{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 984,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003048780487804878,
      "grad_norm": 0.0672889974881053,
      "learning_rate": 5.050505050505052e-07,
      "loss": 0.5879,
      "step": 1
    },
    {
      "epoch": 0.006097560975609756,
      "grad_norm": 0.06658887990303611,
      "learning_rate": 1.0101010101010103e-06,
      "loss": 0.5344,
      "step": 2
    },
    {
      "epoch": 0.009146341463414634,
      "grad_norm": 0.06759856887025624,
      "learning_rate": 1.5151515151515152e-06,
      "loss": 0.5734,
      "step": 3
    },
    {
      "epoch": 0.012195121951219513,
      "grad_norm": 0.07151981947202596,
      "learning_rate": 2.0202020202020206e-06,
      "loss": 0.6133,
      "step": 4
    },
    {
      "epoch": 0.01524390243902439,
      "grad_norm": 0.07851977916778793,
      "learning_rate": 2.5252525252525253e-06,
      "loss": 0.5983,
      "step": 5
    },
    {
      "epoch": 0.018292682926829267,
      "grad_norm": 0.0800165592411649,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 0.6898,
      "step": 6
    },
    {
      "epoch": 0.021341463414634148,
      "grad_norm": 0.07309066641656121,
      "learning_rate": 3.5353535353535352e-06,
      "loss": 0.6453,
      "step": 7
    },
    {
      "epoch": 0.024390243902439025,
      "grad_norm": 0.07569404816722357,
      "learning_rate": 4.040404040404041e-06,
      "loss": 0.6345,
      "step": 8
    },
    {
      "epoch": 0.027439024390243903,
      "grad_norm": 0.07647277110891933,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.6177,
      "step": 9
    },
    {
      "epoch": 0.03048780487804878,
      "grad_norm": 0.07743239710231267,
      "learning_rate": 5.050505050505051e-06,
      "loss": 0.6586,
      "step": 10
    },
    {
      "epoch": 0.03353658536585366,
      "grad_norm": 0.06456324181839977,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.5362,
      "step": 11
    },
    {
      "epoch": 0.036585365853658534,
      "grad_norm": 0.07310464363002138,
      "learning_rate": 6.060606060606061e-06,
      "loss": 0.5606,
      "step": 12
    },
    {
      "epoch": 0.039634146341463415,
      "grad_norm": 0.06589675953784346,
      "learning_rate": 6.565656565656567e-06,
      "loss": 0.5444,
      "step": 13
    },
    {
      "epoch": 0.042682926829268296,
      "grad_norm": 0.06985336076319228,
      "learning_rate": 7.0707070707070704e-06,
      "loss": 0.5802,
      "step": 14
    },
    {
      "epoch": 0.04573170731707317,
      "grad_norm": 0.07816492456588642,
      "learning_rate": 7.5757575757575764e-06,
      "loss": 0.6828,
      "step": 15
    },
    {
      "epoch": 0.04878048780487805,
      "grad_norm": 0.08029559940720289,
      "learning_rate": 8.080808080808082e-06,
      "loss": 0.6571,
      "step": 16
    },
    {
      "epoch": 0.051829268292682924,
      "grad_norm": 0.0716618460526677,
      "learning_rate": 8.585858585858587e-06,
      "loss": 0.5627,
      "step": 17
    },
    {
      "epoch": 0.054878048780487805,
      "grad_norm": 0.07391850146450858,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.5875,
      "step": 18
    },
    {
      "epoch": 0.057926829268292686,
      "grad_norm": 0.0730095019246016,
      "learning_rate": 9.595959595959595e-06,
      "loss": 0.5981,
      "step": 19
    },
    {
      "epoch": 0.06097560975609756,
      "grad_norm": 0.08299791488120488,
      "learning_rate": 1.0101010101010101e-05,
      "loss": 0.6851,
      "step": 20
    },
    {
      "epoch": 0.06402439024390244,
      "grad_norm": 0.08322761474768019,
      "learning_rate": 1.0606060606060607e-05,
      "loss": 0.6895,
      "step": 21
    },
    {
      "epoch": 0.06707317073170732,
      "grad_norm": 0.06946779021320976,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.5958,
      "step": 22
    },
    {
      "epoch": 0.0701219512195122,
      "grad_norm": 0.0840644563494239,
      "learning_rate": 1.1616161616161616e-05,
      "loss": 0.6865,
      "step": 23
    },
    {
      "epoch": 0.07317073170731707,
      "grad_norm": 0.08103942510491499,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 0.6385,
      "step": 24
    },
    {
      "epoch": 0.07621951219512195,
      "grad_norm": 0.06960617777809865,
      "learning_rate": 1.2626262626262628e-05,
      "loss": 0.5766,
      "step": 25
    },
    {
      "epoch": 0.07926829268292683,
      "grad_norm": 0.06350551869608764,
      "learning_rate": 1.3131313131313134e-05,
      "loss": 0.53,
      "step": 26
    },
    {
      "epoch": 0.08231707317073171,
      "grad_norm": 0.06954405913739202,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 0.5804,
      "step": 27
    },
    {
      "epoch": 0.08536585365853659,
      "grad_norm": 0.05801979787649268,
      "learning_rate": 1.4141414141414141e-05,
      "loss": 0.5463,
      "step": 28
    },
    {
      "epoch": 0.08841463414634146,
      "grad_norm": 0.0707113147835271,
      "learning_rate": 1.4646464646464647e-05,
      "loss": 0.6901,
      "step": 29
    },
    {
      "epoch": 0.09146341463414634,
      "grad_norm": 0.07199565636119566,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 0.5657,
      "step": 30
    },
    {
      "epoch": 0.09451219512195122,
      "grad_norm": 0.06031348351647801,
      "learning_rate": 1.565656565656566e-05,
      "loss": 0.5793,
      "step": 31
    },
    {
      "epoch": 0.0975609756097561,
      "grad_norm": 0.051381021893855267,
      "learning_rate": 1.6161616161616165e-05,
      "loss": 0.5032,
      "step": 32
    },
    {
      "epoch": 0.10060975609756098,
      "grad_norm": 0.05224030536199245,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.535,
      "step": 33
    },
    {
      "epoch": 0.10365853658536585,
      "grad_norm": 0.057566908214629696,
      "learning_rate": 1.7171717171717173e-05,
      "loss": 0.5695,
      "step": 34
    },
    {
      "epoch": 0.10670731707317073,
      "grad_norm": 0.050553154622884115,
      "learning_rate": 1.7676767676767676e-05,
      "loss": 0.4747,
      "step": 35
    },
    {
      "epoch": 0.10975609756097561,
      "grad_norm": 0.0555320047190129,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.5131,
      "step": 36
    },
    {
      "epoch": 0.11280487804878049,
      "grad_norm": 0.04980278635904654,
      "learning_rate": 1.8686868686868688e-05,
      "loss": 0.5091,
      "step": 37
    },
    {
      "epoch": 0.11585365853658537,
      "grad_norm": 0.05349296511174711,
      "learning_rate": 1.919191919191919e-05,
      "loss": 0.6005,
      "step": 38
    },
    {
      "epoch": 0.11890243902439024,
      "grad_norm": 0.05531305808997776,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 0.584,
      "step": 39
    },
    {
      "epoch": 0.12195121951219512,
      "grad_norm": 0.05341724693526889,
      "learning_rate": 2.0202020202020203e-05,
      "loss": 0.5734,
      "step": 40
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.04936955763599991,
      "learning_rate": 2.070707070707071e-05,
      "loss": 0.4993,
      "step": 41
    },
    {
      "epoch": 0.12804878048780488,
      "grad_norm": 0.05063294439179141,
      "learning_rate": 2.1212121212121215e-05,
      "loss": 0.5256,
      "step": 42
    },
    {
      "epoch": 0.13109756097560976,
      "grad_norm": 0.04558814730407359,
      "learning_rate": 2.171717171717172e-05,
      "loss": 0.4779,
      "step": 43
    },
    {
      "epoch": 0.13414634146341464,
      "grad_norm": 0.052246397012185086,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.4885,
      "step": 44
    },
    {
      "epoch": 0.13719512195121952,
      "grad_norm": 0.049611145112304265,
      "learning_rate": 2.272727272727273e-05,
      "loss": 0.5412,
      "step": 45
    },
    {
      "epoch": 0.1402439024390244,
      "grad_norm": 0.04616911363965657,
      "learning_rate": 2.3232323232323232e-05,
      "loss": 0.4997,
      "step": 46
    },
    {
      "epoch": 0.14329268292682926,
      "grad_norm": 0.04431786715654772,
      "learning_rate": 2.3737373737373738e-05,
      "loss": 0.4489,
      "step": 47
    },
    {
      "epoch": 0.14634146341463414,
      "grad_norm": 0.049509312908459314,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 0.5677,
      "step": 48
    },
    {
      "epoch": 0.14939024390243902,
      "grad_norm": 0.043303149705511926,
      "learning_rate": 2.474747474747475e-05,
      "loss": 0.5461,
      "step": 49
    },
    {
      "epoch": 0.1524390243902439,
      "grad_norm": 0.04346369573012801,
      "learning_rate": 2.5252525252525256e-05,
      "loss": 0.4852,
      "step": 50
    },
    {
      "epoch": 0.15548780487804878,
      "grad_norm": 0.04010543981497035,
      "learning_rate": 2.575757575757576e-05,
      "loss": 0.4827,
      "step": 51
    },
    {
      "epoch": 0.15853658536585366,
      "grad_norm": 0.04273578550839348,
      "learning_rate": 2.6262626262626268e-05,
      "loss": 0.456,
      "step": 52
    },
    {
      "epoch": 0.16158536585365854,
      "grad_norm": 0.04899472557064514,
      "learning_rate": 2.676767676767677e-05,
      "loss": 0.5793,
      "step": 53
    },
    {
      "epoch": 0.16463414634146342,
      "grad_norm": 0.04318797168345568,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.4915,
      "step": 54
    },
    {
      "epoch": 0.1676829268292683,
      "grad_norm": 0.05006798307827536,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.4948,
      "step": 55
    },
    {
      "epoch": 0.17073170731707318,
      "grad_norm": 0.04034250617082293,
      "learning_rate": 2.8282828282828282e-05,
      "loss": 0.4916,
      "step": 56
    },
    {
      "epoch": 0.17378048780487804,
      "grad_norm": 0.03657488246202251,
      "learning_rate": 2.878787878787879e-05,
      "loss": 0.4937,
      "step": 57
    },
    {
      "epoch": 0.17682926829268292,
      "grad_norm": 0.03213988489946405,
      "learning_rate": 2.9292929292929294e-05,
      "loss": 0.4011,
      "step": 58
    },
    {
      "epoch": 0.1798780487804878,
      "grad_norm": 0.03474129460321165,
      "learning_rate": 2.9797979797979796e-05,
      "loss": 0.4638,
      "step": 59
    },
    {
      "epoch": 0.18292682926829268,
      "grad_norm": 0.038909552802220075,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 0.5359,
      "step": 60
    },
    {
      "epoch": 0.18597560975609756,
      "grad_norm": 0.03437975920891206,
      "learning_rate": 3.080808080808081e-05,
      "loss": 0.4677,
      "step": 61
    },
    {
      "epoch": 0.18902439024390244,
      "grad_norm": 0.03564686452785141,
      "learning_rate": 3.131313131313132e-05,
      "loss": 0.4852,
      "step": 62
    },
    {
      "epoch": 0.19207317073170732,
      "grad_norm": 0.03402175099897292,
      "learning_rate": 3.181818181818182e-05,
      "loss": 0.478,
      "step": 63
    },
    {
      "epoch": 0.1951219512195122,
      "grad_norm": 0.04071628630413561,
      "learning_rate": 3.232323232323233e-05,
      "loss": 0.4804,
      "step": 64
    },
    {
      "epoch": 0.19817073170731708,
      "grad_norm": 0.03133737153925272,
      "learning_rate": 3.282828282828283e-05,
      "loss": 0.4796,
      "step": 65
    },
    {
      "epoch": 0.20121951219512196,
      "grad_norm": 0.030606542892269543,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.4543,
      "step": 66
    },
    {
      "epoch": 0.20426829268292682,
      "grad_norm": 0.028220765431020804,
      "learning_rate": 3.3838383838383844e-05,
      "loss": 0.445,
      "step": 67
    },
    {
      "epoch": 0.2073170731707317,
      "grad_norm": 0.030351542499101217,
      "learning_rate": 3.434343434343435e-05,
      "loss": 0.411,
      "step": 68
    },
    {
      "epoch": 0.21036585365853658,
      "grad_norm": 0.03059399901204726,
      "learning_rate": 3.484848484848485e-05,
      "loss": 0.515,
      "step": 69
    },
    {
      "epoch": 0.21341463414634146,
      "grad_norm": 0.03296963214213472,
      "learning_rate": 3.535353535353535e-05,
      "loss": 0.499,
      "step": 70
    },
    {
      "epoch": 0.21646341463414634,
      "grad_norm": 0.030683874962795153,
      "learning_rate": 3.5858585858585855e-05,
      "loss": 0.4457,
      "step": 71
    },
    {
      "epoch": 0.21951219512195122,
      "grad_norm": 0.03279445122686611,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.4637,
      "step": 72
    },
    {
      "epoch": 0.2225609756097561,
      "grad_norm": 0.027670174967853636,
      "learning_rate": 3.686868686868687e-05,
      "loss": 0.4637,
      "step": 73
    },
    {
      "epoch": 0.22560975609756098,
      "grad_norm": 0.03204511638784144,
      "learning_rate": 3.7373737373737376e-05,
      "loss": 0.4684,
      "step": 74
    },
    {
      "epoch": 0.22865853658536586,
      "grad_norm": 0.02692814426837153,
      "learning_rate": 3.787878787878788e-05,
      "loss": 0.4564,
      "step": 75
    },
    {
      "epoch": 0.23170731707317074,
      "grad_norm": 0.03159646644993244,
      "learning_rate": 3.838383838383838e-05,
      "loss": 0.501,
      "step": 76
    },
    {
      "epoch": 0.2347560975609756,
      "grad_norm": 0.02614313741800732,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.4242,
      "step": 77
    },
    {
      "epoch": 0.23780487804878048,
      "grad_norm": 0.026446688459674766,
      "learning_rate": 3.939393939393939e-05,
      "loss": 0.5611,
      "step": 78
    },
    {
      "epoch": 0.24085365853658536,
      "grad_norm": 0.029235728604449545,
      "learning_rate": 3.98989898989899e-05,
      "loss": 0.4523,
      "step": 79
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 0.029403985571445752,
      "learning_rate": 4.0404040404040405e-05,
      "loss": 0.5413,
      "step": 80
    },
    {
      "epoch": 0.24695121951219512,
      "grad_norm": 0.0248895165085513,
      "learning_rate": 4.0909090909090915e-05,
      "loss": 0.4353,
      "step": 81
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.02514252171483617,
      "learning_rate": 4.141414141414142e-05,
      "loss": 0.4019,
      "step": 82
    },
    {
      "epoch": 0.2530487804878049,
      "grad_norm": 0.02526428258646879,
      "learning_rate": 4.191919191919192e-05,
      "loss": 0.4026,
      "step": 83
    },
    {
      "epoch": 0.25609756097560976,
      "grad_norm": 0.02774638271334102,
      "learning_rate": 4.242424242424243e-05,
      "loss": 0.4443,
      "step": 84
    },
    {
      "epoch": 0.25914634146341464,
      "grad_norm": 0.02552350005556174,
      "learning_rate": 4.292929292929293e-05,
      "loss": 0.532,
      "step": 85
    },
    {
      "epoch": 0.2621951219512195,
      "grad_norm": 0.02337099188494046,
      "learning_rate": 4.343434343434344e-05,
      "loss": 0.3992,
      "step": 86
    },
    {
      "epoch": 0.2652439024390244,
      "grad_norm": 0.023989118355155412,
      "learning_rate": 4.3939393939393944e-05,
      "loss": 0.4473,
      "step": 87
    },
    {
      "epoch": 0.2682926829268293,
      "grad_norm": 0.02687814732230884,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.4594,
      "step": 88
    },
    {
      "epoch": 0.27134146341463417,
      "grad_norm": 0.027896900163840974,
      "learning_rate": 4.494949494949495e-05,
      "loss": 0.4866,
      "step": 89
    },
    {
      "epoch": 0.27439024390243905,
      "grad_norm": 0.025924142515604206,
      "learning_rate": 4.545454545454546e-05,
      "loss": 0.4598,
      "step": 90
    },
    {
      "epoch": 0.2774390243902439,
      "grad_norm": 0.02824961803928781,
      "learning_rate": 4.595959595959596e-05,
      "loss": 0.4895,
      "step": 91
    },
    {
      "epoch": 0.2804878048780488,
      "grad_norm": 0.025064170396782423,
      "learning_rate": 4.6464646464646464e-05,
      "loss": 0.3995,
      "step": 92
    },
    {
      "epoch": 0.28353658536585363,
      "grad_norm": 0.026567289363745077,
      "learning_rate": 4.696969696969697e-05,
      "loss": 0.3898,
      "step": 93
    },
    {
      "epoch": 0.2865853658536585,
      "grad_norm": 0.0235303526093093,
      "learning_rate": 4.7474747474747476e-05,
      "loss": 0.5007,
      "step": 94
    },
    {
      "epoch": 0.2896341463414634,
      "grad_norm": 0.0232550129311821,
      "learning_rate": 4.797979797979798e-05,
      "loss": 0.4353,
      "step": 95
    },
    {
      "epoch": 0.2926829268292683,
      "grad_norm": 0.027355889003192944,
      "learning_rate": 4.848484848484849e-05,
      "loss": 0.4228,
      "step": 96
    },
    {
      "epoch": 0.29573170731707316,
      "grad_norm": 0.026046752710899224,
      "learning_rate": 4.898989898989899e-05,
      "loss": 0.4911,
      "step": 97
    },
    {
      "epoch": 0.29878048780487804,
      "grad_norm": 0.024133091415805927,
      "learning_rate": 4.94949494949495e-05,
      "loss": 0.4194,
      "step": 98
    },
    {
      "epoch": 0.3018292682926829,
      "grad_norm": 0.027499375160356283,
      "learning_rate": 5e-05,
      "loss": 0.4401,
      "step": 99
    },
    {
      "epoch": 0.3048780487804878,
      "grad_norm": 0.02426792472675091,
      "learning_rate": 4.99998424846941e-05,
      "loss": 0.3926,
      "step": 100
    },
    {
      "epoch": 0.3079268292682927,
      "grad_norm": 0.028683991879347375,
      "learning_rate": 4.999936994076129e-05,
      "loss": 0.4317,
      "step": 101
    },
    {
      "epoch": 0.31097560975609756,
      "grad_norm": 0.02515593602285064,
      "learning_rate": 4.999858237415621e-05,
      "loss": 0.4921,
      "step": 102
    },
    {
      "epoch": 0.31402439024390244,
      "grad_norm": 0.02731378150973361,
      "learning_rate": 4.999747979480314e-05,
      "loss": 0.3999,
      "step": 103
    },
    {
      "epoch": 0.3170731707317073,
      "grad_norm": 0.02469325829557163,
      "learning_rate": 4.999606221659595e-05,
      "loss": 0.4291,
      "step": 104
    },
    {
      "epoch": 0.3201219512195122,
      "grad_norm": 0.025887949421610226,
      "learning_rate": 4.999432965739785e-05,
      "loss": 0.4064,
      "step": 105
    },
    {
      "epoch": 0.3231707317073171,
      "grad_norm": 0.029562890120902404,
      "learning_rate": 4.999228213904123e-05,
      "loss": 0.4704,
      "step": 106
    },
    {
      "epoch": 0.32621951219512196,
      "grad_norm": 0.026468507235623037,
      "learning_rate": 4.998991968732729e-05,
      "loss": 0.4291,
      "step": 107
    },
    {
      "epoch": 0.32926829268292684,
      "grad_norm": 0.028415208011529337,
      "learning_rate": 4.9987242332025846e-05,
      "loss": 0.4445,
      "step": 108
    },
    {
      "epoch": 0.3323170731707317,
      "grad_norm": 0.02690846850933201,
      "learning_rate": 4.998425010687484e-05,
      "loss": 0.4413,
      "step": 109
    },
    {
      "epoch": 0.3353658536585366,
      "grad_norm": 0.029434783335942422,
      "learning_rate": 4.9980943049579975e-05,
      "loss": 0.5259,
      "step": 110
    },
    {
      "epoch": 0.3384146341463415,
      "grad_norm": 0.02732698412255557,
      "learning_rate": 4.997732120181423e-05,
      "loss": 0.458,
      "step": 111
    },
    {
      "epoch": 0.34146341463414637,
      "grad_norm": 0.02817361324202221,
      "learning_rate": 4.997338460921731e-05,
      "loss": 0.5046,
      "step": 112
    },
    {
      "epoch": 0.3445121951219512,
      "grad_norm": 0.026598418378303214,
      "learning_rate": 4.9969133321395104e-05,
      "loss": 0.4538,
      "step": 113
    },
    {
      "epoch": 0.3475609756097561,
      "grad_norm": 0.026712012837418288,
      "learning_rate": 4.996456739191905e-05,
      "loss": 0.4333,
      "step": 114
    },
    {
      "epoch": 0.35060975609756095,
      "grad_norm": 0.02664382392284045,
      "learning_rate": 4.995968687832545e-05,
      "loss": 0.3973,
      "step": 115
    },
    {
      "epoch": 0.35365853658536583,
      "grad_norm": 0.026832454561682675,
      "learning_rate": 4.995449184211474e-05,
      "loss": 0.3963,
      "step": 116
    },
    {
      "epoch": 0.3567073170731707,
      "grad_norm": 0.03128497596944388,
      "learning_rate": 4.994898234875075e-05,
      "loss": 0.4612,
      "step": 117
    },
    {
      "epoch": 0.3597560975609756,
      "grad_norm": 0.027920438805892456,
      "learning_rate": 4.994315846765984e-05,
      "loss": 0.4803,
      "step": 118
    },
    {
      "epoch": 0.3628048780487805,
      "grad_norm": 0.02790719813073735,
      "learning_rate": 4.993702027223004e-05,
      "loss": 0.3983,
      "step": 119
    },
    {
      "epoch": 0.36585365853658536,
      "grad_norm": 0.02954525910398473,
      "learning_rate": 4.993056783981013e-05,
      "loss": 0.53,
      "step": 120
    },
    {
      "epoch": 0.36890243902439024,
      "grad_norm": 0.03279451861565122,
      "learning_rate": 4.9923801251708655e-05,
      "loss": 0.4329,
      "step": 121
    },
    {
      "epoch": 0.3719512195121951,
      "grad_norm": 0.02969073542607525,
      "learning_rate": 4.9916720593192925e-05,
      "loss": 0.495,
      "step": 122
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.030046395052057563,
      "learning_rate": 4.990932595348789e-05,
      "loss": 0.5039,
      "step": 123
    },
    {
      "epoch": 0.3780487804878049,
      "grad_norm": 0.030594799975909838,
      "learning_rate": 4.9901617425775067e-05,
      "loss": 0.5416,
