{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5001257018352468,
  "eval_steps": 500,
  "global_step": 2984,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016760244699572614,
      "grad_norm": 0.5471465450210593,
      "learning_rate": 0.0,
      "loss": 1.3574,
      "step": 1
    },
    {
      "epoch": 0.0003352048939914523,
      "grad_norm": 0.5519129609420749,
      "learning_rate": 2.222222222222222e-09,
      "loss": 1.3996,
      "step": 2
    },
    {
      "epoch": 0.0005028073409871785,
      "grad_norm": 0.5605479694391164,
      "learning_rate": 4.444444444444444e-09,
      "loss": 1.3569,
      "step": 3
    },
    {
      "epoch": 0.0006704097879829046,
      "grad_norm": 0.49785009398095087,
      "learning_rate": 6.666666666666666e-09,
      "loss": 1.291,
      "step": 4
    },
    {
      "epoch": 0.0008380122349786307,
      "grad_norm": 0.5326070228483656,
      "learning_rate": 8.888888888888889e-09,
      "loss": 1.3873,
      "step": 5
    },
    {
      "epoch": 0.001005614681974357,
      "grad_norm": 0.5302676775460052,
      "learning_rate": 1.111111111111111e-08,
      "loss": 1.3924,
      "step": 6
    },
    {
      "epoch": 0.001173217128970083,
      "grad_norm": 0.48919895351767373,
      "learning_rate": 1.3333333333333332e-08,
      "loss": 1.2989,
      "step": 7
    },
    {
      "epoch": 0.0013408195759658092,
      "grad_norm": 0.4484156287621507,
      "learning_rate": 1.5555555555555554e-08,
      "loss": 1.2881,
      "step": 8
    },
    {
      "epoch": 0.0015084220229615352,
      "grad_norm": 0.5819201646424759,
      "learning_rate": 1.7777777777777777e-08,
      "loss": 1.4535,
      "step": 9
    },
    {
      "epoch": 0.0016760244699572614,
      "grad_norm": 0.5206315176776741,
      "learning_rate": 2e-08,
      "loss": 1.3581,
      "step": 10
    },
    {
      "epoch": 0.0018436269169529876,
      "grad_norm": 0.5238626327847989,
      "learning_rate": 2.222222222222222e-08,
      "loss": 1.3519,
      "step": 11
    },
    {
      "epoch": 0.002011229363948714,
      "grad_norm": 0.5179443235638943,
      "learning_rate": 2.4444444444444444e-08,
      "loss": 1.3146,
      "step": 12
    },
    {
      "epoch": 0.0021788318109444396,
      "grad_norm": 0.5002382725366042,
      "learning_rate": 2.6666666666666664e-08,
      "loss": 1.3565,
      "step": 13
    },
    {
      "epoch": 0.002346434257940166,
      "grad_norm": 0.5139112659487405,
      "learning_rate": 2.8888888888888884e-08,
      "loss": 1.3073,
      "step": 14
    },
    {
      "epoch": 0.002514036704935892,
      "grad_norm": 0.42868751399043886,
      "learning_rate": 3.111111111111111e-08,
      "loss": 1.2153,
      "step": 15
    },
    {
      "epoch": 0.0026816391519316183,
      "grad_norm": 0.44846615701769793,
      "learning_rate": 3.333333333333333e-08,
      "loss": 1.2875,
      "step": 16
    },
    {
      "epoch": 0.0028492415989273445,
      "grad_norm": 0.48491053800389655,
      "learning_rate": 3.5555555555555554e-08,
      "loss": 1.3364,
      "step": 17
    },
    {
      "epoch": 0.0030168440459230703,
      "grad_norm": 0.5223190277537061,
      "learning_rate": 3.7777777777777774e-08,
      "loss": 1.3554,
      "step": 18
    },
    {
      "epoch": 0.0031844464929187966,
      "grad_norm": 0.5009932674370465,
      "learning_rate": 4e-08,
      "loss": 1.3763,
      "step": 19
    },
    {
      "epoch": 0.0033520489399145228,
      "grad_norm": 0.5728057891805609,
      "learning_rate": 4.222222222222222e-08,
      "loss": 1.3625,
      "step": 20
    },
    {
      "epoch": 0.003519651386910249,
      "grad_norm": 0.500501558012957,
      "learning_rate": 4.444444444444444e-08,
      "loss": 1.2987,
      "step": 21
    },
    {
      "epoch": 0.0036872538339059752,
      "grad_norm": 0.46901855019743255,
      "learning_rate": 4.666666666666667e-08,
      "loss": 1.3135,
      "step": 22
    },
    {
      "epoch": 0.003854856280901701,
      "grad_norm": 0.4913424626442062,
      "learning_rate": 4.888888888888889e-08,
      "loss": 1.3391,
      "step": 23
    },
    {
      "epoch": 0.004022458727897428,
      "grad_norm": 0.4579863048124001,
      "learning_rate": 5.111111111111111e-08,
      "loss": 1.2233,
      "step": 24
    },
    {
      "epoch": 0.004190061174893153,
      "grad_norm": 0.5625040729222139,
      "learning_rate": 5.333333333333333e-08,
      "loss": 1.3878,
      "step": 25
    },
    {
      "epoch": 0.004357663621888879,
      "grad_norm": 0.4947883788811301,
      "learning_rate": 5.5555555555555555e-08,
      "loss": 1.3498,
      "step": 26
    },
    {
      "epoch": 0.0045252660688846055,
      "grad_norm": 0.5650316101529251,
      "learning_rate": 5.777777777777777e-08,
      "loss": 1.3812,
      "step": 27
    },
    {
      "epoch": 0.004692868515880332,
      "grad_norm": 0.49556998973925964,
      "learning_rate": 6e-08,
      "loss": 1.3245,
      "step": 28
    },
    {
      "epoch": 0.004860470962876058,
      "grad_norm": 0.4938699744406017,
      "learning_rate": 6.222222222222221e-08,
      "loss": 1.3239,
      "step": 29
    },
    {
      "epoch": 0.005028073409871784,
      "grad_norm": 0.5038345713881823,
      "learning_rate": 6.444444444444445e-08,
      "loss": 1.3735,
      "step": 30
    },
    {
      "epoch": 0.00519567585686751,
      "grad_norm": 0.48283048586133087,
      "learning_rate": 6.666666666666665e-08,
      "loss": 1.316,
      "step": 31
    },
    {
      "epoch": 0.005363278303863237,
      "grad_norm": 0.48902466716197723,
      "learning_rate": 6.888888888888889e-08,
      "loss": 1.3361,
      "step": 32
    },
    {
      "epoch": 0.005530880750858963,
      "grad_norm": 0.46826154292306865,
      "learning_rate": 7.111111111111111e-08,
      "loss": 1.3014,
      "step": 33
    },
    {
      "epoch": 0.005698483197854689,
      "grad_norm": 0.5623029669528229,
      "learning_rate": 7.333333333333333e-08,
      "loss": 1.4125,
      "step": 34
    },
    {
      "epoch": 0.0058660856448504144,
      "grad_norm": 0.4843183178227428,
      "learning_rate": 7.555555555555555e-08,
      "loss": 1.3083,
      "step": 35
    },
    {
      "epoch": 0.006033688091846141,
      "grad_norm": 0.5327762875693094,
      "learning_rate": 7.777777777777778e-08,
      "loss": 1.4569,
      "step": 36
    },
    {
      "epoch": 0.006201290538841867,
      "grad_norm": 0.4913508354186036,
      "learning_rate": 8e-08,
      "loss": 1.2882,
      "step": 37
    },
    {
      "epoch": 0.006368892985837593,
      "grad_norm": 0.5563721650150807,
      "learning_rate": 8.222222222222221e-08,
      "loss": 1.4041,
      "step": 38
    },
    {
      "epoch": 0.006536495432833319,
      "grad_norm": 0.4990247361603203,
      "learning_rate": 8.444444444444444e-08,
      "loss": 1.382,
      "step": 39
    },
    {
      "epoch": 0.0067040978798290456,
      "grad_norm": 0.39306561442536525,
      "learning_rate": 8.666666666666666e-08,
      "loss": 1.1204,
      "step": 40
    },
    {
      "epoch": 0.006871700326824772,
      "grad_norm": 0.4409159966224185,
      "learning_rate": 8.888888888888888e-08,
      "loss": 1.2633,
      "step": 41
    },
    {
      "epoch": 0.007039302773820498,
      "grad_norm": 0.45028832009827147,
      "learning_rate": 9.11111111111111e-08,
      "loss": 1.2384,
      "step": 42
    },
    {
      "epoch": 0.007206905220816224,
      "grad_norm": 0.45079336889757105,
      "learning_rate": 9.333333333333334e-08,
      "loss": 1.2648,
      "step": 43
    },
    {
      "epoch": 0.0073745076678119505,
      "grad_norm": 0.5489633350965509,
      "learning_rate": 9.555555555555556e-08,
      "loss": 1.4183,
      "step": 44
    },
    {
      "epoch": 0.007542110114807676,
      "grad_norm": 0.6031675160399268,
      "learning_rate": 9.777777777777778e-08,
      "loss": 1.4527,
      "step": 45
    },
    {
      "epoch": 0.007709712561803402,
      "grad_norm": 0.6084660738971998,
      "learning_rate": 1e-07,
      "loss": 1.3991,
      "step": 46
    },
    {
      "epoch": 0.00787731500879913,
      "grad_norm": 0.4837582246332645,
      "learning_rate": 1.0222222222222222e-07,
      "loss": 1.3502,
      "step": 47
    },
    {
      "epoch": 0.008044917455794855,
      "grad_norm": 0.5302046447950054,
      "learning_rate": 1.0444444444444445e-07,
      "loss": 1.3943,
      "step": 48
    },
    {
      "epoch": 0.008212519902790582,
      "grad_norm": 0.4893541426423411,
      "learning_rate": 1.0666666666666666e-07,
      "loss": 1.2904,
      "step": 49
    },
    {
      "epoch": 0.008380122349786306,
      "grad_norm": 0.4752282077738,
      "learning_rate": 1.0888888888888888e-07,
      "loss": 1.325,
      "step": 50
    },
    {
      "epoch": 0.008547724796782032,
      "grad_norm": 0.44346337488294035,
      "learning_rate": 1.1111111111111111e-07,
      "loss": 1.2939,
      "step": 51
    },
    {
      "epoch": 0.008715327243777759,
      "grad_norm": 0.4953251313946097,
      "learning_rate": 1.1333333333333333e-07,
      "loss": 1.3391,
      "step": 52
    },
    {
      "epoch": 0.008882929690773485,
      "grad_norm": 0.5186558211517741,
      "learning_rate": 1.1555555555555554e-07,
      "loss": 1.3271,
      "step": 53
    },
    {
      "epoch": 0.009050532137769211,
      "grad_norm": 0.47548769368143906,
      "learning_rate": 1.1777777777777778e-07,
      "loss": 1.3068,
      "step": 54
    },
    {
      "epoch": 0.009218134584764937,
      "grad_norm": 0.5018425264961417,
      "learning_rate": 1.2e-07,
      "loss": 1.3515,
      "step": 55
    },
    {
      "epoch": 0.009385737031760663,
      "grad_norm": 0.5077532845732232,
      "learning_rate": 1.2222222222222222e-07,
      "loss": 1.3648,
      "step": 56
    },
    {
      "epoch": 0.00955333947875639,
      "grad_norm": 0.5448058293101531,
      "learning_rate": 1.2444444444444443e-07,
      "loss": 1.3967,
      "step": 57
    },
    {
      "epoch": 0.009720941925752116,
      "grad_norm": 0.5213336183942815,
      "learning_rate": 1.2666666666666666e-07,
      "loss": 1.3422,
      "step": 58
    },
    {
      "epoch": 0.009888544372747842,
      "grad_norm": 0.5506085480904805,
      "learning_rate": 1.288888888888889e-07,
      "loss": 1.3702,
      "step": 59
    },
    {
      "epoch": 0.010056146819743568,
      "grad_norm": 0.48956359467022464,
      "learning_rate": 1.311111111111111e-07,
      "loss": 1.3605,
      "step": 60
    },
    {
      "epoch": 0.010223749266739295,
      "grad_norm": 0.5346600932344424,
      "learning_rate": 1.333333333333333e-07,
      "loss": 1.3577,
      "step": 61
    },
    {
      "epoch": 0.01039135171373502,
      "grad_norm": 0.5302917597726902,
      "learning_rate": 1.3555555555555557e-07,
      "loss": 1.4087,
      "step": 62
    },
    {
      "epoch": 0.010558954160730747,
      "grad_norm": 0.4777828235237001,
      "learning_rate": 1.3777777777777778e-07,
      "loss": 1.353,
      "step": 63
    },
    {
      "epoch": 0.010726556607726473,
      "grad_norm": 0.5118939139925528,
      "learning_rate": 1.3999999999999998e-07,
      "loss": 1.3326,
      "step": 64
    },
    {
      "epoch": 0.0108941590547222,
      "grad_norm": 0.4443298461514927,
      "learning_rate": 1.4222222222222222e-07,
      "loss": 1.2474,
      "step": 65
    },
    {
      "epoch": 0.011061761501717926,
      "grad_norm": 0.5926631325692097,
      "learning_rate": 1.4444444444444442e-07,
      "loss": 1.3956,
      "step": 66
    },
    {
      "epoch": 0.011229363948713652,
      "grad_norm": 0.5646629316460601,
      "learning_rate": 1.4666666666666666e-07,
      "loss": 1.4039,
      "step": 67
    },
    {
      "epoch": 0.011396966395709378,
      "grad_norm": 0.4309845467704291,
      "learning_rate": 1.488888888888889e-07,
      "loss": 1.2353,
      "step": 68
    },
    {
      "epoch": 0.011564568842705103,
      "grad_norm": 0.5432259146723719,
      "learning_rate": 1.511111111111111e-07,
      "loss": 1.3499,
      "step": 69
    },
    {
      "epoch": 0.011732171289700829,
      "grad_norm": 0.5346149657408734,
      "learning_rate": 1.5333333333333333e-07,
      "loss": 1.4101,
      "step": 70
    },
    {
      "epoch": 0.011899773736696555,
      "grad_norm": 0.5526317757110711,
      "learning_rate": 1.5555555555555556e-07,
      "loss": 1.3558,
      "step": 71
    },
    {
      "epoch": 0.012067376183692281,
      "grad_norm": 0.5591982488798667,
      "learning_rate": 1.5777777777777777e-07,
      "loss": 1.3334,
      "step": 72
    },
    {
      "epoch": 0.012234978630688008,
      "grad_norm": 0.4480641963444938,
      "learning_rate": 1.6e-07,
      "loss": 1.3038,
      "step": 73
    },
    {
      "epoch": 0.012402581077683734,
      "grad_norm": 0.4155500644488443,
      "learning_rate": 1.622222222222222e-07,
      "loss": 1.2124,
      "step": 74
    },
    {
      "epoch": 0.01257018352467946,
      "grad_norm": 0.5388660420282592,
      "learning_rate": 1.6444444444444442e-07,
      "loss": 1.3742,
      "step": 75
    },
    {
      "epoch": 0.012737785971675186,
      "grad_norm": 0.5702279971482739,
      "learning_rate": 1.6666666666666668e-07,
      "loss": 1.365,
      "step": 76
    },
    {
      "epoch": 0.012905388418670912,
      "grad_norm": 0.48579730177676816,
      "learning_rate": 1.6888888888888888e-07,
      "loss": 1.3572,
      "step": 77
    },
    {
      "epoch": 0.013072990865666639,
      "grad_norm": 0.506449462657981,
      "learning_rate": 1.711111111111111e-07,
      "loss": 1.3091,
      "step": 78
    },
    {
      "epoch": 0.013240593312662365,
      "grad_norm": 0.535555575757291,
      "learning_rate": 1.7333333333333332e-07,
      "loss": 1.4286,
      "step": 79
    },
    {
      "epoch": 0.013408195759658091,
      "grad_norm": 0.4495053460497078,
      "learning_rate": 1.7555555555555556e-07,
      "loss": 1.2345,
      "step": 80
    },
    {
      "epoch": 0.013575798206653817,
      "grad_norm": 0.5653846819159263,
      "learning_rate": 1.7777777777777776e-07,
      "loss": 1.4094,
      "step": 81
    },
    {
      "epoch": 0.013743400653649544,
      "grad_norm": 0.5867538599755305,
      "learning_rate": 1.8e-07,
      "loss": 1.4806,
      "step": 82
    },
    {
      "epoch": 0.01391100310064527,
      "grad_norm": 0.5107691922541985,
      "learning_rate": 1.822222222222222e-07,
      "loss": 1.3696,
      "step": 83
    },
    {
      "epoch": 0.014078605547640996,
      "grad_norm": 0.5004950194857202,
      "learning_rate": 1.8444444444444444e-07,
      "loss": 1.3661,
      "step": 84
    },
    {
      "epoch": 0.014246207994636722,
      "grad_norm": 0.46857398600608646,
      "learning_rate": 1.8666666666666667e-07,
      "loss": 1.3314,
      "step": 85
    },
    {
      "epoch": 0.014413810441632448,
      "grad_norm": 0.5145099476482636,
      "learning_rate": 1.8888888888888888e-07,
      "loss": 1.4221,
      "step": 86
    },
    {
      "epoch": 0.014581412888628175,
      "grad_norm": 0.5613781176424774,
      "learning_rate": 1.911111111111111e-07,
      "loss": 1.4166,
      "step": 87
    },
    {
      "epoch": 0.014749015335623901,
      "grad_norm": 0.48992572401004697,
      "learning_rate": 1.9333333333333332e-07,
      "loss": 1.3407,
      "step": 88
    },
    {
      "epoch": 0.014916617782619625,
      "grad_norm": 0.5439758550070466,
      "learning_rate": 1.9555555555555555e-07,
      "loss": 1.4252,
      "step": 89
    },
    {
      "epoch": 0.015084220229615352,
      "grad_norm": 0.5122581390858262,
      "learning_rate": 1.9777777777777778e-07,
      "loss": 1.3169,
      "step": 90
    },
    {
      "epoch": 0.015251822676611078,
      "grad_norm": 0.6144603034639337,
      "learning_rate": 2e-07,
      "loss": 1.529,
      "step": 91
    },
    {
      "epoch": 0.015419425123606804,
      "grad_norm": 0.5157304324344351,
      "learning_rate": 1.9999994107865475e-07,
      "loss": 1.3672,
      "step": 92
    },
    {
      "epoch": 0.01558702757060253,
      "grad_norm": 0.5597608667224296,
      "learning_rate": 1.999997643146886e-07,
      "loss": 1.3901,
      "step": 93
    },
    {
      "epoch": 0.01575463001759826,
      "grad_norm": 0.4734171427886312,
      "learning_rate": 1.999994697083097e-07,
      "loss": 1.2657,
      "step": 94
    },
    {
      "epoch": 0.015922232464593983,
      "grad_norm": 0.4693480507347577,
      "learning_rate": 1.9999905725986527e-07,
      "loss": 1.31,
      "step": 95
    },
    {
      "epoch": 0.01608983491158971,
      "grad_norm": 0.5330771208193612,
      "learning_rate": 1.9999852696984139e-07,
      "loss": 1.4029,
      "step": 96
    },
    {
      "epoch": 0.016257437358585435,
      "grad_norm": 0.5025157378028816,
      "learning_rate": 1.9999787883886294e-07,
      "loss": 1.2764,
      "step": 97
    },
    {
      "epoch": 0.016425039805581163,
      "grad_norm": 0.5371504340185435,
      "learning_rate": 1.999971128676937e-07,
      "loss": 1.4348,
      "step": 98
    },
    {
      "epoch": 0.016592642252576888,
      "grad_norm": 0.5428300226237261,
      "learning_rate": 1.9999622905723632e-07,
      "loss": 1.4074,
      "step": 99
    },
    {
      "epoch": 0.016760244699572612,
      "grad_norm": 0.5310732872364591,
      "learning_rate": 1.9999522740853226e-07,
      "loss": 1.373,
      "step": 100
    },
    {
      "epoch": 0.01692784714656834,
      "grad_norm": 0.5479222204882496,
      "learning_rate": 1.9999410792276196e-07,
      "loss": 1.4177,
      "step": 101
    },
    {
      "epoch": 0.017095449593564065,
      "grad_norm": 0.5365108219238284,
      "learning_rate": 1.9999287060124462e-07,
      "loss": 1.3589,
      "step": 102
    },
    {
      "epoch": 0.017263052040559793,
      "grad_norm": 0.5430455044460448,
      "learning_rate": 1.9999151544543832e-07,
      "loss": 1.3773,
      "step": 103
    },
    {
      "epoch": 0.017430654487555517,
      "grad_norm": 0.5011512318126362,
      "learning_rate": 1.9999004245694002e-07,
      "loss": 1.3465,
      "step": 104
    },
    {
      "epoch": 0.017598256934551245,
      "grad_norm": 0.5369206425191205,
      "learning_rate": 1.999884516374855e-07,
      "loss": 1.3625,
      "step": 105
    },
    {
      "epoch": 0.01776585938154697,
      "grad_norm": 0.4868956607689913,
      "learning_rate": 1.999867429889495e-07,
      "loss": 1.3166,
      "step": 106
    },
    {
      "epoch": 0.017933461828542697,
      "grad_norm": 0.5121463876329152,
      "learning_rate": 1.9998491651334547e-07,
      "loss": 1.3305,
      "step": 107
    },
    {
      "epoch": 0.018101064275538422,
      "grad_norm": 0.46103466152986083,
      "learning_rate": 1.9998297221282582e-07,
      "loss": 1.3029,
      "step": 108
    },
    {
      "epoch": 0.01826866672253415,
      "grad_norm": 0.5368528653592267,
      "learning_rate": 1.9998091008968174e-07,
      "loss": 1.4279,
      "step": 109
    },
    {
      "epoch": 0.018436269169529874,
      "grad_norm": 0.45454785099935335,
      "learning_rate": 1.9997873014634328e-07,
      "loss": 1.2582,
      "step": 110
    },
    {
      "epoch": 0.018603871616525602,
      "grad_norm": 0.47438920013610203,
      "learning_rate": 1.9997643238537937e-07,
      "loss": 1.3185,
      "step": 111
    },
    {
      "epoch": 0.018771474063521327,
      "grad_norm": 0.41137873162260996,
      "learning_rate": 1.9997401680949776e-07,
      "loss": 1.2715,
      "step": 112
    },
    {
      "epoch": 0.018939076510517055,
      "grad_norm": 0.48770708247143446,
      "learning_rate": 1.9997148342154502e-07,
      "loss": 1.2935,
      "step": 113
    },
    {
      "epoch": 0.01910667895751278,
      "grad_norm": 0.5544614833323931,
      "learning_rate": 1.9996883222450657e-07,
      "loss": 1.4064,
      "step": 114
    },
    {
      "epoch": 0.019274281404508507,
      "grad_norm": 0.5457161929055774,
      "learning_rate": 1.999660632215066e-07,
      "loss": 1.3925,
      "step": 115
    },
    {
      "epoch": 0.019441883851504232,
      "grad_norm": 0.45403992870082993,
      "learning_rate": 1.9996317641580821e-07,
      "loss": 1.2708,
      "step": 116
    },
    {
      "epoch": 0.01960948629849996,
      "grad_norm": 0.5280828251648372,
      "learning_rate": 1.9996017181081335e-07,
      "loss": 1.3756,
      "step": 117
    },
    {
      "epoch": 0.019777088745495684,
      "grad_norm": 0.526481270399972,
      "learning_rate": 1.9995704941006264e-07,
      "loss": 1.349,
      "step": 118
    },
    {
      "epoch": 0.01994469119249141,
      "grad_norm": 0.5274057206160839,
      "learning_rate": 1.9995380921723562e-07,
      "loss": 1.3725,
      "step": 119
    },
    {
      "epoch": 0.020112293639487137,
      "grad_norm": 0.5214269876162836,
      "learning_rate": 1.9995045123615068e-07,
      "loss": 1.3522,
      "step": 120
    },
    {
      "epoch": 0.02027989608648286,
      "grad_norm": 0.4308558077395662,
      "learning_rate": 1.9994697547076487e-07,
      "loss": 1.2161,
      "step": 121
    },
    {
      "epoch": 0.02044749853347859,
      "grad_norm": 0.47516941213508385,
      "learning_rate": 1.9994338192517418e-07,
      "loss": 1.2894,
      "step": 122
    },
    {
      "epoch": 0.020615100980474314,
      "grad_norm": 0.5192577704812598,
      "learning_rate": 1.9993967060361333e-07,
      "loss": 1.3535,
      "step": 123
    },
    {
      "epoch": 0.02078270342747004,
      "grad_norm": 0.5786785036120274,
      "learning_rate": 1.9993584151045586e-07,
      "loss": 1.4183,
      "step": 124
    },
    {
      "epoch": 0.020950305874465766,
      "grad_norm": 0.5001088266610305,
      "learning_rate": 1.9993189465021402e-07,
      "loss": 1.3574,
      "step": 125
    },
    {
      "epoch": 0.021117908321461494,
      "grad_norm": 0.5119027818828682,
      "learning_rate": 1.9992783002753895e-07,
      "loss": 1.3377,
      "step": 126
    },
    {
      "epoch": 0.02128551076845722,
      "grad_norm": 0.5229850931514177,
      "learning_rate": 1.999236476472205e-07,
      "loss": 1.3775,
      "step": 127
    },
    {
      "epoch": 0.021453113215452946,
      "grad_norm": 0.4745883438687319,
      "learning_rate": 1.9991934751418727e-07,
      "loss": 1.3033,
      "step": 128
    },
    {
      "epoch": 0.02162071566244867,
      "grad_norm": 0.47364061766662047,
      "learning_rate": 1.9991492963350667e-07,
      "loss": 1.2799,
      "step": 129
    },
    {
      "epoch": 0.0217883181094444,
      "grad_norm": 0.5240194257343161,
      "learning_rate": 1.9991039401038488e-07,
      "loss": 1.3991,
      "step": 130
    },
    {
      "epoch": 0.021955920556440123,
      "grad_norm": 0.46329923326276984,
      "learning_rate": 1.9990574065016674e-07,
      "loss": 1.2932,
      "step": 131
    },
    {
      "epoch": 0.02212352300343585,
      "grad_norm": 0.504108659949296,
      "learning_rate": 1.9990096955833599e-07,
      "loss": 1.3555,
      "step": 132
    },
    {
      "epoch": 0.022291125450431576,
      "grad_norm": 0.4755555299825361,
      "learning_rate": 1.9989608074051488e-07,
      "loss": 1.3199,
      "step": 133
    },
    {
      "epoch": 0.022458727897427304,
      "grad_norm": 0.4491287448320084,
      "learning_rate": 1.9989107420246462e-07,
      "loss": 1.2484,
      "step": 134
    },
    {
      "epoch": 0.02262633034442303,
      "grad_norm": 0.4876925514597572,
      "learning_rate": 1.9988594995008503e-07,
      "loss": 1.3091,
      "step": 135
    },
    {
      "epoch": 0.022793932791418756,
      "grad_norm": 0.48123882288532327,
      "learning_rate": 1.9988070798941464e-07,
      "loss": 1.314,
      "step": 136
    },
    {
      "epoch": 0.02296153523841448,
      "grad_norm": 0.5106116619961176,
      "learning_rate": 1.998753483266308e-07,
      "loss": 1.3806,
      "step": 137
    },
    {
      "epoch": 0.023129137685410205,
      "grad_norm": 0.49966278045936685,
      "learning_rate": 1.9986987096804937e-07,
      "loss": 1.3051,
      "step": 138
    },
    {
      "epoch": 0.023296740132405933,
      "grad_norm": 0.5158717245394938,
      "learning_rate": 1.998642759201251e-07,
      "loss": 1.3917,
      "step": 139
    },
    {
      "epoch": 0.023464342579401658,
      "grad_norm": 0.5171233776804528,
      "learning_rate": 1.9985856318945127e-07,
      "loss": 1.3309,
      "step": 140
    },
    {
      "epoch": 0.023631945026397386,
      "grad_norm": 0.5693349557609715,
      "learning_rate": 1.9985273278275997e-07,
      "loss": 1.4254,
      "step": 141
    },
    {
      "epoch": 0.02379954747339311,
      "grad_norm": 0.5009823370733918,
      "learning_rate": 1.998467847069219e-07,
      "loss": 1.2624,
      "step": 142
    },
    {
      "epoch": 0.023967149920388838,
      "grad_norm": 0.5297112129292435,
      "learning_rate": 1.9984071896894643e-07,
      "loss": 1.3688,
      "step": 143
    },
    {
      "epoch": 0.024134752367384563,
      "grad_norm": 0.501608550071177,
      "learning_rate": 1.998345355759816e-07,
      "loss": 1.3819,
      "step": 144
    },
    {
      "epoch": 0.02430235481438029,
      "grad_norm": 0.5627396308512745,
      "learning_rate": 1.9982823453531409e-07,
      "loss": 1.3824,
      "step": 145
    },
    {
      "epoch": 0.024469957261376015,
      "grad_norm": 0.49433388572683085,
      "learning_rate": 1.9982181585436914e-07,
      "loss": 1.3411,
      "step": 146
    },
    {
      "epoch": 0.024637559708371743,
      "grad_norm": 0.5246044739714362,
      "learning_rate": 1.9981527954071079e-07,
      "loss": 1.3114,
      "step": 147
    },
    {
      "epoch": 0.024805162155367468,
      "grad_norm": 0.48726899918112826,
      "learning_rate": 1.9980862560204156e-07,
      "loss": 1.3054,
      "step": 148
    },
    {
      "epoch": 0.024972764602363196,
      "grad_norm": 0.4603144674272081,
      "learning_rate": 1.9980185404620266e-07,
      "loss": 1.239,
      "step": 149
    },
    {
      "epoch": 0.02514036704935892,
      "grad_norm": 0.4540208982480958,
      "learning_rate": 1.9979496488117388e-07,
      "loss": 1.3148,
      "step": 150
    },
    {
      "epoch": 0.025307969496354648,
      "grad_norm": 0.5445723262862716,
      "learning_rate": 1.9978795811507352e-07,
      "loss": 1.3808,
      "step": 151
    },
    {
      "epoch": 0.025475571943350372,
      "grad_norm": 0.5876487995785611,
      "learning_rate": 1.9978083375615863e-07,
      "loss": 1.37,
      "step": 152
    },
    {
      "epoch": 0.0256431743903461,
      "grad_norm": 0.49097103820308097,
      "learning_rate": 1.9977359181282473e-07,
      "loss": 1.2766,
      "step": 153
    },
    {
      "epoch": 0.025810776837341825,
      "grad_norm": 0.45496782124749124,
      "learning_rate": 1.9976623229360588e-07,
      "loss": 1.2524,
      "step": 154
    },
    {
      "epoch": 0.025978379284337553,
      "grad_norm": 0.5247976307057965,
      "learning_rate": 1.9975875520717475e-07,
      "loss": 1.3602,
      "step": 155
    },
    {
      "epoch": 0.026145981731333277,
      "grad_norm": 0.4646018072252433,
      "learning_rate": 1.9975116056234255e-07,
      "loss": 1.2306,
      "step": 156
    },
    {
      "epoch": 0.026313584178329005,
      "grad_norm": 0.5122564440466482,
      "learning_rate": 1.9974344836805903e-07,
      "loss": 1.3744,
      "step": 157
    },
    {
      "epoch": 0.02648118662532473,
      "grad_norm": 0.42439007531221484,
      "learning_rate": 1.9973561863341243e-07,
      "loss": 1.2495,
      "step": 158
    },
    {
      "epoch": 0.026648789072320454,
      "grad_norm": 0.5390393285403716,
      "learning_rate": 1.997276713676295e-07,
      "loss": 1.3901,
      "step": 159
    },
    {
      "epoch": 0.026816391519316182,
      "grad_norm": 0.5177834181672666,
      "learning_rate": 1.9971960658007555e-07,
      "loss": 1.3719,
      "step": 160
    },
    {
      "epoch": 0.026983993966311907,
      "grad_norm": 0.5242855334612629,
      "learning_rate": 1.997114242802543e-07,
      "loss": 1.3474,
      "step": 161
    },
    {
      "epoch": 0.027151596413307635,
      "grad_norm": 0.5200533267633689,
      "learning_rate": 1.9970312447780805e-07,
      "loss": 1.3684,
      "step": 162
    },
    {
      "epoch": 0.02731919886030336,
      "grad_norm": 0.5825783377436918,
      "learning_rate": 1.9969470718251745e-07,
      "loss": 1.4479,
      "step": 163
    },
    {
      "epoch": 0.027486801307299087,
      "grad_norm": 0.5085608214401846,
      "learning_rate": 1.996861724043017e-07,
      "loss": 1.2899,
      "step": 164
    },
    {
      "epoch": 0.02765440375429481,
      "grad_norm": 0.5182263528846582,
      "learning_rate": 1.9967752015321844e-07,
      "loss": 1.2769,
      "step": 165
    },
    {
      "epoch": 0.02782200620129054,
      "grad_norm": 0.4907602457997884,
      "learning_rate": 1.9966875043946365e-07,
      "loss": 1.3443,
      "step": 166
    },
    {
      "epoch": 0.027989608648286264,
      "grad_norm": 0.5016885520917153,
      "learning_rate": 1.9965986327337182e-07,
      "loss": 1.3434,
      "step": 167
    },
    {
      "epoch": 0.028157211095281992,
      "grad_norm": 0.5805437190589102,
      "learning_rate": 1.9965085866541586e-07,
      "loss": 1.4184,
      "step": 168
    },
    {
      "epoch": 0.028324813542277717,
      "grad_norm": 0.5417884995144451,
      "learning_rate": 1.9964173662620701e-07,
      "loss": 1.3338,
      "step": 169
    },
    {
      "epoch": 0.028492415989273445,
      "grad_norm": 0.4964943765674787,
      "learning_rate": 1.9963249716649494e-07,
      "loss": 1.335,
      "step": 170
    },
    {
      "epoch": 0.02866001843626917,
      "grad_norm": 0.4498801262262141,
      "learning_rate": 1.9962314029716766e-07,
      "loss": 1.2579,
      "step": 171
    },
    {
      "epoch": 0.028827620883264897,
      "grad_norm": 0.43515947080889394,
      "learning_rate": 1.9961366602925155e-07,
      "loss": 1.2279,
      "step": 172
    },
    {
      "epoch": 0.02899522333026062,
      "grad_norm": 0.49052910038925795,
      "learning_rate": 1.996040743739114e-07,
      "loss": 1.342,
      "step": 173
    },
    {
      "epoch": 0.02916282577725635,
      "grad_norm": 0.48105596041222143,
      "learning_rate": 1.995943653424502e-07,
      "loss": 1.3046,
      "step": 174
    },
    {
      "epoch": 0.029330428224252074,
      "grad_norm": 0.4423989377506611,
      "learning_rate": 1.9958453894630938e-07,
      "loss": 1.2562,
      "step": 175
    },
    {
      "epoch": 0.029498030671247802,
      "grad_norm": 0.4480878104664224,
      "learning_rate": 1.9957459519706862e-07,
      "loss": 1.2463,
      "step": 176
    },
    {
      "epoch": 0.029665633118243526,
      "grad_norm": 0.56449875163635,
      "learning_rate": 1.995645341064459e-07,
      "loss": 1.388,
      "step": 177
    },
    {
      "epoch": 0.02983323556523925,
      "grad_norm": 0.5501637515611523,
      "learning_rate": 1.995543556862975e-07,
      "loss": 1.3225,
      "step": 178
    },
    {
      "epoch": 0.03000083801223498,
      "grad_norm": 0.5138760388716799,
      "learning_rate": 1.9954405994861788e-07,
      "loss": 1.3572,
      "step": 179
    },
    {
      "epoch": 0.030168440459230703,
      "grad_norm": 0.5187932213299531,
      "learning_rate": 1.995336469055399e-07,
      "loss": 1.3574,
      "step": 180
    },
    {
      "epoch": 0.03033604290622643,
      "grad_norm": 0.51837153994281,
      "learning_rate": 1.9952311656933447e-07,
      "loss": 1.3561,
      "step": 181
    },
    {
      "epoch": 0.030503645353222156,
      "grad_norm": 0.42103589305098776,
      "learning_rate": 1.9951246895241094e-07,
      "loss": 1.2882,
      "step": 182
    },
    {
      "epoch": 0.030671247800217884,
      "grad_norm": 0.5352155423826417,
      "learning_rate": 1.9950170406731667e-07,
      "loss": 1.3407,
      "step": 183
    },
    {
      "epoch": 0.030838850247213608,
      "grad_norm": 0.46189682827561745,
      "learning_rate": 1.9949082192673727e-07,
      "loss": 1.2818,
      "step": 184
    },
    {
      "epoch": 0.031006452694209336,
      "grad_norm": 0.45820583325287606,
      "learning_rate": 1.9947982254349665e-07,
      "loss": 1.2231,
      "step": 185
    },
    {
      "epoch": 0.03117405514120506,
      "grad_norm": 0.5420444956972503,
      "learning_rate": 1.9946870593055668e-07,
      "loss": 1.4109,
      "step": 186
    },
    {
      "epoch": 0.03134165758820079,
      "grad_norm": 0.5310941239599197,
      "learning_rate": 1.9945747210101752e-07,
      "loss": 1.3199,
      "step": 187
    },
    {
      "epoch": 0.03150926003519652,
      "grad_norm": 0.43857756950471327,
      "learning_rate": 1.9944612106811737e-07,
      "loss": 1.23,
      "step": 188
    },
    {
      "epoch": 0.03167686248219224,
      "grad_norm": 0.5335009287007911,
      "learning_rate": 1.9943465284523265e-07,
      "loss": 1.3481,
      "step": 189
    },
    {
      "epoch": 0.031844464929187966,
      "grad_norm": 0.5063467700412242,
      "learning_rate": 1.9942306744587783e-07,
      "loss": 1.2878,
      "step": 190
    },
    {
      "epoch": 0.032012067376183694,
      "grad_norm": 0.47719525009922403,
      "learning_rate": 1.994113648837054e-07,
      "loss": 1.2713,
      "step": 191
    },
    {
      "epoch": 0.03217966982317942,
      "grad_norm": 0.5424652604259685,
      "learning_rate": 1.9939954517250603e-07,
      "loss": 1.3511,
      "step": 192
    },
    {
      "epoch": 0.03234727227017514,
      "grad_norm": 0.46520455464853633,
      "learning_rate": 1.9938760832620834e-07,
      "loss": 1.2485,
      "step": 193
    },
    {
      "epoch": 0.03251487471717087,
      "grad_norm": 0.46760883406507714,
      "learning_rate": 1.9937555435887904e-07,
      "loss": 1.3628,
      "step": 194
    },
    {
      "epoch": 0.0326824771641666,
      "grad_norm": 0.5572844292341892,
      "learning_rate": 1.9936338328472287e-07,
      "loss": 1.3631,
      "step": 195
    },
    {
      "epoch": 0.032850079611162326,
      "grad_norm": 0.4323470349624805,
      "learning_rate": 1.9935109511808253e-07,
      "loss": 1.2369,
      "step": 196
    },
    {
      "epoch": 0.03301768205815805,
      "grad_norm": 0.47391064735085864,
      "learning_rate": 1.9933868987343874e-07,
      "loss": 1.3079,
      "step": 197
    },
    {
      "epoch": 0.033185284505153775,
      "grad_norm": 0.512551358988816,
      "learning_rate": 1.9932616756541018e-07,
      "loss": 1.3737,
      "step": 198
    },
    {
      "epoch": 0.0333528869521495,
      "grad_norm": 0.5821390585924014,
      "learning_rate": 1.9931352820875346e-07,
      "loss": 1.4685,
      "step": 199
    },
    {
      "epoch": 0.033520489399145224,
      "grad_norm": 0.49191898587040483,
      "learning_rate": 1.9930077181836314e-07,
      "loss": 1.3226,
      "step": 200
    },
    {
      "epoch": 0.03368809184614095,
      "grad_norm": 0.5012090218492163,
      "learning_rate": 1.992878984092717e-07,
      "loss": 1.3301,
      "step": 201
    },
    {
      "epoch": 0.03385569429313668,
      "grad_norm": 0.4793700357504222,
      "learning_rate": 1.992749079966495e-07,
      "loss": 1.2958,
      "step": 202
    },
    {
      "epoch": 0.03402329674013241,
      "grad_norm": 0.516532531311271,
      "learning_rate": 1.992618005958048e-07,
      "loss": 1.3501,
      "step": 203
    },
    {
      "epoch": 0.03419089918712813,
      "grad_norm": 0.4886897326530053,
      "learning_rate": 1.9924857622218371e-07,
      "loss": 1.2821,
      "step": 204
    },
    {
      "epoch": 0.03435850163412386,
      "grad_norm": 0.4514370913849233,
      "learning_rate": 1.9923523489137022e-07,
      "loss": 1.2779,
      "step": 205
    },
    {
      "epoch": 0.034526104081119585,
      "grad_norm": 0.47776102922470476,
      "learning_rate": 1.9922177661908605e-07,
      "loss": 1.2897,
      "step": 206
    },
    {
      "epoch": 0.03469370652811531,
      "grad_norm": 0.5490963196426525,
      "learning_rate": 1.9920820142119085e-07,
      "loss": 1.3604,
      "step": 207
    },
    {
      "epoch": 0.034861308975111034,
      "grad_norm": 0.5492848272181189,
      "learning_rate": 1.9919450931368195e-07,
      "loss": 1.438,
      "step": 208
    },
    {
      "epoch": 0.03502891142210676,
      "grad_norm": 0.5487395621316232,
      "learning_rate": 1.991807003126945e-07,
      "loss": 1.3707,
      "step": 209
    },
    {
      "epoch": 0.03519651386910249,
      "grad_norm": 0.4461556331729256,
      "learning_rate": 1.9916677443450149e-07,
      "loss": 1.2836,
      "step": 210
    },
    {
      "epoch": 0.03536411631609822,
      "grad_norm": 0.44957882461413834,
      "learning_rate": 1.991527316955134e-07,
      "loss": 1.2581,
      "step": 211
    },
    {
      "epoch": 0.03553171876309394,
      "grad_norm": 0.45094739237428927,
      "learning_rate": 1.991385721122787e-07,
      "loss": 1.1811,
      "step": 212
    },
    {
      "epoch": 0.03569932121008967,
      "grad_norm": 0.5020878863090336,
      "learning_rate": 1.9912429570148337e-07,
      "loss": 1.291,
      "step": 213
    },
    {
      "epoch": 0.035866923657085395,
      "grad_norm": 0.5252056919611553,
      "learning_rate": 1.991099024799511e-07,
      "loss": 1.3548,
      "step": 214
    },
    {
      "epoch": 0.03603452610408112,
      "grad_norm": 0.5310131363869731,
      "learning_rate": 1.9909539246464327e-07,
      "loss": 1.3554,
      "step": 215
    },
    {
      "epoch": 0.036202128551076844,
      "grad_norm": 0.49157075785777443,
      "learning_rate": 1.9908076567265888e-07,
      "loss": 1.3158,
      "step": 216
    },
    {
      "epoch": 0.03636973099807257,
      "grad_norm": 0.48873926633440734,
      "learning_rate": 1.9906602212123455e-07,
      "loss": 1.2855,
      "step": 217
    },
    {
      "epoch": 0.0365373334450683,
      "grad_norm": 0.4631451361175071,
      "learning_rate": 1.9905116182774445e-07,
      "loss": 1.2576,
      "step": 218
    },
    {
      "epoch": 0.03670493589206402,
      "grad_norm": 0.4537436329018545,
      "learning_rate": 1.9903618480970033e-07,
      "loss": 1.2829,
      "step": 219
    },
    {
      "epoch": 0.03687253833905975,
      "grad_norm": 0.4760773774428117,
      "learning_rate": 1.9902109108475153e-07,
      "loss": 1.3076,
      "step": 220
    },
    {
      "epoch": 0.03704014078605548,
      "grad_norm": 0.5072693038524785,
      "learning_rate": 1.990058806706849e-07,
      "loss": 1.361,
      "step": 221
    },
    {
      "epoch": 0.037207743233051205,
      "grad_norm": 0.4679285532669848,
      "learning_rate": 1.9899055358542484e-07,
      "loss": 1.3019,
      "step": 222
    },
    {
      "epoch": 0.037375345680046926,
      "grad_norm": 0.4792610102411162,
      "learning_rate": 1.9897510984703315e-07,
      "loss": 1.3574,
      "step": 223
    },
    {
      "epoch": 0.037542948127042654,
      "grad_norm": 0.46691145662572786,
      "learning_rate": 1.989595494737092e-07,
      "loss": 1.2663,
      "step": 224
    },
    {
      "epoch": 0.03771055057403838,
      "grad_norm": 0.4642357395833113,
      "learning_rate": 1.9894387248378967e-07,
      "loss": 1.2664,
      "step": 225
    },
    {
      "epoch": 0.03787815302103411,
      "grad_norm": 0.5248553463244476,
      "learning_rate": 1.9892807889574881e-07,
      "loss": 1.3279,
      "step": 226
    },
    {
      "epoch": 0.03804575546802983,
      "grad_norm": 0.5096179638928433,
      "learning_rate": 1.9891216872819822e-07,
      "loss": 1.3017,
      "step": 227
    },
    {
      "epoch": 0.03821335791502556,
      "grad_norm": 0.5158242172893513,
      "learning_rate": 1.9889614199988684e-07,
      "loss": 1.3567,
      "step": 228
    },
    {
      "epoch": 0.03838096036202129,
      "grad_norm": 0.48958290629245,
      "learning_rate": 1.9887999872970095e-07,
      "loss": 1.3054,
      "step": 229
    },
    {
      "epoch": 0.038548562809017015,
      "grad_norm": 0.5491977404604751,
      "learning_rate": 1.988637389366643e-07,
      "loss": 1.3951,
      "step": 230
    },
    {
      "epoch": 0.038716165256012736,
      "grad_norm": 0.4837294195757653,
      "learning_rate": 1.988473626399378e-07,
      "loss": 1.2792,
      "step": 231
    },
    {
      "epoch": 0.038883767703008464,
      "grad_norm": 0.4968790383923519,
      "learning_rate": 1.9883086985881977e-07,
      "loss": 1.3247,
      "step": 232
    },
    {
      "epoch": 0.03905137015000419,
      "grad_norm": 0.49911118383380754,
      "learning_rate": 1.9881426061274576e-07,
      "loss": 1.2543,
      "step": 233
    },
    {
      "epoch": 0.03921897259699992,
      "grad_norm": 0.5689967596481516,
      "learning_rate": 1.987975349212885e-07,
      "loss": 1.3643,
      "step": 234
    },
    {
      "epoch": 0.03938657504399564,
      "grad_norm": 0.503177965072902,
      "learning_rate": 1.9878069280415802e-07,
      "loss": 1.3327,
      "step": 235
    },
    {
      "epoch": 0.03955417749099137,
      "grad_norm": 0.4920042856331476,
      "learning_rate": 1.987637342812015e-07,
      "loss": 1.2814,
      "step": 236
    },
    {
      "epoch": 0.039721779937987096,
      "grad_norm": 0.4459588662038392,
      "learning_rate": 1.9874665937240334e-07,
      "loss": 1.283,
      "step": 237
    },
    {
      "epoch": 0.03988938238498282,
      "grad_norm": 0.4707930512856142,
      "learning_rate": 1.987294680978851e-07,
      "loss": 1.2965,
      "step": 238
    },
    {
      "epoch": 0.040056984831978545,
      "grad_norm": 0.5589427154248711,
      "learning_rate": 1.9871216047790534e-07,
      "loss": 1.4077,
      "step": 239
    },
    {
      "epoch": 0.04022458727897427,
      "grad_norm": 0.5299478266307756,
      "learning_rate": 1.9869473653285997e-07,
      "loss": 1.3856,
      "step": 240
    },
    {
      "epoch": 0.04039218972597,
      "grad_norm": 0.4921988532653907,
      "learning_rate": 1.9867719628328175e-07,
      "loss": 1.3399,
      "step": 241
    },
    {
      "epoch": 0.04055979217296572,
      "grad_norm": 0.4984779504999269,
      "learning_rate": 1.9865953974984055e-07,
      "loss": 1.3422,
      "step": 242
    },
    {
      "epoch": 0.04072739461996145,
      "grad_norm": 0.5628551096741782,
      "learning_rate": 1.9864176695334338e-07,
      "loss": 1.4394,
      "step": 243
    },
    {
      "epoch": 0.04089499706695718,
      "grad_norm": 0.42921823914680163,
      "learning_rate": 1.9862387791473417e-07,
      "loss": 1.2781,
      "step": 244
    },
    {
      "epoch": 0.041062599513952906,
      "grad_norm": 0.5446936150254823,
      "learning_rate": 1.9860587265509378e-07,
      "loss": 1.3209,
      "step": 245
    },
    {
      "epoch": 0.04123020196094863,
      "grad_norm": 0.5240284427626032,
      "learning_rate": 1.9858775119564015e-07,
      "loss": 1.3374,
      "step": 246
    },
    {
      "epoch": 0.041397804407944355,
      "grad_norm": 0.4770675079189546,
      "learning_rate": 1.9856951355772812e-07,
      "loss": 1.2586,
      "step": 247
    },
    {
      "epoch": 0.04156540685494008,
      "grad_norm": 0.5405446525136111,
      "learning_rate": 1.9855115976284932e-07,
      "loss": 1.409,
      "step": 248
    },
    {
      "epoch": 0.04173300930193581,
      "grad_norm": 0.5113826698130063,
      "learning_rate": 1.9853268983263244e-07,
      "loss": 1.3184,
      "step": 249
    },
    {
      "epoch": 0.04190061174893153,
      "grad_norm": 0.48812136556792857,
      "learning_rate": 1.9851410378884287e-07,
      "loss": 1.3589,
      "step": 250
    },
    {
      "epoch": 0.04206821419592726,
      "grad_norm": 0.48119543846767393,
      "learning_rate": 1.9849540165338298e-07,
      "loss": 1.2612,
      "step": 251
    },
    {
      "epoch": 0.04223581664292299,
      "grad_norm": 0.49868669028053686,
      "learning_rate": 1.9847658344829185e-07,
      "loss": 1.3578,
      "step": 252
    },
    {
      "epoch": 0.042403419089918716,
      "grad_norm": 0.4201480818452758,
      "learning_rate": 1.9845764919574536e-07,
      "loss": 1.2404,
      "step": 253
    },
    {
      "epoch": 0.04257102153691444,
      "grad_norm": 0.4852735687635263,
      "learning_rate": 1.9843859891805607e-07,
      "loss": 1.2362,
      "step": 254
    },
    {
      "epoch": 0.042738623983910165,
      "grad_norm": 0.5360656529327268,
      "learning_rate": 1.9841943263767343e-07,
      "loss": 1.3865,
      "step": 255
    },
    {
      "epoch": 0.04290622643090589,
      "grad_norm": 0.4839757120008504,
      "learning_rate": 1.9840015037718347e-07,
      "loss": 1.3524,
      "step": 256
    },
    {
      "epoch": 0.043073828877901614,
      "grad_norm": 0.4698290777614883,
      "learning_rate": 1.983807521593089e-07,
      "loss": 1.3171,
      "step": 257
    },
    {
      "epoch": 0.04324143132489734,
      "grad_norm": 0.5096508525171812,
      "learning_rate": 1.9836123800690917e-07,
      "loss": 1.3393,
      "step": 258
    },
    {
      "epoch": 0.04340903377189307,
      "grad_norm": 0.46562356645728714,
      "learning_rate": 1.9834160794298023e-07,
      "loss": 1.2701,
      "step": 259
    },
    {
      "epoch": 0.0435766362188888,
      "grad_norm": 0.46305789080278575,
      "learning_rate": 1.9832186199065463e-07,
      "loss": 1.2632,
      "step": 260
    },
    {
      "epoch": 0.04374423866588452,
      "grad_norm": 0.4961015593483931,
      "learning_rate": 1.9830200017320164e-07,
      "loss": 1.3287,
      "step": 261
    },
    {
      "epoch": 0.04391184111288025,
      "grad_norm": 0.4720854142971024,
      "learning_rate": 1.9828202251402687e-07,
      "loss": 1.2752,
      "step": 262
    },
    {
      "epoch": 0.044079443559875975,
      "grad_norm": 0.4963578174775106,
      "learning_rate": 1.9826192903667257e-07,
      "loss": 1.2895,
      "step": 263
    },
    {
      "epoch": 0.0442470460068717,
      "grad_norm": 0.4776413837115946,
      "learning_rate": 1.9824171976481744e-07,
      "loss": 1.3322,
      "step": 264
    },
    {
      "epoch": 0.044414648453867424,
      "grad_norm": 0.5509098941312381,
      "learning_rate": 1.9822139472227663e-07,
      "loss": 1.3733,
      "step": 265
    },
    {
      "epoch": 0.04458225090086315,
      "grad_norm": 0.5323271950336386,
      "learning_rate": 1.9820095393300169e-07,
      "loss": 1.3389,
      "step": 266
    },
    {
      "epoch": 0.04474985334785888,
      "grad_norm": 0.43282982517946056,
      "learning_rate": 1.9818039742108062e-07,
      "loss": 1.2668,
      "step": 267
    },
    {
      "epoch": 0.04491745579485461,
      "grad_norm": 0.3428884371289743,
      "learning_rate": 1.9815972521073775e-07,
      "loss": 1.1137,
      "step": 268
    },
    {
      "epoch": 0.04508505824185033,
      "grad_norm": 0.5376874867353777,
      "learning_rate": 1.9813893732633374e-07,
      "loss": 1.3308,
      "step": 269
    },
    {
      "epoch": 0.04525266068884606,
      "grad_norm": 0.40381640539067587,
      "learning_rate": 1.9811803379236572e-07,
      "loss": 1.2211,
      "step": 270
    },
    {
      "epoch": 0.045420263135841785,
      "grad_norm": 0.4012320499311086,
      "learning_rate": 1.9809701463346683e-07,
      "loss": 1.2078,
      "step": 271
    },
    {
      "epoch": 0.04558786558283751,
      "grad_norm": 0.5035036559578892,
      "learning_rate": 1.9807587987440666e-07,
      "loss": 1.3377,
      "step": 272
    },
    {
      "epoch": 0.045755468029833234,
      "grad_norm": 0.518805083207451,
      "learning_rate": 1.98054629540091e-07,
      "loss": 1.3441,
      "step": 273
    },
    {
      "epoch": 0.04592307047682896,
      "grad_norm": 0.5097379553945165,
      "learning_rate": 1.9803326365556182e-07,
      "loss": 1.3187,
      "step": 274
    },
    {
      "epoch": 0.04609067292382469,
      "grad_norm": 0.4660576001664114,
      "learning_rate": 1.980117822459972e-07,
      "loss": 1.2472,
      "step": 275
    },
    {
      "epoch": 0.04625827537082041,
      "grad_norm": 0.4982980105057137,
      "learning_rate": 1.9799018533671148e-07,
      "loss": 1.3275,
      "step": 276
    },
    {
      "epoch": 0.04642587781781614,
      "grad_norm": 0.5266357934592201,
      "learning_rate": 1.9796847295315499e-07,
      "loss": 1.2928,
      "step": 277
    },
    {
      "epoch": 0.046593480264811866,
      "grad_norm": 0.45984925906652413,
      "learning_rate": 1.9794664512091422e-07,
      "loss": 1.3019,
      "step": 278
    },
    {
      "epoch": 0.046761082711807594,
      "grad_norm": 0.453048338152402,
      "learning_rate": 1.9792470186571163e-07,
      "loss": 1.2996,
      "step": 279
    },
    {
      "epoch": 0.046928685158803315,
      "grad_norm": 0.4750068109455625,
      "learning_rate": 1.979026432134058e-07,
      "loss": 1.2958,
      "step": 280
    },
    {
      "epoch": 0.04709628760579904,
      "grad_norm": 0.4718940737972919,
      "learning_rate": 1.978804691899912e-07,
      "loss": 1.3259,
      "step": 281
    },
    {
      "epoch": 0.04726389005279477,
      "grad_norm": 0.4551785450649334,
      "learning_rate": 1.978581798215983e-07,
      "loss": 1.2509,
      "step": 282
    },
    {
      "epoch": 0.0474314924997905,
      "grad_norm": 0.5616173069912492,
      "learning_rate": 1.978357751344935e-07,
      "loss": 1.3943,
      "step": 283
    },
    {
      "epoch": 0.04759909494678622,
      "grad_norm": 0.36879720138710986,
      "learning_rate": 1.9781325515507908e-07,
      "loss": 1.1549,
      "step": 284
    },
    {
      "epoch": 0.04776669739378195,
      "grad_norm": 0.5088004444083827,
      "learning_rate": 1.977906199098932e-07,
      "loss": 1.3032,
      "step": 285
    },
    {
      "epoch": 0.047934299840777676,
      "grad_norm": 0.6160382302763678,
      "learning_rate": 1.9776786942560984e-07,
      "loss": 1.4333,
      "step": 286
    },
    {
      "epoch": 0.048101902287773404,
      "grad_norm": 0.4175006080125338,
      "learning_rate": 1.9774500372903878e-07,
      "loss": 1.2158,
      "step": 287
    },
    {
      "epoch": 0.048269504734769125,
      "grad_norm": 0.5258746031129032,
      "learning_rate": 1.9772202284712554e-07,
      "loss": 1.3332,
      "step": 288
    },
    {
      "epoch": 0.04843710718176485,
      "grad_norm": 0.40584134676561745,
      "learning_rate": 1.9769892680695145e-07,
      "loss": 1.2018,
      "step": 289
    },
    {
      "epoch": 0.04860470962876058,
      "grad_norm": 0.4592495416194577,
      "learning_rate": 1.976757156357335e-07,
      "loss": 1.3,
      "step": 290
    },
    {
      "epoch": 0.04877231207575631,
      "grad_norm": 0.440697934525138,
      "learning_rate": 1.9765238936082435e-07,
      "loss": 1.2389,
      "step": 291
    },
    {
      "epoch": 0.04893991452275203,
      "grad_norm": 0.49788064927764003,
      "learning_rate": 1.9762894800971232e-07,
      "loss": 1.329,
      "step": 292
    },
    {
      "epoch": 0.04910751696974776,
      "grad_norm": 0.45386759249479514,
      "learning_rate": 1.9760539161002131e-07,
      "loss": 1.2342,
      "step": 293
    },
    {
      "epoch": 0.049275119416743486,
      "grad_norm": 0.4413459173275765,
      "learning_rate": 1.9758172018951086e-07,
      "loss": 1.2422,
      "step": 294
    },
    {
      "epoch": 0.04944272186373921,
      "grad_norm": 0.4735721610629954,
      "learning_rate": 1.9755793377607593e-07,
      "loss": 1.2051,
      "step": 295
    },
    {
      "epoch": 0.049610324310734935,
      "grad_norm": 0.49133089279174313,
      "learning_rate": 1.9753403239774712e-07,
      "loss": 1.2977,
      "step": 296
    },
    {
      "epoch": 0.04977792675773066,
      "grad_norm": 0.5144693927980712,
      "learning_rate": 1.9751001608269048e-07,
      "loss": 1.3564,
      "step": 297
    },
    {
      "epoch": 0.04994552920472639,
      "grad_norm": 0.42433421518765174,
      "learning_rate": 1.9748588485920746e-07,
      "loss": 1.2975,
      "step": 298
    },
    {
      "epoch": 0.05011313165172211,
      "grad_norm": 0.47161431722454833,
      "learning_rate": 1.974616387557349e-07,
      "loss": 1.3217,
      "step": 299
    },
    {
      "epoch": 0.05028073409871784,
      "grad_norm": 0.45943216952319593,
      "learning_rate": 1.974372778008451e-07,
      "loss": 1.2493,
      "step": 300
    },
    {
      "epoch": 0.05044833654571357,
      "grad_norm": 0.48759577082250427,
      "learning_rate": 1.9741280202324567e-07,
      "loss": 1.3237,
      "step": 301
    },
    {
      "epoch": 0.050615938992709296,
      "grad_norm": 0.4374537054468286,
      "learning_rate": 1.9738821145177952e-07,
      "loss": 1.2517,
      "step": 302
    },
    {
      "epoch": 0.05078354143970502,
      "grad_norm": 0.4813635923305249,
      "learning_rate": 1.9736350611542484e-07,
      "loss": 1.2887,
      "step": 303
    },
    {
      "epoch": 0.050951143886700745,
      "grad_norm": 0.474191199833103,
      "learning_rate": 1.973386860432951e-07,
      "loss": 1.3062,
      "step": 304
    },
    {
      "epoch": 0.05111874633369647,
      "grad_norm": 0.5504816212330593,
      "learning_rate": 1.9731375126463886e-07,
      "loss": 1.3293,
      "step": 305
    },
    {
      "epoch": 0.0512863487806922,
      "grad_norm": 0.47733560390019636,
      "learning_rate": 1.9728870180883995e-07,
      "loss": 1.2904,
      "step": 306
    },
    {
      "epoch": 0.05145395122768792,
      "grad_norm": 0.47099763956794344,
      "learning_rate": 1.972635377054174e-07,
      "loss": 1.2966,
      "step": 307
    },
    {
      "epoch": 0.05162155367468365,
      "grad_norm": 0.4722314501410655,
      "learning_rate": 1.972382589840252e-07,
      "loss": 1.3225,
      "step": 308
    },
    {
      "epoch": 0.05178915612167938,
      "grad_norm": 0.4685332635250827,
      "learning_rate": 1.9721286567445247e-07,
      "loss": 1.2767,
      "step": 309
    },
    {
      "epoch": 0.051956758568675106,
      "grad_norm": 0.46416635880452733,
      "learning_rate": 1.9718735780662344e-07,
      "loss": 1.2766,
      "step": 310
    },
    {
      "epoch": 0.05212436101567083,
      "grad_norm": 0.5426486081825191,
      "learning_rate": 1.9716173541059718e-07,
      "loss": 1.384,
      "step": 311
    },
    {
      "epoch": 0.052291963462666555,
      "grad_norm": 0.4868958889193987,
      "learning_rate": 1.9713599851656783e-07,
      "loss": 1.2987,
      "step": 312
    },
    {
      "epoch": 0.05245956590966228,
      "grad_norm": 0.41824886290688,
      "learning_rate": 1.9711014715486446e-07,
      "loss": 1.2277,
      "step": 313
    },
    {
      "epoch": 0.05262716835665801,
      "grad_norm": 0.43412108190747556,
      "learning_rate": 1.9708418135595105e-07,
      "loss": 1.219,
      "step": 314
    },
    {
      "epoch": 0.05279477080365373,
      "grad_norm": 0.4972492757358146,
      "learning_rate": 1.9705810115042631e-07,
      "loss": 1.3137,
      "step": 315
    },
    {
      "epoch": 0.05296237325064946,
      "grad_norm": 0.4713625712544483,
      "learning_rate": 1.9703190656902394e-07,
      "loss": 1.282,
      "step": 316
    },
    {
      "epoch": 0.05312997569764519,
      "grad_norm": 0.5575630087166197,
      "learning_rate": 1.9700559764261224e-07,
      "loss": 1.3373,
      "step": 317
    },
    {
      "epoch": 0.05329757814464091,
      "grad_norm": 0.4816758824747167,
      "learning_rate": 1.9697917440219446e-07,
      "loss": 1.308,
      "step": 318
    },
    {
      "epoch": 0.05346518059163664,
      "grad_norm": 0.4910734295344755,
      "learning_rate": 1.969526368789084e-07,
      "loss": 1.2803,
      "step": 319
    },
    {
      "epoch": 0.053632783038632365,
      "grad_norm": 0.5009756862105262,
      "learning_rate": 1.969259851040266e-07,
      "loss": 1.3161,
      "step": 320
    },
    {
      "epoch": 0.05380038548562809,
      "grad_norm": 0.45858029899027525,
      "learning_rate": 1.9689921910895623e-07,
      "loss": 1.2806,
      "step": 321
    },
    {
      "epoch": 0.053967987932623813,
      "grad_norm": 0.5349708261186792,
      "learning_rate": 1.968723389252391e-07,
      "loss": 1.4145,
      "step": 322
    },
    {
      "epoch": 0.05413559037961954,
      "grad_norm": 0.4624419391407823,
      "learning_rate": 1.9684534458455143e-07,
      "loss": 1.2339,
      "step": 323
    },
    {
      "epoch": 0.05430319282661527,
      "grad_norm": 0.48805993735047964,
      "learning_rate": 1.968182361187042e-07,
      "loss": 1.3179,
      "step": 324
    },
    {
      "epoch": 0.054470795273611,
      "grad_norm": 0.47045513068453276,
      "learning_rate": 1.967910135596427e-07,
      "loss": 1.3107,
      "step": 325
    },
    {
      "epoch": 0.05463839772060672,
      "grad_norm": 0.5157206536675213,
      "learning_rate": 1.9676367693944672e-07,
      "loss": 1.3065,
      "step": 326
    },
    {
      "epoch": 0.054806000167602446,
      "grad_norm": 0.4821447247740779,
      "learning_rate": 1.967362262903305e-07,
      "loss": 1.3067,
      "step": 327
    },
    {
      "epoch": 0.054973602614598174,
      "grad_norm": 0.4727217769505724,
      "learning_rate": 1.9670866164464258e-07,
      "loss": 1.2686,
      "step": 328
    },
    {
      "epoch": 0.0551412050615939,
      "grad_norm": 0.4351461485604427,
      "learning_rate": 1.9668098303486592e-07,
      "loss": 1.2332,
      "step": 329
    },
    {
      "epoch": 0.05530880750858962,
      "grad_norm": 0.5093998712214538,
      "learning_rate": 1.966531904936177e-07,
      "loss": 1.3266,
      "step": 330
    },
    {
      "epoch": 0.05547640995558535,
      "grad_norm": 0.5195240616194539,
      "learning_rate": 1.9662528405364945e-07,
      "loss": 1.3157,
      "step": 331
    },
    {
      "epoch": 0.05564401240258108,
      "grad_norm": 0.437987108394993,
      "learning_rate": 1.965972637478468e-07,
      "loss": 1.2763,
      "step": 332
    },
    {
      "epoch": 0.05581161484957681,
      "grad_norm": 0.43603350463408874,
      "learning_rate": 1.965691296092297e-07,
      "loss": 1.2602,
      "step": 333
    },
    {
      "epoch": 0.05597921729657253,
      "grad_norm": 0.508838961703096,
      "learning_rate": 1.9654088167095214e-07,
      "loss": 1.327,
      "step": 334
    },
    {
      "epoch": 0.056146819743568256,
      "grad_norm": 0.47218593370723716,
      "learning_rate": 1.965125199663023e-07,
      "loss": 1.2712,
      "step": 335
    },
    {
      "epoch": 0.056314422190563984,
      "grad_norm": 0.4738219025964394,
      "learning_rate": 1.964840445287023e-07,
      "loss": 1.2882,
      "step": 336
    },
    {
      "epoch": 0.056482024637559705,
      "grad_norm": 0.4212518404433323,
      "learning_rate": 1.9645545539170838e-07,
      "loss": 1.2081,
      "step": 337
    },
    {
      "epoch": 0.05664962708455543,
      "grad_norm": 0.47369255495789575,
      "learning_rate": 1.964267525890108e-07,
      "loss": 1.3072,
      "step": 338
    },
    {
      "epoch": 0.05681722953155116,
      "grad_norm": 0.4883415888101317,
      "learning_rate": 1.9639793615443363e-07,
      "loss": 1.3112,
      "step": 339
    },
    {
      "epoch": 0.05698483197854689,
      "grad_norm": 0.45146258111668897,
      "learning_rate": 1.9636900612193502e-07,
      "loss": 1.2547,
      "step": 340
    },
    {
      "epoch": 0.05715243442554261,
      "grad_norm": 0.44500002010700423,
      "learning_rate": 1.9633996252560686e-07,
      "loss": 1.2412,
      "step": 341
    },
    {
      "epoch": 0.05732003687253834,
      "grad_norm": 0.4353977325166278,
      "learning_rate": 1.9631080539967488e-07,
      "loss": 1.1805,
      "step": 342
    },
    {
      "epoch": 0.057487639319534066,
      "grad_norm": 0.4527663219283007,
      "learning_rate": 1.9628153477849865e-07,
      "loss": 1.2551,
      "step": 343
    },
    {
      "epoch": 0.057655241766529794,
      "grad_norm": 0.4160625195474428,
      "learning_rate": 1.9625215069657148e-07,
      "loss": 1.2428,
      "step": 344
    },
    {
      "epoch": 0.057822844213525515,
      "grad_norm": 0.5145222159900323,
      "learning_rate": 1.9622265318852032e-07,
      "loss": 1.3193,
      "step": 345
    },
    {
      "epoch": 0.05799044666052124,
      "grad_norm": 0.4556363211497156,
      "learning_rate": 1.9619304228910583e-07,
      "loss": 1.3035,
      "step": 346
    },
    {
      "epoch": 0.05815804910751697,
      "grad_norm": 0.4825210581234634,
      "learning_rate": 1.9616331803322232e-07,
      "loss": 1.2858,
      "step": 347
    },
    {
      "epoch": 0.0583256515545127,
      "grad_norm": 0.4520104979861744,
      "learning_rate": 1.9613348045589766e-07,
      "loss": 1.2586,
      "step": 348
    },
    {
      "epoch": 0.05849325400150842,
      "grad_norm": 0.4731246292989499,
      "learning_rate": 1.961035295922932e-07,
      "loss": 1.2845,
      "step": 349
    },
    {
      "epoch": 0.05866085644850415,
      "grad_norm": 0.4260794239591328,
      "learning_rate": 1.9607346547770388e-07,
      "loss": 1.2054,
      "step": 350
    },
    {
      "epoch": 0.058828458895499876,
      "grad_norm": 0.5176808038079587,
      "learning_rate": 1.9604328814755805e-07,
      "loss": 1.3191,
      "step": 351
    },
    {
      "epoch": 0.058996061342495604,
      "grad_norm": 0.40898891739035603,
      "learning_rate": 1.960129976374175e-07,
      "loss": 1.2048,
      "step": 352
    },
    {
      "epoch": 0.059163663789491325,
      "grad_norm": 0.4650809765988873,
      "learning_rate": 1.959825939829774e-07,
      "loss": 1.2839,
      "step": 353
    },
    {
      "epoch": 0.05933126623648705,
      "grad_norm": 0.37016758138547373,
      "learning_rate": 1.959520772200662e-07,
      "loss": 1.1233,
      "step": 354
    },
    {
      "epoch": 0.05949886868348278,
      "grad_norm": 0.48792615431078384,
      "learning_rate": 1.9592144738464563e-07,
      "loss": 1.3463,
      "step": 355
    },
    {
      "epoch": 0.0596664711304785,
      "grad_norm": 0.4216737971546497,
      "learning_rate": 1.9589070451281085e-07,
      "loss": 1.2416,
      "step": 356
    },
    {
      "epoch": 0.05983407357747423,
      "grad_norm": 0.4250318358387,
      "learning_rate": 1.9585984864078994e-07,
      "loss": 1.2292,
      "step": 357
    },
    {
      "epoch": 0.06000167602446996,
      "grad_norm": 0.49051779355936637,
      "learning_rate": 1.958288798049444e-07,
      "loss": 1.2785,
      "step": 358
    },
    {
      "epoch": 0.060169278471465686,
      "grad_norm": 0.48165105289660903,
      "learning_rate": 1.9579779804176868e-07,
      "loss": 1.2945,
      "step": 359
    },
    {
      "epoch": 0.06033688091846141,
      "grad_norm": 0.4319659633512919,
      "learning_rate": 1.9576660338789037e-07,
      "loss": 1.2923,
      "step": 360
    },
    {
      "epoch": 0.060504483365457135,
      "grad_norm": 0.44136640136346755,
      "learning_rate": 1.957352958800701e-07,
      "loss": 1.217,
      "step": 361
    },
    {
      "epoch": 0.06067208581245286,
      "grad_norm": 0.42614925051534003,
      "learning_rate": 1.957038755552015e-07,
      "loss": 1.2255,
      "step": 362
    },
    {
      "epoch": 0.06083968825944859,
      "grad_norm": 0.4951513309250324,
      "learning_rate": 1.9567234245031107e-07,
      "loss": 1.2986,
      "step": 363
    },
    {
      "epoch": 0.06100729070644431,
      "grad_norm": 0.46670408329309243,
      "learning_rate": 1.956406966025583e-07,
      "loss": 1.2503,
      "step": 364
    },
    {
      "epoch": 0.06117489315344004,
      "grad_norm": 0.46302924210189317,
      "learning_rate": 1.9560893804923554e-07,
      "loss": 1.3171,
      "step": 365
    },
    {
      "epoch": 0.06134249560043577,
      "grad_norm": 0.5288099435345985,
      "learning_rate": 1.9557706682776788e-07,
      "loss": 1.3576,
      "step": 366
    },
    {
      "epoch": 0.061510098047431495,
      "grad_norm": 0.5222410714914495,
      "learning_rate": 1.9554508297571327e-07,
      "loss": 1.2576,
      "step": 367
    },
    {
      "epoch": 0.061677700494427216,
      "grad_norm": 0.5300660685796754,
      "learning_rate": 1.9551298653076227e-07,
      "loss": 1.3726,
      "step": 368
    },
    {
      "epoch": 0.061845302941422944,
      "grad_norm": 0.48947426537716215,
      "learning_rate": 1.9548077753073826e-07,
      "loss": 1.3294,
      "step": 369
    },
    {
      "epoch": 0.06201290538841867,
      "grad_norm": 0.4427604920705336,
      "learning_rate": 1.9544845601359716e-07,
      "loss": 1.2329,
      "step": 370
    },
    {
      "epoch": 0.0621805078354144,
      "grad_norm": 0.4718292151233188,
      "learning_rate": 1.9541602201742753e-07,
      "loss": 1.2655,
      "step": 371
    },
    {
      "epoch": 0.06234811028241012,
      "grad_norm": 0.4836308351100042,
      "learning_rate": 1.9538347558045048e-07,
      "loss": 1.2967,
      "step": 372
    },
    {
      "epoch": 0.06251571272940586,
      "grad_norm": 0.48029771474412397,
      "learning_rate": 1.9535081674101953e-07,
      "loss": 1.3268,
      "step": 373
    },
    {
      "epoch": 0.06268331517640158,
      "grad_norm": 0.4050352220355886,
      "learning_rate": 1.9531804553762084e-07,
      "loss": 1.2194,
