{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999074845036544,
  "eval_steps": 100,
  "global_step": 2702,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001850309926912758,
      "grad_norm": 0.0,
      "learning_rate": 3.690036900369004e-07,
      "loss": 2.8171,
      "mean_token_accuracy": 0.7238034188034186,
      "mean_token_accuracy_utility": 0.6260905477858321,
      "step": 5
    },
    {
      "epoch": 0.003700619853825516,
      "grad_norm": 0.0,
      "learning_rate": 7.380073800738008e-07,
      "loss": 2.8457,
      "mean_token_accuracy": 0.7138827838827838,
      "mean_token_accuracy_utility": 0.6238034188034186,
      "step": 10
    },
    {
      "epoch": 0.005550929780738274,
      "grad_norm": 0.0,
      "learning_rate": 1.1070110701107011e-06,
      "loss": 2.8883,
      "mean_token_accuracy": 0.7148412698412696,
      "mean_token_accuracy_utility": 0.62019536019536,
      "step": 15
    },
    {
      "epoch": 0.007401239707651032,
      "grad_norm": 0.0,
      "learning_rate": 1.4760147601476015e-06,
      "loss": 2.8438,
      "mean_token_accuracy": 0.7274542124542125,
      "mean_token_accuracy_utility": 0.6184126984126983,
      "step": 20
    },
    {
      "epoch": 0.00925154963456379,
      "grad_norm": 0.0,
      "learning_rate": 1.845018450184502e-06,
      "loss": 2.8529,
      "mean_token_accuracy": 0.7137912087912086,
      "mean_token_accuracy_utility": 0.6252075702075699,
      "step": 25
    },
    {
      "epoch": 0.011101859561476548,
      "grad_norm": 0.0,
      "learning_rate": 2.2140221402214023e-06,
      "loss": 2.8891,
      "mean_token_accuracy": 0.7092063492063491,
      "mean_token_accuracy_utility": 0.6194139194139192,
      "step": 30
    },
    {
      "epoch": 0.012952169488389306,
      "grad_norm": 0.0,
      "learning_rate": 2.5830258302583027e-06,
      "loss": 2.8161,
      "mean_token_accuracy": 0.7136446886446884,
      "mean_token_accuracy_utility": 0.6317521367521366,
      "step": 35
    },
    {
      "epoch": 0.014802479415302064,
      "grad_norm": 0.0,
      "learning_rate": 2.952029520295203e-06,
      "loss": 2.7916,
      "mean_token_accuracy": 0.7236935286935287,
      "mean_token_accuracy_utility": 0.6272010358204574,
      "step": 40
    },
    {
      "epoch": 0.01665278934221482,
      "grad_norm": 0.0,
      "learning_rate": 3.3210332103321034e-06,
      "loss": 2.7992,
      "mean_token_accuracy": 0.7117094017094017,
      "mean_token_accuracy_utility": 0.6384493284493282,
      "step": 45
    },
    {
      "epoch": 0.01850309926912758,
      "grad_norm": 0.0,
      "learning_rate": 3.690036900369004e-06,
      "loss": 2.7869,
      "mean_token_accuracy": 0.7181429681429681,
      "mean_token_accuracy_utility": 0.6290598290598288,
      "step": 50
    },
    {
      "epoch": 0.020353409196040336,
      "grad_norm": 0.0,
      "learning_rate": 4.059040590405905e-06,
      "loss": 2.8084,
      "mean_token_accuracy": 0.7122954822954821,
      "mean_token_accuracy_utility": 0.6367182817182816,
      "step": 55
    },
    {
      "epoch": 0.022203719122953096,
      "grad_norm": 0.0,
      "learning_rate": 4.428044280442805e-06,
      "loss": 2.8437,
      "mean_token_accuracy": 0.7153846153846154,
      "mean_token_accuracy_utility": 0.6183150183150181,
      "step": 60
    },
    {
      "epoch": 0.024054029049865852,
      "grad_norm": 0.0,
      "learning_rate": 4.797047970479705e-06,
      "loss": 2.7564,
      "mean_token_accuracy": 0.7119719169719169,
      "mean_token_accuracy_utility": 0.6417338779579761,
      "step": 65
    },
    {
      "epoch": 0.025904338976778612,
      "grad_norm": 0.0,
      "learning_rate": 5.166051660516605e-06,
      "loss": 2.7147,
      "mean_token_accuracy": 0.7390903540903542,
      "mean_token_accuracy_utility": 0.6293345543345542,
      "step": 70
    },
    {
      "epoch": 0.027754648903691368,
      "grad_norm": 0.0,
      "learning_rate": 5.535055350553506e-06,
      "loss": 2.8073,
      "mean_token_accuracy": 0.7153724053724051,
      "mean_token_accuracy_utility": 0.624133089133089,
      "step": 75
    },
    {
      "epoch": 0.029604958830604128,
      "grad_norm": 0.0,
      "learning_rate": 5.904059040590406e-06,
      "loss": 2.8563,
      "mean_token_accuracy": 0.7114652014652013,
      "mean_token_accuracy_utility": 0.6210561660561659,
      "step": 80
    },
    {
      "epoch": 0.031455268757516884,
      "grad_norm": 0.0,
      "learning_rate": 6.273062730627307e-06,
      "loss": 2.7662,
      "mean_token_accuracy": 0.7116361416361415,
      "mean_token_accuracy_utility": 0.6346658669950106,
      "step": 85
    },
    {
      "epoch": 0.03330557868442964,
      "grad_norm": 0.0,
      "learning_rate": 6.642066420664207e-06,
      "loss": 2.769,
      "mean_token_accuracy": 0.7249145299145299,
      "mean_token_accuracy_utility": 0.6265567765567764,
      "step": 90
    },
    {
      "epoch": 0.0351558886113424,
      "grad_norm": 0.0,
      "learning_rate": 7.011070110701108e-06,
      "loss": 2.8409,
      "mean_token_accuracy": 0.7199572649572648,
      "mean_token_accuracy_utility": 0.6158634819525906,
      "step": 95
    },
    {
      "epoch": 0.03700619853825516,
      "grad_norm": 0.0,
      "learning_rate": 7.380073800738008e-06,
      "loss": 2.7634,
      "mean_token_accuracy": 0.7178144078144078,
      "mean_token_accuracy_utility": 0.6318131868131865,
      "step": 100
    },
    {
      "epoch": 0.03700619853825516,
      "eval_mean_token_accuracy": 0.7045143309849191,
      "eval_reasoning_loss": 1.126886248588562,
      "eval_reasoning_runtime": 29.9871,
      "eval_reasoning_samples_per_second": 4.302,
      "eval_reasoning_steps_per_second": 1.1,
      "step": 100
    },
    {
      "epoch": 0.03700619853825516,
      "eval_mean_token_accuracy": 0.6319856405787203,
      "eval_utility_loss": 1.673980712890625,
      "eval_utility_runtime": 11.0493,
      "eval_utility_samples_per_second": 4.435,
      "eval_utility_steps_per_second": 1.177,
      "step": 100
    },
    {
      "epoch": 0.038856508465167916,
      "grad_norm": 0.0,
      "learning_rate": 7.749077490774908e-06,
      "loss": 2.7357,
      "mean_token_accuracy": 0.7257692307692307,
      "mean_token_accuracy_utility": 0.6388766788766788,
      "step": 105
    },
    {
      "epoch": 0.04070681839208067,
      "grad_norm": 0.0,
      "learning_rate": 8.11808118081181e-06,
      "loss": 2.7987,
      "mean_token_accuracy": 0.7136568986568987,
      "mean_token_accuracy_utility": 0.6325702075702073,
      "step": 110
    },
    {
      "epoch": 0.04255712831899343,
      "grad_norm": 0.0,
      "learning_rate": 8.48708487084871e-06,
      "loss": 2.7058,
      "mean_token_accuracy": 0.7229426129426129,
      "mean_token_accuracy_utility": 0.6381440781440779,
      "step": 115
    },
    {
      "epoch": 0.04440743824590619,
      "grad_norm": 0.0,
      "learning_rate": 8.85608856088561e-06,
      "loss": 2.7723,
      "mean_token_accuracy": 0.7138583638583637,
      "mean_token_accuracy_utility": 0.6309340659340658,
      "step": 120
    },
    {
      "epoch": 0.04625774817281895,
      "grad_norm": 0.0,
      "learning_rate": 9.22509225092251e-06,
      "loss": 2.7693,
      "mean_token_accuracy": 0.7173137973137971,
      "mean_token_accuracy_utility": 0.6265445665445664,
      "step": 125
    },
    {
      "epoch": 0.048108058099731704,
      "grad_norm": 0.0,
      "learning_rate": 9.59409594095941e-06,
      "loss": 2.6987,
      "mean_token_accuracy": 0.7285836385836384,
      "mean_token_accuracy_utility": 0.6380939754012924,
      "step": 130
    },
    {
      "epoch": 0.04995836802664446,
      "grad_norm": 0.0,
      "learning_rate": 9.963099630996312e-06,
      "loss": 2.7419,
      "mean_token_accuracy": 0.7176739926739925,
      "mean_token_accuracy_utility": 0.6292368742368741,
      "step": 135
    },
    {
      "epoch": 0.051808677953557224,
      "grad_norm": 0.0,
      "learning_rate": 1.033210332103321e-05,
      "loss": 2.7083,
      "mean_token_accuracy": 0.7158608058608058,
      "mean_token_accuracy_utility": 0.6339560439560438,
      "step": 140
    },
    {
      "epoch": 0.05365898788046998,
      "grad_norm": 0.0,
      "learning_rate": 1.0701107011070112e-05,
      "loss": 2.6795,
      "mean_token_accuracy": 0.7214957264957264,
      "mean_token_accuracy_utility": 0.6423321123321122,
      "step": 145
    },
    {
      "epoch": 0.055509297807382736,
      "grad_norm": 0.0,
      "learning_rate": 1.1070110701107012e-05,
      "loss": 2.6858,
      "mean_token_accuracy": 0.7193833943833943,
      "mean_token_accuracy_utility": 0.6357081807081805,
      "step": 150
    },
    {
      "epoch": 0.05735960773429549,
      "grad_norm": 0.0,
      "learning_rate": 1.1439114391143913e-05,
      "loss": 2.6414,
      "mean_token_accuracy": 0.7136385836385836,
      "mean_token_accuracy_utility": 0.6473260073260072,
      "step": 155
    },
    {
      "epoch": 0.059209917661208256,
      "grad_norm": 0.0,
      "learning_rate": 1.1808118081180812e-05,
      "loss": 2.7362,
      "mean_token_accuracy": 0.718168498168498,
      "mean_token_accuracy_utility": 0.6304884004884004,
      "step": 160
    },
    {
      "epoch": 0.06106022758812101,
      "grad_norm": 0.0,
      "learning_rate": 1.2177121771217713e-05,
      "loss": 2.7251,
      "mean_token_accuracy": 0.7144200244200242,
      "mean_token_accuracy_utility": 0.6294505494505493,
      "step": 165
    },
    {
      "epoch": 0.06291053751503377,
      "grad_norm": 0.0,
      "learning_rate": 1.2546125461254614e-05,
      "loss": 2.6526,
      "mean_token_accuracy": 0.7165262515262514,
      "mean_token_accuracy_utility": 0.6451953601953601,
      "step": 170
    },
    {
      "epoch": 0.06476084744194653,
      "grad_norm": 0.0,
      "learning_rate": 1.2915129151291515e-05,
      "loss": 2.7255,
      "mean_token_accuracy": 0.7170024420024421,
      "mean_token_accuracy_utility": 0.6314957264957262,
      "step": 175
    },
    {
      "epoch": 0.06661115736885928,
      "grad_norm": 0.0,
      "learning_rate": 1.3284132841328414e-05,
      "loss": 2.6791,
      "mean_token_accuracy": 0.723180708180708,
      "mean_token_accuracy_utility": 0.6379365079365077,
      "step": 180
    },
    {
      "epoch": 0.06846146729577204,
      "grad_norm": 0.0,
      "learning_rate": 1.3653136531365315e-05,
      "loss": 2.798,
      "mean_token_accuracy": 0.7225152625152623,
      "mean_token_accuracy_utility": 0.6155494505494503,
      "step": 185
    },
    {
      "epoch": 0.0703117772226848,
      "grad_norm": 0.0,
      "learning_rate": 1.4022140221402215e-05,
      "loss": 2.6517,
      "mean_token_accuracy": 0.73006105006105,
      "mean_token_accuracy_utility": 0.6356593406593405,
      "step": 190
    },
    {
      "epoch": 0.07216208714959756,
      "grad_norm": 0.0,
      "learning_rate": 1.4391143911439116e-05,
      "loss": 2.735,
      "mean_token_accuracy": 0.716117216117216,
      "mean_token_accuracy_utility": 0.6200505497382951,
      "step": 195
    },
    {
      "epoch": 0.07401239707651032,
      "grad_norm": 0.0,
      "learning_rate": 1.4760147601476015e-05,
      "loss": 2.6683,
      "mean_token_accuracy": 0.717124542124542,
      "mean_token_accuracy_utility": 0.6358669108669106,
      "step": 200
    },
    {
      "epoch": 0.07401239707651032,
      "eval_mean_token_accuracy": 0.7118382706618,
      "eval_reasoning_loss": 1.0721101760864258,
      "eval_reasoning_runtime": 29.3234,
      "eval_reasoning_samples_per_second": 4.399,
      "eval_reasoning_steps_per_second": 1.125,
      "step": 200
    },
    {
      "epoch": 0.07401239707651032,
      "eval_mean_token_accuracy": 0.6395477302540498,
      "eval_utility_loss": 1.5844223499298096,
      "eval_utility_runtime": 11.079,
      "eval_utility_samples_per_second": 4.423,
      "eval_utility_steps_per_second": 1.173,
      "step": 200
    },
    {
      "epoch": 0.07586270700342307,
      "grad_norm": 0.0,
      "learning_rate": 1.5129151291512916e-05,
      "loss": 2.712,
      "mean_token_accuracy": 0.7222649572649571,
      "mean_token_accuracy_utility": 0.6263553113553113,
      "step": 205
    },
    {
      "epoch": 0.07771301693033583,
      "grad_norm": 0.0,
      "learning_rate": 1.5498154981549817e-05,
      "loss": 2.6031,
      "mean_token_accuracy": 0.7240415140415141,
      "mean_token_accuracy_utility": 0.6395054945054943,
      "step": 210
    },
    {
      "epoch": 0.0795633268572486,
      "grad_norm": 0.0,
      "learning_rate": 1.5867158671586716e-05,
      "loss": 2.625,
      "mean_token_accuracy": 0.7275885225885224,
      "mean_token_accuracy_utility": 0.6315590994371482,
      "step": 215
    },
    {
      "epoch": 0.08141363678416134,
      "grad_norm": 0.0,
      "learning_rate": 1.623616236162362e-05,
      "loss": 2.7541,
      "mean_token_accuracy": 0.7075396825396825,
      "mean_token_accuracy_utility": 0.6262743488199876,
      "step": 220
    },
    {
      "epoch": 0.08326394671107411,
      "grad_norm": 0.0,
      "learning_rate": 1.6605166051660518e-05,
      "loss": 2.6358,
      "mean_token_accuracy": 0.7203052503052502,
      "mean_token_accuracy_utility": 0.6376434676434675,
      "step": 225
    },
    {
      "epoch": 0.08511425663798686,
      "grad_norm": 0.0,
      "learning_rate": 1.697416974169742e-05,
      "loss": 2.581,
      "mean_token_accuracy": 0.7305799755799754,
      "mean_token_accuracy_utility": 0.6392063492063491,
      "step": 230
    },
    {
      "epoch": 0.08696456656489962,
      "grad_norm": 0.0,
      "learning_rate": 1.734317343173432e-05,
      "loss": 2.6712,
      "mean_token_accuracy": 0.726208791208791,
      "mean_token_accuracy_utility": 0.6316788766788766,
      "step": 235
    },
    {
      "epoch": 0.08881487649181238,
      "grad_norm": 0.0,
      "learning_rate": 1.771217712177122e-05,
      "loss": 2.4742,
      "mean_token_accuracy": 0.7291758241758244,
      "mean_token_accuracy_utility": 0.6592429792429791,
      "step": 240
    },
    {
      "epoch": 0.09066518641872513,
      "grad_norm": 0.0,
      "learning_rate": 1.8081180811808117e-05,
      "loss": 2.6969,
      "mean_token_accuracy": 0.7155738705738706,
      "mean_token_accuracy_utility": 0.6354273504273503,
      "step": 245
    },
    {
      "epoch": 0.0925154963456379,
      "grad_norm": 0.0,
      "learning_rate": 1.845018450184502e-05,
      "loss": 2.5998,
      "mean_token_accuracy": 0.7233272283272283,
      "mean_token_accuracy_utility": 0.6419352869352867,
      "step": 250
    },
    {
      "epoch": 0.09436580627255066,
      "grad_norm": 0.0,
      "learning_rate": 1.8819188191881922e-05,
      "loss": 2.592,
      "mean_token_accuracy": 0.7305006105006107,
      "mean_token_accuracy_utility": 0.6389307282376588,
      "step": 255
    },
    {
      "epoch": 0.09621611619946341,
      "grad_norm": 0.0,
      "learning_rate": 1.918819188191882e-05,
      "loss": 2.5373,
      "mean_token_accuracy": 0.7257570207570208,
      "mean_token_accuracy_utility": 0.652234432234432,
      "step": 260
    },
    {
      "epoch": 0.09806642612637617,
      "grad_norm": 0.0,
      "learning_rate": 1.955719557195572e-05,
      "loss": 2.5183,
      "mean_token_accuracy": 0.7246398046398044,
      "mean_token_accuracy_utility": 0.6485714285714285,
      "step": 265
    },
    {
      "epoch": 0.09991673605328892,
      "grad_norm": 0.0,
      "learning_rate": 1.9926199261992623e-05,
      "loss": 2.7204,
      "mean_token_accuracy": 0.7118498168498167,
      "mean_token_accuracy_utility": 0.62518315018315,
      "step": 270
    },
    {
      "epoch": 0.10176704598020168,
      "grad_norm": 0.0,
      "learning_rate": 1.9999866396188624e-05,
      "loss": 2.5512,
      "mean_token_accuracy": 0.7198046398046396,
      "mean_token_accuracy_utility": 0.6472161172161169,
      "step": 275
    },
    {
      "epoch": 0.10361735590711445,
      "grad_norm": 0.0,
      "learning_rate": 1.9999323636823398e-05,
      "loss": 2.6397,
      "mean_token_accuracy": 0.7303785103785103,
      "mean_token_accuracy_utility": 0.6312332112332111,
      "step": 280
    },
    {
      "epoch": 0.1054676658340272,
      "grad_norm": 0.0,
      "learning_rate": 1.9998363394309497e-05,
      "loss": 2.4707,
      "mean_token_accuracy": 0.7479609279609276,
      "mean_token_accuracy_utility": 0.6420024420024418,
      "step": 285
    },
    {
      "epoch": 0.10731797576093996,
      "grad_norm": 0.0,
      "learning_rate": 1.9996985708738146e-05,
      "loss": 2.3789,
      "mean_token_accuracy": 0.7481318681318683,
      "mean_token_accuracy_utility": 0.6549450549450547,
      "step": 290
    },
    {
      "epoch": 0.10916828568785271,
      "grad_norm": 0.0,
      "learning_rate": 1.999519063762928e-05,
      "loss": 2.5444,
      "mean_token_accuracy": 0.7266971916971914,
      "mean_token_accuracy_utility": 0.641997667820527,
      "step": 295
    },
    {
      "epoch": 0.11101859561476547,
      "grad_norm": 0.0,
      "learning_rate": 1.9992978255929168e-05,
      "loss": 2.4931,
      "mean_token_accuracy": 0.7213431013431014,
      "mean_token_accuracy_utility": 0.6534371184371182,
      "step": 300
    },
    {
      "epoch": 0.11101859561476547,
      "eval_mean_token_accuracy": 0.7202015849074673,
      "eval_reasoning_loss": 1.0194082260131836,
      "eval_reasoning_runtime": 29.7803,
      "eval_reasoning_samples_per_second": 4.332,
      "eval_reasoning_steps_per_second": 1.108,
      "step": 300
    },
    {
      "epoch": 0.11101859561476547,
      "eval_mean_token_accuracy": 0.6497433523169867,
      "eval_utility_loss": 1.4911507368087769,
      "eval_utility_runtime": 11.2672,
      "eval_utility_samples_per_second": 4.349,
      "eval_utility_steps_per_second": 1.154,
      "step": 300
    },
    {
      "epoch": 0.11286890554167824,
      "grad_norm": 0.0,
      "learning_rate": 1.999034865600726e-05,
      "loss": 2.5898,
      "mean_token_accuracy": 0.7127533577533577,
      "mean_token_accuracy_utility": 0.6475030525030523,
      "step": 305
    },
    {
      "epoch": 0.11471921546859098,
      "grad_norm": 0.0,
      "learning_rate": 1.9987301947652354e-05,
      "loss": 2.4652,
      "mean_token_accuracy": 0.7314896214896216,
      "mean_token_accuracy_utility": 0.6478571428571427,
      "step": 310
    },
    {
      "epoch": 0.11656952539550375,
      "grad_norm": 0.0,
      "learning_rate": 1.998383825806799e-05,
      "loss": 2.5311,
      "mean_token_accuracy": 0.7308485958485959,
      "mean_token_accuracy_utility": 0.6412393162393161,
      "step": 315
    },
    {
      "epoch": 0.11841983532241651,
      "grad_norm": 0.0,
      "learning_rate": 1.9979957731867143e-05,
      "loss": 2.5149,
      "mean_token_accuracy": 0.7372283272283271,
      "mean_token_accuracy_utility": 0.6378021978021976,
      "step": 320
    },
    {
      "epoch": 0.12027014524932926,
      "grad_norm": 0.0,
      "learning_rate": 1.9975660531066215e-05,
      "loss": 2.4625,
      "mean_token_accuracy": 0.7138644688644689,
      "mean_token_accuracy_utility": 0.6601037851037849,
      "step": 325
    },
    {
      "epoch": 0.12212045517624202,
      "grad_norm": 0.0,
      "learning_rate": 1.9970946835078227e-05,
      "loss": 2.4535,
      "mean_token_accuracy": 0.7352258852258851,
      "mean_token_accuracy_utility": 0.6446642246642245,
      "step": 330
    },
    {
      "epoch": 0.12397076510315477,
      "grad_norm": 0.0,
      "learning_rate": 1.9965816840705355e-05,
      "loss": 2.4571,
      "mean_token_accuracy": 0.7426251526251526,
      "mean_token_accuracy_utility": 0.6416788766788764,
      "step": 335
    },
    {
      "epoch": 0.12582107503006754,
      "grad_norm": 0.0,
      "learning_rate": 1.9960270762130705e-05,
      "loss": 2.5053,
      "mean_token_accuracy": 0.7308744285334272,
      "mean_token_accuracy_utility": 0.6397985347985345,
      "step": 340
    },
    {
      "epoch": 0.1276713849569803,