      "step": 124
    },
    {
      "epoch": 0.38109756097560976,
      "grad_norm": 0.02419088359613875,
      "learning_rate": 4.9893595107191355e-05,
      "loss": 0.3206,
      "step": 125
    },
    {
      "epoch": 0.38414634146341464,
      "grad_norm": 0.028896084169673618,
      "learning_rate": 4.988525909882779e-05,
      "loss": 0.4127,
      "step": 126
    },
    {
      "epoch": 0.3871951219512195,
      "grad_norm": 0.04001825579621416,
      "learning_rate": 4.9876609505728266e-05,
      "loss": 0.4392,
      "step": 127
    },
    {
      "epoch": 0.3902439024390244,
      "grad_norm": 0.14115804141301908,
      "learning_rate": 4.9867646436888274e-05,
      "loss": 0.4594,
      "step": 128
    },
    {
      "epoch": 0.3932926829268293,
      "grad_norm": 0.029050747953028554,
      "learning_rate": 4.985837000525343e-05,
      "loss": 0.4337,
      "step": 129
    },
    {
      "epoch": 0.39634146341463417,
      "grad_norm": 0.02920084405101805,
      "learning_rate": 4.984878032771815e-05,
      "loss": 0.4225,
      "step": 130
    },
    {
      "epoch": 0.39939024390243905,
      "grad_norm": 0.03011718239999231,
      "learning_rate": 4.983887752512412e-05,
      "loss": 0.4538,
      "step": 131
    },
    {
      "epoch": 0.4024390243902439,
      "grad_norm": 0.027897119508686768,
      "learning_rate": 4.982866172225875e-05,
      "loss": 0.4071,
      "step": 132
    },
    {
      "epoch": 0.4054878048780488,
      "grad_norm": 0.03577152763319291,
      "learning_rate": 4.981813304785369e-05,
      "loss": 0.5895,
      "step": 133
    },
    {
      "epoch": 0.40853658536585363,
      "grad_norm": 0.03277686846451814,
      "learning_rate": 4.980729163458312e-05,
      "loss": 0.5017,
      "step": 134
    },
    {
      "epoch": 0.4115853658536585,
      "grad_norm": 0.03217006570314937,
      "learning_rate": 4.979613761906212e-05,
      "loss": 0.4384,
      "step": 135
    },
    {
      "epoch": 0.4146341463414634,
      "grad_norm": 0.030949220084112952,
      "learning_rate": 4.978467114184495e-05,
      "loss": 0.4815,
      "step": 136
    },
    {
      "epoch": 0.4176829268292683,
      "grad_norm": 0.030260238522408403,
      "learning_rate": 4.977289234742325e-05,
      "loss": 0.4143,
      "step": 137
    },
    {
      "epoch": 0.42073170731707316,
      "grad_norm": 0.028522608089292575,
      "learning_rate": 4.976080138422428e-05,
      "loss": 0.402,
      "step": 138
    },
    {
      "epoch": 0.42378048780487804,
      "grad_norm": 0.029049158089749227,
      "learning_rate": 4.974839840460895e-05,
      "loss": 0.3804,
      "step": 139
    },
    {
      "epoch": 0.4268292682926829,
      "grad_norm": 0.0298218460769968,
      "learning_rate": 4.973568356487001e-05,
      "loss": 0.4665,
      "step": 140
    },
    {
      "epoch": 0.4298780487804878,
      "grad_norm": 0.030435705321632405,
      "learning_rate": 4.972265702523e-05,
      "loss": 0.4775,
      "step": 141
    },
    {
      "epoch": 0.4329268292682927,
      "grad_norm": 0.033498527194517305,
      "learning_rate": 4.970931894983927e-05,
      "loss": 0.4643,
      "step": 142
    },
    {
      "epoch": 0.43597560975609756,
      "grad_norm": 0.024129978615902534,
      "learning_rate": 4.969566950677391e-05,
      "loss": 0.387,
      "step": 143
    },
    {
      "epoch": 0.43902439024390244,
      "grad_norm": 0.033295978068974476,
      "learning_rate": 4.9681708868033616e-05,
      "loss": 0.4444,
      "step": 144
    },
    {
      "epoch": 0.4420731707317073,
      "grad_norm": 0.0318133787810746,
      "learning_rate": 4.9667437209539525e-05,
      "loss": 0.3922,
      "step": 145
    },
    {
      "epoch": 0.4451219512195122,
      "grad_norm": 0.031838890935437666,
      "learning_rate": 4.965285471113201e-05,
      "loss": 0.43,
      "step": 146
    },
    {
      "epoch": 0.4481707317073171,
      "grad_norm": 0.031279587433315055,
      "learning_rate": 4.9637961556568404e-05,
      "loss": 0.4794,
      "step": 147
    },
    {
      "epoch": 0.45121951219512196,
      "grad_norm": 0.029605363976328944,
      "learning_rate": 4.9622757933520694e-05,
      "loss": 0.4084,
      "step": 148
    },
    {
      "epoch": 0.45426829268292684,
      "grad_norm": 0.03328501411562014,
      "learning_rate": 4.9607244033573156e-05,
      "loss": 0.3862,
      "step": 149
    },
    {
      "epoch": 0.4573170731707317,
      "grad_norm": 0.02977331587790026,
      "learning_rate": 4.959142005221991e-05,
      "loss": 0.4317,
      "step": 150
    },
    {
      "epoch": 0.4603658536585366,
      "grad_norm": 0.03445716661980318,
      "learning_rate": 4.957528618886249e-05,
      "loss": 0.3587,
      "step": 151
    },
    {
      "epoch": 0.4634146341463415,
      "grad_norm": 0.03349464222579777,
      "learning_rate": 4.9558842646807354e-05,
      "loss": 0.4175,
      "step": 152
    },
    {
      "epoch": 0.46646341463414637,
      "grad_norm": 0.03172092320527115,
      "learning_rate": 4.954208963326327e-05,
      "loss": 0.4291,
      "step": 153
    },
    {
      "epoch": 0.4695121951219512,
      "grad_norm": 0.03495378537426485,
      "learning_rate": 4.9525027359338696e-05,
      "loss": 0.4943,
      "step": 154
    },
    {
      "epoch": 0.4725609756097561,
      "grad_norm": 0.02829629512141535,
      "learning_rate": 4.9507656040039186e-05,
      "loss": 0.4044,
      "step": 155
    },
    {
      "epoch": 0.47560975609756095,
      "grad_norm": 0.03372581762913699,
      "learning_rate": 4.9489975894264636e-05,
      "loss": 0.4583,
      "step": 156
    },
    {
      "epoch": 0.47865853658536583,
      "grad_norm": 0.031007930096541795,
      "learning_rate": 4.947198714480653e-05,
      "loss": 0.3455,
      "step": 157
    },
    {
      "epoch": 0.4817073170731707,
      "grad_norm": 0.03294155045988215,
      "learning_rate": 4.9453690018345144e-05,
      "loss": 0.4384,
      "step": 158
    },
    {
      "epoch": 0.4847560975609756,
      "grad_norm": 0.032312853442576034,
      "learning_rate": 4.9435084745446666e-05,
      "loss": 0.4049,
      "step": 159
    },
    {
      "epoch": 0.4878048780487805,
      "grad_norm": 0.030987604019126144,
      "learning_rate": 4.941617156056032e-05,
      "loss": 0.4172,
      "step": 160
    },
    {
      "epoch": 0.49085365853658536,
      "grad_norm": 0.02617581898410179,
      "learning_rate": 4.939695070201541e-05,
      "loss": 0.3645,
      "step": 161
    },
    {
      "epoch": 0.49390243902439024,
      "grad_norm": 0.03584401963601039,
      "learning_rate": 4.937742241201826e-05,
      "loss": 0.4821,
      "step": 162
    },
    {
      "epoch": 0.4969512195121951,
      "grad_norm": 0.03254199704880417,
      "learning_rate": 4.9357586936649255e-05,
      "loss": 0.4286,
      "step": 163
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.03301611426756909,
      "learning_rate": 4.933744452585966e-05,
      "loss": 0.4156,
      "step": 164
    },
    {
      "epoch": 0.5030487804878049,
      "grad_norm": 0.036375576445087335,
      "learning_rate": 4.931699543346854e-05,
      "loss": 0.5381,
      "step": 165
    },
    {
      "epoch": 0.5060975609756098,
      "grad_norm": 0.02629100047771681,
      "learning_rate": 4.929623991715947e-05,
      "loss": 0.3748,
      "step": 166
    },
    {
      "epoch": 0.5091463414634146,
      "grad_norm": 0.03271065383289755,
      "learning_rate": 4.927517823847738e-05,
      "loss": 0.424,
      "step": 167
    },
    {
      "epoch": 0.5121951219512195,
      "grad_norm": 0.02977415364330819,
      "learning_rate": 4.9253810662825216e-05,
      "loss": 0.4107,
      "step": 168
    },
    {
      "epoch": 0.5152439024390244,
      "grad_norm": 0.029592038707982913,
      "learning_rate": 4.923213745946059e-05,
      "loss": 0.4416,
      "step": 169
    },
    {
      "epoch": 0.5182926829268293,
      "grad_norm": 0.034123248186661195,
      "learning_rate": 4.92101589014924e-05,
      "loss": 0.4724,
      "step": 170
    },
    {
      "epoch": 0.5213414634146342,
      "grad_norm": 0.030117950572343562,
      "learning_rate": 4.918787526587739e-05,
      "loss": 0.4342,
      "step": 171
    },
    {
      "epoch": 0.524390243902439,
      "grad_norm": 0.02784489931949698,
      "learning_rate": 4.9165286833416666e-05,
      "loss": 0.414,
      "step": 172
    },
    {
      "epoch": 0.5274390243902439,
      "grad_norm": 0.028838060629043785,
      "learning_rate": 4.9142393888752125e-05,
      "loss": 0.3891,
      "step": 173
    },
    {
      "epoch": 0.5304878048780488,
      "grad_norm": 0.03039770773924711,
      "learning_rate": 4.91191967203629e-05,
      "loss": 0.4696,
      "step": 174
    },
    {
      "epoch": 0.5335365853658537,
      "grad_norm": 0.027932971891971924,
      "learning_rate": 4.909569562056172e-05,
      "loss": 0.3999,
      "step": 175
    },
    {
      "epoch": 0.5365853658536586,
      "grad_norm": 0.032440243318377006,
      "learning_rate": 4.9071890885491224e-05,
      "loss": 0.4376,
      "step": 176
    },
    {
      "epoch": 0.5396341463414634,
      "grad_norm": 0.028239012069124093,
      "learning_rate": 4.9047782815120214e-05,
      "loss": 0.4173,
      "step": 177
    },
    {
      "epoch": 0.5426829268292683,
      "grad_norm": 0.03145028792920836,
      "learning_rate": 4.9023371713239895e-05,
      "loss": 0.4447,
      "step": 178
    },
    {
      "epoch": 0.5457317073170732,
      "grad_norm": 0.0331574276058771,
      "learning_rate": 4.899865788746005e-05,
      "loss": 0.4591,
      "step": 179
    },
    {
      "epoch": 0.5487804878048781,
      "grad_norm": 0.03236199256446896,
      "learning_rate": 4.897364164920514e-05,
      "loss": 0.4119,
      "step": 180
    },
    {
      "epoch": 0.551829268292683,
      "grad_norm": 0.031426884352228006,
      "learning_rate": 4.8948323313710405e-05,
      "loss": 0.3928,
      "step": 181
    },
    {
      "epoch": 0.5548780487804879,
      "grad_norm": 0.029533984825273876,
      "learning_rate": 4.892270320001786e-05,
      "loss": 0.4302,
      "step": 182
    },
    {
      "epoch": 0.5579268292682927,
      "grad_norm": 0.0275533238930569,
      "learning_rate": 4.889678163097233e-05,
      "loss": 0.3847,
      "step": 183
    },
    {
      "epoch": 0.5609756097560976,
      "grad_norm": 0.02879278222254772,
      "learning_rate": 4.88705589332173e-05,
      "loss": 0.3578,
      "step": 184
    },
    {
      "epoch": 0.5640243902439024,
      "grad_norm": 0.02856312748685412,
      "learning_rate": 4.884403543719089e-05,
      "loss": 0.3671,
      "step": 185
    },
    {
      "epoch": 0.5670731707317073,
      "grad_norm": 0.03325654158942621,
      "learning_rate": 4.881721147712162e-05,
      "loss": 0.4458,
      "step": 186
    },
    {
      "epoch": 0.5701219512195121,
      "grad_norm": 0.03203686650016824,
      "learning_rate": 4.879008739102423e-05,
      "loss": 0.4267,
      "step": 187
    },
    {
      "epoch": 0.573170731707317,
      "grad_norm": 0.031235969957028178,
      "learning_rate": 4.876266352069542e-05,
      "loss": 0.4254,
      "step": 188
    },
    {
      "epoch": 0.5762195121951219,
      "grad_norm": 0.03192624303186046,
      "learning_rate": 4.873494021170953e-05,
      "loss": 0.38,
      "step": 189
    },
    {
      "epoch": 0.5792682926829268,
      "grad_norm": 0.030168180809586265,
      "learning_rate": 4.870691781341422e-05,
      "loss": 0.3605,
      "step": 190
    },
    {
      "epoch": 0.5823170731707317,
      "grad_norm": 0.032779575767083055,
      "learning_rate": 4.867859667892599e-05,
      "loss": 0.4632,
      "step": 191
    },
    {
      "epoch": 0.5853658536585366,
      "grad_norm": 0.031635000676225315,
      "learning_rate": 4.864997716512584e-05,
      "loss": 0.4022,
      "step": 192
    },
    {
      "epoch": 0.5884146341463414,
      "grad_norm": 0.03300217321436167,
      "learning_rate": 4.8621059632654684e-05,
      "loss": 0.3976,
      "step": 193
    },
    {
      "epoch": 0.5914634146341463,
      "grad_norm": 0.036533189955443766,
      "learning_rate": 4.859184444590882e-05,
      "loss": 0.4193,
      "step": 194
    },
    {
      "epoch": 0.5945121951219512,
      "grad_norm": 0.03429837316646944,
      "learning_rate": 4.856233197303539e-05,
      "loss": 0.4551,
      "step": 195
    },
    {
      "epoch": 0.5975609756097561,
      "grad_norm": 0.042055757960286724,
      "learning_rate": 4.853252258592769e-05,
      "loss": 0.5209,
      "step": 196
    },
    {
      "epoch": 0.600609756097561,
      "grad_norm": 0.032413833464997835,
      "learning_rate": 4.85024166602205e-05,
      "loss": 0.4542,
      "step": 197
    },
    {
      "epoch": 0.6036585365853658,
      "grad_norm": 0.032486592113239954,
      "learning_rate": 4.847201457528533e-05,
      "loss": 0.4547,
      "step": 198
    },
    {
      "epoch": 0.6067073170731707,
      "grad_norm": 0.03146274594158332,
      "learning_rate": 4.84413167142257e-05,
      "loss": 0.4303,
      "step": 199
    },
    {
      "epoch": 0.6097560975609756,
      "grad_norm": 0.03300614850893232,
      "learning_rate": 4.841032346387223e-05,
      "loss": 0.3996,
      "step": 200
    },
    {
      "epoch": 0.6128048780487805,
      "grad_norm": 0.03033077869518619,
      "learning_rate": 4.8379035214777836e-05,
      "loss": 0.3868,
      "step": 201
    },
    {
      "epoch": 0.6158536585365854,
      "grad_norm": 0.041107227614117905,
      "learning_rate": 4.834745236121276e-05,
      "loss": 0.4447,
      "step": 202
    },
    {
      "epoch": 0.6189024390243902,
      "grad_norm": 0.03258067072978055,
      "learning_rate": 4.8315575301159644e-05,
      "loss": 0.4083,
      "step": 203
    },
    {
      "epoch": 0.6219512195121951,
      "grad_norm": 0.03324141495975344,
      "learning_rate": 4.8283404436308464e-05,
      "loss": 0.3783,
      "step": 204
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.0346026019862594,
      "learning_rate": 4.8250940172051506e-05,
      "loss": 0.4676,
      "step": 205
    },
    {
      "epoch": 0.6280487804878049,
      "grad_norm": 0.034218054427013485,
      "learning_rate": 4.821818291747826e-05,
      "loss": 0.4161,
      "step": 206
    },
    {
      "epoch": 0.6310975609756098,
      "grad_norm": 0.03578241935145395,
      "learning_rate": 4.8185133085370246e-05,
      "loss": 0.4546,
      "step": 207
    },
    {
      "epoch": 0.6341463414634146,
      "grad_norm": 0.029220203681492543,
      "learning_rate": 4.815179109219581e-05,
      "loss": 0.3681,
      "step": 208
    },
    {
      "epoch": 0.6371951219512195,
      "grad_norm": 0.034325656907965206,
      "learning_rate": 4.81181573581049e-05,
      "loss": 0.4236,
      "step": 209
    },
    {
      "epoch": 0.6402439024390244,
      "grad_norm": 0.031648709603629754,
      "learning_rate": 4.808423230692374e-05,
      "loss": 0.4113,
      "step": 210
    },
    {
      "epoch": 0.6432926829268293,
      "grad_norm": 0.04023544907777028,
      "learning_rate": 4.805001636614951e-05,
      "loss": 0.4344,
      "step": 211
    },
    {
      "epoch": 0.6463414634146342,
      "grad_norm": 0.033804893131760355,
      "learning_rate": 4.801550996694498e-05,
      "loss": 0.4015,
      "step": 212
    },
    {
      "epoch": 0.649390243902439,
      "grad_norm": 0.030694782148620977,
      "learning_rate": 4.798071354413302e-05,
      "loss": 0.4318,
      "step": 213
    },
    {
      "epoch": 0.6524390243902439,
      "grad_norm": 0.032401947613213765,
      "learning_rate": 4.7945627536191166e-05,
      "loss": 0.4608,
      "step": 214
    },
    {
      "epoch": 0.6554878048780488,
      "grad_norm": 0.034950291203224994,
      "learning_rate": 4.791025238524608e-05,
      "loss": 0.4746,
      "step": 215
    },
    {
      "epoch": 0.6585365853658537,
      "grad_norm": 0.03170711324948168,
      "learning_rate": 4.787458853706798e-05,
      "loss": 0.4547,
      "step": 216
    },
    {
      "epoch": 0.6615853658536586,
      "grad_norm": 0.03091501658528955,
      "learning_rate": 4.783863644106502e-05,
      "loss": 0.4375,
      "step": 217
    },
    {
      "epoch": 0.6646341463414634,
      "grad_norm": 0.035732798404897494,
      "learning_rate": 4.780239655027764e-05,
      "loss": 0.3974,
      "step": 218
    },
    {
      "epoch": 0.6676829268292683,
      "grad_norm": 0.035871702747555335,
      "learning_rate": 4.7765869321372836e-05,
      "loss": 0.4421,
      "step": 219
    },
    {
      "epoch": 0.6707317073170732,
      "grad_norm": 0.030446573090157556,
      "learning_rate": 4.772905521463841e-05,
      "loss": 0.3724,
      "step": 220
    },
    {
      "epoch": 0.6737804878048781,
      "grad_norm": 0.03069146337023481,
      "learning_rate": 4.769195469397719e-05,
      "loss": 0.4074,
      "step": 221
    },
    {
      "epoch": 0.676829268292683,
      "grad_norm": 0.030890634848987315,
      "learning_rate": 4.765456822690116e-05,
      "loss": 0.4885,
      "step": 222
    },
    {
      "epoch": 0.6798780487804879,
      "grad_norm": 0.03259691598463501,
      "learning_rate": 4.7616896284525594e-05,
      "loss": 0.4435,
      "step": 223
    },
    {
      "epoch": 0.6829268292682927,
      "grad_norm": 0.03610520773760829,
      "learning_rate": 4.7578939341563095e-05,
      "loss": 0.4422,
      "step": 224
    },
    {
      "epoch": 0.6859756097560976,
      "grad_norm": 0.030129525739252205,
      "learning_rate": 4.754069787631761e-05,
      "loss": 0.4183,
      "step": 225
    },
    {
      "epoch": 0.6890243902439024,
      "grad_norm": 0.03226262495076605,
      "learning_rate": 4.7502172370678434e-05,
      "loss": 0.414,
      "step": 226
    },
    {
      "epoch": 0.6920731707317073,
      "grad_norm": 0.038231135681208804,
      "learning_rate": 4.7463363310114106e-05,
      "loss": 0.5366,
      "step": 227
    },
    {
      "epoch": 0.6951219512195121,
      "grad_norm": 0.03350943712591451,
      "learning_rate": 4.742427118366632e-05,
      "loss": 0.5113,
      "step": 228
    },
    {
      "epoch": 0.698170731707317,
      "grad_norm": 0.033377477689505115,
      "learning_rate": 4.738489648394373e-05,
      "loss": 0.3785,
      "step": 229
    },
    {
      "epoch": 0.7012195121951219,
      "grad_norm": 0.02912803902171078,
      "learning_rate": 4.7345239707115764e-05,
      "loss": 0.3561,
      "step": 230
    },
    {
      "epoch": 0.7042682926829268,
      "grad_norm": 0.03279893881426138,
      "learning_rate": 4.7305301352906376e-05,
      "loss": 0.392,
      "step": 231
    },
    {
      "epoch": 0.7073170731707317,
      "grad_norm": 0.033540891496370996,
      "learning_rate": 4.7265081924587715e-05,
      "loss": 0.4123,
      "step": 232
    },
    {
      "epoch": 0.7103658536585366,
      "grad_norm": 0.037955097184342194,
      "learning_rate": 4.722458192897384e-05,
      "loss": 0.3895,
      "step": 233
    },
    {
      "epoch": 0.7134146341463414,
      "grad_norm": 0.03823305877451305,
      "learning_rate": 4.7183801876414294e-05,
      "loss": 0.4963,
      "step": 234
    },
    {
      "epoch": 0.7164634146341463,
      "grad_norm": 0.0325437922967289,
      "learning_rate": 4.7142742280787654e-05,
      "loss": 0.395,
      "step": 235
    },
    {
      "epoch": 0.7195121951219512,
      "grad_norm": 0.033910959532630194,
      "learning_rate": 4.7101403659495114e-05,
      "loss": 0.4213,
      "step": 236
    },
    {
      "epoch": 0.7225609756097561,
      "grad_norm": 0.03988837387160677,
      "learning_rate": 4.705978653345392e-05,
      "loss": 0.4515,
      "step": 237
    },
    {
      "epoch": 0.725609756097561,
      "grad_norm": 0.033382737192550964,
      "learning_rate": 4.701789142709081e-05,
      "loss": 0.4293,
      "step": 238
    },
    {
      "epoch": 0.7286585365853658,
      "grad_norm": 0.03683967072126718,
      "learning_rate": 4.697571886833544e-05,
      "loss": 0.4869,
      "step": 239
    },
    {
      "epoch": 0.7317073170731707,
      "grad_norm": 0.03389014227863287,
      "learning_rate": 4.693326938861366e-05,
      "loss": 0.432,
      "step": 240
    },
    {
      "epoch": 0.7347560975609756,
      "grad_norm": 0.03338844412509684,
      "learning_rate": 4.689054352284094e-05,
      "loss": 0.3734,
      "step": 241
    },
    {
      "epoch": 0.7378048780487805,
      "grad_norm": 0.03350579155466827,
      "learning_rate": 4.684754180941545e-05,
      "loss": 0.4609,
      "step": 242
    },
    {
      "epoch": 0.7408536585365854,
      "grad_norm": 0.0360391450589852,
      "learning_rate": 4.680426479021147e-05,
      "loss": 0.4488,
      "step": 243
    },
    {
      "epoch": 0.7439024390243902,
      "grad_norm": 0.0325562245901543,
      "learning_rate": 4.676071301057243e-05,
      "loss": 0.3746,
      "step": 244
    },
    {
      "epoch": 0.7469512195121951,
      "grad_norm": 0.037989790962959104,
      "learning_rate": 4.6716887019304075e-05,
      "loss": 0.4802,
      "step": 245
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.03446224282201849,
      "learning_rate": 4.6672787368667556e-05,
      "loss": 0.3878,
      "step": 246
    },
    {
      "epoch": 0.7530487804878049,
      "grad_norm": 0.038240603812219866,
      "learning_rate": 4.662841461437247e-05,
