trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9994666666666666,
  "eval_steps": 100,
  "global_step": 937,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "clip_ratio": 0.0,
      "completion_length": 644.2009124755859,
      "epoch": 0.0010666666666666667,
      "grad_norm": 0.7038388582582233,
      "kl": 0.0,
      "learning_rate": 3.191489361702128e-08,
      "loss": 0.0066,
      "reward": 0.3364444375038147,
      "reward_std": 0.4127754420042038,
      "rewards/accuracy_reward": 0.0,
      "rewards/cosine_scaled_reward": 0.33644443191587925,
      "rewards/format_reward": 0.0,
      "rewards/len_reward": -0.04627744515892118,
      "rewards/reasoning_steps_reward": 0.2901785895228386,
      "rewards/tag_count_reward": 0.01562500069849193,
      "step": 1
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 646.2991361618042,
      "epoch": 0.005333333333333333,
      "grad_norm": 1.165178025484845,
      "kl": 0.00015673041343688965,
      "learning_rate": 1.5957446808510638e-07,
      "loss": 0.047,
      "reward": 0.3832849070895463,
      "reward_std": 0.4201222136616707,
      "rewards/accuracy_reward": 0.007812500465661287,
      "rewards/cosine_scaled_reward": 0.37212418764829636,
      "rewards/format_reward": 0.003348214435391128,
      "rewards/len_reward": -0.04301465207754518,
      "rewards/reasoning_steps_reward": 0.3128720454405993,
      "rewards/tag_count_reward": 0.02008928661234677,
      "step": 5
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 651.7919937133789,
      "epoch": 0.010666666666666666,
      "grad_norm": 0.9583817953361469,
      "kl": 0.0002131938934326172,
      "learning_rate": 3.1914893617021275e-07,
      "loss": 0.0412,
      "reward": 0.30802508229389786,
      "reward_std": 0.39543123468756675,
      "rewards/accuracy_reward": 0.0062500002793967726,
      "rewards/cosine_scaled_reward": 0.2964179317001253,
      "rewards/format_reward": 0.0053571430966258046,
      "rewards/len_reward": -0.05133374946017284,
      "rewards/reasoning_steps_reward": 0.30892858691513536,
      "rewards/tag_count_reward": 0.017410715389996768,
      "step": 10
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 634.5687789916992,
      "epoch": 0.016,
      "grad_norm": 1.2665941426985083,
      "kl": 0.0002197742462158203,
      "learning_rate": 4.787234042553192e-07,
      "loss": 0.03,
      "reward": 0.3102074888069183,
      "reward_std": 0.41705508157610893,
      "rewards/accuracy_reward": 0.00714285746216774,
      "rewards/cosine_scaled_reward": 0.2977074964437634,
      "rewards/format_reward": 0.0053571430966258046,
      "rewards/len_reward": -0.04337599159625825,
      "rewards/reasoning_steps_reward": 0.28660716451704504,
      "rewards/tag_count_reward": 0.02656250132713467,
      "step": 15
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 639.2134269714355,
      "epoch": 0.021333333333333333,
      "grad_norm": 0.5388664872711878,
      "kl": 0.00028989315032958987,
      "learning_rate": 6.382978723404255e-07,
      "loss": 0.0341,
      "reward": 0.33970867421012374,
      "reward_std": 0.3754332659766078,
      "rewards/accuracy_reward": 0.008035714738070965,
      "rewards/cosine_scaled_reward": 0.3289943867130205,
      "rewards/format_reward": 0.0026785715483129023,
      "rewards/len_reward": -0.0342805469466839,
      "rewards/reasoning_steps_reward": 0.32232144884765146,
      "rewards/tag_count_reward": 0.02187500111758709,
      "step": 20
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 641.0866355895996,
      "epoch": 0.02666666666666667,
      "grad_norm": 1.33339338523773,
      "kl": 0.00041189193725585935,
      "learning_rate": 7.978723404255319e-07,
      "loss": 0.059,
      "reward": 0.3427321955561638,
      "reward_std": 0.4108327502384782,
      "rewards/accuracy_reward": 0.004464285913854837,
      "rewards/cosine_scaled_reward": 0.3364822005853057,
      "rewards/format_reward": 0.001785714365541935,
      "rewards/len_reward": -0.023951610503718256,
      "rewards/reasoning_steps_reward": 0.30625002086162567,
      "rewards/tag_count_reward": 0.018750000605359675,
      "step": 25
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 641.2518119812012,
      "epoch": 0.032,
      "grad_norm": 0.5777275749493307,
      "kl": 0.0007986545562744141,
      "learning_rate": 9.574468085106384e-07,
      "loss": 0.0497,
      "reward": 0.3393188311718404,
      "reward_std": 0.39051519297063353,
      "rewards/accuracy_reward": 0.00714285746216774,
      "rewards/cosine_scaled_reward": 0.32503311512991784,
      "rewards/format_reward": 0.00714285746216774,
      "rewards/len_reward": -0.029087586002424358,
      "rewards/reasoning_steps_reward": 0.30952382907271386,
      "rewards/tag_count_reward": 0.025446430174633862,
      "step": 30
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 636.4821701049805,
      "epoch": 0.037333333333333336,
      "grad_norm": 1.1015594817672492,
      "kl": 0.053539371490478514,
      "learning_rate": 1.1170212765957447e-06,
      "loss": 0.0637,
      "reward": 0.36175463497638705,
      "reward_std": 0.3966035820543766,
      "rewards/accuracy_reward": 0.0062500002793967726,
      "rewards/cosine_scaled_reward": 0.34925462789833545,
      "rewards/format_reward": 0.0062500002793967726,
      "rewards/len_reward": -0.030977025411993965,
      "rewards/reasoning_steps_reward": 0.3026785932481289,
      "rewards/tag_count_reward": 0.02031250107102096,
      "step": 35
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 662.1411056518555,
      "epoch": 0.042666666666666665,
      "grad_norm": 0.6754960541104272,
      "kl": 0.0025452613830566407,
      "learning_rate": 1.276595744680851e-06,
      "loss": 0.0271,
      "reward": 0.3522125052288175,
      "reward_std": 0.4222590584307909,
      "rewards/accuracy_reward": 0.010714286193251609,
      "rewards/cosine_scaled_reward": 0.3307839307235554,
      "rewards/format_reward": 0.010714286193251609,
      "rewards/len_reward": -0.03555333581316518,
      "rewards/reasoning_steps_reward": 0.3464285958558321,
      "rewards/tag_count_reward": 0.031919644214212894,
      "step": 40
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 634.5562736511231,
      "epoch": 0.048,
      "grad_norm": 0.5612643620767712,
      "kl": 0.003268718719482422,
      "learning_rate": 1.4361702127659576e-06,
      "loss": 0.0534,
      "reward": 0.3792000488378108,
      "reward_std": 0.40308301597833635,
      "rewards/accuracy_reward": 0.0053571430966258046,
      "rewards/cosine_scaled_reward": 0.3649143366143107,
      "rewards/format_reward": 0.008928571920841932,
      "rewards/len_reward": -0.029042343396577054,
      "rewards/reasoning_steps_reward": 0.2964285874739289,
      "rewards/tag_count_reward": 0.02991071599535644,
      "step": 45
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 622.6723480224609,
      "epoch": 0.05333333333333334,
      "grad_norm": 0.7108858441676024,
      "kl": 0.003692150115966797,
      "learning_rate": 1.5957446808510639e-06,
      "loss": 0.0455,
      "reward": 0.371215118841792,
      "reward_std": 0.3538612272590399,
      "rewards/accuracy_reward": 0.004464285913854837,
      "rewards/cosine_scaled_reward": 0.3605008259153692,
      "rewards/format_reward": 0.0062500002793967726,
      "rewards/len_reward": -0.02482663995178882,
      "rewards/reasoning_steps_reward": 0.2991071559488773,
      "rewards/tag_count_reward": 0.023883929662406444,
      "step": 50
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 611.9214553833008,
      "epoch": 0.058666666666666666,
      "grad_norm": 0.43136348282970804,
      "kl": 0.004044151306152344,
      "learning_rate": 1.7553191489361702e-06,
      "loss": 0.0401,
      "reward": 0.4254450174048543,
      "reward_std": 0.3616549573838711,
      "rewards/accuracy_reward": 0.01339285783469677,
      "rewards/cosine_scaled_reward": 0.3995521478354931,
      "rewards/format_reward": 0.012500000558793545,
      "rewards/len_reward": -0.015584439341910183,
      "rewards/reasoning_steps_reward": 0.30208335518836976,
      "rewards/tag_count_reward": 0.030133930058218537,
      "step": 55
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 625.6402084350586,
      "epoch": 0.064,
      "grad_norm": 0.8297564908222154,
      "kl": 0.0055370330810546875,
      "learning_rate": 1.9148936170212767e-06,
      "loss": 0.0352,
      "reward": 0.43232634887099264,
      "reward_std": 0.3845306318253279,
      "rewards/accuracy_reward": 0.011607143376022578,
      "rewards/cosine_scaled_reward": 0.3948263553902507,
      "rewards/format_reward": 0.025892858393490314,
      "rewards/len_reward": -0.009139302224502899,
      "rewards/reasoning_steps_reward": 0.2845238268375397,
      "rewards/tag_count_reward": 0.052455359627492724,
      "step": 60
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 596.3741302490234,
      "epoch": 0.06933333333333333,
      "grad_norm": 0.7228866431898863,
      "kl": 0.012695503234863282,
      "learning_rate": 2.074468085106383e-06,
      "loss": 0.0365,
      "reward": 0.5030742045491934,
      "reward_std": 0.43186259232461455,
      "rewards/accuracy_reward": 0.02767857275903225,
      "rewards/cosine_scaled_reward": 0.4128956302069128,
      "rewards/format_reward": 0.0625000030733645,
      "rewards/len_reward": -0.015213955337821971,
      "rewards/reasoning_steps_reward": 0.3366071626543999,
      "rewards/tag_count_reward": 0.10022321869619191,
      "step": 65
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 585.4160987854004,
      "epoch": 0.07466666666666667,
      "grad_norm": 0.6682633979026431,
      "kl": 0.01843681335449219,
      "learning_rate": 2.2340425531914894e-06,
      "loss": 0.0505,
      "reward": 0.5593858446925879,
      "reward_std": 0.4915604364126921,
      "rewards/accuracy_reward": 0.0419642879627645,
      "rewards/cosine_scaled_reward": 0.3995644087903202,
      "rewards/format_reward": 0.11785714896395802,
      "rewards/len_reward": -0.01337289962102659,
      "rewards/reasoning_steps_reward": 0.3351190686225891,
      "rewards/tag_count_reward": 0.17343750719446688,
      "step": 70
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 557.2544898986816,
      "epoch": 0.08,
      "grad_norm": 0.8532958065384724,
      "kl": 0.014874649047851563,
      "learning_rate": 2.3936170212765957e-06,
      "loss": 0.0318,
      "reward": 0.5684177622199058,
      "reward_std": 0.41847779490053655,
      "rewards/accuracy_reward": 0.025892858393490314,
      "rewards/cosine_scaled_reward": 0.4523463241755962,
      "rewards/format_reward": 0.09017857620492578,
      "rewards/len_reward": -0.018315281078685075,
      "rewards/reasoning_steps_reward": 0.3172619273886085,
      "rewards/tag_count_reward": 0.12142857732251286,
      "step": 75
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 628.3152076721192,
      "epoch": 0.08533333333333333,
      "grad_norm": 0.9448142819988257,
      "kl": 0.009454345703125,
      "learning_rate": 2.553191489361702e-06,
      "loss": 0.0125,
      "reward": 0.5472255704924465,
      "reward_std": 0.4222383305430412,
      "rewards/accuracy_reward": 0.0455357164144516,
      "rewards/cosine_scaled_reward": 0.4150827143341303,
      "rewards/format_reward": 0.08660714719444514,
      "rewards/len_reward": -0.03305647437373409,
      "rewards/reasoning_steps_reward": 0.29196430519223215,
      "rewards/tag_count_reward": 0.10915179166477174,
      "step": 80
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 641.1973541259765,
      "epoch": 0.09066666666666667,
      "grad_norm": 1.2685441941621476,
      "kl": 0.027700042724609374,
      "learning_rate": 2.7127659574468088e-06,
      "loss": 0.0222,
      "reward": 0.711650275811553,
      "reward_std": 0.6027548030018807,
      "rewards/accuracy_reward": 0.07857143254950642,
      "rewards/cosine_scaled_reward": 0.3482574049849063,
      "rewards/format_reward": 0.2848214406520128,
      "rewards/len_reward": -0.04190118510159664,
      "rewards/reasoning_steps_reward": 0.32886906489729884,
      "rewards/tag_count_reward": 0.26964286863803866,
      "step": 85
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 652.3964576721191,
      "epoch": 0.096,
      "grad_norm": 0.7164425205527156,
      "kl": 0.0451019287109375,
      "learning_rate": 2.872340425531915e-06,
      "loss": 0.0152,
      "reward": 0.8936719134449959,
      "reward_std": 0.6759964823722839,
      "rewards/accuracy_reward": 0.13303572190925478,
      "rewards/cosine_scaled_reward": 0.29992188327014446,
      "rewards/format_reward": 0.46071430779993533,
      "rewards/len_reward": -0.044884511404961816,
      "rewards/reasoning_steps_reward": 0.3279762100428343,
      "rewards/tag_count_reward": 0.38973215855658055,
      "step": 90
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 584.2643104553223,
      "epoch": 0.10133333333333333,
      "grad_norm": 0.8742676458030147,
      "kl": 0.0368682861328125,
      "learning_rate": 2.9999895838948146e-06,
      "loss": 0.0333,
      "reward": 1.0263997331261634,
      "reward_std": 0.7044595286250115,
      "rewards/accuracy_reward": 0.18839286491274834,
      "rewards/cosine_scaled_reward": 0.3335425513330847,
      "rewards/format_reward": 0.5044643081724643,
      "rewards/len_reward": -0.031246750250284096,
      "rewards/reasoning_steps_reward": 0.33482144922018053,
      "rewards/tag_count_reward": 0.4535714492201805,
      "step": 95
    },
    {
      "epoch": 0.10666666666666667,
      "grad_norm": 4.100207664946175,
      "learning_rate": 2.9996250354024346e-06,
      "loss": 0.0501,
      "step": 100
    },
    {
      "epoch": 0.10666666666666667,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 572.2930786132813,
      "eval_kl": 0.056494140625,
      "eval_loss": 0.056255146861076355,
      "eval_reward": 1.2980437994003295,
      "eval_reward_std": 0.7676759362220764,
      "eval_rewards/accuracy_reward": 0.325000011920929,
      "eval_rewards/cosine_scaled_reward": 0.3194723308086395,
      "eval_rewards/format_reward": 0.6535714626312256,
      "eval_rewards/len_reward": 0.0353432796895504,
      "eval_rewards/reasoning_steps_reward": 0.3666666865348816,
      "eval_rewards/tag_count_reward": 0.620089304447174,
      "eval_runtime": 113.2606,
      "eval_samples_per_second": 0.671,
      "eval_steps_per_second": 0.009,
      "step": 100
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 580.3339561462402,
      "epoch": 0.112,
      "grad_norm": 1.024357541492345,
      "kl": 0.14632110595703124,
      "learning_rate": 2.9987398263020837e-06,
      "loss": 0.0522,
      "reward": 1.2215609500184654,
      "reward_std": 0.7516257010400296,
      "rewards/accuracy_reward": 0.2843750137835741,
      "rewards/cosine_scaled_reward": 0.30638233484933153,
      "rewards/format_reward": 0.6308036033064127,
      "rewards/len_reward": -0.03830806077457964,
      "rewards/reasoning_steps_reward": 0.3824405036866665,
      "rewards/tag_count_reward": 0.6280134215950965,
      "step": 105
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 613.3071662902832,
      "epoch": 0.11733333333333333,
      "grad_norm": 0.73740128371405,
      "kl": 0.07979736328125,
      "learning_rate": 2.997334263932927e-06,
      "loss": 0.0507,
      "reward": 1.2840619757771492,
      "reward_std": 0.7633979134261608,
      "rewards/accuracy_reward": 0.3741071574389935,
      "rewards/cosine_scaled_reward": 0.31888334788382056,
      "rewards/format_reward": 0.5910714514553547,
      "rewards/len_reward": -0.03648645522480365,
      "rewards/reasoning_steps_reward": 0.4241071671247482,
      "rewards/tag_count_reward": 0.6294643133878708,
      "step": 110
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 623.3973457336426,
      "epoch": 0.12266666666666666,
