{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 1088,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004595588235294118,
      "grad_norm": 75.5,
      "learning_rate": 2.9965533088235297e-05,
      "loss": 7.523,
      "mean_token_accuracy": 0.656203106045723,
      "step": 5,
      "train/kl_loss_qwen": 1.0067849405109883,
      "train/kl_loss_r1": 0.4999275255157499,
      "train/total_kl": 1.5067124910652638
    },
    {
      "epoch": 0.009191176470588236,
      "grad_norm": 29.125,
      "learning_rate": 2.993106617647059e-05,
      "loss": 5.5883,
      "mean_token_accuracy": 0.7142428308725357,
      "step": 10,
      "train/kl_loss_qwen": 0.7231092929840088,
      "train/kl_loss_r1": 0.4078650489449501,
      "train/total_kl": 1.13097435683012
    },
    {
      "epoch": 0.013786764705882353,
      "grad_norm": 38.75,
      "learning_rate": 2.9896599264705885e-05,
      "loss": 4.3,
      "mean_token_accuracy": 0.7153706133365632,
      "step": 15,
      "train/kl_loss_qwen": 0.542685666680336,
      "train/kl_loss_r1": 0.2508307069540024,
      "train/total_kl": 0.7935163721442222
    },
    {
      "epoch": 0.01838235294117647,
      "grad_norm": 14.5,
      "learning_rate": 2.9862132352941178e-05,
      "loss": 3.2403,
      "mean_token_accuracy": 0.7293514549732208,
      "step": 20,
      "train/kl_loss_qwen": 0.3540587030351162,
      "train/kl_loss_r1": 0.18121434897184371,
      "train/total_kl": 0.5352730460464954
    },
    {
      "epoch": 0.02297794117647059,
      "grad_norm": 13.4375,
      "learning_rate": 2.982766544117647e-05,
      "loss": 3.1363,
      "mean_token_accuracy": 0.7384833335876465,
      "step": 25,
      "train/kl_loss_qwen": 0.3636454150080681,
      "train/kl_loss_r1": 0.15524634066969156,
      "train/total_kl": 0.5188917614519596
    },
    {
      "epoch": 0.027573529411764705,
      "grad_norm": 19.25,
      "learning_rate": 2.9793198529411766e-05,
      "loss": 4.2733,
      "mean_token_accuracy": 0.7523588448762893,
      "step": 30,
      "train/kl_loss_qwen": 0.5658364623785019,
      "train/kl_loss_r1": 0.27777326982468364,
      "train/total_kl": 0.843609730899334
    },
    {
      "epoch": 0.03216911764705882,
      "grad_norm": 31.5,
      "learning_rate": 2.975873161764706e-05,
      "loss": 3.7129,
      "mean_token_accuracy": 0.7343137323856354,
      "step": 35,
      "train/kl_loss_qwen": 0.4348488930612803,
      "train/kl_loss_r1": 0.2575880441814661,
      "train/total_kl": 0.6924369387328625
    },
    {
      "epoch": 0.03676470588235294,
      "grad_norm": 8.25,
      "learning_rate": 2.9724264705882355e-05,
      "loss": 2.7708,
      "mean_token_accuracy": 0.728923574090004,
      "step": 40,
      "train/kl_loss_qwen": 0.2851369768381119,
      "train/kl_loss_r1": 0.13467254638671874,
      "train/total_kl": 0.419809527695179
    },
    {
      "epoch": 0.04136029411764706,
      "grad_norm": 9.0625,
      "learning_rate": 2.9689797794117647e-05,
      "loss": 3.1069,
      "mean_token_accuracy": 0.7631141632795334,
      "step": 45,
      "train/kl_loss_qwen": 0.3510517507791519,
      "train/kl_loss_r1": 0.16951459292322396,
      "train/total_kl": 0.5205663420259953
    },
    {
      "epoch": 0.04595588235294118,
      "grad_norm": 20.5,
      "learning_rate": 2.9655330882352943e-05,
      "loss": 2.9952,
      "mean_token_accuracy": 0.7505580186843872,
      "step": 50,
      "train/kl_loss_qwen": 0.3157908026129007,
      "train/kl_loss_r1": 0.15686270054429768,
      "train/total_kl": 0.4726535022258759
    },
    {
      "epoch": 0.050551470588235295,
      "grad_norm": 6.65625,
      "learning_rate": 2.9620863970588236e-05,
      "loss": 2.8788,
      "mean_token_accuracy": 0.7706852316856384,
      "step": 55,
      "train/kl_loss_qwen": 0.2879650827497244,
      "train/kl_loss_r1": 0.16403407491743566,
      "train/total_kl": 0.45199914574623107
    },
    {
      "epoch": 0.05514705882352941,
      "grad_norm": 41.5,
      "learning_rate": 2.958639705882353e-05,
      "loss": 4.1419,
      "mean_token_accuracy": 0.7634413212537765,
      "step": 60,
      "train/kl_loss_qwen": 0.5039678934961558,
      "train/kl_loss_r1": 0.3037068540230393,
      "train/total_kl": 0.8076747469604015
    },
    {
      "epoch": 0.05974264705882353,
      "grad_norm": 8.8125,
      "learning_rate": 2.9551930147058824e-05,
      "loss": 2.847,
      "mean_token_accuracy": 0.7341876327991486,
      "step": 65,
      "train/kl_loss_qwen": 0.296791860088706,
      "train/kl_loss_r1": 0.16430705823004246,
      "train/total_kl": 0.4610989138484001
    },
    {
      "epoch": 0.06433823529411764,
      "grad_norm": 11.5,
      "learning_rate": 2.9517463235294117e-05,
      "loss": 2.524,
      "mean_token_accuracy": 0.747851875424385,
      "step": 70,
      "train/kl_loss_qwen": 0.23588257953524588,
      "train/kl_loss_r1": 0.12017320897430181,
      "train/total_kl": 0.3560557886958122
    },
    {
      "epoch": 0.06893382352941177,
      "grad_norm": 6.8125,
      "learning_rate": 2.9482996323529412e-05,
      "loss": 2.638,
      "mean_token_accuracy": 0.7645364701747894,
      "step": 75,
      "train/kl_loss_qwen": 0.2567322202026844,
      "train/kl_loss_r1": 0.138734588958323,
      "train/total_kl": 0.3954668059945107
    },
    {
      "epoch": 0.07352941176470588,
      "grad_norm": 7.84375,
      "learning_rate": 2.9448529411764705e-05,
      "loss": 3.5239,
      "mean_token_accuracy": 0.760419961810112,
      "step": 80,
      "train/kl_loss_qwen": 0.4126748025417328,
      "train/kl_loss_r1": 0.2278809394687414,
      "train/total_kl": 0.640555739402771
    },
    {
      "epoch": 0.078125,
      "grad_norm": 18.875,
      "learning_rate": 2.94140625e-05,
      "loss": 4.2956,
      "mean_token_accuracy": 0.7762624591588974,
      "step": 85,
      "train/kl_loss_qwen": 0.5349518846720457,
      "train/kl_loss_r1": 0.3311561655253172,
      "train/total_kl": 0.8661080494523048
    },
    {
      "epoch": 0.08272058823529412,
      "grad_norm": 47.0,
      "learning_rate": 2.9379595588235297e-05,
      "loss": 3.2597,
      "mean_token_accuracy": 0.7334744036197662,
      "step": 90,
      "train/kl_loss_qwen": 0.3614440344274044,
      "train/kl_loss_r1": 0.20499854441732168,
      "train/total_kl": 0.5664425753057003
    },
    {
      "epoch": 0.08731617647058823,
      "grad_norm": 11.9375,
      "learning_rate": 2.934512867647059e-05,
      "loss": 3.0669,
      "mean_token_accuracy": 0.7597341448068619,
      "step": 95,
      "train/kl_loss_qwen": 0.3339898385107517,
      "train/kl_loss_r1": 0.19944773185998202,
      "train/total_kl": 0.5334375720471144
    },
    {
      "epoch": 0.09191176470588236,
      "grad_norm": 12.0,
      "learning_rate": 2.9310661764705885e-05,
      "loss": 2.4843,
      "mean_token_accuracy": 0.7619072079658509,
      "step": 100,
      "train/kl_loss_qwen": 0.22200208008289338,
      "train/kl_loss_r1": 0.13626624476164578,
      "train/total_kl": 0.35826832354068755
    },
    {
      "epoch": 0.09650735294117647,
      "grad_norm": 23.375,
      "learning_rate": 2.9276194852941174e-05,
      "loss": 2.7679,
      "mean_token_accuracy": 0.7749551564455033,
      "step": 105,
      "train/kl_loss_qwen": 0.27624217644333837,
      "train/kl_loss_r1": 0.17033643070608379,
      "train/total_kl": 0.4465786039829254
    },
    {
      "epoch": 0.10110294117647059,
      "grad_norm": 11.6875,
      "learning_rate": 2.924172794117647e-05,
      "loss": 3.3113,
      "mean_token_accuracy": 0.7359262406826019,
      "step": 110,
      "train/kl_loss_qwen": 0.386141000315547,
      "train/kl_loss_r1": 0.20178384203463792,
      "train/total_kl": 0.5879248447716237
    },
    {
      "epoch": 0.1056985294117647,
      "grad_norm": 10.375,
      "learning_rate": 2.9207261029411766e-05,
      "loss": 4.194,
      "mean_token_accuracy": 0.740036615729332,
      "step": 115,
      "train/kl_loss_qwen": 0.5264142207801342,
      "train/kl_loss_r1": 0.2895144520327449,
      "train/total_kl": 0.8159286729991436
    },
    {
      "epoch": 0.11029411764705882,
      "grad_norm": 51.5,
      "learning_rate": 2.917279411764706e-05,
      "loss": 4.3143,
      "mean_token_accuracy": 0.7599226355552673,
      "step": 120,
      "train/kl_loss_qwen": 0.5216624522581697,
      "train/kl_loss_r1": 0.3390774505212903,
      "train/total_kl": 0.8607399154454469
    },
    {
      "epoch": 0.11488970588235294,
      "grad_norm": 13.0,
      "learning_rate": 2.9138327205882355e-05,
      "loss": 3.973,
      "mean_token_accuracy": 0.756134757399559,
      "step": 125,
      "train/kl_loss_qwen": 0.48524228297173977,
      "train/kl_loss_r1": 0.27532588355243204,
      "train/total_kl": 0.7605681672692299
    },
    {
      "epoch": 0.11948529411764706,
      "grad_norm": 24.375,
      "learning_rate": 2.9103860294117647e-05,
      "loss": 3.3751,
      "mean_token_accuracy": 0.7322900414466857,
      "step": 130,
      "train/kl_loss_qwen": 0.3569587718695402,
      "train/kl_loss_r1": 0.2533737229183316,
      "train/total_kl": 0.6103324957191945
    },
    {
      "epoch": 0.12408088235294118,
      "grad_norm": 15.25,
      "learning_rate": 2.9069393382352943e-05,
      "loss": 3.8036,
      "mean_token_accuracy": 0.7421320974826813,
      "step": 135,
      "train/kl_loss_qwen": 0.45374844707548617,
      "train/kl_loss_r1": 0.2738330475986004,
      "train/total_kl": 0.7275814957916736
    },
    {