      "grad_norm": 0.0,
      "learning_rate": 1.9954308830909372e-05,
      "loss": 2.5254,
      "mean_token_accuracy": 0.7381684981684982,
      "mean_token_accuracy_utility": 0.6313797313797311,
      "step": 345
    },
    {
      "epoch": 0.12952169488389306,
      "grad_norm": 0.0,
      "learning_rate": 1.9947931295958778e-05,
      "loss": 2.5082,
      "mean_token_accuracy": 0.7352991452991452,
      "mean_token_accuracy_utility": 0.6354945054945053,
      "step": 350
    },
    {
      "epoch": 0.1313720048108058,
      "grad_norm": 0.0,
      "learning_rate": 1.9941138423548266e-05,
      "loss": 2.3778,
      "mean_token_accuracy": 0.735952380952381,
      "mean_token_accuracy_utility": 0.6572527472527471,
      "step": 355
    },
    {
      "epoch": 0.13322231473771856,
      "grad_norm": 0.0,
      "learning_rate": 1.9933930497287996e-05,
      "loss": 2.4797,
      "mean_token_accuracy": 0.7355128205128203,
      "mean_token_accuracy_utility": 0.6426923076923077,
      "step": 360
    },
    {
      "epoch": 0.13507262466463132,
      "grad_norm": 0.0,
      "learning_rate": 1.9926307818117098e-05,
      "loss": 2.4381,
      "mean_token_accuracy": 0.7350854700854699,
      "mean_token_accuracy_utility": 0.6536996336996335,
      "step": 365
    },
    {
      "epoch": 0.1369229345915441,
      "grad_norm": 0.0,
      "learning_rate": 1.9918270704291104e-05,
      "loss": 2.4775,
      "mean_token_accuracy": 0.7395299145299145,
      "mean_token_accuracy_utility": 0.6429914529914529,
      "step": 370
    },
    {
      "epoch": 0.13877324451845685,
      "grad_norm": 0.0,
      "learning_rate": 1.9909819491368677e-05,
      "loss": 2.4364,
      "mean_token_accuracy": 0.7580586080586079,
      "mean_token_accuracy_utility": 0.6366422466422464,
      "step": 375
    },
    {
      "epoch": 0.1406235544453696,
      "grad_norm": 0.0,
      "learning_rate": 1.990095453219757e-05,
      "loss": 2.2875,
      "mean_token_accuracy": 0.7413858363858362,
      "mean_token_accuracy_utility": 0.6729548229548228,
      "step": 380
    },
    {
      "epoch": 0.14247386437228235,
      "grad_norm": 0.0,
      "learning_rate": 1.989167619689993e-05,
      "loss": 2.4546,
      "mean_token_accuracy": 0.7326495726495724,
      "mean_token_accuracy_utility": 0.6500400088617265,
      "step": 385
    },
    {
      "epoch": 0.1443241742991951,
      "grad_norm": 0.0,
      "learning_rate": 1.988198487285682e-05,
      "loss": 2.4514,
      "mean_token_accuracy": 0.72028083028083,
      "mean_token_accuracy_utility": 0.6646214896214894,
      "step": 390
    },
    {
      "epoch": 0.14617448422610788,
      "grad_norm": 0.0,
      "learning_rate": 1.9871880964692055e-05,
      "loss": 2.4245,
      "mean_token_accuracy": 0.7340293040293039,
      "mean_token_accuracy_utility": 0.6563736263736263,
      "step": 395
    },
    {
      "epoch": 0.14802479415302064,
      "grad_norm": 0.0,
      "learning_rate": 1.9861364894255306e-05,
      "loss": 2.4566,
      "mean_token_accuracy": 0.7415018315018316,
      "mean_token_accuracy_utility": 0.6465689865689864,
      "step": 400
    },
    {
      "epoch": 0.14802479415302064,
      "eval_mean_token_accuracy": 0.7259035082564496,
      "eval_reasoning_loss": 0.9811830520629883,
      "eval_reasoning_runtime": 29.7136,
      "eval_reasoning_samples_per_second": 4.341,
      "eval_reasoning_steps_per_second": 1.111,
      "step": 400
    },
    {
      "epoch": 0.14802479415302064,
      "eval_mean_token_accuracy": 0.6556965625166941,
      "eval_utility_loss": 1.466097116470337,
      "eval_utility_runtime": 11.2368,
      "eval_utility_samples_per_second": 4.361,
      "eval_utility_steps_per_second": 1.157,
      "step": 400
    },
    {
      "epoch": 0.1498751040799334,
      "grad_norm": 0.0,
      "learning_rate": 1.985043710060449e-05,
      "loss": 2.3444,
      "mean_token_accuracy": 0.7297374847374847,
      "mean_token_accuracy_utility": 0.6698778998778996,
      "step": 405
    },
    {
      "epoch": 0.15172541400684614,
      "grad_norm": 0.0,
      "learning_rate": 1.9839098039987435e-05,
      "loss": 2.3982,
      "mean_token_accuracy": 0.7448107448107449,
      "mean_token_accuracy_utility": 0.6561111111111109,
      "step": 410
    },
    {
      "epoch": 0.1535757239337589,
      "grad_norm": 0.0,
      "learning_rate": 1.9827348185822834e-05,
      "loss": 2.4641,
      "mean_token_accuracy": 0.7384432234432233,
      "mean_token_accuracy_utility": 0.6460195360195358,
      "step": 415
    },
    {
      "epoch": 0.15542603386067166,
      "grad_norm": 0.0,
      "learning_rate": 1.981518802868048e-05,
      "loss": 2.3537,
      "mean_token_accuracy": 0.7339926739926739,
      "mean_token_accuracy_utility": 0.6724358974358973,
      "step": 420
    },
    {
      "epoch": 0.15727634378758443,
      "grad_norm": 0.0,
      "learning_rate": 1.9802618076260784e-05,
      "loss": 2.383,
      "mean_token_accuracy": 0.7468437118437118,
      "mean_token_accuracy_utility": 0.6575763125763123,
      "step": 425
    },
    {
      "epoch": 0.1591266537144972,
      "grad_norm": 0.0,
      "learning_rate": 1.9789638853373563e-05,
      "loss": 2.2901,
      "mean_token_accuracy": 0.7424603174603175,
      "mean_token_accuracy_utility": 0.6722710622710621,
      "step": 430
    },
    {
      "epoch": 0.16097696364140993,
      "grad_norm": 0.0,
      "learning_rate": 1.9776250901916168e-05,
      "loss": 2.4486,
      "mean_token_accuracy": 0.7194444444444442,
      "mean_token_accuracy_utility": 0.6608119658119656,
      "step": 435
    },
    {
      "epoch": 0.1628272735683227,
      "grad_norm": 0.0,
      "learning_rate": 1.9762454780850807e-05,
      "loss": 2.384,
      "mean_token_accuracy": 0.7404212454212453,
      "mean_token_accuracy_utility": 0.6573015873015872,
      "step": 440
    },
    {
      "epoch": 0.16467758349523545,
      "grad_norm": 0.0,
      "learning_rate": 1.9748251066181247e-05,
      "loss": 2.4278,
      "mean_token_accuracy": 0.7453113553113553,
      "mean_token_accuracy_utility": 0.646214896214896,
      "step": 445
    },
    {
      "epoch": 0.16652789342214822,
      "grad_norm": 0.0,
      "learning_rate": 1.973364035092875e-05,
      "loss": 2.521,
      "mean_token_accuracy": 0.7255555555555552,
      "mean_token_accuracy_utility": 0.6423504273504271,
      "step": 450
    },
    {
      "epoch": 0.16837820334906098,
      "grad_norm": 0.0,
      "learning_rate": 1.971862324510732e-05,
      "loss": 2.4679,
      "mean_token_accuracy": 0.7219291819291819,
      "mean_token_accuracy_utility": 0.6589377289377288,
      "step": 455
    },
    {
      "epoch": 0.17022851327597371,
      "grad_norm": 0.0,
      "learning_rate": 1.9703200375698223e-05,
      "loss": 2.4809,
      "mean_token_accuracy": 0.7290231990231991,
      "mean_token_accuracy_utility": 0.6474664224664223,
      "step": 460
    },
    {
      "epoch": 0.17207882320288648,
      "grad_norm": 0.0,
      "learning_rate": 1.968737238662382e-05,
      "loss": 2.2974,
      "mean_token_accuracy": 0.7510989010989013,
      "mean_token_accuracy_utility": 0.667791466321707,
      "step": 465
    },
    {
      "epoch": 0.17392913312979924,
      "grad_norm": 0.0,
      "learning_rate": 1.9671139938720678e-05,
      "loss": 2.3971,
      "mean_token_accuracy": 0.7455189255189255,
      "mean_token_accuracy_utility": 0.6506166056166054,
      "step": 470
    },
    {
      "epoch": 0.175779443056712,
      "grad_norm": 0.0,
      "learning_rate": 1.9654503709711984e-05,
      "loss": 2.4337,
      "mean_token_accuracy": 0.733217338217338,
      "mean_token_accuracy_utility": 0.65525641025641,
      "step": 475
    },
    {
      "epoch": 0.17762975298362477,
      "grad_norm": 0.0,
      "learning_rate": 1.963746439417924e-05,
      "loss": 2.4329,
      "mean_token_accuracy": 0.7493223443223442,
      "mean_token_accuracy_utility": 0.6497130647130646,
      "step": 480
    },
    {
      "epoch": 0.1794800629105375,
      "grad_norm": 0.0,
      "learning_rate": 1.962002270353328e-05,
      "loss": 2.3609,
      "mean_token_accuracy": 0.7506532356532357,
      "mean_token_accuracy_utility": 0.6516544566544565,
      "step": 485
    },
    {
      "epoch": 0.18133037283745027,
      "grad_norm": 0.0,
      "learning_rate": 1.960217936598454e-05,
      "loss": 2.2523,
      "mean_token_accuracy": 0.7516361416361417,
      "mean_token_accuracy_utility": 0.6669963369963368,
      "step": 490
    },
    {
      "epoch": 0.18318068276436303,
      "grad_norm": 0.0,
      "learning_rate": 1.958393512651269e-05,
      "loss": 2.3433,
      "mean_token_accuracy": 0.7603601953601956,
      "mean_token_accuracy_utility": 0.6469027740574534,
      "step": 495
    },
    {
      "epoch": 0.1850309926912758,
      "grad_norm": 0.0,
      "learning_rate": 1.956529074683551e-05,
      "loss": 2.4142,
      "mean_token_accuracy": 0.738144078144078,
      "mean_token_accuracy_utility": 0.6521672771672768,
      "step": 500
    },
    {
      "epoch": 0.1850309926912758,
      "eval_mean_token_accuracy": 0.7298664298664297,
      "eval_reasoning_loss": 0.9576250314712524,
      "eval_reasoning_runtime": 29.291,
      "eval_reasoning_samples_per_second": 4.404,
      "eval_reasoning_steps_per_second": 1.127,
      "step": 500
    },
    {
      "epoch": 0.1850309926912758,
      "eval_mean_token_accuracy": 0.6577473871726089,
      "eval_utility_loss": 1.4563536643981934,
      "eval_utility_runtime": 11.052,
      "eval_utility_samples_per_second": 4.434,
      "eval_utility_steps_per_second": 1.176,
      "step": 500
    },
    {
      "epoch": 0.18688130261818856,
      "grad_norm": 0.0,
      "learning_rate": 1.9546247005377065e-05,
      "loss": 2.2989,
      "mean_token_accuracy": 0.7442368742368742,
      "mean_token_accuracy_utility": 0.6685470085470084,
      "step": 505
    },
    {
      "epoch": 0.18873161254510132,
      "grad_norm": 0.0,
      "learning_rate": 1.952680469723526e-05,
      "loss": 2.3876,
      "mean_token_accuracy": 0.7468376068376067,
      "mean_token_accuracy_utility": 0.6526068376068375,
      "step": 510
    },
    {
      "epoch": 0.19058192247201405,
      "grad_norm": 0.0,
      "learning_rate": 1.9506964634148597e-05,
      "loss": 2.4415,
      "mean_token_accuracy": 0.7388766788766787,
      "mean_token_accuracy_utility": 0.6493467643467642,
      "step": 515
    },
    {
      "epoch": 0.19243223239892682,
      "grad_norm": 0.0,
      "learning_rate": 1.9486727644462306e-05,
      "loss": 2.3878,
      "mean_token_accuracy": 0.7401526251526253,
      "mean_token_accuracy_utility": 0.6608363858363856,
      "step": 520
    },
    {
      "epoch": 0.19428254232583958,
      "grad_norm": 0.0,
      "learning_rate": 1.9466094573093744e-05,
      "loss": 2.3677,
      "mean_token_accuracy": 0.7586202686202685,
      "mean_token_accuracy_utility": 0.648119658119658,
      "step": 525
    },
    {
      "epoch": 0.19613285225275234,
      "grad_norm": 0.0,
      "learning_rate": 1.9445066281497144e-05,
      "loss": 2.3177,
      "mean_token_accuracy": 0.7513858363858362,
      "mean_token_accuracy_utility": 0.6591330891330889,
      "step": 530
    },
    {
      "epoch": 0.1979831621796651,
      "grad_norm": 0.0,
      "learning_rate": 1.9423643647627625e-05,
      "loss": 2.3999,
      "mean_token_accuracy": 0.7380402930402931,
      "mean_token_accuracy_utility": 0.6579739626215866,
      "step": 535
    },
    {
      "epoch": 0.19983347210657784,
      "grad_norm": 0.0,
      "learning_rate": 1.940182756590454e-05,
      "loss": 2.2979,
      "mean_token_accuracy": 0.7426617826617827,
      "mean_token_accuracy_utility": 0.6688034188034186,
      "step": 540
    },
    {
      "epoch": 0.2016837820334906,
      "grad_norm": 0.0,
      "learning_rate": 1.9379618947174155e-05,
      "loss": 2.3745,
      "mean_token_accuracy": 0.747106227106227,
      "mean_token_accuracy_utility": 0.6605433455433454,
      "step": 545
    },
    {
      "epoch": 0.20353409196040337,
      "grad_norm": 0.0,
      "learning_rate": 1.935701871867158e-05,
      "loss": 2.336,
      "mean_token_accuracy": 0.7463797313797313,
      "mean_token_accuracy_utility": 0.6600122100122099,
      "step": 550
    },
    {
      "epoch": 0.20538440188731613,
      "grad_norm": 0.0,
      "learning_rate": 1.9334027823982103e-05,
      "loss": 2.3233,
      "mean_token_accuracy": 0.746164043164043,
      "mean_token_accuracy_utility": 0.6621794871794869,
      "step": 555
    },
    {
      "epoch": 0.2072347118142289,
      "grad_norm": 0.0,
      "learning_rate": 1.9310647223001752e-05,
      "loss": 2.3795,
      "mean_token_accuracy": 0.7306593406593405,
      "mean_token_accuracy_utility": 0.6624725274725273,
      "step": 560
    },
    {
      "epoch": 0.20908502174114163,
      "grad_norm": 0.0,
      "learning_rate": 1.9286877891897244e-05,
      "loss": 2.364,
      "mean_token_accuracy": 0.759065934065934,
      "mean_token_accuracy_utility": 0.6490415140415139,
      "step": 565
    },
    {
      "epoch": 0.2109353316680544,
      "grad_norm": 0.0,
      "learning_rate": 1.9262720823065217e-05,
      "loss": 2.4158,
      "mean_token_accuracy": 0.7296336996336995,
      "mean_token_accuracy_utility": 0.6575396825396824,
      "step": 570
    },
    {
      "epoch": 0.21278564159496716,
      "grad_norm": 0.0,
      "learning_rate": 1.923817702509081e-05,
      "loss": 2.3535,
      "mean_token_accuracy": 0.7493894993894996,
      "mean_token_accuracy_utility": 0.6538705738705737,
      "step": 575
    },
    {
      "epoch": 0.21463595152187992,
      "grad_norm": 0.0,
      "learning_rate": 1.9213247522705532e-05,
      "loss": 2.3299,
      "mean_token_accuracy": 0.7401037851037852,
      "mean_token_accuracy_utility": 0.668205128205128,
      "step": 580
    },
    {
      "epoch": 0.21648626144879268,
      "grad_norm": 0.0,
      "learning_rate": 1.9187933356744504e-05,
      "loss": 2.3497,
      "mean_token_accuracy": 0.7488583638583638,
      "mean_token_accuracy_utility": 0.6595115995115993,
      "step": 585
    },
    {
      "epoch": 0.21833657137570542,
      "grad_norm": 0.0,
      "learning_rate": 1.9162235584102973e-05,
      "loss": 2.3221,
      "mean_token_accuracy": 0.7435470085470085,
      "mean_token_accuracy_utility": 0.6603235653235652,
      "step": 590
    },
    {
      "epoch": 0.22018688130261818,
      "grad_norm": 0.0,
      "learning_rate": 1.9136155277692215e-05,
      "loss": 2.3133,
      "mean_token_accuracy": 0.7485164835164835,
      "mean_token_accuracy_utility": 0.6579059829059827,
      "step": 595
    },
    {
      "epoch": 0.22203719122953094,
      "grad_norm": 0.0,
      "learning_rate": 1.9109693526394722e-05,
      "loss": 2.3637,
      "mean_token_accuracy": 0.7452075702075702,
      "mean_token_accuracy_utility": 0.6514346764346763,
      "step": 600
    },
    {
      "epoch": 0.22203719122953094,
      "eval_mean_token_accuracy": 0.7326640679581854,
      "eval_reasoning_loss": 0.9425094723701477,
      "eval_reasoning_runtime": 29.3011,
      "eval_reasoning_samples_per_second": 4.403,
      "eval_reasoning_steps_per_second": 1.126,
      "step": 600
    },
    {
      "epoch": 0.22203719122953094,
      "eval_mean_token_accuracy": 0.6587021039837739,
      "eval_utility_loss": 1.4486795663833618,
      "eval_utility_runtime": 11.0803,
      "eval_utility_samples_per_second": 4.422,
      "eval_utility_steps_per_second": 1.173,
      "step": 600
    },
    {
      "epoch": 0.2238875011564437,
      "grad_norm": 0.0,
      "learning_rate": 1.9082851435018743e-05,
      "loss": 2.2923,
      "mean_token_accuracy": 0.7416666666666665,
      "mean_token_accuracy_utility": 0.6719474969474967,
      "step": 605
    },
    {
      "epoch": 0.22573781108335647,
      "grad_norm": 0.0,
      "learning_rate": 1.905563012425216e-05,
      "loss": 2.4276,
      "mean_token_accuracy": 0.7503479853479855,
      "mean_token_accuracy_utility": 0.6413431013431011,
      "step": 610
    },
    {
      "epoch": 0.2275881210102692,
      "grad_norm": 0.0,
      "learning_rate": 1.9028030730615696e-05,
      "loss": 2.4726,
      "mean_token_accuracy": 0.7388888888888889,
      "mean_token_accuracy_utility": 0.6433333333333332,
      "step": 615
    },
    {
      "epoch": 0.22943843093718197,
      "grad_norm": 0.0,
      "learning_rate": 1.9000054406415467e-05,
      "loss": 2.3907,
      "mean_token_accuracy": 0.7496886446886447,
      "mean_token_accuracy_utility": 0.6504884004884003,
      "step": 620
    },
    {
      "epoch": 0.23128874086409473,
      "grad_norm": 0.0,
      "learning_rate": 1.897170231969486e-05,
      "loss": 2.3651,
      "mean_token_accuracy": 0.7432967032967033,
      "mean_token_accuracy_utility": 0.6597619047619047,
      "step": 625
    },
    {
      "epoch": 0.2331390507910075,
      "grad_norm": 0.0,
      "learning_rate": 1.8942975654185788e-05,
      "loss": 2.3086,
      "mean_token_accuracy": 0.7489743589743588,
      "mean_token_accuracy_utility": 0.6643650793650792,
      "step": 630
    },
    {
      "epoch": 0.23498936071792026,
      "grad_norm": 0.0,
      "learning_rate": 1.8913875609259246e-05,
      "loss": 2.2998,
      "mean_token_accuracy": 0.7495970695970695,
      "mean_token_accuracy_utility": 0.6613858363858363,
      "step": 635
    },
    {
      "epoch": 0.23683967064483302,
      "grad_norm": 0.0,
      "learning_rate": 1.8884403399875252e-05,
      "loss": 2.3497,
      "mean_token_accuracy": 0.7414957264957265,
      "mean_token_accuracy_utility": 0.6653227328227327,
      "step": 640
    },
    {
      "epoch": 0.23868998057174576,
      "grad_norm": 0.0,
      "learning_rate": 1.8854560256532098e-05,
      "loss": 2.2372,
      "mean_token_accuracy": 0.7614540903540903,
      "mean_token_accuracy_utility": 0.6674542124542122,
      "step": 645
    },
    {
      "epoch": 0.24054029049865852,
      "grad_norm": 0.0,
      "learning_rate": 1.8824347425215016e-05,
      "loss": 2.3402,
      "mean_token_accuracy": 0.7337179487179487,
      "mean_token_accuracy_utility": 0.6660500610500608,
      "step": 650
    },
    {
      "epoch": 0.24239060042557128,
      "grad_norm": 0.0,
      "learning_rate": 1.8793766167344115e-05,
      "loss": 2.2777,
      "mean_token_accuracy": 0.7496642246642249,
      "mean_token_accuracy_utility": 0.6644871794871794,
      "step": 655
    },
    {
      "epoch": 0.24424091035248405,
      "grad_norm": 0.0,
      "learning_rate": 1.8762817759721735e-05,
      "loss": 2.4199,
      "mean_token_accuracy": 0.7471794871794872,
      "mean_token_accuracy_utility": 0.6443435320198275,
      "step": 660
    },
    {
      "epoch": 0.2460912202793968,
      "grad_norm": 0.0,
      "learning_rate": 1.8731503494479132e-05,
      "loss": 2.3945,
      "mean_token_accuracy": 0.7403357753357753,
      "mean_token_accuracy_utility": 0.6517582417582416,
      "step": 665
    },
    {
      "epoch": 0.24794153020630955,
      "grad_norm": 0.0,
      "learning_rate": 1.869982467902255e-05,
      "loss": 2.3843,
      "mean_token_accuracy": 0.7523199023199025,
      "mean_token_accuracy_utility": 0.6475763125763125,
      "step": 670
    },
    {
      "epoch": 0.2497918401332223,
      "grad_norm": 0.0,
      "learning_rate": 1.8667782635978597e-05,
      "loss": 2.4305,
      "mean_token_accuracy": 0.7348778998778999,
      "mean_token_accuracy_utility": 0.6487188795409086,
      "step": 675
    },
    {
      "epoch": 0.2516421500601351,
      "grad_norm": 0.0,
      "learning_rate": 1.8635378703139066e-05,
      "loss": 2.2509,
      "mean_token_accuracy": 0.7578083028083029,
      "mean_token_accuracy_utility": 0.6640842490842489,
      "step": 680
    },
    {
      "epoch": 0.25349245998704784,
      "grad_norm": 0.0,
      "learning_rate": 1.8602614233405047e-05,
      "loss": 2.3438,
      "mean_token_accuracy": 0.7485592185592187,
