{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 306,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "aux_loss_top_prob": 0.5098888501524925,
      "entropy": 2.268296539783478,
      "epoch": 0.006535947712418301,
      "grad_norm": 1.1260346174240112,
      "learning_rate": 0.0,
      "loss": 6.6642,
      "mean_token_accuracy": 0.43837922372227583,
      "num_tokens": 30447.0,
      "step": 1
    },
    {
      "aux_loss_top_prob": 0.47646450251340866,
      "entropy": 2.584270715713501,
      "epoch": 0.013071895424836602,
      "grad_norm": 1.053179144859314,
      "learning_rate": 3.125e-06,
      "loss": 6.7866,
      "mean_token_accuracy": 0.4098796494453456,
      "num_tokens": 61954.0,
      "step": 2
    },
    {
      "aux_loss_top_prob": 0.5146670043468475,
      "entropy": 2.184733599424362,
      "epoch": 0.0196078431372549,
      "grad_norm": 1.157400131225586,
      "learning_rate": 6.25e-06,
      "loss": 6.7344,
      "mean_token_accuracy": 0.43879726983462924,
      "num_tokens": 92852.0,
      "step": 3
    },
    {
      "aux_loss_top_prob": 0.5053977221250534,
      "entropy": 2.2370023131370544,
      "epoch": 0.026143790849673203,
      "grad_norm": 1.1414951086044312,
      "learning_rate": 9.375000000000001e-06,
      "loss": 6.7082,
      "mean_token_accuracy": 0.4386049506170011,
      "num_tokens": 123272.0,
      "step": 4
    },
    {
      "aux_loss_top_prob": 0.5299825966358185,
      "entropy": 2.0891593396663666,
      "epoch": 0.032679738562091505,
      "grad_norm": 1.3075721263885498,
      "learning_rate": 1.25e-05,
      "loss": 6.7046,
      "mean_token_accuracy": 0.45440419223955114,
      "num_tokens": 153951.0,
      "step": 5
    },
    {
      "aux_loss_top_prob": 0.522010013461113,
      "entropy": 2.1483720541000366,
      "epoch": 0.0392156862745098,
      "grad_norm": 1.2018605470657349,
      "learning_rate": 1.5625e-05,
      "loss": 6.6625,
      "mean_token_accuracy": 0.4476112144824538,
      "num_tokens": 184629.0,
      "step": 6
    },
    {
      "aux_loss_top_prob": 0.4765317738056183,
      "entropy": 2.6441078186035156,
      "epoch": 0.0457516339869281,
      "grad_norm": 1.1445356607437134,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 6.9352,
      "mean_token_accuracy": 0.4198999937236687,
      "num_tokens": 208283.0,
      "step": 7
    },
    {
      "aux_loss_top_prob": 0.48557985574007034,
      "entropy": 2.559255540370941,
      "epoch": 0.05228758169934641,
      "grad_norm": 1.190241813659668,
      "learning_rate": 2.1875e-05,
      "loss": 6.709,
      "mean_token_accuracy": 0.4327983355505834,
      "num_tokens": 238583.0,
      "step": 8
    },
    {
      "aux_loss_top_prob": 0.4938953071832657,
      "entropy": 2.400003135204315,
      "epoch": 0.058823529411764705,
      "grad_norm": 1.2969859838485718,
      "learning_rate": 2.5e-05,
      "loss": 6.6729,
      "mean_token_accuracy": 0.4337025999765826,
      "num_tokens": 268008.0,
      "step": 9
    },
    {
      "aux_loss_top_prob": 0.508846178650856,
      "entropy": 2.2802494168281555,
      "epoch": 0.06535947712418301,
      "grad_norm": 1.4242280721664429,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 6.5721,
      "mean_token_accuracy": 0.45110622557790414,
      "num_tokens": 298780.0,
      "step": 10
    },
    {
      "aux_loss_top_prob": 0.4732913449406624,
      "entropy": 2.682538628578186,
      "epoch": 0.0718954248366013,
      "grad_norm": 1.455923080444336,
      "learning_rate": 3.125e-05,
      "loss": 6.6557,
      "mean_token_accuracy": 0.42359932943177137,
      "num_tokens": 329738.0,
      "step": 11
    },
    {
      "aux_loss_top_prob": 0.4769788235425949,
      "entropy": 2.5329073071479797,
      "epoch": 0.0784313725490196,
      "grad_norm": 1.8878237009048462,
      "learning_rate": 3.4375e-05,
      "loss": 6.5043,
      "mean_token_accuracy": 0.4342515841623504,
      "num_tokens": 360593.0,
      "step": 12
    },
    {
      "aux_loss_top_prob": 0.4772041365504265,
      "entropy": 2.6216312646865845,
      "epoch": 0.08496732026143791,
      "grad_norm": 1.6363930702209473,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 6.4423,
      "mean_token_accuracy": 0.4506732914755866,
      "num_tokens": 390112.0,
      "step": 13
    },
    {
      "aux_loss_top_prob": 0.46650662273168564,
      "entropy": 2.6697844862937927,
      "epoch": 0.0915032679738562,
      "grad_norm": 1.7384305000305176,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 6.292,
      "mean_token_accuracy": 0.4520756506962778,
      "num_tokens": 419497.0,
      "step": 14
    },
    {
      "aux_loss_top_prob": 0.4146432429552078,
      "entropy": 3.192969799041748,
      "epoch": 0.09803921568627451,
      "grad_norm": 1.6363615989685059,
      "learning_rate": 4.375e-05,
      "loss": 6.2355,
      "mean_token_accuracy": 0.4265264154710019,
      "num_tokens": 449296.0,
      "step": 15
    },
    {
      "aux_loss_top_prob": 0.4288252592086792,
      "entropy": 3.1263471245765686,
      "epoch": 0.10457516339869281,
      "grad_norm": 1.705688238143921,
      "learning_rate": 4.6875e-05,
      "loss": 6.1134,
      "mean_token_accuracy": 0.4479966322078331,
      "num_tokens": 480479.0,
      "step": 16
    },
    {
      "aux_loss_top_prob": 0.3538140133023262,
      "entropy": 4.253853261470795,
      "epoch": 0.1111111111111111,
      "grad_norm": 1.205020785331726,
      "learning_rate": 5e-05,
      "loss": 5.7569,
      "mean_token_accuracy": 0.33106382205195656,
      "num_tokens": 504093.0,
      "step": 17
    },
    {
      "aux_loss_top_prob": 0.31766898185014725,
      "entropy": 4.611821413040161,
      "epoch": 0.11764705882352941,
      "grad_norm": 0.9569157361984253,
      "learning_rate": 4.999853306957783e-05,
      "loss": 6.0815,
      "mean_token_accuracy": 0.4111181676203216,
      "num_tokens": 535198.0,
      "step": 18
    },
    {
      "aux_loss_top_prob": 0.31263647601008415,
      "entropy": 4.805905342102051,
      "epoch": 0.12418300653594772,
      "grad_norm": 0.9731336236000061,
      "learning_rate": 4.999413245046211e-05,
      "loss": 5.9268,
      "mean_token_accuracy": 0.44223002079237905,
      "num_tokens": 566576.0,
      "step": 19
    },
    {
      "aux_loss_top_prob": 0.29073501005768776,
      "entropy": 5.195206642150879,
      "epoch": 0.13071895424836602,
      "grad_norm": 1.000673532485962,
      "learning_rate": 4.998679865908499e-05,
      "loss": 5.755,
      "mean_token_accuracy": 0.45028336936768665,
      "num_tokens": 596760.0,
      "step": 20
    },
    {
      "aux_loss_top_prob": 0.24318717420101166,
      "entropy": 5.862072229385376,
      "epoch": 0.13725490196078433,
      "grad_norm": 1.4611996412277222,
      "learning_rate": 4.997653255609942e-05,
      "loss": 5.7117,
      "mean_token_accuracy": 0.4442213915182356,
      "num_tokens": 626761.0,
      "step": 21
    },
    {
      "aux_loss_top_prob": 0.2288288101553917,
      "entropy": 6.385829448699951,
      "epoch": 0.1437908496732026,
      "grad_norm": 1.9518073797225952,
      "learning_rate": 4.996333534627809e-05,
      "loss": 5.7352,
      "mean_token_accuracy": 0.45289150592256444,
      "num_tokens": 657285.0,
      "step": 22
    },
    {
      "aux_loss_top_prob": 0.16969504952430725,
      "entropy": 7.257818102836609,
      "epoch": 0.1503267973856209,
      "grad_norm": 2.939915418624878,
      "learning_rate": 4.994720857837211e-05,
      "loss": 6.2062,
      "mean_token_accuracy": 0.39054983038657454,
      "num_tokens": 681193.0,
      "step": 23
    },
    {
      "aux_loss_top_prob": 0.22713647410273552,
      "entropy": 6.502384185791016,
      "epoch": 0.1568627450980392,
      "grad_norm": 1.6569392681121826,
      "learning_rate": 4.9928154144929175e-05,
      "loss": 5.6526,
      "mean_token_accuracy": 0.45017438551163125,
      "num_tokens": 711571.0,
      "step": 24
    },
    {
      "aux_loss_top_prob": 0.2111387476325035,
      "entropy": 6.783881783485413,
      "epoch": 0.16339869281045752,
      "grad_norm": 1.6402127742767334,
      "learning_rate": 4.9906174282071535e-05,
      "loss": 5.7253,
      "mean_token_accuracy": 0.4364330989699245,
      "num_tokens": 741226.0,
      "step": 25
    },
    {
      "aux_loss_top_prob": 0.21285325288772583,
      "entropy": 6.667588591575623,
      "epoch": 0.16993464052287582,
      "grad_norm": 1.4411511421203613,
      "learning_rate": 4.988127156923355e-05,
      "loss": 5.6431,
      "mean_token_accuracy": 0.4318039306689131,
      "num_tokens": 771762.0,
      "step": 26
    },
    {
      "aux_loss_top_prob": 0.25626177340745926,
      "entropy": 6.218438148498535,
      "epoch": 0.17647058823529413,
      "grad_norm": 0.9314313530921936,
      "learning_rate": 4.985344892885899e-05,
      "loss": 5.5972,
      "mean_token_accuracy": 0.45378688977782466,
      "num_tokens": 802719.0,
      "step": 27
    },
    {
      "aux_loss_top_prob": 0.24144598841667175,
      "entropy": 6.265723466873169,
      "epoch": 0.1830065359477124,
      "grad_norm": 0.6681748032569885,
      "learning_rate": 4.982270962605806e-05,
      "loss": 5.6813,
      "mean_token_accuracy": 0.4210802435213284,
      "num_tokens": 833050.0,
      "step": 28
    },
    {
      "aux_loss_top_prob": 0.23557967320084572,
      "entropy": 6.416491508483887,
      "epoch": 0.1895424836601307,
      "grad_norm": 0.6198770403862,
      "learning_rate": 4.9789057268224234e-05,
      "loss": 5.7495,
      "mean_token_accuracy": 0.4037180604231241,
      "num_tokens": 863458.0,
      "step": 29
    },
    {
      "aux_loss_top_prob": 0.26581963896751404,
      "entropy": 5.879812836647034,
      "epoch": 0.19607843137254902,
      "grad_norm": 0.4309234321117401,
      "learning_rate": 4.9752495804610916e-05,
      "loss": 5.5863,
      "mean_token_accuracy": 0.41638627953440094,
      "num_tokens": 893641.0,
      "step": 30
    },
    {
      "aux_loss_top_prob": 0.29288895428180695,
      "entropy": 5.608330249786377,
      "epoch": 0.20261437908496732,
      "grad_norm": 0.49440765380859375,
      "learning_rate": 4.971302952586796e-05,
      "loss": 5.5723,
      "mean_token_accuracy": 0.4264814096259175,
      "num_tokens": 923945.0,
      "step": 31
    },
    {
      "aux_loss_top_prob": 0.29487766325473785,
      "entropy": 5.628073573112488,
      "epoch": 0.20915032679738563,
      "grad_norm": 0.4724545180797577,
      "learning_rate": 4.967066306353816e-05,
      "loss": 5.6501,
      "mean_token_accuracy": 0.42272184946901253,
      "num_tokens": 954195.0,
      "step": 32
    },
    {
      "aux_loss_top_prob": 0.3069535195827484,
      "entropy": 5.494794249534607,
      "epoch": 0.21568627450980393,
      "grad_norm": 0.5053384304046631,
      "learning_rate": 4.9625401389513715e-05,
      "loss": 5.5243,
      "mean_token_accuracy": 0.42991070516214025,
      "num_tokens": 984506.0,
      "step": 33
    },
    {
      "aux_loss_top_prob": 0.3274266868829727,
      "entropy": 5.28958797454834,
      "epoch": 0.2222222222222222,
      "grad_norm": 0.5244802236557007,
      "learning_rate": 4.957724981545275e-05,
      "loss": 5.3475,
      "mean_token_accuracy": 0.45326393704887535,
      "num_tokens": 1014680.0,
      "step": 34
    },
    {
      "aux_loss_top_prob": 0.3393823355436325,
      "entropy": 5.305056095123291,
      "epoch": 0.22875816993464052,
      "grad_norm": 0.46650102734565735,
      "learning_rate": 4.952621399215598e-05,
      "loss": 5.3576,
      "mean_token_accuracy": 0.46336550141871513,
      "num_tokens": 1045372.0,
      "step": 35
    },
    {
      "aux_loss_top_prob": 0.33368056267499924,
      "entropy": 5.4895665645599365,
      "epoch": 0.23529411764705882,
      "grad_norm": 0.3078683316707611,
      "learning_rate": 4.9472299908903555e-05,
      "loss": 5.3221,
      "mean_token_accuracy": 0.4634719286964809,
      "num_tokens": 1076213.0,
      "step": 36
    },
    {
      "aux_loss_top_prob": 0.2859933450818062,
      "entropy": 6.22998571395874,
      "epoch": 0.24183006535947713,
      "grad_norm": 0.5189754962921143,
      "learning_rate": 4.941551389275217e-05,
      "loss": 5.4572,
      "mean_token_accuracy": 0.43466063501428287,
      "num_tokens": 1107156.0,
      "step": 37
    },
    {
      "aux_loss_top_prob": 0.2993287220597267,
      "entropy": 6.121002793312073,
      "epoch": 0.24836601307189543,
      "grad_norm": 0.4989500939846039,
      "learning_rate": 4.9355862607792614e-05,
      "loss": 5.3422,
      "mean_token_accuracy": 0.45575927526717797,
      "num_tokens": 1137160.0,
      "step": 38
    },
    {
      "aux_loss_top_prob": 0.28258900344371796,
      "entropy": 6.398027062416077,
      "epoch": 0.2549019607843137,
      "grad_norm": 0.6917693018913269,
      "learning_rate": 4.929335305436764e-05,
      "loss": 5.4664,
