{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 100,
  "global_step": 2703,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001849796522382538,
      "grad_norm": 18.25,
      "learning_rate": 3.690036900369004e-07,
      "loss": 2.4352,
      "mean_token_accuracy": 0.733363019086835,
      "mean_token_accuracy_utility": 0.6862420193308187,
      "step": 5
    },
    {
      "epoch": 0.003699593044765076,
      "grad_norm": 14.9375,
      "learning_rate": 7.380073800738008e-07,
      "loss": 2.4333,
      "mean_token_accuracy": 0.7255770839028888,
      "mean_token_accuracy_utility": 0.6903680980750067,
      "step": 10
    },
    {
      "epoch": 0.005549389567147614,
      "grad_norm": 17.125,
      "learning_rate": 1.1070110701107011e-06,
      "loss": 2.4642,
      "mean_token_accuracy": 0.7401319471213864,
      "mean_token_accuracy_utility": 0.6701658008241339,
      "step": 15
    },
    {
      "epoch": 0.007399186089530152,
      "grad_norm": 14.5,
      "learning_rate": 1.4760147601476015e-06,
      "loss": 2.3942,
      "mean_token_accuracy": 0.7349658471081838,
      "mean_token_accuracy_utility": 0.6845740899213989,
      "step": 20
    },
    {
      "epoch": 0.00924898261191269,
      "grad_norm": 16.125,
      "learning_rate": 1.845018450184502e-06,
      "loss": 2.3674,
      "mean_token_accuracy": 0.7208179328443819,
      "mean_token_accuracy_utility": 0.6923890908474004,
      "step": 25
    },
    {
      "epoch": 0.011098779134295227,
      "grad_norm": 13.0,
      "learning_rate": 2.2140221402214023e-06,
      "loss": 2.2495,
      "mean_token_accuracy": 0.7207755271653988,
      "mean_token_accuracy_utility": 0.7040735203106454,
      "step": 30
    },
    {
      "epoch": 0.012948575656677765,
      "grad_norm": 8.125,
      "learning_rate": 2.5830258302583027e-06,
      "loss": 2.0442,
      "mean_token_accuracy": 0.7515207916117835,
      "mean_token_accuracy_utility": 0.709345544507898,
      "step": 35
    },
    {
      "epoch": 0.014798372179060304,
      "grad_norm": 5.40625,
      "learning_rate": 2.952029520295203e-06,
      "loss": 2.1081,
      "mean_token_accuracy": 0.7443690000617026,
      "mean_token_accuracy_utility": 0.6977245873441789,
      "step": 40
    },
    {
      "epoch": 0.016648168701442843,
      "grad_norm": 4.15625,
      "learning_rate": 3.3210332103321034e-06,
      "loss": 2.0328,
      "mean_token_accuracy": 0.7306490348150503,
      "mean_token_accuracy_utility": 0.7198939613138364,
      "step": 45
    },
    {
      "epoch": 0.01849796522382538,
      "grad_norm": 3.53125,
      "learning_rate": 3.690036900369004e-06,
      "loss": 2.0009,
      "mean_token_accuracy": 0.7392133800912574,
      "mean_token_accuracy_utility": 0.7158021623164571,
      "step": 50
    },
    {
      "epoch": 0.020347761746207917,
      "grad_norm": 3.046875,
      "learning_rate": 4.059040590405905e-06,
      "loss": 1.8516,
      "mean_token_accuracy": 0.7588164237600719,
      "mean_token_accuracy_utility": 0.728139955800301,
      "step": 55
    },
    {
      "epoch": 0.022197558268590455,
      "grad_norm": 2.875,
      "learning_rate": 4.428044280442805e-06,
      "loss": 1.9561,
      "mean_token_accuracy": 0.7461067219625048,
      "mean_token_accuracy_utility": 0.7133929595536372,
      "step": 60
    },
    {
      "epoch": 0.024047354790972992,
      "grad_norm": 2.78125,
      "learning_rate": 4.797047970479705e-06,
      "loss": 1.9069,
      "mean_token_accuracy": 0.7463690469627212,
      "mean_token_accuracy_utility": 0.7199195118782864,
      "step": 65
    },
    {
      "epoch": 0.02589715131335553,
      "grad_norm": 2.984375,
      "learning_rate": 5.166051660516605e-06,
      "loss": 1.876,
      "mean_token_accuracy": 0.7440005379505436,
      "mean_token_accuracy_utility": 0.732975300883602,
      "step": 70
    },
    {
      "epoch": 0.02774694783573807,
      "grad_norm": 2.65625,
      "learning_rate": 5.535055350553506e-06,
      "loss": 1.8396,
      "mean_token_accuracy": 0.7552101628985035,
      "mean_token_accuracy_utility": 0.728841651215457,
      "step": 75
    },
    {
      "epoch": 0.029596744358120607,
      "grad_norm": 2.421875,
      "learning_rate": 5.904059040590406e-06,
      "loss": 1.7621,
      "mean_token_accuracy": 0.7657496092767428,
      "mean_token_accuracy_utility": 0.7391688756920565,
      "step": 80
    },
    {
      "epoch": 0.031446540880503145,
      "grad_norm": 2.34375,
      "learning_rate": 6.273062730627307e-06,
      "loss": 1.8067,
      "mean_token_accuracy": 0.7695703771500103,
      "mean_token_accuracy_utility": 0.7217418976514438,
      "step": 85
    },
    {
      "epoch": 0.033296337402885685,
      "grad_norm": 2.359375,
      "learning_rate": 6.642066420664207e-06,
      "loss": 1.8047,
      "mean_token_accuracy": 0.7585375588049337,
      "mean_token_accuracy_utility": 0.7307809707689813,
      "step": 90
    },
    {
      "epoch": 0.03514613392526822,
      "grad_norm": 2.5,
      "learning_rate": 7.011070110701108e-06,
      "loss": 1.7883,
      "mean_token_accuracy": 0.7734939347424241,
      "mean_token_accuracy_utility": 0.7266810592423606,
      "step": 95
    },
    {
      "epoch": 0.03699593044765076,
      "grad_norm": 2.390625,
      "learning_rate": 7.380073800738008e-06,
      "loss": 1.7679,
      "mean_token_accuracy": 0.7649327256435624,
      "mean_token_accuracy_utility": 0.7320843387433315,
      "step": 100
    },
    {
      "epoch": 0.03699593044765076,
      "eval_mean_token_accuracy": 0.7571483577747888,
      "eval_reasoning_loss": 0.8216973543167114,
      "eval_reasoning_runtime": 91.6437,
      "eval_reasoning_samples_per_second": 1.408,
      "eval_reasoning_steps_per_second": 0.36,
      "step": 100
    },
    {
      "epoch": 0.03699593044765076,
      "eval_mean_token_accuracy": 0.7308979247601156,
      "eval_utility_loss": 0.9901875853538513,
      "eval_utility_runtime": 34.1426,
      "eval_utility_samples_per_second": 1.435,
      "eval_utility_steps_per_second": 0.381,
      "step": 100
    },
    {
      "epoch": 0.038845726970033294,
      "grad_norm": 2.3125,
      "learning_rate": 7.749077490774908e-06,
      "loss": 1.6712,
      "mean_token_accuracy": 0.7734131761723025,
      "mean_token_accuracy_utility": 0.7464594291282449,
      "step": 105
    },
    {
      "epoch": 0.040695523492415835,
      "grad_norm": 2.171875,
      "learning_rate": 8.11808118081181e-06,
      "loss": 1.6978,
      "mean_token_accuracy": 0.779535841537834,
      "mean_token_accuracy_utility": 0.731752178254986,
      "step": 110
    },
    {
      "epoch": 0.042545320014798375,
      "grad_norm": 2.46875,
      "learning_rate": 8.48708487084871e-06,
      "loss": 1.7755,
      "mean_token_accuracy": 0.7731289404439617,
      "mean_token_accuracy_utility": 0.722725243799952,
      "step": 115
    },
    {
      "epoch": 0.04439511653718091,
      "grad_norm": 2.234375,
      "learning_rate": 8.85608856088561e-06,
      "loss": 1.7769,
      "mean_token_accuracy": 0.7661713715193729,
      "mean_token_accuracy_utility": 0.7287322530026057,
      "step": 120
    },
    {
      "epoch": 0.04624491305956345,
      "grad_norm": 2.28125,
      "learning_rate": 9.22509225092251e-06,
      "loss": 1.7917,
      "mean_token_accuracy": 0.7644381426849354,
      "mean_token_accuracy_utility": 0.7273824069463871,
      "step": 125
    },
    {
      "epoch": 0.048094709581945984,
      "grad_norm": 2.390625,
      "learning_rate": 9.59409594095941e-06,
      "loss": 1.769,
      "mean_token_accuracy": 0.7684694383781465,
      "mean_token_accuracy_utility": 0.7246125413723525,
      "step": 130
    },
    {
      "epoch": 0.049944506104328525,
      "grad_norm": 2.171875,
      "learning_rate": 9.963099630996312e-06,
      "loss": 1.7008,
      "mean_token_accuracy": 0.7735760693321927,
      "mean_token_accuracy_utility": 0.7341881713165099,
      "step": 135
    },
    {
      "epoch": 0.05179430262671106,
      "grad_norm": 2.4375,
      "learning_rate": 1.033210332103321e-05,
      "loss": 1.8055,
      "mean_token_accuracy": 0.7550621824666877,
      "mean_token_accuracy_utility": 0.729864623321159,
      "step": 140
    },
    {
      "epoch": 0.0536440991490936,
      "grad_norm": 2.4375,
      "learning_rate": 1.0701107011070112e-05,
      "loss": 1.7603,
      "mean_token_accuracy": 0.7713707436438139,
      "mean_token_accuracy_utility": 0.7292619534815439,
      "step": 145
    },
    {
      "epoch": 0.05549389567147614,
      "grad_norm": 2.3125,
      "learning_rate": 1.1070110701107012e-05,
      "loss": 1.7458,
      "mean_token_accuracy": 0.7621733074594857,
      "mean_token_accuracy_utility": 0.7335882348147522,
      "step": 150
    },
    {
      "epoch": 0.057343692193858674,
      "grad_norm": 2.34375,
      "learning_rate": 1.1439114391143913e-05,
      "loss": 1.6387,
      "mean_token_accuracy": 0.7850145445686626,
      "mean_token_accuracy_utility": 0.7421167192184894,
      "step": 155
    },
    {
      "epoch": 0.059193488716241215,
      "grad_norm": 2.640625,
      "learning_rate": 1.1808118081180812e-05,
      "loss": 1.7477,
      "mean_token_accuracy": 0.7734781345454584,
      "mean_token_accuracy_utility": 0.7260915539021525,
      "step": 160
    },
    {
      "epoch": 0.06104328523862375,
      "grad_norm": 2.65625,
      "learning_rate": 1.2177121771217713e-05,
      "loss": 1.7375,
      "mean_token_accuracy": 0.7722359281787945,
      "mean_token_accuracy_utility": 0.7320988726595373,
      "step": 165
    },
    {
      "epoch": 0.06289308176100629,
      "grad_norm": 2.328125,
      "learning_rate": 1.2546125461254614e-05,
      "loss": 1.6903,
      "mean_token_accuracy": 0.782997093679984,
      "mean_token_accuracy_utility": 0.733542601482777,
      "step": 170
    },
    {
      "epoch": 0.06474287828338883,
      "grad_norm": 2.265625,
      "learning_rate": 1.2915129151291515e-05,
      "loss": 1.6712,
      "mean_token_accuracy": 0.7753829073489278,
      "mean_token_accuracy_utility": 0.737767532872154,
      "step": 175
    },
    {
      "epoch": 0.06659267480577137,
      "grad_norm": 2.328125,
      "learning_rate": 1.3284132841328414e-05,
      "loss": 1.7082,
      "mean_token_accuracy": 0.7719573662038298,
      "mean_token_accuracy_utility": 0.7323614996874132,
      "step": 180
    },
    {
      "epoch": 0.0684424713281539,
      "grad_norm": 2.25,
      "learning_rate": 1.3653136531365315e-05,
      "loss": 1.5932,
      "mean_token_accuracy": 0.7893967567108061,
      "mean_token_accuracy_utility": 0.7444797814764765,
      "step": 185
    },
    {
      "epoch": 0.07029226785053644,
      "grad_norm": 2.203125,
      "learning_rate": 1.4022140221402215e-05,
      "loss": 1.6409,
      "mean_token_accuracy": 0.7904701119474233,
      "mean_token_accuracy_utility": 0.7351474746406983,
      "step": 190
    },
    {
      "epoch": 0.07214206437291898,
      "grad_norm": 2.515625,
      "learning_rate": 1.4391143911439116e-05,
      "loss": 1.7163,
      "mean_token_accuracy": 0.7754604153937449,
      "mean_token_accuracy_utility": 0.7294819479937482,
      "step": 195
    },
    {
      "epoch": 0.07399186089530152,
      "grad_norm": 2.3125,
      "learning_rate": 1.4760147601476015e-05,
      "loss": 1.6638,
      "mean_token_accuracy": 0.7788005398372313,
      "mean_token_accuracy_utility": 0.7344039554737974,
      "step": 200
    },
    {
      "epoch": 0.07399186089530152,
      "eval_mean_token_accuracy": 0.7722574687346156,
      "eval_reasoning_loss": 0.7413960099220276,
      "eval_reasoning_runtime": 90.7006,
      "eval_reasoning_samples_per_second": 1.422,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 200
    },
    {
      "epoch": 0.07399186089530152,
      "eval_mean_token_accuracy": 0.7349459787542352,
      "eval_utility_loss": 0.9672914743423462,
      "eval_utility_runtime": 34.1639,
      "eval_utility_samples_per_second": 1.434,
      "eval_utility_steps_per_second": 0.381,
      "step": 200
    },
    {
      "epoch": 0.07584165741768406,
      "grad_norm": 3.578125,
      "learning_rate": 1.5129151291512916e-05,
      "loss": 1.6915,
      "mean_token_accuracy": 0.7675238974663193,
      "mean_token_accuracy_utility": 0.7374232945610754,
      "step": 205
    },
    {
      "epoch": 0.07769145394006659,
      "grad_norm": 2.4375,
      "learning_rate": 1.5498154981549817e-05,
      "loss": 1.7672,
      "mean_token_accuracy": 0.7753124748949457,
      "mean_token_accuracy_utility": 0.7212399194502185,
      "step": 210
    },
    {
      "epoch": 0.07954125046244913,
      "grad_norm": 2.140625,
      "learning_rate": 1.5867158671586716e-05,
      "loss": 1.5718,
      "mean_token_accuracy": 0.785921804157327,
      "mean_token_accuracy_utility": 0.7498165589993043,
      "step": 215
    },
    {
      "epoch": 0.08139104698483167,
      "grad_norm": 2.265625,
      "learning_rate": 1.623616236162362e-05,
      "loss": 1.6835,
      "mean_token_accuracy": 0.7830632042338903,
      "mean_token_accuracy_utility": 0.729833570425148,
      "step": 220
    },
    {
      "epoch": 0.08324084350721421,
      "grad_norm": 2.640625,
      "learning_rate": 1.6605166051660518e-05,
      "loss": 1.6704,
      "mean_token_accuracy": 0.7826464850175741,
      "mean_token_accuracy_utility": 0.7314276114970937,
      "step": 225
    },
    {
      "epoch": 0.08509064002959675,
      "grad_norm": 2.3125,
      "learning_rate": 1.697416974169742e-05,
      "loss": 1.6916,
      "mean_token_accuracy": 0.7732421579578095,
      "mean_token_accuracy_utility": 0.7355732734262601,
      "step": 230
    },
    {
      "epoch": 0.08694043655197928,
      "grad_norm": 2.28125,
      "learning_rate": 1.734317343173432e-05,
      "loss": 1.6575,
      "mean_token_accuracy": 0.7912907262000741,
      "mean_token_accuracy_utility": 0.7322631383197342,
      "step": 235
    },
    {
      "epoch": 0.08879023307436182,
      "grad_norm": 2.390625,
      "learning_rate": 1.771217712177122e-05,
      "loss": 1.6596,
      "mean_token_accuracy": 0.7773446349810542,
      "mean_token_accuracy_utility": 0.7382294911707297,
      "step": 240
    },
    {
      "epoch": 0.09064002959674436,
      "grad_norm": 2.46875,
      "learning_rate": 1.8081180811808117e-05,
      "loss": 1.6068,
      "mean_token_accuracy": 0.7845992982879917,
      "mean_token_accuracy_utility": 0.7470928910315177,
      "step": 245
    },
    {
      "epoch": 0.0924898261191269,
      "grad_norm": 2.328125,
      "learning_rate": 1.845018450184502e-05,
      "loss": 1.7231,
      "mean_token_accuracy": 0.7701165755251836,
      "mean_token_accuracy_utility": 0.7362037221980933,
      "step": 250
    },
    {
      "epoch": 0.09433962264150944,
      "grad_norm": 2.234375,
      "learning_rate": 1.8819188191881922e-05,
      "loss": 1.6047,
      "mean_token_accuracy": 0.7911960879326136,
      "mean_token_accuracy_utility": 0.7417890291881379,
      "step": 255
    },
    {
      "epoch": 0.09618941916389197,
      "grad_norm": 2.109375,
      "learning_rate": 1.918819188191882e-05,
      "loss": 1.6104,
      "mean_token_accuracy": 0.7862329303273177,
      "mean_token_accuracy_utility": 0.7421809753338823,
      "step": 260
    },
    {
      "epoch": 0.09803921568627451,
      "grad_norm": 2.234375,
      "learning_rate": 1.955719557195572e-05,
      "loss": 1.7002,
      "mean_token_accuracy": 0.7848856573585556,
      "mean_token_accuracy_utility": 0.728400324678955,
      "step": 265
    },
    {
      "epoch": 0.09988901220865705,
      "grad_norm": 2.59375,
      "learning_rate": 1.9926199261992623e-05,
      "loss": 1.6052,
      "mean_token_accuracy": 0.8045336314551275,
      "mean_token_accuracy_utility": 0.7279325258360025,
      "step": 270
    },
    {
      "epoch": 0.10173880873103959,
      "grad_norm": 2.390625,
      "learning_rate": 1.9999866506037346e-05,
      "loss": 1.6918,
      "mean_token_accuracy": 0.7816989885685712,
      "mean_token_accuracy_utility": 0.7331254488727733,
      "step": 275
    },
    {
      "epoch": 0.10358860525342212,
      "grad_norm": 2.09375,
      "learning_rate": 1.9999324192922497e-05,
      "loss": 1.6244,
      "mean_token_accuracy": 0.7984723544965504,
      "mean_token_accuracy_utility": 0.735363957914751,
      "step": 280
    },
    {
      "epoch": 0.10543840177580466,
      "grad_norm": 2.3125,
      "learning_rate": 1.9998364739888954e-05,
      "loss": 1.5693,
      "mean_token_accuracy": 0.7881016215782837,
      "mean_token_accuracy_utility": 0.7499378557968293,
      "step": 285
    },
    {
      "epoch": 0.1072881982981872,
      "grad_norm": 3.390625,
      "learning_rate": 1.9996988186962044e-05,
      "loss": 1.576,
      "mean_token_accuracy": 0.7985653391158737,
      "mean_token_accuracy_utility": 0.7401793878757422,
      "step": 290
    },
    {
      "epoch": 0.10913799482056974,
      "grad_norm": 2.34375,
      "learning_rate": 1.999519459156716e-05,
      "loss": 1.5364,
      "mean_token_accuracy": 0.7936807424162721,
      "mean_token_accuracy_utility": 0.7499745888290767,
      "step": 295
    },
    {
      "epoch": 0.11098779134295228,
      "grad_norm": 2.4375,
      "learning_rate": 1.9992984028527388e-05,
      "loss": 1.5584,
      "mean_token_accuracy": 0.7916533167366196,
      "mean_token_accuracy_utility": 0.7491119785020117,
      "step": 300
    },
    {
      "epoch": 0.11098779134295228,
      "eval_mean_token_accuracy": 0.7793672937729486,
      "eval_reasoning_loss": 0.7071473598480225,
      "eval_reasoning_runtime": 90.7314,
      "eval_reasoning_samples_per_second": 1.422,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 300
    },
    {
      "epoch": 0.11098779134295228,
      "eval_mean_token_accuracy": 0.7378149137837154,
      "eval_utility_loss": 0.9572708606719971,
      "eval_utility_runtime": 34.1415,
      "eval_utility_samples_per_second": 1.435,
      "eval_utility_steps_per_second": 0.381,
      "step": 300
    },
    {
      "epoch": 0.1128375878653348,
      "grad_norm": 2.375,
      "learning_rate": 1.9990356590060363e-05,
      "loss": 1.6369,
      "mean_token_accuracy": 0.7844759121768761,
      "mean_token_accuracy_utility": 0.7375486775188693,
      "step": 305
    },
    {
      "epoch": 0.11468738438771735,
      "grad_norm": 2.125,
      "learning_rate": 1.9987312385774444e-05,
      "loss": 1.5281,
      "mean_token_accuracy": 0.8044784712684754,
      "mean_token_accuracy_utility": 0.7488485632898492,
      "step": 310
    },
    {
      "epoch": 0.11653718091009989,
      "grad_norm": 2.171875,
      "learning_rate": 1.9983851542664125e-05,
      "loss": 1.5834,
      "mean_token_accuracy": 0.7935193489329249,
      "mean_token_accuracy_utility": 0.7452218170436391,
      "step": 315
    },
    {
      "epoch": 0.11838697743248243,
      "grad_norm": 2.25,
      "learning_rate": 1.9979974205104743e-05,
      "loss": 1.5818,
      "mean_token_accuracy": 0.7851486157227512,
      "mean_token_accuracy_utility": 0.7439524476316757,
      "step": 320
    },
    {
      "epoch": 0.12023677395486497,
      "grad_norm": 2.203125,
      "learning_rate": 1.9975680534846457e-05,
      "loss": 1.5757,
      "mean_token_accuracy": 0.7924221709191014,
      "mean_token_accuracy_utility": 0.7468762805677509,
      "step": 325
    },
    {
      "epoch": 0.1220865704772475,
      "grad_norm": 2.375,
      "learning_rate": 1.99709707110075e-05,
      "loss": 1.5988,
      "mean_token_accuracy": 0.7894889887765855,
      "mean_token_accuracy_utility": 0.7389159371045535,
      "step": 330
    },
    {
      "epoch": 0.12393636699963004,
      "grad_norm": 2.421875,
      "learning_rate": 1.99658449300667e-05,
      "loss": 1.5691,
      "mean_token_accuracy": 0.7825413830540595,
      "mean_token_accuracy_utility": 0.7524667356716732,
      "step": 335
    },
    {
      "epoch": 0.12578616352201258,
      "grad_norm": 2.609375,
      "learning_rate": 1.9960303405855295e-05,
      "loss": 1.629,
      "mean_token_accuracy": 0.7934933301265643,
      "mean_token_accuracy_utility": 0.7342044729964099,
      "step": 340
    },
    {
      "epoch": 0.12763596004439512,