      "loss": 0.4311,
      "step": 247
    },
    {
      "epoch": 0.7560975609756098,
      "grad_norm": 0.03293308938612019,
      "learning_rate": 4.658376931556986e-05,
      "loss": 0.4569,
      "step": 248
    },
    {
      "epoch": 0.7591463414634146,
      "grad_norm": 0.03479760626237619,
      "learning_rate": 4.653885203484515e-05,
      "loss": 0.3793,
      "step": 249
    },
    {
      "epoch": 0.7621951219512195,
      "grad_norm": 0.037388404653232496,
      "learning_rate": 4.649366333821109e-05,
      "loss": 0.4452,
      "step": 250
    },
    {
      "epoch": 0.7652439024390244,
      "grad_norm": 0.0352761753839379,
      "learning_rate": 4.644820379510058e-05,
      "loss": 0.4568,
      "step": 251
    },
    {
      "epoch": 0.7682926829268293,
      "grad_norm": 0.035571316960728315,
      "learning_rate": 4.640247397835953e-05,
      "loss": 0.4964,
      "step": 252
    },
    {
      "epoch": 0.7713414634146342,
      "grad_norm": 0.042108811767305916,
      "learning_rate": 4.635647446423963e-05,
      "loss": 0.4635,
      "step": 253
    },
    {
      "epoch": 0.774390243902439,
      "grad_norm": 0.03740018008517267,
      "learning_rate": 4.631020583239107e-05,
      "loss": 0.4075,
      "step": 254
    },
    {
      "epoch": 0.7774390243902439,
      "grad_norm": 0.03406740378233361,
      "learning_rate": 4.626366866585528e-05,
      "loss": 0.4123,
      "step": 255
    },
    {
      "epoch": 0.7804878048780488,
      "grad_norm": 0.03534462880308758,
      "learning_rate": 4.621686355105752e-05,
      "loss": 0.5181,
      "step": 256
    },
    {
      "epoch": 0.7835365853658537,
      "grad_norm": 0.039032255535887364,
      "learning_rate": 4.616979107779958e-05,
      "loss": 0.4219,
      "step": 257
    },
    {
      "epoch": 0.7865853658536586,
      "grad_norm": 0.03228976154931515,
      "learning_rate": 4.612245183925225e-05,
      "loss": 0.4264,
      "step": 258
    },
    {
      "epoch": 0.7896341463414634,
      "grad_norm": 0.03359739152114295,
      "learning_rate": 4.607484643194788e-05,
      "loss": 0.39,
      "step": 259
    },
    {
      "epoch": 0.7926829268292683,
      "grad_norm": 0.04618842209757573,
      "learning_rate": 4.602697545577292e-05,
      "loss": 0.4484,
      "step": 260
    },
    {
      "epoch": 0.7957317073170732,
      "grad_norm": 0.03767698901292077,
      "learning_rate": 4.597883951396027e-05,
      "loss": 0.4427,
      "step": 261
    },
    {
      "epoch": 0.7987804878048781,
      "grad_norm": 0.033711281520042095,
      "learning_rate": 4.593043921308175e-05,
      "loss": 0.3961,
      "step": 262
    },
    {
      "epoch": 0.801829268292683,
      "grad_norm": 0.03517505501582254,
      "learning_rate": 4.588177516304042e-05,
      "loss": 0.4626,
      "step": 263
    },
    {
      "epoch": 0.8048780487804879,
      "grad_norm": 0.041222873153210056,
      "learning_rate": 4.5832847977062874e-05,
      "loss": 0.489,
      "step": 264
    },
    {
      "epoch": 0.8079268292682927,
      "grad_norm": 0.036670715661490515,
      "learning_rate": 4.578365827169159e-05,
      "loss": 0.4201,
      "step": 265
    },
    {
      "epoch": 0.8109756097560976,
      "grad_norm": 0.03429356814619947,
      "learning_rate": 4.5734206666777075e-05,
      "loss": 0.4228,
      "step": 266
    },
    {
      "epoch": 0.8140243902439024,
      "grad_norm": 0.03454066782608884,
      "learning_rate": 4.5684493785470105e-05,
      "loss": 0.33,
      "step": 267
    },
    {
      "epoch": 0.8170731707317073,
      "grad_norm": 0.042630367804240565,
      "learning_rate": 4.563452025421386e-05,
      "loss": 0.4113,
      "step": 268
    },
    {
      "epoch": 0.8201219512195121,
      "grad_norm": 0.03518819540379091,
      "learning_rate": 4.558428670273601e-05,
      "loss": 0.4554,
      "step": 269
    },
    {
      "epoch": 0.823170731707317,
      "grad_norm": 0.03059671220624392,
      "learning_rate": 4.5533793764040844e-05,
      "loss": 0.3427,
      "step": 270
    },
    {
      "epoch": 0.8262195121951219,
      "grad_norm": 0.03994211546654242,
      "learning_rate": 4.5483042074401185e-05,
      "loss": 0.4469,
      "step": 271
    },
    {
      "epoch": 0.8292682926829268,
      "grad_norm": 0.042677969772066116,
      "learning_rate": 4.5432032273350484e-05,
      "loss": 0.4816,
      "step": 272
    },
    {
      "epoch": 0.8323170731707317,
      "grad_norm": 0.03453477271523532,
      "learning_rate": 4.538076500367469e-05,
      "loss": 0.4339,
      "step": 273
    },
    {
      "epoch": 0.8353658536585366,
      "grad_norm": 0.033868608488940716,
      "learning_rate": 4.532924091140417e-05,
      "loss": 0.4343,
      "step": 274
    },
    {
      "epoch": 0.8384146341463414,
      "grad_norm": 0.040360291012410086,
      "learning_rate": 4.527746064580559e-05,
      "loss": 0.4379,
      "step": 275
    },
    {
      "epoch": 0.8414634146341463,
      "grad_norm": 0.03915688403846989,
      "learning_rate": 4.522542485937369e-05,
      "loss": 0.4507,
      "step": 276
    },
    {
      "epoch": 0.8445121951219512,
      "grad_norm": 0.03296011686492542,
      "learning_rate": 4.51731342078231e-05,
      "loss": 0.4712,
      "step": 277
    },
    {
      "epoch": 0.8475609756097561,
      "grad_norm": 0.03431476570558208,
      "learning_rate": 4.5120589350080045e-05,
      "loss": 0.4128,
      "step": 278
    },
    {
      "epoch": 0.850609756097561,
      "grad_norm": 0.03296989189683222,
      "learning_rate": 4.5067790948274094e-05,
      "loss": 0.4259,
      "step": 279
    },
    {
      "epoch": 0.8536585365853658,
      "grad_norm": 0.03888989456612432,
      "learning_rate": 4.501473966772974e-05,
      "loss": 0.4363,
      "step": 280
    },
    {
      "epoch": 0.8567073170731707,
      "grad_norm": 0.036013879189621865,
      "learning_rate": 4.4961436176958096e-05,
      "loss": 0.3842,
      "step": 281
    },
    {
      "epoch": 0.8597560975609756,
      "grad_norm": 0.03538488532386994,
      "learning_rate": 4.49078811476484e-05,
      "loss": 0.4286,
      "step": 282
    },
    {
      "epoch": 0.8628048780487805,
      "grad_norm": 0.029695892400757717,
      "learning_rate": 4.485407525465961e-05,
      "loss": 0.326,
      "step": 283
    },
    {
      "epoch": 0.8658536585365854,
      "grad_norm": 0.03618612518943503,
      "learning_rate": 4.480001917601185e-05,
      "loss": 0.4609,
      "step": 284
    },
    {
      "epoch": 0.8689024390243902,
      "grad_norm": 0.03408841436436889,
      "learning_rate": 4.47457135928779e-05,
      "loss": 0.3644,
      "step": 285
    },
    {
      "epoch": 0.8719512195121951,
      "grad_norm": 0.03633183392407226,
      "learning_rate": 4.469115918957462e-05,
      "loss": 0.4926,
      "step": 286
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.03228844576866147,
      "learning_rate": 4.463635665355427e-05,
      "loss": 0.4056,
      "step": 287
    },
    {
      "epoch": 0.8780487804878049,
      "grad_norm": 0.03273543543687427,
      "learning_rate": 4.458130667539592e-05,
      "loss": 0.5236,
      "step": 288
    },
    {
      "epoch": 0.8810975609756098,
      "grad_norm": 0.0367258255673261,
      "learning_rate": 4.4526009948796703e-05,
      "loss": 0.4307,
      "step": 289
    },
    {
      "epoch": 0.8841463414634146,
      "grad_norm": 0.03703121197597588,
      "learning_rate": 4.447046717056308e-05,
      "loss": 0.414,
      "step": 290
    },
    {
      "epoch": 0.8871951219512195,
      "grad_norm": 0.035618837315914385,
      "learning_rate": 4.4414679040602066e-05,
      "loss": 0.4478,
      "step": 291
    },
    {
      "epoch": 0.8902439024390244,
      "grad_norm": 0.03548278274265021,
      "learning_rate": 4.435864626191242e-05,
      "loss": 0.3895,
      "step": 292
    },
    {
      "epoch": 0.8932926829268293,
      "grad_norm": 0.037705061087907554,
      "learning_rate": 4.430236954057575e-05,
      "loss": 0.471,
      "step": 293
    },
    {
      "epoch": 0.8963414634146342,
      "grad_norm": 0.034130715604090416,
      "learning_rate": 4.4245849585747654e-05,
      "loss": 0.4321,
      "step": 294
    },
    {
      "epoch": 0.899390243902439,
      "grad_norm": 0.035178382604340114,
      "learning_rate": 4.4189087109648786e-05,
      "loss": 0.4426,
      "step": 295
    },
    {
      "epoch": 0.9024390243902439,
      "grad_norm": 0.033104141044304335,
      "learning_rate": 4.413208282755583e-05,
      "loss": 0.4569,
      "step": 296
    },
    {
      "epoch": 0.9054878048780488,
      "grad_norm": 0.033911276401756896,
      "learning_rate": 4.4074837457792565e-05,
      "loss": 0.3703,
      "step": 297
    },
    {
      "epoch": 0.9085365853658537,
      "grad_norm": 0.035965621143592474,
      "learning_rate": 4.4017351721720713e-05,
      "loss": 0.4585,
      "step": 298
    },
    {
      "epoch": 0.9115853658536586,
      "grad_norm": 0.03218777441907488,
      "learning_rate": 4.395962634373097e-05,
      "loss": 0.3582,
      "step": 299
    },
    {
      "epoch": 0.9146341463414634,
      "grad_norm": 0.0410793082328498,
      "learning_rate": 4.3901662051233746e-05,
      "loss": 0.4138,
      "step": 300
    },
    {
      "epoch": 0.9176829268292683,
      "grad_norm": 0.03352096380275321,
      "learning_rate": 4.384345957465014e-05,
      "loss": 0.3597,
      "step": 301
    },
    {
      "epoch": 0.9207317073170732,
      "grad_norm": 0.03722187766731886,
      "learning_rate": 4.3785019647402596e-05,
      "loss": 0.3942,
      "step": 302
    },
    {
      "epoch": 0.9237804878048781,
      "grad_norm": 0.0371788079208431,
      "learning_rate": 4.372634300590578e-05,
      "loss": 0.3483,
      "step": 303
    },
    {
      "epoch": 0.926829268292683,
      "grad_norm": 0.03611699462478195,
      "learning_rate": 4.36674303895572e-05,
      "loss": 0.4075,
      "step": 304
    },
    {
      "epoch": 0.9298780487804879,
      "grad_norm": 0.03723998605579338,
      "learning_rate": 4.360828254072796e-05,
      "loss": 0.4151,
      "step": 305
    },
    {
      "epoch": 0.9329268292682927,
      "grad_norm": 0.03959076079536807,
      "learning_rate": 4.35489002047534e-05,
      "loss": 0.4752,
      "step": 306
    },
    {
      "epoch": 0.9359756097560976,
      "grad_norm": 0.03766438805908114,
      "learning_rate": 4.348928412992365e-05,
      "loss": 0.4568,
      "step": 307
    },
    {
      "epoch": 0.9390243902439024,
      "grad_norm": 0.03778927701157187,
      "learning_rate": 4.3429435067474255e-05,
      "loss": 0.487,
      "step": 308
    },
    {
      "epoch": 0.9420731707317073,
      "grad_norm": 0.036336021394423754,
      "learning_rate": 4.336935377157668e-05,
      "loss": 0.4099,
      "step": 309
    },
    {
      "epoch": 0.9451219512195121,
      "grad_norm": 0.0431887089760307,
      "learning_rate": 4.330904099932883e-05,
      "loss": 0.497,
      "step": 310
    },
    {
      "epoch": 0.948170731707317,
      "grad_norm": 0.039944309237872226,
      "learning_rate": 4.324849751074549e-05,
      "loss": 0.443,
      "step": 311
    },
    {
      "epoch": 0.9512195121951219,
      "grad_norm": 0.03854560157459489,
      "learning_rate": 4.318772406874873e-05,
      "loss": 0.3739,
      "step": 312
    },
    {
      "epoch": 0.9542682926829268,
      "grad_norm": 0.03734951750084128,
      "learning_rate": 4.312672143915836e-05,
      "loss": 0.5006,
      "step": 313
    },
    {
      "epoch": 0.9573170731707317,
      "grad_norm": 0.03689350747693368,
      "learning_rate": 4.3065490390682186e-05,
      "loss": 0.3988,
      "step": 314
    },
    {
      "epoch": 0.9603658536585366,
      "grad_norm": 0.038314638731371316,
      "learning_rate": 4.30040316949064e-05,
      "loss": 0.424,
      "step": 315
    },
    {
      "epoch": 0.9634146341463414,
      "grad_norm": 0.04066874086781242,
      "learning_rate": 4.294234612628584e-05,
      "loss": 0.3774,
      "step": 316
    },
    {
      "epoch": 0.9664634146341463,
      "grad_norm": 0.04039695985914129,
      "learning_rate": 4.288043446213418e-05,
      "loss": 0.4616,
      "step": 317
    },
    {
      "epoch": 0.9695121951219512,
      "grad_norm": 0.03660869126341122,
      "learning_rate": 4.2818297482614215e-05,
      "loss": 0.4401,
      "step": 318
    },
    {
      "epoch": 0.9725609756097561,
      "grad_norm": 0.03562869147236029,
      "learning_rate": 4.275593597072796e-05,
      "loss": 0.361,
      "step": 319
    },
    {
      "epoch": 0.975609756097561,
      "grad_norm": 0.03711094322444415,
      "learning_rate": 4.269335071230683e-05,
      "loss": 0.3837,
      "step": 320
    },
    {
      "epoch": 0.9786585365853658,
      "grad_norm": 0.03892312477131053,
      "learning_rate": 4.263054249600172e-05,
      "loss": 0.4111,
      "step": 321
    },
    {
      "epoch": 0.9817073170731707,
      "grad_norm": 0.03865864546953116,
      "learning_rate": 4.2567512113273053e-05,
      "loss": 0.4433,
      "step": 322
    },
    {
      "epoch": 0.9847560975609756,
      "grad_norm": 0.03867539491346883,
      "learning_rate": 4.250426035838083e-05,
      "loss": 0.4212,
      "step": 323
    },
    {
      "epoch": 0.9878048780487805,
      "grad_norm": 0.040465376955033085,
      "learning_rate": 4.2440788028374624e-05,
      "loss": 0.4628,
      "step": 324
    },
    {
      "epoch": 0.9908536585365854,
      "grad_norm": 0.03704119504678098,
      "learning_rate": 4.23770959230835e-05,
      "loss": 0.4128,
      "step": 325
    },
    {
      "epoch": 0.9939024390243902,
      "grad_norm": 0.03839322577794025,
      "learning_rate": 4.231318484510598e-05,
      "loss": 0.44,
      "step": 326
    },
    {
      "epoch": 0.9969512195121951,
      "grad_norm": 0.03315102993940294,
      "learning_rate": 4.2249055599799905e-05,
      "loss": 0.4046,
      "step": 327
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.03625310265220718,
      "learning_rate": 4.218470899527229e-05,
      "loss": 0.4283,
      "step": 328
    },
    {
      "epoch": 1.0030487804878048,
      "grad_norm": 0.039661819040880925,
      "learning_rate": 4.212014584236914e-05,
      "loss": 0.4333,
      "step": 329
    },
    {
      "epoch": 1.0060975609756098,
      "grad_norm": 0.036493365868269324,
      "learning_rate": 4.2055366954665245e-05,
      "loss": 0.3777,
      "step": 330
    },
    {
      "epoch": 1.0091463414634145,
      "grad_norm": 0.035949114268750526,
      "learning_rate": 4.19903731484539e-05,
      "loss": 0.4056,
      "step": 331
    },
    {
      "epoch": 1.0121951219512195,
      "grad_norm": 0.038016945564005054,
      "learning_rate": 4.1925165242736645e-05,
      "loss": 0.3595,
      "step": 332
    },
    {
      "epoch": 1.0152439024390243,
      "grad_norm": 0.043854287453910595,
      "learning_rate": 4.1859744059212945e-05,
      "loss": 0.3806,
      "step": 333
    },
    {
      "epoch": 1.0182926829268293,
      "grad_norm": 0.032489628100366166,
      "learning_rate": 4.1794110422269825e-05,
      "loss": 0.3716,
      "step": 334
    },
    {
      "epoch": 1.021341463414634,
      "grad_norm": 0.034846328027673276,
      "learning_rate": 4.172826515897146e-05,
      "loss": 0.378,
      "step": 335
    },
    {
      "epoch": 1.024390243902439,
      "grad_norm": 0.03617467028978075,
      "learning_rate": 4.16622090990488e-05,
      "loss": 0.4319,
      "step": 336
    },
    {
      "epoch": 1.0274390243902438,
      "grad_norm": 0.036298981743650975,
      "learning_rate": 4.159594307488909e-05,
      "loss": 0.4165,
      "step": 337
    },
    {
      "epoch": 1.0304878048780488,
      "grad_norm": 0.04008248827120437,
      "learning_rate": 4.152946792152537e-05,
      "loss": 0.4598,
      "step": 338
    },
    {
      "epoch": 1.0335365853658536,
      "grad_norm": 0.03413084170917385,
      "learning_rate": 4.146278447662597e-05,
      "loss": 0.3911,
      "step": 339
    },
    {
      "epoch": 1.0365853658536586,
      "grad_norm": 0.0401261349416301,
      "learning_rate": 4.1395893580483946e-05,
      "loss": 0.4359,
      "step": 340
    },
    {
      "epoch": 1.0396341463414633,
      "grad_norm": 0.03712587827307479,
      "learning_rate": 4.13287960760065e-05,
      "loss": 0.3907,
      "step": 341
    },
    {
      "epoch": 1.0426829268292683,
      "grad_norm": 0.03662865750568463,
      "learning_rate": 4.1261492808704336e-05,
      "loss": 0.3809,
      "step": 342
    },
    {
      "epoch": 1.045731707317073,
      "grad_norm": 0.03818973631464467,
      "learning_rate": 4.1193984626681055e-05,
      "loss": 0.392,
      "step": 343
    },
    {
      "epoch": 1.048780487804878,
      "grad_norm": 0.039592831441071695,
      "learning_rate": 4.112627238062239e-05,
      "loss": 0.4329,
      "step": 344
    },
    {
      "epoch": 1.0518292682926829,
      "grad_norm": 0.034611760797123796,
      "learning_rate": 4.105835692378557e-05,
      "loss": 0.3833,
      "step": 345
    },
    {
      "epoch": 1.0548780487804879,
      "grad_norm": 0.041529978989313134,
      "learning_rate": 4.0990239111988495e-05,
      "loss": 0.4961,
      "step": 346
    },
    {
      "epoch": 1.0579268292682926,
      "grad_norm": 0.03915634524818885,
      "learning_rate": 4.092191980359902e-05,
      "loss": 0.367,
      "step": 347
    },
    {
      "epoch": 1.0609756097560976,
      "grad_norm": 0.039821471806743135,
      "learning_rate": 4.085339985952407e-05,
      "loss": 0.4019,
      "step": 348
    },
    {
      "epoch": 1.0640243902439024,
      "grad_norm": 0.0381647553606081,
      "learning_rate": 4.0784680143198836e-05,
      "loss": 0.3899,
      "step": 349
    },
    {
      "epoch": 1.0670731707317074,
      "grad_norm": 0.038813602489438916,
      "learning_rate": 4.071576152057592e-05,
      "loss": 0.4504,
      "step": 350
    },
    {
      "epoch": 1.0701219512195121,
      "grad_norm": 0.03759452805764784,
      "learning_rate": 4.064664486011433e-05,
      "loss": 0.4096,
      "step": 351
    },
    {
      "epoch": 1.0731707317073171,
      "grad_norm": 0.03559570239193551,
      "learning_rate": 4.057733103276862e-05,
      "loss": 0.4448,
      "step": 352
    },
    {
      "epoch": 1.076219512195122,
      "grad_norm": 0.03934894672014586,
      "learning_rate": 4.05078209119779e-05,
      "loss": 0.4315,
      "step": 353
    },
    {
      "epoch": 1.079268292682927,
      "grad_norm": 0.03536865777657963,
      "learning_rate": 4.04381153736548e-05,
      "loss": 0.3166,
      "step": 354
    },
    {
      "epoch": 1.0823170731707317,
      "grad_norm": 0.06112220739940428,
      "learning_rate": 4.036821529617445e-05,
      "loss": 0.4604,
      "step": 355
    },
    {
      "epoch": 1.0853658536585367,
      "grad_norm": 0.042098722892453,
      "learning_rate": 4.0298121560363425e-05,
      "loss": 0.467,
      "step": 356
    },
    {
      "epoch": 1.0884146341463414,
      "grad_norm": 0.033546501683242795,
      "learning_rate": 4.0227835049488615e-05,
      "loss": 0.4055,
      "step": 357
    },
    {
      "epoch": 1.0914634146341464,
      "grad_norm": 0.040738677792441114,
      "learning_rate": 4.015735664924613e-05,
      "loss": 0.4435,
      "step": 358
    },
    {
      "epoch": 1.0945121951219512,
      "grad_norm": 0.03997113538785167,
      "learning_rate": 4.00866872477501e-05,
      "loss": 0.4176,
      "step": 359
    },
    {
      "epoch": 1.0975609756097562,
      "grad_norm": 0.037012001536335506,
      "learning_rate": 4.0015827735521525e-05,
      "loss": 0.4083,
      "step": 360
    },
    {
      "epoch": 1.100609756097561,
      "grad_norm": 0.03848505076268115,
      "learning_rate": 3.994477900547703e-05,
      "loss": 0.4417,
      "step": 361
    },
    {
      "epoch": 1.103658536585366,
      "grad_norm": 0.041316815401570055,
      "learning_rate": 3.98735419529176e-05,
      "loss": 0.4736,
      "step": 362
    },
    {
      "epoch": 1.1067073170731707,
      "grad_norm": 0.03911772664638055,
      "learning_rate": 3.9802117475517333e-05,
      "loss": 0.4158,
      "step": 363
    },
    {
      "epoch": 1.1097560975609757,
      "grad_norm": 0.03522394343528472,
      "learning_rate": 3.973050647331209e-05,
      "loss": 0.4062,
      "step": 364
    },
    {
      "epoch": 1.1128048780487805,
      "grad_norm": 0.03658235791600257,
      "learning_rate": 3.9658709848688204e-05,
      "loss": 0.3585,
      "step": 365
    },
    {
      "epoch": 1.1158536585365855,
      "grad_norm": 0.03982898125923274,
      "learning_rate": 3.9586728506371036e-05,
      "loss": 0.4337,
      "step": 366
    },
    {
      "epoch": 1.1189024390243902,
      "grad_norm": 0.037684991163154595,
      "learning_rate": 3.951456335341365e-05,
      "loss": 0.3462,
      "step": 367
    },
    {
      "epoch": 1.1219512195121952,
      "grad_norm": 0.0397290039170204,
      "learning_rate": 3.944221529918534e-05,
      "loss": 0.4048,
      "step": 368
    },
    {
      "epoch": 1.125,
      "grad_norm": 0.040291593478486015,
      "learning_rate": 3.9369685255360175e-05,
      "loss": 0.4069,
      "step": 369
    },
    {
      "epoch": 1.1280487804878048,
      "grad_norm": 0.03850647603937439,
      "learning_rate": 3.929697413590551e-05,
      "loss": 0.3627,
      "step": 370
    },
    {