      "mean_token_accuracy": 0.4327917288190263,
      "num_tokens": 1167494.0,
      "step": 39
    },
    {
      "aux_loss_top_prob": 0.30563274770975113,
      "entropy": 5.958781957626343,
      "epoch": 0.26143790849673204,
      "grad_norm": 0.3923698365688324,
      "learning_rate": 4.922799256825051e-05,
      "loss": 5.4543,
      "mean_token_accuracy": 0.44169591755257986,
      "num_tokens": 1197348.0,
      "step": 40
    },
    {
      "aux_loss_top_prob": 0.3127927631139755,
      "entropy": 6.02230179309845,
      "epoch": 0.2679738562091503,
      "grad_norm": 0.4428256154060364,
      "learning_rate": 4.9159788819784064e-05,
      "loss": 5.351,
      "mean_token_accuracy": 0.45706104770944966,
      "num_tokens": 1228062.0,
      "step": 41
    },
    {
      "aux_loss_top_prob": 0.3205755650997162,
      "entropy": 5.6522886753082275,
      "epoch": 0.27450980392156865,
      "grad_norm": 0.2621772885322571,
      "learning_rate": 4.908874981298057e-05,
      "loss": 5.2825,
      "mean_token_accuracy": 0.44493984240899503,
      "num_tokens": 1258247.0,
      "step": 42
    },
    {
      "aux_loss_top_prob": 0.34095269441604614,
      "entropy": 5.3686676025390625,
      "epoch": 0.28104575163398693,
      "grad_norm": 0.3995060622692108,
      "learning_rate": 4.901488388458247e-05,
      "loss": 5.3209,
      "mean_token_accuracy": 0.4549872911852822,
      "num_tokens": 1289296.0,
      "step": 43
    },
    {
      "aux_loss_top_prob": 0.3110523819923401,
      "entropy": 5.65318238735199,
      "epoch": 0.2875816993464052,
      "grad_norm": 0.3274209797382355,
      "learning_rate": 4.893819970308394e-05,
      "loss": 5.4744,
      "mean_token_accuracy": 0.42184712807405683,
      "num_tokens": 1319652.0,
      "step": 44
    },
    {
      "aux_loss_top_prob": 0.3718513175845146,
      "entropy": 4.774181485176086,
      "epoch": 0.29411764705882354,
      "grad_norm": 0.7339034080505371,
      "learning_rate": 4.8858706267713704e-05,
      "loss": 5.1577,
      "mean_token_accuracy": 0.48372706972127694,
      "num_tokens": 1349559.0,
      "step": 45
    },
    {
      "aux_loss_top_prob": 0.31519901007413864,
      "entropy": 5.689408898353577,
      "epoch": 0.3006535947712418,
      "grad_norm": 0.2670136094093323,
      "learning_rate": 4.877641290737884e-05,
      "loss": 5.4477,
      "mean_token_accuracy": 0.4324065104478332,
      "num_tokens": 1380281.0,
      "step": 46
    },
    {
      "aux_loss_top_prob": 0.321474090218544,
      "entropy": 5.51611065864563,
      "epoch": 0.30718954248366015,
      "grad_norm": 0.28277069330215454,
      "learning_rate": 4.869132927957007e-05,
      "loss": 5.3375,
      "mean_token_accuracy": 0.4448794698980791,
      "num_tokens": 1410891.0,
      "step": 47
    },
    {
      "aux_loss_top_prob": 0.32388219982385635,
      "entropy": 5.676973104476929,
      "epoch": 0.3137254901960784,
      "grad_norm": 0.2335694134235382,
      "learning_rate": 4.860346536922834e-05,
      "loss": 5.3984,
      "mean_token_accuracy": 0.454445757763335,
      "num_tokens": 1442245.0,
      "step": 48
    },
    {
      "aux_loss_top_prob": 0.31195351481437683,
      "entropy": 5.717365741729736,
      "epoch": 0.3202614379084967,
      "grad_norm": 0.182363361120224,
      "learning_rate": 4.85128314875731e-05,
      "loss": 5.2382,
      "mean_token_accuracy": 0.463343476477589,
      "num_tokens": 1472951.0,
      "step": 49
    },
    {
      "aux_loss_top_prob": 0.320010744035244,
      "entropy": 5.699024558067322,
      "epoch": 0.32679738562091504,
      "grad_norm": 0.1933782696723938,
      "learning_rate": 4.8419438270892226e-05,
      "loss": 5.2419,
      "mean_token_accuracy": 0.47049604397670625,
      "num_tokens": 1503806.0,
      "step": 50
    },
    {
      "aux_loss_top_prob": 0.27193087339401245,
      "entropy": 6.164070725440979,
      "epoch": 0.3333333333333333,
      "grad_norm": 0.3611936867237091,
      "learning_rate": 4.8323296679293774e-05,
      "loss": 5.3588,
      "mean_token_accuracy": 0.4327859002989618,
      "num_tokens": 1534861.0,
      "step": 51
    },
    {
      "aux_loss_top_prob": 0.2984625920653343,
      "entropy": 5.823672294616699,
      "epoch": 0.33986928104575165,
      "grad_norm": 0.3743869662284851,
      "learning_rate": 4.822441799541979e-05,
      "loss": 5.1196,
      "mean_token_accuracy": 0.43119462130017683,
      "num_tokens": 1558497.0,
      "step": 52
    },
    {
      "aux_loss_top_prob": 0.2813778407871723,
      "entropy": 5.982573628425598,
      "epoch": 0.3464052287581699,
      "grad_norm": 0.25157812237739563,
      "learning_rate": 4.8122813823122225e-05,
      "loss": 5.328,
      "mean_token_accuracy": 0.44321849338682706,
      "num_tokens": 1589029.0,
      "step": 53
    },
    {
      "aux_loss_top_prob": 0.26788379997015,
      "entropy": 6.197737693786621,
      "epoch": 0.35294117647058826,
      "grad_norm": 0.32609760761260986,
      "learning_rate": 4.8018496086101194e-05,
      "loss": 5.4415,
      "mean_token_accuracy": 0.43256239102401733,
      "num_tokens": 1619631.0,
      "step": 54
    },
    {
      "aux_loss_top_prob": 0.2655763439834118,
      "entropy": 6.076657772064209,
      "epoch": 0.35947712418300654,
      "grad_norm": 0.1719447374343872,
      "learning_rate": 4.7911477026505654e-05,
      "loss": 5.5382,
      "mean_token_accuracy": 0.4069664660429695,
      "num_tokens": 1650030.0,
      "step": 55
    },
    {
      "aux_loss_top_prob": 0.32412465661764145,
      "entropy": 5.458210825920105,
      "epoch": 0.3660130718954248,
      "grad_norm": 0.2468564361333847,
      "learning_rate": 4.780176920349675e-05,
      "loss": 5.197,
      "mean_token_accuracy": 0.47210909080828395,
      "num_tokens": 1680959.0,
      "step": 56
    },
    {
      "aux_loss_top_prob": 0.27674129232764244,
      "entropy": 5.815967798233032,
      "epoch": 0.37254901960784315,
      "grad_norm": 0.36906665563583374,
      "learning_rate": 4.768938549177393e-05,
      "loss": 5.2456,
      "mean_token_accuracy": 0.3753740144985471,
      "num_tokens": 1704337.0,
      "step": 57
    },
    {
      "aux_loss_top_prob": 0.3141936808824539,
      "entropy": 5.347635507583618,
      "epoch": 0.3790849673202614,
      "grad_norm": 0.4113600254058838,
      "learning_rate": 4.7574339080064044e-05,
      "loss": 5.2765,
      "mean_token_accuracy": 0.45382242899501307,
      "num_tokens": 1734272.0,
      "step": 58
    },
    {
      "aux_loss_top_prob": 0.29988060891628265,
      "entropy": 5.470825672149658,
      "epoch": 0.38562091503267976,
      "grad_norm": 0.37348005175590515,
      "learning_rate": 4.745664346957361e-05,
      "loss": 5.3641,
      "mean_token_accuracy": 0.43680141033822556,
      "num_tokens": 1764501.0,
      "step": 59
    },
    {
      "aux_loss_top_prob": 0.30954184755682945,
      "entropy": 5.673593521118164,
      "epoch": 0.39215686274509803,
      "grad_norm": 0.41574323177337646,
      "learning_rate": 4.7336312472404345e-05,
      "loss": 5.4796,
      "mean_token_accuracy": 0.42860434948558446,
      "num_tokens": 1788206.0,
      "step": 60
    },
    {
      "aux_loss_top_prob": 0.3175780400633812,
      "entropy": 5.430213451385498,
      "epoch": 0.39869281045751637,
      "grad_norm": 0.26925528049468994,
      "learning_rate": 4.7213360209932286e-05,
      "loss": 5.2018,
      "mean_token_accuracy": 0.4713788801097285,
      "num_tokens": 1818601.0,
      "step": 61
    },
    {
      "aux_loss_top_prob": 0.3094403147697449,
      "entropy": 5.379101514816284,
      "epoch": 0.40522875816993464,
      "grad_norm": 0.35771316289901733,
      "learning_rate": 4.708780111115057e-05,
      "loss": 5.2311,
      "mean_token_accuracy": 0.4583288455486584,
      "num_tokens": 1848872.0,
      "step": 62
    },
    {
      "aux_loss_top_prob": 0.258245125412941,
      "entropy": 6.180609226226807,
      "epoch": 0.4117647058823529,
      "grad_norm": 0.42801010608673096,
      "learning_rate": 4.695964991097616e-05,
      "loss": 5.4092,
      "mean_token_accuracy": 0.41897005669717546,
      "num_tokens": 1880255.0,
      "step": 63
    },
    {
      "aux_loss_top_prob": 0.28765855729579926,
      "entropy": 5.824422478675842,
      "epoch": 0.41830065359477125,
      "grad_norm": 0.12480125576257706,
      "learning_rate": 4.682892164852058e-05,
      "loss": 5.2611,
      "mean_token_accuracy": 0.4518620549015767,
      "num_tokens": 1911008.0,
      "step": 64
    },
    {
      "aux_loss_top_prob": 0.2897224947810173,
      "entropy": 5.851225018501282,
      "epoch": 0.42483660130718953,
      "grad_norm": 0.20092909038066864,
      "learning_rate": 4.669563166532503e-05,
      "loss": 5.1985,
      "mean_token_accuracy": 0.4643623760294153,
      "num_tokens": 1941497.0,
      "step": 65
    },
    {
      "aux_loss_top_prob": 0.2729847691953182,
      "entropy": 6.158845901489258,
      "epoch": 0.43137254901960786,
      "grad_norm": 0.4115382134914398,
      "learning_rate": 4.655979560356005e-05,
      "loss": 5.2787,
      "mean_token_accuracy": 0.44821390247041065,
      "num_tokens": 1973574.0,
      "step": 66
    },
    {
      "aux_loss_top_prob": 0.27163349092006683,
      "entropy": 6.023089408874512,
      "epoch": 0.43790849673202614,
      "grad_norm": 0.1619524508714676,
      "learning_rate": 4.642142940418973e-05,
      "loss": 5.394,
      "mean_token_accuracy": 0.43101454248134063,
      "num_tokens": 2003639.0,
      "step": 67
    },
    {
      "aux_loss_top_prob": 0.2917519025504589,
      "entropy": 5.852391719818115,
      "epoch": 0.4444444444444444,
      "grad_norm": 0.26231828331947327,
      "learning_rate": 4.6280549305101063e-05,
      "loss": 5.2939,
      "mean_token_accuracy": 0.4585062336629131,
      "num_tokens": 2034048.0,
      "step": 68
    },
    {
      "aux_loss_top_prob": 0.24277161806821823,
      "entropy": 6.3362345695495605,
      "epoch": 0.45098039215686275,
      "grad_norm": 0.3012363612651825,
      "learning_rate": 4.61371718391983e-05,
      "loss": 5.2868,
      "mean_token_accuracy": 0.3871433452463109,
      "num_tokens": 2057176.0,
      "step": 69
    },
    {
      "aux_loss_top_prob": 0.27032191678881645,
      "entropy": 6.041122317314148,
      "epoch": 0.45751633986928103,
      "grad_norm": 0.13400579988956451,
      "learning_rate": 4.599131383246277e-05,
      "loss": 5.531,
      "mean_token_accuracy": 0.41904271683796435,
      "num_tokens": 2088592.0,
      "step": 70
    },
    {
      "aux_loss_top_prob": 0.2959485277533531,
      "entropy": 5.618825793266296,
      "epoch": 0.46405228758169936,
      "grad_norm": 0.20397238433361053,
      "learning_rate": 4.5842992401978255e-05,
      "loss": 5.3231,
      "mean_token_accuracy": 0.44134252619446884,
      "num_tokens": 2119611.0,
      "step": 71
    },
    {
      "aux_loss_top_prob": 0.3017946779727936,
      "entropy": 5.413687467575073,
      "epoch": 0.47058823529411764,
      "grad_norm": 0.40486910939216614,
      "learning_rate": 4.5692224953922266e-05,
      "loss": 5.3378,
      "mean_token_accuracy": 0.4349554945662131,
      "num_tokens": 2150151.0,
      "step": 72
    },
    {
      "aux_loss_top_prob": 0.31608158349990845,
      "entropy": 5.355222225189209,
      "epoch": 0.477124183006536,
      "grad_norm": 0.41642773151397705,
      "learning_rate": 4.553902918152329e-05,
      "loss": 5.2568,
      "mean_token_accuracy": 0.4601322313555814,
      "num_tokens": 2180705.0,
      "step": 73
    },
    {
      "aux_loss_top_prob": 0.28519682958722115,
      "entropy": 5.80126416683197,
      "epoch": 0.48366013071895425,
      "grad_norm": 0.20127348601818085,
      "learning_rate": 4.5383423062984455e-05,
      "loss": 5.4807,
      "mean_token_accuracy": 0.42747254223780606,
      "num_tokens": 2211311.0,
      "step": 74
    },
    {
      "aux_loss_top_prob": 0.31189220398664474,
      "entropy": 5.426263093948364,
      "epoch": 0.49019607843137253,
      "grad_norm": 0.28286266326904297,
      "learning_rate": 4.522542485937369e-05,
      "loss": 5.1958,
      "mean_token_accuracy": 0.46722617367582203,
      "num_tokens": 2241824.0,
      "step": 75
    },
    {
      "aux_loss_top_prob": 0.285269632935524,
      "entropy": 5.766371369361877,
      "epoch": 0.49673202614379086,
      "grad_norm": 0.09214834868907928,
      "learning_rate": 4.506505311248072e-05,
      "loss": 5.2898,
      "mean_token_accuracy": 0.445447328754576,
      "num_tokens": 2272033.0,
      "step": 76
    },
    {
      "aux_loss_top_prob": 0.28388892486691475,
      "entropy": 5.87073028087616,
      "epoch": 0.5032679738562091,
      "grad_norm": 0.3044549226760864,
      "learning_rate": 4.4902326642641095e-05,
      "loss": 5.3559,
      "mean_token_accuracy": 0.44479302972318535,
      "num_tokens": 2302686.0,
      "step": 77
    },
    {