      "grad_norm": 2.171875,
      "learning_rate": 1.9954346369548002e-05,
      "loss": 1.51,
      "mean_token_accuracy": 0.8093522648839968,
      "mean_token_accuracy_utility": 0.7469265543675234,
      "step": 345
    },
    {
      "epoch": 0.12948575656677766,
      "grad_norm": 2.28125,
      "learning_rate": 1.994797406965338e-05,
      "loss": 1.6025,
      "mean_token_accuracy": 0.7798642598435852,
      "mean_token_accuracy_utility": 0.7486616167398472,
      "step": 350
    },
    {
      "epoch": 0.1313355530891602,
      "grad_norm": 2.3125,
      "learning_rate": 1.9941186772003463e-05,
      "loss": 1.6015,
      "mean_token_accuracy": 0.7750655414272936,
      "mean_token_accuracy_utility": 0.7492139523231998,
      "step": 355
    },
    {
      "epoch": 0.13318534961154274,
      "grad_norm": 2.171875,
      "learning_rate": 1.9933984759742664e-05,
      "loss": 1.5247,
      "mean_token_accuracy": 0.8009225404020031,
      "mean_token_accuracy_utility": 0.7521310993511416,
      "step": 360
    },
    {
      "epoch": 0.13503514613392525,
      "grad_norm": 2.25,
      "learning_rate": 1.9926368333315964e-05,
      "loss": 1.5874,
      "mean_token_accuracy": 0.8010546163583045,
      "mean_token_accuracy_utility": 0.7398091075567463,
      "step": 365
    },
    {
      "epoch": 0.1368849426563078,
      "grad_norm": 1.9921875,
      "learning_rate": 1.991833781045639e-05,
      "loss": 1.4948,
      "mean_token_accuracy": 0.8046166694676806,
      "mean_token_accuracy_utility": 0.753497762095775,
      "step": 370
    },
    {
      "epoch": 0.13873473917869034,
      "grad_norm": 2.3125,
      "learning_rate": 1.9909893526171745e-05,
      "loss": 1.5718,
      "mean_token_accuracy": 0.7986297342587545,
      "mean_token_accuracy_utility": 0.7460903873258029,
      "step": 375
    },
    {
      "epoch": 0.14058453570107288,
      "grad_norm": 2.328125,
      "learning_rate": 1.9901035832730647e-05,
      "loss": 1.5921,
      "mean_token_accuracy": 0.7824574267727593,
      "mean_token_accuracy_utility": 0.754375107141728,
      "step": 380
    },
    {
      "epoch": 0.14243433222345542,
      "grad_norm": 2.140625,
      "learning_rate": 1.989176509964781e-05,
      "loss": 1.4977,
      "mean_token_accuracy": 0.8007794697258095,
      "mean_token_accuracy_utility": 0.7563826030492032,
      "step": 385
    },
    {
      "epoch": 0.14428412874583796,
      "grad_norm": 2.828125,
      "learning_rate": 1.9882081713668665e-05,
      "loss": 1.6747,
      "mean_token_accuracy": 0.7788811271830037,
      "mean_token_accuracy_utility": 0.741330061523101,
      "step": 390
    },
    {
      "epoch": 0.1461339252682205,
      "grad_norm": 2.15625,
      "learning_rate": 1.987198607875319e-05,
      "loss": 1.579,
      "mean_token_accuracy": 0.8084123773160687,
      "mean_token_accuracy_utility": 0.7444845513461931,
      "step": 395
    },
    {
      "epoch": 0.14798372179060304,
      "grad_norm": 2.171875,
      "learning_rate": 1.986147861605908e-05,
      "loss": 1.5055,
      "mean_token_accuracy": 0.7819473730585773,
      "mean_token_accuracy_utility": 0.7657576584844321,
      "step": 400
    },
    {
      "epoch": 0.14798372179060304,
      "eval_mean_token_accuracy": 0.7824864788543128,
      "eval_reasoning_loss": 0.6912236213684082,
      "eval_reasoning_runtime": 90.6859,
      "eval_reasoning_samples_per_second": 1.422,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 400
    },
    {
      "epoch": 0.14798372179060304,
      "eval_mean_token_accuracy": 0.739192953643345,
      "eval_utility_loss": 0.9510657787322998,
      "eval_utility_runtime": 34.1283,
      "eval_utility_samples_per_second": 1.436,
      "eval_utility_steps_per_second": 0.381,
      "step": 400
    },
    {
      "epoch": 0.14983351831298558,
      "grad_norm": 2.34375,
      "learning_rate": 1.9850559763924176e-05,
      "loss": 1.6391,
      "mean_token_accuracy": 0.7896736536868725,
      "mean_token_accuracy_utility": 0.7336327297046571,
      "step": 405
    },
    {
      "epoch": 0.15168331483536812,
      "grad_norm": 2.28125,
      "learning_rate": 1.9839229977848164e-05,
      "loss": 1.6323,
      "mean_token_accuracy": 0.779750710992215,
      "mean_token_accuracy_utility": 0.7422517349789954,
      "step": 410
    },
    {
      "epoch": 0.15353311135775063,
      "grad_norm": 2.34375,
      "learning_rate": 1.9827489730473597e-05,
      "loss": 1.5776,
      "mean_token_accuracy": 0.805219140379361,
      "mean_token_accuracy_utility": 0.7409210830431235,
      "step": 415
    },
    {
      "epoch": 0.15538290788013318,
      "grad_norm": 2.046875,
      "learning_rate": 1.9815339511566148e-05,
      "loss": 1.5898,
      "mean_token_accuracy": 0.7833076918770017,
      "mean_token_accuracy_utility": 0.7462039253888771,
      "step": 420
    },
    {
      "epoch": 0.15723270440251572,
      "grad_norm": 2.125,
      "learning_rate": 1.9802779827994214e-05,
      "loss": 1.5128,
      "mean_token_accuracy": 0.7876728228193355,
      "mean_token_accuracy_utility": 0.7593114851083824,
      "step": 425
    },
    {
      "epoch": 0.15908250092489826,
      "grad_norm": 2.265625,
      "learning_rate": 1.9789811203707736e-05,
      "loss": 1.5804,
      "mean_token_accuracy": 0.7889510006808637,
      "mean_token_accuracy_utility": 0.7488056684685496,
      "step": 430
    },
    {
      "epoch": 0.1609322974472808,
      "grad_norm": 2.140625,
      "learning_rate": 1.9776434179716365e-05,
      "loss": 1.5311,
      "mean_token_accuracy": 0.7864164420331146,
      "mean_token_accuracy_utility": 0.7587389594204634,
      "step": 435
    },
    {
      "epoch": 0.16278209396966334,
      "grad_norm": 2.1875,
      "learning_rate": 1.9762649314066888e-05,
      "loss": 1.4871,
      "mean_token_accuracy": 0.8079452264609884,
      "mean_token_accuracy_utility": 0.7564897754599494,
      "step": 440
    },
    {
      "epoch": 0.16463189049204588,
      "grad_norm": 2.3125,
      "learning_rate": 1.9748457181819937e-05,
      "loss": 1.5418,
      "mean_token_accuracy": 0.7879754391262805,
      "mean_token_accuracy_utility": 0.7546586826161941,
      "step": 445
    },
    {
      "epoch": 0.16648168701442842,
      "grad_norm": 2.234375,
      "learning_rate": 1.9733858375026015e-05,
      "loss": 1.5585,
      "mean_token_accuracy": 0.8038543501161819,
      "mean_token_accuracy_utility": 0.7418649929947887,
      "step": 450
    },
    {
      "epoch": 0.16833148353681096,
      "grad_norm": 2.125,
      "learning_rate": 1.9718853502700783e-05,
      "loss": 1.5577,
      "mean_token_accuracy": 0.7917040020261957,
      "mean_token_accuracy_utility": 0.7543524784030604,
      "step": 455
    },
    {
      "epoch": 0.1701812800591935,
      "grad_norm": 2.21875,
      "learning_rate": 1.9703443190799673e-05,
      "loss": 1.5784,
      "mean_token_accuracy": 0.8018247638490477,
      "mean_token_accuracy_utility": 0.7390648818462731,
      "step": 460
    },
    {
      "epoch": 0.17203107658157601,
      "grad_norm": 2.453125,
      "learning_rate": 1.9687628082191748e-05,
      "loss": 1.6044,
      "mean_token_accuracy": 0.7879587008042465,
      "mean_token_accuracy_utility": 0.7425367423322035,
      "step": 465
    },
    {
      "epoch": 0.17388087310395856,
      "grad_norm": 2.203125,
      "learning_rate": 1.9671408836632915e-05,
      "loss": 1.5345,
      "mean_token_accuracy": 0.808480896864203,
      "mean_token_accuracy_utility": 0.7454854892568232,
      "step": 470
    },
    {
      "epoch": 0.1757306696263411,
      "grad_norm": 2.125,
      "learning_rate": 1.9654786130738372e-05,
      "loss": 1.6447,
      "mean_token_accuracy": 0.7911058157465503,
      "mean_token_accuracy_utility": 0.7364022443775862,
      "step": 475
    },
    {
      "epoch": 0.17758046614872364,
      "grad_norm": 2.265625,
      "learning_rate": 1.96377606579544e-05,
      "loss": 1.5714,
      "mean_token_accuracy": 0.7831565411888421,
      "mean_token_accuracy_utility": 0.7548740674444849,
      "step": 480
    },
    {
      "epoch": 0.17943026267110618,
      "grad_norm": 2.03125,
      "learning_rate": 1.9620333128529436e-05,
      "loss": 1.4684,
      "mean_token_accuracy": 0.8052516468290773,
      "mean_token_accuracy_utility": 0.7600851597099805,
      "step": 485
    },
    {
      "epoch": 0.18128005919348872,
      "grad_norm": 2.359375,
      "learning_rate": 1.9602504269484432e-05,
      "loss": 1.5397,
      "mean_token_accuracy": 0.7838208973127153,
      "mean_token_accuracy_utility": 0.760160893883877,
      "step": 490
    },
    {
      "epoch": 0.18312985571587126,
      "grad_norm": 2.46875,
      "learning_rate": 1.958427482458253e-05,
      "loss": 1.5888,
      "mean_token_accuracy": 0.7879145502660743,
      "mean_token_accuracy_utility": 0.7470239603000838,
      "step": 495
    },
    {
      "epoch": 0.1849796522382538,
      "grad_norm": 2.03125,
      "learning_rate": 1.9565645554298043e-05,
      "loss": 1.5533,
      "mean_token_accuracy": 0.8118774166960818,
      "mean_token_accuracy_utility": 0.7412136093485899,
      "step": 500
    },
    {
      "epoch": 0.1849796522382538,
      "eval_mean_token_accuracy": 0.784884442115606,
      "eval_reasoning_loss": 0.6809573769569397,
      "eval_reasoning_runtime": 90.6702,
      "eval_reasoning_samples_per_second": 1.423,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 500
    },
    {
      "epoch": 0.1849796522382538,
      "eval_mean_token_accuracy": 0.7395856360992847,
      "eval_utility_loss": 0.9482628107070923,
      "eval_utility_runtime": 34.1302,
      "eval_utility_samples_per_second": 1.436,
      "eval_utility_steps_per_second": 0.381,
      "step": 500
    },
    {
      "epoch": 0.18682944876063634,
      "grad_norm": 2.34375,
      "learning_rate": 1.9546617235784716e-05,
      "loss": 1.5528,
      "mean_token_accuracy": 0.7824735765421873,
      "mean_token_accuracy_utility": 0.7555657730338993,
      "step": 505
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 2.203125,
      "learning_rate": 1.952719066284333e-05,
      "loss": 1.533,
      "mean_token_accuracy": 0.7940380575898193,
      "mean_token_accuracy_utility": 0.7511317326708323,
      "step": 510
    },
    {
      "epoch": 0.1905290418054014,
      "grad_norm": 2.125,
      "learning_rate": 1.9507366645888544e-05,
      "loss": 1.5542,
      "mean_token_accuracy": 0.7983090912935927,
      "mean_token_accuracy_utility": 0.7457392987038391,
      "step": 515
    },
    {
      "epoch": 0.19237883832778394,
      "grad_norm": 2.421875,
      "learning_rate": 1.9487146011915142e-05,
      "loss": 1.5243,
      "mean_token_accuracy": 0.7971335868147531,
      "mean_token_accuracy_utility": 0.7537615323745663,
      "step": 520
    },
    {
      "epoch": 0.19422863485016648,
      "grad_norm": 2.09375,
      "learning_rate": 1.9466529604463484e-05,
      "loss": 1.5688,
      "mean_token_accuracy": 0.7946301466224356,
      "mean_token_accuracy_utility": 0.7457635620288553,
      "step": 525
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 2.1875,
      "learning_rate": 1.9445518283584345e-05,
      "loss": 1.4746,
      "mean_token_accuracy": 0.7891994726379796,
      "mean_token_accuracy_utility": 0.770169088293984,
      "step": 530
    },
    {
      "epoch": 0.19792822789493156,
      "grad_norm": 2.125,
      "learning_rate": 1.942411292580304e-05,
      "loss": 1.4469,
      "mean_token_accuracy": 0.808227007974763,
      "mean_token_accuracy_utility": 0.761155358908933,
      "step": 535
    },
    {
      "epoch": 0.1997780244173141,
      "grad_norm": 2.328125,
      "learning_rate": 1.9402314424082824e-05,
      "loss": 1.53,
      "mean_token_accuracy": 0.7917001173003277,
      "mean_token_accuracy_utility": 0.758537490559292,
      "step": 540
    },
    {
      "epoch": 0.20162782093969664,
      "grad_norm": 2.234375,
      "learning_rate": 1.938012368778768e-05,
      "loss": 1.5218,
      "mean_token_accuracy": 0.7911062665437651,
      "mean_token_accuracy_utility": 0.7610327836972651,
      "step": 545
    },
    {
      "epoch": 0.20347761746207918,
      "grad_norm": 2.0625,
      "learning_rate": 1.935754164264437e-05,
      "loss": 1.5124,
      "mean_token_accuracy": 0.7980343145096289,
      "mean_token_accuracy_utility": 0.7598198965191211,
      "step": 550
    },
    {
      "epoch": 0.20532741398446172,
      "grad_norm": 2.1875,
      "learning_rate": 1.9334569230703794e-05,
      "loss": 1.5259,
      "mean_token_accuracy": 0.7957821137880806,
      "mean_token_accuracy_utility": 0.7544856073469484,
      "step": 555
    },
    {
      "epoch": 0.20717721050684423,
      "grad_norm": 2.25,
      "learning_rate": 1.9311207410301728e-05,
      "loss": 1.5564,
      "mean_token_accuracy": 0.7869893103150211,
      "mean_token_accuracy_utility": 0.7564924035422191,
      "step": 560
    },
    {
      "epoch": 0.20902700702922677,
      "grad_norm": 2.15625,
      "learning_rate": 1.9287457156018824e-05,
      "loss": 1.4869,
      "mean_token_accuracy": 0.8072885452555421,
      "mean_token_accuracy_utility": 0.7566970248968835,
      "step": 565
    },
    {
      "epoch": 0.21087680355160932,
      "grad_norm": 2.09375,
      "learning_rate": 1.926331945863995e-05,
      "loss": 1.4937,
      "mean_token_accuracy": 0.7988722906655864,
      "mean_token_accuracy_utility": 0.761134699473422,
      "step": 570
    },
    {
      "epoch": 0.21272660007399186,
      "grad_norm": 2.109375,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 1.4912,
      "mean_token_accuracy": 0.7991658869083185,
      "mean_token_accuracy_utility": 0.7575369440861953,
      "step": 575
    },
    {
      "epoch": 0.2145763965963744,
      "grad_norm": 2.0625,
      "learning_rate": 1.9213885778506228e-05,
      "loss": 1.3933,
      "mean_token_accuracy": 0.8009183443760766,
      "mean_token_accuracy_utility": 0.7805549428406658,
      "step": 580
    },
    {
      "epoch": 0.21642619311875694,
      "grad_norm": 2.203125,
      "learning_rate": 1.9188591857966875e-05,
      "loss": 1.4847,
      "mean_token_accuracy": 0.795048233169968,
      "mean_token_accuracy_utility": 0.7680815438637421,
      "step": 585
    },
    {
      "epoch": 0.21827598964113948,
      "grad_norm": 2.375,
      "learning_rate": 1.916291461867652e-05,
      "loss": 1.4359,
      "mean_token_accuracy": 0.802102027639853,
      "mean_token_accuracy_utility": 0.7737737223850669,
      "step": 590
    },
    {
      "epoch": 0.22012578616352202,
      "grad_norm": 2.390625,
      "learning_rate": 1.9136855131807705e-05,
      "loss": 1.5524,
      "mean_token_accuracy": 0.7947592978321306,
      "mean_token_accuracy_utility": 0.7545689731805246,
      "step": 595
    },
    {
      "epoch": 0.22197558268590456,
      "grad_norm": 2.203125,
      "learning_rate": 1.9110414484479116e-05,
      "loss": 1.5175,
      "mean_token_accuracy": 0.7922366997505506,
      "mean_token_accuracy_utility": 0.75968284848979,
      "step": 600
    },
    {
      "epoch": 0.22197558268590456,
      "eval_mean_token_accuracy": 0.7866712387046192,
      "eval_reasoning_loss": 0.6737954616546631,
      "eval_reasoning_runtime": 90.577,
      "eval_reasoning_samples_per_second": 1.424,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 600
    },
    {
      "epoch": 0.22197558268590456,
      "eval_mean_token_accuracy": 0.7401591829478846,
      "eval_utility_loss": 0.9473322033882141,
      "eval_utility_runtime": 34.0825,
      "eval_utility_samples_per_second": 1.438,
      "eval_utility_steps_per_second": 0.381,
      "step": 600
    },
    {
      "epoch": 0.2238253792082871,
      "grad_norm": 2.1875,
      "learning_rate": 1.908359377971025e-05,
      "loss": 1.5442,
      "mean_token_accuracy": 0.8037209360108017,
      "mean_token_accuracy_utility": 0.7504636765991228,
      "step": 605
    },
    {
      "epoch": 0.2256751757306696,
      "grad_norm": 2.125,
      "learning_rate": 1.9056394136375377e-05,
      "loss": 1.5029,
      "mean_token_accuracy": 0.8038146558228012,
      "mean_token_accuracy_utility": 0.7548630219226108,
      "step": 610
    },
    {
      "epoch": 0.22752497225305215,
      "grad_norm": 2.359375,
      "learning_rate": 1.902881668915688e-05,
      "loss": 1.4988,
      "mean_token_accuracy": 0.7930083611484073,
      "mean_token_accuracy_utility": 0.7644326724412535,
      "step": 615
    },
    {
      "epoch": 0.2293747687754347,
      "grad_norm": 2.109375,
      "learning_rate": 1.9000862588497918e-05,
      "loss": 1.4241,
      "mean_token_accuracy": 0.8006595258965626,
      "mean_token_accuracy_utility": 0.771841119381511,
      "step": 620
    },
    {
      "epoch": 0.23122456529781724,
      "grad_norm": 2.03125,
      "learning_rate": 1.897253300055443e-05,
      "loss": 1.4711,
      "mean_token_accuracy": 0.7970521470274297,
      "mean_token_accuracy_utility": 0.7637335207251746,
      "step": 625
    },
    {
      "epoch": 0.23307436182019978,
      "grad_norm": 2.046875,
      "learning_rate": 1.8943829107146485e-05,
      "loss": 1.5901,
      "mean_token_accuracy": 0.7917324997506512,
      "mean_token_accuracy_utility": 0.74606608210426,
      "step": 630
    },
    {
      "epoch": 0.23492415834258232,
      "grad_norm": 2.203125,
      "learning_rate": 1.891475210570898e-05,
      "loss": 1.4777,
      "mean_token_accuracy": 0.7972993607336061,
      "mean_token_accuracy_utility": 0.7690194593808833,
      "step": 635
    },
    {
      "epoch": 0.23677395486496486,
      "grad_norm": 2.109375,
      "learning_rate": 1.8885303209241694e-05,
      "loss": 1.5381,
      "mean_token_accuracy": 0.7905666633419803,
      "mean_token_accuracy_utility": 0.7585641482586615,
      "step": 640
    },
    {
      "epoch": 0.2386237513873474,
      "grad_norm": 2.453125,
      "learning_rate": 1.8855483646258677e-05,
      "loss": 1.5087,
      "mean_token_accuracy": 0.8054752727519071,
      "mean_token_accuracy_utility": 0.7463697071173068,
      "step": 645
    },
    {
      "epoch": 0.24047354790972994,
      "grad_norm": 2.46875,
      "learning_rate": 1.882529466073701e-05,
      "loss": 1.4627,
      "mean_token_accuracy": 0.8023027718040604,
      "mean_token_accuracy_utility": 0.7613224038130008,
      "step": 650
    },
    {
      "epoch": 0.24232334443211248,
      "grad_norm": 2.234375,
      "learning_rate": 1.879473751206489e-05,
      "loss": 1.5371,
      "mean_token_accuracy": 0.7955562702631545,
      "mean_token_accuracy_utility": 0.7546993213966487,
      "step": 655
    },
    {
      "epoch": 0.244173140954495,
      "grad_norm": 2.1875,
      "learning_rate": 1.876381347498913e-05,
      "loss": 1.5246,
      "mean_token_accuracy": 0.7937458668003463,
      "mean_token_accuracy_utility": 0.7530825995953357,
      "step": 660
    },
    {
      "epoch": 0.24602293747687753,
      "grad_norm": 2.3125,
      "learning_rate": 1.8732523839561934e-05,
      "loss": 1.4903,
      "mean_token_accuracy": 0.7869780327241185,
      "mean_token_accuracy_utility": 0.770935411429585,
      "step": 665
    },
    {
      "epoch": 0.24787273399926008,
      "grad_norm": 2.3125,
      "learning_rate": 1.8700869911087115e-05,
      "loss": 1.5364,
      "mean_token_accuracy": 0.7898664730386754,
      "mean_token_accuracy_utility": 0.7615567923890705,
      "step": 670
    },
    {
      "epoch": 0.24972253052164262,
      "grad_norm": 2.125,
      "learning_rate": 1.8668853010065633e-05,
      "loss": 1.4509,
      "mean_token_accuracy": 0.7992567586334502,
      "mean_token_accuracy_utility": 0.7707564720207654,
      "step": 675
    },
    {
      "epoch": 0.25157232704402516,
      "grad_norm": 2.15625,
      "learning_rate": 1.8636474472140494e-05,
      "loss": 1.6082,
      "mean_token_accuracy": 0.7891003285447369,
      "mean_token_accuracy_utility": 0.7418351620870632,
      "step": 680
    },
    {
      "epoch": 0.2534221235664077,
      "grad_norm": 2.390625,
      "learning_rate": 1.8603735648041054e-05,
      "loss": 1.4466,
      "mean_token_accuracy": 0.7981473241440952,
      "mean_token_accuracy_utility": 0.7698945425656111,