      "mean_token_accuracy_utility": 0.6549340067852177,
      "step": 685
    },
    {
      "epoch": 0.2553427699139606,
      "grad_norm": 0.0,
      "learning_rate": 1.8569490594730474e-05,
      "loss": 2.3325,
      "mean_token_accuracy": 0.7449694749694751,
      "mean_token_accuracy_utility": 0.6651465201465199,
      "step": 690
    },
    {
      "epoch": 0.25719307984087336,
      "grad_norm": 0.0,
      "learning_rate": 1.853600917006497e-05,
      "loss": 2.2368,
      "mean_token_accuracy": 0.7595299145299147,
      "mean_token_accuracy_utility": 0.669157509157509,
      "step": 695
    },
    {
      "epoch": 0.2590433897677861,
      "grad_norm": 0.0,
      "learning_rate": 1.8502171357296144e-05,
      "loss": 2.2649,
      "mean_token_accuracy": 0.7595665445665445,
      "mean_token_accuracy_utility": 0.6639255189255187,
      "step": 700
    },
    {
      "epoch": 0.2590433897677861,
      "eval_mean_token_accuracy": 0.7344359344359348,
      "eval_reasoning_loss": 0.9315915107727051,
      "eval_reasoning_runtime": 29.3427,
      "eval_reasoning_samples_per_second": 4.396,
      "eval_reasoning_steps_per_second": 1.125,
      "step": 700
    },
    {
      "epoch": 0.2590433897677861,
      "eval_mean_token_accuracy": 0.659615951995134,
      "eval_utility_loss": 1.4435120820999146,
      "eval_utility_runtime": 11.0912,
      "eval_utility_samples_per_second": 4.418,
      "eval_utility_steps_per_second": 1.172,
      "step": 700
    },
    {
      "epoch": 0.2608936996946989,
      "grad_norm": 0.0,
      "learning_rate": 1.8467978569191216e-05,
      "loss": 2.3591,
      "mean_token_accuracy": 0.7455006105006102,
      "mean_token_accuracy_utility": 0.6557142857142855,
      "step": 705
    },
    {
      "epoch": 0.2627440096216116,
      "grad_norm": 0.0,
      "learning_rate": 1.8433432233338027e-05,
      "loss": 2.2721,
      "mean_token_accuracy": 0.7560622710622711,
      "mean_token_accuracy_utility": 0.6684310134310132,
      "step": 710
    },
    {
      "epoch": 0.26459431954852436,
      "grad_norm": 0.0,
      "learning_rate": 1.8398533792085436e-05,
      "loss": 2.3266,
      "mean_token_accuracy": 0.7468009768009767,
      "mean_token_accuracy_utility": 0.6601938507774903,
      "step": 715
    },
    {
      "epoch": 0.2664446294754371,
      "grad_norm": 0.0,
      "learning_rate": 1.8363284702483106e-05,
      "loss": 2.3998,
      "mean_token_accuracy": 0.7408485958485957,
      "mean_token_accuracy_utility": 0.6497191697191695,
      "step": 720
    },
    {
      "epoch": 0.2682949394023499,
      "grad_norm": 0.0,
      "learning_rate": 1.832768643622067e-05,
      "loss": 2.3464,
      "mean_token_accuracy": 0.7553418803418805,
      "mean_token_accuracy_utility": 0.6499328449328448,
      "step": 725
    },
    {
      "epoch": 0.27014524932926265,
      "grad_norm": 0.0,
      "learning_rate": 1.8291740479566286e-05,
      "loss": 2.2445,
      "mean_token_accuracy": 0.7537240537240536,
      "mean_token_accuracy_utility": 0.6750366300366298,
      "step": 730
    },
    {
      "epoch": 0.2719955592561754,
      "grad_norm": 0.0,
      "learning_rate": 1.825544833330457e-05,
      "loss": 2.2673,
      "mean_token_accuracy": 0.751013431013431,
      "mean_token_accuracy_utility": 0.667185592185592,
      "step": 735
    },
    {
      "epoch": 0.2738458691830882,
      "grad_norm": 0.0,
      "learning_rate": 1.8218811512673958e-05,
      "loss": 2.3527,
      "mean_token_accuracy": 0.7368742368742369,
      "mean_token_accuracy_utility": 0.6660378510378508,
      "step": 740
    },
    {
      "epoch": 0.27569617911000094,
      "grad_norm": 0.0,
      "learning_rate": 1.818183154730344e-05,
      "loss": 2.3359,
      "mean_token_accuracy": 0.7471978021978024,
      "mean_token_accuracy_utility": 0.6576923076923075,
      "step": 745
    },
    {
      "epoch": 0.2775464890369137,
      "grad_norm": 0.0,
      "learning_rate": 1.8144509981148675e-05,
      "loss": 2.3245,
      "mean_token_accuracy": 0.7504090354090355,
      "mean_token_accuracy_utility": 0.6570146520146518,
      "step": 750
    },
    {
      "epoch": 0.27939679896382646,
      "grad_norm": 0.0,
      "learning_rate": 1.810684837242755e-05,
      "loss": 2.4288,
      "mean_token_accuracy": 0.7450305250305249,
      "mean_token_accuracy_utility": 0.6404305577226468,
      "step": 755
    },
    {
      "epoch": 0.2812471088907392,
      "grad_norm": 0.0,
      "learning_rate": 1.8068848293555118e-05,
      "loss": 2.3205,
      "mean_token_accuracy": 0.7442612942612942,
      "mean_token_accuracy_utility": 0.6648534798534796,
      "step": 760
    },
    {
      "epoch": 0.28309741881765194,
      "grad_norm": 0.0,
      "learning_rate": 1.8030511331077945e-05,
      "loss": 2.2842,
      "mean_token_accuracy": 0.7485286935286938,
      "mean_token_accuracy_utility": 0.669072039072039,
      "step": 765
    },
    {
      "epoch": 0.2849477287445647,
      "grad_norm": 0.0,
      "learning_rate": 1.799183908560787e-05,
      "loss": 2.291,
      "mean_token_accuracy": 0.7368131868131866,
      "mean_token_accuracy_utility": 0.6756837606837605,
      "step": 770
    },
    {
      "epoch": 0.28679803867147746,
      "grad_norm": 0.0,
      "learning_rate": 1.795283317175518e-05,
      "loss": 2.2931,
      "mean_token_accuracy": 0.7562393162393163,
      "mean_token_accuracy_utility": 0.6596214896214895,
      "step": 775
    },
    {
      "epoch": 0.2886483485983902,
      "grad_norm": 0.0,
      "learning_rate": 1.7913495218061202e-05,
      "loss": 2.3788,
      "mean_token_accuracy": 0.7475213675213673,
      "mean_token_accuracy_utility": 0.6528449328449326,
      "step": 780
    },
    {
      "epoch": 0.290498658525303,
      "grad_norm": 0.0,
      "learning_rate": 1.787382686693029e-05,
      "loss": 2.3671,
      "mean_token_accuracy": 0.7524603174603176,
      "mean_token_accuracy_utility": 0.6515018315018313,
      "step": 785
    },
    {
      "epoch": 0.29234896845221575,
      "grad_norm": 0.0,
      "learning_rate": 1.783382977456128e-05,
      "loss": 2.2315,
      "mean_token_accuracy": 0.7545421245421247,
      "mean_token_accuracy_utility": 0.6718070818070816,
      "step": 790
    },
    {
      "epoch": 0.2941992783791285,
      "grad_norm": 0.0,
      "learning_rate": 1.779350561087833e-05,
      "loss": 2.3745,
      "mean_token_accuracy": 0.7402319902319903,
      "mean_token_accuracy_utility": 0.6605799755799754,
      "step": 795
    },
    {
      "epoch": 0.2960495883060413,
      "grad_norm": 0.0,
      "learning_rate": 1.775285605946119e-05,
      "loss": 2.2451,
      "mean_token_accuracy": 0.7574542124542125,
      "mean_token_accuracy_utility": 0.6693467643467642,
      "step": 800
    },
    {
      "epoch": 0.2960495883060413,
      "eval_mean_token_accuracy": 0.7357364857364858,
      "eval_reasoning_loss": 0.9232837557792664,
      "eval_reasoning_runtime": 28.9896,
      "eval_reasoning_samples_per_second": 4.45,
      "eval_reasoning_steps_per_second": 1.138,
      "step": 800
    },
    {
      "epoch": 0.2960495883060413,
      "eval_mean_token_accuracy": 0.6598667513022676,
      "eval_utility_loss": 1.439689040184021,
      "eval_utility_runtime": 10.8148,
      "eval_utility_samples_per_second": 4.531,
      "eval_utility_steps_per_second": 1.202,
      "step": 800
    },
    {
      "epoch": 0.29789989823295404,
      "grad_norm": 0.0,
      "learning_rate": 1.7711882817474922e-05,
      "loss": 2.2976,
      "mean_token_accuracy": 0.7547130647130647,
      "mean_token_accuracy_utility": 0.6604639804639802,
      "step": 805
    },
    {
      "epoch": 0.2997502081598668,
      "grad_norm": 0.0,
      "learning_rate": 1.7670587595599034e-05,
      "loss": 2.2926,
      "mean_token_accuracy": 0.7524969474969474,
      "mean_token_accuracy_utility": 0.6696520146520144,
      "step": 810
    },
    {
      "epoch": 0.3016005180867795,
      "grad_norm": 0.0,
      "learning_rate": 1.762897211795607e-05,
      "loss": 2.3678,
      "mean_token_accuracy": 0.7394078144078142,
      "mean_token_accuracy_utility": 0.6532539682539682,
      "step": 815
    },
    {
      "epoch": 0.3034508280136923,
      "grad_norm": 0.0,
      "learning_rate": 1.758703812203961e-05,
      "loss": 2.3798,
      "mean_token_accuracy": 0.7384126984126984,
      "mean_token_accuracy_utility": 0.6567582417582416,
      "step": 820
    },
    {
      "epoch": 0.30530113794060504,
      "grad_norm": 0.0,
      "learning_rate": 1.7544787358641735e-05,
      "loss": 2.422,
      "mean_token_accuracy": 0.7281074481074482,
      "mean_token_accuracy_utility": 0.657155067155067,
      "step": 825
    },
    {
      "epoch": 0.3071514478675178,
      "grad_norm": 0.0,
      "learning_rate": 1.7502221591779932e-05,
      "loss": 2.3474,
      "mean_token_accuracy": 0.7287301587301588,
      "mean_token_accuracy_utility": 0.6677409401378704,
      "step": 830
    },
    {
      "epoch": 0.30900175779443056,
      "grad_norm": 0.0,
      "learning_rate": 1.7459342598623438e-05,
      "loss": 2.3083,
      "mean_token_accuracy": 0.7496275946275948,
      "mean_token_accuracy_utility": 0.6668925518925517,
      "step": 835
    },
    {
      "epoch": 0.31085206772134333,
      "grad_norm": 0.0,
      "learning_rate": 1.741615216941905e-05,
      "loss": 2.2485,
      "mean_token_accuracy": 0.760964590964591,
      "mean_token_accuracy_utility": 0.6673137973137971,
      "step": 840
    },
    {
      "epoch": 0.3127023776482561,
      "grad_norm": 0.0,
      "learning_rate": 1.7372652107416364e-05,
      "loss": 2.3729,
      "mean_token_accuracy": 0.7434554334554335,
      "mean_token_accuracy_utility": 0.6564224664224663,
      "step": 845
    },
    {
      "epoch": 0.31455268757516885,
      "grad_norm": 0.0,
      "learning_rate": 1.7328844228792513e-05,
      "loss": 2.2667,
      "mean_token_accuracy": 0.7686507936507938,
      "mean_token_accuracy_utility": 0.6587728937728936,
      "step": 850
    },
    {
      "epoch": 0.3164029975020816,
      "grad_norm": 0.0,
      "learning_rate": 1.7284730362576308e-05,
      "loss": 2.3423,
      "mean_token_accuracy": 0.7462759462759463,
      "mean_token_accuracy_utility": 0.6580219780219779,
      "step": 855
    },
    {
      "epoch": 0.3182533074289944,
      "grad_norm": 0.0,
      "learning_rate": 1.7240312350571905e-05,
      "loss": 2.4314,
      "mean_token_accuracy": 0.7527106227106227,
      "mean_token_accuracy_utility": 0.638205128205128,
      "step": 860
    },
    {
      "epoch": 0.3201036173559071,
      "grad_norm": 0.0,
      "learning_rate": 1.719559204728188e-05,
      "loss": 2.3007,
      "mean_token_accuracy": 0.7610744810744811,
      "mean_token_accuracy_utility": 0.6540415140415139,
      "step": 865
    },
    {
      "epoch": 0.32195392728281985,
      "grad_norm": 0.0,
      "learning_rate": 1.715057131982983e-05,
      "loss": 2.3881,
      "mean_token_accuracy": 0.7513980463980463,
      "mean_token_accuracy_utility": 0.6454822954822953,
      "step": 870
    },
    {
      "epoch": 0.3238042372097326,
      "grad_norm": 0.0,
      "learning_rate": 1.710525204788239e-05,
      "loss": 2.2542,
      "mean_token_accuracy": 0.7570940170940172,
      "mean_token_accuracy_utility": 0.6658342426248013,
      "step": 875
    },
    {
      "epoch": 0.3256545471366454,
      "grad_norm": 0.0,
      "learning_rate": 1.7059636123570767e-05,
      "loss": 2.349,
      "mean_token_accuracy": 0.7526495726495728,
      "mean_token_accuracy_utility": 0.651257631257631,
      "step": 880
    },
    {
      "epoch": 0.32750485706355814,
      "grad_norm": 0.0,
      "learning_rate": 1.7013725451411757e-05,
      "loss": 2.2791,
      "mean_token_accuracy": 0.7641880341880344,
      "mean_token_accuracy_utility": 0.6605250305250303,
      "step": 885
    },
    {
      "epoch": 0.3293551669904709,
      "grad_norm": 0.0,
      "learning_rate": 1.696752194822819e-05,
      "loss": 2.3463,
      "mean_token_accuracy": 0.761050061050061,
      "mean_token_accuracy_utility": 0.6459157509157508,
      "step": 890
    },
    {
      "epoch": 0.33120547691738367,
      "grad_norm": 0.0,
      "learning_rate": 1.692102754306895e-05,
      "loss": 2.334,
      "mean_token_accuracy": 0.7614163614163616,
      "mean_token_accuracy_utility": 0.6458119658119656,
      "step": 895
    },
    {
      "epoch": 0.33305578684429643,
      "grad_norm": 0.0,
      "learning_rate": 1.6874244177128395e-05,
      "loss": 2.2319,
      "mean_token_accuracy": 0.7608058608058609,
      "mean_token_accuracy_utility": 0.6675946275946274,
      "step": 900
    },
    {
      "epoch": 0.33305578684429643,
      "eval_mean_token_accuracy": 0.7366462513521337,
      "eval_reasoning_loss": 0.9170047044754028,
      "eval_reasoning_runtime": 29.3855,
      "eval_reasoning_samples_per_second": 4.39,
      "eval_reasoning_steps_per_second": 1.123,
      "step": 900
    },
    {
      "epoch": 0.33305578684429643,
      "eval_mean_token_accuracy": 0.6603344642521082,
      "eval_utility_loss": 1.4367903470993042,
      "eval_utility_runtime": 11.0975,
      "eval_utility_samples_per_second": 4.415,
      "eval_utility_steps_per_second": 1.171,
      "step": 900
    },
    {
      "epoch": 0.3349060967712092,
      "grad_norm": 0.0,
      "learning_rate": 1.6827173803665328e-05,
      "loss": 2.3799,
      "mean_token_accuracy": 0.7354761904761904,
      "mean_token_accuracy_utility": 0.666233211233211,
      "step": 905
    },
    {
      "epoch": 0.33675640669812196,
      "grad_norm": 0.0,
      "learning_rate": 1.677981838792144e-05,
      "loss": 2.3178,
      "mean_token_accuracy": 0.7454761904761906,
      "mean_token_accuracy_utility": 0.6610989010989009,
      "step": 910
    },
    {
      "epoch": 0.3386067166250347,
      "grad_norm": 0.0,
      "learning_rate": 1.6732179907039266e-05,
      "loss": 2.2273,
      "mean_token_accuracy": 0.7507387057387058,
      "mean_token_accuracy_utility": 0.6721794871794871,
      "step": 915
    },
    {
      "epoch": 0.34045702655194743,
      "grad_norm": 0.0,
      "learning_rate": 1.6684260349979637e-05,
      "loss": 2.2861,
      "mean_token_accuracy": 0.7464529914529915,
      "mean_token_accuracy_utility": 0.666190476190476,
      "step": 920
    },
    {
      "epoch": 0.3423073364788602,
      "grad_norm": 0.0,
      "learning_rate": 1.6636061717438626e-05,
      "loss": 2.1852,
      "mean_token_accuracy": 0.7628021978021978,
      "mean_token_accuracy_utility": 0.6739621489621488,
      "step": 925
    },
    {
      "epoch": 0.34415764640577295,
      "grad_norm": 0.0,
      "learning_rate": 1.6587586021764022e-05,
      "loss": 2.2278,
      "mean_token_accuracy": 0.7576495726495727,
      "mean_token_accuracy_utility": 0.6689700109730807,
      "step": 930
    },
    {
      "epoch": 0.3460079563326857,
      "grad_norm": 0.0,
      "learning_rate": 1.653883528687133e-05,
      "loss": 2.3376,
      "mean_token_accuracy": 0.7487296549771991,
      "mean_token_accuracy_utility": 0.6522660051877882,
      "step": 935
    },
    {
      "epoch": 0.3478582662595985,
      "grad_norm": 0.0,
      "learning_rate": 1.6489811548159245e-05,
      "loss": 2.314,
      "mean_token_accuracy": 0.744010989010989,
      "mean_token_accuracy_utility": 0.6598595848595847,
      "step": 940
    },
    {
      "epoch": 0.34970857618651124,
      "grad_norm": 0.0,
      "learning_rate": 1.6440516852424678e-05,
      "loss": 2.3312,
      "mean_token_accuracy": 0.7438888888888888,
      "mean_token_accuracy_utility": 0.6614713064713064,
      "step": 945
    },
    {
      "epoch": 0.351558886113424,
      "grad_norm": 0.0,
      "learning_rate": 1.6390953257777324e-05,
      "loss": 2.2348,
      "mean_token_accuracy": 0.7566056166056165,
      "mean_token_accuracy_utility": 0.6707264957264955,
      "step": 950
    },
    {
      "epoch": 0.35340919604033677,
      "grad_norm": 0.0,
      "learning_rate": 1.634112283355369e-05,
      "loss": 2.3894,
      "mean_token_accuracy": 0.7444383394383395,
      "mean_token_accuracy_utility": 0.6478144078144076,
      "step": 955
    },
    {
      "epoch": 0.35525950596724953,
      "grad_norm": 0.0,
      "learning_rate": 1.6291027660230735e-05,
      "loss": 2.2673,
      "mean_token_accuracy": 0.7532722832722832,
      "mean_token_accuracy_utility": 0.6658584777514818,
      "step": 960
    },
    {
      "epoch": 0.3571098158941623,
      "grad_norm": 0.0,
      "learning_rate": 1.6240669829338992e-05,
      "loss": 2.3823,
      "mean_token_accuracy": 0.7368131868131866,
      "mean_token_accuracy_utility": 0.6511050061050059,
      "step": 965
    },
    {
      "epoch": 0.358960125821075,
      "grad_norm": 0.0,
      "learning_rate": 1.6190051443375248e-05,
      "loss": 2.2525,
      "mean_token_accuracy": 0.7525641025641027,
      "mean_token_accuracy_utility": 0.6703418803418801,
      "step": 970
    },
    {
      "epoch": 0.36081043574798777,
      "grad_norm": 0.0,
      "learning_rate": 1.6139174615714753e-05,
      "loss": 2.3091,
      "mean_token_accuracy": 0.7524481074481073,
      "mean_token_accuracy_utility": 0.6610372384247815,
      "step": 975
    },
    {
      "epoch": 0.36266074567490053,
      "grad_norm": 0.0,
      "learning_rate": 1.6088041470523005e-05,
      "loss": 2.3662,
      "mean_token_accuracy": 0.7376678876678879,
      "mean_token_accuracy_utility": 0.6601587301587298,
      "step": 980
    },
    {
      "epoch": 0.3645110556018133,
      "grad_norm": 0.0,
      "learning_rate": 1.6036654142667043e-05,
      "loss": 2.3741,
      "mean_token_accuracy": 0.7409340659340659,
      "mean_token_accuracy_utility": 0.6571065862242331,
      "step": 985
    },
    {
      "epoch": 0.36636136552872606,
      "grad_norm": 0.0,
      "learning_rate": 1.598501477762632e-05,
      "loss": 2.351,
      "mean_token_accuracy": 0.7456776556776556,
      "mean_token_accuracy_utility": 0.657197802197802,
      "step": 990
    },
    {
      "epoch": 0.3682116754556388,
      "grad_norm": 0.0,
      "learning_rate": 1.5933125531403135e-05,
      "loss": 2.3899,
      "mean_token_accuracy": 0.747815567187807,
      "mean_token_accuracy_utility": 0.6437912087912087,
      "step": 995
    },
    {
      "epoch": 0.3700619853825516,
      "grad_norm": 0.0,
      "learning_rate": 1.5880988570432603e-05,
      "loss": 2.3387,
      "mean_token_accuracy": 0.7626190476190476,
      "mean_token_accuracy_utility": 0.6458485958485957,
      "step": 1000
    },
    {
      "epoch": 0.3700619853825516,
      "eval_mean_token_accuracy": 0.7376177199706612,
      "eval_reasoning_loss": 0.9119777679443359,
      "eval_reasoning_runtime": 29.358,
      "eval_reasoning_samples_per_second": 4.394,
      "eval_reasoning_steps_per_second": 1.124,
      "step": 1000
    },
    {
      "epoch": 0.3700619853825516,
      "eval_mean_token_accuracy": 0.660832257057593,
      "eval_utility_loss": 1.4346266984939575,
      "eval_utility_runtime": 11.0814,
      "eval_utility_samples_per_second": 4.422,
      "eval_utility_steps_per_second": 1.173,
      "step": 1000
    },
    {
      "epoch": 0.37191229530946435,
      "grad_norm": 0.0,
      "learning_rate": 1.582860607149222e-05,
      "loss": 2.2942,
      "mean_token_accuracy": 0.7493589743589743,
      "mean_token_accuracy_utility": 0.6620634920634918,
      "step": 1005
    },
    {
      "epoch": 0.3737626052363771,
      "grad_norm": 0.0,
      "learning_rate": 1.5775980221610966e-05,
      "loss": 2.3481,
      "mean_token_accuracy": 0.7462210012210012,
      "mean_token_accuracy_utility": 0.6581743261188,
      "step": 1010
    },
    {
      "epoch": 0.3756129151632899,
      "grad_norm": 0.0,
      "learning_rate": 1.5723113217978e-05,
      "loss": 2.2593,
      "mean_token_accuracy": 0.7497252747252747,
      "mean_token_accuracy_utility": 0.6754639804639804,
      "step": 1015
    },
    {
      "epoch": 0.37746322509020264,