      "aux_loss_top_prob": 0.31867603957653046,
      "entropy": 5.635192632675171,
      "epoch": 0.5098039215686274,
      "grad_norm": 0.3503623306751251,
      "learning_rate": 4.473726454652755e-05,
      "loss": 5.1219,
      "mean_token_accuracy": 0.4846873077302209,
      "num_tokens": 2334521.0,
      "step": 78
    },
    {
      "aux_loss_top_prob": 0.2572985477745533,
      "entropy": 6.197256088256836,
      "epoch": 0.5163398692810458,
      "grad_norm": 0.41610655188560486,
      "learning_rate": 4.456988619490889e-05,
      "loss": 5.3505,
      "mean_token_accuracy": 0.4341550667650874,
      "num_tokens": 2365439.0,
      "step": 79
    },
    {
      "aux_loss_top_prob": 0.270380362868309,
      "entropy": 6.079758882522583,
      "epoch": 0.5228758169934641,
      "grad_norm": 0.4577791094779968,
      "learning_rate": 4.4400211230376826e-05,
      "loss": 5.1688,
      "mean_token_accuracy": 0.46271364300903256,
      "num_tokens": 2396149.0,
      "step": 80
    },
    {
      "aux_loss_top_prob": 0.27083877474069595,
      "entropy": 6.120800852775574,
      "epoch": 0.5294117647058824,
      "grad_norm": 0.40247684717178345,
      "learning_rate": 4.422825956504072e-05,
      "loss": 5.253,
      "mean_token_accuracy": 0.45487971974433417,
      "num_tokens": 2426808.0,
      "step": 81
    },
    {
      "aux_loss_top_prob": 0.2824140638113022,
      "entropy": 5.872729420661926,
      "epoch": 0.5359477124183006,
      "grad_norm": 0.18734107911586761,
      "learning_rate": 4.4054051378190915e-05,
      "loss": 5.2093,
      "mean_token_accuracy": 0.4577911494397441,
      "num_tokens": 2457428.0,
      "step": 82
    },
    {
      "aux_loss_top_prob": 0.28997989743947983,
      "entropy": 5.7378764152526855,
      "epoch": 0.5424836601307189,
      "grad_norm": 0.11612078547477722,
      "learning_rate": 4.387760711393052e-05,
      "loss": 5.2986,
      "mean_token_accuracy": 0.45413456305698985,
      "num_tokens": 2487777.0,
      "step": 83
    },
    {
      "aux_loss_top_prob": 0.2969295531511307,
      "entropy": 5.473900318145752,
      "epoch": 0.5490196078431373,
      "grad_norm": 0.26367902755737305,
      "learning_rate": 4.3698947478776264e-05,
      "loss": 5.2024,
      "mean_token_accuracy": 0.45511924317079244,
      "num_tokens": 2517356.0,
      "step": 84
    },
    {
      "aux_loss_top_prob": 0.2670605555176735,
      "entropy": 5.915418386459351,
      "epoch": 0.5555555555555556,
      "grad_norm": 0.1559591442346573,
      "learning_rate": 4.351809343922848e-05,
      "loss": 5.5665,
      "mean_token_accuracy": 0.41069108693288375,
      "num_tokens": 2547545.0,
      "step": 85
    },
    {
      "aux_loss_top_prob": 0.27265437692403793,
      "entropy": 5.830083131790161,
      "epoch": 0.5620915032679739,
      "grad_norm": 0.4524465799331665,
      "learning_rate": 4.3335066219310564e-05,
      "loss": 5.1649,
      "mean_token_accuracy": 0.4072992225429335,
      "num_tokens": 2570256.0,
      "step": 86
    },
    {
      "aux_loss_top_prob": 0.2944645807147026,
      "entropy": 5.536332011222839,
      "epoch": 0.5686274509803921,
      "grad_norm": 0.2920515537261963,
      "learning_rate": 4.3149887298078276e-05,
      "loss": 5.3074,
      "mean_token_accuracy": 0.44265441020975116,
      "num_tokens": 2600131.0,
      "step": 87
    },
    {
      "aux_loss_top_prob": 0.2946447804570198,
      "entropy": 5.561327219009399,
      "epoch": 0.5751633986928104,
      "grad_norm": 0.2950854003429413,
      "learning_rate": 4.2962578407099065e-05,
      "loss": 5.3345,
      "mean_token_accuracy": 0.43823468141312844,
      "num_tokens": 2630765.0,
      "step": 88
    },
    {
      "aux_loss_top_prob": 0.28463171422481537,
      "entropy": 5.743519186973572,
      "epoch": 0.5816993464052288,
      "grad_norm": 0.15321138501167297,
      "learning_rate": 4.277316152790177e-05,
      "loss": 5.3559,
      "mean_token_accuracy": 0.43784945480539894,
      "num_tokens": 2660580.0,
      "step": 89
    },
    {
      "aux_loss_top_prob": 0.29848662763834,
      "entropy": 5.595802307128906,
      "epoch": 0.5882352941176471,
      "grad_norm": 0.15545423328876495,
      "learning_rate": 4.2581658889397e-05,
      "loss": 5.2042,
      "mean_token_accuracy": 0.4592900343050329,
      "num_tokens": 2690965.0,
      "step": 90
    },
    {
      "aux_loss_top_prob": 0.26714448630809784,
      "entropy": 6.078777313232422,
      "epoch": 0.5947712418300654,
      "grad_norm": 0.2409888058900833,
      "learning_rate": 4.238809296526847e-05,
      "loss": 5.3718,
      "mean_token_accuracy": 0.4357970394752558,
      "num_tokens": 2721031.0,
      "step": 91
    },
    {
      "aux_loss_top_prob": 0.2648441754281521,
      "entropy": 6.02350640296936,
      "epoch": 0.6013071895424836,
      "grad_norm": 0.15189139544963837,
      "learning_rate": 4.2192486471335585e-05,
      "loss": 5.4196,
      "mean_token_accuracy": 0.42344294941832816,
      "num_tokens": 2751662.0,
      "step": 92
    },
    {
      "aux_loss_top_prob": 0.28125374391674995,
      "entropy": 5.8624207973480225,
      "epoch": 0.6078431372549019,
      "grad_norm": 0.17689332365989685,
      "learning_rate": 4.19948623628877e-05,
      "loss": 5.1259,
      "mean_token_accuracy": 0.45900414089053443,
      "num_tokens": 2776695.0,
      "step": 93
    },
    {
      "aux_loss_top_prob": 0.2708875238895416,
      "entropy": 6.0234140157699585,
      "epoch": 0.6143790849673203,
      "grad_norm": 0.31065353751182556,
      "learning_rate": 4.179524383199016e-05,
      "loss": 5.2638,
      "mean_token_accuracy": 0.4396490464536922,
      "num_tokens": 2799687.0,
      "step": 94
    },
    {
      "aux_loss_top_prob": 0.30954550206661224,
      "entropy": 5.62044084072113,
      "epoch": 0.6209150326797386,
      "grad_norm": 0.09324009716510773,
      "learning_rate": 4.159365430476261e-05,
      "loss": 5.0721,
      "mean_token_accuracy": 0.48882710023188636,
      "num_tokens": 2830362.0,
      "step": 95
    },
    {
      "aux_loss_top_prob": 0.28293993324041367,
      "entropy": 5.882411241531372,
      "epoch": 0.6274509803921569,
      "grad_norm": 0.11088743805885315,
      "learning_rate": 4.139011743862991e-05,
      "loss": 5.3143,
      "mean_token_accuracy": 0.45119141485698366,
      "num_tokens": 2860058.0,
      "step": 96
    },
    {
      "aux_loss_top_prob": 0.2713022716343403,
      "entropy": 6.042698979377747,
      "epoch": 0.6339869281045751,
      "grad_norm": 0.16292473673820496,
      "learning_rate": 4.118465711954569e-05,
      "loss": 5.4366,
      "mean_token_accuracy": 0.4298002330982428,
      "num_tokens": 2890806.0,
      "step": 97
    },
    {
      "aux_loss_top_prob": 0.29711734503507614,
      "entropy": 5.612577557563782,
      "epoch": 0.6405228758169934,
      "grad_norm": 0.17225398123264313,
      "learning_rate": 4.09772974591894e-05,
      "loss": 5.2506,
      "mean_token_accuracy": 0.4573836219411845,
      "num_tokens": 2921256.0,
      "step": 98
    },
    {
      "aux_loss_top_prob": 0.2980693578720093,
      "entropy": 5.616932153701782,
      "epoch": 0.6470588235294118,
      "grad_norm": 0.16407005488872528,
      "learning_rate": 4.0768062792136555e-05,
      "loss": 5.2178,
      "mean_token_accuracy": 0.45805728960458414,
      "num_tokens": 2950751.0,
      "step": 99
    },
    {
      "aux_loss_top_prob": 0.3006795272231102,
      "entropy": 5.696488380432129,
      "epoch": 0.6535947712418301,
      "grad_norm": 0.1020834743976593,
      "learning_rate": 4.055697767300302e-05,
      "loss": 5.2198,
      "mean_token_accuracy": 0.4677996987718374,
      "num_tokens": 2980763.0,
      "step": 100
    },
    {
      "aux_loss_top_prob": 0.31347963213920593,
      "entropy": 5.443450450897217,
      "epoch": 0.6601307189542484,
      "grad_norm": 0.23078453540802002,
      "learning_rate": 4.034406687356344e-05,
      "loss": 5.1059,
      "mean_token_accuracy": 0.48318610303287973,
      "num_tokens": 3011289.0,
      "step": 101
    },
    {
      "aux_loss_top_prob": 0.2865593135356903,
      "entropy": 5.774625897407532,
      "epoch": 0.6666666666666666,
      "grad_norm": 0.09804519265890121,
      "learning_rate": 4.012935537984414e-05,
      "loss": 5.2992,
      "mean_token_accuracy": 0.44632129173898366,
      "num_tokens": 3042184.0,
      "step": 102
    },
    {
      "aux_loss_top_prob": 0.28445518761873245,
      "entropy": 5.831955552101135,
      "epoch": 0.673202614379085,
      "grad_norm": 0.12780314683914185,
      "learning_rate": 3.991286838919086e-05,
      "loss": 5.2426,
      "mean_token_accuracy": 0.4549766149258607,
      "num_tokens": 3073149.0,
      "step": 103
    },
    {
      "aux_loss_top_prob": 0.24654151126742363,
      "entropy": 6.266985058784485,
      "epoch": 0.6797385620915033,
      "grad_norm": 0.2594946324825287,
      "learning_rate": 3.969463130731183e-05,
      "loss": 5.5838,
      "mean_token_accuracy": 0.40122091356010936,
      "num_tokens": 3104601.0,
      "step": 104
    },
    {
      "aux_loss_top_prob": 0.26238182187080383,
      "entropy": 6.126760244369507,
      "epoch": 0.6862745098039216,
      "grad_norm": 0.23843856155872345,
      "learning_rate": 3.9474669745296225e-05,
      "loss": 5.4451,
      "mean_token_accuracy": 0.4278024364045953,
      "num_tokens": 3135161.0,
      "step": 105
    },
    {
      "aux_loss_top_prob": 0.29740121215581894,
      "entropy": 5.54465115070343,
      "epoch": 0.6928104575163399,
      "grad_norm": 0.21591369807720184,
      "learning_rate": 3.925300951660858e-05,
      "loss": 5.2077,
      "mean_token_accuracy": 0.4541610594766511,
      "num_tokens": 3166286.0,
      "step": 106
    },
    {
      "aux_loss_top_prob": 0.29174312204122543,
      "entropy": 5.746402859687805,
      "epoch": 0.6993464052287581,
      "grad_norm": 0.08898267149925232,
      "learning_rate": 3.902967663405956e-05,
      "loss": 5.2397,
      "mean_token_accuracy": 0.4588921011284478,
      "num_tokens": 3197550.0,
      "step": 107
    },
    {
      "aux_loss_top_prob": 0.29267091304063797,
      "entropy": 5.663579702377319,
      "epoch": 0.7058823529411765,
      "grad_norm": 0.10517707467079163,
      "learning_rate": 3.8804697306753114e-05,
      "loss": 5.1914,
      "mean_token_accuracy": 0.4602220828768604,
      "num_tokens": 3228454.0,
      "step": 108
    },
    {
      "aux_loss_top_prob": 0.26256876438856125,
      "entropy": 6.024183750152588,
      "epoch": 0.7124183006535948,
      "grad_norm": 0.12142185866832733,
      "learning_rate": 3.857809793701082e-05,
      "loss": 5.4884,
      "mean_token_accuracy": 0.4188203046598209,
      "num_tokens": 3259326.0,
      "step": 109
    },
    {
      "aux_loss_top_prob": 0.2699594646692276,
      "entropy": 5.979758024215698,
      "epoch": 0.7189542483660131,
      "grad_norm": 0.8644112944602966,
      "learning_rate": 3.834990511727341e-05,
      "loss": 5.4615,
      "mean_token_accuracy": 0.4312578997195964,
      "num_tokens": 3289717.0,
      "step": 110
    },
    {
      "aux_loss_top_prob": 0.28788091242313385,
      "entropy": 5.736933708190918,
      "epoch": 0.7254901960784313,
      "grad_norm": 0.08829255402088165,
      "learning_rate": 3.812014562698002e-05,
      "loss": 5.2301,
      "mean_token_accuracy": 0.4541675329614392,
      "num_tokens": 3319742.0,
      "step": 111
    },
    {
      "aux_loss_top_prob": 0.2886383607983589,
      "entropy": 5.734727621078491,
      "epoch": 0.7320261437908496,
      "grad_norm": 0.1298859715461731,
      "learning_rate": 3.7888846429425546e-05,
      "loss": 5.2959,
      "mean_token_accuracy": 0.44814435971150846,
      "num_tokens": 3350838.0,
      "step": 112
    },
    {
      "aux_loss_top_prob": 0.2763947993516922,
      "entropy": 5.877326965332031,
      "epoch": 0.738562091503268,
      "grad_norm": 0.22810113430023193,
      "learning_rate": 3.765603466859635e-05,
      "loss": 5.1593,
      "mean_token_accuracy": 0.4431907807398087,
      "num_tokens": 3373807.0,
      "step": 113
    },
    {
      "aux_loss_top_prob": 0.30782556533813477,
      "entropy": 5.595377683639526,
      "epoch": 0.7450980392156863,
      "grad_norm": 0.1488320380449295,
      "learning_rate": 3.742173766598481e-05,
      "loss": 5.1568,
      "mean_token_accuracy": 0.4786267173732044,
      "num_tokens": 3405224.0,
      "step": 114
    },
    {
      "aux_loss_top_prob": 0.2952910587191582,
      "entropy": 5.635329246520996,
      "epoch": 0.7516339869281046,
      "grad_norm": 0.1175958663225174,
      "learning_rate": 3.718598291738298e-05,
      "loss": 5.1842,
      "mean_token_accuracy": 0.46561060000401877,
      "num_tokens": 3435251.0,
      "step": 115
    },
    {
      "aux_loss_top_prob": 0.2627361975610256,
      "entropy": 6.018715500831604,
      "epoch": 0.7581699346405228,