      "epoch": 0.12867647058823528,
      "grad_norm": 5.75,
      "learning_rate": 2.9034926470588236e-05,
      "loss": 2.95,
      "mean_token_accuracy": 0.745361715555191,
      "step": 140,
      "train/kl_loss_qwen": 0.3025351382791996,
      "train/kl_loss_r1": 0.17302710115909575,
      "train/total_kl": 0.4755622450262308
    },
    {
      "epoch": 0.1332720588235294,
      "grad_norm": 33.0,
      "learning_rate": 2.900045955882353e-05,
      "loss": 4.0155,
      "mean_token_accuracy": 0.7763338893651962,
      "step": 145,
      "train/kl_loss_qwen": 0.5002008650451899,
      "train/kl_loss_r1": 0.2966492097824812,
      "train/total_kl": 0.7968500778079033
    },
    {
      "epoch": 0.13786764705882354,
      "grad_norm": 13.75,
      "learning_rate": 2.8965992647058824e-05,
      "loss": 3.1621,
      "mean_token_accuracy": 0.7467743963003158,
      "step": 150,
      "train/kl_loss_qwen": 0.34167556203901767,
      "train/kl_loss_r1": 0.2065791456028819,
      "train/total_kl": 0.5482547096908092
    },
    {
      "epoch": 0.14246323529411764,
      "grad_norm": 11.3125,
      "learning_rate": 2.8931525735294117e-05,
      "loss": 4.0015,
      "mean_token_accuracy": 0.7587345838546753,
      "step": 155,
      "train/kl_loss_qwen": 0.46732648052275183,
      "train/kl_loss_r1": 0.3139583099633455,
      "train/total_kl": 0.7812847837805748
    },
    {
      "epoch": 0.14705882352941177,
      "grad_norm": 25.75,
      "learning_rate": 2.8897058823529413e-05,
      "loss": 3.6916,
      "mean_token_accuracy": 0.7547200947999955,
      "step": 160,
      "train/kl_loss_qwen": 0.45896258056163786,
      "train/kl_loss_r1": 0.2407479902729392,
      "train/total_kl": 0.6997105687856674
    },
    {
      "epoch": 0.15165441176470587,
      "grad_norm": 10.375,
      "learning_rate": 2.8862591911764705e-05,
      "loss": 2.8947,
      "mean_token_accuracy": 0.7471380352973938,
      "step": 165,
      "train/kl_loss_qwen": 0.2993515759706497,
      "train/kl_loss_r1": 0.1753220736980438,
      "train/total_kl": 0.47467364743351936
    },
    {
      "epoch": 0.15625,
      "grad_norm": 7.21875,
      "learning_rate": 2.8828125e-05,
      "loss": 3.0875,
      "mean_token_accuracy": 0.7554198652505875,
      "step": 170,
      "train/kl_loss_qwen": 0.3381518451496959,
      "train/kl_loss_r1": 0.1907153930515051,
      "train/total_kl": 0.5288672372698784
    },
    {
      "epoch": 0.16084558823529413,
      "grad_norm": 46.25,
      "learning_rate": 2.8793658088235297e-05,
      "loss": 3.8006,
      "mean_token_accuracy": 0.7543142795562744,
      "step": 175,
      "train/kl_loss_qwen": 0.4342954009771347,
      "train/kl_loss_r1": 0.2781439159065485,
      "train/total_kl": 0.7124393127858639
    },
    {
      "epoch": 0.16544117647058823,
      "grad_norm": 27.625,
      "learning_rate": 2.875919117647059e-05,
      "loss": 3.1037,
      "mean_token_accuracy": 0.7643545418977737,
      "step": 180,
      "train/kl_loss_qwen": 0.32490024231374265,
      "train/kl_loss_r1": 0.2150045122951269,
      "train/total_kl": 0.5399047564715147
    },
    {
      "epoch": 0.17003676470588236,
      "grad_norm": 15.0625,
      "learning_rate": 2.8724724264705885e-05,
      "loss": 3.3512,
      "mean_token_accuracy": 0.7656952410936355,
      "step": 185,
      "train/kl_loss_qwen": 0.39200425930321214,
      "train/kl_loss_r1": 0.21828225292265416,
      "train/total_kl": 0.610286520421505
    },
    {
      "epoch": 0.17463235294117646,
      "grad_norm": 9.5,
      "learning_rate": 2.8690257352941175e-05,
      "loss": 2.4934,
      "mean_token_accuracy": 0.7389886736869812,
      "step": 190,
      "train/kl_loss_qwen": 0.24613040834665298,
      "train/kl_loss_r1": 0.12397182490676642,
      "train/total_kl": 0.37010223269462583
    },
    {
      "epoch": 0.1792279411764706,
      "grad_norm": 17.5,
      "learning_rate": 2.865579044117647e-05,
      "loss": 3.0527,
      "mean_token_accuracy": 0.7645234823226928,
      "step": 195,
      "train/kl_loss_qwen": 0.3280874006450176,
      "train/kl_loss_r1": 0.1920649280771613,
      "train/total_kl": 0.5201523378491402
    },
    {
      "epoch": 0.18382352941176472,
      "grad_norm": 8.3125,
      "learning_rate": 2.8621323529411766e-05,
      "loss": 2.5046,
      "mean_token_accuracy": 0.7753487080335617,
      "step": 200,
      "train/kl_loss_qwen": 0.24486010633409022,
      "train/kl_loss_r1": 0.12939048558473587,
      "train/total_kl": 0.3742505945265293
    },
    {
      "epoch": 0.18841911764705882,
      "grad_norm": 35.0,
      "learning_rate": 2.858685661764706e-05,
      "loss": 3.7333,
      "mean_token_accuracy": 0.7638763725757599,
      "step": 205,
      "train/kl_loss_qwen": 0.4581252448260784,
      "train/kl_loss_r1": 0.25974704790860415,
      "train/total_kl": 0.7178723007440567
    },
    {
      "epoch": 0.19301470588235295,
      "grad_norm": 11.3125,
      "learning_rate": 2.8552389705882355e-05,
      "loss": 3.4325,
      "mean_token_accuracy": 0.7258242666721344,
      "step": 210,
      "train/kl_loss_qwen": 0.3888599809259176,
      "train/kl_loss_r1": 0.22046158649027348,
      "train/total_kl": 0.609321566671133
    },
    {
      "epoch": 0.19761029411764705,
      "grad_norm": 7.28125,
      "learning_rate": 2.8517922794117647e-05,
      "loss": 2.5594,
      "mean_token_accuracy": 0.746726468205452,
      "step": 215,
      "train/kl_loss_qwen": 0.23355862349271775,
      "train/kl_loss_r1": 0.15477345269173384,
      "train/total_kl": 0.38833207711577417
    },
    {
      "epoch": 0.20220588235294118,
      "grad_norm": 15.4375,
      "learning_rate": 2.8483455882352943e-05,
      "loss": 3.2255,
      "mean_token_accuracy": 0.763618940114975,
      "step": 220,
      "train/kl_loss_qwen": 0.3598331693559885,
      "train/kl_loss_r1": 0.22203538492321967,
      "train/total_kl": 0.5818685468286275
    },
    {
      "epoch": 0.20680147058823528,
      "grad_norm": 11.6875,
      "learning_rate": 2.8448988970588236e-05,
      "loss": 3.1035,
      "mean_token_accuracy": 0.7524501204490661,
      "step": 225,
      "train/kl_loss_qwen": 0.32799928411841395,
      "train/kl_loss_r1": 0.2156254580244422,
      "train/total_kl": 0.5436247382313013
    },
    {
      "epoch": 0.2113970588235294,
      "grad_norm": 18.5,
      "learning_rate": 2.841452205882353e-05,
      "loss": 3.5389,
      "mean_token_accuracy": 0.7489451408386231,
      "step": 230,
      "train/kl_loss_qwen": 0.3832179747521877,
      "train/kl_loss_r1": 0.2746766399592161,
      "train/total_kl": 0.6578946128487587
    },
    {
      "epoch": 0.21599264705882354,
      "grad_norm": 32.75,
      "learning_rate": 2.8380055147058824e-05,
      "loss": 3.1394,
      "mean_token_accuracy": 0.762254998087883,
      "step": 235,
      "train/kl_loss_qwen": 0.3588922441005707,
      "train/kl_loss_r1": 0.19826317485421896,
      "train/total_kl": 0.5571554228663445
    },
    {
      "epoch": 0.22058823529411764,
      "grad_norm": 24.0,
      "learning_rate": 2.8345588235294117e-05,
      "loss": 3.1694,
      "mean_token_accuracy": 0.7346105724573135,
      "step": 240,
      "train/kl_loss_qwen": 0.3613942746073008,
      "train/kl_loss_r1": 0.1897347206249833,
      "train/total_kl": 0.5511290028691291
    },
    {
      "epoch": 0.22518382352941177,
      "grad_norm": 15.1875,
      "learning_rate": 2.8311121323529413e-05,
      "loss": 3.3187,
      "mean_token_accuracy": 0.7426173239946365,
      "step": 245,
      "train/kl_loss_qwen": 0.35766145810484884,
      "train/kl_loss_r1": 0.23391314651817083,
      "train/total_kl": 0.5915746077895164
    },
    {
      "epoch": 0.22977941176470587,
      "grad_norm": 8.9375,
      "learning_rate": 2.8276654411764705e-05,
      "loss": 2.8405,
      "mean_token_accuracy": 0.7459217041730881,
      "step": 250,
      "train/kl_loss_qwen": 0.2941163208335638,
      "train/kl_loss_r1": 0.17592173498123884,
      "train/total_kl": 0.47003805860877035
    },
    {
      "epoch": 0.234375,
      "grad_norm": 7.5,
      "learning_rate": 2.82421875e-05,
      "loss": 2.8605,
      "mean_token_accuracy": 0.7551658779382706,
      "step": 255,
      "train/kl_loss_qwen": 0.2726207092404366,
      "train/kl_loss_r1": 0.19635038916021585,
      "train/total_kl": 0.4689711004495621
    },
    {
      "epoch": 0.23897058823529413,
      "grad_norm": 22.875,
      "learning_rate": 2.8207720588235297e-05,
      "loss": 3.3814,
      "mean_token_accuracy": 0.7645096510648728,
      "step": 260,
      "train/kl_loss_qwen": 0.38655108250677583,
      "train/kl_loss_r1": 0.2306967219337821,
      "train/total_kl": 0.6172478139400482
    },
    {
      "epoch": 0.24356617647058823,
      "grad_norm": 8.1875,
      "learning_rate": 2.817325367647059e-05,
      "loss": 3.1262,
      "mean_token_accuracy": 0.7208732426166534,
      "step": 265,
      "train/kl_loss_qwen": 0.3265652749687433,
      "train/kl_loss_r1": 0.19569566920399667,
      "train/total_kl": 0.5222609452903271
    },
    {
      "epoch": 0.24816176470588236,
      "grad_norm": 18.0,
      "learning_rate": 2.8138786764705882e-05,
      "loss": 3.0819,
      "mean_token_accuracy": 0.7513671040534973,
      "step": 270,
      "train/kl_loss_qwen": 0.3243158679455519,
      "train/kl_loss_r1": 0.21789413895457982,
      "train/total_kl": 0.5422100141644478
    },
    {
      "epoch": 0.25275735294117646,
      "grad_norm": 6.09375,
      "learning_rate": 2.8104319852941175e-05,
