{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9998900252941824,
  "eval_steps": 500,
  "global_step": 4546,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00021994941163532388,
      "grad_norm": 0.7960259914398193,
      "learning_rate": 8e-05,
      "loss": 2.2745,
      "step": 1
    },
    {
      "epoch": 0.00043989882327064776,
      "grad_norm": 0.3960127830505371,
      "learning_rate": 8e-05,
      "loss": 1.9155,
      "step": 2
    },
    {
      "epoch": 0.0006598482349059716,
      "grad_norm": 0.3869185745716095,
      "learning_rate": 8e-05,
      "loss": 1.8754,
      "step": 3
    },
    {
      "epoch": 0.0008797976465412955,
      "grad_norm": 0.33234134316444397,
      "learning_rate": 8e-05,
      "loss": 1.9126,
      "step": 4
    },
    {
      "epoch": 0.0010997470581766194,
      "grad_norm": 0.3670472502708435,
      "learning_rate": 8e-05,
      "loss": 1.9912,
      "step": 5
    },
    {
      "epoch": 0.0013196964698119432,
      "grad_norm": 0.32942938804626465,
      "learning_rate": 8e-05,
      "loss": 1.8746,
      "step": 6
    },
    {
      "epoch": 0.001539645881447267,
      "grad_norm": 0.3588086664676666,
      "learning_rate": 8e-05,
      "loss": 1.9545,
      "step": 7
    },
    {
      "epoch": 0.001759595293082591,
      "grad_norm": 0.33002492785453796,
      "learning_rate": 8e-05,
      "loss": 1.8781,
      "step": 8
    },
    {
      "epoch": 0.001979544704717915,
      "grad_norm": 0.3024381101131439,
      "learning_rate": 8e-05,
      "loss": 1.8859,
      "step": 9
    },
    {
      "epoch": 0.002199494116353239,
      "grad_norm": 0.3224199712276459,
      "learning_rate": 8e-05,
      "loss": 1.829,
      "step": 10
    },
    {
      "epoch": 0.0024194435279885624,
      "grad_norm": 0.31481102108955383,
      "learning_rate": 8e-05,
      "loss": 1.817,
      "step": 11
    },
    {
      "epoch": 0.0026393929396238865,
      "grad_norm": 0.3078259825706482,
      "learning_rate": 8e-05,
      "loss": 1.9631,
      "step": 12
    },
    {
      "epoch": 0.0028593423512592105,
      "grad_norm": 0.3141743540763855,
      "learning_rate": 8e-05,
      "loss": 1.8879,
      "step": 13
    },
    {
      "epoch": 0.003079291762894534,
      "grad_norm": 0.29574745893478394,
      "learning_rate": 8e-05,
      "loss": 1.799,
      "step": 14
    },
    {
      "epoch": 0.003299241174529858,
      "grad_norm": 0.3095031976699829,
      "learning_rate": 8e-05,
      "loss": 1.8741,
      "step": 15
    },
    {
      "epoch": 0.003519190586165182,
      "grad_norm": 0.28804337978363037,
      "learning_rate": 8e-05,
      "loss": 1.9931,
      "step": 16
    },
    {
      "epoch": 0.0037391399978005057,
      "grad_norm": 0.25137585401535034,
      "learning_rate": 8e-05,
      "loss": 1.6762,
      "step": 17
    },
    {
      "epoch": 0.00395908940943583,
      "grad_norm": 0.28565698862075806,
      "learning_rate": 8e-05,
      "loss": 1.8489,
      "step": 18
    },
    {
      "epoch": 0.004179038821071153,
      "grad_norm": 0.2877500355243683,
      "learning_rate": 8e-05,
      "loss": 1.8871,
      "step": 19
    },
    {
      "epoch": 0.004398988232706478,
      "grad_norm": 0.28803154826164246,
      "learning_rate": 8e-05,
      "loss": 1.6956,
      "step": 20
    },
    {
      "epoch": 0.004618937644341801,
      "grad_norm": 0.32161301374435425,
      "learning_rate": 8e-05,
      "loss": 1.8243,
      "step": 21
    },
    {
      "epoch": 0.004838887055977125,
      "grad_norm": 0.2959391176700592,
      "learning_rate": 8e-05,
      "loss": 1.8991,
      "step": 22
    },
    {
      "epoch": 0.005058836467612449,
      "grad_norm": 0.3021189868450165,
      "learning_rate": 8e-05,
      "loss": 1.9975,
      "step": 23
    },
    {
      "epoch": 0.005278785879247773,
      "grad_norm": 0.2793104946613312,
      "learning_rate": 8e-05,
      "loss": 1.8792,
      "step": 24
    },
    {
      "epoch": 0.0054987352908830965,
      "grad_norm": 0.2658381760120392,
      "learning_rate": 8e-05,
      "loss": 1.6467,
      "step": 25
    },
    {
      "epoch": 0.005718684702518421,
      "grad_norm": 0.2793010175228119,
      "learning_rate": 8e-05,
      "loss": 1.7479,
      "step": 26
    },
    {
      "epoch": 0.0059386341141537445,
      "grad_norm": 0.2800044119358063,
      "learning_rate": 8e-05,
      "loss": 1.7885,
      "step": 27
    },
    {
      "epoch": 0.006158583525789068,
      "grad_norm": 0.2864585220813751,
      "learning_rate": 8e-05,
      "loss": 1.9257,
      "step": 28
    },
    {
      "epoch": 0.006378532937424393,
      "grad_norm": 0.301496684551239,
      "learning_rate": 8e-05,
      "loss": 1.8586,
      "step": 29
    },
    {
      "epoch": 0.006598482349059716,
      "grad_norm": 0.2858293354511261,
      "learning_rate": 8e-05,
      "loss": 1.8541,
      "step": 30
    },
    {
      "epoch": 0.00681843176069504,
      "grad_norm": 0.31271278858184814,
      "learning_rate": 8e-05,
      "loss": 1.8774,
      "step": 31
    },
    {
      "epoch": 0.007038381172330364,
      "grad_norm": 0.30428266525268555,
      "learning_rate": 8e-05,
      "loss": 1.8464,
      "step": 32
    },
    {
      "epoch": 0.007258330583965688,
      "grad_norm": 0.26637139916419983,
      "learning_rate": 8e-05,
      "loss": 1.7896,
      "step": 33
    },
    {
      "epoch": 0.007478279995601011,
      "grad_norm": 0.2802716791629791,
      "learning_rate": 8e-05,
      "loss": 1.9534,
      "step": 34
    },
    {
      "epoch": 0.007698229407236336,
      "grad_norm": 0.35580113530158997,
      "learning_rate": 8e-05,
      "loss": 1.8236,
      "step": 35
    },
    {
      "epoch": 0.00791817881887166,
      "grad_norm": 0.2794848382472992,
      "learning_rate": 8e-05,
      "loss": 1.8089,
      "step": 36
    },
    {
      "epoch": 0.008138128230506983,
      "grad_norm": 0.27942711114883423,
      "learning_rate": 8e-05,
      "loss": 1.7725,
      "step": 37
    },
    {
      "epoch": 0.008358077642142307,
      "grad_norm": 0.2882610857486725,
      "learning_rate": 8e-05,
      "loss": 1.8279,
      "step": 38
    },
    {
      "epoch": 0.008578027053777632,
      "grad_norm": 0.29375842213630676,
      "learning_rate": 8e-05,
      "loss": 2.0123,
      "step": 39
    },
    {
      "epoch": 0.008797976465412955,
      "grad_norm": 0.26120567321777344,
      "learning_rate": 8e-05,
      "loss": 1.6731,
      "step": 40
    },
    {
      "epoch": 0.009017925877048279,
      "grad_norm": 0.25272971391677856,
      "learning_rate": 8e-05,
      "loss": 1.5723,
      "step": 41
    },
    {
      "epoch": 0.009237875288683603,
      "grad_norm": 0.30548569560050964,
      "learning_rate": 8e-05,
      "loss": 1.9269,
      "step": 42
    },
    {
      "epoch": 0.009457824700318926,
      "grad_norm": 0.2714739441871643,
      "learning_rate": 8e-05,
      "loss": 1.6715,
      "step": 43
    },
    {
      "epoch": 0.00967777411195425,
      "grad_norm": 0.3086313009262085,
      "learning_rate": 8e-05,
      "loss": 1.8903,
      "step": 44
    },
    {
      "epoch": 0.009897723523589575,
      "grad_norm": 0.28676554560661316,
      "learning_rate": 8e-05,
      "loss": 1.8257,
      "step": 45
    },
    {
      "epoch": 0.010117672935224899,
      "grad_norm": 0.2898331880569458,
      "learning_rate": 8e-05,
      "loss": 1.822,
      "step": 46
    },
    {
      "epoch": 0.010337622346860222,
      "grad_norm": 0.2887754440307617,
      "learning_rate": 8e-05,
      "loss": 1.7629,
      "step": 47
    },
    {
      "epoch": 0.010557571758495546,
      "grad_norm": 0.28026437759399414,
      "learning_rate": 8e-05,
      "loss": 1.8874,
      "step": 48
    },
    {
      "epoch": 0.01077752117013087,
      "grad_norm": 0.29256439208984375,
      "learning_rate": 8e-05,
      "loss": 1.9169,
      "step": 49
    },
    {
      "epoch": 0.010997470581766193,
      "grad_norm": 0.29388460516929626,
      "learning_rate": 8e-05,
      "loss": 1.8341,
      "step": 50
    },
    {
      "epoch": 0.011217419993401518,
      "grad_norm": 0.29456326365470886,
      "learning_rate": 8e-05,
      "loss": 1.7088,
      "step": 51
    },
    {
      "epoch": 0.011437369405036842,
      "grad_norm": 0.2810533046722412,
      "learning_rate": 8e-05,
      "loss": 1.8564,
      "step": 52
    },
    {
      "epoch": 0.011657318816672166,
      "grad_norm": 0.3049224019050598,
      "learning_rate": 8e-05,
      "loss": 2.0114,
      "step": 53
    },
    {
      "epoch": 0.011877268228307489,
      "grad_norm": 0.347817987203598,
      "learning_rate": 8e-05,
      "loss": 1.6936,
      "step": 54
    },
    {
      "epoch": 0.012097217639942813,
      "grad_norm": 0.28999242186546326,
      "learning_rate": 8e-05,
      "loss": 1.5852,
      "step": 55
    },
    {
      "epoch": 0.012317167051578136,
      "grad_norm": 0.32856103777885437,
      "learning_rate": 8e-05,
      "loss": 1.8249,
      "step": 56
    },
    {
      "epoch": 0.012537116463213462,
      "grad_norm": 0.3450610339641571,
      "learning_rate": 8e-05,
      "loss": 1.9309,
      "step": 57
    },
    {
      "epoch": 0.012757065874848785,
      "grad_norm": 0.27445971965789795,
      "learning_rate": 8e-05,
      "loss": 1.8153,
      "step": 58
    },
    {
      "epoch": 0.012977015286484109,
      "grad_norm": 0.28595077991485596,
      "learning_rate": 8e-05,
      "loss": 1.8061,
      "step": 59
    },
    {
      "epoch": 0.013196964698119432,
      "grad_norm": 0.2909082770347595,
      "learning_rate": 8e-05,
      "loss": 1.7793,
      "step": 60
    },
    {
      "epoch": 0.013416914109754756,
      "grad_norm": 0.28822049498558044,
      "learning_rate": 8e-05,
      "loss": 1.7218,
      "step": 61
    },
    {
      "epoch": 0.01363686352139008,
      "grad_norm": 0.29159948229789734,
      "learning_rate": 8e-05,
      "loss": 1.7985,
      "step": 62
    },
    {
      "epoch": 0.013856812933025405,
      "grad_norm": 0.29802417755126953,
      "learning_rate": 8e-05,
      "loss": 1.8903,
      "step": 63
    },
    {
      "epoch": 0.014076762344660728,
      "grad_norm": 0.29128944873809814,
      "learning_rate": 8e-05,
      "loss": 1.7928,
      "step": 64
    },
    {
      "epoch": 0.014296711756296052,
      "grad_norm": 0.3093227446079254,
      "learning_rate": 8e-05,
      "loss": 1.8409,
      "step": 65
    },
    {
      "epoch": 0.014516661167931376,
      "grad_norm": 0.2688956558704376,
      "learning_rate": 8e-05,
      "loss": 1.6136,
      "step": 66
    },
    {
      "epoch": 0.0147366105795667,
      "grad_norm": 0.316579133272171,
      "learning_rate": 8e-05,
      "loss": 1.8153,
      "step": 67
    },
    {
      "epoch": 0.014956559991202023,
      "grad_norm": 0.30234795808792114,
      "learning_rate": 8e-05,
      "loss": 1.7311,
      "step": 68
    },
    {
      "epoch": 0.015176509402837348,
      "grad_norm": 0.2790556848049164,
      "learning_rate": 8e-05,
      "loss": 1.7714,
      "step": 69
    },
    {
      "epoch": 0.015396458814472672,
      "grad_norm": 0.29012972116470337,
      "learning_rate": 8e-05,
      "loss": 1.7528,
      "step": 70
    },
    {
      "epoch": 0.015616408226107995,
      "grad_norm": 0.28507527709007263,
      "learning_rate": 8e-05,
      "loss": 1.6452,
      "step": 71
    },
    {
      "epoch": 0.01583635763774332,
      "grad_norm": 0.28862133622169495,
      "learning_rate": 8e-05,
      "loss": 1.7473,
      "step": 72
    },
    {
      "epoch": 0.016056307049378642,
      "grad_norm": 0.2726048231124878,
      "learning_rate": 8e-05,
      "loss": 1.7519,
      "step": 73
    },
    {
      "epoch": 0.016276256461013966,
      "grad_norm": 0.26808786392211914,
      "learning_rate": 8e-05,
      "loss": 1.6332,
      "step": 74
    },
    {
      "epoch": 0.01649620587264929,
      "grad_norm": 0.32144519686698914,
      "learning_rate": 8e-05,
      "loss": 1.7115,
      "step": 75
    },
    {
      "epoch": 0.016716155284284613,
      "grad_norm": 0.26930421590805054,
      "learning_rate": 8e-05,
      "loss": 1.7854,
      "step": 76
    },
    {
      "epoch": 0.016936104695919937,
      "grad_norm": 0.29462486505508423,
      "learning_rate": 8e-05,
      "loss": 1.6919,
      "step": 77
    },
    {
      "epoch": 0.017156054107555264,
      "grad_norm": 0.2780003249645233,
      "learning_rate": 8e-05,
      "loss": 1.6355,
      "step": 78
    },
    {
      "epoch": 0.017376003519190587,
      "grad_norm": 0.29219016432762146,
      "learning_rate": 8e-05,
      "loss": 1.883,
      "step": 79
    },
    {
      "epoch": 0.01759595293082591,
      "grad_norm": 0.2893241047859192,
      "learning_rate": 8e-05,
      "loss": 1.8548,
      "step": 80
    },
    {
      "epoch": 0.017815902342461234,
      "grad_norm": 0.283512145280838,
      "learning_rate": 8e-05,
      "loss": 1.79,
      "step": 81
    },
    {
      "epoch": 0.018035851754096558,
      "grad_norm": 0.2679024040699005,
      "learning_rate": 8e-05,
      "loss": 1.5866,
      "step": 82
    },
    {
      "epoch": 0.01825580116573188,
      "grad_norm": 0.2892123758792877,
      "learning_rate": 8e-05,
      "loss": 1.9033,
      "step": 83
    },
    {
      "epoch": 0.018475750577367205,
      "grad_norm": 0.2680201530456543,
      "learning_rate": 8e-05,
      "loss": 1.8557,
      "step": 84
    },
    {
      "epoch": 0.01869569998900253,
      "grad_norm": 0.30922645330429077,
      "learning_rate": 8e-05,
      "loss": 1.8885,
      "step": 85
    },
    {
      "epoch": 0.018915649400637852,
      "grad_norm": 0.2735271751880646,
      "learning_rate": 8e-05,
      "loss": 1.8765,
      "step": 86
    },
    {
      "epoch": 0.019135598812273176,
      "grad_norm": 0.28639712929725647,
      "learning_rate": 8e-05,
      "loss": 1.9429,
      "step": 87
    },
    {
      "epoch": 0.0193555482239085,
      "grad_norm": 0.28437235951423645,
      "learning_rate": 8e-05,
      "loss": 1.8405,
      "step": 88
    },
    {
      "epoch": 0.019575497635543827,
      "grad_norm": 0.276517778635025,
      "learning_rate": 8e-05,
      "loss": 1.7496,
      "step": 89
    },
    {
      "epoch": 0.01979544704717915,
      "grad_norm": 0.273404598236084,
      "learning_rate": 8e-05,
      "loss": 1.704,
      "step": 90
    },
    {
      "epoch": 0.020015396458814474,
      "grad_norm": 0.2707740366458893,
      "learning_rate": 8e-05,
      "loss": 1.8274,
      "step": 91
    },
    {
      "epoch": 0.020235345870449797,
      "grad_norm": 0.26880595088005066,
      "learning_rate": 8e-05,
      "loss": 1.7695,
      "step": 92
    },
    {
      "epoch": 0.02045529528208512,
      "grad_norm": 0.28712528944015503,
      "learning_rate": 8e-05,
      "loss": 1.9436,
      "step": 93
    },
    {
      "epoch": 0.020675244693720445,
      "grad_norm": 0.26633599400520325,
      "learning_rate": 8e-05,
      "loss": 1.7877,
      "step": 94
    },
    {
      "epoch": 0.020895194105355768,
      "grad_norm": 0.2843431532382965,
      "learning_rate": 8e-05,
      "loss": 1.8389,
      "step": 95
    },
    {
      "epoch": 0.02111514351699109,
      "grad_norm": 0.2597465515136719,
      "learning_rate": 8e-05,
      "loss": 1.7047,
      "step": 96
    },
    {
      "epoch": 0.021335092928626415,
      "grad_norm": 0.2804902493953705,
      "learning_rate": 8e-05,
      "loss": 1.9375,
      "step": 97
    },
    {
      "epoch": 0.02155504234026174,
      "grad_norm": 0.2825285792350769,
      "learning_rate": 8e-05,
      "loss": 1.8348,
      "step": 98
    },
    {
      "epoch": 0.021774991751897062,
      "grad_norm": 0.26459112763404846,
      "learning_rate": 8e-05,
      "loss": 1.7416,
      "step": 99
    },
    {
      "epoch": 0.021994941163532386,
      "grad_norm": 0.28523096442222595,
      "learning_rate": 8e-05,
      "loss": 1.9202,
      "step": 100
    },
    {
      "epoch": 0.022214890575167713,
      "grad_norm": 0.2679818570613861,
      "learning_rate": 8e-05,
      "loss": 1.6741,
      "step": 101
    },
    {
      "epoch": 0.022434839986803037,
      "grad_norm": 0.2798464894294739,
      "learning_rate": 8e-05,
      "loss": 1.6622,
      "step": 102
    },
    {
      "epoch": 0.02265478939843836,
      "grad_norm": 0.2826269567012787,
      "learning_rate": 8e-05,
      "loss": 1.7577,
      "step": 103
    },
    {
      "epoch": 0.022874738810073684,
      "grad_norm": 0.3859495222568512,
      "learning_rate": 8e-05,
      "loss": 1.9705,
      "step": 104
    },
    {
      "epoch": 0.023094688221709007,
      "grad_norm": 0.2766650319099426,
      "learning_rate": 8e-05,
      "loss": 1.7706,
      "step": 105
    },
    {
      "epoch": 0.02331463763334433,
      "grad_norm": 0.2804067134857178,
      "learning_rate": 8e-05,
      "loss": 1.8007,
      "step": 106
    },
    {
      "epoch": 0.023534587044979655,
      "grad_norm": 0.27818629145622253,
      "learning_rate": 8e-05,
      "loss": 1.7913,
      "step": 107
    },
    {
      "epoch": 0.023754536456614978,
      "grad_norm": 0.2697458267211914,
      "learning_rate": 8e-05,
      "loss": 1.8458,
      "step": 108
    },
    {
      "epoch": 0.023974485868250302,
      "grad_norm": 0.28805410861968994,
      "learning_rate": 8e-05,
      "loss": 1.7543,
      "step": 109
    },
    {
      "epoch": 0.024194435279885625,
      "grad_norm": 0.28452396392822266,
      "learning_rate": 8e-05,
      "loss": 1.8499,
      "step": 110
    },
    {
      "epoch": 0.02441438469152095,
      "grad_norm": 0.2837978005409241,
      "learning_rate": 8e-05,
      "loss": 1.797,
      "step": 111
    },
    {
      "epoch": 0.024634334103156273,
      "grad_norm": 0.2965853810310364,
      "learning_rate": 8e-05,
      "loss": 1.7988,
      "step": 112
    },
    {
      "epoch": 0.0248542835147916,
      "grad_norm": 0.28529393672943115,
      "learning_rate": 8e-05,
      "loss": 1.7886,
      "step": 113
    },
    {
      "epoch": 0.025074232926426923,
      "grad_norm": 0.285199910402298,
      "learning_rate": 8e-05,
      "loss": 1.9112,
      "step": 114
    },
    {
      "epoch": 0.025294182338062247,
      "grad_norm": 0.286316454410553,
      "learning_rate": 8e-05,
      "loss": 1.6735,
      "step": 115
    },
    {
      "epoch": 0.02551413174969757,
      "grad_norm": 0.2648874819278717,
      "learning_rate": 8e-05,
      "loss": 1.5333,
      "step": 116
    },
    {
      "epoch": 0.025734081161332894,
      "grad_norm": 0.2834017276763916,
      "learning_rate": 8e-05,
      "loss": 1.7524,
      "step": 117
    },
    {
      "epoch": 0.025954030572968217,
      "grad_norm": 0.27846938371658325,
      "learning_rate": 8e-05,
      "loss": 1.8448,
      "step": 118
    },
    {
      "epoch": 0.02617397998460354,
      "grad_norm": 0.3278025984764099,
      "learning_rate": 8e-05,
      "loss": 1.9158,
      "step": 119
    },
    {
      "epoch": 0.026393929396238865,
      "grad_norm": 0.30259498953819275,
      "learning_rate": 8e-05,
      "loss": 1.7897,
      "step": 120
    },
    {
      "epoch": 0.026613878807874188,
      "grad_norm": 0.27566099166870117,
      "learning_rate": 8e-05,
      "loss": 1.682,
      "step": 121
    },
    {
      "epoch": 0.026833828219509512,
      "grad_norm": 0.2959173321723938,
      "learning_rate": 8e-05,
      "loss": 1.9032,
      "step": 122
    },
    {
      "epoch": 0.027053777631144835,
      "grad_norm": 0.29449525475502014,
      "learning_rate": 8e-05,
      "loss": 1.6174,
      "step": 123
    },
    {
      "epoch": 0.02727372704278016,
      "grad_norm": 0.3012568950653076,
      "learning_rate": 8e-05,
      "loss": 1.6817,
      "step": 124
    },
    {
      "epoch": 0.027493676454415486,
      "grad_norm": 0.29086676239967346,
      "learning_rate": 8e-05,
      "loss": 1.833,
      "step": 125
    },
    {
      "epoch": 0.02771362586605081,
      "grad_norm": 0.2756067216396332,
      "learning_rate": 8e-05,
      "loss": 1.7807,
      "step": 126
    },
    {
      "epoch": 0.027933575277686133,
      "grad_norm": 0.3420695662498474,
      "learning_rate": 8e-05,
      "loss": 1.8652,
      "step": 127
    },
    {
      "epoch": 0.028153524689321457,
      "grad_norm": 0.2899749279022217,
      "learning_rate": 8e-05,
      "loss": 1.7199,
      "step": 128
    },
    {
      "epoch": 0.02837347410095678,
      "grad_norm": 0.274718701839447,
      "learning_rate": 8e-05,
      "loss": 1.7322,
      "step": 129
    },
    {
      "epoch": 0.028593423512592104,
      "grad_norm": 0.3784034848213196,
      "learning_rate": 8e-05,
      "loss": 1.8917,
      "step": 130
    },
    {
      "epoch": 0.028813372924227428,
      "grad_norm": 0.2814437448978424,
      "learning_rate": 8e-05,
      "loss": 1.726,
      "step": 131
    },
    {
      "epoch": 0.02903332233586275,
      "grad_norm": 0.287701815366745,
      "learning_rate": 8e-05,
      "loss": 1.8166,
      "step": 132
    },
    {
      "epoch": 0.029253271747498075,
      "grad_norm": 0.28487101197242737,
      "learning_rate": 8e-05,
      "loss": 1.7183,
      "step": 133
    },
    {
      "epoch": 0.0294732211591334,
      "grad_norm": 0.27141597867012024,
      "learning_rate": 8e-05,
      "loss": 1.7436,
      "step": 134
    },
    {
      "epoch": 0.029693170570768722,
      "grad_norm": 0.2708652913570404,
      "learning_rate": 8e-05,
      "loss": 1.8116,
      "step": 135
    },
    {
      "epoch": 0.029913119982404045,
      "grad_norm": 0.2789991796016693,
      "learning_rate": 8e-05,
      "loss": 1.7942,
      "step": 136
    },
    {
      "epoch": 0.030133069394039372,
      "grad_norm": 0.3053725063800812,
      "learning_rate": 8e-05,
      "loss": 1.8508,
      "step": 137
    },
    {
      "epoch": 0.030353018805674696,
      "grad_norm": 0.30432772636413574,
      "learning_rate": 8e-05,
      "loss": 1.8129,
      "step": 138
    },
    {
      "epoch": 0.03057296821731002,
      "grad_norm": 0.2873070240020752,
      "learning_rate": 8e-05,
      "loss": 1.8713,
      "step": 139
    },
    {
      "epoch": 0.030792917628945343,
      "grad_norm": 0.2777135968208313,
      "learning_rate": 8e-05,
      "loss": 1.7065,
      "step": 140
    },
    {
      "epoch": 0.031012867040580667,
      "grad_norm": 0.29774004220962524,
      "learning_rate": 8e-05,
      "loss": 1.6471,
      "step": 141
    },
    {
      "epoch": 0.03123281645221599,
      "grad_norm": 0.2803782522678375,
      "learning_rate": 8e-05,
      "loss": 1.6992,
      "step": 142
    },
    {
      "epoch": 0.03145276586385132,
      "grad_norm": 0.2777007818222046,
      "learning_rate": 8e-05,
      "loss": 1.8398,
      "step": 143
    },
    {
      "epoch": 0.03167271527548664,
      "grad_norm": 0.26938894391059875,
      "learning_rate": 8e-05,
      "loss": 1.6082,
      "step": 144
    },
    {
      "epoch": 0.031892664687121965,
      "grad_norm": 0.2934747338294983,
      "learning_rate": 8e-05,
      "loss": 1.6929,
      "step": 145
    },
    {
      "epoch": 0.032112614098757285,
      "grad_norm": 0.2687772214412689,
      "learning_rate": 8e-05,
      "loss": 1.6472,
      "step": 146
    },
    {
      "epoch": 0.03233256351039261,
      "grad_norm": 0.2758256793022156,
      "learning_rate": 8e-05,
      "loss": 1.7128,
      "step": 147
    },
    {
      "epoch": 0.03255251292202793,
      "grad_norm": 0.26065707206726074,
      "learning_rate": 8e-05,
      "loss": 1.7108,
      "step": 148
    },
    {
      "epoch": 0.03277246233366326,
      "grad_norm": 0.31668898463249207,
      "learning_rate": 8e-05,
      "loss": 1.9365,
      "step": 149
    },
    {
      "epoch": 0.03299241174529858,
      "grad_norm": 0.2915947437286377,
      "learning_rate": 8e-05,
      "loss": 1.855,
      "step": 150
    },
    {
      "epoch": 0.033212361156933906,
      "grad_norm": 0.2741534113883972,
      "learning_rate": 8e-05,
      "loss": 1.735,
      "step": 151
    },
    {
      "epoch": 0.033432310568569226,
      "grad_norm": 0.300800085067749,
      "learning_rate": 8e-05,
      "loss": 1.7161,
      "step": 152
    },
    {
      "epoch": 0.03365225998020455,
      "grad_norm": 0.26691076159477234,
      "learning_rate": 8e-05,
      "loss": 1.665,
      "step": 153
    },
    {
      "epoch": 0.03387220939183987,
      "grad_norm": 0.2605098485946655,
      "learning_rate": 8e-05,
      "loss": 1.7288,
      "step": 154
    },
    {
      "epoch": 0.0340921588034752,
      "grad_norm": 0.2728619873523712,
      "learning_rate": 8e-05,
      "loss": 1.7237,
      "step": 155
    },
    {
      "epoch": 0.03431210821511053,
      "grad_norm": 0.29627877473831177,
      "learning_rate": 8e-05,
      "loss": 1.8024,
      "step": 156
    },
    {
      "epoch": 0.03453205762674585,
      "grad_norm": 0.27106964588165283,
      "learning_rate": 8e-05,
      "loss": 1.8166,
      "step": 157
    },
    {
      "epoch": 0.034752007038381175,
      "grad_norm": 0.26806893944740295,
      "learning_rate": 8e-05,
      "loss": 1.7061,
      "step": 158
    },
    {
      "epoch": 0.034971956450016495,
      "grad_norm": 0.2509767413139343,
      "learning_rate": 8e-05,
      "loss": 1.6897,
      "step": 159
    },
    {
      "epoch": 0.03519190586165182,
      "grad_norm": 0.34342750906944275,
      "learning_rate": 8e-05,
      "loss": 1.7151,
      "step": 160
    },
    {
      "epoch": 0.03541185527328714,
      "grad_norm": 0.27948594093322754,
      "learning_rate": 8e-05,
      "loss": 1.6574,
      "step": 161
    },
    {
      "epoch": 0.03563180468492247,
      "grad_norm": 0.28651687502861023,
      "learning_rate": 8e-05,
      "loss": 1.839,
      "step": 162
    },
    {
      "epoch": 0.03585175409655779,
      "grad_norm": 0.2787701189517975,
      "learning_rate": 8e-05,
      "loss": 1.8146,
      "step": 163
    },
    {
      "epoch": 0.036071703508193116,
      "grad_norm": 0.2596721351146698,
      "learning_rate": 8e-05,
      "loss": 1.6088,
      "step": 164
    },
    {
      "epoch": 0.036291652919828436,
      "grad_norm": 0.2630285322666168,
      "learning_rate": 8e-05,
      "loss": 1.6941,
      "step": 165
    },
    {
      "epoch": 0.03651160233146376,
      "grad_norm": 0.30072465538978577,
      "learning_rate": 8e-05,
      "loss": 1.8684,
      "step": 166
    },
    {
      "epoch": 0.03673155174309909,
      "grad_norm": 0.2789234519004822,
      "learning_rate": 8e-05,
      "loss": 1.9136,
      "step": 167
    },
    {
      "epoch": 0.03695150115473441,
      "grad_norm": 0.25597283244132996,
      "learning_rate": 8e-05,
      "loss": 1.669,
      "step": 168
    },
    {
      "epoch": 0.03717145056636974,
      "grad_norm": 0.30354219675064087,
      "learning_rate": 8e-05,
      "loss": 1.7845,
      "step": 169
    },
    {
      "epoch": 0.03739139997800506,
      "grad_norm": 0.26998043060302734,
      "learning_rate": 8e-05,
      "loss": 1.6626,
      "step": 170
    },
    {
      "epoch": 0.037611349389640385,
      "grad_norm": 0.27418825030326843,
      "learning_rate": 8e-05,
      "loss": 1.6444,
      "step": 171
    },
    {
      "epoch": 0.037831298801275705,
      "grad_norm": 0.2858507037162781,
      "learning_rate": 8e-05,
      "loss": 1.8584,
      "step": 172
    },
    {
      "epoch": 0.03805124821291103,
      "grad_norm": 0.26513633131980896,
      "learning_rate": 8e-05,
      "loss": 1.7107,
      "step": 173
    },
    {
      "epoch": 0.03827119762454635,
      "grad_norm": 0.3162567913532257,
      "learning_rate": 8e-05,
      "loss": 1.7153,
      "step": 174
    },
    {
      "epoch": 0.03849114703618168,
      "grad_norm": 0.28961601853370667,
      "learning_rate": 8e-05,
      "loss": 1.8455,
      "step": 175
    },
    {
      "epoch": 0.038711096447817,
      "grad_norm": 0.29676249623298645,
      "learning_rate": 8e-05,
      "loss": 1.9303,
      "step": 176
    },
    {
      "epoch": 0.038931045859452326,
      "grad_norm": 0.2863664925098419,
      "learning_rate": 8e-05,
      "loss": 1.6975,
      "step": 177
    },
    {
      "epoch": 0.03915099527108765,
      "grad_norm": 0.2715422213077545,
      "learning_rate": 8e-05,
      "loss": 1.5472,
      "step": 178
    },
    {
      "epoch": 0.03937094468272297,
      "grad_norm": 0.2740415036678314,
      "learning_rate": 8e-05,
      "loss": 1.7113,
      "step": 179
    },
    {
      "epoch": 0.0395908940943583,
      "grad_norm": 0.29612302780151367,
      "learning_rate": 8e-05,
      "loss": 1.8689,
      "step": 180
    },
    {
      "epoch": 0.03981084350599362,
      "grad_norm": 0.26745903491973877,
      "learning_rate": 8e-05,
      "loss": 1.6076,
      "step": 181
    },
    {
      "epoch": 0.04003079291762895,
      "grad_norm": 0.296695739030838,
      "learning_rate": 8e-05,
      "loss": 1.846,
      "step": 182
    },
    {
      "epoch": 0.04025074232926427,
      "grad_norm": 0.27626705169677734,
      "learning_rate": 8e-05,
      "loss": 1.8103,
      "step": 183
    },
    {
      "epoch": 0.040470691740899595,
      "grad_norm": 0.2597677409648895,
      "learning_rate": 8e-05,
      "loss": 1.6432,
      "step": 184
    },
    {
      "epoch": 0.040690641152534915,
      "grad_norm": 0.2738899290561676,
      "learning_rate": 8e-05,
      "loss": 1.8351,
      "step": 185
    },
    {
      "epoch": 0.04091059056417024,
      "grad_norm": 0.2683742344379425,
      "learning_rate": 8e-05,
      "loss": 1.6453,
      "step": 186
    },
    {
      "epoch": 0.04113053997580556,
      "grad_norm": 0.28722816705703735,
      "learning_rate": 8e-05,
      "loss": 1.7685,
      "step": 187
    },
    {
      "epoch": 0.04135048938744089,
      "grad_norm": 0.2851015627384186,
      "learning_rate": 8e-05,
      "loss": 1.8464,
      "step": 188
    },
    {
      "epoch": 0.04157043879907621,
      "grad_norm": 0.2630920112133026,
      "learning_rate": 8e-05,
      "loss": 1.7176,
      "step": 189
    },
    {
      "epoch": 0.041790388210711536,
      "grad_norm": 0.2678779661655426,
      "learning_rate": 8e-05,
      "loss": 1.671,
      "step": 190
    },
    {
      "epoch": 0.04201033762234686,
      "grad_norm": 0.27810946106910706,
      "learning_rate": 8e-05,
      "loss": 1.6467,
      "step": 191
    },
    {
      "epoch": 0.04223028703398218,
      "grad_norm": 0.2831014394760132,
      "learning_rate": 8e-05,
      "loss": 1.8784,
      "step": 192
    },
    {
      "epoch": 0.04245023644561751,
      "grad_norm": 0.2643384635448456,
      "learning_rate": 8e-05,
      "loss": 1.6239,
      "step": 193
    },
    {
      "epoch": 0.04267018585725283,
      "grad_norm": 0.27143070101737976,
      "learning_rate": 8e-05,
      "loss": 1.8012,
      "step": 194
    },
    {
      "epoch": 0.04289013526888816,
      "grad_norm": 0.28524088859558105,
      "learning_rate": 8e-05,
      "loss": 1.7534,
      "step": 195
    },
    {
      "epoch": 0.04311008468052348,
      "grad_norm": 0.27226153016090393,
      "learning_rate": 8e-05,
      "loss": 1.847,
      "step": 196
    },
    {
      "epoch": 0.043330034092158805,
      "grad_norm": 0.27042534947395325,
      "learning_rate": 8e-05,
      "loss": 1.698,
      "step": 197
    },
    {
      "epoch": 0.043549983503794125,
      "grad_norm": 0.2673223912715912,
      "learning_rate": 8e-05,
      "loss": 1.7825,
      "step": 198
    },
    {
      "epoch": 0.04376993291542945,
      "grad_norm": 0.26485180854797363,
      "learning_rate": 8e-05,
      "loss": 1.7755,
      "step": 199
    },
    {
      "epoch": 0.04398988232706477,
      "grad_norm": 0.26945164799690247,
      "learning_rate": 8e-05,
      "loss": 1.8612,
      "step": 200
    },
    {
      "epoch": 0.0442098317387001,
      "grad_norm": 0.30337756872177124,
      "learning_rate": 8e-05,
      "loss": 1.8556,
      "step": 201
    },
    {
      "epoch": 0.044429781150335426,
      "grad_norm": 0.26593855023384094,
      "learning_rate": 8e-05,
      "loss": 1.7633,
      "step": 202
    },
    {
      "epoch": 0.044649730561970746,
      "grad_norm": 0.26703208684921265,
      "learning_rate": 8e-05,
      "loss": 1.7787,
      "step": 203
    },
    {
      "epoch": 0.04486967997360607,
      "grad_norm": 0.2799319922924042,
      "learning_rate": 8e-05,
      "loss": 1.8946,
      "step": 204
    },
    {
      "epoch": 0.04508962938524139,
      "grad_norm": 0.261406809091568,
      "learning_rate": 8e-05,
      "loss": 1.714,
      "step": 205
    },
    {
      "epoch": 0.04530957879687672,
      "grad_norm": 0.30923140048980713,
      "learning_rate": 8e-05,
      "loss": 1.9953,
      "step": 206
    },
    {
      "epoch": 0.04552952820851204,
      "grad_norm": 0.28189903497695923,
      "learning_rate": 8e-05,
      "loss": 1.8068,
      "step": 207
    },
    {
      "epoch": 0.04574947762014737,
      "grad_norm": 0.28659504652023315,
      "learning_rate": 8e-05,
      "loss": 1.7961,
      "step": 208
    },
    {
      "epoch": 0.04596942703178269,
      "grad_norm": 0.27828094363212585,
      "learning_rate": 8e-05,
      "loss": 1.6398,
      "step": 209
    },
    {
      "epoch": 0.046189376443418015,
      "grad_norm": 0.2826248109340668,
      "learning_rate": 8e-05,
      "loss": 1.8442,
      "step": 210
    },
    {
      "epoch": 0.046409325855053335,
      "grad_norm": 0.2596709430217743,
      "learning_rate": 8e-05,
      "loss": 1.7269,
      "step": 211
    },
    {
      "epoch": 0.04662927526668866,
      "grad_norm": 0.26883357763290405,
      "learning_rate": 8e-05,
      "loss": 1.7396,
      "step": 212
    },
    {
      "epoch": 0.04684922467832398,
      "grad_norm": 0.2834852933883667,
      "learning_rate": 8e-05,
      "loss": 1.6992,
      "step": 213
    },
    {
      "epoch": 0.04706917408995931,
      "grad_norm": 0.30232125520706177,
      "learning_rate": 8e-05,
      "loss": 1.8216,
      "step": 214
    },
    {
      "epoch": 0.047289123501594636,
      "grad_norm": 0.2887151539325714,
      "learning_rate": 8e-05,
      "loss": 1.5633,
      "step": 215
    },
    {
      "epoch": 0.047509072913229956,
      "grad_norm": 0.27171874046325684,
      "learning_rate": 8e-05,
      "loss": 1.8272,
      "step": 216
    },
    {
      "epoch": 0.04772902232486528,
      "grad_norm": 0.35441088676452637,
      "learning_rate": 8e-05,
      "loss": 1.8308,
      "step": 217
    },
    {
      "epoch": 0.047948971736500604,
      "grad_norm": 0.28351160883903503,
      "learning_rate": 8e-05,
      "loss": 1.8697,
      "step": 218
    },
    {
      "epoch": 0.04816892114813593,
      "grad_norm": 0.26361364126205444,
      "learning_rate": 8e-05,
      "loss": 1.7044,
      "step": 219
    },
    {
      "epoch": 0.04838887055977125,
      "grad_norm": 0.2720041871070862,
      "learning_rate": 8e-05,
      "loss": 1.7718,
      "step": 220
    },
    {
      "epoch": 0.04860881997140658,
      "grad_norm": 0.28131023049354553,
      "learning_rate": 8e-05,
      "loss": 1.8066,
      "step": 221
    },
    {
      "epoch": 0.0488287693830419,
      "grad_norm": 0.2640543580055237,
      "learning_rate": 8e-05,
      "loss": 1.69,
      "step": 222
    },
    {
      "epoch": 0.049048718794677225,
      "grad_norm": 0.26101046800613403,
      "learning_rate": 8e-05,
      "loss": 1.6372,
      "step": 223
    },
    {
      "epoch": 0.049268668206312545,
      "grad_norm": 0.3021651804447174,
      "learning_rate": 8e-05,
      "loss": 1.8528,
      "step": 224
    },
    {
      "epoch": 0.04948861761794787,
      "grad_norm": 0.2655261158943176,
      "learning_rate": 8e-05,
      "loss": 1.7406,
      "step": 225
    },
    {
      "epoch": 0.0497085670295832,
      "grad_norm": 0.2873914837837219,
      "learning_rate": 8e-05,
      "loss": 1.7643,
      "step": 226
    },
    {
      "epoch": 0.04992851644121852,
      "grad_norm": 0.31813880801200867,
      "learning_rate": 8e-05,
      "loss": 1.8645,
      "step": 227
    },
    {
      "epoch": 0.050148465852853846,
      "grad_norm": 0.2996014654636383,
      "learning_rate": 8e-05,
      "loss": 1.6685,
      "step": 228
    },
    {
      "epoch": 0.050368415264489166,
      "grad_norm": 0.2837509512901306,
      "learning_rate": 8e-05,
      "loss": 1.9227,
      "step": 229
    },
    {
      "epoch": 0.05058836467612449,
      "grad_norm": 0.29532885551452637,
      "learning_rate": 8e-05,
      "loss": 1.9073,
      "step": 230
    },
    {
      "epoch": 0.050808314087759814,
      "grad_norm": 0.285295307636261,
      "learning_rate": 8e-05,
      "loss": 1.8248,
      "step": 231
    },
    {
      "epoch": 0.05102826349939514,
      "grad_norm": 0.26331770420074463,
      "learning_rate": 8e-05,
      "loss": 1.7146,
      "step": 232
    },
    {
      "epoch": 0.05124821291103046,
      "grad_norm": 0.24956567585468292,
      "learning_rate": 8e-05,
      "loss": 1.5574,
      "step": 233
    },
    {
      "epoch": 0.05146816232266579,
      "grad_norm": 0.27515965700149536,
      "learning_rate": 8e-05,
      "loss": 1.7854,
      "step": 234
    },
    {
      "epoch": 0.05168811173430111,
      "grad_norm": 0.28268730640411377,
      "learning_rate": 8e-05,
      "loss": 1.8294,
      "step": 235
    },
    {
      "epoch": 0.051908061145936435,
      "grad_norm": 0.25420427322387695,
      "learning_rate": 8e-05,
      "loss": 1.6735,
      "step": 236
    },
    {
      "epoch": 0.052128010557571755,
      "grad_norm": 0.2869463860988617,
      "learning_rate": 8e-05,
      "loss": 1.808,
      "step": 237
    },
    {
      "epoch": 0.05234795996920708,
      "grad_norm": 0.2574792206287384,
      "learning_rate": 8e-05,
      "loss": 1.7563,
      "step": 238
    },
    {
      "epoch": 0.05256790938084241,
      "grad_norm": 0.26652273535728455,
      "learning_rate": 8e-05,
      "loss": 1.743,
      "step": 239
    },
    {
      "epoch": 0.05278785879247773,
      "grad_norm": 0.2956235408782959,
      "learning_rate": 8e-05,
      "loss": 1.9169,
      "step": 240
    },
    {
      "epoch": 0.053007808204113056,
      "grad_norm": 0.274142861366272,
      "learning_rate": 8e-05,
      "loss": 1.8321,
      "step": 241
    },
    {
      "epoch": 0.053227757615748376,
      "grad_norm": 0.27525436878204346,
      "learning_rate": 8e-05,
      "loss": 1.8206,
      "step": 242
    },
    {
      "epoch": 0.053447707027383703,
      "grad_norm": 0.26323091983795166,
      "learning_rate": 8e-05,
      "loss": 1.7574,
      "step": 243
    },
    {
      "epoch": 0.053667656439019024,
      "grad_norm": 0.28554126620292664,
      "learning_rate": 8e-05,
      "loss": 1.9293,
      "step": 244
    },
    {
      "epoch": 0.05388760585065435,
      "grad_norm": 0.2651476562023163,
      "learning_rate": 8e-05,
      "loss": 1.808,
      "step": 245
    },
    {
      "epoch": 0.05410755526228967,
      "grad_norm": 0.27941837906837463,
      "learning_rate": 8e-05,
      "loss": 1.7838,
      "step": 246
    },
    {
      "epoch": 0.054327504673925,
      "grad_norm": 0.26575711369514465,
      "learning_rate": 8e-05,
      "loss": 1.6117,
      "step": 247
    },
    {
      "epoch": 0.05454745408556032,
      "grad_norm": 0.2620556354522705,
      "learning_rate": 8e-05,
      "loss": 1.7703,
      "step": 248
    },
    {
      "epoch": 0.054767403497195645,
      "grad_norm": 0.2782936990261078,
      "learning_rate": 8e-05,
      "loss": 1.753,
      "step": 249
    },
    {
      "epoch": 0.05498735290883097,
      "grad_norm": 0.28347843885421753,
      "learning_rate": 8e-05,
      "loss": 1.8365,
      "step": 250
    },
    {
      "epoch": 0.05520730232046629,
      "grad_norm": 0.2740314304828644,
      "learning_rate": 8e-05,
      "loss": 1.7448,
      "step": 251
    },
    {
      "epoch": 0.05542725173210162,
      "grad_norm": 0.2779199779033661,
      "learning_rate": 8e-05,
      "loss": 1.8025,
      "step": 252
    },
    {
      "epoch": 0.05564720114373694,
      "grad_norm": 0.27700838446617126,
      "learning_rate": 8e-05,
      "loss": 1.6368,
      "step": 253
    },
    {
      "epoch": 0.055867150555372266,
      "grad_norm": 0.2753797173500061,
      "learning_rate": 8e-05,
      "loss": 1.7058,
      "step": 254
    },
    {
      "epoch": 0.056087099967007586,
      "grad_norm": 0.2677604556083679,
      "learning_rate": 8e-05,
      "loss": 1.772,
      "step": 255
    },
    {
      "epoch": 0.056307049378642914,
      "grad_norm": 0.291358083486557,
      "learning_rate": 8e-05,
      "loss": 1.7229,
      "step": 256
    },
    {
      "epoch": 0.056526998790278234,
      "grad_norm": 0.2605611979961395,
      "learning_rate": 8e-05,
      "loss": 1.6654,
      "step": 257
    },
    {
      "epoch": 0.05674694820191356,
      "grad_norm": 0.2726796865463257,
      "learning_rate": 8e-05,
      "loss": 1.8524,
      "step": 258
    },
    {
      "epoch": 0.05696689761354888,
      "grad_norm": 0.2769307494163513,
      "learning_rate": 8e-05,
      "loss": 1.913,
      "step": 259
    },
    {
      "epoch": 0.05718684702518421,
      "grad_norm": 0.27163514494895935,
      "learning_rate": 8e-05,
      "loss": 1.7076,
      "step": 260
    },
    {
      "epoch": 0.057406796436819535,
      "grad_norm": 0.27037522196769714,
      "learning_rate": 8e-05,
      "loss": 1.7461,
      "step": 261
    },
    {
      "epoch": 0.057626745848454855,
      "grad_norm": 0.2570153772830963,
      "learning_rate": 8e-05,
      "loss": 1.6714,
      "step": 262
    },
    {
      "epoch": 0.05784669526009018,
      "grad_norm": 0.2802227735519409,
      "learning_rate": 8e-05,
      "loss": 1.6782,
      "step": 263
    },
    {
      "epoch": 0.0580666446717255,
      "grad_norm": 0.293969064950943,
      "learning_rate": 8e-05,
      "loss": 1.6253,
      "step": 264
    },
    {
      "epoch": 0.05828659408336083,
      "grad_norm": 0.28199446201324463,
      "learning_rate": 8e-05,
      "loss": 1.791,
      "step": 265
    },
    {
      "epoch": 0.05850654349499615,
      "grad_norm": 0.3037835657596588,
      "learning_rate": 8e-05,
      "loss": 1.8553,
      "step": 266
    },
    {
      "epoch": 0.058726492906631476,
      "grad_norm": 0.2814860939979553,
      "learning_rate": 8e-05,
      "loss": 1.7237,
      "step": 267
    },
    {
      "epoch": 0.0589464423182668,
      "grad_norm": 0.29769864678382874,
      "learning_rate": 8e-05,
      "loss": 1.8635,
      "step": 268
    },
    {
      "epoch": 0.059166391729902124,
      "grad_norm": 0.26650169491767883,
      "learning_rate": 8e-05,
      "loss": 1.8173,
      "step": 269
    },
    {
      "epoch": 0.059386341141537444,
      "grad_norm": 0.29682958126068115,
      "learning_rate": 8e-05,
      "loss": 1.6548,
      "step": 270
    },
    {
      "epoch": 0.05960629055317277,
      "grad_norm": 0.2702498137950897,
      "learning_rate": 8e-05,
      "loss": 1.6022,
      "step": 271
    },
    {
      "epoch": 0.05982623996480809,
      "grad_norm": 0.2940424680709839,
      "learning_rate": 8e-05,
      "loss": 1.7955,
      "step": 272
    },
    {
      "epoch": 0.06004618937644342,
      "grad_norm": 0.2655317485332489,
      "learning_rate": 8e-05,
      "loss": 1.786,
      "step": 273
    },
    {
      "epoch": 0.060266138788078745,
      "grad_norm": 0.28093400597572327,
      "learning_rate": 8e-05,
      "loss": 1.9798,
      "step": 274
    },
    {
      "epoch": 0.060486088199714065,
      "grad_norm": 0.2635514736175537,
      "learning_rate": 8e-05,
      "loss": 1.6737,
      "step": 275
    },
    {
      "epoch": 0.06070603761134939,
      "grad_norm": 0.2648226320743561,
      "learning_rate": 8e-05,
      "loss": 1.8771,
      "step": 276
    },
    {
      "epoch": 0.06092598702298471,
      "grad_norm": 0.2934603691101074,
      "learning_rate": 8e-05,
      "loss": 1.4751,
      "step": 277
    },
    {
      "epoch": 0.06114593643462004,
      "grad_norm": 0.26369500160217285,
      "learning_rate": 8e-05,
      "loss": 1.7832,
      "step": 278
    },
    {
      "epoch": 0.06136588584625536,
      "grad_norm": 0.26159989833831787,
      "learning_rate": 8e-05,
      "loss": 1.7276,
      "step": 279
    },
    {
      "epoch": 0.061585835257890686,
      "grad_norm": 0.2826705873012543,
      "learning_rate": 8e-05,
      "loss": 1.8767,
      "step": 280
    },
    {
      "epoch": 0.06180578466952601,
      "grad_norm": 0.2911459505558014,
      "learning_rate": 8e-05,
      "loss": 1.7795,
      "step": 281
    },
    {
      "epoch": 0.062025734081161334,
      "grad_norm": 0.27846869826316833,
      "learning_rate": 8e-05,
      "loss": 1.838,
      "step": 282
    },
    {
      "epoch": 0.062245683492796654,
      "grad_norm": 0.33195585012435913,
      "learning_rate": 8e-05,
      "loss": 1.8576,
      "step": 283
    },
    {
      "epoch": 0.06246563290443198,
      "grad_norm": 0.26306337118148804,
      "learning_rate": 8e-05,
      "loss": 1.7202,
      "step": 284
    },
    {
      "epoch": 0.0626855823160673,
      "grad_norm": 0.2703022360801697,
      "learning_rate": 8e-05,
      "loss": 1.6962,
      "step": 285
    },
    {
      "epoch": 0.06290553172770263,
      "grad_norm": 0.2754605710506439,
      "learning_rate": 8e-05,
      "loss": 1.6468,
      "step": 286
    },
    {
      "epoch": 0.06312548113933796,
      "grad_norm": 0.2995694577693939,
      "learning_rate": 8e-05,
      "loss": 1.9298,
      "step": 287
    },
    {
      "epoch": 0.06334543055097328,
      "grad_norm": 0.27501800656318665,
      "learning_rate": 8e-05,
      "loss": 1.8152,
      "step": 288
    },
    {
      "epoch": 0.0635653799626086,
      "grad_norm": 0.2668202519416809,
      "learning_rate": 8e-05,
      "loss": 1.8809,
      "step": 289
    },
    {
      "epoch": 0.06378532937424393,
      "grad_norm": 0.26209571957588196,
      "learning_rate": 8e-05,
      "loss": 1.4927,
      "step": 290
    },
    {
      "epoch": 0.06400527878587925,
      "grad_norm": 0.35276591777801514,
      "learning_rate": 8e-05,
      "loss": 1.9654,
      "step": 291
    },
    {
      "epoch": 0.06422522819751457,
      "grad_norm": 0.26070040464401245,
      "learning_rate": 8e-05,
      "loss": 1.7332,
      "step": 292
    },
    {
      "epoch": 0.06444517760914989,
      "grad_norm": 0.26518604159355164,
      "learning_rate": 8e-05,
      "loss": 1.6867,
      "step": 293
    },
    {
      "epoch": 0.06466512702078522,
      "grad_norm": 0.28992095589637756,
      "learning_rate": 8e-05,
      "loss": 1.7498,
      "step": 294
    },
    {
      "epoch": 0.06488507643242054,
      "grad_norm": 0.27465108036994934,
      "learning_rate": 8e-05,
      "loss": 1.6095,
      "step": 295
    },
    {
      "epoch": 0.06510502584405586,
      "grad_norm": 0.2841359078884125,
      "learning_rate": 8e-05,
      "loss": 1.6869,
      "step": 296
    },
    {
      "epoch": 0.0653249752556912,
      "grad_norm": 0.28873759508132935,
      "learning_rate": 8e-05,
      "loss": 1.7954,
      "step": 297
    },
    {
      "epoch": 0.06554492466732652,
      "grad_norm": 0.2542605698108673,
      "learning_rate": 8e-05,
      "loss": 1.6075,
      "step": 298
    },
    {
      "epoch": 0.06576487407896184,
      "grad_norm": 0.270823210477829,
      "learning_rate": 8e-05,
      "loss": 1.7238,
      "step": 299
    },
    {
      "epoch": 0.06598482349059716,
      "grad_norm": 0.2610267102718353,
      "learning_rate": 8e-05,
      "loss": 1.697,
      "step": 300
    },
    {
      "epoch": 0.06620477290223249,
      "grad_norm": 0.28088685870170593,
      "learning_rate": 8e-05,
      "loss": 1.6806,
      "step": 301
    },
    {
      "epoch": 0.06642472231386781,
      "grad_norm": 0.2656930088996887,
      "learning_rate": 8e-05,
      "loss": 1.8744,
      "step": 302
    },
    {
      "epoch": 0.06664467172550313,
      "grad_norm": 0.2721637189388275,
      "learning_rate": 8e-05,
      "loss": 1.6903,
      "step": 303
    },
    {
      "epoch": 0.06686462113713845,
      "grad_norm": 0.2612883746623993,
      "learning_rate": 8e-05,
      "loss": 1.7444,
      "step": 304
    },
    {
      "epoch": 0.06708457054877379,
      "grad_norm": 0.2533530592918396,
      "learning_rate": 8e-05,
      "loss": 1.6427,
      "step": 305
    },
    {
      "epoch": 0.0673045199604091,
      "grad_norm": 0.27200043201446533,
      "learning_rate": 8e-05,
      "loss": 1.769,
      "step": 306
    },
    {
      "epoch": 0.06752446937204443,
      "grad_norm": 0.2626403272151947,
      "learning_rate": 8e-05,
      "loss": 1.64,
      "step": 307
    },
    {
      "epoch": 0.06774441878367975,
      "grad_norm": 0.3720408082008362,
      "learning_rate": 8e-05,
      "loss": 1.9055,
      "step": 308
    },
    {
      "epoch": 0.06796436819531508,
      "grad_norm": 0.2745527923107147,
      "learning_rate": 8e-05,
      "loss": 1.7844,
      "step": 309
    },
    {
      "epoch": 0.0681843176069504,
      "grad_norm": 0.2568323612213135,
      "learning_rate": 8e-05,
      "loss": 1.6728,
      "step": 310
    },
    {
      "epoch": 0.06840426701858572,
      "grad_norm": 0.2704140543937683,
      "learning_rate": 8e-05,
      "loss": 1.7685,
      "step": 311
    },
    {
      "epoch": 0.06862421643022105,
      "grad_norm": 0.27828502655029297,
      "learning_rate": 8e-05,
      "loss": 1.7957,
      "step": 312
    },
    {
      "epoch": 0.06884416584185638,
      "grad_norm": 0.2951858341693878,
      "learning_rate": 8e-05,
      "loss": 1.7709,
      "step": 313
    },
    {
      "epoch": 0.0690641152534917,
      "grad_norm": 0.2756475806236267,
      "learning_rate": 8e-05,
      "loss": 1.6348,
      "step": 314
    },
    {
      "epoch": 0.06928406466512702,
      "grad_norm": 0.2913607954978943,
      "learning_rate": 8e-05,
      "loss": 1.7888,
      "step": 315
    },
    {
      "epoch": 0.06950401407676235,
      "grad_norm": 0.2798636853694916,
      "learning_rate": 8e-05,
      "loss": 1.7806,
      "step": 316
    },
    {
      "epoch": 0.06972396348839767,
      "grad_norm": 0.27596554160118103,
      "learning_rate": 8e-05,
      "loss": 1.7458,
      "step": 317
    },
    {
      "epoch": 0.06994391290003299,
      "grad_norm": 0.26655322313308716,
      "learning_rate": 8e-05,
      "loss": 1.5985,
      "step": 318
    },
    {
      "epoch": 0.07016386231166831,
      "grad_norm": 0.2731332778930664,
      "learning_rate": 8e-05,
      "loss": 1.5995,
      "step": 319
    },
    {
      "epoch": 0.07038381172330364,
      "grad_norm": 0.2769210934638977,
      "learning_rate": 8e-05,
      "loss": 1.6748,
      "step": 320
    },
    {
      "epoch": 0.07060376113493896,
      "grad_norm": 0.290889173746109,
      "learning_rate": 8e-05,
      "loss": 1.9427,
      "step": 321
    },
    {
      "epoch": 0.07082371054657428,
      "grad_norm": 0.2911258339881897,
      "learning_rate": 8e-05,
      "loss": 1.7723,
      "step": 322
    },
    {
      "epoch": 0.07104365995820962,
      "grad_norm": 0.301992267370224,
      "learning_rate": 8e-05,
      "loss": 1.7772,
      "step": 323
    },
    {
      "epoch": 0.07126360936984494,
      "grad_norm": 0.3023516535758972,
      "learning_rate": 8e-05,
      "loss": 1.8363,
      "step": 324
    },
    {
      "epoch": 0.07148355878148026,
      "grad_norm": 0.3058542013168335,
      "learning_rate": 8e-05,
      "loss": 1.8762,
      "step": 325
    },
    {
      "epoch": 0.07170350819311558,
      "grad_norm": 0.3215092718601227,
      "learning_rate": 8e-05,
      "loss": 1.7265,
      "step": 326
    },
    {
      "epoch": 0.07192345760475091,
      "grad_norm": 0.2762998342514038,
      "learning_rate": 8e-05,
      "loss": 1.6361,
      "step": 327
    },
    {
      "epoch": 0.07214340701638623,
      "grad_norm": 0.258635014295578,
      "learning_rate": 8e-05,
      "loss": 1.7031,
      "step": 328
    },
    {
      "epoch": 0.07236335642802155,
      "grad_norm": 0.27160710096359253,
      "learning_rate": 8e-05,
      "loss": 1.6759,
      "step": 329
    },
    {
      "epoch": 0.07258330583965687,
      "grad_norm": 0.31089314818382263,
      "learning_rate": 8e-05,
      "loss": 1.8141,
      "step": 330
    },
    {
      "epoch": 0.0728032552512922,
      "grad_norm": 0.3026575744152069,
      "learning_rate": 8e-05,
      "loss": 1.9513,
      "step": 331
    },
    {
      "epoch": 0.07302320466292753,
      "grad_norm": 0.2692122161388397,
      "learning_rate": 8e-05,
      "loss": 1.8277,
      "step": 332
    },
    {
      "epoch": 0.07324315407456285,
      "grad_norm": 0.27460286021232605,
      "learning_rate": 8e-05,
      "loss": 1.6426,
      "step": 333
    },
    {
      "epoch": 0.07346310348619818,
      "grad_norm": 0.2557325065135956,
      "learning_rate": 8e-05,
      "loss": 1.6418,
      "step": 334
    },
    {
      "epoch": 0.0736830528978335,
      "grad_norm": 0.28074318170547485,
      "learning_rate": 8e-05,
      "loss": 1.79,
      "step": 335
    },
    {
      "epoch": 0.07390300230946882,
      "grad_norm": 0.28538671135902405,
      "learning_rate": 8e-05,
      "loss": 1.7363,
      "step": 336
    },
    {
      "epoch": 0.07412295172110414,
      "grad_norm": 0.27379995584487915,
      "learning_rate": 8e-05,
      "loss": 1.7881,
      "step": 337
    },
    {
      "epoch": 0.07434290113273948,
      "grad_norm": 0.2628316283226013,
      "learning_rate": 8e-05,
      "loss": 1.745,
      "step": 338
    },
    {
      "epoch": 0.0745628505443748,
      "grad_norm": 0.2573058009147644,
      "learning_rate": 8e-05,
      "loss": 1.7997,
      "step": 339
    },
    {
      "epoch": 0.07478279995601012,
      "grad_norm": 0.31905651092529297,
      "learning_rate": 8e-05,
      "loss": 1.8125,
      "step": 340
    },
    {
      "epoch": 0.07500274936764544,
      "grad_norm": 0.2501446604728699,
      "learning_rate": 8e-05,
      "loss": 1.557,
      "step": 341
    },
    {
      "epoch": 0.07522269877928077,
      "grad_norm": 0.26969289779663086,
      "learning_rate": 8e-05,
      "loss": 1.7819,
      "step": 342
    },
    {
      "epoch": 0.07544264819091609,
      "grad_norm": 0.28457415103912354,
      "learning_rate": 8e-05,
      "loss": 1.7682,
      "step": 343
    },
    {
      "epoch": 0.07566259760255141,
      "grad_norm": 0.27833452820777893,
      "learning_rate": 8e-05,
      "loss": 1.8436,
      "step": 344
    },
    {
      "epoch": 0.07588254701418674,
      "grad_norm": 0.2574867010116577,
      "learning_rate": 8e-05,
      "loss": 1.7196,
      "step": 345
    },
    {
      "epoch": 0.07610249642582206,
      "grad_norm": 0.30035245418548584,
      "learning_rate": 8e-05,
      "loss": 1.7159,
      "step": 346
    },
    {
      "epoch": 0.07632244583745738,
      "grad_norm": 0.284169465303421,
      "learning_rate": 8e-05,
      "loss": 1.7238,
      "step": 347
    },
    {
      "epoch": 0.0765423952490927,
      "grad_norm": 0.257168173789978,
      "learning_rate": 8e-05,
      "loss": 1.8531,
      "step": 348
    },
    {
      "epoch": 0.07676234466072804,
      "grad_norm": 0.2611413300037384,
      "learning_rate": 8e-05,
      "loss": 1.7753,
      "step": 349
    },
    {
      "epoch": 0.07698229407236336,
      "grad_norm": 0.26592132449150085,
      "learning_rate": 8e-05,
      "loss": 1.7557,
      "step": 350
    },
    {
      "epoch": 0.07720224348399868,
      "grad_norm": 0.27427396178245544,
      "learning_rate": 8e-05,
      "loss": 1.8699,
      "step": 351
    },
    {
      "epoch": 0.077422192895634,
      "grad_norm": 0.27014485001564026,
      "learning_rate": 8e-05,
      "loss": 1.816,
      "step": 352
    },
    {
      "epoch": 0.07764214230726933,
      "grad_norm": 0.27720019221305847,
      "learning_rate": 8e-05,
      "loss": 1.9601,
      "step": 353
    },
    {
      "epoch": 0.07786209171890465,
      "grad_norm": 0.3222314417362213,
      "learning_rate": 8e-05,
      "loss": 1.6726,
      "step": 354
    },
    {
      "epoch": 0.07808204113053997,
      "grad_norm": 0.2675410211086273,
      "learning_rate": 8e-05,
      "loss": 1.7113,
      "step": 355
    },
    {
      "epoch": 0.0783019905421753,
      "grad_norm": 0.2902251183986664,
      "learning_rate": 8e-05,
      "loss": 1.7734,
      "step": 356
    },
    {
      "epoch": 0.07852193995381063,
      "grad_norm": 0.2985514998435974,
      "learning_rate": 8e-05,
      "loss": 1.9182,
      "step": 357
    },
    {
      "epoch": 0.07874188936544595,
      "grad_norm": 0.30351343750953674,
      "learning_rate": 8e-05,
      "loss": 1.7795,
      "step": 358
    },
    {
      "epoch": 0.07896183877708127,
      "grad_norm": 0.2885829210281372,
      "learning_rate": 8e-05,
      "loss": 1.8054,
      "step": 359
    },
    {
      "epoch": 0.0791817881887166,
      "grad_norm": 0.273366242647171,
      "learning_rate": 8e-05,
      "loss": 1.7903,
      "step": 360
    },
    {
      "epoch": 0.07940173760035192,
      "grad_norm": 0.2959200441837311,
      "learning_rate": 8e-05,
      "loss": 1.9163,
      "step": 361
    },
    {
      "epoch": 0.07962168701198724,
      "grad_norm": 0.2587856948375702,
      "learning_rate": 8e-05,
      "loss": 1.5969,
      "step": 362
    },
    {
      "epoch": 0.07984163642362256,
      "grad_norm": 0.27777665853500366,
      "learning_rate": 8e-05,
      "loss": 1.8769,
      "step": 363
    },
    {
      "epoch": 0.0800615858352579,
      "grad_norm": 0.2635156512260437,
      "learning_rate": 8e-05,
      "loss": 1.8236,
      "step": 364
    },
    {
      "epoch": 0.08028153524689322,
      "grad_norm": 0.26534774899482727,
      "learning_rate": 8e-05,
      "loss": 1.6824,
      "step": 365
    },
    {
      "epoch": 0.08050148465852854,
      "grad_norm": 0.26372772455215454,
      "learning_rate": 8e-05,
      "loss": 1.517,
      "step": 366
    },
    {
      "epoch": 0.08072143407016386,
      "grad_norm": 0.2707895338535309,
      "learning_rate": 8e-05,
      "loss": 1.6757,
      "step": 367
    },
    {
      "epoch": 0.08094138348179919,
      "grad_norm": 0.2712070345878601,
      "learning_rate": 8e-05,
      "loss": 1.7261,
      "step": 368
    },
    {
      "epoch": 0.08116133289343451,
      "grad_norm": 0.2870525121688843,
      "learning_rate": 8e-05,
      "loss": 1.6337,
      "step": 369
    },
    {
      "epoch": 0.08138128230506983,
      "grad_norm": 0.30548396706581116,
      "learning_rate": 8e-05,
      "loss": 1.8733,
      "step": 370
    },
    {
      "epoch": 0.08160123171670516,
      "grad_norm": 0.2853962182998657,
      "learning_rate": 8e-05,
      "loss": 1.7938,
      "step": 371
    },
    {
      "epoch": 0.08182118112834048,
      "grad_norm": 0.2716579735279083,
      "learning_rate": 8e-05,
      "loss": 1.6733,
      "step": 372
    },
    {
      "epoch": 0.0820411305399758,
      "grad_norm": 0.3110131025314331,
      "learning_rate": 8e-05,
      "loss": 1.8554,
      "step": 373
    },
    {
      "epoch": 0.08226107995161112,
      "grad_norm": 0.28003835678100586,
      "learning_rate": 8e-05,
      "loss": 1.8032,
      "step": 374
    },
    {
      "epoch": 0.08248102936324646,
      "grad_norm": 0.28504347801208496,
      "learning_rate": 8e-05,
      "loss": 1.942,
      "step": 375
    },
    {
      "epoch": 0.08270097877488178,
      "grad_norm": 0.2593232989311218,
      "learning_rate": 8e-05,
      "loss": 1.4993,
      "step": 376
    },
    {
      "epoch": 0.0829209281865171,
      "grad_norm": 0.35680094361305237,
      "learning_rate": 8e-05,
      "loss": 1.8997,
      "step": 377
    },
    {
      "epoch": 0.08314087759815242,
      "grad_norm": 0.2747777998447418,
      "learning_rate": 8e-05,
      "loss": 1.7364,
      "step": 378
    },
    {
      "epoch": 0.08336082700978775,
      "grad_norm": 0.26816287636756897,
      "learning_rate": 8e-05,
      "loss": 1.7011,
      "step": 379
    },
    {
      "epoch": 0.08358077642142307,
      "grad_norm": 0.31877851486206055,
      "learning_rate": 8e-05,
      "loss": 1.6131,
      "step": 380
    },
    {
      "epoch": 0.08380072583305839,
      "grad_norm": 0.2845601737499237,
      "learning_rate": 8e-05,
      "loss": 1.6544,
      "step": 381
    },
    {
      "epoch": 0.08402067524469373,
      "grad_norm": 0.27758803963661194,
      "learning_rate": 8e-05,
      "loss": 1.8891,
      "step": 382
    },
    {
      "epoch": 0.08424062465632905,
      "grad_norm": 0.2832657992839813,
      "learning_rate": 8e-05,
      "loss": 1.7505,
      "step": 383
    },
    {
      "epoch": 0.08446057406796437,
      "grad_norm": 0.2901705801486969,
      "learning_rate": 8e-05,
      "loss": 1.7501,
      "step": 384
    },
    {
      "epoch": 0.08468052347959969,
      "grad_norm": 0.31189531087875366,
      "learning_rate": 8e-05,
      "loss": 1.8132,
      "step": 385
    },
    {
      "epoch": 0.08490047289123502,
      "grad_norm": 0.27582603693008423,
      "learning_rate": 8e-05,
      "loss": 1.7693,
      "step": 386
    },
    {
      "epoch": 0.08512042230287034,
      "grad_norm": 0.3030100464820862,
      "learning_rate": 8e-05,
      "loss": 1.7327,
      "step": 387
    },
    {
      "epoch": 0.08534037171450566,
      "grad_norm": 0.26879045367240906,
      "learning_rate": 8e-05,
      "loss": 1.6614,
      "step": 388
    },
    {
      "epoch": 0.08556032112614098,
      "grad_norm": 0.29507508873939514,
      "learning_rate": 8e-05,
      "loss": 1.9483,
      "step": 389
    },
    {
      "epoch": 0.08578027053777632,
      "grad_norm": 0.27386122941970825,
      "learning_rate": 8e-05,
      "loss": 1.8974,
      "step": 390
    },
    {
      "epoch": 0.08600021994941164,
      "grad_norm": 0.27103161811828613,
      "learning_rate": 8e-05,
      "loss": 1.7579,
      "step": 391
    },
    {
      "epoch": 0.08622016936104696,
      "grad_norm": 0.3045141100883484,
      "learning_rate": 8e-05,
      "loss": 1.8175,
      "step": 392
    },
    {
      "epoch": 0.08644011877268229,
      "grad_norm": 0.29032695293426514,
      "learning_rate": 8e-05,
      "loss": 1.7493,
      "step": 393
    },
    {
      "epoch": 0.08666006818431761,
      "grad_norm": 0.27853158116340637,
      "learning_rate": 8e-05,
      "loss": 1.7297,
      "step": 394
    },
    {
      "epoch": 0.08688001759595293,
      "grad_norm": 0.3007650375366211,
      "learning_rate": 8e-05,
      "loss": 1.6736,
      "step": 395
    },
    {
      "epoch": 0.08709996700758825,
      "grad_norm": 0.28009670972824097,
      "learning_rate": 8e-05,
      "loss": 1.9539,
      "step": 396
    },
    {
      "epoch": 0.08731991641922358,
      "grad_norm": 0.2512955665588379,
      "learning_rate": 8e-05,
      "loss": 1.6362,
      "step": 397
    },
    {
      "epoch": 0.0875398658308589,
      "grad_norm": 0.297489732503891,
      "learning_rate": 8e-05,
      "loss": 1.9097,
      "step": 398
    },
    {
      "epoch": 0.08775981524249422,
      "grad_norm": 0.2735532522201538,
      "learning_rate": 8e-05,
      "loss": 1.8348,
      "step": 399
    },
    {
      "epoch": 0.08797976465412954,
      "grad_norm": 0.2559053897857666,
      "learning_rate": 8e-05,
      "loss": 1.685,
      "step": 400
    },
    {
      "epoch": 0.08819971406576488,
      "grad_norm": 0.27982097864151,
      "learning_rate": 8e-05,
      "loss": 1.6801,
      "step": 401
    },
    {
      "epoch": 0.0884196634774002,
      "grad_norm": 0.26066988706588745,
      "learning_rate": 8e-05,
      "loss": 1.7732,
      "step": 402
    },
    {
      "epoch": 0.08863961288903552,
      "grad_norm": 0.26763463020324707,
      "learning_rate": 8e-05,
      "loss": 1.7214,
      "step": 403
    },
    {
      "epoch": 0.08885956230067085,
      "grad_norm": 0.2795925736427307,
      "learning_rate": 8e-05,
      "loss": 1.8387,
      "step": 404
    },
    {
      "epoch": 0.08907951171230617,
      "grad_norm": 0.266305148601532,
      "learning_rate": 8e-05,
      "loss": 1.6515,
      "step": 405
    },
    {
      "epoch": 0.08929946112394149,
      "grad_norm": 0.27049583196640015,
      "learning_rate": 8e-05,
      "loss": 1.7824,
      "step": 406
    },
    {
      "epoch": 0.08951941053557681,
      "grad_norm": 0.2959458529949188,
      "learning_rate": 8e-05,
      "loss": 1.8766,
      "step": 407
    },
    {
      "epoch": 0.08973935994721215,
      "grad_norm": 0.28563347458839417,
      "learning_rate": 8e-05,
      "loss": 1.8618,
      "step": 408
    },
    {
      "epoch": 0.08995930935884747,
      "grad_norm": 0.2840110659599304,
      "learning_rate": 8e-05,
      "loss": 1.6834,
      "step": 409
    },
    {
      "epoch": 0.09017925877048279,
      "grad_norm": 0.25303247570991516,
      "learning_rate": 8e-05,
      "loss": 1.6477,
      "step": 410
    },
    {
      "epoch": 0.09039920818211811,
      "grad_norm": 0.27236899733543396,
      "learning_rate": 8e-05,
      "loss": 1.7004,
      "step": 411
    },
    {
      "epoch": 0.09061915759375344,
      "grad_norm": 0.2795659899711609,
      "learning_rate": 8e-05,
      "loss": 1.7492,
      "step": 412
    },
    {
      "epoch": 0.09083910700538876,
      "grad_norm": 0.26019132137298584,
      "learning_rate": 8e-05,
      "loss": 1.691,
      "step": 413
    },
    {
      "epoch": 0.09105905641702408,
      "grad_norm": 0.26624274253845215,
      "learning_rate": 8e-05,
      "loss": 1.7001,
      "step": 414
    },
    {
      "epoch": 0.09127900582865942,
      "grad_norm": 0.2661585509777069,
      "learning_rate": 8e-05,
      "loss": 1.6762,
      "step": 415
    },
    {
      "epoch": 0.09149895524029474,
      "grad_norm": 0.2719002068042755,
      "learning_rate": 8e-05,
      "loss": 1.6915,
      "step": 416
    },
    {
      "epoch": 0.09171890465193006,
      "grad_norm": 0.24670244753360748,
      "learning_rate": 8e-05,
      "loss": 1.5598,
      "step": 417
    },
    {
      "epoch": 0.09193885406356538,
      "grad_norm": 0.2550405263900757,
      "learning_rate": 8e-05,
      "loss": 1.4817,
      "step": 418
    },
    {
      "epoch": 0.09215880347520071,
      "grad_norm": 0.26272761821746826,
      "learning_rate": 8e-05,
      "loss": 1.7016,
      "step": 419
    },
    {
      "epoch": 0.09237875288683603,
      "grad_norm": 0.2673632502555847,
      "learning_rate": 8e-05,
      "loss": 1.7626,
      "step": 420
    },
    {
      "epoch": 0.09259870229847135,
      "grad_norm": 0.25949448347091675,
      "learning_rate": 8e-05,
      "loss": 1.6273,
      "step": 421
    },
    {
      "epoch": 0.09281865171010667,
      "grad_norm": 0.27953028678894043,
      "learning_rate": 8e-05,
      "loss": 1.8843,
      "step": 422
    },
    {
      "epoch": 0.093038601121742,
      "grad_norm": 0.2534630298614502,
      "learning_rate": 8e-05,
      "loss": 1.7305,
      "step": 423
    },
    {
      "epoch": 0.09325855053337732,
      "grad_norm": 0.2573072910308838,
      "learning_rate": 8e-05,
      "loss": 1.6397,
      "step": 424
    },
    {
      "epoch": 0.09347849994501264,
      "grad_norm": 0.2604135572910309,
      "learning_rate": 8e-05,
      "loss": 1.6696,
      "step": 425
    },
    {
      "epoch": 0.09369844935664796,
      "grad_norm": 0.25805628299713135,
      "learning_rate": 8e-05,
      "loss": 1.6441,
      "step": 426
    },
    {
      "epoch": 0.0939183987682833,
      "grad_norm": 0.2935563027858734,
      "learning_rate": 8e-05,
      "loss": 1.6475,
      "step": 427
    },
    {
      "epoch": 0.09413834817991862,
      "grad_norm": 0.25222933292388916,
      "learning_rate": 8e-05,
      "loss": 1.727,
      "step": 428
    },
    {
      "epoch": 0.09435829759155394,
      "grad_norm": 0.2593076527118683,
      "learning_rate": 8e-05,
      "loss": 1.7066,
      "step": 429
    },
    {
      "epoch": 0.09457824700318927,
      "grad_norm": 0.25259336829185486,
      "learning_rate": 8e-05,
      "loss": 1.6821,
      "step": 430
    },
    {
      "epoch": 0.09479819641482459,
      "grad_norm": 0.2512541115283966,
      "learning_rate": 8e-05,
      "loss": 1.5923,
      "step": 431
    },
    {
      "epoch": 0.09501814582645991,
      "grad_norm": 0.2711183726787567,
      "learning_rate": 8e-05,
      "loss": 1.755,
      "step": 432
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.2782961130142212,
      "learning_rate": 8e-05,
      "loss": 1.8914,
      "step": 433
    },
    {
      "epoch": 0.09545804464973057,
      "grad_norm": 0.25964146852493286,
      "learning_rate": 8e-05,
      "loss": 1.6588,
      "step": 434
    },
    {
      "epoch": 0.09567799406136589,
      "grad_norm": 0.27077510952949524,
      "learning_rate": 8e-05,
      "loss": 1.753,
      "step": 435
    },
    {
      "epoch": 0.09589794347300121,
      "grad_norm": 0.2923937141895294,
      "learning_rate": 8e-05,
      "loss": 1.8218,
      "step": 436
    },
    {
      "epoch": 0.09611789288463653,
      "grad_norm": 0.2513190805912018,
      "learning_rate": 8e-05,
      "loss": 1.6232,
      "step": 437
    },
    {
      "epoch": 0.09633784229627186,
      "grad_norm": 0.28531181812286377,
      "learning_rate": 8e-05,
      "loss": 1.7199,
      "step": 438
    },
    {
      "epoch": 0.09655779170790718,
      "grad_norm": 0.302020400762558,
      "learning_rate": 8e-05,
      "loss": 1.8359,
      "step": 439
    },
    {
      "epoch": 0.0967777411195425,
      "grad_norm": 0.28001338243484497,
      "learning_rate": 8e-05,
      "loss": 1.8434,
      "step": 440
    },
    {
      "epoch": 0.09699769053117784,
      "grad_norm": 0.2990663945674896,
      "learning_rate": 8e-05,
      "loss": 1.6995,
      "step": 441
    },
    {
      "epoch": 0.09721763994281316,
      "grad_norm": 0.266197144985199,
      "learning_rate": 8e-05,
      "loss": 1.6195,
      "step": 442
    },
    {
      "epoch": 0.09743758935444848,
      "grad_norm": 0.28108519315719604,
      "learning_rate": 8e-05,
      "loss": 1.8108,
      "step": 443
    },
    {
      "epoch": 0.0976575387660838,
      "grad_norm": 0.26744788885116577,
      "learning_rate": 8e-05,
      "loss": 1.6497,
      "step": 444
    },
    {
      "epoch": 0.09787748817771913,
      "grad_norm": 0.28030574321746826,
      "learning_rate": 8e-05,
      "loss": 1.8143,
      "step": 445
    },
    {
      "epoch": 0.09809743758935445,
      "grad_norm": 0.27872079610824585,
      "learning_rate": 8e-05,
      "loss": 1.6319,
      "step": 446
    },
    {
      "epoch": 0.09831738700098977,
      "grad_norm": 0.2816067039966583,
      "learning_rate": 8e-05,
      "loss": 1.8385,
      "step": 447
    },
    {
      "epoch": 0.09853733641262509,
      "grad_norm": 0.25677627325057983,
      "learning_rate": 8e-05,
      "loss": 1.6885,
      "step": 448
    },
    {
      "epoch": 0.09875728582426042,
      "grad_norm": 0.276569128036499,
      "learning_rate": 8e-05,
      "loss": 1.7652,
      "step": 449
    },
    {
      "epoch": 0.09897723523589574,
      "grad_norm": 0.2765633463859558,
      "learning_rate": 8e-05,
      "loss": 1.7763,
      "step": 450
    },
    {
      "epoch": 0.09919718464753106,
      "grad_norm": 0.27050015330314636,
      "learning_rate": 8e-05,
      "loss": 1.6459,
      "step": 451
    },
    {
      "epoch": 0.0994171340591664,
      "grad_norm": 0.2552846372127533,
      "learning_rate": 8e-05,
      "loss": 1.6877,
      "step": 452
    },
    {
      "epoch": 0.09963708347080172,
      "grad_norm": 0.2653469741344452,
      "learning_rate": 8e-05,
      "loss": 1.6536,
      "step": 453
    },
    {
      "epoch": 0.09985703288243704,
      "grad_norm": 0.28801941871643066,
      "learning_rate": 8e-05,
      "loss": 1.7643,
      "step": 454
    },
    {
      "epoch": 0.10007698229407236,
      "grad_norm": 0.2930269241333008,
      "learning_rate": 8e-05,
      "loss": 1.7766,
      "step": 455
    },
    {
      "epoch": 0.10029693170570769,
      "grad_norm": 0.2718334496021271,
      "learning_rate": 8e-05,
      "loss": 1.7347,
      "step": 456
    },
    {
      "epoch": 0.10051688111734301,
      "grad_norm": 0.2807629704475403,
      "learning_rate": 8e-05,
      "loss": 1.7245,
      "step": 457
    },
    {
      "epoch": 0.10073683052897833,
      "grad_norm": 0.2801489531993866,
      "learning_rate": 8e-05,
      "loss": 1.7854,
      "step": 458
    },
    {
      "epoch": 0.10095677994061365,
      "grad_norm": 0.2616996765136719,
      "learning_rate": 8e-05,
      "loss": 1.6179,
      "step": 459
    },
    {
      "epoch": 0.10117672935224899,
      "grad_norm": 0.2626480758190155,
      "learning_rate": 8e-05,
      "loss": 1.7475,
      "step": 460
    },
    {
      "epoch": 0.10139667876388431,
      "grad_norm": 0.27338841557502747,
      "learning_rate": 8e-05,
      "loss": 1.8972,
      "step": 461
    },
    {
      "epoch": 0.10161662817551963,
      "grad_norm": 0.2695038616657257,
      "learning_rate": 8e-05,
      "loss": 1.7279,
      "step": 462
    },
    {
      "epoch": 0.10183657758715496,
      "grad_norm": 0.25614050030708313,
      "learning_rate": 8e-05,
      "loss": 1.6,
      "step": 463
    },
    {
      "epoch": 0.10205652699879028,
      "grad_norm": 0.2722180187702179,
      "learning_rate": 8e-05,
      "loss": 1.9241,
      "step": 464
    },
    {
      "epoch": 0.1022764764104256,
      "grad_norm": 0.2580203115940094,
      "learning_rate": 8e-05,
      "loss": 1.693,
      "step": 465
    },
    {
      "epoch": 0.10249642582206092,
      "grad_norm": 0.2848857641220093,
      "learning_rate": 8e-05,
      "loss": 1.9072,
      "step": 466
    },
    {
      "epoch": 0.10271637523369626,
      "grad_norm": 0.2783052325248718,
      "learning_rate": 8e-05,
      "loss": 1.9102,
      "step": 467
    },
    {
      "epoch": 0.10293632464533158,
      "grad_norm": 0.279695987701416,
      "learning_rate": 8e-05,
      "loss": 1.7491,
      "step": 468
    },
    {
      "epoch": 0.1031562740569669,
      "grad_norm": 0.2493034154176712,
      "learning_rate": 8e-05,
      "loss": 1.6789,
      "step": 469
    },
    {
      "epoch": 0.10337622346860222,
      "grad_norm": 0.2751196622848511,
      "learning_rate": 8e-05,
      "loss": 1.8132,
      "step": 470
    },
    {
      "epoch": 0.10359617288023755,
      "grad_norm": 0.2739677131175995,
      "learning_rate": 8e-05,
      "loss": 1.7945,
      "step": 471
    },
    {
      "epoch": 0.10381612229187287,
      "grad_norm": 0.30357351899147034,
      "learning_rate": 8e-05,
      "loss": 1.9113,
      "step": 472
    },
    {
      "epoch": 0.10403607170350819,
      "grad_norm": 0.2646970748901367,
      "learning_rate": 8e-05,
      "loss": 1.811,
      "step": 473
    },
    {
      "epoch": 0.10425602111514351,
      "grad_norm": 0.2626940608024597,
      "learning_rate": 8e-05,
      "loss": 1.6911,
      "step": 474
    },
    {
      "epoch": 0.10447597052677884,
      "grad_norm": 0.2613508701324463,
      "learning_rate": 8e-05,
      "loss": 1.7209,
      "step": 475
    },
    {
      "epoch": 0.10469591993841416,
      "grad_norm": 0.2609264552593231,
      "learning_rate": 8e-05,
      "loss": 1.6303,
      "step": 476
    },
    {
      "epoch": 0.10491586935004948,
      "grad_norm": 0.2549975514411926,
      "learning_rate": 8e-05,
      "loss": 1.7769,
      "step": 477
    },
    {
      "epoch": 0.10513581876168482,
      "grad_norm": 0.2742570638656616,
      "learning_rate": 8e-05,
      "loss": 1.8101,
      "step": 478
    },
    {
      "epoch": 0.10535576817332014,
      "grad_norm": 0.267070472240448,
      "learning_rate": 8e-05,
      "loss": 1.787,
      "step": 479
    },
    {
      "epoch": 0.10557571758495546,
      "grad_norm": 0.2735085189342499,
      "learning_rate": 8e-05,
      "loss": 1.8112,
      "step": 480
    },
    {
      "epoch": 0.10579566699659078,
      "grad_norm": 0.260111540555954,
      "learning_rate": 8e-05,
      "loss": 1.6926,
      "step": 481
    },
    {
      "epoch": 0.10601561640822611,
      "grad_norm": 0.26309284567832947,
      "learning_rate": 8e-05,
      "loss": 1.778,
      "step": 482
    },
    {
      "epoch": 0.10623556581986143,
      "grad_norm": 0.2658458948135376,
      "learning_rate": 8e-05,
      "loss": 1.7179,
      "step": 483
    },
    {
      "epoch": 0.10645551523149675,
      "grad_norm": 0.27498647570610046,
      "learning_rate": 8e-05,
      "loss": 1.6689,
      "step": 484
    },
    {
      "epoch": 0.10667546464313207,
      "grad_norm": 0.2658367156982422,
      "learning_rate": 8e-05,
      "loss": 1.6786,
      "step": 485
    },
    {
      "epoch": 0.10689541405476741,
      "grad_norm": 0.26023292541503906,
      "learning_rate": 8e-05,
      "loss": 1.6995,
      "step": 486
    },
    {
      "epoch": 0.10711536346640273,
      "grad_norm": 0.25749459862709045,
      "learning_rate": 8e-05,
      "loss": 1.6614,
      "step": 487
    },
    {
      "epoch": 0.10733531287803805,
      "grad_norm": 0.26305267214775085,
      "learning_rate": 8e-05,
      "loss": 1.6838,
      "step": 488
    },
    {
      "epoch": 0.10755526228967338,
      "grad_norm": 0.25277695059776306,
      "learning_rate": 8e-05,
      "loss": 1.6975,
      "step": 489
    },
    {
      "epoch": 0.1077752117013087,
      "grad_norm": 0.2584420144557953,
      "learning_rate": 8e-05,
      "loss": 1.7434,
      "step": 490
    },
    {
      "epoch": 0.10799516111294402,
      "grad_norm": 0.28107360005378723,
      "learning_rate": 8e-05,
      "loss": 1.8037,
      "step": 491
    },
    {
      "epoch": 0.10821511052457934,
      "grad_norm": 0.553341269493103,
      "learning_rate": 8e-05,
      "loss": 1.8896,
      "step": 492
    },
    {
      "epoch": 0.10843505993621468,
      "grad_norm": 0.2718677222728729,
      "learning_rate": 8e-05,
      "loss": 1.6646,
      "step": 493
    },
    {
      "epoch": 0.10865500934785,
      "grad_norm": 0.27301734685897827,
      "learning_rate": 8e-05,
      "loss": 1.6663,
      "step": 494
    },
    {
      "epoch": 0.10887495875948532,
      "grad_norm": 0.26952439546585083,
      "learning_rate": 8e-05,
      "loss": 1.7228,
      "step": 495
    },
    {
      "epoch": 0.10909490817112064,
      "grad_norm": 0.3017599582672119,
      "learning_rate": 8e-05,
      "loss": 1.7936,
      "step": 496
    },
    {
      "epoch": 0.10931485758275597,
      "grad_norm": 0.2676602303981781,
      "learning_rate": 8e-05,
      "loss": 1.7861,
      "step": 497
    },
    {
      "epoch": 0.10953480699439129,
      "grad_norm": 0.27192267775535583,
      "learning_rate": 8e-05,
      "loss": 1.8032,
      "step": 498
    },
    {
      "epoch": 0.10975475640602661,
      "grad_norm": 0.2807183861732483,
      "learning_rate": 8e-05,
      "loss": 1.6331,
      "step": 499
    },
    {
      "epoch": 0.10997470581766194,
      "grad_norm": 0.2652963399887085,
      "learning_rate": 8e-05,
      "loss": 1.6231,
      "step": 500
    },
    {
      "epoch": 0.11019465522929726,
      "grad_norm": 0.26010751724243164,
      "learning_rate": 8e-05,
      "loss": 1.729,
      "step": 501
    },
    {
      "epoch": 0.11041460464093258,
      "grad_norm": 0.29573148488998413,
      "learning_rate": 8e-05,
      "loss": 1.8082,
      "step": 502
    },
    {
      "epoch": 0.1106345540525679,
      "grad_norm": 0.28008025884628296,
      "learning_rate": 8e-05,
      "loss": 1.6829,
      "step": 503
    },
    {
      "epoch": 0.11085450346420324,
      "grad_norm": 0.3029135763645172,
      "learning_rate": 8e-05,
      "loss": 1.7699,
      "step": 504
    },
    {
      "epoch": 0.11107445287583856,
      "grad_norm": 0.2821674346923828,
      "learning_rate": 8e-05,
      "loss": 1.7337,
      "step": 505
    },
    {
      "epoch": 0.11129440228747388,
      "grad_norm": 0.274880975484848,
      "learning_rate": 8e-05,
      "loss": 1.7973,
      "step": 506
    },
    {
      "epoch": 0.1115143516991092,
      "grad_norm": 0.28885796666145325,
      "learning_rate": 8e-05,
      "loss": 1.7756,
      "step": 507
    },
    {
      "epoch": 0.11173430111074453,
      "grad_norm": 0.2744079530239105,
      "learning_rate": 8e-05,
      "loss": 1.7991,
      "step": 508
    },
    {
      "epoch": 0.11195425052237985,
      "grad_norm": 0.2645000219345093,
      "learning_rate": 8e-05,
      "loss": 1.6566,
      "step": 509
    },
    {
      "epoch": 0.11217419993401517,
      "grad_norm": 0.2640466094017029,
      "learning_rate": 8e-05,
      "loss": 1.6649,
      "step": 510
    },
    {
      "epoch": 0.11239414934565051,
      "grad_norm": 0.2965867817401886,
      "learning_rate": 8e-05,
      "loss": 1.7733,
      "step": 511
    },
    {
      "epoch": 0.11261409875728583,
      "grad_norm": 0.2533203661441803,
      "learning_rate": 8e-05,
      "loss": 1.7194,
      "step": 512
    },
    {
      "epoch": 0.11283404816892115,
      "grad_norm": 0.261994868516922,
      "learning_rate": 8e-05,
      "loss": 1.7387,
      "step": 513
    },
    {
      "epoch": 0.11305399758055647,
      "grad_norm": 0.2868165969848633,
      "learning_rate": 8e-05,
      "loss": 1.7444,
      "step": 514
    },
    {
      "epoch": 0.1132739469921918,
      "grad_norm": 0.2836281657218933,
      "learning_rate": 8e-05,
      "loss": 1.6507,
      "step": 515
    },
    {
      "epoch": 0.11349389640382712,
      "grad_norm": 0.28675276041030884,
      "learning_rate": 8e-05,
      "loss": 1.7054,
      "step": 516
    },
    {
      "epoch": 0.11371384581546244,
      "grad_norm": 0.2745465040206909,
      "learning_rate": 8e-05,
      "loss": 1.77,
      "step": 517
    },
    {
      "epoch": 0.11393379522709776,
      "grad_norm": 0.27250972390174866,
      "learning_rate": 8e-05,
      "loss": 1.9102,
      "step": 518
    },
    {
      "epoch": 0.1141537446387331,
      "grad_norm": 0.2781262695789337,
      "learning_rate": 8e-05,
      "loss": 1.8126,
      "step": 519
    },
    {
      "epoch": 0.11437369405036842,
      "grad_norm": 0.2691183388233185,
      "learning_rate": 8e-05,
      "loss": 1.5978,
      "step": 520
    },
    {
      "epoch": 0.11459364346200374,
      "grad_norm": 0.29496780037879944,
      "learning_rate": 8e-05,
      "loss": 1.9214,
      "step": 521
    },
    {
      "epoch": 0.11481359287363907,
      "grad_norm": 0.27725401520729065,
      "learning_rate": 8e-05,
      "loss": 1.8722,
      "step": 522
    },
    {
      "epoch": 0.11503354228527439,
      "grad_norm": 0.28819364309310913,
      "learning_rate": 8e-05,
      "loss": 1.6739,
      "step": 523
    },
    {
      "epoch": 0.11525349169690971,
      "grad_norm": 0.278857946395874,
      "learning_rate": 8e-05,
      "loss": 1.8137,
      "step": 524
    },
    {
      "epoch": 0.11547344110854503,
      "grad_norm": 0.26911258697509766,
      "learning_rate": 8e-05,
      "loss": 1.7123,
      "step": 525
    },
    {
      "epoch": 0.11569339052018036,
      "grad_norm": 0.2656850814819336,
      "learning_rate": 8e-05,
      "loss": 1.8124,
      "step": 526
    },
    {
      "epoch": 0.11591333993181568,
      "grad_norm": 0.26521819829940796,
      "learning_rate": 8e-05,
      "loss": 1.8188,
      "step": 527
    },
    {
      "epoch": 0.116133289343451,
      "grad_norm": 0.2821720540523529,
      "learning_rate": 8e-05,
      "loss": 1.7607,
      "step": 528
    },
    {
      "epoch": 0.11635323875508632,
      "grad_norm": 0.294612854719162,
      "learning_rate": 8e-05,
      "loss": 1.8142,
      "step": 529
    },
    {
      "epoch": 0.11657318816672166,
      "grad_norm": 0.29858094453811646,
      "learning_rate": 8e-05,
      "loss": 1.8795,
      "step": 530
    },
    {
      "epoch": 0.11679313757835698,
      "grad_norm": 0.2726878821849823,
      "learning_rate": 8e-05,
      "loss": 1.7988,
      "step": 531
    },
    {
      "epoch": 0.1170130869899923,
      "grad_norm": 0.2651258111000061,
      "learning_rate": 8e-05,
      "loss": 1.8106,
      "step": 532
    },
    {
      "epoch": 0.11723303640162762,
      "grad_norm": 0.2681291997432709,
      "learning_rate": 8e-05,
      "loss": 1.6692,
      "step": 533
    },
    {
      "epoch": 0.11745298581326295,
      "grad_norm": 0.2641060948371887,
      "learning_rate": 8e-05,
      "loss": 1.6479,
      "step": 534
    },
    {
      "epoch": 0.11767293522489827,
      "grad_norm": 0.2850191593170166,
      "learning_rate": 8e-05,
      "loss": 1.7337,
      "step": 535
    },
    {
      "epoch": 0.1178928846365336,
      "grad_norm": 0.2718667685985565,
      "learning_rate": 8e-05,
      "loss": 1.7069,
      "step": 536
    },
    {
      "epoch": 0.11811283404816893,
      "grad_norm": 0.27950581908226013,
      "learning_rate": 8e-05,
      "loss": 1.83,
      "step": 537
    },
    {
      "epoch": 0.11833278345980425,
      "grad_norm": 0.26720213890075684,
      "learning_rate": 8e-05,
      "loss": 1.6787,
      "step": 538
    },
    {
      "epoch": 0.11855273287143957,
      "grad_norm": 0.25440508127212524,
      "learning_rate": 8e-05,
      "loss": 1.5966,
      "step": 539
    },
    {
      "epoch": 0.11877268228307489,
      "grad_norm": 0.2716729938983917,
      "learning_rate": 8e-05,
      "loss": 1.793,
      "step": 540
    },
    {
      "epoch": 0.11899263169471022,
      "grad_norm": 0.26204821467399597,
      "learning_rate": 8e-05,
      "loss": 1.5882,
      "step": 541
    },
    {
      "epoch": 0.11921258110634554,
      "grad_norm": 0.2756775915622711,
      "learning_rate": 8e-05,
      "loss": 1.7529,
      "step": 542
    },
    {
      "epoch": 0.11943253051798086,
      "grad_norm": 0.27235740423202515,
      "learning_rate": 8e-05,
      "loss": 1.7607,
      "step": 543
    },
    {
      "epoch": 0.11965247992961618,
      "grad_norm": 0.27712538838386536,
      "learning_rate": 8e-05,
      "loss": 1.7504,
      "step": 544
    },
    {
      "epoch": 0.11987242934125152,
      "grad_norm": 0.27800193428993225,
      "learning_rate": 8e-05,
      "loss": 1.7421,
      "step": 545
    },
    {
      "epoch": 0.12009237875288684,
      "grad_norm": 0.27911701798439026,
      "learning_rate": 8e-05,
      "loss": 1.6683,
      "step": 546
    },
    {
      "epoch": 0.12031232816452216,
      "grad_norm": 0.27643364667892456,
      "learning_rate": 8e-05,
      "loss": 1.6393,
      "step": 547
    },
    {
      "epoch": 0.12053227757615749,
      "grad_norm": 0.25785166025161743,
      "learning_rate": 8e-05,
      "loss": 1.641,
      "step": 548
    },
    {
      "epoch": 0.12075222698779281,
      "grad_norm": 0.2791956067085266,
      "learning_rate": 8e-05,
      "loss": 1.7585,
      "step": 549
    },
    {
      "epoch": 0.12097217639942813,
      "grad_norm": 0.28245967626571655,
      "learning_rate": 8e-05,
      "loss": 1.8716,
      "step": 550
    },
    {
      "epoch": 0.12119212581106345,
      "grad_norm": 0.27160346508026123,
      "learning_rate": 8e-05,
      "loss": 1.7023,
      "step": 551
    },
    {
      "epoch": 0.12141207522269878,
      "grad_norm": 0.2670506536960602,
      "learning_rate": 8e-05,
      "loss": 1.5844,
      "step": 552
    },
    {
      "epoch": 0.1216320246343341,
      "grad_norm": 0.2762441337108612,
      "learning_rate": 8e-05,
      "loss": 1.7286,
      "step": 553
    },
    {
      "epoch": 0.12185197404596942,
      "grad_norm": 0.29608720541000366,
      "learning_rate": 8e-05,
      "loss": 1.7875,
      "step": 554
    },
    {
      "epoch": 0.12207192345760474,
      "grad_norm": 0.2847777307033539,
      "learning_rate": 8e-05,
      "loss": 1.7388,
      "step": 555
    },
    {
      "epoch": 0.12229187286924008,
      "grad_norm": 0.2769443988800049,
      "learning_rate": 8e-05,
      "loss": 1.8129,
      "step": 556
    },
    {
      "epoch": 0.1225118222808754,
      "grad_norm": 0.27490487694740295,
      "learning_rate": 8e-05,
      "loss": 1.678,
      "step": 557
    },
    {
      "epoch": 0.12273177169251072,
      "grad_norm": 0.2851822078227997,
      "learning_rate": 8e-05,
      "loss": 1.8268,
      "step": 558
    },
    {
      "epoch": 0.12295172110414605,
      "grad_norm": 0.31336653232574463,
      "learning_rate": 8e-05,
      "loss": 1.8247,
      "step": 559
    },
    {
      "epoch": 0.12317167051578137,
      "grad_norm": 0.26455923914909363,
      "learning_rate": 8e-05,
      "loss": 1.5548,
      "step": 560
    },
    {
      "epoch": 0.12339161992741669,
      "grad_norm": 0.2750054597854614,
      "learning_rate": 8e-05,
      "loss": 1.7912,
      "step": 561
    },
    {
      "epoch": 0.12361156933905201,
      "grad_norm": 0.28016433119773865,
      "learning_rate": 8e-05,
      "loss": 1.7367,
      "step": 562
    },
    {
      "epoch": 0.12383151875068735,
      "grad_norm": 0.30594533681869507,
      "learning_rate": 8e-05,
      "loss": 1.7959,
      "step": 563
    },
    {
      "epoch": 0.12405146816232267,
      "grad_norm": 0.2753421664237976,
      "learning_rate": 8e-05,
      "loss": 1.6714,
      "step": 564
    },
    {
      "epoch": 0.12427141757395799,
      "grad_norm": 0.3309609889984131,
      "learning_rate": 8e-05,
      "loss": 1.7632,
      "step": 565
    },
    {
      "epoch": 0.12449136698559331,
      "grad_norm": 0.3116569221019745,
      "learning_rate": 8e-05,
      "loss": 1.8312,
      "step": 566
    },
    {
      "epoch": 0.12471131639722864,
      "grad_norm": 0.27756184339523315,
      "learning_rate": 8e-05,
      "loss": 1.6622,
      "step": 567
    },
    {
      "epoch": 0.12493126580886396,
      "grad_norm": 0.2740349769592285,
      "learning_rate": 8e-05,