      "grad_norm": 0.5315564057866974,
      "kl": 0.0554351806640625,
      "learning_rate": 2.9954088362975936e-06,
      "loss": 0.0422,
      "reward": 1.4692801490426064,
      "reward_std": 0.7539573520421982,
      "rewards/accuracy_reward": 0.4035714466124773,
      "rewards/cosine_scaled_reward": 0.3612444051541388,
      "rewards/format_reward": 0.7044643200933933,
      "rewards/len_reward": -0.017655941796692786,
      "rewards/reasoning_steps_reward": 0.47142860367894174,
      "rewards/tag_count_reward": 0.6227678850293159,
      "step": 115
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 605.333065032959,
      "epoch": 0.128,
      "grad_norm": 0.5958118040109989,
      "kl": 0.074505615234375,
      "learning_rate": 2.99296421189274e-06,
      "loss": 0.0551,
      "reward": 1.4197789639234544,
      "reward_std": 0.7952461540699005,
      "rewards/accuracy_reward": 0.4223214466124773,
      "rewards/cosine_scaled_reward": 0.32067176890559496,
      "rewards/format_reward": 0.6767857477068902,
      "rewards/len_reward": -0.03406066980678588,
      "rewards/reasoning_steps_reward": 0.4580357432365417,
      "rewards/tag_count_reward": 0.6189732424914837,
      "step": 120
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 630.7732460021973,
      "epoch": 0.13333333333333333,
      "grad_norm": 0.8607419693848777,
      "kl": 0.065643310546875,
      "learning_rate": 2.9900012394769546e-06,
      "loss": 0.0506,
      "reward": 1.4135100565850736,
      "reward_std": 0.7826988212764263,
      "rewards/accuracy_reward": 0.42232144698500634,
      "rewards/cosine_scaled_reward": 0.2822600125335157,
      "rewards/format_reward": 0.7089286029338837,
      "rewards/len_reward": -0.04302762195584364,
      "rewards/reasoning_steps_reward": 0.42529764398932457,
      "rewards/tag_count_reward": 0.6520089574158192,
      "step": 125
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 577.2678840637207,
      "epoch": 0.13866666666666666,
      "grad_norm": 1.7581222298002241,
      "kl": 0.0793212890625,
      "learning_rate": 2.986520947776075e-06,
      "loss": 0.0432,
      "reward": 1.540949085354805,
      "reward_std": 0.7308154627680779,
      "rewards/accuracy_reward": 0.45892859138548375,
      "rewards/cosine_scaled_reward": 0.29630617263028397,
      "rewards/format_reward": 0.7857143327593803,
      "rewards/len_reward": -0.040497589367441834,
      "rewards/reasoning_steps_reward": 0.40059526450932026,
      "rewards/tag_count_reward": 0.7185268178582191,
      "step": 130
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 618.2786010742187,
      "epoch": 0.144,
      "grad_norm": 1.2147641283215929,
      "kl": 0.076068115234375,
      "learning_rate": 2.982524545126018e-06,
      "loss": 0.03,
      "reward": 1.4926121950149536,
      "reward_std": 0.7447091139853,
      "rewards/accuracy_reward": 0.46875002086162565,
      "rewards/cosine_scaled_reward": 0.28100501019507645,
      "rewards/format_reward": 0.7428571835160256,
      "rewards/len_reward": -0.04923985062050633,
      "rewards/reasoning_steps_reward": 0.3982143126428127,
      "rewards/tag_count_reward": 0.660714315623045,
      "step": 135
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 631.2741371154785,
      "epoch": 0.14933333333333335,
      "grad_norm": 0.9629578835791102,
      "kl": 0.102850341796875,
      "learning_rate": 2.9780134190532553e-06,
      "loss": 0.0561,
      "reward": 1.442286056280136,
      "reward_std": 0.8065447643399238,
      "rewards/accuracy_reward": 0.423214303702116,
      "rewards/cosine_scaled_reward": 0.25300029437057675,
      "rewards/format_reward": 0.7660714693367481,
      "rewards/len_reward": -0.05252085686661303,
      "rewards/reasoning_steps_reward": 0.43720240965485574,
      "rewards/tag_count_reward": 0.6805803880095482,
      "step": 140
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 664.377709197998,
      "epoch": 0.15466666666666667,
      "grad_norm": 1.5387538457241525,
      "kl": 0.1677978515625,
      "learning_rate": 2.972989135793071e-06,
      "loss": 0.1038,
      "reward": 0.9583968121558428,
      "reward_std": 0.9008305951952934,
      "rewards/accuracy_reward": 0.29732144232839347,
      "rewards/cosine_scaled_reward": 0.11553962156176567,
      "rewards/format_reward": 0.5455357372760773,
      "rewards/len_reward": -0.08262603848124854,
      "rewards/reasoning_steps_reward": 0.43005954995751383,
      "rewards/tag_count_reward": 0.6725446745753288,
      "step": 145
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 669.9571769714355,
      "epoch": 0.16,
      "grad_norm": 8.232445507998719,
      "kl": 1.080517578125,
      "learning_rate": 2.967453439745775e-06,
      "loss": 0.1908,
      "reward": 0.8955060914158821,
      "reward_std": 0.9506111726164818,
      "rewards/accuracy_reward": 0.28571429643779994,
      "rewards/cosine_scaled_reward": 0.10711320142145268,
      "rewards/format_reward": 0.502678594738245,
      "rewards/len_reward": -0.06221853480674326,
      "rewards/reasoning_steps_reward": 0.3529762107878923,
      "rewards/tag_count_reward": 0.7274553924798965,
      "step": 150
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 548.216098022461,
      "epoch": 0.16533333333333333,
      "grad_norm": 5.902897016758618,
      "kl": 7.4263671875,
      "learning_rate": 2.961408252871058e-06,
      "loss": 0.3684,
      "reward": 1.1710864089429378,
      "reward_std": 0.969765692949295,
      "rewards/accuracy_reward": 0.38035716153681276,
      "rewards/cosine_scaled_reward": 0.17108636341872624,
      "rewards/format_reward": 0.6196428865194321,
      "rewards/len_reward": -0.03628237677039579,
      "rewards/reasoning_steps_reward": 0.3818452678620815,
      "rewards/tag_count_reward": 0.8296875432133675,
      "step": 155
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 623.2044944763184,
      "epoch": 0.17066666666666666,
      "grad_norm": 15.230557110031734,
      "kl": 2.2271484375,
      "learning_rate": 2.9548556740207e-06,
      "loss": 0.2607,
      "reward": 1.023178616911173,
      "reward_std": 0.9279091581702232,
      "rewards/accuracy_reward": 0.31607144232839346,
      "rewards/cosine_scaled_reward": 0.11335714326705784,
      "rewards/format_reward": 0.5937500290572644,
      "rewards/len_reward": -0.04749395059770904,
      "rewards/reasoning_steps_reward": 0.33750002086162567,
      "rewards/tag_count_reward": 0.8383928969502449,
      "step": 160
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 735.4348541259766,
      "epoch": 0.176,
      "grad_norm": 5.053812153888101,
      "kl": 1.266015625,
      "learning_rate": 2.9477979782098592e-06,
      "loss": 0.1493,
      "reward": 1.0551368452608585,
      "reward_std": 0.8986638769507408,
      "rewards/accuracy_reward": 0.3321428726427257,
      "rewards/cosine_scaled_reward": 0.1435296577285044,
      "rewards/format_reward": 0.5794643104076386,
      "rewards/len_reward": -0.049430397151081704,
      "rewards/reasoning_steps_reward": 0.33333336040377615,
      "rewards/tag_count_reward": 0.8238839626312255,
      "step": 165
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 845.265217590332,
      "epoch": 0.18133333333333335,
      "grad_norm": 0.6919959400923131,
      "kl": 0.52850341796875,
      "learning_rate": 2.9402376158272022e-06,
      "loss": 0.0567,
      "reward": 1.0284505672752857,
      "reward_std": 0.873851603269577,
      "rewards/accuracy_reward": 0.32767858579754827,
      "rewards/cosine_scaled_reward": 0.1400576631671356,
      "rewards/format_reward": 0.5607143126428127,
      "rewards/len_reward": -0.07766101571614854,
      "rewards/reasoning_steps_reward": 0.3125000227242708,
      "rewards/tag_count_reward": 0.7578125342726707,
      "step": 170
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 948.5634368896484,
      "epoch": 0.18666666666666668,
      "grad_norm": 0.9504642967041296,
      "kl": 0.3015380859375,
      "learning_rate": 2.9321772117841463e-06,
      "loss": 0.019,
      "reward": 0.8819758415222168,
      "reward_std": 0.7948870480060577,
      "rewards/accuracy_reward": 0.27946429857984184,
      "rewards/cosine_scaled_reward": 0.048047243334440284,
      "rewards/format_reward": 0.5544643133878708,
      "rewards/len_reward": -0.12528894301503896,
      "rewards/reasoning_steps_reward": 0.31071430891752244,
      "rewards/tag_count_reward": 0.7254464566707611,
      "step": 175
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 990.125033569336,
      "epoch": 0.192,
      "grad_norm": 0.8747320513862124,
      "kl": 0.17032470703125,
      "learning_rate": 2.923619564603501e-06,
      "loss": -0.0046,
      "reward": 0.9295428976416588,
      "reward_std": 0.7955453962087631,
      "rewards/accuracy_reward": 0.3446428715251386,
      "rewards/cosine_scaled_reward": 0.0063285754295066,
      "rewards/format_reward": 0.5785714522004127,
      "rewards/len_reward": -0.14820685870945455,
      "rewards/reasoning_steps_reward": 0.3136904966086149,
      "rewards/tag_count_reward": 0.7122768208384513,
      "step": 180
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 988.6259307861328,
      "epoch": 0.19733333333333333,
      "grad_norm": 1.7966045934893466,
      "kl": 0.16611328125,
      "learning_rate": 2.9145676454478435e-06,
      "loss": -0.0042,
      "reward": 0.9886309564113617,
      "reward_std": 0.7453033030033112,
      "rewards/accuracy_reward": 0.3205357275903225,
      "rewards/cosine_scaled_reward": -0.0006547938079165761,
      "rewards/format_reward": 0.6687500283122063,
      "rewards/len_reward": -0.1490553854033351,
      "rewards/reasoning_steps_reward": 0.4044643186032772,
      "rewards/tag_count_reward": 0.7602678939700127,
      "step": 185
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 762.9036071777343,
      "epoch": 0.20266666666666666,
      "grad_norm": 100.15512012017349,
      "kl": 7.372412109375,
      "learning_rate": 2.9050245970879456e-06,
      "loss": 0.3495,
      "reward": 0.7870586156845093,
      "reward_std": 0.7903555080294609,
      "rewards/accuracy_reward": 0.20714286556467415,
      "rewards/cosine_scaled_reward": -0.001334278640570119,
      "rewards/format_reward": 0.5812500268220901,
      "rewards/len_reward": -0.1288936346769333,
      "rewards/reasoning_steps_reward": 0.35505954883992674,
      "rewards/tag_count_reward": 0.6861607432365417,
      "step": 190
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 582.100023651123,
      "epoch": 0.208,
      "grad_norm": 3.905714073659836,
      "kl": 1.8900390625,
      "learning_rate": 2.8949937328116252e-06,
      "loss": 0.1685,
      "reward": 0.5571490220725537,
      "reward_std": 0.8787636801600456,
      "rewards/accuracy_reward": 0.10357143403962255,
      "rewards/cosine_scaled_reward": -0.02320814849808812,
      "rewards/format_reward": 0.4767857387661934,
      "rewards/len_reward": -0.09527752730937208,
      "rewards/reasoning_steps_reward": 0.3270833570510149,
      "rewards/tag_count_reward": 0.6017857432365418,
      "step": 195
    },
    {
      "epoch": 0.21333333333333335,
      "grad_norm": 3.9416410412461524,
      "learning_rate": 2.884478535273393e-06,
      "loss": 0.3992,
      "step": 200
    },
    {
      "epoch": 0.21333333333333335,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 499.3030212402344,
      "eval_kl": 3.175,
      "eval_loss": 0.4126989245414734,
      "eval_reward": 0.8562156677246093,
      "eval_reward_std": 0.9723240971565247,
      "eval_rewards/accuracy_reward": 0.21607144176959991,
      "eval_rewards/cosine_scaled_reward": 0.10621565729379653,
      "eval_rewards/format_reward": 0.5339285790920257,
      "eval_rewards/len_reward": 0.021631143055856228,
      "eval_rewards/reasoning_steps_reward": 0.30654761791229246,
      "eval_rewards/tag_count_reward": 0.6245536088943482,
      "eval_runtime": 110.8118,
      "eval_samples_per_second": 0.686,
      "eval_steps_per_second": 0.009,
      "step": 200
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 502.50091552734375,
      "epoch": 0.21866666666666668,
      "grad_norm": 7.328057495766339,
      "kl": 4.2474609375,
      "learning_rate": 2.8734826552852934e-06,
      "loss": 0.4105,
      "reward": 0.8123717293143272,
      "reward_std": 0.9503199592232704,
      "rewards/accuracy_reward": 0.19732143627479673,
      "rewards/cosine_scaled_reward": 0.07755026518425438,
      "rewards/format_reward": 0.5375000230967999,
      "rewards/len_reward": -0.06171439889585599,
      "rewards/reasoning_steps_reward": 0.3178571631200612,
      "rewards/tag_count_reward": 0.635267884656787,
      "step": 205
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 461.1893020629883,
      "epoch": 0.224,
      "grad_norm": 6.661356281614044,
      "kl": 4.8693359375,
      "learning_rate": 2.86200991054937e-06,
      "loss": 0.5026,
      "reward": 1.0250703942030668,
      "reward_std": 0.9945615231990814,
      "rewards/accuracy_reward": 0.26696429792791604,
      "rewards/cosine_scaled_reward": 0.1456060634693131,
      "rewards/format_reward": 0.6125000283122063,
      "rewards/len_reward": -0.038271602327586154,
      "rewards/reasoning_steps_reward": 0.30952383354306223,
      "rewards/tag_count_reward": 0.7133928909897804,
      "step": 210
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 454.26252365112305,
      "epoch": 0.22933333333333333,
      "grad_norm": 29.073733687853476,
      "kl": 5.935546875,
      "learning_rate": 2.850064284332176e-06,
      "loss": 0.4959,
      "reward": 0.9025619432330132,
      "reward_std": 0.9635577172040939,
      "rewards/accuracy_reward": 0.23928572610020638,
      "rewards/cosine_scaled_reward": 0.12399048676597886,
      "rewards/format_reward": 0.5392857387661933,
      "rewards/len_reward": -0.04608473673724802,
      "rewards/reasoning_steps_reward": 0.2455357311293483,
      "rewards/tag_count_reward": 0.6573661014437675,
      "step": 215
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 431.4607322692871,
      "epoch": 0.23466666666666666,
      "grad_norm": 32.81745130495549,
      "kl": 4.9755859375,
      "learning_rate": 2.8376499240818166e-06,
      "loss": 0.4603,
      "reward": 1.0474480047822,
      "reward_std": 1.0046013176441193,
      "rewards/accuracy_reward": 0.30267858877778053,
      "rewards/cosine_scaled_reward": 0.1724479661497753,
      "rewards/format_reward": 0.5723214603960514,
      "rewards/len_reward": -0.03554261325771222,
      "rewards/reasoning_steps_reward": 0.27380954213440417,
      "rewards/tag_count_reward": 0.6997768178582191,
      "step": 220
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 404.6401988983154,
      "epoch": 0.24,
      "grad_norm": 28.1722788422819,
      "kl": 3.94912109375,
      "learning_rate": 2.8247711399879734e-06,
      "loss": 0.3951,
      "reward": 1.0693716026842595,
      "reward_std": 0.9888196289539337,
      "rewards/accuracy_reward": 0.2991071553900838,
      "rewards/cosine_scaled_reward": 0.16133585062343628,
      "rewards/format_reward": 0.6089285999536515,
      "rewards/len_reward": -0.048874559343676086,
      "rewards/reasoning_steps_reward": 0.3071428768336773,
      "rewards/tag_count_reward": 0.722098246216774,
      "step": 225
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 387.7723388671875,
      "epoch": 0.24533333333333332,
      "grad_norm": 13.352200008648655,
      "kl": 4.7109375,
      "learning_rate": 2.8114324034854378e-06,
      "loss": 0.4193,
      "reward": 1.0878922022879123,
      "reward_std": 1.0054454013705254,
      "rewards/accuracy_reward": 0.2955357299186289,
      "rewards/cosine_scaled_reward": 0.15128503099549562,
      "rewards/format_reward": 0.6410714581608772,
      "rewards/len_reward": -0.043122516572475435,
      "rewards/reasoning_steps_reward": 0.28511906825006006,