      "loss": 2.3684,
      "mean_token_accuracy": 0.7698420792818069,
      "step": 275,
      "train/kl_loss_qwen": 0.20914174802601337,
      "train/kl_loss_r1": 0.12488835342228413,
      "train/total_kl": 0.33403009958565233
    },
    {
      "epoch": 0.25735294117647056,
      "grad_norm": 35.5,
      "learning_rate": 2.806985294117647e-05,
      "loss": 2.9746,
      "mean_token_accuracy": 0.7567938059568405,
      "step": 280,
      "train/kl_loss_qwen": 0.2870381150394678,
      "train/kl_loss_r1": 0.21586436461657285,
      "train/total_kl": 0.5029024768620729
    },
    {
      "epoch": 0.2619485294117647,
      "grad_norm": 10.25,
      "learning_rate": 2.8035386029411767e-05,
      "loss": 2.7672,
      "mean_token_accuracy": 0.758710753917694,
      "step": 285,
      "train/kl_loss_qwen": 0.28009881228208544,
      "train/kl_loss_r1": 0.16334772501140832,
      "train/total_kl": 0.4434465430676937
    },
    {
      "epoch": 0.2665441176470588,
      "grad_norm": 13.0625,
      "learning_rate": 2.800091911764706e-05,
      "loss": 2.9307,
      "mean_token_accuracy": 0.7504922091960907,
      "step": 290,
      "train/kl_loss_qwen": 0.3210148632526398,
      "train/kl_loss_r1": 0.1687827054411173,
      "train/total_kl": 0.48979756236076355
    },
    {
      "epoch": 0.2711397058823529,
      "grad_norm": 15.0625,
      "learning_rate": 2.7966452205882355e-05,
      "loss": 2.7678,
      "mean_token_accuracy": 0.7582155168056488,
      "step": 295,
      "train/kl_loss_qwen": 0.291231270134449,
      "train/kl_loss_r1": 0.1674706531688571,
      "train/total_kl": 0.4587019272148609
    },
    {
      "epoch": 0.2757352941176471,
      "grad_norm": 10.1875,
      "learning_rate": 2.7931985294117648e-05,
      "loss": 2.4064,
      "mean_token_accuracy": 0.748086565732956,
      "step": 300,
      "train/kl_loss_qwen": 0.21337708421051502,
      "train/kl_loss_r1": 0.11356487087905406,
      "train/total_kl": 0.3269419576972723
    },
    {
      "epoch": 0.2803308823529412,
      "grad_norm": 12.0,
      "learning_rate": 2.7897518382352943e-05,
      "loss": 3.3713,
      "mean_token_accuracy": 0.7676705598831177,
      "step": 305,
      "train/kl_loss_qwen": 0.38383147977292537,
      "train/kl_loss_r1": 0.23011279180645944,
      "train/total_kl": 0.6139442846179008
    },
    {
      "epoch": 0.2849264705882353,
      "grad_norm": 13.0,
      "learning_rate": 2.7863051470588236e-05,
      "loss": 2.3748,
      "mean_token_accuracy": 0.7754788219928741,
      "step": 310,
      "train/kl_loss_qwen": 0.2166389824822545,
      "train/kl_loss_r1": 0.12619577907025814,
      "train/total_kl": 0.34283475764095783
    },
    {
      "epoch": 0.28952205882352944,
      "grad_norm": 14.3125,
      "learning_rate": 2.782858455882353e-05,
      "loss": 2.7837,
      "mean_token_accuracy": 0.7562883973121644,
      "step": 315,
      "train/kl_loss_qwen": 0.2925012856721878,
      "train/kl_loss_r1": 0.1582660662010312,
      "train/total_kl": 0.45076735615730285
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 24.75,
      "learning_rate": 2.7794117647058824e-05,
      "loss": 2.8939,
      "mean_token_accuracy": 0.7451714336872101,
      "step": 320,
      "train/kl_loss_qwen": 0.2792695712298155,
      "train/kl_loss_r1": 0.18812930770218372,
      "train/total_kl": 0.4673988774418831
    },
    {
      "epoch": 0.29871323529411764,
      "grad_norm": 34.0,
      "learning_rate": 2.7759650735294117e-05,
      "loss": 3.1971,
      "mean_token_accuracy": 0.7753181874752044,
      "step": 325,
      "train/kl_loss_qwen": 0.3638820182532072,
      "train/kl_loss_r1": 0.20677912458777428,
      "train/total_kl": 0.5706611409783363
    },
    {
      "epoch": 0.30330882352941174,
      "grad_norm": 23.5,
      "learning_rate": 2.7725183823529413e-05,
      "loss": 2.5968,
      "mean_token_accuracy": 0.7559898108243942,
      "step": 330,
      "train/kl_loss_qwen": 0.24033832624554635,
      "train/kl_loss_r1": 0.15211960580199957,
      "train/total_kl": 0.3924579367041588
    },
    {
      "epoch": 0.3079044117647059,
      "grad_norm": 12.1875,
      "learning_rate": 2.7690716911764705e-05,
      "loss": 3.1609,
      "mean_token_accuracy": 0.7667332828044892,
      "step": 335,
      "train/kl_loss_qwen": 0.338285806030035,
      "train/kl_loss_r1": 0.21732427142560481,
      "train/total_kl": 0.5556100673973561
    },
    {
      "epoch": 0.3125,
      "grad_norm": 8.6875,
      "learning_rate": 2.765625e-05,
      "loss": 3.4132,
      "mean_token_accuracy": 0.7583595246076584,
      "step": 340,
      "train/kl_loss_qwen": 0.3800947768613696,
      "train/kl_loss_r1": 0.25115054063498976,
      "train/total_kl": 0.6312453065067529
    },
    {
      "epoch": 0.3170955882352941,
      "grad_norm": 8.125,
      "learning_rate": 2.7621783088235297e-05,
      "loss": 2.7595,
      "mean_token_accuracy": 0.7543198734521865,
      "step": 345,
      "train/kl_loss_qwen": 0.26038931421935557,
      "train/kl_loss_r1": 0.17919862512499093,
      "train/total_kl": 0.43958794698119164
    },
    {
      "epoch": 0.32169117647058826,
      "grad_norm": 9.0,
      "learning_rate": 2.758731617647059e-05,
      "loss": 2.8555,
      "mean_token_accuracy": 0.7692651629447937,
      "step": 350,
      "train/kl_loss_qwen": 0.30205008797347543,
      "train/kl_loss_r1": 0.17516350764781236,
      "train/total_kl": 0.47721359729766843
    },
    {
      "epoch": 0.32628676470588236,
      "grad_norm": 14.5,
      "learning_rate": 2.7552849264705882e-05,
      "loss": 2.6488,
      "mean_token_accuracy": 0.7590999722480773,
      "step": 355,
      "train/kl_loss_qwen": 0.25812953151762486,
      "train/kl_loss_r1": 0.15315534509718418,
      "train/total_kl": 0.4112848773598671
    },
    {
      "epoch": 0.33088235294117646,
      "grad_norm": 18.875,
      "learning_rate": 2.7518382352941175e-05,
      "loss": 2.8598,
      "mean_token_accuracy": 0.774398872256279,
      "step": 360,
      "train/kl_loss_qwen": 0.2998722895979881,
      "train/kl_loss_r1": 0.1808535248041153,
      "train/total_kl": 0.48072580993175507
    },
    {
      "epoch": 0.33547794117647056,
      "grad_norm": 15.5625,
      "learning_rate": 2.748391544117647e-05,
      "loss": 3.5063,
      "mean_token_accuracy": 0.7680695742368698,
      "step": 365,
      "train/kl_loss_qwen": 0.3873132593929768,
      "train/kl_loss_r1": 0.26921150721609594,
      "train/total_kl": 0.6565247721970081
    },
    {
      "epoch": 0.3400735294117647,
      "grad_norm": 11.6875,
      "learning_rate": 2.7449448529411767e-05,
      "loss": 3.8119,
      "mean_token_accuracy": 0.7418938785791397,
      "step": 370,
      "train/kl_loss_qwen": 0.47169247530400754,
      "train/kl_loss_r1": 0.26214057188481094,
      "train/total_kl": 0.7338330511003732
    },
    {
      "epoch": 0.3446691176470588,
      "grad_norm": 8.625,
      "learning_rate": 2.741498161764706e-05,
      "loss": 3.0413,
      "mean_token_accuracy": 0.7195575445890426,
      "step": 375,
      "train/kl_loss_qwen": 0.3134828630834818,
      "train/kl_loss_r1": 0.18996739368885757,
      "train/total_kl": 0.5034502573311329
    },
    {
      "epoch": 0.3492647058823529,
      "grad_norm": 21.875,
      "learning_rate": 2.7380514705882355e-05,
      "loss": 2.7333,
      "mean_token_accuracy": 0.7529161244630813,
      "step": 380,
      "train/kl_loss_qwen": 0.26783917360007764,
      "train/kl_loss_r1": 0.16146638356149196,
      "train/total_kl": 0.42930555492639544
    },
    {
      "epoch": 0.3538602941176471,
      "grad_norm": 23.5,
      "learning_rate": 2.7346047794117648e-05,
      "loss": 2.9696,
      "mean_token_accuracy": 0.7752228945493698,
      "step": 385,
      "train/kl_loss_qwen": 0.28993729297071696,
      "train/kl_loss_r1": 0.21835732236504554,
      "train/total_kl": 0.5082946226000786
    },
    {
      "epoch": 0.3584558823529412,
      "grad_norm": 26.5,
      "learning_rate": 2.7311580882352944e-05,
      "loss": 2.4825,
      "mean_token_accuracy": 0.739210131764412,
      "step": 390,
      "train/kl_loss_qwen": 0.23699542619287967,
      "train/kl_loss_r1": 0.13306182324886323,
      "train/total_kl": 0.37005725055933
    },
    {
      "epoch": 0.3630514705882353,
      "grad_norm": 20.375,
      "learning_rate": 2.7277113970588233e-05,
      "loss": 3.0705,
      "mean_token_accuracy": 0.7504832357168197,
      "step": 395,
      "train/kl_loss_qwen": 0.3206694882363081,
      "train/kl_loss_r1": 0.203219941444695,
      "train/total_kl": 0.5238894239068032
    },
    {
      "epoch": 0.36764705882352944,
      "grad_norm": 7.03125,
      "learning_rate": 2.724264705882353e-05,
      "loss": 2.7817,
      "mean_token_accuracy": 0.7592732578516006,
      "step": 400,
      "train/kl_loss_qwen": 0.25980725064873694,
      "train/kl_loss_r1": 0.17893490809947252,
      "train/total_kl": 0.4387421656399965
    },
    {
      "epoch": 0.37224264705882354,
      "grad_norm": 15.8125,
      "learning_rate": 2.7208180147058825e-05,
      "loss": 2.7213,
      "mean_token_accuracy": 0.7661308318376541,
      "step": 405,
      "train/kl_loss_qwen": 0.27035542018711567,
      "train/kl_loss_r1": 0.16476453188806772,
      "train/total_kl": 0.43511994630098344
    },
    {
      "epoch": 0.37683823529411764,
      "grad_norm": 9.5625,
      "learning_rate": 2.7173713235294117e-05,
      "loss": 3.3107,
      "mean_token_accuracy": 0.7672391653060913,