      "grad_norm": 0.0,
      "learning_rate": 1.567000726785093e-05,
      "loss": 2.169,
      "mean_token_accuracy": 0.7610683760683761,
      "mean_token_accuracy_utility": 0.6811111111111108,
      "step": 1020
    },
    {
      "epoch": 0.37931353501711534,
      "grad_norm": 0.0,
      "learning_rate": 1.561666458846365e-05,
      "loss": 2.3628,
      "mean_token_accuracy": 0.7349694749694751,
      "mean_token_accuracy_utility": 0.6594566544566542,
      "step": 1025
    },
    {
      "epoch": 0.3811638449440281,
      "grad_norm": 0.0,
      "learning_rate": 1.5563087406933762e-05,
      "loss": 2.2995,
      "mean_token_accuracy": 0.7427106227106226,
      "mean_token_accuracy_utility": 0.6641330891330889,
      "step": 1030
    },
    {
      "epoch": 0.38301415487094087,
      "grad_norm": 0.0,
      "learning_rate": 1.550927796016961e-05,
      "loss": 2.362,
      "mean_token_accuracy": 0.7538827838827837,
      "mean_token_accuracy_utility": 0.646147741147741,
      "step": 1035
    },
    {
      "epoch": 0.38486446479785363,
      "grad_norm": 0.0,
      "learning_rate": 1.5455238494776876e-05,
      "loss": 2.3385,
      "mean_token_accuracy": 0.7411965811965813,
      "mean_token_accuracy_utility": 0.6603663003663002,
      "step": 1040
    },
    {
      "epoch": 0.3867147747247664,
      "grad_norm": 0.0,
      "learning_rate": 1.5400971266964772e-05,
      "loss": 2.26,
      "mean_token_accuracy": 0.7605372405372408,
      "mean_token_accuracy_utility": 0.6607612615027249,
      "step": 1045
    },
    {
      "epoch": 0.38856508465167916,
      "grad_norm": 0.0,
      "learning_rate": 1.5346478542451862e-05,
      "loss": 2.2436,
      "mean_token_accuracy": 0.7446825396825396,
      "mean_token_accuracy_utility": 0.677600934022772,
      "step": 1050
    },
    {
      "epoch": 0.3904153945785919,
      "grad_norm": 0.0,
      "learning_rate": 1.529176259637145e-05,
      "loss": 2.4048,
      "mean_token_accuracy": 0.7464285714285716,
      "mean_token_accuracy_utility": 0.6486691086691085,
      "step": 1055
    },
    {
      "epoch": 0.3922657045055047,
      "grad_norm": 0.0,
      "learning_rate": 1.5236825713176584e-05,
      "loss": 2.3465,
      "mean_token_accuracy": 0.7436202686202688,
      "mean_token_accuracy_utility": 0.6564590964590962,
      "step": 1060
    },
    {
      "epoch": 0.39411601443241745,
      "grad_norm": 0.0,
      "learning_rate": 1.5181670186544706e-05,
      "loss": 2.388,
      "mean_token_accuracy": 0.7415873015873015,
      "mean_token_accuracy_utility": 0.653357616625208,
      "step": 1065
    },
    {
      "epoch": 0.3959663243593302,
      "grad_norm": 0.0,
      "learning_rate": 1.5126298319281859e-05,
      "loss": 2.2997,
      "mean_token_accuracy": 0.7649572649572651,
      "mean_token_accuracy_utility": 0.6557997557997556,
      "step": 1070
    },
    {
      "epoch": 0.3978166342862429,
      "grad_norm": 0.0,
      "learning_rate": 1.5070712423226552e-05,
      "loss": 2.2408,
      "mean_token_accuracy": 0.7570757020757021,
      "mean_token_accuracy_utility": 0.6701221001220999,
      "step": 1075
    },
    {
      "epoch": 0.3996669442131557,
      "grad_norm": 0.0,
      "learning_rate": 1.5014914819153252e-05,
      "loss": 2.2491,
      "mean_token_accuracy": 0.7396825396825395,
      "mean_token_accuracy_utility": 0.6747802197802196,
      "step": 1080
    },
    {
      "epoch": 0.40151725414006845,
      "grad_norm": 0.0,
      "learning_rate": 1.4958907836675467e-05,
      "loss": 2.315,
      "mean_token_accuracy": 0.7454334554334553,
      "mean_token_accuracy_utility": 0.6616422466422465,
      "step": 1085
    },
    {
      "epoch": 0.4033675640669812,
      "grad_norm": 0.0,
      "learning_rate": 1.490269381414849e-05,
      "loss": 2.2999,
      "mean_token_accuracy": 0.7494749694749696,
      "mean_token_accuracy_utility": 0.6597008547008546,
      "step": 1090
    },
    {
      "epoch": 0.405217873993894,
      "grad_norm": 0.0,
      "learning_rate": 1.484627509857178e-05,
      "loss": 2.2315,
      "mean_token_accuracy": 0.7518070818070819,
      "mean_token_accuracy_utility": 0.6739926739926737,
      "step": 1095
    },
    {
      "epoch": 0.40706818392080674,
      "grad_norm": 0.0,
      "learning_rate": 1.4789654045490957e-05,
      "loss": 2.2808,
      "mean_token_accuracy": 0.7581623931623931,
      "mean_token_accuracy_utility": 0.6565750915750914,
      "step": 1100
    },
    {
      "epoch": 0.40706818392080674,
      "eval_mean_token_accuracy": 0.7382968882968883,
      "eval_reasoning_loss": 0.9083702564239502,
      "eval_reasoning_runtime": 29.3184,
      "eval_reasoning_samples_per_second": 4.4,
      "eval_reasoning_steps_per_second": 1.126,
      "step": 1100
    },
    {
      "epoch": 0.40706818392080674,
      "eval_mean_token_accuracy": 0.6611093304115895,
      "eval_utility_loss": 1.4329625368118286,
      "eval_utility_runtime": 11.0638,
      "eval_utility_samples_per_second": 4.429,
      "eval_utility_steps_per_second": 1.175,
      "step": 1100
    },
    {
      "epoch": 0.4089184938477195,
      "grad_norm": 0.0,
      "learning_rate": 1.4732833018899468e-05,
      "loss": 2.3362,
      "mean_token_accuracy": 0.7439926739926739,
      "mean_token_accuracy_utility": 0.6629059829059827,
      "step": 1105
    },
    {
      "epoch": 0.41076880377463226,
      "grad_norm": 0.0,
      "learning_rate": 1.4675814391139875e-05,
      "loss": 2.2925,
      "mean_token_accuracy": 0.7469597069597069,
      "mean_token_accuracy_utility": 0.6616117216117214,
      "step": 1110
    },
    {
      "epoch": 0.412619113701545,
      "grad_norm": 0.0,
      "learning_rate": 1.4618600542804819e-05,
      "loss": 2.3693,
      "mean_token_accuracy": 0.7297924297924296,
      "mean_token_accuracy_utility": 0.6568498168498167,
      "step": 1115
    },
    {
      "epoch": 0.4144694236284578,
      "grad_norm": 0.0,
      "learning_rate": 1.4561193862637621e-05,
      "loss": 2.2384,
      "mean_token_accuracy": 0.7511538461538462,
      "mean_token_accuracy_utility": 0.6788949938949937,
      "step": 1120
    },
    {
      "epoch": 0.4163197335553705,
      "grad_norm": 0.0,
      "learning_rate": 1.4503596747432554e-05,
      "loss": 2.3474,
      "mean_token_accuracy": 0.7514774114774114,
      "mean_token_accuracy_utility": 0.6547504129857069,
      "step": 1125
    },
    {
      "epoch": 0.41817004348228326,
      "grad_norm": 0.0,
      "learning_rate": 1.4445811601934763e-05,
      "loss": 2.2204,
      "mean_token_accuracy": 0.7524114774114774,
      "mean_token_accuracy_utility": 0.6713003663003662,
      "step": 1130
    },
    {
      "epoch": 0.420020353409196,
      "grad_norm": 0.0,
      "learning_rate": 1.4387840838739875e-05,
      "loss": 2.35,
      "mean_token_accuracy": 0.7437301587301587,
      "mean_token_accuracy_utility": 0.6632173382173381,
      "step": 1135
    },
    {
      "epoch": 0.4218706633361088,
      "grad_norm": 0.0,
      "learning_rate": 1.4329686878193271e-05,
      "loss": 2.2829,
      "mean_token_accuracy": 0.7455250305250306,
      "mean_token_accuracy_utility": 0.6650566790983234,
      "step": 1140
    },
    {
      "epoch": 0.42372097326302155,
      "grad_norm": 0.0,
      "learning_rate": 1.4271352148289025e-05,
      "loss": 2.3158,
      "mean_token_accuracy": 0.7634676434676435,
      "mean_token_accuracy_utility": 0.6520757020757021,
      "step": 1145
    },
    {
      "epoch": 0.4255712831899343,
      "grad_norm": 0.0,
      "learning_rate": 1.421283908456854e-05,
      "loss": 2.3542,
      "mean_token_accuracy": 0.7650366300366301,
      "mean_token_accuracy_utility": 0.6433638583638581,
      "step": 1150
    },
    {
      "epoch": 0.4274215931168471,
      "grad_norm": 0.0,
      "learning_rate": 1.4154150130018867e-05,
      "loss": 2.2623,
      "mean_token_accuracy": 0.7606837606837606,
      "mean_token_accuracy_utility": 0.6615018315018313,
      "step": 1155
    },
    {
      "epoch": 0.42927190304375984,
      "grad_norm": 0.0,
      "learning_rate": 1.4095287734970678e-05,
      "loss": 2.2863,
      "mean_token_accuracy": 0.7538461538461539,
      "mean_token_accuracy_utility": 0.6609768009768009,
      "step": 1160
    },
    {
      "epoch": 0.4311222129706726,
      "grad_norm": 0.0,
      "learning_rate": 1.4036254356996004e-05,
      "loss": 2.3654,
      "mean_token_accuracy": 0.7451831501831501,
      "mean_token_accuracy_utility": 0.6548534798534796,
      "step": 1165
    },
    {
      "epoch": 0.43297252289758537,
      "grad_norm": 0.0,
      "learning_rate": 1.3977052460805597e-05,
      "loss": 2.3551,
      "mean_token_accuracy": 0.7543406593406595,
      "mean_token_accuracy_utility": 0.6475030525030523,
      "step": 1170
    },
    {
      "epoch": 0.43482283282449813,
      "grad_norm": 0.0,
      "learning_rate": 1.3917684518146044e-05,
      "loss": 2.2463,
      "mean_token_accuracy": 0.7522710622710622,
      "mean_token_accuracy_utility": 0.6677167277167275,
      "step": 1175
    },
    {
      "epoch": 0.43667314275141084,
      "grad_norm": 0.0,
      "learning_rate": 1.3858153007696552e-05,
      "loss": 2.2911,
      "mean_token_accuracy": 0.7513858363858362,
      "mean_token_accuracy_utility": 0.6605311355311352,
      "step": 1180
    },
    {
      "epoch": 0.4385234526783236,
      "grad_norm": 0.0,
      "learning_rate": 1.3798460414965475e-05,
      "loss": 2.2155,
      "mean_token_accuracy": 0.7645360195360194,
      "mean_token_accuracy_utility": 0.6667521367521366,
      "step": 1185
    },
    {
      "epoch": 0.44037376260523636,
      "grad_norm": 0.0,
      "learning_rate": 1.3738609232186537e-05,
      "loss": 2.3385,
      "mean_token_accuracy": 0.7363492063492062,
      "mean_token_accuracy_utility": 0.659114774114774,
      "step": 1190
    },
    {
      "epoch": 0.4422240725321491,
      "grad_norm": 0.0,
      "learning_rate": 1.3678601958214779e-05,
      "loss": 2.2011,
      "mean_token_accuracy": 0.7623870573870575,
      "mean_token_accuracy_utility": 0.6690170940170937,
      "step": 1195
    },
    {
      "epoch": 0.4440743824590619,
      "grad_norm": 0.0,
      "learning_rate": 1.3618441098422215e-05,
      "loss": 2.2845,
      "mean_token_accuracy": 0.7614713064713066,
      "mean_token_accuracy_utility": 0.654139194139194,
      "step": 1200
    },
    {
      "epoch": 0.4440743824590619,
      "eval_mean_token_accuracy": 0.7386685210214622,
      "eval_reasoning_loss": 0.905845046043396,
      "eval_reasoning_runtime": 29.0787,
      "eval_reasoning_samples_per_second": 4.436,
      "eval_reasoning_steps_per_second": 1.135,
      "step": 1200
    },
    {
      "epoch": 0.4440743824590619,
      "eval_mean_token_accuracy": 0.6614380615095514,
      "eval_utility_loss": 1.431985855102539,
      "eval_utility_runtime": 11.0052,
      "eval_utility_samples_per_second": 4.452,
      "eval_utility_steps_per_second": 1.181,
      "step": 1200
    },
    {
      "epoch": 0.44592469238597465,
      "grad_norm": 0.0,
      "learning_rate": 1.3558129164593256e-05,
      "loss": 2.3625,
      "mean_token_accuracy": 0.7486568986568987,
      "mean_token_accuracy_utility": 0.6499911314108844,
      "step": 1205
    },
    {
      "epoch": 0.4477750023128874,
      "grad_norm": 0.0,
      "learning_rate": 1.349766867481982e-05,
      "loss": 2.2213,
      "mean_token_accuracy": 0.7429670329670328,
      "mean_token_accuracy_utility": 0.6798351648351646,
      "step": 1210
    },
    {
      "epoch": 0.4496253122398002,
      "grad_norm": 0.0,
      "learning_rate": 1.3437062153396201e-05,
      "loss": 2.2454,
      "mean_token_accuracy": 0.7395238095238096,
      "mean_token_accuracy_utility": 0.6761111111111109,
      "step": 1215
    },
    {
      "epoch": 0.45147562216671294,
      "grad_norm": 0.0,
      "learning_rate": 1.337631213071369e-05,
      "loss": 2.5202,
      "mean_token_accuracy": 0.7384676434676436,
      "mean_token_accuracy_utility": 0.6329059829059828,
      "step": 1220
    },
    {
      "epoch": 0.4533259320936257,
      "grad_norm": 0.0,
      "learning_rate": 1.331542114315491e-05,
      "loss": 2.3088,
      "mean_token_accuracy": 0.756092796092796,
      "mean_token_accuracy_utility": 0.6558511895145558,
      "step": 1225
    },
    {
      "epoch": 0.4551762420205384,
      "grad_norm": 0.0,
      "learning_rate": 1.325439173298793e-05,
      "loss": 2.2719,
      "mean_token_accuracy": 0.7651526251526253,
      "mean_token_accuracy_utility": 0.6559218559218557,
      "step": 1230
    },
    {
      "epoch": 0.4570265519474512,
      "grad_norm": 0.0,
      "learning_rate": 1.3193226448260128e-05,
      "loss": 2.3496,
      "mean_token_accuracy": 0.7455372405372407,
      "mean_token_accuracy_utility": 0.6549256299256297,
      "step": 1235
    },
    {
      "epoch": 0.45887686187436394,
      "grad_norm": 0.0,
      "learning_rate": 1.3131927842691793e-05,
      "loss": 2.369,
      "mean_token_accuracy": 0.7443406593406594,
      "mean_token_accuracy_utility": 0.6480769230769229,
      "step": 1240
    },
    {
      "epoch": 0.4607271718012767,
      "grad_norm": 0.0,
      "learning_rate": 1.3070498475569507e-05,
      "loss": 2.3808,
      "mean_token_accuracy": 0.7286691086691086,
      "mean_token_accuracy_utility": 0.6591941391941389,
      "step": 1245
    },
    {
      "epoch": 0.46257748172818947,
      "grad_norm": 0.0,
      "learning_rate": 1.3008940911639302e-05,
      "loss": 2.3203,
      "mean_token_accuracy": 0.7434493284493281,
      "mean_token_accuracy_utility": 0.6714039228363144,
      "step": 1250
    },
    {
      "epoch": 0.46442779165510223,
      "grad_norm": 0.0,
      "learning_rate": 1.2947257720999577e-05,
      "loss": 2.4362,
      "mean_token_accuracy": 0.7542735042735044,
      "mean_token_accuracy_utility": 0.6411477411477411,
      "step": 1255
    },
    {
      "epoch": 0.466278101582015,
      "grad_norm": 0.0,
      "learning_rate": 1.2885451478993777e-05,
      "loss": 2.4432,
      "mean_token_accuracy": 0.7385103785103784,
      "mean_token_accuracy_utility": 0.6436202686202684,
      "step": 1260
    },
    {
      "epoch": 0.46812841150892776,
      "grad_norm": 0.0,
      "learning_rate": 1.282352476610289e-05,
      "loss": 2.342,
      "mean_token_accuracy": 0.7489255189255187,
      "mean_token_accuracy_utility": 0.6554639804639801,
      "step": 1265
    },
    {
      "epoch": 0.4699787214358405,
      "grad_norm": 0.0,
      "learning_rate": 1.2761480167837705e-05,
      "loss": 2.2096,
      "mean_token_accuracy": 0.76492673992674,
      "mean_token_accuracy_utility": 0.6660073260073258,
      "step": 1270
    },
    {
      "epoch": 0.4718290313627533,
      "grad_norm": 0.0,
      "learning_rate": 1.2699320274630847e-05,
      "loss": 2.2052,
      "mean_token_accuracy": 0.7566483516483518,
      "mean_token_accuracy_utility": 0.6753052503052501,
      "step": 1275
    },
    {
      "epoch": 0.47367934128966604,
      "grad_norm": 0.0,
      "learning_rate": 1.263704768172864e-05,
      "loss": 2.4247,
      "mean_token_accuracy": 0.7346459096459095,
      "mean_token_accuracy_utility": 0.6482967032967031,
      "step": 1280
    },
    {
      "epoch": 0.47552965121657875,
      "grad_norm": 0.0,
      "learning_rate": 1.257466498908276e-05,
      "loss": 2.2103,
      "mean_token_accuracy": 0.7549572649572649,
      "mean_token_accuracy_utility": 0.6760683760683759,
      "step": 1285
    },
    {
      "epoch": 0.4773799611434915,
      "grad_norm": 0.0,
      "learning_rate": 1.2512174801241657e-05,
      "loss": 2.239,
      "mean_token_accuracy": 0.7653907203907205,
      "mean_token_accuracy_utility": 0.6602930402930403,
      "step": 1290
    },
    {
      "epoch": 0.4792302710704043,
      "grad_norm": 0.0,
      "learning_rate": 1.2449579727241834e-05,
      "loss": 2.2724,
      "mean_token_accuracy": 0.7510195360195362,
      "mean_token_accuracy_utility": 0.664224664224664,
      "step": 1295
    },
    {
      "epoch": 0.48108058099731704,
      "grad_norm": 0.0,
      "learning_rate": 1.2386882380498918e-05,
      "loss": 2.3153,
      "mean_token_accuracy": 0.7501770451770451,
      "mean_token_accuracy_utility": 0.6585286935286934,
      "step": 1300
    },
    {
      "epoch": 0.48108058099731704,
      "eval_mean_token_accuracy": 0.7390810714340127,
      "eval_reasoning_loss": 0.9039227962493896,
      "eval_reasoning_runtime": 29.1085,
      "eval_reasoning_samples_per_second": 4.432,
      "eval_reasoning_steps_per_second": 1.134,
      "step": 1300
    },
    {
      "epoch": 0.48108058099731704,
      "eval_mean_token_accuracy": 0.661781771432901,
      "eval_utility_loss": 1.4310052394866943,
      "eval_utility_runtime": 11.0221,
      "eval_utility_samples_per_second": 4.446,
      "eval_utility_steps_per_second": 1.179,
      "step": 1300
    },
    {
      "epoch": 0.4829308909242298,
      "grad_norm": 0.0,
      "learning_rate": 1.2324085378698529e-05,
      "loss": 2.2388,
      "mean_token_accuracy": 0.7716544566544568,
      "mean_token_accuracy_utility": 0.6585470085470084,
      "step": 1305
    },
    {
      "epoch": 0.48478120085114257,
      "grad_norm": 0.0,
      "learning_rate": 1.2261191343687e-05,
      "loss": 2.2564,
      "mean_token_accuracy": 0.751001221001221,
      "mean_token_accuracy_utility": 0.6667032967032964,
      "step": 1310
    },
    {
      "epoch": 0.48663151077805533,
      "grad_norm": 0.0,
      "learning_rate": 1.219820290136192e-05,
      "loss": 2.3089,
      "mean_token_accuracy": 0.7663064713064713,
      "mean_token_accuracy_utility": 0.6536630036630036,
      "step": 1315
    },
    {
      "epoch": 0.4884818207049681,
      "grad_norm": 0.0,
      "learning_rate": 1.2135122681562481e-05,
      "loss": 2.3202,
      "mean_token_accuracy": 0.746990231990232,
      "mean_token_accuracy_utility": 0.6605616605616604,
      "step": 1320
    },
    {
      "epoch": 0.49033213063188086,
      "grad_norm": 0.0,
      "learning_rate": 1.2071953317959692e-05,
      "loss": 2.2488,
      "mean_token_accuracy": 0.7547944829479175,
      "mean_token_accuracy_utility": 0.667185592185592,
      "step": 1325
    },
    {
      "epoch": 0.4921824405587936,
      "grad_norm": 0.0,
      "learning_rate": 1.2008697447946421e-05,
      "loss": 2.2445,
      "mean_token_accuracy": 0.7544688644688646,
      "mean_token_accuracy_utility": 0.6723015873015871,
      "step": 1330
    },
    {
      "epoch": 0.49403275048570633,
      "grad_norm": 0.0,
      "learning_rate": 1.1945357712527273e-05,
      "loss": 2.3376,
      "mean_token_accuracy": 0.7505128205128204,
      "mean_token_accuracy_utility": 0.6567704517704516,
      "step": 1335
    },
    {
      "epoch": 0.4958830604126191,
      "grad_norm": 0.0,
      "learning_rate": 1.1881936756208329e-05,
      "loss": 2.2438,
      "mean_token_accuracy": 0.7548168498168499,
      "mean_token_accuracy_utility": 0.665262515262515,
      "step": 1340
    },
    {
      "epoch": 0.49773337033953186,
      "grad_norm": 0.0,
      "learning_rate": 1.1818437226886738e-05,
      "loss": 2.3341,
      "mean_token_accuracy": 0.7567643467643469,
      "mean_token_accuracy_utility": 0.6514163614163612,
      "step": 1345
    },
    {
      "epoch": 0.4995836802664446,
      "grad_norm": 0.0,
      "learning_rate": 1.1754861775740163e-05,
      "loss": 2.1868,
      "mean_token_accuracy": 0.754041514041514,
      "mean_token_accuracy_utility": 0.6788217338217338,
      "step": 1350
    },
    {
      "epoch": 0.5014339901933574,
      "grad_norm": 0.0,
      "learning_rate": 1.1691213057116082e-05,
      "loss": 2.3109,
      "mean_token_accuracy": 0.759945054945055,
      "mean_token_accuracy_utility": 0.653211233211233,
      "step": 1355
    },
    {
      "epoch": 0.5032843001202701,
      "grad_norm": 0.0,
      "learning_rate": 1.1627493728420978e-05,
      "loss": 2.2506,
      "mean_token_accuracy": 0.7542429792429792,
