{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.22857142857142856,
  "eval_steps": 500,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "clip_fraction": 0.0,
      "completion_length": 2571.2083587646484,
      "epoch": 0.001142857142857143,
      "grad_norm": 0.25632715225219727,
      "kl": 0.0,
      "learning_rate": 0.0,
      "loss": 0.1082,
      "reward": 0.21363236638717353,
      "reward_std": 0.5417899899184704,
      "rewards/accuracy_reward": 0.22916667349636555,
      "rewards/cosine_scaled_reward": -0.015534311532974243,
      "step": 1
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2804.395881652832,
      "epoch": 0.002285714285714286,
      "grad_norm": 0.2158345729112625,
      "kl": 0.0,
      "learning_rate": 5e-08,
      "loss": 0.0476,
      "reward": 0.179365461692214,
      "reward_std": 0.24326776154339314,
      "rewards/accuracy_reward": 0.2291666716337204,
      "rewards/cosine_scaled_reward": -0.04980122856795788,
      "step": 2
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3340.9791717529297,
      "epoch": 0.0034285714285714284,
      "grad_norm": 0.1626892238855362,
      "kl": 4.589557647705078e-05,
      "learning_rate": 1e-07,
      "loss": -0.0326,
      "reward": -0.18680910766124725,
      "reward_std": 0.2629435881972313,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.2076424416154623,
      "step": 3
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2383.8125228881836,
      "epoch": 0.004571428571428572,
      "grad_norm": 0.321321964263916,
      "kl": 3.878772258758545e-05,
      "learning_rate": 1.5e-07,
      "loss": 0.0713,
      "reward": 0.037019677460193634,
      "reward_std": 0.5308115221560001,
      "rewards/accuracy_reward": 0.14583333395421505,
      "rewards/cosine_scaled_reward": -0.10881366301327944,
      "step": 4
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3285.6458740234375,
      "epoch": 0.005714285714285714,
      "grad_norm": 0.16392037272453308,
      "kl": 3.78340482711792e-05,
      "learning_rate": 2e-07,
      "loss": 0.0192,
      "reward": -0.10475460533052683,
      "reward_std": 0.4349679294973612,
      "rewards/accuracy_reward": 0.0833333358168602,
      "rewards/cosine_scaled_reward": -0.18808794301003218,
      "step": 5
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2898.625045776367,
      "epoch": 0.006857142857142857,
      "grad_norm": 0.19641205668449402,
      "kl": 4.819035530090332e-05,
      "learning_rate": 2.5e-07,
      "loss": 0.0638,
      "reward": -0.0739727895706892,
      "reward_std": 0.40896974317729473,
      "rewards/accuracy_reward": 0.0833333358168602,
      "rewards/cosine_scaled_reward": -0.15730612399056554,
      "step": 6
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3123.3333740234375,
      "epoch": 0.008,
      "grad_norm": 0.15763328969478607,
      "kl": 2.3871660232543945e-05,
      "learning_rate": 3e-07,
      "loss": 0.0023,
      "reward": 0.2681586788967252,
      "reward_std": 0.4782935921102762,
      "rewards/accuracy_reward": 0.27083333767950535,
      "rewards/cosine_scaled_reward": -0.0026746466755867004,
      "step": 7
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2718.1458435058594,
      "epoch": 0.009142857142857144,
      "grad_norm": 0.18417485058307648,
      "kl": 2.504885196685791e-05,
      "learning_rate": 3.5e-07,
      "loss": 0.0037,
      "reward": 0.43941989610902965,
      "reward_std": 0.6203235052525997,
      "rewards/accuracy_reward": 0.35416667349636555,
      "rewards/cosine_scaled_reward": 0.08525321818888187,
      "step": 8
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3126.979232788086,
      "epoch": 0.010285714285714285,
      "grad_norm": 0.1806861162185669,
      "kl": 4.266202449798584e-05,
      "learning_rate": 4e-07,
      "loss": 0.0192,
      "reward": 0.06813614349812269,
      "reward_std": 0.5027568060904741,
      "rewards/accuracy_reward": 0.1666666716337204,
      "rewards/cosine_scaled_reward": -0.09853052510879934,
      "step": 9
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2906.416717529297,
      "epoch": 0.011428571428571429,
      "grad_norm": 0.18246354162693024,
      "kl": 2.8778158593922853e-05,
      "learning_rate": 4.5e-07,
      "loss": 0.0043,
      "reward": -0.07218722999095917,
      "reward_std": 0.47773855179548264,
      "rewards/accuracy_reward": 0.10416666977107525,
      "rewards/cosine_scaled_reward": -0.17635390162467957,
      "step": 10
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3303.500030517578,
      "epoch": 0.012571428571428572,
      "grad_norm": 0.15648911893367767,
      "kl": 3.5509467124938965e-05,
      "learning_rate": 5e-07,
      "loss": -0.0166,
      "reward": -0.156071281991899,
      "reward_std": 0.2669069590047002,
      "rewards/accuracy_reward": 0.0416666679084301,
      "rewards/cosine_scaled_reward": -0.1977379499003291,
      "step": 11
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2357.5208740234375,
      "epoch": 0.013714285714285714,
      "grad_norm": 0.2004307210445404,
      "kl": 3.539770841598511e-05,
      "learning_rate": 5.5e-07,
      "loss": 0.0214,
      "reward": -0.02095029642805457,
      "reward_std": 0.5913799963891506,
      "rewards/accuracy_reward": 0.1458333358168602,
      "rewards/cosine_scaled_reward": -0.1667836436536163,
      "step": 12
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2870.541702270508,
      "epoch": 0.014857142857142857,
      "grad_norm": 0.18646906316280365,
      "kl": 3.197789192199707e-05,
      "learning_rate": 6e-07,
      "loss": -0.0017,
      "reward": 0.14097319915890694,
      "reward_std": 0.35468385741114616,
      "rewards/accuracy_reward": 0.1875000074505806,
      "rewards/cosine_scaled_reward": -0.04652681015431881,
      "step": 13
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2858.375015258789,
      "epoch": 0.016,
      "grad_norm": 0.2710908055305481,
      "kl": 2.9653310775756836e-05,
      "learning_rate": 6.5e-07,
      "loss": 0.063,
      "reward": 0.18795185792259872,
      "reward_std": 0.5712584815919399,
      "rewards/accuracy_reward": 0.20833334140479565,
      "rewards/cosine_scaled_reward": -0.0203814753331244,
      "step": 14
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2703.979202270508,
      "epoch": 0.017142857142857144,
      "grad_norm": 0.17628003656864166,
      "kl": 2.5141984224319458e-05,
      "learning_rate": 7e-07,
      "loss": -0.0072,
      "reward": 0.27426825650036335,
      "reward_std": 0.38823096454143524,
      "rewards/accuracy_reward": 0.2500000037252903,
      "rewards/cosine_scaled_reward": 0.02426825277507305,
      "step": 15
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3517.666717529297,
      "epoch": 0.018285714285714287,
      "grad_norm": 0.15789569914340973,
      "kl": 3.6329030990600586e-05,
      "learning_rate": 7.5e-07,
      "loss": -0.0084,
      "reward": -0.18198822857812047,
      "reward_std": 0.3138104174286127,
      "rewards/accuracy_reward": 0.0416666679084301,
      "rewards/cosine_scaled_reward": -0.22365489602088928,
      "step": 16
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2283.0417137145996,
      "epoch": 0.019428571428571427,
      "grad_norm": 0.2678299844264984,
      "kl": 3.6092475056648254e-05,
      "learning_rate": 8e-07,
      "loss": 0.0346,
      "reward": 0.19116858765482903,
      "reward_std": 0.6768512222915888,
      "rewards/accuracy_reward": 0.2291666716337204,
      "rewards/cosine_scaled_reward": -0.03799809608608484,
      "step": 17
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2891.6250534057617,
      "epoch": 0.02057142857142857,
      "grad_norm": 0.1524839550256729,
      "kl": 9.838491678237915e-06,
      "learning_rate": 8.499999999999999e-07,
      "loss": 0.0422,
      "reward": 0.019560502842068672,
      "reward_std": 0.42486423440277576,
      "rewards/accuracy_reward": 0.1250000037252903,
      "rewards/cosine_scaled_reward": -0.10543949622660875,
      "step": 18
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2730.6875381469727,
      "epoch": 0.021714285714285714,
      "grad_norm": 0.1984628587961197,
      "kl": 2.9616057872772217e-05,
      "learning_rate": 9e-07,
      "loss": 0.0244,
      "reward": 0.5108187086880207,
      "reward_std": 0.5117705333977938,
      "rewards/accuracy_reward": 0.3750000111758709,
      "rewards/cosine_scaled_reward": 0.1358186900615692,
      "step": 19
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2329.9583892822266,
      "epoch": 0.022857142857142857,
      "grad_norm": 0.29129523038864136,
      "kl": 1.1567026376724243e-05,
      "learning_rate": 9.499999999999999e-07,
      "loss": 0.1144,
      "reward": 0.3239101804792881,
      "reward_std": 0.5062661934643984,
      "rewards/accuracy_reward": 0.29166666977107525,
      "rewards/cosine_scaled_reward": 0.03224351815879345,
      "step": 20
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2660.2083587646484,
      "epoch": 0.024,
      "grad_norm": 0.19655585289001465,
      "kl": 2.6446767151355743e-05,
      "learning_rate": 1e-06,
      "loss": 0.062,
      "reward": 0.05283654620870948,
      "reward_std": 0.39788419753313065,
      "rewards/accuracy_reward": 0.1458333395421505,
      "rewards/cosine_scaled_reward": -0.09299679077230394,
      "step": 21
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 1825.020896911621,
      "epoch": 0.025142857142857144,
      "grad_norm": 0.2982518970966339,
      "kl": 2.1491199731826782e-05,
      "learning_rate": 9.99931462820376e-07,
      "loss": 0.0703,
      "reward": 0.23684986494481564,
      "reward_std": 0.4532635463401675,
      "rewards/accuracy_reward": 0.27083333767950535,
      "rewards/cosine_scaled_reward": -0.033983476692810655,
      "step": 22
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2637.2708892822266,
      "epoch": 0.026285714285714287,
      "grad_norm": 0.17908190190792084,
      "kl": 2.055056393146515e-05,
      "learning_rate": 9.997258721585931e-07,
      "loss": 0.0216,
      "reward": 0.04677679901942611,
      "reward_std": 0.5608484419062734,
      "rewards/accuracy_reward": 0.1666666679084301,
      "rewards/cosine_scaled_reward": -0.1198898684233427,
      "step": 23
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2836.437530517578,
      "epoch": 0.027428571428571427,
      "grad_norm": 0.18375183641910553,
      "kl": 2.3286789655685425e-05,
      "learning_rate": 9.993832906395582e-07,
      "loss": 0.0363,
      "reward": 0.1852389182895422,
      "reward_std": 0.5924705881625414,
      "rewards/accuracy_reward": 0.22916667349636555,
      "rewards/cosine_scaled_reward": -0.04392773995641619,
      "step": 24
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2892.875045776367,
      "epoch": 0.02857142857142857,
      "grad_norm": 0.19979171454906464,
      "kl": 3.429921343922615e-05,
      "learning_rate": 9.989038226169207e-07,
      "loss": -0.0319,
      "reward": 0.18543785763904452,
      "reward_std": 0.5226433798670769,
      "rewards/accuracy_reward": 0.2083333395421505,
      "rewards/cosine_scaled_reward": -0.022895504254847765,
      "step": 25
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2954.4791870117188,
      "epoch": 0.029714285714285714,