      "rewards/tag_count_reward": 0.7535714656114578,
      "step": 230
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 410.9437679290771,
      "epoch": 0.25066666666666665,
      "grad_norm": 24.250800007195018,
      "kl": 3.6314453125,
      "learning_rate": 2.7976383457016535e-06,
      "loss": 0.3386,
      "reward": 1.1002146422863006,
      "reward_std": 0.9713477358222008,
      "rewards/accuracy_reward": 0.30625001564621923,
      "rewards/cosine_scaled_reward": 0.1180717434406688,
      "rewards/format_reward": 0.6758928917348385,
      "rewards/len_reward": -0.04972980402235407,
      "rewards/reasoning_steps_reward": 0.31785716116428375,
      "rewards/tag_count_reward": 0.770982176065445,
      "step": 235
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 433.2035934448242,
      "epoch": 0.256,
      "grad_norm": 19.63417414565418,
      "kl": 9.439453125,
      "learning_rate": 2.7833937558488187e-06,
      "loss": 0.7573,
      "reward": 1.1284980118274688,
      "reward_std": 1.0023478761315345,
      "rewards/accuracy_reward": 0.29821430118754505,
      "rewards/cosine_scaled_reward": 0.14189081880031154,
      "rewards/format_reward": 0.6883928954601288,
      "rewards/len_reward": -0.04797074495872948,
      "rewards/reasoning_steps_reward": 0.3354166891425848,
      "rewards/tag_count_reward": 0.7850446790456772,
      "step": 240
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 406.71162338256835,
      "epoch": 0.2613333333333333,
      "grad_norm": 24.59188831849529,
      "kl": 4.787109375,
      "learning_rate": 2.7687035795611003e-06,
      "loss": 0.3719,
      "reward": 1.2017367050051688,
      "reward_std": 0.9599395081400871,
      "rewards/accuracy_reward": 0.32857144586741927,
      "rewards/cosine_scaled_reward": 0.15620094180339947,
      "rewards/format_reward": 0.7169643253087997,
      "rewards/len_reward": -0.04119561462430284,
      "rewards/reasoning_steps_reward": 0.32886907272040844,
      "rewards/tag_count_reward": 0.7966518267989159,
      "step": 245
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 408.95180358886716,
      "epoch": 0.26666666666666666,
      "grad_norm": 9.752015710296007,
      "kl": 3.60859375,
      "learning_rate": 2.7535729171775408e-06,
      "loss": 0.4066,
      "reward": 1.3339441634714604,
      "reward_std": 0.9489350289106369,
      "rewards/accuracy_reward": 0.35714287580922244,
      "rewards/cosine_scaled_reward": 0.22412267459294527,
      "rewards/format_reward": 0.7526786118745804,
      "rewards/len_reward": -0.01727719540358521,
      "rewards/reasoning_steps_reward": 0.34851192869246006,
      "rewards/tag_count_reward": 0.8140625387430191,
      "step": 250
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 411.16162490844727,
      "epoch": 0.272,
      "grad_norm": 3.3832375486237614,
      "kl": 3.37255859375,
      "learning_rate": 2.7380070219712514e-06,
      "loss": 0.3354,
      "reward": 1.3407041341066361,
      "reward_std": 0.9241366110742092,
      "rewards/accuracy_reward": 0.3598214445635676,
      "rewards/cosine_scaled_reward": 0.20677551386179402,
      "rewards/format_reward": 0.7741071850061416,
      "rewards/len_reward": -0.027734974736813457,
      "rewards/reasoning_steps_reward": 0.3889881126582623,
      "rewards/tag_count_reward": 0.8216518253087998,
      "step": 255
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 395.68305587768555,
      "epoch": 0.2773333333333333,
      "grad_norm": 5.777425418025543,
      "kl": 2.6078125,
      "learning_rate": 2.722011298325509e-06,
      "loss": 0.2668,
      "reward": 1.363752231001854,
      "reward_std": 0.9073125638067723,
      "rewards/accuracy_reward": 0.37321430230513214,
      "rewards/cosine_scaled_reward": 0.20660932243335992,
      "rewards/format_reward": 0.7839286118745804,
      "rewards/len_reward": -0.01969163056710386,
      "rewards/reasoning_steps_reward": 0.3535714544355869,
      "rewards/tag_count_reward": 0.8475446835160255,
      "step": 260
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 383.4839458465576,
      "epoch": 0.2826666666666667,
      "grad_norm": 6.935633340822481,
      "kl": 3.378515625,
      "learning_rate": 2.705591299857385e-06,
      "loss": 0.285,
      "reward": 1.3742309853434562,
      "reward_std": 0.9106731534004211,
      "rewards/accuracy_reward": 0.39821430314332246,
      "rewards/cosine_scaled_reward": 0.20994521365500987,
      "rewards/format_reward": 0.7660714671015739,
      "rewards/len_reward": -0.03229370064800605,
      "rewards/reasoning_steps_reward": 0.40327384024858476,
      "rewards/tag_count_reward": 0.8417411088943482,
      "step": 265
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 403.61609268188477,
      "epoch": 0.288,
      "grad_norm": 3.039135573350634,
      "kl": 2.783984375,
      "learning_rate": 2.6887527274895657e-06,
      "loss": 0.1881,
      "reward": 1.4360268995165826,
      "reward_std": 0.8577364102005959,
      "rewards/accuracy_reward": 0.4116071589291096,
      "rewards/cosine_scaled_reward": 0.2128125553485006,
      "rewards/format_reward": 0.8116071850061417,
      "rewards/len_reward": -0.044078399677528066,
      "rewards/reasoning_steps_reward": 0.4425595536828041,
      "rewards/tag_count_reward": 0.8636161103844643,
      "step": 270
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 400.08126678466795,
      "epoch": 0.29333333333333333,
      "grad_norm": 2.4388095548591324,
      "kl": 2.45634765625,
      "learning_rate": 2.6715014274710265e-06,
      "loss": 0.1916,
      "reward": 1.5192030668258667,
      "reward_std": 0.8684033416211605,
      "rewards/accuracy_reward": 0.4464285934343934,
      "rewards/cosine_scaled_reward": 0.25313157981727274,
      "rewards/format_reward": 0.8196428954601288,
      "rewards/len_reward": -0.026938176073599607,
      "rewards/reasoning_steps_reward": 0.4386905148625374,
      "rewards/tag_count_reward": 0.8888393223285675,
      "step": 275
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 449.9669845581055,
      "epoch": 0.2986666666666667,
      "grad_norm": 6.321239319821566,
      "kl": 2.86875,
      "learning_rate": 2.65384338934725e-06,
      "loss": 0.2027,
      "reward": 1.4269624769687652,
      "reward_std": 0.8590005040168762,
      "rewards/accuracy_reward": 0.4142857328057289,
      "rewards/cosine_scaled_reward": 0.21446242899401113,
      "rewards/format_reward": 0.7982143282890319,
      "rewards/len_reward": -0.040290398540673775,
      "rewards/reasoning_steps_reward": 0.46130955815315244,
      "rewards/tag_count_reward": 0.8482143253087997,
      "step": 280
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 393.6366256713867,
      "epoch": 0.304,
      "grad_norm": 4.771107753553364,
      "kl": 1.920849609375,
      "learning_rate": 2.6357847438806916e-06,
      "loss": 0.1957,
      "reward": 1.607939650118351,
      "reward_std": 0.8728329673409462,
      "rewards/accuracy_reward": 0.4839285944588482,
      "rewards/cosine_scaled_reward": 0.28115387591533364,
      "rewards/format_reward": 0.84285718947649,
      "rewards/len_reward": -0.020568763601477257,
      "rewards/reasoning_steps_reward": 0.49107146337628366,
      "rewards/tag_count_reward": 0.9120536118745803,
      "step": 285
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 439.46252059936523,
      "epoch": 0.30933333333333335,
      "grad_norm": 2.222415417228654,
      "kl": 1.5540283203125,
      "learning_rate": 2.617331760922218e-06,
      "loss": 0.1929,
      "reward": 1.7294473230838776,
      "reward_std": 0.8087163001298905,
      "rewards/accuracy_reward": 0.5616071686148644,
      "rewards/cosine_scaled_reward": 0.307125816680491,
      "rewards/format_reward": 0.8607143267989159,
      "rewards/len_reward": 0.0014614581130445003,
      "rewards/reasoning_steps_reward": 0.5306547991931438,
      "rewards/tag_count_reward": 0.910267898440361,
      "step": 290
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 502.6964530944824,
      "epoch": 0.31466666666666665,
      "grad_norm": 3.044011469162571,
      "kl": 2.042333984375,
      "learning_rate": 2.598490847234253e-06,
      "loss": 0.1994,
      "reward": 1.5344257444143294,
      "reward_std": 0.8376345105469227,
      "rewards/accuracy_reward": 0.490178593993187,
      "rewards/cosine_scaled_reward": 0.23710425589233636,
      "rewards/format_reward": 0.8071428954601287,
      "rewards/len_reward": -0.023603520775213837,
      "rewards/reasoning_steps_reward": 0.5127976588904858,
      "rewards/tag_count_reward": 0.8689732521772384,
      "step": 295
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.4088716556569,
      "learning_rate": 2.5792685442663883e-06,
      "loss": 0.1325,
      "step": 300
    },
    {
      "epoch": 0.32,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 569.0503173828125,
      "eval_kl": 1.77578125,
      "eval_loss": 0.19349658489227295,
      "eval_reward": 1.5551659345626831,
      "eval_reward_std": 0.8867796778678894,
      "eval_rewards/accuracy_reward": 0.48214287161827085,
      "eval_rewards/cosine_scaled_reward": 0.258737313747406,
      "eval_rewards/format_reward": 0.8142857432365418,
      "eval_rewards/len_reward": 0.03932540193200111,
      "eval_rewards/reasoning_steps_reward": 0.5386904895305633,
      "eval_rewards/tag_count_reward": 0.8683036088943481,
      "eval_runtime": 115.2712,
      "eval_samples_per_second": 0.659,
      "eval_steps_per_second": 0.009,
      "step": 300
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 557.4629703521729,
      "epoch": 0.3253333333333333,
      "grad_norm": 2.3974125622229288,
      "kl": 2.141943359375,
      "learning_rate": 2.559671525884232e-06,
      "loss": 0.1392,
      "reward": 1.4303073339164256,
      "reward_std": 0.930858300626278,
      "rewards/accuracy_reward": 0.4379464481957257,
      "rewards/cosine_scaled_reward": 0.21914655352011322,
      "rewards/format_reward": 0.7732143267989159,
      "rewards/len_reward": -0.04112708342508995,
      "rewards/reasoning_steps_reward": 0.5296131338924169,
      "rewards/tag_count_reward": 0.8465402163565159,
      "step": 305
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 630.7919937133789,
      "epoch": 0.33066666666666666,
      "grad_norm": 2.478264362396331,
      "kl": 2.63065185546875,
      "learning_rate": 2.539706596052286e-06,
      "loss": 0.116,
      "reward": 1.3845888167619704,
      "reward_std": 0.9000825509428978,
      "rewards/accuracy_reward": 0.41964287469163536,
      "rewards/cosine_scaled_reward": 0.19262447137152777,
      "rewards/format_reward": 0.7723214656114579,
      "rewards/len_reward": -0.06281980487401598,
      "rewards/reasoning_steps_reward": 0.5708333693444729,
      "rewards/tag_count_reward": 0.8419643223285675,
      "step": 310
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 646.2580604553223,
      "epoch": 0.336,
      "grad_norm": 6.9435065986378,
      "kl": 2.74915771484375,
      "learning_rate": 2.5193806864716466e-06,
      "loss": 0.1113,
      "reward": 1.655515044927597,
      "reward_std": 0.7905310012400151,
      "rewards/accuracy_reward": 0.512500025331974,
      "rewards/cosine_scaled_reward": 0.26265784676652404,
      "rewards/format_reward": 0.8803571864962578,
      "rewards/len_reward": -0.04766237259027548,
      "rewards/reasoning_steps_reward": 0.6678571827709675,
      "rewards/tag_count_reward": 0.9017857626080513,
      "step": 315
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 594.9205612182617,
      "epoch": 0.3413333333333333,
      "grad_norm": 7.668425562182992,
      "kl": 2.090625,
      "learning_rate": 2.4987008541733663e-06,
      "loss": 0.1401,
      "reward": 1.6221125468611717,
      "reward_std": 0.8562563940882683,
      "rewards/accuracy_reward": 0.5142857370898127,
      "rewards/cosine_scaled_reward": 0.27300534858368336,
      "rewards/format_reward": 0.8348214700818062,
      "rewards/len_reward": -0.033610415155999364,
      "rewards/reasoning_steps_reward": 0.6363095626235008,
      "rewards/tag_count_reward": 0.8723214656114578,
      "step": 320
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 555.5401992797852,
      "epoch": 0.3466666666666667,
      "grad_norm": 18.400557478777237,
      "kl": 3.0746337890625,
      "learning_rate": 2.477674279068291e-06,
      "loss": 0.187,
      "reward": 1.575891900062561,
      "reward_std": 0.9395873740315437,
      "rewards/accuracy_reward": 0.5107143113389612,
      "rewards/cosine_scaled_reward": 0.27589183195959777,
      "rewards/format_reward": 0.7892857536673545,
      "rewards/len_reward": -0.03333336425275775,
      "rewards/reasoning_steps_reward": 0.600000037252903,
      "rewards/tag_count_reward": 0.8345982521772385,
      "step": 325
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 562.6785972595214,
      "epoch": 0.352,
      "grad_norm": 8.132015615402397,
      "kl": 2.1640625,
      "learning_rate": 2.4563082614542412e-06,
      "loss": 0.1433,
      "reward": 1.4302320718765258,
      "reward_std": 0.9858111351728439,
      "rewards/accuracy_reward": 0.4705357376486063,
      "rewards/cosine_scaled_reward": 0.2239820205140859,
      "rewards/format_reward": 0.7357143267989159,
      "rewards/len_reward": -0.05076549501682166,
      "rewards/reasoning_steps_reward": 0.5842262268066406,
      "rewards/tag_count_reward": 0.7995536103844643,
      "step": 330
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 578.0491333007812,
      "epoch": 0.35733333333333334,
      "grad_norm": 6.7174895292302805,
      "kl": 2.395703125,
      "learning_rate": 2.4346102194813937e-06,
      "loss": 0.1352,
      "reward": 1.4722676426172256,
      "reward_std": 0.9315723203122616,
      "rewards/accuracy_reward": 0.48303573820739987,
      "rewards/cosine_scaled_reward": 0.2249461407540366,
      "rewards/format_reward": 0.7642857551574707,
      "rewards/len_reward": -0.04090934251144063,
      "rewards/reasoning_steps_reward": 0.5836309917271137,
      "rewards/tag_count_reward": 0.8129464685916901,
      "step": 335
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 579.7286003112793,
      "epoch": 0.3626666666666667,
      "grad_norm": 14.327208381956924,
      "kl": 2.17421875,
      "learning_rate": 2.4125876865767443e-06,
      "loss": 0.1168,
      "reward": 1.435131347179413,
      "reward_std": 0.9724505253136158,
      "rewards/accuracy_reward": 0.46964288130402565,
      "rewards/cosine_scaled_reward": 0.21459556268528104,
      "rewards/format_reward": 0.7508928999304771,
      "rewards/len_reward": -0.05366924590198323,
      "rewards/reasoning_steps_reward": 0.5889881312847137,
      "rewards/tag_count_reward": 0.8158482506871223,
      "step": 340
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 582.0196670532226,
      "epoch": 0.368,
      "grad_norm": 12.351783358249422,
      "kl": 2.0623779296875,
      "learning_rate": 2.390248308828548e-06,
      "loss": 0.1202,
      "reward": 1.3827464163303376,
      "reward_std": 0.9574041977524758,
      "rewards/accuracy_reward": 0.45535716265439985,
      "rewards/cosine_scaled_reward": 0.19256778084672987,
      "rewards/format_reward": 0.7348214671015739,
      "rewards/len_reward": -0.050123395124683154,
      "rewards/reasoning_steps_reward": 0.6059524111449719,
      "rewards/tag_count_reward": 0.8122768223285675,
      "step": 345
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 600.7643127441406,
      "epoch": 0.37333333333333335,
      "grad_norm": 44.225679044967485,
      "kl": 2.75009765625,
      "learning_rate": 2.367599842331646e-06,
      "loss": 0.1392,
      "reward": 1.3846063792705536,
      "reward_std": 0.9288265883922577,
      "rewards/accuracy_reward": 0.45625002551823857,
      "rewards/cosine_scaled_reward": 0.184606320628518,