      "grad_norm": 0.3944883346557617,
      "learning_rate": 3.694879808965591e-05,
      "loss": 5.2337,
      "mean_token_accuracy": 0.4281979484743697,
      "num_tokens": 3458396.0,
      "step": 116
    },
    {
      "aux_loss_top_prob": 0.2783399894833565,
      "entropy": 5.89261782169342,
      "epoch": 0.7647058823529411,
      "grad_norm": 0.10377000272274017,
      "learning_rate": 3.671021101749476e-05,
      "loss": 5.3467,
      "mean_token_accuracy": 0.44054940207595705,
      "num_tokens": 3488654.0,
      "step": 117
    },
    {
      "aux_loss_top_prob": 0.26020700111985207,
      "entropy": 6.048100471496582,
      "epoch": 0.7712418300653595,
      "grad_norm": 0.10772567987442017,
      "learning_rate": 3.6470249700150274e-05,
      "loss": 5.515,
      "mean_token_accuracy": 0.41456932842848737,
      "num_tokens": 3519495.0,
      "step": 118
    },
    {
      "aux_loss_top_prob": 0.29782455414533615,
      "entropy": 5.650368928909302,
      "epoch": 0.7777777777777778,
      "grad_norm": 0.1058153510093689,
      "learning_rate": 3.622894229814698e-05,
      "loss": 5.1707,
      "mean_token_accuracy": 0.4631255479069383,
      "num_tokens": 3550480.0,
      "step": 119
    },
    {
      "aux_loss_top_prob": 0.27120138704776764,
      "entropy": 5.87645947933197,
      "epoch": 0.7843137254901961,
      "grad_norm": 0.09994731843471527,
      "learning_rate": 3.598631712997841e-05,
      "loss": 5.3302,
      "mean_token_accuracy": 0.43505979189267036,
      "num_tokens": 3580946.0,
      "step": 120
    },
    {
      "aux_loss_top_prob": 0.26725292205810547,
      "entropy": 5.9307966232299805,
      "epoch": 0.7908496732026143,
      "grad_norm": 0.20257869362831116,
      "learning_rate": 3.57424026687838e-05,
      "loss": 5.4671,
      "mean_token_accuracy": 0.42089543578729505,
      "num_tokens": 3611377.0,
      "step": 121
    },
    {
      "aux_loss_top_prob": 0.28511615097522736,
      "entropy": 5.704583048820496,
      "epoch": 0.7973856209150327,
      "grad_norm": 0.13109956681728363,
      "learning_rate": 3.5497227539006614e-05,
      "loss": 5.246,
      "mean_token_accuracy": 0.4477267896470311,
      "num_tokens": 3642046.0,
      "step": 122
    },
    {
      "aux_loss_top_prob": 0.27258673310279846,
      "entropy": 5.903132081031799,
      "epoch": 0.803921568627451,
      "grad_norm": 0.0934256985783577,
      "learning_rate": 3.52508205130354e-05,
      "loss": 5.4332,
      "mean_token_accuracy": 0.42981128126513074,
      "num_tokens": 3672390.0,
      "step": 123
    },
    {
      "aux_loss_top_prob": 0.25118427723646164,
      "entropy": 6.170169949531555,
      "epoch": 0.8104575163398693,
      "grad_norm": 0.27062803506851196,
      "learning_rate": 3.500321050782717e-05,
      "loss": 5.158,
      "mean_token_accuracy": 0.39776051666283785,
      "num_tokens": 3695062.0,
      "step": 124
    },
    {
      "aux_loss_top_prob": 0.28016486018896103,
      "entropy": 5.732437491416931,
      "epoch": 0.8169934640522876,
      "grad_norm": 0.12713772058486938,
      "learning_rate": 3.4754426581513866e-05,
      "loss": 5.3188,
      "mean_token_accuracy": 0.43622297558479045,
      "num_tokens": 3725228.0,
      "step": 125
    },
    {
      "aux_loss_top_prob": 0.26231709495186806,
      "entropy": 6.00349223613739,
      "epoch": 0.8235294117647058,
      "grad_norm": 0.22227346897125244,
      "learning_rate": 3.45044979299923e-05,
      "loss": 5.2175,
      "mean_token_accuracy": 0.4248315195631483,
      "num_tokens": 3748227.0,
      "step": 126
    },
    {
      "aux_loss_top_prob": 0.2830303944647312,
      "entropy": 5.740333795547485,
      "epoch": 0.8300653594771242,
      "grad_norm": 0.09625742584466934,
      "learning_rate": 3.425345388349786e-05,
      "loss": 5.2548,
      "mean_token_accuracy": 0.4427142817500242,
      "num_tokens": 3779517.0,
      "step": 127
    },
    {
      "aux_loss_top_prob": 0.26446598023176193,
      "entropy": 5.978859543800354,
      "epoch": 0.8366013071895425,
      "grad_norm": 0.13702265918254852,
      "learning_rate": 3.400132390316247e-05,
      "loss": 5.5049,
      "mean_token_accuracy": 0.4190405638371341,
      "num_tokens": 3810669.0,
      "step": 128
    },
    {
      "aux_loss_top_prob": 0.27359529584646225,
      "entropy": 5.870739459991455,
      "epoch": 0.8431372549019608,
      "grad_norm": 0.14332513511180878,
      "learning_rate": 3.374813757755721e-05,
      "loss": 5.2749,
      "mean_token_accuracy": 0.44484871365721285,
      "num_tokens": 3840664.0,
      "step": 129
    },
    {
      "aux_loss_top_prob": 0.3053908310830593,
      "entropy": 5.53608500957489,
      "epoch": 0.8496732026143791,
      "grad_norm": 0.30566075444221497,
      "learning_rate": 3.349392461921996e-05,
      "loss": 5.2373,
      "mean_token_accuracy": 0.4678137752208755,
      "num_tokens": 3864733.0,
      "step": 130
    },
    {
      "aux_loss_top_prob": 0.2768132612109184,
      "entropy": 5.785483956336975,
      "epoch": 0.8562091503267973,
      "grad_norm": 0.14120185375213623,
      "learning_rate": 3.323871486116851e-05,
      "loss": 5.3738,
      "mean_token_accuracy": 0.43184680633573996,
      "num_tokens": 3895651.0,
      "step": 131
    },
    {
      "aux_loss_top_prob": 0.2972762957215309,
      "entropy": 5.739743947982788,
      "epoch": 0.8627450980392157,
      "grad_norm": 0.14439521729946136,
      "learning_rate": 3.29825382533995e-05,
      "loss": 5.1072,
      "mean_token_accuracy": 0.47959467645942905,
      "num_tokens": 3925628.0,
      "step": 132
    },
    {
      "aux_loss_top_prob": 0.24690787866711617,
      "entropy": 6.281315445899963,
      "epoch": 0.869281045751634,
      "grad_norm": 0.30964189767837524,
      "learning_rate": 3.272542485937369e-05,
      "loss": 5.2172,
      "mean_token_accuracy": 0.40981597637240663,
      "num_tokens": 3948650.0,
      "step": 133
    },
    {
      "aux_loss_top_prob": 0.30192238837480545,
      "entropy": 5.477472901344299,
      "epoch": 0.8758169934640523,
      "grad_norm": 0.22800329327583313,
      "learning_rate": 3.2467404852487844e-05,
      "loss": 5.1478,
      "mean_token_accuracy": 0.4659123627165386,
      "num_tokens": 3979541.0,
      "step": 134
    },
    {
      "aux_loss_top_prob": 0.2982357442378998,
      "entropy": 5.5648273229599,
      "epoch": 0.8823529411764706,
      "grad_norm": 0.15770165622234344,
      "learning_rate": 3.220850851253377e-05,
      "loss": 5.1659,
      "mean_token_accuracy": 0.4616253411989052,
      "num_tokens": 4009560.0,
      "step": 135
    },
    {
      "aux_loss_top_prob": 0.2918320894241333,
      "entropy": 5.672165155410767,
      "epoch": 0.8888888888888888,
      "grad_norm": 0.1274181604385376,
      "learning_rate": 3.194876622214486e-05,
      "loss": 5.2381,
      "mean_token_accuracy": 0.45542297778589313,
      "num_tokens": 4039828.0,
      "step": 136
    },
    {
      "aux_loss_top_prob": 0.2933979257941246,
      "entropy": 5.640422940254211,
      "epoch": 0.8954248366013072,
      "grad_norm": 0.1309153139591217,
      "learning_rate": 3.168820846323053e-05,
      "loss": 5.205,
      "mean_token_accuracy": 0.45925799023869157,
      "num_tokens": 4070221.0,
      "step": 137
    },
    {
      "aux_loss_top_prob": 0.29129164665937424,
      "entropy": 5.548122763633728,
      "epoch": 0.9019607843137255,
      "grad_norm": 0.1531859189271927,
      "learning_rate": 3.142686581339902e-05,
      "loss": 5.1316,
      "mean_token_accuracy": 0.4564025402919773,
      "num_tokens": 4100569.0,
      "step": 138
    },
    {
      "aux_loss_top_prob": 0.2943243384361267,
      "entropy": 5.735298752784729,
      "epoch": 0.9084967320261438,
      "grad_norm": 0.13864873349666595,
      "learning_rate": 3.1164768942369056e-05,
      "loss": 5.133,
      "mean_token_accuracy": 0.469797830744754,
      "num_tokens": 4131302.0,
      "step": 139
    },
    {
      "aux_loss_top_prob": 0.26606735959649086,
      "entropy": 6.181024670600891,
      "epoch": 0.9150326797385621,
      "grad_norm": 0.4202753007411957,
      "learning_rate": 3.09019486083705e-05,
      "loss": 4.9737,
      "mean_token_accuracy": 0.46598567620797965,
      "num_tokens": 4154186.0,
      "step": 140
    },
    {
      "aux_loss_top_prob": 0.2758731357753277,
      "entropy": 5.9263739585876465,
      "epoch": 0.9215686274509803,
      "grad_norm": 0.19120648503303528,
      "learning_rate": 3.063843565453486e-05,
      "loss": 5.2769,
      "mean_token_accuracy": 0.4439469972576481,
      "num_tokens": 4185121.0,
      "step": 141
    },
    {
      "aux_loss_top_prob": 0.27671580761671066,
      "entropy": 5.873229265213013,
      "epoch": 0.9281045751633987,
      "grad_norm": 0.1959613859653473,
      "learning_rate": 3.0374261005275607e-05,
      "loss": 5.266,
      "mean_token_accuracy": 0.44586291164554914,
      "num_tokens": 4215964.0,
      "step": 142
    },
    {
      "aux_loss_top_prob": 0.23543573170900345,
      "entropy": 6.591423749923706,
      "epoch": 0.934640522875817,
      "grad_norm": 0.48601460456848145,
      "learning_rate": 3.0109455662659124e-05,
      "loss": 5.3152,
      "mean_token_accuracy": 0.41184675098743806,
      "num_tokens": 4238592.0,
      "step": 143
    },
    {
      "aux_loss_top_prob": 0.2799597829580307,
      "entropy": 5.8319196701049805,
      "epoch": 0.9411764705882353,
      "grad_norm": 0.10880076140165329,
      "learning_rate": 2.9844050702766463e-05,
      "loss": 5.3267,
      "mean_token_accuracy": 0.441876293100509,
      "num_tokens": 4269175.0,
      "step": 144
    },
    {
      "aux_loss_top_prob": 0.3056688979268074,
      "entropy": 5.380853652954102,
      "epoch": 0.9477124183006536,
      "grad_norm": 0.3199579119682312,
      "learning_rate": 2.9578077272046408e-05,
      "loss": 5.1387,
      "mean_token_accuracy": 0.4600725234948599,
      "num_tokens": 4298799.0,
      "step": 145
    },
    {
      "aux_loss_top_prob": 0.27907905355095863,
      "entropy": 5.797984957695007,
      "epoch": 0.954248366013072,
      "grad_norm": 0.17270171642303467,
      "learning_rate": 2.9311566583660317e-05,
      "loss": 5.4427,
      "mean_token_accuracy": 0.42895643061863314,
      "num_tokens": 4329345.0,
      "step": 146
    },
    {
      "aux_loss_top_prob": 0.3032563664019108,
      "entropy": 5.522157073020935,
      "epoch": 0.9607843137254902,
      "grad_norm": 0.2984372079372406,
      "learning_rate": 2.9044549913819124e-05,
      "loss": 5.2508,
      "mean_token_accuracy": 0.45798240837377735,
      "num_tokens": 4359814.0,
      "step": 147
    },
    {
      "aux_loss_top_prob": 0.28133249282836914,
      "entropy": 5.688148021697998,
      "epoch": 0.9673202614379085,
      "grad_norm": 0.2108151912689209,
      "learning_rate": 2.8777058598112923e-05,
      "loss": 5.3582,
      "mean_token_accuracy": 0.430391151471764,
      "num_tokens": 4390435.0,
      "step": 148
    },
    {
      "aux_loss_top_prob": 0.2738282345235348,
      "entropy": 5.8304396867752075,
      "epoch": 0.9738562091503268,
      "grad_norm": 0.18369758129119873,
      "learning_rate": 2.850912402783361e-05,
      "loss": 5.4824,
      "mean_token_accuracy": 0.41643854566266175,
      "num_tokens": 4421605.0,
      "step": 149
    },
    {
      "aux_loss_top_prob": 0.27655337378382683,
      "entropy": 5.774801135063171,
      "epoch": 0.9803921568627451,
      "grad_norm": 0.11969313025474548,
      "learning_rate": 2.8240777646290973e-05,
      "loss": 5.3427,
      "mean_token_accuracy": 0.43360380778178886,
      "num_tokens": 4452032.0,
      "step": 150
    },
    {
      "aux_loss_top_prob": 0.27796487510204315,
      "entropy": 5.782013416290283,
      "epoch": 0.9869281045751634,
      "grad_norm": 0.10608270764350891,
      "learning_rate": 2.7972050945122662e-05,
      "loss": 5.2316,
      "mean_token_accuracy": 0.44675124315636106,
      "num_tokens": 4482577.0,
      "step": 151
    },
    {
      "aux_loss_top_prob": 0.2814709097146988,
      "entropy": 5.847845196723938,
      "epoch": 0.9934640522875817,
      "grad_norm": 0.1136578768491745,
      "learning_rate": 2.7702975460598547e-05,
      "loss": 5.2655,
      "mean_token_accuracy": 0.4503836114698603,
      "num_tokens": 4512979.0,
      "step": 152
    },
    {
      "aux_loss_top_prob": 0.2532176449894905,
      "entropy": 6.192783713340759,
      "epoch": 1.0,
      "grad_norm": 0.256334125995636,
      "learning_rate": 2.7433582769919754e-05,
      "loss": 5.5342,
      "mean_token_accuracy": 0.41458234099297203,
      "num_tokens": 4544027.0,
      "step": 153
    },
    {
      "aux_loss_top_prob": 0.280241422355175,
      "entropy": 5.88430392742157,
      "epoch": 1.0065359477124183,
      "grad_norm": 0.18610650300979614,
      "learning_rate": 2.7163904487512944e-05,