      "mean_token_accuracy_utility": 0.6717245119509123,
      "step": 1360
    },
    {
      "epoch": 0.5051346100471829,
      "grad_norm": 0.0,
      "learning_rate": 1.1563706450009391e-05,
      "loss": 2.368,
      "mean_token_accuracy": 0.7515201465201464,
      "mean_token_accuracy_utility": 0.650201465201465,
      "step": 1365
    },
    {
      "epoch": 0.5069849199740957,
      "grad_norm": 0.0,
      "learning_rate": 1.1499853885072827e-05,
      "loss": 2.2694,
      "mean_token_accuracy": 0.7682661782661782,
      "mean_token_accuracy_utility": 0.6525030525030523,
      "step": 1370
    },
    {
      "epoch": 0.5088352299010084,
      "grad_norm": 0.0,
      "learning_rate": 1.1435938699528586e-05,
      "loss": 2.3588,
      "mean_token_accuracy": 0.7355921855921854,
      "mean_token_accuracy_utility": 0.659157509157509,
      "step": 1375
    },
    {
      "epoch": 0.5106855398279212,
      "grad_norm": 0.0,
      "learning_rate": 1.137196356190845e-05,
      "loss": 2.3947,
      "mean_token_accuracy": 0.7474969474969477,
      "mean_token_accuracy_utility": 0.6477055739339476,
      "step": 1380
    },
    {
      "epoch": 0.5125358497548339,
      "grad_norm": 0.0,
      "learning_rate": 1.1307931143247268e-05,
      "loss": 2.2296,
      "mean_token_accuracy": 0.7489743589743589,
      "mean_token_accuracy_utility": 0.6737362637362636,
      "step": 1385
    },
    {
      "epoch": 0.5143861596817467,
      "grad_norm": 0.0,
      "learning_rate": 1.1243844116971433e-05,
      "loss": 2.3122,
      "mean_token_accuracy": 0.7536263736263736,
      "mean_token_accuracy_utility": 0.6573626373626372,
      "step": 1390
    },
    {
      "epoch": 0.5162364696086594,
      "grad_norm": 0.0,
      "learning_rate": 1.1179705158787276e-05,
      "loss": 2.2812,
      "mean_token_accuracy": 0.7447435897435899,
      "mean_token_accuracy_utility": 0.6629345947345446,
      "step": 1395
    },
    {
      "epoch": 0.5180867795355723,
      "grad_norm": 0.0,
      "learning_rate": 1.1115516946569333e-05,
      "loss": 2.3475,
      "mean_token_accuracy": 0.7567887667887667,
      "mean_token_accuracy_utility": 0.6524786324786322,
      "step": 1400
    },
    {
      "epoch": 0.5180867795355723,
      "eval_mean_token_accuracy": 0.7392568216097627,
      "eval_reasoning_loss": 0.9024344682693481,
      "eval_reasoning_runtime": 29.1565,
      "eval_reasoning_samples_per_second": 4.424,
      "eval_reasoning_steps_per_second": 1.132,
      "step": 1400
    },
    {
      "epoch": 0.5180867795355723,
      "eval_mean_token_accuracy": 0.6615192928318463,
      "eval_utility_loss": 1.4302774667739868,
      "eval_utility_runtime": 11.0184,
      "eval_utility_samples_per_second": 4.447,
      "eval_utility_steps_per_second": 1.18,
      "step": 1400
    },
    {
      "epoch": 0.519937089462485,
      "grad_norm": 0.0,
      "learning_rate": 1.105128216024857e-05,
      "loss": 2.259,
      "mean_token_accuracy": 0.7605128205128205,
      "mean_token_accuracy_utility": 0.6664828109130875,
      "step": 1405
    },
    {
      "epoch": 0.5217873993893978,
      "grad_norm": 0.0,
      "learning_rate": 1.0987003481700456e-05,
      "loss": 2.2623,
      "mean_token_accuracy": 0.7601404151404153,
      "mean_token_accuracy_utility": 0.6651831501831499,
      "step": 1410
    },
    {
      "epoch": 0.5236377093163105,
      "grad_norm": 0.0,
      "learning_rate": 1.092268359463302e-05,
      "loss": 2.2614,
      "mean_token_accuracy": 0.762735902869143,
      "mean_token_accuracy_utility": 0.6597069597069595,
      "step": 1415
    },
    {
      "epoch": 0.5254880192432232,
      "grad_norm": 0.0,
      "learning_rate": 1.0858325184474796e-05,
      "loss": 2.2855,
      "mean_token_accuracy": 0.7554945054945055,
      "mean_token_accuracy_utility": 0.661074481074481,
      "step": 1420
    },
    {
      "epoch": 0.527338329170136,
      "grad_norm": 0.0,
      "learning_rate": 1.0793930938262689e-05,
      "loss": 2.3573,
      "mean_token_accuracy": 0.7442796092796092,
      "mean_token_accuracy_utility": 0.6547069597069596,
      "step": 1425
    },
    {
      "epoch": 0.5291886390970487,
      "grad_norm": 0.0,
      "learning_rate": 1.0729503544529814e-05,
      "loss": 2.3155,
      "mean_token_accuracy": 0.7555921855921858,
      "mean_token_accuracy_utility": 0.6555799755799754,
      "step": 1430
    },
    {
      "epoch": 0.5310389490239615,
      "grad_norm": 0.0,
      "learning_rate": 1.0665045693193226e-05,
      "loss": 2.354,
      "mean_token_accuracy": 0.7535958485958487,
      "mean_token_accuracy_utility": 0.6500915750915749,
      "step": 1435
    },
    {
      "epoch": 0.5328892589508742,
      "grad_norm": 0.0,
      "learning_rate": 1.0600560075441617e-05,
      "loss": 2.278,
      "mean_token_accuracy": 0.7668864468864469,
      "mean_token_accuracy_utility": 0.6547191697191695,
      "step": 1440
    },
    {
      "epoch": 0.5347395688777871,
      "grad_norm": 0.0,
      "learning_rate": 1.0536049383622966e-05,
      "loss": 2.3008,
      "mean_token_accuracy": 0.7421733821733822,
      "mean_token_accuracy_utility": 0.6671367521367519,
      "step": 1445
    },
    {
      "epoch": 0.5365898788046998,
      "grad_norm": 0.0,
      "learning_rate": 1.047151631113212e-05,
      "loss": 2.2214,
      "mean_token_accuracy": 0.7599938949938952,
      "mean_token_accuracy_utility": 0.6680586080586078,
      "step": 1450
    },
    {
      "epoch": 0.5384401887316126,
      "grad_norm": 0.0,
      "learning_rate": 1.0406963552298332e-05,
      "loss": 2.2373,
      "mean_token_accuracy": 0.7671184371184372,
      "mean_token_accuracy_utility": 0.6613369963369962,
      "step": 1455
    },
    {
      "epoch": 0.5402904986585253,
      "grad_norm": 0.0,
      "learning_rate": 1.034239380227281e-05,
      "loss": 2.2556,
      "mean_token_accuracy": 0.7641697191697194,
      "mean_token_accuracy_utility": 0.662280052604596,
      "step": 1460
    },
    {
      "epoch": 0.542140808585438,
      "grad_norm": 0.0,
      "learning_rate": 1.0277809756916134e-05,
      "loss": 2.3839,
      "mean_token_accuracy": 0.7436691086691084,
      "mean_token_accuracy_utility": 0.6514896214896213,
      "step": 1465
    },
    {
      "epoch": 0.5439911185123508,
      "grad_norm": 0.0,
      "learning_rate": 1.0213214112685747e-05,
      "loss": 2.2826,
      "mean_token_accuracy": 0.7566239316239315,
      "mean_token_accuracy_utility": 0.6595238095238093,
      "step": 1470
    },
    {
      "epoch": 0.5458414284392635,
      "grad_norm": 0.0,
      "learning_rate": 1.0148609566523358e-05,
      "loss": 2.2796,
      "mean_token_accuracy": 0.7477411477411479,
      "mean_token_accuracy_utility": 0.666074481074481,
      "step": 1475
    },
    {
      "epoch": 0.5476917383661764,
      "grad_norm": 0.0,
      "learning_rate": 1.0083998815742335e-05,
      "loss": 2.2287,
      "mean_token_accuracy": 0.7456410256410254,
      "mean_token_accuracy_utility": 0.6807326007326007,
      "step": 1480
    },
    {
      "epoch": 0.5495420482930891,
      "grad_norm": 0.0,
      "learning_rate": 1.0019384557915099e-05,
      "loss": 2.2741,
      "mean_token_accuracy": 0.7523199023199023,
      "mean_token_accuracy_utility": 0.6624236874236872,
      "step": 1485
    },
    {
      "epoch": 0.5513923582200019,
      "grad_norm": 0.0,
      "learning_rate": 9.9547694907605e-06,
      "loss": 2.2408,
      "mean_token_accuracy": 0.7531807081807083,
      "mean_token_accuracy_utility": 0.6668559218559217,
      "step": 1490
    },
    {
      "epoch": 0.5532426681469146,
      "grad_norm": 0.0,
      "learning_rate": 9.890156312031165e-06,
      "loss": 2.3272,
      "mean_token_accuracy": 0.7567277167277168,
      "mean_token_accuracy_utility": 0.6602354488980003,
      "step": 1495
    },
    {
      "epoch": 0.5550929780738274,
      "grad_norm": 0.0,
      "learning_rate": 9.825547719400889e-06,
      "loss": 2.306,
      "mean_token_accuracy": 0.7595360195360195,
      "mean_token_accuracy_utility": 0.650250305250305,
      "step": 1500
    },
    {
      "epoch": 0.5550929780738274,
      "eval_mean_token_accuracy": 0.7396587073057664,
      "eval_reasoning_loss": 0.9013679027557373,
      "eval_reasoning_runtime": 28.9519,
      "eval_reasoning_samples_per_second": 4.456,
      "eval_reasoning_steps_per_second": 1.14,
      "step": 1500
    },
    {
      "epoch": 0.5550929780738274,
      "eval_mean_token_accuracy": 0.6616568780566492,
      "eval_utility_loss": 1.4297488927841187,
      "eval_utility_runtime": 10.7817,
      "eval_utility_samples_per_second": 4.545,
      "eval_utility_steps_per_second": 1.206,
      "step": 1500
    },
    {
      "epoch": 0.5569432880007401,
      "grad_norm": 0.0,
      "learning_rate": 9.760946410351988e-06,
      "loss": 2.3755,
      "mean_token_accuracy": 0.7404395604395604,
      "mean_token_accuracy_utility": 0.6513125763125761,
      "step": 1505
    },
    {
      "epoch": 0.5587935979276529,
      "grad_norm": 0.0,
      "learning_rate": 9.696355082062679e-06,
      "loss": 2.4039,
      "mean_token_accuracy": 0.7470818070818069,
      "mean_token_accuracy_utility": 0.6390354090354089,
      "step": 1510
    },
    {
      "epoch": 0.5606439078545656,
      "grad_norm": 0.0,
      "learning_rate": 9.631776431294475e-06,
      "loss": 2.3281,
      "mean_token_accuracy": 0.7533272283272284,
      "mean_token_accuracy_utility": 0.6579731379731378,
      "step": 1515
    },
    {
      "epoch": 0.5624942177814783,
      "grad_norm": 0.0,
      "learning_rate": 9.567213154279582e-06,
      "loss": 2.242,
      "mean_token_accuracy": 0.7620024420024422,
      "mean_token_accuracy_utility": 0.6618925518925517,
      "step": 1520
    },
    {
      "epoch": 0.5643445277083912,
      "grad_norm": 0.0,
      "learning_rate": 9.502667946608332e-06,
      "loss": 2.237,
      "mean_token_accuracy": 0.7534432234432236,
      "mean_token_accuracy_utility": 0.6730219780219777,
      "step": 1525
    },
    {
      "epoch": 0.5661948376353039,
      "grad_norm": 0.0,
      "learning_rate": 9.43814350311666e-06,
      "loss": 2.2652,
      "mean_token_accuracy": 0.7513064713064712,
      "mean_token_accuracy_utility": 0.6702165831586047,
      "step": 1530
    },
    {
      "epoch": 0.5680451475622167,
      "grad_norm": 0.0,
      "learning_rate": 9.37364251777355e-06,
      "loss": 2.3215,
      "mean_token_accuracy": 0.7397863247863248,
      "mean_token_accuracy_utility": 0.662319902319902,
      "step": 1535
    },
    {
      "epoch": 0.5698954574891294,
      "grad_norm": 0.0,
      "learning_rate": 9.309167683568597e-06,
      "loss": 2.4096,
      "mean_token_accuracy": 0.7469108669108669,
      "mean_token_accuracy_utility": 0.643241758241758,
      "step": 1540
    },
    {
      "epoch": 0.5717457674160422,
      "grad_norm": 0.0,
      "learning_rate": 9.244721692399545e-06,
      "loss": 2.2191,
      "mean_token_accuracy": 0.7615873015873016,
      "mean_token_accuracy_utility": 0.6666910866910865,
      "step": 1545
    },
    {
      "epoch": 0.5735960773429549,
      "grad_norm": 0.0,
      "learning_rate": 9.180307234959918e-06,
      "loss": 2.2053,
      "mean_token_accuracy": 0.7510989010989009,
      "mean_token_accuracy_utility": 0.6734188034188032,
      "step": 1550
    },
    {
      "epoch": 0.5754463872698677,
      "grad_norm": 0.0,
      "learning_rate": 9.115927000626665e-06,
      "loss": 2.2541,
      "mean_token_accuracy": 0.7492979242979242,
      "mean_token_accuracy_utility": 0.6643345543345542,
      "step": 1555
    },
    {
      "epoch": 0.5772966971967805,
      "grad_norm": 0.0,
      "learning_rate": 9.051583677347879e-06,
      "loss": 2.3247,
      "mean_token_accuracy": 0.7496275946275947,
      "mean_token_accuracy_utility": 0.6526434676434676,
      "step": 1560
    },
    {
      "epoch": 0.5791470071236933,
      "grad_norm": 0.0,
      "learning_rate": 8.987279951530586e-06,
      "loss": 2.2338,
      "mean_token_accuracy": 0.7502930402930403,
      "mean_token_accuracy_utility": 0.6709829059829058,
      "step": 1565
    },
    {
      "epoch": 0.580997317050606,
      "grad_norm": 0.0,
      "learning_rate": 8.923018507928564e-06,
      "loss": 2.3777,
      "mean_token_accuracy": 0.744029304029304,
      "mean_token_accuracy_utility": 0.6517338217338216,
      "step": 1570
    },
    {
      "epoch": 0.5828476269775187,
      "grad_norm": 0.0,
      "learning_rate": 8.85880202953026e-06,
      "loss": 2.2296,
      "mean_token_accuracy": 0.7546092796092798,
      "mean_token_accuracy_utility": 0.6719230769230768,
      "step": 1575
    },
    {
      "epoch": 0.5846979369044315,
      "grad_norm": 0.0,
      "learning_rate": 8.79463319744677e-06,
      "loss": 2.3313,
      "mean_token_accuracy": 0.7371428571428569,
      "mean_token_accuracy_utility": 0.66520757020757,
      "step": 1580
    },
    {
      "epoch": 0.5865482468313442,
      "grad_norm": 0.0,
      "learning_rate": 8.730514690799916e-06,
      "loss": 2.225,
      "mean_token_accuracy": 0.7561294261294262,
      "mean_token_accuracy_utility": 0.6693772893772894,
      "step": 1585
    },
    {
      "epoch": 0.588398556758257,
      "grad_norm": 0.0,
      "learning_rate": 8.666449186610353e-06,
      "loss": 2.2748,
      "mean_token_accuracy": 0.7593101343101345,
      "mean_token_accuracy_utility": 0.6619597069597067,
      "step": 1590
    },
    {
      "epoch": 0.5902488666851697,
      "grad_norm": 0.0,
      "learning_rate": 8.60243935968585e-06,
      "loss": 2.2337,
      "mean_token_accuracy": 0.7570329670329673,
      "mean_token_accuracy_utility": 0.6685592185592184,
      "step": 1595
    },
    {
      "epoch": 0.5920991766120826,
      "grad_norm": 0.0,
      "learning_rate": 8.538487882509568e-06,
      "loss": 2.2455,
      "mean_token_accuracy": 0.7551526251526249,
      "mean_token_accuracy_utility": 0.6701770451770449,
      "step": 1600
    },
    {
      "epoch": 0.5920991766120826,
      "eval_mean_token_accuracy": 0.7396603396603396,
      "eval_reasoning_loss": 0.9006361365318298,
      "eval_reasoning_runtime": 29.3823,
      "eval_reasoning_samples_per_second": 4.39,
      "eval_reasoning_steps_per_second": 1.123,
      "step": 1600
    },
    {
      "epoch": 0.5920991766120826,
      "eval_mean_token_accuracy": 0.6619198629350188,
      "eval_utility_loss": 1.4293965101242065,
      "eval_utility_runtime": 11.1035,
      "eval_utility_samples_per_second": 4.413,
      "eval_utility_steps_per_second": 1.171,
      "step": 1600
    },
    {
      "epoch": 0.5939494865389953,
      "grad_norm": 0.0,
      "learning_rate": 8.474597425128501e-06,
      "loss": 2.3168,
      "mean_token_accuracy": 0.7584065934065932,
      "mean_token_accuracy_utility": 0.6528205128205127,
      "step": 1605
    },
    {
      "epoch": 0.5957997964659081,
      "grad_norm": 0.0,
      "learning_rate": 8.410770655042003e-06,
      "loss": 2.2206,
      "mean_token_accuracy": 0.754102564102564,
      "mean_token_accuracy_utility": 0.6736385836385835,
      "step": 1610
    },
    {
      "epoch": 0.5976501063928208,
      "grad_norm": 0.0,
      "learning_rate": 8.347010237090408e-06,
      "loss": 2.3372,
      "mean_token_accuracy": 0.748931623931624,
      "mean_token_accuracy_utility": 0.6568974747431148,
      "step": 1615
    },
    {
      "epoch": 0.5995004163197336,
      "grad_norm": 0.0,
      "learning_rate": 8.283318833343773e-06,
      "loss": 2.2774,
      "mean_token_accuracy": 0.7426984126984127,
      "mean_token_accuracy_utility": 0.6699084249084246,
      "step": 1620
    },
    {
      "epoch": 0.6013507262466463,
      "grad_norm": 0.0,
      "learning_rate": 8.219699102990735e-06,
      "loss": 2.3377,
      "mean_token_accuracy": 0.7473577355828196,
      "mean_token_accuracy_utility": 0.6583333333333331,
      "step": 1625
    },
    {
      "epoch": 0.603201036173559,
      "grad_norm": 0.0,
      "learning_rate": 8.156153702227484e-06,
      "loss": 2.2504,
      "mean_token_accuracy": 0.7555311355311355,
      "mean_token_accuracy_utility": 0.6697863247863247,
      "step": 1630
    },
    {
      "epoch": 0.6050513461004718,
      "grad_norm": 0.0,
      "learning_rate": 8.092685284146865e-06,
      "loss": 2.3351,
      "mean_token_accuracy": 0.735128205128205,
      "mean_token_accuracy_utility": 0.6641158928098858,
      "step": 1635
    },
    {
      "epoch": 0.6069016560273846,
      "grad_norm": 0.0,
      "learning_rate": 8.029296498627608e-06,
      "loss": 2.3632,
      "mean_token_accuracy": 0.7493711843711844,
      "mean_token_accuracy_utility": 0.6531702886789378,
      "step": 1640
    },
    {
      "epoch": 0.6087519659542974,
      "grad_norm": 0.0,
      "learning_rate": 7.965989992223693e-06,
      "loss": 2.2072,
      "mean_token_accuracy": 0.7572344322344323,
      "mean_token_accuracy_utility": 0.6768620268620266,
      "step": 1645
    },
    {
      "epoch": 0.6106022758812101,
      "grad_norm": 0.0,
      "learning_rate": 7.90276840805385e-06,
      "loss": 2.3663,
      "mean_token_accuracy": 0.7447680097680098,
      "mean_token_accuracy_utility": 0.6548107443382591,
      "step": 1650
    },
    {
      "epoch": 0.6124525858081229,
      "grad_norm": 0.0,
      "learning_rate": 7.839634385691214e-06,
      "loss": 2.2593,
      "mean_token_accuracy": 0.7525091575091574,
      "mean_token_accuracy_utility": 0.6667460317460316,
      "step": 1655
    },
    {
      "epoch": 0.6143028957350356,
      "grad_norm": 0.0,
      "learning_rate": 7.776590561053117e-06,
      "loss": 2.2259,
      "mean_token_accuracy": 0.7476190476190478,
      "mean_token_accuracy_utility": 0.6786263736263733,
      "step": 1660
    },
    {
      "epoch": 0.6161532056619484,
      "grad_norm": 0.0,
      "learning_rate": 7.713639566291028e-06,
      "loss": 2.3812,
      "mean_token_accuracy": 0.7338949938949938,
      "mean_token_accuracy_utility": 0.6549145299145298,
      "step": 1665
    },
    {
      "epoch": 0.6180035155888611,
      "grad_norm": 0.0,
      "learning_rate": 7.650784029680662e-06,
      "loss": 2.2518,
      "mean_token_accuracy": 0.7691147741147741,
      "mean_token_accuracy_utility": 0.6588095238095236,
      "step": 1670
    },
    {
      "epoch": 0.6198538255157738,
      "grad_norm": 0.0,
      "learning_rate": 7.58802657551225e-06,
      "loss": 2.2484,
      "mean_token_accuracy": 0.7631257631257633,
      "mean_token_accuracy_utility": 0.6600854700854699,
      "step": 1675
    },
    {
      "epoch": 0.6217041354426867,
      "grad_norm": 0.0,
      "learning_rate": 7.52536982398097e-06,
      "loss": 2.2908,
      "mean_token_accuracy": 0.7484737484737484,
      "mean_token_accuracy_utility": 0.6653846153846152,
      "step": 1680
    },
    {
      "epoch": 0.6235544453695994,
      "grad_norm": 0.0,
      "learning_rate": 7.46281639107755e-06,
      "loss": 2.4683,
      "mean_token_accuracy": 0.7309951413895336,
      "mean_token_accuracy_utility": 0.6456776556776556,
      "step": 1685
    },
    {
      "epoch": 0.6254047552965122,
      "grad_norm": 0.0,
      "learning_rate": 7.400368888479048e-06,
      "loss": 2.2647,
      "mean_token_accuracy": 0.741007326007326,
      "mean_token_accuracy_utility": 0.6766178266178264,
      "step": 1690
    },
    {
      "epoch": 0.6272550652234249,
      "grad_norm": 0.0,
      "learning_rate": 7.3380299234398076e-06,
      "loss": 2.4061,
      "mean_token_accuracy": 0.7418253968253967,
      "mean_token_accuracy_utility": 0.6488400488400485,
      "step": 1695
    },
    {
      "epoch": 0.6291053751503377,
      "grad_norm": 0.0,
      "learning_rate": 7.275802098682612e-06,
      "loss": 2.3254,
      "mean_token_accuracy": 0.7384981684981683,
      "mean_token_accuracy_utility": 0.665262515262515,
      "step": 1700
    },
    {
      "epoch": 0.6291053751503377,
      "eval_mean_token_accuracy": 0.7398545898545895,
