{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2956,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00033829499323410016,
      "grad_norm": 1.6407462358474731,
      "learning_rate": 5.0000000000000004e-08,
      "loss": 0.8903,
      "step": 1
    },
    {
      "epoch": 0.0006765899864682003,
      "grad_norm": 1.5672248601913452,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 0.8953,
      "step": 2
    },
    {
      "epoch": 0.0010148849797023004,
      "grad_norm": 1.6485013961791992,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 0.8939,
      "step": 3
    },
    {
      "epoch": 0.0013531799729364006,
      "grad_norm": 1.6102432012557983,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.9113,
      "step": 4
    },
    {
      "epoch": 0.0016914749661705007,
      "grad_norm": 1.5657951831817627,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.9247,
      "step": 5
    },
    {
      "epoch": 0.0020297699594046007,
      "grad_norm": 1.5672881603240967,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.9252,
      "step": 6
    },
    {
      "epoch": 0.002368064952638701,
      "grad_norm": 1.6827484369277954,
      "learning_rate": 3.5000000000000004e-07,
      "loss": 0.909,
      "step": 7
    },
    {
      "epoch": 0.0027063599458728013,
      "grad_norm": 1.5518872737884521,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.8979,
      "step": 8
    },
    {
      "epoch": 0.003044654939106901,
      "grad_norm": 1.5625290870666504,
      "learning_rate": 4.5000000000000003e-07,
      "loss": 0.9146,
      "step": 9
    },
    {
      "epoch": 0.0033829499323410014,
      "grad_norm": 1.491262674331665,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.9011,
      "step": 10
    },
    {
      "epoch": 0.0037212449255751017,
      "grad_norm": 1.6137615442276,
      "learning_rate": 5.5e-07,
      "loss": 0.9229,
      "step": 11
    },
    {
      "epoch": 0.0040595399188092015,
      "grad_norm": 1.67592191696167,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.9178,
      "step": 12
    },
    {
      "epoch": 0.004397834912043302,
      "grad_norm": 1.4431407451629639,
      "learning_rate": 6.5e-07,
      "loss": 0.899,
      "step": 13
    },
    {
      "epoch": 0.004736129905277402,
      "grad_norm": 1.4896955490112305,
      "learning_rate": 7.000000000000001e-07,
      "loss": 0.8985,
      "step": 14
    },
    {
      "epoch": 0.005074424898511502,
      "grad_norm": 1.3822637796401978,
      "learning_rate": 7.5e-07,
      "loss": 0.9025,
      "step": 15
    },
    {
      "epoch": 0.005412719891745603,
      "grad_norm": 1.2707531452178955,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.9148,
      "step": 16
    },
    {
      "epoch": 0.005751014884979702,
      "grad_norm": 1.298106074333191,
      "learning_rate": 8.500000000000001e-07,
      "loss": 0.8818,
      "step": 17
    },
    {
      "epoch": 0.006089309878213802,
      "grad_norm": 1.1657626628875732,
      "learning_rate": 9.000000000000001e-07,
      "loss": 0.8882,
      "step": 18
    },
    {
      "epoch": 0.0064276048714479025,
      "grad_norm": 1.1404670476913452,
      "learning_rate": 9.500000000000001e-07,
      "loss": 0.9025,
      "step": 19
    },
    {
      "epoch": 0.006765899864682003,
      "grad_norm": 1.0163322687149048,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.8729,
      "step": 20
    },
    {
      "epoch": 0.007104194857916103,
      "grad_norm": 1.0249444246292114,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 0.867,
      "step": 21
    },
    {
      "epoch": 0.007442489851150203,
      "grad_norm": 0.9234563708305359,
      "learning_rate": 1.1e-06,
      "loss": 0.857,
      "step": 22
    },
    {
      "epoch": 0.007780784844384303,
      "grad_norm": 0.8658290505409241,
      "learning_rate": 1.1500000000000002e-06,
      "loss": 0.8693,
      "step": 23
    },
    {
      "epoch": 0.008119079837618403,
      "grad_norm": 0.8107594847679138,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.8694,
      "step": 24
    },
    {
      "epoch": 0.008457374830852503,
      "grad_norm": 0.8184226751327515,
      "learning_rate": 1.25e-06,
      "loss": 0.8522,
      "step": 25
    },
    {
      "epoch": 0.008795669824086604,
      "grad_norm": 0.7714635729789734,
      "learning_rate": 1.3e-06,
      "loss": 0.8593,
      "step": 26
    },
    {
      "epoch": 0.009133964817320704,
      "grad_norm": 0.7085200548171997,
      "learning_rate": 1.3500000000000002e-06,
      "loss": 0.846,
      "step": 27
    },
    {
      "epoch": 0.009472259810554804,
      "grad_norm": 0.7201448082923889,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.8564,
      "step": 28
    },
    {
      "epoch": 0.009810554803788904,
      "grad_norm": 0.6707621216773987,
      "learning_rate": 1.45e-06,
      "loss": 0.8195,
      "step": 29
    },
    {
      "epoch": 0.010148849797023005,
      "grad_norm": 0.6455383896827698,
      "learning_rate": 1.5e-06,
      "loss": 0.8349,
      "step": 30
    },
    {
      "epoch": 0.010487144790257105,
      "grad_norm": 0.5779524445533752,
      "learning_rate": 1.5500000000000002e-06,
      "loss": 0.819,
      "step": 31
    },
    {
      "epoch": 0.010825439783491205,
      "grad_norm": 0.575465202331543,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.8466,
      "step": 32
    },
    {
      "epoch": 0.011163734776725304,
      "grad_norm": 0.5340820550918579,
      "learning_rate": 1.6500000000000003e-06,
      "loss": 0.8167,
      "step": 33
    },
    {
      "epoch": 0.011502029769959404,
      "grad_norm": 0.4989947974681854,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.8148,
      "step": 34
    },
    {
      "epoch": 0.011840324763193504,
      "grad_norm": 0.4422353208065033,
      "learning_rate": 1.75e-06,
      "loss": 0.806,
      "step": 35
    },
    {
      "epoch": 0.012178619756427604,
      "grad_norm": 0.4404710531234741,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.7935,
      "step": 36
    },
    {
      "epoch": 0.012516914749661705,
      "grad_norm": 0.4069363474845886,
      "learning_rate": 1.85e-06,
      "loss": 0.793,
      "step": 37
    },
    {
      "epoch": 0.012855209742895805,
      "grad_norm": 0.40475258231163025,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 0.7932,
      "step": 38
    },
    {
      "epoch": 0.013193504736129905,
      "grad_norm": 0.38950204849243164,
      "learning_rate": 1.9500000000000004e-06,
      "loss": 0.8042,
      "step": 39
    },
    {
      "epoch": 0.013531799729364006,
      "grad_norm": 0.3722783625125885,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.7772,
      "step": 40
    },
    {
      "epoch": 0.013870094722598106,
      "grad_norm": 0.3677540123462677,
      "learning_rate": 2.05e-06,
      "loss": 0.7688,
      "step": 41
    },
    {
      "epoch": 0.014208389715832206,
      "grad_norm": 0.37267571687698364,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.7631,
      "step": 42
    },
    {
      "epoch": 0.014546684709066306,
      "grad_norm": 0.3498350977897644,
      "learning_rate": 2.15e-06,
      "loss": 0.7653,
      "step": 43
    },
    {
      "epoch": 0.014884979702300407,
      "grad_norm": 0.334867388010025,
      "learning_rate": 2.2e-06,
      "loss": 0.7604,
      "step": 44
    },
    {
      "epoch": 0.015223274695534507,
      "grad_norm": 0.3571762442588806,
      "learning_rate": 2.25e-06,
      "loss": 0.7978,
      "step": 45
    },
    {
      "epoch": 0.015561569688768605,
      "grad_norm": 0.32837843894958496,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 0.7444,
      "step": 46
    },
    {
      "epoch": 0.015899864682002707,
      "grad_norm": 0.34096264839172363,
      "learning_rate": 2.35e-06,
      "loss": 0.784,
      "step": 47
    },
    {
      "epoch": 0.016238159675236806,
      "grad_norm": 0.3243989944458008,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.7727,
      "step": 48
    },
    {
      "epoch": 0.016576454668470908,
      "grad_norm": 0.3283677101135254,
      "learning_rate": 2.4500000000000003e-06,
      "loss": 0.7939,
      "step": 49
    },
    {
      "epoch": 0.016914749661705007,
      "grad_norm": 0.3175937533378601,
      "learning_rate": 2.5e-06,
      "loss": 0.7696,
      "step": 50
    },
    {
      "epoch": 0.01725304465493911,
      "grad_norm": 0.2957337498664856,
      "learning_rate": 2.55e-06,
      "loss": 0.7507,
      "step": 51
    },
    {
      "epoch": 0.017591339648173207,
      "grad_norm": 0.3031109571456909,
      "learning_rate": 2.6e-06,
      "loss": 0.7743,
      "step": 52
    },
    {
      "epoch": 0.017929634641407306,
      "grad_norm": 0.2781732678413391,
      "learning_rate": 2.6500000000000005e-06,
      "loss": 0.7428,
      "step": 53
    },
    {
      "epoch": 0.018267929634641408,
      "grad_norm": 0.28893619775772095,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 0.7814,
      "step": 54
    },
    {
      "epoch": 0.018606224627875506,
      "grad_norm": 0.2668273448944092,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.7836,
      "step": 55
    },
    {
      "epoch": 0.018944519621109608,
      "grad_norm": 0.2713071405887604,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.7601,
      "step": 56
    },
    {
      "epoch": 0.019282814614343707,
      "grad_norm": 0.2615717649459839,
      "learning_rate": 2.85e-06,
      "loss": 0.772,
      "step": 57
    },
    {
      "epoch": 0.01962110960757781,
      "grad_norm": 0.24850614368915558,
      "learning_rate": 2.9e-06,
      "loss": 0.7478,
      "step": 58
    },
    {
      "epoch": 0.019959404600811907,
      "grad_norm": 0.25940048694610596,
      "learning_rate": 2.95e-06,
      "loss": 0.7871,
      "step": 59
    },
    {
      "epoch": 0.02029769959404601,
      "grad_norm": 0.23697689175605774,
      "learning_rate": 3e-06,
      "loss": 0.7599,
      "step": 60
    },
    {
      "epoch": 0.020635994587280108,
      "grad_norm": 0.23364393413066864,
      "learning_rate": 3.05e-06,
      "loss": 0.7619,
      "step": 61
    },
    {
      "epoch": 0.02097428958051421,
      "grad_norm": 0.23482820391654968,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 0.7502,
      "step": 62
    },
    {
      "epoch": 0.02131258457374831,
      "grad_norm": 0.24202880263328552,
      "learning_rate": 3.1500000000000003e-06,
      "loss": 0.7422,
      "step": 63
    },
    {
      "epoch": 0.02165087956698241,
      "grad_norm": 0.2344835251569748,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.734,
      "step": 64
    },
    {
      "epoch": 0.02198917456021651,
      "grad_norm": 0.23511846363544464,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.7659,
      "step": 65
    },
    {
      "epoch": 0.022327469553450607,
      "grad_norm": 0.22822819650173187,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.77,
      "step": 66
    },
    {
      "epoch": 0.02266576454668471,
      "grad_norm": 0.23218046128749847,
      "learning_rate": 3.3500000000000005e-06,
      "loss": 0.7751,
      "step": 67
    },
    {
      "epoch": 0.023004059539918808,
      "grad_norm": 0.22659695148468018,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.7375,
      "step": 68
    },
    {
      "epoch": 0.02334235453315291,
      "grad_norm": 0.23136673867702484,
      "learning_rate": 3.45e-06,
      "loss": 0.7406,
      "step": 69
    },
    {
      "epoch": 0.02368064952638701,
      "grad_norm": 0.2264736145734787,
      "learning_rate": 3.5e-06,
      "loss": 0.7837,
      "step": 70
    },
    {
      "epoch": 0.02401894451962111,
      "grad_norm": 0.21895566582679749,
      "learning_rate": 3.5500000000000003e-06,
      "loss": 0.7698,
      "step": 71
    },
    {
      "epoch": 0.02435723951285521,
      "grad_norm": 0.2110176980495453,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.7692,
      "step": 72
    },
    {
      "epoch": 0.02469553450608931,
      "grad_norm": 0.2077152580022812,
      "learning_rate": 3.65e-06,
      "loss": 0.7274,
      "step": 73
    },
    {
      "epoch": 0.02503382949932341,
      "grad_norm": 0.2114841192960739,
      "learning_rate": 3.7e-06,
      "loss": 0.7438,
      "step": 74
    },
    {
      "epoch": 0.02537212449255751,
      "grad_norm": 0.20656870305538177,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.72,
      "step": 75
    },
    {
      "epoch": 0.02571041948579161,
      "grad_norm": 0.20921222865581512,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.7624,
      "step": 76
    },
    {
      "epoch": 0.026048714479025712,
      "grad_norm": 0.2081456482410431,
      "learning_rate": 3.85e-06,
      "loss": 0.7413,
      "step": 77
    },
    {
      "epoch": 0.02638700947225981,
      "grad_norm": 0.19761289656162262,
      "learning_rate": 3.900000000000001e-06,
      "loss": 0.7468,
      "step": 78
    },
    {
      "epoch": 0.02672530446549391,
      "grad_norm": 0.20001396536827087,
      "learning_rate": 3.95e-06,
      "loss": 0.7538,
      "step": 79
    },
    {
      "epoch": 0.02706359945872801,
      "grad_norm": 0.2047550082206726,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.7576,
      "step": 80
    },
    {
      "epoch": 0.02740189445196211,
      "grad_norm": 0.1974104791879654,
      "learning_rate": 4.05e-06,
      "loss": 0.7558,
      "step": 81
    },
    {
      "epoch": 0.02774018944519621,
      "grad_norm": 0.18975216150283813,
      "learning_rate": 4.1e-06,
      "loss": 0.7245,
      "step": 82
    },
    {
      "epoch": 0.02807848443843031,
      "grad_norm": 0.19672256708145142,
      "learning_rate": 4.15e-06,
      "loss": 0.7299,
      "step": 83
    },
    {
      "epoch": 0.028416779431664412,
      "grad_norm": 0.1959012895822525,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.7412,
      "step": 84
    },
    {
      "epoch": 0.02875507442489851,
      "grad_norm": 0.19008736312389374,
      "learning_rate": 4.25e-06,
      "loss": 0.7514,
      "step": 85
    },
    {
      "epoch": 0.029093369418132613,
      "grad_norm": 0.19592899084091187,
      "learning_rate": 4.3e-06,
      "loss": 0.7386,
      "step": 86
    },
    {
      "epoch": 0.02943166441136671,
      "grad_norm": 0.19850178062915802,
      "learning_rate": 4.350000000000001e-06,
      "loss": 0.745,
      "step": 87
    },
    {
      "epoch": 0.029769959404600813,
      "grad_norm": 0.199819415807724,
      "learning_rate": 4.4e-06,
      "loss": 0.7525,
      "step": 88
    },
    {
      "epoch": 0.030108254397834912,
      "grad_norm": 0.18576590716838837,
      "learning_rate": 4.450000000000001e-06,
      "loss": 0.739,
      "step": 89
    },
    {
      "epoch": 0.030446549391069014,
      "grad_norm": 0.18295684456825256,
      "learning_rate": 4.5e-06,
      "loss": 0.7568,
      "step": 90
    },
    {
      "epoch": 0.030784844384303112,
      "grad_norm": 0.19440272450447083,
      "learning_rate": 4.5500000000000005e-06,
      "loss": 0.7195,
      "step": 91
    },
    {
      "epoch": 0.03112313937753721,
      "grad_norm": 0.19021347165107727,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.7338,
      "step": 92
    },
    {
      "epoch": 0.03146143437077131,
      "grad_norm": 0.18591071665287018,
      "learning_rate": 4.65e-06,
      "loss": 0.7145,
      "step": 93
    },
    {
      "epoch": 0.031799729364005415,
      "grad_norm": 0.19127537310123444,
      "learning_rate": 4.7e-06,
      "loss": 0.7326,
      "step": 94
    },
    {
      "epoch": 0.03213802435723951,
      "grad_norm": 0.20237044990062714,
      "learning_rate": 4.75e-06,
      "loss": 0.7342,
      "step": 95
    },
    {
      "epoch": 0.03247631935047361,
      "grad_norm": 0.1817764937877655,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.6857,
      "step": 96
    },
    {
      "epoch": 0.032814614343707714,
      "grad_norm": 0.18967989087104797,
      "learning_rate": 4.85e-06,
      "loss": 0.7208,
      "step": 97
    },
    {
      "epoch": 0.033152909336941816,
      "grad_norm": 0.1865711659193039,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.7205,
      "step": 98
    },
    {
      "epoch": 0.03349120433017591,
      "grad_norm": 0.18442098796367645,
      "learning_rate": 4.95e-06,
      "loss": 0.7496,
      "step": 99
    },
    {
      "epoch": 0.03382949932341001,
      "grad_norm": 0.18783792853355408,
      "learning_rate": 5e-06,
      "loss": 0.7195,
      "step": 100
    },
    {
      "epoch": 0.034167794316644115,
      "grad_norm": 0.18517974019050598,
      "learning_rate": 4.999999839524525e-06,
      "loss": 0.7014,
      "step": 101
    },
    {
      "epoch": 0.03450608930987822,
      "grad_norm": 0.1914462447166443,
      "learning_rate": 4.999999358098119e-06,
      "loss": 0.7325,
      "step": 102
    },
    {
      "epoch": 0.03484438430311231,
      "grad_norm": 0.18725411593914032,
      "learning_rate": 4.999998555720843e-06,
      "loss": 0.7186,
      "step": 103
    },
    {
      "epoch": 0.035182679296346414,
      "grad_norm": 0.1851007640361786,
      "learning_rate": 4.999997432392803e-06,
      "loss": 0.7244,
      "step": 104
    },
    {
      "epoch": 0.035520974289580516,
      "grad_norm": 0.18126922845840454,
      "learning_rate": 4.99999598811414e-06,
      "loss": 0.7183,
      "step": 105
    },
    {
      "epoch": 0.03585926928281461,
      "grad_norm": 0.18332946300506592,
      "learning_rate": 4.999994222885042e-06,
      "loss": 0.701,
      "step": 106
    },
    {
      "epoch": 0.03619756427604871,
      "grad_norm": 0.19509664177894592,
      "learning_rate": 4.999992136705734e-06,
      "loss": 0.7298,
      "step": 107
    },
    {
      "epoch": 0.036535859269282815,
      "grad_norm": 0.1882726401090622,
      "learning_rate": 4.999989729576485e-06,
      "loss": 0.7258,
      "step": 108
    },
    {
      "epoch": 0.03687415426251692,
      "grad_norm": 0.19097895920276642,
      "learning_rate": 4.999987001497602e-06,
      "loss": 0.7253,
      "step": 109
    },
    {
      "epoch": 0.03721244925575101,
      "grad_norm": 0.19654102623462677,
      "learning_rate": 4.999983952469437e-06,
      "loss": 0.7086,
      "step": 110
    },
    {
      "epoch": 0.037550744248985114,
      "grad_norm": 0.18612295389175415,
      "learning_rate": 4.999980582492382e-06,
      "loss": 0.7197,
      "step": 111
    },
    {
      "epoch": 0.037889039242219216,
      "grad_norm": 0.1974918097257614,
      "learning_rate": 4.999976891566866e-06,
      "loss": 0.7076,
      "step": 112
    },
    {
      "epoch": 0.03822733423545332,
      "grad_norm": 0.1830679476261139,
      "learning_rate": 4.9999728796933684e-06,
      "loss": 0.7242,
      "step": 113
    },
    {
      "epoch": 0.03856562922868741,
      "grad_norm": 0.19517149031162262,
      "learning_rate": 4.9999685468724e-06,
      "loss": 0.7088,
      "step": 114
    },
    {
      "epoch": 0.038903924221921515,
      "grad_norm": 0.18623511493206024,
      "learning_rate": 4.9999638931045186e-06,
      "loss": 0.7168,
      "step": 115
    },
    {
      "epoch": 0.03924221921515562,
      "grad_norm": 0.18677926063537598,
      "learning_rate": 4.999958918390321e-06,
      "loss": 0.7097,
      "step": 116
    },
    {
      "epoch": 0.03958051420838971,
      "grad_norm": 0.18877893686294556,
      "learning_rate": 4.999953622730447e-06,
      "loss": 0.6961,
      "step": 117
    },
    {
      "epoch": 0.039918809201623814,
      "grad_norm": 0.19688422977924347,
      "learning_rate": 4.999948006125575e-06,
      "loss": 0.7143,
      "step": 118
    },
    {
      "epoch": 0.040257104194857916,
      "grad_norm": 0.1940767616033554,
      "learning_rate": 4.999942068576428e-06,
      "loss": 0.7357,
      "step": 119
    },
    {
      "epoch": 0.04059539918809202,
      "grad_norm": 0.18815898895263672,
      "learning_rate": 4.999935810083766e-06,
      "loss": 0.7258,
      "step": 120
    },
    {
      "epoch": 0.040933694181326113,
      "grad_norm": 0.18311698734760284,
      "learning_rate": 4.999929230648394e-06,
      "loss": 0.6921,
      "step": 121
    },
    {
      "epoch": 0.041271989174560215,
      "grad_norm": 0.1848682463169098,
      "learning_rate": 4.999922330271155e-06,
      "loss": 0.7434,
      "step": 122
    },
    {
      "epoch": 0.04161028416779432,
      "grad_norm": 0.18727323412895203,
      "learning_rate": 4.999915108952937e-06,
      "loss": 0.7205,
      "step": 123
    },
    {
      "epoch": 0.04194857916102842,
      "grad_norm": 0.18879395723342896,
      "learning_rate": 4.999907566694667e-06,
      "loss": 0.7029,
      "step": 124
    },
    {
      "epoch": 0.042286874154262515,
      "grad_norm": 0.18784329295158386,
      "learning_rate": 4.999899703497312e-06,
      "loss": 0.7002,
      "step": 125
    },
    {
      "epoch": 0.04262516914749662,
      "grad_norm": 0.1885390430688858,
      "learning_rate": 4.999891519361882e-06,
      "loss": 0.7038,
      "step": 126
    },
    {
      "epoch": 0.04296346414073072,
      "grad_norm": 0.18721194565296173,
      "learning_rate": 4.999883014289426e-06,
      "loss": 0.7227,
      "step": 127
    },
    {
      "epoch": 0.04330175913396482,
      "grad_norm": 0.18127906322479248,
      "learning_rate": 4.999874188281039e-06,
      "loss": 0.7432,
      "step": 128
    },
    {
      "epoch": 0.043640054127198916,
      "grad_norm": 0.18488162755966187,
      "learning_rate": 4.999865041337852e-06,
      "loss": 0.7434,
      "step": 129
    },
    {
      "epoch": 0.04397834912043302,
      "grad_norm": 0.18584181368350983,
      "learning_rate": 4.99985557346104e-06,
      "loss": 0.7445,
      "step": 130
    },
    {
      "epoch": 0.04431664411366712,
      "grad_norm": 0.18478542566299438,
      "learning_rate": 4.999845784651818e-06,
      "loss": 0.7149,
      "step": 131
    },
    {
      "epoch": 0.044654939106901215,
      "grad_norm": 0.1904173046350479,
      "learning_rate": 4.9998356749114434e-06,
      "loss": 0.716,
      "step": 132
    },
    {
      "epoch": 0.04499323410013532,
      "grad_norm": 0.19709834456443787,
      "learning_rate": 4.9998252442412134e-06,
      "loss": 0.758,
      "step": 133
    },
    {
      "epoch": 0.04533152909336942,
      "grad_norm": 0.1878356784582138,
      "learning_rate": 4.999814492642468e-06,
      "loss": 0.7011,
      "step": 134
    },
    {
      "epoch": 0.04566982408660352,
      "grad_norm": 0.1950335055589676,
      "learning_rate": 4.9998034201165866e-06,
      "loss": 0.6946,
      "step": 135
    },
    {
      "epoch": 0.046008119079837616,
      "grad_norm": 0.19865374267101288,
      "learning_rate": 4.999792026664992e-06,
      "loss": 0.7163,
      "step": 136
    },
    {
      "epoch": 0.04634641407307172,
      "grad_norm": 0.1960017830133438,
      "learning_rate": 4.9997803122891446e-06,
      "loss": 0.7347,
      "step": 137
    },
    {
      "epoch": 0.04668470906630582,
      "grad_norm": 0.18267491459846497,
      "learning_rate": 4.99976827699055e-06,
      "loss": 0.7052,
      "step": 138
    },
    {
      "epoch": 0.04702300405953992,
      "grad_norm": 0.18890222907066345,
      "learning_rate": 4.999755920770752e-06,
      "loss": 0.7128,
      "step": 139
    },
    {
      "epoch": 0.04736129905277402,
      "grad_norm": 0.20229989290237427,
      "learning_rate": 4.9997432436313386e-06,
      "loss": 0.7105,
      "step": 140
    },
    {
      "epoch": 0.04769959404600812,
      "grad_norm": 0.20218680799007416,
      "learning_rate": 4.999730245573936e-06,
      "loss": 0.7088,
      "step": 141
    },
    {
      "epoch": 0.04803788903924222,
      "grad_norm": 0.19978873431682587,
      "learning_rate": 4.999716926600215e-06,
      "loss": 0.7154,
      "step": 142
    },
    {
      "epoch": 0.048376184032476316,
      "grad_norm": 0.1951395869255066,
      "learning_rate": 4.999703286711882e-06,
      "loss": 0.7204,
      "step": 143
    },
    {
      "epoch": 0.04871447902571042,
      "grad_norm": 0.18328897655010223,
      "learning_rate": 4.999689325910691e-06,
      "loss": 0.694,
      "step": 144
    },
    {
      "epoch": 0.04905277401894452,
      "grad_norm": 0.20357415080070496,
      "learning_rate": 4.999675044198432e-06,
      "loss": 0.7298,
      "step": 145
    },
    {
      "epoch": 0.04939106901217862,
      "grad_norm": 0.19996631145477295,
      "learning_rate": 4.999660441576941e-06,
      "loss": 0.6968,
      "step": 146
    },
    {
      "epoch": 0.04972936400541272,
      "grad_norm": 0.19135387241840363,
      "learning_rate": 4.999645518048092e-06,
      "loss": 0.7131,
      "step": 147
    },
    {
      "epoch": 0.05006765899864682,
      "grad_norm": 0.1947384476661682,
      "learning_rate": 4.9996302736137994e-06,
      "loss": 0.6915,
      "step": 148
    },
    {
      "epoch": 0.05040595399188092,
      "grad_norm": 0.18958574533462524,
      "learning_rate": 4.999614708276021e-06,
      "loss": 0.6993,
      "step": 149
    },
    {
      "epoch": 0.05074424898511502,
      "grad_norm": 0.1931043118238449,
      "learning_rate": 4.999598822036755e-06,
      "loss": 0.6793,
      "step": 150
    },
    {
      "epoch": 0.05108254397834912,
      "grad_norm": 0.20002217590808868,
      "learning_rate": 4.999582614898042e-06,
      "loss": 0.7188,
      "step": 151
    },
    {
      "epoch": 0.05142083897158322,
      "grad_norm": 0.18821609020233154,
      "learning_rate": 4.9995660868619614e-06,
      "loss": 0.6756,
      "step": 152
    },
    {
      "epoch": 0.05175913396481732,
      "grad_norm": 0.18836663663387299,
      "learning_rate": 4.999549237930636e-06,
      "loss": 0.6953,
      "step": 153
    },
    {
      "epoch": 0.052097428958051424,
      "grad_norm": 0.18696729838848114,
      "learning_rate": 4.999532068106228e-06,
      "loss": 0.7316,
      "step": 154
    },
    {
      "epoch": 0.05243572395128552,
      "grad_norm": 0.18782716989517212,
      "learning_rate": 4.999514577390943e-06,
      "loss": 0.7344,
      "step": 155
    },
    {
      "epoch": 0.05277401894451962,
      "grad_norm": 0.20105955004692078,
      "learning_rate": 4.999496765787024e-06,
      "loss": 0.7039,
      "step": 156
    },
    {
      "epoch": 0.05311231393775372,
      "grad_norm": 0.19204497337341309,
      "learning_rate": 4.999478633296761e-06,
      "loss": 0.6769,
      "step": 157
    },
    {
      "epoch": 0.05345060893098782,
      "grad_norm": 0.1988452970981598,
      "learning_rate": 4.999460179922479e-06,
      "loss": 0.7108,
      "step": 158
    },
    {
      "epoch": 0.05378890392422192,
      "grad_norm": 0.1891714632511139,
      "learning_rate": 4.999441405666547e-06,
      "loss": 0.7266,
      "step": 159
    },
    {
      "epoch": 0.05412719891745602,
      "grad_norm": 0.18919819593429565,
      "learning_rate": 4.999422310531377e-06,
      "loss": 0.7072,
      "step": 160
    },
    {
      "epoch": 0.054465493910690124,
      "grad_norm": 0.19568458199501038,
      "learning_rate": 4.999402894519421e-06,
      "loss": 0.7325,
      "step": 161
    },
    {
      "epoch": 0.05480378890392422,
      "grad_norm": 0.18339653313159943,
      "learning_rate": 4.9993831576331696e-06,
      "loss": 0.6853,
      "step": 162
    },
    {
      "epoch": 0.05514208389715832,
      "grad_norm": 0.19178633391857147,
      "learning_rate": 4.999363099875158e-06,
      "loss": 0.6865,
      "step": 163
    },
    {
      "epoch": 0.05548037889039242,
      "grad_norm": 0.18333321809768677,
      "learning_rate": 4.9993427212479604e-06,
      "loss": 0.7193,
      "step": 164
    },
    {
      "epoch": 0.055818673883626525,
      "grad_norm": 0.1928037852048874,
      "learning_rate": 4.999322021754193e-06,
      "loss": 0.6949,
      "step": 165
    },
    {
      "epoch": 0.05615696887686062,
      "grad_norm": 0.1899040788412094,
      "learning_rate": 4.999301001396513e-06,
      "loss": 0.6957,
      "step": 166
    },
    {
      "epoch": 0.05649526387009472,
      "grad_norm": 0.20395973324775696,
      "learning_rate": 4.99927966017762e-06,
      "loss": 0.7068,
      "step": 167
    },
    {
      "epoch": 0.056833558863328824,
      "grad_norm": 0.1995311826467514,
      "learning_rate": 4.999257998100254e-06,
      "loss": 0.7331,
      "step": 168
    },
    {
      "epoch": 0.05717185385656292,
      "grad_norm": 0.19529490172863007,
      "learning_rate": 4.999236015167195e-06,
      "loss": 0.6982,
      "step": 169
    },
    {
      "epoch": 0.05751014884979702,
      "grad_norm": 0.19061776995658875,
      "learning_rate": 4.999213711381265e-06,
      "loss": 0.6944,
      "step": 170
    },
    {
      "epoch": 0.05784844384303112,
      "grad_norm": 0.1899946630001068,
      "learning_rate": 4.999191086745328e-06,
      "loss": 0.6782,
      "step": 171
    },
    {
      "epoch": 0.058186738836265225,
      "grad_norm": 0.18246881663799286,
      "learning_rate": 4.9991681412622896e-06,
      "loss": 0.7231,
      "step": 172
    },
    {
      "epoch": 0.05852503382949932,
      "grad_norm": 0.18850313127040863,
      "learning_rate": 4.999144874935092e-06,
      "loss": 0.7017,
      "step": 173
    },
    {
      "epoch": 0.05886332882273342,
      "grad_norm": 0.19261717796325684,
      "learning_rate": 4.999121287766726e-06,
      "loss": 0.6849,
      "step": 174
    },
    {
      "epoch": 0.059201623815967525,
      "grad_norm": 0.1908152848482132,
      "learning_rate": 4.999097379760218e-06,
      "loss": 0.7284,
      "step": 175
    },
    {
      "epoch": 0.05953991880920163,
      "grad_norm": 0.19049489498138428,
      "learning_rate": 4.999073150918637e-06,
      "loss": 0.6793,
      "step": 176
    },
    {
      "epoch": 0.05987821380243572,
      "grad_norm": 0.18340197205543518,
      "learning_rate": 4.999048601245095e-06,
      "loss": 0.7046,
      "step": 177
    },
    {
      "epoch": 0.060216508795669824,
      "grad_norm": 0.1974797546863556,
      "learning_rate": 4.999023730742742e-06,
      "loss": 0.7157,
      "step": 178
    },
    {
      "epoch": 0.060554803788903926,
      "grad_norm": 0.19154323637485504,
      "learning_rate": 4.998998539414772e-06,
      "loss": 0.7011,
      "step": 179
    },
    {
      "epoch": 0.06089309878213803,
      "grad_norm": 0.2043318897485733,
      "learning_rate": 4.998973027264419e-06,
      "loss": 0.7123,
      "step": 180
    },
    {
      "epoch": 0.06123139377537212,
      "grad_norm": 0.19144178926944733,
      "learning_rate": 4.998947194294958e-06,
      "loss": 0.7065,
      "step": 181
    },
    {
      "epoch": 0.061569688768606225,
      "grad_norm": 0.18897554278373718,
      "learning_rate": 4.998921040509704e-06,
      "loss": 0.6802,
      "step": 182
    },
    {
      "epoch": 0.06190798376184033,
      "grad_norm": 0.19585008919239044,
      "learning_rate": 4.9988945659120175e-06,
      "loss": 0.6886,
      "step": 183
    },
    {
      "epoch": 0.06224627875507442,
      "grad_norm": 0.19337284564971924,
      "learning_rate": 4.998867770505295e-06,
      "loss": 0.7344,
      "step": 184
    },
    {
      "epoch": 0.06258457374830853,
      "grad_norm": 0.18725362420082092,
      "learning_rate": 4.998840654292978e-06,
      "loss": 0.7053,
      "step": 185
    },
    {
      "epoch": 0.06292286874154263,
      "grad_norm": 0.19687679409980774,
      "learning_rate": 4.998813217278546e-06,
      "loss": 0.6677,
      "step": 186
    },
    {
      "epoch": 0.06326116373477672,
      "grad_norm": 0.19855792820453644,
      "learning_rate": 4.9987854594655236e-06,
      "loss": 0.7214,
      "step": 187
    },
    {
      "epoch": 0.06359945872801083,
      "grad_norm": 0.20533519983291626,
      "learning_rate": 4.998757380857473e-06,
      "loss": 0.7445,
      "step": 188
    },
    {
      "epoch": 0.06393775372124492,
      "grad_norm": 0.20073583722114563,
      "learning_rate": 4.9987289814579984e-06,
      "loss": 0.686,
      "step": 189
    },
    {
      "epoch": 0.06427604871447902,
      "grad_norm": 0.19877952337265015,
      "learning_rate": 4.9987002612707466e-06,
      "loss": 0.6955,
      "step": 190
    },
    {
      "epoch": 0.06461434370771313,
      "grad_norm": 0.1983058601617813,
      "learning_rate": 4.998671220299405e-06,
      "loss": 0.7196,
      "step": 191
    },
    {
      "epoch": 0.06495263870094722,
      "grad_norm": 0.19413228332996368,
      "learning_rate": 4.9986418585477016e-06,
      "loss": 0.6662,
      "step": 192
    },
    {
      "epoch": 0.06529093369418133,
      "grad_norm": 0.19949573278427124,
      "learning_rate": 4.998612176019405e-06,
      "loss": 0.6847,
      "step": 193
    },
    {
      "epoch": 0.06562922868741543,
      "grad_norm": 0.19834288954734802,
      "learning_rate": 4.998582172718327e-06,
      "loss": 0.6724,
      "step": 194
    },
    {
      "epoch": 0.06596752368064952,
      "grad_norm": 0.19697165489196777,
      "learning_rate": 4.9985518486483196e-06,
      "loss": 0.6989,
      "step": 195
    },
    {
      "epoch": 0.06630581867388363,
      "grad_norm": 0.19365616142749786,
      "learning_rate": 4.998521203813275e-06,
      "loss": 0.7187,
      "step": 196
    },
    {
      "epoch": 0.06664411366711773,
      "grad_norm": 0.2025780975818634,
      "learning_rate": 4.998490238217126e-06,
      "loss": 0.7115,
      "step": 197
    },
    {
      "epoch": 0.06698240866035182,
      "grad_norm": 0.18400835990905762,
      "learning_rate": 4.998458951863851e-06,
      "loss": 0.6836,
      "step": 198
    },
    {
      "epoch": 0.06732070365358593,
      "grad_norm": 0.21085551381111145,
      "learning_rate": 4.998427344757465e-06,
      "loss": 0.6975,
      "step": 199
    },
    {
      "epoch": 0.06765899864682003,
      "grad_norm": 0.20779626071453094,
      "learning_rate": 4.998395416902026e-06,
      "loss": 0.7019,
      "step": 200
    },
    {
      "epoch": 0.06799729364005412,
      "grad_norm": 0.19607093930244446,
      "learning_rate": 4.998363168301633e-06,
      "loss": 0.6979,
      "step": 201
    },
    {
      "epoch": 0.06833558863328823,
      "grad_norm": 0.1943480521440506,
      "learning_rate": 4.9983305989604256e-06,
      "loss": 0.7063,
      "step": 202
    },
    {
      "epoch": 0.06867388362652233,
      "grad_norm": 0.1963743269443512,
      "learning_rate": 4.998297708882584e-06,
      "loss": 0.6998,
      "step": 203
    },
    {
      "epoch": 0.06901217861975643,
      "grad_norm": 0.2052895426750183,
      "learning_rate": 4.998264498072334e-06,
      "loss": 0.7093,
      "step": 204
    },
    {
      "epoch": 0.06935047361299053,
      "grad_norm": 0.19581186771392822,
      "learning_rate": 4.998230966533937e-06,
      "loss": 0.6761,
      "step": 205
    },
    {
      "epoch": 0.06968876860622462,
      "grad_norm": 0.20708315074443817,
      "learning_rate": 4.998197114271697e-06,
      "loss": 0.6819,
      "step": 206
    },
    {
      "epoch": 0.07002706359945873,
      "grad_norm": 0.20113201439380646,
      "learning_rate": 4.998162941289961e-06,
      "loss": 0.7028,
      "step": 207
    },
    {
      "epoch": 0.07036535859269283,
      "grad_norm": 0.19930900633335114,
      "learning_rate": 4.998128447593117e-06,
      "loss": 0.6805,
      "step": 208
    },
    {
      "epoch": 0.07070365358592692,
      "grad_norm": 0.19791027903556824,
      "learning_rate": 4.998093633185592e-06,
      "loss": 0.7051,
      "step": 209
    },
    {
      "epoch": 0.07104194857916103,
      "grad_norm": 0.20085057616233826,
      "learning_rate": 4.998058498071856e-06,
      "loss": 0.706,
      "step": 210
    },
    {
      "epoch": 0.07138024357239513,
      "grad_norm": 0.19891057908535004,
      "learning_rate": 4.9980230422564195e-06,
      "loss": 0.6875,
      "step": 211
    },
    {
      "epoch": 0.07171853856562922,
      "grad_norm": 0.19535799324512482,
      "learning_rate": 4.997987265743834e-06,
      "loss": 0.6855,
      "step": 212
    },
    {
      "epoch": 0.07205683355886333,
      "grad_norm": 0.19949902594089508,
      "learning_rate": 4.997951168538693e-06,
      "loss": 0.6651,
      "step": 213
    },
    {
      "epoch": 0.07239512855209743,
      "grad_norm": 0.20261453092098236,
      "learning_rate": 4.99791475064563e-06,
      "loss": 0.6809,
      "step": 214
    },
    {
      "epoch": 0.07273342354533154,
      "grad_norm": 0.19712717831134796,
      "learning_rate": 4.997878012069322e-06,
      "loss": 0.6899,
      "step": 215
    },
    {
      "epoch": 0.07307171853856563,
      "grad_norm": 0.2066112756729126,
      "learning_rate": 4.997840952814484e-06,
      "loss": 0.7037,
      "step": 216
    },
    {
      "epoch": 0.07341001353179973,
      "grad_norm": 0.19402629137039185,
      "learning_rate": 4.997803572885874e-06,
      "loss": 0.68,
      "step": 217
    },
    {
      "epoch": 0.07374830852503383,
      "grad_norm": 0.19789829850196838,
      "learning_rate": 4.997765872288292e-06,
      "loss": 0.7236,
      "step": 218
    },
    {
      "epoch": 0.07408660351826793,
      "grad_norm": 0.20883303880691528,
      "learning_rate": 4.997727851026576e-06,
      "loss": 0.6912,
      "step": 219
    },
    {
      "epoch": 0.07442489851150202,
      "grad_norm": 0.2045092135667801,
      "learning_rate": 4.997689509105608e-06,
      "loss": 0.6858,
      "step": 220
    },
    {
      "epoch": 0.07476319350473613,
      "grad_norm": 0.20657972991466522,
      "learning_rate": 4.997650846530311e-06,
      "loss": 0.7066,
      "step": 221
    },
    {
      "epoch": 0.07510148849797023,
      "grad_norm": 0.2036881297826767,
      "learning_rate": 4.997611863305647e-06,
      "loss": 0.6891,
      "step": 222
    },
    {
      "epoch": 0.07543978349120432,
      "grad_norm": 0.20515350997447968,
      "learning_rate": 4.997572559436623e-06,
      "loss": 0.7125,
      "step": 223
    },
    {
      "epoch": 0.07577807848443843,
      "grad_norm": 0.19229350984096527,
      "learning_rate": 4.997532934928283e-06,
      "loss": 0.6614,
      "step": 224
    },
    {
      "epoch": 0.07611637347767253,
      "grad_norm": 0.20285889506340027,
      "learning_rate": 4.997492989785714e-06,
      "loss": 0.7085,
      "step": 225
    },
    {
      "epoch": 0.07645466847090664,
      "grad_norm": 0.19925077259540558,
      "learning_rate": 4.9974527240140455e-06,
      "loss": 0.7113,
      "step": 226
    },
    {
      "epoch": 0.07679296346414073,
      "grad_norm": 0.1916913241147995,
      "learning_rate": 4.997412137618446e-06,
      "loss": 0.6637,
      "step": 227
    },
    {
      "epoch": 0.07713125845737483,
      "grad_norm": 0.20506256818771362,
      "learning_rate": 4.997371230604126e-06,
      "loss": 0.6923,
      "step": 228
    },
    {
      "epoch": 0.07746955345060894,
      "grad_norm": 0.20287774503231049,
      "learning_rate": 4.997330002976337e-06,
      "loss": 0.6908,
      "step": 229
    },
    {
      "epoch": 0.07780784844384303,
      "grad_norm": 0.20252133905887604,
      "learning_rate": 4.997288454740372e-06,
      "loss": 0.6796,
      "step": 230
    },
    {
      "epoch": 0.07814614343707713,
      "grad_norm": 0.1976034790277481,
      "learning_rate": 4.997246585901565e-06,
      "loss": 0.7089,
      "step": 231
    },
    {
      "epoch": 0.07848443843031123,
      "grad_norm": 0.2120821177959442,
      "learning_rate": 4.997204396465292e-06,
      "loss": 0.6763,
      "step": 232
    },
    {
      "epoch": 0.07882273342354533,
      "grad_norm": 0.20599138736724854,
      "learning_rate": 4.9971618864369676e-06,
      "loss": 0.6834,
      "step": 233
    },
    {
      "epoch": 0.07916102841677942,
      "grad_norm": 0.20934651792049408,
      "learning_rate": 4.99711905582205e-06,
      "loss": 0.6665,
      "step": 234
    },
    {
      "epoch": 0.07949932341001353,
      "grad_norm": 0.204156294465065,
      "learning_rate": 4.9970759046260385e-06,
      "loss": 0.6802,
      "step": 235
    },
    {
      "epoch": 0.07983761840324763,
      "grad_norm": 0.20378589630126953,
      "learning_rate": 4.997032432854472e-06,
      "loss": 0.7043,
      "step": 236
    },
    {
      "epoch": 0.08017591339648174,
      "grad_norm": 0.20697647333145142,
      "learning_rate": 4.996988640512931e-06,
      "loss": 0.6939,
      "step": 237
    },
    {
      "epoch": 0.08051420838971583,
      "grad_norm": 0.2135564535856247,
      "learning_rate": 4.9969445276070395e-06,
      "loss": 0.7218,
      "step": 238
    },
    {
      "epoch": 0.08085250338294993,
      "grad_norm": 0.2033076286315918,
      "learning_rate": 4.996900094142459e-06,
      "loss": 0.6678,
      "step": 239
    },
    {
      "epoch": 0.08119079837618404,
      "grad_norm": 0.20832550525665283,
      "learning_rate": 4.996855340124894e-06,
      "loss": 0.6908,
      "step": 240
    },
    {
      "epoch": 0.08152909336941813,
      "grad_norm": 0.20612551271915436,
      "learning_rate": 4.996810265560091e-06,
      "loss": 0.6838,
      "step": 241
    },
    {
      "epoch": 0.08186738836265223,
      "grad_norm": 0.20617735385894775,
      "learning_rate": 4.996764870453836e-06,
      "loss": 0.6855,
      "step": 242
    },
    {
      "epoch": 0.08220568335588634,
      "grad_norm": 0.21661782264709473,
      "learning_rate": 4.996719154811956e-06,
      "loss": 0.6565,
      "step": 243
    },
    {
      "epoch": 0.08254397834912043,
      "grad_norm": 0.20216654241085052,
      "learning_rate": 4.996673118640323e-06,
      "loss": 0.7049,
      "step": 244
    },
    {
      "epoch": 0.08288227334235454,
      "grad_norm": 0.21835078299045563,
      "learning_rate": 4.996626761944844e-06,
      "loss": 0.6743,
      "step": 245
    },
    {
      "epoch": 0.08322056833558863,
      "grad_norm": 0.21051070094108582,
      "learning_rate": 4.9965800847314705e-06,
      "loss": 0.678,
      "step": 246
    },
    {
      "epoch": 0.08355886332882273,
      "grad_norm": 0.2078743427991867,
      "learning_rate": 4.996533087006197e-06,
      "loss": 0.6791,
      "step": 247
    },
    {
      "epoch": 0.08389715832205684,
      "grad_norm": 0.20969849824905396,
      "learning_rate": 4.996485768775056e-06,
      "loss": 0.6473,
      "step": 248
    },
    {
      "epoch": 0.08423545331529093,
      "grad_norm": 0.20665079355239868,
      "learning_rate": 4.996438130044121e-06,
      "loss": 0.673,
      "step": 249
    },
    {
      "epoch": 0.08457374830852503,
      "grad_norm": 0.2039971500635147,
      "learning_rate": 4.99639017081951e-06,
      "loss": 0.6727,
      "step": 250
    },
    {
      "epoch": 0.08491204330175914,
      "grad_norm": 0.21164476871490479,
      "learning_rate": 4.996341891107379e-06,
      "loss": 0.6526,
      "step": 251
    },
    {
      "epoch": 0.08525033829499323,
      "grad_norm": 0.2099243700504303,
      "learning_rate": 4.9962932909139265e-06,
      "loss": 0.6831,
      "step": 252
    },
    {
      "epoch": 0.08558863328822733,
      "grad_norm": 0.21000251173973083,
      "learning_rate": 4.99624437024539e-06,
      "loss": 0.6917,
      "step": 253
    },
    {
      "epoch": 0.08592692828146144,
      "grad_norm": 0.2100362479686737,
      "learning_rate": 4.996195129108053e-06,
      "loss": 0.698,
      "step": 254
    },
    {
      "epoch": 0.08626522327469553,
      "grad_norm": 0.20553170144557953,
      "learning_rate": 4.996145567508235e-06,
      "loss": 0.6822,
      "step": 255
    },
    {
      "epoch": 0.08660351826792964,
      "grad_norm": 0.20825114846229553,
      "learning_rate": 4.996095685452299e-06,
      "loss": 0.6884,
      "step": 256
    },
    {
      "epoch": 0.08694181326116374,
      "grad_norm": 0.19606679677963257,
      "learning_rate": 4.99604548294665e-06,
      "loss": 0.6707,
      "step": 257
    },
    {
      "epoch": 0.08728010825439783,
      "grad_norm": 0.2105158567428589,
      "learning_rate": 4.995994959997731e-06,
      "loss": 0.7002,
      "step": 258
    },
    {
      "epoch": 0.08761840324763194,
      "grad_norm": 0.21269915997982025,
      "learning_rate": 4.995944116612031e-06,
      "loss": 0.6952,
      "step": 259
    },
    {
      "epoch": 0.08795669824086604,
      "grad_norm": 0.21205005049705505,
      "learning_rate": 4.995892952796074e-06,
      "loss": 0.6893,
      "step": 260
    },
    {
      "epoch": 0.08829499323410013,
      "grad_norm": 0.22121933102607727,
      "learning_rate": 4.995841468556432e-06,
      "loss": 0.6896,
      "step": 261
    },
    {
      "epoch": 0.08863328822733424,
      "grad_norm": 0.20240426063537598,
      "learning_rate": 4.995789663899711e-06,
      "loss": 0.6832,
      "step": 262
    },
    {
      "epoch": 0.08897158322056833,
      "grad_norm": 0.20699365437030792,
      "learning_rate": 4.995737538832564e-06,
      "loss": 0.6788,
      "step": 263
    },
    {
      "epoch": 0.08930987821380243,
      "grad_norm": 0.2231341451406479,
      "learning_rate": 4.995685093361682e-06,
      "loss": 0.6799,
      "step": 264
    },
    {
      "epoch": 0.08964817320703654,
      "grad_norm": 0.21361176669597626,
      "learning_rate": 4.9956323274937975e-06,
      "loss": 0.7036,
      "step": 265
    },
    {
      "epoch": 0.08998646820027063,
      "grad_norm": 0.21375077962875366,
      "learning_rate": 4.9955792412356865e-06,
      "loss": 0.673,
      "step": 266
    },
    {
      "epoch": 0.09032476319350474,
      "grad_norm": 0.2075895369052887,
      "learning_rate": 4.9955258345941625e-06,
      "loss": 0.691,
      "step": 267
    },
    {
      "epoch": 0.09066305818673884,
      "grad_norm": 0.20629572868347168,
      "learning_rate": 4.995472107576083e-06,
      "loss": 0.7092,
      "step": 268
    },
    {
      "epoch": 0.09100135317997293,
      "grad_norm": 0.1996820867061615,
      "learning_rate": 4.995418060188344e-06,
      "loss": 0.6688,
      "step": 269
    },
    {
      "epoch": 0.09133964817320704,
      "grad_norm": 0.20391984283924103,
      "learning_rate": 4.995363692437884e-06,
      "loss": 0.6982,
      "step": 270
    },
    {
      "epoch": 0.09167794316644114,
      "grad_norm": 0.21628272533416748,
      "learning_rate": 4.995309004331686e-06,
      "loss": 0.662,
      "step": 271
    },
    {
      "epoch": 0.09201623815967523,
      "grad_norm": 0.20856939256191254,
      "learning_rate": 4.9952539958767675e-06,
      "loss": 0.6523,
      "step": 272
    },
    {
      "epoch": 0.09235453315290934,
      "grad_norm": 0.21313869953155518,
      "learning_rate": 4.995198667080191e-06,
      "loss": 0.6743,
      "step": 273
    },
    {
      "epoch": 0.09269282814614344,
      "grad_norm": 0.2262369841337204,
      "learning_rate": 4.995143017949061e-06,
      "loss": 0.6894,
      "step": 274
    },
    {
      "epoch": 0.09303112313937753,
      "grad_norm": 0.21210289001464844,
      "learning_rate": 4.995087048490521e-06,
      "loss": 0.6546,
      "step": 275
    },
    {
      "epoch": 0.09336941813261164,
      "grad_norm": 0.2152973711490631,
      "learning_rate": 4.995030758711756e-06,
      "loss": 0.7006,
      "step": 276
    },
    {
      "epoch": 0.09370771312584573,
      "grad_norm": 0.21384137868881226,
      "learning_rate": 4.994974148619994e-06,
      "loss": 0.6973,
      "step": 277
    },
    {
      "epoch": 0.09404600811907984,
      "grad_norm": 0.22062373161315918,
      "learning_rate": 4.9949172182225e-06,
      "loss": 0.6732,
      "step": 278
    },
    {
      "epoch": 0.09438430311231394,
      "grad_norm": 0.21281228959560394,
      "learning_rate": 4.9948599675265855e-06,
      "loss": 0.6917,
      "step": 279
    },
    {
      "epoch": 0.09472259810554803,
      "grad_norm": 0.2118413895368576,
      "learning_rate": 4.994802396539599e-06,
      "loss": 0.6764,
      "step": 280
    },
    {
      "epoch": 0.09506089309878214,
      "grad_norm": 0.20453035831451416,
      "learning_rate": 4.99474450526893e-06,
      "loss": 0.6785,
      "step": 281
    },
    {
      "epoch": 0.09539918809201624,
      "grad_norm": 0.20882472395896912,
      "learning_rate": 4.994686293722014e-06,
      "loss": 0.6935,
      "step": 282
    },
    {
      "epoch": 0.09573748308525033,
      "grad_norm": 0.22508378326892853,
      "learning_rate": 4.9946277619063225e-06,
      "loss": 0.6787,
      "step": 283
    },
    {
      "epoch": 0.09607577807848444,
      "grad_norm": 0.2192172408103943,
      "learning_rate": 4.994568909829369e-06,
      "loss": 0.6843,
      "step": 284
    },
    {
      "epoch": 0.09641407307171854,
      "grad_norm": 0.2155693918466568,
      "learning_rate": 4.9945097374987096e-06,
      "loss": 0.6758,
      "step": 285
    },
    {
      "epoch": 0.09675236806495263,
      "grad_norm": 0.21563497185707092,
      "learning_rate": 4.994450244921941e-06,
      "loss": 0.6792,
      "step": 286
    },
    {
      "epoch": 0.09709066305818674,
      "grad_norm": 0.2330203652381897,
      "learning_rate": 4.994390432106702e-06,
      "loss": 0.6961,
      "step": 287
    },
    {
      "epoch": 0.09742895805142084,
      "grad_norm": 0.21809963881969452,
      "learning_rate": 4.994330299060668e-06,
      "loss": 0.6747,
      "step": 288
    },
    {
      "epoch": 0.09776725304465494,
      "grad_norm": 0.21034228801727295,
      "learning_rate": 4.994269845791563e-06,
      "loss": 0.7135,
      "step": 289
    },
    {
      "epoch": 0.09810554803788904,
      "grad_norm": 0.21551263332366943,
      "learning_rate": 4.9942090723071454e-06,
      "loss": 0.6612,
      "step": 290
    },
    {
      "epoch": 0.09844384303112313,
      "grad_norm": 0.22050173580646515,
      "learning_rate": 4.994147978615219e-06,
      "loss": 0.6871,
      "step": 291
    },
    {
      "epoch": 0.09878213802435724,
      "grad_norm": 0.21225669980049133,
      "learning_rate": 4.9940865647236255e-06,
      "loss": 0.6901,
      "step": 292
    },
    {
      "epoch": 0.09912043301759134,
      "grad_norm": 0.20882059633731842,
      "learning_rate": 4.99402483064025e-06,
      "loss": 0.6834,
      "step": 293
    },
    {
      "epoch": 0.09945872801082543,
      "grad_norm": 0.22327478229999542,
      "learning_rate": 4.993962776373018e-06,
      "loss": 0.6978,
      "step": 294
    },
    {
      "epoch": 0.09979702300405954,
      "grad_norm": 0.21540549397468567,
      "learning_rate": 4.9939004019298955e-06,
      "loss": 0.6869,
      "step": 295
    },
    {
      "epoch": 0.10013531799729364,
      "grad_norm": 0.2194754183292389,
      "learning_rate": 4.99383770731889e-06,
      "loss": 0.6952,
      "step": 296
    },
    {
      "epoch": 0.10047361299052775,
      "grad_norm": 0.21692237257957458,
      "learning_rate": 4.993774692548052e-06,
      "loss": 0.6973,
      "step": 297
    },
    {
      "epoch": 0.10081190798376184,
      "grad_norm": 0.21536946296691895,
      "learning_rate": 4.993711357625469e-06,
      "loss": 0.6721,
      "step": 298
    },
    {
      "epoch": 0.10115020297699594,
      "grad_norm": 0.213889017701149,
      "learning_rate": 4.993647702559274e-06,
      "loss": 0.7035,
      "step": 299
    },
    {
      "epoch": 0.10148849797023005,
      "grad_norm": 0.22723858058452606,
      "learning_rate": 4.993583727357638e-06,
      "loss": 0.677,
      "step": 300
    },
    {
      "epoch": 0.10182679296346414,
      "grad_norm": 0.21247157454490662,
      "learning_rate": 4.993519432028774e-06,
      "loss": 0.6962,
      "step": 301
    },
    {
      "epoch": 0.10216508795669824,
      "grad_norm": 0.21780304610729218,
      "learning_rate": 4.993454816580937e-06,
      "loss": 0.7032,
      "step": 302
    },
    {
      "epoch": 0.10250338294993235,
      "grad_norm": 0.2129039615392685,
      "learning_rate": 4.993389881022421e-06,
      "loss": 0.6654,
      "step": 303
    },
    {
      "epoch": 0.10284167794316644,
      "grad_norm": 0.21771487593650818,
      "learning_rate": 4.9933246253615645e-06,
      "loss": 0.6745,
      "step": 304
    },
    {
      "epoch": 0.10317997293640054,
      "grad_norm": 0.21839098632335663,
      "learning_rate": 4.993259049606743e-06,
      "loss": 0.6797,
      "step": 305
    },
    {
      "epoch": 0.10351826792963464,
      "grad_norm": 0.23275397717952728,
      "learning_rate": 4.9931931537663776e-06,
      "loss": 0.6763,
      "step": 306
    },
    {
      "epoch": 0.10385656292286874,
      "grad_norm": 0.20738539099693298,
      "learning_rate": 4.9931269378489255e-06,
      "loss": 0.6942,
      "step": 307
    },
    {
      "epoch": 0.10419485791610285,
      "grad_norm": 0.22642409801483154,
      "learning_rate": 4.993060401862888e-06,
      "loss": 0.7125,
      "step": 308
    },
    {
      "epoch": 0.10453315290933694,
      "grad_norm": 0.21417498588562012,
      "learning_rate": 4.992993545816809e-06,
      "loss": 0.6696,
      "step": 309
    },
    {
      "epoch": 0.10487144790257104,
      "grad_norm": 0.2231566309928894,
      "learning_rate": 4.9929263697192685e-06,
      "loss": 0.6865,
      "step": 310
    },
    {
      "epoch": 0.10520974289580515,
      "grad_norm": 0.23336437344551086,
      "learning_rate": 4.992858873578893e-06,
      "loss": 0.6897,
      "step": 311
    },
    {
      "epoch": 0.10554803788903924,
      "grad_norm": 0.2189081758260727,
      "learning_rate": 4.992791057404346e-06,
      "loss": 0.6653,
      "step": 312
    },
    {
      "epoch": 0.10588633288227334,
      "grad_norm": 0.21535345911979675,
      "learning_rate": 4.992722921204336e-06,
      "loss": 0.6718,
      "step": 313
    },
    {
      "epoch": 0.10622462787550745,
      "grad_norm": 0.2202640026807785,
      "learning_rate": 4.992654464987608e-06,
      "loss": 0.7185,
      "step": 314
    },
    {
      "epoch": 0.10656292286874154,
      "grad_norm": 0.23366160690784454,
      "learning_rate": 4.992585688762951e-06,
      "loss": 0.7022,
      "step": 315
    },
    {
      "epoch": 0.10690121786197564,
      "grad_norm": 0.21393480896949768,
      "learning_rate": 4.992516592539196e-06,
      "loss": 0.7088,
      "step": 316
    },
    {
      "epoch": 0.10723951285520975,
      "grad_norm": 0.23210909962654114,
      "learning_rate": 4.992447176325212e-06,
      "loss": 0.6905,
      "step": 317
    },
    {
      "epoch": 0.10757780784844384,
      "grad_norm": 0.20842531323432922,
      "learning_rate": 4.992377440129911e-06,
      "loss": 0.6709,
      "step": 318
    },
    {
      "epoch": 0.10791610284167795,
      "grad_norm": 0.21786710619926453,
      "learning_rate": 4.992307383962246e-06,
      "loss": 0.6912,
      "step": 319
    },
    {
      "epoch": 0.10825439783491204,
      "grad_norm": 0.21988864243030548,
      "learning_rate": 4.992237007831211e-06,
      "loss": 0.6735,
      "step": 320
    },
    {
      "epoch": 0.10859269282814614,
      "grad_norm": 0.22171525657176971,
      "learning_rate": 4.992166311745841e-06,
      "loss": 0.6923,
      "step": 321
    },
    {
      "epoch": 0.10893098782138025,
      "grad_norm": 0.22636902332305908,