      "step": 685
    },
    {
      "epoch": 0.25527192008879024,
      "grad_norm": 2.046875,
      "learning_rate": 1.857063790352665e-05,
      "loss": 1.4378,
      "mean_token_accuracy": 0.8174694490444026,
      "mean_token_accuracy_utility": 0.7593309826087697,
      "step": 690
    },
    {
      "epoch": 0.2571217166111728,
      "grad_norm": 2.3125,
      "learning_rate": 1.853718261932964e-05,
      "loss": 1.3942,
      "mean_token_accuracy": 0.8047737648945542,
      "mean_token_accuracy_utility": 0.7804560431246508,
      "step": 695
    },
    {
      "epoch": 0.2589715131335553,
      "grad_norm": 2.546875,
      "learning_rate": 1.85033711910978e-05,
      "loss": 1.4547,
      "mean_token_accuracy": 0.8033916776729889,
      "mean_token_accuracy_utility": 0.7721742331415254,
      "step": 700
    },
    {
      "epoch": 0.2589715131335553,
      "eval_mean_token_accuracy": 0.7875659982604034,
      "eval_reasoning_loss": 0.6690588593482971,
      "eval_reasoning_runtime": 90.5936,
      "eval_reasoning_samples_per_second": 1.424,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 700
    },
    {
      "epoch": 0.2589715131335553,
      "eval_mean_token_accuracy": 0.7411532061255085,
      "eval_utility_loss": 0.9473390579223633,
      "eval_utility_runtime": 34.0539,
      "eval_utility_samples_per_second": 1.439,
      "eval_utility_steps_per_second": 0.382,
      "step": 700
    },
    {
      "epoch": 0.26082130965593786,
      "grad_norm": 2.28125,
      "learning_rate": 1.846920502933609e-05,
      "loss": 1.414,
      "mean_token_accuracy": 0.8055711605675281,
      "mean_token_accuracy_utility": 0.7722408753915214,
      "step": 705
    },
    {
      "epoch": 0.2626711061783204,
      "grad_norm": 2.203125,
      "learning_rate": 1.843468555934783e-05,
      "loss": 1.5029,
      "mean_token_accuracy": 0.7934029811261978,
      "mean_token_accuracy_utility": 0.7605090463244949,
      "step": 710
    },
    {
      "epoch": 0.26452090270070294,
      "grad_norm": 2.328125,
      "learning_rate": 1.839981422117523e-05,
      "loss": 1.4768,
      "mean_token_accuracy": 0.7952129295075439,
      "mean_token_accuracy_utility": 0.7656193281008713,
      "step": 715
    },
    {
      "epoch": 0.2663706992230855,
      "grad_norm": 2.265625,
      "learning_rate": 1.836459246953931e-05,
      "loss": 1.4076,
      "mean_token_accuracy": 0.798071816318851,
      "mean_token_accuracy_utility": 0.7765290186562053,
      "step": 720
    },
    {
      "epoch": 0.268220495745468,
      "grad_norm": 2.140625,
      "learning_rate": 1.8329021773779242e-05,
      "loss": 1.4832,
      "mean_token_accuracy": 0.7967917468409248,
      "mean_token_accuracy_utility": 0.7653929622135168,
      "step": 725
    },
    {
      "epoch": 0.2700702922678505,
      "grad_norm": 2.21875,
      "learning_rate": 1.8293103617791016e-05,
      "loss": 1.5553,
      "mean_token_accuracy": 0.7922974543876347,
      "mean_token_accuracy_utility": 0.7492368209202911,
      "step": 730
    },
    {
      "epoch": 0.27192008879023305,
      "grad_norm": 2.171875,
      "learning_rate": 1.825683949996556e-05,
      "loss": 1.5262,
      "mean_token_accuracy": 0.7907782371837208,
      "mean_token_accuracy_utility": 0.7631783877911531,
      "step": 735
    },
    {
      "epoch": 0.2737698853126156,
      "grad_norm": 2.203125,
      "learning_rate": 1.8220230933126234e-05,
      "loss": 1.4073,
      "mean_token_accuracy": 0.8091891026227614,
      "mean_token_accuracy_utility": 0.7709934169712821,
      "step": 740
    },
    {
      "epoch": 0.27561968183499813,
      "grad_norm": 2.390625,
      "learning_rate": 1.81832794444657e-05,
      "loss": 1.46,
      "mean_token_accuracy": 0.8000897782035349,
      "mean_token_accuracy_utility": 0.7677145556863297,
      "step": 745
    },
    {
      "epoch": 0.27746947835738067,
      "grad_norm": 2.234375,
      "learning_rate": 1.814598657548223e-05,
      "loss": 1.4247,
      "mean_token_accuracy": 0.804800792488229,
      "mean_token_accuracy_utility": 0.7729233316198599,
      "step": 750
    },
    {
      "epoch": 0.2793192748797632,
      "grad_norm": 2.21875,
      "learning_rate": 1.8108353881915403e-05,
      "loss": 1.4773,
      "mean_token_accuracy": 0.7996251723828839,
      "mean_token_accuracy_utility": 0.7633741415020094,
      "step": 755
    },
    {
      "epoch": 0.28116907140214575,
      "grad_norm": 2.265625,
      "learning_rate": 1.8070382933681185e-05,
      "loss": 1.4552,
      "mean_token_accuracy": 0.7960272508026704,
      "mean_token_accuracy_utility": 0.7729572571573959,
      "step": 760
    },
    {
      "epoch": 0.2830188679245283,
      "grad_norm": 2.296875,
      "learning_rate": 1.803207531480645e-05,
      "loss": 1.5039,
      "mean_token_accuracy": 0.7957206962227857,
      "mean_token_accuracy_utility": 0.7572809140257283,
      "step": 765
    },
    {
      "epoch": 0.28486866444691084,
      "grad_norm": 2.28125,
      "learning_rate": 1.79934326233629e-05,
      "loss": 1.4022,
      "mean_token_accuracy": 0.8047799313329056,
      "mean_token_accuracy_utility": 0.7774161064616979,
      "step": 770
    },
    {
      "epoch": 0.2867184609692934,
      "grad_norm": 2.25,
      "learning_rate": 1.7954456471400393e-05,
      "loss": 1.5027,
      "mean_token_accuracy": 0.7925628807756002,
      "mean_token_accuracy_utility": 0.7590824590282541,
      "step": 775
    },
    {
      "epoch": 0.2885682574916759,
      "grad_norm": 2.03125,
      "learning_rate": 1.7915148484879706e-05,
      "loss": 1.4903,
      "mean_token_accuracy": 0.7976968310154874,
      "mean_token_accuracy_utility": 0.7617999687581154,
      "step": 780
    },
    {
      "epoch": 0.29041805401405846,
      "grad_norm": 2.09375,
      "learning_rate": 1.7875510303604678e-05,
      "loss": 1.4787,
      "mean_token_accuracy": 0.8081336680956532,
      "mean_token_accuracy_utility": 0.7640173949391225,
      "step": 785
    },
    {
      "epoch": 0.292267850536441,
      "grad_norm": 2.234375,
      "learning_rate": 1.783554358115384e-05,
      "loss": 1.4282,
      "mean_token_accuracy": 0.7871402421045065,
      "mean_token_accuracy_utility": 0.7827357302428485,
      "step": 790
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 2.203125,
      "learning_rate": 1.7795249984811397e-05,
      "loss": 1.4474,
      "mean_token_accuracy": 0.8087583163972003,
      "mean_token_accuracy_utility": 0.7669730165422537,
      "step": 795
    },
    {
      "epoch": 0.2959674435812061,
      "grad_norm": 2.375,
      "learning_rate": 1.77546311954977e-05,
      "loss": 1.4459,
      "mean_token_accuracy": 0.7915649310925741,
      "mean_token_accuracy_utility": 0.7744157470996265,
      "step": 800
    },
    {
      "epoch": 0.2959674435812061,
      "eval_mean_token_accuracy": 0.7882872999294879,
      "eval_reasoning_loss": 0.6648406386375427,
      "eval_reasoning_runtime": 90.4879,
      "eval_reasoning_samples_per_second": 1.426,
      "eval_reasoning_steps_per_second": 0.365,
      "step": 800
    },
    {
      "epoch": 0.2959674435812061,
      "eval_mean_token_accuracy": 0.7403740604110417,
      "eval_utility_loss": 0.9491404891014099,
      "eval_utility_runtime": 34.0863,
      "eval_utility_samples_per_second": 1.438,
      "eval_utility_steps_per_second": 0.381,
      "step": 800
    },
    {
      "epoch": 0.2978172401035886,
      "grad_norm": 2.140625,
      "learning_rate": 1.7713688907699107e-05,
      "loss": 1.4948,
      "mean_token_accuracy": 0.7846553802392734,
      "mean_token_accuracy_utility": 0.7736424818158273,
      "step": 805
    },
    {
      "epoch": 0.29966703662597116,
      "grad_norm": 2.09375,
      "learning_rate": 1.767242482939731e-05,
      "loss": 1.4369,
      "mean_token_accuracy": 0.7958824243639981,
      "mean_token_accuracy_utility": 0.7761171419573205,
      "step": 810
    },
    {
      "epoch": 0.3015168331483537,
      "grad_norm": 2.09375,
      "learning_rate": 1.7630840681998068e-05,
      "loss": 1.3548,
      "mean_token_accuracy": 0.80736475977383,
      "mean_token_accuracy_utility": 0.7865923218130962,
      "step": 815
    },
    {
      "epoch": 0.30336662967073624,
      "grad_norm": 2.203125,
      "learning_rate": 1.75889382002594e-05,
      "loss": 1.4261,
      "mean_token_accuracy": 0.7939946676546316,
      "mean_token_accuracy_utility": 0.7788754222661263,
      "step": 820
    },
    {
      "epoch": 0.3052164261931188,
      "grad_norm": 2.25,
      "learning_rate": 1.754671913221923e-05,
      "loss": 1.4854,
      "mean_token_accuracy": 0.7950805467979873,
      "mean_token_accuracy_utility": 0.7693706863993756,
      "step": 825
    },
    {
      "epoch": 0.30706622271550127,
      "grad_norm": 2.140625,
      "learning_rate": 1.7504185239122432e-05,
      "loss": 1.4174,
      "mean_token_accuracy": 0.8036131910310509,
      "mean_token_accuracy_utility": 0.771374076122447,
      "step": 830
    },
    {
      "epoch": 0.3089160192378838,
      "grad_norm": 2.1875,
      "learning_rate": 1.7461338295347404e-05,
      "loss": 1.419,
      "mean_token_accuracy": 0.8080622290564146,
      "mean_token_accuracy_utility": 0.7674777309844828,
      "step": 835
    },
    {
      "epoch": 0.31076581576026635,
      "grad_norm": 2.0,
      "learning_rate": 1.7418180088332005e-05,
      "loss": 1.3816,
      "mean_token_accuracy": 0.8020458525715215,
      "mean_token_accuracy_utility": 0.7831976126180913,
      "step": 840
    },
    {
      "epoch": 0.3126156122826489,
      "grad_norm": 2.109375,
      "learning_rate": 1.7374712418498997e-05,
      "loss": 1.4847,
      "mean_token_accuracy": 0.8012010100930299,
      "mean_token_accuracy_utility": 0.7651803201392368,
      "step": 845
    },
    {
      "epoch": 0.31446540880503143,
      "grad_norm": 2.09375,
      "learning_rate": 1.7330937099180956e-05,
      "loss": 1.513,
      "mean_token_accuracy": 0.7911196378329585,
      "mean_token_accuracy_utility": 0.7621944102606613,
      "step": 850
    },
    {
      "epoch": 0.316315205327414,
      "grad_norm": 2.21875,
      "learning_rate": 1.7286855956544616e-05,
      "loss": 1.4456,
      "mean_token_accuracy": 0.8015508888594797,
      "mean_token_accuracy_utility": 0.7693200143246266,
      "step": 855
    },
    {
      "epoch": 0.3181650018497965,
      "grad_norm": 2.140625,
      "learning_rate": 1.7242470829514674e-05,
      "loss": 1.4295,
      "mean_token_accuracy": 0.7913853867390709,
      "mean_token_accuracy_utility": 0.7785532431461258,
      "step": 860
    },
    {
      "epoch": 0.32001479837217905,
      "grad_norm": 2.171875,
      "learning_rate": 1.7197783569697084e-05,
      "loss": 1.4536,
      "mean_token_accuracy": 0.8024162619942639,
      "mean_token_accuracy_utility": 0.7695976461544662,
      "step": 865
    },
    {
      "epoch": 0.3218645948945616,
      "grad_norm": 2.375,
      "learning_rate": 1.715279604130183e-05,
      "loss": 1.4805,
      "mean_token_accuracy": 0.7897790581150643,
      "mean_token_accuracy_utility": 0.766944296030571,
      "step": 870
    },
    {
      "epoch": 0.32371439141694414,
      "grad_norm": 2.1875,
      "learning_rate": 1.7107510121065138e-05,
      "loss": 1.5013,
      "mean_token_accuracy": 0.8044772538237979,
      "mean_token_accuracy_utility": 0.7586207799478826,
      "step": 875
    },
    {
      "epoch": 0.3255641879393267,
      "grad_norm": 2.265625,
      "learning_rate": 1.7061927698171187e-05,
      "loss": 1.3676,
      "mean_token_accuracy": 0.8150121231145508,
      "mean_token_accuracy_utility": 0.7750855161899497,
      "step": 880
    },
    {
      "epoch": 0.3274139844617092,
      "grad_norm": 2.390625,
      "learning_rate": 1.7016050674173304e-05,
      "loss": 1.4642,
      "mean_token_accuracy": 0.8005630816874311,
      "mean_token_accuracy_utility": 0.7705914447234665,
      "step": 885
    },
    {
      "epoch": 0.32926378098409176,
      "grad_norm": 2.125,
      "learning_rate": 1.696988096291463e-05,
      "loss": 1.4991,
      "mean_token_accuracy": 0.7897074839709092,
      "mean_token_accuracy_utility": 0.7753447103459762,
      "step": 890
    },
    {
      "epoch": 0.3311135775064743,
      "grad_norm": 2.09375,
      "learning_rate": 1.6923420490448298e-05,
      "loss": 1.4146,
      "mean_token_accuracy": 0.7982234116193015,
      "mean_token_accuracy_utility": 0.7771191682418085,
      "step": 895
    },
    {
      "epoch": 0.33296337402885684,
      "grad_norm": 2.25,
      "learning_rate": 1.687667119495705e-05,
      "loss": 1.4662,
      "mean_token_accuracy": 0.7898920587065784,
      "mean_token_accuracy_utility": 0.7727022252807991,
      "step": 900
    },
    {
      "epoch": 0.33296337402885684,
      "eval_mean_token_accuracy": 0.789569727361343,
      "eval_reasoning_loss": 0.6612050533294678,
      "eval_reasoning_runtime": 90.5424,
      "eval_reasoning_samples_per_second": 1.425,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 900
    },
    {
      "epoch": 0.33296337402885684,
      "eval_mean_token_accuracy": 0.7405758410814858,
      "eval_utility_loss": 0.9502067565917969,
      "eval_utility_runtime": 34.0891,
      "eval_utility_samples_per_second": 1.437,
      "eval_utility_steps_per_second": 0.381,
      "step": 900
    },
    {
      "epoch": 0.3348131705512394,
      "grad_norm": 2.1875,
      "learning_rate": 1.6829635026672432e-05,
      "loss": 1.375,
      "mean_token_accuracy": 0.8001587831507324,
      "mean_token_accuracy_utility": 0.7861690360357142,
      "step": 905
    },
    {
      "epoch": 0.3366629670736219,
      "grad_norm": 2.09375,
      "learning_rate": 1.6782313947793377e-05,
      "loss": 1.4213,
      "mean_token_accuracy": 0.798133632559186,
      "mean_token_accuracy_utility": 0.7797262848693536,
      "step": 910
    },
    {
      "epoch": 0.33851276359600446,
      "grad_norm": 2.28125,
      "learning_rate": 1.6734709932404404e-05,
      "loss": 1.4424,
      "mean_token_accuracy": 0.7982745516154266,
      "mean_token_accuracy_utility": 0.7706211396109691,
      "step": 915
    },
    {
      "epoch": 0.340362560118387,
      "grad_norm": 2.171875,
      "learning_rate": 1.668682496639323e-05,
      "loss": 1.4792,
      "mean_token_accuracy": 0.7971749439205935,
      "mean_token_accuracy_utility": 0.7686769147446496,
      "step": 920
    },
    {
      "epoch": 0.3422123566407695,
      "grad_norm": 2.3125,
      "learning_rate": 1.663866104736793e-05,
      "loss": 1.4329,
      "mean_token_accuracy": 0.8132711138726995,
      "mean_token_accuracy_utility": 0.7650828475767211,
      "step": 925
    },
    {
      "epoch": 0.34406215316315203,
      "grad_norm": 2.265625,
      "learning_rate": 1.6590220184573623e-05,
      "loss": 1.4952,
      "mean_token_accuracy": 0.787795631258212,
      "mean_token_accuracy_utility": 0.7704060641080204,
      "step": 930
    },
    {
      "epoch": 0.34591194968553457,
      "grad_norm": 2.234375,
      "learning_rate": 1.6541504398808633e-05,
      "loss": 1.4546,
      "mean_token_accuracy": 0.798543437084761,
      "mean_token_accuracy_utility": 0.7717801582017753,
      "step": 935
    },
    {
      "epoch": 0.3477617462079171,
      "grad_norm": 2.03125,
      "learning_rate": 1.6492515722340193e-05,
      "loss": 1.4198,
      "mean_token_accuracy": 0.7837209286462473,
      "mean_token_accuracy_utility": 0.7880696107191522,
      "step": 940
    },
    {
      "epoch": 0.34961154273029965,
      "grad_norm": 2.234375,
      "learning_rate": 1.6443256198819665e-05,
      "loss": 1.4847,
      "mean_token_accuracy": 0.7895385343754591,
      "mean_token_accuracy_utility": 0.7704997210453141,
      "step": 945
    },
    {
      "epoch": 0.3514613392526822,
      "grad_norm": 2.125,
      "learning_rate": 1.6393727883197292e-05,
      "loss": 1.4276,
      "mean_token_accuracy": 0.7997216991279672,
      "mean_token_accuracy_utility": 0.7776914697312188,
      "step": 950
    },
    {
      "epoch": 0.35331113577506473,
      "grad_norm": 2.125,
      "learning_rate": 1.6343932841636455e-05,
      "loss": 1.3404,
      "mean_token_accuracy": 0.7945543957706461,
      "mean_token_accuracy_utility": 0.8048745241435056,
      "step": 955
    },
    {
      "epoch": 0.3551609322974473,
      "grad_norm": 2.09375,
      "learning_rate": 1.6293873151427505e-05,
      "loss": 1.4815,
      "mean_token_accuracy": 0.7919658958584688,
      "mean_token_accuracy_utility": 0.7697878317741453,
      "step": 960
    },
    {
      "epoch": 0.3570107288198298,
      "grad_norm": 2.34375,
      "learning_rate": 1.6243550900901076e-05,
      "loss": 1.5167,
      "mean_token_accuracy": 0.790767021466007,
      "mean_token_accuracy_utility": 0.7646075710408782,
      "step": 965
    },
    {
      "epoch": 0.35886052534221236,
      "grad_norm": 2.140625,
      "learning_rate": 1.6192968189340993e-05,
      "loss": 1.4608,
      "mean_token_accuracy": 0.7938717195700826,
      "mean_token_accuracy_utility": 0.7678621973345724,
      "step": 970
    },
    {
      "epoch": 0.3607103218645949,
      "grad_norm": 2.09375,
      "learning_rate": 1.6142127126896682e-05,
      "loss": 1.4629,
      "mean_token_accuracy": 0.7909464757045104,
      "mean_token_accuracy_utility": 0.776931491069728,
      "step": 975
    },
    {
      "epoch": 0.36256011838697744,
      "grad_norm": 2.140625,
      "learning_rate": 1.6091029834495133e-05,
      "loss": 1.4759,
      "mean_token_accuracy": 0.8049133694176156,
      "mean_token_accuracy_utility": 0.7686070690895958,
      "step": 980
    },
    {
      "epoch": 0.36440991490936,
      "grad_norm": 2.09375,
      "learning_rate": 1.603967844375245e-05,
      "loss": 1.4614,
      "mean_token_accuracy": 0.7986224664234092,
      "mean_token_accuracy_utility": 0.7699008301654866,
      "step": 985
    },
    {
      "epoch": 0.3662597114317425,
      "grad_norm": 2.15625,
      "learning_rate": 1.59880750968849e-05,
      "loss": 1.4013,
      "mean_token_accuracy": 0.798864915358146,
      "mean_token_accuracy_utility": 0.7816328616032392,
      "step": 990
    },
    {
      "epoch": 0.36810950795412506,
      "grad_norm": 2.21875,
      "learning_rate": 1.593622194661956e-05,
      "loss": 1.4032,
      "mean_token_accuracy": 0.812051472816522,
      "mean_token_accuracy_utility": 0.7729357207636134,
      "step": 995
    },
    {
      "epoch": 0.3699593044765076,
      "grad_norm": 2.03125,
      "learning_rate": 1.5884121156104496e-05,
      "loss": 1.3949,
      "mean_token_accuracy": 0.8016069744063803,
      "mean_token_accuracy_utility": 0.7836990857169952,
      "step": 1000
    },
    {
      "epoch": 0.3699593044765076,
      "eval_mean_token_accuracy": 0.7901525403628813,
      "eval_reasoning_loss": 0.658268392086029,
      "eval_reasoning_runtime": 90.5178,
      "eval_reasoning_samples_per_second": 1.425,
      "eval_reasoning_steps_per_second": 0.365,
      "step": 1000
    },
    {
      "epoch": 0.3699593044765076,
      "eval_mean_token_accuracy": 0.7411485717780765,
      "eval_utility_loss": 0.9503855109214783,
      "eval_utility_runtime": 34.0552,
      "eval_utility_samples_per_second": 1.439,
      "eval_utility_steps_per_second": 0.382,
      "step": 1000
    },
    {
      "epoch": 0.37180910099889014,
      "grad_norm": 2.234375,
      "learning_rate": 1.5831774898818558e-05,
      "loss": 1.5227,
      "mean_token_accuracy": 0.7910014454347624,
      "mean_token_accuracy_utility": 0.7644637396542124,
      "step": 1005
    },
    {
      "epoch": 0.3736588975212727,
      "grad_norm": 2.21875,
      "learning_rate": 1.5779185358480666e-05,
      "loss": 1.4469,
      "mean_token_accuracy": 0.796331253832965,
      "mean_token_accuracy_utility": 0.7740732150827497,
      "step": 1010
    },
    {
      "epoch": 0.3755086940436552,
      "grad_norm": 2.140625,
      "learning_rate": 1.5726354728958736e-05,
      "loss": 1.4274,
      "mean_token_accuracy": 0.8020732504465382,
      "mean_token_accuracy_utility": 0.7762363122069827,
      "step": 1015
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 2.140625,