      "epoch": 1.1310975609756098,
      "grad_norm": 0.038135080456534534,
      "learning_rate": 3.92240828570705e-05,
      "loss": 0.4451,
      "step": 371
    },
    {
      "epoch": 1.1341463414634148,
      "grad_norm": 0.03816254670509036,
      "learning_rate": 3.9151012337374495e-05,
      "loss": 0.4146,
      "step": 372
    },
    {
      "epoch": 1.1371951219512195,
      "grad_norm": 0.044412156062597616,
      "learning_rate": 3.907776349759552e-05,
      "loss": 0.5105,
      "step": 373
    },
    {
      "epoch": 1.1402439024390243,
      "grad_norm": 0.04640808798970655,
      "learning_rate": 3.900433726075865e-05,
      "loss": 0.455,
      "step": 374
    },
    {
      "epoch": 1.1432926829268293,
      "grad_norm": 0.038192479317907146,
      "learning_rate": 3.893073455212438e-05,
      "loss": 0.3909,
      "step": 375
    },
    {
      "epoch": 1.146341463414634,
      "grad_norm": 0.04165289486551481,
      "learning_rate": 3.885695629917696e-05,
      "loss": 0.4276,
      "step": 376
    },
    {
      "epoch": 1.149390243902439,
      "grad_norm": 0.04119881876804503,
      "learning_rate": 3.878300343161272e-05,
      "loss": 0.3558,
      "step": 377
    },
    {
      "epoch": 1.1524390243902438,
      "grad_norm": 0.040337725300083456,
      "learning_rate": 3.8708876881328335e-05,
      "loss": 0.3624,
      "step": 378
    },
    {
      "epoch": 1.1554878048780488,
      "grad_norm": 0.04080570289033096,
      "learning_rate": 3.863457758240912e-05,
      "loss": 0.4121,
      "step": 379
    },
    {
      "epoch": 1.1585365853658536,
      "grad_norm": 0.046098474965416324,
      "learning_rate": 3.8560106471117205e-05,
      "loss": 0.442,
      "step": 380
    },
    {
      "epoch": 1.1615853658536586,
      "grad_norm": 0.04336491573458105,
      "learning_rate": 3.848546448587979e-05,
      "loss": 0.4463,
      "step": 381
    },
    {
      "epoch": 1.1646341463414633,
      "grad_norm": 0.0372102108830308,
      "learning_rate": 3.8410652567277274e-05,
      "loss": 0.4164,
      "step": 382
    },
    {
      "epoch": 1.1676829268292683,
      "grad_norm": 0.0411184143025625,
      "learning_rate": 3.8335671658031443e-05,
      "loss": 0.4219,
      "step": 383
    },
    {
      "epoch": 1.170731707317073,
      "grad_norm": 0.03761498632507343,
      "learning_rate": 3.826052270299356e-05,
      "loss": 0.3702,
      "step": 384
    },
    {
      "epoch": 1.173780487804878,
      "grad_norm": 0.04054206319013793,
      "learning_rate": 3.81852066491325e-05,
      "loss": 0.4193,
      "step": 385
    },
    {
      "epoch": 1.1768292682926829,
      "grad_norm": 0.04409636671558726,
      "learning_rate": 3.810972444552273e-05,
      "loss": 0.4673,
      "step": 386
    },
    {
      "epoch": 1.1798780487804879,
      "grad_norm": 0.05497073494441137,
      "learning_rate": 3.8034077043332463e-05,
      "loss": 0.393,
      "step": 387
    },
    {
      "epoch": 1.1829268292682926,
      "grad_norm": 0.035394828898099424,
      "learning_rate": 3.795826539581159e-05,
      "loss": 0.3822,
      "step": 388
    },
    {
      "epoch": 1.1859756097560976,
      "grad_norm": 0.04315585167542129,
      "learning_rate": 3.78822904582797e-05,
      "loss": 0.3952,
      "step": 389
    },
    {
      "epoch": 1.1890243902439024,
      "grad_norm": 0.04172163659394637,
      "learning_rate": 3.7806153188114026e-05,
      "loss": 0.4454,
      "step": 390
    },
    {
      "epoch": 1.1920731707317074,
      "grad_norm": 0.038328909237282424,
      "learning_rate": 3.7729854544737405e-05,
      "loss": 0.4894,
      "step": 391
    },
    {
      "epoch": 1.1951219512195121,
      "grad_norm": 0.04453016859851793,
      "learning_rate": 3.765339548960618e-05,
      "loss": 0.4578,
      "step": 392
    },
    {
      "epoch": 1.1981707317073171,
      "grad_norm": 0.03985170342712862,
      "learning_rate": 3.7576776986198065e-05,
      "loss": 0.4054,
      "step": 393
    },
    {
      "epoch": 1.201219512195122,
      "grad_norm": 0.0363919638493405,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.4151,
      "step": 394
    },
    {
      "epoch": 1.204268292682927,
      "grad_norm": 0.04296486186004094,
      "learning_rate": 3.742306549849605e-05,
      "loss": 0.4092,
      "step": 395
    },
    {
      "epoch": 1.2073170731707317,
      "grad_norm": 0.043399261442856184,
      "learning_rate": 3.7345974451155105e-05,
      "loss": 0.4375,
      "step": 396
    },
    {
      "epoch": 1.2103658536585367,
      "grad_norm": 0.04546310712631861,
      "learning_rate": 3.72687278294188e-05,
      "loss": 0.3944,
      "step": 397
    },
    {
      "epoch": 1.2134146341463414,
      "grad_norm": 0.03869718679031596,
      "learning_rate": 3.719132660668912e-05,
      "loss": 0.38,
      "step": 398
    },
    {
      "epoch": 1.2164634146341464,
      "grad_norm": 0.04423908976746331,
      "learning_rate": 3.711377175831626e-05,
      "loss": 0.4413,
      "step": 399
    },
    {
      "epoch": 1.2195121951219512,
      "grad_norm": 0.03992059124798192,
      "learning_rate": 3.703606426158627e-05,
      "loss": 0.3975,
      "step": 400
    },
    {
      "epoch": 1.2225609756097562,
      "grad_norm": 0.04234926941131931,
      "learning_rate": 3.695820509570876e-05,
      "loss": 0.4182,
      "step": 401
    },
    {
      "epoch": 1.225609756097561,
      "grad_norm": 0.03752584303283549,
      "learning_rate": 3.688019524180457e-05,
      "loss": 0.4336,
      "step": 402
    },
    {
      "epoch": 1.228658536585366,
      "grad_norm": 0.040254488930115774,
      "learning_rate": 3.680203568289335e-05,
      "loss": 0.3102,
      "step": 403
    },
    {
      "epoch": 1.2317073170731707,
      "grad_norm": 0.04126452897719161,
      "learning_rate": 3.6723727403881284e-05,
      "loss": 0.4864,
      "step": 404
    },
    {
      "epoch": 1.2347560975609757,
      "grad_norm": 0.0397168573533502,
      "learning_rate": 3.664527139154854e-05,
      "loss": 0.39,
      "step": 405
    },
    {
      "epoch": 1.2378048780487805,
      "grad_norm": 0.04394740053425026,
      "learning_rate": 3.656666863453697e-05,
      "loss": 0.4457,
      "step": 406
    },
    {
      "epoch": 1.2408536585365852,
      "grad_norm": 0.040165587324300686,
      "learning_rate": 3.6487920123337526e-05,
      "loss": 0.4072,
      "step": 407
    },
    {
      "epoch": 1.2439024390243902,
      "grad_norm": 0.039573002429541836,
      "learning_rate": 3.6409026850277904e-05,
      "loss": 0.4209,
      "step": 408
    },
    {
      "epoch": 1.2469512195121952,
      "grad_norm": 0.04325973648718029,
      "learning_rate": 3.632998980950993e-05,
      "loss": 0.4865,
      "step": 409
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.045634562526459706,
      "learning_rate": 3.625080999699711e-05,
      "loss": 0.3904,
      "step": 410
    },
    {
      "epoch": 1.2530487804878048,
      "grad_norm": 0.04352830294344096,
      "learning_rate": 3.617148841050202e-05,
      "loss": 0.435,
      "step": 411
    },
    {
      "epoch": 1.2560975609756098,
      "grad_norm": 0.04556248919890376,
      "learning_rate": 3.609202604957379e-05,
      "loss": 0.4506,
      "step": 412
    },
    {
      "epoch": 1.2591463414634148,
      "grad_norm": 0.038127895927304076,
      "learning_rate": 3.601242391553546e-05,
      "loss": 0.3466,
      "step": 413
    },
    {
      "epoch": 1.2621951219512195,
      "grad_norm": 0.03894730371478106,
      "learning_rate": 3.593268301147139e-05,
      "loss": 0.4473,
      "step": 414
    },
    {
      "epoch": 1.2652439024390243,
      "grad_norm": 0.03946795235138391,
      "learning_rate": 3.585280434221461e-05,
      "loss": 0.456,
      "step": 415
    },
    {
      "epoch": 1.2682926829268293,
      "grad_norm": 0.04521220263241203,
      "learning_rate": 3.5772788914334165e-05,
      "loss": 0.4479,
      "step": 416
    },
    {
      "epoch": 1.2713414634146343,
      "grad_norm": 0.04497082961254816,
      "learning_rate": 3.5692637736122425e-05,
      "loss": 0.4769,
      "step": 417
    },
    {
      "epoch": 1.274390243902439,
      "grad_norm": 0.041729507940111016,
      "learning_rate": 3.5612351817582375e-05,
      "loss": 0.4383,
      "step": 418
    },
    {
      "epoch": 1.2774390243902438,
      "grad_norm": 0.04030254563385526,
      "learning_rate": 3.5531932170414896e-05,
      "loss": 0.3696,
      "step": 419
    },
    {
      "epoch": 1.2804878048780488,
      "grad_norm": 0.04350184053951066,
      "learning_rate": 3.545137980800601e-05,
      "loss": 0.4084,
      "step": 420
    },
    {
      "epoch": 1.2835365853658536,
      "grad_norm": 0.046162386459619606,
      "learning_rate": 3.5370695745414137e-05,
      "loss": 0.4124,
      "step": 421
    },
    {
      "epoch": 1.2865853658536586,
      "grad_norm": 0.04399090038019498,
      "learning_rate": 3.5289880999357235e-05,
      "loss": 0.433,
      "step": 422
    },
    {
      "epoch": 1.2896341463414633,
      "grad_norm": 0.0398583981998761,
      "learning_rate": 3.520893658820007e-05,
      "loss": 0.484,
      "step": 423
    },
    {
      "epoch": 1.2926829268292683,
      "grad_norm": 0.049515626274668,
      "learning_rate": 3.512786353194134e-05,
      "loss": 0.4273,
      "step": 424
    },
    {
      "epoch": 1.295731707317073,
      "grad_norm": 0.051021698402422024,
      "learning_rate": 3.504666285220083e-05,
      "loss": 0.3773,
      "step": 425
    },
    {
      "epoch": 1.298780487804878,
      "grad_norm": 0.04752736955001905,
      "learning_rate": 3.496533557220651e-05,
      "loss": 0.38,
      "step": 426
    },
    {
      "epoch": 1.3018292682926829,
      "grad_norm": 0.03855841961124006,
      "learning_rate": 3.488388271678173e-05,
      "loss": 0.3691,
      "step": 427
    },
    {
      "epoch": 1.3048780487804879,
      "grad_norm": 0.042396695569388354,
      "learning_rate": 3.480230531233216e-05,
      "loss": 0.3916,
      "step": 428
    },
    {
      "epoch": 1.3079268292682926,
      "grad_norm": 0.04273608922242111,
      "learning_rate": 3.472060438683302e-05,
      "loss": 0.3981,
      "step": 429
    },
    {
      "epoch": 1.3109756097560976,
      "grad_norm": 0.049523730946746974,
      "learning_rate": 3.4638780969816e-05,
      "loss": 0.42,
      "step": 430
    },
    {
      "epoch": 1.3140243902439024,
      "grad_norm": 0.040081605113161875,
      "learning_rate": 3.455683609235633e-05,
      "loss": 0.3995,
      "step": 431
    },
    {
      "epoch": 1.3170731707317074,
      "grad_norm": 0.045376936177541204,
      "learning_rate": 3.447477078705983e-05,
      "loss": 0.3677,
      "step": 432
    },
    {
      "epoch": 1.3201219512195121,
      "grad_norm": 0.04334373283790506,
      "learning_rate": 3.439258608804983e-05,
      "loss": 0.506,
      "step": 433
    },
    {
      "epoch": 1.3231707317073171,
      "grad_norm": 0.04180544711303642,
      "learning_rate": 3.431028303095415e-05,
      "loss": 0.4131,
      "step": 434
    },
    {
      "epoch": 1.326219512195122,
      "grad_norm": 0.040548583882156285,
      "learning_rate": 3.422786265289211e-05,
      "loss": 0.4098,
      "step": 435
    },
    {
      "epoch": 1.329268292682927,
      "grad_norm": 0.0481011172486242,
      "learning_rate": 3.414532599246138e-05,
      "loss": 0.4053,
      "step": 436
    },
    {
      "epoch": 1.3323170731707317,
      "grad_norm": 0.04072709256507153,
      "learning_rate": 3.4062674089724944e-05,
      "loss": 0.4115,
      "step": 437
    },
    {
      "epoch": 1.3353658536585367,
      "grad_norm": 0.04386676391625629,
      "learning_rate": 3.3979907986197996e-05,
      "loss": 0.4421,
      "step": 438
    },
    {
      "epoch": 1.3384146341463414,
      "grad_norm": 0.04109122248944048,
      "learning_rate": 3.389702872483477e-05,
      "loss": 0.4108,
      "step": 439
    },
    {
      "epoch": 1.3414634146341464,
      "grad_norm": 0.03919269914070091,
      "learning_rate": 3.381403735001546e-05,
      "loss": 0.3833,
      "step": 440
    },
    {
      "epoch": 1.3445121951219512,
      "grad_norm": 0.04239685630936478,
      "learning_rate": 3.3730934907533e-05,
      "loss": 0.4242,
      "step": 441
    },
    {
      "epoch": 1.3475609756097562,
      "grad_norm": 0.038258355769465166,
      "learning_rate": 3.364772244457991e-05,
      "loss": 0.3938,
      "step": 442
    },
    {
      "epoch": 1.350609756097561,
      "grad_norm": 0.04096483905702994,
      "learning_rate": 3.356440100973513e-05,
      "loss": 0.4106,
      "step": 443
    },
    {
      "epoch": 1.3536585365853657,
      "grad_norm": 0.047393278834682626,
      "learning_rate": 3.348097165295076e-05,
      "loss": 0.5145,
      "step": 444
    },
    {
      "epoch": 1.3567073170731707,
      "grad_norm": 0.04633439099806606,
      "learning_rate": 3.3397435425538845e-05,
      "loss": 0.3849,
      "step": 445
    },
    {
      "epoch": 1.3597560975609757,
      "grad_norm": 0.039579677850455526,
      "learning_rate": 3.331379338015815e-05,
      "loss": 0.383,
      "step": 446
    },
    {
      "epoch": 1.3628048780487805,
      "grad_norm": 0.039540731669071356,
      "learning_rate": 3.3230046570800864e-05,
      "loss": 0.3098,
      "step": 447
    },
    {
      "epoch": 1.3658536585365852,
      "grad_norm": 0.044430392635034954,
      "learning_rate": 3.3146196052779314e-05,
      "loss": 0.4134,
      "step": 448
    },
    {
      "epoch": 1.3689024390243902,
      "grad_norm": 0.0461209856378901,
      "learning_rate": 3.3062242882712724e-05,
      "loss": 0.4762,
      "step": 449
    },
    {
      "epoch": 1.3719512195121952,
      "grad_norm": 0.04095847664841931,
      "learning_rate": 3.297818811851382e-05,
      "loss": 0.3562,
      "step": 450
    },
    {
      "epoch": 1.375,
      "grad_norm": 0.04167488392555382,
      "learning_rate": 3.289403281937554e-05,
      "loss": 0.3533,
      "step": 451
    },
    {
      "epoch": 1.3780487804878048,
      "grad_norm": 0.04319674461159365,
      "learning_rate": 3.280977804575773e-05,
      "loss": 0.3939,
      "step": 452
    },
    {
      "epoch": 1.3810975609756098,
      "grad_norm": 0.04464600230901868,
      "learning_rate": 3.272542485937369e-05,
      "loss": 0.4278,
      "step": 453
    },
    {
      "epoch": 1.3841463414634148,
      "grad_norm": 0.044632110302958494,
      "learning_rate": 3.2640974323176846e-05,
      "loss": 0.4559,
      "step": 454
    },
    {
      "epoch": 1.3871951219512195,
      "grad_norm": 0.041606750321008176,
      "learning_rate": 3.255642750134738e-05,
      "loss": 0.3438,
      "step": 455
    },
    {
      "epoch": 1.3902439024390243,
      "grad_norm": 0.046396118274942486,
      "learning_rate": 3.2471785459278756e-05,
      "loss": 0.3536,
      "step": 456
    },
    {
      "epoch": 1.3932926829268293,
      "grad_norm": 0.04967698041003248,
      "learning_rate": 3.2387049263564365e-05,
      "loss": 0.4603,
      "step": 457
    },
    {
      "epoch": 1.3963414634146343,
      "grad_norm": 0.04423036929072222,
      "learning_rate": 3.2302219981984014e-05,
      "loss": 0.4108,
      "step": 458
    },
    {
      "epoch": 1.399390243902439,
      "grad_norm": 0.04692015853572872,
      "learning_rate": 3.2217298683490525e-05,
      "loss": 0.44,
      "step": 459
    },
    {
      "epoch": 1.4024390243902438,
      "grad_norm": 0.051126417646320296,
      "learning_rate": 3.2132286438196245e-05,
      "loss": 0.3984,
      "step": 460
    },
    {
      "epoch": 1.4054878048780488,
      "grad_norm": 0.05132235831429306,
      "learning_rate": 3.2047184317359566e-05,
      "loss": 0.4838,
      "step": 461
    },
    {
      "epoch": 1.4085365853658536,
      "grad_norm": 0.0420060500286704,
      "learning_rate": 3.19619933933714e-05,
      "loss": 0.4112,
      "step": 462
    },
    {
      "epoch": 1.4115853658536586,
      "grad_norm": 0.042290123347956725,
      "learning_rate": 3.187671473974172e-05,
      "loss": 0.3994,
      "step": 463
    },
    {
      "epoch": 1.4146341463414633,
      "grad_norm": 0.040748174494785384,
      "learning_rate": 3.179134943108597e-05,
      "loss": 0.3853,
      "step": 464
    },
    {
      "epoch": 1.4176829268292683,
      "grad_norm": 0.04059345235444476,
      "learning_rate": 3.1705898543111575e-05,
      "loss": 0.4084,
      "step": 465
    },
    {
      "epoch": 1.420731707317073,
      "grad_norm": 0.055103011348927346,
      "learning_rate": 3.1620363152604354e-05,
      "loss": 0.4206,
      "step": 466
    },
    {
      "epoch": 1.423780487804878,
      "grad_norm": 0.04158726319207225,
      "learning_rate": 3.153474433741496e-05,
      "loss": 0.3494,
      "step": 467
    },
    {
      "epoch": 1.4268292682926829,
      "grad_norm": 0.04706275403580267,
      "learning_rate": 3.14490431764453e-05,
      "loss": 0.4605,
      "step": 468
    },
    {
      "epoch": 1.4298780487804879,
      "grad_norm": 0.044939505391916955,
      "learning_rate": 3.136326074963494e-05,
      "loss": 0.3611,
      "step": 469
    },
    {
      "epoch": 1.4329268292682926,
      "grad_norm": 0.046842293410835734,
      "learning_rate": 3.1277398137947516e-05,
      "loss": 0.5025,
      "step": 470
    },
    {
      "epoch": 1.4359756097560976,
      "grad_norm": 0.05269740887022689,
      "learning_rate": 3.119145642335705e-05,
      "loss": 0.4378,
      "step": 471
    },
    {
      "epoch": 1.4390243902439024,
      "grad_norm": 0.04498556893820305,
      "learning_rate": 3.110543668883438e-05,
      "loss": 0.5231,
      "step": 472
    },
    {
      "epoch": 1.4420731707317074,
      "grad_norm": 0.041432180188026094,
      "learning_rate": 3.10193400183335e-05,
      "loss": 0.4327,
      "step": 473
    },
    {
      "epoch": 1.4451219512195121,
      "grad_norm": 0.05125275934132454,
      "learning_rate": 3.093316749677788e-05,
      "loss": 0.4129,
      "step": 474
    },
    {
      "epoch": 1.4481707317073171,
      "grad_norm": 0.044066353635275675,
      "learning_rate": 3.0846920210046796e-05,
      "loss": 0.4087,
      "step": 475
    },
    {
      "epoch": 1.451219512195122,
      "grad_norm": 0.04350573606760448,
      "learning_rate": 3.076059924496167e-05,
      "loss": 0.4113,
      "step": 476
    },
    {
      "epoch": 1.454268292682927,
      "grad_norm": 0.039982387903032636,
      "learning_rate": 3.0674205689272375e-05,
      "loss": 0.407,
      "step": 477
    },
    {
      "epoch": 1.4573170731707317,
      "grad_norm": 0.046576254516377445,
      "learning_rate": 3.0587740631643495e-05,
      "loss": 0.4671,
      "step": 478
    },
    {
      "epoch": 1.4603658536585367,
      "grad_norm": 0.04440553082667708,
      "learning_rate": 3.050120516164062e-05,
      "loss": 0.3768,
      "step": 479
    },
    {
      "epoch": 1.4634146341463414,
      "grad_norm": 0.041495054277947314,
      "learning_rate": 3.041460036971664e-05,
      "loss": 0.375,
      "step": 480
    },
    {
      "epoch": 1.4664634146341464,
      "grad_norm": 0.048010338112052395,
      "learning_rate": 3.0327927347197967e-05,
      "loss": 0.4223,
      "step": 481
    },
    {
      "epoch": 1.4695121951219512,
      "grad_norm": 0.04399903311566183,
      "learning_rate": 3.024118718627083e-05,
      "loss": 0.4412,
      "step": 482
    },
    {
      "epoch": 1.4725609756097562,
      "grad_norm": 0.043332913853267936,
      "learning_rate": 3.0154380979967456e-05,
      "loss": 0.4122,
      "step": 483
    },
    {
      "epoch": 1.475609756097561,
      "grad_norm": 0.04979870765528592,
      "learning_rate": 3.006750982215234e-05,
      "loss": 0.4247,
      "step": 484
    },
    {
      "epoch": 1.4786585365853657,
      "grad_norm": 0.04830840196710952,
      "learning_rate": 2.9980574807508443e-05,
      "loss": 0.4054,
      "step": 485
    },
    {
      "epoch": 1.4817073170731707,
      "grad_norm": 0.04493403266733893,
      "learning_rate": 2.98935770315234e-05,
      "loss": 0.4192,
      "step": 486
    },
    {
      "epoch": 1.4847560975609757,
      "grad_norm": 0.04629449676150017,
      "learning_rate": 2.9806517590475707e-05,
      "loss": 0.4177,
      "step": 487
    },
    {
      "epoch": 1.4878048780487805,
      "grad_norm": 0.056540345358252,
      "learning_rate": 2.9719397581420932e-05,
      "loss": 0.3933,
      "step": 488
    },
    {
      "epoch": 1.4908536585365852,
      "grad_norm": 0.044714125976439284,
      "learning_rate": 2.9632218102177862e-05,
      "loss": 0.4328,
      "step": 489
    },
    {
      "epoch": 1.4939024390243902,
      "grad_norm": 0.04581859990723505,
      "learning_rate": 2.954498025131468e-05,
      "loss": 0.4384,
      "step": 490
    },
    {
      "epoch": 1.4969512195121952,
      "grad_norm": 0.04460433072726923,
      "learning_rate": 2.9457685128135136e-05,
      "loss": 0.4242,
      "step": 491
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.043427642512756305,
      "learning_rate": 2.937033383266466e-05,
      "loss": 0.4057,
      "step": 492
    },
    {
      "epoch": 1.5030487804878048,
      "grad_norm": 0.044045306137113785,
      "learning_rate": 2.9282927465636547e-05,
      "loss": 0.4315,
      "step": 493
    },
    {
      "epoch": 1.5060975609756098,
      "grad_norm": 0.046958499553942205,