      "step": 410,
      "train/kl_loss_qwen": 0.37639964669942855,
      "train/kl_loss_r1": 0.22928424850106238,
      "train/total_kl": 0.6056838996708394
    },
    {
      "epoch": 0.38143382352941174,
      "grad_norm": 6.59375,
      "learning_rate": 2.7139246323529413e-05,
      "loss": 2.8039,
      "mean_token_accuracy": 0.7417833805084229,
      "step": 415,
      "train/kl_loss_qwen": 0.28083805963397024,
      "train/kl_loss_r1": 0.18152289222925902,
      "train/total_kl": 0.46236095279455186
    },
    {
      "epoch": 0.3860294117647059,
      "grad_norm": 11.875,
      "learning_rate": 2.7104779411764706e-05,
      "loss": 2.9833,
      "mean_token_accuracy": 0.7545632183551788,
      "step": 420,
      "train/kl_loss_qwen": 0.3026548307389021,
      "train/kl_loss_r1": 0.19777862951159478,
      "train/total_kl": 0.500433461368084
    },
    {
      "epoch": 0.390625,
      "grad_norm": 7.6875,
      "learning_rate": 2.70703125e-05,
      "loss": 2.7921,
      "mean_token_accuracy": 0.7616397023200989,
      "step": 425,
      "train/kl_loss_qwen": 0.2890779687091708,
      "train/kl_loss_r1": 0.17925369031727315,
      "train/total_kl": 0.4683316584676504
    },
    {
      "epoch": 0.3952205882352941,
      "grad_norm": 5.5625,
      "learning_rate": 2.7035845588235297e-05,
      "loss": 2.7383,
      "mean_token_accuracy": 0.7346738040447235,
      "step": 430,
      "train/kl_loss_qwen": 0.28430320098996165,
      "train/kl_loss_r1": 0.16884665340185165,
      "train/total_kl": 0.45314985513687134
    },
    {
      "epoch": 0.39981617647058826,
      "grad_norm": 11.8125,
      "learning_rate": 2.700137867647059e-05,
      "loss": 3.2727,
      "mean_token_accuracy": 0.764810037612915,
      "step": 435,
      "train/kl_loss_qwen": 0.36565163880586626,
      "train/kl_loss_r1": 0.22558138892054558,
      "train/total_kl": 0.5912330389022827
    },
    {
      "epoch": 0.40441176470588236,
      "grad_norm": 8.125,
      "learning_rate": 2.6966911764705883e-05,
      "loss": 3.3381,
      "mean_token_accuracy": 0.7699699103832245,
      "step": 440,
      "train/kl_loss_qwen": 0.35228237994015216,
      "train/kl_loss_r1": 0.2567650653421879,
      "train/total_kl": 0.609047444164753
    },
    {
      "epoch": 0.40900735294117646,
      "grad_norm": 19.5,
      "learning_rate": 2.6932444852941175e-05,
      "loss": 3.0697,
      "mean_token_accuracy": 0.7416951596736908,
      "step": 445,
      "train/kl_loss_qwen": 0.33608091417700053,
      "train/kl_loss_r1": 0.20944351907819508,
      "train/total_kl": 0.5455244343727828
    },
    {
      "epoch": 0.41360294117647056,
      "grad_norm": 16.625,
      "learning_rate": 2.689797794117647e-05,
      "loss": 2.7871,
      "mean_token_accuracy": 0.7526424586772918,
      "step": 450,
      "train/kl_loss_qwen": 0.30512944534420966,
      "train/kl_loss_r1": 0.15808237809687853,
      "train/total_kl": 0.46321182437241076
    },
    {
      "epoch": 0.4181985294117647,
      "grad_norm": 7.375,
      "learning_rate": 2.6863511029411767e-05,
      "loss": 2.8216,
      "mean_token_accuracy": 0.7598502576351166,
      "step": 455,
      "train/kl_loss_qwen": 0.27531296089291574,
      "train/kl_loss_r1": 0.19029796570539476,
      "train/total_kl": 0.46561092883348465
    },
    {
      "epoch": 0.4227941176470588,
      "grad_norm": 9.0625,
      "learning_rate": 2.682904411764706e-05,
      "loss": 2.7403,
      "mean_token_accuracy": 0.7585819303989411,
      "step": 460,
      "train/kl_loss_qwen": 0.2825481027364731,
      "train/kl_loss_r1": 0.17290859669446945,
      "train/total_kl": 0.45545670166611674
    },
    {
      "epoch": 0.4273897058823529,
      "grad_norm": 5.625,
      "learning_rate": 2.6794577205882355e-05,
      "loss": 3.6634,
      "mean_token_accuracy": 0.763472780585289,
      "step": 465,
      "train/kl_loss_qwen": 0.42548854760825633,
      "train/kl_loss_r1": 0.2631331071257591,
      "train/total_kl": 0.6886216450482607
    },
    {
      "epoch": 0.4319852941176471,
      "grad_norm": 7.84375,
      "learning_rate": 2.6760110294117648e-05,
      "loss": 2.3183,
      "mean_token_accuracy": 0.7590369611978531,
      "step": 470,
      "train/kl_loss_qwen": 0.2138752520084381,
      "train/kl_loss_r1": 0.11956894248723984,
      "train/total_kl": 0.3334441937506199
    },
    {
      "epoch": 0.4365808823529412,
      "grad_norm": 9.25,
      "learning_rate": 2.6725643382352944e-05,
      "loss": 2.7691,
      "mean_token_accuracy": 0.7740260541439057,
      "step": 475,
      "train/kl_loss_qwen": 0.28194935582578184,
      "train/kl_loss_r1": 0.17588923051953315,
      "train/total_kl": 0.4578385904431343
    },
    {
      "epoch": 0.4411764705882353,
      "grad_norm": 16.5,
      "learning_rate": 2.6691176470588233e-05,
      "loss": 2.9371,
      "mean_token_accuracy": 0.7553506731987,
      "step": 480,
      "train/kl_loss_qwen": 0.3055728547275066,
      "train/kl_loss_r1": 0.19807562176138163,
      "train/total_kl": 0.5036484815180302
    },
    {
      "epoch": 0.44577205882352944,
      "grad_norm": 8.25,
      "learning_rate": 2.665670955882353e-05,
      "loss": 2.564,
      "mean_token_accuracy": 0.7415809512138367,
      "step": 485,
      "train/kl_loss_qwen": 0.24062816128134729,
      "train/kl_loss_r1": 0.1463530447334051,
      "train/total_kl": 0.38698121197521684
    },
    {
      "epoch": 0.45036764705882354,
      "grad_norm": 12.5,
      "learning_rate": 2.6622242647058825e-05,
      "loss": 2.9048,
      "mean_token_accuracy": 0.7778663486242294,
      "step": 490,
      "train/kl_loss_qwen": 0.30084225051105024,
      "train/kl_loss_r1": 0.19928387235850095,
      "train/total_kl": 0.5001261174678803
    },
    {
      "epoch": 0.45496323529411764,
      "grad_norm": 8.4375,
      "learning_rate": 2.6587775735294117e-05,
      "loss": 3.25,
      "mean_token_accuracy": 0.7694221258163452,
      "step": 495,
      "train/kl_loss_qwen": 0.3652114329859614,
      "train/kl_loss_r1": 0.22170483991503714,
      "train/total_kl": 0.5869162768125534
    },
    {
      "epoch": 0.45955882352941174,
      "grad_norm": 33.25,
      "learning_rate": 2.6553308823529413e-05,
      "loss": 3.0252,
      "mean_token_accuracy": 0.7460946381092072,
      "step": 500,
      "train/kl_loss_qwen": 0.30894157737493516,
      "train/kl_loss_r1": 0.20898498129099607,
      "train/total_kl": 0.5179265588521957
    },
    {
      "epoch": 0.4641544117647059,
      "grad_norm": 8.1875,
      "learning_rate": 2.6518841911764706e-05,
      "loss": 2.8175,
      "mean_token_accuracy": 0.7537802934646607,
      "step": 505,
      "train/kl_loss_qwen": 0.27506737038493156,
      "train/kl_loss_r1": 0.17378015574067832,
      "train/total_kl": 0.4488475173711777
    },
    {
      "epoch": 0.46875,
      "grad_norm": 6.4375,
      "learning_rate": 2.6484375000000002e-05,
      "loss": 2.3492,
      "mean_token_accuracy": 0.7604597806930542,
      "step": 510,
      "train/kl_loss_qwen": 0.21084308940917254,
      "train/kl_loss_r1": 0.12702357918024063,
      "train/total_kl": 0.3378666639328003
    },
    {
      "epoch": 0.4733455882352941,
      "grad_norm": 5.28125,
      "learning_rate": 2.6449908088235298e-05,
      "loss": 2.4155,
      "mean_token_accuracy": 0.7750326544046402,
      "step": 515,
      "train/kl_loss_qwen": 0.22967448085546494,
      "train/kl_loss_r1": 0.14129504952579736,
      "train/total_kl": 0.37096953131258487
    },
    {
      "epoch": 0.47794117647058826,
      "grad_norm": 7.6875,
      "learning_rate": 2.6415441176470587e-05,
      "loss": 2.4723,
      "mean_token_accuracy": 0.7769854992628098,
      "step": 520,
      "train/kl_loss_qwen": 0.2410843674093485,
      "train/kl_loss_r1": 0.14776650853455067,
      "train/total_kl": 0.38885087966918946
    },
    {
      "epoch": 0.48253676470588236,
      "grad_norm": 6.40625,
      "learning_rate": 2.6380974264705883e-05,
      "loss": 2.6232,
      "mean_token_accuracy": 0.7474688351154327,
      "step": 525,
      "train/kl_loss_qwen": 0.273803885653615,
      "train/kl_loss_r1": 0.15615152288228273,
      "train/total_kl": 0.42995540648698805
    },
    {
      "epoch": 0.48713235294117646,
      "grad_norm": 11.4375,
      "learning_rate": 2.6346507352941175e-05,
      "loss": 2.5732,
      "mean_token_accuracy": 0.7471906542778015,
      "step": 530,
      "train/kl_loss_qwen": 0.24260313846170903,
      "train/kl_loss_r1": 0.14697427842766048,
      "train/total_kl": 0.38957741148769853
    },
    {
      "epoch": 0.49172794117647056,
      "grad_norm": 9.4375,
      "learning_rate": 2.631204044117647e-05,
      "loss": 2.4132,
      "mean_token_accuracy": 0.7701083302497864,
      "step": 535,
      "train/kl_loss_qwen": 0.2194656167179346,
      "train/kl_loss_r1": 0.13061367962509393,
      "train/total_kl": 0.35007930025458334
    },
    {
      "epoch": 0.4963235294117647,
      "grad_norm": 14.1875,
      "learning_rate": 2.6277573529411767e-05,
      "loss": 2.8059,
      "mean_token_accuracy": 0.7791450679302215,
      "step": 540,
      "train/kl_loss_qwen": 0.2814615525305271,
      "train/kl_loss_r1": 0.18742658831179143,
      "train/total_kl": 0.468888146430254
    },
    {
      "epoch": 0.5009191176470589,
      "grad_norm": 6.09375,
      "learning_rate": 2.624310661764706e-05,
      "loss": 2.5685,
      "mean_token_accuracy": 0.7711760759353637,
      "step": 545,
      "train/kl_loss_qwen": 0.2479680348187685,