      "eval_reasoning_loss": 0.9001015424728394,
      "eval_reasoning_runtime": 29.368,
      "eval_reasoning_samples_per_second": 4.393,
      "eval_reasoning_steps_per_second": 1.124,
      "step": 1700
    },
    {
      "epoch": 0.6291053751503377,
      "eval_mean_token_accuracy": 0.661672869436667,
      "eval_utility_loss": 1.4291335344314575,
      "eval_utility_runtime": 11.0828,
      "eval_utility_samples_per_second": 4.421,
      "eval_utility_steps_per_second": 1.173,
      "step": 1700
    },
    {
      "epoch": 0.6309556850772504,
      "grad_norm": 0.0,
      "learning_rate": 7.213688012290004e-06,
      "loss": 2.2166,
      "mean_token_accuracy": 0.7475030525030526,
      "mean_token_accuracy_utility": 0.6777594627594625,
      "step": 1705
    },
    {
      "epoch": 0.6328059950041632,
      "grad_norm": 0.0,
      "learning_rate": 7.151690257595826e-06,
      "loss": 2.2314,
      "mean_token_accuracy": 0.7548290598290598,
      "mean_token_accuracy_utility": 0.6704395604395603,
      "step": 1710
    },
    {
      "epoch": 0.6346563049310759,
      "grad_norm": 0.0,
      "learning_rate": 7.089811423076936e-06,
      "loss": 2.2541,
      "mean_token_accuracy": 0.7713797313797314,
      "mean_token_accuracy_utility": 0.6576129426129425,
      "step": 1715
    },
    {
      "epoch": 0.6365066148579888,
      "grad_norm": 0.0,
      "learning_rate": 7.028054092245134e-06,
      "loss": 2.3065,
      "mean_token_accuracy": 0.7436385836385836,
      "mean_token_accuracy_utility": 0.6624725274725273,
      "step": 1720
    },
    {
      "epoch": 0.6383569247849015,
      "grad_norm": 0.0,
      "learning_rate": 6.966420843539321e-06,
      "loss": 2.2275,
      "mean_token_accuracy": 0.7414346764346763,
      "mean_token_accuracy_utility": 0.6792735042735041,
      "step": 1725
    },
    {
      "epoch": 0.6402072347118142,
      "grad_norm": 0.0,
      "learning_rate": 6.90491425021781e-06,
      "loss": 2.3075,
      "mean_token_accuracy": 0.7449877899877901,
      "mean_token_accuracy_utility": 0.6668681318681318,
      "step": 1730
    },
    {
      "epoch": 0.642057544638727,
      "grad_norm": 0.0,
      "learning_rate": 6.843536880250914e-06,
      "loss": 2.281,
      "mean_token_accuracy": 0.7522344322344321,
      "mean_token_accuracy_utility": 0.6645665445665443,
      "step": 1735
    },
    {
      "epoch": 0.6439078545656397,
      "grad_norm": 0.0,
      "learning_rate": 6.7822912962137225e-06,
      "loss": 2.2307,
      "mean_token_accuracy": 0.7457570207570209,
      "mean_token_accuracy_utility": 0.6711355311355309,
      "step": 1740
    },
    {
      "epoch": 0.6457581644925525,
      "grad_norm": 0.0,
      "learning_rate": 6.721180055179113e-06,
      "loss": 2.2719,
      "mean_token_accuracy": 0.7493025754856527,
      "mean_token_accuracy_utility": 0.671196581196581,
      "step": 1745
    },
    {
      "epoch": 0.6476084744194652,
      "grad_norm": 0.0,
      "learning_rate": 6.660205708610987e-06,
      "loss": 2.2449,
      "mean_token_accuracy": 0.7407570207570207,
      "mean_token_accuracy_utility": 0.6794139194139193,
      "step": 1750
    },
    {
      "epoch": 0.649458784346378,
      "grad_norm": 0.0,
      "learning_rate": 6.599370802257755e-06,
      "loss": 2.2331,
      "mean_token_accuracy": 0.7512148962148961,
      "mean_token_accuracy_utility": 0.6736385836385834,
      "step": 1755
    },
    {
      "epoch": 0.6513090942732908,
      "grad_norm": 0.0,
      "learning_rate": 6.5386778760460316e-06,
      "loss": 2.3271,
      "mean_token_accuracy": 0.7625702075702078,
      "mean_token_accuracy_utility": 0.6526129426129426,
      "step": 1760
    },
    {
      "epoch": 0.6531594042002036,
      "grad_norm": 0.0,
      "learning_rate": 6.478129463974598e-06,
      "loss": 2.3739,
      "mean_token_accuracy": 0.7416971916971916,
      "mean_token_accuracy_utility": 0.6553235653235652,
      "step": 1765
    },
    {
      "epoch": 0.6550097141271163,
      "grad_norm": 0.0,
      "learning_rate": 6.417728094008613e-06,
      "loss": 2.1982,
      "mean_token_accuracy": 0.7615995115995117,
      "mean_token_accuracy_utility": 0.6710500610500608,
      "step": 1770
    },
    {
      "epoch": 0.6568600240540291,
      "grad_norm": 0.0,
      "learning_rate": 6.357476287974051e-06,
      "loss": 2.2098,
      "mean_token_accuracy": 0.7501526251526253,
      "mean_token_accuracy_utility": 0.6747252747252745,
      "step": 1775
    },
    {
      "epoch": 0.6587103339809418,
      "grad_norm": 0.0,
      "learning_rate": 6.297376561452428e-06,
      "loss": 2.2446,
      "mean_token_accuracy": 0.7591575091575091,
      "mean_token_accuracy_utility": 0.665506832749631,
      "step": 1780
    },
    {
      "epoch": 0.6605606439078545,
      "grad_norm": 0.0,
      "learning_rate": 6.237431423675764e-06,
      "loss": 2.3132,
      "mean_token_accuracy": 0.7505860805860806,
      "mean_token_accuracy_utility": 0.6606654456654455,
      "step": 1785
    },
    {
      "epoch": 0.6624109538347673,
      "grad_norm": 0.0,
      "learning_rate": 6.177643377421827e-06,
      "loss": 2.3013,
      "mean_token_accuracy": 0.7551159951159949,
      "mean_token_accuracy_utility": 0.657045177045177,
      "step": 1790
    },
    {
      "epoch": 0.66426126376168,
      "grad_norm": 0.0,
      "learning_rate": 6.118014918909633e-06,
      "loss": 2.2911,
      "mean_token_accuracy": 0.7488095238095237,
      "mean_token_accuracy_utility": 0.6606793947596985,
      "step": 1795
    },
    {
      "epoch": 0.6661115736885929,
      "grad_norm": 0.0,
      "learning_rate": 6.058548537695225e-06,
      "loss": 2.293,
      "mean_token_accuracy": 0.7666178266178267,
      "mean_token_accuracy_utility": 0.6554578754578754,
      "step": 1800
    },
    {
      "epoch": 0.6661115736885929,
      "eval_mean_token_accuracy": 0.7397008220537632,
      "eval_reasoning_loss": 0.8998104333877563,
      "eval_reasoning_runtime": 29.3351,
      "eval_reasoning_samples_per_second": 4.397,
      "eval_reasoning_steps_per_second": 1.125,
      "step": 1800
    },
    {
      "epoch": 0.6661115736885929,
      "eval_mean_token_accuracy": 0.6618480239298083,
      "eval_utility_loss": 1.4291177988052368,
      "eval_utility_runtime": 11.0798,
      "eval_utility_samples_per_second": 4.422,
      "eval_utility_steps_per_second": 1.173,
      "step": 1800
    },
    {
      "epoch": 0.6679618836155056,
      "grad_norm": 0.0,
      "learning_rate": 5.999246716567737e-06,
      "loss": 2.2444,
      "mean_token_accuracy": 0.7407692307692305,
      "mean_token_accuracy_utility": 0.6751648351648349,
      "step": 1805
    },
    {
      "epoch": 0.6698121935424184,
      "grad_norm": 0.0,
      "learning_rate": 5.940111931445731e-06,
      "loss": 2.3836,
      "mean_token_accuracy": 0.7327594627594626,
      "mean_token_accuracy_utility": 0.6605555555555553,
      "step": 1810
    },
    {
      "epoch": 0.6716625034693311,
      "grad_norm": 0.0,
      "learning_rate": 5.881146651273825e-06,
      "loss": 2.1994,
      "mean_token_accuracy": 0.7651465201465203,
      "mean_token_accuracy_utility": 0.6678388278388276,
      "step": 1815
    },
    {
      "epoch": 0.6735128133962439,
      "grad_norm": 0.0,
      "learning_rate": 5.822353337919616e-06,
      "loss": 2.2306,
      "mean_token_accuracy": 0.7551892551892554,
      "mean_token_accuracy_utility": 0.6637973137973135,
      "step": 1820
    },
    {
      "epoch": 0.6753631233231566,
      "grad_norm": 0.0,
      "learning_rate": 5.763734446070892e-06,
      "loss": 2.1505,
      "mean_token_accuracy": 0.7681929181929184,
      "mean_token_accuracy_utility": 0.6732295482295482,
      "step": 1825
    },
    {
      "epoch": 0.6772134332500694,
      "grad_norm": 0.0,
      "learning_rate": 5.705292423133133e-06,
      "loss": 2.2727,
      "mean_token_accuracy": 0.768815628815629,
      "mean_token_accuracy_utility": 0.6527045177045175,
      "step": 1830
    },
    {
      "epoch": 0.6790637431769821,
      "grad_norm": 0.0,
      "learning_rate": 5.647029709127355e-06,
      "loss": 2.3546,
      "mean_token_accuracy": 0.7437362637362636,
      "mean_token_accuracy_utility": 0.6512087912087912,
      "step": 1835
    },
    {
      "epoch": 0.6809140531038949,
      "grad_norm": 0.0,
      "learning_rate": 5.5889487365882065e-06,
      "loss": 2.2444,
      "mean_token_accuracy": 0.7570024420024419,
      "mean_token_accuracy_utility": 0.6653296703296702,
      "step": 1840
    },
    {
      "epoch": 0.6827643630308077,
      "grad_norm": 0.0,
      "learning_rate": 5.531051930462437e-06,
      "loss": 2.2738,
      "mean_token_accuracy": 0.7572649572649572,
      "mean_token_accuracy_utility": 0.6627655677655676,
      "step": 1845
    },
    {
      "epoch": 0.6846146729577204,
      "grad_norm": 0.0,
      "learning_rate": 5.4733417080076325e-06,
      "loss": 2.275,
      "mean_token_accuracy": 0.7589560439560441,
      "mean_token_accuracy_utility": 0.6581440781440779,
      "step": 1850
    },
    {
      "epoch": 0.6864649828846332,
      "grad_norm": 0.0,
      "learning_rate": 5.415820478691301e-06,
      "loss": 2.2005,
      "mean_token_accuracy": 0.7586935286935286,
      "mean_token_accuracy_utility": 0.6725702075702074,
      "step": 1855
    },
    {
      "epoch": 0.6883152928115459,
      "grad_norm": 0.0,
      "learning_rate": 5.358490644090263e-06,
      "loss": 2.2994,
      "mean_token_accuracy": 0.7467032967032967,
      "mean_token_accuracy_utility": 0.6619230769230768,
      "step": 1860
    },
    {
      "epoch": 0.6901656027384587,
      "grad_norm": 0.0,
      "learning_rate": 5.3013545977904005e-06,
      "loss": 2.3897,
      "mean_token_accuracy": 0.7386935286935288,
      "mean_token_accuracy_utility": 0.6496336996336994,
      "step": 1865
    },
    {
      "epoch": 0.6920159126653714,
      "grad_norm": 0.0,
      "learning_rate": 5.244414725286717e-06,
      "loss": 2.3579,
      "mean_token_accuracy": 0.7404639804639805,
      "mean_token_accuracy_utility": 0.6600122100122099,
      "step": 1870
    },
    {
      "epoch": 0.6938662225922843,
      "grad_norm": 0.0,
      "learning_rate": 5.187673403883721e-06,
      "loss": 2.2596,
      "mean_token_accuracy": 0.7625213675213677,
      "mean_token_accuracy_utility": 0.659090354090354,
      "step": 1875
    },
    {
      "epoch": 0.695716532519197,
      "grad_norm": 0.0,
      "learning_rate": 5.131133002596199e-06,
      "loss": 2.2752,
      "mean_token_accuracy": 0.7663369963369964,
      "mean_token_accuracy_utility": 0.6523774149445675,
      "step": 1880
    },
    {
      "epoch": 0.6975668424461097,
      "grad_norm": 0.0,
      "learning_rate": 5.074795882050293e-06,
      "loss": 2.2636,
      "mean_token_accuracy": 0.7473992673992674,
      "mean_token_accuracy_utility": 0.6708302808302807,
      "step": 1885
    },
    {
      "epoch": 0.6994171523730225,
      "grad_norm": 0.0,
      "learning_rate": 5.018664394384942e-06,
      "loss": 2.3324,
      "mean_token_accuracy": 0.7616056166056168,
      "mean_token_accuracy_utility": 0.6486324786324784,
      "step": 1890
    },
    {
      "epoch": 0.7012674622999352,
      "grad_norm": 0.0,
      "learning_rate": 4.9627408831536705e-06,
      "loss": 2.2988,
      "mean_token_accuracy": 0.7549633699633702,
      "mean_token_accuracy_utility": 0.6604700854700853,
      "step": 1895
    },
    {
      "epoch": 0.703117772226848,
      "grad_norm": 0.0,
      "learning_rate": 4.907027683226761e-06,
      "loss": 2.2216,
      "mean_token_accuracy": 0.7593930960569331,
      "mean_token_accuracy_utility": 0.6687728937728937,
      "step": 1900
    },
    {
      "epoch": 0.703117772226848,
      "eval_mean_token_accuracy": 0.7399248899248895,
      "eval_reasoning_loss": 0.8996211886405945,
      "eval_reasoning_runtime": 28.917,
      "eval_reasoning_samples_per_second": 4.461,
      "eval_reasoning_steps_per_second": 1.141,
      "step": 1900
    },
    {
      "epoch": 0.703117772226848,
      "eval_mean_token_accuracy": 0.6619856091546112,
      "eval_utility_loss": 1.4288759231567383,
      "eval_utility_runtime": 10.8139,
      "eval_utility_samples_per_second": 4.531,
      "eval_utility_steps_per_second": 1.202,
      "step": 1900
    },
    {
      "epoch": 0.7049680821537607,
      "grad_norm": 0.0,
      "learning_rate": 4.85152712069375e-06,
      "loss": 2.2837,
      "mean_token_accuracy": 0.753785103785104,
      "mean_token_accuracy_utility": 0.6623260073260071,
      "step": 1905
    },
    {
      "epoch": 0.7068183920806735,
      "grad_norm": 0.0,
      "learning_rate": 4.7962415127663265e-06,
      "loss": 2.3197,
      "mean_token_accuracy": 0.7562271062271062,
      "mean_token_accuracy_utility": 0.6564102564102562,
      "step": 1910
    },
    {
      "epoch": 0.7086687020075862,
      "grad_norm": 0.0,
      "learning_rate": 4.74117316768158e-06,
      "loss": 2.3292,
      "mean_token_accuracy": 0.7478266178266177,
      "mean_token_accuracy_utility": 0.6599694749694749,
      "step": 1915
    },
    {
      "epoch": 0.7105190119344991,
      "grad_norm": 0.0,
      "learning_rate": 4.686324384605629e-06,
      "loss": 2.292,
      "mean_token_accuracy": 0.747997557997558,
      "mean_token_accuracy_utility": 0.6648412698412696,
      "step": 1920
    },
    {
      "epoch": 0.7123693218614118,
      "grad_norm": 0.0,
      "learning_rate": 4.631697453537623e-06,
      "loss": 2.3158,
      "mean_token_accuracy": 0.7505433455433456,
      "mean_token_accuracy_utility": 0.6592673992673992,
      "step": 1925
    },
    {
      "epoch": 0.7142196317883246,
      "grad_norm": 0.0,
      "learning_rate": 4.577294655214144e-06,
      "loss": 2.2734,
      "mean_token_accuracy": 0.7464224664224662,
      "mean_token_accuracy_utility": 0.6674668657527112,
      "step": 1930
    },
    {
      "epoch": 0.7160699417152373,
      "grad_norm": 0.0,
      "learning_rate": 4.523118261013969e-06,
      "loss": 2.3228,
      "mean_token_accuracy": 0.7457570207570207,
      "mean_token_accuracy_utility": 0.6625345314407813,
      "step": 1935
    },
    {
      "epoch": 0.71792025164215,
      "grad_norm": 0.0,
      "learning_rate": 4.469170532863254e-06,
      "loss": 2.3306,
      "mean_token_accuracy": 0.7365079365079366,
      "mean_token_accuracy_utility": 0.6666788766788765,
      "step": 1940
    },
    {
      "epoch": 0.7197705615690628,
      "grad_norm": 0.0,
      "learning_rate": 4.415453723141081e-06,
      "loss": 2.4227,
      "mean_token_accuracy": 0.7414529914529915,
      "mean_token_accuracy_utility": 0.6519474969474969,
      "step": 1945
    },
    {
      "epoch": 0.7216208714959755,
      "grad_norm": 0.0,
      "learning_rate": 4.361970074585426e-06,
      "loss": 2.1501,
      "mean_token_accuracy": 0.760854700854701,
      "mean_token_accuracy_utility": 0.6839560439560437,
      "step": 1950
    },
    {
      "epoch": 0.7234711814228884,
      "grad_norm": 0.0,
      "learning_rate": 4.308721820199529e-06,
      "loss": 2.3372,
      "mean_token_accuracy": 0.7446031746031745,
      "mean_token_accuracy_utility": 0.6568009768009768,
      "step": 1955
    },
    {
      "epoch": 0.7253214913498011,
      "grad_norm": 0.0,
      "learning_rate": 4.255711183158635e-06,
      "loss": 2.2423,
      "mean_token_accuracy": 0.7547374847374847,
      "mean_token_accuracy_utility": 0.6689072039072037,
      "step": 1960
    },
    {
      "epoch": 0.7271718012767139,
      "grad_norm": 0.0,
      "learning_rate": 4.2029403767172175e-06,
      "loss": 2.3201,
      "mean_token_accuracy": 0.7517521367521367,
      "mean_token_accuracy_utility": 0.6581746031746031,
      "step": 1965
    },
    {
      "epoch": 0.7290221112036266,
      "grad_norm": 0.0,
      "learning_rate": 4.150411604116531e-06,
      "loss": 2.3402,
      "mean_token_accuracy": 0.7484554334554334,
      "mean_token_accuracy_utility": 0.6543284493284491,
      "step": 1970
    },
    {
      "epoch": 0.7308724211305394,
      "grad_norm": 0.0,
      "learning_rate": 4.098127058492652e-06,
      "loss": 2.23,
      "mean_token_accuracy": 0.7467765567765567,
      "mean_token_accuracy_utility": 0.6737789987789985,
      "step": 1975
    },
    {
      "epoch": 0.7327227310574521,
      "grad_norm": 0.0,
      "learning_rate": 4.0460889227849e-06,
      "loss": 2.306,
      "mean_token_accuracy": 0.7479487179487181,
      "mean_token_accuracy_utility": 0.6618253968253967,
      "step": 1980
    },
    {
      "epoch": 0.7345730409843649,
      "grad_norm": 0.0,
      "learning_rate": 3.9942993696447045e-06,
      "loss": 2.3187,
      "mean_token_accuracy": 0.7530341880341881,
      "mean_token_accuracy_utility": 0.6598229548229546,
      "step": 1985
    },
    {
      "epoch": 0.7364233509112776,
      "grad_norm": 0.0,
      "learning_rate": 3.942760561344877e-06,
      "loss": 2.2232,
      "mean_token_accuracy": 0.7461904761904758,
      "mean_token_accuracy_utility": 0.6770329670329669,
      "step": 1990
    },
    {
      "epoch": 0.7382736608381903,
      "grad_norm": 0.0,
      "learning_rate": 3.891474649689362e-06,
      "loss": 2.293,
      "mean_token_accuracy": 0.7419780219780221,
      "mean_token_accuracy_utility": 0.6670207570207569,
      "step": 1995
    },
    {
      "epoch": 0.7401239707651032,
      "grad_norm": 0.0,
      "learning_rate": 3.840443775923365e-06,
      "loss": 2.2082,
      "mean_token_accuracy": 0.7500610500610503,
      "mean_token_accuracy_utility": 0.6763858363858362,
      "step": 2000
    },
    {
      "epoch": 0.7401239707651032,
      "eval_mean_token_accuracy": 0.7398286898286897,
      "eval_reasoning_loss": 0.8994317054748535,
      "eval_reasoning_runtime": 29.3899,
      "eval_reasoning_samples_per_second": 4.389,
      "eval_reasoning_steps_per_second": 1.123,
      "step": 2000
    },
    {
      "epoch": 0.7401239707651032,
      "eval_mean_token_accuracy": 0.6620137861058649,
      "eval_utility_loss": 1.4288421869277954,
      "eval_utility_runtime": 11.1092,
      "eval_utility_samples_per_second": 4.411,
      "eval_utility_steps_per_second": 1.17,
      "step": 2000
    },
    {
      "epoch": 0.7419742806920159,
      "grad_norm": 0.0,
      "learning_rate": 3.7896700706439826e-06,
      "loss": 2.3686,
      "mean_token_accuracy": 0.7508547008547009,
      "mean_token_accuracy_utility": 0.6483226776287435,
      "step": 2005
    },
    {
      "epoch": 0.7438245906189287,
      "grad_norm": 0.0,
      "learning_rate": 3.7391556537112282e-06,
      "loss": 2.2214,
      "mean_token_accuracy": 0.7592857142857145,
      "mean_token_accuracy_utility": 0.6654761904761904,
      "step": 2010
    },
    {
      "epoch": 0.7456749005458414,
      "grad_norm": 0.0,
      "learning_rate": 3.6889026341595378e-06,
      "loss": 2.2846,
      "mean_token_accuracy": 0.7625457875457877,
      "mean_token_accuracy_utility": 0.6554029304029302,
      "step": 2015
    },
    {
      "epoch": 0.7475252104727542,
      "grad_norm": 0.0,
      "learning_rate": 3.6389131101096953e-06,
      "loss": 2.2816,
      "mean_token_accuracy": 0.7545909645909645,
      "mean_token_accuracy_utility": 0.6585531135531134,
      "step": 2020
    },
    {
      "epoch": 0.7493755203996669,
      "grad_norm": 0.0,
      "learning_rate": 3.5891891686812597e-06,
      "loss": 2.2356,
      "mean_token_accuracy": 0.7429374483020065,
      "mean_token_accuracy_utility": 0.6725803358202255,
      "step": 2025
    },
    {
      "epoch": 0.7512258303265797,
      "grad_norm": 0.0,
      "learning_rate": 3.5397328859054138e-06,
      "loss": 2.2018,
      "mean_token_accuracy": 0.747802197802198,
      "mean_token_accuracy_utility": 0.6813919413919411,
      "step": 2030
    },
    {
      "epoch": 0.7530761402534925,
      "grad_norm": 0.0,
      "learning_rate": 3.490546326638273e-06,
      "loss": 2.3154,
      "mean_token_accuracy": 0.7403706205466756,
