratio/high_max": 0.0024549139125156216,
      "clip_ratio/high_mean": 0.0010258428337692749,
      "clip_ratio/low_mean": 0.0007004829876677832,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0017263258268940262,
      "epoch": 7.807232429279673,
      "grad_norm": 0.24222761392593384,
      "learning_rate": 1e-06,
      "loss": -0.0638,
      "step": 1517
    },
    {
      "clip_ratio/high_max": 0.0028086554957553744,
      "clip_ratio/high_mean": 0.0011056260191253386,
      "clip_ratio/low_mean": 0.0006971920656724251,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0018028180493274704,
      "epoch": 7.811898512685914,
      "grad_norm": 0.16327685117721558,
      "learning_rate": 1e-06,
      "loss": -0.0739,
      "step": 1518
    },
    {
      "clip_ratio/high_max": 0.0028611213856493123,
      "clip_ratio/high_mean": 0.000996094074253051,
      "clip_ratio/low_mean": 0.0008683631622261601,
      "clip_ratio/low_min": 1.674256600381341e-05,
      "clip_ratio/region_mean": 0.0018644572010089178,
      "epoch": 7.816564596092155,
      "grad_norm": 0.17864589393138885,
      "learning_rate": 1e-06,
      "loss": -0.0213,
      "step": 1519
    },
    {
      "clip_ratio/high_max": 0.00227695504509029,
      "clip_ratio/high_mean": 0.0010404180320620071,
      "clip_ratio/low_mean": 0.000932800197006145,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0019732182008738164,
      "epoch": 7.821230679498396,
      "grad_norm": 0.16785147786140442,
      "learning_rate": 1e-06,
      "loss": -0.0632,
      "step": 1520
    },
    {
      "clip_ratio/high_max": 0.005287787120323628,
      "clip_ratio/high_mean": 0.0020277066905691754,
      "clip_ratio/low_mean": 0.0016498571385454852,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036775638873223215,
      "epoch": 7.8258967629046365,
      "grad_norm": 0.09306744486093521,
      "learning_rate": 1e-06,
      "loss": -0.033,
      "step": 1521
    },
    {
      "clip_ratio/high_max": 0.004476631336729042,
      "clip_ratio/high_mean": 0.0017428647479391657,
      "clip_ratio/low_mean": 0.001546897921798518,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0032897626952035353,
      "epoch": 7.830562846310878,
      "grad_norm": 0.11428961157798767,
      "learning_rate": 1e-06,
      "loss": -0.0255,
      "step": 1522
    },
    {
      "clip_ratio/high_max": 0.005534860538318753,
      "clip_ratio/high_mean": 0.0020855005423072726,
      "clip_ratio/low_mean": 0.0016807865795271937,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003766287147300318,
      "epoch": 7.835228929717118,
      "grad_norm": 0.13048382103443146,
      "learning_rate": 1e-06,
      "loss": -0.0717,
      "step": 1523
    },
    {
      "clip_ratio/high_max": 0.005165198905160651,
      "clip_ratio/high_mean": 0.0018588128732517362,
      "clip_ratio/low_mean": 0.0020974801445845515,
      "clip_ratio/low_min": 9.300595411332324e-05,
      "clip_ratio/region_mean": 0.003956292959628627,
      "epoch": 7.83989501312336,
      "grad_norm": 0.12358823418617249,
      "learning_rate": 1e-06,
      "loss": 0.0008,
      "step": 1524
    },
    {
      "clip_ratio/high_max": 0.004943988133163657,
      "clip_ratio/high_mean": 0.0020040498311573174,
      "clip_ratio/low_mean": 0.0015913580173219088,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003595407833927311,
      "epoch": 7.844561096529601,
      "grad_norm": 0.11490639299154282,
      "learning_rate": 1e-06,
      "loss": -0.0915,
      "step": 1525
    },
    {
      "clip_ratio/high_max": 0.005245728782028891,
      "clip_ratio/high_mean": 0.0020393589438754134,
      "clip_ratio/low_mean": 0.001678700245975051,
      "clip_ratio/low_min": 8.982466533780098e-05,
      "clip_ratio/region_mean": 0.0037180591316428035,
      "epoch": 7.849227179935841,
      "grad_norm": 0.13696928322315216,
      "learning_rate": 1e-06,
      "loss": -0.0598,
      "step": 1526
    },
    {
      "clip_ratio/high_max": 0.0034813763631973416,
      "clip_ratio/high_mean": 0.0015270351323124487,
      "clip_ratio/low_mean": 0.0020417072737473063,
      "clip_ratio/low_min": 0.00010003653369494714,
      "clip_ratio/region_mean": 0.00356874248245731,
