{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9979530403371464,
  "eval_steps": 100,
  "global_step": 3114,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00481637567730283,
      "grad_norm": 15.24969482421875,
      "learning_rate": 5.60486511952635e-06,
      "loss": 1.7081,
      "mean_token_accuracy": 0.6025852240622044,
      "step": 5
    },
    {
      "epoch": 0.00963275135460566,
      "grad_norm": 12.314075469970703,
      "learning_rate": 8.018749137669579e-06,
      "loss": 1.7998,
      "mean_token_accuracy": 0.5892860271036625,
      "step": 10
    },
    {
      "epoch": 0.014449127031908489,
      "grad_norm": 3.993947982788086,
      "learning_rate": 9.430780769373472e-06,
      "loss": 1.5099,
      "mean_token_accuracy": 0.6336932629346848,
      "step": 15
    },
    {
      "epoch": 0.01926550270921132,
      "grad_norm": 1.7531483173370361,
      "learning_rate": 1.0432633155812801e-05,
      "loss": 1.4455,
      "mean_token_accuracy": 0.6432643696665764,
      "step": 20
    },
    {
      "epoch": 0.024081878386514148,
      "grad_norm": 0.5668932795524597,
      "learning_rate": 1.12097302390527e-05,
      "loss": 1.351,
      "mean_token_accuracy": 0.6605924591422081,
      "step": 25
    },
    {
      "epoch": 0.028898254063816978,
      "grad_norm": 0.4905467927455902,
      "learning_rate": 1.1844664787516697e-05,
      "loss": 1.2968,
      "mean_token_accuracy": 0.6679659530520439,
      "step": 30
    },
    {
      "epoch": 0.03371462974111981,
      "grad_norm": 0.4706348776817322,
      "learning_rate": 1.2381494299136923e-05,
      "loss": 1.2571,
      "mean_token_accuracy": 0.6737238138914108,
      "step": 35
    },
    {
      "epoch": 0.03853100541842264,
      "grad_norm": 0.43812254071235657,
      "learning_rate": 1.2846517173956028e-05,
      "loss": 1.186,
      "mean_token_accuracy": 0.6864958420395851,
      "step": 40
    },
    {
      "epoch": 0.04334738109572547,
      "grad_norm": 0.47561436891555786,
      "learning_rate": 1.3256696419220589e-05,
      "loss": 1.2086,
      "mean_token_accuracy": 0.67629723995924,
      "step": 45
    },
    {
      "epoch": 0.048163756773028296,
      "grad_norm": 0.4023856222629547,
      "learning_rate": 1.3623614257195928e-05,
      "loss": 1.1375,
      "mean_token_accuracy": 0.6874296680092812,
      "step": 50
    },
    {
      "epoch": 0.052980132450331126,
      "grad_norm": 0.39125627279281616,
      "learning_rate": 1.3955531815614275e-05,
      "loss": 1.1751,
      "mean_token_accuracy": 0.6815471544861793,
      "step": 55
    },
    {
      "epoch": 0.057796508127633955,
      "grad_norm": 0.410149484872818,
      "learning_rate": 1.4258548805659922e-05,
      "loss": 1.1622,
      "mean_token_accuracy": 0.6843628376722336,
      "step": 60
    },
    {
      "epoch": 0.06261288380493679,
      "grad_norm": 0.3972622752189636,
      "learning_rate": 1.4537297415249556e-05,
      "loss": 1.1487,
      "mean_token_accuracy": 0.6880501955747604,
      "step": 65
    },
    {
      "epoch": 0.06742925948223961,
      "grad_norm": 0.39933261275291443,
      "learning_rate": 1.4795378317280148e-05,
      "loss": 1.1235,
      "mean_token_accuracy": 0.6893462091684341,
      "step": 70
    },
    {
      "epoch": 0.07224563515954245,
      "grad_norm": 0.36374545097351074,
      "learning_rate": 1.5035645888899821e-05,
      "loss": 1.1015,
      "mean_token_accuracy": 0.691944369673729,
      "step": 75
    },
    {
      "epoch": 0.07706201083684527,
      "grad_norm": 0.3634852468967438,
      "learning_rate": 1.5260401192099252e-05,
      "loss": 1.1078,
      "mean_token_accuracy": 0.6953662529587745,
      "step": 80
    },
    {
      "epoch": 0.08187838651414811,
      "grad_norm": 0.4331696927547455,
      "learning_rate": 1.547152634677485e-05,
      "loss": 1.0918,
      "mean_token_accuracy": 0.7018570601940155,
      "step": 85
    },
    {
      "epoch": 0.08669476219145093,
      "grad_norm": 0.4048088490962982,
      "learning_rate": 1.5670580437363816e-05,
      "loss": 1.0828,
      "mean_token_accuracy": 0.699838899075985,
      "step": 90
    },
    {
      "epoch": 0.09151113786875377,
      "grad_norm": 0.42811787128448486,
      "learning_rate": 1.5858869454458713e-05,
      "loss": 1.1066,
      "mean_token_accuracy": 0.6937131121754646,
      "step": 95
    },
    {
      "epoch": 0.09632751354605659,
      "grad_norm": 0.39595216512680054,
      "learning_rate": 1.6037498275339157e-05,
      "loss": 1.1185,
      "mean_token_accuracy": 0.6915625333786011,
      "step": 100
    },
    {
      "epoch": 0.09632751354605659,
      "eval_loss": 1.120937466621399,
      "eval_mean_token_accuracy": 0.6942809590926537,
      "eval_runtime": 23.7545,
      "eval_samples_per_second": 4.21,
      "eval_steps_per_second": 0.547,
      "step": 100
    },
    {
      "epoch": 0.10114388922335943,
      "grad_norm": 0.3679651618003845,
      "learning_rate": 1.620740994898404e-05,
      "loss": 1.0867,
      "mean_token_accuracy": 0.695780485868454,
      "step": 105
    },
    {
      "epoch": 0.10596026490066225,
      "grad_norm": 0.3855838179588318,
      "learning_rate": 1.6369415833757502e-05,
      "loss": 1.1044,
      "mean_token_accuracy": 0.6917166858911514,
      "step": 110
    },
    {
      "epoch": 0.11077664057796509,
      "grad_norm": 0.4066142737865448,
      "learning_rate": 1.6524219030333085e-05,
      "loss": 1.0656,
      "mean_token_accuracy": 0.7033417671918869,
      "step": 115
    },
    {
      "epoch": 0.11559301625526791,
      "grad_norm": 0.41730743646621704,
      "learning_rate": 1.6672432823803147e-05,
      "loss": 1.0478,
      "mean_token_accuracy": 0.7045514702796936,
      "step": 120
    },
    {
      "epoch": 0.12040939193257075,
      "grad_norm": 0.41672202944755554,
      "learning_rate": 1.6814595358579055e-05,
      "loss": 1.0736,
      "mean_token_accuracy": 0.701400151848793,
      "step": 125
    },
    {
      "epoch": 0.12522576760987358,
      "grad_norm": 0.3911355137825012,
      "learning_rate": 1.695118143339278e-05,
      "loss": 1.0484,
      "mean_token_accuracy": 0.70690358877182,
      "step": 130
    },
    {
      "epoch": 0.1300421432871764,
      "grad_norm": 0.4797114431858063,
      "learning_rate": 1.708261206906771e-05,
      "loss": 1.0275,
      "mean_token_accuracy": 0.7159404128789901,
      "step": 135
    },
    {
      "epoch": 0.13485851896447923,
      "grad_norm": 0.4215666353702545,
      "learning_rate": 1.7209262335423374e-05,
      "loss": 1.0461,
      "mean_token_accuracy": 0.7075429305434227,
      "step": 140
    },
    {
      "epoch": 0.13967489464178207,
      "grad_norm": 0.39611467719078064,
      "learning_rate": 1.733146780410832e-05,
      "loss": 1.0515,
      "mean_token_accuracy": 0.7017971694469451,
      "step": 145
    },
    {
      "epoch": 0.1444912703190849,
      "grad_norm": 0.40612080693244934,
      "learning_rate": 1.7449529907043045e-05,
      "loss": 1.0619,
      "mean_token_accuracy": 0.7035504311323166,
      "step": 150
    },
    {
      "epoch": 0.1493076459963877,
      "grad_norm": 0.4210834801197052,
      "learning_rate": 1.7563720415913364e-05,
      "loss": 1.0407,
      "mean_token_accuracy": 0.7101730972528457,
      "step": 155
    },
    {
      "epoch": 0.15412402167369055,
      "grad_norm": 0.39545348286628723,
      "learning_rate": 1.7674285210242478e-05,
      "loss": 1.0184,
      "mean_token_accuracy": 0.7077491611242295,
      "step": 160
    },
    {
      "epoch": 0.15894039735099338,
      "grad_norm": 0.38933709263801575,
      "learning_rate": 1.7781447465461393e-05,
      "loss": 1.0496,
      "mean_token_accuracy": 0.7077953517436981,
      "step": 165
    },
    {
      "epoch": 0.16375677302829622,
      "grad_norm": 0.4460107386112213,
      "learning_rate": 1.788541036491807e-05,
      "loss": 1.0444,
      "mean_token_accuracy": 0.7059019073843956,
      "step": 170
    },
    {
      "epoch": 0.16857314870559903,
      "grad_norm": 0.43760132789611816,
      "learning_rate": 1.7986359418663275e-05,
      "loss": 1.0397,
      "mean_token_accuracy": 0.7088375821709633,
      "step": 175
    },
    {
      "epoch": 0.17338952438290187,
      "grad_norm": 0.39786314964294434,
      "learning_rate": 1.8084464455507042e-05,
      "loss": 1.0493,
      "mean_token_accuracy": 0.7046118810772896,
      "step": 180
    },
    {
      "epoch": 0.1782059000602047,
      "grad_norm": 0.36780059337615967,
      "learning_rate": 1.817988134208051e-05,
      "loss": 1.0067,
      "mean_token_accuracy": 0.7162643566727638,
      "step": 185
    },
    {
      "epoch": 0.18302227573750754,
      "grad_norm": 0.3834674656391144,
      "learning_rate": 1.8272753472601942e-05,
      "loss": 1.0074,
      "mean_token_accuracy": 0.7104792430996895,
      "step": 190
    },
    {
      "epoch": 0.18783865141481035,
      "grad_norm": 0.3949407935142517,
      "learning_rate": 1.8363213065096677e-05,
      "loss": 1.0135,
      "mean_token_accuracy": 0.715023210644722,
      "step": 195
    },
    {
      "epoch": 0.19265502709211318,
      "grad_norm": 0.38337242603302,
      "learning_rate": 1.845138229348238e-05,
      "loss": 1.0288,
      "mean_token_accuracy": 0.7040736198425293,
      "step": 200
    },
    {
      "epoch": 0.19265502709211318,
      "eval_loss": 1.064062476158142,
      "eval_mean_token_accuracy": 0.7043925065260667,
      "eval_runtime": 23.1434,
      "eval_samples_per_second": 4.321,
      "eval_steps_per_second": 0.562,
      "step": 200
    },
    {
      "epoch": 0.19747140276941602,
      "grad_norm": 0.39040884375572205,
      "learning_rate": 1.8537374279845146e-05,
      "loss": 0.9799,
      "mean_token_accuracy": 0.7189632415771484,
      "step": 205
    },
    {
      "epoch": 0.20228777844671886,
      "grad_norm": 0.4173654019832611,
      "learning_rate": 1.862129396712727e-05,
      "loss": 1.0342,
      "mean_token_accuracy": 0.7094724953174592,
      "step": 210
    },
    {
      "epoch": 0.20710415412402167,
      "grad_norm": 0.3901558518409729,
      "learning_rate": 1.8703238889115616e-05,
      "loss": 0.9908,
      "mean_token_accuracy": 0.7175863072276115,
      "step": 215
    },
    {
      "epoch": 0.2119205298013245,
      "grad_norm": 0.4490678906440735,
      "learning_rate": 1.8783299851900728e-05,
      "loss": 1.0204,
      "mean_token_accuracy": 0.7106468200683593,
      "step": 220
    },
    {
      "epoch": 0.21673690547862734,
      "grad_norm": 0.43239694833755493,
      "learning_rate": 1.8861561538746943e-05,
      "loss": 1.0393,
      "mean_token_accuracy": 0.7069905534386635,
      "step": 225
    },
    {
      "epoch": 0.22155328115593018,
      "grad_norm": 0.4308166801929474,
      "learning_rate": 1.8938103048476308e-05,
      "loss": 1.0158,
      "mean_token_accuracy": 0.7123534351587295,
      "step": 230
    },
    {
      "epoch": 0.22636965683323299,
      "grad_norm": 0.3894931972026825,
      "learning_rate": 1.901299837594753e-05,
      "loss": 0.9873,
      "mean_token_accuracy": 0.7187181740999222,
      "step": 235
    },
    {
      "epoch": 0.23118603251053582,
      "grad_norm": 0.42207640409469604,
      "learning_rate": 1.9086316841946373e-05,
      "loss": 0.9893,
      "mean_token_accuracy": 0.7170199275016784,
      "step": 240
    },
    {
      "epoch": 0.23600240818783866,
      "grad_norm": 0.45573151111602783,
      "learning_rate": 1.9158123478747496e-05,
      "loss": 1.0173,
      "mean_token_accuracy": 0.7128939390182495,
      "step": 245
    },
    {
      "epoch": 0.2408187838651415,
      "grad_norm": 0.44542551040649414,
      "learning_rate": 1.9228479376722277e-05,
      "loss": 1.0264,
      "mean_token_accuracy": 0.7066132187843323,
      "step": 250
    },
    {
      "epoch": 0.2456351595424443,
      "grad_norm": 0.42006808519363403,
      "learning_rate": 1.9297441996621962e-05,
      "loss": 1.0169,
      "mean_token_accuracy": 0.7104522138834,
      "step": 255
    },
    {
      "epoch": 0.25045153521974717,
      "grad_norm": 0.45401668548583984,
      "learning_rate": 1.9365065451536008e-05,
      "loss": 0.9878,
      "mean_token_accuracy": 0.7117918729782104,
      "step": 260
    },
    {
      "epoch": 0.25526791089705,
      "grad_norm": 0.40076637268066406,
      "learning_rate": 1.9431400761992135e-05,
      "loss": 0.9812,
      "mean_token_accuracy": 0.7238912284374237,
      "step": 265
    },
    {
      "epoch": 0.2600842865743528,
      "grad_norm": 0.39845559000968933,
      "learning_rate": 1.9496496087210937e-05,
      "loss": 0.9924,
      "mean_token_accuracy": 0.7154505327343941,
      "step": 270
    },
    {
      "epoch": 0.26490066225165565,
      "grad_norm": 0.38575807213783264,
      "learning_rate": 1.9560396935140626e-05,
      "loss": 0.9987,
      "mean_token_accuracy": 0.7184735506772995,
      "step": 275
    },
    {
      "epoch": 0.26971703792895846,
      "grad_norm": 0.37453049421310425,
      "learning_rate": 1.9623146353566597e-05,
      "loss": 1.0008,
      "mean_token_accuracy": 0.7148436263203621,
      "step": 280
    },
    {
      "epoch": 0.27453341360626127,
      "grad_norm": 0.379604309797287,
      "learning_rate": 1.9684785104305833e-05,
      "loss": 1.0257,
      "mean_token_accuracy": 0.707456587255001,
      "step": 285
    },
    {
      "epoch": 0.27934978928356413,
      "grad_norm": 0.41674622893333435,
      "learning_rate": 1.9745351822251547e-05,
      "loss": 1.0104,
      "mean_token_accuracy": 0.7121455416083335,
      "step": 290
    },
    {
      "epoch": 0.28416616496086694,
      "grad_norm": 0.36543646454811096,
      "learning_rate": 1.9804883160822233e-05,
      "loss": 1.0018,
      "mean_token_accuracy": 0.7111613750457764,
      "step": 295
    },
    {
      "epoch": 0.2889825406381698,
      "grad_norm": 0.3864189684391022,
      "learning_rate": 1.986341392518627e-05,
      "loss": 0.9896,
      "mean_token_accuracy": 0.7227002561092377,
      "step": 300
    },
    {
      "epoch": 0.2889825406381698,
      "eval_loss": 1.0334374904632568,
      "eval_mean_token_accuracy": 0.7100509542685288,
      "eval_runtime": 23.2119,
      "eval_samples_per_second": 4.308,
      "eval_steps_per_second": 0.56,
      "step": 300
    },
    {
      "epoch": 0.2937989163154726,
      "grad_norm": 0.39042481780052185,
      "learning_rate": 1.9920977194474707e-05,
      "loss": 0.9747,
      "mean_token_accuracy": 0.7194000944495201,
      "step": 305
    },
    {
      "epoch": 0.2986152919927754,
      "grad_norm": 0.41932472586631775,
      "learning_rate": 1.997760443405659e-05,
      "loss": 1.013,
      "mean_token_accuracy": 0.7132839515805245,
      "step": 310
    },
    {
      "epoch": 0.3034316676700783,
      "grad_norm": 0.3739113509654999,
      "learning_rate": 2e-05,
      "loss": 0.998,
      "mean_token_accuracy": 0.7116592064499855,
      "step": 315
    },
    {
      "epoch": 0.3082480433473811,
      "grad_norm": 0.3870009779930115,
      "learning_rate": 2e-05,
      "loss": 1.0472,
      "mean_token_accuracy": 0.7014497891068459,
      "step": 320
    },
    {
      "epoch": 0.3130644190246839,
      "grad_norm": 0.39197883009910583,
      "learning_rate": 2e-05,
      "loss": 0.9896,
      "mean_token_accuracy": 0.7157701522111892,
      "step": 325
    },
    {
      "epoch": 0.31788079470198677,
      "grad_norm": 0.4166513681411743,
      "learning_rate": 2e-05,
      "loss": 0.9832,
      "mean_token_accuracy": 0.7184643030166626,
      "step": 330
    },
    {
      "epoch": 0.3226971703792896,
      "grad_norm": 0.42934778332710266,
      "learning_rate": 2e-05,
      "loss": 0.9779,
      "mean_token_accuracy": 0.7207571104168892,
      "step": 335
    },
    {
      "epoch": 0.32751354605659244,
      "grad_norm": 0.3707175850868225,
      "learning_rate": 2e-05,
      "loss": 1.0075,
      "mean_token_accuracy": 0.7121270015835762,
      "step": 340
    },
    {
      "epoch": 0.33232992173389525,
      "grad_norm": 0.41228577494621277,
      "learning_rate": 2e-05,
      "loss": 0.9755,
      "mean_token_accuracy": 0.7168642222881317,
      "step": 345
    },
    {
      "epoch": 0.33714629741119806,
      "grad_norm": 0.3472833037376404,
      "learning_rate": 2e-05,
      "loss": 0.9505,
      "mean_token_accuracy": 0.7238100618124008,
      "step": 350
    },
    {
      "epoch": 0.3419626730885009,
      "grad_norm": 0.3869970142841339,
      "learning_rate": 2e-05,
      "loss": 0.9567,
      "mean_token_accuracy": 0.7265875309705734,
      "step": 355
    },
    {
      "epoch": 0.34677904876580373,
      "grad_norm": 0.422489196062088,
      "learning_rate": 2e-05,
      "loss": 0.9701,
      "mean_token_accuracy": 0.7184868469834328,
      "step": 360
    },
    {
      "epoch": 0.35159542444310654,
      "grad_norm": 0.43342992663383484,
      "learning_rate": 2e-05,
      "loss": 0.9979,
      "mean_token_accuracy": 0.7126397833228111,
      "step": 365
    },
    {
      "epoch": 0.3564118001204094,
      "grad_norm": 0.3574235737323761,
      "learning_rate": 2e-05,
      "loss": 0.9506,
      "mean_token_accuracy": 0.7254954233765603,
      "step": 370
    },
    {