      "step": 374
    },
    {
      "epoch": 0.0628509176233973,
      "grad_norm": 0.4230701740998826,
      "learning_rate": 1.952851620088728e-07,
      "loss": 1.2315,
      "step": 375
    },
    {
      "epoch": 0.06301852007039303,
      "grad_norm": 0.48051926035712805,
      "learning_rate": 1.952521661935263e-07,
      "loss": 1.3059,
      "step": 376
    },
    {
      "epoch": 0.06318612251738875,
      "grad_norm": 0.4242414339971303,
      "learning_rate": 1.9521905813046443e-07,
      "loss": 1.2145,
      "step": 377
    },
    {
      "epoch": 0.06335372496438448,
      "grad_norm": 0.46317601083184995,
      "learning_rate": 1.9518583785870268e-07,
      "loss": 1.2741,
      "step": 378
    },
    {
      "epoch": 0.06352132741138021,
      "grad_norm": 0.5152968327433587,
      "learning_rate": 1.951525054173887e-07,
      "loss": 1.2936,
      "step": 379
    },
    {
      "epoch": 0.06368892985837593,
      "grad_norm": 0.4492790122750447,
      "learning_rate": 1.951190608458023e-07,
      "loss": 1.2478,
      "step": 380
    },
    {
      "epoch": 0.06385653230537165,
      "grad_norm": 0.40430316422618895,
      "learning_rate": 1.9508550418335553e-07,
      "loss": 1.1427,
      "step": 381
    },
    {
      "epoch": 0.06402413475236739,
      "grad_norm": 0.4578227893248346,
      "learning_rate": 1.9505183546959242e-07,
      "loss": 1.2981,
      "step": 382
    },
    {
      "epoch": 0.06419173719936311,
      "grad_norm": 0.45656321503237546,
      "learning_rate": 1.9501805474418908e-07,
      "loss": 1.2365,
      "step": 383
    },
    {
      "epoch": 0.06435933964635884,
      "grad_norm": 0.39662011496470706,
      "learning_rate": 1.9498416204695366e-07,
      "loss": 1.1676,
      "step": 384
    },
    {
      "epoch": 0.06452694209335456,
      "grad_norm": 0.513161741853131,
      "learning_rate": 1.9495015741782618e-07,
      "loss": 1.3122,
      "step": 385
    },
    {
      "epoch": 0.06469454454035029,
      "grad_norm": 0.4053277021737868,
      "learning_rate": 1.9491604089687865e-07,
      "loss": 1.2241,
      "step": 386
    },
    {
      "epoch": 0.06486214698734602,
      "grad_norm": 0.4657523580597361,
      "learning_rate": 1.948818125243149e-07,
      "loss": 1.2594,
      "step": 387
    },
    {
      "epoch": 0.06502974943434174,
      "grad_norm": 0.4747394385543636,
      "learning_rate": 1.948474723404705e-07,
      "loss": 1.3051,
      "step": 388
    },
    {
      "epoch": 0.06519735188133746,
      "grad_norm": 0.4660694978297882,
      "learning_rate": 1.9481302038581292e-07,
      "loss": 1.2722,
      "step": 389
    },
    {
      "epoch": 0.0653649543283332,
      "grad_norm": 0.4745216414509223,
      "learning_rate": 1.9477845670094127e-07,
      "loss": 1.2667,
      "step": 390
    },
    {
      "epoch": 0.06553255677532892,
      "grad_norm": 0.41606954307180843,
      "learning_rate": 1.9474378132658624e-07,
      "loss": 1.1965,
      "step": 391
    },
    {
      "epoch": 0.06570015922232465,
      "grad_norm": 0.43490109116940256,
      "learning_rate": 1.9470899430361034e-07,
      "loss": 1.2493,
      "step": 392
    },
    {
      "epoch": 0.06586776166932037,
      "grad_norm": 0.45772935995465003,
      "learning_rate": 1.9467409567300742e-07,
      "loss": 1.3065,
      "step": 393
    },
    {
      "epoch": 0.0660353641163161,
      "grad_norm": 0.45247792095803124,
      "learning_rate": 1.946390854759031e-07,
      "loss": 1.254,
      "step": 394
    },
    {
      "epoch": 0.06620296656331183,
      "grad_norm": 0.3906829854858544,
      "learning_rate": 1.946039637535542e-07,
      "loss": 1.2144,
      "step": 395
    },
    {
      "epoch": 0.06637056901030755,
      "grad_norm": 0.4900351608132551,
      "learning_rate": 1.9456873054734917e-07,
      "loss": 1.3086,
      "step": 396
    },
    {
      "epoch": 0.06653817145730327,
      "grad_norm": 0.4114673311806446,
      "learning_rate": 1.945333858988078e-07,
      "loss": 1.2358,
      "step": 397
    },
    {
      "epoch": 0.066705773904299,
      "grad_norm": 0.45825341151060744,
      "learning_rate": 1.9449792984958112e-07,
      "loss": 1.2396,
      "step": 398
    },
    {
      "epoch": 0.06687337635129473,
      "grad_norm": 0.38659583883056525,
      "learning_rate": 1.944623624414515e-07,
      "loss": 1.1778,
      "step": 399
    },
    {
      "epoch": 0.06704097879829045,
      "grad_norm": 0.4597123672618489,
      "learning_rate": 1.9442668371633252e-07,
      "loss": 1.3107,
      "step": 400
    },
    {
      "epoch": 0.06720858124528618,
      "grad_norm": 0.48905660613747143,
      "learning_rate": 1.94390893716269e-07,
      "loss": 1.3405,
      "step": 401
    },
    {
      "epoch": 0.0673761836922819,
      "grad_norm": 0.4932140497476784,
      "learning_rate": 1.943549924834368e-07,
      "loss": 1.334,
      "step": 402
    },
    {
      "epoch": 0.06754378613927764,
      "grad_norm": 0.4499669021306464,
      "learning_rate": 1.943189800601429e-07,
      "loss": 1.2961,
      "step": 403
    },
    {
      "epoch": 0.06771138858627336,
      "grad_norm": 0.40373124068176314,
      "learning_rate": 1.942828564888253e-07,
      "loss": 1.165,
      "step": 404
    },
    {
      "epoch": 0.06787899103326908,
      "grad_norm": 0.4055378096488636,
      "learning_rate": 1.9424662181205303e-07,
      "loss": 1.2359,
      "step": 405
    },
    {
      "epoch": 0.06804659348026482,
      "grad_norm": 0.4268974892452029,
      "learning_rate": 1.94210276072526e-07,
      "loss": 1.26,
      "step": 406
    },
    {
      "epoch": 0.06821419592726054,
      "grad_norm": 0.45392140645308193,
      "learning_rate": 1.9417381931307495e-07,
      "loss": 1.2123,
      "step": 407
    },
    {
      "epoch": 0.06838179837425626,
      "grad_norm": 0.465941560058682,
      "learning_rate": 1.9413725157666156e-07,
      "loss": 1.2543,
      "step": 408
    },
    {
      "epoch": 0.068549400821252,
      "grad_norm": 0.45493404219062966,
      "learning_rate": 1.9410057290637823e-07,
      "loss": 1.2418,
      "step": 409
    },
    {
      "epoch": 0.06871700326824771,
      "grad_norm": 0.4033171808082436,
      "learning_rate": 1.9406378334544806e-07,
      "loss": 1.2673,
      "step": 410
    },
    {
      "epoch": 0.06888460571524345,
      "grad_norm": 0.4800202835811931,
      "learning_rate": 1.940268829372249e-07,
      "loss": 1.3187,
      "step": 411
    },
    {
      "epoch": 0.06905220816223917,
      "grad_norm": 0.41446478422174965,
      "learning_rate": 1.9398987172519315e-07,
      "loss": 1.2314,
      "step": 412
    },
    {
      "epoch": 0.06921981060923489,
      "grad_norm": 0.42990293416561837,
      "learning_rate": 1.9395274975296783e-07,
      "loss": 1.2342,
      "step": 413
    },
    {
      "epoch": 0.06938741305623063,
      "grad_norm": 0.4875700335823143,
      "learning_rate": 1.939155170642945e-07,
      "loss": 1.2936,
      "step": 414
    },
    {
      "epoch": 0.06955501550322635,
      "grad_norm": 0.45109164258275525,
      "learning_rate": 1.9387817370304908e-07,
      "loss": 1.2619,
      "step": 415
    },
    {
      "epoch": 0.06972261795022207,
      "grad_norm": 0.36292761520797234,
      "learning_rate": 1.938407197132381e-07,
      "loss": 1.1132,
      "step": 416
    },
    {
      "epoch": 0.0698902203972178,
      "grad_norm": 0.44669102666910865,
      "learning_rate": 1.9380315513899825e-07,
      "loss": 1.2581,
      "step": 417
    },
    {
      "epoch": 0.07005782284421352,
      "grad_norm": 0.49041801078557656,
      "learning_rate": 1.9376548002459673e-07,
      "loss": 1.2901,
      "step": 418
    },
    {
      "epoch": 0.07022542529120925,
      "grad_norm": 0.4198463673607918,
      "learning_rate": 1.9372769441443083e-07,
      "loss": 1.1975,
      "step": 419
    },
    {
      "epoch": 0.07039302773820498,
      "grad_norm": 0.44921565976807887,
      "learning_rate": 1.9368979835302816e-07,
      "loss": 1.2753,
      "step": 420
    },
    {
      "epoch": 0.0705606301852007,
      "grad_norm": 0.41176105457662016,
      "learning_rate": 1.9365179188504646e-07,
      "loss": 1.1621,
      "step": 421
    },
    {
      "epoch": 0.07072823263219644,
      "grad_norm": 0.4276727039014116,
      "learning_rate": 1.9361367505527357e-07,
      "loss": 1.2291,
      "step": 422
    },
    {
      "epoch": 0.07089583507919216,
      "grad_norm": 0.4332290842521335,
      "learning_rate": 1.9357544790862737e-07,
      "loss": 1.2476,
      "step": 423
    },
    {
      "epoch": 0.07106343752618788,
      "grad_norm": 0.44715040310422377,
      "learning_rate": 1.935371104901558e-07,
      "loss": 1.2623,
      "step": 424
    },
    {
      "epoch": 0.07123103997318361,
      "grad_norm": 0.4371947799269905,
      "learning_rate": 1.934986628450367e-07,
      "loss": 1.2265,
      "step": 425
    },
    {
      "epoch": 0.07139864242017933,
      "grad_norm": 0.4080116565893799,
      "learning_rate": 1.9346010501857776e-07,
      "loss": 1.2265,
      "step": 426
    },
    {
      "epoch": 0.07156624486717506,
      "grad_norm": 0.4641903249759015,
      "learning_rate": 1.9342143705621662e-07,
      "loss": 1.2481,
      "step": 427
    },
    {
      "epoch": 0.07173384731417079,
      "grad_norm": 0.4396494879634481,
      "learning_rate": 1.9338265900352057e-07,
      "loss": 1.2503,
      "step": 428
    },
    {
      "epoch": 0.07190144976116651,
      "grad_norm": 0.39795976096079816,
      "learning_rate": 1.9334377090618682e-07,
      "loss": 1.1936,
      "step": 429
    },
    {
      "epoch": 0.07206905220816225,
      "grad_norm": 0.4715077089926066,
      "learning_rate": 1.9330477281004206e-07,
      "loss": 1.2818,
      "step": 430
    },
    {
      "epoch": 0.07223665465515797,
      "grad_norm": 0.3471493434201732,
      "learning_rate": 1.9326566476104272e-07,
      "loss": 1.1642,
      "step": 431
    },
    {
      "epoch": 0.07240425710215369,
      "grad_norm": 0.4693192415809532,
      "learning_rate": 1.9322644680527477e-07,
      "loss": 1.252,
      "step": 432
    },
    {
      "epoch": 0.07257185954914942,
      "grad_norm": 0.42137897521823237,
      "learning_rate": 1.9318711898895374e-07,
      "loss": 1.2338,
      "step": 433
    },
    {
      "epoch": 0.07273946199614514,
      "grad_norm": 0.44875738802022525,
      "learning_rate": 1.9314768135842457e-07,
      "loss": 1.2511,
      "step": 434
    },
    {
      "epoch": 0.07290706444314086,
      "grad_norm": 0.4323551290555392,
      "learning_rate": 1.931081339601616e-07,
      "loss": 1.2756,
      "step": 435
    },
    {
      "epoch": 0.0730746668901366,
      "grad_norm": 0.4594121071720164,
      "learning_rate": 1.9306847684076858e-07,
      "loss": 1.2515,
      "step": 436
    },
    {
      "epoch": 0.07324226933713232,
      "grad_norm": 0.45978695599297287,
      "learning_rate": 1.930287100469785e-07,
      "loss": 1.2974,
      "step": 437
    },
    {
      "epoch": 0.07340987178412804,
      "grad_norm": 0.35205829231565944,
      "learning_rate": 1.9298883362565365e-07,
      "loss": 1.0749,
      "step": 438
    },
    {
      "epoch": 0.07357747423112378,
      "grad_norm": 0.4285764153387767,
      "learning_rate": 1.9294884762378544e-07,
      "loss": 1.1974,
      "step": 439
    },
    {
      "epoch": 0.0737450766781195,
      "grad_norm": 0.4429799563493179,
      "learning_rate": 1.9290875208849448e-07,
      "loss": 1.2476,
      "step": 440
    },
    {
      "epoch": 0.07391267912511523,
      "grad_norm": 0.3909597272930478,
      "learning_rate": 1.9286854706703043e-07,
      "loss": 1.2037,
      "step": 441
    },
    {
      "epoch": 0.07408028157211095,
      "grad_norm": 0.4123946948060661,
      "learning_rate": 1.9282823260677193e-07,
      "loss": 1.2094,
      "step": 442
    },
    {
      "epoch": 0.07424788401910667,
      "grad_norm": 0.4204652071037692,
      "learning_rate": 1.9278780875522666e-07,
      "loss": 1.1917,
      "step": 443
    },
    {
      "epoch": 0.07441548646610241,
      "grad_norm": 0.3953362171373954,
      "learning_rate": 1.9274727556003118e-07,
      "loss": 1.2003,
      "step": 444
    },
    {
      "epoch": 0.07458308891309813,
      "grad_norm": 0.3703497883346843,
      "learning_rate": 1.9270663306895086e-07,
      "loss": 1.0909,
      "step": 445
    },
    {
      "epoch": 0.07475069136009385,
      "grad_norm": 0.5367266102818821,
      "learning_rate": 1.9266588132987995e-07,
      "loss": 1.2875,
      "step": 446
    },
    {
      "epoch": 0.07491829380708959,
      "grad_norm": 0.4507144123017091,
      "learning_rate": 1.9262502039084137e-07,
      "loss": 1.2143,
      "step": 447
    },
    {
      "epoch": 0.07508589625408531,
      "grad_norm": 0.4241013188962693,
      "learning_rate": 1.9258405029998677e-07,
      "loss": 1.1897,
      "step": 448
    },
    {
      "epoch": 0.07525349870108104,
      "grad_norm": 0.3766335504314604,
      "learning_rate": 1.9254297110559637e-07,
      "loss": 1.1824,
      "step": 449
    },
    {
      "epoch": 0.07542110114807676,
      "grad_norm": 0.45813551079805265,
      "learning_rate": 1.9250178285607904e-07,
      "loss": 1.2788,
      "step": 450
    },
    {
      "epoch": 0.07558870359507248,
      "grad_norm": 0.41061843750261245,
      "learning_rate": 1.9246048559997207e-07,
      "loss": 1.2807,
      "step": 451
    },
    {
      "epoch": 0.07575630604206822,
      "grad_norm": 0.40738495156263205,
      "learning_rate": 1.9241907938594132e-07,
      "loss": 1.1899,
      "step": 452
    },
    {
      "epoch": 0.07592390848906394,
      "grad_norm": 0.42045972160973927,
      "learning_rate": 1.9237756426278094e-07,
      "loss": 1.2593,
      "step": 453
    },
    {
      "epoch": 0.07609151093605966,
      "grad_norm": 0.47907595217983207,
      "learning_rate": 1.923359402794135e-07,
      "loss": 1.2785,
      "step": 454
    },
    {
      "epoch": 0.0762591133830554,
      "grad_norm": 0.4357811423570243,
      "learning_rate": 1.9229420748488975e-07,
      "loss": 1.2172,
      "step": 455
    },
    {
      "epoch": 0.07642671583005112,
      "grad_norm": 0.4491201585113102,
      "learning_rate": 1.9225236592838884e-07,
      "loss": 1.2362,
      "step": 456
    },
    {
      "epoch": 0.07659431827704684,
      "grad_norm": 0.37242496642281636,
      "learning_rate": 1.9221041565921794e-07,
      "loss": 1.1658,
      "step": 457
    },
    {
      "epoch": 0.07676192072404257,
      "grad_norm": 0.39917204774619813,
      "learning_rate": 1.9216835672681237e-07,
      "loss": 1.1755,
      "step": 458
    },
    {
      "epoch": 0.0769295231710383,
      "grad_norm": 0.4197297952786648,
      "learning_rate": 1.9212618918073548e-07,
      "loss": 1.2536,
      "step": 459
    },
    {
      "epoch": 0.07709712561803403,
      "grad_norm": 0.3927194329570493,
      "learning_rate": 1.9208391307067872e-07,
      "loss": 1.1755,
      "step": 460
    },
    {
      "epoch": 0.07726472806502975,
      "grad_norm": 0.44258078943909857,
      "learning_rate": 1.9204152844646133e-07,
      "loss": 1.3215,
      "step": 461
    },
    {
      "epoch": 0.07743233051202547,
      "grad_norm": 0.4513430726655771,
      "learning_rate": 1.9199903535803052e-07,
      "loss": 1.2884,
      "step": 462
    },
    {
      "epoch": 0.0775999329590212,
      "grad_norm": 0.418875714784915,
      "learning_rate": 1.9195643385546125e-07,
      "loss": 1.2689,
      "step": 463
    },
    {
      "epoch": 0.07776753540601693,
      "grad_norm": 0.36232580190199826,
      "learning_rate": 1.9191372398895634e-07,
      "loss": 1.2024,
      "step": 464
    },
    {
      "epoch": 0.07793513785301265,
      "grad_norm": 0.4335371772052694,
      "learning_rate": 1.918709058088462e-07,
      "loss": 1.2196,
      "step": 465
    },
    {
      "epoch": 0.07810274030000838,
      "grad_norm": 0.45303148593194803,
      "learning_rate": 1.9182797936558894e-07,
      "loss": 1.2369,
      "step": 466
    },
    {
      "epoch": 0.0782703427470041,
      "grad_norm": 0.44633550283244783,
      "learning_rate": 1.9178494470977023e-07,
      "loss": 1.2368,
      "step": 467
    },
    {
      "epoch": 0.07843794519399984,
      "grad_norm": 0.4235583922961674,
      "learning_rate": 1.9174180189210324e-07,
      "loss": 1.1955,
      "step": 468
    },
    {
      "epoch": 0.07860554764099556,
      "grad_norm": 0.4382232284364197,
      "learning_rate": 1.9169855096342868e-07,
      "loss": 1.2289,
      "step": 469
    },
    {
      "epoch": 0.07877315008799128,
      "grad_norm": 0.4476446525298324,
      "learning_rate": 1.916551919747146e-07,
      "loss": 1.2819,
      "step": 470
    },
    {
      "epoch": 0.07894075253498702,
      "grad_norm": 0.4739651426643699,
      "learning_rate": 1.9161172497705636e-07,
      "loss": 1.2858,
      "step": 471
    },
    {
      "epoch": 0.07910835498198274,
      "grad_norm": 0.451291884514936,
      "learning_rate": 1.9156815002167666e-07,
      "loss": 1.2586,
      "step": 472
    },
    {
      "epoch": 0.07927595742897846,
      "grad_norm": 0.4652863763232165,
      "learning_rate": 1.915244671599254e-07,
      "loss": 1.2933,
      "step": 473
    },
    {
      "epoch": 0.07944355987597419,
      "grad_norm": 0.4124325236538223,
      "learning_rate": 1.9148067644327967e-07,
      "loss": 1.1873,
      "step": 474
    },
    {
      "epoch": 0.07961116232296991,
      "grad_norm": 0.4065132991232885,
      "learning_rate": 1.9143677792334358e-07,
      "loss": 1.2227,
      "step": 475
    },
    {
      "epoch": 0.07977876476996563,
      "grad_norm": 0.44809545494932285,
      "learning_rate": 1.9139277165184833e-07,
      "loss": 1.2592,
      "step": 476
    },
    {
      "epoch": 0.07994636721696137,
      "grad_norm": 0.4555505806779661,
      "learning_rate": 1.9134865768065215e-07,
      "loss": 1.3075,
      "step": 477
    },
    {
      "epoch": 0.08011396966395709,
      "grad_norm": 0.3931042217936992,
      "learning_rate": 1.9130443606174004e-07,
      "loss": 1.164,
      "step": 478
    },
    {
      "epoch": 0.08028157211095283,
      "grad_norm": 0.4278175439641815,
      "learning_rate": 1.9126010684722405e-07,
      "loss": 1.2081,
      "step": 479
    },
    {
      "epoch": 0.08044917455794855,
      "grad_norm": 0.40683943778366183,
      "learning_rate": 1.9121567008934284e-07,
      "loss": 1.1944,
      "step": 480
    },
    {
      "epoch": 0.08061677700494427,
      "grad_norm": 0.4502212827389735,
      "learning_rate": 1.9117112584046192e-07,
      "loss": 1.298,
      "step": 481
    },
    {
      "epoch": 0.08078437945194,
      "grad_norm": 0.41773117664198833,
      "learning_rate": 1.911264741530734e-07,
      "loss": 1.2418,
      "step": 482
    },
    {
      "epoch": 0.08095198189893572,
      "grad_norm": 0.4401583837038186,
      "learning_rate": 1.9108171507979605e-07,
      "loss": 1.2456,
      "step": 483
    },
    {
      "epoch": 0.08111958434593144,
      "grad_norm": 0.4729258758030939,
      "learning_rate": 1.9103684867337518e-07,
      "loss": 1.2372,
      "step": 484
    },
    {
      "epoch": 0.08128718679292718,
      "grad_norm": 0.4175522117091413,
      "learning_rate": 1.9099187498668254e-07,
      "loss": 1.2435,
      "step": 485
    },
    {
      "epoch": 0.0814547892399229,
      "grad_norm": 0.42759629640830005,
      "learning_rate": 1.9094679407271634e-07,
      "loss": 1.2236,
      "step": 486
    },
    {
      "epoch": 0.08162239168691864,
      "grad_norm": 0.42149443158094313,
      "learning_rate": 1.9090160598460118e-07,
      "loss": 1.2139,
      "step": 487
    },
    {
      "epoch": 0.08178999413391436,
      "grad_norm": 0.42668887783385434,
      "learning_rate": 1.9085631077558787e-07,
      "loss": 1.2304,
      "step": 488
    },
    {
      "epoch": 0.08195759658091008,
      "grad_norm": 0.3439892485977779,
      "learning_rate": 1.9081090849905353e-07,
      "loss": 1.1615,
      "step": 489
    },
    {
      "epoch": 0.08212519902790581,
      "grad_norm": 0.4723202908709293,
      "learning_rate": 1.907653992085014e-07,
      "loss": 1.2894,
      "step": 490
    },
    {
      "epoch": 0.08229280147490153,
      "grad_norm": 0.4415025532030837,
      "learning_rate": 1.9071978295756085e-07,
      "loss": 1.2392,
      "step": 491
    },
    {
      "epoch": 0.08246040392189725,
      "grad_norm": 0.41711312912817816,
      "learning_rate": 1.9067405979998737e-07,
      "loss": 1.2697,
      "step": 492
    },
    {
      "epoch": 0.08262800636889299,
      "grad_norm": 0.42372076738009906,
      "learning_rate": 1.9062822978966226e-07,
      "loss": 1.2206,
      "step": 493
    },
    {
      "epoch": 0.08279560881588871,
      "grad_norm": 0.40965707533544526,
      "learning_rate": 1.905822929805929e-07,
      "loss": 1.1819,
      "step": 494
    },
    {
      "epoch": 0.08296321126288443,
      "grad_norm": 0.4198341286022988,
      "learning_rate": 1.9053624942691245e-07,
      "loss": 1.2415,
      "step": 495
    },
    {
      "epoch": 0.08313081370988017,
      "grad_norm": 0.4124292619676821,
      "learning_rate": 1.9049009918287984e-07,
      "loss": 1.246,
      "step": 496
    },
    {
      "epoch": 0.08329841615687589,
      "grad_norm": 0.4030444224397378,
      "learning_rate": 1.904438423028798e-07,
      "loss": 1.2296,
      "step": 497
    },
    {
      "epoch": 0.08346601860387162,
      "grad_norm": 0.4076435184355045,
      "learning_rate": 1.9039747884142265e-07,
      "loss": 1.1762,
      "step": 498
    },
    {
      "epoch": 0.08363362105086734,
      "grad_norm": 0.4506291613286842,
      "learning_rate": 1.9035100885314438e-07,
      "loss": 1.2443,
      "step": 499
    },
    {
      "epoch": 0.08380122349786306,
      "grad_norm": 0.40705256113880245,
      "learning_rate": 1.9030443239280643e-07,
      "loss": 1.1467,
      "step": 500
    },
    {
      "epoch": 0.0839688259448588,
      "grad_norm": 0.35479524605132895,
      "learning_rate": 1.9025774951529577e-07,
      "loss": 1.1042,
      "step": 501
    },
    {
      "epoch": 0.08413642839185452,
      "grad_norm": 0.3561197588321081,
      "learning_rate": 1.902109602756248e-07,
      "loss": 1.1406,
      "step": 502
    },
    {
      "epoch": 0.08430403083885024,
      "grad_norm": 0.4085176177632358,
      "learning_rate": 1.9016406472893117e-07,
      "loss": 1.2457,
      "step": 503
    },
    {
      "epoch": 0.08447163328584598,
      "grad_norm": 0.4416531721090171,
      "learning_rate": 1.9011706293047784e-07,
      "loss": 1.221,
      "step": 504
    },
    {
      "epoch": 0.0846392357328417,
      "grad_norm": 0.44628753648379327,
      "learning_rate": 1.9006995493565303e-07,
      "loss": 1.244,
      "step": 505
    },
    {
      "epoch": 0.08480683817983743,
      "grad_norm": 0.4547027398570248,
      "learning_rate": 1.9002274079997007e-07,
      "loss": 1.2309,
      "step": 506
    },
    {
      "epoch": 0.08497444062683315,
      "grad_norm": 0.4622995022218325,
      "learning_rate": 1.8997542057906737e-07,
      "loss": 1.2739,
      "step": 507
    },
    {
      "epoch": 0.08514204307382887,
      "grad_norm": 0.4845603952827642,
      "learning_rate": 1.8992799432870832e-07,
      "loss": 1.2832,
      "step": 508
    },
    {
      "epoch": 0.08530964552082461,
      "grad_norm": 0.3721641479502781,
      "learning_rate": 1.898804621047813e-07,
      "loss": 1.1732,
      "step": 509
    },
    {
      "epoch": 0.08547724796782033,
      "grad_norm": 0.4297071965136174,
      "learning_rate": 1.8983282396329957e-07,
      "loss": 1.2582,
      "step": 510
    },
    {
      "epoch": 0.08564485041481605,
      "grad_norm": 0.40929517137987614,
      "learning_rate": 1.8978507996040124e-07,
      "loss": 1.2065,
      "step": 511
    },
    {
      "epoch": 0.08581245286181179,
      "grad_norm": 0.40015249050529283,
      "learning_rate": 1.8973723015234903e-07,
      "loss": 1.2257,
      "step": 512
    },
    {
      "epoch": 0.08598005530880751,
      "grad_norm": 0.3935883605196306,
      "learning_rate": 1.8968927459553053e-07,
      "loss": 1.1807,
      "step": 513
    },
    {
      "epoch": 0.08614765775580323,
      "grad_norm": 0.4290823679588516,
      "learning_rate": 1.8964121334645779e-07,
      "loss": 1.2224,
      "step": 514
    },
    {
      "epoch": 0.08631526020279896,
      "grad_norm": 0.45705980628810483,
      "learning_rate": 1.8959304646176753e-07,
      "loss": 1.2694,
      "step": 515
    },
    {
      "epoch": 0.08648286264979468,
      "grad_norm": 0.3608042056088367,
      "learning_rate": 1.8954477399822088e-07,
      "loss": 1.1393,
      "step": 516
    },
    {
      "epoch": 0.08665046509679042,
      "grad_norm": 0.4291028765173897,
      "learning_rate": 1.8949639601270343e-07,
      "loss": 1.2175,
      "step": 517
    },
    {
      "epoch": 0.08681806754378614,
      "grad_norm": 0.3942099496100389,
      "learning_rate": 1.8944791256222507e-07,
      "loss": 1.2189,
      "step": 518
    },
    {
      "epoch": 0.08698566999078186,
      "grad_norm": 0.40958511456413565,
      "learning_rate": 1.8939932370392003e-07,
      "loss": 1.2624,
      "step": 519
    },
    {
      "epoch": 0.0871532724377776,
      "grad_norm": 0.3744967979858015,
      "learning_rate": 1.893506294950467e-07,
      "loss": 1.1361,
      "step": 520
    },
    {
      "epoch": 0.08732087488477332,
      "grad_norm": 0.4104547755323647,
      "learning_rate": 1.8930182999298768e-07,
      "loss": 1.2062,
      "step": 521
    },
    {
      "epoch": 0.08748847733176904,
      "grad_norm": 0.39292873140828577,
      "learning_rate": 1.892529252552496e-07,
      "loss": 1.1744,
      "step": 522
    },
    {
      "epoch": 0.08765607977876477,
      "grad_norm": 0.4020048457073119,
      "learning_rate": 1.8920391533946308e-07,
      "loss": 1.1918,
      "step": 523
    },
    {
      "epoch": 0.0878236822257605,
      "grad_norm": 0.4124864533605224,
      "learning_rate": 1.891548003033828e-07,
      "loss": 1.2032,
      "step": 524
    },
    {
      "epoch": 0.08799128467275623,
      "grad_norm": 0.33615004371787965,
      "learning_rate": 1.891055802048872e-07,
      "loss": 1.0953,
      "step": 525
    },
    {
      "epoch": 0.08815888711975195,
      "grad_norm": 0.4182319281925786,
      "learning_rate": 1.890562551019785e-07,
      "loss": 1.2346,
      "step": 526
    },
    {
      "epoch": 0.08832648956674767,
      "grad_norm": 0.43759121209956875,
      "learning_rate": 1.8900682505278285e-07,
      "loss": 1.2515,
      "step": 527
    },
    {
      "epoch": 0.0884940920137434,
      "grad_norm": 0.4131350222619009,
      "learning_rate": 1.889572901155499e-07,
      "loss": 1.1541,
      "step": 528
    },
    {
      "epoch": 0.08866169446073913,
      "grad_norm": 0.402143165544033,
      "learning_rate": 1.8890765034865294e-07,
      "loss": 1.164,
      "step": 529
    },
    {
      "epoch": 0.08882929690773485,
      "grad_norm": 0.37878304468308355,
      "learning_rate": 1.888579058105888e-07,
      "loss": 1.1565,
      "step": 530
    },
    {
      "epoch": 0.08899689935473058,
      "grad_norm": 0.4420309232079947,
      "learning_rate": 1.8880805655997782e-07,
      "loss": 1.2046,
      "step": 531
    },
    {
      "epoch": 0.0891645018017263,
      "grad_norm": 0.3546665400109044,
      "learning_rate": 1.8875810265556367e-07,
      "loss": 1.1299,
      "step": 532
    },
    {
      "epoch": 0.08933210424872202,
      "grad_norm": 0.44571792131904453,
      "learning_rate": 1.8870804415621336e-07,
      "loss": 1.258,
      "step": 533
    },
    {
      "epoch": 0.08949970669571776,
      "grad_norm": 0.3967516759818737,
      "learning_rate": 1.8865788112091722e-07,
      "loss": 1.1986,
      "step": 534
    },
    {
      "epoch": 0.08966730914271348,
      "grad_norm": 0.42438859287636505,
      "learning_rate": 1.8860761360878867e-07,
      "loss": 1.2187,
      "step": 535
    },
    {
      "epoch": 0.08983491158970922,
      "grad_norm": 0.426340235745058,
      "learning_rate": 1.8855724167906434e-07,
      "loss": 1.251,
      "step": 536
    },
    {
      "epoch": 0.09000251403670494,
      "grad_norm": 0.41469522136728326,
      "learning_rate": 1.8850676539110383e-07,
      "loss": 1.2128,
      "step": 537
    },
    {
      "epoch": 0.09017011648370066,
      "grad_norm": 0.42224793366174695,
      "learning_rate": 1.884561848043898e-07,
      "loss": 1.2261,
      "step": 538
    },
    {
      "epoch": 0.09033771893069639,
      "grad_norm": 0.4024973714984321,
      "learning_rate": 1.8840549997852775e-07,
      "loss": 1.202,
      "step": 539
    },
    {
      "epoch": 0.09050532137769211,
      "grad_norm": 0.44089732309886204,
      "learning_rate": 1.8835471097324602e-07,
      "loss": 1.2415,
      "step": 540
    },
    {
      "epoch": 0.09067292382468783,
      "grad_norm": 0.4333708775427699,
      "learning_rate": 1.8830381784839578e-07,
      "loss": 1.2519,
      "step": 541
    },
    {
      "epoch": 0.09084052627168357,
      "grad_norm": 0.43526575546162943,
      "learning_rate": 1.8825282066395082e-07,
      "loss": 1.1917,
      "step": 542
    },
    {
      "epoch": 0.09100812871867929,
      "grad_norm": 0.43370903756278184,
      "learning_rate": 1.8820171948000764e-07,
      "loss": 1.2657,
      "step": 543
    },
    {
      "epoch": 0.09117573116567503,
      "grad_norm": 0.4142853369805367,
      "learning_rate": 1.881505143567852e-07,
      "loss": 1.1579,
      "step": 544
    },
    {
      "epoch": 0.09134333361267075,
      "grad_norm": 0.4024364961990316,
      "learning_rate": 1.88099205354625e-07,
      "loss": 1.1483,
      "step": 545
    },
    {
      "epoch": 0.09151093605966647,
      "grad_norm": 0.3943184912751912,
      "learning_rate": 1.88047792533991e-07,
      "loss": 1.1425,
      "step": 546
    },
    {
      "epoch": 0.0916785385066622,
      "grad_norm": 0.40369366671753176,
      "learning_rate": 1.8799627595546942e-07,
      "loss": 1.1844,
      "step": 547
    },
    {
      "epoch": 0.09184614095365792,
      "grad_norm": 0.3750009517651847,
      "learning_rate": 1.8794465567976877e-07,
      "loss": 1.1445,
      "step": 548
    },
    {
      "epoch": 0.09201374340065364,
      "grad_norm": 0.4244471199429254,
      "learning_rate": 1.8789293176771976e-07,
      "loss": 1.1927,
      "step": 549
    },
    {
      "epoch": 0.09218134584764938,
      "grad_norm": 0.4953010770094799,
      "learning_rate": 1.878411042802753e-07,
      "loss": 1.2903,
      "step": 550
    },
    {
      "epoch": 0.0923489482946451,
      "grad_norm": 0.435841844210947,
      "learning_rate": 1.8778917327851024e-07,
      "loss": 1.2107,
      "step": 551
    },
    {
      "epoch": 0.09251655074164082,
      "grad_norm": 0.4503060452171762,
      "learning_rate": 1.8773713882362147e-07,
      "loss": 1.2551,
      "step": 552
    },
    {
      "epoch": 0.09268415318863656,
      "grad_norm": 0.3522081995644739,
      "learning_rate": 1.8768500097692783e-07,
      "loss": 1.1162,
      "step": 553
    },
    {
      "epoch": 0.09285175563563228,
      "grad_norm": 0.36917391246549447,
      "learning_rate": 1.876327597998699e-07,
      "loss": 1.1346,
      "step": 554
    },
    {
      "epoch": 0.09301935808262801,
      "grad_norm": 0.40198065067780697,
      "learning_rate": 1.8758041535401018e-07,
      "loss": 1.1986,
      "step": 555
    },
    {
      "epoch": 0.09318696052962373,
      "grad_norm": 0.38167962709865394,
      "learning_rate": 1.875279677010327e-07,
      "loss": 1.162,
      "step": 556
    },
    {
      "epoch": 0.09335456297661945,
      "grad_norm": 0.4067574433454854,
      "learning_rate": 1.8747541690274323e-07,
      "loss": 1.2007,
      "step": 557
    },
    {
      "epoch": 0.09352216542361519,
      "grad_norm": 0.3940845908190859,
      "learning_rate": 1.8742276302106898e-07,
      "loss": 1.1817,
      "step": 558
    },
    {
      "epoch": 0.09368976787061091,
      "grad_norm": 0.39913585061992474,
      "learning_rate": 1.8737000611805875e-07,
      "loss": 1.2101,
      "step": 559
    },
    {
      "epoch": 0.09385737031760663,
      "grad_norm": 0.3427723552527003,
      "learning_rate": 1.8731714625588268e-07,
      "loss": 1.0941,
      "step": 560
    },
    {
      "epoch": 0.09402497276460237,
      "grad_norm": 0.384828659344871,
      "learning_rate": 1.8726418349683231e-07,
      "loss": 1.1761,
      "step": 561
    },
    {
      "epoch": 0.09419257521159809,
      "grad_norm": 0.4307780104427617,
      "learning_rate": 1.8721111790332027e-07,
      "loss": 1.2155,
      "step": 562
    },
    {
      "epoch": 0.09436017765859382,
      "grad_norm": 0.3789472478324056,
      "learning_rate": 1.8715794953788055e-07,
      "loss": 1.1877,
      "step": 563
    },
    {
      "epoch": 0.09452778010558954,
      "grad_norm": 0.4376766426842412,
      "learning_rate": 1.871046784631682e-07,
      "loss": 1.248,
      "step": 564
    },
    {
      "epoch": 0.09469538255258526,
      "grad_norm": 0.405277985143647,
      "learning_rate": 1.8705130474195927e-07,
      "loss": 1.2223,
      "step": 565
    },
    {
      "epoch": 0.094862984999581,
      "grad_norm": 0.41215329071839724,
      "learning_rate": 1.8699782843715077e-07,
      "loss": 1.1732,
      "step": 566
    },
    {
      "epoch": 0.09503058744657672,
      "grad_norm": 0.4357942340636621,
      "learning_rate": 1.8694424961176064e-07,
      "loss": 1.2443,
      "step": 567
    },
    {
      "epoch": 0.09519818989357244,
      "grad_norm": 0.38149204598335185,
      "learning_rate": 1.8689056832892756e-07,
      "loss": 1.1488,
      "step": 568
    },
    {
      "epoch": 0.09536579234056818,
      "grad_norm": 0.38757206526156257,
      "learning_rate": 1.8683678465191105e-07,
      "loss": 1.195,
      "step": 569
    },
    {
      "epoch": 0.0955333947875639,
      "grad_norm": 0.380370970202923,
      "learning_rate": 1.8678289864409125e-07,
      "loss": 1.1718,
      "step": 570
    },
    {
      "epoch": 0.09570099723455962,
      "grad_norm": 0.40532584457640325,
      "learning_rate": 1.8672891036896882e-07,
      "loss": 1.2037,
      "step": 571
    },
    {
      "epoch": 0.09586859968155535,
      "grad_norm": 0.4626959509608885,
      "learning_rate": 1.8667481989016506e-07,
      "loss": 1.2251,
      "step": 572
    },
    {
      "epoch": 0.09603620212855107,
      "grad_norm": 0.4088038705397123,
      "learning_rate": 1.8662062727142163e-07,
      "loss": 1.171,
      "step": 573
    },
    {
      "epoch": 0.09620380457554681,
      "grad_norm": 0.389096502989025,
      "learning_rate": 1.8656633257660054e-07,
      "loss": 1.1496,
      "step": 574
    },
    {
      "epoch": 0.09637140702254253,
      "grad_norm": 0.41087642018298126,
      "learning_rate": 1.8651193586968414e-07,
      "loss": 1.2203,
      "step": 575
    },
    {
      "epoch": 0.09653900946953825,
      "grad_norm": 0.3973355951228905,
      "learning_rate": 1.8645743721477498e-07,
      "loss": 1.1154,
      "step": 576
    },
    {
      "epoch": 0.09670661191653399,
      "grad_norm": 0.43561120530226255,
      "learning_rate": 1.8640283667609574e-07,
      "loss": 1.1814,
      "step": 577
    },
    {
      "epoch": 0.0968742143635297,
      "grad_norm": 0.43406661835207766,
      "learning_rate": 1.8634813431798914e-07,
      "loss": 1.1791,
      "step": 578
    },
    {
      "epoch": 0.09704181681052543,
      "grad_norm": 0.43356097004402966,
      "learning_rate": 1.8629333020491796e-07,
      "loss": 1.2263,
      "step": 579
    },
    {
      "epoch": 0.09720941925752116,
      "grad_norm": 0.3825481445016694,
      "learning_rate": 1.8623842440146477e-07,
      "loss": 1.1851,
      "step": 580
    },
    {
      "epoch": 0.09737702170451688,
      "grad_norm": 0.39787374236455464,
      "learning_rate": 1.8618341697233212e-07,
      "loss": 1.1547,
      "step": 581
    },
    {
      "epoch": 0.09754462415151262,
      "grad_norm": 0.48438229170167985,
      "learning_rate": 1.8612830798234219e-07,
      "loss": 1.2406,
      "step": 582
    },
    {
      "epoch": 0.09771222659850834,
      "grad_norm": 0.38846486917586037,
      "learning_rate": 1.860730974964369e-07,
      "loss": 1.1502,
      "step": 583
    },
    {
      "epoch": 0.09787982904550406,
      "grad_norm": 0.3915626097944497,
      "learning_rate": 1.860177855796778e-07,
      "loss": 1.2036,
      "step": 584
    },
    {
      "epoch": 0.0980474314924998,
      "grad_norm": 0.44806352630161217,
      "learning_rate": 1.8596237229724594e-07,
      "loss": 1.2383,
      "step": 585
    },
    {
      "epoch": 0.09821503393949552,
      "grad_norm": 0.43929100660128545,
      "learning_rate": 1.859068577144418e-07,
      "loss": 1.2256,
      "step": 586
    },
    {
      "epoch": 0.09838263638649124,
      "grad_norm": 0.36555889570027483,
      "learning_rate": 1.858512418966853e-07,
      "loss": 1.1606,
      "step": 587
    },
    {
      "epoch": 0.09855023883348697,
      "grad_norm": 0.4299690587613859,
      "learning_rate": 1.8579552490951552e-07,
      "loss": 1.1833,
      "step": 588
    },
    {
      "epoch": 0.0987178412804827,
      "grad_norm": 0.4418971783763465,
      "learning_rate": 1.8573970681859097e-07,
      "loss": 1.2066,
      "step": 589
    },
    {
      "epoch": 0.09888544372747841,
      "grad_norm": 0.37210663102022495,
      "learning_rate": 1.8568378768968914e-07,
      "loss": 1.1575,
      "step": 590
    },
    {
      "epoch": 0.09905304617447415,
      "grad_norm": 0.39656878032133047,
      "learning_rate": 1.8562776758870661e-07,
      "loss": 1.1189,
      "step": 591
    },
    {
      "epoch": 0.09922064862146987,
      "grad_norm": 0.37636590031382616,
      "learning_rate": 1.8557164658165904e-07,
      "loss": 1.1549,
      "step": 592
    },
    {
      "epoch": 0.0993882510684656,
      "grad_norm": 0.39066118083869705,
      "learning_rate": 1.855154247346809e-07,
      "loss": 1.167,
      "step": 593
    },
    {
      "epoch": 0.09955585351546133,
      "grad_norm": 0.4352288316224097,
      "learning_rate": 1.8545910211402548e-07,
      "loss": 1.2398,
      "step": 594
    },
    {
      "epoch": 0.09972345596245705,
      "grad_norm": 0.39746479543070257,
      "learning_rate": 1.8540267878606495e-07,
      "loss": 1.2032,
      "step": 595
    },
    {
      "epoch": 0.09989105840945278,
      "grad_norm": 0.42129922538347536,
      "learning_rate": 1.8534615481729002e-07,
      "loss": 1.1818,
      "step": 596
    },
    {
      "epoch": 0.1000586608564485,
      "grad_norm": 0.3687398547820518,
      "learning_rate": 1.852895302743101e-07,
      "loss": 1.1612,
      "step": 597
    },
    {
      "epoch": 0.10022626330344422,
      "grad_norm": 0.37178921249970603,
      "learning_rate": 1.8523280522385306e-07,
      "loss": 1.1265,
      "step": 598
    },
    {
      "epoch": 0.10039386575043996,
      "grad_norm": 0.38818901863182625,
      "learning_rate": 1.8517597973276518e-07,
      "loss": 1.1685,
      "step": 599
    },
    {
      "epoch": 0.10056146819743568,
      "grad_norm": 0.3293159362505437,
      "learning_rate": 1.8511905386801124e-07,
      "loss": 1.0836,
      "step": 600
    },
    {
      "epoch": 0.10072907064443141,
      "grad_norm": 0.3892428793206639,
      "learning_rate": 1.8506202769667413e-07,
      "loss": 1.1598,
      "step": 601
    },
    {
      "epoch": 0.10089667309142714,
      "grad_norm": 0.3688121069044137,
      "learning_rate": 1.8500490128595503e-07,
      "loss": 1.151,
      "step": 602
    },
    {
      "epoch": 0.10106427553842286,
      "grad_norm": 0.3712866385241156,
      "learning_rate": 1.849476747031733e-07,
      "loss": 1.1543,
      "step": 603
    },
    {
      "epoch": 0.10123187798541859,
      "grad_norm": 0.41558527559829966,
      "learning_rate": 1.848903480157662e-07,
      "loss": 1.1967,
      "step": 604
    },
    {
      "epoch": 0.10139948043241431,
      "grad_norm": 0.4260958959006495,
      "learning_rate": 1.848329212912891e-07,
      "loss": 1.191,
      "step": 605
    },
    {
      "epoch": 0.10156708287941003,
      "grad_norm": 0.4080168790605839,
      "learning_rate": 1.8477539459741523e-07,
      "loss": 1.2189,
      "step": 606
    },
    {
      "epoch": 0.10173468532640577,
      "grad_norm": 0.4628743165285627,
      "learning_rate": 1.847177680019355e-07,
      "loss": 1.222,
      "step": 607
    },
    {
      "epoch": 0.10190228777340149,
      "grad_norm": 0.3349762208125001,
      "learning_rate": 1.846600415727587e-07,
      "loss": 1.1077,
      "step": 608
    },
    {
      "epoch": 0.10206989022039722,
      "grad_norm": 0.4035625431119916,
      "learning_rate": 1.846022153779112e-07,
      "loss": 1.1805,
      "step": 609
    },
    {
      "epoch": 0.10223749266739295,
      "grad_norm": 0.3722660520774615,
      "learning_rate": 1.8454428948553694e-07,
      "loss": 1.1338,
      "step": 610
    },
    {
      "epoch": 0.10240509511438867,
      "grad_norm": 0.3728167913821399,
      "learning_rate": 1.8448626396389737e-07,
      "loss": 1.1393,
      "step": 611
    },
    {
      "epoch": 0.1025726975613844,
      "grad_norm": 0.4350665861009182,
      "learning_rate": 1.8442813888137128e-07,
      "loss": 1.2303,
      "step": 612
    },
    {
      "epoch": 0.10274030000838012,
      "grad_norm": 0.3914818593287155,
      "learning_rate": 1.8436991430645486e-07,
      "loss": 1.1481,
      "step": 613
    },
    {
      "epoch": 0.10290790245537584,
      "grad_norm": 0.4684193372021882,
      "learning_rate": 1.843115903077615e-07,
      "loss": 1.2649,
      "step": 614
    },
    {
      "epoch": 0.10307550490237158,
      "grad_norm": 0.3908810837840773,
      "learning_rate": 1.842531669540218e-07,
      "loss": 1.1545,
      "step": 615
    },
    {
      "epoch": 0.1032431073493673,
      "grad_norm": 0.3831612278421622,
      "learning_rate": 1.8419464431408339e-07,
      "loss": 1.1719,
      "step": 616
    },
    {
      "epoch": 0.10341070979636302,
      "grad_norm": 0.42139517184905034,
      "learning_rate": 1.841360224569109e-07,
      "loss": 1.2286,
      "step": 617
    },
    {
      "epoch": 0.10357831224335876,
      "grad_norm": 0.3970470416242366,
      "learning_rate": 1.8407730145158595e-07,
      "loss": 1.1746,
      "step": 618
    },
    {
      "epoch": 0.10374591469035448,
      "grad_norm": 0.40124071163018893,
      "learning_rate": 1.8401848136730694e-07,
      "loss": 1.1259,
      "step": 619
    },
    {
      "epoch": 0.10391351713735021,
      "grad_norm": 0.4317590863735668,
      "learning_rate": 1.8395956227338903e-07,
      "loss": 1.2235,
      "step": 620
    },
    {
      "epoch": 0.10408111958434593,
      "grad_norm": 0.3849934286200893,
      "learning_rate": 1.8390054423926403e-07,
      "loss": 1.2102,
      "step": 621
    },
    {
      "epoch": 0.10424872203134165,
      "grad_norm": 0.383910254835738,
      "learning_rate": 1.8384142733448047e-07,
      "loss": 1.1413,
      "step": 622
    },
    {
      "epoch": 0.10441632447833739,
      "grad_norm": 0.39605382659332844,
      "learning_rate": 1.8378221162870324e-07,
      "loss": 1.2187,
      "step": 623
    },
    {
      "epoch": 0.10458392692533311,
      "grad_norm": 0.3760617426096714,
      "learning_rate": 1.837228971917137e-07,
      "loss": 1.1437,
      "step": 624
    },
    {
      "epoch": 0.10475152937232883,
      "grad_norm": 0.39627670604657833,
      "learning_rate": 1.8366348409340963e-07,
      "loss": 1.1524,
      "step": 625
    },
    {
      "epoch": 0.10491913181932457,
      "grad_norm": 0.4060969758326269,
      "learning_rate": 1.83603972403805e-07,
      "loss": 1.1912,
      "step": 626
    },
    {
      "epoch": 0.10508673426632029,
      "grad_norm": 0.35089441477804695,
      "learning_rate": 1.8354436219302998e-07,
      "loss": 1.1079,
      "step": 627
    },
    {
      "epoch": 0.10525433671331602,
      "grad_norm": 0.3325954563577985,
      "learning_rate": 1.8348465353133084e-07,
      "loss": 1.0833,
      "step": 628
    },
    {
      "epoch": 0.10542193916031174,
      "grad_norm": 0.41535503595650025,
      "learning_rate": 1.8342484648906993e-07,
      "loss": 1.2016,
      "step": 629
    },
    {
      "epoch": 0.10558954160730746,
      "grad_norm": 0.4701665793337646,
      "learning_rate": 1.8336494113672538e-07,
      "loss": 1.2216,
      "step": 630
    },
    {
      "epoch": 0.1057571440543032,
      "grad_norm": 0.4076175260654069,
      "learning_rate": 1.8330493754489137e-07,
      "loss": 1.1594,
      "step": 631
    },
    {
      "epoch": 0.10592474650129892,
      "grad_norm": 0.4096858039565214,
      "learning_rate": 1.832448357842777e-07,
      "loss": 1.1774,
      "step": 632
    },
    {
      "epoch": 0.10609234894829464,
      "grad_norm": 0.4083352704240439,
      "learning_rate": 1.8318463592570985e-07,
      "loss": 1.1558,
      "step": 633
    },
    {
      "epoch": 0.10625995139529038,
      "grad_norm": 0.41400235294414334,
      "learning_rate": 1.8312433804012903e-07,
      "loss": 1.1343,
      "step": 634
    },
    {
      "epoch": 0.1064275538422861,
      "grad_norm": 0.41352695636369113,
      "learning_rate": 1.8306394219859188e-07,
      "loss": 1.2183,
      "step": 635
    },
    {
      "epoch": 0.10659515628928182,
      "grad_norm": 0.4203648579802301,
      "learning_rate": 1.8300344847227044e-07,
      "loss": 1.1988,
      "step": 636
    },
    {
      "epoch": 0.10676275873627755,
      "grad_norm": 0.3873344423673321,
      "learning_rate": 1.8294285693245223e-07,
      "loss": 1.1731,
      "step": 637
    },
    {
      "epoch": 0.10693036118327327,
      "grad_norm": 0.4000444172174203,
      "learning_rate": 1.8288216765053984e-07,
      "loss": 1.1599,
      "step": 638
    },
    {
      "epoch": 0.10709796363026901,
      "grad_norm": 0.4299833887326289,
      "learning_rate": 1.8282138069805123e-07,
      "loss": 1.2007,
      "step": 639
    },
    {
      "epoch": 0.10726556607726473,
      "grad_norm": 0.3867751744727187,
      "learning_rate": 1.8276049614661937e-07,
      "loss": 1.1012,
      "step": 640
    },
    {
      "epoch": 0.10743316852426045,
      "grad_norm": 0.45969178066688293,
      "learning_rate": 1.826995140679922e-07,
      "loss": 1.1882,
      "step": 641
    },
    {
      "epoch": 0.10760077097125618,
      "grad_norm": 0.34207958310732145,
      "learning_rate": 1.826384345340327e-07,
      "loss": 1.0681,
      "step": 642
    },
    {
      "epoch": 0.1077683734182519,
      "grad_norm": 0.37909198156863805,
      "learning_rate": 1.8257725761671864e-07,
      "loss": 1.1154,
      "step": 643
    },
    {
      "epoch": 0.10793597586524763,
      "grad_norm": 0.4047194515797461,
      "learning_rate": 1.8251598338814253e-07,
      "loss": 1.216,
      "step": 644
    },
    {
      "epoch": 0.10810357831224336,
      "grad_norm": 0.4169957712760158,
      "learning_rate": 1.8245461192051154e-07,
      "loss": 1.1764,
      "step": 645
    },
    {
      "epoch": 0.10827118075923908,
      "grad_norm": 0.36548341921221184,
      "learning_rate": 1.823931432861475e-07,
      "loss": 1.096,
      "step": 646
    },
    {
      "epoch": 0.10843878320623482,
      "grad_norm": 0.40853577735360463,
      "learning_rate": 1.8233157755748667e-07,
      "loss": 1.1843,
      "step": 647
    },
    {
      "epoch": 0.10860638565323054,
      "grad_norm": 0.4127377362072803,
      "learning_rate": 1.822699148070798e-07,
      "loss": 1.1769,
      "step": 648
    },
    {
      "epoch": 0.10877398810022626,
      "grad_norm": 0.40899482274637633,
      "learning_rate": 1.822081551075919e-07,
      "loss": 1.2097,
      "step": 649
    },
    {
      "epoch": 0.108941590547222,
      "grad_norm": 0.3909119660622887,
      "learning_rate": 1.8214629853180227e-07,
      "loss": 1.1225,
      "step": 650
    },
    {
      "epoch": 0.10910919299421772,
      "grad_norm": 0.40276917212904717,
      "learning_rate": 1.8208434515260435e-07,
      "loss": 1.146,
      "step": 651
    },
    {
      "epoch": 0.10927679544121344,
      "grad_norm": 0.3738912641517323,
      "learning_rate": 1.8202229504300571e-07,
      "loss": 1.1052,
      "step": 652
    },
    {
      "epoch": 0.10944439788820917,
      "grad_norm": 0.3846016576626563,
      "learning_rate": 1.819601482761278e-07,
      "loss": 1.147,
      "step": 653
    },
    {
      "epoch": 0.10961200033520489,
      "grad_norm": 0.41704514461635855,
      "learning_rate": 1.8189790492520612e-07,
      "loss": 1.1725,
      "step": 654
    },
    {
      "epoch": 0.10977960278220061,
      "grad_norm": 0.40331646444212316,
      "learning_rate": 1.8183556506358988e-07,
      "loss": 1.1371,
      "step": 655
    },
    {
      "epoch": 0.10994720522919635,
      "grad_norm": 0.38914187568288366,
      "learning_rate": 1.81773128764742e-07,
      "loss": 1.1579,
      "step": 656
    },
    {
      "epoch": 0.11011480767619207,
      "grad_norm": 0.40689678267424884,
      "learning_rate": 1.817105961022392e-07,
      "loss": 1.142,
      "step": 657
    },
    {
      "epoch": 0.1102824101231878,
      "grad_norm": 0.38544387297713745,
      "learning_rate": 1.8164796714977157e-07,
      "loss": 1.1085,
      "step": 658
    },
    {
      "epoch": 0.11045001257018353,
      "grad_norm": 0.3870719495392021,
      "learning_rate": 1.8158524198114277e-07,
      "loss": 1.154,
      "step": 659
    },
    {
      "epoch": 0.11061761501717925,
      "grad_norm": 0.40767070105693837,
      "learning_rate": 1.8152242067026981e-07,
      "loss": 1.1605,
      "step": 660
    },
    {
      "epoch": 0.11078521746417498,
      "grad_norm": 0.42664277083802693,
      "learning_rate": 1.8145950329118307e-07,
      "loss": 1.1937,
      "step": 661
    },
    {
      "epoch": 0.1109528199111707,
      "grad_norm": 0.38664695150560424,
      "learning_rate": 1.8139648991802597e-07,
      "loss": 1.1212,
      "step": 662
    },
    {
      "epoch": 0.11112042235816642,
      "grad_norm": 0.40705247220262614,
      "learning_rate": 1.813333806250553e-07,
      "loss": 1.1881,
      "step": 663
    },
    {
      "epoch": 0.11128802480516216,
      "grad_norm": 0.3843110841577225,
      "learning_rate": 1.8127017548664068e-07,
      "loss": 1.1186,
      "step": 664
    },
    {
      "epoch": 0.11145562725215788,
      "grad_norm": 0.3520312094901096,
      "learning_rate": 1.8120687457726477e-07,
      "loss": 1.1157,
      "step": 665
    },
    {
      "epoch": 0.11162322969915361,
      "grad_norm": 0.4027243843167232,
      "learning_rate": 1.8114347797152303e-07,
      "loss": 1.1639,
      "step": 666
    },
    {
      "epoch": 0.11179083214614934,
      "grad_norm": 0.3535328505571573,
      "learning_rate": 1.8107998574412374e-07,
      "loss": 1.0766,
      "step": 667
    },
    {
      "epoch": 0.11195843459314506,
      "grad_norm": 0.4250328530086893,
      "learning_rate": 1.8101639796988785e-07,
      "loss": 1.1559,
      "step": 668
    },
    {
      "epoch": 0.11212603704014079,
      "grad_norm": 0.4154129740104389,
      "learning_rate": 1.809527147237489e-07,
      "loss": 1.1744,
      "step": 669
    },
    {
      "epoch": 0.11229363948713651,
      "grad_norm": 0.41278906116217096,
      "learning_rate": 1.8088893608075298e-07,
      "loss": 1.1475,
      "step": 670
    },
    {
      "epoch": 0.11246124193413223,
      "grad_norm": 0.31562918060009465,
      "learning_rate": 1.8082506211605852e-07,
      "loss": 1.0563,
      "step": 671
    },
    {
      "epoch": 0.11262884438112797,
      "grad_norm": 0.413278395313315,
      "learning_rate": 1.8076109290493628e-07,
      "loss": 1.1717,
      "step": 672
    },
    {
      "epoch": 0.11279644682812369,
      "grad_norm": 0.39538953611785826,
      "learning_rate": 1.806970285227694e-07,
      "loss": 1.1007,
      "step": 673
    },
    {
      "epoch": 0.11296404927511941,
      "grad_norm": 0.4296507184630236,
      "learning_rate": 1.8063286904505297e-07,
      "loss": 1.1569,
      "step": 674
    },
    {
      "epoch": 0.11313165172211515,
      "grad_norm": 0.33910289332580024,
      "learning_rate": 1.805686145473943e-07,
      "loss": 1.0707,
      "step": 675
    },
    {
      "epoch": 0.11329925416911087,
      "grad_norm": 0.40092377723093947,
      "learning_rate": 1.8050426510551264e-07,
      "loss": 1.1431,
      "step": 676
    },
    {
      "epoch": 0.1134668566161066,
      "grad_norm": 0.42115505551029137,
      "learning_rate": 1.8043982079523903e-07,
      "loss": 1.1905,
      "step": 677
    },
    {
      "epoch": 0.11363445906310232,
      "grad_norm": 0.4023635830124107,
      "learning_rate": 1.8037528169251647e-07,
      "loss": 1.1834,
      "step": 678
    },
    {
      "epoch": 0.11380206151009804,
      "grad_norm": 0.4188909678877205,
      "learning_rate": 1.8031064787339946e-07,
      "loss": 1.1405,
      "step": 679
    },
    {
      "epoch": 0.11396966395709378,
      "grad_norm": 0.3766372283719351,
      "learning_rate": 1.8024591941405433e-07,
      "loss": 1.1559,
      "step": 680
    },
    {
      "epoch": 0.1141372664040895,
      "grad_norm": 0.39432454840983194,
      "learning_rate": 1.8018109639075885e-07,
      "loss": 1.065,
      "step": 681
    },
    {
      "epoch": 0.11430486885108522,
      "grad_norm": 0.41848837598159877,
      "learning_rate": 1.8011617887990212e-07,
      "loss": 1.1482,
      "step": 682
    },
    {
      "epoch": 0.11447247129808096,
      "grad_norm": 0.3968061953282843,
      "learning_rate": 1.8005116695798476e-07,
      "loss": 1.119,
      "step": 683
    },
    {
      "epoch": 0.11464007374507668,
      "grad_norm": 0.44345068027875445,
      "learning_rate": 1.799860607016185e-07,
      "loss": 1.2298,
      "step": 684
    },
    {
      "epoch": 0.11480767619207241,
      "grad_norm": 0.3813368374860304,
      "learning_rate": 1.7992086018752638e-07,
      "loss": 1.1355,
      "step": 685
    },
    {
      "epoch": 0.11497527863906813,
      "grad_norm": 0.3809916284012114,
      "learning_rate": 1.7985556549254238e-07,
      "loss": 1.1133,
      "step": 686
    },
    {
      "epoch": 0.11514288108606385,
      "grad_norm": 0.34211820584780006,
      "learning_rate": 1.797901766936116e-07,
      "loss": 1.0942,
      "step": 687
    },
    {
      "epoch": 0.11531048353305959,
      "grad_norm": 0.4399140132934113,
      "learning_rate": 1.7972469386778983e-07,
      "loss": 1.1477,
      "step": 688
    },
    {
      "epoch": 0.11547808598005531,
      "grad_norm": 0.3690391991131506,
      "learning_rate": 1.7965911709224393e-07,
      "loss": 1.1455,
      "step": 689
    },
    {
      "epoch": 0.11564568842705103,
      "grad_norm": 0.39983390788874873,
      "learning_rate": 1.7959344644425127e-07,
      "loss": 1.127,
      "step": 690
    },
    {
      "epoch": 0.11581329087404676,
      "grad_norm": 0.4344908486821997,
      "learning_rate": 1.795276820011999e-07,
      "loss": 1.183,
      "step": 691
    },
    {
      "epoch": 0.11598089332104249,
      "grad_norm": 0.37074278565598495,
      "learning_rate": 1.7946182384058845e-07,
      "loss": 1.1179,
      "step": 692
    },
    {
      "epoch": 0.1161484957680382,
      "grad_norm": 0.3781922420891761,
      "learning_rate": 1.793958720400259e-07,
      "loss": 1.0887,
      "step": 693
    },
    {
      "epoch": 0.11631609821503394,
      "grad_norm": 0.4271240705213637,
      "learning_rate": 1.7932982667723168e-07,
      "loss": 1.1669,
      "step": 694
    },
    {
      "epoch": 0.11648370066202966,
      "grad_norm": 0.3936244921866956,
      "learning_rate": 1.7926368783003536e-07,
      "loss": 1.1412,
      "step": 695
    },
    {
      "epoch": 0.1166513031090254,
      "grad_norm": 0.3281249214576609,
      "learning_rate": 1.791974555763768e-07,
      "loss": 1.0979,
      "step": 696
    },
    {
      "epoch": 0.11681890555602112,
      "grad_norm": 0.34159121302151074,
      "learning_rate": 1.791311299943058e-07,
      "loss": 1.0607,
      "step": 697
    },
    {
      "epoch": 0.11698650800301684,
      "grad_norm": 0.36982639405310525,
      "learning_rate": 1.790647111619823e-07,
      "loss": 1.115,
      "step": 698
    },
    {
      "epoch": 0.11715411045001257,
      "grad_norm": 0.3727021672378182,
      "learning_rate": 1.7899819915767598e-07,
      "loss": 1.1259,
      "step": 699
    },
    {
      "epoch": 0.1173217128970083,
      "grad_norm": 0.3754913405893485,
      "learning_rate": 1.7893159405976637e-07,
      "loss": 1.112,
      "step": 700
    },
    {
      "epoch": 0.11748931534400402,
      "grad_norm": 0.3620266683599068,
      "learning_rate": 1.7886489594674272e-07,
      "loss": 1.0759,
      "step": 701
    },
    {
      "epoch": 0.11765691779099975,
      "grad_norm": 0.3416427359272222,
      "learning_rate": 1.7879810489720392e-07,
      "loss": 1.0991,
      "step": 702
    },
    {
      "epoch": 0.11782452023799547,
      "grad_norm": 0.421468800790263,
      "learning_rate": 1.7873122098985823e-07,
      "loss": 1.1824,
      "step": 703
    },
    {
      "epoch": 0.11799212268499121,
      "grad_norm": 0.3930366563690931,
      "learning_rate": 1.786642443035236e-07,
      "loss": 1.1478,
      "step": 704
    },
    {
      "epoch": 0.11815972513198693,
      "grad_norm": 0.3718173983674682,
      "learning_rate": 1.7859717491712706e-07,
      "loss": 1.0849,
      "step": 705
    },
    {
      "epoch": 0.11832732757898265,
      "grad_norm": 0.37481470845252446,
      "learning_rate": 1.7853001290970498e-07,
      "loss": 1.1108,
      "step": 706
    },
    {
      "epoch": 0.11849493002597838,
      "grad_norm": 0.3412005394804902,
      "learning_rate": 1.784627583604029e-07,
      "loss": 1.0675,
      "step": 707
    },
    {
      "epoch": 0.1186625324729741,
      "grad_norm": 0.43602332600554244,
      "learning_rate": 1.783954113484754e-07,
      "loss": 1.2143,
      "step": 708
    },
    {
      "epoch": 0.11883013491996983,
      "grad_norm": 0.33444839164298523,
      "learning_rate": 1.7832797195328596e-07,
      "loss": 1.0455,
      "step": 709
    },
    {
      "epoch": 0.11899773736696556,
      "grad_norm": 0.42922795245089357,
      "learning_rate": 1.7826044025430707e-07,
      "loss": 1.1435,
      "step": 710
    },
    {
      "epoch": 0.11916533981396128,
      "grad_norm": 0.3763988470150692,
      "learning_rate": 1.7819281633111982e-07,
      "loss": 1.1476,
      "step": 711
    },
    {
      "epoch": 0.119332942260957,
      "grad_norm": 0.3904181078750033,
      "learning_rate": 1.781251002634141e-07,
      "loss": 1.1836,
      "step": 712
    },
    {
      "epoch": 0.11950054470795274,
      "grad_norm": 0.3703808789759766,
      "learning_rate": 1.780572921309883e-07,
      "loss": 1.1024,
      "step": 713
    },
    {
      "epoch": 0.11966814715494846,
      "grad_norm": 0.39756303210761434,
      "learning_rate": 1.779893920137494e-07,
      "loss": 1.1664,
      "step": 714
    },
    {
      "epoch": 0.1198357496019442,
      "grad_norm": 0.4203527275709,
      "learning_rate": 1.779213999917127e-07,
      "loss": 1.1645,
      "step": 715
    },
    {
      "epoch": 0.12000335204893992,
      "grad_norm": 0.3341223334405362,
      "learning_rate": 1.7785331614500183e-07,
      "loss": 1.0772,
      "step": 716
    },
    {
      "epoch": 0.12017095449593564,
      "grad_norm": 0.41647130986749775,
      "learning_rate": 1.7778514055384864e-07,
      "loss": 1.1454,
      "step": 717
    },
    {
      "epoch": 0.12033855694293137,
      "grad_norm": 0.3724204137850256,
      "learning_rate": 1.7771687329859306e-07,
      "loss": 1.1042,
      "step": 718
    },
    {
      "epoch": 0.12050615938992709,
      "grad_norm": 0.3885672418040377,
      "learning_rate": 1.7764851445968308e-07,
      "loss": 1.112,
      "step": 719
    },
    {
      "epoch": 0.12067376183692281,
      "grad_norm": 0.35540885118088367,
      "learning_rate": 1.7758006411767455e-07,
      "loss": 1.0448,
      "step": 720
    },
    {
      "epoch": 0.12084136428391855,
      "grad_norm": 0.35391964761367695,
      "learning_rate": 1.775115223532313e-07,
      "loss": 1.0915,
      "step": 721
    },
    {
      "epoch": 0.12100896673091427,
      "grad_norm": 0.38042556309389014,
      "learning_rate": 1.7744288924712465e-07,
      "loss": 1.114,
      "step": 722
    },
    {
      "epoch": 0.12117656917791,
      "grad_norm": 0.369729941120576,
      "learning_rate": 1.7737416488023383e-07,
      "loss": 1.1106,
      "step": 723
    },
    {
      "epoch": 0.12134417162490573,
      "grad_norm": 0.3800344649163796,
      "learning_rate": 1.7730534933354538e-07,
      "loss": 1.1375,
      "step": 724
    },
    {
      "epoch": 0.12151177407190145,
      "grad_norm": 0.37043967508889253,
      "learning_rate": 1.7723644268815342e-07,
      "loss": 1.0986,
      "step": 725
    },
    {
      "epoch": 0.12167937651889718,
      "grad_norm": 0.47001946077101536,
      "learning_rate": 1.771674450252594e-07,
      "loss": 1.193,
      "step": 726
    },
    {
      "epoch": 0.1218469789658929,
      "grad_norm": 0.3709176319312981,
      "learning_rate": 1.770983564261721e-07,
      "loss": 1.0971,
      "step": 727
    },
    {
      "epoch": 0.12201458141288862,
      "grad_norm": 0.34875157731406287,
      "learning_rate": 1.7702917697230726e-07,
      "loss": 1.0789,
      "step": 728
    },
    {
      "epoch": 0.12218218385988436,
      "grad_norm": 0.3931881731389898,
      "learning_rate": 1.7695990674518787e-07,
      "loss": 1.1502,
      "step": 729
    },
    {
      "epoch": 0.12234978630688008,
      "grad_norm": 0.3427246282082076,
      "learning_rate": 1.7689054582644384e-07,
      "loss": 1.1095,
      "step": 730
    },
    {
      "epoch": 0.1225173887538758,
      "grad_norm": 0.3667428689737461,
      "learning_rate": 1.768210942978119e-07,
      "loss": 1.1237,
      "step": 731
    },
    {
      "epoch": 0.12268499120087153,
      "grad_norm": 0.393339765025784,
      "learning_rate": 1.7675155224113563e-07,
      "loss": 1.1398,
      "step": 732
    },
    {
      "epoch": 0.12285259364786726,
      "grad_norm": 0.32440788430339224,
      "learning_rate": 1.7668191973836528e-07,
      "loss": 1.0499,
      "step": 733
    },
    {
      "epoch": 0.12302019609486299,
      "grad_norm": 0.3788464848362284,
      "learning_rate": 1.766121968715576e-07,
      "loss": 1.1545,
      "step": 734
    },
    {
      "epoch": 0.12318779854185871,
      "grad_norm": 0.38191814892171855,
      "learning_rate": 1.7654238372287598e-07,
      "loss": 1.1153,
      "step": 735
    },
    {
      "epoch": 0.12335540098885443,
      "grad_norm": 0.37521796732167534,
      "learning_rate": 1.7647248037459004e-07,
      "loss": 1.1349,
      "step": 736
    },
    {
      "epoch": 0.12352300343585017,
      "grad_norm": 0.36112202662440385,
      "learning_rate": 1.764024869090758e-07,
      "loss": 1.1105,
      "step": 737
    },
    {
      "epoch": 0.12369060588284589,
      "grad_norm": 0.3358394895299196,
      "learning_rate": 1.7633240340881542e-07,
      "loss": 1.1013,
      "step": 738
    },
    {
      "epoch": 0.12385820832984161,
      "grad_norm": 0.37436778430319007,
      "learning_rate": 1.762622299563972e-07,
      "loss": 1.0919,
      "step": 739
    },
    {
      "epoch": 0.12402581077683734,
      "grad_norm": 0.38079592580527355,
      "learning_rate": 1.7619196663451543e-07,
      "loss": 1.1563,
      "step": 740
    },
    {
      "epoch": 0.12419341322383307,
      "grad_norm": 0.3519887216673953,
      "learning_rate": 1.7612161352597032e-07,
      "loss": 1.1072,
      "step": 741
    },
    {
      "epoch": 0.1243610156708288,
      "grad_norm": 0.35206273979314884,
      "learning_rate": 1.7605117071366777e-07,
      "loss": 1.114,
      "step": 742
    },
    {
      "epoch": 0.12452861811782452,
      "grad_norm": 0.30007419240884364,
      "learning_rate": 1.7598063828061958e-07,
      "loss": 1.0673,
      "step": 743
    },
    {
      "epoch": 0.12469622056482024,
      "grad_norm": 0.3343864159332185,
      "learning_rate": 1.7591001630994304e-07,
      "loss": 1.0335,
      "step": 744
    },
    {
      "epoch": 0.12486382301181598,
      "grad_norm": 0.3780112302278805,
      "learning_rate": 1.75839304884861e-07,
      "loss": 1.1287,
      "step": 745
    },
    {
      "epoch": 0.1250314254588117,
      "grad_norm": 0.3493155398404103,
      "learning_rate": 1.7576850408870163e-07,
      "loss": 1.0806,
      "step": 746
    },
    {
      "epoch": 0.12519902790580742,
      "grad_norm": 0.35904497189277235,
      "learning_rate": 1.756976140048986e-07,
      "loss": 1.1453,
      "step": 747
    },
    {
      "epoch": 0.12536663035280315,
      "grad_norm": 0.3496620733447745,
      "learning_rate": 1.756266347169906e-07,
      "loss": 1.1117,
      "step": 748
    },