      "learning_rate": 4.992095295715211e-06,
      "loss": 0.6709,
      "step": 322
    },
    {
      "epoch": 0.10926928281461434,
      "grad_norm": 0.2151307761669159,
      "learning_rate": 4.992023959748439e-06,
      "loss": 0.6878,
      "step": 323
    },
    {
      "epoch": 0.10960757780784844,
      "grad_norm": 0.22044600546360016,
      "learning_rate": 4.991952303854683e-06,
      "loss": 0.7115,
      "step": 324
    },
    {
      "epoch": 0.10994587280108255,
      "grad_norm": 0.21665897965431213,
      "learning_rate": 4.991880328043141e-06,
      "loss": 0.704,
      "step": 325
    },
    {
      "epoch": 0.11028416779431664,
      "grad_norm": 0.22031225264072418,
      "learning_rate": 4.991808032323056e-06,
      "loss": 0.6744,
      "step": 326
    },
    {
      "epoch": 0.11062246278755074,
      "grad_norm": 0.22540146112442017,
      "learning_rate": 4.991735416703707e-06,
      "loss": 0.6606,
      "step": 327
    },
    {
      "epoch": 0.11096075778078485,
      "grad_norm": 0.2185710072517395,
      "learning_rate": 4.9916624811944175e-06,
      "loss": 0.6654,
      "step": 328
    },
    {
      "epoch": 0.11129905277401894,
      "grad_norm": 0.2213398814201355,
      "learning_rate": 4.991589225804551e-06,
      "loss": 0.6684,
      "step": 329
    },
    {
      "epoch": 0.11163734776725305,
      "grad_norm": 0.22903922200202942,
      "learning_rate": 4.99151565054351e-06,
      "loss": 0.6736,
      "step": 330
    },
    {
      "epoch": 0.11197564276048715,
      "grad_norm": 0.21496374905109406,
      "learning_rate": 4.9914417554207434e-06,
      "loss": 0.6889,
      "step": 331
    },
    {
      "epoch": 0.11231393775372124,
      "grad_norm": 0.216947540640831,
      "learning_rate": 4.991367540445736e-06,
      "loss": 0.6847,
      "step": 332
    },
    {
      "epoch": 0.11265223274695535,
      "grad_norm": 0.2170547991991043,
      "learning_rate": 4.991293005628015e-06,
      "loss": 0.6734,
      "step": 333
    },
    {
      "epoch": 0.11299052774018944,
      "grad_norm": 0.22944290935993195,
      "learning_rate": 4.9912181509771505e-06,
      "loss": 0.6554,
      "step": 334
    },
    {
      "epoch": 0.11332882273342354,
      "grad_norm": 0.22795914113521576,
      "learning_rate": 4.991142976502753e-06,
      "loss": 0.6582,
      "step": 335
    },
    {
      "epoch": 0.11366711772665765,
      "grad_norm": 0.22316622734069824,
      "learning_rate": 4.991067482214471e-06,
      "loss": 0.6872,
      "step": 336
    },
    {
      "epoch": 0.11400541271989174,
      "grad_norm": 0.2117655873298645,
      "learning_rate": 4.990991668121998e-06,
      "loss": 0.6567,
      "step": 337
    },
    {
      "epoch": 0.11434370771312584,
      "grad_norm": 0.22236984968185425,
      "learning_rate": 4.9909155342350675e-06,
      "loss": 0.6716,
      "step": 338
    },
    {
      "epoch": 0.11468200270635995,
      "grad_norm": 0.21970327198505402,
      "learning_rate": 4.990839080563452e-06,
      "loss": 0.6748,
      "step": 339
    },
    {
      "epoch": 0.11502029769959404,
      "grad_norm": 0.23046572506427765,
      "learning_rate": 4.990762307116969e-06,
      "loss": 0.6816,
      "step": 340
    },
    {
      "epoch": 0.11535859269282815,
      "grad_norm": 0.23384122550487518,
      "learning_rate": 4.990685213905472e-06,
      "loss": 0.6724,
      "step": 341
    },
    {
      "epoch": 0.11569688768606225,
      "grad_norm": 0.20976535975933075,
      "learning_rate": 4.99060780093886e-06,
      "loss": 0.6582,
      "step": 342
    },
    {
      "epoch": 0.11603518267929634,
      "grad_norm": 0.21732792258262634,
      "learning_rate": 4.990530068227072e-06,
      "loss": 0.6678,
      "step": 343
    },
    {
      "epoch": 0.11637347767253045,
      "grad_norm": 0.21509791910648346,
      "learning_rate": 4.990452015780085e-06,
      "loss": 0.715,
      "step": 344
    },
    {
      "epoch": 0.11671177266576455,
      "grad_norm": 0.21847152709960938,
      "learning_rate": 4.9903736436079205e-06,
      "loss": 0.6756,
      "step": 345
    },
    {
      "epoch": 0.11705006765899864,
      "grad_norm": 0.21810345351696014,
      "learning_rate": 4.9902949517206415e-06,
      "loss": 0.6664,
      "step": 346
    },
    {
      "epoch": 0.11738836265223275,
      "grad_norm": 0.21546834707260132,
      "learning_rate": 4.990215940128348e-06,
      "loss": 0.6773,
      "step": 347
    },
    {
      "epoch": 0.11772665764546685,
      "grad_norm": 0.2323540896177292,
      "learning_rate": 4.990136608841185e-06,
      "loss": 0.685,
      "step": 348
    },
    {
      "epoch": 0.11806495263870095,
      "grad_norm": 0.22190017998218536,
      "learning_rate": 4.990056957869336e-06,
      "loss": 0.6909,
      "step": 349
    },
    {
      "epoch": 0.11840324763193505,
      "grad_norm": 0.2267863005399704,
      "learning_rate": 4.989976987223027e-06,
      "loss": 0.6703,
      "step": 350
    },
    {
      "epoch": 0.11874154262516914,
      "grad_norm": 0.22832167148590088,
      "learning_rate": 4.9898966969125265e-06,
      "loss": 0.6806,
      "step": 351
    },
    {
      "epoch": 0.11907983761840325,
      "grad_norm": 0.24483756721019745,
      "learning_rate": 4.98981608694814e-06,
      "loss": 0.6898,
      "step": 352
    },
    {
      "epoch": 0.11941813261163735,
      "grad_norm": 0.220206618309021,
      "learning_rate": 4.989735157340218e-06,
      "loss": 0.6824,
      "step": 353
    },
    {
      "epoch": 0.11975642760487144,
      "grad_norm": 0.2350161224603653,
      "learning_rate": 4.989653908099148e-06,
      "loss": 0.6722,
      "step": 354
    },
    {
      "epoch": 0.12009472259810555,
      "grad_norm": 0.23080413043498993,
      "learning_rate": 4.989572339235361e-06,
      "loss": 0.653,
      "step": 355
    },
    {
      "epoch": 0.12043301759133965,
      "grad_norm": 0.23627646267414093,
      "learning_rate": 4.989490450759331e-06,
      "loss": 0.6642,
      "step": 356
    },
    {
      "epoch": 0.12077131258457374,
      "grad_norm": 0.22271862626075745,
      "learning_rate": 4.98940824268157e-06,
      "loss": 0.6768,
      "step": 357
    },
    {
      "epoch": 0.12110960757780785,
      "grad_norm": 0.2121579349040985,
      "learning_rate": 4.989325715012632e-06,
      "loss": 0.6807,
      "step": 358
    },
    {
      "epoch": 0.12144790257104195,
      "grad_norm": 0.23613037168979645,
      "learning_rate": 4.989242867763111e-06,
      "loss": 0.7094,
      "step": 359
    },
    {
      "epoch": 0.12178619756427606,
      "grad_norm": 0.21646465361118317,
      "learning_rate": 4.989159700943643e-06,
      "loss": 0.6835,
      "step": 360
    },
    {
      "epoch": 0.12212449255751015,
      "grad_norm": 0.2305869609117508,
      "learning_rate": 4.989076214564906e-06,
      "loss": 0.6884,
      "step": 361
    },
    {
      "epoch": 0.12246278755074425,
      "grad_norm": 0.2341344654560089,
      "learning_rate": 4.988992408637618e-06,
      "loss": 0.6497,
      "step": 362
    },
    {
      "epoch": 0.12280108254397835,
      "grad_norm": 0.22924962639808655,
      "learning_rate": 4.9889082831725375e-06,
      "loss": 0.6816,
      "step": 363
    },
    {
      "epoch": 0.12313937753721245,
      "grad_norm": 0.2256554514169693,
      "learning_rate": 4.988823838180464e-06,
      "loss": 0.6962,
      "step": 364
    },
    {
      "epoch": 0.12347767253044654,
      "grad_norm": 0.2216321974992752,
      "learning_rate": 4.98873907367224e-06,
      "loss": 0.6787,
      "step": 365
    },
    {
      "epoch": 0.12381596752368065,
      "grad_norm": 0.22423723340034485,
      "learning_rate": 4.988653989658746e-06,
      "loss": 0.6891,
      "step": 366
    },
    {
      "epoch": 0.12415426251691475,
      "grad_norm": 0.23579758405685425,
      "learning_rate": 4.988568586150906e-06,
      "loss": 0.7101,
      "step": 367
    },
    {
      "epoch": 0.12449255751014884,
      "grad_norm": 0.24115969240665436,
      "learning_rate": 4.988482863159684e-06,
      "loss": 0.6762,
      "step": 368
    },
    {
      "epoch": 0.12483085250338295,
      "grad_norm": 0.2256184071302414,
      "learning_rate": 4.988396820696086e-06,
      "loss": 0.71,
      "step": 369
    },
    {
      "epoch": 0.12516914749661706,
      "grad_norm": 0.23354458808898926,
      "learning_rate": 4.988310458771156e-06,
      "loss": 0.6777,
      "step": 370
    },
    {
      "epoch": 0.12550744248985116,
      "grad_norm": 0.23490329086780548,
      "learning_rate": 4.988223777395984e-06,
      "loss": 0.6833,
      "step": 371
    },
    {
      "epoch": 0.12584573748308525,
      "grad_norm": 0.2223309725522995,
      "learning_rate": 4.988136776581697e-06,
      "loss": 0.6554,
      "step": 372
    },
    {
      "epoch": 0.12618403247631935,
      "grad_norm": 0.22902588546276093,
      "learning_rate": 4.988049456339462e-06,
      "loss": 0.6585,
      "step": 373
    },
    {
      "epoch": 0.12652232746955344,
      "grad_norm": 0.22965329885482788,
      "learning_rate": 4.987961816680493e-06,
      "loss": 0.6896,
      "step": 374
    },
    {
      "epoch": 0.12686062246278756,
      "grad_norm": 0.235035702586174,
      "learning_rate": 4.987873857616038e-06,
      "loss": 0.675,
      "step": 375
    },
    {
      "epoch": 0.12719891745602166,
      "grad_norm": 0.23586681485176086,
      "learning_rate": 4.987785579157392e-06,
      "loss": 0.6887,
      "step": 376
    },
    {
      "epoch": 0.12753721244925575,
      "grad_norm": 0.22219079732894897,
      "learning_rate": 4.987696981315885e-06,
      "loss": 0.6761,
      "step": 377
    },
    {
      "epoch": 0.12787550744248985,
      "grad_norm": 0.22536414861679077,
      "learning_rate": 4.987608064102895e-06,
      "loss": 0.6639,
      "step": 378
    },
    {
      "epoch": 0.12821380243572394,
      "grad_norm": 0.2327205091714859,
      "learning_rate": 4.987518827529833e-06,
      "loss": 0.6817,
      "step": 379
    },
    {
      "epoch": 0.12855209742895804,
      "grad_norm": 0.22478225827217102,
      "learning_rate": 4.987429271608159e-06,
      "loss": 0.6531,
      "step": 380
    },
    {
      "epoch": 0.12889039242219216,
      "grad_norm": 0.23206846415996552,
      "learning_rate": 4.987339396349369e-06,
      "loss": 0.6877,
      "step": 381
    },
    {
      "epoch": 0.12922868741542626,
      "grad_norm": 0.23701916635036469,
      "learning_rate": 4.987249201765001e-06,
      "loss": 0.6873,
      "step": 382
    },
    {
      "epoch": 0.12956698240866035,
      "grad_norm": 0.22712261974811554,
      "learning_rate": 4.9871586878666325e-06,
      "loss": 0.6491,
      "step": 383
    },
    {
      "epoch": 0.12990527740189445,
      "grad_norm": 0.23785048723220825,
      "learning_rate": 4.987067854665887e-06,
      "loss": 0.6892,
      "step": 384
    },
    {
      "epoch": 0.13024357239512854,
      "grad_norm": 0.22262883186340332,
      "learning_rate": 4.986976702174423e-06,
      "loss": 0.6904,
      "step": 385
    },
    {
      "epoch": 0.13058186738836267,
      "grad_norm": 0.22985069453716278,
      "learning_rate": 4.986885230403944e-06,
      "loss": 0.6414,
      "step": 386
    },
    {
      "epoch": 0.13092016238159676,
      "grad_norm": 0.21784214675426483,
      "learning_rate": 4.986793439366193e-06,
      "loss": 0.6617,
      "step": 387
    },
    {
      "epoch": 0.13125845737483086,
      "grad_norm": 0.2375217229127884,
      "learning_rate": 4.986701329072954e-06,
      "loss": 0.6718,
      "step": 388
    },
    {
      "epoch": 0.13159675236806495,
      "grad_norm": 0.2305949479341507,
      "learning_rate": 4.9866088995360524e-06,
      "loss": 0.6607,
      "step": 389
    },
    {
      "epoch": 0.13193504736129905,
      "grad_norm": 0.2283347249031067,
      "learning_rate": 4.986516150767354e-06,
      "loss": 0.6797,
      "step": 390
    },
    {
      "epoch": 0.13227334235453314,
      "grad_norm": 0.23237332701683044,
      "learning_rate": 4.986423082778766e-06,
      "loss": 0.6691,
      "step": 391
    },
    {
      "epoch": 0.13261163734776726,
      "grad_norm": 0.23644429445266724,
      "learning_rate": 4.986329695582237e-06,
      "loss": 0.6836,
      "step": 392
    },
    {
      "epoch": 0.13294993234100136,
      "grad_norm": 0.22994762659072876,
      "learning_rate": 4.986235989189756e-06,
      "loss": 0.6724,
      "step": 393
    },
    {
      "epoch": 0.13328822733423545,
      "grad_norm": 0.23970770835876465,
      "learning_rate": 4.986141963613352e-06,
      "loss": 0.6727,
      "step": 394
    },
    {
      "epoch": 0.13362652232746955,
      "grad_norm": 0.2304246574640274,
      "learning_rate": 4.986047618865097e-06,
      "loss": 0.6784,
      "step": 395
    },
    {
      "epoch": 0.13396481732070364,
      "grad_norm": 0.23033125698566437,
      "learning_rate": 4.985952954957103e-06,
      "loss": 0.6693,
      "step": 396
    },
    {
      "epoch": 0.13430311231393777,
      "grad_norm": 0.24728672206401825,
      "learning_rate": 4.985857971901522e-06,
      "loss": 0.7006,
      "step": 397
    },
    {
      "epoch": 0.13464140730717186,
      "grad_norm": 0.23562437295913696,
      "learning_rate": 4.98576266971055e-06,
      "loss": 0.6456,
      "step": 398
    },
    {
      "epoch": 0.13497970230040596,
      "grad_norm": 0.23348332941532135,
      "learning_rate": 4.98566704839642e-06,
      "loss": 0.7073,
      "step": 399
    },
    {
      "epoch": 0.13531799729364005,
      "grad_norm": 0.23460273444652557,
      "learning_rate": 4.985571107971408e-06,
      "loss": 0.6774,
      "step": 400
    },
    {
      "epoch": 0.13565629228687415,
      "grad_norm": 0.23990432918071747,
      "learning_rate": 4.9854748484478324e-06,
      "loss": 0.675,
      "step": 401
    },
    {
      "epoch": 0.13599458728010824,
      "grad_norm": 0.24199119210243225,
      "learning_rate": 4.98537826983805e-06,
      "loss": 0.6657,
      "step": 402
    },
    {
      "epoch": 0.13633288227334237,
      "grad_norm": 0.24207346141338348,
      "learning_rate": 4.985281372154459e-06,
      "loss": 0.6583,
      "step": 403
    },
    {
      "epoch": 0.13667117726657646,
      "grad_norm": 0.24412518739700317,
      "learning_rate": 4.9851841554095e-06,
      "loss": 0.7038,
      "step": 404
    },
    {
      "epoch": 0.13700947225981056,
      "grad_norm": 0.2288157343864441,
      "learning_rate": 4.985086619615655e-06,
      "loss": 0.6766,
      "step": 405
    },
    {
      "epoch": 0.13734776725304465,
      "grad_norm": 0.23269274830818176,
      "learning_rate": 4.984988764785443e-06,
      "loss": 0.6947,
      "step": 406
    },
    {
      "epoch": 0.13768606224627875,
      "grad_norm": 0.23585785925388336,
      "learning_rate": 4.984890590931427e-06,
      "loss": 0.6566,
      "step": 407
    },
    {
      "epoch": 0.13802435723951287,
      "grad_norm": 0.2309969663619995,
      "learning_rate": 4.9847920980662134e-06,
      "loss": 0.6573,
      "step": 408
    },
    {
      "epoch": 0.13836265223274696,
      "grad_norm": 0.23268240690231323,
      "learning_rate": 4.984693286202444e-06,
      "loss": 0.6722,
      "step": 409
    },
    {
      "epoch": 0.13870094722598106,
      "grad_norm": 0.23013322055339813,
      "learning_rate": 4.984594155352805e-06,
      "loss": 0.6639,
      "step": 410
    },
    {
      "epoch": 0.13903924221921515,
      "grad_norm": 0.23997361958026886,
      "learning_rate": 4.984494705530023e-06,
      "loss": 0.6633,
      "step": 411
    },
    {
      "epoch": 0.13937753721244925,
      "grad_norm": 0.23696044087409973,
      "learning_rate": 4.984394936746865e-06,
      "loss": 0.672,
      "step": 412
    },
    {
      "epoch": 0.13971583220568334,
      "grad_norm": 0.2297801524400711,
      "learning_rate": 4.98429484901614e-06,
      "loss": 0.6808,
      "step": 413
    },
    {
      "epoch": 0.14005412719891747,
      "grad_norm": 0.23803795874118805,
      "learning_rate": 4.984194442350698e-06,
      "loss": 0.6413,
      "step": 414
    },
    {
      "epoch": 0.14039242219215156,
      "grad_norm": 0.2315259426832199,
      "learning_rate": 4.984093716763427e-06,
      "loss": 0.6792,
      "step": 415
    },
    {
      "epoch": 0.14073071718538566,
      "grad_norm": 0.2283433973789215,
      "learning_rate": 4.983992672267261e-06,
      "loss": 0.6614,
      "step": 416
    },
    {
      "epoch": 0.14106901217861975,
      "grad_norm": 0.23342324793338776,
      "learning_rate": 4.98389130887517e-06,
      "loss": 0.6929,
      "step": 417
    },
    {
      "epoch": 0.14140730717185385,
      "grad_norm": 0.23194460570812225,
      "learning_rate": 4.983789626600168e-06,
      "loss": 0.6685,
      "step": 418
    },
    {
      "epoch": 0.14174560216508797,
      "grad_norm": 0.23261994123458862,
      "learning_rate": 4.983687625455307e-06,
      "loss": 0.6587,
      "step": 419
    },
    {
      "epoch": 0.14208389715832206,
      "grad_norm": 0.2293771505355835,
      "learning_rate": 4.983585305453685e-06,
      "loss": 0.6794,
      "step": 420
    },
    {
      "epoch": 0.14242219215155616,
      "grad_norm": 0.2353665828704834,
      "learning_rate": 4.983482666608436e-06,
      "loss": 0.6542,
      "step": 421
    },
    {
      "epoch": 0.14276048714479025,
      "grad_norm": 0.23223666846752167,
      "learning_rate": 4.9833797089327375e-06,
      "loss": 0.68,
      "step": 422
    },
    {
      "epoch": 0.14309878213802435,
      "grad_norm": 0.23364655673503876,
      "learning_rate": 4.983276432439808e-06,
      "loss": 0.6553,
      "step": 423
    },
    {
      "epoch": 0.14343707713125844,
      "grad_norm": 0.23806330561637878,
      "learning_rate": 4.983172837142904e-06,
      "loss": 0.6657,
      "step": 424
    },
    {
      "epoch": 0.14377537212449257,
      "grad_norm": 0.24118171632289886,
      "learning_rate": 4.983068923055327e-06,
      "loss": 0.6511,
      "step": 425
    },
    {
      "epoch": 0.14411366711772666,
      "grad_norm": 0.24000559747219086,
      "learning_rate": 4.982964690190417e-06,
      "loss": 0.6915,
      "step": 426
    },
    {
      "epoch": 0.14445196211096076,
      "grad_norm": 0.21783405542373657,
      "learning_rate": 4.982860138561555e-06,
      "loss": 0.6357,
      "step": 427
    },
    {
      "epoch": 0.14479025710419485,
      "grad_norm": 0.23208469152450562,
      "learning_rate": 4.9827552681821645e-06,
      "loss": 0.6702,
      "step": 428
    },
    {
      "epoch": 0.14512855209742895,
      "grad_norm": 0.2403120994567871,
      "learning_rate": 4.982650079065707e-06,
      "loss": 0.6619,
      "step": 429
    },
    {
      "epoch": 0.14546684709066307,
      "grad_norm": 0.24065937101840973,
      "learning_rate": 4.982544571225688e-06,
      "loss": 0.6672,
      "step": 430
    },
    {
      "epoch": 0.14580514208389717,
      "grad_norm": 0.23187507688999176,
      "learning_rate": 4.982438744675653e-06,
      "loss": 0.6505,
      "step": 431
    },
    {
      "epoch": 0.14614343707713126,
      "grad_norm": 0.22386693954467773,
      "learning_rate": 4.982332599429187e-06,
      "loss": 0.6761,
      "step": 432
    },
    {
      "epoch": 0.14648173207036536,
      "grad_norm": 0.23126639425754547,
      "learning_rate": 4.982226135499917e-06,
      "loss": 0.6824,
      "step": 433
    },
    {
      "epoch": 0.14682002706359945,
      "grad_norm": 0.24290819466114044,
      "learning_rate": 4.9821193529015124e-06,
      "loss": 0.6612,
      "step": 434
    },
    {
      "epoch": 0.14715832205683355,
      "grad_norm": 0.24675530195236206,
      "learning_rate": 4.98201225164768e-06,
      "loss": 0.6502,
      "step": 435
    },
    {
      "epoch": 0.14749661705006767,
      "grad_norm": 0.2415398508310318,
      "learning_rate": 4.98190483175217e-06,
      "loss": 0.6635,
      "step": 436
    },
    {
      "epoch": 0.14783491204330176,
      "grad_norm": 0.23464731872081757,
      "learning_rate": 4.981797093228775e-06,
      "loss": 0.6662,
      "step": 437
    },
    {
      "epoch": 0.14817320703653586,
      "grad_norm": 0.23631632328033447,
      "learning_rate": 4.981689036091323e-06,
      "loss": 0.6669,
      "step": 438
    },
    {
      "epoch": 0.14851150202976995,
      "grad_norm": 0.22991792857646942,
      "learning_rate": 4.98158066035369e-06,
      "loss": 0.7033,
      "step": 439
    },
    {
      "epoch": 0.14884979702300405,
      "grad_norm": 0.23498861491680145,
      "learning_rate": 4.981471966029787e-06,
      "loss": 0.6785,
      "step": 440
    },
    {
      "epoch": 0.14918809201623817,
      "grad_norm": 0.238022118806839,
      "learning_rate": 4.981362953133569e-06,
      "loss": 0.6858,
      "step": 441
    },
    {
      "epoch": 0.14952638700947227,
      "grad_norm": 0.22725965082645416,
      "learning_rate": 4.981253621679029e-06,
      "loss": 0.6651,
      "step": 442
    },
    {
      "epoch": 0.14986468200270636,
      "grad_norm": 0.2447727769613266,
      "learning_rate": 4.981143971680208e-06,
      "loss": 0.663,
      "step": 443
    },
    {
      "epoch": 0.15020297699594046,
      "grad_norm": 0.24607551097869873,
      "learning_rate": 4.9810340031511785e-06,
      "loss": 0.6717,
      "step": 444
    },
    {
      "epoch": 0.15054127198917455,
      "grad_norm": 0.22809375822544098,
      "learning_rate": 4.98092371610606e-06,
      "loss": 0.6569,
      "step": 445
    },
    {
      "epoch": 0.15087956698240865,
      "grad_norm": 0.22392520308494568,
      "learning_rate": 4.98081311055901e-06,
      "loss": 0.645,
      "step": 446
    },
    {
      "epoch": 0.15121786197564277,
      "grad_norm": 0.23991553485393524,
      "learning_rate": 4.98070218652423e-06,
      "loss": 0.6891,
      "step": 447
    },
    {
      "epoch": 0.15155615696887687,
      "grad_norm": 0.24203583598136902,
      "learning_rate": 4.9805909440159585e-06,
      "loss": 0.6836,
      "step": 448
    },
    {
      "epoch": 0.15189445196211096,
      "grad_norm": 0.23865444958209991,
      "learning_rate": 4.9804793830484785e-06,
      "loss": 0.6895,
      "step": 449
    },
    {
      "epoch": 0.15223274695534506,
      "grad_norm": 0.23681670427322388,
      "learning_rate": 4.980367503636111e-06,
      "loss": 0.6936,
      "step": 450
    },
    {
      "epoch": 0.15257104194857915,
      "grad_norm": 0.24201667308807373,
      "learning_rate": 4.9802553057932205e-06,
      "loss": 0.6647,
      "step": 451
    },
    {
      "epoch": 0.15290933694181327,
      "grad_norm": 0.23797008395195007,
      "learning_rate": 4.980142789534209e-06,
      "loss": 0.6518,
      "step": 452
    },
    {
      "epoch": 0.15324763193504737,
      "grad_norm": 0.25405654311180115,
      "learning_rate": 4.980029954873524e-06,
      "loss": 0.6798,
      "step": 453
    },
    {
      "epoch": 0.15358592692828146,
      "grad_norm": 0.24635176360607147,
      "learning_rate": 4.9799168018256486e-06,
      "loss": 0.6755,
      "step": 454
    },
    {
      "epoch": 0.15392422192151556,
      "grad_norm": 0.2485082596540451,
      "learning_rate": 4.9798033304051115e-06,
      "loss": 0.6922,
      "step": 455
    },
    {
      "epoch": 0.15426251691474965,
      "grad_norm": 0.26109230518341064,
      "learning_rate": 4.979689540626479e-06,
      "loss": 0.681,
      "step": 456
    },
    {
      "epoch": 0.15460081190798375,
      "grad_norm": 0.23434917628765106,
      "learning_rate": 4.979575432504359e-06,
      "loss": 0.6705,
      "step": 457
    },
    {
      "epoch": 0.15493910690121787,
      "grad_norm": 0.24609756469726562,
      "learning_rate": 4.979461006053403e-06,
      "loss": 0.6757,
      "step": 458
    },
    {
      "epoch": 0.15527740189445197,
      "grad_norm": 0.24599488079547882,
      "learning_rate": 4.979346261288298e-06,
      "loss": 0.6653,
      "step": 459
    },
    {
      "epoch": 0.15561569688768606,
      "grad_norm": 0.24822676181793213,
      "learning_rate": 4.979231198223778e-06,
      "loss": 0.6723,
      "step": 460
    },
    {
      "epoch": 0.15595399188092016,
      "grad_norm": 0.24261929094791412,
      "learning_rate": 4.979115816874613e-06,
      "loss": 0.6642,
      "step": 461
    },
    {
      "epoch": 0.15629228687415425,
      "grad_norm": 0.23589009046554565,
      "learning_rate": 4.979000117255616e-06,
      "loss": 0.6571,
      "step": 462
    },
    {
      "epoch": 0.15663058186738837,
      "grad_norm": 0.24605385959148407,
      "learning_rate": 4.978884099381642e-06,
      "loss": 0.6823,
      "step": 463
    },
    {
      "epoch": 0.15696887686062247,
      "grad_norm": 0.2397337108850479,
      "learning_rate": 4.9787677632675825e-06,
      "loss": 0.6515,
      "step": 464
    },
    {
      "epoch": 0.15730717185385656,
      "grad_norm": 0.22930362820625305,
      "learning_rate": 4.978651108928375e-06,
      "loss": 0.6656,
      "step": 465
    },
    {
      "epoch": 0.15764546684709066,
      "grad_norm": 0.24789506196975708,
      "learning_rate": 4.978534136378995e-06,
      "loss": 0.6671,
      "step": 466
    },
    {
      "epoch": 0.15798376184032475,
      "grad_norm": 0.22828440368175507,
      "learning_rate": 4.9784168456344595e-06,
      "loss": 0.6455,
      "step": 467
    },
    {
      "epoch": 0.15832205683355885,
      "grad_norm": 0.23598957061767578,
      "learning_rate": 4.978299236709826e-06,
      "loss": 0.645,
      "step": 468
    },
    {
      "epoch": 0.15866035182679297,
      "grad_norm": 0.24245992302894592,
      "learning_rate": 4.978181309620195e-06,
      "loss": 0.693,
      "step": 469
    },
    {
      "epoch": 0.15899864682002707,
      "grad_norm": 0.24275842308998108,
      "learning_rate": 4.978063064380703e-06,
      "loss": 0.6519,
      "step": 470
    },
    {
      "epoch": 0.15933694181326116,
      "grad_norm": 0.2353448122739792,
      "learning_rate": 4.977944501006532e-06,
      "loss": 0.6516,
      "step": 471
    },
    {
      "epoch": 0.15967523680649526,
      "grad_norm": 0.2516363859176636,
      "learning_rate": 4.977825619512905e-06,
      "loss": 0.6612,
      "step": 472
    },
    {
      "epoch": 0.16001353179972935,
      "grad_norm": 0.23496970534324646,
      "learning_rate": 4.9777064199150805e-06,
      "loss": 0.6503,
      "step": 473
    },
    {
      "epoch": 0.16035182679296348,
      "grad_norm": 0.23348847031593323,
      "learning_rate": 4.977586902228363e-06,
      "loss": 0.6835,
      "step": 474
    },
    {
      "epoch": 0.16069012178619757,
      "grad_norm": 0.23757898807525635,
      "learning_rate": 4.977467066468097e-06,
      "loss": 0.6397,
      "step": 475
    },
    {
      "epoch": 0.16102841677943167,
      "grad_norm": 0.23906610906124115,
      "learning_rate": 4.977346912649666e-06,
      "loss": 0.6589,
      "step": 476
    },
    {
      "epoch": 0.16136671177266576,
      "grad_norm": 0.23589485883712769,
      "learning_rate": 4.977226440788496e-06,
      "loss": 0.6431,
      "step": 477
    },
    {
      "epoch": 0.16170500676589986,
      "grad_norm": 0.24150773882865906,
      "learning_rate": 4.977105650900053e-06,
      "loss": 0.6895,
      "step": 478
    },
    {
      "epoch": 0.16204330175913398,
      "grad_norm": 0.25467047095298767,
      "learning_rate": 4.9769845429998445e-06,
      "loss": 0.6453,
      "step": 479
    },
    {
      "epoch": 0.16238159675236807,
      "grad_norm": 0.23450668156147003,
      "learning_rate": 4.976863117103417e-06,
      "loss": 0.6632,
      "step": 480
    },
    {
      "epoch": 0.16271989174560217,
      "grad_norm": 0.23753339052200317,
      "learning_rate": 4.976741373226361e-06,
      "loss": 0.6694,
      "step": 481
    },
    {
      "epoch": 0.16305818673883626,
      "grad_norm": 0.25083303451538086,
      "learning_rate": 4.976619311384305e-06,
      "loss": 0.665,
      "step": 482
    },
    {
      "epoch": 0.16339648173207036,
      "grad_norm": 0.23956729471683502,
      "learning_rate": 4.976496931592919e-06,
      "loss": 0.6765,
      "step": 483
    },
    {
      "epoch": 0.16373477672530445,
      "grad_norm": 0.24329128861427307,
      "learning_rate": 4.976374233867915e-06,
      "loss": 0.659,
      "step": 484
    },
    {
      "epoch": 0.16407307171853858,
      "grad_norm": 0.23810063302516937,
      "learning_rate": 4.976251218225045e-06,
      "loss": 0.656,
      "step": 485
    },
    {
      "epoch": 0.16441136671177267,
      "grad_norm": 0.24175657331943512,
      "learning_rate": 4.976127884680101e-06,
      "loss": 0.6672,
      "step": 486
    },
    {
      "epoch": 0.16474966170500677,
      "grad_norm": 0.23622599244117737,
      "learning_rate": 4.976004233248917e-06,
      "loss": 0.6676,
      "step": 487
    },
    {
      "epoch": 0.16508795669824086,
      "grad_norm": 0.24881932139396667,
      "learning_rate": 4.975880263947367e-06,
      "loss": 0.6813,
      "step": 488
    },
    {
      "epoch": 0.16542625169147496,
      "grad_norm": 0.23597869277000427,
      "learning_rate": 4.975755976791368e-06,
      "loss": 0.6424,
      "step": 489
    },
    {
      "epoch": 0.16576454668470908,
      "grad_norm": 0.23728232085704803,
      "learning_rate": 4.975631371796873e-06,
      "loss": 0.6821,
      "step": 490
    },
    {
      "epoch": 0.16610284167794317,
      "grad_norm": 0.2352864295244217,
      "learning_rate": 4.975506448979882e-06,
      "loss": 0.6864,
      "step": 491
    },
    {
      "epoch": 0.16644113667117727,
      "grad_norm": 0.245172381401062,
      "learning_rate": 4.975381208356431e-06,
      "loss": 0.7119,
      "step": 492
    },
    {
      "epoch": 0.16677943166441137,
      "grad_norm": 0.2432330846786499,
      "learning_rate": 4.975255649942598e-06,
      "loss": 0.6212,
      "step": 493
    },
    {
      "epoch": 0.16711772665764546,
      "grad_norm": 0.23403246700763702,
      "learning_rate": 4.975129773754503e-06,
      "loss": 0.6694,
      "step": 494
    },
    {
      "epoch": 0.16745602165087956,
      "grad_norm": 0.2418632060289383,
      "learning_rate": 4.975003579808306e-06,
      "loss": 0.6518,
      "step": 495
    },
    {
      "epoch": 0.16779431664411368,
      "grad_norm": 0.24580247700214386,
      "learning_rate": 4.974877068120208e-06,
      "loss": 0.6554,
      "step": 496
    },
    {
      "epoch": 0.16813261163734777,
      "grad_norm": 0.2346314638853073,
      "learning_rate": 4.97475023870645e-06,
      "loss": 0.6557,
      "step": 497
    },
    {
      "epoch": 0.16847090663058187,
      "grad_norm": 0.24950683116912842,
      "learning_rate": 4.974623091583316e-06,
      "loss": 0.6857,
      "step": 498
    },
    {
      "epoch": 0.16880920162381596,
      "grad_norm": 0.2463996857404709,
      "learning_rate": 4.974495626767126e-06,
      "loss": 0.6805,
      "step": 499
    },
    {
      "epoch": 0.16914749661705006,
      "grad_norm": 0.2514938712120056,
      "learning_rate": 4.974367844274248e-06,
      "loss": 0.6701,
      "step": 500
    },
    {
      "epoch": 0.16948579161028418,
      "grad_norm": 0.2525053322315216,
      "learning_rate": 4.974239744121083e-06,
      "loss": 0.676,
      "step": 501
    },
    {
      "epoch": 0.16982408660351828,
      "grad_norm": 0.2540481686592102,
      "learning_rate": 4.974111326324079e-06,
      "loss": 0.6734,
      "step": 502
    },
    {
      "epoch": 0.17016238159675237,
      "grad_norm": 0.2555442452430725,
      "learning_rate": 4.973982590899722e-06,
      "loss": 0.6532,
      "step": 503
    },
    {
      "epoch": 0.17050067658998647,
      "grad_norm": 0.24633650481700897,
      "learning_rate": 4.973853537864538e-06,
      "loss": 0.6913,
      "step": 504
    },
    {
      "epoch": 0.17083897158322056,
      "grad_norm": 0.24489474296569824,
      "learning_rate": 4.973724167235095e-06,
      "loss": 0.6379,
      "step": 505
    },
    {
      "epoch": 0.17117726657645466,
      "grad_norm": 0.2564601004123688,
      "learning_rate": 4.973594479028003e-06,
      "loss": 0.6836,
      "step": 506
    },
    {
      "epoch": 0.17151556156968878,
      "grad_norm": 0.24805857241153717,
      "learning_rate": 4.973464473259911e-06,
      "loss": 0.6795,
      "step": 507
    },
    {
      "epoch": 0.17185385656292287,
      "grad_norm": 0.24846164882183075,
      "learning_rate": 4.973334149947508e-06,
      "loss": 0.6684,
      "step": 508
    },
    {
      "epoch": 0.17219215155615697,
      "grad_norm": 0.23314347863197327,
      "learning_rate": 4.973203509107527e-06,
      "loss": 0.6417,
      "step": 509
    },
    {
      "epoch": 0.17253044654939106,
      "grad_norm": 0.24465447664260864,
      "learning_rate": 4.973072550756737e-06,
      "loss": 0.6643,
      "step": 510
    },
    {
      "epoch": 0.17286874154262516,
      "grad_norm": 0.2504658102989197,
      "learning_rate": 4.972941274911953e-06,
      "loss": 0.669,
      "step": 511
    },
    {
      "epoch": 0.17320703653585928,
      "grad_norm": 0.24236737191677094,
      "learning_rate": 4.972809681590027e-06,
      "loss": 0.6568,
      "step": 512
    },
    {
      "epoch": 0.17354533152909338,
      "grad_norm": 0.2436477541923523,
      "learning_rate": 4.972677770807853e-06,
      "loss": 0.6959,
      "step": 513
    },
    {
      "epoch": 0.17388362652232747,
      "grad_norm": 0.24689815938472748,
      "learning_rate": 4.972545542582366e-06,
      "loss": 0.6841,
      "step": 514
    },
    {
      "epoch": 0.17422192151556157,
      "grad_norm": 0.24580040574073792,
      "learning_rate": 4.972412996930541e-06,
      "loss": 0.6684,
      "step": 515
    },
    {
      "epoch": 0.17456021650879566,
      "grad_norm": 0.24564984440803528,
      "learning_rate": 4.972280133869396e-06,
      "loss": 0.6322,
      "step": 516
    },
    {
      "epoch": 0.17489851150202976,
      "grad_norm": 0.2577616572380066,
      "learning_rate": 4.972146953415986e-06,
      "loss": 0.6658,
      "step": 517
    },
    {
      "epoch": 0.17523680649526388,
      "grad_norm": 0.23606464266777039,
      "learning_rate": 4.972013455587409e-06,
      "loss": 0.6507,
      "step": 518
    },
    {
      "epoch": 0.17557510148849798,
      "grad_norm": 0.23771150410175323,
      "learning_rate": 4.9718796404008054e-06,
      "loss": 0.6829,
      "step": 519
    },
    {
      "epoch": 0.17591339648173207,
      "grad_norm": 0.25152748823165894,
      "learning_rate": 4.971745507873352e-06,
      "loss": 0.6766,
      "step": 520
    },
    {
      "epoch": 0.17625169147496617,
      "grad_norm": 0.2582070231437683,
      "learning_rate": 4.97161105802227e-06,
      "loss": 0.687,
      "step": 521
    },
    {
      "epoch": 0.17658998646820026,
      "grad_norm": 0.2388746291399002,
      "learning_rate": 4.9714762908648206e-06,
      "loss": 0.6462,
      "step": 522
    },
    {
      "epoch": 0.17692828146143438,
      "grad_norm": 0.2524266839027405,
      "learning_rate": 4.971341206418305e-06,
      "loss": 0.6546,
      "step": 523
    },
    {
      "epoch": 0.17726657645466848,
      "grad_norm": 0.2386075109243393,
      "learning_rate": 4.971205804700064e-06,
      "loss": 0.643,
      "step": 524
    },
    {
      "epoch": 0.17760487144790257,
      "grad_norm": 0.2496464103460312,
      "learning_rate": 4.9710700857274805e-06,
      "loss": 0.6488,
      "step": 525
    },
    {
      "epoch": 0.17794316644113667,
      "grad_norm": 0.25205540657043457,
      "learning_rate": 4.97093404951798e-06,
      "loss": 0.6863,
      "step": 526
    },
    {
      "epoch": 0.17828146143437076,
      "grad_norm": 0.25648388266563416,
      "learning_rate": 4.9707976960890266e-06,
      "loss": 0.6692,
      "step": 527
    },
    {
      "epoch": 0.17861975642760486,
      "grad_norm": 0.24036119878292084,
      "learning_rate": 4.9706610254581255e-06,
      "loss": 0.6605,
      "step": 528
    },
    {
      "epoch": 0.17895805142083898,
      "grad_norm": 0.2514593303203583,
      "learning_rate": 4.970524037642821e-06,
      "loss": 0.6625,
      "step": 529
    },
    {
      "epoch": 0.17929634641407308,
      "grad_norm": 0.2690679132938385,
      "learning_rate": 4.970386732660701e-06,
      "loss": 0.6736,
      "step": 530
    },
    {
      "epoch": 0.17963464140730717,
      "grad_norm": 0.25247398018836975,
      "learning_rate": 4.970249110529392e-06,
      "loss": 0.6712,
      "step": 531
    },
    {
      "epoch": 0.17997293640054127,
      "grad_norm": 0.24959304928779602,
      "learning_rate": 4.970111171266563e-06,
      "loss": 0.6608,
      "step": 532
    },
    {
      "epoch": 0.18031123139377536,
      "grad_norm": 0.2703661024570465,
      "learning_rate": 4.969972914889921e-06,
      "loss": 0.659,
      "step": 533
    },
    {
      "epoch": 0.18064952638700948,
      "grad_norm": 0.25666624307632446,
      "learning_rate": 4.9698343414172165e-06,
      "loss": 0.6777,
      "step": 534
    },
    {
      "epoch": 0.18098782138024358,
      "grad_norm": 0.23413406312465668,
      "learning_rate": 4.9696954508662405e-06,
      "loss": 0.6328,
      "step": 535
    },
    {
      "epoch": 0.18132611637347767,
      "grad_norm": 0.2529549300670624,
      "learning_rate": 4.969556243254822e-06,
      "loss": 0.6612,
      "step": 536
    },
    {
      "epoch": 0.18166441136671177,
      "grad_norm": 0.24946318566799164,
      "learning_rate": 4.969416718600834e-06,
      "loss": 0.6363,
      "step": 537
    },
    {
      "epoch": 0.18200270635994586,
      "grad_norm": 0.24231383204460144,
      "learning_rate": 4.969276876922188e-06,
      "loss": 0.6348,
      "step": 538
    },
    {
      "epoch": 0.18234100135317996,
      "grad_norm": 0.2565390169620514,
      "learning_rate": 4.969136718236836e-06,
      "loss": 0.6792,
      "step": 539
    },
    {
      "epoch": 0.18267929634641408,
      "grad_norm": 0.2580963969230652,
      "learning_rate": 4.968996242562774e-06,
      "loss": 0.6665,
      "step": 540
    },
    {
      "epoch": 0.18301759133964818,
      "grad_norm": 0.24817582964897156,
      "learning_rate": 4.968855449918034e-06,
      "loss": 0.6499,
      "step": 541
    },
    {
      "epoch": 0.18335588633288227,
      "grad_norm": 0.26380398869514465,
      "learning_rate": 4.968714340320692e-06,
      "loss": 0.6347,
      "step": 542
    },
    {
      "epoch": 0.18369418132611637,
      "grad_norm": 0.2705709934234619,
      "learning_rate": 4.968572913788864e-06,
      "loss": 0.6603,
      "step": 543
    },
    {
      "epoch": 0.18403247631935046,
      "grad_norm": 0.2509385943412781,
      "learning_rate": 4.968431170340706e-06,
      "loss": 0.6374,
      "step": 544
    },
    {
      "epoch": 0.18437077131258459,
      "grad_norm": 0.255299836397171,
      "learning_rate": 4.9682891099944155e-06,
      "loss": 0.6557,
      "step": 545
    },
    {
      "epoch": 0.18470906630581868,
      "grad_norm": 0.26094627380371094,
      "learning_rate": 4.9681467327682305e-06,
      "loss": 0.6637,
      "step": 546
    },
    {
      "epoch": 0.18504736129905278,
      "grad_norm": 0.24482354521751404,
      "learning_rate": 4.968004038680428e-06,
      "loss": 0.6459,
      "step": 547
    },
    {
      "epoch": 0.18538565629228687,
      "grad_norm": 0.258472204208374,
      "learning_rate": 4.967861027749327e-06,
      "loss": 0.6663,
      "step": 548
    },
    {
      "epoch": 0.18572395128552097,
      "grad_norm": 0.25342991948127747,
      "learning_rate": 4.967717699993291e-06,
      "loss": 0.6603,
      "step": 549
    },
    {
      "epoch": 0.18606224627875506,
      "grad_norm": 0.2580524981021881,
      "learning_rate": 4.9675740554307155e-06,
      "loss": 0.6748,
      "step": 550
    },
    {
      "epoch": 0.18640054127198918,
      "grad_norm": 0.24921679496765137,
      "learning_rate": 4.967430094080045e-06,
      "loss": 0.6815,
      "step": 551
    },
    {
      "epoch": 0.18673883626522328,
      "grad_norm": 0.2525157928466797,
      "learning_rate": 4.967285815959759e-06,
      "loss": 0.6687,
      "step": 552
    },
    {
      "epoch": 0.18707713125845737,
      "grad_norm": 0.2625119388103485,
      "learning_rate": 4.9671412210883826e-06,
      "loss": 0.6528,
      "step": 553
    },
    {
      "epoch": 0.18741542625169147,
      "grad_norm": 0.27415037155151367,
      "learning_rate": 4.966996309484477e-06,
      "loss": 0.6494,
      "step": 554
    },
    {
      "epoch": 0.18775372124492556,
      "grad_norm": 0.267037957906723,
      "learning_rate": 4.966851081166646e-06,
      "loss": 0.6888,
      "step": 555
    },
    {
      "epoch": 0.1880920162381597,
      "grad_norm": 0.2636261582374573,
      "learning_rate": 4.966705536153536e-06,
      "loss": 0.6489,
      "step": 556
    },
    {
      "epoch": 0.18843031123139378,
      "grad_norm": 0.30301186442375183,
      "learning_rate": 4.9665596744638285e-06,
      "loss": 0.6583,
      "step": 557
    },
    {
      "epoch": 0.18876860622462788,
      "grad_norm": 0.2526191771030426,
      "learning_rate": 4.966413496116254e-06,
      "loss": 0.6708,
      "step": 558
    },
    {
      "epoch": 0.18910690121786197,
      "grad_norm": 0.2610830068588257,
      "learning_rate": 4.966267001129575e-06,
      "loss": 0.6426,
      "step": 559
    },
    {
      "epoch": 0.18944519621109607,
      "grad_norm": 0.24611899256706238,
      "learning_rate": 4.9661201895226e-06,
      "loss": 0.6545,
      "step": 560
    },
    {
      "epoch": 0.18978349120433016,
      "grad_norm": 0.26428115367889404,
      "learning_rate": 4.965973061314178e-06,
      "loss": 0.7069,
      "step": 561
    },
    {
      "epoch": 0.19012178619756429,
      "grad_norm": 0.23961693048477173,
      "learning_rate": 4.965825616523194e-06,
      "loss": 0.6376,
      "step": 562
    },
    {
      "epoch": 0.19046008119079838,
      "grad_norm": 0.2400297075510025,
      "learning_rate": 4.965677855168581e-06,
      "loss": 0.6612,
      "step": 563
    },
    {
      "epoch": 0.19079837618403248,
      "grad_norm": 0.2490435689687729,
      "learning_rate": 4.9655297772693066e-06,
      "loss": 0.6539,
      "step": 564
    },
    {
      "epoch": 0.19113667117726657,
      "grad_norm": 0.2572137415409088,
      "learning_rate": 4.965381382844381e-06,
      "loss": 0.6447,
      "step": 565
    },
    {
      "epoch": 0.19147496617050067,
      "grad_norm": 0.2542514503002167,
      "learning_rate": 4.965232671912855e-06,
      "loss": 0.6692,
      "step": 566
    },
    {
      "epoch": 0.1918132611637348,
      "grad_norm": 0.2554891109466553,
      "learning_rate": 4.965083644493821e-06,
      "loss": 0.6909,
      "step": 567
    },
    {
      "epoch": 0.19215155615696888,
      "grad_norm": 0.2654957175254822,
      "learning_rate": 4.964934300606411e-06,
      "loss": 0.6793,
      "step": 568
    },
    {
      "epoch": 0.19248985115020298,
      "grad_norm": 0.25700655579566956,
      "learning_rate": 4.9647846402697974e-06,
      "loss": 0.6723,
      "step": 569
    },
    {
      "epoch": 0.19282814614343707,
      "grad_norm": 0.2551063299179077,
      "learning_rate": 4.964634663503194e-06,
      "loss": 0.6621,
      "step": 570
    },
    {
      "epoch": 0.19316644113667117,
      "grad_norm": 0.25916796922683716,
      "learning_rate": 4.964484370325855e-06,
      "loss": 0.6729,
      "step": 571
    },
    {
      "epoch": 0.19350473612990526,
      "grad_norm": 0.24621054530143738,
      "learning_rate": 4.964333760757074e-06,
      "loss": 0.6571,
      "step": 572
    },
    {
      "epoch": 0.1938430311231394,
      "grad_norm": 0.24834199249744415,
      "learning_rate": 4.9641828348161874e-06,
      "loss": 0.6636,
      "step": 573
    },
    {
      "epoch": 0.19418132611637348,
      "grad_norm": 0.25093671679496765,
      "learning_rate": 4.9640315925225715e-06,
      "loss": 0.6407,
      "step": 574
    },
    {
      "epoch": 0.19451962110960758,
      "grad_norm": 0.2574840486049652,
      "learning_rate": 4.9638800338956415e-06,
      "loss": 0.6799,
      "step": 575
    },
    {
      "epoch": 0.19485791610284167,
      "grad_norm": 0.2423202246427536,
      "learning_rate": 4.9637281589548555e-06,
      "loss": 0.654,
      "step": 576
    },
    {
      "epoch": 0.19519621109607577,
      "grad_norm": 0.2530333995819092,
      "learning_rate": 4.963575967719711e-06,
      "loss": 0.647,
      "step": 577
    },
    {
      "epoch": 0.1955345060893099,
      "grad_norm": 0.24557620286941528,
      "learning_rate": 4.963423460209747e-06,
      "loss": 0.6482,
      "step": 578
    },
    {
      "epoch": 0.19587280108254398,
      "grad_norm": 0.2524111866950989,
      "learning_rate": 4.963270636444542e-06,
      "loss": 0.6562,
      "step": 579
    },
    {
      "epoch": 0.19621109607577808,
      "grad_norm": 0.2587498426437378,
      "learning_rate": 4.963117496443715e-06,
      "loss": 0.6445,
      "step": 580
    },
    {
      "epoch": 0.19654939106901217,
      "grad_norm": 0.2540567219257355,
      "learning_rate": 4.962964040226927e-06,
      "loss": 0.6789,
      "step": 581
    },
    {
      "epoch": 0.19688768606224627,
      "grad_norm": 0.2543244957923889,
      "learning_rate": 4.962810267813878e-06,
      "loss": 0.6623,
      "step": 582
    },
    {
      "epoch": 0.1972259810554804,
      "grad_norm": 0.2576369643211365,
      "learning_rate": 4.96265617922431e-06,
      "loss": 0.6481,
      "step": 583
    },
    {
      "epoch": 0.1975642760487145,
      "grad_norm": 0.2497120052576065,
      "learning_rate": 4.962501774478005e-06,
      "loss": 0.6584,
      "step": 584
    },
    {
      "epoch": 0.19790257104194858,
      "grad_norm": 0.24584496021270752,
      "learning_rate": 4.962347053594784e-06,
      "loss": 0.6596,
      "step": 585
    },
    {
      "epoch": 0.19824086603518268,
      "grad_norm": 0.2562602758407593,
      "learning_rate": 4.9621920165945125e-06,
      "loss": 0.6449,
      "step": 586
    },
    {
      "epoch": 0.19857916102841677,
      "grad_norm": 0.2451668381690979,
      "learning_rate": 4.962036663497093e-06,
      "loss": 0.6362,
      "step": 587
    },
    {
      "epoch": 0.19891745602165087,
      "grad_norm": 0.24884480237960815,
      "learning_rate": 4.96188099432247e-06,
      "loss": 0.6191,
      "step": 588
    },
    {
      "epoch": 0.199255751014885,
      "grad_norm": 0.24974341690540314,
      "learning_rate": 4.961725009090627e-06,
      "loss": 0.6672,
      "step": 589
    },
    {
      "epoch": 0.19959404600811909,
      "grad_norm": 0.25356993079185486,
      "learning_rate": 4.9615687078215925e-06,
      "loss": 0.6598,
      "step": 590
    },
    {
      "epoch": 0.19993234100135318,
      "grad_norm": 0.2501576542854309,
      "learning_rate": 4.96141209053543e-06,
      "loss": 0.6595,
      "step": 591
    },
    {
      "epoch": 0.20027063599458728,
      "grad_norm": 0.2463545799255371,
      "learning_rate": 4.9612551572522465e-06,
      "loss": 0.6488,
      "step": 592
    },
    {
      "epoch": 0.20060893098782137,
      "grad_norm": 0.25693491101264954,
      "learning_rate": 4.96109790799219e-06,
      "loss": 0.6575,
      "step": 593
    },
    {
      "epoch": 0.2009472259810555,
      "grad_norm": 0.251827597618103,
      "learning_rate": 4.960940342775448e-06,
      "loss": 0.6565,
      "step": 594
    },
    {
      "epoch": 0.2012855209742896,
      "grad_norm": 0.2766624689102173,
      "learning_rate": 4.960782461622247e-06,
      "loss": 0.6515,
      "step": 595
    },
    {
      "epoch": 0.20162381596752368,
      "grad_norm": 0.24949146807193756,
      "learning_rate": 4.960624264552858e-06,
      "loss": 0.6789,
      "step": 596
    },
    {
      "epoch": 0.20196211096075778,
      "grad_norm": 0.2712768316268921,
      "learning_rate": 4.9604657515875894e-06,
      "loss": 0.6403,
      "step": 597
    },
    {
      "epoch": 0.20230040595399187,
      "grad_norm": 0.24750161170959473,
      "learning_rate": 4.960306922746793e-06,
      "loss": 0.6671,
      "step": 598
    },
    {
      "epoch": 0.20263870094722597,
      "grad_norm": 0.2518949806690216,
      "learning_rate": 4.960147778050856e-06,
      "loss": 0.6559,
      "step": 599
    },
    {
      "epoch": 0.2029769959404601,
      "grad_norm": 0.26023411750793457,
      "learning_rate": 4.959988317520213e-06,
      "loss": 0.6844,
      "step": 600
    },
    {
      "epoch": 0.2033152909336942,
      "grad_norm": 0.26342353224754333,
      "learning_rate": 4.959828541175332e-06,
      "loss": 0.6627,
      "step": 601
    },
    {
      "epoch": 0.20365358592692828,
      "grad_norm": 0.26747390627861023,
      "learning_rate": 4.959668449036728e-06,
      "loss": 0.6379,
      "step": 602
    },
    {
      "epoch": 0.20399188092016238,
      "grad_norm": 0.262066513299942,
      "learning_rate": 4.959508041124952e-06,
      "loss": 0.6667,
      "step": 603
    },
    {
      "epoch": 0.20433017591339647,
      "grad_norm": 0.26433098316192627,
      "learning_rate": 4.9593473174605976e-06,
      "loss": 0.6713,
      "step": 604
    },
    {
      "epoch": 0.2046684709066306,
      "grad_norm": 0.27428731322288513,
      "learning_rate": 4.959186278064299e-06,
      "loss": 0.627,
      "step": 605
    },
    {
      "epoch": 0.2050067658998647,
      "grad_norm": 0.25257450342178345,
      "learning_rate": 4.959024922956732e-06,
      "loss": 0.6573,
      "step": 606
    },
    {
      "epoch": 0.20534506089309879,
      "grad_norm": 0.25500935316085815,
      "learning_rate": 4.958863252158608e-06,
      "loss": 0.6821,
      "step": 607
    },
    {
      "epoch": 0.20568335588633288,
      "grad_norm": 0.2703934907913208,
      "learning_rate": 4.958701265690685e-06,
      "loss": 0.6471,
      "step": 608
    },
    {
      "epoch": 0.20602165087956698,
      "grad_norm": 0.26352187991142273,
      "learning_rate": 4.958538963573758e-06,
      "loss": 0.6738,
      "step": 609
    },
    {
      "epoch": 0.20635994587280107,
      "grad_norm": 0.26434412598609924,
      "learning_rate": 4.958376345828662e-06,
      "loss": 0.666,
      "step": 610
    },
    {
      "epoch": 0.2066982408660352,
      "grad_norm": 0.26020878553390503,
      "learning_rate": 4.9582134124762775e-06,
      "loss": 0.6866,
      "step": 611
    },
    {
      "epoch": 0.2070365358592693,
      "grad_norm": 0.2619779109954834,
      "learning_rate": 4.958050163537519e-06,
      "loss": 0.6566,
      "step": 612
    },
    {
      "epoch": 0.20737483085250338,
      "grad_norm": 0.2512053847312927,
      "learning_rate": 4.957886599033345e-06,
      "loss": 0.6648,
      "step": 613
    },
    {
      "epoch": 0.20771312584573748,
      "grad_norm": 0.27655842900276184,
      "learning_rate": 4.957722718984754e-06,
      "loss": 0.6549,
      "step": 614
    },
    {
      "epoch": 0.20805142083897157,
      "grad_norm": 0.2558527886867523,
      "learning_rate": 4.957558523412786e-06,
      "loss": 0.6734,
      "step": 615
    },
    {
      "epoch": 0.2083897158322057,
      "grad_norm": 0.26270580291748047,
      "learning_rate": 4.957394012338519e-06,
      "loss": 0.6546,
      "step": 616
    },
    {
      "epoch": 0.2087280108254398,
      "grad_norm": 0.2507633864879608,
      "learning_rate": 4.957229185783074e-06,
      "loss": 0.6549,
      "step": 617
    },
    {
      "epoch": 0.2090663058186739,
      "grad_norm": 0.2596471309661865,
      "learning_rate": 4.957064043767611e-06,
      "loss": 0.6698,
      "step": 618
    },
    {
      "epoch": 0.20940460081190798,
      "grad_norm": 0.24986913800239563,
      "learning_rate": 4.956898586313332e-06,
      "loss": 0.639,
      "step": 619
    },
    {
      "epoch": 0.20974289580514208,
      "grad_norm": 0.24691560864448547,
      "learning_rate": 4.956732813441477e-06,
      "loss": 0.6757,
      "step": 620
    },
    {
      "epoch": 0.21008119079837617,
      "grad_norm": 0.2614850699901581,
      "learning_rate": 4.956566725173329e-06,
      "loss": 0.6553,
      "step": 621
    },
    {
      "epoch": 0.2104194857916103,
      "grad_norm": 0.26026490330696106,
      "learning_rate": 4.956400321530211e-06,
      "loss": 0.6743,
      "step": 622
    },
    {
      "epoch": 0.2107577807848444,
      "grad_norm": 0.25970712304115295,
      "learning_rate": 4.956233602533483e-06,
      "loss": 0.6715,
      "step": 623
    },
    {
      "epoch": 0.21109607577807848,
      "grad_norm": 0.2530291974544525,
      "learning_rate": 4.956066568204551e-06,
      "loss": 0.639,
      "step": 624
    },
    {
      "epoch": 0.21143437077131258,
      "grad_norm": 0.2687148153781891,
      "learning_rate": 4.95589921856486e-06,
      "loss": 0.6565,
      "step": 625
    },
    {
      "epoch": 0.21177266576454667,
      "grad_norm": 0.2534351944923401,
      "learning_rate": 4.955731553635892e-06,
      "loss": 0.64,
      "step": 626
    },
    {
      "epoch": 0.2121109607577808,
      "grad_norm": 0.25353437662124634,
      "learning_rate": 4.9555635734391725e-06,
      "loss": 0.684,
      "step": 627
    },
    {
      "epoch": 0.2124492557510149,
      "grad_norm": 0.26010623574256897,
      "learning_rate": 4.955395277996268e-06,
      "loss": 0.6516,
      "step": 628
    },
    {
      "epoch": 0.212787550744249,
      "grad_norm": 0.2603456974029541,
      "learning_rate": 4.955226667328783e-06,
      "loss": 0.6619,
      "step": 629
    },
    {
      "epoch": 0.21312584573748308,
      "grad_norm": 0.25239628553390503,
      "learning_rate": 4.9550577414583635e-06,
      "loss": 0.673,
      "step": 630
    },
    {
      "epoch": 0.21346414073071718,
      "grad_norm": 0.25560659170150757,
      "learning_rate": 4.954888500406699e-06,
      "loss": 0.6576,
      "step": 631
    },
    {
      "epoch": 0.21380243572395127,
      "grad_norm": 0.26723402738571167,
      "learning_rate": 4.9547189441955124e-06,
      "loss": 0.6821,
      "step": 632
    },
    {
      "epoch": 0.2141407307171854,
      "grad_norm": 0.24569633603096008,
      "learning_rate": 4.9545490728465754e-06,
      "loss": 0.6669,
      "step": 633
    },
    {
      "epoch": 0.2144790257104195,
      "grad_norm": 0.26862242817878723,
      "learning_rate": 4.9543788863816944e-06,
      "loss": 0.6476,
      "step": 634
    },
    {
      "epoch": 0.21481732070365359,
      "grad_norm": 0.24638478457927704,
      "learning_rate": 4.954208384822717e-06,
      "loss": 0.6506,
      "step": 635
    },
    {
      "epoch": 0.21515561569688768,
      "grad_norm": 0.24835918843746185,
      "learning_rate": 4.954037568191534e-06,
      "loss": 0.6584,
      "step": 636
    },
    {
      "epoch": 0.21549391069012178,
      "grad_norm": 0.24750922620296478,
      "learning_rate": 4.9538664365100755e-06,
      "loss": 0.6486,
      "step": 637
    },
    {
      "epoch": 0.2158322056833559,
      "grad_norm": 0.27238205075263977,
      "learning_rate": 4.953694989800309e-06,
      "loss": 0.6549,
      "step": 638
    },
    {
      "epoch": 0.21617050067659,
      "grad_norm": 0.2620726227760315,