      "rewards/format_reward": 0.743750037252903,
      "rewards/len_reward": -0.0557435173453996,
      "rewards/reasoning_steps_reward": 0.597916704416275,
      "rewards/tag_count_reward": 0.8087053939700126,
      "step": 350
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 521.9241317749023,
      "epoch": 0.37866666666666665,
      "grad_norm": 2.5071157826124266,
      "kl": 2.10361328125,
      "learning_rate": 2.344650150494596e-06,
      "loss": 0.1433,
      "reward": 1.4640185952186584,
      "reward_std": 0.9381368611007929,
      "rewards/accuracy_reward": 0.49196430630981924,
      "rewards/cosine_scaled_reward": 0.21937568094581367,
      "rewards/format_reward": 0.7526786088943481,
      "rewards/len_reward": -0.05196416466642404,
      "rewards/reasoning_steps_reward": 0.5964286036789417,
      "rewards/tag_count_reward": 0.8064732551574707,
      "step": 355
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 496.58126907348634,
      "epoch": 0.384,
      "grad_norm": 3.557634724061136,
      "kl": 2.18310546875,
      "learning_rate": 2.3214072013095436e-06,
      "loss": 0.0958,
      "reward": 1.468079997599125,
      "reward_std": 0.909414467215538,
      "rewards/accuracy_reward": 0.4651785932481289,
      "rewards/cosine_scaled_reward": 0.22075851147892536,
      "rewards/format_reward": 0.7821428999304771,
      "rewards/len_reward": -0.054533878527581695,
      "rewards/reasoning_steps_reward": 0.577380983531475,
      "rewards/tag_count_reward": 0.8406250387430191,
      "step": 360
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 542.5785934448243,
      "epoch": 0.3893333333333333,
      "grad_norm": 3.8879973791198053,
      "kl": 3.174609375,
      "learning_rate": 2.2978790645857867e-06,
      "loss": 0.2324,
      "reward": 1.0243786200881004,
      "reward_std": 0.8560890629887581,
      "rewards/accuracy_reward": 0.28839287189766766,
      "rewards/cosine_scaled_reward": 0.08598572693299503,
      "rewards/format_reward": 0.6500000312924386,
      "rewards/len_reward": -0.08332871415186674,
      "rewards/reasoning_steps_reward": 0.5497024253010749,
      "rewards/tag_count_reward": 0.7598214596509933,
      "step": 365
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 502.62145309448243,
      "epoch": 0.39466666666666667,
      "grad_norm": 3.506874575960076,
      "kl": 1.89091796875,
      "learning_rate": 2.274073909147986e-06,
      "loss": 0.1493,
      "reward": 1.2348149195313454,
      "reward_std": 0.9403831019997597,
      "rewards/accuracy_reward": 0.3812500167638063,
      "rewards/cosine_scaled_reward": 0.1330291626858525,
      "rewards/format_reward": 0.7205357491970062,
      "rewards/len_reward": -0.07397339158924296,
      "rewards/reasoning_steps_reward": 0.5413690857589245,
      "rewards/tag_count_reward": 0.7933036088943481,
      "step": 370
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 516.9330604553222,
      "epoch": 0.4,
      "grad_norm": 2.9176316838758143,
      "kl": 3.01181640625,
      "learning_rate": 2.25e-06,
      "loss": 0.2186,
      "reward": 1.2647677563130855,
      "reward_std": 0.951485538482666,
      "rewards/accuracy_reward": 0.3955357325263321,
      "rewards/cosine_scaled_reward": 0.1781605551484972,
      "rewards/format_reward": 0.6910714544355869,
      "rewards/len_reward": -0.058914582757279274,
      "rewards/reasoning_steps_reward": 0.5461309857666492,
      "rewards/tag_count_reward": 0.7575893193483353,
      "step": 375
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 511.66430816650393,
      "epoch": 0.4053333333333333,
      "grad_norm": 2.696075434837073,
      "kl": 2.28759765625,
      "learning_rate": 2.225665695455325e-06,
      "loss": 0.1772,
      "reward": 1.3005409233272076,
      "reward_std": 0.9499507501721383,
      "rewards/accuracy_reward": 0.4178571649827063,
      "rewards/cosine_scaled_reward": 0.15232657552696766,
      "rewards/format_reward": 0.7303571812808514,
      "rewards/len_reward": -0.05240586331638042,
      "rewards/reasoning_steps_reward": 0.5726190879940987,
      "rewards/tag_count_reward": 0.8000000312924385,
      "step": 380
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 538.5303825378418,
      "epoch": 0.4106666666666667,
      "grad_norm": 2.894911697402106,
      "kl": 2.70791015625,
      "learning_rate": 2.20107944423514e-06,
      "loss": 0.1687,
      "reward": 1.2143270827829837,
      "reward_std": 0.9277090534567833,
      "rewards/accuracy_reward": 0.39285716190934183,
      "rewards/cosine_scaled_reward": 0.1196841929873699,
      "rewards/format_reward": 0.701785746961832,
      "rewards/len_reward": -0.08161826087161898,
      "rewards/reasoning_steps_reward": 0.6038690775632858,
      "rewards/tag_count_reward": 0.7930803939700126,
      "step": 385
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 530.4901977539063,
      "epoch": 0.416,
      "grad_norm": 2.1893043577623335,
      "kl": 1.98134765625,
      "learning_rate": 2.1762497825349665e-06,
      "loss": 0.1139,
      "reward": 1.451311932504177,
      "reward_std": 0.8446490153670311,
      "rewards/accuracy_reward": 0.4669643074274063,
      "rewards/cosine_scaled_reward": 0.16202617147937418,
      "rewards/format_reward": 0.8223214730620384,
      "rewards/len_reward": -0.049353477614931765,
      "rewards/reasoning_steps_reward": 0.6681548073887825,
      "rewards/tag_count_reward": 0.8658482521772385,
      "step": 390
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 515.6937721252441,
      "epoch": 0.42133333333333334,
      "grad_norm": 2.108843718460042,
      "kl": 1.975244140625,
      "learning_rate": 2.1511853310609467e-06,
      "loss": 0.1375,
      "reward": 1.566932386159897,
      "reward_std": 0.796729838848114,
      "rewards/accuracy_reward": 0.5142857372760773,
      "rewards/cosine_scaled_reward": 0.20353946691611782,
      "rewards/format_reward": 0.8491071820259094,
      "rewards/len_reward": -0.04666417014668696,
      "rewards/reasoning_steps_reward": 0.6708333812654018,
      "rewards/tag_count_reward": 0.894196467101574,
      "step": 395
    },
    {
      "epoch": 0.4266666666666667,
      "grad_norm": 2.6528834721585386,
      "learning_rate": 2.1258947920367943e-06,
      "loss": 0.142,
      "step": 400
    },
    {
      "epoch": 0.4266666666666667,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 549.1377075195312,
      "eval_kl": 2.1953125,
      "eval_loss": 0.14695748686790466,
      "eval_reward": 1.5166082382202148,
      "eval_reward_std": 0.9115615248680115,
      "eval_rewards/accuracy_reward": 0.5214285969734191,
      "eval_rewards/cosine_scaled_reward": 0.19160813689231873,
      "eval_rewards/format_reward": 0.8035714626312256,
      "eval_rewards/len_reward": 0.03922285344451666,
      "eval_rewards/reasoning_steps_reward": 0.6208333611488343,
      "eval_rewards/tag_count_reward": 0.8433036088943482,
      "eval_runtime": 119.8458,
      "eval_samples_per_second": 0.634,
      "eval_steps_per_second": 0.008,
      "step": 400
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 561.7768123626709,
      "epoch": 0.432,
      "grad_norm": 1.8155192088908143,
      "kl": 2.64248046875,
      "learning_rate": 2.100386946182431e-06,
      "loss": 0.1025,
      "reward": 1.4179199129343032,
      "reward_std": 0.8790018357336521,
      "rewards/accuracy_reward": 0.4616071647964418,
      "rewards/cosine_scaled_reward": 0.18131270299782046,
      "rewards/format_reward": 0.7750000402331352,
      "rewards/len_reward": -0.05932185428849053,
      "rewards/reasoning_steps_reward": 0.6251488503068685,
      "rewards/tag_count_reward": 0.8315848618745804,
      "step": 405
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 577.8580657958985,
      "epoch": 0.43733333333333335,
      "grad_norm": 6.06476868467511,
      "kl": 2.2408203125,
      "learning_rate": 2.0746706496653765e-06,
      "loss": 0.1245,
      "reward": 1.3582762971520423,
      "reward_std": 0.8962713375687599,
      "rewards/accuracy_reward": 0.43214288000017403,
      "rewards/cosine_scaled_reward": 0.1823833931237459,
      "rewards/format_reward": 0.743750037252903,
      "rewards/len_reward": -0.06492970008366683,
      "rewards/reasoning_steps_reward": 0.6160714574158191,
      "rewards/tag_count_reward": 0.8098214656114578,
      "step": 410
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 607.6991371154785,
      "epoch": 0.44266666666666665,
      "grad_norm": 3.369320135705493,
      "kl": 2.0812744140625,
      "learning_rate": 2.048754831025942e-06,
      "loss": 0.1126,
      "reward": 1.353794051706791,
      "reward_std": 0.8936590984463691,
      "rewards/accuracy_reward": 0.4241071607917547,
      "rewards/cosine_scaled_reward": 0.16986540658399463,
      "rewards/format_reward": 0.7598214641213417,
      "rewards/len_reward": -0.07249546247767284,
      "rewards/reasoning_steps_reward": 0.6014881297945976,
      "rewards/tag_count_reward": 0.8178571820259094,
      "step": 415
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 649.4482482910156,
      "epoch": 0.448,
      "grad_norm": 3.4251131757872786,
      "kl": 1.88330078125,
      "learning_rate": 2.0226484880772943e-06,
      "loss": 0.0423,
      "reward": 1.4434038519859314,
      "reward_std": 0.8622494846582412,
      "rewards/accuracy_reward": 0.4553571665659547,
      "rewards/cosine_scaled_reward": 0.18536810133846301,
      "rewards/format_reward": 0.8026786103844643,
      "rewards/len_reward": -0.0708045683815726,
      "rewards/reasoning_steps_reward": 0.615773843973875,
      "rewards/tag_count_reward": 0.8500000402331352,
      "step": 420
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 646.4428848266601,
      "epoch": 0.4533333333333333,
      "grad_norm": 3.9373036793479965,
      "kl": 1.9150634765625,
      "learning_rate": 1.9963606847814702e-06,
      "loss": 0.0719,
      "reward": 1.4559296056628228,
      "reward_std": 0.8618416309356689,
      "rewards/accuracy_reward": 0.4446428772062063,
      "rewards/cosine_scaled_reward": 0.20235811360180378,
      "rewards/format_reward": 0.8089286148548126,
      "rewards/len_reward": -0.0718243672628887,
      "rewards/reasoning_steps_reward": 0.648511940985918,
      "rewards/tag_count_reward": 0.861830398440361,
      "step": 425
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 625.8178810119629,
      "epoch": 0.45866666666666667,
      "grad_norm": 1.5199869338814753,
      "kl": 1.849365234375,
      "learning_rate": 1.9699005481024273e-06,
      "loss": 0.0543,
      "reward": 1.5006039649248124,
      "reward_std": 0.8628779232501984,
      "rewards/accuracy_reward": 0.4571428752504289,
      "rewards/cosine_scaled_reward": 0.25953247884754094,
      "rewards/format_reward": 0.7839286133646965,
      "rewards/len_reward": -0.06937549803406,
      "rewards/reasoning_steps_reward": 0.6339286088943481,
      "rewards/tag_count_reward": 0.8537946835160255,
      "step": 430
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 628.6714561462402,
      "epoch": 0.464,
      "grad_norm": 3.1795113531999535,
      "kl": 1.884423828125,
      "learning_rate": 1.943277264837214e-06,
      "loss": 0.0435,
      "reward": 1.4665423482656479,
      "reward_std": 0.8692625299096107,
      "rewards/accuracy_reward": 0.45803573401644826,
      "rewards/cosine_scaled_reward": 0.22100658505223691,
      "rewards/format_reward": 0.7875000417232514,
      "rewards/len_reward": -0.08461722731226473,
      "rewards/reasoning_steps_reward": 0.6083333678543568,
      "rewards/tag_count_reward": 0.859375037252903,
      "step": 435
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 650.9866348266602,
      "epoch": 0.4693333333333333,
      "grad_norm": 2.3194432630180817,
      "kl": 2.84091796875,
      "learning_rate": 1.9165000784263734e-06,
      "loss": 0.0896,
      "reward": 1.4741526886820793,
      "reward_std": 0.9216743379831314,
      "rewards/accuracy_reward": 0.4535714493133128,
      "rewards/cosine_scaled_reward": 0.2571883347816765,
      "rewards/format_reward": 0.7633929014205932,
      "rewards/len_reward": -0.06349783511832356,
      "rewards/reasoning_steps_reward": 0.5604166954755783,
      "rewards/tag_count_reward": 0.822991105914116,
      "step": 440
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 643.0214553833008,
      "epoch": 0.4746666666666667,
      "grad_norm": 3.2555576351928543,
      "kl": 1.411083984375,
      "learning_rate": 1.8895782857446754e-06,
      "loss": 0.0528,
      "reward": 1.5955371528863906,
      "reward_std": 0.8531403854489327,
      "rewards/accuracy_reward": 0.5044643089175225,
      "rewards/cosine_scaled_reward": 0.26964423903264106,
      "rewards/format_reward": 0.8214286103844642,
      "rewards/len_reward": -0.05830043089517858,
      "rewards/reasoning_steps_reward": 0.5845238514244556,
      "rewards/tag_count_reward": 0.8455357491970062,
      "step": 445
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 645.4437805175781,
      "epoch": 0.48,
      "grad_norm": 3.0639176041989487,
      "kl": 1.577734375,
      "learning_rate": 1.8625212338733005e-06,
      "loss": 0.0851,
      "reward": 1.6924126744270325,
      "reward_std": 0.8420305147767067,
      "rewards/accuracy_reward": 0.5589285969734192,
      "rewards/cosine_scaled_reward": 0.3040197363588959,
      "rewards/format_reward": 0.8294643297791481,
      "rewards/len_reward": -0.04013834063371178,
      "rewards/reasoning_steps_reward": 0.549404788017273,
      "rewards/tag_count_reward": 0.8446428969502449,
      "step": 450
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 629.6714530944824,
      "epoch": 0.48533333333333334,
      "grad_norm": 5.415079728456791,
      "kl": 1.687841796875,
      "learning_rate": 1.835338316854588e-06,
      "loss": 0.1255,
      "reward": 1.7355066567659378,
      "reward_std": 0.8314376153051853,
      "rewards/accuracy_reward": 0.5687500245869159,
      "rewards/cosine_scaled_reward": 0.32300658551976086,
      "rewards/format_reward": 0.8437500402331353,
      "rewards/len_reward": -0.03221394362917636,
      "rewards/reasoning_steps_reward": 0.5708333767950535,
      "rewards/tag_count_reward": 0.8620536133646965,
      "step": 455
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 620.2669906616211,
      "epoch": 0.49066666666666664,
      "grad_norm": 3.3718675514560483,
      "kl": 1.591455078125,
      "learning_rate": 1.8080389724304863e-06,
      "loss": 0.0807,
      "reward": 1.7503222301602364,
      "reward_std": 0.8645094059407711,
      "rewards/accuracy_reward": 0.5830357387661934,
      "rewards/cosine_scaled_reward": 0.3387150165159255,
      "rewards/format_reward": 0.8285714715719223,
      "rewards/len_reward": -0.03689943939389195,
      "rewards/reasoning_steps_reward": 0.551190510392189,
      "rewards/tag_count_reward": 0.8562500432133675,
      "step": 460
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 678.2982467651367,
      "epoch": 0.496,
      "grad_norm": 1.4238603128331906,
      "kl": 1.46229248046875,
      "learning_rate": 1.7806326787658219e-06,
      "loss": 0.0696,
      "reward": 1.7878277271986007,
      "reward_std": 0.8226681739091873,
      "rewards/accuracy_reward": 0.6151786014437676,
      "rewards/cosine_scaled_reward": 0.33604192789644005,
      "rewards/format_reward": 0.8366071790456772,
      "rewards/len_reward": -0.04087298092490528,
      "rewards/reasoning_steps_reward": 0.551190511137247,
      "rewards/tag_count_reward": 0.8636161163449287,
      "step": 465
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 713.3116371154786,
      "epoch": 0.5013333333333333,
      "grad_norm": 2.222503543302329,
      "kl": 1.5125,
      "learning_rate": 1.7531289511575427e-06,
      "loss": 0.0596,
      "reward": 1.6392380952835084,