    {
      "epoch": 0.1255342327997989,
      "grad_norm": 0.3225845990119082,
      "learning_rate": 1.755555663086216e-07,
      "loss": 1.0613,
      "step": 749
    },
    {
      "epoch": 0.1257018352467946,
      "grad_norm": 0.33542367443563625,
      "learning_rate": 1.754844088635405e-07,
      "loss": 1.0748,
      "step": 750
    },
    {
      "epoch": 0.12586943769379033,
      "grad_norm": 0.3890147868051163,
      "learning_rate": 1.7541316246560108e-07,
      "loss": 1.1467,
      "step": 751
    },
    {
      "epoch": 0.12603704014078607,
      "grad_norm": 0.3408207435795358,
      "learning_rate": 1.7534182719876215e-07,
      "loss": 1.0749,
      "step": 752
    },
    {
      "epoch": 0.12620464258778177,
      "grad_norm": 0.30317551240551727,
      "learning_rate": 1.7527040314708702e-07,
      "loss": 1.0044,
      "step": 753
    },
    {
      "epoch": 0.1263722450347775,
      "grad_norm": 0.3498146830747345,
      "learning_rate": 1.7519889039474372e-07,
      "loss": 1.1078,
      "step": 754
    },
    {
      "epoch": 0.12653984748177324,
      "grad_norm": 0.3606029704823278,
      "learning_rate": 1.751272890260048e-07,
      "loss": 1.1073,
      "step": 755
    },
    {
      "epoch": 0.12670744992876895,
      "grad_norm": 0.36212540205330523,
      "learning_rate": 1.7505559912524725e-07,
      "loss": 1.0825,
      "step": 756
    },
    {
      "epoch": 0.12687505237576469,
      "grad_norm": 0.3772105032567963,
      "learning_rate": 1.7498382077695236e-07,
      "loss": 1.141,
      "step": 757
    },
    {
      "epoch": 0.12704265482276042,
      "grad_norm": 0.41342435282293094,
      "learning_rate": 1.749119540657057e-07,
      "loss": 1.2253,
      "step": 758
    },
    {
      "epoch": 0.12721025726975613,
      "grad_norm": 0.37193885879944233,
      "learning_rate": 1.7483999907619693e-07,
      "loss": 1.1204,
      "step": 759
    },
    {
      "epoch": 0.12737785971675186,
      "grad_norm": 0.3188373713161176,
      "learning_rate": 1.747679558932197e-07,
      "loss": 1.0418,
      "step": 760
    },
    {
      "epoch": 0.1275454621637476,
      "grad_norm": 0.3509244229219662,
      "learning_rate": 1.746958246016717e-07,
      "loss": 1.057,
      "step": 761
    },
    {
      "epoch": 0.1277130646107433,
      "grad_norm": 0.3908272746677847,
      "learning_rate": 1.7462360528655436e-07,
      "loss": 1.095,
      "step": 762
    },
    {
      "epoch": 0.12788066705773904,
      "grad_norm": 0.3857131930697678,
      "learning_rate": 1.7455129803297286e-07,
      "loss": 1.132,
      "step": 763
    },
    {
      "epoch": 0.12804826950473477,
      "grad_norm": 0.3496053453019053,
      "learning_rate": 1.7447890292613598e-07,
      "loss": 1.0647,
      "step": 764
    },
    {
      "epoch": 0.1282158719517305,
      "grad_norm": 0.37330685934769947,
      "learning_rate": 1.744064200513561e-07,
      "loss": 1.0662,
      "step": 765
    },
    {
      "epoch": 0.12838347439872622,
      "grad_norm": 0.3634498280036985,
      "learning_rate": 1.74333849494049e-07,
      "loss": 1.0788,
      "step": 766
    },
    {
      "epoch": 0.12855107684572195,
      "grad_norm": 0.3090350132219015,
      "learning_rate": 1.7426119133973375e-07,
      "loss": 1.0336,
      "step": 767
    },
    {
      "epoch": 0.12871867929271769,
      "grad_norm": 0.39028355446648694,
      "learning_rate": 1.7418844567403269e-07,
      "loss": 1.1154,
      "step": 768
    },
    {
      "epoch": 0.1288862817397134,
      "grad_norm": 0.34937497107007937,
      "learning_rate": 1.7411561258267126e-07,
      "loss": 1.1136,
      "step": 769
    },
    {
      "epoch": 0.12905388418670913,
      "grad_norm": 0.3987026491667282,
      "learning_rate": 1.7404269215147793e-07,
      "loss": 1.198,
      "step": 770
    },
    {
      "epoch": 0.12922148663370486,
      "grad_norm": 0.38852567427201834,
      "learning_rate": 1.739696844663841e-07,
      "loss": 1.1323,
      "step": 771
    },
    {
      "epoch": 0.12938908908070057,
      "grad_norm": 0.3683004319341922,
      "learning_rate": 1.7389658961342402e-07,
      "loss": 1.1324,
      "step": 772
    },
    {
      "epoch": 0.1295566915276963,
      "grad_norm": 0.3324546712402692,
      "learning_rate": 1.7382340767873457e-07,
      "loss": 1.0931,
      "step": 773
    },
    {
      "epoch": 0.12972429397469204,
      "grad_norm": 0.390524423942671,
      "learning_rate": 1.7375013874855536e-07,
      "loss": 1.1209,
      "step": 774
    },
    {
      "epoch": 0.12989189642168775,
      "grad_norm": 0.3455674633611906,
      "learning_rate": 1.736767829092285e-07,
      "loss": 1.0628,
      "step": 775
    },
    {
      "epoch": 0.13005949886868348,
      "grad_norm": 0.4274396331109979,
      "learning_rate": 1.736033402471984e-07,
      "loss": 1.1538,
      "step": 776
    },
    {
      "epoch": 0.13022710131567922,
      "grad_norm": 0.35192934847631313,
      "learning_rate": 1.7352981084901193e-07,
      "loss": 1.1039,
      "step": 777
    },
    {
      "epoch": 0.13039470376267492,
      "grad_norm": 0.33100493722168467,
      "learning_rate": 1.7345619480131805e-07,
      "loss": 1.049,
      "step": 778
    },
    {
      "epoch": 0.13056230620967066,
      "grad_norm": 0.37372223679297983,
      "learning_rate": 1.73382492190868e-07,
      "loss": 1.1383,
      "step": 779
    },
    {
      "epoch": 0.1307299086566664,
      "grad_norm": 0.38726611017784385,
      "learning_rate": 1.733087031045148e-07,
      "loss": 1.0837,
      "step": 780
    },
    {
      "epoch": 0.1308975111036621,
      "grad_norm": 0.30627719980994655,
      "learning_rate": 1.7323482762921354e-07,
      "loss": 1.0731,
      "step": 781
    },
    {
      "epoch": 0.13106511355065784,
      "grad_norm": 0.32871972546016015,
      "learning_rate": 1.7316086585202108e-07,
      "loss": 1.0567,
      "step": 782
    },
    {
      "epoch": 0.13123271599765357,
      "grad_norm": 0.34520786928471475,
      "learning_rate": 1.7308681786009599e-07,
      "loss": 1.0792,
      "step": 783
    },
    {
      "epoch": 0.1314003184446493,
      "grad_norm": 0.3019215019273035,
      "learning_rate": 1.7301268374069834e-07,
      "loss": 1.0303,
      "step": 784
    },
    {
      "epoch": 0.131567920891645,
      "grad_norm": 0.3692506040919821,
      "learning_rate": 1.7293846358118985e-07,
      "loss": 1.1103,
      "step": 785
    },
    {
      "epoch": 0.13173552333864075,
      "grad_norm": 0.34971852324575453,
      "learning_rate": 1.7286415746903353e-07,
      "loss": 1.064,
      "step": 786
    },
    {
      "epoch": 0.13190312578563648,
      "grad_norm": 0.32263668089921127,
      "learning_rate": 1.7278976549179368e-07,
      "loss": 1.1113,
      "step": 787
    },
    {
      "epoch": 0.1320707282326322,
      "grad_norm": 0.349011709162542,
      "learning_rate": 1.7271528773713585e-07,
      "loss": 1.1149,
      "step": 788
    },
    {
      "epoch": 0.13223833067962792,
      "grad_norm": 0.39816027515013624,
      "learning_rate": 1.7264072429282654e-07,
      "loss": 1.1134,
      "step": 789
    },
    {
      "epoch": 0.13240593312662366,
      "grad_norm": 0.3712874353897627,
      "learning_rate": 1.725660752467334e-07,
      "loss": 1.1229,
      "step": 790
    },
    {
      "epoch": 0.13257353557361937,
      "grad_norm": 0.3161194703086633,
      "learning_rate": 1.7249134068682487e-07,
      "loss": 1.0342,
      "step": 791
    },
    {
      "epoch": 0.1327411380206151,
      "grad_norm": 0.336663509023671,
      "learning_rate": 1.7241652070117015e-07,
      "loss": 1.063,
      "step": 792
    },
    {
      "epoch": 0.13290874046761084,
      "grad_norm": 0.3399741999068847,
      "learning_rate": 1.7234161537793912e-07,
      "loss": 1.083,
      "step": 793
    },
    {
      "epoch": 0.13307634291460654,
      "grad_norm": 0.36372760921768577,
      "learning_rate": 1.7226662480540223e-07,
      "loss": 1.1484,
      "step": 794
    },
    {
      "epoch": 0.13324394536160228,
      "grad_norm": 0.368599191557536,
      "learning_rate": 1.721915490719304e-07,
      "loss": 1.0972,
      "step": 795
    },
    {
      "epoch": 0.133411547808598,
      "grad_norm": 0.34182109114405895,
      "learning_rate": 1.7211638826599485e-07,
      "loss": 1.1528,
      "step": 796
    },
    {
      "epoch": 0.13357915025559372,
      "grad_norm": 0.34977544683135203,
      "learning_rate": 1.7204114247616715e-07,
      "loss": 1.1048,
      "step": 797
    },
    {
      "epoch": 0.13374675270258946,
      "grad_norm": 0.37451020589881967,
      "learning_rate": 1.7196581179111895e-07,
      "loss": 1.0673,
      "step": 798
    },
    {
      "epoch": 0.1339143551495852,
      "grad_norm": 0.3336499713834984,
      "learning_rate": 1.718903962996219e-07,
      "loss": 1.1066,
      "step": 799
    },
    {
      "epoch": 0.1340819575965809,
      "grad_norm": 0.32321158847276504,
      "learning_rate": 1.7181489609054777e-07,
      "loss": 1.0316,
      "step": 800
    },
    {
      "epoch": 0.13424956004357663,
      "grad_norm": 0.3304715662588373,
      "learning_rate": 1.717393112528679e-07,
      "loss": 1.083,
      "step": 801
    },
    {
      "epoch": 0.13441716249057237,
      "grad_norm": 0.33737954062443715,
      "learning_rate": 1.716636418756536e-07,
      "loss": 1.0602,
      "step": 802
    },
    {
      "epoch": 0.1345847649375681,
      "grad_norm": 0.34262268347477376,
      "learning_rate": 1.7158788804807564e-07,
      "loss": 1.1272,
      "step": 803
    },
    {
      "epoch": 0.1347523673845638,
      "grad_norm": 0.3556608651966313,
      "learning_rate": 1.715120498594044e-07,
      "loss": 1.0976,
      "step": 804
    },
    {
      "epoch": 0.13491996983155954,
      "grad_norm": 0.34635630118088845,
      "learning_rate": 1.714361273990096e-07,
      "loss": 1.0918,
      "step": 805
    },
    {
      "epoch": 0.13508757227855528,
      "grad_norm": 0.39616954665650284,
      "learning_rate": 1.7136012075636038e-07,
      "loss": 1.145,
      "step": 806
    },
    {
      "epoch": 0.135255174725551,
      "grad_norm": 0.3396833446205391,
      "learning_rate": 1.712840300210249e-07,
      "loss": 1.1188,
      "step": 807
    },
    {
      "epoch": 0.13542277717254672,
      "grad_norm": 0.3429116020128108,
      "learning_rate": 1.7120785528267065e-07,
      "loss": 1.1032,
      "step": 808
    },
    {
      "epoch": 0.13559037961954246,
      "grad_norm": 0.3455452625108052,
      "learning_rate": 1.7113159663106394e-07,
      "loss": 1.0889,
      "step": 809
    },
    {
      "epoch": 0.13575798206653816,
      "grad_norm": 0.3414842091708314,
      "learning_rate": 1.7105525415607002e-07,
      "loss": 1.0717,
      "step": 810
    },
    {
      "epoch": 0.1359255845135339,
      "grad_norm": 0.34400544881628703,
      "learning_rate": 1.709788279476529e-07,
      "loss": 1.0849,
      "step": 811
    },
    {
      "epoch": 0.13609318696052963,
      "grad_norm": 0.39186856047981633,
      "learning_rate": 1.7090231809587528e-07,
      "loss": 1.1365,
      "step": 812
    },
    {
      "epoch": 0.13626078940752534,
      "grad_norm": 0.3165067750966069,
      "learning_rate": 1.7082572469089844e-07,
      "loss": 1.0303,
      "step": 813
    },
    {
      "epoch": 0.13642839185452107,
      "grad_norm": 0.3335130959451929,
      "learning_rate": 1.7074904782298214e-07,
      "loss": 1.0913,
      "step": 814
    },
    {
      "epoch": 0.1365959943015168,
      "grad_norm": 0.3462072526395098,
      "learning_rate": 1.7067228758248443e-07,
      "loss": 1.089,
      "step": 815
    },
    {
      "epoch": 0.13676359674851252,
      "grad_norm": 0.3728520401376406,
      "learning_rate": 1.7059544405986164e-07,
      "loss": 1.1354,
      "step": 816
    },
    {
      "epoch": 0.13693119919550825,
      "grad_norm": 0.3327371644243139,
      "learning_rate": 1.7051851734566826e-07,
      "loss": 1.0951,
      "step": 817
    },
    {
      "epoch": 0.137098801642504,
      "grad_norm": 0.34911367292205875,
      "learning_rate": 1.704415075305568e-07,
      "loss": 1.0885,
      "step": 818
    },
    {
      "epoch": 0.1372664040894997,
      "grad_norm": 0.3396796203698726,
      "learning_rate": 1.7036441470527766e-07,
      "loss": 1.0918,
      "step": 819
    },
    {
      "epoch": 0.13743400653649543,
      "grad_norm": 0.37920757159604657,
      "learning_rate": 1.7028723896067913e-07,
      "loss": 1.1273,
      "step": 820
    },
    {
      "epoch": 0.13760160898349116,
      "grad_norm": 0.3159641703760726,
      "learning_rate": 1.7020998038770722e-07,
      "loss": 1.0294,
      "step": 821
    },
    {
      "epoch": 0.1377692114304869,
      "grad_norm": 0.3737031554251462,
      "learning_rate": 1.7013263907740544e-07,
      "loss": 1.0914,
      "step": 822
    },
    {
      "epoch": 0.1379368138774826,
      "grad_norm": 0.3455537831702491,
      "learning_rate": 1.7005521512091491e-07,
      "loss": 1.0854,
      "step": 823
    },
    {
      "epoch": 0.13810441632447834,
      "grad_norm": 0.30544921506204437,
      "learning_rate": 1.699777086094741e-07,
      "loss": 1.0068,
      "step": 824
    },
    {
      "epoch": 0.13827201877147408,
      "grad_norm": 0.31980575239295134,
      "learning_rate": 1.699001196344188e-07,
      "loss": 1.0336,
      "step": 825
    },
    {
      "epoch": 0.13843962121846978,
      "grad_norm": 0.3732573206412432,
      "learning_rate": 1.6982244828718187e-07,
      "loss": 1.1165,
      "step": 826
    },
    {
      "epoch": 0.13860722366546552,
      "grad_norm": 0.3224989502777276,
      "learning_rate": 1.6974469465929335e-07,
      "loss": 1.0998,
      "step": 827
    },
    {
      "epoch": 0.13877482611246125,
      "grad_norm": 0.30209997011227957,
      "learning_rate": 1.6966685884238027e-07,
      "loss": 1.0591,
      "step": 828
    },
    {
      "epoch": 0.13894242855945696,
      "grad_norm": 0.2883891602708135,
      "learning_rate": 1.6958894092816633e-07,
      "loss": 1.022,
      "step": 829
    },
    {
      "epoch": 0.1391100310064527,
      "grad_norm": 0.30977325062799077,
      "learning_rate": 1.695109410084722e-07,
      "loss": 1.0613,
      "step": 830
    },
    {
      "epoch": 0.13927763345344843,
      "grad_norm": 0.3179220442773947,
      "learning_rate": 1.6943285917521504e-07,
      "loss": 1.0569,
      "step": 831
    },
    {
      "epoch": 0.13944523590044414,
      "grad_norm": 0.29876745670910115,
      "learning_rate": 1.693546955204086e-07,
      "loss": 1.0105,
      "step": 832
    },
    {
      "epoch": 0.13961283834743987,
      "grad_norm": 0.3166889686338121,
      "learning_rate": 1.6927645013616299e-07,
      "loss": 1.0731,
      "step": 833
    },
    {
      "epoch": 0.1397804407944356,
      "grad_norm": 0.3284259131677626,
      "learning_rate": 1.6919812311468477e-07,
      "loss": 1.0334,
      "step": 834
    },
    {
      "epoch": 0.1399480432414313,
      "grad_norm": 0.3500420755436527,
      "learning_rate": 1.6911971454827648e-07,
      "loss": 1.1347,
      "step": 835
    },
    {
      "epoch": 0.14011564568842705,
      "grad_norm": 0.33885612910954505,
      "learning_rate": 1.69041224529337e-07,
      "loss": 1.0964,
      "step": 836
    },
    {
      "epoch": 0.14028324813542278,
      "grad_norm": 0.3519947945776944,
      "learning_rate": 1.68962653150361e-07,
      "loss": 1.0428,
      "step": 837
    },
    {
      "epoch": 0.1404508505824185,
      "grad_norm": 0.32174363164685305,
      "learning_rate": 1.6888400050393915e-07,
      "loss": 1.0561,
      "step": 838
    },
    {
      "epoch": 0.14061845302941423,
      "grad_norm": 0.3258711949681964,
      "learning_rate": 1.6880526668275784e-07,
      "loss": 1.0241,
      "step": 839
    },
    {
      "epoch": 0.14078605547640996,
      "grad_norm": 0.3252437026515319,
      "learning_rate": 1.6872645177959909e-07,
      "loss": 1.0511,
      "step": 840
    },
    {
      "epoch": 0.1409536579234057,
      "grad_norm": 0.3042682273535404,
      "learning_rate": 1.6864755588734053e-07,
      "loss": 1.0421,
      "step": 841
    },
    {
      "epoch": 0.1411212603704014,
      "grad_norm": 0.33222490712869146,
      "learning_rate": 1.6856857909895522e-07,
      "loss": 1.1011,
      "step": 842
    },
    {
      "epoch": 0.14128886281739714,
      "grad_norm": 0.3295184218092705,
      "learning_rate": 1.6848952150751152e-07,
      "loss": 1.0402,
      "step": 843
    },
    {
      "epoch": 0.14145646526439287,
      "grad_norm": 0.3032979761769794,
      "learning_rate": 1.68410383206173e-07,
      "loss": 1.0314,
      "step": 844
    },
    {
      "epoch": 0.14162406771138858,
      "grad_norm": 0.30067715703299425,
      "learning_rate": 1.6833116428819837e-07,
      "loss": 1.0675,
      "step": 845
    },
    {
      "epoch": 0.14179167015838431,
      "grad_norm": 0.3142593912243732,
      "learning_rate": 1.682518648469414e-07,
      "loss": 1.0735,
      "step": 846
    },
    {
      "epoch": 0.14195927260538005,
      "grad_norm": 0.3116658623540809,
      "learning_rate": 1.6817248497585058e-07,
      "loss": 1.0624,
      "step": 847
    },
    {
      "epoch": 0.14212687505237576,
      "grad_norm": 0.317524709524535,
      "learning_rate": 1.6809302476846935e-07,
      "loss": 1.0853,
      "step": 848
    },
    {
      "epoch": 0.1422944774993715,
      "grad_norm": 0.33016495694258813,
      "learning_rate": 1.6801348431843573e-07,
      "loss": 1.1004,
      "step": 849
    },
    {
      "epoch": 0.14246207994636723,
      "grad_norm": 0.3205694559565464,
      "learning_rate": 1.6793386371948234e-07,
      "loss": 1.075,
      "step": 850
    },
    {
      "epoch": 0.14262968239336293,
      "grad_norm": 0.34678875272347703,
      "learning_rate": 1.6785416306543626e-07,
      "loss": 1.1055,
      "step": 851
    },
    {
      "epoch": 0.14279728484035867,
      "grad_norm": 0.3157229445811686,
      "learning_rate": 1.6777438245021883e-07,
      "loss": 1.0386,
      "step": 852
    },
    {
      "epoch": 0.1429648872873544,
      "grad_norm": 0.37468155738721715,
      "learning_rate": 1.676945219678457e-07,
      "loss": 1.1536,
      "step": 853
    },
    {
      "epoch": 0.1431324897343501,
      "grad_norm": 0.3343958863119367,
      "learning_rate": 1.676145817124266e-07,
      "loss": 1.072,
      "step": 854
    },
    {
      "epoch": 0.14330009218134585,
      "grad_norm": 0.3256478960922459,
      "learning_rate": 1.6753456177816535e-07,
      "loss": 1.0863,
      "step": 855
    },
    {
      "epoch": 0.14346769462834158,
      "grad_norm": 0.2957262556554951,
      "learning_rate": 1.6745446225935947e-07,
      "loss": 1.0136,
      "step": 856
    },
    {
      "epoch": 0.1436352970753373,
      "grad_norm": 0.27506970429427,
      "learning_rate": 1.6737428325040046e-07,
      "loss": 1.0097,
      "step": 857
    },
    {
      "epoch": 0.14380289952233302,
      "grad_norm": 0.28445832901585993,
      "learning_rate": 1.6729402484577343e-07,
      "loss": 0.9982,
      "step": 858
    },
    {
      "epoch": 0.14397050196932876,
      "grad_norm": 0.29021826071619394,
      "learning_rate": 1.6721368714005704e-07,
      "loss": 1.0188,
      "step": 859
    },
    {
      "epoch": 0.1441381044163245,
      "grad_norm": 0.29516926548085504,
      "learning_rate": 1.6713327022792335e-07,
      "loss": 1.0533,
      "step": 860
    },
    {
      "epoch": 0.1443057068633202,
      "grad_norm": 0.2967141670345929,
      "learning_rate": 1.6705277420413789e-07,
      "loss": 1.0331,
      "step": 861
    },
    {
      "epoch": 0.14447330931031593,
      "grad_norm": 0.308919284860164,
      "learning_rate": 1.6697219916355926e-07,
      "loss": 1.0565,
      "step": 862
    },
    {
      "epoch": 0.14464091175731167,
      "grad_norm": 0.3186476216693793,
      "learning_rate": 1.6689154520113938e-07,
      "loss": 1.1011,
      "step": 863
    },
    {
      "epoch": 0.14480851420430738,
      "grad_norm": 0.311710820614857,
      "learning_rate": 1.6681081241192287e-07,
      "loss": 1.0554,
      "step": 864
    },
    {
      "epoch": 0.1449761166513031,
      "grad_norm": 0.3301715985447259,
      "learning_rate": 1.667300008910476e-07,
      "loss": 1.0899,
      "step": 865
    },
    {
      "epoch": 0.14514371909829885,
      "grad_norm": 0.3430324066982608,
      "learning_rate": 1.666491107337439e-07,
      "loss": 1.1159,
      "step": 866
    },
    {
      "epoch": 0.14531132154529455,
      "grad_norm": 0.31285351724519267,
      "learning_rate": 1.66568142035335e-07,
      "loss": 1.0241,
      "step": 867
    },
    {
      "epoch": 0.1454789239922903,
      "grad_norm": 0.2929038277686386,
      "learning_rate": 1.6648709489123655e-07,
      "loss": 1.0022,
      "step": 868
    },
    {
      "epoch": 0.14564652643928602,
      "grad_norm": 0.34385763064418795,
      "learning_rate": 1.664059693969567e-07,
      "loss": 1.0801,
      "step": 869
    },
    {
      "epoch": 0.14581412888628173,
      "grad_norm": 0.28370084143812596,
      "learning_rate": 1.663247656480959e-07,
      "loss": 1.0525,
      "step": 870
    },
    {
      "epoch": 0.14598173133327746,
      "grad_norm": 0.3281317714101836,
      "learning_rate": 1.6624348374034684e-07,
      "loss": 1.0788,
      "step": 871
    },
    {
      "epoch": 0.1461493337802732,
      "grad_norm": 0.3107495688575775,
      "learning_rate": 1.661621237694943e-07,
      "loss": 1.0967,
      "step": 872
    },
    {
      "epoch": 0.1463169362272689,
      "grad_norm": 0.31278773153085504,
      "learning_rate": 1.660806858314151e-07,
      "loss": 1.0444,
      "step": 873
    },
    {
      "epoch": 0.14648453867426464,
      "grad_norm": 0.32559490411630054,
      "learning_rate": 1.6599917002207785e-07,
      "loss": 1.03,
      "step": 874
    },
    {
      "epoch": 0.14665214112126038,
      "grad_norm": 0.34656542281047903,
      "learning_rate": 1.6591757643754298e-07,
      "loss": 1.1519,
      "step": 875
    },
    {
      "epoch": 0.14681974356825608,
      "grad_norm": 0.2957951852705467,
      "learning_rate": 1.6583590517396257e-07,
      "loss": 1.0308,
      "step": 876
    },
    {
      "epoch": 0.14698734601525182,
      "grad_norm": 0.2825128101416877,
      "learning_rate": 1.6575415632758025e-07,
      "loss": 1.0463,
      "step": 877
    },
    {
      "epoch": 0.14715494846224755,
      "grad_norm": 0.3313919154639437,
      "learning_rate": 1.6567232999473104e-07,
      "loss": 1.1147,
      "step": 878
    },
    {
      "epoch": 0.1473225509092433,
      "grad_norm": 0.30216609577826337,
      "learning_rate": 1.655904262718413e-07,
      "loss": 1.0579,
      "step": 879
    },
    {
      "epoch": 0.147490153356239,
      "grad_norm": 0.34580502772934035,
      "learning_rate": 1.6550844525542858e-07,
      "loss": 1.111,
      "step": 880
    },
    {
      "epoch": 0.14765775580323473,
      "grad_norm": 0.291941563361908,
      "learning_rate": 1.6542638704210152e-07,
      "loss": 1.0217,
      "step": 881
    },
    {
      "epoch": 0.14782535825023047,
      "grad_norm": 0.26209806442629663,
      "learning_rate": 1.653442517285597e-07,
      "loss": 1.0201,
      "step": 882
    },
    {
      "epoch": 0.14799296069722617,
      "grad_norm": 0.3377951326110435,
      "learning_rate": 1.6526203941159363e-07,
      "loss": 1.1208,
      "step": 883
    },
    {
      "epoch": 0.1481605631442219,
      "grad_norm": 0.3164560147804531,
      "learning_rate": 1.651797501880845e-07,
      "loss": 1.0502,
      "step": 884
    },
    {
      "epoch": 0.14832816559121764,
      "grad_norm": 0.2965758121653322,
      "learning_rate": 1.6509738415500412e-07,
      "loss": 1.0665,
      "step": 885
    },
    {
      "epoch": 0.14849576803821335,
      "grad_norm": 0.3370158410246872,
      "learning_rate": 1.6501494140941485e-07,
      "loss": 1.0492,
      "step": 886
    },
    {
      "epoch": 0.14866337048520908,
      "grad_norm": 0.2850671862654452,
      "learning_rate": 1.6493242204846945e-07,
      "loss": 1.0167,
      "step": 887
    },
    {
      "epoch": 0.14883097293220482,
      "grad_norm": 0.29195313536875106,
      "learning_rate": 1.6484982616941093e-07,
      "loss": 1.0255,
      "step": 888
    },
    {
      "epoch": 0.14899857537920053,
      "grad_norm": 0.2749754526204745,
      "learning_rate": 1.6476715386957255e-07,
      "loss": 1.0109,
      "step": 889
    },
    {
      "epoch": 0.14916617782619626,
      "grad_norm": 0.35418466592074155,
      "learning_rate": 1.646844052463775e-07,
      "loss": 1.1011,
      "step": 890
    },
    {
      "epoch": 0.149333780273192,
      "grad_norm": 0.31510022699870227,
      "learning_rate": 1.6460158039733907e-07,
      "loss": 1.0414,
      "step": 891
    },
    {
      "epoch": 0.1495013827201877,
      "grad_norm": 0.29637762251466804,
      "learning_rate": 1.645186794200602e-07,
      "loss": 1.1068,
      "step": 892
    },
    {
      "epoch": 0.14966898516718344,
      "grad_norm": 0.3636672874654298,
      "learning_rate": 1.644357024122337e-07,
      "loss": 1.1316,
      "step": 893
    },
    {
      "epoch": 0.14983658761417917,
      "grad_norm": 0.28024828427197046,
      "learning_rate": 1.6435264947164188e-07,
      "loss": 1.0464,
      "step": 894
    },
    {
      "epoch": 0.15000419006117488,
      "grad_norm": 0.3098603593428195,
      "learning_rate": 1.6426952069615653e-07,
      "loss": 1.0763,
      "step": 895
    },
    {
      "epoch": 0.15017179250817062,
      "grad_norm": 0.34419956595143375,
      "learning_rate": 1.641863161837389e-07,
      "loss": 1.0674,
      "step": 896
    },
    {
      "epoch": 0.15033939495516635,
      "grad_norm": 0.31822799023676573,
      "learning_rate": 1.641030360324394e-07,
      "loss": 1.0992,
      "step": 897
    },
    {
      "epoch": 0.15050699740216208,
      "grad_norm": 0.2963399536596775,
      "learning_rate": 1.6401968034039757e-07,
      "loss": 1.0297,
      "step": 898
    },
    {
      "epoch": 0.1506745998491578,
      "grad_norm": 0.30583679277157594,
      "learning_rate": 1.6393624920584208e-07,
      "loss": 1.0057,
      "step": 899
    },
    {
      "epoch": 0.15084220229615353,
      "grad_norm": 0.2738539928686454,
      "learning_rate": 1.638527427270903e-07,
      "loss": 0.9934,
      "step": 900
    },
    {
      "epoch": 0.15100980474314926,
      "grad_norm": 0.3526142104358342,
      "learning_rate": 1.6376916100254863e-07,
      "loss": 1.1022,
      "step": 901
    },
    {
      "epoch": 0.15117740719014497,
      "grad_norm": 0.3228461210034373,
      "learning_rate": 1.6368550413071196e-07,
      "loss": 1.1276,
      "step": 902
    },
    {
      "epoch": 0.1513450096371407,
      "grad_norm": 0.2995116540682773,
      "learning_rate": 1.6360177221016378e-07,
      "loss": 1.0526,
      "step": 903
    },
    {
      "epoch": 0.15151261208413644,
      "grad_norm": 0.3072949615837976,
      "learning_rate": 1.635179653395761e-07,
      "loss": 1.0765,
      "step": 904
    },
    {
      "epoch": 0.15168021453113215,
      "grad_norm": 0.29621020610435034,
      "learning_rate": 1.6343408361770914e-07,
      "loss": 1.0505,
      "step": 905
    },
    {
      "epoch": 0.15184781697812788,
      "grad_norm": 0.3398138349781299,
      "learning_rate": 1.6335012714341141e-07,
      "loss": 1.13,
      "step": 906
    },
    {
      "epoch": 0.15201541942512362,
      "grad_norm": 0.3189651008141494,
      "learning_rate": 1.632660960156195e-07,
      "loss": 1.0978,
      "step": 907
    },
    {
      "epoch": 0.15218302187211932,
      "grad_norm": 0.27409308756214906,
      "learning_rate": 1.6318199033335786e-07,
      "loss": 1.0098,
      "step": 908
    },
    {
      "epoch": 0.15235062431911506,
      "grad_norm": 0.27108673552871637,
      "learning_rate": 1.6309781019573899e-07,
      "loss": 0.9905,
      "step": 909
    },
    {
      "epoch": 0.1525182267661108,
      "grad_norm": 0.29434221798592936,
      "learning_rate": 1.6301355570196294e-07,
      "loss": 1.0867,
      "step": 910
    },
    {
      "epoch": 0.1526858292131065,
      "grad_norm": 0.29966264271561627,
      "learning_rate": 1.6292922695131754e-07,
      "loss": 1.0807,
      "step": 911
    },
    {
      "epoch": 0.15285343166010223,
      "grad_norm": 0.34858228587905604,
      "learning_rate": 1.6284482404317808e-07,
      "loss": 1.1314,
      "step": 912
    },
    {
      "epoch": 0.15302103410709797,
      "grad_norm": 0.28932957495039663,
      "learning_rate": 1.6276034707700712e-07,
      "loss": 1.0498,
      "step": 913
    },
    {
      "epoch": 0.15318863655409368,
      "grad_norm": 0.2969012059768926,
      "learning_rate": 1.6267579615235465e-07,
      "loss": 1.0377,
      "step": 914
    },
    {
      "epoch": 0.1533562390010894,
      "grad_norm": 0.3039557679064843,
      "learning_rate": 1.6259117136885777e-07,
      "loss": 1.0233,
      "step": 915
    },
    {
      "epoch": 0.15352384144808515,
      "grad_norm": 0.32017195563321293,
      "learning_rate": 1.6250647282624057e-07,
      "loss": 1.0872,
      "step": 916
    },
    {
      "epoch": 0.15369144389508088,
      "grad_norm": 0.3300902868758511,
      "learning_rate": 1.6242170062431407e-07,
      "loss": 1.0724,
      "step": 917
    },
    {
      "epoch": 0.1538590463420766,
      "grad_norm": 0.28219864274115986,
      "learning_rate": 1.623368548629762e-07,
      "loss": 1.0507,
      "step": 918
    },
    {
      "epoch": 0.15402664878907232,
      "grad_norm": 0.3071139797026571,
      "learning_rate": 1.622519356422114e-07,
      "loss": 1.0368,
      "step": 919
    },
    {
      "epoch": 0.15419425123606806,
      "grad_norm": 0.3253279571158471,
      "learning_rate": 1.6216694306209083e-07,
      "loss": 1.0918,
      "step": 920
    },
    {
      "epoch": 0.15436185368306377,
      "grad_norm": 0.2866696047398289,
      "learning_rate": 1.6208187722277198e-07,
      "loss": 1.0424,
      "step": 921
    },
    {
      "epoch": 0.1545294561300595,
      "grad_norm": 0.3353380147399203,
      "learning_rate": 1.6199673822449872e-07,
      "loss": 1.0693,
      "step": 922
    },
    {
      "epoch": 0.15469705857705524,
      "grad_norm": 0.3258636335367968,
      "learning_rate": 1.619115261676012e-07,
      "loss": 1.1263,
      "step": 923
    },
    {
      "epoch": 0.15486466102405094,
      "grad_norm": 0.2641023181216732,
      "learning_rate": 1.6182624115249553e-07,
      "loss": 1.0272,
      "step": 924
    },
    {
      "epoch": 0.15503226347104668,
      "grad_norm": 0.29173846251658886,
      "learning_rate": 1.6174088327968392e-07,
      "loss": 1.0556,
      "step": 925
    },
    {
      "epoch": 0.1551998659180424,
      "grad_norm": 0.32663457686802433,
      "learning_rate": 1.6165545264975434e-07,
      "loss": 1.0898,
      "step": 926
    },
    {
      "epoch": 0.15536746836503812,
      "grad_norm": 0.3098657992525131,
      "learning_rate": 1.6156994936338059e-07,
      "loss": 1.094,
      "step": 927
    },
    {
      "epoch": 0.15553507081203385,
      "grad_norm": 0.354957914358628,
      "learning_rate": 1.61484373521322e-07,
      "loss": 1.0994,
      "step": 928
    },
    {
      "epoch": 0.1557026732590296,
      "grad_norm": 0.30121266604032265,
      "learning_rate": 1.6139872522442345e-07,
      "loss": 1.0477,
      "step": 929
    },
    {
      "epoch": 0.1558702757060253,
      "grad_norm": 0.32228335514963097,
      "learning_rate": 1.613130045736152e-07,
      "loss": 1.0935,
      "step": 930
    },
    {
      "epoch": 0.15603787815302103,
      "grad_norm": 0.2881833054508771,
      "learning_rate": 1.6122721166991284e-07,
      "loss": 1.0284,
      "step": 931
    },
    {
      "epoch": 0.15620548060001677,
      "grad_norm": 0.3200907490413666,
      "learning_rate": 1.611413466144169e-07,
      "loss": 1.0431,
      "step": 932
    },
    {
      "epoch": 0.15637308304701247,
      "grad_norm": 0.291030231291893,
      "learning_rate": 1.610554095083132e-07,
      "loss": 1.0555,
      "step": 933
    },
    {
      "epoch": 0.1565406854940082,
      "grad_norm": 0.31081504499178453,
      "learning_rate": 1.609694004528723e-07,
      "loss": 1.0617,
      "step": 934
    },
    {
      "epoch": 0.15670828794100394,
      "grad_norm": 0.3315125034640804,
      "learning_rate": 1.608833195494495e-07,
      "loss": 1.088,
      "step": 935
    },
    {
      "epoch": 0.15687589038799968,
      "grad_norm": 0.26031212440386725,
      "learning_rate": 1.6079716689948496e-07,
      "loss": 0.9703,
      "step": 936
    },
    {
      "epoch": 0.15704349283499539,
      "grad_norm": 0.3088445920782914,
      "learning_rate": 1.6071094260450326e-07,
      "loss": 1.0865,
      "step": 937
    },
    {
      "epoch": 0.15721109528199112,
      "grad_norm": 0.2802022603180517,
      "learning_rate": 1.6062464676611339e-07,
      "loss": 1.0173,
      "step": 938
    },
    {
      "epoch": 0.15737869772898685,
      "grad_norm": 0.3330147850057007,
      "learning_rate": 1.6053827948600872e-07,
      "loss": 1.078,
      "step": 939
    },
    {
      "epoch": 0.15754630017598256,
      "grad_norm": 0.25084536421487663,
      "learning_rate": 1.6045184086596674e-07,
      "loss": 1.0256,
      "step": 940
    },
    {
      "epoch": 0.1577139026229783,
      "grad_norm": 0.2873624064636402,
      "learning_rate": 1.6036533100784911e-07,
      "loss": 1.0275,
      "step": 941
    },
    {
      "epoch": 0.15788150506997403,
      "grad_norm": 0.2359796670194329,
      "learning_rate": 1.6027875001360133e-07,
      "loss": 0.9458,
      "step": 942
    },
    {
      "epoch": 0.15804910751696974,
      "grad_norm": 0.26342549771561796,
      "learning_rate": 1.6019209798525278e-07,
      "loss": 1.0109,
      "step": 943
    },
    {
      "epoch": 0.15821670996396547,
      "grad_norm": 0.2709038274513712,
      "learning_rate": 1.6010537502491653e-07,
      "loss": 0.9862,
      "step": 944
    },
    {
      "epoch": 0.1583843124109612,
      "grad_norm": 0.3631589003613457,
      "learning_rate": 1.6001858123478926e-07,
      "loss": 1.0834,
      "step": 945
    },
    {
      "epoch": 0.15855191485795692,
      "grad_norm": 0.31521183706570327,
      "learning_rate": 1.5993171671715112e-07,
      "loss": 1.0531,
      "step": 946
    },
    {
      "epoch": 0.15871951730495265,
      "grad_norm": 0.3246662269081074,
      "learning_rate": 1.5984478157436556e-07,
      "loss": 1.0909,
      "step": 947
    },
    {
      "epoch": 0.15888711975194839,
      "grad_norm": 0.25831522820507297,
      "learning_rate": 1.5975777590887932e-07,
      "loss": 1.0108,
      "step": 948
    },
    {
      "epoch": 0.1590547221989441,
      "grad_norm": 0.263836296206583,
      "learning_rate": 1.5967069982322224e-07,
      "loss": 1.053,
      "step": 949
    },
    {
      "epoch": 0.15922232464593983,
      "grad_norm": 0.2823451164701269,
      "learning_rate": 1.5958355342000703e-07,
      "loss": 1.0493,
      "step": 950
    },
    {
      "epoch": 0.15938992709293556,
      "grad_norm": 0.2957421199833698,
      "learning_rate": 1.594963368019295e-07,
      "loss": 1.0546,
      "step": 951
    },
    {
      "epoch": 0.15955752953993127,
      "grad_norm": 0.2931265430279669,
      "learning_rate": 1.5940905007176793e-07,
      "loss": 1.0243,
      "step": 952
    },
    {
      "epoch": 0.159725131986927,
      "grad_norm": 0.25950552527326737,
      "learning_rate": 1.5932169333238338e-07,
      "loss": 1.0077,
      "step": 953
    },
    {
      "epoch": 0.15989273443392274,
      "grad_norm": 0.27441633688175415,
      "learning_rate": 1.5923426668671944e-07,
      "loss": 1.0249,
      "step": 954
    },
    {
      "epoch": 0.16006033688091847,
      "grad_norm": 0.3156426402998638,
      "learning_rate": 1.59146770237802e-07,
      "loss": 1.0611,
      "step": 955
    },
    {
      "epoch": 0.16022793932791418,
      "grad_norm": 0.2978134019587279,
      "learning_rate": 1.5905920408873914e-07,
      "loss": 1.0831,
      "step": 956
    },
    {
      "epoch": 0.16039554177490992,
      "grad_norm": 0.3186296011339433,
      "learning_rate": 1.589715683427213e-07,
      "loss": 1.0693,
      "step": 957
    },
    {
      "epoch": 0.16056314422190565,
      "grad_norm": 0.2877841928650721,
      "learning_rate": 1.5888386310302071e-07,
      "loss": 1.0401,
      "step": 958
    },
    {
      "epoch": 0.16073074666890136,
      "grad_norm": 0.2470598717556598,
      "learning_rate": 1.5879608847299163e-07,
      "loss": 0.9553,
      "step": 959
    },
    {
      "epoch": 0.1608983491158971,
      "grad_norm": 0.2987978392644795,
      "learning_rate": 1.5870824455607002e-07,
      "loss": 1.0588,
      "step": 960
    },
    {
      "epoch": 0.16106595156289283,
      "grad_norm": 0.2285736807090564,
      "learning_rate": 1.586203314557735e-07,
      "loss": 0.9029,
      "step": 961
    },
    {
      "epoch": 0.16123355400988854,
      "grad_norm": 0.27999878766780467,
      "learning_rate": 1.5853234927570129e-07,
      "loss": 1.0252,
      "step": 962
    },
    {
      "epoch": 0.16140115645688427,
      "grad_norm": 0.2801711923274267,
      "learning_rate": 1.5844429811953393e-07,
      "loss": 0.996,
      "step": 963
    },
    {
      "epoch": 0.16156875890388,
      "grad_norm": 0.3035864113423551,
      "learning_rate": 1.5835617809103323e-07,
      "loss": 1.0653,
      "step": 964
    },
    {
      "epoch": 0.1617363613508757,
      "grad_norm": 0.2831216758605725,
      "learning_rate": 1.5826798929404226e-07,
      "loss": 1.0186,
      "step": 965
    },
    {
      "epoch": 0.16190396379787145,
      "grad_norm": 0.28075406626187543,
      "learning_rate": 1.5817973183248503e-07,
      "loss": 1.0432,
      "step": 966
    },
    {
      "epoch": 0.16207156624486718,
      "grad_norm": 0.2708041914160719,
      "learning_rate": 1.5809140581036658e-07,
      "loss": 1.0795,
      "step": 967
    },
    {
      "epoch": 0.1622391686918629,
      "grad_norm": 0.264119558896937,
      "learning_rate": 1.5800301133177256e-07,
      "loss": 1.0015,
      "step": 968
    },
    {
      "epoch": 0.16240677113885862,
      "grad_norm": 0.2931121712032405,
      "learning_rate": 1.579145485008695e-07,
      "loss": 1.0826,
      "step": 969
    },
    {
      "epoch": 0.16257437358585436,
      "grad_norm": 0.30379067490284056,
      "learning_rate": 1.5782601742190431e-07,
      "loss": 1.0748,
      "step": 970
    },
    {
      "epoch": 0.16274197603285007,
      "grad_norm": 0.3056805305490091,
      "learning_rate": 1.5773741819920443e-07,
      "loss": 1.0611,
      "step": 971
    },
    {
      "epoch": 0.1629095784798458,
      "grad_norm": 0.2769666237953121,
      "learning_rate": 1.5764875093717756e-07,
      "loss": 1.0183,
      "step": 972
    },
    {
      "epoch": 0.16307718092684154,
      "grad_norm": 0.310274972470976,
      "learning_rate": 1.575600157403116e-07,
      "loss": 1.0969,
      "step": 973
    },
    {
      "epoch": 0.16324478337383727,
      "grad_norm": 0.3053474884509601,
      "learning_rate": 1.5747121271317445e-07,
      "loss": 1.0134,
      "step": 974
    },
    {
      "epoch": 0.16341238582083298,
      "grad_norm": 0.24712268591332956,
      "learning_rate": 1.5738234196041407e-07,
      "loss": 0.9673,
      "step": 975
    },
    {
      "epoch": 0.1635799882678287,
      "grad_norm": 0.29742621537467595,
      "learning_rate": 1.5729340358675808e-07,
      "loss": 1.0469,
      "step": 976
    },
    {
      "epoch": 0.16374759071482445,
      "grad_norm": 0.2826205594224282,
      "learning_rate": 1.5720439769701382e-07,
      "loss": 1.0696,
      "step": 977
    },
    {
      "epoch": 0.16391519316182016,
      "grad_norm": 0.2644256794424601,
      "learning_rate": 1.5711532439606834e-07,
      "loss": 1.0291,
      "step": 978
    },
    {
      "epoch": 0.1640827956088159,
      "grad_norm": 0.31609370993042,
      "learning_rate": 1.570261837888879e-07,
      "loss": 1.0828,
      "step": 979
    },
    {
      "epoch": 0.16425039805581162,
      "grad_norm": 0.2586421205077691,
      "learning_rate": 1.5693697598051826e-07,
      "loss": 1.0555,
      "step": 980
    },
    {
      "epoch": 0.16441800050280733,
      "grad_norm": 0.2863383002424727,
      "learning_rate": 1.5684770107608426e-07,
      "loss": 1.0365,
      "step": 981
    },
    {
      "epoch": 0.16458560294980307,
      "grad_norm": 0.2689674024203239,
      "learning_rate": 1.5675835918078988e-07,
      "loss": 1.0017,
      "step": 982
    },
    {
      "epoch": 0.1647532053967988,
      "grad_norm": 0.30151609663862355,
      "learning_rate": 1.56668950399918e-07,
      "loss": 1.0857,
      "step": 983
    },
    {
      "epoch": 0.1649208078437945,
      "grad_norm": 0.27999296001672086,
      "learning_rate": 1.5657947483883032e-07,
      "loss": 1.022,
      "step": 984
    },
    {
      "epoch": 0.16508841029079024,
      "grad_norm": 0.27292709077950805,
      "learning_rate": 1.5648993260296727e-07,
      "loss": 1.0425,
      "step": 985
    },
    {
      "epoch": 0.16525601273778598,
      "grad_norm": 0.2540175397833907,
      "learning_rate": 1.5640032379784781e-07,
      "loss": 1.0199,
      "step": 986
    },
    {
      "epoch": 0.16542361518478169,
      "grad_norm": 0.31143117955437577,
      "learning_rate": 1.563106485290694e-07,
      "loss": 1.0901,
      "step": 987
    },
    {
      "epoch": 0.16559121763177742,
      "grad_norm": 0.24934550056209842,
      "learning_rate": 1.5622090690230776e-07,
      "loss": 1.011,
      "step": 988
    },
    {
      "epoch": 0.16575882007877316,
      "grad_norm": 0.2893437902343497,
      "learning_rate": 1.5613109902331686e-07,
      "loss": 1.0248,
      "step": 989
    },
    {
      "epoch": 0.16592642252576886,
      "grad_norm": 0.28096669991236417,
      "learning_rate": 1.5604122499792867e-07,
      "loss": 1.0261,
      "step": 990
    },
    {
      "epoch": 0.1660940249727646,
      "grad_norm": 0.27345261007914995,
      "learning_rate": 1.5595128493205323e-07,
      "loss": 1.0364,
      "step": 991
    },
    {
      "epoch": 0.16626162741976033,
      "grad_norm": 0.28356078334570567,
      "learning_rate": 1.5586127893167827e-07,
      "loss": 1.0783,
      "step": 992
    },
    {
      "epoch": 0.16642922986675607,
      "grad_norm": 0.2635134335894332,
      "learning_rate": 1.557712071028693e-07,
      "loss": 1.0273,
      "step": 993
    },
    {
      "epoch": 0.16659683231375177,
      "grad_norm": 0.3190335174108025,
      "learning_rate": 1.5568106955176945e-07,
      "loss": 1.0869,
      "step": 994
    },
    {
      "epoch": 0.1667644347607475,
      "grad_norm": 0.2726771526954252,
      "learning_rate": 1.5559086638459914e-07,
      "loss": 1.0417,
      "step": 995
    },
    {
      "epoch": 0.16693203720774324,
      "grad_norm": 0.2803317706274094,
      "learning_rate": 1.5550059770765627e-07,
      "loss": 1.0463,
      "step": 996
    },
    {
      "epoch": 0.16709963965473895,
      "grad_norm": 0.2844076793644102,
      "learning_rate": 1.5541026362731584e-07,
      "loss": 1.0697,
      "step": 997
    },
    {
      "epoch": 0.1672672421017347,
      "grad_norm": 0.28541857207240906,
      "learning_rate": 1.5531986425003002e-07,
      "loss": 1.0482,
      "step": 998
    },
    {
      "epoch": 0.16743484454873042,
      "grad_norm": 0.2812860369738872,
      "learning_rate": 1.5522939968232777e-07,
      "loss": 1.0536,
      "step": 999
    },
    {
      "epoch": 0.16760244699572613,
      "grad_norm": 0.2832244813845103,
      "learning_rate": 1.5513887003081506e-07,
      "loss": 1.024,
      "step": 1000
    },
    {
      "epoch": 0.16777004944272186,
      "grad_norm": 0.27633406521881915,
      "learning_rate": 1.5504827540217444e-07,
      "loss": 1.0213,
      "step": 1001
    },
    {
      "epoch": 0.1679376518897176,
      "grad_norm": 0.29766838392508993,
      "learning_rate": 1.5495761590316506e-07,
      "loss": 1.0759,
      "step": 1002
    },
    {
      "epoch": 0.1681052543367133,
      "grad_norm": 0.28868737801863437,
      "learning_rate": 1.548668916406225e-07,
      "loss": 1.0216,
      "step": 1003
    },
    {
      "epoch": 0.16827285678370904,
      "grad_norm": 0.25951171000906825,
      "learning_rate": 1.5477610272145868e-07,
      "loss": 1.0189,
      "step": 1004
    },
    {
      "epoch": 0.16844045923070478,
      "grad_norm": 0.22798585451155878,
      "learning_rate": 1.5468524925266168e-07,
      "loss": 0.9309,
      "step": 1005
    },
    {
      "epoch": 0.16860806167770048,
      "grad_norm": 0.25828816808455535,
      "learning_rate": 1.5459433134129572e-07,
      "loss": 1.0479,
      "step": 1006
    },
    {
      "epoch": 0.16877566412469622,
      "grad_norm": 0.2607470485726105,
      "learning_rate": 1.5450334909450086e-07,
      "loss": 1.0353,
      "step": 1007
    },
    {
      "epoch": 0.16894326657169195,
      "grad_norm": 0.2624312318496608,
      "learning_rate": 1.5441230261949306e-07,
      "loss": 0.9834,
      "step": 1008
    },
    {
      "epoch": 0.16911086901868766,
      "grad_norm": 0.2355085997760303,
      "learning_rate": 1.5432119202356396e-07,
      "loss": 0.9525,
      "step": 1009
    },
    {
      "epoch": 0.1692784714656834,
      "grad_norm": 0.24998916440392552,
      "learning_rate": 1.542300174140807e-07,
      "loss": 1.0238,
      "step": 1010
    },
    {
      "epoch": 0.16944607391267913,
      "grad_norm": 0.2714446767912772,
      "learning_rate": 1.5413877889848587e-07,
      "loss": 1.045,
      "step": 1011
    },
    {
      "epoch": 0.16961367635967486,
      "grad_norm": 0.2765349653683857,
      "learning_rate": 1.5404747658429744e-07,
      "loss": 1.0771,
      "step": 1012
    },
    {
      "epoch": 0.16978127880667057,
      "grad_norm": 0.273030468065554,
      "learning_rate": 1.5395611057910852e-07,
      "loss": 1.0477,
      "step": 1013
    },
    {
      "epoch": 0.1699488812536663,
      "grad_norm": 0.2387297296083996,
      "learning_rate": 1.5386468099058723e-07,
      "loss": 0.9471,
      "step": 1014
    },
    {
      "epoch": 0.17011648370066204,
      "grad_norm": 0.25959935955042646,
      "learning_rate": 1.5377318792647667e-07,
      "loss": 1.0321,
      "step": 1015
    },
    {
      "epoch": 0.17028408614765775,
      "grad_norm": 0.2675273739798068,
      "learning_rate": 1.5368163149459474e-07,
      "loss": 1.0352,
      "step": 1016
    },
    {
      "epoch": 0.17045168859465348,
      "grad_norm": 0.24167167718328716,
      "learning_rate": 1.5359001180283398e-07,
      "loss": 0.9937,
      "step": 1017
    },
    {
      "epoch": 0.17061929104164922,
      "grad_norm": 0.2832564198814399,
      "learning_rate": 1.5349832895916152e-07,
      "loss": 1.0218,
      "step": 1018
    },
    {
      "epoch": 0.17078689348864493,
      "grad_norm": 0.2835442333722525,
      "learning_rate": 1.5340658307161885e-07,
      "loss": 1.0421,
      "step": 1019
    },
    {
      "epoch": 0.17095449593564066,
      "grad_norm": 0.26637422650788356,
      "learning_rate": 1.5331477424832186e-07,
      "loss": 1.0385,
      "step": 1020
    },
    {
      "epoch": 0.1711220983826364,
      "grad_norm": 0.2751486272491116,
      "learning_rate": 1.532229025974605e-07,
      "loss": 1.0538,
      "step": 1021
    },
    {
      "epoch": 0.1712897008296321,
      "grad_norm": 0.24237974746821028,
      "learning_rate": 1.5313096822729876e-07,
      "loss": 1.0196,
      "step": 1022
    },
    {
      "epoch": 0.17145730327662784,
      "grad_norm": 0.23888699922330472,
      "learning_rate": 1.5303897124617467e-07,
      "loss": 0.9579,
      "step": 1023
    },
    {
      "epoch": 0.17162490572362357,
      "grad_norm": 0.2589740843291592,
      "learning_rate": 1.5294691176249982e-07,
      "loss": 1.0381,
      "step": 1024
    },
    {
      "epoch": 0.17179250817061928,
      "grad_norm": 0.3006210946098925,
      "learning_rate": 1.528547898847597e-07,
      "loss": 1.0467,
      "step": 1025
    },
    {
      "epoch": 0.17196011061761501,
      "grad_norm": 0.27299185516428215,
      "learning_rate": 1.5276260572151317e-07,
      "loss": 1.026,
      "step": 1026
    },
    {
      "epoch": 0.17212771306461075,
      "grad_norm": 0.27012647483839514,
      "learning_rate": 1.526703593813925e-07,
      "loss": 1.0369,
      "step": 1027
    },
    {
      "epoch": 0.17229531551160646,
      "grad_norm": 0.2819441987067369,
      "learning_rate": 1.5257805097310327e-07,
      "loss": 1.1208,
      "step": 1028
    },
    {
      "epoch": 0.1724629179586022,
      "grad_norm": 0.28590137893234846,
      "learning_rate": 1.5248568060542422e-07,
      "loss": 1.0394,
      "step": 1029
    },
    {
      "epoch": 0.17263052040559793,
      "grad_norm": 0.2804558788927519,
      "learning_rate": 1.5239324838720704e-07,
      "loss": 1.0457,
      "step": 1030
    },
    {
      "epoch": 0.17279812285259366,
      "grad_norm": 0.22213833769098085,
      "learning_rate": 1.5230075442737637e-07,
      "loss": 0.9565,
      "step": 1031
    },
    {
      "epoch": 0.17296572529958937,
      "grad_norm": 0.2781655637298193,
      "learning_rate": 1.5220819883492955e-07,
      "loss": 1.0784,
      "step": 1032
    },
    {
      "epoch": 0.1731333277465851,
      "grad_norm": 0.2878268495922409,
      "learning_rate": 1.521155817189366e-07,
      "loss": 1.0699,
      "step": 1033
    },
    {
      "epoch": 0.17330093019358084,
      "grad_norm": 0.20698359182712298,
      "learning_rate": 1.5202290318854006e-07,
      "loss": 0.9281,
      "step": 1034
    },
    {
      "epoch": 0.17346853264057654,
      "grad_norm": 0.26909283081641505,
      "learning_rate": 1.5193016335295473e-07,
      "loss": 1.0598,
      "step": 1035
    },
    {
      "epoch": 0.17363613508757228,
      "grad_norm": 0.26638518759147894,
      "learning_rate": 1.518373623214678e-07,
      "loss": 1.0323,
      "step": 1036
    },
    {
      "epoch": 0.17380373753456801,
      "grad_norm": 0.2548805883645317,
      "learning_rate": 1.5174450020343841e-07,
      "loss": 1.0014,
      "step": 1037
    },
    {
      "epoch": 0.17397133998156372,
      "grad_norm": 0.27459415610375,
      "learning_rate": 1.516515771082979e-07,
      "loss": 1.0485,
      "step": 1038
    },
    {
      "epoch": 0.17413894242855946,
      "grad_norm": 0.27673603831290294,
      "learning_rate": 1.515585931455492e-07,
      "loss": 1.0153,
      "step": 1039
    },
    {
      "epoch": 0.1743065448755552,
      "grad_norm": 0.255931999737098,
      "learning_rate": 1.5146554842476724e-07,
      "loss": 1.049,
      "step": 1040
    },
    {
      "epoch": 0.1744741473225509,
      "grad_norm": 0.2611744556173021,
      "learning_rate": 1.5137244305559838e-07,
      "loss": 1.0192,
      "step": 1041
    },
    {
      "epoch": 0.17464174976954663,
      "grad_norm": 0.2613706932629094,
      "learning_rate": 1.5127927714776046e-07,
      "loss": 1.0383,
      "step": 1042
    },
    {
      "epoch": 0.17480935221654237,
      "grad_norm": 0.27020654720542053,
      "learning_rate": 1.5118605081104274e-07,
      "loss": 1.0182,
      "step": 1043
    },
    {
      "epoch": 0.17497695466353808,
      "grad_norm": 0.2659508394529833,
      "learning_rate": 1.5109276415530561e-07,
      "loss": 0.9948,
      "step": 1044
    },
    {
      "epoch": 0.1751445571105338,
      "grad_norm": 0.26920580983411274,
      "learning_rate": 1.5099941729048057e-07,
      "loss": 1.0157,
      "step": 1045
    },
    {
      "epoch": 0.17531215955752955,
      "grad_norm": 0.2511636222748831,
      "learning_rate": 1.5090601032657016e-07,
      "loss": 1.0016,
      "step": 1046
    },
    {
      "epoch": 0.17547976200452525,
      "grad_norm": 0.2252297078355041,
      "learning_rate": 1.5081254337364754e-07,
      "loss": 0.9765,
      "step": 1047
    },
    {
      "epoch": 0.175647364451521,
      "grad_norm": 0.25000575001712205,
      "learning_rate": 1.5071901654185674e-07,
      "loss": 1.0177,
      "step": 1048
    },
    {
      "epoch": 0.17581496689851672,
      "grad_norm": 0.2963833172837157,
      "learning_rate": 1.506254299414123e-07,
      "loss": 1.0947,
      "step": 1049
    },
    {
      "epoch": 0.17598256934551246,
      "grad_norm": 0.2237887419424654,
      "learning_rate": 1.5053178368259912e-07,
      "loss": 0.9319,
      "step": 1050
    },
    {
      "epoch": 0.17615017179250816,
      "grad_norm": 0.23129497081475286,
      "learning_rate": 1.5043807787577262e-07,
      "loss": 0.9663,
      "step": 1051
    },
    {
      "epoch": 0.1763177742395039,
      "grad_norm": 0.2285972989850153,
      "learning_rate": 1.503443126313581e-07,
      "loss": 0.929,
      "step": 1052
    },
    {
      "epoch": 0.17648537668649963,
      "grad_norm": 0.2823538985237284,
      "learning_rate": 1.502504880598511e-07,
      "loss": 1.0117,
      "step": 1053
    },
    {
      "epoch": 0.17665297913349534,
      "grad_norm": 0.2670370870611528,
      "learning_rate": 1.5015660427181703e-07,
      "loss": 1.0284,
      "step": 1054
    },
    {
      "epoch": 0.17682058158049108,
      "grad_norm": 0.2912167806837974,
      "learning_rate": 1.5006266137789107e-07,
      "loss": 1.0558,
      "step": 1055
    },
    {
      "epoch": 0.1769881840274868,
      "grad_norm": 0.2542399522039058,
      "learning_rate": 1.4996865948877802e-07,
      "loss": 1.0156,
      "step": 1056
    },
    {
      "epoch": 0.17715578647448252,
      "grad_norm": 0.2794999048157347,
      "learning_rate": 1.4987459871525228e-07,
      "loss": 1.0435,
      "step": 1057
    },
    {
      "epoch": 0.17732338892147825,
      "grad_norm": 0.2983956425860928,
      "learning_rate": 1.4978047916815756e-07,
      "loss": 1.0669,
      "step": 1058
    },
    {
      "epoch": 0.177490991368474,
      "grad_norm": 0.2733518519505484,
      "learning_rate": 1.4968630095840688e-07,
      "loss": 0.9987,
      "step": 1059
    },
    {
      "epoch": 0.1776585938154697,
      "grad_norm": 0.24999463283725823,
      "learning_rate": 1.495920641969824e-07,
      "loss": 0.9979,
      "step": 1060
    },
    {
      "epoch": 0.17782619626246543,
      "grad_norm": 0.23959393139470128,
      "learning_rate": 1.494977689949352e-07,
      "loss": 0.9984,
      "step": 1061
    },
    {
      "epoch": 0.17799379870946117,
      "grad_norm": 0.2652546315682368,
      "learning_rate": 1.4940341546338536e-07,
      "loss": 1.0786,
      "step": 1062
    },
    {
      "epoch": 0.17816140115645687,
      "grad_norm": 0.26325427906499665,
      "learning_rate": 1.4930900371352156e-07,
      "loss": 1.0341,
      "step": 1063
    },
    {
      "epoch": 0.1783290036034526,
      "grad_norm": 0.2458009074066654,
      "learning_rate": 1.4921453385660111e-07,
      "loss": 1.0046,
      "step": 1064
    },
    {
      "epoch": 0.17849660605044834,
      "grad_norm": 0.23864333713668254,
      "learning_rate": 1.4912000600394995e-07,
      "loss": 1.0004,
      "step": 1065
    },
    {
      "epoch": 0.17866420849744405,
      "grad_norm": 0.2626804122574614,
      "learning_rate": 1.4902542026696214e-07,
      "loss": 1.0184,
      "step": 1066
    },
    {
      "epoch": 0.17883181094443978,
      "grad_norm": 0.26118417563589985,
      "learning_rate": 1.489307767571001e-07,
      "loss": 1.0484,
      "step": 1067
    },
    {
      "epoch": 0.17899941339143552,
      "grad_norm": 0.2544374586836278,
      "learning_rate": 1.488360755858943e-07,
      "loss": 1.026,
      "step": 1068
    },
    {
      "epoch": 0.17916701583843125,
      "grad_norm": 0.23871586549263565,
      "learning_rate": 1.4874131686494314e-07,
      "loss": 0.9361,
      "step": 1069
    },
    {
      "epoch": 0.17933461828542696,
      "grad_norm": 0.23580017921253843,
      "learning_rate": 1.486465007059128e-07,
      "loss": 0.9948,
      "step": 1070
    },
    {
      "epoch": 0.1795022207324227,
      "grad_norm": 0.2480974938588441,
      "learning_rate": 1.4855162722053725e-07,
      "loss": 1.0242,
      "step": 1071
    },
    {
      "epoch": 0.17966982317941843,
      "grad_norm": 0.2518013398551965,
      "learning_rate": 1.4845669652061795e-07,
      "loss": 1.0089,
      "step": 1072
    },
    {
      "epoch": 0.17983742562641414,
      "grad_norm": 0.20547334091468286,
      "learning_rate": 1.4836170871802376e-07,
      "loss": 0.9563,
      "step": 1073
    },
    {
      "epoch": 0.18000502807340987,
      "grad_norm": 0.2724869549544852,
      "learning_rate": 1.4826666392469087e-07,
      "loss": 1.0204,
      "step": 1074
    },
    {
      "epoch": 0.1801726305204056,
      "grad_norm": 0.2799944797095893,
      "learning_rate": 1.481715622526227e-07,
      "loss": 0.9867,
      "step": 1075
    },
    {
      "epoch": 0.18034023296740131,
      "grad_norm": 0.28763108080482414,
      "learning_rate": 1.480764038138895e-07,
      "loss": 1.0484,
      "step": 1076
    },
    {
      "epoch": 0.18050783541439705,
      "grad_norm": 0.2395959707909104,
      "learning_rate": 1.4798118872062855e-07,
      "loss": 0.9559,
      "step": 1077
    },
    {
      "epoch": 0.18067543786139278,
      "grad_norm": 0.2878530521571105,
      "learning_rate": 1.4788591708504398e-07,
      "loss": 1.0527,
      "step": 1078
    },
    {
      "epoch": 0.1808430403083885,
      "grad_norm": 0.2631214633543428,
      "learning_rate": 1.4779058901940634e-07,
      "loss": 0.9917,
      "step": 1079
    },
    {
      "epoch": 0.18101064275538423,
      "grad_norm": 0.24564795623923621,
      "learning_rate": 1.4769520463605281e-07,
      "loss": 1.0064,
      "step": 1080
    },
    {
      "epoch": 0.18117824520237996,
      "grad_norm": 0.2708066966665084,
      "learning_rate": 1.4759976404738698e-07,
      "loss": 1.0377,
      "step": 1081
    },
    {
      "epoch": 0.18134584764937567,
      "grad_norm": 0.2501881970548862,
      "learning_rate": 1.4750426736587857e-07,
      "loss": 0.9943,
      "step": 1082
    },
    {
      "epoch": 0.1815134500963714,
      "grad_norm": 0.23468217558100948,
      "learning_rate": 1.474087147040634e-07,
      "loss": 0.9849,
      "step": 1083
    },
    {
      "epoch": 0.18168105254336714,
      "grad_norm": 0.27713324777226267,
      "learning_rate": 1.4731310617454335e-07,
      "loss": 1.0423,
      "step": 1084
    },
    {
      "epoch": 0.18184865499036285,
      "grad_norm": 0.26916007950525805,
      "learning_rate": 1.4721744188998604e-07,
      "loss": 1.0323,
      "step": 1085
    },
    {
      "epoch": 0.18201625743735858,
      "grad_norm": 0.2544889640954427,
      "learning_rate": 1.4712172196312485e-07,
      "loss": 0.9962,
      "step": 1086
    },
    {
      "epoch": 0.18218385988435432,
      "grad_norm": 0.29619227215709737,
      "learning_rate": 1.4702594650675871e-07,
      "loss": 1.0779,
      "step": 1087
    },
    {
      "epoch": 0.18235146233135005,
      "grad_norm": 0.26972151801626837,
      "learning_rate": 1.4693011563375206e-07,
      "loss": 1.0123,
      "step": 1088
    },
    {
      "epoch": 0.18251906477834576,
      "grad_norm": 0.2690904256191142,
      "learning_rate": 1.468342294570345e-07,
      "loss": 1.0186,
      "step": 1089
    },
    {
      "epoch": 0.1826866672253415,
      "grad_norm": 0.22156594102301178,
      "learning_rate": 1.467382880896009e-07,
      "loss": 0.9851,
      "step": 1090
    },
    {
      "epoch": 0.18285426967233723,
      "grad_norm": 0.26752061840684743,
      "learning_rate": 1.4664229164451118e-07,
      "loss": 1.0457,
      "step": 1091
    },
    {
      "epoch": 0.18302187211933293,
      "grad_norm": 0.2612119744381757,
      "learning_rate": 1.465462402348901e-07,
      "loss": 1.0267,
      "step": 1092
    },
    {
      "epoch": 0.18318947456632867,
      "grad_norm": 0.24463056416918802,
      "learning_rate": 1.4645013397392721e-07,
      "loss": 1.0077,
      "step": 1093
    },
    {
      "epoch": 0.1833570770133244,
      "grad_norm": 0.28007441848430437,
      "learning_rate": 1.4635397297487676e-07,
      "loss": 1.0275,
      "step": 1094
    },
    {
      "epoch": 0.1835246794603201,
      "grad_norm": 0.26828686072417063,
      "learning_rate": 1.4625775735105744e-07,
      "loss": 0.995,
      "step": 1095
    },
    {
      "epoch": 0.18369228190731585,
      "grad_norm": 0.26199553644561757,
      "learning_rate": 1.4616148721585232e-07,
      "loss": 0.9965,
      "step": 1096
    },
    {
      "epoch": 0.18385988435431158,
      "grad_norm": 0.23496710369965765,
      "learning_rate": 1.4606516268270872e-07,
      "loss": 0.9881,
      "step": 1097
    },
    {
      "epoch": 0.1840274868013073,
      "grad_norm": 0.2565175484215671,
      "learning_rate": 1.4596878386513807e-07,
      "loss": 1.0456,
      "step": 1098
    },
    {
      "epoch": 0.18419508924830302,
      "grad_norm": 0.2416757780426304,
      "learning_rate": 1.4587235087671576e-07,
      "loss": 0.9935,
      "step": 1099
    },
    {
      "epoch": 0.18436269169529876,
      "grad_norm": 0.2653592189105421,
      "learning_rate": 1.4577586383108101e-07,
      "loss": 1.0789,
      "step": 1100
    },
    {
      "epoch": 0.18453029414229447,
      "grad_norm": 0.25566657407203264,
      "learning_rate": 1.4567932284193677e-07,
      "loss": 1.0251,
      "step": 1101
    },
    {
      "epoch": 0.1846978965892902,
      "grad_norm": 0.24136632326407534,
      "learning_rate": 1.4558272802304952e-07,
      "loss": 0.9865,
      "step": 1102
    },
    {
      "epoch": 0.18486549903628594,
      "grad_norm": 0.2637162293353835,
      "learning_rate": 1.4548607948824917e-07,
      "loss": 1.0353,
      "step": 1103
    },
    {
      "epoch": 0.18503310148328164,
      "grad_norm": 0.24844178837381548,
      "learning_rate": 1.4538937735142904e-07,
      "loss": 1.01,
      "step": 1104
    },
    {
      "epoch": 0.18520070393027738,
      "grad_norm": 0.22492957218595588,
      "learning_rate": 1.4529262172654543e-07,
      "loss": 0.9774,
      "step": 1105
    },
    {
      "epoch": 0.1853683063772731,
      "grad_norm": 0.2717220195962178,
      "learning_rate": 1.4519581272761782e-07,
      "loss": 1.0516,
      "step": 1106
    },
    {
      "epoch": 0.18553590882426885,
      "grad_norm": 0.235977865064619,
      "learning_rate": 1.4509895046872852e-07,
      "loss": 0.9928,
      "step": 1107
    },
    {
      "epoch": 0.18570351127126455,
      "grad_norm": 0.25226820444282233,
      "learning_rate": 1.4500203506402265e-07,
      "loss": 0.962,
      "step": 1108
    },
    {
      "epoch": 0.1858711137182603,
      "grad_norm": 0.2713321910868741,
      "learning_rate": 1.4490506662770794e-07,
      "loss": 1.0767,
      "step": 1109
    },
    {
      "epoch": 0.18603871616525602,
      "grad_norm": 0.22352584549567095,
      "learning_rate": 1.4480804527405456e-07,
      "loss": 0.9782,
      "step": 1110
    },
    {
      "epoch": 0.18620631861225173,
      "grad_norm": 0.2245759408305602,
      "learning_rate": 1.4471097111739507e-07,
      "loss": 0.955,
      "step": 1111
    },
    {
      "epoch": 0.18637392105924747,
      "grad_norm": 0.25754939731615667,
      "learning_rate": 1.4461384427212433e-07,
      "loss": 0.9869,
      "step": 1112
    },
    {
      "epoch": 0.1865415235062432,
      "grad_norm": 0.25019132604214167,
      "learning_rate": 1.445166648526992e-07,
      "loss": 1.0029,
      "step": 1113
    },
    {
      "epoch": 0.1867091259532389,
      "grad_norm": 0.2577580867419062,
      "learning_rate": 1.444194329736385e-07,
      "loss": 0.974,
      "step": 1114
    },
    {
      "epoch": 0.18687672840023464,
      "grad_norm": 0.2530094521227813,
      "learning_rate": 1.4432214874952293e-07,
      "loss": 1.0132,
      "step": 1115
    },
    {
      "epoch": 0.18704433084723038,
      "grad_norm": 0.2643250602021817,
      "learning_rate": 1.442248122949948e-07,
      "loss": 0.9881,
      "step": 1116
    },
    {
      "epoch": 0.18721193329422608,
      "grad_norm": 0.23775876893397455,
      "learning_rate": 1.4412742372475805e-07,
      "loss": 0.9834,
      "step": 1117
    },
    {
      "epoch": 0.18737953574122182,
      "grad_norm": 0.24487387070649283,
      "learning_rate": 1.4402998315357797e-07,
      "loss": 1.0094,
      "step": 1118
    },
    {
      "epoch": 0.18754713818821755,
      "grad_norm": 0.27221098497752383,
      "learning_rate": 1.4393249069628112e-07,
      "loss": 1.0517,
      "step": 1119
    },
    {
      "epoch": 0.18771474063521326,
      "grad_norm": 0.2900350162811489,
      "learning_rate": 1.4383494646775526e-07,
      "loss": 1.0573,
      "step": 1120
    },
    {
      "epoch": 0.187882343082209,
      "grad_norm": 0.26778978508142015,
      "learning_rate": 1.4373735058294918e-07,
      "loss": 0.9827,
      "step": 1121
    },
    {
      "epoch": 0.18804994552920473,
      "grad_norm": 0.2354308626614222,
      "learning_rate": 1.436397031568724e-07,