      "grad_norm": 0.1611883044242859,
      "kl": 2.4512410163879395e-05,
      "learning_rate": 9.982876141412855e-07,
      "loss": -0.0069,
      "reward": 0.2805785769596696,
      "reward_std": 0.33157524187117815,
      "rewards/accuracy_reward": 0.25000000558793545,
      "rewards/cosine_scaled_reward": 0.030578549951314926,
      "step": 26
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3068.104232788086,
      "epoch": 0.030857142857142857,
      "grad_norm": 0.1950128972530365,
      "kl": 8.122995495796204e-06,
      "learning_rate": 9.975348529157229e-07,
      "loss": 0.0453,
      "reward": 0.07182890921831131,
      "reward_std": 0.5058779399842024,
      "rewards/accuracy_reward": 0.1666666679084301,
      "rewards/cosine_scaled_reward": -0.09483776055276394,
      "step": 27
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2946.7083587646484,
      "epoch": 0.032,
      "grad_norm": 0.17087508738040924,
      "kl": 1.4003482647240162e-05,
      "learning_rate": 9.96645768238595e-07,
      "loss": -0.041,
      "reward": 0.18860503658652306,
      "reward_std": 0.5205447860062122,
      "rewards/accuracy_reward": 0.2083333395421505,
      "rewards/cosine_scaled_reward": -0.01972830155864358,
      "step": 28
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3220.2708740234375,
      "epoch": 0.03314285714285714,
      "grad_norm": 0.24361945688724518,
      "kl": 3.1463801860809326e-05,
      "learning_rate": 9.956206309337066e-07,
      "loss": -0.1063,
      "reward": -0.17443929053843021,
      "reward_std": 0.3365236781537533,
      "rewards/accuracy_reward": 0.0416666679084301,
      "rewards/cosine_scaled_reward": -0.2161059557693079,
      "step": 29
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2893.104248046875,
      "epoch": 0.03428571428571429,
      "grad_norm": 0.2155693769454956,
      "kl": 5.615316331386566e-05,
      "learning_rate": 9.944597532678119e-07,
      "loss": 0.0487,
      "reward": 0.27425617165863514,
      "reward_std": 0.6831754259765148,
      "rewards/accuracy_reward": 0.25000000558793545,
      "rewards/cosine_scaled_reward": 0.024256166070699692,
      "step": 30
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3071.4375,
      "epoch": 0.03542857142857143,
      "grad_norm": 0.17095845937728882,
      "kl": 4.3095555156469345e-05,
      "learning_rate": 9.931634888554935e-07,
      "loss": -0.0257,
      "reward": -0.04954687878489494,
      "reward_std": 0.3689954075962305,
      "rewards/accuracy_reward": 0.10416666977107525,
      "rewards/cosine_scaled_reward": -0.15371354296803474,
      "step": 31
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3270.375030517578,
      "epoch": 0.036571428571428574,
      "grad_norm": 0.1867983490228653,
      "kl": 4.005682421848178e-05,
      "learning_rate": 9.917322325514487e-07,
      "loss": -0.0114,
      "reward": 0.33223521208856255,
      "reward_std": 0.5476495027542114,
      "rewards/accuracy_reward": 0.2916666716337204,
      "rewards/cosine_scaled_reward": 0.04056852776557207,
      "step": 32
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3300.2708740234375,
      "epoch": 0.037714285714285714,
      "grad_norm": 0.15105699002742767,
      "kl": 0.00013676844537258148,
      "learning_rate": 9.901664203302124e-07,
      "loss": 0.0197,
      "reward": 0.19714780524373055,
      "reward_std": 0.5669918619096279,
      "rewards/accuracy_reward": 0.22916667349636555,
      "rewards/cosine_scaled_reward": -0.032018861966207623,
      "step": 33
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2588.1875610351562,
      "epoch": 0.038857142857142854,
      "grad_norm": 0.47045713663101196,
      "kl": 0.00015014410018920898,
      "learning_rate": 9.88466529153356e-07,
      "loss": 0.1528,
      "reward": 0.3600555928424001,
      "reward_std": 0.6696161534637213,
      "rewards/accuracy_reward": 0.31250000931322575,
      "rewards/cosine_scaled_reward": 0.0475555956363678,
      "step": 34
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2979.000030517578,
      "epoch": 0.04,
      "grad_norm": 0.21019528806209564,
      "kl": 0.00010133162140846252,
      "learning_rate": 9.866330768241983e-07,
      "loss": 0.0436,
      "reward": 0.08721516001969576,
      "reward_std": 0.6073270477354527,
      "rewards/accuracy_reward": 0.16666666977107525,
      "rewards/cosine_scaled_reward": -0.07945150946034119,
      "step": 35
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3349.1041870117188,
      "epoch": 0.04114285714285714,
      "grad_norm": 0.1755942404270172,
      "kl": 0.00011576339602470398,
      "learning_rate": 9.846666218300807e-07,
      "loss": 0.0059,
      "reward": -0.17237537261098623,
      "reward_std": 0.22124690702185035,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.19320870749652386,
      "step": 36
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3302.3333435058594,
      "epoch": 0.04228571428571429,
      "grad_norm": 0.172617107629776,
      "kl": 7.249228656291962e-05,
      "learning_rate": 9.825677631722435e-07,
      "loss": -0.0307,
      "reward": -0.2602706290781498,
      "reward_std": 0.16562421713024378,
      "rewards/accuracy_reward": 0.0,
      "rewards/cosine_scaled_reward": -0.26027063466608524,
      "step": 37
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3232.187515258789,
      "epoch": 0.04342857142857143,
      "grad_norm": 0.16284486651420593,
      "kl": 8.244812488555908e-05,
      "learning_rate": 9.80337140183366e-07,
      "loss": -0.0182,
      "reward": -0.027252301573753357,
      "reward_std": 0.2725130654871464,
      "rewards/accuracy_reward": 0.1041666716337204,
      "rewards/cosine_scaled_reward": -0.13141895458102226,
      "step": 38
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2854.1042098999023,
      "epoch": 0.044571428571428574,
      "grad_norm": 0.17649812996387482,
      "kl": 9.210407733917236e-05,
      "learning_rate": 9.779754323328192e-07,
      "loss": 0.0249,
      "reward": 0.1576530784368515,
      "reward_std": 0.24235367169603705,
      "rewards/accuracy_reward": 0.2083333358168602,
      "rewards/cosine_scaled_reward": -0.05068026855587959,
      "step": 39
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2369.416702270508,
      "epoch": 0.045714285714285714,
      "grad_norm": 0.18827204406261444,
      "kl": 0.00034741777926683426,
      "learning_rate": 9.754833590196926e-07,
      "loss": -0.0217,
      "reward": 0.06249742116779089,
      "reward_std": 0.2527109682559967,
      "rewards/accuracy_reward": 0.14583333395421505,
      "rewards/cosine_scaled_reward": -0.0833359295502305,
      "step": 40
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2972.229217529297,
      "epoch": 0.046857142857142854,
      "grad_norm": 0.17488084733486176,
      "kl": 0.0001214444637298584,
      "learning_rate": 9.728616793536587e-07,
      "loss": -0.0016,
      "reward": 0.055548351956531405,
      "reward_std": 0.5645285593345761,
      "rewards/accuracy_reward": 0.16666666977107525,
      "rewards/cosine_scaled_reward": -0.11111831851303577,
      "step": 41
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2735.8958473205566,
      "epoch": 0.048,
      "grad_norm": 0.28289923071861267,
      "kl": 8.103251457214355e-05,
      "learning_rate": 9.701111919237408e-07,
      "loss": -0.0087,
      "reward": -0.2322952365502715,
      "reward_std": 0.2050471631810069,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.25312857422977686,
      "step": 42
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3113.8958587646484,
      "epoch": 0.04914285714285714,
      "grad_norm": 0.20076647400856018,
      "kl": 8.933991193771362e-05,
      "learning_rate": 9.672327345550543e-07,
      "loss": 0.0291,
      "reward": -0.03336653020232916,
      "reward_std": 0.36820667795836926,
      "rewards/accuracy_reward": 0.10416666977107525,
      "rewards/cosine_scaled_reward": -0.1375331999734044,
      "step": 43
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2842.6250228881836,
      "epoch": 0.05028571428571429,
      "grad_norm": 0.32278913259506226,
      "kl": 0.0004453342407941818,
      "learning_rate": 9.64227184053598e-07,
      "loss": 0.0449,
      "reward": 0.2544646831229329,
      "reward_std": 0.47283124178647995,
      "rewards/accuracy_reward": 0.2291666753590107,
      "rewards/cosine_scaled_reward": 0.025297993794083595,
      "step": 44
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3480.7708740234375,
      "epoch": 0.05142857142857143,
      "grad_norm": 0.1464184671640396,
      "kl": 0.00017410144209861755,
      "learning_rate": 9.610954559391704e-07,
      "loss": 0.0163,
      "reward": 0.20797571539878845,
      "reward_std": 0.4654997680336237,
      "rewards/accuracy_reward": 0.2083333395421505,
      "rewards/cosine_scaled_reward": -0.0003576315939426422,
      "step": 45
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3276.7916717529297,
      "epoch": 0.052571428571428575,
      "grad_norm": 0.20637758076190948,
      "kl": 0.0003883976023644209,
      "learning_rate": 9.578385041664925e-07,
      "loss": -0.0218,
      "reward": -0.2070446526631713,
      "reward_std": 0.15773496124893427,
      "rewards/accuracy_reward": 0.0,
      "rewards/cosine_scaled_reward": -0.20704465080052614,
      "step": 46
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2968.0833740234375,
      "epoch": 0.053714285714285714,
      "grad_norm": 0.20068301260471344,
      "kl": 0.0002564918249845505,
      "learning_rate": 9.54457320834625e-07,
      "loss": 0.1738,
      "reward": 0.22800862789154053,
      "reward_std": 0.6421588193625212,
      "rewards/accuracy_reward": 0.2291666716337204,
      "rewards/cosine_scaled_reward": -0.0011580577120184898,
      "step": 47
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2857.354232788086,
      "epoch": 0.054857142857142854,
      "grad_norm": 0.20215612649917603,
      "kl": 0.0009041652083396912,
      "learning_rate": 9.509529358847654e-07,
      "loss": -0.0166,
      "reward": 0.11733034905046225,
      "reward_std": 0.36523033399134874,
      "rewards/accuracy_reward": 0.1666666679084301,
      "rewards/cosine_scaled_reward": -0.04933632561005652,
      "step": 48
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2516.229217529297,
      "epoch": 0.056,
      "grad_norm": 0.19533751904964447,
      "kl": 0.00025360286235809326,
      "learning_rate": 9.473264167865171e-07,
      "loss": -0.0101,
      "reward": 0.2620775690302253,
      "reward_std": 0.6330471076071262,
      "rewards/accuracy_reward": 0.2500000074505806,
      "rewards/cosine_scaled_reward": 0.012077568098902702,
      "step": 49
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2981.4375,
      "epoch": 0.05714285714285714,
      "grad_norm": 0.165849506855011,
      "kl": 0.00045928824692964554,
      "learning_rate": 9.43578868212728e-07,
      "loss": -0.0219,
      "reward": 0.291345676407218,
      "reward_std": 0.4359543416649103,
      "rewards/accuracy_reward": 0.29166667349636555,
      "rewards/cosine_scaled_reward": -0.0003209911519661546,
      "step": 50
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2388.0625381469727,
      "epoch": 0.05828571428571429,
      "grad_norm": 0.22810906171798706,