      "epoch": 0.3612281757977122,
      "grad_norm": 0.4248846173286438,
      "learning_rate": 2e-05,
      "loss": 0.9657,
      "mean_token_accuracy": 0.7162819489836693,
      "step": 375
    },
    {
      "epoch": 0.3660445514750151,
      "grad_norm": 0.4223816692829132,
      "learning_rate": 2e-05,
      "loss": 0.9546,
      "mean_token_accuracy": 0.7249079883098603,
      "step": 380
    },
    {
      "epoch": 0.3708609271523179,
      "grad_norm": 0.45861685276031494,
      "learning_rate": 2e-05,
      "loss": 0.9349,
      "mean_token_accuracy": 0.7322170153260231,
      "step": 385
    },
    {
      "epoch": 0.3756773028296207,
      "grad_norm": 0.39118438959121704,
      "learning_rate": 2e-05,
      "loss": 0.9674,
      "mean_token_accuracy": 0.7160528540611267,
      "step": 390
    },
    {
      "epoch": 0.38049367850692356,
      "grad_norm": 0.3915044069290161,
      "learning_rate": 2e-05,
      "loss": 0.9661,
      "mean_token_accuracy": 0.7225557699799537,
      "step": 395
    },
    {
      "epoch": 0.38531005418422637,
      "grad_norm": 0.41501498222351074,
      "learning_rate": 2e-05,
      "loss": 0.9475,
      "mean_token_accuracy": 0.7268393576145172,
      "step": 400
    },
    {
      "epoch": 0.38531005418422637,
      "eval_loss": 1.0129687786102295,
      "eval_mean_token_accuracy": 0.7142733885691717,
      "eval_runtime": 23.1853,
      "eval_samples_per_second": 4.313,
      "eval_steps_per_second": 0.561,
      "step": 400
    },
    {
      "epoch": 0.3901264298615292,
      "grad_norm": 0.48722606897354126,
      "learning_rate": 2e-05,
      "loss": 0.9985,
      "mean_token_accuracy": 0.712544234097004,
      "step": 405
    },
    {
      "epoch": 0.39494280553883204,
      "grad_norm": 0.42091795802116394,
      "learning_rate": 2e-05,
      "loss": 0.9292,
      "mean_token_accuracy": 0.7308942764997483,
      "step": 410
    },
    {
      "epoch": 0.39975918121613485,
      "grad_norm": 0.4148479998111725,
      "learning_rate": 2e-05,
      "loss": 0.998,
      "mean_token_accuracy": 0.7112752854824066,
      "step": 415
    },
    {
      "epoch": 0.4045755568934377,
      "grad_norm": 0.3901733160018921,
      "learning_rate": 2e-05,
      "loss": 0.9559,
      "mean_token_accuracy": 0.7231112748384476,
      "step": 420
    },
    {
      "epoch": 0.4093919325707405,
      "grad_norm": 0.4191341996192932,
      "learning_rate": 2e-05,
      "loss": 0.9793,
      "mean_token_accuracy": 0.7161881670355796,
      "step": 425
    },
    {
      "epoch": 0.41420830824804333,
      "grad_norm": 0.3945569097995758,
      "learning_rate": 2e-05,
      "loss": 0.9623,
      "mean_token_accuracy": 0.7208192080259324,
      "step": 430
    },
    {
      "epoch": 0.4190246839253462,
      "grad_norm": 0.44316938519477844,
      "learning_rate": 2e-05,
      "loss": 0.9391,
      "mean_token_accuracy": 0.7293342381715775,
      "step": 435
    },
    {
      "epoch": 0.423841059602649,
      "grad_norm": 0.4176236391067505,
      "learning_rate": 2e-05,
      "loss": 0.9879,
      "mean_token_accuracy": 0.7138759195804596,
      "step": 440
    },
    {
      "epoch": 0.4286574352799518,
      "grad_norm": 0.38419532775878906,
      "learning_rate": 2e-05,
      "loss": 0.9332,
      "mean_token_accuracy": 0.7278907805681228,
      "step": 445
    },
    {
      "epoch": 0.4334738109572547,
      "grad_norm": 0.3864535093307495,
      "learning_rate": 2e-05,
      "loss": 0.9838,
      "mean_token_accuracy": 0.7188102528452873,
      "step": 450
    },
    {
      "epoch": 0.4382901866345575,
      "grad_norm": 0.40301385521888733,
      "learning_rate": 2e-05,
      "loss": 0.9727,
      "mean_token_accuracy": 0.7179606407880783,
      "step": 455
    },
    {
      "epoch": 0.44310656231186035,
      "grad_norm": 0.41534721851348877,
      "learning_rate": 2e-05,
      "loss": 0.9444,
      "mean_token_accuracy": 0.7267005681991577,
      "step": 460
    },
    {
      "epoch": 0.44792293798916316,
      "grad_norm": 0.3811696767807007,
      "learning_rate": 2e-05,
      "loss": 0.931,
      "mean_token_accuracy": 0.7301761016249657,
      "step": 465
    },
    {
      "epoch": 0.45273931366646597,
      "grad_norm": 0.38206103444099426,
      "learning_rate": 2e-05,
      "loss": 0.9205,
      "mean_token_accuracy": 0.7324171543121338,
      "step": 470
    },
    {
      "epoch": 0.45755568934376883,
      "grad_norm": 0.36854949593544006,
      "learning_rate": 2e-05,
      "loss": 0.933,
      "mean_token_accuracy": 0.7262160122394562,
      "step": 475
    },
    {
      "epoch": 0.46237206502107164,
      "grad_norm": 0.36893048882484436,
      "learning_rate": 2e-05,
      "loss": 0.9566,
      "mean_token_accuracy": 0.7235715687274933,
      "step": 480
    },
    {
      "epoch": 0.46718844069837445,
      "grad_norm": 0.3726271092891693,
      "learning_rate": 2e-05,
      "loss": 0.9731,
      "mean_token_accuracy": 0.7182789981365204,
      "step": 485
    },
    {
      "epoch": 0.4720048163756773,
      "grad_norm": 0.36003515124320984,
      "learning_rate": 2e-05,
      "loss": 0.9331,
      "mean_token_accuracy": 0.7282133370637893,
      "step": 490
    },
    {
      "epoch": 0.4768211920529801,
      "grad_norm": 0.3744361698627472,
      "learning_rate": 2e-05,
      "loss": 0.943,
      "mean_token_accuracy": 0.727964948117733,
      "step": 495
    },
    {
      "epoch": 0.481637567730283,
      "grad_norm": 0.40966275334358215,
      "learning_rate": 2e-05,
      "loss": 0.9205,
      "mean_token_accuracy": 0.7308084771037102,
      "step": 500
    },
    {
      "epoch": 0.481637567730283,
      "eval_loss": 0.9973437786102295,
      "eval_mean_token_accuracy": 0.7170403462189895,
      "eval_runtime": 23.158,
      "eval_samples_per_second": 4.318,
      "eval_steps_per_second": 0.561,
      "step": 500
    },
    {
      "epoch": 0.4864539434075858,
      "grad_norm": 0.38875406980514526,
      "learning_rate": 2e-05,
      "loss": 0.9812,
      "mean_token_accuracy": 0.7161158278584481,
      "step": 505
    },
    {
      "epoch": 0.4912703190848886,
      "grad_norm": 0.39988937973976135,
      "learning_rate": 2e-05,
      "loss": 0.9406,
      "mean_token_accuracy": 0.7260874435305595,
      "step": 510
    },
    {
      "epoch": 0.49608669476219147,
      "grad_norm": 0.40625178813934326,
      "learning_rate": 2e-05,
      "loss": 0.9645,
      "mean_token_accuracy": 0.7189718931913376,
      "step": 515
    },
    {
      "epoch": 0.5009030704394943,
      "grad_norm": 0.4269265830516815,
      "learning_rate": 2e-05,
      "loss": 0.9302,
      "mean_token_accuracy": 0.731168407201767,
      "step": 520
    },
    {
      "epoch": 0.5057194461167971,
      "grad_norm": 0.39293113350868225,
      "learning_rate": 2e-05,
      "loss": 0.9231,
      "mean_token_accuracy": 0.735957583785057,
      "step": 525
    },
    {
      "epoch": 0.5105358217941,
      "grad_norm": 0.4136344790458679,
      "learning_rate": 2e-05,
      "loss": 0.9396,
      "mean_token_accuracy": 0.7235330045223236,
      "step": 530
    },
    {
      "epoch": 0.5153521974714028,
      "grad_norm": 0.3826082944869995,
      "learning_rate": 2e-05,
      "loss": 0.9618,
      "mean_token_accuracy": 0.7200583890080452,
      "step": 535
    },
    {
      "epoch": 0.5201685731487056,
      "grad_norm": 0.40097910165786743,
      "learning_rate": 2e-05,
      "loss": 0.9442,
      "mean_token_accuracy": 0.72747682929039,
      "step": 540
    },
    {
      "epoch": 0.5249849488260084,
      "grad_norm": 0.40495404601097107,
      "learning_rate": 2e-05,
      "loss": 0.9654,
      "mean_token_accuracy": 0.7136755734682083,
      "step": 545
    },
    {
      "epoch": 0.5298013245033113,
      "grad_norm": 0.4751788079738617,
      "learning_rate": 2e-05,
      "loss": 0.9431,
      "mean_token_accuracy": 0.7281810432672501,
      "step": 550
    },
    {
      "epoch": 0.534617700180614,
      "grad_norm": 0.3656327724456787,
      "learning_rate": 2e-05,
      "loss": 0.9285,
      "mean_token_accuracy": 0.7332020610570907,
      "step": 555
    },
    {
      "epoch": 0.5394340758579169,
      "grad_norm": 0.3560369312763214,
      "learning_rate": 2e-05,
      "loss": 0.9382,
      "mean_token_accuracy": 0.7254460051655769,
      "step": 560
    },
    {
      "epoch": 0.5442504515352198,
      "grad_norm": 0.38857781887054443,
      "learning_rate": 2e-05,
      "loss": 0.9634,
      "mean_token_accuracy": 0.7230386212468147,
      "step": 565
    },
    {
      "epoch": 0.5490668272125225,
      "grad_norm": 0.3445562422275543,
      "learning_rate": 2e-05,
      "loss": 0.932,
      "mean_token_accuracy": 0.7287012577056885,
      "step": 570
    },
    {
      "epoch": 0.5538832028898254,
      "grad_norm": 0.4291544258594513,
      "learning_rate": 2e-05,
      "loss": 0.9771,
      "mean_token_accuracy": 0.7165321439504624,
      "step": 575
    },
    {
      "epoch": 0.5586995785671283,
      "grad_norm": 0.4487311542034149,
      "learning_rate": 2e-05,
      "loss": 0.9168,
      "mean_token_accuracy": 0.7317699804902077,
      "step": 580
    },
    {
      "epoch": 0.563515954244431,
      "grad_norm": 0.40127262473106384,
      "learning_rate": 2e-05,
      "loss": 0.9564,
      "mean_token_accuracy": 0.7205055937170982,
      "step": 585
    },
    {
      "epoch": 0.5683323299217339,
      "grad_norm": 0.3540896475315094,
      "learning_rate": 2e-05,
      "loss": 0.9416,
      "mean_token_accuracy": 0.7258080333471298,
      "step": 590
    },
    {
      "epoch": 0.5731487055990367,
      "grad_norm": 0.36789658665657043,
      "learning_rate": 2e-05,
      "loss": 0.9602,
      "mean_token_accuracy": 0.7191953271627426,
      "step": 595
    },
    {
      "epoch": 0.5779650812763396,
      "grad_norm": 0.42882364988327026,
      "learning_rate": 2e-05,
      "loss": 0.9277,
      "mean_token_accuracy": 0.7274743899703026,
      "step": 600
    },
    {
      "epoch": 0.5779650812763396,
      "eval_loss": 0.9842187762260437,
      "eval_mean_token_accuracy": 0.7194240184930655,
      "eval_runtime": 23.1812,
      "eval_samples_per_second": 4.314,
      "eval_steps_per_second": 0.561,
      "step": 600
    },
    {
      "epoch": 0.5827814569536424,
      "grad_norm": 0.3456374406814575,
      "learning_rate": 2e-05,
      "loss": 0.9583,
      "mean_token_accuracy": 0.7262210890650749,
      "step": 605
    },
    {
      "epoch": 0.5875978326309452,
      "grad_norm": 0.39239296317100525,
      "learning_rate": 2e-05,
      "loss": 0.9355,
      "mean_token_accuracy": 0.7298878788948059,
      "step": 610
    },
    {
      "epoch": 0.5924142083082481,
      "grad_norm": 0.3922767639160156,
      "learning_rate": 2e-05,
      "loss": 0.9119,
      "mean_token_accuracy": 0.731063374876976,
      "step": 615
    },
    {
      "epoch": 0.5972305839855508,
      "grad_norm": 0.4066627025604248,
      "learning_rate": 2e-05,
      "loss": 0.9721,
      "mean_token_accuracy": 0.7155601590871811,
      "step": 620
    },
    {
      "epoch": 0.6020469596628537,
      "grad_norm": 0.3650158643722534,
      "learning_rate": 2e-05,
      "loss": 0.9183,
      "mean_token_accuracy": 0.7321714773774147,
      "step": 625
    },
    {
      "epoch": 0.6068633353401566,
      "grad_norm": 0.3967447876930237,
      "learning_rate": 2e-05,
      "loss": 0.9125,
      "mean_token_accuracy": 0.7360529005527496,
      "step": 630
    },
    {
      "epoch": 0.6116797110174593,
      "grad_norm": 0.4131582975387573,
      "learning_rate": 2e-05,
      "loss": 0.9537,
      "mean_token_accuracy": 0.7238716900348663,
      "step": 635
    },
    {
      "epoch": 0.6164960866947622,
      "grad_norm": 0.3441373109817505,
      "learning_rate": 2e-05,
      "loss": 0.942,
      "mean_token_accuracy": 0.7241286799311638,
      "step": 640
    },
    {
      "epoch": 0.621312462372065,
      "grad_norm": 0.4480727016925812,
      "learning_rate": 2e-05,
      "loss": 0.9481,
      "mean_token_accuracy": 0.7221115916967392,
      "step": 645
    },
    {
      "epoch": 0.6261288380493678,
      "grad_norm": 0.3634999394416809,
      "learning_rate": 2e-05,
      "loss": 0.9463,
      "mean_token_accuracy": 0.7219712048768997,
      "step": 650
    },
    {
      "epoch": 0.6309452137266707,
      "grad_norm": 0.3939569890499115,
      "learning_rate": 2e-05,
      "loss": 0.9309,
      "mean_token_accuracy": 0.7307459726929665,
      "step": 655
    },
    {
      "epoch": 0.6357615894039735,
      "grad_norm": 0.3766266405582428,
      "learning_rate": 2e-05,
      "loss": 0.9568,
      "mean_token_accuracy": 0.7186618536710739,
      "step": 660
    },
    {
      "epoch": 0.6405779650812763,
      "grad_norm": 0.40739813446998596,
      "learning_rate": 2e-05,
      "loss": 0.9012,
      "mean_token_accuracy": 0.7370730206370354,
      "step": 665
    },
    {
      "epoch": 0.6453943407585792,
      "grad_norm": 0.4568629860877991,
      "learning_rate": 2e-05,
      "loss": 0.934,
      "mean_token_accuracy": 0.7295185983181,
      "step": 670
    },
    {
      "epoch": 0.650210716435882,
      "grad_norm": 0.4078063666820526,
      "learning_rate": 2e-05,
      "loss": 0.8946,
      "mean_token_accuracy": 0.7382076278328895,
      "step": 675
    },
    {
      "epoch": 0.6550270921131849,
      "grad_norm": 0.39687684178352356,
      "learning_rate": 2e-05,
      "loss": 0.9299,
      "mean_token_accuracy": 0.726612477004528,
      "step": 680
    },
    {
      "epoch": 0.6598434677904876,
      "grad_norm": 0.42360830307006836,
      "learning_rate": 2e-05,
      "loss": 0.9168,
      "mean_token_accuracy": 0.7292766153812409,
      "step": 685
    },
    {
      "epoch": 0.6646598434677905,
      "grad_norm": 0.391550213098526,
      "learning_rate": 2e-05,
      "loss": 0.9434,
      "mean_token_accuracy": 0.7261136472225189,
      "step": 690
    },
    {
      "epoch": 0.6694762191450934,
      "grad_norm": 0.380113422870636,
      "learning_rate": 2e-05,
      "loss": 0.9304,
      "mean_token_accuracy": 0.7198660045862197,
      "step": 695
    },
    {
      "epoch": 0.6742925948223961,
      "grad_norm": 0.37536484003067017,
      "learning_rate": 2e-05,
      "loss": 0.9371,
      "mean_token_accuracy": 0.7235382974147797,
      "step": 700
    },
    {
      "epoch": 0.6742925948223961,
      "eval_loss": 0.973437488079071,
      "eval_mean_token_accuracy": 0.7215940081156217,
      "eval_runtime": 23.1338,
      "eval_samples_per_second": 4.323,
      "eval_steps_per_second": 0.562,
      "step": 700
    },
    {
      "epoch": 0.679108970499699,
      "grad_norm": 0.39021074771881104,
      "learning_rate": 2e-05,
      "loss": 0.9437,
      "mean_token_accuracy": 0.7243467301130295,
      "step": 705
    },
    {
      "epoch": 0.6839253461770018,
      "grad_norm": 0.391563355922699,
      "learning_rate": 2e-05,
      "loss": 0.9232,
      "mean_token_accuracy": 0.7265035390853882,
      "step": 710
    },
    {
      "epoch": 0.6887417218543046,
      "grad_norm": 0.3613850772380829,
      "learning_rate": 2e-05,
      "loss": 0.9233,
      "mean_token_accuracy": 0.7278270050883293,
      "step": 715
    },
    {
      "epoch": 0.6935580975316075,
      "grad_norm": 0.4116956293582916,
      "learning_rate": 2e-05,
      "loss": 0.9495,
      "mean_token_accuracy": 0.7207832634449005,
      "step": 720
    },
    {
      "epoch": 0.6983744732089103,
      "grad_norm": 0.3616946339607239,
      "learning_rate": 2e-05,
      "loss": 0.9772,
      "mean_token_accuracy": 0.7127095699310303,
      "step": 725
    },
    {
      "epoch": 0.7031908488862131,
      "grad_norm": 0.40902280807495117,
      "learning_rate": 2e-05,
      "loss": 0.9146,
      "mean_token_accuracy": 0.7307567968964577,
      "step": 730
    },
    {
      "epoch": 0.708007224563516,
      "grad_norm": 0.3533954918384552,
      "learning_rate": 2e-05,
      "loss": 0.9187,
      "mean_token_accuracy": 0.7295649662613869,
      "step": 735
    },
    {
      "epoch": 0.7128236002408188,
      "grad_norm": 0.3901711702346802,
      "learning_rate": 2e-05,
      "loss": 0.9299,
      "mean_token_accuracy": 0.7287529841065407,
      "step": 740
    },
    {
      "epoch": 0.7176399759181216,
      "grad_norm": 0.45047813653945923,
      "learning_rate": 2e-05,
      "loss": 0.9104,
      "mean_token_accuracy": 0.7317241802811623,
      "step": 745
    },
    {
      "epoch": 0.7224563515954244,
      "grad_norm": 0.38531169295310974,
      "learning_rate": 2e-05,
      "loss": 0.9146,
      "mean_token_accuracy": 0.728726564347744,
      "step": 750
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.437835693359375,
      "learning_rate": 2e-05,
      "loss": 0.9102,
      "mean_token_accuracy": 0.7315579026937484,
      "step": 755
    },
    {
      "epoch": 0.7320891029500302,
      "grad_norm": 0.3801443874835968,
      "learning_rate": 2e-05,
      "loss": 0.9416,
      "mean_token_accuracy": 0.725105169415474,
      "step": 760
    },
    {
      "epoch": 0.7369054786273329,
      "grad_norm": 0.39513495564460754,
      "learning_rate": 2e-05,
      "loss": 0.8933,
      "mean_token_accuracy": 0.7324695363640785,