      "learning_rate": 4.953523228084246e-06,
      "loss": 0.6545,
      "step": 639
    },
    {
      "epoch": 0.2165087956698241,
      "grad_norm": 0.2717283070087433,
      "learning_rate": 4.953351151383938e-06,
      "loss": 0.6326,
      "step": 640
    },
    {
      "epoch": 0.21684709066305818,
      "grad_norm": 0.28660696744918823,
      "learning_rate": 4.9531787597214756e-06,
      "loss": 0.662,
      "step": 641
    },
    {
      "epoch": 0.21718538565629228,
      "grad_norm": 0.26013630628585815,
      "learning_rate": 4.953006053118991e-06,
      "loss": 0.6644,
      "step": 642
    },
    {
      "epoch": 0.21752368064952637,
      "grad_norm": 0.2518508732318878,
      "learning_rate": 4.952833031598655e-06,
      "loss": 0.6412,
      "step": 643
    },
    {
      "epoch": 0.2178619756427605,
      "grad_norm": 0.2623193860054016,
      "learning_rate": 4.952659695182682e-06,
      "loss": 0.6527,
      "step": 644
    },
    {
      "epoch": 0.2182002706359946,
      "grad_norm": 0.2599288821220398,
      "learning_rate": 4.952486043893324e-06,
      "loss": 0.6508,
      "step": 645
    },
    {
      "epoch": 0.2185385656292287,
      "grad_norm": 0.25234729051589966,
      "learning_rate": 4.952312077752875e-06,
      "loss": 0.6416,
      "step": 646
    },
    {
      "epoch": 0.21887686062246278,
      "grad_norm": 0.2652762234210968,
      "learning_rate": 4.952137796783667e-06,
      "loss": 0.6537,
      "step": 647
    },
    {
      "epoch": 0.21921515561569688,
      "grad_norm": 0.26549839973449707,
      "learning_rate": 4.9519632010080765e-06,
      "loss": 0.6722,
      "step": 648
    },
    {
      "epoch": 0.219553450608931,
      "grad_norm": 0.2607754170894623,
      "learning_rate": 4.951788290448517e-06,
      "loss": 0.6602,
      "step": 649
    },
    {
      "epoch": 0.2198917456021651,
      "grad_norm": 0.26408514380455017,
      "learning_rate": 4.951613065127445e-06,
      "loss": 0.6671,
      "step": 650
    },
    {
      "epoch": 0.2202300405953992,
      "grad_norm": 0.2623929977416992,
      "learning_rate": 4.951437525067354e-06,
      "loss": 0.6523,
      "step": 651
    },
    {
      "epoch": 0.22056833558863329,
      "grad_norm": 0.25325843691825867,
      "learning_rate": 4.951261670290781e-06,
      "loss": 0.6434,
      "step": 652
    },
    {
      "epoch": 0.22090663058186738,
      "grad_norm": 0.25801414251327515,
      "learning_rate": 4.951085500820302e-06,
      "loss": 0.665,
      "step": 653
    },
    {
      "epoch": 0.22124492557510148,
      "grad_norm": 0.25949883460998535,
      "learning_rate": 4.950909016678534e-06,
      "loss": 0.6339,
      "step": 654
    },
    {
      "epoch": 0.2215832205683356,
      "grad_norm": 0.26575353741645813,
      "learning_rate": 4.950732217888134e-06,
      "loss": 0.6508,
      "step": 655
    },
    {
      "epoch": 0.2219215155615697,
      "grad_norm": 0.2612333297729492,
      "learning_rate": 4.950555104471799e-06,
      "loss": 0.6499,
      "step": 656
    },
    {
      "epoch": 0.2222598105548038,
      "grad_norm": 0.26511678099632263,
      "learning_rate": 4.950377676452267e-06,
      "loss": 0.6475,
      "step": 657
    },
    {
      "epoch": 0.22259810554803788,
      "grad_norm": 0.264708936214447,
      "learning_rate": 4.950199933852318e-06,
      "loss": 0.632,
      "step": 658
    },
    {
      "epoch": 0.22293640054127198,
      "grad_norm": 0.2613477408885956,
      "learning_rate": 4.950021876694768e-06,
      "loss": 0.6399,
      "step": 659
    },
    {
      "epoch": 0.2232746955345061,
      "grad_norm": 0.2623678743839264,
      "learning_rate": 4.949843505002478e-06,
      "loss": 0.657,
      "step": 660
    },
    {
      "epoch": 0.2236129905277402,
      "grad_norm": 0.2551998198032379,
      "learning_rate": 4.949664818798347e-06,
      "loss": 0.6564,
      "step": 661
    },
    {
      "epoch": 0.2239512855209743,
      "grad_norm": 0.27580368518829346,
      "learning_rate": 4.949485818105313e-06,
      "loss": 0.68,
      "step": 662
    },
    {
      "epoch": 0.2242895805142084,
      "grad_norm": 0.2604037821292877,
      "learning_rate": 4.949306502946358e-06,
      "loss": 0.6649,
      "step": 663
    },
    {
      "epoch": 0.22462787550744248,
      "grad_norm": 0.27373746037483215,
      "learning_rate": 4.949126873344504e-06,
      "loss": 0.6637,
      "step": 664
    },
    {
      "epoch": 0.22496617050067658,
      "grad_norm": 0.256798654794693,
      "learning_rate": 4.948946929322809e-06,
      "loss": 0.6679,
      "step": 665
    },
    {
      "epoch": 0.2253044654939107,
      "grad_norm": 0.26641178131103516,
      "learning_rate": 4.948766670904375e-06,
      "loss": 0.6541,
      "step": 666
    },
    {
      "epoch": 0.2256427604871448,
      "grad_norm": 0.2707980275154114,
      "learning_rate": 4.948586098112344e-06,
      "loss": 0.6644,
      "step": 667
    },
    {
      "epoch": 0.2259810554803789,
      "grad_norm": 0.26444727182388306,
      "learning_rate": 4.9484052109698984e-06,
      "loss": 0.6547,
      "step": 668
    },
    {
      "epoch": 0.22631935047361298,
      "grad_norm": 0.26225337386131287,
      "learning_rate": 4.948224009500261e-06,
      "loss": 0.6686,
      "step": 669
    },
    {
      "epoch": 0.22665764546684708,
      "grad_norm": 0.26892396807670593,
      "learning_rate": 4.948042493726694e-06,
      "loss": 0.6394,
      "step": 670
    },
    {
      "epoch": 0.2269959404600812,
      "grad_norm": 0.25884899497032166,
      "learning_rate": 4.947860663672499e-06,
      "loss": 0.6215,
      "step": 671
    },
    {
      "epoch": 0.2273342354533153,
      "grad_norm": 0.26995548605918884,
      "learning_rate": 4.947678519361021e-06,
      "loss": 0.6476,
      "step": 672
    },
    {
      "epoch": 0.2276725304465494,
      "grad_norm": 0.26591798663139343,
      "learning_rate": 4.947496060815643e-06,
      "loss": 0.6653,
      "step": 673
    },
    {
      "epoch": 0.2280108254397835,
      "grad_norm": 0.2683780789375305,
      "learning_rate": 4.94731328805979e-06,
      "loss": 0.6593,
      "step": 674
    },
    {
      "epoch": 0.22834912043301758,
      "grad_norm": 0.2754748463630676,
      "learning_rate": 4.947130201116926e-06,
      "loss": 0.6589,
      "step": 675
    },
    {
      "epoch": 0.22868741542625168,
      "grad_norm": 0.25586527585983276,
      "learning_rate": 4.946946800010556e-06,
      "loss": 0.6178,
      "step": 676
    },
    {
      "epoch": 0.2290257104194858,
      "grad_norm": 0.252361536026001,
      "learning_rate": 4.946763084764225e-06,
      "loss": 0.6348,
      "step": 677
    },
    {
      "epoch": 0.2293640054127199,
      "grad_norm": 0.2510469853878021,
      "learning_rate": 4.946579055401517e-06,
      "loss": 0.6216,
      "step": 678
    },
    {
      "epoch": 0.229702300405954,
      "grad_norm": 0.2685176432132721,
      "learning_rate": 4.94639471194606e-06,
      "loss": 0.6469,
      "step": 679
    },
    {
      "epoch": 0.23004059539918809,
      "grad_norm": 0.2603522539138794,
      "learning_rate": 4.946210054421519e-06,
      "loss": 0.6434,
      "step": 680
    },
    {
      "epoch": 0.23037889039242218,
      "grad_norm": 0.2632562220096588,
      "learning_rate": 4.9460250828516e-06,
      "loss": 0.6483,
      "step": 681
    },
    {
      "epoch": 0.2307171853856563,
      "grad_norm": 0.2648269236087799,
      "learning_rate": 4.9458397972600505e-06,
      "loss": 0.645,
      "step": 682
    },
    {
      "epoch": 0.2310554803788904,
      "grad_norm": 0.254108190536499,
      "learning_rate": 4.945654197670657e-06,
      "loss": 0.6604,
      "step": 683
    },
    {
      "epoch": 0.2313937753721245,
      "grad_norm": 0.269853413105011,
      "learning_rate": 4.945468284107246e-06,
      "loss": 0.6617,
      "step": 684
    },
    {
      "epoch": 0.2317320703653586,
      "grad_norm": 0.2735177278518677,
      "learning_rate": 4.945282056593689e-06,
      "loss": 0.6613,
      "step": 685
    },
    {
      "epoch": 0.23207036535859268,
      "grad_norm": 0.2711608409881592,
      "learning_rate": 4.945095515153889e-06,
      "loss": 0.647,
      "step": 686
    },
    {
      "epoch": 0.2324086603518268,
      "grad_norm": 0.2601723074913025,
      "learning_rate": 4.944908659811796e-06,
      "loss": 0.6584,
      "step": 687
    },
    {
      "epoch": 0.2327469553450609,
      "grad_norm": 0.2672250270843506,
      "learning_rate": 4.944721490591401e-06,
      "loss": 0.6529,
      "step": 688
    },
    {
      "epoch": 0.233085250338295,
      "grad_norm": 0.2668798863887787,
      "learning_rate": 4.94453400751673e-06,
      "loss": 0.6695,
      "step": 689
    },
    {
      "epoch": 0.2334235453315291,
      "grad_norm": 0.26739099621772766,
      "learning_rate": 4.944346210611853e-06,
      "loss": 0.6493,
      "step": 690
    },
    {
      "epoch": 0.2337618403247632,
      "grad_norm": 0.2629740834236145,
      "learning_rate": 4.9441580999008794e-06,
      "loss": 0.6472,
      "step": 691
    },
    {
      "epoch": 0.23410013531799728,
      "grad_norm": 0.2657390534877777,
      "learning_rate": 4.94396967540796e-06,
      "loss": 0.6516,
      "step": 692
    },
    {
      "epoch": 0.2344384303112314,
      "grad_norm": 0.26596182584762573,
      "learning_rate": 4.943780937157283e-06,
      "loss": 0.6571,
      "step": 693
    },
    {
      "epoch": 0.2347767253044655,
      "grad_norm": 0.26130345463752747,
      "learning_rate": 4.943591885173079e-06,
      "loss": 0.6407,
      "step": 694
    },
    {
      "epoch": 0.2351150202976996,
      "grad_norm": 0.24915356934070587,
      "learning_rate": 4.943402519479619e-06,
      "loss": 0.6739,
      "step": 695
    },
    {
      "epoch": 0.2354533152909337,
      "grad_norm": 0.27618497610092163,
      "learning_rate": 4.943212840101215e-06,
      "loss": 0.658,
      "step": 696
    },
    {
      "epoch": 0.23579161028416779,
      "grad_norm": 0.2530587613582611,
      "learning_rate": 4.943022847062216e-06,
      "loss": 0.6126,
      "step": 697
    },
    {
      "epoch": 0.2361299052774019,
      "grad_norm": 0.2686367928981781,
      "learning_rate": 4.942832540387015e-06,
      "loss": 0.6383,
      "step": 698
    },
    {
      "epoch": 0.236468200270636,
      "grad_norm": 0.27149710059165955,
      "learning_rate": 4.942641920100042e-06,
      "loss": 0.6494,
      "step": 699
    },
    {
      "epoch": 0.2368064952638701,
      "grad_norm": 0.2659294903278351,
      "learning_rate": 4.942450986225771e-06,
      "loss": 0.6707,
      "step": 700
    },
    {
      "epoch": 0.2371447902571042,
      "grad_norm": 0.2757556438446045,
      "learning_rate": 4.9422597387887126e-06,
      "loss": 0.6336,
      "step": 701
    },
    {
      "epoch": 0.2374830852503383,
      "grad_norm": 0.2766989767551422,
      "learning_rate": 4.942068177813419e-06,
      "loss": 0.6496,
      "step": 702
    },
    {
      "epoch": 0.23782138024357238,
      "grad_norm": 0.2604788541793823,
      "learning_rate": 4.941876303324486e-06,
      "loss": 0.6459,
      "step": 703
    },
    {
      "epoch": 0.2381596752368065,
      "grad_norm": 0.2848023474216461,
      "learning_rate": 4.941684115346541e-06,
      "loss": 0.6673,
      "step": 704
    },
    {
      "epoch": 0.2384979702300406,
      "grad_norm": 0.27619898319244385,
      "learning_rate": 4.941491613904263e-06,
      "loss": 0.6474,
      "step": 705
    },
    {
      "epoch": 0.2388362652232747,
      "grad_norm": 0.25733035802841187,
      "learning_rate": 4.941298799022361e-06,
      "loss": 0.6355,
      "step": 706
    },
    {
      "epoch": 0.2391745602165088,
      "grad_norm": 0.25964587926864624,
      "learning_rate": 4.94110567072559e-06,
      "loss": 0.6675,
      "step": 707
    },
    {
      "epoch": 0.2395128552097429,
      "grad_norm": 0.2735098898410797,
      "learning_rate": 4.940912229038746e-06,
      "loss": 0.665,
      "step": 708
    },
    {
      "epoch": 0.239851150202977,
      "grad_norm": 0.28130871057510376,
      "learning_rate": 4.94071847398666e-06,
      "loss": 0.6356,
      "step": 709
    },
    {
      "epoch": 0.2401894451962111,
      "grad_norm": 0.2739136219024658,
      "learning_rate": 4.940524405594208e-06,
      "loss": 0.6485,
      "step": 710
    },
    {
      "epoch": 0.2405277401894452,
      "grad_norm": 0.27404212951660156,
      "learning_rate": 4.940330023886304e-06,
      "loss": 0.6505,
      "step": 711
    },
    {
      "epoch": 0.2408660351826793,
      "grad_norm": 0.2825517952442169,
      "learning_rate": 4.940135328887903e-06,
      "loss": 0.6816,
      "step": 712
    },
    {
      "epoch": 0.2412043301759134,
      "grad_norm": 0.26147276163101196,
      "learning_rate": 4.939940320624e-06,
      "loss": 0.6614,
      "step": 713
    },
    {
      "epoch": 0.24154262516914748,
      "grad_norm": 0.26788076758384705,
      "learning_rate": 4.9397449991196305e-06,
      "loss": 0.6508,
      "step": 714
    },
    {
      "epoch": 0.2418809201623816,
      "grad_norm": 0.26942339539527893,
      "learning_rate": 4.93954936439987e-06,
      "loss": 0.6447,
      "step": 715
    },
    {
      "epoch": 0.2422192151556157,
      "grad_norm": 0.26093873381614685,
      "learning_rate": 4.939353416489834e-06,
      "loss": 0.6644,
      "step": 716
    },
    {
      "epoch": 0.2425575101488498,
      "grad_norm": 0.2670133411884308,
      "learning_rate": 4.9391571554146775e-06,
      "loss": 0.6457,
      "step": 717
    },
    {
      "epoch": 0.2428958051420839,
      "grad_norm": 0.2692917287349701,
      "learning_rate": 4.938960581199598e-06,
      "loss": 0.6626,
      "step": 718
    },
    {
      "epoch": 0.243234100135318,
      "grad_norm": 0.2619960606098175,
      "learning_rate": 4.93876369386983e-06,
      "loss": 0.6708,
      "step": 719
    },
    {
      "epoch": 0.2435723951285521,
      "grad_norm": 0.25591549277305603,
      "learning_rate": 4.938566493450653e-06,
      "loss": 0.6589,
      "step": 720
    },
    {
      "epoch": 0.2439106901217862,
      "grad_norm": 0.2719472348690033,
      "learning_rate": 4.9383689799673805e-06,
      "loss": 0.6549,
      "step": 721
    },
    {
      "epoch": 0.2442489851150203,
      "grad_norm": 0.2731001675128937,
      "learning_rate": 4.938171153445371e-06,
      "loss": 0.669,
      "step": 722
    },
    {
      "epoch": 0.2445872801082544,
      "grad_norm": 0.2630438506603241,
      "learning_rate": 4.937973013910021e-06,
      "loss": 0.6771,
      "step": 723
    },
    {
      "epoch": 0.2449255751014885,
      "grad_norm": 0.2734837234020233,
      "learning_rate": 4.937774561386768e-06,
      "loss": 0.6661,
      "step": 724
    },
    {
      "epoch": 0.24526387009472259,
      "grad_norm": 0.2699190378189087,
      "learning_rate": 4.937575795901089e-06,
      "loss": 0.6566,
      "step": 725
    },
    {
      "epoch": 0.2456021650879567,
      "grad_norm": 0.27050119638442993,
      "learning_rate": 4.937376717478502e-06,
      "loss": 0.6205,
      "step": 726
    },
    {
      "epoch": 0.2459404600811908,
      "grad_norm": 0.26535895466804504,
      "learning_rate": 4.937177326144564e-06,
      "loss": 0.6516,
      "step": 727
    },
    {
      "epoch": 0.2462787550744249,
      "grad_norm": 0.2628813683986664,
      "learning_rate": 4.936977621924875e-06,
      "loss": 0.6572,
      "step": 728
    },
    {
      "epoch": 0.246617050067659,
      "grad_norm": 0.27562880516052246,
      "learning_rate": 4.936777604845071e-06,
      "loss": 0.6686,
      "step": 729
    },
    {
      "epoch": 0.2469553450608931,
      "grad_norm": 0.26140421628952026,
      "learning_rate": 4.936577274930831e-06,
      "loss": 0.6538,
      "step": 730
    },
    {
      "epoch": 0.2472936400541272,
      "grad_norm": 0.26812177896499634,
      "learning_rate": 4.9363766322078735e-06,
      "loss": 0.6526,
      "step": 731
    },
    {
      "epoch": 0.2476319350473613,
      "grad_norm": 0.2756509780883789,
      "learning_rate": 4.936175676701956e-06,
      "loss": 0.648,
      "step": 732
    },
    {
      "epoch": 0.2479702300405954,
      "grad_norm": 0.27314281463623047,
      "learning_rate": 4.935974408438879e-06,
      "loss": 0.7115,
      "step": 733
    },
    {
      "epoch": 0.2483085250338295,
      "grad_norm": 0.2648617625236511,
      "learning_rate": 4.9357728274444806e-06,
      "loss": 0.6563,
      "step": 734
    },
    {
      "epoch": 0.2486468200270636,
      "grad_norm": 0.27088579535484314,
      "learning_rate": 4.93557093374464e-06,
      "loss": 0.6569,
      "step": 735
    },
    {
      "epoch": 0.2489851150202977,
      "grad_norm": 0.2711031436920166,
      "learning_rate": 4.935368727365276e-06,
      "loss": 0.6514,
      "step": 736
    },
    {
      "epoch": 0.2493234100135318,
      "grad_norm": 0.2798830270767212,
      "learning_rate": 4.935166208332348e-06,
      "loss": 0.6791,
      "step": 737
    },
    {
      "epoch": 0.2496617050067659,
      "grad_norm": 0.26322317123413086,
      "learning_rate": 4.934963376671857e-06,
      "loss": 0.6176,
      "step": 738
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2789643108844757,
      "learning_rate": 4.934760232409839e-06,
      "loss": 0.6578,
      "step": 739
    },
    {
      "epoch": 0.2503382949932341,
      "grad_norm": 0.26852181553840637,
      "learning_rate": 4.934556775572377e-06,
      "loss": 0.6377,
      "step": 740
    },
    {
      "epoch": 0.2506765899864682,
      "grad_norm": 0.2695555090904236,
      "learning_rate": 4.9343530061855906e-06,
      "loss": 0.6618,
      "step": 741
    },
    {
      "epoch": 0.2510148849797023,
      "grad_norm": 0.272977739572525,
      "learning_rate": 4.934148924275638e-06,
      "loss": 0.6486,
      "step": 742
    },
    {
      "epoch": 0.2513531799729364,
      "grad_norm": 0.2690495550632477,
      "learning_rate": 4.933944529868721e-06,
      "loss": 0.6457,
      "step": 743
    },
    {
      "epoch": 0.2516914749661705,
      "grad_norm": 0.281653493642807,
      "learning_rate": 4.933739822991079e-06,
      "loss": 0.6658,
      "step": 744
    },
    {
      "epoch": 0.2520297699594046,
      "grad_norm": 0.27239227294921875,
      "learning_rate": 4.933534803668992e-06,
      "loss": 0.64,
      "step": 745
    },
    {
      "epoch": 0.2523680649526387,
      "grad_norm": 0.2809191346168518,
      "learning_rate": 4.9333294719287815e-06,
      "loss": 0.6798,
      "step": 746
    },
    {
      "epoch": 0.2527063599458728,
      "grad_norm": 0.2619982659816742,
      "learning_rate": 4.933123827796808e-06,
      "loss": 0.6456,
      "step": 747
    },
    {
      "epoch": 0.2530446549391069,
      "grad_norm": 0.26253849267959595,
      "learning_rate": 4.932917871299471e-06,
      "loss": 0.6387,
      "step": 748
    },
    {
      "epoch": 0.253382949932341,
      "grad_norm": 0.27713361382484436,
      "learning_rate": 4.932711602463212e-06,
      "loss": 0.6593,
      "step": 749
    },
    {
      "epoch": 0.25372124492557513,
      "grad_norm": 0.285663366317749,
      "learning_rate": 4.932505021314513e-06,
      "loss": 0.6451,
      "step": 750
    },
    {
      "epoch": 0.2540595399188092,
      "grad_norm": 0.27399569749832153,
      "learning_rate": 4.932298127879893e-06,
      "loss": 0.6649,
      "step": 751
    },
    {
      "epoch": 0.2543978349120433,
      "grad_norm": 0.2679443955421448,
      "learning_rate": 4.932090922185914e-06,
      "loss": 0.6687,
      "step": 752
    },
    {
      "epoch": 0.2547361299052774,
      "grad_norm": 0.26548048853874207,
      "learning_rate": 4.9318834042591765e-06,
      "loss": 0.6627,
      "step": 753
    },
    {
      "epoch": 0.2550744248985115,
      "grad_norm": 0.274859219789505,
      "learning_rate": 4.9316755741263225e-06,
      "loss": 0.6615,
      "step": 754
    },
    {
      "epoch": 0.2554127198917456,
      "grad_norm": 0.2755845785140991,
      "learning_rate": 4.931467431814034e-06,
      "loss": 0.6382,
      "step": 755
    },
    {
      "epoch": 0.2557510148849797,
      "grad_norm": 0.2709038555622101,
      "learning_rate": 4.93125897734903e-06,
      "loss": 0.6393,
      "step": 756
    },
    {
      "epoch": 0.2560893098782138,
      "grad_norm": 0.26591774821281433,
      "learning_rate": 4.931050210758075e-06,
      "loss": 0.6749,
      "step": 757
    },
    {
      "epoch": 0.2564276048714479,
      "grad_norm": 0.2669565677642822,
      "learning_rate": 4.9308411320679685e-06,
      "loss": 0.6301,
      "step": 758
    },
    {
      "epoch": 0.256765899864682,
      "grad_norm": 0.26058095693588257,
      "learning_rate": 4.930631741305552e-06,
      "loss": 0.661,
      "step": 759
    },
    {
      "epoch": 0.2571041948579161,
      "grad_norm": 0.27673038840293884,
      "learning_rate": 4.930422038497708e-06,
      "loss": 0.6619,
      "step": 760
    },
    {
      "epoch": 0.2574424898511502,
      "grad_norm": 0.27486756443977356,
      "learning_rate": 4.930212023671358e-06,
      "loss": 0.652,
      "step": 761
    },
    {
      "epoch": 0.2577807848443843,
      "grad_norm": 0.27896642684936523,
      "learning_rate": 4.930001696853463e-06,
      "loss": 0.6478,
      "step": 762
    },
    {
      "epoch": 0.2581190798376184,
      "grad_norm": 0.26254603266716003,
      "learning_rate": 4.929791058071027e-06,
      "loss": 0.663,
      "step": 763
    },
    {
      "epoch": 0.2584573748308525,
      "grad_norm": 0.2756212055683136,
      "learning_rate": 4.92958010735109e-06,
      "loss": 0.6576,
      "step": 764
    },
    {
      "epoch": 0.2587956698240866,
      "grad_norm": 0.2646501362323761,
      "learning_rate": 4.929368844720734e-06,
      "loss": 0.677,
      "step": 765
    },
    {
      "epoch": 0.2591339648173207,
      "grad_norm": 0.2635130286216736,
      "learning_rate": 4.929157270207082e-06,
      "loss": 0.6584,
      "step": 766
    },
    {
      "epoch": 0.25947225981055483,
      "grad_norm": 0.26653656363487244,
      "learning_rate": 4.928945383837295e-06,
      "loss": 0.6664,
      "step": 767
    },
    {
      "epoch": 0.2598105548037889,
      "grad_norm": 0.2647206485271454,
      "learning_rate": 4.928733185638576e-06,
      "loss": 0.6653,
      "step": 768
    },
    {
      "epoch": 0.260148849797023,
      "grad_norm": 0.26432615518569946,
      "learning_rate": 4.928520675638166e-06,
      "loss": 0.6444,
      "step": 769
    },
    {
      "epoch": 0.2604871447902571,
      "grad_norm": 0.26040834188461304,
      "learning_rate": 4.928307853863348e-06,
      "loss": 0.6244,
      "step": 770
    },
    {
      "epoch": 0.2608254397834912,
      "grad_norm": 0.2694340944290161,
      "learning_rate": 4.928094720341443e-06,
      "loss": 0.6523,
      "step": 771
    },
    {
      "epoch": 0.26116373477672533,
      "grad_norm": 0.27743464708328247,
      "learning_rate": 4.927881275099815e-06,
      "loss": 0.6625,
      "step": 772
    },
    {
      "epoch": 0.2615020297699594,
      "grad_norm": 0.2654753625392914,
      "learning_rate": 4.927667518165865e-06,
      "loss": 0.6122,
      "step": 773
    },
    {
      "epoch": 0.2618403247631935,
      "grad_norm": 0.25984954833984375,
      "learning_rate": 4.927453449567034e-06,
      "loss": 0.659,
      "step": 774
    },
    {
      "epoch": 0.2621786197564276,
      "grad_norm": 0.2867293357849121,
      "learning_rate": 4.927239069330807e-06,
      "loss": 0.6761,
      "step": 775
    },
    {
      "epoch": 0.2625169147496617,
      "grad_norm": 0.2768084704875946,
      "learning_rate": 4.927024377484705e-06,
      "loss": 0.6616,
      "step": 776
    },
    {
      "epoch": 0.2628552097428958,
      "grad_norm": 0.27421021461486816,
      "learning_rate": 4.926809374056289e-06,
      "loss": 0.6375,
      "step": 777
    },
    {
      "epoch": 0.2631935047361299,
      "grad_norm": 0.2622529864311218,
      "learning_rate": 4.926594059073163e-06,
      "loss": 0.6457,
      "step": 778
    },
    {
      "epoch": 0.263531799729364,
      "grad_norm": 0.27969247102737427,
      "learning_rate": 4.926378432562967e-06,
      "loss": 0.6683,
      "step": 779
    },
    {
      "epoch": 0.2638700947225981,
      "grad_norm": 0.2766921818256378,
      "learning_rate": 4.926162494553386e-06,
      "loss": 0.664,
      "step": 780
    },
    {
      "epoch": 0.2642083897158322,
      "grad_norm": 0.26445505023002625,
      "learning_rate": 4.925946245072141e-06,
      "loss": 0.6448,
      "step": 781
    },
    {
      "epoch": 0.2645466847090663,
      "grad_norm": 0.2706442177295685,
      "learning_rate": 4.925729684146993e-06,
      "loss": 0.6505,
      "step": 782
    },
    {
      "epoch": 0.2648849797023004,
      "grad_norm": 0.27379173040390015,
      "learning_rate": 4.925512811805746e-06,
      "loss": 0.6697,
      "step": 783
    },
    {
      "epoch": 0.2652232746955345,
      "grad_norm": 0.26962971687316895,
      "learning_rate": 4.925295628076241e-06,
      "loss": 0.6092,
      "step": 784
    },
    {
      "epoch": 0.2655615696887686,
      "grad_norm": 0.2711654007434845,
      "learning_rate": 4.925078132986361e-06,
      "loss": 0.6708,
      "step": 785
    },
    {
      "epoch": 0.2658998646820027,
      "grad_norm": 0.26972121000289917,
      "learning_rate": 4.924860326564027e-06,
      "loss": 0.6659,
      "step": 786
    },
    {
      "epoch": 0.2662381596752368,
      "grad_norm": 0.27795904874801636,
      "learning_rate": 4.9246422088372015e-06,
      "loss": 0.6511,
      "step": 787
    },
    {
      "epoch": 0.2665764546684709,
      "grad_norm": 0.26935580372810364,
      "learning_rate": 4.924423779833887e-06,
      "loss": 0.6264,
      "step": 788
    },
    {
      "epoch": 0.26691474966170503,
      "grad_norm": 0.26369181275367737,
      "learning_rate": 4.924205039582125e-06,
      "loss": 0.6679,
      "step": 789
    },
    {
      "epoch": 0.2672530446549391,
      "grad_norm": 0.2865050733089447,
      "learning_rate": 4.9239859881099984e-06,
      "loss": 0.6591,
      "step": 790
    },
    {
      "epoch": 0.2675913396481732,
      "grad_norm": 0.3010486960411072,
      "learning_rate": 4.923766625445627e-06,
      "loss": 0.6298,
      "step": 791
    },
    {
      "epoch": 0.2679296346414073,
      "grad_norm": 0.26568490266799927,
      "learning_rate": 4.9235469516171754e-06,
      "loss": 0.6755,
      "step": 792
    },
    {
      "epoch": 0.2682679296346414,
      "grad_norm": 0.27336952090263367,
      "learning_rate": 4.923326966652844e-06,
      "loss": 0.6797,
      "step": 793
    },
    {
      "epoch": 0.26860622462787553,
      "grad_norm": 0.29760247468948364,
      "learning_rate": 4.923106670580874e-06,
      "loss": 0.6558,
      "step": 794
    },
    {
      "epoch": 0.2689445196211096,
      "grad_norm": 0.2981927692890167,
      "learning_rate": 4.922886063429547e-06,
      "loss": 0.6575,
      "step": 795
    },
    {
      "epoch": 0.2692828146143437,
      "grad_norm": 0.2672402858734131,
      "learning_rate": 4.9226651452271865e-06,
      "loss": 0.6547,
      "step": 796
    },
    {
      "epoch": 0.2696211096075778,
      "grad_norm": 0.2725027799606323,
      "learning_rate": 4.922443916002153e-06,
      "loss": 0.6709,
      "step": 797
    },
    {
      "epoch": 0.2699594046008119,
      "grad_norm": 0.2933136224746704,
      "learning_rate": 4.922222375782847e-06,
      "loss": 0.6582,
      "step": 798
    },
    {
      "epoch": 0.270297699594046,
      "grad_norm": 0.26434826850891113,
      "learning_rate": 4.922000524597712e-06,
      "loss": 0.6615,
      "step": 799
    },
    {
      "epoch": 0.2706359945872801,
      "grad_norm": 0.24848580360412598,
      "learning_rate": 4.921778362475227e-06,
      "loss": 0.6071,
      "step": 800
    },
    {
      "epoch": 0.2709742895805142,
      "grad_norm": 0.26173582673072815,
      "learning_rate": 4.921555889443914e-06,
      "loss": 0.6705,
      "step": 801
    },
    {
      "epoch": 0.2713125845737483,
      "grad_norm": 0.29065439105033875,
      "learning_rate": 4.9213331055323356e-06,
      "loss": 0.6351,
      "step": 802
    },
    {
      "epoch": 0.2716508795669824,
      "grad_norm": 0.287558913230896,
      "learning_rate": 4.921110010769091e-06,
      "loss": 0.6432,
      "step": 803
    },
    {
      "epoch": 0.2719891745602165,
      "grad_norm": 0.2773423492908478,
      "learning_rate": 4.920886605182823e-06,
      "loss": 0.6495,
      "step": 804
    },
    {
      "epoch": 0.2723274695534506,
      "grad_norm": 0.2841305434703827,
      "learning_rate": 4.9206628888022116e-06,
      "loss": 0.6729,
      "step": 805
    },
    {
      "epoch": 0.27266576454668473,
      "grad_norm": 0.28614985942840576,
      "learning_rate": 4.920438861655977e-06,
      "loss": 0.6524,
      "step": 806
    },
    {
      "epoch": 0.2730040595399188,
      "grad_norm": 0.2771781086921692,
      "learning_rate": 4.9202145237728805e-06,
      "loss": 0.6564,
      "step": 807
    },
    {
      "epoch": 0.2733423545331529,
      "grad_norm": 0.2679559886455536,
      "learning_rate": 4.919989875181722e-06,
      "loss": 0.6593,
      "step": 808
    },
    {
      "epoch": 0.273680649526387,
      "grad_norm": 0.27360039949417114,
      "learning_rate": 4.919764915911344e-06,
      "loss": 0.6367,
      "step": 809
    },
    {
      "epoch": 0.2740189445196211,
      "grad_norm": 0.295267254114151,
      "learning_rate": 4.919539645990624e-06,
      "loss": 0.6485,
      "step": 810
    },
    {
      "epoch": 0.27435723951285523,
      "grad_norm": 0.2728036046028137,
      "learning_rate": 4.919314065448484e-06,
      "loss": 0.6688,
      "step": 811
    },
    {
      "epoch": 0.2746955345060893,
      "grad_norm": 0.2643529772758484,
      "learning_rate": 4.919088174313884e-06,
      "loss": 0.6599,
      "step": 812
    },
    {
      "epoch": 0.2750338294993234,
      "grad_norm": 0.2759469151496887,
      "learning_rate": 4.918861972615824e-06,
      "loss": 0.6354,
      "step": 813
    },
    {
      "epoch": 0.2753721244925575,
      "grad_norm": 0.29973581433296204,
      "learning_rate": 4.918635460383343e-06,
      "loss": 0.6638,
      "step": 814
    },
    {
      "epoch": 0.2757104194857916,
      "grad_norm": 0.2820885479450226,
      "learning_rate": 4.918408637645522e-06,
      "loss": 0.6475,
      "step": 815
    },
    {
      "epoch": 0.27604871447902574,
      "grad_norm": 0.2663920223712921,
      "learning_rate": 4.91818150443148e-06,
      "loss": 0.648,
      "step": 816
    },
    {
      "epoch": 0.2763870094722598,
      "grad_norm": 0.29732027649879456,
      "learning_rate": 4.917954060770376e-06,
      "loss": 0.6461,
      "step": 817
    },
    {
      "epoch": 0.2767253044654939,
      "grad_norm": 0.27227893471717834,
      "learning_rate": 4.917726306691409e-06,
      "loss": 0.6449,
      "step": 818
    },
    {
      "epoch": 0.277063599458728,
      "grad_norm": 0.27002593874931335,
      "learning_rate": 4.91749824222382e-06,
      "loss": 0.6624,
      "step": 819
    },
    {
      "epoch": 0.2774018944519621,
      "grad_norm": 0.2894153594970703,
      "learning_rate": 4.917269867396887e-06,
      "loss": 0.6217,
      "step": 820
    },
    {
      "epoch": 0.2777401894451962,
      "grad_norm": 0.263049453496933,
      "learning_rate": 4.917041182239927e-06,
      "loss": 0.6274,
      "step": 821
    },
    {
      "epoch": 0.2780784844384303,
      "grad_norm": 0.2681529223918915,
      "learning_rate": 4.916812186782303e-06,
      "loss": 0.6323,
      "step": 822
    },
    {
      "epoch": 0.27841677943166443,
      "grad_norm": 0.27024561166763306,
      "learning_rate": 4.9165828810534085e-06,
      "loss": 0.6493,
      "step": 823
    },
    {
      "epoch": 0.2787550744248985,
      "grad_norm": 0.28334054350852966,
      "learning_rate": 4.916353265082686e-06,
      "loss": 0.6519,
      "step": 824
    },
    {
      "epoch": 0.2790933694181326,
      "grad_norm": 0.2738794982433319,
      "learning_rate": 4.916123338899611e-06,
      "loss": 0.6228,
      "step": 825
    },
    {
      "epoch": 0.2794316644113667,
      "grad_norm": 0.2756909430027008,
      "learning_rate": 4.915893102533702e-06,
      "loss": 0.6367,
      "step": 826
    },
    {
      "epoch": 0.2797699594046008,
      "grad_norm": 0.272421270608902,
      "learning_rate": 4.9156625560145185e-06,
      "loss": 0.6688,
      "step": 827
    },
    {
      "epoch": 0.28010825439783493,
      "grad_norm": 0.2805042266845703,
      "learning_rate": 4.915431699371656e-06,
      "loss": 0.6588,
      "step": 828
    },
    {
      "epoch": 0.280446549391069,
      "grad_norm": 0.27977386116981506,
      "learning_rate": 4.915200532634754e-06,
      "loss": 0.6351,
      "step": 829
    },
    {
      "epoch": 0.2807848443843031,
      "grad_norm": 0.26902759075164795,
      "learning_rate": 4.914969055833488e-06,
      "loss": 0.6649,
      "step": 830
    },
    {
      "epoch": 0.2811231393775372,
      "grad_norm": 0.26842644810676575,
      "learning_rate": 4.914737268997576e-06,
      "loss": 0.6335,
      "step": 831
    },
    {
      "epoch": 0.2814614343707713,
      "grad_norm": 0.2780354619026184,
      "learning_rate": 4.914505172156774e-06,
      "loss": 0.6641,
      "step": 832
    },
    {
      "epoch": 0.28179972936400544,
      "grad_norm": 0.276547372341156,
      "learning_rate": 4.91427276534088e-06,
      "loss": 0.6778,
      "step": 833
    },
    {
      "epoch": 0.2821380243572395,
      "grad_norm": 0.26204949617385864,
      "learning_rate": 4.91404004857973e-06,
      "loss": 0.6253,
      "step": 834
    },
    {
      "epoch": 0.2824763193504736,
      "grad_norm": 0.2651567757129669,
      "learning_rate": 4.913807021903199e-06,
      "loss": 0.6357,
      "step": 835
    },
    {
      "epoch": 0.2828146143437077,
      "grad_norm": 0.26619213819503784,
      "learning_rate": 4.913573685341205e-06,
      "loss": 0.6679,
      "step": 836
    },
    {
      "epoch": 0.2831529093369418,
      "grad_norm": 0.26389190554618835,
      "learning_rate": 4.913340038923703e-06,
      "loss": 0.6428,
      "step": 837
    },
    {
      "epoch": 0.28349120433017594,
      "grad_norm": 0.264245867729187,
      "learning_rate": 4.913106082680688e-06,
      "loss": 0.6489,
      "step": 838
    },
    {
      "epoch": 0.28382949932341,
      "grad_norm": 0.27729716897010803,
      "learning_rate": 4.912871816642196e-06,
      "loss": 0.6338,
      "step": 839
    },
    {
      "epoch": 0.28416779431664413,
      "grad_norm": 0.2706574499607086,
      "learning_rate": 4.912637240838302e-06,
      "loss": 0.6252,
      "step": 840
    },
    {
      "epoch": 0.2845060893098782,
      "grad_norm": 0.2795005142688751,
      "learning_rate": 4.912402355299122e-06,
      "loss": 0.6473,
      "step": 841
    },
    {
      "epoch": 0.2848443843031123,
      "grad_norm": 0.2789385914802551,
      "learning_rate": 4.9121671600548094e-06,
      "loss": 0.6632,
      "step": 842
    },
    {
      "epoch": 0.2851826792963464,
      "grad_norm": 0.27970561385154724,
      "learning_rate": 4.911931655135559e-06,
      "loss": 0.6352,
      "step": 843
    },
    {
      "epoch": 0.2855209742895805,
      "grad_norm": 0.2838321030139923,
      "learning_rate": 4.911695840571605e-06,
      "loss": 0.6298,
      "step": 844
    },
    {
      "epoch": 0.28585926928281463,
      "grad_norm": 0.2680176794528961,
      "learning_rate": 4.911459716393221e-06,
      "loss": 0.6386,
      "step": 845
    },
    {
      "epoch": 0.2861975642760487,
      "grad_norm": 0.2741706669330597,
      "learning_rate": 4.911223282630722e-06,
      "loss": 0.642,
      "step": 846
    },
    {
      "epoch": 0.2865358592692828,
      "grad_norm": 0.27544164657592773,
      "learning_rate": 4.91098653931446e-06,
      "loss": 0.6226,
      "step": 847
    },
    {
      "epoch": 0.2868741542625169,
      "grad_norm": 0.2681983709335327,
      "learning_rate": 4.9107494864748285e-06,
      "loss": 0.6822,
      "step": 848
    },
    {
      "epoch": 0.287212449255751,
      "grad_norm": 0.27470800280570984,
      "learning_rate": 4.91051212414226e-06,
      "loss": 0.6666,
      "step": 849
    },
    {
      "epoch": 0.28755074424898514,
      "grad_norm": 0.27337849140167236,
      "learning_rate": 4.91027445234723e-06,
      "loss": 0.6535,
      "step": 850
    },
    {
      "epoch": 0.2878890392422192,
      "grad_norm": 0.26465293765068054,
      "learning_rate": 4.910036471120247e-06,
      "loss": 0.6492,
      "step": 851
    },
    {
      "epoch": 0.2882273342354533,
      "grad_norm": 0.2718878984451294,
      "learning_rate": 4.909798180491865e-06,
      "loss": 0.656,
      "step": 852
    },
    {
      "epoch": 0.2885656292286874,
      "grad_norm": 0.27970167994499207,
      "learning_rate": 4.909559580492677e-06,
      "loss": 0.6487,
      "step": 853
    },
    {
      "epoch": 0.2889039242219215,
      "grad_norm": 0.2778087854385376,
      "learning_rate": 4.909320671153311e-06,
      "loss": 0.6309,
      "step": 854
    },
    {
      "epoch": 0.28924221921515564,
      "grad_norm": 0.29702576994895935,
      "learning_rate": 4.909081452504443e-06,
      "loss": 0.6452,
      "step": 855
    },
    {
      "epoch": 0.2895805142083897,
      "grad_norm": 0.28228092193603516,
      "learning_rate": 4.908841924576781e-06,
      "loss": 0.6476,
      "step": 856
    },
    {
      "epoch": 0.28991880920162383,
      "grad_norm": 0.279044508934021,
      "learning_rate": 4.908602087401075e-06,
      "loss": 0.6609,
      "step": 857
    },
    {
      "epoch": 0.2902571041948579,
      "grad_norm": 0.2703443467617035,
      "learning_rate": 4.9083619410081165e-06,
      "loss": 0.6392,
      "step": 858
    },
    {
      "epoch": 0.290595399188092,
      "grad_norm": 0.273985356092453,
      "learning_rate": 4.908121485428737e-06,
      "loss": 0.6312,
      "step": 859
    },
    {
      "epoch": 0.29093369418132614,
      "grad_norm": 0.271116703748703,
      "learning_rate": 4.907880720693804e-06,
      "loss": 0.6329,
      "step": 860
    },
    {
      "epoch": 0.2912719891745602,
      "grad_norm": 0.2669009864330292,
      "learning_rate": 4.907639646834229e-06,
      "loss": 0.6406,
      "step": 861
    },
    {
      "epoch": 0.29161028416779433,
      "grad_norm": 0.262957900762558,
      "learning_rate": 4.907398263880959e-06,
      "loss": 0.6125,
      "step": 862
    },
    {
      "epoch": 0.2919485791610284,
      "grad_norm": 0.2755131423473358,
      "learning_rate": 4.907156571864984e-06,
      "loss": 0.654,
      "step": 863
    },
    {
      "epoch": 0.2922868741542625,
      "grad_norm": 0.28943315148353577,
      "learning_rate": 4.906914570817333e-06,
      "loss": 0.6432,
      "step": 864
    },
    {
      "epoch": 0.29262516914749664,
      "grad_norm": 0.2823984622955322,
      "learning_rate": 4.9066722607690734e-06,
      "loss": 0.6548,
      "step": 865
    },
    {
      "epoch": 0.2929634641407307,
      "grad_norm": 0.2662210464477539,
      "learning_rate": 4.906429641751313e-06,
      "loss": 0.6448,
      "step": 866
    },
    {
      "epoch": 0.29330175913396483,
      "grad_norm": 0.27130478620529175,
      "learning_rate": 4.9061867137952006e-06,
      "loss": 0.6261,
      "step": 867
    },
    {
      "epoch": 0.2936400541271989,
      "grad_norm": 0.2734956741333008,
      "learning_rate": 4.9059434769319205e-06,
      "loss": 0.6509,
      "step": 868
    },
    {
      "epoch": 0.293978349120433,
      "grad_norm": 0.2768975496292114,
      "learning_rate": 4.905699931192703e-06,
      "loss": 0.6399,
      "step": 869
    },
    {
      "epoch": 0.2943166441136671,
      "grad_norm": 0.2712252140045166,
      "learning_rate": 4.905456076608812e-06,
      "loss": 0.6462,
      "step": 870
    },
    {
      "epoch": 0.2946549391069012,
      "grad_norm": 0.2708142101764679,
      "learning_rate": 4.905211913211555e-06,
      "loss": 0.6442,
      "step": 871
    },
    {
      "epoch": 0.29499323410013534,
      "grad_norm": 0.2733133137226105,
      "learning_rate": 4.9049674410322775e-06,
      "loss": 0.6661,
      "step": 872
    },
    {
      "epoch": 0.2953315290933694,
      "grad_norm": 0.28249210119247437,
      "learning_rate": 4.904722660102366e-06,
      "loss": 0.6331,
      "step": 873
    },
    {
      "epoch": 0.2956698240866035,
      "grad_norm": 0.28206610679626465,
      "learning_rate": 4.904477570453243e-06,
      "loss": 0.6507,
      "step": 874
    },
    {
      "epoch": 0.2960081190798376,
      "grad_norm": 0.26992663741111755,
      "learning_rate": 4.904232172116375e-06,
      "loss": 0.6725,
      "step": 875
    },
    {
      "epoch": 0.2963464140730717,
      "grad_norm": 0.27740415930747986,
      "learning_rate": 4.903986465123266e-06,
      "loss": 0.6486,
      "step": 876
    },
    {
      "epoch": 0.29668470906630584,
      "grad_norm": 0.2828187346458435,
      "learning_rate": 4.903740449505461e-06,
      "loss": 0.6728,
      "step": 877
    },
    {
      "epoch": 0.2970230040595399,
      "grad_norm": 0.27313926815986633,
      "learning_rate": 4.903494125294541e-06,
      "loss": 0.6422,
      "step": 878
    },
    {
      "epoch": 0.29736129905277403,
      "grad_norm": 0.2670018970966339,
      "learning_rate": 4.903247492522131e-06,
      "loss": 0.6512,
      "step": 879
    },
    {
      "epoch": 0.2976995940460081,
      "grad_norm": 0.2771807312965393,
      "learning_rate": 4.903000551219894e-06,
      "loss": 0.6448,
      "step": 880
    },
    {
      "epoch": 0.2980378890392422,
      "grad_norm": 0.2740825414657593,
      "learning_rate": 4.902753301419532e-06,
      "loss": 0.6555,
      "step": 881
    },
    {
      "epoch": 0.29837618403247634,
      "grad_norm": 0.26953041553497314,
      "learning_rate": 4.902505743152786e-06,
      "loss": 0.616,
      "step": 882
    },
    {
      "epoch": 0.2987144790257104,
      "grad_norm": 0.2877902686595917,
      "learning_rate": 4.90225787645144e-06,
      "loss": 0.6762,
      "step": 883
    },
    {
      "epoch": 0.29905277401894453,
      "grad_norm": 0.275796115398407,
      "learning_rate": 4.902009701347313e-06,
      "loss": 0.6637,
      "step": 884
    },
    {
      "epoch": 0.2993910690121786,
      "grad_norm": 0.28130945563316345,
      "learning_rate": 4.9017612178722665e-06,
      "loss": 0.652,
      "step": 885
    },
    {
      "epoch": 0.2997293640054127,
      "grad_norm": 0.29672494530677795,
      "learning_rate": 4.901512426058202e-06,
      "loss": 0.659,
      "step": 886
    },
    {
      "epoch": 0.30006765899864685,
      "grad_norm": 0.27471110224723816,
      "learning_rate": 4.901263325937057e-06,
      "loss": 0.6305,
      "step": 887
    },
    {
      "epoch": 0.3004059539918809,
      "grad_norm": 0.26989707350730896,
      "learning_rate": 4.901013917540814e-06,
      "loss": 0.6655,
      "step": 888
    },
    {
      "epoch": 0.30074424898511504,
      "grad_norm": 0.2895686626434326,
      "learning_rate": 4.90076420090149e-06,
      "loss": 0.6305,
      "step": 889
    },
    {
      "epoch": 0.3010825439783491,
      "grad_norm": 0.2914600074291229,
      "learning_rate": 4.900514176051144e-06,
      "loss": 0.6623,
      "step": 890
    },
    {
      "epoch": 0.3014208389715832,
      "grad_norm": 0.27903082966804504,
      "learning_rate": 4.900263843021876e-06,
      "loss": 0.6194,
      "step": 891
    },
    {
      "epoch": 0.3017591339648173,
      "grad_norm": 0.2799786925315857,
      "learning_rate": 4.900013201845821e-06,
      "loss": 0.6553,
      "step": 892
    },
    {
      "epoch": 0.3020974289580514,
      "grad_norm": 0.27043768763542175,
      "learning_rate": 4.89976225255516e-06,
      "loss": 0.6641,
      "step": 893
    },
    {
      "epoch": 0.30243572395128554,
      "grad_norm": 0.27861344814300537,
      "learning_rate": 4.899510995182107e-06,
      "loss": 0.6405,
      "step": 894
    },
    {
      "epoch": 0.3027740189445196,
      "grad_norm": 0.28648340702056885,
      "learning_rate": 4.8992594297589194e-06,
      "loss": 0.6421,
      "step": 895
    },
    {
      "epoch": 0.30311231393775373,
      "grad_norm": 0.27372077107429504,
      "learning_rate": 4.899007556317893e-06,
      "loss": 0.6481,
      "step": 896
    },
    {
      "epoch": 0.3034506089309878,
      "grad_norm": 0.2841525673866272,
      "learning_rate": 4.898755374891364e-06,
      "loss": 0.6849,
      "step": 897
    },
    {
      "epoch": 0.3037889039242219,
      "grad_norm": 0.2740134298801422,
      "learning_rate": 4.898502885511707e-06,
      "loss": 0.6327,
      "step": 898
    },
    {
      "epoch": 0.30412719891745604,
      "grad_norm": 0.264041930437088,
      "learning_rate": 4.898250088211337e-06,
      "loss": 0.6478,
      "step": 899
    },
    {
      "epoch": 0.3044654939106901,
      "grad_norm": 0.27168142795562744,
      "learning_rate": 4.897996983022709e-06,
      "loss": 0.6394,
      "step": 900
    },
    {
      "epoch": 0.30480378890392423,
      "grad_norm": 0.2795570194721222,
      "learning_rate": 4.897743569978315e-06,
      "loss": 0.659,
      "step": 901
    },
    {
      "epoch": 0.3051420838971583,
      "grad_norm": 0.2912237346172333,
      "learning_rate": 4.8974898491106895e-06,
      "loss": 0.681,
      "step": 902
    },
    {
      "epoch": 0.3054803788903924,
      "grad_norm": 0.27849677205085754,
      "learning_rate": 4.897235820452405e-06,
      "loss": 0.6362,
      "step": 903
    },
    {
      "epoch": 0.30581867388362655,
      "grad_norm": 0.274339497089386,
      "learning_rate": 4.896981484036074e-06,
      "loss": 0.6646,
      "step": 904
    },
    {
      "epoch": 0.3061569688768606,
      "grad_norm": 0.29278630018234253,
      "learning_rate": 4.896726839894348e-06,
      "loss": 0.6681,
      "step": 905
    },
    {
      "epoch": 0.30649526387009474,
      "grad_norm": 0.27960678935050964,
      "learning_rate": 4.896471888059918e-06,
      "loss": 0.6597,
      "step": 906
    },
    {
      "epoch": 0.3068335588633288,
      "grad_norm": 0.2835066318511963,
      "learning_rate": 4.896216628565515e-06,
      "loss": 0.639,
      "step": 907
    },
    {
      "epoch": 0.3071718538565629,
      "grad_norm": 0.27574077248573303,
      "learning_rate": 4.895961061443912e-06,
      "loss": 0.6654,
      "step": 908
    },
    {
      "epoch": 0.30751014884979705,
      "grad_norm": 0.29681262373924255,
      "learning_rate": 4.895705186727913e-06,
      "loss": 0.6216,
      "step": 909
    },
    {
      "epoch": 0.3078484438430311,
      "grad_norm": 0.2979026436805725,
      "learning_rate": 4.8954490044503726e-06,
      "loss": 0.6562,
      "step": 910
    },
    {
      "epoch": 0.30818673883626524,
      "grad_norm": 0.28780779242515564,
      "learning_rate": 4.895192514644176e-06,
      "loss": 0.636,
      "step": 911
    },
    {
      "epoch": 0.3085250338294993,
      "grad_norm": 0.2931484580039978,
      "learning_rate": 4.894935717342255e-06,
      "loss": 0.6479,
      "step": 912
    },
    {
      "epoch": 0.30886332882273343,
      "grad_norm": 0.29892581701278687,
      "learning_rate": 4.894678612577574e-06,
      "loss": 0.654,
      "step": 913
    },
    {
      "epoch": 0.3092016238159675,
      "grad_norm": 0.29458022117614746,
      "learning_rate": 4.894421200383142e-06,
      "loss": 0.6537,
      "step": 914
    },
    {
      "epoch": 0.3095399188092016,
      "grad_norm": 0.2971714735031128,
      "learning_rate": 4.894163480792006e-06,
      "loss": 0.6514,
      "step": 915
    },
    {
      "epoch": 0.30987821380243574,
      "grad_norm": 0.28501057624816895,
      "learning_rate": 4.89390545383725e-06,
      "loss": 0.6881,
      "step": 916
    },
    {
      "epoch": 0.3102165087956698,
      "grad_norm": 0.2875775098800659,
      "learning_rate": 4.893647119552002e-06,
      "loss": 0.6531,
      "step": 917
    },
    {
      "epoch": 0.31055480378890393,
      "grad_norm": 0.28826165199279785,
      "learning_rate": 4.893388477969425e-06,
      "loss": 0.6387,
      "step": 918
    },
    {
      "epoch": 0.310893098782138,
      "grad_norm": 0.2894171476364136,
      "learning_rate": 4.893129529122725e-06,
      "loss": 0.643,
      "step": 919
    },
    {
      "epoch": 0.3112313937753721,
      "grad_norm": 0.2911660969257355,
      "learning_rate": 4.892870273045146e-06,
      "loss": 0.6703,
      "step": 920
    },
    {
      "epoch": 0.31156968876860625,
      "grad_norm": 0.2881146967411041,
      "learning_rate": 4.892610709769971e-06,
      "loss": 0.6624,
      "step": 921
    },
    {
      "epoch": 0.3119079837618403,
      "grad_norm": 0.288736492395401,
      "learning_rate": 4.8923508393305224e-06,
      "loss": 0.6588,
      "step": 922
    },
    {
      "epoch": 0.31224627875507444,
      "grad_norm": 0.2821895480155945,
      "learning_rate": 4.892090661760163e-06,
      "loss": 0.6462,
      "step": 923
    },
    {
      "epoch": 0.3125845737483085,
      "grad_norm": 0.2836967706680298,
      "learning_rate": 4.891830177092294e-06,
      "loss": 0.6612,
      "step": 924
    },
    {
      "epoch": 0.3129228687415426,
      "grad_norm": 0.2861119210720062,
      "learning_rate": 4.8915693853603576e-06,
      "loss": 0.65,
      "step": 925
    },
    {
      "epoch": 0.31326116373477675,
      "grad_norm": 0.2934957444667816,
      "learning_rate": 4.891308286597832e-06,
      "loss": 0.662,
      "step": 926
    },
    {
      "epoch": 0.3135994587280108,
      "grad_norm": 0.280984103679657,
      "learning_rate": 4.891046880838241e-06,
      "loss": 0.6539,
      "step": 927
    },
    {
      "epoch": 0.31393775372124494,
      "grad_norm": 0.27881526947021484,
      "learning_rate": 4.89078516811514e-06,
      "loss": 0.6465,
      "step": 928
    },
    {
      "epoch": 0.314276048714479,
      "grad_norm": 0.2846492826938629,
      "learning_rate": 4.89052314846213e-06,
      "loss": 0.6333,
      "step": 929
    },
    {
      "epoch": 0.31461434370771313,
      "grad_norm": 0.2940545380115509,
      "learning_rate": 4.890260821912848e-06,
      "loss": 0.6067,
      "step": 930
    },
    {
      "epoch": 0.31495263870094725,
      "grad_norm": 0.28840309381484985,
      "learning_rate": 4.889998188500973e-06,
      "loss": 0.6233,
      "step": 931
    },
    {
      "epoch": 0.3152909336941813,
      "grad_norm": 0.26783108711242676,
      "learning_rate": 4.889735248260221e-06,
      "loss": 0.6431,
      "step": 932
    },
    {
      "epoch": 0.31562922868741544,
      "grad_norm": 0.28233930468559265,
      "learning_rate": 4.889472001224348e-06,
      "loss": 0.6512,
      "step": 933
    },
    {
      "epoch": 0.3159675236806495,
      "grad_norm": 0.2793835699558258,
      "learning_rate": 4.8892084474271515e-06,
      "loss": 0.6427,
      "step": 934
    },
    {
      "epoch": 0.31630581867388363,
      "grad_norm": 0.29794105887413025,
      "learning_rate": 4.888944586902464e-06,
      "loss": 0.6256,
      "step": 935
    },
    {
      "epoch": 0.3166441136671177,
      "grad_norm": 0.2742573618888855,
      "learning_rate": 4.8886804196841634e-06,
      "loss": 0.6423,
      "step": 936
    },
    {
      "epoch": 0.3169824086603518,
      "grad_norm": 0.29450520873069763,
      "learning_rate": 4.8884159458061604e-06,
      "loss": 0.6399,
      "step": 937
    },
    {
      "epoch": 0.31732070365358594,
      "grad_norm": 0.28303098678588867,
      "learning_rate": 4.888151165302409e-06,
      "loss": 0.6366,
      "step": 938
    },
    {
      "epoch": 0.31765899864682,
      "grad_norm": 0.28655314445495605,
      "learning_rate": 4.887886078206904e-06,
      "loss": 0.655,
      "step": 939
    },
    {
      "epoch": 0.31799729364005414,
      "grad_norm": 0.28076592087745667,
      "learning_rate": 4.887620684553675e-06,
      "loss": 0.6391,
      "step": 940
    },
    {
      "epoch": 0.3183355886332882,
      "grad_norm": 0.2662147283554077,
      "learning_rate": 4.8873549843767935e-06,
      "loss": 0.6394,
      "step": 941
    },
    {
      "epoch": 0.3186738836265223,
      "grad_norm": 0.28641197085380554,
      "learning_rate": 4.887088977710371e-06,
      "loss": 0.6288,
      "step": 942
    },
    {
      "epoch": 0.31901217861975645,
      "grad_norm": 0.2831316292285919,
      "learning_rate": 4.886822664588558e-06,
      "loss": 0.6378,
      "step": 943
    },
    {
      "epoch": 0.3193504736129905,
      "grad_norm": 0.27543431520462036,
      "learning_rate": 4.886556045045542e-06,
      "loss": 0.6125,
      "step": 944
    },
    {
      "epoch": 0.31968876860622464,
      "grad_norm": 0.2942037880420685,
      "learning_rate": 4.886289119115554e-06,
      "loss": 0.6516,
      "step": 945
    },
    {
      "epoch": 0.3200270635994587,
      "grad_norm": 0.28229376673698425,
      "learning_rate": 4.886021886832861e-06,
      "loss": 0.6216,
      "step": 946
    },
    {
      "epoch": 0.32036535859269283,
      "grad_norm": 0.27998530864715576,
      "learning_rate": 4.885754348231769e-06,
      "loss": 0.6612,
      "step": 947
    },
    {
      "epoch": 0.32070365358592695,
      "grad_norm": 0.29919928312301636,
      "learning_rate": 4.885486503346628e-06,
      "loss": 0.6532,
      "step": 948
    },
    {
      "epoch": 0.321041948579161,
      "grad_norm": 0.3005315363407135,
      "learning_rate": 4.88521835221182e-06,
      "loss": 0.6491,
      "step": 949
    },
    {
      "epoch": 0.32138024357239514,
      "grad_norm": 0.30181828141212463,
      "learning_rate": 4.884949894861774e-06,
      "loss": 0.6376,
      "step": 950
    },
    {
      "epoch": 0.3217185385656292,
      "grad_norm": 0.2998781204223633,
      "learning_rate": 4.884681131330953e-06,
      "loss": 0.676,
      "step": 951
    },
    {
      "epoch": 0.32205683355886333,
      "grad_norm": 0.27801403403282166,
      "learning_rate": 4.88441206165386e-06,
      "loss": 0.6399,
      "step": 952
    },
    {
      "epoch": 0.32239512855209745,
      "grad_norm": 0.2941897511482239,
      "learning_rate": 4.88414268586504e-06,
      "loss": 0.6376,
      "step": 953
    },
    {
      "epoch": 0.3227334235453315,
      "grad_norm": 0.2913419008255005,
      "learning_rate": 4.883873003999075e-06,
      "loss": 0.6379,
      "step": 954
    },
    {
      "epoch": 0.32307171853856564,
      "grad_norm": 0.2757265865802765,
      "learning_rate": 4.883603016090586e-06,
      "loss": 0.6473,
      "step": 955
    },
    {
      "epoch": 0.3234100135317997,
      "grad_norm": 0.28613418340682983,
      "learning_rate": 4.883332722174236e-06,
      "loss": 0.6618,
      "step": 956
    },
    {
      "epoch": 0.32374830852503383,
      "grad_norm": 0.2763930857181549,
      "learning_rate": 4.883062122284723e-06,
      "loss": 0.6264,
      "step": 957
    },
    {
      "epoch": 0.32408660351826796,
      "grad_norm": 0.27556413412094116,