      "loss": 5.2313,
      "mean_token_accuracy": 0.45606894315315905,
      "num_tokens": 4574131.0,
      "step": 154
    },
    {
      "aux_loss_top_prob": 0.25071264058351517,
      "entropy": 6.383748292922974,
      "epoch": 1.0130718954248366,
      "grad_norm": 0.43775466084480286,
      "learning_rate": 2.6893972261320266e-05,
      "loss": 5.2084,
      "mean_token_accuracy": 0.4247283724778649,
      "num_tokens": 4596918.0,
      "step": 155
    },
    {
      "aux_loss_top_prob": 0.2792350463569164,
      "entropy": 5.864920616149902,
      "epoch": 1.0196078431372548,
      "grad_norm": 0.10352112352848053,
      "learning_rate": 2.662381776908527e-05,
      "loss": 5.3785,
      "mean_token_accuracy": 0.4353010431114778,
      "num_tokens": 4628179.0,
      "step": 156
    },
    {
      "aux_loss_top_prob": 0.2826084792613983,
      "entropy": 5.778141140937805,
      "epoch": 1.026143790849673,
      "grad_norm": 0.11390064656734467,
      "learning_rate": 2.635347271463544e-05,
      "loss": 5.2132,
      "mean_token_accuracy": 0.45673764642213055,
      "num_tokens": 4658227.0,
      "step": 157
    },
    {
      "aux_loss_top_prob": 0.28147098422050476,
      "entropy": 5.709586024284363,
      "epoch": 1.0326797385620916,
      "grad_norm": 0.13762709498405457,
      "learning_rate": 2.6082968824161557e-05,
      "loss": 5.2667,
      "mean_token_accuracy": 0.4472941826180451,
      "num_tokens": 4688925.0,
      "step": 158
    },
    {
      "aux_loss_top_prob": 0.2917625680565834,
      "entropy": 5.566564679145813,
      "epoch": 1.0392156862745099,
      "grad_norm": 0.19779188930988312,
      "learning_rate": 2.5812337842494516e-05,
      "loss": 5.1812,
      "mean_token_accuracy": 0.45432654132720074,
      "num_tokens": 4719223.0,
      "step": 159
    },
    {
      "aux_loss_top_prob": 0.30887170135974884,
      "entropy": 5.4031288623809814,
      "epoch": 1.0457516339869282,
      "grad_norm": 0.2922874987125397,
      "learning_rate": 2.554161152937994e-05,
      "loss": 5.1072,
      "mean_token_accuracy": 0.47278477949298103,
      "num_tokens": 4750171.0,
      "step": 160
    },
    {
      "aux_loss_top_prob": 0.29656875878572464,
      "entropy": 5.682553768157959,
      "epoch": 1.0522875816993464,
      "grad_norm": 0.10858976095914841,
      "learning_rate": 2.5270821655751e-05,
      "loss": 5.1808,
      "mean_token_accuracy": 0.47104811371926736,
      "num_tokens": 4780089.0,
      "step": 161
    },
    {
      "aux_loss_top_prob": 0.2817157730460167,
      "entropy": 5.803648948669434,
      "epoch": 1.0588235294117647,
      "grad_norm": 0.10089188069105148,
      "learning_rate": 2.5e-05,
      "loss": 5.3304,
      "mean_token_accuracy": 0.44400369367408543,
      "num_tokens": 4810611.0,
      "step": 162
    },
    {
      "aux_loss_top_prob": 0.27601689100265503,
      "entropy": 5.8239521980285645,
      "epoch": 1.065359477124183,
      "grad_norm": 0.1516360640525818,
      "learning_rate": 2.4729178344249006e-05,
      "loss": 5.2393,
      "mean_token_accuracy": 0.44875673947614325,
      "num_tokens": 4840089.0,
      "step": 163
    },
    {
      "aux_loss_top_prob": 0.27090488374233246,
      "entropy": 6.032192945480347,
      "epoch": 1.0718954248366013,
      "grad_norm": 0.2205819934606552,
      "learning_rate": 2.4458388470620067e-05,
      "loss": 5.3754,
      "mean_token_accuracy": 0.4400788437131369,
      "num_tokens": 4871957.0,
      "step": 164
    },
    {
      "aux_loss_top_prob": 0.2783363088965416,
      "entropy": 5.913723468780518,
      "epoch": 1.0784313725490196,
      "grad_norm": 0.1869364082813263,
      "learning_rate": 2.418766215750549e-05,
      "loss": 5.2818,
      "mean_token_accuracy": 0.45564274302793006,
      "num_tokens": 4901588.0,
      "step": 165
    },
    {
      "aux_loss_top_prob": 0.2840847298502922,
      "entropy": 5.783800721168518,
      "epoch": 1.0849673202614378,
      "grad_norm": 0.10851762443780899,
      "learning_rate": 2.391703117583845e-05,
      "loss": 5.2329,
      "mean_token_accuracy": 0.4559316791214359,
      "num_tokens": 4932409.0,
      "step": 166
    },
    {
      "aux_loss_top_prob": 0.2840278595685959,
      "entropy": 5.720959901809692,
      "epoch": 1.091503267973856,
      "grad_norm": 0.1325722187757492,
      "learning_rate": 2.3646527285364565e-05,
      "loss": 5.288,
      "mean_token_accuracy": 0.4464286355694884,
      "num_tokens": 4962625.0,
      "step": 167
    },
    {
      "aux_loss_top_prob": 0.2849148064851761,
      "entropy": 5.815640687942505,
      "epoch": 1.0980392156862746,
      "grad_norm": 0.10391997545957565,
      "learning_rate": 2.337618223091473e-05,
      "loss": 5.2732,
      "mean_token_accuracy": 0.45383645072722256,
      "num_tokens": 4992497.0,
      "step": 168
    },
    {
      "aux_loss_top_prob": 0.2904425486922264,
      "entropy": 5.657232046127319,
      "epoch": 1.1045751633986929,
      "grad_norm": 0.14013437926769257,
      "learning_rate": 2.3106027738679743e-05,
      "loss": 5.2231,
      "mean_token_accuracy": 0.45307925103163477,
      "num_tokens": 5023139.0,
      "step": 169
    },
    {
      "aux_loss_top_prob": 0.278327539563179,
      "entropy": 5.747777342796326,
      "epoch": 1.1111111111111112,
      "grad_norm": 0.16339990496635437,
      "learning_rate": 2.283609551248706e-05,
      "loss": 5.3652,
      "mean_token_accuracy": 0.4279139505387279,
      "num_tokens": 5053420.0,
      "step": 170
    },
    {
      "aux_loss_top_prob": 0.2910778447985649,
      "entropy": 5.654794931411743,
      "epoch": 1.1176470588235294,
      "grad_norm": 0.13097801804542542,
      "learning_rate": 2.2566417230080262e-05,
      "loss": 5.1922,
      "mean_token_accuracy": 0.4547990082962557,
      "num_tokens": 5083754.0,
      "step": 171
    },
    {
      "aux_loss_top_prob": 0.2809377536177635,
      "entropy": 5.79895293712616,
      "epoch": 1.1241830065359477,
      "grad_norm": 0.09876923263072968,
      "learning_rate": 2.2297024539401463e-05,
      "loss": 5.3069,
      "mean_token_accuracy": 0.4404381798775594,
      "num_tokens": 5114796.0,
      "step": 172
    },
    {
      "aux_loss_top_prob": 0.2802402749657631,
      "entropy": 5.8849116563797,
      "epoch": 1.130718954248366,
      "grad_norm": 0.17107968032360077,
      "learning_rate": 2.202794905487734e-05,
      "loss": 5.249,
      "mean_token_accuracy": 0.45283557268301433,
      "num_tokens": 5144790.0,
      "step": 173
    },
    {
      "aux_loss_top_prob": 0.26414455100893974,
      "entropy": 6.00560736656189,
      "epoch": 1.1372549019607843,
      "grad_norm": 0.2380240559577942,
      "learning_rate": 2.175922235370904e-05,
      "loss": 5.3349,
      "mean_token_accuracy": 0.4366737057428298,
      "num_tokens": 5175904.0,
      "step": 174
    },
    {
      "aux_loss_top_prob": 0.2814266011118889,
      "entropy": 5.798683762550354,
      "epoch": 1.1437908496732025,
      "grad_norm": 0.14212630689144135,
      "learning_rate": 2.1490875972166395e-05,
      "loss": 5.206,
      "mean_token_accuracy": 0.45370829396913304,
      "num_tokens": 5205737.0,
      "step": 175
    },
    {
      "aux_loss_top_prob": 0.26877105981111526,
      "entropy": 5.88240647315979,
      "epoch": 1.1503267973856208,
      "grad_norm": 0.10077331215143204,
      "learning_rate": 2.1222941401887086e-05,
      "loss": 5.4131,
      "mean_token_accuracy": 0.42383778870742056,
      "num_tokens": 5236594.0,
      "step": 176
    },
    {
      "aux_loss_top_prob": 0.29696179181337357,
      "entropy": 5.529824256896973,
      "epoch": 1.156862745098039,
      "grad_norm": 0.23808398842811584,
      "learning_rate": 2.0955450086180882e-05,
      "loss": 5.17,
      "mean_token_accuracy": 0.463474998478525,
      "num_tokens": 5267642.0,
      "step": 177
    },
    {
      "aux_loss_top_prob": 0.28522709757089615,
      "entropy": 5.719590783119202,
      "epoch": 1.1633986928104576,
      "grad_norm": 0.10429779440164566,
      "learning_rate": 2.0688433416339692e-05,
      "loss": 5.1838,
      "mean_token_accuracy": 0.45776626866964826,
      "num_tokens": 5297307.0,
      "step": 178
    },
    {
      "aux_loss_top_prob": 0.2800040915608406,
      "entropy": 5.777139782905579,
      "epoch": 1.1699346405228759,
      "grad_norm": 0.1022808700799942,
      "learning_rate": 2.0421922727953595e-05,
      "loss": 5.2479,
      "mean_token_accuracy": 0.45033869549089256,
      "num_tokens": 5327964.0,
      "step": 179
    },
    {
      "aux_loss_top_prob": 0.29455822333693504,
      "entropy": 5.726712703704834,
      "epoch": 1.1764705882352942,
      "grad_norm": 0.1025228202342987,
      "learning_rate": 2.0155949297233543e-05,
      "loss": 5.2455,
      "mean_token_accuracy": 0.46535804234335065,
      "num_tokens": 5359115.0,
      "step": 180
    },
    {
      "aux_loss_top_prob": 0.30131253600120544,
      "entropy": 5.533236742019653,
      "epoch": 1.1830065359477124,
      "grad_norm": 0.1306033581495285,
      "learning_rate": 1.9890544337340882e-05,
      "loss": 5.0737,
      "mean_token_accuracy": 0.4735176627634492,
      "num_tokens": 5389863.0,
      "step": 181
    },
    {
      "aux_loss_top_prob": 0.2848794609308243,
      "entropy": 5.697869300842285,
      "epoch": 1.1895424836601307,
      "grad_norm": 0.09805908054113388,
      "learning_rate": 1.96257389947244e-05,
      "loss": 5.1805,
      "mean_token_accuracy": 0.45311685031718457,
      "num_tokens": 5420294.0,
      "step": 182
    },
    {
      "aux_loss_top_prob": 0.25775984302163124,
      "entropy": 5.986367464065552,
      "epoch": 1.196078431372549,
      "grad_norm": 0.10857228189706802,
      "learning_rate": 1.9361564345465148e-05,
      "loss": 5.4858,
      "mean_token_accuracy": 0.4143642526469295,
      "num_tokens": 5451425.0,
      "step": 183
    },
    {
      "aux_loss_top_prob": 0.3028993606567383,
      "entropy": 5.554796576499939,
      "epoch": 1.2026143790849673,
      "grad_norm": 0.10437875986099243,
      "learning_rate": 1.9098051391629497e-05,
      "loss": 5.0454,
      "mean_token_accuracy": 0.47777103247854485,
      "num_tokens": 5482546.0,
      "step": 184
    },
    {
      "aux_loss_top_prob": 0.2857842817902565,
      "entropy": 5.757816314697266,
      "epoch": 1.2091503267973855,
      "grad_norm": 0.12509027123451233,
      "learning_rate": 1.8835231057630953e-05,
      "loss": 5.152,
      "mean_token_accuracy": 0.4643214137523283,
      "num_tokens": 5512893.0,
      "step": 185
    },
    {
      "aux_loss_top_prob": 0.2636541686952114,
      "entropy": 6.016231298446655,
      "epoch": 1.215686274509804,
      "grad_norm": 0.2238379716873169,
      "learning_rate": 1.8573134186600975e-05,
      "loss": 5.3554,
      "mean_token_accuracy": 0.43563644409839,
      "num_tokens": 5542867.0,
      "step": 186
    },
    {
      "aux_loss_top_prob": 0.286666639149189,
      "entropy": 5.803942561149597,
      "epoch": 1.2222222222222223,
      "grad_norm": 0.18973374366760254,
      "learning_rate": 1.8311791536769483e-05,
      "loss": 5.1417,
      "mean_token_accuracy": 0.4707162417477901,
      "num_tokens": 5573179.0,
      "step": 187
    },
    {
      "aux_loss_top_prob": 0.2730545848608017,
      "entropy": 5.844292521476746,
      "epoch": 1.2287581699346406,
      "grad_norm": 0.11710316687822342,
      "learning_rate": 1.8051233777855148e-05,
      "loss": 5.2763,
      "mean_token_accuracy": 0.440268440892692,
      "num_tokens": 5603658.0,
      "step": 188
    },
    {
      "aux_loss_top_prob": 0.2942955866456032,
      "entropy": 5.5773738622665405,
      "epoch": 1.2352941176470589,
      "grad_norm": 0.13799655437469482,
      "learning_rate": 1.7791491487466234e-05,
      "loss": 5.1338,
      "mean_token_accuracy": 0.46307624152275684,
      "num_tokens": 5633627.0,
      "step": 189
    },
    {
      "aux_loss_top_prob": 0.2908540517091751,
      "entropy": 5.784247279167175,
      "epoch": 1.2418300653594772,
      "grad_norm": 0.13116711378097534,
      "learning_rate": 1.7532595147512165e-05,
      "loss": 5.282,
      "mean_token_accuracy": 0.45693084264753026,
      "num_tokens": 5664615.0,
      "step": 190
    },
    {
      "aux_loss_top_prob": 0.2845427319407463,
      "entropy": 5.651019215583801,
      "epoch": 1.2483660130718954,
      "grad_norm": 0.1246897429227829,
      "learning_rate": 1.7274575140626318e-05,
      "loss": 5.1983,
      "mean_token_accuracy": 0.44669259779799797,
      "num_tokens": 5696003.0,
      "step": 191
    },
    {
      "aux_loss_top_prob": 0.2926280349493027,
      "entropy": 5.643812894821167,
      "epoch": 1.2549019607843137,
      "grad_norm": 0.12411350756883621,
      "learning_rate": 1.7017461746600506e-05,
      "loss": 5.1858,
      "mean_token_accuracy": 0.46038126721050093,
      "num_tokens": 5726573.0,