      "learning_rate": 1.5673285214178176e-05,
      "loss": 1.4782,
      "mean_token_accuracy": 0.8016800825946373,
      "mean_token_accuracy_utility": 0.7620441024088671,
      "step": 1020
    },
    {
      "epoch": 0.37920828708842025,
      "grad_norm": 2.140625,
      "learning_rate": 1.5619979028029898e-05,
      "loss": 1.3919,
      "mean_token_accuracy": 0.8027623811982091,
      "mean_token_accuracy_utility": 0.7818722671791012,
      "step": 1025
    },
    {
      "epoch": 0.3810580836108028,
      "grad_norm": 2.0625,
      "learning_rate": 1.5566438394278005e-05,
      "loss": 1.4667,
      "mean_token_accuracy": 0.810734226995782,
      "mean_token_accuracy_utility": 0.7580320907980502,
      "step": 1030
    },
    {
      "epoch": 0.38290788013318533,
      "grad_norm": 2.171875,
      "learning_rate": 1.5512665546467008e-05,
      "loss": 1.6306,
      "mean_token_accuracy": 0.7809220643189005,
      "mean_token_accuracy_utility": 0.7534598392435468,
      "step": 1035
    },
    {
      "epoch": 0.38475767665556787,
      "grad_norm": 2.25,
      "learning_rate": 1.5458662727828633e-05,
      "loss": 1.4896,
      "mean_token_accuracy": 0.8033384531961218,
      "mean_token_accuracy_utility": 0.7632901645189228,
      "step": 1040
    },
    {
      "epoch": 0.3866074731779504,
      "grad_norm": 2.390625,
      "learning_rate": 1.540443219118827e-05,
      "loss": 1.3965,
      "mean_token_accuracy": 0.7961841542117015,
      "mean_token_accuracy_utility": 0.784776891345165,
      "step": 1045
    },
    {
      "epoch": 0.38845726970033295,
      "grad_norm": 2.21875,
      "learning_rate": 1.5349976198870974e-05,
      "loss": 1.4106,
      "mean_token_accuracy": 0.7973692085973345,
      "mean_token_accuracy_utility": 0.7849606946116621,
      "step": 1050
    },
    {
      "epoch": 0.3903070662227155,
      "grad_norm": 2.234375,
      "learning_rate": 1.529529702260709e-05,
      "loss": 1.4875,
      "mean_token_accuracy": 0.79482126959765,
      "mean_token_accuracy_utility": 0.76556137759991,
      "step": 1055
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 2.078125,
      "learning_rate": 1.5240396943437483e-05,
      "loss": 1.4173,
      "mean_token_accuracy": 0.7891225184234906,
      "mean_token_accuracy_utility": 0.788036756873641,
      "step": 1060
    },
    {
      "epoch": 0.3940066592674806,
      "grad_norm": 2.078125,
      "learning_rate": 1.5185278251618391e-05,
      "loss": 1.4574,
      "mean_token_accuracy": 0.7951387864187255,
      "mean_token_accuracy_utility": 0.7735260437227001,
      "step": 1065
    },
    {
      "epoch": 0.3958564557898631,
      "grad_norm": 2.21875,
      "learning_rate": 1.5129943246525871e-05,
      "loss": 1.4416,
      "mean_token_accuracy": 0.7941486943411826,
      "mean_token_accuracy_utility": 0.7769904041299937,
      "step": 1070
    },
    {
      "epoch": 0.39770625231224566,
      "grad_norm": 2.046875,
      "learning_rate": 1.5074394236559871e-05,
      "loss": 1.4006,
      "mean_token_accuracy": 0.813519846542906,
      "mean_token_accuracy_utility": 0.7711771684556186,
      "step": 1075
    },
    {
      "epoch": 0.3995560488346282,
      "grad_norm": 2.28125,
      "learning_rate": 1.5018633539047955e-05,
      "loss": 1.5712,
      "mean_token_accuracy": 0.7920979343243886,
      "mean_token_accuracy_utility": 0.7632444981984154,
      "step": 1080
    },
    {
      "epoch": 0.40140584535701074,
      "grad_norm": 2.046875,
      "learning_rate": 1.4962663480148606e-05,
      "loss": 1.3993,
      "mean_token_accuracy": 0.8048860981850087,
      "mean_token_accuracy_utility": 0.7789162401343684,
      "step": 1085
    },
    {
      "epoch": 0.4032556418793933,
      "grad_norm": 2.0625,
      "learning_rate": 1.4906486394754198e-05,
      "loss": 1.3859,
      "mean_token_accuracy": 0.8099425353173636,
      "mean_token_accuracy_utility": 0.7775224541708716,
      "step": 1090
    },
    {
      "epoch": 0.4051054384017758,
      "grad_norm": 2.390625,
      "learning_rate": 1.4850104626393598e-05,
      "loss": 1.4492,
      "mean_token_accuracy": 0.8057702939451559,
      "mean_token_accuracy_utility": 0.7682619681960898,
      "step": 1095
    },
    {
      "epoch": 0.40695523492415836,
      "grad_norm": 2.203125,
      "learning_rate": 1.4793520527134381e-05,
      "loss": 1.4438,
      "mean_token_accuracy": 0.8035946826973902,
      "mean_token_accuracy_utility": 0.7718347896136136,
      "step": 1100
    },
    {
      "epoch": 0.40695523492415836,
      "eval_mean_token_accuracy": 0.7905838130427169,
      "eval_reasoning_loss": 0.655839204788208,
      "eval_reasoning_runtime": 90.5297,
      "eval_reasoning_samples_per_second": 1.425,
      "eval_reasoning_steps_per_second": 0.365,
      "step": 1100
    },
    {
      "epoch": 0.40695523492415836,
      "eval_mean_token_accuracy": 0.7411495995376872,
      "eval_utility_loss": 0.9466155171394348,
      "eval_utility_runtime": 34.0734,
      "eval_utility_samples_per_second": 1.438,
      "eval_utility_steps_per_second": 0.382,
      "step": 1100
    },
    {
      "epoch": 0.4088050314465409,
      "grad_norm": 2.15625,
      "learning_rate": 1.473673645748473e-05,
      "loss": 1.4098,
      "mean_token_accuracy": 0.797641920318691,
      "mean_token_accuracy_utility": 0.7800954912883329,
      "step": 1105
    },
    {
      "epoch": 0.41065482796892344,
      "grad_norm": 2.171875,
      "learning_rate": 1.4679754786294954e-05,
      "loss": 1.4309,
      "mean_token_accuracy": 0.8016690686515314,
      "mean_token_accuracy_utility": 0.7759983051497436,
      "step": 1110
    },
    {
      "epoch": 0.412504624491306,
      "grad_norm": 2.03125,
      "learning_rate": 1.4622577890658668e-05,
      "loss": 1.3973,
      "mean_token_accuracy": 0.8043382060119489,
      "mean_token_accuracy_utility": 0.7790453831123709,
      "step": 1115
    },
    {
      "epoch": 0.41435442101368847,
      "grad_norm": 2.21875,
      "learning_rate": 1.4565208155813626e-05,
      "loss": 1.4033,
      "mean_token_accuracy": 0.8057985800872608,
      "mean_token_accuracy_utility": 0.7781032521991118,
      "step": 1120
    },
    {
      "epoch": 0.416204217536071,
      "grad_norm": 2.234375,
      "learning_rate": 1.4507647975042221e-05,
      "loss": 1.4014,
      "mean_token_accuracy": 0.7980797366781365,
      "mean_token_accuracy_utility": 0.7818549844948925,
      "step": 1125
    },
    {
      "epoch": 0.41805401405845355,
      "grad_norm": 2.109375,
      "learning_rate": 1.4449899749571643e-05,
      "loss": 1.2847,
      "mean_token_accuracy": 0.8021328598998556,
      "mean_token_accuracy_utility": 0.8117612965354034,
      "step": 1130
    },
    {
      "epoch": 0.4199038105808361,
      "grad_norm": 2.0625,
      "learning_rate": 1.4391965888473705e-05,
      "loss": 1.3306,
      "mean_token_accuracy": 0.8054404423386936,
      "mean_token_accuracy_utility": 0.7944913507914053,
      "step": 1135
    },
    {
      "epoch": 0.42175360710321863,
      "grad_norm": 2.03125,
      "learning_rate": 1.4333848808564353e-05,
      "loss": 1.3534,
      "mean_token_accuracy": 0.7963831829179484,
      "mean_token_accuracy_utility": 0.7912167100843395,
      "step": 1140
    },
    {
      "epoch": 0.42360340362560117,
      "grad_norm": 2.3125,
      "learning_rate": 1.4275550934302822e-05,
      "loss": 1.4055,
      "mean_token_accuracy": 0.8077453902740684,
      "mean_token_accuracy_utility": 0.7771742575664373,
      "step": 1145
    },
    {
      "epoch": 0.4254532001479837,
      "grad_norm": 2.390625,
      "learning_rate": 1.4217074697690532e-05,
      "loss": 1.5094,
      "mean_token_accuracy": 0.7923901574228422,
      "mean_token_accuracy_utility": 0.7654522120231234,
      "step": 1150
    },
    {
      "epoch": 0.42730299667036625,
      "grad_norm": 2.0625,
      "learning_rate": 1.4158422538169596e-05,
      "loss": 1.3914,
      "mean_token_accuracy": 0.7955600394290965,
      "mean_token_accuracy_utility": 0.7852925132983064,
      "step": 1155
    },
    {
      "epoch": 0.4291527931927488,
      "grad_norm": 2.140625,
      "learning_rate": 1.4099596902521073e-05,
      "loss": 1.3775,
      "mean_token_accuracy": 0.799777113771585,
      "mean_token_accuracy_utility": 0.7859889672887033,
      "step": 1160
    },
    {
      "epoch": 0.43100258971513133,
      "grad_norm": 2.078125,
      "learning_rate": 1.40406002447629e-05,
      "loss": 1.3535,
      "mean_token_accuracy": 0.7926819637530038,
      "mean_token_accuracy_utility": 0.7926749980766165,
      "step": 1165
    },
    {
      "epoch": 0.4328523862375139,
      "grad_norm": 2.09375,
      "learning_rate": 1.39814350260475e-05,
      "loss": 1.3896,
      "mean_token_accuracy": 0.8005776037300152,
      "mean_token_accuracy_utility": 0.7791530047341093,
      "step": 1170
    },
    {
      "epoch": 0.4347021827598964,
      "grad_norm": 2.390625,
      "learning_rate": 1.392210371455913e-05,
      "loss": 1.3764,
      "mean_token_accuracy": 0.8001032692078637,
      "mean_token_accuracy_utility": 0.7874627612834961,
      "step": 1175
    },
    {
      "epoch": 0.43655197928227896,
      "grad_norm": 2.140625,
      "learning_rate": 1.3862608785410903e-05,
      "loss": 1.4205,
      "mean_token_accuracy": 0.8061254738848019,
      "mean_token_accuracy_utility": 0.7721669361504112,
      "step": 1180
    },
    {
      "epoch": 0.4384017758046615,
      "grad_norm": 1.9609375,
      "learning_rate": 1.3802952720541543e-05,
      "loss": 1.2947,
      "mean_token_accuracy": 0.8133601366209563,
      "mean_token_accuracy_utility": 0.7985106256107891,
      "step": 1185
    },
    {
      "epoch": 0.44025157232704404,
      "grad_norm": 2.078125,
      "learning_rate": 1.3743138008611838e-05,
      "loss": 1.3641,
      "mean_token_accuracy": 0.8036737600393156,
      "mean_token_accuracy_utility": 0.7872050374717365,
      "step": 1190
    },
    {
      "epoch": 0.4421013688494266,
      "grad_norm": 2.125,
      "learning_rate": 1.3683167144900833e-05,
      "loss": 1.3757,
      "mean_token_accuracy": 0.8049544801687022,
      "mean_token_accuracy_utility": 0.7832934496308497,
      "step": 1195
    },
    {
      "epoch": 0.4439511653718091,
      "grad_norm": 2.125,
      "learning_rate": 1.3623042631201713e-05,
      "loss": 1.3705,
      "mean_token_accuracy": 0.7898726439069291,
      "mean_token_accuracy_utility": 0.7955650427763133,
      "step": 1200
    },
    {
      "epoch": 0.4439511653718091,
      "eval_mean_token_accuracy": 0.7913399065837978,
      "eval_reasoning_loss": 0.6541661620140076,
      "eval_reasoning_runtime": 90.4825,
      "eval_reasoning_samples_per_second": 1.426,
      "eval_reasoning_steps_per_second": 0.365,
      "step": 1200
    },
    {
      "epoch": 0.4439511653718091,
      "eval_mean_token_accuracy": 0.7409661409335714,
      "eval_utility_loss": 0.9521284103393555,
      "eval_utility_runtime": 34.055,
      "eval_utility_samples_per_second": 1.439,
      "eval_utility_steps_per_second": 0.382,
      "step": 1200
    },
    {
      "epoch": 0.44580096189419166,
      "grad_norm": 2.203125,
      "learning_rate": 1.3562766975717468e-05,
      "loss": 1.385,
      "mean_token_accuracy": 0.8002209767721199,
      "mean_token_accuracy_utility": 0.7836235889512201,
      "step": 1205
    },
    {
      "epoch": 0.4476507584165742,
      "grad_norm": 2.015625,
      "learning_rate": 1.3502342692956229e-05,
      "loss": 1.3053,
      "mean_token_accuracy": 0.8204219947484555,
      "mean_token_accuracy_utility": 0.7858187278799871,
      "step": 1210
    },
    {
      "epoch": 0.44950055493895674,
      "grad_norm": 2.3125,
      "learning_rate": 1.3441772303626387e-05,
      "loss": 1.433,
      "mean_token_accuracy": 0.7939850328753155,
      "mean_token_accuracy_utility": 0.7791943067719371,
      "step": 1215
    },
    {
      "epoch": 0.4513503514613392,
      "grad_norm": 2.234375,
      "learning_rate": 1.3381058334531436e-05,
      "loss": 1.3158,
      "mean_token_accuracy": 0.8134116836829088,
      "mean_token_accuracy_utility": 0.7882023875929918,
      "step": 1220
    },
    {
      "epoch": 0.45320014798372177,
      "grad_norm": 2.203125,
      "learning_rate": 1.3320203318464552e-05,
      "loss": 1.3949,
      "mean_token_accuracy": 0.7971792987648367,
      "mean_token_accuracy_utility": 0.7841586478936058,
      "step": 1225
    },
    {
      "epoch": 0.4550499445061043,
      "grad_norm": 1.9921875,
      "learning_rate": 1.3259209794102952e-05,
      "loss": 1.2873,
      "mean_token_accuracy": 0.8034199438847549,
      "mean_token_accuracy_utility": 0.8072989705829426,
      "step": 1230
    },
    {
      "epoch": 0.45689974102848685,
      "grad_norm": 2.203125,
      "learning_rate": 1.319808030590197e-05,
      "loss": 1.4679,
      "mean_token_accuracy": 0.8059433061287953,
      "mean_token_accuracy_utility": 0.768150655534322,
      "step": 1235
    },
    {
      "epoch": 0.4587495375508694,
      "grad_norm": 2.0625,
      "learning_rate": 1.3136817403988918e-05,
      "loss": 1.3245,
      "mean_token_accuracy": 0.7946016489153382,
      "mean_token_accuracy_utility": 0.8012968893824703,
      "step": 1240
    },
    {
      "epoch": 0.46059933407325193,
      "grad_norm": 2.125,
      "learning_rate": 1.3075423644056699e-05,
      "loss": 1.4555,
      "mean_token_accuracy": 0.7852164315289644,
      "mean_token_accuracy_utility": 0.7799166186388254,
      "step": 1245
    },
    {
      "epoch": 0.46244913059563447,
      "grad_norm": 2.0625,
      "learning_rate": 1.3013901587257207e-05,
      "loss": 1.3206,
      "mean_token_accuracy": 0.799543774178223,
      "mean_token_accuracy_utility": 0.8021937811114466,
      "step": 1250
    },
    {
      "epoch": 0.464298927118017,
      "grad_norm": 2.0625,
      "learning_rate": 1.2952253800094467e-05,
      "loss": 1.4366,
      "mean_token_accuracy": 0.7966514061427604,
      "mean_token_accuracy_utility": 0.7756003363346806,
      "step": 1255
    },
    {
      "epoch": 0.46614872364039955,
      "grad_norm": 2.34375,
      "learning_rate": 1.2890482854317565e-05,
      "loss": 1.515,
      "mean_token_accuracy": 0.7800464864208256,
      "mean_token_accuracy_utility": 0.7747564880004861,
      "step": 1260
    },
    {
      "epoch": 0.4679985201627821,
      "grad_norm": 2.078125,
      "learning_rate": 1.2828591326813382e-05,
      "loss": 1.3888,
      "mean_token_accuracy": 0.7948308415517878,
      "mean_token_accuracy_utility": 0.7900095084385188,
      "step": 1265
    },
    {
      "epoch": 0.46984831668516464,
      "grad_norm": 2.015625,
      "learning_rate": 1.2766581799499078e-05,
      "loss": 1.3545,
      "mean_token_accuracy": 0.8000872061772641,
      "mean_token_accuracy_utility": 0.7977183729226588,
      "step": 1270
    },
    {
      "epoch": 0.4716981132075472,
      "grad_norm": 2.25,
      "learning_rate": 1.2704456859214397e-05,
      "loss": 1.5467,
      "mean_token_accuracy": 0.787494110234478,
      "mean_token_accuracy_utility": 0.7609952717503742,
      "step": 1275
    },
    {
      "epoch": 0.4735479097299297,
      "grad_norm": 2.1875,
      "learning_rate": 1.264221909761374e-05,
      "loss": 1.3403,
      "mean_token_accuracy": 0.8173088847125782,
      "mean_token_accuracy_utility": 0.7830747434474324,
      "step": 1280
    },
    {
      "epoch": 0.47539770625231226,
      "grad_norm": 2.09375,
      "learning_rate": 1.2579871111058042e-05,
      "loss": 1.4106,
      "mean_token_accuracy": 0.790460870373084,
      "mean_token_accuracy_utility": 0.7842697889747962,
      "step": 1285
    },
    {
      "epoch": 0.4772475027746948,
      "grad_norm": 1.984375,
      "learning_rate": 1.251741550050649e-05,
      "loss": 1.2787,
      "mean_token_accuracy": 0.8098426317501441,
      "mean_token_accuracy_utility": 0.8063230772512915,
      "step": 1290
    },
    {
      "epoch": 0.47909729929707734,
      "grad_norm": 2.0,
      "learning_rate": 1.2454854871407993e-05,
      "loss": 1.4354,
      "mean_token_accuracy": 0.7957743941259954,
      "mean_token_accuracy_utility": 0.7807218002742429,
      "step": 1295
    },
    {
      "epoch": 0.4809470958194599,
      "grad_norm": 2.140625,
      "learning_rate": 1.2392191833592495e-05,
      "loss": 1.3806,
      "mean_token_accuracy": 0.8075084997924646,
      "mean_token_accuracy_utility": 0.7814047827215039,
      "step": 1300
    },
    {
      "epoch": 0.4809470958194599,
      "eval_mean_token_accuracy": 0.7914766216523565,
      "eval_reasoning_loss": 0.6522796154022217,
      "eval_reasoning_runtime": 90.5195,
      "eval_reasoning_samples_per_second": 1.425,
      "eval_reasoning_steps_per_second": 0.365,
      "step": 1300
    },
    {
      "epoch": 0.4809470958194599,
      "eval_mean_token_accuracy": 0.741120285533632,
      "eval_utility_loss": 0.9527794122695923,
      "eval_utility_runtime": 34.0668,
      "eval_utility_samples_per_second": 1.438,
      "eval_utility_steps_per_second": 0.382,
      "step": 1300
    },
    {
      "epoch": 0.4827968923418424,
      "grad_norm": 2.140625,
      "learning_rate": 1.2329429001162114e-05,
      "loss": 1.3345,
      "mean_token_accuracy": 0.8054217000652498,
      "mean_token_accuracy_utility": 0.792612358366271,
      "step": 1305
    },
    {
      "epoch": 0.48464668886422496,
      "grad_norm": 2.125,
      "learning_rate": 1.226656899238207e-05,
      "loss": 1.3175,
      "mean_token_accuracy": 0.8098936299065743,
      "mean_token_accuracy_utility": 0.7920122553735919,
      "step": 1310
    },
    {
      "epoch": 0.48649648538660745,
      "grad_norm": 2.15625,
      "learning_rate": 1.2203614429571475e-05,
      "loss": 1.3841,
      "mean_token_accuracy": 0.7892449483841746,
      "mean_token_accuracy_utility": 0.7945355933757255,
      "step": 1315
    },
    {
      "epoch": 0.48834628190899,
      "grad_norm": 2.046875,
      "learning_rate": 1.214056793899394e-05,
      "loss": 1.3112,
      "mean_token_accuracy": 0.8031152367575354,
      "mean_token_accuracy_utility": 0.7973513798175675,
      "step": 1320
    },
    {
      "epoch": 0.49019607843137253,
      "grad_norm": 2.15625,
      "learning_rate": 1.207743215074801e-05,
      "loss": 1.3468,
      "mean_token_accuracy": 0.8067241640348358,
      "mean_token_accuracy_utility": 0.7883982077081637,
      "step": 1325
    },
    {
      "epoch": 0.49204587495375507,
      "grad_norm": 2.078125,
      "learning_rate": 1.2014209698657424e-05,
      "loss": 1.3163,
      "mean_token_accuracy": 0.7973014350008075,
      "mean_token_accuracy_utility": 0.8009108904043314,
      "step": 1330
    },
    {
      "epoch": 0.4938956714761376,
      "grad_norm": 2.375,
      "learning_rate": 1.1950903220161286e-05,
      "loss": 1.4494,
      "mean_token_accuracy": 0.7973125980944369,
      "mean_token_accuracy_utility": 0.7691566963748965,
      "step": 1335
    },
    {
      "epoch": 0.49574546799852015,
      "grad_norm": 2.15625,
      "learning_rate": 1.1887515356203994e-05,
      "loss": 1.423,
      "mean_token_accuracy": 0.7958103429676744,
      "mean_token_accuracy_utility": 0.7822347798572423,
      "step": 1340
    },
    {
      "epoch": 0.4975952645209027,
      "grad_norm": 1.921875,
      "learning_rate": 1.1824048751125101e-05,
      "loss": 1.377,
      "mean_token_accuracy": 0.7936567596063777,
      "mean_token_accuracy_utility": 0.7887679773668583,
      "step": 1345
    },
    {
      "epoch": 0.49944506104328523,
      "grad_norm": 2.296875,
      "learning_rate": 1.1760506052548984e-05,
      "loss": 1.4055,
      "mean_token_accuracy": 0.794537959702917,
      "mean_token_accuracy_utility": 0.7888921740062197,
      "step": 1350
    },
    {
      "epoch": 0.5012948575656678,
      "grad_norm": 2.0,
      "learning_rate": 1.1696889911274394e-05,
      "loss": 1.3604,
      "mean_token_accuracy": 0.8039158355283801,
      "mean_token_accuracy_utility": 0.787205595719523,
      "step": 1355
    },
    {
      "epoch": 0.5031446540880503,
      "grad_norm": 2.1875,
      "learning_rate": 1.1633202981163882e-05,
      "loss": 1.3966,