      "reward_std": 0.8160237230360508,
      "rewards/accuracy_reward": 0.5241071671247483,
      "rewards/cosine_scaled_reward": 0.2669166113249958,
      "rewards/format_reward": 0.8482143312692643,
      "rewards/len_reward": -0.06106900977320038,
      "rewards/reasoning_steps_reward": 0.5258928924798966,
      "rewards/tag_count_reward": 0.875669677555561,
      "step": 470
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 672.4080635070801,
      "epoch": 0.5066666666666667,
      "grad_norm": 3.1430403238439366,
      "kl": 2.708740234375,
      "learning_rate": 1.7255373387310633e-06,
      "loss": 0.0805,
      "reward": 1.5507970213890077,
      "reward_std": 0.8910624407231807,
      "rewards/accuracy_reward": 0.49107144922018053,
      "rewards/cosine_scaled_reward": 0.26954693003499414,
      "rewards/format_reward": 0.7901786148548127,
      "rewards/len_reward": -0.05758061000378802,
      "rewards/reasoning_steps_reward": 0.5047619365155697,
      "rewards/tag_count_reward": 0.83482146859169,
      "step": 475
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 680.8928863525391,
      "epoch": 0.512,
      "grad_norm": 1.2963031388346116,
      "kl": 1.945947265625,
      "learning_rate": 1.6978674211248676e-06,
      "loss": 0.0875,
      "reward": 1.604718978703022,
      "reward_std": 0.8831877142190934,
      "rewards/accuracy_reward": 0.49553573653101923,
      "rewards/cosine_scaled_reward": 0.2832903404778335,
      "rewards/format_reward": 0.8258928969502449,
      "rewards/len_reward": -0.06320558707229793,
      "rewards/reasoning_steps_reward": 0.48422621935606003,
      "rewards/tag_count_reward": 0.8361607521772385,
      "step": 480
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 677.7125305175781,
      "epoch": 0.5173333333333333,
      "grad_norm": 1.0370713662213589,
      "kl": 1.35882568359375,
      "learning_rate": 1.6701288051645182e-06,
      "loss": 0.0922,
      "reward": 1.792074230313301,
      "reward_std": 0.79517278149724,
      "rewards/accuracy_reward": 0.5866071701049804,
      "rewards/cosine_scaled_reward": 0.3420741647016257,
      "rewards/format_reward": 0.8633928969502449,
      "rewards/len_reward": -0.03288009662137483,
      "rewards/reasoning_steps_reward": 0.4857143238186836,
      "rewards/tag_count_reward": 0.8662946850061417,
      "step": 485
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 676.1411018371582,
      "epoch": 0.5226666666666666,
      "grad_norm": 1.2794408915109077,
      "kl": 1.83466796875,
      "learning_rate": 1.642331121527223e-06,
      "loss": 0.0709,
      "reward": 1.6959030866622924,
      "reward_std": 0.8808967053890229,
      "rewards/accuracy_reward": 0.5696428790688515,
      "rewards/cosine_scaled_reward": 0.29679586752317844,
      "rewards/format_reward": 0.8294643342494965,
      "rewards/len_reward": -0.05175146399415098,
      "rewards/reasoning_steps_reward": 0.47678574323654177,
      "rewards/tag_count_reward": 0.8544643208384514,
      "step": 490
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 685.1571792602539,
      "epoch": 0.528,
      "grad_norm": 2.014915380678631,
      "kl": 1.9997802734375,
      "learning_rate": 1.6144840213981257e-06,
      "loss": 0.0355,
      "reward": 1.5696246579289437,
      "reward_std": 0.8558191373944283,
      "rewards/accuracy_reward": 0.508928595483303,
      "rewards/cosine_scaled_reward": 0.2562317634001374,
      "rewards/format_reward": 0.8044643253087997,
      "rewards/len_reward": -0.07657566228881478,
      "rewards/reasoning_steps_reward": 0.48779764398932457,
      "rewards/tag_count_reward": 0.846651828289032,
      "step": 495
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 1.4151878603780967,
      "learning_rate": 1.5865971731194738e-06,
      "loss": 0.0949,
      "step": 500
    },
    {
      "epoch": 0.5333333333333333,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 705.310546875,
      "eval_kl": 3.2875,
      "eval_loss": 0.10477960109710693,
      "eval_reward": 1.4186731815338134,
      "eval_reward_std": 0.9728380680084229,
      "eval_rewards/accuracy_reward": 0.47500001788139345,
      "eval_rewards/cosine_scaled_reward": 0.2133159816265106,
      "eval_rewards/format_reward": 0.7303571701049805,
      "eval_rewards/len_reward": -0.04307665973901749,
      "eval_rewards/reasoning_steps_reward": 0.46190478801727297,
      "eval_rewards/tag_count_reward": 0.7973214626312256,
      "eval_runtime": 123.234,
      "eval_samples_per_second": 0.617,
      "eval_steps_per_second": 0.008,
      "step": 500
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 709.8607475280762,
      "epoch": 0.5386666666666666,
      "grad_norm": 64.5985203796494,
      "kl": 2.5929443359375,
      "learning_rate": 1.5586802588338262e-06,
      "loss": 0.094,
      "reward": 1.4848765015602112,
      "reward_std": 0.9199745565652847,
      "rewards/accuracy_reward": 0.4910714510828257,
      "rewards/cosine_scaled_reward": 0.22773357991827653,
      "rewards/format_reward": 0.7660714708268642,
      "rewards/len_reward": -0.07428886514389887,
      "rewards/reasoning_steps_reward": 0.4977678868919611,
      "rewards/tag_count_reward": 0.8183036133646965,
      "step": 505
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 710.7768203735352,
      "epoch": 0.544,
      "grad_norm": 1.0652888427475398,
      "kl": 1.95126953125,
      "learning_rate": 1.5307429711224756e-06,
      "loss": 0.0823,
      "reward": 1.6024441167712211,
      "reward_std": 0.9890345253050328,
      "rewards/accuracy_reward": 0.550000024959445,
      "rewards/cosine_scaled_reward": 0.26494405455887315,
      "rewards/format_reward": 0.7875000387430191,
      "rewards/len_reward": -0.0605491196794901,
      "rewards/reasoning_steps_reward": 0.5348214656114578,
      "rewards/tag_count_reward": 0.8069196820259095,
      "step": 510
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 693.4411056518554,
      "epoch": 0.5493333333333333,
      "grad_norm": 3.159918602831137,
      "kl": 2.3521484375,
      "learning_rate": 1.5027950096402447e-06,
      "loss": 0.0542,
      "reward": 1.5701138600707054,
      "reward_std": 0.8735249437391758,
      "rewards/accuracy_reward": 0.5089285969734192,
      "rewards/cosine_scaled_reward": 0.27189951852487865,
      "rewards/format_reward": 0.7892857536673545,
      "rewards/len_reward": -0.07036974684160668,
      "rewards/reasoning_steps_reward": 0.5255952708423137,
      "rewards/tag_count_reward": 0.793303607404232,
      "step": 515
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 716.8518157958985,
      "epoch": 0.5546666666666666,
      "grad_norm": 1.361987334147322,
      "kl": 2.1235595703125,
      "learning_rate": 1.474846077747821e-06,
      "loss": 0.0603,
      "reward": 1.6362058356404305,
      "reward_std": 0.8912258110940456,
      "rewards/accuracy_reward": 0.5241071671247483,
      "rewards/cosine_scaled_reward": 0.28977720933035017,
      "rewards/format_reward": 0.8223214656114578,
      "rewards/len_reward": -0.06673234046902507,
      "rewards/reasoning_steps_reward": 0.5232143193483353,
      "rewards/tag_count_reward": 0.8129464626312256,
      "step": 520
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 696.7134223937989,
      "epoch": 0.56,
      "grad_norm": 2.896870550734319,
      "kl": 2.0576904296875,
      "learning_rate": 1.4469058791428154e-06,
      "loss": 0.0856,
      "reward": 1.648724715411663,
      "reward_std": 0.8711829155683517,
      "rewards/accuracy_reward": 0.5375000245869159,
      "rewards/cosine_scaled_reward": 0.2808674838131992,
      "rewards/format_reward": 0.8303571864962578,
      "rewards/len_reward": -0.05326956428325502,
      "rewards/reasoning_steps_reward": 0.48779765255749225,
      "rewards/tag_count_reward": 0.8171875372529029,
      "step": 525
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 729.4321807861328,
      "epoch": 0.5653333333333334,
      "grad_norm": 2.0390779532818843,
      "kl": 1.3226318359375,
      "learning_rate": 1.4189841144906928e-06,
      "loss": 0.0481,
      "reward": 1.725620111823082,
      "reward_std": 0.8606425553560257,
      "rewards/accuracy_reward": 0.5705357432365418,
      "rewards/cosine_scaled_reward": 0.3131200488656759,
      "rewards/format_reward": 0.8419643253087997,
      "rewards/len_reward": -0.06525407417793758,
      "rewards/reasoning_steps_reward": 0.4889881335198879,
      "rewards/tag_count_reward": 0.8185268193483353,
      "step": 530
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 694.4687828063965,
      "epoch": 0.5706666666666667,
      "grad_norm": 3.2275030286756503,
      "kl": 1.80302734375,
      "learning_rate": 1.3910904780567642e-06,
      "loss": 0.0862,
      "reward": 1.5835659071803092,
      "reward_std": 0.9309476897120476,
      "rewards/accuracy_reward": 0.5133928820490837,
      "rewards/cosine_scaled_reward": 0.2826729838096071,
      "rewards/format_reward": 0.7875000402331352,
      "rewards/len_reward": -0.05334139431070071,
      "rewards/reasoning_steps_reward": 0.4339285997673869,
      "rewards/tag_count_reward": 0.7857143238186837,
      "step": 535
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 674.718782043457,
      "epoch": 0.576,
      "grad_norm": 2.445582670371733,
      "kl": 2.04580078125,
      "learning_rate": 1.3632346543403946e-06,
      "loss": 0.0442,
      "reward": 1.7607878595590591,
      "reward_std": 0.8948764257133007,
      "rewards/accuracy_reward": 0.5901785979047418,
      "rewards/cosine_scaled_reward": 0.3402521046809852,
      "rewards/format_reward": 0.8303571909666061,
      "rewards/len_reward": -0.0463905618264107,
      "rewards/reasoning_steps_reward": 0.47976194024086,
      "rewards/tag_count_reward": 0.8104911088943482,
      "step": 540
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 672.7214630126953,
      "epoch": 0.5813333333333334,
      "grad_norm": 2.105824534589838,
      "kl": 1.8478271484375,
      "learning_rate": 1.335426314712607e-06,
      "loss": 0.0705,
      "reward": 1.643703442811966,
      "reward_std": 0.8904561165720224,
      "rewards/accuracy_reward": 0.541071455925703,
      "rewards/cosine_scaled_reward": 0.3079891032539308,
      "rewards/format_reward": 0.7946428984403611,
      "rewards/len_reward": -0.05454629746964201,
      "rewards/reasoning_steps_reward": 0.4824405059218407,
      "rewards/tag_count_reward": 0.78794646859169,
      "step": 545
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 661.1991333007812,
      "epoch": 0.5866666666666667,
      "grad_norm": 2.7373033343485553,
      "kl": 2.600390625,
      "learning_rate": 1.3076751140582396e-06,
      "loss": 0.0498,
      "reward": 1.5583188071846963,
      "reward_std": 0.9974759891629219,
      "rewards/accuracy_reward": 0.5366071701049805,
      "rewards/cosine_scaled_reward": 0.2967116108164191,
      "rewards/format_reward": 0.725000036507845,
      "rewards/len_reward": -0.06344892352935858,
      "rewards/reasoning_steps_reward": 0.46160717457532885,
      "rewards/tag_count_reward": 0.7638393208384514,
      "step": 550
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 665.4652076721192,
      "epoch": 0.592,
      "grad_norm": 1.1852476234067224,
      "kl": 1.458251953125,
      "learning_rate": 1.2799906874238297e-06,
      "loss": 0.0173,
      "reward": 1.7244828373193741,
      "reward_std": 0.8695188030600548,
      "rewards/accuracy_reward": 0.5767857413738966,
      "rewards/cosine_scaled_reward": 0.3334113570395857,
      "rewards/format_reward": 0.8142857536673546,
      "rewards/len_reward": -0.060599689168157056,
      "rewards/reasoning_steps_reward": 0.5214286014437676,
      "rewards/tag_count_reward": 0.7852678894996643,
      "step": 555
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 669.0857482910156,
      "epoch": 0.5973333333333334,
      "grad_norm": 2.914690295232921,
      "kl": 1.308984375,
      "learning_rate": 1.2523826466723843e-06,
      "loss": 0.0526,
      "reward": 1.798653519153595,
      "reward_std": 0.7890060037374497,
      "rewards/accuracy_reward": 0.5723214579746128,
      "rewards/cosine_scaled_reward": 0.35133201819844545,
      "rewards/format_reward": 0.8750000417232513,
      "rewards/len_reward": -0.050365234701894225,
      "rewards/reasoning_steps_reward": 0.561607176065445,
      "rewards/tag_count_reward": 0.8348214671015739,
      "step": 560
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 663.8428840637207,
      "epoch": 0.6026666666666667,
      "grad_norm": 3.266880748687459,
      "kl": 3.2771484375,
      "learning_rate": 1.2248605771462016e-06,
      "loss": 0.0748,
      "reward": 1.5902897894382477,
      "reward_std": 0.9371627844870091,
      "rewards/accuracy_reward": 0.5294643044471741,
      "rewards/cosine_scaled_reward": 0.2885039990535006,
      "rewards/format_reward": 0.7723214700818062,
      "rewards/len_reward": -0.07391626983880997,
      "rewards/reasoning_steps_reward": 0.5761905089020729,
      "rewards/tag_count_reward": 0.7772321790456772,
      "step": 565
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 649.5839538574219,
      "epoch": 0.608,
      "grad_norm": 1.1323574314355938,
      "kl": 1.91318359375,
      "learning_rate": 1.1974340343388974e-06,
      "loss": 0.0716,
      "reward": 1.7778286993503571,
      "reward_std": 0.8691867522895336,
      "rewards/accuracy_reward": 0.6008928859606385,
      "rewards/cosine_scaled_reward": 0.3430071874521673,
      "rewards/format_reward": 0.8339286118745803,
      "rewards/len_reward": -0.05089021619896812,
      "rewards/reasoning_steps_reward": 0.5464286044239998,
      "rewards/tag_count_reward": 0.8194196790456771,
      "step": 570
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 664.774137878418,
      "epoch": 0.6133333333333333,
      "grad_norm": 1.4388282162154258,
      "kl": 1.21708984375,
      "learning_rate": 1.1701125405777965e-06,
      "loss": 0.0413,
      "reward": 1.888391876220703,
      "reward_std": 0.8107489451766015,
      "rewards/accuracy_reward": 0.6410714574158192,
      "rewards/cosine_scaled_reward": 0.37678469233214856,
      "rewards/format_reward": 0.8705357581377029,
      "rewards/len_reward": -0.03641782412887551,
      "rewards/reasoning_steps_reward": 0.6273809865117073,
      "rewards/tag_count_reward": 0.8319196745753288,
      "step": 575
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 668.4446685791015,
      "epoch": 0.6186666666666667,
      "grad_norm": 1.3341419974163,
      "kl": 1.6546875,
      "learning_rate": 1.142905581717841e-06,
      "loss": 0.0385,
      "reward": 1.6953963339328766,
      "reward_std": 0.8444600582122803,
      "rewards/accuracy_reward": 0.5392857410013676,
      "rewards/cosine_scaled_reward": 0.31414626743644475,
      "rewards/format_reward": 0.841964328289032,
      "rewards/len_reward": -0.052902453497517855,
      "rewards/reasoning_steps_reward": 0.5791667021811009,
      "rewards/tag_count_reward": 0.80357146859169,
      "step": 580
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 676.8196731567383,
      "epoch": 0.624,
      "grad_norm": 4.640349038075715,
      "kl": 2.029638671875,
      "learning_rate": 1.1158226038481584e-06,
      "loss": 0.09,
      "reward": 1.795636734366417,
      "reward_std": 0.8532808139920235,
      "rewards/accuracy_reward": 0.6071428872644902,
      "rewards/cosine_scaled_reward": 0.3384937860071659,
      "rewards/format_reward": 0.8500000461935997,
      "rewards/len_reward": -0.044303331701667045,
      "rewards/reasoning_steps_reward": 0.6011905141174794,
      "rewards/tag_count_reward": 0.7720982536673546,
      "step": 585
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 699.8062805175781,
      "epoch": 0.6293333333333333,
      "grad_norm": 1.798885837943298,