      "loss": 1.7015,
      "step": 568
    },
    {
      "epoch": 0.1251512152204993,
      "grad_norm": 0.2696126401424408,
      "learning_rate": 8e-05,
      "loss": 1.6063,
      "step": 569
    },
    {
      "epoch": 0.1253711646321346,
      "grad_norm": 0.29191461205482483,
      "learning_rate": 8e-05,
      "loss": 1.8429,
      "step": 570
    },
    {
      "epoch": 0.12559111404376994,
      "grad_norm": 0.2984013855457306,
      "learning_rate": 8e-05,
      "loss": 1.8194,
      "step": 571
    },
    {
      "epoch": 0.12581106345540527,
      "grad_norm": 0.27315613627433777,
      "learning_rate": 8e-05,
      "loss": 1.7027,
      "step": 572
    },
    {
      "epoch": 0.12603101286704058,
      "grad_norm": 0.28547149896621704,
      "learning_rate": 8e-05,
      "loss": 1.694,
      "step": 573
    },
    {
      "epoch": 0.1262509622786759,
      "grad_norm": 0.26458805799484253,
      "learning_rate": 8e-05,
      "loss": 1.7978,
      "step": 574
    },
    {
      "epoch": 0.12647091169031122,
      "grad_norm": 0.29676830768585205,
      "learning_rate": 8e-05,
      "loss": 1.8295,
      "step": 575
    },
    {
      "epoch": 0.12669086110194655,
      "grad_norm": 0.28077611327171326,
      "learning_rate": 8e-05,
      "loss": 1.7711,
      "step": 576
    },
    {
      "epoch": 0.12691081051358188,
      "grad_norm": 0.256736159324646,
      "learning_rate": 8e-05,
      "loss": 1.5371,
      "step": 577
    },
    {
      "epoch": 0.1271307599252172,
      "grad_norm": 0.2888578474521637,
      "learning_rate": 8e-05,
      "loss": 1.7532,
      "step": 578
    },
    {
      "epoch": 0.12735070933685252,
      "grad_norm": 0.29349133372306824,
      "learning_rate": 8e-05,
      "loss": 1.856,
      "step": 579
    },
    {
      "epoch": 0.12757065874848786,
      "grad_norm": 0.2626110911369324,
      "learning_rate": 8e-05,
      "loss": 1.5482,
      "step": 580
    },
    {
      "epoch": 0.12779060816012316,
      "grad_norm": 0.2715248167514801,
      "learning_rate": 8e-05,
      "loss": 1.7003,
      "step": 581
    },
    {
      "epoch": 0.1280105575717585,
      "grad_norm": 0.2800534963607788,
      "learning_rate": 8e-05,
      "loss": 1.7065,
      "step": 582
    },
    {
      "epoch": 0.12823050698339383,
      "grad_norm": 0.3190186619758606,
      "learning_rate": 8e-05,
      "loss": 1.8099,
      "step": 583
    },
    {
      "epoch": 0.12845045639502914,
      "grad_norm": 0.2689470648765564,
      "learning_rate": 8e-05,
      "loss": 1.7824,
      "step": 584
    },
    {
      "epoch": 0.12867040580666447,
      "grad_norm": 0.2715473473072052,
      "learning_rate": 8e-05,
      "loss": 1.7721,
      "step": 585
    },
    {
      "epoch": 0.12889035521829978,
      "grad_norm": 0.27956798672676086,
      "learning_rate": 8e-05,
      "loss": 1.7888,
      "step": 586
    },
    {
      "epoch": 0.1291103046299351,
      "grad_norm": 0.2842330038547516,
      "learning_rate": 8e-05,
      "loss": 1.7131,
      "step": 587
    },
    {
      "epoch": 0.12933025404157045,
      "grad_norm": 0.2888692021369934,
      "learning_rate": 8e-05,
      "loss": 1.7509,
      "step": 588
    },
    {
      "epoch": 0.12955020345320575,
      "grad_norm": 0.27673423290252686,
      "learning_rate": 8e-05,
      "loss": 1.7235,
      "step": 589
    },
    {
      "epoch": 0.1297701528648411,
      "grad_norm": 0.26007330417633057,
      "learning_rate": 8e-05,
      "loss": 1.7157,
      "step": 590
    },
    {
      "epoch": 0.12999010227647642,
      "grad_norm": 0.27521616220474243,
      "learning_rate": 8e-05,
      "loss": 1.7519,
      "step": 591
    },
    {
      "epoch": 0.13021005168811173,
      "grad_norm": 0.2753496766090393,
      "learning_rate": 8e-05,
      "loss": 1.6956,
      "step": 592
    },
    {
      "epoch": 0.13043000109974706,
      "grad_norm": 0.25559505820274353,
      "learning_rate": 8e-05,
      "loss": 1.522,
      "step": 593
    },
    {
      "epoch": 0.1306499505113824,
      "grad_norm": 0.26815375685691833,
      "learning_rate": 8e-05,
      "loss": 1.7658,
      "step": 594
    },
    {
      "epoch": 0.1308698999230177,
      "grad_norm": 0.26870042085647583,
      "learning_rate": 8e-05,
      "loss": 1.779,
      "step": 595
    },
    {
      "epoch": 0.13108984933465304,
      "grad_norm": 0.27346327900886536,
      "learning_rate": 8e-05,
      "loss": 1.7397,
      "step": 596
    },
    {
      "epoch": 0.13130979874628834,
      "grad_norm": 0.26674172282218933,
      "learning_rate": 8e-05,
      "loss": 1.906,
      "step": 597
    },
    {
      "epoch": 0.13152974815792368,
      "grad_norm": 0.266916960477829,
      "learning_rate": 8e-05,
      "loss": 1.6896,
      "step": 598
    },
    {
      "epoch": 0.131749697569559,
      "grad_norm": 0.2620035707950592,
      "learning_rate": 8e-05,
      "loss": 1.8032,
      "step": 599
    },
    {
      "epoch": 0.13196964698119432,
      "grad_norm": 0.2721168100833893,
      "learning_rate": 8e-05,
      "loss": 1.7992,
      "step": 600
    },
    {
      "epoch": 0.13218959639282965,
      "grad_norm": 0.2902929186820984,
      "learning_rate": 8e-05,
      "loss": 1.8392,
      "step": 601
    },
    {
      "epoch": 0.13240954580446498,
      "grad_norm": 0.267459899187088,
      "learning_rate": 8e-05,
      "loss": 1.8469,
      "step": 602
    },
    {
      "epoch": 0.1326294952161003,
      "grad_norm": 0.25643131136894226,
      "learning_rate": 8e-05,
      "loss": 1.5562,
      "step": 603
    },
    {
      "epoch": 0.13284944462773562,
      "grad_norm": 0.2919185757637024,
      "learning_rate": 8e-05,
      "loss": 1.7108,
      "step": 604
    },
    {
      "epoch": 0.13306939403937096,
      "grad_norm": 0.2631925046443939,
      "learning_rate": 8e-05,
      "loss": 1.4344,
      "step": 605
    },
    {
      "epoch": 0.13328934345100626,
      "grad_norm": 0.2710738182067871,
      "learning_rate": 8e-05,
      "loss": 1.6774,
      "step": 606
    },
    {
      "epoch": 0.1335092928626416,
      "grad_norm": 0.2641798257827759,
      "learning_rate": 8e-05,
      "loss": 1.8032,
      "step": 607
    },
    {
      "epoch": 0.1337292422742769,
      "grad_norm": 0.2571311891078949,
      "learning_rate": 8e-05,
      "loss": 1.63,
      "step": 608
    },
    {
      "epoch": 0.13394919168591224,
      "grad_norm": 0.24528057873249054,
      "learning_rate": 8e-05,
      "loss": 1.4576,
      "step": 609
    },
    {
      "epoch": 0.13416914109754757,
      "grad_norm": 0.270641028881073,
      "learning_rate": 8e-05,
      "loss": 1.7896,
      "step": 610
    },
    {
      "epoch": 0.13438909050918288,
      "grad_norm": 0.2723008990287781,
      "learning_rate": 8e-05,
      "loss": 1.7894,
      "step": 611
    },
    {
      "epoch": 0.1346090399208182,
      "grad_norm": 0.26487669348716736,
      "learning_rate": 8e-05,
      "loss": 1.7646,
      "step": 612
    },
    {
      "epoch": 0.13482898933245355,
      "grad_norm": 0.26771143078804016,
      "learning_rate": 8e-05,
      "loss": 1.8015,
      "step": 613
    },
    {
      "epoch": 0.13504893874408885,
      "grad_norm": 0.2585919499397278,
      "learning_rate": 8e-05,
      "loss": 1.6487,
      "step": 614
    },
    {
      "epoch": 0.1352688881557242,
      "grad_norm": 0.28161996603012085,
      "learning_rate": 8e-05,
      "loss": 1.7813,
      "step": 615
    },
    {
      "epoch": 0.1354888375673595,
      "grad_norm": 0.25246456265449524,
      "learning_rate": 8e-05,
      "loss": 1.5549,
      "step": 616
    },
    {
      "epoch": 0.13570878697899483,
      "grad_norm": 0.2803630530834198,
      "learning_rate": 8e-05,
      "loss": 1.7545,
      "step": 617
    },
    {
      "epoch": 0.13592873639063016,
      "grad_norm": 0.2587769031524658,
      "learning_rate": 8e-05,
      "loss": 1.6755,
      "step": 618
    },
    {
      "epoch": 0.13614868580226547,
      "grad_norm": 0.2890148162841797,
      "learning_rate": 8e-05,
      "loss": 1.9753,
      "step": 619
    },
    {
      "epoch": 0.1363686352139008,
      "grad_norm": 0.2924948036670685,
      "learning_rate": 8e-05,
      "loss": 1.7611,
      "step": 620
    },
    {
      "epoch": 0.13658858462553614,
      "grad_norm": 0.2594594359397888,
      "learning_rate": 8e-05,
      "loss": 1.6945,
      "step": 621
    },
    {
      "epoch": 0.13680853403717144,
      "grad_norm": 0.2853068709373474,
      "learning_rate": 8e-05,
      "loss": 1.8637,
      "step": 622
    },
    {
      "epoch": 0.13702848344880678,
      "grad_norm": 0.2696111798286438,
      "learning_rate": 8e-05,
      "loss": 1.7777,
      "step": 623
    },
    {
      "epoch": 0.1372484328604421,
      "grad_norm": 0.3137861490249634,
      "learning_rate": 8e-05,
      "loss": 1.8799,
      "step": 624
    },
    {
      "epoch": 0.13746838227207742,
      "grad_norm": 0.25645750761032104,
      "learning_rate": 8e-05,
      "loss": 1.5023,
      "step": 625
    },
    {
      "epoch": 0.13768833168371275,
      "grad_norm": 0.29853489995002747,
      "learning_rate": 8e-05,
      "loss": 1.9131,
      "step": 626
    },
    {
      "epoch": 0.13790828109534806,
      "grad_norm": 0.2653225362300873,
      "learning_rate": 8e-05,
      "loss": 1.6835,
      "step": 627
    },
    {
      "epoch": 0.1381282305069834,
      "grad_norm": 0.26686328649520874,
      "learning_rate": 8e-05,
      "loss": 1.7667,
      "step": 628
    },
    {
      "epoch": 0.13834817991861872,
      "grad_norm": 0.26114073395729065,
      "learning_rate": 8e-05,
      "loss": 1.6925,
      "step": 629
    },
    {
      "epoch": 0.13856812933025403,
      "grad_norm": 0.2520682215690613,
      "learning_rate": 8e-05,
      "loss": 1.6065,
      "step": 630
    },
    {
      "epoch": 0.13878807874188936,
      "grad_norm": 0.2676456868648529,
      "learning_rate": 8e-05,
      "loss": 1.7353,
      "step": 631
    },
    {
      "epoch": 0.1390080281535247,
      "grad_norm": 0.2525452673435211,
      "learning_rate": 8e-05,
      "loss": 1.5993,
      "step": 632
    },
    {
      "epoch": 0.13922797756516,
      "grad_norm": 0.25620371103286743,
      "learning_rate": 8e-05,
      "loss": 1.7188,
      "step": 633
    },
    {
      "epoch": 0.13944792697679534,
      "grad_norm": 0.4071904420852661,
      "learning_rate": 8e-05,
      "loss": 1.9348,
      "step": 634
    },
    {
      "epoch": 0.13966787638843067,
      "grad_norm": 0.2656376361846924,
      "learning_rate": 8e-05,
      "loss": 1.7833,
      "step": 635
    },
    {
      "epoch": 0.13988782580006598,
      "grad_norm": 0.25558993220329285,
      "learning_rate": 8e-05,
      "loss": 1.715,
      "step": 636
    },
    {
      "epoch": 0.1401077752117013,
      "grad_norm": 0.28318601846694946,
      "learning_rate": 8e-05,
      "loss": 1.8012,
      "step": 637
    },
    {
      "epoch": 0.14032772462333662,
      "grad_norm": 0.2558564245700836,
      "learning_rate": 8e-05,
      "loss": 1.5802,
      "step": 638
    },
    {
      "epoch": 0.14054767403497195,
      "grad_norm": 0.26874974370002747,
      "learning_rate": 8e-05,
      "loss": 1.7884,
      "step": 639
    },
    {
      "epoch": 0.1407676234466073,
      "grad_norm": 0.2960795760154724,
      "learning_rate": 8e-05,
      "loss": 1.7884,
      "step": 640
    },
    {
      "epoch": 0.1409875728582426,
      "grad_norm": 0.3098964989185333,
      "learning_rate": 8e-05,
      "loss": 1.8844,
      "step": 641
    },
    {
      "epoch": 0.14120752226987793,
      "grad_norm": 0.2819165885448456,
      "learning_rate": 8e-05,
      "loss": 1.7111,
      "step": 642
    },
    {
      "epoch": 0.14142747168151326,
      "grad_norm": 0.26352617144584656,
      "learning_rate": 8e-05,
      "loss": 1.7337,
      "step": 643
    },
    {
      "epoch": 0.14164742109314857,
      "grad_norm": 0.2622654139995575,
      "learning_rate": 8e-05,
      "loss": 1.7284,
      "step": 644
    },
    {
      "epoch": 0.1418673705047839,
      "grad_norm": 0.2793010473251343,
      "learning_rate": 8e-05,
      "loss": 1.8534,
      "step": 645
    },
    {
      "epoch": 0.14208731991641924,
      "grad_norm": 0.27972397208213806,
      "learning_rate": 8e-05,
      "loss": 1.7658,
      "step": 646
    },
    {
      "epoch": 0.14230726932805454,
      "grad_norm": 0.25940972566604614,
      "learning_rate": 8e-05,
      "loss": 1.6676,
      "step": 647
    },
    {
      "epoch": 0.14252721873968988,
      "grad_norm": 0.29578897356987,
      "learning_rate": 8e-05,
      "loss": 1.8002,
      "step": 648
    },
    {
      "epoch": 0.14274716815132518,
      "grad_norm": 0.2577681541442871,
      "learning_rate": 8e-05,
      "loss": 1.6154,
      "step": 649
    },
    {
      "epoch": 0.14296711756296052,
      "grad_norm": 0.2615002989768982,
      "learning_rate": 8e-05,
      "loss": 1.7539,
      "step": 650
    },
    {
      "epoch": 0.14318706697459585,
      "grad_norm": 0.26044437289237976,
      "learning_rate": 8e-05,
      "loss": 1.5284,
      "step": 651
    },
    {
      "epoch": 0.14340701638623116,
      "grad_norm": 0.28386443853378296,
      "learning_rate": 8e-05,
      "loss": 1.7188,
      "step": 652
    },
    {
      "epoch": 0.1436269657978665,
      "grad_norm": 0.2579086124897003,
      "learning_rate": 8e-05,
      "loss": 1.6758,
      "step": 653
    },
    {
      "epoch": 0.14384691520950182,
      "grad_norm": 0.263192743062973,
      "learning_rate": 8e-05,
      "loss": 1.7013,
      "step": 654
    },
    {
      "epoch": 0.14406686462113713,
      "grad_norm": 0.26551106572151184,
      "learning_rate": 8e-05,
      "loss": 1.7314,
      "step": 655
    },
    {
      "epoch": 0.14428681403277246,
      "grad_norm": 0.26143091917037964,
      "learning_rate": 8e-05,
      "loss": 1.7041,
      "step": 656
    },
    {
      "epoch": 0.1445067634444078,
      "grad_norm": 0.26432663202285767,
      "learning_rate": 8e-05,
      "loss": 1.601,
      "step": 657
    },
    {
      "epoch": 0.1447267128560431,
      "grad_norm": 0.2831920087337494,
      "learning_rate": 8e-05,
      "loss": 1.8573,
      "step": 658
    },
    {
      "epoch": 0.14494666226767844,
      "grad_norm": 0.3045855462551117,
      "learning_rate": 8e-05,
      "loss": 1.7853,
      "step": 659
    },
    {
      "epoch": 0.14516661167931375,
      "grad_norm": 0.28249257802963257,
      "learning_rate": 8e-05,
      "loss": 1.7525,
      "step": 660
    },
    {
      "epoch": 0.14538656109094908,
      "grad_norm": 0.27501189708709717,
      "learning_rate": 8e-05,
      "loss": 1.6939,
      "step": 661
    },
    {
      "epoch": 0.1456065105025844,
      "grad_norm": 0.28419750928878784,
      "learning_rate": 8e-05,
      "loss": 1.837,
      "step": 662
    },
    {
      "epoch": 0.14582645991421972,
      "grad_norm": 0.28872454166412354,
      "learning_rate": 8e-05,
      "loss": 1.623,
      "step": 663
    },
    {
      "epoch": 0.14604640932585505,
      "grad_norm": 0.2926316559314728,
      "learning_rate": 8e-05,
      "loss": 1.7438,
      "step": 664
    },
    {
      "epoch": 0.1462663587374904,
      "grad_norm": 0.2716543972492218,
      "learning_rate": 8e-05,
      "loss": 1.8925,
      "step": 665
    },
    {
      "epoch": 0.1464863081491257,
      "grad_norm": 0.2707289159297943,
      "learning_rate": 8e-05,
      "loss": 1.8218,
      "step": 666
    },
    {
      "epoch": 0.14670625756076103,
      "grad_norm": 0.2609579265117645,
      "learning_rate": 8e-05,
      "loss": 1.4612,
      "step": 667
    },
    {
      "epoch": 0.14692620697239636,
      "grad_norm": 0.2958548367023468,
      "learning_rate": 8e-05,
      "loss": 1.6191,
      "step": 668
    },
    {
      "epoch": 0.14714615638403167,
      "grad_norm": 0.2585492730140686,
      "learning_rate": 8e-05,
      "loss": 1.7161,
      "step": 669
    },
    {
      "epoch": 0.147366105795667,
      "grad_norm": 0.2637808322906494,
      "learning_rate": 8e-05,
      "loss": 1.6534,
      "step": 670
    },
    {
      "epoch": 0.1475860552073023,
      "grad_norm": 0.2885671854019165,
      "learning_rate": 8e-05,
      "loss": 1.7663,
      "step": 671
    },
    {
      "epoch": 0.14780600461893764,
      "grad_norm": 0.27028244733810425,
      "learning_rate": 8e-05,
      "loss": 1.7718,
      "step": 672
    },
    {
      "epoch": 0.14802595403057298,
      "grad_norm": 0.27723586559295654,
      "learning_rate": 8e-05,
      "loss": 1.7762,
      "step": 673
    },
    {
      "epoch": 0.14824590344220828,
      "grad_norm": 0.26336848735809326,
      "learning_rate": 8e-05,
      "loss": 1.6114,
      "step": 674
    },
    {
      "epoch": 0.14846585285384362,
      "grad_norm": 0.26031750440597534,
      "learning_rate": 8e-05,
      "loss": 1.7259,
      "step": 675
    },
    {
      "epoch": 0.14868580226547895,
      "grad_norm": 0.30176040530204773,
      "learning_rate": 8e-05,
      "loss": 1.7007,
      "step": 676
    },
    {
      "epoch": 0.14890575167711426,
      "grad_norm": 0.25952771306037903,
      "learning_rate": 8e-05,
      "loss": 1.6573,
      "step": 677
    },
    {
      "epoch": 0.1491257010887496,
      "grad_norm": 0.2727009356021881,
      "learning_rate": 8e-05,
      "loss": 1.7725,
      "step": 678
    },
    {
      "epoch": 0.14934565050038492,
      "grad_norm": 0.26398420333862305,
      "learning_rate": 8e-05,
      "loss": 1.7245,
      "step": 679
    },
    {
      "epoch": 0.14956559991202023,
      "grad_norm": 0.273967981338501,
      "learning_rate": 8e-05,
      "loss": 1.7231,
      "step": 680
    },
    {
      "epoch": 0.14978554932365556,
      "grad_norm": 0.27241724729537964,
      "learning_rate": 8e-05,
      "loss": 1.6896,
      "step": 681
    },
    {
      "epoch": 0.15000549873529087,
      "grad_norm": 0.26996085047721863,
      "learning_rate": 8e-05,
      "loss": 1.6767,
      "step": 682
    },
    {
      "epoch": 0.1502254481469262,
      "grad_norm": 0.27165672183036804,
      "learning_rate": 8e-05,
      "loss": 1.7747,
      "step": 683
    },
    {
      "epoch": 0.15044539755856154,
      "grad_norm": 0.26840028166770935,
      "learning_rate": 8e-05,
      "loss": 1.7616,
      "step": 684
    },
    {
      "epoch": 0.15066534697019685,
      "grad_norm": 0.27101555466651917,
      "learning_rate": 8e-05,
      "loss": 1.622,
      "step": 685
    },
    {
      "epoch": 0.15088529638183218,
      "grad_norm": 0.2691043019294739,
      "learning_rate": 8e-05,
      "loss": 1.7514,
      "step": 686
    },
    {
      "epoch": 0.1511052457934675,
      "grad_norm": 0.2926357090473175,
      "learning_rate": 8e-05,
      "loss": 1.6953,
      "step": 687
    },
    {
      "epoch": 0.15132519520510282,
      "grad_norm": 0.2730226516723633,
      "learning_rate": 8e-05,
      "loss": 1.6286,
      "step": 688
    },
    {
      "epoch": 0.15154514461673815,
      "grad_norm": 0.2618841826915741,
      "learning_rate": 8e-05,
      "loss": 1.7194,
      "step": 689
    },
    {
      "epoch": 0.1517650940283735,
      "grad_norm": 0.2584119737148285,
      "learning_rate": 8e-05,
      "loss": 1.6032,
      "step": 690
    },
    {
      "epoch": 0.1519850434400088,
      "grad_norm": 0.26063093543052673,
      "learning_rate": 8e-05,
      "loss": 1.63,
      "step": 691
    },
    {
      "epoch": 0.15220499285164413,
      "grad_norm": 0.267938494682312,
      "learning_rate": 8e-05,
      "loss": 1.7087,
      "step": 692
    },
    {
      "epoch": 0.15242494226327943,
      "grad_norm": 0.2709169089794159,
      "learning_rate": 8e-05,
      "loss": 1.6663,
      "step": 693
    },
    {
      "epoch": 0.15264489167491477,
      "grad_norm": 0.3015836775302887,
      "learning_rate": 8e-05,
      "loss": 1.6797,
      "step": 694
    },
    {
      "epoch": 0.1528648410865501,
      "grad_norm": 0.27824944257736206,
      "learning_rate": 8e-05,
      "loss": 1.7972,
      "step": 695
    },
    {
      "epoch": 0.1530847904981854,
      "grad_norm": 0.31089073419570923,
      "learning_rate": 8e-05,
      "loss": 1.7352,
      "step": 696
    },
    {
      "epoch": 0.15330473990982074,
      "grad_norm": 0.2804546654224396,
      "learning_rate": 8e-05,
      "loss": 1.6898,
      "step": 697
    },
    {
      "epoch": 0.15352468932145608,
      "grad_norm": 0.2804514765739441,
      "learning_rate": 8e-05,
      "loss": 1.8409,
      "step": 698
    },
    {
      "epoch": 0.15374463873309138,
      "grad_norm": 0.31666815280914307,
      "learning_rate": 8e-05,
      "loss": 1.6569,
      "step": 699
    },
    {
      "epoch": 0.15396458814472672,
      "grad_norm": 0.2846215069293976,
      "learning_rate": 8e-05,
      "loss": 1.8081,
      "step": 700
    },
    {
      "epoch": 0.15418453755636205,
      "grad_norm": 0.2656068801879883,
      "learning_rate": 8e-05,
      "loss": 1.5747,
      "step": 701
    },
    {
      "epoch": 0.15440448696799736,
      "grad_norm": 0.2633317708969116,
      "learning_rate": 8e-05,
      "loss": 1.6027,
      "step": 702
    },
    {
      "epoch": 0.1546244363796327,
      "grad_norm": 0.2669740319252014,
      "learning_rate": 8e-05,
      "loss": 1.6964,
      "step": 703
    },
    {
      "epoch": 0.154844385791268,
      "grad_norm": 0.2878497540950775,
      "learning_rate": 8e-05,
      "loss": 1.677,
      "step": 704
    },
    {
      "epoch": 0.15506433520290333,
      "grad_norm": 0.2624325156211853,
      "learning_rate": 8e-05,
      "loss": 1.6247,
      "step": 705
    },
    {
      "epoch": 0.15528428461453866,
      "grad_norm": 0.2894291579723358,
      "learning_rate": 8e-05,
      "loss": 1.7271,
      "step": 706
    },
    {
      "epoch": 0.15550423402617397,
      "grad_norm": 0.2924456298351288,
      "learning_rate": 8e-05,
      "loss": 1.7475,
      "step": 707
    },
    {
      "epoch": 0.1557241834378093,
      "grad_norm": 0.2519112229347229,
      "learning_rate": 8e-05,
      "loss": 1.6306,
      "step": 708
    },
    {
      "epoch": 0.15594413284944464,
      "grad_norm": 0.2831405699253082,
      "learning_rate": 8e-05,
      "loss": 1.7571,
      "step": 709
    },
    {
      "epoch": 0.15616408226107995,
      "grad_norm": 0.2804257273674011,
      "learning_rate": 8e-05,
      "loss": 1.6721,
      "step": 710
    },
    {
      "epoch": 0.15638403167271528,
      "grad_norm": 0.27130362391471863,
      "learning_rate": 8e-05,
      "loss": 1.7451,
      "step": 711
    },
    {
      "epoch": 0.1566039810843506,
      "grad_norm": 0.27843937277793884,
      "learning_rate": 8e-05,
      "loss": 1.7187,
      "step": 712
    },
    {
      "epoch": 0.15682393049598592,
      "grad_norm": 0.26205387711524963,
      "learning_rate": 8e-05,
      "loss": 1.7667,
      "step": 713
    },
    {
      "epoch": 0.15704387990762125,
      "grad_norm": 0.25978967547416687,
      "learning_rate": 8e-05,
      "loss": 1.6595,
      "step": 714
    },
    {
      "epoch": 0.15726382931925656,
      "grad_norm": 0.26331478357315063,
      "learning_rate": 8e-05,
      "loss": 1.8067,
      "step": 715
    },
    {
      "epoch": 0.1574837787308919,
      "grad_norm": 0.26023924350738525,
      "learning_rate": 8e-05,
      "loss": 1.8533,
      "step": 716
    },
    {
      "epoch": 0.15770372814252723,
      "grad_norm": 0.27147844433784485,
      "learning_rate": 8e-05,
      "loss": 1.6309,
      "step": 717
    },
    {
      "epoch": 0.15792367755416253,
      "grad_norm": 0.286035418510437,
      "learning_rate": 8e-05,
      "loss": 1.72,
      "step": 718
    },
    {
      "epoch": 0.15814362696579787,
      "grad_norm": 0.3167229890823364,
      "learning_rate": 8e-05,
      "loss": 1.9007,
      "step": 719
    },
    {
      "epoch": 0.1583635763774332,
      "grad_norm": 0.283975750207901,
      "learning_rate": 8e-05,
      "loss": 1.6662,
      "step": 720
    },
    {
      "epoch": 0.1585835257890685,
      "grad_norm": 0.2812137007713318,
      "learning_rate": 8e-05,
      "loss": 1.7651,
      "step": 721
    },
    {
      "epoch": 0.15880347520070384,
      "grad_norm": 0.2737642526626587,
      "learning_rate": 8e-05,
      "loss": 1.7679,
      "step": 722
    },
    {
      "epoch": 0.15902342461233915,
      "grad_norm": 0.30812978744506836,
      "learning_rate": 8e-05,
      "loss": 1.8408,
      "step": 723
    },
    {
      "epoch": 0.15924337402397448,
      "grad_norm": 0.27026352286338806,
      "learning_rate": 8e-05,
      "loss": 1.7362,
      "step": 724
    },
    {
      "epoch": 0.15946332343560982,
      "grad_norm": 0.2788861393928528,
      "learning_rate": 8e-05,
      "loss": 1.8371,
      "step": 725
    },
    {
      "epoch": 0.15968327284724512,
      "grad_norm": 0.2623996138572693,
      "learning_rate": 8e-05,
      "loss": 1.5855,
      "step": 726
    },
    {
      "epoch": 0.15990322225888046,
      "grad_norm": 0.2764820158481598,
      "learning_rate": 8e-05,
      "loss": 1.8185,
      "step": 727
    },
    {
      "epoch": 0.1601231716705158,
      "grad_norm": 0.27394816279411316,
      "learning_rate": 8e-05,
      "loss": 1.641,
      "step": 728
    },
    {
      "epoch": 0.1603431210821511,
      "grad_norm": 0.2726307511329651,
      "learning_rate": 8e-05,
      "loss": 1.6128,
      "step": 729
    },
    {
      "epoch": 0.16056307049378643,
      "grad_norm": 0.28221258521080017,
      "learning_rate": 8e-05,
      "loss": 1.8413,
      "step": 730
    },
    {
      "epoch": 0.16078301990542176,
      "grad_norm": 0.2649543881416321,
      "learning_rate": 8e-05,
      "loss": 1.5707,
      "step": 731
    },
    {
      "epoch": 0.16100296931705707,
      "grad_norm": 0.2659435570240021,
      "learning_rate": 8e-05,
      "loss": 1.6761,
      "step": 732
    },
    {
      "epoch": 0.1612229187286924,
      "grad_norm": 0.3131570518016815,
      "learning_rate": 8e-05,
      "loss": 1.9439,
      "step": 733
    },
    {
      "epoch": 0.1614428681403277,
      "grad_norm": 0.263069748878479,
      "learning_rate": 8e-05,
      "loss": 1.7069,
      "step": 734
    },
    {
      "epoch": 0.16166281755196305,
      "grad_norm": 0.2708505392074585,
      "learning_rate": 8e-05,
      "loss": 1.8031,
      "step": 735
    },
    {
      "epoch": 0.16188276696359838,
      "grad_norm": 0.26446613669395447,
      "learning_rate": 8e-05,
      "loss": 1.6419,
      "step": 736
    },
    {
      "epoch": 0.16210271637523369,
      "grad_norm": 0.27720367908477783,
      "learning_rate": 8e-05,
      "loss": 1.8291,
      "step": 737
    },
    {
      "epoch": 0.16232266578686902,
      "grad_norm": 0.25950226187705994,
      "learning_rate": 8e-05,
      "loss": 1.7498,
      "step": 738
    },
    {
      "epoch": 0.16254261519850435,
      "grad_norm": 0.25445327162742615,
      "learning_rate": 8e-05,
      "loss": 1.6804,
      "step": 739
    },
    {
      "epoch": 0.16276256461013966,
      "grad_norm": 0.2868766784667969,
      "learning_rate": 8e-05,
      "loss": 1.8058,
      "step": 740
    },
    {
      "epoch": 0.162982514021775,
      "grad_norm": 0.2775559425354004,
      "learning_rate": 8e-05,
      "loss": 1.7971,
      "step": 741
    },
    {
      "epoch": 0.16320246343341033,
      "grad_norm": 0.2822381556034088,
      "learning_rate": 8e-05,
      "loss": 1.7294,
      "step": 742
    },
    {
      "epoch": 0.16342241284504563,
      "grad_norm": 0.26617857813835144,
      "learning_rate": 8e-05,
      "loss": 1.8011,
      "step": 743
    },
    {
      "epoch": 0.16364236225668097,
      "grad_norm": 0.25615090131759644,
      "learning_rate": 8e-05,
      "loss": 1.6328,
      "step": 744
    },
    {
      "epoch": 0.16386231166831627,
      "grad_norm": 0.25831338763237,
      "learning_rate": 8e-05,
      "loss": 1.6174,
      "step": 745
    },
    {
      "epoch": 0.1640822610799516,
      "grad_norm": 0.2707291543483734,
      "learning_rate": 8e-05,
      "loss": 1.8217,
      "step": 746
    },
    {
      "epoch": 0.16430221049158694,
      "grad_norm": 0.3028862774372101,
      "learning_rate": 8e-05,
      "loss": 1.5852,
      "step": 747
    },
    {
      "epoch": 0.16452215990322225,
      "grad_norm": 0.26598575711250305,
      "learning_rate": 8e-05,
      "loss": 1.7213,
      "step": 748
    },
    {
      "epoch": 0.16474210931485758,
      "grad_norm": 0.27408871054649353,
      "learning_rate": 8e-05,
      "loss": 1.7109,
      "step": 749
    },
    {
      "epoch": 0.16496205872649292,
      "grad_norm": 0.27065837383270264,
      "learning_rate": 8e-05,
      "loss": 1.6696,
      "step": 750
    },
    {
      "epoch": 0.16518200813812822,
      "grad_norm": 0.2721879184246063,
      "learning_rate": 8e-05,
      "loss": 1.7055,
      "step": 751
    },
    {
      "epoch": 0.16540195754976356,
      "grad_norm": 0.29569125175476074,
      "learning_rate": 8e-05,
      "loss": 1.5921,
      "step": 752
    },
    {
      "epoch": 0.1656219069613989,
      "grad_norm": 0.28580978512763977,
      "learning_rate": 8e-05,
      "loss": 1.7518,
      "step": 753
    },
    {
      "epoch": 0.1658418563730342,
      "grad_norm": 0.2869469225406647,
      "learning_rate": 8e-05,
      "loss": 1.8164,
      "step": 754
    },
    {
      "epoch": 0.16606180578466953,
      "grad_norm": 0.2796071171760559,
      "learning_rate": 8e-05,
      "loss": 1.8325,
      "step": 755
    },
    {
      "epoch": 0.16628175519630484,
      "grad_norm": 0.27365031838417053,
      "learning_rate": 8e-05,
      "loss": 1.7287,
      "step": 756
    },
    {
      "epoch": 0.16650170460794017,
      "grad_norm": 0.2524491846561432,
      "learning_rate": 8e-05,
      "loss": 1.5379,
      "step": 757
    },
    {
      "epoch": 0.1667216540195755,
      "grad_norm": 0.259860634803772,
      "learning_rate": 8e-05,
      "loss": 1.5204,
      "step": 758
    },
    {
      "epoch": 0.1669416034312108,
      "grad_norm": 0.2714100182056427,
      "learning_rate": 8e-05,
      "loss": 1.7245,
      "step": 759
    },
    {
      "epoch": 0.16716155284284615,
      "grad_norm": 0.2729417383670807,
      "learning_rate": 8e-05,
      "loss": 1.6889,
      "step": 760
    },
    {
      "epoch": 0.16738150225448148,
      "grad_norm": 0.2753896415233612,
      "learning_rate": 8e-05,
      "loss": 1.7345,
      "step": 761
    },
    {
      "epoch": 0.16760145166611679,
      "grad_norm": 0.2830727994441986,
      "learning_rate": 8e-05,
      "loss": 1.6884,
      "step": 762
    },
    {
      "epoch": 0.16782140107775212,
      "grad_norm": 0.27818116545677185,
      "learning_rate": 8e-05,
      "loss": 1.7819,
      "step": 763
    },
    {
      "epoch": 0.16804135048938745,
      "grad_norm": 0.2601570785045624,
      "learning_rate": 8e-05,
      "loss": 1.6323,
      "step": 764
    },
    {
      "epoch": 0.16826129990102276,
      "grad_norm": 0.2638706564903259,
      "learning_rate": 8e-05,
      "loss": 1.5957,
      "step": 765
    },
    {
      "epoch": 0.1684812493126581,
      "grad_norm": 0.2798631489276886,
      "learning_rate": 8e-05,
      "loss": 1.7946,
      "step": 766
    },
    {
      "epoch": 0.1687011987242934,
      "grad_norm": 0.2975100874900818,
      "learning_rate": 8e-05,
      "loss": 1.871,
      "step": 767
    },
    {
      "epoch": 0.16892114813592873,
      "grad_norm": 0.28308364748954773,
      "learning_rate": 8e-05,
      "loss": 1.7184,
      "step": 768
    },
    {
      "epoch": 0.16914109754756407,
      "grad_norm": 0.2594911456108093,
      "learning_rate": 8e-05,
      "loss": 1.5867,
      "step": 769
    },
    {
      "epoch": 0.16936104695919937,
      "grad_norm": 0.27594470977783203,
      "learning_rate": 8e-05,
      "loss": 1.7722,
      "step": 770
    },
    {
      "epoch": 0.1695809963708347,
      "grad_norm": 0.2783298194408417,
      "learning_rate": 8e-05,
      "loss": 1.7891,
      "step": 771
    },
    {
      "epoch": 0.16980094578247004,
      "grad_norm": 0.2863733172416687,
      "learning_rate": 8e-05,
      "loss": 1.6274,
      "step": 772
    },
    {
      "epoch": 0.17002089519410535,
      "grad_norm": 0.27953147888183594,
      "learning_rate": 8e-05,
      "loss": 1.7287,
      "step": 773
    },
    {
      "epoch": 0.17024084460574068,
      "grad_norm": 0.2736772894859314,
      "learning_rate": 8e-05,
      "loss": 1.6802,
      "step": 774
    },
    {
      "epoch": 0.17046079401737602,
      "grad_norm": 0.27663713693618774,
      "learning_rate": 8e-05,
      "loss": 1.6607,
      "step": 775
    },
    {
      "epoch": 0.17068074342901132,
      "grad_norm": 0.3064086437225342,
      "learning_rate": 8e-05,
      "loss": 1.8244,
      "step": 776
    },
    {
      "epoch": 0.17090069284064666,
      "grad_norm": 0.29848581552505493,
      "learning_rate": 8e-05,
      "loss": 1.7702,
      "step": 777
    },
    {
      "epoch": 0.17112064225228196,
      "grad_norm": 0.3101220726966858,
      "learning_rate": 8e-05,
      "loss": 1.7714,
      "step": 778
    },
    {
      "epoch": 0.1713405916639173,
      "grad_norm": 0.2754581868648529,
      "learning_rate": 8e-05,
      "loss": 1.6367,
      "step": 779
    },
    {
      "epoch": 0.17156054107555263,
      "grad_norm": 0.2706362307071686,
      "learning_rate": 8e-05,
      "loss": 1.6236,
      "step": 780
    },
    {
      "epoch": 0.17178049048718794,
      "grad_norm": 0.29135438799858093,
      "learning_rate": 8e-05,
      "loss": 1.8478,
      "step": 781
    },
    {
      "epoch": 0.17200043989882327,
      "grad_norm": 0.2751868963241577,
      "learning_rate": 8e-05,
      "loss": 1.751,
      "step": 782
    },
    {
      "epoch": 0.1722203893104586,
      "grad_norm": 0.2871004045009613,
      "learning_rate": 8e-05,
      "loss": 1.6793,
      "step": 783
    },
    {
      "epoch": 0.1724403387220939,
      "grad_norm": 0.31024861335754395,
      "learning_rate": 8e-05,
      "loss": 1.7419,
      "step": 784
    },
    {
      "epoch": 0.17266028813372924,
      "grad_norm": 0.2917722165584564,
      "learning_rate": 8e-05,
      "loss": 1.8913,
      "step": 785
    },
    {
      "epoch": 0.17288023754536458,
      "grad_norm": 0.25443291664123535,
      "learning_rate": 8e-05,
      "loss": 1.6991,
      "step": 786
    },
    {
      "epoch": 0.17310018695699989,
      "grad_norm": 0.2827921211719513,
      "learning_rate": 8e-05,
      "loss": 1.8408,
      "step": 787
    },
    {
      "epoch": 0.17332013636863522,
      "grad_norm": 0.26190435886383057,
      "learning_rate": 8e-05,
      "loss": 1.6841,
      "step": 788
    },
    {
      "epoch": 0.17354008578027053,
      "grad_norm": 0.31557098031044006,
      "learning_rate": 8e-05,
      "loss": 1.8838,
      "step": 789
    },
    {
      "epoch": 0.17376003519190586,
      "grad_norm": 0.27622002363204956,
      "learning_rate": 8e-05,
      "loss": 1.6558,
      "step": 790
    },
    {
      "epoch": 0.1739799846035412,
      "grad_norm": 0.3161294758319855,
      "learning_rate": 8e-05,
      "loss": 1.5771,
      "step": 791
    },
    {
      "epoch": 0.1741999340151765,
      "grad_norm": 0.3014603555202484,
      "learning_rate": 8e-05,
      "loss": 1.7742,
      "step": 792
    },
    {
      "epoch": 0.17441988342681183,
      "grad_norm": 0.24996457993984222,
      "learning_rate": 8e-05,
      "loss": 1.5667,
      "step": 793
    },
    {
      "epoch": 0.17463983283844717,
      "grad_norm": 0.29180648922920227,
      "learning_rate": 8e-05,
      "loss": 1.7703,
      "step": 794
    },
    {
      "epoch": 0.17485978225008247,
      "grad_norm": 0.26707547903060913,
      "learning_rate": 8e-05,
      "loss": 1.7964,
      "step": 795
    },
    {
      "epoch": 0.1750797316617178,
      "grad_norm": 0.24924349784851074,
      "learning_rate": 8e-05,
      "loss": 1.6619,
      "step": 796
    },
    {
      "epoch": 0.17529968107335314,
      "grad_norm": 0.29872292280197144,
      "learning_rate": 8e-05,
      "loss": 1.8561,
      "step": 797
    },
    {
      "epoch": 0.17551963048498845,
      "grad_norm": 0.2770175337791443,
      "learning_rate": 8e-05,
      "loss": 1.6352,
      "step": 798
    },
    {
      "epoch": 0.17573957989662378,
      "grad_norm": 0.26890453696250916,
      "learning_rate": 8e-05,
      "loss": 1.885,
      "step": 799
    },
    {
      "epoch": 0.1759595293082591,
      "grad_norm": 0.2830483317375183,
      "learning_rate": 8e-05,
      "loss": 1.6029,
      "step": 800
    },
    {
      "epoch": 0.17617947871989442,
      "grad_norm": 0.27421921491622925,
      "learning_rate": 8e-05,
      "loss": 1.6845,
      "step": 801
    },
    {
      "epoch": 0.17639942813152976,
      "grad_norm": 0.29273220896720886,
      "learning_rate": 8e-05,
      "loss": 1.8135,
      "step": 802
    },
    {
      "epoch": 0.17661937754316506,
      "grad_norm": 0.2675575315952301,
      "learning_rate": 8e-05,
      "loss": 1.571,
      "step": 803
    },
    {
      "epoch": 0.1768393269548004,
      "grad_norm": 0.2821138799190521,
      "learning_rate": 8e-05,
      "loss": 1.9244,
      "step": 804
    },
    {
      "epoch": 0.17705927636643573,
      "grad_norm": 0.28082311153411865,
      "learning_rate": 8e-05,
      "loss": 1.7395,
      "step": 805
    },
    {
      "epoch": 0.17727922577807104,
      "grad_norm": 0.27897313237190247,
      "learning_rate": 8e-05,
      "loss": 1.6347,
      "step": 806
    },
    {
      "epoch": 0.17749917518970637,
      "grad_norm": 0.27358707785606384,
      "learning_rate": 8e-05,
      "loss": 1.7643,
      "step": 807
    },
    {
      "epoch": 0.1777191246013417,
      "grad_norm": 0.284059077501297,
      "learning_rate": 8e-05,
      "loss": 1.5789,
      "step": 808
    },
    {
      "epoch": 0.177939074012977,
      "grad_norm": 0.26125824451446533,
      "learning_rate": 8e-05,
      "loss": 1.7029,
      "step": 809
    },
    {
      "epoch": 0.17815902342461234,
      "grad_norm": 0.26438888907432556,
      "learning_rate": 8e-05,
      "loss": 1.6424,
      "step": 810
    },
    {
      "epoch": 0.17837897283624765,
      "grad_norm": 0.2746163010597229,
      "learning_rate": 8e-05,
      "loss": 1.7992,
      "step": 811
    },
    {
      "epoch": 0.17859892224788299,
      "grad_norm": 0.27717527747154236,
      "learning_rate": 8e-05,
      "loss": 1.7603,
      "step": 812
    },
    {
      "epoch": 0.17881887165951832,
      "grad_norm": 0.28336596488952637,
      "learning_rate": 8e-05,
      "loss": 1.7133,
      "step": 813
    },
    {
      "epoch": 0.17903882107115363,
      "grad_norm": 0.2701306939125061,
      "learning_rate": 8e-05,
      "loss": 1.7724,
      "step": 814
    },
    {
      "epoch": 0.17925877048278896,
      "grad_norm": 0.2807336449623108,
      "learning_rate": 8e-05,
      "loss": 1.7832,
      "step": 815
    },
    {
      "epoch": 0.1794787198944243,
      "grad_norm": 0.2847912907600403,
      "learning_rate": 8e-05,
      "loss": 1.7042,
      "step": 816
    },
    {
      "epoch": 0.1796986693060596,
      "grad_norm": 0.2836345434188843,
      "learning_rate": 8e-05,
      "loss": 1.8506,
      "step": 817
    },
    {
      "epoch": 0.17991861871769493,
      "grad_norm": 0.30620551109313965,
      "learning_rate": 8e-05,
      "loss": 1.7695,
      "step": 818
    },
    {
      "epoch": 0.18013856812933027,
      "grad_norm": 0.2698993980884552,
      "learning_rate": 8e-05,
      "loss": 1.6388,
      "step": 819
    },
    {
      "epoch": 0.18035851754096557,
      "grad_norm": 0.2937266528606415,
      "learning_rate": 8e-05,
      "loss": 1.8648,
      "step": 820
    },
    {
      "epoch": 0.1805784669526009,
      "grad_norm": 0.2661988139152527,
      "learning_rate": 8e-05,
      "loss": 1.7563,
      "step": 821
    },
    {
      "epoch": 0.18079841636423621,
      "grad_norm": 0.2944018840789795,
      "learning_rate": 8e-05,
      "loss": 1.882,
      "step": 822
    },
    {
      "epoch": 0.18101836577587155,
      "grad_norm": 0.2774435579776764,
      "learning_rate": 8e-05,
      "loss": 1.8117,
      "step": 823
    },
    {
      "epoch": 0.18123831518750688,
      "grad_norm": 0.27865204215049744,
      "learning_rate": 8e-05,
      "loss": 1.8815,
      "step": 824
    },
    {
      "epoch": 0.1814582645991422,
      "grad_norm": 0.26444011926651,
      "learning_rate": 8e-05,
      "loss": 1.5844,
      "step": 825
    },
    {
      "epoch": 0.18167821401077752,
      "grad_norm": 0.27044716477394104,
      "learning_rate": 8e-05,
      "loss": 1.7403,
      "step": 826
    },
    {
      "epoch": 0.18189816342241286,
      "grad_norm": 0.28727805614471436,
      "learning_rate": 8e-05,
      "loss": 1.8556,
      "step": 827
    },
    {
      "epoch": 0.18211811283404816,
      "grad_norm": 0.26131972670555115,
      "learning_rate": 8e-05,
      "loss": 1.7727,
      "step": 828
    },
    {
      "epoch": 0.1823380622456835,
      "grad_norm": 0.269638329744339,
      "learning_rate": 8e-05,
      "loss": 1.6795,
      "step": 829
    },
    {
      "epoch": 0.18255801165731883,
      "grad_norm": 0.2671653628349304,
      "learning_rate": 8e-05,
      "loss": 1.6811,
      "step": 830
    },
    {
      "epoch": 0.18277796106895414,
      "grad_norm": 0.2659014165401459,
      "learning_rate": 8e-05,
      "loss": 1.7166,
      "step": 831
    },
    {
      "epoch": 0.18299791048058947,
      "grad_norm": 0.2719801962375641,
      "learning_rate": 8e-05,
      "loss": 1.6938,
      "step": 832
    },
    {
      "epoch": 0.18321785989222478,
      "grad_norm": 0.3272366225719452,
      "learning_rate": 8e-05,
      "loss": 1.8213,
      "step": 833
    },
    {
      "epoch": 0.1834378093038601,
      "grad_norm": 0.2635113000869751,
      "learning_rate": 8e-05,
      "loss": 1.6291,
      "step": 834
    },
    {
      "epoch": 0.18365775871549544,
      "grad_norm": 0.29401281476020813,
      "learning_rate": 8e-05,
      "loss": 1.8234,
      "step": 835
    },
    {
      "epoch": 0.18387770812713075,
      "grad_norm": 0.29188451170921326,
      "learning_rate": 8e-05,
      "loss": 1.7359,
      "step": 836
    },
    {
      "epoch": 0.18409765753876609,
      "grad_norm": 0.2688080072402954,
      "learning_rate": 8e-05,
      "loss": 1.7088,
      "step": 837
    },
    {
      "epoch": 0.18431760695040142,
      "grad_norm": 0.27907344698905945,
      "learning_rate": 8e-05,
      "loss": 1.6762,
      "step": 838
    },
    {
      "epoch": 0.18453755636203673,
      "grad_norm": 0.2875908315181732,
      "learning_rate": 8e-05,
      "loss": 1.7612,
      "step": 839
    },
    {
      "epoch": 0.18475750577367206,
      "grad_norm": 0.2683177888393402,
      "learning_rate": 8e-05,
      "loss": 1.5965,
      "step": 840
    },
    {
      "epoch": 0.18497745518530737,
      "grad_norm": 0.29948660731315613,
      "learning_rate": 8e-05,
      "loss": 1.7358,
      "step": 841
    },
    {
      "epoch": 0.1851974045969427,
      "grad_norm": 0.28153204917907715,
      "learning_rate": 8e-05,
      "loss": 1.8089,
      "step": 842
    },
    {
      "epoch": 0.18541735400857803,
      "grad_norm": 0.29185283184051514,
      "learning_rate": 8e-05,
      "loss": 1.8357,
      "step": 843
    },
    {
      "epoch": 0.18563730342021334,
      "grad_norm": 0.27565860748291016,
      "learning_rate": 8e-05,
      "loss": 1.9565,
      "step": 844
    },
    {
      "epoch": 0.18585725283184867,
      "grad_norm": 0.2811479866504669,
      "learning_rate": 8e-05,
      "loss": 1.8493,
      "step": 845
    },
    {
      "epoch": 0.186077202243484,
      "grad_norm": 0.271893173456192,
      "learning_rate": 8e-05,
      "loss": 1.7622,
      "step": 846
    },
    {
      "epoch": 0.18629715165511931,
      "grad_norm": 0.26383113861083984,
      "learning_rate": 8e-05,
      "loss": 1.7392,
      "step": 847
    },
    {
      "epoch": 0.18651710106675465,
      "grad_norm": 0.2863881289958954,
      "learning_rate": 8e-05,
      "loss": 1.7367,
      "step": 848
    },
    {
      "epoch": 0.18673705047838998,
      "grad_norm": 0.28036433458328247,
      "learning_rate": 8e-05,
      "loss": 1.6587,
      "step": 849
    },
    {
      "epoch": 0.1869569998900253,
      "grad_norm": 0.2938581705093384,
      "learning_rate": 8e-05,
      "loss": 1.7411,
      "step": 850
    },
    {
      "epoch": 0.18717694930166062,
      "grad_norm": 0.27487799525260925,
      "learning_rate": 8e-05,
      "loss": 1.8054,
      "step": 851
    },
    {
      "epoch": 0.18739689871329593,
      "grad_norm": 0.2693670690059662,
      "learning_rate": 8e-05,
      "loss": 1.7361,
      "step": 852
    },
    {
      "epoch": 0.18761684812493126,
      "grad_norm": 0.2999705970287323,
      "learning_rate": 8e-05,
      "loss": 1.909,
      "step": 853
    },
    {
      "epoch": 0.1878367975365666,
      "grad_norm": 0.28235265612602234,
      "learning_rate": 8e-05,
      "loss": 1.8611,
      "step": 854
    },
    {
      "epoch": 0.1880567469482019,
      "grad_norm": 0.28417298197746277,
      "learning_rate": 8e-05,
      "loss": 1.683,
      "step": 855
    },
    {
      "epoch": 0.18827669635983724,
      "grad_norm": 0.2697356045246124,
      "learning_rate": 8e-05,
      "loss": 1.7138,
      "step": 856
    },
    {
      "epoch": 0.18849664577147257,
      "grad_norm": 0.26900357007980347,
      "learning_rate": 8e-05,
      "loss": 1.5579,
      "step": 857
    },
    {
      "epoch": 0.18871659518310788,
      "grad_norm": 0.259941041469574,
      "learning_rate": 8e-05,
      "loss": 1.7106,
      "step": 858
    },
    {
      "epoch": 0.1889365445947432,
      "grad_norm": 0.26958781480789185,
      "learning_rate": 8e-05,
      "loss": 1.6454,
      "step": 859
    },
    {
      "epoch": 0.18915649400637854,
      "grad_norm": 0.26425305008888245,
      "learning_rate": 8e-05,
      "loss": 1.6408,
      "step": 860
    },
    {
      "epoch": 0.18937644341801385,
      "grad_norm": 0.26996907591819763,
      "learning_rate": 8e-05,
      "loss": 1.6949,
      "step": 861
    },
    {
      "epoch": 0.18959639282964919,
      "grad_norm": 0.25882837176322937,
      "learning_rate": 8e-05,
      "loss": 1.6142,
      "step": 862
    },
    {
      "epoch": 0.1898163422412845,
      "grad_norm": 0.28000783920288086,
      "learning_rate": 8e-05,
      "loss": 1.8007,
      "step": 863
    },
    {
      "epoch": 0.19003629165291983,
      "grad_norm": 0.2744222581386566,
      "learning_rate": 8e-05,
      "loss": 1.6604,
      "step": 864
    },
    {
      "epoch": 0.19025624106455516,
      "grad_norm": 0.2791576683521271,
      "learning_rate": 8e-05,
      "loss": 1.7061,
      "step": 865
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.27878084778785706,
      "learning_rate": 8e-05,
      "loss": 1.8604,
      "step": 866
    },
    {
      "epoch": 0.1906961398878258,
      "grad_norm": 0.3818608820438385,
      "learning_rate": 8e-05,
      "loss": 1.8616,
      "step": 867
    },
    {
      "epoch": 0.19091608929946113,
      "grad_norm": 0.27952665090560913,
      "learning_rate": 8e-05,
      "loss": 1.7616,
      "step": 868
    },
    {
      "epoch": 0.19113603871109644,
      "grad_norm": 0.2711832523345947,
      "learning_rate": 8e-05,
      "loss": 1.7974,
      "step": 869
    },
    {
      "epoch": 0.19135598812273177,
      "grad_norm": 0.2572176456451416,
      "learning_rate": 8e-05,
      "loss": 1.584,
      "step": 870
    },
    {
      "epoch": 0.1915759375343671,
      "grad_norm": 0.2847760319709778,
      "learning_rate": 8e-05,
      "loss": 1.8598,
      "step": 871
    },
    {
      "epoch": 0.19179588694600241,
      "grad_norm": 0.29798731207847595,
      "learning_rate": 8e-05,
      "loss": 1.6689,
      "step": 872
    },
    {
      "epoch": 0.19201583635763775,
      "grad_norm": 0.2674097716808319,
      "learning_rate": 8e-05,
      "loss": 1.6694,
      "step": 873
    },
    {
      "epoch": 0.19223578576927305,
      "grad_norm": 0.27707335352897644,
      "learning_rate": 8e-05,
      "loss": 1.7251,
      "step": 874
    },
    {
      "epoch": 0.1924557351809084,
      "grad_norm": 0.2801666259765625,
      "learning_rate": 8e-05,
      "loss": 1.7251,
      "step": 875
    },
    {
      "epoch": 0.19267568459254372,
      "grad_norm": 0.2656191885471344,
      "learning_rate": 8e-05,
      "loss": 1.6232,
      "step": 876
    },
    {
      "epoch": 0.19289563400417903,
      "grad_norm": 0.2588733732700348,
      "learning_rate": 8e-05,
      "loss": 1.7595,
      "step": 877
    },
    {
      "epoch": 0.19311558341581436,
      "grad_norm": 0.2999958097934723,
      "learning_rate": 8e-05,
      "loss": 1.9095,
      "step": 878
    },
    {
      "epoch": 0.1933355328274497,
      "grad_norm": 0.27143120765686035,
      "learning_rate": 8e-05,
      "loss": 1.6698,
      "step": 879
    },
    {
      "epoch": 0.193555482239085,
      "grad_norm": 0.29155731201171875,
      "learning_rate": 8e-05,
      "loss": 1.6437,
      "step": 880
    },
    {
      "epoch": 0.19377543165072034,
      "grad_norm": 0.26307716965675354,
      "learning_rate": 8e-05,
      "loss": 1.6161,
      "step": 881
    },
    {
      "epoch": 0.19399538106235567,
      "grad_norm": 0.27041196823120117,
      "learning_rate": 8e-05,
      "loss": 1.5374,
      "step": 882
    },
    {
      "epoch": 0.19421533047399098,
      "grad_norm": 0.2752692699432373,
      "learning_rate": 8e-05,
      "loss": 1.6543,
      "step": 883
    },
    {
      "epoch": 0.1944352798856263,
      "grad_norm": 0.2883388102054596,
      "learning_rate": 8e-05,
      "loss": 1.7503,
      "step": 884
    },
    {
      "epoch": 0.19465522929726162,
      "grad_norm": 0.27332282066345215,
      "learning_rate": 8e-05,
      "loss": 1.8456,
      "step": 885
    },
    {
      "epoch": 0.19487517870889695,
      "grad_norm": 0.26226627826690674,
      "learning_rate": 8e-05,
      "loss": 1.2577,
      "step": 886
    },
    {
      "epoch": 0.19509512812053229,
      "grad_norm": 0.2709749639034271,
      "learning_rate": 8e-05,
      "loss": 1.7854,
      "step": 887
    },
    {
      "epoch": 0.1953150775321676,
      "grad_norm": 0.28380879759788513,
      "learning_rate": 8e-05,
      "loss": 1.7151,
      "step": 888
    },
    {
      "epoch": 0.19553502694380293,
      "grad_norm": 0.2702254354953766,
      "learning_rate": 8e-05,
      "loss": 1.6779,
      "step": 889
    },
    {
      "epoch": 0.19575497635543826,
      "grad_norm": 0.2620486617088318,
      "learning_rate": 8e-05,
      "loss": 1.7166,
      "step": 890
    },
    {
      "epoch": 0.19597492576707357,
      "grad_norm": 0.27195873856544495,
      "learning_rate": 8e-05,
      "loss": 1.7263,
      "step": 891
    },
    {
      "epoch": 0.1961948751787089,
      "grad_norm": 0.2719867527484894,
      "learning_rate": 8e-05,
      "loss": 1.6982,
      "step": 892
    },
    {
      "epoch": 0.19641482459034423,
      "grad_norm": 0.27889111638069153,
      "learning_rate": 8e-05,
      "loss": 1.7726,
      "step": 893
    },
    {
      "epoch": 0.19663477400197954,
      "grad_norm": 0.2745397686958313,
      "learning_rate": 8e-05,
      "loss": 1.7737,
      "step": 894
    },
    {
      "epoch": 0.19685472341361487,
      "grad_norm": 0.2698670029640198,
      "learning_rate": 8e-05,
      "loss": 1.7149,
      "step": 895
    },
    {
      "epoch": 0.19707467282525018,
      "grad_norm": 0.27113667130470276,
      "learning_rate": 8e-05,
      "loss": 1.7887,
      "step": 896
    },
    {
      "epoch": 0.19729462223688551,
      "grad_norm": 0.2772979140281677,
      "learning_rate": 8e-05,
      "loss": 1.8163,
      "step": 897
    },
    {
      "epoch": 0.19751457164852085,
      "grad_norm": 0.2757657766342163,
      "learning_rate": 8e-05,
      "loss": 1.636,
      "step": 898
    },
    {
      "epoch": 0.19773452106015615,
      "grad_norm": 0.26945242285728455,
      "learning_rate": 8e-05,
      "loss": 1.7639,
      "step": 899
    },
    {
      "epoch": 0.1979544704717915,
      "grad_norm": 0.27328991889953613,
      "learning_rate": 8e-05,
      "loss": 1.8421,
      "step": 900
    },
    {
      "epoch": 0.19817441988342682,
      "grad_norm": 0.2721468210220337,
      "learning_rate": 8e-05,
      "loss": 1.6926,
      "step": 901
    },
    {
      "epoch": 0.19839436929506213,
      "grad_norm": 0.2633766233921051,
      "learning_rate": 8e-05,
      "loss": 1.7629,
      "step": 902
    },
    {
      "epoch": 0.19861431870669746,
      "grad_norm": 0.26183879375457764,
      "learning_rate": 8e-05,
      "loss": 1.7149,
      "step": 903
    },
    {
      "epoch": 0.1988342681183328,
      "grad_norm": 0.2837960422039032,
      "learning_rate": 8e-05,
      "loss": 1.7923,
      "step": 904
    },
    {
      "epoch": 0.1990542175299681,
      "grad_norm": 0.30745571851730347,
      "learning_rate": 8e-05,
      "loss": 1.8315,
      "step": 905
    },
    {
      "epoch": 0.19927416694160344,
      "grad_norm": 0.2734341323375702,
      "learning_rate": 8e-05,
      "loss": 1.7424,
      "step": 906
    },
    {
      "epoch": 0.19949411635323874,
      "grad_norm": 0.2613460123538971,
      "learning_rate": 8e-05,
      "loss": 1.6445,
      "step": 907
    },
    {
      "epoch": 0.19971406576487408,
      "grad_norm": 0.27867522835731506,
      "learning_rate": 8e-05,
      "loss": 1.7075,
      "step": 908
    },
    {
      "epoch": 0.1999340151765094,
      "grad_norm": 0.269789457321167,
      "learning_rate": 8e-05,
      "loss": 1.6801,
      "step": 909
    },
    {
      "epoch": 0.20015396458814472,
      "grad_norm": 0.2684427797794342,
      "learning_rate": 8e-05,
      "loss": 1.6862,
      "step": 910
    },
    {
      "epoch": 0.20037391399978005,
      "grad_norm": 0.2929883897304535,
      "learning_rate": 8e-05,
      "loss": 1.7972,
      "step": 911
    },
    {
      "epoch": 0.20059386341141539,
      "grad_norm": 0.2757764756679535,
      "learning_rate": 8e-05,
      "loss": 1.6198,
      "step": 912
    },
    {
      "epoch": 0.2008138128230507,
      "grad_norm": 0.28071129322052,
      "learning_rate": 8e-05,
      "loss": 1.8268,
      "step": 913
    },
    {
      "epoch": 0.20103376223468603,
      "grad_norm": 0.2964448928833008,
      "learning_rate": 8e-05,
      "loss": 1.7806,
      "step": 914
    },
    {
      "epoch": 0.20125371164632136,
      "grad_norm": 0.2682490050792694,
      "learning_rate": 8e-05,
      "loss": 1.7016,
      "step": 915
    },
    {
      "epoch": 0.20147366105795667,
      "grad_norm": 0.2838338613510132,
      "learning_rate": 8e-05,
      "loss": 1.7402,
      "step": 916
    },
    {
      "epoch": 0.201693610469592,
      "grad_norm": 0.27621790766716003,
      "learning_rate": 8e-05,
      "loss": 1.7442,
      "step": 917
    },
    {
      "epoch": 0.2019135598812273,
      "grad_norm": 0.29265734553337097,
      "learning_rate": 8e-05,
      "loss": 1.6924,
      "step": 918
    },
    {
      "epoch": 0.20213350929286264,
      "grad_norm": 0.27404630184173584,
      "learning_rate": 8e-05,
      "loss": 1.7312,
      "step": 919
    },
    {
      "epoch": 0.20235345870449797,
      "grad_norm": 0.2742730975151062,
      "learning_rate": 8e-05,
      "loss": 1.8645,
      "step": 920
    },
    {
      "epoch": 0.20257340811613328,
      "grad_norm": 0.28536343574523926,
      "learning_rate": 8e-05,
      "loss": 1.717,
      "step": 921
    },
    {
      "epoch": 0.20279335752776861,
      "grad_norm": 0.28739288449287415,
      "learning_rate": 8e-05,
      "loss": 1.8356,
      "step": 922
    },
    {
      "epoch": 0.20301330693940395,
      "grad_norm": 0.2650564908981323,
      "learning_rate": 8e-05,
      "loss": 1.717,
      "step": 923
    },
    {
      "epoch": 0.20323325635103925,
      "grad_norm": 0.28638410568237305,
      "learning_rate": 8e-05,
      "loss": 1.8822,
      "step": 924
    },
    {
      "epoch": 0.2034532057626746,
      "grad_norm": 0.25474488735198975,
      "learning_rate": 8e-05,
      "loss": 1.5533,
      "step": 925
    },
    {
      "epoch": 0.20367315517430992,
      "grad_norm": 0.2719588279724121,
      "learning_rate": 8e-05,
      "loss": 1.7887,
      "step": 926
    },
    {
      "epoch": 0.20389310458594523,
      "grad_norm": 0.2572193741798401,
      "learning_rate": 8e-05,
      "loss": 1.5735,
      "step": 927
    },
    {
      "epoch": 0.20411305399758056,
      "grad_norm": 0.2975933253765106,
      "learning_rate": 8e-05,
      "loss": 1.7875,
      "step": 928
    },
    {
      "epoch": 0.20433300340921587,
      "grad_norm": 0.2562117874622345,
      "learning_rate": 8e-05,
      "loss": 1.5977,
      "step": 929
    },
    {
      "epoch": 0.2045529528208512,
      "grad_norm": 0.2524821162223816,
      "learning_rate": 8e-05,
      "loss": 1.6356,
      "step": 930
    },
    {
      "epoch": 0.20477290223248654,
      "grad_norm": 0.2621130347251892,
      "learning_rate": 8e-05,
      "loss": 1.6082,
      "step": 931
    },
    {
      "epoch": 0.20499285164412184,
      "grad_norm": 0.27930355072021484,
      "learning_rate": 8e-05,
      "loss": 1.7618,
      "step": 932
    },
    {
      "epoch": 0.20521280105575718,
      "grad_norm": 0.29147934913635254,
      "learning_rate": 8e-05,
      "loss": 1.8223,
      "step": 933
    },
    {
      "epoch": 0.2054327504673925,
      "grad_norm": 0.2584928870201111,
      "learning_rate": 8e-05,
      "loss": 1.6916,
      "step": 934
    },
    {
      "epoch": 0.20565269987902782,
      "grad_norm": 0.27299705147743225,
      "learning_rate": 8e-05,
      "loss": 1.5535,
      "step": 935
    },
    {
      "epoch": 0.20587264929066315,
      "grad_norm": 0.2682443857192993,
      "learning_rate": 8e-05,
      "loss": 1.7119,
      "step": 936
    },
    {
      "epoch": 0.20609259870229849,
      "grad_norm": 0.29716598987579346,
      "learning_rate": 8e-05,
      "loss": 2.0561,
      "step": 937
    },
    {
      "epoch": 0.2063125481139338,
      "grad_norm": 0.27801281213760376,
      "learning_rate": 8e-05,
      "loss": 1.7605,
      "step": 938
    },
    {
      "epoch": 0.20653249752556913,
      "grad_norm": 0.26767662167549133,
      "learning_rate": 8e-05,
      "loss": 1.6462,
      "step": 939
    },
    {
      "epoch": 0.20675244693720443,
      "grad_norm": 0.27354639768600464,
      "learning_rate": 8e-05,
      "loss": 1.7241,
      "step": 940
    },
    {
      "epoch": 0.20697239634883977,
      "grad_norm": 0.2684631049633026,
      "learning_rate": 8e-05,
      "loss": 1.8066,
      "step": 941
    },
    {
      "epoch": 0.2071923457604751,
      "grad_norm": 0.27846816182136536,
      "learning_rate": 8e-05,
      "loss": 1.7553,
      "step": 942
    },
    {
      "epoch": 0.2074122951721104,
      "grad_norm": 0.2820284366607666,
      "learning_rate": 8e-05,
      "loss": 1.8302,
      "step": 943
    },
    {
      "epoch": 0.20763224458374574,
      "grad_norm": 0.28080835938453674,
      "learning_rate": 8e-05,
      "loss": 1.7544,
      "step": 944
    },
    {
      "epoch": 0.20785219399538107,
      "grad_norm": 0.28095102310180664,
      "learning_rate": 8e-05,
      "loss": 1.7271,
      "step": 945
    },
    {
      "epoch": 0.20807214340701638,
      "grad_norm": 0.27856144309043884,
      "learning_rate": 8e-05,
      "loss": 1.8441,
      "step": 946
    },
    {
      "epoch": 0.20829209281865171,
      "grad_norm": 0.27816230058670044,
      "learning_rate": 8e-05,
      "loss": 1.981,
      "step": 947
    },
    {
      "epoch": 0.20851204223028702,
      "grad_norm": 0.2954215705394745,
      "learning_rate": 8e-05,
      "loss": 1.8001,
      "step": 948
    },
    {
      "epoch": 0.20873199164192235,
      "grad_norm": 0.24413350224494934,
      "learning_rate": 8e-05,
      "loss": 1.5436,
      "step": 949
    },
    {
      "epoch": 0.2089519410535577,
      "grad_norm": 0.2849874198436737,
      "learning_rate": 8e-05,
      "loss": 1.7027,
      "step": 950
    },
    {
      "epoch": 0.209171890465193,
      "grad_norm": 0.2710252106189728,
      "learning_rate": 8e-05,
      "loss": 1.7222,
      "step": 951
    },
    {
      "epoch": 0.20939183987682833,
      "grad_norm": 0.2557348608970642,
      "learning_rate": 8e-05,
      "loss": 1.6469,
      "step": 952
    },
    {
      "epoch": 0.20961178928846366,
      "grad_norm": 0.2688618004322052,
      "learning_rate": 8e-05,
      "loss": 1.6471,
      "step": 953
    },
    {
      "epoch": 0.20983173870009897,
      "grad_norm": 0.28641626238822937,
      "learning_rate": 8e-05,
      "loss": 1.8796,
      "step": 954
    },
    {
      "epoch": 0.2100516881117343,
      "grad_norm": 0.2582222521305084,
      "learning_rate": 8e-05,
      "loss": 1.5961,
      "step": 955
    },
    {
      "epoch": 0.21027163752336964,
      "grad_norm": 0.2615504562854767,
      "learning_rate": 8e-05,
      "loss": 1.6668,
      "step": 956
    },
    {
      "epoch": 0.21049158693500494,
      "grad_norm": 0.2669670879840851,
      "learning_rate": 8e-05,
      "loss": 1.663,
      "step": 957
    },
    {
      "epoch": 0.21071153634664028,
      "grad_norm": 0.2649092972278595,
      "learning_rate": 8e-05,
      "loss": 1.5377,
      "step": 958
    },
    {
      "epoch": 0.21093148575827558,
      "grad_norm": 0.2936461865901947,
      "learning_rate": 8e-05,
      "loss": 1.5659,
      "step": 959
    },
    {
      "epoch": 0.21115143516991092,
      "grad_norm": 0.2878846824169159,
      "learning_rate": 8e-05,
      "loss": 1.8567,
      "step": 960
    },
    {
      "epoch": 0.21137138458154625,
      "grad_norm": 0.2928799092769623,
      "learning_rate": 8e-05,
      "loss": 1.8423,
      "step": 961
    },
    {
      "epoch": 0.21159133399318156,
      "grad_norm": 0.2641200125217438,
      "learning_rate": 8e-05,
      "loss": 1.6403,
      "step": 962
    },
    {
      "epoch": 0.2118112834048169,
      "grad_norm": 0.26553985476493835,
      "learning_rate": 8e-05,
      "loss": 1.7436,
      "step": 963
    },
    {
      "epoch": 0.21203123281645223,
      "grad_norm": 0.25616276264190674,
      "learning_rate": 8e-05,
      "loss": 1.5959,
      "step": 964
    },
    {
      "epoch": 0.21225118222808753,
      "grad_norm": 0.29729175567626953,
      "learning_rate": 8e-05,
      "loss": 1.7164,
      "step": 965
    },
    {
      "epoch": 0.21247113163972287,
      "grad_norm": 0.2739759683609009,
      "learning_rate": 8e-05,
      "loss": 1.7797,
      "step": 966
    },
    {
      "epoch": 0.2126910810513582,
      "grad_norm": 0.2686353921890259,
      "learning_rate": 8e-05,
      "loss": 1.6974,
      "step": 967
    },
    {
      "epoch": 0.2129110304629935,
      "grad_norm": 0.261820912361145,
      "learning_rate": 8e-05,
      "loss": 1.6864,
      "step": 968
    },
    {
      "epoch": 0.21313097987462884,
      "grad_norm": 0.26877105236053467,
      "learning_rate": 8e-05,
      "loss": 1.6164,
      "step": 969
    },
    {
      "epoch": 0.21335092928626415,
      "grad_norm": 0.2555043399333954,
      "learning_rate": 8e-05,
      "loss": 1.6898,
      "step": 970
    },
    {
      "epoch": 0.21357087869789948,
      "grad_norm": 0.28584909439086914,
      "learning_rate": 8e-05,
      "loss": 1.9125,
      "step": 971
    },
    {
      "epoch": 0.21379082810953481,
      "grad_norm": 0.2830945551395416,
      "learning_rate": 8e-05,
      "loss": 1.6416,
      "step": 972
    },
    {
      "epoch": 0.21401077752117012,
      "grad_norm": 0.27979904413223267,
      "learning_rate": 8e-05,
      "loss": 1.8355,
      "step": 973
    },
    {
      "epoch": 0.21423072693280545,
      "grad_norm": 0.2672286033630371,
      "learning_rate": 8e-05,
      "loss": 1.685,
      "step": 974
    },
    {
      "epoch": 0.2144506763444408,
      "grad_norm": 0.26699069142341614,
      "learning_rate": 8e-05,
      "loss": 1.5951,
      "step": 975
    },
    {
      "epoch": 0.2146706257560761,
      "grad_norm": 0.2720418870449066,
      "learning_rate": 8e-05,
      "loss": 1.7558,
      "step": 976
    },
    {
      "epoch": 0.21489057516771143,
      "grad_norm": 0.26792463660240173,
      "learning_rate": 8e-05,
      "loss": 1.7407,
      "step": 977
    },
    {
      "epoch": 0.21511052457934676,
      "grad_norm": 0.2763652503490448,
      "learning_rate": 8e-05,
      "loss": 1.7525,
      "step": 978
    },
    {
      "epoch": 0.21533047399098207,
      "grad_norm": 0.2952554225921631,
      "learning_rate": 8e-05,
      "loss": 1.6535,
      "step": 979
    },
    {
      "epoch": 0.2155504234026174,
      "grad_norm": 0.24981874227523804,
      "learning_rate": 8e-05,
      "loss": 1.6055,
      "step": 980
    },
    {
      "epoch": 0.2157703728142527,
      "grad_norm": 0.29071807861328125,
      "learning_rate": 8e-05,
      "loss": 1.7461,
      "step": 981
    },
    {
      "epoch": 0.21599032222588804,
      "grad_norm": 0.26875782012939453,
      "learning_rate": 8e-05,
      "loss": 1.5809,
      "step": 982
    },
    {
      "epoch": 0.21621027163752338,
      "grad_norm": 0.2519072890281677,
      "learning_rate": 8e-05,
      "loss": 1.7001,
      "step": 983
    },
    {
      "epoch": 0.21643022104915868,
      "grad_norm": 0.2748781144618988,
      "learning_rate": 8e-05,
      "loss": 1.8367,
      "step": 984
    },
    {
      "epoch": 0.21665017046079402,
      "grad_norm": 0.274047315120697,
      "learning_rate": 8e-05,
      "loss": 1.7698,
      "step": 985
    },
    {
      "epoch": 0.21687011987242935,
      "grad_norm": 0.2614712119102478,
      "learning_rate": 8e-05,
      "loss": 1.5411,
      "step": 986
    },
    {
      "epoch": 0.21709006928406466,
      "grad_norm": 0.2714536190032959,
      "learning_rate": 8e-05,
      "loss": 1.6058,
      "step": 987
    },
    {
      "epoch": 0.2173100186957,
      "grad_norm": 0.28763729333877563,
      "learning_rate": 8e-05,
      "loss": 1.6711,
      "step": 988
    },
    {
      "epoch": 0.21752996810733533,
      "grad_norm": 0.26780402660369873,
      "learning_rate": 8e-05,
      "loss": 1.549,
      "step": 989
    },
    {
      "epoch": 0.21774991751897063,
      "grad_norm": 0.28782159090042114,
      "learning_rate": 8e-05,
      "loss": 1.8305,
      "step": 990
    },
    {
      "epoch": 0.21796986693060597,
      "grad_norm": 0.2859013080596924,
      "learning_rate": 8e-05,
      "loss": 1.7794,
      "step": 991
    },
    {
      "epoch": 0.21818981634224127,
      "grad_norm": 0.2893369197845459,
      "learning_rate": 8e-05,
      "loss": 1.6284,
      "step": 992
    },
    {
      "epoch": 0.2184097657538766,
      "grad_norm": 0.2809627652168274,
      "learning_rate": 8e-05,
      "loss": 1.6401,
      "step": 993
    },
    {
      "epoch": 0.21862971516551194,
      "grad_norm": 0.2700895667076111,
      "learning_rate": 8e-05,
      "loss": 1.7153,
      "step": 994
    },
    {
      "epoch": 0.21884966457714725,
      "grad_norm": 0.26506903767585754,
      "learning_rate": 8e-05,
      "loss": 1.4688,
      "step": 995
    },
    {
      "epoch": 0.21906961398878258,
      "grad_norm": 0.28202024102211,
      "learning_rate": 8e-05,
      "loss": 1.7009,
      "step": 996
    },
    {
      "epoch": 0.21928956340041791,
      "grad_norm": 0.2625409960746765,
      "learning_rate": 8e-05,
      "loss": 1.6491,
      "step": 997
    },
    {
      "epoch": 0.21950951281205322,
      "grad_norm": 0.29967787861824036,
      "learning_rate": 8e-05,
      "loss": 1.7231,
      "step": 998
    },
    {
      "epoch": 0.21972946222368855,
      "grad_norm": 0.2992357909679413,
      "learning_rate": 8e-05,
      "loss": 1.7028,
      "step": 999
    },
    {
      "epoch": 0.2199494116353239,
      "grad_norm": 0.28712475299835205,
      "learning_rate": 8e-05,
      "loss": 1.763,
      "step": 1000
    },
    {
      "epoch": 0.2201693610469592,
      "grad_norm": 0.26186901330947876,
      "learning_rate": 8e-05,
      "loss": 1.5695,
      "step": 1001
    },
    {
      "epoch": 0.22038931045859453,
      "grad_norm": 0.2897952198982239,
      "learning_rate": 8e-05,
      "loss": 1.6303,
      "step": 1002
    },
    {
      "epoch": 0.22060925987022983,
      "grad_norm": 0.2761494815349579,
      "learning_rate": 8e-05,
      "loss": 1.7448,
      "step": 1003
    },
    {
      "epoch": 0.22082920928186517,
      "grad_norm": 0.2604154944419861,
      "learning_rate": 8e-05,
      "loss": 1.53,
      "step": 1004
    },
    {
      "epoch": 0.2210491586935005,
      "grad_norm": 0.2897418737411499,
      "learning_rate": 8e-05,
      "loss": 1.7639,
      "step": 1005
    },
    {
      "epoch": 0.2212691081051358,
      "grad_norm": 0.28289687633514404,
      "learning_rate": 8e-05,
      "loss": 1.7202,
      "step": 1006
    },
    {
      "epoch": 0.22148905751677114,
      "grad_norm": 0.26917099952697754,
      "learning_rate": 8e-05,
      "loss": 1.7183,
      "step": 1007
    },
    {
      "epoch": 0.22170900692840648,
      "grad_norm": 0.26708024740219116,
      "learning_rate": 8e-05,
      "loss": 1.636,
      "step": 1008
    },
    {
      "epoch": 0.22192895634004178,
      "grad_norm": 0.2759459316730499,
      "learning_rate": 8e-05,
      "loss": 1.6537,
      "step": 1009
    },
    {
      "epoch": 0.22214890575167712,
      "grad_norm": 0.3040393590927124,
      "learning_rate": 8e-05,
      "loss": 1.7849,
      "step": 1010
    },
    {
      "epoch": 0.22236885516331245,
      "grad_norm": 0.2729750871658325,
      "learning_rate": 8e-05,
      "loss": 1.8199,
      "step": 1011
    },
    {
      "epoch": 0.22258880457494776,
      "grad_norm": 0.28002965450286865,
      "learning_rate": 8e-05,
      "loss": 1.8286,
      "step": 1012
    },
    {
      "epoch": 0.2228087539865831,
      "grad_norm": 0.27389100193977356,
      "learning_rate": 8e-05,
      "loss": 1.6472,
      "step": 1013
    },
    {
      "epoch": 0.2230287033982184,
      "grad_norm": 0.2610195279121399,
      "learning_rate": 8e-05,
      "loss": 1.6096,
      "step": 1014
    },
    {
      "epoch": 0.22324865280985373,
      "grad_norm": 0.2683162987232208,
      "learning_rate": 8e-05,
      "loss": 1.6477,
      "step": 1015
    },
    {
      "epoch": 0.22346860222148907,
      "grad_norm": 0.26524773240089417,
      "learning_rate": 8e-05,
      "loss": 1.6224,
      "step": 1016
    },
    {
      "epoch": 0.22368855163312437,
      "grad_norm": 0.26295366883277893,
      "learning_rate": 8e-05,
      "loss": 1.7316,
      "step": 1017
    },
    {
      "epoch": 0.2239085010447597,
      "grad_norm": 0.2837565243244171,
      "learning_rate": 8e-05,
      "loss": 1.9452,
      "step": 1018
    },
    {
      "epoch": 0.22412845045639504,
      "grad_norm": 0.28365132212638855,
      "learning_rate": 8e-05,
      "loss": 1.6577,
      "step": 1019
    },
    {
      "epoch": 0.22434839986803035,
      "grad_norm": 0.2736522853374481,
      "learning_rate": 8e-05,
      "loss": 1.6644,
      "step": 1020
    },
    {
      "epoch": 0.22456834927966568,
      "grad_norm": 0.2878374755382538,
      "learning_rate": 8e-05,
      "loss": 1.5844,
      "step": 1021
    },
    {
      "epoch": 0.22478829869130101,
      "grad_norm": 0.28223422169685364,
      "learning_rate": 8e-05,
      "loss": 1.881,
      "step": 1022
    },
    {
      "epoch": 0.22500824810293632,
      "grad_norm": 0.26408734917640686,
      "learning_rate": 8e-05,
      "loss": 1.6201,
      "step": 1023
    },
    {
      "epoch": 0.22522819751457165,
      "grad_norm": 0.28506824374198914,
      "learning_rate": 8e-05,
      "loss": 1.8146,
      "step": 1024
    },
    {
      "epoch": 0.22544814692620696,
      "grad_norm": 0.2808188796043396,
      "learning_rate": 8e-05,
      "loss": 1.8394,
      "step": 1025
    },
    {
      "epoch": 0.2256680963378423,
      "grad_norm": 0.2950645387172699,
      "learning_rate": 8e-05,
      "loss": 1.7993,
      "step": 1026
    },
    {
      "epoch": 0.22588804574947763,
      "grad_norm": 0.27935850620269775,
      "learning_rate": 8e-05,
      "loss": 1.6506,
      "step": 1027
    },
    {
      "epoch": 0.22610799516111293,
      "grad_norm": 0.2576957643032074,
      "learning_rate": 8e-05,
      "loss": 1.6987,
      "step": 1028
    },
    {
      "epoch": 0.22632794457274827,
      "grad_norm": 0.2719384729862213,
      "learning_rate": 8e-05,
      "loss": 1.6407,
      "step": 1029
    },
    {
      "epoch": 0.2265478939843836,
      "grad_norm": 0.25457167625427246,
      "learning_rate": 8e-05,
      "loss": 1.6877,
      "step": 1030
    },
    {
      "epoch": 0.2267678433960189,
      "grad_norm": 0.2758035659790039,
      "learning_rate": 8e-05,
      "loss": 1.6739,
      "step": 1031
    },
    {
      "epoch": 0.22698779280765424,
      "grad_norm": 0.27135321497917175,
      "learning_rate": 8e-05,
      "loss": 1.7124,
      "step": 1032
    },
    {
      "epoch": 0.22720774221928958,
      "grad_norm": 0.2675740420818329,
      "learning_rate": 8e-05,
      "loss": 1.7857,
      "step": 1033
    },
    {
      "epoch": 0.22742769163092488,
      "grad_norm": 0.28627943992614746,
      "learning_rate": 8e-05,
      "loss": 1.7012,
      "step": 1034
    },
    {
      "epoch": 0.22764764104256022,
      "grad_norm": 0.2710109353065491,
      "learning_rate": 8e-05,
      "loss": 1.6463,
      "step": 1035
    },
    {
      "epoch": 0.22786759045419552,
      "grad_norm": 0.27190473675727844,
      "learning_rate": 8e-05,
      "loss": 1.7288,
      "step": 1036
    },
    {
      "epoch": 0.22808753986583086,
      "grad_norm": 0.2503564953804016,
      "learning_rate": 8e-05,
      "loss": 1.566,
      "step": 1037
    },
    {
      "epoch": 0.2283074892774662,
      "grad_norm": 0.26503992080688477,
      "learning_rate": 8e-05,
      "loss": 1.7034,
      "step": 1038
    },
    {
      "epoch": 0.2285274386891015,
      "grad_norm": 0.29445260763168335,
      "learning_rate": 8e-05,
      "loss": 1.6739,
      "step": 1039
    },
    {
      "epoch": 0.22874738810073683,
      "grad_norm": 0.25705471634864807,
      "learning_rate": 8e-05,
      "loss": 1.6503,
      "step": 1040
    },
    {
      "epoch": 0.22896733751237217,
      "grad_norm": 0.27109014987945557,
      "learning_rate": 8e-05,
      "loss": 1.8045,
      "step": 1041
    },
    {
      "epoch": 0.22918728692400747,
      "grad_norm": 0.2972055673599243,
      "learning_rate": 8e-05,
      "loss": 1.6439,
      "step": 1042
    },
    {
      "epoch": 0.2294072363356428,
      "grad_norm": 0.27126485109329224,
      "learning_rate": 8e-05,
      "loss": 1.672,
      "step": 1043
    },
    {
      "epoch": 0.22962718574727814,
      "grad_norm": 0.2731145918369293,
      "learning_rate": 8e-05,
      "loss": 1.7795,
      "step": 1044
    },
    {
      "epoch": 0.22984713515891345,
      "grad_norm": 0.2768365442752838,
      "learning_rate": 8e-05,
      "loss": 1.7145,
      "step": 1045
    },
    {
      "epoch": 0.23006708457054878,
      "grad_norm": 0.2606940269470215,
      "learning_rate": 8e-05,
      "loss": 1.6169,
      "step": 1046
    },
    {
      "epoch": 0.2302870339821841,
      "grad_norm": 0.2898729741573334,
      "learning_rate": 8e-05,
      "loss": 1.7315,
      "step": 1047
    },
    {
      "epoch": 0.23050698339381942,
      "grad_norm": 0.2772413194179535,
      "learning_rate": 8e-05,
      "loss": 1.8632,
      "step": 1048
    },
    {
      "epoch": 0.23072693280545475,
      "grad_norm": 0.25808605551719666,
      "learning_rate": 8e-05,
      "loss": 1.6626,
      "step": 1049
    },
    {
      "epoch": 0.23094688221709006,
      "grad_norm": 0.2727161645889282,
      "learning_rate": 8e-05,
      "loss": 1.7848,
      "step": 1050
    },
    {
      "epoch": 0.2311668316287254,
      "grad_norm": 0.25677087903022766,
      "learning_rate": 8e-05,
      "loss": 1.6168,
      "step": 1051
    },
    {
      "epoch": 0.23138678104036073,
      "grad_norm": 0.2761050760746002,
      "learning_rate": 8e-05,
      "loss": 1.8615,
      "step": 1052
    },
    {
      "epoch": 0.23160673045199603,
      "grad_norm": 0.2862778604030609,
      "learning_rate": 8e-05,
      "loss": 1.8728,
      "step": 1053
    },
    {
      "epoch": 0.23182667986363137,
      "grad_norm": 0.27526941895484924,
      "learning_rate": 8e-05,
      "loss": 1.7627,
      "step": 1054
    },
    {
      "epoch": 0.23204662927526667,
      "grad_norm": 0.2932235896587372,
      "learning_rate": 8e-05,
      "loss": 1.8539,
      "step": 1055
    },
    {
      "epoch": 0.232266578686902,
      "grad_norm": 0.2770839035511017,
      "learning_rate": 8e-05,
      "loss": 1.7393,
      "step": 1056
    },
    {
      "epoch": 0.23248652809853734,
      "grad_norm": 0.2741580307483673,
      "learning_rate": 8e-05,
      "loss": 1.6076,
      "step": 1057
    },
    {
      "epoch": 0.23270647751017265,
      "grad_norm": 0.2788783311843872,
      "learning_rate": 8e-05,
      "loss": 1.7615,
      "step": 1058
    },
    {
      "epoch": 0.23292642692180798,
      "grad_norm": 0.28565406799316406,
      "learning_rate": 8e-05,
      "loss": 1.6266,
      "step": 1059
    },
    {
      "epoch": 0.23314637633344332,
      "grad_norm": 0.26543545722961426,
      "learning_rate": 8e-05,
      "loss": 1.7192,
      "step": 1060
    },
    {
      "epoch": 0.23336632574507862,
      "grad_norm": 0.2770478129386902,
      "learning_rate": 8e-05,
      "loss": 1.8056,
      "step": 1061
    },
    {
      "epoch": 0.23358627515671396,
      "grad_norm": 0.27805015444755554,
      "learning_rate": 8e-05,
      "loss": 1.6735,
      "step": 1062
    },
    {
      "epoch": 0.2338062245683493,
      "grad_norm": 0.309862345457077,
      "learning_rate": 8e-05,
      "loss": 1.7235,
      "step": 1063
    },
    {
      "epoch": 0.2340261739799846,
      "grad_norm": 0.27140697836875916,
      "learning_rate": 8e-05,
      "loss": 1.6883,
      "step": 1064
    },
    {
      "epoch": 0.23424612339161993,
      "grad_norm": 0.3052090108394623,
      "learning_rate": 8e-05,
      "loss": 1.8792,
      "step": 1065
    },
    {
      "epoch": 0.23446607280325524,
      "grad_norm": 0.2995065450668335,
      "learning_rate": 8e-05,
      "loss": 1.6632,
      "step": 1066
    },
    {
      "epoch": 0.23468602221489057,
      "grad_norm": 0.2782532870769501,
      "learning_rate": 8e-05,
      "loss": 1.7395,
      "step": 1067
    },
    {
      "epoch": 0.2349059716265259,
      "grad_norm": 0.28436902165412903,
      "learning_rate": 8e-05,
      "loss": 1.8416,
      "step": 1068
    },
    {
      "epoch": 0.2351259210381612,
      "grad_norm": 0.2740377187728882,
      "learning_rate": 8e-05,
      "loss": 1.9026,
      "step": 1069
    },
    {
      "epoch": 0.23534587044979655,
      "grad_norm": 0.2978285849094391,
      "learning_rate": 8e-05,
      "loss": 1.7277,
      "step": 1070
    },
    {
      "epoch": 0.23556581986143188,
      "grad_norm": 0.27265986800193787,
      "learning_rate": 8e-05,
      "loss": 1.7376,
      "step": 1071
    },
    {
      "epoch": 0.2357857692730672,
      "grad_norm": 0.24915599822998047,
      "learning_rate": 8e-05,
      "loss": 1.6151,
      "step": 1072
    },
    {
      "epoch": 0.23600571868470252,
      "grad_norm": 0.28203171491622925,
      "learning_rate": 8e-05,
      "loss": 1.7713,
      "step": 1073
    },
    {
      "epoch": 0.23622566809633785,
      "grad_norm": 0.278793066740036,
      "learning_rate": 8e-05,
      "loss": 1.6717,
      "step": 1074
    },
    {
      "epoch": 0.23644561750797316,
      "grad_norm": 0.2760609984397888,
      "learning_rate": 8e-05,
      "loss": 1.5866,
      "step": 1075
    },
    {
      "epoch": 0.2366655669196085,
      "grad_norm": 0.2726036012172699,
      "learning_rate": 8e-05,
      "loss": 1.6774,
      "step": 1076
    },
    {
      "epoch": 0.2368855163312438,
      "grad_norm": 0.27443891763687134,
      "learning_rate": 8e-05,
      "loss": 1.7615,
      "step": 1077
    },
    {
      "epoch": 0.23710546574287913,
      "grad_norm": 0.2818880081176758,
      "learning_rate": 8e-05,
      "loss": 1.7433,
      "step": 1078
    },
    {
      "epoch": 0.23732541515451447,
      "grad_norm": 0.2646252512931824,
      "learning_rate": 8e-05,
      "loss": 1.5498,
      "step": 1079
    },
    {
      "epoch": 0.23754536456614977,
      "grad_norm": 0.2964784502983093,
      "learning_rate": 8e-05,
      "loss": 1.6162,
      "step": 1080
    },
    {
      "epoch": 0.2377653139777851,
      "grad_norm": 0.3044411242008209,
      "learning_rate": 8e-05,
      "loss": 1.7395,
      "step": 1081
    },
    {
      "epoch": 0.23798526338942044,
      "grad_norm": 0.28679221868515015,
      "learning_rate": 8e-05,
      "loss": 1.8126,
      "step": 1082
    },
    {
      "epoch": 0.23820521280105575,
      "grad_norm": 0.26326417922973633,
      "learning_rate": 8e-05,
      "loss": 1.6451,
      "step": 1083
    },
    {
      "epoch": 0.23842516221269108,
      "grad_norm": 0.28527480363845825,
      "learning_rate": 8e-05,
      "loss": 1.8442,
      "step": 1084
    },
    {
      "epoch": 0.23864511162432642,
      "grad_norm": 0.28897759318351746,
      "learning_rate": 8e-05,
      "loss": 1.8224,
      "step": 1085
    },
    {
      "epoch": 0.23886506103596172,
      "grad_norm": 0.2955721616744995,
      "learning_rate": 8e-05,
      "loss": 1.7304,
      "step": 1086
    },
    {
      "epoch": 0.23908501044759706,
      "grad_norm": 0.26267075538635254,
      "learning_rate": 8e-05,
      "loss": 1.67,
      "step": 1087
    },
    {
      "epoch": 0.23930495985923236,
      "grad_norm": 0.27105912566185,
      "learning_rate": 8e-05,
      "loss": 1.7461,
      "step": 1088
    },
    {
      "epoch": 0.2395249092708677,
      "grad_norm": 0.26483941078186035,
      "learning_rate": 8e-05,
      "loss": 1.6215,
      "step": 1089
    },
    {
      "epoch": 0.23974485868250303,
      "grad_norm": 0.2804373800754547,
      "learning_rate": 8e-05,
      "loss": 1.6618,
      "step": 1090
    },
    {
      "epoch": 0.23996480809413834,
      "grad_norm": 0.26146185398101807,
      "learning_rate": 8e-05,
      "loss": 1.6641,
      "step": 1091
    },
    {
      "epoch": 0.24018475750577367,
      "grad_norm": 0.2839837372303009,
      "learning_rate": 8e-05,
      "loss": 1.5898,
      "step": 1092
    },
    {
      "epoch": 0.240404706917409,
      "grad_norm": 0.26833322644233704,
      "learning_rate": 8e-05,
      "loss": 1.8341,
      "step": 1093
    },
    {
      "epoch": 0.2406246563290443,
      "grad_norm": 0.2779574394226074,
      "learning_rate": 8e-05,
      "loss": 1.7142,
      "step": 1094
    },
    {
      "epoch": 0.24084460574067965,
      "grad_norm": 0.2821759879589081,
      "learning_rate": 8e-05,
      "loss": 1.7261,
      "step": 1095
    },
    {
      "epoch": 0.24106455515231498,
      "grad_norm": 0.2849150002002716,
      "learning_rate": 8e-05,
      "loss": 1.6834,
      "step": 1096
    },
    {
      "epoch": 0.24128450456395029,
      "grad_norm": 0.277148574590683,
      "learning_rate": 8e-05,
      "loss": 1.5617,
      "step": 1097
    },
    {
      "epoch": 0.24150445397558562,
      "grad_norm": 0.28307756781578064,
      "learning_rate": 8e-05,
      "loss": 1.8104,
      "step": 1098
    },
    {
      "epoch": 0.24172440338722093,
      "grad_norm": 0.28540289402008057,
      "learning_rate": 8e-05,
      "loss": 1.7331,
      "step": 1099
    },
    {
      "epoch": 0.24194435279885626,
      "grad_norm": 0.277544766664505,
      "learning_rate": 8e-05,
      "loss": 1.762,
      "step": 1100
    },
    {
      "epoch": 0.2421643022104916,
      "grad_norm": 0.259435772895813,
      "learning_rate": 8e-05,
      "loss": 1.5474,
      "step": 1101
    },
    {
      "epoch": 0.2423842516221269,
      "grad_norm": 0.2759372591972351,
      "learning_rate": 8e-05,
      "loss": 1.6535,
      "step": 1102
    },
    {
      "epoch": 0.24260420103376223,
      "grad_norm": 0.27163347601890564,
      "learning_rate": 8e-05,
      "loss": 1.6035,
      "step": 1103
    },
    {
      "epoch": 0.24282415044539757,
      "grad_norm": 0.26722922921180725,
      "learning_rate": 8e-05,
      "loss": 1.7607,
      "step": 1104
    },
    {
      "epoch": 0.24304409985703287,
      "grad_norm": 0.2925039529800415,
      "learning_rate": 8e-05,
      "loss": 1.6441,
      "step": 1105
    },
    {
      "epoch": 0.2432640492686682,
      "grad_norm": 0.271672785282135,
      "learning_rate": 8e-05,
      "loss": 1.658,
      "step": 1106
    },
    {
      "epoch": 0.24348399868030354,
      "grad_norm": 0.2827896773815155,
      "learning_rate": 8e-05,
      "loss": 1.6258,
      "step": 1107
    },
    {
      "epoch": 0.24370394809193885,
      "grad_norm": 0.2732497751712799,
      "learning_rate": 8e-05,
      "loss": 1.6379,
      "step": 1108
    },
    {
      "epoch": 0.24392389750357418,
      "grad_norm": 0.28081193566322327,
      "learning_rate": 8e-05,
      "loss": 1.7901,
      "step": 1109
    },
    {
      "epoch": 0.2441438469152095,
      "grad_norm": 0.2799675762653351,
      "learning_rate": 8e-05,
      "loss": 1.8323,
      "step": 1110
    },
    {
      "epoch": 0.24436379632684482,
      "grad_norm": 0.2677648961544037,
      "learning_rate": 8e-05,
      "loss": 1.7372,
      "step": 1111
    },
    {
      "epoch": 0.24458374573848016,
      "grad_norm": 0.2644648551940918,
      "learning_rate": 8e-05,
      "loss": 1.6594,
      "step": 1112
    },
    {
      "epoch": 0.24480369515011546,
      "grad_norm": 0.2704750895500183,
      "learning_rate": 8e-05,
      "loss": 1.706,
      "step": 1113
    },
    {
      "epoch": 0.2450236445617508,
      "grad_norm": 0.2762587368488312,
      "learning_rate": 8e-05,
      "loss": 1.7445,
      "step": 1114
    },
    {
      "epoch": 0.24524359397338613,
      "grad_norm": 0.2578018307685852,
      "learning_rate": 8e-05,
      "loss": 1.5707,
      "step": 1115
    },
    {
      "epoch": 0.24546354338502144,
      "grad_norm": 0.2892129719257355,
      "learning_rate": 8e-05,
      "loss": 1.805,
      "step": 1116
    },
    {
      "epoch": 0.24568349279665677,
      "grad_norm": 0.2868081033229828,
      "learning_rate": 8e-05,
      "loss": 1.7756,
      "step": 1117
    },
    {
      "epoch": 0.2459034422082921,
      "grad_norm": 0.2820534110069275,
      "learning_rate": 8e-05,
      "loss": 1.7826,
      "step": 1118
    },
    {
      "epoch": 0.2461233916199274,
      "grad_norm": 0.2824958264827728,
      "learning_rate": 8e-05,
      "loss": 1.6752,
      "step": 1119
    },
    {
      "epoch": 0.24634334103156275,
      "grad_norm": 0.2782610356807709,
      "learning_rate": 8e-05,
      "loss": 1.7536,
      "step": 1120
    },
    {
      "epoch": 0.24656329044319805,
      "grad_norm": 0.27147912979125977,
      "learning_rate": 8e-05,
      "loss": 1.6783,
      "step": 1121
    },
    {
      "epoch": 0.24678323985483339,
      "grad_norm": 0.2740795612335205,
      "learning_rate": 8e-05,
      "loss": 1.7702,
      "step": 1122
    },
    {
      "epoch": 0.24700318926646872,
      "grad_norm": 0.2922619879245758,
      "learning_rate": 8e-05,
      "loss": 1.8204,
      "step": 1123
    },
    {
      "epoch": 0.24722313867810403,
      "grad_norm": 0.2872619926929474,
      "learning_rate": 8e-05,
      "loss": 1.714,
      "step": 1124
    },
    {
      "epoch": 0.24744308808973936,
      "grad_norm": 0.27333369851112366,
      "learning_rate": 8e-05,
      "loss": 1.6575,
      "step": 1125
    },
    {
      "epoch": 0.2476630375013747,
      "grad_norm": 0.28192320466041565,
      "learning_rate": 8e-05,
      "loss": 1.7221,
      "step": 1126
    },
    {
      "epoch": 0.24788298691301,
      "grad_norm": 0.26607248187065125,
      "learning_rate": 8e-05,
      "loss": 1.7262,
      "step": 1127
    },
    {
      "epoch": 0.24810293632464533,
      "grad_norm": 0.279690682888031,
      "learning_rate": 8e-05,
      "loss": 1.7004,
      "step": 1128
    },
    {
      "epoch": 0.24832288573628067,
      "grad_norm": 0.27289190888404846,
      "learning_rate": 8e-05,
      "loss": 1.6916,
      "step": 1129
    },
    {
      "epoch": 0.24854283514791597,
      "grad_norm": 0.27388349175453186,
      "learning_rate": 8e-05,
      "loss": 1.6656,
      "step": 1130
    },
    {
      "epoch": 0.2487627845595513,
      "grad_norm": 0.2912501096725464,
      "learning_rate": 8e-05,
      "loss": 1.8086,
      "step": 1131
    },
    {
      "epoch": 0.24898273397118661,
      "grad_norm": 0.2999799847602844,
      "learning_rate": 8e-05,
      "loss": 1.7659,
      "step": 1132
    },
    {
      "epoch": 0.24920268338282195,
      "grad_norm": 0.262207955121994,
      "learning_rate": 8e-05,
      "loss": 1.6581,
      "step": 1133
    },
    {
      "epoch": 0.24942263279445728,
      "grad_norm": 0.2571624517440796,
      "learning_rate": 8e-05,
      "loss": 1.6509,
      "step": 1134
    },
    {
      "epoch": 0.2496425822060926,
      "grad_norm": 0.26213690638542175,
      "learning_rate": 8e-05,
      "loss": 1.6044,
      "step": 1135
    },
    {
      "epoch": 0.24986253161772792,
      "grad_norm": 0.2870398461818695,
      "learning_rate": 8e-05,
      "loss": 1.6678,
      "step": 1136
    },
    {
      "epoch": 0.25008248102936326,
      "grad_norm": 0.2672583758831024,