      "mean_token_accuracy_utility": 0.6659768009768008,
      "step": 2035
    },
    {
      "epoch": 0.7549264501804053,
      "grad_norm": 0.0,
      "learning_rate": 3.441631544474705e-06,
      "loss": 2.2539,
      "mean_token_accuracy": 0.7663369963369963,
      "mean_token_accuracy_utility": 0.6600793650793649,
      "step": 2040
    },
    {
      "epoch": 0.756776760107318,
      "grad_norm": 0.0,
      "learning_rate": 3.3929905816625653e-06,
      "loss": 2.4234,
      "mean_token_accuracy": 0.7519474969474969,
      "mean_token_accuracy_utility": 0.6320757020757019,
      "step": 2045
    },
    {
      "epoch": 0.7586270700342307,
      "grad_norm": 0.0,
      "learning_rate": 3.344625469017445e-06,
      "loss": 2.237,
      "mean_token_accuracy": 0.7472832722832722,
      "mean_token_accuracy_utility": 0.6783699633699631,
      "step": 2050
    },
    {
      "epoch": 0.7604773799611435,
      "grad_norm": 0.0,
      "learning_rate": 3.2965382258378674e-06,
      "loss": 2.2201,
      "mean_token_accuracy": 0.7670390720390721,
      "mean_token_accuracy_utility": 0.6644200244200242,
      "step": 2055
    },
    {
      "epoch": 0.7623276898880562,
      "grad_norm": 0.0,
      "learning_rate": 3.248730859821002e-06,
      "loss": 2.3168,
      "mean_token_accuracy": 0.7493162393162394,
      "mean_token_accuracy_utility": 0.6635958485958484,
      "step": 2060
    },
    {
      "epoch": 0.764177999814969,
      "grad_norm": 0.0,
      "learning_rate": 3.2012053669788136e-06,
      "loss": 2.2972,
      "mean_token_accuracy": 0.7467399267399267,
      "mean_token_accuracy_utility": 0.6590781440781438,
      "step": 2065
    },
    {
      "epoch": 0.7660283097418817,
      "grad_norm": 0.0,
      "learning_rate": 3.1539637315547524e-06,
      "loss": 2.1553,
      "mean_token_accuracy": 0.7551831501831503,
      "mean_token_accuracy_utility": 0.6828144078144076,
      "step": 2070
    },
    {
      "epoch": 0.7678786196687946,
      "grad_norm": 0.0,
      "learning_rate": 3.1070079259408934e-06,
      "loss": 2.2337,
      "mean_token_accuracy": 0.7452319902319902,
      "mean_token_accuracy_utility": 0.676263736263736,
      "step": 2075
    },
    {
      "epoch": 0.7697289295957073,
      "grad_norm": 0.0,
      "learning_rate": 3.0603399105955966e-06,
      "loss": 2.3723,
      "mean_token_accuracy": 0.7506227106227107,
      "mean_token_accuracy_utility": 0.6535103785103783,
      "step": 2080
    },
    {
      "epoch": 0.7715792395226201,
      "grad_norm": 0.0,
      "learning_rate": 3.0139616339616394e-06,
      "loss": 2.2697,
      "mean_token_accuracy": 0.7542796092796094,
      "mean_token_accuracy_utility": 0.6672486636839627,
      "step": 2085
    },
    {
      "epoch": 0.7734295494495328,
      "grad_norm": 0.0,
      "learning_rate": 2.9678750323848893e-06,
      "loss": 2.2166,
      "mean_token_accuracy": 0.7557387057387058,
      "mean_token_accuracy_utility": 0.6749206349206347,
      "step": 2090
    },
    {
      "epoch": 0.7752798593764455,
      "grad_norm": 0.0,
      "learning_rate": 2.922082030033446e-06,
      "loss": 2.2738,
      "mean_token_accuracy": 0.7394200244200243,
      "mean_token_accuracy_utility": 0.6698079393552643,
      "step": 2095
    },
    {
      "epoch": 0.7771301693033583,
      "grad_norm": 0.0,
      "learning_rate": 2.8765845388172955e-06,
      "loss": 2.2887,
      "mean_token_accuracy": 0.7608485958485959,
      "mean_token_accuracy_utility": 0.6553537566688183,
      "step": 2100
    },
    {
      "epoch": 0.7771301693033583,
      "eval_mean_token_accuracy": 0.7400731077201663,
      "eval_reasoning_loss": 0.8994479179382324,
      "eval_reasoning_runtime": 29.3324,
      "eval_reasoning_samples_per_second": 4.398,
      "eval_reasoning_steps_per_second": 1.125,
      "step": 2100
    },
    {
      "epoch": 0.7771301693033583,
      "eval_mean_token_accuracy": 0.6617306199662504,
      "eval_utility_loss": 1.4288012981414795,
      "eval_utility_runtime": 11.0947,
      "eval_utility_samples_per_second": 4.417,
      "eval_utility_steps_per_second": 1.172,
      "step": 2100
    },
    {
      "epoch": 0.778980479230271,
      "grad_norm": 0.0,
      "learning_rate": 2.831384458308518e-06,
      "loss": 2.2394,
      "mean_token_accuracy": 0.7661538461538463,
      "mean_token_accuracy_utility": 0.6584004884004881,
      "step": 2105
    },
    {
      "epoch": 0.7808307891571838,
      "grad_norm": 0.0,
      "learning_rate": 2.7864836756619407e-06,
      "loss": 2.2988,
      "mean_token_accuracy": 0.7526678876678876,
      "mean_token_accuracy_utility": 0.655982905982906,
      "step": 2110
    },
    {
      "epoch": 0.7826810990840966,
      "grad_norm": 0.0,
      "learning_rate": 2.741884065536373e-06,
      "loss": 2.2746,
      "mean_token_accuracy": 0.7664163614163615,
      "mean_token_accuracy_utility": 0.6539010989010987,
      "step": 2115
    },
    {
      "epoch": 0.7845314090110094,
      "grad_norm": 0.0,
      "learning_rate": 2.6975874900163223e-06,
      "loss": 2.3167,
      "mean_token_accuracy": 0.7515995115995115,
      "mean_token_accuracy_utility": 0.6562820512820511,
      "step": 2120
    },
    {
      "epoch": 0.7863817189379221,
      "grad_norm": 0.0,
      "learning_rate": 2.6535957985342653e-06,
      "loss": 2.2185,
      "mean_token_accuracy": 0.7581440781440782,
      "mean_token_accuracy_utility": 0.6711477411477411,
      "step": 2125
    },
    {
      "epoch": 0.7882320288648349,
      "grad_norm": 0.0,
      "learning_rate": 2.6099108277934105e-06,
      "loss": 2.1986,
      "mean_token_accuracy": 0.7664407814407814,
      "mean_token_accuracy_utility": 0.6669093079783565,
      "step": 2130
    },
    {
      "epoch": 0.7900823387917476,
      "grad_norm": 0.0,
      "learning_rate": 2.5665344016910367e-06,
      "loss": 2.1425,
      "mean_token_accuracy": 0.759023199023199,
      "mean_token_accuracy_utility": 0.6779426129426127,
      "step": 2135
    },
    {
      "epoch": 0.7919326487186604,
      "grad_norm": 0.0,
      "learning_rate": 2.523468331242329e-06,
      "loss": 2.2063,
      "mean_token_accuracy": 0.7515140415140416,
      "mean_token_accuracy_utility": 0.6749694749694748,
      "step": 2140
    },
    {
      "epoch": 0.7937829586455731,
      "grad_norm": 0.0,
      "learning_rate": 2.4807144145047734e-06,
      "loss": 2.2713,
      "mean_token_accuracy": 0.7384859584859584,
      "mean_token_accuracy_utility": 0.6720634920634919,
      "step": 2145
    },
    {
      "epoch": 0.7956332685724858,
      "grad_norm": 0.0,
      "learning_rate": 2.438274436503074e-06,
      "loss": 2.2397,
      "mean_token_accuracy": 0.7666544566544566,
      "mean_token_accuracy_utility": 0.6613208675932494,
      "step": 2150
    },
    {
      "epoch": 0.7974835784993987,
      "grad_norm": 0.0,
      "learning_rate": 2.396150169154644e-06,
      "loss": 2.3663,
      "mean_token_accuracy": 0.7450061050061051,
      "mean_token_accuracy_utility": 0.6537057387057384,
      "step": 2155
    },
    {
      "epoch": 0.7993338884263114,
      "grad_norm": 0.0,
      "learning_rate": 2.3543433711956197e-06,
      "loss": 2.3028,
      "mean_token_accuracy": 0.7480830280830281,
      "mean_token_accuracy_utility": 0.6612881562881561,
      "step": 2160
    },
    {
      "epoch": 0.8011841983532242,
      "grad_norm": 0.0,
      "learning_rate": 2.3128557881074153e-06,
      "loss": 2.2831,
      "mean_token_accuracy": 0.7574542124542124,
      "mean_token_accuracy_utility": 0.6654151404151403,
      "step": 2165
    },
    {
      "epoch": 0.8030345082801369,
      "grad_norm": 0.0,
      "learning_rate": 2.271689152043873e-06,
      "loss": 2.2997,
      "mean_token_accuracy": 0.7505067155067154,
      "mean_token_accuracy_utility": 0.6584615384615383,
      "step": 2170
    },
    {
      "epoch": 0.8048848182070497,
      "grad_norm": 0.0,
      "learning_rate": 2.230845181758928e-06,
      "loss": 2.2977,
      "mean_token_accuracy": 0.7347191697191696,
      "mean_token_accuracy_utility": 0.6737179487179485,
      "step": 2175
    },
    {
      "epoch": 0.8067351281339624,
      "grad_norm": 0.0,
      "learning_rate": 2.1903255825348533e-06,
      "loss": 2.3804,
      "mean_token_accuracy": 0.7593711843711846,
      "mean_token_accuracy_utility": 0.6427960927960926,
      "step": 2180
    },
    {
      "epoch": 0.8085854380608752,
      "grad_norm": 0.0,
      "learning_rate": 2.150132046111054e-06,
      "loss": 2.2272,
      "mean_token_accuracy": 0.7494139194139193,
      "mean_token_accuracy_utility": 0.6737667887667885,
      "step": 2185
    },
    {
      "epoch": 0.810435747987788,
      "grad_norm": 0.0,
      "learning_rate": 2.1102662506134506e-06,
      "loss": 2.3686,
      "mean_token_accuracy": 0.7344627594627593,
      "mean_token_accuracy_utility": 0.6560012210012209,
      "step": 2190
    },
    {
      "epoch": 0.8122860579147008,
      "grad_norm": 0.0,
      "learning_rate": 2.0707298604843964e-06,
      "loss": 2.3091,
      "mean_token_accuracy": 0.7539377289377287,
      "mean_token_accuracy_utility": 0.6572283272283269,
      "step": 2195
    },
    {
      "epoch": 0.8141363678416135,
      "grad_norm": 0.0,
      "learning_rate": 2.03152452641321e-06,
      "loss": 2.2581,
      "mean_token_accuracy": 0.7640964590964592,
      "mean_token_accuracy_utility": 0.656178266178266,
      "step": 2200
    },
    {
      "epoch": 0.8141363678416135,
      "eval_mean_token_accuracy": 0.7398138898138898,
      "eval_reasoning_loss": 0.8994264602661133,
      "eval_reasoning_runtime": 29.3783,
      "eval_reasoning_samples_per_second": 4.391,
      "eval_reasoning_steps_per_second": 1.123,
      "step": 2200
    },
    {
      "epoch": 0.8141363678416135,
      "eval_mean_token_accuracy": 0.6619842125275354,
      "eval_utility_loss": 1.4287770986557007,
      "eval_utility_runtime": 11.0948,
      "eval_utility_samples_per_second": 4.416,
      "eval_utility_steps_per_second": 1.172,
      "step": 2200
    },
    {
      "epoch": 0.8159866777685262,
      "grad_norm": 0.0,
      "learning_rate": 1.9926518852672294e-06,
      "loss": 2.2139,
      "mean_token_accuracy": 0.7505799755799757,
      "mean_token_accuracy_utility": 0.6744139194139193,
      "step": 2205
    },
    {
      "epoch": 0.817836987695439,
      "grad_norm": 0.0,
      "learning_rate": 1.9541135600234917e-06,
      "loss": 2.227,
      "mean_token_accuracy": 0.7606776556776558,
      "mean_token_accuracy_utility": 0.6705189255189253,
      "step": 2210
    },
    {
      "epoch": 0.8196872976223517,
      "grad_norm": 0.0,
      "learning_rate": 1.9159111597009584e-06,
      "loss": 2.2952,
      "mean_token_accuracy": 0.7595726495726496,
      "mean_token_accuracy_utility": 0.6565934065934064,
      "step": 2215
    },
    {
      "epoch": 0.8215376075492645,
      "grad_norm": 0.0,
      "learning_rate": 1.8780462792933473e-06,
      "loss": 2.324,
      "mean_token_accuracy": 0.7610744810744811,
      "mean_token_accuracy_utility": 0.6467582417582417,
      "step": 2220
    },
    {
      "epoch": 0.8233879174761772,
      "grad_norm": 0.0,
      "learning_rate": 1.8405204997025394e-06,
      "loss": 2.3546,
      "mean_token_accuracy": 0.7304945054945053,
      "mean_token_accuracy_utility": 0.6607875457875456,
      "step": 2225
    },
    {
      "epoch": 0.82523822740309,
      "grad_norm": 0.0,
      "learning_rate": 1.8033353876725578e-06,
      "loss": 2.3003,
      "mean_token_accuracy": 0.7351648351648349,
      "mean_token_accuracy_utility": 0.6728205128205127,
      "step": 2230
    },
    {
      "epoch": 0.8270885373300028,
      "grad_norm": 0.0,
      "learning_rate": 1.766492495724178e-06,
      "loss": 2.2881,
      "mean_token_accuracy": 0.7581196581196582,
      "mean_token_accuracy_utility": 0.6567460317460315,
      "step": 2235
    },
    {
      "epoch": 0.8289388472569156,
      "grad_norm": 0.0,
      "learning_rate": 1.7299933620900945e-06,
      "loss": 2.2675,
      "mean_token_accuracy": 0.7465201465201465,
      "mean_token_accuracy_utility": 0.6678449328449327,
      "step": 2240
    },
    {
      "epoch": 0.8307891571838283,
      "grad_norm": 0.0,
      "learning_rate": 1.6938395106507034e-06,
      "loss": 2.1656,
      "mean_token_accuracy": 0.7748107448107449,
      "mean_token_accuracy_utility": 0.6691147741147738,
      "step": 2245
    },
    {
      "epoch": 0.832639467110741,
      "grad_norm": 0.0,
      "learning_rate": 1.658032450870467e-06,
      "loss": 2.2901,
      "mean_token_accuracy": 0.754188034188034,
      "mean_token_accuracy_utility": 0.6593589743589743,
      "step": 2250
    },
    {
      "epoch": 0.8344897770376538,
      "grad_norm": 0.0,
      "learning_rate": 1.622573677734911e-06,
      "loss": 2.2392,
      "mean_token_accuracy": 0.7634249084249084,
      "mean_token_accuracy_utility": 0.6624297924297923,
      "step": 2255
    },
    {
      "epoch": 0.8363400869645665,
      "grad_norm": 0.0,
      "learning_rate": 1.587464671688187e-06,
      "loss": 2.2142,
      "mean_token_accuracy": 0.7453296703296702,
      "mean_token_accuracy_utility": 0.677203907203907,
      "step": 2260
    },
    {
      "epoch": 0.8381903968914793,
      "grad_norm": 0.0,
      "learning_rate": 1.552706898571288e-06,
      "loss": 2.2723,
      "mean_token_accuracy": 0.7430647130647131,
      "mean_token_accuracy_utility": 0.6689941134097975,
      "step": 2265
    },
    {
      "epoch": 0.840040706818392,
      "grad_norm": 0.0,
      "learning_rate": 1.5183018095608138e-06,
      "loss": 2.2822,
      "mean_token_accuracy": 0.7460736633778631,
      "mean_token_accuracy_utility": 0.6662184937014203,
      "step": 2270
    },
    {
      "epoch": 0.8418910167453049,
      "grad_norm": 0.0,
      "learning_rate": 1.4842508411084145e-06,
      "loss": 2.1612,
      "mean_token_accuracy": 0.7613614163614165,
      "mean_token_accuracy_utility": 0.6850915750915749,
      "step": 2275
    },
    {
      "epoch": 0.8437413266722176,
      "grad_norm": 0.0,
      "learning_rate": 1.4505554148807954e-06,
      "loss": 2.4233,
      "mean_token_accuracy": 0.7430769230769231,
      "mean_token_accuracy_utility": 0.6580402930402929,
      "step": 2280
    },
    {
      "epoch": 0.8455916365991304,
      "grad_norm": 0.0,
      "learning_rate": 1.4172169377003775e-06,
      "loss": 2.2481,
      "mean_token_accuracy": 0.752014652014652,
      "mean_token_accuracy_utility": 0.6766666666666665,
      "step": 2285
    },
    {
      "epoch": 0.8474419465260431,
      "grad_norm": 0.0,
      "learning_rate": 1.3842368014865414e-06,
      "loss": 2.2582,
      "mean_token_accuracy": 0.7567948717948717,
      "mean_token_accuracy_utility": 0.6702188221381091,
      "step": 2290
    },
    {
      "epoch": 0.8492922564529559,
      "grad_norm": 0.0,
      "learning_rate": 1.3516163831975337e-06,
      "loss": 2.2908,
      "mean_token_accuracy": 0.7575030525030525,
      "mean_token_accuracy_utility": 0.659084249084249,
      "step": 2295
    },
    {
      "epoch": 0.8511425663798686,
      "grad_norm": 0.0,
      "learning_rate": 1.3193570447729642e-06,
      "loss": 2.2929,
      "mean_token_accuracy": 0.7446886446886445,
      "mean_token_accuracy_utility": 0.6662578370095754,
      "step": 2300
    },
    {
      "epoch": 0.8511425663798686,
      "eval_mean_token_accuracy": 0.74000999000999,
      "eval_reasoning_loss": 0.8993231058120728,
      "eval_reasoning_runtime": 29.1733,
      "eval_reasoning_samples_per_second": 4.422,
      "eval_reasoning_steps_per_second": 1.131,
      "step": 2300
    },
    {
      "epoch": 0.8511425663798686,
      "eval_mean_token_accuracy": 0.6619043778323158,
      "eval_utility_loss": 1.4287939071655273,
      "eval_utility_runtime": 10.8812,
      "eval_utility_samples_per_second": 4.503,
      "eval_utility_steps_per_second": 1.195,
      "step": 2300
    },
    {
      "epoch": 0.8529928763067813,
      "grad_norm": 0.0,
      "learning_rate": 1.2874601330769488e-06,
      "loss": 2.3174,
      "mean_token_accuracy": 0.7414285714285713,
      "mean_token_accuracy_utility": 0.6647728658419136,
      "step": 2305
    },
    {
      "epoch": 0.8548431862336942,
      "grad_norm": 0.0,
      "learning_rate": 1.255926979841876e-06,
      "loss": 2.2759,
      "mean_token_accuracy": 0.7596398046398045,
      "mean_token_accuracy_utility": 0.6614163614163613,
      "step": 2310
    },
    {
      "epoch": 0.8566934961606069,
      "grad_norm": 0.0,
      "learning_rate": 1.224758901612796e-06,
      "loss": 2.3207,
      "mean_token_accuracy": 0.7554884004884007,
      "mean_token_accuracy_utility": 0.654163614163614,
      "step": 2315
    },
    {
      "epoch": 0.8585438060875197,
      "grad_norm": 0.0,
      "learning_rate": 1.1939571996924738e-06,
      "loss": 2.2717,
      "mean_token_accuracy": 0.7642246642246643,
      "mean_token_accuracy_utility": 0.6599328449328448,
      "step": 2320
    },
    {
      "epoch": 0.8603941160144324,
      "grad_norm": 0.0,
      "learning_rate": 1.1635231600870334e-06,
      "loss": 2.3426,
      "mean_token_accuracy": 0.742020757020757,
      "mean_token_accuracy_utility": 0.653156288156288,
      "step": 2325
    },
    {
      "epoch": 0.8622444259413452,
      "grad_norm": 0.0,
      "learning_rate": 1.1334580534522932e-06,
      "loss": 2.2573,
      "mean_token_accuracy": 0.7509829059829058,
      "mean_token_accuracy_utility": 0.668278388278388,
      "step": 2330
    },
    {
      "epoch": 0.8640947358682579,
      "grad_norm": 0.0,
      "learning_rate": 1.1037631350406874e-06,
      "loss": 2.2368,
      "mean_token_accuracy": 0.7784004884004884,
      "mean_token_accuracy_utility": 0.6527953026812584,
      "step": 2335
    },
    {
      "epoch": 0.8659450457951707,
      "grad_norm": 0.0,
      "learning_rate": 1.0744396446488781e-06,
      "loss": 2.3101,
      "mean_token_accuracy": 0.7480769230769233,
      "mean_token_accuracy_utility": 0.6611416361416359,
      "step": 2340
    },
    {
      "epoch": 0.8677953557220834,
      "grad_norm": 0.0,
      "learning_rate": 1.0454888065659775e-06,
      "loss": 2.3352,
      "mean_token_accuracy": 0.7361965811965812,
      "mean_token_accuracy_utility": 0.6597374847374845,
      "step": 2345
    },
    {
      "epoch": 0.8696456656489963,
      "grad_norm": 0.0,
      "learning_rate": 1.0169118295224488e-06,
      "loss": 2.4115,
      "mean_token_accuracy": 0.7417460317460316,
      "mean_token_accuracy_utility": 0.6498962148962149,
      "step": 2350
    },
    {
      "epoch": 0.871495975575909,
      "grad_norm": 0.0,
      "learning_rate": 9.887099066396178e-07,
      "loss": 2.3149,
      "mean_token_accuracy": 0.7412087912087911,
      "mean_token_accuracy_utility": 0.6617338217338216,
      "step": 2355
    },
    {
      "epoch": 0.8733462855028217,
      "grad_norm": 0.0,
      "learning_rate": 9.608842153798903e-07,
      "loss": 2.3647,
      "mean_token_accuracy": 0.7344078144078143,
      "mean_token_accuracy_utility": 0.6638411313658836,
      "step": 2360
    },
    {
      "epoch": 0.8751965954297345,
      "grad_norm": 0.0,
      "learning_rate": 9.33435917497556e-07,
      "loss": 2.2358,
      "mean_token_accuracy": 0.7551953601953602,
      "mean_token_accuracy_utility": 0.6680525030525029,
      "step": 2365
    },
    {
      "epoch": 0.8770469053566472,
      "grad_norm": 0.0,
      "learning_rate": 9.063661589903116e-07,
      "loss": 2.2427,
      "mean_token_accuracy": 0.7527716727716729,
      "mean_token_accuracy_utility": 0.6683940507348027,
      "step": 2370
    },
    {
      "epoch": 0.87889721528356,
      "grad_norm": 0.0,
      "learning_rate": 8.796760700513984e-07,
      "loss": 2.2547,
      "mean_token_accuracy": 0.738107448107448,
      "mean_token_accuracy_utility": 0.6815626040626039,
      "step": 2375
    },
    {
      "epoch": 0.8807475252104727,
      "grad_norm": 0.0,
      "learning_rate": 8.533667650224253e-07,