      "step": 192
    },
    {
      "aux_loss_top_prob": 0.2911618947982788,
      "entropy": 5.682378530502319,
      "epoch": 1.261437908496732,
      "grad_norm": 0.11092296987771988,
      "learning_rate": 1.6761285138831492e-05,
      "loss": 5.1965,
      "mean_token_accuracy": 0.4604546322020468,
      "num_tokens": 5756665.0,
      "step": 193
    },
    {
      "aux_loss_top_prob": 0.28558604419231415,
      "entropy": 5.727760672569275,
      "epoch": 1.2679738562091503,
      "grad_norm": 0.1259036362171173,
      "learning_rate": 1.6506075380780042e-05,
      "loss": 5.2756,
      "mean_token_accuracy": 0.44578778362089444,
      "num_tokens": 5787400.0,
      "step": 194
    },
    {
      "aux_loss_top_prob": 0.28636419028043747,
      "entropy": 5.682798504829407,
      "epoch": 1.2745098039215685,
      "grad_norm": 0.15135331451892853,
      "learning_rate": 1.625186242244279e-05,
      "loss": 5.2493,
      "mean_token_accuracy": 0.45213492107779235,
      "num_tokens": 5818096.0,
      "step": 195
    },
    {
      "aux_loss_top_prob": 0.2845500558614731,
      "entropy": 5.712936878204346,
      "epoch": 1.2810457516339868,
      "grad_norm": 0.09489395469427109,
      "learning_rate": 1.5998676096837534e-05,
      "loss": 5.216,
      "mean_token_accuracy": 0.45270093043475756,
      "num_tokens": 5848120.0,
      "step": 196
    },
    {
      "aux_loss_top_prob": 0.28729623556137085,
      "entropy": 5.825428485870361,
      "epoch": 1.287581699346405,
      "grad_norm": 0.13823136687278748,
      "learning_rate": 1.574654611650214e-05,
      "loss": 5.2614,
      "mean_token_accuracy": 0.45973229369725305,
      "num_tokens": 5879484.0,
      "step": 197
    },
    {
      "aux_loss_top_prob": 0.29280292242765427,
      "entropy": 5.567736268043518,
      "epoch": 1.2941176470588236,
      "grad_norm": 0.21606343984603882,
      "learning_rate": 1.54955020700077e-05,
      "loss": 5.1873,
      "mean_token_accuracy": 0.4557080171355149,
      "num_tokens": 5910759.0,
      "step": 198
    },
    {
      "aux_loss_top_prob": 0.26599927991628647,
      "entropy": 5.923296093940735,
      "epoch": 1.3006535947712419,
      "grad_norm": 0.11450164020061493,
      "learning_rate": 1.5245573418486136e-05,
      "loss": 5.3794,
      "mean_token_accuracy": 0.42418440375953476,
      "num_tokens": 5941052.0,
      "step": 199
    },
    {
      "aux_loss_top_prob": 0.28016459569334984,
      "entropy": 5.772648334503174,
      "epoch": 1.3071895424836601,
      "grad_norm": 0.10269754379987717,
      "learning_rate": 1.4996789492172836e-05,
      "loss": 5.2118,
      "mean_token_accuracy": 0.45035657816340163,
      "num_tokens": 5970571.0,
      "step": 200
    },
    {
      "aux_loss_top_prob": 0.2950386181473732,
      "entropy": 5.732956886291504,
      "epoch": 1.3137254901960784,
      "grad_norm": 0.16262753307819366,
      "learning_rate": 1.4749179486964598e-05,
      "loss": 5.0776,
      "mean_token_accuracy": 0.4831611690813547,
      "num_tokens": 6001234.0,
      "step": 201
    },
    {
      "aux_loss_top_prob": 0.28757135570049286,
      "entropy": 5.828650236129761,
      "epoch": 1.3202614379084967,
      "grad_norm": 0.8942100405693054,
      "learning_rate": 1.4502772460993385e-05,
      "loss": 5.1648,
      "mean_token_accuracy": 0.4661281497346823,
      "num_tokens": 6032924.0,
      "step": 202
    },
    {
      "aux_loss_top_prob": 0.27078814804553986,
      "entropy": 5.822415709495544,
      "epoch": 1.326797385620915,
      "grad_norm": 0.09996708482503891,
      "learning_rate": 1.425759733121621e-05,
      "loss": 5.2933,
      "mean_token_accuracy": 0.4387624058707818,
      "num_tokens": 6064312.0,
      "step": 203
    },
    {
      "aux_loss_top_prob": 0.27583659440279007,
      "entropy": 5.881180763244629,
      "epoch": 1.3333333333333333,
      "grad_norm": 0.14784276485443115,
      "learning_rate": 1.4013682870021594e-05,
      "loss": 5.2911,
      "mean_token_accuracy": 0.444652655888536,
      "num_tokens": 6094605.0,
      "step": 204
    },
    {
      "aux_loss_top_prob": 0.2768149822950363,
      "entropy": 5.841611266136169,
      "epoch": 1.3398692810457518,
      "grad_norm": 0.14392530918121338,
      "learning_rate": 1.3771057701853032e-05,
      "loss": 5.2368,
      "mean_token_accuracy": 0.4496015916765662,
      "num_tokens": 6124795.0,
      "step": 205
    },
    {
      "aux_loss_top_prob": 0.300910048186779,
      "entropy": 5.627748250961304,
      "epoch": 1.34640522875817,
      "grad_norm": 0.15599419176578522,
      "learning_rate": 1.3529750299849738e-05,
      "loss": 5.1989,
      "mean_token_accuracy": 0.4685505474162256,
      "num_tokens": 6155870.0,
      "step": 206
    },
    {
      "aux_loss_top_prob": 0.2739686332643032,
      "entropy": 5.819380044937134,
      "epoch": 1.3529411764705883,
      "grad_norm": 0.10587356239557266,
      "learning_rate": 1.328978898250525e-05,
      "loss": 5.3635,
      "mean_token_accuracy": 0.4327491423778534,
      "num_tokens": 6186391.0,
      "step": 207
    },
    {
      "aux_loss_top_prob": 0.2944241762161255,
      "entropy": 5.690333962440491,
      "epoch": 1.3594771241830066,
      "grad_norm": 0.09930114448070526,
      "learning_rate": 1.3051201910344089e-05,
      "loss": 5.181,
      "mean_token_accuracy": 0.46289401498427085,
      "num_tokens": 6217306.0,
      "step": 208
    },
    {
      "aux_loss_top_prob": 0.2546944059431553,
      "entropy": 6.1714030504226685,
      "epoch": 1.3660130718954249,
      "grad_norm": 0.3597683608531952,
      "learning_rate": 1.2814017082617025e-05,
      "loss": 5.2575,
      "mean_token_accuracy": 0.43351696474583346,
      "num_tokens": 6240376.0,
      "step": 209
    },
    {
      "aux_loss_top_prob": 0.27523647248744965,
      "entropy": 5.888242483139038,
      "epoch": 1.3725490196078431,
      "grad_norm": 0.10635742545127869,
      "learning_rate": 1.25782623340152e-05,
      "loss": 5.3712,
      "mean_token_accuracy": 0.43651786953071225,
      "num_tokens": 6270613.0,
      "step": 210
    },
    {
      "aux_loss_top_prob": 0.30731917172670364,
      "entropy": 5.579526424407959,
      "epoch": 1.3790849673202614,
      "grad_norm": 0.10765237361192703,
      "learning_rate": 1.2343965331403651e-05,
      "loss": 5.0764,
      "mean_token_accuracy": 0.4873237692726303,
      "num_tokens": 6300891.0,
      "step": 211
    },
    {
      "aux_loss_top_prob": 0.2886655703186989,
      "entropy": 5.741968274116516,
      "epoch": 1.3856209150326797,
      "grad_norm": 0.10023419559001923,
      "learning_rate": 1.2111153570574454e-05,
      "loss": 5.225,
      "mean_token_accuracy": 0.4580497412221804,
      "num_tokens": 6331603.0,
      "step": 212
    },
    {
      "aux_loss_top_prob": 0.2787569463253021,
      "entropy": 5.714465856552124,
      "epoch": 1.392156862745098,
      "grad_norm": 0.16637979447841644,
      "learning_rate": 1.1879854373019988e-05,
      "loss": 5.3326,
      "mean_token_accuracy": 0.4362636886189561,
      "num_tokens": 6363146.0,
      "step": 213
    },
    {
      "aux_loss_top_prob": 0.2567136026918888,
      "entropy": 6.032349944114685,
      "epoch": 1.3986928104575163,
      "grad_norm": 0.12060339003801346,
      "learning_rate": 1.1650094882726598e-05,
      "loss": 5.6011,
      "mean_token_accuracy": 0.4046537975802545,
      "num_tokens": 6394164.0,
      "step": 214
    },
    {
      "aux_loss_top_prob": 0.2645033597946167,
      "entropy": 5.894779920578003,
      "epoch": 1.4052287581699345,
      "grad_norm": 0.12390998005867004,
      "learning_rate": 1.1421902062989179e-05,
      "loss": 5.48,
      "mean_token_accuracy": 0.41729506815017925,
      "num_tokens": 6424957.0,
      "step": 215
    },
    {
      "aux_loss_top_prob": 0.27425268292427063,
      "entropy": 5.835648059844971,
      "epoch": 1.4117647058823528,
      "grad_norm": 0.10449754446744919,
      "learning_rate": 1.1195302693246878e-05,
      "loss": 5.3294,
      "mean_token_accuracy": 0.43852067880267764,
      "num_tokens": 6454126.0,
      "step": 216
    },
    {
      "aux_loss_top_prob": 0.21997729316353798,
      "entropy": 6.757992744445801,
      "epoch": 1.4183006535947713,
      "grad_norm": 0.6302598714828491,
      "learning_rate": 1.0970323365940444e-05,
      "loss": 4.7214,
      "mean_token_accuracy": 0.3884124912300632,
      "num_tokens": 6469667.0,
      "step": 217
    },
    {
      "aux_loss_top_prob": 0.26251307129859924,
      "entropy": 5.902698993682861,
      "epoch": 1.4248366013071896,
      "grad_norm": 0.11810523271560669,
      "learning_rate": 1.0746990483391415e-05,
      "loss": 5.468,
      "mean_token_accuracy": 0.4146786391657097,
      "num_tokens": 6500355.0,
      "step": 218
    },
    {
      "aux_loss_top_prob": 0.2304418757557869,
      "entropy": 6.456882119178772,
      "epoch": 1.4313725490196079,
      "grad_norm": 0.7557706832885742,
      "learning_rate": 1.052533025470379e-05,
      "loss": 4.997,
      "mean_token_accuracy": 0.3318053193959388,
      "num_tokens": 6522558.0,
      "step": 219
    },
    {
      "aux_loss_top_prob": 0.2926318496465683,
      "entropy": 5.5905855894088745,
      "epoch": 1.4379084967320261,
      "grad_norm": 0.2737779915332794,
      "learning_rate": 1.0305368692688174e-05,
      "loss": 5.275,
      "mean_token_accuracy": 0.4442215658844073,
      "num_tokens": 6554012.0,
      "step": 220
    },
    {
      "aux_loss_top_prob": 0.3002009391784668,
      "entropy": 5.438734531402588,
      "epoch": 1.4444444444444444,
      "grad_norm": 0.3349403142929077,
      "learning_rate": 1.0087131610809153e-05,
      "loss": 5.179,
      "mean_token_accuracy": 0.4560533214153689,
      "num_tokens": 6584257.0,
      "step": 221
    },
    {
      "aux_loss_top_prob": 0.27831168100237846,
      "entropy": 5.735527753829956,
      "epoch": 1.4509803921568627,
      "grad_norm": 0.2199510782957077,
      "learning_rate": 9.870644620155877e-06,
      "loss": 5.3952,
      "mean_token_accuracy": 0.4301956686785762,
      "num_tokens": 6614883.0,
      "step": 222
    },
    {
      "aux_loss_top_prob": 0.281936377286911,
      "entropy": 5.802496075630188,
      "epoch": 1.457516339869281,
      "grad_norm": 0.12029431760311127,
      "learning_rate": 9.655933126436564e-06,
      "loss": 5.3448,
      "mean_token_accuracy": 0.4400415855158149,
      "num_tokens": 6645317.0,
      "step": 223
    },
    {
      "aux_loss_top_prob": 0.28136735782027245,
      "entropy": 5.806631684303284,
      "epoch": 1.4640522875816995,
      "grad_norm": 0.11811744421720505,
      "learning_rate": 9.443022326996984e-06,
      "loss": 5.3234,
      "mean_token_accuracy": 0.44566655195146876,
      "num_tokens": 6675326.0,
      "step": 224
    },
    {
      "aux_loss_top_prob": 0.2804422117769718,
      "entropy": 5.725481748580933,
      "epoch": 1.4705882352941178,
      "grad_norm": 0.15430091321468353,
      "learning_rate": 9.231937207863459e-06,
      "loss": 5.318,
      "mean_token_accuracy": 0.43893657470761516,
      "num_tokens": 6706131.0,
      "step": 225
    },
    {
      "aux_loss_top_prob": 0.28404561430215836,
      "entropy": 5.754128932952881,
      "epoch": 1.477124183006536,
      "grad_norm": 0.11856935918331146,
      "learning_rate": 9.022702540810607e-06,
      "loss": 5.2953,
      "mean_token_accuracy": 0.4503447683942945,
      "num_tokens": 6736447.0,
      "step": 226
    },
    {
      "aux_loss_top_prob": 0.28312472254037857,
      "entropy": 5.765999674797058,
      "epoch": 1.4836601307189543,
      "grad_norm": 0.10468044131994247,
      "learning_rate": 8.815342880454311e-06,
      "loss": 5.2832,
      "mean_token_accuracy": 0.4497058070022713,
      "num_tokens": 6767134.0,
      "step": 227
    },
    {
      "aux_loss_top_prob": 0.27276143059134483,
      "entropy": 5.9160462617874146,
      "epoch": 1.4901960784313726,
      "grad_norm": 0.15895818173885345,
      "learning_rate": 8.6098825613701e-06,
      "loss": 5.3389,
      "mean_token_accuracy": 0.4400420864826853,
      "num_tokens": 6798329.0,
      "step": 228
    },
    {
      "aux_loss_top_prob": 0.2809566482901573,
      "entropy": 5.76944887638092,
      "epoch": 1.4967320261437909,
      "grad_norm": 0.10363449901342392,
      "learning_rate": 8.406345695237393e-06,
      "loss": 5.2433,
      "mean_token_accuracy": 0.4481067941917617,
      "num_tokens": 6828876.0,
      "step": 229
    },
    {
      "aux_loss_top_prob": 0.3016709089279175,
      "entropy": 5.540647745132446,
      "epoch": 1.5032679738562091,
      "grad_norm": 0.13936801254749298,
      "learning_rate": 8.20475616800985e-06,
      "loss": 5.0706,
      "mean_token_accuracy": 0.47480149010596345,
      "num_tokens": 6858300.0,
      "step": 230
    },
    {
      "aux_loss_top_prob": 0.2883935496211052,