      "kl": 0.0011309236288070679,
      "learning_rate": 9.397114317029974e-07,
      "loss": 0.0198,
      "reward": 0.11073957197368145,
      "reward_std": 0.4677628371864557,
      "rewards/accuracy_reward": 0.1666666716337204,
      "rewards/cosine_scaled_reward": -0.055927105247974396,
      "step": 51
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2879.5417289733887,
      "epoch": 0.05942857142857143,
      "grad_norm": 0.22546225786209106,
      "kl": 0.0010811500251293182,
      "learning_rate": 9.357252853159505e-07,
      "loss": 0.0439,
      "reward": 0.2536798194050789,
      "reward_std": 0.6959168985486031,
      "rewards/accuracy_reward": 0.2708333395421505,
      "rewards/cosine_scaled_reward": -0.01715349592268467,
      "step": 52
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2816.3125610351562,
      "epoch": 0.060571428571428575,
      "grad_norm": 0.22971203923225403,
      "kl": 0.0007025115191936493,
      "learning_rate": 9.316216432703916e-07,
      "loss": 0.0842,
      "reward": 0.30444883555173874,
      "reward_std": 0.6236651353538036,
      "rewards/accuracy_reward": 0.2916666753590107,
      "rewards/cosine_scaled_reward": 0.01278215617639944,
      "step": 53
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2990.541732788086,
      "epoch": 0.061714285714285715,
      "grad_norm": 0.17176969349384308,
      "kl": 0.00022144615650177002,
      "learning_rate": 9.274017555754407e-07,
      "loss": 0.0564,
      "reward": 0.36173775792121887,
      "reward_std": 0.7157497424632311,
      "rewards/accuracy_reward": 0.31250000931322575,
      "rewards/cosine_scaled_reward": 0.04923774115741253,
      "step": 54
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3098.687530517578,
      "epoch": 0.06285714285714286,
      "grad_norm": 0.1664225459098816,
      "kl": 0.0008269157260656357,
      "learning_rate": 9.230669076497687e-07,
      "loss": 0.055,
      "reward": 0.33386562392115593,
      "reward_std": 0.4932790081948042,
      "rewards/accuracy_reward": 0.27083333395421505,
      "rewards/cosine_scaled_reward": 0.06303227692842484,
      "step": 55
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2980.1667098999023,
      "epoch": 0.064,
      "grad_norm": 0.22248487174510956,
      "kl": 0.0004000365734100342,
      "learning_rate": 9.186184199300463e-07,
      "loss": 0.0463,
      "reward": 0.1612101886421442,
      "reward_std": 0.3384838867932558,
      "rewards/accuracy_reward": 0.2083333432674408,
      "rewards/cosine_scaled_reward": -0.047123152762651443,
      "step": 56
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3345.5833740234375,
      "epoch": 0.06514285714285714,
      "grad_norm": 0.14005281031131744,
      "kl": 0.00021963752806186676,
      "learning_rate": 9.140576474687263e-07,
      "loss": 0.02,
      "reward": -0.10464224513270892,
      "reward_std": 0.44751227274537086,
      "rewards/accuracy_reward": 0.0833333358168602,
      "rewards/cosine_scaled_reward": -0.1879755798727274,
      "step": 57
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2249.7917098999023,
      "epoch": 0.06628571428571428,
      "grad_norm": 0.21027207374572754,
      "kl": 0.00269511342048645,
      "learning_rate": 9.093859795212817e-07,
      "loss": 0.0657,
      "reward": 0.3534491113387048,
      "reward_std": 0.5972686447203159,
      "rewards/accuracy_reward": 0.31250000186264515,
      "rewards/cosine_scaled_reward": 0.04094908758997917,
      "step": 58
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2903.854179382324,
      "epoch": 0.06742857142857143,
      "grad_norm": 0.205511212348938,
      "kl": 0.0010725334286689758,
      "learning_rate": 9.046048391230247e-07,
      "loss": -0.0834,
      "reward": -0.007946477271616459,
      "reward_std": 0.4581656437367201,
      "rewards/accuracy_reward": 0.12500000186264515,
      "rewards/cosine_scaled_reward": -0.13294648192822933,
      "step": 59
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3034.270866394043,
      "epoch": 0.06857142857142857,
      "grad_norm": 0.16402272880077362,
      "kl": 0.00045564770698547363,
      "learning_rate": 8.997156826556369e-07,
      "loss": 0.0219,
      "reward": -0.010321232955902815,
      "reward_std": 0.44436920061707497,
      "rewards/accuracy_reward": 0.12500000186264515,
      "rewards/cosine_scaled_reward": -0.13532123528420925,
      "step": 60
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3173.125030517578,
      "epoch": 0.06971428571428571,
      "grad_norm": 0.15932059288024902,
      "kl": 0.0008110031485557556,
      "learning_rate": 8.9471999940354e-07,
      "loss": 0.0541,
      "reward": -0.0845769364386797,
      "reward_std": 0.42896613012999296,
      "rewards/accuracy_reward": 0.12500000186264515,
      "rewards/cosine_scaled_reward": -0.20957693923264742,
      "step": 61
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2611.666717529297,
      "epoch": 0.07085714285714285,
      "grad_norm": 0.2158014327287674,
      "kl": 0.07185456156730652,
      "learning_rate": 8.896193111002475e-07,
      "loss": 0.0293,
      "reward": 0.34560922905802727,
      "reward_std": 0.7172879008576274,
      "rewards/accuracy_reward": 0.31250000558793545,
      "rewards/cosine_scaled_reward": 0.033109224401414394,
      "step": 62
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2483.729232788086,
      "epoch": 0.072,
      "grad_norm": 0.17524653673171997,
      "kl": 0.003895312547683716,
      "learning_rate": 8.844151714648274e-07,
      "loss": 0.0736,
      "reward": 0.3518992383033037,
      "reward_std": 0.4679460935294628,
      "rewards/accuracy_reward": 0.29166666977107525,
      "rewards/cosine_scaled_reward": 0.060232581570744514,
      "step": 63
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3119.6666870117188,
      "epoch": 0.07314285714285715,
      "grad_norm": 0.17840062081813812,
      "kl": 0.001272439956665039,
      "learning_rate": 8.791091657286267e-07,
      "loss": 0.0552,
      "reward": 0.021638838574290276,
      "reward_std": 0.5125578781589866,
      "rewards/accuracy_reward": 0.1458333358168602,
      "rewards/cosine_scaled_reward": -0.12419449165463448,
      "step": 64
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2688.958366394043,
      "epoch": 0.07428571428571429,
      "grad_norm": 0.1985544115304947,
      "kl": 0.001712799072265625,
      "learning_rate": 8.737029101523929e-07,
      "loss": -0.0062,
      "reward": 0.08414293639361858,
      "reward_std": 0.45870870165526867,
      "rewards/accuracy_reward": 0.1875000074505806,
      "rewards/cosine_scaled_reward": -0.10335706360638142,
      "step": 65
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2155.270839691162,
      "epoch": 0.07542857142857143,
      "grad_norm": 0.24202020466327667,
      "kl": 0.0014235973358154297,
      "learning_rate": 8.681980515339463e-07,
      "loss": 0.0342,
      "reward": 0.2718210807070136,
      "reward_std": 0.3149276631884277,
      "rewards/accuracy_reward": 0.2916666716337204,
      "rewards/cosine_scaled_reward": -0.019845602102577686,
      "step": 66
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3514.5,
      "epoch": 0.07657142857142857,
      "grad_norm": 0.13506484031677246,
      "kl": 0.0011570192873477936,
      "learning_rate": 8.625962667065487e-07,
      "loss": 0.0197,
      "reward": -0.17880290560424328,
      "reward_std": 0.3006857465952635,
      "rewards/accuracy_reward": 0.0625,
      "rewards/cosine_scaled_reward": -0.24130290746688843,
      "step": 67
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2224.1250228881836,
      "epoch": 0.07771428571428571,
      "grad_norm": 0.26422637701034546,
      "kl": 0.008173942565917969,
      "learning_rate": 8.568992620281243e-07,
      "loss": 0.1289,
      "reward": 0.26997855119407177,
      "reward_std": 0.5711956042796373,
      "rewards/accuracy_reward": 0.2500000074505806,
      "rewards/cosine_scaled_reward": 0.019978564232587814,
      "step": 68
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2771.104202270508,
      "epoch": 0.07885714285714286,
      "grad_norm": 0.22578439116477966,
      "kl": 0.004560351371765137,
      "learning_rate": 8.511087728614862e-07,
      "loss": -0.0186,
      "reward": -0.20487310830503702,
      "reward_std": 0.2483804766088724,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.22570644319057465,
      "step": 69
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3143.8333587646484,
      "epoch": 0.08,
      "grad_norm": 0.2783580422401428,
      "kl": 0.0023155808448791504,
      "learning_rate": 8.452265630457282e-07,
      "loss": 0.0232,
      "reward": -0.22550072893500328,
      "reward_std": 0.24843326956033707,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.24633406475186348,
      "step": 70
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2701.6666870117188,
      "epoch": 0.08114285714285714,
      "grad_norm": 0.24257808923721313,
      "kl": 0.0030676722526550293,
      "learning_rate": 8.392544243589427e-07,
      "loss": 0.0883,
      "reward": 0.10035606473684311,
      "reward_std": 0.5035490561276674,
      "rewards/accuracy_reward": 0.1666666679084301,
      "rewards/cosine_scaled_reward": -0.0663105882704258,
      "step": 71
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3351.541748046875,
      "epoch": 0.08228571428571428,
      "grad_norm": 0.1588461697101593,
      "kl": 0.00363922119140625,
      "learning_rate": 8.331941759724268e-07,
      "loss": 0.0452,
      "reward": -0.19743284268770367,
      "reward_std": 0.27725750021636486,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.2182661797851324,
      "step": 72
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3507.666717529297,
      "epoch": 0.08342857142857144,
      "grad_norm": 0.1444426029920578,
      "kl": 0.0006330609321594238,
      "learning_rate": 8.270476638965461e-07,
      "loss": 0.0331,
      "reward": -0.05398387461900711,
      "reward_std": 0.49269056133925915,
      "rewards/accuracy_reward": 0.10416666977107525,
      "rewards/cosine_scaled_reward": -0.15815055184066296,
      "step": 73
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3344.041717529297,
      "epoch": 0.08457142857142858,
      "grad_norm": 0.15516585111618042,
      "kl": 0.0021051764488220215,
      "learning_rate": 8.208167604184217e-07,
      "loss": 0.0095,
      "reward": 0.008867893368005753,
      "reward_std": 0.26112012285739183,
      "rewards/accuracy_reward": 0.14583333395421505,
      "rewards/cosine_scaled_reward": -0.1369654512964189,
      "step": 74
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2982.4583435058594,
      "epoch": 0.08571428571428572,
      "grad_norm": 0.13954143226146698,
      "kl": 0.011302351951599121,
      "learning_rate": 8.145033635316128e-07,
      "loss": 0.0333,
      "reward": 0.28953575156629086,
      "reward_std": 0.3247816953808069,
      "rewards/accuracy_reward": 0.2708333358168602,
      "rewards/cosine_scaled_reward": 0.018702420871704817,
      "step": 75
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3167.2083587646484,
      "epoch": 0.08685714285714285,
      "grad_norm": 0.1654420793056488,
      "kl": 0.0009011179208755493,
      "learning_rate": 8.081093963579707e-07,