      "epoch": 7.853893263342083,
      "grad_norm": 0.10824873298406601,
      "learning_rate": 1e-06,
      "loss": 0.0092,
      "step": 1527
    },
    {
      "clip_ratio/high_max": 0.00502075794065604,
      "clip_ratio/high_mean": 0.001747989321302157,
      "clip_ratio/low_mean": 0.0018746527634903032,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.00362264207797125,
      "epoch": 7.858559346748323,
      "grad_norm": 0.10474837571382523,
      "learning_rate": 1e-06,
      "loss": -0.0221,
      "step": 1528
    },
    {
      "clip_ratio/high_max": 0.0041661158902570605,
      "clip_ratio/high_mean": 0.0016701023087080102,
      "clip_ratio/low_mean": 0.001712045577733079,
      "clip_ratio/low_min": 5.383290408644825e-05,
      "clip_ratio/region_mean": 0.003382147966476623,
      "epoch": 7.863225430154564,
      "grad_norm": 0.12958404421806335,
      "learning_rate": 1e-06,
      "loss": -0.0182,
      "step": 1529
    },
    {
      "clip_ratio/high_max": 0.004341957544966135,
      "clip_ratio/high_mean": 0.0017415515503671486,
      "clip_ratio/low_mean": 0.0016735713061279966,
      "clip_ratio/low_min": 6.16735705989413e-05,
      "clip_ratio/region_mean": 0.0034151228464907035,
      "epoch": 7.867891513560805,
      "grad_norm": 0.1064004898071289,
      "learning_rate": 1e-06,
      "loss": -0.0408,
      "step": 1530
    },
    {
      "clip_ratio/high_max": 0.004208986523735803,
      "clip_ratio/high_mean": 0.0014789962715440197,
      "clip_ratio/low_mean": 0.0017891750176204368,
      "clip_ratio/low_min": 5.764903289673384e-05,
      "clip_ratio/region_mean": 0.0032681712473277003,
      "epoch": 7.872557596967046,
      "grad_norm": 0.10802465677261353,
      "learning_rate": 1e-06,
      "loss": 0.0152,
      "step": 1531
    },
    {
      "clip_ratio/high_max": 0.004379143705591559,
      "clip_ratio/high_mean": 0.0018261924305988941,
      "clip_ratio/low_mean": 0.0016290333787765121,
      "clip_ratio/low_min": 3.2688283681636676e-05,
      "clip_ratio/region_mean": 0.003455225822108332,
      "epoch": 7.8772236803732865,
      "grad_norm": 0.1274808943271637,
      "learning_rate": 1e-06,
      "loss": -0.0625,
      "step": 1532
    },
    {
      "clip_ratio/high_max": 0.00491440951009281,
      "clip_ratio/high_mean": 0.0020928465128235985,
      "clip_ratio/low_mean": 0.00162763443950098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0037204809195827693,
      "epoch": 7.881889763779528,
      "grad_norm": 0.12007849663496017,
      "learning_rate": 1e-06,
      "loss": -0.0649,
      "step": 1533
    },
    {
      "clip_ratio/high_max": 0.005424270071671344,
      "clip_ratio/high_mean": 0.002143973426427692,
      "clip_ratio/low_mean": 0.0017188498586619971,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0038628233887720853,
      "epoch": 7.886555847185768,
      "grad_norm": 0.11042458564043045,
      "learning_rate": 1e-06,
      "loss": -0.0748,
      "step": 1534
    },
    {
      "clip_ratio/high_max": 0.004832319049455691,
      "clip_ratio/high_mean": 0.0018328824280615663,
      "clip_ratio/low_mean": 0.002159742532967357,
      "clip_ratio/low_min": 0.00013025767111685127,
      "clip_ratio/region_mean": 0.003992625017417595,
      "epoch": 7.8912219305920095,
      "grad_norm": 0.11553628742694855,
      "learning_rate": 1e-06,
      "loss": -0.0223,
      "step": 1535
    },
    {
      "clip_ratio/high_max": 0.004251797858159989,
      "clip_ratio/high_mean": 0.0017704329166008392,
      "clip_ratio/low_mean": 0.001913725460326532,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0036841584151261486,
      "epoch": 7.89588801399825,
      "grad_norm": 0.11286855489015579,
      "learning_rate": 1e-06,
      "loss": -0.064,
      "step": 1536
    },
    {
      "epoch": 7.89588801399825,
      "step": 1536,
      "total_flos": 0.0,
      "train_loss": -0.017879978650705047,
      "train_runtime": 58472.4999,
      "train_samples_per_second": 24.518,
      "train_steps_per_second": 0.027
    }
  ],
  "logging_steps": 1,
  "max_steps": 1600,
  "num_input_tokens_seen": 430715788,
  "num_train_epochs": 8,
  "save_steps": 160,
  "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
}