      "step": 765
    },
    {
      "epoch": 0.7417218543046358,
      "grad_norm": 0.4051715135574341,
      "learning_rate": 2e-05,
      "loss": 0.9193,
      "mean_token_accuracy": 0.727542620897293,
      "step": 770
    },
    {
      "epoch": 0.7465382299819386,
      "grad_norm": 0.37806564569473267,
      "learning_rate": 2e-05,
      "loss": 0.9447,
      "mean_token_accuracy": 0.7283599302172661,
      "step": 775
    },
    {
      "epoch": 0.7513546056592414,
      "grad_norm": 0.4053668975830078,
      "learning_rate": 2e-05,
      "loss": 0.8943,
      "mean_token_accuracy": 0.7407664448022843,
      "step": 780
    },
    {
      "epoch": 0.7561709813365443,
      "grad_norm": 0.377989798784256,
      "learning_rate": 2e-05,
      "loss": 0.897,
      "mean_token_accuracy": 0.7331721663475037,
      "step": 785
    },
    {
      "epoch": 0.7609873570138471,
      "grad_norm": 0.3637182414531708,
      "learning_rate": 2e-05,
      "loss": 0.9274,
      "mean_token_accuracy": 0.7238499134778976,
      "step": 790
    },
    {
      "epoch": 0.7658037326911499,
      "grad_norm": 0.3860657513141632,
      "learning_rate": 2e-05,
      "loss": 0.9038,
      "mean_token_accuracy": 0.7350373581051827,
      "step": 795
    },
    {
      "epoch": 0.7706201083684527,
      "grad_norm": 0.3994176387786865,
      "learning_rate": 2e-05,
      "loss": 0.9671,
      "mean_token_accuracy": 0.7155893579125404,
      "step": 800
    },
    {
      "epoch": 0.7706201083684527,
      "eval_loss": 0.964062511920929,
      "eval_mean_token_accuracy": 0.7233101450479947,
      "eval_runtime": 23.129,
      "eval_samples_per_second": 4.324,
      "eval_steps_per_second": 0.562,
      "step": 800
    },
    {
      "epoch": 0.7754364840457556,
      "grad_norm": 0.4492134153842926,
      "learning_rate": 2e-05,
      "loss": 0.9353,
      "mean_token_accuracy": 0.7261847347021103,
      "step": 805
    },
    {
      "epoch": 0.7802528597230584,
      "grad_norm": 0.45549172163009644,
      "learning_rate": 2e-05,
      "loss": 0.9376,
      "mean_token_accuracy": 0.7261715829372406,
      "step": 810
    },
    {
      "epoch": 0.7850692354003612,
      "grad_norm": 0.378998726606369,
      "learning_rate": 2e-05,
      "loss": 0.9256,
      "mean_token_accuracy": 0.7321177124977112,
      "step": 815
    },
    {
      "epoch": 0.7898856110776641,
      "grad_norm": 0.35385045409202576,
      "learning_rate": 2e-05,
      "loss": 0.908,
      "mean_token_accuracy": 0.7342436522245407,
      "step": 820
    },
    {
      "epoch": 0.7947019867549668,
      "grad_norm": 0.3778769373893738,
      "learning_rate": 2e-05,
      "loss": 0.919,
      "mean_token_accuracy": 0.7232446476817131,
      "step": 825
    },
    {
      "epoch": 0.7995183624322697,
      "grad_norm": 0.36614373326301575,
      "learning_rate": 2e-05,
      "loss": 0.9074,
      "mean_token_accuracy": 0.7344337999820709,
      "step": 830
    },
    {
      "epoch": 0.8043347381095726,
      "grad_norm": 0.36225926876068115,
      "learning_rate": 2e-05,
      "loss": 0.8927,
      "mean_token_accuracy": 0.7368348643183709,
      "step": 835
    },
    {
      "epoch": 0.8091511137868754,
      "grad_norm": 0.3926050662994385,
      "learning_rate": 2e-05,
      "loss": 0.8856,
      "mean_token_accuracy": 0.7389758110046387,
      "step": 840
    },
    {
      "epoch": 0.8139674894641782,
      "grad_norm": 0.3587627708911896,
      "learning_rate": 2e-05,
      "loss": 0.9458,
      "mean_token_accuracy": 0.7253442853689194,
      "step": 845
    },
    {
      "epoch": 0.818783865141481,
      "grad_norm": 0.33780330419540405,
      "learning_rate": 2e-05,
      "loss": 0.9201,
      "mean_token_accuracy": 0.7290245488286018,
      "step": 850
    },
    {
      "epoch": 0.8236002408187839,
      "grad_norm": 0.36326298117637634,
      "learning_rate": 2e-05,
      "loss": 0.8672,
      "mean_token_accuracy": 0.7469315692782402,
      "step": 855
    },
    {
      "epoch": 0.8284166164960867,
      "grad_norm": 0.3772122859954834,
      "learning_rate": 2e-05,
      "loss": 0.9153,
      "mean_token_accuracy": 0.7269368454813957,
      "step": 860
    },
    {
      "epoch": 0.8332329921733895,
      "grad_norm": 0.3824109137058258,
      "learning_rate": 2e-05,
      "loss": 0.922,
      "mean_token_accuracy": 0.7278964444994926,
      "step": 865
    },
    {
      "epoch": 0.8380493678506924,
      "grad_norm": 0.42278799414634705,
      "learning_rate": 2e-05,
      "loss": 0.921,
      "mean_token_accuracy": 0.730242757499218,
      "step": 870
    },
    {
      "epoch": 0.8428657435279951,
      "grad_norm": 0.37090981006622314,
      "learning_rate": 2e-05,
      "loss": 0.8888,
      "mean_token_accuracy": 0.7369868889451027,
      "step": 875
    },
    {
      "epoch": 0.847682119205298,
      "grad_norm": 0.3563065230846405,
      "learning_rate": 2e-05,
      "loss": 0.88,
      "mean_token_accuracy": 0.7384928792715073,
      "step": 880
    },
    {
      "epoch": 0.8524984948826009,
      "grad_norm": 0.3902587592601776,
      "learning_rate": 2e-05,
      "loss": 0.9187,
      "mean_token_accuracy": 0.7276976093649864,
      "step": 885
    },
    {
      "epoch": 0.8573148705599036,
      "grad_norm": 0.4171347916126251,
      "learning_rate": 2e-05,
      "loss": 0.9003,
      "mean_token_accuracy": 0.7332590997219086,
      "step": 890
    },
    {
      "epoch": 0.8621312462372065,
      "grad_norm": 0.39231064915657043,
      "learning_rate": 2e-05,
      "loss": 0.9098,
      "mean_token_accuracy": 0.7335718736052513,
      "step": 895
    },
    {
      "epoch": 0.8669476219145094,
      "grad_norm": 0.37322694063186646,
      "learning_rate": 2e-05,
      "loss": 0.8997,
      "mean_token_accuracy": 0.7329329386353493,
      "step": 900
    },
    {
      "epoch": 0.8669476219145094,
      "eval_loss": 0.9554687738418579,
      "eval_mean_token_accuracy": 0.7253300134952252,
      "eval_runtime": 23.1858,
      "eval_samples_per_second": 4.313,
      "eval_steps_per_second": 0.561,
      "step": 900
    },
    {
      "epoch": 0.8717639975918121,
      "grad_norm": 0.4675639271736145,
      "learning_rate": 2e-05,
      "loss": 0.9078,
      "mean_token_accuracy": 0.7307937040925025,
      "step": 905
    },
    {
      "epoch": 0.876580373269115,
      "grad_norm": 0.37796643376350403,
      "learning_rate": 2e-05,
      "loss": 0.915,
      "mean_token_accuracy": 0.7300151839852334,
      "step": 910
    },
    {
      "epoch": 0.8813967489464178,
      "grad_norm": 0.3831245005130768,
      "learning_rate": 2e-05,
      "loss": 0.8853,
      "mean_token_accuracy": 0.7340099573135376,
      "step": 915
    },
    {
      "epoch": 0.8862131246237207,
      "grad_norm": 0.4262712299823761,
      "learning_rate": 2e-05,
      "loss": 0.8898,
      "mean_token_accuracy": 0.7373559743165969,
      "step": 920
    },
    {
      "epoch": 0.8910295003010235,
      "grad_norm": 0.3827621638774872,
      "learning_rate": 2e-05,
      "loss": 0.9313,
      "mean_token_accuracy": 0.721035298705101,
      "step": 925
    },
    {
      "epoch": 0.8958458759783263,
      "grad_norm": 0.358642041683197,
      "learning_rate": 2e-05,
      "loss": 0.8993,
      "mean_token_accuracy": 0.7342952102422714,
      "step": 930
    },
    {
      "epoch": 0.9006622516556292,
      "grad_norm": 0.45269307494163513,
      "learning_rate": 2e-05,
      "loss": 0.9201,
      "mean_token_accuracy": 0.7350223436951637,
      "step": 935
    },
    {
      "epoch": 0.9054786273329319,
      "grad_norm": 0.4106045663356781,
      "learning_rate": 2e-05,
      "loss": 0.8975,
      "mean_token_accuracy": 0.7355057403445244,
      "step": 940
    },
    {
      "epoch": 0.9102950030102348,
      "grad_norm": 0.43229761719703674,
      "learning_rate": 2e-05,
      "loss": 0.9174,
      "mean_token_accuracy": 0.7289028167724609,
      "step": 945
    },
    {
      "epoch": 0.9151113786875377,
      "grad_norm": 0.4135943353176117,
      "learning_rate": 2e-05,
      "loss": 0.8568,
      "mean_token_accuracy": 0.7484120547771453,
      "step": 950
    },
    {
      "epoch": 0.9199277543648404,
      "grad_norm": 0.3772605359554291,
      "learning_rate": 2e-05,
      "loss": 0.8858,
      "mean_token_accuracy": 0.7402070254087448,
      "step": 955
    },
    {
      "epoch": 0.9247441300421433,
      "grad_norm": 0.38806506991386414,
      "learning_rate": 2e-05,
      "loss": 0.8839,
      "mean_token_accuracy": 0.738246065378189,
      "step": 960
    },
    {
      "epoch": 0.9295605057194462,
      "grad_norm": 0.37721186876296997,
      "learning_rate": 2e-05,
      "loss": 0.9446,
      "mean_token_accuracy": 0.7229430004954338,
      "step": 965
    },
    {
      "epoch": 0.9343768813967489,
      "grad_norm": 0.37863266468048096,
      "learning_rate": 2e-05,
      "loss": 0.8968,
      "mean_token_accuracy": 0.735011525452137,
      "step": 970
    },
    {
      "epoch": 0.9391932570740518,
      "grad_norm": 0.35967886447906494,
      "learning_rate": 2e-05,
      "loss": 0.9053,
      "mean_token_accuracy": 0.7334525480866432,
      "step": 975
    },
    {
      "epoch": 0.9440096327513546,
      "grad_norm": 0.36415719985961914,
      "learning_rate": 2e-05,
      "loss": 0.8855,
      "mean_token_accuracy": 0.7396774977445603,
      "step": 980
    },
    {
      "epoch": 0.9488260084286574,
      "grad_norm": 0.3698769509792328,
      "learning_rate": 2e-05,
      "loss": 0.8916,
      "mean_token_accuracy": 0.7383552804589272,
      "step": 985
    },
    {
      "epoch": 0.9536423841059603,
      "grad_norm": 0.38771921396255493,
      "learning_rate": 2e-05,
      "loss": 0.9274,
      "mean_token_accuracy": 0.7202894672751426,
      "step": 990
    },
    {
      "epoch": 0.9584587597832631,
      "grad_norm": 0.42747798562049866,
      "learning_rate": 2e-05,
      "loss": 0.9301,
      "mean_token_accuracy": 0.7242205336689949,
      "step": 995
    },
    {
      "epoch": 0.963275135460566,
      "grad_norm": 0.36320194602012634,
      "learning_rate": 2e-05,
      "loss": 0.9062,
      "mean_token_accuracy": 0.7277265936136246,
      "step": 1000
    },
    {
      "epoch": 0.963275135460566,
      "eval_loss": 0.948437511920929,
      "eval_mean_token_accuracy": 0.7268086809378403,
      "eval_runtime": 23.2366,
      "eval_samples_per_second": 4.304,
      "eval_steps_per_second": 0.559,
      "step": 1000
    },
    {
      "epoch": 0.9680915111378687,
      "grad_norm": 0.38512277603149414,
      "learning_rate": 2e-05,
      "loss": 0.8589,
      "mean_token_accuracy": 0.7454344347119332,
      "step": 1005
    },
    {
      "epoch": 0.9729078868151716,
      "grad_norm": 0.35500526428222656,
      "learning_rate": 2e-05,
      "loss": 0.8943,
      "mean_token_accuracy": 0.7319369062781333,
      "step": 1010
    },
    {
      "epoch": 0.9777242624924745,
      "grad_norm": 0.4057998061180115,
      "learning_rate": 2e-05,
      "loss": 0.8865,
      "mean_token_accuracy": 0.7375865921378135,
      "step": 1015
    },
    {
      "epoch": 0.9825406381697772,
      "grad_norm": 0.40733131766319275,
      "learning_rate": 2e-05,
      "loss": 0.9033,
      "mean_token_accuracy": 0.7339003443717956,
      "step": 1020
    },
    {
      "epoch": 0.9873570138470801,
      "grad_norm": 0.3689517676830292,
      "learning_rate": 2e-05,
      "loss": 0.9072,
      "mean_token_accuracy": 0.7332539647817612,
      "step": 1025
    },
    {
      "epoch": 0.9921733895243829,
      "grad_norm": 0.3920786678791046,
      "learning_rate": 2e-05,
      "loss": 0.8831,
      "mean_token_accuracy": 0.7399213343858719,
      "step": 1030
    },
    {
      "epoch": 0.9969897652016857,
      "grad_norm": 0.3332076072692871,
      "learning_rate": 2e-05,
      "loss": 0.8808,
      "mean_token_accuracy": 0.7407150834798812,
      "step": 1035
    },
    {
      "epoch": 1.0009632751354605,
      "grad_norm": 0.43358659744262695,
      "learning_rate": 2e-05,
      "loss": 0.7348,
      "mean_token_accuracy": 0.7383806271986528,
      "step": 1040
    },
    {
      "epoch": 1.0057796508127634,
      "grad_norm": 0.4309595823287964,
      "learning_rate": 2e-05,
      "loss": 0.7893,
      "mean_token_accuracy": 0.7559069380164146,
      "step": 1045
    },
    {
      "epoch": 1.0105960264900662,
      "grad_norm": 0.4139677584171295,
      "learning_rate": 2e-05,
      "loss": 0.8363,
      "mean_token_accuracy": 0.7487693518400192,
      "step": 1050
    },
    {
      "epoch": 1.0154124021673692,
      "grad_norm": 0.37942108511924744,
      "learning_rate": 2e-05,
      "loss": 0.8459,
      "mean_token_accuracy": 0.7475774019956589,
      "step": 1055
    },
    {
      "epoch": 1.020228777844672,
      "grad_norm": 0.42199230194091797,
      "learning_rate": 2e-05,
      "loss": 0.8702,
      "mean_token_accuracy": 0.7373737841844559,
      "step": 1060
    },
    {
      "epoch": 1.0250451535219747,
      "grad_norm": 0.36780673265457153,
      "learning_rate": 2e-05,
      "loss": 0.8534,
      "mean_token_accuracy": 0.7421116903424263,
      "step": 1065
    },
    {
      "epoch": 1.0298615291992776,
      "grad_norm": 0.3655712306499481,
      "learning_rate": 2e-05,
      "loss": 0.8861,
      "mean_token_accuracy": 0.7346468567848206,
      "step": 1070
    },
    {
      "epoch": 1.0346779048765804,
      "grad_norm": 0.3763968348503113,
      "learning_rate": 2e-05,
      "loss": 0.869,
      "mean_token_accuracy": 0.7416014015674591,
      "step": 1075
    },
    {
      "epoch": 1.0394942805538832,
      "grad_norm": 0.3777642548084259,
      "learning_rate": 2e-05,
      "loss": 0.882,
      "mean_token_accuracy": 0.7347766742110252,
      "step": 1080
    },
    {
      "epoch": 1.0443106562311861,
      "grad_norm": 0.37617817521095276,
      "learning_rate": 2e-05,
      "loss": 0.8808,
      "mean_token_accuracy": 0.7349648147821426,
      "step": 1085
    },
    {
      "epoch": 1.0491270319084889,
      "grad_norm": 0.37556421756744385,
      "learning_rate": 2e-05,
      "loss": 0.8426,
      "mean_token_accuracy": 0.7498120263218879,
      "step": 1090
    },
    {
      "epoch": 1.0539434075857916,
      "grad_norm": 0.3898829221725464,
      "learning_rate": 2e-05,
      "loss": 0.8365,
      "mean_token_accuracy": 0.746685053408146,
      "step": 1095
    },
    {
      "epoch": 1.0587597832630946,
      "grad_norm": 0.3657301068305969,
      "learning_rate": 2e-05,
      "loss": 0.865,
      "mean_token_accuracy": 0.7422873020172119,
      "step": 1100
    },
    {
      "epoch": 1.0587597832630946,
      "eval_loss": 0.9439062476158142,
      "eval_mean_token_accuracy": 0.7276775607695947,
      "eval_runtime": 23.2077,
      "eval_samples_per_second": 4.309,
      "eval_steps_per_second": 0.56,
      "step": 1100
    },
    {
      "epoch": 1.0635761589403974,
      "grad_norm": 0.3528231084346771,
      "learning_rate": 2e-05,
      "loss": 0.8491,
      "mean_token_accuracy": 0.7472317442297935,
      "step": 1105
    },
    {
      "epoch": 1.0683925346177001,
      "grad_norm": 0.3855115473270416,
      "learning_rate": 2e-05,
      "loss": 0.8343,
      "mean_token_accuracy": 0.7537892758846283,
      "step": 1110
    },
    {
      "epoch": 1.073208910295003,
      "grad_norm": 0.37757667899131775,
      "learning_rate": 2e-05,
      "loss": 0.8523,
      "mean_token_accuracy": 0.7377029240131379,
      "step": 1115
    },
    {
      "epoch": 1.0780252859723058,
      "grad_norm": 0.363437682390213,
      "learning_rate": 2e-05,
      "loss": 0.8576,
      "mean_token_accuracy": 0.7391376078128815,
      "step": 1120
    },
    {
      "epoch": 1.0828416616496086,
      "grad_norm": 0.366401344537735,
      "learning_rate": 2e-05,
      "loss": 0.854,
      "mean_token_accuracy": 0.7467326611280442,
      "step": 1125
    },
    {
      "epoch": 1.0876580373269116,
      "grad_norm": 0.3566599488258362,
      "learning_rate": 2e-05,
      "loss": 0.8801,
      "mean_token_accuracy": 0.7326125755906105,
      "step": 1130
    },
    {
      "epoch": 1.0924744130042143,
      "grad_norm": 0.34054407477378845,
      "learning_rate": 2e-05,
      "loss": 0.8527,
      "mean_token_accuracy": 0.7444597333669662,
      "step": 1135
    },
    {
      "epoch": 1.097290788681517,
      "grad_norm": 0.4111313223838806,
      "learning_rate": 2e-05,
      "loss": 0.8497,
      "mean_token_accuracy": 0.7466902017593384,
      "step": 1140
    },
    {
      "epoch": 1.10210716435882,
      "grad_norm": 0.3806932270526886,
      "learning_rate": 2e-05,
      "loss": 0.8485,
      "mean_token_accuracy": 0.7481859296560287,
      "step": 1145
    },
    {
      "epoch": 1.1069235400361228,
      "grad_norm": 0.3866956830024719,
      "learning_rate": 2e-05,
      "loss": 0.8138,
      "mean_token_accuracy": 0.7499373897910118,
      "step": 1150
    },
    {
      "epoch": 1.1117399157134256,
      "grad_norm": 0.37800461053848267,
      "learning_rate": 2e-05,
      "loss": 0.8537,
      "mean_token_accuracy": 0.7427974909543991,
      "step": 1155
    },
    {
      "epoch": 1.1165562913907285,
      "grad_norm": 0.3657369911670685,