      "mean_token_accuracy": 0.7928294579887477,
      "mean_token_accuracy_utility": 0.7916047231008297,
      "step": 1360
    },
    {
      "epoch": 0.5049944506104328,
      "grad_norm": 2.109375,
      "learning_rate": 1.1569447919033086e-05,
      "loss": 1.3155,
      "mean_token_accuracy": 0.7932566218745195,
      "mean_token_accuracy_utility": 0.8041895890338282,
      "step": 1365
    },
    {
      "epoch": 0.5068442471328154,
      "grad_norm": 2.078125,
      "learning_rate": 1.1505627384539888e-05,
      "loss": 1.4186,
      "mean_token_accuracy": 0.7997352134002677,
      "mean_token_accuracy_utility": 0.7790007069821939,
      "step": 1370
    },
    {
      "epoch": 0.5086940436551979,
      "grad_norm": 2.328125,
      "learning_rate": 1.1441744040073469e-05,
      "loss": 1.3748,
      "mean_token_accuracy": 0.7933830331908165,
      "mean_token_accuracy_utility": 0.7913311711056046,
      "step": 1375
    },
    {
      "epoch": 0.5105438401775805,
      "grad_norm": 2.171875,
      "learning_rate": 1.1377800550643246e-05,
      "loss": 1.4171,
      "mean_token_accuracy": 0.8005048033162889,
      "mean_token_accuracy_utility": 0.7798922720825783,
      "step": 1380
    },
    {
      "epoch": 0.512393636699963,
      "grad_norm": 2.046875,
      "learning_rate": 1.1313799583767693e-05,
      "loss": 1.2975,
      "mean_token_accuracy": 0.8024866624031277,
      "mean_token_accuracy_utility": 0.8020589843443743,
      "step": 1385
    },
    {
      "epoch": 0.5142434332223456,
      "grad_norm": 2.140625,
      "learning_rate": 1.124974380936305e-05,
      "loss": 1.3776,
      "mean_token_accuracy": 0.7971979741249758,
      "mean_token_accuracy_utility": 0.7868303150530993,
      "step": 1390
    },
    {
      "epoch": 0.516093229744728,
      "grad_norm": 2.15625,
      "learning_rate": 1.1185635899631963e-05,
      "loss": 1.4356,
      "mean_token_accuracy": 0.7986483675399658,
      "mean_token_accuracy_utility": 0.7801712573025522,
      "step": 1395
    },
    {
      "epoch": 0.5179430262671106,
      "grad_norm": 2.1875,
      "learning_rate": 1.1121478528951994e-05,
      "loss": 1.3329,
      "mean_token_accuracy": 0.804626148757243,
      "mean_token_accuracy_utility": 0.7973895081520931,
      "step": 1400
    },
    {
      "epoch": 0.5179430262671106,
      "eval_mean_token_accuracy": 0.791889067342275,
      "eval_reasoning_loss": 0.6509132981300354,
      "eval_reasoning_runtime": 90.5592,
      "eval_reasoning_samples_per_second": 1.424,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 1400
    },
    {
      "epoch": 0.5179430262671106,
      "eval_mean_token_accuracy": 0.741063573654075,
      "eval_utility_loss": 0.953432559967041,
      "eval_utility_runtime": 34.0701,
      "eval_utility_samples_per_second": 1.438,
      "eval_utility_steps_per_second": 0.382,
      "step": 1400
    },
    {
      "epoch": 0.5197928227894931,
      "grad_norm": 2.109375,
      "learning_rate": 1.1057274373764056e-05,
      "loss": 1.3659,
      "mean_token_accuracy": 0.8008705436601915,
      "mean_token_accuracy_utility": 0.789967762598635,
      "step": 1405
    },
    {
      "epoch": 0.5216426193118757,
      "grad_norm": 2.0625,
      "learning_rate": 1.0993026112460765e-05,
      "loss": 1.3784,
      "mean_token_accuracy": 0.8023630577977103,
      "mean_token_accuracy_utility": 0.7847640782557118,
      "step": 1410
    },
    {
      "epoch": 0.5234924158342582,
      "grad_norm": 2.09375,
      "learning_rate": 1.0928736425274702e-05,
      "loss": 1.3696,
      "mean_token_accuracy": 0.8010188862890365,
      "mean_token_accuracy_utility": 0.7875353507982814,
      "step": 1415
    },
    {
      "epoch": 0.5253422123566408,
      "grad_norm": 2.125,
      "learning_rate": 1.0864407994166609e-05,
      "loss": 1.3778,
      "mean_token_accuracy": 0.8008206604111084,
      "mean_token_accuracy_utility": 0.7911287691619452,
      "step": 1420
    },
    {
      "epoch": 0.5271920088790233,
      "grad_norm": 1.96875,
      "learning_rate": 1.0800043502713486e-05,
      "loss": 1.3334,
      "mean_token_accuracy": 0.8025326467352432,
      "mean_token_accuracy_utility": 0.7915591273348467,
      "step": 1425
    },
    {
      "epoch": 0.5290418054014059,
      "grad_norm": 2.140625,
      "learning_rate": 1.0735645635996676e-05,
      "loss": 1.4547,
      "mean_token_accuracy": 0.7954996298201868,
      "mean_token_accuracy_utility": 0.7717019300276056,
      "step": 1430
    },
    {
      "epoch": 0.5308916019237884,
      "grad_norm": 2.25,
      "learning_rate": 1.0671217080489816e-05,
      "loss": 1.3393,
      "mean_token_accuracy": 0.8066134663455257,
      "mean_token_accuracy_utility": 0.7895356548404592,
      "step": 1435
    },
    {
      "epoch": 0.532741398446171,
      "grad_norm": 2.109375,
      "learning_rate": 1.0606760523946789e-05,
      "loss": 1.3064,
      "mean_token_accuracy": 0.8144749852169557,
      "mean_token_accuracy_utility": 0.7927559719123365,
      "step": 1440
    },
    {
      "epoch": 0.5345911949685535,
      "grad_norm": 2.0,
      "learning_rate": 1.0542278655289588e-05,
      "loss": 1.3649,
      "mean_token_accuracy": 0.8028855124970257,
      "mean_token_accuracy_utility": 0.7863737099734542,
      "step": 1445
    },
    {
      "epoch": 0.536440991490936,
      "grad_norm": 2.140625,
      "learning_rate": 1.047777416449615e-05,
      "loss": 1.4463,
      "mean_token_accuracy": 0.8034415318491968,
      "mean_token_accuracy_utility": 0.7706630892617602,
      "step": 1450
    },
    {
      "epoch": 0.5382907880133185,
      "grad_norm": 2.296875,
      "learning_rate": 1.0413249742488132e-05,
      "loss": 1.3927,
      "mean_token_accuracy": 0.8015230583417996,
      "mean_token_accuracy_utility": 0.780653631071303,
      "step": 1455
    },
    {
      "epoch": 0.540140584535701,
      "grad_norm": 2.28125,
      "learning_rate": 1.034870808101866e-05,
      "loss": 1.3485,
      "mean_token_accuracy": 0.8012200406309324,
      "mean_token_accuracy_utility": 0.7922680052268601,
      "step": 1460
    },
    {
      "epoch": 0.5419903810580836,
      "grad_norm": 2.453125,
      "learning_rate": 1.0284151872560042e-05,
      "loss": 1.4649,
      "mean_token_accuracy": 0.8004178023950453,
      "mean_token_accuracy_utility": 0.7780262161594432,
      "step": 1465
    },
    {
      "epoch": 0.5438401775804661,
      "grad_norm": 1.9453125,
      "learning_rate": 1.0219583810191422e-05,
      "loss": 1.3765,
      "mean_token_accuracy": 0.8043348001556586,
      "mean_token_accuracy_utility": 0.7866759622581634,
      "step": 1470
    },
    {
      "epoch": 0.5456899741028487,
      "grad_norm": 2.21875,
      "learning_rate": 1.0155006587486468e-05,
      "loss": 1.4307,
      "mean_token_accuracy": 0.7897137003927182,
      "mean_token_accuracy_utility": 0.7819908969450304,
      "step": 1475
    },
    {
      "epoch": 0.5475397706252312,
      "grad_norm": 2.109375,
      "learning_rate": 1.0090422898400985e-05,
      "loss": 1.4339,
      "mean_token_accuracy": 0.8019959795099123,
      "mean_token_accuracy_utility": 0.7757578849005731,
      "step": 1480
    },
    {
      "epoch": 0.5493895671476138,
      "grad_norm": 2.265625,
      "learning_rate": 1.0025835437160523e-05,
      "loss": 1.5733,
      "mean_token_accuracy": 0.7944739171491417,
      "mean_token_accuracy_utility": 0.7576656792402287,
      "step": 1485
    },
    {
      "epoch": 0.5512393636699963,
      "grad_norm": 2.109375,
      "learning_rate": 9.96124689814801e-06,
      "loss": 1.315,
      "mean_token_accuracy": 0.7918607971412663,
      "mean_token_accuracy_utility": 0.8089928312622352,
      "step": 1490
    },
    {
      "epoch": 0.5530891601923789,
      "grad_norm": 2.328125,
      "learning_rate": 9.89665997579133e-06,
      "loss": 1.3496,
      "mean_token_accuracy": 0.7969397698364442,
      "mean_token_accuracy_utility": 0.7957842673639044,
      "step": 1495
    },
    {
      "epoch": 0.5549389567147613,
      "grad_norm": 2.109375,
      "learning_rate": 9.832077364450915e-06,
      "loss": 1.3189,
      "mean_token_accuracy": 0.8170009590507192,
      "mean_token_accuracy_utility": 0.7872159053166248,
      "step": 1500
    },
    {
      "epoch": 0.5549389567147613,
      "eval_mean_token_accuracy": 0.7923780416856482,
      "eval_reasoning_loss": 0.6500461101531982,
      "eval_reasoning_runtime": 90.5567,
      "eval_reasoning_samples_per_second": 1.425,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 1500
    },
    {
      "epoch": 0.5549389567147613,
      "eval_mean_token_accuracy": 0.7403061334324441,
      "eval_utility_loss": 0.9589380621910095,
      "eval_utility_runtime": 34.0738,
      "eval_utility_samples_per_second": 1.438,
      "eval_utility_steps_per_second": 0.382,
      "step": 1500
    },
    {
      "epoch": 0.5567887532371439,
      "grad_norm": 2.1875,
      "learning_rate": 9.767501758307376e-06,
      "loss": 1.321,
      "mean_token_accuracy": 0.8065425873262608,
      "mean_token_accuracy_utility": 0.7943946774521984,
      "step": 1505
    },
    {
      "epoch": 0.5586385497595264,
      "grad_norm": 2.234375,
      "learning_rate": 9.70293585124907e-06,
      "loss": 1.3184,
      "mean_token_accuracy": 0.8116000328169324,
      "mean_token_accuracy_utility": 0.7905118073450413,
      "step": 1510
    },
    {
      "epoch": 0.560488346281909,
      "grad_norm": 2.140625,
      "learning_rate": 9.638382336759749e-06,
      "loss": 1.3019,
      "mean_token_accuracy": 0.7978014581844312,
      "mean_token_accuracy_utility": 0.80674210746209,
      "step": 1515
    },
    {
      "epoch": 0.5623381428042915,
      "grad_norm": 2.171875,
      "learning_rate": 9.57384390780618e-06,
      "loss": 1.3531,
      "mean_token_accuracy": 0.802696855631807,
      "mean_token_accuracy_utility": 0.7895470973447456,
      "step": 1520
    },
    {
      "epoch": 0.5641879393266741,
      "grad_norm": 2.140625,
      "learning_rate": 9.50932325672582e-06,
      "loss": 1.3908,
      "mean_token_accuracy": 0.7917780336323472,
      "mean_token_accuracy_utility": 0.7865452618287391,
      "step": 1525
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 2.109375,
      "learning_rate": 9.444823075114481e-06,
      "loss": 1.3399,
      "mean_token_accuracy": 0.8031762500778408,
      "mean_token_accuracy_utility": 0.7933752142172894,
      "step": 1530
    },
    {
      "epoch": 0.5678875323714392,
      "grad_norm": 2.0625,
      "learning_rate": 9.380346053714055e-06,
      "loss": 1.2882,
      "mean_token_accuracy": 0.8122247756863163,
      "mean_token_accuracy_utility": 0.795472353764872,
      "step": 1535
    },
    {
      "epoch": 0.5697373288938217,
      "grad_norm": 2.125,
      "learning_rate": 9.315894882300269e-06,
      "loss": 1.2777,
      "mean_token_accuracy": 0.8209194562809525,
      "mean_token_accuracy_utility": 0.7953350879960319,
      "step": 1540
    },
    {
      "epoch": 0.5715871254162043,
      "grad_norm": 2.125,
      "learning_rate": 9.251472249570458e-06,
      "loss": 1.2966,
      "mean_token_accuracy": 0.7995781693217205,
      "mean_token_accuracy_utility": 0.8059219645793518,
      "step": 1545
    },
    {
      "epoch": 0.5734369219385868,
      "grad_norm": 2.03125,
      "learning_rate": 9.187080843031436e-06,
      "loss": 1.3336,
      "mean_token_accuracy": 0.8078994886901645,
      "mean_token_accuracy_utility": 0.7925450299739355,
      "step": 1550
    },
    {
      "epoch": 0.5752867184609693,
      "grad_norm": 2.296875,
      "learning_rate": 9.12272334888734e-06,
      "loss": 1.3922,
      "mean_token_accuracy": 0.7938785032983747,
      "mean_token_accuracy_utility": 0.7878480918395927,
      "step": 1555
    },
    {
      "epoch": 0.5771365149833518,
      "grad_norm": 2.0625,
      "learning_rate": 9.058402451927599e-06,
      "loss": 1.333,
      "mean_token_accuracy": 0.8105665894667903,
      "mean_token_accuracy_utility": 0.7904233630310513,
      "step": 1560
    },
    {
      "epoch": 0.5789863115057343,
      "grad_norm": 2.125,
      "learning_rate": 8.99412083541492e-06,
      "loss": 1.2798,
      "mean_token_accuracy": 0.8050580130670177,
      "mean_token_accuracy_utility": 0.8056968040677562,
      "step": 1565
    },
    {
      "epoch": 0.5808361080281169,
      "grad_norm": 2.265625,
      "learning_rate": 8.929881180973365e-06,
      "loss": 1.3255,
      "mean_token_accuracy": 0.7958576118148469,
      "mean_token_accuracy_utility": 0.801321578085453,
      "step": 1570
    },
    {
      "epoch": 0.5826859045504994,
      "grad_norm": 2.015625,
      "learning_rate": 8.865686168476458e-06,
      "loss": 1.3969,
      "mean_token_accuracy": 0.8003371273786305,
      "mean_token_accuracy_utility": 0.7833881697692724,
      "step": 1575
    },
    {
      "epoch": 0.584535701072882,
      "grad_norm": 2.078125,
      "learning_rate": 8.801538475935413e-06,
      "loss": 1.3296,
      "mean_token_accuracy": 0.8085386633850957,
      "mean_token_accuracy_utility": 0.7900983002694009,
      "step": 1580
    },
    {
      "epoch": 0.5863854975952645,
      "grad_norm": 2.1875,
      "learning_rate": 8.737440779387398e-06,
      "loss": 1.3578,
      "mean_token_accuracy": 0.800503784523884,
      "mean_token_accuracy_utility": 0.7926277128622738,
      "step": 1585
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 2.03125,
      "learning_rate": 8.673395752783924e-06,
      "loss": 1.4175,
      "mean_token_accuracy": 0.7990452865824158,
      "mean_token_accuracy_utility": 0.7798174330828442,
      "step": 1590
    },
    {
      "epoch": 0.5900850906400296,
      "grad_norm": 2.265625,
      "learning_rate": 8.609406067879258e-06,
      "loss": 1.3406,
      "mean_token_accuracy": 0.8055202971638427,
      "mean_token_accuracy_utility": 0.7910098208897741,
      "step": 1595
    },
    {
      "epoch": 0.5919348871624122,
      "grad_norm": 2.015625,
      "learning_rate": 8.54547439411901e-06,
      "loss": 1.368,
      "mean_token_accuracy": 0.7953000188002811,
      "mean_token_accuracy_utility": 0.7974292341883114,
      "step": 1600
    },
    {
      "epoch": 0.5919348871624122,
      "eval_mean_token_accuracy": 0.7925261556759985,
      "eval_reasoning_loss": 0.6492166519165039,
      "eval_reasoning_runtime": 90.5742,
      "eval_reasoning_samples_per_second": 1.424,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 1600
    },
    {
      "epoch": 0.5919348871624122,
      "eval_mean_token_accuracy": 0.7403092636767853,
      "eval_utility_loss": 0.9578368663787842,
      "eval_utility_runtime": 34.097,
      "eval_utility_samples_per_second": 1.437,
      "eval_utility_steps_per_second": 0.381,
      "step": 1600
    },
    {
      "epoch": 0.5937846836847946,
      "grad_norm": 2.203125,
      "learning_rate": 8.481603398528727e-06,
      "loss": 1.3579,
      "mean_token_accuracy": 0.8145502276379137,
      "mean_token_accuracy_utility": 0.782437774228584,
      "step": 1605
    },
    {
      "epoch": 0.5956344802071772,
      "grad_norm": 2.078125,
      "learning_rate": 8.417795745602679e-06,
      "loss": 1.3078,
      "mean_token_accuracy": 0.7980142436029243,
      "mean_token_accuracy_utility": 0.80417087960444,
      "step": 1610
    },
    {
      "epoch": 0.5974842767295597,
      "grad_norm": 2.125,
      "learning_rate": 8.35405409719266e-06,
      "loss": 1.3797,
      "mean_token_accuracy": 0.7949205720970914,
      "mean_token_accuracy_utility": 0.7862279385738071,
      "step": 1615
    },
    {
      "epoch": 0.5993340732519423,
      "grad_norm": 2.21875,
      "learning_rate": 8.290381112396989e-06,
      "loss": 1.3053,
      "mean_token_accuracy": 0.8080786918191107,
      "mean_token_accuracy_utility": 0.7976886381713114,
      "step": 1620
    },
    {
      "epoch": 0.6011838697743248,
      "grad_norm": 2.203125,
      "learning_rate": 8.226779447449538e-06,
      "loss": 1.2912,
      "mean_token_accuracy": 0.8098351265033081,
      "mean_token_accuracy_utility": 0.7999871407557483,
      "step": 1625
    },
    {
      "epoch": 0.6030336662967074,
      "grad_norm": 2.1875,
      "learning_rate": 8.163251755608954e-06,
      "loss": 1.3906,
      "mean_token_accuracy": 0.8029297433015149,
      "mean_token_accuracy_utility": 0.7832563580945605,
      "step": 1630
    },
    {
      "epoch": 0.6048834628190899,
      "grad_norm": 2.09375,
      "learning_rate": 8.099800687047958e-06,
      "loss": 1.3243,
      "mean_token_accuracy": 0.8049684240138373,
      "mean_token_accuracy_utility": 0.7962895986192665,
      "step": 1635
    },
    {
      "epoch": 0.6067332593414725,
      "grad_norm": 2.078125,
      "learning_rate": 8.036428888742789e-06,
      "loss": 1.3077,
      "mean_token_accuracy": 0.8040647849454332,
      "mean_token_accuracy_utility": 0.7996635911951229,
      "step": 1640
    },
    {
      "epoch": 0.608583055863855,
      "grad_norm": 2.15625,
      "learning_rate": 7.973139004362794e-06,
      "loss": 1.3509,
      "mean_token_accuracy": 0.800697612110152,
      "mean_token_accuracy_utility": 0.797439166718959,
      "step": 1645
    },
    {
      "epoch": 0.6104328523862376,
      "grad_norm": 2.03125,
      "learning_rate": 7.909933674160118e-06,
      "loss": 1.3665,
      "mean_token_accuracy": 0.7948504811887983,
      "mean_token_accuracy_utility": 0.7907601108356854,
      "step": 1650
    },
    {
      "epoch": 0.61228264890862,
      "grad_norm": 2.265625,
      "learning_rate": 7.846815534859592e-06,
      "loss": 1.415,
      "mean_token_accuracy": 0.798837663367585,
      "mean_token_accuracy_utility": 0.7817465613685279,
      "step": 1655
    },
    {
      "epoch": 0.6141324454310025,
      "grad_norm": 1.9453125,
      "learning_rate": 7.7837872195487e-06,
      "loss": 1.2852,
      "mean_token_accuracy": 0.8020185296915073,
      "mean_token_accuracy_utility": 0.8063624177216726,
      "step": 1660
    },
    {
      "epoch": 0.6159822419533851,
      "grad_norm": 2.125,
      "learning_rate": 7.720851357567778e-06,
      "loss": 1.3518,
      "mean_token_accuracy": 0.8088078625446713,
      "mean_token_accuracy_utility": 0.7854860897574434,
      "step": 1665
    },
    {
      "epoch": 0.6178320384757676,
      "grad_norm": 2.1875,
      "learning_rate": 7.658010574400285e-06,
      "loss": 1.3655,
      "mean_token_accuracy": 0.8060380565301009,
      "mean_token_accuracy_utility": 0.7892182639091554,
      "step": 1670
    },
    {
      "epoch": 0.6196818349981502,
      "grad_norm": 2.21875,
      "learning_rate": 7.595267491563311e-06,
      "loss": 1.3272,
      "mean_token_accuracy": 0.8061799237733718,
      "mean_token_accuracy_utility": 0.7924209078358079,
      "step": 1675
    },
    {
      "epoch": 0.6215316315205327,
      "grad_norm": 2.203125,
      "learning_rate": 7.532624726498181e-06,
      "loss": 1.3668,
      "mean_token_accuracy": 0.8012336562621154,
      "mean_token_accuracy_utility": 0.7901877497852416,
      "step": 1680
    },
    {
      "epoch": 0.6233814280429153,
      "grad_norm": 2.046875,
      "learning_rate": 7.470084892461305e-06,
      "loss": 1.3787,
      "mean_token_accuracy": 0.7948494546003289,
      "mean_token_accuracy_utility": 0.7910775347357658,
      "step": 1685
    },
    {
      "epoch": 0.6252312245652978,
      "grad_norm": 1.9921875,
      "learning_rate": 7.407650598415119e-06,
      "loss": 1.3347,
      "mean_token_accuracy": 0.8011234683285204,
      "mean_token_accuracy_utility": 0.7951748180619677,
      "step": 1690
    },
    {
      "epoch": 0.6270810210876804,
      "grad_norm": 2.140625,
      "learning_rate": 7.34532444891928e-06,
      "loss": 1.3563,
      "mean_token_accuracy": 0.8046421443358713,
      "mean_token_accuracy_utility": 0.7920589701115381,
      "step": 1695
    },
    {
      "epoch": 0.6289308176100629,
      "grad_norm": 2.109375,
      "learning_rate": 7.283109044021987e-06,
      "loss": 1.4774,
      "mean_token_accuracy": 0.7827012290982961,
      "mean_token_accuracy_utility": 0.7775017189480519,
      "step": 1700
    },
    {
      "epoch": 0.6289308176100629,
      "eval_mean_token_accuracy": 0.7925760950578022,