      "learning_rate": 2.9195467128478044e-05,
      "loss": 0.4443,
      "step": 494
    },
    {
      "epoch": 1.5091463414634148,
      "grad_norm": 0.04320799042144439,
      "learning_rate": 2.9107953923296488e-05,
      "loss": 0.3808,
      "step": 495
    },
    {
      "epoch": 1.5121951219512195,
      "grad_norm": 0.04109444723893145,
      "learning_rate": 2.9020388952865423e-05,
      "loss": 0.3906,
      "step": 496
    },
    {
      "epoch": 1.5152439024390243,
      "grad_norm": 0.049646393822732345,
      "learning_rate": 2.89327733206107e-05,
      "loss": 0.4675,
      "step": 497
    },
    {
      "epoch": 1.5182926829268293,
      "grad_norm": 0.042206105467635986,
      "learning_rate": 2.884510813059657e-05,
      "loss": 0.385,
      "step": 498
    },
    {
      "epoch": 1.5213414634146343,
      "grad_norm": 0.04242629061154276,
      "learning_rate": 2.875739448751176e-05,
      "loss": 0.4142,
      "step": 499
    },
    {
      "epoch": 1.524390243902439,
      "grad_norm": 0.04508722521275613,
      "learning_rate": 2.866963349665559e-05,
      "loss": 0.4049,
      "step": 500
    },
    {
      "epoch": 1.5274390243902438,
      "grad_norm": 0.04226906162135462,
      "learning_rate": 2.8581826263923993e-05,
      "loss": 0.3626,
      "step": 501
    },
    {
      "epoch": 1.5304878048780488,
      "grad_norm": 0.04361915151450245,
      "learning_rate": 2.849397389579564e-05,
      "loss": 0.4577,
      "step": 502
    },
    {
      "epoch": 1.5335365853658538,
      "grad_norm": 0.04649015076624045,
      "learning_rate": 2.8406077499317912e-05,
      "loss": 0.3971,
      "step": 503
    },
    {
      "epoch": 1.5365853658536586,
      "grad_norm": 0.05228072608444702,
      "learning_rate": 2.8318138182093052e-05,
      "loss": 0.4525,
      "step": 504
    },
    {
      "epoch": 1.5396341463414633,
      "grad_norm": 0.03764229090200894,
      "learning_rate": 2.8230157052264127e-05,
      "loss": 0.31,
      "step": 505
    },
    {
      "epoch": 1.5426829268292683,
      "grad_norm": 0.04785158948760201,
      "learning_rate": 2.8142135218501116e-05,
      "loss": 0.4055,
      "step": 506
    },
    {
      "epoch": 1.5457317073170733,
      "grad_norm": 0.04267301527731101,
      "learning_rate": 2.8054073789986883e-05,
      "loss": 0.4174,
      "step": 507
    },
    {
      "epoch": 1.548780487804878,
      "grad_norm": 0.049619578175530116,
      "learning_rate": 2.796597387640328e-05,
      "loss": 0.431,
      "step": 508
    },
    {
      "epoch": 1.5518292682926829,
      "grad_norm": 0.04530815374929139,
      "learning_rate": 2.7877836587917072e-05,
      "loss": 0.4632,
      "step": 509
    },
    {
      "epoch": 1.5548780487804879,
      "grad_norm": 0.042163759856576145,
      "learning_rate": 2.7789663035166034e-05,
      "loss": 0.426,
      "step": 510
    },
    {
      "epoch": 1.5579268292682928,
      "grad_norm": 0.04282500694359744,
      "learning_rate": 2.770145432924489e-05,
      "loss": 0.4057,
      "step": 511
    },
    {
      "epoch": 1.5609756097560976,
      "grad_norm": 0.04368531663527112,
      "learning_rate": 2.761321158169134e-05,
      "loss": 0.4924,
      "step": 512
    },
    {
      "epoch": 1.5640243902439024,
      "grad_norm": 0.04332755295462451,
      "learning_rate": 2.7524935904472054e-05,
      "loss": 0.3845,
      "step": 513
    },
    {
      "epoch": 1.5670731707317072,
      "grad_norm": 0.0502150335523684,
      "learning_rate": 2.7436628409968664e-05,
      "loss": 0.416,
      "step": 514
    },
    {
      "epoch": 1.5701219512195121,
      "grad_norm": 0.04826923090280151,
      "learning_rate": 2.7348290210963723e-05,
      "loss": 0.3989,
      "step": 515
    },
    {
      "epoch": 1.5731707317073171,
      "grad_norm": 0.04054443877837543,
      "learning_rate": 2.7259922420626705e-05,
      "loss": 0.4146,
      "step": 516
    },
    {
      "epoch": 1.576219512195122,
      "grad_norm": 0.04116187892553754,
      "learning_rate": 2.7171526152499978e-05,
      "loss": 0.3766,
      "step": 517
    },
    {
      "epoch": 1.5792682926829267,
      "grad_norm": 0.04717761075694039,
      "learning_rate": 2.708310252048476e-05,
      "loss": 0.439,
      "step": 518
    },
    {
      "epoch": 1.5823170731707317,
      "grad_norm": 0.045062272355279945,
      "learning_rate": 2.6994652638827078e-05,
      "loss": 0.3296,
      "step": 519
    },
    {
      "epoch": 1.5853658536585367,
      "grad_norm": 0.04444154739237893,
      "learning_rate": 2.6906177622103752e-05,
      "loss": 0.4123,
      "step": 520
    },
    {
      "epoch": 1.5884146341463414,
      "grad_norm": 0.046567991907138076,
      "learning_rate": 2.6817678585208327e-05,
      "loss": 0.3991,
      "step": 521
    },
    {
      "epoch": 1.5914634146341462,
      "grad_norm": 0.043720022432954565,
      "learning_rate": 2.672915664333704e-05,
      "loss": 0.3864,
      "step": 522
    },
    {
      "epoch": 1.5945121951219512,
      "grad_norm": 0.0514525696971225,
      "learning_rate": 2.6640612911974733e-05,
      "loss": 0.4243,
      "step": 523
    },
    {
      "epoch": 1.5975609756097562,
      "grad_norm": 0.043857465570069955,
      "learning_rate": 2.655204850688085e-05,
      "loss": 0.4074,
      "step": 524
    },
    {
      "epoch": 1.600609756097561,
      "grad_norm": 0.04552516973100716,
      "learning_rate": 2.6463464544075344e-05,
      "loss": 0.4147,
      "step": 525
    },
    {
      "epoch": 1.6036585365853657,
      "grad_norm": 0.049473995207474815,
      "learning_rate": 2.6374862139824618e-05,
      "loss": 0.4709,
      "step": 526
    },
    {
      "epoch": 1.6067073170731707,
      "grad_norm": 0.04820818030504909,
      "learning_rate": 2.628624241062744e-05,
      "loss": 0.4169,
      "step": 527
    },
    {
      "epoch": 1.6097560975609757,
      "grad_norm": 0.050450825910037425,
      "learning_rate": 2.6197606473200924e-05,
      "loss": 0.5224,
      "step": 528
    },
    {
      "epoch": 1.6128048780487805,
      "grad_norm": 0.04550000497484494,
      "learning_rate": 2.610895544446641e-05,
      "loss": 0.4492,
      "step": 529
    },
    {
      "epoch": 1.6158536585365852,
      "grad_norm": 0.04200381471392651,
      "learning_rate": 2.6020290441535417e-05,
      "loss": 0.3978,
      "step": 530
    },
    {
      "epoch": 1.6189024390243902,
      "grad_norm": 0.04822670808590297,
      "learning_rate": 2.5931612581695537e-05,
      "loss": 0.4099,
      "step": 531
    },
    {
      "epoch": 1.6219512195121952,
      "grad_norm": 0.04500687842031658,
      "learning_rate": 2.5842922982396395e-05,
      "loss": 0.3042,
      "step": 532
    },
    {
      "epoch": 1.625,
      "grad_norm": 0.05239766736226129,
      "learning_rate": 2.5754222761235535e-05,
      "loss": 0.4081,
      "step": 533
    },
    {
      "epoch": 1.6280487804878048,
      "grad_norm": 0.04602743501218835,
      "learning_rate": 2.566551303594437e-05,
      "loss": 0.3982,
      "step": 534
    },
    {
      "epoch": 1.6310975609756098,
      "grad_norm": 0.049006753443694435,
      "learning_rate": 2.557679492437404e-05,
      "loss": 0.4353,
      "step": 535
    },
    {
      "epoch": 1.6341463414634148,
      "grad_norm": 0.04588961339707511,
      "learning_rate": 2.54880695444814e-05,
      "loss": 0.4979,
      "step": 536
    },
    {
      "epoch": 1.6371951219512195,
      "grad_norm": 0.043562453539744625,
      "learning_rate": 2.539933801431487e-05,
      "loss": 0.4229,
      "step": 537
    },
    {
      "epoch": 1.6402439024390243,
      "grad_norm": 0.04108293903835419,
      "learning_rate": 2.5310601452000383e-05,
      "loss": 0.3284,
      "step": 538
    },
    {
      "epoch": 1.6432926829268293,
      "grad_norm": 0.04784461798032342,
      "learning_rate": 2.5221860975727275e-05,
      "loss": 0.416,
      "step": 539
    },
    {
      "epoch": 1.6463414634146343,
      "grad_norm": 0.055611309808498224,
      "learning_rate": 2.5133117703734212e-05,
      "loss": 0.4033,
      "step": 540
    },
    {
      "epoch": 1.649390243902439,
      "grad_norm": 0.042125329738103255,
      "learning_rate": 2.504437275429508e-05,
      "loss": 0.4342,
      "step": 541
    },
    {
      "epoch": 1.6524390243902438,
      "grad_norm": 0.043957874421772736,
      "learning_rate": 2.4955627245704918e-05,
      "loss": 0.3941,
      "step": 542
    },
    {
      "epoch": 1.6554878048780488,
      "grad_norm": 0.049378467127122065,
      "learning_rate": 2.4866882296265794e-05,
      "loss": 0.4344,
      "step": 543
    },
    {
      "epoch": 1.6585365853658538,
      "grad_norm": 0.05139749501618102,
      "learning_rate": 2.4778139024272724e-05,
      "loss": 0.45,
      "step": 544
    },
    {
      "epoch": 1.6615853658536586,
      "grad_norm": 0.04540333353423334,
      "learning_rate": 2.468939854799962e-05,
      "loss": 0.422,
      "step": 545
    },
    {
      "epoch": 1.6646341463414633,
      "grad_norm": 0.04781977767818636,
      "learning_rate": 2.4600661985685132e-05,
      "loss": 0.4636,
      "step": 546
    },
    {
      "epoch": 1.6676829268292683,
      "grad_norm": 0.04791256952355809,
      "learning_rate": 2.4511930455518605e-05,
      "loss": 0.3755,
      "step": 547
    },
    {
      "epoch": 1.6707317073170733,
      "grad_norm": 0.04622700218481359,
      "learning_rate": 2.4423205075625955e-05,
      "loss": 0.4243,
      "step": 548
    },
    {
      "epoch": 1.673780487804878,
      "grad_norm": 0.0535233632473163,
      "learning_rate": 2.433448696405563e-05,
      "loss": 0.4589,
      "step": 549
    },
    {
      "epoch": 1.6768292682926829,
      "grad_norm": 0.04930305119098795,
      "learning_rate": 2.424577723876446e-05,
      "loss": 0.4351,
      "step": 550
    },
    {
      "epoch": 1.6798780487804879,
      "grad_norm": 0.050525972340958514,
      "learning_rate": 2.415707701760361e-05,
      "loss": 0.3527,
      "step": 551
    },
    {
      "epoch": 1.6829268292682928,
      "grad_norm": 0.04658338095694016,
      "learning_rate": 2.4068387418304462e-05,
      "loss": 0.419,
      "step": 552
    },
    {
      "epoch": 1.6859756097560976,
      "grad_norm": 0.0459308673167424,
      "learning_rate": 2.397970955846459e-05,
      "loss": 0.3947,
      "step": 553
    },
    {
      "epoch": 1.6890243902439024,
      "grad_norm": 0.04483255537595823,
      "learning_rate": 2.3891044555533588e-05,
      "loss": 0.3736,
      "step": 554
    },
    {
      "epoch": 1.6920731707317072,
      "grad_norm": 0.045520663386516586,
      "learning_rate": 2.380239352679908e-05,
      "loss": 0.39,
      "step": 555
    },
    {
      "epoch": 1.6951219512195121,
      "grad_norm": 0.05169918882299193,
      "learning_rate": 2.3713757589372557e-05,
      "loss": 0.5003,
      "step": 556
    },
    {
      "epoch": 1.6981707317073171,
      "grad_norm": 0.0461102833224252,
      "learning_rate": 2.3625137860175385e-05,
      "loss": 0.4633,
      "step": 557
    },
    {
      "epoch": 1.701219512195122,
      "grad_norm": 0.04613663732250752,
      "learning_rate": 2.3536535455924655e-05,
      "loss": 0.3678,
      "step": 558
    },
    {
      "epoch": 1.7042682926829267,
      "grad_norm": 0.041673432713452745,
      "learning_rate": 2.3447951493119152e-05,
      "loss": 0.4118,
      "step": 559
    },
    {
      "epoch": 1.7073170731707317,
      "grad_norm": 0.04431275177649695,
      "learning_rate": 2.3359387088025266e-05,
      "loss": 0.3673,
      "step": 560
    },
    {
      "epoch": 1.7103658536585367,
      "grad_norm": 0.06367990054330565,
      "learning_rate": 2.327084335666297e-05,
      "loss": 0.4002,
      "step": 561
    },
    {
      "epoch": 1.7134146341463414,
      "grad_norm": 0.048050644260211994,
      "learning_rate": 2.3182321414791672e-05,
      "loss": 0.4634,
      "step": 562
    },
    {
      "epoch": 1.7164634146341462,
      "grad_norm": 0.04923233496697778,
      "learning_rate": 2.309382237789625e-05,
      "loss": 0.3871,
      "step": 563
    },
    {
      "epoch": 1.7195121951219512,
      "grad_norm": 0.05364103691899248,
      "learning_rate": 2.300534736117292e-05,
      "loss": 0.4803,
      "step": 564
    },
    {
      "epoch": 1.7225609756097562,
      "grad_norm": 0.048579593446969827,
      "learning_rate": 2.2916897479515244e-05,
      "loss": 0.4196,
      "step": 565
    },
    {
      "epoch": 1.725609756097561,
      "grad_norm": 0.0461136811151973,
      "learning_rate": 2.2828473847500024e-05,
      "loss": 0.4513,
      "step": 566
    },
    {
      "epoch": 1.7286585365853657,
      "grad_norm": 0.04940628622994503,
      "learning_rate": 2.27400775793733e-05,
      "loss": 0.4208,
      "step": 567
    },
    {
      "epoch": 1.7317073170731707,
      "grad_norm": 0.04972357687250847,
      "learning_rate": 2.265170978903628e-05,
      "loss": 0.3931,
      "step": 568
    },
    {
      "epoch": 1.7347560975609757,
      "grad_norm": 0.0444698806113803,
      "learning_rate": 2.256337159003134e-05,
      "loss": 0.435,
      "step": 569
    },
    {
      "epoch": 1.7378048780487805,
      "grad_norm": 0.04255075307190636,
      "learning_rate": 2.2475064095527948e-05,
      "loss": 0.3906,
      "step": 570
    },
    {
      "epoch": 1.7408536585365852,
      "grad_norm": 0.04535567324227825,
      "learning_rate": 2.238678841830867e-05,
      "loss": 0.3901,
      "step": 571
    },
    {
      "epoch": 1.7439024390243902,
      "grad_norm": 0.041821101755344973,
      "learning_rate": 2.229854567075511e-05,
      "loss": 0.3818,
      "step": 572
    },
    {
      "epoch": 1.7469512195121952,
      "grad_norm": 0.04572808806362685,
      "learning_rate": 2.2210336964833968e-05,
      "loss": 0.4141,
      "step": 573
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.04622786206882963,
      "learning_rate": 2.2122163412082927e-05,
      "loss": 0.5004,
      "step": 574
    },
    {
      "epoch": 1.7530487804878048,
      "grad_norm": 0.041755188750875076,
      "learning_rate": 2.203402612359673e-05,
      "loss": 0.369,
      "step": 575
    },
    {
      "epoch": 1.7560975609756098,
      "grad_norm": 0.04754782156490283,
      "learning_rate": 2.1945926210013112e-05,
      "loss": 0.4536,
      "step": 576
    },
    {
      "epoch": 1.7591463414634148,
      "grad_norm": 0.050433137791283965,
      "learning_rate": 2.185786478149889e-05,
      "loss": 0.4506,
      "step": 577
    },
    {
      "epoch": 1.7621951219512195,
      "grad_norm": 0.04384522351312468,
      "learning_rate": 2.1769842947735876e-05,
      "loss": 0.3789,
      "step": 578
    },
    {
      "epoch": 1.7652439024390243,
      "grad_norm": 0.051693646701183,
      "learning_rate": 2.1681861817906954e-05,
      "loss": 0.4029,
      "step": 579
    },
    {
      "epoch": 1.7682926829268293,
      "grad_norm": 0.04878335306364114,
      "learning_rate": 2.1593922500682087e-05,
      "loss": 0.424,
      "step": 580
    },
    {
      "epoch": 1.7713414634146343,
      "grad_norm": 0.050862021634430564,
      "learning_rate": 2.150602610420437e-05,
      "loss": 0.3739,
      "step": 581
    },
    {
      "epoch": 1.774390243902439,
      "grad_norm": 0.04605607998701594,
      "learning_rate": 2.141817373607601e-05,
      "loss": 0.4437,
      "step": 582
    },
    {
      "epoch": 1.7774390243902438,
      "grad_norm": 0.051381647631965954,
      "learning_rate": 2.1330366503344415e-05,
      "loss": 0.3766,
      "step": 583
    },
    {
      "epoch": 1.7804878048780488,
      "grad_norm": 0.05272844434906908,
      "learning_rate": 2.1242605512488248e-05,
      "loss": 0.3929,
      "step": 584
    },
    {
      "epoch": 1.7835365853658538,
      "grad_norm": 0.05311986437136711,
      "learning_rate": 2.1154891869403435e-05,
      "loss": 0.4092,
      "step": 585
    },
    {
      "epoch": 1.7865853658536586,
      "grad_norm": 0.05502515824691199,
      "learning_rate": 2.1067226679389306e-05,
      "loss": 0.4104,
      "step": 586
    },
    {
      "epoch": 1.7896341463414633,
      "grad_norm": 0.044502014868971776,
      "learning_rate": 2.097961104713458e-05,
      "loss": 0.3761,
      "step": 587
    },
    {
      "epoch": 1.7926829268292683,
      "grad_norm": 0.04774171126141972,
      "learning_rate": 2.089204607670352e-05,
      "loss": 0.4288,
      "step": 588
    },
    {
      "epoch": 1.7957317073170733,
      "grad_norm": 0.045225558382394004,
      "learning_rate": 2.080453287152196e-05,
      "loss": 0.4577,
      "step": 589
    },
    {
      "epoch": 1.798780487804878,
      "grad_norm": 0.04835401158227742,
      "learning_rate": 2.0717072534363456e-05,
      "loss": 0.4534,
      "step": 590
    },
    {
      "epoch": 1.8018292682926829,
      "grad_norm": 0.045025439868481315,
      "learning_rate": 2.0629666167335344e-05,
      "loss": 0.4066,
      "step": 591
    },
    {
      "epoch": 1.8048780487804879,
      "grad_norm": 0.049456094850128966,
      "learning_rate": 2.0542314871864877e-05,
      "loss": 0.3749,
      "step": 592
    },
    {
      "epoch": 1.8079268292682928,
      "grad_norm": 0.04201275484991142,
      "learning_rate": 2.0455019748685322e-05,
      "loss": 0.3862,
      "step": 593
    },
    {
      "epoch": 1.8109756097560976,
      "grad_norm": 0.04819807866358233,
      "learning_rate": 2.0367781897822147e-05,
      "loss": 0.4145,
      "step": 594
    },
    {
      "epoch": 1.8140243902439024,
      "grad_norm": 0.045799317561101595,
      "learning_rate": 2.0280602418579074e-05,
      "loss": 0.3821,
      "step": 595
    },
    {
      "epoch": 1.8170731707317072,
      "grad_norm": 0.04751188506756341,
      "learning_rate": 2.01934824095243e-05,
      "loss": 0.4254,
      "step": 596
    },
    {
      "epoch": 1.8201219512195121,
      "grad_norm": 0.04593743100660145,
      "learning_rate": 2.0106422968476606e-05,
      "loss": 0.3731,
      "step": 597
    },
    {
      "epoch": 1.8231707317073171,
      "grad_norm": 0.04921016078030088,
      "learning_rate": 2.0019425192491563e-05,
      "loss": 0.3503,
      "step": 598
    },
    {
      "epoch": 1.826219512195122,
      "grad_norm": 0.045878356807498875,
      "learning_rate": 1.993249017784766e-05,
      "loss": 0.3802,
      "step": 599
    },
    {
      "epoch": 1.8292682926829267,
      "grad_norm": 0.05193684501902608,
      "learning_rate": 1.9845619020032553e-05,
      "loss": 0.4092,
      "step": 600
    },
    {
      "epoch": 1.8323170731707317,
      "grad_norm": 0.05027784444090783,
      "learning_rate": 1.975881281372917e-05,
      "loss": 0.4722,
      "step": 601
    },
    {
      "epoch": 1.8353658536585367,
      "grad_norm": 0.04514834962956475,
      "learning_rate": 1.9672072652802036e-05,
      "loss": 0.46,
      "step": 602
    },
    {
      "epoch": 1.8384146341463414,
      "grad_norm": 0.05451938225894702,
      "learning_rate": 1.958539963028337e-05,
      "loss": 0.4368,
      "step": 603
    },
    {
      "epoch": 1.8414634146341462,
      "grad_norm": 0.040106913103446724,
      "learning_rate": 1.949879483835939e-05,
      "loss": 0.3257,
      "step": 604
    },
    {
      "epoch": 1.8445121951219512,
      "grad_norm": 0.04926311870177447,
      "learning_rate": 1.9412259368356507e-05,
      "loss": 0.4183,
      "step": 605
    },
    {
      "epoch": 1.8475609756097562,
      "grad_norm": 0.04422209752149198,
      "learning_rate": 1.9325794310727627e-05,
      "loss": 0.3639,
      "step": 606
    },
    {
      "epoch": 1.850609756097561,
      "grad_norm": 0.048679581480980755,
      "learning_rate": 1.923940075503833e-05,
      "loss": 0.4157,
      "step": 607
    },
    {
      "epoch": 1.8536585365853657,
      "grad_norm": 0.05499776323028079,
      "learning_rate": 1.9153079789953217e-05,
      "loss": 0.4656,
      "step": 608
    },
    {
      "epoch": 1.8567073170731707,
      "grad_norm": 0.06922562178056044,
      "learning_rate": 1.9066832503222128e-05,
      "loss": 0.3813,
      "step": 609
    },
    {
      "epoch": 1.8597560975609757,
      "grad_norm": 0.04395445795822402,
      "learning_rate": 1.898065998166651e-05,
      "loss": 0.3411,
      "step": 610
    },
    {
      "epoch": 1.8628048780487805,
      "grad_norm": 0.04207124907862645,
      "learning_rate": 1.8894563311165626e-05,
      "loss": 0.3592,
      "step": 611
    },
    {
      "epoch": 1.8658536585365852,
      "grad_norm": 0.05054981775482249,
      "learning_rate": 1.8808543576642963e-05,
      "loss": 0.4706,
      "step": 612
    },
    {
      "epoch": 1.8689024390243902,
      "grad_norm": 0.05019817217403753,
      "learning_rate": 1.872260186205249e-05,
      "loss": 0.4213,
      "step": 613
    },
    {
      "epoch": 1.8719512195121952,
      "grad_norm": 0.0568861338806866,
      "learning_rate": 1.863673925036506e-05,
      "loss": 0.3852,
      "step": 614
    },
    {
      "epoch": 1.875,
      "grad_norm": 0.04821125446126471,
      "learning_rate": 1.8550956823554707e-05,
      "loss": 0.4304,
      "step": 615
    },
    {
      "epoch": 1.8780487804878048,
      "grad_norm": 0.047092779650319404,
      "learning_rate": 1.8465255662585053e-05,
      "loss": 0.4227,
      "step": 616
    },
    {
      "epoch": 1.8810975609756098,
      "grad_norm": 0.056212137174365545,