      "train/kl_loss_r1": 0.15079462714493275,
      "train/total_kl": 0.3987626597285271
    },
    {
      "epoch": 0.5055147058823529,
      "grad_norm": 13.9375,
      "learning_rate": 2.6208639705882356e-05,
      "loss": 2.9214,
      "mean_token_accuracy": 0.777749553322792,
      "step": 550,
      "train/kl_loss_qwen": 0.3169623063877225,
      "train/kl_loss_r1": 0.18846932239830494,
      "train/total_kl": 0.5054316341876983
    },
    {
      "epoch": 0.5101102941176471,
      "grad_norm": 10.9375,
      "learning_rate": 2.6174172794117648e-05,
      "loss": 3.4257,
      "mean_token_accuracy": 0.7404684275388718,
      "step": 555,
      "train/kl_loss_qwen": 0.3698401369154453,
      "train/kl_loss_r1": 0.25148618761450053,
      "train/total_kl": 0.6213263235986233
    },
    {
      "epoch": 0.5147058823529411,
      "grad_norm": 11.125,
      "learning_rate": 2.613970588235294e-05,
      "loss": 2.2958,
      "mean_token_accuracy": 0.7467740803956986,
      "step": 560,
      "train/kl_loss_qwen": 0.1958090879023075,
      "train/kl_loss_r1": 0.12171240914613009,
      "train/total_kl": 0.31752149723470213
    },
    {
      "epoch": 0.5193014705882353,
      "grad_norm": 6.25,
      "learning_rate": 2.6105238970588233e-05,
      "loss": 3.0122,
      "mean_token_accuracy": 0.7664730340242386,
      "step": 565,
      "train/kl_loss_qwen": 0.3262002274394035,
      "train/kl_loss_r1": 0.19693421963602303,
      "train/total_kl": 0.5231344491243363
    },
    {
      "epoch": 0.5238970588235294,
      "grad_norm": 5.9375,
      "learning_rate": 2.607077205882353e-05,
      "loss": 2.6851,
      "mean_token_accuracy": 0.7639783143997192,
      "step": 570,
      "train/kl_loss_qwen": 0.2612082902342081,
      "train/kl_loss_r1": 0.16572732012718916,
      "train/total_kl": 0.42693560793995855
    },
    {
      "epoch": 0.5284926470588235,
      "grad_norm": 6.25,
      "learning_rate": 2.6036305147058825e-05,
      "loss": 2.6604,
      "mean_token_accuracy": 0.7705452233552933,
      "step": 575,
      "train/kl_loss_qwen": 0.272306926548481,
      "train/kl_loss_r1": 0.15905672162771226,
      "train/total_kl": 0.4313636474311352
    },
    {
      "epoch": 0.5330882352941176,
      "grad_norm": 8.4375,
      "learning_rate": 2.6001838235294117e-05,
      "loss": 3.255,
      "mean_token_accuracy": 0.7644341349601745,
      "step": 580,
      "train/kl_loss_qwen": 0.3845398426055908,
      "train/kl_loss_r1": 0.1971276355907321,
      "train/total_kl": 0.581667472794652
    },
    {
      "epoch": 0.5376838235294118,
      "grad_norm": 5.46875,
      "learning_rate": 2.5967371323529413e-05,
      "loss": 2.321,
      "mean_token_accuracy": 0.73127601146698,
      "step": 585,
      "train/kl_loss_qwen": 0.18589173927903174,
      "train/kl_loss_r1": 0.11133090611547232,
      "train/total_kl": 0.2972226433455944
    },
    {
      "epoch": 0.5422794117647058,
      "grad_norm": 8.1875,
      "learning_rate": 2.5932904411764706e-05,
      "loss": 3.1372,
      "mean_token_accuracy": 0.766996106505394,
      "step": 590,
      "train/kl_loss_qwen": 0.32438481897115706,
      "train/kl_loss_r1": 0.23537467941641807,
      "train/total_kl": 0.5597595013678074
    },
    {
      "epoch": 0.546875,
      "grad_norm": 14.5,
      "learning_rate": 2.5898437500000002e-05,
      "loss": 2.838,
      "mean_token_accuracy": 0.7733567923307418,
      "step": 595,
      "train/kl_loss_qwen": 0.3003999553620815,
      "train/kl_loss_r1": 0.18438877165317535,
      "train/total_kl": 0.4847887299954891
    },
    {
      "epoch": 0.5514705882352942,
      "grad_norm": 8.1875,
      "learning_rate": 2.5863970588235298e-05,
      "loss": 3.119,
      "mean_token_accuracy": 0.7530891060829162,
      "step": 600,
      "train/kl_loss_qwen": 0.3513902911916375,
      "train/kl_loss_r1": 0.2060710046440363,
      "train/total_kl": 0.5574612963944674
    },
    {
      "epoch": 0.5560661764705882,
      "grad_norm": 7.375,
      "learning_rate": 2.5829503676470587e-05,
      "loss": 3.1272,
      "mean_token_accuracy": 0.7233130037784576,
      "step": 605,
      "train/kl_loss_qwen": 0.32725715525448323,
      "train/kl_loss_r1": 0.21415164284408092,
      "train/total_kl": 0.5414088018238544
    },
    {
      "epoch": 0.5606617647058824,
      "grad_norm": 7.78125,
      "learning_rate": 2.5795036764705883e-05,
      "loss": 2.65,
      "mean_token_accuracy": 0.7488237828016281,
      "step": 610,
      "train/kl_loss_qwen": 0.25779034085571767,
      "train/kl_loss_r1": 0.15438018105924128,
      "train/total_kl": 0.4121705275028944
    },
    {
      "epoch": 0.5652573529411765,
      "grad_norm": 7.1875,
      "learning_rate": 2.5760569852941175e-05,
      "loss": 3.1002,
      "mean_token_accuracy": 0.7656900107860565,
      "step": 615,
      "train/kl_loss_qwen": 0.32317040748894216,
      "train/kl_loss_r1": 0.22581094168126584,
      "train/total_kl": 0.5489813484251499
    },
    {
      "epoch": 0.5698529411764706,
      "grad_norm": 10.3125,
      "learning_rate": 2.572610294117647e-05,
      "loss": 3.6832,
      "mean_token_accuracy": 0.7709728151559829,
      "step": 620,
      "train/kl_loss_qwen": 0.43695655167102815,
      "train/kl_loss_r1": 0.26164604611694814,
      "train/total_kl": 0.6986026015132666
    },
    {
      "epoch": 0.5744485294117647,
      "grad_norm": 9.75,
      "learning_rate": 2.5691636029411767e-05,
      "loss": 2.8011,
      "mean_token_accuracy": 0.7490771561861038,
      "step": 625,
      "train/kl_loss_qwen": 0.27099971994757655,
      "train/kl_loss_r1": 0.1779295329004526,
      "train/total_kl": 0.4489292375743389
    },
    {
      "epoch": 0.5790441176470589,
      "grad_norm": 5.34375,
      "learning_rate": 2.565716911764706e-05,
      "loss": 2.5584,
      "mean_token_accuracy": 0.7660718470811844,
      "step": 630,
      "train/kl_loss_qwen": 0.23552559316158295,
      "train/kl_loss_r1": 0.14556924644857644,
      "train/total_kl": 0.3810948390513659
    },
    {
      "epoch": 0.5836397058823529,
      "grad_norm": 6.96875,
      "learning_rate": 2.5622702205882356e-05,
      "loss": 2.8204,
      "mean_token_accuracy": 0.7795866638422012,
      "step": 635,
      "train/kl_loss_qwen": 0.2912254322320223,
      "train/kl_loss_r1": 0.18594579454511403,
      "train/total_kl": 0.477171229571104
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 26.375,
      "learning_rate": 2.5588235294117648e-05,
      "loss": 3.5962,
      "mean_token_accuracy": 0.7579382926225662,
      "step": 640,
      "train/kl_loss_qwen": 0.4152888409793377,
      "train/kl_loss_r1": 0.2703390533104539,
      "train/total_kl": 0.6856279008090496
    },
    {
      "epoch": 0.5928308823529411,
      "grad_norm": 6.5625,
      "learning_rate": 2.555376838235294e-05,
      "loss": 2.2892,
      "mean_token_accuracy": 0.7371518105268479,
      "step": 645,
      "train/kl_loss_qwen": 0.2000577926635742,
      "train/kl_loss_r1": 0.1173404160887003,
      "train/total_kl": 0.3173982098698616
    },
    {
      "epoch": 0.5974264705882353,
      "grad_norm": 5.34375,
      "learning_rate": 2.5519301470588233e-05,
      "loss": 2.9149,
      "mean_token_accuracy": 0.7631843984127045,
      "step": 650,
      "train/kl_loss_qwen": 0.30258313585072755,
      "train/kl_loss_r1": 0.17346918024122715,
      "train/total_kl": 0.476052313297987
    },
    {
      "epoch": 0.6020220588235294,
      "grad_norm": 5.625,
      "learning_rate": 2.548483455882353e-05,
      "loss": 2.3347,
      "mean_token_accuracy": 0.7706077218055725,
      "step": 655,
      "train/kl_loss_qwen": 0.21414661556482315,
      "train/kl_loss_r1": 0.12977762762457132,
      "train/total_kl": 0.3439242452383041
    },
    {
      "epoch": 0.6066176470588235,
      "grad_norm": 18.875,
      "learning_rate": 2.5450367647058825e-05,
      "loss": 3.5523,
      "mean_token_accuracy": 0.7638006836175919,
      "step": 660,
      "train/kl_loss_qwen": 0.4221733514219522,
      "train/kl_loss_r1": 0.2539014684036374,
      "train/total_kl": 0.6760748248547316
    },
    {
      "epoch": 0.6112132352941176,
      "grad_norm": 20.375,
      "learning_rate": 2.5415900735294118e-05,
      "loss": 2.58,
      "mean_token_accuracy": 0.7460006952285767,
      "step": 665,
      "train/kl_loss_qwen": 0.25217038616538046,
      "train/kl_loss_r1": 0.1543192882090807,
      "train/total_kl": 0.40648967288434507
    },
    {
      "epoch": 0.6158088235294118,
      "grad_norm": 6.4375,
      "learning_rate": 2.5381433823529414e-05,
      "loss": 2.7286,
      "mean_token_accuracy": 0.7695958316326141,
      "step": 670,
      "train/kl_loss_qwen": 0.28523584045469763,
      "train/kl_loss_r1": 0.16746240444481372,
      "train/total_kl": 0.45269824415445326
    },
    {
      "epoch": 0.6204044117647058,
      "grad_norm": 9.125,
      "learning_rate": 2.5346966911764706e-05,
      "loss": 2.7487,
      "mean_token_accuracy": 0.7600745111703873,
      "step": 675,
      "train/kl_loss_qwen": 0.26131963096559047,
      "train/kl_loss_r1": 0.18360938951373101,
      "train/total_kl": 0.4449290160089731
    },
    {
      "epoch": 0.625,
      "grad_norm": 5.6875,
      "learning_rate": 2.5312500000000002e-05,
      "loss": 2.5675,
      "mean_token_accuracy": 0.7691577523946762,
      "step": 680,
      "train/kl_loss_qwen": 0.24630658701062202,
      "train/kl_loss_r1": 0.15804811622947454,
      "train/total_kl": 0.404354702308774