      "learning_rate": 4.882791216456789e-06,
      "loss": 0.6497,
      "step": 958
    },
    {
      "epoch": 0.324424898511502,
      "grad_norm": 0.28852880001068115,
      "learning_rate": 4.882520004725212e-06,
      "loss": 0.6384,
      "step": 959
    },
    {
      "epoch": 0.32476319350473615,
      "grad_norm": 0.29310303926467896,
      "learning_rate": 4.88224848712481e-06,
      "loss": 0.6453,
      "step": 960
    },
    {
      "epoch": 0.3251014884979702,
      "grad_norm": 0.2852711081504822,
      "learning_rate": 4.881976663690441e-06,
      "loss": 0.662,
      "step": 961
    },
    {
      "epoch": 0.32543978349120434,
      "grad_norm": 0.30317068099975586,
      "learning_rate": 4.881704534457001e-06,
      "loss": 0.6489,
      "step": 962
    },
    {
      "epoch": 0.3257780784844384,
      "grad_norm": 0.29215002059936523,
      "learning_rate": 4.881432099459427e-06,
      "loss": 0.6365,
      "step": 963
    },
    {
      "epoch": 0.3261163734776725,
      "grad_norm": 0.2727847099304199,
      "learning_rate": 4.881159358732695e-06,
      "loss": 0.6483,
      "step": 964
    },
    {
      "epoch": 0.32645466847090665,
      "grad_norm": 0.28112176060676575,
      "learning_rate": 4.880886312311817e-06,
      "loss": 0.6496,
      "step": 965
    },
    {
      "epoch": 0.3267929634641407,
      "grad_norm": 0.2979572117328644,
      "learning_rate": 4.880612960231849e-06,
      "loss": 0.6684,
      "step": 966
    },
    {
      "epoch": 0.32713125845737484,
      "grad_norm": 0.2899046838283539,
      "learning_rate": 4.880339302527883e-06,
      "loss": 0.6463,
      "step": 967
    },
    {
      "epoch": 0.3274695534506089,
      "grad_norm": 0.28542327880859375,
      "learning_rate": 4.880065339235053e-06,
      "loss": 0.6415,
      "step": 968
    },
    {
      "epoch": 0.32780784844384303,
      "grad_norm": 0.27839645743370056,
      "learning_rate": 4.8797910703885275e-06,
      "loss": 0.6607,
      "step": 969
    },
    {
      "epoch": 0.32814614343707715,
      "grad_norm": 0.3090164065361023,
      "learning_rate": 4.8795164960235206e-06,
      "loss": 0.644,
      "step": 970
    },
    {
      "epoch": 0.3284844384303112,
      "grad_norm": 0.2935158610343933,
      "learning_rate": 4.879241616175279e-06,
      "loss": 0.6398,
      "step": 971
    },
    {
      "epoch": 0.32882273342354534,
      "grad_norm": 0.2757309377193451,
      "learning_rate": 4.878966430879094e-06,
      "loss": 0.6296,
      "step": 972
    },
    {
      "epoch": 0.3291610284167794,
      "grad_norm": 0.3061336874961853,
      "learning_rate": 4.8786909401702944e-06,
      "loss": 0.6394,
      "step": 973
    },
    {
      "epoch": 0.32949932341001353,
      "grad_norm": 0.297267347574234,
      "learning_rate": 4.8784151440842455e-06,
      "loss": 0.6492,
      "step": 974
    },
    {
      "epoch": 0.32983761840324766,
      "grad_norm": 0.29051071405410767,
      "learning_rate": 4.878139042656356e-06,
      "loss": 0.6316,
      "step": 975
    },
    {
      "epoch": 0.3301759133964817,
      "grad_norm": 0.2814705967903137,
      "learning_rate": 4.877862635922071e-06,
      "loss": 0.6727,
      "step": 976
    },
    {
      "epoch": 0.33051420838971585,
      "grad_norm": 0.282053142786026,
      "learning_rate": 4.877585923916877e-06,
      "loss": 0.6201,
      "step": 977
    },
    {
      "epoch": 0.3308525033829499,
      "grad_norm": 0.27852943539619446,
      "learning_rate": 4.877308906676297e-06,
      "loss": 0.6466,
      "step": 978
    },
    {
      "epoch": 0.33119079837618404,
      "grad_norm": 0.2765594720840454,
      "learning_rate": 4.877031584235895e-06,
      "loss": 0.6282,
      "step": 979
    },
    {
      "epoch": 0.33152909336941816,
      "grad_norm": 0.28560516238212585,
      "learning_rate": 4.876753956631274e-06,
      "loss": 0.6581,
      "step": 980
    },
    {
      "epoch": 0.3318673883626522,
      "grad_norm": 0.27953213453292847,
      "learning_rate": 4.876476023898076e-06,
      "loss": 0.6319,
      "step": 981
    },
    {
      "epoch": 0.33220568335588635,
      "grad_norm": 0.29077577590942383,
      "learning_rate": 4.876197786071981e-06,
      "loss": 0.6608,
      "step": 982
    },
    {
      "epoch": 0.3325439783491204,
      "grad_norm": 0.2820235788822174,
      "learning_rate": 4.87591924318871e-06,
      "loss": 0.6464,
      "step": 983
    },
    {
      "epoch": 0.33288227334235454,
      "grad_norm": 0.297158420085907,
      "learning_rate": 4.875640395284023e-06,
      "loss": 0.6358,
      "step": 984
    },
    {
      "epoch": 0.3332205683355886,
      "grad_norm": 0.2829797863960266,
      "learning_rate": 4.875361242393719e-06,
      "loss": 0.6352,
      "step": 985
    },
    {
      "epoch": 0.33355886332882273,
      "grad_norm": 0.284898579120636,
      "learning_rate": 4.875081784553634e-06,
      "loss": 0.6305,
      "step": 986
    },
    {
      "epoch": 0.33389715832205685,
      "grad_norm": 0.2893214523792267,
      "learning_rate": 4.8748020217996465e-06,
      "loss": 0.6286,
      "step": 987
    },
    {
      "epoch": 0.3342354533152909,
      "grad_norm": 0.29545867443084717,
      "learning_rate": 4.874521954167671e-06,
      "loss": 0.6374,
      "step": 988
    },
    {
      "epoch": 0.33457374830852504,
      "grad_norm": 0.2779877185821533,
      "learning_rate": 4.874241581693664e-06,
      "loss": 0.6334,
      "step": 989
    },
    {
      "epoch": 0.3349120433017591,
      "grad_norm": 0.2803078591823578,
      "learning_rate": 4.87396090441362e-06,
      "loss": 0.6608,
      "step": 990
    },
    {
      "epoch": 0.33525033829499323,
      "grad_norm": 0.27986928820610046,
      "learning_rate": 4.873679922363571e-06,
      "loss": 0.6628,
      "step": 991
    },
    {
      "epoch": 0.33558863328822736,
      "grad_norm": 0.30071017146110535,
      "learning_rate": 4.87339863557959e-06,
      "loss": 0.6446,
      "step": 992
    },
    {
      "epoch": 0.3359269282814614,
      "grad_norm": 0.2871248424053192,
      "learning_rate": 4.87311704409779e-06,
      "loss": 0.6347,
      "step": 993
    },
    {
      "epoch": 0.33626522327469555,
      "grad_norm": 0.2827127277851105,
      "learning_rate": 4.872835147954321e-06,
      "loss": 0.6415,
      "step": 994
    },
    {
      "epoch": 0.3366035182679296,
      "grad_norm": 0.28385990858078003,
      "learning_rate": 4.8725529471853726e-06,
      "loss": 0.5986,
      "step": 995
    },
    {
      "epoch": 0.33694181326116374,
      "grad_norm": 0.26931530237197876,
      "learning_rate": 4.872270441827174e-06,
      "loss": 0.6573,
      "step": 996
    },
    {
      "epoch": 0.33728010825439786,
      "grad_norm": 0.29211145639419556,
      "learning_rate": 4.871987631915994e-06,
      "loss": 0.6367,
      "step": 997
    },
    {
      "epoch": 0.3376184032476319,
      "grad_norm": 0.2922752797603607,
      "learning_rate": 4.871704517488139e-06,
      "loss": 0.6503,
      "step": 998
    },
    {
      "epoch": 0.33795669824086605,
      "grad_norm": 0.2984164357185364,
      "learning_rate": 4.8714210985799556e-06,
      "loss": 0.6605,
      "step": 999
    },
    {
      "epoch": 0.3382949932341001,
      "grad_norm": 0.28730762004852295,
      "learning_rate": 4.871137375227829e-06,
      "loss": 0.6365,
      "step": 1000
    },
    {
      "epoch": 0.33863328822733424,
      "grad_norm": 0.3080568015575409,
      "learning_rate": 4.8708533474681844e-06,
      "loss": 0.6491,
      "step": 1001
    },
    {
      "epoch": 0.33897158322056836,
      "grad_norm": 0.3268609642982483,
      "learning_rate": 4.870569015337485e-06,
      "loss": 0.6398,
      "step": 1002
    },
    {
      "epoch": 0.33930987821380243,
      "grad_norm": 0.28499341011047363,
      "learning_rate": 4.8702843788722335e-06,
      "loss": 0.6574,
      "step": 1003
    },
    {
      "epoch": 0.33964817320703655,
      "grad_norm": 0.3010106384754181,
      "learning_rate": 4.8699994381089715e-06,
      "loss": 0.6277,
      "step": 1004
    },
    {
      "epoch": 0.3399864682002706,
      "grad_norm": 0.29605910181999207,
      "learning_rate": 4.86971419308428e-06,
      "loss": 0.6473,
      "step": 1005
    },
    {
      "epoch": 0.34032476319350474,
      "grad_norm": 0.2929319441318512,
      "learning_rate": 4.869428643834779e-06,
      "loss": 0.6541,
      "step": 1006
    },
    {
      "epoch": 0.3406630581867388,
      "grad_norm": 0.2899619936943054,
      "learning_rate": 4.869142790397126e-06,
      "loss": 0.6425,
      "step": 1007
    },
    {
      "epoch": 0.34100135317997293,
      "grad_norm": 0.29885226488113403,
      "learning_rate": 4.8688566328080215e-06,
      "loss": 0.6462,
      "step": 1008
    },
    {
      "epoch": 0.34133964817320706,
      "grad_norm": 0.28795602917671204,
      "learning_rate": 4.868570171104201e-06,
      "loss": 0.6526,
      "step": 1009
    },
    {
      "epoch": 0.3416779431664411,
      "grad_norm": 0.27857333421707153,
      "learning_rate": 4.86828340532244e-06,
      "loss": 0.6579,
      "step": 1010
    },
    {
      "epoch": 0.34201623815967525,
      "grad_norm": 0.30258554220199585,
      "learning_rate": 4.867996335499555e-06,
      "loss": 0.6285,
      "step": 1011
    },
    {
      "epoch": 0.3423545331529093,
      "grad_norm": 0.284223735332489,
      "learning_rate": 4.867708961672399e-06,
      "loss": 0.6259,
      "step": 1012
    },
    {
      "epoch": 0.34269282814614344,
      "grad_norm": 0.29295721650123596,
      "learning_rate": 4.867421283877866e-06,
      "loss": 0.6311,
      "step": 1013
    },
    {
      "epoch": 0.34303112313937756,
      "grad_norm": 0.2908198833465576,
      "learning_rate": 4.867133302152888e-06,
      "loss": 0.6339,
      "step": 1014
    },
    {
      "epoch": 0.3433694181326116,
      "grad_norm": 0.2774893641471863,
      "learning_rate": 4.866845016534436e-06,
      "loss": 0.6468,
      "step": 1015
    },
    {
      "epoch": 0.34370771312584575,
      "grad_norm": 0.2836349904537201,
      "learning_rate": 4.86655642705952e-06,
      "loss": 0.6219,
      "step": 1016
    },
    {
      "epoch": 0.3440460081190798,
      "grad_norm": 0.2842315435409546,
      "learning_rate": 4.8662675337651886e-06,
      "loss": 0.6248,
      "step": 1017
    },
    {
      "epoch": 0.34438430311231394,
      "grad_norm": 0.2972695827484131,
      "learning_rate": 4.865978336688532e-06,
      "loss": 0.676,
      "step": 1018
    },
    {
      "epoch": 0.34472259810554806,
      "grad_norm": 0.2845618724822998,
      "learning_rate": 4.865688835866677e-06,
      "loss": 0.6475,
      "step": 1019
    },
    {
      "epoch": 0.34506089309878213,
      "grad_norm": 0.2947195768356323,
      "learning_rate": 4.8653990313367875e-06,
      "loss": 0.6472,
      "step": 1020
    },
    {
      "epoch": 0.34539918809201625,
      "grad_norm": 0.28122177720069885,
      "learning_rate": 4.865108923136071e-06,
      "loss": 0.6306,
      "step": 1021
    },
    {
      "epoch": 0.3457374830852503,
      "grad_norm": 0.29748567938804626,
      "learning_rate": 4.864818511301771e-06,
      "loss": 0.6618,
      "step": 1022
    },
    {
      "epoch": 0.34607577807848444,
      "grad_norm": 0.2848156690597534,
      "learning_rate": 4.864527795871172e-06,
      "loss": 0.6294,
      "step": 1023
    },
    {
      "epoch": 0.34641407307171856,
      "grad_norm": 0.2786930799484253,
      "learning_rate": 4.864236776881593e-06,
      "loss": 0.67,
      "step": 1024
    },
    {
      "epoch": 0.34675236806495263,
      "grad_norm": 0.28665855526924133,
      "learning_rate": 4.863945454370398e-06,
      "loss": 0.665,
      "step": 1025
    },
    {
      "epoch": 0.34709066305818675,
      "grad_norm": 0.30270564556121826,
      "learning_rate": 4.863653828374986e-06,
      "loss": 0.6347,
      "step": 1026
    },
    {
      "epoch": 0.3474289580514208,
      "grad_norm": 0.28528571128845215,
      "learning_rate": 4.863361898932796e-06,
      "loss": 0.6698,
      "step": 1027
    },
    {
      "epoch": 0.34776725304465494,
      "grad_norm": 0.28242045640945435,
      "learning_rate": 4.863069666081307e-06,
      "loss": 0.6253,
      "step": 1028
    },
    {
      "epoch": 0.348105548037889,
      "grad_norm": 0.2775231599807739,
      "learning_rate": 4.862777129858034e-06,
      "loss": 0.606,
      "step": 1029
    },
    {
      "epoch": 0.34844384303112313,
      "grad_norm": 0.2802172601222992,
      "learning_rate": 4.862484290300536e-06,
      "loss": 0.6675,
      "step": 1030
    },
    {
      "epoch": 0.34878213802435726,
      "grad_norm": 0.2965852916240692,
      "learning_rate": 4.862191147446403e-06,
      "loss": 0.6503,
      "step": 1031
    },
    {
      "epoch": 0.3491204330175913,
      "grad_norm": 0.2785798907279968,
      "learning_rate": 4.8618977013332744e-06,
      "loss": 0.6122,
      "step": 1032
    },
    {
      "epoch": 0.34945872801082545,
      "grad_norm": 0.2883950471878052,
      "learning_rate": 4.861603951998819e-06,
      "loss": 0.6526,
      "step": 1033
    },
    {
      "epoch": 0.3497970230040595,
      "grad_norm": 0.2888367474079132,
      "learning_rate": 4.86130989948075e-06,
      "loss": 0.6418,
      "step": 1034
    },
    {
      "epoch": 0.35013531799729364,
      "grad_norm": 0.30539917945861816,
      "learning_rate": 4.8610155438168175e-06,
      "loss": 0.6232,
      "step": 1035
    },
    {
      "epoch": 0.35047361299052776,
      "grad_norm": 0.28636616468429565,
      "learning_rate": 4.860720885044811e-06,
      "loss": 0.6353,
      "step": 1036
    },
    {
      "epoch": 0.35081190798376183,
      "grad_norm": 0.2941320240497589,
      "learning_rate": 4.86042592320256e-06,
      "loss": 0.6699,
      "step": 1037
    },
    {
      "epoch": 0.35115020297699595,
      "grad_norm": 0.28965914249420166,
      "learning_rate": 4.860130658327929e-06,
      "loss": 0.6387,
      "step": 1038
    },
    {
      "epoch": 0.35148849797023,
      "grad_norm": 0.28881844878196716,
      "learning_rate": 4.859835090458827e-06,
      "loss": 0.6532,
      "step": 1039
    },
    {
      "epoch": 0.35182679296346414,
      "grad_norm": 0.2799035906791687,
      "learning_rate": 4.859539219633199e-06,
      "loss": 0.635,
      "step": 1040
    },
    {
      "epoch": 0.35216508795669826,
      "grad_norm": 0.2909943759441376,
      "learning_rate": 4.859243045889028e-06,
      "loss": 0.6236,
      "step": 1041
    },
    {
      "epoch": 0.35250338294993233,
      "grad_norm": 0.3185156583786011,
      "learning_rate": 4.858946569264337e-06,
      "loss": 0.6398,
      "step": 1042
    },
    {
      "epoch": 0.35284167794316645,
      "grad_norm": 0.3125178813934326,
      "learning_rate": 4.858649789797188e-06,
      "loss": 0.6224,
      "step": 1043
    },
    {
      "epoch": 0.3531799729364005,
      "grad_norm": 0.365818053483963,
      "learning_rate": 4.858352707525681e-06,
      "loss": 0.665,
      "step": 1044
    },
    {
      "epoch": 0.35351826792963464,
      "grad_norm": 0.2922377288341522,
      "learning_rate": 4.858055322487957e-06,
      "loss": 0.6528,
      "step": 1045
    },
    {
      "epoch": 0.35385656292286877,
      "grad_norm": 0.2925521731376648,
      "learning_rate": 4.857757634722193e-06,
      "loss": 0.6496,
      "step": 1046
    },
    {
      "epoch": 0.35419485791610283,
      "grad_norm": 0.32205817103385925,
      "learning_rate": 4.857459644266606e-06,
      "loss": 0.6299,
      "step": 1047
    },
    {
      "epoch": 0.35453315290933696,
      "grad_norm": 0.32586464285850525,
      "learning_rate": 4.857161351159454e-06,
      "loss": 0.6439,
      "step": 1048
    },
    {
      "epoch": 0.354871447902571,
      "grad_norm": 0.29772520065307617,
      "learning_rate": 4.8568627554390304e-06,
      "loss": 0.6637,
      "step": 1049
    },
    {
      "epoch": 0.35520974289580515,
      "grad_norm": 0.38943392038345337,
      "learning_rate": 4.85656385714367e-06,
      "loss": 0.6257,
      "step": 1050
    },
    {
      "epoch": 0.3555480378890392,
      "grad_norm": 0.2854504883289337,
      "learning_rate": 4.856264656311745e-06,
      "loss": 0.618,
      "step": 1051
    },
    {
      "epoch": 0.35588633288227334,
      "grad_norm": 0.3191981315612793,
      "learning_rate": 4.855965152981667e-06,
      "loss": 0.6259,
      "step": 1052
    },
    {
      "epoch": 0.35622462787550746,
      "grad_norm": 0.326311320066452,
      "learning_rate": 4.855665347191885e-06,
      "loss": 0.6669,
      "step": 1053
    },
    {
      "epoch": 0.3565629228687415,
      "grad_norm": 0.2891475260257721,
      "learning_rate": 4.85536523898089e-06,
      "loss": 0.6516,
      "step": 1054
    },
    {
      "epoch": 0.35690121786197565,
      "grad_norm": 0.29754310846328735,
      "learning_rate": 4.85506482838721e-06,
      "loss": 0.6429,
      "step": 1055
    },
    {
      "epoch": 0.3572395128552097,
      "grad_norm": 0.3007654547691345,
      "learning_rate": 4.85476411544941e-06,
      "loss": 0.6733,
      "step": 1056
    },
    {
      "epoch": 0.35757780784844384,
      "grad_norm": 0.30116209387779236,
      "learning_rate": 4.8544631002060995e-06,
      "loss": 0.6343,
      "step": 1057
    },
    {
      "epoch": 0.35791610284167796,
      "grad_norm": 0.27309471368789673,
      "learning_rate": 4.8541617826959186e-06,
      "loss": 0.6562,
      "step": 1058
    },
    {
      "epoch": 0.35825439783491203,
      "grad_norm": 0.2854005694389343,
      "learning_rate": 4.8538601629575525e-06,
      "loss": 0.6266,
      "step": 1059
    },
    {
      "epoch": 0.35859269282814615,
      "grad_norm": 0.29010218381881714,
      "learning_rate": 4.853558241029723e-06,
      "loss": 0.6342,
      "step": 1060
    },
    {
      "epoch": 0.3589309878213802,
      "grad_norm": 0.28771087527275085,
      "learning_rate": 4.853256016951191e-06,
      "loss": 0.6483,
      "step": 1061
    },
    {
      "epoch": 0.35926928281461434,
      "grad_norm": 0.2850498855113983,
      "learning_rate": 4.852953490760757e-06,
      "loss": 0.6505,
      "step": 1062
    },
    {
      "epoch": 0.35960757780784847,
      "grad_norm": 0.2764224708080292,
      "learning_rate": 4.8526506624972576e-06,
      "loss": 0.6644,
      "step": 1063
    },
    {
      "epoch": 0.35994587280108253,
      "grad_norm": 0.27553361654281616,
      "learning_rate": 4.852347532199571e-06,
      "loss": 0.6562,
      "step": 1064
    },
    {
      "epoch": 0.36028416779431666,
      "grad_norm": 0.28768324851989746,
      "learning_rate": 4.8520440999066135e-06,
      "loss": 0.6411,
      "step": 1065
    },
    {
      "epoch": 0.3606224627875507,
      "grad_norm": 0.29022908210754395,
      "learning_rate": 4.85174036565734e-06,
      "loss": 0.6808,
      "step": 1066
    },
    {
      "epoch": 0.36096075778078485,
      "grad_norm": 0.28811946511268616,
      "learning_rate": 4.851436329490743e-06,
      "loss": 0.6348,
      "step": 1067
    },
    {
      "epoch": 0.36129905277401897,
      "grad_norm": 0.28230148553848267,
      "learning_rate": 4.851131991445856e-06,
      "loss": 0.6294,
      "step": 1068
    },
    {
      "epoch": 0.36163734776725304,
      "grad_norm": 0.2874899208545685,
      "learning_rate": 4.850827351561749e-06,
      "loss": 0.6418,
      "step": 1069
    },
    {
      "epoch": 0.36197564276048716,
      "grad_norm": 0.28801053762435913,
      "learning_rate": 4.850522409877532e-06,
      "loss": 0.6389,
      "step": 1070
    },
    {
      "epoch": 0.3623139377537212,
      "grad_norm": 0.28445324301719666,
      "learning_rate": 4.850217166432354e-06,
      "loss": 0.615,
      "step": 1071
    },
    {
      "epoch": 0.36265223274695535,
      "grad_norm": 0.29263418912887573,
      "learning_rate": 4.849911621265401e-06,
      "loss": 0.6432,
      "step": 1072
    },
    {
      "epoch": 0.3629905277401895,
      "grad_norm": 0.2812570035457611,
      "learning_rate": 4.849605774415901e-06,
      "loss": 0.6478,
      "step": 1073
    },
    {
      "epoch": 0.36332882273342354,
      "grad_norm": 0.27595818042755127,
      "learning_rate": 4.849299625923117e-06,
      "loss": 0.6271,
      "step": 1074
    },
    {
      "epoch": 0.36366711772665766,
      "grad_norm": 0.2792835533618927,
      "learning_rate": 4.848993175826354e-06,
      "loss": 0.6411,
      "step": 1075
    },
    {
      "epoch": 0.36400541271989173,
      "grad_norm": 0.2848093509674072,
      "learning_rate": 4.848686424164953e-06,
      "loss": 0.6452,
      "step": 1076
    },
    {
      "epoch": 0.36434370771312585,
      "grad_norm": 0.2892758846282959,
      "learning_rate": 4.8483793709782955e-06,
      "loss": 0.6333,
      "step": 1077
    },
    {
      "epoch": 0.3646820027063599,
      "grad_norm": 0.2910735607147217,
      "learning_rate": 4.8480720163058e-06,
      "loss": 0.6263,
      "step": 1078
    },
    {
      "epoch": 0.36502029769959404,
      "grad_norm": 0.2763359546661377,
      "learning_rate": 4.8477643601869265e-06,
      "loss": 0.6252,
      "step": 1079
    },
    {
      "epoch": 0.36535859269282817,
      "grad_norm": 0.28197360038757324,
      "learning_rate": 4.84745640266117e-06,
      "loss": 0.619,
      "step": 1080
    },
    {
      "epoch": 0.36569688768606223,
      "grad_norm": 0.29176396131515503,
      "learning_rate": 4.847148143768069e-06,
      "loss": 0.6784,
      "step": 1081
    },
    {
      "epoch": 0.36603518267929636,
      "grad_norm": 0.28936561942100525,
      "learning_rate": 4.846839583547195e-06,
      "loss": 0.627,
      "step": 1082
    },
    {
      "epoch": 0.3663734776725304,
      "grad_norm": 0.2793295979499817,
      "learning_rate": 4.846530722038163e-06,
      "loss": 0.6333,
      "step": 1083
    },
    {
      "epoch": 0.36671177266576455,
      "grad_norm": 0.295923113822937,
      "learning_rate": 4.846221559280624e-06,
      "loss": 0.6628,
      "step": 1084
    },
    {
      "epoch": 0.36705006765899867,
      "grad_norm": 0.28163689374923706,
      "learning_rate": 4.845912095314269e-06,
      "loss": 0.6323,
      "step": 1085
    },
    {
      "epoch": 0.36738836265223274,
      "grad_norm": 0.26855286955833435,
      "learning_rate": 4.845602330178826e-06,
      "loss": 0.6075,
      "step": 1086
    },
    {
      "epoch": 0.36772665764546686,
      "grad_norm": 0.3333419859409332,
      "learning_rate": 4.845292263914063e-06,
      "loss": 0.6323,
      "step": 1087
    },
    {
      "epoch": 0.3680649526387009,
      "grad_norm": 0.2788902223110199,
      "learning_rate": 4.844981896559787e-06,
      "loss": 0.6374,
      "step": 1088
    },
    {
      "epoch": 0.36840324763193505,
      "grad_norm": 0.3036459982395172,
      "learning_rate": 4.844671228155844e-06,
      "loss": 0.6248,
      "step": 1089
    },
    {
      "epoch": 0.36874154262516917,
      "grad_norm": 0.27933943271636963,
      "learning_rate": 4.844360258742115e-06,
      "loss": 0.6524,
      "step": 1090
    },
    {
      "epoch": 0.36907983761840324,
      "grad_norm": 0.2957841157913208,
      "learning_rate": 4.844048988358525e-06,
      "loss": 0.6441,
      "step": 1091
    },
    {
      "epoch": 0.36941813261163736,
      "grad_norm": 0.2745031714439392,
      "learning_rate": 4.843737417045035e-06,
      "loss": 0.6421,
      "step": 1092
    },
    {
      "epoch": 0.36975642760487143,
      "grad_norm": 0.2944257855415344,
      "learning_rate": 4.843425544841642e-06,
      "loss": 0.6331,
      "step": 1093
    },
    {
      "epoch": 0.37009472259810555,
      "grad_norm": 0.2939079999923706,
      "learning_rate": 4.843113371788386e-06,
      "loss": 0.6425,
      "step": 1094
    },
    {
      "epoch": 0.3704330175913397,
      "grad_norm": 0.2723322808742523,
      "learning_rate": 4.842800897925344e-06,
      "loss": 0.6283,
      "step": 1095
    },
    {
      "epoch": 0.37077131258457374,
      "grad_norm": 0.2891277074813843,
      "learning_rate": 4.842488123292632e-06,
      "loss": 0.634,
      "step": 1096
    },
    {
      "epoch": 0.37110960757780787,
      "grad_norm": 0.27961093187332153,
      "learning_rate": 4.842175047930403e-06,
      "loss": 0.6664,
      "step": 1097
    },
    {
      "epoch": 0.37144790257104193,
      "grad_norm": 0.2886982858181,
      "learning_rate": 4.841861671878851e-06,
      "loss": 0.6424,
      "step": 1098
    },
    {
      "epoch": 0.37178619756427606,
      "grad_norm": 0.28926536440849304,
      "learning_rate": 4.8415479951782054e-06,
      "loss": 0.642,
      "step": 1099
    },
    {
      "epoch": 0.3721244925575101,
      "grad_norm": 0.28860124945640564,
      "learning_rate": 4.841234017868738e-06,
      "loss": 0.6579,
      "step": 1100
    },
    {
      "epoch": 0.37246278755074425,
      "grad_norm": 0.2937227189540863,
      "learning_rate": 4.840919739990756e-06,
      "loss": 0.6384,
      "step": 1101
    },
    {
      "epoch": 0.37280108254397837,
      "grad_norm": 0.31778162717819214,
      "learning_rate": 4.840605161584607e-06,
      "loss": 0.6537,
      "step": 1102
    },
    {
      "epoch": 0.37313937753721244,
      "grad_norm": 0.2775851786136627,
      "learning_rate": 4.840290282690678e-06,
      "loss": 0.6401,
      "step": 1103
    },
    {
      "epoch": 0.37347767253044656,
      "grad_norm": 0.28459596633911133,
      "learning_rate": 4.839975103349391e-06,
      "loss": 0.6417,
      "step": 1104
    },
    {
      "epoch": 0.3738159675236806,
      "grad_norm": 0.30100637674331665,
      "learning_rate": 4.83965962360121e-06,
      "loss": 0.6175,
      "step": 1105
    },
    {
      "epoch": 0.37415426251691475,
      "grad_norm": 0.27300921082496643,
      "learning_rate": 4.839343843486636e-06,
      "loss": 0.6477,
      "step": 1106
    },
    {
      "epoch": 0.37449255751014887,
      "grad_norm": 0.2922280728816986,
      "learning_rate": 4.8390277630462105e-06,
      "loss": 0.6143,
      "step": 1107
    },
    {
      "epoch": 0.37483085250338294,
      "grad_norm": 0.2937355041503906,
      "learning_rate": 4.8387113823205104e-06,
      "loss": 0.6433,
      "step": 1108
    },
    {
      "epoch": 0.37516914749661706,
      "grad_norm": 0.27802449464797974,
      "learning_rate": 4.838394701350153e-06,
      "loss": 0.6202,
      "step": 1109
    },
    {
      "epoch": 0.37550744248985113,
      "grad_norm": 0.2949639856815338,
      "learning_rate": 4.838077720175793e-06,
      "loss": 0.6614,
      "step": 1110
    },
    {
      "epoch": 0.37584573748308525,
      "grad_norm": 0.3015292286872864,
      "learning_rate": 4.837760438838128e-06,
      "loss": 0.6589,
      "step": 1111
    },
    {
      "epoch": 0.3761840324763194,
      "grad_norm": 0.30569127202033997,
      "learning_rate": 4.837442857377886e-06,
      "loss": 0.652,
      "step": 1112
    },
    {
      "epoch": 0.37652232746955344,
      "grad_norm": 0.27839574217796326,
      "learning_rate": 4.837124975835842e-06,
      "loss": 0.6272,
      "step": 1113
    },
    {
      "epoch": 0.37686062246278756,
      "grad_norm": 0.29350098967552185,
      "learning_rate": 4.8368067942528045e-06,
      "loss": 0.6202,
      "step": 1114
    },
    {
      "epoch": 0.37719891745602163,
      "grad_norm": 0.2903738021850586,
      "learning_rate": 4.836488312669621e-06,
      "loss": 0.6172,
      "step": 1115
    },
    {
      "epoch": 0.37753721244925575,
      "grad_norm": 0.28676044940948486,
      "learning_rate": 4.8361695311271795e-06,
      "loss": 0.6215,
      "step": 1116
    },
    {
      "epoch": 0.3778755074424899,
      "grad_norm": 0.2821767032146454,
      "learning_rate": 4.835850449666405e-06,
      "loss": 0.6195,
      "step": 1117
    },
    {
      "epoch": 0.37821380243572394,
      "grad_norm": 0.29364868998527527,
      "learning_rate": 4.83553106832826e-06,
      "loss": 0.6345,
      "step": 1118
    },
    {
      "epoch": 0.37855209742895807,
      "grad_norm": 0.2790587246417999,
      "learning_rate": 4.835211387153749e-06,
      "loss": 0.659,
      "step": 1119
    },
    {
      "epoch": 0.37889039242219213,
      "grad_norm": 0.2861728072166443,
      "learning_rate": 4.83489140618391e-06,
      "loss": 0.6375,
      "step": 1120
    },
    {
      "epoch": 0.37922868741542626,
      "grad_norm": 0.2797929048538208,
      "learning_rate": 4.834571125459826e-06,
      "loss": 0.6465,
      "step": 1121
    },
    {
      "epoch": 0.3795669824086603,
      "grad_norm": 0.30132776498794556,
      "learning_rate": 4.834250545022611e-06,
      "loss": 0.6353,
      "step": 1122
    },
    {
      "epoch": 0.37990527740189445,
      "grad_norm": 0.2855285704135895,
      "learning_rate": 4.833929664913423e-06,
      "loss": 0.6558,
      "step": 1123
    },
    {
      "epoch": 0.38024357239512857,
      "grad_norm": 0.2976146638393402,
      "learning_rate": 4.833608485173458e-06,
      "loss": 0.6267,
      "step": 1124
    },
    {
      "epoch": 0.38058186738836264,
      "grad_norm": 0.2943631410598755,
      "learning_rate": 4.8332870058439465e-06,
      "loss": 0.6269,
      "step": 1125
    },
    {
      "epoch": 0.38092016238159676,
      "grad_norm": 0.2879925072193146,
      "learning_rate": 4.832965226966161e-06,
      "loss": 0.6487,
      "step": 1126
    },
    {
      "epoch": 0.38125845737483083,
      "grad_norm": 0.2811349034309387,
      "learning_rate": 4.832643148581413e-06,
      "loss": 0.6268,
      "step": 1127
    },
    {
      "epoch": 0.38159675236806495,
      "grad_norm": 0.30948612093925476,
      "learning_rate": 4.83232077073105e-06,
      "loss": 0.638,
      "step": 1128
    },
    {
      "epoch": 0.3819350473612991,
      "grad_norm": 0.3030785620212555,
      "learning_rate": 4.8319980934564595e-06,
      "loss": 0.6621,
      "step": 1129
    },
    {
      "epoch": 0.38227334235453314,
      "grad_norm": 0.29253870248794556,
      "learning_rate": 4.831675116799065e-06,
      "loss": 0.623,
      "step": 1130
    },
    {
      "epoch": 0.38261163734776726,
      "grad_norm": 0.28105953335762024,
      "learning_rate": 4.831351840800333e-06,
      "loss": 0.6375,
      "step": 1131
    },
    {
      "epoch": 0.38294993234100133,
      "grad_norm": 0.307233989238739,
      "learning_rate": 4.831028265501764e-06,
      "loss": 0.6732,
      "step": 1132
    },
    {
      "epoch": 0.38328822733423545,
      "grad_norm": 0.2878662943840027,
      "learning_rate": 4.8307043909449e-06,
      "loss": 0.6332,
      "step": 1133
    },
    {
      "epoch": 0.3836265223274696,
      "grad_norm": 0.30148088932037354,
      "learning_rate": 4.830380217171319e-06,
      "loss": 0.6477,
      "step": 1134
    },
    {
      "epoch": 0.38396481732070364,
      "grad_norm": 0.3052041232585907,
      "learning_rate": 4.830055744222639e-06,
      "loss": 0.6243,
      "step": 1135
    },
    {
      "epoch": 0.38430311231393777,
      "grad_norm": 0.2861061990261078,
      "learning_rate": 4.8297309721405175e-06,
      "loss": 0.6257,
      "step": 1136
    },
    {
      "epoch": 0.38464140730717183,
      "grad_norm": 0.2866950035095215,
      "learning_rate": 4.8294059009666455e-06,
      "loss": 0.623,
      "step": 1137
    },
    {
      "epoch": 0.38497970230040596,
      "grad_norm": 0.28932666778564453,
      "learning_rate": 4.8290805307427585e-06,
      "loss": 0.6312,
      "step": 1138
    },
    {
      "epoch": 0.3853179972936401,
      "grad_norm": 0.2892247438430786,
      "learning_rate": 4.828754861510627e-06,
      "loss": 0.6162,
      "step": 1139
    },
    {
      "epoch": 0.38565629228687415,
      "grad_norm": 0.30072498321533203,
      "learning_rate": 4.8284288933120595e-06,
      "loss": 0.6235,
      "step": 1140
    },
    {
      "epoch": 0.38599458728010827,
      "grad_norm": 0.2970368266105652,
      "learning_rate": 4.828102626188905e-06,
      "loss": 0.6561,
      "step": 1141
    },
    {
      "epoch": 0.38633288227334234,
      "grad_norm": 0.2955375611782074,
      "learning_rate": 4.82777606018305e-06,
      "loss": 0.6355,
      "step": 1142
    },
    {
      "epoch": 0.38667117726657646,
      "grad_norm": 0.27442505955696106,
      "learning_rate": 4.827449195336419e-06,
      "loss": 0.6217,
      "step": 1143
    },
    {
      "epoch": 0.3870094722598105,
      "grad_norm": 0.28408247232437134,
      "learning_rate": 4.827122031690974e-06,
      "loss": 0.6251,
      "step": 1144
    },
    {
      "epoch": 0.38734776725304465,
      "grad_norm": 0.30623340606689453,
      "learning_rate": 4.826794569288717e-06,
      "loss": 0.6523,
      "step": 1145
    },
    {
      "epoch": 0.3876860622462788,
      "grad_norm": 0.2974097430706024,
      "learning_rate": 4.826466808171689e-06,
      "loss": 0.6364,
      "step": 1146
    },
    {
      "epoch": 0.38802435723951284,
      "grad_norm": 0.2743532061576843,
      "learning_rate": 4.826138748381966e-06,
      "loss": 0.6601,
      "step": 1147
    },
    {
      "epoch": 0.38836265223274696,
      "grad_norm": 0.2808363437652588,
      "learning_rate": 4.825810389961666e-06,
      "loss": 0.6576,
      "step": 1148
    },
    {
      "epoch": 0.38870094722598103,
      "grad_norm": 0.29762133955955505,
      "learning_rate": 4.825481732952943e-06,
      "loss": 0.6562,
      "step": 1149
    },
    {
      "epoch": 0.38903924221921515,
      "grad_norm": 0.2852347791194916,
      "learning_rate": 4.82515277739799e-06,
      "loss": 0.6076,
      "step": 1150
    },
    {
      "epoch": 0.3893775372124493,
      "grad_norm": 0.2761669158935547,
      "learning_rate": 4.824823523339039e-06,
      "loss": 0.6341,
      "step": 1151
    },
    {
      "epoch": 0.38971583220568334,
      "grad_norm": 0.3076622784137726,
      "learning_rate": 4.824493970818359e-06,
      "loss": 0.6332,
      "step": 1152
    },
    {
      "epoch": 0.39005412719891747,
      "grad_norm": 0.28125861287117004,
      "learning_rate": 4.82416411987826e-06,
      "loss": 0.6297,
      "step": 1153
    },
    {
      "epoch": 0.39039242219215153,
      "grad_norm": 0.2835453748703003,
      "learning_rate": 4.823833970561086e-06,
      "loss": 0.6394,
      "step": 1154
    },
    {
      "epoch": 0.39073071718538566,
      "grad_norm": 0.28459158539772034,
      "learning_rate": 4.823503522909223e-06,
      "loss": 0.618,
      "step": 1155
    },
    {
      "epoch": 0.3910690121786198,
      "grad_norm": 0.27935102581977844,
      "learning_rate": 4.8231727769650934e-06,
      "loss": 0.6447,
      "step": 1156
    },
    {
      "epoch": 0.39140730717185385,
      "grad_norm": 0.2861892580986023,
      "learning_rate": 4.822841732771159e-06,
      "loss": 0.6379,
      "step": 1157
    },
    {
      "epoch": 0.39174560216508797,
      "grad_norm": 0.290678471326828,
      "learning_rate": 4.8225103903699185e-06,
      "loss": 0.6247,
      "step": 1158
    },
    {
      "epoch": 0.39208389715832204,
      "grad_norm": 0.2962476313114166,
      "learning_rate": 4.822178749803912e-06,
      "loss": 0.6452,
      "step": 1159
    },
    {
      "epoch": 0.39242219215155616,
      "grad_norm": 0.283066987991333,
      "learning_rate": 4.821846811115713e-06,
      "loss": 0.6398,
      "step": 1160
    },
    {
      "epoch": 0.3927604871447903,
      "grad_norm": 0.28135359287261963,
      "learning_rate": 4.821514574347937e-06,
      "loss": 0.618,
      "step": 1161
    },
    {
      "epoch": 0.39309878213802435,
      "grad_norm": 0.29703113436698914,
      "learning_rate": 4.821182039543236e-06,
      "loss": 0.6371,
      "step": 1162
    },
    {
      "epoch": 0.3934370771312585,
      "grad_norm": 0.33292141556739807,
      "learning_rate": 4.820849206744303e-06,
      "loss": 0.6329,
      "step": 1163
    },
    {
      "epoch": 0.39377537212449254,
      "grad_norm": 0.29253455996513367,
      "learning_rate": 4.820516075993865e-06,
      "loss": 0.65,
      "step": 1164
    },
    {
      "epoch": 0.39411366711772666,
      "grad_norm": 0.3138137757778168,
      "learning_rate": 4.82018264733469e-06,
      "loss": 0.6401,
      "step": 1165
    },
    {
      "epoch": 0.3944519621109608,
      "grad_norm": 0.3234380781650543,
      "learning_rate": 4.819848920809584e-06,
      "loss": 0.6454,
      "step": 1166
    },
    {
      "epoch": 0.39479025710419485,
      "grad_norm": 0.3004589080810547,
      "learning_rate": 4.8195148964613916e-06,
      "loss": 0.6386,
      "step": 1167
    },
    {
      "epoch": 0.395128552097429,
      "grad_norm": 0.2853522300720215,
      "learning_rate": 4.8191805743329945e-06,
      "loss": 0.6326,
      "step": 1168
    },
    {
      "epoch": 0.39546684709066304,
      "grad_norm": 0.3297440707683563,
      "learning_rate": 4.8188459544673114e-06,
      "loss": 0.6675,
      "step": 1169
    },
    {
      "epoch": 0.39580514208389717,
      "grad_norm": 0.3187149167060852,
      "learning_rate": 4.818511036907304e-06,
      "loss": 0.6379,
      "step": 1170
    },
    {
      "epoch": 0.39614343707713123,
      "grad_norm": 0.28473100066185,
      "learning_rate": 4.8181758216959665e-06,
      "loss": 0.6199,
      "step": 1171
    },
    {
      "epoch": 0.39648173207036536,
      "grad_norm": 0.29906320571899414,
      "learning_rate": 4.817840308876335e-06,
      "loss": 0.6314,
      "step": 1172
    },
    {
      "epoch": 0.3968200270635995,
      "grad_norm": 0.3337453603744507,
      "learning_rate": 4.817504498491483e-06,
      "loss": 0.6325,
      "step": 1173
    },
    {
      "epoch": 0.39715832205683355,
      "grad_norm": 0.29629188776016235,
      "learning_rate": 4.817168390584522e-06,
      "loss": 0.6378,
      "step": 1174
    },
    {
      "epoch": 0.39749661705006767,
      "grad_norm": 0.28483545780181885,
      "learning_rate": 4.816831985198601e-06,
      "loss": 0.6383,
      "step": 1175
    },
    {
      "epoch": 0.39783491204330174,
      "grad_norm": 0.28717607259750366,
      "learning_rate": 4.8164952823769085e-06,
      "loss": 0.6127,
      "step": 1176
    },
    {
      "epoch": 0.39817320703653586,
      "grad_norm": 0.30153074860572815,
      "learning_rate": 4.816158282162671e-06,
      "loss": 0.6759,
      "step": 1177
    },
    {
      "epoch": 0.39851150202977,
      "grad_norm": 0.3128243088722229,
      "learning_rate": 4.815820984599151e-06,
      "loss": 0.6336,
      "step": 1178
    },
    {
      "epoch": 0.39884979702300405,
      "grad_norm": 0.29970309138298035,
      "learning_rate": 4.815483389729652e-06,
      "loss": 0.6306,
      "step": 1179
    },
    {
      "epoch": 0.39918809201623817,
      "grad_norm": 0.2995675802230835,
      "learning_rate": 4.815145497597514e-06,
      "loss": 0.657,
      "step": 1180
    },
    {
      "epoch": 0.39952638700947224,
      "grad_norm": 0.33796557784080505,
      "learning_rate": 4.814807308246116e-06,
      "loss": 0.6151,
      "step": 1181
    },
    {
      "epoch": 0.39986468200270636,
      "grad_norm": 0.2935757637023926,
      "learning_rate": 4.8144688217188765e-06,
      "loss": 0.6291,
      "step": 1182
    },
    {
      "epoch": 0.4002029769959405,
      "grad_norm": 0.2939322590827942,
      "learning_rate": 4.814130038059248e-06,
      "loss": 0.6379,
      "step": 1183
    },
    {
      "epoch": 0.40054127198917455,
      "grad_norm": 0.32517513632774353,
      "learning_rate": 4.813790957310725e-06,
      "loss": 0.6222,
      "step": 1184
    },
    {
      "epoch": 0.4008795669824087,
      "grad_norm": 0.28321826457977295,
      "learning_rate": 4.813451579516839e-06,
      "loss": 0.6199,
      "step": 1185
    },
    {
      "epoch": 0.40121786197564274,
      "grad_norm": 0.287484735250473,
      "learning_rate": 4.8131119047211585e-06,
      "loss": 0.6533,
      "step": 1186
    },
    {
      "epoch": 0.40155615696887687,
      "grad_norm": 0.30498993396759033,
      "learning_rate": 4.8127719329672914e-06,
      "loss": 0.6154,
      "step": 1187
    },
    {
      "epoch": 0.401894451962111,
      "grad_norm": 0.28674793243408203,
      "learning_rate": 4.812431664298884e-06,
      "loss": 0.6355,
      "step": 1188
    },
    {
      "epoch": 0.40223274695534506,
      "grad_norm": 0.294114887714386,
      "learning_rate": 4.812091098759619e-06,
      "loss": 0.6637,
      "step": 1189
    },
    {
      "epoch": 0.4025710419485792,
      "grad_norm": 0.2917622923851013,
      "learning_rate": 4.8117502363932195e-06,
      "loss": 0.6409,
      "step": 1190
    },
    {
      "epoch": 0.40290933694181325,
      "grad_norm": 0.2833552658557892,
      "learning_rate": 4.811409077243445e-06,
      "loss": 0.6262,
      "step": 1191
    },
    {
      "epoch": 0.40324763193504737,
      "grad_norm": 0.28660398721694946,
      "learning_rate": 4.811067621354094e-06,
      "loss": 0.6387,
      "step": 1192
    },
    {
      "epoch": 0.40358592692828144,
      "grad_norm": 0.28230899572372437,
      "learning_rate": 4.8107258687690015e-06,
      "loss": 0.6346,
      "step": 1193
    },
    {
      "epoch": 0.40392422192151556,
      "grad_norm": 0.2882273495197296,
      "learning_rate": 4.810383819532044e-06,
      "loss": 0.6639,
      "step": 1194
    },
    {
      "epoch": 0.4042625169147497,
      "grad_norm": 0.3016176223754883,
      "learning_rate": 4.810041473687132e-06,
      "loss": 0.6336,
      "step": 1195
    },
    {
      "epoch": 0.40460081190798375,
      "grad_norm": 0.2878057360649109,
      "learning_rate": 4.809698831278217e-06,
      "loss": 0.6015,
      "step": 1196
    },
    {
      "epoch": 0.40493910690121787,
      "grad_norm": 0.29583847522735596,
      "learning_rate": 4.809355892349288e-06,
      "loss": 0.6413,
      "step": 1197
    },
    {
      "epoch": 0.40527740189445194,
      "grad_norm": 0.29840466380119324,
      "learning_rate": 4.80901265694437e-06,
      "loss": 0.6442,
      "step": 1198
    },
    {
      "epoch": 0.40561569688768606,
      "grad_norm": 0.2863212525844574,
      "learning_rate": 4.808669125107528e-06,
      "loss": 0.6521,
      "step": 1199
    },
    {
      "epoch": 0.4059539918809202,
      "grad_norm": 0.29659488797187805,
      "learning_rate": 4.808325296882866e-06,
      "loss": 0.6321,
      "step": 1200
    },
    {
      "epoch": 0.40629228687415425,
      "grad_norm": 0.29904645681381226,
      "learning_rate": 4.807981172314524e-06,
      "loss": 0.6282,
      "step": 1201
    },
    {
      "epoch": 0.4066305818673884,
      "grad_norm": 0.29009896516799927,
      "learning_rate": 4.8076367514466815e-06,
      "loss": 0.6455,
      "step": 1202
    },
    {
      "epoch": 0.40696887686062244,
      "grad_norm": 0.2847822308540344,
      "learning_rate": 4.807292034323554e-06,
      "loss": 0.6386,
      "step": 1203
    },
    {
      "epoch": 0.40730717185385656,
      "grad_norm": 0.316760390996933,
      "learning_rate": 4.8069470209893976e-06,
      "loss": 0.6163,
      "step": 1204
    },
    {
      "epoch": 0.4076454668470907,
      "grad_norm": 0.28848010301589966,
      "learning_rate": 4.806601711488505e-06,
      "loss": 0.6422,
      "step": 1205
    },
    {
      "epoch": 0.40798376184032475,
      "grad_norm": 0.28792351484298706,
      "learning_rate": 4.806256105865207e-06,
      "loss": 0.645,
      "step": 1206
    },
    {
      "epoch": 0.4083220568335589,
      "grad_norm": 0.2956791818141937,
      "learning_rate": 4.805910204163873e-06,
      "loss": 0.6307,
      "step": 1207
    },
    {
      "epoch": 0.40866035182679294,
      "grad_norm": 0.27958643436431885,
      "learning_rate": 4.805564006428909e-06,
      "loss": 0.6482,
      "step": 1208
    },
    {
      "epoch": 0.40899864682002707,
      "grad_norm": 0.29864054918289185,
      "learning_rate": 4.80521751270476e-06,
      "loss": 0.6393,
      "step": 1209
    },
    {
      "epoch": 0.4093369418132612,
      "grad_norm": 0.28602924942970276,
      "learning_rate": 4.804870723035912e-06,
      "loss": 0.6451,
      "step": 1210
    },
    {
      "epoch": 0.40967523680649526,
      "grad_norm": 0.2817135155200958,
      "learning_rate": 4.804523637466882e-06,
      "loss": 0.6507,
      "step": 1211
    },
    {
      "epoch": 0.4100135317997294,
      "grad_norm": 0.2887398302555084,
      "learning_rate": 4.80417625604223e-06,
      "loss": 0.6357,
      "step": 1212
    },
    {
      "epoch": 0.41035182679296345,
      "grad_norm": 0.29490935802459717,
      "learning_rate": 4.803828578806555e-06,
      "loss": 0.6125,
      "step": 1213
    },
    {
      "epoch": 0.41069012178619757,
      "grad_norm": 0.28504541516304016,
      "learning_rate": 4.8034806058044895e-06,
      "loss": 0.6331,
      "step": 1214
    },
    {
      "epoch": 0.41102841677943164,
      "grad_norm": 0.3002273738384247,
      "learning_rate": 4.803132337080708e-06,
      "loss": 0.6351,
      "step": 1215
    },
    {
      "epoch": 0.41136671177266576,
      "grad_norm": 0.3103099465370178,
      "learning_rate": 4.80278377267992e-06,
      "loss": 0.6436,
      "step": 1216
    },
    {
      "epoch": 0.4117050067658999,
      "grad_norm": 0.291058212518692,
      "learning_rate": 4.802434912646876e-06,
      "loss": 0.6405,
      "step": 1217
    },
    {
      "epoch": 0.41204330175913395,
      "grad_norm": 0.28568235039711,
      "learning_rate": 4.802085757026361e-06,
      "loss": 0.6553,
      "step": 1218
    },
    {
      "epoch": 0.4123815967523681,
      "grad_norm": 0.303610622882843,
      "learning_rate": 4.801736305863202e-06,
      "loss": 0.6947,
      "step": 1219
    },
    {
      "epoch": 0.41271989174560214,
      "grad_norm": 0.2856765687465668,
      "learning_rate": 4.801386559202259e-06,
      "loss": 0.6226,
      "step": 1220
    },
    {
      "epoch": 0.41305818673883626,
      "grad_norm": 0.30076685547828674,
      "learning_rate": 4.801036517088434e-06,
      "loss": 0.6668,
      "step": 1221
    },
    {
      "epoch": 0.4133964817320704,
      "grad_norm": 0.2825355529785156,
      "learning_rate": 4.800686179566666e-06,
      "loss": 0.6337,
      "step": 1222
    },
    {
      "epoch": 0.41373477672530445,
      "grad_norm": 0.29440179467201233,
      "learning_rate": 4.8003355466819306e-06,
      "loss": 0.6355,
      "step": 1223
    },
    {
      "epoch": 0.4140730717185386,
      "grad_norm": 0.28212273120880127,
      "learning_rate": 4.7999846184792424e-06,
      "loss": 0.6378,
      "step": 1224
    },
    {
      "epoch": 0.41441136671177264,
      "grad_norm": 0.3060968518257141,
      "learning_rate": 4.799633395003655e-06,
      "loss": 0.637,
      "step": 1225
    },
    {
      "epoch": 0.41474966170500677,
      "grad_norm": 0.2914919853210449,
      "learning_rate": 4.799281876300256e-06,
      "loss": 0.6605,
      "step": 1226
    },
    {
      "epoch": 0.4150879566982409,
      "grad_norm": 0.29111871123313904,
      "learning_rate": 4.7989300624141745e-06,
      "loss": 0.6086,
      "step": 1227
    },
    {
      "epoch": 0.41542625169147496,
      "grad_norm": 0.2990707457065582,
      "learning_rate": 4.798577953390577e-06,
      "loss": 0.6452,
      "step": 1228
    },
    {
      "epoch": 0.4157645466847091,
      "grad_norm": 0.29072779417037964,
      "learning_rate": 4.7982255492746685e-06,
      "loss": 0.6408,
      "step": 1229
    },
    {
      "epoch": 0.41610284167794315,
      "grad_norm": 0.28976115584373474,
      "learning_rate": 4.797872850111689e-06,
      "loss": 0.6294,
      "step": 1230
    },
    {
      "epoch": 0.41644113667117727,
      "grad_norm": 0.291080504655838,
      "learning_rate": 4.797519855946918e-06,
      "loss": 0.6231,
      "step": 1231
    },
    {
      "epoch": 0.4167794316644114,
      "grad_norm": 0.2862843871116638,
      "learning_rate": 4.797166566825675e-06,
      "loss": 0.626,
      "step": 1232
    },
    {
      "epoch": 0.41711772665764546,
      "grad_norm": 0.28314611315727234,
      "learning_rate": 4.7968129827933135e-06,
      "loss": 0.6495,
      "step": 1233
    },
    {
      "epoch": 0.4174560216508796,
      "grad_norm": 0.29271024465560913,
      "learning_rate": 4.7964591038952276e-06,
      "loss": 0.6341,
      "step": 1234
    },
    {
      "epoch": 0.41779431664411365,
      "grad_norm": 0.2987518310546875,
      "learning_rate": 4.796104930176848e-06,
      "loss": 0.6417,
      "step": 1235
    },
    {
      "epoch": 0.4181326116373478,
      "grad_norm": 0.29295241832733154,
      "learning_rate": 4.795750461683643e-06,
      "loss": 0.6201,
      "step": 1236
    },
    {
      "epoch": 0.41847090663058184,
      "grad_norm": 0.29133695363998413,
      "learning_rate": 4.7953956984611215e-06,
      "loss": 0.6094,
      "step": 1237
    },
    {
      "epoch": 0.41880920162381596,
      "grad_norm": 0.2857004702091217,
      "learning_rate": 4.795040640554827e-06,
      "loss": 0.6453,
      "step": 1238
    },
    {
      "epoch": 0.4191474966170501,
      "grad_norm": 0.2874625623226166,
      "learning_rate": 4.7946852880103414e-06,
      "loss": 0.6183,
      "step": 1239
    },
    {
      "epoch": 0.41948579161028415,
      "grad_norm": 0.29205045104026794,
      "learning_rate": 4.794329640873285e-06,
      "loss": 0.6329,
      "step": 1240
    },
    {
      "epoch": 0.4198240866035183,
      "grad_norm": 0.2948002219200134,
      "learning_rate": 4.793973699189317e-06,
      "loss": 0.6295,
      "step": 1241
    },
    {
      "epoch": 0.42016238159675234,
      "grad_norm": 0.2800194025039673,
      "learning_rate": 4.793617463004132e-06,
      "loss": 0.6322,
      "step": 1242
    },
    {
      "epoch": 0.42050067658998647,
      "grad_norm": 0.27651315927505493,
      "learning_rate": 4.793260932363465e-06,
      "loss": 0.6343,
      "step": 1243
    },
    {
      "epoch": 0.4208389715832206,
      "grad_norm": 0.2835398316383362,
      "learning_rate": 4.792904107313086e-06,
      "loss": 0.6401,
      "step": 1244
    },
    {
      "epoch": 0.42117726657645466,
      "grad_norm": 0.28559476137161255,
      "learning_rate": 4.792546987898807e-06,
      "loss": 0.625,
      "step": 1245
    },
    {
      "epoch": 0.4215155615696888,
      "grad_norm": 0.27733302116394043,
      "learning_rate": 4.792189574166473e-06,
      "loss": 0.6432,
      "step": 1246
    },
    {
      "epoch": 0.42185385656292285,
      "grad_norm": 0.29606395959854126,
      "learning_rate": 4.791831866161968e-06,
      "loss": 0.643,
      "step": 1247
    },
    {
      "epoch": 0.42219215155615697,
      "grad_norm": 0.30032095313072205,
      "learning_rate": 4.7914738639312165e-06,
      "loss": 0.612,
      "step": 1248
    },
    {
      "epoch": 0.4225304465493911,
      "grad_norm": 0.3007729649543762,
      "learning_rate": 4.791115567520179e-06,
      "loss": 0.6394,
      "step": 1249
    },
    {
      "epoch": 0.42286874154262516,
      "grad_norm": 0.290590763092041,
      "learning_rate": 4.790756976974853e-06,
      "loss": 0.6427,
      "step": 1250
    },
    {
      "epoch": 0.4232070365358593,
      "grad_norm": 0.2848972976207733,
      "learning_rate": 4.790398092341275e-06,
      "loss": 0.6471,
      "step": 1251
    },
    {
      "epoch": 0.42354533152909335,
      "grad_norm": 0.29602229595184326,
      "learning_rate": 4.790038913665519e-06,
      "loss": 0.613,
      "step": 1252
    },
    {
      "epoch": 0.4238836265223275,
      "grad_norm": 0.2961762547492981,
      "learning_rate": 4.789679440993695e-06,
      "loss": 0.6486,
      "step": 1253
    },
    {
      "epoch": 0.4242219215155616,
      "grad_norm": 0.2972279489040375,
      "learning_rate": 4.789319674371953e-06,
      "loss": 0.6368,
      "step": 1254
    },
    {
      "epoch": 0.42456021650879566,
      "grad_norm": 0.28891313076019287,
      "learning_rate": 4.788959613846481e-06,
      "loss": 0.6802,
      "step": 1255
    },
    {
      "epoch": 0.4248985115020298,
      "grad_norm": 0.28022804856300354,
      "learning_rate": 4.7885992594635025e-06,
      "loss": 0.627,
      "step": 1256
    },
    {
      "epoch": 0.42523680649526385,
      "grad_norm": 0.2823637127876282,
      "learning_rate": 4.78823861126928e-06,
      "loss": 0.6288,
      "step": 1257
    },
    {
      "epoch": 0.425575101488498,
      "grad_norm": 0.29723280668258667,
      "learning_rate": 4.7878776693101146e-06,
      "loss": 0.6202,
      "step": 1258
    },
    {
      "epoch": 0.42591339648173204,
      "grad_norm": 0.2898513376712799,
      "learning_rate": 4.787516433632342e-06,
      "loss": 0.6265,
      "step": 1259
    },
    {
      "epoch": 0.42625169147496617,
      "grad_norm": 0.31341275572776794,
      "learning_rate": 4.787154904282341e-06,
      "loss": 0.6241,
      "step": 1260
    },
    {
      "epoch": 0.4265899864682003,
      "grad_norm": 0.28693923354148865,
      "learning_rate": 4.786793081306523e-06,
      "loss": 0.6273,
      "step": 1261
    },
    {
      "epoch": 0.42692828146143436,
      "grad_norm": 0.305917888879776,
      "learning_rate": 4.786430964751338e-06,
      "loss": 0.6224,
      "step": 1262
    },
    {
      "epoch": 0.4272665764546685,
      "grad_norm": 0.2971956133842468,
      "learning_rate": 4.7860685546632775e-06,
      "loss": 0.6333,
      "step": 1263
    },
    {
      "epoch": 0.42760487144790255,
      "grad_norm": 0.3205307424068451,
      "learning_rate": 4.785705851088865e-06,
      "loss": 0.6435,
      "step": 1264
    },
    {
      "epoch": 0.42794316644113667,
      "grad_norm": 0.3163212835788727,
      "learning_rate": 4.785342854074666e-06,
      "loss": 0.6457,
      "step": 1265
    },
    {
      "epoch": 0.4282814614343708,
      "grad_norm": 0.28766465187072754,
      "learning_rate": 4.784979563667281e-06,
      "loss": 0.6522,
      "step": 1266
    },
    {
      "epoch": 0.42861975642760486,
      "grad_norm": 0.298446387052536,
      "learning_rate": 4.784615979913352e-06,
      "loss": 0.6412,
      "step": 1267
    },
    {
      "epoch": 0.428958051420839,
      "grad_norm": 0.30019763112068176,
      "learning_rate": 4.784252102859553e-06,
      "loss": 0.6241,
      "step": 1268
    },
    {
      "epoch": 0.42929634641407305,
      "grad_norm": 0.2901851534843445,
      "learning_rate": 4.7838879325526e-06,
      "loss": 0.6138,
      "step": 1269
    },
    {
      "epoch": 0.42963464140730717,
      "grad_norm": 0.2932225167751312,
      "learning_rate": 4.783523469039245e-06,
      "loss": 0.5911,
      "step": 1270
    },
    {
      "epoch": 0.4299729364005413,
      "grad_norm": 0.29000577330589294,
      "learning_rate": 4.7831587123662794e-06,
      "loss": 0.6138,
      "step": 1271
    },
    {
      "epoch": 0.43031123139377536,
      "grad_norm": 0.2969661355018616,
      "learning_rate": 4.782793662580528e-06,
      "loss": 0.6491,
      "step": 1272
    },
    {
      "epoch": 0.4306495263870095,
      "grad_norm": 0.2832574248313904,
      "learning_rate": 4.782428319728859e-06,
      "loss": 0.663,
      "step": 1273
    },
    {
      "epoch": 0.43098782138024355,
      "grad_norm": 0.28881847858428955,
      "learning_rate": 4.782062683858174e-06,
      "loss": 0.6298,
      "step": 1274