      "learning_rate": 2e-05,
      "loss": 0.8643,
      "mean_token_accuracy": 0.7425652265548706,
      "step": 1160
    },
    {
      "epoch": 1.1213726670680313,
      "grad_norm": 0.34589388966560364,
      "learning_rate": 2e-05,
      "loss": 0.8236,
      "mean_token_accuracy": 0.746645887196064,
      "step": 1165
    },
    {
      "epoch": 1.126189042745334,
      "grad_norm": 0.3889903426170349,
      "learning_rate": 2e-05,
      "loss": 0.8502,
      "mean_token_accuracy": 0.7441315606236458,
      "step": 1170
    },
    {
      "epoch": 1.131005418422637,
      "grad_norm": 0.3654171824455261,
      "learning_rate": 2e-05,
      "loss": 0.8639,
      "mean_token_accuracy": 0.7370901137590409,
      "step": 1175
    },
    {
      "epoch": 1.1358217940999398,
      "grad_norm": 0.3497660756111145,
      "learning_rate": 2e-05,
      "loss": 0.8617,
      "mean_token_accuracy": 0.741532389819622,
      "step": 1180
    },
    {
      "epoch": 1.1406381697772425,
      "grad_norm": 0.36609187722206116,
      "learning_rate": 2e-05,
      "loss": 0.8618,
      "mean_token_accuracy": 0.7446050852537155,
      "step": 1185
    },
    {
      "epoch": 1.1454545454545455,
      "grad_norm": 0.4400254786014557,
      "learning_rate": 2e-05,
      "loss": 0.8326,
      "mean_token_accuracy": 0.7501450702548027,
      "step": 1190
    },
    {
      "epoch": 1.1502709211318483,
      "grad_norm": 0.3326510190963745,
      "learning_rate": 2e-05,
      "loss": 0.8418,
      "mean_token_accuracy": 0.744844700396061,
      "step": 1195
    },
    {
      "epoch": 1.155087296809151,
      "grad_norm": 0.37238839268684387,
      "learning_rate": 2e-05,
      "loss": 0.8586,
      "mean_token_accuracy": 0.7447129786014557,
      "step": 1200
    },
    {
      "epoch": 1.155087296809151,
      "eval_loss": 0.9375,
      "eval_mean_token_accuracy": 0.7290752346699054,
      "eval_runtime": 23.1543,
      "eval_samples_per_second": 4.319,
      "eval_steps_per_second": 0.561,
      "step": 1200
    },
    {
      "epoch": 1.159903672486454,
      "grad_norm": 0.3780873119831085,
      "learning_rate": 2e-05,
      "loss": 0.8274,
      "mean_token_accuracy": 0.7483693122863769,
      "step": 1205
    },
    {
      "epoch": 1.1647200481637567,
      "grad_norm": 0.404081255197525,
      "learning_rate": 2e-05,
      "loss": 0.8249,
      "mean_token_accuracy": 0.7522385701537132,
      "step": 1210
    },
    {
      "epoch": 1.1695364238410595,
      "grad_norm": 0.3591543734073639,
      "learning_rate": 2e-05,
      "loss": 0.8128,
      "mean_token_accuracy": 0.7550233393907547,
      "step": 1215
    },
    {
      "epoch": 1.1743527995183625,
      "grad_norm": 0.3817455768585205,
      "learning_rate": 2e-05,
      "loss": 0.8393,
      "mean_token_accuracy": 0.7501140117645264,
      "step": 1220
    },
    {
      "epoch": 1.1791691751956652,
      "grad_norm": 0.3948347270488739,
      "learning_rate": 2e-05,
      "loss": 0.8172,
      "mean_token_accuracy": 0.7537783578038215,
      "step": 1225
    },
    {
      "epoch": 1.183985550872968,
      "grad_norm": 0.39961639046669006,
      "learning_rate": 2e-05,
      "loss": 0.8531,
      "mean_token_accuracy": 0.7440052881836892,
      "step": 1230
    },
    {
      "epoch": 1.188801926550271,
      "grad_norm": 0.3994446098804474,
      "learning_rate": 2e-05,
      "loss": 0.8193,
      "mean_token_accuracy": 0.7571675822138786,
      "step": 1235
    },
    {
      "epoch": 1.1936183022275737,
      "grad_norm": 0.3635159730911255,
      "learning_rate": 2e-05,
      "loss": 0.8042,
      "mean_token_accuracy": 0.7555163532495499,
      "step": 1240
    },
    {
      "epoch": 1.1984346779048767,
      "grad_norm": 0.38725772500038147,
      "learning_rate": 2e-05,
      "loss": 0.867,
      "mean_token_accuracy": 0.7381075561046601,
      "step": 1245
    },
    {
      "epoch": 1.2032510535821794,
      "grad_norm": 0.4206209182739258,
      "learning_rate": 2e-05,
      "loss": 0.9091,
      "mean_token_accuracy": 0.727660009264946,
      "step": 1250
    },
    {
      "epoch": 1.2080674292594822,
      "grad_norm": 0.3780755400657654,
      "learning_rate": 2e-05,
      "loss": 0.8501,
      "mean_token_accuracy": 0.740609385073185,
      "step": 1255
    },
    {
      "epoch": 1.2128838049367852,
      "grad_norm": 0.3919096887111664,
      "learning_rate": 2e-05,
      "loss": 0.8262,
      "mean_token_accuracy": 0.7502545565366745,
      "step": 1260
    },
    {
      "epoch": 1.217700180614088,
      "grad_norm": 0.4055858850479126,
      "learning_rate": 2e-05,
      "loss": 0.8326,
      "mean_token_accuracy": 0.746037882566452,
      "step": 1265
    },
    {
      "epoch": 1.2225165562913907,
      "grad_norm": 0.3605825901031494,
      "learning_rate": 2e-05,
      "loss": 0.8352,
      "mean_token_accuracy": 0.751497520506382,
      "step": 1270
    },
    {
      "epoch": 1.2273329319686936,
      "grad_norm": 0.3745580017566681,
      "learning_rate": 2e-05,
      "loss": 0.9102,
      "mean_token_accuracy": 0.7277455866336823,
      "step": 1275
    },
    {
      "epoch": 1.2321493076459964,
      "grad_norm": 0.4068564772605896,
      "learning_rate": 2e-05,
      "loss": 0.8439,
      "mean_token_accuracy": 0.7488078013062477,
      "step": 1280
    },
    {
      "epoch": 1.2369656833232991,
      "grad_norm": 0.33656877279281616,
      "learning_rate": 2e-05,
      "loss": 0.853,
      "mean_token_accuracy": 0.7449796929955482,
      "step": 1285
    },
    {
      "epoch": 1.2417820590006021,
      "grad_norm": 0.38442090153694153,
      "learning_rate": 2e-05,
      "loss": 0.8105,
      "mean_token_accuracy": 0.7490809202194214,
      "step": 1290
    },
    {
      "epoch": 1.2465984346779049,
      "grad_norm": 0.4118853211402893,
      "learning_rate": 2e-05,
      "loss": 0.8409,
      "mean_token_accuracy": 0.7445864990353585,
      "step": 1295
    },
    {
      "epoch": 1.2514148103552076,
      "grad_norm": 0.40977755188941956,
      "learning_rate": 2e-05,
      "loss": 0.8131,
      "mean_token_accuracy": 0.7540942341089248,
      "step": 1300
    },
    {
      "epoch": 1.2514148103552076,
      "eval_loss": 0.9334375262260437,
      "eval_mean_token_accuracy": 0.7299059858688941,
      "eval_runtime": 23.2243,
      "eval_samples_per_second": 4.306,
      "eval_steps_per_second": 0.56,
      "step": 1300
    },
    {
      "epoch": 1.2562311860325106,
      "grad_norm": 0.37511029839515686,
      "learning_rate": 2e-05,
      "loss": 0.8374,
      "mean_token_accuracy": 0.7419778451323509,
      "step": 1305
    },
    {
      "epoch": 1.2610475617098134,
      "grad_norm": 0.3589267134666443,
      "learning_rate": 2e-05,
      "loss": 0.8379,
      "mean_token_accuracy": 0.7468772605061531,
      "step": 1310
    },
    {
      "epoch": 1.2658639373871163,
      "grad_norm": 0.39658141136169434,
      "learning_rate": 2e-05,
      "loss": 0.8297,
      "mean_token_accuracy": 0.7469786286354065,
      "step": 1315
    },
    {
      "epoch": 1.270680313064419,
      "grad_norm": 0.42367202043533325,
      "learning_rate": 2e-05,
      "loss": 0.8337,
      "mean_token_accuracy": 0.7481065571308136,
      "step": 1320
    },
    {
      "epoch": 1.2754966887417218,
      "grad_norm": 0.3799798786640167,
      "learning_rate": 2e-05,
      "loss": 0.8699,
      "mean_token_accuracy": 0.7395881712436676,
      "step": 1325
    },
    {
      "epoch": 1.2803130644190248,
      "grad_norm": 0.35513991117477417,
      "learning_rate": 2e-05,
      "loss": 0.842,
      "mean_token_accuracy": 0.7494134768843651,
      "step": 1330
    },
    {
      "epoch": 1.2851294400963276,
      "grad_norm": 0.3369906544685364,
      "learning_rate": 2e-05,
      "loss": 0.8505,
      "mean_token_accuracy": 0.7420820742845535,
      "step": 1335
    },
    {
      "epoch": 1.2899458157736303,
      "grad_norm": 0.37939414381980896,
      "learning_rate": 2e-05,
      "loss": 0.8575,
      "mean_token_accuracy": 0.7391051918268203,
      "step": 1340
    },
    {
      "epoch": 1.2947621914509333,
      "grad_norm": 0.4896661043167114,
      "learning_rate": 2e-05,
      "loss": 0.8597,
      "mean_token_accuracy": 0.7398741409182549,
      "step": 1345
    },
    {
      "epoch": 1.299578567128236,
      "grad_norm": 0.36100226640701294,
      "learning_rate": 2e-05,
      "loss": 0.8282,
      "mean_token_accuracy": 0.7494728401303291,
      "step": 1350
    },
    {
      "epoch": 1.3043949428055388,
      "grad_norm": 0.35935041308403015,
      "learning_rate": 2e-05,
      "loss": 0.8384,
      "mean_token_accuracy": 0.7416606619954109,
      "step": 1355
    },
    {
      "epoch": 1.3092113184828418,
      "grad_norm": 0.3769769072532654,
      "learning_rate": 2e-05,
      "loss": 0.8533,
      "mean_token_accuracy": 0.7399677246809006,
      "step": 1360
    },
    {
      "epoch": 1.3140276941601445,
      "grad_norm": 0.3857510983943939,
      "learning_rate": 2e-05,
      "loss": 0.8071,
      "mean_token_accuracy": 0.7550437197089195,
      "step": 1365
    },
    {
      "epoch": 1.3188440698374473,
      "grad_norm": 0.3967873752117157,
      "learning_rate": 2e-05,
      "loss": 0.8172,
      "mean_token_accuracy": 0.7505990505218506,
      "step": 1370
    },
    {
      "epoch": 1.3236604455147503,
      "grad_norm": 0.36245691776275635,
      "learning_rate": 2e-05,
      "loss": 0.8492,
      "mean_token_accuracy": 0.7389700740575791,
      "step": 1375
    },
    {
      "epoch": 1.328476821192053,
      "grad_norm": 0.35738521814346313,
      "learning_rate": 2e-05,
      "loss": 0.8273,
      "mean_token_accuracy": 0.74720598757267,
      "step": 1380
    },
    {
      "epoch": 1.3332931968693558,
      "grad_norm": 0.3659331202507019,
      "learning_rate": 2e-05,
      "loss": 0.8284,
      "mean_token_accuracy": 0.7490911051630974,
      "step": 1385
    },
    {
      "epoch": 1.3381095725466587,
      "grad_norm": 0.4177457392215729,
      "learning_rate": 2e-05,
      "loss": 0.8674,
      "mean_token_accuracy": 0.7377855911850929,
      "step": 1390
    },
    {
      "epoch": 1.3429259482239615,
      "grad_norm": 0.35989993810653687,
      "learning_rate": 2e-05,
      "loss": 0.8047,
      "mean_token_accuracy": 0.7577745035290718,
      "step": 1395
    },
    {
      "epoch": 1.3477423239012642,
      "grad_norm": 0.3731432557106018,
      "learning_rate": 2e-05,
      "loss": 0.8361,
      "mean_token_accuracy": 0.7453808888792992,
      "step": 1400
    },
    {
      "epoch": 1.3477423239012642,
      "eval_loss": 0.9284374713897705,
      "eval_mean_token_accuracy": 0.7310354984723605,
      "eval_runtime": 23.1671,
      "eval_samples_per_second": 4.316,
      "eval_steps_per_second": 0.561,
      "step": 1400
    },
    {
      "epoch": 1.3525586995785672,
      "grad_norm": 0.3571363389492035,
      "learning_rate": 2e-05,
      "loss": 0.8256,
      "mean_token_accuracy": 0.7432364627718926,
      "step": 1405
    },
    {
      "epoch": 1.35737507525587,
      "grad_norm": 0.3314119279384613,
      "learning_rate": 2e-05,
      "loss": 0.8363,
      "mean_token_accuracy": 0.7436547055840492,
      "step": 1410
    },
    {
      "epoch": 1.3621914509331727,
      "grad_norm": 0.3417344093322754,
      "learning_rate": 2e-05,
      "loss": 0.8472,
      "mean_token_accuracy": 0.7448557123541832,
      "step": 1415
    },
    {
      "epoch": 1.3670078266104757,
      "grad_norm": 0.3343771994113922,
      "learning_rate": 2e-05,
      "loss": 0.8758,
      "mean_token_accuracy": 0.7367025226354599,
      "step": 1420
    },
    {
      "epoch": 1.3718242022877785,
      "grad_norm": 0.40026524662971497,
      "learning_rate": 2e-05,
      "loss": 0.847,
      "mean_token_accuracy": 0.7458574742078781,
      "step": 1425
    },
    {
      "epoch": 1.3766405779650812,
      "grad_norm": 0.3898317813873291,
      "learning_rate": 2e-05,
      "loss": 0.8698,
      "mean_token_accuracy": 0.7364508762955666,
      "step": 1430
    },
    {
      "epoch": 1.3814569536423842,
      "grad_norm": 0.38850364089012146,
      "learning_rate": 2e-05,
      "loss": 0.8153,
      "mean_token_accuracy": 0.7526135891675949,
      "step": 1435
    },
    {
      "epoch": 1.386273329319687,
      "grad_norm": 0.3574685752391815,
      "learning_rate": 2e-05,
      "loss": 0.8464,
      "mean_token_accuracy": 0.7421124458312989,
      "step": 1440
    },
    {
      "epoch": 1.3910897049969897,
      "grad_norm": 0.3952126204967499,
      "learning_rate": 2e-05,
      "loss": 0.8529,
      "mean_token_accuracy": 0.7436041012406349,
      "step": 1445
    },
    {
      "epoch": 1.3959060806742927,
      "grad_norm": 0.3442822992801666,
      "learning_rate": 2e-05,
      "loss": 0.8072,
      "mean_token_accuracy": 0.7531652361154556,
      "step": 1450
    },
    {
      "epoch": 1.4007224563515954,
      "grad_norm": 0.39355242252349854,
      "learning_rate": 2e-05,
      "loss": 0.8571,
      "mean_token_accuracy": 0.7427126541733742,
      "step": 1455
    },
    {
      "epoch": 1.4055388320288982,
      "grad_norm": 0.37595421075820923,
      "learning_rate": 2e-05,
      "loss": 0.8479,
      "mean_token_accuracy": 0.742530232667923,
      "step": 1460
    },
    {
      "epoch": 1.4103552077062012,
      "grad_norm": 0.37319841980934143,
      "learning_rate": 2e-05,
      "loss": 0.8995,
      "mean_token_accuracy": 0.7270940363407135,
      "step": 1465
    },
    {
      "epoch": 1.415171583383504,
      "grad_norm": 0.4596470892429352,
      "learning_rate": 2e-05,
      "loss": 0.8115,
      "mean_token_accuracy": 0.7522259518504143,
      "step": 1470
    },
    {
      "epoch": 1.4199879590608067,
      "grad_norm": 0.3937012553215027,
      "learning_rate": 2e-05,
      "loss": 0.818,
      "mean_token_accuracy": 0.7493147507309914,
      "step": 1475
    },
    {
      "epoch": 1.4248043347381096,
      "grad_norm": 0.37263306975364685,
      "learning_rate": 2e-05,
      "loss": 0.8398,
      "mean_token_accuracy": 0.7486390605568886,
      "step": 1480
    },
    {
      "epoch": 1.4296207104154124,
      "grad_norm": 0.3784118592739105,
      "learning_rate": 2e-05,
      "loss": 0.8002,
      "mean_token_accuracy": 0.7569744929671287,
      "step": 1485
    },
    {
      "epoch": 1.4344370860927151,
      "grad_norm": 0.36756840348243713,
      "learning_rate": 2e-05,
      "loss": 0.8808,
      "mean_token_accuracy": 0.735991969704628,
      "step": 1490
    },
    {
      "epoch": 1.4392534617700181,
      "grad_norm": 0.37854087352752686,
      "learning_rate": 2e-05,
      "loss": 0.8555,
      "mean_token_accuracy": 0.7391902863979339,
      "step": 1495
    },
    {
      "epoch": 1.4440698374473209,
      "grad_norm": 0.37972742319107056,
      "learning_rate": 2e-05,
      "loss": 0.8449,
      "mean_token_accuracy": 0.7450771510601044,
      "step": 1500
    },
    {
      "epoch": 1.4440698374473209,
      "eval_loss": 0.9237499833106995,
      "eval_mean_token_accuracy": 0.7318991560202378,
      "eval_runtime": 23.0893,
      "eval_samples_per_second": 4.331,
      "eval_steps_per_second": 0.563,
      "step": 1500
    },
    {
      "epoch": 1.4488862131246236,
      "grad_norm": 0.37376001477241516,
      "learning_rate": 2e-05,
      "loss": 0.8485,
      "mean_token_accuracy": 0.7402021259069442,
      "step": 1505
    },
    {
      "epoch": 1.4537025888019266,
      "grad_norm": 0.38736093044281006,
      "learning_rate": 2e-05,
      "loss": 0.8693,
      "mean_token_accuracy": 0.7362229198217392,
      "step": 1510
    },
    {
      "epoch": 1.4585189644792294,
      "grad_norm": 0.3877595067024231,
      "learning_rate": 2e-05,
      "loss": 0.8424,
      "mean_token_accuracy": 0.745487704873085,
      "step": 1515
    },
    {
      "epoch": 1.463335340156532,
      "grad_norm": 0.35137495398521423,
      "learning_rate": 2e-05,
      "loss": 0.8278,
      "mean_token_accuracy": 0.7506787940859795,
      "step": 1520
    },
    {
      "epoch": 1.468151715833835,
      "grad_norm": 0.37618595361709595,
      "learning_rate": 2e-05,
      "loss": 0.8311,
      "mean_token_accuracy": 0.7513900756835937,
      "step": 1525
    },
    {
      "epoch": 1.4729680915111378,
      "grad_norm": 0.3877982795238495,
      "learning_rate": 2e-05,
      "loss": 0.8406,
      "mean_token_accuracy": 0.746095435321331,
      "step": 1530
    },
    {
      "epoch": 1.4777844671884406,
      "grad_norm": 0.3873893618583679,
      "learning_rate": 2e-05,
      "loss": 0.8465,
      "mean_token_accuracy": 0.7434879496693612,
      "step": 1535
    },
    {
      "epoch": 1.4826008428657436,
      "grad_norm": 0.37615063786506653,
      "learning_rate": 2e-05,
      "loss": 0.8214,
      "mean_token_accuracy": 0.7459251716732979,
      "step": 1540
    },
    {
      "epoch": 1.4874172185430463,
      "grad_norm": 0.3867892324924469,
      "learning_rate": 2e-05,
      "loss": 0.8237,
      "mean_token_accuracy": 0.7544758155941963,
      "step": 1545
    },
    {
      "epoch": 1.492233594220349,
      "grad_norm": 0.3590559959411621,
      "learning_rate": 2e-05,
      "loss": 0.8196,
      "mean_token_accuracy": 0.7523276194930076,