    },
    {
      "epoch": 0.6295955882352942,
      "grad_norm": 7.90625,
      "learning_rate": 2.5278033088235295e-05,
      "loss": 2.9306,
      "mean_token_accuracy": 0.7747731059789658,
      "step": 685,
      "train/kl_loss_qwen": 0.3065579190850258,
      "train/kl_loss_r1": 0.1966759165748954,
      "train/total_kl": 0.5032338403165341
    },
    {
      "epoch": 0.6341911764705882,
      "grad_norm": 14.25,
      "learning_rate": 2.5243566176470587e-05,
      "loss": 3.0556,
      "mean_token_accuracy": 0.7747425675392151,
      "step": 690,
      "train/kl_loss_qwen": 0.34722540378570554,
      "train/kl_loss_r1": 0.20382858254015446,
      "train/total_kl": 0.5510539896786213
    },
    {
      "epoch": 0.6387867647058824,
      "grad_norm": 8.25,
      "learning_rate": 2.5209099264705883e-05,
      "loss": 2.9003,
      "mean_token_accuracy": 0.7529223769903183,
      "step": 695,
      "train/kl_loss_qwen": 0.30322663113474846,
      "train/kl_loss_r1": 0.1966948075219989,
      "train/total_kl": 0.4999214388430119
    },
    {
      "epoch": 0.6433823529411765,
      "grad_norm": 20.625,
      "learning_rate": 2.5174632352941176e-05,
      "loss": 2.7603,
      "mean_token_accuracy": 0.7450964391231537,
      "step": 700,
      "train/kl_loss_qwen": 0.2775906875729561,
      "train/kl_loss_r1": 0.1681121250614524,
      "train/total_kl": 0.44570281244814397
    },
    {
      "epoch": 0.6479779411764706,
      "grad_norm": 7.46875,
      "learning_rate": 2.514016544117647e-05,
      "loss": 2.2999,
      "mean_token_accuracy": 0.7612766295671463,
      "step": 705,
      "train/kl_loss_qwen": 0.20588579438626767,
      "train/kl_loss_r1": 0.11731699910014867,
      "train/total_kl": 0.3232027903199196
    },
    {
      "epoch": 0.6525735294117647,
      "grad_norm": 24.25,
      "learning_rate": 2.5105698529411767e-05,
      "loss": 3.0425,
      "mean_token_accuracy": 0.7618170917034149,
      "step": 710,
      "train/kl_loss_qwen": 0.33920769318938254,
      "train/kl_loss_r1": 0.20069157797843218,
      "train/total_kl": 0.5398992709815502
    },
    {
      "epoch": 0.6571691176470589,
      "grad_norm": 6.34375,
      "learning_rate": 2.507123161764706e-05,
      "loss": 3.3906,
      "mean_token_accuracy": 0.7389930814504624,
      "step": 715,
      "train/kl_loss_qwen": 0.3495714729651809,
      "train/kl_loss_r1": 0.2725962562486529,
      "train/total_kl": 0.622167743742466
    },
    {
      "epoch": 0.6617647058823529,
      "grad_norm": 13.625,
      "learning_rate": 2.5036764705882356e-05,
      "loss": 2.5055,
      "mean_token_accuracy": 0.7673314332962036,
      "step": 720,
      "train/kl_loss_qwen": 0.24460462778806685,
      "train/kl_loss_r1": 0.15297702196985483,
      "train/total_kl": 0.39758165180683136
    },
    {
      "epoch": 0.6663602941176471,
      "grad_norm": 8.0,
      "learning_rate": 2.500229779411765e-05,
      "loss": 2.9579,
      "mean_token_accuracy": 0.7666648864746094,
      "step": 725,
      "train/kl_loss_qwen": 0.3130575641989708,
      "train/kl_loss_r1": 0.20369201358407735,
      "train/total_kl": 0.5167495809495449
    },
    {
      "epoch": 0.6709558823529411,
      "grad_norm": 32.25,
      "learning_rate": 2.496783088235294e-05,
      "loss": 3.303,
      "mean_token_accuracy": 0.7632827788591385,
      "step": 730,
      "train/kl_loss_qwen": 0.37421940360218287,
      "train/kl_loss_r1": 0.23571279421448707,
      "train/total_kl": 0.6099322006106377
    },
    {
      "epoch": 0.6755514705882353,
      "grad_norm": 6.0625,
      "learning_rate": 2.4933363970588233e-05,
      "loss": 2.3364,
      "mean_token_accuracy": 0.7545148015022278,
      "step": 735,
      "train/kl_loss_qwen": 0.20344098135828972,
      "train/kl_loss_r1": 0.12272758428007365,
      "train/total_kl": 0.3261685647070408
    },
    {
      "epoch": 0.6801470588235294,
      "grad_norm": 5.625,
      "learning_rate": 2.489889705882353e-05,
      "loss": 2.5848,
      "mean_token_accuracy": 0.773073548078537,
      "step": 740,
      "train/kl_loss_qwen": 0.24243146367371082,
      "train/kl_loss_r1": 0.1647475201636553,
      "train/total_kl": 0.4071789853274822
    },
    {
      "epoch": 0.6847426470588235,
      "grad_norm": 5.71875,
      "learning_rate": 2.4864430147058825e-05,
      "loss": 3.2043,
      "mean_token_accuracy": 0.7741522669792176,
      "step": 745,
      "train/kl_loss_qwen": 0.34202426187694074,
      "train/kl_loss_r1": 0.23942313343286514,
      "train/total_kl": 0.5814473945647478
    },
    {
      "epoch": 0.6893382352941176,
      "grad_norm": 16.875,
      "learning_rate": 2.4829963235294118e-05,
      "loss": 3.1311,
      "mean_token_accuracy": 0.759959951043129,
      "step": 750,
      "train/kl_loss_qwen": 0.33796292170882225,
      "train/kl_loss_r1": 0.21734277866780757,
      "train/total_kl": 0.5553057052195072
    },
    {
      "epoch": 0.6939338235294118,
      "grad_norm": 6.125,
      "learning_rate": 2.4795496323529414e-05,
      "loss": 2.2645,
      "mean_token_accuracy": 0.729770639538765,
      "step": 755,
      "train/kl_loss_qwen": 0.19255626071244478,
      "train/kl_loss_r1": 0.11135205067694187,
      "train/total_kl": 0.30390830971300603
    },
    {
      "epoch": 0.6985294117647058,
      "grad_norm": 12.875,
      "learning_rate": 2.4761029411764706e-05,
      "loss": 2.7247,
      "mean_token_accuracy": 0.767203563451767,
      "step": 760,
      "train/kl_loss_qwen": 0.2671534739434719,
      "train/kl_loss_r1": 0.16766709443181754,
      "train/total_kl": 0.43482056371867656
    },
    {
      "epoch": 0.703125,
      "grad_norm": 21.5,
      "learning_rate": 2.4726562500000002e-05,
      "loss": 3.0395,
      "mean_token_accuracy": 0.7792555302381515,
      "step": 765,
      "train/kl_loss_qwen": 0.3042921770364046,
      "train/kl_loss_r1": 0.23475912101566793,
      "train/total_kl": 0.5390512943267822
    },
    {
      "epoch": 0.7077205882352942,
      "grad_norm": 18.25,
      "learning_rate": 2.4692095588235295e-05,
      "loss": 3.132,
      "mean_token_accuracy": 0.7637085497379303,
      "step": 770,
      "train/kl_loss_qwen": 0.3571842934936285,
      "train/kl_loss_r1": 0.2055378383025527,
      "train/total_kl": 0.5627221390604973
    },
    {
      "epoch": 0.7123161764705882,
      "grad_norm": 7.53125,
      "learning_rate": 2.4657628676470587e-05,
      "loss": 2.8827,
      "mean_token_accuracy": 0.7434431910514832,
      "step": 775,
      "train/kl_loss_qwen": 0.3152502715587616,
      "train/kl_loss_r1": 0.18244544900953769,
      "train/total_kl": 0.4976957157254219
    },
    {
      "epoch": 0.7169117647058824,
      "grad_norm": 20.5,
      "learning_rate": 2.4623161764705883e-05,
      "loss": 2.8948,
      "mean_token_accuracy": 0.7385075211524963,
      "step": 780,
      "train/kl_loss_qwen": 0.301816888153553,
      "train/kl_loss_r1": 0.18196456991136073,
      "train/total_kl": 0.48378146141767503
    },
    {
      "epoch": 0.7215073529411765,
      "grad_norm": 6.03125,
      "learning_rate": 2.4588694852941176e-05,
      "loss": 2.5771,
      "mean_token_accuracy": 0.753093707561493,
      "step": 785,
      "train/kl_loss_qwen": 0.24673266094177962,
      "train/kl_loss_r1": 0.15335584823042153,
      "train/total_kl": 0.4000885050743818
    },
    {
      "epoch": 0.7261029411764706,
      "grad_norm": 12.5625,
      "learning_rate": 2.455422794117647e-05,
      "loss": 2.6954,
      "mean_token_accuracy": 0.7686734408140182,
      "step": 790,
      "train/kl_loss_qwen": 0.2682993970811367,
      "train/kl_loss_r1": 0.1770503718405962,
      "train/total_kl": 0.44534977376461027
    },
    {
      "epoch": 0.7306985294117647,
      "grad_norm": 5.84375,
      "learning_rate": 2.4519761029411768e-05,
      "loss": 2.399,
      "mean_token_accuracy": 0.7646504133939743,
      "step": 795,
      "train/kl_loss_qwen": 0.2142568849027157,
      "train/kl_loss_r1": 0.13209631983190775,
      "train/total_kl": 0.34635320864617825
    },
    {
      "epoch": 0.7352941176470589,
      "grad_norm": 4.53125,
      "learning_rate": 2.448529411764706e-05,
      "loss": 2.6112,
      "mean_token_accuracy": 0.760350888967514,
      "step": 800,
      "train/kl_loss_qwen": 0.24145320765674114,
      "train/kl_loss_r1": 0.1602572899311781,
      "train/total_kl": 0.40171049572527406
    },
    {
      "epoch": 0.7398897058823529,
      "grad_norm": 5.53125,
      "learning_rate": 2.4450827205882356e-05,
      "loss": 2.7783,
      "mean_token_accuracy": 0.7784444391727448,
      "step": 805,
      "train/kl_loss_qwen": 0.2839724402874708,
      "train/kl_loss_r1": 0.18588781580328942,
      "train/total_kl": 0.46986025720834734
    },
    {
      "epoch": 0.7444852941176471,
      "grad_norm": 13.8125,
      "learning_rate": 2.4416360294117645e-05,
      "loss": 3.2653,
      "mean_token_accuracy": 0.7738717645406723,
      "step": 810,
      "train/kl_loss_qwen": 0.36536430157721045,
      "train/kl_loss_r1": 0.24171638526022435,
      "train/total_kl": 0.6070806853473186
    },
    {
      "epoch": 0.7490808823529411,
      "grad_norm": 6.625,
      "learning_rate": 2.438189338235294e-05,
      "loss": 2.3927,
      "mean_token_accuracy": 0.7411265462636948,
      "step": 815,
      "train/kl_loss_qwen": 0.21468257531523705,
      "train/kl_loss_r1": 0.1389029709622264,
      "train/total_kl": 0.35358554795384406
    },
    {
      "epoch": 0.7536764705882353,