    },
    {
      "epoch": 0.4313261163734777,
      "grad_norm": 0.301044225692749,
      "learning_rate": 4.781696755015412e-06,
      "loss": 0.6531,
      "step": 1275
    },
    {
      "epoch": 0.4316644113667118,
      "grad_norm": 0.29331329464912415,
      "learning_rate": 4.781330533247554e-06,
      "loss": 0.652,
      "step": 1276
    },
    {
      "epoch": 0.43200270635994586,
      "grad_norm": 0.2893969714641571,
      "learning_rate": 4.780964018601613e-06,
      "loss": 0.6428,
      "step": 1277
    },
    {
      "epoch": 0.43234100135318,
      "grad_norm": 0.2999071180820465,
      "learning_rate": 4.780597211124643e-06,
      "loss": 0.6286,
      "step": 1278
    },
    {
      "epoch": 0.43267929634641406,
      "grad_norm": 0.29314321279525757,
      "learning_rate": 4.780230110863737e-06,
      "loss": 0.6701,
      "step": 1279
    },
    {
      "epoch": 0.4330175913396482,
      "grad_norm": 0.28190457820892334,
      "learning_rate": 4.779862717866021e-06,
      "loss": 0.6192,
      "step": 1280
    },
    {
      "epoch": 0.4333558863328823,
      "grad_norm": 0.2893087863922119,
      "learning_rate": 4.779495032178662e-06,
      "loss": 0.6405,
      "step": 1281
    },
    {
      "epoch": 0.43369418132611637,
      "grad_norm": 0.2829298973083496,
      "learning_rate": 4.779127053848863e-06,
      "loss": 0.6296,
      "step": 1282
    },
    {
      "epoch": 0.4340324763193505,
      "grad_norm": 0.2973841726779938,
      "learning_rate": 4.778758782923866e-06,
      "loss": 0.6828,
      "step": 1283
    },
    {
      "epoch": 0.43437077131258456,
      "grad_norm": 0.298194020986557,
      "learning_rate": 4.778390219450949e-06,
      "loss": 0.648,
      "step": 1284
    },
    {
      "epoch": 0.4347090663058187,
      "grad_norm": 0.2952548861503601,
      "learning_rate": 4.778021363477429e-06,
      "loss": 0.6675,
      "step": 1285
    },
    {
      "epoch": 0.43504736129905275,
      "grad_norm": 0.28702524304389954,
      "learning_rate": 4.777652215050659e-06,
      "loss": 0.6434,
      "step": 1286
    },
    {
      "epoch": 0.43538565629228687,
      "grad_norm": 0.28044795989990234,
      "learning_rate": 4.777282774218033e-06,
      "loss": 0.6356,
      "step": 1287
    },
    {
      "epoch": 0.435723951285521,
      "grad_norm": 0.2795807123184204,
      "learning_rate": 4.776913041026976e-06,
      "loss": 0.6434,
      "step": 1288
    },
    {
      "epoch": 0.43606224627875506,
      "grad_norm": 0.2964823544025421,
      "learning_rate": 4.776543015524957e-06,
      "loss": 0.6511,
      "step": 1289
    },
    {
      "epoch": 0.4364005412719892,
      "grad_norm": 0.29234057664871216,
      "learning_rate": 4.77617269775948e-06,
      "loss": 0.6316,
      "step": 1290
    },
    {
      "epoch": 0.43673883626522325,
      "grad_norm": 0.2864561080932617,
      "learning_rate": 4.775802087778085e-06,
      "loss": 0.6307,
      "step": 1291
    },
    {
      "epoch": 0.4370771312584574,
      "grad_norm": 0.2859666049480438,
      "learning_rate": 4.775431185628353e-06,
      "loss": 0.607,
      "step": 1292
    },
    {
      "epoch": 0.4374154262516915,
      "grad_norm": 0.2826318144798279,
      "learning_rate": 4.775059991357899e-06,
      "loss": 0.6114,
      "step": 1293
    },
    {
      "epoch": 0.43775372124492556,
      "grad_norm": 0.2815684676170349,
      "learning_rate": 4.774688505014379e-06,
      "loss": 0.6453,
      "step": 1294
    },
    {
      "epoch": 0.4380920162381597,
      "grad_norm": 0.3040962219238281,
      "learning_rate": 4.7743167266454815e-06,
      "loss": 0.6587,
      "step": 1295
    },
    {
      "epoch": 0.43843031123139375,
      "grad_norm": 0.2755117416381836,
      "learning_rate": 4.773944656298939e-06,
      "loss": 0.6284,
      "step": 1296
    },
    {
      "epoch": 0.4387686062246279,
      "grad_norm": 0.31091102957725525,
      "learning_rate": 4.773572294022515e-06,
      "loss": 0.6014,
      "step": 1297
    },
    {
      "epoch": 0.439106901217862,
      "grad_norm": 0.29069027304649353,
      "learning_rate": 4.7731996398640155e-06,
      "loss": 0.6281,
      "step": 1298
    },
    {
      "epoch": 0.43944519621109607,
      "grad_norm": 0.27709999680519104,
      "learning_rate": 4.7728266938712806e-06,
      "loss": 0.6052,
      "step": 1299
    },
    {
      "epoch": 0.4397834912043302,
      "grad_norm": 0.2892073690891266,
      "learning_rate": 4.772453456092191e-06,
      "loss": 0.6333,
      "step": 1300
    },
    {
      "epoch": 0.44012178619756426,
      "grad_norm": 0.30395960807800293,
      "learning_rate": 4.772079926574662e-06,
      "loss": 0.6376,
      "step": 1301
    },
    {
      "epoch": 0.4404600811907984,
      "grad_norm": 0.3262636363506317,
      "learning_rate": 4.771706105366647e-06,
      "loss": 0.6489,
      "step": 1302
    },
    {
      "epoch": 0.4407983761840325,
      "grad_norm": 0.2975095510482788,
      "learning_rate": 4.771331992516137e-06,
      "loss": 0.6259,
      "step": 1303
    },
    {
      "epoch": 0.44113667117726657,
      "grad_norm": 0.3001902103424072,
      "learning_rate": 4.770957588071163e-06,
      "loss": 0.6566,
      "step": 1304
    },
    {
      "epoch": 0.4414749661705007,
      "grad_norm": 0.315866619348526,
      "learning_rate": 4.7705828920797895e-06,
      "loss": 0.6424,
      "step": 1305
    },
    {
      "epoch": 0.44181326116373476,
      "grad_norm": 0.2861289978027344,
      "learning_rate": 4.770207904590121e-06,
      "loss": 0.6457,
      "step": 1306
    },
    {
      "epoch": 0.4421515561569689,
      "grad_norm": 0.30767688155174255,
      "learning_rate": 4.7698326256502966e-06,
      "loss": 0.6309,
      "step": 1307
    },
    {
      "epoch": 0.44248985115020295,
      "grad_norm": 0.3229195773601532,
      "learning_rate": 4.7694570553084974e-06,
      "loss": 0.6264,
      "step": 1308
    },
    {
      "epoch": 0.4428281461434371,
      "grad_norm": 0.29174500703811646,
      "learning_rate": 4.769081193612937e-06,
      "loss": 0.6583,
      "step": 1309
    },
    {
      "epoch": 0.4431664411366712,
      "grad_norm": 0.28457409143447876,
      "learning_rate": 4.76870504061187e-06,
      "loss": 0.6225,
      "step": 1310
    },
    {
      "epoch": 0.44350473612990526,
      "grad_norm": 0.285459965467453,
      "learning_rate": 4.768328596353586e-06,
      "loss": 0.6216,
      "step": 1311
    },
    {
      "epoch": 0.4438430311231394,
      "grad_norm": 0.29809266328811646,
      "learning_rate": 4.767951860886415e-06,
      "loss": 0.6375,
      "step": 1312
    },
    {
      "epoch": 0.44418132611637345,
      "grad_norm": 0.29426756501197815,
      "learning_rate": 4.767574834258721e-06,
      "loss": 0.6495,
      "step": 1313
    },
    {
      "epoch": 0.4445196211096076,
      "grad_norm": 0.277817040681839,
      "learning_rate": 4.7671975165189065e-06,
      "loss": 0.6251,
      "step": 1314
    },
    {
      "epoch": 0.4448579161028417,
      "grad_norm": 0.28166523575782776,
      "learning_rate": 4.766819907715412e-06,
      "loss": 0.6252,
      "step": 1315
    },
    {
      "epoch": 0.44519621109607577,
      "grad_norm": 0.2760406732559204,
      "learning_rate": 4.766442007896716e-06,
      "loss": 0.639,
      "step": 1316
    },
    {
      "epoch": 0.4455345060893099,
      "grad_norm": 0.28585439920425415,
      "learning_rate": 4.766063817111332e-06,
      "loss": 0.633,
      "step": 1317
    },
    {
      "epoch": 0.44587280108254396,
      "grad_norm": 0.2784285247325897,
      "learning_rate": 4.765685335407814e-06,
      "loss": 0.6203,
      "step": 1318
    },
    {
      "epoch": 0.4462110960757781,
      "grad_norm": 0.28521108627319336,
      "learning_rate": 4.765306562834749e-06,
      "loss": 0.6029,
      "step": 1319
    },
    {
      "epoch": 0.4465493910690122,
      "grad_norm": 0.2915601432323456,
      "learning_rate": 4.764927499440767e-06,
      "loss": 0.6454,
      "step": 1320
    },
    {
      "epoch": 0.44688768606224627,
      "grad_norm": 0.3053300976753235,
      "learning_rate": 4.7645481452745305e-06,
      "loss": 0.6349,
      "step": 1321
    },
    {
      "epoch": 0.4472259810554804,
      "grad_norm": 0.29973912239074707,
      "learning_rate": 4.764168500384741e-06,
      "loss": 0.6618,
      "step": 1322
    },
    {
      "epoch": 0.44756427604871446,
      "grad_norm": 0.2932901084423065,
      "learning_rate": 4.763788564820138e-06,
      "loss": 0.6189,
      "step": 1323
    },
    {
      "epoch": 0.4479025710419486,
      "grad_norm": 0.29192137718200684,
      "learning_rate": 4.7634083386294975e-06,
      "loss": 0.6457,
      "step": 1324
    },
    {
      "epoch": 0.4482408660351827,
      "grad_norm": 0.29242607951164246,
      "learning_rate": 4.763027821861634e-06,
      "loss": 0.6253,
      "step": 1325
    },
    {
      "epoch": 0.4485791610284168,
      "grad_norm": 0.28581079840660095,
      "learning_rate": 4.7626470145653975e-06,
      "loss": 0.6435,
      "step": 1326
    },
    {
      "epoch": 0.4489174560216509,
      "grad_norm": 0.3019113540649414,
      "learning_rate": 4.762265916789676e-06,
      "loss": 0.6543,
      "step": 1327
    },
    {
      "epoch": 0.44925575101488496,
      "grad_norm": 0.2888141870498657,
      "learning_rate": 4.7618845285833955e-06,
      "loss": 0.6301,
      "step": 1328
    },
    {
      "epoch": 0.4495940460081191,
      "grad_norm": 0.2920967936515808,
      "learning_rate": 4.761502849995519e-06,
      "loss": 0.619,
      "step": 1329
    },
    {
      "epoch": 0.44993234100135315,
      "grad_norm": 0.29467645287513733,
      "learning_rate": 4.761120881075046e-06,
      "loss": 0.6471,
      "step": 1330
    },
    {
      "epoch": 0.4502706359945873,
      "grad_norm": 0.302282452583313,
      "learning_rate": 4.760738621871014e-06,
      "loss": 0.6422,
      "step": 1331
    },
    {
      "epoch": 0.4506089309878214,
      "grad_norm": 0.32091787457466125,
      "learning_rate": 4.7603560724324985e-06,
      "loss": 0.6119,
      "step": 1332
    },
    {
      "epoch": 0.45094722598105547,
      "grad_norm": 0.2797046899795532,
      "learning_rate": 4.759973232808609e-06,
      "loss": 0.6276,
      "step": 1333
    },
    {
      "epoch": 0.4512855209742896,
      "grad_norm": 0.2929953634738922,
      "learning_rate": 4.7595901030484955e-06,
      "loss": 0.6542,
      "step": 1334
    },
    {
      "epoch": 0.45162381596752366,
      "grad_norm": 0.3046846091747284,
      "learning_rate": 4.759206683201346e-06,
      "loss": 0.6434,
      "step": 1335
    },
    {
      "epoch": 0.4519621109607578,
      "grad_norm": 0.29917415976524353,
      "learning_rate": 4.758822973316383e-06,
      "loss": 0.6491,
      "step": 1336
    },
    {
      "epoch": 0.4523004059539919,
      "grad_norm": 0.29974162578582764,
      "learning_rate": 4.758438973442867e-06,
      "loss": 0.641,
      "step": 1337
    },
    {
      "epoch": 0.45263870094722597,
      "grad_norm": 0.28453171253204346,
      "learning_rate": 4.758054683630097e-06,
      "loss": 0.6292,
      "step": 1338
    },
    {
      "epoch": 0.4529769959404601,
      "grad_norm": 0.2985386550426483,
      "learning_rate": 4.757670103927406e-06,
      "loss": 0.6235,
      "step": 1339
    },
    {
      "epoch": 0.45331529093369416,
      "grad_norm": 0.2876027226448059,
      "learning_rate": 4.757285234384169e-06,
      "loss": 0.6396,
      "step": 1340
    },
    {
      "epoch": 0.4536535859269283,
      "grad_norm": 0.29215019941329956,
      "learning_rate": 4.756900075049794e-06,
      "loss": 0.6409,
      "step": 1341
    },
    {
      "epoch": 0.4539918809201624,
      "grad_norm": 0.29155978560447693,
      "learning_rate": 4.75651462597373e-06,
      "loss": 0.6111,
      "step": 1342
    },
    {
      "epoch": 0.4543301759133965,
      "grad_norm": 0.28371426463127136,
      "learning_rate": 4.756128887205459e-06,
      "loss": 0.6385,
      "step": 1343
    },
    {
      "epoch": 0.4546684709066306,
      "grad_norm": 0.29578351974487305,
      "learning_rate": 4.755742858794503e-06,
      "loss": 0.6335,
      "step": 1344
    },
    {
      "epoch": 0.45500676589986466,
      "grad_norm": 0.2908320426940918,
      "learning_rate": 4.755356540790421e-06,
      "loss": 0.649,
      "step": 1345
    },
    {
      "epoch": 0.4553450608930988,
      "grad_norm": 0.28711384534835815,
      "learning_rate": 4.754969933242808e-06,
      "loss": 0.6248,
      "step": 1346
    },
    {
      "epoch": 0.4556833558863329,
      "grad_norm": 0.2830841839313507,
      "learning_rate": 4.754583036201297e-06,
      "loss": 0.6144,
      "step": 1347
    },
    {
      "epoch": 0.456021650879567,
      "grad_norm": 0.2901039123535156,
      "learning_rate": 4.754195849715557e-06,
      "loss": 0.6547,
      "step": 1348
    },
    {
      "epoch": 0.4563599458728011,
      "grad_norm": 0.29270651936531067,
      "learning_rate": 4.753808373835297e-06,
      "loss": 0.6314,
      "step": 1349
    },
    {
      "epoch": 0.45669824086603517,
      "grad_norm": 0.3053850829601288,
      "learning_rate": 4.7534206086102604e-06,
      "loss": 0.6672,
      "step": 1350
    },
    {
      "epoch": 0.4570365358592693,
      "grad_norm": 0.3080502450466156,
      "learning_rate": 4.753032554090229e-06,
      "loss": 0.658,
      "step": 1351
    },
    {
      "epoch": 0.45737483085250336,
      "grad_norm": 0.2952156662940979,
      "learning_rate": 4.75264421032502e-06,
      "loss": 0.6457,
      "step": 1352
    },
    {
      "epoch": 0.4577131258457375,
      "grad_norm": 0.29784637689590454,
      "learning_rate": 4.752255577364492e-06,
      "loss": 0.6259,
      "step": 1353
    },
    {
      "epoch": 0.4580514208389716,
      "grad_norm": 0.2907617688179016,
      "learning_rate": 4.751866655258535e-06,
      "loss": 0.6375,
      "step": 1354
    },
    {
      "epoch": 0.45838971583220567,
      "grad_norm": 0.29962706565856934,
      "learning_rate": 4.75147744405708e-06,
      "loss": 0.6405,
      "step": 1355
    },
    {
      "epoch": 0.4587280108254398,
      "grad_norm": 0.28614145517349243,
      "learning_rate": 4.751087943810094e-06,
      "loss": 0.6156,
      "step": 1356
    },
    {
      "epoch": 0.45906630581867386,
      "grad_norm": 0.2923153042793274,
      "learning_rate": 4.7506981545675815e-06,
      "loss": 0.6227,
      "step": 1357
    },
    {
      "epoch": 0.459404600811908,
      "grad_norm": 0.29524093866348267,
      "learning_rate": 4.7503080763795836e-06,
      "loss": 0.6504,
      "step": 1358
    },
    {
      "epoch": 0.4597428958051421,
      "grad_norm": 0.2971424460411072,
      "learning_rate": 4.749917709296178e-06,
      "loss": 0.6445,
      "step": 1359
    },
    {
      "epoch": 0.46008119079837617,
      "grad_norm": 0.29231759905815125,
      "learning_rate": 4.749527053367482e-06,
      "loss": 0.6015,
      "step": 1360
    },
    {
      "epoch": 0.4604194857916103,
      "grad_norm": 0.2905111014842987,
      "learning_rate": 4.749136108643646e-06,
      "loss": 0.6244,
      "step": 1361
    },
    {
      "epoch": 0.46075778078484436,
      "grad_norm": 0.3043908178806305,
      "learning_rate": 4.748744875174861e-06,
      "loss": 0.6537,
      "step": 1362
    },
    {
      "epoch": 0.4610960757780785,
      "grad_norm": 0.297868013381958,
      "learning_rate": 4.748353353011353e-06,
      "loss": 0.6261,
      "step": 1363
    },
    {
      "epoch": 0.4614343707713126,
      "grad_norm": 0.2935093641281128,
      "learning_rate": 4.747961542203387e-06,
      "loss": 0.6272,
      "step": 1364
    },
    {
      "epoch": 0.4617726657645467,
      "grad_norm": 0.2828471064567566,
      "learning_rate": 4.747569442801262e-06,
      "loss": 0.6421,
      "step": 1365
    },
    {
      "epoch": 0.4621109607577808,
      "grad_norm": 0.29044583439826965,
      "learning_rate": 4.747177054855316e-06,
      "loss": 0.6221,
      "step": 1366
    },
    {
      "epoch": 0.46244925575101486,
      "grad_norm": 0.2825765907764435,
      "learning_rate": 4.746784378415926e-06,
      "loss": 0.6054,
      "step": 1367
    },
    {
      "epoch": 0.462787550744249,
      "grad_norm": 0.30992552638053894,
      "learning_rate": 4.746391413533503e-06,
      "loss": 0.621,
      "step": 1368
    },
    {
      "epoch": 0.4631258457374831,
      "grad_norm": 0.30108433961868286,
      "learning_rate": 4.745998160258495e-06,
      "loss": 0.6174,
      "step": 1369
    },
    {
      "epoch": 0.4634641407307172,
      "grad_norm": 0.27994582056999207,
      "learning_rate": 4.745604618641388e-06,
      "loss": 0.6526,
      "step": 1370
    },
    {
      "epoch": 0.4638024357239513,
      "grad_norm": 0.3178618848323822,
      "learning_rate": 4.745210788732707e-06,
      "loss": 0.6011,
      "step": 1371
    },
    {
      "epoch": 0.46414073071718537,
      "grad_norm": 0.29102611541748047,
      "learning_rate": 4.74481667058301e-06,
      "loss": 0.6517,
      "step": 1372
    },
    {
      "epoch": 0.4644790257104195,
      "grad_norm": 0.2938326597213745,
      "learning_rate": 4.744422264242895e-06,
      "loss": 0.6288,
      "step": 1373
    },
    {
      "epoch": 0.4648173207036536,
      "grad_norm": 0.2730098366737366,
      "learning_rate": 4.744027569762996e-06,
      "loss": 0.6241,
      "step": 1374
    },
    {
      "epoch": 0.4651556156968877,
      "grad_norm": 0.30259615182876587,
      "learning_rate": 4.743632587193983e-06,
      "loss": 0.6524,
      "step": 1375
    },
    {
      "epoch": 0.4654939106901218,
      "grad_norm": 0.2964252233505249,
      "learning_rate": 4.743237316586564e-06,
      "loss": 0.6453,
      "step": 1376
    },
    {
      "epoch": 0.46583220568335587,
      "grad_norm": 0.2919650673866272,
      "learning_rate": 4.742841757991486e-06,
      "loss": 0.6295,
      "step": 1377
    },
    {
      "epoch": 0.46617050067659,
      "grad_norm": 0.3055798411369324,
      "learning_rate": 4.74244591145953e-06,
      "loss": 0.6531,
      "step": 1378
    },
    {
      "epoch": 0.46650879566982406,
      "grad_norm": 0.2998957335948944,
      "learning_rate": 4.742049777041514e-06,
      "loss": 0.6265,
      "step": 1379
    },
    {
      "epoch": 0.4668470906630582,
      "grad_norm": 0.29530641436576843,
      "learning_rate": 4.741653354788295e-06,
      "loss": 0.6321,
      "step": 1380
    },
    {
      "epoch": 0.4671853856562923,
      "grad_norm": 0.2901705503463745,
      "learning_rate": 4.741256644750766e-06,
      "loss": 0.6425,
      "step": 1381
    },
    {
      "epoch": 0.4675236806495264,
      "grad_norm": 0.2957158386707306,
      "learning_rate": 4.7408596469798555e-06,
      "loss": 0.6347,
      "step": 1382
    },
    {
      "epoch": 0.4678619756427605,
      "grad_norm": 0.30493056774139404,
      "learning_rate": 4.740462361526531e-06,
      "loss": 0.6378,
      "step": 1383
    },
    {
      "epoch": 0.46820027063599456,
      "grad_norm": 0.30551135540008545,
      "learning_rate": 4.740064788441796e-06,
      "loss": 0.644,
      "step": 1384
    },
    {
      "epoch": 0.4685385656292287,
      "grad_norm": 0.3051181435585022,
      "learning_rate": 4.739666927776691e-06,
      "loss": 0.6409,
      "step": 1385
    },
    {
      "epoch": 0.4688768606224628,
      "grad_norm": 0.2968808114528656,
      "learning_rate": 4.739268779582294e-06,
      "loss": 0.6492,
      "step": 1386
    },
    {
      "epoch": 0.4692151556156969,
      "grad_norm": 0.3039150834083557,
      "learning_rate": 4.738870343909719e-06,
      "loss": 0.6412,
      "step": 1387
    },
    {
      "epoch": 0.469553450608931,
      "grad_norm": 0.2963270843029022,
      "learning_rate": 4.738471620810116e-06,
      "loss": 0.6187,
      "step": 1388
    },
    {
      "epoch": 0.46989174560216507,
      "grad_norm": 0.2898751199245453,
      "learning_rate": 4.738072610334677e-06,
      "loss": 0.6273,
      "step": 1389
    },
    {
      "epoch": 0.4702300405953992,
      "grad_norm": 0.30511993169784546,
      "learning_rate": 4.737673312534624e-06,
      "loss": 0.6142,
      "step": 1390
    },
    {
      "epoch": 0.4705683355886333,
      "grad_norm": 0.2916101813316345,
      "learning_rate": 4.737273727461219e-06,
      "loss": 0.6232,
      "step": 1391
    },
    {
      "epoch": 0.4709066305818674,
      "grad_norm": 0.29085099697113037,
      "learning_rate": 4.736873855165762e-06,
      "loss": 0.6198,
      "step": 1392
    },
    {
      "epoch": 0.4712449255751015,
      "grad_norm": 0.28201308846473694,
      "learning_rate": 4.7364736956995885e-06,
      "loss": 0.6311,
      "step": 1393
    },
    {
      "epoch": 0.47158322056833557,
      "grad_norm": 0.30364304780960083,
      "learning_rate": 4.736073249114071e-06,
      "loss": 0.6485,
      "step": 1394
    },
    {
      "epoch": 0.4719215155615697,
      "grad_norm": 0.3099037706851959,
      "learning_rate": 4.735672515460619e-06,
      "loss": 0.6662,
      "step": 1395
    },
    {
      "epoch": 0.4722598105548038,
      "grad_norm": 0.3040882647037506,
      "learning_rate": 4.7352714947906785e-06,
      "loss": 0.65,
      "step": 1396
    },
    {
      "epoch": 0.4725981055480379,
      "grad_norm": 0.29287227988243103,
      "learning_rate": 4.7348701871557335e-06,
      "loss": 0.6391,
      "step": 1397
    },
    {
      "epoch": 0.472936400541272,
      "grad_norm": 0.309793084859848,
      "learning_rate": 4.734468592607304e-06,
      "loss": 0.6349,
      "step": 1398
    },
    {
      "epoch": 0.4732746955345061,
      "grad_norm": 0.32687175273895264,
      "learning_rate": 4.7340667111969455e-06,
      "loss": 0.6093,
      "step": 1399
    },
    {
      "epoch": 0.4736129905277402,
      "grad_norm": 0.2984752953052521,
      "learning_rate": 4.7336645429762525e-06,
      "loss": 0.6214,
      "step": 1400
    },
    {
      "epoch": 0.47395128552097426,
      "grad_norm": 0.3069734573364258,
      "learning_rate": 4.733262087996856e-06,
      "loss": 0.6316,
      "step": 1401
    },
    {
      "epoch": 0.4742895805142084,
      "grad_norm": 0.3015008270740509,
      "learning_rate": 4.732859346310423e-06,
      "loss": 0.6207,
      "step": 1402
    },
    {
      "epoch": 0.4746278755074425,
      "grad_norm": 0.2884121835231781,
      "learning_rate": 4.732456317968657e-06,
      "loss": 0.6189,
      "step": 1403
    },
    {
      "epoch": 0.4749661705006766,
      "grad_norm": 0.3015613555908203,
      "learning_rate": 4.732053003023301e-06,
      "loss": 0.6479,
      "step": 1404
    },
    {
      "epoch": 0.4753044654939107,
      "grad_norm": 0.309891939163208,
      "learning_rate": 4.73164940152613e-06,
      "loss": 0.6508,
      "step": 1405
    },
    {
      "epoch": 0.47564276048714477,
      "grad_norm": 0.2924291491508484,
      "learning_rate": 4.73124551352896e-06,
      "loss": 0.6394,
      "step": 1406
    },
    {
      "epoch": 0.4759810554803789,
      "grad_norm": 0.3016890585422516,
      "learning_rate": 4.730841339083642e-06,
      "loss": 0.6225,
      "step": 1407
    },
    {
      "epoch": 0.476319350473613,
      "grad_norm": 0.32249557971954346,
      "learning_rate": 4.730436878242064e-06,
      "loss": 0.6343,
      "step": 1408
    },
    {
      "epoch": 0.4766576454668471,
      "grad_norm": 0.29110148549079895,
      "learning_rate": 4.730032131056152e-06,
      "loss": 0.6443,
      "step": 1409
    },
    {
      "epoch": 0.4769959404600812,
      "grad_norm": 0.2962009906768799,
      "learning_rate": 4.7296270975778644e-06,
      "loss": 0.6331,
      "step": 1410
    },
    {
      "epoch": 0.47733423545331527,
      "grad_norm": 0.31483912467956543,
      "learning_rate": 4.729221777859203e-06,
      "loss": 0.6251,
      "step": 1411
    },
    {
      "epoch": 0.4776725304465494,
      "grad_norm": 0.2956962585449219,
      "learning_rate": 4.728816171952202e-06,
      "loss": 0.6193,
      "step": 1412
    },
    {
      "epoch": 0.4780108254397835,
      "grad_norm": 0.29781419038772583,
      "learning_rate": 4.7284102799089325e-06,
      "loss": 0.6516,
      "step": 1413
    },
    {
      "epoch": 0.4783491204330176,
      "grad_norm": 0.31204479932785034,
      "learning_rate": 4.728004101781503e-06,
      "loss": 0.6504,
      "step": 1414
    },
    {
      "epoch": 0.4786874154262517,
      "grad_norm": 0.29530659317970276,
      "learning_rate": 4.7275976376220585e-06,
      "loss": 0.6497,
      "step": 1415
    },
    {
      "epoch": 0.4790257104194858,
      "grad_norm": 0.2870703339576721,
      "learning_rate": 4.727190887482783e-06,
      "loss": 0.6241,
      "step": 1416
    },
    {
      "epoch": 0.4793640054127199,
      "grad_norm": 0.2797526717185974,
      "learning_rate": 4.726783851415893e-06,
      "loss": 0.6024,
      "step": 1417
    },
    {
      "epoch": 0.479702300405954,
      "grad_norm": 0.31374120712280273,
      "learning_rate": 4.7263765294736465e-06,
      "loss": 0.6248,
      "step": 1418
    },
    {
      "epoch": 0.4800405953991881,
      "grad_norm": 0.3121717870235443,
      "learning_rate": 4.7259689217083325e-06,
      "loss": 0.6621,
      "step": 1419
    },
    {
      "epoch": 0.4803788903924222,
      "grad_norm": 0.29636168479919434,
      "learning_rate": 4.725561028172282e-06,
      "loss": 0.6427,
      "step": 1420
    },
    {
      "epoch": 0.4807171853856563,
      "grad_norm": 0.2872474789619446,
      "learning_rate": 4.72515284891786e-06,
      "loss": 0.6355,
      "step": 1421
    },
    {
      "epoch": 0.4810554803788904,
      "grad_norm": 0.3191288411617279,
      "learning_rate": 4.724744383997469e-06,
      "loss": 0.6253,
      "step": 1422
    },
    {
      "epoch": 0.48139377537212447,
      "grad_norm": 0.2883187234401703,
      "learning_rate": 4.724335633463547e-06,
      "loss": 0.6315,
      "step": 1423
    },
    {
      "epoch": 0.4817320703653586,
      "grad_norm": 0.2939365804195404,
      "learning_rate": 4.72392659736857e-06,
      "loss": 0.6391,
      "step": 1424
    },
    {
      "epoch": 0.4820703653585927,
      "grad_norm": 0.2954607903957367,
      "learning_rate": 4.72351727576505e-06,
      "loss": 0.6202,
      "step": 1425
    },
    {
      "epoch": 0.4824086603518268,
      "grad_norm": 0.2945690453052521,
      "learning_rate": 4.723107668705537e-06,
      "loss": 0.6501,
      "step": 1426
    },
    {
      "epoch": 0.4827469553450609,
      "grad_norm": 0.29526031017303467,
      "learning_rate": 4.722697776242615e-06,
      "loss": 0.6231,
      "step": 1427
    },
    {
      "epoch": 0.48308525033829497,
      "grad_norm": 0.29499900341033936,
      "learning_rate": 4.722287598428907e-06,
      "loss": 0.6362,
      "step": 1428
    },
    {
      "epoch": 0.4834235453315291,
      "grad_norm": 0.2970479726791382,
      "learning_rate": 4.721877135317071e-06,
      "loss": 0.642,
      "step": 1429
    },
    {
      "epoch": 0.4837618403247632,
      "grad_norm": 0.30396947264671326,
      "learning_rate": 4.721466386959805e-06,
      "loss": 0.6774,
      "step": 1430
    },
    {
      "epoch": 0.4841001353179973,
      "grad_norm": 0.29407331347465515,
      "learning_rate": 4.721055353409837e-06,
      "loss": 0.6475,
      "step": 1431
    },
    {
      "epoch": 0.4844384303112314,
      "grad_norm": 0.2962198555469513,
      "learning_rate": 4.720644034719938e-06,
      "loss": 0.6495,
      "step": 1432
    },
    {
      "epoch": 0.4847767253044655,
      "grad_norm": 0.30761298537254333,
      "learning_rate": 4.720232430942912e-06,
      "loss": 0.6254,
      "step": 1433
    },
    {
      "epoch": 0.4851150202976996,
      "grad_norm": 0.29770034551620483,
      "learning_rate": 4.719820542131604e-06,
      "loss": 0.6483,
      "step": 1434
    },
    {
      "epoch": 0.4854533152909337,
      "grad_norm": 0.2793305814266205,
      "learning_rate": 4.7194083683388885e-06,
      "loss": 0.6207,
      "step": 1435
    },
    {
      "epoch": 0.4857916102841678,
      "grad_norm": 0.29254627227783203,
      "learning_rate": 4.718995909617683e-06,
      "loss": 0.6026,
      "step": 1436
    },
    {
      "epoch": 0.4861299052774019,
      "grad_norm": 0.30059441924095154,
      "learning_rate": 4.718583166020938e-06,
      "loss": 0.6157,
      "step": 1437
    },
    {
      "epoch": 0.486468200270636,
      "grad_norm": 0.3180963397026062,
      "learning_rate": 4.718170137601642e-06,
      "loss": 0.6643,
      "step": 1438
    },
    {
      "epoch": 0.4868064952638701,
      "grad_norm": 0.3006470203399658,
      "learning_rate": 4.71775682441282e-06,
      "loss": 0.6524,
      "step": 1439
    },
    {
      "epoch": 0.4871447902571042,
      "grad_norm": 0.2820841670036316,
      "learning_rate": 4.717343226507533e-06,
      "loss": 0.6185,
      "step": 1440
    },
    {
      "epoch": 0.4874830852503383,
      "grad_norm": 0.28982532024383545,
      "learning_rate": 4.71692934393888e-06,
      "loss": 0.6269,
      "step": 1441
    },
    {
      "epoch": 0.4878213802435724,
      "grad_norm": 0.29427915811538696,
      "learning_rate": 4.716515176759992e-06,
      "loss": 0.6493,
      "step": 1442
    },
    {
      "epoch": 0.4881596752368065,
      "grad_norm": 0.29777371883392334,
      "learning_rate": 4.716100725024044e-06,
      "loss": 0.6418,
      "step": 1443
    },
    {
      "epoch": 0.4884979702300406,
      "grad_norm": 0.29938769340515137,
      "learning_rate": 4.7156859887842416e-06,
      "loss": 0.6325,
      "step": 1444
    },
    {
      "epoch": 0.48883626522327467,
      "grad_norm": 0.305256724357605,
      "learning_rate": 4.715270968093828e-06,
      "loss": 0.6552,
      "step": 1445
    },
    {
      "epoch": 0.4891745602165088,
      "grad_norm": 0.29933780431747437,
      "learning_rate": 4.714855663006086e-06,
      "loss": 0.6296,
      "step": 1446
    },
    {
      "epoch": 0.4895128552097429,
      "grad_norm": 0.2861994802951813,
      "learning_rate": 4.71444007357433e-06,
      "loss": 0.6133,
      "step": 1447
    },
    {
      "epoch": 0.489851150202977,
      "grad_norm": 0.2853560745716095,
      "learning_rate": 4.7140241998519156e-06,
      "loss": 0.654,
      "step": 1448
    },
    {
      "epoch": 0.4901894451962111,
      "grad_norm": 0.2986946403980255,
      "learning_rate": 4.713608041892232e-06,
      "loss": 0.6182,
      "step": 1449
    },
    {
      "epoch": 0.49052774018944517,
      "grad_norm": 0.2984166145324707,
      "learning_rate": 4.713191599748706e-06,
      "loss": 0.6207,
      "step": 1450
    },
    {
      "epoch": 0.4908660351826793,
      "grad_norm": 0.32026952505111694,
      "learning_rate": 4.712774873474799e-06,
      "loss": 0.6257,
      "step": 1451
    },
    {
      "epoch": 0.4912043301759134,
      "grad_norm": 0.31071943044662476,
      "learning_rate": 4.712357863124013e-06,
      "loss": 0.6277,
      "step": 1452
    },
    {
      "epoch": 0.4915426251691475,
      "grad_norm": 0.2902413308620453,
      "learning_rate": 4.711940568749883e-06,
      "loss": 0.6342,
      "step": 1453
    },
    {
      "epoch": 0.4918809201623816,
      "grad_norm": 0.29815566539764404,
      "learning_rate": 4.71152299040598e-06,
      "loss": 0.633,
      "step": 1454
    },
    {
      "epoch": 0.4922192151556157,
      "grad_norm": 0.3356594741344452,
      "learning_rate": 4.7111051281459155e-06,
      "loss": 0.633,
      "step": 1455
    },
    {
      "epoch": 0.4925575101488498,
      "grad_norm": 0.2962017059326172,
      "learning_rate": 4.7106869820233324e-06,
      "loss": 0.6481,
      "step": 1456
    },
    {
      "epoch": 0.4928958051420839,
      "grad_norm": 0.30914294719696045,
      "learning_rate": 4.710268552091913e-06,
      "loss": 0.633,
      "step": 1457
    },
    {
      "epoch": 0.493234100135318,
      "grad_norm": 0.29593202471733093,
      "learning_rate": 4.709849838405377e-06,
      "loss": 0.6326,
      "step": 1458
    },
    {
      "epoch": 0.4935723951285521,
      "grad_norm": 0.29423394799232483,
      "learning_rate": 4.709430841017477e-06,
      "loss": 0.6414,
      "step": 1459
    },
    {
      "epoch": 0.4939106901217862,
      "grad_norm": 0.2971169650554657,
      "learning_rate": 4.709011559982006e-06,
      "loss": 0.6301,
      "step": 1460
    },
    {
      "epoch": 0.4942489851150203,
      "grad_norm": 0.30087241530418396,
      "learning_rate": 4.70859199535279e-06,
      "loss": 0.6349,
      "step": 1461
    },
    {
      "epoch": 0.4945872801082544,
      "grad_norm": 0.30483344197273254,
      "learning_rate": 4.7081721471836934e-06,
      "loss": 0.6544,
      "step": 1462
    },
    {
      "epoch": 0.4949255751014885,
      "grad_norm": 0.2955804765224457,
      "learning_rate": 4.707752015528616e-06,
      "loss": 0.6165,
      "step": 1463
    },
    {
      "epoch": 0.4952638700947226,
      "grad_norm": 0.2943514287471771,
      "learning_rate": 4.707331600441496e-06,
      "loss": 0.6206,
      "step": 1464
    },
    {
      "epoch": 0.4956021650879567,
      "grad_norm": 0.29659345746040344,
      "learning_rate": 4.706910901976304e-06,
      "loss": 0.6441,
      "step": 1465
    },
    {
      "epoch": 0.4959404600811908,
      "grad_norm": 0.29106733202934265,
      "learning_rate": 4.706489920187051e-06,
      "loss": 0.654,
      "step": 1466
    },
    {
      "epoch": 0.49627875507442487,
      "grad_norm": 0.30567026138305664,
      "learning_rate": 4.706068655127783e-06,
      "loss": 0.6589,
      "step": 1467
    },
    {
      "epoch": 0.496617050067659,
      "grad_norm": 0.3061468303203583,
      "learning_rate": 4.705647106852581e-06,
      "loss": 0.6408,
      "step": 1468
    },
    {
      "epoch": 0.4969553450608931,
      "grad_norm": 0.30140402913093567,
      "learning_rate": 4.705225275415565e-06,
      "loss": 0.6068,
      "step": 1469
    },
    {
      "epoch": 0.4972936400541272,
      "grad_norm": 0.29500529170036316,
      "learning_rate": 4.704803160870888e-06,
      "loss": 0.6431,
      "step": 1470
    },
    {
      "epoch": 0.4976319350473613,
      "grad_norm": 0.295264333486557,
      "learning_rate": 4.704380763272743e-06,
      "loss": 0.6281,
      "step": 1471
    },
    {
      "epoch": 0.4979702300405954,
      "grad_norm": 0.2895084023475647,
      "learning_rate": 4.703958082675357e-06,
      "loss": 0.6677,
      "step": 1472
    },
    {
      "epoch": 0.4983085250338295,
      "grad_norm": 0.2979891896247864,
      "learning_rate": 4.7035351191329935e-06,
      "loss": 0.6417,
      "step": 1473
    },
    {
      "epoch": 0.4986468200270636,
      "grad_norm": 0.29371345043182373,
      "learning_rate": 4.703111872699953e-06,
      "loss": 0.5812,
      "step": 1474
    },
    {
      "epoch": 0.4989851150202977,
      "grad_norm": 0.3126574754714966,
      "learning_rate": 4.702688343430573e-06,
      "loss": 0.6373,
      "step": 1475
    },
    {
      "epoch": 0.4993234100135318,
      "grad_norm": 0.30721336603164673,
      "learning_rate": 4.7022645313792235e-06,
      "loss": 0.6307,
      "step": 1476
    },
    {
      "epoch": 0.4996617050067659,
      "grad_norm": 0.296394020318985,
      "learning_rate": 4.701840436600317e-06,
      "loss": 0.647,
      "step": 1477
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2912466526031494,
      "learning_rate": 4.701416059148297e-06,
      "loss": 0.6266,
      "step": 1478
    },
    {
      "epoch": 0.5003382949932341,
      "grad_norm": 0.30898961424827576,
      "learning_rate": 4.700991399077646e-06,
      "loss": 0.637,
      "step": 1479
    },
    {
      "epoch": 0.5006765899864682,
      "grad_norm": 0.3106672763824463,
      "learning_rate": 4.700566456442882e-06,
      "loss": 0.6161,
      "step": 1480
    },
    {
      "epoch": 0.5010148849797023,
      "grad_norm": 0.2974197268486023,
      "learning_rate": 4.700141231298558e-06,
      "loss": 0.6181,
      "step": 1481
    },
    {
      "epoch": 0.5013531799729364,
      "grad_norm": 0.3075565695762634,
      "learning_rate": 4.699715723699266e-06,
      "loss": 0.6344,
      "step": 1482
    },
    {
      "epoch": 0.5016914749661705,
      "grad_norm": 0.2872021198272705,
      "learning_rate": 4.699289933699633e-06,
      "loss": 0.6409,
      "step": 1483
    },
    {
      "epoch": 0.5020297699594046,
      "grad_norm": 0.28962892293930054,
      "learning_rate": 4.698863861354322e-06,
      "loss": 0.5999,
      "step": 1484
    },
    {
      "epoch": 0.5023680649526387,
      "grad_norm": 0.2840323746204376,
      "learning_rate": 4.698437506718031e-06,
      "loss": 0.608,
      "step": 1485
    },
    {
      "epoch": 0.5027063599458728,
      "grad_norm": 0.2985817790031433,
      "learning_rate": 4.6980108698454965e-06,
      "loss": 0.6344,
      "step": 1486
    },
    {
      "epoch": 0.5030446549391069,
      "grad_norm": 0.295597642660141,
      "learning_rate": 4.69758395079149e-06,
      "loss": 0.6484,
      "step": 1487
    },
    {
      "epoch": 0.503382949932341,
      "grad_norm": 0.28852009773254395,
      "learning_rate": 4.69715674961082e-06,
      "loss": 0.6124,
      "step": 1488
    },
    {
      "epoch": 0.5037212449255751,
      "grad_norm": 0.2893640995025635,
      "learning_rate": 4.696729266358331e-06,
      "loss": 0.6167,
      "step": 1489
    },
    {
      "epoch": 0.5040595399188093,
      "grad_norm": 0.30135470628738403,
      "learning_rate": 4.6963015010889025e-06,
      "loss": 0.6498,
      "step": 1490
    },
    {
      "epoch": 0.5043978349120433,
      "grad_norm": 0.2909967303276062,
      "learning_rate": 4.695873453857451e-06,
      "loss": 0.6422,
      "step": 1491
    },
    {
      "epoch": 0.5047361299052774,
      "grad_norm": 0.299000084400177,
      "learning_rate": 4.695445124718932e-06,
      "loss": 0.6358,
      "step": 1492
    },
    {
      "epoch": 0.5050744248985115,
      "grad_norm": 0.29471755027770996,
      "learning_rate": 4.695016513728331e-06,
      "loss": 0.6081,
      "step": 1493
    },
    {
      "epoch": 0.5054127198917456,
      "grad_norm": 0.30111798644065857,
      "learning_rate": 4.694587620940675e-06,
      "loss": 0.6396,
      "step": 1494
    },
    {
      "epoch": 0.5057510148849798,
      "grad_norm": 0.29957783222198486,
      "learning_rate": 4.694158446411025e-06,
      "loss": 0.6298,
      "step": 1495
    },
    {
      "epoch": 0.5060893098782138,
      "grad_norm": 0.2970424294471741,
      "learning_rate": 4.69372899019448e-06,
      "loss": 0.6281,
      "step": 1496
    },
    {
      "epoch": 0.5064276048714479,
      "grad_norm": 0.3034062385559082,
      "learning_rate": 4.69329925234617e-06,
      "loss": 0.6496,
      "step": 1497
    },
    {
      "epoch": 0.506765899864682,
      "grad_norm": 0.31338825821876526,
      "learning_rate": 4.69286923292127e-06,
      "loss": 0.6452,
      "step": 1498
    },
    {
      "epoch": 0.5071041948579161,
      "grad_norm": 0.2992928624153137,
      "learning_rate": 4.692438931974983e-06,
      "loss": 0.6376,
      "step": 1499
    },
    {
      "epoch": 0.5074424898511503,
      "grad_norm": 0.29723188281059265,
      "learning_rate": 4.692008349562551e-06,
      "loss": 0.636,
      "step": 1500
    },
    {
      "epoch": 0.5077807848443843,
      "grad_norm": 0.29501450061798096,
      "learning_rate": 4.691577485739254e-06,
      "loss": 0.6134,
      "step": 1501
    },
    {
      "epoch": 0.5081190798376184,
      "grad_norm": 0.3050749897956848,
      "learning_rate": 4.691146340560405e-06,
      "loss": 0.639,
      "step": 1502
    },
    {
      "epoch": 0.5084573748308525,
      "grad_norm": 0.29698941111564636,
      "learning_rate": 4.690714914081355e-06,
      "loss": 0.6379,
      "step": 1503
    },
    {
      "epoch": 0.5087956698240866,
      "grad_norm": 0.29626375436782837,
      "learning_rate": 4.690283206357491e-06,
      "loss": 0.6353,
      "step": 1504
    },
    {
      "epoch": 0.5091339648173207,
      "grad_norm": 0.3038589060306549,
      "learning_rate": 4.689851217444236e-06,
      "loss": 0.6212,
      "step": 1505
    },
    {
      "epoch": 0.5094722598105548,
      "grad_norm": 0.3010556697845459,
      "learning_rate": 4.689418947397047e-06,
      "loss": 0.6423,
      "step": 1506
    },
    {
      "epoch": 0.5098105548037889,
      "grad_norm": 0.28557172417640686,
      "learning_rate": 4.688986396271422e-06,
      "loss": 0.6096,
      "step": 1507
    },
    {
      "epoch": 0.510148849797023,
      "grad_norm": 0.3025129735469818,
      "learning_rate": 4.688553564122891e-06,
      "loss": 0.6154,
      "step": 1508
    },
    {
      "epoch": 0.5104871447902571,
      "grad_norm": 0.31298425793647766,
      "learning_rate": 4.68812045100702e-06,
      "loss": 0.6312,
      "step": 1509
    },
    {
      "epoch": 0.5108254397834912,
      "grad_norm": 0.28845450282096863,
      "learning_rate": 4.687687056979413e-06,
      "loss": 0.616,
      "step": 1510
    },
    {
      "epoch": 0.5111637347767253,
      "grad_norm": 0.3298991024494171,
      "learning_rate": 4.687253382095709e-06,
      "loss": 0.6243,
      "step": 1511
    },
    {
      "epoch": 0.5115020297699594,
      "grad_norm": 0.3234930634498596,
      "learning_rate": 4.686819426411583e-06,
      "loss": 0.632,
      "step": 1512
    },
    {
      "epoch": 0.5118403247631935,
      "grad_norm": 0.3078064024448395,
      "learning_rate": 4.686385189982748e-06,
      "loss": 0.6429,
      "step": 1513
    },
    {
      "epoch": 0.5121786197564276,
      "grad_norm": 0.3139505088329315,
      "learning_rate": 4.6859506728649506e-06,
      "loss": 0.6035,
      "step": 1514
    },
    {
      "epoch": 0.5125169147496617,
      "grad_norm": 0.3039025366306305,
      "learning_rate": 4.685515875113974e-06,
      "loss": 0.6127,
      "step": 1515
    },
    {
      "epoch": 0.5128552097428958,
      "grad_norm": 0.2968452572822571,
      "learning_rate": 4.685080796785637e-06,
      "loss": 0.6297,
      "step": 1516
    },
    {
      "epoch": 0.5131935047361299,
      "grad_norm": 0.34408167004585266,
      "learning_rate": 4.6846454379357965e-06,
      "loss": 0.6367,
      "step": 1517
    },
    {
      "epoch": 0.513531799729364,
      "grad_norm": 0.29630929231643677,
      "learning_rate": 4.684209798620343e-06,
      "loss": 0.6253,
      "step": 1518
    },
    {
      "epoch": 0.5138700947225981,
      "grad_norm": 0.32279467582702637,
      "learning_rate": 4.683773878895206e-06,
      "loss": 0.6103,
      "step": 1519
    },
    {
      "epoch": 0.5142083897158322,
      "grad_norm": 0.2961653470993042,
      "learning_rate": 4.683337678816345e-06,
      "loss": 0.6256,
      "step": 1520
    },
    {
      "epoch": 0.5145466847090663,
      "grad_norm": 0.30008238554000854,
      "learning_rate": 4.682901198439764e-06,
      "loss": 0.636,
      "step": 1521
    },
    {
      "epoch": 0.5148849797023004,
      "grad_norm": 0.29182377457618713,
      "learning_rate": 4.682464437821497e-06,
      "loss": 0.6359,
      "step": 1522
    },
    {
      "epoch": 0.5152232746955345,
      "grad_norm": 0.3136955499649048,
      "learning_rate": 4.682027397017614e-06,
      "loss": 0.6213,
      "step": 1523
    },
    {
      "epoch": 0.5155615696887687,
      "grad_norm": 0.3067409098148346,
      "learning_rate": 4.681590076084224e-06,
      "loss": 0.6088,
      "step": 1524
    },
    {
      "epoch": 0.5158998646820027,
      "grad_norm": 0.31430402398109436,
      "learning_rate": 4.6811524750774695e-06,
      "loss": 0.6581,
      "step": 1525
    },
    {
      "epoch": 0.5162381596752368,
      "grad_norm": 0.29381296038627625,
      "learning_rate": 4.68071459405353e-06,
      "loss": 0.5942,
      "step": 1526
    },
    {
      "epoch": 0.5165764546684709,
      "grad_norm": 0.29115763306617737,
      "learning_rate": 4.680276433068623e-06,
      "loss": 0.6393,
      "step": 1527
    },
    {
      "epoch": 0.516914749661705,
      "grad_norm": 0.27736565470695496,
      "learning_rate": 4.679837992178996e-06,
      "loss": 0.6567,
      "step": 1528
    },
    {
      "epoch": 0.5172530446549392,
      "grad_norm": 0.2848239243030548,
      "learning_rate": 4.679399271440939e-06,
      "loss": 0.6144,
      "step": 1529
    },
    {
      "epoch": 0.5175913396481732,
      "grad_norm": 0.30477553606033325,
      "learning_rate": 4.678960270910774e-06,
      "loss": 0.6314,
      "step": 1530
    },
    {
      "epoch": 0.5179296346414073,
      "grad_norm": 0.2981856167316437,
      "learning_rate": 4.678520990644862e-06,
      "loss": 0.6189,
      "step": 1531
    },
    {
      "epoch": 0.5182679296346414,
      "grad_norm": 0.29681396484375,
      "learning_rate": 4.678081430699594e-06,
      "loss": 0.6103,
      "step": 1532
    },
    {
      "epoch": 0.5186062246278755,
      "grad_norm": 0.2962600290775299,
      "learning_rate": 4.677641591131404e-06,
      "loss": 0.6481,
      "step": 1533
    },
    {
      "epoch": 0.5189445196211097,
      "grad_norm": 0.28115904331207275,
      "learning_rate": 4.6772014719967585e-06,
      "loss": 0.6124,
      "step": 1534
    },
    {
      "epoch": 0.5192828146143437,
      "grad_norm": 0.306104451417923,
      "learning_rate": 4.6767610733521594e-06,
      "loss": 0.652,
      "step": 1535
    },
    {
      "epoch": 0.5196211096075778,
      "grad_norm": 0.30700162053108215,
      "learning_rate": 4.6763203952541455e-06,
      "loss": 0.6487,
      "step": 1536
    },
    {
      "epoch": 0.5199594046008119,
      "grad_norm": 0.2890525162220001,
      "learning_rate": 4.675879437759292e-06,
      "loss": 0.6282,
      "step": 1537
    },
    {
      "epoch": 0.520297699594046,
      "grad_norm": 0.3027341365814209,
      "learning_rate": 4.675438200924207e-06,
      "loss": 0.6332,
      "step": 1538
    },
    {
      "epoch": 0.5206359945872802,
      "grad_norm": 0.29767075181007385,
      "learning_rate": 4.674996684805538e-06,
      "loss": 0.6543,
      "step": 1539
    },
    {
      "epoch": 0.5209742895805142,
      "grad_norm": 0.2975456118583679,
      "learning_rate": 4.674554889459969e-06,
      "loss": 0.6186,
      "step": 1540
    },
    {
      "epoch": 0.5213125845737483,
      "grad_norm": 0.3017563819885254,
      "learning_rate": 4.674112814944214e-06,
      "loss": 0.6455,
      "step": 1541
    },
    {
      "epoch": 0.5216508795669824,
      "grad_norm": 0.29811638593673706,
      "learning_rate": 4.6736704613150295e-06,
      "loss": 0.6343,
      "step": 1542
    },
    {
      "epoch": 0.5219891745602165,
      "grad_norm": 0.30314648151397705,
      "learning_rate": 4.673227828629203e-06,
      "loss": 0.6222,
      "step": 1543
    },
    {
      "epoch": 0.5223274695534507,
      "grad_norm": 0.3316211402416229,
      "learning_rate": 4.672784916943562e-06,
      "loss": 0.6362,
      "step": 1544
    },
    {
      "epoch": 0.5226657645466847,
      "grad_norm": 0.31122979521751404,
      "learning_rate": 4.672341726314968e-06,
      "loss": 0.6484,
      "step": 1545
    },
    {
      "epoch": 0.5230040595399188,
      "grad_norm": 0.304100900888443,
      "learning_rate": 4.671898256800314e-06,
      "loss": 0.6546,
      "step": 1546
    },
    {
      "epoch": 0.5233423545331529,
      "grad_norm": 0.2922753691673279,
      "learning_rate": 4.671454508456536e-06,
      "loss": 0.6038,
      "step": 1547
    },
    {
      "epoch": 0.523680649526387,
      "grad_norm": 0.2911527454853058,
      "learning_rate": 4.671010481340603e-06,
      "loss": 0.6247,
      "step": 1548
    },
    {
      "epoch": 0.524018944519621,
      "grad_norm": 0.2970218360424042,
      "learning_rate": 4.670566175509518e-06,
      "loss": 0.6106,
      "step": 1549
    },
    {
      "epoch": 0.5243572395128552,
      "grad_norm": 0.3152008056640625,
      "learning_rate": 4.6701215910203225e-06,
      "loss": 0.6001,
      "step": 1550
    },
    {
      "epoch": 0.5246955345060893,
      "grad_norm": 0.3008357286453247,
      "learning_rate": 4.66967672793009e-06,
      "loss": 0.6501,
      "step": 1551
    },
    {
      "epoch": 0.5250338294993234,
      "grad_norm": 0.300070196390152,
      "learning_rate": 4.669231586295934e-06,
      "loss": 0.6292,
      "step": 1552
    },
    {
      "epoch": 0.5253721244925575,
      "grad_norm": 0.3507230579853058,
      "learning_rate": 4.668786166175002e-06,
      "loss": 0.616,
      "step": 1553
    },
    {
      "epoch": 0.5257104194857916,
      "grad_norm": 0.2806701362133026,
      "learning_rate": 4.668340467624476e-06,
      "loss": 0.6118,
      "step": 1554
    },
    {
      "epoch": 0.5260487144790257,
      "grad_norm": 0.3331495225429535,
      "learning_rate": 4.667894490701577e-06,
      "loss": 0.6,
      "step": 1555
    },
    {
      "epoch": 0.5263870094722598,
      "grad_norm": 0.32306334376335144,
      "learning_rate": 4.6674482354635576e-06,
      "loss": 0.6329,
      "step": 1556
    },
    {
      "epoch": 0.5267253044654939,
      "grad_norm": 0.292290061712265,
      "learning_rate": 4.667001701967709e-06,
      "loss": 0.6428,
      "step": 1557
    },
    {
      "epoch": 0.527063599458728,
      "grad_norm": 0.28629037737846375,
      "learning_rate": 4.666554890271357e-06,
      "loss": 0.6499,
      "step": 1558
    },
    {
      "epoch": 0.5274018944519621,
      "grad_norm": 0.31190797686576843,
      "learning_rate": 4.666107800431865e-06,
      "loss": 0.6019,
      "step": 1559
    },
    {
      "epoch": 0.5277401894451962,
      "grad_norm": 0.35739222168922424,
      "learning_rate": 4.665660432506629e-06,
      "loss": 0.6221,
      "step": 1560
    },
    {
      "epoch": 0.5280784844384303,
      "grad_norm": 0.29792535305023193,
      "learning_rate": 4.665212786553083e-06,
      "loss": 0.6332,
      "step": 1561
    },
    {
      "epoch": 0.5284167794316644,
      "grad_norm": 0.3818718492984772,
      "learning_rate": 4.664764862628695e-06,
      "loss": 0.6064,
      "step": 1562
    },
    {
      "epoch": 0.5287550744248986,
      "grad_norm": 0.3109496533870697,
      "learning_rate": 4.664316660790971e-06,
      "loss": 0.649,
      "step": 1563
    },
    {
      "epoch": 0.5290933694181326,
      "grad_norm": 0.2961856424808502,
      "learning_rate": 4.6638681810974495e-06,
      "loss": 0.6346,
      "step": 1564
    },
    {
      "epoch": 0.5294316644113667,
      "grad_norm": 0.31769058108329773,
      "learning_rate": 4.663419423605708e-06,
      "loss": 0.6269,
      "step": 1565
    },
    {
      "epoch": 0.5297699594046008,
      "grad_norm": 0.30717143416404724,
      "learning_rate": 4.662970388373359e-06,
      "loss": 0.6222,
      "step": 1566
    },
    {
      "epoch": 0.5301082543978349,
      "grad_norm": 0.3136506676673889,
      "learning_rate": 4.662521075458048e-06,
      "loss": 0.6372,
      "step": 1567
    },
    {
      "epoch": 0.530446549391069,
      "grad_norm": 0.2984481751918793,
      "learning_rate": 4.662071484917458e-06,
      "loss": 0.6397,
      "step": 1568
    },
    {
      "epoch": 0.5307848443843031,
      "grad_norm": 0.30340978503227234,
      "learning_rate": 4.661621616809308e-06,
      "loss": 0.6271,
      "step": 1569
    },
    {
      "epoch": 0.5311231393775372,
      "grad_norm": 0.3092040717601776,
      "learning_rate": 4.661171471191352e-06,
      "loss": 0.6111,
      "step": 1570
    },
    {
      "epoch": 0.5314614343707713,
      "grad_norm": 0.2942693829536438,
      "learning_rate": 4.660721048121382e-06,
      "loss": 0.6371,
      "step": 1571
    },
    {
      "epoch": 0.5317997293640054,
      "grad_norm": 0.2909970283508301,
      "learning_rate": 4.66027034765722e-06,
      "loss": 0.6365,
      "step": 1572
    },
    {
      "epoch": 0.5321380243572396,
      "grad_norm": 0.29246649146080017,
      "learning_rate": 4.65981936985673e-06,
      "loss": 0.6155,
      "step": 1573
    },
    {
      "epoch": 0.5324763193504736,
      "grad_norm": 0.3133332133293152,
      "learning_rate": 4.659368114777807e-06,
      "loss": 0.6308,
      "step": 1574
    },
    {
      "epoch": 0.5328146143437077,
      "grad_norm": 0.3009655177593231,
      "learning_rate": 4.658916582478384e-06,
      "loss": 0.6051,
      "step": 1575
    },
    {
      "epoch": 0.5331529093369418,
      "grad_norm": 0.29506585001945496,
      "learning_rate": 4.658464773016428e-06,
      "loss": 0.6222,
      "step": 1576
    },
    {
      "epoch": 0.5334912043301759,
      "grad_norm": 0.2916436791419983,
      "learning_rate": 4.658012686449944e-06,
      "loss": 0.6388,
      "step": 1577
    },
    {
      "epoch": 0.5338294993234101,
      "grad_norm": 0.3032309412956238,
      "learning_rate": 4.657560322836969e-06,
      "loss": 0.6428,
      "step": 1578
    },
    {
      "epoch": 0.5341677943166441,
      "grad_norm": 0.3248575031757355,
      "learning_rate": 4.657107682235581e-06,
      "loss": 0.6429,
      "step": 1579
    },
    {
      "epoch": 0.5345060893098782,
      "grad_norm": 0.30649298429489136,
      "learning_rate": 4.6566547647038865e-06,
      "loss": 0.6281,
      "step": 1580
    },
    {
      "epoch": 0.5348443843031123,
      "grad_norm": 0.28190237283706665,
      "learning_rate": 4.656201570300033e-06,
      "loss": 0.6185,
      "step": 1581
    },
    {
      "epoch": 0.5351826792963464,
      "grad_norm": 0.310199111700058,
      "learning_rate": 4.655748099082201e-06,
      "loss": 0.6097,
      "step": 1582
    },
    {
      "epoch": 0.5355209742895806,
      "grad_norm": 0.30557340383529663,
      "learning_rate": 4.655294351108609e-06,
      "loss": 0.6575,
      "step": 1583
    },
    {
      "epoch": 0.5358592692828146,
      "grad_norm": 0.30597639083862305,
      "learning_rate": 4.654840326437508e-06,
      "loss": 0.6497,
      "step": 1584
    },
    {
      "epoch": 0.5361975642760487,
      "grad_norm": 0.2925478219985962,
      "learning_rate": 4.654386025127186e-06,
      "loss": 0.6135,
      "step": 1585
    },
    {
      "epoch": 0.5365358592692828,
      "grad_norm": 0.3115440905094147,
      "learning_rate": 4.6539314472359655e-06,
      "loss": 0.612,
      "step": 1586
    },
    {
      "epoch": 0.536874154262517,
      "grad_norm": 0.3000645935535431,
      "learning_rate": 4.653476592822207e-06,
      "loss": 0.6333,
      "step": 1587
    },
    {
      "epoch": 0.5372124492557511,
      "grad_norm": 0.34186699986457825,
      "learning_rate": 4.653021461944304e-06,
      "loss": 0.6348,
      "step": 1588
    },
    {
      "epoch": 0.5375507442489851,
      "grad_norm": 0.3043736517429352,
      "learning_rate": 4.652566054660686e-06,
      "loss": 0.6465,
      "step": 1589
    },
    {
      "epoch": 0.5378890392422192,
      "grad_norm": 0.30343347787857056,
      "learning_rate": 4.652110371029819e-06,
      "loss": 0.6419,
      "step": 1590
    },
    {
      "epoch": 0.5382273342354533,
      "grad_norm": 0.3004716634750366,
      "learning_rate": 4.651654411110205e-06,
      "loss": 0.6448,
      "step": 1591
    },
    {
      "epoch": 0.5385656292286874,