      "step": 1550
    },
    {
      "epoch": 1.497049969897652,
      "grad_norm": 0.33671319484710693,
      "learning_rate": 2e-05,
      "loss": 0.8477,
      "mean_token_accuracy": 0.7436845555901528,
      "step": 1555
    },
    {
      "epoch": 1.5018663455749548,
      "grad_norm": 0.3445674180984497,
      "learning_rate": 2e-05,
      "loss": 0.8554,
      "mean_token_accuracy": 0.7416912838816643,
      "step": 1560
    },
    {
      "epoch": 1.5066827212522576,
      "grad_norm": 0.38447070121765137,
      "learning_rate": 2e-05,
      "loss": 0.8339,
      "mean_token_accuracy": 0.7485029339790344,
      "step": 1565
    },
    {
      "epoch": 1.5114990969295605,
      "grad_norm": 0.40366995334625244,
      "learning_rate": 2e-05,
      "loss": 0.8211,
      "mean_token_accuracy": 0.7500607624650002,
      "step": 1570
    },
    {
      "epoch": 1.5163154726068635,
      "grad_norm": 0.39034920930862427,
      "learning_rate": 2e-05,
      "loss": 0.8145,
      "mean_token_accuracy": 0.7497797414660454,
      "step": 1575
    },
    {
      "epoch": 1.521131848284166,
      "grad_norm": 0.35405203700065613,
      "learning_rate": 2e-05,
      "loss": 0.8527,
      "mean_token_accuracy": 0.7429638877511024,
      "step": 1580
    },
    {
      "epoch": 1.525948223961469,
      "grad_norm": 0.3475797474384308,
      "learning_rate": 2e-05,
      "loss": 0.8116,
      "mean_token_accuracy": 0.7482677146792411,
      "step": 1585
    },
    {
      "epoch": 1.530764599638772,
      "grad_norm": 0.38322359323501587,
      "learning_rate": 2e-05,
      "loss": 0.8077,
      "mean_token_accuracy": 0.7507022231817245,
      "step": 1590
    },
    {
      "epoch": 1.5355809753160745,
      "grad_norm": 0.4077315032482147,
      "learning_rate": 2e-05,
      "loss": 0.8108,
      "mean_token_accuracy": 0.7523728922009468,
      "step": 1595
    },
    {
      "epoch": 1.5403973509933775,
      "grad_norm": 0.41062474250793457,
      "learning_rate": 2e-05,
      "loss": 0.816,
      "mean_token_accuracy": 0.7487197533249855,
      "step": 1600
    },
    {
      "epoch": 1.5403973509933775,
      "eval_loss": 0.9193750023841858,
      "eval_mean_token_accuracy": 0.7336924580427316,
      "eval_runtime": 23.1661,
      "eval_samples_per_second": 4.317,
      "eval_steps_per_second": 0.561,
      "step": 1600
    },
    {
      "epoch": 1.5452137266706805,
      "grad_norm": 0.3489004671573639,
      "learning_rate": 2e-05,
      "loss": 0.8252,
      "mean_token_accuracy": 0.7542433932423591,
      "step": 1605
    },
    {
      "epoch": 1.550030102347983,
      "grad_norm": 0.42114558815956116,
      "learning_rate": 2e-05,
      "loss": 0.8513,
      "mean_token_accuracy": 0.7369220793247223,
      "step": 1610
    },
    {
      "epoch": 1.554846478025286,
      "grad_norm": 0.40832996368408203,
      "learning_rate": 2e-05,
      "loss": 0.8532,
      "mean_token_accuracy": 0.7461734786629677,
      "step": 1615
    },
    {
      "epoch": 1.559662853702589,
      "grad_norm": 0.369779497385025,
      "learning_rate": 2e-05,
      "loss": 0.8416,
      "mean_token_accuracy": 0.7418405711650848,
      "step": 1620
    },
    {
      "epoch": 1.5644792293798915,
      "grad_norm": 0.36488446593284607,
      "learning_rate": 2e-05,
      "loss": 0.8424,
      "mean_token_accuracy": 0.7447330921888351,
      "step": 1625
    },
    {
      "epoch": 1.5692956050571945,
      "grad_norm": 0.4260449707508087,
      "learning_rate": 2e-05,
      "loss": 0.8483,
      "mean_token_accuracy": 0.7396710678935051,
      "step": 1630
    },
    {
      "epoch": 1.5741119807344974,
      "grad_norm": 0.44844576716423035,
      "learning_rate": 2e-05,
      "loss": 0.832,
      "mean_token_accuracy": 0.7461599200963974,
      "step": 1635
    },
    {
      "epoch": 1.5789283564118,
      "grad_norm": 0.3739905059337616,
      "learning_rate": 2e-05,
      "loss": 0.7956,
      "mean_token_accuracy": 0.7564267441630363,
      "step": 1640
    },
    {
      "epoch": 1.583744732089103,
      "grad_norm": 0.3647367060184479,
      "learning_rate": 2e-05,
      "loss": 0.8269,
      "mean_token_accuracy": 0.7473939165472985,
      "step": 1645
    },
    {
      "epoch": 1.588561107766406,
      "grad_norm": 0.32268261909484863,
      "learning_rate": 2e-05,
      "loss": 0.8051,
      "mean_token_accuracy": 0.7524646311998368,
      "step": 1650
    },
    {
      "epoch": 1.5933774834437087,
      "grad_norm": 0.38236913084983826,
      "learning_rate": 2e-05,
      "loss": 0.8193,
      "mean_token_accuracy": 0.7506327718496323,
      "step": 1655
    },
    {
      "epoch": 1.5981938591210114,
      "grad_norm": 0.33767271041870117,
      "learning_rate": 2e-05,
      "loss": 0.8471,
      "mean_token_accuracy": 0.7457294464111328,
      "step": 1660
    },
    {
      "epoch": 1.6030102347983144,
      "grad_norm": 0.36715444922447205,
      "learning_rate": 2e-05,
      "loss": 0.827,
      "mean_token_accuracy": 0.7489886403083801,
      "step": 1665
    },
    {
      "epoch": 1.6078266104756171,
      "grad_norm": 0.3837079107761383,
      "learning_rate": 2e-05,
      "loss": 0.8183,
      "mean_token_accuracy": 0.750606831908226,
      "step": 1670
    },
    {
      "epoch": 1.61264298615292,
      "grad_norm": 0.4105582535266876,
      "learning_rate": 2e-05,
      "loss": 0.7927,
      "mean_token_accuracy": 0.7599633246660232,
      "step": 1675
    },
    {
      "epoch": 1.6174593618302229,
      "grad_norm": 0.328166127204895,
      "learning_rate": 2e-05,
      "loss": 0.8538,
      "mean_token_accuracy": 0.7361954420804977,
      "step": 1680
    },
    {
      "epoch": 1.6222757375075256,
      "grad_norm": 0.34064626693725586,
      "learning_rate": 2e-05,
      "loss": 0.8232,
      "mean_token_accuracy": 0.7506048232316971,
      "step": 1685
    },
    {
      "epoch": 1.6270921131848284,
      "grad_norm": 0.3536415100097656,
      "learning_rate": 2e-05,
      "loss": 0.8128,
      "mean_token_accuracy": 0.7551938936114311,
      "step": 1690
    },
    {
      "epoch": 1.6319084888621314,
      "grad_norm": 0.3961397707462311,
      "learning_rate": 2e-05,
      "loss": 0.845,
      "mean_token_accuracy": 0.743786184489727,
      "step": 1695
    },
    {
      "epoch": 1.636724864539434,
      "grad_norm": 0.3730252981185913,
      "learning_rate": 2e-05,
      "loss": 0.8177,
      "mean_token_accuracy": 0.7538556814193725,
      "step": 1700
    },
    {
      "epoch": 1.636724864539434,
      "eval_loss": 0.9150000214576721,
      "eval_mean_token_accuracy": 0.7336668784801776,
      "eval_runtime": 23.1715,
      "eval_samples_per_second": 4.316,
      "eval_steps_per_second": 0.561,
      "step": 1700
    },
    {
      "epoch": 1.6415412402167369,
      "grad_norm": 0.40690943598747253,
      "learning_rate": 2e-05,
      "loss": 0.8164,
      "mean_token_accuracy": 0.7524816766381264,
      "step": 1705
    },
    {
      "epoch": 1.6463576158940398,
      "grad_norm": 0.40426430106163025,
      "learning_rate": 2e-05,
      "loss": 0.8629,
      "mean_token_accuracy": 0.7397387072443962,
      "step": 1710
    },
    {
      "epoch": 1.6511739915713426,
      "grad_norm": 0.3870543837547302,
      "learning_rate": 2e-05,
      "loss": 0.8344,
      "mean_token_accuracy": 0.7449035495519638,
      "step": 1715
    },
    {
      "epoch": 1.6559903672486453,
      "grad_norm": 0.3909133970737457,
      "learning_rate": 2e-05,
      "loss": 0.8416,
      "mean_token_accuracy": 0.7425220400094986,
      "step": 1720
    },
    {
      "epoch": 1.6608067429259483,
      "grad_norm": 0.3573339581489563,
      "learning_rate": 2e-05,
      "loss": 0.831,
      "mean_token_accuracy": 0.749519394338131,
      "step": 1725
    },
    {
      "epoch": 1.665623118603251,
      "grad_norm": 0.4339994788169861,
      "learning_rate": 2e-05,
      "loss": 0.83,
      "mean_token_accuracy": 0.7467664986848831,
      "step": 1730
    },
    {
      "epoch": 1.6704394942805538,
      "grad_norm": 0.3833540081977844,
      "learning_rate": 2e-05,
      "loss": 0.8376,
      "mean_token_accuracy": 0.7486771330237388,
      "step": 1735
    },
    {
      "epoch": 1.6752558699578568,
      "grad_norm": 0.4031335711479187,
      "learning_rate": 2e-05,
      "loss": 0.7827,
      "mean_token_accuracy": 0.7610711231827736,
      "step": 1740
    },
    {
      "epoch": 1.6800722456351596,
      "grad_norm": 0.40794068574905396,
      "learning_rate": 2e-05,
      "loss": 0.824,
      "mean_token_accuracy": 0.7416556209325791,
      "step": 1745
    },
    {
      "epoch": 1.6848886213124623,
      "grad_norm": 0.3737964332103729,
      "learning_rate": 2e-05,
      "loss": 0.8187,
      "mean_token_accuracy": 0.7503210604190826,
      "step": 1750
    },
    {
      "epoch": 1.6897049969897653,
      "grad_norm": 0.33517488837242126,
      "learning_rate": 2e-05,
      "loss": 0.8333,
      "mean_token_accuracy": 0.7439797177910805,
      "step": 1755
    },
    {
      "epoch": 1.694521372667068,
      "grad_norm": 0.41663119196891785,
      "learning_rate": 2e-05,
      "loss": 0.888,
      "mean_token_accuracy": 0.7334117487072944,
      "step": 1760
    },
    {
      "epoch": 1.6993377483443708,
      "grad_norm": 0.339233934879303,
      "learning_rate": 2e-05,
      "loss": 0.8314,
      "mean_token_accuracy": 0.7454211816191674,
      "step": 1765
    },
    {
      "epoch": 1.7041541240216738,
      "grad_norm": 0.36906835436820984,
      "learning_rate": 2e-05,
      "loss": 0.8427,
      "mean_token_accuracy": 0.7440360963344574,
      "step": 1770
    },
    {
      "epoch": 1.7089704996989765,
      "grad_norm": 0.38175931572914124,
      "learning_rate": 2e-05,
      "loss": 0.8298,
      "mean_token_accuracy": 0.7456939503550529,
      "step": 1775
    },
    {
      "epoch": 1.7137868753762793,
      "grad_norm": 0.3851398527622223,
      "learning_rate": 2e-05,
      "loss": 0.8381,
      "mean_token_accuracy": 0.7432308256626129,
      "step": 1780
    },
    {
      "epoch": 1.7186032510535822,
      "grad_norm": 0.4011867046356201,
      "learning_rate": 2e-05,
      "loss": 0.8172,
      "mean_token_accuracy": 0.7504382818937302,
      "step": 1785
    },
    {
      "epoch": 1.723419626730885,
      "grad_norm": 0.4010428786277771,
      "learning_rate": 2e-05,
      "loss": 0.8278,
      "mean_token_accuracy": 0.7499980315566063,
      "step": 1790
    },
    {
      "epoch": 1.7282360024081878,
      "grad_norm": 0.35039839148521423,
      "learning_rate": 2e-05,
      "loss": 0.8456,
      "mean_token_accuracy": 0.7438850581645966,
      "step": 1795
    },
    {
      "epoch": 1.7330523780854907,
      "grad_norm": 0.37959855794906616,
      "learning_rate": 2e-05,
      "loss": 0.8414,
      "mean_token_accuracy": 0.748087365925312,
      "step": 1800
    },
    {
      "epoch": 1.7330523780854907,
      "eval_loss": 0.9110937714576721,
      "eval_mean_token_accuracy": 0.7353770962128272,
      "eval_runtime": 23.1605,
      "eval_samples_per_second": 4.318,
      "eval_steps_per_second": 0.561,
      "step": 1800
    },
    {
      "epoch": 1.7378687537627935,
      "grad_norm": 0.33767440915107727,
      "learning_rate": 2e-05,
      "loss": 0.8569,
      "mean_token_accuracy": 0.7423548147082328,
      "step": 1805
    },
    {
      "epoch": 1.7426851294400962,
      "grad_norm": 0.3997369408607483,
      "learning_rate": 2e-05,
      "loss": 0.8201,
      "mean_token_accuracy": 0.7456977978348732,
      "step": 1810
    },
    {
      "epoch": 1.7475015051173992,
      "grad_norm": 0.38806620240211487,
      "learning_rate": 2e-05,
      "loss": 0.8488,
      "mean_token_accuracy": 0.7421660631895065,
      "step": 1815
    },
    {
      "epoch": 1.752317880794702,
      "grad_norm": 0.3360239863395691,
      "learning_rate": 2e-05,
      "loss": 0.8266,
      "mean_token_accuracy": 0.7481269389390945,
      "step": 1820
    },
    {
      "epoch": 1.7571342564720047,
      "grad_norm": 0.41784462332725525,
      "learning_rate": 2e-05,
      "loss": 0.8148,
      "mean_token_accuracy": 0.7510803028941154,
      "step": 1825
    },
    {
      "epoch": 1.7619506321493077,
      "grad_norm": 0.3966754078865051,
      "learning_rate": 2e-05,
      "loss": 0.8139,
      "mean_token_accuracy": 0.7494689762592316,
      "step": 1830
    },
    {
      "epoch": 1.7667670078266104,
      "grad_norm": 0.36949893832206726,
      "learning_rate": 2e-05,
      "loss": 0.8528,
      "mean_token_accuracy": 0.7427579507231712,
      "step": 1835
    },
    {
      "epoch": 1.7715833835039132,
      "grad_norm": 0.37594905495643616,
      "learning_rate": 2e-05,
      "loss": 0.8478,
      "mean_token_accuracy": 0.746378231048584,
      "step": 1840
    },
    {
      "epoch": 1.7763997591812162,
      "grad_norm": 0.3548242151737213,
      "learning_rate": 2e-05,
      "loss": 0.8191,
      "mean_token_accuracy": 0.7453361362218857,
      "step": 1845
    },
    {
      "epoch": 1.781216134858519,
      "grad_norm": 0.3962383270263672,
      "learning_rate": 2e-05,
      "loss": 0.8207,
      "mean_token_accuracy": 0.7496239960193634,
      "step": 1850
    },
    {
      "epoch": 1.7860325105358217,
      "grad_norm": 0.3505515158176422,
      "learning_rate": 2e-05,
      "loss": 0.8609,
      "mean_token_accuracy": 0.7414278611540794,
      "step": 1855
    },
    {
      "epoch": 1.7908488862131247,
      "grad_norm": 0.3898848295211792,
      "learning_rate": 2e-05,
      "loss": 0.7912,
      "mean_token_accuracy": 0.755648897588253,
      "step": 1860
    },
    {
      "epoch": 1.7956652618904274,
      "grad_norm": 0.3252900242805481,
      "learning_rate": 2e-05,
      "loss": 0.8272,
      "mean_token_accuracy": 0.7487868279218673,
      "step": 1865
    },
    {
      "epoch": 1.8004816375677302,
      "grad_norm": 0.3451252281665802,
      "learning_rate": 2e-05,
      "loss": 0.8074,
      "mean_token_accuracy": 0.7537914827466011,
      "step": 1870
    },
    {
      "epoch": 1.8052980132450331,
      "grad_norm": 0.388134241104126,
      "learning_rate": 2e-05,
      "loss": 0.8195,
      "mean_token_accuracy": 0.7505655080080033,
      "step": 1875
    },
    {
      "epoch": 1.810114388922336,
      "grad_norm": 0.3933774530887604,
      "learning_rate": 2e-05,
      "loss": 0.828,
      "mean_token_accuracy": 0.7499305367469787,
      "step": 1880
    },
    {
      "epoch": 1.8149307645996386,
      "grad_norm": 0.41381701827049255,
      "learning_rate": 2e-05,
      "loss": 0.8403,
      "mean_token_accuracy": 0.74359240680933,
      "step": 1885
    },
    {
      "epoch": 1.8197471402769416,
      "grad_norm": 0.3744955062866211,
      "learning_rate": 2e-05,
      "loss": 0.8285,
      "mean_token_accuracy": 0.7447529405355453,
      "step": 1890
    },
    {
      "epoch": 1.8245635159542446,
      "grad_norm": 0.37023985385894775,
      "learning_rate": 2e-05,
      "loss": 0.8395,
      "mean_token_accuracy": 0.7450712293386459,
      "step": 1895
    },
    {
      "epoch": 1.8293798916315471,
      "grad_norm": 0.3904387354850769,
      "learning_rate": 2e-05,
      "loss": 0.806,
      "mean_token_accuracy": 0.7523078992962837,
      "step": 1900
    },
    {
      "epoch": 1.8293798916315471,
      "eval_loss": 0.9075000286102295,
      "eval_mean_token_accuracy": 0.7357059946426978,
      "eval_runtime": 23.1147,
      "eval_samples_per_second": 4.326,
      "eval_steps_per_second": 0.562,
      "step": 1900
    },
    {
      "epoch": 1.83419626730885,
      "grad_norm": 0.36936506628990173,
      "learning_rate": 2e-05,
      "loss": 0.8049,
      "mean_token_accuracy": 0.7516879752278328,
      "step": 1905
    },
    {
      "epoch": 1.839012642986153,
      "grad_norm": 0.37847042083740234,
      "learning_rate": 2e-05,
      "loss": 0.8447,
      "mean_token_accuracy": 0.7475872814655304,
      "step": 1910
    },
    {
      "epoch": 1.8438290186634556,
      "grad_norm": 0.3323959708213806,
      "learning_rate": 2e-05,
      "loss": 0.8357,
      "mean_token_accuracy": 0.7477746129035949,
      "step": 1915
    },
    {
      "epoch": 1.8486453943407586,
      "grad_norm": 0.39311644434928894,
      "learning_rate": 2e-05,
      "loss": 0.8016,
      "mean_token_accuracy": 0.7572365298867225,
      "step": 1920
    },
    {
      "epoch": 1.8534617700180616,
      "grad_norm": 0.4057285487651825,
      "learning_rate": 2e-05,
      "loss": 0.8245,
      "mean_token_accuracy": 0.7479765564203262,
      "step": 1925
    },
    {
      "epoch": 1.858278145695364,
      "grad_norm": 0.35648080706596375,
      "learning_rate": 2e-05,
      "loss": 0.828,
      "mean_token_accuracy": 0.7472658976912498,
      "step": 1930
    },
    {
      "epoch": 1.863094521372667,
      "grad_norm": 0.4065404236316681,
      "learning_rate": 2e-05,
      "loss": 0.8149,
      "mean_token_accuracy": 0.7524289473891258,
      "step": 1935
    },
    {
      "epoch": 1.86791089704997,
      "grad_norm": 0.3594423234462738,
      "learning_rate": 2e-05,
      "loss": 0.8431,
      "mean_token_accuracy": 0.7439246565103531,
      "step": 1940
    },
    {
      "epoch": 1.8727272727272726,