      "loss": 0.014,
      "reward": -0.2234128573909402,
      "reward_std": 0.27190146408975124,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.24424619507044554,
      "step": 76
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3288.125030517578,
      "epoch": 0.088,
      "grad_norm": 0.15181006491184235,
      "kl": 0.0010012537240982056,
      "learning_rate": 8.01636806561836e-07,
      "loss": -0.013,
      "reward": 0.1649619173258543,
      "reward_std": 0.2921932004392147,
      "rewards/accuracy_reward": 0.1875,
      "rewards/cosine_scaled_reward": -0.022538071498274803,
      "step": 77
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3191.187530517578,
      "epoch": 0.08914285714285715,
      "grad_norm": 0.1391523778438568,
      "kl": 0.0008662641048431396,
      "learning_rate": 7.950875657567621e-07,
      "loss": 0.0498,
      "reward": 0.156300887465477,
      "reward_std": 0.28907839488238096,
      "rewards/accuracy_reward": 0.1875,
      "rewards/cosine_scaled_reward": -0.031199105083942413,
      "step": 78
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2525.6459159851074,
      "epoch": 0.09028571428571429,
      "grad_norm": 0.22027799487113953,
      "kl": 0.0032279491424560547,
      "learning_rate": 7.884636689049422e-07,
      "loss": -0.0856,
      "reward": 0.27039410918951035,
      "reward_std": 0.4748187121003866,
      "rewards/accuracy_reward": 0.29166666977107525,
      "rewards/cosine_scaled_reward": -0.021272567100822926,
      "step": 79
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3310.000030517578,
      "epoch": 0.09142857142857143,
      "grad_norm": 0.15355749428272247,
      "kl": 0.0017953217029571533,
      "learning_rate": 7.817671337095244e-07,
      "loss": 0.0254,
      "reward": 0.15592915751039982,
      "reward_std": 0.45574192702770233,
      "rewards/accuracy_reward": 0.20833334140479565,
      "rewards/cosine_scaled_reward": -0.05240418389439583,
      "step": 80
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3161.375030517578,
      "epoch": 0.09257142857142857,
      "grad_norm": 0.20919400453567505,
      "kl": 0.004012107849121094,
      "learning_rate": 7.75e-07,
      "loss": 0.0254,
      "reward": -0.12468860670924187,
      "reward_std": 0.35988178849220276,
      "rewards/accuracy_reward": 0.06250000186264515,
      "rewards/cosine_scaled_reward": -0.1871886090375483,
      "step": 81
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2771.8958435058594,
      "epoch": 0.09371428571428571,
      "grad_norm": 0.184526726603508,
      "kl": 0.0013185292482376099,
      "learning_rate": 7.681643291108517e-07,
      "loss": 0.0203,
      "reward": 0.25131775764748454,
      "reward_std": 0.3994978480041027,
      "rewards/accuracy_reward": 0.2500000037252903,
      "rewards/cosine_scaled_reward": 0.0013177692890167236,
      "step": 82
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2935.8958435058594,
      "epoch": 0.09485714285714286,
      "grad_norm": 0.21173416078090668,
      "kl": 0.0016698837280273438,
      "learning_rate": 7.612622032536507e-07,
      "loss": 0.0217,
      "reward": -0.1123672490939498,
      "reward_std": 0.27851989958435297,
      "rewards/accuracy_reward": 0.0625,
      "rewards/cosine_scaled_reward": -0.17486725561320782,
      "step": 83
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3135.541702270508,
      "epoch": 0.096,
      "grad_norm": 0.16200174391269684,
      "kl": 0.0007222890853881836,
      "learning_rate": 7.54295724882796e-07,
      "loss": 0.0357,
      "reward": 0.19974466552957892,
      "reward_std": 0.48104337602853775,
      "rewards/accuracy_reward": 0.25000000186264515,
      "rewards/cosine_scaled_reward": -0.05025534983724356,
      "step": 84
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3218.375045776367,
      "epoch": 0.09714285714285714,
      "grad_norm": 0.15586286783218384,
      "kl": 0.0010872483253479004,
      "learning_rate": 7.472670160550848e-07,
      "loss": 0.0099,
      "reward": 0.09634784981608391,
      "reward_std": 0.6995302718132734,
      "rewards/accuracy_reward": 0.18750000558793545,
      "rewards/cosine_scaled_reward": -0.09115215530619025,
      "step": 85
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3214.8958587646484,
      "epoch": 0.09828571428571428,
      "grad_norm": 0.16115757822990417,
      "kl": 0.0023956298828125,
      "learning_rate": 7.401782177833147e-07,
      "loss": 0.0367,
      "reward": 0.044234735891222954,
      "reward_std": 0.4710828922688961,
      "rewards/accuracy_reward": 0.14583333767950535,
      "rewards/cosine_scaled_reward": -0.10159859777195379,
      "step": 86
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3030.041748046875,
      "epoch": 0.09942857142857142,
      "grad_norm": 0.23102359473705292,
      "kl": 0.007025090977549553,
      "learning_rate": 7.330314893841101e-07,
      "loss": 0.1063,
      "reward": 0.1451429482549429,
      "reward_std": 0.5002225376665592,
      "rewards/accuracy_reward": 0.2083333395421505,
      "rewards/cosine_scaled_reward": -0.0631904061883688,
      "step": 87
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3002.729248046875,
      "epoch": 0.10057142857142858,
      "grad_norm": 0.19206282496452332,
      "kl": 0.008913278579711914,
      "learning_rate": 7.258290078201731e-07,
      "loss": 0.0245,
      "reward": 0.16737306024879217,
      "reward_std": 0.46823734790086746,
      "rewards/accuracy_reward": 0.2083333358168602,
      "rewards/cosine_scaled_reward": -0.040960275335237384,
      "step": 88
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3434.5208435058594,
      "epoch": 0.10171428571428572,
      "grad_norm": 0.15944458544254303,
      "kl": 0.0026093721389770508,
      "learning_rate": 7.185729670371604e-07,
      "loss": 0.0369,
      "reward": 0.0876035988330841,
      "reward_std": 0.35983937978744507,
      "rewards/accuracy_reward": 0.1666666716337204,
      "rewards/cosine_scaled_reward": -0.07906308211386204,
      "step": 89
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2749.937515258789,
      "epoch": 0.10285714285714286,
      "grad_norm": 0.3216039538383484,
      "kl": 0.008787989616394043,
      "learning_rate": 7.11265577295385e-07,
      "loss": -0.007,
      "reward": -0.27455414831638336,
      "reward_std": 0.2550176875665784,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.29538749624043703,
      "step": 90
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3049.4166870117188,
      "epoch": 0.104,
      "grad_norm": 0.21130341291427612,
      "kl": 0.003309965133666992,
      "learning_rate": 7.039090644965509e-07,
      "loss": 0.0554,
      "reward": 0.2680956181138754,
      "reward_std": 0.5306598274037242,
      "rewards/accuracy_reward": 0.22916667349636555,
      "rewards/cosine_scaled_reward": 0.038928942289203405,
      "step": 91
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3023.8959045410156,
      "epoch": 0.10514285714285715,
      "grad_norm": 0.22417984902858734,
      "kl": 0.0150984525680542,
      "learning_rate": 6.965056695057204e-07,
      "loss": 0.0827,
      "reward": 0.06401701085269451,
      "reward_std": 0.4534417111426592,
      "rewards/accuracy_reward": 0.16666666977107525,
      "rewards/cosine_scaled_reward": -0.10264966636896133,
      "step": 92
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3390.0,
      "epoch": 0.10628571428571429,
      "grad_norm": 0.16805854439735413,
      "kl": 0.002878427505493164,
      "learning_rate": 6.890576474687263e-07,
      "loss": 0.0208,
      "reward": -0.25338721089065075,
      "reward_std": 0.18274102546274662,
      "rewards/accuracy_reward": 0.0,
      "rewards/cosine_scaled_reward": -0.25338720716536045,
      "step": 93
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3132.7291870117188,
      "epoch": 0.10742857142857143,
      "grad_norm": 0.23390114307403564,
      "kl": 0.004430115222930908,
      "learning_rate": 6.815672671252315e-07,
      "loss": 0.0391,
      "reward": 0.06390632130205631,
      "reward_std": 0.2537277704104781,
      "rewards/accuracy_reward": 0.1666666679084301,
      "rewards/cosine_scaled_reward": -0.10276034381240606,
      "step": 94
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3494.0208435058594,
      "epoch": 0.10857142857142857,
      "grad_norm": 0.13700152933597565,
      "kl": 0.0011189579963684082,
      "learning_rate": 6.740368101176495e-07,
      "loss": 0.0066,
      "reward": -0.24112762324512005,
      "reward_std": 0.25422961357980967,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.2619609571993351,
      "step": 95
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3054.812530517578,
      "epoch": 0.10971428571428571,
      "grad_norm": 0.1651657223701477,
      "kl": 0.00439077615737915,
      "learning_rate": 6.664685702961344e-07,
      "loss": 0.0172,
      "reward": 0.043599123833701015,
      "reward_std": 0.5089055569842458,
      "rewards/accuracy_reward": 0.16666667349636555,
      "rewards/cosine_scaled_reward": -0.12306755485769827,
      "step": 96
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3421.5833740234375,
      "epoch": 0.11085714285714286,
      "grad_norm": 0.1608337163925171,
      "kl": 0.00201570987701416,
      "learning_rate": 6.588648530198504e-07,
      "loss": 0.0159,
      "reward": 0.07680944725871086,
      "reward_std": 0.4438377758488059,
      "rewards/accuracy_reward": 0.2083333395421505,
      "rewards/cosine_scaled_reward": -0.13152388390153646,
      "step": 97
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3099.375045776367,
      "epoch": 0.112,
      "grad_norm": 0.20414039492607117,
      "kl": 0.0015641450881958008,
      "learning_rate": 6.512279744547392e-07,
      "loss": 0.0324,
      "reward": 0.009776812046766281,
      "reward_std": 0.18312601745128632,
      "rewards/accuracy_reward": 0.125,
      "rewards/cosine_scaled_reward": -0.11522318236529827,
      "step": 98
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2837.979217529297,
      "epoch": 0.11314285714285714,
      "grad_norm": 0.21932294964790344,
      "kl": 0.0025873184204101562,
      "learning_rate": 6.435602608679916e-07,
      "loss": -0.043,
      "reward": 0.07655952998902649,
      "reward_std": 0.24549889098852873,
      "rewards/accuracy_reward": 0.14583333395421505,
      "rewards/cosine_scaled_reward": -0.06927378475666046,
      "step": 99
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3029.0208587646484,
      "epoch": 0.11428571428571428,
      "grad_norm": 0.17687632143497467,
      "kl": 0.005702972412109375,
      "learning_rate": 6.358640479194451e-07,
      "loss": 0.0584,
      "reward": 0.3614894598722458,
      "reward_std": 0.4475015373900533,
      "rewards/accuracy_reward": 0.3125000074505806,
      "rewards/cosine_scaled_reward": 0.048989441245794296,
      "step": 100
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3111.666702270508,
      "epoch": 0.11542857142857142,
      "grad_norm": 0.19027626514434814,
      "kl": 0.0019845962524414062,
      "learning_rate": 6.281416799501187e-07,
      "loss": 0.025,
      "reward": 0.06912354379892349,
      "reward_std": 0.28665477968752384,
      "rewards/accuracy_reward": 0.1666666679084301,