      "grad_norm": 10.5625,
      "learning_rate": 2.4347426470588234e-05,
      "loss": 2.6545,
      "mean_token_accuracy": 0.7369672119617462,
      "step": 820,
      "train/kl_loss_qwen": 0.23823342826217414,
      "train/kl_loss_r1": 0.16688897851854562,
      "train/total_kl": 0.4051224060356617
    },
    {
      "epoch": 0.7582720588235294,
      "grad_norm": 9.5,
      "learning_rate": 2.431295955882353e-05,
      "loss": 2.221,
      "mean_token_accuracy": 0.7671397745609283,
      "step": 825,
      "train/kl_loss_qwen": 0.1944459293037653,
      "train/kl_loss_r1": 0.11296823415905237,
      "train/total_kl": 0.30741416439414027
    },
    {
      "epoch": 0.7628676470588235,
      "grad_norm": 9.8125,
      "learning_rate": 2.4278492647058825e-05,
      "loss": 2.8627,
      "mean_token_accuracy": 0.7721496760845185,
      "step": 830,
      "train/kl_loss_qwen": 0.29709434546530245,
      "train/kl_loss_r1": 0.19235175009816885,
      "train/total_kl": 0.4894461028277874
    },
    {
      "epoch": 0.7674632352941176,
      "grad_norm": 9.0625,
      "learning_rate": 2.4244025735294118e-05,
      "loss": 2.5478,
      "mean_token_accuracy": 0.777729532122612,
      "step": 835,
      "train/kl_loss_qwen": 0.24631783794611692,
      "train/kl_loss_r1": 0.16106365360319613,
      "train/total_kl": 0.40738149620592595
    },
    {
      "epoch": 0.7720588235294118,
      "grad_norm": 5.8125,
      "learning_rate": 2.4209558823529414e-05,
      "loss": 3.015,
      "mean_token_accuracy": 0.7657909542322159,
      "step": 840,
      "train/kl_loss_qwen": 0.3283080026507378,
      "train/kl_loss_r1": 0.20606368985027074,
      "train/total_kl": 0.5343716844916344
    },
    {
      "epoch": 0.7766544117647058,
      "grad_norm": 11.25,
      "learning_rate": 2.4175091911764706e-05,
      "loss": 2.4779,
      "mean_token_accuracy": 0.7260577827692032,
      "step": 845,
      "train/kl_loss_qwen": 0.224995918571949,
      "train/kl_loss_r1": 0.13734962455928326,
      "train/total_kl": 0.36234553903341293
    },
    {
      "epoch": 0.78125,
      "grad_norm": 12.4375,
      "learning_rate": 2.4140625e-05,
      "loss": 2.5439,
      "mean_token_accuracy": 0.7665543109178543,
      "step": 850,
      "train/kl_loss_qwen": 0.2431650910526514,
      "train/kl_loss_r1": 0.163537354208529,
      "train/total_kl": 0.40670243836939335
    },
    {
      "epoch": 0.7858455882352942,
      "grad_norm": 4.90625,
      "learning_rate": 2.4106158088235295e-05,
      "loss": 3.2831,
      "mean_token_accuracy": 0.7692991495132446,
      "step": 855,
      "train/kl_loss_qwen": 0.3613782159984112,
      "train/kl_loss_r1": 0.23370234090834857,
      "train/total_kl": 0.595080553740263
    },
    {
      "epoch": 0.7904411764705882,
      "grad_norm": 11.625,
      "learning_rate": 2.4071691176470587e-05,
      "loss": 3.0925,
      "mean_token_accuracy": 0.7705245614051819,
      "step": 860,
      "train/kl_loss_qwen": 0.32698389384895565,
      "train/kl_loss_r1": 0.21528607234358788,
      "train/total_kl": 0.5422699719667434
    },
    {
      "epoch": 0.7950367647058824,
      "grad_norm": 10.0,
      "learning_rate": 2.4037224264705883e-05,
      "loss": 2.5964,
      "mean_token_accuracy": 0.7585686951875686,
      "step": 865,
      "train/kl_loss_qwen": 0.24837292842566966,
      "train/kl_loss_r1": 0.1542820379137993,
      "train/total_kl": 0.40265496596693995
    },
    {
      "epoch": 0.7996323529411765,
      "grad_norm": 23.125,
      "learning_rate": 2.4002757352941176e-05,
      "loss": 2.4837,
      "mean_token_accuracy": 0.7597948163747787,
      "step": 870,
      "train/kl_loss_qwen": 0.2187556240707636,
      "train/kl_loss_r1": 0.13237967137247325,
      "train/total_kl": 0.3511353015899658
    },
    {
      "epoch": 0.8042279411764706,
      "grad_norm": 5.15625,
      "learning_rate": 2.3968290441176472e-05,
      "loss": 2.5718,
      "mean_token_accuracy": 0.7743768215179443,
      "step": 875,
      "train/kl_loss_qwen": 0.2609509777277708,
      "train/kl_loss_r1": 0.1498867817223072,
      "train/total_kl": 0.4108377553522587
    },
    {
      "epoch": 0.8088235294117647,
      "grad_norm": 7.65625,
      "learning_rate": 2.3933823529411768e-05,
      "loss": 3.0424,
      "mean_token_accuracy": 0.76705681681633,
      "step": 880,
      "train/kl_loss_qwen": 0.328360665217042,
      "train/kl_loss_r1": 0.20937101989984513,
      "train/total_kl": 0.5377316907048225
    },
    {
      "epoch": 0.8134191176470589,
      "grad_norm": 38.5,
      "learning_rate": 2.389935661764706e-05,
      "loss": 3.3681,
      "mean_token_accuracy": 0.7505431562662125,
      "step": 885,
      "train/kl_loss_qwen": 0.3830601923167706,
      "train/kl_loss_r1": 0.23940850496292115,
      "train/total_kl": 0.6224686868488789
    },
    {
      "epoch": 0.8180147058823529,
      "grad_norm": 6.78125,
      "learning_rate": 2.3864889705882356e-05,
      "loss": 2.5873,
      "mean_token_accuracy": 0.7314470678567886,
      "step": 890,
      "train/kl_loss_qwen": 0.22989343330264092,
      "train/kl_loss_r1": 0.17244479935616255,
      "train/total_kl": 0.4023382365703583
    },
    {
      "epoch": 0.8226102941176471,
      "grad_norm": 24.125,
      "learning_rate": 2.3830422794117645e-05,
      "loss": 3.0858,
      "mean_token_accuracy": 0.7681948751211166,
      "step": 895,
      "train/kl_loss_qwen": 0.32556538172066213,
      "train/kl_loss_r1": 0.2322130450978875,
      "train/total_kl": 0.5577784314751625
    },
    {
      "epoch": 0.8272058823529411,
      "grad_norm": 8.8125,
      "learning_rate": 2.379595588235294e-05,
      "loss": 2.7973,
      "mean_token_accuracy": 0.758452445268631,
      "step": 900,
      "train/kl_loss_qwen": 0.2798971727490425,
      "train/kl_loss_r1": 0.17839549258351325,
      "train/total_kl": 0.45829266123473644
    },
    {
      "epoch": 0.8318014705882353,
      "grad_norm": 7.03125,
      "learning_rate": 2.3761488970588234e-05,
      "loss": 2.946,
      "mean_token_accuracy": 0.7686445116996765,
      "step": 905,
      "train/kl_loss_qwen": 0.2969953244552016,
      "train/kl_loss_r1": 0.2133618885651231,
      "train/total_kl": 0.5103572174906731
    },
    {
      "epoch": 0.8363970588235294,
      "grad_norm": 5.6875,
      "learning_rate": 2.372702205882353e-05,
      "loss": 2.9284,
      "mean_token_accuracy": 0.763682508468628,
      "step": 910,
      "train/kl_loss_qwen": 0.28750825896859167,
      "train/kl_loss_r1": 0.20824656039476394,
      "train/total_kl": 0.4957548208534718
    },
    {
      "epoch": 0.8409926470588235,
      "grad_norm": 6.0,
      "learning_rate": 2.3692555147058826e-05,
      "loss": 2.4715,
      "mean_token_accuracy": 0.7754847556352615,
      "step": 915,
      "train/kl_loss_qwen": 0.23572093360126017,
      "train/kl_loss_r1": 0.15165257677435875,
      "train/total_kl": 0.3873735036700964
    },
    {
      "epoch": 0.8455882352941176,
      "grad_norm": 37.0,
      "learning_rate": 2.3658088235294118e-05,
      "loss": 3.8152,
      "mean_token_accuracy": 0.7668180167675018,
      "step": 920,
      "train/kl_loss_qwen": 0.43432158417999744,
      "train/kl_loss_r1": 0.29057041853666304,
      "train/total_kl": 0.7248920112848282
    },
    {
      "epoch": 0.8501838235294118,
      "grad_norm": 8.4375,
      "learning_rate": 2.3623621323529414e-05,
      "loss": 2.2838,
      "mean_token_accuracy": 0.756912013888359,
      "step": 925,
      "train/kl_loss_qwen": 0.19460431225597857,
      "train/kl_loss_r1": 0.12254806980490685,
      "train/total_kl": 0.31715237982571126
    },
    {
      "epoch": 0.8547794117647058,
      "grad_norm": 6.09375,
      "learning_rate": 2.3589154411764707e-05,
      "loss": 2.8498,
      "mean_token_accuracy": 0.7745060354471207,
      "step": 930,
      "train/kl_loss_qwen": 0.31135519929230215,
      "train/kl_loss_r1": 0.16893052253872157,
      "train/total_kl": 0.48028572648763657
    },
    {
      "epoch": 0.859375,
      "grad_norm": 9.25,
      "learning_rate": 2.35546875e-05,
      "loss": 2.5034,
      "mean_token_accuracy": 0.7658473283052445,
      "step": 935,
      "train/kl_loss_qwen": 0.23895751722157002,
      "train/kl_loss_r1": 0.16066635940223933,
      "train/total_kl": 0.39962387681007383
    },
    {
      "epoch": 0.8639705882352942,
      "grad_norm": 4.90625,
      "learning_rate": 2.3520220588235295e-05,
      "loss": 2.3413,
      "mean_token_accuracy": 0.7497002571821213,
      "step": 940,
      "train/kl_loss_qwen": 0.22181288599967958,
      "train/kl_loss_r1": 0.13520750738680362,
      "train/total_kl": 0.3570203959941864
    },
    {
      "epoch": 0.8685661764705882,
      "grad_norm": 7.78125,
      "learning_rate": 2.3485753676470588e-05,
      "loss": 3.0273,
      "mean_token_accuracy": 0.7652308195829391,
      "step": 945,
      "train/kl_loss_qwen": 0.31859342865645884,
      "train/kl_loss_r1": 0.2173796972259879,
      "train/total_kl": 0.5359731260687113
    },
    {
      "epoch": 0.8731617647058824,
      "grad_norm": 6.0,
      "learning_rate": 2.3451286764705883e-05,
      "loss": 3.2067,
      "mean_token_accuracy": 0.7529262214899063,
      "step": 950,
      "train/kl_loss_qwen": 0.34089123979210856,
      "train/kl_loss_r1": 0.22591239623725415,
      "train/total_kl": 0.5668036345392465
    },
    {
      "epoch": 0.8777573529411765,
      "grad_norm": 13.0625,
      "learning_rate": 2.3416819852941176e-05,