      "learning_rate": 8e-05,
      "loss": 1.6563,
      "step": 1137
    },
    {
      "epoch": 0.2503024304409986,
      "grad_norm": 0.29864680767059326,
      "learning_rate": 8e-05,
      "loss": 1.858,
      "step": 1138
    },
    {
      "epoch": 0.25052237985263387,
      "grad_norm": 0.3096907436847687,
      "learning_rate": 8e-05,
      "loss": 1.7731,
      "step": 1139
    },
    {
      "epoch": 0.2507423292642692,
      "grad_norm": 0.2668014466762543,
      "learning_rate": 8e-05,
      "loss": 1.6173,
      "step": 1140
    },
    {
      "epoch": 0.25096227867590454,
      "grad_norm": 0.275074303150177,
      "learning_rate": 8e-05,
      "loss": 1.704,
      "step": 1141
    },
    {
      "epoch": 0.25118222808753987,
      "grad_norm": 0.29657119512557983,
      "learning_rate": 8e-05,
      "loss": 1.9789,
      "step": 1142
    },
    {
      "epoch": 0.2514021774991752,
      "grad_norm": 0.26117807626724243,
      "learning_rate": 8e-05,
      "loss": 1.6815,
      "step": 1143
    },
    {
      "epoch": 0.25162212691081054,
      "grad_norm": 0.2738019824028015,
      "learning_rate": 8e-05,
      "loss": 1.7031,
      "step": 1144
    },
    {
      "epoch": 0.2518420763224458,
      "grad_norm": 0.27922967076301575,
      "learning_rate": 8e-05,
      "loss": 1.7914,
      "step": 1145
    },
    {
      "epoch": 0.25206202573408115,
      "grad_norm": 0.2876172661781311,
      "learning_rate": 8e-05,
      "loss": 1.721,
      "step": 1146
    },
    {
      "epoch": 0.2522819751457165,
      "grad_norm": 0.28017961978912354,
      "learning_rate": 8e-05,
      "loss": 1.6731,
      "step": 1147
    },
    {
      "epoch": 0.2525019245573518,
      "grad_norm": 0.2898389399051666,
      "learning_rate": 8e-05,
      "loss": 1.8749,
      "step": 1148
    },
    {
      "epoch": 0.25272187396898715,
      "grad_norm": 0.2742408812046051,
      "learning_rate": 8e-05,
      "loss": 1.6811,
      "step": 1149
    },
    {
      "epoch": 0.25294182338062243,
      "grad_norm": 0.2806207835674286,
      "learning_rate": 8e-05,
      "loss": 1.7082,
      "step": 1150
    },
    {
      "epoch": 0.25316177279225777,
      "grad_norm": 0.27871328592300415,
      "learning_rate": 8e-05,
      "loss": 1.7142,
      "step": 1151
    },
    {
      "epoch": 0.2533817222038931,
      "grad_norm": 0.2792799472808838,
      "learning_rate": 8e-05,
      "loss": 1.5703,
      "step": 1152
    },
    {
      "epoch": 0.25360167161552843,
      "grad_norm": 0.27358901500701904,
      "learning_rate": 8e-05,
      "loss": 1.7576,
      "step": 1153
    },
    {
      "epoch": 0.25382162102716377,
      "grad_norm": 0.26983192563056946,
      "learning_rate": 8e-05,
      "loss": 1.6646,
      "step": 1154
    },
    {
      "epoch": 0.2540415704387991,
      "grad_norm": 0.2711959183216095,
      "learning_rate": 8e-05,
      "loss": 1.7698,
      "step": 1155
    },
    {
      "epoch": 0.2542615198504344,
      "grad_norm": 0.28412333130836487,
      "learning_rate": 8e-05,
      "loss": 1.7446,
      "step": 1156
    },
    {
      "epoch": 0.2544814692620697,
      "grad_norm": 0.2698575258255005,
      "learning_rate": 8e-05,
      "loss": 1.6861,
      "step": 1157
    },
    {
      "epoch": 0.25470141867370505,
      "grad_norm": 0.2806732952594757,
      "learning_rate": 8e-05,
      "loss": 1.7308,
      "step": 1158
    },
    {
      "epoch": 0.2549213680853404,
      "grad_norm": 0.2715948522090912,
      "learning_rate": 8e-05,
      "loss": 1.852,
      "step": 1159
    },
    {
      "epoch": 0.2551413174969757,
      "grad_norm": 0.33048170804977417,
      "learning_rate": 8e-05,
      "loss": 1.881,
      "step": 1160
    },
    {
      "epoch": 0.255361266908611,
      "grad_norm": 0.27907994389533997,
      "learning_rate": 8e-05,
      "loss": 1.6501,
      "step": 1161
    },
    {
      "epoch": 0.25558121632024633,
      "grad_norm": 0.2747988998889923,
      "learning_rate": 8e-05,
      "loss": 1.7265,
      "step": 1162
    },
    {
      "epoch": 0.25580116573188166,
      "grad_norm": 0.28321677446365356,
      "learning_rate": 8e-05,
      "loss": 1.8602,
      "step": 1163
    },
    {
      "epoch": 0.256021115143517,
      "grad_norm": 0.2695465683937073,
      "learning_rate": 8e-05,
      "loss": 1.6091,
      "step": 1164
    },
    {
      "epoch": 0.25624106455515233,
      "grad_norm": 0.272135466337204,
      "learning_rate": 8e-05,
      "loss": 1.6236,
      "step": 1165
    },
    {
      "epoch": 0.25646101396678767,
      "grad_norm": 0.2715020775794983,
      "learning_rate": 8e-05,
      "loss": 1.674,
      "step": 1166
    },
    {
      "epoch": 0.25668096337842294,
      "grad_norm": 0.2879820764064789,
      "learning_rate": 8e-05,
      "loss": 1.8393,
      "step": 1167
    },
    {
      "epoch": 0.2569009127900583,
      "grad_norm": 0.2616657018661499,
      "learning_rate": 8e-05,
      "loss": 1.6391,
      "step": 1168
    },
    {
      "epoch": 0.2571208622016936,
      "grad_norm": 0.2558441460132599,
      "learning_rate": 8e-05,
      "loss": 1.606,
      "step": 1169
    },
    {
      "epoch": 0.25734081161332895,
      "grad_norm": 0.26944512128829956,
      "learning_rate": 8e-05,
      "loss": 1.7288,
      "step": 1170
    },
    {
      "epoch": 0.2575607610249643,
      "grad_norm": 0.26958367228507996,
      "learning_rate": 8e-05,
      "loss": 1.6233,
      "step": 1171
    },
    {
      "epoch": 0.25778071043659956,
      "grad_norm": 0.29003527760505676,
      "learning_rate": 8e-05,
      "loss": 1.777,
      "step": 1172
    },
    {
      "epoch": 0.2580006598482349,
      "grad_norm": 0.2677457630634308,
      "learning_rate": 8e-05,
      "loss": 1.6835,
      "step": 1173
    },
    {
      "epoch": 0.2582206092598702,
      "grad_norm": 0.28062689304351807,
      "learning_rate": 8e-05,
      "loss": 1.726,
      "step": 1174
    },
    {
      "epoch": 0.25844055867150556,
      "grad_norm": 0.26764920353889465,
      "learning_rate": 8e-05,
      "loss": 1.6575,
      "step": 1175
    },
    {
      "epoch": 0.2586605080831409,
      "grad_norm": 0.28183332085609436,
      "learning_rate": 8e-05,
      "loss": 1.784,
      "step": 1176
    },
    {
      "epoch": 0.25888045749477623,
      "grad_norm": 0.25718390941619873,
      "learning_rate": 8e-05,
      "loss": 1.6317,
      "step": 1177
    },
    {
      "epoch": 0.2591004069064115,
      "grad_norm": 0.25523149967193604,
      "learning_rate": 8e-05,
      "loss": 1.5634,
      "step": 1178
    },
    {
      "epoch": 0.25932035631804684,
      "grad_norm": 0.2539874315261841,
      "learning_rate": 8e-05,
      "loss": 1.5878,
      "step": 1179
    },
    {
      "epoch": 0.2595403057296822,
      "grad_norm": 0.2868393659591675,
      "learning_rate": 8e-05,
      "loss": 1.7301,
      "step": 1180
    },
    {
      "epoch": 0.2597602551413175,
      "grad_norm": 0.27819645404815674,
      "learning_rate": 8e-05,
      "loss": 1.6895,
      "step": 1181
    },
    {
      "epoch": 0.25998020455295284,
      "grad_norm": 0.27499255537986755,
      "learning_rate": 8e-05,
      "loss": 1.7462,
      "step": 1182
    },
    {
      "epoch": 0.2602001539645881,
      "grad_norm": 0.2858695685863495,
      "learning_rate": 8e-05,
      "loss": 1.8199,
      "step": 1183
    },
    {
      "epoch": 0.26042010337622346,
      "grad_norm": 0.2646760642528534,
      "learning_rate": 8e-05,
      "loss": 1.6597,
      "step": 1184
    },
    {
      "epoch": 0.2606400527878588,
      "grad_norm": 0.2831268310546875,
      "learning_rate": 8e-05,
      "loss": 1.8383,
      "step": 1185
    },
    {
      "epoch": 0.2608600021994941,
      "grad_norm": 0.2593746483325958,
      "learning_rate": 8e-05,
      "loss": 1.6115,
      "step": 1186
    },
    {
      "epoch": 0.26107995161112946,
      "grad_norm": 0.26519641280174255,
      "learning_rate": 8e-05,
      "loss": 1.5959,
      "step": 1187
    },
    {
      "epoch": 0.2612999010227648,
      "grad_norm": 0.2733252942562103,
      "learning_rate": 8e-05,
      "loss": 1.6318,
      "step": 1188
    },
    {
      "epoch": 0.26151985043440007,
      "grad_norm": 0.27299511432647705,
      "learning_rate": 8e-05,
      "loss": 1.7313,
      "step": 1189
    },
    {
      "epoch": 0.2617397998460354,
      "grad_norm": 0.2684955894947052,
      "learning_rate": 8e-05,
      "loss": 1.5826,
      "step": 1190
    },
    {
      "epoch": 0.26195974925767074,
      "grad_norm": 0.2747553586959839,
      "learning_rate": 8e-05,
      "loss": 1.7008,
      "step": 1191
    },
    {
      "epoch": 0.26217969866930607,
      "grad_norm": 0.26033639907836914,
      "learning_rate": 8e-05,
      "loss": 1.5571,
      "step": 1192
    },
    {
      "epoch": 0.2623996480809414,
      "grad_norm": 0.2640804350376129,
      "learning_rate": 8e-05,
      "loss": 1.5317,
      "step": 1193
    },
    {
      "epoch": 0.2626195974925767,
      "grad_norm": 0.27063700556755066,
      "learning_rate": 8e-05,
      "loss": 1.5501,
      "step": 1194
    },
    {
      "epoch": 0.262839546904212,
      "grad_norm": 0.2677111029624939,
      "learning_rate": 8e-05,
      "loss": 1.5894,
      "step": 1195
    },
    {
      "epoch": 0.26305949631584735,
      "grad_norm": 0.28144168853759766,
      "learning_rate": 8e-05,
      "loss": 1.7496,
      "step": 1196
    },
    {
      "epoch": 0.2632794457274827,
      "grad_norm": 0.2602388858795166,
      "learning_rate": 8e-05,
      "loss": 1.571,
      "step": 1197
    },
    {
      "epoch": 0.263499395139118,
      "grad_norm": 0.2941505014896393,
      "learning_rate": 8e-05,
      "loss": 1.6692,
      "step": 1198
    },
    {
      "epoch": 0.26371934455075335,
      "grad_norm": 0.264433354139328,
      "learning_rate": 8e-05,
      "loss": 1.6922,
      "step": 1199
    },
    {
      "epoch": 0.26393929396238863,
      "grad_norm": 0.25587090849876404,
      "learning_rate": 8e-05,
      "loss": 1.5599,
      "step": 1200
    },
    {
      "epoch": 0.26415924337402397,
      "grad_norm": 0.3012869358062744,
      "learning_rate": 8e-05,
      "loss": 1.9195,
      "step": 1201
    },
    {
      "epoch": 0.2643791927856593,
      "grad_norm": 0.2762719392776489,
      "learning_rate": 8e-05,
      "loss": 1.898,
      "step": 1202
    },
    {
      "epoch": 0.26459914219729463,
      "grad_norm": 0.2701188325881958,
      "learning_rate": 8e-05,
      "loss": 1.7312,
      "step": 1203
    },
    {
      "epoch": 0.26481909160892997,
      "grad_norm": 0.29665982723236084,
      "learning_rate": 8e-05,
      "loss": 1.8089,
      "step": 1204
    },
    {
      "epoch": 0.26503904102056525,
      "grad_norm": 0.26700517535209656,
      "learning_rate": 8e-05,
      "loss": 1.8401,
      "step": 1205
    },
    {
      "epoch": 0.2652589904322006,
      "grad_norm": 0.2828493118286133,
      "learning_rate": 8e-05,
      "loss": 1.8622,
      "step": 1206
    },
    {
      "epoch": 0.2654789398438359,
      "grad_norm": 0.2746271789073944,
      "learning_rate": 8e-05,
      "loss": 1.6521,
      "step": 1207
    },
    {
      "epoch": 0.26569888925547125,
      "grad_norm": 0.2882270812988281,
      "learning_rate": 8e-05,
      "loss": 1.7168,
      "step": 1208
    },
    {
      "epoch": 0.2659188386671066,
      "grad_norm": 0.29784512519836426,
      "learning_rate": 8e-05,
      "loss": 1.6968,
      "step": 1209
    },
    {
      "epoch": 0.2661387880787419,
      "grad_norm": 0.2807427942752838,
      "learning_rate": 8e-05,
      "loss": 1.6004,
      "step": 1210
    },
    {
      "epoch": 0.2663587374903772,
      "grad_norm": 0.2956424951553345,
      "learning_rate": 8e-05,
      "loss": 1.8325,
      "step": 1211
    },
    {
      "epoch": 0.26657868690201253,
      "grad_norm": 0.2647739350795746,
      "learning_rate": 8e-05,
      "loss": 1.6391,
      "step": 1212
    },
    {
      "epoch": 0.26679863631364786,
      "grad_norm": 0.2955171465873718,
      "learning_rate": 8e-05,
      "loss": 1.7893,
      "step": 1213
    },
    {
      "epoch": 0.2670185857252832,
      "grad_norm": 0.27241894602775574,
      "learning_rate": 8e-05,
      "loss": 1.781,
      "step": 1214
    },
    {
      "epoch": 0.26723853513691853,
      "grad_norm": 0.2841251492500305,
      "learning_rate": 8e-05,
      "loss": 1.8612,
      "step": 1215
    },
    {
      "epoch": 0.2674584845485538,
      "grad_norm": 0.327891081571579,
      "learning_rate": 8e-05,
      "loss": 1.844,
      "step": 1216
    },
    {
      "epoch": 0.26767843396018914,
      "grad_norm": 0.26434099674224854,
      "learning_rate": 8e-05,
      "loss": 1.6325,
      "step": 1217
    },
    {
      "epoch": 0.2678983833718245,
      "grad_norm": 0.2868417799472809,
      "learning_rate": 8e-05,
      "loss": 1.7087,
      "step": 1218
    },
    {
      "epoch": 0.2681183327834598,
      "grad_norm": 0.27408069372177124,
      "learning_rate": 8e-05,
      "loss": 1.6006,
      "step": 1219
    },
    {
      "epoch": 0.26833828219509515,
      "grad_norm": 0.2697390019893646,
      "learning_rate": 8e-05,
      "loss": 1.6833,
      "step": 1220
    },
    {
      "epoch": 0.2685582316067304,
      "grad_norm": 0.27598559856414795,
      "learning_rate": 8e-05,
      "loss": 1.7192,
      "step": 1221
    },
    {
      "epoch": 0.26877818101836576,
      "grad_norm": 0.26871007680892944,
      "learning_rate": 8e-05,
      "loss": 1.6301,
      "step": 1222
    },
    {
      "epoch": 0.2689981304300011,
      "grad_norm": 0.2739337980747223,
      "learning_rate": 8e-05,
      "loss": 1.6828,
      "step": 1223
    },
    {
      "epoch": 0.2692180798416364,
      "grad_norm": 0.286530464887619,
      "learning_rate": 8e-05,
      "loss": 1.6484,
      "step": 1224
    },
    {
      "epoch": 0.26943802925327176,
      "grad_norm": 0.27509886026382446,
      "learning_rate": 8e-05,
      "loss": 1.6647,
      "step": 1225
    },
    {
      "epoch": 0.2696579786649071,
      "grad_norm": 0.2916969358921051,
      "learning_rate": 8e-05,
      "loss": 1.7908,
      "step": 1226
    },
    {
      "epoch": 0.2698779280765424,
      "grad_norm": 0.26566174626350403,
      "learning_rate": 8e-05,
      "loss": 1.6075,
      "step": 1227
    },
    {
      "epoch": 0.2700978774881777,
      "grad_norm": 0.27648022770881653,
      "learning_rate": 8e-05,
      "loss": 1.7536,
      "step": 1228
    },
    {
      "epoch": 0.27031782689981304,
      "grad_norm": 0.27313023805618286,
      "learning_rate": 8e-05,
      "loss": 1.6978,
      "step": 1229
    },
    {
      "epoch": 0.2705377763114484,
      "grad_norm": 0.2755061388015747,
      "learning_rate": 8e-05,
      "loss": 1.7196,
      "step": 1230
    },
    {
      "epoch": 0.2707577257230837,
      "grad_norm": 0.25907769799232483,
      "learning_rate": 8e-05,
      "loss": 1.5518,
      "step": 1231
    },
    {
      "epoch": 0.270977675134719,
      "grad_norm": 0.26485681533813477,
      "learning_rate": 8e-05,
      "loss": 1.5053,
      "step": 1232
    },
    {
      "epoch": 0.2711976245463543,
      "grad_norm": 0.27980178594589233,
      "learning_rate": 8e-05,
      "loss": 1.7824,
      "step": 1233
    },
    {
      "epoch": 0.27141757395798966,
      "grad_norm": 0.2750954329967499,
      "learning_rate": 8e-05,
      "loss": 1.6973,
      "step": 1234
    },
    {
      "epoch": 0.271637523369625,
      "grad_norm": 0.27367594838142395,
      "learning_rate": 8e-05,
      "loss": 1.6691,
      "step": 1235
    },
    {
      "epoch": 0.2718574727812603,
      "grad_norm": 0.27089521288871765,
      "learning_rate": 8e-05,
      "loss": 1.7532,
      "step": 1236
    },
    {
      "epoch": 0.27207742219289566,
      "grad_norm": 0.30656641721725464,
      "learning_rate": 8e-05,
      "loss": 1.8411,
      "step": 1237
    },
    {
      "epoch": 0.27229737160453094,
      "grad_norm": 0.25732672214508057,
      "learning_rate": 8e-05,
      "loss": 1.5599,
      "step": 1238
    },
    {
      "epoch": 0.27251732101616627,
      "grad_norm": 0.2643807828426361,
      "learning_rate": 8e-05,
      "loss": 1.6654,
      "step": 1239
    },
    {
      "epoch": 0.2727372704278016,
      "grad_norm": 0.2703326344490051,
      "learning_rate": 8e-05,
      "loss": 1.594,
      "step": 1240
    },
    {
      "epoch": 0.27295721983943694,
      "grad_norm": 0.27907243371009827,
      "learning_rate": 8e-05,
      "loss": 1.7531,
      "step": 1241
    },
    {
      "epoch": 0.27317716925107227,
      "grad_norm": 0.2482902854681015,
      "learning_rate": 8e-05,
      "loss": 1.3586,
      "step": 1242
    },
    {
      "epoch": 0.27339711866270755,
      "grad_norm": 0.2879469394683838,
      "learning_rate": 8e-05,
      "loss": 1.76,
      "step": 1243
    },
    {
      "epoch": 0.2736170680743429,
      "grad_norm": 0.26334571838378906,
      "learning_rate": 8e-05,
      "loss": 1.536,
      "step": 1244
    },
    {
      "epoch": 0.2738370174859782,
      "grad_norm": 0.27328065037727356,
      "learning_rate": 8e-05,
      "loss": 1.7199,
      "step": 1245
    },
    {
      "epoch": 0.27405696689761355,
      "grad_norm": 0.27392926812171936,
      "learning_rate": 8e-05,
      "loss": 1.7731,
      "step": 1246
    },
    {
      "epoch": 0.2742769163092489,
      "grad_norm": 0.29755476117134094,
      "learning_rate": 8e-05,
      "loss": 1.7184,
      "step": 1247
    },
    {
      "epoch": 0.2744968657208842,
      "grad_norm": 0.29554107785224915,
      "learning_rate": 8e-05,
      "loss": 1.7442,
      "step": 1248
    },
    {
      "epoch": 0.2747168151325195,
      "grad_norm": 0.2562367618083954,
      "learning_rate": 8e-05,
      "loss": 1.63,
      "step": 1249
    },
    {
      "epoch": 0.27493676454415483,
      "grad_norm": 0.27746453881263733,
      "learning_rate": 8e-05,
      "loss": 1.7396,
      "step": 1250
    },
    {
      "epoch": 0.27515671395579017,
      "grad_norm": 0.2747843265533447,
      "learning_rate": 8e-05,
      "loss": 1.6628,
      "step": 1251
    },
    {
      "epoch": 0.2753766633674255,
      "grad_norm": 0.2650463581085205,
      "learning_rate": 8e-05,
      "loss": 1.6409,
      "step": 1252
    },
    {
      "epoch": 0.27559661277906083,
      "grad_norm": 0.30537328124046326,
      "learning_rate": 8e-05,
      "loss": 1.4927,
      "step": 1253
    },
    {
      "epoch": 0.2758165621906961,
      "grad_norm": 0.26015424728393555,
      "learning_rate": 8e-05,
      "loss": 1.718,
      "step": 1254
    },
    {
      "epoch": 0.27603651160233145,
      "grad_norm": 0.2512992322444916,
      "learning_rate": 8e-05,
      "loss": 1.4757,
      "step": 1255
    },
    {
      "epoch": 0.2762564610139668,
      "grad_norm": 0.28478461503982544,
      "learning_rate": 8e-05,
      "loss": 1.9081,
      "step": 1256
    },
    {
      "epoch": 0.2764764104256021,
      "grad_norm": 0.28490516543388367,
      "learning_rate": 8e-05,
      "loss": 1.8495,
      "step": 1257
    },
    {
      "epoch": 0.27669635983723745,
      "grad_norm": 0.2758481204509735,
      "learning_rate": 8e-05,
      "loss": 1.767,
      "step": 1258
    },
    {
      "epoch": 0.2769163092488728,
      "grad_norm": 0.28743213415145874,
      "learning_rate": 8e-05,
      "loss": 1.6548,
      "step": 1259
    },
    {
      "epoch": 0.27713625866050806,
      "grad_norm": 0.2738385796546936,
      "learning_rate": 8e-05,
      "loss": 1.5616,
      "step": 1260
    },
    {
      "epoch": 0.2773562080721434,
      "grad_norm": 0.27758583426475525,
      "learning_rate": 8e-05,
      "loss": 1.7793,
      "step": 1261
    },
    {
      "epoch": 0.27757615748377873,
      "grad_norm": 0.2830480635166168,
      "learning_rate": 8e-05,
      "loss": 1.8048,
      "step": 1262
    },
    {
      "epoch": 0.27779610689541406,
      "grad_norm": 0.296036034822464,
      "learning_rate": 8e-05,
      "loss": 1.7844,
      "step": 1263
    },
    {
      "epoch": 0.2780160563070494,
      "grad_norm": 0.28651297092437744,
      "learning_rate": 8e-05,
      "loss": 1.7239,
      "step": 1264
    },
    {
      "epoch": 0.2782360057186847,
      "grad_norm": 0.2826116979122162,
      "learning_rate": 8e-05,
      "loss": 1.8415,
      "step": 1265
    },
    {
      "epoch": 0.27845595513032,
      "grad_norm": 0.27445724606513977,
      "learning_rate": 8e-05,
      "loss": 1.6738,
      "step": 1266
    },
    {
      "epoch": 0.27867590454195534,
      "grad_norm": 0.28153640031814575,
      "learning_rate": 8e-05,
      "loss": 1.6519,
      "step": 1267
    },
    {
      "epoch": 0.2788958539535907,
      "grad_norm": 0.27389946579933167,
      "learning_rate": 8e-05,
      "loss": 1.681,
      "step": 1268
    },
    {
      "epoch": 0.279115803365226,
      "grad_norm": 0.2639203667640686,
      "learning_rate": 8e-05,
      "loss": 1.6398,
      "step": 1269
    },
    {
      "epoch": 0.27933575277686135,
      "grad_norm": 0.2787509560585022,
      "learning_rate": 8e-05,
      "loss": 1.7199,
      "step": 1270
    },
    {
      "epoch": 0.2795557021884966,
      "grad_norm": 0.28468430042266846,
      "learning_rate": 8e-05,
      "loss": 1.8668,
      "step": 1271
    },
    {
      "epoch": 0.27977565160013196,
      "grad_norm": 0.2907005250453949,
      "learning_rate": 8e-05,
      "loss": 1.9328,
      "step": 1272
    },
    {
      "epoch": 0.2799956010117673,
      "grad_norm": 0.2607463300228119,
      "learning_rate": 8e-05,
      "loss": 1.5958,
      "step": 1273
    },
    {
      "epoch": 0.2802155504234026,
      "grad_norm": 0.2695181965827942,
      "learning_rate": 8e-05,
      "loss": 1.6708,
      "step": 1274
    },
    {
      "epoch": 0.28043549983503796,
      "grad_norm": 0.28671538829803467,
      "learning_rate": 8e-05,
      "loss": 1.7736,
      "step": 1275
    },
    {
      "epoch": 0.28065544924667324,
      "grad_norm": 0.3246489465236664,
      "learning_rate": 8e-05,
      "loss": 1.8145,
      "step": 1276
    },
    {
      "epoch": 0.2808753986583086,
      "grad_norm": 0.2879314720630646,
      "learning_rate": 8e-05,
      "loss": 1.782,
      "step": 1277
    },
    {
      "epoch": 0.2810953480699439,
      "grad_norm": 0.27141574025154114,
      "learning_rate": 8e-05,
      "loss": 1.8069,
      "step": 1278
    },
    {
      "epoch": 0.28131529748157924,
      "grad_norm": 0.2893892228603363,
      "learning_rate": 8e-05,
      "loss": 1.7893,
      "step": 1279
    },
    {
      "epoch": 0.2815352468932146,
      "grad_norm": 0.2985538840293884,
      "learning_rate": 8e-05,
      "loss": 1.7804,
      "step": 1280
    },
    {
      "epoch": 0.2817551963048499,
      "grad_norm": 0.2664276957511902,
      "learning_rate": 8e-05,
      "loss": 1.6785,
      "step": 1281
    },
    {
      "epoch": 0.2819751457164852,
      "grad_norm": 0.3002198040485382,
      "learning_rate": 8e-05,
      "loss": 1.6109,
      "step": 1282
    },
    {
      "epoch": 0.2821950951281205,
      "grad_norm": 0.27687907218933105,
      "learning_rate": 8e-05,
      "loss": 1.6322,
      "step": 1283
    },
    {
      "epoch": 0.28241504453975586,
      "grad_norm": 0.28822144865989685,
      "learning_rate": 8e-05,
      "loss": 1.6785,
      "step": 1284
    },
    {
      "epoch": 0.2826349939513912,
      "grad_norm": 0.2801685333251953,
      "learning_rate": 8e-05,
      "loss": 1.69,
      "step": 1285
    },
    {
      "epoch": 0.2828549433630265,
      "grad_norm": 0.27876734733581543,
      "learning_rate": 8e-05,
      "loss": 1.6442,
      "step": 1286
    },
    {
      "epoch": 0.2830748927746618,
      "grad_norm": 0.2990095317363739,
      "learning_rate": 8e-05,
      "loss": 1.7439,
      "step": 1287
    },
    {
      "epoch": 0.28329484218629714,
      "grad_norm": 0.2710682451725006,
      "learning_rate": 8e-05,
      "loss": 1.6908,
      "step": 1288
    },
    {
      "epoch": 0.28351479159793247,
      "grad_norm": 0.2922731935977936,
      "learning_rate": 8e-05,
      "loss": 1.8361,
      "step": 1289
    },
    {
      "epoch": 0.2837347410095678,
      "grad_norm": 0.2638223171234131,
      "learning_rate": 8e-05,
      "loss": 1.6233,
      "step": 1290
    },
    {
      "epoch": 0.28395469042120314,
      "grad_norm": 0.27564552426338196,
      "learning_rate": 8e-05,
      "loss": 1.7624,
      "step": 1291
    },
    {
      "epoch": 0.28417463983283847,
      "grad_norm": 0.28238940238952637,
      "learning_rate": 8e-05,
      "loss": 1.8649,
      "step": 1292
    },
    {
      "epoch": 0.28439458924447375,
      "grad_norm": 0.27798035740852356,
      "learning_rate": 8e-05,
      "loss": 1.7877,
      "step": 1293
    },
    {
      "epoch": 0.2846145386561091,
      "grad_norm": 0.29618534445762634,
      "learning_rate": 8e-05,
      "loss": 1.816,
      "step": 1294
    },
    {
      "epoch": 0.2848344880677444,
      "grad_norm": 0.27669045329093933,
      "learning_rate": 8e-05,
      "loss": 1.7014,
      "step": 1295
    },
    {
      "epoch": 0.28505443747937975,
      "grad_norm": 0.27973508834838867,
      "learning_rate": 8e-05,
      "loss": 1.7491,
      "step": 1296
    },
    {
      "epoch": 0.2852743868910151,
      "grad_norm": 0.28833356499671936,
      "learning_rate": 8e-05,
      "loss": 1.6948,
      "step": 1297
    },
    {
      "epoch": 0.28549433630265036,
      "grad_norm": 0.2751030921936035,
      "learning_rate": 8e-05,
      "loss": 1.6846,
      "step": 1298
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.2766781449317932,
      "learning_rate": 8e-05,
      "loss": 1.5442,
      "step": 1299
    },
    {
      "epoch": 0.28593423512592103,
      "grad_norm": 0.29664894938468933,
      "learning_rate": 8e-05,
      "loss": 1.6884,
      "step": 1300
    },
    {
      "epoch": 0.28615418453755637,
      "grad_norm": 0.2771795392036438,
      "learning_rate": 8e-05,
      "loss": 1.6479,
      "step": 1301
    },
    {
      "epoch": 0.2863741339491917,
      "grad_norm": 0.2623322904109955,
      "learning_rate": 8e-05,
      "loss": 1.5803,
      "step": 1302
    },
    {
      "epoch": 0.28659408336082703,
      "grad_norm": 0.2821153998374939,
      "learning_rate": 8e-05,
      "loss": 1.7758,
      "step": 1303
    },
    {
      "epoch": 0.2868140327724623,
      "grad_norm": 0.29058384895324707,
      "learning_rate": 8e-05,
      "loss": 1.7244,
      "step": 1304
    },
    {
      "epoch": 0.28703398218409765,
      "grad_norm": 0.2811940312385559,
      "learning_rate": 8e-05,
      "loss": 1.6708,
      "step": 1305
    },
    {
      "epoch": 0.287253931595733,
      "grad_norm": 0.2773367762565613,
      "learning_rate": 8e-05,
      "loss": 1.7857,
      "step": 1306
    },
    {
      "epoch": 0.2874738810073683,
      "grad_norm": 0.2689999043941498,
      "learning_rate": 8e-05,
      "loss": 1.7432,
      "step": 1307
    },
    {
      "epoch": 0.28769383041900365,
      "grad_norm": 0.26896870136260986,
      "learning_rate": 8e-05,
      "loss": 1.6389,
      "step": 1308
    },
    {
      "epoch": 0.2879137798306389,
      "grad_norm": 0.2981964349746704,
      "learning_rate": 8e-05,
      "loss": 1.8771,
      "step": 1309
    },
    {
      "epoch": 0.28813372924227426,
      "grad_norm": 0.2872856855392456,
      "learning_rate": 8e-05,
      "loss": 1.785,
      "step": 1310
    },
    {
      "epoch": 0.2883536786539096,
      "grad_norm": 0.3186649680137634,
      "learning_rate": 8e-05,
      "loss": 1.9051,
      "step": 1311
    },
    {
      "epoch": 0.28857362806554493,
      "grad_norm": 0.2802119255065918,
      "learning_rate": 8e-05,
      "loss": 1.6532,
      "step": 1312
    },
    {
      "epoch": 0.28879357747718026,
      "grad_norm": 0.2864134907722473,
      "learning_rate": 8e-05,
      "loss": 1.7373,
      "step": 1313
    },
    {
      "epoch": 0.2890135268888156,
      "grad_norm": 0.2739737331867218,
      "learning_rate": 8e-05,
      "loss": 1.5365,
      "step": 1314
    },
    {
      "epoch": 0.2892334763004509,
      "grad_norm": 0.2707555294036865,
      "learning_rate": 8e-05,
      "loss": 1.6516,
      "step": 1315
    },
    {
      "epoch": 0.2894534257120862,
      "grad_norm": 0.2895212173461914,
      "learning_rate": 8e-05,
      "loss": 1.5634,
      "step": 1316
    },
    {
      "epoch": 0.28967337512372154,
      "grad_norm": 0.26424047350883484,
      "learning_rate": 8e-05,
      "loss": 1.6543,
      "step": 1317
    },
    {
      "epoch": 0.2898933245353569,
      "grad_norm": 0.26237159967422485,
      "learning_rate": 8e-05,
      "loss": 1.6814,
      "step": 1318
    },
    {
      "epoch": 0.2901132739469922,
      "grad_norm": 0.27964159846305847,
      "learning_rate": 8e-05,
      "loss": 1.7505,
      "step": 1319
    },
    {
      "epoch": 0.2903332233586275,
      "grad_norm": 0.27128270268440247,
      "learning_rate": 8e-05,
      "loss": 1.7229,
      "step": 1320
    },
    {
      "epoch": 0.2905531727702628,
      "grad_norm": 0.3012688159942627,
      "learning_rate": 8e-05,
      "loss": 1.6851,
      "step": 1321
    },
    {
      "epoch": 0.29077312218189816,
      "grad_norm": 0.2725695073604584,
      "learning_rate": 8e-05,
      "loss": 1.6552,
      "step": 1322
    },
    {
      "epoch": 0.2909930715935335,
      "grad_norm": 0.2855455279350281,
      "learning_rate": 8e-05,
      "loss": 1.7779,
      "step": 1323
    },
    {
      "epoch": 0.2912130210051688,
      "grad_norm": 0.2906174659729004,
      "learning_rate": 8e-05,
      "loss": 1.8209,
      "step": 1324
    },
    {
      "epoch": 0.29143297041680416,
      "grad_norm": 0.26015472412109375,
      "learning_rate": 8e-05,
      "loss": 1.5403,
      "step": 1325
    },
    {
      "epoch": 0.29165291982843944,
      "grad_norm": 0.29065820574760437,
      "learning_rate": 8e-05,
      "loss": 1.8499,
      "step": 1326
    },
    {
      "epoch": 0.2918728692400748,
      "grad_norm": 0.28715917468070984,
      "learning_rate": 8e-05,
      "loss": 1.854,
      "step": 1327
    },
    {
      "epoch": 0.2920928186517101,
      "grad_norm": 0.26932859420776367,
      "learning_rate": 8e-05,
      "loss": 1.6254,
      "step": 1328
    },
    {
      "epoch": 0.29231276806334544,
      "grad_norm": 0.2757404148578644,
      "learning_rate": 8e-05,
      "loss": 1.6077,
      "step": 1329
    },
    {
      "epoch": 0.2925327174749808,
      "grad_norm": 0.26532551646232605,
      "learning_rate": 8e-05,
      "loss": 1.6551,
      "step": 1330
    },
    {
      "epoch": 0.29275266688661605,
      "grad_norm": 0.2754289209842682,
      "learning_rate": 8e-05,
      "loss": 1.7276,
      "step": 1331
    },
    {
      "epoch": 0.2929726162982514,
      "grad_norm": 0.290568470954895,
      "learning_rate": 8e-05,
      "loss": 1.7622,
      "step": 1332
    },
    {
      "epoch": 0.2931925657098867,
      "grad_norm": 0.3045903742313385,
      "learning_rate": 8e-05,
      "loss": 1.7937,
      "step": 1333
    },
    {
      "epoch": 0.29341251512152206,
      "grad_norm": 0.2594483196735382,
      "learning_rate": 8e-05,
      "loss": 1.6163,
      "step": 1334
    },
    {
      "epoch": 0.2936324645331574,
      "grad_norm": 0.3054102957248688,
      "learning_rate": 8e-05,
      "loss": 1.7767,
      "step": 1335
    },
    {
      "epoch": 0.2938524139447927,
      "grad_norm": 0.27347666025161743,
      "learning_rate": 8e-05,
      "loss": 1.682,
      "step": 1336
    },
    {
      "epoch": 0.294072363356428,
      "grad_norm": 0.2639494836330414,
      "learning_rate": 8e-05,
      "loss": 1.4616,
      "step": 1337
    },
    {
      "epoch": 0.29429231276806334,
      "grad_norm": 0.2842942178249359,
      "learning_rate": 8e-05,
      "loss": 1.7625,
      "step": 1338
    },
    {
      "epoch": 0.29451226217969867,
      "grad_norm": 0.2895960509777069,
      "learning_rate": 8e-05,
      "loss": 1.7127,
      "step": 1339
    },
    {
      "epoch": 0.294732211591334,
      "grad_norm": 0.2836678624153137,
      "learning_rate": 8e-05,
      "loss": 1.7765,
      "step": 1340
    },
    {
      "epoch": 0.29495216100296934,
      "grad_norm": 0.26315444707870483,
      "learning_rate": 8e-05,
      "loss": 1.6592,
      "step": 1341
    },
    {
      "epoch": 0.2951721104146046,
      "grad_norm": 0.2601313591003418,
      "learning_rate": 8e-05,
      "loss": 1.5803,
      "step": 1342
    },
    {
      "epoch": 0.29539205982623995,
      "grad_norm": 0.28084784746170044,
      "learning_rate": 8e-05,
      "loss": 1.6172,
      "step": 1343
    },
    {
      "epoch": 0.2956120092378753,
      "grad_norm": 0.27707698941230774,
      "learning_rate": 8e-05,
      "loss": 1.6774,
      "step": 1344
    },
    {
      "epoch": 0.2958319586495106,
      "grad_norm": 0.28750407695770264,
      "learning_rate": 8e-05,
      "loss": 1.7775,
      "step": 1345
    },
    {
      "epoch": 0.29605190806114595,
      "grad_norm": 0.27315664291381836,
      "learning_rate": 8e-05,
      "loss": 1.6578,
      "step": 1346
    },
    {
      "epoch": 0.2962718574727813,
      "grad_norm": 0.26131486892700195,
      "learning_rate": 8e-05,
      "loss": 1.6429,
      "step": 1347
    },
    {
      "epoch": 0.29649180688441656,
      "grad_norm": 0.27198976278305054,
      "learning_rate": 8e-05,
      "loss": 1.6594,
      "step": 1348
    },
    {
      "epoch": 0.2967117562960519,
      "grad_norm": 0.2785218060016632,
      "learning_rate": 8e-05,
      "loss": 1.6959,
      "step": 1349
    },
    {
      "epoch": 0.29693170570768723,
      "grad_norm": 0.26987215876579285,
      "learning_rate": 8e-05,
      "loss": 1.6561,
      "step": 1350
    },
    {
      "epoch": 0.29715165511932257,
      "grad_norm": 0.2634013295173645,
      "learning_rate": 8e-05,
      "loss": 1.6817,
      "step": 1351
    },
    {
      "epoch": 0.2973716045309579,
      "grad_norm": 0.2584557831287384,
      "learning_rate": 8e-05,
      "loss": 1.5104,
      "step": 1352
    },
    {
      "epoch": 0.2975915539425932,
      "grad_norm": 0.28787991404533386,
      "learning_rate": 8e-05,
      "loss": 1.8217,
      "step": 1353
    },
    {
      "epoch": 0.2978115033542285,
      "grad_norm": 0.5047094225883484,
      "learning_rate": 8e-05,
      "loss": 1.7733,
      "step": 1354
    },
    {
      "epoch": 0.29803145276586385,
      "grad_norm": 0.26776471734046936,
      "learning_rate": 8e-05,
      "loss": 1.6961,
      "step": 1355
    },
    {
      "epoch": 0.2982514021774992,
      "grad_norm": 0.30351778864860535,
      "learning_rate": 8e-05,
      "loss": 1.7104,
      "step": 1356
    },
    {
      "epoch": 0.2984713515891345,
      "grad_norm": 0.27889010310173035,
      "learning_rate": 8e-05,
      "loss": 1.7276,
      "step": 1357
    },
    {
      "epoch": 0.29869130100076985,
      "grad_norm": 0.2656184136867523,
      "learning_rate": 8e-05,
      "loss": 1.7438,
      "step": 1358
    },
    {
      "epoch": 0.2989112504124051,
      "grad_norm": 0.27338340878486633,
      "learning_rate": 8e-05,
      "loss": 1.7526,
      "step": 1359
    },
    {
      "epoch": 0.29913119982404046,
      "grad_norm": 0.3266398310661316,
      "learning_rate": 8e-05,
      "loss": 1.8091,
      "step": 1360
    },
    {
      "epoch": 0.2993511492356758,
      "grad_norm": 0.309469997882843,
      "learning_rate": 8e-05,
      "loss": 2.0485,
      "step": 1361
    },
    {
      "epoch": 0.29957109864731113,
      "grad_norm": 0.2768929600715637,
      "learning_rate": 8e-05,
      "loss": 1.7977,
      "step": 1362
    },
    {
      "epoch": 0.29979104805894646,
      "grad_norm": 0.27685433626174927,
      "learning_rate": 8e-05,
      "loss": 1.5712,
      "step": 1363
    },
    {
      "epoch": 0.30001099747058174,
      "grad_norm": 0.26404622197151184,
      "learning_rate": 8e-05,
      "loss": 1.6639,
      "step": 1364
    },
    {
      "epoch": 0.3002309468822171,
      "grad_norm": 0.2719237208366394,
      "learning_rate": 8e-05,
      "loss": 1.788,
      "step": 1365
    },
    {
      "epoch": 0.3004508962938524,
      "grad_norm": 0.27983394265174866,
      "learning_rate": 8e-05,
      "loss": 1.7361,
      "step": 1366
    },
    {
      "epoch": 0.30067084570548774,
      "grad_norm": 0.2673875689506531,
      "learning_rate": 8e-05,
      "loss": 1.6288,
      "step": 1367
    },
    {
      "epoch": 0.3008907951171231,
      "grad_norm": 0.2850426435470581,
      "learning_rate": 8e-05,
      "loss": 1.8328,
      "step": 1368
    },
    {
      "epoch": 0.3011107445287584,
      "grad_norm": 0.2577967345714569,
      "learning_rate": 8e-05,
      "loss": 1.6267,
      "step": 1369
    },
    {
      "epoch": 0.3013306939403937,
      "grad_norm": 0.276094913482666,
      "learning_rate": 8e-05,
      "loss": 1.7673,
      "step": 1370
    },
    {
      "epoch": 0.301550643352029,
      "grad_norm": 0.2834344208240509,
      "learning_rate": 8e-05,
      "loss": 1.6692,
      "step": 1371
    },
    {
      "epoch": 0.30177059276366436,
      "grad_norm": 0.2617560029029846,
      "learning_rate": 8e-05,
      "loss": 1.7734,
      "step": 1372
    },
    {
      "epoch": 0.3019905421752997,
      "grad_norm": 0.27122870087623596,
      "learning_rate": 8e-05,
      "loss": 1.6988,
      "step": 1373
    },
    {
      "epoch": 0.302210491586935,
      "grad_norm": 0.26526594161987305,
      "learning_rate": 8e-05,
      "loss": 1.7459,
      "step": 1374
    },
    {
      "epoch": 0.3024304409985703,
      "grad_norm": 0.2893051207065582,
      "learning_rate": 8e-05,
      "loss": 1.8214,
      "step": 1375
    },
    {
      "epoch": 0.30265039041020564,
      "grad_norm": 0.2735356092453003,
      "learning_rate": 8e-05,
      "loss": 1.8437,
      "step": 1376
    },
    {
      "epoch": 0.302870339821841,
      "grad_norm": 0.2743459939956665,
      "learning_rate": 8e-05,
      "loss": 1.8365,
      "step": 1377
    },
    {
      "epoch": 0.3030902892334763,
      "grad_norm": 0.28047019243240356,
      "learning_rate": 8e-05,
      "loss": 1.6143,
      "step": 1378
    },
    {
      "epoch": 0.30331023864511164,
      "grad_norm": 0.268197238445282,
      "learning_rate": 8e-05,
      "loss": 1.591,
      "step": 1379
    },
    {
      "epoch": 0.303530188056747,
      "grad_norm": 0.2890843451023102,
      "learning_rate": 8e-05,
      "loss": 1.7757,
      "step": 1380
    },
    {
      "epoch": 0.30375013746838225,
      "grad_norm": 0.2765072286128998,
      "learning_rate": 8e-05,
      "loss": 1.6363,
      "step": 1381
    },
    {
      "epoch": 0.3039700868800176,
      "grad_norm": 0.290147602558136,
      "learning_rate": 8e-05,
      "loss": 1.7615,
      "step": 1382
    },
    {
      "epoch": 0.3041900362916529,
      "grad_norm": 0.2721220850944519,
      "learning_rate": 8e-05,
      "loss": 1.7101,
      "step": 1383
    },
    {
      "epoch": 0.30440998570328826,
      "grad_norm": 0.27125662565231323,
      "learning_rate": 8e-05,
      "loss": 1.7291,
      "step": 1384
    },
    {
      "epoch": 0.3046299351149236,
      "grad_norm": 0.2594304084777832,
      "learning_rate": 8e-05,
      "loss": 1.6754,
      "step": 1385
    },
    {
      "epoch": 0.30484988452655887,
      "grad_norm": 0.28582707047462463,
      "learning_rate": 8e-05,
      "loss": 1.6808,
      "step": 1386
    },
    {
      "epoch": 0.3050698339381942,
      "grad_norm": 0.2853895425796509,
      "learning_rate": 8e-05,
      "loss": 1.779,
      "step": 1387
    },
    {
      "epoch": 0.30528978334982954,
      "grad_norm": 0.2580530345439911,
      "learning_rate": 8e-05,
      "loss": 1.6316,
      "step": 1388
    },
    {
      "epoch": 0.30550973276146487,
      "grad_norm": 0.2793220281600952,
      "learning_rate": 8e-05,
      "loss": 1.7326,
      "step": 1389
    },
    {
      "epoch": 0.3057296821731002,
      "grad_norm": 0.2672085165977478,
      "learning_rate": 8e-05,
      "loss": 1.6544,
      "step": 1390
    },
    {
      "epoch": 0.30594963158473554,
      "grad_norm": 0.27718111872673035,
      "learning_rate": 8e-05,
      "loss": 1.6307,
      "step": 1391
    },
    {
      "epoch": 0.3061695809963708,
      "grad_norm": 0.29295554757118225,
      "learning_rate": 8e-05,
      "loss": 1.502,
      "step": 1392
    },
    {
      "epoch": 0.30638953040800615,
      "grad_norm": 0.2840512990951538,
      "learning_rate": 8e-05,
      "loss": 1.6326,
      "step": 1393
    },
    {
      "epoch": 0.3066094798196415,
      "grad_norm": 0.2897029519081116,
      "learning_rate": 8e-05,
      "loss": 1.7543,
      "step": 1394
    },
    {
      "epoch": 0.3068294292312768,
      "grad_norm": 0.28060710430145264,
      "learning_rate": 8e-05,
      "loss": 1.7227,
      "step": 1395
    },
    {
      "epoch": 0.30704937864291215,
      "grad_norm": 0.27874305844306946,
      "learning_rate": 8e-05,
      "loss": 1.6639,
      "step": 1396
    },
    {
      "epoch": 0.30726932805454743,
      "grad_norm": 0.2679193615913391,
      "learning_rate": 8e-05,
      "loss": 1.7226,
      "step": 1397
    },
    {
      "epoch": 0.30748927746618276,
      "grad_norm": 0.2769779562950134,
      "learning_rate": 8e-05,
      "loss": 1.6384,
      "step": 1398
    },
    {
      "epoch": 0.3077092268778181,
      "grad_norm": 0.26620879769325256,
      "learning_rate": 8e-05,
      "loss": 1.7134,
      "step": 1399
    },
    {
      "epoch": 0.30792917628945343,
      "grad_norm": 0.277423620223999,
      "learning_rate": 8e-05,
      "loss": 1.7376,
      "step": 1400
    },
    {
      "epoch": 0.30814912570108877,
      "grad_norm": 0.2629416882991791,
      "learning_rate": 8e-05,
      "loss": 1.598,
      "step": 1401
    },
    {
      "epoch": 0.3083690751127241,
      "grad_norm": 0.2844812572002411,
      "learning_rate": 8e-05,
      "loss": 1.7067,
      "step": 1402
    },
    {
      "epoch": 0.3085890245243594,
      "grad_norm": 0.2731526494026184,
      "learning_rate": 8e-05,
      "loss": 1.8571,
      "step": 1403
    },
    {
      "epoch": 0.3088089739359947,
      "grad_norm": 0.287438303232193,
      "learning_rate": 8e-05,
      "loss": 1.7612,
      "step": 1404
    },
    {
      "epoch": 0.30902892334763005,
      "grad_norm": 0.266718327999115,
      "learning_rate": 8e-05,
      "loss": 1.6106,
      "step": 1405
    },
    {
      "epoch": 0.3092488727592654,
      "grad_norm": 0.28080686926841736,
      "learning_rate": 8e-05,
      "loss": 1.8281,
      "step": 1406
    },
    {
      "epoch": 0.3094688221709007,
      "grad_norm": 0.27558308839797974,
      "learning_rate": 8e-05,
      "loss": 1.8677,
      "step": 1407
    },
    {
      "epoch": 0.309688771582536,
      "grad_norm": 0.2798183262348175,
      "learning_rate": 8e-05,
      "loss": 1.7867,
      "step": 1408
    },
    {
      "epoch": 0.3099087209941713,
      "grad_norm": 0.25823187828063965,
      "learning_rate": 8e-05,
      "loss": 1.6743,
      "step": 1409
    },
    {
      "epoch": 0.31012867040580666,
      "grad_norm": 0.27356335520744324,
      "learning_rate": 8e-05,
      "loss": 1.7039,
      "step": 1410
    },
    {
      "epoch": 0.310348619817442,
      "grad_norm": 0.2842661440372467,
      "learning_rate": 8e-05,
      "loss": 1.7046,
      "step": 1411
    },
    {
      "epoch": 0.31056856922907733,
      "grad_norm": 0.2561197876930237,
      "learning_rate": 8e-05,
      "loss": 1.4887,
      "step": 1412
    },
    {
      "epoch": 0.31078851864071266,
      "grad_norm": 0.2851184904575348,
      "learning_rate": 8e-05,
      "loss": 1.7074,
      "step": 1413
    },
    {
      "epoch": 0.31100846805234794,
      "grad_norm": 0.2655506432056427,
      "learning_rate": 8e-05,
      "loss": 1.6049,
      "step": 1414
    },
    {
      "epoch": 0.3112284174639833,
      "grad_norm": 0.26412099599838257,
      "learning_rate": 8e-05,
      "loss": 1.6052,
      "step": 1415
    },
    {
      "epoch": 0.3114483668756186,
      "grad_norm": 0.3026227056980133,
      "learning_rate": 8e-05,
      "loss": 1.7085,
      "step": 1416
    },
    {
      "epoch": 0.31166831628725394,
      "grad_norm": 0.28821703791618347,
      "learning_rate": 8e-05,
      "loss": 1.7573,
      "step": 1417
    },
    {
      "epoch": 0.3118882656988893,
      "grad_norm": 0.26806455850601196,
      "learning_rate": 8e-05,
      "loss": 1.7136,
      "step": 1418
    },
    {
      "epoch": 0.31210821511052456,
      "grad_norm": 0.28336799144744873,
      "learning_rate": 8e-05,
      "loss": 1.8445,
      "step": 1419
    },
    {
      "epoch": 0.3123281645221599,
      "grad_norm": 0.2772139012813568,
      "learning_rate": 8e-05,
      "loss": 1.692,
      "step": 1420
    },
    {
      "epoch": 0.3125481139337952,
      "grad_norm": 0.2815256714820862,
      "learning_rate": 8e-05,
      "loss": 1.77,
      "step": 1421
    },
    {
      "epoch": 0.31276806334543056,
      "grad_norm": 0.4029920697212219,
      "learning_rate": 8e-05,
      "loss": 1.8103,
      "step": 1422
    },
    {
      "epoch": 0.3129880127570659,
      "grad_norm": 0.2677610218524933,
      "learning_rate": 8e-05,
      "loss": 1.5898,
      "step": 1423
    },
    {
      "epoch": 0.3132079621687012,
      "grad_norm": 0.2605397701263428,
      "learning_rate": 8e-05,
      "loss": 1.5735,
      "step": 1424
    },
    {
      "epoch": 0.3134279115803365,
      "grad_norm": 0.2831586003303528,
      "learning_rate": 8e-05,
      "loss": 1.6641,
      "step": 1425
    },
    {
      "epoch": 0.31364786099197184,
      "grad_norm": 0.2746485471725464,
      "learning_rate": 8e-05,
      "loss": 1.618,
      "step": 1426
    },
    {
      "epoch": 0.3138678104036072,
      "grad_norm": 0.283342182636261,
      "learning_rate": 8e-05,
      "loss": 1.6963,
      "step": 1427
    },
    {
      "epoch": 0.3140877598152425,
      "grad_norm": 0.27635300159454346,
      "learning_rate": 8e-05,
      "loss": 1.6911,
      "step": 1428
    },
    {
      "epoch": 0.31430770922687784,
      "grad_norm": 0.2719132900238037,
      "learning_rate": 8e-05,
      "loss": 1.7063,
      "step": 1429
    },
    {
      "epoch": 0.3145276586385131,
      "grad_norm": 0.27162256836891174,
      "learning_rate": 8e-05,
      "loss": 1.6397,
      "step": 1430
    },
    {
      "epoch": 0.31474760805014845,
      "grad_norm": 0.2934938073158264,
      "learning_rate": 8e-05,
      "loss": 1.7555,
      "step": 1431
    },
    {
      "epoch": 0.3149675574617838,
      "grad_norm": 0.3060123920440674,
      "learning_rate": 8e-05,
      "loss": 1.642,
      "step": 1432
    },
    {
      "epoch": 0.3151875068734191,
      "grad_norm": 0.280846506357193,
      "learning_rate": 8e-05,
      "loss": 1.6805,
      "step": 1433
    },
    {
      "epoch": 0.31540745628505445,
      "grad_norm": 0.2768997550010681,
      "learning_rate": 8e-05,
      "loss": 1.7359,
      "step": 1434
    },
    {
      "epoch": 0.3156274056966898,
      "grad_norm": 0.29172810912132263,
      "learning_rate": 8e-05,
      "loss": 1.821,
      "step": 1435
    },
    {
      "epoch": 0.31584735510832507,
      "grad_norm": 0.30742648243904114,
      "learning_rate": 8e-05,
      "loss": 1.8198,
      "step": 1436
    },
    {
      "epoch": 0.3160673045199604,
      "grad_norm": 0.2889997065067291,
      "learning_rate": 8e-05,
      "loss": 1.6733,
      "step": 1437
    },
    {
      "epoch": 0.31628725393159574,
      "grad_norm": 0.2859675884246826,
      "learning_rate": 8e-05,
      "loss": 1.7655,
      "step": 1438
    },
    {
      "epoch": 0.31650720334323107,
      "grad_norm": 0.2926831543445587,
      "learning_rate": 8e-05,
      "loss": 1.7871,
      "step": 1439
    },
    {
      "epoch": 0.3167271527548664,
      "grad_norm": 0.28924524784088135,
      "learning_rate": 8e-05,
      "loss": 1.665,
      "step": 1440
    },
    {
      "epoch": 0.3169471021665017,
      "grad_norm": 0.2940097749233246,
      "learning_rate": 8e-05,
      "loss": 1.8364,
      "step": 1441
    },
    {
      "epoch": 0.317167051578137,
      "grad_norm": 0.2923974096775055,
      "learning_rate": 8e-05,
      "loss": 1.8071,
      "step": 1442
    },
    {
      "epoch": 0.31738700098977235,
      "grad_norm": 0.28991878032684326,
      "learning_rate": 8e-05,
      "loss": 1.7445,
      "step": 1443
    },
    {
      "epoch": 0.3176069504014077,
      "grad_norm": 0.283600777387619,
      "learning_rate": 8e-05,
      "loss": 1.8043,
      "step": 1444
    },
    {
      "epoch": 0.317826899813043,
      "grad_norm": 0.3082323372364044,
      "learning_rate": 8e-05,
      "loss": 1.7858,
      "step": 1445
    },
    {
      "epoch": 0.3180468492246783,
      "grad_norm": 0.28433462977409363,
      "learning_rate": 8e-05,
      "loss": 1.6911,
      "step": 1446
    },
    {
      "epoch": 0.31826679863631363,
      "grad_norm": 0.27776578068733215,
      "learning_rate": 8e-05,
      "loss": 1.7212,
      "step": 1447
    },
    {
      "epoch": 0.31848674804794896,
      "grad_norm": 0.29395151138305664,
      "learning_rate": 8e-05,
      "loss": 1.7221,
      "step": 1448
    },
    {
      "epoch": 0.3187066974595843,
      "grad_norm": 0.27507245540618896,
      "learning_rate": 8e-05,
      "loss": 1.7358,
      "step": 1449
    },
    {
      "epoch": 0.31892664687121963,
      "grad_norm": 0.25614190101623535,
      "learning_rate": 8e-05,
      "loss": 1.5138,
      "step": 1450
    },
    {
      "epoch": 0.31914659628285497,
      "grad_norm": 0.2908024489879608,
      "learning_rate": 8e-05,
      "loss": 1.756,
      "step": 1451
    },
    {
      "epoch": 0.31936654569449024,
      "grad_norm": 0.2729463577270508,
      "learning_rate": 8e-05,
      "loss": 1.5542,
      "step": 1452
    },
    {
      "epoch": 0.3195864951061256,
      "grad_norm": 0.27094194293022156,
      "learning_rate": 8e-05,
      "loss": 1.5917,
      "step": 1453
    },
    {
      "epoch": 0.3198064445177609,
      "grad_norm": 0.28125494718551636,
      "learning_rate": 8e-05,
      "loss": 1.6584,
      "step": 1454
    },
    {
      "epoch": 0.32002639392939625,
      "grad_norm": 0.29033198952674866,
      "learning_rate": 8e-05,
      "loss": 1.7332,
      "step": 1455
    },
    {
      "epoch": 0.3202463433410316,
      "grad_norm": 0.26570284366607666,
      "learning_rate": 8e-05,
      "loss": 1.6159,
      "step": 1456
    },
    {
      "epoch": 0.32046629275266686,
      "grad_norm": 0.307412713766098,
      "learning_rate": 8e-05,
      "loss": 1.7351,
      "step": 1457
    },
    {
      "epoch": 0.3206862421643022,
      "grad_norm": 0.29387474060058594,
      "learning_rate": 8e-05,
      "loss": 1.9386,
      "step": 1458
    },
    {
      "epoch": 0.3209061915759375,
      "grad_norm": 0.26545315980911255,
      "learning_rate": 8e-05,
      "loss": 1.6343,
      "step": 1459
    },
    {
      "epoch": 0.32112614098757286,
      "grad_norm": 0.279238224029541,
      "learning_rate": 8e-05,
      "loss": 1.6245,
      "step": 1460
    },
    {
      "epoch": 0.3213460903992082,
      "grad_norm": 0.2766862213611603,
      "learning_rate": 8e-05,
      "loss": 1.7135,
      "step": 1461
    },
    {
      "epoch": 0.32156603981084353,
      "grad_norm": 0.2705351412296295,
      "learning_rate": 8e-05,
      "loss": 1.6526,
      "step": 1462
    },
    {
      "epoch": 0.3217859892224788,
      "grad_norm": 0.27870967984199524,
      "learning_rate": 8e-05,
      "loss": 1.6512,
      "step": 1463
    },
    {
      "epoch": 0.32200593863411414,
      "grad_norm": 0.284407377243042,
      "learning_rate": 8e-05,
      "loss": 1.755,
      "step": 1464
    },
    {
      "epoch": 0.3222258880457495,
      "grad_norm": 0.2897641062736511,
      "learning_rate": 8e-05,
      "loss": 1.8383,
      "step": 1465
    },
    {
      "epoch": 0.3224458374573848,
      "grad_norm": 0.2667568624019623,
      "learning_rate": 8e-05,
      "loss": 1.6989,
      "step": 1466
    },
    {
      "epoch": 0.32266578686902014,
      "grad_norm": 0.26580294966697693,
      "learning_rate": 8e-05,
      "loss": 1.5895,
      "step": 1467
    },
    {
      "epoch": 0.3228857362806554,
      "grad_norm": 0.26188549399375916,
      "learning_rate": 8e-05,
      "loss": 1.5799,
      "step": 1468
    },
    {
      "epoch": 0.32310568569229076,
      "grad_norm": 0.27703747153282166,
      "learning_rate": 8e-05,
      "loss": 1.8306,
      "step": 1469
    },
    {
      "epoch": 0.3233256351039261,
      "grad_norm": 0.27643802762031555,
      "learning_rate": 8e-05,
      "loss": 1.6864,
      "step": 1470
    },
    {
      "epoch": 0.3235455845155614,
      "grad_norm": 0.27216553688049316,
      "learning_rate": 8e-05,
      "loss": 1.6006,
      "step": 1471
    },
    {
      "epoch": 0.32376553392719676,
      "grad_norm": 0.2984940707683563,
      "learning_rate": 8e-05,
      "loss": 1.7548,
      "step": 1472
    },
    {
      "epoch": 0.3239854833388321,
      "grad_norm": 0.30579298734664917,
      "learning_rate": 8e-05,
      "loss": 1.8307,
      "step": 1473
    },
    {
      "epoch": 0.32420543275046737,
      "grad_norm": 0.27524709701538086,
      "learning_rate": 8e-05,
      "loss": 1.6134,
      "step": 1474
    },
    {
      "epoch": 0.3244253821621027,
      "grad_norm": 0.2788650393486023,
      "learning_rate": 8e-05,
      "loss": 1.8194,
      "step": 1475
    },
    {
      "epoch": 0.32464533157373804,
      "grad_norm": 0.28263744711875916,
      "learning_rate": 8e-05,
      "loss": 1.7633,
      "step": 1476
    },
    {
      "epoch": 0.3248652809853734,
      "grad_norm": 0.30234408378601074,
      "learning_rate": 8e-05,
      "loss": 1.6057,
      "step": 1477
    },
    {
      "epoch": 0.3250852303970087,
      "grad_norm": 0.2820134162902832,
      "learning_rate": 8e-05,
      "loss": 1.6913,
      "step": 1478
    },
    {
      "epoch": 0.325305179808644,
      "grad_norm": 0.28929245471954346,
      "learning_rate": 8e-05,
      "loss": 1.9538,
      "step": 1479
    },
    {
      "epoch": 0.3255251292202793,
      "grad_norm": 0.26399463415145874,
      "learning_rate": 8e-05,
      "loss": 1.7309,
      "step": 1480
    },
    {
      "epoch": 0.32574507863191465,
      "grad_norm": 0.2722630202770233,
      "learning_rate": 8e-05,
      "loss": 1.5595,
      "step": 1481
    },
    {
      "epoch": 0.32596502804355,
      "grad_norm": 0.2759261727333069,
      "learning_rate": 8e-05,
      "loss": 1.6272,
      "step": 1482
    },
    {
      "epoch": 0.3261849774551853,
      "grad_norm": 0.28047022223472595,
      "learning_rate": 8e-05,
      "loss": 1.6933,
      "step": 1483
    },
    {
      "epoch": 0.32640492686682065,
      "grad_norm": 0.2835995554924011,
      "learning_rate": 8e-05,
      "loss": 1.7165,
      "step": 1484
    },
    {
      "epoch": 0.32662487627845593,
      "grad_norm": 0.28965097665786743,
      "learning_rate": 8e-05,
      "loss": 1.6399,
      "step": 1485
    },
    {
      "epoch": 0.32684482569009127,
      "grad_norm": 0.2729817032814026,
      "learning_rate": 8e-05,
      "loss": 1.7397,
      "step": 1486
    },
    {
      "epoch": 0.3270647751017266,
      "grad_norm": 0.26809874176979065,
      "learning_rate": 8e-05,
      "loss": 1.6954,
      "step": 1487
    },
    {
      "epoch": 0.32728472451336194,
      "grad_norm": 0.29766684770584106,
      "learning_rate": 8e-05,
      "loss": 1.6947,
      "step": 1488
    },
    {
      "epoch": 0.32750467392499727,
      "grad_norm": 0.27032670378685,
      "learning_rate": 8e-05,
      "loss": 1.8036,
      "step": 1489
    },
    {
      "epoch": 0.32772462333663255,
      "grad_norm": 0.2694716453552246,
      "learning_rate": 8e-05,
      "loss": 1.6856,
      "step": 1490
    },
    {
      "epoch": 0.3279445727482679,
      "grad_norm": 0.27968841791152954,
      "learning_rate": 8e-05,
      "loss": 1.7466,
      "step": 1491
    },
    {
      "epoch": 0.3281645221599032,
      "grad_norm": 0.2956348955631256,
      "learning_rate": 8e-05,
      "loss": 1.833,
      "step": 1492
    },
    {
      "epoch": 0.32838447157153855,
      "grad_norm": 0.27069491147994995,
      "learning_rate": 8e-05,
      "loss": 1.715,
      "step": 1493
    },
    {
      "epoch": 0.3286044209831739,
      "grad_norm": 0.26747795939445496,
      "learning_rate": 8e-05,
      "loss": 1.6663,
      "step": 1494
    },
    {
      "epoch": 0.3288243703948092,
      "grad_norm": 0.2619915008544922,
      "learning_rate": 8e-05,
      "loss": 1.6503,
      "step": 1495
    },
    {
      "epoch": 0.3290443198064445,
      "grad_norm": 0.2720276117324829,
      "learning_rate": 8e-05,
      "loss": 1.7174,
      "step": 1496
    },
    {
      "epoch": 0.32926426921807983,
      "grad_norm": 0.26874253153800964,
      "learning_rate": 8e-05,
      "loss": 1.6503,
      "step": 1497
    },
    {
      "epoch": 0.32948421862971516,
      "grad_norm": 0.28397336602211,
      "learning_rate": 8e-05,
      "loss": 1.67,
      "step": 1498
    },
    {
      "epoch": 0.3297041680413505,
      "grad_norm": 0.2544403076171875,
      "learning_rate": 8e-05,
      "loss": 1.5153,
      "step": 1499
    },
    {
      "epoch": 0.32992411745298583,
      "grad_norm": 0.2819180488586426,
      "learning_rate": 8e-05,
      "loss": 1.6704,
      "step": 1500
    },
    {
      "epoch": 0.3301440668646211,
      "grad_norm": 0.28150951862335205,
      "learning_rate": 8e-05,
      "loss": 1.8451,
      "step": 1501
    },
    {
      "epoch": 0.33036401627625644,
      "grad_norm": 0.27396339178085327,
      "learning_rate": 8e-05,
      "loss": 1.7631,
      "step": 1502
    },
    {
      "epoch": 0.3305839656878918,
      "grad_norm": 0.2954351007938385,
      "learning_rate": 8e-05,
      "loss": 1.8101,
      "step": 1503
    },
    {
      "epoch": 0.3308039150995271,
      "grad_norm": 0.27129319310188293,
      "learning_rate": 8e-05,
      "loss": 1.6484,
      "step": 1504
    },
    {
      "epoch": 0.33102386451116245,
      "grad_norm": 0.27612754702568054,
      "learning_rate": 8e-05,
      "loss": 1.6178,
      "step": 1505
    },
    {
      "epoch": 0.3312438139227978,
      "grad_norm": 0.26097655296325684,
      "learning_rate": 8e-05,
      "loss": 1.5781,
      "step": 1506
    },
    {
      "epoch": 0.33146376333443306,
      "grad_norm": 0.2704753577709198,
      "learning_rate": 8e-05,
      "loss": 1.6919,
      "step": 1507
    },
    {
      "epoch": 0.3316837127460684,
      "grad_norm": 0.26866593956947327,
      "learning_rate": 8e-05,
      "loss": 1.6795,
      "step": 1508
    },
    {
      "epoch": 0.3319036621577037,
      "grad_norm": 0.31797948479652405,
      "learning_rate": 8e-05,
      "loss": 1.7511,
      "step": 1509
    },
    {
      "epoch": 0.33212361156933906,
      "grad_norm": 0.29456841945648193,
      "learning_rate": 8e-05,
      "loss": 1.7041,
      "step": 1510
    },
    {
      "epoch": 0.3323435609809744,
      "grad_norm": 0.28345033526420593,
      "learning_rate": 8e-05,
      "loss": 1.7499,
      "step": 1511
    },
    {
      "epoch": 0.3325635103926097,
      "grad_norm": 0.28679129481315613,
      "learning_rate": 8e-05,
      "loss": 1.8304,
      "step": 1512
    },
    {
      "epoch": 0.332783459804245,
      "grad_norm": 0.2799399793148041,
      "learning_rate": 8e-05,
      "loss": 1.6461,
      "step": 1513
    },
    {
      "epoch": 0.33300340921588034,
      "grad_norm": 0.3234422206878662,
      "learning_rate": 8e-05,
      "loss": 1.5622,
      "step": 1514
    },
    {
      "epoch": 0.3332233586275157,
      "grad_norm": 0.27786344289779663,
      "learning_rate": 8e-05,
      "loss": 1.6718,
      "step": 1515
    },
    {
      "epoch": 0.333443308039151,
      "grad_norm": 0.27040839195251465,
      "learning_rate": 8e-05,
      "loss": 1.7428,
      "step": 1516
    },
    {
      "epoch": 0.33366325745078634,
      "grad_norm": 0.2837252616882324,
      "learning_rate": 8e-05,
      "loss": 1.6929,
      "step": 1517
    },
    {
      "epoch": 0.3338832068624216,
      "grad_norm": 0.27352792024612427,
      "learning_rate": 8e-05,
      "loss": 1.7804,
      "step": 1518
    },
    {
      "epoch": 0.33410315627405696,
      "grad_norm": 0.27237218618392944,
      "learning_rate": 8e-05,
      "loss": 1.7652,
      "step": 1519
    },
    {
      "epoch": 0.3343231056856923,
      "grad_norm": 0.3166270852088928,
      "learning_rate": 8e-05,
      "loss": 1.6363,
      "step": 1520
    },
    {
      "epoch": 0.3345430550973276,
      "grad_norm": 0.2650817930698395,
      "learning_rate": 8e-05,
      "loss": 1.6954,
      "step": 1521
    },
    {
      "epoch": 0.33476300450896296,
      "grad_norm": 0.2907481789588928,
      "learning_rate": 8e-05,
      "loss": 1.809,
      "step": 1522
    },
    {
      "epoch": 0.33498295392059824,
      "grad_norm": 0.2754502296447754,
      "learning_rate": 8e-05,
      "loss": 1.8143,
      "step": 1523
    },
    {
      "epoch": 0.33520290333223357,
      "grad_norm": 0.2890012264251709,
      "learning_rate": 8e-05,
      "loss": 1.6603,
      "step": 1524
    },
    {
      "epoch": 0.3354228527438689,
      "grad_norm": 0.271720826625824,
      "learning_rate": 8e-05,
      "loss": 1.7186,
      "step": 1525
    },
    {
      "epoch": 0.33564280215550424,
      "grad_norm": 0.2845331132411957,
      "learning_rate": 8e-05,
      "loss": 1.7739,
      "step": 1526
    },
    {
      "epoch": 0.3358627515671396,
      "grad_norm": 0.2787776291370392,
      "learning_rate": 8e-05,
      "loss": 1.6146,
      "step": 1527
    },
    {
      "epoch": 0.3360827009787749,
      "grad_norm": 0.2612919211387634,
      "learning_rate": 8e-05,
      "loss": 1.5575,
      "step": 1528
    },
    {
      "epoch": 0.3363026503904102,
      "grad_norm": 0.279220849275589,
      "learning_rate": 8e-05,
      "loss": 1.7661,
      "step": 1529
    },
    {
      "epoch": 0.3365225998020455,
      "grad_norm": 0.2812168300151825,
      "learning_rate": 8e-05,
      "loss": 1.7011,
      "step": 1530
    },
    {
      "epoch": 0.33674254921368085,
      "grad_norm": 0.28216826915740967,
      "learning_rate": 8e-05,
      "loss": 1.7856,
      "step": 1531
    },
    {
      "epoch": 0.3369624986253162,
      "grad_norm": 0.279895156621933,
      "learning_rate": 8e-05,
      "loss": 1.6793,
      "step": 1532
    },
    {
      "epoch": 0.3371824480369515,
      "grad_norm": 0.2694056034088135,
      "learning_rate": 8e-05,
      "loss": 1.6289,
      "step": 1533
    },
    {
      "epoch": 0.3374023974485868,
      "grad_norm": 0.2692592740058899,
      "learning_rate": 8e-05,
      "loss": 1.5595,
      "step": 1534
    },
    {
      "epoch": 0.33762234686022213,
      "grad_norm": 0.32149383425712585,
      "learning_rate": 8e-05,
      "loss": 1.6667,
      "step": 1535
    },
    {
      "epoch": 0.33784229627185747,
      "grad_norm": 0.28884437680244446,
      "learning_rate": 8e-05,
      "loss": 1.7836,
      "step": 1536
    },
    {
      "epoch": 0.3380622456834928,
      "grad_norm": 0.276017963886261,
      "learning_rate": 8e-05,
      "loss": 1.712,
      "step": 1537
    },
    {
      "epoch": 0.33828219509512814,
      "grad_norm": 0.26901450753211975,
      "learning_rate": 8e-05,
      "loss": 1.6442,
      "step": 1538
    },
    {
      "epoch": 0.33850214450676347,
      "grad_norm": 0.29827412962913513,
      "learning_rate": 8e-05,
      "loss": 1.7619,
      "step": 1539
    },
    {
      "epoch": 0.33872209391839875,
      "grad_norm": 0.2763231098651886,
      "learning_rate": 8e-05,
      "loss": 1.6344,
      "step": 1540
    },
    {
      "epoch": 0.3389420433300341,
      "grad_norm": 0.26493677496910095,
      "learning_rate": 8e-05,
      "loss": 1.6964,
      "step": 1541
    },
    {
      "epoch": 0.3391619927416694,
      "grad_norm": 0.2956371605396271,
      "learning_rate": 8e-05,
      "loss": 1.7328,
      "step": 1542
    },
    {
      "epoch": 0.33938194215330475,
      "grad_norm": 0.2845339775085449,
      "learning_rate": 8e-05,
      "loss": 1.6477,
      "step": 1543
    },
    {
      "epoch": 0.3396018915649401,
      "grad_norm": 0.29501214623451233,
      "learning_rate": 8e-05,
      "loss": 1.7951,
      "step": 1544
    },
    {
      "epoch": 0.33982184097657536,
      "grad_norm": 0.2859644591808319,
      "learning_rate": 8e-05,
      "loss": 1.6607,
      "step": 1545
    },
    {
      "epoch": 0.3400417903882107,
      "grad_norm": 0.2733168303966522,
      "learning_rate": 8e-05,
      "loss": 1.6397,
      "step": 1546
    },
    {
      "epoch": 0.34026173979984603,
      "grad_norm": 0.2580598294734955,
      "learning_rate": 8e-05,
      "loss": 1.5692,
      "step": 1547
    },
    {
      "epoch": 0.34048168921148136,
      "grad_norm": 0.3042803406715393,
      "learning_rate": 8e-05,
      "loss": 1.7063,
      "step": 1548
    },
    {
      "epoch": 0.3407016386231167,
      "grad_norm": 0.2833859324455261,
      "learning_rate": 8e-05,
      "loss": 1.7531,
      "step": 1549
    },
    {
      "epoch": 0.34092158803475203,
      "grad_norm": 0.259620726108551,
      "learning_rate": 8e-05,
      "loss": 1.6179,
      "step": 1550
    },
    {
      "epoch": 0.3411415374463873,
      "grad_norm": 0.268355131149292,
      "learning_rate": 8e-05,
      "loss": 1.6009,
      "step": 1551
    },
    {
      "epoch": 0.34136148685802264,
      "grad_norm": 0.2858780324459076,
      "learning_rate": 8e-05,
      "loss": 1.7033,
      "step": 1552
    },
    {
      "epoch": 0.341581436269658,
      "grad_norm": 0.2777354121208191,
      "learning_rate": 8e-05,
      "loss": 1.7615,
      "step": 1553
    },
    {
      "epoch": 0.3418013856812933,
      "grad_norm": 0.27899524569511414,
      "learning_rate": 8e-05,
      "loss": 1.6684,
      "step": 1554
    },
    {
      "epoch": 0.34202133509292865,
      "grad_norm": 0.3156200349330902,
      "learning_rate": 8e-05,
      "loss": 1.5658,
      "step": 1555
    },
    {
      "epoch": 0.3422412845045639,
      "grad_norm": 0.27549582719802856,
      "learning_rate": 8e-05,
      "loss": 1.692,
      "step": 1556
    },
    {
      "epoch": 0.34246123391619926,
      "grad_norm": 0.27770310640335083,
      "learning_rate": 8e-05,
      "loss": 1.5891,
      "step": 1557
    },
    {
      "epoch": 0.3426811833278346,
      "grad_norm": 0.28138646483421326,
      "learning_rate": 8e-05,
      "loss": 1.5949,
      "step": 1558
    },
    {
      "epoch": 0.3429011327394699,
      "grad_norm": 0.2790684998035431,
      "learning_rate": 8e-05,
      "loss": 1.6371,
      "step": 1559
    },
    {
      "epoch": 0.34312108215110526,
      "grad_norm": 0.303230345249176,
      "learning_rate": 8e-05,
      "loss": 1.7416,
      "step": 1560
    },
    {
      "epoch": 0.3433410315627406,
      "grad_norm": 0.26891767978668213,
      "learning_rate": 8e-05,
      "loss": 1.8044,
      "step": 1561
    },
    {
      "epoch": 0.3435609809743759,
      "grad_norm": 0.2734631896018982,
      "learning_rate": 8e-05,
      "loss": 1.7171,
      "step": 1562
    },
    {
      "epoch": 0.3437809303860112,
      "grad_norm": 0.29556018114089966,
      "learning_rate": 8e-05,
      "loss": 1.9085,
      "step": 1563
    },
    {
      "epoch": 0.34400087979764654,
      "grad_norm": 0.26478004455566406,
      "learning_rate": 8e-05,
      "loss": 1.6153,
      "step": 1564
    },
    {
      "epoch": 0.3442208292092819,
      "grad_norm": 0.27655404806137085,
      "learning_rate": 8e-05,
      "loss": 1.7384,
      "step": 1565
    },
    {
      "epoch": 0.3444407786209172,
      "grad_norm": 0.2902698218822479,
      "learning_rate": 8e-05,
      "loss": 1.6589,
      "step": 1566
    },
    {
      "epoch": 0.3446607280325525,
      "grad_norm": 0.2857147455215454,
      "learning_rate": 8e-05,
      "loss": 1.6598,
      "step": 1567
    },
    {
      "epoch": 0.3448806774441878,
      "grad_norm": 0.28339943289756775,
      "learning_rate": 8e-05,
      "loss": 1.7356,
      "step": 1568
    },
    {
      "epoch": 0.34510062685582316,
      "grad_norm": 0.29340776801109314,
      "learning_rate": 8e-05,
      "loss": 1.8316,
      "step": 1569
    },
    {
      "epoch": 0.3453205762674585,
      "grad_norm": 0.26669397950172424,
      "learning_rate": 8e-05,
      "loss": 1.6803,
      "step": 1570
    },
    {
      "epoch": 0.3455405256790938,
      "grad_norm": 0.28508248925209045,
      "learning_rate": 8e-05,
      "loss": 1.7702,
      "step": 1571
    },
    {
      "epoch": 0.34576047509072916,
      "grad_norm": 0.25610047578811646,
      "learning_rate": 8e-05,
      "loss": 1.6343,
      "step": 1572
    },
    {
      "epoch": 0.34598042450236444,
      "grad_norm": 0.2758273482322693,
      "learning_rate": 8e-05,
      "loss": 1.7875,
      "step": 1573
    },
    {
      "epoch": 0.34620037391399977,
      "grad_norm": 0.2674688398838043,
      "learning_rate": 8e-05,
      "loss": 1.6804,
      "step": 1574
    },
    {
      "epoch": 0.3464203233256351,
      "grad_norm": 0.2796163558959961,
      "learning_rate": 8e-05,
      "loss": 1.6135,
      "step": 1575
    },
    {
      "epoch": 0.34664027273727044,
      "grad_norm": 0.26260775327682495,
      "learning_rate": 8e-05,
      "loss": 1.6752,
      "step": 1576
    },
    {
      "epoch": 0.3468602221489058,
      "grad_norm": 0.2897137403488159,
      "learning_rate": 8e-05,
      "loss": 1.6743,
      "step": 1577
    },
    {
      "epoch": 0.34708017156054105,
      "grad_norm": 0.27681732177734375,
      "learning_rate": 8e-05,
      "loss": 1.6436,
      "step": 1578
    },
    {
      "epoch": 0.3473001209721764,
      "grad_norm": 0.2694265842437744,
      "learning_rate": 8e-05,
      "loss": 1.6343,
      "step": 1579
    },
    {
      "epoch": 0.3475200703838117,
      "grad_norm": 0.28179508447647095,
      "learning_rate": 8e-05,
      "loss": 1.676,
      "step": 1580
    },
    {
      "epoch": 0.34774001979544705,
      "grad_norm": 0.29600057005882263,
      "learning_rate": 8e-05,
      "loss": 1.786,
      "step": 1581
    },
    {
      "epoch": 0.3479599692070824,
      "grad_norm": 0.28932616114616394,
      "learning_rate": 8e-05,
      "loss": 1.7151,
      "step": 1582
    },
    {
      "epoch": 0.3481799186187177,
      "grad_norm": 0.2912417948246002,
      "learning_rate": 8e-05,
      "loss": 1.7788,
      "step": 1583
    },
    {
      "epoch": 0.348399868030353,
      "grad_norm": 0.2844431698322296,
      "learning_rate": 8e-05,
      "loss": 1.5585,
      "step": 1584
    },
    {
      "epoch": 0.34861981744198833,
      "grad_norm": 0.2916630804538727,
      "learning_rate": 8e-05,
      "loss": 1.7484,
      "step": 1585
    },
    {
      "epoch": 0.34883976685362367,
      "grad_norm": 0.2785089612007141,
      "learning_rate": 8e-05,
      "loss": 1.61,
      "step": 1586
    },
    {
      "epoch": 0.349059716265259,
      "grad_norm": 0.2777422368526459,
      "learning_rate": 8e-05,
      "loss": 1.7183,
      "step": 1587
    },
    {
      "epoch": 0.34927966567689434,
      "grad_norm": 0.28772565722465515,
      "learning_rate": 8e-05,
      "loss": 1.7161,
      "step": 1588
    },
    {
      "epoch": 0.3494996150885296,
      "grad_norm": 0.28452831506729126,
      "learning_rate": 8e-05,
      "loss": 1.8004,
      "step": 1589
    },
    {
      "epoch": 0.34971956450016495,
      "grad_norm": 0.2837449014186859,
      "learning_rate": 8e-05,
      "loss": 1.7992,
      "step": 1590
    },
    {
      "epoch": 0.3499395139118003,
      "grad_norm": 0.2874920666217804,
      "learning_rate": 8e-05,
      "loss": 1.6408,
      "step": 1591
    },
    {
      "epoch": 0.3501594633234356,
      "grad_norm": 0.26615065336227417,
      "learning_rate": 8e-05,
      "loss": 1.658,
      "step": 1592
    },
    {
      "epoch": 0.35037941273507095,
      "grad_norm": 0.27493569254875183,
      "learning_rate": 8e-05,
      "loss": 1.6843,
      "step": 1593
    },
    {
      "epoch": 0.3505993621467063,
      "grad_norm": 0.291886568069458,
      "learning_rate": 8e-05,
      "loss": 1.7683,
      "step": 1594
    },
    {
      "epoch": 0.35081931155834156,
      "grad_norm": 0.2868814468383789,
      "learning_rate": 8e-05,
      "loss": 1.7825,
      "step": 1595
    },
    {
      "epoch": 0.3510392609699769,
      "grad_norm": 0.30988067388534546,
      "learning_rate": 8e-05,
      "loss": 1.7511,
      "step": 1596
    },
    {
      "epoch": 0.35125921038161223,
      "grad_norm": 0.2746553122997284,
      "learning_rate": 8e-05,
      "loss": 1.6298,
      "step": 1597
    },
    {
      "epoch": 0.35147915979324756,
      "grad_norm": 0.3013536036014557,
      "learning_rate": 8e-05,
      "loss": 1.7883,
      "step": 1598
    },
    {
      "epoch": 0.3516991092048829,
      "grad_norm": 0.2906748056411743,
      "learning_rate": 8e-05,
      "loss": 1.5819,
      "step": 1599
    },
    {
      "epoch": 0.3519190586165182,
      "grad_norm": 0.28082364797592163,
      "learning_rate": 8e-05,
      "loss": 1.7525,
      "step": 1600
    },
    {
      "epoch": 0.3521390080281535,
      "grad_norm": 0.28713324666023254,
      "learning_rate": 8e-05,
      "loss": 1.79,
      "step": 1601
    },
    {
      "epoch": 0.35235895743978884,
      "grad_norm": 0.2819896638393402,
      "learning_rate": 8e-05,
      "loss": 1.6514,
      "step": 1602
    },
    {
      "epoch": 0.3525789068514242,
      "grad_norm": 0.27669310569763184,
      "learning_rate": 8e-05,
      "loss": 1.5888,
      "step": 1603
    },
    {
      "epoch": 0.3527988562630595,
      "grad_norm": 0.2873641848564148,
      "learning_rate": 8e-05,
      "loss": 1.8206,
      "step": 1604
    },
    {
      "epoch": 0.35301880567469485,
      "grad_norm": 0.28426647186279297,
      "learning_rate": 8e-05,
      "loss": 1.7736,
      "step": 1605
    },
    {
      "epoch": 0.3532387550863301,
      "grad_norm": 0.2733590602874756,
      "learning_rate": 8e-05,
      "loss": 1.6653,
      "step": 1606
    },
    {
      "epoch": 0.35345870449796546,
      "grad_norm": 0.26751479506492615,
      "learning_rate": 8e-05,
      "loss": 1.5841,
      "step": 1607
    },
    {
      "epoch": 0.3536786539096008,
      "grad_norm": 0.2767663598060608,
      "learning_rate": 8e-05,
      "loss": 1.6859,
      "step": 1608
    },
    {
      "epoch": 0.3538986033212361,
      "grad_norm": 0.28359255194664,
      "learning_rate": 8e-05,
      "loss": 1.8799,
      "step": 1609
    },
    {
      "epoch": 0.35411855273287146,
      "grad_norm": 0.27551594376564026,
      "learning_rate": 8e-05,
      "loss": 1.6429,
      "step": 1610
    },
    {
      "epoch": 0.35433850214450674,
      "grad_norm": 0.26260972023010254,
      "learning_rate": 8e-05,
      "loss": 1.6068,
      "step": 1611
    },
    {
      "epoch": 0.3545584515561421,
      "grad_norm": 0.2778937518596649,
      "learning_rate": 8e-05,
      "loss": 1.8057,
      "step": 1612
    },
    {
      "epoch": 0.3547784009677774,
      "grad_norm": 0.27607765793800354,
      "learning_rate": 8e-05,
      "loss": 1.7439,
      "step": 1613
    },
    {
      "epoch": 0.35499835037941274,
      "grad_norm": 0.2628287076950073,
      "learning_rate": 8e-05,
      "loss": 1.6916,
      "step": 1614
    },
    {
      "epoch": 0.3552182997910481,
      "grad_norm": 0.2767592966556549,
      "learning_rate": 8e-05,
      "loss": 1.7185,
      "step": 1615
    },
    {
      "epoch": 0.3554382492026834,
      "grad_norm": 0.2666943669319153,
      "learning_rate": 8e-05,
      "loss": 1.7351,
      "step": 1616
    },
    {
      "epoch": 0.3556581986143187,
      "grad_norm": 0.28780093789100647,
      "learning_rate": 8e-05,
      "loss": 1.757,
      "step": 1617
    },
    {
      "epoch": 0.355878148025954,
      "grad_norm": 0.30761584639549255,
      "learning_rate": 8e-05,
      "loss": 1.8096,
      "step": 1618
    },
    {
      "epoch": 0.35609809743758936,
      "grad_norm": 0.2926090359687805,
      "learning_rate": 8e-05,
      "loss": 1.8609,
      "step": 1619
    },
    {
      "epoch": 0.3563180468492247,
      "grad_norm": 0.27546852827072144,
      "learning_rate": 8e-05,
      "loss": 1.6422,
      "step": 1620
    },
    {
      "epoch": 0.35653799626086,
      "grad_norm": 0.28559309244155884,
      "learning_rate": 8e-05,
      "loss": 1.8225,
      "step": 1621
    },
    {
      "epoch": 0.3567579456724953,
      "grad_norm": 0.2804494798183441,
      "learning_rate": 8e-05,
      "loss": 1.9108,
      "step": 1622
    },
    {
      "epoch": 0.35697789508413064,
      "grad_norm": 0.2643645703792572,
      "learning_rate": 8e-05,
      "loss": 1.5462,
      "step": 1623
    },
    {
      "epoch": 0.35719784449576597,
      "grad_norm": 0.2888531982898712,
      "learning_rate": 8e-05,
      "loss": 1.701,
      "step": 1624
    },
    {
      "epoch": 0.3574177939074013,
      "grad_norm": 0.28601035475730896,
      "learning_rate": 8e-05,
      "loss": 1.628,
      "step": 1625
    },
    {
      "epoch": 0.35763774331903664,
      "grad_norm": 0.2877524197101593,
      "learning_rate": 8e-05,
      "loss": 1.8403,
      "step": 1626
    },
    {
      "epoch": 0.357857692730672,
      "grad_norm": 0.2658945918083191,
      "learning_rate": 8e-05,
      "loss": 1.4552,
      "step": 1627
    },
    {
      "epoch": 0.35807764214230725,
      "grad_norm": 0.2911885976791382,
      "learning_rate": 8e-05,
      "loss": 1.7753,
      "step": 1628
    },
    {
      "epoch": 0.3582975915539426,
      "grad_norm": 0.29072439670562744,
      "learning_rate": 8e-05,
      "loss": 1.7229,
      "step": 1629
    },
    {
      "epoch": 0.3585175409655779,
      "grad_norm": 0.29961150884628296,
      "learning_rate": 8e-05,
      "loss": 1.7694,
      "step": 1630
    },
    {
      "epoch": 0.35873749037721325,
      "grad_norm": 0.2760653793811798,
      "learning_rate": 8e-05,
      "loss": 1.719,
      "step": 1631
    },
    {
      "epoch": 0.3589574397888486,
      "grad_norm": 0.2739832103252411,
      "learning_rate": 8e-05,
      "loss": 1.7367,
      "step": 1632
    },
    {
      "epoch": 0.35917738920048387,
      "grad_norm": 0.2669771611690521,
      "learning_rate": 8e-05,
      "loss": 1.5306,
      "step": 1633
    },
    {
      "epoch": 0.3593973386121192,
      "grad_norm": 0.2744583189487457,
      "learning_rate": 8e-05,
      "loss": 1.5713,
      "step": 1634
    },
    {
      "epoch": 0.35961728802375453,
      "grad_norm": 0.2943086326122284,
      "learning_rate": 8e-05,
      "loss": 1.6569,
      "step": 1635
    },
    {
      "epoch": 0.35983723743538987,
      "grad_norm": 0.2873243987560272,
      "learning_rate": 8e-05,
      "loss": 1.6864,
      "step": 1636
    },
    {
      "epoch": 0.3600571868470252,
      "grad_norm": 0.27217867970466614,
      "learning_rate": 8e-05,
      "loss": 1.7519,
      "step": 1637
    },
    {
      "epoch": 0.36027713625866054,
      "grad_norm": 0.28656938672065735,
      "learning_rate": 8e-05,
      "loss": 1.7892,
      "step": 1638
    },
    {
      "epoch": 0.3604970856702958,
      "grad_norm": 0.2876884937286377,
      "learning_rate": 8e-05,
      "loss": 1.6709,
      "step": 1639
    },
    {
      "epoch": 0.36071703508193115,
      "grad_norm": 0.2873481512069702,
      "learning_rate": 8e-05,
      "loss": 1.8336,
      "step": 1640
    },
    {
      "epoch": 0.3609369844935665,
      "grad_norm": 0.28285419940948486,
      "learning_rate": 8e-05,
      "loss": 1.5887,
      "step": 1641
    },
    {
      "epoch": 0.3611569339052018,
      "grad_norm": 0.2624582052230835,
      "learning_rate": 8e-05,
      "loss": 1.6248,
      "step": 1642
    },
    {
      "epoch": 0.36137688331683715,
      "grad_norm": 0.2794424891471863,
      "learning_rate": 8e-05,
      "loss": 1.7191,
      "step": 1643
    },
    {
      "epoch": 0.36159683272847243,
      "grad_norm": 0.2890479862689972,
      "learning_rate": 8e-05,
      "loss": 1.905,
      "step": 1644
    },
    {
      "epoch": 0.36181678214010776,
      "grad_norm": 0.28444570302963257,
      "learning_rate": 8e-05,
      "loss": 1.6948,
      "step": 1645
    },
    {
      "epoch": 0.3620367315517431,
      "grad_norm": 0.27037203311920166,
      "learning_rate": 8e-05,
      "loss": 1.6245,
      "step": 1646
    },
    {
      "epoch": 0.36225668096337843,
      "grad_norm": 0.2864437699317932,
      "learning_rate": 8e-05,
      "loss": 1.688,
      "step": 1647
    },
    {
      "epoch": 0.36247663037501376,
      "grad_norm": 0.27912065386772156,
      "learning_rate": 8e-05,
      "loss": 1.6056,
      "step": 1648
    },
    {
      "epoch": 0.3626965797866491,
      "grad_norm": 0.26467230916023254,
      "learning_rate": 8e-05,
      "loss": 1.5786,
      "step": 1649
    },
    {
      "epoch": 0.3629165291982844,
      "grad_norm": 0.2793690264225006,
      "learning_rate": 8e-05,
      "loss": 1.6003,
      "step": 1650
    },
    {
      "epoch": 0.3631364786099197,
      "grad_norm": 0.288629949092865,
      "learning_rate": 8e-05,
      "loss": 1.6752,
      "step": 1651
    },
    {
      "epoch": 0.36335642802155504,
      "grad_norm": 0.283195823431015,
      "learning_rate": 8e-05,
      "loss": 1.6854,
      "step": 1652
    },
    {
      "epoch": 0.3635763774331904,
      "grad_norm": 0.2929665446281433,
      "learning_rate": 8e-05,
      "loss": 1.7191,
      "step": 1653
    },
    {
      "epoch": 0.3637963268448257,
      "grad_norm": 0.28676289319992065,
      "learning_rate": 8e-05,
      "loss": 1.6959,
      "step": 1654
    },
    {
      "epoch": 0.364016276256461,
      "grad_norm": 0.264635294675827,
      "learning_rate": 8e-05,
      "loss": 1.6232,
      "step": 1655
    },
    {
      "epoch": 0.3642362256680963,
      "grad_norm": 0.2763380706310272,
      "learning_rate": 8e-05,
      "loss": 1.7631,
      "step": 1656
    },
    {
      "epoch": 0.36445617507973166,
      "grad_norm": 0.2624233365058899,
      "learning_rate": 8e-05,
      "loss": 1.6635,
      "step": 1657
    },
    {
      "epoch": 0.364676124491367,
      "grad_norm": 0.2564058303833008,
      "learning_rate": 8e-05,
      "loss": 1.4745,
      "step": 1658
    },
    {
      "epoch": 0.3648960739030023,
      "grad_norm": 0.2966236174106598,
      "learning_rate": 8e-05,
      "loss": 1.6892,
      "step": 1659
    },
    {
      "epoch": 0.36511602331463766,
      "grad_norm": 0.30588555335998535,
      "learning_rate": 8e-05,
      "loss": 1.6884,
      "step": 1660
    },
    {
      "epoch": 0.36533597272627294,
      "grad_norm": 0.2692076861858368,
      "learning_rate": 8e-05,
      "loss": 1.7158,
      "step": 1661
    },
    {
      "epoch": 0.3655559221379083,
      "grad_norm": 0.29388558864593506,
      "learning_rate": 8e-05,
      "loss": 1.7133,
      "step": 1662
    },
    {
      "epoch": 0.3657758715495436,
      "grad_norm": 0.28685635328292847,
      "learning_rate": 8e-05,
      "loss": 1.7444,
      "step": 1663
    },
    {
      "epoch": 0.36599582096117894,
      "grad_norm": 0.2885795831680298,
      "learning_rate": 8e-05,
      "loss": 1.7537,
      "step": 1664
    },
    {
      "epoch": 0.3662157703728143,
      "grad_norm": 0.3066631853580475,
      "learning_rate": 8e-05,
      "loss": 1.7843,
      "step": 1665
    },
    {
      "epoch": 0.36643571978444955,
      "grad_norm": 0.31112298369407654,
      "learning_rate": 8e-05,
      "loss": 1.4934,
      "step": 1666
    },
    {
      "epoch": 0.3666556691960849,
      "grad_norm": 0.2751656472682953,
      "learning_rate": 8e-05,
      "loss": 1.7463,
      "step": 1667
    },
    {
      "epoch": 0.3668756186077202,
      "grad_norm": 0.2834889590740204,
      "learning_rate": 8e-05,
      "loss": 1.7757,
      "step": 1668
    },
    {
      "epoch": 0.36709556801935556,
      "grad_norm": 0.2778145968914032,
      "learning_rate": 8e-05,
      "loss": 1.7423,
      "step": 1669
    },
    {
      "epoch": 0.3673155174309909,
      "grad_norm": 0.32161521911621094,
      "learning_rate": 8e-05,
      "loss": 1.8312,
      "step": 1670
    },
    {
      "epoch": 0.36753546684262617,
      "grad_norm": 0.27995115518569946,
      "learning_rate": 8e-05,
      "loss": 1.7694,
      "step": 1671
    },
    {
      "epoch": 0.3677554162542615,
      "grad_norm": 0.27701541781425476,
      "learning_rate": 8e-05,
      "loss": 1.8054,
      "step": 1672
    },
    {
      "epoch": 0.36797536566589684,
      "grad_norm": 0.2757355570793152,
      "learning_rate": 8e-05,
      "loss": 1.7036,
      "step": 1673
    },
    {
      "epoch": 0.36819531507753217,
      "grad_norm": 0.27305907011032104,
      "learning_rate": 8e-05,
      "loss": 1.6627,
      "step": 1674
    },
    {
      "epoch": 0.3684152644891675,
      "grad_norm": 0.299679696559906,
      "learning_rate": 8e-05,
      "loss": 1.7552,
      "step": 1675
    },
    {
      "epoch": 0.36863521390080284,
      "grad_norm": 0.2728777825832367,
      "learning_rate": 8e-05,
      "loss": 1.7649,
      "step": 1676
    },
    {
      "epoch": 0.3688551633124381,
      "grad_norm": 0.26330089569091797,
      "learning_rate": 8e-05,
      "loss": 1.5887,
      "step": 1677
    },
    {
      "epoch": 0.36907511272407345,
      "grad_norm": 0.2850317060947418,
      "learning_rate": 8e-05,
      "loss": 1.6255,
      "step": 1678
    },
    {
      "epoch": 0.3692950621357088,
      "grad_norm": 0.2784862220287323,
      "learning_rate": 8e-05,
      "loss": 1.7123,
      "step": 1679
    },
    {
      "epoch": 0.3695150115473441,
      "grad_norm": 0.284298300743103,
      "learning_rate": 8e-05,
      "loss": 1.5809,
      "step": 1680
    },
    {
      "epoch": 0.36973496095897945,
      "grad_norm": 0.2725334167480469,
      "learning_rate": 8e-05,
      "loss": 1.7037,
      "step": 1681
    },
    {
      "epoch": 0.36995491037061473,
      "grad_norm": 0.2760758399963379,
      "learning_rate": 8e-05,
      "loss": 1.6827,
      "step": 1682
    },
    {
      "epoch": 0.37017485978225007,
      "grad_norm": 0.2661541700363159,
      "learning_rate": 8e-05,
      "loss": 1.7042,
      "step": 1683
    },
    {
      "epoch": 0.3703948091938854,
      "grad_norm": 0.27737516164779663,
      "learning_rate": 8e-05,
      "loss": 1.7689,
      "step": 1684
    },
    {
      "epoch": 0.37061475860552073,
      "grad_norm": 0.2607424259185791,
      "learning_rate": 8e-05,
      "loss": 1.6356,
      "step": 1685
    },
    {
      "epoch": 0.37083470801715607,
      "grad_norm": 0.2802969217300415,
      "learning_rate": 8e-05,
      "loss": 1.7004,
      "step": 1686
    },
    {
      "epoch": 0.3710546574287914,
      "grad_norm": 0.2660817801952362,
      "learning_rate": 8e-05,
      "loss": 1.5539,
      "step": 1687
    },
    {
      "epoch": 0.3712746068404267,
      "grad_norm": 0.27867192029953003,
      "learning_rate": 8e-05,
      "loss": 1.6531,
      "step": 1688
    },
    {
      "epoch": 0.371494556252062,
      "grad_norm": 0.27857083082199097,
      "learning_rate": 8e-05,
      "loss": 1.8023,
      "step": 1689
    },
    {
      "epoch": 0.37171450566369735,
      "grad_norm": 0.2689161002635956,
      "learning_rate": 8e-05,
      "loss": 1.7601,
      "step": 1690
    },
    {
      "epoch": 0.3719344550753327,
      "grad_norm": 0.297826886177063,
      "learning_rate": 8e-05,
      "loss": 1.7627,
      "step": 1691
    },
    {
      "epoch": 0.372154404486968,
      "grad_norm": 0.2592705190181732,
      "learning_rate": 8e-05,
      "loss": 1.7132,
      "step": 1692
    },
    {
      "epoch": 0.3723743538986033,
      "grad_norm": 0.28288522362709045,
      "learning_rate": 8e-05,
      "loss": 1.7604,
      "step": 1693
    },
    {
      "epoch": 0.37259430331023863,
      "grad_norm": 0.30823859572410583,
      "learning_rate": 8e-05,
      "loss": 1.8563,
      "step": 1694
    },
    {
      "epoch": 0.37281425272187396,
      "grad_norm": 0.27835527062416077,
      "learning_rate": 8e-05,
      "loss": 1.6816,
      "step": 1695
    },
    {
      "epoch": 0.3730342021335093,
      "grad_norm": 0.2626672089099884,
      "learning_rate": 8e-05,
      "loss": 1.6185,
      "step": 1696
    },
    {
      "epoch": 0.37325415154514463,
      "grad_norm": 0.2489227056503296,
      "learning_rate": 8e-05,
      "loss": 1.6119,
      "step": 1697
    },
    {
      "epoch": 0.37347410095677996,
      "grad_norm": 0.28637897968292236,
      "learning_rate": 8e-05,
      "loss": 1.6695,
      "step": 1698
    },
    {
      "epoch": 0.37369405036841524,
      "grad_norm": 0.27077022194862366,
      "learning_rate": 8e-05,
      "loss": 1.6095,
      "step": 1699
    },
    {
      "epoch": 0.3739139997800506,
      "grad_norm": 0.32049357891082764,
      "learning_rate": 8e-05,
      "loss": 1.875,
      "step": 1700
    },
    {
      "epoch": 0.3741339491916859,
      "grad_norm": 0.2890382707118988,
      "learning_rate": 8e-05,
      "loss": 1.7129,
      "step": 1701
    },
    {
      "epoch": 0.37435389860332124,
      "grad_norm": 0.2785224914550781,
      "learning_rate": 8e-05,
      "loss": 1.7162,
      "step": 1702
    },
    {
      "epoch": 0.3745738480149566,
      "grad_norm": 0.2685299217700958,
      "learning_rate": 8e-05,
      "loss": 1.7358,
      "step": 1703
    },
    {
      "epoch": 0.37479379742659186,
      "grad_norm": 0.2840120494365692,
      "learning_rate": 8e-05,