      "rewards/cosine_scaled_reward": -0.09754312154836953,
      "step": 101
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2907.312530517578,
      "epoch": 0.11657142857142858,
      "grad_norm": 0.2261529117822647,
      "kl": 0.006736278533935547,
      "learning_rate": 6.203955092681039e-07,
      "loss": 0.0423,
      "reward": 0.11842460185289383,
      "reward_std": 0.49445752426981926,
      "rewards/accuracy_reward": 0.16666667349636555,
      "rewards/cosine_scaled_reward": -0.0482420613989234,
      "step": 102
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3219.6458587646484,
      "epoch": 0.11771428571428572,
      "grad_norm": 0.20831391215324402,
      "kl": 0.004827260971069336,
      "learning_rate": 6.126278954320294e-07,
      "loss": 0.065,
      "reward": -0.018325381679460406,
      "reward_std": 0.4717335421591997,
      "rewards/accuracy_reward": 0.12500000186264515,
      "rewards/cosine_scaled_reward": -0.14332537725567818,
      "step": 103
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2912.9791870117188,
      "epoch": 0.11885714285714286,
      "grad_norm": 0.17986515164375305,
      "kl": 0.034677982330322266,
      "learning_rate": 6.048412045323164e-07,
      "loss": 0.0421,
      "reward": -0.11950392462313175,
      "reward_std": 0.2809145264327526,
      "rewards/accuracy_reward": 0.0625,
      "rewards/cosine_scaled_reward": -0.18200392089784145,
      "step": 104
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3010.8958435058594,
      "epoch": 0.12,
      "grad_norm": 0.21148350834846497,
      "kl": 0.0034750699996948242,
      "learning_rate": 5.97037808470444e-07,
      "loss": 0.012,
      "reward": 0.14130431599915028,
      "reward_std": 0.47796807438135147,
      "rewards/accuracy_reward": 0.18750000558793545,
      "rewards/cosine_scaled_reward": -0.04619568458292633,
      "step": 105
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2606.0416870117188,
      "epoch": 0.12114285714285715,
      "grad_norm": 0.18517449498176575,
      "kl": 0.014556169509887695,
      "learning_rate": 5.892200842364462e-07,
      "loss": 0.0554,
      "reward": 0.554547032341361,
      "reward_std": 0.45337705919519067,
      "rewards/accuracy_reward": 0.41666666977107525,
      "rewards/cosine_scaled_reward": 0.1378803295083344,
      "step": 106
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3122.312515258789,
      "epoch": 0.12228571428571429,
      "grad_norm": 0.2288726419210434,
      "kl": 0.0027909278869628906,
      "learning_rate": 5.813904131848564e-07,
      "loss": 0.0421,
      "reward": 0.014997333288192749,
      "reward_std": 0.35810419358313084,
      "rewards/accuracy_reward": 0.1458333395421505,
      "rewards/cosine_scaled_reward": -0.1308360043913126,
      "step": 107
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3150.2083587646484,
      "epoch": 0.12342857142857143,
      "grad_norm": 0.22030171751976013,
      "kl": 0.0030778050422668457,
      "learning_rate": 5.735511803093248e-07,
      "loss": 0.0993,
      "reward": 0.12523731286637485,
      "reward_std": 0.47955003939568996,
      "rewards/accuracy_reward": 0.18750000558793545,
      "rewards/cosine_scaled_reward": -0.06226269342005253,
      "step": 108
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3190.062530517578,
      "epoch": 0.12457142857142857,
      "grad_norm": 0.16835717856884003,
      "kl": 0.001190185546875,
      "learning_rate": 5.657047735161255e-07,
      "loss": 0.0296,
      "reward": 0.11127902567386627,
      "reward_std": 0.30289303325116634,
      "rewards/accuracy_reward": 0.1666666679084301,
      "rewards/cosine_scaled_reward": -0.05538764409720898,
      "step": 109
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3129.8750762939453,
      "epoch": 0.12571428571428572,
      "grad_norm": 0.18525457382202148,
      "kl": 0.0027114152908325195,
      "learning_rate": 5.578535828967777e-07,
      "loss": 0.0241,
      "reward": 0.1006783633492887,
      "reward_std": 0.5584467053413391,
      "rewards/accuracy_reward": 0.18750000558793545,
      "rewards/cosine_scaled_reward": -0.08682164316996932,
      "step": 110
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3500.687530517578,
      "epoch": 0.12685714285714286,
      "grad_norm": 0.16640202701091766,
      "kl": 0.003269195556640625,
      "learning_rate": 5.5e-07,
      "loss": 0.0407,
      "reward": 0.06599505990743637,
      "reward_std": 0.5910971527919173,
      "rewards/accuracy_reward": 0.16666666977107525,
      "rewards/cosine_scaled_reward": -0.10067159991012886,
      "step": 111
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3325.8125610351562,
      "epoch": 0.128,
      "grad_norm": 0.16706104576587677,
      "kl": 0.0014832019805908203,
      "learning_rate": 5.421464171032224e-07,
      "loss": 0.0229,
      "reward": 0.19466983899474144,
      "reward_std": 0.5429200492799282,
      "rewards/accuracy_reward": 0.2291666753590107,
      "rewards/cosine_scaled_reward": -0.03449681680649519,
      "step": 112
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3154.7916870117188,
      "epoch": 0.12914285714285714,
      "grad_norm": 0.18158937990665436,
      "kl": 0.011591911315917969,
      "learning_rate": 5.342952264838747e-07,
      "loss": 0.0169,
      "reward": 0.09023613715544343,
      "reward_std": 0.4198597203940153,
      "rewards/accuracy_reward": 0.16666667349636555,
      "rewards/cosine_scaled_reward": -0.07643054611980915,
      "step": 113
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2874.5000381469727,
      "epoch": 0.13028571428571428,
      "grad_norm": 0.15838727355003357,
      "kl": 0.003212451934814453,
      "learning_rate": 5.264488196906752e-07,
      "loss": -0.0056,
      "reward": -0.14509500935673714,
      "reward_std": 0.27863842714577913,
      "rewards/accuracy_reward": 0.0833333358168602,
      "rewards/cosine_scaled_reward": -0.22842834889888763,
      "step": 114
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2977.4791870117188,
      "epoch": 0.13142857142857142,
      "grad_norm": 0.20102404057979584,
      "kl": 0.005918979644775391,
      "learning_rate": 5.186095868151436e-07,
      "loss": 0.0203,
      "reward": 0.08041055500507355,
      "reward_std": 0.34110954217612743,
      "rewards/accuracy_reward": 0.1875,
      "rewards/cosine_scaled_reward": -0.10708944126963615,
      "step": 115
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3477.7708740234375,
      "epoch": 0.13257142857142856,
      "grad_norm": 0.1574009358882904,
      "kl": 0.002073526382446289,
      "learning_rate": 5.107799157635538e-07,
      "loss": -0.0103,
      "reward": 0.10368268750607967,
      "reward_std": 0.2848002705723047,
      "rewards/accuracy_reward": 0.1875,
      "rewards/cosine_scaled_reward": -0.08381731063127518,
      "step": 116
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3352.8125,
      "epoch": 0.1337142857142857,
      "grad_norm": 0.2221246063709259,
      "kl": 0.003772258758544922,
      "learning_rate": 5.02962191529556e-07,
      "loss": 0.046,
      "reward": -0.11389691138174385,
      "reward_std": 0.4279285650700331,
      "rewards/accuracy_reward": 0.06250000186264515,
      "rewards/cosine_scaled_reward": -0.17639691312797368,
      "step": 117
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3320.604248046875,
      "epoch": 0.13485714285714287,
      "grad_norm": 0.19471098482608795,
      "kl": 0.002338886260986328,
      "learning_rate": 4.951587954676837e-07,
      "loss": 0.0529,
      "reward": 0.3609738126397133,
      "reward_std": 0.551423080265522,
      "rewards/accuracy_reward": 0.31250000558793545,
      "rewards/cosine_scaled_reward": 0.04847381100989878,
      "step": 118
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2834.875030517578,
      "epoch": 0.136,
      "grad_norm": 0.18936385214328766,
      "kl": 0.013357400894165039,
      "learning_rate": 4.873721045679706e-07,
      "loss": 0.0535,
      "reward": -0.012741273269057274,
      "reward_std": 0.41901483573019505,
      "rewards/accuracy_reward": 0.1250000037252903,
      "rewards/cosine_scaled_reward": -0.13774127699434757,
      "step": 119
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2937.7916984558105,
      "epoch": 0.13714285714285715,
      "grad_norm": 0.24001307785511017,
      "kl": 0.007732391357421875,
      "learning_rate": 4.79604490731896e-07,
      "loss": -0.0484,
      "reward": -0.006065825000405312,
      "reward_std": 0.15845976071432233,
      "rewards/accuracy_reward": 0.125,
      "rewards/cosine_scaled_reward": -0.13106583431363106,
      "step": 120
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2497.1042098999023,
      "epoch": 0.1382857142857143,
      "grad_norm": 0.17908261716365814,
      "kl": 0.026035308837890625,
      "learning_rate": 4.7185832004988133e-07,
      "loss": 0.0151,
      "reward": 0.11876576207578182,
      "reward_std": 0.3727921061217785,
      "rewards/accuracy_reward": 0.18750000186264515,
      "rewards/cosine_scaled_reward": -0.06873424537479877,
      "step": 121
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3167.729217529297,
      "epoch": 0.13942857142857143,
      "grad_norm": 0.20142368972301483,
      "kl": 0.0050547122955322266,
      "learning_rate": 4.641359520805548e-07,
      "loss": 0.0011,
      "reward": 0.2649596631526947,
      "reward_std": 0.5385044906288385,
      "rewards/accuracy_reward": 0.25000000931322575,
      "rewards/cosine_scaled_reward": 0.014959652442485094,
      "step": 122
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3255.562530517578,
      "epoch": 0.14057142857142857,
      "grad_norm": 0.15558990836143494,
      "kl": 0.0027800798416137695,
      "learning_rate": 4.5643973913200837e-07,
      "loss": 0.0281,
      "reward": -0.0696017425507307,
      "reward_std": 0.36961269890889525,
      "rewards/accuracy_reward": 0.1041666679084301,
      "rewards/cosine_scaled_reward": -0.17376840766519308,
      "step": 123
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2784.2709045410156,
      "epoch": 0.1417142857142857,
      "grad_norm": 0.2072015255689621,
      "kl": 0.026355981826782227,
      "learning_rate": 4.4877202554526084e-07,
      "loss": 0.0621,
      "reward": 0.23046518303453922,
      "reward_std": 0.6176506169140339,
      "rewards/accuracy_reward": 0.2500000074505806,
      "rewards/cosine_scaled_reward": -0.019534816965460777,
      "step": 124
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2942.375,
      "epoch": 0.14285714285714285,
      "grad_norm": 0.17982633411884308,
      "kl": 0.07889890670776367,
      "learning_rate": 4.4113514698014953e-07,
      "loss": -0.0356,
      "reward": 0.20839283242821693,
      "reward_std": 0.37462584115564823,
      "rewards/accuracy_reward": 0.22916666977107525,
      "rewards/cosine_scaled_reward": -0.02077384665608406,
      "step": 125
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2955.916702270508,
      "epoch": 0.144,
      "grad_norm": 0.16790178418159485,
      "kl": 0.0015211105346679688,
      "learning_rate": 4.3353142970386557e-07,
      "loss": 0.0345,
      "reward": 0.08628061786293983,
      "reward_std": 0.33532324619591236,
      "rewards/accuracy_reward": 0.2083333432674408,