      "eval_reasoning_loss": 0.6487041115760803,
      "eval_reasoning_runtime": 90.6006,
      "eval_reasoning_samples_per_second": 1.424,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 1700
    },
    {
      "epoch": 0.6289308176100629,
      "eval_mean_token_accuracy": 0.7411323225698231,
      "eval_utility_loss": 0.9562034010887146,
      "eval_utility_runtime": 34.0868,
      "eval_utility_samples_per_second": 1.438,
      "eval_utility_steps_per_second": 0.381,
      "step": 1700
    },
    {
      "epoch": 0.6307806141324455,
      "grad_norm": 2.25,
      "learning_rate": 7.221006979151546e-06,
      "loss": 1.3762,
      "mean_token_accuracy": 0.800469282206661,
      "mean_token_accuracy_utility": 0.7927948731942174,
      "step": 1705
    },
    {
      "epoch": 0.632630410654828,
      "grad_norm": 2.25,
      "learning_rate": 7.159020845008056e-06,
      "loss": 1.449,
      "mean_token_accuracy": 0.7809441330604621,
      "mean_token_accuracy_utility": 0.7842533883434072,
      "step": 1710
    },
    {
      "epoch": 0.6344802071772105,
      "grad_norm": 2.09375,
      "learning_rate": 7.097153227455379e-06,
      "loss": 1.3603,
      "mean_token_accuracy": 0.8030618625070838,
      "mean_token_accuracy_utility": 0.7863666156726808,
      "step": 1715
    },
    {
      "epoch": 0.636330003699593,
      "grad_norm": 2.1875,
      "learning_rate": 7.035406707413221e-06,
      "loss": 1.3875,
      "mean_token_accuracy": 0.7920137236068808,
      "mean_token_accuracy_utility": 0.7910021044098411,
      "step": 1720
    },
    {
      "epoch": 0.6381798002219756,
      "grad_norm": 1.9921875,
      "learning_rate": 6.973783860749499e-06,
      "loss": 1.3482,
      "mean_token_accuracy": 0.7998231174477181,
      "mean_token_accuracy_utility": 0.7929809072492697,
      "step": 1725
    },
    {
      "epoch": 0.6400295967443581,
      "grad_norm": 1.921875,
      "learning_rate": 6.912287258172869e-06,
      "loss": 1.2612,
      "mean_token_accuracy": 0.8214457849269147,
      "mean_token_accuracy_utility": 0.7948503455614556,
      "step": 1730
    },
    {
      "epoch": 0.6418793932667407,
      "grad_norm": 2.109375,
      "learning_rate": 6.8509194651254825e-06,
      "loss": 1.4422,
      "mean_token_accuracy": 0.7937506263068088,
      "mean_token_accuracy_utility": 0.7816574067328353,
      "step": 1735
    },
    {
      "epoch": 0.6437291897891232,
      "grad_norm": 2.125,
      "learning_rate": 6.789683041675977e-06,
      "loss": 1.3441,
      "mean_token_accuracy": 0.79584180663546,
      "mean_token_accuracy_utility": 0.7966595718333576,
      "step": 1740
    },
    {
      "epoch": 0.6455789863115058,
      "grad_norm": 2.03125,
      "learning_rate": 6.7285805424126585e-06,
      "loss": 1.3267,
      "mean_token_accuracy": 0.7943188661507775,
      "mean_token_accuracy_utility": 0.8018615991654967,
      "step": 1745
    },
    {
      "epoch": 0.6474287828338883,
      "grad_norm": 2.109375,
      "learning_rate": 6.667614516336949e-06,
      "loss": 1.3371,
      "mean_token_accuracy": 0.8022821706238247,
      "mean_token_accuracy_utility": 0.7979826685517469,
      "step": 1750
    },
    {
      "epoch": 0.6492785793562708,
      "grad_norm": 2.0625,
      "learning_rate": 6.60678750675704e-06,
      "loss": 1.3819,
      "mean_token_accuracy": 0.7930077650713101,
      "mean_token_accuracy_utility": 0.7919821706658248,
      "step": 1755
    },
    {
      "epoch": 0.6511283758786534,
      "grad_norm": 2.015625,
      "learning_rate": 6.546102051181808e-06,
      "loss": 1.381,
      "mean_token_accuracy": 0.8141920458544218,
      "mean_token_accuracy_utility": 0.7832829003877934,
      "step": 1760
    },
    {
      "epoch": 0.6529781724010358,
      "grad_norm": 2.1875,
      "learning_rate": 6.485560681214933e-06,
      "loss": 1.3716,
      "mean_token_accuracy": 0.7986495252052358,
      "mean_token_accuracy_utility": 0.7920585742830759,
      "step": 1765
    },
    {
      "epoch": 0.6548279689234184,
      "grad_norm": 2.203125,
      "learning_rate": 6.425165922449319e-06,
      "loss": 1.3662,
      "mean_token_accuracy": 0.7924919475987822,
      "mean_token_accuracy_utility": 0.7963610155137015,
      "step": 1770
    },
    {
      "epoch": 0.6566777654458009,
      "grad_norm": 2.1875,
      "learning_rate": 6.364920294361701e-06,
      "loss": 1.3215,
      "mean_token_accuracy": 0.8007657524785177,
      "mean_token_accuracy_utility": 0.8011471607651757,
      "step": 1775
    },
    {
      "epoch": 0.6585275619681835,
      "grad_norm": 2.0625,
      "learning_rate": 6.3048263102075766e-06,
      "loss": 1.3229,
      "mean_token_accuracy": 0.7981398244509355,
      "mean_token_accuracy_utility": 0.7997776575698075,
      "step": 1780
    },
    {
      "epoch": 0.660377358490566,
      "grad_norm": 2.109375,
      "learning_rate": 6.244886476916325e-06,
      "loss": 1.3866,
      "mean_token_accuracy": 0.7973975781183975,
      "mean_token_accuracy_utility": 0.7849776024448584,
      "step": 1785
    },
    {
      "epoch": 0.6622271550129486,
      "grad_norm": 2.203125,
      "learning_rate": 6.185103294986654e-06,
      "loss": 1.4423,
      "mean_token_accuracy": 0.8011142273890952,
      "mean_token_accuracy_utility": 0.7821272314008417,
      "step": 1790
    },
    {
      "epoch": 0.6640769515353311,
      "grad_norm": 2.0625,
      "learning_rate": 6.125479258382268e-06,
      "loss": 1.3986,
      "mean_token_accuracy": 0.7999277288072351,
      "mean_token_accuracy_utility": 0.7780036991314967,
      "step": 1795
    },
    {
      "epoch": 0.6659267480577137,
      "grad_norm": 1.9765625,
      "learning_rate": 6.066016854427846e-06,
      "loss": 1.2981,
      "mean_token_accuracy": 0.8043779214968637,
      "mean_token_accuracy_utility": 0.8018171473392608,
      "step": 1800
    },
    {
      "epoch": 0.6659267480577137,
      "eval_mean_token_accuracy": 0.7927238344290986,
      "eval_reasoning_loss": 0.6481964588165283,
      "eval_reasoning_runtime": 90.5563,
      "eval_reasoning_samples_per_second": 1.425,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 1800
    },
    {
      "epoch": 0.6659267480577137,
      "eval_mean_token_accuracy": 0.7410897931562227,
      "eval_utility_loss": 0.9560399055480957,
      "eval_utility_runtime": 34.0819,
      "eval_utility_samples_per_second": 1.438,
      "eval_utility_steps_per_second": 0.381,
      "step": 1800
    },
    {
      "epoch": 0.6677765445800962,
      "grad_norm": 2.109375,
      "learning_rate": 6.006718563705258e-06,
      "loss": 1.3551,
      "mean_token_accuracy": 0.7921908977854335,
      "mean_token_accuracy_utility": 0.7981032519209205,
      "step": 1805
    },
    {
      "epoch": 0.6696263411024788,
      "grad_norm": 1.984375,
      "learning_rate": 5.947586859950103e-06,
      "loss": 1.2888,
      "mean_token_accuracy": 0.8079180742997863,
      "mean_token_accuracy_utility": 0.803279863227635,
      "step": 1810
    },
    {
      "epoch": 0.6714761376248612,
      "grad_norm": 2.140625,
      "learning_rate": 5.888624209948495e-06,
      "loss": 1.4265,
      "mean_token_accuracy": 0.8013951035950253,
      "mean_token_accuracy_utility": 0.7796599074785686,
      "step": 1815
    },
    {
      "epoch": 0.6733259341472438,
      "grad_norm": 2.03125,
      "learning_rate": 5.82983307343416e-06,
      "loss": 1.2874,
      "mean_token_accuracy": 0.810361913380062,
      "mean_token_accuracy_utility": 0.799601371521528,
      "step": 1820
    },
    {
      "epoch": 0.6751757306696263,
      "grad_norm": 1.9296875,
      "learning_rate": 5.771215902985848e-06,
      "loss": 1.3267,
      "mean_token_accuracy": 0.8060440499758691,
      "mean_token_accuracy_utility": 0.7918362396418577,
      "step": 1825
    },
    {
      "epoch": 0.6770255271920089,
      "grad_norm": 2.171875,
      "learning_rate": 5.712775143924978e-06,
      "loss": 1.36,
      "mean_token_accuracy": 0.8106614665200743,
      "mean_token_accuracy_utility": 0.7825484530844851,
      "step": 1830
    },
    {
      "epoch": 0.6788753237143914,
      "grad_norm": 2.0625,
      "learning_rate": 5.654513234213663e-06,
      "loss": 1.3151,
      "mean_token_accuracy": 0.8031215377826865,
      "mean_token_accuracy_utility": 0.8017431223213999,
      "step": 1835
    },
    {
      "epoch": 0.680725120236774,
      "grad_norm": 2.296875,
      "learning_rate": 5.596432604352982e-06,
      "loss": 1.3964,
      "mean_token_accuracy": 0.8083849954406773,
      "mean_token_accuracy_utility": 0.7758357538198013,
      "step": 1840
    },
    {
      "epoch": 0.6825749167591565,
      "grad_norm": 2.015625,
      "learning_rate": 5.538535677281608e-06,
      "loss": 1.3838,
      "mean_token_accuracy": 0.7987950191401969,
      "mean_token_accuracy_utility": 0.7878698132405043,
      "step": 1845
    },
    {
      "epoch": 0.684424713281539,
      "grad_norm": 2.15625,
      "learning_rate": 5.480824868274699e-06,
      "loss": 1.3747,
      "mean_token_accuracy": 0.7994545545779098,
      "mean_token_accuracy_utility": 0.7903508432204391,
      "step": 1850
    },
    {
      "epoch": 0.6862745098039216,
      "grad_norm": 2.21875,
      "learning_rate": 5.423302584843186e-06,
      "loss": 1.4335,
      "mean_token_accuracy": 0.794139923963461,
      "mean_token_accuracy_utility": 0.7803640946969455,
      "step": 1855
    },
    {
      "epoch": 0.6881243063263041,
      "grad_norm": 2.0625,
      "learning_rate": 5.365971226633294e-06,
      "loss": 1.3243,
      "mean_token_accuracy": 0.8104008240426811,
      "mean_token_accuracy_utility": 0.7911373692734933,
      "step": 1860
    },
    {
      "epoch": 0.6899741028486867,
      "grad_norm": 2.40625,
      "learning_rate": 5.308833185326472e-06,
      "loss": 1.4114,
      "mean_token_accuracy": 0.7990287645261619,
      "mean_token_accuracy_utility": 0.7838049535159783,
      "step": 1865
    },
    {
      "epoch": 0.6918238993710691,
      "grad_norm": 2.1875,
      "learning_rate": 5.25189084453959e-06,
      "loss": 1.3788,
      "mean_token_accuracy": 0.8002921145313044,
      "mean_token_accuracy_utility": 0.7829996575439292,
      "step": 1870
    },
    {
      "epoch": 0.6936736958934517,
      "grad_norm": 2.046875,
      "learning_rate": 5.19514657972553e-06,
      "loss": 1.3009,
      "mean_token_accuracy": 0.8165886986414606,
      "mean_token_accuracy_utility": 0.7965793256125144,
      "step": 1875
    },
    {
      "epoch": 0.6955234924158342,
      "grad_norm": 2.0,
      "learning_rate": 5.138602758074062e-06,
      "loss": 1.2897,
      "mean_token_accuracy": 0.816310115770124,
      "mean_token_accuracy_utility": 0.7942299658365094,
      "step": 1880
    },
    {
      "epoch": 0.6973732889382168,
      "grad_norm": 2.328125,
      "learning_rate": 5.082261738413124e-06,
      "loss": 1.3785,
      "mean_token_accuracy": 0.7889815777939976,
      "mean_token_accuracy_utility": 0.7948403247802525,
      "step": 1885
    },
    {
      "epoch": 0.6992230854605993,
      "grad_norm": 2.15625,
      "learning_rate": 5.026125871110383e-06,
      "loss": 1.4074,
      "mean_token_accuracy": 0.8004481689204509,
      "mean_token_accuracy_utility": 0.7824528460866943,
      "step": 1890
    },
    {
      "epoch": 0.7010728819829819,
      "grad_norm": 2.0,
      "learning_rate": 4.970197497975216e-06,
      "loss": 1.362,
      "mean_token_accuracy": 0.8110918535537721,
      "mean_token_accuracy_utility": 0.7837108926939016,
      "step": 1895
    },
    {
      "epoch": 0.7029226785053644,
      "grad_norm": 1.984375,
      "learning_rate": 4.914478952160997e-06,
      "loss": 1.3648,
      "mean_token_accuracy": 0.7967510057137687,
      "mean_token_accuracy_utility": 0.7993606696799402,
      "step": 1900
    },
    {
      "epoch": 0.7029226785053644,
      "eval_mean_token_accuracy": 0.7926497609823351,
      "eval_reasoning_loss": 0.6480225324630737,
      "eval_reasoning_runtime": 90.6447,
      "eval_reasoning_samples_per_second": 1.423,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 1900
    },
    {
      "epoch": 0.7029226785053644,
      "eval_mean_token_accuracy": 0.7404904043716379,
      "eval_utility_loss": 0.9595364332199097,
      "eval_utility_runtime": 34.0998,
      "eval_utility_samples_per_second": 1.437,
      "eval_utility_steps_per_second": 0.381,
      "step": 1900
    },
    {
      "epoch": 0.704772475027747,
      "grad_norm": 2.015625,
      "learning_rate": 4.858972558067784e-06,
      "loss": 1.3013,
      "mean_token_accuracy": 0.8169469603513704,
      "mean_token_accuracy_utility": 0.791147850289918,
      "step": 1905
    },
    {
      "epoch": 0.7066222715501295,
      "grad_norm": 2.0,
      "learning_rate": 4.803680631245333e-06,
      "loss": 1.2684,
      "mean_token_accuracy": 0.812279459297873,
      "mean_token_accuracy_utility": 0.8049015418289768,
      "step": 1910
    },
    {
      "epoch": 0.7084720680725121,
      "grad_norm": 1.9140625,
      "learning_rate": 4.748605478296508e-06,
      "loss": 1.3021,
      "mean_token_accuracy": 0.8121748492190749,
      "mean_token_accuracy_utility": 0.796444983216289,
      "step": 1915
    },
    {
      "epoch": 0.7103218645948945,
      "grad_norm": 2.171875,
      "learning_rate": 4.6937493967810665e-06,
      "loss": 1.3939,
      "mean_token_accuracy": 0.8060775498081345,
      "mean_token_accuracy_utility": 0.7818387532762099,
      "step": 1920
    },
    {
      "epoch": 0.7121716611172771,
      "grad_norm": 2.28125,
      "learning_rate": 4.639114675119797e-06,
      "loss": 1.422,
      "mean_token_accuracy": 0.7926502701694929,
      "mean_token_accuracy_utility": 0.7831150168941006,
      "step": 1925
    },
    {
      "epoch": 0.7140214576396596,
      "grad_norm": 2.125,
      "learning_rate": 4.584703592499068e-06,
      "loss": 1.3563,
      "mean_token_accuracy": 0.8017924338641181,
      "mean_token_accuracy_utility": 0.7917728371775985,
      "step": 1930
    },
    {
      "epoch": 0.7158712541620422,
      "grad_norm": 2.03125,
      "learning_rate": 4.530518418775734e-06,
      "loss": 1.3377,
      "mean_token_accuracy": 0.803092108837126,
      "mean_token_accuracy_utility": 0.7951072801794123,
      "step": 1935
    },
    {
      "epoch": 0.7177210506844247,
      "grad_norm": 1.953125,
      "learning_rate": 4.476561414382451e-06,
      "loss": 1.2108,
      "mean_token_accuracy": 0.8194250033746033,
      "mean_token_accuracy_utility": 0.8136789162922969,
      "step": 1940
    },
    {
      "epoch": 0.7195708472068072,
      "grad_norm": 2.046875,
      "learning_rate": 4.422834830233378e-06,
      "loss": 1.3177,
      "mean_token_accuracy": 0.8047896269310687,
      "mean_token_accuracy_utility": 0.7989028654635364,
      "step": 1945
    },
    {
      "epoch": 0.7214206437291898,
      "grad_norm": 2.46875,
      "learning_rate": 4.369340907630284e-06,
      "loss": 1.3774,
      "mean_token_accuracy": 0.7926865233429574,
      "mean_token_accuracy_utility": 0.7936197535699014,
      "step": 1950
    },
    {
      "epoch": 0.7232704402515723,
      "grad_norm": 2.078125,
      "learning_rate": 4.3160818781690286e-06,
      "loss": 1.3404,
      "mean_token_accuracy": 0.7978618710010525,
      "mean_token_accuracy_utility": 0.7989650159989556,
      "step": 1955
    },
    {
      "epoch": 0.7251202367739549,
      "grad_norm": 2.171875,
      "learning_rate": 4.26305996364649e-06,
      "loss": 1.3653,
      "mean_token_accuracy": 0.8059205921387333,
      "mean_token_accuracy_utility": 0.7875991148527812,
      "step": 1960
    },
    {
      "epoch": 0.7269700332963374,
      "grad_norm": 2.03125,
      "learning_rate": 4.210277375967855e-06,
      "loss": 1.31,
      "mean_token_accuracy": 0.7999238055921607,
      "mean_token_accuracy_utility": 0.8002782194020931,
      "step": 1965
    },
    {
      "epoch": 0.72881982981872,
      "grad_norm": 2.171875,
      "learning_rate": 4.157736317054373e-06,
      "loss": 1.4154,
      "mean_token_accuracy": 0.8005453326519281,
      "mean_token_accuracy_utility": 0.7803789151261222,
      "step": 1970
    },
    {
      "epoch": 0.7306696263411024,
      "grad_norm": 1.9140625,
      "learning_rate": 4.105438978751465e-06,
      "loss": 1.3837,
      "mean_token_accuracy": 0.8065559599030621,
      "mean_token_accuracy_utility": 0.7778438778428487,
      "step": 1975
    },
    {
      "epoch": 0.732519422863485,
      "grad_norm": 2.078125,
      "learning_rate": 4.053387542737323e-06,
      "loss": 1.3092,
      "mean_token_accuracy": 0.8063409217517858,
      "mean_token_accuracy_utility": 0.802143084769485,
      "step": 1980
    },
    {
      "epoch": 0.7343692193858675,
      "grad_norm": 2.140625,
      "learning_rate": 4.001584180431869e-06,
      "loss": 1.3452,
      "mean_token_accuracy": 0.8110673841751822,
      "mean_token_accuracy_utility": 0.7885131195623156,
      "step": 1985
    },
    {
      "epoch": 0.7362190159082501,
      "grad_norm": 2.09375,
      "learning_rate": 3.950031052906183e-06,
      "loss": 1.3559,
      "mean_token_accuracy": 0.8041449058507519,
      "mean_token_accuracy_utility": 0.7894576914910736,
      "step": 1990
    },
    {
      "epoch": 0.7380688124306326,
      "grad_norm": 2.125,
      "learning_rate": 3.898730310792346e-06,
      "loss": 1.337,
      "mean_token_accuracy": 0.7984253456910807,
      "mean_token_accuracy_utility": 0.7999499578044301,
      "step": 1995
    },
    {
      "epoch": 0.7399186089530152,
      "grad_norm": 2.171875,
      "learning_rate": 3.847684094193733e-06,
      "loss": 1.3258,
      "mean_token_accuracy": 0.7939965494239809,
      "mean_token_accuracy_utility": 0.8017990482964082,
      "step": 2000
    },
    {
      "epoch": 0.7399186089530152,
      "eval_mean_token_accuracy": 0.7926664751110114,
      "eval_reasoning_loss": 0.6478474736213684,
      "eval_reasoning_runtime": 90.6353,
      "eval_reasoning_samples_per_second": 1.423,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 2000
    },
    {
      "epoch": 0.7399186089530152,
      "eval_mean_token_accuracy": 0.7404963659243423,
      "eval_utility_loss": 0.9593599438667297,
      "eval_utility_runtime": 34.1146,
      "eval_utility_samples_per_second": 1.436,
      "eval_utility_steps_per_second": 0.381,
      "step": 2000
    },
    {
      "epoch": 0.7417684054753977,
      "grad_norm": 2.15625,
      "learning_rate": 3.7968945325957175e-06,
      "loss": 1.3748,
      "mean_token_accuracy": 0.7967967103718052,
      "mean_token_accuracy_utility": 0.7930525367100172,
      "step": 2005
    },
    {
      "epoch": 0.7436182019977803,
      "grad_norm": 2.171875,
      "learning_rate": 3.746363744776844e-06,
      "loss": 1.3734,
      "mean_token_accuracy": 0.7963965004034452,
      "mean_token_accuracy_utility": 0.7905338918354674,
      "step": 2010
    },
    {
      "epoch": 0.7454679985201628,
      "grad_norm": 2.03125,
      "learning_rate": 3.69609383872045e-06,
      "loss": 1.3541,
      "mean_token_accuracy": 0.8025840210210049,
      "mean_token_accuracy_utility": 0.7892266839570831,
      "step": 2015
    },
    {
      "epoch": 0.7473177950425454,
      "grad_norm": 2.015625,
      "learning_rate": 3.646086911526706e-06,
      "loss": 1.2669,
      "mean_token_accuracy": 0.8091728387604407,
      "mean_token_accuracy_utility": 0.8044403262517277,
      "step": 2020
    },
    {
      "epoch": 0.7491675915649278,
      "grad_norm": 2.125,
      "learning_rate": 3.5963450493251552e-06,
      "loss": 1.2636,
      "mean_token_accuracy": 0.7967616553790182,
      "mean_token_accuracy_utility": 0.8169434278498944,
      "step": 2025
    },
    {
      "epoch": 0.7510173880873104,
      "grad_norm": 2.109375,
      "learning_rate": 3.5468703271876626e-06,
      "loss": 1.3113,
      "mean_token_accuracy": 0.8110927693336019,
      "mean_token_accuracy_utility": 0.7949574657901184,
      "step": 2030
    },
    {
      "epoch": 0.7528671846096929,
      "grad_norm": 2.390625,
      "learning_rate": 3.4976648090418685e-06,
      "loss": 1.3721,
      "mean_token_accuracy": 0.8047257112541091,