      "entropy": 5.739317059516907,
      "epoch": 1.5098039215686274,
      "grad_norm": 0.17810212075710297,
      "learning_rate": 8.005137637112303e-06,
      "loss": 5.0285,
      "mean_token_accuracy": 0.4709248419383637,
      "num_tokens": 6883256.0,
      "step": 231
    },
    {
      "aux_loss_top_prob": 0.2829754576086998,
      "entropy": 5.840222477912903,
      "epoch": 1.5163398692810457,
      "grad_norm": 0.13936778903007507,
      "learning_rate": 7.807513528664414e-06,
      "loss": 5.2376,
      "mean_token_accuracy": 0.4567169447665652,
      "num_tokens": 6913091.0,
      "step": 232
    },
    {
      "aux_loss_top_prob": 0.3014686331152916,
      "entropy": 5.6900246143341064,
      "epoch": 1.522875816993464,
      "grad_norm": 0.14062733948230743,
      "learning_rate": 7.611907034731539e-06,
      "loss": 5.0783,
      "mean_token_accuracy": 0.4843632068798777,
      "num_tokens": 6943006.0,
      "step": 233
    },
    {
      "aux_loss_top_prob": 0.27456242963671684,
      "entropy": 5.902594566345215,
      "epoch": 1.5294117647058822,
      "grad_norm": 0.1538599282503128,
      "learning_rate": 7.418341110603e-06,
      "loss": 5.3131,
      "mean_token_accuracy": 0.4428775787583682,
      "num_tokens": 6973577.0,
      "step": 234
    },
    {
      "aux_loss_top_prob": 0.2886464484035969,
      "entropy": 5.729029178619385,
      "epoch": 1.5359477124183005,
      "grad_norm": 0.10246621817350388,
      "learning_rate": 7.226838472098238e-06,
      "loss": 5.2206,
      "mean_token_accuracy": 0.46003355182197725,
      "num_tokens": 7003970.0,
      "step": 235
    },
    {
      "aux_loss_top_prob": 0.2903565391898155,
      "entropy": 5.728077054023743,
      "epoch": 1.5424836601307188,
      "grad_norm": 0.12496346235275269,
      "learning_rate": 7.037421592900942e-06,
      "loss": 5.164,
      "mean_token_accuracy": 0.46285299087205745,
      "num_tokens": 7034503.0,
      "step": 236
    },
    {
      "aux_loss_top_prob": 0.28790057450532913,
      "entropy": 5.794499516487122,
      "epoch": 1.5490196078431373,
      "grad_norm": 0.16668422520160675,
      "learning_rate": 6.8501127019217346e-06,
      "loss": 5.1799,
      "mean_token_accuracy": 0.4693444884714622,
      "num_tokens": 7065916.0,
      "step": 237
    },
    {
      "aux_loss_top_prob": 0.21127384342253208,
      "entropy": 6.963577747344971,
      "epoch": 1.5555555555555556,
      "grad_norm": 0.8092356324195862,
      "learning_rate": 6.664933780689445e-06,
      "loss": 4.8121,
      "mean_token_accuracy": 0.4072748088353719,
      "num_tokens": 7080666.0,
      "step": 238
    },
    {
      "aux_loss_top_prob": 0.28685609996318817,
      "entropy": 5.664153218269348,
      "epoch": 1.5620915032679739,
      "grad_norm": 0.1706911325454712,
      "learning_rate": 6.481906560771525e-06,
      "loss": 5.2601,
      "mean_token_accuracy": 0.45253409025587366,
      "num_tokens": 7110944.0,
      "step": 239
    },
    {
      "aux_loss_top_prob": 0.2834233418107033,
      "entropy": 5.696601390838623,
      "epoch": 1.5686274509803921,
      "grad_norm": 0.11385975778102875,
      "learning_rate": 6.301052521223735e-06,
      "loss": 5.2219,
      "mean_token_accuracy": 0.4524409609911124,
      "num_tokens": 7141418.0,
      "step": 240
    },
    {
      "aux_loss_top_prob": 0.2827957645058632,
      "entropy": 5.756961941719055,
      "epoch": 1.5751633986928104,
      "grad_norm": 0.32700660824775696,
      "learning_rate": 6.122392886069486e-06,
      "loss": 5.254,
      "mean_token_accuracy": 0.45220576065938156,
      "num_tokens": 7171589.0,
      "step": 241
    },
    {
      "aux_loss_top_prob": 0.26048360019922256,
      "entropy": 5.9918729066848755,
      "epoch": 1.581699346405229,
      "grad_norm": 0.13625770807266235,
      "learning_rate": 5.945948621809091e-06,
      "loss": 5.5275,
      "mean_token_accuracy": 0.41320437610446953,
      "num_tokens": 7202592.0,
      "step": 242
    },
    {
      "aux_loss_top_prob": 0.24250251799821854,
      "entropy": 6.481173038482666,
      "epoch": 1.5882352941176472,
      "grad_norm": 0.5599669218063354,
      "learning_rate": 5.771740434959278e-06,
      "loss": 5.1925,
      "mean_token_accuracy": 0.43364270512751313,
      "num_tokens": 7218376.0,
      "step": 243
    },
    {
      "aux_loss_top_prob": 0.30172795057296753,
      "entropy": 5.441320538520813,
      "epoch": 1.5947712418300655,
      "grad_norm": 0.29380130767822266,
      "learning_rate": 5.5997887696231735e-06,
      "loss": 5.1347,
      "mean_token_accuracy": 0.46140535806705696,
      "num_tokens": 7249777.0,
      "step": 244
    },
    {
      "aux_loss_top_prob": 0.23225780203938484,
      "entropy": 6.40979540348053,
      "epoch": 1.6013071895424837,
      "grad_norm": 0.3881549835205078,
      "learning_rate": 5.430113805091111e-06,
      "loss": 5.0605,
      "mean_token_accuracy": 0.3645892205434231,
      "num_tokens": 7272344.0,
      "step": 245
    },
    {
      "aux_loss_top_prob": 0.2863045856356621,
      "entropy": 5.740433692932129,
      "epoch": 1.607843137254902,
      "grad_norm": 0.11012943089008331,
      "learning_rate": 5.262735453472459e-06,
      "loss": 5.2597,
      "mean_token_accuracy": 0.45309580936409893,
      "num_tokens": 7302346.0,
      "step": 246
    },
    {
      "aux_loss_top_prob": 0.2888062372803688,
      "entropy": 5.63741934299469,
      "epoch": 1.6143790849673203,
      "grad_norm": 0.19898571074008942,
      "learning_rate": 5.097673357358907e-06,
      "loss": 5.2689,
      "mean_token_accuracy": 0.4523140972296816,
      "num_tokens": 7332868.0,
      "step": 247
    },
    {
      "aux_loss_top_prob": 0.30033355206251144,
      "entropy": 5.54612922668457,
      "epoch": 1.6209150326797386,
      "grad_norm": 0.19067813456058502,
      "learning_rate": 4.934946887519279e-06,
      "loss": 5.1408,
      "mean_token_accuracy": 0.46576083706558363,
      "num_tokens": 7364141.0,
      "step": 248
    },
    {
      "aux_loss_top_prob": 0.29766593128442764,
      "entropy": 5.563416600227356,
      "epoch": 1.6274509803921569,
      "grad_norm": 0.22993546724319458,
      "learning_rate": 4.7745751406263165e-06,
      "loss": 5.2061,
      "mean_token_accuracy": 0.4605940654021242,
      "num_tokens": 7394849.0,
      "step": 249
    },
    {
      "aux_loss_top_prob": 0.29042842984199524,
      "entropy": 5.569738864898682,
      "epoch": 1.6339869281045751,
      "grad_norm": 0.2654423117637634,
      "learning_rate": 4.616576937015551e-06,
      "loss": 5.2511,
      "mean_token_accuracy": 0.4471975956053214,
      "num_tokens": 7425593.0,
      "step": 250
    },
    {
      "aux_loss_top_prob": 0.3035343065857887,
      "entropy": 5.734548449516296,
      "epoch": 1.6405228758169934,
      "grad_norm": 0.39061594009399414,
      "learning_rate": 4.460970818476718e-06,
      "loss": 5.0648,
      "mean_token_accuracy": 0.48812669030305145,
      "num_tokens": 7449599.0,
      "step": 251
    },
    {
      "aux_loss_top_prob": 0.2750558629631996,
      "entropy": 5.776683568954468,
      "epoch": 1.6470588235294117,
      "grad_norm": 0.17301222681999207,
      "learning_rate": 4.307775046077739e-06,
      "loss": 5.4022,
      "mean_token_accuracy": 0.427896966370192,
      "num_tokens": 7480414.0,
      "step": 252
    },
    {
      "aux_loss_top_prob": 0.30104003101587296,
      "entropy": 5.490870356559753,
      "epoch": 1.65359477124183,
      "grad_norm": 0.22275079786777496,
      "learning_rate": 4.15700759802175e-06,
      "loss": 5.1077,
      "mean_token_accuracy": 0.4689421850501184,
      "num_tokens": 7511412.0,
      "step": 253
    },
    {
      "aux_loss_top_prob": 0.3032931461930275,
      "entropy": 5.55763566493988,
      "epoch": 1.6601307189542482,
      "grad_norm": 0.12481661885976791,
      "learning_rate": 4.008686167537243e-06,
      "loss": 5.064,
      "mean_token_accuracy": 0.47837400578440853,
      "num_tokens": 7542078.0,
      "step": 254
    },
    {
      "aux_loss_top_prob": 0.2732524275779724,
      "entropy": 5.796776175498962,
      "epoch": 1.6666666666666665,
      "grad_norm": 0.10881564021110535,
      "learning_rate": 3.862828160801707e-06,
      "loss": 5.2981,
      "mean_token_accuracy": 0.43391887273782787,
      "num_tokens": 7573160.0,
      "step": 255
    },
    {
      "aux_loss_top_prob": 0.26295750588178635,
      "entropy": 6.001337885856628,
      "epoch": 1.673202614379085,
      "grad_norm": 0.13999661803245544,
      "learning_rate": 3.7194506948989406e-06,
      "loss": 5.446,
      "mean_token_accuracy": 0.42783732446455985,
      "num_tokens": 7603517.0,
      "step": 256
    },
    {
      "aux_loss_top_prob": 0.2949695438146591,
      "entropy": 5.587567925453186,
      "epoch": 1.6797385620915033,
      "grad_norm": 0.15612345933914185,
      "learning_rate": 3.578570595810274e-06,
      "loss": 5.17,
      "mean_token_accuracy": 0.4642786041864468,
      "num_tokens": 7633620.0,
      "step": 257
    },
    {
      "aux_loss_top_prob": 0.29168836772441864,
      "entropy": 5.70826256275177,
      "epoch": 1.6862745098039216,
      "grad_norm": 0.11891917884349823,
      "learning_rate": 3.4402043964399527e-06,
      "loss": 5.1241,
      "mean_token_accuracy": 0.4697517899667902,
      "num_tokens": 7664535.0,
      "step": 258
    },
    {
      "aux_loss_top_prob": 0.2646394334733486,
      "entropy": 6.00098991394043,
      "epoch": 1.6928104575163399,
      "grad_norm": 0.1710331290960312,
      "learning_rate": 3.3043683346749647e-06,
      "loss": 5.4344,
      "mean_token_accuracy": 0.4268672311519721,
      "num_tokens": 7694994.0,
      "step": 259
    },
    {
      "aux_loss_top_prob": 0.2491995245218277,
      "entropy": 6.350843667984009,
      "epoch": 1.6993464052287581,
      "grad_norm": 0.3792577385902405,
      "learning_rate": 3.1710783514794257e-06,
      "loss": 5.2944,
      "mean_token_accuracy": 0.42888093581318854,
      "num_tokens": 7718149.0,
      "step": 260
    },
    {
      "aux_loss_top_prob": 0.2524934597313404,
      "entropy": 6.3930100202560425,
      "epoch": 1.7058823529411766,
      "grad_norm": 0.3932129144668579,
      "learning_rate": 3.040350089023844e-06,
      "loss": 4.9908,
      "mean_token_accuracy": 0.42401203400356724,
      "num_tokens": 7741137.0,
      "step": 261
    },
    {
      "aux_loss_top_prob": 0.2684704586863518,
      "entropy": 5.847909569740295,
      "epoch": 1.712418300653595,
      "grad_norm": 0.1097669005393982,
      "learning_rate": 2.9121988888494297e-06,
      "loss": 5.3535,
      "mean_token_accuracy": 0.42689535374785337,
      "num_tokens": 7772023.0,
      "step": 262
    },
    {
      "aux_loss_top_prob": 0.27034761011600494,
      "entropy": 6.010856628417969,
      "epoch": 1.7189542483660132,
      "grad_norm": 0.25189605355262756,
      "learning_rate": 2.786639790067719e-06,
      "loss": 5.3355,
      "mean_token_accuracy": 0.4439473556248371,
      "num_tokens": 7803193.0,
      "step": 263
    },
    {
      "aux_loss_top_prob": 0.27195480465888977,
      "entropy": 5.850176811218262,
      "epoch": 1.7254901960784315,
      "grad_norm": 0.12924204766750336,
      "learning_rate": 2.663687527595657e-06,
      "loss": 5.2922,
      "mean_token_accuracy": 0.43906703360068716,
      "num_tokens": 7833449.0,
      "step": 264
    },
    {
      "aux_loss_top_prob": 0.2837466448545456,
      "entropy": 5.7709022760391235,
      "epoch": 1.7320261437908497,
      "grad_norm": 0.13096733391284943,
      "learning_rate": 2.543356530426394e-06,
      "loss": 5.2052,
      "mean_token_accuracy": 0.4522660403327277,
      "num_tokens": 7863878.0,
      "step": 265
    },
    {
      "aux_loss_top_prob": 0.27897416800260544,
      "entropy": 5.793304681777954,
      "epoch": 1.738562091503268,
      "grad_norm": 0.13209740817546844,
      "learning_rate": 2.425660919935954e-06,
      "loss": 5.2141,
      "mean_token_accuracy": 0.4519914775579275,
      "num_tokens": 7894539.0,
      "step": 266
    },
    {
      "aux_loss_top_prob": 0.28243637830018997,
      "entropy": 5.826195001602173,
      "epoch": 1.7450980392156863,
      "grad_norm": 0.23233957588672638,
      "learning_rate": 2.310614508226078e-06,
      "loss": 5.2637,
      "mean_token_accuracy": 0.4496695810123375,
      "num_tokens": 7925376.0,
      "step": 267
    },
    {
      "aux_loss_top_prob": 0.23658869415521622,
      "entropy": 6.356955409049988,
      "epoch": 1.7516339869281046,
      "grad_norm": 0.23825743794441223,
      "learning_rate": 2.198230796503256e-06,
      "loss": 5.4112,
      "mean_token_accuracy": 0.40131729146281847,
      "num_tokens": 7948504.0,
      "step": 268
    },
    {
      "aux_loss_top_prob": 0.26828988641500473,
      "entropy": 5.885715126991272,
      "epoch": 1.7581699346405228,