      "loss": 0.9622,
      "step": 1122
    },
    {
      "epoch": 0.18821754797620044,
      "grad_norm": 0.2510879350060905,
      "learning_rate": 1.4354200430459534e-07,
      "loss": 0.9868,
      "step": 1123
    },
    {
      "epoch": 0.18838515042319617,
      "grad_norm": 0.2474473736959406,
      "learning_rate": 1.4344425414124896e-07,
      "loss": 0.9963,
      "step": 1124
    },
    {
      "epoch": 0.1885527528701919,
      "grad_norm": 0.2687826307237104,
      "learning_rate": 1.433464527820247e-07,
      "loss": 1.0044,
      "step": 1125
    },
    {
      "epoch": 0.18872035531718764,
      "grad_norm": 0.24518827818119004,
      "learning_rate": 1.432486003421742e-07,
      "loss": 0.9787,
      "step": 1126
    },
    {
      "epoch": 0.18888795776418335,
      "grad_norm": 0.25378639137658454,
      "learning_rate": 1.4315069693700954e-07,
      "loss": 0.9632,
      "step": 1127
    },
    {
      "epoch": 0.18905556021117909,
      "grad_norm": 0.24246671027594144,
      "learning_rate": 1.4305274268190265e-07,
      "loss": 0.9879,
      "step": 1128
    },
    {
      "epoch": 0.18922316265817482,
      "grad_norm": 0.2648352353087182,
      "learning_rate": 1.4295473769228545e-07,
      "loss": 0.9924,
      "step": 1129
    },
    {
      "epoch": 0.18939076510517053,
      "grad_norm": 0.24780773203717021,
      "learning_rate": 1.4285668208364975e-07,
      "loss": 0.9785,
      "step": 1130
    },
    {
      "epoch": 0.18955836755216626,
      "grad_norm": 0.22838764358455846,
      "learning_rate": 1.4275857597154683e-07,
      "loss": 0.9232,
      "step": 1131
    },
    {
      "epoch": 0.189725969999162,
      "grad_norm": 0.2482787451403922,
      "learning_rate": 1.426604194715876e-07,
      "loss": 0.967,
      "step": 1132
    },
    {
      "epoch": 0.1898935724461577,
      "grad_norm": 0.24437034522663417,
      "learning_rate": 1.425622126994423e-07,
      "loss": 1.0078,
      "step": 1133
    },
    {
      "epoch": 0.19006117489315344,
      "grad_norm": 0.23740214581120506,
      "learning_rate": 1.424639557708405e-07,
      "loss": 0.9761,
      "step": 1134
    },
    {
      "epoch": 0.19022877734014917,
      "grad_norm": 0.26642883473494755,
      "learning_rate": 1.423656488015707e-07,
      "loss": 1.0023,
      "step": 1135
    },
    {
      "epoch": 0.19039637978714488,
      "grad_norm": 0.2452004857527195,
      "learning_rate": 1.4226729190748054e-07,
      "loss": 0.9906,
      "step": 1136
    },
    {
      "epoch": 0.19056398223414062,
      "grad_norm": 0.2573518735042166,
      "learning_rate": 1.4216888520447647e-07,
      "loss": 0.9861,
      "step": 1137
    },
    {
      "epoch": 0.19073158468113635,
      "grad_norm": 0.24049829297500894,
      "learning_rate": 1.4207042880852354e-07,
      "loss": 1.0186,
      "step": 1138
    },
    {
      "epoch": 0.19089918712813206,
      "grad_norm": 0.2465095339765797,
      "learning_rate": 1.4197192283564541e-07,
      "loss": 0.9785,
      "step": 1139
    },
    {
      "epoch": 0.1910667895751278,
      "grad_norm": 0.22964382096072583,
      "learning_rate": 1.418733674019242e-07,
      "loss": 0.9495,
      "step": 1140
    },
    {
      "epoch": 0.19123439202212353,
      "grad_norm": 0.27823354948041673,
      "learning_rate": 1.4177476262350026e-07,
      "loss": 1.0056,
      "step": 1141
    },
    {
      "epoch": 0.19140199446911924,
      "grad_norm": 0.24379227155102312,
      "learning_rate": 1.4167610861657214e-07,
      "loss": 1.019,
      "step": 1142
    },
    {
      "epoch": 0.19156959691611497,
      "grad_norm": 0.2046503041571023,
      "learning_rate": 1.4157740549739634e-07,
      "loss": 0.922,
      "step": 1143
    },
    {
      "epoch": 0.1917371993631107,
      "grad_norm": 0.26471109739585397,
      "learning_rate": 1.4147865338228733e-07,
      "loss": 1.0156,
      "step": 1144
    },
    {
      "epoch": 0.19190480181010644,
      "grad_norm": 0.23459863307168086,
      "learning_rate": 1.4137985238761716e-07,
      "loss": 1.0296,
      "step": 1145
    },
    {
      "epoch": 0.19207240425710215,
      "grad_norm": 0.27805824337829166,
      "learning_rate": 1.4128100262981568e-07,
      "loss": 1.042,
      "step": 1146
    },
    {
      "epoch": 0.19224000670409788,
      "grad_norm": 0.23196064422751697,
      "learning_rate": 1.411821042253701e-07,
      "loss": 0.9966,
      "step": 1147
    },
    {
      "epoch": 0.19240760915109362,
      "grad_norm": 0.24605986890632264,
      "learning_rate": 1.4108315729082488e-07,
      "loss": 0.9688,
      "step": 1148
    },
    {
      "epoch": 0.19257521159808932,
      "grad_norm": 0.2395093967862174,
      "learning_rate": 1.409841619427818e-07,
      "loss": 1.0121,
      "step": 1149
    },
    {
      "epoch": 0.19274281404508506,
      "grad_norm": 0.2411222714776577,
      "learning_rate": 1.4088511829789967e-07,
      "loss": 0.9787,
      "step": 1150
    },
    {
      "epoch": 0.1929104164920808,
      "grad_norm": 0.2555480579650938,
      "learning_rate": 1.4078602647289412e-07,
      "loss": 1.005,
      "step": 1151
    },
    {
      "epoch": 0.1930780189390765,
      "grad_norm": 0.23251443845265748,
      "learning_rate": 1.4068688658453768e-07,
      "loss": 1.0018,
      "step": 1152
    },
    {
      "epoch": 0.19324562138607224,
      "grad_norm": 0.21789596745888143,
      "learning_rate": 1.4058769874965942e-07,
      "loss": 0.9771,
      "step": 1153
    },
    {
      "epoch": 0.19341322383306797,
      "grad_norm": 0.24255660656630235,
      "learning_rate": 1.40488463085145e-07,
      "loss": 1.0134,
      "step": 1154
    },
    {
      "epoch": 0.19358082628006368,
      "grad_norm": 0.2340392999024737,
      "learning_rate": 1.4038917970793633e-07,
      "loss": 0.9749,
      "step": 1155
    },
    {
      "epoch": 0.1937484287270594,
      "grad_norm": 0.21736633643936276,
      "learning_rate": 1.4028984873503169e-07,
      "loss": 1.0225,
      "step": 1156
    },
    {
      "epoch": 0.19391603117405515,
      "grad_norm": 0.24086441542192163,
      "learning_rate": 1.401904702834853e-07,
      "loss": 0.9684,
      "step": 1157
    },
    {
      "epoch": 0.19408363362105086,
      "grad_norm": 0.2648442747771526,
      "learning_rate": 1.4009104447040743e-07,
      "loss": 1.0134,
      "step": 1158
    },
    {
      "epoch": 0.1942512360680466,
      "grad_norm": 0.22837327691551038,
      "learning_rate": 1.3999157141296417e-07,
      "loss": 0.9995,
      "step": 1159
    },
    {
      "epoch": 0.19441883851504232,
      "grad_norm": 0.2284092075925194,
      "learning_rate": 1.3989205122837723e-07,
      "loss": 0.9651,
      "step": 1160
    },
    {
      "epoch": 0.19458644096203803,
      "grad_norm": 0.26070898107223905,
      "learning_rate": 1.397924840339238e-07,
      "loss": 1.0136,
      "step": 1161
    },
    {
      "epoch": 0.19475404340903377,
      "grad_norm": 0.24512665743574064,
      "learning_rate": 1.396928699469366e-07,
      "loss": 0.9674,
      "step": 1162
    },
    {
      "epoch": 0.1949216458560295,
      "grad_norm": 0.26196199099225287,
      "learning_rate": 1.3959320908480358e-07,
      "loss": 1.0461,
      "step": 1163
    },
    {
      "epoch": 0.19508924830302524,
      "grad_norm": 0.2504511231023409,
      "learning_rate": 1.3949350156496773e-07,
      "loss": 0.967,
      "step": 1164
    },
    {
      "epoch": 0.19525685075002094,
      "grad_norm": 0.23945754959420792,
      "learning_rate": 1.393937475049271e-07,
      "loss": 1.0399,
      "step": 1165
    },
    {
      "epoch": 0.19542445319701668,
      "grad_norm": 0.25396298073723306,
      "learning_rate": 1.3929394702223456e-07,
      "loss": 1.031,
      "step": 1166
    },
    {
      "epoch": 0.1955920556440124,
      "grad_norm": 0.20106252827442964,
      "learning_rate": 1.3919410023449767e-07,
      "loss": 0.9318,
      "step": 1167
    },
    {
      "epoch": 0.19575965809100812,
      "grad_norm": 0.25181474958712685,
      "learning_rate": 1.3909420725937856e-07,
      "loss": 1.0215,
      "step": 1168
    },
    {
      "epoch": 0.19592726053800386,
      "grad_norm": 0.19359126620147205,
      "learning_rate": 1.3899426821459384e-07,
      "loss": 0.8932,
      "step": 1169
    },
    {
      "epoch": 0.1960948629849996,
      "grad_norm": 0.24484805964949408,
      "learning_rate": 1.3889428321791433e-07,
      "loss": 1.0035,
      "step": 1170
    },
    {
      "epoch": 0.1962624654319953,
      "grad_norm": 0.2584996672355118,
      "learning_rate": 1.3879425238716503e-07,
      "loss": 1.0169,
      "step": 1171
    },
    {
      "epoch": 0.19643006787899103,
      "grad_norm": 0.2387116642788933,
      "learning_rate": 1.38694175840225e-07,
      "loss": 0.9673,
      "step": 1172
    },
    {
      "epoch": 0.19659767032598677,
      "grad_norm": 0.2475741321592713,
      "learning_rate": 1.3859405369502718e-07,
      "loss": 0.9992,
      "step": 1173
    },
    {
      "epoch": 0.19676527277298247,
      "grad_norm": 0.24331476860831863,
      "learning_rate": 1.3849388606955806e-07,
      "loss": 0.9776,
      "step": 1174
    },
    {
      "epoch": 0.1969328752199782,
      "grad_norm": 0.22610316758286486,
      "learning_rate": 1.38393673081858e-07,
      "loss": 0.9841,
      "step": 1175
    },
    {
      "epoch": 0.19710047766697394,
      "grad_norm": 0.21527177127577757,
      "learning_rate": 1.3829341485002062e-07,
      "loss": 0.9305,
      "step": 1176
    },
    {
      "epoch": 0.19726808011396965,
      "grad_norm": 0.25563345626654227,
      "learning_rate": 1.3819311149219288e-07,
      "loss": 1.003,
      "step": 1177
    },
    {
      "epoch": 0.1974356825609654,
      "grad_norm": 0.2565001991986457,
      "learning_rate": 1.3809276312657505e-07,
      "loss": 0.9751,
      "step": 1178
    },
    {
      "epoch": 0.19760328500796112,
      "grad_norm": 0.21187063544525112,
      "learning_rate": 1.3799236987142027e-07,
      "loss": 0.9663,
      "step": 1179
    },
    {
      "epoch": 0.19777088745495683,
      "grad_norm": 0.2135815271005737,
      "learning_rate": 1.3789193184503468e-07,
      "loss": 0.9943,
      "step": 1180
    },
    {
      "epoch": 0.19793848990195256,
      "grad_norm": 0.26083712341757515,
      "learning_rate": 1.3779144916577717e-07,
      "loss": 1.0222,
      "step": 1181
    },
    {
      "epoch": 0.1981060923489483,
      "grad_norm": 0.2543975421839596,
      "learning_rate": 1.3769092195205917e-07,
      "loss": 1.01,
      "step": 1182
    },
    {
      "epoch": 0.19827369479594403,
      "grad_norm": 0.2160634601856732,
      "learning_rate": 1.3759035032234474e-07,
      "loss": 0.9586,
      "step": 1183
    },
    {
      "epoch": 0.19844129724293974,
      "grad_norm": 0.20311516200896093,
      "learning_rate": 1.3748973439515012e-07,
      "loss": 0.9179,
      "step": 1184
    },
    {
      "epoch": 0.19860889968993548,
      "grad_norm": 0.24095882948238503,
      "learning_rate": 1.3738907428904387e-07,
      "loss": 1.0055,
      "step": 1185
    },
    {
      "epoch": 0.1987765021369312,
      "grad_norm": 0.230855335499202,
      "learning_rate": 1.3728837012264656e-07,
      "loss": 0.9527,
      "step": 1186
    },
    {
      "epoch": 0.19894410458392692,
      "grad_norm": 0.23659439206259875,
      "learning_rate": 1.3718762201463064e-07,
      "loss": 0.9911,
      "step": 1187
    },
    {
      "epoch": 0.19911170703092265,
      "grad_norm": 0.22707224706851917,
      "learning_rate": 1.3708683008372048e-07,
      "loss": 0.9693,
      "step": 1188
    },
    {
      "epoch": 0.1992793094779184,
      "grad_norm": 0.2796046587100497,
      "learning_rate": 1.3698599444869197e-07,
      "loss": 1.0184,
      "step": 1189
    },
    {
      "epoch": 0.1994469119249141,
      "grad_norm": 0.25044602434953445,
      "learning_rate": 1.3688511522837246e-07,
      "loss": 0.9944,
      "step": 1190
    },
    {
      "epoch": 0.19961451437190983,
      "grad_norm": 0.25225723815893397,
      "learning_rate": 1.3678419254164085e-07,
      "loss": 1.0203,
      "step": 1191
    },
    {
      "epoch": 0.19978211681890556,
      "grad_norm": 0.2317633297526905,
      "learning_rate": 1.3668322650742708e-07,
      "loss": 0.9834,
      "step": 1192
    },
    {
      "epoch": 0.19994971926590127,
      "grad_norm": 0.251993608697388,
      "learning_rate": 1.3658221724471226e-07,
      "loss": 1.0062,
      "step": 1193
    },
    {
      "epoch": 0.200117321712897,
      "grad_norm": 0.21468012951531518,
      "learning_rate": 1.364811648725284e-07,
      "loss": 0.9633,
      "step": 1194
    },
    {
      "epoch": 0.20028492415989274,
      "grad_norm": 0.2519393516840195,
      "learning_rate": 1.363800695099584e-07,
      "loss": 0.9846,
      "step": 1195
    },
    {
      "epoch": 0.20045252660688845,
      "grad_norm": 0.2241063140336694,
      "learning_rate": 1.3627893127613567e-07,
      "loss": 0.9379,
      "step": 1196
    },
    {
      "epoch": 0.20062012905388418,
      "grad_norm": 0.24124120418261247,
      "learning_rate": 1.3617775029024427e-07,
      "loss": 1.0092,
      "step": 1197
    },
    {
      "epoch": 0.20078773150087992,
      "grad_norm": 0.230926091982134,
      "learning_rate": 1.360765266715186e-07,
      "loss": 0.9776,
      "step": 1198
    },
    {
      "epoch": 0.20095533394787565,
      "grad_norm": 0.23855572481087772,
      "learning_rate": 1.3597526053924332e-07,
      "loss": 0.9933,
      "step": 1199
    },
    {
      "epoch": 0.20112293639487136,
      "grad_norm": 0.22916408303211788,
      "learning_rate": 1.358739520127531e-07,
      "loss": 0.9738,
      "step": 1200
    },
    {
      "epoch": 0.2012905388418671,
      "grad_norm": 0.22919865012896962,
      "learning_rate": 1.3577260121143265e-07,
      "loss": 1.0047,
      "step": 1201
    },
    {
      "epoch": 0.20145814128886283,
      "grad_norm": 0.25725831920688014,
      "learning_rate": 1.3567120825471653e-07,
      "loss": 0.9867,
      "step": 1202
    },
    {
      "epoch": 0.20162574373585854,
      "grad_norm": 0.240138598239353,
      "learning_rate": 1.355697732620889e-07,
      "loss": 0.9824,
      "step": 1203
    },
    {
      "epoch": 0.20179334618285427,
      "grad_norm": 0.2496559360332462,
      "learning_rate": 1.3546829635308345e-07,
      "loss": 0.9972,
      "step": 1204
    },
    {
      "epoch": 0.20196094862985,
      "grad_norm": 0.2526212064383878,
      "learning_rate": 1.3536677764728334e-07,
      "loss": 0.9799,
      "step": 1205
    },
    {
      "epoch": 0.2021285510768457,
      "grad_norm": 0.2364059375663139,
      "learning_rate": 1.3526521726432095e-07,
      "loss": 1.0001,
      "step": 1206
    },
    {
      "epoch": 0.20229615352384145,
      "grad_norm": 0.24710324046431176,
      "learning_rate": 1.3516361532387772e-07,
      "loss": 0.9848,
      "step": 1207
    },
    {
      "epoch": 0.20246375597083718,
      "grad_norm": 0.23850334883898222,
      "learning_rate": 1.3506197194568417e-07,
      "loss": 0.9625,
      "step": 1208
    },
    {
      "epoch": 0.2026313584178329,
      "grad_norm": 0.2396690482206315,
      "learning_rate": 1.3496028724951954e-07,
      "loss": 0.96,
      "step": 1209
    },
    {
      "epoch": 0.20279896086482863,
      "grad_norm": 0.25399188079007085,
      "learning_rate": 1.3485856135521187e-07,
      "loss": 0.9987,
      "step": 1210
    },
    {
      "epoch": 0.20296656331182436,
      "grad_norm": 0.2374458692682214,
      "learning_rate": 1.3475679438263762e-07,
      "loss": 1.0175,
      "step": 1211
    },
    {
      "epoch": 0.20313416575882007,
      "grad_norm": 0.2503536364067803,
      "learning_rate": 1.346549864517218e-07,
      "loss": 1.0024,
      "step": 1212
    },
    {
      "epoch": 0.2033017682058158,
      "grad_norm": 0.2461747978852277,
      "learning_rate": 1.3455313768243755e-07,
      "loss": 0.981,
      "step": 1213
    },
    {
      "epoch": 0.20346937065281154,
      "grad_norm": 0.22042758395977047,
      "learning_rate": 1.344512481948062e-07,
      "loss": 0.9488,
      "step": 1214
    },
    {
      "epoch": 0.20363697309980724,
      "grad_norm": 0.26108942153381937,
      "learning_rate": 1.3434931810889717e-07,
      "loss": 0.9963,
      "step": 1215
    },
    {
      "epoch": 0.20380457554680298,
      "grad_norm": 0.22698321481693282,
      "learning_rate": 1.3424734754482746e-07,
      "loss": 0.9451,
      "step": 1216
    },
    {
      "epoch": 0.20397217799379871,
      "grad_norm": 0.1978668594938756,
      "learning_rate": 1.3414533662276206e-07,
      "loss": 0.9239,
      "step": 1217
    },
    {
      "epoch": 0.20413978044079445,
      "grad_norm": 0.24400327459154192,
      "learning_rate": 1.3404328546291333e-07,
      "loss": 0.9871,
      "step": 1218
    },
    {
      "epoch": 0.20430738288779016,
      "grad_norm": 0.23450896707182467,
      "learning_rate": 1.3394119418554107e-07,
      "loss": 0.9807,
      "step": 1219
    },
    {
      "epoch": 0.2044749853347859,
      "grad_norm": 0.22885465953624523,
      "learning_rate": 1.338390629109524e-07,
      "loss": 0.971,
      "step": 1220
    },
    {
      "epoch": 0.20464258778178163,
      "grad_norm": 0.2515270044958081,
      "learning_rate": 1.3373689175950162e-07,
      "loss": 0.9756,
      "step": 1221
    },
    {
      "epoch": 0.20481019022877733,
      "grad_norm": 0.22286991596084224,
      "learning_rate": 1.336346808515899e-07,
      "loss": 0.9931,
      "step": 1222
    },
    {
      "epoch": 0.20497779267577307,
      "grad_norm": 0.2501365103247085,
      "learning_rate": 1.335324303076654e-07,
      "loss": 1.0258,
      "step": 1223
    },
    {
      "epoch": 0.2051453951227688,
      "grad_norm": 0.23947347895869472,
      "learning_rate": 1.3343014024822282e-07,
      "loss": 0.9557,
      "step": 1224
    },
    {
      "epoch": 0.2053129975697645,
      "grad_norm": 0.18695576322991975,
      "learning_rate": 1.3332781079380356e-07,
      "loss": 0.9027,
      "step": 1225
    },
    {
      "epoch": 0.20548060001676025,
      "grad_norm": 0.21533603629151477,
      "learning_rate": 1.332254420649954e-07,
      "loss": 0.9367,
      "step": 1226
    },
    {
      "epoch": 0.20564820246375598,
      "grad_norm": 0.20836936064583397,
      "learning_rate": 1.3312303418243238e-07,
      "loss": 0.9685,
      "step": 1227
    },
    {
      "epoch": 0.2058158049107517,
      "grad_norm": 0.21701885811520122,
      "learning_rate": 1.3302058726679478e-07,
      "loss": 0.9445,
      "step": 1228
    },
    {
      "epoch": 0.20598340735774742,
      "grad_norm": 0.2208069044582316,
      "learning_rate": 1.329181014388087e-07,
      "loss": 0.9884,
      "step": 1229
    },
    {
      "epoch": 0.20615100980474316,
      "grad_norm": 0.20002837962749884,
      "learning_rate": 1.3281557681924626e-07,
      "loss": 0.9265,
      "step": 1230
    },
    {
      "epoch": 0.20631861225173886,
      "grad_norm": 0.23071708236778068,
      "learning_rate": 1.3271301352892528e-07,
      "loss": 0.9984,
      "step": 1231
    },
    {
      "epoch": 0.2064862146987346,
      "grad_norm": 0.2233505505972024,
      "learning_rate": 1.3261041168870897e-07,
      "loss": 0.9928,
      "step": 1232
    },
    {
      "epoch": 0.20665381714573033,
      "grad_norm": 0.24297031904018726,
      "learning_rate": 1.3250777141950617e-07,
      "loss": 0.9575,
      "step": 1233
    },
    {
      "epoch": 0.20682141959272604,
      "grad_norm": 0.24288707985286162,
      "learning_rate": 1.3240509284227097e-07,
      "loss": 0.9986,
      "step": 1234
    },
    {
      "epoch": 0.20698902203972178,
      "grad_norm": 0.2172713227681449,
      "learning_rate": 1.323023760780025e-07,
      "loss": 0.9875,
      "step": 1235
    },
    {
      "epoch": 0.2071566244867175,
      "grad_norm": 0.28125193492724426,
      "learning_rate": 1.32199621247745e-07,
      "loss": 1.0112,
      "step": 1236
    },
    {
      "epoch": 0.20732422693371325,
      "grad_norm": 0.22958238818357182,
      "learning_rate": 1.320968284725875e-07,
      "loss": 0.9869,
      "step": 1237
    },
    {
      "epoch": 0.20749182938070895,
      "grad_norm": 0.23877569285360634,
      "learning_rate": 1.3199399787366382e-07,
      "loss": 1.0137,
      "step": 1238
    },
    {
      "epoch": 0.2076594318277047,
      "grad_norm": 0.22340490070463057,
      "learning_rate": 1.3189112957215224e-07,
      "loss": 0.9926,
      "step": 1239
    },
    {
      "epoch": 0.20782703427470042,
      "grad_norm": 0.23409010648588296,
      "learning_rate": 1.3178822368927555e-07,
      "loss": 0.9587,
      "step": 1240
    },
    {
      "epoch": 0.20799463672169613,
      "grad_norm": 0.2312586140567972,
      "learning_rate": 1.3168528034630088e-07,
      "loss": 1.013,
      "step": 1241
    },
    {
      "epoch": 0.20816223916869186,
      "grad_norm": 0.22736818761893907,
      "learning_rate": 1.3158229966453935e-07,
      "loss": 0.9617,
      "step": 1242
    },
    {
      "epoch": 0.2083298416156876,
      "grad_norm": 0.22249982679037675,
      "learning_rate": 1.314792817653462e-07,
      "loss": 0.9365,
      "step": 1243
    },
    {
      "epoch": 0.2084974440626833,
      "grad_norm": 0.20016127213284,
      "learning_rate": 1.313762267701205e-07,
      "loss": 0.9387,
      "step": 1244
    },
    {
      "epoch": 0.20866504650967904,
      "grad_norm": 0.20104946952245914,
      "learning_rate": 1.31273134800305e-07,
      "loss": 0.9229,
      "step": 1245
    },
    {
      "epoch": 0.20883264895667478,
      "grad_norm": 0.23249236033180695,
      "learning_rate": 1.3117000597738607e-07,
      "loss": 1.0043,
      "step": 1246
    },
    {
      "epoch": 0.20900025140367048,
      "grad_norm": 0.21877799670135623,
      "learning_rate": 1.310668404228935e-07,
      "loss": 0.9848,
      "step": 1247
    },
    {
      "epoch": 0.20916785385066622,
      "grad_norm": 0.22610479994322308,
      "learning_rate": 1.3096363825840036e-07,
      "loss": 1.0014,
      "step": 1248
    },
    {
      "epoch": 0.20933545629766195,
      "grad_norm": 0.2403774119545526,
      "learning_rate": 1.3086039960552282e-07,
      "loss": 1.0069,
      "step": 1249
    },
    {
      "epoch": 0.20950305874465766,
      "grad_norm": 0.2517376561223675,
      "learning_rate": 1.3075712458592012e-07,
      "loss": 1.039,
      "step": 1250
    },
    {
      "epoch": 0.2096706611916534,
      "grad_norm": 0.23141651582047068,
      "learning_rate": 1.3065381332129432e-07,
      "loss": 0.9889,
      "step": 1251
    },
    {
      "epoch": 0.20983826363864913,
      "grad_norm": 0.2213908650043683,
      "learning_rate": 1.3055046593339019e-07,
      "loss": 0.9871,
      "step": 1252
    },
    {
      "epoch": 0.21000586608564484,
      "grad_norm": 0.21112292785825962,
      "learning_rate": 1.30447082543995e-07,
      "loss": 0.9438,
      "step": 1253
    },
    {
      "epoch": 0.21017346853264057,
      "grad_norm": 0.22811040885750322,
      "learning_rate": 1.3034366327493868e-07,
      "loss": 1.0018,
      "step": 1254
    },
    {
      "epoch": 0.2103410709796363,
      "grad_norm": 0.213598725384136,
      "learning_rate": 1.3024020824809314e-07,
      "loss": 1.0041,
      "step": 1255
    },
    {
      "epoch": 0.21050867342663204,
      "grad_norm": 0.22076535802221786,
      "learning_rate": 1.3013671758537259e-07,
      "loss": 0.9745,
      "step": 1256
    },
    {
      "epoch": 0.21067627587362775,
      "grad_norm": 0.24625753391474609,
      "learning_rate": 1.3003319140873328e-07,
      "loss": 1.0144,
      "step": 1257
    },
    {
      "epoch": 0.21084387832062348,
      "grad_norm": 0.24891080229062815,
      "learning_rate": 1.2992962984017312e-07,
      "loss": 0.984,
      "step": 1258
    },
    {
      "epoch": 0.21101148076761922,
      "grad_norm": 0.2135309488599901,
      "learning_rate": 1.2982603300173197e-07,
      "loss": 0.9529,
      "step": 1259
    },
    {
      "epoch": 0.21117908321461493,
      "grad_norm": 0.2314322700486935,
      "learning_rate": 1.297224010154911e-07,
      "loss": 0.9745,
      "step": 1260
    },
    {
      "epoch": 0.21134668566161066,
      "grad_norm": 0.23313308350675857,
      "learning_rate": 1.2961873400357316e-07,
      "loss": 0.9804,
      "step": 1261
    },
    {
      "epoch": 0.2115142881086064,
      "grad_norm": 0.24606366906234478,
      "learning_rate": 1.2951503208814226e-07,
      "loss": 0.9566,
      "step": 1262
    },
    {
      "epoch": 0.2116818905556021,
      "grad_norm": 0.23543600055038913,
      "learning_rate": 1.2941129539140345e-07,
      "loss": 0.9544,
      "step": 1263
    },
    {
      "epoch": 0.21184949300259784,
      "grad_norm": 0.23374525374615282,
      "learning_rate": 1.2930752403560284e-07,
      "loss": 0.9533,
      "step": 1264
    },
    {
      "epoch": 0.21201709544959357,
      "grad_norm": 0.21488578071735512,
      "learning_rate": 1.2920371814302742e-07,
      "loss": 0.9618,
      "step": 1265
    },
    {
      "epoch": 0.21218469789658928,
      "grad_norm": 0.22187410141762648,
      "learning_rate": 1.2909987783600483e-07,
      "loss": 0.9677,
      "step": 1266
    },
    {
      "epoch": 0.21235230034358502,
      "grad_norm": 0.20600044597304665,
      "learning_rate": 1.2899600323690332e-07,
      "loss": 0.9132,
      "step": 1267
    },
    {
      "epoch": 0.21251990279058075,
      "grad_norm": 0.24228020897119557,
      "learning_rate": 1.2889209446813146e-07,
      "loss": 0.9806,
      "step": 1268
    },
    {
      "epoch": 0.21268750523757646,
      "grad_norm": 0.20871771048333437,
      "learning_rate": 1.2878815165213812e-07,
      "loss": 0.9417,
      "step": 1269
    },
    {
      "epoch": 0.2128551076845722,
      "grad_norm": 0.21986687174763897,
      "learning_rate": 1.286841749114124e-07,
      "loss": 0.9507,
      "step": 1270
    },
    {
      "epoch": 0.21302271013156793,
      "grad_norm": 0.24486003518831956,
      "learning_rate": 1.285801643684832e-07,
      "loss": 1.0112,
      "step": 1271
    },
    {
      "epoch": 0.21319031257856363,
      "grad_norm": 0.22328937975165555,
      "learning_rate": 1.2847612014591943e-07,
      "loss": 0.966,
      "step": 1272
    },
    {
      "epoch": 0.21335791502555937,
      "grad_norm": 0.21578595561995811,
      "learning_rate": 1.283720423663295e-07,
      "loss": 0.9733,
      "step": 1273
    },
    {
      "epoch": 0.2135255174725551,
      "grad_norm": 0.21947071341026533,
      "learning_rate": 1.2826793115236154e-07,
      "loss": 0.9411,
      "step": 1274
    },
    {
      "epoch": 0.21369311991955084,
      "grad_norm": 0.23600220415450174,
      "learning_rate": 1.28163786626703e-07,
      "loss": 0.9985,
      "step": 1275
    },
    {
      "epoch": 0.21386072236654655,
      "grad_norm": 0.21073425623163428,
      "learning_rate": 1.2805960891208053e-07,
      "loss": 0.9304,
      "step": 1276
    },
    {
      "epoch": 0.21402832481354228,
      "grad_norm": 0.22913915045804628,
      "learning_rate": 1.2795539813126003e-07,
      "loss": 0.963,
      "step": 1277
    },
    {
      "epoch": 0.21419592726053802,
      "grad_norm": 0.23807050245687525,
      "learning_rate": 1.2785115440704626e-07,
      "loss": 0.9663,
      "step": 1278
    },
    {
      "epoch": 0.21436352970753372,
      "grad_norm": 0.22703070619413396,
      "learning_rate": 1.2774687786228279e-07,
      "loss": 0.979,
      "step": 1279
    },
    {
      "epoch": 0.21453113215452946,
      "grad_norm": 0.2168932927126946,
      "learning_rate": 1.2764256861985196e-07,
      "loss": 0.9185,
      "step": 1280
    },
    {
      "epoch": 0.2146987346015252,
      "grad_norm": 0.21535871458058461,
      "learning_rate": 1.2753822680267456e-07,
      "loss": 0.9755,
      "step": 1281
    },
    {
      "epoch": 0.2148663370485209,
      "grad_norm": 0.2245258717097903,
      "learning_rate": 1.274338525337098e-07,
      "loss": 0.9608,
      "step": 1282
    },
    {
      "epoch": 0.21503393949551663,
      "grad_norm": 0.21494277951150312,
      "learning_rate": 1.2732944593595514e-07,
      "loss": 0.9752,
      "step": 1283
    },
    {
      "epoch": 0.21520154194251237,
      "grad_norm": 0.21705861096523738,
      "learning_rate": 1.272250071324461e-07,
      "loss": 0.9414,
      "step": 1284
    },
    {
      "epoch": 0.21536914438950808,
      "grad_norm": 0.21030716101731814,
      "learning_rate": 1.2712053624625617e-07,
      "loss": 0.938,
      "step": 1285
    },
    {
      "epoch": 0.2155367468365038,
      "grad_norm": 0.21376799847785907,
      "learning_rate": 1.270160334004967e-07,
      "loss": 0.9595,
      "step": 1286
    },
    {
      "epoch": 0.21570434928349955,
      "grad_norm": 0.22278392841316932,
      "learning_rate": 1.269114987183166e-07,
      "loss": 0.9652,
      "step": 1287
    },
    {
      "epoch": 0.21587195173049525,
      "grad_norm": 0.23801259779757067,
      "learning_rate": 1.2680693232290238e-07,
      "loss": 1.001,
      "step": 1288
    },
    {
      "epoch": 0.216039554177491,
      "grad_norm": 0.241419244383179,
      "learning_rate": 1.2670233433747787e-07,
      "loss": 1.0186,
      "step": 1289
    },
    {
      "epoch": 0.21620715662448672,
      "grad_norm": 0.22849636168937415,
      "learning_rate": 1.2659770488530422e-07,
      "loss": 0.9669,
      "step": 1290
    },
    {
      "epoch": 0.21637475907148243,
      "grad_norm": 0.20648026578617815,
      "learning_rate": 1.264930440896795e-07,
      "loss": 0.9568,
      "step": 1291
    },
    {
      "epoch": 0.21654236151847817,
      "grad_norm": 0.2149005719625435,
      "learning_rate": 1.2638835207393887e-07,
      "loss": 0.9775,
      "step": 1292
    },
    {
      "epoch": 0.2167099639654739,
      "grad_norm": 0.20836982814729557,
      "learning_rate": 1.262836289614542e-07,
      "loss": 0.9221,
      "step": 1293
    },
    {
      "epoch": 0.21687756641246964,
      "grad_norm": 0.21828248741401815,
      "learning_rate": 1.2617887487563402e-07,
      "loss": 0.9089,
      "step": 1294
    },
    {
      "epoch": 0.21704516885946534,
      "grad_norm": 0.24489170933484686,
      "learning_rate": 1.2607408993992334e-07,
      "loss": 1.0214,
      "step": 1295
    },
    {
      "epoch": 0.21721277130646108,
      "grad_norm": 0.19590993148659527,
      "learning_rate": 1.2596927427780362e-07,
      "loss": 0.9567,
      "step": 1296
    },
    {
      "epoch": 0.2173803737534568,
      "grad_norm": 0.21718482241850062,
      "learning_rate": 1.2586442801279238e-07,
      "loss": 0.9743,
      "step": 1297
    },
    {
      "epoch": 0.21754797620045252,
      "grad_norm": 0.21641488063805553,
      "learning_rate": 1.257595512684433e-07,
      "loss": 0.9747,
      "step": 1298
    },
    {
      "epoch": 0.21771557864744825,
      "grad_norm": 0.23243387554504236,
      "learning_rate": 1.2565464416834595e-07,
      "loss": 0.989,
      "step": 1299
    },
    {
      "epoch": 0.217883181094444,
      "grad_norm": 0.2234773406090437,
      "learning_rate": 1.255497068361257e-07,
      "loss": 0.97,
      "step": 1300
    },
    {
      "epoch": 0.2180507835414397,
      "grad_norm": 0.22965862929105052,
      "learning_rate": 1.2544473939544353e-07,
      "loss": 0.9768,
      "step": 1301
    },
    {
      "epoch": 0.21821838598843543,
      "grad_norm": 0.21138850335202766,
      "learning_rate": 1.2533974196999587e-07,
      "loss": 0.9488,
      "step": 1302
    },
    {
      "epoch": 0.21838598843543117,
      "grad_norm": 0.23698462723531488,
      "learning_rate": 1.2523471468351455e-07,
      "loss": 1.0176,
      "step": 1303
    },
    {
      "epoch": 0.21855359088242687,
      "grad_norm": 0.20549895929770967,
      "learning_rate": 1.2512965765976652e-07,
      "loss": 0.9447,
      "step": 1304
    },
    {
      "epoch": 0.2187211933294226,
      "grad_norm": 0.22776017522814468,
      "learning_rate": 1.250245710225538e-07,
      "loss": 0.9877,
      "step": 1305
    },
    {
      "epoch": 0.21888879577641834,
      "grad_norm": 0.1926335930999807,
      "learning_rate": 1.2491945489571332e-07,
      "loss": 0.9333,
      "step": 1306
    },
    {
      "epoch": 0.21905639822341405,
      "grad_norm": 0.20986948383462278,
      "learning_rate": 1.2481430940311677e-07,
      "loss": 0.9695,
      "step": 1307
    },
    {
      "epoch": 0.21922400067040979,
      "grad_norm": 0.19136341646100924,
      "learning_rate": 1.2470913466867038e-07,
      "loss": 0.8981,
      "step": 1308
    },
    {
      "epoch": 0.21939160311740552,
      "grad_norm": 0.2122088783848289,
      "learning_rate": 1.2460393081631493e-07,
      "loss": 0.9368,
      "step": 1309
    },
    {
      "epoch": 0.21955920556440123,
      "grad_norm": 0.24274161636086264,
      "learning_rate": 1.2449869797002546e-07,
      "loss": 0.9878,
      "step": 1310
    },
    {
      "epoch": 0.21972680801139696,
      "grad_norm": 0.24033199032600822,
      "learning_rate": 1.2439343625381116e-07,
      "loss": 0.9848,
      "step": 1311
    },
    {
      "epoch": 0.2198944104583927,
      "grad_norm": 0.22238133902414722,
      "learning_rate": 1.2428814579171532e-07,
      "loss": 0.9933,
      "step": 1312
    },
    {
      "epoch": 0.22006201290538843,
      "grad_norm": 0.22782175786476247,
      "learning_rate": 1.24182826707815e-07,
      "loss": 1.0151,
      "step": 1313
    },
    {
      "epoch": 0.22022961535238414,
      "grad_norm": 0.2101993667069179,
      "learning_rate": 1.240774791262211e-07,
      "loss": 0.9703,
      "step": 1314
    },
    {
      "epoch": 0.22039721779937987,
      "grad_norm": 0.2398258134138547,
      "learning_rate": 1.2397210317107797e-07,
      "loss": 1.0044,
      "step": 1315
    },
    {
      "epoch": 0.2205648202463756,
      "grad_norm": 0.21044811133933125,
      "learning_rate": 1.2386669896656353e-07,
      "loss": 0.9309,
      "step": 1316
    },
    {
      "epoch": 0.22073242269337132,
      "grad_norm": 0.23373436787476506,
      "learning_rate": 1.2376126663688892e-07,
      "loss": 0.976,
      "step": 1317
    },
    {
      "epoch": 0.22090002514036705,
      "grad_norm": 0.219616607058818,
      "learning_rate": 1.236558063062984e-07,
      "loss": 0.9474,
      "step": 1318
    },
    {
      "epoch": 0.22106762758736279,
      "grad_norm": 0.20321542622133804,
      "learning_rate": 1.2355031809906932e-07,
      "loss": 0.9835,
      "step": 1319
    },
    {
      "epoch": 0.2212352300343585,
      "grad_norm": 0.22692544923106336,
      "learning_rate": 1.234448021395118e-07,
      "loss": 0.945,
      "step": 1320
    },
    {
      "epoch": 0.22140283248135423,
      "grad_norm": 0.23417316517782363,
      "learning_rate": 1.2333925855196862e-07,
      "loss": 0.98,
      "step": 1321
    },
    {
      "epoch": 0.22157043492834996,
      "grad_norm": 0.21295412318766993,
      "learning_rate": 1.2323368746081528e-07,
      "loss": 0.9711,
      "step": 1322
    },
    {
      "epoch": 0.22173803737534567,
      "grad_norm": 0.21285299289041557,
      "learning_rate": 1.2312808899045954e-07,
      "loss": 0.9757,
      "step": 1323
    },
    {
      "epoch": 0.2219056398223414,
      "grad_norm": 0.2224353429831012,
      "learning_rate": 1.2302246326534147e-07,
      "loss": 0.9881,
      "step": 1324
    },
    {
      "epoch": 0.22207324226933714,
      "grad_norm": 0.2457063312090721,
      "learning_rate": 1.229168104099333e-07,
      "loss": 0.9665,
      "step": 1325
    },
    {
      "epoch": 0.22224084471633285,
      "grad_norm": 0.18931303874420913,
      "learning_rate": 1.2281113054873923e-07,
      "loss": 0.9419,
      "step": 1326
    },
    {
      "epoch": 0.22240844716332858,
      "grad_norm": 0.23182209683621757,
      "learning_rate": 1.2270542380629515e-07,
      "loss": 1.0281,
      "step": 1327
    },
    {
      "epoch": 0.22257604961032432,
      "grad_norm": 0.20347520382029777,
      "learning_rate": 1.2259969030716878e-07,
      "loss": 1.0003,
      "step": 1328
    },
    {
      "epoch": 0.22274365205732002,
      "grad_norm": 0.21305117816127528,
      "learning_rate": 1.2249393017595934e-07,
      "loss": 0.9529,
      "step": 1329
    },
    {
      "epoch": 0.22291125450431576,
      "grad_norm": 0.2151381048504401,
      "learning_rate": 1.2238814353729742e-07,
      "loss": 0.9716,
      "step": 1330
    },
    {
      "epoch": 0.2230788569513115,
      "grad_norm": 0.23169091516642956,
      "learning_rate": 1.2228233051584477e-07,
      "loss": 0.9879,
      "step": 1331
    },
    {
      "epoch": 0.22324645939830723,
      "grad_norm": 0.24112478284680086,
      "learning_rate": 1.2217649123629438e-07,
      "loss": 0.9426,
      "step": 1332
    },
    {
      "epoch": 0.22341406184530294,
      "grad_norm": 0.22501536287671492,
      "learning_rate": 1.220706258233701e-07,
      "loss": 1.0192,
      "step": 1333
    },
    {
      "epoch": 0.22358166429229867,
      "grad_norm": 0.22574390539704434,
      "learning_rate": 1.219647344018265e-07,
      "loss": 0.9696,
      "step": 1334
    },
    {
      "epoch": 0.2237492667392944,
      "grad_norm": 0.16927491165183403,
      "learning_rate": 1.2185881709644897e-07,
      "loss": 0.9013,
      "step": 1335
    },
    {
      "epoch": 0.2239168691862901,
      "grad_norm": 0.19301425490295698,
      "learning_rate": 1.217528740320533e-07,
      "loss": 0.8987,
      "step": 1336
    },
    {
      "epoch": 0.22408447163328585,
      "grad_norm": 0.2336734079655653,
      "learning_rate": 1.2164690533348556e-07,
      "loss": 0.9904,
      "step": 1337
    },
    {
      "epoch": 0.22425207408028158,
      "grad_norm": 0.22228635926495316,
      "learning_rate": 1.2154091112562222e-07,
      "loss": 0.9393,
      "step": 1338
    },
    {
      "epoch": 0.2244196765272773,
      "grad_norm": 0.21653929766812835,
      "learning_rate": 1.214348915333697e-07,
      "loss": 0.9833,
      "step": 1339
    },
    {
      "epoch": 0.22458727897427302,
      "grad_norm": 0.24062298722578135,
      "learning_rate": 1.2132884668166425e-07,
      "loss": 0.9451,
      "step": 1340
    },
    {
      "epoch": 0.22475488142126876,
      "grad_norm": 0.19574126447243248,
      "learning_rate": 1.2122277669547208e-07,
      "loss": 0.9585,
      "step": 1341
    },
    {
      "epoch": 0.22492248386826447,
      "grad_norm": 0.21502996616617207,
      "learning_rate": 1.2111668169978884e-07,
      "loss": 1.0086,
      "step": 1342
    },
    {
      "epoch": 0.2250900863152602,
      "grad_norm": 0.2412599776692445,
      "learning_rate": 1.2101056181963976e-07,
      "loss": 1.0388,
      "step": 1343
    },
    {
      "epoch": 0.22525768876225594,
      "grad_norm": 0.17842197610566793,
      "learning_rate": 1.2090441718007935e-07,
      "loss": 0.9023,
      "step": 1344
    },
    {
      "epoch": 0.22542529120925164,
      "grad_norm": 0.22710678459364367,
      "learning_rate": 1.2079824790619136e-07,
      "loss": 0.9747,
      "step": 1345
    },
    {
      "epoch": 0.22559289365624738,
      "grad_norm": 0.24422237034747696,
      "learning_rate": 1.2069205412308844e-07,
      "loss": 0.9722,
      "step": 1346
    },
    {
      "epoch": 0.2257604961032431,
      "grad_norm": 0.2072293447904605,
      "learning_rate": 1.2058583595591226e-07,
      "loss": 0.9426,
      "step": 1347
    },
    {
      "epoch": 0.22592809855023882,
      "grad_norm": 0.22065371571550924,
      "learning_rate": 1.204795935298331e-07,
      "loss": 0.9825,
      "step": 1348
    },
    {
      "epoch": 0.22609570099723456,
      "grad_norm": 0.19464531785487485,
      "learning_rate": 1.2037332697004998e-07,
      "loss": 0.8868,
      "step": 1349
    },
    {
      "epoch": 0.2262633034442303,
      "grad_norm": 0.23655739587039398,
      "learning_rate": 1.2026703640179018e-07,
      "loss": 0.9953,
      "step": 1350
    },
    {
      "epoch": 0.22643090589122603,
      "grad_norm": 0.17828364271487715,
      "learning_rate": 1.2016072195030948e-07,
      "loss": 0.8937,
      "step": 1351
    },
    {
      "epoch": 0.22659850833822173,
      "grad_norm": 0.20283119201905822,
      "learning_rate": 1.2005438374089158e-07,
      "loss": 0.9367,
      "step": 1352
    },
    {
      "epoch": 0.22676611078521747,
      "grad_norm": 0.20969278248214265,
      "learning_rate": 1.1994802189884831e-07,
      "loss": 0.9805,
      "step": 1353
    },
    {
      "epoch": 0.2269337132322132,
      "grad_norm": 0.18031339651124975,
      "learning_rate": 1.198416365495194e-07,
      "loss": 0.9132,
      "step": 1354
    },
    {
      "epoch": 0.2271013156792089,
      "grad_norm": 0.20973101112854642,
      "learning_rate": 1.1973522781827216e-07,
      "loss": 0.9632,
      "step": 1355
    },
    {
      "epoch": 0.22726891812620464,
      "grad_norm": 0.1999293624481844,
      "learning_rate": 1.1962879583050145e-07,
      "loss": 0.9293,
      "step": 1356
    },
    {
      "epoch": 0.22743652057320038,
      "grad_norm": 0.21773625628191973,
      "learning_rate": 1.1952234071162967e-07,
      "loss": 0.9733,
      "step": 1357
    },
    {
      "epoch": 0.2276041230201961,
      "grad_norm": 0.21006410067633619,
      "learning_rate": 1.1941586258710635e-07,
      "loss": 0.9733,
      "step": 1358
    },
    {
      "epoch": 0.22777172546719182,
      "grad_norm": 0.22404096336078266,
      "learning_rate": 1.193093615824082e-07,
      "loss": 0.9495,
      "step": 1359
    },
    {
      "epoch": 0.22793932791418756,
      "grad_norm": 0.2118026677723084,
      "learning_rate": 1.1920283782303884e-07,
      "loss": 0.9498,
      "step": 1360
    },
    {
      "epoch": 0.22810693036118326,
      "grad_norm": 0.1937399719951624,
      "learning_rate": 1.1909629143452876e-07,
      "loss": 0.9475,
      "step": 1361
    },
    {
      "epoch": 0.228274532808179,
      "grad_norm": 0.18621792850659596,
      "learning_rate": 1.1898972254243506e-07,
      "loss": 0.9004,
      "step": 1362
    },
    {
      "epoch": 0.22844213525517473,
      "grad_norm": 0.2223776076803597,
      "learning_rate": 1.1888313127234144e-07,
      "loss": 0.9421,
      "step": 1363
    },
    {
      "epoch": 0.22860973770217044,
      "grad_norm": 0.1918367009300926,
      "learning_rate": 1.1877651774985789e-07,
      "loss": 0.9268,
      "step": 1364
    },
    {
      "epoch": 0.22877734014916618,
      "grad_norm": 0.21481289755801689,
      "learning_rate": 1.1866988210062063e-07,
      "loss": 0.9526,
      "step": 1365
    },
    {
      "epoch": 0.2289449425961619,
      "grad_norm": 0.22792409106246755,
      "learning_rate": 1.1856322445029201e-07,
      "loss": 0.9633,
      "step": 1366
    },
    {
      "epoch": 0.22911254504315762,
      "grad_norm": 0.1985644490824707,
      "learning_rate": 1.184565449245603e-07,
      "loss": 0.9369,
      "step": 1367
    },
    {
      "epoch": 0.22928014749015335,
      "grad_norm": 0.20102583665035217,
      "learning_rate": 1.1834984364913947e-07,
      "loss": 0.9517,
      "step": 1368
    },
    {
      "epoch": 0.2294477499371491,
      "grad_norm": 0.2224792431898555,
      "learning_rate": 1.182431207497692e-07,
      "loss": 0.9888,
      "step": 1369
    },
    {
      "epoch": 0.22961535238414482,
      "grad_norm": 0.19479460592209374,
      "learning_rate": 1.1813637635221461e-07,
      "loss": 0.9026,
      "step": 1370
    },
    {
      "epoch": 0.22978295483114053,
      "grad_norm": 0.2118419881951463,
      "learning_rate": 1.1802961058226619e-07,
      "loss": 1.0061,
      "step": 1371
    },
    {
      "epoch": 0.22995055727813626,
      "grad_norm": 0.2432551556401157,
      "learning_rate": 1.1792282356573958e-07,
      "loss": 0.9545,
      "step": 1372
    },
    {
      "epoch": 0.230118159725132,
      "grad_norm": 0.2067724477861119,
      "learning_rate": 1.1781601542847548e-07,
      "loss": 0.9325,
      "step": 1373
    },
    {
      "epoch": 0.2302857621721277,
      "grad_norm": 0.1903242165983211,
      "learning_rate": 1.1770918629633948e-07,
      "loss": 0.9108,
      "step": 1374
    },
    {
      "epoch": 0.23045336461912344,
      "grad_norm": 0.2158650821235818,
      "learning_rate": 1.176023362952219e-07,
      "loss": 0.9807,
      "step": 1375
    },
    {
      "epoch": 0.23062096706611918,
      "grad_norm": 0.22998178943845665,
      "learning_rate": 1.1749546555103764e-07,
      "loss": 0.9921,
      "step": 1376
    },
    {
      "epoch": 0.23078856951311488,
      "grad_norm": 0.2070906606787951,
      "learning_rate": 1.1738857418972607e-07,
      "loss": 0.9333,
      "step": 1377
    },
    {
      "epoch": 0.23095617196011062,
      "grad_norm": 0.20201840419731473,
      "learning_rate": 1.1728166233725088e-07,
      "loss": 0.9869,
      "step": 1378
    },
    {
      "epoch": 0.23112377440710635,
      "grad_norm": 0.20187251305135215,
      "learning_rate": 1.1717473011959977e-07,
      "loss": 0.9372,
      "step": 1379
    },
    {
      "epoch": 0.23129137685410206,
      "grad_norm": 0.1945184480230038,
      "learning_rate": 1.1706777766278468e-07,
      "loss": 0.9267,
      "step": 1380
    },
    {
      "epoch": 0.2314589793010978,
      "grad_norm": 0.2083269106253129,
      "learning_rate": 1.1696080509284118e-07,
      "loss": 0.9471,
      "step": 1381
    },
    {
      "epoch": 0.23162658174809353,
      "grad_norm": 0.19652147048864876,
      "learning_rate": 1.1685381253582865e-07,
      "loss": 0.9368,
      "step": 1382
    },
    {
      "epoch": 0.23179418419508924,
      "grad_norm": 0.18729352074769914,
      "learning_rate": 1.1674680011782995e-07,
      "loss": 0.8826,
      "step": 1383
    },
    {
      "epoch": 0.23196178664208497,
      "grad_norm": 0.1839587992587994,
      "learning_rate": 1.166397679649515e-07,
      "loss": 0.9229,
      "step": 1384
    },
    {
      "epoch": 0.2321293890890807,
      "grad_norm": 0.1898761592972123,
      "learning_rate": 1.1653271620332272e-07,
      "loss": 0.9226,
      "step": 1385
    },
    {
      "epoch": 0.2322969915360764,
      "grad_norm": 0.2103908707079103,
      "learning_rate": 1.1642564495909642e-07,
      "loss": 0.9575,
      "step": 1386
    },
    {
      "epoch": 0.23246459398307215,
      "grad_norm": 0.23057052675458078,
      "learning_rate": 1.1631855435844815e-07,
      "loss": 0.9761,
      "step": 1387
    },
    {
      "epoch": 0.23263219643006788,
      "grad_norm": 0.1979171773330782,
      "learning_rate": 1.1621144452757641e-07,
      "loss": 0.943,
      "step": 1388
    },
    {
      "epoch": 0.23279979887706362,
      "grad_norm": 0.20380038790924418,
      "learning_rate": 1.1610431559270226e-07,
      "loss": 0.9515,
      "step": 1389
    },
    {
      "epoch": 0.23296740132405933,
      "grad_norm": 0.23073657484956267,
      "learning_rate": 1.1599716768006937e-07,
      "loss": 0.9579,
      "step": 1390
    },
    {
      "epoch": 0.23313500377105506,
      "grad_norm": 0.20603225434150804,
      "learning_rate": 1.1589000091594367e-07,
      "loss": 0.9325,
      "step": 1391
    },
    {
      "epoch": 0.2333026062180508,
      "grad_norm": 0.24389060808073706,
      "learning_rate": 1.1578281542661336e-07,
      "loss": 1.0039,
      "step": 1392
    },
    {
      "epoch": 0.2334702086650465,
      "grad_norm": 0.22528060914510012,
      "learning_rate": 1.1567561133838876e-07,
      "loss": 0.9762,
      "step": 1393
    },
    {
      "epoch": 0.23363781111204224,
      "grad_norm": 0.19211936773083563,
      "learning_rate": 1.1556838877760204e-07,
      "loss": 0.8818,
      "step": 1394
    },
    {
      "epoch": 0.23380541355903797,
      "grad_norm": 0.22296141066140893,
      "learning_rate": 1.1546114787060707e-07,
      "loss": 0.9632,
      "step": 1395
    },
    {
      "epoch": 0.23397301600603368,
      "grad_norm": 0.22012320066921834,
      "learning_rate": 1.1535388874377952e-07,
      "loss": 0.9962,
      "step": 1396
    },
    {
      "epoch": 0.23414061845302941,
      "grad_norm": 0.20985958732569135,
      "learning_rate": 1.1524661152351641e-07,
      "loss": 0.9567,
      "step": 1397
    },
    {
      "epoch": 0.23430822090002515,
      "grad_norm": 0.21585955354913272,
      "learning_rate": 1.1513931633623602e-07,
      "loss": 0.9885,
      "step": 1398
    },
    {
      "epoch": 0.23447582334702086,
      "grad_norm": 0.22478511836863507,
      "learning_rate": 1.1503200330837799e-07,
      "loss": 0.9565,
      "step": 1399
    },
    {
      "epoch": 0.2346434257940166,
      "grad_norm": 0.2327922796331205,
      "learning_rate": 1.1492467256640281e-07,
      "loss": 0.9856,
      "step": 1400
    },
    {
      "epoch": 0.23481102824101233,
      "grad_norm": 0.18230094104425043,
      "learning_rate": 1.1481732423679195e-07,
      "loss": 0.8966,
      "step": 1401
    },
    {
      "epoch": 0.23497863068800803,
      "grad_norm": 0.20587310266547088,
      "learning_rate": 1.1470995844604758e-07,
      "loss": 0.9649,
      "step": 1402
    },
    {
      "epoch": 0.23514623313500377,
      "grad_norm": 0.20015979817143378,
      "learning_rate": 1.146025753206924e-07,
      "loss": 0.9229,
      "step": 1403
    },
    {
      "epoch": 0.2353138355819995,
      "grad_norm": 0.19009847057431548,
      "learning_rate": 1.1449517498726959e-07,
      "loss": 0.9401,
      "step": 1404
    },
    {
      "epoch": 0.2354814380289952,
      "grad_norm": 0.21421460705878254,
      "learning_rate": 1.1438775757234261e-07,
      "loss": 0.9859,
      "step": 1405
    },
    {
      "epoch": 0.23564904047599095,
      "grad_norm": 0.20222849860253245,
      "learning_rate": 1.1428032320249496e-07,
      "loss": 0.9279,
      "step": 1406
    },
    {
      "epoch": 0.23581664292298668,
      "grad_norm": 0.20986219216174531,
      "learning_rate": 1.1417287200433032e-07,
      "loss": 0.9626,
      "step": 1407
    },
    {
      "epoch": 0.23598424536998241,
      "grad_norm": 0.19900656545860393,
      "learning_rate": 1.1406540410447194e-07,
      "loss": 0.9229,
      "step": 1408
    },
    {
      "epoch": 0.23615184781697812,
      "grad_norm": 0.21164209774890974,
      "learning_rate": 1.1395791962956299e-07,
      "loss": 0.9394,
      "step": 1409
    },
    {
      "epoch": 0.23631945026397386,
      "grad_norm": 0.22476892298523768,
      "learning_rate": 1.1385041870626602e-07,
      "loss": 0.9356,
      "step": 1410
    },
    {
      "epoch": 0.2364870527109696,
      "grad_norm": 0.20000951746537993,
      "learning_rate": 1.1374290146126297e-07,
      "loss": 0.9308,
      "step": 1411
    },
    {
      "epoch": 0.2366546551579653,
      "grad_norm": 0.23994444947840793,
      "learning_rate": 1.1363536802125512e-07,
      "loss": 0.9922,
      "step": 1412
    },
    {
      "epoch": 0.23682225760496103,
      "grad_norm": 0.2331015866654547,
      "learning_rate": 1.1352781851296275e-07,
      "loss": 1.0065,
      "step": 1413
    },
    {
      "epoch": 0.23698986005195677,
      "grad_norm": 0.20357835975677763,
      "learning_rate": 1.1342025306312509e-07,
      "loss": 0.9422,
      "step": 1414
    },
    {
      "epoch": 0.23715746249895248,
      "grad_norm": 0.20973394507987322,
      "learning_rate": 1.1331267179850014e-07,
      "loss": 0.9216,
      "step": 1415
    },
    {
      "epoch": 0.2373250649459482,
      "grad_norm": 0.1920580860060001,
      "learning_rate": 1.1320507484586459e-07,
      "loss": 0.8392,
      "step": 1416
    },
    {
      "epoch": 0.23749266739294395,
      "grad_norm": 0.19687821438754363,
      "learning_rate": 1.1309746233201356e-07,
      "loss": 0.9237,
      "step": 1417
    },
    {
      "epoch": 0.23766026983993965,
      "grad_norm": 0.22132461443315718,
      "learning_rate": 1.1298983438376055e-07,
      "loss": 0.9735,
      "step": 1418
    },
    {
      "epoch": 0.2378278722869354,
      "grad_norm": 0.20783342384105805,
      "learning_rate": 1.128821911279372e-07,
      "loss": 0.9505,
      "step": 1419
    },
    {
      "epoch": 0.23799547473393112,
      "grad_norm": 0.22535863092930633,
      "learning_rate": 1.1277453269139328e-07,
      "loss": 0.9762,
      "step": 1420
    },
    {
      "epoch": 0.23816307718092683,
      "grad_norm": 0.20723156917643806,
      "learning_rate": 1.126668592009963e-07,
      "loss": 0.9269,
      "step": 1421
    },
    {
      "epoch": 0.23833067962792256,
      "grad_norm": 0.19548064906434723,
      "learning_rate": 1.1255917078363166e-07,
      "loss": 0.897,
      "step": 1422
    },
    {
      "epoch": 0.2384982820749183,
      "grad_norm": 0.232157457200605,
      "learning_rate": 1.1245146756620232e-07,
      "loss": 0.9616,
      "step": 1423
    },
    {
      "epoch": 0.238665884521914,
      "grad_norm": 0.16847895726665968,
      "learning_rate": 1.1234374967562854e-07,
      "loss": 0.9165,
      "step": 1424
    },
    {
      "epoch": 0.23883348696890974,
      "grad_norm": 0.22557578318660307,
      "learning_rate": 1.1223601723884805e-07,
      "loss": 0.9939,
      "step": 1425
    },
    {
      "epoch": 0.23900108941590548,
      "grad_norm": 0.21045223846443675,
      "learning_rate": 1.1212827038281568e-07,
      "loss": 0.9571,
      "step": 1426
    },
    {
      "epoch": 0.2391686918629012,
      "grad_norm": 0.18960920266283143,
      "learning_rate": 1.1202050923450316e-07,
      "loss": 0.9078,
      "step": 1427
    },
    {
      "epoch": 0.23933629430989692,
      "grad_norm": 0.20073094918781922,
      "learning_rate": 1.1191273392089917e-07,
      "loss": 0.9448,
      "step": 1428
    },
    {
      "epoch": 0.23950389675689265,
      "grad_norm": 0.20095106693836196,
      "learning_rate": 1.1180494456900902e-07,
      "loss": 0.9163,
      "step": 1429
    },
    {
      "epoch": 0.2396714992038884,
      "grad_norm": 0.19939007786516402,
      "learning_rate": 1.1169714130585459e-07,
      "loss": 0.9502,
      "step": 1430
    },
    {
      "epoch": 0.2398391016508841,
      "grad_norm": 0.20870232060482874,
      "learning_rate": 1.1158932425847413e-07,
      "loss": 0.9494,
      "step": 1431
    },
    {
      "epoch": 0.24000670409787983,
      "grad_norm": 0.21576902789790345,
      "learning_rate": 1.1148149355392215e-07,
      "loss": 0.9532,
      "step": 1432
    },
    {
      "epoch": 0.24017430654487557,
      "grad_norm": 0.2179745947556558,
      "learning_rate": 1.1137364931926931e-07,
      "loss": 1.005,
      "step": 1433
    },
    {
      "epoch": 0.24034190899187127,
      "grad_norm": 0.19262699870028058,
      "learning_rate": 1.1126579168160209e-07,
      "loss": 0.9357,
      "step": 1434
    },
    {
      "epoch": 0.240509511438867,
      "grad_norm": 0.21274760876510185,
      "learning_rate": 1.1115792076802284e-07,
      "loss": 0.9765,
      "step": 1435
    },
    {
      "epoch": 0.24067711388586274,
      "grad_norm": 0.20975995005508816,
      "learning_rate": 1.1105003670564959e-07,
      "loss": 0.9862,
      "step": 1436
    },
    {
      "epoch": 0.24084471633285845,
      "grad_norm": 0.21238734168737392,
      "learning_rate": 1.1094213962161576e-07,
      "loss": 0.9648,
      "step": 1437
    },
    {
      "epoch": 0.24101231877985418,
      "grad_norm": 0.19041807642878672,
      "learning_rate": 1.1083422964307023e-07,
      "loss": 0.9668,
      "step": 1438
    },
    {
      "epoch": 0.24117992122684992,
      "grad_norm": 0.20201547170588757,
      "learning_rate": 1.1072630689717699e-07,
      "loss": 0.9117,
      "step": 1439
    },
    {
      "epoch": 0.24134752367384563,
      "grad_norm": 0.2043618749465753,
      "learning_rate": 1.1061837151111513e-07,
      "loss": 0.941,
      "step": 1440
    },
    {
      "epoch": 0.24151512612084136,
      "grad_norm": 0.18151869550469638,
      "learning_rate": 1.1051042361207861e-07,
      "loss": 0.8901,
      "step": 1441
    },
    {
      "epoch": 0.2416827285678371,
      "grad_norm": 0.20313509558606266,
      "learning_rate": 1.1040246332727612e-07,
      "loss": 0.96,
      "step": 1442
    },
    {
      "epoch": 0.2418503310148328,
      "grad_norm": 0.21523932265696447,
      "learning_rate": 1.1029449078393096e-07,
      "loss": 0.9547,
      "step": 1443
    },
    {
      "epoch": 0.24201793346182854,
      "grad_norm": 0.2227112668160891,
      "learning_rate": 1.1018650610928094e-07,
      "loss": 0.9915,
      "step": 1444
    },
    {
      "epoch": 0.24218553590882427,
      "grad_norm": 0.20576247813926501,
      "learning_rate": 1.10078509430578e-07,
      "loss": 0.9297,
      "step": 1445
    },
    {
      "epoch": 0.24235313835582,
      "grad_norm": 0.21423678343174182,
      "learning_rate": 1.0997050087508844e-07,
      "loss": 0.9102,
      "step": 1446
    },
    {
      "epoch": 0.24252074080281572,
      "grad_norm": 0.1967636763233004,
      "learning_rate": 1.0986248057009238e-07,
      "loss": 0.9111,
      "step": 1447
    },
    {
      "epoch": 0.24268834324981145,
      "grad_norm": 0.24351308049172843,
      "learning_rate": 1.0975444864288386e-07,
      "loss": 1.0177,
      "step": 1448
    },
    {
      "epoch": 0.24285594569680719,
      "grad_norm": 0.20866641796440077,
      "learning_rate": 1.0964640522077064e-07,
      "loss": 0.9274,
      "step": 1449
    },
    {
      "epoch": 0.2430235481438029,
      "grad_norm": 0.22244392452175782,
      "learning_rate": 1.0953835043107398e-07,
      "loss": 0.9508,
      "step": 1450
    },
    {
      "epoch": 0.24319115059079863,
      "grad_norm": 0.18864064244445303,
      "learning_rate": 1.0943028440112854e-07,
      "loss": 0.8814,
      "step": 1451
    },
    {
      "epoch": 0.24335875303779436,
      "grad_norm": 0.22539950321647706,
      "learning_rate": 1.0932220725828225e-07,
      "loss": 0.9772,
      "step": 1452
    },
    {
      "epoch": 0.24352635548479007,
      "grad_norm": 0.20410483739694232,
      "learning_rate": 1.0921411912989614e-07,
      "loss": 0.9444,
      "step": 1453
    },
    {
      "epoch": 0.2436939579317858,
      "grad_norm": 0.20615535209358718,
      "learning_rate": 1.0910602014334412e-07,
      "loss": 0.9977,
      "step": 1454
    },
    {
      "epoch": 0.24386156037878154,
      "grad_norm": 0.21055882198064715,
      "learning_rate": 1.0899791042601298e-07,
      "loss": 0.9362,
      "step": 1455
    },
    {
      "epoch": 0.24402916282577725,
      "grad_norm": 0.2339778737666371,
      "learning_rate": 1.0888979010530213e-07,
      "loss": 0.9907,
      "step": 1456
    },
    {
      "epoch": 0.24419676527277298,
      "grad_norm": 0.2051538023286299,
      "learning_rate": 1.0878165930862342e-07,
      "loss": 0.9555,
      "step": 1457
    },
    {
      "epoch": 0.24436436771976872,
      "grad_norm": 0.19475765387497346,
      "learning_rate": 1.0867351816340111e-07,
      "loss": 0.9083,
      "step": 1458
    },
    {
      "epoch": 0.24453197016676442,
      "grad_norm": 0.2018151907114244,
      "learning_rate": 1.0856536679707169e-07,
      "loss": 0.9748,
      "step": 1459
    },
    {
      "epoch": 0.24469957261376016,
      "grad_norm": 0.21144919789948358,
      "learning_rate": 1.0845720533708356e-07,
      "loss": 0.9433,
      "step": 1460
    },
    {
      "epoch": 0.2448671750607559,
      "grad_norm": 0.17969275783470115,
      "learning_rate": 1.0834903391089711e-07,
      "loss": 0.8978,
      "step": 1461
    },
    {
      "epoch": 0.2450347775077516,
      "grad_norm": 0.2361936183852747,
      "learning_rate": 1.0824085264598453e-07,
      "loss": 0.9616,
      "step": 1462
    },
    {
      "epoch": 0.24520237995474733,
      "grad_norm": 0.1876737464507639,
      "learning_rate": 1.0813266166982944e-07,
      "loss": 0.9075,
      "step": 1463
    },
    {
      "epoch": 0.24536998240174307,
      "grad_norm": 0.1752482265345877,
      "learning_rate": 1.08024461109927e-07,
      "loss": 0.8869,
      "step": 1464
    },
    {
      "epoch": 0.2455375848487388,
      "grad_norm": 0.21018981162988915,
      "learning_rate": 1.0791625109378375e-07,
      "loss": 0.9863,
      "step": 1465
    },
    {
      "epoch": 0.2457051872957345,
      "grad_norm": 0.1887404242201157,
      "learning_rate": 1.0780803174891722e-07,
      "loss": 0.9033,
      "step": 1466
    },
    {
      "epoch": 0.24587278974273025,
      "grad_norm": 0.1785105842960758,
      "learning_rate": 1.0769980320285599e-07,
      "loss": 0.8788,
      "step": 1467
    },
    {
      "epoch": 0.24604039218972598,
      "grad_norm": 0.20959559203107594,
      "learning_rate": 1.0759156558313951e-07,
      "loss": 0.9593,
      "step": 1468
    },
    {
      "epoch": 0.2462079946367217,
      "grad_norm": 0.205888576576009,
      "learning_rate": 1.0748331901731794e-07,
      "loss": 0.924,
      "step": 1469
    },
    {
      "epoch": 0.24637559708371742,
      "grad_norm": 0.18903285613131462,
      "learning_rate": 1.0737506363295185e-07,
      "loss": 0.9472,
      "step": 1470
    },
    {
      "epoch": 0.24654319953071316,
      "grad_norm": 0.2121611150816383,
      "learning_rate": 1.0726679955761238e-07,
      "loss": 0.9398,
      "step": 1471
    },
    {
      "epoch": 0.24671080197770887,
      "grad_norm": 0.21122351869792255,
      "learning_rate": 1.0715852691888081e-07,
      "loss": 0.9292,
      "step": 1472
    },
    {
      "epoch": 0.2468784044247046,
      "grad_norm": 0.18801395420769768,
      "learning_rate": 1.0705024584434853e-07,
      "loss": 0.9114,
      "step": 1473
    },
    {
      "epoch": 0.24704600687170034,
      "grad_norm": 0.2070573957408405,
      "learning_rate": 1.0694195646161683e-07,
      "loss": 0.9445,
      "step": 1474
    },
    {
      "epoch": 0.24721360931869604,
      "grad_norm": 0.1967811372926684,
      "learning_rate": 1.0683365889829692e-07,
      "loss": 0.9545,
      "step": 1475
    },
    {
      "epoch": 0.24738121176569178,
      "grad_norm": 0.19426059620245903,
      "learning_rate": 1.0672535328200949e-07,
      "loss": 0.9225,
      "step": 1476
    },
    {
      "epoch": 0.2475488142126875,
      "grad_norm": 0.21938474954312004,
      "learning_rate": 1.0661703974038478e-07,
      "loss": 0.9161,
      "step": 1477
    },
    {
      "epoch": 0.24771641665968322,
      "grad_norm": 0.20073153036937214,
      "learning_rate": 1.0650871840106246e-07,
      "loss": 0.9397,
      "step": 1478
    },
    {
      "epoch": 0.24788401910667895,
      "grad_norm": 0.21423146434044446,
      "learning_rate": 1.0640038939169123e-07,
      "loss": 0.9371,
      "step": 1479
    },
    {
      "epoch": 0.2480516215536747,
      "grad_norm": 0.20335062149690392,
      "learning_rate": 1.0629205283992895e-07,
      "loss": 0.9308,
      "step": 1480
    },
    {
      "epoch": 0.2482192240006704,
      "grad_norm": 0.18495626524405812,
      "learning_rate": 1.0618370887344233e-07,
      "loss": 0.9086,
      "step": 1481
    },
    {
      "epoch": 0.24838682644766613,
      "grad_norm": 0.1962927438304182,
      "learning_rate": 1.0607535761990678e-07,
      "loss": 0.9204,
      "step": 1482
    },
    {
      "epoch": 0.24855442889466187,
      "grad_norm": 0.16774619418694672,
      "learning_rate": 1.0596699920700638e-07,
      "loss": 0.8846,
      "step": 1483
    },
    {
      "epoch": 0.2487220313416576,
      "grad_norm": 0.19611800589051567,
      "learning_rate": 1.0585863376243357e-07,
      "loss": 0.9484,
      "step": 1484
    },
    {
      "epoch": 0.2488896337886533,
      "grad_norm": 0.2017653062729045,
      "learning_rate": 1.0575026141388915e-07,
      "loss": 0.8986,
      "step": 1485
    },
    {
      "epoch": 0.24905723623564904,
      "grad_norm": 0.20348537845380224,
      "learning_rate": 1.0564188228908193e-07,
      "loss": 0.9238,
      "step": 1486
    },
    {
      "epoch": 0.24922483868264478,
      "grad_norm": 0.17530206416788402,
      "learning_rate": 1.055334965157288e-07,
      "loss": 0.9039,
      "step": 1487
    },
    {
      "epoch": 0.24939244112964049,
      "grad_norm": 0.2158790606353716,
      "learning_rate": 1.0542510422155457e-07,
      "loss": 0.9388,
      "step": 1488
    },
    {
      "epoch": 0.24956004357663622,
      "grad_norm": 0.20332920369888474,
      "learning_rate": 1.0531670553429155e-07,
      "loss": 0.956,
      "step": 1489
    },
    {
      "epoch": 0.24972764602363196,
      "grad_norm": 0.19420995480352177,
      "learning_rate": 1.0520830058167964e-07,
      "loss": 0.9501,
      "step": 1490
    },
    {
      "epoch": 0.24989524847062766,
      "grad_norm": 0.17148594455598862,
      "learning_rate": 1.0509988949146627e-07,
      "loss": 0.8662,
      "step": 1491
    },
    {
      "epoch": 0.2500628509176234,
      "grad_norm": 0.20005804271626657,
      "learning_rate": 1.0499147239140586e-07,
      "loss": 0.8899,
      "step": 1492
    },
    {
      "epoch": 0.25023045336461913,
      "grad_norm": 0.19287540463740618,
      "learning_rate": 1.0488304940926012e-07,
      "loss": 0.9023,
      "step": 1493
    },
    {
      "epoch": 0.25039805581161484,