      "learning_rate": 1.8379636847395655e-05,
      "loss": 0.4862,
      "step": 617
    },
    {
      "epoch": 1.8841463414634148,
      "grad_norm": 0.057774981328737894,
      "learning_rate": 1.829410145688843e-05,
      "loss": 0.452,
      "step": 618
    },
    {
      "epoch": 1.8871951219512195,
      "grad_norm": 0.053132370814335673,
      "learning_rate": 1.8208650568914033e-05,
      "loss": 0.4183,
      "step": 619
    },
    {
      "epoch": 1.8902439024390243,
      "grad_norm": 0.04462487543154932,
      "learning_rate": 1.812328526025829e-05,
      "loss": 0.3585,
      "step": 620
    },
    {
      "epoch": 1.8932926829268293,
      "grad_norm": 0.04333309998936559,
      "learning_rate": 1.80380066066286e-05,
      "loss": 0.4107,
      "step": 621
    },
    {
      "epoch": 1.8963414634146343,
      "grad_norm": 0.05489562977125714,
      "learning_rate": 1.795281568264044e-05,
      "loss": 0.396,
      "step": 622
    },
    {
      "epoch": 1.899390243902439,
      "grad_norm": 0.048000398098398726,
      "learning_rate": 1.7867713561803758e-05,
      "loss": 0.4085,
      "step": 623
    },
    {
      "epoch": 1.9024390243902438,
      "grad_norm": 0.048282605255221815,
      "learning_rate": 1.778270131650948e-05,
      "loss": 0.3363,
      "step": 624
    },
    {
      "epoch": 1.9054878048780488,
      "grad_norm": 0.05716822504858685,
      "learning_rate": 1.7697780018015985e-05,
      "loss": 0.4386,
      "step": 625
    },
    {
      "epoch": 1.9085365853658538,
      "grad_norm": 0.04794543022496826,
      "learning_rate": 1.761295073643564e-05,
      "loss": 0.3821,
      "step": 626
    },
    {
      "epoch": 1.9115853658536586,
      "grad_norm": 0.053218946583905274,
      "learning_rate": 1.752821454072124e-05,
      "loss": 0.4508,
      "step": 627
    },
    {
      "epoch": 1.9146341463414633,
      "grad_norm": 0.049908536883833524,
      "learning_rate": 1.7443572498652626e-05,
      "loss": 0.3659,
      "step": 628
    },
    {
      "epoch": 1.9176829268292683,
      "grad_norm": 0.04638106740630859,
      "learning_rate": 1.735902567682315e-05,
      "loss": 0.391,
      "step": 629
    },
    {
      "epoch": 1.9207317073170733,
      "grad_norm": 0.040265654554186424,
      "learning_rate": 1.7274575140626318e-05,
      "loss": 0.3738,
      "step": 630
    },
    {
      "epoch": 1.923780487804878,
      "grad_norm": 0.04596084571609373,
      "learning_rate": 1.719022195424227e-05,
      "loss": 0.3629,
      "step": 631
    },
    {
      "epoch": 1.9268292682926829,
      "grad_norm": 0.045024822777817844,
      "learning_rate": 1.710596718062446e-05,
      "loss": 0.4327,
      "step": 632
    },
    {
      "epoch": 1.9298780487804879,
      "grad_norm": 0.04732414273870265,
      "learning_rate": 1.7021811881486186e-05,
      "loss": 0.4537,
      "step": 633
    },
    {
      "epoch": 1.9329268292682928,
      "grad_norm": 0.059279293870337126,
      "learning_rate": 1.6937757117287278e-05,
      "loss": 0.384,
      "step": 634
    },
    {
      "epoch": 1.9359756097560976,
      "grad_norm": 0.054739792171745454,
      "learning_rate": 1.685380394722068e-05,
      "loss": 0.4748,
      "step": 635
    },
    {
      "epoch": 1.9390243902439024,
      "grad_norm": 0.05392035736905421,
      "learning_rate": 1.676995342919914e-05,
      "loss": 0.4268,
      "step": 636
    },
    {
      "epoch": 1.9420731707317072,
      "grad_norm": 0.054225627701012045,
      "learning_rate": 1.6686206619841844e-05,
      "loss": 0.4547,
      "step": 637
    },
    {
      "epoch": 1.9451219512195121,
      "grad_norm": 0.04760347853249887,
      "learning_rate": 1.6602564574461154e-05,
      "loss": 0.363,
      "step": 638
    },
    {
      "epoch": 1.9481707317073171,
      "grad_norm": 0.04725199886274887,
      "learning_rate": 1.651902834704924e-05,
      "loss": 0.4266,
      "step": 639
    },
    {
      "epoch": 1.951219512195122,
      "grad_norm": 0.04599398148342138,
      "learning_rate": 1.6435598990264877e-05,
      "loss": 0.4332,
      "step": 640
    },
    {
      "epoch": 1.9542682926829267,
      "grad_norm": 0.04607811862896777,
      "learning_rate": 1.635227755542009e-05,
      "loss": 0.3435,
      "step": 641
    },
    {
      "epoch": 1.9573170731707317,
      "grad_norm": 0.04753790065032661,
      "learning_rate": 1.626906509246701e-05,
      "loss": 0.3416,
      "step": 642
    },
    {
      "epoch": 1.9603658536585367,
      "grad_norm": 0.042068522853175805,
      "learning_rate": 1.6185962649984542e-05,
      "loss": 0.3723,
      "step": 643
    },
    {
      "epoch": 1.9634146341463414,
      "grad_norm": 0.05315456275684181,
      "learning_rate": 1.6102971275165228e-05,
      "loss": 0.4321,
      "step": 644
    },
    {
      "epoch": 1.9664634146341462,
      "grad_norm": 0.04863243472200057,
      "learning_rate": 1.6020092013802003e-05,
      "loss": 0.4262,
      "step": 645
    },
    {
      "epoch": 1.9695121951219512,
      "grad_norm": 0.04645463781107319,
      "learning_rate": 1.593732591027506e-05,
      "loss": 0.4141,
      "step": 646
    },
    {
      "epoch": 1.9725609756097562,
      "grad_norm": 0.04630974142223754,
      "learning_rate": 1.585467400753863e-05,
      "loss": 0.4527,
      "step": 647
    },
    {
      "epoch": 1.975609756097561,
      "grad_norm": 0.04647892123563765,
      "learning_rate": 1.57721373471079e-05,
      "loss": 0.3633,
      "step": 648
    },
    {
      "epoch": 1.9786585365853657,
      "grad_norm": 0.054863057319935696,
      "learning_rate": 1.5689716969045848e-05,
      "loss": 0.4567,
      "step": 649
    },
    {
      "epoch": 1.9817073170731707,
      "grad_norm": 0.047476319487593684,
      "learning_rate": 1.5607413911950177e-05,
      "loss": 0.3634,
      "step": 650
    },
    {
      "epoch": 1.9847560975609757,
      "grad_norm": 0.05498886155793677,
      "learning_rate": 1.5525229212940167e-05,
      "loss": 0.4834,
      "step": 651
    },
    {
      "epoch": 1.9878048780487805,
      "grad_norm": 0.04146670137787259,
      "learning_rate": 1.544316390764367e-05,
      "loss": 0.3592,
      "step": 652
    },
    {
      "epoch": 1.9908536585365852,
      "grad_norm": 0.04448592709004556,
      "learning_rate": 1.5361219030184005e-05,
      "loss": 0.3881,
      "step": 653
    },
    {
      "epoch": 1.9939024390243902,
      "grad_norm": 0.04921396944888872,
      "learning_rate": 1.5279395613166986e-05,
      "loss": 0.3893,
      "step": 654
    },
    {
      "epoch": 1.9969512195121952,
      "grad_norm": 0.04399703962269756,
      "learning_rate": 1.519769468766784e-05,
      "loss": 0.3972,
      "step": 655
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.05225803349065435,
      "learning_rate": 1.5116117283218284e-05,
      "loss": 0.3982,
      "step": 656
    },
    {
      "epoch": 2.0030487804878048,
      "grad_norm": 0.047057971420935575,
      "learning_rate": 1.5034664427793485e-05,
      "loss": 0.3979,
      "step": 657
    },
    {
      "epoch": 2.0060975609756095,
      "grad_norm": 0.039924127005037634,
      "learning_rate": 1.495333714779918e-05,
      "loss": 0.3487,
      "step": 658
    },
    {
      "epoch": 2.0091463414634148,
      "grad_norm": 0.05227671668342273,
      "learning_rate": 1.4872136468058661e-05,
      "loss": 0.4956,
      "step": 659
    },
    {
      "epoch": 2.0121951219512195,
      "grad_norm": 0.04465904605331701,
      "learning_rate": 1.4791063411799938e-05,
      "loss": 0.3784,
      "step": 660
    },
    {
      "epoch": 2.0152439024390243,
      "grad_norm": 0.04843350015064368,
      "learning_rate": 1.471011900064277e-05,
      "loss": 0.4202,
      "step": 661
    },
    {
      "epoch": 2.018292682926829,
      "grad_norm": 0.04711367212265629,
      "learning_rate": 1.4629304254585874e-05,
      "loss": 0.4296,
      "step": 662
    },
    {
      "epoch": 2.0213414634146343,
      "grad_norm": 0.051282377670104844,
      "learning_rate": 1.4548620191994e-05,
      "loss": 0.405,
      "step": 663
    },
    {
      "epoch": 2.024390243902439,
      "grad_norm": 0.04516937696168246,
      "learning_rate": 1.4468067829585108e-05,
      "loss": 0.3849,
      "step": 664
    },
    {
      "epoch": 2.027439024390244,
      "grad_norm": 0.10404174025420722,
      "learning_rate": 1.4387648182417634e-05,
      "loss": 0.3091,
      "step": 665
    },
    {
      "epoch": 2.0304878048780486,
      "grad_norm": 0.04851496450594152,
      "learning_rate": 1.4307362263877582e-05,
      "loss": 0.3326,
      "step": 666
    },
    {
      "epoch": 2.033536585365854,
      "grad_norm": 0.04897987402822538,
      "learning_rate": 1.4227211085665843e-05,
      "loss": 0.3917,
      "step": 667
    },
    {
      "epoch": 2.0365853658536586,
      "grad_norm": 0.05250855661884853,
      "learning_rate": 1.4147195657785394e-05,
      "loss": 0.413,
      "step": 668
    },
    {
      "epoch": 2.0396341463414633,
      "grad_norm": 0.04485267998459715,
      "learning_rate": 1.4067316988528617e-05,
      "loss": 0.414,
      "step": 669
    },
    {
      "epoch": 2.042682926829268,
      "grad_norm": 0.05050211588874968,
      "learning_rate": 1.3987576084464544e-05,
      "loss": 0.4114,
      "step": 670
    },
    {
      "epoch": 2.0457317073170733,
      "grad_norm": 0.049062688965448895,
      "learning_rate": 1.3907973950426223e-05,
      "loss": 0.436,
      "step": 671
    },
    {
      "epoch": 2.048780487804878,
      "grad_norm": 0.04934816967138291,
      "learning_rate": 1.3828511589497979e-05,
      "loss": 0.3806,
      "step": 672
    },
    {
      "epoch": 2.051829268292683,
      "grad_norm": 0.04846301665690881,
      "learning_rate": 1.37491900030029e-05,
      "loss": 0.3782,
      "step": 673
    },
    {
      "epoch": 2.0548780487804876,
      "grad_norm": 0.0487751457696009,
      "learning_rate": 1.3670010190490073e-05,
      "loss": 0.3684,
      "step": 674
    },
    {
      "epoch": 2.057926829268293,
      "grad_norm": 0.05011236219413779,
      "learning_rate": 1.3590973149722103e-05,
      "loss": 0.4557,
      "step": 675
    },
    {
      "epoch": 2.0609756097560976,
      "grad_norm": 0.0433306137614156,
      "learning_rate": 1.3512079876662476e-05,
      "loss": 0.398,
      "step": 676
    },
    {
      "epoch": 2.0640243902439024,
      "grad_norm": 0.04720276909475184,
      "learning_rate": 1.3433331365463042e-05,
      "loss": 0.4615,
      "step": 677
    },
    {
      "epoch": 2.067073170731707,
      "grad_norm": 0.051702815490710055,
      "learning_rate": 1.3354728608451462e-05,
      "loss": 0.4537,
      "step": 678
    },
    {
      "epoch": 2.0701219512195124,
      "grad_norm": 0.049061114526012986,
      "learning_rate": 1.327627259611873e-05,
      "loss": 0.3737,
      "step": 679
    },
    {
      "epoch": 2.073170731707317,
      "grad_norm": 0.05016460137745102,
      "learning_rate": 1.3197964317106645e-05,
      "loss": 0.4094,
      "step": 680
    },
    {
      "epoch": 2.076219512195122,
      "grad_norm": 0.04496857184174542,
      "learning_rate": 1.3119804758195443e-05,
      "loss": 0.3212,
      "step": 681
    },
    {
      "epoch": 2.0792682926829267,
      "grad_norm": 0.04565158070603903,
      "learning_rate": 1.3041794904291244e-05,
      "loss": 0.3478,
      "step": 682
    },
    {
      "epoch": 2.082317073170732,
      "grad_norm": 0.05118300622487708,
      "learning_rate": 1.2963935738413737e-05,
      "loss": 0.4798,
      "step": 683
    },
    {
      "epoch": 2.0853658536585367,
      "grad_norm": 0.041350376374063275,
      "learning_rate": 1.2886228241683749e-05,
      "loss": 0.3147,
      "step": 684
    },
    {
      "epoch": 2.0884146341463414,
      "grad_norm": 0.05145828794964563,
      "learning_rate": 1.2808673393310888e-05,
      "loss": 0.4323,
      "step": 685
    },
    {
      "epoch": 2.091463414634146,
      "grad_norm": 0.046360376128903494,
      "learning_rate": 1.2731272170581204e-05,
      "loss": 0.4051,
      "step": 686
    },
    {
      "epoch": 2.0945121951219514,
      "grad_norm": 0.04566477398251874,
      "learning_rate": 1.26540255488449e-05,
      "loss": 0.4433,
      "step": 687
    },
    {
      "epoch": 2.097560975609756,
      "grad_norm": 0.04654212012479589,
      "learning_rate": 1.2576934501503957e-05,
      "loss": 0.3933,
      "step": 688
    },
    {
      "epoch": 2.100609756097561,
      "grad_norm": 0.044474358744242146,
      "learning_rate": 1.2500000000000006e-05,
      "loss": 0.3542,
      "step": 689
    },
    {
      "epoch": 2.1036585365853657,
      "grad_norm": 0.08086234904184278,
      "learning_rate": 1.2423223013801946e-05,
      "loss": 0.4152,
      "step": 690
    },
    {
      "epoch": 2.106707317073171,
      "grad_norm": 0.04578439970593931,
      "learning_rate": 1.2346604510393821e-05,
      "loss": 0.3802,
      "step": 691
    },
    {
      "epoch": 2.1097560975609757,
      "grad_norm": 0.045355970498272065,
      "learning_rate": 1.2270145455262594e-05,
      "loss": 0.4265,
      "step": 692
    },
    {
      "epoch": 2.1128048780487805,
      "grad_norm": 0.046546974689879166,
      "learning_rate": 1.2193846811885978e-05,
      "loss": 0.4156,
      "step": 693
    },
    {
      "epoch": 2.1158536585365852,
      "grad_norm": 0.04928163113897007,
      "learning_rate": 1.2117709541720306e-05,
      "loss": 0.4878,
      "step": 694
    },
    {
      "epoch": 2.1189024390243905,
      "grad_norm": 0.04679052446479647,
      "learning_rate": 1.204173460418842e-05,
      "loss": 0.495,
      "step": 695
    },
    {
      "epoch": 2.1219512195121952,
      "grad_norm": 0.060718380403180575,
      "learning_rate": 1.1965922956667536e-05,
      "loss": 0.4401,
      "step": 696
    },
    {
      "epoch": 2.125,
      "grad_norm": 0.04352857277050566,
      "learning_rate": 1.1890275554477278e-05,
      "loss": 0.4842,
      "step": 697
    },
    {
      "epoch": 2.1280487804878048,
      "grad_norm": 0.047483675913447015,
      "learning_rate": 1.1814793350867511e-05,
      "loss": 0.4075,
      "step": 698
    },
    {
      "epoch": 2.1310975609756095,
      "grad_norm": 0.044241685897127424,
      "learning_rate": 1.173947729700644e-05,
      "loss": 0.4077,
      "step": 699
    },
    {
      "epoch": 2.1341463414634148,
      "grad_norm": 0.04856861739756317,
      "learning_rate": 1.1664328341968564e-05,
      "loss": 0.4312,
      "step": 700
    },
    {
      "epoch": 2.1371951219512195,
      "grad_norm": 0.04330635555839946,
      "learning_rate": 1.1589347432722734e-05,
      "loss": 0.3697,
      "step": 701
    },
    {
      "epoch": 2.1402439024390243,
      "grad_norm": 0.047497521607233716,
      "learning_rate": 1.1514535514120217e-05,
      "loss": 0.4053,
      "step": 702
    },
    {
      "epoch": 2.143292682926829,
      "grad_norm": 0.06097670818580457,
      "learning_rate": 1.1439893528882803e-05,
      "loss": 0.471,
      "step": 703
    },
    {
      "epoch": 2.1463414634146343,
      "grad_norm": 0.04556035518561954,
      "learning_rate": 1.1365422417590878e-05,
      "loss": 0.4122,
      "step": 704
    },
    {
      "epoch": 2.149390243902439,
      "grad_norm": 0.052903392746045705,
      "learning_rate": 1.1291123118671665e-05,
      "loss": 0.3995,
      "step": 705
    },
    {
      "epoch": 2.152439024390244,
      "grad_norm": 0.04777545800892784,
      "learning_rate": 1.1216996568387284e-05,
      "loss": 0.4108,
      "step": 706
    },
    {
      "epoch": 2.1554878048780486,
      "grad_norm": 0.04820369744975627,
      "learning_rate": 1.1143043700823039e-05,
      "loss": 0.5138,
      "step": 707
    },
    {
      "epoch": 2.158536585365854,
      "grad_norm": 0.04305534978570109,
      "learning_rate": 1.1069265447875618e-05,
      "loss": 0.3452,
      "step": 708
    },
    {
      "epoch": 2.1615853658536586,
      "grad_norm": 0.049468850693703,
      "learning_rate": 1.0995662739241347e-05,
      "loss": 0.4621,
      "step": 709
    },
    {
      "epoch": 2.1646341463414633,
      "grad_norm": 0.051986125434277566,
      "learning_rate": 1.092223650240448e-05,
      "loss": 0.3518,
      "step": 710
    },
    {
      "epoch": 2.167682926829268,
      "grad_norm": 0.045548926482217825,
      "learning_rate": 1.0848987662625515e-05,
      "loss": 0.3681,
      "step": 711
    },
    {
      "epoch": 2.1707317073170733,
      "grad_norm": 0.051992076545282505,
      "learning_rate": 1.0775917142929501e-05,
      "loss": 0.4154,
      "step": 712
    },
    {
      "epoch": 2.173780487804878,
      "grad_norm": 0.048057122442053686,
      "learning_rate": 1.0703025864094494e-05,
      "loss": 0.4414,
      "step": 713
    },
    {
      "epoch": 2.176829268292683,
      "grad_norm": 0.05323915865529368,
      "learning_rate": 1.063031474463983e-05,
      "loss": 0.409,
      "step": 714
    },
    {
      "epoch": 2.1798780487804876,
      "grad_norm": 0.04632056825317641,
      "learning_rate": 1.0557784700814663e-05,
      "loss": 0.3771,
      "step": 715
    },
    {
      "epoch": 2.182926829268293,
      "grad_norm": 0.041087767138815914,
      "learning_rate": 1.0485436646586349e-05,
      "loss": 0.3533,
      "step": 716
    },
    {
      "epoch": 2.1859756097560976,
      "grad_norm": 0.04858659822072648,
      "learning_rate": 1.0413271493628965e-05,
      "loss": 0.3772,
      "step": 717
    },
    {
      "epoch": 2.1890243902439024,
      "grad_norm": 0.04566631819528302,
      "learning_rate": 1.03412901513118e-05,
      "loss": 0.4482,
      "step": 718
    },
    {
      "epoch": 2.192073170731707,
      "grad_norm": 0.05591022083048744,
      "learning_rate": 1.0269493526687915e-05,
      "loss": 0.4326,
      "step": 719
    },
    {
      "epoch": 2.1951219512195124,
      "grad_norm": 0.04838645871812315,
      "learning_rate": 1.0197882524482669e-05,
      "loss": 0.3609,
      "step": 720
    },
    {
      "epoch": 2.198170731707317,
      "grad_norm": 0.047677260894461095,
      "learning_rate": 1.0126458047082405e-05,
      "loss": 0.4344,
      "step": 721
    },
    {
      "epoch": 2.201219512195122,
      "grad_norm": 0.049886331786466606,
      "learning_rate": 1.0055220994522977e-05,
      "loss": 0.4234,
      "step": 722
    },
    {
      "epoch": 2.2042682926829267,
      "grad_norm": 0.05109177654726143,
      "learning_rate": 9.984172264478476e-06,
      "loss": 0.4465,
      "step": 723
    },
    {
      "epoch": 2.207317073170732,
      "grad_norm": 0.05367217987712921,
      "learning_rate": 9.913312752249903e-06,
      "loss": 0.472,
      "step": 724
    },
    {
      "epoch": 2.2103658536585367,
      "grad_norm": 0.04842354910885158,
      "learning_rate": 9.842643350753877e-06,
      "loss": 0.3958,
      "step": 725
    },
    {
      "epoch": 2.2134146341463414,
      "grad_norm": 0.04554936876201704,
      "learning_rate": 9.772164950511387e-06,
      "loss": 0.3897,
      "step": 726
    },
    {
      "epoch": 2.216463414634146,
      "grad_norm": 0.051586649959186465,
      "learning_rate": 9.701878439636586e-06,
      "loss": 0.434,
      "step": 727
    },
    {
      "epoch": 2.2195121951219514,
      "grad_norm": 0.047742332583085245,
      "learning_rate": 9.631784703825547e-06,
      "loss": 0.4433,
      "step": 728
    },
    {
      "epoch": 2.222560975609756,
      "grad_norm": 0.04567788571876452,
      "learning_rate": 9.561884626345205e-06,
      "loss": 0.4215,
      "step": 729
    },
    {
      "epoch": 2.225609756097561,
      "grad_norm": 0.04315451101759397,
      "learning_rate": 9.492179088022102e-06,
      "loss": 0.3366,
      "step": 730
    },
    {
      "epoch": 2.2286585365853657,
      "grad_norm": 0.0488469738860539,
      "learning_rate": 9.422668967231379e-06,
      "loss": 0.4103,
      "step": 731
    },
    {
      "epoch": 2.231707317073171,
      "grad_norm": 0.06002146852937906,
      "learning_rate": 9.353355139885673e-06,
      "loss": 0.4298,
      "step": 732
    },
    {
      "epoch": 2.2347560975609757,
      "grad_norm": 0.044891336885385796,
      "learning_rate": 9.284238479424081e-06,
      "loss": 0.3422,
      "step": 733
    },
    {
      "epoch": 2.2378048780487805,
      "grad_norm": 0.04393420507675689,
      "learning_rate": 9.215319856801158e-06,
      "loss": 0.3937,
      "step": 734
    },
    {
      "epoch": 2.2408536585365852,
      "grad_norm": 0.047101192915488074,
      "learning_rate": 9.146600140475945e-06,
      "loss": 0.4188,
      "step": 735
    },
    {
      "epoch": 2.2439024390243905,
      "grad_norm": 0.058962750452763446,
      "learning_rate": 9.078080196400985e-06,
      "loss": 0.4283,
      "step": 736
    },
    {
      "epoch": 2.2469512195121952,
      "grad_norm": 0.051835093932849644,
      "learning_rate": 9.009760888011512e-06,
      "loss": 0.4219,
      "step": 737
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.049881513783743586,
      "learning_rate": 8.941643076214437e-06,
      "loss": 0.3981,
      "step": 738
    },
    {
      "epoch": 2.2530487804878048,
      "grad_norm": 0.04685764539849008,
      "learning_rate": 8.873727619377611e-06,
      "loss": 0.4235,
      "step": 739
    },
    {
      "epoch": 2.2560975609756095,
      "grad_norm": 0.05463586193578815,