      "grad_norm": 0.10782335698604584,
      "learning_rate": 2.0885229734943502e-06,
      "loss": 5.3596,
      "mean_token_accuracy": 0.4325719984446089,
      "num_tokens": 7979321.0,
      "step": 269
    },
    {
      "aux_loss_top_prob": 0.26850707456469536,
      "entropy": 5.78294050693512,
      "epoch": 1.7647058823529411,
      "grad_norm": 0.14268767833709717,
      "learning_rate": 1.9815039138988135e-06,
      "loss": 5.3807,
      "mean_token_accuracy": 0.4199526438410289,
      "num_tokens": 8010334.0,
      "step": 270
    },
    {
      "aux_loss_top_prob": 0.2814105600118637,
      "entropy": 5.77380907535553,
      "epoch": 1.7712418300653594,
      "grad_norm": 0.1044280007481575,
      "learning_rate": 1.8771861768777792e-06,
      "loss": 5.2499,
      "mean_token_accuracy": 0.4497251254838286,
      "num_tokens": 8041098.0,
      "step": 271
    },
    {
      "aux_loss_top_prob": 0.27226322144269943,
      "entropy": 5.848519802093506,
      "epoch": 1.7777777777777777,
      "grad_norm": 0.11296819150447845,
      "learning_rate": 1.7755820045802145e-06,
      "loss": 5.3353,
      "mean_token_accuracy": 0.43767801686252855,
      "num_tokens": 8071635.0,
      "step": 272
    },
    {
      "aux_loss_top_prob": 0.2992687448859215,
      "entropy": 5.632307291030884,
      "epoch": 1.784313725490196,
      "grad_norm": 0.10923735797405243,
      "learning_rate": 1.67670332070623e-06,
      "loss": 5.1175,
      "mean_token_accuracy": 0.47929374124394913,
      "num_tokens": 8101273.0,
      "step": 273
    },
    {
      "aux_loss_top_prob": 0.2710805758833885,
      "entropy": 5.8428720235824585,
      "epoch": 1.7908496732026142,
      "grad_norm": 0.10545022040605545,
      "learning_rate": 1.5805617291077768e-06,
      "loss": 5.3655,
      "mean_token_accuracy": 0.42581676544762764,
      "num_tokens": 8132291.0,
      "step": 274
    },
    {
      "aux_loss_top_prob": 0.2828085534274578,
      "entropy": 5.743242502212524,
      "epoch": 1.7973856209150327,
      "grad_norm": 0.1612960398197174,
      "learning_rate": 1.4871685124269008e-06,
      "loss": 5.3443,
      "mean_token_accuracy": 0.4412462906258824,
      "num_tokens": 8163713.0,
      "step": 275
    },
    {
      "aux_loss_top_prob": 0.2768475115299225,
      "entropy": 5.767160177230835,
      "epoch": 1.803921568627451,
      "grad_norm": 0.12148293852806091,
      "learning_rate": 1.3965346307716675e-06,
      "loss": 5.26,
      "mean_token_accuracy": 0.44290794081979257,
      "num_tokens": 8195007.0,
      "step": 276
    },
    {
      "aux_loss_top_prob": 0.2839002460241318,
      "entropy": 5.67188036441803,
      "epoch": 1.8104575163398693,
      "grad_norm": 0.2284420132637024,
      "learning_rate": 1.3086707204299414e-06,
      "loss": 5.2926,
      "mean_token_accuracy": 0.4409179993702006,
      "num_tokens": 8226054.0,
      "step": 277
    },
    {
      "aux_loss_top_prob": 0.29005027562379837,
      "entropy": 5.631163120269775,
      "epoch": 1.8169934640522876,
      "grad_norm": 0.16713319718837738,
      "learning_rate": 1.2235870926211619e-06,
      "loss": 5.2295,
      "mean_token_accuracy": 0.4555899856200759,
      "num_tokens": 8255721.0,
      "step": 278
    },
    {
      "aux_loss_top_prob": 0.28684523701667786,
      "entropy": 5.69948148727417,
      "epoch": 1.8235294117647058,
      "grad_norm": 0.1433834582567215,
      "learning_rate": 1.141293732286297e-06,
      "loss": 5.2634,
      "mean_token_accuracy": 0.4483790314211955,
      "num_tokens": 8285725.0,
      "step": 279
    },
    {
      "aux_loss_top_prob": 0.25583869218826294,
      "entropy": 6.006671786308289,
      "epoch": 1.8300653594771243,
      "grad_norm": 0.111579030752182,
      "learning_rate": 1.0618002969160544e-06,
      "loss": 5.539,
      "mean_token_accuracy": 0.4072580889401174,
      "num_tokens": 8316334.0,
      "step": 280
    },
    {
      "aux_loss_top_prob": 0.28158367425203323,
      "entropy": 5.809409856796265,
      "epoch": 1.8366013071895426,
      "grad_norm": 0.11823819577693939,
      "learning_rate": 9.851161154175337e-07,
      "loss": 5.2483,
      "mean_token_accuracy": 0.44901155750431615,
      "num_tokens": 8347669.0,
      "step": 281
    },
    {
      "aux_loss_top_prob": 0.23087788000702858,
      "entropy": 6.540221214294434,
      "epoch": 1.843137254901961,
      "grad_norm": 0.34318944811820984,
      "learning_rate": 9.112501870194273e-07,
      "loss": 5.4637,
      "mean_token_accuracy": 0.39349939489159863,
      "num_tokens": 8371308.0,
      "step": 282
    },
    {
      "aux_loss_top_prob": 0.28495974093675613,
      "entropy": 5.718922734260559,
      "epoch": 1.8496732026143792,
      "grad_norm": 0.11739230901002884,
      "learning_rate": 8.402111802159412e-07,
      "loss": 5.2489,
      "mean_token_accuracy": 0.44704093716697146,
      "num_tokens": 8401603.0,
      "step": 283
    },
    {
      "aux_loss_top_prob": 0.27856648713350296,
      "entropy": 5.845695734024048,
      "epoch": 1.8562091503267975,
      "grad_norm": 0.10946676135063171,
      "learning_rate": 7.720074317494913e-07,
      "loss": 5.3426,
      "mean_token_accuracy": 0.44350032090329883,
      "num_tokens": 8432205.0,
      "step": 284
    },
    {
      "aux_loss_top_prob": 0.29073021188378334,
      "entropy": 5.774864912033081,
      "epoch": 1.8627450980392157,
      "grad_norm": 0.12754598259925842,
      "learning_rate": 7.06646945632361e-07,
      "loss": 5.3145,
      "mean_token_accuracy": 0.4579413572742992,
      "num_tokens": 8464030.0,
      "step": 285
    },
    {
      "aux_loss_top_prob": 0.297753244638443,
      "entropy": 5.613563179969788,
      "epoch": 1.869281045751634,
      "grad_norm": 0.1126393973827362,
      "learning_rate": 6.441373922073945e-07,
      "loss": 5.1227,
      "mean_token_accuracy": 0.47021022694903203,
      "num_tokens": 8494775.0,
      "step": 286
    },
    {
      "aux_loss_top_prob": 0.261444553732872,
      "entropy": 5.884864091873169,
      "epoch": 1.8758169934640523,
      "grad_norm": 0.11973244696855545,
      "learning_rate": 5.844861072478336e-07,
      "loss": 5.4703,
      "mean_token_accuracy": 0.41280821187597405,
      "num_tokens": 8525647.0,
      "step": 287
    },
    {
      "aux_loss_top_prob": 0.28679419681429863,
      "entropy": 5.7279897928237915,
      "epoch": 1.8823529411764706,
      "grad_norm": 0.1341501623392105,
      "learning_rate": 5.27700091096453e-07,
      "loss": 5.2816,
      "mean_token_accuracy": 0.45094768382649847,
      "num_tokens": 8555964.0,
      "step": 288
    },
    {
      "aux_loss_top_prob": 0.25558295100927353,
      "entropy": 6.062047243118286,
      "epoch": 1.8888888888888888,
      "grad_norm": 0.1221308708190918,
      "learning_rate": 4.7378600784402093e-07,
      "loss": 5.5502,
      "mean_token_accuracy": 0.41198137279268116,
      "num_tokens": 8587190.0,
      "step": 289
    },
    {
      "aux_loss_top_prob": 0.2740677446126938,
      "entropy": 5.813249111175537,
      "epoch": 1.8954248366013071,
      "grad_norm": 0.12060526758432388,
      "learning_rate": 4.2275018454725203e-07,
      "loss": 5.3765,
      "mean_token_accuracy": 0.4305072224139843,
      "num_tokens": 8617856.0,
      "step": 290
    },
    {
      "aux_loss_top_prob": 0.28896818310022354,
      "entropy": 5.717164158821106,
      "epoch": 1.9019607843137254,
      "grad_norm": 0.12941595911979675,
      "learning_rate": 3.745986104862903e-07,
      "loss": 5.1343,
      "mean_token_accuracy": 0.4675023437426693,
      "num_tokens": 8646768.0,
      "step": 291
    },
    {
      "aux_loss_top_prob": 0.27895141392946243,
      "entropy": 5.743113279342651,
      "epoch": 1.9084967320261437,
      "grad_norm": 0.1239534243941307,
      "learning_rate": 3.293369364618465e-07,
      "loss": 5.2961,
      "mean_token_accuracy": 0.4390467134215964,
      "num_tokens": 8677666.0,
      "step": 292
    },
    {
      "aux_loss_top_prob": 0.28562700748443604,
      "entropy": 5.787073254585266,
      "epoch": 1.915032679738562,
      "grad_norm": 0.1021827980875969,
      "learning_rate": 2.869704741320478e-07,
      "loss": 5.2555,
      "mean_token_accuracy": 0.45694924348088584,
      "num_tokens": 8707683.0,
      "step": 293
    },
    {
      "aux_loss_top_prob": 0.2357088141143322,
      "entropy": 6.399084210395813,
      "epoch": 1.9215686274509802,
      "grad_norm": 0.34307026863098145,
      "learning_rate": 2.4750419538908663e-07,
      "loss": 5.1942,
      "mean_token_accuracy": 0.39500068153038576,
      "num_tokens": 8730717.0,
      "step": 294
    },
    {
      "aux_loss_top_prob": 0.27967261150479317,
      "entropy": 5.73197066783905,
      "epoch": 1.9281045751633987,
      "grad_norm": 0.10405728965997696,
      "learning_rate": 2.1094273177576507e-07,
      "loss": 5.2334,
      "mean_token_accuracy": 0.44409212777565893,
      "num_tokens": 8761531.0,
      "step": 295
    },
    {
      "aux_loss_top_prob": 0.28506751358509064,
      "entropy": 5.708837628364563,
      "epoch": 1.934640522875817,
      "grad_norm": 0.1517622470855713,
      "learning_rate": 1.7729037394193793e-07,
      "loss": 5.1532,
      "mean_token_accuracy": 0.4563712468464015,
      "num_tokens": 8791846.0,
      "step": 296
    },
    {
      "aux_loss_top_prob": 0.29964617639780045,
      "entropy": 5.602272272109985,
      "epoch": 1.9411764705882353,
      "grad_norm": 0.10091777890920639,
      "learning_rate": 1.4655107114101007e-07,
      "loss": 5.078,
      "mean_token_accuracy": 0.4776847157290369,
      "num_tokens": 8823107.0,
      "step": 297
    },
    {
      "aux_loss_top_prob": 0.24369057826697826,
      "entropy": 6.384117364883423,
      "epoch": 1.9477124183006536,
      "grad_norm": 0.44275346398353577,
      "learning_rate": 1.1872843076645157e-07,
      "loss": 4.9468,
      "mean_token_accuracy": 0.3807038111473259,
      "num_tokens": 8846273.0,
      "step": 298
    },
    {
      "aux_loss_top_prob": 0.288282785564661,
      "entropy": 5.763324975967407,
      "epoch": 1.954248366013072,
      "grad_norm": 0.2068706452846527,
      "learning_rate": 9.382571792846961e-08,
      "loss": 5.2902,
      "mean_token_accuracy": 0.45460217484126403,
      "num_tokens": 8877244.0,
      "step": 299
    },
    {
      "aux_loss_top_prob": 0.2847488448023796,
      "entropy": 5.740633249282837,
      "epoch": 1.9607843137254903,
      "grad_norm": 0.10678768903017044,
      "learning_rate": 7.184585507082897e-08,
      "loss": 5.1838,
      "mean_token_accuracy": 0.46031675996402166,
      "num_tokens": 8907724.0,
      "step": 300
    },
    {
      "aux_loss_top_prob": 0.2850300073623657,
      "entropy": 5.659717321395874,
      "epoch": 1.9673202614379086,
      "grad_norm": 0.16874495148658752,
      "learning_rate": 5.279142162789019e-08,
      "loss": 5.2662,
      "mean_token_accuracy": 0.44445266431964664,
      "num_tokens": 8936861.0,
      "step": 301
    },
    {
      "aux_loss_top_prob": 0.2974099814891815,
      "entropy": 5.543270945549011,
      "epoch": 1.973856209150327,
      "grad_norm": 0.24483275413513184,
      "learning_rate": 3.666465372190453e-08,
      "loss": 5.1941,
      "mean_token_accuracy": 0.4614864787288934,
      "num_tokens": 8967693.0,
      "step": 302
    },
    {
      "aux_loss_top_prob": 0.30284444242715836,
      "entropy": 5.544885277748108,
      "epoch": 1.9803921568627452,
      "grad_norm": 0.1883031278848648,
      "learning_rate": 2.3467443900582198e-08,
      "loss": 5.1352,
      "mean_token_accuracy": 0.4743697105702206,
      "num_tokens": 8998018.0,
      "step": 303
    },
    {
      "aux_loss_top_prob": 0.28624632209539413,
      "entropy": 5.745443820953369,
      "epoch": 1.9869281045751634,
      "grad_norm": 0.10352321714162827,
      "learning_rate": 1.3201340915011683e-08,
      "loss": 5.2246,
      "mean_token_accuracy": 0.4549818011920458,
      "num_tokens": 9028228.0,
      "step": 304
    },
    {
      "aux_loss_top_prob": 0.27371224388480186,
      "entropy": 5.8770962953567505,
      "epoch": 1.9934640522875817,
      "grad_norm": 0.12151777744293213,
      "learning_rate": 5.86754953789681e-09,
      "loss": 5.3228,
      "mean_token_accuracy": 0.4387709594495228,
      "num_tokens": 9058428.0,
      "step": 305
    },
    {
      "aux_loss_top_prob": 0.2918546050786972,
      "entropy": 5.699133276939392,
      "epoch": 2.0,
      "grad_norm": 0.13471631705760956,
      "learning_rate": 1.4669304221726077e-09,
      "loss": 5.1723,
      "mean_token_accuracy": 0.4679350526488095,
      "num_tokens": 9088054.0,
      "step": 306
    }
  ],
  "logging_steps": 1,
  "max_steps": 306,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 76,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.7586827527566336e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}