      "grad_norm": 0.34858575463294983,
      "learning_rate": 2e-05,
      "loss": 0.833,
      "mean_token_accuracy": 0.749679408967495,
      "step": 1945
    },
    {
      "epoch": 1.8775436484045755,
      "grad_norm": 0.37988269329071045,
      "learning_rate": 2e-05,
      "loss": 0.7625,
      "mean_token_accuracy": 0.7687516495585441,
      "step": 1950
    },
    {
      "epoch": 1.8823600240818785,
      "grad_norm": 0.36146003007888794,
      "learning_rate": 2e-05,
      "loss": 0.8401,
      "mean_token_accuracy": 0.743234421312809,
      "step": 1955
    },
    {
      "epoch": 1.887176399759181,
      "grad_norm": 0.3551631569862366,
      "learning_rate": 2e-05,
      "loss": 0.8378,
      "mean_token_accuracy": 0.7408182680606842,
      "step": 1960
    },
    {
      "epoch": 1.891992775436484,
      "grad_norm": 0.3982977569103241,
      "learning_rate": 2e-05,
      "loss": 0.8227,
      "mean_token_accuracy": 0.751612164080143,
      "step": 1965
    },
    {
      "epoch": 1.896809151113787,
      "grad_norm": 0.3863884508609772,
      "learning_rate": 2e-05,
      "loss": 0.8463,
      "mean_token_accuracy": 0.7427176803350448,
      "step": 1970
    },
    {
      "epoch": 1.9016255267910898,
      "grad_norm": 0.3617970645427704,
      "learning_rate": 2e-05,
      "loss": 0.8255,
      "mean_token_accuracy": 0.7477703839540482,
      "step": 1975
    },
    {
      "epoch": 1.9064419024683925,
      "grad_norm": 0.41495469212532043,
      "learning_rate": 2e-05,
      "loss": 0.8384,
      "mean_token_accuracy": 0.7448007896542549,
      "step": 1980
    },
    {
      "epoch": 1.9112582781456955,
      "grad_norm": 0.3848714828491211,
      "learning_rate": 2e-05,
      "loss": 0.8407,
      "mean_token_accuracy": 0.7430221542716027,
      "step": 1985
    },
    {
      "epoch": 1.9160746538229982,
      "grad_norm": 0.36609163880348206,
      "learning_rate": 2e-05,
      "loss": 0.8299,
      "mean_token_accuracy": 0.7459353134036064,
      "step": 1990
    },
    {
      "epoch": 1.920891029500301,
      "grad_norm": 0.3701276183128357,
      "learning_rate": 2e-05,
      "loss": 0.8325,
      "mean_token_accuracy": 0.7488315016031265,
      "step": 1995
    },
    {
      "epoch": 1.925707405177604,
      "grad_norm": 0.3757995367050171,
      "learning_rate": 2e-05,
      "loss": 0.8018,
      "mean_token_accuracy": 0.7553416565060616,
      "step": 2000
    },
    {
      "epoch": 1.925707405177604,
      "eval_loss": 0.9051562547683716,
      "eval_mean_token_accuracy": 0.7356420801236079,
      "eval_runtime": 23.1611,
      "eval_samples_per_second": 4.318,
      "eval_steps_per_second": 0.561,
      "step": 2000
    },
    {
      "epoch": 1.9305237808549067,
      "grad_norm": 0.36982470750808716,
      "learning_rate": 2e-05,
      "loss": 0.8334,
      "mean_token_accuracy": 0.7408973425626755,
      "step": 2005
    },
    {
      "epoch": 1.9353401565322095,
      "grad_norm": 0.40300092101097107,
      "learning_rate": 2e-05,
      "loss": 0.8294,
      "mean_token_accuracy": 0.748587854206562,
      "step": 2010
    },
    {
      "epoch": 1.9401565322095125,
      "grad_norm": 0.3860663175582886,
      "learning_rate": 2e-05,
      "loss": 0.7914,
      "mean_token_accuracy": 0.7596546769142151,
      "step": 2015
    },
    {
      "epoch": 1.9449729078868152,
      "grad_norm": 0.4126531481742859,
      "learning_rate": 2e-05,
      "loss": 0.8144,
      "mean_token_accuracy": 0.7499192729592323,
      "step": 2020
    },
    {
      "epoch": 1.949789283564118,
      "grad_norm": 0.36783406138420105,
      "learning_rate": 2e-05,
      "loss": 0.8136,
      "mean_token_accuracy": 0.7480936273932457,
      "step": 2025
    },
    {
      "epoch": 1.954605659241421,
      "grad_norm": 0.3961862623691559,
      "learning_rate": 2e-05,
      "loss": 0.8175,
      "mean_token_accuracy": 0.7519208177924156,
      "step": 2030
    },
    {
      "epoch": 1.9594220349187237,
      "grad_norm": 0.39503687620162964,
      "learning_rate": 2e-05,
      "loss": 0.7986,
      "mean_token_accuracy": 0.7552447572350502,
      "step": 2035
    },
    {
      "epoch": 1.9642384105960264,
      "grad_norm": 0.383649080991745,
      "learning_rate": 2e-05,
      "loss": 0.816,
      "mean_token_accuracy": 0.7495980471372604,
      "step": 2040
    },
    {
      "epoch": 1.9690547862733294,
      "grad_norm": 0.36074724793434143,
      "learning_rate": 2e-05,
      "loss": 0.8157,
      "mean_token_accuracy": 0.7535919129848481,
      "step": 2045
    },
    {
      "epoch": 1.9738711619506322,
      "grad_norm": 0.38681527972221375,
      "learning_rate": 2e-05,
      "loss": 0.8404,
      "mean_token_accuracy": 0.7456748083233833,
      "step": 2050
    },
    {
      "epoch": 1.978687537627935,
      "grad_norm": 0.36571046710014343,
      "learning_rate": 2e-05,
      "loss": 0.8167,
      "mean_token_accuracy": 0.7556032404303551,
      "step": 2055
    },
    {
      "epoch": 1.983503913305238,
      "grad_norm": 0.43187224864959717,
      "learning_rate": 2e-05,
      "loss": 0.8022,
      "mean_token_accuracy": 0.7503258436918259,
      "step": 2060
    },
    {
      "epoch": 1.9883202889825407,
      "grad_norm": 0.3371817469596863,
      "learning_rate": 2e-05,
      "loss": 0.8379,
      "mean_token_accuracy": 0.7471990913152695,
      "step": 2065
    },
    {
      "epoch": 1.9931366646598434,
      "grad_norm": 0.3704403042793274,
      "learning_rate": 2e-05,
      "loss": 0.8468,
      "mean_token_accuracy": 0.7404090151190758,
      "step": 2070
    },
    {
      "epoch": 1.9979530403371464,
      "grad_norm": 0.37344101071357727,
      "learning_rate": 2e-05,
      "loss": 0.8149,
      "mean_token_accuracy": 0.7522780746221542,
      "step": 2075
    },
    {
      "epoch": 2.001926550270921,
      "grad_norm": 0.4436754882335663,
      "learning_rate": 2e-05,
      "loss": 0.6846,
      "mean_token_accuracy": 0.7422799016490127,
      "step": 2080
    },
    {
      "epoch": 2.006742925948224,
      "grad_norm": 0.3635530471801758,
      "learning_rate": 2e-05,
      "loss": 0.7807,
      "mean_token_accuracy": 0.7589672565460205,
      "step": 2085
    },
    {
      "epoch": 2.011559301625527,
      "grad_norm": 0.3698291778564453,
      "learning_rate": 2e-05,
      "loss": 0.778,
      "mean_token_accuracy": 0.7537376001477242,
      "step": 2090
    },
    {
      "epoch": 2.0163756773028294,
      "grad_norm": 0.3593323230743408,
      "learning_rate": 2e-05,
      "loss": 0.7656,
      "mean_token_accuracy": 0.7640043929219246,
      "step": 2095
    },
    {
      "epoch": 2.0211920529801324,
      "grad_norm": 0.38869941234588623,
      "learning_rate": 2e-05,
      "loss": 0.7958,
      "mean_token_accuracy": 0.7520812705159188,
      "step": 2100
    },
    {
      "epoch": 2.0211920529801324,
      "eval_loss": 0.9059374928474426,
      "eval_mean_token_accuracy": 0.7361722084192129,
      "eval_runtime": 23.1994,
      "eval_samples_per_second": 4.31,
      "eval_steps_per_second": 0.56,
      "step": 2100
    },
    {
      "epoch": 2.0260084286574354,
      "grad_norm": 0.324433833360672,
      "learning_rate": 2e-05,
      "loss": 0.7856,
      "mean_token_accuracy": 0.7576806485652924,
      "step": 2105
    },
    {
      "epoch": 2.0308248043347383,
      "grad_norm": 0.35579246282577515,
      "learning_rate": 2e-05,
      "loss": 0.7895,
      "mean_token_accuracy": 0.7542310237884522,
      "step": 2110
    },
    {
      "epoch": 2.035641180012041,
      "grad_norm": 0.4416930377483368,
      "learning_rate": 2e-05,
      "loss": 0.7471,
      "mean_token_accuracy": 0.7662236809730529,
      "step": 2115
    },
    {
      "epoch": 2.040457555689344,
      "grad_norm": 0.3690590262413025,
      "learning_rate": 2e-05,
      "loss": 0.7557,
      "mean_token_accuracy": 0.7577535837888718,
      "step": 2120
    },
    {
      "epoch": 2.045273931366647,
      "grad_norm": 0.3864422142505646,
      "learning_rate": 2e-05,
      "loss": 0.7441,
      "mean_token_accuracy": 0.7661188945174218,
      "step": 2125
    },
    {
      "epoch": 2.0500903070439493,
      "grad_norm": 0.3759057819843292,
      "learning_rate": 2e-05,
      "loss": 0.7791,
      "mean_token_accuracy": 0.7549151331186295,
      "step": 2130
    },
    {
      "epoch": 2.0549066827212523,
      "grad_norm": 0.35572880506515503,
      "learning_rate": 2e-05,
      "loss": 0.795,
      "mean_token_accuracy": 0.7541061162948608,
      "step": 2135
    },
    {
      "epoch": 2.0597230583985553,
      "grad_norm": 0.3675136864185333,
      "learning_rate": 2e-05,
      "loss": 0.7965,
      "mean_token_accuracy": 0.7506997480988502,
      "step": 2140
    },
    {
      "epoch": 2.064539434075858,
      "grad_norm": 0.3544418215751648,
      "learning_rate": 2e-05,
      "loss": 0.8036,
      "mean_token_accuracy": 0.7521514996886254,
      "step": 2145
    },
    {
      "epoch": 2.069355809753161,
      "grad_norm": 0.3543994426727295,
      "learning_rate": 2e-05,
      "loss": 0.7788,
      "mean_token_accuracy": 0.7620324581861496,
      "step": 2150
    },
    {
      "epoch": 2.0741721854304638,
      "grad_norm": 0.35432854294776917,
      "learning_rate": 2e-05,
      "loss": 0.774,
      "mean_token_accuracy": 0.7557155057787895,
      "step": 2155
    },
    {
      "epoch": 2.0789885611077663,
      "grad_norm": 0.30006834864616394,
      "learning_rate": 2e-05,
      "loss": 0.7463,
      "mean_token_accuracy": 0.7641932889819145,
      "step": 2160
    },
    {
      "epoch": 2.0838049367850693,
      "grad_norm": 0.34351858496665955,
      "learning_rate": 2e-05,
      "loss": 0.7893,
      "mean_token_accuracy": 0.7499398306012154,
      "step": 2165
    },
    {
      "epoch": 2.0886213124623723,
      "grad_norm": 0.40961310267448425,
      "learning_rate": 2e-05,
      "loss": 0.7827,
      "mean_token_accuracy": 0.7576932951807975,
      "step": 2170
    },
    {
      "epoch": 2.093437688139675,
      "grad_norm": 0.35560116171836853,
      "learning_rate": 2e-05,
      "loss": 0.7698,
      "mean_token_accuracy": 0.7573257461190224,
      "step": 2175
    },
    {
      "epoch": 2.0982540638169778,
      "grad_norm": 0.35371777415275574,
      "learning_rate": 2e-05,
      "loss": 0.7816,
      "mean_token_accuracy": 0.7532159358263015,
      "step": 2180
    },
    {
      "epoch": 2.1030704394942807,
      "grad_norm": 0.3600073456764221,
      "learning_rate": 2e-05,
      "loss": 0.7881,
      "mean_token_accuracy": 0.7561186358332634,
      "step": 2185
    },
    {
      "epoch": 2.1078868151715833,
      "grad_norm": 0.37766149640083313,
      "learning_rate": 2e-05,
      "loss": 0.7797,
      "mean_token_accuracy": 0.7565206706523895,
      "step": 2190
    },
    {
      "epoch": 2.1127031908488862,
      "grad_norm": 0.35926687717437744,
      "learning_rate": 2e-05,
      "loss": 0.8197,
      "mean_token_accuracy": 0.7425670564174652,
      "step": 2195
    },
    {
      "epoch": 2.1175195665261892,
      "grad_norm": 0.37030982971191406,
      "learning_rate": 2e-05,
      "loss": 0.7642,
      "mean_token_accuracy": 0.7601809903979302,
      "step": 2200
    },
    {
      "epoch": 2.1175195665261892,
      "eval_loss": 0.9032812714576721,
      "eval_mean_token_accuracy": 0.7368296614060035,
      "eval_runtime": 23.1974,
      "eval_samples_per_second": 4.311,
      "eval_steps_per_second": 0.56,
      "step": 2200
    },
    {
      "epoch": 2.1223359422034918,
      "grad_norm": 0.37453070282936096,
      "learning_rate": 2e-05,
      "loss": 0.714,
      "mean_token_accuracy": 0.777516531944275,
      "step": 2205
    },
    {
      "epoch": 2.1271523178807947,
      "grad_norm": 0.39775583148002625,
      "learning_rate": 2e-05,
      "loss": 0.7447,
      "mean_token_accuracy": 0.7646070823073388,
      "step": 2210
    },
    {
      "epoch": 2.1319686935580977,
      "grad_norm": 0.35055381059646606,
      "learning_rate": 2e-05,
      "loss": 0.7748,
      "mean_token_accuracy": 0.7598061427474022,
      "step": 2215
    },
    {
      "epoch": 2.1367850692354002,
      "grad_norm": 0.3668005168437958,
      "learning_rate": 2e-05,
      "loss": 0.7791,
      "mean_token_accuracy": 0.7541374906897544,
      "step": 2220
    },
    {
      "epoch": 2.141601444912703,
      "grad_norm": 0.3544014096260071,
      "learning_rate": 2e-05,
      "loss": 0.7835,
      "mean_token_accuracy": 0.7573020115494729,
      "step": 2225
    },
    {
      "epoch": 2.146417820590006,
      "grad_norm": 0.3782559037208557,
      "learning_rate": 2e-05,
      "loss": 0.7682,
      "mean_token_accuracy": 0.7608320832252502,
      "step": 2230
    },
    {
      "epoch": 2.1512341962673087,
      "grad_norm": 0.3616378605365753,
      "learning_rate": 2e-05,
      "loss": 0.7991,
      "mean_token_accuracy": 0.7570154413580894,
      "step": 2235
    },
    {
      "epoch": 2.1560505719446117,
      "grad_norm": 0.3906887471675873,
      "learning_rate": 2e-05,
      "loss": 0.7757,
      "mean_token_accuracy": 0.7537379145622254,
      "step": 2240
    },
    {
      "epoch": 2.1608669476219147,
      "grad_norm": 0.39540261030197144,
      "learning_rate": 2e-05,
      "loss": 0.7531,
      "mean_token_accuracy": 0.765704445540905,
      "step": 2245
    },
    {
      "epoch": 2.165683323299217,
      "grad_norm": 0.3733433187007904,
      "learning_rate": 2e-05,
      "loss": 0.7745,
      "mean_token_accuracy": 0.7563082128763199,
      "step": 2250
    },
    {
      "epoch": 2.17049969897652,
      "grad_norm": 0.3424307107925415,
      "learning_rate": 2e-05,
      "loss": 0.7655,
      "mean_token_accuracy": 0.7576767817139626,
      "step": 2255
    },
    {
      "epoch": 2.175316074653823,
      "grad_norm": 0.4548731744289398,
      "learning_rate": 2e-05,
      "loss": 0.7269,
      "mean_token_accuracy": 0.7734593346714973,
      "step": 2260
    },
    {
      "epoch": 2.1801324503311257,
      "grad_norm": 0.37536028027534485,
      "learning_rate": 2e-05,
      "loss": 0.7452,
      "mean_token_accuracy": 0.7681032359600067,
      "step": 2265
    },
    {
      "epoch": 2.1849488260084287,
      "grad_norm": 0.3755234479904175,
      "learning_rate": 2e-05,
      "loss": 0.7654,
      "mean_token_accuracy": 0.7575892135500908,
      "step": 2270
    },
    {
      "epoch": 2.1897652016857316,
      "grad_norm": 0.38675227761268616,
      "learning_rate": 2e-05,
      "loss": 0.7479,
      "mean_token_accuracy": 0.7688836097717285,
      "step": 2275
    },
    {
      "epoch": 2.194581577363034,
      "grad_norm": 0.3535851240158081,
      "learning_rate": 2e-05,
      "loss": 0.7733,
      "mean_token_accuracy": 0.7607254981994629,
      "step": 2280
    },
    {
      "epoch": 2.199397953040337,
      "grad_norm": 0.39500853419303894,
      "learning_rate": 2e-05,
      "loss": 0.7917,
      "mean_token_accuracy": 0.7558043077588081,
      "step": 2285
    },
    {
      "epoch": 2.20421432871764,
      "grad_norm": 0.3839825987815857,
      "learning_rate": 2e-05,
      "loss": 0.7498,
      "mean_token_accuracy": 0.7639773234724998,
      "step": 2290
    },
    {
      "epoch": 2.2090307043949426,
      "grad_norm": 0.36488768458366394,
      "learning_rate": 2e-05,
      "loss": 0.7335,
      "mean_token_accuracy": 0.7700603663921356,
      "step": 2295
    },
    {
      "epoch": 2.2138470800722456,
      "grad_norm": 0.3318578898906708,
      "learning_rate": 2e-05,
      "loss": 0.784,
      "mean_token_accuracy": 0.7567041218280792,
      "step": 2300
    },
    {
      "epoch": 2.2138470800722456,
      "eval_loss": 0.9012500047683716,
      "eval_mean_token_accuracy": 0.7370858192443848,
      "eval_runtime": 23.18,
      "eval_samples_per_second": 4.314,
      "eval_steps_per_second": 0.561,
      "step": 2300
    },
    {
      "epoch": 2.2186634557495486,
      "grad_norm": 0.4050207734107971,
      "learning_rate": 2e-05,
      "loss": 0.7634,
      "mean_token_accuracy": 0.7616067379713058,
      "step": 2305
    },
    {
      "epoch": 2.223479831426851,
      "grad_norm": 0.3558892011642456,
      "learning_rate": 2e-05,
      "loss": 0.7733,
      "mean_token_accuracy": 0.7565241128206253,
      "step": 2310
    },
    {
      "epoch": 2.228296207104154,
      "grad_norm": 0.39691296219825745,
      "learning_rate": 2e-05,
      "loss": 0.7773,
      "mean_token_accuracy": 0.755375012755394,
      "step": 2315
    },
    {
      "epoch": 2.233112582781457,
      "grad_norm": 0.39223188161849976,
      "learning_rate": 2e-05,
      "loss": 0.7261,
      "mean_token_accuracy": 0.77184529453516,
      "step": 2320
    },
    {
      "epoch": 2.2379289584587596,
      "grad_norm": 0.3514679968357086,
      "learning_rate": 2e-05,
      "loss": 0.7958,
      "mean_token_accuracy": 0.7533982694149017,
      "step": 2325
    },
    {
      "epoch": 2.2427453341360626,
      "grad_norm": 0.3668522536754608,
      "learning_rate": 2e-05,
      "loss": 0.7359,
      "mean_token_accuracy": 0.7672333508729935,
      "step": 2330
    },
    {
      "epoch": 2.2475617098133656,
      "grad_norm": 0.36801326274871826,
      "learning_rate": 2e-05,
      "loss": 0.7852,