      "loss": 1.9123,
      "step": 1704
    },
    {
      "epoch": 0.3750137468382272,
      "grad_norm": 0.27426856756210327,
      "learning_rate": 8e-05,
      "loss": 1.7144,
      "step": 1705
    },
    {
      "epoch": 0.3752336962498625,
      "grad_norm": 0.2707318663597107,
      "learning_rate": 8e-05,
      "loss": 1.6961,
      "step": 1706
    },
    {
      "epoch": 0.37545364566149786,
      "grad_norm": 0.3059745728969574,
      "learning_rate": 8e-05,
      "loss": 1.7491,
      "step": 1707
    },
    {
      "epoch": 0.3756735950731332,
      "grad_norm": 0.27109962701797485,
      "learning_rate": 8e-05,
      "loss": 1.6515,
      "step": 1708
    },
    {
      "epoch": 0.3758935444847685,
      "grad_norm": 0.26874709129333496,
      "learning_rate": 8e-05,
      "loss": 1.7119,
      "step": 1709
    },
    {
      "epoch": 0.3761134938964038,
      "grad_norm": 0.27959340810775757,
      "learning_rate": 8e-05,
      "loss": 1.5449,
      "step": 1710
    },
    {
      "epoch": 0.37633344330803914,
      "grad_norm": 0.284386545419693,
      "learning_rate": 8e-05,
      "loss": 1.8336,
      "step": 1711
    },
    {
      "epoch": 0.3765533927196745,
      "grad_norm": 0.27861231565475464,
      "learning_rate": 8e-05,
      "loss": 1.7547,
      "step": 1712
    },
    {
      "epoch": 0.3767733421313098,
      "grad_norm": 0.26845625042915344,
      "learning_rate": 8e-05,
      "loss": 1.6838,
      "step": 1713
    },
    {
      "epoch": 0.37699329154294514,
      "grad_norm": 0.31240981817245483,
      "learning_rate": 8e-05,
      "loss": 1.7489,
      "step": 1714
    },
    {
      "epoch": 0.3772132409545804,
      "grad_norm": 0.2878013253211975,
      "learning_rate": 8e-05,
      "loss": 1.7533,
      "step": 1715
    },
    {
      "epoch": 0.37743319036621575,
      "grad_norm": 0.27676892280578613,
      "learning_rate": 8e-05,
      "loss": 1.6218,
      "step": 1716
    },
    {
      "epoch": 0.3776531397778511,
      "grad_norm": 0.2782065272331238,
      "learning_rate": 8e-05,
      "loss": 1.6311,
      "step": 1717
    },
    {
      "epoch": 0.3778730891894864,
      "grad_norm": 0.2829797863960266,
      "learning_rate": 8e-05,
      "loss": 1.5863,
      "step": 1718
    },
    {
      "epoch": 0.37809303860112176,
      "grad_norm": 0.2851261794567108,
      "learning_rate": 8e-05,
      "loss": 1.8365,
      "step": 1719
    },
    {
      "epoch": 0.3783129880127571,
      "grad_norm": 0.2844488322734833,
      "learning_rate": 8e-05,
      "loss": 1.7765,
      "step": 1720
    },
    {
      "epoch": 0.37853293742439237,
      "grad_norm": 0.2976120412349701,
      "learning_rate": 8e-05,
      "loss": 1.7334,
      "step": 1721
    },
    {
      "epoch": 0.3787528868360277,
      "grad_norm": 0.27947840094566345,
      "learning_rate": 8e-05,
      "loss": 1.6641,
      "step": 1722
    },
    {
      "epoch": 0.37897283624766304,
      "grad_norm": 0.2986278831958771,
      "learning_rate": 8e-05,
      "loss": 1.8201,
      "step": 1723
    },
    {
      "epoch": 0.37919278565929837,
      "grad_norm": 0.26200374960899353,
      "learning_rate": 8e-05,
      "loss": 1.5835,
      "step": 1724
    },
    {
      "epoch": 0.3794127350709337,
      "grad_norm": 0.2846388816833496,
      "learning_rate": 8e-05,
      "loss": 1.7863,
      "step": 1725
    },
    {
      "epoch": 0.379632684482569,
      "grad_norm": 0.2809320390224457,
      "learning_rate": 8e-05,
      "loss": 1.6667,
      "step": 1726
    },
    {
      "epoch": 0.3798526338942043,
      "grad_norm": 0.28523099422454834,
      "learning_rate": 8e-05,
      "loss": 1.6647,
      "step": 1727
    },
    {
      "epoch": 0.38007258330583965,
      "grad_norm": 0.2719436287879944,
      "learning_rate": 8e-05,
      "loss": 1.6,
      "step": 1728
    },
    {
      "epoch": 0.380292532717475,
      "grad_norm": 0.2762429118156433,
      "learning_rate": 8e-05,
      "loss": 1.6888,
      "step": 1729
    },
    {
      "epoch": 0.3805124821291103,
      "grad_norm": 0.30161863565444946,
      "learning_rate": 8e-05,
      "loss": 1.6659,
      "step": 1730
    },
    {
      "epoch": 0.38073243154074565,
      "grad_norm": 0.27962687611579895,
      "learning_rate": 8e-05,
      "loss": 1.629,
      "step": 1731
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.27580323815345764,
      "learning_rate": 8e-05,
      "loss": 1.689,
      "step": 1732
    },
    {
      "epoch": 0.38117233036401627,
      "grad_norm": 0.2676113545894623,
      "learning_rate": 8e-05,
      "loss": 1.7195,
      "step": 1733
    },
    {
      "epoch": 0.3813922797756516,
      "grad_norm": 0.27840152382850647,
      "learning_rate": 8e-05,
      "loss": 1.6433,
      "step": 1734
    },
    {
      "epoch": 0.38161222918728693,
      "grad_norm": 0.27100005745887756,
      "learning_rate": 8e-05,
      "loss": 1.6517,
      "step": 1735
    },
    {
      "epoch": 0.38183217859892227,
      "grad_norm": 0.2874828577041626,
      "learning_rate": 8e-05,
      "loss": 1.9139,
      "step": 1736
    },
    {
      "epoch": 0.38205212801055755,
      "grad_norm": 0.2685931324958801,
      "learning_rate": 8e-05,
      "loss": 1.7373,
      "step": 1737
    },
    {
      "epoch": 0.3822720774221929,
      "grad_norm": 0.2895548641681671,
      "learning_rate": 8e-05,
      "loss": 1.7828,
      "step": 1738
    },
    {
      "epoch": 0.3824920268338282,
      "grad_norm": 0.29109206795692444,
      "learning_rate": 8e-05,
      "loss": 1.6347,
      "step": 1739
    },
    {
      "epoch": 0.38271197624546355,
      "grad_norm": 0.2804923951625824,
      "learning_rate": 8e-05,
      "loss": 1.5978,
      "step": 1740
    },
    {
      "epoch": 0.3829319256570989,
      "grad_norm": 0.2829732894897461,
      "learning_rate": 8e-05,
      "loss": 1.6271,
      "step": 1741
    },
    {
      "epoch": 0.3831518750687342,
      "grad_norm": 0.28979840874671936,
      "learning_rate": 8e-05,
      "loss": 1.7244,
      "step": 1742
    },
    {
      "epoch": 0.3833718244803695,
      "grad_norm": 0.30159792304039,
      "learning_rate": 8e-05,
      "loss": 1.8074,
      "step": 1743
    },
    {
      "epoch": 0.38359177389200483,
      "grad_norm": 0.28228580951690674,
      "learning_rate": 8e-05,
      "loss": 1.6669,
      "step": 1744
    },
    {
      "epoch": 0.38381172330364016,
      "grad_norm": 0.27950945496559143,
      "learning_rate": 8e-05,
      "loss": 1.6583,
      "step": 1745
    },
    {
      "epoch": 0.3840316727152755,
      "grad_norm": 0.2708896994590759,
      "learning_rate": 8e-05,
      "loss": 1.5793,
      "step": 1746
    },
    {
      "epoch": 0.38425162212691083,
      "grad_norm": 0.27368029952049255,
      "learning_rate": 8e-05,
      "loss": 1.6371,
      "step": 1747
    },
    {
      "epoch": 0.3844715715385461,
      "grad_norm": 0.27621379494667053,
      "learning_rate": 8e-05,
      "loss": 1.5737,
      "step": 1748
    },
    {
      "epoch": 0.38469152095018144,
      "grad_norm": 0.27143922448158264,
      "learning_rate": 8e-05,
      "loss": 1.7289,
      "step": 1749
    },
    {
      "epoch": 0.3849114703618168,
      "grad_norm": 0.28887274861335754,
      "learning_rate": 8e-05,
      "loss": 1.7262,
      "step": 1750
    },
    {
      "epoch": 0.3851314197734521,
      "grad_norm": 0.26516541838645935,
      "learning_rate": 8e-05,
      "loss": 1.6358,
      "step": 1751
    },
    {
      "epoch": 0.38535136918508744,
      "grad_norm": 0.31475701928138733,
      "learning_rate": 8e-05,
      "loss": 1.8599,
      "step": 1752
    },
    {
      "epoch": 0.3855713185967228,
      "grad_norm": 0.27711552381515503,
      "learning_rate": 8e-05,
      "loss": 1.632,
      "step": 1753
    },
    {
      "epoch": 0.38579126800835806,
      "grad_norm": 0.27542901039123535,
      "learning_rate": 8e-05,
      "loss": 1.6741,
      "step": 1754
    },
    {
      "epoch": 0.3860112174199934,
      "grad_norm": 0.2941054701805115,
      "learning_rate": 8e-05,
      "loss": 1.6396,
      "step": 1755
    },
    {
      "epoch": 0.3862311668316287,
      "grad_norm": 0.27836698293685913,
      "learning_rate": 8e-05,
      "loss": 1.5689,
      "step": 1756
    },
    {
      "epoch": 0.38645111624326406,
      "grad_norm": 0.29147645831108093,
      "learning_rate": 8e-05,
      "loss": 1.7523,
      "step": 1757
    },
    {
      "epoch": 0.3866710656548994,
      "grad_norm": 0.30084285140037537,
      "learning_rate": 8e-05,
      "loss": 1.6932,
      "step": 1758
    },
    {
      "epoch": 0.38689101506653467,
      "grad_norm": 0.2850727140903473,
      "learning_rate": 8e-05,
      "loss": 1.7276,
      "step": 1759
    },
    {
      "epoch": 0.38711096447817,
      "grad_norm": 0.27011391520500183,
      "learning_rate": 8e-05,
      "loss": 1.6635,
      "step": 1760
    },
    {
      "epoch": 0.38733091388980534,
      "grad_norm": 0.28682348132133484,
      "learning_rate": 8e-05,
      "loss": 1.7642,
      "step": 1761
    },
    {
      "epoch": 0.3875508633014407,
      "grad_norm": 0.27676117420196533,
      "learning_rate": 8e-05,
      "loss": 1.7406,
      "step": 1762
    },
    {
      "epoch": 0.387770812713076,
      "grad_norm": 0.2654523551464081,
      "learning_rate": 8e-05,
      "loss": 1.6484,
      "step": 1763
    },
    {
      "epoch": 0.38799076212471134,
      "grad_norm": 0.28026026487350464,
      "learning_rate": 8e-05,
      "loss": 1.6714,
      "step": 1764
    },
    {
      "epoch": 0.3882107115363466,
      "grad_norm": 0.3003789782524109,
      "learning_rate": 8e-05,
      "loss": 1.8121,
      "step": 1765
    },
    {
      "epoch": 0.38843066094798195,
      "grad_norm": 0.35523107647895813,
      "learning_rate": 8e-05,
      "loss": 1.9299,
      "step": 1766
    },
    {
      "epoch": 0.3886506103596173,
      "grad_norm": 0.26844245195388794,
      "learning_rate": 8e-05,
      "loss": 1.6358,
      "step": 1767
    },
    {
      "epoch": 0.3888705597712526,
      "grad_norm": 0.27308356761932373,
      "learning_rate": 8e-05,
      "loss": 1.6104,
      "step": 1768
    },
    {
      "epoch": 0.38909050918288796,
      "grad_norm": 0.2775373160839081,
      "learning_rate": 8e-05,
      "loss": 1.5679,
      "step": 1769
    },
    {
      "epoch": 0.38931045859452323,
      "grad_norm": 0.29753705859184265,
      "learning_rate": 8e-05,
      "loss": 1.7678,
      "step": 1770
    },
    {
      "epoch": 0.38953040800615857,
      "grad_norm": 0.2798722982406616,
      "learning_rate": 8e-05,
      "loss": 1.7034,
      "step": 1771
    },
    {
      "epoch": 0.3897503574177939,
      "grad_norm": 0.2842818796634674,
      "learning_rate": 8e-05,
      "loss": 1.727,
      "step": 1772
    },
    {
      "epoch": 0.38997030682942924,
      "grad_norm": 0.27555832266807556,
      "learning_rate": 8e-05,
      "loss": 1.7438,
      "step": 1773
    },
    {
      "epoch": 0.39019025624106457,
      "grad_norm": 0.2824547588825226,
      "learning_rate": 8e-05,
      "loss": 1.5733,
      "step": 1774
    },
    {
      "epoch": 0.3904102056526999,
      "grad_norm": 0.2658035159111023,
      "learning_rate": 8e-05,
      "loss": 1.5997,
      "step": 1775
    },
    {
      "epoch": 0.3906301550643352,
      "grad_norm": 0.27601394057273865,
      "learning_rate": 8e-05,
      "loss": 1.7025,
      "step": 1776
    },
    {
      "epoch": 0.3908501044759705,
      "grad_norm": 0.2990022897720337,
      "learning_rate": 8e-05,
      "loss": 1.8278,
      "step": 1777
    },
    {
      "epoch": 0.39107005388760585,
      "grad_norm": 0.29378873109817505,
      "learning_rate": 8e-05,
      "loss": 1.7549,
      "step": 1778
    },
    {
      "epoch": 0.3912900032992412,
      "grad_norm": 0.29202136397361755,
      "learning_rate": 8e-05,
      "loss": 1.6602,
      "step": 1779
    },
    {
      "epoch": 0.3915099527108765,
      "grad_norm": 0.28191903233528137,
      "learning_rate": 8e-05,
      "loss": 1.6618,
      "step": 1780
    },
    {
      "epoch": 0.3917299021225118,
      "grad_norm": 0.26916682720184326,
      "learning_rate": 8e-05,
      "loss": 1.6523,
      "step": 1781
    },
    {
      "epoch": 0.39194985153414713,
      "grad_norm": 0.2886850833892822,
      "learning_rate": 8e-05,
      "loss": 1.5752,
      "step": 1782
    },
    {
      "epoch": 0.39216980094578247,
      "grad_norm": 0.2749246656894684,
      "learning_rate": 8e-05,
      "loss": 1.6731,
      "step": 1783
    },
    {
      "epoch": 0.3923897503574178,
      "grad_norm": 0.28945374488830566,
      "learning_rate": 8e-05,
      "loss": 1.7404,
      "step": 1784
    },
    {
      "epoch": 0.39260969976905313,
      "grad_norm": 0.27297836542129517,
      "learning_rate": 8e-05,
      "loss": 1.7176,
      "step": 1785
    },
    {
      "epoch": 0.39282964918068847,
      "grad_norm": 0.2738782465457916,
      "learning_rate": 8e-05,
      "loss": 1.6844,
      "step": 1786
    },
    {
      "epoch": 0.39304959859232375,
      "grad_norm": 0.2897050082683563,
      "learning_rate": 8e-05,
      "loss": 1.6522,
      "step": 1787
    },
    {
      "epoch": 0.3932695480039591,
      "grad_norm": 0.31031668186187744,
      "learning_rate": 8e-05,
      "loss": 1.8216,
      "step": 1788
    },
    {
      "epoch": 0.3934894974155944,
      "grad_norm": 0.2869516909122467,
      "learning_rate": 8e-05,
      "loss": 1.7598,
      "step": 1789
    },
    {
      "epoch": 0.39370944682722975,
      "grad_norm": 0.3080596625804901,
      "learning_rate": 8e-05,
      "loss": 1.7645,
      "step": 1790
    },
    {
      "epoch": 0.3939293962388651,
      "grad_norm": 0.27992716431617737,
      "learning_rate": 8e-05,
      "loss": 1.8373,
      "step": 1791
    },
    {
      "epoch": 0.39414934565050036,
      "grad_norm": 0.2761777341365814,
      "learning_rate": 8e-05,
      "loss": 1.6767,
      "step": 1792
    },
    {
      "epoch": 0.3943692950621357,
      "grad_norm": 0.30193084478378296,
      "learning_rate": 8e-05,
      "loss": 1.7394,
      "step": 1793
    },
    {
      "epoch": 0.39458924447377103,
      "grad_norm": 0.29375529289245605,
      "learning_rate": 8e-05,
      "loss": 1.761,
      "step": 1794
    },
    {
      "epoch": 0.39480919388540636,
      "grad_norm": 0.32190364599227905,
      "learning_rate": 8e-05,
      "loss": 1.7045,
      "step": 1795
    },
    {
      "epoch": 0.3950291432970417,
      "grad_norm": 0.27505311369895935,
      "learning_rate": 8e-05,
      "loss": 1.6678,
      "step": 1796
    },
    {
      "epoch": 0.39524909270867703,
      "grad_norm": 0.28678107261657715,
      "learning_rate": 8e-05,
      "loss": 1.7155,
      "step": 1797
    },
    {
      "epoch": 0.3954690421203123,
      "grad_norm": 0.28372088074684143,
      "learning_rate": 8e-05,
      "loss": 1.8833,
      "step": 1798
    },
    {
      "epoch": 0.39568899153194764,
      "grad_norm": 0.27803388237953186,
      "learning_rate": 8e-05,
      "loss": 1.7125,
      "step": 1799
    },
    {
      "epoch": 0.395908940943583,
      "grad_norm": 0.278728187084198,
      "learning_rate": 8e-05,
      "loss": 1.7019,
      "step": 1800
    },
    {
      "epoch": 0.3961288903552183,
      "grad_norm": 0.29563671350479126,
      "learning_rate": 8e-05,
      "loss": 1.6588,
      "step": 1801
    },
    {
      "epoch": 0.39634883976685364,
      "grad_norm": 0.35585105419158936,
      "learning_rate": 8e-05,
      "loss": 1.8864,
      "step": 1802
    },
    {
      "epoch": 0.3965687891784889,
      "grad_norm": 0.27399691939353943,
      "learning_rate": 8e-05,
      "loss": 1.6222,
      "step": 1803
    },
    {
      "epoch": 0.39678873859012426,
      "grad_norm": 0.2557234764099121,
      "learning_rate": 8e-05,
      "loss": 1.4835,
      "step": 1804
    },
    {
      "epoch": 0.3970086880017596,
      "grad_norm": 0.2929818332195282,
      "learning_rate": 8e-05,
      "loss": 1.7737,
      "step": 1805
    },
    {
      "epoch": 0.3972286374133949,
      "grad_norm": 0.279729425907135,
      "learning_rate": 8e-05,
      "loss": 1.6073,
      "step": 1806
    },
    {
      "epoch": 0.39744858682503026,
      "grad_norm": 0.2622847259044647,
      "learning_rate": 8e-05,
      "loss": 1.5368,
      "step": 1807
    },
    {
      "epoch": 0.3976685362366656,
      "grad_norm": 0.2704038619995117,
      "learning_rate": 8e-05,
      "loss": 1.5809,
      "step": 1808
    },
    {
      "epoch": 0.39788848564830087,
      "grad_norm": 0.2785516679286957,
      "learning_rate": 8e-05,
      "loss": 1.6416,
      "step": 1809
    },
    {
      "epoch": 0.3981084350599362,
      "grad_norm": 0.29611852765083313,
      "learning_rate": 8e-05,
      "loss": 1.7186,
      "step": 1810
    },
    {
      "epoch": 0.39832838447157154,
      "grad_norm": 0.28127896785736084,
      "learning_rate": 8e-05,
      "loss": 1.6248,
      "step": 1811
    },
    {
      "epoch": 0.3985483338832069,
      "grad_norm": 0.2746615707874298,
      "learning_rate": 8e-05,
      "loss": 1.7029,
      "step": 1812
    },
    {
      "epoch": 0.3987682832948422,
      "grad_norm": 0.2650880515575409,
      "learning_rate": 8e-05,
      "loss": 1.7828,
      "step": 1813
    },
    {
      "epoch": 0.3989882327064775,
      "grad_norm": 0.28278401494026184,
      "learning_rate": 8e-05,
      "loss": 1.6383,
      "step": 1814
    },
    {
      "epoch": 0.3992081821181128,
      "grad_norm": 0.2749755382537842,
      "learning_rate": 8e-05,
      "loss": 1.7551,
      "step": 1815
    },
    {
      "epoch": 0.39942813152974815,
      "grad_norm": 0.26788878440856934,
      "learning_rate": 8e-05,
      "loss": 1.5206,
      "step": 1816
    },
    {
      "epoch": 0.3996480809413835,
      "grad_norm": 0.28166842460632324,
      "learning_rate": 8e-05,
      "loss": 1.7134,
      "step": 1817
    },
    {
      "epoch": 0.3998680303530188,
      "grad_norm": 0.2781674563884735,
      "learning_rate": 8e-05,
      "loss": 1.7864,
      "step": 1818
    },
    {
      "epoch": 0.40008797976465416,
      "grad_norm": 0.2810186445713043,
      "learning_rate": 8e-05,
      "loss": 1.7017,
      "step": 1819
    },
    {
      "epoch": 0.40030792917628943,
      "grad_norm": 0.2872167229652405,
      "learning_rate": 8e-05,
      "loss": 1.563,
      "step": 1820
    },
    {
      "epoch": 0.40052787858792477,
      "grad_norm": 0.2864447832107544,
      "learning_rate": 8e-05,
      "loss": 1.8261,
      "step": 1821
    },
    {
      "epoch": 0.4007478279995601,
      "grad_norm": 0.2633639872074127,
      "learning_rate": 8e-05,
      "loss": 1.5864,
      "step": 1822
    },
    {
      "epoch": 0.40096777741119544,
      "grad_norm": 0.3556326925754547,
      "learning_rate": 8e-05,
      "loss": 1.5781,
      "step": 1823
    },
    {
      "epoch": 0.40118772682283077,
      "grad_norm": 0.2832813560962677,
      "learning_rate": 8e-05,
      "loss": 1.7145,
      "step": 1824
    },
    {
      "epoch": 0.40140767623446605,
      "grad_norm": 0.2862699627876282,
      "learning_rate": 8e-05,
      "loss": 1.7753,
      "step": 1825
    },
    {
      "epoch": 0.4016276256461014,
      "grad_norm": 0.3274460732936859,
      "learning_rate": 8e-05,
      "loss": 1.7979,
      "step": 1826
    },
    {
      "epoch": 0.4018475750577367,
      "grad_norm": 0.277118444442749,
      "learning_rate": 8e-05,
      "loss": 1.6089,
      "step": 1827
    },
    {
      "epoch": 0.40206752446937205,
      "grad_norm": 0.278337687253952,
      "learning_rate": 8e-05,
      "loss": 1.5788,
      "step": 1828
    },
    {
      "epoch": 0.4022874738810074,
      "grad_norm": 0.28072914481163025,
      "learning_rate": 8e-05,
      "loss": 1.6735,
      "step": 1829
    },
    {
      "epoch": 0.4025074232926427,
      "grad_norm": 0.2815505564212799,
      "learning_rate": 8e-05,
      "loss": 1.7593,
      "step": 1830
    },
    {
      "epoch": 0.402727372704278,
      "grad_norm": 0.2957006096839905,
      "learning_rate": 8e-05,
      "loss": 1.865,
      "step": 1831
    },
    {
      "epoch": 0.40294732211591333,
      "grad_norm": 0.3079582452774048,
      "learning_rate": 8e-05,
      "loss": 1.7421,
      "step": 1832
    },
    {
      "epoch": 0.40316727152754867,
      "grad_norm": 0.2924387454986572,
      "learning_rate": 8e-05,
      "loss": 1.7462,
      "step": 1833
    },
    {
      "epoch": 0.403387220939184,
      "grad_norm": 0.28879454731941223,
      "learning_rate": 8e-05,
      "loss": 1.7433,
      "step": 1834
    },
    {
      "epoch": 0.40360717035081933,
      "grad_norm": 0.27446237206459045,
      "learning_rate": 8e-05,
      "loss": 1.5869,
      "step": 1835
    },
    {
      "epoch": 0.4038271197624546,
      "grad_norm": 0.3164878487586975,
      "learning_rate": 8e-05,
      "loss": 1.7505,
      "step": 1836
    },
    {
      "epoch": 0.40404706917408995,
      "grad_norm": 0.25979530811309814,
      "learning_rate": 8e-05,
      "loss": 1.6001,
      "step": 1837
    },
    {
      "epoch": 0.4042670185857253,
      "grad_norm": 0.30625709891319275,
      "learning_rate": 8e-05,
      "loss": 1.7907,
      "step": 1838
    },
    {
      "epoch": 0.4044869679973606,
      "grad_norm": 0.27351540327072144,
      "learning_rate": 8e-05,
      "loss": 1.5835,
      "step": 1839
    },
    {
      "epoch": 0.40470691740899595,
      "grad_norm": 0.302372545003891,
      "learning_rate": 8e-05,
      "loss": 1.7821,
      "step": 1840
    },
    {
      "epoch": 0.4049268668206313,
      "grad_norm": 0.2910183370113373,
      "learning_rate": 8e-05,
      "loss": 1.7993,
      "step": 1841
    },
    {
      "epoch": 0.40514681623226656,
      "grad_norm": 0.2934883236885071,
      "learning_rate": 8e-05,
      "loss": 1.5928,
      "step": 1842
    },
    {
      "epoch": 0.4053667656439019,
      "grad_norm": 0.2586327791213989,
      "learning_rate": 8e-05,
      "loss": 1.5714,
      "step": 1843
    },
    {
      "epoch": 0.40558671505553723,
      "grad_norm": 0.27952027320861816,
      "learning_rate": 8e-05,
      "loss": 1.8168,
      "step": 1844
    },
    {
      "epoch": 0.40580666446717256,
      "grad_norm": 0.2987437844276428,
      "learning_rate": 8e-05,
      "loss": 1.555,
      "step": 1845
    },
    {
      "epoch": 0.4060266138788079,
      "grad_norm": 0.29165002703666687,
      "learning_rate": 8e-05,
      "loss": 1.6865,
      "step": 1846
    },
    {
      "epoch": 0.4062465632904432,
      "grad_norm": 0.2825503945350647,
      "learning_rate": 8e-05,
      "loss": 1.8814,
      "step": 1847
    },
    {
      "epoch": 0.4064665127020785,
      "grad_norm": 0.27995482087135315,
      "learning_rate": 8e-05,
      "loss": 1.6897,
      "step": 1848
    },
    {
      "epoch": 0.40668646211371384,
      "grad_norm": 0.2735064923763275,
      "learning_rate": 8e-05,
      "loss": 1.7279,
      "step": 1849
    },
    {
      "epoch": 0.4069064115253492,
      "grad_norm": 0.2850511074066162,
      "learning_rate": 8e-05,
      "loss": 1.6459,
      "step": 1850
    },
    {
      "epoch": 0.4071263609369845,
      "grad_norm": 0.3000599145889282,
      "learning_rate": 8e-05,
      "loss": 1.7301,
      "step": 1851
    },
    {
      "epoch": 0.40734631034861984,
      "grad_norm": 0.2768002152442932,
      "learning_rate": 8e-05,
      "loss": 1.5748,
      "step": 1852
    },
    {
      "epoch": 0.4075662597602551,
      "grad_norm": 0.26737141609191895,
      "learning_rate": 8e-05,
      "loss": 1.5895,
      "step": 1853
    },
    {
      "epoch": 0.40778620917189046,
      "grad_norm": 0.26408424973487854,
      "learning_rate": 8e-05,
      "loss": 1.5611,
      "step": 1854
    },
    {
      "epoch": 0.4080061585835258,
      "grad_norm": 0.2646276354789734,
      "learning_rate": 8e-05,
      "loss": 1.5865,
      "step": 1855
    },
    {
      "epoch": 0.4082261079951611,
      "grad_norm": 0.27871212363243103,
      "learning_rate": 8e-05,
      "loss": 1.8202,
      "step": 1856
    },
    {
      "epoch": 0.40844605740679646,
      "grad_norm": 0.3234533965587616,
      "learning_rate": 8e-05,
      "loss": 1.8213,
      "step": 1857
    },
    {
      "epoch": 0.40866600681843174,
      "grad_norm": 0.2705099284648895,
      "learning_rate": 8e-05,
      "loss": 1.6637,
      "step": 1858
    },
    {
      "epoch": 0.40888595623006707,
      "grad_norm": 0.28647711873054504,
      "learning_rate": 8e-05,
      "loss": 1.7396,
      "step": 1859
    },
    {
      "epoch": 0.4091059056417024,
      "grad_norm": 0.2812083959579468,
      "learning_rate": 8e-05,
      "loss": 1.5663,
      "step": 1860
    },
    {
      "epoch": 0.40932585505333774,
      "grad_norm": 0.2818193733692169,
      "learning_rate": 8e-05,
      "loss": 1.6073,
      "step": 1861
    },
    {
      "epoch": 0.4095458044649731,
      "grad_norm": 0.29906994104385376,
      "learning_rate": 8e-05,
      "loss": 1.7061,
      "step": 1862
    },
    {
      "epoch": 0.4097657538766084,
      "grad_norm": 0.27941465377807617,
      "learning_rate": 8e-05,
      "loss": 1.7282,
      "step": 1863
    },
    {
      "epoch": 0.4099857032882437,
      "grad_norm": 0.27629899978637695,
      "learning_rate": 8e-05,
      "loss": 1.6879,
      "step": 1864
    },
    {
      "epoch": 0.410205652699879,
      "grad_norm": 0.2792319059371948,
      "learning_rate": 8e-05,
      "loss": 1.7196,
      "step": 1865
    },
    {
      "epoch": 0.41042560211151435,
      "grad_norm": 0.2763090431690216,
      "learning_rate": 8e-05,
      "loss": 1.673,
      "step": 1866
    },
    {
      "epoch": 0.4106455515231497,
      "grad_norm": 0.2930999994277954,
      "learning_rate": 8e-05,
      "loss": 1.6919,
      "step": 1867
    },
    {
      "epoch": 0.410865500934785,
      "grad_norm": 0.2748461365699768,
      "learning_rate": 8e-05,
      "loss": 1.7553,
      "step": 1868
    },
    {
      "epoch": 0.4110854503464203,
      "grad_norm": 0.2742187976837158,
      "learning_rate": 8e-05,
      "loss": 1.6786,
      "step": 1869
    },
    {
      "epoch": 0.41130539975805563,
      "grad_norm": 0.3050731420516968,
      "learning_rate": 8e-05,
      "loss": 1.4902,
      "step": 1870
    },
    {
      "epoch": 0.41152534916969097,
      "grad_norm": 0.29456627368927,
      "learning_rate": 8e-05,
      "loss": 1.758,
      "step": 1871
    },
    {
      "epoch": 0.4117452985813263,
      "grad_norm": 0.2844219505786896,
      "learning_rate": 8e-05,
      "loss": 1.6206,
      "step": 1872
    },
    {
      "epoch": 0.41196524799296164,
      "grad_norm": 0.28889915347099304,
      "learning_rate": 8e-05,
      "loss": 1.6907,
      "step": 1873
    },
    {
      "epoch": 0.41218519740459697,
      "grad_norm": 0.27245181798934937,
      "learning_rate": 8e-05,
      "loss": 1.6749,
      "step": 1874
    },
    {
      "epoch": 0.41240514681623225,
      "grad_norm": 0.2927252948284149,
      "learning_rate": 8e-05,
      "loss": 1.6382,
      "step": 1875
    },
    {
      "epoch": 0.4126250962278676,
      "grad_norm": 0.27153030037879944,
      "learning_rate": 8e-05,
      "loss": 1.6011,
      "step": 1876
    },
    {
      "epoch": 0.4128450456395029,
      "grad_norm": 0.2807110846042633,
      "learning_rate": 8e-05,
      "loss": 1.7126,
      "step": 1877
    },
    {
      "epoch": 0.41306499505113825,
      "grad_norm": 0.27375784516334534,
      "learning_rate": 8e-05,
      "loss": 1.7443,
      "step": 1878
    },
    {
      "epoch": 0.4132849444627736,
      "grad_norm": 0.27330929040908813,
      "learning_rate": 8e-05,
      "loss": 1.6305,
      "step": 1879
    },
    {
      "epoch": 0.41350489387440886,
      "grad_norm": 0.27126336097717285,
      "learning_rate": 8e-05,
      "loss": 1.6688,
      "step": 1880
    },
    {
      "epoch": 0.4137248432860442,
      "grad_norm": 0.2768147885799408,
      "learning_rate": 8e-05,
      "loss": 1.7274,
      "step": 1881
    },
    {
      "epoch": 0.41394479269767953,
      "grad_norm": 0.2686031460762024,
      "learning_rate": 8e-05,
      "loss": 1.6445,
      "step": 1882
    },
    {
      "epoch": 0.41416474210931487,
      "grad_norm": 0.27737778425216675,
      "learning_rate": 8e-05,
      "loss": 1.5226,
      "step": 1883
    },
    {
      "epoch": 0.4143846915209502,
      "grad_norm": 0.2761901319026947,
      "learning_rate": 8e-05,
      "loss": 1.6884,
      "step": 1884
    },
    {
      "epoch": 0.41460464093258553,
      "grad_norm": 0.28609856963157654,
      "learning_rate": 8e-05,
      "loss": 1.7719,
      "step": 1885
    },
    {
      "epoch": 0.4148245903442208,
      "grad_norm": 0.2904943525791168,
      "learning_rate": 8e-05,
      "loss": 1.6979,
      "step": 1886
    },
    {
      "epoch": 0.41504453975585615,
      "grad_norm": 0.3016435503959656,
      "learning_rate": 8e-05,
      "loss": 1.7912,
      "step": 1887
    },
    {
      "epoch": 0.4152644891674915,
      "grad_norm": 0.27562782168388367,
      "learning_rate": 8e-05,
      "loss": 1.5822,
      "step": 1888
    },
    {
      "epoch": 0.4154844385791268,
      "grad_norm": 0.2841348648071289,
      "learning_rate": 8e-05,
      "loss": 1.7524,
      "step": 1889
    },
    {
      "epoch": 0.41570438799076215,
      "grad_norm": 0.26393935084342957,
      "learning_rate": 8e-05,
      "loss": 1.6219,
      "step": 1890
    },
    {
      "epoch": 0.4159243374023974,
      "grad_norm": 0.2792678773403168,
      "learning_rate": 8e-05,
      "loss": 1.7243,
      "step": 1891
    },
    {
      "epoch": 0.41614428681403276,
      "grad_norm": 0.291425496339798,
      "learning_rate": 8e-05,
      "loss": 1.7499,
      "step": 1892
    },
    {
      "epoch": 0.4163642362256681,
      "grad_norm": 0.2737634778022766,
      "learning_rate": 8e-05,
      "loss": 1.6565,
      "step": 1893
    },
    {
      "epoch": 0.41658418563730343,
      "grad_norm": 0.26807767152786255,
      "learning_rate": 8e-05,
      "loss": 1.6149,
      "step": 1894
    },
    {
      "epoch": 0.41680413504893876,
      "grad_norm": 0.28826507925987244,
      "learning_rate": 8e-05,
      "loss": 1.6857,
      "step": 1895
    },
    {
      "epoch": 0.41702408446057404,
      "grad_norm": 0.27604466676712036,
      "learning_rate": 8e-05,
      "loss": 1.7689,
      "step": 1896
    },
    {
      "epoch": 0.4172440338722094,
      "grad_norm": 0.27355703711509705,
      "learning_rate": 8e-05,
      "loss": 1.7254,
      "step": 1897
    },
    {
      "epoch": 0.4174639832838447,
      "grad_norm": 0.26692044734954834,
      "learning_rate": 8e-05,
      "loss": 1.6372,
      "step": 1898
    },
    {
      "epoch": 0.41768393269548004,
      "grad_norm": 0.27527916431427,
      "learning_rate": 8e-05,
      "loss": 1.6913,
      "step": 1899
    },
    {
      "epoch": 0.4179038821071154,
      "grad_norm": 0.26881837844848633,
      "learning_rate": 8e-05,
      "loss": 1.6663,
      "step": 1900
    },
    {
      "epoch": 0.4181238315187507,
      "grad_norm": 0.27977946400642395,
      "learning_rate": 8e-05,
      "loss": 1.8117,
      "step": 1901
    },
    {
      "epoch": 0.418343780930386,
      "grad_norm": 0.2958911955356598,
      "learning_rate": 8e-05,
      "loss": 1.6603,
      "step": 1902
    },
    {
      "epoch": 0.4185637303420213,
      "grad_norm": 0.2845151424407959,
      "learning_rate": 8e-05,
      "loss": 1.7517,
      "step": 1903
    },
    {
      "epoch": 0.41878367975365666,
      "grad_norm": 0.2804581820964813,
      "learning_rate": 8e-05,
      "loss": 1.765,
      "step": 1904
    },
    {
      "epoch": 0.419003629165292,
      "grad_norm": 0.29568520188331604,
      "learning_rate": 8e-05,
      "loss": 1.6995,
      "step": 1905
    },
    {
      "epoch": 0.4192235785769273,
      "grad_norm": 0.303100049495697,
      "learning_rate": 8e-05,
      "loss": 1.77,
      "step": 1906
    },
    {
      "epoch": 0.4194435279885626,
      "grad_norm": 0.26847636699676514,
      "learning_rate": 8e-05,
      "loss": 1.6964,
      "step": 1907
    },
    {
      "epoch": 0.41966347740019794,
      "grad_norm": 0.2791590094566345,
      "learning_rate": 8e-05,
      "loss": 1.5912,
      "step": 1908
    },
    {
      "epoch": 0.41988342681183327,
      "grad_norm": 0.2687268555164337,
      "learning_rate": 8e-05,
      "loss": 1.6163,
      "step": 1909
    },
    {
      "epoch": 0.4201033762234686,
      "grad_norm": 0.29087433218955994,
      "learning_rate": 8e-05,
      "loss": 1.6588,
      "step": 1910
    },
    {
      "epoch": 0.42032332563510394,
      "grad_norm": 0.29639971256256104,
      "learning_rate": 8e-05,
      "loss": 1.6911,
      "step": 1911
    },
    {
      "epoch": 0.4205432750467393,
      "grad_norm": 0.27669841051101685,
      "learning_rate": 8e-05,
      "loss": 1.7078,
      "step": 1912
    },
    {
      "epoch": 0.42076322445837455,
      "grad_norm": 0.2851327955722809,
      "learning_rate": 8e-05,
      "loss": 1.7217,
      "step": 1913
    },
    {
      "epoch": 0.4209831738700099,
      "grad_norm": 0.27069011330604553,
      "learning_rate": 8e-05,
      "loss": 1.7076,
      "step": 1914
    },
    {
      "epoch": 0.4212031232816452,
      "grad_norm": 0.26195240020751953,
      "learning_rate": 8e-05,
      "loss": 1.6647,
      "step": 1915
    },
    {
      "epoch": 0.42142307269328055,
      "grad_norm": 0.3046209216117859,
      "learning_rate": 8e-05,
      "loss": 1.5303,
      "step": 1916
    },
    {
      "epoch": 0.4216430221049159,
      "grad_norm": 0.29437899589538574,
      "learning_rate": 8e-05,
      "loss": 1.6589,
      "step": 1917
    },
    {
      "epoch": 0.42186297151655117,
      "grad_norm": 0.2954728603363037,
      "learning_rate": 8e-05,
      "loss": 1.777,
      "step": 1918
    },
    {
      "epoch": 0.4220829209281865,
      "grad_norm": 0.2612738609313965,
      "learning_rate": 8e-05,
      "loss": 1.5668,
      "step": 1919
    },
    {
      "epoch": 0.42230287033982183,
      "grad_norm": 0.3015122413635254,
      "learning_rate": 8e-05,
      "loss": 1.7861,
      "step": 1920
    },
    {
      "epoch": 0.42252281975145717,
      "grad_norm": 0.3785838484764099,
      "learning_rate": 8e-05,
      "loss": 1.8979,
      "step": 1921
    },
    {
      "epoch": 0.4227427691630925,
      "grad_norm": 0.2849038541316986,
      "learning_rate": 8e-05,
      "loss": 1.6891,
      "step": 1922
    },
    {
      "epoch": 0.42296271857472784,
      "grad_norm": 0.278728723526001,
      "learning_rate": 8e-05,
      "loss": 1.7891,
      "step": 1923
    },
    {
      "epoch": 0.4231826679863631,
      "grad_norm": 0.27032172679901123,
      "learning_rate": 8e-05,
      "loss": 1.6963,
      "step": 1924
    },
    {
      "epoch": 0.42340261739799845,
      "grad_norm": 0.2731832265853882,
      "learning_rate": 8e-05,
      "loss": 1.632,
      "step": 1925
    },
    {
      "epoch": 0.4236225668096338,
      "grad_norm": 0.30378425121307373,
      "learning_rate": 8e-05,
      "loss": 1.7823,
      "step": 1926
    },
    {
      "epoch": 0.4238425162212691,
      "grad_norm": 0.27693971991539,
      "learning_rate": 8e-05,
      "loss": 1.483,
      "step": 1927
    },
    {
      "epoch": 0.42406246563290445,
      "grad_norm": 0.2719477415084839,
      "learning_rate": 8e-05,
      "loss": 1.6708,
      "step": 1928
    },
    {
      "epoch": 0.42428241504453973,
      "grad_norm": 0.26625335216522217,
      "learning_rate": 8e-05,
      "loss": 1.4946,
      "step": 1929
    },
    {
      "epoch": 0.42450236445617506,
      "grad_norm": 0.2843473553657532,
      "learning_rate": 8e-05,
      "loss": 1.722,
      "step": 1930
    },
    {
      "epoch": 0.4247223138678104,
      "grad_norm": 0.3453083336353302,
      "learning_rate": 8e-05,
      "loss": 1.7238,
      "step": 1931
    },
    {
      "epoch": 0.42494226327944573,
      "grad_norm": 0.25626078248023987,
      "learning_rate": 8e-05,
      "loss": 1.4706,
      "step": 1932
    },
    {
      "epoch": 0.42516221269108107,
      "grad_norm": 0.2908123731613159,
      "learning_rate": 8e-05,
      "loss": 1.7105,
      "step": 1933
    },
    {
      "epoch": 0.4253821621027164,
      "grad_norm": 0.33517104387283325,
      "learning_rate": 8e-05,
      "loss": 1.8023,
      "step": 1934
    },
    {
      "epoch": 0.4256021115143517,
      "grad_norm": 0.28047069907188416,
      "learning_rate": 8e-05,
      "loss": 1.6266,
      "step": 1935
    },
    {
      "epoch": 0.425822060925987,
      "grad_norm": 0.2778942584991455,
      "learning_rate": 8e-05,
      "loss": 1.6866,
      "step": 1936
    },
    {
      "epoch": 0.42604201033762235,
      "grad_norm": 0.3038877248764038,
      "learning_rate": 8e-05,
      "loss": 1.6075,
      "step": 1937
    },
    {
      "epoch": 0.4262619597492577,
      "grad_norm": 0.2814297378063202,
      "learning_rate": 8e-05,
      "loss": 1.5939,
      "step": 1938
    },
    {
      "epoch": 0.426481909160893,
      "grad_norm": 0.27854403853416443,
      "learning_rate": 8e-05,
      "loss": 1.5943,
      "step": 1939
    },
    {
      "epoch": 0.4267018585725283,
      "grad_norm": 0.2924019694328308,
      "learning_rate": 8e-05,
      "loss": 1.8193,
      "step": 1940
    },
    {
      "epoch": 0.4269218079841636,
      "grad_norm": 0.2862766683101654,
      "learning_rate": 8e-05,
      "loss": 1.6065,
      "step": 1941
    },
    {
      "epoch": 0.42714175739579896,
      "grad_norm": 0.2696346342563629,
      "learning_rate": 8e-05,
      "loss": 1.5343,
      "step": 1942
    },
    {
      "epoch": 0.4273617068074343,
      "grad_norm": 0.2578338384628296,
      "learning_rate": 8e-05,
      "loss": 1.5786,
      "step": 1943
    },
    {
      "epoch": 0.42758165621906963,
      "grad_norm": 0.28594937920570374,
      "learning_rate": 8e-05,
      "loss": 1.725,
      "step": 1944
    },
    {
      "epoch": 0.42780160563070496,
      "grad_norm": 0.2808282971382141,
      "learning_rate": 8e-05,
      "loss": 1.7951,
      "step": 1945
    },
    {
      "epoch": 0.42802155504234024,
      "grad_norm": 0.32533401250839233,
      "learning_rate": 8e-05,
      "loss": 1.9645,
      "step": 1946
    },
    {
      "epoch": 0.4282415044539756,
      "grad_norm": 0.2737642228603363,
      "learning_rate": 8e-05,
      "loss": 1.6243,
      "step": 1947
    },
    {
      "epoch": 0.4284614538656109,
      "grad_norm": 0.2885657250881195,
      "learning_rate": 8e-05,
      "loss": 1.7338,
      "step": 1948
    },
    {
      "epoch": 0.42868140327724624,
      "grad_norm": 0.2788100242614746,
      "learning_rate": 8e-05,
      "loss": 1.76,
      "step": 1949
    },
    {
      "epoch": 0.4289013526888816,
      "grad_norm": 0.2899073362350464,
      "learning_rate": 8e-05,
      "loss": 1.7739,
      "step": 1950
    },
    {
      "epoch": 0.42912130210051685,
      "grad_norm": 0.2874782681465149,
      "learning_rate": 8e-05,
      "loss": 1.8283,
      "step": 1951
    },
    {
      "epoch": 0.4293412515121522,
      "grad_norm": 0.2757413685321808,
      "learning_rate": 8e-05,
      "loss": 1.641,
      "step": 1952
    },
    {
      "epoch": 0.4295612009237875,
      "grad_norm": 0.2811121940612793,
      "learning_rate": 8e-05,
      "loss": 1.7231,
      "step": 1953
    },
    {
      "epoch": 0.42978115033542286,
      "grad_norm": 0.3400493860244751,
      "learning_rate": 8e-05,
      "loss": 1.8431,
      "step": 1954
    },
    {
      "epoch": 0.4300010997470582,
      "grad_norm": 0.29006627202033997,
      "learning_rate": 8e-05,
      "loss": 1.7438,
      "step": 1955
    },
    {
      "epoch": 0.4302210491586935,
      "grad_norm": 0.30233392119407654,
      "learning_rate": 8e-05,
      "loss": 1.6603,
      "step": 1956
    },
    {
      "epoch": 0.4304409985703288,
      "grad_norm": 0.2921263873577118,
      "learning_rate": 8e-05,
      "loss": 1.6604,
      "step": 1957
    },
    {
      "epoch": 0.43066094798196414,
      "grad_norm": 0.27695250511169434,
      "learning_rate": 8e-05,
      "loss": 1.7536,
      "step": 1958
    },
    {
      "epoch": 0.43088089739359947,
      "grad_norm": 0.2827337980270386,
      "learning_rate": 8e-05,
      "loss": 1.6324,
      "step": 1959
    },
    {
      "epoch": 0.4311008468052348,
      "grad_norm": 0.27993375062942505,
      "learning_rate": 8e-05,
      "loss": 1.7168,
      "step": 1960
    },
    {
      "epoch": 0.43132079621687014,
      "grad_norm": 0.2801220417022705,
      "learning_rate": 8e-05,
      "loss": 1.705,
      "step": 1961
    },
    {
      "epoch": 0.4315407456285054,
      "grad_norm": 0.27520567178726196,
      "learning_rate": 8e-05,
      "loss": 1.664,
      "step": 1962
    },
    {
      "epoch": 0.43176069504014075,
      "grad_norm": 0.26910632848739624,
      "learning_rate": 8e-05,
      "loss": 1.3616,
      "step": 1963
    },
    {
      "epoch": 0.4319806444517761,
      "grad_norm": 0.27770352363586426,
      "learning_rate": 8e-05,
      "loss": 1.6689,
      "step": 1964
    },
    {
      "epoch": 0.4322005938634114,
      "grad_norm": 0.27606719732284546,
      "learning_rate": 8e-05,
      "loss": 1.6644,
      "step": 1965
    },
    {
      "epoch": 0.43242054327504675,
      "grad_norm": 0.27787330746650696,
      "learning_rate": 8e-05,
      "loss": 1.8854,
      "step": 1966
    },
    {
      "epoch": 0.4326404926866821,
      "grad_norm": 0.26479870080947876,
      "learning_rate": 8e-05,
      "loss": 1.5904,
      "step": 1967
    },
    {
      "epoch": 0.43286044209831737,
      "grad_norm": 0.27598053216934204,
      "learning_rate": 8e-05,
      "loss": 1.5666,
      "step": 1968
    },
    {
      "epoch": 0.4330803915099527,
      "grad_norm": 0.27461937069892883,
      "learning_rate": 8e-05,
      "loss": 1.5487,
      "step": 1969
    },
    {
      "epoch": 0.43330034092158803,
      "grad_norm": 0.2928270399570465,
      "learning_rate": 8e-05,
      "loss": 1.8173,
      "step": 1970
    },
    {
      "epoch": 0.43352029033322337,
      "grad_norm": 0.30754199624061584,
      "learning_rate": 8e-05,
      "loss": 1.6762,
      "step": 1971
    },
    {
      "epoch": 0.4337402397448587,
      "grad_norm": 0.2676936089992523,
      "learning_rate": 8e-05,
      "loss": 1.7314,
      "step": 1972
    },
    {
      "epoch": 0.433960189156494,
      "grad_norm": 0.2919710576534271,
      "learning_rate": 8e-05,
      "loss": 1.8586,
      "step": 1973
    },
    {
      "epoch": 0.4341801385681293,
      "grad_norm": 0.28165963292121887,
      "learning_rate": 8e-05,
      "loss": 1.7943,
      "step": 1974
    },
    {
      "epoch": 0.43440008797976465,
      "grad_norm": 0.2700537443161011,
      "learning_rate": 8e-05,
      "loss": 1.613,
      "step": 1975
    },
    {
      "epoch": 0.4346200373914,
      "grad_norm": 0.26830658316612244,
      "learning_rate": 8e-05,
      "loss": 1.5854,
      "step": 1976
    },
    {
      "epoch": 0.4348399868030353,
      "grad_norm": 0.28799256682395935,
      "learning_rate": 8e-05,
      "loss": 1.8246,
      "step": 1977
    },
    {
      "epoch": 0.43505993621467065,
      "grad_norm": 0.27226150035858154,
      "learning_rate": 8e-05,
      "loss": 1.6252,
      "step": 1978
    },
    {
      "epoch": 0.43527988562630593,
      "grad_norm": 0.2646162807941437,
      "learning_rate": 8e-05,
      "loss": 1.5699,
      "step": 1979
    },
    {
      "epoch": 0.43549983503794126,
      "grad_norm": 0.27331140637397766,
      "learning_rate": 8e-05,
      "loss": 1.6893,
      "step": 1980
    },
    {
      "epoch": 0.4357197844495766,
      "grad_norm": 0.26996269822120667,
      "learning_rate": 8e-05,
      "loss": 1.6004,
      "step": 1981
    },
    {
      "epoch": 0.43593973386121193,
      "grad_norm": 0.29484307765960693,
      "learning_rate": 8e-05,
      "loss": 1.551,
      "step": 1982
    },
    {
      "epoch": 0.43615968327284727,
      "grad_norm": 0.28224268555641174,
      "learning_rate": 8e-05,
      "loss": 1.6898,
      "step": 1983
    },
    {
      "epoch": 0.43637963268448254,
      "grad_norm": 0.26172178983688354,
      "learning_rate": 8e-05,
      "loss": 1.4375,
      "step": 1984
    },
    {
      "epoch": 0.4365995820961179,
      "grad_norm": 0.2603735029697418,
      "learning_rate": 8e-05,
      "loss": 1.4528,
      "step": 1985
    },
    {
      "epoch": 0.4368195315077532,
      "grad_norm": 0.30643707513809204,
      "learning_rate": 8e-05,
      "loss": 1.688,
      "step": 1986
    },
    {
      "epoch": 0.43703948091938855,
      "grad_norm": 0.2951216995716095,
      "learning_rate": 8e-05,
      "loss": 1.7769,
      "step": 1987
    },
    {
      "epoch": 0.4372594303310239,
      "grad_norm": 0.2939329445362091,
      "learning_rate": 8e-05,
      "loss": 1.8161,
      "step": 1988
    },
    {
      "epoch": 0.4374793797426592,
      "grad_norm": 0.27539846301078796,
      "learning_rate": 8e-05,
      "loss": 1.6019,
      "step": 1989
    },
    {
      "epoch": 0.4376993291542945,
      "grad_norm": 0.2770693898200989,
      "learning_rate": 8e-05,
      "loss": 1.5972,
      "step": 1990
    },
    {
      "epoch": 0.4379192785659298,
      "grad_norm": 0.2832552492618561,
      "learning_rate": 8e-05,
      "loss": 1.7467,
      "step": 1991
    },
    {
      "epoch": 0.43813922797756516,
      "grad_norm": 0.2983148992061615,
      "learning_rate": 8e-05,
      "loss": 1.7181,
      "step": 1992
    },
    {
      "epoch": 0.4383591773892005,
      "grad_norm": 0.2829340994358063,
      "learning_rate": 8e-05,
      "loss": 1.5984,
      "step": 1993
    },
    {
      "epoch": 0.43857912680083583,
      "grad_norm": 0.2857687473297119,
      "learning_rate": 8e-05,
      "loss": 1.6471,
      "step": 1994
    },
    {
      "epoch": 0.4387990762124711,
      "grad_norm": 0.2669824957847595,
      "learning_rate": 8e-05,
      "loss": 1.6215,
      "step": 1995
    },
    {
      "epoch": 0.43901902562410644,
      "grad_norm": 0.28832894563674927,
      "learning_rate": 8e-05,
      "loss": 1.6884,
      "step": 1996
    },
    {
      "epoch": 0.4392389750357418,
      "grad_norm": 0.2919970154762268,
      "learning_rate": 8e-05,
      "loss": 1.7462,
      "step": 1997
    },
    {
      "epoch": 0.4394589244473771,
      "grad_norm": 0.2998509109020233,
      "learning_rate": 8e-05,
      "loss": 1.7219,
      "step": 1998
    },
    {
      "epoch": 0.43967887385901244,
      "grad_norm": 0.2780647575855255,
      "learning_rate": 8e-05,
      "loss": 1.8219,
      "step": 1999
    },
    {
      "epoch": 0.4398988232706478,
      "grad_norm": 0.2833268940448761,
      "learning_rate": 8e-05,
      "loss": 1.6873,
      "step": 2000
    },
    {
      "epoch": 0.44011877268228305,
      "grad_norm": 0.2802470624446869,
      "learning_rate": 8e-05,
      "loss": 1.7249,
      "step": 2001
    },
    {
      "epoch": 0.4403387220939184,
      "grad_norm": 0.2767699658870697,
      "learning_rate": 8e-05,
      "loss": 1.6789,
      "step": 2002
    },
    {
      "epoch": 0.4405586715055537,
      "grad_norm": 0.28534451127052307,
      "learning_rate": 8e-05,
      "loss": 1.6269,
      "step": 2003
    },
    {
      "epoch": 0.44077862091718906,
      "grad_norm": 0.28716540336608887,
      "learning_rate": 8e-05,
      "loss": 1.8097,
      "step": 2004
    },
    {
      "epoch": 0.4409985703288244,
      "grad_norm": 0.27516409754753113,
      "learning_rate": 8e-05,
      "loss": 1.837,
      "step": 2005
    },
    {
      "epoch": 0.44121851974045967,
      "grad_norm": 0.2910866439342499,
      "learning_rate": 8e-05,
      "loss": 1.61,
      "step": 2006
    },
    {
      "epoch": 0.441438469152095,
      "grad_norm": 0.2851128578186035,
      "learning_rate": 8e-05,
      "loss": 1.6939,
      "step": 2007
    },
    {
      "epoch": 0.44165841856373034,
      "grad_norm": 0.2709331214427948,
      "learning_rate": 8e-05,
      "loss": 1.6352,
      "step": 2008
    },
    {
      "epoch": 0.44187836797536567,
      "grad_norm": 0.28077712655067444,
      "learning_rate": 8e-05,
      "loss": 1.6119,
      "step": 2009
    },
    {
      "epoch": 0.442098317387001,
      "grad_norm": 0.2804681956768036,
      "learning_rate": 8e-05,
      "loss": 1.706,
      "step": 2010
    },
    {
      "epoch": 0.44231826679863634,
      "grad_norm": 0.28305575251579285,
      "learning_rate": 8e-05,
      "loss": 1.7501,
      "step": 2011
    },
    {
      "epoch": 0.4425382162102716,
      "grad_norm": 0.30372944474220276,
      "learning_rate": 8e-05,
      "loss": 1.69,
      "step": 2012
    },
    {
      "epoch": 0.44275816562190695,
      "grad_norm": 0.2695739269256592,
      "learning_rate": 8e-05,
      "loss": 1.4976,
      "step": 2013
    },
    {
      "epoch": 0.4429781150335423,
      "grad_norm": 0.27175822854042053,
      "learning_rate": 8e-05,
      "loss": 1.6941,
      "step": 2014
    },
    {
      "epoch": 0.4431980644451776,
      "grad_norm": 0.2786177396774292,
      "learning_rate": 8e-05,
      "loss": 1.719,
      "step": 2015
    },
    {
      "epoch": 0.44341801385681295,
      "grad_norm": 0.26625001430511475,
      "learning_rate": 8e-05,
      "loss": 1.6197,
      "step": 2016
    },
    {
      "epoch": 0.44363796326844823,
      "grad_norm": 0.29516807198524475,
      "learning_rate": 8e-05,
      "loss": 1.7305,
      "step": 2017
    },
    {
      "epoch": 0.44385791268008357,
      "grad_norm": 0.39562076330184937,
      "learning_rate": 8e-05,
      "loss": 1.783,
      "step": 2018
    },
    {
      "epoch": 0.4440778620917189,
      "grad_norm": 0.27659425139427185,
      "learning_rate": 8e-05,
      "loss": 1.6948,
      "step": 2019
    },
    {
      "epoch": 0.44429781150335423,
      "grad_norm": 0.2787366211414337,
      "learning_rate": 8e-05,
      "loss": 1.6237,
      "step": 2020
    },
    {
      "epoch": 0.44451776091498957,
      "grad_norm": 0.27939459681510925,
      "learning_rate": 8e-05,
      "loss": 1.7628,
      "step": 2021
    },
    {
      "epoch": 0.4447377103266249,
      "grad_norm": 0.27395081520080566,
      "learning_rate": 8e-05,
      "loss": 1.5619,
      "step": 2022
    },
    {
      "epoch": 0.4449576597382602,
      "grad_norm": 0.28255096077919006,
      "learning_rate": 8e-05,
      "loss": 1.7556,
      "step": 2023
    },
    {
      "epoch": 0.4451776091498955,
      "grad_norm": 0.2922489643096924,
      "learning_rate": 8e-05,
      "loss": 1.624,
      "step": 2024
    },
    {
      "epoch": 0.44539755856153085,
      "grad_norm": 0.4039583206176758,
      "learning_rate": 8e-05,
      "loss": 1.507,
      "step": 2025
    },
    {
      "epoch": 0.4456175079731662,
      "grad_norm": 0.28025928139686584,
      "learning_rate": 8e-05,
      "loss": 1.8057,
      "step": 2026
    },
    {
      "epoch": 0.4458374573848015,
      "grad_norm": 0.2777588963508606,
      "learning_rate": 8e-05,
      "loss": 1.6104,
      "step": 2027
    },
    {
      "epoch": 0.4460574067964368,
      "grad_norm": 0.2689501941204071,
      "learning_rate": 8e-05,
      "loss": 1.5339,
      "step": 2028
    },
    {
      "epoch": 0.44627735620807213,
      "grad_norm": 0.28021785616874695,
      "learning_rate": 8e-05,
      "loss": 1.7819,
      "step": 2029
    },
    {
      "epoch": 0.44649730561970746,
      "grad_norm": 0.26980918645858765,
      "learning_rate": 8e-05,
      "loss": 1.5995,
      "step": 2030
    },
    {
      "epoch": 0.4467172550313428,
      "grad_norm": 0.293047696352005,
      "learning_rate": 8e-05,
      "loss": 1.8571,
      "step": 2031
    },
    {
      "epoch": 0.44693720444297813,
      "grad_norm": 0.2841939330101013,
      "learning_rate": 8e-05,
      "loss": 1.5794,
      "step": 2032
    },
    {
      "epoch": 0.44715715385461346,
      "grad_norm": 0.2845712900161743,
      "learning_rate": 8e-05,
      "loss": 1.6635,
      "step": 2033
    },
    {
      "epoch": 0.44737710326624874,
      "grad_norm": 0.27919885516166687,
      "learning_rate": 8e-05,
      "loss": 1.7094,
      "step": 2034
    },
    {
      "epoch": 0.4475970526778841,
      "grad_norm": 0.30076712369918823,
      "learning_rate": 8e-05,
      "loss": 1.6989,
      "step": 2035
    },
    {
      "epoch": 0.4478170020895194,
      "grad_norm": 0.2666080892086029,
      "learning_rate": 8e-05,
      "loss": 1.689,
      "step": 2036
    },
    {
      "epoch": 0.44803695150115475,
      "grad_norm": 0.29133087396621704,
      "learning_rate": 8e-05,
      "loss": 1.8559,
      "step": 2037
    },
    {
      "epoch": 0.4482569009127901,
      "grad_norm": 0.28610843420028687,
      "learning_rate": 8e-05,
      "loss": 1.6556,
      "step": 2038
    },
    {
      "epoch": 0.44847685032442536,
      "grad_norm": 0.26724278926849365,
      "learning_rate": 8e-05,
      "loss": 1.6104,
      "step": 2039
    },
    {
      "epoch": 0.4486967997360607,
      "grad_norm": 0.2724173665046692,
      "learning_rate": 8e-05,
      "loss": 1.645,
      "step": 2040
    },
    {
      "epoch": 0.448916749147696,
      "grad_norm": 0.26570823788642883,
      "learning_rate": 8e-05,
      "loss": 1.4611,
      "step": 2041
    },
    {
      "epoch": 0.44913669855933136,
      "grad_norm": 0.29498788714408875,
      "learning_rate": 8e-05,
      "loss": 1.8787,
      "step": 2042
    },
    {
      "epoch": 0.4493566479709667,
      "grad_norm": 0.28211459517478943,
      "learning_rate": 8e-05,
      "loss": 1.6062,
      "step": 2043
    },
    {
      "epoch": 0.44957659738260203,
      "grad_norm": 0.3148192763328552,
      "learning_rate": 8e-05,
      "loss": 1.8734,
      "step": 2044
    },
    {
      "epoch": 0.4497965467942373,
      "grad_norm": 0.27721115946769714,
      "learning_rate": 8e-05,
      "loss": 1.7274,
      "step": 2045
    },
    {
      "epoch": 0.45001649620587264,
      "grad_norm": 0.29178541898727417,
      "learning_rate": 8e-05,
      "loss": 1.7568,
      "step": 2046
    },
    {
      "epoch": 0.450236445617508,
      "grad_norm": 0.27845948934555054,
      "learning_rate": 8e-05,
      "loss": 1.6615,
      "step": 2047
    },
    {
      "epoch": 0.4504563950291433,
      "grad_norm": 0.2741856873035431,
      "learning_rate": 8e-05,
      "loss": 1.5667,
      "step": 2048
    },
    {
      "epoch": 0.45067634444077864,
      "grad_norm": 0.28572753071784973,
      "learning_rate": 8e-05,
      "loss": 1.7115,
      "step": 2049
    },
    {
      "epoch": 0.4508962938524139,
      "grad_norm": 0.2769505977630615,
      "learning_rate": 8e-05,
      "loss": 1.6297,
      "step": 2050
    },
    {
      "epoch": 0.45111624326404925,
      "grad_norm": 0.27633893489837646,
      "learning_rate": 8e-05,
      "loss": 1.6455,
      "step": 2051
    },
    {
      "epoch": 0.4513361926756846,
      "grad_norm": 0.28455862402915955,
      "learning_rate": 8e-05,
      "loss": 1.7496,
      "step": 2052
    },
    {
      "epoch": 0.4515561420873199,
      "grad_norm": 0.2920532524585724,
      "learning_rate": 8e-05,
      "loss": 1.7671,
      "step": 2053
    },
    {
      "epoch": 0.45177609149895526,
      "grad_norm": 0.27528077363967896,
      "learning_rate": 8e-05,
      "loss": 1.6535,
      "step": 2054
    },
    {
      "epoch": 0.4519960409105906,
      "grad_norm": 0.2949519157409668,
      "learning_rate": 8e-05,
      "loss": 1.6796,
      "step": 2055
    },
    {
      "epoch": 0.45221599032222587,
      "grad_norm": 0.2740989923477173,
      "learning_rate": 8e-05,
      "loss": 1.4341,
      "step": 2056
    },
    {
      "epoch": 0.4524359397338612,
      "grad_norm": 0.31732696294784546,
      "learning_rate": 8e-05,
      "loss": 1.717,
      "step": 2057
    },
    {
      "epoch": 0.45265588914549654,
      "grad_norm": 0.2747776210308075,
      "learning_rate": 8e-05,
      "loss": 1.7162,
      "step": 2058
    },
    {
      "epoch": 0.45287583855713187,
      "grad_norm": 0.3037000596523285,
      "learning_rate": 8e-05,
      "loss": 1.6399,
      "step": 2059
    },
    {
      "epoch": 0.4530957879687672,
      "grad_norm": 0.29499107599258423,
      "learning_rate": 8e-05,
      "loss": 1.6306,
      "step": 2060
    },
    {
      "epoch": 0.4533157373804025,
      "grad_norm": 0.2890235483646393,
      "learning_rate": 8e-05,
      "loss": 1.8048,
      "step": 2061
    },
    {
      "epoch": 0.4535356867920378,
      "grad_norm": 0.28515708446502686,
      "learning_rate": 8e-05,
      "loss": 1.843,
      "step": 2062
    },
    {
      "epoch": 0.45375563620367315,
      "grad_norm": 0.28525930643081665,
      "learning_rate": 8e-05,
      "loss": 1.6568,
      "step": 2063
    },
    {
      "epoch": 0.4539755856153085,
      "grad_norm": 0.27117711305618286,
      "learning_rate": 8e-05,
      "loss": 1.6605,
      "step": 2064
    },
    {
      "epoch": 0.4541955350269438,
      "grad_norm": 0.2771861255168915,
      "learning_rate": 8e-05,
      "loss": 1.7553,
      "step": 2065
    },
    {
      "epoch": 0.45441548443857915,
      "grad_norm": 0.29263827204704285,
      "learning_rate": 8e-05,
      "loss": 1.7919,
      "step": 2066
    },
    {
      "epoch": 0.45463543385021443,
      "grad_norm": 0.2785603702068329,
      "learning_rate": 8e-05,
      "loss": 1.6686,
      "step": 2067
    },
    {
      "epoch": 0.45485538326184977,
      "grad_norm": 0.2752209007740021,
      "learning_rate": 8e-05,
      "loss": 1.5984,
      "step": 2068
    },
    {
      "epoch": 0.4550753326734851,
      "grad_norm": 0.279784232378006,
      "learning_rate": 8e-05,
      "loss": 1.6903,
      "step": 2069
    },
    {
      "epoch": 0.45529528208512043,
      "grad_norm": 0.2957722246646881,
      "learning_rate": 8e-05,
      "loss": 1.8176,
      "step": 2070
    },
    {
      "epoch": 0.45551523149675577,
      "grad_norm": 0.2798726260662079,
      "learning_rate": 8e-05,
      "loss": 1.6637,
      "step": 2071
    },
    {
      "epoch": 0.45573518090839105,
      "grad_norm": 0.26509538292884827,
      "learning_rate": 8e-05,
      "loss": 1.5034,
      "step": 2072
    },
    {
      "epoch": 0.4559551303200264,
      "grad_norm": 0.2984442114830017,
      "learning_rate": 8e-05,
      "loss": 1.8225,
      "step": 2073
    },
    {
      "epoch": 0.4561750797316617,
      "grad_norm": 0.28242239356040955,
      "learning_rate": 8e-05,
      "loss": 1.6707,
      "step": 2074
    },
    {
      "epoch": 0.45639502914329705,
      "grad_norm": 0.2722650468349457,
      "learning_rate": 8e-05,
      "loss": 1.5723,
      "step": 2075
    },
    {
      "epoch": 0.4566149785549324,
      "grad_norm": 0.25942909717559814,
      "learning_rate": 8e-05,
      "loss": 1.5417,
      "step": 2076
    },
    {
      "epoch": 0.4568349279665677,
      "grad_norm": 0.2782632112503052,
      "learning_rate": 8e-05,
      "loss": 1.5857,
      "step": 2077
    },
    {
      "epoch": 0.457054877378203,
      "grad_norm": 0.28298354148864746,
      "learning_rate": 8e-05,
      "loss": 1.6124,
      "step": 2078
    },
    {
      "epoch": 0.45727482678983833,
      "grad_norm": 0.2920227348804474,
      "learning_rate": 8e-05,
      "loss": 1.6247,
      "step": 2079
    },
    {
      "epoch": 0.45749477620147366,
      "grad_norm": 0.30804532766342163,
      "learning_rate": 8e-05,
      "loss": 1.8689,
      "step": 2080
    },
    {
      "epoch": 0.457714725613109,
      "grad_norm": 0.2759280204772949,
      "learning_rate": 8e-05,
      "loss": 1.6366,
      "step": 2081
    },
    {
      "epoch": 0.45793467502474433,
      "grad_norm": 0.27967414259910583,
      "learning_rate": 8e-05,
      "loss": 1.6391,
      "step": 2082
    },
    {
      "epoch": 0.4581546244363796,
      "grad_norm": 0.30624908208847046,
      "learning_rate": 8e-05,
      "loss": 1.8188,
      "step": 2083
    },
    {
      "epoch": 0.45837457384801494,
      "grad_norm": 0.2747632563114166,
      "learning_rate": 8e-05,
      "loss": 1.6394,
      "step": 2084
    },
    {
      "epoch": 0.4585945232596503,
      "grad_norm": 0.29921606183052063,
      "learning_rate": 8e-05,
      "loss": 1.6264,
      "step": 2085
    },
    {
      "epoch": 0.4588144726712856,
      "grad_norm": 0.27374643087387085,
      "learning_rate": 8e-05,
      "loss": 1.5955,
      "step": 2086
    },
    {
      "epoch": 0.45903442208292095,
      "grad_norm": 0.2804218530654907,
      "learning_rate": 8e-05,
      "loss": 1.6936,
      "step": 2087
    },
    {
      "epoch": 0.4592543714945563,
      "grad_norm": 0.288095623254776,
      "learning_rate": 8e-05,
      "loss": 1.658,
      "step": 2088
    },
    {
      "epoch": 0.45947432090619156,
      "grad_norm": 0.2622469961643219,
      "learning_rate": 8e-05,
      "loss": 1.5468,
      "step": 2089
    },
    {
      "epoch": 0.4596942703178269,
      "grad_norm": 0.298968106508255,
      "learning_rate": 8e-05,
      "loss": 1.8415,
      "step": 2090
    },
    {
      "epoch": 0.4599142197294622,
      "grad_norm": 0.27127620577812195,
      "learning_rate": 8e-05,
      "loss": 1.5154,
      "step": 2091
    },
    {
      "epoch": 0.46013416914109756,
      "grad_norm": 0.3025810122489929,
      "learning_rate": 8e-05,
      "loss": 1.7251,
      "step": 2092
    },
    {
      "epoch": 0.4603541185527329,
      "grad_norm": 0.2805241346359253,
      "learning_rate": 8e-05,
      "loss": 1.786,
      "step": 2093
    },
    {
      "epoch": 0.4605740679643682,
      "grad_norm": 0.28292620182037354,
      "learning_rate": 8e-05,
      "loss": 1.6205,
      "step": 2094
    },
    {
      "epoch": 0.4607940173760035,
      "grad_norm": 0.27890294790267944,
      "learning_rate": 8e-05,
      "loss": 1.6182,
      "step": 2095
    },
    {
      "epoch": 0.46101396678763884,
      "grad_norm": 0.2704887390136719,
      "learning_rate": 8e-05,
      "loss": 1.643,
      "step": 2096
    },
    {
      "epoch": 0.4612339161992742,
      "grad_norm": 0.27034714818000793,
      "learning_rate": 8e-05,
      "loss": 1.6918,
      "step": 2097
    },
    {
      "epoch": 0.4614538656109095,
      "grad_norm": 0.2763729691505432,
      "learning_rate": 8e-05,
      "loss": 1.6218,
      "step": 2098
    },
    {
      "epoch": 0.46167381502254484,
      "grad_norm": 0.28457143902778625,
      "learning_rate": 8e-05,
      "loss": 1.5602,
      "step": 2099
    },
    {
      "epoch": 0.4618937644341801,
      "grad_norm": 0.3102862536907196,
      "learning_rate": 8e-05,
      "loss": 1.6946,
      "step": 2100
    },
    {
      "epoch": 0.46211371384581545,
      "grad_norm": 0.2817099690437317,
      "learning_rate": 8e-05,
      "loss": 1.69,
      "step": 2101
    },
    {
      "epoch": 0.4623336632574508,
      "grad_norm": 0.2816404104232788,
      "learning_rate": 8e-05,
      "loss": 1.6651,
      "step": 2102
    },
    {
      "epoch": 0.4625536126690861,
      "grad_norm": 0.2756252884864807,
      "learning_rate": 8e-05,
      "loss": 1.6935,
      "step": 2103
    },
    {
      "epoch": 0.46277356208072146,
      "grad_norm": 0.28443071246147156,
      "learning_rate": 8e-05,
      "loss": 1.8278,
      "step": 2104
    },
    {
      "epoch": 0.46299351149235674,
      "grad_norm": 0.2955114543437958,
      "learning_rate": 8e-05,
      "loss": 1.7654,
      "step": 2105
    },
    {
      "epoch": 0.46321346090399207,
      "grad_norm": 0.30527764558792114,
      "learning_rate": 8e-05,
      "loss": 1.921,
      "step": 2106
    },
    {
      "epoch": 0.4634334103156274,
      "grad_norm": 0.28985050320625305,
      "learning_rate": 8e-05,
      "loss": 1.7637,
      "step": 2107
    },
    {
      "epoch": 0.46365335972726274,
      "grad_norm": 0.2904118299484253,
      "learning_rate": 8e-05,
      "loss": 1.7645,
      "step": 2108
    },
    {
      "epoch": 0.46387330913889807,
      "grad_norm": 0.3137964606285095,
      "learning_rate": 8e-05,
      "loss": 1.8301,
      "step": 2109
    },
    {
      "epoch": 0.46409325855053335,
      "grad_norm": 0.2634297013282776,
      "learning_rate": 8e-05,
      "loss": 1.5916,
      "step": 2110
    },
    {
      "epoch": 0.4643132079621687,
      "grad_norm": 0.26435586810112,
      "learning_rate": 8e-05,
      "loss": 1.5805,
      "step": 2111
    },
    {
      "epoch": 0.464533157373804,
      "grad_norm": 0.2845149040222168,
      "learning_rate": 8e-05,
      "loss": 1.6894,
      "step": 2112
    },
    {
      "epoch": 0.46475310678543935,
      "grad_norm": 0.3034592568874359,
      "learning_rate": 8e-05,
      "loss": 1.7122,
      "step": 2113
    },
    {
      "epoch": 0.4649730561970747,
      "grad_norm": 0.2862027585506439,
      "learning_rate": 8e-05,
      "loss": 1.5886,
      "step": 2114
    },
    {
      "epoch": 0.46519300560871,
      "grad_norm": 0.2709789574146271,
      "learning_rate": 8e-05,
      "loss": 1.6112,
      "step": 2115
    },
    {
      "epoch": 0.4654129550203453,
      "grad_norm": 0.3048953115940094,
      "learning_rate": 8e-05,
      "loss": 1.718,
      "step": 2116
    },
    {
      "epoch": 0.46563290443198063,
      "grad_norm": 0.295149564743042,
      "learning_rate": 8e-05,
      "loss": 1.7339,
      "step": 2117
    },
    {
      "epoch": 0.46585285384361597,
      "grad_norm": 0.27533626556396484,
      "learning_rate": 8e-05,
      "loss": 1.6143,
      "step": 2118
    },
    {
      "epoch": 0.4660728032552513,
      "grad_norm": 0.30383235216140747,
      "learning_rate": 8e-05,
      "loss": 1.7154,
      "step": 2119
    },
    {
      "epoch": 0.46629275266688663,
      "grad_norm": 0.2834450304508209,
      "learning_rate": 8e-05,
      "loss": 1.6892,
      "step": 2120
    },
    {
      "epoch": 0.4665127020785219,
      "grad_norm": 0.27407264709472656,
      "learning_rate": 8e-05,
      "loss": 1.7138,
      "step": 2121
    },
    {
      "epoch": 0.46673265149015725,
      "grad_norm": 0.28417688608169556,
      "learning_rate": 8e-05,
      "loss": 1.7416,
      "step": 2122
    },
    {
      "epoch": 0.4669526009017926,
      "grad_norm": 0.2915797233581543,
      "learning_rate": 8e-05,
      "loss": 1.7104,
      "step": 2123
    },
    {
      "epoch": 0.4671725503134279,
      "grad_norm": 0.29155269265174866,
      "learning_rate": 8e-05,
      "loss": 1.7374,
      "step": 2124
    },
    {
      "epoch": 0.46739249972506325,
      "grad_norm": 0.27683204412460327,
      "learning_rate": 8e-05,
      "loss": 1.5669,
      "step": 2125
    },
    {
      "epoch": 0.4676124491366986,
      "grad_norm": 0.2835148870944977,
      "learning_rate": 8e-05,
      "loss": 1.7426,
      "step": 2126
    },
    {
      "epoch": 0.46783239854833386,
      "grad_norm": 0.27906641364097595,
      "learning_rate": 8e-05,
      "loss": 1.6297,
      "step": 2127
    },
    {
      "epoch": 0.4680523479599692,
      "grad_norm": 0.28407955169677734,
      "learning_rate": 8e-05,
      "loss": 1.7351,
      "step": 2128
    },
    {
      "epoch": 0.46827229737160453,
      "grad_norm": 0.2793600261211395,
      "learning_rate": 8e-05,
      "loss": 1.5786,
      "step": 2129
    },
    {
      "epoch": 0.46849224678323986,
      "grad_norm": 0.2806802988052368,
      "learning_rate": 8e-05,
      "loss": 1.7466,
      "step": 2130
    },
    {
      "epoch": 0.4687121961948752,
      "grad_norm": 0.30251967906951904,
      "learning_rate": 8e-05,
      "loss": 1.6703,
      "step": 2131
    },
    {
      "epoch": 0.4689321456065105,
      "grad_norm": 0.275473415851593,
      "learning_rate": 8e-05,
      "loss": 1.5881,
      "step": 2132
    },
    {
      "epoch": 0.4691520950181458,
      "grad_norm": 0.28925517201423645,
      "learning_rate": 8e-05,
      "loss": 1.7369,
      "step": 2133
    },
    {
      "epoch": 0.46937204442978114,
      "grad_norm": 0.26768866181373596,
      "learning_rate": 8e-05,
      "loss": 1.5469,
      "step": 2134
    },
    {
      "epoch": 0.4695919938414165,
      "grad_norm": 0.272969514131546,
      "learning_rate": 8e-05,
      "loss": 1.6235,
      "step": 2135
    },
    {
      "epoch": 0.4698119432530518,
      "grad_norm": 0.39006346464157104,
      "learning_rate": 8e-05,
      "loss": 1.6943,
      "step": 2136
    },
    {
      "epoch": 0.47003189266468715,
      "grad_norm": 0.2898694574832916,
      "learning_rate": 8e-05,
      "loss": 1.7196,
      "step": 2137
    },
    {
      "epoch": 0.4702518420763224,
      "grad_norm": 0.28824204206466675,
      "learning_rate": 8e-05,
      "loss": 1.6807,
      "step": 2138
    },
    {
      "epoch": 0.47047179148795776,
      "grad_norm": 0.3024749755859375,
      "learning_rate": 8e-05,
      "loss": 1.7435,
      "step": 2139
    },
    {
      "epoch": 0.4706917408995931,
      "grad_norm": 0.2894933521747589,
      "learning_rate": 8e-05,
      "loss": 1.8047,
      "step": 2140
    },
    {
      "epoch": 0.4709116903112284,
      "grad_norm": 0.2900967299938202,
      "learning_rate": 8e-05,
      "loss": 1.6536,
      "step": 2141
    },
    {
      "epoch": 0.47113163972286376,
      "grad_norm": 0.2727701961994171,
      "learning_rate": 8e-05,
      "loss": 1.6187,
      "step": 2142
    },
    {
      "epoch": 0.47135158913449904,
      "grad_norm": 0.2630798816680908,
      "learning_rate": 8e-05,
      "loss": 1.5558,
      "step": 2143
    },
    {
      "epoch": 0.4715715385461344,
      "grad_norm": 0.2755641043186188,
      "learning_rate": 8e-05,
      "loss": 1.6247,
      "step": 2144
    },
    {
      "epoch": 0.4717914879577697,
      "grad_norm": 0.26855289936065674,
      "learning_rate": 8e-05,
      "loss": 1.6908,
      "step": 2145
    },
    {
      "epoch": 0.47201143736940504,
      "grad_norm": 0.26333558559417725,
      "learning_rate": 8e-05,
      "loss": 1.6467,
      "step": 2146
    },
    {
      "epoch": 0.4722313867810404,
      "grad_norm": 0.2696126103401184,
      "learning_rate": 8e-05,
      "loss": 1.646,
      "step": 2147
    },
    {
      "epoch": 0.4724513361926757,
      "grad_norm": 0.2838461101055145,
      "learning_rate": 8e-05,
      "loss": 1.7269,
      "step": 2148
    },
    {
      "epoch": 0.472671285604311,
      "grad_norm": 0.27359622716903687,
      "learning_rate": 8e-05,
      "loss": 1.6578,
      "step": 2149
    },
    {
      "epoch": 0.4728912350159463,
      "grad_norm": 0.28489992022514343,
      "learning_rate": 8e-05,
      "loss": 1.7451,
      "step": 2150
    },
    {
      "epoch": 0.47311118442758165,
      "grad_norm": 0.30069005489349365,
      "learning_rate": 8e-05,
      "loss": 1.7789,
      "step": 2151
    },
    {
      "epoch": 0.473331133839217,
      "grad_norm": 0.2787550091743469,
      "learning_rate": 8e-05,
      "loss": 1.8532,
      "step": 2152
    },
    {
      "epoch": 0.4735510832508523,
      "grad_norm": 0.28521937131881714,
      "learning_rate": 8e-05,
      "loss": 1.6202,
      "step": 2153
    },
    {
      "epoch": 0.4737710326624876,
      "grad_norm": 0.27512073516845703,
      "learning_rate": 8e-05,
      "loss": 1.6982,
      "step": 2154
    },
    {
      "epoch": 0.47399098207412294,
      "grad_norm": 0.28500398993492126,
      "learning_rate": 8e-05,
      "loss": 1.7262,
      "step": 2155
    },
    {
      "epoch": 0.47421093148575827,
      "grad_norm": 0.2889910340309143,
      "learning_rate": 8e-05,
      "loss": 1.7001,
      "step": 2156
    },
    {
      "epoch": 0.4744308808973936,
      "grad_norm": 0.2868637144565582,
      "learning_rate": 8e-05,
      "loss": 1.6569,
      "step": 2157
    },
    {
      "epoch": 0.47465083030902894,
      "grad_norm": 0.27974042296409607,
      "learning_rate": 8e-05,
      "loss": 1.7308,
      "step": 2158
    },
    {
      "epoch": 0.47487077972066427,
      "grad_norm": 0.2812412977218628,
      "learning_rate": 8e-05,
      "loss": 1.6586,
      "step": 2159
    },
    {
      "epoch": 0.47509072913229955,
      "grad_norm": 0.27973487973213196,
      "learning_rate": 8e-05,
      "loss": 1.7688,
      "step": 2160
    },
    {
      "epoch": 0.4753106785439349,
      "grad_norm": 0.2852223515510559,
      "learning_rate": 8e-05,
      "loss": 1.7773,
      "step": 2161
    },
    {
      "epoch": 0.4755306279555702,
      "grad_norm": 0.2702232301235199,
      "learning_rate": 8e-05,
      "loss": 1.5842,
      "step": 2162
    },
    {
      "epoch": 0.47575057736720555,
      "grad_norm": 0.26885986328125,
      "learning_rate": 8e-05,
      "loss": 1.6179,
      "step": 2163
    },
    {
      "epoch": 0.4759705267788409,
      "grad_norm": 0.26561740040779114,
      "learning_rate": 8e-05,
      "loss": 1.6105,
      "step": 2164
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 0.3136744499206543,
      "learning_rate": 8e-05,
      "loss": 1.7263,
      "step": 2165
    },
    {
      "epoch": 0.4764104256021115,
      "grad_norm": 0.29980844259262085,
      "learning_rate": 8e-05,
      "loss": 1.6428,
      "step": 2166
    },
    {
      "epoch": 0.47663037501374683,
      "grad_norm": 0.2782588303089142,
      "learning_rate": 8e-05,
      "loss": 1.5905,
      "step": 2167
    },
    {
      "epoch": 0.47685032442538217,
      "grad_norm": 0.3134911358356476,
      "learning_rate": 8e-05,
      "loss": 1.7437,
      "step": 2168
    },
    {
      "epoch": 0.4770702738370175,
      "grad_norm": 0.28305792808532715,
      "learning_rate": 8e-05,
      "loss": 1.6187,
      "step": 2169
    },
    {
      "epoch": 0.47729022324865283,
      "grad_norm": 0.2741806209087372,
      "learning_rate": 8e-05,
      "loss": 1.7017,
      "step": 2170
    },
    {
      "epoch": 0.4775101726602881,
      "grad_norm": 0.2861132323741913,
      "learning_rate": 8e-05,
      "loss": 1.7167,
      "step": 2171
    },
    {
      "epoch": 0.47773012207192345,
      "grad_norm": 0.2796178162097931,
      "learning_rate": 8e-05,
      "loss": 1.7747,
      "step": 2172
    },
    {
      "epoch": 0.4779500714835588,
      "grad_norm": 0.3019583821296692,
      "learning_rate": 8e-05,
      "loss": 1.9244,
      "step": 2173
    },
    {
      "epoch": 0.4781700208951941,
      "grad_norm": 0.2874825894832611,
      "learning_rate": 8e-05,
      "loss": 1.676,
      "step": 2174
    },
    {
      "epoch": 0.47838997030682945,
      "grad_norm": 0.2864963412284851,
      "learning_rate": 8e-05,
      "loss": 1.8289,
      "step": 2175
    },
    {
      "epoch": 0.4786099197184647,
      "grad_norm": 0.3347536623477936,
      "learning_rate": 8e-05,
      "loss": 1.8204,
      "step": 2176
    },
    {
      "epoch": 0.47882986913010006,
      "grad_norm": 0.2859993577003479,
      "learning_rate": 8e-05,
      "loss": 1.5472,
      "step": 2177
    },
    {
      "epoch": 0.4790498185417354,
      "grad_norm": 0.2972160875797272,
      "learning_rate": 8e-05,
      "loss": 1.6577,
      "step": 2178
    },
    {
      "epoch": 0.47926976795337073,
      "grad_norm": 0.26402032375335693,
      "learning_rate": 8e-05,
      "loss": 1.4936,
      "step": 2179
    },
    {
      "epoch": 0.47948971736500606,
      "grad_norm": 0.28069886565208435,
      "learning_rate": 8e-05,
      "loss": 1.8158,
      "step": 2180
    },
    {
      "epoch": 0.4797096667766414,
      "grad_norm": 0.2630525529384613,
      "learning_rate": 8e-05,
      "loss": 1.5814,
      "step": 2181
    },
    {
      "epoch": 0.4799296161882767,
      "grad_norm": 0.2999349534511566,
      "learning_rate": 8e-05,
      "loss": 1.7879,
      "step": 2182
    },
    {
      "epoch": 0.480149565599912,
      "grad_norm": 0.28722846508026123,
      "learning_rate": 8e-05,
      "loss": 1.7143,
      "step": 2183
    },
    {
      "epoch": 0.48036951501154734,
      "grad_norm": 0.2746049165725708,
      "learning_rate": 8e-05,
      "loss": 1.5197,
      "step": 2184
    },
    {
      "epoch": 0.4805894644231827,
      "grad_norm": 0.2719694972038269,
      "learning_rate": 8e-05,
      "loss": 1.6642,
      "step": 2185
    },
    {
      "epoch": 0.480809413834818,
      "grad_norm": 0.286636084318161,
      "learning_rate": 8e-05,
      "loss": 1.6885,
      "step": 2186
    },
    {
      "epoch": 0.4810293632464533,
      "grad_norm": 0.3076469600200653,
      "learning_rate": 8e-05,
      "loss": 1.7681,
      "step": 2187
    },
    {
      "epoch": 0.4812493126580886,
      "grad_norm": 0.3075680434703827,
      "learning_rate": 8e-05,
      "loss": 1.8007,
      "step": 2188
    },
    {
      "epoch": 0.48146926206972396,
      "grad_norm": 0.2793465852737427,
      "learning_rate": 8e-05,
      "loss": 1.6552,
      "step": 2189
    },
    {
      "epoch": 0.4816892114813593,
      "grad_norm": 0.2886781692504883,
      "learning_rate": 8e-05,
      "loss": 1.7827,
      "step": 2190
    },
    {
      "epoch": 0.4819091608929946,
      "grad_norm": 0.27432283759117126,
      "learning_rate": 8e-05,
      "loss": 1.7724,
      "step": 2191
    },
    {
      "epoch": 0.48212911030462996,
      "grad_norm": 0.26780393719673157,
      "learning_rate": 8e-05,
      "loss": 1.4613,
      "step": 2192
    },
    {
      "epoch": 0.48234905971626524,
      "grad_norm": 0.27178165316581726,
      "learning_rate": 8e-05,
      "loss": 1.6439,
      "step": 2193
    },
    {
      "epoch": 0.48256900912790057,
      "grad_norm": 0.27651992440223694,
      "learning_rate": 8e-05,
      "loss": 1.6672,
      "step": 2194
    },
    {
      "epoch": 0.4827889585395359,
      "grad_norm": 0.26919931173324585,
      "learning_rate": 8e-05,
      "loss": 1.594,
      "step": 2195
    },
    {
      "epoch": 0.48300890795117124,
      "grad_norm": 0.267678439617157,
      "learning_rate": 8e-05,
      "loss": 1.5852,
      "step": 2196
    },
    {
      "epoch": 0.4832288573628066,
      "grad_norm": 0.2895921766757965,
      "learning_rate": 8e-05,
      "loss": 1.6456,
      "step": 2197
    },
    {
      "epoch": 0.48344880677444185,
      "grad_norm": 0.27052855491638184,
      "learning_rate": 8e-05,
      "loss": 1.5393,
      "step": 2198
    },
    {
      "epoch": 0.4836687561860772,
      "grad_norm": 0.2793048024177551,
      "learning_rate": 8e-05,
      "loss": 1.7006,
      "step": 2199
    },
    {
      "epoch": 0.4838887055977125,
      "grad_norm": 0.2838841676712036,
      "learning_rate": 8e-05,
      "loss": 1.6547,
      "step": 2200
    },
    {
      "epoch": 0.48410865500934785,
      "grad_norm": 0.29814717173576355,
      "learning_rate": 8e-05,
      "loss": 1.7392,
      "step": 2201
    },
    {
      "epoch": 0.4843286044209832,
      "grad_norm": 0.28823426365852356,
      "learning_rate": 8e-05,
      "loss": 1.6434,
      "step": 2202
    },
    {
      "epoch": 0.4845485538326185,
      "grad_norm": 0.2645476460456848,
      "learning_rate": 8e-05,
      "loss": 1.5663,
      "step": 2203
    },
    {
      "epoch": 0.4847685032442538,
      "grad_norm": 0.3046651780605316,
      "learning_rate": 8e-05,
      "loss": 1.5941,
      "step": 2204
    },
    {
      "epoch": 0.48498845265588914,
      "grad_norm": 0.28958046436309814,
      "learning_rate": 8e-05,
      "loss": 1.7348,
      "step": 2205
    },
    {
      "epoch": 0.48520840206752447,
      "grad_norm": 0.25703537464141846,
      "learning_rate": 8e-05,
      "loss": 1.5712,
      "step": 2206
    },
    {
      "epoch": 0.4854283514791598,
      "grad_norm": 0.2969980537891388,
      "learning_rate": 8e-05,
      "loss": 1.7966,
      "step": 2207
    },
    {
      "epoch": 0.48564830089079514,
      "grad_norm": 0.27638381719589233,
      "learning_rate": 8e-05,
      "loss": 1.6478,
      "step": 2208
    },
    {
      "epoch": 0.4858682503024304,
      "grad_norm": 0.283682644367218,
      "learning_rate": 8e-05,
      "loss": 1.6524,
      "step": 2209
    },
    {
      "epoch": 0.48608819971406575,
      "grad_norm": 0.2837259769439697,
      "learning_rate": 8e-05,
      "loss": 1.7302,
      "step": 2210
    },
    {
      "epoch": 0.4863081491257011,
      "grad_norm": 0.3042410612106323,
      "learning_rate": 8e-05,
      "loss": 1.5969,
      "step": 2211
    },
    {
      "epoch": 0.4865280985373364,
      "grad_norm": 0.2819627821445465,
      "learning_rate": 8e-05,
      "loss": 1.5621,
      "step": 2212
    },
    {
      "epoch": 0.48674804794897175,
      "grad_norm": 0.28049173951148987,
      "learning_rate": 8e-05,
      "loss": 1.6831,
      "step": 2213
    },
    {
      "epoch": 0.4869679973606071,
      "grad_norm": 0.29762500524520874,
      "learning_rate": 8e-05,
      "loss": 1.7883,
      "step": 2214
    },
    {
      "epoch": 0.48718794677224236,
      "grad_norm": 0.3022189140319824,
      "learning_rate": 8e-05,
      "loss": 1.5744,
      "step": 2215
    },
    {
      "epoch": 0.4874078961838777,
      "grad_norm": 0.28249025344848633,
      "learning_rate": 8e-05,
      "loss": 1.6514,
      "step": 2216
    },
    {
      "epoch": 0.48762784559551303,
      "grad_norm": 0.3398612141609192,
      "learning_rate": 8e-05,
      "loss": 1.4215,
      "step": 2217
    },
    {
      "epoch": 0.48784779500714837,
      "grad_norm": 0.28481197357177734,
      "learning_rate": 8e-05,
      "loss": 1.7065,
      "step": 2218
    },
    {
      "epoch": 0.4880677444187837,
      "grad_norm": 0.3076950013637543,
      "learning_rate": 8e-05,
      "loss": 1.7665,
      "step": 2219
    },
    {
      "epoch": 0.488287693830419,
      "grad_norm": 0.28533896803855896,
      "learning_rate": 8e-05,
      "loss": 1.6156,
      "step": 2220
    },
    {
      "epoch": 0.4885076432420543,
      "grad_norm": 0.2940129041671753,
      "learning_rate": 8e-05,
      "loss": 1.6957,
      "step": 2221
    },
    {
      "epoch": 0.48872759265368965,
      "grad_norm": 0.30342915654182434,
      "learning_rate": 8e-05,
      "loss": 1.6014,
      "step": 2222
    },
    {
      "epoch": 0.488947542065325,
      "grad_norm": 0.2851787507534027,
      "learning_rate": 8e-05,
      "loss": 1.8816,
      "step": 2223
    },
    {
      "epoch": 0.4891674914769603,
      "grad_norm": 0.3813328146934509,
      "learning_rate": 8e-05,
      "loss": 1.6608,
      "step": 2224
    },
    {
      "epoch": 0.48938744088859565,
      "grad_norm": 0.329373300075531,
      "learning_rate": 8e-05,
      "loss": 1.7455,
      "step": 2225
    },
    {
      "epoch": 0.4896073903002309,
      "grad_norm": 0.2751460373401642,
      "learning_rate": 8e-05,
      "loss": 1.6854,
      "step": 2226
    },
    {
      "epoch": 0.48982733971186626,
      "grad_norm": 0.283033549785614,
      "learning_rate": 8e-05,
      "loss": 1.7018,
      "step": 2227
    },
    {
      "epoch": 0.4900472891235016,
      "grad_norm": 0.2869894504547119,
      "learning_rate": 8e-05,
      "loss": 1.7062,
      "step": 2228
    },
    {
      "epoch": 0.49026723853513693,
      "grad_norm": 0.2895123064517975,
      "learning_rate": 8e-05,
      "loss": 1.625,
      "step": 2229
    },
    {
      "epoch": 0.49048718794677226,
      "grad_norm": 0.28288763761520386,
      "learning_rate": 8e-05,
      "loss": 1.6901,
      "step": 2230
    },
    {
      "epoch": 0.49070713735840754,
      "grad_norm": 0.28054291009902954,
      "learning_rate": 8e-05,
      "loss": 1.7782,
      "step": 2231
    },
    {
      "epoch": 0.4909270867700429,
      "grad_norm": 0.2727196216583252,
      "learning_rate": 8e-05,
      "loss": 1.531,
      "step": 2232
    },
    {
      "epoch": 0.4911470361816782,
      "grad_norm": 0.28192493319511414,
      "learning_rate": 8e-05,
      "loss": 1.7144,
      "step": 2233
    },
    {
      "epoch": 0.49136698559331354,
      "grad_norm": 0.28133469820022583,
      "learning_rate": 8e-05,
      "loss": 1.6448,
      "step": 2234
    },
    {
      "epoch": 0.4915869350049489,
      "grad_norm": 0.3175356686115265,
      "learning_rate": 8e-05,
      "loss": 1.648,
      "step": 2235
    },
    {
      "epoch": 0.4918068844165842,
      "grad_norm": 0.2990395128726959,
      "learning_rate": 8e-05,
      "loss": 1.6871,
      "step": 2236
    },
    {
      "epoch": 0.4920268338282195,
      "grad_norm": 0.3121372163295746,
      "learning_rate": 8e-05,
      "loss": 1.7671,
      "step": 2237
    },
    {
      "epoch": 0.4922467832398548,
      "grad_norm": 0.2772499918937683,
      "learning_rate": 8e-05,
      "loss": 1.674,
      "step": 2238
    },
    {
      "epoch": 0.49246673265149016,
      "grad_norm": 0.2990477383136749,
      "learning_rate": 8e-05,
      "loss": 1.8202,
      "step": 2239
    },
    {
      "epoch": 0.4926866820631255,
      "grad_norm": 0.28028279542922974,
      "learning_rate": 8e-05,
      "loss": 1.7255,
      "step": 2240
    },
    {
      "epoch": 0.4929066314747608,
      "grad_norm": 0.28883370757102966,
      "learning_rate": 8e-05,
      "loss": 1.6102,
      "step": 2241
    },
    {
      "epoch": 0.4931265808863961,
      "grad_norm": 0.2775261700153351,
      "learning_rate": 8e-05,
      "loss": 1.8572,
      "step": 2242
    },
    {
      "epoch": 0.49334653029803144,
      "grad_norm": 0.2821192443370819,
      "learning_rate": 8e-05,
      "loss": 1.8068,
      "step": 2243
    },
    {
      "epoch": 0.49356647970966677,
      "grad_norm": 0.29555544257164,
      "learning_rate": 8e-05,
      "loss": 1.6875,
      "step": 2244
    },
    {
      "epoch": 0.4937864291213021,
      "grad_norm": 0.28827783465385437,
      "learning_rate": 8e-05,
      "loss": 1.7123,
      "step": 2245
    },
    {
      "epoch": 0.49400637853293744,
      "grad_norm": 0.27672290802001953,
      "learning_rate": 8e-05,
      "loss": 1.4739,
      "step": 2246
    },
    {
      "epoch": 0.4942263279445728,
      "grad_norm": 0.27185946702957153,
      "learning_rate": 8e-05,
      "loss": 1.7748,
      "step": 2247
    },
    {
      "epoch": 0.49444627735620805,
      "grad_norm": 0.2972213327884674,
      "learning_rate": 8e-05,
      "loss": 1.7122,
      "step": 2248
    },
    {
      "epoch": 0.4946662267678434,
      "grad_norm": 0.30817538499832153,
      "learning_rate": 8e-05,
      "loss": 1.926,
      "step": 2249
    },
    {
      "epoch": 0.4948861761794787,
      "grad_norm": 0.2821509838104248,
      "learning_rate": 8e-05,
      "loss": 1.7608,
      "step": 2250
    },
    {
      "epoch": 0.49510612559111405,
      "grad_norm": 0.29807963967323303,
      "learning_rate": 8e-05,
      "loss": 1.8233,
      "step": 2251
    },
    {
      "epoch": 0.4953260750027494,
      "grad_norm": 0.29549431800842285,
      "learning_rate": 8e-05,
      "loss": 1.7963,
      "step": 2252
    },
    {
      "epoch": 0.49554602441438467,
      "grad_norm": 0.3025868535041809,
      "learning_rate": 8e-05,
      "loss": 1.7597,
      "step": 2253
    },
    {
      "epoch": 0.49576597382602,
      "grad_norm": 0.30950862169265747,
      "learning_rate": 8e-05,
      "loss": 1.8901,
      "step": 2254
    },
    {
      "epoch": 0.49598592323765534,
      "grad_norm": 0.3357299864292145,
      "learning_rate": 8e-05,
      "loss": 1.7692,
      "step": 2255
    },
    {
      "epoch": 0.49620587264929067,
      "grad_norm": 0.2873973548412323,
      "learning_rate": 8e-05,
      "loss": 1.6934,
      "step": 2256
    },
    {
      "epoch": 0.496425822060926,
      "grad_norm": 0.2997465431690216,
      "learning_rate": 8e-05,
      "loss": 1.5939,
      "step": 2257
    },
    {
      "epoch": 0.49664577147256134,
      "grad_norm": 0.269217312335968,
      "learning_rate": 8e-05,
      "loss": 1.5687,
      "step": 2258
    },
    {
      "epoch": 0.4968657208841966,
      "grad_norm": 0.27386826276779175,
      "learning_rate": 8e-05,
      "loss": 1.6858,
      "step": 2259
    },
    {
      "epoch": 0.49708567029583195,
      "grad_norm": 0.2911466658115387,
      "learning_rate": 8e-05,
      "loss": 1.701,
      "step": 2260
    },
    {
      "epoch": 0.4973056197074673,
      "grad_norm": 0.2837962508201599,
      "learning_rate": 8e-05,
      "loss": 1.6886,
      "step": 2261
    },
    {
      "epoch": 0.4975255691191026,
      "grad_norm": 0.3071229159832001,
      "learning_rate": 8e-05,
      "loss": 1.73,
      "step": 2262
    },
    {
      "epoch": 0.49774551853073795,
      "grad_norm": 0.303252249956131,
      "learning_rate": 8e-05,
      "loss": 1.7693,
      "step": 2263
    },
    {
      "epoch": 0.49796546794237323,
      "grad_norm": 0.2802221179008484,
      "learning_rate": 8e-05,
      "loss": 1.6394,
      "step": 2264
    },
    {
      "epoch": 0.49818541735400856,
      "grad_norm": 0.28856000304222107,
      "learning_rate": 8e-05,
      "loss": 1.6035,
      "step": 2265
    },
    {
      "epoch": 0.4984053667656439,
      "grad_norm": 0.28943875432014465,
      "learning_rate": 8e-05,
      "loss": 1.7989,
      "step": 2266
    },
    {
      "epoch": 0.49862531617727923,
      "grad_norm": 0.26969149708747864,
      "learning_rate": 8e-05,
      "loss": 1.6607,
      "step": 2267
    },
    {
      "epoch": 0.49884526558891457,
      "grad_norm": 0.311819851398468,
      "learning_rate": 8e-05,
      "loss": 1.7912,
      "step": 2268
    },
    {
      "epoch": 0.4990652150005499,
      "grad_norm": 0.296274334192276,
      "learning_rate": 8e-05,
      "loss": 1.6953,
      "step": 2269
    },
    {
      "epoch": 0.4992851644121852,
      "grad_norm": 0.26551195979118347,
      "learning_rate": 8e-05,
      "loss": 1.3549,
      "step": 2270
    },
    {
      "epoch": 0.4995051138238205,
      "grad_norm": 0.28540030121803284,
      "learning_rate": 8e-05,
      "loss": 1.7142,
      "step": 2271