      "kl": 2.5541015625,
      "learning_rate": 1.0888730100124355e-06,
      "loss": 0.0852,
      "reward": 1.640977455675602,
      "reward_std": 0.9245925977826118,
      "rewards/accuracy_reward": 0.550892885029316,
      "rewards/cosine_scaled_reward": 0.2829416638240218,
      "rewards/format_reward": 0.8071428954601287,
      "rewards/len_reward": -0.058377658866811545,
      "rewards/reasoning_steps_reward": 0.5770833715796471,
      "rewards/tag_count_reward": 0.7515625357627869,
      "step": 590
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 699.8830703735351,
      "epoch": 0.6346666666666667,
      "grad_norm": 1.4180247445536809,
      "kl": 2.21064453125,
      "learning_rate": 1.062066156944242e-06,
      "loss": 0.0715,
      "reward": 1.5858480513095856,
      "reward_std": 0.9440520867705345,
      "rewards/accuracy_reward": 0.5116071663796902,
      "rewards/cosine_scaled_reward": 0.28941939915530385,
      "rewards/format_reward": 0.7848214745521546,
      "rewards/len_reward": -0.06778505290567409,
      "rewards/reasoning_steps_reward": 0.5654762282967567,
      "rewards/tag_count_reward": 0.7723214596509933,
      "step": 595
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5382009317793046,
      "learning_rate": 1.0354113518184304e-06,
      "loss": 0.0362,
      "step": 600
    },
    {
      "epoch": 0.64,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 692.8052856445313,
      "eval_kl": 2.0015625,
      "eval_loss": 0.02467462234199047,
      "eval_reward": 1.579683256149292,
      "eval_reward_std": 0.9282516837120056,
      "eval_rewards/accuracy_reward": 0.5410714447498322,
      "eval_rewards/cosine_scaled_reward": 0.28146889209747317,
      "eval_rewards/format_reward": 0.7571428894996644,
      "eval_rewards/len_reward": -0.033683009818196295,
      "eval_rewards/reasoning_steps_reward": 0.5428571820259094,
      "eval_rewards/tag_count_reward": 0.7959821939468383,
      "eval_runtime": 123.4097,
      "eval_samples_per_second": 0.616,
      "eval_steps_per_second": 0.008,
      "step": 600
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 698.9357452392578,
      "epoch": 0.6453333333333333,
      "grad_norm": 0.8999801432514335,
      "kl": 2.011328125,
      "learning_rate": 1.008917849019739e-06,
      "loss": 0.0246,
      "reward": 1.6242469638586043,
      "reward_std": 0.9202400345355273,
      "rewards/accuracy_reward": 0.5218750248197466,
      "rewards/cosine_scaled_reward": 0.3135326229268685,
      "rewards/format_reward": 0.7888393275439739,
      "rewards/len_reward": -0.06526912819986137,
      "rewards/reasoning_steps_reward": 0.5854167021811009,
      "rewards/tag_count_reward": 0.8075893238186836,
      "step": 605
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 750.725926208496,
      "epoch": 0.6506666666666666,
      "grad_norm": 2.1187107981918802,
      "kl": 1.882958984375,
      "learning_rate": 9.825948469297303e-07,
      "loss": 0.0314,
      "reward": 1.64074095338583,
      "reward_std": 0.9081449903547764,
      "rewards/accuracy_reward": 0.5330357388593256,
      "rewards/cosine_scaled_reward": 0.3068123304285109,
      "rewards/format_reward": 0.8008928969502449,
      "rewards/len_reward": -0.06153649939224124,
      "rewards/reasoning_steps_reward": 0.6250000432133674,
      "rewards/tag_count_reward": 0.8250000357627869,
      "step": 610
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 748.2152130126954,
      "epoch": 0.656,
      "grad_norm": 2.131590013783802,
      "kl": 1.9369140625,
      "learning_rate": 9.564514847331647e-07,
      "loss": 0.0395,
      "reward": 1.6662242144346238,
      "reward_std": 0.8884774453938007,
      "rewards/accuracy_reward": 0.5437500268220902,
      "rewards/cosine_scaled_reward": 0.30372417061589657,
      "rewards/format_reward": 0.8187500432133674,
      "rewards/len_reward": -0.07084276499663247,
      "rewards/reasoning_steps_reward": 0.586309564858675,
      "rewards/tag_count_reward": 0.8140625357627869,
      "step": 615
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 722.0428909301758,
      "epoch": 0.6613333333333333,
      "grad_norm": 0.947846170528357,
      "kl": 1.95106201171875,
      "learning_rate": 9.304968392449361e-07,
      "loss": 0.0166,
      "reward": 1.6543820381164551,
      "reward_std": 0.8385958023369312,
      "rewards/accuracy_reward": 0.5223214563913643,
      "rewards/cosine_scaled_reward": 0.3061676881741732,
      "rewards/format_reward": 0.8258928969502449,
      "rewards/len_reward": -0.07246997265610844,
      "rewards/reasoning_steps_reward": 0.5738095574080944,
      "rewards/tag_count_reward": 0.8107143238186836,
      "step": 620
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 716.7339599609375,
      "epoch": 0.6666666666666666,
      "grad_norm": 3.0578518576987914,
      "kl": 1.99375,
      "learning_rate": 9.047399217586552e-07,
      "loss": 0.0515,
      "reward": 1.6329590931534768,
      "reward_std": 0.8712823033332825,
      "rewards/accuracy_reward": 0.4964285962283611,
      "rewards/cosine_scaled_reward": 0.32670902004465463,
      "rewards/format_reward": 0.8098214730620384,
      "rewards/len_reward": -0.05736397755099461,
      "rewards/reasoning_steps_reward": 0.5723214700818062,
      "rewards/tag_count_reward": 0.7941964656114578,
      "step": 625
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 768.658071899414,
      "epoch": 0.672,
      "grad_norm": 1.9045908740529087,
      "kl": 1.85390625,
      "learning_rate": 8.791896749179831e-07,
      "loss": 0.0319,
      "reward": 1.545591439306736,
      "reward_std": 0.9249462381005287,
      "rewards/accuracy_reward": 0.49910716637969016,
      "rewards/cosine_scaled_reward": 0.2473770938348025,
      "rewards/format_reward": 0.7991071864962578,
      "rewards/len_reward": -0.09455443265615031,
      "rewards/reasoning_steps_reward": 0.5520833678543567,
      "rewards/tag_count_reward": 0.8082589685916901,
      "step": 630
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 764.3089599609375,
      "epoch": 0.6773333333333333,
      "grad_norm": 1.489830070872437,
      "kl": 1.666015625,
      "learning_rate": 8.538549696118023e-07,
      "loss": 0.035,
      "reward": 1.6950147479772568,
      "reward_std": 0.8494538977742195,
      "rewards/accuracy_reward": 0.5785714577883482,
      "rewards/cosine_scaled_reward": 0.28787180441431703,
      "rewards/format_reward": 0.8285714700818062,
      "rewards/len_reward": -0.07843075728160329,
      "rewards/reasoning_steps_reward": 0.5327381260693074,
      "rewards/tag_count_reward": 0.825446467101574,
      "step": 635
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 771.7393203735352,
      "epoch": 0.6826666666666666,
      "grad_norm": 1.408555863280269,
      "kl": 1.8015625,
      "learning_rate": 8.287446018942973e-07,
      "loss": 0.0166,
      "reward": 1.6653958648443221,
      "reward_std": 0.8103255234658718,
      "rewards/accuracy_reward": 0.5616071740165353,
      "rewards/cosine_scaled_reward": 0.2725386595353484,
      "rewards/format_reward": 0.8312500461935997,
      "rewards/len_reward": -0.08096674757543951,
      "rewards/reasoning_steps_reward": 0.5892857603728772,
      "rewards/tag_count_reward": 0.8274553939700127,
      "step": 640
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 744.9634246826172,
      "epoch": 0.688,
      "grad_norm": 3.237395884870554,
      "kl": 1.91787109375,
      "learning_rate": 8.038672899310176e-07,
      "loss": 0.0597,
      "reward": 1.6257297798991204,
      "reward_std": 0.8151006668806076,
      "rewards/accuracy_reward": 0.5276785969734192,
      "rewards/cosine_scaled_reward": 0.24983684375183657,
      "rewards/format_reward": 0.8482143342494964,
      "rewards/len_reward": -0.07792667767498643,
      "rewards/reasoning_steps_reward": 0.5824405170977116,
      "rewards/tag_count_reward": 0.8095982536673546,
      "step": 645
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 734.9598541259766,
      "epoch": 0.6933333333333334,
      "grad_norm": 2.539952667963002,
      "kl": 1.850830078125,
      "learning_rate": 7.792316709719875e-07,
      "loss": 0.0642,
      "reward": 1.649840420484543,
      "reward_std": 0.8533309623599052,
      "rewards/accuracy_reward": 0.5375000268220902,
      "rewards/cosine_scaled_reward": 0.28555465580429884,
      "rewards/format_reward": 0.8267857581377029,
      "rewards/len_reward": -0.06069966709183063,
      "rewards/reasoning_steps_reward": 0.5857143230736256,
      "rewards/tag_count_reward": 0.7982143208384513,
      "step": 650
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 723.6411041259765,
      "epoch": 0.6986666666666667,
      "grad_norm": 1.8599505779512118,
      "kl": 1.41396484375,
      "learning_rate": 7.548462983529016e-07,
      "loss": 0.0594,
      "reward": 1.759677691757679,
      "reward_std": 0.8695821583271026,
      "rewards/accuracy_reward": 0.5821428894996643,
      "rewards/cosine_scaled_reward": 0.3382490785326809,
      "rewards/format_reward": 0.839285759627819,
      "rewards/len_reward": -0.05585277818609029,
      "rewards/reasoning_steps_reward": 0.5860119491815567,
      "rewards/tag_count_reward": 0.8180803880095482,
      "step": 655
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 694.9009216308593,
      "epoch": 0.704,
      "grad_norm": 1.356477272292917,
      "kl": 1.7875,
      "learning_rate": 7.307196385254621e-07,
      "loss": 0.0349,
      "reward": 1.6008909553289414,
      "reward_std": 0.9093584820628167,
      "rewards/accuracy_reward": 0.5258928779512644,
      "rewards/cosine_scaled_reward": 0.2687480329768732,
      "rewards/format_reward": 0.8062500432133675,
      "rewards/len_reward": -0.07140515584032983,
      "rewards/reasoning_steps_reward": 0.5979167088866234,
      "rewards/tag_count_reward": 0.7738839656114578,
      "step": 660
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 728.4169921875,
      "epoch": 0.7093333333333334,
      "grad_norm": 2.3455305274577944,
      "kl": 1.65068359375,
      "learning_rate": 7.068600681178772e-07,
      "loss": 0.0147,
      "reward": 1.760453712940216,
      "reward_std": 0.869930399954319,
      "rewards/accuracy_reward": 0.5937500312924385,
      "rewards/cosine_scaled_reward": 0.32741790837608276,
      "rewards/format_reward": 0.8392857626080513,
      "rewards/len_reward": -0.05782133540487848,
      "rewards/reasoning_steps_reward": 0.5970238469541073,
      "rewards/tag_count_reward": 0.8209821820259094,
      "step": 665
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 724.6794952392578,
      "epoch": 0.7146666666666667,
      "grad_norm": 1.4838696193874328,
      "kl": 1.65224609375,
      "learning_rate": 6.832758710265492e-07,
      "loss": 0.0393,
      "reward": 1.6467845141887665,
      "reward_std": 0.8451258420944214,
      "rewards/accuracy_reward": 0.5276785969734192,
      "rewards/cosine_scaled_reward": 0.29232015907764436,
      "rewards/format_reward": 0.8267857551574707,
      "rewards/len_reward": -0.05971484867623076,
      "rewards/reasoning_steps_reward": 0.5928571835160256,
      "rewards/tag_count_reward": 0.799330398440361,
      "step": 670
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 730.0205703735352,
      "epoch": 0.72,
      "grad_norm": 1.7671955478088914,
      "kl": 1.96552734375,
      "learning_rate": 6.599752355399538e-07,
      "loss": 0.0535,
      "reward": 1.6977156594395637,
      "reward_std": 0.8604207873344422,
      "rewards/accuracy_reward": 0.559821455180645,
      "rewards/cosine_scaled_reward": 0.31289418824017046,
      "rewards/format_reward": 0.8250000461935997,
      "rewards/len_reward": -0.06840839698270429,
      "rewards/reasoning_steps_reward": 0.5937500350177288,
      "rewards/tag_count_reward": 0.806919677555561,
      "step": 675
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 701.5803939819336,
      "epoch": 0.7253333333333334,
      "grad_norm": 2.711356467895934,
      "kl": 2.33935546875,
      "learning_rate": 6.369662514957191e-07,
      "loss": 0.0471,
      "reward": 1.6881353616714478,
      "reward_std": 0.9093373231589794,
      "rewards/accuracy_reward": 0.5616071753203868,
      "rewards/cosine_scaled_reward": 0.3033138638362288,
      "rewards/format_reward": 0.8232143267989158,
      "rewards/len_reward": -0.06628427920097693,
      "rewards/reasoning_steps_reward": 0.5904762275516987,
      "rewards/tag_count_reward": 0.7837053939700127,
      "step": 680
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 717.6080688476562,
      "epoch": 0.7306666666666667,
      "grad_norm": 2.3553290067273727,
      "kl": 1.72353515625,
      "learning_rate": 6.142569074718818e-07,
      "loss": 0.0314,
      "reward": 1.6499603599309922,
      "reward_std": 0.8875117138028145,
      "rewards/accuracy_reward": 0.5419643139466643,
      "rewards/cosine_scaled_reward": 0.27763886260800064,
      "rewards/format_reward": 0.8303571879863739,
      "rewards/len_reward": -0.06650189743377269,
      "rewards/reasoning_steps_reward": 0.6178571805357933,
      "rewards/tag_count_reward": 0.7953125342726708,
      "step": 685
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 701.9911003112793,
      "epoch": 0.736,
      "grad_norm": 1.4466892795867712,
      "kl": 1.6891357421875,
      "learning_rate": 5.918550880133018e-07,
      "loss": 0.022,
      "reward": 1.8273341655731201,
      "reward_std": 0.8836588777601719,
      "rewards/accuracy_reward": 0.6133928835391999,
      "rewards/cosine_scaled_reward": 0.37644122838974,
      "rewards/format_reward": 0.8375000417232513,
      "rewards/len_reward": -0.05486658178269863,
      "rewards/reasoning_steps_reward": 0.6154762282967567,
      "rewards/tag_count_reward": 0.7799107506871223,
      "step": 690
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 709.267894744873,
      "epoch": 0.7413333333333333,
      "grad_norm": 1.8058065674655572,
      "kl": 1.5920654296875,
      "learning_rate": 5.697685708941996e-07,
      "loss": 0.0249,
      "reward": 1.6933651700615884,
      "reward_std": 0.8785169780254364,
      "rewards/accuracy_reward": 0.5642857432365418,
      "rewards/cosine_scaled_reward": 0.3067579740891233,
      "rewards/format_reward": 0.8223214775323868,
      "rewards/len_reward": -0.06730828973231837,
      "rewards/reasoning_steps_reward": 0.5997024156153202,
      "rewards/tag_count_reward": 0.7810268208384514,
      "step": 695
    },
    {
      "epoch": 0.7466666666666667,
      "grad_norm": 2.0040117357136964,
      "learning_rate": 5.480050244177573e-07,
      "loss": 0.0404,
      "step": 700
    },
    {
      "epoch": 0.7466666666666667,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 689.7265380859375,
      "eval_kl": 1.8734375,
      "eval_loss": 0.05176318436861038,
      "eval_reward": 1.7034537076950074,
      "eval_reward_std": 0.9170334577560425,
      "eval_rewards/accuracy_reward": 0.5625000417232513,
      "eval_rewards/cosine_scaled_reward": 0.3195250898599625,
      "eval_rewards/format_reward": 0.8214286088943481,
      "eval_rewards/len_reward": -0.014184250682592391,
      "eval_rewards/reasoning_steps_reward": 0.598809564113617,
      "eval_rewards/tag_count_reward": 0.7892857432365418,
      "eval_runtime": 121.9627,
      "eval_samples_per_second": 0.623,
      "eval_steps_per_second": 0.008,
      "step": 700
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 684.2286022186279,
      "epoch": 0.752,
      "grad_norm": 1.7320693073402242,
      "kl": 2.126708984375,
      "learning_rate": 5.265720047537318e-07,
      "loss": 0.0423,
      "reward": 1.5781065180897713,
      "reward_std": 0.9164291121065616,
      "rewards/accuracy_reward": 0.529017882142216,
      "rewards/cosine_scaled_reward": 0.274981448915787,