      "mean_token_accuracy": 0.7535142377018929,
      "step": 2335
    },
    {
      "epoch": 2.252378085490668,
      "grad_norm": 0.39534881711006165,
      "learning_rate": 2e-05,
      "loss": 0.7792,
      "mean_token_accuracy": 0.7582185581326485,
      "step": 2340
    },
    {
      "epoch": 2.257194461167971,
      "grad_norm": 0.40263575315475464,
      "learning_rate": 2e-05,
      "loss": 0.7334,
      "mean_token_accuracy": 0.773960568010807,
      "step": 2345
    },
    {
      "epoch": 2.262010836845274,
      "grad_norm": 0.3632088005542755,
      "learning_rate": 2e-05,
      "loss": 0.7542,
      "mean_token_accuracy": 0.7653770089149475,
      "step": 2350
    },
    {
      "epoch": 2.2668272125225766,
      "grad_norm": 0.37849465012550354,
      "learning_rate": 2e-05,
      "loss": 0.7661,
      "mean_token_accuracy": 0.7620905011892318,
      "step": 2355
    },
    {
      "epoch": 2.2716435881998795,
      "grad_norm": 0.43345385789871216,
      "learning_rate": 2e-05,
      "loss": 0.786,
      "mean_token_accuracy": 0.7564262568950653,
      "step": 2360
    },
    {
      "epoch": 2.2764599638771825,
      "grad_norm": 0.3616976737976074,
      "learning_rate": 2e-05,
      "loss": 0.7895,
      "mean_token_accuracy": 0.756988562643528,
      "step": 2365
    },
    {
      "epoch": 2.281276339554485,
      "grad_norm": 0.36480623483657837,
      "learning_rate": 2e-05,
      "loss": 0.7702,
      "mean_token_accuracy": 0.7585639297962189,
      "step": 2370
    },
    {
      "epoch": 2.286092715231788,
      "grad_norm": 0.36813783645629883,
      "learning_rate": 2e-05,
      "loss": 0.7542,
      "mean_token_accuracy": 0.7651529818773269,
      "step": 2375
    },
    {
      "epoch": 2.290909090909091,
      "grad_norm": 0.3709932565689087,
      "learning_rate": 2e-05,
      "loss": 0.7678,
      "mean_token_accuracy": 0.7603297665715217,
      "step": 2380
    },
    {
      "epoch": 2.2957254665863935,
      "grad_norm": 0.3750036954879761,
      "learning_rate": 2e-05,
      "loss": 0.7467,
      "mean_token_accuracy": 0.7639903724193573,
      "step": 2385
    },
    {
      "epoch": 2.3005418422636965,
      "grad_norm": 0.4516039788722992,
      "learning_rate": 2e-05,
      "loss": 0.7618,
      "mean_token_accuracy": 0.7606613889336586,
      "step": 2390
    },
    {
      "epoch": 2.3053582179409995,
      "grad_norm": 0.4092274010181427,
      "learning_rate": 2e-05,
      "loss": 0.742,
      "mean_token_accuracy": 0.763524578511715,
      "step": 2395
    },
    {
      "epoch": 2.310174593618302,
      "grad_norm": 0.3847052752971649,
      "learning_rate": 2e-05,
      "loss": 0.7589,
      "mean_token_accuracy": 0.7599444389343262,
      "step": 2400
    },
    {
      "epoch": 2.310174593618302,
      "eval_loss": 0.8995312452316284,
      "eval_mean_token_accuracy": 0.7376233431009146,
      "eval_runtime": 23.063,
      "eval_samples_per_second": 4.336,
      "eval_steps_per_second": 0.564,
      "step": 2400
    },
    {
      "epoch": 2.314990969295605,
      "grad_norm": 0.39317360520362854,
      "learning_rate": 2e-05,
      "loss": 0.7473,
      "mean_token_accuracy": 0.7662416860461235,
      "step": 2405
    },
    {
      "epoch": 2.319807344972908,
      "grad_norm": 0.3822837471961975,
      "learning_rate": 2e-05,
      "loss": 0.7337,
      "mean_token_accuracy": 0.7616621315479278,
      "step": 2410
    },
    {
      "epoch": 2.3246237206502105,
      "grad_norm": 0.40895310044288635,
      "learning_rate": 2e-05,
      "loss": 0.7632,
      "mean_token_accuracy": 0.765174786746502,
      "step": 2415
    },
    {
      "epoch": 2.3294400963275135,
      "grad_norm": 0.3547898530960083,
      "learning_rate": 2e-05,
      "loss": 0.7343,
      "mean_token_accuracy": 0.767419508099556,
      "step": 2420
    },
    {
      "epoch": 2.3342564720048165,
      "grad_norm": 0.3836051821708679,
      "learning_rate": 2e-05,
      "loss": 0.769,
      "mean_token_accuracy": 0.7603886500000954,
      "step": 2425
    },
    {
      "epoch": 2.339072847682119,
      "grad_norm": 0.374479204416275,
      "learning_rate": 2e-05,
      "loss": 0.7685,
      "mean_token_accuracy": 0.7552028700709343,
      "step": 2430
    },
    {
      "epoch": 2.343889223359422,
      "grad_norm": 0.37519872188568115,
      "learning_rate": 2e-05,
      "loss": 0.738,
      "mean_token_accuracy": 0.7714676797389984,
      "step": 2435
    },
    {
      "epoch": 2.348705599036725,
      "grad_norm": 0.41698017716407776,
      "learning_rate": 2e-05,
      "loss": 0.7647,
      "mean_token_accuracy": 0.7613590836524964,
      "step": 2440
    },
    {
      "epoch": 2.3535219747140275,
      "grad_norm": 0.3842358887195587,
      "learning_rate": 2e-05,
      "loss": 0.7809,
      "mean_token_accuracy": 0.7568897753953934,
      "step": 2445
    },
    {
      "epoch": 2.3583383503913304,
      "grad_norm": 0.42433464527130127,
      "learning_rate": 2e-05,
      "loss": 0.726,
      "mean_token_accuracy": 0.7667565375566483,
      "step": 2450
    },
    {
      "epoch": 2.3631547260686334,
      "grad_norm": 0.3653852641582489,
      "learning_rate": 2e-05,
      "loss": 0.7759,
      "mean_token_accuracy": 0.7587065488100052,
      "step": 2455
    },
    {
      "epoch": 2.367971101745936,
      "grad_norm": 0.4034559726715088,
      "learning_rate": 2e-05,
      "loss": 0.756,
      "mean_token_accuracy": 0.7609921962022781,
      "step": 2460
    },
    {
      "epoch": 2.372787477423239,
      "grad_norm": 0.39348727464675903,
      "learning_rate": 2e-05,
      "loss": 0.755,
      "mean_token_accuracy": 0.7636922210454941,
      "step": 2465
    },
    {
      "epoch": 2.377603853100542,
      "grad_norm": 0.4438548684120178,
      "learning_rate": 2e-05,
      "loss": 0.7267,
      "mean_token_accuracy": 0.7742465406656265,
      "step": 2470
    },
    {
      "epoch": 2.3824202287778444,
      "grad_norm": 0.455318808555603,
      "learning_rate": 2e-05,
      "loss": 0.7524,
      "mean_token_accuracy": 0.7645120665431022,
      "step": 2475
    },
    {
      "epoch": 2.3872366044551474,
      "grad_norm": 0.36987900733947754,
      "learning_rate": 2e-05,
      "loss": 0.749,
      "mean_token_accuracy": 0.7676509037613869,
      "step": 2480
    },
    {
      "epoch": 2.3920529801324504,
      "grad_norm": 0.38009896874427795,
      "learning_rate": 2e-05,
      "loss": 0.7927,
      "mean_token_accuracy": 0.7539709448814392,
      "step": 2485
    },
    {
      "epoch": 2.3968693558097534,
      "grad_norm": 0.3641670048236847,
      "learning_rate": 2e-05,
      "loss": 0.7412,
      "mean_token_accuracy": 0.7636485174298286,
      "step": 2490
    },
    {
      "epoch": 2.401685731487056,
      "grad_norm": 0.35153478384017944,
      "learning_rate": 2e-05,
      "loss": 0.7788,
      "mean_token_accuracy": 0.7512114852666855,
      "step": 2495
    },
    {
      "epoch": 2.406502107164359,
      "grad_norm": 0.3910485506057739,
      "learning_rate": 2e-05,
      "loss": 0.7558,
      "mean_token_accuracy": 0.7611356824636459,
      "step": 2500
    },
    {
      "epoch": 2.406502107164359,
      "eval_loss": 0.8964062333106995,
      "eval_mean_token_accuracy": 0.7384276848572952,
      "eval_runtime": 23.1541,
      "eval_samples_per_second": 4.319,
      "eval_steps_per_second": 0.561,
      "step": 2500
    },
    {
      "epoch": 2.411318482841662,
      "grad_norm": 0.35887807607650757,
      "learning_rate": 2e-05,
      "loss": 0.7755,
      "mean_token_accuracy": 0.7549531042575837,
      "step": 2505
    },
    {
      "epoch": 2.4161348585189644,
      "grad_norm": 0.39550381898880005,
      "learning_rate": 2e-05,
      "loss": 0.7881,
      "mean_token_accuracy": 0.7503395497798919,
      "step": 2510
    },
    {
      "epoch": 2.4209512341962673,
      "grad_norm": 0.4064529836177826,
      "learning_rate": 2e-05,
      "loss": 0.7608,
      "mean_token_accuracy": 0.7628593772649765,
      "step": 2515
    },
    {
      "epoch": 2.4257676098735703,
      "grad_norm": 0.3830781579017639,
      "learning_rate": 2e-05,
      "loss": 0.7672,
      "mean_token_accuracy": 0.7605699598789215,
      "step": 2520
    },
    {
      "epoch": 2.430583985550873,
      "grad_norm": 0.36953088641166687,
      "learning_rate": 2e-05,
      "loss": 0.7333,
      "mean_token_accuracy": 0.7733596116304398,
      "step": 2525
    },
    {
      "epoch": 2.435400361228176,
      "grad_norm": 0.42198288440704346,
      "learning_rate": 2e-05,
      "loss": 0.7165,
      "mean_token_accuracy": 0.7725221425294876,
      "step": 2530
    },
    {
      "epoch": 2.440216736905479,
      "grad_norm": 0.41048187017440796,
      "learning_rate": 2e-05,
      "loss": 0.7745,
      "mean_token_accuracy": 0.7596460074186325,
      "step": 2535
    },
    {
      "epoch": 2.4450331125827813,
      "grad_norm": 0.4293424189090729,
      "learning_rate": 2e-05,
      "loss": 0.7548,
      "mean_token_accuracy": 0.7634711265563965,
      "step": 2540
    },
    {
      "epoch": 2.4498494882600843,
      "grad_norm": 0.3512209355831146,
      "learning_rate": 2e-05,
      "loss": 0.7663,
      "mean_token_accuracy": 0.7589974060654641,
      "step": 2545
    },
    {
      "epoch": 2.4546658639373873,
      "grad_norm": 0.3962377607822418,
      "learning_rate": 2e-05,
      "loss": 0.7541,
      "mean_token_accuracy": 0.7592899754643441,
      "step": 2550
    },
    {
      "epoch": 2.45948223961469,
      "grad_norm": 0.3469253182411194,
      "learning_rate": 2e-05,
      "loss": 0.7522,
      "mean_token_accuracy": 0.7652729198336601,
      "step": 2555
    },
    {
      "epoch": 2.464298615291993,
      "grad_norm": 0.3742927312850952,
      "learning_rate": 2e-05,
      "loss": 0.7787,
      "mean_token_accuracy": 0.7575361758470536,
      "step": 2560
    },
    {
      "epoch": 2.4691149909692958,
      "grad_norm": 0.35377809405326843,
      "learning_rate": 2e-05,
      "loss": 0.8101,
      "mean_token_accuracy": 0.7458323091268539,
      "step": 2565
    },
    {
      "epoch": 2.4739313666465983,
      "grad_norm": 0.3490973711013794,
      "learning_rate": 2e-05,
      "loss": 0.7427,
      "mean_token_accuracy": 0.7698989689350129,
      "step": 2570
    },
    {
      "epoch": 2.4787477423239013,
      "grad_norm": 0.4195067584514618,
      "learning_rate": 2e-05,
      "loss": 0.775,
      "mean_token_accuracy": 0.760101044178009,
      "step": 2575
    },
    {
      "epoch": 2.4835641180012042,
      "grad_norm": 0.40223532915115356,
      "learning_rate": 2e-05,
      "loss": 0.7506,
      "mean_token_accuracy": 0.7645360633730889,
      "step": 2580
    },
    {
      "epoch": 2.4883804936785068,
      "grad_norm": 0.38401681184768677,
      "learning_rate": 2e-05,
      "loss": 0.7746,
      "mean_token_accuracy": 0.7553236082196235,
      "step": 2585
    },
    {
      "epoch": 2.4931968693558098,
      "grad_norm": 0.3794625997543335,
      "learning_rate": 2e-05,
      "loss": 0.7866,
      "mean_token_accuracy": 0.7584537342190742,
      "step": 2590
    },
    {
      "epoch": 2.4980132450331127,
      "grad_norm": 0.3422956168651581,
      "learning_rate": 2e-05,
      "loss": 0.7529,
      "mean_token_accuracy": 0.7641922473907471,
      "step": 2595
    },
    {
      "epoch": 2.5028296207104153,
      "grad_norm": 0.3222048580646515,
      "learning_rate": 2e-05,
      "loss": 0.7641,
      "mean_token_accuracy": 0.764282739162445,
      "step": 2600
    },
    {
      "epoch": 2.5028296207104153,
      "eval_loss": 0.8935937285423279,
      "eval_mean_token_accuracy": 0.738965378357814,
      "eval_runtime": 23.1976,
      "eval_samples_per_second": 4.311,
      "eval_steps_per_second": 0.56,
      "step": 2600
    },
    {
      "epoch": 2.5076459963877182,
      "grad_norm": 0.37517258524894714,
      "learning_rate": 2e-05,
      "loss": 0.7487,
      "mean_token_accuracy": 0.7686536237597466,
      "step": 2605
    },
    {
      "epoch": 2.512462372065021,
      "grad_norm": 0.39009490609169006,
      "learning_rate": 2e-05,
      "loss": 0.7905,
      "mean_token_accuracy": 0.7552632004022598,
      "step": 2610
    },
    {
      "epoch": 2.5172787477423237,
      "grad_norm": 0.3636697232723236,
      "learning_rate": 2e-05,
      "loss": 0.7804,
      "mean_token_accuracy": 0.757335577905178,
      "step": 2615
    },
    {
      "epoch": 2.5220951234196267,
      "grad_norm": 0.35198771953582764,
      "learning_rate": 2e-05,
      "loss": 0.74,
      "mean_token_accuracy": 0.7654551029205322,
      "step": 2620
    },
    {
      "epoch": 2.5269114990969297,
      "grad_norm": 0.37345898151397705,
      "learning_rate": 2e-05,
      "loss": 0.755,
      "mean_token_accuracy": 0.7649663195014,
      "step": 2625
    },
    {
      "epoch": 2.5317278747742327,
      "grad_norm": 0.34071609377861023,
      "learning_rate": 2e-05,
      "loss": 0.7371,
      "mean_token_accuracy": 0.7710829094052315,
      "step": 2630
    },
    {
      "epoch": 2.536544250451535,
      "grad_norm": 0.36361777782440186,
      "learning_rate": 2e-05,
      "loss": 0.7923,
      "mean_token_accuracy": 0.7498438969254494,
      "step": 2635
    },
    {
      "epoch": 2.541360626128838,
      "grad_norm": 0.3340684771537781,
      "learning_rate": 2e-05,
      "loss": 0.753,
      "mean_token_accuracy": 0.763702517747879,
      "step": 2640
    },
    {
      "epoch": 2.546177001806141,
      "grad_norm": 0.3246862590312958,
      "learning_rate": 2e-05,
      "loss": 0.7392,
      "mean_token_accuracy": 0.7666044220328331,
      "step": 2645
    },
    {
      "epoch": 2.5509933774834437,
      "grad_norm": 0.3731483221054077,
      "learning_rate": 2e-05,
      "loss": 0.8032,
      "mean_token_accuracy": 0.7451361432671547,
      "step": 2650
    },
    {
      "epoch": 2.5558097531607467,
      "grad_norm": 0.3968595266342163,
      "learning_rate": 2e-05,
      "loss": 0.7558,
      "mean_token_accuracy": 0.7619862273335457,
      "step": 2655
    },
    {
      "epoch": 2.5606261288380496,
      "grad_norm": 0.36753612756729126,
      "learning_rate": 2e-05,
      "loss": 0.7936,
      "mean_token_accuracy": 0.7513890892267228,
      "step": 2660
    },
    {
      "epoch": 2.565442504515352,
      "grad_norm": 0.38044974207878113,
      "learning_rate": 2e-05,
      "loss": 0.7652,
      "mean_token_accuracy": 0.7612641602754593,
      "step": 2665
    },
    {
      "epoch": 2.570258880192655,
      "grad_norm": 0.4354467988014221,
      "learning_rate": 2e-05,
      "loss": 0.7662,
      "mean_token_accuracy": 0.7602713435888291,
      "step": 2670
    },
    {
      "epoch": 2.575075255869958,
      "grad_norm": 0.34920626878738403,
      "learning_rate": 2e-05,
      "loss": 0.7802,
      "mean_token_accuracy": 0.7527212247252464,
      "step": 2675
    },
    {
      "epoch": 2.5798916315472606,
      "grad_norm": 0.3853444457054138,
      "learning_rate": 2e-05,
      "loss": 0.7929,
      "mean_token_accuracy": 0.7507719174027443,
      "step": 2680
    },
    {
      "epoch": 2.5847080072245636,
      "grad_norm": 0.37398529052734375,
      "learning_rate": 2e-05,
      "loss": 0.7335,
      "mean_token_accuracy": 0.7689384967088699,
      "step": 2685
    },
    {
      "epoch": 2.5895243829018666,
      "grad_norm": 0.37192922830581665,
      "learning_rate": 2e-05,
      "loss": 0.7588,
      "mean_token_accuracy": 0.7620333895087242,
      "step": 2690
    },
    {
      "epoch": 2.594340758579169,
      "grad_norm": 0.39118680357933044,
      "learning_rate": 2e-05,
      "loss": 0.7804,
      "mean_token_accuracy": 0.7577536061406136,
      "step": 2695
    },
    {
      "epoch": 2.599157134256472,
      "grad_norm": 0.3549438416957855,
      "learning_rate": 2e-05,
      "loss": 0.7601,
      "mean_token_accuracy": 0.7577766343951226,
      "step": 2700
    },
    {
      "epoch": 2.599157134256472,
      "eval_loss": 0.8917187452316284,
      "eval_mean_token_accuracy": 0.7396010435544528,
      "eval_runtime": 23.1125,
      "eval_samples_per_second": 4.327,
      "eval_steps_per_second": 0.562,
      "step": 2700
    },
    {
      "epoch": 2.603973509933775,
      "grad_norm": 0.3113292455673218,
      "learning_rate": 2e-05,
      "loss": 0.8017,
      "mean_token_accuracy": 0.7544750049710274,
      "step": 2705
    },
    {
      "epoch": 2.6087898856110776,
      "grad_norm": 0.3961426615715027,
      "learning_rate": 2e-05,
      "loss": 0.7835,
      "mean_token_accuracy": 0.757952244579792,
      "step": 2710
    },
    {
      "epoch": 2.6136062612883806,
      "grad_norm": 0.38531556725502014,
      "learning_rate": 2e-05,
      "loss": 0.7448,
      "mean_token_accuracy": 0.7656066238880157,
      "step": 2715
    },
    {
      "epoch": 2.6184226369656836,
      "grad_norm": 0.35683271288871765,
      "learning_rate": 2e-05,
      "loss": 0.764,
      "mean_token_accuracy": 0.7579592734575271,
      "step": 2720
    },
    {
      "epoch": 2.623239012642986,
      "grad_norm": 0.3948013484477997,
      "learning_rate": 2e-05,
      "loss": 0.7455,
      "mean_token_accuracy": 0.7686024904251099,
      "step": 2725
    },
    {
      "epoch": 2.628055388320289,