      "grad_norm": 0.2967135012149811,
      "learning_rate": 4.651198174960378e-06,
      "loss": 0.6284,
      "step": 1592
    },
    {
      "epoch": 0.5389039242219216,
      "grad_norm": 0.2954842150211334,
      "learning_rate": 4.65074166263891e-06,
      "loss": 0.6322,
      "step": 1593
    },
    {
      "epoch": 0.5392422192151556,
      "grad_norm": 0.2938835322856903,
      "learning_rate": 4.65028487420441e-06,
      "loss": 0.6332,
      "step": 1594
    },
    {
      "epoch": 0.5395805142083897,
      "grad_norm": 0.3059346079826355,
      "learning_rate": 4.64982780971552e-06,
      "loss": 0.6341,
      "step": 1595
    },
    {
      "epoch": 0.5399188092016238,
      "grad_norm": 0.31491121649742126,
      "learning_rate": 4.6493704692309175e-06,
      "loss": 0.6307,
      "step": 1596
    },
    {
      "epoch": 0.540257104194858,
      "grad_norm": 0.31054824590682983,
      "learning_rate": 4.648912852809317e-06,
      "loss": 0.6353,
      "step": 1597
    },
    {
      "epoch": 0.540595399188092,
      "grad_norm": 0.2955755591392517,
      "learning_rate": 4.648454960509466e-06,
      "loss": 0.6051,
      "step": 1598
    },
    {
      "epoch": 0.5409336941813261,
      "grad_norm": 0.3091103136539459,
      "learning_rate": 4.647996792390151e-06,
      "loss": 0.6422,
      "step": 1599
    },
    {
      "epoch": 0.5412719891745602,
      "grad_norm": 0.29176416993141174,
      "learning_rate": 4.647538348510189e-06,
      "loss": 0.658,
      "step": 1600
    },
    {
      "epoch": 0.5416102841677943,
      "grad_norm": 0.3021467626094818,
      "learning_rate": 4.647079628928437e-06,
      "loss": 0.6146,
      "step": 1601
    },
    {
      "epoch": 0.5419485791610285,
      "grad_norm": 0.29566964507102966,
      "learning_rate": 4.646620633703786e-06,
      "loss": 0.6163,
      "step": 1602
    },
    {
      "epoch": 0.5422868741542625,
      "grad_norm": 0.3090681731700897,
      "learning_rate": 4.64616136289516e-06,
      "loss": 0.638,
      "step": 1603
    },
    {
      "epoch": 0.5426251691474966,
      "grad_norm": 0.3095795214176178,
      "learning_rate": 4.645701816561523e-06,
      "loss": 0.6195,
      "step": 1604
    },
    {
      "epoch": 0.5429634641407307,
      "grad_norm": 0.3343175947666168,
      "learning_rate": 4.64524199476187e-06,
      "loss": 0.6299,
      "step": 1605
    },
    {
      "epoch": 0.5433017591339648,
      "grad_norm": 0.2911258041858673,
      "learning_rate": 4.644781897555233e-06,
      "loss": 0.6221,
      "step": 1606
    },
    {
      "epoch": 0.543640054127199,
      "grad_norm": 0.29694199562072754,
      "learning_rate": 4.644321525000681e-06,
      "loss": 0.6053,
      "step": 1607
    },
    {
      "epoch": 0.543978349120433,
      "grad_norm": 0.3226844370365143,
      "learning_rate": 4.643860877157314e-06,
      "loss": 0.6264,
      "step": 1608
    },
    {
      "epoch": 0.5443166441136671,
      "grad_norm": 0.308302104473114,
      "learning_rate": 4.643399954084274e-06,
      "loss": 0.6098,
      "step": 1609
    },
    {
      "epoch": 0.5446549391069012,
      "grad_norm": 0.33216387033462524,
      "learning_rate": 4.642938755840731e-06,
      "loss": 0.6029,
      "step": 1610
    },
    {
      "epoch": 0.5449932341001353,
      "grad_norm": 0.2923735976219177,
      "learning_rate": 4.642477282485896e-06,
      "loss": 0.6131,
      "step": 1611
    },
    {
      "epoch": 0.5453315290933695,
      "grad_norm": 0.3002014458179474,
      "learning_rate": 4.642015534079012e-06,
      "loss": 0.6412,
      "step": 1612
    },
    {
      "epoch": 0.5456698240866035,
      "grad_norm": 0.30318304896354675,
      "learning_rate": 4.641553510679358e-06,
      "loss": 0.62,
      "step": 1613
    },
    {
      "epoch": 0.5460081190798376,
      "grad_norm": 0.3148590326309204,
      "learning_rate": 4.64109121234625e-06,
      "loss": 0.6435,
      "step": 1614
    },
    {
      "epoch": 0.5463464140730717,
      "grad_norm": 0.29820773005485535,
      "learning_rate": 4.640628639139037e-06,
      "loss": 0.6331,
      "step": 1615
    },
    {
      "epoch": 0.5466847090663058,
      "grad_norm": 0.308205246925354,
      "learning_rate": 4.640165791117106e-06,
      "loss": 0.6775,
      "step": 1616
    },
    {
      "epoch": 0.54702300405954,
      "grad_norm": 0.3078860342502594,
      "learning_rate": 4.639702668339876e-06,
      "loss": 0.6472,
      "step": 1617
    },
    {
      "epoch": 0.547361299052774,
      "grad_norm": 0.3088330924510956,
      "learning_rate": 4.6392392708668035e-06,
      "loss": 0.623,
      "step": 1618
    },
    {
      "epoch": 0.5476995940460081,
      "grad_norm": 0.2907048463821411,
      "learning_rate": 4.638775598757379e-06,
      "loss": 0.6284,
      "step": 1619
    },
    {
      "epoch": 0.5480378890392422,
      "grad_norm": 0.31769198179244995,
      "learning_rate": 4.63831165207113e-06,
      "loss": 0.6483,
      "step": 1620
    },
    {
      "epoch": 0.5483761840324763,
      "grad_norm": 0.30527791380882263,
      "learning_rate": 4.637847430867617e-06,
      "loss": 0.644,
      "step": 1621
    },
    {
      "epoch": 0.5487144790257105,
      "grad_norm": 0.3103595972061157,
      "learning_rate": 4.6373829352064375e-06,
      "loss": 0.6223,
      "step": 1622
    },
    {
      "epoch": 0.5490527740189445,
      "grad_norm": 0.2978458106517792,
      "learning_rate": 4.636918165147224e-06,
      "loss": 0.6252,
      "step": 1623
    },
    {
      "epoch": 0.5493910690121786,
      "grad_norm": 0.3217633366584778,
      "learning_rate": 4.6364531207496426e-06,
      "loss": 0.6436,
      "step": 1624
    },
    {
      "epoch": 0.5497293640054127,
      "grad_norm": 0.2983613610267639,
      "learning_rate": 4.635987802073397e-06,
      "loss": 0.6162,
      "step": 1625
    },
    {
      "epoch": 0.5500676589986468,
      "grad_norm": 0.297369122505188,
      "learning_rate": 4.6355222091782245e-06,
      "loss": 0.6494,
      "step": 1626
    },
    {
      "epoch": 0.550405953991881,
      "grad_norm": 0.2905776798725128,
      "learning_rate": 4.635056342123899e-06,
      "loss": 0.6155,
      "step": 1627
    },
    {
      "epoch": 0.550744248985115,
      "grad_norm": 0.3083232045173645,
      "learning_rate": 4.634590200970228e-06,
      "loss": 0.6471,
      "step": 1628
    },
    {
      "epoch": 0.5510825439783491,
      "grad_norm": 0.30626025795936584,
      "learning_rate": 4.634123785777054e-06,
      "loss": 0.6376,
      "step": 1629
    },
    {
      "epoch": 0.5514208389715832,
      "grad_norm": 0.29137128591537476,
      "learning_rate": 4.633657096604256e-06,
      "loss": 0.6381,
      "step": 1630
    },
    {
      "epoch": 0.5517591339648173,
      "grad_norm": 0.29958054423332214,
      "learning_rate": 4.633190133511749e-06,
      "loss": 0.642,
      "step": 1631
    },
    {
      "epoch": 0.5520974289580515,
      "grad_norm": 0.317205548286438,
      "learning_rate": 4.632722896559481e-06,
      "loss": 0.6412,
      "step": 1632
    },
    {
      "epoch": 0.5524357239512855,
      "grad_norm": 0.328844279050827,
      "learning_rate": 4.632255385807436e-06,
      "loss": 0.625,
      "step": 1633
    },
    {
      "epoch": 0.5527740189445196,
      "grad_norm": 0.3044576644897461,
      "learning_rate": 4.631787601315633e-06,
      "loss": 0.6228,
      "step": 1634
    },
    {
      "epoch": 0.5531123139377537,
      "grad_norm": 0.3005305230617523,
      "learning_rate": 4.631319543144127e-06,
      "loss": 0.6022,
      "step": 1635
    },
    {
      "epoch": 0.5534506089309879,
      "grad_norm": 0.3218366801738739,
      "learning_rate": 4.630851211353007e-06,
      "loss": 0.6252,
      "step": 1636
    },
    {
      "epoch": 0.553788903924222,
      "grad_norm": 0.3036673069000244,
      "learning_rate": 4.6303826060023975e-06,
      "loss": 0.6553,
      "step": 1637
    },
    {
      "epoch": 0.554127198917456,
      "grad_norm": 0.2918086349964142,
      "learning_rate": 4.629913727152459e-06,
      "loss": 0.6138,
      "step": 1638
    },
    {
      "epoch": 0.5544654939106901,
      "grad_norm": 0.29163315892219543,
      "learning_rate": 4.629444574863385e-06,
      "loss": 0.6228,
      "step": 1639
    },
    {
      "epoch": 0.5548037889039242,
      "grad_norm": 0.31965193152427673,
      "learning_rate": 4.6289751491954076e-06,
      "loss": 0.6039,
      "step": 1640
    },
    {
      "epoch": 0.5551420838971584,
      "grad_norm": 0.313843309879303,
      "learning_rate": 4.628505450208789e-06,
      "loss": 0.6416,
      "step": 1641
    },
    {
      "epoch": 0.5554803788903924,
      "grad_norm": 0.29786884784698486,
      "learning_rate": 4.628035477963832e-06,
      "loss": 0.6026,
      "step": 1642
    },
    {
      "epoch": 0.5558186738836265,
      "grad_norm": 0.30769261717796326,
      "learning_rate": 4.627565232520871e-06,
      "loss": 0.6294,
      "step": 1643
    },
    {
      "epoch": 0.5561569688768606,
      "grad_norm": 0.3097287714481354,
      "learning_rate": 4.627094713940274e-06,
      "loss": 0.6159,
      "step": 1644
    },
    {
      "epoch": 0.5564952638700947,
      "grad_norm": 0.30283322930336,
      "learning_rate": 4.626623922282451e-06,
      "loss": 0.628,
      "step": 1645
    },
    {
      "epoch": 0.5568335588633289,
      "grad_norm": 0.3172987699508667,
      "learning_rate": 4.626152857607837e-06,
      "loss": 0.6285,
      "step": 1646
    },
    {
      "epoch": 0.5571718538565629,
      "grad_norm": 0.3280118703842163,
      "learning_rate": 4.625681519976912e-06,
      "loss": 0.6326,
      "step": 1647
    },
    {
      "epoch": 0.557510148849797,
      "grad_norm": 0.30398502945899963,
      "learning_rate": 4.625209909450183e-06,
      "loss": 0.6218,
      "step": 1648
    },
    {
      "epoch": 0.5578484438430311,
      "grad_norm": 0.34335753321647644,
      "learning_rate": 4.6247380260881995e-06,
      "loss": 0.5902,
      "step": 1649
    },
    {
      "epoch": 0.5581867388362652,
      "grad_norm": 0.2943098843097687,
      "learning_rate": 4.624265869951539e-06,
      "loss": 0.6408,
      "step": 1650
    },
    {
      "epoch": 0.5585250338294994,
      "grad_norm": 0.31536927819252014,
      "learning_rate": 4.623793441100816e-06,
      "loss": 0.6236,
      "step": 1651
    },
    {
      "epoch": 0.5588633288227334,
      "grad_norm": 0.3096260130405426,
      "learning_rate": 4.623320739596685e-06,
      "loss": 0.6412,
      "step": 1652
    },
    {
      "epoch": 0.5592016238159675,
      "grad_norm": 0.3125949501991272,
      "learning_rate": 4.6228477654998294e-06,
      "loss": 0.6222,
      "step": 1653
    },
    {
      "epoch": 0.5595399188092016,
      "grad_norm": 0.34240230917930603,
      "learning_rate": 4.622374518870969e-06,
      "loss": 0.6293,
      "step": 1654
    },
    {
      "epoch": 0.5598782138024357,
      "grad_norm": 0.30551519989967346,
      "learning_rate": 4.621900999770862e-06,
      "loss": 0.6609,
      "step": 1655
    },
    {
      "epoch": 0.5602165087956699,
      "grad_norm": 0.31002277135849,
      "learning_rate": 4.621427208260296e-06,
      "loss": 0.5994,
      "step": 1656
    },
    {
      "epoch": 0.5605548037889039,
      "grad_norm": 0.2998320460319519,
      "learning_rate": 4.6209531444000975e-06,
      "loss": 0.6245,
      "step": 1657
    },
    {
      "epoch": 0.560893098782138,
      "grad_norm": 0.32024434208869934,
      "learning_rate": 4.620478808251129e-06,
      "loss": 0.6083,
      "step": 1658
    },
    {
      "epoch": 0.5612313937753721,
      "grad_norm": 0.2951360046863556,
      "learning_rate": 4.620004199874283e-06,
      "loss": 0.6236,
      "step": 1659
    },
    {
      "epoch": 0.5615696887686062,
      "grad_norm": 0.29551962018013,
      "learning_rate": 4.619529319330492e-06,
      "loss": 0.6338,
      "step": 1660
    },
    {
      "epoch": 0.5619079837618404,
      "grad_norm": 0.29304054379463196,
      "learning_rate": 4.619054166680718e-06,
      "loss": 0.6175,
      "step": 1661
    },
    {
      "epoch": 0.5622462787550744,
      "grad_norm": 0.31250274181365967,
      "learning_rate": 4.6185787419859675e-06,
      "loss": 0.651,
      "step": 1662
    },
    {
      "epoch": 0.5625845737483085,
      "grad_norm": 0.3109738528728485,
      "learning_rate": 4.6181030453072695e-06,
      "loss": 0.6302,
      "step": 1663
    },
    {
      "epoch": 0.5629228687415426,
      "grad_norm": 0.2980029284954071,
      "learning_rate": 4.617627076705697e-06,
      "loss": 0.6325,
      "step": 1664
    },
    {
      "epoch": 0.5632611637347767,
      "grad_norm": 0.28963977098464966,
      "learning_rate": 4.617150836242356e-06,
      "loss": 0.6178,
      "step": 1665
    },
    {
      "epoch": 0.5635994587280109,
      "grad_norm": 0.2933375835418701,
      "learning_rate": 4.616674323978384e-06,
      "loss": 0.6361,
      "step": 1666
    },
    {
      "epoch": 0.5639377537212449,
      "grad_norm": 0.2948474884033203,
      "learning_rate": 4.616197539974957e-06,
      "loss": 0.6377,
      "step": 1667
    },
    {
      "epoch": 0.564276048714479,
      "grad_norm": 0.29074469208717346,
      "learning_rate": 4.615720484293286e-06,
      "loss": 0.6263,
      "step": 1668
    },
    {
      "epoch": 0.5646143437077131,
      "grad_norm": 0.28231802582740784,
      "learning_rate": 4.615243156994613e-06,
      "loss": 0.6354,
      "step": 1669
    },
    {
      "epoch": 0.5649526387009473,
      "grad_norm": 0.3019883632659912,
      "learning_rate": 4.614765558140219e-06,
      "loss": 0.6089,
      "step": 1670
    },
    {
      "epoch": 0.5652909336941814,
      "grad_norm": 0.29071640968322754,
      "learning_rate": 4.6142876877914175e-06,
      "loss": 0.6215,
      "step": 1671
    },
    {
      "epoch": 0.5656292286874154,
      "grad_norm": 0.2982822060585022,
      "learning_rate": 4.613809546009558e-06,
      "loss": 0.6117,
      "step": 1672
    },
    {
      "epoch": 0.5659675236806495,
      "grad_norm": 0.29466015100479126,
      "learning_rate": 4.613331132856026e-06,
      "loss": 0.6109,
      "step": 1673
    },
    {
      "epoch": 0.5663058186738836,
      "grad_norm": 0.2973395884037018,
      "learning_rate": 4.612852448392238e-06,
      "loss": 0.615,
      "step": 1674
    },
    {
      "epoch": 0.5666441136671178,
      "grad_norm": 0.3011510968208313,
      "learning_rate": 4.612373492679649e-06,
      "loss": 0.644,
      "step": 1675
    },
    {
      "epoch": 0.5669824086603519,
      "grad_norm": 0.2894744575023651,
      "learning_rate": 4.611894265779748e-06,
      "loss": 0.6086,
      "step": 1676
    },
    {
      "epoch": 0.5673207036535859,
      "grad_norm": 0.2948070466518402,
      "learning_rate": 4.611414767754057e-06,
      "loss": 0.6314,
      "step": 1677
    },
    {
      "epoch": 0.56765899864682,
      "grad_norm": 0.2949257493019104,
      "learning_rate": 4.610934998664134e-06,
      "loss": 0.6324,
      "step": 1678
    },
    {
      "epoch": 0.5679972936400541,
      "grad_norm": 0.30842041969299316,
      "learning_rate": 4.610454958571574e-06,
      "loss": 0.6016,
      "step": 1679
    },
    {
      "epoch": 0.5683355886332883,
      "grad_norm": 0.28693491220474243,
      "learning_rate": 4.609974647538003e-06,
      "loss": 0.6331,
      "step": 1680
    },
    {
      "epoch": 0.5686738836265224,
      "grad_norm": 0.2898673415184021,
      "learning_rate": 4.6094940656250845e-06,
      "loss": 0.633,
      "step": 1681
    },
    {
      "epoch": 0.5690121786197564,
      "grad_norm": 0.28437864780426025,
      "learning_rate": 4.6090132128945146e-06,
      "loss": 0.5995,
      "step": 1682
    },
    {
      "epoch": 0.5693504736129905,
      "grad_norm": 0.36713582277297974,
      "learning_rate": 4.608532089408026e-06,
      "loss": 0.5904,
      "step": 1683
    },
    {
      "epoch": 0.5696887686062246,
      "grad_norm": 0.31164711713790894,
      "learning_rate": 4.6080506952273855e-06,
      "loss": 0.6251,
      "step": 1684
    },
    {
      "epoch": 0.5700270635994588,
      "grad_norm": 0.31584176421165466,
      "learning_rate": 4.607569030414394e-06,
      "loss": 0.6547,
      "step": 1685
    },
    {
      "epoch": 0.5703653585926928,
      "grad_norm": 0.34662458300590515,
      "learning_rate": 4.60708709503089e-06,
      "loss": 0.6161,
      "step": 1686
    },
    {
      "epoch": 0.5707036535859269,
      "grad_norm": 0.30349597334861755,
      "learning_rate": 4.606604889138742e-06,
      "loss": 0.6173,
      "step": 1687
    },
    {
      "epoch": 0.571041948579161,
      "grad_norm": 0.32037651538848877,
      "learning_rate": 4.606122412799857e-06,
      "loss": 0.6024,
      "step": 1688
    },
    {
      "epoch": 0.5713802435723951,
      "grad_norm": 0.3367273211479187,
      "learning_rate": 4.6056396660761755e-06,
      "loss": 0.6313,
      "step": 1689
    },
    {
      "epoch": 0.5717185385656293,
      "grad_norm": 0.3002927601337433,
      "learning_rate": 4.6051566490296726e-06,
      "loss": 0.6454,
      "step": 1690
    },
    {
      "epoch": 0.5720568335588633,
      "grad_norm": 0.32096385955810547,
      "learning_rate": 4.604673361722358e-06,
      "loss": 0.6168,
      "step": 1691
    },
    {
      "epoch": 0.5723951285520974,
      "grad_norm": 0.3034731149673462,
      "learning_rate": 4.604189804216276e-06,
      "loss": 0.6364,
      "step": 1692
    },
    {
      "epoch": 0.5727334235453315,
      "grad_norm": 0.29405683279037476,
      "learning_rate": 4.603705976573507e-06,
      "loss": 0.6073,
      "step": 1693
    },
    {
      "epoch": 0.5730717185385656,
      "grad_norm": 0.2853470742702484,
      "learning_rate": 4.603221878856164e-06,
      "loss": 0.6218,
      "step": 1694
    },
    {
      "epoch": 0.5734100135317998,
      "grad_norm": 0.3522389829158783,
      "learning_rate": 4.602737511126396e-06,
      "loss": 0.6441,
      "step": 1695
    },
    {
      "epoch": 0.5737483085250338,
      "grad_norm": 0.32325372099876404,
      "learning_rate": 4.602252873446386e-06,
      "loss": 0.664,
      "step": 1696
    },
    {
      "epoch": 0.5740866035182679,
      "grad_norm": 0.2904968857765198,
      "learning_rate": 4.601767965878352e-06,
      "loss": 0.6285,
      "step": 1697
    },
    {
      "epoch": 0.574424898511502,
      "grad_norm": 0.297441691160202,
      "learning_rate": 4.601282788484548e-06,
      "loss": 0.6422,
      "step": 1698
    },
    {
      "epoch": 0.5747631935047361,
      "grad_norm": 0.3208168148994446,
      "learning_rate": 4.60079734132726e-06,
      "loss": 0.6446,
      "step": 1699
    },
    {
      "epoch": 0.5751014884979703,
      "grad_norm": 0.30540528893470764,
      "learning_rate": 4.6003116244688095e-06,
      "loss": 0.6093,
      "step": 1700
    },
    {
      "epoch": 0.5754397834912043,
      "grad_norm": 0.2912519872188568,
      "learning_rate": 4.599825637971554e-06,
      "loss": 0.6016,
      "step": 1701
    },
    {
      "epoch": 0.5757780784844384,
      "grad_norm": 0.30942586064338684,
      "learning_rate": 4.599339381897883e-06,
      "loss": 0.6199,
      "step": 1702
    },
    {
      "epoch": 0.5761163734776725,
      "grad_norm": 0.2984882891178131,
      "learning_rate": 4.598852856310224e-06,
      "loss": 0.6337,
      "step": 1703
    },
    {
      "epoch": 0.5764546684709067,
      "grad_norm": 0.28245341777801514,
      "learning_rate": 4.5983660612710365e-06,
      "loss": 0.6185,
      "step": 1704
    },
    {
      "epoch": 0.5767929634641408,
      "grad_norm": 0.29131874442100525,
      "learning_rate": 4.5978789968428165e-06,
      "loss": 0.6113,
      "step": 1705
    },
    {
      "epoch": 0.5771312584573748,
      "grad_norm": 0.3047155439853668,
      "learning_rate": 4.5973916630880924e-06,
      "loss": 0.587,
      "step": 1706
    },
    {
      "epoch": 0.5774695534506089,
      "grad_norm": 0.2988587021827698,
      "learning_rate": 4.596904060069428e-06,
      "loss": 0.6216,
      "step": 1707
    },
    {
      "epoch": 0.577807848443843,
      "grad_norm": 0.28836652636528015,
      "learning_rate": 4.5964161878494225e-06,
      "loss": 0.635,
      "step": 1708
    },
    {
      "epoch": 0.5781461434370772,
      "grad_norm": 0.2934615910053253,
      "learning_rate": 4.59592804649071e-06,
      "loss": 0.5957,
      "step": 1709
    },
    {
      "epoch": 0.5784844384303113,
      "grad_norm": 0.2974570691585541,
      "learning_rate": 4.595439636055957e-06,
      "loss": 0.6348,
      "step": 1710
    },
    {
      "epoch": 0.5788227334235453,
      "grad_norm": 0.28336742520332336,
      "learning_rate": 4.594950956607867e-06,
      "loss": 0.6351,
      "step": 1711
    },
    {
      "epoch": 0.5791610284167794,
      "grad_norm": 0.29395925998687744,
      "learning_rate": 4.594462008209175e-06,
      "loss": 0.623,
      "step": 1712
    },
    {
      "epoch": 0.5794993234100135,
      "grad_norm": 0.28559985756874084,
      "learning_rate": 4.593972790922654e-06,
      "loss": 0.6327,
      "step": 1713
    },
    {
      "epoch": 0.5798376184032477,
      "grad_norm": 0.28623858094215393,
      "learning_rate": 4.593483304811109e-06,
      "loss": 0.6377,
      "step": 1714
    },
    {
      "epoch": 0.5801759133964818,
      "grad_norm": 0.3167601525783539,
      "learning_rate": 4.592993549937381e-06,
      "loss": 0.6398,
      "step": 1715
    },
    {
      "epoch": 0.5805142083897158,
      "grad_norm": 0.30663159489631653,
      "learning_rate": 4.592503526364344e-06,
      "loss": 0.641,
      "step": 1716
    },
    {
      "epoch": 0.5808525033829499,
      "grad_norm": 0.2943854331970215,
      "learning_rate": 4.592013234154909e-06,
      "loss": 0.6207,
      "step": 1717
    },
    {
      "epoch": 0.581190798376184,
      "grad_norm": 0.29026129841804504,
      "learning_rate": 4.591522673372019e-06,
      "loss": 0.618,
      "step": 1718
    },
    {
      "epoch": 0.5815290933694182,
      "grad_norm": 0.28923630714416504,
      "learning_rate": 4.591031844078653e-06,
      "loss": 0.6366,
      "step": 1719
    },
    {
      "epoch": 0.5818673883626523,
      "grad_norm": 0.29991474747657776,
      "learning_rate": 4.590540746337823e-06,
      "loss": 0.6473,
      "step": 1720
    },
    {
      "epoch": 0.5822056833558863,
      "grad_norm": 0.2991500496864319,
      "learning_rate": 4.590049380212576e-06,
      "loss": 0.5944,
      "step": 1721
    },
    {
      "epoch": 0.5825439783491204,
      "grad_norm": 0.2879493832588196,
      "learning_rate": 4.589557745765994e-06,
      "loss": 0.6345,
      "step": 1722
    },
    {
      "epoch": 0.5828822733423545,
      "grad_norm": 0.31147581338882446,
      "learning_rate": 4.5890658430611944e-06,
      "loss": 0.6371,
      "step": 1723
    },
    {
      "epoch": 0.5832205683355887,
      "grad_norm": 0.3145962059497833,
      "learning_rate": 4.588573672161326e-06,
      "loss": 0.6414,
      "step": 1724
    },
    {
      "epoch": 0.5835588633288228,
      "grad_norm": 0.3050592243671417,
      "learning_rate": 4.588081233129576e-06,
      "loss": 0.6349,
      "step": 1725
    },
    {
      "epoch": 0.5838971583220568,
      "grad_norm": 0.3005061745643616,
      "learning_rate": 4.5875885260291625e-06,
      "loss": 0.637,
      "step": 1726
    },
    {
      "epoch": 0.5842354533152909,
      "grad_norm": 0.3005605638027191,
      "learning_rate": 4.5870955509233394e-06,
      "loss": 0.6203,
      "step": 1727
    },
    {
      "epoch": 0.584573748308525,
      "grad_norm": 0.28691211342811584,
      "learning_rate": 4.586602307875396e-06,
      "loss": 0.6112,
      "step": 1728
    },
    {
      "epoch": 0.5849120433017592,
      "grad_norm": 0.292736291885376,
      "learning_rate": 4.586108796948654e-06,
      "loss": 0.6356,
      "step": 1729
    },
    {
      "epoch": 0.5852503382949933,
      "grad_norm": 0.31532350182533264,
      "learning_rate": 4.585615018206471e-06,
      "loss": 0.6378,
      "step": 1730
    },
    {
      "epoch": 0.5855886332882273,
      "grad_norm": 0.31929609179496765,
      "learning_rate": 4.585120971712239e-06,
      "loss": 0.6412,
      "step": 1731
    },
    {
      "epoch": 0.5859269282814614,
      "grad_norm": 0.30665725469589233,
      "learning_rate": 4.584626657529382e-06,
      "loss": 0.6164,
      "step": 1732
    },
    {
      "epoch": 0.5862652232746955,
      "grad_norm": 0.29724955558776855,
      "learning_rate": 4.584132075721362e-06,
      "loss": 0.614,
      "step": 1733
    },
    {
      "epoch": 0.5866035182679297,
      "grad_norm": 0.32470571994781494,
      "learning_rate": 4.5836372263516734e-06,
      "loss": 0.6302,
      "step": 1734
    },
    {
      "epoch": 0.5869418132611637,
      "grad_norm": 0.31625714898109436,
      "learning_rate": 4.583142109483846e-06,
      "loss": 0.6166,
      "step": 1735
    },
    {
      "epoch": 0.5872801082543978,
      "grad_norm": 0.28975987434387207,
      "learning_rate": 4.582646725181441e-06,
      "loss": 0.6248,
      "step": 1736
    },
    {
      "epoch": 0.5876184032476319,
      "grad_norm": 0.31003639101982117,
      "learning_rate": 4.582151073508057e-06,
      "loss": 0.61,
      "step": 1737
    },
    {
      "epoch": 0.587956698240866,
      "grad_norm": 0.2950305640697479,
      "learning_rate": 4.581655154527326e-06,
      "loss": 0.6081,
      "step": 1738
    },
    {
      "epoch": 0.5882949932341002,
      "grad_norm": 0.309980183839798,
      "learning_rate": 4.581158968302913e-06,
      "loss": 0.6307,
      "step": 1739
    },
    {
      "epoch": 0.5886332882273342,
      "grad_norm": 0.2961738109588623,
      "learning_rate": 4.580662514898522e-06,
      "loss": 0.635,
      "step": 1740
    },
    {
      "epoch": 0.5889715832205683,
      "grad_norm": 0.29777249693870544,
      "learning_rate": 4.580165794377884e-06,
      "loss": 0.6219,
      "step": 1741
    },
    {
      "epoch": 0.5893098782138024,
      "grad_norm": 0.2928023338317871,
      "learning_rate": 4.57966880680477e-06,
      "loss": 0.6453,
      "step": 1742
    },
    {
      "epoch": 0.5896481732070366,
      "grad_norm": 0.3013555109500885,
      "learning_rate": 4.579171552242984e-06,
      "loss": 0.6127,
      "step": 1743
    },
    {
      "epoch": 0.5899864682002707,
      "grad_norm": 0.30283522605895996,
      "learning_rate": 4.578674030756364e-06,
      "loss": 0.6303,
      "step": 1744
    },
    {
      "epoch": 0.5903247631935047,
      "grad_norm": 0.3043537735939026,
      "learning_rate": 4.5781762424087794e-06,
      "loss": 0.6316,
      "step": 1745
    },
    {
      "epoch": 0.5906630581867388,
      "grad_norm": 0.2953866422176361,
      "learning_rate": 4.577678187264139e-06,
      "loss": 0.6209,
      "step": 1746
    },
    {
      "epoch": 0.5910013531799729,
      "grad_norm": 0.2934488356113434,
      "learning_rate": 4.577179865386382e-06,
      "loss": 0.614,
      "step": 1747
    },
    {
      "epoch": 0.591339648173207,
      "grad_norm": 0.30841347575187683,
      "learning_rate": 4.576681276839483e-06,
      "loss": 0.5964,
      "step": 1748
    },
    {
      "epoch": 0.5916779431664412,
      "grad_norm": 0.29147687554359436,
      "learning_rate": 4.576182421687452e-06,
      "loss": 0.6147,
      "step": 1749
    },
    {
      "epoch": 0.5920162381596752,
      "grad_norm": 0.3026900589466095,
      "learning_rate": 4.5756832999943326e-06,
      "loss": 0.641,
      "step": 1750
    },
    {
      "epoch": 0.5923545331529093,
      "grad_norm": 0.3127942681312561,
      "learning_rate": 4.5751839118242e-06,
      "loss": 0.6295,
      "step": 1751
    },
    {
      "epoch": 0.5926928281461434,
      "grad_norm": 0.2994759976863861,
      "learning_rate": 4.5746842572411685e-06,
      "loss": 0.6025,
      "step": 1752
    },
    {
      "epoch": 0.5930311231393776,
      "grad_norm": 0.3004252016544342,
      "learning_rate": 4.5741843363093814e-06,
      "loss": 0.6157,
      "step": 1753
    },
    {
      "epoch": 0.5933694181326117,
      "grad_norm": 0.2946246862411499,
      "learning_rate": 4.573684149093021e-06,
      "loss": 0.6033,
      "step": 1754
    },
    {
      "epoch": 0.5937077131258457,
      "grad_norm": 0.30164048075675964,
      "learning_rate": 4.5731836956563e-06,
      "loss": 0.6602,
      "step": 1755
    },
    {
      "epoch": 0.5940460081190798,
      "grad_norm": 0.3036515414714813,
      "learning_rate": 4.5726829760634685e-06,
      "loss": 0.6136,
      "step": 1756
    },
    {
      "epoch": 0.5943843031123139,
      "grad_norm": 0.29374492168426514,
      "learning_rate": 4.572181990378807e-06,
      "loss": 0.6153,
      "step": 1757
    },
    {
      "epoch": 0.5947225981055481,
      "grad_norm": 0.2989799380302429,
      "learning_rate": 4.571680738666633e-06,
      "loss": 0.6277,
      "step": 1758
    },
    {
      "epoch": 0.5950608930987822,
      "grad_norm": 0.31341370940208435,
      "learning_rate": 4.571179220991298e-06,
      "loss": 0.6217,
      "step": 1759
    },
    {
      "epoch": 0.5953991880920162,
      "grad_norm": 0.298629492521286,
      "learning_rate": 4.570677437417186e-06,
      "loss": 0.6235,
      "step": 1760
    },
    {
      "epoch": 0.5957374830852503,
      "grad_norm": 0.29496222734451294,
      "learning_rate": 4.570175388008717e-06,
      "loss": 0.6526,
      "step": 1761
    },
    {
      "epoch": 0.5960757780784844,
      "grad_norm": 0.2959974408149719,
      "learning_rate": 4.569673072830344e-06,
      "loss": 0.629,
      "step": 1762
    },
    {
      "epoch": 0.5964140730717186,
      "grad_norm": 0.29397693276405334,
      "learning_rate": 4.569170491946554e-06,
      "loss": 0.6256,
      "step": 1763
    },
    {
      "epoch": 0.5967523680649527,
      "grad_norm": 0.29911816120147705,
      "learning_rate": 4.56866764542187e-06,
      "loss": 0.6418,
      "step": 1764
    },
    {
      "epoch": 0.5970906630581867,
      "grad_norm": 0.30201956629753113,
      "learning_rate": 4.568164533320846e-06,
      "loss": 0.6336,
      "step": 1765
    },
    {
      "epoch": 0.5974289580514208,
      "grad_norm": 0.3169977366924286,
      "learning_rate": 4.567661155708072e-06,
      "loss": 0.6216,
      "step": 1766
    },
    {
      "epoch": 0.597767253044655,
      "grad_norm": 0.2975189983844757,
      "learning_rate": 4.567157512648173e-06,
      "loss": 0.621,
      "step": 1767
    },
    {
      "epoch": 0.5981055480378891,
      "grad_norm": 0.3022186756134033,
      "learning_rate": 4.566653604205805e-06,
      "loss": 0.6125,
      "step": 1768
    },
    {
      "epoch": 0.5984438430311232,
      "grad_norm": 0.3099009692668915,
      "learning_rate": 4.566149430445662e-06,
      "loss": 0.6065,
      "step": 1769
    },
    {
      "epoch": 0.5987821380243572,
      "grad_norm": 0.3135291635990143,
      "learning_rate": 4.565644991432468e-06,
      "loss": 0.6071,
      "step": 1770
    },
    {
      "epoch": 0.5991204330175913,
      "grad_norm": 0.3010227084159851,
      "learning_rate": 4.565140287230985e-06,
      "loss": 0.6388,
      "step": 1771
    },
    {
      "epoch": 0.5994587280108254,
      "grad_norm": 0.2999197840690613,
      "learning_rate": 4.564635317906006e-06,
      "loss": 0.6355,
      "step": 1772
    },
    {
      "epoch": 0.5997970230040596,
      "grad_norm": 0.29599177837371826,
      "learning_rate": 4.564130083522359e-06,
      "loss": 0.633,
      "step": 1773
    },
    {
      "epoch": 0.6001353179972937,
      "grad_norm": 0.3240567743778229,
      "learning_rate": 4.563624584144907e-06,
      "loss": 0.6322,
      "step": 1774
    },
    {
      "epoch": 0.6004736129905277,
      "grad_norm": 0.3124981224536896,
      "learning_rate": 4.563118819838545e-06,
      "loss": 0.6321,
      "step": 1775
    },
    {
      "epoch": 0.6008119079837618,
      "grad_norm": 0.29879269003868103,
      "learning_rate": 4.562612790668204e-06,
      "loss": 0.6083,
      "step": 1776
    },
    {
      "epoch": 0.601150202976996,
      "grad_norm": 0.2953958213329315,
      "learning_rate": 4.5621064966988485e-06,
      "loss": 0.6133,
      "step": 1777
    },
    {
      "epoch": 0.6014884979702301,
      "grad_norm": 0.2944336533546448,
      "learning_rate": 4.561599937995477e-06,
      "loss": 0.6442,
      "step": 1778
    },
    {
      "epoch": 0.6018267929634641,
      "grad_norm": 0.30257466435432434,
      "learning_rate": 4.561093114623119e-06,
      "loss": 0.6165,
      "step": 1779
    },
    {
      "epoch": 0.6021650879566982,
      "grad_norm": 0.28953441977500916,
      "learning_rate": 4.560586026646845e-06,
      "loss": 0.6142,
      "step": 1780
    },
    {
      "epoch": 0.6025033829499323,
      "grad_norm": 0.30395811796188354,
      "learning_rate": 4.560078674131752e-06,
      "loss": 0.6217,
      "step": 1781
    },
    {
      "epoch": 0.6028416779431665,
      "grad_norm": 0.28763335943222046,
      "learning_rate": 4.559571057142975e-06,
      "loss": 0.5914,
      "step": 1782
    },
    {
      "epoch": 0.6031799729364006,
      "grad_norm": 0.3176610767841339,
      "learning_rate": 4.559063175745682e-06,
      "loss": 0.6282,
      "step": 1783
    },
    {
      "epoch": 0.6035182679296346,
      "grad_norm": 0.2906036674976349,
      "learning_rate": 4.558555030005075e-06,
      "loss": 0.616,
      "step": 1784
    },
    {
      "epoch": 0.6038565629228687,
      "grad_norm": 0.2957024872303009,
      "learning_rate": 4.55804661998639e-06,
      "loss": 0.651,
      "step": 1785
    },
    {
      "epoch": 0.6041948579161028,
      "grad_norm": 0.30133163928985596,
      "learning_rate": 4.557537945754898e-06,
      "loss": 0.6547,
      "step": 1786
    },
    {
      "epoch": 0.604533152909337,
      "grad_norm": 0.3015631139278412,
      "learning_rate": 4.5570290073759e-06,
      "loss": 0.6307,
      "step": 1787
    },
    {
      "epoch": 0.6048714479025711,
      "grad_norm": 0.28448137640953064,
      "learning_rate": 4.556519804914736e-06,
      "loss": 0.6074,
      "step": 1788
    },
    {
      "epoch": 0.6052097428958051,
      "grad_norm": 0.286736398935318,
      "learning_rate": 4.556010338436777e-06,
      "loss": 0.6448,
      "step": 1789
    },
    {
      "epoch": 0.6055480378890392,
      "grad_norm": 0.30498284101486206,
      "learning_rate": 4.555500608007428e-06,
      "loss": 0.6177,
      "step": 1790
    },
    {
      "epoch": 0.6058863328822733,
      "grad_norm": 0.32178056240081787,
      "learning_rate": 4.55499061369213e-06,
      "loss": 0.6371,
      "step": 1791
    },
    {
      "epoch": 0.6062246278755075,
      "grad_norm": 0.2903531491756439,
      "learning_rate": 4.554480355556354e-06,
      "loss": 0.6149,
      "step": 1792
    },
    {
      "epoch": 0.6065629228687416,
      "grad_norm": 0.30403220653533936,
      "learning_rate": 4.553969833665609e-06,
      "loss": 0.6309,
      "step": 1793
    },
    {
      "epoch": 0.6069012178619756,
      "grad_norm": 0.3077298104763031,
      "learning_rate": 4.553459048085435e-06,
      "loss": 0.5958,
      "step": 1794
    },
    {
      "epoch": 0.6072395128552097,
      "grad_norm": 0.292605459690094,
      "learning_rate": 4.552947998881407e-06,
      "loss": 0.6282,
      "step": 1795
    },
    {
      "epoch": 0.6075778078484438,
      "grad_norm": 0.29466307163238525,
      "learning_rate": 4.5524366861191345e-06,
      "loss": 0.6057,
      "step": 1796
    },
    {
      "epoch": 0.607916102841678,
      "grad_norm": 0.29204457998275757,
      "learning_rate": 4.551925109864259e-06,
      "loss": 0.6301,
      "step": 1797
    },
    {
      "epoch": 0.6082543978349121,
      "grad_norm": 0.31700393557548523,
      "learning_rate": 4.551413270182457e-06,
      "loss": 0.6191,
      "step": 1798
    },
    {
      "epoch": 0.6085926928281461,
      "grad_norm": 0.3024339973926544,
      "learning_rate": 4.550901167139439e-06,
      "loss": 0.6192,
      "step": 1799
    },
    {
      "epoch": 0.6089309878213802,
      "grad_norm": 0.3239328861236572,
      "learning_rate": 4.550388800800948e-06,
      "loss": 0.6272,
      "step": 1800
    },
    {
      "epoch": 0.6092692828146143,
      "grad_norm": 0.3086513876914978,
      "learning_rate": 4.549876171232763e-06,
      "loss": 0.6001,
      "step": 1801
    },
    {
      "epoch": 0.6096075778078485,
      "grad_norm": 0.3088929057121277,
      "learning_rate": 4.549363278500696e-06,
      "loss": 0.6319,
      "step": 1802
    },
    {
      "epoch": 0.6099458728010826,
      "grad_norm": 0.3044373095035553,
      "learning_rate": 4.548850122670591e-06,
      "loss": 0.6127,
      "step": 1803
    },
    {
      "epoch": 0.6102841677943166,
      "grad_norm": 0.31143665313720703,
      "learning_rate": 4.548336703808328e-06,
      "loss": 0.6246,
      "step": 1804
    },
    {
      "epoch": 0.6106224627875507,
      "grad_norm": 0.3061283826828003,
      "learning_rate": 4.547823021979819e-06,
      "loss": 0.6111,
      "step": 1805
    },
    {
      "epoch": 0.6109607577807848,
      "grad_norm": 0.31211286783218384,
      "learning_rate": 4.547309077251012e-06,
      "loss": 0.6391,
      "step": 1806
    },
    {
      "epoch": 0.611299052774019,
      "grad_norm": 0.29427969455718994,
      "learning_rate": 4.546794869687887e-06,
      "loss": 0.6101,
      "step": 1807
    },
    {
      "epoch": 0.6116373477672531,
      "grad_norm": 0.314834326505661,
      "learning_rate": 4.546280399356457e-06,
      "loss": 0.6223,
      "step": 1808
    },
    {
      "epoch": 0.6119756427604871,
      "grad_norm": 0.28986093401908875,
      "learning_rate": 4.5457656663227716e-06,
      "loss": 0.6101,
      "step": 1809
    },
    {
      "epoch": 0.6123139377537212,
      "grad_norm": 0.3071696162223816,
      "learning_rate": 4.54525067065291e-06,
      "loss": 0.6115,
      "step": 1810
    },
    {
      "epoch": 0.6126522327469553,
      "grad_norm": 0.30934277176856995,
      "learning_rate": 4.54473541241299e-06,
      "loss": 0.636,
      "step": 1811
    },
    {
      "epoch": 0.6129905277401895,
      "grad_norm": 0.30213168263435364,
      "learning_rate": 4.54421989166916e-06,
      "loss": 0.6123,
      "step": 1812
    },
    {
      "epoch": 0.6133288227334236,
      "grad_norm": 0.3003959059715271,
      "learning_rate": 4.5437041084876024e-06,
      "loss": 0.6321,
      "step": 1813
    },
    {
      "epoch": 0.6136671177266576,
      "grad_norm": 0.2956574261188507,
      "learning_rate": 4.543188062934534e-06,
      "loss": 0.6395,
      "step": 1814
    },
    {
      "epoch": 0.6140054127198917,
      "grad_norm": 0.3125058710575104,
      "learning_rate": 4.542671755076205e-06,
      "loss": 0.5955,
      "step": 1815
    },
    {
      "epoch": 0.6143437077131259,
      "grad_norm": 0.30371323227882385,
      "learning_rate": 4.542155184978898e-06,
      "loss": 0.6501,
      "step": 1816
    },
    {
      "epoch": 0.61468200270636,
      "grad_norm": 0.3121377229690552,
      "learning_rate": 4.541638352708931e-06,
      "loss": 0.6133,
      "step": 1817
    },
    {
      "epoch": 0.6150202976995941,
      "grad_norm": 0.30062562227249146,
      "learning_rate": 4.541121258332657e-06,
      "loss": 0.6259,
      "step": 1818
    },
    {
      "epoch": 0.6153585926928281,
      "grad_norm": 0.306066632270813,
      "learning_rate": 4.540603901916458e-06,
      "loss": 0.611,
      "step": 1819
    },
    {
      "epoch": 0.6156968876860622,
      "grad_norm": 0.3028794229030609,
      "learning_rate": 4.540086283526755e-06,
      "loss": 0.6574,
      "step": 1820
    },
    {
      "epoch": 0.6160351826792964,
      "grad_norm": 0.3133271038532257,
      "learning_rate": 4.539568403229998e-06,
      "loss": 0.6078,
      "step": 1821
    },
    {
      "epoch": 0.6163734776725305,
      "grad_norm": 0.28491806983947754,
      "learning_rate": 4.539050261092672e-06,
      "loss": 0.6401,
      "step": 1822
    },
    {
      "epoch": 0.6167117726657646,
      "grad_norm": 0.30179521441459656,
      "learning_rate": 4.5385318571812994e-06,
      "loss": 0.6206,
      "step": 1823
    },
    {
      "epoch": 0.6170500676589986,
      "grad_norm": 0.2998371124267578,
      "learning_rate": 4.538013191562431e-06,
      "loss": 0.5907,
      "step": 1824
    },
    {
      "epoch": 0.6173883626522327,
      "grad_norm": 0.30136269330978394,
      "learning_rate": 4.537494264302653e-06,
      "loss": 0.6058,
      "step": 1825
    },
    {
      "epoch": 0.6177266576454669,
      "grad_norm": 0.295851469039917,
      "learning_rate": 4.536975075468587e-06,
      "loss": 0.6031,
      "step": 1826
    },
    {
      "epoch": 0.618064952638701,
      "grad_norm": 0.3003900349140167,
      "learning_rate": 4.5364556251268855e-06,
      "loss": 0.625,
      "step": 1827
    },
    {
      "epoch": 0.618403247631935,
      "grad_norm": 0.29929232597351074,
      "learning_rate": 4.535935913344236e-06,
      "loss": 0.6188,
      "step": 1828
    },
    {
      "epoch": 0.6187415426251691,
      "grad_norm": 0.30480554699897766,
      "learning_rate": 4.535415940187359e-06,
      "loss": 0.6096,
      "step": 1829
    },
    {
      "epoch": 0.6190798376184032,
      "grad_norm": 0.29189208149909973,
      "learning_rate": 4.534895705723009e-06,
      "loss": 0.6328,
      "step": 1830
    },
    {
      "epoch": 0.6194181326116374,
      "grad_norm": 0.290440171957016,
      "learning_rate": 4.534375210017975e-06,
      "loss": 0.6324,
      "step": 1831
    },
    {
      "epoch": 0.6197564276048715,
      "grad_norm": 0.3073995113372803,
      "learning_rate": 4.5338544531390775e-06,
      "loss": 0.6349,
      "step": 1832
    },
    {
      "epoch": 0.6200947225981055,
      "grad_norm": 0.29721468687057495,
      "learning_rate": 4.5333334351531695e-06,
      "loss": 0.6122,
      "step": 1833
    },
    {
      "epoch": 0.6204330175913396,
      "grad_norm": 0.30599215626716614,
      "learning_rate": 4.5328121561271435e-06,
      "loss": 0.6392,
      "step": 1834
    },
    {
      "epoch": 0.6207713125845737,
      "grad_norm": 0.2927095592021942,
      "learning_rate": 4.532290616127919e-06,
      "loss": 0.6602,
      "step": 1835
    },
    {
      "epoch": 0.6211096075778079,
      "grad_norm": 0.307754784822464,
      "learning_rate": 4.531768815222452e-06,
      "loss": 0.6291,
      "step": 1836
    },
    {
      "epoch": 0.621447902571042,
      "grad_norm": 0.3044011890888214,
      "learning_rate": 4.531246753477731e-06,
      "loss": 0.6317,
      "step": 1837
    },
    {
      "epoch": 0.621786197564276,
      "grad_norm": 0.2979058623313904,
      "learning_rate": 4.530724430960779e-06,
      "loss": 0.637,
      "step": 1838
    },
    {
      "epoch": 0.6221244925575101,
      "grad_norm": 0.31330716609954834,
      "learning_rate": 4.530201847738652e-06,
      "loss": 0.6338,
      "step": 1839
    },
    {
      "epoch": 0.6224627875507442,
      "grad_norm": 0.3006748557090759,
      "learning_rate": 4.52967900387844e-06,
      "loss": 0.6259,
      "step": 1840
    },
    {
      "epoch": 0.6228010825439784,
      "grad_norm": 0.2865881323814392,
      "learning_rate": 4.529155899447265e-06,
      "loss": 0.6105,
      "step": 1841
    },
    {
      "epoch": 0.6231393775372125,
      "grad_norm": 0.30335038900375366,
      "learning_rate": 4.528632534512283e-06,
      "loss": 0.6053,
      "step": 1842
    },
    {
      "epoch": 0.6234776725304465,
      "grad_norm": 0.30165183544158936,
      "learning_rate": 4.5281089091406845e-06,
      "loss": 0.6585,
      "step": 1843
    },
    {
      "epoch": 0.6238159675236806,
      "grad_norm": 0.3047686517238617,
      "learning_rate": 4.527585023399693e-06,
      "loss": 0.6155,
      "step": 1844
    },
    {
      "epoch": 0.6241542625169147,
      "grad_norm": 0.2917962372303009,
      "learning_rate": 4.527060877356564e-06,
      "loss": 0.6174,
      "step": 1845
    },
    {
      "epoch": 0.6244925575101489,
      "grad_norm": 0.3098011314868927,
      "learning_rate": 4.526536471078589e-06,
      "loss": 0.6362,
      "step": 1846
    },
    {
      "epoch": 0.624830852503383,
      "grad_norm": 0.3431854248046875,
      "learning_rate": 4.526011804633091e-06,
      "loss": 0.6487,
      "step": 1847
    },
    {
      "epoch": 0.625169147496617,
      "grad_norm": 0.3013041615486145,
      "learning_rate": 4.525486878087426e-06,
      "loss": 0.6037,
      "step": 1848
    },
    {
      "epoch": 0.6255074424898511,
      "grad_norm": 0.3066675066947937,
      "learning_rate": 4.5249616915089846e-06,
      "loss": 0.6335,
      "step": 1849
    },
    {
      "epoch": 0.6258457374830853,
      "grad_norm": 0.34530869126319885,
      "learning_rate": 4.5244362449651915e-06,
      "loss": 0.6089,
      "step": 1850
    },
    {
      "epoch": 0.6261840324763194,
      "grad_norm": 0.3135780394077301,
      "learning_rate": 4.5239105385235035e-06,
      "loss": 0.627,
      "step": 1851
    },
    {
      "epoch": 0.6265223274695535,
      "grad_norm": 0.3080191910266876,
      "learning_rate": 4.523384572251409e-06,
      "loss": 0.6062,
      "step": 1852
    },
    {
      "epoch": 0.6268606224627875,
      "grad_norm": 0.3695083558559418,
      "learning_rate": 4.5228583462164345e-06,
      "loss": 0.6111,
      "step": 1853
    },
    {
      "epoch": 0.6271989174560216,
      "grad_norm": 0.3244759738445282,
      "learning_rate": 4.522331860486136e-06,
      "loss": 0.6436,
      "step": 1854
    },
    {
      "epoch": 0.6275372124492558,
      "grad_norm": 0.3044929802417755,
      "learning_rate": 4.521805115128103e-06,
      "loss": 0.6357,
      "step": 1855
    },
    {
      "epoch": 0.6278755074424899,
      "grad_norm": 0.31515035033226013,
      "learning_rate": 4.52127811020996e-06,
      "loss": 0.5963,
      "step": 1856
    },
    {
      "epoch": 0.628213802435724,
      "grad_norm": 0.3156471252441406,
      "learning_rate": 4.520750845799364e-06,
      "loss": 0.6174,
      "step": 1857
    },
    {
      "epoch": 0.628552097428958,
      "grad_norm": 0.3069937527179718,
      "learning_rate": 4.520223321964007e-06,
      "loss": 0.6293,
      "step": 1858
    },
    {
      "epoch": 0.6288903924221921,
      "grad_norm": 0.2919570207595825,
      "learning_rate": 4.51969553877161e-06,
      "loss": 0.6259,
      "step": 1859
    },
    {
      "epoch": 0.6292286874154263,
      "grad_norm": 0.3123161494731903,
      "learning_rate": 4.5191674962899314e-06,
      "loss": 0.6201,
      "step": 1860
    },
    {
      "epoch": 0.6295669824086604,
      "grad_norm": 0.30140236020088196,
      "learning_rate": 4.518639194586762e-06,
      "loss": 0.6178,
      "step": 1861
    },
    {
      "epoch": 0.6299052774018945,
      "grad_norm": 0.2950606644153595,
      "learning_rate": 4.518110633729924e-06,
      "loss": 0.6103,
      "step": 1862
    },
    {
      "epoch": 0.6302435723951285,
      "grad_norm": 0.30651673674583435,
      "learning_rate": 4.517581813787275e-06,
      "loss": 0.6492,
      "step": 1863
    },
    {
      "epoch": 0.6305818673883626,
      "grad_norm": 0.3081085681915283,
      "learning_rate": 4.517052734826706e-06,
      "loss": 0.6305,
      "step": 1864
    },
    {
      "epoch": 0.6309201623815968,
      "grad_norm": 0.30926793813705444,
      "learning_rate": 4.5165233969161394e-06,
      "loss": 0.6182,
      "step": 1865
    },
    {
      "epoch": 0.6312584573748309,
      "grad_norm": 0.295867383480072,
      "learning_rate": 4.515993800123531e-06,
      "loss": 0.6139,
      "step": 1866
    },
    {
      "epoch": 0.631596752368065,
      "grad_norm": 0.3241640627384186,
      "learning_rate": 4.515463944516872e-06,
      "loss": 0.6298,
      "step": 1867
    },
    {
      "epoch": 0.631935047361299,
      "grad_norm": 0.32582196593284607,
      "learning_rate": 4.5149338301641845e-06,
      "loss": 0.6066,
      "step": 1868
    },
    {
      "epoch": 0.6322733423545331,
      "grad_norm": 0.3094950020313263,
      "learning_rate": 4.514403457133526e-06,
      "loss": 0.6115,
      "step": 1869
    },
    {
      "epoch": 0.6326116373477673,
      "grad_norm": 0.3076767921447754,
      "learning_rate": 4.513872825492984e-06,
      "loss": 0.6482,
      "step": 1870
    },
    {
      "epoch": 0.6329499323410014,
      "grad_norm": 0.3136141002178192,
      "learning_rate": 4.513341935310684e-06,
      "loss": 0.6598,
      "step": 1871
    },
    {
      "epoch": 0.6332882273342354,
      "grad_norm": 0.34437096118927,
      "learning_rate": 4.5128107866547795e-06,
      "loss": 0.6144,
      "step": 1872
    },
    {
      "epoch": 0.6336265223274695,
      "grad_norm": 0.29778510332107544,
      "learning_rate": 4.512279379593461e-06,
      "loss": 0.6197,
      "step": 1873
    },
    {
      "epoch": 0.6339648173207036,
      "grad_norm": 0.32931333780288696,
      "learning_rate": 4.5117477141949485e-06,
      "loss": 0.6288,
      "step": 1874
    },
    {
      "epoch": 0.6343031123139378,
      "grad_norm": 0.335248202085495,
      "learning_rate": 4.511215790527501e-06,
      "loss": 0.6151,
      "step": 1875
    },
    {
      "epoch": 0.6346414073071719,
      "grad_norm": 0.2965134382247925,
      "learning_rate": 4.510683608659404e-06,
      "loss": 0.62,
      "step": 1876
    },
    {
      "epoch": 0.6349797023004059,
      "grad_norm": 0.294978529214859,
      "learning_rate": 4.51015116865898e-06,
      "loss": 0.6222,
      "step": 1877
    },
    {
      "epoch": 0.63531799729364,
      "grad_norm": 0.33229756355285645,
      "learning_rate": 4.509618470594584e-06,
      "loss": 0.6474,
      "step": 1878
    },
    {
      "epoch": 0.6356562922868741,
      "grad_norm": 0.3117801547050476,
      "learning_rate": 4.509085514534606e-06,
      "loss": 0.6074,
      "step": 1879
    },
    {
      "epoch": 0.6359945872801083,
      "grad_norm": 0.30003616213798523,
      "learning_rate": 4.508552300547463e-06,
      "loss": 0.6186,
      "step": 1880
    },
    {
      "epoch": 0.6363328822733424,
      "grad_norm": 0.29257848858833313,
      "learning_rate": 4.508018828701613e-06,
      "loss": 0.6297,
      "step": 1881
    },
    {
      "epoch": 0.6366711772665764,
      "grad_norm": 0.4021831750869751,
      "learning_rate": 4.507485099065541e-06,
      "loss": 0.613,
      "step": 1882
    },
    {
      "epoch": 0.6370094722598105,
      "grad_norm": 0.3081733286380768,
      "learning_rate": 4.506951111707768e-06,
      "loss": 0.6497,
      "step": 1883
    },
    {
      "epoch": 0.6373477672530447,
      "grad_norm": 0.29754048585891724,
      "learning_rate": 4.506416866696849e-06,
      "loss": 0.6135,
      "step": 1884
    },
    {
      "epoch": 0.6376860622462788,
      "grad_norm": 0.3275603950023651,
      "learning_rate": 4.505882364101367e-06,
      "loss": 0.6273,
      "step": 1885
    },
    {
      "epoch": 0.6380243572395129,
      "grad_norm": 0.32382166385650635,
      "learning_rate": 4.505347603989946e-06,
      "loss": 0.6264,
      "step": 1886
    },
    {
      "epoch": 0.6383626522327469,
      "grad_norm": 0.31701892614364624,
      "learning_rate": 4.504812586431236e-06,
      "loss": 0.6106,
      "step": 1887
    },
    {
      "epoch": 0.638700947225981,
      "grad_norm": 0.2922547161579132,
      "learning_rate": 4.5042773114939225e-06,
      "loss": 0.6341,
      "step": 1888
    },
    {
      "epoch": 0.6390392422192152,
      "grad_norm": 0.3007252812385559,
      "learning_rate": 4.503741779246726e-06,
      "loss": 0.6408,
      "step": 1889
    },
    {
      "epoch": 0.6393775372124493,
      "grad_norm": 0.3614402413368225,
      "learning_rate": 4.503205989758397e-06,
      "loss": 0.6023,
      "step": 1890
    },
    {
      "epoch": 0.6397158322056834,
      "grad_norm": 0.3172839283943176,
      "learning_rate": 4.502669943097721e-06,
      "loss": 0.6552,
      "step": 1891
    },
    {
      "epoch": 0.6400541271989174,
      "grad_norm": 0.3163464367389679,
      "learning_rate": 4.502133639333516e-06,
      "loss": 0.6254,
      "step": 1892
    },
    {
      "epoch": 0.6403924221921515,
      "grad_norm": 0.3306618928909302,
      "learning_rate": 4.501597078534633e-06,
      "loss": 0.5947,
      "step": 1893
    },
    {
      "epoch": 0.6407307171853857,
      "grad_norm": 0.31476470828056335,
      "learning_rate": 4.501060260769955e-06,
      "loss": 0.6453,
      "step": 1894
    },
    {
      "epoch": 0.6410690121786198,
      "grad_norm": 0.3184593915939331,
      "learning_rate": 4.5005231861083994e-06,
      "loss": 0.6365,
      "step": 1895
    },
    {
      "epoch": 0.6414073071718539,
      "grad_norm": 0.3020038306713104,
      "learning_rate": 4.499985854618916e-06,
      "loss": 0.6286,
      "step": 1896
    },
    {
      "epoch": 0.6417456021650879,
      "grad_norm": 0.40171536803245544,
      "learning_rate": 4.499448266370487e-06,
      "loss": 0.628,
      "step": 1897
    },
    {
      "epoch": 0.642083897158322,
      "grad_norm": 0.30737072229385376,
      "learning_rate": 4.49891042143213e-06,
      "loss": 0.6304,
      "step": 1898
    },
    {
      "epoch": 0.6424221921515562,
      "grad_norm": 0.31211888790130615,
      "learning_rate": 4.498372319872892e-06,
      "loss": 0.6183,
      "step": 1899
    },
    {
      "epoch": 0.6427604871447903,
      "grad_norm": 0.31268051266670227,
      "learning_rate": 4.497833961761855e-06,
      "loss": 0.6368,
      "step": 1900
    },
    {
      "epoch": 0.6430987821380244,
      "grad_norm": 0.3052826225757599,
      "learning_rate": 4.497295347168134e-06,
      "loss": 0.6023,
      "step": 1901
    },
    {
      "epoch": 0.6434370771312584,
      "grad_norm": 0.30680912733078003,
      "learning_rate": 4.496756476160876e-06,
      "loss": 0.6088,
      "step": 1902
    },
    {
      "epoch": 0.6437753721244925,
      "grad_norm": 0.2901969850063324,
      "learning_rate": 4.496217348809263e-06,
      "loss": 0.6357,
      "step": 1903
    },
    {
      "epoch": 0.6441136671177267,
      "grad_norm": 0.2912886440753937,
      "learning_rate": 4.495677965182506e-06,
      "loss": 0.6027,
      "step": 1904
    },
    {
      "epoch": 0.6444519621109608,
      "grad_norm": 0.3001358211040497,
      "learning_rate": 4.495138325349854e-06,
      "loss": 0.6284,
      "step": 1905
    },
    {
      "epoch": 0.6447902571041949,
      "grad_norm": 0.32089850306510925,
      "learning_rate": 4.494598429380583e-06,
      "loss": 0.6001,
      "step": 1906
    },
    {
      "epoch": 0.6451285520974289,
      "grad_norm": 0.3067314326763153,
      "learning_rate": 4.494058277344008e-06,
      "loss": 0.6002,
      "step": 1907
    },
    {
      "epoch": 0.645466847090663,
      "grad_norm": 0.29842349886894226,
      "learning_rate": 4.493517869309472e-06,
      "loss": 0.5946,
      "step": 1908
    },
    {
      "epoch": 0.6458051420838972,
      "grad_norm": 0.3103325068950653,
      "learning_rate": 4.492977205346354e-06,
      "loss": 0.6184,
      "step": 1909
    },