      "loss": 3.0626,
      "mean_token_accuracy": 0.7502310097217559,
      "step": 955,
      "train/kl_loss_qwen": 0.3198371402919292,
      "train/kl_loss_r1": 0.2195627711713314,
      "train/total_kl": 0.5393999077379703
    },
    {
      "epoch": 0.8823529411764706,
      "grad_norm": 5.09375,
      "learning_rate": 2.3382352941176472e-05,
      "loss": 2.7594,
      "mean_token_accuracy": 0.7421835958957672,
      "step": 960,
      "train/kl_loss_qwen": 0.27218600660562514,
      "train/kl_loss_r1": 0.17624912671744825,
      "train/total_kl": 0.44843512773513794
    },
    {
      "epoch": 0.8869485294117647,
      "grad_norm": 5.3125,
      "learning_rate": 2.3347886029411768e-05,
      "loss": 3.4481,
      "mean_token_accuracy": 0.7635885417461395,
      "step": 965,
      "train/kl_loss_qwen": 0.3790398512035608,
      "train/kl_loss_r1": 0.2533990180119872,
      "train/total_kl": 0.6324388731271029
    },
    {
      "epoch": 0.8915441176470589,
      "grad_norm": 9.5,
      "learning_rate": 2.331341911764706e-05,
      "loss": 2.3847,
      "mean_token_accuracy": 0.7719403833150864,
      "step": 970,
      "train/kl_loss_qwen": 0.22646238952875136,
      "train/kl_loss_r1": 0.1279384709894657,
      "train/total_kl": 0.3544008556753397
    },
    {
      "epoch": 0.8961397058823529,
      "grad_norm": 5.90625,
      "learning_rate": 2.3278952205882353e-05,
      "loss": 2.9426,
      "mean_token_accuracy": 0.7727076560258865,
      "step": 975,
      "train/kl_loss_qwen": 0.304868184402585,
      "train/kl_loss_r1": 0.19793193209916354,
      "train/total_kl": 0.5028001211583615
    },
    {
      "epoch": 0.9007352941176471,
      "grad_norm": 9.5625,
      "learning_rate": 2.3244485294117645e-05,
      "loss": 2.9464,
      "mean_token_accuracy": 0.7676802188158035,
      "step": 980,
      "train/kl_loss_qwen": 0.3127247016876936,
      "train/kl_loss_r1": 0.1998186567798257,
      "train/total_kl": 0.5125433541834354
    },
    {
      "epoch": 0.9053308823529411,
      "grad_norm": 7.0625,
      "learning_rate": 2.321001838235294e-05,
      "loss": 2.9416,
      "mean_token_accuracy": 0.7612376570701599,
      "step": 985,
      "train/kl_loss_qwen": 0.29801272638142107,
      "train/kl_loss_r1": 0.19902198743075133,
      "train/total_kl": 0.49703471437096597
    },
    {
      "epoch": 0.9099264705882353,
      "grad_norm": 7.5625,
      "learning_rate": 2.3175551470588234e-05,
      "loss": 2.413,
      "mean_token_accuracy": 0.7658768028020859,
      "step": 990,
      "train/kl_loss_qwen": 0.2236352551728487,
      "train/kl_loss_r1": 0.13950190246105193,
      "train/total_kl": 0.3631371632218361
    },
    {
      "epoch": 0.9145220588235294,
      "grad_norm": 12.125,
      "learning_rate": 2.314108455882353e-05,
      "loss": 2.669,
      "mean_token_accuracy": 0.7687429904937744,
      "step": 995,
      "train/kl_loss_qwen": 0.27499873116612433,
      "train/kl_loss_r1": 0.16382879298180342,
      "train/total_kl": 0.43882752135396
    },
    {
      "epoch": 0.9191176470588235,
      "grad_norm": 34.0,
      "learning_rate": 2.3106617647058826e-05,
      "loss": 3.0599,
      "mean_token_accuracy": 0.7651446223258972,
      "step": 1000,
      "train/kl_loss_qwen": 0.32082758136093614,
      "train/kl_loss_r1": 0.20667899027466774,
      "train/total_kl": 0.5275065772235393
    },
    {
      "epoch": 0.9237132352941176,
      "grad_norm": 6.53125,
      "learning_rate": 2.3072150735294118e-05,
      "loss": 2.5313,
      "mean_token_accuracy": 0.76262626349926,
      "step": 1005,
      "train/kl_loss_qwen": 0.2558111008256674,
      "train/kl_loss_r1": 0.15045844614505768,
      "train/total_kl": 0.4062695436179638
    },
    {
      "epoch": 0.9283088235294118,
      "grad_norm": 5.90625,
      "learning_rate": 2.3037683823529414e-05,
      "loss": 3.1849,
      "mean_token_accuracy": 0.7305946588516236,
      "step": 1010,
      "train/kl_loss_qwen": 0.3190811090171337,
      "train/kl_loss_r1": 0.20359136033803224,
      "train/total_kl": 0.5226724751293659
    },
    {
      "epoch": 0.9329044117647058,
      "grad_norm": 12.375,
      "learning_rate": 2.3003216911764707e-05,
      "loss": 3.1059,
      "mean_token_accuracy": 0.7705761641263962,
      "step": 1015,
      "train/kl_loss_qwen": 0.3244234949350357,
      "train/kl_loss_r1": 0.23338106833398342,
      "train/total_kl": 0.557804561406374
    },
    {
      "epoch": 0.9375,
      "grad_norm": 6.0625,
      "learning_rate": 2.296875e-05,
      "loss": 3.0695,
      "mean_token_accuracy": 0.7646759301424026,
      "step": 1020,
      "train/kl_loss_qwen": 0.31649671010673047,
      "train/kl_loss_r1": 0.2347225785255432,
      "train/total_kl": 0.5512192968279124
    },
    {
      "epoch": 0.9420955882352942,
      "grad_norm": 8.125,
      "learning_rate": 2.2934283088235295e-05,
      "loss": 2.4066,
      "mean_token_accuracy": 0.7482253730297088,
      "step": 1025,
      "train/kl_loss_qwen": 0.2247611027210951,
      "train/kl_loss_r1": 0.13678079079836608,
      "train/total_kl": 0.3615418955683708
    },
    {
      "epoch": 0.9466911764705882,
      "grad_norm": 10.5,
      "learning_rate": 2.2899816176470588e-05,
      "loss": 2.5618,
      "mean_token_accuracy": 0.7486281037330628,
      "step": 1030,
      "train/kl_loss_qwen": 0.24083134531974792,
      "train/kl_loss_r1": 0.1521694902330637,
      "train/total_kl": 0.39300084486603737
    },
    {
      "epoch": 0.9512867647058824,
      "grad_norm": 6.375,
      "learning_rate": 2.2865349264705884e-05,
      "loss": 2.528,
      "mean_token_accuracy": 0.7573053300380707,
      "step": 1035,
      "train/kl_loss_qwen": 0.2377445988357067,
      "train/kl_loss_r1": 0.15604632087051867,
      "train/total_kl": 0.39379091933369637
    },
    {
      "epoch": 0.9558823529411765,
      "grad_norm": 9.9375,
      "learning_rate": 2.2830882352941176e-05,
      "loss": 2.4803,
      "mean_token_accuracy": 0.7767712444067001,
      "step": 1040,
      "train/kl_loss_qwen": 0.23854669630527497,
      "train/kl_loss_r1": 0.15197529271245003,
      "train/total_kl": 0.39052199125289916
    },
    {
      "epoch": 0.9604779411764706,
      "grad_norm": 4.71875,
      "learning_rate": 2.2796415441176472e-05,
      "loss": 2.5356,
      "mean_token_accuracy": 0.7657318651676178,
      "step": 1045,
      "train/kl_loss_qwen": 0.2492308657616377,
      "train/kl_loss_r1": 0.15562628637999296,
      "train/total_kl": 0.4048571538180113
    },
    {
      "epoch": 0.9650735294117647,
      "grad_norm": 6.125,
      "learning_rate": 2.2761948529411768e-05,
      "loss": 2.7125,
      "mean_token_accuracy": 0.7519971400499343,
      "step": 1050,
      "train/kl_loss_qwen": 0.2670097855851054,
      "train/kl_loss_r1": 0.17238117679953574,
      "train/total_kl": 0.43939096443355086
    },
    {
      "epoch": 0.9696691176470589,
      "grad_norm": 5.75,
      "learning_rate": 2.272748161764706e-05,
      "loss": 2.28,
      "mean_token_accuracy": 0.7458152621984482,
      "step": 1055,
      "train/kl_loss_qwen": 0.21117937862873076,
      "train/kl_loss_r1": 0.12069441135972739,
      "train/total_kl": 0.33187379613518714
    },
    {
      "epoch": 0.9742647058823529,
      "grad_norm": 22.375,
      "learning_rate": 2.2693014705882353e-05,
      "loss": 2.6571,
      "mean_token_accuracy": 0.762286388874054,
      "step": 1060,
      "train/kl_loss_qwen": 0.2774271484464407,
      "train/kl_loss_r1": 0.16375370547175408,
      "train/total_kl": 0.4411808494478464
    },
    {
      "epoch": 0.9788602941176471,
      "grad_norm": 14.6875,
      "learning_rate": 2.2658547794117646e-05,
      "loss": 2.9265,
      "mean_token_accuracy": 0.758323472738266,
      "step": 1065,
      "train/kl_loss_qwen": 0.30878260508179667,
      "train/kl_loss_r1": 0.19694424830377102,
      "train/total_kl": 0.5057268541306257
    },
    {
      "epoch": 0.9834558823529411,
      "grad_norm": 14.5625,
      "learning_rate": 2.262408088235294e-05,
      "loss": 2.5522,
      "mean_token_accuracy": 0.7652994871139527,
      "step": 1070,
      "train/kl_loss_qwen": 0.24689169451594353,
      "train/kl_loss_r1": 0.15555771812796593,
      "train/total_kl": 0.4024494141340256
    },
    {
      "epoch": 0.9880514705882353,
      "grad_norm": 8.4375,
      "learning_rate": 2.2589613970588234e-05,
      "loss": 2.9886,
      "mean_token_accuracy": 0.7621013224124908,
      "step": 1075,
      "train/kl_loss_qwen": 0.30935199521481993,
      "train/kl_loss_r1": 0.21715811733156443,
      "train/total_kl": 0.5265101172029972
    },
    {
      "epoch": 0.9926470588235294,
      "grad_norm": 12.75,
      "learning_rate": 2.255514705882353e-05,
      "loss": 2.7839,
      "mean_token_accuracy": 0.7489064037799835,
      "step": 1080,
      "train/kl_loss_qwen": 0.28402559775859115,
      "train/kl_loss_r1": 0.16853774935007096,
      "train/total_kl": 0.45256335139274595
    },
    {
      "epoch": 0.9972426470588235,
      "grad_norm": 29.625,
      "learning_rate": 2.2520680147058826e-05,
      "loss": 2.9104,
      "mean_token_accuracy": 0.7622999787330628,
      "step": 1085,
      "train/kl_loss_qwen": 0.30061994940042497,
      "train/kl_loss_r1": 0.19877788815647363,
      "train/total_kl": 0.4993978329002857
    }
  ],
  "logging_steps": 5,
  "max_steps": 4352,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6991100809098752e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}