    },
    {
      "epoch": 0.49972506323545585,
      "grad_norm": 0.2834233045578003,
      "learning_rate": 8e-05,
      "loss": 1.7293,
      "step": 2272
    },
    {
      "epoch": 0.4999450126470912,
      "grad_norm": 0.34650975465774536,
      "learning_rate": 8e-05,
      "loss": 1.7652,
      "step": 2273
    },
    {
      "epoch": 0.5001649620587265,
      "grad_norm": 0.2988453507423401,
      "learning_rate": 8e-05,
      "loss": 1.8895,
      "step": 2274
    },
    {
      "epoch": 0.5003849114703618,
      "grad_norm": 0.2912983000278473,
      "learning_rate": 8e-05,
      "loss": 1.6431,
      "step": 2275
    },
    {
      "epoch": 0.5006048608819972,
      "grad_norm": 0.28406545519828796,
      "learning_rate": 8e-05,
      "loss": 1.7805,
      "step": 2276
    },
    {
      "epoch": 0.5008248102936325,
      "grad_norm": 0.2748315632343292,
      "learning_rate": 8e-05,
      "loss": 1.5514,
      "step": 2277
    },
    {
      "epoch": 0.5010447597052677,
      "grad_norm": 0.3016912341117859,
      "learning_rate": 8e-05,
      "loss": 1.6271,
      "step": 2278
    },
    {
      "epoch": 0.5012647091169031,
      "grad_norm": 0.3006996512413025,
      "learning_rate": 8e-05,
      "loss": 1.7195,
      "step": 2279
    },
    {
      "epoch": 0.5014846585285384,
      "grad_norm": 0.27598950266838074,
      "learning_rate": 8e-05,
      "loss": 1.4975,
      "step": 2280
    },
    {
      "epoch": 0.5017046079401738,
      "grad_norm": 0.2810399830341339,
      "learning_rate": 8e-05,
      "loss": 1.6498,
      "step": 2281
    },
    {
      "epoch": 0.5019245573518091,
      "grad_norm": 0.3018679916858673,
      "learning_rate": 8e-05,
      "loss": 1.6711,
      "step": 2282
    },
    {
      "epoch": 0.5021445067634444,
      "grad_norm": 0.2889658510684967,
      "learning_rate": 8e-05,
      "loss": 1.7318,
      "step": 2283
    },
    {
      "epoch": 0.5023644561750797,
      "grad_norm": 0.28475597500801086,
      "learning_rate": 8e-05,
      "loss": 1.7573,
      "step": 2284
    },
    {
      "epoch": 0.502584405586715,
      "grad_norm": 0.3001987338066101,
      "learning_rate": 8e-05,
      "loss": 1.8463,
      "step": 2285
    },
    {
      "epoch": 0.5028043549983504,
      "grad_norm": 0.273773193359375,
      "learning_rate": 8e-05,
      "loss": 1.6465,
      "step": 2286
    },
    {
      "epoch": 0.5030243044099857,
      "grad_norm": 0.34727615118026733,
      "learning_rate": 8e-05,
      "loss": 1.9431,
      "step": 2287
    },
    {
      "epoch": 0.5032442538216211,
      "grad_norm": 0.2818615138530731,
      "learning_rate": 8e-05,
      "loss": 1.6113,
      "step": 2288
    },
    {
      "epoch": 0.5034642032332564,
      "grad_norm": 0.26619333028793335,
      "learning_rate": 8e-05,
      "loss": 1.5167,
      "step": 2289
    },
    {
      "epoch": 0.5036841526448916,
      "grad_norm": 0.2748722434043884,
      "learning_rate": 8e-05,
      "loss": 1.6384,
      "step": 2290
    },
    {
      "epoch": 0.503904102056527,
      "grad_norm": 0.29873546957969666,
      "learning_rate": 8e-05,
      "loss": 1.7309,
      "step": 2291
    },
    {
      "epoch": 0.5041240514681623,
      "grad_norm": 0.28378361463546753,
      "learning_rate": 8e-05,
      "loss": 1.6788,
      "step": 2292
    },
    {
      "epoch": 0.5043440008797977,
      "grad_norm": 0.28786107897758484,
      "learning_rate": 8e-05,
      "loss": 1.7281,
      "step": 2293
    },
    {
      "epoch": 0.504563950291433,
      "grad_norm": 0.2831546366214752,
      "learning_rate": 8e-05,
      "loss": 1.7644,
      "step": 2294
    },
    {
      "epoch": 0.5047838997030683,
      "grad_norm": 0.28964316844940186,
      "learning_rate": 8e-05,
      "loss": 1.6409,
      "step": 2295
    },
    {
      "epoch": 0.5050038491147036,
      "grad_norm": 0.2778918743133545,
      "learning_rate": 8e-05,
      "loss": 1.6078,
      "step": 2296
    },
    {
      "epoch": 0.5052237985263389,
      "grad_norm": 0.2749491035938263,
      "learning_rate": 8e-05,
      "loss": 1.668,
      "step": 2297
    },
    {
      "epoch": 0.5054437479379743,
      "grad_norm": 0.2856389880180359,
      "learning_rate": 8e-05,
      "loss": 1.7052,
      "step": 2298
    },
    {
      "epoch": 0.5056636973496096,
      "grad_norm": 0.28082379698753357,
      "learning_rate": 8e-05,
      "loss": 1.6627,
      "step": 2299
    },
    {
      "epoch": 0.5058836467612449,
      "grad_norm": 0.27894240617752075,
      "learning_rate": 8e-05,
      "loss": 1.6408,
      "step": 2300
    },
    {
      "epoch": 0.5061035961728803,
      "grad_norm": 0.31712663173675537,
      "learning_rate": 8e-05,
      "loss": 1.6148,
      "step": 2301
    },
    {
      "epoch": 0.5063235455845155,
      "grad_norm": 0.28600454330444336,
      "learning_rate": 8e-05,
      "loss": 1.7695,
      "step": 2302
    },
    {
      "epoch": 0.5065434949961509,
      "grad_norm": 0.3285694718360901,
      "learning_rate": 8e-05,
      "loss": 1.7882,
      "step": 2303
    },
    {
      "epoch": 0.5067634444077862,
      "grad_norm": 0.27823877334594727,
      "learning_rate": 8e-05,
      "loss": 1.4648,
      "step": 2304
    },
    {
      "epoch": 0.5069833938194215,
      "grad_norm": 0.3157597482204437,
      "learning_rate": 8e-05,
      "loss": 1.8441,
      "step": 2305
    },
    {
      "epoch": 0.5072033432310569,
      "grad_norm": 0.2913108170032501,
      "learning_rate": 8e-05,
      "loss": 1.7584,
      "step": 2306
    },
    {
      "epoch": 0.5074232926426921,
      "grad_norm": 0.28753626346588135,
      "learning_rate": 8e-05,
      "loss": 1.7045,
      "step": 2307
    },
    {
      "epoch": 0.5076432420543275,
      "grad_norm": 0.2981377840042114,
      "learning_rate": 8e-05,
      "loss": 1.54,
      "step": 2308
    },
    {
      "epoch": 0.5078631914659628,
      "grad_norm": 0.2911180853843689,
      "learning_rate": 8e-05,
      "loss": 1.8266,
      "step": 2309
    },
    {
      "epoch": 0.5080831408775982,
      "grad_norm": 0.2862488031387329,
      "learning_rate": 8e-05,
      "loss": 1.7503,
      "step": 2310
    },
    {
      "epoch": 0.5083030902892335,
      "grad_norm": 0.3015568256378174,
      "learning_rate": 8e-05,
      "loss": 1.8209,
      "step": 2311
    },
    {
      "epoch": 0.5085230397008688,
      "grad_norm": 0.27230823040008545,
      "learning_rate": 8e-05,
      "loss": 1.7977,
      "step": 2312
    },
    {
      "epoch": 0.5087429891125042,
      "grad_norm": 0.27871981263160706,
      "learning_rate": 8e-05,
      "loss": 1.5971,
      "step": 2313
    },
    {
      "epoch": 0.5089629385241394,
      "grad_norm": 0.2884436249732971,
      "learning_rate": 8e-05,
      "loss": 1.7726,
      "step": 2314
    },
    {
      "epoch": 0.5091828879357748,
      "grad_norm": 0.2778714895248413,
      "learning_rate": 8e-05,
      "loss": 1.7104,
      "step": 2315
    },
    {
      "epoch": 0.5094028373474101,
      "grad_norm": 0.30202385783195496,
      "learning_rate": 8e-05,
      "loss": 1.658,
      "step": 2316
    },
    {
      "epoch": 0.5096227867590454,
      "grad_norm": 0.2785106599330902,
      "learning_rate": 8e-05,
      "loss": 1.717,
      "step": 2317
    },
    {
      "epoch": 0.5098427361706808,
      "grad_norm": 0.28846096992492676,
      "learning_rate": 8e-05,
      "loss": 1.7456,
      "step": 2318
    },
    {
      "epoch": 0.510062685582316,
      "grad_norm": 0.27753984928131104,
      "learning_rate": 8e-05,
      "loss": 1.6569,
      "step": 2319
    },
    {
      "epoch": 0.5102826349939514,
      "grad_norm": 0.2834784686565399,
      "learning_rate": 8e-05,
      "loss": 1.6348,
      "step": 2320
    },
    {
      "epoch": 0.5105025844055867,
      "grad_norm": 0.27789169549942017,
      "learning_rate": 8e-05,
      "loss": 1.5586,
      "step": 2321
    },
    {
      "epoch": 0.510722533817222,
      "grad_norm": 0.28466710448265076,
      "learning_rate": 8e-05,
      "loss": 1.6793,
      "step": 2322
    },
    {
      "epoch": 0.5109424832288574,
      "grad_norm": 0.2759189009666443,
      "learning_rate": 8e-05,
      "loss": 1.5311,
      "step": 2323
    },
    {
      "epoch": 0.5111624326404927,
      "grad_norm": 0.2931334674358368,
      "learning_rate": 8e-05,
      "loss": 1.7258,
      "step": 2324
    },
    {
      "epoch": 0.511382382052128,
      "grad_norm": 0.2740546464920044,
      "learning_rate": 8e-05,
      "loss": 1.558,
      "step": 2325
    },
    {
      "epoch": 0.5116023314637633,
      "grad_norm": 0.29584407806396484,
      "learning_rate": 8e-05,
      "loss": 1.7777,
      "step": 2326
    },
    {
      "epoch": 0.5118222808753986,
      "grad_norm": 0.2948019504547119,
      "learning_rate": 8e-05,
      "loss": 1.7722,
      "step": 2327
    },
    {
      "epoch": 0.512042230287034,
      "grad_norm": 0.27409225702285767,
      "learning_rate": 8e-05,
      "loss": 1.6524,
      "step": 2328
    },
    {
      "epoch": 0.5122621796986693,
      "grad_norm": 0.26278048753738403,
      "learning_rate": 8e-05,
      "loss": 1.5945,
      "step": 2329
    },
    {
      "epoch": 0.5124821291103047,
      "grad_norm": 0.29483261704444885,
      "learning_rate": 8e-05,
      "loss": 1.8132,
      "step": 2330
    },
    {
      "epoch": 0.5127020785219399,
      "grad_norm": 0.27037349343299866,
      "learning_rate": 8e-05,
      "loss": 1.5657,
      "step": 2331
    },
    {
      "epoch": 0.5129220279335753,
      "grad_norm": 0.2826361060142517,
      "learning_rate": 8e-05,
      "loss": 1.6955,
      "step": 2332
    },
    {
      "epoch": 0.5131419773452106,
      "grad_norm": 0.2957696318626404,
      "learning_rate": 8e-05,
      "loss": 1.814,
      "step": 2333
    },
    {
      "epoch": 0.5133619267568459,
      "grad_norm": 0.2752826511859894,
      "learning_rate": 8e-05,
      "loss": 1.636,
      "step": 2334
    },
    {
      "epoch": 0.5135818761684813,
      "grad_norm": 0.28523313999176025,
      "learning_rate": 8e-05,
      "loss": 1.5444,
      "step": 2335
    },
    {
      "epoch": 0.5138018255801166,
      "grad_norm": 0.286304235458374,
      "learning_rate": 8e-05,
      "loss": 1.4665,
      "step": 2336
    },
    {
      "epoch": 0.514021774991752,
      "grad_norm": 0.28738734126091003,
      "learning_rate": 8e-05,
      "loss": 1.6802,
      "step": 2337
    },
    {
      "epoch": 0.5142417244033872,
      "grad_norm": 0.2669237554073334,
      "learning_rate": 8e-05,
      "loss": 1.6011,
      "step": 2338
    },
    {
      "epoch": 0.5144616738150225,
      "grad_norm": 0.274325430393219,
      "learning_rate": 8e-05,
      "loss": 1.6248,
      "step": 2339
    },
    {
      "epoch": 0.5146816232266579,
      "grad_norm": 0.2798522710800171,
      "learning_rate": 8e-05,
      "loss": 1.7636,
      "step": 2340
    },
    {
      "epoch": 0.5149015726382932,
      "grad_norm": 0.27266305685043335,
      "learning_rate": 8e-05,
      "loss": 1.6434,
      "step": 2341
    },
    {
      "epoch": 0.5151215220499286,
      "grad_norm": 0.2740791440010071,
      "learning_rate": 8e-05,
      "loss": 1.7084,
      "step": 2342
    },
    {
      "epoch": 0.5153414714615638,
      "grad_norm": 0.28098320960998535,
      "learning_rate": 8e-05,
      "loss": 1.6356,
      "step": 2343
    },
    {
      "epoch": 0.5155614208731991,
      "grad_norm": 0.2760515809059143,
      "learning_rate": 8e-05,
      "loss": 1.661,
      "step": 2344
    },
    {
      "epoch": 0.5157813702848345,
      "grad_norm": 0.27894532680511475,
      "learning_rate": 8e-05,
      "loss": 1.6794,
      "step": 2345
    },
    {
      "epoch": 0.5160013196964698,
      "grad_norm": 0.2972679138183594,
      "learning_rate": 8e-05,
      "loss": 1.6943,
      "step": 2346
    },
    {
      "epoch": 0.5162212691081052,
      "grad_norm": 0.3100125789642334,
      "learning_rate": 8e-05,
      "loss": 1.7214,
      "step": 2347
    },
    {
      "epoch": 0.5164412185197405,
      "grad_norm": 0.2743578255176544,
      "learning_rate": 8e-05,
      "loss": 1.7021,
      "step": 2348
    },
    {
      "epoch": 0.5166611679313757,
      "grad_norm": 0.29266777634620667,
      "learning_rate": 8e-05,
      "loss": 1.7585,
      "step": 2349
    },
    {
      "epoch": 0.5168811173430111,
      "grad_norm": 0.2791600227355957,
      "learning_rate": 8e-05,
      "loss": 1.7012,
      "step": 2350
    },
    {
      "epoch": 0.5171010667546464,
      "grad_norm": 0.28535401821136475,
      "learning_rate": 8e-05,
      "loss": 1.6695,
      "step": 2351
    },
    {
      "epoch": 0.5173210161662818,
      "grad_norm": 0.2860865592956543,
      "learning_rate": 8e-05,
      "loss": 1.7419,
      "step": 2352
    },
    {
      "epoch": 0.5175409655779171,
      "grad_norm": 0.27693790197372437,
      "learning_rate": 8e-05,
      "loss": 1.5459,
      "step": 2353
    },
    {
      "epoch": 0.5177609149895525,
      "grad_norm": 0.2858433723449707,
      "learning_rate": 8e-05,
      "loss": 1.799,
      "step": 2354
    },
    {
      "epoch": 0.5179808644011877,
      "grad_norm": 0.30761632323265076,
      "learning_rate": 8e-05,
      "loss": 1.5971,
      "step": 2355
    },
    {
      "epoch": 0.518200813812823,
      "grad_norm": 0.2943046987056732,
      "learning_rate": 8e-05,
      "loss": 1.7399,
      "step": 2356
    },
    {
      "epoch": 0.5184207632244584,
      "grad_norm": 0.2748922109603882,
      "learning_rate": 8e-05,
      "loss": 1.7202,
      "step": 2357
    },
    {
      "epoch": 0.5186407126360937,
      "grad_norm": 0.2818071246147156,
      "learning_rate": 8e-05,
      "loss": 1.5918,
      "step": 2358
    },
    {
      "epoch": 0.5188606620477291,
      "grad_norm": 0.28235137462615967,
      "learning_rate": 8e-05,
      "loss": 1.5728,
      "step": 2359
    },
    {
      "epoch": 0.5190806114593643,
      "grad_norm": 0.27833959460258484,
      "learning_rate": 8e-05,
      "loss": 1.5966,
      "step": 2360
    },
    {
      "epoch": 0.5193005608709996,
      "grad_norm": 0.2731468677520752,
      "learning_rate": 8e-05,
      "loss": 1.6518,
      "step": 2361
    },
    {
      "epoch": 0.519520510282635,
      "grad_norm": 0.2777821719646454,
      "learning_rate": 8e-05,
      "loss": 1.6885,
      "step": 2362
    },
    {
      "epoch": 0.5197404596942703,
      "grad_norm": 0.2685951590538025,
      "learning_rate": 8e-05,
      "loss": 1.6315,
      "step": 2363
    },
    {
      "epoch": 0.5199604091059057,
      "grad_norm": 0.3087875545024872,
      "learning_rate": 8e-05,
      "loss": 1.7525,
      "step": 2364
    },
    {
      "epoch": 0.520180358517541,
      "grad_norm": 0.2693195044994354,
      "learning_rate": 8e-05,
      "loss": 1.5993,
      "step": 2365
    },
    {
      "epoch": 0.5204003079291762,
      "grad_norm": 0.2832968235015869,
      "learning_rate": 8e-05,
      "loss": 1.5798,
      "step": 2366
    },
    {
      "epoch": 0.5206202573408116,
      "grad_norm": 0.2873738706111908,
      "learning_rate": 8e-05,
      "loss": 1.7373,
      "step": 2367
    },
    {
      "epoch": 0.5208402067524469,
      "grad_norm": 0.2888682782649994,
      "learning_rate": 8e-05,
      "loss": 1.6995,
      "step": 2368
    },
    {
      "epoch": 0.5210601561640823,
      "grad_norm": 0.2788809537887573,
      "learning_rate": 8e-05,
      "loss": 1.5928,
      "step": 2369
    },
    {
      "epoch": 0.5212801055757176,
      "grad_norm": 0.28021880984306335,
      "learning_rate": 8e-05,
      "loss": 1.6692,
      "step": 2370
    },
    {
      "epoch": 0.5215000549873529,
      "grad_norm": 0.3041718006134033,
      "learning_rate": 8e-05,
      "loss": 1.7378,
      "step": 2371
    },
    {
      "epoch": 0.5217200043989882,
      "grad_norm": 0.2774748206138611,
      "learning_rate": 8e-05,
      "loss": 1.5802,
      "step": 2372
    },
    {
      "epoch": 0.5219399538106235,
      "grad_norm": 0.2876451015472412,
      "learning_rate": 8e-05,
      "loss": 1.8057,
      "step": 2373
    },
    {
      "epoch": 0.5221599032222589,
      "grad_norm": 0.2740166485309601,
      "learning_rate": 8e-05,
      "loss": 1.6694,
      "step": 2374
    },
    {
      "epoch": 0.5223798526338942,
      "grad_norm": 0.288555771112442,
      "learning_rate": 8e-05,
      "loss": 1.792,
      "step": 2375
    },
    {
      "epoch": 0.5225998020455296,
      "grad_norm": 0.2777664065361023,
      "learning_rate": 8e-05,
      "loss": 1.6781,
      "step": 2376
    },
    {
      "epoch": 0.5228197514571649,
      "grad_norm": 0.27108079195022583,
      "learning_rate": 8e-05,
      "loss": 1.4881,
      "step": 2377
    },
    {
      "epoch": 0.5230397008688001,
      "grad_norm": 0.2909669578075409,
      "learning_rate": 8e-05,
      "loss": 1.7174,
      "step": 2378
    },
    {
      "epoch": 0.5232596502804355,
      "grad_norm": 0.2978494167327881,
      "learning_rate": 8e-05,
      "loss": 1.8641,
      "step": 2379
    },
    {
      "epoch": 0.5234795996920708,
      "grad_norm": 0.2649437487125397,
      "learning_rate": 8e-05,
      "loss": 1.473,
      "step": 2380
    },
    {
      "epoch": 0.5236995491037062,
      "grad_norm": 0.28939372301101685,
      "learning_rate": 8e-05,
      "loss": 1.5567,
      "step": 2381
    },
    {
      "epoch": 0.5239194985153415,
      "grad_norm": 0.2740820646286011,
      "learning_rate": 8e-05,
      "loss": 1.6321,
      "step": 2382
    },
    {
      "epoch": 0.5241394479269768,
      "grad_norm": 0.28426122665405273,
      "learning_rate": 8e-05,
      "loss": 1.5952,
      "step": 2383
    },
    {
      "epoch": 0.5243593973386121,
      "grad_norm": 0.28176257014274597,
      "learning_rate": 8e-05,
      "loss": 1.6231,
      "step": 2384
    },
    {
      "epoch": 0.5245793467502474,
      "grad_norm": 0.29681360721588135,
      "learning_rate": 8e-05,
      "loss": 1.8203,
      "step": 2385
    },
    {
      "epoch": 0.5247992961618828,
      "grad_norm": 0.272658109664917,
      "learning_rate": 8e-05,
      "loss": 1.6942,
      "step": 2386
    },
    {
      "epoch": 0.5250192455735181,
      "grad_norm": 0.27786141633987427,
      "learning_rate": 8e-05,
      "loss": 1.6081,
      "step": 2387
    },
    {
      "epoch": 0.5252391949851534,
      "grad_norm": 0.2938309609889984,
      "learning_rate": 8e-05,
      "loss": 1.7454,
      "step": 2388
    },
    {
      "epoch": 0.5254591443967888,
      "grad_norm": 0.2710343599319458,
      "learning_rate": 8e-05,
      "loss": 1.6391,
      "step": 2389
    },
    {
      "epoch": 0.525679093808424,
      "grad_norm": 0.2757870554924011,
      "learning_rate": 8e-05,
      "loss": 1.6526,
      "step": 2390
    },
    {
      "epoch": 0.5258990432200594,
      "grad_norm": 0.2581859827041626,
      "learning_rate": 8e-05,
      "loss": 1.4344,
      "step": 2391
    },
    {
      "epoch": 0.5261189926316947,
      "grad_norm": 0.2732166647911072,
      "learning_rate": 8e-05,
      "loss": 1.5987,
      "step": 2392
    },
    {
      "epoch": 0.52633894204333,
      "grad_norm": 0.2859753370285034,
      "learning_rate": 8e-05,
      "loss": 1.6654,
      "step": 2393
    },
    {
      "epoch": 0.5265588914549654,
      "grad_norm": 0.2680748701095581,
      "learning_rate": 8e-05,
      "loss": 1.5764,
      "step": 2394
    },
    {
      "epoch": 0.5267788408666007,
      "grad_norm": 0.2866816818714142,
      "learning_rate": 8e-05,
      "loss": 1.7725,
      "step": 2395
    },
    {
      "epoch": 0.526998790278236,
      "grad_norm": 0.27792397141456604,
      "learning_rate": 8e-05,
      "loss": 1.5932,
      "step": 2396
    },
    {
      "epoch": 0.5272187396898713,
      "grad_norm": 0.26985982060432434,
      "learning_rate": 8e-05,
      "loss": 1.5955,
      "step": 2397
    },
    {
      "epoch": 0.5274386891015067,
      "grad_norm": 0.28183454275131226,
      "learning_rate": 8e-05,
      "loss": 1.553,
      "step": 2398
    },
    {
      "epoch": 0.527658638513142,
      "grad_norm": 0.29282763600349426,
      "learning_rate": 8e-05,
      "loss": 1.7858,
      "step": 2399
    },
    {
      "epoch": 0.5278785879247773,
      "grad_norm": 0.30619367957115173,
      "learning_rate": 8e-05,
      "loss": 1.718,
      "step": 2400
    },
    {
      "epoch": 0.5280985373364127,
      "grad_norm": 0.26707130670547485,
      "learning_rate": 8e-05,
      "loss": 1.5899,
      "step": 2401
    },
    {
      "epoch": 0.5283184867480479,
      "grad_norm": 0.3182383179664612,
      "learning_rate": 8e-05,
      "loss": 1.7268,
      "step": 2402
    },
    {
      "epoch": 0.5285384361596833,
      "grad_norm": 0.3178313374519348,
      "learning_rate": 8e-05,
      "loss": 1.7282,
      "step": 2403
    },
    {
      "epoch": 0.5287583855713186,
      "grad_norm": 0.26504799723625183,
      "learning_rate": 8e-05,
      "loss": 1.6046,
      "step": 2404
    },
    {
      "epoch": 0.5289783349829539,
      "grad_norm": 0.2749512195587158,
      "learning_rate": 8e-05,
      "loss": 1.6251,
      "step": 2405
    },
    {
      "epoch": 0.5291982843945893,
      "grad_norm": 0.27312803268432617,
      "learning_rate": 8e-05,
      "loss": 1.746,
      "step": 2406
    },
    {
      "epoch": 0.5294182338062245,
      "grad_norm": 0.26339027285575867,
      "learning_rate": 8e-05,
      "loss": 1.5214,
      "step": 2407
    },
    {
      "epoch": 0.5296381832178599,
      "grad_norm": 0.28254935145378113,
      "learning_rate": 8e-05,
      "loss": 1.629,
      "step": 2408
    },
    {
      "epoch": 0.5298581326294952,
      "grad_norm": 0.2761283218860626,
      "learning_rate": 8e-05,
      "loss": 1.7202,
      "step": 2409
    },
    {
      "epoch": 0.5300780820411305,
      "grad_norm": 0.27570095658302307,
      "learning_rate": 8e-05,
      "loss": 1.7042,
      "step": 2410
    },
    {
      "epoch": 0.5302980314527659,
      "grad_norm": 0.2886349856853485,
      "learning_rate": 8e-05,
      "loss": 1.7923,
      "step": 2411
    },
    {
      "epoch": 0.5305179808644012,
      "grad_norm": 0.29611504077911377,
      "learning_rate": 8e-05,
      "loss": 1.7219,
      "step": 2412
    },
    {
      "epoch": 0.5307379302760366,
      "grad_norm": 0.28122174739837646,
      "learning_rate": 8e-05,
      "loss": 1.6888,
      "step": 2413
    },
    {
      "epoch": 0.5309578796876718,
      "grad_norm": 0.2690391540527344,
      "learning_rate": 8e-05,
      "loss": 1.2745,
      "step": 2414
    },
    {
      "epoch": 0.5311778290993071,
      "grad_norm": 0.2676471173763275,
      "learning_rate": 8e-05,
      "loss": 1.6422,
      "step": 2415
    },
    {
      "epoch": 0.5313977785109425,
      "grad_norm": 0.2947712540626526,
      "learning_rate": 8e-05,
      "loss": 1.654,
      "step": 2416
    },
    {
      "epoch": 0.5316177279225778,
      "grad_norm": 0.27766644954681396,
      "learning_rate": 8e-05,
      "loss": 1.6208,
      "step": 2417
    },
    {
      "epoch": 0.5318376773342132,
      "grad_norm": 0.28579944372177124,
      "learning_rate": 8e-05,
      "loss": 1.8171,
      "step": 2418
    },
    {
      "epoch": 0.5320576267458484,
      "grad_norm": 0.2734217345714569,
      "learning_rate": 8e-05,
      "loss": 1.658,
      "step": 2419
    },
    {
      "epoch": 0.5322775761574838,
      "grad_norm": 0.28343021869659424,
      "learning_rate": 8e-05,
      "loss": 1.6291,
      "step": 2420
    },
    {
      "epoch": 0.5324975255691191,
      "grad_norm": 0.2881801128387451,
      "learning_rate": 8e-05,
      "loss": 1.5212,
      "step": 2421
    },
    {
      "epoch": 0.5327174749807544,
      "grad_norm": 0.27267688512802124,
      "learning_rate": 8e-05,
      "loss": 1.6599,
      "step": 2422
    },
    {
      "epoch": 0.5329374243923898,
      "grad_norm": 0.29100489616394043,
      "learning_rate": 8e-05,
      "loss": 1.6636,
      "step": 2423
    },
    {
      "epoch": 0.5331573738040251,
      "grad_norm": 0.301812082529068,
      "learning_rate": 8e-05,
      "loss": 1.9097,
      "step": 2424
    },
    {
      "epoch": 0.5333773232156604,
      "grad_norm": 0.2864093482494354,
      "learning_rate": 8e-05,
      "loss": 1.6535,
      "step": 2425
    },
    {
      "epoch": 0.5335972726272957,
      "grad_norm": 0.28721320629119873,
      "learning_rate": 8e-05,
      "loss": 1.7307,
      "step": 2426
    },
    {
      "epoch": 0.533817222038931,
      "grad_norm": 0.3100323975086212,
      "learning_rate": 8e-05,
      "loss": 1.7155,
      "step": 2427
    },
    {
      "epoch": 0.5340371714505664,
      "grad_norm": 0.2595236301422119,
      "learning_rate": 8e-05,
      "loss": 1.4525,
      "step": 2428
    },
    {
      "epoch": 0.5342571208622017,
      "grad_norm": 0.27981269359588623,
      "learning_rate": 8e-05,
      "loss": 1.6821,
      "step": 2429
    },
    {
      "epoch": 0.5344770702738371,
      "grad_norm": 0.28523892164230347,
      "learning_rate": 8e-05,
      "loss": 1.6213,
      "step": 2430
    },
    {
      "epoch": 0.5346970196854723,
      "grad_norm": 0.2951820194721222,
      "learning_rate": 8e-05,
      "loss": 1.7798,
      "step": 2431
    },
    {
      "epoch": 0.5349169690971076,
      "grad_norm": 0.27744752168655396,
      "learning_rate": 8e-05,
      "loss": 1.664,
      "step": 2432
    },
    {
      "epoch": 0.535136918508743,
      "grad_norm": 0.2700327932834625,
      "learning_rate": 8e-05,
      "loss": 1.6476,
      "step": 2433
    },
    {
      "epoch": 0.5353568679203783,
      "grad_norm": 0.3043116331100464,
      "learning_rate": 8e-05,
      "loss": 1.8377,
      "step": 2434
    },
    {
      "epoch": 0.5355768173320137,
      "grad_norm": 0.2886519730091095,
      "learning_rate": 8e-05,
      "loss": 1.7098,
      "step": 2435
    },
    {
      "epoch": 0.535796766743649,
      "grad_norm": 0.28121626377105713,
      "learning_rate": 8e-05,
      "loss": 1.5902,
      "step": 2436
    },
    {
      "epoch": 0.5360167161552842,
      "grad_norm": 0.28657859563827515,
      "learning_rate": 8e-05,
      "loss": 1.6769,
      "step": 2437
    },
    {
      "epoch": 0.5362366655669196,
      "grad_norm": 0.3111754059791565,
      "learning_rate": 8e-05,
      "loss": 1.8352,
      "step": 2438
    },
    {
      "epoch": 0.5364566149785549,
      "grad_norm": 0.27172762155532837,
      "learning_rate": 8e-05,
      "loss": 1.5897,
      "step": 2439
    },
    {
      "epoch": 0.5366765643901903,
      "grad_norm": 0.28469017148017883,
      "learning_rate": 8e-05,
      "loss": 1.6976,
      "step": 2440
    },
    {
      "epoch": 0.5368965138018256,
      "grad_norm": 0.29801180958747864,
      "learning_rate": 8e-05,
      "loss": 1.7811,
      "step": 2441
    },
    {
      "epoch": 0.5371164632134608,
      "grad_norm": 0.2860267758369446,
      "learning_rate": 8e-05,
      "loss": 1.7026,
      "step": 2442
    },
    {
      "epoch": 0.5373364126250962,
      "grad_norm": 0.3069910705089569,
      "learning_rate": 8e-05,
      "loss": 1.823,
      "step": 2443
    },
    {
      "epoch": 0.5375563620367315,
      "grad_norm": 0.29847028851509094,
      "learning_rate": 8e-05,
      "loss": 1.6314,
      "step": 2444
    },
    {
      "epoch": 0.5377763114483669,
      "grad_norm": 0.2970685660839081,
      "learning_rate": 8e-05,
      "loss": 1.8591,
      "step": 2445
    },
    {
      "epoch": 0.5379962608600022,
      "grad_norm": 0.28767916560173035,
      "learning_rate": 8e-05,
      "loss": 1.6244,
      "step": 2446
    },
    {
      "epoch": 0.5382162102716376,
      "grad_norm": 0.2858954966068268,
      "learning_rate": 8e-05,
      "loss": 1.6002,
      "step": 2447
    },
    {
      "epoch": 0.5384361596832729,
      "grad_norm": 0.25083082914352417,
      "learning_rate": 8e-05,
      "loss": 1.4772,
      "step": 2448
    },
    {
      "epoch": 0.5386561090949081,
      "grad_norm": 0.28772327303886414,
      "learning_rate": 8e-05,
      "loss": 1.7885,
      "step": 2449
    },
    {
      "epoch": 0.5388760585065435,
      "grad_norm": 0.300503671169281,
      "learning_rate": 8e-05,
      "loss": 1.8074,
      "step": 2450
    },
    {
      "epoch": 0.5390960079181788,
      "grad_norm": 0.29243797063827515,
      "learning_rate": 8e-05,
      "loss": 1.7033,
      "step": 2451
    },
    {
      "epoch": 0.5393159573298142,
      "grad_norm": 0.28921830654144287,
      "learning_rate": 8e-05,
      "loss": 1.734,
      "step": 2452
    },
    {
      "epoch": 0.5395359067414495,
      "grad_norm": 0.2754501700401306,
      "learning_rate": 8e-05,
      "loss": 1.577,
      "step": 2453
    },
    {
      "epoch": 0.5397558561530847,
      "grad_norm": 0.26824522018432617,
      "learning_rate": 8e-05,
      "loss": 1.6434,
      "step": 2454
    },
    {
      "epoch": 0.5399758055647201,
      "grad_norm": 0.26851388812065125,
      "learning_rate": 8e-05,
      "loss": 1.6706,
      "step": 2455
    },
    {
      "epoch": 0.5401957549763554,
      "grad_norm": 0.2697846293449402,
      "learning_rate": 8e-05,
      "loss": 1.6052,
      "step": 2456
    },
    {
      "epoch": 0.5404157043879908,
      "grad_norm": 0.27774059772491455,
      "learning_rate": 8e-05,
      "loss": 1.663,
      "step": 2457
    },
    {
      "epoch": 0.5406356537996261,
      "grad_norm": 0.2799103558063507,
      "learning_rate": 8e-05,
      "loss": 1.9412,
      "step": 2458
    },
    {
      "epoch": 0.5408556032112614,
      "grad_norm": 0.2874007523059845,
      "learning_rate": 8e-05,
      "loss": 1.6366,
      "step": 2459
    },
    {
      "epoch": 0.5410755526228967,
      "grad_norm": 0.29054176807403564,
      "learning_rate": 8e-05,
      "loss": 1.7546,
      "step": 2460
    },
    {
      "epoch": 0.541295502034532,
      "grad_norm": 0.29359421133995056,
      "learning_rate": 8e-05,
      "loss": 1.8274,
      "step": 2461
    },
    {
      "epoch": 0.5415154514461674,
      "grad_norm": 0.29589033126831055,
      "learning_rate": 8e-05,
      "loss": 1.8619,
      "step": 2462
    },
    {
      "epoch": 0.5417354008578027,
      "grad_norm": 0.2997150421142578,
      "learning_rate": 8e-05,
      "loss": 1.7685,
      "step": 2463
    },
    {
      "epoch": 0.541955350269438,
      "grad_norm": 0.2759319543838501,
      "learning_rate": 8e-05,
      "loss": 1.5652,
      "step": 2464
    },
    {
      "epoch": 0.5421752996810734,
      "grad_norm": 0.27741214632987976,
      "learning_rate": 8e-05,
      "loss": 1.6247,
      "step": 2465
    },
    {
      "epoch": 0.5423952490927086,
      "grad_norm": 0.29365673661231995,
      "learning_rate": 8e-05,
      "loss": 1.7768,
      "step": 2466
    },
    {
      "epoch": 0.542615198504344,
      "grad_norm": 0.2897026836872101,
      "learning_rate": 8e-05,
      "loss": 1.7435,
      "step": 2467
    },
    {
      "epoch": 0.5428351479159793,
      "grad_norm": 0.2963312566280365,
      "learning_rate": 8e-05,
      "loss": 1.7792,
      "step": 2468
    },
    {
      "epoch": 0.5430550973276147,
      "grad_norm": 0.3142043948173523,
      "learning_rate": 8e-05,
      "loss": 1.8348,
      "step": 2469
    },
    {
      "epoch": 0.54327504673925,
      "grad_norm": 0.28869184851646423,
      "learning_rate": 8e-05,
      "loss": 1.6916,
      "step": 2470
    },
    {
      "epoch": 0.5434949961508853,
      "grad_norm": 0.27220281958580017,
      "learning_rate": 8e-05,
      "loss": 1.5963,
      "step": 2471
    },
    {
      "epoch": 0.5437149455625206,
      "grad_norm": 0.3002524971961975,
      "learning_rate": 8e-05,
      "loss": 1.7516,
      "step": 2472
    },
    {
      "epoch": 0.5439348949741559,
      "grad_norm": 0.27016308903694153,
      "learning_rate": 8e-05,
      "loss": 1.5655,
      "step": 2473
    },
    {
      "epoch": 0.5441548443857913,
      "grad_norm": 0.2886146903038025,
      "learning_rate": 8e-05,
      "loss": 1.6563,
      "step": 2474
    },
    {
      "epoch": 0.5443747937974266,
      "grad_norm": 0.2743261158466339,
      "learning_rate": 8e-05,
      "loss": 1.7916,
      "step": 2475
    },
    {
      "epoch": 0.5445947432090619,
      "grad_norm": 0.27933475375175476,
      "learning_rate": 8e-05,
      "loss": 1.7397,
      "step": 2476
    },
    {
      "epoch": 0.5448146926206973,
      "grad_norm": 0.2805885672569275,
      "learning_rate": 8e-05,
      "loss": 1.6608,
      "step": 2477
    },
    {
      "epoch": 0.5450346420323325,
      "grad_norm": 0.26985716819763184,
      "learning_rate": 8e-05,
      "loss": 1.6551,
      "step": 2478
    },
    {
      "epoch": 0.5452545914439679,
      "grad_norm": 0.2778765857219696,
      "learning_rate": 8e-05,
      "loss": 1.6792,
      "step": 2479
    },
    {
      "epoch": 0.5454745408556032,
      "grad_norm": 0.27623313665390015,
      "learning_rate": 8e-05,
      "loss": 1.6867,
      "step": 2480
    },
    {
      "epoch": 0.5456944902672385,
      "grad_norm": 0.27185389399528503,
      "learning_rate": 8e-05,
      "loss": 1.6184,
      "step": 2481
    },
    {
      "epoch": 0.5459144396788739,
      "grad_norm": 0.29302138090133667,
      "learning_rate": 8e-05,
      "loss": 1.7075,
      "step": 2482
    },
    {
      "epoch": 0.5461343890905092,
      "grad_norm": 0.26639193296432495,
      "learning_rate": 8e-05,
      "loss": 1.6138,
      "step": 2483
    },
    {
      "epoch": 0.5463543385021445,
      "grad_norm": 0.28048211336135864,
      "learning_rate": 8e-05,
      "loss": 1.6672,
      "step": 2484
    },
    {
      "epoch": 0.5465742879137798,
      "grad_norm": 0.2844570577144623,
      "learning_rate": 8e-05,
      "loss": 1.6888,
      "step": 2485
    },
    {
      "epoch": 0.5467942373254151,
      "grad_norm": 0.2801128923892975,
      "learning_rate": 8e-05,
      "loss": 1.8506,
      "step": 2486
    },
    {
      "epoch": 0.5470141867370505,
      "grad_norm": 0.2718241810798645,
      "learning_rate": 8e-05,
      "loss": 1.6105,
      "step": 2487
    },
    {
      "epoch": 0.5472341361486858,
      "grad_norm": 0.28759825229644775,
      "learning_rate": 8e-05,
      "loss": 1.7449,
      "step": 2488
    },
    {
      "epoch": 0.5474540855603212,
      "grad_norm": 0.29218876361846924,
      "learning_rate": 8e-05,
      "loss": 1.8732,
      "step": 2489
    },
    {
      "epoch": 0.5476740349719564,
      "grad_norm": 0.29760751128196716,
      "learning_rate": 8e-05,
      "loss": 1.7804,
      "step": 2490
    },
    {
      "epoch": 0.5478939843835918,
      "grad_norm": 0.28636956214904785,
      "learning_rate": 8e-05,
      "loss": 1.6994,
      "step": 2491
    },
    {
      "epoch": 0.5481139337952271,
      "grad_norm": 0.2892046570777893,
      "learning_rate": 8e-05,
      "loss": 1.759,
      "step": 2492
    },
    {
      "epoch": 0.5483338832068624,
      "grad_norm": 0.280556857585907,
      "learning_rate": 8e-05,
      "loss": 1.8084,
      "step": 2493
    },
    {
      "epoch": 0.5485538326184978,
      "grad_norm": 0.2733471691608429,
      "learning_rate": 8e-05,
      "loss": 1.7293,
      "step": 2494
    },
    {
      "epoch": 0.548773782030133,
      "grad_norm": 0.2813643515110016,
      "learning_rate": 8e-05,
      "loss": 1.6178,
      "step": 2495
    },
    {
      "epoch": 0.5489937314417684,
      "grad_norm": 0.27255943417549133,
      "learning_rate": 8e-05,
      "loss": 1.6827,
      "step": 2496
    },
    {
      "epoch": 0.5492136808534037,
      "grad_norm": 0.2690375745296478,
      "learning_rate": 8e-05,
      "loss": 1.684,
      "step": 2497
    },
    {
      "epoch": 0.549433630265039,
      "grad_norm": 0.30036401748657227,
      "learning_rate": 8e-05,
      "loss": 1.8676,
      "step": 2498
    },
    {
      "epoch": 0.5496535796766744,
      "grad_norm": 0.27924251556396484,
      "learning_rate": 8e-05,
      "loss": 1.6619,
      "step": 2499
    },
    {
      "epoch": 0.5498735290883097,
      "grad_norm": 0.2792947590351105,
      "learning_rate": 8e-05,
      "loss": 1.6121,
      "step": 2500
    },
    {
      "epoch": 0.550093478499945,
      "grad_norm": 0.27976930141448975,
      "learning_rate": 8e-05,
      "loss": 1.5815,
      "step": 2501
    },
    {
      "epoch": 0.5503134279115803,
      "grad_norm": 0.28429850935935974,
      "learning_rate": 8e-05,
      "loss": 1.6713,
      "step": 2502
    },
    {
      "epoch": 0.5505333773232156,
      "grad_norm": 0.2669944763183594,
      "learning_rate": 8e-05,
      "loss": 1.5065,
      "step": 2503
    },
    {
      "epoch": 0.550753326734851,
      "grad_norm": 0.2846994400024414,
      "learning_rate": 8e-05,
      "loss": 1.7238,
      "step": 2504
    },
    {
      "epoch": 0.5509732761464863,
      "grad_norm": 0.27598071098327637,
      "learning_rate": 8e-05,
      "loss": 1.5364,
      "step": 2505
    },
    {
      "epoch": 0.5511932255581217,
      "grad_norm": 0.27275460958480835,
      "learning_rate": 8e-05,
      "loss": 1.6171,
      "step": 2506
    },
    {
      "epoch": 0.551413174969757,
      "grad_norm": 0.2846895456314087,
      "learning_rate": 8e-05,
      "loss": 1.7082,
      "step": 2507
    },
    {
      "epoch": 0.5516331243813922,
      "grad_norm": 0.3010547161102295,
      "learning_rate": 8e-05,
      "loss": 1.7946,
      "step": 2508
    },
    {
      "epoch": 0.5518530737930276,
      "grad_norm": 0.28405773639678955,
      "learning_rate": 8e-05,
      "loss": 1.6063,
      "step": 2509
    },
    {
      "epoch": 0.5520730232046629,
      "grad_norm": 0.2855536639690399,
      "learning_rate": 8e-05,
      "loss": 1.6138,
      "step": 2510
    },
    {
      "epoch": 0.5522929726162983,
      "grad_norm": 0.2949456572532654,
      "learning_rate": 8e-05,
      "loss": 1.7734,
      "step": 2511
    },
    {
      "epoch": 0.5525129220279336,
      "grad_norm": 0.31665512919425964,
      "learning_rate": 8e-05,
      "loss": 1.7163,
      "step": 2512
    },
    {
      "epoch": 0.552732871439569,
      "grad_norm": 0.2881389260292053,
      "learning_rate": 8e-05,
      "loss": 1.5617,
      "step": 2513
    },
    {
      "epoch": 0.5529528208512042,
      "grad_norm": 0.26758721470832825,
      "learning_rate": 8e-05,
      "loss": 1.6714,
      "step": 2514
    },
    {
      "epoch": 0.5531727702628395,
      "grad_norm": 0.29549580812454224,
      "learning_rate": 8e-05,
      "loss": 1.6516,
      "step": 2515
    },
    {
      "epoch": 0.5533927196744749,
      "grad_norm": 0.2811340391635895,
      "learning_rate": 8e-05,
      "loss": 1.6026,
      "step": 2516
    },
    {
      "epoch": 0.5536126690861102,
      "grad_norm": 0.2837204039096832,
      "learning_rate": 8e-05,
      "loss": 1.8413,
      "step": 2517
    },
    {
      "epoch": 0.5538326184977456,
      "grad_norm": 0.276216983795166,
      "learning_rate": 8e-05,
      "loss": 1.6671,
      "step": 2518
    },
    {
      "epoch": 0.5540525679093808,
      "grad_norm": 0.2781767249107361,
      "learning_rate": 8e-05,
      "loss": 1.5886,
      "step": 2519
    },
    {
      "epoch": 0.5542725173210161,
      "grad_norm": 0.2830861210823059,
      "learning_rate": 8e-05,
      "loss": 1.6097,
      "step": 2520
    },
    {
      "epoch": 0.5544924667326515,
      "grad_norm": 0.2746805250644684,
      "learning_rate": 8e-05,
      "loss": 1.7153,
      "step": 2521
    },
    {
      "epoch": 0.5547124161442868,
      "grad_norm": 0.2781994640827179,
      "learning_rate": 8e-05,
      "loss": 1.6597,
      "step": 2522
    },
    {
      "epoch": 0.5549323655559222,
      "grad_norm": 0.2919979393482208,
      "learning_rate": 8e-05,
      "loss": 1.715,
      "step": 2523
    },
    {
      "epoch": 0.5551523149675575,
      "grad_norm": 0.27563661336898804,
      "learning_rate": 8e-05,
      "loss": 1.6129,
      "step": 2524
    },
    {
      "epoch": 0.5553722643791927,
      "grad_norm": 0.3070942163467407,
      "learning_rate": 8e-05,
      "loss": 1.875,
      "step": 2525
    },
    {
      "epoch": 0.5555922137908281,
      "grad_norm": 0.278039813041687,
      "learning_rate": 8e-05,
      "loss": 1.6894,
      "step": 2526
    },
    {
      "epoch": 0.5558121632024634,
      "grad_norm": 0.2709571421146393,
      "learning_rate": 8e-05,
      "loss": 1.5268,
      "step": 2527
    },
    {
      "epoch": 0.5560321126140988,
      "grad_norm": 0.27659523487091064,
      "learning_rate": 8e-05,
      "loss": 1.5817,
      "step": 2528
    },
    {
      "epoch": 0.5562520620257341,
      "grad_norm": 0.33376970887184143,
      "learning_rate": 8e-05,
      "loss": 1.6275,
      "step": 2529
    },
    {
      "epoch": 0.5564720114373694,
      "grad_norm": 0.28663134574890137,
      "learning_rate": 8e-05,
      "loss": 1.6497,
      "step": 2530
    },
    {
      "epoch": 0.5566919608490047,
      "grad_norm": 0.27400556206703186,
      "learning_rate": 8e-05,
      "loss": 1.6768,
      "step": 2531
    },
    {
      "epoch": 0.55691191026064,
      "grad_norm": 0.3359694182872772,
      "learning_rate": 8e-05,
      "loss": 1.7628,
      "step": 2532
    },
    {
      "epoch": 0.5571318596722754,
      "grad_norm": 0.3009445071220398,
      "learning_rate": 8e-05,
      "loss": 1.6368,
      "step": 2533
    },
    {
      "epoch": 0.5573518090839107,
      "grad_norm": 0.2951606512069702,
      "learning_rate": 8e-05,
      "loss": 1.6468,
      "step": 2534
    },
    {
      "epoch": 0.5575717584955461,
      "grad_norm": 0.298835426568985,
      "learning_rate": 8e-05,
      "loss": 1.679,
      "step": 2535
    },
    {
      "epoch": 0.5577917079071814,
      "grad_norm": 0.29196399450302124,
      "learning_rate": 8e-05,
      "loss": 1.6865,
      "step": 2536
    },
    {
      "epoch": 0.5580116573188166,
      "grad_norm": 0.3057127296924591,
      "learning_rate": 8e-05,
      "loss": 1.7979,
      "step": 2537
    },
    {
      "epoch": 0.558231606730452,
      "grad_norm": 0.3170565664768219,
      "learning_rate": 8e-05,
      "loss": 1.7815,
      "step": 2538
    },
    {
      "epoch": 0.5584515561420873,
      "grad_norm": 0.28287273645401,
      "learning_rate": 8e-05,
      "loss": 1.7151,
      "step": 2539
    },
    {
      "epoch": 0.5586715055537227,
      "grad_norm": 0.30313780903816223,
      "learning_rate": 8e-05,
      "loss": 1.6501,
      "step": 2540
    },
    {
      "epoch": 0.558891454965358,
      "grad_norm": 0.28195586800575256,
      "learning_rate": 8e-05,
      "loss": 1.7281,
      "step": 2541
    },
    {
      "epoch": 0.5591114043769932,
      "grad_norm": 0.2734014391899109,
      "learning_rate": 8e-05,
      "loss": 1.6504,
      "step": 2542
    },
    {
      "epoch": 0.5593313537886286,
      "grad_norm": 0.28178513050079346,
      "learning_rate": 8e-05,
      "loss": 1.5946,
      "step": 2543
    },
    {
      "epoch": 0.5595513032002639,
      "grad_norm": 0.2800062894821167,
      "learning_rate": 8e-05,
      "loss": 1.7498,
      "step": 2544
    },
    {
      "epoch": 0.5597712526118993,
      "grad_norm": 0.28368762135505676,
      "learning_rate": 8e-05,
      "loss": 1.6732,
      "step": 2545
    },
    {
      "epoch": 0.5599912020235346,
      "grad_norm": 0.3069396913051605,
      "learning_rate": 8e-05,
      "loss": 1.7647,
      "step": 2546
    },
    {
      "epoch": 0.5602111514351699,
      "grad_norm": 0.27336394786834717,
      "learning_rate": 8e-05,
      "loss": 1.4557,
      "step": 2547
    },
    {
      "epoch": 0.5604311008468053,
      "grad_norm": 0.28363245725631714,
      "learning_rate": 8e-05,
      "loss": 1.7301,
      "step": 2548
    },
    {
      "epoch": 0.5606510502584405,
      "grad_norm": 0.3097067177295685,
      "learning_rate": 8e-05,
      "loss": 1.7322,
      "step": 2549
    },
    {
      "epoch": 0.5608709996700759,
      "grad_norm": 0.28125154972076416,
      "learning_rate": 8e-05,
      "loss": 1.5551,
      "step": 2550
    },
    {
      "epoch": 0.5610909490817112,
      "grad_norm": 0.3111821413040161,
      "learning_rate": 8e-05,
      "loss": 1.8965,
      "step": 2551
    },
    {
      "epoch": 0.5613108984933465,
      "grad_norm": 0.2920529842376709,
      "learning_rate": 8e-05,
      "loss": 1.6835,
      "step": 2552
    },
    {
      "epoch": 0.5615308479049819,
      "grad_norm": 0.27278631925582886,
      "learning_rate": 8e-05,
      "loss": 1.7175,
      "step": 2553
    },
    {
      "epoch": 0.5617507973166171,
      "grad_norm": 0.2742355763912201,
      "learning_rate": 8e-05,
      "loss": 1.6745,
      "step": 2554
    },
    {
      "epoch": 0.5619707467282525,
      "grad_norm": 0.2675003707408905,
      "learning_rate": 8e-05,
      "loss": 1.5771,
      "step": 2555
    },
    {
      "epoch": 0.5621906961398878,
      "grad_norm": 0.2805350422859192,
      "learning_rate": 8e-05,
      "loss": 1.7503,
      "step": 2556
    },
    {
      "epoch": 0.5624106455515232,
      "grad_norm": 0.27205830812454224,
      "learning_rate": 8e-05,
      "loss": 1.6591,
      "step": 2557
    },
    {
      "epoch": 0.5626305949631585,
      "grad_norm": 0.26984983682632446,
      "learning_rate": 8e-05,
      "loss": 1.6351,
      "step": 2558
    },
    {
      "epoch": 0.5628505443747938,
      "grad_norm": 0.3067481517791748,
      "learning_rate": 8e-05,
      "loss": 1.5304,
      "step": 2559
    },
    {
      "epoch": 0.5630704937864291,
      "grad_norm": 0.28945624828338623,
      "learning_rate": 8e-05,
      "loss": 1.7099,
      "step": 2560
    },
    {
      "epoch": 0.5632904431980644,
      "grad_norm": 0.269144743680954,
      "learning_rate": 8e-05,
      "loss": 1.557,
      "step": 2561
    },
    {
      "epoch": 0.5635103926096998,
      "grad_norm": 0.329520583152771,
      "learning_rate": 8e-05,
      "loss": 1.7867,
      "step": 2562
    },
    {
      "epoch": 0.5637303420213351,
      "grad_norm": 0.35944700241088867,
      "learning_rate": 8e-05,
      "loss": 1.8146,
      "step": 2563
    },
    {
      "epoch": 0.5639502914329704,
      "grad_norm": 0.30693116784095764,
      "learning_rate": 8e-05,
      "loss": 1.7709,
      "step": 2564
    },
    {
      "epoch": 0.5641702408446058,
      "grad_norm": 0.31814631819725037,
      "learning_rate": 8e-05,
      "loss": 1.8679,
      "step": 2565
    },
    {
      "epoch": 0.564390190256241,
      "grad_norm": 0.2988479435443878,
      "learning_rate": 8e-05,
      "loss": 1.7675,
      "step": 2566
    },
    {
      "epoch": 0.5646101396678764,
      "grad_norm": 0.2955850064754486,
      "learning_rate": 8e-05,
      "loss": 1.7082,
      "step": 2567
    },
    {
      "epoch": 0.5648300890795117,
      "grad_norm": 0.27773404121398926,
      "learning_rate": 8e-05,
      "loss": 1.5378,
      "step": 2568
    },
    {
      "epoch": 0.565050038491147,
      "grad_norm": 0.2847524583339691,
      "learning_rate": 8e-05,
      "loss": 1.792,
      "step": 2569
    },
    {
      "epoch": 0.5652699879027824,
      "grad_norm": 0.29024967551231384,
      "learning_rate": 8e-05,
      "loss": 1.6185,
      "step": 2570
    },
    {
      "epoch": 0.5654899373144177,
      "grad_norm": 0.27534323930740356,
      "learning_rate": 8e-05,
      "loss": 1.7044,
      "step": 2571
    },
    {
      "epoch": 0.565709886726053,
      "grad_norm": 0.28059902787208557,
      "learning_rate": 8e-05,
      "loss": 1.7199,
      "step": 2572
    },
    {
      "epoch": 0.5659298361376883,
      "grad_norm": 0.29140958189964294,
      "learning_rate": 8e-05,
      "loss": 1.6534,
      "step": 2573
    },
    {
      "epoch": 0.5661497855493236,
      "grad_norm": 0.303821861743927,
      "learning_rate": 8e-05,
      "loss": 1.7163,
      "step": 2574
    },
    {
      "epoch": 0.566369734960959,
      "grad_norm": 0.3073093295097351,
      "learning_rate": 8e-05,
      "loss": 1.7885,
      "step": 2575
    },
    {
      "epoch": 0.5665896843725943,
      "grad_norm": 0.2976214289665222,
      "learning_rate": 8e-05,
      "loss": 1.7059,
      "step": 2576
    },
    {
      "epoch": 0.5668096337842297,
      "grad_norm": 0.3081284761428833,
      "learning_rate": 8e-05,
      "loss": 1.5529,
      "step": 2577
    },
    {
      "epoch": 0.5670295831958649,
      "grad_norm": 0.2893354594707489,
      "learning_rate": 8e-05,
      "loss": 1.8564,
      "step": 2578
    },
    {
      "epoch": 0.5672495326075003,
      "grad_norm": 0.2904176115989685,
      "learning_rate": 8e-05,
      "loss": 1.6903,
      "step": 2579
    },
    {
      "epoch": 0.5674694820191356,
      "grad_norm": 0.2907819449901581,
      "learning_rate": 8e-05,
      "loss": 1.5663,
      "step": 2580
    },
    {
      "epoch": 0.5676894314307709,
      "grad_norm": 0.27938172221183777,
      "learning_rate": 8e-05,
      "loss": 1.6031,
      "step": 2581
    },
    {
      "epoch": 0.5679093808424063,
      "grad_norm": 0.28864786028862,
      "learning_rate": 8e-05,
      "loss": 1.666,
      "step": 2582
    },
    {
      "epoch": 0.5681293302540416,
      "grad_norm": 0.29587891697883606,
      "learning_rate": 8e-05,
      "loss": 1.7545,
      "step": 2583
    },
    {
      "epoch": 0.5683492796656769,
      "grad_norm": 0.26541203260421753,
      "learning_rate": 8e-05,
      "loss": 1.6059,
      "step": 2584
    },
    {
      "epoch": 0.5685692290773122,
      "grad_norm": 0.2819576561450958,
      "learning_rate": 8e-05,
      "loss": 1.7227,
      "step": 2585
    },
    {
      "epoch": 0.5687891784889475,
      "grad_norm": 0.2920463979244232,
      "learning_rate": 8e-05,
      "loss": 1.7553,
      "step": 2586
    },
    {
      "epoch": 0.5690091279005829,
      "grad_norm": 0.29490089416503906,
      "learning_rate": 8e-05,
      "loss": 1.7117,
      "step": 2587
    },
    {
      "epoch": 0.5692290773122182,
      "grad_norm": 0.29847970604896545,
      "learning_rate": 8e-05,
      "loss": 1.8931,
      "step": 2588
    },
    {
      "epoch": 0.5694490267238536,
      "grad_norm": 0.28575995564460754,
      "learning_rate": 8e-05,
      "loss": 1.595,
      "step": 2589
    },
    {
      "epoch": 0.5696689761354888,
      "grad_norm": 0.28053271770477295,
      "learning_rate": 8e-05,
      "loss": 1.6089,
      "step": 2590
    },
    {
      "epoch": 0.5698889255471241,
      "grad_norm": 0.27538979053497314,
      "learning_rate": 8e-05,
      "loss": 1.7808,
      "step": 2591
    },
    {
      "epoch": 0.5701088749587595,
      "grad_norm": 0.2819748520851135,
      "learning_rate": 8e-05,
      "loss": 1.7355,
      "step": 2592
    },
    {
      "epoch": 0.5703288243703948,
      "grad_norm": 0.3023085594177246,
      "learning_rate": 8e-05,
      "loss": 1.759,
      "step": 2593
    },
    {
      "epoch": 0.5705487737820302,
      "grad_norm": 0.28369995951652527,
      "learning_rate": 8e-05,
      "loss": 1.7796,
      "step": 2594
    },
    {
      "epoch": 0.5707687231936655,
      "grad_norm": 0.27062156796455383,
      "learning_rate": 8e-05,
      "loss": 1.6206,
      "step": 2595
    },
    {
      "epoch": 0.5709886726053007,
      "grad_norm": 0.2928752303123474,
      "learning_rate": 8e-05,
      "loss": 1.7578,
      "step": 2596
    },
    {
      "epoch": 0.5712086220169361,
      "grad_norm": 0.28366369009017944,
      "learning_rate": 8e-05,
      "loss": 1.6367,
      "step": 2597
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.2794798016548157,
      "learning_rate": 8e-05,
      "loss": 1.7006,
      "step": 2598
    },
    {
      "epoch": 0.5716485208402068,
      "grad_norm": 0.278814435005188,
      "learning_rate": 8e-05,
      "loss": 1.6883,
      "step": 2599
    },
    {
      "epoch": 0.5718684702518421,
      "grad_norm": 0.28789058327674866,
      "learning_rate": 8e-05,
      "loss": 1.76,
      "step": 2600
    },
    {
      "epoch": 0.5720884196634775,
      "grad_norm": 0.289120614528656,
      "learning_rate": 8e-05,
      "loss": 1.5449,
      "step": 2601
    },
    {
      "epoch": 0.5723083690751127,
      "grad_norm": 0.27491265535354614,
      "learning_rate": 8e-05,
      "loss": 1.6287,
      "step": 2602
    },
    {
      "epoch": 0.572528318486748,
      "grad_norm": 0.2837536931037903,
      "learning_rate": 8e-05,
      "loss": 1.6618,
      "step": 2603
    },
    {
      "epoch": 0.5727482678983834,
      "grad_norm": 0.27386194467544556,
      "learning_rate": 8e-05,
      "loss": 1.5815,
      "step": 2604
    },
    {
      "epoch": 0.5729682173100187,
      "grad_norm": 0.2818918228149414,
      "learning_rate": 8e-05,
      "loss": 1.511,
      "step": 2605
    },
    {
      "epoch": 0.5731881667216541,
      "grad_norm": 0.29329514503479004,
      "learning_rate": 8e-05,
      "loss": 1.5494,
      "step": 2606
    },
    {
      "epoch": 0.5734081161332893,
      "grad_norm": 0.29942408204078674,
      "learning_rate": 8e-05,
      "loss": 1.8049,
      "step": 2607
    },
    {
      "epoch": 0.5736280655449246,
      "grad_norm": 0.30527159571647644,
      "learning_rate": 8e-05,
      "loss": 1.8735,
      "step": 2608
    },
    {
      "epoch": 0.57384801495656,
      "grad_norm": 0.2842453122138977,
      "learning_rate": 8e-05,
      "loss": 1.7124,
      "step": 2609
    },
    {
      "epoch": 0.5740679643681953,
      "grad_norm": 0.3305295407772064,
      "learning_rate": 8e-05,
      "loss": 1.6555,
      "step": 2610
    },
    {
      "epoch": 0.5742879137798307,
      "grad_norm": 0.28134140372276306,
      "learning_rate": 8e-05,
      "loss": 1.639,
      "step": 2611
    },
    {
      "epoch": 0.574507863191466,
      "grad_norm": 0.2862444818019867,
      "learning_rate": 8e-05,
      "loss": 1.7949,
      "step": 2612
    },
    {
      "epoch": 0.5747278126031012,
      "grad_norm": 0.3089071214199066,
      "learning_rate": 8e-05,
      "loss": 1.7782,
      "step": 2613
    },
    {
      "epoch": 0.5749477620147366,
      "grad_norm": 0.3113284111022949,
      "learning_rate": 8e-05,
      "loss": 1.5588,
      "step": 2614
    },
    {
      "epoch": 0.5751677114263719,
      "grad_norm": 0.2865052819252014,
      "learning_rate": 8e-05,
      "loss": 1.6919,
      "step": 2615
    },
    {
      "epoch": 0.5753876608380073,
      "grad_norm": 0.26997220516204834,
      "learning_rate": 8e-05,
      "loss": 1.6853,
      "step": 2616
    },
    {
      "epoch": 0.5756076102496426,
      "grad_norm": 0.3056239187717438,
      "learning_rate": 8e-05,
      "loss": 1.7971,
      "step": 2617
    },
    {
      "epoch": 0.5758275596612779,
      "grad_norm": 0.3041035234928131,
      "learning_rate": 8e-05,
      "loss": 1.7889,
      "step": 2618
    },
    {
      "epoch": 0.5760475090729132,
      "grad_norm": 0.2829764187335968,
      "learning_rate": 8e-05,
      "loss": 1.6371,
      "step": 2619
    },
    {
      "epoch": 0.5762674584845485,
      "grad_norm": 0.29050111770629883,
      "learning_rate": 8e-05,
      "loss": 1.8909,
      "step": 2620
    },
    {
      "epoch": 0.5764874078961839,
      "grad_norm": 0.29888811707496643,
      "learning_rate": 8e-05,
      "loss": 1.6974,
      "step": 2621
    },
    {
      "epoch": 0.5767073573078192,
      "grad_norm": 0.3193587362766266,
      "learning_rate": 8e-05,
      "loss": 1.7083,
      "step": 2622
    },
    {
      "epoch": 0.5769273067194546,
      "grad_norm": 0.2855699360370636,
      "learning_rate": 8e-05,
      "loss": 1.6106,
      "step": 2623
    },
    {
      "epoch": 0.5771472561310899,
      "grad_norm": 0.29608815908432007,
      "learning_rate": 8e-05,
      "loss": 1.7813,
      "step": 2624
    },
    {
      "epoch": 0.5773672055427251,
      "grad_norm": 0.2846873700618744,
      "learning_rate": 8e-05,
      "loss": 1.6164,
      "step": 2625
    },
    {
      "epoch": 0.5775871549543605,
      "grad_norm": 0.3074873983860016,
      "learning_rate": 8e-05,
      "loss": 1.7278,
      "step": 2626
    },
    {
      "epoch": 0.5778071043659958,
      "grad_norm": 0.3016159236431122,
      "learning_rate": 8e-05,
      "loss": 1.6286,
      "step": 2627
    },
    {
      "epoch": 0.5780270537776312,
      "grad_norm": 0.28926798701286316,
      "learning_rate": 8e-05,
      "loss": 1.746,
      "step": 2628
    },
    {
      "epoch": 0.5782470031892665,
      "grad_norm": 0.3222711682319641,
      "learning_rate": 8e-05,
      "loss": 1.8108,
      "step": 2629
    },
    {
      "epoch": 0.5784669526009018,
      "grad_norm": 0.30052945017814636,
      "learning_rate": 8e-05,
      "loss": 1.7374,
      "step": 2630
    },
    {
      "epoch": 0.5786869020125371,
      "grad_norm": 0.2880706787109375,
      "learning_rate": 8e-05,
      "loss": 1.627,
      "step": 2631
    },
    {
      "epoch": 0.5789068514241724,
      "grad_norm": 0.30028629302978516,
      "learning_rate": 8e-05,
      "loss": 1.8345,
      "step": 2632
    },
    {
      "epoch": 0.5791268008358078,
      "grad_norm": 0.3164263665676117,
      "learning_rate": 8e-05,
      "loss": 1.9713,
      "step": 2633
    },
    {
      "epoch": 0.5793467502474431,
      "grad_norm": 0.294114887714386,
      "learning_rate": 8e-05,
      "loss": 1.6083,
      "step": 2634
    },
    {
      "epoch": 0.5795666996590784,
      "grad_norm": 0.31809002161026,
      "learning_rate": 8e-05,
      "loss": 1.794,
      "step": 2635
    },
    {
      "epoch": 0.5797866490707138,
      "grad_norm": 0.3005049526691437,
      "learning_rate": 8e-05,
      "loss": 1.469,
      "step": 2636
    },
    {
      "epoch": 0.580006598482349,
      "grad_norm": 0.2874310314655304,
      "learning_rate": 8e-05,
      "loss": 1.7345,
      "step": 2637
    },
    {
      "epoch": 0.5802265478939844,
      "grad_norm": 0.295523077249527,
      "learning_rate": 8e-05,
      "loss": 1.6995,
      "step": 2638
    },
    {
      "epoch": 0.5804464973056197,
      "grad_norm": 0.29120928049087524,
      "learning_rate": 8e-05,
      "loss": 1.7736,
      "step": 2639
    },
    {
      "epoch": 0.580666446717255,
      "grad_norm": 0.2916790246963501,
      "learning_rate": 8e-05,
      "loss": 1.74,
      "step": 2640
    },
    {
      "epoch": 0.5808863961288904,
      "grad_norm": 0.285230427980423,
      "learning_rate": 8e-05,
      "loss": 1.7685,
      "step": 2641
    },
    {
      "epoch": 0.5811063455405256,
      "grad_norm": 0.2743189334869385,
      "learning_rate": 8e-05,
      "loss": 1.6751,
      "step": 2642
    },
    {
      "epoch": 0.581326294952161,
      "grad_norm": 0.2997332811355591,
      "learning_rate": 8e-05,
      "loss": 1.5959,
      "step": 2643
    },
    {
      "epoch": 0.5815462443637963,
      "grad_norm": 0.28394201397895813,
      "learning_rate": 8e-05,
      "loss": 1.6288,
      "step": 2644
    },
    {
      "epoch": 0.5817661937754317,
      "grad_norm": 0.2787470519542694,
      "learning_rate": 8e-05,
      "loss": 1.7496,
      "step": 2645
    },
    {
      "epoch": 0.581986143187067,
      "grad_norm": 0.2853599488735199,
      "learning_rate": 8e-05,
      "loss": 1.6439,
      "step": 2646
    },
    {
      "epoch": 0.5822060925987023,
      "grad_norm": 0.2939299941062927,
      "learning_rate": 8e-05,
      "loss": 1.7293,
      "step": 2647
    },
    {
      "epoch": 0.5824260420103377,
      "grad_norm": 0.27831408381462097,
      "learning_rate": 8e-05,
      "loss": 1.6748,
      "step": 2648
    },
    {
      "epoch": 0.5826459914219729,
      "grad_norm": 0.296762615442276,
      "learning_rate": 8e-05,
      "loss": 1.6735,
      "step": 2649
    },
    {
      "epoch": 0.5828659408336083,
      "grad_norm": 0.27961719036102295,
      "learning_rate": 8e-05,
      "loss": 1.6837,
      "step": 2650
    },
    {
      "epoch": 0.5830858902452436,
      "grad_norm": 0.27915704250335693,
      "learning_rate": 8e-05,
      "loss": 1.6745,
      "step": 2651
    },
    {
      "epoch": 0.5833058396568789,
      "grad_norm": 0.273799329996109,
      "learning_rate": 8e-05,
      "loss": 1.5609,
      "step": 2652
    },
    {
      "epoch": 0.5835257890685143,
      "grad_norm": 0.287383109331131,
      "learning_rate": 8e-05,
      "loss": 1.7569,
      "step": 2653
    },
    {
      "epoch": 0.5837457384801495,
      "grad_norm": 0.27745500206947327,
      "learning_rate": 8e-05,
      "loss": 1.6093,
      "step": 2654
    },
    {
      "epoch": 0.5839656878917849,
      "grad_norm": 0.2954557240009308,
      "learning_rate": 8e-05,
      "loss": 1.788,
      "step": 2655
    },
    {
      "epoch": 0.5841856373034202,
      "grad_norm": 0.28464850783348083,
      "learning_rate": 8e-05,
      "loss": 1.7079,
      "step": 2656
    },
    {
      "epoch": 0.5844055867150555,
      "grad_norm": 0.27475497126579285,
      "learning_rate": 8e-05,
      "loss": 1.6137,
      "step": 2657
    },
    {
      "epoch": 0.5846255361266909,
      "grad_norm": 0.27928462624549866,
      "learning_rate": 8e-05,
      "loss": 1.5776,
      "step": 2658
    },
    {
      "epoch": 0.5848454855383262,
      "grad_norm": 0.2889251708984375,
      "learning_rate": 8e-05,
      "loss": 1.7871,
      "step": 2659
    },
    {
      "epoch": 0.5850654349499615,
      "grad_norm": 0.29489466547966003,
      "learning_rate": 8e-05,
      "loss": 1.7299,
      "step": 2660
    },
    {
      "epoch": 0.5852853843615968,
      "grad_norm": 0.27761825919151306,
      "learning_rate": 8e-05,
      "loss": 1.6772,
      "step": 2661
    },
    {
      "epoch": 0.5855053337732321,
      "grad_norm": 0.2886674702167511,
      "learning_rate": 8e-05,
      "loss": 1.6718,
      "step": 2662
    },
    {
      "epoch": 0.5857252831848675,
      "grad_norm": 0.2736080586910248,
      "learning_rate": 8e-05,
      "loss": 1.5834,
      "step": 2663
    },
    {
      "epoch": 0.5859452325965028,
      "grad_norm": 0.29493847489356995,
      "learning_rate": 8e-05,
      "loss": 1.7123,
      "step": 2664
    },
    {
      "epoch": 0.5861651820081382,
      "grad_norm": 0.2919282615184784,
      "learning_rate": 8e-05,
      "loss": 1.7192,
      "step": 2665
    },
    {
      "epoch": 0.5863851314197734,
      "grad_norm": 0.2883647680282593,
      "learning_rate": 8e-05,
      "loss": 1.6271,
      "step": 2666
    },
    {
      "epoch": 0.5866050808314087,
      "grad_norm": 0.2852446734905243,
      "learning_rate": 8e-05,
      "loss": 1.7295,
      "step": 2667
    },
    {
      "epoch": 0.5868250302430441,
      "grad_norm": 0.3113778531551361,
      "learning_rate": 8e-05,
      "loss": 1.6605,
      "step": 2668
    },
    {
      "epoch": 0.5870449796546794,
      "grad_norm": 0.2629379630088806,
      "learning_rate": 8e-05,
      "loss": 1.3457,
      "step": 2669
    },
    {
      "epoch": 0.5872649290663148,
      "grad_norm": 0.28648287057876587,
      "learning_rate": 8e-05,
      "loss": 1.7137,
      "step": 2670
    },
    {
      "epoch": 0.5874848784779501,
      "grad_norm": 0.30140426754951477,
      "learning_rate": 8e-05,
      "loss": 1.7612,
      "step": 2671
    },
    {
      "epoch": 0.5877048278895854,
      "grad_norm": 0.29059261083602905,
      "learning_rate": 8e-05,
      "loss": 1.8526,
      "step": 2672
    },
    {
      "epoch": 0.5879247773012207,
      "grad_norm": 0.2913878560066223,
      "learning_rate": 8e-05,
      "loss": 1.7214,
      "step": 2673
    },
    {
      "epoch": 0.588144726712856,
      "grad_norm": 0.3046487271785736,
      "learning_rate": 8e-05,
      "loss": 1.8342,
      "step": 2674
    },
    {
      "epoch": 0.5883646761244914,
      "grad_norm": 0.2699670195579529,
      "learning_rate": 8e-05,
      "loss": 1.6057,
      "step": 2675
    },
    {
      "epoch": 0.5885846255361267,
      "grad_norm": 0.2722747027873993,
      "learning_rate": 8e-05,
      "loss": 1.5067,
      "step": 2676
    },
    {
      "epoch": 0.5888045749477621,
      "grad_norm": 0.27758973836898804,
      "learning_rate": 8e-05,
      "loss": 1.613,
      "step": 2677
    },
    {
      "epoch": 0.5890245243593973,
      "grad_norm": 0.30234992504119873,
      "learning_rate": 8e-05,
      "loss": 1.7266,
      "step": 2678
    },
    {
      "epoch": 0.5892444737710326,
      "grad_norm": 0.3146234452724457,
      "learning_rate": 8e-05,
      "loss": 1.6588,
      "step": 2679
    },
    {
      "epoch": 0.589464423182668,
      "grad_norm": 0.2867683470249176,
      "learning_rate": 8e-05,
      "loss": 1.6726,
      "step": 2680
    },
    {
      "epoch": 0.5896843725943033,
      "grad_norm": 0.28295040130615234,
      "learning_rate": 8e-05,
      "loss": 1.7338,
      "step": 2681
    },
    {
      "epoch": 0.5899043220059387,
      "grad_norm": 0.28655725717544556,
      "learning_rate": 8e-05,
      "loss": 1.6791,
      "step": 2682
    },
    {
      "epoch": 0.590124271417574,
      "grad_norm": 0.2897862493991852,
      "learning_rate": 8e-05,
      "loss": 1.7127,
      "step": 2683
    },
    {
      "epoch": 0.5903442208292092,
      "grad_norm": 0.278427392244339,
      "learning_rate": 8e-05,
      "loss": 1.7166,
      "step": 2684
    },
    {
      "epoch": 0.5905641702408446,
      "grad_norm": 0.28383758664131165,
      "learning_rate": 8e-05,
      "loss": 1.8498,
      "step": 2685
    },
    {
      "epoch": 0.5907841196524799,
      "grad_norm": 0.2690020501613617,
      "learning_rate": 8e-05,
      "loss": 1.5887,
      "step": 2686
    },
    {
      "epoch": 0.5910040690641153,
      "grad_norm": 0.2910546362400055,
      "learning_rate": 8e-05,
      "loss": 1.7525,
      "step": 2687
    },
    {
      "epoch": 0.5912240184757506,
      "grad_norm": 0.2932651937007904,
      "learning_rate": 8e-05,
      "loss": 1.6113,
      "step": 2688
    },
    {
      "epoch": 0.5914439678873858,
      "grad_norm": 0.275622695684433,
      "learning_rate": 8e-05,
      "loss": 1.6322,
      "step": 2689
    },
    {
      "epoch": 0.5916639172990212,
      "grad_norm": 0.2838039696216583,
      "learning_rate": 8e-05,
      "loss": 1.8021,
      "step": 2690
    },
    {
      "epoch": 0.5918838667106565,
      "grad_norm": 0.290005087852478,
      "learning_rate": 8e-05,
      "loss": 1.6378,
      "step": 2691
    },
    {
      "epoch": 0.5921038161222919,
      "grad_norm": 0.2730334401130676,
      "learning_rate": 8e-05,
      "loss": 1.6665,
      "step": 2692
    },
    {
      "epoch": 0.5923237655339272,
      "grad_norm": 0.27828192710876465,
      "learning_rate": 8e-05,
      "loss": 1.6905,
      "step": 2693
    },
    {
      "epoch": 0.5925437149455626,
      "grad_norm": 0.26481491327285767,
      "learning_rate": 8e-05,
      "loss": 1.5056,
      "step": 2694
    },
    {
      "epoch": 0.5927636643571979,
      "grad_norm": 0.2684583365917206,
      "learning_rate": 8e-05,
      "loss": 1.4181,
      "step": 2695
    },
    {
      "epoch": 0.5929836137688331,
      "grad_norm": 0.2848527431488037,
      "learning_rate": 8e-05,
      "loss": 1.6243,
      "step": 2696
    },
    {
      "epoch": 0.5932035631804685,
      "grad_norm": 0.2943567931652069,
      "learning_rate": 8e-05,
      "loss": 1.7296,
      "step": 2697
    },
    {
      "epoch": 0.5934235125921038,
      "grad_norm": 0.2790435552597046,
      "learning_rate": 8e-05,
      "loss": 1.611,
      "step": 2698
    },
    {
      "epoch": 0.5936434620037392,
      "grad_norm": 0.3020678460597992,
      "learning_rate": 8e-05,
      "loss": 1.8619,
      "step": 2699
    },
    {
      "epoch": 0.5938634114153745,
      "grad_norm": 0.2809624969959259,
      "learning_rate": 8e-05,
      "loss": 1.7253,
      "step": 2700
    },
    {
      "epoch": 0.5940833608270097,
      "grad_norm": 0.2655926048755646,
      "learning_rate": 8e-05,
      "loss": 1.5514,
      "step": 2701
    },
    {
      "epoch": 0.5943033102386451,
      "grad_norm": 0.28663522005081177,
      "learning_rate": 8e-05,
      "loss": 1.6708,
      "step": 2702
    },
    {
      "epoch": 0.5945232596502804,
      "grad_norm": 0.28419819474220276,
      "learning_rate": 8e-05,
      "loss": 1.6551,
      "step": 2703
    },
    {
      "epoch": 0.5947432090619158,
      "grad_norm": 0.29084041714668274,
      "learning_rate": 8e-05,
      "loss": 1.7509,
      "step": 2704
    },
    {
      "epoch": 0.5949631584735511,
      "grad_norm": 0.27892929315567017,
      "learning_rate": 8e-05,
      "loss": 1.6507,
      "step": 2705
    },
    {
      "epoch": 0.5951831078851864,
      "grad_norm": 0.29692748188972473,
      "learning_rate": 8e-05,
      "loss": 1.8667,
      "step": 2706
    },
    {
      "epoch": 0.5954030572968217,
      "grad_norm": 0.2867085933685303,
      "learning_rate": 8e-05,
      "loss": 1.6157,
      "step": 2707
    },
    {
      "epoch": 0.595623006708457,
      "grad_norm": 0.29867735505104065,
      "learning_rate": 8e-05,
      "loss": 1.7813,
      "step": 2708
    },
    {
      "epoch": 0.5958429561200924,
      "grad_norm": 0.28061944246292114,
      "learning_rate": 8e-05,
      "loss": 1.6157,
      "step": 2709
    },
    {
      "epoch": 0.5960629055317277,
      "grad_norm": 0.2807196080684662,
      "learning_rate": 8e-05,
      "loss": 1.5745,
      "step": 2710
    },
    {
      "epoch": 0.596282854943363,
      "grad_norm": 0.2854728698730469,
      "learning_rate": 8e-05,
      "loss": 1.78,
      "step": 2711
    },
    {
      "epoch": 0.5965028043549984,
      "grad_norm": 0.2980540990829468,
      "learning_rate": 8e-05,
      "loss": 1.8421,
      "step": 2712
    },
    {
      "epoch": 0.5967227537666336,
      "grad_norm": 0.2892910838127136,
      "learning_rate": 8e-05,
      "loss": 1.6555,
      "step": 2713
    },
    {
      "epoch": 0.596942703178269,
      "grad_norm": 0.2773078680038452,
      "learning_rate": 8e-05,
      "loss": 1.609,
      "step": 2714
    },
    {
      "epoch": 0.5971626525899043,
      "grad_norm": 0.29283806681632996,
      "learning_rate": 8e-05,
      "loss": 1.7398,
      "step": 2715
    },
    {
      "epoch": 0.5973826020015397,
      "grad_norm": 0.2872734069824219,
      "learning_rate": 8e-05,
      "loss": 1.6864,
      "step": 2716
    },
    {
      "epoch": 0.597602551413175,
      "grad_norm": 0.26770031452178955,
      "learning_rate": 8e-05,
      "loss": 1.5877,
      "step": 2717
    },
    {
      "epoch": 0.5978225008248103,
      "grad_norm": 0.2958748936653137,
      "learning_rate": 8e-05,
      "loss": 1.6565,
      "step": 2718
    },
    {
      "epoch": 0.5980424502364456,
      "grad_norm": 0.30203044414520264,
      "learning_rate": 8e-05,
      "loss": 1.7878,
      "step": 2719
    },
    {
      "epoch": 0.5982623996480809,
      "grad_norm": 0.29320842027664185,
      "learning_rate": 8e-05,
      "loss": 1.5706,
      "step": 2720
    },
    {
      "epoch": 0.5984823490597163,
      "grad_norm": 0.29835638403892517,
      "learning_rate": 8e-05,
      "loss": 1.7271,
      "step": 2721
    },
    {
      "epoch": 0.5987022984713516,
      "grad_norm": 0.36251741647720337,
      "learning_rate": 8e-05,
      "loss": 1.5832,
      "step": 2722
    },
    {
      "epoch": 0.5989222478829869,
      "grad_norm": 0.28875645995140076,
      "learning_rate": 8e-05,
      "loss": 1.8148,
      "step": 2723
    },
    {
      "epoch": 0.5991421972946223,
      "grad_norm": 0.27607399225234985,
      "learning_rate": 8e-05,
      "loss": 1.6024,
      "step": 2724
    },
    {
      "epoch": 0.5993621467062575,
      "grad_norm": 0.290351539850235,
      "learning_rate": 8e-05,
      "loss": 1.7287,
      "step": 2725
    },
    {
      "epoch": 0.5995820961178929,
      "grad_norm": 0.28432413935661316,
      "learning_rate": 8e-05,
      "loss": 1.8603,
      "step": 2726
    },
    {
      "epoch": 0.5998020455295282,
      "grad_norm": 0.2780609130859375,
      "learning_rate": 8e-05,
      "loss": 1.6839,
      "step": 2727
    },
    {
      "epoch": 0.6000219949411635,
      "grad_norm": 0.31952062249183655,
      "learning_rate": 8e-05,
      "loss": 1.585,
      "step": 2728
    },
    {
      "epoch": 0.6002419443527989,
      "grad_norm": 0.2631243169307709,
      "learning_rate": 8e-05,
      "loss": 1.6074,
      "step": 2729
    },
    {
      "epoch": 0.6004618937644342,
      "grad_norm": 0.28518691658973694,
      "learning_rate": 8e-05,
      "loss": 1.6944,
      "step": 2730
    },
    {
      "epoch": 0.6006818431760695,
      "grad_norm": 0.29021504521369934,
      "learning_rate": 8e-05,
      "loss": 1.5919,
      "step": 2731
    },
    {
      "epoch": 0.6009017925877048,
      "grad_norm": 0.2772546410560608,
      "learning_rate": 8e-05,
      "loss": 1.6372,
      "step": 2732
    },
    {
      "epoch": 0.6011217419993401,
      "grad_norm": 0.27938538789749146,
      "learning_rate": 8e-05,
      "loss": 1.7311,
      "step": 2733
    },
    {
      "epoch": 0.6013416914109755,
      "grad_norm": 0.2936658561229706,
      "learning_rate": 8e-05,
      "loss": 1.695,
      "step": 2734
    },
    {
      "epoch": 0.6015616408226108,
      "grad_norm": 0.2893039286136627,
      "learning_rate": 8e-05,
      "loss": 1.6837,
      "step": 2735
    },
    {
      "epoch": 0.6017815902342462,
      "grad_norm": 0.28634974360466003,
      "learning_rate": 8e-05,
      "loss": 1.6155,
      "step": 2736
    },
    {
      "epoch": 0.6020015396458814,
      "grad_norm": 0.2868409752845764,
      "learning_rate": 8e-05,
      "loss": 1.5901,
      "step": 2737
    },
    {
      "epoch": 0.6022214890575168,
      "grad_norm": 0.28888818621635437,
      "learning_rate": 8e-05,
      "loss": 1.6951,
      "step": 2738
    },
    {
      "epoch": 0.6024414384691521,
      "grad_norm": 0.2881872355937958,
      "learning_rate": 8e-05,
      "loss": 1.7648,
      "step": 2739
    },
    {
      "epoch": 0.6026613878807874,
      "grad_norm": 0.29601114988327026,
      "learning_rate": 8e-05,
      "loss": 1.6499,
      "step": 2740
    },
    {
      "epoch": 0.6028813372924228,
      "grad_norm": 0.28861135244369507,
      "learning_rate": 8e-05,
      "loss": 1.6896,
      "step": 2741
    },
    {
      "epoch": 0.603101286704058,
      "grad_norm": 0.30852892994880676,
      "learning_rate": 8e-05,
      "loss": 1.8539,
      "step": 2742
    },
    {
      "epoch": 0.6033212361156934,
      "grad_norm": 0.2659029960632324,
      "learning_rate": 8e-05,
      "loss": 1.5581,
      "step": 2743
    },
    {
      "epoch": 0.6035411855273287,
      "grad_norm": 0.2938629686832428,
      "learning_rate": 8e-05,
      "loss": 1.6893,
      "step": 2744
    },
    {
      "epoch": 0.603761134938964,
      "grad_norm": 0.3215024769306183,
      "learning_rate": 8e-05,
      "loss": 1.809,
      "step": 2745
    },
    {
      "epoch": 0.6039810843505994,
      "grad_norm": 0.3320122957229614,
      "learning_rate": 8e-05,
      "loss": 1.7137,
      "step": 2746
    },
    {
      "epoch": 0.6042010337622347,
      "grad_norm": 0.2901141047477722,
      "learning_rate": 8e-05,
      "loss": 1.7675,
      "step": 2747
    },
    {
      "epoch": 0.60442098317387,
      "grad_norm": 0.2905901074409485,
      "learning_rate": 8e-05,
      "loss": 1.7454,
      "step": 2748
    },
    {
      "epoch": 0.6046409325855053,
      "grad_norm": 0.2820628583431244,
      "learning_rate": 8e-05,
      "loss": 1.7143,
      "step": 2749
    },
    {
      "epoch": 0.6048608819971406,
      "grad_norm": 0.29754188656806946,
      "learning_rate": 8e-05,
      "loss": 1.6957,
      "step": 2750
    },
    {
      "epoch": 0.605080831408776,
      "grad_norm": 0.28644484281539917,
      "learning_rate": 8e-05,
      "loss": 1.864,
      "step": 2751
    },
    {
      "epoch": 0.6053007808204113,
      "grad_norm": 0.2816253900527954,
      "learning_rate": 8e-05,
      "loss": 1.6956,
      "step": 2752
    },
    {
      "epoch": 0.6055207302320467,
      "grad_norm": 0.27785420417785645,
      "learning_rate": 8e-05,
      "loss": 1.6618,
      "step": 2753
    },
    {
      "epoch": 0.605740679643682,
      "grad_norm": 0.2993432283401489,
      "learning_rate": 8e-05,
      "loss": 1.6782,
      "step": 2754
    },
    {
      "epoch": 0.6059606290553172,
      "grad_norm": 0.2837073802947998,
      "learning_rate": 8e-05,
      "loss": 1.6096,
      "step": 2755
    },
    {
      "epoch": 0.6061805784669526,
      "grad_norm": 0.2930501401424408,
      "learning_rate": 8e-05,
      "loss": 1.7036,
      "step": 2756
    },
    {
      "epoch": 0.6064005278785879,
      "grad_norm": 0.2830953299999237,
      "learning_rate": 8e-05,
      "loss": 1.7393,
      "step": 2757
    },
    {
      "epoch": 0.6066204772902233,
      "grad_norm": 0.3069010078907013,
      "learning_rate": 8e-05,
      "loss": 1.5636,
      "step": 2758
    },
    {
      "epoch": 0.6068404267018586,
      "grad_norm": 0.2761766314506531,
      "learning_rate": 8e-05,
      "loss": 1.7233,
      "step": 2759
    },
    {
      "epoch": 0.607060376113494,
      "grad_norm": 0.28254058957099915,
      "learning_rate": 8e-05,
      "loss": 1.7132,
      "step": 2760
    },
    {
      "epoch": 0.6072803255251292,
      "grad_norm": 0.27911651134490967,
      "learning_rate": 8e-05,
      "loss": 1.7782,
      "step": 2761
    },
    {
      "epoch": 0.6075002749367645,
      "grad_norm": 0.2875358462333679,
      "learning_rate": 8e-05,
      "loss": 1.6974,
      "step": 2762
    },
    {
      "epoch": 0.6077202243483999,
      "grad_norm": 0.28940457105636597,
      "learning_rate": 8e-05,
      "loss": 1.6274,
      "step": 2763
    },
    {
      "epoch": 0.6079401737600352,
      "grad_norm": 0.27163782715797424,
      "learning_rate": 8e-05,
      "loss": 1.6507,
      "step": 2764
    },
    {
      "epoch": 0.6081601231716706,
      "grad_norm": 0.2914412021636963,
      "learning_rate": 8e-05,
      "loss": 1.6826,
      "step": 2765
    },
    {
      "epoch": 0.6083800725833058,
      "grad_norm": 0.31414681673049927,
      "learning_rate": 8e-05,
      "loss": 1.8466,
      "step": 2766
    },
    {
      "epoch": 0.6086000219949411,
      "grad_norm": 0.3015105426311493,
      "learning_rate": 8e-05,
      "loss": 1.599,
      "step": 2767
    },
    {
      "epoch": 0.6088199714065765,
      "grad_norm": 0.27743127942085266,
      "learning_rate": 8e-05,
      "loss": 1.5278,
      "step": 2768
    },
    {
      "epoch": 0.6090399208182118,
      "grad_norm": 0.2868049442768097,
      "learning_rate": 8e-05,
      "loss": 1.7008,
      "step": 2769
    },
    {
      "epoch": 0.6092598702298472,
      "grad_norm": 0.2832272946834564,
      "learning_rate": 8e-05,
      "loss": 1.6368,
      "step": 2770
    },
    {
      "epoch": 0.6094798196414825,
      "grad_norm": 0.28054770827293396,
      "learning_rate": 8e-05,
      "loss": 1.6475,
      "step": 2771
    },
    {
      "epoch": 0.6096997690531177,
      "grad_norm": 0.28185421228408813,
      "learning_rate": 8e-05,
      "loss": 1.7731,
      "step": 2772
    },
    {
      "epoch": 0.6099197184647531,
      "grad_norm": 0.2819845676422119,
      "learning_rate": 8e-05,
      "loss": 1.6824,
      "step": 2773
    },
    {
      "epoch": 0.6101396678763884,
      "grad_norm": 0.2764539420604706,
      "learning_rate": 8e-05,
      "loss": 1.7001,
      "step": 2774
    },
    {
      "epoch": 0.6103596172880238,
      "grad_norm": 0.30475977063179016,
      "learning_rate": 8e-05,
      "loss": 1.8297,
      "step": 2775
    },
    {
      "epoch": 0.6105795666996591,
      "grad_norm": 0.2848237454891205,
      "learning_rate": 8e-05,
      "loss": 1.7453,
      "step": 2776
    },
    {
      "epoch": 0.6107995161112943,
      "grad_norm": 0.28268033266067505,
      "learning_rate": 8e-05,
      "loss": 1.5241,
      "step": 2777
    },
    {
      "epoch": 0.6110194655229297,
      "grad_norm": 0.27673062682151794,
      "learning_rate": 8e-05,
      "loss": 1.5273,
      "step": 2778
    },
    {
      "epoch": 0.611239414934565,
      "grad_norm": 0.28202882409095764,
      "learning_rate": 8e-05,
      "loss": 1.5769,
      "step": 2779
    },
    {
      "epoch": 0.6114593643462004,
      "grad_norm": 0.28480303287506104,
      "learning_rate": 8e-05,
      "loss": 1.7456,
      "step": 2780
    },
    {
      "epoch": 0.6116793137578357,
      "grad_norm": 0.3028055727481842,
      "learning_rate": 8e-05,
      "loss": 1.6652,
      "step": 2781
    },
    {
      "epoch": 0.6118992631694711,
      "grad_norm": 0.28677237033843994,
      "learning_rate": 8e-05,
      "loss": 1.6877,
      "step": 2782
    },
    {
      "epoch": 0.6121192125811064,
      "grad_norm": 0.3057413399219513,
      "learning_rate": 8e-05,
      "loss": 1.9811,
      "step": 2783
    },
    {
      "epoch": 0.6123391619927416,
      "grad_norm": 0.2802276313304901,
      "learning_rate": 8e-05,
      "loss": 1.5965,
      "step": 2784
    },
    {
      "epoch": 0.612559111404377,
      "grad_norm": 0.27934229373931885,
      "learning_rate": 8e-05,
      "loss": 1.5959,
      "step": 2785
    },
    {
      "epoch": 0.6127790608160123,
      "grad_norm": 0.2864493429660797,
      "learning_rate": 8e-05,
      "loss": 1.6289,
      "step": 2786
    },
    {
      "epoch": 0.6129990102276477,
      "grad_norm": 0.26668915152549744,
      "learning_rate": 8e-05,
      "loss": 1.4584,
      "step": 2787
    },
    {
      "epoch": 0.613218959639283,
      "grad_norm": 0.28092291951179504,
      "learning_rate": 8e-05,
      "loss": 1.6641,
      "step": 2788
    },
    {
      "epoch": 0.6134389090509182,
      "grad_norm": 0.2933676242828369,
      "learning_rate": 8e-05,
      "loss": 1.9453,
      "step": 2789
    },
    {
      "epoch": 0.6136588584625536,
      "grad_norm": 0.31618431210517883,
      "learning_rate": 8e-05,
      "loss": 1.594,
      "step": 2790
    },
    {
      "epoch": 0.6138788078741889,
      "grad_norm": 0.28090760111808777,
      "learning_rate": 8e-05,
      "loss": 1.7531,
      "step": 2791
    },
    {
      "epoch": 0.6140987572858243,
      "grad_norm": 0.3137405216693878,
      "learning_rate": 8e-05,
      "loss": 1.7243,
      "step": 2792
    },
    {
      "epoch": 0.6143187066974596,
      "grad_norm": 0.2949986755847931,
      "learning_rate": 8e-05,
      "loss": 1.7581,
      "step": 2793
    },
    {
      "epoch": 0.6145386561090949,
      "grad_norm": 0.28396037220954895,
      "learning_rate": 8e-05,
      "loss": 1.6995,
      "step": 2794
    },
    {
      "epoch": 0.6147586055207303,
      "grad_norm": 0.26976051926612854,
      "learning_rate": 8e-05,
      "loss": 1.654,
      "step": 2795
    },
    {
      "epoch": 0.6149785549323655,
      "grad_norm": 0.27323633432388306,
      "learning_rate": 8e-05,
      "loss": 1.6944,
      "step": 2796
    },
    {
      "epoch": 0.6151985043440009,
      "grad_norm": 0.29849350452423096,
      "learning_rate": 8e-05,
      "loss": 1.6127,
      "step": 2797
    },
    {
      "epoch": 0.6154184537556362,
      "grad_norm": 0.28575918078422546,
      "learning_rate": 8e-05,
      "loss": 1.7579,
      "step": 2798
    },
    {
      "epoch": 0.6156384031672715,
      "grad_norm": 0.26723456382751465,
      "learning_rate": 8e-05,
      "loss": 1.5461,
      "step": 2799
    },
    {
      "epoch": 0.6158583525789069,
      "grad_norm": 0.29076528549194336,
      "learning_rate": 8e-05,
      "loss": 1.7001,
      "step": 2800
    },
    {
      "epoch": 0.6160783019905421,
      "grad_norm": 0.27913492918014526,
      "learning_rate": 8e-05,
      "loss": 1.3878,
      "step": 2801
    },
    {
      "epoch": 0.6162982514021775,
      "grad_norm": 0.2841816246509552,
      "learning_rate": 8e-05,
      "loss": 1.6871,
      "step": 2802
    },
    {
      "epoch": 0.6165182008138128,
      "grad_norm": 0.26845458149909973,
      "learning_rate": 8e-05,
      "loss": 1.5518,
      "step": 2803
    },
    {
      "epoch": 0.6167381502254482,
      "grad_norm": 0.30308809876441956,
      "learning_rate": 8e-05,
      "loss": 1.7735,
      "step": 2804
    },
    {
      "epoch": 0.6169580996370835,
      "grad_norm": 0.2812938690185547,
      "learning_rate": 8e-05,
      "loss": 1.7043,
      "step": 2805
    },
    {
      "epoch": 0.6171780490487188,
      "grad_norm": 0.27101054787635803,
      "learning_rate": 8e-05,
      "loss": 1.6156,
      "step": 2806
    },
    {
      "epoch": 0.6173979984603541,
      "grad_norm": 0.2900649607181549,
      "learning_rate": 8e-05,
      "loss": 1.7119,
      "step": 2807
    },
    {
      "epoch": 0.6176179478719894,
      "grad_norm": 0.3011523187160492,
      "learning_rate": 8e-05,
      "loss": 1.7085,
      "step": 2808
    },
    {
      "epoch": 0.6178378972836248,
      "grad_norm": 0.2845047116279602,
      "learning_rate": 8e-05,
      "loss": 1.6691,
      "step": 2809
    },
    {
      "epoch": 0.6180578466952601,
      "grad_norm": 0.31060662865638733,
      "learning_rate": 8e-05,
      "loss": 1.831,
      "step": 2810
    },
    {
      "epoch": 0.6182777961068954,
      "grad_norm": 0.27987706661224365,
      "learning_rate": 8e-05,
      "loss": 1.4881,
      "step": 2811
    },
    {
      "epoch": 0.6184977455185308,
      "grad_norm": 0.3197080194950104,
      "learning_rate": 8e-05,
      "loss": 1.7112,
      "step": 2812
    },
    {
      "epoch": 0.618717694930166,
      "grad_norm": 0.31402066349983215,
      "learning_rate": 8e-05,
      "loss": 1.6535,
      "step": 2813
    },
    {
      "epoch": 0.6189376443418014,
      "grad_norm": 0.303529292345047,
      "learning_rate": 8e-05,
      "loss": 1.6563,
      "step": 2814
    },
    {
      "epoch": 0.6191575937534367,
      "grad_norm": 0.26674556732177734,
      "learning_rate": 8e-05,
      "loss": 1.5202,
      "step": 2815
    },
    {
      "epoch": 0.619377543165072,
      "grad_norm": 0.30466997623443604,
      "learning_rate": 8e-05,
      "loss": 1.6014,
      "step": 2816
    },
    {
      "epoch": 0.6195974925767074,
      "grad_norm": 0.2991195619106293,
      "learning_rate": 8e-05,
      "loss": 1.765,
      "step": 2817
    },
    {
      "epoch": 0.6198174419883427,
      "grad_norm": 0.30000337958335876,
      "learning_rate": 8e-05,
      "loss": 1.7794,
      "step": 2818
    },
    {
      "epoch": 0.620037391399978,
      "grad_norm": 0.29237842559814453,
      "learning_rate": 8e-05,
      "loss": 1.8753,
      "step": 2819
    },
    {
      "epoch": 0.6202573408116133,
      "grad_norm": 0.2896344065666199,
      "learning_rate": 8e-05,
      "loss": 1.6137,
      "step": 2820
    },
    {
      "epoch": 0.6204772902232486,
      "grad_norm": 0.34269601106643677,
      "learning_rate": 8e-05,
      "loss": 1.7693,
      "step": 2821
    },
    {
      "epoch": 0.620697239634884,
      "grad_norm": 0.30044153332710266,
      "learning_rate": 8e-05,
      "loss": 1.7286,
      "step": 2822
    },
    {
      "epoch": 0.6209171890465193,
      "grad_norm": 0.2616185247898102,
      "learning_rate": 8e-05,
      "loss": 1.5373,
      "step": 2823
    },
    {
      "epoch": 0.6211371384581547,
      "grad_norm": 0.3217238485813141,
      "learning_rate": 8e-05,
      "loss": 1.7681,
      "step": 2824
    },
    {
      "epoch": 0.6213570878697899,
      "grad_norm": 0.284446120262146,
      "learning_rate": 8e-05,
      "loss": 1.5597,
      "step": 2825
    },
    {
      "epoch": 0.6215770372814253,
      "grad_norm": 0.28698036074638367,
      "learning_rate": 8e-05,
      "loss": 1.6844,
      "step": 2826
    },
    {
      "epoch": 0.6217969866930606,
      "grad_norm": 0.2828524708747864,
      "learning_rate": 8e-05,
      "loss": 1.7142,
      "step": 2827
    },
    {
      "epoch": 0.6220169361046959,
      "grad_norm": 0.3004125952720642,
      "learning_rate": 8e-05,
      "loss": 1.6812,
      "step": 2828
    },
    {
      "epoch": 0.6222368855163313,
      "grad_norm": 0.30438825488090515,
      "learning_rate": 8e-05,
      "loss": 1.6696,
      "step": 2829
    },
    {
      "epoch": 0.6224568349279666,
      "grad_norm": 0.2654431164264679,
      "learning_rate": 8e-05,
      "loss": 1.5656,
      "step": 2830
    },
    {
      "epoch": 0.6226767843396019,
      "grad_norm": 0.28561410307884216,
      "learning_rate": 8e-05,
      "loss": 1.5924,
      "step": 2831
    },
    {
      "epoch": 0.6228967337512372,
      "grad_norm": 0.29075953364372253,
      "learning_rate": 8e-05,
      "loss": 1.6026,
      "step": 2832
    },
    {
      "epoch": 0.6231166831628725,
      "grad_norm": 0.3002355098724365,
      "learning_rate": 8e-05,
      "loss": 1.7783,
      "step": 2833
    },
    {
      "epoch": 0.6233366325745079,
      "grad_norm": 0.2757151424884796,
      "learning_rate": 8e-05,
      "loss": 1.6404,
      "step": 2834
    },
    {
      "epoch": 0.6235565819861432,
      "grad_norm": 0.28108781576156616,
      "learning_rate": 8e-05,
      "loss": 1.6727,
      "step": 2835
    },
    {
      "epoch": 0.6237765313977786,
      "grad_norm": 0.31818297505378723,
      "learning_rate": 8e-05,
      "loss": 1.8859,
      "step": 2836
    },
    {
      "epoch": 0.6239964808094138,
      "grad_norm": 0.30283015966415405,
      "learning_rate": 8e-05,
      "loss": 1.6967,
      "step": 2837
    },
    {
      "epoch": 0.6242164302210491,
      "grad_norm": 0.28991082310676575,
      "learning_rate": 8e-05,
      "loss": 1.7564,
      "step": 2838
    },
    {
      "epoch": 0.6244363796326845,
      "grad_norm": 0.27985113859176636,
      "learning_rate": 8e-05,
      "loss": 1.645,
      "step": 2839
    },
    {
      "epoch": 0.6246563290443198,
      "grad_norm": 0.28599318861961365,
      "learning_rate": 8e-05,
      "loss": 1.7069,