    {
      "epoch": 0.6461434370771313,
      "grad_norm": 0.29811927676200867,
      "learning_rate": 4.492436285524063e-06,
      "loss": 0.6304,
      "step": 1910
    },
    {
      "epoch": 0.6464817320703654,
      "grad_norm": 0.2993214726448059,
      "learning_rate": 4.4918951099120445e-06,
      "loss": 0.6072,
      "step": 1911
    },
    {
      "epoch": 0.6468200270635994,
      "grad_norm": 0.2939266562461853,
      "learning_rate": 4.491353678579774e-06,
      "loss": 0.6187,
      "step": 1912
    },
    {
      "epoch": 0.6471583220568335,
      "grad_norm": 0.3164563775062561,
      "learning_rate": 4.49081199159676e-06,
      "loss": 0.6248,
      "step": 1913
    },
    {
      "epoch": 0.6474966170500677,
      "grad_norm": 0.30701878666877747,
      "learning_rate": 4.490270049032546e-06,
      "loss": 0.6345,
      "step": 1914
    },
    {
      "epoch": 0.6478349120433018,
      "grad_norm": 0.28954970836639404,
      "learning_rate": 4.489727850956705e-06,
      "loss": 0.6065,
      "step": 1915
    },
    {
      "epoch": 0.6481732070365359,
      "grad_norm": 0.2986772656440735,
      "learning_rate": 4.489185397438846e-06,
      "loss": 0.614,
      "step": 1916
    },
    {
      "epoch": 0.6485115020297699,
      "grad_norm": 0.2876133620738983,
      "learning_rate": 4.4886426885486075e-06,
      "loss": 0.6303,
      "step": 1917
    },
    {
      "epoch": 0.648849797023004,
      "grad_norm": 0.30175134539604187,
      "learning_rate": 4.4880997243556654e-06,
      "loss": 0.5836,
      "step": 1918
    },
    {
      "epoch": 0.6491880920162382,
      "grad_norm": 0.2978866398334503,
      "learning_rate": 4.4875565049297235e-06,
      "loss": 0.6164,
      "step": 1919
    },
    {
      "epoch": 0.6495263870094723,
      "grad_norm": 0.2987023890018463,
      "learning_rate": 4.487013030340522e-06,
      "loss": 0.6097,
      "step": 1920
    },
    {
      "epoch": 0.6498646820027063,
      "grad_norm": 0.3010125160217285,
      "learning_rate": 4.4864693006578305e-06,
      "loss": 0.6113,
      "step": 1921
    },
    {
      "epoch": 0.6502029769959404,
      "grad_norm": 0.30306074023246765,
      "learning_rate": 4.485925315951456e-06,
      "loss": 0.6327,
      "step": 1922
    },
    {
      "epoch": 0.6505412719891746,
      "grad_norm": 0.3026984930038452,
      "learning_rate": 4.4853810762912335e-06,
      "loss": 0.614,
      "step": 1923
    },
    {
      "epoch": 0.6508795669824087,
      "grad_norm": 0.2981746196746826,
      "learning_rate": 4.484836581747032e-06,
      "loss": 0.6354,
      "step": 1924
    },
    {
      "epoch": 0.6512178619756428,
      "grad_norm": 0.2966022491455078,
      "learning_rate": 4.4842918323887555e-06,
      "loss": 0.6432,
      "step": 1925
    },
    {
      "epoch": 0.6515561569688768,
      "grad_norm": 0.286921888589859,
      "learning_rate": 4.483746828286339e-06,
      "loss": 0.6112,
      "step": 1926
    },
    {
      "epoch": 0.6518944519621109,
      "grad_norm": 0.3007039427757263,
      "learning_rate": 4.4832015695097484e-06,
      "loss": 0.6263,
      "step": 1927
    },
    {
      "epoch": 0.652232746955345,
      "grad_norm": 0.30770787596702576,
      "learning_rate": 4.482656056128986e-06,
      "loss": 0.6096,
      "step": 1928
    },
    {
      "epoch": 0.6525710419485792,
      "grad_norm": 0.2954249978065491,
      "learning_rate": 4.482110288214087e-06,
      "loss": 0.6103,
      "step": 1929
    },
    {
      "epoch": 0.6529093369418133,
      "grad_norm": 0.30537140369415283,
      "learning_rate": 4.481564265835113e-06,
      "loss": 0.6444,
      "step": 1930
    },
    {
      "epoch": 0.6532476319350473,
      "grad_norm": 0.29392582178115845,
      "learning_rate": 4.481017989062165e-06,
      "loss": 0.619,
      "step": 1931
    },
    {
      "epoch": 0.6535859269282814,
      "grad_norm": 0.3098503053188324,
      "learning_rate": 4.480471457965374e-06,
      "loss": 0.6141,
      "step": 1932
    },
    {
      "epoch": 0.6539242219215156,
      "grad_norm": 0.3279946446418762,
      "learning_rate": 4.479924672614903e-06,
      "loss": 0.6255,
      "step": 1933
    },
    {
      "epoch": 0.6542625169147497,
      "grad_norm": 0.30103981494903564,
      "learning_rate": 4.4793776330809504e-06,
      "loss": 0.6332,
      "step": 1934
    },
    {
      "epoch": 0.6546008119079838,
      "grad_norm": 0.32124587893486023,
      "learning_rate": 4.478830339433744e-06,
      "loss": 0.6409,
      "step": 1935
    },
    {
      "epoch": 0.6549391069012178,
      "grad_norm": 0.31162700057029724,
      "learning_rate": 4.478282791743546e-06,
      "loss": 0.6432,
      "step": 1936
    },
    {
      "epoch": 0.6552774018944519,
      "grad_norm": 0.3058624565601349,
      "learning_rate": 4.477734990080649e-06,
      "loss": 0.6413,
      "step": 1937
    },
    {
      "epoch": 0.6556156968876861,
      "grad_norm": 0.29996979236602783,
      "learning_rate": 4.477186934515383e-06,
      "loss": 0.6187,
      "step": 1938
    },
    {
      "epoch": 0.6559539918809202,
      "grad_norm": 0.2978936433792114,
      "learning_rate": 4.4766386251181055e-06,
      "loss": 0.6186,
      "step": 1939
    },
    {
      "epoch": 0.6562922868741543,
      "grad_norm": 0.2992279529571533,
      "learning_rate": 4.476090061959209e-06,
      "loss": 0.6053,
      "step": 1940
    },
    {
      "epoch": 0.6566305818673883,
      "grad_norm": 0.320601224899292,
      "learning_rate": 4.475541245109118e-06,
      "loss": 0.5902,
      "step": 1941
    },
    {
      "epoch": 0.6569688768606224,
      "grad_norm": 0.32168227434158325,
      "learning_rate": 4.474992174638292e-06,
      "loss": 0.642,
      "step": 1942
    },
    {
      "epoch": 0.6573071718538566,
      "grad_norm": 0.29337483644485474,
      "learning_rate": 4.474442850617218e-06,
      "loss": 0.6184,
      "step": 1943
    },
    {
      "epoch": 0.6576454668470907,
      "grad_norm": 0.2891935408115387,
      "learning_rate": 4.47389327311642e-06,
      "loss": 0.6125,
      "step": 1944
    },
    {
      "epoch": 0.6579837618403248,
      "grad_norm": 0.3021150529384613,
      "learning_rate": 4.473343442206452e-06,
      "loss": 0.6144,
      "step": 1945
    },
    {
      "epoch": 0.6583220568335588,
      "grad_norm": 0.30139896273612976,
      "learning_rate": 4.472793357957902e-06,
      "loss": 0.65,
      "step": 1946
    },
    {
      "epoch": 0.658660351826793,
      "grad_norm": 0.3164536654949188,
      "learning_rate": 4.4722430204413905e-06,
      "loss": 0.6289,
      "step": 1947
    },
    {
      "epoch": 0.6589986468200271,
      "grad_norm": 0.29989132285118103,
      "learning_rate": 4.47169242972757e-06,
      "loss": 0.6043,
      "step": 1948
    },
    {
      "epoch": 0.6593369418132612,
      "grad_norm": 0.30992090702056885,
      "learning_rate": 4.4711415858871245e-06,
      "loss": 0.6203,
      "step": 1949
    },
    {
      "epoch": 0.6596752368064953,
      "grad_norm": 0.31831830739974976,
      "learning_rate": 4.470590488990773e-06,
      "loss": 0.6217,
      "step": 1950
    },
    {
      "epoch": 0.6600135317997293,
      "grad_norm": 0.30647245049476624,
      "learning_rate": 4.470039139109265e-06,
      "loss": 0.5992,
      "step": 1951
    },
    {
      "epoch": 0.6603518267929634,
      "grad_norm": 0.30912190675735474,
      "learning_rate": 4.469487536313381e-06,
      "loss": 0.6199,
      "step": 1952
    },
    {
      "epoch": 0.6606901217861976,
      "grad_norm": 0.31395310163497925,
      "learning_rate": 4.46893568067394e-06,
      "loss": 0.6285,
      "step": 1953
    },
    {
      "epoch": 0.6610284167794317,
      "grad_norm": 0.3426959812641144,
      "learning_rate": 4.468383572261785e-06,
      "loss": 0.6317,
      "step": 1954
    },
    {
      "epoch": 0.6613667117726658,
      "grad_norm": 0.3148653209209442,
      "learning_rate": 4.467831211147799e-06,
      "loss": 0.6212,
      "step": 1955
    },
    {
      "epoch": 0.6617050067658998,
      "grad_norm": 0.3014954626560211,
      "learning_rate": 4.4672785974028935e-06,
      "loss": 0.6046,
      "step": 1956
    },
    {
      "epoch": 0.662043301759134,
      "grad_norm": 0.33388400077819824,
      "learning_rate": 4.466725731098013e-06,
      "loss": 0.6218,
      "step": 1957
    },
    {
      "epoch": 0.6623815967523681,
      "grad_norm": 0.3054538071155548,
      "learning_rate": 4.466172612304135e-06,
      "loss": 0.6546,
      "step": 1958
    },
    {
      "epoch": 0.6627198917456022,
      "grad_norm": 0.31339314579963684,
      "learning_rate": 4.46561924109227e-06,
      "loss": 0.627,
      "step": 1959
    },
    {
      "epoch": 0.6630581867388363,
      "grad_norm": 0.2950597405433655,
      "learning_rate": 4.465065617533457e-06,
      "loss": 0.6252,
      "step": 1960
    },
    {
      "epoch": 0.6633964817320703,
      "grad_norm": 0.3085661828517914,
      "learning_rate": 4.464511741698773e-06,
      "loss": 0.6179,
      "step": 1961
    },
    {
      "epoch": 0.6637347767253045,
      "grad_norm": 0.3314128518104553,
      "learning_rate": 4.463957613659325e-06,
      "loss": 0.6212,
      "step": 1962
    },
    {
      "epoch": 0.6640730717185386,
      "grad_norm": 0.3055969774723053,
      "learning_rate": 4.46340323348625e-06,
      "loss": 0.6241,
      "step": 1963
    },
    {
      "epoch": 0.6644113667117727,
      "grad_norm": 0.3067176938056946,
      "learning_rate": 4.462848601250721e-06,
      "loss": 0.6261,
      "step": 1964
    },
    {
      "epoch": 0.6647496617050067,
      "grad_norm": 0.3050393760204315,
      "learning_rate": 4.462293717023943e-06,
      "loss": 0.6316,
      "step": 1965
    },
    {
      "epoch": 0.6650879566982408,
      "grad_norm": 0.32629260420799255,
      "learning_rate": 4.461738580877151e-06,
      "loss": 0.6313,
      "step": 1966
    },
    {
      "epoch": 0.665426251691475,
      "grad_norm": 0.29928138852119446,
      "learning_rate": 4.461183192881612e-06,
      "loss": 0.6434,
      "step": 1967
    },
    {
      "epoch": 0.6657645466847091,
      "grad_norm": 0.308493435382843,
      "learning_rate": 4.46062755310863e-06,
      "loss": 0.6266,
      "step": 1968
    },
    {
      "epoch": 0.6661028416779432,
      "grad_norm": 0.31029900908470154,
      "learning_rate": 4.4600716616295355e-06,
      "loss": 0.6187,
      "step": 1969
    },
    {
      "epoch": 0.6664411366711772,
      "grad_norm": 0.30931907892227173,
      "learning_rate": 4.459515518515696e-06,
      "loss": 0.6202,
      "step": 1970
    },
    {
      "epoch": 0.6667794316644113,
      "grad_norm": 0.2888835370540619,
      "learning_rate": 4.4589591238385085e-06,
      "loss": 0.6092,
      "step": 1971
    },
    {
      "epoch": 0.6671177266576455,
      "grad_norm": 0.2925364077091217,
      "learning_rate": 4.458402477669403e-06,
      "loss": 0.5867,
      "step": 1972
    },
    {
      "epoch": 0.6674560216508796,
      "grad_norm": 0.31146425008773804,
      "learning_rate": 4.4578455800798435e-06,
      "loss": 0.644,
      "step": 1973
    },
    {
      "epoch": 0.6677943166441137,
      "grad_norm": 0.28718727827072144,
      "learning_rate": 4.457288431141322e-06,
      "loss": 0.5907,
      "step": 1974
    },
    {
      "epoch": 0.6681326116373477,
      "grad_norm": 0.3077245354652405,
      "learning_rate": 4.456731030925368e-06,
      "loss": 0.624,
      "step": 1975
    },
    {
      "epoch": 0.6684709066305818,
      "grad_norm": 0.3103635311126709,
      "learning_rate": 4.45617337950354e-06,
      "loss": 0.6181,
      "step": 1976
    },
    {
      "epoch": 0.668809201623816,
      "grad_norm": 0.31054601073265076,
      "learning_rate": 4.455615476947428e-06,
      "loss": 0.6207,
      "step": 1977
    },
    {
      "epoch": 0.6691474966170501,
      "grad_norm": 0.3209282457828522,
      "learning_rate": 4.455057323328658e-06,
      "loss": 0.6072,
      "step": 1978
    },
    {
      "epoch": 0.6694857916102842,
      "grad_norm": 0.3089926540851593,
      "learning_rate": 4.4544989187188845e-06,
      "loss": 0.629,
      "step": 1979
    },
    {
      "epoch": 0.6698240866035182,
      "grad_norm": 0.3040165901184082,
      "learning_rate": 4.453940263189797e-06,
      "loss": 0.6393,
      "step": 1980
    },
    {
      "epoch": 0.6701623815967523,
      "grad_norm": 0.30424657464027405,
      "learning_rate": 4.4533813568131145e-06,
      "loss": 0.6386,
      "step": 1981
    },
    {
      "epoch": 0.6705006765899865,
      "grad_norm": 0.31967079639434814,
      "learning_rate": 4.4528221996605905e-06,
      "loss": 0.6047,
      "step": 1982
    },
    {
      "epoch": 0.6708389715832206,
      "grad_norm": 0.3124231994152069,
      "learning_rate": 4.452262791804009e-06,
      "loss": 0.6324,
      "step": 1983
    },
    {
      "epoch": 0.6711772665764547,
      "grad_norm": 0.320998877286911,
      "learning_rate": 4.451703133315187e-06,
      "loss": 0.6223,
      "step": 1984
    },
    {
      "epoch": 0.6715155615696887,
      "grad_norm": 0.30833613872528076,
      "learning_rate": 4.451143224265975e-06,
      "loss": 0.628,
      "step": 1985
    },
    {
      "epoch": 0.6718538565629228,
      "grad_norm": 0.2876376211643219,
      "learning_rate": 4.450583064728253e-06,
      "loss": 0.6139,
      "step": 1986
    },
    {
      "epoch": 0.672192151556157,
      "grad_norm": 0.29824233055114746,
      "learning_rate": 4.450022654773936e-06,
      "loss": 0.6364,
      "step": 1987
    },
    {
      "epoch": 0.6725304465493911,
      "grad_norm": 0.29504624009132385,
      "learning_rate": 4.449461994474968e-06,
      "loss": 0.6165,
      "step": 1988
    },
    {
      "epoch": 0.6728687415426252,
      "grad_norm": 0.3090459406375885,
      "learning_rate": 4.448901083903327e-06,
      "loss": 0.6539,
      "step": 1989
    },
    {
      "epoch": 0.6732070365358592,
      "grad_norm": 0.30286070704460144,
      "learning_rate": 4.4483399231310245e-06,
      "loss": 0.6271,
      "step": 1990
    },
    {
      "epoch": 0.6735453315290933,
      "grad_norm": 0.2875913977622986,
      "learning_rate": 4.4477785122301e-06,
      "loss": 0.6078,
      "step": 1991
    },
    {
      "epoch": 0.6738836265223275,
      "grad_norm": 0.3007805347442627,
      "learning_rate": 4.44721685127263e-06,
      "loss": 0.6229,
      "step": 1992
    },
    {
      "epoch": 0.6742219215155616,
      "grad_norm": 0.30888527631759644,
      "learning_rate": 4.4466549403307195e-06,
      "loss": 0.6038,
      "step": 1993
    },
    {
      "epoch": 0.6745602165087957,
      "grad_norm": 0.3012210428714752,
      "learning_rate": 4.446092779476507e-06,
      "loss": 0.604,
      "step": 1994
    },
    {
      "epoch": 0.6748985115020297,
      "grad_norm": 0.3052041232585907,
      "learning_rate": 4.445530368782163e-06,
      "loss": 0.6271,
      "step": 1995
    },
    {
      "epoch": 0.6752368064952639,
      "grad_norm": 0.30048370361328125,
      "learning_rate": 4.44496770831989e-06,
      "loss": 0.6322,
      "step": 1996
    },
    {
      "epoch": 0.675575101488498,
      "grad_norm": 0.315618097782135,
      "learning_rate": 4.444404798161922e-06,
      "loss": 0.6246,
      "step": 1997
    },
    {
      "epoch": 0.6759133964817321,
      "grad_norm": 0.3120715916156769,
      "learning_rate": 4.443841638380527e-06,
      "loss": 0.6105,
      "step": 1998
    },
    {
      "epoch": 0.6762516914749662,
      "grad_norm": 0.2947427034378052,
      "learning_rate": 4.443278229048002e-06,
      "loss": 0.6094,
      "step": 1999
    },
    {
      "epoch": 0.6765899864682002,
      "grad_norm": 0.31929072737693787,
      "learning_rate": 4.4427145702366804e-06,
      "loss": 0.6565,
      "step": 2000
    },
    {
      "epoch": 0.6769282814614344,
      "grad_norm": 0.30863380432128906,
      "learning_rate": 4.442150662018921e-06,
      "loss": 0.6154,
      "step": 2001
    },
    {
      "epoch": 0.6772665764546685,
      "grad_norm": 0.323507159948349,
      "learning_rate": 4.441586504467122e-06,
      "loss": 0.6491,
      "step": 2002
    },
    {
      "epoch": 0.6776048714479026,
      "grad_norm": 0.32120445370674133,
      "learning_rate": 4.441022097653709e-06,
      "loss": 0.633,
      "step": 2003
    },
    {
      "epoch": 0.6779431664411367,
      "grad_norm": 0.3001091182231903,
      "learning_rate": 4.440457441651139e-06,
      "loss": 0.6174,
      "step": 2004
    },
    {
      "epoch": 0.6782814614343707,
      "grad_norm": 0.3134676516056061,
      "learning_rate": 4.439892536531906e-06,
      "loss": 0.6355,
      "step": 2005
    },
    {
      "epoch": 0.6786197564276049,
      "grad_norm": 0.32472702860832214,
      "learning_rate": 4.43932738236853e-06,
      "loss": 0.6161,
      "step": 2006
    },
    {
      "epoch": 0.678958051420839,
      "grad_norm": 0.2919924259185791,
      "learning_rate": 4.438761979233568e-06,
      "loss": 0.6225,
      "step": 2007
    },
    {
      "epoch": 0.6792963464140731,
      "grad_norm": 0.28722047805786133,
      "learning_rate": 4.438196327199604e-06,
      "loss": 0.6087,
      "step": 2008
    },
    {
      "epoch": 0.6796346414073072,
      "grad_norm": 0.3046940863132477,
      "learning_rate": 4.437630426339259e-06,
      "loss": 0.6443,
      "step": 2009
    },
    {
      "epoch": 0.6799729364005412,
      "grad_norm": 0.32804107666015625,
      "learning_rate": 4.437064276725183e-06,
      "loss": 0.6077,
      "step": 2010
    },
    {
      "epoch": 0.6803112313937754,
      "grad_norm": 0.31065163016319275,
      "learning_rate": 4.436497878430057e-06,
      "loss": 0.6456,
      "step": 2011
    },
    {
      "epoch": 0.6806495263870095,
      "grad_norm": 0.3122207224369049,
      "learning_rate": 4.435931231526597e-06,
      "loss": 0.647,
      "step": 2012
    },
    {
      "epoch": 0.6809878213802436,
      "grad_norm": 0.2977723479270935,
      "learning_rate": 4.435364336087549e-06,
      "loss": 0.6344,
      "step": 2013
    },
    {
      "epoch": 0.6813261163734776,
      "grad_norm": 0.30897650122642517,
      "learning_rate": 4.434797192185691e-06,
      "loss": 0.6284,
      "step": 2014
    },
    {
      "epoch": 0.6816644113667117,
      "grad_norm": 0.31061431765556335,
      "learning_rate": 4.434229799893833e-06,
      "loss": 0.5913,
      "step": 2015
    },
    {
      "epoch": 0.6820027063599459,
      "grad_norm": 0.30263346433639526,
      "learning_rate": 4.433662159284818e-06,
      "loss": 0.6329,
      "step": 2016
    },
    {
      "epoch": 0.68234100135318,
      "grad_norm": 0.29153409600257874,
      "learning_rate": 4.43309427043152e-06,
      "loss": 0.6289,
      "step": 2017
    },
    {
      "epoch": 0.6826792963464141,
      "grad_norm": 0.29121798276901245,
      "learning_rate": 4.432526133406843e-06,
      "loss": 0.6246,
      "step": 2018
    },
    {
      "epoch": 0.6830175913396481,
      "grad_norm": 0.29549071192741394,
      "learning_rate": 4.431957748283725e-06,
      "loss": 0.6361,
      "step": 2019
    },
    {
      "epoch": 0.6833558863328822,
      "grad_norm": 0.2968063950538635,
      "learning_rate": 4.431389115135138e-06,
      "loss": 0.6018,
      "step": 2020
    },
    {
      "epoch": 0.6836941813261164,
      "grad_norm": 0.30572426319122314,
      "learning_rate": 4.430820234034081e-06,
      "loss": 0.6287,
      "step": 2021
    },
    {
      "epoch": 0.6840324763193505,
      "grad_norm": 0.29529130458831787,
      "learning_rate": 4.430251105053586e-06,
      "loss": 0.6009,
      "step": 2022
    },
    {
      "epoch": 0.6843707713125846,
      "grad_norm": 0.29441121220588684,
      "learning_rate": 4.429681728266721e-06,
      "loss": 0.6015,
      "step": 2023
    },
    {
      "epoch": 0.6847090663058186,
      "grad_norm": 0.30199480056762695,
      "learning_rate": 4.4291121037465825e-06,
      "loss": 0.6377,
      "step": 2024
    },
    {
      "epoch": 0.6850473612990527,
      "grad_norm": 0.30264589190483093,
      "learning_rate": 4.428542231566297e-06,
      "loss": 0.6215,
      "step": 2025
    },
    {
      "epoch": 0.6853856562922869,
      "grad_norm": 0.2855820953845978,
      "learning_rate": 4.427972111799026e-06,
      "loss": 0.6178,
      "step": 2026
    },
    {
      "epoch": 0.685723951285521,
      "grad_norm": 0.31441786885261536,
      "learning_rate": 4.427401744517963e-06,
      "loss": 0.6324,
      "step": 2027
    },
    {
      "epoch": 0.6860622462787551,
      "grad_norm": 0.29725101590156555,
      "learning_rate": 4.42683112979633e-06,
      "loss": 0.6196,
      "step": 2028
    },
    {
      "epoch": 0.6864005412719891,
      "grad_norm": 0.3108838200569153,
      "learning_rate": 4.426260267707382e-06,
      "loss": 0.6123,
      "step": 2029
    },
    {
      "epoch": 0.6867388362652233,
      "grad_norm": 0.3054821491241455,
      "learning_rate": 4.42568915832441e-06,
      "loss": 0.6051,
      "step": 2030
    },
    {
      "epoch": 0.6870771312584574,
      "grad_norm": 0.2996203899383545,
      "learning_rate": 4.42511780172073e-06,
      "loss": 0.6078,
      "step": 2031
    },
    {
      "epoch": 0.6874154262516915,
      "grad_norm": 0.3160284459590912,
      "learning_rate": 4.424546197969694e-06,
      "loss": 0.6135,
      "step": 2032
    },
    {
      "epoch": 0.6877537212449256,
      "grad_norm": 0.3138665556907654,
      "learning_rate": 4.423974347144685e-06,
      "loss": 0.6368,
      "step": 2033
    },
    {
      "epoch": 0.6880920162381596,
      "grad_norm": 0.2967749536037445,
      "learning_rate": 4.423402249319118e-06,
      "loss": 0.6203,
      "step": 2034
    },
    {
      "epoch": 0.6884303112313938,
      "grad_norm": 0.3004557490348816,
      "learning_rate": 4.422829904566438e-06,
      "loss": 0.6269,
      "step": 2035
    },
    {
      "epoch": 0.6887686062246279,
      "grad_norm": 0.3086588680744171,
      "learning_rate": 4.422257312960123e-06,
      "loss": 0.6152,
      "step": 2036
    },
    {
      "epoch": 0.689106901217862,
      "grad_norm": 0.3136051297187805,
      "learning_rate": 4.421684474573683e-06,
      "loss": 0.6186,
      "step": 2037
    },
    {
      "epoch": 0.6894451962110961,
      "grad_norm": 0.29930639266967773,
      "learning_rate": 4.421111389480659e-06,
      "loss": 0.6097,
      "step": 2038
    },
    {
      "epoch": 0.6897834912043301,
      "grad_norm": 0.2962172031402588,
      "learning_rate": 4.420538057754624e-06,
      "loss": 0.6144,
      "step": 2039
    },
    {
      "epoch": 0.6901217861975643,
      "grad_norm": 0.3190195560455322,
      "learning_rate": 4.419964479469182e-06,
      "loss": 0.6146,
      "step": 2040
    },
    {
      "epoch": 0.6904600811907984,
      "grad_norm": 0.29853519797325134,
      "learning_rate": 4.41939065469797e-06,
      "loss": 0.6209,
      "step": 2041
    },
    {
      "epoch": 0.6907983761840325,
      "grad_norm": 0.30535730719566345,
      "learning_rate": 4.418816583514656e-06,
      "loss": 0.6155,
      "step": 2042
    },
    {
      "epoch": 0.6911366711772666,
      "grad_norm": 0.3163444697856903,
      "learning_rate": 4.418242265992938e-06,
      "loss": 0.6461,
      "step": 2043
    },
    {
      "epoch": 0.6914749661705006,
      "grad_norm": 0.3203446567058563,
      "learning_rate": 4.417667702206548e-06,
      "loss": 0.6407,
      "step": 2044
    },
    {
      "epoch": 0.6918132611637348,
      "grad_norm": 0.30420753359794617,
      "learning_rate": 4.417092892229249e-06,
      "loss": 0.6118,
      "step": 2045
    },
    {
      "epoch": 0.6921515561569689,
      "grad_norm": 0.29890942573547363,
      "learning_rate": 4.416517836134836e-06,
      "loss": 0.5982,
      "step": 2046
    },
    {
      "epoch": 0.692489851150203,
      "grad_norm": 0.3120978772640228,
      "learning_rate": 4.415942533997133e-06,
      "loss": 0.6162,
      "step": 2047
    },
    {
      "epoch": 0.6928281461434371,
      "grad_norm": 0.30351585149765015,
      "learning_rate": 4.415366985889998e-06,
      "loss": 0.6016,
      "step": 2048
    },
    {
      "epoch": 0.6931664411366711,
      "grad_norm": 0.3094577491283417,
      "learning_rate": 4.4147911918873225e-06,
      "loss": 0.6405,
      "step": 2049
    },
    {
      "epoch": 0.6935047361299053,
      "grad_norm": 0.31087616086006165,
      "learning_rate": 4.414215152063023e-06,
      "loss": 0.5978,
      "step": 2050
    },
    {
      "epoch": 0.6938430311231394,
      "grad_norm": 0.30971506237983704,
      "learning_rate": 4.413638866491056e-06,
      "loss": 0.6391,
      "step": 2051
    },
    {
      "epoch": 0.6941813261163735,
      "grad_norm": 0.2959864139556885,
      "learning_rate": 4.413062335245402e-06,
      "loss": 0.6182,
      "step": 2052
    },
    {
      "epoch": 0.6945196211096076,
      "grad_norm": 0.30373695492744446,
      "learning_rate": 4.4124855584000766e-06,
      "loss": 0.6128,
      "step": 2053
    },
    {
      "epoch": 0.6948579161028416,
      "grad_norm": 0.3113371431827545,
      "learning_rate": 4.41190853602913e-06,
      "loss": 0.6014,
      "step": 2054
    },
    {
      "epoch": 0.6951962110960758,
      "grad_norm": 0.30726322531700134,
      "learning_rate": 4.411331268206637e-06,
      "loss": 0.6317,
      "step": 2055
    },
    {
      "epoch": 0.6955345060893099,
      "grad_norm": 0.31048333644866943,
      "learning_rate": 4.410753755006708e-06,
      "loss": 0.6007,
      "step": 2056
    },
    {
      "epoch": 0.695872801082544,
      "grad_norm": 0.3138802647590637,
      "learning_rate": 4.410175996503485e-06,
      "loss": 0.6174,
      "step": 2057
    },
    {
      "epoch": 0.696211096075778,
      "grad_norm": 0.2968793511390686,
      "learning_rate": 4.409597992771141e-06,
      "loss": 0.6037,
      "step": 2058
    },
    {
      "epoch": 0.6965493910690121,
      "grad_norm": 0.3008846640586853,
      "learning_rate": 4.40901974388388e-06,
      "loss": 0.6303,
      "step": 2059
    },
    {
      "epoch": 0.6968876860622463,
      "grad_norm": 0.32319778203964233,
      "learning_rate": 4.408441249915938e-06,
      "loss": 0.6435,
      "step": 2060
    },
    {
      "epoch": 0.6972259810554804,
      "grad_norm": 0.3118629455566406,
      "learning_rate": 4.4078625109415825e-06,
      "loss": 0.6247,
      "step": 2061
    },
    {
      "epoch": 0.6975642760487145,
      "grad_norm": 0.296186238527298,
      "learning_rate": 4.407283527035112e-06,
      "loss": 0.6303,
      "step": 2062
    },
    {
      "epoch": 0.6979025710419485,
      "grad_norm": 0.31817618012428284,
      "learning_rate": 4.406704298270856e-06,
      "loss": 0.6305,
      "step": 2063
    },
    {
      "epoch": 0.6982408660351827,
      "grad_norm": 0.3056838810443878,
      "learning_rate": 4.406124824723178e-06,
      "loss": 0.6104,
      "step": 2064
    },
    {
      "epoch": 0.6985791610284168,
      "grad_norm": 0.3022473156452179,
      "learning_rate": 4.405545106466469e-06,
      "loss": 0.6289,
      "step": 2065
    },
    {
      "epoch": 0.6989174560216509,
      "grad_norm": 0.2971550524234772,
      "learning_rate": 4.404965143575154e-06,
      "loss": 0.6225,
      "step": 2066
    },
    {
      "epoch": 0.699255751014885,
      "grad_norm": 0.3262900412082672,
      "learning_rate": 4.404384936123689e-06,
      "loss": 0.6323,
      "step": 2067
    },
    {
      "epoch": 0.699594046008119,
      "grad_norm": 0.29921954870224,
      "learning_rate": 4.403804484186561e-06,
      "loss": 0.6421,
      "step": 2068
    },
    {
      "epoch": 0.6999323410013532,
      "grad_norm": 0.296154260635376,
      "learning_rate": 4.40322378783829e-06,
      "loss": 0.6158,
      "step": 2069
    },
    {
      "epoch": 0.7002706359945873,
      "grad_norm": 0.2893257737159729,
      "learning_rate": 4.402642847153424e-06,
      "loss": 0.6101,
      "step": 2070
    },
    {
      "epoch": 0.7006089309878214,
      "grad_norm": 0.2974931001663208,
      "learning_rate": 4.402061662206546e-06,
      "loss": 0.6193,
      "step": 2071
    },
    {
      "epoch": 0.7009472259810555,
      "grad_norm": 0.30189311504364014,
      "learning_rate": 4.401480233072269e-06,
      "loss": 0.6276,
      "step": 2072
    },
    {
      "epoch": 0.7012855209742895,
      "grad_norm": 0.31920090317726135,
      "learning_rate": 4.400898559825234e-06,
      "loss": 0.6193,
      "step": 2073
    },
    {
      "epoch": 0.7016238159675237,
      "grad_norm": 0.31290367245674133,
      "learning_rate": 4.40031664254012e-06,
      "loss": 0.6413,
      "step": 2074
    },
    {
      "epoch": 0.7019621109607578,
      "grad_norm": 0.3053489029407501,
      "learning_rate": 4.399734481291632e-06,
      "loss": 0.5726,
      "step": 2075
    },
    {
      "epoch": 0.7023004059539919,
      "grad_norm": 0.29699379205703735,
      "learning_rate": 4.399152076154509e-06,
      "loss": 0.5966,
      "step": 2076
    },
    {
      "epoch": 0.702638700947226,
      "grad_norm": 0.29986077547073364,
      "learning_rate": 4.39856942720352e-06,
      "loss": 0.6061,
      "step": 2077
    },
    {
      "epoch": 0.70297699594046,
      "grad_norm": 0.30691906809806824,
      "learning_rate": 4.397986534513465e-06,
      "loss": 0.6097,
      "step": 2078
    },
    {
      "epoch": 0.7033152909336942,
      "grad_norm": 0.2931588888168335,
      "learning_rate": 4.397403398159176e-06,
      "loss": 0.5868,
      "step": 2079
    },
    {
      "epoch": 0.7036535859269283,
      "grad_norm": 0.29350966215133667,
      "learning_rate": 4.396820018215518e-06,
      "loss": 0.6257,
      "step": 2080
    },
    {
      "epoch": 0.7039918809201624,
      "grad_norm": 0.2993786036968231,
      "learning_rate": 4.396236394757384e-06,
      "loss": 0.6153,
      "step": 2081
    },
    {
      "epoch": 0.7043301759133965,
      "grad_norm": 0.31493711471557617,
      "learning_rate": 4.3956525278597e-06,
      "loss": 0.5981,
      "step": 2082
    },
    {
      "epoch": 0.7046684709066305,
      "grad_norm": 0.3099541664123535,
      "learning_rate": 4.395068417597423e-06,
      "loss": 0.6051,
      "step": 2083
    },
    {
      "epoch": 0.7050067658998647,
      "grad_norm": 0.30024874210357666,
      "learning_rate": 4.394484064045542e-06,
      "loss": 0.6399,
      "step": 2084
    },
    {
      "epoch": 0.7053450608930988,
      "grad_norm": 0.2888277471065521,
      "learning_rate": 4.393899467279076e-06,
      "loss": 0.6009,
      "step": 2085
    },
    {
      "epoch": 0.7056833558863329,
      "grad_norm": 0.31682926416397095,
      "learning_rate": 4.393314627373075e-06,
      "loss": 0.5759,
      "step": 2086
    },
    {
      "epoch": 0.706021650879567,
      "grad_norm": 0.2965947687625885,
      "learning_rate": 4.392729544402623e-06,
      "loss": 0.6215,
      "step": 2087
    },
    {
      "epoch": 0.706359945872801,
      "grad_norm": 0.2999950349330902,
      "learning_rate": 4.392144218442831e-06,
      "loss": 0.5964,
      "step": 2088
    },
    {
      "epoch": 0.7066982408660352,
      "grad_norm": 0.30933696031570435,
      "learning_rate": 4.391558649568844e-06,
      "loss": 0.5839,
      "step": 2089
    },
    {
      "epoch": 0.7070365358592693,
      "grad_norm": 0.2939186990261078,
      "learning_rate": 4.390972837855839e-06,
      "loss": 0.606,
      "step": 2090
    },
    {
      "epoch": 0.7073748308525034,
      "grad_norm": 0.3039042353630066,
      "learning_rate": 4.390386783379021e-06,
      "loss": 0.6371,
      "step": 2091
    },
    {
      "epoch": 0.7077131258457375,
      "grad_norm": 0.32648152112960815,
      "learning_rate": 4.389800486213629e-06,
      "loss": 0.641,
      "step": 2092
    },
    {
      "epoch": 0.7080514208389715,
      "grad_norm": 0.3150835931301117,
      "learning_rate": 4.389213946434932e-06,
      "loss": 0.6315,
      "step": 2093
    },
    {
      "epoch": 0.7083897158322057,
      "grad_norm": 0.3053288757801056,
      "learning_rate": 4.388627164118228e-06,
      "loss": 0.6338,
      "step": 2094
    },
    {
      "epoch": 0.7087280108254398,
      "grad_norm": 0.3217431902885437,
      "learning_rate": 4.3880401393388515e-06,
      "loss": 0.6092,
      "step": 2095
    },
    {
      "epoch": 0.7090663058186739,
      "grad_norm": 0.3065360486507416,
      "learning_rate": 4.387452872172163e-06,
      "loss": 0.6318,
      "step": 2096
    },
    {
      "epoch": 0.709404600811908,
      "grad_norm": 0.31731465458869934,
      "learning_rate": 4.386865362693556e-06,
      "loss": 0.6102,
      "step": 2097
    },
    {
      "epoch": 0.709742895805142,
      "grad_norm": 0.31071433424949646,
      "learning_rate": 4.386277610978457e-06,
      "loss": 0.6446,
      "step": 2098
    },
    {
      "epoch": 0.7100811907983762,
      "grad_norm": 0.32428860664367676,
      "learning_rate": 4.38568961710232e-06,
      "loss": 0.6522,
      "step": 2099
    },
    {
      "epoch": 0.7104194857916103,
      "grad_norm": 0.3046785295009613,
      "learning_rate": 4.385101381140633e-06,
      "loss": 0.6293,
      "step": 2100
    },
    {
      "epoch": 0.7107577807848444,
      "grad_norm": 0.34561029076576233,
      "learning_rate": 4.384512903168913e-06,
      "loss": 0.6367,
      "step": 2101
    },
    {
      "epoch": 0.7110960757780784,
      "grad_norm": 0.3116186559200287,
      "learning_rate": 4.383924183262709e-06,
      "loss": 0.6268,
      "step": 2102
    },
    {
      "epoch": 0.7114343707713126,
      "grad_norm": 0.30118685960769653,
      "learning_rate": 4.383335221497602e-06,
      "loss": 0.6293,
      "step": 2103
    },
    {
      "epoch": 0.7117726657645467,
      "grad_norm": 0.3198152482509613,
      "learning_rate": 4.382746017949203e-06,
      "loss": 0.6086,
      "step": 2104
    },
    {
      "epoch": 0.7121109607577808,
      "grad_norm": 0.33087316155433655,
      "learning_rate": 4.382156572693154e-06,
      "loss": 0.6012,
      "step": 2105
    },
    {
      "epoch": 0.7124492557510149,
      "grad_norm": 0.29373490810394287,
      "learning_rate": 4.3815668858051274e-06,
      "loss": 0.6399,
      "step": 2106
    },
    {
      "epoch": 0.7127875507442489,
      "grad_norm": 0.35433363914489746,
      "learning_rate": 4.380976957360829e-06,
      "loss": 0.6371,
      "step": 2107
    },
    {
      "epoch": 0.713125845737483,
      "grad_norm": 0.3058205246925354,
      "learning_rate": 4.3803867874359925e-06,
      "loss": 0.6296,
      "step": 2108
    },
    {
      "epoch": 0.7134641407307172,
      "grad_norm": 0.297324001789093,
      "learning_rate": 4.379796376106384e-06,
      "loss": 0.6044,
      "step": 2109
    },
    {
      "epoch": 0.7138024357239513,
      "grad_norm": 0.31194958090782166,
      "learning_rate": 4.379205723447803e-06,
      "loss": 0.6151,
      "step": 2110
    },
    {
      "epoch": 0.7141407307171854,
      "grad_norm": 0.32881027460098267,
      "learning_rate": 4.3786148295360755e-06,
      "loss": 0.6278,
      "step": 2111
    },
    {
      "epoch": 0.7144790257104194,
      "grad_norm": 0.29545143246650696,
      "learning_rate": 4.3780236944470615e-06,
      "loss": 0.6063,
      "step": 2112
    },
    {
      "epoch": 0.7148173207036536,
      "grad_norm": 0.31030046939849854,
      "learning_rate": 4.377432318256651e-06,
      "loss": 0.6279,
      "step": 2113
    },
    {
      "epoch": 0.7151556156968877,
      "grad_norm": 0.31192392110824585,
      "learning_rate": 4.3768407010407655e-06,
      "loss": 0.6201,
      "step": 2114
    },
    {
      "epoch": 0.7154939106901218,
      "grad_norm": 0.3663148880004883,
      "learning_rate": 4.3762488428753565e-06,
      "loss": 0.6217,
      "step": 2115
    },
    {
      "epoch": 0.7158322056833559,
      "grad_norm": 0.29583805799484253,
      "learning_rate": 4.3756567438364075e-06,
      "loss": 0.6118,
      "step": 2116
    },
    {
      "epoch": 0.7161705006765899,
      "grad_norm": 0.32086995244026184,
      "learning_rate": 4.375064403999931e-06,
      "loss": 0.6265,
      "step": 2117
    },
    {
      "epoch": 0.7165087956698241,
      "grad_norm": 0.32434090971946716,
      "learning_rate": 4.3744718234419735e-06,
      "loss": 0.6001,
      "step": 2118
    },
    {
      "epoch": 0.7168470906630582,
      "grad_norm": 0.316076397895813,
      "learning_rate": 4.3738790022386105e-06,
      "loss": 0.6157,
      "step": 2119
    },
    {
      "epoch": 0.7171853856562923,
      "grad_norm": 0.3119267523288727,
      "learning_rate": 4.373285940465948e-06,
      "loss": 0.5994,
      "step": 2120
    },
    {
      "epoch": 0.7175236806495264,
      "grad_norm": 0.29969432950019836,
      "learning_rate": 4.372692638200124e-06,
      "loss": 0.6253,
      "step": 2121
    },
    {
      "epoch": 0.7178619756427604,
      "grad_norm": 0.3249845802783966,
      "learning_rate": 4.372099095517306e-06,
      "loss": 0.6368,
      "step": 2122
    },
    {
      "epoch": 0.7182002706359946,
      "grad_norm": 0.31541869044303894,
      "learning_rate": 4.371505312493694e-06,
      "loss": 0.6028,
      "step": 2123
    },
    {
      "epoch": 0.7185385656292287,
      "grad_norm": 0.3059319257736206,
      "learning_rate": 4.370911289205518e-06,
      "loss": 0.6261,
      "step": 2124
    },
    {
      "epoch": 0.7188768606224628,
      "grad_norm": 0.3207683265209198,
      "learning_rate": 4.370317025729039e-06,
      "loss": 0.617,
      "step": 2125
    },
    {
      "epoch": 0.7192151556156969,
      "grad_norm": 0.31733056902885437,
      "learning_rate": 4.369722522140548e-06,
      "loss": 0.5974,
      "step": 2126
    },
    {
      "epoch": 0.719553450608931,
      "grad_norm": 0.31061556935310364,
      "learning_rate": 4.369127778516369e-06,
      "loss": 0.6519,
      "step": 2127
    },
    {
      "epoch": 0.7198917456021651,
      "grad_norm": 0.3010721802711487,
      "learning_rate": 4.368532794932854e-06,
      "loss": 0.6385,
      "step": 2128
    },
    {
      "epoch": 0.7202300405953992,
      "grad_norm": 0.30090805888175964,
      "learning_rate": 4.367937571466388e-06,
      "loss": 0.6291,
      "step": 2129
    },
    {
      "epoch": 0.7205683355886333,
      "grad_norm": 0.3015844225883484,
      "learning_rate": 4.3673421081933865e-06,
      "loss": 0.6117,
      "step": 2130
    },
    {
      "epoch": 0.7209066305818674,
      "grad_norm": 0.2912375330924988,
      "learning_rate": 4.366746405190294e-06,
      "loss": 0.6218,
      "step": 2131
    },
    {
      "epoch": 0.7212449255751014,
      "grad_norm": 0.30901646614074707,
      "learning_rate": 4.366150462533588e-06,
      "loss": 0.6179,
      "step": 2132
    },
    {
      "epoch": 0.7215832205683356,
      "grad_norm": 0.31542983651161194,
      "learning_rate": 4.3655542802997755e-06,
      "loss": 0.6134,
      "step": 2133
    },
    {
      "epoch": 0.7219215155615697,
      "grad_norm": 0.3000514805316925,
      "learning_rate": 4.3649578585653944e-06,
      "loss": 0.6181,
      "step": 2134
    },
    {
      "epoch": 0.7222598105548038,
      "grad_norm": 0.30705004930496216,
      "learning_rate": 4.364361197407014e-06,
      "loss": 0.6206,
      "step": 2135
    },
    {
      "epoch": 0.7225981055480379,
      "grad_norm": 0.29016777873039246,
      "learning_rate": 4.363764296901234e-06,
      "loss": 0.5988,
      "step": 2136
    },
    {
      "epoch": 0.722936400541272,
      "grad_norm": 0.3026140034198761,
      "learning_rate": 4.363167157124685e-06,
      "loss": 0.6365,
      "step": 2137
    },
    {
      "epoch": 0.7232746955345061,
      "grad_norm": 0.3143080174922943,
      "learning_rate": 4.362569778154026e-06,
      "loss": 0.638,
      "step": 2138
    },
    {
      "epoch": 0.7236129905277402,
      "grad_norm": 0.30859896540641785,
      "learning_rate": 4.361972160065951e-06,
      "loss": 0.6199,
      "step": 2139
    },
    {
      "epoch": 0.7239512855209743,
      "grad_norm": 0.3060165047645569,
      "learning_rate": 4.361374302937182e-06,
      "loss": 0.6386,
      "step": 2140
    },
    {
      "epoch": 0.7242895805142084,
      "grad_norm": 0.30037814378738403,
      "learning_rate": 4.360776206844472e-06,
      "loss": 0.5968,
      "step": 2141
    },
    {
      "epoch": 0.7246278755074425,
      "grad_norm": 0.3116171061992645,
      "learning_rate": 4.360177871864604e-06,
      "loss": 0.6025,
      "step": 2142
    },
    {
      "epoch": 0.7249661705006766,
      "grad_norm": 0.30952051281929016,
      "learning_rate": 4.359579298074392e-06,
      "loss": 0.5999,
      "step": 2143
    },
    {
      "epoch": 0.7253044654939107,
      "grad_norm": 0.2957140803337097,
      "learning_rate": 4.358980485550684e-06,
      "loss": 0.6096,
      "step": 2144
    },
    {
      "epoch": 0.7256427604871448,
      "grad_norm": 0.2928706109523773,
      "learning_rate": 4.358381434370353e-06,
      "loss": 0.6406,
      "step": 2145
    },
    {
      "epoch": 0.725981055480379,
      "grad_norm": 0.2954714298248291,
      "learning_rate": 4.357782144610306e-06,
      "loss": 0.6224,
      "step": 2146
    },
    {
      "epoch": 0.726319350473613,
      "grad_norm": 0.2990542948246002,
      "learning_rate": 4.357182616347482e-06,
      "loss": 0.619,
      "step": 2147
    },
    {
      "epoch": 0.7266576454668471,
      "grad_norm": 0.29735761880874634,
      "learning_rate": 4.356582849658845e-06,
      "loss": 0.6148,
      "step": 2148
    },
    {
      "epoch": 0.7269959404600812,
      "grad_norm": 0.30296590924263,
      "learning_rate": 4.355982844621397e-06,
      "loss": 0.6442,
      "step": 2149
    },
    {
      "epoch": 0.7273342354533153,
      "grad_norm": 0.289024293422699,
      "learning_rate": 4.355382601312164e-06,
      "loss": 0.6015,
      "step": 2150
    },
    {
      "epoch": 0.7276725304465493,
      "grad_norm": 0.29794010519981384,
      "learning_rate": 4.354782119808208e-06,
      "loss": 0.6026,
      "step": 2151
    },
    {
      "epoch": 0.7280108254397835,
      "grad_norm": 0.3014563024044037,
      "learning_rate": 4.3541814001866175e-06,
      "loss": 0.6069,
      "step": 2152
    },
    {
      "epoch": 0.7283491204330176,
      "grad_norm": 0.31807276606559753,
      "learning_rate": 4.353580442524513e-06,
      "loss": 0.581,
      "step": 2153
    },
    {
      "epoch": 0.7286874154262517,
      "grad_norm": 0.2963697612285614,
      "learning_rate": 4.352979246899046e-06,
      "loss": 0.616,
      "step": 2154
    },
    {
      "epoch": 0.7290257104194858,
      "grad_norm": 0.29154714941978455,
      "learning_rate": 4.352377813387398e-06,
      "loss": 0.6223,
      "step": 2155
    },
    {
      "epoch": 0.7293640054127198,
      "grad_norm": 0.30207982659339905,
      "learning_rate": 4.351776142066782e-06,
      "loss": 0.5977,
      "step": 2156
    },
    {
      "epoch": 0.729702300405954,
      "grad_norm": 0.33041995763778687,
      "learning_rate": 4.35117423301444e-06,
      "loss": 0.6263,
      "step": 2157
    },
    {
      "epoch": 0.7300405953991881,
      "grad_norm": 0.3013479709625244,
      "learning_rate": 4.350572086307646e-06,
      "loss": 0.613,
      "step": 2158
    },
    {
      "epoch": 0.7303788903924222,
      "grad_norm": 0.29995208978652954,
      "learning_rate": 4.349969702023703e-06,
      "loss": 0.6212,
      "step": 2159
    },
    {
      "epoch": 0.7307171853856563,
      "grad_norm": 0.30284181237220764,
      "learning_rate": 4.349367080239946e-06,
      "loss": 0.5916,
      "step": 2160
    },
    {
      "epoch": 0.7310554803788903,
      "grad_norm": 0.30870866775512695,
      "learning_rate": 4.34876422103374e-06,
      "loss": 0.645,
      "step": 2161
    },
    {
      "epoch": 0.7313937753721245,
      "grad_norm": 0.3040004074573517,
      "learning_rate": 4.3481611244824794e-06,
      "loss": 0.5821,
      "step": 2162
    },
    {
      "epoch": 0.7317320703653586,
      "grad_norm": 0.30425241589546204,
      "learning_rate": 4.347557790663591e-06,
      "loss": 0.6022,
      "step": 2163
    },
    {
      "epoch": 0.7320703653585927,
      "grad_norm": 0.3005933165550232,
      "learning_rate": 4.34695421965453e-06,
      "loss": 0.613,
      "step": 2164
    },
    {
      "epoch": 0.7324086603518268,
      "grad_norm": 0.30306586623191833,
      "learning_rate": 4.346350411532784e-06,
      "loss": 0.6193,
      "step": 2165
    },
    {
      "epoch": 0.7327469553450608,
      "grad_norm": 0.32029709219932556,
      "learning_rate": 4.34574636637587e-06,
      "loss": 0.6395,
      "step": 2166
    },
    {
      "epoch": 0.733085250338295,
      "grad_norm": 0.2901228070259094,
      "learning_rate": 4.345142084261333e-06,
      "loss": 0.6225,
      "step": 2167
    },
    {
      "epoch": 0.7334235453315291,
      "grad_norm": 0.30677056312561035,
      "learning_rate": 4.344537565266755e-06,
      "loss": 0.6142,
      "step": 2168
    },
    {
      "epoch": 0.7337618403247632,
      "grad_norm": 0.3195612132549286,
      "learning_rate": 4.343932809469742e-06,
      "loss": 0.616,
      "step": 2169
    },
    {
      "epoch": 0.7341001353179973,
      "grad_norm": 0.3008933365345001,
      "learning_rate": 4.3433278169479346e-06,
      "loss": 0.6205,
      "step": 2170
    },
    {
      "epoch": 0.7344384303112313,
      "grad_norm": 0.29755228757858276,
      "learning_rate": 4.342722587778999e-06,
      "loss": 0.6119,
      "step": 2171
    },
    {
      "epoch": 0.7347767253044655,
      "grad_norm": 0.3149365186691284,
      "learning_rate": 4.342117122040637e-06,
      "loss": 0.6121,
      "step": 2172
    },
    {
      "epoch": 0.7351150202976996,
      "grad_norm": 0.29928284883499146,
      "learning_rate": 4.341511419810578e-06,
      "loss": 0.6049,
      "step": 2173
    },
    {
      "epoch": 0.7354533152909337,
      "grad_norm": 0.2991817891597748,
      "learning_rate": 4.340905481166583e-06,
      "loss": 0.6285,
      "step": 2174
    },
    {
      "epoch": 0.7357916102841678,
      "grad_norm": 0.30570951104164124,
      "learning_rate": 4.340299306186441e-06,
      "loss": 0.6381,
      "step": 2175
    },
    {
      "epoch": 0.7361299052774019,
      "grad_norm": 0.31654542684555054,
      "learning_rate": 4.339692894947975e-06,
      "loss": 0.6396,
      "step": 2176
    },
    {
      "epoch": 0.736468200270636,
      "grad_norm": 0.29424408078193665,
      "learning_rate": 4.339086247529034e-06,
      "loss": 0.6155,
      "step": 2177
    },
    {
      "epoch": 0.7368064952638701,
      "grad_norm": 0.30464109778404236,
      "learning_rate": 4.338479364007501e-06,
      "loss": 0.6512,
      "step": 2178
    },
    {
      "epoch": 0.7371447902571042,
      "grad_norm": 0.29115793108940125,
      "learning_rate": 4.3378722444612875e-06,
      "loss": 0.6139,
      "step": 2179
    },
    {
      "epoch": 0.7374830852503383,
      "grad_norm": 0.32231682538986206,
      "learning_rate": 4.337264888968337e-06,
      "loss": 0.6391,
      "step": 2180
    },
    {
      "epoch": 0.7378213802435724,
      "grad_norm": 0.29485422372817993,
      "learning_rate": 4.33665729760662e-06,
      "loss": 0.5805,
      "step": 2181
    },
    {
      "epoch": 0.7381596752368065,
      "grad_norm": 0.2966797351837158,
      "learning_rate": 4.3360494704541415e-06,
      "loss": 0.6164,
      "step": 2182
    },
    {
      "epoch": 0.7384979702300406,
      "grad_norm": 0.3220391869544983,
      "learning_rate": 4.335441407588933e-06,
      "loss": 0.6149,
      "step": 2183
    },
    {
      "epoch": 0.7388362652232747,
      "grad_norm": 0.31245046854019165,
      "learning_rate": 4.334833109089057e-06,
      "loss": 0.6176,
      "step": 2184
    },
    {
      "epoch": 0.7391745602165088,
      "grad_norm": 0.3080689013004303,
      "learning_rate": 4.33422457503261e-06,
      "loss": 0.626,
      "step": 2185
    },
    {
      "epoch": 0.7395128552097429,
      "grad_norm": 0.3042171895503998,
      "learning_rate": 4.333615805497713e-06,
      "loss": 0.6094,
      "step": 2186
    },
    {
      "epoch": 0.739851150202977,
      "grad_norm": 0.30676788091659546,
      "learning_rate": 4.333006800562521e-06,
      "loss": 0.6181,
      "step": 2187
    },
    {
      "epoch": 0.7401894451962111,
      "grad_norm": 0.2994125783443451,
      "learning_rate": 4.33239756030522e-06,
      "loss": 0.6035,
      "step": 2188
    },
    {
      "epoch": 0.7405277401894452,
      "grad_norm": 0.2977219223976135,
      "learning_rate": 4.331788084804022e-06,
      "loss": 0.6145,
      "step": 2189
    },
    {
      "epoch": 0.7408660351826793,
      "grad_norm": 0.3188292980194092,
      "learning_rate": 4.331178374137173e-06,
      "loss": 0.6049,
      "step": 2190
    },
    {
      "epoch": 0.7412043301759134,
      "grad_norm": 0.28873610496520996,
      "learning_rate": 4.330568428382948e-06,
      "loss": 0.5933,
      "step": 2191
    },
    {
      "epoch": 0.7415426251691475,
      "grad_norm": 0.30604344606399536,
      "learning_rate": 4.329958247619651e-06,
      "loss": 0.6132,
      "step": 2192
    },
    {
      "epoch": 0.7418809201623816,
      "grad_norm": 0.3060908913612366,
      "learning_rate": 4.329347831925618e-06,
      "loss": 0.6088,
      "step": 2193
    },
    {
      "epoch": 0.7422192151556157,
      "grad_norm": 0.29733046889305115,
      "learning_rate": 4.328737181379215e-06,
      "loss": 0.6181,
      "step": 2194
    },
    {
      "epoch": 0.7425575101488497,
      "grad_norm": 0.2979923188686371,
      "learning_rate": 4.3281262960588355e-06,
      "loss": 0.6164,
      "step": 2195
    },
    {
      "epoch": 0.7428958051420839,
      "grad_norm": 0.29377368092536926,
      "learning_rate": 4.327515176042908e-06,
      "loss": 0.6361,
      "step": 2196
    },
    {
      "epoch": 0.743234100135318,
      "grad_norm": 0.30877503752708435,
      "learning_rate": 4.326903821409885e-06,
      "loss": 0.6199,
      "step": 2197
    },
    {
      "epoch": 0.7435723951285521,
      "grad_norm": 0.3117124140262604,
      "learning_rate": 4.326292232238256e-06,
      "loss": 0.6265,
      "step": 2198
    },
    {
      "epoch": 0.7439106901217862,
      "grad_norm": 0.3066393733024597,
      "learning_rate": 4.325680408606535e-06,
      "loss": 0.6211,
      "step": 2199
    },
    {
      "epoch": 0.7442489851150202,
      "grad_norm": 0.299986332654953,
      "learning_rate": 4.325068350593269e-06,
      "loss": 0.5956,
      "step": 2200
    },
    {
      "epoch": 0.7445872801082544,
      "grad_norm": 0.2944273352622986,
      "learning_rate": 4.324456058277033e-06,
      "loss": 0.6349,
      "step": 2201
    },
    {
      "epoch": 0.7449255751014885,
      "grad_norm": 0.3012370467185974,
      "learning_rate": 4.323843531736434e-06,
      "loss": 0.629,
      "step": 2202
    },
    {
      "epoch": 0.7452638700947226,
      "grad_norm": 0.3050519526004791,
      "learning_rate": 4.323230771050109e-06,
      "loss": 0.6096,
      "step": 2203
    },
    {
      "epoch": 0.7456021650879567,
      "grad_norm": 0.30067360401153564,
      "learning_rate": 4.322617776296723e-06,
      "loss": 0.6296,
      "step": 2204
    },
    {
      "epoch": 0.7459404600811907,
      "grad_norm": 0.30389589071273804,
      "learning_rate": 4.322004547554974e-06,
      "loss": 0.6307,
      "step": 2205
    },
    {
      "epoch": 0.7462787550744249,
      "grad_norm": 0.3155769109725952,
      "learning_rate": 4.321391084903588e-06,
      "loss": 0.6216,
      "step": 2206
    },
    {
      "epoch": 0.746617050067659,
      "grad_norm": 0.3266087770462036,
      "learning_rate": 4.320777388421321e-06,
      "loss": 0.6411,
      "step": 2207
    },
    {
      "epoch": 0.7469553450608931,
      "grad_norm": 0.2882659435272217,
      "learning_rate": 4.320163458186961e-06,
      "loss": 0.6252,
      "step": 2208
    },
    {
      "epoch": 0.7472936400541272,
      "grad_norm": 0.3083643317222595,
      "learning_rate": 4.319549294279323e-06,
      "loss": 0.6268,
      "step": 2209
    },
    {
      "epoch": 0.7476319350473613,
      "grad_norm": 0.2978135645389557,
      "learning_rate": 4.318934896777255e-06,
      "loss": 0.6321,
      "step": 2210
    },
    {
      "epoch": 0.7479702300405954,
      "grad_norm": 0.2990574538707733,
      "learning_rate": 4.318320265759633e-06,
      "loss": 0.6463,
      "step": 2211
    },
    {
      "epoch": 0.7483085250338295,
      "grad_norm": 0.31218650937080383,
      "learning_rate": 4.317705401305362e-06,
      "loss": 0.616,
      "step": 2212
    },
    {
      "epoch": 0.7486468200270636,
      "grad_norm": 0.30175313353538513,
      "learning_rate": 4.317090303493381e-06,
      "loss": 0.6045,
      "step": 2213
    },
    {
      "epoch": 0.7489851150202977,
      "grad_norm": 0.2993379831314087,
      "learning_rate": 4.316474972402655e-06,
      "loss": 0.5937,
      "step": 2214
    },
    {
      "epoch": 0.7493234100135318,
      "grad_norm": 0.3067174255847931,
      "learning_rate": 4.315859408112182e-06,
      "loss": 0.6261,
      "step": 2215
    },
    {
      "epoch": 0.7496617050067659,
      "grad_norm": 0.3013573884963989,
      "learning_rate": 4.315243610700987e-06,
      "loss": 0.6173,
      "step": 2216
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31558793783187866,
      "learning_rate": 4.3146275802481254e-06,
      "loss": 0.6196,
      "step": 2217
    },
    {
      "epoch": 0.7503382949932341,
      "grad_norm": 0.32746657729148865,
      "learning_rate": 4.3140113168326856e-06,
      "loss": 0.6211,
      "step": 2218
    },
    {
      "epoch": 0.7506765899864682,
      "grad_norm": 0.2910456657409668,
      "learning_rate": 4.313394820533783e-06,
      "loss": 0.5915,
      "step": 2219
    },
    {
      "epoch": 0.7510148849797023,
      "grad_norm": 0.29211342334747314,
      "learning_rate": 4.312778091430563e-06,
      "loss": 0.6037,
      "step": 2220
    },
    {
      "epoch": 0.7513531799729364,
      "grad_norm": 0.2989952862262726,
      "learning_rate": 4.312161129602201e-06,
      "loss": 0.634,
      "step": 2221
    },
    {
      "epoch": 0.7516914749661705,
      "grad_norm": 0.3006637394428253,
      "learning_rate": 4.3115439351279045e-06,
      "loss": 0.6103,
      "step": 2222
    },
    {
      "epoch": 0.7520297699594046,
      "grad_norm": 0.3160168528556824,
      "learning_rate": 4.310926508086909e-06,
      "loss": 0.6193,
      "step": 2223
    },
    {
      "epoch": 0.7523680649526387,
      "grad_norm": 0.2798728048801422,
      "learning_rate": 4.3103088485584784e-06,
      "loss": 0.593,
      "step": 2224
    },
    {
      "epoch": 0.7527063599458728,
      "grad_norm": 0.31656160950660706,
      "learning_rate": 4.309690956621909e-06,
      "loss": 0.6264,
      "step": 2225
    },
    {
      "epoch": 0.7530446549391069,
      "grad_norm": 0.30439293384552,
      "learning_rate": 4.309072832356527e-06,
      "loss": 0.6268,
      "step": 2226
    },
    {
      "epoch": 0.753382949932341,
      "grad_norm": 0.294822096824646,
      "learning_rate": 4.308454475841686e-06,