      "grad_norm": 0.19723028060543246,
      "learning_rate": 1.0477462067279756e-07,
      "loss": 0.9307,
      "step": 1494
    },
    {
      "epoch": 0.2505656582586106,
      "grad_norm": 0.18088720501374259,
      "learning_rate": 1.0466618630979356e-07,
      "loss": 0.9203,
      "step": 1495
    },
    {
      "epoch": 0.2507332607056063,
      "grad_norm": 0.19171155119701028,
      "learning_rate": 1.0455774644803006e-07,
      "loss": 0.8843,
      "step": 1496
    },
    {
      "epoch": 0.250900863152602,
      "grad_norm": 0.23304749193452523,
      "learning_rate": 1.0444930121529554e-07,
      "loss": 0.9711,
      "step": 1497
    },
    {
      "epoch": 0.2510684655995978,
      "grad_norm": 0.2210400295755124,
      "learning_rate": 1.0434085073938474e-07,
      "loss": 0.9556,
      "step": 1498
    },
    {
      "epoch": 0.2512360680465935,
      "grad_norm": 0.2009649747313283,
      "learning_rate": 1.0423239514809865e-07,
      "loss": 0.98,
      "step": 1499
    },
    {
      "epoch": 0.2514036704935892,
      "grad_norm": 0.19242302617875495,
      "learning_rate": 1.0412393456924424e-07,
      "loss": 0.9147,
      "step": 1500
    },
    {
      "epoch": 0.25157127294058496,
      "grad_norm": 0.19457758886024298,
      "learning_rate": 1.0401546913063441e-07,
      "loss": 0.93,
      "step": 1501
    },
    {
      "epoch": 0.25173887538758066,
      "grad_norm": 0.19528594316551473,
      "learning_rate": 1.0390699896008772e-07,
      "loss": 0.9402,
      "step": 1502
    },
    {
      "epoch": 0.25190647783457637,
      "grad_norm": 0.19720723757722322,
      "learning_rate": 1.0379852418542829e-07,
      "loss": 0.9416,
      "step": 1503
    },
    {
      "epoch": 0.25207408028157213,
      "grad_norm": 0.18567424346746972,
      "learning_rate": 1.0369004493448583e-07,
      "loss": 0.9254,
      "step": 1504
    },
    {
      "epoch": 0.25224168272856784,
      "grad_norm": 0.19525724869719688,
      "learning_rate": 1.0358156133509513e-07,
      "loss": 0.9068,
      "step": 1505
    },
    {
      "epoch": 0.25240928517556355,
      "grad_norm": 0.20901640832925794,
      "learning_rate": 1.0347307351509613e-07,
      "loss": 0.9521,
      "step": 1506
    },
    {
      "epoch": 0.2525768876225593,
      "grad_norm": 0.2034957493522597,
      "learning_rate": 1.0336458160233392e-07,
      "loss": 0.9403,
      "step": 1507
    },
    {
      "epoch": 0.252744490069555,
      "grad_norm": 0.2207467262794117,
      "learning_rate": 1.032560857246582e-07,
      "loss": 0.9903,
      "step": 1508
    },
    {
      "epoch": 0.2529120925165507,
      "grad_norm": 0.21698260269483743,
      "learning_rate": 1.0314758600992348e-07,
      "loss": 0.9578,
      "step": 1509
    },
    {
      "epoch": 0.2530796949635465,
      "grad_norm": 0.19762364508829575,
      "learning_rate": 1.0303908258598873e-07,
      "loss": 0.9819,
      "step": 1510
    },
    {
      "epoch": 0.2532472974105422,
      "grad_norm": 0.18756759965316114,
      "learning_rate": 1.0293057558071727e-07,
      "loss": 0.9038,
      "step": 1511
    },
    {
      "epoch": 0.2534148998575379,
      "grad_norm": 0.2225845619955166,
      "learning_rate": 1.028220651219767e-07,
      "loss": 0.9759,
      "step": 1512
    },
    {
      "epoch": 0.25358250230453366,
      "grad_norm": 0.19471467747123442,
      "learning_rate": 1.0271355133763868e-07,
      "loss": 0.9106,
      "step": 1513
    },
    {
      "epoch": 0.25375010475152937,
      "grad_norm": 0.251981224142755,
      "learning_rate": 1.0260503435557877e-07,
      "loss": 0.9013,
      "step": 1514
    },
    {
      "epoch": 0.2539177071985251,
      "grad_norm": 0.2080083108330489,
      "learning_rate": 1.0249651430367627e-07,
      "loss": 0.9432,
      "step": 1515
    },
    {
      "epoch": 0.25408530964552084,
      "grad_norm": 0.19660997668873526,
      "learning_rate": 1.0238799130981413e-07,
      "loss": 0.9362,
      "step": 1516
    },
    {
      "epoch": 0.25425291209251655,
      "grad_norm": 0.20721895629063886,
      "learning_rate": 1.0227946550187883e-07,
      "loss": 0.9301,
      "step": 1517
    },
    {
      "epoch": 0.25442051453951225,
      "grad_norm": 0.2035532916468481,
      "learning_rate": 1.0217093700776003e-07,
      "loss": 0.9377,
      "step": 1518
    },
    {
      "epoch": 0.254588116986508,
      "grad_norm": 0.19451573905133004,
      "learning_rate": 1.0206240595535062e-07,
      "loss": 0.8601,
      "step": 1519
    },
    {
      "epoch": 0.2547557194335037,
      "grad_norm": 0.1767406726336927,
      "learning_rate": 1.0195387247254659e-07,
      "loss": 0.8635,
      "step": 1520
    },
    {
      "epoch": 0.25492332188049943,
      "grad_norm": 0.212067648617125,
      "learning_rate": 1.0184533668724666e-07,
      "loss": 0.9348,
      "step": 1521
    },
    {
      "epoch": 0.2550909243274952,
      "grad_norm": 0.21719985037245676,
      "learning_rate": 1.0173679872735233e-07,
      "loss": 0.9571,
      "step": 1522
    },
    {
      "epoch": 0.2552585267744909,
      "grad_norm": 0.19571672000859552,
      "learning_rate": 1.0162825872076766e-07,
      "loss": 0.912,
      "step": 1523
    },
    {
      "epoch": 0.2554261292214866,
      "grad_norm": 0.1903141580852671,
      "learning_rate": 1.0151971679539908e-07,
      "loss": 0.9278,
      "step": 1524
    },
    {
      "epoch": 0.25559373166848237,
      "grad_norm": 0.20102934818080695,
      "learning_rate": 1.0141117307915536e-07,
      "loss": 0.9255,
      "step": 1525
    },
    {
      "epoch": 0.2557613341154781,
      "grad_norm": 0.20615172773254153,
      "learning_rate": 1.0130262769994733e-07,
      "loss": 0.9257,
      "step": 1526
    },
    {
      "epoch": 0.2559289365624738,
      "grad_norm": 0.20407330643334914,
      "learning_rate": 1.0119408078568774e-07,
      "loss": 0.9423,
      "step": 1527
    },
    {
      "epoch": 0.25609653900946955,
      "grad_norm": 0.18476585966882575,
      "learning_rate": 1.0108553246429126e-07,
      "loss": 0.9616,
      "step": 1528
    },
    {
      "epoch": 0.25626414145646526,
      "grad_norm": 0.20081592906920473,
      "learning_rate": 1.0097698286367409e-07,
      "loss": 0.9647,
      "step": 1529
    },
    {
      "epoch": 0.256431743903461,
      "grad_norm": 0.21853509167745683,
      "learning_rate": 1.0086843211175404e-07,
      "loss": 0.9365,
      "step": 1530
    },
    {
      "epoch": 0.2565993463504567,
      "grad_norm": 0.15516198045257962,
      "learning_rate": 1.0075988033645025e-07,
      "loss": 0.8471,
      "step": 1531
    },
    {
      "epoch": 0.25676694879745243,
      "grad_norm": 0.2348636960514574,
      "learning_rate": 1.0065132766568296e-07,
      "loss": 0.978,
      "step": 1532
    },
    {
      "epoch": 0.2569345512444482,
      "grad_norm": 0.20730662246574572,
      "learning_rate": 1.0054277422737368e-07,
      "loss": 0.9277,
      "step": 1533
    },
    {
      "epoch": 0.2571021536914439,
      "grad_norm": 0.20262940065679788,
      "learning_rate": 1.0043422014944465e-07,
      "loss": 0.9433,
      "step": 1534
    },
    {
      "epoch": 0.2572697561384396,
      "grad_norm": 0.1951226783865686,
      "learning_rate": 1.003256655598189e-07,
      "loss": 0.929,
      "step": 1535
    },
    {
      "epoch": 0.25743735858543537,
      "grad_norm": 0.19414898959669752,
      "learning_rate": 1.0021711058642008e-07,
      "loss": 0.9258,
      "step": 1536
    },
    {
      "epoch": 0.2576049610324311,
      "grad_norm": 0.19083667202820348,
      "learning_rate": 1.0010855535717231e-07,
      "loss": 0.9252,
      "step": 1537
    },
    {
      "epoch": 0.2577725634794268,
      "grad_norm": 0.18832960487736777,
      "learning_rate": 1e-07,
      "loss": 0.92,
      "step": 1538
    },
    {
      "epoch": 0.25794016592642255,
      "grad_norm": 0.22637917784977682,
      "learning_rate": 9.989144464282768e-08,
      "loss": 0.9417,
      "step": 1539
    },
    {
      "epoch": 0.25810776837341826,
      "grad_norm": 0.1707129064881496,
      "learning_rate": 9.978288941357991e-08,
      "loss": 0.8862,
      "step": 1540
    },
    {
      "epoch": 0.25827537082041396,
      "grad_norm": 0.182680571617709,
      "learning_rate": 9.96743344401811e-08,
      "loss": 0.9323,
      "step": 1541
    },
    {
      "epoch": 0.2584429732674097,
      "grad_norm": 0.20790270838631467,
      "learning_rate": 9.956577985055535e-08,
      "loss": 0.9196,
      "step": 1542
    },
    {
      "epoch": 0.25861057571440543,
      "grad_norm": 0.17797093358279306,
      "learning_rate": 9.945722577262631e-08,
      "loss": 0.8971,
      "step": 1543
    },
    {
      "epoch": 0.25877817816140114,
      "grad_norm": 0.17043742757511154,
      "learning_rate": 9.934867233431704e-08,
      "loss": 0.8875,
      "step": 1544
    },
    {
      "epoch": 0.2589457806083969,
      "grad_norm": 0.2184941065069256,
      "learning_rate": 9.92401196635498e-08,
      "loss": 0.9623,
      "step": 1545
    },
    {
      "epoch": 0.2591133830553926,
      "grad_norm": 0.19770029407449635,
      "learning_rate": 9.913156788824595e-08,
      "loss": 0.9305,
      "step": 1546
    },
    {
      "epoch": 0.2592809855023883,
      "grad_norm": 0.20368688671735885,
      "learning_rate": 9.90230171363259e-08,
      "loss": 0.9348,
      "step": 1547
    },
    {
      "epoch": 0.2594485879493841,
      "grad_norm": 0.17107958188135522,
      "learning_rate": 9.891446753570875e-08,
      "loss": 0.8769,
      "step": 1548
    },
    {
      "epoch": 0.2596161903963798,
      "grad_norm": 0.19282534633937676,
      "learning_rate": 9.880591921431226e-08,
      "loss": 0.9302,
      "step": 1549
    },
    {
      "epoch": 0.2597837928433755,
      "grad_norm": 0.19255724373796193,
      "learning_rate": 9.86973723000527e-08,
      "loss": 0.9319,
      "step": 1550
    },
    {
      "epoch": 0.25995139529037126,
      "grad_norm": 0.1843064717014745,
      "learning_rate": 9.858882692084466e-08,
      "loss": 0.9509,
      "step": 1551
    },
    {
      "epoch": 0.26011899773736696,
      "grad_norm": 0.19274496440080743,
      "learning_rate": 9.848028320460089e-08,
      "loss": 0.9143,
      "step": 1552
    },
    {
      "epoch": 0.26028660018436267,
      "grad_norm": 0.2385023060315162,
      "learning_rate": 9.837174127923235e-08,
      "loss": 0.9677,
      "step": 1553
    },
    {
      "epoch": 0.26045420263135843,
      "grad_norm": 0.1886372892768397,
      "learning_rate": 9.826320127264768e-08,
      "loss": 0.9195,
      "step": 1554
    },
    {
      "epoch": 0.26062180507835414,
      "grad_norm": 0.19331697844037152,
      "learning_rate": 9.815466331275335e-08,
      "loss": 0.9674,
      "step": 1555
    },
    {
      "epoch": 0.26078940752534985,
      "grad_norm": 0.17879959081696348,
      "learning_rate": 9.804612752745341e-08,
      "loss": 0.8907,
      "step": 1556
    },
    {
      "epoch": 0.2609570099723456,
      "grad_norm": 0.18828916419094446,
      "learning_rate": 9.793759404464935e-08,
      "loss": 0.926,
      "step": 1557
    },
    {
      "epoch": 0.2611246124193413,
      "grad_norm": 0.1904244876032736,
      "learning_rate": 9.782906299223997e-08,
      "loss": 0.8819,
      "step": 1558
    },
    {
      "epoch": 0.261292214866337,
      "grad_norm": 0.2059349050280002,
      "learning_rate": 9.772053449812117e-08,
      "loss": 0.9545,
      "step": 1559
    },
    {
      "epoch": 0.2614598173133328,
      "grad_norm": 0.21766674271593425,
      "learning_rate": 9.761200869018587e-08,
      "loss": 0.9185,
      "step": 1560
    },
    {
      "epoch": 0.2616274197603285,
      "grad_norm": 0.1773966086891064,
      "learning_rate": 9.750348569632375e-08,
      "loss": 0.9171,
      "step": 1561
    },
    {
      "epoch": 0.2617950222073242,
      "grad_norm": 0.17440978872209137,
      "learning_rate": 9.739496564442124e-08,
      "loss": 0.9197,
      "step": 1562
    },
    {
      "epoch": 0.26196262465431996,
      "grad_norm": 0.18162223251503282,
      "learning_rate": 9.728644866236131e-08,
      "loss": 0.9141,
      "step": 1563
    },
    {
      "epoch": 0.26213022710131567,
      "grad_norm": 0.21683625292181233,
      "learning_rate": 9.71779348780233e-08,
      "loss": 0.9759,
      "step": 1564
    },
    {
      "epoch": 0.2622978295483114,
      "grad_norm": 0.2191918815893909,
      "learning_rate": 9.706942441928273e-08,
      "loss": 1.0042,
      "step": 1565
    },
    {
      "epoch": 0.26246543199530714,
      "grad_norm": 0.18688277356145636,
      "learning_rate": 9.696091741401129e-08,
      "loss": 0.9242,
      "step": 1566
    },
    {
      "epoch": 0.26263303444230285,
      "grad_norm": 0.21303770903121402,
      "learning_rate": 9.68524139900765e-08,
      "loss": 0.9345,
      "step": 1567
    },
    {
      "epoch": 0.2628006368892986,
      "grad_norm": 0.21170938937716133,
      "learning_rate": 9.674391427534177e-08,
      "loss": 0.9572,
      "step": 1568
    },
    {
      "epoch": 0.2629682393362943,
      "grad_norm": 0.17737460050008796,
      "learning_rate": 9.663541839766609e-08,
      "loss": 0.9423,
      "step": 1569
    },
    {
      "epoch": 0.26313584178329,
      "grad_norm": 0.1822940552553249,
      "learning_rate": 9.652692648490386e-08,
      "loss": 0.9411,
      "step": 1570
    },
    {
      "epoch": 0.2633034442302858,
      "grad_norm": 0.1929078820669615,
      "learning_rate": 9.641843866490491e-08,
      "loss": 0.933,
      "step": 1571
    },
    {
      "epoch": 0.2634710466772815,
      "grad_norm": 0.17819771017761385,
      "learning_rate": 9.630995506551419e-08,
      "loss": 0.9326,
      "step": 1572
    },
    {
      "epoch": 0.2636386491242772,
      "grad_norm": 0.19739714978953807,
      "learning_rate": 9.620147581457168e-08,
      "loss": 0.9242,
      "step": 1573
    },
    {
      "epoch": 0.26380625157127296,
      "grad_norm": 0.19035437577248757,
      "learning_rate": 9.609300103991229e-08,
      "loss": 0.9213,
      "step": 1574
    },
    {
      "epoch": 0.26397385401826867,
      "grad_norm": 0.18649720671401437,
      "learning_rate": 9.598453086936559e-08,
      "loss": 0.8902,
      "step": 1575
    },
    {
      "epoch": 0.2641414564652644,
      "grad_norm": 0.2048837539798187,
      "learning_rate": 9.587606543075576e-08,
      "loss": 0.9391,
      "step": 1576
    },
    {
      "epoch": 0.26430905891226014,
      "grad_norm": 0.22441053989580248,
      "learning_rate": 9.576760485190137e-08,
      "loss": 1.0011,
      "step": 1577
    },
    {
      "epoch": 0.26447666135925585,
      "grad_norm": 0.17994549823226189,
      "learning_rate": 9.565914926061524e-08,
      "loss": 0.8949,
      "step": 1578
    },
    {
      "epoch": 0.26464426380625156,
      "grad_norm": 0.16506058439592947,
      "learning_rate": 9.555069878470447e-08,
      "loss": 0.8557,
      "step": 1579
    },
    {
      "epoch": 0.2648118662532473,
      "grad_norm": 0.21129849984547378,
      "learning_rate": 9.544225355196994e-08,
      "loss": 0.9599,
      "step": 1580
    },
    {
      "epoch": 0.264979468700243,
      "grad_norm": 0.19539910820795098,
      "learning_rate": 9.533381369020645e-08,
      "loss": 0.9524,
      "step": 1581
    },
    {
      "epoch": 0.26514707114723873,
      "grad_norm": 0.20566987022360084,
      "learning_rate": 9.522537932720244e-08,
      "loss": 0.9223,
      "step": 1582
    },
    {
      "epoch": 0.2653146735942345,
      "grad_norm": 0.21107407154393534,
      "learning_rate": 9.511695059073987e-08,
      "loss": 0.9475,
      "step": 1583
    },
    {
      "epoch": 0.2654822760412302,
      "grad_norm": 0.17775305202357117,
      "learning_rate": 9.500852760859413e-08,
      "loss": 0.8768,
      "step": 1584
    },
    {
      "epoch": 0.2656498784882259,
      "grad_norm": 0.20613089751577765,
      "learning_rate": 9.490011050853374e-08,
      "loss": 0.9801,
      "step": 1585
    },
    {
      "epoch": 0.2658174809352217,
      "grad_norm": 0.19769061047747716,
      "learning_rate": 9.479169941832035e-08,
      "loss": 0.9249,
      "step": 1586
    },
    {
      "epoch": 0.2659850833822174,
      "grad_norm": 0.17686050066231795,
      "learning_rate": 9.468329446570848e-08,
      "loss": 0.9349,
      "step": 1587
    },
    {
      "epoch": 0.2661526858292131,
      "grad_norm": 0.1839179381034633,
      "learning_rate": 9.457489577844543e-08,
      "loss": 0.8827,
      "step": 1588
    },
    {
      "epoch": 0.26632028827620885,
      "grad_norm": 0.2127202406260534,
      "learning_rate": 9.446650348427116e-08,
      "loss": 0.9343,
      "step": 1589
    },
    {
      "epoch": 0.26648789072320456,
      "grad_norm": 0.21336929874337063,
      "learning_rate": 9.435811771091808e-08,
      "loss": 0.9153,
      "step": 1590
    },
    {
      "epoch": 0.26665549317020026,
      "grad_norm": 0.1951265776442967,
      "learning_rate": 9.424973858611087e-08,
      "loss": 0.922,
      "step": 1591
    },
    {
      "epoch": 0.266823095617196,
      "grad_norm": 0.1974936397149584,
      "learning_rate": 9.414136623756642e-08,
      "loss": 0.9374,
      "step": 1592
    },
    {
      "epoch": 0.26699069806419173,
      "grad_norm": 0.2003108930726311,
      "learning_rate": 9.403300079299363e-08,
      "loss": 0.984,
      "step": 1593
    },
    {
      "epoch": 0.26715830051118744,
      "grad_norm": 0.20518562878297888,
      "learning_rate": 9.392464238009318e-08,
      "loss": 0.9432,
      "step": 1594
    },
    {
      "epoch": 0.2673259029581832,
      "grad_norm": 0.1797757134818971,
      "learning_rate": 9.381629112655769e-08,
      "loss": 0.8818,
      "step": 1595
    },
    {
      "epoch": 0.2674935054051789,
      "grad_norm": 0.19736699651810435,
      "learning_rate": 9.370794716007107e-08,
      "loss": 0.9258,
      "step": 1596
    },
    {
      "epoch": 0.2676611078521746,
      "grad_norm": 0.1795931049980068,
      "learning_rate": 9.359961060830879e-08,
      "loss": 0.9292,
      "step": 1597
    },
    {
      "epoch": 0.2678287102991704,
      "grad_norm": 0.21802023895765313,
      "learning_rate": 9.349128159893757e-08,
      "loss": 0.971,
      "step": 1598
    },
    {
      "epoch": 0.2679963127461661,
      "grad_norm": 0.21189314020658392,
      "learning_rate": 9.338296025961521e-08,
      "loss": 0.9291,
      "step": 1599
    },
    {
      "epoch": 0.2681639151931618,
      "grad_norm": 0.192337514634436,
      "learning_rate": 9.327464671799053e-08,
      "loss": 0.9317,
      "step": 1600
    },
    {
      "epoch": 0.26833151764015756,
      "grad_norm": 0.21269064288966222,
      "learning_rate": 9.316634110170308e-08,
      "loss": 0.9631,
      "step": 1601
    },
    {
      "epoch": 0.26849912008715326,
      "grad_norm": 0.19399029311739346,
      "learning_rate": 9.305804353838316e-08,
      "loss": 0.956,
      "step": 1602
    },
    {
      "epoch": 0.26866672253414897,
      "grad_norm": 0.17946597418479754,
      "learning_rate": 9.294975415565149e-08,
      "loss": 0.9032,
      "step": 1603
    },
    {
      "epoch": 0.26883432498114473,
      "grad_norm": 0.20756840840208046,
      "learning_rate": 9.284147308111915e-08,
      "loss": 0.9378,
      "step": 1604
    },
    {
      "epoch": 0.26900192742814044,
      "grad_norm": 0.19026871969133147,
      "learning_rate": 9.273320044238761e-08,
      "loss": 0.9339,
      "step": 1605
    },
    {
      "epoch": 0.2691695298751362,
      "grad_norm": 0.1838173856224204,
      "learning_rate": 9.262493636704814e-08,
      "loss": 0.8682,
      "step": 1606
    },
    {
      "epoch": 0.2693371323221319,
      "grad_norm": 0.1854496062676279,
      "learning_rate": 9.251668098268208e-08,
      "loss": 0.914,
      "step": 1607
    },
    {
      "epoch": 0.2695047347691276,
      "grad_norm": 0.19015231912688538,
      "learning_rate": 9.240843441686048e-08,
      "loss": 0.9302,
      "step": 1608
    },
    {
      "epoch": 0.2696723372161234,
      "grad_norm": 0.19134282903154914,
      "learning_rate": 9.230019679714404e-08,
      "loss": 0.9197,
      "step": 1609
    },
    {
      "epoch": 0.2698399396631191,
      "grad_norm": 0.1826582906672556,
      "learning_rate": 9.219196825108277e-08,
      "loss": 0.9371,
      "step": 1610
    },
    {
      "epoch": 0.2700075421101148,
      "grad_norm": 0.18210180467585738,
      "learning_rate": 9.208374890621627e-08,
      "loss": 0.9394,
      "step": 1611
    },
    {
      "epoch": 0.27017514455711056,
      "grad_norm": 0.19832718436883764,
      "learning_rate": 9.197553889007299e-08,
      "loss": 0.9743,
      "step": 1612
    },
    {
      "epoch": 0.27034274700410627,
      "grad_norm": 0.20210206430624825,
      "learning_rate": 9.186733833017059e-08,
      "loss": 0.9167,
      "step": 1613
    },
    {
      "epoch": 0.270510349451102,
      "grad_norm": 0.18344240558544397,
      "learning_rate": 9.175914735401551e-08,
      "loss": 0.9233,
      "step": 1614
    },
    {
      "epoch": 0.27067795189809774,
      "grad_norm": 0.2101165521181376,
      "learning_rate": 9.165096608910286e-08,
      "loss": 0.9171,
      "step": 1615
    },
    {
      "epoch": 0.27084555434509344,
      "grad_norm": 0.194899272291471,
      "learning_rate": 9.154279466291643e-08,
      "loss": 0.9349,
      "step": 1616
    },
    {
      "epoch": 0.27101315679208915,
      "grad_norm": 0.19556552254859813,
      "learning_rate": 9.143463320292832e-08,
      "loss": 0.9135,
      "step": 1617
    },
    {
      "epoch": 0.2711807592390849,
      "grad_norm": 0.19964025237322555,
      "learning_rate": 9.13264818365989e-08,
      "loss": 0.8937,
      "step": 1618
    },
    {
      "epoch": 0.2713483616860806,
      "grad_norm": 0.18530147158220658,
      "learning_rate": 9.12183406913766e-08,
      "loss": 0.8894,
      "step": 1619
    },
    {
      "epoch": 0.2715159641330763,
      "grad_norm": 0.20394780907653887,
      "learning_rate": 9.111020989469785e-08,
      "loss": 0.9237,
      "step": 1620
    },
    {
      "epoch": 0.2716835665800721,
      "grad_norm": 0.21274019221240012,
      "learning_rate": 9.100208957398702e-08,
      "loss": 0.9115,
      "step": 1621
    },
    {
      "epoch": 0.2718511690270678,
      "grad_norm": 0.19962306269292351,
      "learning_rate": 9.089397985665587e-08,
      "loss": 0.9571,
      "step": 1622
    },
    {
      "epoch": 0.2720187714740635,
      "grad_norm": 0.17188069641525935,
      "learning_rate": 9.078588087010388e-08,
      "loss": 0.88,
      "step": 1623
    },
    {
      "epoch": 0.27218637392105927,
      "grad_norm": 0.1755794324374738,
      "learning_rate": 9.067779274171775e-08,
      "loss": 0.8781,
      "step": 1624
    },
    {
      "epoch": 0.272353976368055,
      "grad_norm": 0.18693435668573558,
      "learning_rate": 9.056971559887144e-08,
      "loss": 0.9129,
      "step": 1625
    },
    {
      "epoch": 0.2725215788150507,
      "grad_norm": 0.21964219737508164,
      "learning_rate": 9.046164956892601e-08,
      "loss": 0.9777,
      "step": 1626
    },
    {
      "epoch": 0.27268918126204644,
      "grad_norm": 0.2245923018347746,
      "learning_rate": 9.035359477922935e-08,
      "loss": 0.9435,
      "step": 1627
    },
    {
      "epoch": 0.27285678370904215,
      "grad_norm": 0.1707984354848887,
      "learning_rate": 9.024555135711616e-08,
      "loss": 0.8483,
      "step": 1628
    },
    {
      "epoch": 0.27302438615603786,
      "grad_norm": 0.1971448389206231,
      "learning_rate": 9.013751942990765e-08,
      "loss": 0.9078,
      "step": 1629
    },
    {
      "epoch": 0.2731919886030336,
      "grad_norm": 0.20508133622330807,
      "learning_rate": 9.002949912491159e-08,
      "loss": 0.9242,
      "step": 1630
    },
    {
      "epoch": 0.2733595910500293,
      "grad_norm": 0.22461454442947523,
      "learning_rate": 8.992149056942198e-08,
      "loss": 0.9686,
      "step": 1631
    },
    {
      "epoch": 0.27352719349702503,
      "grad_norm": 0.20301988546560065,
      "learning_rate": 8.981349389071908e-08,
      "loss": 0.93,
      "step": 1632
    },
    {
      "epoch": 0.2736947959440208,
      "grad_norm": 0.19448573733781707,
      "learning_rate": 8.970550921606903e-08,
      "loss": 0.906,
      "step": 1633
    },
    {
      "epoch": 0.2738623983910165,
      "grad_norm": 0.18358407472187518,
      "learning_rate": 8.95975366727239e-08,
      "loss": 0.9043,
      "step": 1634
    },
    {
      "epoch": 0.2740300008380122,
      "grad_norm": 0.1799001189805115,
      "learning_rate": 8.948957638792142e-08,
      "loss": 0.9297,
      "step": 1635
    },
    {
      "epoch": 0.274197603285008,
      "grad_norm": 0.16316707863861096,
      "learning_rate": 8.938162848888485e-08,
      "loss": 0.8974,
      "step": 1636
    },
    {
      "epoch": 0.2743652057320037,
      "grad_norm": 0.19009324967738842,
      "learning_rate": 8.927369310282302e-08,
      "loss": 0.9042,
      "step": 1637
    },
    {
      "epoch": 0.2745328081789994,
      "grad_norm": 0.2053753687491867,
      "learning_rate": 8.916577035692978e-08,
      "loss": 0.9372,
      "step": 1638
    },
    {
      "epoch": 0.27470041062599515,
      "grad_norm": 0.20089341740835504,
      "learning_rate": 8.905786037838425e-08,
      "loss": 0.9044,
      "step": 1639
    },
    {
      "epoch": 0.27486801307299086,
      "grad_norm": 0.2085685529988518,
      "learning_rate": 8.894996329435044e-08,
      "loss": 0.9149,
      "step": 1640
    },
    {
      "epoch": 0.27503561551998656,
      "grad_norm": 0.16285600963360483,
      "learning_rate": 8.884207923197716e-08,
      "loss": 0.8648,
      "step": 1641
    },
    {
      "epoch": 0.2752032179669823,
      "grad_norm": 0.18024074403976034,
      "learning_rate": 8.87342083183979e-08,
      "loss": 0.8935,
      "step": 1642
    },
    {
      "epoch": 0.27537082041397803,
      "grad_norm": 0.18831164706503856,
      "learning_rate": 8.862635068073068e-08,
      "loss": 0.9026,
      "step": 1643
    },
    {
      "epoch": 0.2755384228609738,
      "grad_norm": 0.18405938656144646,
      "learning_rate": 8.851850644607784e-08,
      "loss": 0.8978,
      "step": 1644
    },
    {
      "epoch": 0.2757060253079695,
      "grad_norm": 0.1988197194351685,
      "learning_rate": 8.841067574152588e-08,
      "loss": 0.9299,
      "step": 1645
    },
    {
      "epoch": 0.2758736277549652,
      "grad_norm": 0.17975409656278932,
      "learning_rate": 8.83028586941454e-08,
      "loss": 0.9176,
      "step": 1646
    },
    {
      "epoch": 0.276041230201961,
      "grad_norm": 0.1823547925714971,
      "learning_rate": 8.819505543099099e-08,
      "loss": 0.8918,
      "step": 1647
    },
    {
      "epoch": 0.2762088326489567,
      "grad_norm": 0.18340663271027363,
      "learning_rate": 8.808726607910085e-08,
      "loss": 0.8918,
      "step": 1648
    },
    {
      "epoch": 0.2763764350959524,
      "grad_norm": 0.18390908563506778,
      "learning_rate": 8.797949076549684e-08,
      "loss": 0.9092,
      "step": 1649
    },
    {
      "epoch": 0.27654403754294815,
      "grad_norm": 0.1843808267197851,
      "learning_rate": 8.787172961718433e-08,
      "loss": 0.9088,
      "step": 1650
    },
    {
      "epoch": 0.27671163998994386,
      "grad_norm": 0.17670101989328946,
      "learning_rate": 8.776398276115196e-08,
      "loss": 0.9102,
      "step": 1651
    },
    {
      "epoch": 0.27687924243693957,
      "grad_norm": 0.19505719037784383,
      "learning_rate": 8.765625032437145e-08,
      "loss": 0.9097,
      "step": 1652
    },
    {
      "epoch": 0.27704684488393533,
      "grad_norm": 0.1821949047630325,
      "learning_rate": 8.754853243379769e-08,
      "loss": 0.9187,
      "step": 1653
    },
    {
      "epoch": 0.27721444733093104,
      "grad_norm": 0.19952259231287975,
      "learning_rate": 8.744082921636834e-08,
      "loss": 0.9229,
      "step": 1654
    },
    {
      "epoch": 0.27738204977792674,
      "grad_norm": 0.18804416028123033,
      "learning_rate": 8.733314079900371e-08,
      "loss": 0.8778,
      "step": 1655
    },
    {
      "epoch": 0.2775496522249225,
      "grad_norm": 0.19298212662293493,
      "learning_rate": 8.722546730860674e-08,
      "loss": 0.9114,
      "step": 1656
    },
    {
      "epoch": 0.2777172546719182,
      "grad_norm": 0.20711718000789242,
      "learning_rate": 8.711780887206278e-08,
      "loss": 0.9483,
      "step": 1657
    },
    {
      "epoch": 0.2778848571189139,
      "grad_norm": 0.19168241251017443,
      "learning_rate": 8.701016561623944e-08,
      "loss": 0.9276,
      "step": 1658
    },
    {
      "epoch": 0.2780524595659097,
      "grad_norm": 0.19099878191156067,
      "learning_rate": 8.690253766798644e-08,
      "loss": 0.9311,
      "step": 1659
    },
    {
      "epoch": 0.2782200620129054,
      "grad_norm": 0.20026460216671424,
      "learning_rate": 8.679492515413542e-08,
      "loss": 0.919,
      "step": 1660
    },
    {
      "epoch": 0.2783876644599011,
      "grad_norm": 0.20997791685686407,
      "learning_rate": 8.668732820149989e-08,
      "loss": 0.9649,
      "step": 1661
    },
    {
      "epoch": 0.27855526690689686,
      "grad_norm": 0.19437544876372037,
      "learning_rate": 8.65797469368749e-08,
      "loss": 0.9109,
      "step": 1662
    },
    {
      "epoch": 0.27872286935389257,
      "grad_norm": 0.21435037090771908,
      "learning_rate": 8.647218148703725e-08,
      "loss": 0.9368,
      "step": 1663
    },
    {
      "epoch": 0.2788904718008883,
      "grad_norm": 0.18112568537085497,
      "learning_rate": 8.636463197874488e-08,
      "loss": 0.8876,
      "step": 1664
    },
    {
      "epoch": 0.27905807424788404,
      "grad_norm": 0.19109573014557543,
      "learning_rate": 8.625709853873704e-08,
      "loss": 0.9428,
      "step": 1665
    },
    {
      "epoch": 0.27922567669487974,
      "grad_norm": 0.18798991026069653,
      "learning_rate": 8.614958129373401e-08,
      "loss": 0.9047,
      "step": 1666
    },
    {
      "epoch": 0.27939327914187545,
      "grad_norm": 0.1802452337236893,
      "learning_rate": 8.604208037043699e-08,
      "loss": 0.8876,
      "step": 1667
    },
    {
      "epoch": 0.2795608815888712,
      "grad_norm": 0.17774589995292878,
      "learning_rate": 8.593459589552803e-08,
      "loss": 0.8855,
      "step": 1668
    },
    {
      "epoch": 0.2797284840358669,
      "grad_norm": 0.22987402259203799,
      "learning_rate": 8.582712799566968e-08,
      "loss": 0.9372,
      "step": 1669
    },
    {
      "epoch": 0.2798960864828626,
      "grad_norm": 0.1765884695664785,
      "learning_rate": 8.571967679750503e-08,
      "loss": 0.8875,
      "step": 1670
    },
    {
      "epoch": 0.2800636889298584,
      "grad_norm": 0.17047655068694212,
      "learning_rate": 8.561224242765742e-08,
      "loss": 0.8975,
      "step": 1671
    },
    {
      "epoch": 0.2802312913768541,
      "grad_norm": 0.19619706513457774,
      "learning_rate": 8.550482501273043e-08,
      "loss": 0.909,
      "step": 1672
    },
    {
      "epoch": 0.2803988938238498,
      "grad_norm": 0.1812658131389931,
      "learning_rate": 8.539742467930761e-08,
      "loss": 0.9059,
      "step": 1673
    },
    {
      "epoch": 0.28056649627084557,
      "grad_norm": 0.1893821993579802,
      "learning_rate": 8.529004155395243e-08,
      "loss": 0.9364,
      "step": 1674
    },
    {
      "epoch": 0.2807340987178413,
      "grad_norm": 0.18223526058583314,
      "learning_rate": 8.518267576320804e-08,
      "loss": 0.9122,
      "step": 1675
    },
    {
      "epoch": 0.280901701164837,
      "grad_norm": 0.19123948738227442,
      "learning_rate": 8.507532743359719e-08,
      "loss": 0.9004,
      "step": 1676
    },
    {
      "epoch": 0.28106930361183274,
      "grad_norm": 0.1955820124323723,
      "learning_rate": 8.496799669162205e-08,
      "loss": 0.9335,
      "step": 1677
    },
    {
      "epoch": 0.28123690605882845,
      "grad_norm": 0.21098391504332242,
      "learning_rate": 8.486068366376397e-08,
      "loss": 0.9479,
      "step": 1678
    },
    {
      "epoch": 0.28140450850582416,
      "grad_norm": 0.18026994991025186,
      "learning_rate": 8.475338847648361e-08,
      "loss": 0.8931,
      "step": 1679
    },
    {
      "epoch": 0.2815721109528199,
      "grad_norm": 0.17828472992623526,
      "learning_rate": 8.464611125622049e-08,
      "loss": 0.896,
      "step": 1680
    },
    {
      "epoch": 0.28173971339981563,
      "grad_norm": 0.18166609715458373,
      "learning_rate": 8.453885212939292e-08,
      "loss": 0.883,
      "step": 1681
    },
    {
      "epoch": 0.2819073158468114,
      "grad_norm": 0.1989884413729368,
      "learning_rate": 8.443161122239799e-08,
      "loss": 0.9188,
      "step": 1682
    },
    {
      "epoch": 0.2820749182938071,
      "grad_norm": 0.203580776710553,
      "learning_rate": 8.432438866161123e-08,
      "loss": 0.8726,
      "step": 1683
    },
    {
      "epoch": 0.2822425207408028,
      "grad_norm": 0.16739704904790262,
      "learning_rate": 8.421718457338662e-08,
      "loss": 0.8418,
      "step": 1684
    },
    {
      "epoch": 0.28241012318779857,
      "grad_norm": 0.19456813432490327,
      "learning_rate": 8.410999908405633e-08,
      "loss": 0.9342,
      "step": 1685
    },
    {
      "epoch": 0.2825777256347943,
      "grad_norm": 0.21044109890773685,
      "learning_rate": 8.400283231993064e-08,
      "loss": 0.9262,
      "step": 1686
    },
    {
      "epoch": 0.28274532808179,
      "grad_norm": 0.19651992620067232,
      "learning_rate": 8.389568440729776e-08,
      "loss": 0.9245,
      "step": 1687
    },
    {
      "epoch": 0.28291293052878574,
      "grad_norm": 0.195138265161823,
      "learning_rate": 8.378855547242357e-08,
      "loss": 0.9333,
      "step": 1688
    },
    {
      "epoch": 0.28308053297578145,
      "grad_norm": 0.19496593668319734,
      "learning_rate": 8.368144564155184e-08,
      "loss": 0.8997,
      "step": 1689
    },
    {
      "epoch": 0.28324813542277716,
      "grad_norm": 0.19918002366104548,
      "learning_rate": 8.357435504090359e-08,
      "loss": 0.901,
      "step": 1690
    },
    {
      "epoch": 0.2834157378697729,
      "grad_norm": 0.1618569621225145,
      "learning_rate": 8.346728379667727e-08,
      "loss": 0.9109,
      "step": 1691
    },
    {
      "epoch": 0.28358334031676863,
      "grad_norm": 0.1723133345429209,
      "learning_rate": 8.336023203504852e-08,
      "loss": 0.8558,
      "step": 1692
    },
    {
      "epoch": 0.28375094276376434,
      "grad_norm": 0.17540594305549545,
      "learning_rate": 8.325319988217004e-08,
      "loss": 0.9015,
      "step": 1693
    },
    {
      "epoch": 0.2839185452107601,
      "grad_norm": 0.18990866187089317,
      "learning_rate": 8.314618746417134e-08,
      "loss": 0.9529,
      "step": 1694
    },
    {
      "epoch": 0.2840861476577558,
      "grad_norm": 0.15648485022882547,
      "learning_rate": 8.303919490715881e-08,
      "loss": 0.8362,
      "step": 1695
    },
    {
      "epoch": 0.2842537501047515,
      "grad_norm": 0.2092507644931485,
      "learning_rate": 8.293222233721533e-08,
      "loss": 0.9503,
      "step": 1696
    },
    {
      "epoch": 0.2844213525517473,
      "grad_norm": 0.19118817099691096,
      "learning_rate": 8.282526988040022e-08,
      "loss": 0.9325,
      "step": 1697
    },
    {
      "epoch": 0.284588954998743,
      "grad_norm": 0.19394011210787987,
      "learning_rate": 8.271833766274916e-08,
      "loss": 0.9412,
      "step": 1698
    },
    {
      "epoch": 0.2847565574457387,
      "grad_norm": 0.17437553182655507,
      "learning_rate": 8.261142581027392e-08,
      "loss": 0.9138,
      "step": 1699
    },
    {
      "epoch": 0.28492415989273445,
      "grad_norm": 0.22285418984901875,
      "learning_rate": 8.250453444896236e-08,
      "loss": 0.947,
      "step": 1700
    },
    {
      "epoch": 0.28509176233973016,
      "grad_norm": 0.20230930405055367,
      "learning_rate": 8.23976637047781e-08,
      "loss": 0.9364,
      "step": 1701
    },
    {
      "epoch": 0.28525936478672587,
      "grad_norm": 0.19905461591494056,
      "learning_rate": 8.229081370366053e-08,
      "loss": 0.9107,
      "step": 1702
    },
    {
      "epoch": 0.28542696723372163,
      "grad_norm": 0.18837913065937154,
      "learning_rate": 8.218398457152453e-08,
      "loss": 0.9151,
      "step": 1703
    },
    {
      "epoch": 0.28559456968071734,
      "grad_norm": 0.1937812323209485,
      "learning_rate": 8.20771764342604e-08,
      "loss": 0.922,
      "step": 1704
    },
    {
      "epoch": 0.28576217212771304,
      "grad_norm": 0.20982455055206156,
      "learning_rate": 8.19703894177338e-08,
      "loss": 0.9857,
      "step": 1705
    },
    {
      "epoch": 0.2859297745747088,
      "grad_norm": 0.1662685171700633,
      "learning_rate": 8.18636236477854e-08,
      "loss": 0.894,
      "step": 1706
    },
    {
      "epoch": 0.2860973770217045,
      "grad_norm": 0.18711858493376915,
      "learning_rate": 8.175687925023081e-08,
      "loss": 0.9071,
      "step": 1707
    },
    {
      "epoch": 0.2862649794687002,
      "grad_norm": 0.18937945378403026,
      "learning_rate": 8.165015635086055e-08,
      "loss": 0.9446,
      "step": 1708
    },
    {
      "epoch": 0.286432581915696,
      "grad_norm": 0.187062541196743,
      "learning_rate": 8.154345507543972e-08,
      "loss": 0.9105,
      "step": 1709
    },
    {
      "epoch": 0.2866001843626917,
      "grad_norm": 0.1727534438768274,
      "learning_rate": 8.143677554970795e-08,
      "loss": 0.9015,
      "step": 1710
    },
    {
      "epoch": 0.2867677868096874,
      "grad_norm": 0.19437101109485125,
      "learning_rate": 8.133011789937936e-08,
      "loss": 0.8944,
      "step": 1711
    },
    {
      "epoch": 0.28693538925668316,
      "grad_norm": 0.18314431028556893,
      "learning_rate": 8.122348225014212e-08,
      "loss": 0.8966,
      "step": 1712
    },
    {
      "epoch": 0.28710299170367887,
      "grad_norm": 0.1990536222301002,
      "learning_rate": 8.111686872765858e-08,
      "loss": 0.9112,
      "step": 1713
    },
    {
      "epoch": 0.2872705941506746,
      "grad_norm": 0.19089733774448508,
      "learning_rate": 8.101027745756496e-08,
      "loss": 0.9222,
      "step": 1714
    },
    {
      "epoch": 0.28743819659767034,
      "grad_norm": 0.19423726057015558,
      "learning_rate": 8.090370856547125e-08,
      "loss": 0.9057,
      "step": 1715
    },
    {
      "epoch": 0.28760579904466604,
      "grad_norm": 0.18559511372448606,
      "learning_rate": 8.079716217696118e-08,
      "loss": 0.9109,
      "step": 1716
    },
    {
      "epoch": 0.28777340149166175,
      "grad_norm": 0.17791746285141788,
      "learning_rate": 8.069063841759181e-08,
      "loss": 0.9192,
      "step": 1717
    },
    {
      "epoch": 0.2879410039386575,
      "grad_norm": 0.18034099839998993,
      "learning_rate": 8.058413741289366e-08,
      "loss": 0.8608,
      "step": 1718
    },
    {
      "epoch": 0.2881086063856532,
      "grad_norm": 0.19970208645375764,
      "learning_rate": 8.047765928837034e-08,
      "loss": 0.938,
      "step": 1719
    },
    {
      "epoch": 0.288276208832649,
      "grad_norm": 0.19211335735343124,
      "learning_rate": 8.037120416949853e-08,
      "loss": 0.899,
      "step": 1720
    },
    {
      "epoch": 0.2884438112796447,
      "grad_norm": 0.180711666906609,
      "learning_rate": 8.026477218172785e-08,
      "loss": 0.9065,
      "step": 1721
    },
    {
      "epoch": 0.2886114137266404,
      "grad_norm": 0.19337102762409958,
      "learning_rate": 8.01583634504806e-08,
      "loss": 0.9227,
      "step": 1722
    },
    {
      "epoch": 0.28877901617363616,
      "grad_norm": 0.19463218263300075,
      "learning_rate": 8.005197810115168e-08,
      "loss": 0.9391,
      "step": 1723
    },
    {
      "epoch": 0.28894661862063187,
      "grad_norm": 0.18278492769902668,
      "learning_rate": 7.994561625910844e-08,
      "loss": 0.9354,
      "step": 1724
    },
    {
      "epoch": 0.2891142210676276,
      "grad_norm": 0.20066202156701515,
      "learning_rate": 7.983927804969053e-08,
      "loss": 0.9546,
      "step": 1725
    },
    {
      "epoch": 0.28928182351462334,
      "grad_norm": 0.17687239499847646,
      "learning_rate": 7.973296359820981e-08,
      "loss": 0.893,
      "step": 1726
    },
    {
      "epoch": 0.28944942596161904,
      "grad_norm": 0.17691340573031883,
      "learning_rate": 7.962667302995003e-08,
      "loss": 0.9135,
      "step": 1727
    },
    {
      "epoch": 0.28961702840861475,
      "grad_norm": 0.1970945636837427,
      "learning_rate": 7.95204064701669e-08,
      "loss": 0.9424,
      "step": 1728
    },
    {
      "epoch": 0.2897846308556105,
      "grad_norm": 0.20564628071580124,
      "learning_rate": 7.941416404408779e-08,
      "loss": 0.9397,
      "step": 1729
    },
    {
      "epoch": 0.2899522333026062,
      "grad_norm": 0.16103034667845073,
      "learning_rate": 7.930794587691158e-08,
      "loss": 0.8889,
      "step": 1730
    },
    {
      "epoch": 0.29011983574960193,
      "grad_norm": 0.17447815632011063,
      "learning_rate": 7.920175209380864e-08,
      "loss": 0.8883,
      "step": 1731
    },
    {
      "epoch": 0.2902874381965977,
      "grad_norm": 0.18856015223760833,
      "learning_rate": 7.909558281992064e-08,
      "loss": 0.9391,
      "step": 1732
    },
    {
      "epoch": 0.2904550406435934,
      "grad_norm": 0.18881958603832374,
      "learning_rate": 7.898943818036023e-08,
      "loss": 0.9308,
      "step": 1733
    },
    {
      "epoch": 0.2906226430905891,
      "grad_norm": 0.18657949096768478,
      "learning_rate": 7.888331830021117e-08,
      "loss": 0.9213,
      "step": 1734
    },
    {
      "epoch": 0.29079024553758487,
      "grad_norm": 0.19671096381016437,
      "learning_rate": 7.877722330452794e-08,
      "loss": 0.9314,
      "step": 1735
    },
    {
      "epoch": 0.2909578479845806,
      "grad_norm": 0.1702300516903879,
      "learning_rate": 7.867115331833572e-08,
      "loss": 0.8915,
      "step": 1736
    },
    {
      "epoch": 0.2911254504315763,
      "grad_norm": 0.16404367043680737,
      "learning_rate": 7.85651084666303e-08,
      "loss": 0.8766,
      "step": 1737
    },
    {
      "epoch": 0.29129305287857205,
      "grad_norm": 0.17220781299920318,
      "learning_rate": 7.845908887437778e-08,
      "loss": 0.8938,
      "step": 1738
    },
    {
      "epoch": 0.29146065532556775,
      "grad_norm": 0.17909629474360225,
      "learning_rate": 7.835309466651443e-08,
      "loss": 0.8898,
      "step": 1739
    },
    {
      "epoch": 0.29162825777256346,
      "grad_norm": 0.18835212824496234,
      "learning_rate": 7.824712596794673e-08,
      "loss": 0.9221,
      "step": 1740
    },
    {
      "epoch": 0.2917958602195592,
      "grad_norm": 0.19196102102404675,
      "learning_rate": 7.814118290355102e-08,
      "loss": 0.9245,
      "step": 1741
    },
    {
      "epoch": 0.29196346266655493,
      "grad_norm": 0.15623012187124316,
      "learning_rate": 7.803526559817348e-08,
      "loss": 0.8458,
      "step": 1742
    },
    {
      "epoch": 0.29213106511355064,
      "grad_norm": 0.2025610895960505,
      "learning_rate": 7.792937417662993e-08,
      "loss": 0.9093,
      "step": 1743
    },
    {
      "epoch": 0.2922986675605464,
      "grad_norm": 0.17490776983881537,
      "learning_rate": 7.782350876370562e-08,
      "loss": 0.8896,
      "step": 1744
    },
    {
      "epoch": 0.2924662700075421,
      "grad_norm": 0.2082933166223772,
      "learning_rate": 7.771766948415524e-08,
      "loss": 0.9706,
      "step": 1745
    },
    {
      "epoch": 0.2926338724545378,
      "grad_norm": 0.17701960613163079,
      "learning_rate": 7.761185646270259e-08,
      "loss": 0.8899,
      "step": 1746
    },
    {
      "epoch": 0.2928014749015336,
      "grad_norm": 0.17963869008102964,
      "learning_rate": 7.750606982404064e-08,
      "loss": 0.8594,
      "step": 1747
    },
    {
      "epoch": 0.2929690773485293,
      "grad_norm": 0.18493309827761023,
      "learning_rate": 7.740030969283122e-08,
      "loss": 0.909,
      "step": 1748
    },
    {
      "epoch": 0.293136679795525,
      "grad_norm": 0.18183272457280858,
      "learning_rate": 7.729457619370487e-08,
      "loss": 0.922,
      "step": 1749
    },
    {
      "epoch": 0.29330428224252075,
      "grad_norm": 0.1823747493030349,
      "learning_rate": 7.718886945126079e-08,
      "loss": 0.8676,
      "step": 1750
    },
    {
      "epoch": 0.29347188468951646,
      "grad_norm": 0.19441872090788845,
      "learning_rate": 7.708318959006668e-08,
      "loss": 0.9088,
      "step": 1751
    },
    {
      "epoch": 0.29363948713651217,
      "grad_norm": 0.19064215694759745,
      "learning_rate": 7.69775367346585e-08,
      "loss": 0.9188,
      "step": 1752
    },
    {
      "epoch": 0.29380708958350793,
      "grad_norm": 0.1767745813760241,
      "learning_rate": 7.687191100954045e-08,
      "loss": 0.9239,
      "step": 1753
    },
    {
      "epoch": 0.29397469203050364,
      "grad_norm": 0.2145141453273042,
      "learning_rate": 7.67663125391847e-08,
      "loss": 0.9551,
      "step": 1754
    },
    {
      "epoch": 0.29414229447749934,
      "grad_norm": 0.18492294831977432,
      "learning_rate": 7.666074144803138e-08,
      "loss": 0.8653,
      "step": 1755
    },
    {
      "epoch": 0.2943098969244951,
      "grad_norm": 0.18368708638264417,
      "learning_rate": 7.655519786048822e-08,
      "loss": 0.881,
      "step": 1756
    },
    {
      "epoch": 0.2944774993714908,
      "grad_norm": 0.1837974444268386,
      "learning_rate": 7.644968190093067e-08,
      "loss": 0.8921,
      "step": 1757
    },
    {
      "epoch": 0.2946451018184866,
      "grad_norm": 0.21262425422351705,
      "learning_rate": 7.634419369370157e-08,
      "loss": 0.938,
      "step": 1758
    },
    {
      "epoch": 0.2948127042654823,
      "grad_norm": 0.20582003806121144,
      "learning_rate": 7.623873336311107e-08,
      "loss": 0.937,
      "step": 1759
    },
    {
      "epoch": 0.294980306712478,
      "grad_norm": 0.1881526432148505,
      "learning_rate": 7.613330103343646e-08,
      "loss": 0.8997,
      "step": 1760
    },
    {
      "epoch": 0.29514790915947375,
      "grad_norm": 0.1946466545884575,
      "learning_rate": 7.602789682892202e-08,
      "loss": 0.9037,
      "step": 1761
    },
    {
      "epoch": 0.29531551160646946,
      "grad_norm": 0.1683435840188057,
      "learning_rate": 7.592252087377889e-08,
      "loss": 0.8872,
      "step": 1762
    },
    {
      "epoch": 0.29548311405346517,
      "grad_norm": 0.18238204430081634,
      "learning_rate": 7.581717329218498e-08,
      "loss": 0.8978,
      "step": 1763
    },
    {
      "epoch": 0.29565071650046093,
      "grad_norm": 0.18907355032607745,
      "learning_rate": 7.57118542082847e-08,
      "loss": 0.9032,
      "step": 1764
    },
    {
      "epoch": 0.29581831894745664,
      "grad_norm": 0.2010795642313979,
      "learning_rate": 7.560656374618884e-08,
      "loss": 0.8898,
      "step": 1765
    },
    {
      "epoch": 0.29598592139445234,
      "grad_norm": 0.19629209842765652,
      "learning_rate": 7.550130202997456e-08,
      "loss": 0.9004,
      "step": 1766
    },
    {
      "epoch": 0.2961535238414481,
      "grad_norm": 0.18482393457639906,
      "learning_rate": 7.539606918368506e-08,
      "loss": 0.9036,
      "step": 1767
    },
    {
      "epoch": 0.2963211262884438,
      "grad_norm": 0.17150334909723225,
      "learning_rate": 7.529086533132963e-08,
      "loss": 0.8798,
      "step": 1768
    },
    {
      "epoch": 0.2964887287354395,
      "grad_norm": 0.1833005399280927,
      "learning_rate": 7.518569059688324e-08,
      "loss": 0.9131,
      "step": 1769
    },
    {
      "epoch": 0.2966563311824353,
      "grad_norm": 0.20365632980422754,
      "learning_rate": 7.508054510428667e-08,
      "loss": 0.917,
      "step": 1770
    },
    {
      "epoch": 0.296823933629431,
      "grad_norm": 0.1958440751365385,
      "learning_rate": 7.497542897744621e-08,
      "loss": 0.9108,
      "step": 1771
    },
    {
      "epoch": 0.2969915360764267,
      "grad_norm": 0.19249875544381448,
      "learning_rate": 7.48703423402335e-08,
      "loss": 0.9579,
      "step": 1772
    },
    {
      "epoch": 0.29715913852342246,
      "grad_norm": 0.19101447684577044,
      "learning_rate": 7.476528531648543e-08,
      "loss": 0.9173,
      "step": 1773
    },
    {
      "epoch": 0.29732674097041817,
      "grad_norm": 0.2012917374788699,
      "learning_rate": 7.466025803000412e-08,
      "loss": 0.9495,
      "step": 1774
    },
    {
      "epoch": 0.2974943434174139,
      "grad_norm": 0.17263945710461207,
      "learning_rate": 7.455526060455646e-08,
      "loss": 0.8906,
      "step": 1775
    },
    {
      "epoch": 0.29766194586440964,
      "grad_norm": 0.17439918389130038,
      "learning_rate": 7.44502931638743e-08,
      "loss": 0.8653,
      "step": 1776
    },
    {
      "epoch": 0.29782954831140535,
      "grad_norm": 0.19849223452840237,
      "learning_rate": 7.434535583165407e-08,
      "loss": 0.931,
      "step": 1777
    },
    {
      "epoch": 0.29799715075840105,
      "grad_norm": 0.18544937919472884,
      "learning_rate": 7.42404487315567e-08,
      "loss": 0.8954,
      "step": 1778
    },
    {
      "epoch": 0.2981647532053968,
      "grad_norm": 0.16615380570261412,
      "learning_rate": 7.413557198720764e-08,
      "loss": 0.8674,
      "step": 1779
    },
    {
      "epoch": 0.2983323556523925,
      "grad_norm": 0.18780637062981312,
      "learning_rate": 7.403072572219639e-08,
      "loss": 0.9081,
      "step": 1780
    },
    {
      "epoch": 0.29849995809938823,
      "grad_norm": 0.2021901828757338,
      "learning_rate": 7.392591006007666e-08,
      "loss": 0.9342,
      "step": 1781
    },
    {
      "epoch": 0.298667560546384,
      "grad_norm": 0.18323854070151818,
      "learning_rate": 7.3821125124366e-08,
      "loss": 0.8705,
      "step": 1782
    },
    {
      "epoch": 0.2988351629933797,
      "grad_norm": 0.2052151400183302,
      "learning_rate": 7.37163710385458e-08,
      "loss": 0.9489,
      "step": 1783
    },
    {
      "epoch": 0.2990027654403754,
      "grad_norm": 0.1787046836665105,
      "learning_rate": 7.361164792606112e-08,
      "loss": 0.8784,
      "step": 1784
    },
    {
      "epoch": 0.29917036788737117,
      "grad_norm": 0.17945340070990273,
      "learning_rate": 7.350695591032048e-08,
      "loss": 0.8837,
      "step": 1785
    },
    {
      "epoch": 0.2993379703343669,
      "grad_norm": 0.1970989661753106,
      "learning_rate": 7.340229511469577e-08,
      "loss": 0.8757,
      "step": 1786
    },
    {
      "epoch": 0.2995055727813626,
      "grad_norm": 0.1695389182433334,
      "learning_rate": 7.329766566252211e-08,
      "loss": 0.8651,
      "step": 1787
    },
    {
      "epoch": 0.29967317522835835,
      "grad_norm": 0.2128887058354598,
      "learning_rate": 7.319306767709761e-08,
      "loss": 0.9236,
      "step": 1788
    },
    {
      "epoch": 0.29984077767535405,
      "grad_norm": 0.1957536241934109,
      "learning_rate": 7.308850128168339e-08,
      "loss": 0.8847,
      "step": 1789
    },
    {
      "epoch": 0.30000838012234976,
      "grad_norm": 0.1966865793987873,
      "learning_rate": 7.298396659950333e-08,
      "loss": 0.9644,
      "step": 1790
    },
    {
      "epoch": 0.3001759825693455,
      "grad_norm": 0.20502159174596035,
      "learning_rate": 7.287946375374385e-08,
      "loss": 0.9433,
      "step": 1791
    },
    {
      "epoch": 0.30034358501634123,
      "grad_norm": 0.19008700024575173,
      "learning_rate": 7.277499286755393e-08,
      "loss": 0.9222,
      "step": 1792
    },
    {
      "epoch": 0.30051118746333694,
      "grad_norm": 0.19166330024072356,
      "learning_rate": 7.267055406404489e-08,
      "loss": 0.9345,
      "step": 1793
    },
    {
      "epoch": 0.3006787899103327,
      "grad_norm": 0.20181420464376787,
      "learning_rate": 7.25661474662902e-08,
      "loss": 0.8974,
      "step": 1794
    },
    {
      "epoch": 0.3008463923573284,
      "grad_norm": 0.18251866067495615,
      "learning_rate": 7.246177319732543e-08,
      "loss": 0.917,
      "step": 1795
    },
    {
      "epoch": 0.30101399480432417,
      "grad_norm": 0.17269370702336215,
      "learning_rate": 7.235743138014802e-08,
      "loss": 0.8785,
      "step": 1796
    },
    {
      "epoch": 0.3011815972513199,
      "grad_norm": 0.18110666455754695,
      "learning_rate": 7.225312213771722e-08,
      "loss": 0.9007,
      "step": 1797
    },
    {
      "epoch": 0.3013491996983156,
      "grad_norm": 0.17267323643207222,
      "learning_rate": 7.214884559295377e-08,
      "loss": 0.9001,
      "step": 1798
    },
    {
      "epoch": 0.30151680214531135,
      "grad_norm": 0.17954416663171704,
      "learning_rate": 7.204460186873994e-08,
      "loss": 0.8874,
      "step": 1799
    },
    {
      "epoch": 0.30168440459230705,
      "grad_norm": 0.15420362096633164,
      "learning_rate": 7.194039108791945e-08,
      "loss": 0.8554,
      "step": 1800
    },
    {
      "epoch": 0.30185200703930276,
      "grad_norm": 0.20492724245421695,
      "learning_rate": 7.183621337329702e-08,
      "loss": 0.9278,
      "step": 1801
    },
    {
      "epoch": 0.3020196094862985,
      "grad_norm": 0.18375990003076914,
      "learning_rate": 7.173206884763845e-08,
      "loss": 0.9241,
      "step": 1802
    },
    {
      "epoch": 0.30218721193329423,
      "grad_norm": 0.19685195065558733,
      "learning_rate": 7.162795763367048e-08,
      "loss": 0.8916,
      "step": 1803
    },
    {
      "epoch": 0.30235481438028994,
      "grad_norm": 0.17776205101550868,
      "learning_rate": 7.152387985408058e-08,
      "loss": 0.9125,
      "step": 1804
    },
    {
      "epoch": 0.3025224168272857,
      "grad_norm": 0.20192454315287645,
      "learning_rate": 7.141983563151677e-08,
      "loss": 0.9139,
      "step": 1805
    },
    {
      "epoch": 0.3026900192742814,
      "grad_norm": 0.20016823425606553,
      "learning_rate": 7.13158250885876e-08,
      "loss": 0.9148,
      "step": 1806
    },
    {
      "epoch": 0.3028576217212771,
      "grad_norm": 0.18812267147006673,
      "learning_rate": 7.121184834786188e-08,
      "loss": 0.9128,
      "step": 1807
    },
    {
      "epoch": 0.3030252241682729,
      "grad_norm": 0.1893989872847782,
      "learning_rate": 7.110790553186857e-08,
      "loss": 0.9276,
      "step": 1808
    },
    {
      "epoch": 0.3031928266152686,
      "grad_norm": 0.17946202955064539,
      "learning_rate": 7.10039967630967e-08,
      "loss": 0.9067,
      "step": 1809
    },
    {
      "epoch": 0.3033604290622643,
      "grad_norm": 0.1956450151126611,
      "learning_rate": 7.090012216399516e-08,
      "loss": 0.9435,
      "step": 1810
    },
    {
      "epoch": 0.30352803150926005,
      "grad_norm": 0.18871973337414094,
      "learning_rate": 7.079628185697257e-08,
      "loss": 0.9294,
      "step": 1811
    },
    {
      "epoch": 0.30369563395625576,
      "grad_norm": 0.18157832851138947,
      "learning_rate": 7.069247596439716e-08,
      "loss": 0.9094,
      "step": 1812
    },
    {
      "epoch": 0.30386323640325147,
      "grad_norm": 0.1690779995032822,
      "learning_rate": 7.058870460859656e-08,
      "loss": 0.8688,
      "step": 1813
    },
    {
      "epoch": 0.30403083885024723,
      "grad_norm": 0.18129325437896426,
      "learning_rate": 7.048496791185777e-08,
      "loss": 0.9227,
      "step": 1814
    },
    {
      "epoch": 0.30419844129724294,
      "grad_norm": 0.17812503064183466,
      "learning_rate": 7.03812659964268e-08,
      "loss": 0.881,
      "step": 1815
    },
    {
      "epoch": 0.30436604374423865,
      "grad_norm": 0.17006811126061158,
      "learning_rate": 7.027759898450891e-08,
      "loss": 0.8661,
      "step": 1816
    },
    {
      "epoch": 0.3045336461912344,
      "grad_norm": 0.19716845570795666,
      "learning_rate": 7.017396699826803e-08,
      "loss": 0.9035,
      "step": 1817
    },
    {
      "epoch": 0.3047012486382301,
      "grad_norm": 0.19959869759494425,
      "learning_rate": 7.007037015982687e-08,
      "loss": 0.9127,
      "step": 1818
    },
    {
      "epoch": 0.3048688510852258,
      "grad_norm": 0.2168203571774005,
      "learning_rate": 6.996680859126676e-08,
      "loss": 0.9707,
      "step": 1819
    },
    {
      "epoch": 0.3050364535322216,
      "grad_norm": 0.20794255202850442,
      "learning_rate": 6.98632824146274e-08,
      "loss": 0.9485,
      "step": 1820
    },
    {
      "epoch": 0.3052040559792173,
      "grad_norm": 0.20168825607345156,
      "learning_rate": 6.975979175190687e-08,
      "loss": 0.9243,
      "step": 1821
    },
    {
      "epoch": 0.305371658426213,
      "grad_norm": 0.1928146059131723,
      "learning_rate": 6.965633672506133e-08,
      "loss": 0.933,
      "step": 1822
    },
    {
      "epoch": 0.30553926087320876,
      "grad_norm": 0.17298609168895787,
      "learning_rate": 6.955291745600498e-08,
      "loss": 0.8817,
      "step": 1823
    },
    {
      "epoch": 0.30570686332020447,
      "grad_norm": 0.16440977205765644,
      "learning_rate": 6.944953406660984e-08,
      "loss": 0.8514,
      "step": 1824
    },
    {
      "epoch": 0.3058744657672002,
      "grad_norm": 0.19652337960218208,
      "learning_rate": 6.934618667870567e-08,
      "loss": 0.8959,
      "step": 1825
    },
    {
      "epoch": 0.30604206821419594,
      "grad_norm": 0.18248445853779097,
      "learning_rate": 6.924287541407987e-08,
      "loss": 0.9284,
      "step": 1826
    },
    {
      "epoch": 0.30620967066119165,
      "grad_norm": 0.20154263939537617,
      "learning_rate": 6.913960039447717e-08,
      "loss": 0.9337,
      "step": 1827
    },
    {
      "epoch": 0.30637727310818735,
      "grad_norm": 0.18234163503040238,
      "learning_rate": 6.903636174159964e-08,
      "loss": 0.9223,
      "step": 1828
    },
    {
      "epoch": 0.3065448755551831,
      "grad_norm": 0.1727666179585064,
      "learning_rate": 6.893315957710649e-08,
      "loss": 0.9362,
      "step": 1829
    },
    {
      "epoch": 0.3067124780021788,
      "grad_norm": 0.18590993372817025,
      "learning_rate": 6.882999402261395e-08,
      "loss": 0.8791,
      "step": 1830
    },
    {
      "epoch": 0.30688008044917453,
      "grad_norm": 0.1903181072152287,
      "learning_rate": 6.8726865199695e-08,
      "loss": 0.8944,
      "step": 1831
    },
    {
      "epoch": 0.3070476828961703,
      "grad_norm": 0.2072758106494689,
      "learning_rate": 6.86237732298795e-08,
      "loss": 0.9203,
      "step": 1832
    },
    {
      "epoch": 0.307215285343166,
      "grad_norm": 0.18830909037940483,
      "learning_rate": 6.852071823465382e-08,
      "loss": 0.9053,
      "step": 1833
    },
    {
      "epoch": 0.30738288779016176,
      "grad_norm": 0.1873099099696868,
      "learning_rate": 6.841770033546066e-08,
      "loss": 0.8908,
      "step": 1834
    },
    {
      "epoch": 0.30755049023715747,
      "grad_norm": 0.16883542916575725,
      "learning_rate": 6.831471965369913e-08,
      "loss": 0.8711,
      "step": 1835
    },
    {
      "epoch": 0.3077180926841532,
      "grad_norm": 0.17611566747148133,
      "learning_rate": 6.821177631072441e-08,
      "loss": 0.9216,
      "step": 1836
    },
    {
      "epoch": 0.30788569513114894,
      "grad_norm": 0.1943452524709528,
      "learning_rate": 6.810887042784776e-08,
      "loss": 0.9227,
      "step": 1837
    },
    {
      "epoch": 0.30805329757814465,
      "grad_norm": 0.18528780435949133,
      "learning_rate": 6.800600212633618e-08,
      "loss": 0.8753,
      "step": 1838
    },
    {
      "epoch": 0.30822090002514035,
      "grad_norm": 0.19085256905500586,
      "learning_rate": 6.790317152741248e-08,
      "loss": 0.9353,
      "step": 1839
    },
    {
      "epoch": 0.3083885024721361,
      "grad_norm": 0.22682001996432027,
      "learning_rate": 6.780037875225502e-08,
      "loss": 0.9632,
      "step": 1840
    },
    {
      "epoch": 0.3085561049191318,
      "grad_norm": 0.1778897658641129,
      "learning_rate": 6.769762392199746e-08,
      "loss": 0.8935,
      "step": 1841
    },
    {
      "epoch": 0.30872370736612753,
      "grad_norm": 0.1987059213069955,
      "learning_rate": 6.759490715772905e-08,
      "loss": 0.9242,
      "step": 1842
    },
    {
      "epoch": 0.3088913098131233,
      "grad_norm": 0.19138526993588573,
      "learning_rate": 6.749222858049382e-08,
      "loss": 0.9147,
      "step": 1843
    },
    {
      "epoch": 0.309058912260119,
      "grad_norm": 0.1624824758574282,
      "learning_rate": 6.738958831129105e-08,
      "loss": 0.8804,
      "step": 1844
    },
    {
      "epoch": 0.3092265147071147,
      "grad_norm": 0.1867597918691358,
      "learning_rate": 6.728698647107474e-08,
      "loss": 0.9034,
      "step": 1845
    },
    {
      "epoch": 0.30939411715411047,
      "grad_norm": 0.16979942965007852,
      "learning_rate": 6.718442318075371e-08,
      "loss": 0.8765,
      "step": 1846
    },
    {
      "epoch": 0.3095617196011062,
      "grad_norm": 0.19335939352050313,
      "learning_rate": 6.708189856119127e-08,
      "loss": 0.9791,
      "step": 1847
    },
    {
      "epoch": 0.3097293220481019,
      "grad_norm": 0.19952837047422325,
      "learning_rate": 6.697941273320522e-08,
      "loss": 0.9187,
      "step": 1848
    },
    {
      "epoch": 0.30989692449509765,
      "grad_norm": 0.17927313375935594,
      "learning_rate": 6.687696581756763e-08,
      "loss": 0.9112,
      "step": 1849
    },
    {
      "epoch": 0.31006452694209335,
      "grad_norm": 0.1802180363451423,
      "learning_rate": 6.677455793500463e-08,
      "loss": 0.9062,
      "step": 1850
    },
    {
      "epoch": 0.31023212938908906,
      "grad_norm": 0.18496941596207997,
      "learning_rate": 6.667218920619649e-08,
      "loss": 0.8916,
      "step": 1851
    },
    {
      "epoch": 0.3103997318360848,
      "grad_norm": 0.19647492971358704,
      "learning_rate": 6.656985975177721e-08,
      "loss": 0.9341,
      "step": 1852
    },
    {
      "epoch": 0.31056733428308053,
      "grad_norm": 0.20544640066662945,
      "learning_rate": 6.646756969233463e-08,
      "loss": 0.9483,
      "step": 1853
    },
    {
      "epoch": 0.31073493673007624,
      "grad_norm": 0.19699014498684997,
      "learning_rate": 6.636531914841008e-08,
      "loss": 0.9221,
      "step": 1854
    },
    {
      "epoch": 0.310902539177072,
      "grad_norm": 0.18929202631135747,
      "learning_rate": 6.626310824049838e-08,
      "loss": 0.8916,
      "step": 1855
    },
    {
      "epoch": 0.3110701416240677,
      "grad_norm": 0.1907867225149302,
      "learning_rate": 6.616093708904762e-08,
      "loss": 0.925,
      "step": 1856
    },
    {
      "epoch": 0.3112377440710634,
      "grad_norm": 0.18669757968140244,
      "learning_rate": 6.605880581445892e-08,
      "loss": 0.9049,
      "step": 1857
    },
    {
      "epoch": 0.3114053465180592,
      "grad_norm": 0.1963990071662736,
      "learning_rate": 6.595671453708667e-08,
      "loss": 0.9489,
      "step": 1858
    },
    {
      "epoch": 0.3115729489650549,
      "grad_norm": 0.1901005245169179,
      "learning_rate": 6.585466337723795e-08,
      "loss": 0.8963,
      "step": 1859
    },
    {
      "epoch": 0.3117405514120506,
      "grad_norm": 0.18340319292783044,
      "learning_rate": 6.575265245517253e-08,
      "loss": 0.8955,
      "step": 1860
    },
    {
      "epoch": 0.31190815385904636,
      "grad_norm": 0.17789508261099524,
      "learning_rate": 6.565068189110286e-08,
      "loss": 0.8794,
      "step": 1861
    },
    {
      "epoch": 0.31207575630604206,
      "grad_norm": 0.19826258069153888,
      "learning_rate": 6.554875180519376e-08,
      "loss": 0.9545,
      "step": 1862
    },
    {
      "epoch": 0.31224335875303777,
      "grad_norm": 0.18716216565166957,
      "learning_rate": 6.544686231756245e-08,
      "loss": 0.8755,
      "step": 1863
    },
    {
      "epoch": 0.31241096120003353,
      "grad_norm": 0.16835949937174902,
      "learning_rate": 6.534501354827821e-08,
      "loss": 0.886,
      "step": 1864
    },
    {
      "epoch": 0.31257856364702924,
      "grad_norm": 0.1692938821061581,
      "learning_rate": 6.524320561736239e-08,
      "loss": 0.9025,
      "step": 1865
    },
    {
      "epoch": 0.31274616609402495,
      "grad_norm": 0.1770631226207543,
      "learning_rate": 6.514143864478816e-08,
      "loss": 0.9121,
      "step": 1866
    },
    {
      "epoch": 0.3129137685410207,
      "grad_norm": 0.2213602991130013,
      "learning_rate": 6.503971275048041e-08,
      "loss": 0.9375,
      "step": 1867
    },
    {
      "epoch": 0.3130813709880164,