      "rewards/cosine_scaled_reward": -0.12205271050333977,
      "step": 126
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3365.2083435058594,
      "epoch": 0.14514285714285713,
      "grad_norm": 0.24060513079166412,
      "kl": 0.0030236244201660156,
      "learning_rate": 4.2596318988235037e-07,
      "loss": 0.0356,
      "reward": -0.12239565665367991,
      "reward_std": 0.304832368157804,
      "rewards/accuracy_reward": 0.0416666679084301,
      "rewards/cosine_scaled_reward": -0.16406232165172696,
      "step": 127
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3064.354217529297,
      "epoch": 0.1462857142857143,
      "grad_norm": 0.17565199732780457,
      "kl": 0.004384756088256836,
      "learning_rate": 4.1843273287476854e-07,
      "loss": 0.0906,
      "reward": 0.28212980553507805,
      "reward_std": 0.6925715450197458,
      "rewards/accuracy_reward": 0.27083333767950535,
      "rewards/cosine_scaled_reward": 0.011296482291072607,
      "step": 128
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3541.0625,
      "epoch": 0.14742857142857144,
      "grad_norm": 0.15770180523395538,
      "kl": 0.0032634735107421875,
      "learning_rate": 4.1094235253127374e-07,
      "loss": 0.0054,
      "reward": -0.060815065167844296,
      "reward_std": 0.25299688428640366,
      "rewards/accuracy_reward": 0.1041666716337204,
      "rewards/cosine_scaled_reward": -0.1649817330762744,
      "step": 129
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3418.1458435058594,
      "epoch": 0.14857142857142858,
      "grad_norm": 0.16763123869895935,
      "kl": 0.0033943653106689453,
      "learning_rate": 4.034943304942796e-07,
      "loss": 0.0367,
      "reward": 0.07925572944805026,
      "reward_std": 0.4248347468674183,
      "rewards/accuracy_reward": 0.1458333395421505,
      "rewards/cosine_scaled_reward": -0.06657761428505182,
      "step": 130
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3062.2292098999023,
      "epoch": 0.14971428571428572,
      "grad_norm": 0.169542133808136,
      "kl": 0.019721031188964844,
      "learning_rate": 3.9609093550344907e-07,
      "loss": -0.0339,
      "reward": 0.32605051435530186,
      "reward_std": 0.5189148262143135,
      "rewards/accuracy_reward": 0.2916666716337204,
      "rewards/cosine_scaled_reward": 0.034383836667984724,
      "step": 131
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3312.1875610351562,
      "epoch": 0.15085714285714286,
      "grad_norm": 0.1781223863363266,
      "kl": 0.0033979415893554688,
      "learning_rate": 3.8873442270461485e-07,
      "loss": 0.0421,
      "reward": 0.27052065124735236,
      "reward_std": 0.7929235212504864,
      "rewards/accuracy_reward": 0.27083334140479565,
      "rewards/cosine_scaled_reward": -0.0003126785159111023,
      "step": 132
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3373.3541870117188,
      "epoch": 0.152,
      "grad_norm": 0.19108432531356812,
      "kl": 0.0029697418212890625,
      "learning_rate": 3.8142703296283953e-07,
      "loss": -0.0076,
      "reward": -0.12809189222753048,
      "reward_std": 0.2327743861824274,
      "rewards/accuracy_reward": 0.14583333395421505,
      "rewards/cosine_scaled_reward": -0.2739252429455519,
      "step": 133
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3061.6666870117188,
      "epoch": 0.15314285714285714,
      "grad_norm": 0.1890539675951004,
      "kl": 0.01691436767578125,
      "learning_rate": 3.7417099217982686e-07,
      "loss": -0.0643,
      "reward": -0.005162534303963184,
      "reward_std": 0.384098834823817,
      "rewards/accuracy_reward": 0.1458333395421505,
      "rewards/cosine_scaled_reward": -0.15099586872383952,
      "step": 134
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2299.3333892822266,
      "epoch": 0.15428571428571428,
      "grad_norm": 0.2244781255722046,
      "kl": 0.014550924301147461,
      "learning_rate": 3.6696851061588994e-07,
      "loss": 0.0863,
      "reward": 0.7467391490936279,
      "reward_std": 0.5308007836574689,
      "rewards/accuracy_reward": 0.5000000093132257,
      "rewards/cosine_scaled_reward": 0.24673915409948677,
      "step": 135
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3300.187530517578,
      "epoch": 0.15542857142857142,
      "grad_norm": 0.18033601343631744,
      "kl": 0.004876136779785156,
      "learning_rate": 3.5982178221668533e-07,
      "loss": 0.0506,
      "reward": 0.16450288891792297,
      "reward_std": 0.5824453700333834,
      "rewards/accuracy_reward": 0.20833333767950535,
      "rewards/cosine_scaled_reward": -0.04383046319708228,
      "step": 136
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3515.1041870117188,
      "epoch": 0.15657142857142858,
      "grad_norm": 0.14998017251491547,
      "kl": 0.005837440490722656,
      "learning_rate": 3.5273298394491515e-07,
      "loss": 0.0241,
      "reward": -0.27306424523703754,
      "reward_std": 0.2572753308340907,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.29389757476747036,
      "step": 137
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3021.7083740234375,
      "epoch": 0.15771428571428572,
      "grad_norm": 0.17564791440963745,
      "kl": 0.0214078426361084,
      "learning_rate": 3.45704275117204e-07,
      "loss": -0.0187,
      "reward": 0.028833418153226376,
      "reward_std": 0.3300330266356468,
      "rewards/accuracy_reward": 0.12500000558793545,
      "rewards/cosine_scaled_reward": -0.09616657719016075,
      "step": 138
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3443.187530517578,
      "epoch": 0.15885714285714286,
      "grad_norm": 0.17364783585071564,
      "kl": 0.003993988037109375,
      "learning_rate": 3.387377967463493e-07,
      "loss": -0.0154,
      "reward": -0.1509188860654831,
      "reward_std": 0.3400895958766341,
      "rewards/accuracy_reward": 0.0416666679084301,
      "rewards/cosine_scaled_reward": -0.19258555606938899,
      "step": 139
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3458.5833740234375,
      "epoch": 0.16,
      "grad_norm": 0.2179875522851944,
      "kl": 0.009866714477539062,
      "learning_rate": 3.3183567088914833e-07,
      "loss": -0.0423,
      "reward": 0.11468263063579798,
      "reward_std": 0.3303603548556566,
      "rewards/accuracy_reward": 0.1666666679084301,
      "rewards/cosine_scaled_reward": -0.05198403773829341,
      "step": 140
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3385.3958740234375,
      "epoch": 0.16114285714285714,
      "grad_norm": 0.17362990975379944,
      "kl": 0.0052642822265625,
      "learning_rate": 3.250000000000001e-07,
      "loss": -0.0233,
      "reward": 0.12504707090556622,
      "reward_std": 0.5917222537100315,
      "rewards/accuracy_reward": 0.20833333767950535,
      "rewards/cosine_scaled_reward": -0.08328627422451973,
      "step": 141
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3200.6459045410156,
      "epoch": 0.16228571428571428,
      "grad_norm": 0.19476763904094696,
      "kl": 0.005651950836181641,
      "learning_rate": 3.182328662904756e-07,
      "loss": 0.0829,
      "reward": -0.07443058118224144,
      "reward_std": 0.4768957942724228,
      "rewards/accuracy_reward": 0.1250000037252903,
      "rewards/cosine_scaled_reward": -0.1994305873522535,
      "step": 142
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3258.6875610351562,
      "epoch": 0.16342857142857142,
      "grad_norm": 0.23263806104660034,
      "kl": 0.00710296630859375,
      "learning_rate": 3.115363310950578e-07,
      "loss": 0.0686,
      "reward": -0.07686329772695899,
      "reward_std": 0.355038208886981,
      "rewards/accuracy_reward": 0.10416666977107525,
      "rewards/cosine_scaled_reward": -0.18102996051311493,
      "step": 143
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3194.75,
      "epoch": 0.16457142857142856,
      "grad_norm": 0.17562638223171234,
      "kl": 0.0062198638916015625,
      "learning_rate": 3.0491243424323783e-07,
      "loss": -0.0268,
      "reward": -0.06668785959482193,
      "reward_std": 0.3318067956715822,
      "rewards/accuracy_reward": 0.12500000558793545,
      "rewards/cosine_scaled_reward": -0.19168786704540253,
      "step": 144
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2559.2500076293945,
      "epoch": 0.1657142857142857,
      "grad_norm": 0.24671532213687897,
      "kl": 0.009662628173828125,
      "learning_rate": 2.9836319343816397e-07,
      "loss": -0.0379,
      "reward": 0.37313266890123487,
      "reward_std": 0.33964273426681757,
      "rewards/accuracy_reward": 0.31250000186264515,
      "rewards/cosine_scaled_reward": 0.060632668901234865,
      "step": 145
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3086.375015258789,
      "epoch": 0.16685714285714287,
      "grad_norm": 0.15435300767421722,
      "kl": 0.0024802684783935547,
      "learning_rate": 2.918906036420294e-07,
      "loss": 0.0125,
      "reward": -0.19754038006067276,
      "reward_std": 0.2727511157281697,
      "rewards/accuracy_reward": 0.0416666679084301,
      "rewards/cosine_scaled_reward": -0.23920704424381256,
      "step": 146
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3410.8541870117188,
      "epoch": 0.168,
      "grad_norm": 0.2101191282272339,
      "kl": 0.004631996154785156,
      "learning_rate": 2.854966364683872e-07,
      "loss": 0.0229,
      "reward": -0.13149442267604172,
      "reward_std": 0.3892161212861538,
      "rewards/accuracy_reward": 0.06250000186264515,
      "rewards/cosine_scaled_reward": -0.19399441499263048,
      "step": 147
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2876.1250610351562,
      "epoch": 0.16914285714285715,
      "grad_norm": 0.18126218020915985,
      "kl": 0.008637428283691406,
      "learning_rate": 2.791832395815782e-07,
      "loss": 0.0202,
      "reward": -0.013470180332660675,
      "reward_std": 0.2522149607539177,
      "rewards/accuracy_reward": 0.14583333395421505,
      "rewards/cosine_scaled_reward": -0.15930352546274662,
      "step": 148
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3289.3750610351562,
      "epoch": 0.1702857142857143,
      "grad_norm": 0.16630330681800842,
      "kl": 0.004746913909912109,
      "learning_rate": 2.729523361034538e-07,
      "loss": 0.0387,
      "reward": 0.05808070907369256,
      "reward_std": 0.4934985414147377,
      "rewards/accuracy_reward": 0.16666666977107525,
      "rewards/cosine_scaled_reward": -0.10858596302568913,
      "step": 149
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3029.5417098999023,
      "epoch": 0.17142857142857143,
      "grad_norm": 0.16553689539432526,
      "kl": 0.007030487060546875,
      "learning_rate": 2.6680582402757324e-07,
      "loss": 0.0283,
      "reward": 0.3249948853626847,
      "reward_std": 0.5702541470527649,
      "rewards/accuracy_reward": 0.29166667722165585,
      "rewards/cosine_scaled_reward": 0.03332819044589996,
      "step": 150
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3161.3750610351562,
      "epoch": 0.17257142857142857,
      "grad_norm": 0.18477413058280945,
      "kl": 0.005667924880981445,
      "learning_rate": 2.6074557564105724e-07,
      "loss": 0.0607,
      "reward": 0.26313829235732555,
      "reward_std": 0.5168293453752995,