      "mean_token_accuracy_utility": 0.7863134667719687,
      "step": 2035
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 2.125,
      "learning_rate": 3.4487305475850754e-06,
      "loss": 1.4658,
      "mean_token_accuracy": 0.7983925250945354,
      "mean_token_accuracy_utility": 0.7714898902944402,
      "step": 2040
    },
    {
      "epoch": 0.756566777654458,
      "grad_norm": 2.109375,
      "learning_rate": 3.400069584198633e-06,
      "loss": 1.4057,
      "mean_token_accuracy": 0.8024300576299673,
      "mean_token_accuracy_utility": 0.7813070494785829,
      "step": 2045
    },
    {
      "epoch": 0.7584165741768405,
      "grad_norm": 2.140625,
      "learning_rate": 3.351683948862753e-06,
      "loss": 1.3564,
      "mean_token_accuracy": 0.794366387565353,
      "mean_token_accuracy_utility": 0.7911765410780138,
      "step": 2050
    },
    {
      "epoch": 0.7602663706992231,
      "grad_norm": 2.046875,
      "learning_rate": 3.3035756600718515e-06,
      "loss": 1.3353,
      "mean_token_accuracy": 0.8062225161511398,
      "mean_token_accuracy_utility": 0.7922772133044487,
      "step": 2055
    },
    {
      "epoch": 0.7621161672216056,
      "grad_norm": 2.140625,
      "learning_rate": 3.2557467247503205e-06,
      "loss": 1.3387,
      "mean_token_accuracy": 0.8010375761296924,
      "mean_token_accuracy_utility": 0.7986957777608679,
      "step": 2060
    },
    {
      "epoch": 0.7639659637439882,
      "grad_norm": 2.125,
      "learning_rate": 3.208199138168826e-06,
      "loss": 1.3978,
      "mean_token_accuracy": 0.7886469130371976,
      "mean_token_accuracy_utility": 0.7945956422287713,
      "step": 2065
    },
    {
      "epoch": 0.7658157602663707,
      "grad_norm": 1.953125,
      "learning_rate": 3.160934883861053e-06,
      "loss": 1.3262,
      "mean_token_accuracy": 0.7958408629022024,
      "mean_token_accuracy_utility": 0.8026362481205386,
      "step": 2070
    },
    {
      "epoch": 0.7676655567887533,
      "grad_norm": 2.09375,
      "learning_rate": 3.113955933540973e-06,
      "loss": 1.3006,
      "mean_token_accuracy": 0.8210821636859421,
      "mean_token_accuracy_utility": 0.7906894871843215,
      "step": 2075
    },
    {
      "epoch": 0.7695153533111357,
      "grad_norm": 2.0625,
      "learning_rate": 3.0672642470205794e-06,
      "loss": 1.4577,
      "mean_token_accuracy": 0.8035695254739089,
      "mean_token_accuracy_utility": 0.7741234719985413,
      "step": 2080
    },
    {
      "epoch": 0.7713651498335183,
      "grad_norm": 2.203125,
      "learning_rate": 3.020861772128145e-06,
      "loss": 1.3804,
      "mean_token_accuracy": 0.7874552773479941,
      "mean_token_accuracy_utility": 0.7928091662420463,
      "step": 2085
    },
    {
      "epoch": 0.7732149463559008,
      "grad_norm": 2.03125,
      "learning_rate": 2.9747504446269447e-06,
      "loss": 1.3256,
      "mean_token_accuracy": 0.8075855806970814,
      "mean_token_accuracy_utility": 0.7939750857455345,
      "step": 2090
    },
    {
      "epoch": 0.7750647428782834,
      "grad_norm": 1.9453125,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 1.3687,
      "mean_token_accuracy": 0.7967592996787118,
      "mean_token_accuracy_utility": 0.7954397631980554,
      "step": 2095
    },
    {
      "epoch": 0.7769145394006659,
      "grad_norm": 2.203125,
      "learning_rate": 2.883408914042435e-06,
      "loss": 1.3573,
      "mean_token_accuracy": 0.7891181235708553,
      "mean_token_accuracy_utility": 0.8007137579181836,
      "step": 2100
    },
    {
      "epoch": 0.7769145394006659,
      "eval_mean_token_accuracy": 0.7927830862773846,
      "eval_reasoning_loss": 0.6477451920509338,
      "eval_reasoning_runtime": 90.7226,
      "eval_reasoning_samples_per_second": 1.422,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 2100
    },
    {
      "epoch": 0.7769145394006659,
      "eval_mean_token_accuracy": 0.7405307024319199,
      "eval_utility_loss": 0.9600812196731567,
      "eval_utility_runtime": 34.152,
      "eval_utility_samples_per_second": 1.435,
      "eval_utility_steps_per_second": 0.381,
      "step": 2100
    },
    {
      "epoch": 0.7787643359230485,
      "grad_norm": 2.09375,
      "learning_rate": 2.838182521436498e-06,
      "loss": 1.3652,
      "mean_token_accuracy": 0.8023040908848087,
      "mean_token_accuracy_utility": 0.7872767604923137,
      "step": 2105
    },
    {
      "epoch": 0.780614132445431,
      "grad_norm": 2.0625,
      "learning_rate": 2.7932548970175977e-06,
      "loss": 1.3208,
      "mean_token_accuracy": 0.7924853857875809,
      "mean_token_accuracy_utility": 0.8084556845017123,
      "step": 2110
    },
    {
      "epoch": 0.7824639289678136,
      "grad_norm": 1.984375,
      "learning_rate": 2.74862791502295e-06,
      "loss": 1.3144,
      "mean_token_accuracy": 0.8048684355597491,
      "mean_token_accuracy_utility": 0.7990936850885102,
      "step": 2115
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 2.1875,
      "learning_rate": 2.7043034371479404e-06,
      "loss": 1.2926,
      "mean_token_accuracy": 0.7954699238211357,
      "mean_token_accuracy_utility": 0.8111990657456694,
      "step": 2120
    },
    {
      "epoch": 0.7861635220125787,
      "grad_norm": 2.015625,
      "learning_rate": 2.660283312468438e-06,
      "loss": 1.3208,
      "mean_token_accuracy": 0.8172235943874242,
      "mean_token_accuracy_utility": 0.7883929906922252,
      "step": 2125
    },
    {
      "epoch": 0.7880133185349611,
      "grad_norm": 2.140625,
      "learning_rate": 2.616569377363668e-06,
      "loss": 1.4205,
      "mean_token_accuracy": 0.7905158388772555,
      "mean_token_accuracy_utility": 0.7875326225439481,
      "step": 2130
    },
    {
      "epoch": 0.7898631150573437,
      "grad_norm": 2.125,
      "learning_rate": 2.573163455439601e-06,
      "loss": 1.3711,
      "mean_token_accuracy": 0.7985778334570766,
      "mean_token_accuracy_utility": 0.7957669823513743,
      "step": 2135
    },
    {
      "epoch": 0.7917129115797262,
      "grad_norm": 2.171875,
      "learning_rate": 2.530067357452892e-06,
      "loss": 1.4555,
      "mean_token_accuracy": 0.8006896656818215,
      "mean_token_accuracy_utility": 0.7717196724516585,
      "step": 2140
    },
    {
      "epoch": 0.7935627081021087,
      "grad_norm": 1.9765625,
      "learning_rate": 2.4872828812353146e-06,
      "loss": 1.2232,
      "mean_token_accuracy": 0.8080836491588498,
      "mean_token_accuracy_utility": 0.8156791995136394,
      "step": 2145
    },
    {
      "epoch": 0.7954125046244913,
      "grad_norm": 2.140625,
      "learning_rate": 2.4448118116187847e-06,
      "loss": 1.3432,
      "mean_token_accuracy": 0.7915575267410201,
      "mean_token_accuracy_utility": 0.7986785520962193,
      "step": 2150
    },
    {
      "epoch": 0.7972623011468738,
      "grad_norm": 2.171875,
      "learning_rate": 2.402655920360889e-06,
      "loss": 1.3247,
      "mean_token_accuracy": 0.8039421585804896,
      "mean_token_accuracy_utility": 0.799006137253576,
      "step": 2155
    },
    {
      "epoch": 0.7991120976692564,
      "grad_norm": 2.078125,
      "learning_rate": 2.360816966070986e-06,
      "loss": 1.2956,
      "mean_token_accuracy": 0.8050194155837609,
      "mean_token_accuracy_utility": 0.802147858420146,
      "step": 2160
    },
    {
      "epoch": 0.8009618941916389,
      "grad_norm": 2.109375,
      "learning_rate": 2.3192966941368247e-06,
      "loss": 1.3668,
      "mean_token_accuracy": 0.7982761303754906,
      "mean_token_accuracy_utility": 0.7912824589491727,
      "step": 2165
    },
    {
      "epoch": 0.8028116907140215,
      "grad_norm": 1.9765625,
      "learning_rate": 2.2780968366517476e-06,
      "loss": 1.3612,
      "mean_token_accuracy": 0.8149195683674229,
      "mean_token_accuracy_utility": 0.7790149146677751,
      "step": 2170
    },
    {
      "epoch": 0.804661487236404,
      "grad_norm": 2.03125,
      "learning_rate": 2.237219112342426e-06,
      "loss": 1.3589,
      "mean_token_accuracy": 0.8043794254361968,
      "mean_token_accuracy_utility": 0.787811558101065,
      "step": 2175
    },
    {
      "epoch": 0.8065112837587866,
      "grad_norm": 2.109375,
      "learning_rate": 2.196665226497169e-06,
      "loss": 1.3838,
      "mean_token_accuracy": 0.8005341453598236,
      "mean_token_accuracy_utility": 0.7861639292843782,
      "step": 2180
    },
    {
      "epoch": 0.808361080281169,
      "grad_norm": 2.171875,
      "learning_rate": 2.156436870894767e-06,
      "loss": 1.4209,
      "mean_token_accuracy": 0.7923612984972455,
      "mean_token_accuracy_utility": 0.7835607670987231,
      "step": 2185
    },
    {
      "epoch": 0.8102108768035516,
      "grad_norm": 2.0625,
      "learning_rate": 2.116535723733938e-06,
      "loss": 1.262,
      "mean_token_accuracy": 0.8117126230214062,
      "mean_token_accuracy_utility": 0.8011139000795223,
      "step": 2190
    },
    {
      "epoch": 0.8120606733259341,
      "grad_norm": 2.203125,
      "learning_rate": 2.0769634495632986e-06,
      "loss": 1.4021,
      "mean_token_accuracy": 0.7999497387598418,
      "mean_token_accuracy_utility": 0.7852134152506763,
      "step": 2195
    },
    {
      "epoch": 0.8139104698483167,
      "grad_norm": 2.203125,
      "learning_rate": 2.037721699211933e-06,
      "loss": 1.3998,
      "mean_token_accuracy": 0.7937094446453241,
      "mean_token_accuracy_utility": 0.7915124776647481,
      "step": 2200
    },
    {
      "epoch": 0.8139104698483167,
      "eval_mean_token_accuracy": 0.7928515538377074,
      "eval_reasoning_loss": 0.6475924849510193,
      "eval_reasoning_runtime": 90.744,
      "eval_reasoning_samples_per_second": 1.422,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 2200
    },
    {
      "epoch": 0.8139104698483167,
      "eval_mean_token_accuracy": 0.7407037174073458,
      "eval_utility_loss": 0.959997296333313,
      "eval_utility_runtime": 34.1415,
      "eval_utility_samples_per_second": 1.435,
      "eval_utility_steps_per_second": 0.381,
      "step": 2200
    },
    {
      "epoch": 0.8157602663706992,
      "grad_norm": 2.015625,
      "learning_rate": 1.998812109720535e-06,
      "loss": 1.3143,
      "mean_token_accuracy": 0.7974378966294962,
      "mean_token_accuracy_utility": 0.8035489760691596,
      "step": 2205
    },
    {
      "epoch": 0.8176100628930818,
      "grad_norm": 2.078125,
      "learning_rate": 1.9602363042730964e-06,
      "loss": 1.3689,
      "mean_token_accuracy": 0.7911903721542584,
      "mean_token_accuracy_utility": 0.7967285543772705,
      "step": 2210
    },
    {
      "epoch": 0.8194598594154643,
      "grad_norm": 2.09375,
      "learning_rate": 1.921995892129208e-06,
      "loss": 1.2611,
      "mean_token_accuracy": 0.8023804961466252,
      "mean_token_accuracy_utility": 0.8132386043005694,
      "step": 2215
    },
    {
      "epoch": 0.8213096559378469,
      "grad_norm": 2.140625,
      "learning_rate": 1.8840924685569162e-06,
      "loss": 1.3151,
      "mean_token_accuracy": 0.8082170759568686,
      "mean_token_accuracy_utility": 0.7960541661322054,
      "step": 2220
    },
    {
      "epoch": 0.8231594524602294,
      "grad_norm": 2.21875,
      "learning_rate": 1.8465276147661905e-06,
      "loss": 1.2977,
      "mean_token_accuracy": 0.8140005525237524,
      "mean_token_accuracy_utility": 0.7939823559672725,
      "step": 2225
    },
    {
      "epoch": 0.825009248982612,
      "grad_norm": 2.109375,
      "learning_rate": 1.809302897842934e-06,
      "loss": 1.3452,
      "mean_token_accuracy": 0.7984935464918711,
      "mean_token_accuracy_utility": 0.8005471678301719,
      "step": 2230
    },
    {
      "epoch": 0.8268590455049944,
      "grad_norm": 2.0625,
      "learning_rate": 1.7724198706836372e-06,
      "loss": 1.2693,
      "mean_token_accuracy": 0.8117025214187354,
      "mean_token_accuracy_utility": 0.8089383000044499,
      "step": 2235
    },
    {
      "epoch": 0.8287088420273769,
      "grad_norm": 2.140625,
      "learning_rate": 1.7358800719305735e-06,
      "loss": 1.3406,
      "mean_token_accuracy": 0.7947607438251226,
      "mean_token_accuracy_utility": 0.7974042841238483,
      "step": 2240
    },
    {
      "epoch": 0.8305586385497595,
      "grad_norm": 2.09375,
      "learning_rate": 1.6996850259076303e-06,
      "loss": 1.3858,
      "mean_token_accuracy": 0.8068708091191515,
      "mean_token_accuracy_utility": 0.7838768374402202,
      "step": 2245
    },
    {
      "epoch": 0.832408435072142,
      "grad_norm": 2.109375,
      "learning_rate": 1.6638362425567012e-06,
      "loss": 1.4251,
      "mean_token_accuracy": 0.7912954092210812,
      "mean_token_accuracy_utility": 0.779652357701173,
      "step": 2250
    },
    {
      "epoch": 0.8342582315945246,
      "grad_norm": 1.8984375,
      "learning_rate": 1.6283352173747148e-06,
      "loss": 1.3198,
      "mean_token_accuracy": 0.8026324256324904,
      "mean_token_accuracy_utility": 0.8023336567087199,
      "step": 2255
    },
    {
      "epoch": 0.8361080281169071,
      "grad_norm": 2.078125,
      "learning_rate": 1.5931834313512307e-06,
      "loss": 1.3898,
      "mean_token_accuracy": 0.7948547983426106,
      "mean_token_accuracy_utility": 0.7891938245395896,
      "step": 2260
    },
    {
      "epoch": 0.8379578246392897,
      "grad_norm": 2.03125,
      "learning_rate": 1.5583823509066665e-06,
      "loss": 1.3085,
      "mean_token_accuracy": 0.8014040128950051,
      "mean_token_accuracy_utility": 0.804059453821241,
      "step": 2265
    },
    {
      "epoch": 0.8398076211616722,
      "grad_norm": 2.125,
      "learning_rate": 1.5239334278311214e-06,
      "loss": 1.3026,
      "mean_token_accuracy": 0.8044230407588243,
      "mean_token_accuracy_utility": 0.8043497276564379,
      "step": 2270
    },
    {
      "epoch": 0.8416574176840548,
      "grad_norm": 2.09375,
      "learning_rate": 1.489838099223816e-06,
      "loss": 1.3249,
      "mean_token_accuracy": 0.8031085317499432,
      "mean_token_accuracy_utility": 0.7979779617881892,
      "step": 2275
    },
    {
      "epoch": 0.8435072142064373,
      "grad_norm": 1.984375,
      "learning_rate": 1.4560977874331284e-06,
      "loss": 1.2993,
      "mean_token_accuracy": 0.80747925146387,
      "mean_token_accuracy_utility": 0.8012276467728181,
      "step": 2280
    },
    {
      "epoch": 0.8453570107288199,
      "grad_norm": 2.125,
      "learning_rate": 1.4227138999972801e-06,
      "loss": 1.2921,
      "mean_token_accuracy": 0.8060897387019559,
      "mean_token_accuracy_utility": 0.8065221697056817,
      "step": 2285
    },
    {
      "epoch": 0.8472068072512023,
      "grad_norm": 2.265625,
      "learning_rate": 1.389687829585592e-06,
      "loss": 1.3884,
      "mean_token_accuracy": 0.7931530355887474,
      "mean_token_accuracy_utility": 0.7937198926471063,
      "step": 2290
    },
    {
      "epoch": 0.8490566037735849,
      "grad_norm": 2.171875,
      "learning_rate": 1.3570209539404067e-06,
      "loss": 1.3505,
      "mean_token_accuracy": 0.810897009218966,
      "mean_token_accuracy_utility": 0.7852190567936245,
      "step": 2295
    },
    {
      "epoch": 0.8509064002959674,
      "grad_norm": 2.1875,
      "learning_rate": 1.3247146358196073e-06,
      "loss": 1.384,
      "mean_token_accuracy": 0.7992858038117768,
      "mean_token_accuracy_utility": 0.7854254182993216,
      "step": 2300
    },
    {
      "epoch": 0.8509064002959674,
      "eval_mean_token_accuracy": 0.7927642956512375,
      "eval_reasoning_loss": 0.6476864814758301,
      "eval_reasoning_runtime": 90.7004,
      "eval_reasoning_samples_per_second": 1.422,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 2300
    },
    {
      "epoch": 0.8509064002959674,
      "eval_mean_token_accuracy": 0.7403685207798727,
      "eval_utility_loss": 0.9602022171020508,
      "eval_utility_runtime": 34.1528,
      "eval_utility_samples_per_second": 1.435,
      "eval_utility_steps_per_second": 0.381,
      "step": 2300
    },
    {
      "epoch": 0.85275619681835,
      "grad_norm": 1.9609375,
      "learning_rate": 1.2927702229397633e-06,
      "loss": 1.3234,
      "mean_token_accuracy": 0.79884900789307,
      "mean_token_accuracy_utility": 0.7990931723600154,
      "step": 2305
    },
    {
      "epoch": 0.8546059933407325,
      "grad_norm": 2.0,
      "learning_rate": 1.261189047919913e-06,
      "loss": 1.3343,
      "mean_token_accuracy": 0.7958460039033601,
      "mean_token_accuracy_utility": 0.8006394979490272,
      "step": 2310
    },
    {
      "epoch": 0.8564557898631151,
      "grad_norm": 2.09375,
      "learning_rate": 1.2299724282259685e-06,
      "loss": 1.3558,
      "mean_token_accuracy": 0.7961573445030325,
      "mean_token_accuracy_utility": 0.7921575495605173,
      "step": 2315
    },
    {
      "epoch": 0.8583055863854976,
      "grad_norm": 2.078125,
      "learning_rate": 1.1991216661157611e-06,
      "loss": 1.3341,
      "mean_token_accuracy": 0.8126465613539315,
      "mean_token_accuracy_utility": 0.791415381599222,
      "step": 2320
    },
    {
      "epoch": 0.8601553829078802,
      "grad_norm": 2.109375,
      "learning_rate": 1.1686380485847027e-06,
      "loss": 1.3225,
      "mean_token_accuracy": 0.7965170159134856,
      "mean_token_accuracy_utility": 0.8025845449765849,
      "step": 2325
    },
    {
      "epoch": 0.8620051794302627,
      "grad_norm": 2.03125,
      "learning_rate": 1.1385228473121113e-06,
      "loss": 1.2764,
      "mean_token_accuracy": 0.8150161819532178,
      "mean_token_accuracy_utility": 0.8009832261267468,
      "step": 2330
    },
    {
      "epoch": 0.8638549759526452,
      "grad_norm": 2.125,
      "learning_rate": 1.1087773186081474e-06,
      "loss": 1.3617,
      "mean_token_accuracy": 0.7948024073073308,
      "mean_token_accuracy_utility": 0.7977715757107269,
      "step": 2335
    },
    {
      "epoch": 0.8657047724750278,
      "grad_norm": 2.03125,
      "learning_rate": 1.0794027033614175e-06,
      "loss": 1.3102,
      "mean_token_accuracy": 0.8046025198952952,
      "mean_token_accuracy_utility": 0.8009814706046345,
      "step": 2340
    },
    {
      "epoch": 0.8675545689974102,
      "grad_norm": 2.0,
      "learning_rate": 1.0504002269871927e-06,
      "loss": 1.3224,
      "mean_token_accuracy": 0.8075330622268723,
      "mean_token_accuracy_utility": 0.796108317445136,
      "step": 2345
    },
    {
      "epoch": 0.8694043655197928,
      "grad_norm": 2.046875,
      "learning_rate": 1.0217710993763086e-06,
      "loss": 1.2603,
      "mean_token_accuracy": 0.8148699771361647,
      "mean_token_accuracy_utility": 0.8070580691856346,
      "step": 2350
    },
    {
      "epoch": 0.8712541620421753,
      "grad_norm": 2.0625,
      "learning_rate": 9.935165148446658e-07,
      "loss": 1.3343,
      "mean_token_accuracy": 0.7901211028193563,
      "mean_token_accuracy_utility": 0.8028768894121772,
      "step": 2355
    },
    {
      "epoch": 0.8731039585645579,
      "grad_norm": 2.015625,
      "learning_rate": 9.656376520834365e-07,
      "loss": 1.3835,
      "mean_token_accuracy": 0.8022637890357414,
      "mean_token_accuracy_utility": 0.7864816552373237,
      "step": 2360
    },
    {
      "epoch": 0.8749537550869404,
      "grad_norm": 2.15625,
      "learning_rate": 9.381356741098702e-07,
      "loss": 1.2603,
      "mean_token_accuracy": 0.8140391061412172,
      "mean_token_accuracy_utility": 0.8049883801344269,
      "step": 2365
    },
    {
      "epoch": 0.876803551609323,
      "grad_norm": 2.0625,
      "learning_rate": 9.110117282187892e-07,
      "loss": 1.2952,
      "mean_token_accuracy": 0.8207917792526354,
      "mean_token_accuracy_utility": 0.7909131887208832,
      "step": 2370
    },
    {
      "epoch": 0.8786533481317055,
      "grad_norm": 2.015625,
      "learning_rate": 8.842669459347186e-07,
      "loss": 1.2726,
      "mean_token_accuracy": 0.7983337852042187,
      "mean_token_accuracy_utility": 0.8118552781606153,
      "step": 2375
    },
    {
      "epoch": 0.8805031446540881,
      "grad_norm": 2.078125,