      "rewards/format_reward": 0.7741071842610836,
      "rewards/len_reward": -0.0748695431771921,
      "rewards/reasoning_steps_reward": 0.6022321883589029,
      "rewards/tag_count_reward": 0.761272357404232,
      "step": 705
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 668.6464576721191,
      "epoch": 0.7573333333333333,
      "grad_norm": 2.851783458060785,
      "kl": 1.795166015625,
      "learning_rate": 5.054769533149999e-07,
      "loss": 0.0354,
      "reward": 1.7714153915643691,
      "reward_std": 0.9144653856754303,
      "rewards/accuracy_reward": 0.5973214589059352,
      "rewards/cosine_scaled_reward": 0.3535581724718213,
      "rewards/format_reward": 0.8205357566475868,
      "rewards/len_reward": -0.057637330750003456,
      "rewards/reasoning_steps_reward": 0.621130996197462,
      "rewards/tag_count_reward": 0.7683036014437675,
      "step": 710
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 708.1009216308594,
      "epoch": 0.7626666666666667,
      "grad_norm": 2.260299938099763,
      "kl": 1.759228515625,
      "learning_rate": 4.847271941739458e-07,
      "loss": 0.0182,
      "reward": 1.6718120127916336,
      "reward_std": 0.8970876529812812,
      "rewards/accuracy_reward": 0.5437500262632966,
      "rewards/cosine_scaled_reward": 0.32538338992744686,
      "rewards/format_reward": 0.8026786133646965,
      "rewards/len_reward": -0.062959678506013,
      "rewards/reasoning_steps_reward": 0.6113095685839653,
      "rewards/tag_count_reward": 0.7883928954601288,
      "step": 715
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 683.4464584350586,
      "epoch": 0.768,
      "grad_norm": 2.074272987776867,
      "kl": 1.8629638671875,
      "learning_rate": 4.643299315195855e-07,
      "loss": 0.0045,
      "reward": 1.6655956655740738,
      "reward_std": 0.9097626186907292,
      "rewards/accuracy_reward": 0.5223214536905288,
      "rewards/cosine_scaled_reward": 0.3227384569123387,
      "rewards/format_reward": 0.820535758137703,
      "rewards/len_reward": -0.06934529592399485,
      "rewards/reasoning_steps_reward": 0.6032738454639912,
      "rewards/tag_count_reward": 0.7993303939700127,
      "step": 720
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 678.5500289916993,
      "epoch": 0.7733333333333333,
      "grad_norm": 1.6540501406095172,
      "kl": 1.699609375,
      "learning_rate": 4.442922471563205e-07,
      "loss": 0.0157,
      "reward": 1.7752037733793258,
      "reward_std": 0.8400040708482266,
      "rewards/accuracy_reward": 0.5580357387661934,
      "rewards/cosine_scaled_reward": 0.3662751629948616,
      "rewards/format_reward": 0.8508929058909416,
      "rewards/len_reward": -0.05521610935102217,
      "rewards/reasoning_steps_reward": 0.6130952782928943,
      "rewards/tag_count_reward": 0.8223214641213417,
      "step": 725
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 725.4411071777344,
      "epoch": 0.7786666666666666,
      "grad_norm": 8.410594372712595,
      "kl": 1.941552734375,
      "learning_rate": 4.24621098045175e-07,
      "loss": 0.0076,
      "reward": 1.7029503554105758,
      "reward_std": 0.8638798981904984,
      "rewards/accuracy_reward": 0.5294643077999354,
      "rewards/cosine_scaled_reward": 0.3404503061901778,
      "rewards/format_reward": 0.8330357566475868,
      "rewards/len_reward": -0.06691470365040005,
      "rewards/reasoning_steps_reward": 0.5958333618938922,
      "rewards/tag_count_reward": 0.8234375402331352,
      "step": 730
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 713.8982528686523,
      "epoch": 0.784,
      "grad_norm": 1.9391448788727794,
      "kl": 1.941015625,
      "learning_rate": 4.053233138883835e-07,
      "loss": 0.0489,
      "reward": 1.77894726395607,
      "reward_std": 0.9000619798898697,
      "rewards/accuracy_reward": 0.5982143111526966,
      "rewards/cosine_scaled_reward": 0.3539472218602896,
      "rewards/format_reward": 0.8267857581377029,
      "rewards/len_reward": -0.05374447105568834,
      "rewards/reasoning_steps_reward": 0.6297619432210922,
      "rewards/tag_count_reward": 0.8029018208384514,
      "step": 735
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 702.009854888916,
      "epoch": 0.7893333333333333,
      "grad_norm": 1.355632286643051,
      "kl": 2.266845703125,
      "learning_rate": 3.864055947581605e-07,
      "loss": 0.0316,
      "reward": 1.6201725795865058,
      "reward_std": 0.8785072147846222,
      "rewards/accuracy_reward": 0.5321428820490837,
      "rewards/cosine_scaled_reward": 0.29070827066898347,
      "rewards/format_reward": 0.7973214685916901,
      "rewards/len_reward": -0.07668145237257704,
      "rewards/reasoning_steps_reward": 0.5699405044317245,
      "rewards/tag_count_reward": 0.8022321805357933,
      "step": 740
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 708.0571716308593,
      "epoch": 0.7946666666666666,
      "grad_norm": 3.3069120908956227,
      "kl": 2.06044921875,
      "learning_rate": 3.6787450877047543e-07,
      "loss": 0.0349,
      "reward": 1.6467604607343673,
      "reward_std": 0.8980094000697136,
      "rewards/accuracy_reward": 0.543750024586916,
      "rewards/cosine_scaled_reward": 0.2985461330041289,
      "rewards/format_reward": 0.8044643297791481,
      "rewards/len_reward": -0.06843005996015564,
      "rewards/reasoning_steps_reward": 0.598214327543974,
      "rewards/tag_count_reward": 0.8011161029338837,
      "step": 745
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 693.400032043457,
      "epoch": 0.8,
      "grad_norm": 1.7291219089700667,
      "kl": 2.141796875,
      "learning_rate": 3.4973648980464454e-07,
      "loss": 0.0398,
      "reward": 1.7774270474910736,
      "reward_std": 0.9566665157675743,
      "rewards/accuracy_reward": 0.6017857499420642,
      "rewards/cosine_scaled_reward": 0.3622484166175127,
      "rewards/format_reward": 0.813392898440361,
      "rewards/len_reward": -0.06579462469671853,
      "rewards/reasoning_steps_reward": 0.6119048051536083,
      "rewards/tag_count_reward": 0.7859375327825546,
      "step": 750
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 697.6071731567383,
      "epoch": 0.8053333333333333,
      "grad_norm": 2.1537011975481626,
      "kl": 2.369970703125,
      "learning_rate": 3.3199783526952656e-07,
      "loss": 0.0285,
      "reward": 1.626606747508049,
      "reward_std": 0.8798683725297451,
      "rewards/accuracy_reward": 0.546428595483303,
      "rewards/cosine_scaled_reward": 0.2810709737765137,
      "rewards/format_reward": 0.7991071850061416,
      "rewards/len_reward": -0.08888059532619082,
      "rewards/reasoning_steps_reward": 0.6297619499266147,
      "rewards/tag_count_reward": 0.7700893223285675,
      "step": 755
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 669.0143157958985,
      "epoch": 0.8106666666666666,
      "grad_norm": 2.365610096938751,
      "kl": 2.06806640625,
      "learning_rate": 3.146647039171002e-07,
      "loss": 0.0252,
      "reward": 1.7266648024320603,
      "reward_std": 0.8433036901056766,
      "rewards/accuracy_reward": 0.5705357395112515,
      "rewards/cosine_scaled_reward": 0.33559329714626074,
      "rewards/format_reward": 0.8205357536673545,
      "rewards/len_reward": -0.058183514599659245,
      "rewards/reasoning_steps_reward": 0.6181547954678536,
      "rewards/tag_count_reward": 0.7935268253087997,
      "step": 760
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 718.244677734375,
      "epoch": 0.816,
      "grad_norm": 2.229073726941259,
      "kl": 1.7294921875,
      "learning_rate": 2.977431137041848e-07,
      "loss": 0.0401,
      "reward": 1.6797782227396965,
      "reward_std": 0.8497840896248817,
      "rewards/accuracy_reward": 0.5312500277534127,
      "rewards/cosine_scaled_reward": 0.32263530092313886,
      "rewards/format_reward": 0.8258928999304771,
      "rewards/len_reward": -0.060991953429766,
      "rewards/reasoning_steps_reward": 0.6047619454562664,
      "rewards/tag_count_reward": 0.8078125342726707,
      "step": 765
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 698.1232482910157,
      "epoch": 0.8213333333333334,
      "grad_norm": 1.619053512727917,
      "kl": 1.646826171875,
      "learning_rate": 2.8123893970304154e-07,
      "loss": 0.0266,
      "reward": 1.7365109011530877,
      "reward_std": 0.8401576727628708,
      "rewards/accuracy_reward": 0.5803571723401546,
      "rewards/cosine_scaled_reward": 0.3257965755648911,
      "rewards/format_reward": 0.8303571820259095,
      "rewards/len_reward": -0.05876578897587024,
      "rewards/reasoning_steps_reward": 0.6479167118668556,
      "rewards/tag_count_reward": 0.8062500417232513,
      "step": 770
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 693.6268112182618,
      "epoch": 0.8266666666666667,
      "grad_norm": 1.9920542202647236,
      "kl": 1.9521728515625,
      "learning_rate": 2.651579120615855e-07,
      "loss": 0.043,
      "reward": 1.746559591591358,
      "reward_std": 0.8663071312010289,
      "rewards/accuracy_reward": 0.584821455925703,
      "rewards/cosine_scaled_reward": 0.3394166727666743,
      "rewards/format_reward": 0.8223214671015739,
      "rewards/len_reward": -0.05387566906865686,
      "rewards/reasoning_steps_reward": 0.6288690894842148,
      "rewards/tag_count_reward": 0.7970982506871224,
      "step": 775
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 703.2910995483398,
      "epoch": 0.832,
      "grad_norm": 1.999916137698683,
      "kl": 1.91943359375,
      "learning_rate": 2.495056140139119e-07,
      "loss": 0.0445,
      "reward": 1.650632557272911,
      "reward_std": 0.812701889872551,
      "rewards/accuracy_reward": 0.5187500238418579,
      "rewards/cosine_scaled_reward": 0.30152534758672117,
      "rewards/format_reward": 0.8303571879863739,
      "rewards/len_reward": -0.0650079416227527,
      "rewards/reasoning_steps_reward": 0.6523810014128685,
      "rewards/tag_count_reward": 0.7991071790456772,
      "step": 780
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 675.2893173217774,
      "epoch": 0.8373333333333334,
      "grad_norm": 1.4300723881782424,
      "kl": 2.218310546875,
      "learning_rate": 2.3428747994183364e-07,
      "loss": 0.0414,
      "reward": 1.7487051695585252,
      "reward_std": 0.8682012394070625,
      "rewards/accuracy_reward": 0.5741071701049805,
      "rewards/cosine_scaled_reward": 0.34424083852209153,
      "rewards/format_reward": 0.8303571879863739,
      "rewards/len_reward": -0.05801073173061013,
      "rewards/reasoning_steps_reward": 0.6276786103844643,
      "rewards/tag_count_reward": 0.8100446790456772,
      "step": 785
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 681.7848541259766,
      "epoch": 0.8426666666666667,
      "grad_norm": 3.1714728291900087,
      "kl": 2.08974609375,
      "learning_rate": 2.1950879348809548e-07,
      "loss": 0.0119,
      "reward": 1.751221266388893,
      "reward_std": 0.8795267082750797,
      "rewards/accuracy_reward": 0.5803571723401546,
      "rewards/cosine_scaled_reward": 0.34943547430448235,
      "rewards/format_reward": 0.8214286148548127,
      "rewards/len_reward": -0.07035618057707324,
      "rewards/reasoning_steps_reward": 0.635119092464447,
      "rewards/tag_count_reward": 0.8131696820259094,
      "step": 790
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 680.6027069091797,
      "epoch": 0.848,
      "grad_norm": 1.5849755112353832,
      "kl": 1.878271484375,
      "learning_rate": 2.0517468572192632e-07,
      "loss": 0.035,
      "reward": 1.7385686576366424,
      "reward_std": 0.892936672270298,
      "rewards/accuracy_reward": 0.5535714536905288,
      "rewards/cosine_scaled_reward": 0.35374716023216024,
      "rewards/format_reward": 0.8312500447034836,
      "rewards/len_reward": -0.06039702805610432,
      "rewards/reasoning_steps_reward": 0.6625000402331352,
      "rewards/tag_count_reward": 0.8140625387430191,
      "step": 795
    },
    {
      "epoch": 0.8533333333333334,
      "grad_norm": 1.9274338875069446,
      "learning_rate": 1.9129013335756317e-07,
      "loss": 0.0496,
      "step": 800
    },
    {
      "epoch": 0.8533333333333334,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 684.6899658203125,
      "eval_kl": 2.2203125,
      "eval_loss": 0.019189750775694847,
      "eval_reward": 1.7292620897293092,
      "eval_reward_std": 0.872122836112976,
      "eval_rewards/accuracy_reward": 0.573214304447174,
      "eval_rewards/cosine_scaled_reward": 0.34890486896038053,
      "eval_rewards/format_reward": 0.8071428775787354,
      "eval_rewards/len_reward": -0.008494636416435242,
      "eval_rewards/reasoning_steps_reward": 0.6005952835083008,
      "eval_rewards/tag_count_reward": 0.8008928894996643,
      "eval_runtime": 122.1397,
      "eval_samples_per_second": 0.622,
      "eval_steps_per_second": 0.008,
      "step": 800
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 691.9308334350586,
      "epoch": 0.8586666666666667,
      "grad_norm": 3.224793392092436,
      "kl": 1.940557861328125,
      "learning_rate": 1.7785995702636698e-07,
      "loss": 0.0121,
      "reward": 1.7232433706521988,
      "reward_std": 0.8734276548027993,
      "rewards/accuracy_reward": 0.5656250283122063,
      "rewards/cosine_scaled_reward": 0.3406540274620056,
      "rewards/format_reward": 0.816964328289032,
      "rewards/len_reward": -0.05547755345760379,
      "rewards/reasoning_steps_reward": 0.6305059880018234,
      "rewards/tag_count_reward": 0.8243303947150707,
      "step": 805
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 708.755387878418,
      "epoch": 0.864,
      "grad_norm": 3.5260608802965816,
      "kl": 1.724560546875,
      "learning_rate": 1.64888819603129e-07,
      "loss": 0.0148,
      "reward": 1.6694684252142906,
      "reward_std": 0.8598788410425187,
      "rewards/accuracy_reward": 0.5410714522004128,
      "rewards/cosine_scaled_reward": 0.31500408379361033,
      "rewards/format_reward": 0.813392898440361,
      "rewards/len_reward": -0.06669973691023187,
      "rewards/reasoning_steps_reward": 0.6187500357627869,
      "rewards/tag_count_reward": 0.8223214700818062,
      "step": 810
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 704.1750305175781,
      "epoch": 0.8693333333333333,
      "grad_norm": 1.8205557040783624,
      "kl": 1.6877685546875,
      "learning_rate": 1.5238122458714925e-07,
      "loss": -0.0007,
      "reward": 1.778202649950981,
      "reward_std": 0.8024675074964762,
      "rewards/accuracy_reward": 0.5928571715950965,
      "rewards/cosine_scaled_reward": 0.3424883014522493,
      "rewards/format_reward": 0.8428571850061417,
      "rewards/len_reward": -0.07196439338222263,
      "rewards/reasoning_steps_reward": 0.6529762379825115,
      "rewards/tag_count_reward": 0.8439732566475868,
      "step": 815
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 703.4661041259766,
      "epoch": 0.8746666666666667,
      "grad_norm": 1.1390547789955021,
      "kl": 1.820703125,
      "learning_rate": 1.4034151453864846e-07,
      "loss": 0.0058,
      "reward": 1.7741546720266341,
      "reward_std": 0.887740996479988,
      "rewards/accuracy_reward": 0.5687500283122062,
      "rewards/cosine_scaled_reward": 0.3643331742845476,
      "rewards/format_reward": 0.84107146859169,
      "rewards/len_reward": -0.06238816555123776,
      "rewards/reasoning_steps_reward": 0.6565476641058922,
      "rewards/tag_count_reward": 0.810044676065445,
      "step": 820
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 711.9839569091797,
      "epoch": 0.88,
      "grad_norm": 3.201601066169082,
      "kl": 2.0060546875,
      "learning_rate": 1.287738695710592e-07,
      "loss": 0.0445,
      "reward": 1.741848286986351,
      "reward_std": 0.8842388309538365,