      "rewards/accuracy_reward": 0.25000000186264515,
      "rewards/cosine_scaled_reward": 0.013138292357325554,
      "step": 151
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3020.750011444092,
      "epoch": 0.1737142857142857,
      "grad_norm": 0.25039243698120117,
      "kl": 0.0038559436798095703,
      "learning_rate": 2.547734369542718e-07,
      "loss": -0.0343,
      "reward": -0.17000860534608364,
      "reward_std": 0.290488900616765,
      "rewards/accuracy_reward": 0.0416666679084301,
      "rewards/cosine_scaled_reward": -0.21167527325451374,
      "step": 152
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3283.4583587646484,
      "epoch": 0.17485714285714285,
      "grad_norm": 0.2211620956659317,
      "kl": 0.025887489318847656,
      "learning_rate": 2.488912271385139e-07,
      "loss": 0.0627,
      "reward": -0.1608979816082865,
      "reward_std": 0.30768819246441126,
      "rewards/accuracy_reward": 0.06250000186264515,
      "rewards/cosine_scaled_reward": -0.22339799162000418,
      "step": 153
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3509.625,
      "epoch": 0.176,
      "grad_norm": 0.15299846231937408,
      "kl": 0.0022106170654296875,
      "learning_rate": 2.4310073797187573e-07,
      "loss": 0.0166,
      "reward": 0.2281611803919077,
      "reward_std": 0.6786817125976086,
      "rewards/accuracy_reward": 0.2291666716337204,
      "rewards/cosine_scaled_reward": -0.0010055112652480602,
      "step": 154
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2825.2708587646484,
      "epoch": 0.17714285714285713,
      "grad_norm": 0.22317463159561157,
      "kl": 0.022104263305664062,
      "learning_rate": 2.374037332934512e-07,
      "loss": -0.0181,
      "reward": 0.3611069116741419,
      "reward_std": 0.5537243597209454,
      "rewards/accuracy_reward": 0.3125000111758709,
      "rewards/cosine_scaled_reward": 0.04860687721520662,
      "step": 155
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3509.3333435058594,
      "epoch": 0.1782857142857143,
      "grad_norm": 0.20389185845851898,
      "kl": 0.004364371299743652,
      "learning_rate": 2.3180194846605364e-07,
      "loss": 0.0435,
      "reward": -0.07293254625983536,
      "reward_std": 0.3572534155100584,
      "rewards/accuracy_reward": 0.08333333395421505,
      "rewards/cosine_scaled_reward": -0.15626587718725204,
      "step": 156
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3416.7291870117188,
      "epoch": 0.17942857142857144,
      "grad_norm": 0.16730445623397827,
      "kl": 0.005460262298583984,
      "learning_rate": 2.2629708984760706e-07,
      "loss": 0.0057,
      "reward": -0.07636097725480795,
      "reward_std": 0.2699170485138893,
      "rewards/accuracy_reward": 0.0833333358168602,
      "rewards/cosine_scaled_reward": -0.15969432052224874,
      "step": 157
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3422.0625610351562,
      "epoch": 0.18057142857142858,
      "grad_norm": 0.14682160317897797,
      "kl": 0.0034723281860351562,
      "learning_rate": 2.2089083427137329e-07,
      "loss": 0.0179,
      "reward": 0.32586812041699886,
      "reward_std": 0.3167658653110266,
      "rewards/accuracy_reward": 0.25000000558793545,
      "rewards/cosine_scaled_reward": 0.07586812693625689,
      "step": 158
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3442.6041870117188,
      "epoch": 0.18171428571428572,
      "grad_norm": 0.15924254059791565,
      "kl": 0.014659881591796875,
      "learning_rate": 2.1558482853517253e-07,
      "loss": 0.0161,
      "reward": -0.2088061198592186,
      "reward_std": 0.26268018409609795,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.2296394552104175,
      "step": 159
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3276.9791870117188,
      "epoch": 0.18285714285714286,
      "grad_norm": 0.1880086064338684,
      "kl": 0.008440017700195312,
      "learning_rate": 2.1038068889975259e-07,
      "loss": 0.0752,
      "reward": 0.1678475480293855,
      "reward_std": 0.5140087231993675,
      "rewards/accuracy_reward": 0.2291666753590107,
      "rewards/cosine_scaled_reward": -0.06131910812109709,
      "step": 160
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3382.1875610351562,
      "epoch": 0.184,
      "grad_norm": 0.1801372915506363,
      "kl": 0.015121936798095703,
      "learning_rate": 2.0528000059645995e-07,
      "loss": 0.0314,
      "reward": 0.0975832249969244,
      "reward_std": 0.4436509357765317,
      "rewards/accuracy_reward": 0.16666667349636555,
      "rewards/cosine_scaled_reward": -0.06908343941904604,
      "step": 161
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3415.6875,
      "epoch": 0.18514285714285714,
      "grad_norm": 0.235793799161911,
      "kl": 0.00775146484375,
      "learning_rate": 2.0028431734436308e-07,
      "loss": 0.0693,
      "reward": 0.02907794527709484,
      "reward_std": 0.47673553973436356,
      "rewards/accuracy_reward": 0.14583333395421505,
      "rewards/cosine_scaled_reward": -0.11675538681447506,
      "step": 162
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2728.625045776367,
      "epoch": 0.18628571428571428,
      "grad_norm": 0.2469216287136078,
      "kl": 0.004032135009765625,
      "learning_rate": 1.9539516087697517e-07,
      "loss": 0.1104,
      "reward": 0.4551220340654254,
      "reward_std": 0.6010043127462268,
      "rewards/accuracy_reward": 0.33333333767950535,
      "rewards/cosine_scaled_reward": 0.12178870104253292,
      "step": 163
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2826.9166870117188,
      "epoch": 0.18742857142857142,
      "grad_norm": 0.2510416507720947,
      "kl": 0.010884761810302734,
      "learning_rate": 1.9061402047871833e-07,
      "loss": 0.0495,
      "reward": 0.29104864224791527,
      "reward_std": 0.41844242811203003,
      "rewards/accuracy_reward": 0.2708333395421505,
      "rewards/cosine_scaled_reward": 0.020215285941958427,
      "step": 164
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3338.875030517578,
      "epoch": 0.18857142857142858,
      "grad_norm": 0.17247295379638672,
      "kl": 0.005005836486816406,
      "learning_rate": 1.8594235253127372e-07,
      "loss": 0.0091,
      "reward": -0.10422082059085369,
      "reward_std": 0.4068410564213991,
      "rewards/accuracy_reward": 0.08333333395421505,
      "rewards/cosine_scaled_reward": -0.18755416944622993,
      "step": 165
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3450.0833740234375,
      "epoch": 0.18971428571428572,
      "grad_norm": 0.13382409512996674,
      "kl": 0.002165555953979492,
      "learning_rate": 1.8138158006995363e-07,
      "loss": 0.0106,
      "reward": 0.05564507842063904,
      "reward_std": 0.5360854268074036,
      "rewards/accuracy_reward": 0.1458333358168602,
      "rewards/cosine_scaled_reward": -0.09018826205283403,
      "step": 166
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2906.791748046875,
      "epoch": 0.19085714285714286,
      "grad_norm": 0.17852716147899628,
      "kl": 0.008731842041015625,
      "learning_rate": 1.7693309235023127e-07,
      "loss": 0.0562,
      "reward": 0.060366696037817746,
      "reward_std": 0.4651718642562628,
      "rewards/accuracy_reward": 0.14583333395421505,
      "rewards/cosine_scaled_reward": -0.08546665217727423,
      "step": 167
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3323.2291870117188,
      "epoch": 0.192,
      "grad_norm": 0.14015178382396698,
      "kl": 0.00559234619140625,
      "learning_rate": 1.7259824442455923e-07,
      "loss": 0.0001,
      "reward": -0.00011509843170642853,
      "reward_std": 0.48929045908153057,
      "rewards/accuracy_reward": 0.12500000186264515,
      "rewards/cosine_scaled_reward": -0.12511511333286762,
      "step": 168
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2785.3333587646484,
      "epoch": 0.19314285714285714,
      "grad_norm": 0.17836594581604004,
      "kl": 0.0029582977294921875,
      "learning_rate": 1.6837835672960831e-07,
      "loss": 0.0415,
      "reward": 0.7883107475936413,
      "reward_std": 0.31396112497895956,
      "rewards/accuracy_reward": 0.5416666679084301,
      "rewards/cosine_scaled_reward": 0.24664408713579178,
      "step": 169
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2772.9375228881836,
      "epoch": 0.19428571428571428,
      "grad_norm": 0.1806839257478714,
      "kl": 0.006674766540527344,
      "learning_rate": 1.6427471468404952e-07,
      "loss": 0.0447,
      "reward": 0.19183126650750637,
      "reward_std": 0.3847146909683943,
      "rewards/accuracy_reward": 0.2291666679084301,
      "rewards/cosine_scaled_reward": -0.03733538277447224,
      "step": 170
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3159.541717529297,
      "epoch": 0.19542857142857142,
      "grad_norm": 0.17659954726696014,
      "kl": 0.002955913543701172,
      "learning_rate": 1.6028856829700258e-07,
      "loss": 0.0338,
      "reward": 0.261606702581048,
      "reward_std": 0.4677132572978735,
      "rewards/accuracy_reward": 0.2500000074505806,
      "rewards/cosine_scaled_reward": 0.011606710031628609,
      "step": 171
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3158.8958435058594,
      "epoch": 0.19657142857142856,
      "grad_norm": 0.1956152766942978,
      "kl": 0.006173133850097656,
      "learning_rate": 1.5642113178727193e-07,
      "loss": 0.0097,
      "reward": 0.2144802352413535,
      "reward_std": 0.46392463706433773,
      "rewards/accuracy_reward": 0.2500000037252903,
      "rewards/cosine_scaled_reward": -0.035519770928658545,
      "step": 172
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2472.437530517578,
      "epoch": 0.1977142857142857,
      "grad_norm": 0.254905641078949,
      "kl": 0.008039474487304688,
      "learning_rate": 1.5267358321348285e-07,
      "loss": 0.0255,
      "reward": -0.041594887152314186,
      "reward_std": 0.4095227625221014,
      "rewards/accuracy_reward": 0.1250000037252903,
      "rewards/cosine_scaled_reward": -0.16659490356687456,
      "step": 173
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3101.9584045410156,
      "epoch": 0.19885714285714284,
      "grad_norm": 0.18561899662017822,
      "kl": 0.0075778961181640625,
      "learning_rate": 1.4904706411523448e-07,
      "loss": 0.035,
      "reward": 0.24006008356809616,
      "reward_std": 0.5043942555785179,
      "rewards/accuracy_reward": 0.22916666977107525,
      "rewards/cosine_scaled_reward": 0.010893410071730614,
      "step": 174
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2890.500030517578,
      "epoch": 0.2,
      "grad_norm": 0.17110423743724823,
      "kl": 0.01293182373046875,
      "learning_rate": 1.4554267916537495e-07,
      "loss": 0.0116,
      "reward": 0.32882001250982285,
      "reward_std": 0.2852998124435544,
      "rewards/accuracy_reward": 0.3125,
      "rewards/cosine_scaled_reward": 0.016320008784532547,
      "step": 175
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3115.0208892822266,
      "epoch": 0.20114285714285715,
      "grad_norm": 0.2752262055873871,
      "kl": 0.008016586303710938,
      "learning_rate": 1.4216149583350755e-07,
      "loss": 0.0955,
      "reward": 0.15877557545900345,
      "reward_std": 0.6403312431648374,