      "grad_norm": 0.4119192659854889,
      "learning_rate": 2e-05,
      "loss": 0.7726,
      "mean_token_accuracy": 0.7610606983304024,
      "step": 2730
    },
    {
      "epoch": 2.632871763997592,
      "grad_norm": 0.31112074851989746,
      "learning_rate": 2e-05,
      "loss": 0.7557,
      "mean_token_accuracy": 0.7639082312583924,
      "step": 2735
    },
    {
      "epoch": 2.6376881396748946,
      "grad_norm": 0.371297150850296,
      "learning_rate": 2e-05,
      "loss": 0.7477,
      "mean_token_accuracy": 0.76564369648695,
      "step": 2740
    },
    {
      "epoch": 2.6425045153521975,
      "grad_norm": 0.4188891351222992,
      "learning_rate": 2e-05,
      "loss": 0.7785,
      "mean_token_accuracy": 0.7570346057415008,
      "step": 2745
    },
    {
      "epoch": 2.6473208910295005,
      "grad_norm": 0.34406188130378723,
      "learning_rate": 2e-05,
      "loss": 0.7626,
      "mean_token_accuracy": 0.7595376938581466,
      "step": 2750
    },
    {
      "epoch": 2.652137266706803,
      "grad_norm": 0.3494187295436859,
      "learning_rate": 2e-05,
      "loss": 0.7572,
      "mean_token_accuracy": 0.7598767071962357,
      "step": 2755
    },
    {
      "epoch": 2.656953642384106,
      "grad_norm": 0.40831491351127625,
      "learning_rate": 2e-05,
      "loss": 0.7581,
      "mean_token_accuracy": 0.7597803384065628,
      "step": 2760
    },
    {
      "epoch": 2.661770018061409,
      "grad_norm": 0.44571653008461,
      "learning_rate": 2e-05,
      "loss": 0.7731,
      "mean_token_accuracy": 0.7607214003801346,
      "step": 2765
    },
    {
      "epoch": 2.6665863937387115,
      "grad_norm": 0.4143577814102173,
      "learning_rate": 2e-05,
      "loss": 0.7426,
      "mean_token_accuracy": 0.7718357488512992,
      "step": 2770
    },
    {
      "epoch": 2.6714027694160145,
      "grad_norm": 0.38475990295410156,
      "learning_rate": 2e-05,
      "loss": 0.7494,
      "mean_token_accuracy": 0.7667113184928894,
      "step": 2775
    },
    {
      "epoch": 2.6762191450933175,
      "grad_norm": 0.3438257575035095,
      "learning_rate": 2e-05,
      "loss": 0.7726,
      "mean_token_accuracy": 0.7581563532352448,
      "step": 2780
    },
    {
      "epoch": 2.68103552077062,
      "grad_norm": 0.3562197983264923,
      "learning_rate": 2e-05,
      "loss": 0.7758,
      "mean_token_accuracy": 0.7572029024362564,
      "step": 2785
    },
    {
      "epoch": 2.685851896447923,
      "grad_norm": 0.38419750332832336,
      "learning_rate": 2e-05,
      "loss": 0.7738,
      "mean_token_accuracy": 0.7595375701785088,
      "step": 2790
    },
    {
      "epoch": 2.690668272125226,
      "grad_norm": 0.3750392198562622,
      "learning_rate": 2e-05,
      "loss": 0.7642,
      "mean_token_accuracy": 0.7576078772544861,
      "step": 2795
    },
    {
      "epoch": 2.6954846478025285,
      "grad_norm": 0.4193848669528961,
      "learning_rate": 2e-05,
      "loss": 0.7517,
      "mean_token_accuracy": 0.7621890246868134,
      "step": 2800
    },
    {
      "epoch": 2.6954846478025285,
      "eval_loss": 0.8895312547683716,
      "eval_mean_token_accuracy": 0.7398839684633108,
      "eval_runtime": 23.179,
      "eval_samples_per_second": 4.314,
      "eval_steps_per_second": 0.561,
      "step": 2800
    },
    {
      "epoch": 2.7003010234798315,
      "grad_norm": 0.4397479295730591,
      "learning_rate": 2e-05,
      "loss": 0.7713,
      "mean_token_accuracy": 0.7680559456348419,
      "step": 2805
    },
    {
      "epoch": 2.7051173991571345,
      "grad_norm": 0.3621998727321625,
      "learning_rate": 2e-05,
      "loss": 0.7212,
      "mean_token_accuracy": 0.77404675334692,
      "step": 2810
    },
    {
      "epoch": 2.709933774834437,
      "grad_norm": 0.3363446891307831,
      "learning_rate": 2e-05,
      "loss": 0.7884,
      "mean_token_accuracy": 0.7583529412746429,
      "step": 2815
    },
    {
      "epoch": 2.71475015051174,
      "grad_norm": 0.36584141850471497,
      "learning_rate": 2e-05,
      "loss": 0.7697,
      "mean_token_accuracy": 0.756261146068573,
      "step": 2820
    },
    {
      "epoch": 2.719566526189043,
      "grad_norm": 0.46295201778411865,
      "learning_rate": 2e-05,
      "loss": 0.7241,
      "mean_token_accuracy": 0.7722676262259484,
      "step": 2825
    },
    {
      "epoch": 2.7243829018663455,
      "grad_norm": 0.3580615520477295,
      "learning_rate": 2e-05,
      "loss": 0.8185,
      "mean_token_accuracy": 0.7469978988170624,
      "step": 2830
    },
    {
      "epoch": 2.7291992775436484,
      "grad_norm": 0.3244786858558655,
      "learning_rate": 2e-05,
      "loss": 0.7581,
      "mean_token_accuracy": 0.7603826940059661,
      "step": 2835
    },
    {
      "epoch": 2.7340156532209514,
      "grad_norm": 0.39725518226623535,
      "learning_rate": 2e-05,
      "loss": 0.767,
      "mean_token_accuracy": 0.7605881199240685,
      "step": 2840
    },
    {
      "epoch": 2.738832028898254,
      "grad_norm": 0.40936702489852905,
      "learning_rate": 2e-05,
      "loss": 0.7682,
      "mean_token_accuracy": 0.7618032857775688,
      "step": 2845
    },
    {
      "epoch": 2.743648404575557,
      "grad_norm": 0.357181191444397,
      "learning_rate": 2e-05,
      "loss": 0.7275,
      "mean_token_accuracy": 0.7717493593692779,
      "step": 2850
    },
    {
      "epoch": 2.74846478025286,
      "grad_norm": 0.409920871257782,
      "learning_rate": 2e-05,
      "loss": 0.7478,
      "mean_token_accuracy": 0.7658723518252373,
      "step": 2855
    },
    {
      "epoch": 2.7532811559301624,
      "grad_norm": 0.4086902439594269,
      "learning_rate": 2e-05,
      "loss": 0.7401,
      "mean_token_accuracy": 0.7642821088433266,
      "step": 2860
    },
    {
      "epoch": 2.7580975316074654,
      "grad_norm": 0.3491520285606384,
      "learning_rate": 2e-05,
      "loss": 0.7711,
      "mean_token_accuracy": 0.7554255187511444,
      "step": 2865
    },
    {
      "epoch": 2.7629139072847684,
      "grad_norm": 0.41609054803848267,
      "learning_rate": 2e-05,
      "loss": 0.7615,
      "mean_token_accuracy": 0.7587140306830407,
      "step": 2870
    },
    {
      "epoch": 2.767730282962071,
      "grad_norm": 0.3808140754699707,
      "learning_rate": 2e-05,
      "loss": 0.8011,
      "mean_token_accuracy": 0.7536677673459053,
      "step": 2875
    },
    {
      "epoch": 2.772546658639374,
      "grad_norm": 0.37478163838386536,
      "learning_rate": 2e-05,
      "loss": 0.7766,
      "mean_token_accuracy": 0.7533249914646148,
      "step": 2880
    },
    {
      "epoch": 2.777363034316677,
      "grad_norm": 0.43278947472572327,
      "learning_rate": 2e-05,
      "loss": 0.7367,
      "mean_token_accuracy": 0.7670947164297104,
      "step": 2885
    },
    {
      "epoch": 2.7821794099939794,
      "grad_norm": 0.3967084288597107,
      "learning_rate": 2e-05,
      "loss": 0.7541,
      "mean_token_accuracy": 0.760562515258789,
      "step": 2890
    },
    {
      "epoch": 2.7869957856712824,
      "grad_norm": 0.4235425293445587,
      "learning_rate": 2e-05,
      "loss": 0.7525,
      "mean_token_accuracy": 0.7589347019791604,
      "step": 2895
    },
    {
      "epoch": 2.7918121613485853,
      "grad_norm": 0.4071233570575714,
      "learning_rate": 2e-05,
      "loss": 0.7481,
      "mean_token_accuracy": 0.7661397770047188,
      "step": 2900
    },
    {
      "epoch": 2.7918121613485853,
      "eval_loss": 0.8868749737739563,
      "eval_mean_token_accuracy": 0.7403522042127756,
      "eval_runtime": 23.1245,
      "eval_samples_per_second": 4.324,
      "eval_steps_per_second": 0.562,
      "step": 2900
    },
    {
      "epoch": 2.796628537025888,
      "grad_norm": 0.38018038868904114,
      "learning_rate": 2e-05,
      "loss": 0.75,
      "mean_token_accuracy": 0.7664709404110909,
      "step": 2905
    },
    {
      "epoch": 2.801444912703191,
      "grad_norm": 0.412870317697525,
      "learning_rate": 2e-05,
      "loss": 0.7623,
      "mean_token_accuracy": 0.7559232786297798,
      "step": 2910
    },
    {
      "epoch": 2.806261288380494,
      "grad_norm": 0.41684240102767944,
      "learning_rate": 2e-05,
      "loss": 0.7232,
      "mean_token_accuracy": 0.7719547718763351,
      "step": 2915
    },
    {
      "epoch": 2.8110776640577964,
      "grad_norm": 0.35981109738349915,
      "learning_rate": 2e-05,
      "loss": 0.7832,
      "mean_token_accuracy": 0.7560476765036583,
      "step": 2920
    },
    {
      "epoch": 2.8158940397350993,
      "grad_norm": 0.40023863315582275,
      "learning_rate": 2e-05,
      "loss": 0.7567,
      "mean_token_accuracy": 0.765048947930336,
      "step": 2925
    },
    {
      "epoch": 2.8207104154124023,
      "grad_norm": 0.39661338925361633,
      "learning_rate": 2e-05,
      "loss": 0.7377,
      "mean_token_accuracy": 0.7675866693258285,
      "step": 2930
    },
    {
      "epoch": 2.825526791089705,
      "grad_norm": 0.35321974754333496,
      "learning_rate": 2e-05,
      "loss": 0.8157,
      "mean_token_accuracy": 0.746397140622139,
      "step": 2935
    },
    {
      "epoch": 2.830343166767008,
      "grad_norm": 0.39979612827301025,
      "learning_rate": 2e-05,
      "loss": 0.7824,
      "mean_token_accuracy": 0.7587210416793824,
      "step": 2940
    },
    {
      "epoch": 2.835159542444311,
      "grad_norm": 0.3821732699871063,
      "learning_rate": 2e-05,
      "loss": 0.7895,
      "mean_token_accuracy": 0.7537524312734604,
      "step": 2945
    },
    {
      "epoch": 2.8399759181216133,
      "grad_norm": 0.3334910571575165,
      "learning_rate": 2e-05,
      "loss": 0.7674,
      "mean_token_accuracy": 0.7562859013676644,
      "step": 2950
    },
    {
      "epoch": 2.8447922937989163,
      "grad_norm": 0.37625160813331604,
      "learning_rate": 2e-05,
      "loss": 0.7397,
      "mean_token_accuracy": 0.7702266126871109,
      "step": 2955
    },
    {
      "epoch": 2.8496086694762193,
      "grad_norm": 0.38816380500793457,
      "learning_rate": 2e-05,
      "loss": 0.7463,
      "mean_token_accuracy": 0.767043624818325,
      "step": 2960
    },
    {
      "epoch": 2.854425045153522,
      "grad_norm": 0.38124793767929077,
      "learning_rate": 2e-05,
      "loss": 0.7188,
      "mean_token_accuracy": 0.7760316908359528,
      "step": 2965
    },
    {
      "epoch": 2.8592414208308248,
      "grad_norm": 0.3955833315849304,
      "learning_rate": 2e-05,
      "loss": 0.7405,
      "mean_token_accuracy": 0.7660493075847625,
      "step": 2970
    },
    {
      "epoch": 2.8640577965081278,
      "grad_norm": 0.36449986696243286,
      "learning_rate": 2e-05,
      "loss": 0.7639,
      "mean_token_accuracy": 0.7551109924912452,
      "step": 2975
    },
    {
      "epoch": 2.8688741721854303,
      "grad_norm": 0.3809254765510559,
      "learning_rate": 2e-05,
      "loss": 0.7378,
      "mean_token_accuracy": 0.7687376230955124,
      "step": 2980
    },
    {
      "epoch": 2.8736905478627333,
      "grad_norm": 0.347232460975647,
      "learning_rate": 2e-05,
      "loss": 0.717,
      "mean_token_accuracy": 0.7766376540064812,
      "step": 2985
    },
    {
      "epoch": 2.8785069235400362,
      "grad_norm": 0.35575151443481445,
      "learning_rate": 2e-05,
      "loss": 0.7485,
      "mean_token_accuracy": 0.7674190372228622,
      "step": 2990
    },
    {
      "epoch": 2.8833232992173388,
      "grad_norm": 0.3710139989852905,
      "learning_rate": 2e-05,
      "loss": 0.7296,
      "mean_token_accuracy": 0.7713616698980331,
      "step": 2995
    },
    {
      "epoch": 2.8881396748946417,
      "grad_norm": 0.37481653690338135,
      "learning_rate": 2e-05,
      "loss": 0.759,
      "mean_token_accuracy": 0.7613216653466225,
      "step": 3000
    },
    {
      "epoch": 2.8881396748946417,
      "eval_loss": 0.8843749761581421,
      "eval_mean_token_accuracy": 0.7408330532220694,
      "eval_runtime": 23.1361,
      "eval_samples_per_second": 4.322,
      "eval_steps_per_second": 0.562,
      "step": 3000
    },
    {
      "epoch": 2.8929560505719447,
      "grad_norm": 0.3814287781715393,
      "learning_rate": 2e-05,
      "loss": 0.7218,
      "mean_token_accuracy": 0.7693289056420326,
      "step": 3005
    },
    {
      "epoch": 2.8977724262492472,
      "grad_norm": 0.38517627120018005,
      "learning_rate": 2e-05,
      "loss": 0.7829,
      "mean_token_accuracy": 0.7556496039032936,
      "step": 3010
    },
    {
      "epoch": 2.90258880192655,
      "grad_norm": 0.3561043441295624,
      "learning_rate": 2e-05,
      "loss": 0.7406,
      "mean_token_accuracy": 0.7704349234700203,
      "step": 3015
    },
    {
      "epoch": 2.907405177603853,
      "grad_norm": 0.3632068932056427,
      "learning_rate": 2e-05,
      "loss": 0.7639,
      "mean_token_accuracy": 0.7629840105772019,
      "step": 3020
    },
    {
      "epoch": 2.9122215532811557,
      "grad_norm": 0.39246779680252075,
      "learning_rate": 2e-05,
      "loss": 0.7497,
      "mean_token_accuracy": 0.7630103766918183,
      "step": 3025
    },
    {
      "epoch": 2.9170379289584587,
      "grad_norm": 0.3949568569660187,
      "learning_rate": 2e-05,
      "loss": 0.7395,
      "mean_token_accuracy": 0.7638208046555519,
      "step": 3030
    },
    {
      "epoch": 2.9218543046357617,
      "grad_norm": 0.3802914619445801,
      "learning_rate": 2e-05,
      "loss": 0.7666,
      "mean_token_accuracy": 0.7584463760256768,
      "step": 3035
    },
    {
      "epoch": 2.926670680313064,
      "grad_norm": 0.3917528986930847,
      "learning_rate": 2e-05,
      "loss": 0.7556,
      "mean_token_accuracy": 0.7628441542387009,
      "step": 3040
    },
    {
      "epoch": 2.931487055990367,
      "grad_norm": 0.3582029640674591,
      "learning_rate": 2e-05,
      "loss": 0.7634,
      "mean_token_accuracy": 0.762310317158699,
      "step": 3045
    },
    {
      "epoch": 2.93630343166767,
      "grad_norm": 0.3874087929725647,
      "learning_rate": 2e-05,
      "loss": 0.7719,
      "mean_token_accuracy": 0.7597872078418731,
      "step": 3050
    },
    {
      "epoch": 2.9411198073449727,
      "grad_norm": 0.3633805215358734,
      "learning_rate": 2e-05,
      "loss": 0.761,
      "mean_token_accuracy": 0.7614316448569298,
      "step": 3055
    },
    {
      "epoch": 2.9459361830222757,
      "grad_norm": 0.4053983688354492,
      "learning_rate": 2e-05,
      "loss": 0.7396,
      "mean_token_accuracy": 0.7678497686982155,
      "step": 3060
    },
    {
      "epoch": 2.9507525586995786,
      "grad_norm": 0.37059614062309265,
      "learning_rate": 2e-05,
      "loss": 0.7771,
      "mean_token_accuracy": 0.756655989587307,
      "step": 3065
    },
    {
      "epoch": 2.955568934376881,
      "grad_norm": 0.3870275020599365,
      "learning_rate": 2e-05,
      "loss": 0.8507,
      "mean_token_accuracy": 0.7373094201087952,
      "step": 3070
    },
    {
      "epoch": 2.960385310054184,
      "grad_norm": 0.32037490606307983,
      "learning_rate": 2e-05,
      "loss": 0.7344,
      "mean_token_accuracy": 0.7677667886018753,
      "step": 3075
    },
    {
      "epoch": 2.965201685731487,
      "grad_norm": 0.39370909333229065,
      "learning_rate": 2e-05,
      "loss": 0.7814,
      "mean_token_accuracy": 0.7587102144956589,
      "step": 3080
    },
    {
      "epoch": 2.9700180614087897,
      "grad_norm": 0.39840105175971985,
      "learning_rate": 2e-05,
      "loss": 0.7745,
      "mean_token_accuracy": 0.7543177887797355,
      "step": 3085
    },
    {
      "epoch": 2.9748344370860926,
      "grad_norm": 0.40780118107795715,
      "learning_rate": 2e-05,
      "loss": 0.7405,
      "mean_token_accuracy": 0.7715254962444306,
      "step": 3090
    },
    {
      "epoch": 2.9796508127633956,
      "grad_norm": 0.41411417722702026,
      "learning_rate": 2e-05,
      "loss": 0.7468,
      "mean_token_accuracy": 0.7587345033884049,
      "step": 3095
    },
    {
      "epoch": 2.984467188440698,
      "grad_norm": 0.36079859733581543,
      "learning_rate": 2e-05,
      "loss": 0.7635,
      "mean_token_accuracy": 0.757730433344841,
      "step": 3100
    },
    {
      "epoch": 2.984467188440698,
      "eval_loss": 0.8817187547683716,
      "eval_mean_token_accuracy": 0.7412376220409687,
      "eval_runtime": 23.1884,
      "eval_samples_per_second": 4.313,
      "eval_steps_per_second": 0.561,
      "step": 3100
    },
    {
      "epoch": 2.989283564118001,
      "grad_norm": 0.33232828974723816,
      "learning_rate": 2e-05,
      "loss": 0.7523,
      "mean_token_accuracy": 0.7645888909697532,
      "step": 3105
    },
    {
      "epoch": 2.994099939795304,
      "grad_norm": 0.3581736981868744,
      "learning_rate": 2e-05,
      "loss": 0.7617,
      "mean_token_accuracy": 0.7637889131903648,
      "step": 3110
    },
    {
      "epoch": 2.9979530403371464,
      "step": 3114,
      "total_flos": 0.0,
      "train_loss": 0.0,
      "train_runtime": 1.7052,
      "train_samples_per_second": 29222.257,
      "train_steps_per_second": 1826.171
    }
  ],
  "logging_steps": 5,
  "max_steps": 3114,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}