      "grad_norm": 0.16433855957050067,
      "learning_rate": 6.493802805431583e-08,
      "loss": 0.8666,
      "step": 1868
    },
    {
      "epoch": 0.3132489734350121,
      "grad_norm": 0.1921742290327212,
      "learning_rate": 6.483638467612227e-08,
      "loss": 0.9188,
      "step": 1869
    },
    {
      "epoch": 0.3134165758820079,
      "grad_norm": 0.1838414469835517,
      "learning_rate": 6.473478273567905e-08,
      "loss": 0.8815,
      "step": 1870
    },
    {
      "epoch": 0.3135841783290036,
      "grad_norm": 0.17213391298881606,
      "learning_rate": 6.463322235271666e-08,
      "loss": 0.8919,
      "step": 1871
    },
    {
      "epoch": 0.31375178077599936,
      "grad_norm": 0.19313169587223947,
      "learning_rate": 6.453170364691658e-08,
      "loss": 0.9096,
      "step": 1872
    },
    {
      "epoch": 0.31391938322299506,
      "grad_norm": 0.18439681044790898,
      "learning_rate": 6.44302267379111e-08,
      "loss": 0.9238,
      "step": 1873
    },
    {
      "epoch": 0.31408698566999077,
      "grad_norm": 0.18269452319458826,
      "learning_rate": 6.432879174528345e-08,
      "loss": 0.8578,
      "step": 1874
    },
    {
      "epoch": 0.31425458811698653,
      "grad_norm": 0.18830418224843054,
      "learning_rate": 6.422739878856734e-08,
      "loss": 0.9559,
      "step": 1875
    },
    {
      "epoch": 0.31442219056398224,
      "grad_norm": 0.1967375799369618,
      "learning_rate": 6.412604798724692e-08,
      "loss": 0.8983,
      "step": 1876
    },
    {
      "epoch": 0.31458979301097795,
      "grad_norm": 0.19027063613751413,
      "learning_rate": 6.40247394607567e-08,
      "loss": 0.9311,
      "step": 1877
    },
    {
      "epoch": 0.3147573954579737,
      "grad_norm": 0.1734327459746675,
      "learning_rate": 6.392347332848138e-08,
      "loss": 0.8978,
      "step": 1878
    },
    {
      "epoch": 0.3149249979049694,
      "grad_norm": 0.18795776805407327,
      "learning_rate": 6.382224970975571e-08,
      "loss": 0.8909,
      "step": 1879
    },
    {
      "epoch": 0.3150926003519651,
      "grad_norm": 0.1801182284491048,
      "learning_rate": 6.372106872386433e-08,
      "loss": 0.8591,
      "step": 1880
    },
    {
      "epoch": 0.3152602027989609,
      "grad_norm": 0.18033135280364987,
      "learning_rate": 6.361993049004161e-08,
      "loss": 0.8928,
      "step": 1881
    },
    {
      "epoch": 0.3154278052459566,
      "grad_norm": 0.20300025625301274,
      "learning_rate": 6.35188351274716e-08,
      "loss": 0.9449,
      "step": 1882
    },
    {
      "epoch": 0.3155954076929523,
      "grad_norm": 0.18216602306258967,
      "learning_rate": 6.341778275528773e-08,
      "loss": 0.8843,
      "step": 1883
    },
    {
      "epoch": 0.31576301013994806,
      "grad_norm": 0.18101420316179434,
      "learning_rate": 6.331677349257293e-08,
      "loss": 0.894,
      "step": 1884
    },
    {
      "epoch": 0.31593061258694377,
      "grad_norm": 0.18274133696884676,
      "learning_rate": 6.321580745835915e-08,
      "loss": 0.8883,
      "step": 1885
    },
    {
      "epoch": 0.3160982150339395,
      "grad_norm": 0.20093652727292605,
      "learning_rate": 6.311488477162753e-08,
      "loss": 0.8943,
      "step": 1886
    },
    {
      "epoch": 0.31626581748093524,
      "grad_norm": 0.18758688658902803,
      "learning_rate": 6.301400555130805e-08,
      "loss": 0.943,
      "step": 1887
    },
    {
      "epoch": 0.31643341992793095,
      "grad_norm": 0.1749198512281565,
      "learning_rate": 6.291316991627949e-08,
      "loss": 0.903,
      "step": 1888
    },
    {
      "epoch": 0.31660102237492666,
      "grad_norm": 0.17832553933995132,
      "learning_rate": 6.281237798536931e-08,
      "loss": 0.8897,
      "step": 1889
    },
    {
      "epoch": 0.3167686248219224,
      "grad_norm": 0.19021560333570248,
      "learning_rate": 6.271162987735343e-08,
      "loss": 0.925,
      "step": 1890
    },
    {
      "epoch": 0.3169362272689181,
      "grad_norm": 0.17285370376051357,
      "learning_rate": 6.261092571095615e-08,
      "loss": 0.884,
      "step": 1891
    },
    {
      "epoch": 0.31710382971591383,
      "grad_norm": 0.17723828521061494,
      "learning_rate": 6.251026560484989e-08,
      "loss": 0.911,
      "step": 1892
    },
    {
      "epoch": 0.3172714321629096,
      "grad_norm": 0.1773574285353817,
      "learning_rate": 6.240964967765527e-08,
      "loss": 0.9141,
      "step": 1893
    },
    {
      "epoch": 0.3174390346099053,
      "grad_norm": 0.1812698488260321,
      "learning_rate": 6.23090780479408e-08,
      "loss": 0.8401,
      "step": 1894
    },
    {
      "epoch": 0.317606637056901,
      "grad_norm": 0.1835950394407376,
      "learning_rate": 6.220855083422284e-08,
      "loss": 0.8914,
      "step": 1895
    },
    {
      "epoch": 0.31777423950389677,
      "grad_norm": 0.17937917444668225,
      "learning_rate": 6.210806815496532e-08,
      "loss": 0.9157,
      "step": 1896
    },
    {
      "epoch": 0.3179418419508925,
      "grad_norm": 0.16477756636677957,
      "learning_rate": 6.200763012857972e-08,
      "loss": 0.8289,
      "step": 1897
    },
    {
      "epoch": 0.3181094443978882,
      "grad_norm": 0.21729552288103077,
      "learning_rate": 6.190723687342498e-08,
      "loss": 0.9063,
      "step": 1898
    },
    {
      "epoch": 0.31827704684488395,
      "grad_norm": 0.18946319700974576,
      "learning_rate": 6.18068885078071e-08,
      "loss": 0.9297,
      "step": 1899
    },
    {
      "epoch": 0.31844464929187966,
      "grad_norm": 0.18017675842724148,
      "learning_rate": 6.17065851499794e-08,
      "loss": 0.8822,
      "step": 1900
    },
    {
      "epoch": 0.31861225173887536,
      "grad_norm": 0.16844258723781086,
      "learning_rate": 6.160632691814203e-08,
      "loss": 0.9072,
      "step": 1901
    },
    {
      "epoch": 0.3187798541858711,
      "grad_norm": 0.1614795618170478,
      "learning_rate": 6.150611393044194e-08,
      "loss": 0.8945,
      "step": 1902
    },
    {
      "epoch": 0.31894745663286683,
      "grad_norm": 0.18927582418509478,
      "learning_rate": 6.140594630497287e-08,
      "loss": 0.9176,
      "step": 1903
    },
    {
      "epoch": 0.31911505907986254,
      "grad_norm": 0.1698713012690334,
      "learning_rate": 6.130582415977497e-08,
      "loss": 0.912,
      "step": 1904
    },
    {
      "epoch": 0.3192826615268583,
      "grad_norm": 0.1789837757319818,
      "learning_rate": 6.120574761283496e-08,
      "loss": 0.892,
      "step": 1905
    },
    {
      "epoch": 0.319450263973854,
      "grad_norm": 0.19025083792634379,
      "learning_rate": 6.110571678208568e-08,
      "loss": 0.9022,
      "step": 1906
    },
    {
      "epoch": 0.3196178664208497,
      "grad_norm": 0.14067530991108704,
      "learning_rate": 6.100573178540618e-08,
      "loss": 0.8231,
      "step": 1907
    },
    {
      "epoch": 0.3197854688678455,
      "grad_norm": 0.16142354624448033,
      "learning_rate": 6.090579274062146e-08,
      "loss": 0.8823,
      "step": 1908
    },
    {
      "epoch": 0.3199530713148412,
      "grad_norm": 0.19672995192218737,
      "learning_rate": 6.080589976550232e-08,
      "loss": 0.933,
      "step": 1909
    },
    {
      "epoch": 0.32012067376183695,
      "grad_norm": 0.19332062387333443,
      "learning_rate": 6.070605297776546e-08,
      "loss": 0.9309,
      "step": 1910
    },
    {
      "epoch": 0.32028827620883266,
      "grad_norm": 0.1625516594269731,
      "learning_rate": 6.06062524950729e-08,
      "loss": 0.8898,
      "step": 1911
    },
    {
      "epoch": 0.32045587865582836,
      "grad_norm": 0.1629606349533714,
      "learning_rate": 6.050649843503228e-08,
      "loss": 0.8353,
      "step": 1912
    },
    {
      "epoch": 0.3206234811028241,
      "grad_norm": 0.17347880824213663,
      "learning_rate": 6.040679091519643e-08,
      "loss": 0.8836,
      "step": 1913
    },
    {
      "epoch": 0.32079108354981983,
      "grad_norm": 0.17317078516169745,
      "learning_rate": 6.03071300530634e-08,
      "loss": 0.9326,
      "step": 1914
    },
    {
      "epoch": 0.32095868599681554,
      "grad_norm": 0.1954336730555435,
      "learning_rate": 6.020751596607621e-08,
      "loss": 0.9049,
      "step": 1915
    },
    {
      "epoch": 0.3211262884438113,
      "grad_norm": 0.20276153156179952,
      "learning_rate": 6.010794877162279e-08,
      "loss": 0.9026,
      "step": 1916
    },
    {
      "epoch": 0.321293890890807,
      "grad_norm": 0.2158090915570876,
      "learning_rate": 6.000842858703585e-08,
      "loss": 0.9095,
      "step": 1917
    },
    {
      "epoch": 0.3214614933378027,
      "grad_norm": 0.21092184298558278,
      "learning_rate": 5.990895552959256e-08,
      "loss": 0.9434,
      "step": 1918
    },
    {
      "epoch": 0.3216290957847985,
      "grad_norm": 0.1829529376603461,
      "learning_rate": 5.980952971651472e-08,
      "loss": 0.8928,
      "step": 1919
    },
    {
      "epoch": 0.3217966982317942,
      "grad_norm": 0.1781442369292984,
      "learning_rate": 5.971015126496832e-08,
      "loss": 0.9447,
      "step": 1920
    },
    {
      "epoch": 0.3219643006787899,
      "grad_norm": 0.17655489258739757,
      "learning_rate": 5.961082029206366e-08,
      "loss": 0.884,
      "step": 1921
    },
    {
      "epoch": 0.32213190312578566,
      "grad_norm": 0.17140419686836778,
      "learning_rate": 5.951153691485502e-08,
      "loss": 0.8445,
      "step": 1922
    },
    {
      "epoch": 0.32229950557278136,
      "grad_norm": 0.21059792498175162,
      "learning_rate": 5.941230125034058e-08,
      "loss": 0.9612,
      "step": 1923
    },
    {
      "epoch": 0.32246710801977707,
      "grad_norm": 0.18662045627147109,
      "learning_rate": 5.9313113415462344e-08,
      "loss": 0.8585,
      "step": 1924
    },
    {
      "epoch": 0.32263471046677283,
      "grad_norm": 0.18403925751224276,
      "learning_rate": 5.921397352710587e-08,
      "loss": 0.8746,
      "step": 1925
    },
    {
      "epoch": 0.32280231291376854,
      "grad_norm": 0.1988560209778446,
      "learning_rate": 5.9114881702100325e-08,
      "loss": 0.9251,
      "step": 1926
    },
    {
      "epoch": 0.32296991536076425,
      "grad_norm": 0.19182778570785747,
      "learning_rate": 5.901583805721819e-08,
      "loss": 0.8957,
      "step": 1927
    },
    {
      "epoch": 0.32313751780776,
      "grad_norm": 0.17855078703647675,
      "learning_rate": 5.891684270917513e-08,
      "loss": 0.9074,
      "step": 1928
    },
    {
      "epoch": 0.3233051202547557,
      "grad_norm": 0.17071050840722715,
      "learning_rate": 5.881789577462992e-08,
      "loss": 0.8953,
      "step": 1929
    },
    {
      "epoch": 0.3234727227017514,
      "grad_norm": 0.16871671581301578,
      "learning_rate": 5.871899737018431e-08,
      "loss": 0.92,
      "step": 1930
    },
    {
      "epoch": 0.3236403251487472,
      "grad_norm": 0.18465390069087212,
      "learning_rate": 5.8620147612382807e-08,
      "loss": 0.8804,
      "step": 1931
    },
    {
      "epoch": 0.3238079275957429,
      "grad_norm": 0.18492648090361904,
      "learning_rate": 5.852134661771268e-08,
      "loss": 0.9205,
      "step": 1932
    },
    {
      "epoch": 0.3239755300427386,
      "grad_norm": 0.18908270025621482,
      "learning_rate": 5.842259450260365e-08,
      "loss": 0.9096,
      "step": 1933
    },
    {
      "epoch": 0.32414313248973436,
      "grad_norm": 0.1590651695441171,
      "learning_rate": 5.832389138342788e-08,
      "loss": 0.8277,
      "step": 1934
    },
    {
      "epoch": 0.32431073493673007,
      "grad_norm": 0.20038307687357101,
      "learning_rate": 5.822523737649974e-08,
      "loss": 0.9163,
      "step": 1935
    },
    {
      "epoch": 0.3244783373837258,
      "grad_norm": 0.18629494872503233,
      "learning_rate": 5.81266325980758e-08,
      "loss": 0.918,
      "step": 1936
    },
    {
      "epoch": 0.32464593983072154,
      "grad_norm": 0.1735555900193119,
      "learning_rate": 5.80280771643546e-08,
      "loss": 0.885,
      "step": 1937
    },
    {
      "epoch": 0.32481354227771725,
      "grad_norm": 0.17854379023309142,
      "learning_rate": 5.79295711914765e-08,
      "loss": 0.8615,
      "step": 1938
    },
    {
      "epoch": 0.32498114472471296,
      "grad_norm": 0.18991427312004222,
      "learning_rate": 5.783111479552354e-08,
      "loss": 0.9434,
      "step": 1939
    },
    {
      "epoch": 0.3251487471717087,
      "grad_norm": 0.1814759348222561,
      "learning_rate": 5.773270809251947e-08,
      "loss": 0.9153,
      "step": 1940
    },
    {
      "epoch": 0.3253163496187044,
      "grad_norm": 0.16306874202292004,
      "learning_rate": 5.76343511984293e-08,
      "loss": 0.8402,
      "step": 1941
    },
    {
      "epoch": 0.32548395206570013,
      "grad_norm": 0.19571174236303376,
      "learning_rate": 5.753604422915951e-08,
      "loss": 0.9327,
      "step": 1942
    },
    {
      "epoch": 0.3256515545126959,
      "grad_norm": 0.1527089062307572,
      "learning_rate": 5.743778730055769e-08,
      "loss": 0.8311,
      "step": 1943
    },
    {
      "epoch": 0.3258191569596916,
      "grad_norm": 0.16940880949654052,
      "learning_rate": 5.733958052841239e-08,
      "loss": 0.8879,
      "step": 1944
    },
    {
      "epoch": 0.32598675940668737,
      "grad_norm": 0.17627213878799888,
      "learning_rate": 5.724142402845318e-08,
      "loss": 0.9243,
      "step": 1945
    },
    {
      "epoch": 0.3261543618536831,
      "grad_norm": 0.18976311238166843,
      "learning_rate": 5.714331791635024e-08,
      "loss": 0.9137,
      "step": 1946
    },
    {
      "epoch": 0.3263219643006788,
      "grad_norm": 0.17581915171061266,
      "learning_rate": 5.704526230771449e-08,
      "loss": 0.9293,
      "step": 1947
    },
    {
      "epoch": 0.32648956674767454,
      "grad_norm": 0.17546417062931263,
      "learning_rate": 5.694725731809735e-08,
      "loss": 0.856,
      "step": 1948
    },
    {
      "epoch": 0.32665716919467025,
      "grad_norm": 0.19817917388774683,
      "learning_rate": 5.6849303062990494e-08,
      "loss": 0.9698,
      "step": 1949
    },
    {
      "epoch": 0.32682477164166596,
      "grad_norm": 0.1681145657015865,
      "learning_rate": 5.675139965782578e-08,
      "loss": 0.8975,
      "step": 1950
    },
    {
      "epoch": 0.3269923740886617,
      "grad_norm": 0.16475177252423778,
      "learning_rate": 5.665354721797535e-08,
      "loss": 0.8817,
      "step": 1951
    },
    {
      "epoch": 0.3271599765356574,
      "grad_norm": 0.18402192430957556,
      "learning_rate": 5.655574585875103e-08,
      "loss": 0.8771,
      "step": 1952
    },
    {
      "epoch": 0.32732757898265313,
      "grad_norm": 0.19195105247932043,
      "learning_rate": 5.645799569540462e-08,
      "loss": 0.9331,
      "step": 1953
    },
    {
      "epoch": 0.3274951814296489,
      "grad_norm": 0.1920686970951378,
      "learning_rate": 5.6360296843127594e-08,
      "loss": 0.9548,
      "step": 1954
    },
    {
      "epoch": 0.3276627838766446,
      "grad_norm": 0.20565839822876156,
      "learning_rate": 5.6262649417050856e-08,
      "loss": 0.9503,
      "step": 1955
    },
    {
      "epoch": 0.3278303863236403,
      "grad_norm": 0.1870607006058217,
      "learning_rate": 5.616505353224475e-08,
      "loss": 0.9582,
      "step": 1956
    },
    {
      "epoch": 0.3279979887706361,
      "grad_norm": 0.16579402676131882,
      "learning_rate": 5.606750930371887e-08,
      "loss": 0.86,
      "step": 1957
    },
    {
      "epoch": 0.3281655912176318,
      "grad_norm": 0.21970829892739463,
      "learning_rate": 5.597001684642206e-08,
      "loss": 0.9858,
      "step": 1958
    },
    {
      "epoch": 0.3283331936646275,
      "grad_norm": 0.17807490702111692,
      "learning_rate": 5.5872576275241944e-08,
      "loss": 0.8795,
      "step": 1959
    },
    {
      "epoch": 0.32850079611162325,
      "grad_norm": 0.17758992351895067,
      "learning_rate": 5.57751877050052e-08,
      "loss": 0.8538,
      "step": 1960
    },
    {
      "epoch": 0.32866839855861896,
      "grad_norm": 0.18401195275062188,
      "learning_rate": 5.567785125047707e-08,
      "loss": 0.8881,
      "step": 1961
    },
    {
      "epoch": 0.32883600100561466,
      "grad_norm": 0.17641074035262178,
      "learning_rate": 5.558056702636147e-08,
      "loss": 0.8902,
      "step": 1962
    },
    {
      "epoch": 0.3290036034526104,
      "grad_norm": 0.17618332575320866,
      "learning_rate": 5.5483335147300815e-08,
      "loss": 0.8995,
      "step": 1963
    },
    {
      "epoch": 0.32917120589960613,
      "grad_norm": 0.19804352875955528,
      "learning_rate": 5.538615572787569e-08,
      "loss": 0.9509,
      "step": 1964
    },
    {
      "epoch": 0.32933880834660184,
      "grad_norm": 0.16770633541334057,
      "learning_rate": 5.5289028882604925e-08,
      "loss": 0.8551,
      "step": 1965
    },
    {
      "epoch": 0.3295064107935976,
      "grad_norm": 0.18363669687899384,
      "learning_rate": 5.519195472594549e-08,
      "loss": 0.8949,
      "step": 1966
    },
    {
      "epoch": 0.3296740132405933,
      "grad_norm": 0.17489287859632915,
      "learning_rate": 5.5094933372292077e-08,
      "loss": 0.8914,
      "step": 1967
    },
    {
      "epoch": 0.329841615687589,
      "grad_norm": 0.17115601246953896,
      "learning_rate": 5.4997964935977325e-08,
      "loss": 0.9068,
      "step": 1968
    },
    {
      "epoch": 0.3300092181345848,
      "grad_norm": 0.16156460863304217,
      "learning_rate": 5.490104953127147e-08,
      "loss": 0.8591,
      "step": 1969
    },
    {
      "epoch": 0.3301768205815805,
      "grad_norm": 0.16626165638067733,
      "learning_rate": 5.480418727238216e-08,
      "loss": 0.8947,
      "step": 1970
    },
    {
      "epoch": 0.3303444230285762,
      "grad_norm": 0.1703418915244161,
      "learning_rate": 5.470737827345457e-08,
      "loss": 0.8991,
      "step": 1971
    },
    {
      "epoch": 0.33051202547557196,
      "grad_norm": 0.17491649668932646,
      "learning_rate": 5.4610622648571e-08,
      "loss": 0.8982,
      "step": 1972
    },
    {
      "epoch": 0.33067962792256766,
      "grad_norm": 0.18809092111131653,
      "learning_rate": 5.4513920511750786e-08,
      "loss": 0.9388,
      "step": 1973
    },
    {
      "epoch": 0.33084723036956337,
      "grad_norm": 0.18345778693055725,
      "learning_rate": 5.441727197695047e-08,
      "loss": 0.917,
      "step": 1974
    },
    {
      "epoch": 0.33101483281655913,
      "grad_norm": 0.1789381890304888,
      "learning_rate": 5.4320677158063243e-08,
      "loss": 0.8921,
      "step": 1975
    },
    {
      "epoch": 0.33118243526355484,
      "grad_norm": 0.18963120917623458,
      "learning_rate": 5.4224136168918976e-08,
      "loss": 0.9048,
      "step": 1976
    },
    {
      "epoch": 0.33135003771055055,
      "grad_norm": 0.16875232246511318,
      "learning_rate": 5.412764912328426e-08,
      "loss": 0.865,
      "step": 1977
    },
    {
      "epoch": 0.3315176401575463,
      "grad_norm": 0.19642787603997183,
      "learning_rate": 5.4031216134861924e-08,
      "loss": 0.9234,
      "step": 1978
    },
    {
      "epoch": 0.331685242604542,
      "grad_norm": 0.17269463865640156,
      "learning_rate": 5.3934837317291273e-08,
      "loss": 0.8626,
      "step": 1979
    },
    {
      "epoch": 0.3318528450515377,
      "grad_norm": 0.14685406347036056,
      "learning_rate": 5.383851278414769e-08,
      "loss": 0.8419,
      "step": 1980
    },
    {
      "epoch": 0.3320204474985335,
      "grad_norm": 0.17786092089028493,
      "learning_rate": 5.37422426489426e-08,
      "loss": 0.8639,
      "step": 1981
    },
    {
      "epoch": 0.3321880499455292,
      "grad_norm": 0.15365838217882086,
      "learning_rate": 5.364602702512325e-08,
      "loss": 0.867,
      "step": 1982
    },
    {
      "epoch": 0.33235565239252496,
      "grad_norm": 0.18792661103021854,
      "learning_rate": 5.354986602607279e-08,
      "loss": 0.9447,
      "step": 1983
    },
    {
      "epoch": 0.33252325483952067,
      "grad_norm": 0.15837147662955084,
      "learning_rate": 5.345375976510993e-08,
      "loss": 0.8407,
      "step": 1984
    },
    {
      "epoch": 0.3326908572865164,
      "grad_norm": 0.17314144542821183,
      "learning_rate": 5.335770835548883e-08,
      "loss": 0.8966,
      "step": 1985
    },
    {
      "epoch": 0.33285845973351214,
      "grad_norm": 0.19253078653610894,
      "learning_rate": 5.326171191039912e-08,
      "loss": 0.9435,
      "step": 1986
    },
    {
      "epoch": 0.33302606218050784,
      "grad_norm": 0.17617199988004154,
      "learning_rate": 5.3165770542965504e-08,
      "loss": 0.8718,
      "step": 1987
    },
    {
      "epoch": 0.33319366462750355,
      "grad_norm": 0.17063077116496989,
      "learning_rate": 5.306988436624793e-08,
      "loss": 0.8905,
      "step": 1988
    },
    {
      "epoch": 0.3333612670744993,
      "grad_norm": 0.18382647469000007,
      "learning_rate": 5.2974053493241263e-08,
      "loss": 0.915,
      "step": 1989
    },
    {
      "epoch": 0.333528869521495,
      "grad_norm": 0.19133914933153873,
      "learning_rate": 5.287827803687517e-08,
      "loss": 0.8962,
      "step": 1990
    },
    {
      "epoch": 0.3336964719684907,
      "grad_norm": 0.18485974226605018,
      "learning_rate": 5.278255811001398e-08,
      "loss": 0.8718,
      "step": 1991
    },
    {
      "epoch": 0.3338640744154865,
      "grad_norm": 0.20807997966223737,
      "learning_rate": 5.268689382545669e-08,
      "loss": 0.9447,
      "step": 1992
    },
    {
      "epoch": 0.3340316768624822,
      "grad_norm": 0.17173302098548346,
      "learning_rate": 5.25912852959366e-08,
      "loss": 0.8571,
      "step": 1993
    },
    {
      "epoch": 0.3341992793094779,
      "grad_norm": 0.18988489444031134,
      "learning_rate": 5.249573263412141e-08,
      "loss": 0.9274,
      "step": 1994
    },
    {
      "epoch": 0.33436688175647367,
      "grad_norm": 0.1748348676261372,
      "learning_rate": 5.2400235952613e-08,
      "loss": 0.8983,
      "step": 1995
    },
    {
      "epoch": 0.3345344842034694,
      "grad_norm": 0.15483811480330764,
      "learning_rate": 5.230479536394714e-08,
      "loss": 0.8506,
      "step": 1996
    },
    {
      "epoch": 0.3347020866504651,
      "grad_norm": 0.19877187264617643,
      "learning_rate": 5.220941098059367e-08,
      "loss": 0.9191,
      "step": 1997
    },
    {
      "epoch": 0.33486968909746084,
      "grad_norm": 0.16010978451587654,
      "learning_rate": 5.211408291495607e-08,
      "loss": 0.8336,
      "step": 1998
    },
    {
      "epoch": 0.33503729154445655,
      "grad_norm": 0.17498737018628666,
      "learning_rate": 5.201881127937141e-08,
      "loss": 0.9012,
      "step": 1999
    },
    {
      "epoch": 0.33520489399145226,
      "grad_norm": 0.18517737887290248,
      "learning_rate": 5.192359618611052e-08,
      "loss": 0.9165,
      "step": 2000
    },
    {
      "epoch": 0.335372496438448,
      "grad_norm": 0.17907262914142727,
      "learning_rate": 5.1828437747377353e-08,
      "loss": 0.91,
      "step": 2001
    },
    {
      "epoch": 0.3355400988854437,
      "grad_norm": 0.16530629764293722,
      "learning_rate": 5.173333607530911e-08,
      "loss": 0.886,
      "step": 2002
    },
    {
      "epoch": 0.33570770133243943,
      "grad_norm": 0.17454238329781716,
      "learning_rate": 5.1638291281976255e-08,
      "loss": 0.9011,
      "step": 2003
    },
    {
      "epoch": 0.3358753037794352,
      "grad_norm": 0.18192280322728818,
      "learning_rate": 5.154330347938205e-08,
      "loss": 0.8864,
      "step": 2004
    },
    {
      "epoch": 0.3360429062264309,
      "grad_norm": 0.1972403282615578,
      "learning_rate": 5.1448372779462725e-08,
      "loss": 0.9066,
      "step": 2005
    },
    {
      "epoch": 0.3362105086734266,
      "grad_norm": 0.190710681833568,
      "learning_rate": 5.13534992940872e-08,
      "loss": 0.8877,
      "step": 2006
    },
    {
      "epoch": 0.3363781111204224,
      "grad_norm": 0.1737745145224343,
      "learning_rate": 5.12586831350569e-08,
      "loss": 0.9031,
      "step": 2007
    },
    {
      "epoch": 0.3365457135674181,
      "grad_norm": 0.195195646170488,
      "learning_rate": 5.116392441410571e-08,
      "loss": 0.8922,
      "step": 2008
    },
    {
      "epoch": 0.3367133160144138,
      "grad_norm": 0.1880499683295383,
      "learning_rate": 5.1069223242899873e-08,
      "loss": 0.9162,
      "step": 2009
    },
    {
      "epoch": 0.33688091846140955,
      "grad_norm": 0.20445493263071657,
      "learning_rate": 5.097457973303787e-08,
      "loss": 0.9635,
      "step": 2010
    },
    {
      "epoch": 0.33704852090840526,
      "grad_norm": 0.18827548851780734,
      "learning_rate": 5.0879993996050055e-08,
      "loss": 0.9167,
      "step": 2011
    },
    {
      "epoch": 0.33721612335540097,
      "grad_norm": 0.20958035010737952,
      "learning_rate": 5.0785466143398884e-08,
      "loss": 0.9207,
      "step": 2012
    },
    {
      "epoch": 0.33738372580239673,
      "grad_norm": 0.17840386290094257,
      "learning_rate": 5.069099628647846e-08,
      "loss": 0.8975,
      "step": 2013
    },
    {
      "epoch": 0.33755132824939243,
      "grad_norm": 0.17530039686394053,
      "learning_rate": 5.059658453661466e-08,
      "loss": 0.8635,
      "step": 2014
    },
    {
      "epoch": 0.33771893069638814,
      "grad_norm": 0.16184837384385203,
      "learning_rate": 5.050223100506479e-08,
      "loss": 0.8665,
      "step": 2015
    },
    {
      "epoch": 0.3378865331433839,
      "grad_norm": 0.17941748292280082,
      "learning_rate": 5.040793580301762e-08,
      "loss": 0.8436,
      "step": 2016
    },
    {
      "epoch": 0.3380541355903796,
      "grad_norm": 0.20204425202829304,
      "learning_rate": 5.0313699041593106e-08,
      "loss": 0.9178,
      "step": 2017
    },
    {
      "epoch": 0.3382217380373753,
      "grad_norm": 0.18386199023327648,
      "learning_rate": 5.021952083184245e-08,
      "loss": 0.8853,
      "step": 2018
    },
    {
      "epoch": 0.3383893404843711,
      "grad_norm": 0.20099102938977995,
      "learning_rate": 5.012540128474773e-08,
      "loss": 0.8853,
      "step": 2019
    },
    {
      "epoch": 0.3385569429313668,
      "grad_norm": 0.18126139799279467,
      "learning_rate": 5.003134051122195e-08,
      "loss": 0.9181,
      "step": 2020
    },
    {
      "epoch": 0.33872454537836255,
      "grad_norm": 0.1793705066336132,
      "learning_rate": 4.993733862210894e-08,
      "loss": 0.9068,
      "step": 2021
    },
    {
      "epoch": 0.33889214782535826,
      "grad_norm": 0.19421346217260652,
      "learning_rate": 4.9843395728182944e-08,
      "loss": 0.9216,
      "step": 2022
    },
    {
      "epoch": 0.33905975027235397,
      "grad_norm": 0.19734331483846582,
      "learning_rate": 4.974951194014889e-08,
      "loss": 0.8809,
      "step": 2023
    },
    {
      "epoch": 0.33922735271934973,
      "grad_norm": 0.18358884200998069,
      "learning_rate": 4.965568736864193e-08,
      "loss": 0.9015,
      "step": 2024
    },
    {
      "epoch": 0.33939495516634544,
      "grad_norm": 0.16942775795395004,
      "learning_rate": 4.956192212422736e-08,
      "loss": 0.8976,
      "step": 2025
    },
    {
      "epoch": 0.33956255761334114,
      "grad_norm": 0.1936413514807987,
      "learning_rate": 4.9468216317400834e-08,
      "loss": 0.9165,
      "step": 2026
    },
    {
      "epoch": 0.3397301600603369,
      "grad_norm": 0.18430352676952366,
      "learning_rate": 4.9374570058587726e-08,
      "loss": 0.8759,
      "step": 2027
    },
    {
      "epoch": 0.3398977625073326,
      "grad_norm": 0.19336783456876813,
      "learning_rate": 4.928098345814327e-08,
      "loss": 0.8835,
      "step": 2028
    },
    {
      "epoch": 0.3400653649543283,
      "grad_norm": 0.17058265728045574,
      "learning_rate": 4.9187456626352484e-08,
      "loss": 0.8771,
      "step": 2029
    },
    {
      "epoch": 0.3402329674013241,
      "grad_norm": 0.19161507783698303,
      "learning_rate": 4.909398967342985e-08,
      "loss": 0.8783,
      "step": 2030
    },
    {
      "epoch": 0.3404005698483198,
      "grad_norm": 0.1657604199220094,
      "learning_rate": 4.900058270951938e-08,
      "loss": 0.8618,
      "step": 2031
    },
    {
      "epoch": 0.3405681722953155,
      "grad_norm": 0.17936652731288055,
      "learning_rate": 4.890723584469438e-08,
      "loss": 0.9053,
      "step": 2032
    },
    {
      "epoch": 0.34073577474231126,
      "grad_norm": 0.20393384638858147,
      "learning_rate": 4.881394918895727e-08,
      "loss": 0.9155,
      "step": 2033
    },
    {
      "epoch": 0.34090337718930697,
      "grad_norm": 0.16314811252549938,
      "learning_rate": 4.872072285223953e-08,
      "loss": 0.8685,
      "step": 2034
    },
    {
      "epoch": 0.3410709796363027,
      "grad_norm": 0.17346954457420383,
      "learning_rate": 4.8627556944401636e-08,
      "loss": 0.8967,
      "step": 2035
    },
    {
      "epoch": 0.34123858208329844,
      "grad_norm": 0.16654937133701278,
      "learning_rate": 4.853445157523275e-08,
      "loss": 0.859,
      "step": 2036
    },
    {
      "epoch": 0.34140618453029414,
      "grad_norm": 0.20891061356872279,
      "learning_rate": 4.844140685445076e-08,
      "loss": 0.9274,
      "step": 2037
    },
    {
      "epoch": 0.34157378697728985,
      "grad_norm": 0.17607214997479528,
      "learning_rate": 4.834842289170212e-08,
      "loss": 0.888,
      "step": 2038
    },
    {
      "epoch": 0.3417413894242856,
      "grad_norm": 0.20045129299964612,
      "learning_rate": 4.8255499796561564e-08,
      "loss": 0.9055,
      "step": 2039
    },
    {
      "epoch": 0.3419089918712813,
      "grad_norm": 0.16488648186862218,
      "learning_rate": 4.8162637678532226e-08,
      "loss": 0.9009,
      "step": 2040
    },
    {
      "epoch": 0.342076594318277,
      "grad_norm": 0.17437348628367838,
      "learning_rate": 4.8069836647045246e-08,
      "loss": 0.8844,
      "step": 2041
    },
    {
      "epoch": 0.3422441967652728,
      "grad_norm": 0.17395534278553212,
      "learning_rate": 4.797709681145995e-08,
      "loss": 0.8703,
      "step": 2042
    },
    {
      "epoch": 0.3424117992122685,
      "grad_norm": 0.18680589682427187,
      "learning_rate": 4.7884418281063374e-08,
      "loss": 0.9094,
      "step": 2043
    },
    {
      "epoch": 0.3425794016592642,
      "grad_norm": 0.18430353907097097,
      "learning_rate": 4.779180116507046e-08,
      "loss": 0.9439,
      "step": 2044
    },
    {
      "epoch": 0.34274700410625997,
      "grad_norm": 0.17466385060406453,
      "learning_rate": 4.769924557262364e-08,
      "loss": 0.8459,
      "step": 2045
    },
    {
      "epoch": 0.3429146065532557,
      "grad_norm": 0.17310392650006676,
      "learning_rate": 4.760675161279294e-08,
      "loss": 0.8775,
      "step": 2046
    },
    {
      "epoch": 0.3430822090002514,
      "grad_norm": 0.1726086139678355,
      "learning_rate": 4.7514319394575783e-08,
      "loss": 0.8992,
      "step": 2047
    },
    {
      "epoch": 0.34324981144724714,
      "grad_norm": 0.17081354791946604,
      "learning_rate": 4.742194902689671e-08,
      "loss": 0.918,
      "step": 2048
    },
    {
      "epoch": 0.34341741389424285,
      "grad_norm": 0.17652398211210538,
      "learning_rate": 4.7329640618607516e-08,
      "loss": 0.897,
      "step": 2049
    },
    {
      "epoch": 0.34358501634123856,
      "grad_norm": 0.1638611493596011,
      "learning_rate": 4.7237394278486865e-08,
      "loss": 0.835,
      "step": 2050
    },
    {
      "epoch": 0.3437526187882343,
      "grad_norm": 0.16804736150882055,
      "learning_rate": 4.71452101152403e-08,
      "loss": 0.8676,
      "step": 2051
    },
    {
      "epoch": 0.34392022123523003,
      "grad_norm": 0.18860799186410693,
      "learning_rate": 4.705308823750015e-08,
      "loss": 0.8817,
      "step": 2052
    },
    {
      "epoch": 0.34408782368222574,
      "grad_norm": 0.19990304078512147,
      "learning_rate": 4.696102875382536e-08,
      "loss": 0.9669,
      "step": 2053
    },
    {
      "epoch": 0.3442554261292215,
      "grad_norm": 0.19640236074333867,
      "learning_rate": 4.6869031772701215e-08,
      "loss": 0.943,
      "step": 2054
    },
    {
      "epoch": 0.3444230285762172,
      "grad_norm": 0.19482698556421724,
      "learning_rate": 4.677709740253952e-08,
      "loss": 0.9327,
      "step": 2055
    },
    {
      "epoch": 0.3445906310232129,
      "grad_norm": 0.195734179367742,
      "learning_rate": 4.668522575167818e-08,
      "loss": 0.942,
      "step": 2056
    },
    {
      "epoch": 0.3447582334702087,
      "grad_norm": 0.16756617958787473,
      "learning_rate": 4.6593416928381124e-08,
      "loss": 0.8483,
      "step": 2057
    },
    {
      "epoch": 0.3449258359172044,
      "grad_norm": 0.18946937101559175,
      "learning_rate": 4.6501671040838495e-08,
      "loss": 0.9208,
      "step": 2058
    },
    {
      "epoch": 0.34509343836420014,
      "grad_norm": 0.15458347729819621,
      "learning_rate": 4.640998819716605e-08,
      "loss": 0.8935,
      "step": 2059
    },
    {
      "epoch": 0.34526104081119585,
      "grad_norm": 0.19174712906059993,
      "learning_rate": 4.631836850540527e-08,
      "loss": 0.9091,
      "step": 2060
    },
    {
      "epoch": 0.34542864325819156,
      "grad_norm": 0.19465663805856334,
      "learning_rate": 4.622681207352335e-08,
      "loss": 0.8975,
      "step": 2061
    },
    {
      "epoch": 0.3455962457051873,
      "grad_norm": 0.17310167699061024,
      "learning_rate": 4.6135319009412774e-08,
      "loss": 0.8691,
      "step": 2062
    },
    {
      "epoch": 0.34576384815218303,
      "grad_norm": 0.16867228774285037,
      "learning_rate": 4.604388942089146e-08,
      "loss": 0.8863,
      "step": 2063
    },
    {
      "epoch": 0.34593145059917874,
      "grad_norm": 0.18774495406416433,
      "learning_rate": 4.595252341570254e-08,
      "loss": 0.9092,
      "step": 2064
    },
    {
      "epoch": 0.3460990530461745,
      "grad_norm": 0.17780875234690768,
      "learning_rate": 4.5861221101514136e-08,
      "loss": 0.886,
      "step": 2065
    },
    {
      "epoch": 0.3462666554931702,
      "grad_norm": 0.21419461459591532,
      "learning_rate": 4.576998258591931e-08,
      "loss": 0.9435,
      "step": 2066
    },
    {
      "epoch": 0.3464342579401659,
      "grad_norm": 0.210749912858005,
      "learning_rate": 4.5678807976436026e-08,
      "loss": 0.9215,
      "step": 2067
    },
    {
      "epoch": 0.3466018603871617,
      "grad_norm": 0.19045599989671333,
      "learning_rate": 4.558769738050693e-08,
      "loss": 0.9227,
      "step": 2068
    },
    {
      "epoch": 0.3467694628341574,
      "grad_norm": 0.19155053179169246,
      "learning_rate": 4.5496650905499126e-08,
      "loss": 0.9422,
      "step": 2069
    },
    {
      "epoch": 0.3469370652811531,
      "grad_norm": 0.1671593005426575,
      "learning_rate": 4.540566865870431e-08,
      "loss": 0.8245,
      "step": 2070
    },
    {
      "epoch": 0.34710466772814885,
      "grad_norm": 0.1710069966539945,
      "learning_rate": 4.531475074733831e-08,
      "loss": 0.8753,
      "step": 2071
    },
    {
      "epoch": 0.34727227017514456,
      "grad_norm": 0.1739353559362194,
      "learning_rate": 4.522389727854136e-08,
      "loss": 0.8705,
      "step": 2072
    },
    {
      "epoch": 0.34743987262214027,
      "grad_norm": 0.1739473704236476,
      "learning_rate": 4.5133108359377504e-08,
      "loss": 0.881,
      "step": 2073
    },
    {
      "epoch": 0.34760747506913603,
      "grad_norm": 0.17293512070432515,
      "learning_rate": 4.504238409683493e-08,
      "loss": 0.8875,
      "step": 2074
    },
    {
      "epoch": 0.34777507751613174,
      "grad_norm": 0.17813153256842004,
      "learning_rate": 4.495172459782556e-08,
      "loss": 0.8692,
      "step": 2075
    },
    {
      "epoch": 0.34794267996312744,
      "grad_norm": 0.181141979219082,
      "learning_rate": 4.486112996918495e-08,
      "loss": 0.9133,
      "step": 2076
    },
    {
      "epoch": 0.3481102824101232,
      "grad_norm": 0.1685625287082402,
      "learning_rate": 4.4770600317672225e-08,
      "loss": 0.8791,
      "step": 2077
    },
    {
      "epoch": 0.3482778848571189,
      "grad_norm": 0.1864602728944833,
      "learning_rate": 4.468013574996998e-08,
      "loss": 0.8944,
      "step": 2078
    },
    {
      "epoch": 0.3484454873041146,
      "grad_norm": 0.18018559251405086,
      "learning_rate": 4.458973637268416e-08,
      "loss": 0.9168,
      "step": 2079
    },
    {
      "epoch": 0.3486130897511104,
      "grad_norm": 0.17361960370411306,
      "learning_rate": 4.449940229234372e-08,
      "loss": 0.8942,
      "step": 2080
    },
    {
      "epoch": 0.3487806921981061,
      "grad_norm": 0.16889080199916293,
      "learning_rate": 4.440913361540086e-08,
      "loss": 0.8459,
      "step": 2081
    },
    {
      "epoch": 0.3489482946451018,
      "grad_norm": 0.15407265566200737,
      "learning_rate": 4.4318930448230584e-08,
      "loss": 0.843,
      "step": 2082
    },
    {
      "epoch": 0.34911589709209756,
      "grad_norm": 0.16567155833501213,
      "learning_rate": 4.422879289713065e-08,
      "loss": 0.8892,
      "step": 2083
    },
    {
      "epoch": 0.34928349953909327,
      "grad_norm": 0.1949276281185936,
      "learning_rate": 4.4138721068321727e-08,
      "loss": 0.9085,
      "step": 2084
    },
    {
      "epoch": 0.349451101986089,
      "grad_norm": 0.1652307152286719,
      "learning_rate": 4.40487150679468e-08,
      "loss": 0.8489,
      "step": 2085
    },
    {
      "epoch": 0.34961870443308474,
      "grad_norm": 0.14863507457290892,
      "learning_rate": 4.395877500207133e-08,
      "loss": 0.8495,
      "step": 2086
    },
    {
      "epoch": 0.34978630688008044,
      "grad_norm": 0.18573693659661678,
      "learning_rate": 4.386890097668317e-08,
      "loss": 0.9135,
      "step": 2087
    },
    {
      "epoch": 0.34995390932707615,
      "grad_norm": 0.1641243806856041,
      "learning_rate": 4.377909309769223e-08,
      "loss": 0.8454,
      "step": 2088
    },
    {
      "epoch": 0.3501215117740719,
      "grad_norm": 0.17773304590146602,
      "learning_rate": 4.368935147093057e-08,
      "loss": 0.8968,
      "step": 2089
    },
    {
      "epoch": 0.3502891142210676,
      "grad_norm": 0.1765437048722387,
      "learning_rate": 4.359967620215217e-08,
      "loss": 0.8762,
      "step": 2090
    },
    {
      "epoch": 0.35045671666806333,
      "grad_norm": 0.16596355241153435,
      "learning_rate": 4.3510067397032746e-08,
      "loss": 0.8984,
      "step": 2091
    },
    {
      "epoch": 0.3506243191150591,
      "grad_norm": 0.1767402545214463,
      "learning_rate": 4.342052516116968e-08,
      "loss": 0.906,
      "step": 2092
    },
    {
      "epoch": 0.3507919215620548,
      "grad_norm": 0.15524918650085648,
      "learning_rate": 4.3331049600082025e-08,
      "loss": 0.85,
      "step": 2093
    },
    {
      "epoch": 0.3509595240090505,
      "grad_norm": 0.20274431125872422,
      "learning_rate": 4.324164081921012e-08,
      "loss": 0.9028,
      "step": 2094
    },
    {
      "epoch": 0.35112712645604627,
      "grad_norm": 0.19997733613405108,
      "learning_rate": 4.315229892391573e-08,
      "loss": 0.9262,
      "step": 2095
    },
    {
      "epoch": 0.351294728903042,
      "grad_norm": 0.1810675551771918,
      "learning_rate": 4.306302401948175e-08,
      "loss": 0.8694,
      "step": 2096
    },
    {
      "epoch": 0.35146233135003774,
      "grad_norm": 0.16642348922158182,
      "learning_rate": 4.297381621111209e-08,
      "loss": 0.881,
      "step": 2097
    },
    {
      "epoch": 0.35162993379703344,
      "grad_norm": 0.19945967988080926,
      "learning_rate": 4.288467560393169e-08,
      "loss": 0.9091,
      "step": 2098
    },
    {
      "epoch": 0.35179753624402915,
      "grad_norm": 0.1681074281309261,
      "learning_rate": 4.2795602302986154e-08,
      "loss": 0.8656,
      "step": 2099
    },
    {
      "epoch": 0.3519651386910249,
      "grad_norm": 0.19411985611536733,
      "learning_rate": 4.2706596413241914e-08,
      "loss": 0.9175,
      "step": 2100
    },
    {
      "epoch": 0.3521327411380206,
      "grad_norm": 0.1843351573081995,
      "learning_rate": 4.2617658039585934e-08,
      "loss": 0.9342,
      "step": 2101
    },
    {
      "epoch": 0.35230034358501633,
      "grad_norm": 0.18864941005925698,
      "learning_rate": 4.252878728682555e-08,
      "loss": 0.9417,
      "step": 2102
    },
    {
      "epoch": 0.3524679460320121,
      "grad_norm": 0.16907467199878679,
      "learning_rate": 4.243998425968841e-08,
      "loss": 0.8857,
      "step": 2103
    },
    {
      "epoch": 0.3526355484790078,
      "grad_norm": 0.18150744799938598,
      "learning_rate": 4.2351249062822425e-08,
      "loss": 0.8809,
      "step": 2104
    },
    {
      "epoch": 0.3528031509260035,
      "grad_norm": 0.1661011947575896,
      "learning_rate": 4.2262581800795584e-08,
      "loss": 0.8796,
      "step": 2105
    },
    {
      "epoch": 0.35297075337299927,
      "grad_norm": 0.1549558038927998,
      "learning_rate": 4.217398257809568e-08,
      "loss": 0.8335,
      "step": 2106
    },
    {
      "epoch": 0.353138355819995,
      "grad_norm": 0.14679159105670375,
      "learning_rate": 4.208545149913052e-08,
      "loss": 0.8348,
      "step": 2107
    },
    {
      "epoch": 0.3533059582669907,
      "grad_norm": 0.15535717564758458,
      "learning_rate": 4.199698866822746e-08,
      "loss": 0.8025,
      "step": 2108
    },
    {
      "epoch": 0.35347356071398645,
      "grad_norm": 0.1845921081037009,
      "learning_rate": 4.19085941896334e-08,
      "loss": 0.8504,
      "step": 2109
    },
    {
      "epoch": 0.35364116316098215,
      "grad_norm": 0.15872953188372094,
      "learning_rate": 4.182026816751494e-08,
      "loss": 0.8681,
      "step": 2110
    },
    {
      "epoch": 0.35380876560797786,
      "grad_norm": 0.18247416519615334,
      "learning_rate": 4.1732010705957746e-08,
      "loss": 0.8957,
      "step": 2111
    },
    {
      "epoch": 0.3539763680549736,
      "grad_norm": 0.1801979697349846,
      "learning_rate": 4.164382190896676e-08,
      "loss": 0.9032,
      "step": 2112
    },
    {
      "epoch": 0.35414397050196933,
      "grad_norm": 0.18403562409713373,
      "learning_rate": 4.15557018804661e-08,
      "loss": 0.9246,
      "step": 2113
    },
    {
      "epoch": 0.35431157294896504,
      "grad_norm": 0.1755654363471981,
      "learning_rate": 4.1467650724298706e-08,
      "loss": 0.893,
      "step": 2114
    },
    {
      "epoch": 0.3544791753959608,
      "grad_norm": 0.19389847465020288,
      "learning_rate": 4.137966854422647e-08,
      "loss": 0.8946,
      "step": 2115
    },
    {
      "epoch": 0.3546467778429565,
      "grad_norm": 0.16903649401349513,
      "learning_rate": 4.129175544392999e-08,
      "loss": 0.8826,
      "step": 2116
    },
    {
      "epoch": 0.3548143802899522,
      "grad_norm": 0.18205144183098748,
      "learning_rate": 4.1203911527008406e-08,
      "loss": 0.9059,
      "step": 2117
    },
    {
      "epoch": 0.354981982736948,
      "grad_norm": 0.1682013575655726,
      "learning_rate": 4.111613689697929e-08,
      "loss": 0.8418,
      "step": 2118
    },
    {
      "epoch": 0.3551495851839437,
      "grad_norm": 0.19085444817405678,
      "learning_rate": 4.102843165727873e-08,
      "loss": 0.9193,
      "step": 2119
    },
    {
      "epoch": 0.3553171876309394,
      "grad_norm": 0.17665830824182685,
      "learning_rate": 4.094079591126085e-08,
      "loss": 0.9055,
      "step": 2120
    },
    {
      "epoch": 0.35548479007793515,
      "grad_norm": 0.1871582040396501,
      "learning_rate": 4.085322976219802e-08,
      "loss": 0.9175,
      "step": 2121
    },
    {
      "epoch": 0.35565239252493086,
      "grad_norm": 0.1664488789419297,
      "learning_rate": 4.076573331328057e-08,
      "loss": 0.8769,
      "step": 2122
    },
    {
      "epoch": 0.35581999497192657,
      "grad_norm": 0.19656609019286525,
      "learning_rate": 4.06783066676166e-08,
      "loss": 0.8877,
      "step": 2123
    },
    {
      "epoch": 0.35598759741892233,
      "grad_norm": 0.17926223423883966,
      "learning_rate": 4.059094992823209e-08,
      "loss": 0.8797,
      "step": 2124
    },
    {
      "epoch": 0.35615519986591804,
      "grad_norm": 0.15550756900146853,
      "learning_rate": 4.0503663198070514e-08,
      "loss": 0.8376,
      "step": 2125
    },
    {
      "epoch": 0.35632280231291374,
      "grad_norm": 0.19462323271232584,
      "learning_rate": 4.0416446579992926e-08,
      "loss": 0.8752,
      "step": 2126
    },
    {
      "epoch": 0.3564904047599095,
      "grad_norm": 0.18200034888429378,
      "learning_rate": 4.032930017677777e-08,
      "loss": 0.8959,
      "step": 2127
    },
    {
      "epoch": 0.3566580072069052,
      "grad_norm": 0.17473948891910607,
      "learning_rate": 4.024222409112069e-08,
      "loss": 0.8568,
      "step": 2128
    },
    {
      "epoch": 0.3568256096539009,
      "grad_norm": 0.1769764691915139,
      "learning_rate": 4.015521842563444e-08,
      "loss": 0.8589,
      "step": 2129
    },
    {
      "epoch": 0.3569932121008967,
      "grad_norm": 0.17056855358088532,
      "learning_rate": 4.006828328284887e-08,
      "loss": 0.8737,
      "step": 2130
    },
    {
      "epoch": 0.3571608145478924,
      "grad_norm": 0.17444710292049947,
      "learning_rate": 3.998141876521074e-08,
      "loss": 0.8706,
      "step": 2131
    },
    {
      "epoch": 0.3573284169948881,
      "grad_norm": 0.18968516252330983,
      "learning_rate": 3.989462497508346e-08,
      "loss": 0.879,
      "step": 2132
    },
    {
      "epoch": 0.35749601944188386,
      "grad_norm": 0.15316810273429235,
      "learning_rate": 3.980790201474723e-08,
      "loss": 0.8436,
      "step": 2133
    },
    {
      "epoch": 0.35766362188887957,
      "grad_norm": 0.1766392895914881,
      "learning_rate": 3.9721249986398685e-08,
      "loss": 0.8638,
      "step": 2134
    },
    {
      "epoch": 0.35783122433587533,
      "grad_norm": 0.18040030136974342,
      "learning_rate": 3.963466899215089e-08,
      "loss": 0.9241,
      "step": 2135
    },
    {
      "epoch": 0.35799882678287104,
      "grad_norm": 0.15257876804704082,
      "learning_rate": 3.954815913403322e-08,
      "loss": 0.8624,
      "step": 2136
    },
    {
      "epoch": 0.35816642922986675,
      "grad_norm": 0.16580478407929336,
      "learning_rate": 3.946172051399128e-08,
      "loss": 0.8622,
      "step": 2137
    },
    {
      "epoch": 0.3583340316768625,
      "grad_norm": 0.1925896709464674,
      "learning_rate": 3.93753532338866e-08,
      "loss": 0.9161,
      "step": 2138
    },
    {
      "epoch": 0.3585016341238582,
      "grad_norm": 0.164989306088513,
      "learning_rate": 3.9289057395496746e-08,
      "loss": 0.8638,
      "step": 2139
    },
    {
      "epoch": 0.3586692365708539,
      "grad_norm": 0.17902397892562744,
      "learning_rate": 3.920283310051502e-08,
      "loss": 0.8652,
      "step": 2140
    },
    {
      "epoch": 0.3588368390178497,
      "grad_norm": 0.19383732377403204,
      "learning_rate": 3.9116680450550466e-08,
      "loss": 0.8872,
      "step": 2141
    },
    {
      "epoch": 0.3590044414648454,
      "grad_norm": 0.17119347649329522,
      "learning_rate": 3.903059954712772e-08,
      "loss": 0.8667,
      "step": 2142
    },
    {
      "epoch": 0.3591720439118411,
      "grad_norm": 0.17390364589778318,
      "learning_rate": 3.8944590491686814e-08,
      "loss": 0.917,
      "step": 2143
    },
    {
      "epoch": 0.35933964635883686,
      "grad_norm": 0.19217793650492,
      "learning_rate": 3.885865338558307e-08,
      "loss": 0.8951,
      "step": 2144
    },
    {
      "epoch": 0.35950724880583257,
      "grad_norm": 0.1685086653497738,
      "learning_rate": 3.877278833008719e-08,
      "loss": 0.8711,
      "step": 2145
    },
    {
      "epoch": 0.3596748512528283,
      "grad_norm": 0.19182705082949653,
      "learning_rate": 3.8686995426384775e-08,
      "loss": 0.865,
      "step": 2146
    },
    {
      "epoch": 0.35984245369982404,
      "grad_norm": 0.17015867625698294,
      "learning_rate": 3.860127477557653e-08,
      "loss": 0.8775,
      "step": 2147
    },
    {
      "epoch": 0.36001005614681975,
      "grad_norm": 0.17644234663031103,
      "learning_rate": 3.851562647867801e-08,
      "loss": 0.8979,
      "step": 2148
    },
    {
      "epoch": 0.36017765859381545,
      "grad_norm": 0.1928406147457813,
      "learning_rate": 3.8430050636619405e-08,
      "loss": 0.9374,
      "step": 2149
    },
    {
      "epoch": 0.3603452610408112,
      "grad_norm": 0.19607836991624863,
      "learning_rate": 3.834454735024566e-08,
      "loss": 0.9001,
      "step": 2150
    },
    {
      "epoch": 0.3605128634878069,
      "grad_norm": 0.17552110937241946,
      "learning_rate": 3.8259116720316106e-08,
      "loss": 0.8618,
      "step": 2151
    },
    {
      "epoch": 0.36068046593480263,
      "grad_norm": 0.19155163502103917,
      "learning_rate": 3.817375884750443e-08,
      "loss": 0.9405,
      "step": 2152
    },
    {
      "epoch": 0.3608480683817984,
      "grad_norm": 0.1577190522925032,
      "learning_rate": 3.8088473832398804e-08,
      "loss": 0.8488,
      "step": 2153
    },
    {
      "epoch": 0.3610156708287941,
      "grad_norm": 0.1773964536934643,
      "learning_rate": 3.8003261775501286e-08,
      "loss": 0.8977,
      "step": 2154
    },
    {
      "epoch": 0.3611832732757898,
      "grad_norm": 0.2065642254050472,
      "learning_rate": 3.7918122777228024e-08,
      "loss": 0.911,
      "step": 2155
    },
    {
      "epoch": 0.36135087572278557,
      "grad_norm": 0.1730474326158049,
      "learning_rate": 3.783305693790919e-08,
      "loss": 0.8339,
      "step": 2156
    },
    {
      "epoch": 0.3615184781697813,
      "grad_norm": 0.17628576336095025,
      "learning_rate": 3.774806435778858e-08,
      "loss": 0.9004,
      "step": 2157
    },
    {
      "epoch": 0.361686080616777,
      "grad_norm": 0.1934437674110247,
      "learning_rate": 3.766314513702378e-08,
      "loss": 0.881,
      "step": 2158
    },
    {
      "epoch": 0.36185368306377275,
      "grad_norm": 0.18120509157071985,
      "learning_rate": 3.7578299375685906e-08,
      "loss": 0.8979,
      "step": 2159
    },
    {
      "epoch": 0.36202128551076845,
      "grad_norm": 0.16972712931759143,
      "learning_rate": 3.7493527173759455e-08,
      "loss": 0.8921,
      "step": 2160
    },
    {
      "epoch": 0.36218888795776416,
      "grad_norm": 0.1680360003685835,
      "learning_rate": 3.740882863114223e-08,
      "loss": 0.8622,
      "step": 2161
    },
    {
      "epoch": 0.3623564904047599,
      "grad_norm": 0.18916748237358316,
      "learning_rate": 3.732420384764533e-08,
      "loss": 0.9133,
      "step": 2162
    },
    {
      "epoch": 0.36252409285175563,
      "grad_norm": 0.1825212314286375,
      "learning_rate": 3.723965292299289e-08,
      "loss": 0.9034,
      "step": 2163
    },
    {
      "epoch": 0.36269169529875134,
      "grad_norm": 0.1505212942245953,
      "learning_rate": 3.715517595682193e-08,
      "loss": 0.8525,
      "step": 2164
    },
    {
      "epoch": 0.3628592977457471,
      "grad_norm": 0.18899763656950005,
      "learning_rate": 3.707077304868245e-08,
      "loss": 0.9093,
      "step": 2165
    },
    {
      "epoch": 0.3630269001927428,
      "grad_norm": 0.17242359387432465,
      "learning_rate": 3.6986444298037046e-08,
      "loss": 0.8892,
      "step": 2166
    },
    {
      "epoch": 0.3631945026397385,
      "grad_norm": 0.17218934731753946,
      "learning_rate": 3.6902189804261005e-08,
      "loss": 0.8798,
      "step": 2167
    },
    {
      "epoch": 0.3633621050867343,
      "grad_norm": 0.16649788140797417,
      "learning_rate": 3.681800966664213e-08,
      "loss": 0.8619,
      "step": 2168
    },
    {
      "epoch": 0.36352970753373,
      "grad_norm": 0.18719360211790737,
      "learning_rate": 3.6733903984380546e-08,
      "loss": 0.893,
      "step": 2169
    },
    {
      "epoch": 0.3636973099807257,
      "grad_norm": 0.18752977447461286,
      "learning_rate": 3.6649872856588584e-08,
      "loss": 0.9116,
      "step": 2170
    },
    {
      "epoch": 0.36386491242772145,
      "grad_norm": 0.18433890036890743,
      "learning_rate": 3.6565916382290875e-08,
      "loss": 0.9557,
      "step": 2171
    },
    {
      "epoch": 0.36403251487471716,
      "grad_norm": 0.19446646780493426,
      "learning_rate": 3.648203466042391e-08,
      "loss": 0.8908,
      "step": 2172
    },
    {
      "epoch": 0.3642001173217129,
      "grad_norm": 0.16901579540705894,
      "learning_rate": 3.63982277898362e-08,
      "loss": 0.8567,
      "step": 2173
    },
    {
      "epoch": 0.36436771976870863,
      "grad_norm": 0.16718391209090014,
      "learning_rate": 3.631449586928806e-08,
      "loss": 0.8727,
      "step": 2174
    },
    {
      "epoch": 0.36453532221570434,
      "grad_norm": 0.16661311573653634,
      "learning_rate": 3.6230838997451364e-08,
      "loss": 0.8609,
      "step": 2175
    },
    {
      "epoch": 0.3647029246627001,
      "grad_norm": 0.1675364616044119,
      "learning_rate": 3.614725727290969e-08,
      "loss": 0.8277,
      "step": 2176
    },
    {
      "epoch": 0.3648705271096958,
      "grad_norm": 0.16561400222469183,
      "learning_rate": 3.6063750794157965e-08,
      "loss": 0.8472,
      "step": 2177
    },
    {
      "epoch": 0.3650381295566915,
      "grad_norm": 0.1805295901828147,
      "learning_rate": 3.598031965960238e-08,
      "loss": 0.9021,
      "step": 2178
    },
    {
      "epoch": 0.3652057320036873,
      "grad_norm": 0.18171348033767232,
      "learning_rate": 3.589696396756058e-08,
      "loss": 0.9002,
      "step": 2179
    },
    {
      "epoch": 0.365373334450683,
      "grad_norm": 0.17510880182187763,
      "learning_rate": 3.581368381626109e-08,
      "loss": 0.8753,
      "step": 2180
    },
    {
      "epoch": 0.3655409368976787,
      "grad_norm": 0.17604923239201012,
      "learning_rate": 3.5730479303843444e-08,
      "loss": 0.8748,
      "step": 2181
    },
    {
      "epoch": 0.36570853934467445,
      "grad_norm": 0.1884533005998993,
      "learning_rate": 3.5647350528358136e-08,
      "loss": 0.9235,
      "step": 2182
    },
    {
      "epoch": 0.36587614179167016,
      "grad_norm": 0.15779686940845022,
      "learning_rate": 3.556429758776629e-08,
      "loss": 0.8681,
      "step": 2183
    },
    {
      "epoch": 0.36604374423866587,
      "grad_norm": 0.17793804885640252,
      "learning_rate": 3.548132057993976e-08,
      "loss": 0.875,
      "step": 2184
    },
    {
      "epoch": 0.36621134668566163,
      "grad_norm": 0.1686952747938433,
      "learning_rate": 3.539841960266093e-08,
      "loss": 0.8721,
      "step": 2185
    },
    {
      "epoch": 0.36637894913265734,
      "grad_norm": 0.1715341357149523,
      "learning_rate": 3.5315594753622504e-08,
      "loss": 0.9054,
      "step": 2186
    },
    {
      "epoch": 0.36654655157965305,
      "grad_norm": 0.17804440704891325,
      "learning_rate": 3.523284613042745e-08,
      "loss": 0.8819,
      "step": 2187
    },
    {
      "epoch": 0.3667141540266488,
      "grad_norm": 0.1907345414294638,
      "learning_rate": 3.5150173830589046e-08,
      "loss": 0.9162,
      "step": 2188
    },
    {
      "epoch": 0.3668817564736445,
      "grad_norm": 0.20043818963040969,
      "learning_rate": 3.5067577951530556e-08,
      "loss": 0.945,
      "step": 2189
    },
    {
      "epoch": 0.3670493589206402,
      "grad_norm": 0.21081120250502414,
      "learning_rate": 3.498505859058515e-08,
      "loss": 0.9175,
      "step": 2190
    },
    {
      "epoch": 0.367216961367636,
      "grad_norm": 0.1713478448845793,
      "learning_rate": 3.49026158449959e-08,
      "loss": 0.8774,
      "step": 2191
    },
    {
      "epoch": 0.3673845638146317,
      "grad_norm": 0.18751486599490216,
      "learning_rate": 3.482024981191554e-08,
      "loss": 0.8929,
      "step": 2192
    },
    {
      "epoch": 0.3675521662616274,
      "grad_norm": 0.16566173143693333,
      "learning_rate": 3.473796058840637e-08,
      "loss": 0.8741,
      "step": 2193
    },
    {
      "epoch": 0.36771976870862316,
      "grad_norm": 0.1714289151317342,
      "learning_rate": 3.465574827144028e-08,
      "loss": 0.9011,
      "step": 2194
    },
    {
      "epoch": 0.36788737115561887,
      "grad_norm": 0.2042427499870332,
      "learning_rate": 3.4573612957898494e-08,
      "loss": 0.9124,
      "step": 2195
    },
    {
      "epoch": 0.3680549736026146,
      "grad_norm": 0.19352040418515498,
      "learning_rate": 3.4491554744571414e-08,
      "loss": 0.9039,
      "step": 2196
    },
    {
      "epoch": 0.36822257604961034,
      "grad_norm": 0.18115117696437996,
      "learning_rate": 3.440957372815872e-08,
      "loss": 0.8863,
      "step": 2197
    },
    {
      "epoch": 0.36839017849660605,
      "grad_norm": 0.17830887291100417,
      "learning_rate": 3.432767000526896e-08,
      "loss": 0.911,
      "step": 2198
    },
    {
      "epoch": 0.36855778094360175,
      "grad_norm": 0.1737286888247694,
      "learning_rate": 3.4245843672419726e-08,
      "loss": 0.8634,
      "step": 2199
    },
    {
      "epoch": 0.3687253833905975,
      "grad_norm": 0.18304378448389955,
      "learning_rate": 3.416409482603743e-08,
      "loss": 0.8977,
      "step": 2200
    },
    {
      "epoch": 0.3688929858375932,
      "grad_norm": 0.17986638961316326,
      "learning_rate": 3.408242356245701e-08,
      "loss": 0.9174,
      "step": 2201
    },
    {
      "epoch": 0.36906058828458893,
      "grad_norm": 0.18776256172127803,
      "learning_rate": 3.4000829977922166e-08,
      "loss": 0.8927,
      "step": 2202
    },
    {
      "epoch": 0.3692281907315847,
      "grad_norm": 0.15615759176925195,
      "learning_rate": 3.391931416858492e-08,
      "loss": 0.8222,
      "step": 2203
    },
    {
      "epoch": 0.3693957931785804,
      "grad_norm": 0.17207304723719555,
      "learning_rate": 3.383787623050569e-08,
      "loss": 0.8614,
      "step": 2204
    },
    {
      "epoch": 0.3695633956255761,
      "grad_norm": 0.16595703613855728,
      "learning_rate": 3.3756516259653146e-08,
      "loss": 0.8802,
      "step": 2205
    },
    {
      "epoch": 0.36973099807257187,
      "grad_norm": 0.18270561939439015,
      "learning_rate": 3.367523435190412e-08,
      "loss": 0.9083,
      "step": 2206
    },
    {
      "epoch": 0.3698986005195676,
      "grad_norm": 0.16638223544701272,
      "learning_rate": 3.35940306030433e-08,
      "loss": 0.8686,
      "step": 2207
    },
    {
      "epoch": 0.3700662029665633,
      "grad_norm": 0.1834670139606147,
      "learning_rate": 3.351290510876347e-08,
      "loss": 0.9212,
      "step": 2208
    },
    {
      "epoch": 0.37023380541355905,
      "grad_norm": 0.17625472429829106,
      "learning_rate": 3.3431857964665e-08,
      "loss": 0.9103,
      "step": 2209
    },
    {
      "epoch": 0.37040140786055475,
      "grad_norm": 0.1714630241459331,
      "learning_rate": 3.335088926625608e-08,
      "loss": 0.8607,
      "step": 2210
    },
    {
      "epoch": 0.3705690103075505,
      "grad_norm": 0.15251951201488115,
      "learning_rate": 3.326999910895242e-08,
      "loss": 0.8464,
      "step": 2211
    },
    {
      "epoch": 0.3707366127545462,
      "grad_norm": 0.16264159765518466,
      "learning_rate": 3.3189187588077125e-08,
      "loss": 0.8215,
      "step": 2212
    },
    {
      "epoch": 0.37090421520154193,
      "grad_norm": 0.18738252149438595,
      "learning_rate": 3.3108454798860655e-08,
      "loss": 0.895,
      "step": 2213
    },
    {
      "epoch": 0.3710718176485377,
      "grad_norm": 0.186457347540789,
      "learning_rate": 3.302780083644073e-08,
      "loss": 0.9324,
      "step": 2214
    },
    {
      "epoch": 0.3712394200955334,
      "grad_norm": 0.17716758175817368,
      "learning_rate": 3.2947225795862125e-08,
      "loss": 0.8898,
      "step": 2215
    },
    {
      "epoch": 0.3714070225425291,
      "grad_norm": 0.17807910694330867,
      "learning_rate": 3.286672977207663e-08,
      "loss": 0.9408,
      "step": 2216
    },
    {
      "epoch": 0.37157462498952487,
      "grad_norm": 0.15403711837099562,
      "learning_rate": 3.278631285994298e-08,
      "loss": 0.8726,
      "step": 2217
    },
    {
      "epoch": 0.3717422274365206,
      "grad_norm": 0.17454824669869312,
      "learning_rate": 3.2705975154226585e-08,
      "loss": 0.8761,
      "step": 2218
    },
    {
      "epoch": 0.3719098298835163,
      "grad_norm": 0.16773454589479786,
      "learning_rate": 3.262571674959954e-08,
      "loss": 0.8906,
      "step": 2219
    },
    {
      "epoch": 0.37207743233051205,
      "grad_norm": 0.18866296981541444,
      "learning_rate": 3.254553774064053e-08,
      "loss": 0.8822,
      "step": 2220
    },
    {
      "epoch": 0.37224503477750776,
      "grad_norm": 0.17966402730652575,
      "learning_rate": 3.246543822183469e-08,
      "loss": 0.9073,
      "step": 2221
    },
    {
      "epoch": 0.37241263722450346,
      "grad_norm": 0.18083601075492017,
      "learning_rate": 3.2385418287573386e-08,
      "loss": 0.8691,
      "step": 2222
    },
    {
      "epoch": 0.3725802396714992,
      "grad_norm": 0.1606641525820176,
      "learning_rate": 3.230547803215432e-08,
      "loss": 0.8614,
      "step": 2223
    },
    {
      "epoch": 0.37274784211849493,
      "grad_norm": 0.17457680094922753,
      "learning_rate": 3.222561754978118e-08,
      "loss": 0.8617,
      "step": 2224
    },
    {
      "epoch": 0.37291544456549064,
      "grad_norm": 0.1568807823430049,
      "learning_rate": 3.214583693456374e-08,
      "loss": 0.8726,
      "step": 2225
    },
    {
      "epoch": 0.3730830470124864,
      "grad_norm": 0.1921099428435798,
      "learning_rate": 3.206613628051765e-08,
      "loss": 0.9048,
      "step": 2226
    },
    {
      "epoch": 0.3732506494594821,
      "grad_norm": 0.18499013625396807,
      "learning_rate": 3.1986515681564255e-08,
      "loss": 0.8949,
      "step": 2227
    },
    {
      "epoch": 0.3734182519064778,
      "grad_norm": 0.17149660067065628,
      "learning_rate": 3.190697523153065e-08,
      "loss": 0.8935,
      "step": 2228
    },
    {
      "epoch": 0.3735858543534736,
      "grad_norm": 0.16824879068503104,
      "learning_rate": 3.1827515024149445e-08,
      "loss": 0.8743,
      "step": 2229
    },
    {
      "epoch": 0.3737534568004693,
      "grad_norm": 0.16970546649213136,
      "learning_rate": 3.174813515305861e-08,
      "loss": 0.8718,
      "step": 2230
    },
    {
      "epoch": 0.373921059247465,
      "grad_norm": 0.16936797543190665,
      "learning_rate": 3.166883571180159e-08,
      "loss": 0.8649,
      "step": 2231
    },
    {
      "epoch": 0.37408866169446076,
      "grad_norm": 0.18621513977382143,
      "learning_rate": 3.1589616793827e-08,
      "loss": 0.9187,
      "step": 2232
    },
    {
      "epoch": 0.37425626414145646,
      "grad_norm": 0.15513984079894672,
      "learning_rate": 3.151047849248847e-08,
      "loss": 0.8562,
      "step": 2233
    },
    {
      "epoch": 0.37442386658845217,
      "grad_norm": 0.1483370725712312,
      "learning_rate": 3.143142090104478e-08,
      "loss": 0.8151,
      "step": 2234
    },
    {
      "epoch": 0.37459146903544793,
      "grad_norm": 0.1702854513449247,
      "learning_rate": 3.135244411265948e-08,
      "loss": 0.845,
      "step": 2235
    },
    {
      "epoch": 0.37475907148244364,
      "grad_norm": 0.18357018516316495,
      "learning_rate": 3.127354822040089e-08,
      "loss": 0.9216,
      "step": 2236
    },
    {
      "epoch": 0.37492667392943935,
      "grad_norm": 0.18110506646598837,
      "learning_rate": 3.119473331724218e-08,
      "loss": 0.8998,
      "step": 2237
    },
    {
      "epoch": 0.3750942763764351,
      "grad_norm": 0.16089906815749275,
      "learning_rate": 3.1115999496060884e-08,
      "loss": 0.8487,
      "step": 2238
    },
    {
      "epoch": 0.3752618788234308,
      "grad_norm": 0.17989112560606,
      "learning_rate": 3.103734684963901e-08,
      "loss": 0.885,
      "step": 2239
    },
    {
      "epoch": 0.3754294812704265,
      "grad_norm": 0.17790081659599938,
      "learning_rate": 3.0958775470663046e-08,
      "loss": 0.8772,
      "step": 2240
    },
    {
      "epoch": 0.3755970837174223,
      "grad_norm": 0.17666897291020323,
      "learning_rate": 3.088028545172352e-08,