      "rewards/accuracy_reward": 0.5696428827941418,
      "rewards/cosine_scaled_reward": 0.3463124948553741,
      "rewards/format_reward": 0.8258928999304771,
      "rewards/len_reward": -0.061798423925574754,
      "rewards/reasoning_steps_reward": 0.6389881439507008,
      "rewards/tag_count_reward": 0.8033482506871223,
      "step": 825
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 699.556282043457,
      "epoch": 0.8853333333333333,
      "grad_norm": 2.047674415321257,
      "kl": 1.7311279296875,
      "learning_rate": 1.1768230589971457e-07,
      "loss": 0.0151,
      "reward": 1.7691504821181296,
      "reward_std": 0.8467303015291691,
      "rewards/accuracy_reward": 0.581250025331974,
      "rewards/cosine_scaled_reward": 0.3566504124552011,
      "rewards/format_reward": 0.8312500342726707,
      "rewards/len_reward": -0.05797590847651009,
      "rewards/reasoning_steps_reward": 0.6645833745598793,
      "rewards/tag_count_reward": 0.813169677555561,
      "step": 830
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 718.4741348266601,
      "epoch": 0.8906666666666667,
      "grad_norm": 1.2710201599686817,
      "kl": 1.734326171875,
      "learning_rate": 1.0707067444744439e-07,
      "loss": 0.0118,
      "reward": 1.7898413628339767,
      "reward_std": 0.8654466308653355,
      "rewards/accuracy_reward": 0.5803571701049804,
      "rewards/cosine_scaled_reward": 0.36394843012094497,
      "rewards/format_reward": 0.845535758137703,
      "rewards/len_reward": -0.06752787398800138,
      "rewards/reasoning_steps_reward": 0.6630952812731266,
      "rewards/tag_count_reward": 0.8261161103844643,
      "step": 835
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 699.3634231567382,
      "epoch": 0.896,
      "grad_norm": 1.9737307274220788,
      "kl": 1.7364990234375,
      "learning_rate": 9.69426595075566e-08,
      "loss": -0.0073,
      "reward": 1.728901642560959,
      "reward_std": 0.8724588222801686,
      "rewards/accuracy_reward": 0.5598214510828257,
      "rewards/cosine_scaled_reward": 0.33693731487728656,
      "rewards/format_reward": 0.8321429014205932,
      "rewards/len_reward": -0.06747927672695368,
      "rewards/reasoning_steps_reward": 0.6422619573771954,
      "rewards/tag_count_reward": 0.8299107551574707,
      "step": 840
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 737.0732482910156,
      "epoch": 0.9013333333333333,
      "grad_norm": 1.9784493712749323,
      "kl": 1.6184814453125,
      "learning_rate": 8.730177746467616e-08,
      "loss": 0.0245,
      "reward": 1.6512254253029823,
      "reward_std": 0.8224760733544827,
      "rewards/accuracy_reward": 0.5133928835391999,
      "rewards/cosine_scaled_reward": 0.2949753848835826,
      "rewards/format_reward": 0.84285718947649,
      "rewards/len_reward": -0.06837809896096587,
      "rewards/reasoning_steps_reward": 0.6458333730697632,
      "rewards/tag_count_reward": 0.8607143267989159,
      "step": 845
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 717.0750335693359,
      "epoch": 0.9066666666666666,
      "grad_norm": 1.1451369234582136,
      "kl": 1.67718505859375,
      "learning_rate": 7.81513755738742e-08,
      "loss": 0.0011,
      "reward": 1.742706833779812,
      "reward_std": 0.8420647040009499,
      "rewards/accuracy_reward": 0.5696428827941418,
      "rewards/cosine_scaled_reward": 0.3373496507178061,
      "rewards/format_reward": 0.8357143253087997,
      "rewards/len_reward": -0.06677207143511624,
      "rewards/reasoning_steps_reward": 0.6532738506793976,
      "rewards/tag_count_reward": 0.8330357551574707,
      "step": 850
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 739.9223541259765,
      "epoch": 0.912,
      "grad_norm": 2.5094540072798353,
      "kl": 1.945263671875,
      "learning_rate": 6.949463079852491e-08,
      "loss": 0.0302,
      "reward": 1.7340825259685517,
      "reward_std": 0.8873489238321781,
      "rewards/accuracy_reward": 0.5642857439815998,
      "rewards/cosine_scaled_reward": 0.33586815614253285,
      "rewards/format_reward": 0.8339286178350449,
      "rewards/len_reward": -0.05986913426313549,
      "rewards/reasoning_steps_reward": 0.6571428991854191,
      "rewards/tag_count_reward": 0.8406250417232514,
      "step": 855
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 696.5928855895996,
      "epoch": 0.9173333333333333,
      "grad_norm": 1.9398788694913773,
      "kl": 2.00673828125,
      "learning_rate": 6.133454870728111e-08,
      "loss": 0.0432,
      "reward": 1.7525239706039428,
      "reward_std": 0.9240262523293495,
      "rewards/accuracy_reward": 0.5803571745753289,
      "rewards/cosine_scaled_reward": 0.3480596005916595,
      "rewards/format_reward": 0.8241071864962578,
      "rewards/len_reward": -0.05318943413440138,
      "rewards/reasoning_steps_reward": 0.6342262253165245,
      "rewards/tag_count_reward": 0.830803605914116,
      "step": 860
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 737.1518157958984,
      "epoch": 0.9226666666666666,
      "grad_norm": 2.0067649784156574,
      "kl": 1.8587890625,
      "learning_rate": 5.367396243056022e-08,
      "loss": 0.0586,
      "reward": 1.5764021649956703,
      "reward_std": 0.9064562991261482,
      "rewards/accuracy_reward": 0.5071428824216128,
      "rewards/cosine_scaled_reward": 0.2781878274225164,
      "rewards/format_reward": 0.7910714671015739,
      "rewards/len_reward": -0.0710936440504156,
      "rewards/reasoning_steps_reward": 0.6214286148548126,
      "rewards/tag_count_reward": 0.8029018253087997,
      "step": 865
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 702.3964599609375,
      "epoch": 0.928,
      "grad_norm": 1.3626379779643958,
      "kl": 1.9628662109375,
      "learning_rate": 4.6515531676899316e-08,
      "loss": 0.0235,
      "reward": 1.8264706060290337,
      "reward_std": 0.896178449690342,
      "rewards/accuracy_reward": 0.603571455180645,
      "rewards/cosine_scaled_reward": 0.38807767661637627,
      "rewards/format_reward": 0.8348214730620385,
      "rewards/len_reward": -0.053661848261253906,
      "rewards/reasoning_steps_reward": 0.656547662615776,
      "rewards/tag_count_reward": 0.8383928894996643,
      "step": 870
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 688.1732444763184,
      "epoch": 0.9333333333333333,
      "grad_norm": 1.9694362633321498,
      "kl": 2.45517578125,
      "learning_rate": 3.986174180951896e-08,
      "loss": 0.0403,
      "reward": 1.7715040892362595,
      "reward_std": 0.875535361468792,
      "rewards/accuracy_reward": 0.5866071682423353,
      "rewards/cosine_scaled_reward": 0.3420397279784083,
      "rewards/format_reward": 0.8428571850061417,
      "rewards/len_reward": -0.07309366355184466,
      "rewards/reasoning_steps_reward": 0.6440476626157761,
      "rewards/tag_count_reward": 0.829687537252903,
      "step": 875
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 704.2875289916992,
      "epoch": 0.9386666666666666,
      "grad_norm": 1.8297011568150405,
      "kl": 2.25908203125,
      "learning_rate": 3.3714902983421944e-08,
      "loss": 0.0123,
      "reward": 1.7017309829592704,
      "reward_std": 0.9257866092026233,
      "rewards/accuracy_reward": 0.5705357387661933,
      "rewards/cosine_scaled_reward": 0.32851662803441284,
      "rewards/format_reward": 0.8026786118745803,
      "rewards/len_reward": -0.06342671204765793,
      "rewards/reasoning_steps_reward": 0.6318452827632427,
      "rewards/tag_count_reward": 0.8151786103844643,
      "step": 880
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 709.7786010742187,
      "epoch": 0.944,
      "grad_norm": 2.0419278966924916,
      "kl": 2.3875,
      "learning_rate": 2.807714934332073e-08,
      "loss": 0.0495,
      "reward": 1.659614458680153,
      "reward_std": 0.9322795614600181,
      "rewards/accuracy_reward": 0.5294643096625805,
      "rewards/cosine_scaled_reward": 0.3292572578415275,
      "rewards/format_reward": 0.8008929014205932,
      "rewards/len_reward": -0.07157147889083718,
      "rewards/reasoning_steps_reward": 0.6122024178504943,
      "rewards/tag_count_reward": 0.8060268223285675,
      "step": 885
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 715.5553909301758,
      "epoch": 0.9493333333333334,
      "grad_norm": 2.8653022841959257,
      "kl": 2.29814453125,
      "learning_rate": 2.2950438282676455e-08,
      "loss": -0.0014,
      "reward": 1.6036544814705849,
      "reward_std": 0.9239592231810093,
      "rewards/accuracy_reward": 0.5241071704775095,
      "rewards/cosine_scaled_reward": 0.2813329972326756,
      "rewards/format_reward": 0.7982143253087998,
      "rewards/len_reward": -0.08350599388359115,
      "rewards/reasoning_steps_reward": 0.603571467846632,
      "rewards/tag_count_reward": 0.816741107404232,
      "step": 890
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 709.7669952392578,
      "epoch": 0.9546666666666667,
      "grad_norm": 3.1175284258712788,
      "kl": 2.15888671875,
      "learning_rate": 1.8336549764102594e-08,
      "loss": 0.038,
      "reward": 1.7244937881827354,
      "reward_std": 0.8962258003652096,
      "rewards/accuracy_reward": 0.5776786021888256,
      "rewards/cosine_scaled_reward": 0.34592231740243734,
      "rewards/format_reward": 0.800892898440361,
      "rewards/len_reward": -0.05511092038359493,
      "rewards/reasoning_steps_reward": 0.6098214618861675,
      "rewards/tag_count_reward": 0.8035714671015739,
      "step": 895
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9860913993698874,
      "learning_rate": 1.4237085701374109e-08,
      "loss": 0.0147,
      "step": 900
    },
    {
      "epoch": 0.96,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 710.0548217773437,
      "eval_kl": 2.0609375,
      "eval_loss": 0.08885510265827179,
      "eval_reward": 1.7470547199249267,
      "eval_reward_std": 0.9193769574165345,
      "eval_rewards/accuracy_reward": 0.5785714566707612,
      "eval_rewards/cosine_scaled_reward": 0.3541975736618042,
      "eval_rewards/format_reward": 0.8142857432365418,
      "eval_rewards/len_reward": 0.017873572744429113,
      "eval_rewards/reasoning_steps_reward": 0.6309524178504944,
      "eval_rewards/tag_count_reward": 0.8183036088943482,
      "eval_runtime": 125.1026,
      "eval_samples_per_second": 0.608,
      "eval_steps_per_second": 0.008,
      "step": 900
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 694.5861915588379,
      "epoch": 0.9653333333333334,
      "grad_norm": 1.6253024075355416,
      "kl": 2.132666015625,
      "learning_rate": 1.0653469403252015e-08,
      "loss": 0.0295,
      "reward": 1.7099345169961453,
      "reward_std": 0.9086461085826159,
      "rewards/accuracy_reward": 0.5669643130153418,
      "rewards/cosine_scaled_reward": 0.3335951896267943,
      "rewards/format_reward": 0.8093750402331352,
      "rewards/len_reward": -0.06031820691714529,
      "rewards/reasoning_steps_reward": 0.6342262338846922,
      "rewards/tag_count_reward": 0.8136161088943481,
      "step": 905
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 711.0339614868165,
      "epoch": 0.9706666666666667,
      "grad_norm": 3.0891558552999125,
      "kl": 2.23671875,
      "learning_rate": 7.586945079319673e-09,
      "loss": 0.027,
      "reward": 1.7360344290733338,
      "reward_std": 0.8696943923830986,
      "rewards/accuracy_reward": 0.5678571678698063,
      "rewards/cosine_scaled_reward": 0.33514149505645036,
      "rewards/format_reward": 0.8330357581377029,
      "rewards/len_reward": -0.06922432167921215,
      "rewards/reasoning_steps_reward": 0.6446429014205932,
      "rewards/tag_count_reward": 0.8270089656114579,
      "step": 910
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 702.9910995483399,
      "epoch": 0.976,
      "grad_norm": 1.4845427370224942,
      "kl": 2.0302734375,
      "learning_rate": 5.038577408000844e-09,
      "loss": 0.0143,
      "reward": 1.6590574353933334,
      "reward_std": 0.9159045577049255,
      "rewards/accuracy_reward": 0.5455357421189546,
      "rewards/cosine_scaled_reward": 0.3045931006781757,
      "rewards/format_reward": 0.8089286133646965,
      "rewards/len_reward": -0.0816710107261315,
      "rewards/reasoning_steps_reward": 0.6232143297791481,
      "rewards/tag_count_reward": 0.808258967101574,
      "step": 915
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 699.5473541259765,
      "epoch": 0.9813333333333333,
      "grad_norm": 1.7313125978534194,
      "kl": 2.3384765625,
      "learning_rate": 3.009251166909699e-09,
      "loss": 0.0087,
      "reward": 1.5918783232569695,
      "reward_std": 0.9757736340165138,
      "rewards/accuracy_reward": 0.5142857357859612,
      "rewards/cosine_scaled_reward": 0.2990211246535182,
      "rewards/format_reward": 0.778571467101574,
      "rewards/len_reward": -0.08776085836580023,
      "rewards/reasoning_steps_reward": 0.6345238506793975,
      "rewards/tag_count_reward": 0.7993303939700127,
      "step": 920
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 699.5911056518555,
      "epoch": 0.9866666666666667,
      "grad_norm": 1.739888955006767,
      "kl": 2.113232421875,
      "learning_rate": 1.4996709256617225e-09,
      "loss": 0.0032,
      "reward": 1.8000961601734162,
      "reward_std": 0.9336124449968338,
      "rewards/accuracy_reward": 0.6107143178582192,
      "rewards/cosine_scaled_reward": 0.36973895924165845,
      "rewards/format_reward": 0.8196429014205933,
      "rewards/len_reward": -0.06013430766906822,
      "rewards/reasoning_steps_reward": 0.6375000424683094,
      "rewards/tag_count_reward": 0.8087053939700126,
      "step": 925
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 704.1303871154785,
      "epoch": 0.992,
      "grad_norm": 1.0971767587313153,
      "kl": 1.8646484375,
      "learning_rate": 5.103608012512195e-10,
      "loss": 0.025,
      "reward": 1.7941276371479034,
      "reward_std": 0.9121661514043808,
      "rewards/accuracy_reward": 0.6089286014437676,
      "rewards/cosine_scaled_reward": 0.36734184846282003,
      "rewards/format_reward": 0.8178571909666061,
      "rewards/len_reward": -0.04802298827562481,
      "rewards/reasoning_steps_reward": 0.6273809909820557,
      "rewards/tag_count_reward": 0.8111607506871223,
      "step": 930
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 715.4536041259765,
      "epoch": 0.9973333333333333,
      "grad_norm": 1.472648685003733,
      "kl": 2.31025390625,
      "learning_rate": 4.1664276081376796e-11,
      "loss": 0.0476,
      "reward": 1.6112597823143004,
      "reward_std": 0.9594747319817543,
      "rewards/accuracy_reward": 0.5142857380211353,
      "rewards/cosine_scaled_reward": 0.294295444060117,
      "rewards/format_reward": 0.8026786193251609,
      "rewards/len_reward": -0.06919930868316441,
      "rewards/reasoning_steps_reward": 0.62529766112566,
      "rewards/tag_count_reward": 0.8127232551574707,
      "step": 935
    },
    {
      "clip_ratio": 0.0,
      "completion_length": 724.0409545898438,
      "epoch": 0.9994666666666666,
      "kl": 2.156005859375,
      "reward": 1.7690049931406975,
      "reward_std": 0.8843330852687359,
      "rewards/accuracy_reward": 0.6093750298023224,
      "rewards/cosine_scaled_reward": 0.35382634587585926,
      "rewards/format_reward": 0.8058036118745804,
      "rewards/len_reward": -0.057325188361573964,
      "rewards/reasoning_steps_reward": 0.6279762387275696,
      "rewards/tag_count_reward": 0.8113839626312256,
      "step": 937,
      "total_flos": 0.0,
      "train_loss": 0.09447722357977083,
      "train_runtime": 119506.5825,
      "train_samples_per_second": 0.251,
      "train_steps_per_second": 0.008
    }
  ],
  "logging_steps": 5,
  "max_steps": 937,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}