      "step": 2840
    },
    {
      "epoch": 0.6248762784559552,
      "grad_norm": 0.291962593793869,
      "learning_rate": 8e-05,
      "loss": 1.7303,
      "step": 2841
    },
    {
      "epoch": 0.6250962278675904,
      "grad_norm": 0.2977605164051056,
      "learning_rate": 8e-05,
      "loss": 1.6251,
      "step": 2842
    },
    {
      "epoch": 0.6253161772792257,
      "grad_norm": 0.280979186296463,
      "learning_rate": 8e-05,
      "loss": 1.6132,
      "step": 2843
    },
    {
      "epoch": 0.6255361266908611,
      "grad_norm": 0.30565154552459717,
      "learning_rate": 8e-05,
      "loss": 1.8351,
      "step": 2844
    },
    {
      "epoch": 0.6257560761024964,
      "grad_norm": 0.2870398759841919,
      "learning_rate": 8e-05,
      "loss": 1.7169,
      "step": 2845
    },
    {
      "epoch": 0.6259760255141318,
      "grad_norm": 0.2740568518638611,
      "learning_rate": 8e-05,
      "loss": 1.6666,
      "step": 2846
    },
    {
      "epoch": 0.6261959749257671,
      "grad_norm": 0.27255693078041077,
      "learning_rate": 8e-05,
      "loss": 1.6112,
      "step": 2847
    },
    {
      "epoch": 0.6264159243374025,
      "grad_norm": 0.2785317003726959,
      "learning_rate": 8e-05,
      "loss": 1.6532,
      "step": 2848
    },
    {
      "epoch": 0.6266358737490377,
      "grad_norm": 0.2979902923107147,
      "learning_rate": 8e-05,
      "loss": 1.7981,
      "step": 2849
    },
    {
      "epoch": 0.626855823160673,
      "grad_norm": 0.29625701904296875,
      "learning_rate": 8e-05,
      "loss": 1.7212,
      "step": 2850
    },
    {
      "epoch": 0.6270757725723084,
      "grad_norm": 0.2768239676952362,
      "learning_rate": 8e-05,
      "loss": 1.6358,
      "step": 2851
    },
    {
      "epoch": 0.6272957219839437,
      "grad_norm": 0.2931036055088043,
      "learning_rate": 8e-05,
      "loss": 1.7728,
      "step": 2852
    },
    {
      "epoch": 0.6275156713955791,
      "grad_norm": 0.2883271872997284,
      "learning_rate": 8e-05,
      "loss": 1.6403,
      "step": 2853
    },
    {
      "epoch": 0.6277356208072143,
      "grad_norm": 0.31137219071388245,
      "learning_rate": 8e-05,
      "loss": 1.6454,
      "step": 2854
    },
    {
      "epoch": 0.6279555702188496,
      "grad_norm": 0.3026840388774872,
      "learning_rate": 8e-05,
      "loss": 1.6534,
      "step": 2855
    },
    {
      "epoch": 0.628175519630485,
      "grad_norm": 0.2950657606124878,
      "learning_rate": 8e-05,
      "loss": 1.7298,
      "step": 2856
    },
    {
      "epoch": 0.6283954690421203,
      "grad_norm": 0.29347553849220276,
      "learning_rate": 8e-05,
      "loss": 1.58,
      "step": 2857
    },
    {
      "epoch": 0.6286154184537557,
      "grad_norm": 0.28075262904167175,
      "learning_rate": 8e-05,
      "loss": 1.6262,
      "step": 2858
    },
    {
      "epoch": 0.628835367865391,
      "grad_norm": 0.26556506752967834,
      "learning_rate": 8e-05,
      "loss": 1.5666,
      "step": 2859
    },
    {
      "epoch": 0.6290553172770262,
      "grad_norm": 0.28918468952178955,
      "learning_rate": 8e-05,
      "loss": 1.6918,
      "step": 2860
    },
    {
      "epoch": 0.6292752666886616,
      "grad_norm": 0.2816839814186096,
      "learning_rate": 8e-05,
      "loss": 1.6387,
      "step": 2861
    },
    {
      "epoch": 0.6294952161002969,
      "grad_norm": 0.2819633185863495,
      "learning_rate": 8e-05,
      "loss": 1.6945,
      "step": 2862
    },
    {
      "epoch": 0.6297151655119323,
      "grad_norm": 0.2847195863723755,
      "learning_rate": 8e-05,
      "loss": 1.744,
      "step": 2863
    },
    {
      "epoch": 0.6299351149235676,
      "grad_norm": 0.2706061899662018,
      "learning_rate": 8e-05,
      "loss": 1.6246,
      "step": 2864
    },
    {
      "epoch": 0.6301550643352029,
      "grad_norm": 0.281125545501709,
      "learning_rate": 8e-05,
      "loss": 1.49,
      "step": 2865
    },
    {
      "epoch": 0.6303750137468382,
      "grad_norm": 0.2861780822277069,
      "learning_rate": 8e-05,
      "loss": 1.7375,
      "step": 2866
    },
    {
      "epoch": 0.6305949631584735,
      "grad_norm": 0.2654918134212494,
      "learning_rate": 8e-05,
      "loss": 1.5997,
      "step": 2867
    },
    {
      "epoch": 0.6308149125701089,
      "grad_norm": 0.29169219732284546,
      "learning_rate": 8e-05,
      "loss": 1.8921,
      "step": 2868
    },
    {
      "epoch": 0.6310348619817442,
      "grad_norm": 0.2858426570892334,
      "learning_rate": 8e-05,
      "loss": 1.8006,
      "step": 2869
    },
    {
      "epoch": 0.6312548113933796,
      "grad_norm": 0.2712969183921814,
      "learning_rate": 8e-05,
      "loss": 1.5761,
      "step": 2870
    },
    {
      "epoch": 0.6314747608050149,
      "grad_norm": 0.28961536288261414,
      "learning_rate": 8e-05,
      "loss": 1.618,
      "step": 2871
    },
    {
      "epoch": 0.6316947102166501,
      "grad_norm": 0.2879860997200012,
      "learning_rate": 8e-05,
      "loss": 1.6116,
      "step": 2872
    },
    {
      "epoch": 0.6319146596282855,
      "grad_norm": 0.3009500801563263,
      "learning_rate": 8e-05,
      "loss": 1.8099,
      "step": 2873
    },
    {
      "epoch": 0.6321346090399208,
      "grad_norm": 0.3012961149215698,
      "learning_rate": 8e-05,
      "loss": 1.8123,
      "step": 2874
    },
    {
      "epoch": 0.6323545584515562,
      "grad_norm": 0.27382341027259827,
      "learning_rate": 8e-05,
      "loss": 1.5486,
      "step": 2875
    },
    {
      "epoch": 0.6325745078631915,
      "grad_norm": 0.45538756251335144,
      "learning_rate": 8e-05,
      "loss": 1.7382,
      "step": 2876
    },
    {
      "epoch": 0.6327944572748267,
      "grad_norm": 0.27454543113708496,
      "learning_rate": 8e-05,
      "loss": 1.6488,
      "step": 2877
    },
    {
      "epoch": 0.6330144066864621,
      "grad_norm": 0.28111204504966736,
      "learning_rate": 8e-05,
      "loss": 1.7262,
      "step": 2878
    },
    {
      "epoch": 0.6332343560980974,
      "grad_norm": 0.2855817675590515,
      "learning_rate": 8e-05,
      "loss": 1.6841,
      "step": 2879
    },
    {
      "epoch": 0.6334543055097328,
      "grad_norm": 0.3017145097255707,
      "learning_rate": 8e-05,
      "loss": 1.759,
      "step": 2880
    },
    {
      "epoch": 0.6336742549213681,
      "grad_norm": 0.27578651905059814,
      "learning_rate": 8e-05,
      "loss": 1.5341,
      "step": 2881
    },
    {
      "epoch": 0.6338942043330034,
      "grad_norm": 0.28522011637687683,
      "learning_rate": 8e-05,
      "loss": 1.5042,
      "step": 2882
    },
    {
      "epoch": 0.6341141537446388,
      "grad_norm": 0.28013676404953003,
      "learning_rate": 8e-05,
      "loss": 1.591,
      "step": 2883
    },
    {
      "epoch": 0.634334103156274,
      "grad_norm": 0.30440640449523926,
      "learning_rate": 8e-05,
      "loss": 1.6698,
      "step": 2884
    },
    {
      "epoch": 0.6345540525679094,
      "grad_norm": 0.28555527329444885,
      "learning_rate": 8e-05,
      "loss": 1.6199,
      "step": 2885
    },
    {
      "epoch": 0.6347740019795447,
      "grad_norm": 0.31451916694641113,
      "learning_rate": 8e-05,
      "loss": 1.6717,
      "step": 2886
    },
    {
      "epoch": 0.63499395139118,
      "grad_norm": 0.3116842806339264,
      "learning_rate": 8e-05,
      "loss": 1.7043,
      "step": 2887
    },
    {
      "epoch": 0.6352139008028154,
      "grad_norm": 0.30441299080848694,
      "learning_rate": 8e-05,
      "loss": 1.6953,
      "step": 2888
    },
    {
      "epoch": 0.6354338502144506,
      "grad_norm": 0.2890806496143341,
      "learning_rate": 8e-05,
      "loss": 1.7363,
      "step": 2889
    },
    {
      "epoch": 0.635653799626086,
      "grad_norm": 0.2715187072753906,
      "learning_rate": 8e-05,
      "loss": 1.5412,
      "step": 2890
    },
    {
      "epoch": 0.6358737490377213,
      "grad_norm": 0.32213905453681946,
      "learning_rate": 8e-05,
      "loss": 1.7735,
      "step": 2891
    },
    {
      "epoch": 0.6360936984493566,
      "grad_norm": 0.28850191831588745,
      "learning_rate": 8e-05,
      "loss": 1.5156,
      "step": 2892
    },
    {
      "epoch": 0.636313647860992,
      "grad_norm": 0.2934744358062744,
      "learning_rate": 8e-05,
      "loss": 1.6822,
      "step": 2893
    },
    {
      "epoch": 0.6365335972726273,
      "grad_norm": 0.29068851470947266,
      "learning_rate": 8e-05,
      "loss": 1.7723,
      "step": 2894
    },
    {
      "epoch": 0.6367535466842627,
      "grad_norm": 0.28490251302719116,
      "learning_rate": 8e-05,
      "loss": 1.7666,
      "step": 2895
    },
    {
      "epoch": 0.6369734960958979,
      "grad_norm": 0.28677448630332947,
      "learning_rate": 8e-05,
      "loss": 1.6792,
      "step": 2896
    },
    {
      "epoch": 0.6371934455075333,
      "grad_norm": 0.29424387216567993,
      "learning_rate": 8e-05,
      "loss": 1.6224,
      "step": 2897
    },
    {
      "epoch": 0.6374133949191686,
      "grad_norm": 0.2872456908226013,
      "learning_rate": 8e-05,
      "loss": 1.6677,
      "step": 2898
    },
    {
      "epoch": 0.6376333443308039,
      "grad_norm": 0.26886799931526184,
      "learning_rate": 8e-05,
      "loss": 1.5086,
      "step": 2899
    },
    {
      "epoch": 0.6378532937424393,
      "grad_norm": 0.2737233638763428,
      "learning_rate": 8e-05,
      "loss": 1.6832,
      "step": 2900
    },
    {
      "epoch": 0.6380732431540745,
      "grad_norm": 0.2912994623184204,
      "learning_rate": 8e-05,
      "loss": 1.5945,
      "step": 2901
    },
    {
      "epoch": 0.6382931925657099,
      "grad_norm": 0.2800372540950775,
      "learning_rate": 8e-05,
      "loss": 1.5661,
      "step": 2902
    },
    {
      "epoch": 0.6385131419773452,
      "grad_norm": 0.3248150050640106,
      "learning_rate": 8e-05,
      "loss": 1.6098,
      "step": 2903
    },
    {
      "epoch": 0.6387330913889805,
      "grad_norm": 0.27953609824180603,
      "learning_rate": 8e-05,
      "loss": 1.7255,
      "step": 2904
    },
    {
      "epoch": 0.6389530408006159,
      "grad_norm": 0.276395708322525,
      "learning_rate": 8e-05,
      "loss": 1.6782,
      "step": 2905
    },
    {
      "epoch": 0.6391729902122512,
      "grad_norm": 0.2754693031311035,
      "learning_rate": 8e-05,
      "loss": 1.5135,
      "step": 2906
    },
    {
      "epoch": 0.6393929396238865,
      "grad_norm": 0.2756873369216919,
      "learning_rate": 8e-05,
      "loss": 1.6208,
      "step": 2907
    },
    {
      "epoch": 0.6396128890355218,
      "grad_norm": 0.3032161593437195,
      "learning_rate": 8e-05,
      "loss": 1.7675,
      "step": 2908
    },
    {
      "epoch": 0.6398328384471571,
      "grad_norm": 0.2915925085544586,
      "learning_rate": 8e-05,
      "loss": 1.6899,
      "step": 2909
    },
    {
      "epoch": 0.6400527878587925,
      "grad_norm": 0.28415006399154663,
      "learning_rate": 8e-05,
      "loss": 1.7218,
      "step": 2910
    },
    {
      "epoch": 0.6402727372704278,
      "grad_norm": 0.3023785352706909,
      "learning_rate": 8e-05,
      "loss": 1.7959,
      "step": 2911
    },
    {
      "epoch": 0.6404926866820632,
      "grad_norm": 0.2656283974647522,
      "learning_rate": 8e-05,
      "loss": 1.6287,
      "step": 2912
    },
    {
      "epoch": 0.6407126360936984,
      "grad_norm": 0.2835081219673157,
      "learning_rate": 8e-05,
      "loss": 1.6597,
      "step": 2913
    },
    {
      "epoch": 0.6409325855053337,
      "grad_norm": 0.2756771743297577,
      "learning_rate": 8e-05,
      "loss": 1.625,
      "step": 2914
    },
    {
      "epoch": 0.6411525349169691,
      "grad_norm": 0.283149778842926,
      "learning_rate": 8e-05,
      "loss": 1.662,
      "step": 2915
    },
    {
      "epoch": 0.6413724843286044,
      "grad_norm": 0.28902921080589294,
      "learning_rate": 8e-05,
      "loss": 1.4862,
      "step": 2916
    },
    {
      "epoch": 0.6415924337402398,
      "grad_norm": 0.28932076692581177,
      "learning_rate": 8e-05,
      "loss": 1.6109,
      "step": 2917
    },
    {
      "epoch": 0.641812383151875,
      "grad_norm": 0.30964934825897217,
      "learning_rate": 8e-05,
      "loss": 1.7021,
      "step": 2918
    },
    {
      "epoch": 0.6420323325635104,
      "grad_norm": 0.28258854150772095,
      "learning_rate": 8e-05,
      "loss": 1.6536,
      "step": 2919
    },
    {
      "epoch": 0.6422522819751457,
      "grad_norm": 0.2939313054084778,
      "learning_rate": 8e-05,
      "loss": 1.6299,
      "step": 2920
    },
    {
      "epoch": 0.642472231386781,
      "grad_norm": 0.31722190976142883,
      "learning_rate": 8e-05,
      "loss": 1.6789,
      "step": 2921
    },
    {
      "epoch": 0.6426921807984164,
      "grad_norm": 0.29024428129196167,
      "learning_rate": 8e-05,
      "loss": 1.7429,
      "step": 2922
    },
    {
      "epoch": 0.6429121302100517,
      "grad_norm": 0.2716485559940338,
      "learning_rate": 8e-05,
      "loss": 1.619,
      "step": 2923
    },
    {
      "epoch": 0.6431320796216871,
      "grad_norm": 0.2986311614513397,
      "learning_rate": 8e-05,
      "loss": 1.7574,
      "step": 2924
    },
    {
      "epoch": 0.6433520290333223,
      "grad_norm": 0.29542550444602966,
      "learning_rate": 8e-05,
      "loss": 1.7322,
      "step": 2925
    },
    {
      "epoch": 0.6435719784449576,
      "grad_norm": 0.27215078473091125,
      "learning_rate": 8e-05,
      "loss": 1.5544,
      "step": 2926
    },
    {
      "epoch": 0.643791927856593,
      "grad_norm": 0.29105404019355774,
      "learning_rate": 8e-05,
      "loss": 1.7499,
      "step": 2927
    },
    {
      "epoch": 0.6440118772682283,
      "grad_norm": 0.2990782558917999,
      "learning_rate": 8e-05,
      "loss": 1.7987,
      "step": 2928
    },
    {
      "epoch": 0.6442318266798637,
      "grad_norm": 0.27296003699302673,
      "learning_rate": 8e-05,
      "loss": 1.6271,
      "step": 2929
    },
    {
      "epoch": 0.644451776091499,
      "grad_norm": 0.27194517850875854,
      "learning_rate": 8e-05,
      "loss": 1.5178,
      "step": 2930
    },
    {
      "epoch": 0.6446717255031342,
      "grad_norm": 0.2720150053501129,
      "learning_rate": 8e-05,
      "loss": 1.5665,
      "step": 2931
    },
    {
      "epoch": 0.6448916749147696,
      "grad_norm": 0.2825513780117035,
      "learning_rate": 8e-05,
      "loss": 1.6276,
      "step": 2932
    },
    {
      "epoch": 0.6451116243264049,
      "grad_norm": 0.2869420349597931,
      "learning_rate": 8e-05,
      "loss": 1.5306,
      "step": 2933
    },
    {
      "epoch": 0.6453315737380403,
      "grad_norm": 0.2829979360103607,
      "learning_rate": 8e-05,
      "loss": 1.7132,
      "step": 2934
    },
    {
      "epoch": 0.6455515231496756,
      "grad_norm": 0.28047260642051697,
      "learning_rate": 8e-05,
      "loss": 1.5932,
      "step": 2935
    },
    {
      "epoch": 0.6457714725613108,
      "grad_norm": 0.287765234708786,
      "learning_rate": 8e-05,
      "loss": 1.6029,
      "step": 2936
    },
    {
      "epoch": 0.6459914219729462,
      "grad_norm": 0.2858487665653229,
      "learning_rate": 8e-05,
      "loss": 1.5959,
      "step": 2937
    },
    {
      "epoch": 0.6462113713845815,
      "grad_norm": 0.29041311144828796,
      "learning_rate": 8e-05,
      "loss": 1.6348,
      "step": 2938
    },
    {
      "epoch": 0.6464313207962169,
      "grad_norm": 0.2873425781726837,
      "learning_rate": 8e-05,
      "loss": 1.6514,
      "step": 2939
    },
    {
      "epoch": 0.6466512702078522,
      "grad_norm": 0.2767978310585022,
      "learning_rate": 8e-05,
      "loss": 1.5471,
      "step": 2940
    },
    {
      "epoch": 0.6468712196194876,
      "grad_norm": 0.27061501145362854,
      "learning_rate": 8e-05,
      "loss": 1.5439,
      "step": 2941
    },
    {
      "epoch": 0.6470911690311228,
      "grad_norm": 0.2724677324295044,
      "learning_rate": 8e-05,
      "loss": 1.6808,
      "step": 2942
    },
    {
      "epoch": 0.6473111184427581,
      "grad_norm": 0.2804121971130371,
      "learning_rate": 8e-05,
      "loss": 1.7923,
      "step": 2943
    },
    {
      "epoch": 0.6475310678543935,
      "grad_norm": 0.2881599962711334,
      "learning_rate": 8e-05,
      "loss": 1.745,
      "step": 2944
    },
    {
      "epoch": 0.6477510172660288,
      "grad_norm": 0.3064921200275421,
      "learning_rate": 8e-05,
      "loss": 1.7387,
      "step": 2945
    },
    {
      "epoch": 0.6479709666776642,
      "grad_norm": 0.2994825541973114,
      "learning_rate": 8e-05,
      "loss": 1.7143,
      "step": 2946
    },
    {
      "epoch": 0.6481909160892995,
      "grad_norm": 0.29315468668937683,
      "learning_rate": 8e-05,
      "loss": 1.7712,
      "step": 2947
    },
    {
      "epoch": 0.6484108655009347,
      "grad_norm": 0.2923111915588379,
      "learning_rate": 8e-05,
      "loss": 1.6861,
      "step": 2948
    },
    {
      "epoch": 0.6486308149125701,
      "grad_norm": 0.2861957848072052,
      "learning_rate": 8e-05,
      "loss": 1.5951,
      "step": 2949
    },
    {
      "epoch": 0.6488507643242054,
      "grad_norm": 0.2978787422180176,
      "learning_rate": 8e-05,
      "loss": 1.6617,
      "step": 2950
    },
    {
      "epoch": 0.6490707137358408,
      "grad_norm": 0.28596314787864685,
      "learning_rate": 8e-05,
      "loss": 1.7623,
      "step": 2951
    },
    {
      "epoch": 0.6492906631474761,
      "grad_norm": 0.29754844307899475,
      "learning_rate": 8e-05,
      "loss": 1.6497,
      "step": 2952
    },
    {
      "epoch": 0.6495106125591114,
      "grad_norm": 0.2931132912635803,
      "learning_rate": 8e-05,
      "loss": 1.5972,
      "step": 2953
    },
    {
      "epoch": 0.6497305619707467,
      "grad_norm": 0.2667228877544403,
      "learning_rate": 8e-05,
      "loss": 1.5598,
      "step": 2954
    },
    {
      "epoch": 0.649950511382382,
      "grad_norm": 0.2866271436214447,
      "learning_rate": 8e-05,
      "loss": 1.5908,
      "step": 2955
    },
    {
      "epoch": 0.6501704607940174,
      "grad_norm": 0.28429698944091797,
      "learning_rate": 8e-05,
      "loss": 1.6582,
      "step": 2956
    },
    {
      "epoch": 0.6503904102056527,
      "grad_norm": 0.27636975049972534,
      "learning_rate": 8e-05,
      "loss": 1.6778,
      "step": 2957
    },
    {
      "epoch": 0.650610359617288,
      "grad_norm": 0.3005516529083252,
      "learning_rate": 8e-05,
      "loss": 1.7191,
      "step": 2958
    },
    {
      "epoch": 0.6508303090289234,
      "grad_norm": 0.28478094935417175,
      "learning_rate": 8e-05,
      "loss": 1.5651,
      "step": 2959
    },
    {
      "epoch": 0.6510502584405586,
      "grad_norm": 0.2879832088947296,
      "learning_rate": 8e-05,
      "loss": 1.7055,
      "step": 2960
    },
    {
      "epoch": 0.651270207852194,
      "grad_norm": 0.2899249196052551,
      "learning_rate": 8e-05,
      "loss": 1.6682,
      "step": 2961
    },
    {
      "epoch": 0.6514901572638293,
      "grad_norm": 0.26806798577308655,
      "learning_rate": 8e-05,
      "loss": 1.6962,
      "step": 2962
    },
    {
      "epoch": 0.6517101066754647,
      "grad_norm": 0.2929481565952301,
      "learning_rate": 8e-05,
      "loss": 1.6388,
      "step": 2963
    },
    {
      "epoch": 0.6519300560871,
      "grad_norm": 0.2920469641685486,
      "learning_rate": 8e-05,
      "loss": 1.8584,
      "step": 2964
    },
    {
      "epoch": 0.6521500054987353,
      "grad_norm": 0.285696417093277,
      "learning_rate": 8e-05,
      "loss": 1.6491,
      "step": 2965
    },
    {
      "epoch": 0.6523699549103706,
      "grad_norm": 0.2991807162761688,
      "learning_rate": 8e-05,
      "loss": 1.6565,
      "step": 2966
    },
    {
      "epoch": 0.6525899043220059,
      "grad_norm": 0.27987217903137207,
      "learning_rate": 8e-05,
      "loss": 1.6274,
      "step": 2967
    },
    {
      "epoch": 0.6528098537336413,
      "grad_norm": 0.2810576856136322,
      "learning_rate": 8e-05,
      "loss": 1.5294,
      "step": 2968
    },
    {
      "epoch": 0.6530298031452766,
      "grad_norm": 0.2755715847015381,
      "learning_rate": 8e-05,
      "loss": 1.6319,
      "step": 2969
    },
    {
      "epoch": 0.6532497525569119,
      "grad_norm": 0.3041331171989441,
      "learning_rate": 8e-05,
      "loss": 1.5308,
      "step": 2970
    },
    {
      "epoch": 0.6534697019685473,
      "grad_norm": 0.2858032286167145,
      "learning_rate": 8e-05,
      "loss": 1.6557,
      "step": 2971
    },
    {
      "epoch": 0.6536896513801825,
      "grad_norm": 0.3001968562602997,
      "learning_rate": 8e-05,
      "loss": 1.6841,
      "step": 2972
    },
    {
      "epoch": 0.6539096007918179,
      "grad_norm": 0.29567384719848633,
      "learning_rate": 8e-05,
      "loss": 1.6329,
      "step": 2973
    },
    {
      "epoch": 0.6541295502034532,
      "grad_norm": 0.29874905943870544,
      "learning_rate": 8e-05,
      "loss": 1.7178,
      "step": 2974
    },
    {
      "epoch": 0.6543494996150885,
      "grad_norm": 0.34721627831459045,
      "learning_rate": 8e-05,
      "loss": 1.8192,
      "step": 2975
    },
    {
      "epoch": 0.6545694490267239,
      "grad_norm": 0.2965874671936035,
      "learning_rate": 8e-05,
      "loss": 1.8156,
      "step": 2976
    },
    {
      "epoch": 0.6547893984383591,
      "grad_norm": 0.2710880637168884,
      "learning_rate": 8e-05,
      "loss": 1.5101,
      "step": 2977
    },
    {
      "epoch": 0.6550093478499945,
      "grad_norm": 0.2852049171924591,
      "learning_rate": 8e-05,
      "loss": 1.5835,
      "step": 2978
    },
    {
      "epoch": 0.6552292972616298,
      "grad_norm": 0.2898092567920685,
      "learning_rate": 8e-05,
      "loss": 1.599,
      "step": 2979
    },
    {
      "epoch": 0.6554492466732651,
      "grad_norm": 0.2880117893218994,
      "learning_rate": 8e-05,
      "loss": 1.5904,
      "step": 2980
    },
    {
      "epoch": 0.6556691960849005,
      "grad_norm": 0.2850951850414276,
      "learning_rate": 8e-05,
      "loss": 1.5551,
      "step": 2981
    },
    {
      "epoch": 0.6558891454965358,
      "grad_norm": 0.292883038520813,
      "learning_rate": 8e-05,
      "loss": 1.5587,
      "step": 2982
    },
    {
      "epoch": 0.6561090949081712,
      "grad_norm": 0.3050660490989685,
      "learning_rate": 8e-05,
      "loss": 1.7904,
      "step": 2983
    },
    {
      "epoch": 0.6563290443198064,
      "grad_norm": 0.31059566140174866,
      "learning_rate": 8e-05,
      "loss": 1.7716,
      "step": 2984
    },
    {
      "epoch": 0.6565489937314418,
      "grad_norm": 0.33118265867233276,
      "learning_rate": 8e-05,
      "loss": 1.7878,
      "step": 2985
    },
    {
      "epoch": 0.6567689431430771,
      "grad_norm": 0.27835318446159363,
      "learning_rate": 8e-05,
      "loss": 1.7069,
      "step": 2986
    },
    {
      "epoch": 0.6569888925547124,
      "grad_norm": 0.33706921339035034,
      "learning_rate": 8e-05,
      "loss": 1.7135,
      "step": 2987
    },
    {
      "epoch": 0.6572088419663478,
      "grad_norm": 0.3115323781967163,
      "learning_rate": 8e-05,
      "loss": 1.7396,
      "step": 2988
    },
    {
      "epoch": 0.657428791377983,
      "grad_norm": 0.2827862501144409,
      "learning_rate": 8e-05,
      "loss": 1.6889,
      "step": 2989
    },
    {
      "epoch": 0.6576487407896184,
      "grad_norm": 0.29057440161705017,
      "learning_rate": 8e-05,
      "loss": 1.7208,
      "step": 2990
    },
    {
      "epoch": 0.6578686902012537,
      "grad_norm": 0.2977316081523895,
      "learning_rate": 8e-05,
      "loss": 1.7192,
      "step": 2991
    },
    {
      "epoch": 0.658088639612889,
      "grad_norm": 0.296475887298584,
      "learning_rate": 8e-05,
      "loss": 1.6815,
      "step": 2992
    },
    {
      "epoch": 0.6583085890245244,
      "grad_norm": 0.29579752683639526,
      "learning_rate": 8e-05,
      "loss": 1.781,
      "step": 2993
    },
    {
      "epoch": 0.6585285384361597,
      "grad_norm": 0.2853552997112274,
      "learning_rate": 8e-05,
      "loss": 1.69,
      "step": 2994
    },
    {
      "epoch": 0.658748487847795,
      "grad_norm": 0.2831558883190155,
      "learning_rate": 8e-05,
      "loss": 1.7118,
      "step": 2995
    },
    {
      "epoch": 0.6589684372594303,
      "grad_norm": 0.30975469946861267,
      "learning_rate": 8e-05,
      "loss": 1.7153,
      "step": 2996
    },
    {
      "epoch": 0.6591883866710656,
      "grad_norm": 0.287047803401947,
      "learning_rate": 8e-05,
      "loss": 1.7241,
      "step": 2997
    },
    {
      "epoch": 0.659408336082701,
      "grad_norm": 0.2812976837158203,
      "learning_rate": 8e-05,
      "loss": 1.6536,
      "step": 2998
    },
    {
      "epoch": 0.6596282854943363,
      "grad_norm": 0.2794138491153717,
      "learning_rate": 8e-05,
      "loss": 1.6176,
      "step": 2999
    },
    {
      "epoch": 0.6598482349059717,
      "grad_norm": 0.2949649691581726,
      "learning_rate": 8e-05,
      "loss": 1.6499,
      "step": 3000
    },
    {
      "epoch": 0.6600681843176069,
      "grad_norm": 0.2770403027534485,
      "learning_rate": 8e-05,
      "loss": 1.6235,
      "step": 3001
    },
    {
      "epoch": 0.6602881337292422,
      "grad_norm": 0.30226513743400574,
      "learning_rate": 8e-05,
      "loss": 1.7714,
      "step": 3002
    },
    {
      "epoch": 0.6605080831408776,
      "grad_norm": 0.27673766016960144,
      "learning_rate": 8e-05,
      "loss": 1.6372,
      "step": 3003
    },
    {
      "epoch": 0.6607280325525129,
      "grad_norm": 0.2805183231830597,
      "learning_rate": 8e-05,
      "loss": 1.5495,
      "step": 3004
    },
    {
      "epoch": 0.6609479819641483,
      "grad_norm": 0.33441999554634094,
      "learning_rate": 8e-05,
      "loss": 1.7972,
      "step": 3005
    },
    {
      "epoch": 0.6611679313757836,
      "grad_norm": 0.29396557807922363,
      "learning_rate": 8e-05,
      "loss": 1.7203,
      "step": 3006
    },
    {
      "epoch": 0.661387880787419,
      "grad_norm": 0.27457180619239807,
      "learning_rate": 8e-05,
      "loss": 1.5044,
      "step": 3007
    },
    {
      "epoch": 0.6616078301990542,
      "grad_norm": 0.2927014231681824,
      "learning_rate": 8e-05,
      "loss": 1.6993,
      "step": 3008
    },
    {
      "epoch": 0.6618277796106895,
      "grad_norm": 0.2836621403694153,
      "learning_rate": 8e-05,
      "loss": 1.6748,
      "step": 3009
    },
    {
      "epoch": 0.6620477290223249,
      "grad_norm": 0.28402644395828247,
      "learning_rate": 8e-05,
      "loss": 1.6562,
      "step": 3010
    },
    {
      "epoch": 0.6622676784339602,
      "grad_norm": 0.2613022029399872,
      "learning_rate": 8e-05,
      "loss": 1.5386,
      "step": 3011
    },
    {
      "epoch": 0.6624876278455956,
      "grad_norm": 0.2824957072734833,
      "learning_rate": 8e-05,
      "loss": 1.6881,
      "step": 3012
    },
    {
      "epoch": 0.6627075772572308,
      "grad_norm": 0.301944375038147,
      "learning_rate": 8e-05,
      "loss": 1.6021,
      "step": 3013
    },
    {
      "epoch": 0.6629275266688661,
      "grad_norm": 0.2858961820602417,
      "learning_rate": 8e-05,
      "loss": 1.7544,
      "step": 3014
    },
    {
      "epoch": 0.6631474760805015,
      "grad_norm": 0.3019895851612091,
      "learning_rate": 8e-05,
      "loss": 1.611,
      "step": 3015
    },
    {
      "epoch": 0.6633674254921368,
      "grad_norm": 0.30784347653388977,
      "learning_rate": 8e-05,
      "loss": 1.5494,
      "step": 3016
    },
    {
      "epoch": 0.6635873749037722,
      "grad_norm": 0.2946648597717285,
      "learning_rate": 8e-05,
      "loss": 1.6664,
      "step": 3017
    },
    {
      "epoch": 0.6638073243154075,
      "grad_norm": 0.29524630308151245,
      "learning_rate": 8e-05,
      "loss": 1.6755,
      "step": 3018
    },
    {
      "epoch": 0.6640272737270427,
      "grad_norm": 0.2957313656806946,
      "learning_rate": 8e-05,
      "loss": 1.7904,
      "step": 3019
    },
    {
      "epoch": 0.6642472231386781,
      "grad_norm": 0.2753024697303772,
      "learning_rate": 8e-05,
      "loss": 1.5431,
      "step": 3020
    },
    {
      "epoch": 0.6644671725503134,
      "grad_norm": 0.27610090374946594,
      "learning_rate": 8e-05,
      "loss": 1.5585,
      "step": 3021
    },
    {
      "epoch": 0.6646871219619488,
      "grad_norm": 0.33159691095352173,
      "learning_rate": 8e-05,
      "loss": 1.6951,
      "step": 3022
    },
    {
      "epoch": 0.6649070713735841,
      "grad_norm": 0.3014865219593048,
      "learning_rate": 8e-05,
      "loss": 1.8178,
      "step": 3023
    },
    {
      "epoch": 0.6651270207852193,
      "grad_norm": 0.29317575693130493,
      "learning_rate": 8e-05,
      "loss": 1.699,
      "step": 3024
    },
    {
      "epoch": 0.6653469701968547,
      "grad_norm": 0.28347086906433105,
      "learning_rate": 8e-05,
      "loss": 1.658,
      "step": 3025
    },
    {
      "epoch": 0.66556691960849,
      "grad_norm": 0.27827227115631104,
      "learning_rate": 8e-05,
      "loss": 1.6284,
      "step": 3026
    },
    {
      "epoch": 0.6657868690201254,
      "grad_norm": 0.2806004285812378,
      "learning_rate": 8e-05,
      "loss": 1.6494,
      "step": 3027
    },
    {
      "epoch": 0.6660068184317607,
      "grad_norm": 0.27863019704818726,
      "learning_rate": 8e-05,
      "loss": 1.5617,
      "step": 3028
    },
    {
      "epoch": 0.6662267678433961,
      "grad_norm": 0.3057478368282318,
      "learning_rate": 8e-05,
      "loss": 1.7866,
      "step": 3029
    },
    {
      "epoch": 0.6664467172550314,
      "grad_norm": 0.29229068756103516,
      "learning_rate": 8e-05,
      "loss": 1.5757,
      "step": 3030
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.35514572262763977,
      "learning_rate": 8e-05,
      "loss": 1.8164,
      "step": 3031
    },
    {
      "epoch": 0.666886616078302,
      "grad_norm": 0.2878423035144806,
      "learning_rate": 8e-05,
      "loss": 1.6587,
      "step": 3032
    },
    {
      "epoch": 0.6671065654899373,
      "grad_norm": 0.34859347343444824,
      "learning_rate": 8e-05,
      "loss": 1.8458,
      "step": 3033
    },
    {
      "epoch": 0.6673265149015727,
      "grad_norm": 0.2848225235939026,
      "learning_rate": 8e-05,
      "loss": 1.6178,
      "step": 3034
    },
    {
      "epoch": 0.667546464313208,
      "grad_norm": 0.29574286937713623,
      "learning_rate": 8e-05,
      "loss": 1.5993,
      "step": 3035
    },
    {
      "epoch": 0.6677664137248432,
      "grad_norm": 0.2836141586303711,
      "learning_rate": 8e-05,
      "loss": 1.696,
      "step": 3036
    },
    {
      "epoch": 0.6679863631364786,
      "grad_norm": 0.32046374678611755,
      "learning_rate": 8e-05,
      "loss": 1.8144,
      "step": 3037
    },
    {
      "epoch": 0.6682063125481139,
      "grad_norm": 0.2817988991737366,
      "learning_rate": 8e-05,
      "loss": 1.675,
      "step": 3038
    },
    {
      "epoch": 0.6684262619597493,
      "grad_norm": 0.3105819523334503,
      "learning_rate": 8e-05,
      "loss": 1.756,
      "step": 3039
    },
    {
      "epoch": 0.6686462113713846,
      "grad_norm": 0.28836145997047424,
      "learning_rate": 8e-05,
      "loss": 1.7165,
      "step": 3040
    },
    {
      "epoch": 0.6688661607830199,
      "grad_norm": 0.35061419010162354,
      "learning_rate": 8e-05,
      "loss": 1.6084,
      "step": 3041
    },
    {
      "epoch": 0.6690861101946552,
      "grad_norm": 0.2887910008430481,
      "learning_rate": 8e-05,
      "loss": 1.7425,
      "step": 3042
    },
    {
      "epoch": 0.6693060596062905,
      "grad_norm": 0.34143608808517456,
      "learning_rate": 8e-05,
      "loss": 1.8527,
      "step": 3043
    },
    {
      "epoch": 0.6695260090179259,
      "grad_norm": 0.2963760197162628,
      "learning_rate": 8e-05,
      "loss": 1.7126,
      "step": 3044
    },
    {
      "epoch": 0.6697459584295612,
      "grad_norm": 0.27499714493751526,
      "learning_rate": 8e-05,
      "loss": 1.5678,
      "step": 3045
    },
    {
      "epoch": 0.6699659078411965,
      "grad_norm": 0.2994243800640106,
      "learning_rate": 8e-05,
      "loss": 1.7,
      "step": 3046
    },
    {
      "epoch": 0.6701858572528319,
      "grad_norm": 0.28100645542144775,
      "learning_rate": 8e-05,
      "loss": 1.5831,
      "step": 3047
    },
    {
      "epoch": 0.6704058066644671,
      "grad_norm": 0.3079579472541809,
      "learning_rate": 8e-05,
      "loss": 1.5761,
      "step": 3048
    },
    {
      "epoch": 0.6706257560761025,
      "grad_norm": 0.30679425597190857,
      "learning_rate": 8e-05,
      "loss": 1.8592,
      "step": 3049
    },
    {
      "epoch": 0.6708457054877378,
      "grad_norm": 0.2816236913204193,
      "learning_rate": 8e-05,
      "loss": 1.61,
      "step": 3050
    },
    {
      "epoch": 0.6710656548993732,
      "grad_norm": 0.29027682542800903,
      "learning_rate": 8e-05,
      "loss": 1.7003,
      "step": 3051
    },
    {
      "epoch": 0.6712856043110085,
      "grad_norm": 0.29806220531463623,
      "learning_rate": 8e-05,
      "loss": 1.7301,
      "step": 3052
    },
    {
      "epoch": 0.6715055537226438,
      "grad_norm": 0.3061826527118683,
      "learning_rate": 8e-05,
      "loss": 1.7959,
      "step": 3053
    },
    {
      "epoch": 0.6717255031342791,
      "grad_norm": 0.2916790544986725,
      "learning_rate": 8e-05,
      "loss": 1.6444,
      "step": 3054
    },
    {
      "epoch": 0.6719454525459144,
      "grad_norm": 0.29990294575691223,
      "learning_rate": 8e-05,
      "loss": 1.6154,
      "step": 3055
    },
    {
      "epoch": 0.6721654019575498,
      "grad_norm": 0.2841048240661621,
      "learning_rate": 8e-05,
      "loss": 1.604,
      "step": 3056
    },
    {
      "epoch": 0.6723853513691851,
      "grad_norm": 0.2929285764694214,
      "learning_rate": 8e-05,
      "loss": 1.7562,
      "step": 3057
    },
    {
      "epoch": 0.6726053007808204,
      "grad_norm": 0.3035488724708557,
      "learning_rate": 8e-05,
      "loss": 1.7122,
      "step": 3058
    },
    {
      "epoch": 0.6728252501924558,
      "grad_norm": 0.2829075753688812,
      "learning_rate": 8e-05,
      "loss": 1.5672,
      "step": 3059
    },
    {
      "epoch": 0.673045199604091,
      "grad_norm": 0.2959376871585846,
      "learning_rate": 8e-05,
      "loss": 1.6747,
      "step": 3060
    },
    {
      "epoch": 0.6732651490157264,
      "grad_norm": 0.2833213806152344,
      "learning_rate": 8e-05,
      "loss": 1.6152,
      "step": 3061
    },
    {
      "epoch": 0.6734850984273617,
      "grad_norm": 0.30541667342185974,
      "learning_rate": 8e-05,
      "loss": 1.8179,
      "step": 3062
    },
    {
      "epoch": 0.673705047838997,
      "grad_norm": 0.2929080128669739,
      "learning_rate": 8e-05,
      "loss": 1.7548,
      "step": 3063
    },
    {
      "epoch": 0.6739249972506324,
      "grad_norm": 0.28685638308525085,
      "learning_rate": 8e-05,
      "loss": 1.6509,
      "step": 3064
    },
    {
      "epoch": 0.6741449466622677,
      "grad_norm": 0.2862042486667633,
      "learning_rate": 8e-05,
      "loss": 1.6994,
      "step": 3065
    },
    {
      "epoch": 0.674364896073903,
      "grad_norm": 0.298554927110672,
      "learning_rate": 8e-05,
      "loss": 1.7516,
      "step": 3066
    },
    {
      "epoch": 0.6745848454855383,
      "grad_norm": 0.27519044280052185,
      "learning_rate": 8e-05,
      "loss": 1.6233,
      "step": 3067
    },
    {
      "epoch": 0.6748047948971736,
      "grad_norm": 0.2909212112426758,
      "learning_rate": 8e-05,
      "loss": 1.6986,
      "step": 3068
    },
    {
      "epoch": 0.675024744308809,
      "grad_norm": 0.3514604866504669,
      "learning_rate": 8e-05,
      "loss": 1.5948,
      "step": 3069
    },
    {
      "epoch": 0.6752446937204443,
      "grad_norm": 0.2866019308567047,
      "learning_rate": 8e-05,
      "loss": 1.6945,
      "step": 3070
    },
    {
      "epoch": 0.6754646431320797,
      "grad_norm": 0.28309282660484314,
      "learning_rate": 8e-05,
      "loss": 1.5978,
      "step": 3071
    },
    {
      "epoch": 0.6756845925437149,
      "grad_norm": 0.29281336069107056,
      "learning_rate": 8e-05,
      "loss": 1.7229,
      "step": 3072
    },
    {
      "epoch": 0.6759045419553503,
      "grad_norm": 0.2977350056171417,
      "learning_rate": 8e-05,
      "loss": 1.6669,
      "step": 3073
    },
    {
      "epoch": 0.6761244913669856,
      "grad_norm": 0.2664392590522766,
      "learning_rate": 8e-05,
      "loss": 1.5452,
      "step": 3074
    },
    {
      "epoch": 0.6763444407786209,
      "grad_norm": 0.27810007333755493,
      "learning_rate": 8e-05,
      "loss": 1.5437,
      "step": 3075
    },
    {
      "epoch": 0.6765643901902563,
      "grad_norm": 0.28598442673683167,
      "learning_rate": 8e-05,
      "loss": 1.6154,
      "step": 3076
    },
    {
      "epoch": 0.6767843396018915,
      "grad_norm": 0.2768900692462921,
      "learning_rate": 8e-05,
      "loss": 1.7086,
      "step": 3077
    },
    {
      "epoch": 0.6770042890135269,
      "grad_norm": 0.2813318073749542,
      "learning_rate": 8e-05,
      "loss": 1.638,
      "step": 3078
    },
    {
      "epoch": 0.6772242384251622,
      "grad_norm": 0.2905293107032776,
      "learning_rate": 8e-05,
      "loss": 1.8223,
      "step": 3079
    },
    {
      "epoch": 0.6774441878367975,
      "grad_norm": 0.3056020140647888,
      "learning_rate": 8e-05,
      "loss": 1.6754,
      "step": 3080
    },
    {
      "epoch": 0.6776641372484329,
      "grad_norm": 0.29252004623413086,
      "learning_rate": 8e-05,
      "loss": 1.764,
      "step": 3081
    },
    {
      "epoch": 0.6778840866600682,
      "grad_norm": 0.29702186584472656,
      "learning_rate": 8e-05,
      "loss": 1.707,
      "step": 3082
    },
    {
      "epoch": 0.6781040360717036,
      "grad_norm": 0.286920964717865,
      "learning_rate": 8e-05,
      "loss": 1.6433,
      "step": 3083
    },
    {
      "epoch": 0.6783239854833388,
      "grad_norm": 0.28408947587013245,
      "learning_rate": 8e-05,
      "loss": 1.5688,
      "step": 3084
    },
    {
      "epoch": 0.6785439348949741,
      "grad_norm": 0.30250096321105957,
      "learning_rate": 8e-05,
      "loss": 1.6018,
      "step": 3085
    },
    {
      "epoch": 0.6787638843066095,
      "grad_norm": 0.28146034479141235,
      "learning_rate": 8e-05,
      "loss": 1.5733,
      "step": 3086
    },
    {
      "epoch": 0.6789838337182448,
      "grad_norm": 0.27325439453125,
      "learning_rate": 8e-05,
      "loss": 1.5913,
      "step": 3087
    },
    {
      "epoch": 0.6792037831298802,
      "grad_norm": 0.2941316068172455,
      "learning_rate": 8e-05,
      "loss": 1.8549,
      "step": 3088
    },
    {
      "epoch": 0.6794237325415154,
      "grad_norm": 0.29817789793014526,
      "learning_rate": 8e-05,
      "loss": 1.8493,
      "step": 3089
    },
    {
      "epoch": 0.6796436819531507,
      "grad_norm": 0.29364877939224243,
      "learning_rate": 8e-05,
      "loss": 1.6039,
      "step": 3090
    },
    {
      "epoch": 0.6798636313647861,
      "grad_norm": 0.3067622482776642,
      "learning_rate": 8e-05,
      "loss": 1.7219,
      "step": 3091
    },
    {
      "epoch": 0.6800835807764214,
      "grad_norm": 0.2989698648452759,
      "learning_rate": 8e-05,
      "loss": 1.6856,
      "step": 3092
    },
    {
      "epoch": 0.6803035301880568,
      "grad_norm": 0.27398690581321716,
      "learning_rate": 8e-05,
      "loss": 1.6872,
      "step": 3093
    },
    {
      "epoch": 0.6805234795996921,
      "grad_norm": 0.27870994806289673,
      "learning_rate": 8e-05,
      "loss": 1.5519,
      "step": 3094
    },
    {
      "epoch": 0.6807434290113274,
      "grad_norm": 0.27137231826782227,
      "learning_rate": 8e-05,
      "loss": 1.703,
      "step": 3095
    },
    {
      "epoch": 0.6809633784229627,
      "grad_norm": 0.30240193009376526,
      "learning_rate": 8e-05,
      "loss": 1.7728,
      "step": 3096
    },
    {
      "epoch": 0.681183327834598,
      "grad_norm": 0.29213348031044006,
      "learning_rate": 8e-05,
      "loss": 1.7236,
      "step": 3097
    },
    {
      "epoch": 0.6814032772462334,
      "grad_norm": 0.2953360974788666,
      "learning_rate": 8e-05,
      "loss": 1.6609,
      "step": 3098
    },
    {
      "epoch": 0.6816232266578687,
      "grad_norm": 0.29802900552749634,
      "learning_rate": 8e-05,
      "loss": 1.6969,
      "step": 3099
    },
    {
      "epoch": 0.6818431760695041,
      "grad_norm": 0.2967725396156311,
      "learning_rate": 8e-05,
      "loss": 1.8439,
      "step": 3100
    },
    {
      "epoch": 0.6820631254811393,
      "grad_norm": 0.2971365451812744,
      "learning_rate": 8e-05,
      "loss": 1.7695,
      "step": 3101
    },
    {
      "epoch": 0.6822830748927746,
      "grad_norm": 0.2880017161369324,
      "learning_rate": 8e-05,
      "loss": 1.6916,
      "step": 3102
    },
    {
      "epoch": 0.68250302430441,
      "grad_norm": 0.2691252529621124,
      "learning_rate": 8e-05,
      "loss": 1.5439,
      "step": 3103
    },
    {
      "epoch": 0.6827229737160453,
      "grad_norm": 0.2928919494152069,
      "learning_rate": 8e-05,
      "loss": 1.6325,
      "step": 3104
    },
    {
      "epoch": 0.6829429231276807,
      "grad_norm": 0.28234854340553284,
      "learning_rate": 8e-05,
      "loss": 1.6778,
      "step": 3105
    },
    {
      "epoch": 0.683162872539316,
      "grad_norm": 0.2900925278663635,
      "learning_rate": 8e-05,
      "loss": 1.7047,
      "step": 3106
    },
    {
      "epoch": 0.6833828219509512,
      "grad_norm": 0.30338430404663086,
      "learning_rate": 8e-05,
      "loss": 1.7422,
      "step": 3107
    },
    {
      "epoch": 0.6836027713625866,
      "grad_norm": 0.2576045095920563,
      "learning_rate": 8e-05,
      "loss": 1.2735,
      "step": 3108
    },
    {
      "epoch": 0.6838227207742219,
      "grad_norm": 0.2839035987854004,
      "learning_rate": 8e-05,
      "loss": 1.6443,
      "step": 3109
    },
    {
      "epoch": 0.6840426701858573,
      "grad_norm": 0.29431045055389404,
      "learning_rate": 8e-05,
      "loss": 1.7928,
      "step": 3110
    },
    {
      "epoch": 0.6842626195974926,
      "grad_norm": 0.301283597946167,
      "learning_rate": 8e-05,
      "loss": 1.7339,
      "step": 3111
    },
    {
      "epoch": 0.6844825690091279,
      "grad_norm": 0.27778249979019165,
      "learning_rate": 8e-05,
      "loss": 1.6481,
      "step": 3112
    },
    {
      "epoch": 0.6847025184207632,
      "grad_norm": 0.317439466714859,
      "learning_rate": 8e-05,
      "loss": 1.7339,
      "step": 3113
    },
    {
      "epoch": 0.6849224678323985,
      "grad_norm": 0.26562100648880005,
      "learning_rate": 8e-05,
      "loss": 1.5077,
      "step": 3114
    },
    {
      "epoch": 0.6851424172440339,
      "grad_norm": 0.27444925904273987,
      "learning_rate": 8e-05,
      "loss": 1.6178,
      "step": 3115
    },
    {
      "epoch": 0.6853623666556692,
      "grad_norm": 0.302569180727005,
      "learning_rate": 8e-05,
      "loss": 1.7989,
      "step": 3116
    },
    {
      "epoch": 0.6855823160673045,
      "grad_norm": 0.27267923951148987,
      "learning_rate": 8e-05,
      "loss": 1.5604,
      "step": 3117
    },
    {
      "epoch": 0.6858022654789399,
      "grad_norm": 0.295436292886734,
      "learning_rate": 8e-05,
      "loss": 1.6847,
      "step": 3118
    },
    {
      "epoch": 0.6860222148905751,
      "grad_norm": 0.29056796431541443,
      "learning_rate": 8e-05,
      "loss": 1.5835,
      "step": 3119
    },
    {
      "epoch": 0.6862421643022105,
      "grad_norm": 0.29942628741264343,
      "learning_rate": 8e-05,
      "loss": 1.692,
      "step": 3120
    },
    {
      "epoch": 0.6864621137138458,
      "grad_norm": 0.29042521119117737,
      "learning_rate": 8e-05,
      "loss": 1.7639,
      "step": 3121
    },
    {
      "epoch": 0.6866820631254812,
      "grad_norm": 0.2988620102405548,
      "learning_rate": 8e-05,
      "loss": 1.7096,
      "step": 3122
    },
    {
      "epoch": 0.6869020125371165,
      "grad_norm": 0.30227869749069214,
      "learning_rate": 8e-05,
      "loss": 1.7103,
      "step": 3123
    },
    {
      "epoch": 0.6871219619487517,
      "grad_norm": 0.2963883578777313,
      "learning_rate": 8e-05,
      "loss": 1.6252,
      "step": 3124
    },
    {
      "epoch": 0.6873419113603871,
      "grad_norm": 0.267903596162796,
      "learning_rate": 8e-05,
      "loss": 1.5459,
      "step": 3125
    },
    {
      "epoch": 0.6875618607720224,
      "grad_norm": 0.34292858839035034,
      "learning_rate": 8e-05,
      "loss": 1.7557,
      "step": 3126
    },
    {
      "epoch": 0.6877818101836578,
      "grad_norm": 0.31191956996917725,
      "learning_rate": 8e-05,
      "loss": 1.8102,
      "step": 3127
    },
    {
      "epoch": 0.6880017595952931,
      "grad_norm": 0.2759200632572174,
      "learning_rate": 8e-05,
      "loss": 1.4965,
      "step": 3128
    },
    {
      "epoch": 0.6882217090069284,
      "grad_norm": 0.279722660779953,
      "learning_rate": 8e-05,
      "loss": 1.6268,
      "step": 3129
    },
    {
      "epoch": 0.6884416584185638,
      "grad_norm": 0.296171635389328,
      "learning_rate": 8e-05,
      "loss": 1.8015,
      "step": 3130
    },
    {
      "epoch": 0.688661607830199,
      "grad_norm": 0.2832602560520172,
      "learning_rate": 8e-05,
      "loss": 1.6554,
      "step": 3131
    },
    {
      "epoch": 0.6888815572418344,
      "grad_norm": 0.28423500061035156,
      "learning_rate": 8e-05,
      "loss": 1.5898,
      "step": 3132
    },
    {
      "epoch": 0.6891015066534697,
      "grad_norm": 0.27690091729164124,
      "learning_rate": 8e-05,
      "loss": 1.5216,
      "step": 3133
    },
    {
      "epoch": 0.689321456065105,
      "grad_norm": 0.36139482259750366,
      "learning_rate": 8e-05,
      "loss": 1.7153,
      "step": 3134
    },
    {
      "epoch": 0.6895414054767404,
      "grad_norm": 0.3002482056617737,
      "learning_rate": 8e-05,
      "loss": 1.8588,
      "step": 3135
    },
    {
      "epoch": 0.6897613548883756,
      "grad_norm": 0.29645755887031555,
      "learning_rate": 8e-05,
      "loss": 1.6894,
      "step": 3136
    },
    {
      "epoch": 0.689981304300011,
      "grad_norm": 0.27819156646728516,
      "learning_rate": 8e-05,
      "loss": 1.5448,
      "step": 3137
    },
    {
      "epoch": 0.6902012537116463,
      "grad_norm": 0.2958146035671234,
      "learning_rate": 8e-05,
      "loss": 1.7416,
      "step": 3138
    },
    {
      "epoch": 0.6904212031232816,
      "grad_norm": 0.30198779702186584,
      "learning_rate": 8e-05,
      "loss": 1.7765,
      "step": 3139
    },
    {
      "epoch": 0.690641152534917,
      "grad_norm": 0.2883966565132141,
      "learning_rate": 8e-05,
      "loss": 1.6467,
      "step": 3140
    },
    {
      "epoch": 0.6908611019465523,
      "grad_norm": 0.2731442451477051,
      "learning_rate": 8e-05,
      "loss": 1.5186,
      "step": 3141
    },
    {
      "epoch": 0.6910810513581876,
      "grad_norm": 0.2791924774646759,
      "learning_rate": 8e-05,
      "loss": 1.7721,
      "step": 3142
    },
    {
      "epoch": 0.6913010007698229,
      "grad_norm": 0.2815117835998535,
      "learning_rate": 8e-05,
      "loss": 1.5362,
      "step": 3143
    },
    {
      "epoch": 0.6915209501814583,
      "grad_norm": 0.2724231481552124,
      "learning_rate": 8e-05,
      "loss": 1.7015,
      "step": 3144
    },
    {
      "epoch": 0.6917408995930936,
      "grad_norm": 0.3006933629512787,
      "learning_rate": 8e-05,
      "loss": 1.7298,
      "step": 3145
    },
    {
      "epoch": 0.6919608490047289,
      "grad_norm": 0.3087989389896393,
      "learning_rate": 8e-05,
      "loss": 1.6349,
      "step": 3146
    },
    {
      "epoch": 0.6921807984163643,
      "grad_norm": 0.2806054949760437,
      "learning_rate": 8e-05,
      "loss": 1.6347,
      "step": 3147
    },
    {
      "epoch": 0.6924007478279995,
      "grad_norm": 0.28338587284088135,
      "learning_rate": 8e-05,
      "loss": 1.7277,
      "step": 3148
    },
    {
      "epoch": 0.6926206972396349,
      "grad_norm": 0.27475735545158386,
      "learning_rate": 8e-05,
      "loss": 1.6705,
      "step": 3149
    },
    {
      "epoch": 0.6928406466512702,
      "grad_norm": 0.3012952506542206,
      "learning_rate": 8e-05,
      "loss": 1.8124,
      "step": 3150
    },
    {
      "epoch": 0.6930605960629055,
      "grad_norm": 0.28966784477233887,
      "learning_rate": 8e-05,
      "loss": 1.6413,
      "step": 3151
    },
    {
      "epoch": 0.6932805454745409,
      "grad_norm": 0.27356377243995667,
      "learning_rate": 8e-05,
      "loss": 1.5894,
      "step": 3152
    },
    {
      "epoch": 0.6935004948861762,
      "grad_norm": 0.3299154043197632,
      "learning_rate": 8e-05,
      "loss": 1.6931,
      "step": 3153
    },
    {
      "epoch": 0.6937204442978115,
      "grad_norm": 0.2960734963417053,
      "learning_rate": 8e-05,
      "loss": 1.7089,
      "step": 3154
    },
    {
      "epoch": 0.6939403937094468,
      "grad_norm": 0.2994215190410614,
      "learning_rate": 8e-05,
      "loss": 1.6326,
      "step": 3155
    },
    {
      "epoch": 0.6941603431210821,
      "grad_norm": 0.2847454249858856,
      "learning_rate": 8e-05,
      "loss": 1.7982,
      "step": 3156
    },
    {
      "epoch": 0.6943802925327175,
      "grad_norm": 0.28896382451057434,
      "learning_rate": 8e-05,
      "loss": 1.6695,
      "step": 3157
    },
    {
      "epoch": 0.6946002419443528,
      "grad_norm": 0.2861023545265198,
      "learning_rate": 8e-05,
      "loss": 1.7209,
      "step": 3158
    },
    {
      "epoch": 0.6948201913559882,
      "grad_norm": 0.26986655592918396,
      "learning_rate": 8e-05,
      "loss": 1.535,
      "step": 3159
    },
    {
      "epoch": 0.6950401407676234,
      "grad_norm": 0.3060385286808014,
      "learning_rate": 8e-05,
      "loss": 1.6242,
      "step": 3160
    },
    {
      "epoch": 0.6952600901792587,
      "grad_norm": 0.2831737995147705,
      "learning_rate": 8e-05,
      "loss": 1.6341,
      "step": 3161
    },
    {
      "epoch": 0.6954800395908941,
      "grad_norm": 0.30474355816841125,
      "learning_rate": 8e-05,
      "loss": 1.7048,
      "step": 3162
    },
    {
      "epoch": 0.6956999890025294,
      "grad_norm": 0.2686191499233246,
      "learning_rate": 8e-05,
      "loss": 1.5553,
      "step": 3163
    },
    {
      "epoch": 0.6959199384141648,
      "grad_norm": 0.2899799048900604,
      "learning_rate": 8e-05,
      "loss": 1.5918,
      "step": 3164
    },
    {
      "epoch": 0.6961398878258,
      "grad_norm": 0.2757590115070343,
      "learning_rate": 8e-05,
      "loss": 1.6582,
      "step": 3165
    },
    {
      "epoch": 0.6963598372374354,
      "grad_norm": 0.2934139668941498,
      "learning_rate": 8e-05,
      "loss": 1.6196,
      "step": 3166
    },
    {
      "epoch": 0.6965797866490707,
      "grad_norm": 0.2803792953491211,
      "learning_rate": 8e-05,
      "loss": 1.4932,
      "step": 3167
    },
    {
      "epoch": 0.696799736060706,
      "grad_norm": 0.27724161744117737,
      "learning_rate": 8e-05,
      "loss": 1.6452,
      "step": 3168
    },
    {
      "epoch": 0.6970196854723414,
      "grad_norm": 0.31233644485473633,
      "learning_rate": 8e-05,
      "loss": 1.7572,
      "step": 3169
    },
    {
      "epoch": 0.6972396348839767,
      "grad_norm": 0.2862301766872406,
      "learning_rate": 8e-05,
      "loss": 1.663,
      "step": 3170
    },
    {
      "epoch": 0.6974595842956121,
      "grad_norm": 0.284385085105896,
      "learning_rate": 8e-05,
      "loss": 1.7268,
      "step": 3171
    },
    {
      "epoch": 0.6976795337072473,
      "grad_norm": 0.2832756042480469,
      "learning_rate": 8e-05,
      "loss": 1.6006,
      "step": 3172
    },
    {
      "epoch": 0.6978994831188826,
      "grad_norm": 0.2860611081123352,
      "learning_rate": 8e-05,
      "loss": 1.6932,
      "step": 3173
    },
    {
      "epoch": 0.698119432530518,
      "grad_norm": 0.2913680970668793,
      "learning_rate": 8e-05,
      "loss": 1.6705,
      "step": 3174
    },
    {
      "epoch": 0.6983393819421533,
      "grad_norm": 0.2927596867084503,
      "learning_rate": 8e-05,
      "loss": 1.7578,
      "step": 3175
    },
    {
      "epoch": 0.6985593313537887,
      "grad_norm": 0.2793924808502197,
      "learning_rate": 8e-05,
      "loss": 1.6293,
      "step": 3176
    },
    {
      "epoch": 0.698779280765424,
      "grad_norm": 0.28402742743492126,
      "learning_rate": 8e-05,
      "loss": 1.6703,
      "step": 3177
    },
    {
      "epoch": 0.6989992301770592,
      "grad_norm": 0.2879323661327362,
      "learning_rate": 8e-05,
      "loss": 1.5645,
      "step": 3178
    },
    {
      "epoch": 0.6992191795886946,
      "grad_norm": 0.28840363025665283,
      "learning_rate": 8e-05,
      "loss": 1.6608,
      "step": 3179
    },
    {
      "epoch": 0.6994391290003299,
      "grad_norm": 0.2971773147583008,
      "learning_rate": 8e-05,
      "loss": 1.7994,
      "step": 3180
    },
    {
      "epoch": 0.6996590784119653,
      "grad_norm": 0.2872611880302429,
      "learning_rate": 8e-05,
      "loss": 1.6931,
      "step": 3181
    },
    {
      "epoch": 0.6998790278236006,
      "grad_norm": 0.2892748713493347,
      "learning_rate": 8e-05,
      "loss": 1.6808,
      "step": 3182
    },
    {
      "epoch": 0.7000989772352358,
      "grad_norm": 0.3782177269458771,
      "learning_rate": 8e-05,
      "loss": 1.7109,
      "step": 3183
    },
    {
      "epoch": 0.7003189266468712,
      "grad_norm": 0.2824467420578003,
      "learning_rate": 8e-05,
      "loss": 1.6967,
      "step": 3184
    },
    {
      "epoch": 0.7005388760585065,
      "grad_norm": 0.3341313898563385,
      "learning_rate": 8e-05,
      "loss": 1.8551,
      "step": 3185
    },
    {
      "epoch": 0.7007588254701419,
      "grad_norm": 0.2792004346847534,
      "learning_rate": 8e-05,
      "loss": 1.6054,
      "step": 3186
    },
    {
      "epoch": 0.7009787748817772,
      "grad_norm": 0.2860063910484314,
      "learning_rate": 8e-05,
      "loss": 1.7671,
      "step": 3187
    },
    {
      "epoch": 0.7011987242934126,
      "grad_norm": 0.2876088321208954,
      "learning_rate": 8e-05,
      "loss": 1.6843,
      "step": 3188
    },
    {
      "epoch": 0.7014186737050478,
      "grad_norm": 0.28482869267463684,
      "learning_rate": 8e-05,
      "loss": 1.42,
      "step": 3189
    },
    {
      "epoch": 0.7016386231166831,
      "grad_norm": 0.30407440662384033,
      "learning_rate": 8e-05,
      "loss": 1.7995,
      "step": 3190
    },
    {
      "epoch": 0.7018585725283185,
      "grad_norm": 0.28979554772377014,
      "learning_rate": 8e-05,
      "loss": 1.6972,
      "step": 3191
    },
    {
      "epoch": 0.7020785219399538,
      "grad_norm": 0.2807050049304962,
      "learning_rate": 8e-05,
      "loss": 1.6641,
      "step": 3192
    },
    {
      "epoch": 0.7022984713515892,
      "grad_norm": 0.3025875985622406,
      "learning_rate": 8e-05,
      "loss": 1.5833,
      "step": 3193
    },
    {
      "epoch": 0.7025184207632245,
      "grad_norm": 0.3045402467250824,
      "learning_rate": 8e-05,
      "loss": 1.5986,
      "step": 3194
    },
    {
      "epoch": 0.7027383701748597,
      "grad_norm": 0.2938578128814697,
      "learning_rate": 8e-05,
      "loss": 1.5732,
      "step": 3195
    },
    {
      "epoch": 0.7029583195864951,
      "grad_norm": 0.270671546459198,
      "learning_rate": 8e-05,
      "loss": 1.4949,
      "step": 3196
    },
    {
      "epoch": 0.7031782689981304,
      "grad_norm": 0.2783917784690857,
      "learning_rate": 8e-05,
      "loss": 1.6004,
      "step": 3197
    },
    {
      "epoch": 0.7033982184097658,
      "grad_norm": 0.26526427268981934,
      "learning_rate": 8e-05,
      "loss": 1.5312,
      "step": 3198
    },
    {
      "epoch": 0.7036181678214011,
      "grad_norm": 0.2944982647895813,
      "learning_rate": 8e-05,
      "loss": 1.7079,
      "step": 3199
    },
    {
      "epoch": 0.7038381172330364,
      "grad_norm": 0.2870388627052307,
      "learning_rate": 8e-05,
      "loss": 1.5144,
      "step": 3200
    },
    {
      "epoch": 0.7040580666446717,
      "grad_norm": 0.28370967507362366,
      "learning_rate": 8e-05,
      "loss": 1.5572,
      "step": 3201
    },
    {
      "epoch": 0.704278016056307,
      "grad_norm": 0.2962104380130768,
      "learning_rate": 8e-05,
      "loss": 1.768,
      "step": 3202
    },
    {
      "epoch": 0.7044979654679424,
      "grad_norm": 0.30153971910476685,
      "learning_rate": 8e-05,
      "loss": 1.5896,
      "step": 3203
    },
    {
      "epoch": 0.7047179148795777,
      "grad_norm": 0.27244117856025696,
      "learning_rate": 8e-05,
      "loss": 1.6106,
      "step": 3204
    },
    {
      "epoch": 0.704937864291213,
      "grad_norm": 0.27167847752571106,
      "learning_rate": 8e-05,
      "loss": 1.6106,
      "step": 3205
    },
    {
      "epoch": 0.7051578137028484,
      "grad_norm": 0.28725895285606384,
      "learning_rate": 8e-05,
      "loss": 1.6029,
      "step": 3206
    },
    {
      "epoch": 0.7053777631144836,
      "grad_norm": 0.2840006947517395,
      "learning_rate": 8e-05,
      "loss": 1.5254,
      "step": 3207
    },
    {
      "epoch": 0.705597712526119,
      "grad_norm": 0.30838871002197266,
      "learning_rate": 8e-05,
      "loss": 1.711,
      "step": 3208
    },
    {
      "epoch": 0.7058176619377543,
      "grad_norm": 0.28863799571990967,
      "learning_rate": 8e-05,
      "loss": 1.6534,
      "step": 3209
    },
    {
      "epoch": 0.7060376113493897,
      "grad_norm": 0.27136942744255066,
      "learning_rate": 8e-05,
      "loss": 1.6272,
      "step": 3210
    },
    {
      "epoch": 0.706257560761025,
      "grad_norm": 0.29609814286231995,
      "learning_rate": 8e-05,
      "loss": 1.7805,
      "step": 3211
    },
    {
      "epoch": 0.7064775101726603,
      "grad_norm": 0.2786168158054352,
      "learning_rate": 8e-05,
      "loss": 1.5998,
      "step": 3212
    },
    {
      "epoch": 0.7066974595842956,
      "grad_norm": 0.2873217761516571,
      "learning_rate": 8e-05,
      "loss": 1.8394,
      "step": 3213
    },
    {
      "epoch": 0.7069174089959309,
      "grad_norm": 0.28696301579475403,
      "learning_rate": 8e-05,
      "loss": 1.6836,
      "step": 3214
    },
    {
      "epoch": 0.7071373584075663,
      "grad_norm": 0.2748275101184845,
      "learning_rate": 8e-05,
      "loss": 1.632,
      "step": 3215
    },
    {
      "epoch": 0.7073573078192016,
      "grad_norm": 0.30858665704727173,
      "learning_rate": 8e-05,
      "loss": 1.6296,
      "step": 3216
    },
    {
      "epoch": 0.7075772572308369,
      "grad_norm": 0.28793784976005554,
      "learning_rate": 8e-05,
      "loss": 1.7079,
      "step": 3217
    },
    {
      "epoch": 0.7077972066424723,
      "grad_norm": 0.3041824996471405,
      "learning_rate": 8e-05,
      "loss": 1.756,
      "step": 3218
    },
    {
      "epoch": 0.7080171560541075,
      "grad_norm": 0.27776944637298584,
      "learning_rate": 8e-05,
      "loss": 1.5994,
      "step": 3219
    },
    {
      "epoch": 0.7082371054657429,
      "grad_norm": 0.291361927986145,
      "learning_rate": 8e-05,
      "loss": 1.6744,
      "step": 3220
    },
    {
      "epoch": 0.7084570548773782,
      "grad_norm": 0.2865651249885559,
      "learning_rate": 8e-05,
      "loss": 1.7511,
      "step": 3221
    },
    {
      "epoch": 0.7086770042890135,
      "grad_norm": 0.28433775901794434,
      "learning_rate": 8e-05,
      "loss": 1.7272,
      "step": 3222
    },
    {
      "epoch": 0.7088969537006489,
      "grad_norm": 0.2853262722492218,
      "learning_rate": 8e-05,
      "loss": 1.6281,
      "step": 3223
    },
    {
      "epoch": 0.7091169031122841,
      "grad_norm": 0.2622532844543457,
      "learning_rate": 8e-05,
      "loss": 1.6179,
      "step": 3224
    },
    {
      "epoch": 0.7093368525239195,
      "grad_norm": 0.32692235708236694,
      "learning_rate": 8e-05,
      "loss": 1.784,
      "step": 3225
    },
    {
      "epoch": 0.7095568019355548,
      "grad_norm": 0.3035888671875,
      "learning_rate": 8e-05,
      "loss": 1.6944,
      "step": 3226
    },
    {
      "epoch": 0.7097767513471901,
      "grad_norm": 0.28113415837287903,
      "learning_rate": 8e-05,
      "loss": 1.5492,
      "step": 3227
    },
    {
      "epoch": 0.7099967007588255,
      "grad_norm": 0.282117635011673,
      "learning_rate": 8e-05,
      "loss": 1.5972,
      "step": 3228
    },
    {
      "epoch": 0.7102166501704608,
      "grad_norm": 0.28914523124694824,
      "learning_rate": 8e-05,
      "loss": 1.6345,
      "step": 3229
    },
    {
      "epoch": 0.7104365995820962,
      "grad_norm": 0.2931251525878906,
      "learning_rate": 8e-05,
      "loss": 1.7021,
      "step": 3230
    },
    {
      "epoch": 0.7106565489937314,
      "grad_norm": 0.29080888628959656,
      "learning_rate": 8e-05,
      "loss": 1.6678,
      "step": 3231
    },
    {
      "epoch": 0.7108764984053668,
      "grad_norm": 0.29373523592948914,
      "learning_rate": 8e-05,
      "loss": 1.6674,
      "step": 3232
    },
    {
      "epoch": 0.7110964478170021,
      "grad_norm": 0.2820954918861389,
      "learning_rate": 8e-05,
      "loss": 1.6162,
      "step": 3233
    },
    {
      "epoch": 0.7113163972286374,
      "grad_norm": 0.2817211449146271,
      "learning_rate": 8e-05,
      "loss": 1.572,
      "step": 3234
    },
    {
      "epoch": 0.7115363466402728,
      "grad_norm": 0.28479206562042236,
      "learning_rate": 8e-05,
      "loss": 1.4741,
      "step": 3235
    },
    {
      "epoch": 0.711756296051908,
      "grad_norm": 0.2774266004562378,
      "learning_rate": 8e-05,
      "loss": 1.5465,
      "step": 3236
    },
    {
      "epoch": 0.7119762454635434,
      "grad_norm": 0.2834816873073578,
      "learning_rate": 8e-05,
      "loss": 1.6505,
      "step": 3237
    },
    {
      "epoch": 0.7121961948751787,
      "grad_norm": 0.28653544187545776,
      "learning_rate": 8e-05,
      "loss": 1.6307,
      "step": 3238
    },
    {
      "epoch": 0.712416144286814,
      "grad_norm": 0.28194478154182434,
      "learning_rate": 8e-05,
      "loss": 1.7033,
      "step": 3239
    },
    {
      "epoch": 0.7126360936984494,
      "grad_norm": 0.2983120083808899,
      "learning_rate": 8e-05,
      "loss": 1.7034,
      "step": 3240
    },
    {
      "epoch": 0.7128560431100847,
      "grad_norm": 0.3055756390094757,
      "learning_rate": 8e-05,
      "loss": 1.6579,
      "step": 3241
    },
    {
      "epoch": 0.71307599252172,
      "grad_norm": 0.3022952377796173,
      "learning_rate": 8e-05,
      "loss": 1.7379,
      "step": 3242
    },
    {
      "epoch": 0.7132959419333553,
      "grad_norm": 0.28210487961769104,
      "learning_rate": 8e-05,
      "loss": 1.519,
      "step": 3243
    },
    {
      "epoch": 0.7135158913449906,
      "grad_norm": 0.28924715518951416,
      "learning_rate": 8e-05,
      "loss": 1.6344,
      "step": 3244
    },
    {
      "epoch": 0.713735840756626,
      "grad_norm": 0.285334974527359,
      "learning_rate": 8e-05,
      "loss": 1.7545,
      "step": 3245
    },
    {
      "epoch": 0.7139557901682613,
      "grad_norm": 0.29755088686943054,
      "learning_rate": 8e-05,
      "loss": 1.7103,
      "step": 3246
    },
    {
      "epoch": 0.7141757395798967,
      "grad_norm": 0.30941107869148254,
      "learning_rate": 8e-05,
      "loss": 1.6855,
      "step": 3247
    },
    {
      "epoch": 0.7143956889915319,
      "grad_norm": 0.29600104689598083,
      "learning_rate": 8e-05,
      "loss": 1.8132,
      "step": 3248
    },
    {
      "epoch": 0.7146156384031672,
      "grad_norm": 0.290912926197052,
      "learning_rate": 8e-05,
      "loss": 1.7253,
      "step": 3249
    },
    {
      "epoch": 0.7148355878148026,
      "grad_norm": 0.2880486845970154,
      "learning_rate": 8e-05,
      "loss": 1.7233,
      "step": 3250
    },
    {
      "epoch": 0.7150555372264379,
      "grad_norm": 0.27967116236686707,
      "learning_rate": 8e-05,
      "loss": 1.4805,
      "step": 3251
    },
    {
      "epoch": 0.7152754866380733,
      "grad_norm": 0.2888146638870239,
      "learning_rate": 8e-05,
      "loss": 1.7523,
      "step": 3252
    },
    {
      "epoch": 0.7154954360497086,
      "grad_norm": 0.2760499119758606,
      "learning_rate": 8e-05,
      "loss": 1.6595,
      "step": 3253
    },
    {
      "epoch": 0.715715385461344,
      "grad_norm": 0.29169565439224243,
      "learning_rate": 8e-05,
      "loss": 1.6981,
      "step": 3254
    },
    {
      "epoch": 0.7159353348729792,
      "grad_norm": 0.2904864251613617,
      "learning_rate": 8e-05,
      "loss": 1.708,
      "step": 3255
    },
    {
      "epoch": 0.7161552842846145,
      "grad_norm": 0.2833339273929596,
      "learning_rate": 8e-05,
      "loss": 1.5798,
      "step": 3256
    },
    {
      "epoch": 0.7163752336962499,
      "grad_norm": 0.3214869499206543,
      "learning_rate": 8e-05,
      "loss": 1.823,
      "step": 3257
    },
    {
      "epoch": 0.7165951831078852,
      "grad_norm": 0.3174017071723938,
      "learning_rate": 8e-05,
      "loss": 1.7109,
      "step": 3258
    },
    {
      "epoch": 0.7168151325195206,
      "grad_norm": 0.2913649082183838,
      "learning_rate": 8e-05,
      "loss": 1.5767,
      "step": 3259
    },
    {
      "epoch": 0.7170350819311558,
      "grad_norm": 0.2901788651943207,
      "learning_rate": 8e-05,
      "loss": 1.6476,
      "step": 3260
    },
    {
      "epoch": 0.7172550313427911,
      "grad_norm": 0.2945404052734375,
      "learning_rate": 8e-05,
      "loss": 1.6527,
      "step": 3261
    },
    {
      "epoch": 0.7174749807544265,
      "grad_norm": 0.29439011216163635,
      "learning_rate": 8e-05,
      "loss": 1.7253,
      "step": 3262
    },
    {
      "epoch": 0.7176949301660618,
      "grad_norm": 0.2744368016719818,
      "learning_rate": 8e-05,
      "loss": 1.5482,
      "step": 3263
    },
    {
      "epoch": 0.7179148795776972,
      "grad_norm": 0.3027225434780121,
      "learning_rate": 8e-05,
      "loss": 1.646,
      "step": 3264
    },
    {
      "epoch": 0.7181348289893325,
      "grad_norm": 0.271297425031662,
      "learning_rate": 8e-05,
      "loss": 1.4867,
      "step": 3265
    },
    {
      "epoch": 0.7183547784009677,
      "grad_norm": 0.32890835404396057,
      "learning_rate": 8e-05,
      "loss": 1.8444,
      "step": 3266
    },
    {
      "epoch": 0.7185747278126031,
      "grad_norm": 0.2833615839481354,
      "learning_rate": 8e-05,
      "loss": 1.7631,
      "step": 3267
    },
    {
      "epoch": 0.7187946772242384,
      "grad_norm": 0.2767006456851959,
      "learning_rate": 8e-05,
      "loss": 1.7149,
      "step": 3268
    },
    {
      "epoch": 0.7190146266358738,
      "grad_norm": 0.2924560308456421,
      "learning_rate": 8e-05,
      "loss": 1.6564,
      "step": 3269
    },
    {
      "epoch": 0.7192345760475091,
      "grad_norm": 0.28660503029823303,
      "learning_rate": 8e-05,
      "loss": 1.5485,
      "step": 3270
    },
    {
      "epoch": 0.7194545254591443,
      "grad_norm": 0.28452107310295105,
      "learning_rate": 8e-05,
      "loss": 1.4871,
      "step": 3271
    },
    {
      "epoch": 0.7196744748707797,
      "grad_norm": 0.2906745374202728,
      "learning_rate": 8e-05,
      "loss": 1.7993,
      "step": 3272
    },
    {
      "epoch": 0.719894424282415,
      "grad_norm": 0.28533515334129333,
      "learning_rate": 8e-05,
      "loss": 1.6419,
      "step": 3273
    },
    {
      "epoch": 0.7201143736940504,
      "grad_norm": 0.282487690448761,
      "learning_rate": 8e-05,
      "loss": 1.6759,
      "step": 3274
    },
    {
      "epoch": 0.7203343231056857,
      "grad_norm": 0.3005145490169525,
      "learning_rate": 8e-05,
      "loss": 1.7566,
      "step": 3275
    },
    {
      "epoch": 0.7205542725173211,
      "grad_norm": 0.2659122347831726,
      "learning_rate": 8e-05,
      "loss": 1.443,
      "step": 3276
    },
    {
      "epoch": 0.7207742219289563,
      "grad_norm": 0.30296915769577026,
      "learning_rate": 8e-05,
      "loss": 1.6464,
      "step": 3277
    },
    {
      "epoch": 0.7209941713405916,
      "grad_norm": 0.2925507128238678,
      "learning_rate": 8e-05,
      "loss": 1.5286,
      "step": 3278
    },
    {
      "epoch": 0.721214120752227,
      "grad_norm": 0.27778342366218567,
      "learning_rate": 8e-05,
      "loss": 1.5417,
      "step": 3279
    },
    {
      "epoch": 0.7214340701638623,
      "grad_norm": 0.3245250880718231,
      "learning_rate": 8e-05,
      "loss": 1.9343,
      "step": 3280
    },
    {
      "epoch": 0.7216540195754977,
      "grad_norm": 0.2841079831123352,
      "learning_rate": 8e-05,
      "loss": 1.6095,
      "step": 3281
    },
    {
      "epoch": 0.721873968987133,
      "grad_norm": 0.2818967401981354,
      "learning_rate": 8e-05,
      "loss": 1.775,
      "step": 3282
    },
    {
      "epoch": 0.7220939183987682,
      "grad_norm": 0.30750346183776855,
      "learning_rate": 8e-05,
      "loss": 1.837,
      "step": 3283
    },
    {
      "epoch": 0.7223138678104036,
      "grad_norm": 0.2764554023742676,
      "learning_rate": 8e-05,
      "loss": 1.5843,
      "step": 3284
    },
    {
      "epoch": 0.7225338172220389,
      "grad_norm": 0.2906947433948517,
      "learning_rate": 8e-05,
      "loss": 1.748,
      "step": 3285
    },
    {
      "epoch": 0.7227537666336743,
      "grad_norm": 0.30081701278686523,
      "learning_rate": 8e-05,
      "loss": 1.6306,
      "step": 3286
    },
    {
      "epoch": 0.7229737160453096,
      "grad_norm": 0.2862705886363983,
      "learning_rate": 8e-05,
      "loss": 1.7076,
      "step": 3287
    },
    {
      "epoch": 0.7231936654569449,
      "grad_norm": 0.29947561025619507,
      "learning_rate": 8e-05,
      "loss": 1.6987,
      "step": 3288
    },
    {
      "epoch": 0.7234136148685802,
      "grad_norm": 0.2880048453807831,
      "learning_rate": 8e-05,
      "loss": 1.6921,
      "step": 3289
    },
    {
      "epoch": 0.7236335642802155,
      "grad_norm": 0.2952638268470764,
      "learning_rate": 8e-05,
      "loss": 1.533,
      "step": 3290
    },
    {
      "epoch": 0.7238535136918509,
      "grad_norm": 0.29756438732147217,
      "learning_rate": 8e-05,
      "loss": 1.6473,
      "step": 3291
    },
    {
      "epoch": 0.7240734631034862,
      "grad_norm": 0.29125919938087463,
      "learning_rate": 8e-05,
      "loss": 1.6892,
      "step": 3292
    },
    {
      "epoch": 0.7242934125151215,
      "grad_norm": 0.26599353551864624,
      "learning_rate": 8e-05,
      "loss": 1.4273,
      "step": 3293
    },
    {
      "epoch": 0.7245133619267569,
      "grad_norm": 0.2843374013900757,
      "learning_rate": 8e-05,
      "loss": 1.7237,
      "step": 3294
    },
    {
      "epoch": 0.7247333113383921,
      "grad_norm": 0.28228116035461426,
      "learning_rate": 8e-05,
      "loss": 1.7063,
      "step": 3295
    },
    {
      "epoch": 0.7249532607500275,
      "grad_norm": 0.2917938828468323,
      "learning_rate": 8e-05,
      "loss": 1.7018,
      "step": 3296
    },
    {
      "epoch": 0.7251732101616628,
      "grad_norm": 0.28480616211891174,
      "learning_rate": 8e-05,
      "loss": 1.5869,
      "step": 3297
    },
    {
      "epoch": 0.7253931595732982,
      "grad_norm": 0.3087390661239624,
      "learning_rate": 8e-05,
      "loss": 1.8078,
      "step": 3298
    },
    {
      "epoch": 0.7256131089849335,
      "grad_norm": 0.2808952331542969,
      "learning_rate": 8e-05,
      "loss": 1.7724,
      "step": 3299
    },
    {
      "epoch": 0.7258330583965688,
      "grad_norm": 0.30944177508354187,
      "learning_rate": 8e-05,
      "loss": 1.7574,
      "step": 3300
    },
    {
      "epoch": 0.7260530078082041,
      "grad_norm": 0.2862800657749176,
      "learning_rate": 8e-05,
      "loss": 1.5506,
      "step": 3301
    },
    {
      "epoch": 0.7262729572198394,
      "grad_norm": 0.29587674140930176,
      "learning_rate": 8e-05,
      "loss": 1.6555,
      "step": 3302
    },
    {
      "epoch": 0.7264929066314748,
      "grad_norm": 0.279253214597702,
      "learning_rate": 8e-05,
      "loss": 1.5783,
      "step": 3303
    },
    {
      "epoch": 0.7267128560431101,
      "grad_norm": 0.2954663634300232,
      "learning_rate": 8e-05,
      "loss": 1.898,
      "step": 3304
    },
    {
      "epoch": 0.7269328054547454,
      "grad_norm": 0.27713724970817566,
      "learning_rate": 8e-05,
      "loss": 1.6111,
      "step": 3305
    },
    {
      "epoch": 0.7271527548663808,
      "grad_norm": 0.2909323275089264,
      "learning_rate": 8e-05,
      "loss": 1.7006,
      "step": 3306
    },
    {
      "epoch": 0.727372704278016,
      "grad_norm": 0.28384488821029663,
      "learning_rate": 8e-05,
      "loss": 1.5633,
      "step": 3307
    },
    {
      "epoch": 0.7275926536896514,
      "grad_norm": 0.30551770329475403,
      "learning_rate": 8e-05,
      "loss": 1.7208,
      "step": 3308
    },
    {
      "epoch": 0.7278126031012867,
      "grad_norm": 0.29794755578041077,
      "learning_rate": 8e-05,
      "loss": 1.7839,
      "step": 3309
    },
    {
      "epoch": 0.728032552512922,
      "grad_norm": 0.30165839195251465,
      "learning_rate": 8e-05,
      "loss": 1.7569,
      "step": 3310
    },
    {
      "epoch": 0.7282525019245574,
      "grad_norm": 0.2731498181819916,
      "learning_rate": 8e-05,
      "loss": 1.4881,
      "step": 3311
    },
    {
      "epoch": 0.7284724513361927,
      "grad_norm": 0.29713213443756104,
      "learning_rate": 8e-05,
      "loss": 1.7023,
      "step": 3312
    },
    {
      "epoch": 0.728692400747828,
      "grad_norm": 0.30032122135162354,
      "learning_rate": 8e-05,
      "loss": 1.7717,
      "step": 3313
    },
    {
      "epoch": 0.7289123501594633,
      "grad_norm": 0.2873935401439667,
      "learning_rate": 8e-05,
      "loss": 1.8062,
      "step": 3314
    },
    {
      "epoch": 0.7291322995710986,
      "grad_norm": 0.2813476026058197,
      "learning_rate": 8e-05,
      "loss": 1.6503,
      "step": 3315
    },
    {
      "epoch": 0.729352248982734,
      "grad_norm": 0.28851833939552307,
      "learning_rate": 8e-05,
      "loss": 1.5839,
      "step": 3316
    },
    {
      "epoch": 0.7295721983943693,
      "grad_norm": 0.29246899485588074,
      "learning_rate": 8e-05,
      "loss": 1.7905,
      "step": 3317
    },
    {
      "epoch": 0.7297921478060047,
      "grad_norm": 0.2823907434940338,
      "learning_rate": 8e-05,
      "loss": 1.695,
      "step": 3318
    },
    {
      "epoch": 0.7300120972176399,
      "grad_norm": 0.2800486981868744,
      "learning_rate": 8e-05,
      "loss": 1.7131,
      "step": 3319
    },
    {
      "epoch": 0.7302320466292753,
      "grad_norm": 0.28211459517478943,
      "learning_rate": 8e-05,
      "loss": 1.5183,
      "step": 3320
    },
    {
      "epoch": 0.7304519960409106,
      "grad_norm": 0.28503546118736267,
      "learning_rate": 8e-05,
      "loss": 1.4416,
      "step": 3321
    },
    {
      "epoch": 0.7306719454525459,
      "grad_norm": 0.2907847464084625,
      "learning_rate": 8e-05,
      "loss": 1.6119,
      "step": 3322
    },
    {
      "epoch": 0.7308918948641813,
      "grad_norm": 0.2849101722240448,
      "learning_rate": 8e-05,
      "loss": 1.5296,
      "step": 3323
    },
    {
      "epoch": 0.7311118442758165,
      "grad_norm": 0.2884346842765808,
      "learning_rate": 8e-05,
      "loss": 1.6141,
      "step": 3324
    },
    {
      "epoch": 0.7313317936874519,
      "grad_norm": 0.30081263184547424,
      "learning_rate": 8e-05,
      "loss": 1.7284,
      "step": 3325
    },
    {
      "epoch": 0.7315517430990872,
      "grad_norm": 0.2910812497138977,
      "learning_rate": 8e-05,
      "loss": 1.5262,
      "step": 3326
    },
    {
      "epoch": 0.7317716925107225,
      "grad_norm": 0.2986319661140442,
      "learning_rate": 8e-05,
      "loss": 1.6879,
      "step": 3327
    },
    {
      "epoch": 0.7319916419223579,
      "grad_norm": 0.2961769700050354,
      "learning_rate": 8e-05,
      "loss": 1.7522,
      "step": 3328
    },
    {
      "epoch": 0.7322115913339932,
      "grad_norm": 0.2825765907764435,
      "learning_rate": 8e-05,
      "loss": 1.7106,
      "step": 3329
    },
    {
      "epoch": 0.7324315407456286,
      "grad_norm": 0.2914576530456543,
      "learning_rate": 8e-05,
      "loss": 1.6491,
      "step": 3330
    },
    {
      "epoch": 0.7326514901572638,
      "grad_norm": 0.2786273658275604,
      "learning_rate": 8e-05,
      "loss": 1.7187,
      "step": 3331
    },
    {
      "epoch": 0.7328714395688991,
      "grad_norm": 0.27736160159111023,
      "learning_rate": 8e-05,
      "loss": 1.5219,
      "step": 3332
    },
    {
      "epoch": 0.7330913889805345,
      "grad_norm": 0.289495587348938,
      "learning_rate": 8e-05,
      "loss": 1.7318,
      "step": 3333
    },
    {
      "epoch": 0.7333113383921698,
      "grad_norm": 0.31175169348716736,
      "learning_rate": 8e-05,
      "loss": 1.8213,
      "step": 3334
    },
    {
      "epoch": 0.7335312878038052,
      "grad_norm": 0.29931631684303284,
      "learning_rate": 8e-05,
      "loss": 1.8126,
      "step": 3335
    },
    {
      "epoch": 0.7337512372154404,
      "grad_norm": 0.2750184237957001,
      "learning_rate": 8e-05,
      "loss": 1.5831,
      "step": 3336
    },
    {
      "epoch": 0.7339711866270757,
      "grad_norm": 0.30192333459854126,
      "learning_rate": 8e-05,
      "loss": 1.7635,
      "step": 3337
    },
    {
      "epoch": 0.7341911360387111,
      "grad_norm": 0.28221380710601807,
      "learning_rate": 8e-05,
      "loss": 1.5748,
      "step": 3338
    },
    {
      "epoch": 0.7344110854503464,
      "grad_norm": 0.2903328239917755,
      "learning_rate": 8e-05,
      "loss": 1.6313,
      "step": 3339
    },
    {
      "epoch": 0.7346310348619818,
      "grad_norm": 0.2983735501766205,
      "learning_rate": 8e-05,
      "loss": 1.6551,
      "step": 3340
    },
    {
      "epoch": 0.7348509842736171,
      "grad_norm": 0.29272475838661194,
      "learning_rate": 8e-05,
      "loss": 1.7012,
      "step": 3341
    },
    {
      "epoch": 0.7350709336852523,
      "grad_norm": 0.3180513083934784,
      "learning_rate": 8e-05,
      "loss": 1.7281,
      "step": 3342
    },
    {
      "epoch": 0.7352908830968877,
      "grad_norm": 0.3015908896923065,
      "learning_rate": 8e-05,
      "loss": 1.7734,
      "step": 3343
    },
    {
      "epoch": 0.735510832508523,
      "grad_norm": 0.303092360496521,
      "learning_rate": 8e-05,
      "loss": 1.708,
      "step": 3344
    },
    {
      "epoch": 0.7357307819201584,
      "grad_norm": 0.2860123813152313,
      "learning_rate": 8e-05,
      "loss": 1.7253,
      "step": 3345
    },
    {
      "epoch": 0.7359507313317937,
      "grad_norm": 0.2848500907421112,
      "learning_rate": 8e-05,
      "loss": 1.6338,
      "step": 3346
    },
    {
      "epoch": 0.7361706807434291,
      "grad_norm": 0.2954801321029663,
      "learning_rate": 8e-05,
      "loss": 1.7751,
      "step": 3347
    },
    {
      "epoch": 0.7363906301550643,
      "grad_norm": 0.36568450927734375,
      "learning_rate": 8e-05,
      "loss": 1.9713,
      "step": 3348
    },
    {
      "epoch": 0.7366105795666996,
      "grad_norm": 0.2932252585887909,
      "learning_rate": 8e-05,
      "loss": 1.6413,
      "step": 3349
    },
    {
      "epoch": 0.736830528978335,
      "grad_norm": 0.2913023829460144,
      "learning_rate": 8e-05,
      "loss": 1.6776,
      "step": 3350
    },
    {
      "epoch": 0.7370504783899703,
      "grad_norm": 0.30034127831459045,
      "learning_rate": 8e-05,
      "loss": 1.8318,
      "step": 3351
    },
    {
      "epoch": 0.7372704278016057,
      "grad_norm": 0.28994688391685486,
      "learning_rate": 8e-05,
      "loss": 1.6315,
      "step": 3352
    },
    {
      "epoch": 0.737490377213241,
      "grad_norm": 0.30393558740615845,
      "learning_rate": 8e-05,
      "loss": 1.7876,
      "step": 3353
    },
    {
      "epoch": 0.7377103266248762,
      "grad_norm": 0.3077211380004883,
      "learning_rate": 8e-05,
      "loss": 1.7739,
      "step": 3354
    },
    {
      "epoch": 0.7379302760365116,
      "grad_norm": 0.2963906526565552,
      "learning_rate": 8e-05,
      "loss": 1.7134,
      "step": 3355
    },
    {
      "epoch": 0.7381502254481469,
      "grad_norm": 0.30240267515182495,
      "learning_rate": 8e-05,
      "loss": 1.7058,
      "step": 3356
    },
    {
      "epoch": 0.7383701748597823,
      "grad_norm": 0.2841428518295288,
      "learning_rate": 8e-05,
      "loss": 1.5885,
      "step": 3357
    },
    {
      "epoch": 0.7385901242714176,
      "grad_norm": 0.2864632308483124,
      "learning_rate": 8e-05,
      "loss": 1.4941,
      "step": 3358
    },
    {
      "epoch": 0.7388100736830528,
      "grad_norm": 0.3023029565811157,
      "learning_rate": 8e-05,
      "loss": 1.7757,
      "step": 3359
    },
    {
      "epoch": 0.7390300230946882,
      "grad_norm": 0.29607269167900085,
      "learning_rate": 8e-05,
      "loss": 1.7288,
      "step": 3360
    },
    {
      "epoch": 0.7392499725063235,
      "grad_norm": 0.28938642144203186,
      "learning_rate": 8e-05,
      "loss": 1.6216,
      "step": 3361
    },
    {
      "epoch": 0.7394699219179589,
      "grad_norm": 0.2934838831424713,
      "learning_rate": 8e-05,
      "loss": 1.7762,
      "step": 3362
    },
    {
      "epoch": 0.7396898713295942,
      "grad_norm": 0.3048613369464874,
      "learning_rate": 8e-05,
      "loss": 1.7296,
      "step": 3363
    },
    {
      "epoch": 0.7399098207412295,
      "grad_norm": 0.2915562689304352,
      "learning_rate": 8e-05,
      "loss": 1.6688,
      "step": 3364
    },
    {
      "epoch": 0.7401297701528649,
      "grad_norm": 0.2817826271057129,
      "learning_rate": 8e-05,
      "loss": 1.7981,
      "step": 3365
    },
    {
      "epoch": 0.7403497195645001,
      "grad_norm": 0.31884127855300903,
      "learning_rate": 8e-05,
      "loss": 1.7536,
      "step": 3366
    },
    {
      "epoch": 0.7405696689761355,
      "grad_norm": 0.2934412956237793,
      "learning_rate": 8e-05,
      "loss": 1.6895,
      "step": 3367
    },
    {
      "epoch": 0.7407896183877708,
      "grad_norm": 0.279256671667099,
      "learning_rate": 8e-05,
      "loss": 1.681,
      "step": 3368
    },
    {
      "epoch": 0.7410095677994062,
      "grad_norm": 0.2802670896053314,
      "learning_rate": 8e-05,
      "loss": 1.6775,
      "step": 3369
    },
    {
      "epoch": 0.7412295172110415,
      "grad_norm": 0.28715866804122925,
      "learning_rate": 8e-05,
      "loss": 1.7328,
      "step": 3370
    },
    {
      "epoch": 0.7414494666226767,
      "grad_norm": 0.28366830945014954,
      "learning_rate": 8e-05,
      "loss": 1.6503,
      "step": 3371
    },
    {
      "epoch": 0.7416694160343121,
      "grad_norm": 0.27768293023109436,
      "learning_rate": 8e-05,
      "loss": 1.6495,
      "step": 3372
    },
    {
      "epoch": 0.7418893654459474,
      "grad_norm": 0.27918893098831177,
      "learning_rate": 8e-05,
      "loss": 1.5964,
      "step": 3373
    },
    {
      "epoch": 0.7421093148575828,
      "grad_norm": 0.31968846917152405,
      "learning_rate": 8e-05,
      "loss": 1.7707,
      "step": 3374
    },
    {
      "epoch": 0.7423292642692181,
      "grad_norm": 0.2908543348312378,
      "learning_rate": 8e-05,
      "loss": 1.7626,
      "step": 3375
    },
    {
      "epoch": 0.7425492136808534,
      "grad_norm": 0.2790231704711914,
      "learning_rate": 8e-05,
      "loss": 1.571,
      "step": 3376
    },
    {
      "epoch": 0.7427691630924887,
      "grad_norm": 0.2957472503185272,
      "learning_rate": 8e-05,
      "loss": 1.6063,
      "step": 3377
    },
    {
      "epoch": 0.742989112504124,
      "grad_norm": 0.2858772277832031,
      "learning_rate": 8e-05,
      "loss": 1.5614,
      "step": 3378
    },
    {
      "epoch": 0.7432090619157594,
      "grad_norm": 0.2928299605846405,
      "learning_rate": 8e-05,
      "loss": 1.6377,
      "step": 3379
    },
    {
      "epoch": 0.7434290113273947,
      "grad_norm": 0.28481486439704895,
      "learning_rate": 8e-05,
      "loss": 1.5916,
      "step": 3380
    },
    {
      "epoch": 0.74364896073903,
      "grad_norm": 0.30001428723335266,
      "learning_rate": 8e-05,
      "loss": 1.6408,
      "step": 3381
    },
    {
      "epoch": 0.7438689101506654,
      "grad_norm": 0.31076952815055847,
      "learning_rate": 8e-05,
      "loss": 1.7663,
      "step": 3382
    },
    {
      "epoch": 0.7440888595623006,
      "grad_norm": 0.2954659163951874,
      "learning_rate": 8e-05,
      "loss": 1.8063,
      "step": 3383
    },
    {
      "epoch": 0.744308808973936,
      "grad_norm": 0.2676610052585602,
      "learning_rate": 8e-05,
      "loss": 1.6039,
      "step": 3384
    },
    {
      "epoch": 0.7445287583855713,
      "grad_norm": 0.28663912415504456,
      "learning_rate": 8e-05,
      "loss": 1.486,
      "step": 3385
    },
    {
      "epoch": 0.7447487077972066,
      "grad_norm": 0.28007110953330994,
      "learning_rate": 8e-05,
      "loss": 1.5365,
      "step": 3386
    },
    {
      "epoch": 0.744968657208842,
      "grad_norm": 0.28811848163604736,
      "learning_rate": 8e-05,
      "loss": 1.6558,
      "step": 3387
    },
    {
      "epoch": 0.7451886066204773,
      "grad_norm": 0.313472181558609,
      "learning_rate": 8e-05,
      "loss": 1.7308,
      "step": 3388
    },
    {
      "epoch": 0.7454085560321126,
      "grad_norm": 0.28725314140319824,
      "learning_rate": 8e-05,
      "loss": 1.7269,
      "step": 3389
    },
    {
      "epoch": 0.7456285054437479,
      "grad_norm": 0.3030818998813629,
      "learning_rate": 8e-05,
      "loss": 1.6803,
      "step": 3390
    },
    {
      "epoch": 0.7458484548553833,
      "grad_norm": 0.34181302785873413,
      "learning_rate": 8e-05,
      "loss": 1.8658,
      "step": 3391
    },
    {
      "epoch": 0.7460684042670186,
      "grad_norm": 0.29316088557243347,
      "learning_rate": 8e-05,
      "loss": 1.6458,
      "step": 3392
    },
    {
      "epoch": 0.7462883536786539,
      "grad_norm": 0.3064456582069397,
      "learning_rate": 8e-05,
      "loss": 1.8798,
      "step": 3393
    },
    {
      "epoch": 0.7465083030902893,
      "grad_norm": 0.28297048807144165,
      "learning_rate": 8e-05,
      "loss": 1.6796,
      "step": 3394
    },
    {
      "epoch": 0.7467282525019245,
      "grad_norm": 0.2969253659248352,
      "learning_rate": 8e-05,
      "loss": 1.6369,
      "step": 3395
    },
    {
      "epoch": 0.7469482019135599,
      "grad_norm": 0.3293825089931488,
      "learning_rate": 8e-05,
      "loss": 1.9732,
      "step": 3396
    },
    {
      "epoch": 0.7471681513251952,
      "grad_norm": 0.2745886445045471,
      "learning_rate": 8e-05,
      "loss": 1.537,
      "step": 3397
    },
    {
      "epoch": 0.7473881007368305,
      "grad_norm": 0.28334876894950867,
      "learning_rate": 8e-05,
      "loss": 1.582,
      "step": 3398
    },
    {
      "epoch": 0.7476080501484659,
      "grad_norm": 0.29402390122413635,
      "learning_rate": 8e-05,
      "loss": 1.8205,
      "step": 3399
    },
    {
      "epoch": 0.7478279995601012,
      "grad_norm": 0.3036934733390808,
      "learning_rate": 8e-05,
      "loss": 1.6388,
      "step": 3400
    },
    {
      "epoch": 0.7480479489717365,
      "grad_norm": 0.2951960563659668,
      "learning_rate": 8e-05,
      "loss": 1.8011,
      "step": 3401
    },
    {
      "epoch": 0.7482678983833718,
      "grad_norm": 0.29489466547966003,
      "learning_rate": 8e-05,
      "loss": 1.8069,
      "step": 3402
    },
    {
      "epoch": 0.7484878477950071,
      "grad_norm": 0.2927097976207733,
      "learning_rate": 8e-05,
      "loss": 1.7201,
      "step": 3403
    },
    {
      "epoch": 0.7487077972066425,
      "grad_norm": 0.277259886264801,
      "learning_rate": 8e-05,
      "loss": 1.6,
      "step": 3404
    },
    {
      "epoch": 0.7489277466182778,
      "grad_norm": 0.2830856144428253,
      "learning_rate": 8e-05,
      "loss": 1.6689,
      "step": 3405
    },
    {
      "epoch": 0.7491476960299132,
      "grad_norm": 0.3067016899585724,
      "learning_rate": 8e-05,
      "loss": 1.8059,
      "step": 3406
    },
    {
      "epoch": 0.7493676454415484,
      "grad_norm": 0.2836841940879822,
      "learning_rate": 8e-05,
      "loss": 1.6509,
      "step": 3407
    },
    {
      "epoch": 0.7495875948531837,
      "grad_norm": 0.2890060842037201,
      "learning_rate": 8e-05,
      "loss": 1.7764,
      "step": 3408
    },
    {
      "epoch": 0.7498075442648191,
      "grad_norm": 0.29502347111701965,
      "learning_rate": 8e-05,
      "loss": 1.6978,
      "step": 3409
    },