      "loss": 0.8731,
      "step": 2241
    },
    {
      "epoch": 0.375764686164418,
      "grad_norm": 0.19415166179364915,
      "learning_rate": 3.080187688531525e-08,
      "loss": 0.8863,
      "step": 2242
    },
    {
      "epoch": 0.3759322886114137,
      "grad_norm": 0.1915772673350963,
      "learning_rate": 3.072354986383699e-08,
      "loss": 0.9015,
      "step": 2243
    },
    {
      "epoch": 0.37609989105840946,
      "grad_norm": 0.17530437866676965,
      "learning_rate": 3.064530447959143e-08,
      "loss": 0.8851,
      "step": 2244
    },
    {
      "epoch": 0.37626749350540517,
      "grad_norm": 0.15103869371794737,
      "learning_rate": 3.0567140824784955e-08,
      "loss": 0.812,
      "step": 2245
    },
    {
      "epoch": 0.3764350959524009,
      "grad_norm": 0.17714270495530085,
      "learning_rate": 3.048905899152778e-08,
      "loss": 0.9435,
      "step": 2246
    },
    {
      "epoch": 0.37660269839939664,
      "grad_norm": 0.18955421044156795,
      "learning_rate": 3.0411059071833664e-08,
      "loss": 0.9031,
      "step": 2247
    },
    {
      "epoch": 0.37677030084639235,
      "grad_norm": 0.16876234827892694,
      "learning_rate": 3.0333141157619735e-08,
      "loss": 0.9451,
      "step": 2248
    },
    {
      "epoch": 0.3769379032933881,
      "grad_norm": 0.19347880622756114,
      "learning_rate": 3.025530534070664e-08,
      "loss": 0.8907,
      "step": 2249
    },
    {
      "epoch": 0.3771055057403838,
      "grad_norm": 0.18066980561736856,
      "learning_rate": 3.017755171281813e-08,
      "loss": 0.918,
      "step": 2250
    },
    {
      "epoch": 0.3772731081873795,
      "grad_norm": 0.18092823163345048,
      "learning_rate": 3.00998803655812e-08,
      "loss": 0.9212,
      "step": 2251
    },
    {
      "epoch": 0.3774407106343753,
      "grad_norm": 0.1921489100218455,
      "learning_rate": 3.002229139052589e-08,
      "loss": 0.8819,
      "step": 2252
    },
    {
      "epoch": 0.377608313081371,
      "grad_norm": 0.1717038935932481,
      "learning_rate": 2.994478487908506e-08,
      "loss": 0.8726,
      "step": 2253
    },
    {
      "epoch": 0.3777759155283667,
      "grad_norm": 0.18420555885558343,
      "learning_rate": 2.9867360922594554e-08,
      "loss": 0.9058,
      "step": 2254
    },
    {
      "epoch": 0.37794351797536246,
      "grad_norm": 0.18760807987440856,
      "learning_rate": 2.9790019612292804e-08,
      "loss": 0.8904,
      "step": 2255
    },
    {
      "epoch": 0.37811112042235817,
      "grad_norm": 0.18050968599389863,
      "learning_rate": 2.9712761039320854e-08,
      "loss": 0.8981,
      "step": 2256
    },
    {
      "epoch": 0.3782787228693539,
      "grad_norm": 0.17850317264000953,
      "learning_rate": 2.9635585294722333e-08,
      "loss": 0.8913,
      "step": 2257
    },
    {
      "epoch": 0.37844632531634964,
      "grad_norm": 0.1745866508830542,
      "learning_rate": 2.955849246944322e-08,
      "loss": 0.9134,
      "step": 2258
    },
    {
      "epoch": 0.37861392776334535,
      "grad_norm": 0.19407603447413693,
      "learning_rate": 2.9481482654331737e-08,
      "loss": 0.8992,
      "step": 2259
    },
    {
      "epoch": 0.37878153021034106,
      "grad_norm": 0.18791855535821458,
      "learning_rate": 2.9404555940138363e-08,
      "loss": 0.8876,
      "step": 2260
    },
    {
      "epoch": 0.3789491326573368,
      "grad_norm": 0.17048588091022074,
      "learning_rate": 2.9327712417515593e-08,
      "loss": 0.8808,
      "step": 2261
    },
    {
      "epoch": 0.3791167351043325,
      "grad_norm": 0.16603842992275455,
      "learning_rate": 2.925095217701783e-08,
      "loss": 0.8626,
      "step": 2262
    },
    {
      "epoch": 0.37928433755132823,
      "grad_norm": 0.1647231327110639,
      "learning_rate": 2.9174275309101538e-08,
      "loss": 0.8558,
      "step": 2263
    },
    {
      "epoch": 0.379451939998324,
      "grad_norm": 0.18323572552161577,
      "learning_rate": 2.9097681904124726e-08,
      "loss": 0.8986,
      "step": 2264
    },
    {
      "epoch": 0.3796195424453197,
      "grad_norm": 0.1482157106352014,
      "learning_rate": 2.902117205234711e-08,
      "loss": 0.8651,
      "step": 2265
    },
    {
      "epoch": 0.3797871448923154,
      "grad_norm": 0.17026721355467872,
      "learning_rate": 2.8944745843930008e-08,
      "loss": 0.9152,
      "step": 2266
    },
    {
      "epoch": 0.37995474733931117,
      "grad_norm": 0.15691372224584074,
      "learning_rate": 2.8868403368936056e-08,
      "loss": 0.8591,
      "step": 2267
    },
    {
      "epoch": 0.3801223497863069,
      "grad_norm": 0.1701792200046499,
      "learning_rate": 2.8792144717329327e-08,
      "loss": 0.8776,
      "step": 2268
    },
    {
      "epoch": 0.3802899522333026,
      "grad_norm": 0.16698291571220691,
      "learning_rate": 2.8715969978975085e-08,
      "loss": 0.8307,
      "step": 2269
    },
    {
      "epoch": 0.38045755468029835,
      "grad_norm": 0.2080990951070255,
      "learning_rate": 2.8639879243639665e-08,
      "loss": 0.9047,
      "step": 2270
    },
    {
      "epoch": 0.38062515712729406,
      "grad_norm": 0.15545152372776444,
      "learning_rate": 2.856387260099039e-08,
      "loss": 0.823,
      "step": 2271
    },
    {
      "epoch": 0.38079275957428976,
      "grad_norm": 0.2026495800309594,
      "learning_rate": 2.8487950140595628e-08,
      "loss": 0.9258,
      "step": 2272
    },
    {
      "epoch": 0.3809603620212855,
      "grad_norm": 0.16874953368016174,
      "learning_rate": 2.8412111951924355e-08,
      "loss": 0.8688,
      "step": 2273
    },
    {
      "epoch": 0.38112796446828123,
      "grad_norm": 0.17519168066946691,
      "learning_rate": 2.8336358124346384e-08,
      "loss": 0.8976,
      "step": 2274
    },
    {
      "epoch": 0.38129556691527694,
      "grad_norm": 0.1787294551528771,
      "learning_rate": 2.826068874713208e-08,
      "loss": 0.8786,
      "step": 2275
    },
    {
      "epoch": 0.3814631693622727,
      "grad_norm": 0.18369346470770426,
      "learning_rate": 2.8185103909452225e-08,
      "loss": 0.9126,
      "step": 2276
    },
    {
      "epoch": 0.3816307718092684,
      "grad_norm": 0.16727703509273584,
      "learning_rate": 2.810960370037807e-08,
      "loss": 0.8732,
      "step": 2277
    },
    {
      "epoch": 0.3817983742562641,
      "grad_norm": 0.19706090912856158,
      "learning_rate": 2.803418820888105e-08,
      "loss": 0.9099,
      "step": 2278
    },
    {
      "epoch": 0.3819659767032599,
      "grad_norm": 0.17440455696605076,
      "learning_rate": 2.7958857523832835e-08,
      "loss": 0.8744,
      "step": 2279
    },
    {
      "epoch": 0.3821335791502556,
      "grad_norm": 0.17371705781961927,
      "learning_rate": 2.7883611734005154e-08,
      "loss": 0.8886,
      "step": 2280
    },
    {
      "epoch": 0.3823011815972513,
      "grad_norm": 0.17341990770021284,
      "learning_rate": 2.780845092806964e-08,
      "loss": 0.8951,
      "step": 2281
    },
    {
      "epoch": 0.38246878404424706,
      "grad_norm": 0.16439231567709037,
      "learning_rate": 2.7733375194597774e-08,
      "loss": 0.8573,
      "step": 2282
    },
    {
      "epoch": 0.38263638649124276,
      "grad_norm": 0.1686469164160783,
      "learning_rate": 2.765838462206087e-08,
      "loss": 0.8438,
      "step": 2283
    },
    {
      "epoch": 0.38280398893823847,
      "grad_norm": 0.16609839270352966,
      "learning_rate": 2.758347929882985e-08,
      "loss": 0.9121,
      "step": 2284
    },
    {
      "epoch": 0.38297159138523423,
      "grad_norm": 0.16467894223558185,
      "learning_rate": 2.750865931317511e-08,
      "loss": 0.8691,
      "step": 2285
    },
    {
      "epoch": 0.38313919383222994,
      "grad_norm": 0.18856586250362356,
      "learning_rate": 2.743392475326659e-08,
      "loss": 0.9299,
      "step": 2286
    },
    {
      "epoch": 0.3833067962792257,
      "grad_norm": 0.17928298526704828,
      "learning_rate": 2.735927570717348e-08,
      "loss": 0.9241,
      "step": 2287
    },
    {
      "epoch": 0.3834743987262214,
      "grad_norm": 0.2520697503687935,
      "learning_rate": 2.728471226286415e-08,
      "loss": 0.8722,
      "step": 2288
    },
    {
      "epoch": 0.3836420011732171,
      "grad_norm": 0.17156018833428563,
      "learning_rate": 2.721023450820631e-08,
      "loss": 0.9082,
      "step": 2289
    },
    {
      "epoch": 0.3838096036202129,
      "grad_norm": 0.16995700060211938,
      "learning_rate": 2.713584253096648e-08,
      "loss": 0.87,
      "step": 2290
    },
    {
      "epoch": 0.3839772060672086,
      "grad_norm": 0.17889734184930398,
      "learning_rate": 2.7061536418810126e-08,
      "loss": 0.8766,
      "step": 2291
    },
    {
      "epoch": 0.3841448085142043,
      "grad_norm": 0.16653270046821722,
      "learning_rate": 2.698731625930165e-08,
      "loss": 0.8608,
      "step": 2292
    },
    {
      "epoch": 0.38431241096120006,
      "grad_norm": 0.17306560295830234,
      "learning_rate": 2.691318213990401e-08,
      "loss": 0.8801,
      "step": 2293
    },
    {
      "epoch": 0.38448001340819576,
      "grad_norm": 0.15708083285289107,
      "learning_rate": 2.6839134147978892e-08,
      "loss": 0.8876,
      "step": 2294
    },
    {
      "epoch": 0.38464761585519147,
      "grad_norm": 0.1859155283345047,
      "learning_rate": 2.676517237078646e-08,
      "loss": 0.9044,
      "step": 2295
    },
    {
      "epoch": 0.38481521830218723,
      "grad_norm": 0.1922905445089486,
      "learning_rate": 2.669129689548524e-08,
      "loss": 0.8971,
      "step": 2296
    },
    {
      "epoch": 0.38498282074918294,
      "grad_norm": 0.1901983596750971,
      "learning_rate": 2.6617507809132033e-08,
      "loss": 0.9223,
      "step": 2297
    },
    {
      "epoch": 0.38515042319617865,
      "grad_norm": 0.14504199504802415,
      "learning_rate": 2.6543805198681957e-08,
      "loss": 0.8486,
      "step": 2298
    },
    {
      "epoch": 0.3853180256431744,
      "grad_norm": 0.17814457009957632,
      "learning_rate": 2.647018915098809e-08,
      "loss": 0.8687,
      "step": 2299
    },
    {
      "epoch": 0.3854856280901701,
      "grad_norm": 0.17682298068387914,
      "learning_rate": 2.6396659752801586e-08,
      "loss": 0.8612,
      "step": 2300
    },
    {
      "epoch": 0.3856532305371658,
      "grad_norm": 0.17837455405539013,
      "learning_rate": 2.632321709077151e-08,
      "loss": 0.8594,
      "step": 2301
    },
    {
      "epoch": 0.3858208329841616,
      "grad_norm": 0.17356155260197395,
      "learning_rate": 2.6249861251444605e-08,
      "loss": 0.8787,
      "step": 2302
    },
    {
      "epoch": 0.3859884354311573,
      "grad_norm": 0.17990869134862636,
      "learning_rate": 2.617659232126542e-08,
      "loss": 0.9025,
      "step": 2303
    },
    {
      "epoch": 0.386156037878153,
      "grad_norm": 0.17628577436552162,
      "learning_rate": 2.6103410386575984e-08,
      "loss": 0.8652,
      "step": 2304
    },
    {
      "epoch": 0.38632364032514876,
      "grad_norm": 0.17509648425811153,
      "learning_rate": 2.6030315533615877e-08,
      "loss": 0.8752,
      "step": 2305
    },
    {
      "epoch": 0.38649124277214447,
      "grad_norm": 0.1711885825807002,
      "learning_rate": 2.5957307848522068e-08,
      "loss": 0.8925,
      "step": 2306
    },
    {
      "epoch": 0.3866588452191402,
      "grad_norm": 0.18938990307371967,
      "learning_rate": 2.5884387417328755e-08,
      "loss": 0.9232,
      "step": 2307
    },
    {
      "epoch": 0.38682644766613594,
      "grad_norm": 0.17927016486999994,
      "learning_rate": 2.5811554325967312e-08,
      "loss": 0.8985,
      "step": 2308
    },
    {
      "epoch": 0.38699405011313165,
      "grad_norm": 0.1559917522225125,
      "learning_rate": 2.573880866026623e-08,
      "loss": 0.8577,
      "step": 2309
    },
    {
      "epoch": 0.38716165256012736,
      "grad_norm": 0.1760030889423118,
      "learning_rate": 2.5666150505951002e-08,
      "loss": 0.9073,
      "step": 2310
    },
    {
      "epoch": 0.3873292550071231,
      "grad_norm": 0.19316010186977933,
      "learning_rate": 2.5593579948643872e-08,
      "loss": 0.932,
      "step": 2311
    },
    {
      "epoch": 0.3874968574541188,
      "grad_norm": 0.17122988311189843,
      "learning_rate": 2.5521097073864018e-08,
      "loss": 0.8713,
      "step": 2312
    },
    {
      "epoch": 0.38766445990111453,
      "grad_norm": 0.18700425946225172,
      "learning_rate": 2.5448701967027164e-08,
      "loss": 0.8887,
      "step": 2313
    },
    {
      "epoch": 0.3878320623481103,
      "grad_norm": 0.18684354442358336,
      "learning_rate": 2.5376394713445636e-08,
      "loss": 0.8938,
      "step": 2314
    },
    {
      "epoch": 0.387999664795106,
      "grad_norm": 0.1552313784893093,
      "learning_rate": 2.5304175398328277e-08,
      "loss": 0.8767,
      "step": 2315
    },
    {
      "epoch": 0.3881672672421017,
      "grad_norm": 0.1830031793506012,
      "learning_rate": 2.523204410678028e-08,
      "loss": 0.9275,
      "step": 2316
    },
    {
      "epoch": 0.3883348696890975,
      "grad_norm": 0.1979489729578907,
      "learning_rate": 2.516000092380306e-08,
      "loss": 0.9194,
      "step": 2317
    },
    {
      "epoch": 0.3885024721360932,
      "grad_norm": 0.17195318654233074,
      "learning_rate": 2.50880459342943e-08,
      "loss": 0.8648,
      "step": 2318
    },
    {
      "epoch": 0.3886700745830889,
      "grad_norm": 0.17850020416955845,
      "learning_rate": 2.501617922304763e-08,
      "loss": 0.9075,
      "step": 2319
    },
    {
      "epoch": 0.38883767703008465,
      "grad_norm": 0.17299800244787994,
      "learning_rate": 2.4944400874752736e-08,
      "loss": 0.8492,
      "step": 2320
    },
    {
      "epoch": 0.38900527947708036,
      "grad_norm": 0.1850091775445275,
      "learning_rate": 2.4872710973995203e-08,
      "loss": 0.8953,
      "step": 2321
    },
    {
      "epoch": 0.38917288192407606,
      "grad_norm": 0.18152264061539491,
      "learning_rate": 2.480110960525631e-08,
      "loss": 0.8872,
      "step": 2322
    },
    {
      "epoch": 0.3893404843710718,
      "grad_norm": 0.2035850113946039,
      "learning_rate": 2.4729596852912994e-08,
      "loss": 0.9009,
      "step": 2323
    },
    {
      "epoch": 0.38950808681806753,
      "grad_norm": 0.17170410132005148,
      "learning_rate": 2.4658172801237863e-08,
      "loss": 0.891,
      "step": 2324
    },
    {
      "epoch": 0.3896756892650633,
      "grad_norm": 0.1526390765850433,
      "learning_rate": 2.4586837534398898e-08,
      "loss": 0.8351,
      "step": 2325
    },
    {
      "epoch": 0.389843291712059,
      "grad_norm": 0.20713732595712694,
      "learning_rate": 2.4515591136459514e-08,
      "loss": 0.8773,
      "step": 2326
    },
    {
      "epoch": 0.3900108941590547,
      "grad_norm": 0.1839327226086983,
      "learning_rate": 2.4444433691378407e-08,
      "loss": 0.9018,
      "step": 2327
    },
    {
      "epoch": 0.3901784966060505,
      "grad_norm": 0.17929362877468777,
      "learning_rate": 2.4373365283009384e-08,
      "loss": 0.8559,
      "step": 2328
    },
    {
      "epoch": 0.3903460990530462,
      "grad_norm": 0.1415883657345691,
      "learning_rate": 2.4302385995101415e-08,
      "loss": 0.828,
      "step": 2329
    },
    {
      "epoch": 0.3905137015000419,
      "grad_norm": 0.2215476224210223,
      "learning_rate": 2.423149591129835e-08,
      "loss": 0.8937,
      "step": 2330
    },
    {
      "epoch": 0.39068130394703765,
      "grad_norm": 0.18761512747288475,
      "learning_rate": 2.4160695115138998e-08,
      "loss": 0.8927,
      "step": 2331
    },
    {
      "epoch": 0.39084890639403336,
      "grad_norm": 0.17424970481505475,
      "learning_rate": 2.4089983690056957e-08,
      "loss": 0.8873,
      "step": 2332
    },
    {
      "epoch": 0.39101650884102906,
      "grad_norm": 0.1532898943172838,
      "learning_rate": 2.4019361719380428e-08,
      "loss": 0.8564,
      "step": 2333
    },
    {
      "epoch": 0.3911841112880248,
      "grad_norm": 0.17839642669442848,
      "learning_rate": 2.394882928633223e-08,
      "loss": 0.8935,
      "step": 2334
    },
    {
      "epoch": 0.39135171373502053,
      "grad_norm": 0.1644976191375398,
      "learning_rate": 2.387838647402972e-08,
      "loss": 0.8502,
      "step": 2335
    },
    {
      "epoch": 0.39151931618201624,
      "grad_norm": 0.1632338794672582,
      "learning_rate": 2.380803336548457e-08,
      "loss": 0.8817,
      "step": 2336
    },
    {
      "epoch": 0.391686918629012,
      "grad_norm": 0.17309495582943732,
      "learning_rate": 2.3737770043602778e-08,
      "loss": 0.8629,
      "step": 2337
    },
    {
      "epoch": 0.3918545210760077,
      "grad_norm": 0.1727554430184219,
      "learning_rate": 2.366759659118458e-08,
      "loss": 0.8778,
      "step": 2338
    },
    {
      "epoch": 0.3920221235230034,
      "grad_norm": 0.17215110682701018,
      "learning_rate": 2.3597513090924227e-08,
      "loss": 0.8787,
      "step": 2339
    },
    {
      "epoch": 0.3921897259699992,
      "grad_norm": 0.19275483802708968,
      "learning_rate": 2.3527519625409962e-08,
      "loss": 0.8714,
      "step": 2340
    },
    {
      "epoch": 0.3923573284169949,
      "grad_norm": 0.17907090653553662,
      "learning_rate": 2.3457616277124015e-08,
      "loss": 0.8593,
      "step": 2341
    },
    {
      "epoch": 0.3925249308639906,
      "grad_norm": 0.16789549702452056,
      "learning_rate": 2.3387803128442386e-08,
      "loss": 0.8873,
      "step": 2342
    },
    {
      "epoch": 0.39269253331098636,
      "grad_norm": 0.16225073568043916,
      "learning_rate": 2.331808026163472e-08,
      "loss": 0.8749,
      "step": 2343
    },
    {
      "epoch": 0.39286013575798207,
      "grad_norm": 0.19270041449726605,
      "learning_rate": 2.3248447758864366e-08,
      "loss": 0.9325,
      "step": 2344
    },
    {
      "epoch": 0.3930277382049778,
      "grad_norm": 0.1713395971861851,
      "learning_rate": 2.3178905702188124e-08,
      "loss": 0.86,
      "step": 2345
    },
    {
      "epoch": 0.39319534065197354,
      "grad_norm": 0.1684988154553444,
      "learning_rate": 2.3109454173556152e-08,
      "loss": 0.8726,
      "step": 2346
    },
    {
      "epoch": 0.39336294309896924,
      "grad_norm": 0.17352567636522673,
      "learning_rate": 2.304009325481212e-08,
      "loss": 0.8863,
      "step": 2347
    },
    {
      "epoch": 0.39353054554596495,
      "grad_norm": 0.13942097749158697,
      "learning_rate": 2.2970823027692754e-08,
      "loss": 0.821,
      "step": 2348
    },
    {
      "epoch": 0.3936981479929607,
      "grad_norm": 0.1778196976595698,
      "learning_rate": 2.290164357382791e-08,
      "loss": 0.9238,
      "step": 2349
    },
    {
      "epoch": 0.3938657504399564,
      "grad_norm": 0.1773780140671596,
      "learning_rate": 2.283255497474058e-08,
      "loss": 0.8432,
      "step": 2350
    },
    {
      "epoch": 0.3940333528869521,
      "grad_norm": 0.17505262647039765,
      "learning_rate": 2.276355731184658e-08,
      "loss": 0.9125,
      "step": 2351
    },
    {
      "epoch": 0.3942009553339479,
      "grad_norm": 0.1861124571602958,
      "learning_rate": 2.269465066645463e-08,
      "loss": 0.8937,
      "step": 2352
    },
    {
      "epoch": 0.3943685577809436,
      "grad_norm": 0.16182097694159722,
      "learning_rate": 2.26258351197662e-08,
      "loss": 0.8715,
      "step": 2353
    },
    {
      "epoch": 0.3945361602279393,
      "grad_norm": 0.17952624391656963,
      "learning_rate": 2.255711075287533e-08,
      "loss": 0.8852,
      "step": 2354
    },
    {
      "epoch": 0.39470376267493507,
      "grad_norm": 0.1748416862747054,
      "learning_rate": 2.2488477646768722e-08,
      "loss": 0.9066,
      "step": 2355
    },
    {
      "epoch": 0.3948713651219308,
      "grad_norm": 0.17499567307724948,
      "learning_rate": 2.2419935882325446e-08,
      "loss": 0.8868,
      "step": 2356
    },
    {
      "epoch": 0.3950389675689265,
      "grad_norm": 0.17799662539213398,
      "learning_rate": 2.235148554031694e-08,
      "loss": 0.8961,
      "step": 2357
    },
    {
      "epoch": 0.39520657001592224,
      "grad_norm": 0.17710521334591092,
      "learning_rate": 2.228312670140694e-08,
      "loss": 0.8856,
      "step": 2358
    },
    {
      "epoch": 0.39537417246291795,
      "grad_norm": 0.16902080286458035,
      "learning_rate": 2.221485944615137e-08,
      "loss": 0.8518,
      "step": 2359
    },
    {
      "epoch": 0.39554177490991366,
      "grad_norm": 0.18973345213948195,
      "learning_rate": 2.2146683854998162e-08,
      "loss": 0.9104,
      "step": 2360
    },
    {
      "epoch": 0.3957093773569094,
      "grad_norm": 0.16820192560140992,
      "learning_rate": 2.2078600008287308e-08,
      "loss": 0.8579,
      "step": 2361
    },
    {
      "epoch": 0.3958769798039051,
      "grad_norm": 0.1773054573579452,
      "learning_rate": 2.2010607986250596e-08,
      "loss": 0.8831,
      "step": 2362
    },
    {
      "epoch": 0.3960445822509009,
      "grad_norm": 0.16360070969807664,
      "learning_rate": 2.1942707869011678e-08,
      "loss": 0.8678,
      "step": 2363
    },
    {
      "epoch": 0.3962121846978966,
      "grad_norm": 0.17068175310418665,
      "learning_rate": 2.1874899736585916e-08,
      "loss": 0.8839,
      "step": 2364
    },
    {
      "epoch": 0.3963797871448923,
      "grad_norm": 0.17831978039572047,
      "learning_rate": 2.180718366888019e-08,
      "loss": 0.8853,
      "step": 2365
    },
    {
      "epoch": 0.39654738959188807,
      "grad_norm": 0.1773758038022879,
      "learning_rate": 2.1739559745692926e-08,
      "loss": 0.8829,
      "step": 2366
    },
    {
      "epoch": 0.3967149920388838,
      "grad_norm": 0.18958426621967878,
      "learning_rate": 2.1672028046713996e-08,
      "loss": 0.8832,
      "step": 2367
    },
    {
      "epoch": 0.3968825944858795,
      "grad_norm": 0.1613748238343166,
      "learning_rate": 2.1604588651524602e-08,
      "loss": 0.8371,
      "step": 2368
    },
    {
      "epoch": 0.39705019693287524,
      "grad_norm": 0.17633009167544691,
      "learning_rate": 2.153724163959708e-08,
      "loss": 0.8863,
      "step": 2369
    },
    {
      "epoch": 0.39721779937987095,
      "grad_norm": 0.18268215654367892,
      "learning_rate": 2.1469987090295028e-08,
      "loss": 0.8703,
      "step": 2370
    },
    {
      "epoch": 0.39738540182686666,
      "grad_norm": 0.16155031593500588,
      "learning_rate": 2.1402825082872966e-08,
      "loss": 0.8713,
      "step": 2371
    },
    {
      "epoch": 0.3975530042738624,
      "grad_norm": 0.19447407979074133,
      "learning_rate": 2.1335755696476376e-08,
      "loss": 0.8868,
      "step": 2372
    },
    {
      "epoch": 0.3977206067208581,
      "grad_norm": 0.18440969887263986,
      "learning_rate": 2.1268779010141736e-08,
      "loss": 0.9144,
      "step": 2373
    },
    {
      "epoch": 0.39788820916785383,
      "grad_norm": 0.15205898040393373,
      "learning_rate": 2.1201895102796107e-08,
      "loss": 0.8304,
      "step": 2374
    },
    {
      "epoch": 0.3980558116148496,
      "grad_norm": 0.1765435067894015,
      "learning_rate": 2.113510405325727e-08,
      "loss": 0.8653,
      "step": 2375
    },
    {
      "epoch": 0.3982234140618453,
      "grad_norm": 0.18186773647408772,
      "learning_rate": 2.1068405940233647e-08,
      "loss": 0.8913,
      "step": 2376
    },
    {
      "epoch": 0.398391016508841,
      "grad_norm": 0.1702476098729444,
      "learning_rate": 2.1001800842324024e-08,
      "loss": 0.8704,
      "step": 2377
    },
    {
      "epoch": 0.3985586189558368,
      "grad_norm": 0.1925681113440292,
      "learning_rate": 2.0935288838017685e-08,
      "loss": 0.9065,
      "step": 2378
    },
    {
      "epoch": 0.3987262214028325,
      "grad_norm": 0.17072058082821207,
      "learning_rate": 2.0868870005694174e-08,
      "loss": 0.894,
      "step": 2379
    },
    {
      "epoch": 0.3988938238498282,
      "grad_norm": 0.16192539319382748,
      "learning_rate": 2.080254442362319e-08,
      "loss": 0.9116,
      "step": 2380
    },
    {
      "epoch": 0.39906142629682395,
      "grad_norm": 0.14570051394828568,
      "learning_rate": 2.073631216996463e-08,
      "loss": 0.8461,
      "step": 2381
    },
    {
      "epoch": 0.39922902874381966,
      "grad_norm": 0.1807656249008454,
      "learning_rate": 2.0670173322768348e-08,
      "loss": 0.8901,
      "step": 2382
    },
    {
      "epoch": 0.39939663119081537,
      "grad_norm": 0.1669247549857809,
      "learning_rate": 2.0604127959974105e-08,
      "loss": 0.8832,
      "step": 2383
    },
    {
      "epoch": 0.39956423363781113,
      "grad_norm": 0.17708636358135363,
      "learning_rate": 2.0538176159411546e-08,
      "loss": 0.8839,
      "step": 2384
    },
    {
      "epoch": 0.39973183608480684,
      "grad_norm": 0.1777035213157198,
      "learning_rate": 2.0472317998800103e-08,
      "loss": 0.8935,
      "step": 2385
    },
    {
      "epoch": 0.39989943853180254,
      "grad_norm": 0.1668412847791677,
      "learning_rate": 2.0406553555748728e-08,
      "loss": 0.8715,
      "step": 2386
    },
    {
      "epoch": 0.4000670409787983,
      "grad_norm": 0.17284003272224532,
      "learning_rate": 2.0340882907756073e-08,
      "loss": 0.8947,
      "step": 2387
    },
    {
      "epoch": 0.400234643425794,
      "grad_norm": 0.1809516504353806,
      "learning_rate": 2.0275306132210146e-08,
      "loss": 0.8601,
      "step": 2388
    },
    {
      "epoch": 0.4004022458727897,
      "grad_norm": 0.192095807080101,
      "learning_rate": 2.0209823306388407e-08,
      "loss": 0.8786,
      "step": 2389
    },
    {
      "epoch": 0.4005698483197855,
      "grad_norm": 0.19058631530729028,
      "learning_rate": 2.014443450745761e-08,
      "loss": 0.9113,
      "step": 2390
    },
    {
      "epoch": 0.4007374507667812,
      "grad_norm": 0.16767141633969734,
      "learning_rate": 2.0079139812473634e-08,
      "loss": 0.8621,
      "step": 2391
    },
    {
      "epoch": 0.4009050532137769,
      "grad_norm": 0.15513373716232323,
      "learning_rate": 2.0013939298381487e-08,
      "loss": 0.8473,
      "step": 2392
    },
    {
      "epoch": 0.40107265566077266,
      "grad_norm": 0.17702976921536173,
      "learning_rate": 1.9948833042015268e-08,
      "loss": 0.8533,
      "step": 2393
    },
    {
      "epoch": 0.40124025810776837,
      "grad_norm": 0.19078688408342237,
      "learning_rate": 1.9883821120097877e-08,
      "loss": 0.9115,
      "step": 2394
    },
    {
      "epoch": 0.4014078605547641,
      "grad_norm": 0.15550706276144644,
      "learning_rate": 1.9818903609241155e-08,
      "loss": 0.8514,
      "step": 2395
    },
    {
      "epoch": 0.40157546300175984,
      "grad_norm": 0.16514863196571725,
      "learning_rate": 1.9754080585945644e-08,
      "loss": 0.8917,
      "step": 2396
    },
    {
      "epoch": 0.40174306544875554,
      "grad_norm": 0.17026880568218158,
      "learning_rate": 1.9689352126600534e-08,
      "loss": 0.8572,
      "step": 2397
    },
    {
      "epoch": 0.4019106678957513,
      "grad_norm": 0.17798451426053075,
      "learning_rate": 1.9624718307483556e-08,
      "loss": 0.8613,
      "step": 2398
    },
    {
      "epoch": 0.402078270342747,
      "grad_norm": 0.16916512349110455,
      "learning_rate": 1.9560179204760952e-08,
      "loss": 0.8995,
      "step": 2399
    },
    {
      "epoch": 0.4022458727897427,
      "grad_norm": 0.1696612501233044,
      "learning_rate": 1.9495734894487382e-08,
      "loss": 0.8956,
      "step": 2400
    },
    {
      "epoch": 0.4024134752367385,
      "grad_norm": 0.17611336805640904,
      "learning_rate": 1.9431385452605687e-08,
      "loss": 0.8712,
      "step": 2401
    },
    {
      "epoch": 0.4025810776837342,
      "grad_norm": 0.17618376836841457,
      "learning_rate": 1.936713095494703e-08,
      "loss": 0.905,
      "step": 2402
    },
    {
      "epoch": 0.4027486801307299,
      "grad_norm": 0.1612648719594171,
      "learning_rate": 1.9302971477230613e-08,
      "loss": 0.8277,
      "step": 2403
    },
    {
      "epoch": 0.40291628257772566,
      "grad_norm": 0.18224754878263677,
      "learning_rate": 1.9238907095063693e-08,
      "loss": 0.8875,
      "step": 2404
    },
    {
      "epoch": 0.40308388502472137,
      "grad_norm": 0.1733977712482349,
      "learning_rate": 1.91749378839415e-08,
      "loss": 0.8787,
      "step": 2405
    },
    {
      "epoch": 0.4032514874717171,
      "grad_norm": 0.16609210365102411,
      "learning_rate": 1.9111063919247004e-08,
      "loss": 0.9208,
      "step": 2406
    },
    {
      "epoch": 0.40341908991871284,
      "grad_norm": 0.17197178414964445,
      "learning_rate": 1.904728527625108e-08,
      "loss": 0.8634,
      "step": 2407
    },
    {
      "epoch": 0.40358669236570854,
      "grad_norm": 0.162575543328405,
      "learning_rate": 1.8983602030112156e-08,
      "loss": 0.8775,
      "step": 2408
    },
    {
      "epoch": 0.40375429481270425,
      "grad_norm": 0.16114767581764436,
      "learning_rate": 1.892001425587626e-08,
      "loss": 0.8914,
      "step": 2409
    },
    {
      "epoch": 0.4039218972597,
      "grad_norm": 0.1663773530331093,
      "learning_rate": 1.885652202847696e-08,
      "loss": 0.857,
      "step": 2410
    },
    {
      "epoch": 0.4040894997066957,
      "grad_norm": 0.19425422659899966,
      "learning_rate": 1.8793125422735233e-08,
      "loss": 0.9126,
      "step": 2411
    },
    {
      "epoch": 0.4042571021536914,
      "grad_norm": 0.17921561047741671,
      "learning_rate": 1.8729824513359293e-08,
      "loss": 0.8978,
      "step": 2412
    },
    {
      "epoch": 0.4044247046006872,
      "grad_norm": 0.1789593572733817,
      "learning_rate": 1.866661937494468e-08,
      "loss": 0.8967,
      "step": 2413
    },
    {
      "epoch": 0.4045923070476829,
      "grad_norm": 0.17809293670113346,
      "learning_rate": 1.8603510081974017e-08,
      "loss": 0.8902,
      "step": 2414
    },
    {
      "epoch": 0.4047599094946786,
      "grad_norm": 0.20077932718140845,
      "learning_rate": 1.8540496708816932e-08,
      "loss": 0.926,
      "step": 2415
    },
    {
      "epoch": 0.40492751194167437,
      "grad_norm": 0.17919456524686425,
      "learning_rate": 1.8477579329730186e-08,
      "loss": 0.8949,
      "step": 2416
    },
    {
      "epoch": 0.4050951143886701,
      "grad_norm": 0.1715429538055682,
      "learning_rate": 1.8414758018857245e-08,
      "loss": 0.8629,
      "step": 2417
    },
    {
      "epoch": 0.4052627168356658,
      "grad_norm": 0.18179834666414144,
      "learning_rate": 1.8352032850228437e-08,
      "loss": 0.9029,
      "step": 2418
    },
    {
      "epoch": 0.40543031928266154,
      "grad_norm": 0.17250477668381164,
      "learning_rate": 1.8289403897760813e-08,
      "loss": 0.8885,
      "step": 2419
    },
    {
      "epoch": 0.40559792172965725,
      "grad_norm": 0.15693142080059974,
      "learning_rate": 1.822687123525797e-08,
      "loss": 0.8386,
      "step": 2420
    },
    {
      "epoch": 0.40576552417665296,
      "grad_norm": 0.17030127664006675,
      "learning_rate": 1.8164434936410113e-08,
      "loss": 0.8827,
      "step": 2421
    },
    {
      "epoch": 0.4059331266236487,
      "grad_norm": 0.17639403062031833,
      "learning_rate": 1.810209507479388e-08,
      "loss": 0.8738,
      "step": 2422
    },
    {
      "epoch": 0.40610072907064443,
      "grad_norm": 0.1891190771515339,
      "learning_rate": 1.8039851723872203e-08,
      "loss": 0.9137,
      "step": 2423
    },
    {
      "epoch": 0.40626833151764014,
      "grad_norm": 0.14513961085721708,
      "learning_rate": 1.797770495699431e-08,
      "loss": 0.8303,
      "step": 2424
    },
    {
      "epoch": 0.4064359339646359,
      "grad_norm": 0.18880835996028397,
      "learning_rate": 1.7915654847395645e-08,
      "loss": 0.8937,
      "step": 2425
    },
    {
      "epoch": 0.4066035364116316,
      "grad_norm": 0.18580677840435625,
      "learning_rate": 1.7853701468197746e-08,
      "loss": 0.8926,
      "step": 2426
    },
    {
      "epoch": 0.4067711388586273,
      "grad_norm": 0.18881768736095697,
      "learning_rate": 1.7791844892408103e-08,
      "loss": 0.8799,
      "step": 2427
    },
    {
      "epoch": 0.4069387413056231,
      "grad_norm": 0.19625159605226203,
      "learning_rate": 1.773008519292022e-08,
      "loss": 0.918,
      "step": 2428
    },
    {
      "epoch": 0.4071063437526188,
      "grad_norm": 0.17595106176317032,
      "learning_rate": 1.766842244251332e-08,
      "loss": 0.8903,
      "step": 2429
    },
    {
      "epoch": 0.4072739461996145,
      "grad_norm": 0.19440549568993817,
      "learning_rate": 1.7606856713852482e-08,
      "loss": 0.89,
      "step": 2430
    },
    {
      "epoch": 0.40744154864661025,
      "grad_norm": 0.1484960039055631,
      "learning_rate": 1.7545388079488454e-08,
      "loss": 0.8459,
      "step": 2431
    },
    {
      "epoch": 0.40760915109360596,
      "grad_norm": 0.16651552862803295,
      "learning_rate": 1.7484016611857454e-08,
      "loss": 0.8879,
      "step": 2432
    },
    {
      "epoch": 0.40777675354060167,
      "grad_norm": 0.17968234368662564,
      "learning_rate": 1.7422742383281354e-08,
      "loss": 0.8533,
      "step": 2433
    },
    {
      "epoch": 0.40794435598759743,
      "grad_norm": 0.17933316961502965,
      "learning_rate": 1.7361565465967287e-08,
      "loss": 0.8912,
      "step": 2434
    },
    {
      "epoch": 0.40811195843459314,
      "grad_norm": 0.19348365082370753,
      "learning_rate": 1.730048593200779e-08,
      "loss": 0.9232,
      "step": 2435
    },
    {
      "epoch": 0.4082795608815889,
      "grad_norm": 0.18005267643357045,
      "learning_rate": 1.7239503853380633e-08,
      "loss": 0.9012,
      "step": 2436
    },
    {
      "epoch": 0.4084471633285846,
      "grad_norm": 0.17107723772971897,
      "learning_rate": 1.7178619301948772e-08,
      "loss": 0.8911,
      "step": 2437
    },
    {
      "epoch": 0.4086147657755803,
      "grad_norm": 0.19633983219168177,
      "learning_rate": 1.711783234946015e-08,
      "loss": 0.9393,
      "step": 2438
    },
    {
      "epoch": 0.4087823682225761,
      "grad_norm": 0.17857470729778172,
      "learning_rate": 1.7057143067547785e-08,
      "loss": 0.865,
      "step": 2439
    },
    {
      "epoch": 0.4089499706695718,
      "grad_norm": 0.16506886912185845,
      "learning_rate": 1.6996551527729563e-08,
      "loss": 0.8998,
      "step": 2440
    },
    {
      "epoch": 0.4091175731165675,
      "grad_norm": 0.17371339961536988,
      "learning_rate": 1.69360578014081e-08,
      "loss": 0.8827,
      "step": 2441
    },
    {
      "epoch": 0.40928517556356325,
      "grad_norm": 0.19397659557387073,
      "learning_rate": 1.6875661959870957e-08,
      "loss": 0.9096,
      "step": 2442
    },
    {
      "epoch": 0.40945277801055896,
      "grad_norm": 0.19374926413862775,
      "learning_rate": 1.6815364074290152e-08,
      "loss": 0.8787,
      "step": 2443
    },
    {
      "epoch": 0.40962038045755467,
      "grad_norm": 0.17691743848054642,
      "learning_rate": 1.6755164215722317e-08,
      "loss": 0.8931,
      "step": 2444
    },
    {
      "epoch": 0.40978798290455043,
      "grad_norm": 0.1776583768383468,
      "learning_rate": 1.6695062455108643e-08,
      "loss": 0.9143,
      "step": 2445
    },
    {
      "epoch": 0.40995558535154614,
      "grad_norm": 0.17942386156874982,
      "learning_rate": 1.6635058863274597e-08,
      "loss": 0.892,
      "step": 2446
    },
    {
      "epoch": 0.41012318779854184,
      "grad_norm": 0.15754803471651954,
      "learning_rate": 1.6575153510930062e-08,
      "loss": 0.8809,
      "step": 2447
    },
    {
      "epoch": 0.4102907902455376,
      "grad_norm": 0.17890040287059208,
      "learning_rate": 1.6515346468669134e-08,
      "loss": 0.896,
      "step": 2448
    },
    {
      "epoch": 0.4104583926925333,
      "grad_norm": 0.1902690645073458,
      "learning_rate": 1.6455637806970025e-08,
      "loss": 0.9364,
      "step": 2449
    },
    {
      "epoch": 0.410625995139529,
      "grad_norm": 0.1837993991573053,
      "learning_rate": 1.6396027596194995e-08,
      "loss": 0.8885,
      "step": 2450
    },
    {
      "epoch": 0.4107935975865248,
      "grad_norm": 0.17566598879076376,
      "learning_rate": 1.6336515906590352e-08,
      "loss": 0.8946,
      "step": 2451
    },
    {
      "epoch": 0.4109612000335205,
      "grad_norm": 0.16627418173235797,
      "learning_rate": 1.6277102808286292e-08,
      "loss": 0.8827,
      "step": 2452
    },
    {
      "epoch": 0.4111288024805162,
      "grad_norm": 0.18003033098564733,
      "learning_rate": 1.621778837129676e-08,
      "loss": 0.8543,
      "step": 2453
    },
    {
      "epoch": 0.41129640492751196,
      "grad_norm": 0.19056442503901797,
      "learning_rate": 1.6158572665519523e-08,
      "loss": 0.8746,
      "step": 2454
    },
    {
      "epoch": 0.41146400737450767,
      "grad_norm": 0.16554125801402822,
      "learning_rate": 1.6099455760735937e-08,
      "loss": 0.8415,
      "step": 2455
    },
    {
      "epoch": 0.4116316098215034,
      "grad_norm": 0.17018317663286334,
      "learning_rate": 1.6040437726610978e-08,
      "loss": 0.8636,
      "step": 2456
    },
    {
      "epoch": 0.41179921226849914,
      "grad_norm": 0.16351136136844868,
      "learning_rate": 1.598151863269306e-08,
      "loss": 0.851,
      "step": 2457
    },
    {
      "epoch": 0.41196681471549484,
      "grad_norm": 0.14933607299128424,
      "learning_rate": 1.5922698548414026e-08,
      "loss": 0.855,
      "step": 2458
    },
    {
      "epoch": 0.41213441716249055,
      "grad_norm": 0.1938203616314078,
      "learning_rate": 1.5863977543089092e-08,
      "loss": 0.8978,
      "step": 2459
    },
    {
      "epoch": 0.4123020196094863,
      "grad_norm": 0.1638512750690715,
      "learning_rate": 1.5805355685916632e-08,
      "loss": 0.8601,
      "step": 2460
    },
    {
      "epoch": 0.412469622056482,
      "grad_norm": 0.1874562842543399,
      "learning_rate": 1.5746833045978214e-08,
      "loss": 0.8599,
      "step": 2461
    },
    {
      "epoch": 0.41263722450347773,
      "grad_norm": 0.16226272221328253,
      "learning_rate": 1.5688409692238493e-08,
      "loss": 0.8738,
      "step": 2462
    },
    {
      "epoch": 0.4128048269504735,
      "grad_norm": 0.17183792263178474,
      "learning_rate": 1.563008569354516e-08,
      "loss": 0.8587,
      "step": 2463
    },
    {
      "epoch": 0.4129724293974692,
      "grad_norm": 0.17456990269281344,
      "learning_rate": 1.5571861118628725e-08,
      "loss": 0.8371,
      "step": 2464
    },
    {
      "epoch": 0.4131400318444649,
      "grad_norm": 0.1808853137762258,
      "learning_rate": 1.5513736036102643e-08,
      "loss": 0.8958,
      "step": 2465
    },
    {
      "epoch": 0.41330763429146067,
      "grad_norm": 0.16627754378388687,
      "learning_rate": 1.5455710514463073e-08,
      "loss": 0.8821,
      "step": 2466
    },
    {
      "epoch": 0.4134752367384564,
      "grad_norm": 0.19301224841872203,
      "learning_rate": 1.539778462208877e-08,
      "loss": 0.8916,
      "step": 2467
    },
    {
      "epoch": 0.4136428391854521,
      "grad_norm": 0.17965679298638096,
      "learning_rate": 1.5339958427241285e-08,
      "loss": 0.853,
      "step": 2468
    },
    {
      "epoch": 0.41381044163244785,
      "grad_norm": 0.18681431499478604,
      "learning_rate": 1.52822319980645e-08,
      "loss": 0.8815,
      "step": 2469
    },
    {
      "epoch": 0.41397804407944355,
      "grad_norm": 0.2055187129340397,
      "learning_rate": 1.522460540258478e-08,
      "loss": 0.925,
      "step": 2470
    },
    {
      "epoch": 0.41414564652643926,
      "grad_norm": 0.17223345271416685,
      "learning_rate": 1.516707870871089e-08,
      "loss": 0.8834,
      "step": 2471
    },
    {
      "epoch": 0.414313248973435,
      "grad_norm": 0.15666229815456836,
      "learning_rate": 1.5109651984233794e-08,
      "loss": 0.8292,
      "step": 2472
    },
    {
      "epoch": 0.41448085142043073,
      "grad_norm": 0.17838654953023284,
      "learning_rate": 1.5052325296826708e-08,
      "loss": 0.8869,
      "step": 2473
    },
    {
      "epoch": 0.4146484538674265,
      "grad_norm": 0.18454300002556648,
      "learning_rate": 1.4995098714044974e-08,
      "loss": 0.9224,
      "step": 2474
    },
    {
      "epoch": 0.4148160563144222,
      "grad_norm": 0.1726034907291927,
      "learning_rate": 1.493797230332591e-08,
      "loss": 0.8759,
      "step": 2475
    },
    {
      "epoch": 0.4149836587614179,
      "grad_norm": 0.16473877613987276,
      "learning_rate": 1.488094613198878e-08,
      "loss": 0.874,
      "step": 2476
    },
    {
      "epoch": 0.41515126120841367,
      "grad_norm": 0.15181224936629248,
      "learning_rate": 1.4824020267234827e-08,
      "loss": 0.8446,
      "step": 2477
    },
    {
      "epoch": 0.4153188636554094,
      "grad_norm": 0.16769114302452975,
      "learning_rate": 1.4767194776146963e-08,
      "loss": 0.8676,
      "step": 2478
    },
    {
      "epoch": 0.4154864661024051,
      "grad_norm": 0.17602661563837535,
      "learning_rate": 1.4710469725689889e-08,
      "loss": 0.8705,
      "step": 2479
    },
    {
      "epoch": 0.41565406854940085,
      "grad_norm": 0.17613958373428698,
      "learning_rate": 1.4653845182709978e-08,
      "loss": 0.8632,
      "step": 2480
    },
    {
      "epoch": 0.41582167099639655,
      "grad_norm": 0.1860846536832876,
      "learning_rate": 1.4597321213935043e-08,
      "loss": 0.9094,
      "step": 2481
    },
    {
      "epoch": 0.41598927344339226,
      "grad_norm": 0.1487199045955726,
      "learning_rate": 1.4540897885974524e-08,
      "loss": 0.8084,
      "step": 2482
    },
    {
      "epoch": 0.416156875890388,
      "grad_norm": 0.18279022625848668,
      "learning_rate": 1.448457526531911e-08,
      "loss": 0.8751,
      "step": 2483
    },
    {
      "epoch": 0.41632447833738373,
      "grad_norm": 0.1603792346112557,
      "learning_rate": 1.4428353418340967e-08,
      "loss": 0.8851,
      "step": 2484
    },
    {
      "epoch": 0.41649208078437944,
      "grad_norm": 0.18233590171912092,
      "learning_rate": 1.4372232411293372e-08,
      "loss": 0.8914,
      "step": 2485
    },
    {
      "epoch": 0.4166596832313752,
      "grad_norm": 0.1583026380662961,
      "learning_rate": 1.4316212310310871e-08,
      "loss": 0.8764,
      "step": 2486
    },
    {
      "epoch": 0.4168272856783709,
      "grad_norm": 0.17509149381024866,
      "learning_rate": 1.4260293181409022e-08,
      "loss": 0.8839,
      "step": 2487
    },
    {
      "epoch": 0.4169948881253666,
      "grad_norm": 0.18436525350513933,
      "learning_rate": 1.4204475090484459e-08,
      "loss": 0.8794,
      "step": 2488
    },
    {
      "epoch": 0.4171624905723624,
      "grad_norm": 0.16504823605175215,
      "learning_rate": 1.4148758103314728e-08,
      "loss": 0.9197,
      "step": 2489
    },
    {
      "epoch": 0.4173300930193581,
      "grad_norm": 0.1878257367371666,
      "learning_rate": 1.409314228555819e-08,
      "loss": 0.9049,
      "step": 2490
    },
    {
      "epoch": 0.4174976954663538,
      "grad_norm": 0.16117033877792467,
      "learning_rate": 1.4037627702754062e-08,
      "loss": 0.8541,
      "step": 2491
    },
    {
      "epoch": 0.41766529791334955,
      "grad_norm": 0.18981623046239873,
      "learning_rate": 1.3982214420322203e-08,
      "loss": 0.8932,
      "step": 2492
    },
    {
      "epoch": 0.41783290036034526,
      "grad_norm": 0.18078124273875543,
      "learning_rate": 1.3926902503563099e-08,
      "loss": 0.8834,
      "step": 2493
    },
    {
      "epoch": 0.41800050280734097,
      "grad_norm": 0.15933313433763646,
      "learning_rate": 1.387169201765781e-08,
      "loss": 0.8182,
      "step": 2494
    },
    {
      "epoch": 0.41816810525433673,
      "grad_norm": 0.15308282256150696,
      "learning_rate": 1.3816583027667894e-08,
      "loss": 0.8364,
      "step": 2495
    },
    {
      "epoch": 0.41833570770133244,
      "grad_norm": 0.17921843164279433,
      "learning_rate": 1.3761575598535213e-08,
      "loss": 0.8565,
      "step": 2496
    },
    {
      "epoch": 0.41850331014832814,
      "grad_norm": 0.16680874662079387,
      "learning_rate": 1.370666979508206e-08,
      "loss": 0.8932,
      "step": 2497
    },
    {
      "epoch": 0.4186709125953239,
      "grad_norm": 0.18671579318881382,
      "learning_rate": 1.3651865682010865e-08,
      "loss": 0.8723,
      "step": 2498
    },
    {
      "epoch": 0.4188385150423196,
      "grad_norm": 0.16791780866114944,
      "learning_rate": 1.3597163323904238e-08,
      "loss": 0.8901,
      "step": 2499
    },
    {
      "epoch": 0.4190061174893153,
      "grad_norm": 0.16116774470939202,
      "learning_rate": 1.354256278522501e-08,
      "loss": 0.8868,
      "step": 2500
    },
    {
      "epoch": 0.4191737199363111,
      "grad_norm": 0.17231392069041632,
      "learning_rate": 1.3488064130315857e-08,
      "loss": 0.9051,
      "step": 2501
    },
    {
      "epoch": 0.4193413223833068,
      "grad_norm": 0.1816277874100691,
      "learning_rate": 1.3433667423399453e-08,
      "loss": 0.8854,
      "step": 2502
    },
    {
      "epoch": 0.4195089248303025,
      "grad_norm": 0.1580605484308184,
      "learning_rate": 1.3379372728578387e-08,
      "loss": 0.8363,
      "step": 2503
    },
    {
      "epoch": 0.41967652727729826,
      "grad_norm": 0.17006693484114116,
      "learning_rate": 1.3325180109834932e-08,
      "loss": 0.8833,
      "step": 2504
    },
    {
      "epoch": 0.41984412972429397,
      "grad_norm": 0.16545786913417068,
      "learning_rate": 1.3271089631031152e-08,
      "loss": 0.8797,
      "step": 2505
    },
    {
      "epoch": 0.4200117321712897,
      "grad_norm": 0.16997748964819398,
      "learning_rate": 1.3217101355908755e-08,
      "loss": 0.8819,
      "step": 2506
    },
    {
      "epoch": 0.42017933461828544,
      "grad_norm": 0.17500682749923463,
      "learning_rate": 1.316321534808893e-08,
      "loss": 0.8705,
      "step": 2507
    },
    {
      "epoch": 0.42034693706528115,
      "grad_norm": 0.16503785167821663,
      "learning_rate": 1.3109431671072446e-08,
      "loss": 0.8858,
      "step": 2508
    },
    {
      "epoch": 0.42051453951227685,
      "grad_norm": 0.17853572974173554,
      "learning_rate": 1.3055750388239372e-08,
      "loss": 0.8551,
      "step": 2509
    },
    {
      "epoch": 0.4206821419592726,
      "grad_norm": 0.1751072294737565,
      "learning_rate": 1.300217156284924e-08,
      "loss": 0.8982,
      "step": 2510
    },
    {
      "epoch": 0.4208497444062683,
      "grad_norm": 0.18557579890964712,
      "learning_rate": 1.2948695258040732e-08,
      "loss": 0.8893,
      "step": 2511
    },
    {
      "epoch": 0.4210173468532641,
      "grad_norm": 0.18733060499755985,
      "learning_rate": 1.2895321536831794e-08,
      "loss": 0.9082,
      "step": 2512
    },
    {
      "epoch": 0.4211849493002598,
      "grad_norm": 0.1461365874329596,
      "learning_rate": 1.2842050462119424e-08,
      "loss": 0.7994,
      "step": 2513
    },
    {
      "epoch": 0.4213525517472555,
      "grad_norm": 0.16318397479053526,
      "learning_rate": 1.2788882096679732e-08,
      "loss": 0.8548,
      "step": 2514
    },
    {
      "epoch": 0.42152015419425126,
      "grad_norm": 0.17625595410784234,
      "learning_rate": 1.2735816503167706e-08,
      "loss": 0.8693,
      "step": 2515
    },
    {
      "epoch": 0.42168775664124697,
      "grad_norm": 0.17247249191019562,
      "learning_rate": 1.2682853744117294e-08,
      "loss": 0.8616,
      "step": 2516
    },
    {
      "epoch": 0.4218553590882427,
      "grad_norm": 0.1704910688135697,
      "learning_rate": 1.2629993881941247e-08,
      "loss": 0.9198,
      "step": 2517
    },
    {
      "epoch": 0.42202296153523844,
      "grad_norm": 0.18730524474349386,
      "learning_rate": 1.2577236978931038e-08,
      "loss": 0.8526,
      "step": 2518
    },
    {
      "epoch": 0.42219056398223415,
      "grad_norm": 0.17442054653954708,
      "learning_rate": 1.2524583097256792e-08,
      "loss": 0.8908,
      "step": 2519
    },
    {
      "epoch": 0.42235816642922985,
      "grad_norm": 0.18324482556561317,
      "learning_rate": 1.2472032298967294e-08,
      "loss": 0.8853,
      "step": 2520
    },
    {
      "epoch": 0.4225257688762256,
      "grad_norm": 0.16896411999852964,
      "learning_rate": 1.2419584645989822e-08,
      "loss": 0.863,
      "step": 2521
    },
    {
      "epoch": 0.4226933713232213,
      "grad_norm": 0.16686256182739548,
      "learning_rate": 1.2367240200130069e-08,
      "loss": 0.8337,
      "step": 2522
    },
    {
      "epoch": 0.42286097377021703,
      "grad_norm": 0.16505926389131662,
      "learning_rate": 1.2314999023072181e-08,
      "loss": 0.8648,
      "step": 2523
    },
    {
      "epoch": 0.4230285762172128,
      "grad_norm": 0.17409568520040306,
      "learning_rate": 1.2262861176378546e-08,
      "loss": 0.8961,
      "step": 2524
    },
    {
      "epoch": 0.4231961786642085,
      "grad_norm": 0.17875806158570556,
      "learning_rate": 1.2210826721489764e-08,
      "loss": 0.8692,
      "step": 2525
    },
    {
      "epoch": 0.4233637811112042,
      "grad_norm": 0.16494301737325157,
      "learning_rate": 1.2158895719724704e-08,
      "loss": 0.8807,
      "step": 2526
    },
    {
      "epoch": 0.42353138355819997,
      "grad_norm": 0.16045150381422454,
      "learning_rate": 1.2107068232280238e-08,
      "loss": 0.8565,
      "step": 2527
    },
    {
      "epoch": 0.4236989860051957,
      "grad_norm": 0.1937176981729721,
      "learning_rate": 1.2055344320231243e-08,
      "loss": 0.9492,
      "step": 2528
    },
    {
      "epoch": 0.4238665884521914,
      "grad_norm": 0.16813495230591455,
      "learning_rate": 1.2003724044530594e-08,
      "loss": 0.8823,
      "step": 2529
    },
    {
      "epoch": 0.42403419089918715,
      "grad_norm": 0.17695318507547453,
      "learning_rate": 1.1952207466009001e-08,
      "loss": 0.8633,
      "step": 2530
    },
    {
      "epoch": 0.42420179334618285,
      "grad_norm": 0.16821340354568454,
      "learning_rate": 1.1900794645374978e-08,
      "loss": 0.8762,
      "step": 2531
    },
    {
      "epoch": 0.42436939579317856,
      "grad_norm": 0.18136565693757853,
      "learning_rate": 1.1849485643214807e-08,
      "loss": 0.8912,
      "step": 2532
    },
    {
      "epoch": 0.4245369982401743,
      "grad_norm": 0.1946775784252961,
      "learning_rate": 1.1798280519992366e-08,
      "loss": 0.9221,
      "step": 2533
    },
    {
      "epoch": 0.42470460068717003,
      "grad_norm": 0.18907521845123584,
      "learning_rate": 1.1747179336049185e-08,
      "loss": 0.9044,
      "step": 2534
    },
    {
      "epoch": 0.42487220313416574,
      "grad_norm": 0.18343953893156797,
      "learning_rate": 1.169618215160424e-08,
      "loss": 0.887,
      "step": 2535
    },
    {
      "epoch": 0.4250398055811615,
      "grad_norm": 0.1527388989673738,
      "learning_rate": 1.164528902675398e-08,
      "loss": 0.8193,
      "step": 2536
    },
    {
      "epoch": 0.4252074080281572,
      "grad_norm": 0.15435802865681084,
      "learning_rate": 1.1594500021472242e-08,
      "loss": 0.831,
      "step": 2537
    },
    {
      "epoch": 0.4253750104751529,
      "grad_norm": 0.17162298939291,
      "learning_rate": 1.1543815195610195e-08,
      "loss": 0.8642,
      "step": 2538
    },
    {
      "epoch": 0.4255426129221487,
      "grad_norm": 0.18419251661970804,
      "learning_rate": 1.149323460889614e-08,
      "loss": 0.8981,
      "step": 2539
    },
    {
      "epoch": 0.4257102153691444,
      "grad_norm": 0.18195657702503698,
      "learning_rate": 1.1442758320935653e-08,
      "loss": 0.8955,
      "step": 2540
    },
    {
      "epoch": 0.4258778178161401,
      "grad_norm": 0.17083920058566232,
      "learning_rate": 1.1392386391211306e-08,
      "loss": 0.8837,
      "step": 2541
    },
    {
      "epoch": 0.42604542026313585,
      "grad_norm": 0.20258579067061333,
      "learning_rate": 1.1342118879082763e-08,
      "loss": 0.8859,
      "step": 2542
    },
    {
      "epoch": 0.42621302271013156,
      "grad_norm": 0.1828986132066237,
      "learning_rate": 1.1291955843786616e-08,
      "loss": 0.9037,
      "step": 2543
    },
    {
      "epoch": 0.42638062515712727,
      "grad_norm": 0.17677821866655502,
      "learning_rate": 1.1241897344436335e-08,
      "loss": 0.8849,
      "step": 2544
    },
    {
      "epoch": 0.42654822760412303,
      "grad_norm": 0.173574033396664,
      "learning_rate": 1.1191943440022178e-08,
      "loss": 0.8977,
      "step": 2545
    },
    {
      "epoch": 0.42671583005111874,
      "grad_norm": 0.17609363474682685,
      "learning_rate": 1.1142094189411178e-08,
      "loss": 0.8725,
      "step": 2546
    },
    {
      "epoch": 0.42688343249811445,
      "grad_norm": 0.18404148388033698,
      "learning_rate": 1.1092349651347054e-08,
      "loss": 0.8406,
      "step": 2547
    },
    {
      "epoch": 0.4270510349451102,
      "grad_norm": 0.15443912733095186,
      "learning_rate": 1.1042709884450085e-08,
      "loss": 0.869,
      "step": 2548
    },
    {
      "epoch": 0.4272186373921059,
      "grad_norm": 0.16382660112598577,
      "learning_rate": 1.0993174947217143e-08,
      "loss": 0.8268,
      "step": 2549
    },
    {
      "epoch": 0.4273862398391017,
      "grad_norm": 0.17447302515234023,
      "learning_rate": 1.0943744898021501e-08,
      "loss": 0.8638,
      "step": 2550
    },
    {
      "epoch": 0.4275538422860974,
      "grad_norm": 0.17653119891134836,
      "learning_rate": 1.089441979511282e-08,
      "loss": 0.8563,
      "step": 2551
    },
    {
      "epoch": 0.4277214447330931,
      "grad_norm": 0.1705472856539476,
      "learning_rate": 1.0845199696617202e-08,
      "loss": 0.8631,
      "step": 2552
    },
    {
      "epoch": 0.42788904718008886,
      "grad_norm": 0.15951026565978424,
      "learning_rate": 1.0796084660536919e-08,
      "loss": 0.86,
      "step": 2553
    },
    {
      "epoch": 0.42805664962708456,
      "grad_norm": 0.1510007910529139,
      "learning_rate": 1.0747074744750417e-08,
      "loss": 0.8513,
      "step": 2554
    },
    {
      "epoch": 0.42822425207408027,
      "grad_norm": 0.15795398658448057,
      "learning_rate": 1.069817000701233e-08,
      "loss": 0.8572,
      "step": 2555
    },
    {
      "epoch": 0.42839185452107603,
      "grad_norm": 0.1695415379855283,
      "learning_rate": 1.0649370504953292e-08,
      "loss": 0.8555,
      "step": 2556
    },
    {
      "epoch": 0.42855945696807174,
      "grad_norm": 0.18218008485580325,
      "learning_rate": 1.0600676296079957e-08,
      "loss": 0.8883,
      "step": 2557
    },
    {
      "epoch": 0.42872705941506745,
      "grad_norm": 0.18133164129679055,
      "learning_rate": 1.0552087437774925e-08,
      "loss": 0.926,
      "step": 2558
    },
    {
      "epoch": 0.4288946618620632,
      "grad_norm": 0.1814474210844024,
      "learning_rate": 1.0503603987296561e-08,
      "loss": 0.8358,
      "step": 2559
    },
    {
      "epoch": 0.4290622643090589,
      "grad_norm": 0.1706629971215161,
      "learning_rate": 1.0455226001779116e-08,
      "loss": 0.8661,
      "step": 2560
    },
    {
      "epoch": 0.4292298667560546,
      "grad_norm": 0.17594608431774705,
      "learning_rate": 1.0406953538232477e-08,
      "loss": 0.8762,
      "step": 2561
    },
    {
      "epoch": 0.4293974692030504,
      "grad_norm": 0.1759562700943578,
      "learning_rate": 1.0358786653542207e-08,
      "loss": 0.8986,
      "step": 2562
    },
    {
      "epoch": 0.4295650716500461,
      "grad_norm": 0.16867882986880173,
      "learning_rate": 1.0310725404469478e-08,
      "loss": 0.8562,
      "step": 2563
    },
    {
      "epoch": 0.4297326740970418,
      "grad_norm": 0.1896981334515377,
      "learning_rate": 1.0262769847650977e-08,
      "loss": 0.8761,
      "step": 2564
    },
    {
      "epoch": 0.42990027654403756,
      "grad_norm": 0.1730481666215826,
      "learning_rate": 1.0214920039598774e-08,
      "loss": 0.9034,
      "step": 2565
    },
    {
      "epoch": 0.43006787899103327,
      "grad_norm": 0.1741780211921027,
      "learning_rate": 1.016717603670042e-08,
      "loss": 0.8646,
      "step": 2566
    },
    {
      "epoch": 0.430235481438029,
      "grad_norm": 0.17632538726959646,
      "learning_rate": 1.0119537895218688e-08,
      "loss": 0.8797,
      "step": 2567
    },
    {
      "epoch": 0.43040308388502474,
      "grad_norm": 0.18344133833375967,
      "learning_rate": 1.0072005671291672e-08,
      "loss": 0.8412,
      "step": 2568
    },
    {
      "epoch": 0.43057068633202045,
      "grad_norm": 0.20370322654545928,
      "learning_rate": 1.0024579420932633e-08,
      "loss": 0.8884,
      "step": 2569
    },
    {
      "epoch": 0.43073828877901615,
      "grad_norm": 0.1752323081901467,
      "learning_rate": 9.977259200029931e-09,
      "loss": 0.8986,
      "step": 2570
    },
    {
      "epoch": 0.4309058912260119,
      "grad_norm": 0.1605460793928903,
      "learning_rate": 9.930045064346959e-09,
      "loss": 0.8329,
      "step": 2571
    },
    {
      "epoch": 0.4310734936730076,
      "grad_norm": 0.20039570589816988,
      "learning_rate": 9.882937069522146e-09,
      "loss": 0.8935,
      "step": 2572
    },
    {
      "epoch": 0.43124109612000333,
      "grad_norm": 0.19179612196320656,
      "learning_rate": 9.83593527106884e-09,
      "loss": 0.9401,
      "step": 2573
    },
    {
      "epoch": 0.4314086985669991,
      "grad_norm": 0.19653995545855715,
      "learning_rate": 9.789039724375191e-09,
      "loss": 0.8889,
      "step": 2574
    },
    {
      "epoch": 0.4315763010139948,
      "grad_norm": 0.17828326080476414,
      "learning_rate": 9.742250484704217e-09,
      "loss": 0.8781,
      "step": 2575
    },
    {
      "epoch": 0.4317439034609905,
      "grad_norm": 0.18437302073724618,
      "learning_rate": 9.695567607193577e-09,
      "loss": 0.8885,
      "step": 2576
    },
    {
      "epoch": 0.43191150590798627,
      "grad_norm": 0.1904386613209284,
      "learning_rate": 9.648991146855634e-09,
      "loss": 0.882,
      "step": 2577
    },
    {
      "epoch": 0.432079108354982,
      "grad_norm": 0.15818952116137766,
      "learning_rate": 9.602521158577337e-09,
      "loss": 0.8523,
      "step": 2578
    },
    {
      "epoch": 0.4322467108019777,
      "grad_norm": 0.17464445564262282,
      "learning_rate": 9.556157697120216e-09,
      "loss": 0.8933,
      "step": 2579
    },
    {
      "epoch": 0.43241431324897345,
      "grad_norm": 0.16261254497202832,
      "learning_rate": 9.509900817120154e-09,
      "loss": 0.8496,
      "step": 2580
    },
    {
      "epoch": 0.43258191569596915,
      "grad_norm": 0.1909485112146064,
      "learning_rate": 9.46375057308757e-09,
      "loss": 0.8515,
      "step": 2581
    },
    {
      "epoch": 0.43274951814296486,
      "grad_norm": 0.16003865574612072,
      "learning_rate": 9.417707019407094e-09,
      "loss": 0.8383,
      "step": 2582
    },
    {
      "epoch": 0.4329171205899606,
      "grad_norm": 0.17562741575168106,
      "learning_rate": 9.371770210337727e-09,
      "loss": 0.8798,
      "step": 2583
    },
    {
      "epoch": 0.43308472303695633,
      "grad_norm": 0.16727353994189692,
      "learning_rate": 9.325940200012639e-09,
      "loss": 0.8649,
      "step": 2584
    },
    {
      "epoch": 0.43325232548395204,
      "grad_norm": 0.1764277260181294,
      "learning_rate": 9.28021704243912e-09,
      "loss": 0.8389,
      "step": 2585
    },
    {
      "epoch": 0.4334199279309478,
      "grad_norm": 0.16642350177195372,
      "learning_rate": 9.234600791498603e-09,
      "loss": 0.832,
      "step": 2586
    },
    {
      "epoch": 0.4335875303779435,
      "grad_norm": 0.17590247075021212,
      "learning_rate": 9.189091500946488e-09,
      "loss": 0.9252,
      "step": 2587
    },
    {
      "epoch": 0.43375513282493927,
      "grad_norm": 0.1755088179094247,
      "learning_rate": 9.143689224412132e-09,
      "loss": 0.8665,
      "step": 2588
    },
    {
      "epoch": 0.433922735271935,
      "grad_norm": 0.157417825483718,
      "learning_rate": 9.098394015398814e-09,
      "loss": 0.8319,
      "step": 2589
    },
    {
      "epoch": 0.4340903377189307,
      "grad_norm": 0.16569634408268238,
      "learning_rate": 9.053205927283647e-09,
      "loss": 0.8725,
      "step": 2590
    },
    {
      "epoch": 0.43425794016592645,
      "grad_norm": 0.1791248601076958,
      "learning_rate": 9.008125013317446e-09,
      "loss": 0.8948,
      "step": 2591
    },
    {
      "epoch": 0.43442554261292216,
      "grad_norm": 0.20882721313193112,
      "learning_rate": 8.963151326624829e-09,
      "loss": 0.9226,
      "step": 2592
    },
    {
      "epoch": 0.43459314505991786,
      "grad_norm": 0.1791136448016004,
      "learning_rate": 8.918284920203933e-09,
      "loss": 0.921,
      "step": 2593
    },
    {
      "epoch": 0.4347607475069136,
      "grad_norm": 0.1836388654956824,
      "learning_rate": 8.873525846926588e-09,
      "loss": 0.8582,
      "step": 2594
    },
    {
      "epoch": 0.43492834995390933,
      "grad_norm": 0.15063759604116764,
      "learning_rate": 8.828874159538092e-09,
      "loss": 0.8838,
      "step": 2595
    },
    {
      "epoch": 0.43509595240090504,
      "grad_norm": 0.1811298234524105,
      "learning_rate": 8.784329910657162e-09,
      "loss": 0.8818,
      "step": 2596
    },
    {
      "epoch": 0.4352635548479008,
      "grad_norm": 0.1594770620001015,
      "learning_rate": 8.739893152775956e-09,
      "loss": 0.8944,
      "step": 2597
    },
    {
      "epoch": 0.4354311572948965,
      "grad_norm": 0.20001747166392986,
      "learning_rate": 8.695563938259953e-09,
      "loss": 0.9149,
      "step": 2598
    },
    {
      "epoch": 0.4355987597418922,
      "grad_norm": 0.16274925694612508,
      "learning_rate": 8.651342319347864e-09,
      "loss": 0.8563,
      "step": 2599
    },
    {
      "epoch": 0.435766362188888,
      "grad_norm": 0.17853382351293656,
      "learning_rate": 8.607228348151651e-09,
      "loss": 0.8996,
      "step": 2600
    },
    {
      "epoch": 0.4359339646358837,
      "grad_norm": 0.19849907856665996,
      "learning_rate": 8.563222076656429e-09,
      "loss": 0.9168,
      "step": 2601
    },
    {
      "epoch": 0.4361015670828794,
      "grad_norm": 0.16184080012932384,
      "learning_rate": 8.519323556720348e-09,
      "loss": 0.8671,
      "step": 2602
    },
    {
      "epoch": 0.43626916952987516,
      "grad_norm": 0.154759556094354,
      "learning_rate": 8.475532840074584e-09,
      "loss": 0.8505,
      "step": 2603
    },
    {
      "epoch": 0.43643677197687086,
      "grad_norm": 0.1789582300899027,
      "learning_rate": 8.431849978323324e-09,
      "loss": 0.884,
      "step": 2604
    },
    {
      "epoch": 0.43660437442386657,
      "grad_norm": 0.19170898442740497,
      "learning_rate": 8.388275022943648e-09,
      "loss": 0.9205,
      "step": 2605
    },
    {
      "epoch": 0.43677197687086233,
      "grad_norm": 0.15827980571592343,
      "learning_rate": 8.344808025285399e-09,
      "loss": 0.8539,
      "step": 2606
    },
    {
      "epoch": 0.43693957931785804,
      "grad_norm": 0.1811407114795912,
      "learning_rate": 8.301449036571318e-09,
      "loss": 0.8782,
      "step": 2607
    },
    {
      "epoch": 0.43710718176485375,
      "grad_norm": 0.1828967898111119,
      "learning_rate": 8.258198107896753e-09,
      "loss": 0.8788,
      "step": 2608
    },
    {
      "epoch": 0.4372747842118495,
      "grad_norm": 0.16720751321696814,
      "learning_rate": 8.215055290229778e-09,
      "loss": 0.8516,
      "step": 2609
    },
    {
      "epoch": 0.4374423866588452,
      "grad_norm": 0.21477743353551373,
      "learning_rate": 8.17202063441107e-09,
      "loss": 0.9057,
      "step": 2610
    },
    {
      "epoch": 0.4376099891058409,
      "grad_norm": 0.16423600985145345,
      "learning_rate": 8.129094191153808e-09,
      "loss": 0.8413,
      "step": 2611
    },
    {
      "epoch": 0.4377775915528367,
      "grad_norm": 0.17401285778698758,
      "learning_rate": 8.086276011043658e-09,
      "loss": 0.8694,
      "step": 2612
    },
    {
      "epoch": 0.4379451939998324,
      "grad_norm": 0.17513500471379534,
      "learning_rate": 8.043566144538737e-09,
      "loss": 0.8686,
      "step": 2613
    },
    {
      "epoch": 0.4381127964468281,