      "learning_rate": 8.80601537331895e-06,
      "loss": 0.4941,
      "step": 740
    },
    {
      "epoch": 2.2591463414634148,
      "grad_norm": 0.047851225684963736,
      "learning_rate": 8.73850719129566e-06,
      "loss": 0.4333,
      "step": 741
    },
    {
      "epoch": 2.2621951219512195,
      "grad_norm": 0.04784242396571518,
      "learning_rate": 8.671203923993501e-06,
      "loss": 0.4482,
      "step": 742
    },
    {
      "epoch": 2.2652439024390243,
      "grad_norm": 0.04804999183223516,
      "learning_rate": 8.604106419516062e-06,
      "loss": 0.3663,
      "step": 743
    },
    {
      "epoch": 2.2682926829268295,
      "grad_norm": 0.05381821577825733,
      "learning_rate": 8.537215523374038e-06,
      "loss": 0.4166,
      "step": 744
    },
    {
      "epoch": 2.2713414634146343,
      "grad_norm": 0.04969839859509168,
      "learning_rate": 8.470532078474635e-06,
      "loss": 0.4872,
      "step": 745
    },
    {
      "epoch": 2.274390243902439,
      "grad_norm": 0.04271271764381146,
      "learning_rate": 8.404056925110914e-06,
      "loss": 0.4392,
      "step": 746
    },
    {
      "epoch": 2.277439024390244,
      "grad_norm": 0.04992579157152129,
      "learning_rate": 8.3377909009512e-06,
      "loss": 0.3657,
      "step": 747
    },
    {
      "epoch": 2.2804878048780486,
      "grad_norm": 0.04818703661052087,
      "learning_rate": 8.271734841028553e-06,
      "loss": 0.3295,
      "step": 748
    },
    {
      "epoch": 2.283536585365854,
      "grad_norm": 0.044986665548360116,
      "learning_rate": 8.20588957773018e-06,
      "loss": 0.3813,
      "step": 749
    },
    {
      "epoch": 2.2865853658536586,
      "grad_norm": 0.04234902786859893,
      "learning_rate": 8.14025594078706e-06,
      "loss": 0.3131,
      "step": 750
    },
    {
      "epoch": 2.2896341463414633,
      "grad_norm": 0.045016594884271365,
      "learning_rate": 8.074834757263361e-06,
      "loss": 0.3419,
      "step": 751
    },
    {
      "epoch": 2.292682926829268,
      "grad_norm": 0.047781975300300185,
      "learning_rate": 8.00962685154611e-06,
      "loss": 0.3678,
      "step": 752
    },
    {
      "epoch": 2.2957317073170733,
      "grad_norm": 0.06553586697371824,
      "learning_rate": 7.944633045334762e-06,
      "loss": 0.4017,
      "step": 753
    },
    {
      "epoch": 2.298780487804878,
      "grad_norm": 0.04606601862679692,
      "learning_rate": 7.879854157630861e-06,
      "loss": 0.3969,
      "step": 754
    },
    {
      "epoch": 2.301829268292683,
      "grad_norm": 0.04300968673061053,
      "learning_rate": 7.815291004727713e-06,
      "loss": 0.3166,
      "step": 755
    },
    {
      "epoch": 2.3048780487804876,
      "grad_norm": 0.04765947333057751,
      "learning_rate": 7.750944400200102e-06,
      "loss": 0.3845,
      "step": 756
    },
    {
      "epoch": 2.307926829268293,
      "grad_norm": 0.04497924678023646,
      "learning_rate": 7.68681515489402e-06,
      "loss": 0.4085,
      "step": 757
    },
    {
      "epoch": 2.3109756097560976,
      "grad_norm": 0.05015532439252871,
      "learning_rate": 7.622904076916509e-06,
      "loss": 0.4564,
      "step": 758
    },
    {
      "epoch": 2.3140243902439024,
      "grad_norm": 0.042988505749526125,
      "learning_rate": 7.5592119716253855e-06,
      "loss": 0.3383,
      "step": 759
    },
    {
      "epoch": 2.317073170731707,
      "grad_norm": 0.04305861474154292,
      "learning_rate": 7.495739641619176e-06,
      "loss": 0.4107,
      "step": 760
    },
    {
      "epoch": 2.3201219512195124,
      "grad_norm": 0.04920016219288475,
      "learning_rate": 7.432487886726955e-06,
      "loss": 0.4005,
      "step": 761
    },
    {
      "epoch": 2.323170731707317,
      "grad_norm": 0.048799915611591145,
      "learning_rate": 7.369457503998287e-06,
      "loss": 0.4202,
      "step": 762
    },
    {
      "epoch": 2.326219512195122,
      "grad_norm": 0.044791697449636486,
      "learning_rate": 7.3066492876931705e-06,
      "loss": 0.4262,
      "step": 763
    },
    {
      "epoch": 2.3292682926829267,
      "grad_norm": 0.04249567961276802,
      "learning_rate": 7.244064029272049e-06,
      "loss": 0.3719,
      "step": 764
    },
    {
      "epoch": 2.332317073170732,
      "grad_norm": 0.051851485763910736,
      "learning_rate": 7.181702517385788e-06,
      "loss": 0.3861,
      "step": 765
    },
    {
      "epoch": 2.3353658536585367,
      "grad_norm": 0.044010285547474354,
      "learning_rate": 7.119565537865827e-06,
      "loss": 0.367,
      "step": 766
    },
    {
      "epoch": 2.3384146341463414,
      "grad_norm": 0.043359373882457634,
      "learning_rate": 7.057653873714171e-06,
      "loss": 0.4148,
      "step": 767
    },
    {
      "epoch": 2.341463414634146,
      "grad_norm": 0.05279502635927622,
      "learning_rate": 6.995968305093603e-06,
      "loss": 0.3696,
      "step": 768
    },
    {
      "epoch": 2.3445121951219514,
      "grad_norm": 0.04883108675178682,
      "learning_rate": 6.934509609317821e-06,
      "loss": 0.402,
      "step": 769
    },
    {
      "epoch": 2.347560975609756,
      "grad_norm": 0.043977626763723576,
      "learning_rate": 6.8732785608416476e-06,
      "loss": 0.3727,
      "step": 770
    },
    {
      "epoch": 2.350609756097561,
      "grad_norm": 0.04588882631244444,
      "learning_rate": 6.812275931251269e-06,
      "loss": 0.3991,
      "step": 771
    },
    {
      "epoch": 2.3536585365853657,
      "grad_norm": 0.04496378231139958,
      "learning_rate": 6.751502489254521e-06,
      "loss": 0.3228,
      "step": 772
    },
    {
      "epoch": 2.3567073170731705,
      "grad_norm": 0.04808438680353696,
      "learning_rate": 6.6909590006711655e-06,
      "loss": 0.4434,
      "step": 773
    },
    {
      "epoch": 2.3597560975609757,
      "grad_norm": 0.05620186191151277,
      "learning_rate": 6.6306462284233234e-06,
      "loss": 0.3159,
      "step": 774
    },
    {
      "epoch": 2.3628048780487805,
      "grad_norm": 0.046014700942548746,
      "learning_rate": 6.57056493252575e-06,
      "loss": 0.3341,
      "step": 775
    },
    {
      "epoch": 2.3658536585365852,
      "grad_norm": 0.04317894848169251,
      "learning_rate": 6.510715870076354e-06,
      "loss": 0.3533,
      "step": 776
    },
    {
      "epoch": 2.3689024390243905,
      "grad_norm": 0.0660246778046814,
      "learning_rate": 6.451099795246604e-06,
      "loss": 0.3837,
      "step": 777
    },
    {
      "epoch": 2.3719512195121952,
      "grad_norm": 0.049699048047795104,
      "learning_rate": 6.39171745927204e-06,
      "loss": 0.4546,
      "step": 778
    },
    {
      "epoch": 2.375,
      "grad_norm": 0.04647481627923815,
      "learning_rate": 6.332569610442807e-06,
      "loss": 0.4114,
      "step": 779
    },
    {
      "epoch": 2.3780487804878048,
      "grad_norm": 0.053410613463110365,
      "learning_rate": 6.273656994094232e-06,
      "loss": 0.4131,
      "step": 780
    },
    {
      "epoch": 2.3810975609756095,
      "grad_norm": 0.04635611109855328,
      "learning_rate": 6.214980352597399e-06,
      "loss": 0.4087,
      "step": 781
    },
    {
      "epoch": 2.3841463414634148,
      "grad_norm": 0.04998162717060871,
      "learning_rate": 6.1565404253498685e-06,
      "loss": 0.4286,
      "step": 782
    },
    {
      "epoch": 2.3871951219512195,
      "grad_norm": 0.04858537108941669,
      "learning_rate": 6.098337948766255e-06,
      "loss": 0.4074,
      "step": 783
    },
    {
      "epoch": 2.3902439024390243,
      "grad_norm": 0.04766628235578669,
      "learning_rate": 6.040373656269041e-06,
      "loss": 0.4047,
      "step": 784
    },
    {
      "epoch": 2.3932926829268295,
      "grad_norm": 0.04566031966872623,
      "learning_rate": 5.982648278279287e-06,
      "loss": 0.3623,
      "step": 785
    },
    {
      "epoch": 2.3963414634146343,
      "grad_norm": 0.04220225746659264,
      "learning_rate": 5.925162542207441e-06,
      "loss": 0.4055,
      "step": 786
    },
    {
      "epoch": 2.399390243902439,
      "grad_norm": 0.04753754420384983,
      "learning_rate": 5.867917172444165e-06,
      "loss": 0.4484,
      "step": 787
    },
    {
      "epoch": 2.402439024390244,
      "grad_norm": 0.044606797958348605,
      "learning_rate": 5.810912890351219e-06,
      "loss": 0.3544,
      "step": 788
    },
    {
      "epoch": 2.4054878048780486,
      "grad_norm": 0.05429191520954466,
      "learning_rate": 5.75415041425234e-06,
      "loss": 0.4157,
      "step": 789
    },
    {
      "epoch": 2.408536585365854,
      "grad_norm": 0.04841822893600714,
      "learning_rate": 5.697630459424258e-06,
      "loss": 0.4028,
      "step": 790
    },
    {
      "epoch": 2.4115853658536586,
      "grad_norm": 0.05269493013761591,
      "learning_rate": 5.641353738087588e-06,
      "loss": 0.3775,
      "step": 791
    },
    {
      "epoch": 2.4146341463414633,
      "grad_norm": 0.04113581222389823,
      "learning_rate": 5.585320959397935e-06,
      "loss": 0.3377,
      "step": 792
    },
    {
      "epoch": 2.417682926829268,
      "grad_norm": 0.05014248840706332,
      "learning_rate": 5.529532829436923e-06,
      "loss": 0.3736,
      "step": 793
    },
    {
      "epoch": 2.4207317073170733,
      "grad_norm": 0.05375871646459066,
      "learning_rate": 5.473990051203298e-06,
      "loss": 0.4044,
      "step": 794
    },
    {
      "epoch": 2.423780487804878,
      "grad_norm": 0.04562745676951761,
      "learning_rate": 5.418693324604082e-06,
      "loss": 0.3851,
      "step": 795
    },
    {
      "epoch": 2.426829268292683,
      "grad_norm": 0.04607000577575072,
      "learning_rate": 5.36364334644574e-06,
      "loss": 0.3204,
      "step": 796
    },
    {
      "epoch": 2.4298780487804876,
      "grad_norm": 0.05406018016723344,
      "learning_rate": 5.308840810425386e-06,
      "loss": 0.4035,
      "step": 797
    },
    {
      "epoch": 2.432926829268293,
      "grad_norm": 0.047760483793245496,
      "learning_rate": 5.2542864071221025e-06,
      "loss": 0.4563,
      "step": 798
    },
    {
      "epoch": 2.4359756097560976,
      "grad_norm": 0.0476426880340459,
      "learning_rate": 5.199980823988157e-06,
      "loss": 0.438,
      "step": 799
    },
    {
      "epoch": 2.4390243902439024,
      "grad_norm": 0.04751953066554679,
      "learning_rate": 5.145924745340394e-06,
      "loss": 0.3799,
      "step": 800
    },
    {
      "epoch": 2.442073170731707,
      "grad_norm": 0.046302307478752316,
      "learning_rate": 5.092118852351599e-06,
      "loss": 0.3826,
      "step": 801
    },
    {
      "epoch": 2.4451219512195124,
      "grad_norm": 0.04692143633572673,
      "learning_rate": 5.038563823041903e-06,
      "loss": 0.3318,
      "step": 802
    },
    {
      "epoch": 2.448170731707317,
      "grad_norm": 0.04435149632876141,
      "learning_rate": 4.985260332270256e-06,
      "loss": 0.3337,
      "step": 803
    },
    {
      "epoch": 2.451219512195122,
      "grad_norm": 0.04336607743154857,
      "learning_rate": 4.932209051725914e-06,
      "loss": 0.3668,
      "step": 804
    },
    {
      "epoch": 2.4542682926829267,
      "grad_norm": 0.04553355750341298,
      "learning_rate": 4.879410649919952e-06,
      "loss": 0.3869,
      "step": 805
    },
    {
      "epoch": 2.457317073170732,
      "grad_norm": 0.06575400522455546,
      "learning_rate": 4.82686579217691e-06,
      "loss": 0.402,
      "step": 806
    },
    {
      "epoch": 2.4603658536585367,
      "grad_norm": 0.05025978450812776,
      "learning_rate": 4.7745751406263165e-06,
      "loss": 0.3956,
      "step": 807
    },
    {
      "epoch": 2.4634146341463414,
      "grad_norm": 0.04207130067854438,
      "learning_rate": 4.722539354194414e-06,
      "loss": 0.2913,
      "step": 808
    },
    {
      "epoch": 2.466463414634146,
      "grad_norm": 0.0463599669859819,
      "learning_rate": 4.67075908859583e-06,
      "loss": 0.3902,
      "step": 809
    },
    {
      "epoch": 2.4695121951219514,
      "grad_norm": 0.049170400057323115,
      "learning_rate": 4.619234996325314e-06,
      "loss": 0.4393,
      "step": 810
    },
    {
      "epoch": 2.472560975609756,
      "grad_norm": 0.055986709128824926,
      "learning_rate": 4.5679677266495194e-06,
      "loss": 0.4592,
      "step": 811
    },
    {
      "epoch": 2.475609756097561,
      "grad_norm": 0.04850790136281466,
      "learning_rate": 4.516957925598822e-06,
      "loss": 0.3765,
      "step": 812
    },
    {
      "epoch": 2.4786585365853657,
      "grad_norm": 0.04548455150924458,
      "learning_rate": 4.4662062359591585e-06,
      "loss": 0.3892,
      "step": 813
    },
    {
      "epoch": 2.4817073170731705,
      "grad_norm": 0.048166202410053933,
      "learning_rate": 4.415713297263987e-06,
      "loss": 0.4386,
      "step": 814
    },
    {
      "epoch": 2.4847560975609757,
      "grad_norm": 0.052189487836794404,
      "learning_rate": 4.365479745786147e-06,
      "loss": 0.3875,
      "step": 815
    },
    {
      "epoch": 2.4878048780487805,
      "grad_norm": 0.04797712854047369,
      "learning_rate": 4.315506214529899e-06,
      "loss": 0.422,
      "step": 816
    },
    {
      "epoch": 2.4908536585365852,
      "grad_norm": 0.04730083089291941,
      "learning_rate": 4.265793333222928e-06,
      "loss": 0.3352,
      "step": 817
    },
    {
      "epoch": 2.4939024390243905,
      "grad_norm": 0.04406355303948544,
      "learning_rate": 4.216341728308412e-06,
      "loss": 0.3597,
      "step": 818
    },
    {
      "epoch": 2.4969512195121952,
      "grad_norm": 0.046402880449595424,
      "learning_rate": 4.167152022937124e-06,
      "loss": 0.47,
      "step": 819
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.048301472887954396,
      "learning_rate": 4.118224836959589e-06,
      "loss": 0.3518,
      "step": 820
    },
    {
      "epoch": 2.5030487804878048,
      "grad_norm": 0.04996830725135542,
      "learning_rate": 4.0695607869182445e-06,
      "loss": 0.43,
      "step": 821
    },
    {
      "epoch": 2.5060975609756095,
      "grad_norm": 0.05401545257800483,
      "learning_rate": 4.02116048603973e-06,
      "loss": 0.434,
      "step": 822
    },
    {
      "epoch": 2.5091463414634148,
      "grad_norm": 0.05119011813317202,
      "learning_rate": 3.973024544227086e-06,
      "loss": 0.3394,
      "step": 823
    },
    {
      "epoch": 2.5121951219512195,
      "grad_norm": 0.04804793913213503,
      "learning_rate": 3.925153568052123e-06,
      "loss": 0.3541,
      "step": 824
    },
    {
      "epoch": 2.5152439024390243,
      "grad_norm": 0.04570138233548225,
      "learning_rate": 3.877548160747768e-06,
      "loss": 0.4105,
      "step": 825
    },
    {
      "epoch": 2.5182926829268295,
      "grad_norm": 0.05180433023610628,
      "learning_rate": 3.830208922200421e-06,
      "loss": 0.4355,
      "step": 826
    },
    {
      "epoch": 2.5213414634146343,
      "grad_norm": 0.04510265543541877,
      "learning_rate": 3.7831364489424816e-06,
      "loss": 0.421,
      "step": 827
    },
    {
      "epoch": 2.524390243902439,
      "grad_norm": 0.04553571711513919,
      "learning_rate": 3.7363313341447326e-06,
      "loss": 0.4308,
      "step": 828
    },
    {
      "epoch": 2.527439024390244,
      "grad_norm": 0.05157642728956852,
      "learning_rate": 3.689794167608937e-06,
      "loss": 0.434,
      "step": 829
    },
    {
      "epoch": 2.5304878048780486,
      "grad_norm": 0.05098353856392178,
      "learning_rate": 3.643525535760378e-06,
      "loss": 0.4203,
      "step": 830
    },
    {
      "epoch": 2.533536585365854,
      "grad_norm": 0.041115017707901325,
      "learning_rate": 3.597526021640471e-06,
      "loss": 0.3213,
      "step": 831
    },
    {
      "epoch": 2.5365853658536586,
      "grad_norm": 0.04677981809027256,
      "learning_rate": 3.5517962048994213e-06,
      "loss": 0.3828,
      "step": 832
    },
    {
      "epoch": 2.5396341463414633,
      "grad_norm": 0.045752089961378764,
      "learning_rate": 3.5063366617889143e-06,
      "loss": 0.4117,
      "step": 833
    },
    {
      "epoch": 2.5426829268292686,
      "grad_norm": 0.04701033883496135,
      "learning_rate": 3.4611479651548457e-06,
      "loss": 0.3884,
      "step": 834
    },
    {
      "epoch": 2.5457317073170733,
      "grad_norm": 0.048123442696396505,
      "learning_rate": 3.4162306844301465e-06,
      "loss": 0.4071,
      "step": 835
    },
    {
      "epoch": 2.548780487804878,
      "grad_norm": 0.04713491381100571,
      "learning_rate": 3.371585385627535e-06,
      "loss": 0.449,
      "step": 836
    },
    {
      "epoch": 2.551829268292683,
      "grad_norm": 0.045698119900967754,
      "learning_rate": 3.327212631332452e-06,
      "loss": 0.391,
      "step": 837
    },
    {
      "epoch": 2.5548780487804876,
      "grad_norm": 0.045309930585614173,
      "learning_rate": 3.283112980695932e-06,
      "loss": 0.3458,
      "step": 838
    },
    {
      "epoch": 2.557926829268293,
      "grad_norm": 0.04898751105733046,
      "learning_rate": 3.239286989427573e-06,
      "loss": 0.4192,
      "step": 839
    },
    {
      "epoch": 2.5609756097560976,
      "grad_norm": 0.0509124926611376,
      "learning_rate": 3.195735209788528e-06,
      "loss": 0.4381,
      "step": 840
    },
    {
      "epoch": 2.5640243902439024,
      "grad_norm": 0.05226387101278107,
      "learning_rate": 3.1524581905845536e-06,
      "loss": 0.4069,
      "step": 841
    },
    {
      "epoch": 2.567073170731707,
      "grad_norm": 0.052460384189690755,
      "learning_rate": 3.1094564771590703e-06,
      "loss": 0.4304,
      "step": 842
    },
    {
      "epoch": 2.5701219512195124,
      "grad_norm": 0.04576303428634451,
      "learning_rate": 3.066730611386337e-06,
      "loss": 0.4097,
      "step": 843
    },
    {
      "epoch": 2.573170731707317,
      "grad_norm": 0.0454758335824736,
      "learning_rate": 3.024281131664569e-06,
      "loss": 0.3863,
      "step": 844
    },
    {
      "epoch": 2.576219512195122,
      "grad_norm": 0.04635517134671258,
      "learning_rate": 2.9821085729091924e-06,
      "loss": 0.4123,
      "step": 845
    },
    {
      "epoch": 2.5792682926829267,
      "grad_norm": 0.04489456701562646,
      "learning_rate": 2.9402134665460853e-06,
      "loss": 0.3966,
      "step": 846
    },
    {
      "epoch": 2.5823170731707314,
      "grad_norm": 0.05588564719009066,
      "learning_rate": 2.898596340504886e-06,
      "loss": 0.422,
      "step": 847
    },
    {
      "epoch": 2.5853658536585367,
      "grad_norm": 0.05079351877436922,
      "learning_rate": 2.857257719212347e-06,
      "loss": 0.3938,
      "step": 848
    },
    {
      "epoch": 2.5884146341463414,
      "grad_norm": 0.048725942156966844,
      "learning_rate": 2.8161981235857143e-06,
      "loss": 0.3921,
      "step": 849
    },
    {
      "epoch": 2.591463414634146,
      "grad_norm": 0.05354775440497977,
      "learning_rate": 2.775418071026156e-06,
      "loss": 0.3606,
      "step": 850
    },
    {
      "epoch": 2.5945121951219514,
      "grad_norm": 0.04385956729202263,
      "learning_rate": 2.7349180754122906e-06,
      "loss": 0.3814,
      "step": 851
    },
    {
      "epoch": 2.597560975609756,
      "grad_norm": 0.04889690374749316,
      "learning_rate": 2.6946986470936353e-06,
      "loss": 0.4797,
      "step": 852
    },
    {
      "epoch": 2.600609756097561,
      "grad_norm": 0.04295496415305672,
      "learning_rate": 2.6547602928842396e-06,
      "loss": 0.3818,
      "step": 853
    },
    {
      "epoch": 2.6036585365853657,
      "grad_norm": 0.04348352239438333,
      "learning_rate": 2.615103516056275e-06,
      "loss": 0.3598,
      "step": 854
    },
    {
      "epoch": 2.6067073170731705,
      "grad_norm": 0.047044801096703746,
      "learning_rate": 2.5757288163336807e-06,
      "loss": 0.4477,
      "step": 855
    },
    {
      "epoch": 2.6097560975609757,
      "grad_norm": 0.045721796420320274,
      "learning_rate": 2.5366366898858935e-06,
      "loss": 0.4007,
      "step": 856
    },
    {
      "epoch": 2.6128048780487805,
      "grad_norm": 0.04578531053202925,
      "learning_rate": 2.4978276293215725e-06,
      "loss": 0.3475,
      "step": 857
    },
    {
      "epoch": 2.6158536585365852,
      "grad_norm": 0.045594774709890595,
      "learning_rate": 2.4593021236823914e-06,
      "loss": 0.422,
      "step": 858
    },
    {
      "epoch": 2.6189024390243905,
      "grad_norm": 0.04700300164236975,
      "learning_rate": 2.4210606584369104e-06,
      "loss": 0.3931,
      "step": 859
    },
    {
      "epoch": 2.6219512195121952,
      "grad_norm": 0.059320405162608614,
      "learning_rate": 2.383103715474408e-06,
      "loss": 0.4289,
      "step": 860
    },
    {
      "epoch": 2.625,
      "grad_norm": 0.05020753814940063,
      "learning_rate": 2.345431773098841e-06,
      "loss": 0.4138,
      "step": 861
    },
    {
      "epoch": 2.6280487804878048,
      "grad_norm": 0.04601472647317754,
      "learning_rate": 2.3080453060228157e-06,
      "loss": 0.386,
      "step": 862
    },
    {
      "epoch": 2.6310975609756095,
      "grad_norm": 0.04866574147542356,
      "learning_rate": 2.270944785361598e-06,
      "loss": 0.4103,
      "step": 863
    },