      "loss": 2.2948,
      "mean_token_accuracy": 0.752002442002442,
      "mean_token_accuracy_utility": 0.6607264957264956,
      "step": 2380
    },
    {
      "epoch": 0.8825978351373855,
      "grad_norm": 0.0,
      "learning_rate": 8.274393423468385e-07,
      "loss": 2.385,
      "mean_token_accuracy": 0.7360805860805859,
      "mean_token_accuracy_utility": 0.6580586080586077,
      "step": 2385
    },
    {
      "epoch": 0.8844481450642983,
      "grad_norm": 0.0,
      "learning_rate": 8.018948845240538e-07,
      "loss": 2.3518,
      "mean_token_accuracy": 0.7453968253968254,
      "mean_token_accuracy_utility": 0.6559096459096457,
      "step": 2390
    },
    {
      "epoch": 0.8862984549912111,
      "grad_norm": 0.0,
      "learning_rate": 7.767344580642821e-07,
      "loss": 2.3408,
      "mean_token_accuracy": 0.7547863247863249,
      "mean_token_accuracy_utility": 0.652081807081807,
      "step": 2395
    },
    {
      "epoch": 0.8881487649181238,
      "grad_norm": 0.0,
      "learning_rate": 7.519591134439753e-07,
      "loss": 2.3213,
      "mean_token_accuracy": 0.7533394383394384,
      "mean_token_accuracy_utility": 0.6570024420024418,
      "step": 2400
    },
    {
      "epoch": 0.8881487649181238,
      "eval_mean_token_accuracy": 0.7397267220796631,
      "eval_reasoning_loss": 0.8993381261825562,
      "eval_reasoning_runtime": 29.3691,
      "eval_reasoning_samples_per_second": 4.392,
      "eval_reasoning_steps_per_second": 1.124,
      "step": 2400
    },
    {
      "epoch": 0.8881487649181238,
      "eval_mean_token_accuracy": 0.6620781356983815,
      "eval_utility_loss": 1.4288201332092285,
      "eval_utility_runtime": 11.0928,
      "eval_utility_samples_per_second": 4.417,
      "eval_utility_steps_per_second": 1.172,
      "step": 2400
    },
    {
      "epoch": 0.8899990748450366,
      "grad_norm": 0.0,
      "learning_rate": 7.275698850619861e-07,
      "loss": 2.3118,
      "mean_token_accuracy": 0.7552075702075702,
      "mean_token_accuracy_utility": 0.6567826617826615,
      "step": 2405
    },
    {
      "epoch": 0.8918493847719493,
      "grad_norm": 0.0,
      "learning_rate": 7.035677911963712e-07,
      "loss": 2.2082,
      "mean_token_accuracy": 0.7634004884004885,
      "mean_token_accuracy_utility": 0.6713247863247862,
      "step": 2410
    },
    {
      "epoch": 0.893699694698862,
      "grad_norm": 0.0,
      "learning_rate": 6.799538339618838e-07,
      "loss": 2.2137,
      "mean_token_accuracy": 0.7457448107448105,
      "mean_token_accuracy_utility": 0.6789560439560438,
      "step": 2415
    },
    {
      "epoch": 0.8955500046257748,
      "grad_norm": 0.0,
      "learning_rate": 6.567289992681258e-07,
      "loss": 2.2683,
      "mean_token_accuracy": 0.7573443223443224,
      "mean_token_accuracy_utility": 0.6596153846153845,
      "step": 2420
    },
    {
      "epoch": 0.8974003145526875,
      "grad_norm": 0.0,
      "learning_rate": 6.33894256778399e-07,
      "loss": 2.3471,
      "mean_token_accuracy": 0.746214896214896,
      "mean_token_accuracy_utility": 0.6555067155067154,
      "step": 2425
    },
    {
      "epoch": 0.8992506244796004,
      "grad_norm": 0.0,
      "learning_rate": 6.114505598692011e-07,
      "loss": 2.3233,
      "mean_token_accuracy": 0.7486874236874237,
      "mean_token_accuracy_utility": 0.6556410256410254,
      "step": 2430
    },
    {
      "epoch": 0.9011009344065131,
      "grad_norm": 0.0,
      "learning_rate": 5.893988455904387e-07,
      "loss": 2.2261,
      "mean_token_accuracy": 0.7630097680097682,
      "mean_token_accuracy_utility": 0.6643115856958768,
      "step": 2435
    },
    {
      "epoch": 0.9029512443334259,
      "grad_norm": 0.0,
      "learning_rate": 5.677400346262918e-07,
      "loss": 2.3002,
      "mean_token_accuracy": 0.7579670329670333,
      "mean_token_accuracy_utility": 0.656178266178266,
      "step": 2440
    },
    {
      "epoch": 0.9048015542603386,
      "grad_norm": 0.0,
      "learning_rate": 5.464750312567835e-07,
      "loss": 2.1402,
      "mean_token_accuracy": 0.7759279609279611,
      "mean_token_accuracy_utility": 0.6743410574355363,
      "step": 2445
    },
    {
      "epoch": 0.9066518641872514,
      "grad_norm": 0.0,
      "learning_rate": 5.256047233200201e-07,
      "loss": 2.3067,
      "mean_token_accuracy": 0.7514346764346765,
      "mean_token_accuracy_utility": 0.6583577533577532,
      "step": 2450
    },
    {
      "epoch": 0.9085021741141641,
      "grad_norm": 0.0,
      "learning_rate": 5.051299821751254e-07,
      "loss": 2.1758,
      "mean_token_accuracy": 0.7480097680097682,
      "mean_token_accuracy_utility": 0.6856288156288154,
      "step": 2455
    },
    {
      "epoch": 0.9103524840410768,
      "grad_norm": 0.0,
      "learning_rate": 4.850516626658585e-07,
      "loss": 2.3627,
      "mean_token_accuracy": 0.7398290598290598,
      "mean_token_accuracy_utility": 0.662167277167277,
      "step": 2460
    },
    {
      "epoch": 0.9122027939679896,
      "grad_norm": 0.0,
      "learning_rate": 4.653706030849214e-07,
      "loss": 2.2734,
      "mean_token_accuracy": 0.7510989010989011,
      "mean_token_accuracy_utility": 0.6656105006105004,
      "step": 2465
    },
    {
      "epoch": 0.9140531038949024,
      "grad_norm": 0.0,
      "learning_rate": 4.4608762513896455e-07,
      "loss": 2.3149,
      "mean_token_accuracy": 0.7344932844932843,
      "mean_token_accuracy_utility": 0.669188034188034,
      "step": 2470
    },
    {
      "epoch": 0.9159034138218152,
      "grad_norm": 0.0,
      "learning_rate": 4.2720353391427547e-07,
      "loss": 2.2289,
      "mean_token_accuracy": 0.7567094017094018,
      "mean_token_accuracy_utility": 0.6635965544484813,
      "step": 2475
    },
    {
      "epoch": 0.9177537237487279,
      "grad_norm": 0.0,
      "learning_rate": 4.087191178431682e-07,
      "loss": 2.2257,
      "mean_token_accuracy": 0.746819291819292,
      "mean_token_accuracy_utility": 0.6742124542124541,
      "step": 2480
    },
    {
      "epoch": 0.9196040336756407,
      "grad_norm": 0.0,
      "learning_rate": 3.9063514867105914e-07,
      "loss": 2.3834,
      "mean_token_accuracy": 0.7292735042735041,
      "mean_token_accuracy_utility": 0.6636446886446885,
      "step": 2485
    },
    {
      "epoch": 0.9214543436025534,
      "grad_norm": 0.0,
      "learning_rate": 3.729523814242608e-07,
      "loss": 2.2982,
      "mean_token_accuracy": 0.749078144078144,
      "mean_token_accuracy_utility": 0.6620390720390719,
      "step": 2490
    },
    {
      "epoch": 0.9233046535294662,
      "grad_norm": 0.0,
      "learning_rate": 3.5567155437843725e-07,
      "loss": 2.1864,
      "mean_token_accuracy": 0.7579731379731379,
      "mean_token_accuracy_utility": 0.6765750915750914,
      "step": 2495
    },
    {
      "epoch": 0.9251549634563789,
      "grad_norm": 0.0,
      "learning_rate": 3.3879338902779945e-07,
      "loss": 2.2504,
      "mean_token_accuracy": 0.7681684981684981,
      "mean_token_accuracy_utility": 0.6516666666666665,
      "step": 2500
    },
    {
      "epoch": 0.9251549634563789,
      "eval_mean_token_accuracy": 0.7398469721999134,
      "eval_reasoning_loss": 0.8993061780929565,
      "eval_reasoning_runtime": 29.0283,
      "eval_reasoning_samples_per_second": 4.444,
      "eval_reasoning_steps_per_second": 1.137,
      "step": 2500
    },
    {
      "epoch": 0.9251549634563789,
      "eval_mean_token_accuracy": 0.6617949695587672,
      "eval_utility_loss": 1.428654670715332,
      "eval_utility_runtime": 10.9198,
      "eval_utility_samples_per_second": 4.487,
      "eval_utility_steps_per_second": 1.191,
      "step": 2500
    },
    {
      "epoch": 0.9270052733832918,
      "grad_norm": 0.0,
      "learning_rate": 3.223185900549686e-07,
      "loss": 2.2201,
      "mean_token_accuracy": 0.7464163614163611,
      "mean_token_accuracy_utility": 0.6797802197802196,
      "step": 2505
    },
    {
      "epoch": 0.9288555833102045,
      "grad_norm": 0.0,
      "learning_rate": 3.0624784530156384e-07,
      "loss": 2.3454,
      "mean_token_accuracy": 0.7486263736263735,
      "mean_token_accuracy_utility": 0.6567032967032965,
      "step": 2510
    },
    {
      "epoch": 0.9307058932371172,
      "grad_norm": 0.0,
      "learning_rate": 2.905818257394799e-07,
      "loss": 2.1133,
      "mean_token_accuracy": 0.7655311355311356,
      "mean_token_accuracy_utility": 0.688009768009768,
      "step": 2515
    },
    {
      "epoch": 0.93255620316403,
      "grad_norm": 0.0,
      "learning_rate": 2.753211854428728e-07,
      "loss": 2.2232,
      "mean_token_accuracy": 0.7642735042735044,
      "mean_token_accuracy_utility": 0.6618742368742366,
      "step": 2520
    },
    {
      "epoch": 0.9344065130909427,
      "grad_norm": 0.0,
      "learning_rate": 2.604665615608526e-07,
      "loss": 2.3759,
      "mean_token_accuracy": 0.7497096057409374,
      "mean_token_accuracy_utility": 0.6477885315290037,
      "step": 2525
    },
    {
      "epoch": 0.9362568230178555,
      "grad_norm": 0.0,
      "learning_rate": 2.460185742908816e-07,
      "loss": 2.2691,
      "mean_token_accuracy": 0.738015873015873,
      "mean_token_accuracy_utility": 0.6721123321123319,
      "step": 2530
    },
    {
      "epoch": 0.9381071329447682,
      "grad_norm": 0.0,
      "learning_rate": 2.3197782685288385e-07,
      "loss": 2.3434,
      "mean_token_accuracy": 0.7459157509157508,
      "mean_token_accuracy_utility": 0.654139194139194,
      "step": 2535
    },
    {
      "epoch": 0.939957442871681,
      "grad_norm": 0.0,
      "learning_rate": 2.1834490546405186e-07,
      "loss": 2.2703,
      "mean_token_accuracy": 0.7497435897435896,
      "mean_token_accuracy_utility": 0.6676168784350843,
      "step": 2540
    },
    {
      "epoch": 0.9418077527985937,
      "grad_norm": 0.0,
      "learning_rate": 2.0512037931437855e-07,
      "loss": 2.3566,
      "mean_token_accuracy": 0.7428327228327228,
      "mean_token_accuracy_utility": 0.6556105006105004,
      "step": 2545
    },
    {
      "epoch": 0.9436580627255066,
      "grad_norm": 0.0,
      "learning_rate": 1.9230480054288958e-07,
      "loss": 2.2313,
      "mean_token_accuracy": 0.7680891330891331,
      "mean_token_accuracy_utility": 0.6617094017094016,
      "step": 2550
    },
    {
      "epoch": 0.9455083726524193,
      "grad_norm": 0.0,
      "learning_rate": 1.7989870421459498e-07,
      "loss": 2.2016,
      "mean_token_accuracy": 0.7593040293040294,
      "mean_token_accuracy_utility": 0.6722090662887863,
      "step": 2555
    },
    {
      "epoch": 0.9473586825793321,
      "grad_norm": 0.0,
      "learning_rate": 1.6790260829814053e-07,
      "loss": 2.212,
      "mean_token_accuracy": 0.7432783882783882,
      "mean_token_accuracy_utility": 0.6853724053724052,
      "step": 2560
    },
    {
      "epoch": 0.9492089925062448,
      "grad_norm": 0.0,
      "learning_rate": 1.5631701364419492e-07,
      "loss": 2.095,
      "mean_token_accuracy": 0.776068376068376,
      "mean_token_accuracy_utility": 0.6827289377289375,
      "step": 2565
    },
    {
      "epoch": 0.9510593024331575,
      "grad_norm": 0.0,
      "learning_rate": 1.4514240396452438e-07,
      "loss": 2.3291,
      "mean_token_accuracy": 0.7581440781440782,
      "mean_token_accuracy_utility": 0.6518192918192917,
      "step": 2570
    },
    {
      "epoch": 0.9529096123600703,
      "grad_norm": 0.0,
      "learning_rate": 1.3437924581181205e-07,
      "loss": 2.1923,
      "mean_token_accuracy": 0.7684798534798538,
      "mean_token_accuracy_utility": 0.6659523809523809,
      "step": 2575
    },
    {
      "epoch": 0.954759922286983,
      "grad_norm": 0.0,
      "learning_rate": 1.2402798856016474e-07,
      "loss": 2.2955,
      "mean_token_accuracy": 0.7370390720390719,
      "mean_token_accuracy_utility": 0.6734676434676434,
      "step": 2580
    },
    {
      "epoch": 0.9566102322138959,
      "grad_norm": 0.0,
      "learning_rate": 1.1408906438636236e-07,
      "loss": 2.3725,
      "mean_token_accuracy": 0.7595665445665446,
      "mean_token_accuracy_utility": 0.6459401709401709,
      "step": 2585
    },
    {
      "epoch": 0.9584605421408086,
      "grad_norm": 0.0,
      "learning_rate": 1.045628882518046e-07,
      "loss": 2.2056,
      "mean_token_accuracy": 0.7626129426129429,
      "mean_token_accuracy_utility": 0.6693772893772891,
      "step": 2590
    },
    {
      "epoch": 0.9603108520677214,
      "grad_norm": 0.0,
      "learning_rate": 9.544985788519589e-08,
      "loss": 2.2799,
      "mean_token_accuracy": 0.7408119658119658,
      "mean_token_accuracy_utility": 0.6775903624316018,
      "step": 2595
    },
    {
      "epoch": 0.9621611619946341,
      "grad_norm": 0.0,
      "learning_rate": 8.675035376593088e-08,
      "loss": 2.3295,
      "mean_token_accuracy": 0.7515079365079365,
      "mean_token_accuracy_utility": 0.6529792429792428,
      "step": 2600
    },
    {
      "epoch": 0.9621611619946341,
      "eval_mean_token_accuracy": 0.7397748221277635,
      "eval_reasoning_loss": 0.8993598818778992,
      "eval_reasoning_runtime": 29.3659,
      "eval_reasoning_samples_per_second": 4.393,
      "eval_reasoning_steps_per_second": 1.124,
      "step": 2600
    },
    {
      "epoch": 0.9621611619946341,
      "eval_mean_token_accuracy": 0.6621030131181691,
      "eval_utility_loss": 1.4287171363830566,
      "eval_utility_runtime": 11.0998,
      "eval_utility_samples_per_second": 4.414,
      "eval_utility_steps_per_second": 1.171,
      "step": 2600
    },
    {
      "epoch": 0.9640114719215469,
      "grad_norm": 0.0,
      "learning_rate": 7.846473910821162e-08,
      "loss": 2.2486,
      "mean_token_accuracy": 0.7471550671550672,
      "mean_token_accuracy_utility": 0.6714166712039595,
      "step": 2605
    },
    {
      "epoch": 0.9658617818484596,
      "grad_norm": 0.0,
      "learning_rate": 7.059335984588634e-08,
      "loss": 2.2628,
      "mean_token_accuracy": 0.7706227106227107,
      "mean_token_accuracy_utility": 0.6527497057699296,
      "step": 2610
    },
    {
      "epoch": 0.9677120917753724,
      "grad_norm": 0.0,
      "learning_rate": 6.313654461800322e-08,
      "loss": 2.2989,
      "mean_token_accuracy": 0.7343101343101341,
      "mean_token_accuracy_utility": 0.6753601953601951,
      "step": 2615
    },
    {
      "epoch": 0.9695624017022851,
      "grad_norm": 0.0,
      "learning_rate": 5.609460475509032e-08,
      "loss": 2.2027,
      "mean_token_accuracy": 0.7615140415140415,
      "mean_token_accuracy_utility": 0.6656532356532354,
      "step": 2620
    },
    {
      "epoch": 0.9714127116291978,
      "grad_norm": 0.0,
      "learning_rate": 4.9467834266154756e-08,
      "loss": 2.2567,
      "mean_token_accuracy": 0.7432905982905985,
      "mean_token_accuracy_utility": 0.6723015873015872,
      "step": 2625
    },
    {
      "epoch": 0.9732630215561107,
      "grad_norm": 0.0,
      "learning_rate": 4.325650982641039e-08,
      "loss": 2.3557,
      "mean_token_accuracy": 0.748919413919414,
      "mean_token_accuracy_utility": 0.6597416042977706,
      "step": 2630
    },
    {
      "epoch": 0.9751133314830234,
      "grad_norm": 0.0,
      "learning_rate": 3.746089076572701e-08,
      "loss": 2.3918,
      "mean_token_accuracy": 0.755042735042735,
      "mean_token_accuracy_utility": 0.6394993894993893,
      "step": 2635
    },
    {
      "epoch": 0.9769636414099362,
      "grad_norm": 0.0,
      "learning_rate": 3.208121905779904e-08,
      "loss": 2.221,
      "mean_token_accuracy": 0.7664835164835166,
      "mean_token_accuracy_utility": 0.6612881562881562,
      "step": 2640
    },
    {
      "epoch": 0.9788139513368489,
      "grad_norm": 0.0,
      "learning_rate": 2.711771931004692e-08,
      "loss": 2.3136,
      "mean_token_accuracy": 0.7556288156288155,
      "mean_token_accuracy_utility": 0.6547557997557997,
      "step": 2645
    },
    {
      "epoch": 0.9806642612637617,
      "grad_norm": 0.0,
      "learning_rate": 2.257059875423795e-08,
      "loss": 2.3479,
      "mean_token_accuracy": 0.7338400488400489,
      "mean_token_accuracy_utility": 0.666221001221001,
      "step": 2650
    },
    {
      "epoch": 0.9825145711906744,
      "grad_norm": 0.0,
      "learning_rate": 1.8440047237832105e-08,
      "loss": 2.1329,
      "mean_token_accuracy": 0.7772710622710622,
      "mean_token_accuracy_utility": 0.6685347985347985,
      "step": 2655
    },
    {
      "epoch": 0.9843648811175872,
      "grad_norm": 0.0,
      "learning_rate": 1.472623721606059e-08,
      "loss": 2.2234,
      "mean_token_accuracy": 0.7575518925518927,
      "mean_token_accuracy_utility": 0.6747252747252744,
      "step": 2660
    },
    {
      "epoch": 0.9862151910445,
      "grad_norm": 0.0,
      "learning_rate": 1.1429323744720499e-08,
      "loss": 2.2785,
      "mean_token_accuracy": 0.7536202686202687,
      "mean_token_accuracy_utility": 0.6605372405372404,
      "step": 2665
    },
    {
      "epoch": 0.9880655009714127,
      "grad_norm": 0.0,
      "learning_rate": 8.549444473702207e-09,
      "loss": 2.3606,
      "mean_token_accuracy": 0.7369841269841269,
      "mean_token_accuracy_utility": 0.6568925518925518,
      "step": 2670
    },
    {
      "epoch": 0.9899158108983255,
      "grad_norm": 0.0,
      "learning_rate": 6.086719641246186e-09,
      "loss": 2.1985,
      "mean_token_accuracy": 0.7687423687423689,
      "mean_token_accuracy_utility": 0.6650183150183147,
      "step": 2675
    },
    {
      "epoch": 0.9917661208252382,
      "grad_norm": 0.0,
      "learning_rate": 4.041252068918145e-09,
      "loss": 2.2288,
      "mean_token_accuracy": 0.765006105006105,
      "mean_token_accuracy_utility": 0.660115995115995,
      "step": 2680
    },
    {
      "epoch": 0.993616430752151,
      "grad_norm": 0.0,
      "learning_rate": 2.4131271573191172e-09,
      "loss": 2.4015,
      "mean_token_accuracy": 0.7568192918192919,
      "mean_token_accuracy_utility": 0.6437851037851037,
      "step": 2685
    },
    {
      "epoch": 0.9954667406790637,
      "grad_norm": 0.0,
      "learning_rate": 1.2024128825172121e-09,
      "loss": 2.1219,
      "mean_token_accuracy": 0.7567582417582417,
      "mean_token_accuracy_utility": 0.6870085470085467,
      "step": 2690
    },
    {
      "epoch": 0.9973170506059765,
      "grad_norm": 0.0,
      "learning_rate": 4.0915979321320967e-10,
      "loss": 2.1901,
      "mean_token_accuracy": 0.7490659340659339,
      "mean_token_accuracy_utility": 0.6753296703296702,
      "step": 2695
    },
    {
      "epoch": 0.9991673605328892,
      "grad_norm": 0.0,
      "learning_rate": 3.3401008625588706e-11,
      "loss": 2.3192,
      "mean_token_accuracy": 0.7562637362637364,
      "mean_token_accuracy_utility": 0.6532905982905981,
      "step": 2700
    },
    {
      "epoch": 0.9991673605328892,
      "eval_mean_token_accuracy": 0.7398192221721633,
      "eval_reasoning_loss": 0.8993186950683594,
      "eval_reasoning_runtime": 29.3694,
      "eval_reasoning_samples_per_second": 4.392,
      "eval_reasoning_steps_per_second": 1.124,
      "step": 2700
    },
    {
      "epoch": 0.9991673605328892,
      "eval_mean_token_accuracy": 0.6619668245204421,
      "eval_utility_loss": 1.4286295175552368,
      "eval_utility_runtime": 11.0962,
      "eval_utility_samples_per_second": 4.416,
      "eval_utility_steps_per_second": 1.172,
      "step": 2700
    },
    {
      "epoch": 0.9999074845036544,
      "mean_token_accuracy": 0.7819749694749694,
      "mean_token_accuracy_utility": 0.6718253968253969,
      "step": 2702,
      "total_flos": 0.0,
      "train_loss": 2.351434097388865,
      "train_runtime": 33703.7845,
      "train_samples_per_second": 0.641,
      "train_steps_per_second": 0.08
    }
  ],
  "logging_steps": 5,
  "max_steps": 2702,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}