    {
      "epoch": 0.7500274936764544,
      "grad_norm": 0.3138415217399597,
      "learning_rate": 8e-05,
      "loss": 1.6973,
      "step": 3410
    },
    {
      "epoch": 0.7502474430880898,
      "grad_norm": 0.2842028737068176,
      "learning_rate": 8e-05,
      "loss": 1.5305,
      "step": 3411
    },
    {
      "epoch": 0.750467392499725,
      "grad_norm": 0.3058598041534424,
      "learning_rate": 8e-05,
      "loss": 1.7835,
      "step": 3412
    },
    {
      "epoch": 0.7506873419113604,
      "grad_norm": 0.2859271466732025,
      "learning_rate": 8e-05,
      "loss": 1.6406,
      "step": 3413
    },
    {
      "epoch": 0.7509072913229957,
      "grad_norm": 0.2928830683231354,
      "learning_rate": 8e-05,
      "loss": 1.5819,
      "step": 3414
    },
    {
      "epoch": 0.751127240734631,
      "grad_norm": 0.3138205409049988,
      "learning_rate": 8e-05,
      "loss": 1.5155,
      "step": 3415
    },
    {
      "epoch": 0.7513471901462664,
      "grad_norm": 0.3159571886062622,
      "learning_rate": 8e-05,
      "loss": 1.8749,
      "step": 3416
    },
    {
      "epoch": 0.7515671395579017,
      "grad_norm": 0.2998178005218506,
      "learning_rate": 8e-05,
      "loss": 1.6651,
      "step": 3417
    },
    {
      "epoch": 0.751787088969537,
      "grad_norm": 0.30106493830680847,
      "learning_rate": 8e-05,
      "loss": 1.7626,
      "step": 3418
    },
    {
      "epoch": 0.7520070383811723,
      "grad_norm": 0.2966890037059784,
      "learning_rate": 8e-05,
      "loss": 1.6505,
      "step": 3419
    },
    {
      "epoch": 0.7522269877928076,
      "grad_norm": 0.28747057914733887,
      "learning_rate": 8e-05,
      "loss": 1.5537,
      "step": 3420
    },
    {
      "epoch": 0.752446937204443,
      "grad_norm": 0.3139256536960602,
      "learning_rate": 8e-05,
      "loss": 1.65,
      "step": 3421
    },
    {
      "epoch": 0.7526668866160783,
      "grad_norm": 0.2905026972293854,
      "learning_rate": 8e-05,
      "loss": 1.7456,
      "step": 3422
    },
    {
      "epoch": 0.7528868360277137,
      "grad_norm": 0.29663148522377014,
      "learning_rate": 8e-05,
      "loss": 1.6986,
      "step": 3423
    },
    {
      "epoch": 0.753106785439349,
      "grad_norm": 0.2942657470703125,
      "learning_rate": 8e-05,
      "loss": 1.7396,
      "step": 3424
    },
    {
      "epoch": 0.7533267348509842,
      "grad_norm": 0.2846404016017914,
      "learning_rate": 8e-05,
      "loss": 1.5673,
      "step": 3425
    },
    {
      "epoch": 0.7535466842626196,
      "grad_norm": 0.3121284544467926,
      "learning_rate": 8e-05,
      "loss": 1.7852,
      "step": 3426
    },
    {
      "epoch": 0.7537666336742549,
      "grad_norm": 0.3095207214355469,
      "learning_rate": 8e-05,
      "loss": 1.7747,
      "step": 3427
    },
    {
      "epoch": 0.7539865830858903,
      "grad_norm": 0.28364792466163635,
      "learning_rate": 8e-05,
      "loss": 1.6513,
      "step": 3428
    },
    {
      "epoch": 0.7542065324975256,
      "grad_norm": 0.2978142201900482,
      "learning_rate": 8e-05,
      "loss": 1.7227,
      "step": 3429
    },
    {
      "epoch": 0.7544264819091608,
      "grad_norm": 0.2860453724861145,
      "learning_rate": 8e-05,
      "loss": 1.5699,
      "step": 3430
    },
    {
      "epoch": 0.7546464313207962,
      "grad_norm": 0.30176058411598206,
      "learning_rate": 8e-05,
      "loss": 1.683,
      "step": 3431
    },
    {
      "epoch": 0.7548663807324315,
      "grad_norm": 0.2823827266693115,
      "learning_rate": 8e-05,
      "loss": 1.4892,
      "step": 3432
    },
    {
      "epoch": 0.7550863301440669,
      "grad_norm": 0.2879229784011841,
      "learning_rate": 8e-05,
      "loss": 1.5821,
      "step": 3433
    },
    {
      "epoch": 0.7553062795557022,
      "grad_norm": 0.29188740253448486,
      "learning_rate": 8e-05,
      "loss": 1.5946,
      "step": 3434
    },
    {
      "epoch": 0.7555262289673376,
      "grad_norm": 0.29899924993515015,
      "learning_rate": 8e-05,
      "loss": 1.7849,
      "step": 3435
    },
    {
      "epoch": 0.7557461783789728,
      "grad_norm": 0.29541707038879395,
      "learning_rate": 8e-05,
      "loss": 1.6732,
      "step": 3436
    },
    {
      "epoch": 0.7559661277906081,
      "grad_norm": 0.2920849919319153,
      "learning_rate": 8e-05,
      "loss": 1.5479,
      "step": 3437
    },
    {
      "epoch": 0.7561860772022435,
      "grad_norm": 0.3060014545917511,
      "learning_rate": 8e-05,
      "loss": 1.7094,
      "step": 3438
    },
    {
      "epoch": 0.7564060266138788,
      "grad_norm": 0.28043729066848755,
      "learning_rate": 8e-05,
      "loss": 1.4717,
      "step": 3439
    },
    {
      "epoch": 0.7566259760255142,
      "grad_norm": 0.29130029678344727,
      "learning_rate": 8e-05,
      "loss": 1.6355,
      "step": 3440
    },
    {
      "epoch": 0.7568459254371495,
      "grad_norm": 0.2969004213809967,
      "learning_rate": 8e-05,
      "loss": 1.6808,
      "step": 3441
    },
    {
      "epoch": 0.7570658748487847,
      "grad_norm": 0.3089723289012909,
      "learning_rate": 8e-05,
      "loss": 1.7368,
      "step": 3442
    },
    {
      "epoch": 0.7572858242604201,
      "grad_norm": 0.29811516404151917,
      "learning_rate": 8e-05,
      "loss": 1.6317,
      "step": 3443
    },
    {
      "epoch": 0.7575057736720554,
      "grad_norm": 0.2753433883190155,
      "learning_rate": 8e-05,
      "loss": 1.598,
      "step": 3444
    },
    {
      "epoch": 0.7577257230836908,
      "grad_norm": 0.2914634943008423,
      "learning_rate": 8e-05,
      "loss": 1.7117,
      "step": 3445
    },
    {
      "epoch": 0.7579456724953261,
      "grad_norm": 0.3008229732513428,
      "learning_rate": 8e-05,
      "loss": 1.7158,
      "step": 3446
    },
    {
      "epoch": 0.7581656219069614,
      "grad_norm": 0.297671377658844,
      "learning_rate": 8e-05,
      "loss": 1.793,
      "step": 3447
    },
    {
      "epoch": 0.7583855713185967,
      "grad_norm": 0.27893882989883423,
      "learning_rate": 8e-05,
      "loss": 1.6701,
      "step": 3448
    },
    {
      "epoch": 0.758605520730232,
      "grad_norm": 0.2874872386455536,
      "learning_rate": 8e-05,
      "loss": 1.5559,
      "step": 3449
    },
    {
      "epoch": 0.7588254701418674,
      "grad_norm": 0.30345162749290466,
      "learning_rate": 8e-05,
      "loss": 1.6925,
      "step": 3450
    },
    {
      "epoch": 0.7590454195535027,
      "grad_norm": 0.3018907308578491,
      "learning_rate": 8e-05,
      "loss": 1.6325,
      "step": 3451
    },
    {
      "epoch": 0.759265368965138,
      "grad_norm": 0.30021166801452637,
      "learning_rate": 8e-05,
      "loss": 1.6077,
      "step": 3452
    },
    {
      "epoch": 0.7594853183767734,
      "grad_norm": 0.2934636175632477,
      "learning_rate": 8e-05,
      "loss": 1.681,
      "step": 3453
    },
    {
      "epoch": 0.7597052677884086,
      "grad_norm": 0.30079886317253113,
      "learning_rate": 8e-05,
      "loss": 1.7378,
      "step": 3454
    },
    {
      "epoch": 0.759925217200044,
      "grad_norm": 0.29279422760009766,
      "learning_rate": 8e-05,
      "loss": 1.5918,
      "step": 3455
    },
    {
      "epoch": 0.7601451666116793,
      "grad_norm": 0.2980215847492218,
      "learning_rate": 8e-05,
      "loss": 1.7888,
      "step": 3456
    },
    {
      "epoch": 0.7603651160233147,
      "grad_norm": 0.27554240822792053,
      "learning_rate": 8e-05,
      "loss": 1.5351,
      "step": 3457
    },
    {
      "epoch": 0.76058506543495,
      "grad_norm": 0.2831701338291168,
      "learning_rate": 8e-05,
      "loss": 1.7558,
      "step": 3458
    },
    {
      "epoch": 0.7608050148465852,
      "grad_norm": 0.29401731491088867,
      "learning_rate": 8e-05,
      "loss": 1.69,
      "step": 3459
    },
    {
      "epoch": 0.7610249642582206,
      "grad_norm": 0.29017552733421326,
      "learning_rate": 8e-05,
      "loss": 1.6145,
      "step": 3460
    },
    {
      "epoch": 0.7612449136698559,
      "grad_norm": 0.2881031036376953,
      "learning_rate": 8e-05,
      "loss": 1.6464,
      "step": 3461
    },
    {
      "epoch": 0.7614648630814913,
      "grad_norm": 0.3012145757675171,
      "learning_rate": 8e-05,
      "loss": 1.6904,
      "step": 3462
    },
    {
      "epoch": 0.7616848124931266,
      "grad_norm": 0.29406794905662537,
      "learning_rate": 8e-05,
      "loss": 1.6375,
      "step": 3463
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 0.28518983721733093,
      "learning_rate": 8e-05,
      "loss": 1.5478,
      "step": 3464
    },
    {
      "epoch": 0.7621247113163973,
      "grad_norm": 0.3052254021167755,
      "learning_rate": 8e-05,
      "loss": 1.8076,
      "step": 3465
    },
    {
      "epoch": 0.7623446607280325,
      "grad_norm": 0.2873866856098175,
      "learning_rate": 8e-05,
      "loss": 1.7547,
      "step": 3466
    },
    {
      "epoch": 0.7625646101396679,
      "grad_norm": 0.28812938928604126,
      "learning_rate": 8e-05,
      "loss": 1.6502,
      "step": 3467
    },
    {
      "epoch": 0.7627845595513032,
      "grad_norm": 0.28548526763916016,
      "learning_rate": 8e-05,
      "loss": 1.7744,
      "step": 3468
    },
    {
      "epoch": 0.7630045089629385,
      "grad_norm": 0.30266743898391724,
      "learning_rate": 8e-05,
      "loss": 1.7142,
      "step": 3469
    },
    {
      "epoch": 0.7632244583745739,
      "grad_norm": 0.29223671555519104,
      "learning_rate": 8e-05,
      "loss": 1.5969,
      "step": 3470
    },
    {
      "epoch": 0.7634444077862091,
      "grad_norm": 0.2857435643672943,
      "learning_rate": 8e-05,
      "loss": 1.6398,
      "step": 3471
    },
    {
      "epoch": 0.7636643571978445,
      "grad_norm": 0.28153562545776367,
      "learning_rate": 8e-05,
      "loss": 1.5967,
      "step": 3472
    },
    {
      "epoch": 0.7638843066094798,
      "grad_norm": 0.28425905108451843,
      "learning_rate": 8e-05,
      "loss": 1.5901,
      "step": 3473
    },
    {
      "epoch": 0.7641042560211151,
      "grad_norm": 0.3140738308429718,
      "learning_rate": 8e-05,
      "loss": 1.772,
      "step": 3474
    },
    {
      "epoch": 0.7643242054327505,
      "grad_norm": 0.2958965301513672,
      "learning_rate": 8e-05,
      "loss": 1.8743,
      "step": 3475
    },
    {
      "epoch": 0.7645441548443858,
      "grad_norm": 0.3059069514274597,
      "learning_rate": 8e-05,
      "loss": 1.6789,
      "step": 3476
    },
    {
      "epoch": 0.7647641042560211,
      "grad_norm": 0.3015092611312866,
      "learning_rate": 8e-05,
      "loss": 1.7492,
      "step": 3477
    },
    {
      "epoch": 0.7649840536676564,
      "grad_norm": 0.2840091586112976,
      "learning_rate": 8e-05,
      "loss": 1.6442,
      "step": 3478
    },
    {
      "epoch": 0.7652040030792918,
      "grad_norm": 0.2928544282913208,
      "learning_rate": 8e-05,
      "loss": 1.5534,
      "step": 3479
    },
    {
      "epoch": 0.7654239524909271,
      "grad_norm": 0.2872399389743805,
      "learning_rate": 8e-05,
      "loss": 1.7049,
      "step": 3480
    },
    {
      "epoch": 0.7656439019025624,
      "grad_norm": 0.29002323746681213,
      "learning_rate": 8e-05,
      "loss": 1.6036,
      "step": 3481
    },
    {
      "epoch": 0.7658638513141978,
      "grad_norm": 0.28830286860466003,
      "learning_rate": 8e-05,
      "loss": 1.6805,
      "step": 3482
    },
    {
      "epoch": 0.766083800725833,
      "grad_norm": 0.2776120901107788,
      "learning_rate": 8e-05,
      "loss": 1.609,
      "step": 3483
    },
    {
      "epoch": 0.7663037501374684,
      "grad_norm": 0.2959437072277069,
      "learning_rate": 8e-05,
      "loss": 1.6657,
      "step": 3484
    },
    {
      "epoch": 0.7665236995491037,
      "grad_norm": 0.29850873351097107,
      "learning_rate": 8e-05,
      "loss": 1.4705,
      "step": 3485
    },
    {
      "epoch": 0.766743648960739,
      "grad_norm": 0.2813439667224884,
      "learning_rate": 8e-05,
      "loss": 1.5541,
      "step": 3486
    },
    {
      "epoch": 0.7669635983723744,
      "grad_norm": 0.3008776009082794,
      "learning_rate": 8e-05,
      "loss": 1.7506,
      "step": 3487
    },
    {
      "epoch": 0.7671835477840097,
      "grad_norm": 0.28104785084724426,
      "learning_rate": 8e-05,
      "loss": 1.5551,
      "step": 3488
    },
    {
      "epoch": 0.767403497195645,
      "grad_norm": 0.28244099020957947,
      "learning_rate": 8e-05,
      "loss": 1.644,
      "step": 3489
    },
    {
      "epoch": 0.7676234466072803,
      "grad_norm": 0.2998720407485962,
      "learning_rate": 8e-05,
      "loss": 1.7757,
      "step": 3490
    },
    {
      "epoch": 0.7678433960189156,
      "grad_norm": 0.28794363141059875,
      "learning_rate": 8e-05,
      "loss": 1.5814,
      "step": 3491
    },
    {
      "epoch": 0.768063345430551,
      "grad_norm": 0.2936362326145172,
      "learning_rate": 8e-05,
      "loss": 1.602,
      "step": 3492
    },
    {
      "epoch": 0.7682832948421863,
      "grad_norm": 0.282718300819397,
      "learning_rate": 8e-05,
      "loss": 1.5837,
      "step": 3493
    },
    {
      "epoch": 0.7685032442538217,
      "grad_norm": 0.2850697934627533,
      "learning_rate": 8e-05,
      "loss": 1.708,
      "step": 3494
    },
    {
      "epoch": 0.7687231936654569,
      "grad_norm": 0.3146105110645294,
      "learning_rate": 8e-05,
      "loss": 1.7864,
      "step": 3495
    },
    {
      "epoch": 0.7689431430770922,
      "grad_norm": 0.295669823884964,
      "learning_rate": 8e-05,
      "loss": 1.7563,
      "step": 3496
    },
    {
      "epoch": 0.7691630924887276,
      "grad_norm": 0.2910419702529907,
      "learning_rate": 8e-05,
      "loss": 1.7159,
      "step": 3497
    },
    {
      "epoch": 0.7693830419003629,
      "grad_norm": 0.2934516668319702,
      "learning_rate": 8e-05,
      "loss": 1.7183,
      "step": 3498
    },
    {
      "epoch": 0.7696029913119983,
      "grad_norm": 0.2876817584037781,
      "learning_rate": 8e-05,
      "loss": 1.652,
      "step": 3499
    },
    {
      "epoch": 0.7698229407236336,
      "grad_norm": 0.2993420660495758,
      "learning_rate": 8e-05,
      "loss": 1.8512,
      "step": 3500
    },
    {
      "epoch": 0.7700428901352689,
      "grad_norm": 0.28241509199142456,
      "learning_rate": 8e-05,
      "loss": 1.6526,
      "step": 3501
    },
    {
      "epoch": 0.7702628395469042,
      "grad_norm": 0.2991597056388855,
      "learning_rate": 8e-05,
      "loss": 1.656,
      "step": 3502
    },
    {
      "epoch": 0.7704827889585395,
      "grad_norm": 0.26715072989463806,
      "learning_rate": 8e-05,
      "loss": 1.4417,
      "step": 3503
    },
    {
      "epoch": 0.7707027383701749,
      "grad_norm": 0.28751102089881897,
      "learning_rate": 8e-05,
      "loss": 1.6471,
      "step": 3504
    },
    {
      "epoch": 0.7709226877818102,
      "grad_norm": 0.297297865152359,
      "learning_rate": 8e-05,
      "loss": 1.6943,
      "step": 3505
    },
    {
      "epoch": 0.7711426371934456,
      "grad_norm": 0.2984234690666199,
      "learning_rate": 8e-05,
      "loss": 1.6245,
      "step": 3506
    },
    {
      "epoch": 0.7713625866050808,
      "grad_norm": 0.29237473011016846,
      "learning_rate": 8e-05,
      "loss": 1.6624,
      "step": 3507
    },
    {
      "epoch": 0.7715825360167161,
      "grad_norm": 0.2652399241924286,
      "learning_rate": 8e-05,
      "loss": 1.3897,
      "step": 3508
    },
    {
      "epoch": 0.7718024854283515,
      "grad_norm": 0.28701311349868774,
      "learning_rate": 8e-05,
      "loss": 1.5524,
      "step": 3509
    },
    {
      "epoch": 0.7720224348399868,
      "grad_norm": 0.2850416600704193,
      "learning_rate": 8e-05,
      "loss": 1.6508,
      "step": 3510
    },
    {
      "epoch": 0.7722423842516222,
      "grad_norm": 0.30162402987480164,
      "learning_rate": 8e-05,
      "loss": 1.7517,
      "step": 3511
    },
    {
      "epoch": 0.7724623336632574,
      "grad_norm": 0.29350200295448303,
      "learning_rate": 8e-05,
      "loss": 1.7633,
      "step": 3512
    },
    {
      "epoch": 0.7726822830748927,
      "grad_norm": 0.28733333945274353,
      "learning_rate": 8e-05,
      "loss": 1.5516,
      "step": 3513
    },
    {
      "epoch": 0.7729022324865281,
      "grad_norm": 0.32462671399116516,
      "learning_rate": 8e-05,
      "loss": 1.7854,
      "step": 3514
    },
    {
      "epoch": 0.7731221818981634,
      "grad_norm": 0.3113280236721039,
      "learning_rate": 8e-05,
      "loss": 1.8449,
      "step": 3515
    },
    {
      "epoch": 0.7733421313097988,
      "grad_norm": 0.2827896773815155,
      "learning_rate": 8e-05,
      "loss": 1.7656,
      "step": 3516
    },
    {
      "epoch": 0.7735620807214341,
      "grad_norm": 0.2928074598312378,
      "learning_rate": 8e-05,
      "loss": 1.6439,
      "step": 3517
    },
    {
      "epoch": 0.7737820301330693,
      "grad_norm": 0.2791458070278168,
      "learning_rate": 8e-05,
      "loss": 1.618,
      "step": 3518
    },
    {
      "epoch": 0.7740019795447047,
      "grad_norm": 0.3217780590057373,
      "learning_rate": 8e-05,
      "loss": 1.7436,
      "step": 3519
    },
    {
      "epoch": 0.77422192895634,
      "grad_norm": 0.28555285930633545,
      "learning_rate": 8e-05,
      "loss": 1.6563,
      "step": 3520
    },
    {
      "epoch": 0.7744418783679754,
      "grad_norm": 0.30171915888786316,
      "learning_rate": 8e-05,
      "loss": 1.5573,
      "step": 3521
    },
    {
      "epoch": 0.7746618277796107,
      "grad_norm": 0.29558464884757996,
      "learning_rate": 8e-05,
      "loss": 1.7783,
      "step": 3522
    },
    {
      "epoch": 0.7748817771912461,
      "grad_norm": 0.28546035289764404,
      "learning_rate": 8e-05,
      "loss": 1.5793,
      "step": 3523
    },
    {
      "epoch": 0.7751017266028813,
      "grad_norm": 0.30701640248298645,
      "learning_rate": 8e-05,
      "loss": 1.8204,
      "step": 3524
    },
    {
      "epoch": 0.7753216760145166,
      "grad_norm": 0.2884214222431183,
      "learning_rate": 8e-05,
      "loss": 1.6179,
      "step": 3525
    },
    {
      "epoch": 0.775541625426152,
      "grad_norm": 0.3010881245136261,
      "learning_rate": 8e-05,
      "loss": 1.7521,
      "step": 3526
    },
    {
      "epoch": 0.7757615748377873,
      "grad_norm": 0.2960352301597595,
      "learning_rate": 8e-05,
      "loss": 1.6982,
      "step": 3527
    },
    {
      "epoch": 0.7759815242494227,
      "grad_norm": 0.3347279131412506,
      "learning_rate": 8e-05,
      "loss": 1.8114,
      "step": 3528
    },
    {
      "epoch": 0.776201473661058,
      "grad_norm": 0.298384428024292,
      "learning_rate": 8e-05,
      "loss": 1.676,
      "step": 3529
    },
    {
      "epoch": 0.7764214230726932,
      "grad_norm": 0.30956321954727173,
      "learning_rate": 8e-05,
      "loss": 1.7505,
      "step": 3530
    },
    {
      "epoch": 0.7766413724843286,
      "grad_norm": 0.29873716831207275,
      "learning_rate": 8e-05,
      "loss": 1.6829,
      "step": 3531
    },
    {
      "epoch": 0.7768613218959639,
      "grad_norm": 0.2934054136276245,
      "learning_rate": 8e-05,
      "loss": 1.6946,
      "step": 3532
    },
    {
      "epoch": 0.7770812713075993,
      "grad_norm": 0.3284706175327301,
      "learning_rate": 8e-05,
      "loss": 1.881,
      "step": 3533
    },
    {
      "epoch": 0.7773012207192346,
      "grad_norm": 0.29005488753318787,
      "learning_rate": 8e-05,
      "loss": 1.6785,
      "step": 3534
    },
    {
      "epoch": 0.7775211701308699,
      "grad_norm": 0.2805337607860565,
      "learning_rate": 8e-05,
      "loss": 1.6025,
      "step": 3535
    },
    {
      "epoch": 0.7777411195425052,
      "grad_norm": 0.308052122592926,
      "learning_rate": 8e-05,
      "loss": 1.8216,
      "step": 3536
    },
    {
      "epoch": 0.7779610689541405,
      "grad_norm": 0.2906242907047272,
      "learning_rate": 8e-05,
      "loss": 1.7335,
      "step": 3537
    },
    {
      "epoch": 0.7781810183657759,
      "grad_norm": 0.28683236241340637,
      "learning_rate": 8e-05,
      "loss": 1.6598,
      "step": 3538
    },
    {
      "epoch": 0.7784009677774112,
      "grad_norm": 0.3011239469051361,
      "learning_rate": 8e-05,
      "loss": 1.5983,
      "step": 3539
    },
    {
      "epoch": 0.7786209171890465,
      "grad_norm": 0.2807599604129791,
      "learning_rate": 8e-05,
      "loss": 1.5252,
      "step": 3540
    },
    {
      "epoch": 0.7788408666006819,
      "grad_norm": 0.3378777503967285,
      "learning_rate": 8e-05,
      "loss": 1.6314,
      "step": 3541
    },
    {
      "epoch": 0.7790608160123171,
      "grad_norm": 0.28623417019844055,
      "learning_rate": 8e-05,
      "loss": 1.7219,
      "step": 3542
    },
    {
      "epoch": 0.7792807654239525,
      "grad_norm": 0.2920507788658142,
      "learning_rate": 8e-05,
      "loss": 1.5725,
      "step": 3543
    },
    {
      "epoch": 0.7795007148355878,
      "grad_norm": 0.3120638132095337,
      "learning_rate": 8e-05,
      "loss": 1.637,
      "step": 3544
    },
    {
      "epoch": 0.7797206642472232,
      "grad_norm": 0.29561811685562134,
      "learning_rate": 8e-05,
      "loss": 1.6626,
      "step": 3545
    },
    {
      "epoch": 0.7799406136588585,
      "grad_norm": 0.299492746591568,
      "learning_rate": 8e-05,
      "loss": 1.7206,
      "step": 3546
    },
    {
      "epoch": 0.7801605630704938,
      "grad_norm": 0.3011503517627716,
      "learning_rate": 8e-05,
      "loss": 1.695,
      "step": 3547
    },
    {
      "epoch": 0.7803805124821291,
      "grad_norm": 0.3122527003288269,
      "learning_rate": 8e-05,
      "loss": 1.8032,
      "step": 3548
    },
    {
      "epoch": 0.7806004618937644,
      "grad_norm": 0.28565728664398193,
      "learning_rate": 8e-05,
      "loss": 1.6263,
      "step": 3549
    },
    {
      "epoch": 0.7808204113053998,
      "grad_norm": 0.28905171155929565,
      "learning_rate": 8e-05,
      "loss": 1.7008,
      "step": 3550
    },
    {
      "epoch": 0.7810403607170351,
      "grad_norm": 0.2950139343738556,
      "learning_rate": 8e-05,
      "loss": 1.6454,
      "step": 3551
    },
    {
      "epoch": 0.7812603101286704,
      "grad_norm": 0.29627618193626404,
      "learning_rate": 8e-05,
      "loss": 1.6792,
      "step": 3552
    },
    {
      "epoch": 0.7814802595403058,
      "grad_norm": 0.28478100895881653,
      "learning_rate": 8e-05,
      "loss": 1.65,
      "step": 3553
    },
    {
      "epoch": 0.781700208951941,
      "grad_norm": 0.2886268198490143,
      "learning_rate": 8e-05,
      "loss": 1.655,
      "step": 3554
    },
    {
      "epoch": 0.7819201583635764,
      "grad_norm": 0.30898573994636536,
      "learning_rate": 8e-05,
      "loss": 1.6732,
      "step": 3555
    },
    {
      "epoch": 0.7821401077752117,
      "grad_norm": 0.2944103479385376,
      "learning_rate": 8e-05,
      "loss": 1.6625,
      "step": 3556
    },
    {
      "epoch": 0.782360057186847,
      "grad_norm": 0.2840811610221863,
      "learning_rate": 8e-05,
      "loss": 1.7059,
      "step": 3557
    },
    {
      "epoch": 0.7825800065984824,
      "grad_norm": 0.31668660044670105,
      "learning_rate": 8e-05,
      "loss": 1.7343,
      "step": 3558
    },
    {
      "epoch": 0.7827999560101176,
      "grad_norm": 0.2928016483783722,
      "learning_rate": 8e-05,
      "loss": 1.5946,
      "step": 3559
    },
    {
      "epoch": 0.783019905421753,
      "grad_norm": 0.30465641617774963,
      "learning_rate": 8e-05,
      "loss": 1.8133,
      "step": 3560
    },
    {
      "epoch": 0.7832398548333883,
      "grad_norm": 0.3166663646697998,
      "learning_rate": 8e-05,
      "loss": 1.7594,
      "step": 3561
    },
    {
      "epoch": 0.7834598042450236,
      "grad_norm": 0.2896936535835266,
      "learning_rate": 8e-05,
      "loss": 1.7585,
      "step": 3562
    },
    {
      "epoch": 0.783679753656659,
      "grad_norm": 0.313091903924942,
      "learning_rate": 8e-05,
      "loss": 1.8129,
      "step": 3563
    },
    {
      "epoch": 0.7838997030682943,
      "grad_norm": 0.306034117937088,
      "learning_rate": 8e-05,
      "loss": 1.6265,
      "step": 3564
    },
    {
      "epoch": 0.7841196524799297,
      "grad_norm": 0.2841499149799347,
      "learning_rate": 8e-05,
      "loss": 1.7271,
      "step": 3565
    },
    {
      "epoch": 0.7843396018915649,
      "grad_norm": 0.29752469062805176,
      "learning_rate": 8e-05,
      "loss": 1.5481,
      "step": 3566
    },
    {
      "epoch": 0.7845595513032002,
      "grad_norm": 0.29329928755760193,
      "learning_rate": 8e-05,
      "loss": 1.7561,
      "step": 3567
    },
    {
      "epoch": 0.7847795007148356,
      "grad_norm": 0.290488600730896,
      "learning_rate": 8e-05,
      "loss": 1.7908,
      "step": 3568
    },
    {
      "epoch": 0.7849994501264709,
      "grad_norm": 0.2861136198043823,
      "learning_rate": 8e-05,
      "loss": 1.6646,
      "step": 3569
    },
    {
      "epoch": 0.7852193995381063,
      "grad_norm": 0.28423643112182617,
      "learning_rate": 8e-05,
      "loss": 1.5909,
      "step": 3570
    },
    {
      "epoch": 0.7854393489497415,
      "grad_norm": 0.3002358675003052,
      "learning_rate": 8e-05,
      "loss": 1.6773,
      "step": 3571
    },
    {
      "epoch": 0.7856592983613769,
      "grad_norm": 0.2944103479385376,
      "learning_rate": 8e-05,
      "loss": 1.5379,
      "step": 3572
    },
    {
      "epoch": 0.7858792477730122,
      "grad_norm": 0.28983041644096375,
      "learning_rate": 8e-05,
      "loss": 1.5886,
      "step": 3573
    },
    {
      "epoch": 0.7860991971846475,
      "grad_norm": 0.297059565782547,
      "learning_rate": 8e-05,
      "loss": 1.6874,
      "step": 3574
    },
    {
      "epoch": 0.7863191465962829,
      "grad_norm": 0.2881814241409302,
      "learning_rate": 8e-05,
      "loss": 1.6546,
      "step": 3575
    },
    {
      "epoch": 0.7865390960079182,
      "grad_norm": 0.30335813760757446,
      "learning_rate": 8e-05,
      "loss": 1.6688,
      "step": 3576
    },
    {
      "epoch": 0.7867590454195535,
      "grad_norm": 0.2669346332550049,
      "learning_rate": 8e-05,
      "loss": 1.414,
      "step": 3577
    },
    {
      "epoch": 0.7869789948311888,
      "grad_norm": 0.29250532388687134,
      "learning_rate": 8e-05,
      "loss": 1.6435,
      "step": 3578
    },
    {
      "epoch": 0.7871989442428241,
      "grad_norm": 0.3067176043987274,
      "learning_rate": 8e-05,
      "loss": 1.7231,
      "step": 3579
    },
    {
      "epoch": 0.7874188936544595,
      "grad_norm": 0.2805663049221039,
      "learning_rate": 8e-05,
      "loss": 1.5512,
      "step": 3580
    },
    {
      "epoch": 0.7876388430660948,
      "grad_norm": 0.2998615503311157,
      "learning_rate": 8e-05,
      "loss": 1.6355,
      "step": 3581
    },
    {
      "epoch": 0.7878587924777302,
      "grad_norm": 0.28142982721328735,
      "learning_rate": 8e-05,
      "loss": 1.605,
      "step": 3582
    },
    {
      "epoch": 0.7880787418893654,
      "grad_norm": 0.30738621950149536,
      "learning_rate": 8e-05,
      "loss": 1.6741,
      "step": 3583
    },
    {
      "epoch": 0.7882986913010007,
      "grad_norm": 0.3007453382015228,
      "learning_rate": 8e-05,
      "loss": 1.8258,
      "step": 3584
    },
    {
      "epoch": 0.7885186407126361,
      "grad_norm": 0.29813793301582336,
      "learning_rate": 8e-05,
      "loss": 1.6551,
      "step": 3585
    },
    {
      "epoch": 0.7887385901242714,
      "grad_norm": 0.29949578642845154,
      "learning_rate": 8e-05,
      "loss": 1.6033,
      "step": 3586
    },
    {
      "epoch": 0.7889585395359068,
      "grad_norm": 0.2934853136539459,
      "learning_rate": 8e-05,
      "loss": 1.5856,
      "step": 3587
    },
    {
      "epoch": 0.7891784889475421,
      "grad_norm": 0.3007175326347351,
      "learning_rate": 8e-05,
      "loss": 1.5335,
      "step": 3588
    },
    {
      "epoch": 0.7893984383591773,
      "grad_norm": 0.2819620370864868,
      "learning_rate": 8e-05,
      "loss": 1.6833,
      "step": 3589
    },
    {
      "epoch": 0.7896183877708127,
      "grad_norm": 0.27381354570388794,
      "learning_rate": 8e-05,
      "loss": 1.5009,
      "step": 3590
    },
    {
      "epoch": 0.789838337182448,
      "grad_norm": 0.2849341928958893,
      "learning_rate": 8e-05,
      "loss": 1.7086,
      "step": 3591
    },
    {
      "epoch": 0.7900582865940834,
      "grad_norm": 0.3121960759162903,
      "learning_rate": 8e-05,
      "loss": 1.8541,
      "step": 3592
    },
    {
      "epoch": 0.7902782360057187,
      "grad_norm": 0.28709515929222107,
      "learning_rate": 8e-05,
      "loss": 1.6198,
      "step": 3593
    },
    {
      "epoch": 0.7904981854173541,
      "grad_norm": 0.29474472999572754,
      "learning_rate": 8e-05,
      "loss": 1.6322,
      "step": 3594
    },
    {
      "epoch": 0.7907181348289893,
      "grad_norm": 0.2754751443862915,
      "learning_rate": 8e-05,
      "loss": 1.5421,
      "step": 3595
    },
    {
      "epoch": 0.7909380842406246,
      "grad_norm": 0.3164171576499939,
      "learning_rate": 8e-05,
      "loss": 1.8512,
      "step": 3596
    },
    {
      "epoch": 0.79115803365226,
      "grad_norm": 0.2999406158924103,
      "learning_rate": 8e-05,
      "loss": 1.6801,
      "step": 3597
    },
    {
      "epoch": 0.7913779830638953,
      "grad_norm": 0.2944709360599518,
      "learning_rate": 8e-05,
      "loss": 1.6419,
      "step": 3598
    },
    {
      "epoch": 0.7915979324755307,
      "grad_norm": 0.28523892164230347,
      "learning_rate": 8e-05,
      "loss": 1.4772,
      "step": 3599
    },
    {
      "epoch": 0.791817881887166,
      "grad_norm": 0.3085593581199646,
      "learning_rate": 8e-05,
      "loss": 1.643,
      "step": 3600
    },
    {
      "epoch": 0.7920378312988012,
      "grad_norm": 0.2771039605140686,
      "learning_rate": 8e-05,
      "loss": 1.701,
      "step": 3601
    },
    {
      "epoch": 0.7922577807104366,
      "grad_norm": 0.2826726734638214,
      "learning_rate": 8e-05,
      "loss": 1.5694,
      "step": 3602
    },
    {
      "epoch": 0.7924777301220719,
      "grad_norm": 0.28886765241622925,
      "learning_rate": 8e-05,
      "loss": 1.6117,
      "step": 3603
    },
    {
      "epoch": 0.7926976795337073,
      "grad_norm": 0.30170392990112305,
      "learning_rate": 8e-05,
      "loss": 1.7294,
      "step": 3604
    },
    {
      "epoch": 0.7929176289453426,
      "grad_norm": 0.2870231866836548,
      "learning_rate": 8e-05,
      "loss": 1.5244,
      "step": 3605
    },
    {
      "epoch": 0.7931375783569778,
      "grad_norm": 0.28652969002723694,
      "learning_rate": 8e-05,
      "loss": 1.5694,
      "step": 3606
    },
    {
      "epoch": 0.7933575277686132,
      "grad_norm": 0.2963985204696655,
      "learning_rate": 8e-05,
      "loss": 1.7553,
      "step": 3607
    },
    {
      "epoch": 0.7935774771802485,
      "grad_norm": 0.302688866853714,
      "learning_rate": 8e-05,
      "loss": 1.689,
      "step": 3608
    },
    {
      "epoch": 0.7937974265918839,
      "grad_norm": 0.32751408219337463,
      "learning_rate": 8e-05,
      "loss": 1.8021,
      "step": 3609
    },
    {
      "epoch": 0.7940173760035192,
      "grad_norm": 0.3023033142089844,
      "learning_rate": 8e-05,
      "loss": 1.7521,
      "step": 3610
    },
    {
      "epoch": 0.7942373254151545,
      "grad_norm": 0.2758845090866089,
      "learning_rate": 8e-05,
      "loss": 1.5365,
      "step": 3611
    },
    {
      "epoch": 0.7944572748267898,
      "grad_norm": 0.2906113564968109,
      "learning_rate": 8e-05,
      "loss": 1.7001,
      "step": 3612
    },
    {
      "epoch": 0.7946772242384251,
      "grad_norm": 0.32935893535614014,
      "learning_rate": 8e-05,
      "loss": 1.8016,
      "step": 3613
    },
    {
      "epoch": 0.7948971736500605,
      "grad_norm": 0.29360225796699524,
      "learning_rate": 8e-05,
      "loss": 1.5381,
      "step": 3614
    },
    {
      "epoch": 0.7951171230616958,
      "grad_norm": 0.3403272330760956,
      "learning_rate": 8e-05,
      "loss": 1.6639,
      "step": 3615
    },
    {
      "epoch": 0.7953370724733312,
      "grad_norm": 0.27274513244628906,
      "learning_rate": 8e-05,
      "loss": 1.5926,
      "step": 3616
    },
    {
      "epoch": 0.7955570218849665,
      "grad_norm": 0.2820468544960022,
      "learning_rate": 8e-05,
      "loss": 1.5791,
      "step": 3617
    },
    {
      "epoch": 0.7957769712966017,
      "grad_norm": 0.3011898398399353,
      "learning_rate": 8e-05,
      "loss": 1.7195,
      "step": 3618
    },
    {
      "epoch": 0.7959969207082371,
      "grad_norm": 0.2950305938720703,
      "learning_rate": 8e-05,
      "loss": 1.6335,
      "step": 3619
    },
    {
      "epoch": 0.7962168701198724,
      "grad_norm": 0.27936869859695435,
      "learning_rate": 8e-05,
      "loss": 1.5635,
      "step": 3620
    },
    {
      "epoch": 0.7964368195315078,
      "grad_norm": 0.3040826916694641,
      "learning_rate": 8e-05,
      "loss": 1.672,
      "step": 3621
    },
    {
      "epoch": 0.7966567689431431,
      "grad_norm": 0.3088817894458771,
      "learning_rate": 8e-05,
      "loss": 1.7677,
      "step": 3622
    },
    {
      "epoch": 0.7968767183547784,
      "grad_norm": 0.28857895731925964,
      "learning_rate": 8e-05,
      "loss": 1.6923,
      "step": 3623
    },
    {
      "epoch": 0.7970966677664137,
      "grad_norm": 0.3066072165966034,
      "learning_rate": 8e-05,
      "loss": 1.5091,
      "step": 3624
    },
    {
      "epoch": 0.797316617178049,
      "grad_norm": 0.3056313991546631,
      "learning_rate": 8e-05,
      "loss": 1.8183,
      "step": 3625
    },
    {
      "epoch": 0.7975365665896844,
      "grad_norm": 0.27834954857826233,
      "learning_rate": 8e-05,
      "loss": 1.6914,
      "step": 3626
    },
    {
      "epoch": 0.7977565160013197,
      "grad_norm": 0.2966832220554352,
      "learning_rate": 8e-05,
      "loss": 1.6422,
      "step": 3627
    },
    {
      "epoch": 0.797976465412955,
      "grad_norm": 0.31216248869895935,
      "learning_rate": 8e-05,
      "loss": 1.7629,
      "step": 3628
    },
    {
      "epoch": 0.7981964148245904,
      "grad_norm": 0.3048402965068817,
      "learning_rate": 8e-05,
      "loss": 1.7893,
      "step": 3629
    },
    {
      "epoch": 0.7984163642362256,
      "grad_norm": 0.3240011930465698,
      "learning_rate": 8e-05,
      "loss": 1.7933,
      "step": 3630
    },
    {
      "epoch": 0.798636313647861,
      "grad_norm": 0.3020577132701874,
      "learning_rate": 8e-05,
      "loss": 1.6685,
      "step": 3631
    },
    {
      "epoch": 0.7988562630594963,
      "grad_norm": 0.3189142644405365,
      "learning_rate": 8e-05,
      "loss": 1.7346,
      "step": 3632
    },
    {
      "epoch": 0.7990762124711316,
      "grad_norm": 0.2925734519958496,
      "learning_rate": 8e-05,
      "loss": 1.6465,
      "step": 3633
    },
    {
      "epoch": 0.799296161882767,
      "grad_norm": 0.32868942618370056,
      "learning_rate": 8e-05,
      "loss": 1.6621,
      "step": 3634
    },
    {
      "epoch": 0.7995161112944023,
      "grad_norm": 0.28033214807510376,
      "learning_rate": 8e-05,
      "loss": 1.6006,
      "step": 3635
    },
    {
      "epoch": 0.7997360607060376,
      "grad_norm": 0.28859859704971313,
      "learning_rate": 8e-05,
      "loss": 1.7644,
      "step": 3636
    },
    {
      "epoch": 0.7999560101176729,
      "grad_norm": 0.2980223000049591,
      "learning_rate": 8e-05,
      "loss": 1.699,
      "step": 3637
    },
    {
      "epoch": 0.8001759595293083,
      "grad_norm": 0.288064181804657,
      "learning_rate": 8e-05,
      "loss": 1.6435,
      "step": 3638
    },
    {
      "epoch": 0.8003959089409436,
      "grad_norm": 0.29023095965385437,
      "learning_rate": 8e-05,
      "loss": 1.8842,
      "step": 3639
    },
    {
      "epoch": 0.8006158583525789,
      "grad_norm": 0.2843625247478485,
      "learning_rate": 8e-05,
      "loss": 1.6155,
      "step": 3640
    },
    {
      "epoch": 0.8008358077642143,
      "grad_norm": 0.292680948972702,
      "learning_rate": 8e-05,
      "loss": 1.7985,
      "step": 3641
    },
    {
      "epoch": 0.8010557571758495,
      "grad_norm": 0.2843727171421051,
      "learning_rate": 8e-05,
      "loss": 1.8262,
      "step": 3642
    },
    {
      "epoch": 0.8012757065874849,
      "grad_norm": 0.3109283745288849,
      "learning_rate": 8e-05,
      "loss": 1.719,
      "step": 3643
    },
    {
      "epoch": 0.8014956559991202,
      "grad_norm": 0.2911458909511566,
      "learning_rate": 8e-05,
      "loss": 1.6543,
      "step": 3644
    },
    {
      "epoch": 0.8017156054107555,
      "grad_norm": 0.29368388652801514,
      "learning_rate": 8e-05,
      "loss": 1.6606,
      "step": 3645
    },
    {
      "epoch": 0.8019355548223909,
      "grad_norm": 0.27842018008232117,
      "learning_rate": 8e-05,
      "loss": 1.6225,
      "step": 3646
    },
    {
      "epoch": 0.8021555042340262,
      "grad_norm": 0.2872051000595093,
      "learning_rate": 8e-05,
      "loss": 1.5608,
      "step": 3647
    },
    {
      "epoch": 0.8023754536456615,
      "grad_norm": 0.2749423086643219,
      "learning_rate": 8e-05,
      "loss": 1.579,
      "step": 3648
    },
    {
      "epoch": 0.8025954030572968,
      "grad_norm": 0.28996387124061584,
      "learning_rate": 8e-05,
      "loss": 1.6742,
      "step": 3649
    },
    {
      "epoch": 0.8028153524689321,
      "grad_norm": 0.3018381595611572,
      "learning_rate": 8e-05,
      "loss": 1.7604,
      "step": 3650
    },
    {
      "epoch": 0.8030353018805675,
      "grad_norm": 0.29183822870254517,
      "learning_rate": 8e-05,
      "loss": 1.7273,
      "step": 3651
    },
    {
      "epoch": 0.8032552512922028,
      "grad_norm": 0.2697616517543793,
      "learning_rate": 8e-05,
      "loss": 1.4865,
      "step": 3652
    },
    {
      "epoch": 0.8034752007038382,
      "grad_norm": 0.29433682560920715,
      "learning_rate": 8e-05,
      "loss": 1.7453,
      "step": 3653
    },
    {
      "epoch": 0.8036951501154734,
      "grad_norm": 0.3100878596305847,
      "learning_rate": 8e-05,
      "loss": 1.6805,
      "step": 3654
    },
    {
      "epoch": 0.8039150995271087,
      "grad_norm": 0.3025181293487549,
      "learning_rate": 8e-05,
      "loss": 1.6999,
      "step": 3655
    },
    {
      "epoch": 0.8041350489387441,
      "grad_norm": 0.29453882575035095,
      "learning_rate": 8e-05,
      "loss": 1.5694,
      "step": 3656
    },
    {
      "epoch": 0.8043549983503794,
      "grad_norm": 0.30523207783699036,
      "learning_rate": 8e-05,
      "loss": 1.6819,
      "step": 3657
    },
    {
      "epoch": 0.8045749477620148,
      "grad_norm": 0.28335437178611755,
      "learning_rate": 8e-05,
      "loss": 1.7137,
      "step": 3658
    },
    {
      "epoch": 0.80479489717365,
      "grad_norm": 0.30658838152885437,
      "learning_rate": 8e-05,
      "loss": 1.6486,
      "step": 3659
    },
    {
      "epoch": 0.8050148465852854,
      "grad_norm": 0.30316153168678284,
      "learning_rate": 8e-05,
      "loss": 1.7012,
      "step": 3660
    },
    {
      "epoch": 0.8052347959969207,
      "grad_norm": 0.2958810329437256,
      "learning_rate": 8e-05,
      "loss": 1.5989,
      "step": 3661
    },
    {
      "epoch": 0.805454745408556,
      "grad_norm": 0.2858383059501648,
      "learning_rate": 8e-05,
      "loss": 1.6466,
      "step": 3662
    },
    {
      "epoch": 0.8056746948201914,
      "grad_norm": 0.3087503910064697,
      "learning_rate": 8e-05,
      "loss": 1.7894,
      "step": 3663
    },
    {
      "epoch": 0.8058946442318267,
      "grad_norm": 0.29160773754119873,
      "learning_rate": 8e-05,
      "loss": 1.5268,
      "step": 3664
    },
    {
      "epoch": 0.806114593643462,
      "grad_norm": 0.3402816653251648,
      "learning_rate": 8e-05,
      "loss": 1.728,
      "step": 3665
    },
    {
      "epoch": 0.8063345430550973,
      "grad_norm": 0.2856731712818146,
      "learning_rate": 8e-05,
      "loss": 1.5064,
      "step": 3666
    },
    {
      "epoch": 0.8065544924667326,
      "grad_norm": 0.29663363099098206,
      "learning_rate": 8e-05,
      "loss": 1.7799,
      "step": 3667
    },
    {
      "epoch": 0.806774441878368,
      "grad_norm": 0.2893684208393097,
      "learning_rate": 8e-05,
      "loss": 1.6961,
      "step": 3668
    },
    {
      "epoch": 0.8069943912900033,
      "grad_norm": 0.2832998037338257,
      "learning_rate": 8e-05,
      "loss": 1.514,
      "step": 3669
    },
    {
      "epoch": 0.8072143407016387,
      "grad_norm": 0.3098543584346771,
      "learning_rate": 8e-05,
      "loss": 1.7395,
      "step": 3670
    },
    {
      "epoch": 0.807434290113274,
      "grad_norm": 0.30097678303718567,
      "learning_rate": 8e-05,
      "loss": 1.6402,
      "step": 3671
    },
    {
      "epoch": 0.8076542395249092,
      "grad_norm": 0.3011092245578766,
      "learning_rate": 8e-05,
      "loss": 1.6554,
      "step": 3672
    },
    {
      "epoch": 0.8078741889365446,
      "grad_norm": 0.3125768303871155,
      "learning_rate": 8e-05,
      "loss": 1.6503,
      "step": 3673
    },
    {
      "epoch": 0.8080941383481799,
      "grad_norm": 0.3440975546836853,
      "learning_rate": 8e-05,
      "loss": 1.6666,
      "step": 3674
    },
    {
      "epoch": 0.8083140877598153,
      "grad_norm": 0.3339664340019226,
      "learning_rate": 8e-05,
      "loss": 1.7333,
      "step": 3675
    },
    {
      "epoch": 0.8085340371714506,
      "grad_norm": 0.3038443922996521,
      "learning_rate": 8e-05,
      "loss": 1.6352,
      "step": 3676
    },
    {
      "epoch": 0.8087539865830858,
      "grad_norm": 0.2839770019054413,
      "learning_rate": 8e-05,
      "loss": 1.6714,
      "step": 3677
    },
    {
      "epoch": 0.8089739359947212,
      "grad_norm": 0.28624406456947327,
      "learning_rate": 8e-05,
      "loss": 1.5826,
      "step": 3678
    },
    {
      "epoch": 0.8091938854063565,
      "grad_norm": 0.28209447860717773,
      "learning_rate": 8e-05,
      "loss": 1.5905,
      "step": 3679
    },
    {
      "epoch": 0.8094138348179919,
      "grad_norm": 0.3045743703842163,
      "learning_rate": 8e-05,
      "loss": 1.5201,
      "step": 3680
    },
    {
      "epoch": 0.8096337842296272,
      "grad_norm": 0.2729533910751343,
      "learning_rate": 8e-05,
      "loss": 1.5097,
      "step": 3681
    },
    {
      "epoch": 0.8098537336412626,
      "grad_norm": 0.3302726447582245,
      "learning_rate": 8e-05,
      "loss": 1.9036,
      "step": 3682
    },
    {
      "epoch": 0.8100736830528978,
      "grad_norm": 0.3054613471031189,
      "learning_rate": 8e-05,
      "loss": 1.6474,
      "step": 3683
    },
    {
      "epoch": 0.8102936324645331,
      "grad_norm": 0.306883305311203,
      "learning_rate": 8e-05,
      "loss": 1.8867,
      "step": 3684
    },
    {
      "epoch": 0.8105135818761685,
      "grad_norm": 0.2889968156814575,
      "learning_rate": 8e-05,
      "loss": 1.7218,
      "step": 3685
    },
    {
      "epoch": 0.8107335312878038,
      "grad_norm": 0.2770795226097107,
      "learning_rate": 8e-05,
      "loss": 1.5795,
      "step": 3686
    },
    {
      "epoch": 0.8109534806994392,
      "grad_norm": 0.29530447721481323,
      "learning_rate": 8e-05,
      "loss": 1.681,
      "step": 3687
    },
    {
      "epoch": 0.8111734301110745,
      "grad_norm": 0.28968799114227295,
      "learning_rate": 8e-05,
      "loss": 1.6409,
      "step": 3688
    },
    {
      "epoch": 0.8113933795227097,
      "grad_norm": 0.2868937849998474,
      "learning_rate": 8e-05,
      "loss": 1.6418,
      "step": 3689
    },
    {
      "epoch": 0.8116133289343451,
      "grad_norm": 0.29399046301841736,
      "learning_rate": 8e-05,
      "loss": 1.6872,
      "step": 3690
    },
    {
      "epoch": 0.8118332783459804,
      "grad_norm": 0.2806103229522705,
      "learning_rate": 8e-05,
      "loss": 1.5778,
      "step": 3691
    },
    {
      "epoch": 0.8120532277576158,
      "grad_norm": 0.3034190535545349,
      "learning_rate": 8e-05,
      "loss": 1.605,
      "step": 3692
    },
    {
      "epoch": 0.8122731771692511,
      "grad_norm": 0.29207345843315125,
      "learning_rate": 8e-05,
      "loss": 1.7604,
      "step": 3693
    },
    {
      "epoch": 0.8124931265808863,
      "grad_norm": 0.28544682264328003,
      "learning_rate": 8e-05,
      "loss": 1.5085,
      "step": 3694
    },
    {
      "epoch": 0.8127130759925217,
      "grad_norm": 0.30136585235595703,
      "learning_rate": 8e-05,
      "loss": 1.7208,
      "step": 3695
    },
    {
      "epoch": 0.812933025404157,
      "grad_norm": 0.29252931475639343,
      "learning_rate": 8e-05,
      "loss": 1.5529,
      "step": 3696
    },
    {
      "epoch": 0.8131529748157924,
      "grad_norm": 0.3112747371196747,
      "learning_rate": 8e-05,
      "loss": 1.6788,
      "step": 3697
    },
    {
      "epoch": 0.8133729242274277,
      "grad_norm": 0.28653424978256226,
      "learning_rate": 8e-05,
      "loss": 1.6755,
      "step": 3698
    },
    {
      "epoch": 0.813592873639063,
      "grad_norm": 0.2881905436515808,
      "learning_rate": 8e-05,
      "loss": 1.598,
      "step": 3699
    },
    {
      "epoch": 0.8138128230506984,
      "grad_norm": 0.3594631850719452,
      "learning_rate": 8e-05,
      "loss": 1.7263,
      "step": 3700
    },
    {
      "epoch": 0.8140327724623336,
      "grad_norm": 0.31845301389694214,
      "learning_rate": 8e-05,
      "loss": 1.7378,
      "step": 3701
    },
    {
      "epoch": 0.814252721873969,
      "grad_norm": 0.29628971219062805,
      "learning_rate": 8e-05,
      "loss": 1.6484,
      "step": 3702
    },
    {
      "epoch": 0.8144726712856043,
      "grad_norm": 0.2896782159805298,
      "learning_rate": 8e-05,
      "loss": 1.6006,
      "step": 3703
    },
    {
      "epoch": 0.8146926206972397,
      "grad_norm": 0.32776281237602234,
      "learning_rate": 8e-05,
      "loss": 1.8014,
      "step": 3704
    },
    {
      "epoch": 0.814912570108875,
      "grad_norm": 0.29571112990379333,
      "learning_rate": 8e-05,
      "loss": 1.7807,
      "step": 3705
    },
    {
      "epoch": 0.8151325195205102,
      "grad_norm": 0.2930830717086792,
      "learning_rate": 8e-05,
      "loss": 1.6407,
      "step": 3706
    },
    {
      "epoch": 0.8153524689321456,
      "grad_norm": 0.30788877606391907,
      "learning_rate": 8e-05,
      "loss": 1.8575,
      "step": 3707
    },
    {
      "epoch": 0.8155724183437809,
      "grad_norm": 0.2772351801395416,
      "learning_rate": 8e-05,
      "loss": 1.7,
      "step": 3708
    },
    {
      "epoch": 0.8157923677554163,
      "grad_norm": 0.2957149147987366,
      "learning_rate": 8e-05,
      "loss": 1.6791,
      "step": 3709
    },
    {
      "epoch": 0.8160123171670516,
      "grad_norm": 0.29258304834365845,
      "learning_rate": 8e-05,
      "loss": 1.6694,
      "step": 3710
    },
    {
      "epoch": 0.8162322665786869,
      "grad_norm": 0.3325083255767822,
      "learning_rate": 8e-05,
      "loss": 1.649,
      "step": 3711
    },
    {
      "epoch": 0.8164522159903222,
      "grad_norm": 0.2934175431728363,
      "learning_rate": 8e-05,
      "loss": 1.5983,
      "step": 3712
    },
    {
      "epoch": 0.8166721654019575,
      "grad_norm": 0.2800202965736389,
      "learning_rate": 8e-05,
      "loss": 1.6778,
      "step": 3713
    },
    {
      "epoch": 0.8168921148135929,
      "grad_norm": 0.28423386812210083,
      "learning_rate": 8e-05,
      "loss": 1.4981,
      "step": 3714
    },
    {
      "epoch": 0.8171120642252282,
      "grad_norm": 0.3425268530845642,
      "learning_rate": 8e-05,
      "loss": 1.7839,
      "step": 3715
    },
    {
      "epoch": 0.8173320136368635,
      "grad_norm": 0.3047974705696106,
      "learning_rate": 8e-05,
      "loss": 1.6118,
      "step": 3716
    },
    {
      "epoch": 0.8175519630484989,
      "grad_norm": 0.2974521815776825,
      "learning_rate": 8e-05,
      "loss": 1.6564,
      "step": 3717
    },
    {
      "epoch": 0.8177719124601341,
      "grad_norm": 0.27390167117118835,
      "learning_rate": 8e-05,
      "loss": 1.5317,
      "step": 3718
    },
    {
      "epoch": 0.8179918618717695,
      "grad_norm": 0.3012712597846985,
      "learning_rate": 8e-05,
      "loss": 1.5497,
      "step": 3719
    },
    {
      "epoch": 0.8182118112834048,
      "grad_norm": 0.2798818349838257,
      "learning_rate": 8e-05,
      "loss": 1.6036,
      "step": 3720
    },
    {
      "epoch": 0.8184317606950401,
      "grad_norm": 0.2981497049331665,
      "learning_rate": 8e-05,
      "loss": 1.7168,
      "step": 3721
    },
    {
      "epoch": 0.8186517101066755,
      "grad_norm": 0.30722394585609436,
      "learning_rate": 8e-05,
      "loss": 1.5669,
      "step": 3722
    },
    {
      "epoch": 0.8188716595183108,
      "grad_norm": 0.317390114068985,
      "learning_rate": 8e-05,
      "loss": 1.7005,
      "step": 3723
    },
    {
      "epoch": 0.8190916089299461,
      "grad_norm": 0.306482195854187,
      "learning_rate": 8e-05,
      "loss": 1.7386,
      "step": 3724
    },
    {
      "epoch": 0.8193115583415814,
      "grad_norm": 0.2971460521221161,
      "learning_rate": 8e-05,
      "loss": 1.6786,
      "step": 3725
    },
    {
      "epoch": 0.8195315077532168,
      "grad_norm": 0.30549904704093933,
      "learning_rate": 8e-05,
      "loss": 1.7212,
      "step": 3726
    },
    {
      "epoch": 0.8197514571648521,
      "grad_norm": 0.2948817312717438,
      "learning_rate": 8e-05,
      "loss": 1.5523,
      "step": 3727
    },
    {
      "epoch": 0.8199714065764874,
      "grad_norm": 0.29448559880256653,
      "learning_rate": 8e-05,
      "loss": 1.6445,
      "step": 3728
    },
    {
      "epoch": 0.8201913559881228,
      "grad_norm": 0.31158149242401123,
      "learning_rate": 8e-05,
      "loss": 1.7997,
      "step": 3729
    },
    {
      "epoch": 0.820411305399758,
      "grad_norm": 0.29270637035369873,
      "learning_rate": 8e-05,
      "loss": 1.6418,
      "step": 3730
    },
    {
      "epoch": 0.8206312548113934,
      "grad_norm": 0.29338178038597107,
      "learning_rate": 8e-05,
      "loss": 1.6839,
      "step": 3731
    },
    {
      "epoch": 0.8208512042230287,
      "grad_norm": 0.2772715389728546,
      "learning_rate": 8e-05,
      "loss": 1.6326,
      "step": 3732
    },
    {
      "epoch": 0.821071153634664,
      "grad_norm": 0.2976597547531128,
      "learning_rate": 8e-05,
      "loss": 1.6613,
      "step": 3733
    },
    {
      "epoch": 0.8212911030462994,
      "grad_norm": 0.3194999396800995,
      "learning_rate": 8e-05,
      "loss": 1.665,
      "step": 3734
    },
    {
      "epoch": 0.8215110524579347,
      "grad_norm": 0.28494882583618164,
      "learning_rate": 8e-05,
      "loss": 1.5707,
      "step": 3735
    },
    {
      "epoch": 0.82173100186957,
      "grad_norm": 0.29339084029197693,
      "learning_rate": 8e-05,
      "loss": 1.6276,
      "step": 3736
    },
    {
      "epoch": 0.8219509512812053,
      "grad_norm": 0.2872979938983917,
      "learning_rate": 8e-05,
      "loss": 1.5653,
      "step": 3737
    },
    {
      "epoch": 0.8221709006928406,
      "grad_norm": 0.2818096876144409,
      "learning_rate": 8e-05,
      "loss": 1.5631,
      "step": 3738
    },
    {
      "epoch": 0.822390850104476,
      "grad_norm": 0.2971639633178711,
      "learning_rate": 8e-05,
      "loss": 1.6698,
      "step": 3739
    },
    {
      "epoch": 0.8226107995161113,
      "grad_norm": 0.27923282980918884,
      "learning_rate": 8e-05,
      "loss": 1.6127,
      "step": 3740
    },
    {
      "epoch": 0.8228307489277467,
      "grad_norm": 0.2899787425994873,
      "learning_rate": 8e-05,
      "loss": 1.6323,
      "step": 3741
    },
    {
      "epoch": 0.8230506983393819,
      "grad_norm": 0.2971678376197815,
      "learning_rate": 8e-05,
      "loss": 1.6495,
      "step": 3742
    },
    {
      "epoch": 0.8232706477510172,
      "grad_norm": 0.3197123408317566,
      "learning_rate": 8e-05,
      "loss": 1.6807,
      "step": 3743
    },
    {
      "epoch": 0.8234905971626526,
      "grad_norm": 0.2743538022041321,
      "learning_rate": 8e-05,
      "loss": 1.659,
      "step": 3744
    },
    {
      "epoch": 0.8237105465742879,
      "grad_norm": 0.29337090253829956,
      "learning_rate": 8e-05,
      "loss": 1.7992,
      "step": 3745
    },
    {
      "epoch": 0.8239304959859233,
      "grad_norm": 0.295331209897995,
      "learning_rate": 8e-05,
      "loss": 1.5459,
      "step": 3746
    },
    {
      "epoch": 0.8241504453975586,
      "grad_norm": 0.2844454050064087,
      "learning_rate": 8e-05,
      "loss": 1.7079,
      "step": 3747
    },
    {
      "epoch": 0.8243703948091939,
      "grad_norm": 0.32962703704833984,
      "learning_rate": 8e-05,
      "loss": 1.8193,
      "step": 3748
    },
    {
      "epoch": 0.8245903442208292,
      "grad_norm": 0.2843455374240875,
      "learning_rate": 8e-05,
      "loss": 1.6964,
      "step": 3749
    },
    {
      "epoch": 0.8248102936324645,
      "grad_norm": 0.28151237964630127,
      "learning_rate": 8e-05,
      "loss": 1.7152,
      "step": 3750
    },
    {
      "epoch": 0.8250302430440999,
      "grad_norm": 0.30013763904571533,
      "learning_rate": 8e-05,
      "loss": 1.7262,
      "step": 3751
    },
    {
      "epoch": 0.8252501924557352,
      "grad_norm": 0.2816784083843231,
      "learning_rate": 8e-05,
      "loss": 1.7431,
      "step": 3752
    },
    {
      "epoch": 0.8254701418673706,
      "grad_norm": 0.32148584723472595,
      "learning_rate": 8e-05,
      "loss": 1.7909,
      "step": 3753
    },
    {
      "epoch": 0.8256900912790058,
      "grad_norm": 0.2902558147907257,
      "learning_rate": 8e-05,
      "loss": 1.5782,
      "step": 3754
    },
    {
      "epoch": 0.8259100406906411,
      "grad_norm": 0.3017018735408783,
      "learning_rate": 8e-05,
      "loss": 1.7332,
      "step": 3755
    },
    {
      "epoch": 0.8261299901022765,
      "grad_norm": 0.29928895831108093,
      "learning_rate": 8e-05,
      "loss": 1.6972,
      "step": 3756
    },
    {
      "epoch": 0.8263499395139118,
      "grad_norm": 0.2922762632369995,
      "learning_rate": 8e-05,
      "loss": 1.5663,
      "step": 3757
    },
    {
      "epoch": 0.8265698889255472,
      "grad_norm": 0.2903006672859192,
      "learning_rate": 8e-05,
      "loss": 1.6253,
      "step": 3758
    },
    {
      "epoch": 0.8267898383371824,
      "grad_norm": 0.2981746196746826,
      "learning_rate": 8e-05,
      "loss": 1.69,
      "step": 3759
    },
    {
      "epoch": 0.8270097877488177,
      "grad_norm": 0.2905546724796295,
      "learning_rate": 8e-05,
      "loss": 1.5513,
      "step": 3760
    },
    {
      "epoch": 0.8272297371604531,
      "grad_norm": 0.30684736371040344,
      "learning_rate": 8e-05,
      "loss": 1.7188,
      "step": 3761
    },
    {
      "epoch": 0.8274496865720884,
      "grad_norm": 0.29893192648887634,
      "learning_rate": 8e-05,
      "loss": 1.7759,
      "step": 3762
    },
    {
      "epoch": 0.8276696359837238,
      "grad_norm": 0.2762022316455841,
      "learning_rate": 8e-05,
      "loss": 1.5835,
      "step": 3763
    },
    {
      "epoch": 0.8278895853953591,
      "grad_norm": 0.3005400002002716,
      "learning_rate": 8e-05,
      "loss": 1.7313,
      "step": 3764
    },
    {
      "epoch": 0.8281095348069943,
      "grad_norm": 0.33047348260879517,
      "learning_rate": 8e-05,
      "loss": 1.7504,
      "step": 3765
    },
    {
      "epoch": 0.8283294842186297,
      "grad_norm": 0.29011067748069763,
      "learning_rate": 8e-05,
      "loss": 1.677,
      "step": 3766
    },
    {
      "epoch": 0.828549433630265,
      "grad_norm": 0.28470665216445923,
      "learning_rate": 8e-05,
      "loss": 1.5375,
      "step": 3767
    },
    {
      "epoch": 0.8287693830419004,
      "grad_norm": 0.47327253222465515,
      "learning_rate": 8e-05,
      "loss": 1.9681,
      "step": 3768
    },
    {
      "epoch": 0.8289893324535357,
      "grad_norm": 0.3004605174064636,
      "learning_rate": 8e-05,
      "loss": 1.6489,
      "step": 3769
    },
    {
      "epoch": 0.8292092818651711,
      "grad_norm": 0.29962432384490967,
      "learning_rate": 8e-05,
      "loss": 1.5485,
      "step": 3770
    },
    {
      "epoch": 0.8294292312768063,
      "grad_norm": 0.3226446509361267,
      "learning_rate": 8e-05,
      "loss": 1.5853,
      "step": 3771
    },
    {
      "epoch": 0.8296491806884416,
      "grad_norm": 0.2956733703613281,
      "learning_rate": 8e-05,
      "loss": 1.6783,
      "step": 3772
    },
    {
      "epoch": 0.829869130100077,
      "grad_norm": 0.30481475591659546,
      "learning_rate": 8e-05,
      "loss": 1.7441,
      "step": 3773
    },
    {
      "epoch": 0.8300890795117123,
      "grad_norm": 0.29181525111198425,
      "learning_rate": 8e-05,
      "loss": 1.7026,
      "step": 3774
    },
    {
      "epoch": 0.8303090289233477,
      "grad_norm": 0.28580138087272644,
      "learning_rate": 8e-05,
      "loss": 1.6131,
      "step": 3775
    },
    {
      "epoch": 0.830528978334983,
      "grad_norm": 0.28029665350914,
      "learning_rate": 8e-05,
      "loss": 1.618,
      "step": 3776
    },
    {
      "epoch": 0.8307489277466182,
      "grad_norm": 0.31312623620033264,
      "learning_rate": 8e-05,
      "loss": 1.629,
      "step": 3777
    },
    {
      "epoch": 0.8309688771582536,
      "grad_norm": 0.31540897488594055,
      "learning_rate": 8e-05,
      "loss": 1.8215,
      "step": 3778
    },
    {
      "epoch": 0.8311888265698889,
      "grad_norm": 0.3031502664089203,
      "learning_rate": 8e-05,
      "loss": 1.6722,
      "step": 3779
    },
    {
      "epoch": 0.8314087759815243,
      "grad_norm": 0.31536054611206055,
      "learning_rate": 8e-05,
      "loss": 1.753,
      "step": 3780
    },
    {
      "epoch": 0.8316287253931596,
      "grad_norm": 0.31534263491630554,
      "learning_rate": 8e-05,
      "loss": 1.6841,
      "step": 3781
    },
    {
      "epoch": 0.8318486748047949,
      "grad_norm": 0.29810741543769836,
      "learning_rate": 8e-05,
      "loss": 1.8143,
      "step": 3782
    },
    {
      "epoch": 0.8320686242164302,
      "grad_norm": 0.2857937812805176,
      "learning_rate": 8e-05,
      "loss": 1.6004,
      "step": 3783
    },
    {
      "epoch": 0.8322885736280655,
      "grad_norm": 0.2927904427051544,
      "learning_rate": 8e-05,
      "loss": 1.6385,
      "step": 3784
    },
    {
      "epoch": 0.8325085230397009,
      "grad_norm": 0.29359960556030273,
      "learning_rate": 8e-05,
      "loss": 1.762,
      "step": 3785
    },
    {
      "epoch": 0.8327284724513362,
      "grad_norm": 0.2871841788291931,
      "learning_rate": 8e-05,
      "loss": 1.6141,
      "step": 3786
    },
    {
      "epoch": 0.8329484218629715,
      "grad_norm": 0.32198578119277954,
      "learning_rate": 8e-05,
      "loss": 1.8785,
      "step": 3787
    },
    {
      "epoch": 0.8331683712746069,
      "grad_norm": 0.2883181571960449,
      "learning_rate": 8e-05,
      "loss": 1.6136,
      "step": 3788
    },
    {
      "epoch": 0.8333883206862421,
      "grad_norm": 0.2934539020061493,
      "learning_rate": 8e-05,
      "loss": 1.6878,
      "step": 3789
    },
    {
      "epoch": 0.8336082700978775,
      "grad_norm": 0.29860734939575195,
      "learning_rate": 8e-05,
      "loss": 1.5412,
      "step": 3790
    },
    {
      "epoch": 0.8338282195095128,
      "grad_norm": 0.32730698585510254,
      "learning_rate": 8e-05,
      "loss": 1.8215,
      "step": 3791
    },
    {
      "epoch": 0.8340481689211481,
      "grad_norm": 0.2875627875328064,
      "learning_rate": 8e-05,
      "loss": 1.571,
      "step": 3792
    },
    {
      "epoch": 0.8342681183327835,
      "grad_norm": 0.33305448293685913,
      "learning_rate": 8e-05,
      "loss": 1.8454,
      "step": 3793
    },
    {
      "epoch": 0.8344880677444187,
      "grad_norm": 0.2842869162559509,
      "learning_rate": 8e-05,
      "loss": 1.6616,
      "step": 3794
    },
    {
      "epoch": 0.8347080171560541,
      "grad_norm": 0.30781999230384827,
      "learning_rate": 8e-05,
      "loss": 1.622,
      "step": 3795
    },
    {
      "epoch": 0.8349279665676894,
      "grad_norm": 0.2892124056816101,
      "learning_rate": 8e-05,
      "loss": 1.5731,
      "step": 3796
    },
    {
      "epoch": 0.8351479159793248,
      "grad_norm": 0.2950188219547272,
      "learning_rate": 8e-05,
      "loss": 1.6374,
      "step": 3797
    },
    {
      "epoch": 0.8353678653909601,
      "grad_norm": 0.283393919467926,
      "learning_rate": 8e-05,
      "loss": 1.5603,
      "step": 3798
    },
    {
      "epoch": 0.8355878148025954,
      "grad_norm": 0.28449615836143494,
      "learning_rate": 8e-05,
      "loss": 1.5694,
      "step": 3799
    },
    {
      "epoch": 0.8358077642142308,
      "grad_norm": 0.27379027009010315,
      "learning_rate": 8e-05,
      "loss": 1.4331,
      "step": 3800
    },
    {
      "epoch": 0.836027713625866,
      "grad_norm": 0.31976261734962463,
      "learning_rate": 8e-05,
      "loss": 1.7362,
      "step": 3801
    },
    {
      "epoch": 0.8362476630375014,
      "grad_norm": 0.27755287289619446,
      "learning_rate": 8e-05,
      "loss": 1.577,
      "step": 3802
    },
    {
      "epoch": 0.8364676124491367,
      "grad_norm": 0.2974432110786438,
      "learning_rate": 8e-05,
      "loss": 1.7651,
      "step": 3803
    },
    {
      "epoch": 0.836687561860772,
      "grad_norm": 0.320019006729126,
      "learning_rate": 8e-05,
      "loss": 1.8053,
      "step": 3804
    },
    {
      "epoch": 0.8369075112724074,
      "grad_norm": 0.29677972197532654,
      "learning_rate": 8e-05,
      "loss": 1.7302,
      "step": 3805
    },
    {
      "epoch": 0.8371274606840426,
      "grad_norm": 0.28835824131965637,
      "learning_rate": 8e-05,
      "loss": 1.6395,
      "step": 3806
    },
    {
      "epoch": 0.837347410095678,
      "grad_norm": 0.29761141538619995,
      "learning_rate": 8e-05,
      "loss": 1.5725,
      "step": 3807
    },
    {
      "epoch": 0.8375673595073133,
      "grad_norm": 0.30719003081321716,
      "learning_rate": 8e-05,
      "loss": 1.6291,
      "step": 3808
    },
    {
      "epoch": 0.8377873089189486,
      "grad_norm": 0.3065665662288666,
      "learning_rate": 8e-05,
      "loss": 1.6901,
      "step": 3809
    },
    {
      "epoch": 0.838007258330584,
      "grad_norm": 0.3014419376850128,
      "learning_rate": 8e-05,
      "loss": 1.7154,
      "step": 3810
    },
    {
      "epoch": 0.8382272077422193,
      "grad_norm": 0.29619207978248596,
      "learning_rate": 8e-05,
      "loss": 1.6578,
      "step": 3811
    },
    {
      "epoch": 0.8384471571538546,
      "grad_norm": 0.2940784692764282,
      "learning_rate": 8e-05,
      "loss": 1.7499,
      "step": 3812
    },
    {
      "epoch": 0.8386671065654899,
      "grad_norm": 0.29198816418647766,
      "learning_rate": 8e-05,
      "loss": 1.6565,
      "step": 3813
    },
    {
      "epoch": 0.8388870559771252,
      "grad_norm": 0.28611230850219727,
      "learning_rate": 8e-05,
      "loss": 1.7194,
      "step": 3814
    },
    {
      "epoch": 0.8391070053887606,
      "grad_norm": 0.2848166227340698,
      "learning_rate": 8e-05,
      "loss": 1.6492,
      "step": 3815
    },
    {
      "epoch": 0.8393269548003959,
      "grad_norm": 0.31025707721710205,
      "learning_rate": 8e-05,
      "loss": 1.7402,
      "step": 3816
    },
    {
      "epoch": 0.8395469042120313,
      "grad_norm": 0.2825357913970947,
      "learning_rate": 8e-05,
      "loss": 1.6927,
      "step": 3817
    },
    {
      "epoch": 0.8397668536236665,
      "grad_norm": 0.27436748147010803,
      "learning_rate": 8e-05,
      "loss": 1.551,
      "step": 3818
    },
    {
      "epoch": 0.8399868030353019,
      "grad_norm": 0.26923510432243347,
      "learning_rate": 8e-05,
      "loss": 1.5776,
      "step": 3819
    },
    {
      "epoch": 0.8402067524469372,
      "grad_norm": 0.29636648297309875,
      "learning_rate": 8e-05,
      "loss": 1.5779,
      "step": 3820
    },
    {
      "epoch": 0.8404267018585725,
      "grad_norm": 0.2913194000720978,
      "learning_rate": 8e-05,
      "loss": 1.6921,
      "step": 3821
    },
    {
      "epoch": 0.8406466512702079,
      "grad_norm": 0.28964900970458984,
      "learning_rate": 8e-05,
      "loss": 1.6866,
      "step": 3822
    },
    {
      "epoch": 0.8408666006818432,
      "grad_norm": 0.2947314381599426,
      "learning_rate": 8e-05,
      "loss": 1.7541,
      "step": 3823
    },
    {
      "epoch": 0.8410865500934785,
      "grad_norm": 0.2964346706867218,
      "learning_rate": 8e-05,
      "loss": 1.7498,
      "step": 3824
    },
    {
      "epoch": 0.8413064995051138,
      "grad_norm": 0.2943279445171356,
      "learning_rate": 8e-05,
      "loss": 1.6294,
      "step": 3825
    },
    {
      "epoch": 0.8415264489167491,
      "grad_norm": 0.305585116147995,
      "learning_rate": 8e-05,
      "loss": 1.6573,
      "step": 3826
    },
    {
      "epoch": 0.8417463983283845,
      "grad_norm": 0.2842620313167572,
      "learning_rate": 8e-05,
      "loss": 1.6747,
      "step": 3827
    },
    {
      "epoch": 0.8419663477400198,
      "grad_norm": 0.29704785346984863,
      "learning_rate": 8e-05,
      "loss": 1.7986,
      "step": 3828
    },
    {
      "epoch": 0.8421862971516552,
      "grad_norm": 0.28527146577835083,
      "learning_rate": 8e-05,
      "loss": 1.722,
      "step": 3829
    },
    {
      "epoch": 0.8424062465632904,
      "grad_norm": 0.2996181547641754,
      "learning_rate": 8e-05,
      "loss": 1.8015,
      "step": 3830
    },
    {
      "epoch": 0.8426261959749257,
      "grad_norm": 0.3034017086029053,
      "learning_rate": 8e-05,
      "loss": 1.7774,
      "step": 3831
    },
    {
      "epoch": 0.8428461453865611,
      "grad_norm": 0.2914513647556305,
      "learning_rate": 8e-05,
      "loss": 1.672,
      "step": 3832
    },
    {
      "epoch": 0.8430660947981964,
      "grad_norm": 0.2868506610393524,
      "learning_rate": 8e-05,
      "loss": 1.6906,
      "step": 3833
    },
    {
      "epoch": 0.8432860442098318,
      "grad_norm": 0.2967677712440491,
      "learning_rate": 8e-05,
      "loss": 1.4859,
      "step": 3834
    },
    {
      "epoch": 0.843505993621467,
      "grad_norm": 0.2904664874076843,
      "learning_rate": 8e-05,
      "loss": 1.7559,
      "step": 3835
    },
    {
      "epoch": 0.8437259430331023,
      "grad_norm": 0.2900542616844177,
      "learning_rate": 8e-05,
      "loss": 1.6315,
      "step": 3836
    },
    {
      "epoch": 0.8439458924447377,
      "grad_norm": 0.28981515765190125,
      "learning_rate": 8e-05,
      "loss": 1.6254,
      "step": 3837
    },
    {
      "epoch": 0.844165841856373,
      "grad_norm": 0.29468080401420593,
      "learning_rate": 8e-05,
      "loss": 1.6417,
      "step": 3838
    },
    {
      "epoch": 0.8443857912680084,
      "grad_norm": 0.3301861584186554,
      "learning_rate": 8e-05,
      "loss": 1.7982,
      "step": 3839
    },
    {
      "epoch": 0.8446057406796437,
      "grad_norm": 0.28386393189430237,
      "learning_rate": 8e-05,
      "loss": 1.7077,
      "step": 3840
    },
    {
      "epoch": 0.8448256900912791,
      "grad_norm": 0.27421340346336365,
      "learning_rate": 8e-05,
      "loss": 1.5492,
      "step": 3841
    },
    {
      "epoch": 0.8450456395029143,
      "grad_norm": 0.30234119296073914,
      "learning_rate": 8e-05,
      "loss": 1.6162,
      "step": 3842
    },
    {
      "epoch": 0.8452655889145496,
      "grad_norm": 0.3157059848308563,
      "learning_rate": 8e-05,
      "loss": 1.7228,
      "step": 3843
    },
    {
      "epoch": 0.845485538326185,
      "grad_norm": 0.2899254858493805,
      "learning_rate": 8e-05,
      "loss": 1.451,
      "step": 3844
    },
    {
      "epoch": 0.8457054877378203,
      "grad_norm": 0.3020762801170349,
      "learning_rate": 8e-05,
      "loss": 1.8273,
      "step": 3845
    },
    {
      "epoch": 0.8459254371494557,
      "grad_norm": 0.3134917616844177,
      "learning_rate": 8e-05,
      "loss": 1.8578,
      "step": 3846
    },
    {
      "epoch": 0.846145386561091,
      "grad_norm": 0.29482507705688477,
      "learning_rate": 8e-05,
      "loss": 1.6039,
      "step": 3847
    },
    {
      "epoch": 0.8463653359727262,
      "grad_norm": 0.3106936514377594,
      "learning_rate": 8e-05,
      "loss": 1.7019,
      "step": 3848
    },
    {
      "epoch": 0.8465852853843616,
      "grad_norm": 0.2869921028614044,
      "learning_rate": 8e-05,
      "loss": 1.6515,
      "step": 3849
    },
    {
      "epoch": 0.8468052347959969,
      "grad_norm": 0.32278892397880554,
      "learning_rate": 8e-05,
      "loss": 1.955,
      "step": 3850
    },
    {
      "epoch": 0.8470251842076323,
      "grad_norm": 0.28249430656433105,
      "learning_rate": 8e-05,
      "loss": 1.575,
      "step": 3851
    },
    {
      "epoch": 0.8472451336192676,
      "grad_norm": 0.3081803321838379,
      "learning_rate": 8e-05,
      "loss": 1.5634,
      "step": 3852
    },
    {
      "epoch": 0.8474650830309028,
      "grad_norm": 0.3033173978328705,
      "learning_rate": 8e-05,
      "loss": 1.6182,
      "step": 3853
    },
    {
      "epoch": 0.8476850324425382,
      "grad_norm": 0.3237481415271759,
      "learning_rate": 8e-05,
      "loss": 1.5605,
      "step": 3854
    },
    {
      "epoch": 0.8479049818541735,
      "grad_norm": 0.3091139495372772,
      "learning_rate": 8e-05,
      "loss": 1.7551,
      "step": 3855
    },
    {
      "epoch": 0.8481249312658089,
      "grad_norm": 0.2904627025127411,
      "learning_rate": 8e-05,
      "loss": 1.6982,
      "step": 3856
    },
    {
      "epoch": 0.8483448806774442,
      "grad_norm": 0.3025551438331604,
      "learning_rate": 8e-05,
      "loss": 1.6156,
      "step": 3857
    },
    {
      "epoch": 0.8485648300890795,
      "grad_norm": 0.28311145305633545,
      "learning_rate": 8e-05,
      "loss": 1.6919,
      "step": 3858
    },
    {
      "epoch": 0.8487847795007148,
      "grad_norm": 0.2805883288383484,
      "learning_rate": 8e-05,
      "loss": 1.5491,
      "step": 3859
    },
    {
      "epoch": 0.8490047289123501,
      "grad_norm": 0.31418681144714355,
      "learning_rate": 8e-05,
      "loss": 1.7864,
      "step": 3860
    },
    {
      "epoch": 0.8492246783239855,
      "grad_norm": 0.2807353734970093,
      "learning_rate": 8e-05,
      "loss": 1.4794,
      "step": 3861
    },
    {
      "epoch": 0.8494446277356208,
      "grad_norm": 0.29171043634414673,
      "learning_rate": 8e-05,
      "loss": 1.7223,
      "step": 3862
    },
    {
      "epoch": 0.8496645771472562,
      "grad_norm": 0.30100691318511963,
      "learning_rate": 8e-05,
      "loss": 1.8106,
      "step": 3863
    },
    {
      "epoch": 0.8498845265588915,
      "grad_norm": 0.29176101088523865,
      "learning_rate": 8e-05,
      "loss": 1.6441,
      "step": 3864
    },
    {
      "epoch": 0.8501044759705267,
      "grad_norm": 0.27088651061058044,
      "learning_rate": 8e-05,
      "loss": 1.4872,
      "step": 3865
    },
    {
      "epoch": 0.8503244253821621,
      "grad_norm": 0.28077685832977295,
      "learning_rate": 8e-05,
      "loss": 1.6243,
      "step": 3866
    },
    {
      "epoch": 0.8505443747937974,
      "grad_norm": 0.30027422308921814,
      "learning_rate": 8e-05,
      "loss": 1.7244,
      "step": 3867
    },
    {
      "epoch": 0.8507643242054328,
      "grad_norm": 0.29100343585014343,
      "learning_rate": 8e-05,
      "loss": 1.6324,
      "step": 3868
    },
    {
      "epoch": 0.8509842736170681,
      "grad_norm": 0.29306820034980774,
      "learning_rate": 8e-05,
      "loss": 1.7432,
      "step": 3869
    },
    {
      "epoch": 0.8512042230287034,
      "grad_norm": 0.29503849148750305,
      "learning_rate": 8e-05,
      "loss": 1.5348,
      "step": 3870
    },
    {
      "epoch": 0.8514241724403387,
      "grad_norm": 0.28499239683151245,
      "learning_rate": 8e-05,
      "loss": 1.6653,
      "step": 3871
    },
    {
      "epoch": 0.851644121851974,
      "grad_norm": 0.3126235008239746,
      "learning_rate": 8e-05,
      "loss": 1.6023,
      "step": 3872
    },
    {
      "epoch": 0.8518640712636094,
      "grad_norm": 0.2873520255088806,
      "learning_rate": 8e-05,
      "loss": 1.6659,
      "step": 3873
    },
    {
      "epoch": 0.8520840206752447,
      "grad_norm": 0.28033289313316345,
      "learning_rate": 8e-05,
      "loss": 1.6133,
      "step": 3874
    },
    {
      "epoch": 0.85230397008688,
      "grad_norm": 0.28731170296669006,
      "learning_rate": 8e-05,
      "loss": 1.6344,
      "step": 3875
    },
    {
      "epoch": 0.8525239194985154,
      "grad_norm": 0.28616607189178467,
      "learning_rate": 8e-05,
      "loss": 1.5874,
      "step": 3876
    },
    {
      "epoch": 0.8527438689101506,
      "grad_norm": 0.29681506752967834,
      "learning_rate": 8e-05,
      "loss": 1.6628,
      "step": 3877
    },
    {
      "epoch": 0.852963818321786,
      "grad_norm": 0.3163807690143585,
      "learning_rate": 8e-05,
      "loss": 1.7927,
      "step": 3878
    },
    {
      "epoch": 0.8531837677334213,
      "grad_norm": 0.28302204608917236,
      "learning_rate": 8e-05,
      "loss": 1.6406,
      "step": 3879
    },
    {
      "epoch": 0.8534037171450566,
      "grad_norm": 0.29161083698272705,
      "learning_rate": 8e-05,
      "loss": 1.7386,
      "step": 3880
    },
    {
      "epoch": 0.853623666556692,
      "grad_norm": 0.29691028594970703,
      "learning_rate": 8e-05,
      "loss": 1.6582,
      "step": 3881
    },
    {
      "epoch": 0.8538436159683273,
      "grad_norm": 0.28576239943504333,
      "learning_rate": 8e-05,
      "loss": 1.6866,
      "step": 3882
    },
    {
      "epoch": 0.8540635653799626,
      "grad_norm": 0.2867792844772339,
      "learning_rate": 8e-05,
      "loss": 1.4684,
      "step": 3883
    },
    {
      "epoch": 0.8542835147915979,
      "grad_norm": 0.2788017690181732,
      "learning_rate": 8e-05,
      "loss": 1.6202,
      "step": 3884
    },
    {
      "epoch": 0.8545034642032333,
      "grad_norm": 0.2767166197299957,
      "learning_rate": 8e-05,
      "loss": 1.5813,
      "step": 3885
    },
    {
      "epoch": 0.8547234136148686,
      "grad_norm": 0.29414990544319153,
      "learning_rate": 8e-05,
      "loss": 1.6617,
      "step": 3886
    },
    {
      "epoch": 0.8549433630265039,
      "grad_norm": 0.2955497205257416,
      "learning_rate": 8e-05,
      "loss": 1.6456,
      "step": 3887
    },
    {
      "epoch": 0.8551633124381393,
      "grad_norm": 0.31308290362358093,
      "learning_rate": 8e-05,
      "loss": 1.6532,
      "step": 3888
    },
    {
      "epoch": 0.8553832618497745,
      "grad_norm": 0.301305890083313,
      "learning_rate": 8e-05,
      "loss": 1.6877,
      "step": 3889
    },
    {
      "epoch": 0.8556032112614099,
      "grad_norm": 0.290628582239151,
      "learning_rate": 8e-05,
      "loss": 1.7098,
      "step": 3890
    },
    {
      "epoch": 0.8558231606730452,
      "grad_norm": 0.30904853343963623,
      "learning_rate": 8e-05,
      "loss": 1.6352,
      "step": 3891
    },
    {
      "epoch": 0.8560431100846805,
      "grad_norm": 0.29854530096054077,
      "learning_rate": 8e-05,
      "loss": 1.6971,
      "step": 3892
    },
    {
      "epoch": 0.8562630594963159,
      "grad_norm": 0.2870941460132599,
      "learning_rate": 8e-05,
      "loss": 1.6655,
      "step": 3893
    },
    {
      "epoch": 0.8564830089079511,
      "grad_norm": 0.29070189595222473,
      "learning_rate": 8e-05,
      "loss": 1.5808,
      "step": 3894
    },
    {
      "epoch": 0.8567029583195865,
      "grad_norm": 0.3046536445617676,
      "learning_rate": 8e-05,
      "loss": 1.7834,
      "step": 3895
    },
    {
      "epoch": 0.8569229077312218,
      "grad_norm": 0.2883049249649048,
      "learning_rate": 8e-05,
      "loss": 1.5898,
      "step": 3896
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.29613447189331055,
      "learning_rate": 8e-05,
      "loss": 1.7619,
      "step": 3897
    },
    {
      "epoch": 0.8573628065544925,
      "grad_norm": 0.27996566891670227,
      "learning_rate": 8e-05,
      "loss": 1.5863,
      "step": 3898
    },
    {
      "epoch": 0.8575827559661278,
      "grad_norm": 0.3026728332042694,
      "learning_rate": 8e-05,
      "loss": 1.7488,
      "step": 3899
    },
    {
      "epoch": 0.8578027053777632,
      "grad_norm": 0.28238317370414734,
      "learning_rate": 8e-05,
      "loss": 1.6908,
      "step": 3900
    },
    {
      "epoch": 0.8580226547893984,
      "grad_norm": 0.271651953458786,
      "learning_rate": 8e-05,
      "loss": 1.5228,
      "step": 3901
    },
    {
      "epoch": 0.8582426042010337,
      "grad_norm": 0.2936142385005951,
      "learning_rate": 8e-05,
      "loss": 1.7332,
      "step": 3902
    },
    {
      "epoch": 0.8584625536126691,
      "grad_norm": 0.29982343316078186,
      "learning_rate": 8e-05,
      "loss": 1.6792,
      "step": 3903
    },
    {
      "epoch": 0.8586825030243044,
      "grad_norm": 0.30300748348236084,
      "learning_rate": 8e-05,
      "loss": 1.7689,
      "step": 3904
    },
    {
      "epoch": 0.8589024524359398,
      "grad_norm": 0.3021508455276489,
      "learning_rate": 8e-05,
      "loss": 1.7082,
      "step": 3905
    },
    {
      "epoch": 0.859122401847575,
      "grad_norm": 0.2989715039730072,
      "learning_rate": 8e-05,
      "loss": 1.6786,
      "step": 3906
    },
    {
      "epoch": 0.8593423512592104,
      "grad_norm": 0.2968369424343109,
      "learning_rate": 8e-05,
      "loss": 1.7742,
      "step": 3907
    },
    {
      "epoch": 0.8595623006708457,
      "grad_norm": 0.2943005859851837,
      "learning_rate": 8e-05,
      "loss": 1.6512,
      "step": 3908
    },
    {
      "epoch": 0.859782250082481,
      "grad_norm": 0.2851683497428894,
      "learning_rate": 8e-05,
      "loss": 1.6554,
      "step": 3909
    },
    {
      "epoch": 0.8600021994941164,
      "grad_norm": 0.295810729265213,
      "learning_rate": 8e-05,
      "loss": 1.7791,
      "step": 3910
    },
    {
      "epoch": 0.8602221489057517,
      "grad_norm": 0.28922411799430847,
      "learning_rate": 8e-05,
      "loss": 1.7462,
      "step": 3911
    },
    {
      "epoch": 0.860442098317387,
      "grad_norm": 0.2724452316761017,
      "learning_rate": 8e-05,
      "loss": 1.3677,
      "step": 3912
    },
    {
      "epoch": 0.8606620477290223,
      "grad_norm": 0.28607258200645447,
      "learning_rate": 8e-05,
      "loss": 1.7326,
      "step": 3913
    },
    {
      "epoch": 0.8608819971406576,
      "grad_norm": 0.26861658692359924,
      "learning_rate": 8e-05,
      "loss": 1.4499,
      "step": 3914
    },
    {
      "epoch": 0.861101946552293,
      "grad_norm": 0.2905106246471405,
      "learning_rate": 8e-05,
      "loss": 1.6292,
      "step": 3915
    },
    {
      "epoch": 0.8613218959639283,
      "grad_norm": 0.28987225890159607,
      "learning_rate": 8e-05,
      "loss": 1.5838,
      "step": 3916
    },
    {
      "epoch": 0.8615418453755637,
      "grad_norm": 0.3006553053855896,
      "learning_rate": 8e-05,
      "loss": 1.5658,
      "step": 3917
    },
    {
      "epoch": 0.8617617947871989,
      "grad_norm": 0.2888564467430115,
      "learning_rate": 8e-05,
      "loss": 1.6067,
      "step": 3918
    },
    {
      "epoch": 0.8619817441988342,
      "grad_norm": 0.3040676414966583,
      "learning_rate": 8e-05,
      "loss": 1.8178,
      "step": 3919
    },
    {
      "epoch": 0.8622016936104696,
      "grad_norm": 0.29286056756973267,
      "learning_rate": 8e-05,
      "loss": 1.6235,
      "step": 3920
    },
    {
      "epoch": 0.8624216430221049,
      "grad_norm": 0.33138954639434814,
      "learning_rate": 8e-05,
      "loss": 1.6806,
      "step": 3921
    },
    {
      "epoch": 0.8626415924337403,
      "grad_norm": 0.2946946620941162,
      "learning_rate": 8e-05,
      "loss": 1.631,
      "step": 3922
    },
    {
      "epoch": 0.8628615418453756,
      "grad_norm": 0.3030078411102295,
      "learning_rate": 8e-05,
      "loss": 1.7346,
      "step": 3923
    },
    {
      "epoch": 0.8630814912570108,
      "grad_norm": 0.2995181381702423,
      "learning_rate": 8e-05,
      "loss": 1.6159,
      "step": 3924
    },
    {
      "epoch": 0.8633014406686462,
      "grad_norm": 0.31188464164733887,
      "learning_rate": 8e-05,
      "loss": 1.7165,
      "step": 3925
    },
    {
      "epoch": 0.8635213900802815,
      "grad_norm": 0.2989533245563507,
      "learning_rate": 8e-05,
      "loss": 1.7557,
      "step": 3926
    },
    {
      "epoch": 0.8637413394919169,
      "grad_norm": 0.2965574264526367,
      "learning_rate": 8e-05,
      "loss": 1.678,
      "step": 3927
    },
    {
      "epoch": 0.8639612889035522,
      "grad_norm": 0.28041690587997437,
      "learning_rate": 8e-05,
      "loss": 1.553,
      "step": 3928
    },
    {
      "epoch": 0.8641812383151876,
      "grad_norm": 0.2907460033893585,
      "learning_rate": 8e-05,
      "loss": 1.6468,
      "step": 3929
    },
    {
      "epoch": 0.8644011877268228,
      "grad_norm": 0.3051550090312958,
      "learning_rate": 8e-05,
      "loss": 1.7234,
      "step": 3930
    },
    {
      "epoch": 0.8646211371384581,
      "grad_norm": 0.2832759618759155,
      "learning_rate": 8e-05,
      "loss": 1.5426,
      "step": 3931
    },
    {
      "epoch": 0.8648410865500935,
      "grad_norm": 0.2915174961090088,
      "learning_rate": 8e-05,
      "loss": 1.649,
      "step": 3932
    },
    {
      "epoch": 0.8650610359617288,
      "grad_norm": 0.3146245777606964,
      "learning_rate": 8e-05,
      "loss": 1.7486,
      "step": 3933
    },
    {
      "epoch": 0.8652809853733642,
      "grad_norm": 0.2949977517127991,
      "learning_rate": 8e-05,
      "loss": 1.6221,
      "step": 3934
    },
    {
      "epoch": 0.8655009347849995,
      "grad_norm": 0.34237998723983765,
      "learning_rate": 8e-05,
      "loss": 1.6641,
      "step": 3935
    },
    {
      "epoch": 0.8657208841966347,
      "grad_norm": 0.3072294592857361,
      "learning_rate": 8e-05,
      "loss": 1.6981,
      "step": 3936
    },
    {
      "epoch": 0.8659408336082701,
      "grad_norm": 0.2796785235404968,
      "learning_rate": 8e-05,
      "loss": 1.5687,
      "step": 3937
    },
    {
      "epoch": 0.8661607830199054,
      "grad_norm": 0.27884992957115173,
      "learning_rate": 8e-05,
      "loss": 1.4457,
      "step": 3938
    },
    {
      "epoch": 0.8663807324315408,
      "grad_norm": 0.30189958214759827,
      "learning_rate": 8e-05,
      "loss": 1.7328,
      "step": 3939
    },
    {
      "epoch": 0.8666006818431761,
      "grad_norm": 0.3078850209712982,
      "learning_rate": 8e-05,
      "loss": 1.8462,
      "step": 3940
    },
    {
      "epoch": 0.8668206312548113,
      "grad_norm": 0.27258560061454773,
      "learning_rate": 8e-05,
      "loss": 1.501,
      "step": 3941
    },
    {
      "epoch": 0.8670405806664467,
      "grad_norm": 0.31592684984207153,
      "learning_rate": 8e-05,
      "loss": 1.573,
      "step": 3942
    },
    {
      "epoch": 0.867260530078082,
      "grad_norm": 0.2867029309272766,
      "learning_rate": 8e-05,
      "loss": 1.6152,
      "step": 3943
    },
    {
      "epoch": 0.8674804794897174,
      "grad_norm": 0.31841394305229187,
      "learning_rate": 8e-05,
      "loss": 1.6078,
      "step": 3944
    },
    {
      "epoch": 0.8677004289013527,
      "grad_norm": 0.3039727509021759,
      "learning_rate": 8e-05,
      "loss": 1.6775,
      "step": 3945
    },
    {
      "epoch": 0.867920378312988,
      "grad_norm": 0.3054428994655609,
      "learning_rate": 8e-05,
      "loss": 1.8358,
      "step": 3946
    },
    {
      "epoch": 0.8681403277246234,
      "grad_norm": 0.2971208691596985,
      "learning_rate": 8e-05,
      "loss": 1.6011,
      "step": 3947
    },
    {
      "epoch": 0.8683602771362586,
      "grad_norm": 0.30911338329315186,
      "learning_rate": 8e-05,
      "loss": 1.7928,
      "step": 3948
    },
    {
      "epoch": 0.868580226547894,
      "grad_norm": 0.3277186453342438,
      "learning_rate": 8e-05,
      "loss": 1.5783,
      "step": 3949
    },
    {
      "epoch": 0.8688001759595293,
      "grad_norm": 0.29121798276901245,
      "learning_rate": 8e-05,
      "loss": 1.6368,
      "step": 3950
    },
    {
      "epoch": 0.8690201253711647,
      "grad_norm": 0.30979645252227783,
      "learning_rate": 8e-05,
      "loss": 1.6603,
      "step": 3951
    },
    {
      "epoch": 0.8692400747828,
      "grad_norm": 0.29589441418647766,
      "learning_rate": 8e-05,
      "loss": 1.7536,
      "step": 3952
    },
    {
      "epoch": 0.8694600241944352,
      "grad_norm": 0.3051590323448181,
      "learning_rate": 8e-05,
      "loss": 1.6639,
      "step": 3953
    },
    {
      "epoch": 0.8696799736060706,
      "grad_norm": 0.30917656421661377,
      "learning_rate": 8e-05,
      "loss": 1.8418,
      "step": 3954
    },
    {
      "epoch": 0.8698999230177059,
      "grad_norm": 0.27990633249282837,
      "learning_rate": 8e-05,
      "loss": 1.7211,
      "step": 3955
    },
    {
      "epoch": 0.8701198724293413,
      "grad_norm": 0.28773045539855957,
      "learning_rate": 8e-05,
      "loss": 1.5785,
      "step": 3956
    },
    {
      "epoch": 0.8703398218409766,
      "grad_norm": 0.2980699837207794,
      "learning_rate": 8e-05,
      "loss": 1.5905,
      "step": 3957
    },
    {
      "epoch": 0.8705597712526119,
      "grad_norm": 0.30660519003868103,
      "learning_rate": 8e-05,
      "loss": 1.6561,
      "step": 3958
    },
    {
      "epoch": 0.8707797206642472,
      "grad_norm": 0.2980074882507324,
      "learning_rate": 8e-05,
      "loss": 1.5939,
      "step": 3959
    },
    {
      "epoch": 0.8709996700758825,
      "grad_norm": 0.30249714851379395,
      "learning_rate": 8e-05,
      "loss": 1.7815,
      "step": 3960
    },
    {
      "epoch": 0.8712196194875179,
      "grad_norm": 0.3185308277606964,
      "learning_rate": 8e-05,
      "loss": 1.6994,
      "step": 3961
    },
    {
      "epoch": 0.8714395688991532,
      "grad_norm": 0.29345861077308655,
      "learning_rate": 8e-05,
      "loss": 1.6363,
      "step": 3962
    },
    {
      "epoch": 0.8716595183107885,
      "grad_norm": 0.29790154099464417,
      "learning_rate": 8e-05,
      "loss": 1.6433,
      "step": 3963
    },
    {
      "epoch": 0.8718794677224239,
      "grad_norm": 0.30221304297447205,
      "learning_rate": 8e-05,
      "loss": 1.6036,
      "step": 3964
    },
    {
      "epoch": 0.8720994171340591,
      "grad_norm": 0.3027768135070801,
      "learning_rate": 8e-05,
      "loss": 1.7376,
      "step": 3965
    },
    {
      "epoch": 0.8723193665456945,
      "grad_norm": 0.32072335481643677,
      "learning_rate": 8e-05,
      "loss": 1.7161,
      "step": 3966
    },
    {
      "epoch": 0.8725393159573298,
      "grad_norm": 0.3282933831214905,
      "learning_rate": 8e-05,
      "loss": 1.8066,
      "step": 3967
    },
    {
      "epoch": 0.8727592653689651,
      "grad_norm": 0.2951596677303314,
      "learning_rate": 8e-05,
      "loss": 1.5761,
      "step": 3968
    },
    {
      "epoch": 0.8729792147806005,
      "grad_norm": 0.3005920946598053,
      "learning_rate": 8e-05,
      "loss": 1.6086,
      "step": 3969
    },
    {
      "epoch": 0.8731991641922358,
      "grad_norm": 0.2846725881099701,
      "learning_rate": 8e-05,
      "loss": 1.6208,
      "step": 3970
    },
    {
      "epoch": 0.8734191136038711,
      "grad_norm": 0.3003213107585907,
      "learning_rate": 8e-05,
      "loss": 1.6404,
      "step": 3971
    },
    {
      "epoch": 0.8736390630155064,
      "grad_norm": 0.2906554639339447,
      "learning_rate": 8e-05,
      "loss": 1.5947,
      "step": 3972
    },
    {
      "epoch": 0.8738590124271418,
      "grad_norm": 0.2852460741996765,
      "learning_rate": 8e-05,
      "loss": 1.6473,
      "step": 3973
    },
    {
      "epoch": 0.8740789618387771,
      "grad_norm": 0.2833881378173828,
      "learning_rate": 8e-05,
      "loss": 1.6245,
      "step": 3974
    },
    {
      "epoch": 0.8742989112504124,
      "grad_norm": 0.3263465464115143,
      "learning_rate": 8e-05,
      "loss": 1.6252,
      "step": 3975
    },
    {
      "epoch": 0.8745188606620478,
      "grad_norm": 0.3010908365249634,
      "learning_rate": 8e-05,
      "loss": 1.6072,
      "step": 3976
    },
    {
      "epoch": 0.874738810073683,
      "grad_norm": 0.29418328404426575,
      "learning_rate": 8e-05,
      "loss": 1.7066,
      "step": 3977
    },
    {
      "epoch": 0.8749587594853184,
      "grad_norm": 0.31706956028938293,
      "learning_rate": 8e-05,
      "loss": 1.6545,
      "step": 3978
    },
    {
      "epoch": 0.8751787088969537,