      "grad_norm": 0.18012341139443927,
      "learning_rate": 8.000964641969477e-09,
      "loss": 0.8996,
      "step": 2614
    },
    {
      "epoch": 0.43828039889382386,
      "grad_norm": 0.162923613011852,
      "learning_rate": 7.958471553538649e-09,
      "loss": 0.8613,
      "step": 2615
    },
    {
      "epoch": 0.43844800134081957,
      "grad_norm": 0.16694260800198973,
      "learning_rate": 7.916086929321275e-09,
      "loss": 0.872,
      "step": 2616
    },
    {
      "epoch": 0.4386156037878153,
      "grad_norm": 0.16632485084897417,
      "learning_rate": 7.873810819264481e-09,
      "loss": 0.8311,
      "step": 2617
    },
    {
      "epoch": 0.43878320623481104,
      "grad_norm": 0.18043310210896532,
      "learning_rate": 7.831643273187638e-09,
      "loss": 0.8785,
      "step": 2618
    },
    {
      "epoch": 0.43895080868180675,
      "grad_norm": 0.15814250074962374,
      "learning_rate": 7.789584340782074e-09,
      "loss": 0.8873,
      "step": 2619
    },
    {
      "epoch": 0.43911841112880245,
      "grad_norm": 0.19315739515428715,
      "learning_rate": 7.747634071611142e-09,
      "loss": 0.8734,
      "step": 2620
    },
    {
      "epoch": 0.4392860135757982,
      "grad_norm": 0.1581917408888513,
      "learning_rate": 7.705792515110232e-09,
      "loss": 0.8543,
      "step": 2621
    },
    {
      "epoch": 0.4394536160227939,
      "grad_norm": 0.17069339388118304,
      "learning_rate": 7.664059720586525e-09,
      "loss": 0.8683,
      "step": 2622
    },
    {
      "epoch": 0.43962121846978963,
      "grad_norm": 0.18401824182523877,
      "learning_rate": 7.622435737219068e-09,
      "loss": 0.8703,
      "step": 2623
    },
    {
      "epoch": 0.4397888209167854,
      "grad_norm": 0.1807916743682259,
      "learning_rate": 7.580920614058695e-09,
      "loss": 0.9129,
      "step": 2624
    },
    {
      "epoch": 0.4399564233637811,
      "grad_norm": 0.17702364884715593,
      "learning_rate": 7.539514400027925e-09,
      "loss": 0.8629,
      "step": 2625
    },
    {
      "epoch": 0.44012402581077686,
      "grad_norm": 0.16278986163589554,
      "learning_rate": 7.49821714392096e-09,
      "loss": 0.8594,
      "step": 2626
    },
    {
      "epoch": 0.44029162825777257,
      "grad_norm": 0.15484054442398273,
      "learning_rate": 7.457028894403628e-09,
      "loss": 0.814,
      "step": 2627
    },
    {
      "epoch": 0.4404592307047683,
      "grad_norm": 0.1627744772464351,
      "learning_rate": 7.415949700013246e-09,
      "loss": 0.8814,
      "step": 2628
    },
    {
      "epoch": 0.44062683315176404,
      "grad_norm": 0.18142072151095093,
      "learning_rate": 7.374979609158616e-09,
      "loss": 0.9095,
      "step": 2629
    },
    {
      "epoch": 0.44079443559875975,
      "grad_norm": 0.17183649136258278,
      "learning_rate": 7.334118670120026e-09,
      "loss": 0.8715,
      "step": 2630
    },
    {
      "epoch": 0.44096203804575546,
      "grad_norm": 0.15582373672378688,
      "learning_rate": 7.2933669310491274e-09,
      "loss": 0.8502,
      "step": 2631
    },
    {
      "epoch": 0.4411296404927512,
      "grad_norm": 0.18520661139819816,
      "learning_rate": 7.252724439968816e-09,
      "loss": 0.8601,
      "step": 2632
    },
    {
      "epoch": 0.4412972429397469,
      "grad_norm": 0.19884083176470113,
      "learning_rate": 7.212191244773336e-09,
      "loss": 0.9205,
      "step": 2633
    },
    {
      "epoch": 0.44146484538674263,
      "grad_norm": 0.1568097837679988,
      "learning_rate": 7.1717673932280675e-09,
      "loss": 0.8346,
      "step": 2634
    },
    {
      "epoch": 0.4416324478337384,
      "grad_norm": 0.1789812146835891,
      "learning_rate": 7.131452932969595e-09,
      "loss": 0.9052,
      "step": 2635
    },
    {
      "epoch": 0.4418000502807341,
      "grad_norm": 0.16173589494919066,
      "learning_rate": 7.091247911505516e-09,
      "loss": 0.828,
      "step": 2636
    },
    {
      "epoch": 0.4419676527277298,
      "grad_norm": 0.19131161742128677,
      "learning_rate": 7.051152376214564e-09,
      "loss": 0.9335,
      "step": 2637
    },
    {
      "epoch": 0.44213525517472557,
      "grad_norm": 0.17448743451086926,
      "learning_rate": 7.011166374346356e-09,
      "loss": 0.9153,
      "step": 2638
    },
    {
      "epoch": 0.4423028576217213,
      "grad_norm": 0.1667761438011156,
      "learning_rate": 6.971289953021509e-09,
      "loss": 0.8773,
      "step": 2639
    },
    {
      "epoch": 0.442470460068717,
      "grad_norm": 0.1631529562748383,
      "learning_rate": 6.931523159231422e-09,
      "loss": 0.839,
      "step": 2640
    },
    {
      "epoch": 0.44263806251571275,
      "grad_norm": 0.17198659729410837,
      "learning_rate": 6.89186603983839e-09,
      "loss": 0.8888,
      "step": 2641
    },
    {
      "epoch": 0.44280566496270846,
      "grad_norm": 0.16293405664828428,
      "learning_rate": 6.852318641575439e-09,
      "loss": 0.868,
      "step": 2642
    },
    {
      "epoch": 0.44297326740970416,
      "grad_norm": 0.16301314106324952,
      "learning_rate": 6.812881011046246e-09,
      "loss": 0.8305,
      "step": 2643
    },
    {
      "epoch": 0.4431408698566999,
      "grad_norm": 0.15743889777530498,
      "learning_rate": 6.773553194725224e-09,
      "loss": 0.8605,
      "step": 2644
    },
    {
      "epoch": 0.44330847230369563,
      "grad_norm": 0.19526118443848783,
      "learning_rate": 6.7343352389573e-09,
      "loss": 0.896,
      "step": 2645
    },
    {
      "epoch": 0.44347607475069134,
      "grad_norm": 0.19162319070149464,
      "learning_rate": 6.6952271899579415e-09,
      "loss": 0.8515,
      "step": 2646
    },
    {
      "epoch": 0.4436436771976871,
      "grad_norm": 0.17078848435057215,
      "learning_rate": 6.656229093813182e-09,
      "loss": 0.8498,
      "step": 2647
    },
    {
      "epoch": 0.4438112796446828,
      "grad_norm": 0.16427421294288985,
      "learning_rate": 6.617340996479414e-09,
      "loss": 0.8609,
      "step": 2648
    },
    {
      "epoch": 0.4439788820916785,
      "grad_norm": 0.19379626160478478,
      "learning_rate": 6.578562943783383e-09,
      "loss": 0.8964,
      "step": 2649
    },
    {
      "epoch": 0.4441464845386743,
      "grad_norm": 0.16363068250629734,
      "learning_rate": 6.539894981422234e-09,
      "loss": 0.8352,
      "step": 2650
    },
    {
      "epoch": 0.44431408698567,
      "grad_norm": 0.16481154008340765,
      "learning_rate": 6.501337154963304e-09,
      "loss": 0.8284,
      "step": 2651
    },
    {
      "epoch": 0.4444816894326657,
      "grad_norm": 0.17216471351500823,
      "learning_rate": 6.4628895098441715e-09,
      "loss": 0.8635,
      "step": 2652
    },
    {
      "epoch": 0.44464929187966146,
      "grad_norm": 0.17772135274343767,
      "learning_rate": 6.424552091372604e-09,
      "loss": 0.848,
      "step": 2653
    },
    {
      "epoch": 0.44481689432665716,
      "grad_norm": 0.15512905616872108,
      "learning_rate": 6.3863249447264286e-09,
      "loss": 0.8447,
      "step": 2654
    },
    {
      "epoch": 0.44498449677365287,
      "grad_norm": 0.19769272359950094,
      "learning_rate": 6.348208114953535e-09,
      "loss": 0.9076,
      "step": 2655
    },
    {
      "epoch": 0.44515209922064863,
      "grad_norm": 0.17193366591171866,
      "learning_rate": 6.310201646971847e-09,
      "loss": 0.8759,
      "step": 2656
    },
    {
      "epoch": 0.44531970166764434,
      "grad_norm": 0.16114479950841287,
      "learning_rate": 6.272305585569171e-09,
      "loss": 0.8702,
      "step": 2657
    },
    {
      "epoch": 0.44548730411464005,
      "grad_norm": 0.18428191584739106,
      "learning_rate": 6.234519975403263e-09,
      "loss": 0.8959,
      "step": 2658
    },
    {
      "epoch": 0.4456549065616358,
      "grad_norm": 0.19428285830089667,
      "learning_rate": 6.1968448610017265e-09,
      "loss": 0.9017,
      "step": 2659
    },
    {
      "epoch": 0.4458225090086315,
      "grad_norm": 0.18830790271839606,
      "learning_rate": 6.159280286761892e-09,
      "loss": 0.8585,
      "step": 2660
    },
    {
      "epoch": 0.4459901114556272,
      "grad_norm": 0.15972055905920027,
      "learning_rate": 6.121826296950905e-09,
      "loss": 0.827,
      "step": 2661
    },
    {
      "epoch": 0.446157713902623,
      "grad_norm": 0.17528223430686724,
      "learning_rate": 6.084482935705504e-09,
      "loss": 0.8602,
      "step": 2662
    },
    {
      "epoch": 0.4463253163496187,
      "grad_norm": 0.17835855204491202,
      "learning_rate": 6.047250247032165e-09,
      "loss": 0.9006,
      "step": 2663
    },
    {
      "epoch": 0.44649291879661446,
      "grad_norm": 0.2079127560543362,
      "learning_rate": 6.010128274806847e-09,
      "loss": 0.923,
      "step": 2664
    },
    {
      "epoch": 0.44666052124361016,
      "grad_norm": 0.1816384932328198,
      "learning_rate": 5.973117062775112e-09,
      "loss": 0.8975,
      "step": 2665
    },
    {
      "epoch": 0.44682812369060587,
      "grad_norm": 0.15425074292914093,
      "learning_rate": 5.936216654551929e-09,
      "loss": 0.878,
      "step": 2666
    },
    {
      "epoch": 0.44699572613760163,
      "grad_norm": 0.18181941003595417,
      "learning_rate": 5.8994270936217675e-09,
      "loss": 0.8607,
      "step": 2667
    },
    {
      "epoch": 0.44716332858459734,
      "grad_norm": 0.16941455056592924,
      "learning_rate": 5.862748423338437e-09,
      "loss": 0.882,
      "step": 2668
    },
    {
      "epoch": 0.44733093103159305,
      "grad_norm": 0.18477461437444784,
      "learning_rate": 5.826180686925031e-09,
      "loss": 0.9149,
      "step": 2669
    },
    {
      "epoch": 0.4474985334785888,
      "grad_norm": 0.17760507781390794,
      "learning_rate": 5.7897239274740105e-09,
      "loss": 0.8868,
      "step": 2670
    },
    {
      "epoch": 0.4476661359255845,
      "grad_norm": 0.17491155187102542,
      "learning_rate": 5.753378187946967e-09,
      "loss": 0.8834,
      "step": 2671
    },
    {
      "epoch": 0.4478337383725802,
      "grad_norm": 0.1933325215030335,
      "learning_rate": 5.71714351117466e-09,
      "loss": 0.9518,
      "step": 2672
    },
    {
      "epoch": 0.448001340819576,
      "grad_norm": 0.1526519860900119,
      "learning_rate": 5.68101993985709e-09,
      "loss": 0.8547,
      "step": 2673
    },
    {
      "epoch": 0.4481689432665717,
      "grad_norm": 0.15875853689959254,
      "learning_rate": 5.645007516563205e-09,
      "loss": 0.8824,
      "step": 2674
    },
    {
      "epoch": 0.4483365457135674,
      "grad_norm": 0.17263550311440393,
      "learning_rate": 5.609106283730991e-09,
      "loss": 0.8465,
      "step": 2675
    },
    {
      "epoch": 0.44850414816056317,
      "grad_norm": 0.182456680066194,
      "learning_rate": 5.573316283667473e-09,
      "loss": 0.9028,
      "step": 2676
    },
    {
      "epoch": 0.4486717506075589,
      "grad_norm": 0.16801161002346393,
      "learning_rate": 5.5376375585485245e-09,
      "loss": 0.8665,
      "step": 2677
    },
    {
      "epoch": 0.4488393530545546,
      "grad_norm": 0.169416101080608,
      "learning_rate": 5.50207015041888e-09,
      "loss": 0.8529,
      "step": 2678
    },
    {
      "epoch": 0.44900695550155034,
      "grad_norm": 0.15909713310464974,
      "learning_rate": 5.466614101192202e-09,
      "loss": 0.8599,
      "step": 2679
    },
    {
      "epoch": 0.44917455794854605,
      "grad_norm": 0.1746254673674983,
      "learning_rate": 5.431269452650811e-09,
      "loss": 0.898,
      "step": 2680
    },
    {
      "epoch": 0.44934216039554176,
      "grad_norm": 0.1752043702684684,
      "learning_rate": 5.39603624644579e-09,
      "loss": 0.9024,
      "step": 2681
    },
    {
      "epoch": 0.4495097628425375,
      "grad_norm": 0.17080207144985615,
      "learning_rate": 5.360914524096915e-09,
      "loss": 0.8535,
      "step": 2682
    },
    {
      "epoch": 0.4496773652895332,
      "grad_norm": 0.19607189134809236,
      "learning_rate": 5.325904326992547e-09,
      "loss": 0.9441,
      "step": 2683
    },
    {
      "epoch": 0.44984496773652893,
      "grad_norm": 0.18223658150452057,
      "learning_rate": 5.291005696389661e-09,
      "loss": 0.8697,
      "step": 2684
    },
    {
      "epoch": 0.4500125701835247,
      "grad_norm": 0.14248083211005666,
      "learning_rate": 5.256218673413748e-09,
      "loss": 0.8415,
      "step": 2685
    },
    {
      "epoch": 0.4501801726305204,
      "grad_norm": 0.18528920812734215,
      "learning_rate": 5.221543299058739e-09,
      "loss": 0.8902,
      "step": 2686
    },
    {
      "epoch": 0.4503477750775161,
      "grad_norm": 0.17039473948517725,
      "learning_rate": 5.18697961418707e-09,
      "loss": 0.8859,
      "step": 2687
    },
    {
      "epoch": 0.4505153775245119,
      "grad_norm": 0.17881841233101656,
      "learning_rate": 5.152527659529482e-09,
      "loss": 0.897,
      "step": 2688
    },
    {
      "epoch": 0.4506829799715076,
      "grad_norm": 0.1423863672984811,
      "learning_rate": 5.11818747568512e-09,
      "loss": 0.8413,
      "step": 2689
    },
    {
      "epoch": 0.4508505824185033,
      "grad_norm": 0.17967024979857027,
      "learning_rate": 5.083959103121338e-09,
      "loss": 0.9,
      "step": 2690
    },
    {
      "epoch": 0.45101818486549905,
      "grad_norm": 0.17311565811153928,
      "learning_rate": 5.049842582173814e-09,
      "loss": 0.8654,
      "step": 2691
    },
    {
      "epoch": 0.45118578731249476,
      "grad_norm": 0.185597237911278,
      "learning_rate": 5.015837953046342e-09,
      "loss": 0.8541,
      "step": 2692
    },
    {
      "epoch": 0.45135338975949046,
      "grad_norm": 0.18402889362079772,
      "learning_rate": 4.981945255810904e-09,
      "loss": 0.8981,
      "step": 2693
    },
    {
      "epoch": 0.4515209922064862,
      "grad_norm": 0.16519583606022226,
      "learning_rate": 4.948164530407572e-09,
      "loss": 0.8972,
      "step": 2694
    },
    {
      "epoch": 0.45168859465348193,
      "grad_norm": 0.17651609474541238,
      "learning_rate": 4.914495816644448e-09,
      "loss": 0.8823,
      "step": 2695
    },
    {
      "epoch": 0.45185619710047764,
      "grad_norm": 0.17639451513434212,
      "learning_rate": 4.8809391541976716e-09,
      "loss": 0.9016,
      "step": 2696
    },
    {
      "epoch": 0.4520237995474734,
      "grad_norm": 0.1751685074490951,
      "learning_rate": 4.8474945826113e-09,
      "loss": 0.8663,
      "step": 2697
    },
    {
      "epoch": 0.4521914019944691,
      "grad_norm": 0.16028846653614007,
      "learning_rate": 4.814162141297307e-09,
      "loss": 0.8684,
      "step": 2698
    },
    {
      "epoch": 0.4523590044414648,
      "grad_norm": 0.15470780834570433,
      "learning_rate": 4.78094186953556e-09,
      "loss": 0.8459,
      "step": 2699
    },
    {
      "epoch": 0.4525266068884606,
      "grad_norm": 0.20689712440085384,
      "learning_rate": 4.747833806473711e-09,
      "loss": 0.8739,
      "step": 2700
    },
    {
      "epoch": 0.4526942093354563,
      "grad_norm": 0.16446388165605658,
      "learning_rate": 4.714837991127185e-09,
      "loss": 0.8695,
      "step": 2701
    },
    {
      "epoch": 0.45286181178245205,
      "grad_norm": 0.16861027583497337,
      "learning_rate": 4.681954462379167e-09,
      "loss": 0.85,
      "step": 2702
    },
    {
      "epoch": 0.45302941422944776,
      "grad_norm": 0.16043131571464198,
      "learning_rate": 4.6491832589804605e-09,
      "loss": 0.8632,
      "step": 2703
    },
    {
      "epoch": 0.45319701667644346,
      "grad_norm": 0.19760375628269505,
      "learning_rate": 4.616524419549528e-09,
      "loss": 0.9188,
      "step": 2704
    },
    {
      "epoch": 0.4533646191234392,
      "grad_norm": 0.18106770314531984,
      "learning_rate": 4.583977982572462e-09,
      "loss": 0.8668,
      "step": 2705
    },
    {
      "epoch": 0.45353222157043493,
      "grad_norm": 0.16774015988462584,
      "learning_rate": 4.551543986402839e-09,
      "loss": 0.8677,
      "step": 2706
    },
    {
      "epoch": 0.45369982401743064,
      "grad_norm": 0.18197505265663882,
      "learning_rate": 4.51922246926173e-09,
      "loss": 0.8973,
      "step": 2707
    },
    {
      "epoch": 0.4538674264644264,
      "grad_norm": 0.15641633592368515,
      "learning_rate": 4.487013469237721e-09,
      "loss": 0.8204,
      "step": 2708
    },
    {
      "epoch": 0.4540350289114221,
      "grad_norm": 0.16470067768644064,
      "learning_rate": 4.454917024286742e-09,
      "loss": 0.8611,
      "step": 2709
    },
    {
      "epoch": 0.4542026313584178,
      "grad_norm": 0.1593089612319892,
      "learning_rate": 4.422933172232101e-09,
      "loss": 0.8514,
      "step": 2710
    },
    {
      "epoch": 0.4543702338054136,
      "grad_norm": 0.17277916855116313,
      "learning_rate": 4.391061950764452e-09,
      "loss": 0.8711,
      "step": 2711
    },
    {
      "epoch": 0.4545378362524093,
      "grad_norm": 0.18170020473281576,
      "learning_rate": 4.359303397441671e-09,
      "loss": 0.9025,
      "step": 2712
    },
    {
      "epoch": 0.454705438699405,
      "grad_norm": 0.18624133507655405,
      "learning_rate": 4.3276575496889346e-09,
      "loss": 0.8905,
      "step": 2713
    },
    {
      "epoch": 0.45487304114640076,
      "grad_norm": 0.19096318260992973,
      "learning_rate": 4.296124444798521e-09,
      "loss": 0.9123,
      "step": 2714
    },
    {
      "epoch": 0.45504064359339647,
      "grad_norm": 0.14958845682630376,
      "learning_rate": 4.264704119929896e-09,
      "loss": 0.8099,
      "step": 2715
    },
    {
      "epoch": 0.4552082460403922,
      "grad_norm": 0.16581490325292256,
      "learning_rate": 4.2333966121096186e-09,
      "loss": 0.8658,
      "step": 2716
    },
    {
      "epoch": 0.45537584848738794,
      "grad_norm": 0.18304537699526796,
      "learning_rate": 4.2022019582313215e-09,
      "loss": 0.9156,
      "step": 2717
    },
    {
      "epoch": 0.45554345093438364,
      "grad_norm": 0.1793211665082464,
      "learning_rate": 4.171120195055588e-09,
      "loss": 0.8834,
      "step": 2718
    },
    {
      "epoch": 0.45571105338137935,
      "grad_norm": 0.18268111983725183,
      "learning_rate": 4.140151359210042e-09,
      "loss": 0.9055,
      "step": 2719
    },
    {
      "epoch": 0.4558786558283751,
      "grad_norm": 0.18664322854719564,
      "learning_rate": 4.109295487189146e-09,
      "loss": 0.8937,
      "step": 2720
    },
    {
      "epoch": 0.4560462582753708,
      "grad_norm": 0.1784807310606192,
      "learning_rate": 4.078552615354325e-09,
      "loss": 0.8838,
      "step": 2721
    },
    {
      "epoch": 0.4562138607223665,
      "grad_norm": 0.18188978843256867,
      "learning_rate": 4.0479227799338104e-09,
      "loss": 0.8894,
      "step": 2722
    },
    {
      "epoch": 0.4563814631693623,
      "grad_norm": 0.16395117005421456,
      "learning_rate": 4.017406017022606e-09,
      "loss": 0.8717,
      "step": 2723
    },
    {
      "epoch": 0.456549065616358,
      "grad_norm": 0.17420575333369637,
      "learning_rate": 3.987002362582481e-09,
      "loss": 0.8862,
      "step": 2724
    },
    {
      "epoch": 0.4567166680633537,
      "grad_norm": 0.15547270856850115,
      "learning_rate": 3.956711852441929e-09,
      "loss": 0.8631,
      "step": 2725
    },
    {
      "epoch": 0.45688427051034947,
      "grad_norm": 0.17541840230989833,
      "learning_rate": 3.926534522296121e-09,
      "loss": 0.8732,
      "step": 2726
    },
    {
      "epoch": 0.4570518729573452,
      "grad_norm": 0.17505612847365906,
      "learning_rate": 3.896470407706798e-09,
      "loss": 0.9033,
      "step": 2727
    },
    {
      "epoch": 0.4572194754043409,
      "grad_norm": 0.16015204311669784,
      "learning_rate": 3.8665195441023535e-09,
      "loss": 0.8582,
      "step": 2728
    },
    {
      "epoch": 0.45738707785133664,
      "grad_norm": 0.18508393974956158,
      "learning_rate": 3.836681966777677e-09,
      "loss": 0.8641,
      "step": 2729
    },
    {
      "epoch": 0.45755468029833235,
      "grad_norm": 0.21063771450665533,
      "learning_rate": 3.806957710894143e-09,
      "loss": 0.9172,
      "step": 2730
    },
    {
      "epoch": 0.45772228274532806,
      "grad_norm": 0.17333458598266574,
      "learning_rate": 3.777346811479676e-09,
      "loss": 0.8788,
      "step": 2731
    },
    {
      "epoch": 0.4578898851923238,
      "grad_norm": 0.17428375467495516,
      "learning_rate": 3.7478493034285295e-09,
      "loss": 0.8964,
      "step": 2732
    },
    {
      "epoch": 0.4580574876393195,
      "grad_norm": 0.19330703153884393,
      "learning_rate": 3.7184652215013437e-09,
      "loss": 0.9302,
      "step": 2733
    },
    {
      "epoch": 0.45822509008631523,
      "grad_norm": 0.17595194944069517,
      "learning_rate": 3.6891946003251295e-09,
      "loss": 0.8881,
      "step": 2734
    },
    {
      "epoch": 0.458392692533311,
      "grad_norm": 0.18554153741041268,
      "learning_rate": 3.6600374743931515e-09,
      "loss": 0.8996,
      "step": 2735
    },
    {
      "epoch": 0.4585602949803067,
      "grad_norm": 0.17890976694005112,
      "learning_rate": 3.6309938780649805e-09,
      "loss": 0.8824,
      "step": 2736
    },
    {
      "epoch": 0.4587278974273024,
      "grad_norm": 0.188620761699398,
      "learning_rate": 3.602063845566361e-09,
      "loss": 0.8715,
      "step": 2737
    },
    {
      "epoch": 0.4588954998742982,
      "grad_norm": 0.1708322870489815,
      "learning_rate": 3.573247410989233e-09,
      "loss": 0.873,
      "step": 2738
    },
    {
      "epoch": 0.4590631023212939,
      "grad_norm": 0.15503087548815225,
      "learning_rate": 3.544544608291622e-09,
      "loss": 0.8313,
      "step": 2739
    },
    {
      "epoch": 0.45923070476828964,
      "grad_norm": 0.17342320652196425,
      "learning_rate": 3.5159554712977267e-09,
      "loss": 0.8641,
      "step": 2740
    },
    {
      "epoch": 0.45939830721528535,
      "grad_norm": 0.17170146675412318,
      "learning_rate": 3.4874800336977206e-09,
      "loss": 0.8728,
      "step": 2741
    },
    {
      "epoch": 0.45956590966228106,
      "grad_norm": 0.1922765707196611,
      "learning_rate": 3.4591183290478388e-09,
      "loss": 0.9103,
      "step": 2742
    },
    {
      "epoch": 0.4597335121092768,
      "grad_norm": 0.16439067125955445,
      "learning_rate": 3.430870390770291e-09,
      "loss": 0.8974,
      "step": 2743
    },
    {
      "epoch": 0.45990111455627253,
      "grad_norm": 0.17575572342909107,
      "learning_rate": 3.402736252153182e-09,
      "loss": 0.8965,
      "step": 2744
    },
    {
      "epoch": 0.46006871700326823,
      "grad_norm": 0.1631083455006615,
      "learning_rate": 3.3747159463505594e-09,
      "loss": 0.8714,
      "step": 2745
    },
    {
      "epoch": 0.460236319450264,
      "grad_norm": 0.1792370546662929,
      "learning_rate": 3.3468095063822863e-09,
      "loss": 0.9155,
      "step": 2746
    },
    {
      "epoch": 0.4604039218972597,
      "grad_norm": 0.17975649733706273,
      "learning_rate": 3.3190169651340806e-09,
      "loss": 0.8832,
      "step": 2747
    },
    {
      "epoch": 0.4605715243442554,
      "grad_norm": 0.15441803507230054,
      "learning_rate": 3.2913383553574203e-09,
      "loss": 0.8475,
      "step": 2748
    },
    {
      "epoch": 0.4607391267912512,
      "grad_norm": 0.149226238741696,
      "learning_rate": 3.263773709669515e-09,
      "loss": 0.8543,
      "step": 2749
    },
    {
      "epoch": 0.4609067292382469,
      "grad_norm": 0.1800752013612399,
      "learning_rate": 3.23632306055327e-09,
      "loss": 0.8804,
      "step": 2750
    },
    {
      "epoch": 0.4610743316852426,
      "grad_norm": 0.17903153641848427,
      "learning_rate": 3.2089864403572973e-09,
      "loss": 0.8448,
      "step": 2751
    },
    {
      "epoch": 0.46124193413223835,
      "grad_norm": 0.17193014123000555,
      "learning_rate": 3.1817638812957957e-09,
      "loss": 0.8484,
      "step": 2752
    },
    {
      "epoch": 0.46140953657923406,
      "grad_norm": 0.19048654621637642,
      "learning_rate": 3.1546554154485485e-09,
      "loss": 0.9301,
      "step": 2753
    },
    {
      "epoch": 0.46157713902622977,
      "grad_norm": 0.1708903772347989,
      "learning_rate": 3.127661074760923e-09,
      "loss": 0.8578,
      "step": 2754
    },
    {
      "epoch": 0.46174474147322553,
      "grad_norm": 0.1789336457904136,
      "learning_rate": 3.100780891043764e-09,
      "loss": 0.8682,
      "step": 2755
    },
    {
      "epoch": 0.46191234392022124,
      "grad_norm": 0.1705376861008378,
      "learning_rate": 3.074014895973398e-09,
      "loss": 0.8914,
      "step": 2756
    },
    {
      "epoch": 0.46207994636721694,
      "grad_norm": 0.16598963135754702,
      "learning_rate": 3.047363121091595e-09,
      "loss": 0.8989,
      "step": 2757
    },
    {
      "epoch": 0.4622475488142127,
      "grad_norm": 0.17139612330036827,
      "learning_rate": 3.020825597805532e-09,
      "loss": 0.8796,
      "step": 2758
    },
    {
      "epoch": 0.4624151512612084,
      "grad_norm": 0.17080670191598238,
      "learning_rate": 2.9944023573877377e-09,
      "loss": 0.8955,
      "step": 2759
    },
    {
      "epoch": 0.4625827537082041,
      "grad_norm": 0.17748932302222445,
      "learning_rate": 2.968093430976082e-09,
      "loss": 0.8678,
      "step": 2760
    },
    {
      "epoch": 0.4627503561551999,
      "grad_norm": 0.17519284871626872,
      "learning_rate": 2.9418988495736765e-09,
      "loss": 0.8702,
      "step": 2761
    },
    {
      "epoch": 0.4629179586021956,
      "grad_norm": 0.16887827214530063,
      "learning_rate": 2.9158186440489508e-09,
      "loss": 0.8627,
      "step": 2762
    },
    {
      "epoch": 0.4630855610491913,
      "grad_norm": 0.1752119007779156,
      "learning_rate": 2.8898528451355208e-09,
      "loss": 0.8774,
      "step": 2763
    },
    {
      "epoch": 0.46325316349618706,
      "grad_norm": 0.16373107932261904,
      "learning_rate": 2.864001483432166e-09,
      "loss": 0.8576,
      "step": 2764
    },
    {
      "epoch": 0.46342076594318277,
      "grad_norm": 0.17314261160080063,
      "learning_rate": 2.838264589402839e-09,
      "loss": 0.8676,
      "step": 2765
    },
    {
      "epoch": 0.4635883683901785,
      "grad_norm": 0.16586522182486113,
      "learning_rate": 2.8126421933765798e-09,
      "loss": 0.8929,
      "step": 2766
    },
    {
      "epoch": 0.46375597083717424,
      "grad_norm": 0.17665003219002634,
      "learning_rate": 2.787134325547513e-09,
      "loss": 0.8944,
      "step": 2767
    },
    {
      "epoch": 0.46392357328416994,
      "grad_norm": 0.16428845634884853,
      "learning_rate": 2.7617410159747944e-09,
      "loss": 0.8608,
      "step": 2768
    },
    {
      "epoch": 0.46409117573116565,
      "grad_norm": 0.16251132038890756,
      "learning_rate": 2.7364622945825975e-09,
      "loss": 0.8311,
      "step": 2769
    },
    {
      "epoch": 0.4642587781781614,
      "grad_norm": 0.16240715937446043,
      "learning_rate": 2.7112981911600165e-09,
      "loss": 0.8635,
      "step": 2770
    },
    {
      "epoch": 0.4644263806251571,
      "grad_norm": 0.17813576932226152,
      "learning_rate": 2.6862487353611406e-09,
      "loss": 0.9071,
      "step": 2771
    },
    {
      "epoch": 0.4645939830721528,
      "grad_norm": 0.17556347173745748,
      "learning_rate": 2.6613139567049025e-09,
      "loss": 0.8898,
      "step": 2772
    },
    {
      "epoch": 0.4647615855191486,
      "grad_norm": 0.17317943134906932,
      "learning_rate": 2.6364938845751196e-09,
      "loss": 0.8715,
      "step": 2773
    },
    {
      "epoch": 0.4649291879661443,
      "grad_norm": 0.16614154815144766,
      "learning_rate": 2.6117885482204505e-09,
      "loss": 0.8612,
      "step": 2774
    },
    {
      "epoch": 0.46509679041314,
      "grad_norm": 0.15438644332787727,
      "learning_rate": 2.5871979767543075e-09,
      "loss": 0.8101,
      "step": 2775
    },
    {
      "epoch": 0.46526439286013577,
      "grad_norm": 0.1681104312912706,
      "learning_rate": 2.5627221991548765e-09,
      "loss": 0.8581,
      "step": 2776
    },
    {
      "epoch": 0.4654319953071315,
      "grad_norm": 0.1738805368922363,
      "learning_rate": 2.538361244265097e-09,
      "loss": 0.8948,
      "step": 2777
    },
    {
      "epoch": 0.46559959775412724,
      "grad_norm": 0.181631621192094,
      "learning_rate": 2.5141151407925497e-09,
      "loss": 0.8854,
      "step": 2778
    },
    {
      "epoch": 0.46576720020112294,
      "grad_norm": 0.18445365055118357,
      "learning_rate": 2.489983917309502e-09,
      "loss": 0.8688,
      "step": 2779
    },
    {
      "epoch": 0.46593480264811865,
      "grad_norm": 0.170621328376433,
      "learning_rate": 2.4659676022528607e-09,
      "loss": 0.865,
      "step": 2780
    },
    {
      "epoch": 0.4661024050951144,
      "grad_norm": 0.19686685759531167,
      "learning_rate": 2.4420662239240664e-09,
      "loss": 0.9097,
      "step": 2781
    },
    {
      "epoch": 0.4662700075421101,
      "grad_norm": 0.18154299833662982,
      "learning_rate": 2.4182798104891543e-09,
      "loss": 0.9254,
      "step": 2782
    },
    {
      "epoch": 0.46643760998910583,
      "grad_norm": 0.17688439169682305,
      "learning_rate": 2.3946083899786694e-09,
      "loss": 0.891,
      "step": 2783
    },
    {
      "epoch": 0.4666052124361016,
      "grad_norm": 0.16183811086202757,
      "learning_rate": 2.3710519902876756e-09,
      "loss": 0.8664,
      "step": 2784
    },
    {
      "epoch": 0.4667728148830973,
      "grad_norm": 0.15926321352869763,
      "learning_rate": 2.3476106391756346e-09,
      "loss": 0.8743,
      "step": 2785
    },
    {
      "epoch": 0.466940417330093,
      "grad_norm": 0.1725340623250181,
      "learning_rate": 2.3242843642664935e-09,
      "loss": 0.8817,
      "step": 2786
    },
    {
      "epoch": 0.46710801977708877,
      "grad_norm": 0.1829611908764473,
      "learning_rate": 2.3010731930485416e-09,
      "loss": 0.866,
      "step": 2787
    },
    {
      "epoch": 0.4672756222240845,
      "grad_norm": 0.1799238572291312,
      "learning_rate": 2.277977152874444e-09,
      "loss": 0.8724,
      "step": 2788
    },
    {
      "epoch": 0.4674432246710802,
      "grad_norm": 0.1725570827680606,
      "learning_rate": 2.2549962709612293e-09,
      "loss": 0.9195,
      "step": 2789
    },
    {
      "epoch": 0.46761082711807594,
      "grad_norm": 0.18629510247338246,
      "learning_rate": 2.2321305743901673e-09,
      "loss": 0.9074,
      "step": 2790
    },
    {
      "epoch": 0.46777842956507165,
      "grad_norm": 0.18742701269605466,
      "learning_rate": 2.209380090106794e-09,
      "loss": 0.8884,
      "step": 2791
    },
    {
      "epoch": 0.46794603201206736,
      "grad_norm": 0.19300082706826188,
      "learning_rate": 2.186744844920907e-09,
      "loss": 0.9036,
      "step": 2792
    },
    {
      "epoch": 0.4681136344590631,
      "grad_norm": 0.15909063178128854,
      "learning_rate": 2.164224865506492e-09,
      "loss": 0.8529,
      "step": 2793
    },
    {
      "epoch": 0.46828123690605883,
      "grad_norm": 0.1613108449973953,
      "learning_rate": 2.1418201784016876e-09,
      "loss": 0.866,
      "step": 2794
    },
    {
      "epoch": 0.46844883935305454,
      "grad_norm": 0.17010116616536305,
      "learning_rate": 2.119530810008796e-09,
      "loss": 0.879,
      "step": 2795
    },
    {
      "epoch": 0.4686164418000503,
      "grad_norm": 0.17315353390773866,
      "learning_rate": 2.097356786594184e-09,
      "loss": 0.8748,
      "step": 2796
    },
    {
      "epoch": 0.468784044247046,
      "grad_norm": 0.1721809018186193,
      "learning_rate": 2.07529813428835e-09,
      "loss": 0.8801,
      "step": 2797
    },
    {
      "epoch": 0.4689516466940417,
      "grad_norm": 0.14757647311287342,
      "learning_rate": 2.053354879085789e-09,
      "loss": 0.837,
      "step": 2798
    },
    {
      "epoch": 0.4691192491410375,
      "grad_norm": 0.15651317681411653,
      "learning_rate": 2.031527046844994e-09,
      "loss": 0.8446,
      "step": 2799
    },
    {
      "epoch": 0.4692868515880332,
      "grad_norm": 0.17989890320912058,
      "learning_rate": 2.0098146632885117e-09,
      "loss": 0.8703,
      "step": 2800
    },
    {
      "epoch": 0.4694544540350289,
      "grad_norm": 0.1738126879126536,
      "learning_rate": 1.9882177540027854e-09,
      "loss": 0.8633,
      "step": 2801
    },
    {
      "epoch": 0.46962205648202465,
      "grad_norm": 0.20305210755731118,
      "learning_rate": 1.96673634443818e-09,
      "loss": 0.9195,
      "step": 2802
    },
    {
      "epoch": 0.46978965892902036,
      "grad_norm": 0.17024818786291857,
      "learning_rate": 1.9453704599089905e-09,
      "loss": 0.8778,
      "step": 2803
    },
    {
      "epoch": 0.46995726137601607,
      "grad_norm": 0.17332403462333662,
      "learning_rate": 1.9241201255933315e-09,
      "loss": 0.8855,
      "step": 2804
    },
    {
      "epoch": 0.47012486382301183,
      "grad_norm": 0.174912482273094,
      "learning_rate": 1.902985366533172e-09,
      "loss": 0.8566,
      "step": 2805
    },
    {
      "epoch": 0.47029246627000754,
      "grad_norm": 0.15515168291293588,
      "learning_rate": 1.8819662076343e-09,
      "loss": 0.8311,
      "step": 2806
    },
    {
      "epoch": 0.47046006871700324,
      "grad_norm": 0.1819648354235075,
      "learning_rate": 1.861062673666225e-09,
      "loss": 0.9029,
      "step": 2807
    },
    {
      "epoch": 0.470627671163999,
      "grad_norm": 0.1761075598709931,
      "learning_rate": 1.8402747892622528e-09,
      "loss": 0.8604,
      "step": 2808
    },
    {
      "epoch": 0.4707952736109947,
      "grad_norm": 0.18082551305456235,
      "learning_rate": 1.819602578919377e-09,
      "loss": 0.9064,
      "step": 2809
    },
    {
      "epoch": 0.4709628760579904,
      "grad_norm": 0.1642120603861888,
      "learning_rate": 1.799046066998311e-09,
      "loss": 0.8474,
      "step": 2810
    },
    {
      "epoch": 0.4711304785049862,
      "grad_norm": 0.1676330483150995,
      "learning_rate": 1.7786052777233663e-09,
      "loss": 0.8895,
      "step": 2811
    },
    {
      "epoch": 0.4712980809519819,
      "grad_norm": 0.165618999896715,
      "learning_rate": 1.758280235182541e-09,
      "loss": 0.8967,
      "step": 2812
    },
    {
      "epoch": 0.4714656833989776,
      "grad_norm": 0.16850005115775207,
      "learning_rate": 1.7380709633274094e-09,
      "loss": 0.8571,
      "step": 2813
    },
    {
      "epoch": 0.47163328584597336,
      "grad_norm": 0.1828121720231751,
      "learning_rate": 1.71797748597311e-09,
      "loss": 0.8996,
      "step": 2814
    },
    {
      "epoch": 0.47180088829296907,
      "grad_norm": 0.17875137991729323,
      "learning_rate": 1.6979998267983575e-09,
      "loss": 0.8614,
      "step": 2815
    },
    {
      "epoch": 0.47196849073996483,
      "grad_norm": 0.17604296756844515,
      "learning_rate": 1.6781380093453535e-09,
      "loss": 0.859,
      "step": 2816
    },
    {
      "epoch": 0.47213609318696054,
      "grad_norm": 0.16677447780427623,
      "learning_rate": 1.6583920570197751e-09,
      "loss": 0.8655,
      "step": 2817
    },
    {
      "epoch": 0.47230369563395624,
      "grad_norm": 0.17976950370114803,
      "learning_rate": 1.6387619930908203e-09,
      "loss": 0.8946,
      "step": 2818
    },
    {
      "epoch": 0.472471298080952,
      "grad_norm": 0.1659218262572627,
      "learning_rate": 1.6192478406910625e-09,
      "loss": 0.8488,
      "step": 2819
    },
    {
      "epoch": 0.4726389005279477,
      "grad_norm": 0.1852444658922347,
      "learning_rate": 1.599849622816507e-09,
      "loss": 0.8985,
      "step": 2820
    },
    {
      "epoch": 0.4728065029749434,
      "grad_norm": 0.18844005702578548,
      "learning_rate": 1.580567362326557e-09,
      "loss": 0.9405,
      "step": 2821
    },
    {
      "epoch": 0.4729741054219392,
      "grad_norm": 0.16367649207828144,
      "learning_rate": 1.5614010819439028e-09,
      "loss": 0.8592,
      "step": 2822
    },
    {
      "epoch": 0.4731417078689349,
      "grad_norm": 0.16990315158458716,
      "learning_rate": 1.5423508042546552e-09,
      "loss": 0.8694,
      "step": 2823
    },
    {
      "epoch": 0.4733093103159306,
      "grad_norm": 0.18677325572729017,
      "learning_rate": 1.5234165517081344e-09,
      "loss": 0.8992,
      "step": 2824
    },
    {
      "epoch": 0.47347691276292636,
      "grad_norm": 0.15581950736224578,
      "learning_rate": 1.5045983466169809e-09,
      "loss": 0.8343,
      "step": 2825
    },
    {
      "epoch": 0.47364451520992207,
      "grad_norm": 0.17212114746475288,
      "learning_rate": 1.4858962111571004e-09,
      "loss": 0.8765,
      "step": 2826
    },
    {
      "epoch": 0.4738121176569178,
      "grad_norm": 0.18139030688991417,
      "learning_rate": 1.4673101673675636e-09,
      "loss": 0.8689,
      "step": 2827
    },
    {
      "epoch": 0.47397972010391354,
      "grad_norm": 0.1868864473656384,
      "learning_rate": 1.4488402371506725e-09,
      "loss": 0.8606,
      "step": 2828
    },
    {
      "epoch": 0.47414732255090924,
      "grad_norm": 0.16060312980223443,
      "learning_rate": 1.4304864422718943e-09,
      "loss": 0.8271,
      "step": 2829
    },
    {
      "epoch": 0.47431492499790495,
      "grad_norm": 0.1657402893035018,
      "learning_rate": 1.4122488043598279e-09,
      "loss": 0.8808,
      "step": 2830
    },
    {
      "epoch": 0.4744825274449007,
      "grad_norm": 0.1752410976733369,
      "learning_rate": 1.394127344906204e-09,
      "loss": 0.8708,
      "step": 2831
    },
    {
      "epoch": 0.4746501298918964,
      "grad_norm": 0.16782093643823878,
      "learning_rate": 1.3761220852658295e-09,
      "loss": 0.8403,
      "step": 2832
    },
    {
      "epoch": 0.47481773233889213,
      "grad_norm": 0.1682768335037804,
      "learning_rate": 1.3582330466565983e-09,
      "loss": 0.8829,
      "step": 2833
    },
    {
      "epoch": 0.4749853347858879,
      "grad_norm": 0.17595969319188934,
      "learning_rate": 1.3404602501594252e-09,
      "loss": 0.8668,
      "step": 2834
    },
    {
      "epoch": 0.4751529372328836,
      "grad_norm": 0.16308142162864875,
      "learning_rate": 1.3228037167182571e-09,
      "loss": 0.8727,
      "step": 2835
    },
    {
      "epoch": 0.4753205396798793,
      "grad_norm": 0.19110763393048302,
      "learning_rate": 1.3052634671400165e-09,
      "loss": 0.9281,
      "step": 2836
    },
    {
      "epoch": 0.47548814212687507,
      "grad_norm": 0.1592367033201745,
      "learning_rate": 1.2878395220946247e-09,
      "loss": 0.8358,
      "step": 2837
    },
    {
      "epoch": 0.4756557445738708,
      "grad_norm": 0.16850024712652711,
      "learning_rate": 1.270531902114913e-09,
      "loss": 0.8827,
      "step": 2838
    },
    {
      "epoch": 0.4758233470208665,
      "grad_norm": 0.14043267253444688,
      "learning_rate": 1.2533406275966442e-09,
      "loss": 0.8063,
      "step": 2839
    },
    {
      "epoch": 0.47599094946786225,
      "grad_norm": 0.16846980492690375,
      "learning_rate": 1.2362657187985014e-09,
      "loss": 0.8703,
      "step": 2840
    },
    {
      "epoch": 0.47615855191485795,
      "grad_norm": 0.16736144689695007,
      "learning_rate": 1.2193071958419787e-09,
      "loss": 0.8567,
      "step": 2841
    },
    {
      "epoch": 0.47632615436185366,
      "grad_norm": 0.17437581549217665,
      "learning_rate": 1.2024650787115009e-09,
      "loss": 0.9098,
      "step": 2842
    },
    {
      "epoch": 0.4764937568088494,
      "grad_norm": 0.18487077229924745,
      "learning_rate": 1.1857393872542253e-09,
      "loss": 0.8884,
      "step": 2843
    },
    {
      "epoch": 0.47666135925584513,
      "grad_norm": 0.1500087995088469,
      "learning_rate": 1.1691301411801968e-09,
      "loss": 0.8136,
      "step": 2844
    },
    {
      "epoch": 0.47682896170284084,
      "grad_norm": 0.16642945402430978,
      "learning_rate": 1.1526373600621697e-09,
      "loss": 0.8464,
      "step": 2845
    },
    {
      "epoch": 0.4769965641498366,
      "grad_norm": 0.1720985804310414,
      "learning_rate": 1.1362610633356974e-09,
      "loss": 0.8393,
      "step": 2846
    },
    {
      "epoch": 0.4771641665968323,
      "grad_norm": 0.16785718471458852,
      "learning_rate": 1.1200012702990425e-09,
      "loss": 0.8508,
      "step": 2847
    },
    {
      "epoch": 0.477331769043828,
      "grad_norm": 0.16610706511105203,
      "learning_rate": 1.103858000113178e-09,
      "loss": 0.8448,
      "step": 2848
    },
    {
      "epoch": 0.4774993714908238,
      "grad_norm": 0.18099999095723493,
      "learning_rate": 1.0878312718017868e-09,
      "loss": 0.8972,
      "step": 2849
    },
    {
      "epoch": 0.4776669739378195,
      "grad_norm": 0.16727839529312952,
      "learning_rate": 1.0719211042511722e-09,
      "loss": 0.8526,
      "step": 2850
    },
    {
      "epoch": 0.4778345763848152,
      "grad_norm": 0.1671142894360413,
      "learning_rate": 1.0561275162103033e-09,
      "loss": 0.8571,
      "step": 2851
    },
    {
      "epoch": 0.47800217883181095,
      "grad_norm": 0.16663179311352289,
      "learning_rate": 1.0404505262908036e-09,
      "loss": 0.8391,
      "step": 2852
    },
    {
      "epoch": 0.47816978127880666,
      "grad_norm": 0.15428337583086071,
      "learning_rate": 1.0248901529668287e-09,
      "loss": 0.8705,
      "step": 2853
    },
    {
      "epoch": 0.4783373837258024,
      "grad_norm": 0.1788833922122076,
      "learning_rate": 1.0094464145751436e-09,
      "loss": 0.8742,
      "step": 2854
    },
    {
      "epoch": 0.47850498617279813,
      "grad_norm": 0.17652689794931872,
      "learning_rate": 9.941193293150796e-10,
      "loss": 0.8798,
      "step": 2855
    },
    {
      "epoch": 0.47867258861979384,
      "grad_norm": 0.17635252688909572,
      "learning_rate": 9.789089152484664e-10,
      "loss": 0.8804,
      "step": 2856
    },
    {
      "epoch": 0.4788401910667896,
      "grad_norm": 0.16281052158589915,
      "learning_rate": 9.638151902996772e-10,
      "loss": 0.844,
      "step": 2857
    },
    {
      "epoch": 0.4790077935137853,
      "grad_norm": 0.17796369141007665,
      "learning_rate": 9.488381722555728e-10,
      "loss": 0.8593,
      "step": 2858
    },
    {
      "epoch": 0.479175395960781,
      "grad_norm": 0.17622134958530436,
      "learning_rate": 9.339778787654573e-10,
      "loss": 0.835,
      "step": 2859
    },
    {
      "epoch": 0.4793429984077768,
      "grad_norm": 0.17983013553787222,
      "learning_rate": 9.192343273411007e-10,
      "loss": 0.9445,
      "step": 2860
    },
    {
      "epoch": 0.4795106008547725,
      "grad_norm": 0.1878623442029024,
      "learning_rate": 9.046075353567162e-10,
      "loss": 0.8925,
      "step": 2861
    },
    {
      "epoch": 0.4796782033017682,
      "grad_norm": 0.1673502646592197,
      "learning_rate": 8.900975200488936e-10,
      "loss": 0.8872,
      "step": 2862
    },
    {
      "epoch": 0.47984580574876395,
      "grad_norm": 0.19321849791481763,
      "learning_rate": 8.757042985166329e-10,
      "loss": 0.9346,
      "step": 2863
    },
    {
      "epoch": 0.48001340819575966,
      "grad_norm": 0.1844628282137391,
      "learning_rate": 8.614278877212888e-10,
      "loss": 0.917,
      "step": 2864
    },
    {
      "epoch": 0.48018101064275537,
      "grad_norm": 0.16585086364984536,
      "learning_rate": 8.472683044865814e-10,
      "loss": 0.8578,
      "step": 2865
    },
    {
      "epoch": 0.48034861308975113,
      "grad_norm": 0.16857640112813846,
      "learning_rate": 8.332255654985299e-10,
      "loss": 0.8714,
      "step": 2866
    },
    {
      "epoch": 0.48051621553674684,
      "grad_norm": 0.19025876443609938,
      "learning_rate": 8.192996873054747e-10,
      "loss": 0.8837,
      "step": 2867
    },
    {
      "epoch": 0.48068381798374255,
      "grad_norm": 0.17675752782882595,
      "learning_rate": 8.054906863180555e-10,
      "loss": 0.894,
      "step": 2868
    },
    {
      "epoch": 0.4808514204307383,
      "grad_norm": 0.18435786963434278,
      "learning_rate": 7.917985788091552e-10,
      "loss": 0.8826,
      "step": 2869
    },
    {
      "epoch": 0.481019022877734,
      "grad_norm": 0.1675042212956494,
      "learning_rate": 7.782233809139449e-10,
      "loss": 0.8743,
      "step": 2870
    },
    {
      "epoch": 0.4811866253247297,
      "grad_norm": 0.16960889738202675,
      "learning_rate": 7.647651086297835e-10,
      "loss": 0.8538,
      "step": 2871
    },
    {
      "epoch": 0.4813542277717255,
      "grad_norm": 0.17159744705771657,
      "learning_rate": 7.514237778162735e-10,
      "loss": 0.8425,
      "step": 2872
    },
    {
      "epoch": 0.4815218302187212,
      "grad_norm": 0.14553944144957537,
      "learning_rate": 7.381994041951944e-10,
      "loss": 0.8316,
      "step": 2873
    },
    {
      "epoch": 0.4816894326657169,
      "grad_norm": 0.18130310931306365,
      "learning_rate": 7.250920033504915e-10,
      "loss": 0.8781,
      "step": 2874
    },
    {
      "epoch": 0.48185703511271266,
      "grad_norm": 0.18600933444717235,
      "learning_rate": 7.121015907282979e-10,
      "loss": 0.9093,
      "step": 2875
    },
    {
      "epoch": 0.48202463755970837,
      "grad_norm": 0.17061362128077523,
      "learning_rate": 6.992281816368573e-10,
      "loss": 0.8595,
      "step": 2876
    },
    {
      "epoch": 0.4821922400067041,
      "grad_norm": 0.18417160179251063,
      "learning_rate": 6.864717912465345e-10,
      "loss": 0.8514,
      "step": 2877
    },
    {
      "epoch": 0.48235984245369984,
      "grad_norm": 0.1598372035823503,
      "learning_rate": 6.73832434589805e-10,
      "loss": 0.8516,
      "step": 2878
    },
    {
      "epoch": 0.48252744490069555,
      "grad_norm": 0.1811705404946187,
      "learning_rate": 6.61310126561243e-10,
      "loss": 0.9069,
      "step": 2879
    },
    {
      "epoch": 0.48269504734769125,
      "grad_norm": 0.18103410542497042,
      "learning_rate": 6.489048819174559e-10,
      "loss": 0.8562,
      "step": 2880
    },
    {
      "epoch": 0.482862649794687,
      "grad_norm": 0.16060803886761177,
      "learning_rate": 6.366167152771274e-10,
      "loss": 0.8491,
      "step": 2881
    },
    {
      "epoch": 0.4830302522416827,
      "grad_norm": 0.17255818578531015,
      "learning_rate": 6.24445641120952e-10,
      "loss": 0.8511,
      "step": 2882
    },
    {
      "epoch": 0.48319785468867843,
      "grad_norm": 0.15887174919570507,
      "learning_rate": 6.123916737916568e-10,
      "loss": 0.8605,
      "step": 2883
    },
    {
      "epoch": 0.4833654571356742,
      "grad_norm": 0.16576046493938676,
      "learning_rate": 6.004548274939681e-10,
      "loss": 0.8528,
      "step": 2884
    },
    {
      "epoch": 0.4835330595826699,
      "grad_norm": 0.19353717973784634,
      "learning_rate": 5.88635116294589e-10,
      "loss": 0.8974,
      "step": 2885
    },
    {
      "epoch": 0.4837006620296656,
      "grad_norm": 0.16725603498814323,
      "learning_rate": 5.76932554122167e-10,
      "loss": 0.8538,
      "step": 2886
    },
    {
      "epoch": 0.48386826447666137,
      "grad_norm": 0.16718514137957122,
      "learning_rate": 5.65347154767326e-10,
      "loss": 0.8646,
      "step": 2887
    },
    {
      "epoch": 0.4840358669236571,
      "grad_norm": 0.17635009574871505,
      "learning_rate": 5.53878931882612e-10,
      "loss": 0.8853,
      "step": 2888
    },
    {
      "epoch": 0.48420346937065284,
      "grad_norm": 0.1915271926798497,
      "learning_rate": 5.425278989824922e-10,
      "loss": 0.8821,
      "step": 2889
    },
    {
      "epoch": 0.48437107181764855,
      "grad_norm": 0.1742560959612863,
      "learning_rate": 5.312940694433333e-10,
      "loss": 0.8928,
      "step": 2890
    },
    {
      "epoch": 0.48453867426464425,
      "grad_norm": 0.18396835429229907,
      "learning_rate": 5.201774565033679e-10,
      "loss": 0.9225,
      "step": 2891
    },
    {
      "epoch": 0.48470627671164,
      "grad_norm": 0.18574695522647786,
      "learning_rate": 5.091780732627061e-10,
      "loss": 0.9283,
      "step": 2892
    },
    {
      "epoch": 0.4848738791586357,
      "grad_norm": 0.18773550003469733,
      "learning_rate": 4.982959326833347e-10,
      "loss": 0.8996,
      "step": 2893
    },
    {
      "epoch": 0.48504148160563143,
      "grad_norm": 0.16782444750024178,
      "learning_rate": 4.875310475890626e-10,
      "loss": 0.8704,
      "step": 2894
    },
    {
      "epoch": 0.4852090840526272,
      "grad_norm": 0.15818465673138565,
      "learning_rate": 4.76883430665509e-10,
      "loss": 0.8598,
      "step": 2895
    },
    {
      "epoch": 0.4853766864996229,
      "grad_norm": 0.17702513018439348,
      "learning_rate": 4.663530944601258e-10,
      "loss": 0.9052,
      "step": 2896
    },
    {
      "epoch": 0.4855442889466186,
      "grad_norm": 0.17927054339138335,
      "learning_rate": 4.559400513821199e-10,
      "loss": 0.8919,
      "step": 2897
    },
    {
      "epoch": 0.48571189139361437,
      "grad_norm": 0.18053873647743046,
      "learning_rate": 4.456443137025201e-10,
      "loss": 0.8579,
      "step": 2898
    },
    {
      "epoch": 0.4858794938406101,
      "grad_norm": 0.15477164680687258,
      "learning_rate": 4.354658935540989e-10,
      "loss": 0.8401,
      "step": 2899
    },
    {
      "epoch": 0.4860470962876058,
      "grad_norm": 0.1591099825602417,
      "learning_rate": 4.254048029313728e-10,
      "loss": 0.8613,
      "step": 2900
    },
    {
      "epoch": 0.48621469873460155,
      "grad_norm": 0.17220891817037304,
      "learning_rate": 4.1546105369061334e-10,
      "loss": 0.8387,
      "step": 2901
    },
    {
      "epoch": 0.48638230118159725,
      "grad_norm": 0.17645461377803398,
      "learning_rate": 4.0563465754979156e-10,
      "loss": 0.9252,
      "step": 2902
    },
    {
      "epoch": 0.48654990362859296,
      "grad_norm": 0.17332562017084266,
      "learning_rate": 3.9592562608861125e-10,
      "loss": 0.8386,
      "step": 2903
    },
    {
      "epoch": 0.4867175060755887,
      "grad_norm": 0.17029930227340065,
      "learning_rate": 3.863339707484314e-10,
      "loss": 0.8247,
      "step": 2904
    },
    {
      "epoch": 0.48688510852258443,
      "grad_norm": 0.17536725274393086,
      "learning_rate": 3.768597028323439e-10,
      "loss": 0.911,
      "step": 2905
    },
    {
      "epoch": 0.48705271096958014,
      "grad_norm": 0.18150712270245276,
      "learning_rate": 3.6750283350506226e-10,
      "loss": 0.9204,
      "step": 2906
    },
    {
      "epoch": 0.4872203134165759,
      "grad_norm": 0.17371730444149563,
      "learning_rate": 3.5826337379297743e-10,
      "loss": 0.8977,
      "step": 2907
    },
    {
      "epoch": 0.4873879158635716,
      "grad_norm": 0.19233814666470958,
      "learning_rate": 3.491413345841354e-10,
      "loss": 0.8985,
      "step": 2908
    },
    {
      "epoch": 0.4875555183105673,
      "grad_norm": 0.19117405830926062,
      "learning_rate": 3.401367266281596e-10,
      "loss": 0.8658,
      "step": 2909
    },
    {
      "epoch": 0.4877231207575631,
      "grad_norm": 0.1539239541086167,
      "learning_rate": 3.312495605363397e-10,
      "loss": 0.8671,
      "step": 2910
    },
    {
      "epoch": 0.4878907232045588,
      "grad_norm": 0.17641749639631527,
      "learning_rate": 3.2247984678155373e-10,
      "loss": 0.9001,
      "step": 2911
    },
    {
      "epoch": 0.4880583256515545,
      "grad_norm": 0.19349742065424472,
      "learning_rate": 3.138275956982683e-10,
      "loss": 0.9017,
      "step": 2912
    },
    {
      "epoch": 0.48822592809855025,
      "grad_norm": 0.16915684010343982,
      "learning_rate": 3.0529281748252754e-10,
      "loss": 0.8938,
      "step": 2913
    },
    {
      "epoch": 0.48839353054554596,
      "grad_norm": 0.15977123900456175,
      "learning_rate": 2.968755221919417e-10,
      "loss": 0.8625,
      "step": 2914
    },
    {
      "epoch": 0.48856113299254167,
      "grad_norm": 0.17913748441880345,
      "learning_rate": 2.885757197456762e-10,
      "loss": 0.8856,
      "step": 2915
    },
    {
      "epoch": 0.48872873543953743,
      "grad_norm": 0.17208218580038487,
      "learning_rate": 2.8039341992444065e-10,
      "loss": 0.8207,
      "step": 2916
    },
    {
      "epoch": 0.48889633788653314,
      "grad_norm": 0.18348075060843572,
      "learning_rate": 2.7232863237048873e-10,
      "loss": 0.9099,
      "step": 2917
    },
    {
      "epoch": 0.48906394033352885,
      "grad_norm": 0.15624206911454094,
      "learning_rate": 2.6438136658757384e-10,
      "loss": 0.8691,
      "step": 2918
    },
    {
      "epoch": 0.4892315427805246,
      "grad_norm": 0.1705877867933035,
      "learning_rate": 2.5655163194097107e-10,
      "loss": 0.8568,
      "step": 2919
    },
    {
      "epoch": 0.4893991452275203,
      "grad_norm": 0.16849393853038944,
      "learning_rate": 2.488394376574332e-10,
      "loss": 0.8519,
      "step": 2920
    },
    {
      "epoch": 0.489566747674516,
      "grad_norm": 0.15713071816735658,
      "learning_rate": 2.412447928252459e-10,
      "loss": 0.8722,
      "step": 2921
    },
    {
      "epoch": 0.4897343501215118,
      "grad_norm": 0.1712448430012013,
      "learning_rate": 2.337677063941279e-10,
      "loss": 0.8665,
      "step": 2922
    },
    {
      "epoch": 0.4899019525685075,
      "grad_norm": 0.18391351981409632,
      "learning_rate": 2.2640818717527543e-10,
      "loss": 0.8666,
      "step": 2923
    },
    {
      "epoch": 0.4900695550155032,
      "grad_norm": 0.16379881207224936,
      "learning_rate": 2.1916624384136217e-10,
      "loss": 0.8279,
      "step": 2924
    },
    {
      "epoch": 0.49023715746249896,
      "grad_norm": 0.17608848662003046,
      "learning_rate": 2.1204188492646157e-10,
      "loss": 0.8924,
      "step": 2925
    },
    {
      "epoch": 0.49040475990949467,
      "grad_norm": 0.1721614707114467,
      "learning_rate": 2.0503511882612457e-10,
      "loss": 0.8858,
      "step": 2926
    },
    {
      "epoch": 0.49057236235649043,
      "grad_norm": 0.17510323228984553,
      "learning_rate": 1.98145953797324e-10,
      "loss": 0.8807,
      "step": 2927
    },
    {
      "epoch": 0.49073996480348614,
      "grad_norm": 0.1830917635653012,
      "learning_rate": 1.9137439795842148e-10,
      "loss": 0.8782,
      "step": 2928
    },
    {
      "epoch": 0.49090756725048185,
      "grad_norm": 0.17433567820483173,
      "learning_rate": 1.8472045928920043e-10,
      "loss": 0.8694,
      "step": 2929
    },
    {
      "epoch": 0.4910751696974776,
      "grad_norm": 0.1937723700014669,
      "learning_rate": 1.7818414563084416e-10,
      "loss": 0.8908,
      "step": 2930
    },
    {
      "epoch": 0.4912427721444733,
      "grad_norm": 0.19881472231275615,
      "learning_rate": 1.7176546468592457e-10,
      "loss": 0.9093,
      "step": 2931
    },
    {
      "epoch": 0.491410374591469,
      "grad_norm": 0.16527511424008978,
      "learning_rate": 1.654644240183911e-10,
      "loss": 0.8408,
      "step": 2932
    },
    {
      "epoch": 0.4915779770384648,
      "grad_norm": 0.15709938083909356,
      "learning_rate": 1.5928103105354862e-10,
      "loss": 0.8661,
      "step": 2933
    },
    {
      "epoch": 0.4917455794854605,
      "grad_norm": 0.16930027535059233,
      "learning_rate": 1.5321529307807944e-10,
      "loss": 0.8917,
      "step": 2934
    },
    {
      "epoch": 0.4919131819324562,
      "grad_norm": 0.1825285469498783,
      "learning_rate": 1.4726721724001024e-10,
      "loss": 0.8913,
      "step": 2935
    },
    {
      "epoch": 0.49208078437945196,
      "grad_norm": 0.1706972363115364,
      "learning_rate": 1.414368105487229e-10,
      "loss": 0.8528,
      "step": 2936
    },
    {
      "epoch": 0.49224838682644767,
      "grad_norm": 0.1627389814918238,
      "learning_rate": 1.3572407987491041e-10,
      "loss": 0.8596,
      "step": 2937
    },
    {
      "epoch": 0.4924159892734434,
      "grad_norm": 0.1837384031906574,
      "learning_rate": 1.3012903195062097e-10,
      "loss": 0.8758,
      "step": 2938
    },
    {
      "epoch": 0.49258359172043914,
      "grad_norm": 0.19454706428777696,
      "learning_rate": 1.246516733692027e-10,
      "loss": 0.8915,
      "step": 2939
    },
    {
      "epoch": 0.49275119416743485,
      "grad_norm": 0.16829161381402702,
      "learning_rate": 1.1929201058532569e-10,
      "loss": 0.8426,
      "step": 2940
    },
    {
      "epoch": 0.49291879661443055,
      "grad_norm": 0.18069595951370082,
      "learning_rate": 1.1405004991495993e-10,
      "loss": 0.9037,
      "step": 2941
    },
    {
      "epoch": 0.4930863990614263,
      "grad_norm": 0.1667719344874642,
      "learning_rate": 1.0892579753536413e-10,
      "loss": 0.874,
      "step": 2942
    },
    {
      "epoch": 0.493254001508422,
      "grad_norm": 0.17324927947457816,
      "learning_rate": 1.0391925948511904e-10,
      "loss": 0.8662,
      "step": 2943
    },
    {
      "epoch": 0.49342160395541773,
      "grad_norm": 0.1838388752804786,
      "learning_rate": 9.903044166402752e-11,
      "loss": 0.8821,
      "step": 2944
    },
    {
      "epoch": 0.4935892064024135,
      "grad_norm": 0.17750126496098492,
      "learning_rate": 9.425934983323668e-11,
      "loss": 0.8988,
      "step": 2945
    },
    {
      "epoch": 0.4937568088494092,
      "grad_norm": 0.16674035140396565,
      "learning_rate": 8.960598961511578e-11,
      "loss": 0.8603,
      "step": 2946
    },
    {
      "epoch": 0.4939244112964049,
      "grad_norm": 0.16182923594996834,
      "learning_rate": 8.507036649331167e-11,
      "loss": 0.8724,
      "step": 2947
    },
    {
      "epoch": 0.49409201374340067,
      "grad_norm": 0.20076003710042253,
      "learning_rate": 8.065248581272666e-11,
      "loss": 0.9171,
      "step": 2948
    },
    {
      "epoch": 0.4942596161903964,
      "grad_norm": 0.18040039525777585,
      "learning_rate": 7.635235277950735e-11,
      "loss": 0.8746,
      "step": 2949
    },
    {
      "epoch": 0.4944272186373921,
      "grad_norm": 0.19247336420631492,
      "learning_rate": 7.216997246104473e-11,
      "loss": 0.9304,
      "step": 2950
    },
    {
      "epoch": 0.49459482108438785,
      "grad_norm": 0.1933656428679124,
      "learning_rate": 6.810534978597404e-11,
      "loss": 0.9008,
      "step": 2951
    },
    {
      "epoch": 0.49476242353138356,
      "grad_norm": 0.16208639385655532,
      "learning_rate": 6.41584895441416e-11,
      "loss": 0.8812,
      "step": 2952
    },
    {
      "epoch": 0.49493002597837926,
      "grad_norm": 0.17378305345925674,
      "learning_rate": 6.032939638664914e-11,
      "loss": 0.8722,
      "step": 2953
    },
    {
      "epoch": 0.495097628425375,
      "grad_norm": 0.15384190656985877,
      "learning_rate": 5.661807482580938e-11,
      "loss": 0.8346,
      "step": 2954
    },
    {
      "epoch": 0.49526523087237073,
      "grad_norm": 0.155375646593035,
      "learning_rate": 5.30245292351128e-11,
      "loss": 0.8794,
      "step": 2955
    },
    {
      "epoch": 0.49543283331936644,
      "grad_norm": 0.18029750623999521,
      "learning_rate": 4.954876384932749e-11,
      "loss": 0.8841,
      "step": 2956
    },
    {
      "epoch": 0.4956004357663622,
      "grad_norm": 0.19576339607777393,
      "learning_rate": 4.619078276436594e-11,
      "loss": 0.8677,
      "step": 2957
    },
    {
      "epoch": 0.4957680382133579,
      "grad_norm": 0.18265498931301835,
      "learning_rate": 4.295058993736278e-11,
      "loss": 0.8655,
      "step": 2958
    },
    {
      "epoch": 0.4959356406603536,
      "grad_norm": 0.1770778755817075,
      "learning_rate": 3.982818918665254e-11,
      "loss": 0.8652,
      "step": 2959
    },
    {
      "epoch": 0.4961032431073494,
      "grad_norm": 0.18666816649844864,
      "learning_rate": 3.682358419176967e-11,
      "loss": 0.8819,
      "step": 2960
    },
    {
      "epoch": 0.4962708455543451,
      "grad_norm": 0.1554804789330606,
      "learning_rate": 3.393677849340415e-11,
      "loss": 0.8416,
      "step": 2961
    },
    {
      "epoch": 0.4964384480013408,
      "grad_norm": 0.1876621599451238,
      "learning_rate": 3.1167775493445847e-11,
      "loss": 0.9059,
      "step": 2962
    },
    {
      "epoch": 0.49660605044833656,
      "grad_norm": 0.176330013906115,
      "learning_rate": 2.8516578454973462e-11,
      "loss": 0.8553,
      "step": 2963
    },
    {
      "epoch": 0.49677365289533226,
      "grad_norm": 0.15228259979966144,
      "learning_rate": 2.5983190502221197e-11,
      "loss": 0.8542,
      "step": 2964
    },
    {
      "epoch": 0.496941255342328,
      "grad_norm": 0.14830596234173596,
      "learning_rate": 2.3567614620600974e-11,
      "loss": 0.8159,
      "step": 2965
    },
    {
      "epoch": 0.49710885778932373,
      "grad_norm": 0.1576744172221426,
      "learning_rate": 2.1269853656702417e-11,
      "loss": 0.8643,
      "step": 2966
    },
    {
      "epoch": 0.49727646023631944,
      "grad_norm": 0.17493490108216514,
      "learning_rate": 1.908991031825957e-11,
      "loss": 0.8796,
      "step": 2967
    },
    {
      "epoch": 0.4974440626833152,
      "grad_norm": 0.17650076754831764,
      "learning_rate": 1.702778717417308e-11,
      "loss": 0.851,
      "step": 2968
    },
    {
      "epoch": 0.4976116651303109,
      "grad_norm": 0.18015793491192852,
      "learning_rate": 1.50834866545102e-11,
      "loss": 0.8647,
      "step": 2969
    },
    {
      "epoch": 0.4977792675773066,
      "grad_norm": 0.169189104078413,
      "learning_rate": 1.3257011050493705e-11,
      "loss": 0.8549,
      "step": 2970
    },
    {
      "epoch": 0.4979468700243024,
      "grad_norm": 0.18929665382052452,
      "learning_rate": 1.1548362514479659e-11,
      "loss": 0.876,
      "step": 2971
    },
    {
      "epoch": 0.4981144724712981,
      "grad_norm": 0.16145798248133275,
      "learning_rate": 9.957543059979645e-12,
      "loss": 0.8442,
      "step": 2972
    },
    {
      "epoch": 0.4982820749182938,
      "grad_norm": 0.18764045204075666,
      "learning_rate": 8.484554561682955e-12,
      "loss": 0.9043,
      "step": 2973
    },
    {
      "epoch": 0.49844967736528956,
      "grad_norm": 0.16780072296511686,
      "learning_rate": 7.129398755378879e-12,
      "loss": 0.8599,
      "step": 2974
    },
    {
      "epoch": 0.49861727981228526,
      "grad_norm": 0.18513500701390914,
      "learning_rate": 5.892077238023319e-12,
      "loss": 0.8809,
      "step": 2975
    },
    {
      "epoch": 0.49878488225928097,
      "grad_norm": 0.16959613485475766,
      "learning_rate": 4.772591467705478e-12,
      "loss": 0.9074,
      "step": 2976
    },
    {
      "epoch": 0.49895248470627673,
      "grad_norm": 0.16024081746045518,
      "learning_rate": 3.770942763670071e-12,
      "loss": 0.8633,
      "step": 2977
    },
    {
      "epoch": 0.49912008715327244,
      "grad_norm": 0.16217344776246936,
      "learning_rate": 2.8871323062840125e-12,
      "loss": 0.8816,
      "step": 2978
    },
    {
      "epoch": 0.49928768960026815,
      "grad_norm": 0.18492581809506337,
      "learning_rate": 2.1211611370475226e-12,
      "loss": 0.9158,
      "step": 2979
    },
    {
      "epoch": 0.4994552920472639,
      "grad_norm": 0.1966774666977336,
      "learning_rate": 1.473030158594124e-12,
      "loss": 0.8738,
      "step": 2980
    },
    {
      "epoch": 0.4996228944942596,
      "grad_norm": 0.17134510760359373,
      "learning_rate": 9.427401347128494e-13,
      "loss": 0.881,
      "step": 2981
    },
    {
      "epoch": 0.4997904969412553,
      "grad_norm": 0.18897466687207629,
      "learning_rate": 5.302916903038302e-13,
      "loss": 0.8816,
      "step": 2982
    },
    {
      "epoch": 0.4999580993882511,
      "grad_norm": 0.17357883227373952,
      "learning_rate": 2.356853114116042e-13,
      "loss": 0.8469,
      "step": 2983
    },
    {
      "epoch": 0.5001257018352468,
      "grad_norm": 0.17167124251127178,
      "learning_rate": 5.892134521401359e-14,
      "loss": 0.8907,
      "step": 2984
    },
    {
      "epoch": 0.5001257018352468,
      "step": 2984,
      "total_flos": 178280838049792.0,
      "train_loss": 1.0106905110120454,
      "train_runtime": 35875.3338,
      "train_samples_per_second": 2.661,
      "train_steps_per_second": 0.083
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2984,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 178280838049792.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}