      "rewards/accuracy_reward": 0.20833333767950535,
      "rewards/cosine_scaled_reward": -0.049557752907276154,
      "step": 176
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3386.5833435058594,
      "epoch": 0.2022857142857143,
      "grad_norm": 0.1783076673746109,
      "kl": 0.019231557846069336,
      "learning_rate": 1.3890454406082956e-07,
      "loss": -0.0247,
      "reward": 0.05885206814855337,
      "reward_std": 0.3794629080221057,
      "rewards/accuracy_reward": 0.125,
      "rewards/cosine_scaled_reward": -0.06614794302731752,
      "step": 177
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3214.7291870117188,
      "epoch": 0.20342857142857143,
      "grad_norm": 0.1741504669189453,
      "kl": 0.021752357482910156,
      "learning_rate": 1.3577281594640182e-07,
      "loss": -0.019,
      "reward": 0.19468790292739868,
      "reward_std": 0.5135045610368252,
      "rewards/accuracy_reward": 0.2083333395421505,
      "rewards/cosine_scaled_reward": -0.013645432889461517,
      "step": 178
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3233.4166870117188,
      "epoch": 0.20457142857142857,
      "grad_norm": 0.16726212203502655,
      "kl": 0.0071485042572021484,
      "learning_rate": 1.3276726544494571e-07,
      "loss": 0.0061,
      "reward": -0.21683532558381557,
      "reward_std": 0.26457902044057846,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.23766866483492777,
      "step": 179
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2661.5000228881836,
      "epoch": 0.2057142857142857,
      "grad_norm": 0.30925920605659485,
      "kl": 0.015074729919433594,
      "learning_rate": 1.2988880807625927e-07,
      "loss": 0.0328,
      "reward": 0.512437904253602,
      "reward_std": 0.387176813557744,
      "rewards/accuracy_reward": 0.3958333358168602,
      "rewards/cosine_scaled_reward": 0.11660455353558064,
      "step": 180
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3386.437530517578,
      "epoch": 0.20685714285714285,
      "grad_norm": 0.2104531228542328,
      "kl": 0.0052337646484375,
      "learning_rate": 1.2713832064634125e-07,
      "loss": 0.0401,
      "reward": -0.050248525105416775,
      "reward_std": 0.3244944280013442,
      "rewards/accuracy_reward": 0.10416666977107525,
      "rewards/cosine_scaled_reward": -0.15441520046442747,
      "step": 181
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2823.104179382324,
      "epoch": 0.208,
      "grad_norm": 0.1794274002313614,
      "kl": 0.02399587631225586,
      "learning_rate": 1.2451664098030743e-07,
      "loss": -0.0603,
      "reward": 0.1379284020513296,
      "reward_std": 0.4621511232107878,
      "rewards/accuracy_reward": 0.18750000558793545,
      "rewards/cosine_scaled_reward": -0.049571617506444454,
      "step": 182
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3123.6458435058594,
      "epoch": 0.20914285714285713,
      "grad_norm": 0.17760169506072998,
      "kl": 0.023395538330078125,
      "learning_rate": 1.220245676671809e-07,
      "loss": 0.0078,
      "reward": 0.16959657333791256,
      "reward_std": 0.3798244297504425,
      "rewards/accuracy_reward": 0.2291666679084301,
      "rewards/cosine_scaled_reward": -0.059570083394646645,
      "step": 183
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3245.2708435058594,
      "epoch": 0.2102857142857143,
      "grad_norm": 0.19478560984134674,
      "kl": 0.005706787109375,
      "learning_rate": 1.1966285981663407e-07,
      "loss": -0.0293,
      "reward": -0.055914007127285004,
      "reward_std": 0.2492837980389595,
      "rewards/accuracy_reward": 0.1041666716337204,
      "rewards/cosine_scaled_reward": -0.1600806824862957,
      "step": 184
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3321.2708740234375,
      "epoch": 0.21142857142857144,
      "grad_norm": 0.17917311191558838,
      "kl": 0.008937358856201172,
      "learning_rate": 1.1743223682775649e-07,
      "loss": 0.0092,
      "reward": -0.04576564393937588,
      "reward_std": 0.39649237506091595,
      "rewards/accuracy_reward": 0.1041666679084301,
      "rewards/cosine_scaled_reward": -0.14993230905383825,
      "step": 185
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3297.6458740234375,
      "epoch": 0.21257142857142858,
      "grad_norm": 0.1722106635570526,
      "kl": 0.003765106201171875,
      "learning_rate": 1.1533337816991931e-07,
      "loss": 0.0326,
      "reward": 0.4093305990099907,
      "reward_std": 0.3512581354007125,
      "rewards/accuracy_reward": 0.3333333432674408,
      "rewards/cosine_scaled_reward": 0.0759972594678402,
      "step": 186
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3079.541717529297,
      "epoch": 0.21371428571428572,
      "grad_norm": 0.27707189321517944,
      "kl": 0.00933837890625,
      "learning_rate": 1.1336692317580158e-07,
      "loss": -0.0215,
      "reward": -0.08524070866405964,
      "reward_std": 0.38815517351031303,
      "rewards/accuracy_reward": 0.08333333395421505,
      "rewards/cosine_scaled_reward": -0.16857404820621014,
      "step": 187
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3571.2916870117188,
      "epoch": 0.21485714285714286,
      "grad_norm": 0.1520886868238449,
      "kl": 0.0035920143127441406,
      "learning_rate": 1.1153347084664419e-07,
      "loss": 0.002,
      "reward": -0.1857756022363901,
      "reward_std": 0.2816652413457632,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.20660893432796001,
      "step": 188
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3236.5208587646484,
      "epoch": 0.216,
      "grad_norm": 0.19879254698753357,
      "kl": 0.006503105163574219,
      "learning_rate": 1.0983357966978745e-07,
      "loss": 0.0444,
      "reward": -0.016374904662370682,
      "reward_std": 0.4003318352624774,
      "rewards/accuracy_reward": 0.1458333358168602,
      "rewards/cosine_scaled_reward": -0.1622082181274891,
      "step": 189
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3082.4583740234375,
      "epoch": 0.21714285714285714,
      "grad_norm": 0.16634653508663177,
      "kl": 0.00673985481262207,
      "learning_rate": 1.0826776744855121e-07,
      "loss": 0.0302,
      "reward": 0.24105202872306108,
      "reward_std": 0.5610934533178806,
      "rewards/accuracy_reward": 0.2708333358168602,
      "rewards/cosine_scaled_reward": -0.02978132525458932,
      "step": 190
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2954.7708740234375,
      "epoch": 0.21828571428571428,
      "grad_norm": 0.2336689531803131,
      "kl": 0.009424209594726562,
      "learning_rate": 1.068365111445064e-07,
      "loss": 0.121,
      "reward": 0.170956801623106,
      "reward_std": 0.3355375565588474,
      "rewards/accuracy_reward": 0.2083333432674408,
      "rewards/cosine_scaled_reward": -0.037376536056399345,
      "step": 191
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3462.2916870117188,
      "epoch": 0.21942857142857142,
      "grad_norm": 0.1604224145412445,
      "kl": 0.0033931732177734375,
      "learning_rate": 1.0554024673218806e-07,
      "loss": 0.0184,
      "reward": -0.08346282877027988,
      "reward_std": 0.28180871857330203,
      "rewards/accuracy_reward": 0.0833333358168602,
      "rewards/cosine_scaled_reward": -0.16679617203772068,
      "step": 192
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3324.479202270508,
      "epoch": 0.22057142857142858,
      "grad_norm": 0.23563383519649506,
      "kl": 0.004616260528564453,
      "learning_rate": 1.0437936906629334e-07,
      "loss": 0.0425,
      "reward": -0.1967187790432945,
      "reward_std": 0.2698577819392085,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.21755211427807808,
      "step": 193
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3275.7916870117188,
      "epoch": 0.22171428571428572,
      "grad_norm": 0.14834217727184296,
      "kl": 0.002389192581176758,
      "learning_rate": 1.0335423176140511e-07,
      "loss": -0.012,
      "reward": 0.502600722014904,
      "reward_std": 0.4794109221547842,
      "rewards/accuracy_reward": 0.3541666679084301,
      "rewards/cosine_scaled_reward": 0.14843403454869986,
      "step": 194
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3356.8541870117188,
      "epoch": 0.22285714285714286,
      "grad_norm": 0.16246356070041656,
      "kl": 0.0031523704528808594,
      "learning_rate": 1.0246514708427701e-07,
      "loss": 0.0019,
      "reward": 0.14120314689353108,
      "reward_std": 0.36129687167704105,
      "rewards/accuracy_reward": 0.20833333395421505,
      "rewards/cosine_scaled_reward": -0.06713018752634525,
      "step": 195
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3576.2291870117188,
      "epoch": 0.224,
      "grad_norm": 0.15277273952960968,
      "kl": 0.0033049583435058594,
      "learning_rate": 1.017123858587145e-07,
      "loss": 0.0026,
      "reward": -0.04466534685343504,
      "reward_std": 0.37323744408786297,
      "rewards/accuracy_reward": 0.08333333395421505,
      "rewards/cosine_scaled_reward": -0.12799868267029524,
      "step": 196
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2975.020835876465,
      "epoch": 0.22514285714285714,
      "grad_norm": 0.24708448350429535,
      "kl": 0.008154869079589844,
      "learning_rate": 1.0109617738307911e-07,
      "loss": 0.0281,
      "reward": 0.3771647736430168,
      "reward_std": 0.6537368502467871,
      "rewards/accuracy_reward": 0.2916666753590107,
      "rewards/cosine_scaled_reward": 0.08549809700343758,
      "step": 197
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3131.687530517578,
      "epoch": 0.22628571428571428,
      "grad_norm": 0.17502129077911377,
      "kl": 0.016117095947265625,
      "learning_rate": 1.0061670936044178e-07,
      "loss": -0.0111,
      "reward": 0.12027511559426785,
      "reward_std": 0.26299865636974573,
      "rewards/accuracy_reward": 0.2083333358168602,
      "rewards/cosine_scaled_reward": -0.08805821277201176,
      "step": 198
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 3566.75,
      "epoch": 0.22742857142857142,
      "grad_norm": 0.14533254504203796,
      "kl": 0.005496978759765625,
      "learning_rate": 1.002741278414069e-07,
      "loss": 0.0099,
      "reward": -0.23100331239402294,
      "reward_std": 0.262982377782464,
      "rewards/accuracy_reward": 0.02083333395421505,
      "rewards/cosine_scaled_reward": -0.25183664448559284,
      "step": 199
    },
    {
      "clip_fraction": 0.0,
      "completion_length": 2976.791717529297,
      "epoch": 0.22857142857142856,
      "grad_norm": 0.18751339614391327,
      "kl": 0.0031397342681884766,
      "learning_rate": 1.0006853717962393e-07,
      "loss": 0.0796,
      "reward": 0.2756543140858412,
      "reward_std": 0.398221081122756,
      "rewards/accuracy_reward": 0.2708333358168602,
      "rewards/cosine_scaled_reward": 0.004820965230464935,
      "step": 200
    },
    {
      "epoch": 0.22857142857142856,
      "step": 200,
      "total_flos": 0.0,
      "train_loss": 0.024151897984629613,
      "train_runtime": 32217.4606,
      "train_samples_per_second": 0.298,
      "train_steps_per_second": 0.006
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 10,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}