    {
      "epoch": 2.6341463414634148,
      "grad_norm": 0.04658640783031281,
      "learning_rate": 2.2341306786271695e-06,
      "loss": 0.4522,
      "step": 864
    },
    {
      "epoch": 2.6371951219512195,
      "grad_norm": 0.0487913591888631,
      "learning_rate": 2.197603449722363e-06,
      "loss": 0.4711,
      "step": 865
    },
    {
      "epoch": 2.6402439024390243,
      "grad_norm": 0.05756759392886848,
      "learning_rate": 2.1613635589349756e-06,
      "loss": 0.391,
      "step": 866
    },
    {
      "epoch": 2.6432926829268295,
      "grad_norm": 0.05197045765940042,
      "learning_rate": 2.125411462932023e-06,
      "loss": 0.4897,
      "step": 867
    },
    {
      "epoch": 2.6463414634146343,
      "grad_norm": 0.0500854850405243,
      "learning_rate": 2.089747614753923e-06,
      "loss": 0.3817,
      "step": 868
    },
    {
      "epoch": 2.649390243902439,
      "grad_norm": 0.04590140661248139,
      "learning_rate": 2.0543724638088347e-06,
      "loss": 0.417,
      "step": 869
    },
    {
      "epoch": 2.652439024390244,
      "grad_norm": 0.048277191569880015,
      "learning_rate": 2.019286455866981e-06,
      "loss": 0.4101,
      "step": 870
    },
    {
      "epoch": 2.6554878048780486,
      "grad_norm": 0.0509909437544304,
      "learning_rate": 1.98449003305502e-06,
      "loss": 0.4346,
      "step": 871
    },
    {
      "epoch": 2.658536585365854,
      "grad_norm": 0.046959947573902566,
      "learning_rate": 1.9499836338504886e-06,
      "loss": 0.4033,
      "step": 872
    },
    {
      "epoch": 2.6615853658536586,
      "grad_norm": 0.044990086629584425,
      "learning_rate": 1.91576769307627e-06,
      "loss": 0.4074,
      "step": 873
    },
    {
      "epoch": 2.6646341463414633,
      "grad_norm": 0.046010218256964684,
      "learning_rate": 1.881842641895104e-06,
      "loss": 0.3816,
      "step": 874
    },
    {
      "epoch": 2.6676829268292686,
      "grad_norm": 0.04704431532728486,
      "learning_rate": 1.8482089078041915e-06,
      "loss": 0.4273,
      "step": 875
    },
    {
      "epoch": 2.6707317073170733,
      "grad_norm": 0.047115973314242464,
      "learning_rate": 1.8148669146297565e-06,
      "loss": 0.4154,
      "step": 876
    },
    {
      "epoch": 2.673780487804878,
      "grad_norm": 0.052659028245325507,
      "learning_rate": 1.781817082521739e-06,
      "loss": 0.4727,
      "step": 877
    },
    {
      "epoch": 2.676829268292683,
      "grad_norm": 0.049586231251581646,
      "learning_rate": 1.7490598279484933e-06,
      "loss": 0.4617,
      "step": 878
    },
    {
      "epoch": 2.6798780487804876,
      "grad_norm": 0.04619796705926887,
      "learning_rate": 1.7165955636915392e-06,
      "loss": 0.3888,
      "step": 879
    },
    {
      "epoch": 2.682926829268293,
      "grad_norm": 0.04857595937650091,
      "learning_rate": 1.6844246988403561e-06,
      "loss": 0.3977,
      "step": 880
    },
    {
      "epoch": 2.6859756097560976,
      "grad_norm": 0.04587078872794076,
      "learning_rate": 1.6525476387872384e-06,
      "loss": 0.3635,
      "step": 881
    },
    {
      "epoch": 2.6890243902439024,
      "grad_norm": 0.04749578656301489,
      "learning_rate": 1.6209647852221622e-06,
      "loss": 0.4613,
      "step": 882
    },
    {
      "epoch": 2.692073170731707,
      "grad_norm": 0.045361324797590356,
      "learning_rate": 1.589676536127771e-06,
      "loss": 0.3525,
      "step": 883
    },
    {
      "epoch": 2.6951219512195124,
      "grad_norm": 0.04894264008622803,
      "learning_rate": 1.558683285774304e-06,
      "loss": 0.4656,
      "step": 884
    },
    {
      "epoch": 2.698170731707317,
      "grad_norm": 0.04596961558750641,
      "learning_rate": 1.5279854247146702e-06,
      "loss": 0.3885,
      "step": 885
    },
    {
      "epoch": 2.701219512195122,
      "grad_norm": 0.04877091621740549,
      "learning_rate": 1.497583339779507e-06,
      "loss": 0.3995,
      "step": 886
    },
    {
      "epoch": 2.7042682926829267,
      "grad_norm": 0.046603063123663,
      "learning_rate": 1.4674774140723103e-06,
      "loss": 0.3794,
      "step": 887
    },
    {
      "epoch": 2.7073170731707314,
      "grad_norm": 0.049078136886864604,
      "learning_rate": 1.4376680269646086e-06,
      "loss": 0.3831,
      "step": 888
    },
    {
      "epoch": 2.7103658536585367,
      "grad_norm": 0.046551620610760534,
      "learning_rate": 1.408155554091184e-06,
      "loss": 0.3683,
      "step": 889
    },
    {
      "epoch": 2.7134146341463414,
      "grad_norm": 0.04769564867133674,
      "learning_rate": 1.3789403673453216e-06,
      "loss": 0.4364,
      "step": 890
    },
    {
      "epoch": 2.716463414634146,
      "grad_norm": 0.05275442418545579,
      "learning_rate": 1.3500228348741594e-06,
      "loss": 0.4007,
      "step": 891
    },
    {
      "epoch": 2.7195121951219514,
      "grad_norm": 0.0487800880348598,
      "learning_rate": 1.3214033210740079e-06,
      "loss": 0.3825,
      "step": 892
    },
    {
      "epoch": 2.722560975609756,
      "grad_norm": 0.052257228902126626,
      "learning_rate": 1.2930821865857867e-06,
      "loss": 0.4774,
      "step": 893
    },
    {
      "epoch": 2.725609756097561,
      "grad_norm": 0.04598036087813201,
      "learning_rate": 1.265059788290468e-06,
      "loss": 0.4152,
      "step": 894
    },
    {
      "epoch": 2.7286585365853657,
      "grad_norm": 0.05163087663792444,
      "learning_rate": 1.2373364793045821e-06,
      "loss": 0.4959,
      "step": 895
    },
    {
      "epoch": 2.7317073170731705,
      "grad_norm": 0.0514975087513403,
      "learning_rate": 1.2099126089757746e-06,
      "loss": 0.3535,
      "step": 896
    },
    {
      "epoch": 2.7347560975609757,
      "grad_norm": 0.043893617709315234,
      "learning_rate": 1.1827885228783863e-06,
      "loss": 0.4188,
      "step": 897
    },
    {
      "epoch": 2.7378048780487805,
      "grad_norm": 0.04818454241881662,
      "learning_rate": 1.1559645628091115e-06,
      "loss": 0.3196,
      "step": 898
    },
    {
      "epoch": 2.7408536585365852,
      "grad_norm": 0.04398367015233411,
      "learning_rate": 1.129441066782702e-06,
      "loss": 0.3253,
      "step": 899
    },
    {
      "epoch": 2.7439024390243905,
      "grad_norm": 0.04017783128096445,
      "learning_rate": 1.1032183690276754e-06,
      "loss": 0.3725,
      "step": 900
    },
    {
      "epoch": 2.7469512195121952,
      "grad_norm": 0.051719053632263305,
      "learning_rate": 1.0772967999821387e-06,
      "loss": 0.3862,
      "step": 901
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.04677169001526996,
      "learning_rate": 1.0516766862895988e-06,
      "loss": 0.4183,
      "step": 902
    },
    {
      "epoch": 2.7530487804878048,
      "grad_norm": 0.05608282238185744,
      "learning_rate": 1.0263583507948592e-06,
      "loss": 0.4227,
      "step": 903
    },
    {
      "epoch": 2.7560975609756095,
      "grad_norm": 0.050428851178379905,
      "learning_rate": 1.0013421125399519e-06,
      "loss": 0.3616,
      "step": 904
    },
    {
      "epoch": 2.7591463414634148,
      "grad_norm": 0.05034269092062958,
      "learning_rate": 9.766282867601062e-07,
      "loss": 0.3844,
      "step": 905
    },
    {
      "epoch": 2.7621951219512195,
      "grad_norm": 0.04504676946969238,
      "learning_rate": 9.522171848797918e-07,
      "loss": 0.391,
      "step": 906
    },
    {
      "epoch": 2.7652439024390243,
      "grad_norm": 0.04567291042842459,
      "learning_rate": 9.281091145087794e-07,
      "loss": 0.3731,
      "step": 907
    },
    {
      "epoch": 2.7682926829268295,
      "grad_norm": 0.04760877660304163,
      "learning_rate": 9.043043794382811e-07,
      "loss": 0.3975,
      "step": 908
    },
    {
      "epoch": 2.7713414634146343,
      "grad_norm": 0.04886230944889475,
      "learning_rate": 8.808032796371019e-07,
      "loss": 0.4345,
      "step": 909
    },
    {
      "epoch": 2.774390243902439,
      "grad_norm": 0.046761963685432256,
      "learning_rate": 8.5760611124788e-07,
      "loss": 0.3587,
      "step": 910
    },
    {
      "epoch": 2.777439024390244,
      "grad_norm": 0.05012257767113604,
      "learning_rate": 8.34713166583334e-07,
      "loss": 0.4435,
      "step": 911
    },
    {
      "epoch": 2.7804878048780486,
      "grad_norm": 0.0485857061164425,
      "learning_rate": 8.121247341226073e-07,
      "loss": 0.4457,
      "step": 912
    },
    {
      "epoch": 2.783536585365854,
      "grad_norm": 0.047771627365448335,
      "learning_rate": 7.898410985076016e-07,
      "loss": 0.3506,
      "step": 913
    },
    {
      "epoch": 2.7865853658536586,
      "grad_norm": 0.04524184109786172,
      "learning_rate": 7.678625405394157e-07,
      "loss": 0.368,
      "step": 914
    },
    {
      "epoch": 2.7896341463414633,
      "grad_norm": 0.04638346083359397,
      "learning_rate": 7.46189337174788e-07,
      "loss": 0.4712,
      "step": 915
    },
    {
      "epoch": 2.7926829268292686,
      "grad_norm": 0.04795283141258439,
      "learning_rate": 7.248217615226232e-07,
      "loss": 0.4073,
      "step": 916
    },
    {
      "epoch": 2.7957317073170733,
      "grad_norm": 0.047896666710965284,
      "learning_rate": 7.037600828405349e-07,
      "loss": 0.4145,
      "step": 917
    },
    {
      "epoch": 2.798780487804878,
      "grad_norm": 0.04476687658273201,
      "learning_rate": 6.830045665314671e-07,
      "loss": 0.3853,
      "step": 918
    },
    {
      "epoch": 2.801829268292683,
      "grad_norm": 0.045070434638962945,
      "learning_rate": 6.625554741403333e-07,
      "loss": 0.3998,
      "step": 919
    },
    {
      "epoch": 2.8048780487804876,
      "grad_norm": 0.05006029469711144,
      "learning_rate": 6.424130633507497e-07,
      "loss": 0.4101,
      "step": 920
    },
    {
      "epoch": 2.807926829268293,
      "grad_norm": 0.05376575151245747,
      "learning_rate": 6.22577587981743e-07,
      "loss": 0.4244,
      "step": 921
    },
    {
      "epoch": 2.8109756097560976,
      "grad_norm": 0.051072787008947136,
      "learning_rate": 6.030492979845953e-07,
      "loss": 0.4588,
      "step": 922
    },
    {
      "epoch": 2.8140243902439024,
      "grad_norm": 0.05446211637366452,
      "learning_rate": 5.838284394396764e-07,
      "loss": 0.425,
      "step": 923
    },
    {
      "epoch": 2.817073170731707,
      "grad_norm": 0.04809228041596783,
      "learning_rate": 5.649152545533332e-07,
      "loss": 0.4626,
      "step": 924
    },
    {
      "epoch": 2.8201219512195124,
      "grad_norm": 0.0443872011957413,
      "learning_rate": 5.463099816548579e-07,
      "loss": 0.3606,
      "step": 925
    },
    {
      "epoch": 2.823170731707317,
      "grad_norm": 0.0456727521695989,
      "learning_rate": 5.280128551934743e-07,
      "loss": 0.3553,
      "step": 926
    },
    {
      "epoch": 2.826219512195122,
      "grad_norm": 0.04370934239171925,
      "learning_rate": 5.100241057353682e-07,
      "loss": 0.3541,
      "step": 927
    },
    {
      "epoch": 2.8292682926829267,
      "grad_norm": 0.052759703472349234,
      "learning_rate": 4.923439599608227e-07,
      "loss": 0.4907,
      "step": 928
    },
    {
      "epoch": 2.8323170731707314,
      "grad_norm": 0.04862415168463701,
      "learning_rate": 4.749726406613142e-07,
      "loss": 0.3889,
      "step": 929
    },
    {
      "epoch": 2.8353658536585367,
      "grad_norm": 0.0518156119357567,
      "learning_rate": 4.5791036673673846e-07,
      "loss": 0.4403,
      "step": 930
    },
    {
      "epoch": 2.8384146341463414,
      "grad_norm": 0.04371517773149062,
      "learning_rate": 4.411573531926422e-07,
      "loss": 0.3311,
      "step": 931
    },
    {
      "epoch": 2.841463414634146,
      "grad_norm": 0.0444702244148316,
      "learning_rate": 4.2471381113750876e-07,
      "loss": 0.3891,
      "step": 932
    },
    {
      "epoch": 2.8445121951219514,
      "grad_norm": 0.050165104913087226,
      "learning_rate": 4.0857994778009944e-07,
      "loss": 0.391,
      "step": 933
    },
    {
      "epoch": 2.847560975609756,
      "grad_norm": 0.04413473680201035,
      "learning_rate": 3.9275596642685543e-07,
      "loss": 0.3764,
      "step": 934
    },
    {
      "epoch": 2.850609756097561,
      "grad_norm": 0.04934776501624898,
      "learning_rate": 3.7724206647930803e-07,
      "loss": 0.3838,
      "step": 935
    },
    {
      "epoch": 2.8536585365853657,
      "grad_norm": 0.051994396400409246,
      "learning_rate": 3.6203844343160044e-07,
      "loss": 0.4299,
      "step": 936
    },
    {
      "epoch": 2.8567073170731705,
      "grad_norm": 0.05086329496351862,
      "learning_rate": 3.4714528886799494e-07,
      "loss": 0.4624,
      "step": 937
    },
    {
      "epoch": 2.8597560975609757,
      "grad_norm": 0.04546567367722647,
      "learning_rate": 3.3256279046048053e-07,
      "loss": 0.4194,
      "step": 938
    },
    {
      "epoch": 2.8628048780487805,
      "grad_norm": 0.05351102581600541,
      "learning_rate": 3.1829113196638614e-07,
      "loss": 0.4779,
      "step": 939
    },
    {
      "epoch": 2.8658536585365852,
      "grad_norm": 0.046764695752408325,
      "learning_rate": 3.0433049322608767e-07,
      "loss": 0.3816,
      "step": 940
    },
    {
      "epoch": 2.8689024390243905,
      "grad_norm": 0.04223339280911066,
      "learning_rate": 2.906810501607293e-07,
      "loss": 0.3986,
      "step": 941
    },
    {
      "epoch": 2.8719512195121952,
      "grad_norm": 0.04356884108337029,
      "learning_rate": 2.7734297477000627e-07,
      "loss": 0.3702,
      "step": 942
    },
    {
      "epoch": 2.875,
      "grad_norm": 0.04622932575023138,
      "learning_rate": 2.643164351299965e-07,
      "loss": 0.4075,
      "step": 943
    },
    {
      "epoch": 2.8780487804878048,
      "grad_norm": 0.05178690372987777,
      "learning_rate": 2.5160159539105443e-07,
      "loss": 0.4318,
      "step": 944
    },
    {
      "epoch": 2.8810975609756095,
      "grad_norm": 0.044948193037368314,
      "learning_rate": 2.391986157757292e-07,
      "loss": 0.4114,
      "step": 945
    },
    {
      "epoch": 2.8841463414634148,
      "grad_norm": 0.04997670450445459,
      "learning_rate": 2.2710765257674958e-07,
      "loss": 0.483,
      "step": 946
    },
    {
      "epoch": 2.8871951219512195,
      "grad_norm": 0.05332715471259246,
      "learning_rate": 2.153288581550561e-07,
      "loss": 0.4037,
      "step": 947
    },
    {
      "epoch": 2.8902439024390243,
      "grad_norm": 0.043304309516807464,
      "learning_rate": 2.038623809378859e-07,
      "loss": 0.4034,
      "step": 948
    },
    {
      "epoch": 2.8932926829268295,
      "grad_norm": 0.05034997634317469,
      "learning_rate": 1.927083654168854e-07,
      "loss": 0.445,
      "step": 949
    },
    {
      "epoch": 2.8963414634146343,
      "grad_norm": 0.04656387288683428,
      "learning_rate": 1.8186695214631733e-07,
      "loss": 0.4002,
      "step": 950
    },
    {
      "epoch": 2.899390243902439,
      "grad_norm": 0.05685909635754947,
      "learning_rate": 1.7133827774125367e-07,
      "loss": 0.4089,
      "step": 951
    },
    {
      "epoch": 2.902439024390244,
      "grad_norm": 0.05105550552370391,
      "learning_rate": 1.6112247487589105e-07,
      "loss": 0.3904,
      "step": 952
    },
    {
      "epoch": 2.9054878048780486,
      "grad_norm": 0.04499414356958469,
      "learning_rate": 1.512196722818493e-07,
      "loss": 0.3904,
      "step": 953
    },
    {
      "epoch": 2.908536585365854,
      "grad_norm": 0.04763536407866678,
      "learning_rate": 1.4162999474657268e-07,
      "loss": 0.381,
      "step": 954
    },
    {
      "epoch": 2.9115853658536586,
      "grad_norm": 0.04829185976705493,
      "learning_rate": 1.3235356311173397e-07,
      "loss": 0.4471,
      "step": 955
    },
    {
      "epoch": 2.9146341463414633,
      "grad_norm": 0.046717109208157435,
      "learning_rate": 1.2339049427173843e-07,
      "loss": 0.46,
      "step": 956
    },
    {
      "epoch": 2.9176829268292686,
      "grad_norm": 0.04764383813122349,
      "learning_rate": 1.1474090117221947e-07,
      "loss": 0.3908,
      "step": 957
    },
    {
      "epoch": 2.9207317073170733,
      "grad_norm": 0.04617300216734002,
      "learning_rate": 1.0640489280865085e-07,
      "loss": 0.4213,
      "step": 958
    },
    {
      "epoch": 2.923780487804878,
      "grad_norm": 0.04930526528152621,
      "learning_rate": 9.838257422493668e-08,
      "loss": 0.4257,
      "step": 959
    },
    {
      "epoch": 2.926829268292683,
      "grad_norm": 0.04763207735191656,
      "learning_rate": 9.067404651211808e-08,
      "loss": 0.369,
      "step": 960
    },
    {
      "epoch": 2.9298780487804876,
      "grad_norm": 0.049033279860561683,
      "learning_rate": 8.327940680708246e-08,
      "loss": 0.4256,
      "step": 961
    },
    {
      "epoch": 2.932926829268293,
      "grad_norm": 0.051564111935322204,
      "learning_rate": 7.61987482913451e-08,
      "loss": 0.4545,
      "step": 962
    },
    {
      "epoch": 2.9359756097560976,
      "grad_norm": 0.05251292272917081,
      "learning_rate": 6.943216018987508e-08,
      "loss": 0.4666,
      "step": 963
    },
    {
      "epoch": 2.9390243902439024,
      "grad_norm": 0.04823175777273723,
      "learning_rate": 6.297972776996286e-08,
      "loss": 0.4674,
      "step": 964
    },
    {
      "epoch": 2.942073170731707,
      "grad_norm": 0.04891409789807861,
      "learning_rate": 5.6841532340162764e-08,
      "loss": 0.3941,
      "step": 965
    },
    {
      "epoch": 2.9451219512195124,
      "grad_norm": 0.05083495908280979,
      "learning_rate": 5.1017651249252175e-08,
      "loss": 0.408,
      "step": 966
    },
    {
      "epoch": 2.948170731707317,
      "grad_norm": 0.048667783893621294,
      "learning_rate": 4.550815788526008e-08,
      "loss": 0.4246,
      "step": 967
    },
    {
      "epoch": 2.951219512195122,
      "grad_norm": 0.04938238004854509,
      "learning_rate": 4.03131216745567e-08,
      "loss": 0.4088,
      "step": 968
    },
    {
      "epoch": 2.9542682926829267,
      "grad_norm": 0.044065171800084124,
      "learning_rate": 3.543260808095139e-08,
      "loss": 0.3315,
      "step": 969
    },
    {
      "epoch": 2.9573170731707314,
      "grad_norm": 0.04128399852179956,
      "learning_rate": 3.0866678604896116e-08,
      "loss": 0.3572,
      "step": 970
    },
    {
      "epoch": 2.9603658536585367,
      "grad_norm": 0.05213292796988347,
      "learning_rate": 2.6615390782691596e-08,
      "loss": 0.4389,
      "step": 971
    },
    {
      "epoch": 2.9634146341463414,
      "grad_norm": 0.05712450278932026,
      "learning_rate": 2.2678798185771232e-08,
      "loss": 0.3561,
      "step": 972
    },
    {
      "epoch": 2.966463414634146,
      "grad_norm": 0.04612117104693992,
      "learning_rate": 1.905695042002109e-08,
      "loss": 0.4228,
      "step": 973
    },
    {
      "epoch": 2.9695121951219514,
      "grad_norm": 0.05064073173011995,
      "learning_rate": 1.5749893125160954e-08,
      "loss": 0.4459,
      "step": 974
    },
    {
      "epoch": 2.972560975609756,
      "grad_norm": 0.04260762810954784,
      "learning_rate": 1.2757667974155896e-08,
      "loss": 0.4114,
      "step": 975
    },
    {
      "epoch": 2.975609756097561,
      "grad_norm": 0.06403331793633539,
      "learning_rate": 1.0080312672711145e-08,
      "loss": 0.4859,
      "step": 976
    },
    {
      "epoch": 2.9786585365853657,
      "grad_norm": 0.04714572968910608,
      "learning_rate": 7.717860958780798e-09,
      "loss": 0.4298,
      "step": 977
    },
    {
      "epoch": 2.9817073170731705,
      "grad_norm": 0.04845961498863735,
      "learning_rate": 5.670342602148715e-09,
      "loss": 0.3839,
      "step": 978
    },
    {
      "epoch": 2.9847560975609757,
      "grad_norm": 0.049494663259983175,
      "learning_rate": 3.9377834040538185e-09,
      "loss": 0.3712,
      "step": 979
    },
    {
      "epoch": 2.9878048780487805,
      "grad_norm": 0.04949515354248563,
      "learning_rate": 2.5202051968625797e-09,
      "loss": 0.3833,
      "step": 980
    },
    {
      "epoch": 2.9908536585365852,
      "grad_norm": 0.05067059935337459,
      "learning_rate": 1.4176258437970103e-09,
      "loss": 0.4035,
      "step": 981
    },
    {
      "epoch": 2.9939024390243905,
      "grad_norm": 0.05569874207800916,
      "learning_rate": 6.300592387098458e-10,
      "loss": 0.4107,
      "step": 982
    },
    {
      "epoch": 2.9969512195121952,
      "grad_norm": 0.048017333907845926,
      "learning_rate": 1.575153058985812e-10,
      "loss": 0.3739,
      "step": 983
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.05063228298134809,
      "learning_rate": 0.0,
      "loss": 0.4038,
      "step": 984
    },
    {
      "epoch": 3.0,
      "step": 984,
      "total_flos": 3793981707124736.0,
      "train_loss": 0.4249174514315962,
      "train_runtime": 22985.8992,
      "train_samples_per_second": 1.366,
      "train_steps_per_second": 0.043
    }
  ],
  "logging_steps": 1,
  "max_steps": 984,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3793981707124736.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}