      "grad_norm": 0.31312480568885803,
      "learning_rate": 8e-05,
      "loss": 1.6229,
      "step": 3979
    },
    {
      "epoch": 0.875398658308589,
      "grad_norm": 0.27910348773002625,
      "learning_rate": 8e-05,
      "loss": 1.7239,
      "step": 3980
    },
    {
      "epoch": 0.8756186077202244,
      "grad_norm": 0.33857640624046326,
      "learning_rate": 8e-05,
      "loss": 1.7275,
      "step": 3981
    },
    {
      "epoch": 0.8758385571318597,
      "grad_norm": 0.2809359133243561,
      "learning_rate": 8e-05,
      "loss": 1.482,
      "step": 3982
    },
    {
      "epoch": 0.876058506543495,
      "grad_norm": 0.3103906214237213,
      "learning_rate": 8e-05,
      "loss": 1.7277,
      "step": 3983
    },
    {
      "epoch": 0.8762784559551303,
      "grad_norm": 0.2988683581352234,
      "learning_rate": 8e-05,
      "loss": 1.7167,
      "step": 3984
    },
    {
      "epoch": 0.8764984053667656,
      "grad_norm": 0.3110254406929016,
      "learning_rate": 8e-05,
      "loss": 1.6631,
      "step": 3985
    },
    {
      "epoch": 0.876718354778401,
      "grad_norm": 0.2963966429233551,
      "learning_rate": 8e-05,
      "loss": 1.7406,
      "step": 3986
    },
    {
      "epoch": 0.8769383041900363,
      "grad_norm": 0.3057374954223633,
      "learning_rate": 8e-05,
      "loss": 1.5424,
      "step": 3987
    },
    {
      "epoch": 0.8771582536016717,
      "grad_norm": 0.32236918807029724,
      "learning_rate": 8e-05,
      "loss": 1.6419,
      "step": 3988
    },
    {
      "epoch": 0.8773782030133069,
      "grad_norm": 0.35666584968566895,
      "learning_rate": 8e-05,
      "loss": 1.7104,
      "step": 3989
    },
    {
      "epoch": 0.8775981524249422,
      "grad_norm": 0.2982276380062103,
      "learning_rate": 8e-05,
      "loss": 1.5707,
      "step": 3990
    },
    {
      "epoch": 0.8778181018365776,
      "grad_norm": 0.28403279185295105,
      "learning_rate": 8e-05,
      "loss": 1.592,
      "step": 3991
    },
    {
      "epoch": 0.8780380512482129,
      "grad_norm": 0.3072356581687927,
      "learning_rate": 8e-05,
      "loss": 1.7964,
      "step": 3992
    },
    {
      "epoch": 0.8782580006598483,
      "grad_norm": 0.29571470618247986,
      "learning_rate": 8e-05,
      "loss": 1.7442,
      "step": 3993
    },
    {
      "epoch": 0.8784779500714835,
      "grad_norm": 0.30304571986198425,
      "learning_rate": 8e-05,
      "loss": 1.7008,
      "step": 3994
    },
    {
      "epoch": 0.8786978994831188,
      "grad_norm": 0.2830870747566223,
      "learning_rate": 8e-05,
      "loss": 1.6545,
      "step": 3995
    },
    {
      "epoch": 0.8789178488947542,
      "grad_norm": 0.3138338029384613,
      "learning_rate": 8e-05,
      "loss": 1.649,
      "step": 3996
    },
    {
      "epoch": 0.8791377983063895,
      "grad_norm": 0.28111809492111206,
      "learning_rate": 8e-05,
      "loss": 1.6918,
      "step": 3997
    },
    {
      "epoch": 0.8793577477180249,
      "grad_norm": 0.30100250244140625,
      "learning_rate": 8e-05,
      "loss": 1.7132,
      "step": 3998
    },
    {
      "epoch": 0.8795776971296602,
      "grad_norm": 0.3129050135612488,
      "learning_rate": 8e-05,
      "loss": 1.7433,
      "step": 3999
    },
    {
      "epoch": 0.8797976465412956,
      "grad_norm": 0.2925567030906677,
      "learning_rate": 8e-05,
      "loss": 1.7867,
      "step": 4000
    },
    {
      "epoch": 0.8800175959529308,
      "grad_norm": 0.28086230158805847,
      "learning_rate": 8e-05,
      "loss": 1.5538,
      "step": 4001
    },
    {
      "epoch": 0.8802375453645661,
      "grad_norm": 0.29163795709609985,
      "learning_rate": 8e-05,
      "loss": 1.6125,
      "step": 4002
    },
    {
      "epoch": 0.8804574947762015,
      "grad_norm": 0.278808057308197,
      "learning_rate": 8e-05,
      "loss": 1.4841,
      "step": 4003
    },
    {
      "epoch": 0.8806774441878368,
      "grad_norm": 0.3127332031726837,
      "learning_rate": 8e-05,
      "loss": 1.821,
      "step": 4004
    },
    {
      "epoch": 0.8808973935994722,
      "grad_norm": 0.3092391788959503,
      "learning_rate": 8e-05,
      "loss": 1.4779,
      "step": 4005
    },
    {
      "epoch": 0.8811173430111074,
      "grad_norm": 0.30426523089408875,
      "learning_rate": 8e-05,
      "loss": 1.7699,
      "step": 4006
    },
    {
      "epoch": 0.8813372924227427,
      "grad_norm": 0.28605568408966064,
      "learning_rate": 8e-05,
      "loss": 1.7,
      "step": 4007
    },
    {
      "epoch": 0.8815572418343781,
      "grad_norm": 0.2959664762020111,
      "learning_rate": 8e-05,
      "loss": 1.733,
      "step": 4008
    },
    {
      "epoch": 0.8817771912460134,
      "grad_norm": 0.33198082447052,
      "learning_rate": 8e-05,
      "loss": 1.7399,
      "step": 4009
    },
    {
      "epoch": 0.8819971406576488,
      "grad_norm": 0.3039609491825104,
      "learning_rate": 8e-05,
      "loss": 1.7289,
      "step": 4010
    },
    {
      "epoch": 0.8822170900692841,
      "grad_norm": 0.3058362603187561,
      "learning_rate": 8e-05,
      "loss": 1.7897,
      "step": 4011
    },
    {
      "epoch": 0.8824370394809193,
      "grad_norm": 0.3194586932659149,
      "learning_rate": 8e-05,
      "loss": 1.806,
      "step": 4012
    },
    {
      "epoch": 0.8826569888925547,
      "grad_norm": 0.31681060791015625,
      "learning_rate": 8e-05,
      "loss": 1.7131,
      "step": 4013
    },
    {
      "epoch": 0.88287693830419,
      "grad_norm": 0.2874566912651062,
      "learning_rate": 8e-05,
      "loss": 1.5243,
      "step": 4014
    },
    {
      "epoch": 0.8830968877158254,
      "grad_norm": 0.3297685980796814,
      "learning_rate": 8e-05,
      "loss": 1.9256,
      "step": 4015
    },
    {
      "epoch": 0.8833168371274607,
      "grad_norm": 0.30325692892074585,
      "learning_rate": 8e-05,
      "loss": 1.7809,
      "step": 4016
    },
    {
      "epoch": 0.883536786539096,
      "grad_norm": 0.38110217452049255,
      "learning_rate": 8e-05,
      "loss": 1.7246,
      "step": 4017
    },
    {
      "epoch": 0.8837567359507313,
      "grad_norm": 0.3073844313621521,
      "learning_rate": 8e-05,
      "loss": 1.6379,
      "step": 4018
    },
    {
      "epoch": 0.8839766853623666,
      "grad_norm": 0.3144959509372711,
      "learning_rate": 8e-05,
      "loss": 1.7369,
      "step": 4019
    },
    {
      "epoch": 0.884196634774002,
      "grad_norm": 0.29416364431381226,
      "learning_rate": 8e-05,
      "loss": 1.6181,
      "step": 4020
    },
    {
      "epoch": 0.8844165841856373,
      "grad_norm": 0.31616780161857605,
      "learning_rate": 8e-05,
      "loss": 1.6965,
      "step": 4021
    },
    {
      "epoch": 0.8846365335972727,
      "grad_norm": 0.3114292621612549,
      "learning_rate": 8e-05,
      "loss": 1.7151,
      "step": 4022
    },
    {
      "epoch": 0.884856483008908,
      "grad_norm": 0.2938895523548126,
      "learning_rate": 8e-05,
      "loss": 1.6919,
      "step": 4023
    },
    {
      "epoch": 0.8850764324205432,
      "grad_norm": 0.3129356801509857,
      "learning_rate": 8e-05,
      "loss": 1.8082,
      "step": 4024
    },
    {
      "epoch": 0.8852963818321786,
      "grad_norm": 0.2698216140270233,
      "learning_rate": 8e-05,
      "loss": 1.5458,
      "step": 4025
    },
    {
      "epoch": 0.8855163312438139,
      "grad_norm": 0.3061481714248657,
      "learning_rate": 8e-05,
      "loss": 1.7474,
      "step": 4026
    },
    {
      "epoch": 0.8857362806554493,
      "grad_norm": 0.3081664741039276,
      "learning_rate": 8e-05,
      "loss": 1.7645,
      "step": 4027
    },
    {
      "epoch": 0.8859562300670846,
      "grad_norm": 0.29356658458709717,
      "learning_rate": 8e-05,
      "loss": 1.7544,
      "step": 4028
    },
    {
      "epoch": 0.8861761794787198,
      "grad_norm": 0.287942498922348,
      "learning_rate": 8e-05,
      "loss": 1.5793,
      "step": 4029
    },
    {
      "epoch": 0.8863961288903552,
      "grad_norm": 0.2899230718612671,
      "learning_rate": 8e-05,
      "loss": 1.7263,
      "step": 4030
    },
    {
      "epoch": 0.8866160783019905,
      "grad_norm": 0.28524360060691833,
      "learning_rate": 8e-05,
      "loss": 1.6558,
      "step": 4031
    },
    {
      "epoch": 0.8868360277136259,
      "grad_norm": 0.34471920132637024,
      "learning_rate": 8e-05,
      "loss": 1.7586,
      "step": 4032
    },
    {
      "epoch": 0.8870559771252612,
      "grad_norm": 0.2993306815624237,
      "learning_rate": 8e-05,
      "loss": 1.8712,
      "step": 4033
    },
    {
      "epoch": 0.8872759265368965,
      "grad_norm": 0.30809563398361206,
      "learning_rate": 8e-05,
      "loss": 1.4949,
      "step": 4034
    },
    {
      "epoch": 0.8874958759485319,
      "grad_norm": 0.3870789110660553,
      "learning_rate": 8e-05,
      "loss": 1.7098,
      "step": 4035
    },
    {
      "epoch": 0.8877158253601671,
      "grad_norm": 0.27530455589294434,
      "learning_rate": 8e-05,
      "loss": 1.5475,
      "step": 4036
    },
    {
      "epoch": 0.8879357747718025,
      "grad_norm": 0.3203597366809845,
      "learning_rate": 8e-05,
      "loss": 1.5524,
      "step": 4037
    },
    {
      "epoch": 0.8881557241834378,
      "grad_norm": 0.2910875678062439,
      "learning_rate": 8e-05,
      "loss": 1.6297,
      "step": 4038
    },
    {
      "epoch": 0.8883756735950731,
      "grad_norm": 0.362775057554245,
      "learning_rate": 8e-05,
      "loss": 2.0182,
      "step": 4039
    },
    {
      "epoch": 0.8885956230067085,
      "grad_norm": 0.31645676493644714,
      "learning_rate": 8e-05,
      "loss": 1.7838,
      "step": 4040
    },
    {
      "epoch": 0.8888155724183437,
      "grad_norm": 0.3013925552368164,
      "learning_rate": 8e-05,
      "loss": 1.7254,
      "step": 4041
    },
    {
      "epoch": 0.8890355218299791,
      "grad_norm": 0.3068082928657532,
      "learning_rate": 8e-05,
      "loss": 1.7238,
      "step": 4042
    },
    {
      "epoch": 0.8892554712416144,
      "grad_norm": 0.29745063185691833,
      "learning_rate": 8e-05,
      "loss": 1.8604,
      "step": 4043
    },
    {
      "epoch": 0.8894754206532498,
      "grad_norm": 0.30625858902931213,
      "learning_rate": 8e-05,
      "loss": 1.8517,
      "step": 4044
    },
    {
      "epoch": 0.8896953700648851,
      "grad_norm": 0.2904675006866455,
      "learning_rate": 8e-05,
      "loss": 1.5672,
      "step": 4045
    },
    {
      "epoch": 0.8899153194765204,
      "grad_norm": 0.28286129236221313,
      "learning_rate": 8e-05,
      "loss": 1.6136,
      "step": 4046
    },
    {
      "epoch": 0.8901352688881558,
      "grad_norm": 0.2881507873535156,
      "learning_rate": 8e-05,
      "loss": 1.6415,
      "step": 4047
    },
    {
      "epoch": 0.890355218299791,
      "grad_norm": 0.2884863018989563,
      "learning_rate": 8e-05,
      "loss": 1.7578,
      "step": 4048
    },
    {
      "epoch": 0.8905751677114264,
      "grad_norm": 0.2926565706729889,
      "learning_rate": 8e-05,
      "loss": 1.6658,
      "step": 4049
    },
    {
      "epoch": 0.8907951171230617,
      "grad_norm": 0.3014330565929413,
      "learning_rate": 8e-05,
      "loss": 1.6047,
      "step": 4050
    },
    {
      "epoch": 0.891015066534697,
      "grad_norm": 0.3148139417171478,
      "learning_rate": 8e-05,
      "loss": 1.4756,
      "step": 4051
    },
    {
      "epoch": 0.8912350159463324,
      "grad_norm": 0.31396764516830444,
      "learning_rate": 8e-05,
      "loss": 1.8031,
      "step": 4052
    },
    {
      "epoch": 0.8914549653579676,
      "grad_norm": 0.2879650890827179,
      "learning_rate": 8e-05,
      "loss": 1.7601,
      "step": 4053
    },
    {
      "epoch": 0.891674914769603,
      "grad_norm": 0.2978127896785736,
      "learning_rate": 8e-05,
      "loss": 1.5957,
      "step": 4054
    },
    {
      "epoch": 0.8918948641812383,
      "grad_norm": 0.28600072860717773,
      "learning_rate": 8e-05,
      "loss": 1.629,
      "step": 4055
    },
    {
      "epoch": 0.8921148135928736,
      "grad_norm": 0.30629444122314453,
      "learning_rate": 8e-05,
      "loss": 1.7707,
      "step": 4056
    },
    {
      "epoch": 0.892334763004509,
      "grad_norm": 0.2934803366661072,
      "learning_rate": 8e-05,
      "loss": 1.669,
      "step": 4057
    },
    {
      "epoch": 0.8925547124161443,
      "grad_norm": 0.2665422558784485,
      "learning_rate": 8e-05,
      "loss": 1.4797,
      "step": 4058
    },
    {
      "epoch": 0.8927746618277796,
      "grad_norm": 0.2825581431388855,
      "learning_rate": 8e-05,
      "loss": 1.662,
      "step": 4059
    },
    {
      "epoch": 0.8929946112394149,
      "grad_norm": 0.2984130382537842,
      "learning_rate": 8e-05,
      "loss": 1.7903,
      "step": 4060
    },
    {
      "epoch": 0.8932145606510502,
      "grad_norm": 0.2899414300918579,
      "learning_rate": 8e-05,
      "loss": 1.6148,
      "step": 4061
    },
    {
      "epoch": 0.8934345100626856,
      "grad_norm": 0.29609423875808716,
      "learning_rate": 8e-05,
      "loss": 1.7082,
      "step": 4062
    },
    {
      "epoch": 0.8936544594743209,
      "grad_norm": 0.2987944185733795,
      "learning_rate": 8e-05,
      "loss": 1.7379,
      "step": 4063
    },
    {
      "epoch": 0.8938744088859563,
      "grad_norm": 0.3441619873046875,
      "learning_rate": 8e-05,
      "loss": 1.5096,
      "step": 4064
    },
    {
      "epoch": 0.8940943582975915,
      "grad_norm": 0.27636975049972534,
      "learning_rate": 8e-05,
      "loss": 1.4661,
      "step": 4065
    },
    {
      "epoch": 0.8943143077092269,
      "grad_norm": 0.2951783835887909,
      "learning_rate": 8e-05,
      "loss": 1.801,
      "step": 4066
    },
    {
      "epoch": 0.8945342571208622,
      "grad_norm": 0.2693262994289398,
      "learning_rate": 8e-05,
      "loss": 1.5644,
      "step": 4067
    },
    {
      "epoch": 0.8947542065324975,
      "grad_norm": 0.283433198928833,
      "learning_rate": 8e-05,
      "loss": 1.6118,
      "step": 4068
    },
    {
      "epoch": 0.8949741559441329,
      "grad_norm": 0.32869231700897217,
      "learning_rate": 8e-05,
      "loss": 1.7622,
      "step": 4069
    },
    {
      "epoch": 0.8951941053557682,
      "grad_norm": 0.29133233428001404,
      "learning_rate": 8e-05,
      "loss": 1.6443,
      "step": 4070
    },
    {
      "epoch": 0.8954140547674035,
      "grad_norm": 0.3477209210395813,
      "learning_rate": 8e-05,
      "loss": 1.9147,
      "step": 4071
    },
    {
      "epoch": 0.8956340041790388,
      "grad_norm": 0.30423182249069214,
      "learning_rate": 8e-05,
      "loss": 1.6524,
      "step": 4072
    },
    {
      "epoch": 0.8958539535906741,
      "grad_norm": 0.3046300411224365,
      "learning_rate": 8e-05,
      "loss": 1.5708,
      "step": 4073
    },
    {
      "epoch": 0.8960739030023095,
      "grad_norm": 0.31017544865608215,
      "learning_rate": 8e-05,
      "loss": 1.8136,
      "step": 4074
    },
    {
      "epoch": 0.8962938524139448,
      "grad_norm": 0.2947177588939667,
      "learning_rate": 8e-05,
      "loss": 1.6837,
      "step": 4075
    },
    {
      "epoch": 0.8965138018255802,
      "grad_norm": 0.3107057213783264,
      "learning_rate": 8e-05,
      "loss": 1.6765,
      "step": 4076
    },
    {
      "epoch": 0.8967337512372154,
      "grad_norm": 0.3040451109409332,
      "learning_rate": 8e-05,
      "loss": 1.6128,
      "step": 4077
    },
    {
      "epoch": 0.8969537006488507,
      "grad_norm": 0.31244924664497375,
      "learning_rate": 8e-05,
      "loss": 1.6391,
      "step": 4078
    },
    {
      "epoch": 0.8971736500604861,
      "grad_norm": 0.2981853485107422,
      "learning_rate": 8e-05,
      "loss": 1.6423,
      "step": 4079
    },
    {
      "epoch": 0.8973935994721214,
      "grad_norm": 0.3044477701187134,
      "learning_rate": 8e-05,
      "loss": 1.6362,
      "step": 4080
    },
    {
      "epoch": 0.8976135488837568,
      "grad_norm": 0.30110621452331543,
      "learning_rate": 8e-05,
      "loss": 1.6289,
      "step": 4081
    },
    {
      "epoch": 0.897833498295392,
      "grad_norm": 0.2834107577800751,
      "learning_rate": 8e-05,
      "loss": 1.593,
      "step": 4082
    },
    {
      "epoch": 0.8980534477070273,
      "grad_norm": 0.34450557827949524,
      "learning_rate": 8e-05,
      "loss": 1.7081,
      "step": 4083
    },
    {
      "epoch": 0.8982733971186627,
      "grad_norm": 0.27985870838165283,
      "learning_rate": 8e-05,
      "loss": 1.6028,
      "step": 4084
    },
    {
      "epoch": 0.898493346530298,
      "grad_norm": 0.31170904636383057,
      "learning_rate": 8e-05,
      "loss": 1.67,
      "step": 4085
    },
    {
      "epoch": 0.8987132959419334,
      "grad_norm": 0.30392032861709595,
      "learning_rate": 8e-05,
      "loss": 1.6134,
      "step": 4086
    },
    {
      "epoch": 0.8989332453535687,
      "grad_norm": 0.30208954215049744,
      "learning_rate": 8e-05,
      "loss": 1.7703,
      "step": 4087
    },
    {
      "epoch": 0.8991531947652041,
      "grad_norm": 0.2879715859889984,
      "learning_rate": 8e-05,
      "loss": 1.6479,
      "step": 4088
    },
    {
      "epoch": 0.8993731441768393,
      "grad_norm": 0.3037298619747162,
      "learning_rate": 8e-05,
      "loss": 1.5704,
      "step": 4089
    },
    {
      "epoch": 0.8995930935884746,
      "grad_norm": 0.3177938163280487,
      "learning_rate": 8e-05,
      "loss": 1.6422,
      "step": 4090
    },
    {
      "epoch": 0.89981304300011,
      "grad_norm": 0.28877830505371094,
      "learning_rate": 8e-05,
      "loss": 1.5779,
      "step": 4091
    },
    {
      "epoch": 0.9000329924117453,
      "grad_norm": 0.2998355031013489,
      "learning_rate": 8e-05,
      "loss": 1.6163,
      "step": 4092
    },
    {
      "epoch": 0.9002529418233807,
      "grad_norm": 0.28364381194114685,
      "learning_rate": 8e-05,
      "loss": 1.6757,
      "step": 4093
    },
    {
      "epoch": 0.900472891235016,
      "grad_norm": 0.31235194206237793,
      "learning_rate": 8e-05,
      "loss": 1.7467,
      "step": 4094
    },
    {
      "epoch": 0.9006928406466512,
      "grad_norm": 0.2888253927230835,
      "learning_rate": 8e-05,
      "loss": 1.6793,
      "step": 4095
    },
    {
      "epoch": 0.9009127900582866,
      "grad_norm": 0.3084707260131836,
      "learning_rate": 8e-05,
      "loss": 1.7183,
      "step": 4096
    },
    {
      "epoch": 0.9011327394699219,
      "grad_norm": 0.2737601697444916,
      "learning_rate": 8e-05,
      "loss": 1.4579,
      "step": 4097
    },
    {
      "epoch": 0.9013526888815573,
      "grad_norm": 0.2910381257534027,
      "learning_rate": 8e-05,
      "loss": 1.6289,
      "step": 4098
    },
    {
      "epoch": 0.9015726382931926,
      "grad_norm": 0.3083101809024811,
      "learning_rate": 8e-05,
      "loss": 1.8007,
      "step": 4099
    },
    {
      "epoch": 0.9017925877048278,
      "grad_norm": 0.2874706983566284,
      "learning_rate": 8e-05,
      "loss": 1.7087,
      "step": 4100
    },
    {
      "epoch": 0.9020125371164632,
      "grad_norm": 0.27825212478637695,
      "learning_rate": 8e-05,
      "loss": 1.5046,
      "step": 4101
    },
    {
      "epoch": 0.9022324865280985,
      "grad_norm": 0.30026623606681824,
      "learning_rate": 8e-05,
      "loss": 1.7011,
      "step": 4102
    },
    {
      "epoch": 0.9024524359397339,
      "grad_norm": 0.3117727041244507,
      "learning_rate": 8e-05,
      "loss": 1.8456,
      "step": 4103
    },
    {
      "epoch": 0.9026723853513692,
      "grad_norm": 0.2942452132701874,
      "learning_rate": 8e-05,
      "loss": 1.6479,
      "step": 4104
    },
    {
      "epoch": 0.9028923347630045,
      "grad_norm": 0.29466378688812256,
      "learning_rate": 8e-05,
      "loss": 1.6942,
      "step": 4105
    },
    {
      "epoch": 0.9031122841746398,
      "grad_norm": 0.2969743013381958,
      "learning_rate": 8e-05,
      "loss": 1.6088,
      "step": 4106
    },
    {
      "epoch": 0.9033322335862751,
      "grad_norm": 0.29155898094177246,
      "learning_rate": 8e-05,
      "loss": 1.7225,
      "step": 4107
    },
    {
      "epoch": 0.9035521829979105,
      "grad_norm": 0.27798759937286377,
      "learning_rate": 8e-05,
      "loss": 1.5386,
      "step": 4108
    },
    {
      "epoch": 0.9037721324095458,
      "grad_norm": 0.29623040556907654,
      "learning_rate": 8e-05,
      "loss": 1.7057,
      "step": 4109
    },
    {
      "epoch": 0.9039920818211812,
      "grad_norm": 0.28695133328437805,
      "learning_rate": 8e-05,
      "loss": 1.6513,
      "step": 4110
    },
    {
      "epoch": 0.9042120312328165,
      "grad_norm": 0.283668577671051,
      "learning_rate": 8e-05,
      "loss": 1.8331,
      "step": 4111
    },
    {
      "epoch": 0.9044319806444517,
      "grad_norm": 0.3282906115055084,
      "learning_rate": 8e-05,
      "loss": 1.7565,
      "step": 4112
    },
    {
      "epoch": 0.9046519300560871,
      "grad_norm": 0.2832440435886383,
      "learning_rate": 8e-05,
      "loss": 1.4466,
      "step": 4113
    },
    {
      "epoch": 0.9048718794677224,
      "grad_norm": 0.2924807369709015,
      "learning_rate": 8e-05,
      "loss": 1.6251,
      "step": 4114
    },
    {
      "epoch": 0.9050918288793578,
      "grad_norm": 0.30800655484199524,
      "learning_rate": 8e-05,
      "loss": 1.6581,
      "step": 4115
    },
    {
      "epoch": 0.9053117782909931,
      "grad_norm": 0.2765842080116272,
      "learning_rate": 8e-05,
      "loss": 1.6272,
      "step": 4116
    },
    {
      "epoch": 0.9055317277026284,
      "grad_norm": 0.29153522849082947,
      "learning_rate": 8e-05,
      "loss": 1.5286,
      "step": 4117
    },
    {
      "epoch": 0.9057516771142637,
      "grad_norm": 0.2986176013946533,
      "learning_rate": 8e-05,
      "loss": 1.5265,
      "step": 4118
    },
    {
      "epoch": 0.905971626525899,
      "grad_norm": 0.28605547547340393,
      "learning_rate": 8e-05,
      "loss": 1.6859,
      "step": 4119
    },
    {
      "epoch": 0.9061915759375344,
      "grad_norm": 0.28253173828125,
      "learning_rate": 8e-05,
      "loss": 1.6302,
      "step": 4120
    },
    {
      "epoch": 0.9064115253491697,
      "grad_norm": 0.2970622479915619,
      "learning_rate": 8e-05,
      "loss": 1.7511,
      "step": 4121
    },
    {
      "epoch": 0.906631474760805,
      "grad_norm": 0.2855907082557678,
      "learning_rate": 8e-05,
      "loss": 1.6366,
      "step": 4122
    },
    {
      "epoch": 0.9068514241724404,
      "grad_norm": 0.29107728600502014,
      "learning_rate": 8e-05,
      "loss": 1.7313,
      "step": 4123
    },
    {
      "epoch": 0.9070713735840756,
      "grad_norm": 0.2963238060474396,
      "learning_rate": 8e-05,
      "loss": 1.7331,
      "step": 4124
    },
    {
      "epoch": 0.907291322995711,
      "grad_norm": 0.27756524085998535,
      "learning_rate": 8e-05,
      "loss": 1.6023,
      "step": 4125
    },
    {
      "epoch": 0.9075112724073463,
      "grad_norm": 0.30239957571029663,
      "learning_rate": 8e-05,
      "loss": 1.7414,
      "step": 4126
    },
    {
      "epoch": 0.9077312218189816,
      "grad_norm": 0.28329378366470337,
      "learning_rate": 8e-05,
      "loss": 1.6604,
      "step": 4127
    },
    {
      "epoch": 0.907951171230617,
      "grad_norm": 0.2846319377422333,
      "learning_rate": 8e-05,
      "loss": 1.5306,
      "step": 4128
    },
    {
      "epoch": 0.9081711206422522,
      "grad_norm": 0.294981986284256,
      "learning_rate": 8e-05,
      "loss": 1.6668,
      "step": 4129
    },
    {
      "epoch": 0.9083910700538876,
      "grad_norm": 0.30630192160606384,
      "learning_rate": 8e-05,
      "loss": 1.789,
      "step": 4130
    },
    {
      "epoch": 0.9086110194655229,
      "grad_norm": 0.29193127155303955,
      "learning_rate": 8e-05,
      "loss": 1.7436,
      "step": 4131
    },
    {
      "epoch": 0.9088309688771583,
      "grad_norm": 0.28358370065689087,
      "learning_rate": 8e-05,
      "loss": 1.6301,
      "step": 4132
    },
    {
      "epoch": 0.9090509182887936,
      "grad_norm": 0.29850831627845764,
      "learning_rate": 8e-05,
      "loss": 1.7101,
      "step": 4133
    },
    {
      "epoch": 0.9092708677004289,
      "grad_norm": 0.3096507787704468,
      "learning_rate": 8e-05,
      "loss": 1.7025,
      "step": 4134
    },
    {
      "epoch": 0.9094908171120643,
      "grad_norm": 0.3215219974517822,
      "learning_rate": 8e-05,
      "loss": 1.5914,
      "step": 4135
    },
    {
      "epoch": 0.9097107665236995,
      "grad_norm": 0.3059280812740326,
      "learning_rate": 8e-05,
      "loss": 1.6822,
      "step": 4136
    },
    {
      "epoch": 0.9099307159353349,
      "grad_norm": 0.3068895936012268,
      "learning_rate": 8e-05,
      "loss": 1.63,
      "step": 4137
    },
    {
      "epoch": 0.9101506653469702,
      "grad_norm": 0.2992387115955353,
      "learning_rate": 8e-05,
      "loss": 1.7165,
      "step": 4138
    },
    {
      "epoch": 0.9103706147586055,
      "grad_norm": 0.29406991600990295,
      "learning_rate": 8e-05,
      "loss": 1.65,
      "step": 4139
    },
    {
      "epoch": 0.9105905641702409,
      "grad_norm": 0.2930237650871277,
      "learning_rate": 8e-05,
      "loss": 1.7292,
      "step": 4140
    },
    {
      "epoch": 0.9108105135818761,
      "grad_norm": 0.2610776722431183,
      "learning_rate": 8e-05,
      "loss": 1.4707,
      "step": 4141
    },
    {
      "epoch": 0.9110304629935115,
      "grad_norm": 0.2843222916126251,
      "learning_rate": 8e-05,
      "loss": 1.5374,
      "step": 4142
    },
    {
      "epoch": 0.9112504124051468,
      "grad_norm": 0.30256927013397217,
      "learning_rate": 8e-05,
      "loss": 1.7271,
      "step": 4143
    },
    {
      "epoch": 0.9114703618167821,
      "grad_norm": 0.29254963994026184,
      "learning_rate": 8e-05,
      "loss": 1.656,
      "step": 4144
    },
    {
      "epoch": 0.9116903112284175,
      "grad_norm": 0.2959951162338257,
      "learning_rate": 8e-05,
      "loss": 1.5848,
      "step": 4145
    },
    {
      "epoch": 0.9119102606400528,
      "grad_norm": 0.3133392632007599,
      "learning_rate": 8e-05,
      "loss": 1.6898,
      "step": 4146
    },
    {
      "epoch": 0.9121302100516882,
      "grad_norm": 0.31604188680648804,
      "learning_rate": 8e-05,
      "loss": 1.727,
      "step": 4147
    },
    {
      "epoch": 0.9123501594633234,
      "grad_norm": 0.28543493151664734,
      "learning_rate": 8e-05,
      "loss": 1.6329,
      "step": 4148
    },
    {
      "epoch": 0.9125701088749587,
      "grad_norm": 0.28241634368896484,
      "learning_rate": 8e-05,
      "loss": 1.5312,
      "step": 4149
    },
    {
      "epoch": 0.9127900582865941,
      "grad_norm": 0.2962937653064728,
      "learning_rate": 8e-05,
      "loss": 1.6362,
      "step": 4150
    },
    {
      "epoch": 0.9130100076982294,
      "grad_norm": 0.299084335565567,
      "learning_rate": 8e-05,
      "loss": 1.6091,
      "step": 4151
    },
    {
      "epoch": 0.9132299571098648,
      "grad_norm": 0.28923937678337097,
      "learning_rate": 8e-05,
      "loss": 1.6693,
      "step": 4152
    },
    {
      "epoch": 0.9134499065215,
      "grad_norm": 0.2867174446582794,
      "learning_rate": 8e-05,
      "loss": 1.5203,
      "step": 4153
    },
    {
      "epoch": 0.9136698559331354,
      "grad_norm": 0.28517308831214905,
      "learning_rate": 8e-05,
      "loss": 1.6159,
      "step": 4154
    },
    {
      "epoch": 0.9138898053447707,
      "grad_norm": 0.29169774055480957,
      "learning_rate": 8e-05,
      "loss": 1.707,
      "step": 4155
    },
    {
      "epoch": 0.914109754756406,
      "grad_norm": 0.28574511408805847,
      "learning_rate": 8e-05,
      "loss": 1.6871,
      "step": 4156
    },
    {
      "epoch": 0.9143297041680414,
      "grad_norm": 0.29836469888687134,
      "learning_rate": 8e-05,
      "loss": 1.6859,
      "step": 4157
    },
    {
      "epoch": 0.9145496535796767,
      "grad_norm": 0.32184654474258423,
      "learning_rate": 8e-05,
      "loss": 1.667,
      "step": 4158
    },
    {
      "epoch": 0.914769602991312,
      "grad_norm": 0.28499841690063477,
      "learning_rate": 8e-05,
      "loss": 1.6998,
      "step": 4159
    },
    {
      "epoch": 0.9149895524029473,
      "grad_norm": 0.28232523798942566,
      "learning_rate": 8e-05,
      "loss": 1.5786,
      "step": 4160
    },
    {
      "epoch": 0.9152095018145826,
      "grad_norm": 0.29549431800842285,
      "learning_rate": 8e-05,
      "loss": 1.6957,
      "step": 4161
    },
    {
      "epoch": 0.915429451226218,
      "grad_norm": 0.30055925250053406,
      "learning_rate": 8e-05,
      "loss": 1.6295,
      "step": 4162
    },
    {
      "epoch": 0.9156494006378533,
      "grad_norm": 0.30925673246383667,
      "learning_rate": 8e-05,
      "loss": 1.5685,
      "step": 4163
    },
    {
      "epoch": 0.9158693500494887,
      "grad_norm": 0.30817949771881104,
      "learning_rate": 8e-05,
      "loss": 1.732,
      "step": 4164
    },
    {
      "epoch": 0.9160892994611239,
      "grad_norm": 0.28672513365745544,
      "learning_rate": 8e-05,
      "loss": 1.6787,
      "step": 4165
    },
    {
      "epoch": 0.9163092488727592,
      "grad_norm": 0.30071011185646057,
      "learning_rate": 8e-05,
      "loss": 1.7844,
      "step": 4166
    },
    {
      "epoch": 0.9165291982843946,
      "grad_norm": 0.292388916015625,
      "learning_rate": 8e-05,
      "loss": 1.579,
      "step": 4167
    },
    {
      "epoch": 0.9167491476960299,
      "grad_norm": 0.31311121582984924,
      "learning_rate": 8e-05,
      "loss": 1.6822,
      "step": 4168
    },
    {
      "epoch": 0.9169690971076653,
      "grad_norm": 0.31428346037864685,
      "learning_rate": 8e-05,
      "loss": 1.7232,
      "step": 4169
    },
    {
      "epoch": 0.9171890465193006,
      "grad_norm": 0.30659395456314087,
      "learning_rate": 8e-05,
      "loss": 1.7332,
      "step": 4170
    },
    {
      "epoch": 0.9174089959309358,
      "grad_norm": 0.32651838660240173,
      "learning_rate": 8e-05,
      "loss": 1.7149,
      "step": 4171
    },
    {
      "epoch": 0.9176289453425712,
      "grad_norm": 0.3228572607040405,
      "learning_rate": 8e-05,
      "loss": 1.6848,
      "step": 4172
    },
    {
      "epoch": 0.9178488947542065,
      "grad_norm": 0.2901211380958557,
      "learning_rate": 8e-05,
      "loss": 1.7337,
      "step": 4173
    },
    {
      "epoch": 0.9180688441658419,
      "grad_norm": 0.2871779799461365,
      "learning_rate": 8e-05,
      "loss": 1.6134,
      "step": 4174
    },
    {
      "epoch": 0.9182887935774772,
      "grad_norm": 0.30921611189842224,
      "learning_rate": 8e-05,
      "loss": 1.7896,
      "step": 4175
    },
    {
      "epoch": 0.9185087429891126,
      "grad_norm": 0.2913828492164612,
      "learning_rate": 8e-05,
      "loss": 1.7384,
      "step": 4176
    },
    {
      "epoch": 0.9187286924007478,
      "grad_norm": 0.31601426005363464,
      "learning_rate": 8e-05,
      "loss": 1.6391,
      "step": 4177
    },
    {
      "epoch": 0.9189486418123831,
      "grad_norm": 0.2996780276298523,
      "learning_rate": 8e-05,
      "loss": 1.5581,
      "step": 4178
    },
    {
      "epoch": 0.9191685912240185,
      "grad_norm": 0.3079301416873932,
      "learning_rate": 8e-05,
      "loss": 1.5813,
      "step": 4179
    },
    {
      "epoch": 0.9193885406356538,
      "grad_norm": 0.28514519333839417,
      "learning_rate": 8e-05,
      "loss": 1.702,
      "step": 4180
    },
    {
      "epoch": 0.9196084900472892,
      "grad_norm": 0.29393357038497925,
      "learning_rate": 8e-05,
      "loss": 1.6957,
      "step": 4181
    },
    {
      "epoch": 0.9198284394589245,
      "grad_norm": 0.2881946563720703,
      "learning_rate": 8e-05,
      "loss": 1.6014,
      "step": 4182
    },
    {
      "epoch": 0.9200483888705597,
      "grad_norm": 0.2976199686527252,
      "learning_rate": 8e-05,
      "loss": 1.7531,
      "step": 4183
    },
    {
      "epoch": 0.9202683382821951,
      "grad_norm": 0.3015999495983124,
      "learning_rate": 8e-05,
      "loss": 1.6589,
      "step": 4184
    },
    {
      "epoch": 0.9204882876938304,
      "grad_norm": 0.2981117367744446,
      "learning_rate": 8e-05,
      "loss": 1.6802,
      "step": 4185
    },
    {
      "epoch": 0.9207082371054658,
      "grad_norm": 0.29663896560668945,
      "learning_rate": 8e-05,
      "loss": 1.7327,
      "step": 4186
    },
    {
      "epoch": 0.9209281865171011,
      "grad_norm": 0.28317975997924805,
      "learning_rate": 8e-05,
      "loss": 1.6129,
      "step": 4187
    },
    {
      "epoch": 0.9211481359287363,
      "grad_norm": 0.27301955223083496,
      "learning_rate": 8e-05,
      "loss": 1.3794,
      "step": 4188
    },
    {
      "epoch": 0.9213680853403717,
      "grad_norm": 0.28590822219848633,
      "learning_rate": 8e-05,
      "loss": 1.6962,
      "step": 4189
    },
    {
      "epoch": 0.921588034752007,
      "grad_norm": 0.29811057448387146,
      "learning_rate": 8e-05,
      "loss": 1.8319,
      "step": 4190
    },
    {
      "epoch": 0.9218079841636424,
      "grad_norm": 0.2986351549625397,
      "learning_rate": 8e-05,
      "loss": 1.6403,
      "step": 4191
    },
    {
      "epoch": 0.9220279335752777,
      "grad_norm": 0.29516395926475525,
      "learning_rate": 8e-05,
      "loss": 1.5754,
      "step": 4192
    },
    {
      "epoch": 0.922247882986913,
      "grad_norm": 0.29312869906425476,
      "learning_rate": 8e-05,
      "loss": 1.6533,
      "step": 4193
    },
    {
      "epoch": 0.9224678323985483,
      "grad_norm": 0.28571614623069763,
      "learning_rate": 8e-05,
      "loss": 1.6711,
      "step": 4194
    },
    {
      "epoch": 0.9226877818101836,
      "grad_norm": 0.28665637969970703,
      "learning_rate": 8e-05,
      "loss": 1.6586,
      "step": 4195
    },
    {
      "epoch": 0.922907731221819,
      "grad_norm": 0.3119889795780182,
      "learning_rate": 8e-05,
      "loss": 1.8049,
      "step": 4196
    },
    {
      "epoch": 0.9231276806334543,
      "grad_norm": 0.2987707853317261,
      "learning_rate": 8e-05,
      "loss": 1.8385,
      "step": 4197
    },
    {
      "epoch": 0.9233476300450897,
      "grad_norm": 0.2872779071331024,
      "learning_rate": 8e-05,
      "loss": 1.7374,
      "step": 4198
    },
    {
      "epoch": 0.923567579456725,
      "grad_norm": 0.30723926424980164,
      "learning_rate": 8e-05,
      "loss": 1.7982,
      "step": 4199
    },
    {
      "epoch": 0.9237875288683602,
      "grad_norm": 0.2957054376602173,
      "learning_rate": 8e-05,
      "loss": 1.712,
      "step": 4200
    },
    {
      "epoch": 0.9240074782799956,
      "grad_norm": 0.2941366136074066,
      "learning_rate": 8e-05,
      "loss": 1.6782,
      "step": 4201
    },
    {
      "epoch": 0.9242274276916309,
      "grad_norm": 0.3279740512371063,
      "learning_rate": 8e-05,
      "loss": 1.7321,
      "step": 4202
    },
    {
      "epoch": 0.9244473771032663,
      "grad_norm": 0.3030930459499359,
      "learning_rate": 8e-05,
      "loss": 1.6227,
      "step": 4203
    },
    {
      "epoch": 0.9246673265149016,
      "grad_norm": 0.29185232520103455,
      "learning_rate": 8e-05,
      "loss": 1.6529,
      "step": 4204
    },
    {
      "epoch": 0.9248872759265369,
      "grad_norm": 0.31919175386428833,
      "learning_rate": 8e-05,
      "loss": 1.6879,
      "step": 4205
    },
    {
      "epoch": 0.9251072253381722,
      "grad_norm": 0.2989010214805603,
      "learning_rate": 8e-05,
      "loss": 1.7958,
      "step": 4206
    },
    {
      "epoch": 0.9253271747498075,
      "grad_norm": 0.28767868876457214,
      "learning_rate": 8e-05,
      "loss": 1.6155,
      "step": 4207
    },
    {
      "epoch": 0.9255471241614429,
      "grad_norm": 0.30420902371406555,
      "learning_rate": 8e-05,
      "loss": 1.74,
      "step": 4208
    },
    {
      "epoch": 0.9257670735730782,
      "grad_norm": 0.28389647603034973,
      "learning_rate": 8e-05,
      "loss": 1.6953,
      "step": 4209
    },
    {
      "epoch": 0.9259870229847135,
      "grad_norm": 0.2950130105018616,
      "learning_rate": 8e-05,
      "loss": 1.8119,
      "step": 4210
    },
    {
      "epoch": 0.9262069723963489,
      "grad_norm": 0.28202083706855774,
      "learning_rate": 8e-05,
      "loss": 1.6032,
      "step": 4211
    },
    {
      "epoch": 0.9264269218079841,
      "grad_norm": 0.2873939871788025,
      "learning_rate": 8e-05,
      "loss": 1.6406,
      "step": 4212
    },
    {
      "epoch": 0.9266468712196195,
      "grad_norm": 0.29845505952835083,
      "learning_rate": 8e-05,
      "loss": 1.7685,
      "step": 4213
    },
    {
      "epoch": 0.9268668206312548,
      "grad_norm": 0.3224455714225769,
      "learning_rate": 8e-05,
      "loss": 1.8116,
      "step": 4214
    },
    {
      "epoch": 0.9270867700428901,
      "grad_norm": 0.2965445816516876,
      "learning_rate": 8e-05,
      "loss": 1.6149,
      "step": 4215
    },
    {
      "epoch": 0.9273067194545255,
      "grad_norm": 0.2944619357585907,
      "learning_rate": 8e-05,
      "loss": 1.5827,
      "step": 4216
    },
    {
      "epoch": 0.9275266688661608,
      "grad_norm": 0.29647067189216614,
      "learning_rate": 8e-05,
      "loss": 1.6907,
      "step": 4217
    },
    {
      "epoch": 0.9277466182777961,
      "grad_norm": 0.27989983558654785,
      "learning_rate": 8e-05,
      "loss": 1.5422,
      "step": 4218
    },
    {
      "epoch": 0.9279665676894314,
      "grad_norm": 0.30345508456230164,
      "learning_rate": 8e-05,
      "loss": 1.6439,
      "step": 4219
    },
    {
      "epoch": 0.9281865171010667,
      "grad_norm": 0.29629063606262207,
      "learning_rate": 8e-05,
      "loss": 1.5947,
      "step": 4220
    },
    {
      "epoch": 0.9284064665127021,
      "grad_norm": 0.29597344994544983,
      "learning_rate": 8e-05,
      "loss": 1.6203,
      "step": 4221
    },
    {
      "epoch": 0.9286264159243374,
      "grad_norm": 0.2957248389720917,
      "learning_rate": 8e-05,
      "loss": 1.5987,
      "step": 4222
    },
    {
      "epoch": 0.9288463653359728,
      "grad_norm": 0.2960350513458252,
      "learning_rate": 8e-05,
      "loss": 1.5776,
      "step": 4223
    },
    {
      "epoch": 0.929066314747608,
      "grad_norm": 0.29710423946380615,
      "learning_rate": 8e-05,
      "loss": 1.5518,
      "step": 4224
    },
    {
      "epoch": 0.9292862641592434,
      "grad_norm": 0.29832521080970764,
      "learning_rate": 8e-05,
      "loss": 1.6414,
      "step": 4225
    },
    {
      "epoch": 0.9295062135708787,
      "grad_norm": 0.26050087809562683,
      "learning_rate": 8e-05,
      "loss": 1.3936,
      "step": 4226
    },
    {
      "epoch": 0.929726162982514,
      "grad_norm": 0.3182241916656494,
      "learning_rate": 8e-05,
      "loss": 1.8863,
      "step": 4227
    },
    {
      "epoch": 0.9299461123941494,
      "grad_norm": 0.2917816936969757,
      "learning_rate": 8e-05,
      "loss": 1.7444,
      "step": 4228
    },
    {
      "epoch": 0.9301660618057846,
      "grad_norm": 0.30499497056007385,
      "learning_rate": 8e-05,
      "loss": 1.6962,
      "step": 4229
    },
    {
      "epoch": 0.93038601121742,
      "grad_norm": 0.30994856357574463,
      "learning_rate": 8e-05,
      "loss": 1.8838,
      "step": 4230
    },
    {
      "epoch": 0.9306059606290553,
      "grad_norm": 0.3050210177898407,
      "learning_rate": 8e-05,
      "loss": 1.6575,
      "step": 4231
    },
    {
      "epoch": 0.9308259100406906,
      "grad_norm": 0.2949892282485962,
      "learning_rate": 8e-05,
      "loss": 1.7391,
      "step": 4232
    },
    {
      "epoch": 0.931045859452326,
      "grad_norm": 0.2921696603298187,
      "learning_rate": 8e-05,
      "loss": 1.8789,
      "step": 4233
    },
    {
      "epoch": 0.9312658088639613,
      "grad_norm": 0.2918970584869385,
      "learning_rate": 8e-05,
      "loss": 1.5761,
      "step": 4234
    },
    {
      "epoch": 0.9314857582755967,
      "grad_norm": 0.2940721809864044,
      "learning_rate": 8e-05,
      "loss": 1.6322,
      "step": 4235
    },
    {
      "epoch": 0.9317057076872319,
      "grad_norm": 0.27402326464653015,
      "learning_rate": 8e-05,
      "loss": 1.5636,
      "step": 4236
    },
    {
      "epoch": 0.9319256570988672,
      "grad_norm": 0.34060537815093994,
      "learning_rate": 8e-05,
      "loss": 1.693,
      "step": 4237
    },
    {
      "epoch": 0.9321456065105026,
      "grad_norm": 0.2948283851146698,
      "learning_rate": 8e-05,
      "loss": 1.7477,
      "step": 4238
    },
    {
      "epoch": 0.9323655559221379,
      "grad_norm": 0.29711946845054626,
      "learning_rate": 8e-05,
      "loss": 1.654,
      "step": 4239
    },
    {
      "epoch": 0.9325855053337733,
      "grad_norm": 0.28953954577445984,
      "learning_rate": 8e-05,
      "loss": 1.6674,
      "step": 4240
    },
    {
      "epoch": 0.9328054547454085,
      "grad_norm": 0.30088186264038086,
      "learning_rate": 8e-05,
      "loss": 1.5417,
      "step": 4241
    },
    {
      "epoch": 0.9330254041570438,
      "grad_norm": 0.3125405013561249,
      "learning_rate": 8e-05,
      "loss": 1.7275,
      "step": 4242
    },
    {
      "epoch": 0.9332453535686792,
      "grad_norm": 0.2886843979358673,
      "learning_rate": 8e-05,
      "loss": 1.6494,
      "step": 4243
    },
    {
      "epoch": 0.9334653029803145,
      "grad_norm": 0.30982518196105957,
      "learning_rate": 8e-05,
      "loss": 1.8119,
      "step": 4244
    },
    {
      "epoch": 0.9336852523919499,
      "grad_norm": 0.27964144945144653,
      "learning_rate": 8e-05,
      "loss": 1.6068,
      "step": 4245
    },
    {
      "epoch": 0.9339052018035852,
      "grad_norm": 0.29574427008628845,
      "learning_rate": 8e-05,
      "loss": 1.7418,
      "step": 4246
    },
    {
      "epoch": 0.9341251512152206,
      "grad_norm": 0.3052821755409241,
      "learning_rate": 8e-05,
      "loss": 1.6943,
      "step": 4247
    },
    {
      "epoch": 0.9343451006268558,
      "grad_norm": 0.28099381923675537,
      "learning_rate": 8e-05,
      "loss": 1.5774,
      "step": 4248
    },
    {
      "epoch": 0.9345650500384911,
      "grad_norm": 0.2942703664302826,
      "learning_rate": 8e-05,
      "loss": 1.7297,
      "step": 4249
    },
    {
      "epoch": 0.9347849994501265,
      "grad_norm": 0.2937106490135193,
      "learning_rate": 8e-05,
      "loss": 1.6643,
      "step": 4250
    },
    {
      "epoch": 0.9350049488617618,
      "grad_norm": 0.30792105197906494,
      "learning_rate": 8e-05,
      "loss": 1.8709,
      "step": 4251
    },
    {
      "epoch": 0.9352248982733972,
      "grad_norm": 0.2843385338783264,
      "learning_rate": 8e-05,
      "loss": 1.6057,
      "step": 4252
    },
    {
      "epoch": 0.9354448476850324,
      "grad_norm": 0.30580613017082214,
      "learning_rate": 8e-05,
      "loss": 1.5744,
      "step": 4253
    },
    {
      "epoch": 0.9356647970966677,
      "grad_norm": 0.29129788279533386,
      "learning_rate": 8e-05,
      "loss": 1.6922,
      "step": 4254
    },
    {
      "epoch": 0.9358847465083031,
      "grad_norm": 0.31255224347114563,
      "learning_rate": 8e-05,
      "loss": 1.8804,
      "step": 4255
    },
    {
      "epoch": 0.9361046959199384,
      "grad_norm": 0.30788132548332214,
      "learning_rate": 8e-05,
      "loss": 1.7001,
      "step": 4256
    },
    {
      "epoch": 0.9363246453315738,
      "grad_norm": 0.2987869083881378,
      "learning_rate": 8e-05,
      "loss": 1.6763,
      "step": 4257
    },
    {
      "epoch": 0.9365445947432091,
      "grad_norm": 0.28915441036224365,
      "learning_rate": 8e-05,
      "loss": 1.5499,
      "step": 4258
    },
    {
      "epoch": 0.9367645441548443,
      "grad_norm": 0.3005789518356323,
      "learning_rate": 8e-05,
      "loss": 1.7255,
      "step": 4259
    },
    {
      "epoch": 0.9369844935664797,
      "grad_norm": 0.299762099981308,
      "learning_rate": 8e-05,
      "loss": 1.6768,
      "step": 4260
    },
    {
      "epoch": 0.937204442978115,
      "grad_norm": 0.3155803680419922,
      "learning_rate": 8e-05,
      "loss": 1.684,
      "step": 4261
    },
    {
      "epoch": 0.9374243923897504,
      "grad_norm": 0.2879777252674103,
      "learning_rate": 8e-05,
      "loss": 1.5999,
      "step": 4262
    },
    {
      "epoch": 0.9376443418013857,
      "grad_norm": 0.304275244474411,
      "learning_rate": 8e-05,
      "loss": 1.6666,
      "step": 4263
    },
    {
      "epoch": 0.937864291213021,
      "grad_norm": 0.30361175537109375,
      "learning_rate": 8e-05,
      "loss": 1.6459,
      "step": 4264
    },
    {
      "epoch": 0.9380842406246563,
      "grad_norm": 0.2961748242378235,
      "learning_rate": 8e-05,
      "loss": 1.6006,
      "step": 4265
    },
    {
      "epoch": 0.9383041900362916,
      "grad_norm": 0.30298399925231934,
      "learning_rate": 8e-05,
      "loss": 1.7134,
      "step": 4266
    },
    {
      "epoch": 0.938524139447927,
      "grad_norm": 0.2866550087928772,
      "learning_rate": 8e-05,
      "loss": 1.6643,
      "step": 4267
    },
    {
      "epoch": 0.9387440888595623,
      "grad_norm": 0.3346495032310486,
      "learning_rate": 8e-05,
      "loss": 1.6863,
      "step": 4268
    },
    {
      "epoch": 0.9389640382711977,
      "grad_norm": 0.2898212671279907,
      "learning_rate": 8e-05,
      "loss": 1.5883,
      "step": 4269
    },
    {
      "epoch": 0.939183987682833,
      "grad_norm": 0.29734665155410767,
      "learning_rate": 8e-05,
      "loss": 1.7226,
      "step": 4270
    },
    {
      "epoch": 0.9394039370944682,
      "grad_norm": 0.3220420181751251,
      "learning_rate": 8e-05,
      "loss": 1.663,
      "step": 4271
    },
    {
      "epoch": 0.9396238865061036,
      "grad_norm": 0.30133485794067383,
      "learning_rate": 8e-05,
      "loss": 1.4772,
      "step": 4272
    },
    {
      "epoch": 0.9398438359177389,
      "grad_norm": 0.28944581747055054,
      "learning_rate": 8e-05,
      "loss": 1.5403,
      "step": 4273
    },
    {
      "epoch": 0.9400637853293743,
      "grad_norm": 0.3234364092350006,
      "learning_rate": 8e-05,
      "loss": 1.8958,
      "step": 4274
    },
    {
      "epoch": 0.9402837347410096,
      "grad_norm": 0.3140423595905304,
      "learning_rate": 8e-05,
      "loss": 1.6434,
      "step": 4275
    },
    {
      "epoch": 0.9405036841526448,
      "grad_norm": 0.31199538707733154,
      "learning_rate": 8e-05,
      "loss": 1.7016,
      "step": 4276
    },
    {
      "epoch": 0.9407236335642802,
      "grad_norm": 0.31812262535095215,
      "learning_rate": 8e-05,
      "loss": 1.6401,
      "step": 4277
    },
    {
      "epoch": 0.9409435829759155,
      "grad_norm": 0.30784544348716736,
      "learning_rate": 8e-05,
      "loss": 1.745,
      "step": 4278
    },
    {
      "epoch": 0.9411635323875509,
      "grad_norm": 0.2914595901966095,
      "learning_rate": 8e-05,
      "loss": 1.6515,
      "step": 4279
    },
    {
      "epoch": 0.9413834817991862,
      "grad_norm": 0.3110421597957611,
      "learning_rate": 8e-05,
      "loss": 1.5437,
      "step": 4280
    },
    {
      "epoch": 0.9416034312108215,
      "grad_norm": 0.32427704334259033,
      "learning_rate": 8e-05,
      "loss": 1.7521,
      "step": 4281
    },
    {
      "epoch": 0.9418233806224569,
      "grad_norm": 0.3069719672203064,
      "learning_rate": 8e-05,
      "loss": 1.7473,
      "step": 4282
    },
    {
      "epoch": 0.9420433300340921,
      "grad_norm": 0.3008507192134857,
      "learning_rate": 8e-05,
      "loss": 1.6536,
      "step": 4283
    },
    {
      "epoch": 0.9422632794457275,
      "grad_norm": 0.28315114974975586,
      "learning_rate": 8e-05,
      "loss": 1.5788,
      "step": 4284
    },
    {
      "epoch": 0.9424832288573628,
      "grad_norm": 0.2987407445907593,
      "learning_rate": 8e-05,
      "loss": 1.7148,
      "step": 4285
    },
    {
      "epoch": 0.9427031782689981,
      "grad_norm": 0.2861090302467346,
      "learning_rate": 8e-05,
      "loss": 1.6858,
      "step": 4286
    },
    {
      "epoch": 0.9429231276806335,
      "grad_norm": 0.297574907541275,
      "learning_rate": 8e-05,
      "loss": 1.7461,
      "step": 4287
    },
    {
      "epoch": 0.9431430770922687,
      "grad_norm": 0.28463122248649597,
      "learning_rate": 8e-05,
      "loss": 1.6208,
      "step": 4288
    },
    {
      "epoch": 0.9433630265039041,
      "grad_norm": 0.3073120415210724,
      "learning_rate": 8e-05,
      "loss": 1.6601,
      "step": 4289
    },
    {
      "epoch": 0.9435829759155394,
      "grad_norm": 0.2836720049381256,
      "learning_rate": 8e-05,
      "loss": 1.525,
      "step": 4290
    },
    {
      "epoch": 0.9438029253271748,
      "grad_norm": 0.2875766456127167,
      "learning_rate": 8e-05,
      "loss": 1.6625,
      "step": 4291
    },
    {
      "epoch": 0.9440228747388101,
      "grad_norm": 0.3201231062412262,
      "learning_rate": 8e-05,
      "loss": 1.7481,
      "step": 4292
    },
    {
      "epoch": 0.9442428241504454,
      "grad_norm": 0.30221429467201233,
      "learning_rate": 8e-05,
      "loss": 1.6486,
      "step": 4293
    },
    {
      "epoch": 0.9444627735620807,
      "grad_norm": 0.3243577182292938,
      "learning_rate": 8e-05,
      "loss": 1.6943,
      "step": 4294
    },
    {
      "epoch": 0.944682722973716,
      "grad_norm": 0.2791222035884857,
      "learning_rate": 8e-05,
      "loss": 1.6036,
      "step": 4295
    },
    {
      "epoch": 0.9449026723853514,
      "grad_norm": 0.2943405210971832,
      "learning_rate": 8e-05,
      "loss": 1.6321,
      "step": 4296
    },
    {
      "epoch": 0.9451226217969867,
      "grad_norm": 0.303776353597641,
      "learning_rate": 8e-05,
      "loss": 1.6788,
      "step": 4297
    },
    {
      "epoch": 0.945342571208622,
      "grad_norm": 0.2904314398765564,
      "learning_rate": 8e-05,
      "loss": 1.5924,
      "step": 4298
    },
    {
      "epoch": 0.9455625206202574,
      "grad_norm": 0.29499560594558716,
      "learning_rate": 8e-05,
      "loss": 1.767,
      "step": 4299
    },
    {
      "epoch": 0.9457824700318926,
      "grad_norm": 0.2957536578178406,
      "learning_rate": 8e-05,
      "loss": 1.6784,
      "step": 4300
    },
    {
      "epoch": 0.946002419443528,
      "grad_norm": 0.3278692364692688,
      "learning_rate": 8e-05,
      "loss": 1.7183,
      "step": 4301
    },
    {
      "epoch": 0.9462223688551633,
      "grad_norm": 0.32448363304138184,
      "learning_rate": 8e-05,
      "loss": 1.7439,
      "step": 4302
    },
    {
      "epoch": 0.9464423182667986,
      "grad_norm": 0.29863062500953674,
      "learning_rate": 8e-05,
      "loss": 1.8053,
      "step": 4303
    },
    {
      "epoch": 0.946662267678434,
      "grad_norm": 0.29405081272125244,
      "learning_rate": 8e-05,
      "loss": 1.5501,
      "step": 4304
    },
    {
      "epoch": 0.9468822170900693,
      "grad_norm": 0.29146045446395874,
      "learning_rate": 8e-05,
      "loss": 1.6857,
      "step": 4305
    },
    {
      "epoch": 0.9471021665017046,
      "grad_norm": 0.31284937262535095,
      "learning_rate": 8e-05,
      "loss": 1.6328,
      "step": 4306
    },
    {
      "epoch": 0.9473221159133399,
      "grad_norm": 0.29520636796951294,
      "learning_rate": 8e-05,
      "loss": 1.6156,
      "step": 4307
    },
    {
      "epoch": 0.9475420653249752,
      "grad_norm": 0.3117775022983551,
      "learning_rate": 8e-05,
      "loss": 1.6738,
      "step": 4308
    },
    {
      "epoch": 0.9477620147366106,
      "grad_norm": 0.3293139338493347,
      "learning_rate": 8e-05,
      "loss": 1.6706,
      "step": 4309
    },
    {
      "epoch": 0.9479819641482459,
      "grad_norm": 0.30195489525794983,
      "learning_rate": 8e-05,
      "loss": 1.6393,
      "step": 4310
    },
    {
      "epoch": 0.9482019135598813,
      "grad_norm": 0.2867063581943512,
      "learning_rate": 8e-05,
      "loss": 1.6381,
      "step": 4311
    },
    {
      "epoch": 0.9484218629715165,
      "grad_norm": 0.31880491971969604,
      "learning_rate": 8e-05,
      "loss": 1.6891,
      "step": 4312
    },
    {
      "epoch": 0.9486418123831519,
      "grad_norm": 0.28975731134414673,
      "learning_rate": 8e-05,
      "loss": 1.6369,
      "step": 4313
    },
    {
      "epoch": 0.9488617617947872,
      "grad_norm": 0.30534425377845764,
      "learning_rate": 8e-05,
      "loss": 1.6479,
      "step": 4314
    },
    {
      "epoch": 0.9490817112064225,
      "grad_norm": 0.29459723830223083,
      "learning_rate": 8e-05,
      "loss": 1.6362,
      "step": 4315
    },
    {
      "epoch": 0.9493016606180579,
      "grad_norm": 0.31398990750312805,
      "learning_rate": 8e-05,
      "loss": 1.5502,
      "step": 4316
    },
    {
      "epoch": 0.9495216100296932,
      "grad_norm": 0.3026418089866638,
      "learning_rate": 8e-05,
      "loss": 1.8244,
      "step": 4317
    },
    {
      "epoch": 0.9497415594413285,
      "grad_norm": 0.29010850191116333,
      "learning_rate": 8e-05,
      "loss": 1.6547,
      "step": 4318
    },
    {
      "epoch": 0.9499615088529638,
      "grad_norm": 0.28357642889022827,
      "learning_rate": 8e-05,
      "loss": 1.562,
      "step": 4319
    },
    {
      "epoch": 0.9501814582645991,
      "grad_norm": 0.3127571940422058,
      "learning_rate": 8e-05,
      "loss": 1.5851,
      "step": 4320
    },
    {
      "epoch": 0.9504014076762345,
      "grad_norm": 0.31412258744239807,
      "learning_rate": 8e-05,
      "loss": 1.7569,
      "step": 4321
    },
    {
      "epoch": 0.9506213570878698,
      "grad_norm": 0.29025739431381226,
      "learning_rate": 8e-05,
      "loss": 1.6701,
      "step": 4322
    },
    {
      "epoch": 0.9508413064995052,
      "grad_norm": 0.36086446046829224,
      "learning_rate": 8e-05,
      "loss": 1.8504,
      "step": 4323
    },
    {
      "epoch": 0.9510612559111404,
      "grad_norm": 0.29903754591941833,
      "learning_rate": 8e-05,
      "loss": 1.8401,
      "step": 4324
    },
    {
      "epoch": 0.9512812053227757,
      "grad_norm": 0.31516164541244507,
      "learning_rate": 8e-05,
      "loss": 1.6723,
      "step": 4325
    },
    {
      "epoch": 0.9515011547344111,
      "grad_norm": 0.2996736168861389,
      "learning_rate": 8e-05,
      "loss": 1.8117,
      "step": 4326
    },
    {
      "epoch": 0.9517211041460464,
      "grad_norm": 0.2747475504875183,
      "learning_rate": 8e-05,
      "loss": 1.4716,
      "step": 4327
    },
    {
      "epoch": 0.9519410535576818,
      "grad_norm": 0.2930557429790497,
      "learning_rate": 8e-05,
      "loss": 1.7099,
      "step": 4328
    },
    {
      "epoch": 0.952161002969317,
      "grad_norm": 0.3209344446659088,
      "learning_rate": 8e-05,
      "loss": 1.7772,
      "step": 4329
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 0.2976103127002716,
      "learning_rate": 8e-05,
      "loss": 1.6168,
      "step": 4330
    },
    {
      "epoch": 0.9526009017925877,
      "grad_norm": 0.2832762598991394,
      "learning_rate": 8e-05,
      "loss": 1.5194,
      "step": 4331
    },
    {
      "epoch": 0.952820851204223,
      "grad_norm": 0.3471873998641968,
      "learning_rate": 8e-05,
      "loss": 1.6147,
      "step": 4332
    },
    {
      "epoch": 0.9530408006158584,
      "grad_norm": 0.323137491941452,
      "learning_rate": 8e-05,
      "loss": 1.6678,
      "step": 4333
    },
    {
      "epoch": 0.9532607500274937,
      "grad_norm": 0.3110518157482147,
      "learning_rate": 8e-05,
      "loss": 1.7733,
      "step": 4334
    },
    {
      "epoch": 0.953480699439129,
      "grad_norm": 0.30080094933509827,
      "learning_rate": 8e-05,
      "loss": 1.6189,
      "step": 4335
    },
    {
      "epoch": 0.9537006488507643,
      "grad_norm": 0.3108077645301819,
      "learning_rate": 8e-05,
      "loss": 1.6818,
      "step": 4336
    },
    {
      "epoch": 0.9539205982623996,
      "grad_norm": 0.32039645314216614,
      "learning_rate": 8e-05,
      "loss": 1.7102,
      "step": 4337
    },
    {
      "epoch": 0.954140547674035,
      "grad_norm": 0.29062017798423767,
      "learning_rate": 8e-05,
      "loss": 1.6722,
      "step": 4338
    },
    {
      "epoch": 0.9543604970856703,
      "grad_norm": 0.314155250787735,
      "learning_rate": 8e-05,
      "loss": 1.7383,
      "step": 4339
    },
    {
      "epoch": 0.9545804464973057,
      "grad_norm": 0.29008060693740845,
      "learning_rate": 8e-05,
      "loss": 1.5542,
      "step": 4340
    },
    {
      "epoch": 0.954800395908941,
      "grad_norm": 0.31939879059791565,
      "learning_rate": 8e-05,
      "loss": 1.7417,
      "step": 4341
    },
    {
      "epoch": 0.9550203453205762,
      "grad_norm": 0.32870885729789734,
      "learning_rate": 8e-05,
      "loss": 1.8094,
      "step": 4342
    },
    {
      "epoch": 0.9552402947322116,
      "grad_norm": 0.3048943877220154,
      "learning_rate": 8e-05,
      "loss": 1.5526,
      "step": 4343
    },
    {
      "epoch": 0.9554602441438469,
      "grad_norm": 0.3018760681152344,
      "learning_rate": 8e-05,
      "loss": 1.7414,
      "step": 4344
    },
    {
      "epoch": 0.9556801935554823,
      "grad_norm": 0.299770325422287,
      "learning_rate": 8e-05,
      "loss": 1.7656,
      "step": 4345
    },
    {
      "epoch": 0.9559001429671176,
      "grad_norm": 0.3222806751728058,
      "learning_rate": 8e-05,
      "loss": 1.6876,
      "step": 4346
    },
    {
      "epoch": 0.9561200923787528,
      "grad_norm": 0.2997550666332245,
      "learning_rate": 8e-05,
      "loss": 1.6424,
      "step": 4347
    },
    {
      "epoch": 0.9563400417903882,
      "grad_norm": 0.298745721578598,
      "learning_rate": 8e-05,
      "loss": 1.6877,
      "step": 4348
    },
    {
      "epoch": 0.9565599912020235,
      "grad_norm": 0.3091779947280884,
      "learning_rate": 8e-05,
      "loss": 1.6664,
      "step": 4349
    },
    {
      "epoch": 0.9567799406136589,
      "grad_norm": 0.29578372836112976,
      "learning_rate": 8e-05,
      "loss": 1.7281,
      "step": 4350
    },
    {
      "epoch": 0.9569998900252942,
      "grad_norm": 0.29080474376678467,
      "learning_rate": 8e-05,
      "loss": 1.7009,
      "step": 4351
    },
    {
      "epoch": 0.9572198394369295,
      "grad_norm": 0.30178776383399963,
      "learning_rate": 8e-05,
      "loss": 1.7002,
      "step": 4352
    },
    {
      "epoch": 0.9574397888485648,
      "grad_norm": 0.3175758719444275,
      "learning_rate": 8e-05,
      "loss": 1.8114,
      "step": 4353
    },
    {
      "epoch": 0.9576597382602001,
      "grad_norm": 0.2928202748298645,
      "learning_rate": 8e-05,
      "loss": 1.5807,
      "step": 4354
    },
    {
      "epoch": 0.9578796876718355,
      "grad_norm": 0.30198556184768677,
      "learning_rate": 8e-05,
      "loss": 1.6466,
      "step": 4355
    },
    {
      "epoch": 0.9580996370834708,
      "grad_norm": 0.30202385783195496,
      "learning_rate": 8e-05,
      "loss": 1.7117,
      "step": 4356
    },
    {
      "epoch": 0.9583195864951062,
      "grad_norm": 0.30362236499786377,
      "learning_rate": 8e-05,
      "loss": 1.7272,
      "step": 4357
    },
    {
      "epoch": 0.9585395359067415,
      "grad_norm": 0.29718053340911865,
      "learning_rate": 8e-05,
      "loss": 1.5618,
      "step": 4358
    },
    {
      "epoch": 0.9587594853183767,
      "grad_norm": 0.28676435351371765,
      "learning_rate": 8e-05,
      "loss": 1.5514,
      "step": 4359
    },
    {
      "epoch": 0.9589794347300121,
      "grad_norm": 0.35272395610809326,
      "learning_rate": 8e-05,
      "loss": 1.6748,
      "step": 4360
    },
    {
      "epoch": 0.9591993841416474,
      "grad_norm": 0.28872179985046387,
      "learning_rate": 8e-05,
      "loss": 1.6616,
      "step": 4361
    },
    {
      "epoch": 0.9594193335532828,
      "grad_norm": 0.3032558262348175,
      "learning_rate": 8e-05,
      "loss": 1.7008,
      "step": 4362
    },
    {
      "epoch": 0.9596392829649181,
      "grad_norm": 0.307427316904068,
      "learning_rate": 8e-05,
      "loss": 1.6507,
      "step": 4363
    },
    {
      "epoch": 0.9598592323765534,
      "grad_norm": 0.28583332896232605,
      "learning_rate": 8e-05,
      "loss": 1.578,
      "step": 4364
    },
    {
      "epoch": 0.9600791817881887,
      "grad_norm": 0.3125254809856415,
      "learning_rate": 8e-05,
      "loss": 1.6767,
      "step": 4365
    },
    {
      "epoch": 0.960299131199824,
      "grad_norm": 0.30561694502830505,
      "learning_rate": 8e-05,
      "loss": 1.797,
      "step": 4366
    },
    {
      "epoch": 0.9605190806114594,
      "grad_norm": 0.30248764157295227,
      "learning_rate": 8e-05,
      "loss": 1.5862,
      "step": 4367
    },
    {
      "epoch": 0.9607390300230947,
      "grad_norm": 0.3098229169845581,
      "learning_rate": 8e-05,
      "loss": 1.7568,
      "step": 4368
    },
    {
      "epoch": 0.96095897943473,
      "grad_norm": 0.30071017146110535,
      "learning_rate": 8e-05,
      "loss": 1.6039,
      "step": 4369
    },
    {
      "epoch": 0.9611789288463654,
      "grad_norm": 0.3008226454257965,
      "learning_rate": 8e-05,
      "loss": 1.7515,
      "step": 4370
    },
    {
      "epoch": 0.9613988782580006,
      "grad_norm": 0.3019086420536041,
      "learning_rate": 8e-05,
      "loss": 1.7186,
      "step": 4371
    },
    {
      "epoch": 0.961618827669636,
      "grad_norm": 0.2837902307510376,
      "learning_rate": 8e-05,
      "loss": 1.5705,
      "step": 4372
    },
    {
      "epoch": 0.9618387770812713,
      "grad_norm": 0.27051079273223877,
      "learning_rate": 8e-05,
      "loss": 1.5068,
      "step": 4373
    },
    {
      "epoch": 0.9620587264929066,
      "grad_norm": 0.3019634485244751,
      "learning_rate": 8e-05,
      "loss": 1.8051,
      "step": 4374
    },
    {
      "epoch": 0.962278675904542,
      "grad_norm": 0.29091939330101013,
      "learning_rate": 8e-05,
      "loss": 1.5171,
      "step": 4375
    },
    {
      "epoch": 0.9624986253161772,
      "grad_norm": 0.3006875514984131,
      "learning_rate": 8e-05,
      "loss": 1.619,
      "step": 4376
    },
    {
      "epoch": 0.9627185747278126,
      "grad_norm": 0.2964954078197479,
      "learning_rate": 8e-05,
      "loss": 1.5904,
      "step": 4377
    },
    {
      "epoch": 0.9629385241394479,
      "grad_norm": 0.2927916944026947,
      "learning_rate": 8e-05,
      "loss": 1.6793,
      "step": 4378
    },
    {
      "epoch": 0.9631584735510833,
      "grad_norm": 0.2932109236717224,
      "learning_rate": 8e-05,
      "loss": 1.655,
      "step": 4379
    },
    {
      "epoch": 0.9633784229627186,
      "grad_norm": 0.29698604345321655,
      "learning_rate": 8e-05,
      "loss": 1.6694,
      "step": 4380
    },
    {
      "epoch": 0.9635983723743539,
      "grad_norm": 0.28351879119873047,
      "learning_rate": 8e-05,
      "loss": 1.6355,
      "step": 4381
    },
    {
      "epoch": 0.9638183217859893,
      "grad_norm": 0.3210040330886841,
      "learning_rate": 8e-05,
      "loss": 1.7633,
      "step": 4382
    },
    {
      "epoch": 0.9640382711976245,
      "grad_norm": 0.2798837423324585,
      "learning_rate": 8e-05,
      "loss": 1.6254,
      "step": 4383
    },
    {
      "epoch": 0.9642582206092599,
      "grad_norm": 0.2931169867515564,
      "learning_rate": 8e-05,
      "loss": 1.6197,
      "step": 4384
    },
    {
      "epoch": 0.9644781700208952,
      "grad_norm": 0.2779824137687683,
      "learning_rate": 8e-05,
      "loss": 1.623,
      "step": 4385
    },
    {
      "epoch": 0.9646981194325305,
      "grad_norm": 0.3030729293823242,
      "learning_rate": 8e-05,
      "loss": 1.5768,
      "step": 4386
    },
    {
      "epoch": 0.9649180688441659,
      "grad_norm": 0.2920215427875519,
      "learning_rate": 8e-05,
      "loss": 1.5899,
      "step": 4387
    },
    {
      "epoch": 0.9651380182558011,
      "grad_norm": 0.30520427227020264,
      "learning_rate": 8e-05,
      "loss": 1.6882,
      "step": 4388
    },
    {
      "epoch": 0.9653579676674365,
      "grad_norm": 0.3132491409778595,
      "learning_rate": 8e-05,
      "loss": 1.8433,
      "step": 4389
    },
    {
      "epoch": 0.9655779170790718,
      "grad_norm": 0.305482417345047,
      "learning_rate": 8e-05,
      "loss": 1.7519,
      "step": 4390
    },
    {
      "epoch": 0.9657978664907071,
      "grad_norm": 0.3076193630695343,
      "learning_rate": 8e-05,
      "loss": 1.6502,
      "step": 4391
    },
    {
      "epoch": 0.9660178159023425,
      "grad_norm": 0.30150118470191956,
      "learning_rate": 8e-05,
      "loss": 1.6548,
      "step": 4392
    },
    {
      "epoch": 0.9662377653139778,
      "grad_norm": 0.3134932219982147,
      "learning_rate": 8e-05,
      "loss": 1.6598,
      "step": 4393
    },
    {
      "epoch": 0.9664577147256131,
      "grad_norm": 0.31480950117111206,
      "learning_rate": 8e-05,
      "loss": 1.654,
      "step": 4394
    },
    {
      "epoch": 0.9666776641372484,
      "grad_norm": 0.2868606150150299,
      "learning_rate": 8e-05,
      "loss": 1.7108,
      "step": 4395
    },
    {
      "epoch": 0.9668976135488837,
      "grad_norm": 0.2981705069541931,
      "learning_rate": 8e-05,
      "loss": 1.6678,
      "step": 4396
    },
    {
      "epoch": 0.9671175629605191,
      "grad_norm": 0.30914050340652466,
      "learning_rate": 8e-05,
      "loss": 1.7751,
      "step": 4397
    },
    {
      "epoch": 0.9673375123721544,
      "grad_norm": 0.2951439619064331,
      "learning_rate": 8e-05,
      "loss": 1.7323,
      "step": 4398
    },
    {
      "epoch": 0.9675574617837898,
      "grad_norm": 0.3048958480358124,
      "learning_rate": 8e-05,
      "loss": 1.6566,
      "step": 4399
    },
    {
      "epoch": 0.967777411195425,
      "grad_norm": 0.2965305745601654,
      "learning_rate": 8e-05,
      "loss": 1.7116,
      "step": 4400
    },
    {
      "epoch": 0.9679973606070604,
      "grad_norm": 0.2833409011363983,
      "learning_rate": 8e-05,
      "loss": 1.5868,
      "step": 4401
    },
    {
      "epoch": 0.9682173100186957,
      "grad_norm": 0.3302156925201416,
      "learning_rate": 8e-05,
      "loss": 1.7029,
      "step": 4402
    },
    {
      "epoch": 0.968437259430331,
      "grad_norm": 0.3424098789691925,
      "learning_rate": 8e-05,
      "loss": 1.7407,
      "step": 4403
    },
    {
      "epoch": 0.9686572088419664,
      "grad_norm": 0.3008859157562256,
      "learning_rate": 8e-05,
      "loss": 1.6784,
      "step": 4404
    },
    {
      "epoch": 0.9688771582536017,
      "grad_norm": 0.3263012170791626,
      "learning_rate": 8e-05,
      "loss": 1.759,
      "step": 4405
    },
    {
      "epoch": 0.969097107665237,
      "grad_norm": 0.2940591871738434,
      "learning_rate": 8e-05,
      "loss": 1.6771,
      "step": 4406
    },
    {
      "epoch": 0.9693170570768723,
      "grad_norm": 0.2776276767253876,
      "learning_rate": 8e-05,
      "loss": 1.5939,
      "step": 4407
    },
    {
      "epoch": 0.9695370064885076,
      "grad_norm": 0.3084886968135834,
      "learning_rate": 8e-05,
      "loss": 1.8447,
      "step": 4408
    },
    {
      "epoch": 0.969756955900143,
      "grad_norm": 0.30279961228370667,
      "learning_rate": 8e-05,
      "loss": 1.6919,
      "step": 4409
    },
    {
      "epoch": 0.9699769053117783,
      "grad_norm": 0.2889377772808075,
      "learning_rate": 8e-05,
      "loss": 1.6302,
      "step": 4410
    },
    {
      "epoch": 0.9701968547234137,
      "grad_norm": 0.3116249740123749,
      "learning_rate": 8e-05,
      "loss": 1.6945,
      "step": 4411
    },
    {
      "epoch": 0.9704168041350489,
      "grad_norm": 0.2997124493122101,
      "learning_rate": 8e-05,
      "loss": 1.5636,
      "step": 4412
    },
    {
      "epoch": 0.9706367535466842,
      "grad_norm": 0.29871103167533875,
      "learning_rate": 8e-05,
      "loss": 1.6513,
      "step": 4413
    },
    {
      "epoch": 0.9708567029583196,
      "grad_norm": 0.2757253348827362,
      "learning_rate": 8e-05,
      "loss": 1.5724,
      "step": 4414
    },
    {
      "epoch": 0.9710766523699549,
      "grad_norm": 0.29090118408203125,
      "learning_rate": 8e-05,
      "loss": 1.7293,
      "step": 4415
    },
    {
      "epoch": 0.9712966017815903,
      "grad_norm": 0.2890031337738037,
      "learning_rate": 8e-05,
      "loss": 1.6034,
      "step": 4416
    },
    {
      "epoch": 0.9715165511932256,
      "grad_norm": 0.2936271131038666,
      "learning_rate": 8e-05,
      "loss": 1.5668,
      "step": 4417
    },
    {
      "epoch": 0.9717365006048608,
      "grad_norm": 0.336303174495697,
      "learning_rate": 8e-05,
      "loss": 1.6313,
      "step": 4418
    },
    {
      "epoch": 0.9719564500164962,
      "grad_norm": 0.31786924600601196,
      "learning_rate": 8e-05,
      "loss": 1.7597,
      "step": 4419
    },
    {
      "epoch": 0.9721763994281315,
      "grad_norm": 0.3035045266151428,
      "learning_rate": 8e-05,
      "loss": 1.6743,
      "step": 4420
    },
    {
      "epoch": 0.9723963488397669,
      "grad_norm": 0.27620363235473633,
      "learning_rate": 8e-05,
      "loss": 1.4299,
      "step": 4421
    },
    {
      "epoch": 0.9726162982514022,
      "grad_norm": 0.2746250629425049,
      "learning_rate": 8e-05,
      "loss": 1.5365,
      "step": 4422
    },
    {
      "epoch": 0.9728362476630376,
      "grad_norm": 0.3262247145175934,
      "learning_rate": 8e-05,
      "loss": 1.7204,
      "step": 4423
    },
    {
      "epoch": 0.9730561970746728,
      "grad_norm": 0.31040772795677185,
      "learning_rate": 8e-05,
      "loss": 1.648,
      "step": 4424
    },
    {
      "epoch": 0.9732761464863081,
      "grad_norm": 0.2918784022331238,
      "learning_rate": 8e-05,
      "loss": 1.6991,
      "step": 4425
    },
    {
      "epoch": 0.9734960958979435,
      "grad_norm": 0.3132600784301758,
      "learning_rate": 8e-05,
      "loss": 1.682,
      "step": 4426
    },
    {
      "epoch": 0.9737160453095788,
      "grad_norm": 0.2929805517196655,
      "learning_rate": 8e-05,
      "loss": 1.6979,
      "step": 4427
    },
    {
      "epoch": 0.9739359947212142,
      "grad_norm": 0.30973419547080994,
      "learning_rate": 8e-05,
      "loss": 1.5817,
      "step": 4428
    },
    {
      "epoch": 0.9741559441328494,
      "grad_norm": 0.3428287208080292,
      "learning_rate": 8e-05,
      "loss": 1.8223,
      "step": 4429
    },
    {
      "epoch": 0.9743758935444847,
      "grad_norm": 0.29221585392951965,
      "learning_rate": 8e-05,
      "loss": 1.5699,
      "step": 4430
    },
    {
      "epoch": 0.9745958429561201,
      "grad_norm": 0.28617948293685913,
      "learning_rate": 8e-05,
      "loss": 1.5019,
      "step": 4431
    },
    {
      "epoch": 0.9748157923677554,
      "grad_norm": 0.31226715445518494,
      "learning_rate": 8e-05,
      "loss": 1.581,
      "step": 4432
    },
    {
      "epoch": 0.9750357417793908,
      "grad_norm": 0.3052622675895691,
      "learning_rate": 8e-05,
      "loss": 1.6488,
      "step": 4433
    },
    {
      "epoch": 0.9752556911910261,
      "grad_norm": 0.29430046677589417,
      "learning_rate": 8e-05,
      "loss": 1.5932,
      "step": 4434
    },
    {
      "epoch": 0.9754756406026613,
      "grad_norm": 0.2936963140964508,
      "learning_rate": 8e-05,
      "loss": 1.5609,
      "step": 4435
    },
    {
      "epoch": 0.9756955900142967,
      "grad_norm": 0.2764322757720947,
      "learning_rate": 8e-05,
      "loss": 1.6329,
      "step": 4436
    },
    {
      "epoch": 0.975915539425932,
      "grad_norm": 0.28910988569259644,
      "learning_rate": 8e-05,
      "loss": 1.59,
      "step": 4437
    },
    {
      "epoch": 0.9761354888375674,
      "grad_norm": 0.28807327151298523,
      "learning_rate": 8e-05,
      "loss": 1.6683,
      "step": 4438
    },
    {
      "epoch": 0.9763554382492027,
      "grad_norm": 0.29345518350601196,
      "learning_rate": 8e-05,
      "loss": 1.6477,
      "step": 4439
    },
    {
      "epoch": 0.976575387660838,
      "grad_norm": 0.28553545475006104,
      "learning_rate": 8e-05,
      "loss": 1.6333,
      "step": 4440
    },
    {
      "epoch": 0.9767953370724733,
      "grad_norm": 0.3055817484855652,
      "learning_rate": 8e-05,
      "loss": 1.7352,
      "step": 4441
    },
    {
      "epoch": 0.9770152864841086,
      "grad_norm": 0.26843395829200745,
      "learning_rate": 8e-05,
      "loss": 1.5031,
      "step": 4442
    },
    {
      "epoch": 0.977235235895744,
      "grad_norm": 0.29834216833114624,
      "learning_rate": 8e-05,
      "loss": 1.7358,
      "step": 4443
    },
    {
      "epoch": 0.9774551853073793,
      "grad_norm": 0.32550013065338135,
      "learning_rate": 8e-05,
      "loss": 1.5645,
      "step": 4444
    },
    {
      "epoch": 0.9776751347190146,
      "grad_norm": 0.29521921277046204,
      "learning_rate": 8e-05,
      "loss": 1.6498,
      "step": 4445
    },
    {
      "epoch": 0.97789508413065,
      "grad_norm": 0.29316624999046326,
      "learning_rate": 8e-05,
      "loss": 1.5499,
      "step": 4446
    },
    {
      "epoch": 0.9781150335422852,
      "grad_norm": 0.2895122468471527,
      "learning_rate": 8e-05,
      "loss": 1.6961,
      "step": 4447
    },
    {
      "epoch": 0.9783349829539206,
      "grad_norm": 0.3027707040309906,
      "learning_rate": 8e-05,
      "loss": 1.9138,
      "step": 4448
    },
    {
      "epoch": 0.9785549323655559,
      "grad_norm": 0.2765319049358368,
      "learning_rate": 8e-05,
      "loss": 1.6376,
      "step": 4449
    },
    {
      "epoch": 0.9787748817771913,
      "grad_norm": 0.28939464688301086,
      "learning_rate": 8e-05,
      "loss": 1.6433,
      "step": 4450
    },
    {
      "epoch": 0.9789948311888266,
      "grad_norm": 0.3009035289287567,
      "learning_rate": 8e-05,
      "loss": 1.6584,
      "step": 4451
    },
    {
      "epoch": 0.9792147806004619,
      "grad_norm": 0.30962783098220825,
      "learning_rate": 8e-05,
      "loss": 1.673,
      "step": 4452
    },
    {
      "epoch": 0.9794347300120972,
      "grad_norm": 0.27727317810058594,
      "learning_rate": 8e-05,
      "loss": 1.5244,
      "step": 4453
    },
    {
      "epoch": 0.9796546794237325,
      "grad_norm": 0.28220412135124207,
      "learning_rate": 8e-05,
      "loss": 1.4242,
      "step": 4454
    },
    {
      "epoch": 0.9798746288353679,
      "grad_norm": 0.29269176721572876,
      "learning_rate": 8e-05,
      "loss": 1.5952,
      "step": 4455
    },
    {
      "epoch": 0.9800945782470032,
      "grad_norm": 0.30379146337509155,
      "learning_rate": 8e-05,
      "loss": 1.6384,
      "step": 4456
    },
    {
      "epoch": 0.9803145276586385,
      "grad_norm": 0.3239672780036926,
      "learning_rate": 8e-05,
      "loss": 1.5808,
      "step": 4457
    },
    {
      "epoch": 0.9805344770702739,
      "grad_norm": 0.28674960136413574,
      "learning_rate": 8e-05,
      "loss": 1.4705,
      "step": 4458
    },
    {
      "epoch": 0.9807544264819091,
      "grad_norm": 0.2956444025039673,
      "learning_rate": 8e-05,
      "loss": 1.7746,
      "step": 4459
    },
    {
      "epoch": 0.9809743758935445,
      "grad_norm": 0.2885272800922394,
      "learning_rate": 8e-05,
      "loss": 1.5695,
      "step": 4460
    },
    {
      "epoch": 0.9811943253051798,
      "grad_norm": 0.29144173860549927,
      "learning_rate": 8e-05,
      "loss": 1.7118,
      "step": 4461
    },
    {
      "epoch": 0.9814142747168151,
      "grad_norm": 0.31238171458244324,
      "learning_rate": 8e-05,
      "loss": 1.7893,
      "step": 4462
    },
    {
      "epoch": 0.9816342241284505,
      "grad_norm": 0.2996987998485565,
      "learning_rate": 8e-05,
      "loss": 1.7206,
      "step": 4463
    },
    {
      "epoch": 0.9818541735400858,
      "grad_norm": 0.29437190294265747,
      "learning_rate": 8e-05,
      "loss": 1.6398,
      "step": 4464
    },
    {
      "epoch": 0.9820741229517211,
      "grad_norm": 0.28236857056617737,
      "learning_rate": 8e-05,
      "loss": 1.477,
      "step": 4465
    },
    {
      "epoch": 0.9822940723633564,
      "grad_norm": 0.290854811668396,
      "learning_rate": 8e-05,
      "loss": 1.6445,
      "step": 4466
    },
    {
      "epoch": 0.9825140217749917,
      "grad_norm": 0.28919798135757446,
      "learning_rate": 8e-05,
      "loss": 1.5369,
      "step": 4467
    },
    {
      "epoch": 0.9827339711866271,
      "grad_norm": 0.3495406210422516,
      "learning_rate": 8e-05,
      "loss": 1.8034,
      "step": 4468
    },
    {
      "epoch": 0.9829539205982624,
      "grad_norm": 0.2776128053665161,
      "learning_rate": 8e-05,
      "loss": 1.4966,
      "step": 4469
    },
    {
      "epoch": 0.9831738700098978,
      "grad_norm": 0.32572057843208313,
      "learning_rate": 8e-05,
      "loss": 1.6874,
      "step": 4470
    },
    {
      "epoch": 0.983393819421533,
      "grad_norm": 0.3179634213447571,
      "learning_rate": 8e-05,
      "loss": 1.5736,
      "step": 4471
    },
    {
      "epoch": 0.9836137688331684,
      "grad_norm": 0.30530181527137756,
      "learning_rate": 8e-05,
      "loss": 1.7474,
      "step": 4472
    },
    {
      "epoch": 0.9838337182448037,
      "grad_norm": 0.30747902393341064,
      "learning_rate": 8e-05,
      "loss": 1.7044,
      "step": 4473
    },
    {
      "epoch": 0.984053667656439,
      "grad_norm": 0.2948608100414276,
      "learning_rate": 8e-05,
      "loss": 1.6953,
      "step": 4474
    },
    {
      "epoch": 0.9842736170680744,
      "grad_norm": 0.32307469844818115,
      "learning_rate": 8e-05,
      "loss": 1.6535,
      "step": 4475
    },
    {
      "epoch": 0.9844935664797096,
      "grad_norm": 0.314513623714447,
      "learning_rate": 8e-05,
      "loss": 1.6058,
      "step": 4476
    },
    {
      "epoch": 0.984713515891345,
      "grad_norm": 0.30157166719436646,
      "learning_rate": 8e-05,
      "loss": 1.7344,
      "step": 4477
    },
    {
      "epoch": 0.9849334653029803,
      "grad_norm": 0.32073357701301575,
      "learning_rate": 8e-05,
      "loss": 1.5717,
      "step": 4478
    },
    {
      "epoch": 0.9851534147146156,
      "grad_norm": 0.29884982109069824,
      "learning_rate": 8e-05,
      "loss": 1.6779,
      "step": 4479
    },
    {
      "epoch": 0.985373364126251,
      "grad_norm": 0.3231385052204132,
      "learning_rate": 8e-05,
      "loss": 1.8388,
      "step": 4480
    },
    {
      "epoch": 0.9855933135378863,
      "grad_norm": 0.2887023091316223,
      "learning_rate": 8e-05,
      "loss": 1.6754,
      "step": 4481
    },
    {
      "epoch": 0.9858132629495217,
      "grad_norm": 0.2872041165828705,
      "learning_rate": 8e-05,
      "loss": 1.6078,
      "step": 4482
    },
    {
      "epoch": 0.9860332123611569,
      "grad_norm": 0.2921448051929474,
      "learning_rate": 8e-05,
      "loss": 1.5941,
      "step": 4483
    },
    {
      "epoch": 0.9862531617727922,
      "grad_norm": 0.3134671747684479,
      "learning_rate": 8e-05,
      "loss": 1.6015,
      "step": 4484
    },
    {
      "epoch": 0.9864731111844276,
      "grad_norm": 0.31237107515335083,
      "learning_rate": 8e-05,
      "loss": 1.6342,
      "step": 4485
    },
    {
      "epoch": 0.9866930605960629,
      "grad_norm": 0.28489091992378235,
      "learning_rate": 8e-05,
      "loss": 1.564,
      "step": 4486
    },
    {
      "epoch": 0.9869130100076983,
      "grad_norm": 0.300727903842926,
      "learning_rate": 8e-05,
      "loss": 1.7434,
      "step": 4487
    },
    {
      "epoch": 0.9871329594193335,
      "grad_norm": 0.28028106689453125,
      "learning_rate": 8e-05,
      "loss": 1.4857,
      "step": 4488
    },
    {
      "epoch": 0.9873529088309688,
      "grad_norm": 0.28882718086242676,
      "learning_rate": 8e-05,
      "loss": 1.6234,
      "step": 4489
    },
    {
      "epoch": 0.9875728582426042,
      "grad_norm": 0.3193677067756653,
      "learning_rate": 8e-05,
      "loss": 1.7581,
      "step": 4490
    },
    {
      "epoch": 0.9877928076542395,
      "grad_norm": 0.2990550398826599,
      "learning_rate": 8e-05,
      "loss": 1.55,
      "step": 4491
    },
    {
      "epoch": 0.9880127570658749,
      "grad_norm": 0.3127286434173584,
      "learning_rate": 8e-05,
      "loss": 1.8427,
      "step": 4492
    },
    {
      "epoch": 0.9882327064775102,
      "grad_norm": 0.3024497926235199,
      "learning_rate": 8e-05,
      "loss": 1.6602,
      "step": 4493
    },
    {
      "epoch": 0.9884526558891455,
      "grad_norm": 0.2969488799571991,
      "learning_rate": 8e-05,
      "loss": 1.7345,
      "step": 4494
    },
    {
      "epoch": 0.9886726053007808,
      "grad_norm": 0.2769564092159271,
      "learning_rate": 8e-05,
      "loss": 1.5995,
      "step": 4495
    },
    {
      "epoch": 0.9888925547124161,
      "grad_norm": 0.29575562477111816,
      "learning_rate": 8e-05,
      "loss": 1.6056,
      "step": 4496
    },
    {
      "epoch": 0.9891125041240515,
      "grad_norm": 0.3045266270637512,
      "learning_rate": 8e-05,
      "loss": 1.5789,
      "step": 4497
    },
    {
      "epoch": 0.9893324535356868,
      "grad_norm": 0.2926347553730011,
      "learning_rate": 8e-05,
      "loss": 1.5467,
      "step": 4498
    },
    {
      "epoch": 0.9895524029473222,
      "grad_norm": 0.2982698976993561,
      "learning_rate": 8e-05,
      "loss": 1.7507,
      "step": 4499
    },
    {
      "epoch": 0.9897723523589574,
      "grad_norm": 0.31570032238960266,
      "learning_rate": 8e-05,
      "loss": 1.6138,
      "step": 4500
    },
    {
      "epoch": 0.9899923017705927,
      "grad_norm": 0.29251834750175476,
      "learning_rate": 8e-05,
      "loss": 1.6789,
      "step": 4501
    },
    {
      "epoch": 0.9902122511822281,
      "grad_norm": 0.28332188725471497,
      "learning_rate": 8e-05,
      "loss": 1.517,
      "step": 4502
    },
    {
      "epoch": 0.9904322005938634,
      "grad_norm": 0.31072255969047546,
      "learning_rate": 8e-05,
      "loss": 1.8599,
      "step": 4503
    },
    {
      "epoch": 0.9906521500054988,
      "grad_norm": 0.3108634650707245,
      "learning_rate": 8e-05,
      "loss": 1.7261,
      "step": 4504
    },
    {
      "epoch": 0.9908720994171341,
      "grad_norm": 0.29612377285957336,
      "learning_rate": 8e-05,
      "loss": 1.5988,
      "step": 4505
    },
    {
      "epoch": 0.9910920488287693,
      "grad_norm": 0.29026466608047485,
      "learning_rate": 8e-05,
      "loss": 1.6283,
      "step": 4506
    },
    {
      "epoch": 0.9913119982404047,
      "grad_norm": 0.31100019812583923,
      "learning_rate": 8e-05,
      "loss": 1.7241,
      "step": 4507
    },
    {
      "epoch": 0.99153194765204,
      "grad_norm": 0.29672232270240784,
      "learning_rate": 8e-05,
      "loss": 1.49,
      "step": 4508
    },
    {
      "epoch": 0.9917518970636754,
      "grad_norm": 0.28797340393066406,
      "learning_rate": 8e-05,
      "loss": 1.6308,
      "step": 4509
    },
    {
      "epoch": 0.9919718464753107,
      "grad_norm": 0.3070496618747711,
      "learning_rate": 8e-05,
      "loss": 1.6575,
      "step": 4510
    },
    {
      "epoch": 0.992191795886946,
      "grad_norm": 0.2943253517150879,
      "learning_rate": 8e-05,
      "loss": 1.6108,
      "step": 4511
    },
    {
      "epoch": 0.9924117452985813,
      "grad_norm": 0.311239093542099,
      "learning_rate": 8e-05,
      "loss": 1.7555,
      "step": 4512
    },
    {
      "epoch": 0.9926316947102166,
      "grad_norm": 0.3025715947151184,
      "learning_rate": 8e-05,
      "loss": 1.6224,
      "step": 4513
    },
    {
      "epoch": 0.992851644121852,
      "grad_norm": 0.3275901675224304,
      "learning_rate": 8e-05,
      "loss": 1.7496,
      "step": 4514
    },
    {
      "epoch": 0.9930715935334873,
      "grad_norm": 0.2779105603694916,
      "learning_rate": 8e-05,
      "loss": 1.5484,
      "step": 4515
    },
    {
      "epoch": 0.9932915429451227,
      "grad_norm": 0.2887669503688812,
      "learning_rate": 8e-05,
      "loss": 1.6076,
      "step": 4516
    },
    {
      "epoch": 0.993511492356758,
      "grad_norm": 0.28616613149642944,
      "learning_rate": 8e-05,
      "loss": 1.6341,
      "step": 4517
    },
    {
      "epoch": 0.9937314417683932,
      "grad_norm": 0.2875531017780304,
      "learning_rate": 8e-05,
      "loss": 1.5669,
      "step": 4518
    },
    {
      "epoch": 0.9939513911800286,
      "grad_norm": 0.28946763277053833,
      "learning_rate": 8e-05,
      "loss": 1.634,
      "step": 4519
    },
    {
      "epoch": 0.9941713405916639,
      "grad_norm": 0.2973674535751343,
      "learning_rate": 8e-05,
      "loss": 1.8391,
      "step": 4520
    },
    {
      "epoch": 0.9943912900032993,
      "grad_norm": 0.3605339229106903,
      "learning_rate": 8e-05,
      "loss": 1.7529,
      "step": 4521
    },
    {
      "epoch": 0.9946112394149346,
      "grad_norm": 0.28719213604927063,
      "learning_rate": 8e-05,
      "loss": 1.6061,
      "step": 4522
    },
    {
      "epoch": 0.9948311888265698,
      "grad_norm": 0.29998138546943665,
      "learning_rate": 8e-05,
      "loss": 1.7375,
      "step": 4523
    },
    {
      "epoch": 0.9950511382382052,
      "grad_norm": 0.3062797486782074,
      "learning_rate": 8e-05,
      "loss": 1.625,
      "step": 4524
    },
    {
      "epoch": 0.9952710876498405,
      "grad_norm": 0.28792884945869446,
      "learning_rate": 8e-05,
      "loss": 1.6572,
      "step": 4525
    },
    {
      "epoch": 0.9954910370614759,
      "grad_norm": 0.33734482526779175,
      "learning_rate": 8e-05,
      "loss": 1.9469,
      "step": 4526
    },
    {
      "epoch": 0.9957109864731112,
      "grad_norm": 0.2947852313518524,
      "learning_rate": 8e-05,
      "loss": 1.6837,
      "step": 4527
    },
    {
      "epoch": 0.9959309358847465,
      "grad_norm": 0.30927664041519165,
      "learning_rate": 8e-05,
      "loss": 1.7504,
      "step": 4528
    },
    {
      "epoch": 0.9961508852963818,
      "grad_norm": 0.2859261631965637,
      "learning_rate": 8e-05,
      "loss": 1.7234,
      "step": 4529
    },
    {
      "epoch": 0.9963708347080171,
      "grad_norm": 0.29213234782218933,
      "learning_rate": 8e-05,
      "loss": 1.6802,
      "step": 4530
    },
    {
      "epoch": 0.9965907841196525,
      "grad_norm": 0.30377697944641113,
      "learning_rate": 8e-05,
      "loss": 1.7073,
      "step": 4531
    },
    {
      "epoch": 0.9968107335312878,
      "grad_norm": 0.32840946316719055,
      "learning_rate": 8e-05,
      "loss": 1.7767,
      "step": 4532
    },
    {
      "epoch": 0.9970306829429231,
      "grad_norm": 0.29963135719299316,
      "learning_rate": 8e-05,
      "loss": 1.6862,
      "step": 4533
    },
    {
      "epoch": 0.9972506323545585,
      "grad_norm": 0.28282636404037476,
      "learning_rate": 8e-05,
      "loss": 1.6298,
      "step": 4534
    },
    {
      "epoch": 0.9974705817661937,
      "grad_norm": 0.29182666540145874,
      "learning_rate": 8e-05,
      "loss": 1.6182,
      "step": 4535
    },
    {
      "epoch": 0.9976905311778291,
      "grad_norm": 0.290402352809906,
      "learning_rate": 8e-05,
      "loss": 1.7541,
      "step": 4536
    },
    {
      "epoch": 0.9979104805894644,
      "grad_norm": 0.2885053753852844,
      "learning_rate": 8e-05,
      "loss": 1.619,
      "step": 4537
    },
    {
      "epoch": 0.9981304300010998,
      "grad_norm": 0.30916741490364075,
      "learning_rate": 8e-05,
      "loss": 1.6159,
      "step": 4538
    },
    {
      "epoch": 0.9983503794127351,
      "grad_norm": 0.2906371057033539,
      "learning_rate": 8e-05,
      "loss": 1.5887,
      "step": 4539
    },
    {
      "epoch": 0.9985703288243704,
      "grad_norm": 0.2910885810852051,
      "learning_rate": 8e-05,
      "loss": 1.6174,
      "step": 4540
    },
    {
      "epoch": 0.9987902782360057,
      "grad_norm": 0.325016051530838,
      "learning_rate": 8e-05,
      "loss": 1.6684,
      "step": 4541
    },
    {
      "epoch": 0.999010227647641,
      "grad_norm": 0.29865211248397827,
      "learning_rate": 8e-05,
      "loss": 1.6947,
      "step": 4542
    },
    {
      "epoch": 0.9992301770592764,
      "grad_norm": 0.29670125246047974,
      "learning_rate": 8e-05,
      "loss": 1.678,
      "step": 4543
    },
    {
      "epoch": 0.9994501264709117,
      "grad_norm": 0.28062552213668823,
      "learning_rate": 8e-05,
      "loss": 1.5117,
      "step": 4544
    },
    {
      "epoch": 0.999670075882547,
      "grad_norm": 0.3313736319541931,
      "learning_rate": 8e-05,
      "loss": 1.7289,
      "step": 4545
    },
    {
      "epoch": 0.9998900252941824,
      "grad_norm": 0.3050709068775177,
      "learning_rate": 8e-05,
      "loss": 1.6675,
      "step": 4546
    }
  ],
  "logging_steps": 1,
  "max_steps": 4546,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.572128411223327e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}