      "learning_rate": 8.579024429646932e-07,
      "loss": 1.3697,
      "mean_token_accuracy": 0.7946229322228235,
      "mean_token_accuracy_utility": 0.7903302916210349,
      "step": 2380
    },
    {
      "epoch": 0.8823529411764706,
      "grad_norm": 1.984375,
      "learning_rate": 8.319193191517016e-07,
      "loss": 1.2937,
      "mean_token_accuracy": 0.8054194577517924,
      "mean_token_accuracy_utility": 0.8033666611375772,
      "step": 2385
    },
    {
      "epoch": 0.8842027376988532,
      "grad_norm": 2.078125,
      "learning_rate": 8.063186584288108e-07,
      "loss": 1.3904,
      "mean_token_accuracy": 0.805003163192651,
      "mean_token_accuracy_utility": 0.782178453603551,
      "step": 2390
    },
    {
      "epoch": 0.8860525342212356,
      "grad_norm": 1.9375,
      "learning_rate": 7.81101528773951e-07,
      "loss": 1.3538,
      "mean_token_accuracy": 0.7944743635407095,
      "mean_token_accuracy_utility": 0.7968451836241457,
      "step": 2395
    },
    {
      "epoch": 0.8879023307436182,
      "grad_norm": 2.109375,
      "learning_rate": 7.562689821653568e-07,
      "loss": 1.3158,
      "mean_token_accuracy": 0.8091690753542258,
      "mean_token_accuracy_utility": 0.7943476478815767,
      "step": 2400
    },
    {
      "epoch": 0.8879023307436182,
      "eval_mean_token_accuracy": 0.7928386088237782,
      "eval_reasoning_loss": 0.6475443243980408,
      "eval_reasoning_runtime": 90.6838,
      "eval_reasoning_samples_per_second": 1.423,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 2400
    },
    {
      "epoch": 0.8879023307436182,
      "eval_mean_token_accuracy": 0.7404651506476744,
      "eval_utility_loss": 0.9601895213127136,
      "eval_utility_runtime": 34.1338,
      "eval_utility_samples_per_second": 1.436,
      "eval_utility_steps_per_second": 0.381,
      "step": 2400
    },
    {
      "epoch": 0.8897521272660007,
      "grad_norm": 1.96875,
      "learning_rate": 7.318220545376842e-07,
      "loss": 1.3329,
      "mean_token_accuracy": 0.8074815971816038,
      "mean_token_accuracy_utility": 0.7941731888871604,
      "step": 2405
    },
    {
      "epoch": 0.8916019237883833,
      "grad_norm": 2.1875,
      "learning_rate": 7.077617657387936e-07,
      "loss": 1.3027,
      "mean_token_accuracy": 0.8020902510979642,
      "mean_token_accuracy_utility": 0.804947720745195,
      "step": 2410
    },
    {
      "epoch": 0.8934517203107658,
      "grad_norm": 2.265625,
      "learning_rate": 6.840891194872112e-07,
      "loss": 1.4256,
      "mean_token_accuracy": 0.7963904804142474,
      "mean_token_accuracy_utility": 0.7830443016960035,
      "step": 2415
    },
    {
      "epoch": 0.8953015168331484,
      "grad_norm": 2.203125,
      "learning_rate": 6.608051033302476e-07,
      "loss": 1.3236,
      "mean_token_accuracy": 0.7979192797878141,
      "mean_token_accuracy_utility": 0.7996772891981083,
      "step": 2420
    },
    {
      "epoch": 0.8971513133555309,
      "grad_norm": 2.1875,
      "learning_rate": 6.379106886028086e-07,
      "loss": 1.2877,
      "mean_token_accuracy": 0.8144404345303226,
      "mean_token_accuracy_utility": 0.7983129061935019,
      "step": 2425
    },
    {
      "epoch": 0.8990011098779135,
      "grad_norm": 2.1875,
      "learning_rate": 6.154068303868666e-07,
      "loss": 1.3228,
      "mean_token_accuracy": 0.8041690907429858,
      "mean_token_accuracy_utility": 0.7962275692699524,
      "step": 2430
    },
    {
      "epoch": 0.900850906400296,
      "grad_norm": 2.125,
      "learning_rate": 5.932944674716279e-07,
      "loss": 1.3271,
      "mean_token_accuracy": 0.8034212666861897,
      "mean_token_accuracy_utility": 0.7935484120103417,
      "step": 2435
    },
    {
      "epoch": 0.9027007029226785,
      "grad_norm": 2.09375,
      "learning_rate": 5.715745223143566e-07,
      "loss": 1.3949,
      "mean_token_accuracy": 0.7998082044000647,
      "mean_token_accuracy_utility": 0.7827027222696041,
      "step": 2440
    },
    {
      "epoch": 0.904550499445061,
      "grad_norm": 1.9375,
      "learning_rate": 5.502479010019046e-07,
      "loss": 1.2358,
      "mean_token_accuracy": 0.8120394482826283,
      "mean_token_accuracy_utility": 0.8052053945871629,
      "step": 2445
    },
    {
      "epoch": 0.9064002959674435,
      "grad_norm": 2.203125,
      "learning_rate": 5.293154932129019e-07,
      "loss": 1.4335,
      "mean_token_accuracy": 0.8005157481625839,
      "mean_token_accuracy_utility": 0.7795003430646112,
      "step": 2450
    },
    {
      "epoch": 0.9082500924898261,
      "grad_norm": 2.171875,
      "learning_rate": 5.087781721806539e-07,
      "loss": 1.3197,
      "mean_token_accuracy": 0.8132823967681272,
      "mean_token_accuracy_utility": 0.7910058051921174,
      "step": 2455
    },
    {
      "epoch": 0.9100998890122086,
      "grad_norm": 2.125,
      "learning_rate": 4.886367946567005e-07,
      "loss": 1.3784,
      "mean_token_accuracy": 0.7972726133746608,
      "mean_token_accuracy_utility": 0.7921476638367259,
      "step": 2460
    },
    {
      "epoch": 0.9119496855345912,
      "grad_norm": 2.109375,
      "learning_rate": 4.6889220087508514e-07,
      "loss": 1.3195,
      "mean_token_accuracy": 0.8121462835679445,
      "mean_token_accuracy_utility": 0.7940813836969067,
      "step": 2465
    },
    {
      "epoch": 0.9137994820569737,
      "grad_norm": 1.9296875,
      "learning_rate": 4.4954521451729626e-07,
      "loss": 1.2443,
      "mean_token_accuracy": 0.8274707774845401,
      "mean_token_accuracy_utility": 0.7982814392244546,
      "step": 2470
    },
    {
      "epoch": 0.9156492785793563,
      "grad_norm": 2.046875,
      "learning_rate": 4.305966426779118e-07,
      "loss": 1.2853,
      "mean_token_accuracy": 0.8013762204682727,
      "mean_token_accuracy_utility": 0.8117358727788435,
      "step": 2475
    },
    {
      "epoch": 0.9174990751017388,
      "grad_norm": 2.015625,
      "learning_rate": 4.120472758309213e-07,
      "loss": 1.3471,
      "mean_token_accuracy": 0.8164077018711019,
      "mean_token_accuracy_utility": 0.792575651630505,
      "step": 2480
    },
    {
      "epoch": 0.9193488716241214,
      "grad_norm": 2.171875,
      "learning_rate": 3.9389788779675806e-07,
      "loss": 1.3982,
      "mean_token_accuracy": 0.793197670462478,
      "mean_token_accuracy_utility": 0.7877046887906711,
      "step": 2485
    },
    {
      "epoch": 0.9211986681465039,
      "grad_norm": 2.109375,
      "learning_rate": 3.761492357100172e-07,
      "loss": 1.3779,
      "mean_token_accuracy": 0.7983691498016581,
      "mean_token_accuracy_utility": 0.7893925110020866,
      "step": 2490
    },
    {
      "epoch": 0.9230484646688865,
      "grad_norm": 2.015625,
      "learning_rate": 3.588020599878639e-07,
      "loss": 1.4103,
      "mean_token_accuracy": 0.8007592577794143,
      "mean_token_accuracy_utility": 0.7829125873179767,
      "step": 2495
    },
    {
      "epoch": 0.9248982611912689,
      "grad_norm": 2.21875,
      "learning_rate": 3.41857084299152e-07,
      "loss": 1.2876,
      "mean_token_accuracy": 0.8101339857151121,
      "mean_token_accuracy_utility": 0.8052246822561159,
      "step": 2500
    },
    {
      "epoch": 0.9248982611912689,
      "eval_mean_token_accuracy": 0.792688428863009,
      "eval_reasoning_loss": 0.6475629210472107,
      "eval_reasoning_runtime": 90.6407,
      "eval_reasoning_samples_per_second": 1.423,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 2500
    },
    {
      "epoch": 0.9248982611912689,
      "eval_mean_token_accuracy": 0.7404477043549316,
      "eval_utility_loss": 0.9602048397064209,
      "eval_utility_runtime": 34.1215,
      "eval_utility_samples_per_second": 1.436,
      "eval_utility_steps_per_second": 0.381,
      "step": 2500
    },
    {
      "epoch": 0.9267480577136515,
      "grad_norm": 2.0625,
      "learning_rate": 3.2531501553422884e-07,
      "loss": 1.3432,
      "mean_token_accuracy": 0.8112249583707861,
      "mean_token_accuracy_utility": 0.7839806408730025,
      "step": 2505
    },
    {
      "epoch": 0.928597854236034,
      "grad_norm": 2.203125,
      "learning_rate": 3.0917654377545567e-07,
      "loss": 1.3657,
      "mean_token_accuracy": 0.808913193451388,
      "mean_token_accuracy_utility": 0.7837544133032475,
      "step": 2510
    },
    {
      "epoch": 0.9304476507584166,
      "grad_norm": 2.078125,
      "learning_rate": 2.9344234226840964e-07,
      "loss": 1.3752,
      "mean_token_accuracy": 0.8005793037003992,
      "mean_token_accuracy_utility": 0.7896063801178025,
      "step": 2515
    },
    {
      "epoch": 0.9322974472807991,
      "grad_norm": 2.1875,
      "learning_rate": 2.781130673938037e-07,
      "loss": 1.3858,
      "mean_token_accuracy": 0.8031196358270938,
      "mean_token_accuracy_utility": 0.7829944003912842,
      "step": 2520
    },
    {
      "epoch": 0.9341472438031817,
      "grad_norm": 2.1875,
      "learning_rate": 2.6318935864010133e-07,
      "loss": 1.3281,
      "mean_token_accuracy": 0.8004446199178867,
      "mean_token_accuracy_utility": 0.7976781624158031,
      "step": 2525
    },
    {
      "epoch": 0.9359970403255642,
      "grad_norm": 2.0625,
      "learning_rate": 2.4867183857684385e-07,
      "loss": 1.2599,
      "mean_token_accuracy": 0.8107408664825971,
      "mean_token_accuracy_utility": 0.8041227308802208,
      "step": 2530
    },
    {
      "epoch": 0.9378468368479467,
      "grad_norm": 2.265625,
      "learning_rate": 2.3456111282867178e-07,
      "loss": 1.377,
      "mean_token_accuracy": 0.8054094923623157,
      "mean_token_accuracy_utility": 0.7832387518073368,
      "step": 2535
    },
    {
      "epoch": 0.9396966333703293,
      "grad_norm": 2.15625,
      "learning_rate": 2.2085777005006692e-07,
      "loss": 1.335,
      "mean_token_accuracy": 0.8060093327629587,
      "mean_token_accuracy_utility": 0.7961333668019623,
      "step": 2540
    },
    {
      "epoch": 0.9415464298927118,
      "grad_norm": 2.0625,
      "learning_rate": 2.0756238190078991e-07,
      "loss": 1.2947,
      "mean_token_accuracy": 0.8079537393292574,
      "mean_token_accuracy_utility": 0.8011977077943351,
      "step": 2545
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 2.1875,
      "learning_rate": 1.946755030220393e-07,
      "loss": 1.3954,
      "mean_token_accuracy": 0.7899217276621129,
      "mean_token_accuracy_utility": 0.7926439798761742,
      "step": 2550
    },
    {
      "epoch": 0.9452460229374768,
      "grad_norm": 2.125,
      "learning_rate": 1.8219767101330442e-07,
      "loss": 1.3537,
      "mean_token_accuracy": 0.8070224589873398,
      "mean_token_accuracy_utility": 0.787691196043897,
      "step": 2555
    },
    {
      "epoch": 0.9470958194598594,
      "grad_norm": 2.171875,
      "learning_rate": 1.7012940640995012e-07,
      "loss": 1.316,
      "mean_token_accuracy": 0.803463582327861,
      "mean_token_accuracy_utility": 0.7983702518389384,
      "step": 2560
    },
    {
      "epoch": 0.9489456159822419,
      "grad_norm": 2.109375,
      "learning_rate": 1.5847121266148847e-07,
      "loss": 1.3086,
      "mean_token_accuracy": 0.8006108332175084,
      "mean_token_accuracy_utility": 0.8015104170605636,
      "step": 2565
    },
    {
      "epoch": 0.9507954125046245,
      "grad_norm": 2.125,
      "learning_rate": 1.472235761105878e-07,
      "loss": 1.334,
      "mean_token_accuracy": 0.8195085582923701,
      "mean_token_accuracy_utility": 0.7819722856770938,
      "step": 2570
    },
    {
      "epoch": 0.952645209027007,
      "grad_norm": 2.1875,
      "learning_rate": 1.3638696597277678e-07,
      "loss": 1.3637,
      "mean_token_accuracy": 0.7994650657889049,
      "mean_token_accuracy_utility": 0.7914023438118674,
      "step": 2575
    },
    {
      "epoch": 0.9544950055493896,
      "grad_norm": 2.203125,
      "learning_rate": 1.2596183431687336e-07,
      "loss": 1.3839,
      "mean_token_accuracy": 0.8044488397411811,
      "mean_token_accuracy_utility": 0.7854891970216745,
      "step": 2580
    },
    {
      "epoch": 0.9563448020717721,
      "grad_norm": 2.078125,
      "learning_rate": 1.159486160461265e-07,
      "loss": 1.3068,
      "mean_token_accuracy": 0.8008146551705233,
      "mean_token_accuracy_utility": 0.8040750246006809,
      "step": 2585
    },
    {
      "epoch": 0.9581945985941547,
      "grad_norm": 2.0625,
      "learning_rate": 1.0634772888007073e-07,
      "loss": 1.3164,
      "mean_token_accuracy": 0.8073699696853301,
      "mean_token_accuracy_utility": 0.7964189916514004,
      "step": 2590
    },
    {
      "epoch": 0.9600443951165372,
      "grad_norm": 1.9609375,
      "learning_rate": 9.715957333710447e-08,
      "loss": 1.2556,
      "mean_token_accuracy": 0.8100692682897138,
      "mean_token_accuracy_utility": 0.8044697556725794,
      "step": 2595
    },
    {
      "epoch": 0.9618941916389198,
      "grad_norm": 2.15625,
      "learning_rate": 8.838453271777347e-08,
      "loss": 1.3415,
      "mean_token_accuracy": 0.8126434404907373,
      "mean_token_accuracy_utility": 0.7844376744098309,
      "step": 2600
    },
    {
      "epoch": 0.9618941916389198,
      "eval_mean_token_accuracy": 0.7926735738912333,
      "eval_reasoning_loss": 0.647609293460846,
      "eval_reasoning_runtime": 90.775,
      "eval_reasoning_samples_per_second": 1.421,
      "eval_reasoning_steps_per_second": 0.364,
      "step": 2600
    },
    {
      "epoch": 0.9618941916389198,
      "eval_mean_token_accuracy": 0.7405340908997903,
      "eval_utility_loss": 0.9601435661315918,
      "eval_utility_runtime": 34.1536,
      "eval_utility_samples_per_second": 1.435,
      "eval_utility_steps_per_second": 0.381,
      "step": 2600
    },
    {
      "epoch": 0.9637439881613022,
      "grad_norm": 2.09375,
      "learning_rate": 8.002297308879359e-08,
      "loss": 1.3414,
      "mean_token_accuracy": 0.7982381292364685,
      "mean_token_accuracy_utility": 0.7969557520431243,
      "step": 2605
    },
    {
      "epoch": 0.9655937846836848,
      "grad_norm": 2.015625,
      "learning_rate": 7.207524326776628e-08,
      "loss": 1.3294,
      "mean_token_accuracy": 0.8091265791897619,
      "mean_token_accuracy_utility": 0.7917696554481748,
      "step": 2610
    },
    {
      "epoch": 0.9674435812060673,
      "grad_norm": 2.15625,
      "learning_rate": 6.454167480863694e-08,
      "loss": 1.4093,
      "mean_token_accuracy": 0.7917380993879756,
      "mean_token_accuracy_utility": 0.7865796768873289,
      "step": 2615
    },
    {
      "epoch": 0.9692933777284499,
      "grad_norm": 2.09375,
      "learning_rate": 5.7422581987856e-08,
      "loss": 1.3482,
      "mean_token_accuracy": 0.8006220066139669,
      "mean_token_accuracy_utility": 0.7922571469172082,
      "step": 2620
    },
    {
      "epoch": 0.9711431742508324,
      "grad_norm": 2.09375,
      "learning_rate": 5.0718261791274924e-08,
      "loss": 1.3844,
      "mean_token_accuracy": 0.7894971043628515,
      "mean_token_accuracy_utility": 0.7946752681575147,
      "step": 2625
    },
    {
      "epoch": 0.9729929707732149,
      "grad_norm": 1.953125,
      "learning_rate": 4.4428993901753924e-08,
      "loss": 1.2519,
      "mean_token_accuracy": 0.809205545085071,
      "mean_token_accuracy_utility": 0.8108405738811172,
      "step": 2630
    },
    {
      "epoch": 0.9748427672955975,
      "grad_norm": 2.171875,
      "learning_rate": 3.8555040687493494e-08,
      "loss": 1.2873,
      "mean_token_accuracy": 0.8105078750933608,
      "mean_token_accuracy_utility": 0.8017168241276004,
      "step": 2635
    },
    {
      "epoch": 0.97669256381798,
      "grad_norm": 2.109375,
      "learning_rate": 3.309664719108874e-08,
      "loss": 1.3394,
      "mean_token_accuracy": 0.8161221120346462,
      "mean_token_accuracy_utility": 0.7858386918507687,
      "step": 2640
    },
    {
      "epoch": 0.9785423603403626,
      "grad_norm": 2.125,
      "learning_rate": 2.805404111931198e-08,
      "loss": 1.335,
      "mean_token_accuracy": 0.8043003969251844,
      "mean_token_accuracy_utility": 0.7942905181062777,
      "step": 2645
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 2.03125,
      "learning_rate": 2.3427432833607046e-08,
      "loss": 1.2863,
      "mean_token_accuracy": 0.8039570071561505,
      "mean_token_accuracy_utility": 0.8057494355729823,
      "step": 2650
    },
    {
      "epoch": 0.9822419533851277,
      "grad_norm": 2.015625,
      "learning_rate": 1.9217015341318478e-08,
      "loss": 1.3767,
      "mean_token_accuracy": 0.7908223190175888,
      "mean_token_accuracy_utility": 0.7963824621813479,
      "step": 2655
    },
    {
      "epoch": 0.9840917499075101,
      "grad_norm": 2.09375,
      "learning_rate": 1.542296428763801e-08,
      "loss": 1.3166,
      "mean_token_accuracy": 0.8116719772043324,
      "mean_token_accuracy_utility": 0.7875966952066182,
      "step": 2660
    },
    {
      "epoch": 0.9859415464298927,
      "grad_norm": 2.234375,
      "learning_rate": 1.2045437948275952e-08,
      "loss": 1.4301,
      "mean_token_accuracy": 0.7978998845166878,
      "mean_token_accuracy_utility": 0.7779689476847308,
      "step": 2665
    },
    {
      "epoch": 0.9877913429522752,
      "grad_norm": 1.9921875,
      "learning_rate": 9.084577222860935e-09,
      "loss": 1.3021,
      "mean_token_accuracy": 0.8118116148153934,
      "mean_token_accuracy_utility": 0.7954338914586361,
      "step": 2670
    },
    {
      "epoch": 0.9896411394746578,
      "grad_norm": 2.1875,
      "learning_rate": 6.540505629061278e-09,
      "loss": 1.311,
      "mean_token_accuracy": 0.8032091885789677,
      "mean_token_accuracy_utility": 0.80144399947995,
      "step": 2675
    },
    {
      "epoch": 0.9914909359970403,
      "grad_norm": 2.125,
      "learning_rate": 4.413329297430213e-09,
      "loss": 1.3235,
      "mean_token_accuracy": 0.8049698645045489,
      "mean_token_accuracy_utility": 0.7955331372224959,
      "step": 2680
    },
    {
      "epoch": 0.9933407325194229,
      "grad_norm": 2.0,
      "learning_rate": 2.7031369669816566e-09,
      "loss": 1.3907,
      "mean_token_accuracy": 0.7864592914525759,
      "mean_token_accuracy_utility": 0.7946361387352286,
      "step": 2685
    },
    {
      "epoch": 0.9951905290418054,
      "grad_norm": 2.109375,
      "learning_rate": 1.4099999814853883e-09,
      "loss": 1.3942,
      "mean_token_accuracy": 0.8007344816976947,
      "mean_token_accuracy_utility": 0.7844585765435285,
      "step": 2690
    },
    {
      "epoch": 0.997040325564188,
      "grad_norm": 2.046875,
      "learning_rate": 5.339722864927677e-10,
      "loss": 1.369,
      "mean_token_accuracy": 0.7972730100006065,
      "mean_token_accuracy_utility": 0.7903819274915397,
      "step": 2695
    },
    {
      "epoch": 0.9988901220865705,
      "grad_norm": 2.375,
      "learning_rate": 7.509042708631065e-11,
      "loss": 1.4021,
      "mean_token_accuracy": 0.8007453038644611,
      "mean_token_accuracy_utility": 0.7837229198694624,
      "step": 2700
    },
    {
      "epoch": 0.9988901220865705,
      "eval_mean_token_accuracy": 0.7928682081206948,
      "eval_reasoning_loss": 0.6476022005081177,
      "eval_reasoning_runtime": 90.7955,
      "eval_reasoning_samples_per_second": 1.421,
      "eval_reasoning_steps_per_second": 0.363,
      "step": 2700
    },
    {
      "epoch": 0.9988901220865705,
      "eval_mean_token_accuracy": 0.7404319290195404,
      "eval_utility_loss": 0.9602180123329163,
      "eval_utility_runtime": 34.1677,
      "eval_utility_samples_per_second": 1.434,
      "eval_utility_steps_per_second": 0.38,
      "step": 2700
    },
    {
      "epoch": 1.0,
      "mean_token_accuracy": 0.7995414138028116,
      "mean_token_accuracy_utility": 0.8161031719952763,
      "step": 2703,
      "total_flos": 0.0,
      "train_loss": 1.4448112659087413,
      "train_runtime": 91860.5158,
      "train_samples_per_second": 0.235,
      "train_steps_per_second": 0.029
    }
  ],
  "logging_steps": 5,
  "max_steps": 2703,
  "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
}