      "loss": 0.6165,
      "step": 2227
    },
    {
      "epoch": 0.7537212449255751,
      "grad_norm": 0.3071642220020294,
      "learning_rate": 4.307835887156771e-06,
      "loss": 0.6089,
      "step": 2228
    },
    {
      "epoch": 0.7540595399188093,
      "grad_norm": 0.29992255568504333,
      "learning_rate": 4.307217066381198e-06,
      "loss": 0.6274,
      "step": 2229
    },
    {
      "epoch": 0.7543978349120433,
      "grad_norm": 0.3074096441268921,
      "learning_rate": 4.3065980135944095e-06,
      "loss": 0.6513,
      "step": 2230
    },
    {
      "epoch": 0.7547361299052774,
      "grad_norm": 0.3223959505558014,
      "learning_rate": 4.30597872887588e-06,
      "loss": 0.6385,
      "step": 2231
    },
    {
      "epoch": 0.7550744248985115,
      "grad_norm": 0.29632002115249634,
      "learning_rate": 4.305359212305115e-06,
      "loss": 0.5938,
      "step": 2232
    },
    {
      "epoch": 0.7554127198917456,
      "grad_norm": 0.29177916049957275,
      "learning_rate": 4.304739463961647e-06,
      "loss": 0.6213,
      "step": 2233
    },
    {
      "epoch": 0.7557510148849798,
      "grad_norm": 0.29772329330444336,
      "learning_rate": 4.304119483925041e-06,
      "loss": 0.6098,
      "step": 2234
    },
    {
      "epoch": 0.7560893098782138,
      "grad_norm": 0.3082590401172638,
      "learning_rate": 4.303499272274888e-06,
      "loss": 0.5982,
      "step": 2235
    },
    {
      "epoch": 0.7564276048714479,
      "grad_norm": 0.2956467866897583,
      "learning_rate": 4.302878829090813e-06,
      "loss": 0.6481,
      "step": 2236
    },
    {
      "epoch": 0.756765899864682,
      "grad_norm": 0.2921627461910248,
      "learning_rate": 4.302258154452468e-06,
      "loss": 0.6251,
      "step": 2237
    },
    {
      "epoch": 0.7571041948579161,
      "grad_norm": 0.32271668314933777,
      "learning_rate": 4.301637248439534e-06,
      "loss": 0.5771,
      "step": 2238
    },
    {
      "epoch": 0.7574424898511503,
      "grad_norm": 0.3013802170753479,
      "learning_rate": 4.301016111131726e-06,
      "loss": 0.584,
      "step": 2239
    },
    {
      "epoch": 0.7577807848443843,
      "grad_norm": 0.3379766345024109,
      "learning_rate": 4.300394742608785e-06,
      "loss": 0.5867,
      "step": 2240
    },
    {
      "epoch": 0.7581190798376184,
      "grad_norm": 0.3216727375984192,
      "learning_rate": 4.29977314295048e-06,
      "loss": 0.6012,
      "step": 2241
    },
    {
      "epoch": 0.7584573748308525,
      "grad_norm": 0.29771074652671814,
      "learning_rate": 4.299151312236615e-06,
      "loss": 0.6351,
      "step": 2242
    },
    {
      "epoch": 0.7587956698240866,
      "grad_norm": 0.2906126379966736,
      "learning_rate": 4.29852925054702e-06,
      "loss": 0.5994,
      "step": 2243
    },
    {
      "epoch": 0.7591339648173207,
      "grad_norm": 0.3101195991039276,
      "learning_rate": 4.2979069579615566e-06,
      "loss": 0.6384,
      "step": 2244
    },
    {
      "epoch": 0.7594722598105548,
      "grad_norm": 0.316148966550827,
      "learning_rate": 4.297284434560112e-06,
      "loss": 0.6193,
      "step": 2245
    },
    {
      "epoch": 0.7598105548037889,
      "grad_norm": 0.295263409614563,
      "learning_rate": 4.296661680422609e-06,
      "loss": 0.6227,
      "step": 2246
    },
    {
      "epoch": 0.760148849797023,
      "grad_norm": 0.2999385595321655,
      "learning_rate": 4.296038695628996e-06,
      "loss": 0.6089,
      "step": 2247
    },
    {
      "epoch": 0.7604871447902571,
      "grad_norm": 0.30205249786376953,
      "learning_rate": 4.295415480259251e-06,
      "loss": 0.6349,
      "step": 2248
    },
    {
      "epoch": 0.7608254397834912,
      "grad_norm": 0.29788750410079956,
      "learning_rate": 4.294792034393384e-06,
      "loss": 0.6112,
      "step": 2249
    },
    {
      "epoch": 0.7611637347767253,
      "grad_norm": 0.2968345284461975,
      "learning_rate": 4.294168358111433e-06,
      "loss": 0.614,
      "step": 2250
    },
    {
      "epoch": 0.7615020297699594,
      "grad_norm": 0.30285030603408813,
      "learning_rate": 4.293544451493465e-06,
      "loss": 0.5997,
      "step": 2251
    },
    {
      "epoch": 0.7618403247631935,
      "grad_norm": 0.2902609407901764,
      "learning_rate": 4.292920314619579e-06,
      "loss": 0.594,
      "step": 2252
    },
    {
      "epoch": 0.7621786197564276,
      "grad_norm": 0.31383830308914185,
      "learning_rate": 4.2922959475699e-06,
      "loss": 0.6165,
      "step": 2253
    },
    {
      "epoch": 0.7625169147496617,
      "grad_norm": 0.2986510694026947,
      "learning_rate": 4.291671350424586e-06,
      "loss": 0.6325,
      "step": 2254
    },
    {
      "epoch": 0.7628552097428958,
      "grad_norm": 0.30191484093666077,
      "learning_rate": 4.291046523263821e-06,
      "loss": 0.6332,
      "step": 2255
    },
    {
      "epoch": 0.7631935047361299,
      "grad_norm": 0.2976613938808441,
      "learning_rate": 4.290421466167822e-06,
      "loss": 0.6081,
      "step": 2256
    },
    {
      "epoch": 0.763531799729364,
      "grad_norm": 0.3022184371948242,
      "learning_rate": 4.289796179216835e-06,
      "loss": 0.6107,
      "step": 2257
    },
    {
      "epoch": 0.7638700947225981,
      "grad_norm": 0.29924991726875305,
      "learning_rate": 4.2891706624911335e-06,
      "loss": 0.6248,
      "step": 2258
    },
    {
      "epoch": 0.7642083897158322,
      "grad_norm": 0.29953059554100037,
      "learning_rate": 4.288544916071022e-06,
      "loss": 0.5876,
      "step": 2259
    },
    {
      "epoch": 0.7645466847090663,
      "grad_norm": 0.2986959218978882,
      "learning_rate": 4.287918940036831e-06,
      "loss": 0.61,
      "step": 2260
    },
    {
      "epoch": 0.7648849797023004,
      "grad_norm": 0.30967196822166443,
      "learning_rate": 4.287292734468928e-06,
      "loss": 0.6271,
      "step": 2261
    },
    {
      "epoch": 0.7652232746955345,
      "grad_norm": 0.3109535872936249,
      "learning_rate": 4.286666299447705e-06,
      "loss": 0.595,
      "step": 2262
    },
    {
      "epoch": 0.7655615696887687,
      "grad_norm": 0.3030044734477997,
      "learning_rate": 4.286039635053582e-06,
      "loss": 0.6136,
      "step": 2263
    },
    {
      "epoch": 0.7658998646820027,
      "grad_norm": 0.3069053292274475,
      "learning_rate": 4.28541274136701e-06,
      "loss": 0.6211,
      "step": 2264
    },
    {
      "epoch": 0.7662381596752368,
      "grad_norm": 0.30615657567977905,
      "learning_rate": 4.284785618468471e-06,
      "loss": 0.6138,
      "step": 2265
    },
    {
      "epoch": 0.7665764546684709,
      "grad_norm": 0.3103742003440857,
      "learning_rate": 4.284158266438476e-06,
      "loss": 0.6073,
      "step": 2266
    },
    {
      "epoch": 0.766914749661705,
      "grad_norm": 0.30274397134780884,
      "learning_rate": 4.283530685357564e-06,
      "loss": 0.6354,
      "step": 2267
    },
    {
      "epoch": 0.7672530446549392,
      "grad_norm": 0.31919652223587036,
      "learning_rate": 4.282902875306304e-06,
      "loss": 0.614,
      "step": 2268
    },
    {
      "epoch": 0.7675913396481732,
      "grad_norm": 0.28906887769699097,
      "learning_rate": 4.282274836365294e-06,
      "loss": 0.6114,
      "step": 2269
    },
    {
      "epoch": 0.7679296346414073,
      "grad_norm": 0.2996445894241333,
      "learning_rate": 4.281646568615163e-06,
      "loss": 0.6156,
      "step": 2270
    },
    {
      "epoch": 0.7682679296346414,
      "grad_norm": 0.3094944357872009,
      "learning_rate": 4.281018072136567e-06,
      "loss": 0.6188,
      "step": 2271
    },
    {
      "epoch": 0.7686062246278755,
      "grad_norm": 0.3552551567554474,
      "learning_rate": 4.280389347010193e-06,
      "loss": 0.6019,
      "step": 2272
    },
    {
      "epoch": 0.7689445196211097,
      "grad_norm": 0.2939181625843048,
      "learning_rate": 4.279760393316759e-06,
      "loss": 0.6293,
      "step": 2273
    },
    {
      "epoch": 0.7692828146143437,
      "grad_norm": 0.30944308638572693,
      "learning_rate": 4.279131211137007e-06,
      "loss": 0.6104,
      "step": 2274
    },
    {
      "epoch": 0.7696211096075778,
      "grad_norm": 0.3241821527481079,
      "learning_rate": 4.278501800551713e-06,
      "loss": 0.6269,
      "step": 2275
    },
    {
      "epoch": 0.7699594046008119,
      "grad_norm": 0.2961249053478241,
      "learning_rate": 4.277872161641682e-06,
      "loss": 0.6052,
      "step": 2276
    },
    {
      "epoch": 0.770297699594046,
      "grad_norm": 0.30989935994148254,
      "learning_rate": 4.277242294487746e-06,
      "loss": 0.6172,
      "step": 2277
    },
    {
      "epoch": 0.7706359945872802,
      "grad_norm": 0.2998683452606201,
      "learning_rate": 4.276612199170768e-06,
      "loss": 0.6073,
      "step": 2278
    },
    {
      "epoch": 0.7709742895805142,
      "grad_norm": 0.30103355646133423,
      "learning_rate": 4.27598187577164e-06,
      "loss": 0.6458,
      "step": 2279
    },
    {
      "epoch": 0.7713125845737483,
      "grad_norm": 0.31310510635375977,
      "learning_rate": 4.275351324371283e-06,
      "loss": 0.6023,
      "step": 2280
    },
    {
      "epoch": 0.7716508795669824,
      "grad_norm": 0.30720946192741394,
      "learning_rate": 4.274720545050648e-06,
      "loss": 0.6116,
      "step": 2281
    },
    {
      "epoch": 0.7719891745602165,
      "grad_norm": 0.30738264322280884,
      "learning_rate": 4.2740895378907135e-06,
      "loss": 0.6165,
      "step": 2282
    },
    {
      "epoch": 0.7723274695534507,
      "grad_norm": 0.32559826970100403,
      "learning_rate": 4.27345830297249e-06,
      "loss": 0.6144,
      "step": 2283
    },
    {
      "epoch": 0.7726657645466847,
      "grad_norm": 0.3149685859680176,
      "learning_rate": 4.272826840377014e-06,
      "loss": 0.6285,
      "step": 2284
    },
    {
      "epoch": 0.7730040595399188,
      "grad_norm": 0.30466142296791077,
      "learning_rate": 4.2721951501853544e-06,
      "loss": 0.6275,
      "step": 2285
    },
    {
      "epoch": 0.7733423545331529,
      "grad_norm": 0.30216631293296814,
      "learning_rate": 4.271563232478606e-06,
      "loss": 0.6071,
      "step": 2286
    },
    {
      "epoch": 0.773680649526387,
      "grad_norm": 0.33934286236763,
      "learning_rate": 4.270931087337897e-06,
      "loss": 0.6108,
      "step": 2287
    },
    {
      "epoch": 0.774018944519621,
      "grad_norm": 0.31253665685653687,
      "learning_rate": 4.270298714844381e-06,
      "loss": 0.6435,
      "step": 2288
    },
    {
      "epoch": 0.7743572395128552,
      "grad_norm": 0.292512983083725,
      "learning_rate": 4.269666115079242e-06,
      "loss": 0.6193,
      "step": 2289
    },
    {
      "epoch": 0.7746955345060893,
      "grad_norm": 0.32233932614326477,
      "learning_rate": 4.269033288123694e-06,
      "loss": 0.6033,
      "step": 2290
    },
    {
      "epoch": 0.7750338294993234,
      "grad_norm": 0.3083772361278534,
      "learning_rate": 4.268400234058979e-06,
      "loss": 0.6366,
      "step": 2291
    },
    {
      "epoch": 0.7753721244925575,
      "grad_norm": 0.30060258507728577,
      "learning_rate": 4.267766952966369e-06,
      "loss": 0.5928,
      "step": 2292
    },
    {
      "epoch": 0.7757104194857916,
      "grad_norm": 0.32433411478996277,
      "learning_rate": 4.267133444927166e-06,
      "loss": 0.6308,
      "step": 2293
    },
    {
      "epoch": 0.7760487144790257,
      "grad_norm": 0.29127776622772217,
      "learning_rate": 4.266499710022698e-06,
      "loss": 0.5992,
      "step": 2294
    },
    {
      "epoch": 0.7763870094722598,
      "grad_norm": 0.30229055881500244,
      "learning_rate": 4.2658657483343255e-06,
      "loss": 0.6312,
      "step": 2295
    },
    {
      "epoch": 0.7767253044654939,
      "grad_norm": 0.31788191199302673,
      "learning_rate": 4.2652315599434355e-06,
      "loss": 0.628,
      "step": 2296
    },
    {
      "epoch": 0.777063599458728,
      "grad_norm": 0.3147873282432556,
      "learning_rate": 4.264597144931448e-06,
      "loss": 0.6439,
      "step": 2297
    },
    {
      "epoch": 0.7774018944519621,
      "grad_norm": 0.3166278302669525,
      "learning_rate": 4.263962503379807e-06,
      "loss": 0.6432,
      "step": 2298
    },
    {
      "epoch": 0.7777401894451962,
      "grad_norm": 0.3072454035282135,
      "learning_rate": 4.263327635369988e-06,
      "loss": 0.6057,
      "step": 2299
    },
    {
      "epoch": 0.7780784844384303,
      "grad_norm": 0.3119280934333801,
      "learning_rate": 4.262692540983496e-06,
      "loss": 0.6412,
      "step": 2300
    },
    {
      "epoch": 0.7784167794316644,
      "grad_norm": 0.30572327971458435,
      "learning_rate": 4.262057220301866e-06,
      "loss": 0.6125,
      "step": 2301
    },
    {
      "epoch": 0.7787550744248986,
      "grad_norm": 0.30425435304641724,
      "learning_rate": 4.261421673406659e-06,
      "loss": 0.6278,
      "step": 2302
    },
    {
      "epoch": 0.7790933694181326,
      "grad_norm": 0.2952420711517334,
      "learning_rate": 4.260785900379468e-06,
      "loss": 0.6134,
      "step": 2303
    },
    {
      "epoch": 0.7794316644113667,
      "grad_norm": 0.3194384276866913,
      "learning_rate": 4.260149901301913e-06,
      "loss": 0.6092,
      "step": 2304
    },
    {
      "epoch": 0.7797699594046008,
      "grad_norm": 0.3062346279621124,
      "learning_rate": 4.2595136762556435e-06,
      "loss": 0.6025,
      "step": 2305
    },
    {
      "epoch": 0.7801082543978349,
      "grad_norm": 0.3019559979438782,
      "learning_rate": 4.258877225322339e-06,
      "loss": 0.5973,
      "step": 2306
    },
    {
      "epoch": 0.780446549391069,
      "grad_norm": 0.2972407341003418,
      "learning_rate": 4.258240548583707e-06,
      "loss": 0.6136,
      "step": 2307
    },
    {
      "epoch": 0.7807848443843031,
      "grad_norm": 0.2973707914352417,
      "learning_rate": 4.2576036461214844e-06,
      "loss": 0.6184,
      "step": 2308
    },
    {
      "epoch": 0.7811231393775372,
      "grad_norm": 0.30419808626174927,
      "learning_rate": 4.256966518017437e-06,
      "loss": 0.6301,
      "step": 2309
    },
    {
      "epoch": 0.7814614343707713,
      "grad_norm": 0.2952224612236023,
      "learning_rate": 4.25632916435336e-06,
      "loss": 0.6127,
      "step": 2310
    },
    {
      "epoch": 0.7817997293640054,
      "grad_norm": 0.3098551630973816,
      "learning_rate": 4.2556915852110756e-06,
      "loss": 0.6042,
      "step": 2311
    },
    {
      "epoch": 0.7821380243572396,
      "grad_norm": 0.31795603036880493,
      "learning_rate": 4.255053780672438e-06,
      "loss": 0.6115,
      "step": 2312
    },
    {
      "epoch": 0.7824763193504736,
      "grad_norm": 0.29997360706329346,
      "learning_rate": 4.254415750819328e-06,
      "loss": 0.6205,
      "step": 2313
    },
    {
      "epoch": 0.7828146143437077,
      "grad_norm": 0.31599006056785583,
      "learning_rate": 4.253777495733658e-06,
      "loss": 0.628,
      "step": 2314
    },
    {
      "epoch": 0.7831529093369418,
      "grad_norm": 0.30331894755363464,
      "learning_rate": 4.253139015497364e-06,
      "loss": 0.635,
      "step": 2315
    },
    {
      "epoch": 0.7834912043301759,
      "grad_norm": 0.2977330684661865,
      "learning_rate": 4.252500310192417e-06,
      "loss": 0.6199,
      "step": 2316
    },
    {
      "epoch": 0.7838294993234101,
      "grad_norm": 0.2929128110408783,
      "learning_rate": 4.251861379900813e-06,
      "loss": 0.6375,
      "step": 2317
    },
    {
      "epoch": 0.7841677943166441,
      "grad_norm": 0.2989046573638916,
      "learning_rate": 4.251222224704578e-06,
      "loss": 0.6071,
      "step": 2318
    },
    {
      "epoch": 0.7845060893098782,
      "grad_norm": 0.28972285985946655,
      "learning_rate": 4.2505828446857685e-06,
      "loss": 0.6161,
      "step": 2319
    },
    {
      "epoch": 0.7848443843031123,
      "grad_norm": 0.3031672239303589,
      "learning_rate": 4.249943239926467e-06,
      "loss": 0.6261,
      "step": 2320
    },
    {
      "epoch": 0.7851826792963464,
      "grad_norm": 0.30067288875579834,
      "learning_rate": 4.249303410508787e-06,
      "loss": 0.6453,
      "step": 2321
    },
    {
      "epoch": 0.7855209742895806,
      "grad_norm": 0.2904297113418579,
      "learning_rate": 4.248663356514869e-06,
      "loss": 0.6202,
      "step": 2322
    },
    {
      "epoch": 0.7858592692828146,
      "grad_norm": 0.3080545663833618,
      "learning_rate": 4.248023078026884e-06,
      "loss": 0.6249,
      "step": 2323
    },
    {
      "epoch": 0.7861975642760487,
      "grad_norm": 0.29530856013298035,
      "learning_rate": 4.247382575127031e-06,
      "loss": 0.6149,
      "step": 2324
    },
    {
      "epoch": 0.7865358592692828,
      "grad_norm": 0.30119362473487854,
      "learning_rate": 4.246741847897538e-06,
      "loss": 0.609,
      "step": 2325
    },
    {
      "epoch": 0.786874154262517,
      "grad_norm": 0.32412683963775635,
      "learning_rate": 4.2461008964206626e-06,
      "loss": 0.6127,
      "step": 2326
    },
    {
      "epoch": 0.7872124492557511,
      "grad_norm": 0.298534095287323,
      "learning_rate": 4.24545972077869e-06,
      "loss": 0.6073,
      "step": 2327
    },
    {
      "epoch": 0.7875507442489851,
      "grad_norm": 0.30755627155303955,
      "learning_rate": 4.244818321053933e-06,
      "loss": 0.6207,
      "step": 2328
    },
    {
      "epoch": 0.7878890392422192,
      "grad_norm": 0.30254215002059937,
      "learning_rate": 4.244176697328738e-06,
      "loss": 0.6297,
      "step": 2329
    },
    {
      "epoch": 0.7882273342354533,
      "grad_norm": 0.3068731427192688,
      "learning_rate": 4.243534849685473e-06,
      "loss": 0.6512,
      "step": 2330
    },
    {
      "epoch": 0.7885656292286874,
      "grad_norm": 0.3131106197834015,
      "learning_rate": 4.242892778206542e-06,
      "loss": 0.628,
      "step": 2331
    },
    {
      "epoch": 0.7889039242219216,
      "grad_norm": 0.2996361255645752,
      "learning_rate": 4.242250482974373e-06,
      "loss": 0.5914,
      "step": 2332
    },
    {
      "epoch": 0.7892422192151556,
      "grad_norm": 0.3127598762512207,
      "learning_rate": 4.241607964071423e-06,
      "loss": 0.6097,
      "step": 2333
    },
    {
      "epoch": 0.7895805142083897,
      "grad_norm": 0.3156915605068207,
      "learning_rate": 4.240965221580181e-06,
      "loss": 0.625,
      "step": 2334
    },
    {
      "epoch": 0.7899188092016238,
      "grad_norm": 0.2918887138366699,
      "learning_rate": 4.24032225558316e-06,
      "loss": 0.6121,
      "step": 2335
    },
    {
      "epoch": 0.790257104194858,
      "grad_norm": 0.3149941563606262,
      "learning_rate": 4.239679066162908e-06,
      "loss": 0.583,
      "step": 2336
    },
    {
      "epoch": 0.790595399188092,
      "grad_norm": 0.3043084740638733,
      "learning_rate": 4.239035653401994e-06,
      "loss": 0.5879,
      "step": 2337
    },
    {
      "epoch": 0.7909336941813261,
      "grad_norm": 0.29296600818634033,
      "learning_rate": 4.238392017383021e-06,
      "loss": 0.5865,
      "step": 2338
    },
    {
      "epoch": 0.7912719891745602,
      "grad_norm": 0.3000551164150238,
      "learning_rate": 4.2377481581886195e-06,
      "loss": 0.6241,
      "step": 2339
    },
    {
      "epoch": 0.7916102841677943,
      "grad_norm": 0.3253296911716461,
      "learning_rate": 4.237104075901449e-06,
      "loss": 0.6208,
      "step": 2340
    },
    {
      "epoch": 0.7919485791610285,
      "grad_norm": 0.3103972375392914,
      "learning_rate": 4.236459770604196e-06,
      "loss": 0.6126,
      "step": 2341
    },
    {
      "epoch": 0.7922868741542625,
      "grad_norm": 0.29499924182891846,
      "learning_rate": 4.235815242379577e-06,
      "loss": 0.5787,
      "step": 2342
    },
    {
      "epoch": 0.7926251691474966,
      "grad_norm": 0.3240320086479187,
      "learning_rate": 4.235170491310336e-06,
      "loss": 0.6374,
      "step": 2343
    },
    {
      "epoch": 0.7929634641407307,
      "grad_norm": 0.3204994797706604,
      "learning_rate": 4.234525517479248e-06,
      "loss": 0.5895,
      "step": 2344
    },
    {
      "epoch": 0.7933017591339648,
      "grad_norm": 0.30940183997154236,
      "learning_rate": 4.233880320969114e-06,
      "loss": 0.6014,
      "step": 2345
    },
    {
      "epoch": 0.793640054127199,
      "grad_norm": 0.30504274368286133,
      "learning_rate": 4.233234901862765e-06,
      "loss": 0.6476,
      "step": 2346
    },
    {
      "epoch": 0.793978349120433,
      "grad_norm": 0.31054219603538513,
      "learning_rate": 4.232589260243058e-06,
      "loss": 0.645,
      "step": 2347
    },
    {
      "epoch": 0.7943166441136671,
      "grad_norm": 0.34609827399253845,
      "learning_rate": 4.231943396192885e-06,
      "loss": 0.6053,
      "step": 2348
    },
    {
      "epoch": 0.7946549391069012,
      "grad_norm": 0.32586246728897095,
      "learning_rate": 4.231297309795158e-06,
      "loss": 0.6133,
      "step": 2349
    },
    {
      "epoch": 0.7949932341001353,
      "grad_norm": 0.3187881112098694,
      "learning_rate": 4.2306510011328245e-06,
      "loss": 0.6034,
      "step": 2350
    },
    {
      "epoch": 0.7953315290933695,
      "grad_norm": 0.3788541257381439,
      "learning_rate": 4.230004470288857e-06,
      "loss": 0.5957,
      "step": 2351
    },
    {
      "epoch": 0.7956698240866035,
      "grad_norm": 0.3172380328178406,
      "learning_rate": 4.229357717346257e-06,
      "loss": 0.626,
      "step": 2352
    },
    {
      "epoch": 0.7960081190798376,
      "grad_norm": 0.3106147348880768,
      "learning_rate": 4.2287107423880555e-06,
      "loss": 0.6233,
      "step": 2353
    },
    {
      "epoch": 0.7963464140730717,
      "grad_norm": 0.32800528407096863,
      "learning_rate": 4.228063545497312e-06,
      "loss": 0.6105,
      "step": 2354
    },
    {
      "epoch": 0.7966847090663058,
      "grad_norm": 0.3131791353225708,
      "learning_rate": 4.227416126757112e-06,
      "loss": 0.5973,
      "step": 2355
    },
    {
      "epoch": 0.79702300405954,
      "grad_norm": 0.31193608045578003,
      "learning_rate": 4.226768486250573e-06,
      "loss": 0.6071,
      "step": 2356
    },
    {
      "epoch": 0.797361299052774,
      "grad_norm": 0.3071676194667816,
      "learning_rate": 4.226120624060838e-06,
      "loss": 0.6198,
      "step": 2357
    },
    {
      "epoch": 0.7976995940460081,
      "grad_norm": 0.3355953097343445,
      "learning_rate": 4.2254725402710806e-06,
      "loss": 0.6179,
      "step": 2358
    },
    {
      "epoch": 0.7980378890392422,
      "grad_norm": 0.297477662563324,
      "learning_rate": 4.224824234964502e-06,
      "loss": 0.6068,
      "step": 2359
    },
    {
      "epoch": 0.7983761840324763,
      "grad_norm": 0.3031257092952728,
      "learning_rate": 4.224175708224332e-06,
      "loss": 0.6184,
      "step": 2360
    },
    {
      "epoch": 0.7987144790257105,
      "grad_norm": 0.3067512810230255,
      "learning_rate": 4.223526960133829e-06,
      "loss": 0.6228,
      "step": 2361
    },
    {
      "epoch": 0.7990527740189445,
      "grad_norm": 0.32556161284446716,
      "learning_rate": 4.222877990776278e-06,
      "loss": 0.6174,
      "step": 2362
    },
    {
      "epoch": 0.7993910690121786,
      "grad_norm": 0.3196711838245392,
      "learning_rate": 4.222228800234997e-06,
      "loss": 0.6003,
      "step": 2363
    },
    {
      "epoch": 0.7997293640054127,
      "grad_norm": 0.30310988426208496,
      "learning_rate": 4.221579388593326e-06,
      "loss": 0.6244,
      "step": 2364
    },
    {
      "epoch": 0.8000676589986468,
      "grad_norm": 0.30188992619514465,
      "learning_rate": 4.220929755934638e-06,
      "loss": 0.6086,
      "step": 2365
    },
    {
      "epoch": 0.800405953991881,
      "grad_norm": 0.2987558841705322,
      "learning_rate": 4.220279902342334e-06,
      "loss": 0.6066,
      "step": 2366
    },
    {
      "epoch": 0.800744248985115,
      "grad_norm": 0.31522136926651,
      "learning_rate": 4.2196298278998405e-06,
      "loss": 0.6217,
      "step": 2367
    },
    {
      "epoch": 0.8010825439783491,
      "grad_norm": 0.3034696877002716,
      "learning_rate": 4.218979532690616e-06,
      "loss": 0.6142,
      "step": 2368
    },
    {
      "epoch": 0.8014208389715832,
      "grad_norm": 0.3127799332141876,
      "learning_rate": 4.218329016798145e-06,
      "loss": 0.6414,
      "step": 2369
    },
    {
      "epoch": 0.8017591339648173,
      "grad_norm": 0.3001839816570282,
      "learning_rate": 4.217678280305941e-06,
      "loss": 0.6126,
      "step": 2370
    },
    {
      "epoch": 0.8020974289580515,
      "grad_norm": 0.3111492395401001,
      "learning_rate": 4.2170273232975455e-06,
      "loss": 0.6222,
      "step": 2371
    },
    {
      "epoch": 0.8024357239512855,
      "grad_norm": 0.3090807795524597,
      "learning_rate": 4.216376145856529e-06,
      "loss": 0.63,
      "step": 2372
    },
    {
      "epoch": 0.8027740189445196,
      "grad_norm": 0.30586564540863037,
      "learning_rate": 4.215724748066491e-06,
      "loss": 0.5959,
      "step": 2373
    },
    {
      "epoch": 0.8031123139377537,
      "grad_norm": 0.30838629603385925,
      "learning_rate": 4.215073130011056e-06,
      "loss": 0.6192,
      "step": 2374
    },
    {
      "epoch": 0.8034506089309879,
      "grad_norm": 0.29361045360565186,
      "learning_rate": 4.21442129177388e-06,
      "loss": 0.5953,
      "step": 2375
    },
    {
      "epoch": 0.803788903924222,
      "grad_norm": 0.29793581366539,
      "learning_rate": 4.213769233438647e-06,
      "loss": 0.6129,
      "step": 2376
    },
    {
      "epoch": 0.804127198917456,
      "grad_norm": 0.3093686103820801,
      "learning_rate": 4.213116955089066e-06,
      "loss": 0.6105,
      "step": 2377
    },
    {
      "epoch": 0.8044654939106901,
      "grad_norm": 0.3063512444496155,
      "learning_rate": 4.21246445680888e-06,
      "loss": 0.5978,
      "step": 2378
    },
    {
      "epoch": 0.8048037889039242,
      "grad_norm": 0.31297993659973145,
      "learning_rate": 4.211811738681855e-06,
      "loss": 0.6156,
      "step": 2379
    },
    {
      "epoch": 0.8051420838971584,
      "grad_norm": 0.31399765610694885,
      "learning_rate": 4.211158800791788e-06,
      "loss": 0.6097,
      "step": 2380
    },
    {
      "epoch": 0.8054803788903924,
      "grad_norm": 0.2975424528121948,
      "learning_rate": 4.210505643222504e-06,
      "loss": 0.6481,
      "step": 2381
    },
    {
      "epoch": 0.8058186738836265,
      "grad_norm": 0.3167499303817749,
      "learning_rate": 4.209852266057853e-06,
      "loss": 0.6228,
      "step": 2382
    },
    {
      "epoch": 0.8061569688768606,
      "grad_norm": 0.31156760454177856,
      "learning_rate": 4.209198669381718e-06,
      "loss": 0.6102,
      "step": 2383
    },
    {
      "epoch": 0.8064952638700947,
      "grad_norm": 0.29924094676971436,
      "learning_rate": 4.208544853278008e-06,
      "loss": 0.5892,
      "step": 2384
    },
    {
      "epoch": 0.8068335588633289,
      "grad_norm": 0.30753546953201294,
      "learning_rate": 4.207890817830659e-06,
      "loss": 0.6128,
      "step": 2385
    },
    {
      "epoch": 0.8071718538565629,
      "grad_norm": 0.3182028532028198,
      "learning_rate": 4.207236563123638e-06,
      "loss": 0.6134,
      "step": 2386
    },
    {
      "epoch": 0.807510148849797,
      "grad_norm": 0.3230728507041931,
      "learning_rate": 4.206582089240936e-06,
      "loss": 0.6068,
      "step": 2387
    },
    {
      "epoch": 0.8078484438430311,
      "grad_norm": 0.3081299662590027,
      "learning_rate": 4.205927396266577e-06,
      "loss": 0.6069,
      "step": 2388
    },
    {
      "epoch": 0.8081867388362652,
      "grad_norm": 0.3078051507472992,
      "learning_rate": 4.20527248428461e-06,
      "loss": 0.6207,
      "step": 2389
    },
    {
      "epoch": 0.8085250338294994,
      "grad_norm": 0.30221062898635864,
      "learning_rate": 4.2046173533791126e-06,
      "loss": 0.6069,
      "step": 2390
    },
    {
      "epoch": 0.8088633288227334,
      "grad_norm": 0.3207438886165619,
      "learning_rate": 4.203962003634191e-06,
      "loss": 0.627,
      "step": 2391
    },
    {
      "epoch": 0.8092016238159675,
      "grad_norm": 0.30132848024368286,
      "learning_rate": 4.203306435133978e-06,
      "loss": 0.6253,
      "step": 2392
    },
    {
      "epoch": 0.8095399188092016,
      "grad_norm": 0.3093264102935791,
      "learning_rate": 4.2026506479626375e-06,
      "loss": 0.6189,
      "step": 2393
    },
    {
      "epoch": 0.8098782138024357,
      "grad_norm": 0.30848339200019836,
      "learning_rate": 4.201994642204358e-06,
      "loss": 0.6296,
      "step": 2394
    },
    {
      "epoch": 0.8102165087956699,
      "grad_norm": 0.3058469295501709,
      "learning_rate": 4.20133841794336e-06,
      "loss": 0.5895,
      "step": 2395
    },
    {
      "epoch": 0.8105548037889039,
      "grad_norm": 0.313781201839447,
      "learning_rate": 4.200681975263889e-06,
      "loss": 0.632,
      "step": 2396
    },
    {
      "epoch": 0.810893098782138,
      "grad_norm": 0.3138314187526703,
      "learning_rate": 4.2000253142502175e-06,
      "loss": 0.6164,
      "step": 2397
    },
    {
      "epoch": 0.8112313937753721,
      "grad_norm": 0.31005388498306274,
      "learning_rate": 4.199368434986651e-06,
      "loss": 0.6143,
      "step": 2398
    },
    {
      "epoch": 0.8115696887686062,
      "grad_norm": 0.29906076192855835,
      "learning_rate": 4.198711337557516e-06,
      "loss": 0.6004,
      "step": 2399
    },
    {
      "epoch": 0.8119079837618404,
      "grad_norm": 0.301923543214798,
      "learning_rate": 4.198054022047175e-06,
      "loss": 0.6091,
      "step": 2400
    },
    {
      "epoch": 0.8122462787550744,
      "grad_norm": 0.3076381981372833,
      "learning_rate": 4.197396488540011e-06,
      "loss": 0.5867,
      "step": 2401
    },
    {
      "epoch": 0.8125845737483085,
      "grad_norm": 0.3071296811103821,
      "learning_rate": 4.1967387371204415e-06,
      "loss": 0.5892,
      "step": 2402
    },
    {
      "epoch": 0.8129228687415426,
      "grad_norm": 0.30223989486694336,
      "learning_rate": 4.196080767872907e-06,
      "loss": 0.609,
      "step": 2403
    },
    {
      "epoch": 0.8132611637347767,
      "grad_norm": 0.2886689007282257,
      "learning_rate": 4.195422580881878e-06,
      "loss": 0.5915,
      "step": 2404
    },
    {
      "epoch": 0.8135994587280109,
      "grad_norm": 0.2973400354385376,
      "learning_rate": 4.194764176231852e-06,
      "loss": 0.6031,
      "step": 2405
    },
    {
      "epoch": 0.8139377537212449,
      "grad_norm": 0.3109757900238037,
      "learning_rate": 4.194105554007358e-06,
      "loss": 0.6461,
      "step": 2406
    },
    {
      "epoch": 0.814276048714479,
      "grad_norm": 0.32744237780570984,
      "learning_rate": 4.193446714292948e-06,
      "loss": 0.6348,
      "step": 2407
    },
    {
      "epoch": 0.8146143437077131,
      "grad_norm": 0.3049798309803009,
      "learning_rate": 4.192787657173204e-06,
      "loss": 0.6183,
      "step": 2408
    },
    {
      "epoch": 0.8149526387009473,
      "grad_norm": 0.30525657534599304,
      "learning_rate": 4.192128382732737e-06,
      "loss": 0.5968,
      "step": 2409
    },
    {
      "epoch": 0.8152909336941814,
      "grad_norm": 0.31621846556663513,
      "learning_rate": 4.191468891056184e-06,
      "loss": 0.6138,
      "step": 2410
    },
    {
      "epoch": 0.8156292286874154,
      "grad_norm": 0.2994174659252167,
      "learning_rate": 4.19080918222821e-06,
      "loss": 0.6164,
      "step": 2411
    },
    {
      "epoch": 0.8159675236806495,
      "grad_norm": 0.3103436529636383,
      "learning_rate": 4.1901492563335115e-06,
      "loss": 0.6322,
      "step": 2412
    },
    {
      "epoch": 0.8163058186738836,
      "grad_norm": 0.3064984381198883,
      "learning_rate": 4.189489113456808e-06,
      "loss": 0.5964,
      "step": 2413
    },
    {
      "epoch": 0.8166441136671178,
      "grad_norm": 0.30544745922088623,
      "learning_rate": 4.18882875368285e-06,
      "loss": 0.616,
      "step": 2414
    },
    {
      "epoch": 0.8169824086603519,
      "grad_norm": 0.31149283051490784,
      "learning_rate": 4.188168177096414e-06,
      "loss": 0.6021,
      "step": 2415
    },
    {
      "epoch": 0.8173207036535859,
      "grad_norm": 0.290450781583786,
      "learning_rate": 4.187507383782303e-06,
      "loss": 0.5883,
      "step": 2416
    },
    {
      "epoch": 0.81765899864682,
      "grad_norm": 0.31284961104393005,
      "learning_rate": 4.186846373825354e-06,
      "loss": 0.6312,
      "step": 2417
    },
    {
      "epoch": 0.8179972936400541,
      "grad_norm": 0.31574201583862305,
      "learning_rate": 4.186185147310425e-06,
      "loss": 0.6195,
      "step": 2418
    },
    {
      "epoch": 0.8183355886332883,
      "grad_norm": 0.30727875232696533,
      "learning_rate": 4.185523704322406e-06,
      "loss": 0.6322,
      "step": 2419
    },
    {
      "epoch": 0.8186738836265224,
      "grad_norm": 0.30774906277656555,
      "learning_rate": 4.1848620449462115e-06,
      "loss": 0.585,
      "step": 2420
    },
    {
      "epoch": 0.8190121786197564,
      "grad_norm": 0.3038913905620575,
      "learning_rate": 4.184200169266786e-06,
      "loss": 0.6191,
      "step": 2421
    },
    {
      "epoch": 0.8193504736129905,
      "grad_norm": 0.3066258430480957,
      "learning_rate": 4.183538077369104e-06,
      "loss": 0.6158,
      "step": 2422
    },
    {
      "epoch": 0.8196887686062246,
      "grad_norm": 0.3115321695804596,
      "learning_rate": 4.1828757693381625e-06,
      "loss": 0.6229,
      "step": 2423
    },
    {
      "epoch": 0.8200270635994588,
      "grad_norm": 0.29962000250816345,
      "learning_rate": 4.182213245258989e-06,
      "loss": 0.6226,
      "step": 2424
    },
    {
      "epoch": 0.8203653585926928,
      "grad_norm": 0.2973437011241913,
      "learning_rate": 4.181550505216639e-06,
      "loss": 0.6197,
      "step": 2425
    },
    {
      "epoch": 0.8207036535859269,
      "grad_norm": 0.31722259521484375,
      "learning_rate": 4.180887549296195e-06,
      "loss": 0.6128,
      "step": 2426
    },
    {
      "epoch": 0.821041948579161,
      "grad_norm": 0.2999642789363861,
      "learning_rate": 4.180224377582769e-06,
      "loss": 0.6114,
      "step": 2427
    },
    {
      "epoch": 0.8213802435723951,
      "grad_norm": 0.30792975425720215,
      "learning_rate": 4.179560990161496e-06,
      "loss": 0.59,
      "step": 2428
    },
    {
      "epoch": 0.8217185385656293,
      "grad_norm": 0.30365827679634094,
      "learning_rate": 4.178897387117547e-06,
      "loss": 0.6202,
      "step": 2429
    },
    {
      "epoch": 0.8220568335588633,
      "grad_norm": 0.29755693674087524,
      "learning_rate": 4.178233568536111e-06,
      "loss": 0.5968,
      "step": 2430
    },
    {
      "epoch": 0.8223951285520974,
      "grad_norm": 0.31009364128112793,
      "learning_rate": 4.17756953450241e-06,
      "loss": 0.616,
      "step": 2431
    },
    {
      "epoch": 0.8227334235453315,
      "grad_norm": 0.3053271770477295,
      "learning_rate": 4.176905285101696e-06,
      "loss": 0.5933,
      "step": 2432
    },
    {
      "epoch": 0.8230717185385656,
      "grad_norm": 0.3323591947555542,
      "learning_rate": 4.176240820419242e-06,
      "loss": 0.6047,
      "step": 2433
    },
    {
      "epoch": 0.8234100135317998,
      "grad_norm": 0.31013503670692444,
      "learning_rate": 4.175576140540354e-06,
      "loss": 0.6263,
      "step": 2434
    },
    {
      "epoch": 0.8237483085250338,
      "grad_norm": 0.30118370056152344,
      "learning_rate": 4.1749112455503635e-06,
      "loss": 0.6161,
      "step": 2435
    },
    {
      "epoch": 0.8240866035182679,
      "grad_norm": 0.29885560274124146,
      "learning_rate": 4.17424613553463e-06,
      "loss": 0.6328,
      "step": 2436
    },
    {
      "epoch": 0.824424898511502,
      "grad_norm": 0.30362245440483093,
      "learning_rate": 4.173580810578541e-06,
      "loss": 0.6067,
      "step": 2437
    },
    {
      "epoch": 0.8247631935047361,
      "grad_norm": 0.3012903928756714,
      "learning_rate": 4.17291527076751e-06,
      "loss": 0.609,
      "step": 2438
    },
    {
      "epoch": 0.8251014884979703,
      "grad_norm": 0.29213324189186096,
      "learning_rate": 4.172249516186982e-06,
      "loss": 0.6196,
      "step": 2439
    },
    {
      "epoch": 0.8254397834912043,
      "grad_norm": 0.3087073862552643,
      "learning_rate": 4.171583546922423e-06,
      "loss": 0.6174,
      "step": 2440
    },
    {
      "epoch": 0.8257780784844384,
      "grad_norm": 0.30654779076576233,
      "learning_rate": 4.170917363059333e-06,
      "loss": 0.5877,
      "step": 2441
    },
    {
      "epoch": 0.8261163734776725,
      "grad_norm": 0.2954830229282379,
      "learning_rate": 4.1702509646832355e-06,
      "loss": 0.6132,
      "step": 2442
    },
    {
      "epoch": 0.8264546684709067,
      "grad_norm": 0.2958783507347107,
      "learning_rate": 4.169584351879684e-06,
      "loss": 0.6359,
      "step": 2443
    },
    {
      "epoch": 0.8267929634641408,
      "grad_norm": 0.31136995553970337,
      "learning_rate": 4.168917524734259e-06,
      "loss": 0.633,
      "step": 2444
    },
    {
      "epoch": 0.8271312584573748,
      "grad_norm": 0.2963434159755707,
      "learning_rate": 4.168250483332565e-06,
      "loss": 0.6307,
      "step": 2445
    },
    {
      "epoch": 0.8274695534506089,
      "grad_norm": 0.3102148473262787,
      "learning_rate": 4.167583227760242e-06,
      "loss": 0.627,
      "step": 2446
    },
    {
      "epoch": 0.827807848443843,
      "grad_norm": 0.3160640001296997,
      "learning_rate": 4.166915758102947e-06,
      "loss": 0.6167,
      "step": 2447
    },
    {
      "epoch": 0.8281461434370772,
      "grad_norm": 0.3037266433238983,
      "learning_rate": 4.166248074446374e-06,
      "loss": 0.623,
      "step": 2448
    },
    {
      "epoch": 0.8284844384303113,
      "grad_norm": 0.30871710181236267,
      "learning_rate": 4.16558017687624e-06,
      "loss": 0.6234,
      "step": 2449
    },
    {
      "epoch": 0.8288227334235453,
      "grad_norm": 0.3070647120475769,
      "learning_rate": 4.164912065478288e-06,
      "loss": 0.5829,
      "step": 2450
    },
    {
      "epoch": 0.8291610284167794,
      "grad_norm": 0.3087005615234375,
      "learning_rate": 4.164243740338293e-06,
      "loss": 0.6116,
      "step": 2451
    },
    {
      "epoch": 0.8294993234100135,
      "grad_norm": 0.3228028416633606,
      "learning_rate": 4.163575201542052e-06,
      "loss": 0.6041,
      "step": 2452
    },
    {
      "epoch": 0.8298376184032477,
      "grad_norm": 0.31015029549598694,
      "learning_rate": 4.162906449175395e-06,
      "loss": 0.6173,
      "step": 2453
    },
    {
      "epoch": 0.8301759133964818,
      "grad_norm": 0.31407010555267334,
      "learning_rate": 4.162237483324174e-06,
      "loss": 0.6049,
      "step": 2454
    },
    {
      "epoch": 0.8305142083897158,
      "grad_norm": 0.30077072978019714,
      "learning_rate": 4.161568304074272e-06,
      "loss": 0.628,
      "step": 2455
    },
    {
      "epoch": 0.8308525033829499,
      "grad_norm": 0.31255874037742615,
      "learning_rate": 4.1608989115116005e-06,
      "loss": 0.5983,
      "step": 2456
    },
    {
      "epoch": 0.831190798376184,
      "grad_norm": 0.33020806312561035,
      "learning_rate": 4.160229305722094e-06,
      "loss": 0.5942,
      "step": 2457
    },
    {
      "epoch": 0.8315290933694182,
      "grad_norm": 0.31188005208969116,
      "learning_rate": 4.159559486791718e-06,
      "loss": 0.6375,
      "step": 2458
    },
    {
      "epoch": 0.8318673883626523,
      "grad_norm": 0.3300791084766388,
      "learning_rate": 4.158889454806464e-06,
      "loss": 0.5908,
      "step": 2459
    },
    {
      "epoch": 0.8322056833558863,
      "grad_norm": 0.3007451295852661,
      "learning_rate": 4.15821920985235e-06,
      "loss": 0.6051,
      "step": 2460
    },
    {
      "epoch": 0.8325439783491204,
      "grad_norm": 0.308019757270813,
      "learning_rate": 4.157548752015422e-06,
      "loss": 0.5893,
      "step": 2461
    },
    {
      "epoch": 0.8328822733423545,
      "grad_norm": 0.3098035454750061,
      "learning_rate": 4.156878081381756e-06,
      "loss": 0.6111,
      "step": 2462
    },
    {
      "epoch": 0.8332205683355887,
      "grad_norm": 0.2945764660835266,
      "learning_rate": 4.156207198037451e-06,
      "loss": 0.6372,
      "step": 2463
    },
    {
      "epoch": 0.8335588633288228,
      "grad_norm": 0.3222520053386688,
      "learning_rate": 4.155536102068636e-06,
      "loss": 0.6387,
      "step": 2464
    },
    {
      "epoch": 0.8338971583220568,
      "grad_norm": 0.3068825900554657,
      "learning_rate": 4.154864793561465e-06,
      "loss": 0.6057,
      "step": 2465
    },
    {
      "epoch": 0.8342354533152909,
      "grad_norm": 0.32351332902908325,
      "learning_rate": 4.154193272602124e-06,
      "loss": 0.6065,
      "step": 2466
    },
    {
      "epoch": 0.834573748308525,
      "grad_norm": 0.30833709239959717,
      "learning_rate": 4.15352153927682e-06,
      "loss": 0.5877,
      "step": 2467
    },
    {
      "epoch": 0.8349120433017592,
      "grad_norm": 0.3007957935333252,
      "learning_rate": 4.152849593671793e-06,
      "loss": 0.5955,
      "step": 2468
    },
    {
      "epoch": 0.8352503382949933,
      "grad_norm": 0.31723707914352417,
      "learning_rate": 4.152177435873305e-06,
      "loss": 0.64,
      "step": 2469
    },
    {
      "epoch": 0.8355886332882273,
      "grad_norm": 0.29329389333724976,
      "learning_rate": 4.15150506596765e-06,
      "loss": 0.6218,
      "step": 2470
    },
    {
      "epoch": 0.8359269282814614,
      "grad_norm": 0.31001290678977966,
      "learning_rate": 4.150832484041145e-06,
      "loss": 0.6336,
      "step": 2471
    },
    {
      "epoch": 0.8362652232746955,
      "grad_norm": 0.30639132857322693,
      "learning_rate": 4.150159690180139e-06,
      "loss": 0.6338,
      "step": 2472
    },
    {
      "epoch": 0.8366035182679297,
      "grad_norm": 0.3088569641113281,
      "learning_rate": 4.149486684471004e-06,
      "loss": 0.6046,
      "step": 2473
    },
    {
      "epoch": 0.8369418132611637,
      "grad_norm": 0.3085624873638153,
      "learning_rate": 4.1488134670001404e-06,
      "loss": 0.6125,
      "step": 2474
    },
    {
      "epoch": 0.8372801082543978,
      "grad_norm": 0.294872909784317,
      "learning_rate": 4.148140037853976e-06,
      "loss": 0.606,
      "step": 2475
    },
    {
      "epoch": 0.8376184032476319,
      "grad_norm": 0.29635724425315857,
      "learning_rate": 4.147466397118968e-06,
      "loss": 0.6239,
      "step": 2476
    },
    {
      "epoch": 0.837956698240866,
      "grad_norm": 0.31449973583221436,
      "learning_rate": 4.146792544881595e-06,
      "loss": 0.6107,
      "step": 2477
    },
    {
      "epoch": 0.8382949932341002,
      "grad_norm": 0.31982365250587463,
      "learning_rate": 4.146118481228371e-06,
      "loss": 0.6202,
      "step": 2478
    },
    {
      "epoch": 0.8386332882273342,
      "grad_norm": 0.30596596002578735,
      "learning_rate": 4.145444206245828e-06,
      "loss": 0.6054,
      "step": 2479
    },
    {
      "epoch": 0.8389715832205683,
      "grad_norm": 0.2982724905014038,
      "learning_rate": 4.144769720020533e-06,
      "loss": 0.5951,
      "step": 2480
    },
    {
      "epoch": 0.8393098782138024,
      "grad_norm": 0.29678595066070557,
      "learning_rate": 4.144095022639075e-06,
      "loss": 0.6166,
      "step": 2481
    },
    {
      "epoch": 0.8396481732070366,
      "grad_norm": 0.3216749131679535,
      "learning_rate": 4.1434201141880726e-06,
      "loss": 0.6016,
      "step": 2482
    },
    {
      "epoch": 0.8399864682002707,
      "grad_norm": 0.31580859422683716,
      "learning_rate": 4.142744994754171e-06,
      "loss": 0.5961,
      "step": 2483
    },
    {
      "epoch": 0.8403247631935047,
      "grad_norm": 0.3653385043144226,
      "learning_rate": 4.142069664424041e-06,
      "loss": 0.6222,
      "step": 2484
    },
    {
      "epoch": 0.8406630581867388,
      "grad_norm": 0.29260021448135376,
      "learning_rate": 4.141394123284383e-06,
      "loss": 0.6045,
      "step": 2485
    },
    {
      "epoch": 0.8410013531799729,
      "grad_norm": 0.3153046667575836,
      "learning_rate": 4.140718371421923e-06,
      "loss": 0.609,
      "step": 2486
    },
    {
      "epoch": 0.841339648173207,
      "grad_norm": 0.3311583697795868,
      "learning_rate": 4.140042408923415e-06,
      "loss": 0.5924,
      "step": 2487
    },
    {
      "epoch": 0.8416779431664412,
      "grad_norm": 0.31460800766944885,
      "learning_rate": 4.139366235875637e-06,
      "loss": 0.6028,
      "step": 2488
    },
    {
      "epoch": 0.8420162381596752,
      "grad_norm": 0.28972992300987244,
      "learning_rate": 4.1386898523654e-06,
      "loss": 0.608,
      "step": 2489
    },
    {
      "epoch": 0.8423545331529093,
      "grad_norm": 0.31510454416275024,
      "learning_rate": 4.138013258479535e-06,
      "loss": 0.6159,
      "step": 2490
    },
    {
      "epoch": 0.8426928281461434,
      "grad_norm": 0.2981686592102051,
      "learning_rate": 4.137336454304904e-06,
      "loss": 0.6174,
      "step": 2491
    },
    {
      "epoch": 0.8430311231393776,
      "grad_norm": 0.3189791738986969,
      "learning_rate": 4.136659439928397e-06,
      "loss": 0.6039,
      "step": 2492
    },
    {
      "epoch": 0.8433694181326117,
      "grad_norm": 0.31273263692855835,
      "learning_rate": 4.135982215436928e-06,
      "loss": 0.6068,
      "step": 2493
    },
    {
      "epoch": 0.8437077131258457,
      "grad_norm": 0.3063289523124695,
      "learning_rate": 4.135304780917439e-06,
      "loss": 0.609,
      "step": 2494
    },
    {
      "epoch": 0.8440460081190798,
      "grad_norm": 0.3077285885810852,
      "learning_rate": 4.134627136456902e-06,
      "loss": 0.619,
      "step": 2495
    },
    {
      "epoch": 0.8443843031123139,
      "grad_norm": 0.303930401802063,
      "learning_rate": 4.13394928214231e-06,
      "loss": 0.6134,
      "step": 2496
    },
    {
      "epoch": 0.8447225981055481,
      "grad_norm": 0.31286007165908813,
      "learning_rate": 4.1332712180606885e-06,
      "loss": 0.6041,
      "step": 2497
    },
    {
      "epoch": 0.8450608930987822,
      "grad_norm": 0.3026021420955658,
      "learning_rate": 4.1325929442990855e-06,
      "loss": 0.6058,
      "step": 2498
    },
    {
      "epoch": 0.8453991880920162,
      "grad_norm": 0.30549052357673645,
      "learning_rate": 4.131914460944579e-06,
      "loss": 0.6017,
      "step": 2499
    },
    {
      "epoch": 0.8457374830852503,
      "grad_norm": 0.3061017394065857,
      "learning_rate": 4.1312357680842735e-06,
      "loss": 0.581,
      "step": 2500
    },
    {
      "epoch": 0.8460757780784844,
      "grad_norm": 0.3075471818447113,
      "learning_rate": 4.1305568658053e-06,
      "loss": 0.6344,
      "step": 2501
    },
    {
      "epoch": 0.8464140730717186,
      "grad_norm": 0.29023295640945435,
      "learning_rate": 4.129877754194815e-06,
      "loss": 0.6298,
      "step": 2502
    },
    {
      "epoch": 0.8467523680649527,
      "grad_norm": 0.29860207438468933,
      "learning_rate": 4.129198433340004e-06,
      "loss": 0.6224,
      "step": 2503
    },
    {
      "epoch": 0.8470906630581867,
      "grad_norm": 0.29904088377952576,
      "learning_rate": 4.128518903328078e-06,
      "loss": 0.6118,
      "step": 2504
    },
    {
      "epoch": 0.8474289580514208,
      "grad_norm": 0.3165917694568634,
      "learning_rate": 4.1278391642462755e-06,
      "loss": 0.5659,
      "step": 2505
    },
    {
      "epoch": 0.847767253044655,
      "grad_norm": 0.30058714747428894,
      "learning_rate": 4.127159216181862e-06,
      "loss": 0.642,
      "step": 2506
    },
    {
      "epoch": 0.8481055480378891,
      "grad_norm": 0.31144291162490845,
      "learning_rate": 4.126479059222129e-06,
      "loss": 0.6078,
      "step": 2507
    },
    {
      "epoch": 0.8484438430311232,
      "grad_norm": 0.2997061610221863,
      "learning_rate": 4.125798693454396e-06,
      "loss": 0.6131,
      "step": 2508
    },
    {
      "epoch": 0.8487821380243572,
      "grad_norm": 0.29972556233406067,
      "learning_rate": 4.125118118966008e-06,
      "loss": 0.6195,
      "step": 2509
    },
    {
      "epoch": 0.8491204330175913,
      "grad_norm": 0.30682137608528137,
      "learning_rate": 4.124437335844337e-06,
      "loss": 0.6238,
      "step": 2510
    },
    {
      "epoch": 0.8494587280108254,
      "grad_norm": 0.3116617202758789,
      "learning_rate": 4.123756344176783e-06,
      "loss": 0.6114,
      "step": 2511
    },
    {
      "epoch": 0.8497970230040596,
      "grad_norm": 0.3055778443813324,
      "learning_rate": 4.123075144050772e-06,
      "loss": 0.5937,
      "step": 2512
    },
    {
      "epoch": 0.8501353179972937,
      "grad_norm": 0.31854158639907837,
      "learning_rate": 4.122393735553757e-06,
      "loss": 0.6331,
      "step": 2513
    },
    {
      "epoch": 0.8504736129905277,
      "grad_norm": 0.3081203103065491,
      "learning_rate": 4.121712118773216e-06,
      "loss": 0.6282,
      "step": 2514
    },
    {
      "epoch": 0.8508119079837618,
      "grad_norm": 0.31456461548805237,
      "learning_rate": 4.121030293796656e-06,
      "loss": 0.6218,
      "step": 2515
    },
    {
      "epoch": 0.851150202976996,
      "grad_norm": 0.3050321936607361,
      "learning_rate": 4.1203482607116105e-06,
      "loss": 0.6169,
      "step": 2516
    },
    {
      "epoch": 0.8514884979702301,
      "grad_norm": 0.30027446150779724,
      "learning_rate": 4.119666019605639e-06,
      "loss": 0.6361,
      "step": 2517
    },
    {
      "epoch": 0.8518267929634641,
      "grad_norm": 0.30657586455345154,
      "learning_rate": 4.118983570566328e-06,
      "loss": 0.6254,
      "step": 2518
    },
    {
      "epoch": 0.8521650879566982,
      "grad_norm": 0.3064988851547241,
      "learning_rate": 4.118300913681289e-06,
      "loss": 0.6033,
      "step": 2519
    },
    {
      "epoch": 0.8525033829499323,
      "grad_norm": 0.30324336886405945,
      "learning_rate": 4.117618049038165e-06,
      "loss": 0.6433,
      "step": 2520
    },
    {
      "epoch": 0.8528416779431665,
      "grad_norm": 0.3076494038105011,
      "learning_rate": 4.116934976724618e-06,
      "loss": 0.6265,
      "step": 2521
    },
    {
      "epoch": 0.8531799729364006,
      "grad_norm": 0.31042566895484924,
      "learning_rate": 4.116251696828345e-06,
      "loss": 0.6,
      "step": 2522
    },
    {
      "epoch": 0.8535182679296346,
      "grad_norm": 0.303231805562973,
      "learning_rate": 4.115568209437064e-06,
      "loss": 0.6164,
      "step": 2523
    },
    {
      "epoch": 0.8538565629228687,
      "grad_norm": 0.3145078420639038,
      "learning_rate": 4.114884514638522e-06,
      "loss": 0.6188,
      "step": 2524
    },
    {
      "epoch": 0.8541948579161028,
      "grad_norm": 0.308861643075943,
      "learning_rate": 4.11420061252049e-06,
      "loss": 0.6097,
      "step": 2525
    },
    {
      "epoch": 0.854533152909337,
      "grad_norm": 0.29943719506263733,
      "learning_rate": 4.11351650317077e-06,
      "loss": 0.6117,
      "step": 2526
    },
    {
      "epoch": 0.8548714479025711,
      "grad_norm": 0.3042926490306854,
      "learning_rate": 4.112832186677188e-06,
      "loss": 0.6499,
      "step": 2527
    },
    {
      "epoch": 0.8552097428958051,
      "grad_norm": 0.3146323561668396,
      "learning_rate": 4.112147663127597e-06,
      "loss": 0.5948,
      "step": 2528
    },
    {
      "epoch": 0.8555480378890392,
      "grad_norm": 0.3046211004257202,
      "learning_rate": 4.111462932609874e-06,
      "loss": 0.6268,
      "step": 2529
    },
    {
      "epoch": 0.8558863328822733,
      "grad_norm": 0.31243595480918884,
      "learning_rate": 4.110777995211926e-06,
      "loss": 0.6056,
      "step": 2530
    },
    {
      "epoch": 0.8562246278755075,
      "grad_norm": 0.31479743123054504,
      "learning_rate": 4.110092851021688e-06,
      "loss": 0.6181,
      "step": 2531
    },
    {
      "epoch": 0.8565629228687416,
      "grad_norm": 0.31839272379875183,
      "learning_rate": 4.1094075001271164e-06,
      "loss": 0.6067,
      "step": 2532
    },
    {
      "epoch": 0.8569012178619756,
      "grad_norm": 0.30680611729621887,
      "learning_rate": 4.108721942616197e-06,
      "loss": 0.5856,
      "step": 2533
    },
    {
      "epoch": 0.8572395128552097,
      "grad_norm": 0.3028895854949951,
      "learning_rate": 4.108036178576944e-06,
      "loss": 0.5957,
      "step": 2534
    },
    {
      "epoch": 0.8575778078484438,
      "grad_norm": 0.2930106818675995,
      "learning_rate": 4.107350208097393e-06,
      "loss": 0.6151,
      "step": 2535
    },
    {
      "epoch": 0.857916102841678,
      "grad_norm": 0.3081621825695038,
      "learning_rate": 4.106664031265611e-06,
      "loss": 0.6248,
      "step": 2536
    },
    {
      "epoch": 0.8582543978349121,
      "grad_norm": 0.3085257112979889,
      "learning_rate": 4.10597764816969e-06,
      "loss": 0.5873,
      "step": 2537
    },
    {
      "epoch": 0.8585926928281461,
      "grad_norm": 0.29894325137138367,
      "learning_rate": 4.105291058897747e-06,
      "loss": 0.5895,
      "step": 2538
    },
    {
      "epoch": 0.8589309878213802,
      "grad_norm": 0.30664926767349243,
      "learning_rate": 4.104604263537927e-06,
      "loss": 0.6536,
      "step": 2539
    },
    {
      "epoch": 0.8592692828146143,
      "grad_norm": 0.2995365560054779,
      "learning_rate": 4.103917262178402e-06,
      "loss": 0.6293,
      "step": 2540
    },
    {
      "epoch": 0.8596075778078485,
      "grad_norm": 0.30400002002716064,
      "learning_rate": 4.103230054907368e-06,
      "loss": 0.5964,
      "step": 2541
    },
    {
      "epoch": 0.8599458728010826,
      "grad_norm": 0.31295180320739746,
      "learning_rate": 4.10254264181305e-06,
      "loss": 0.6275,
      "step": 2542
    },
    {
      "epoch": 0.8602841677943166,
      "grad_norm": 0.29022055864334106,
      "learning_rate": 4.101855022983697e-06,
      "loss": 0.6004,
      "step": 2543
    },
    {
      "epoch": 0.8606224627875507,
      "grad_norm": 0.31110915541648865,
      "learning_rate": 4.101167198507587e-06,
      "loss": 0.6207,
      "step": 2544
    },
    {
      "epoch": 0.8609607577807848,
      "grad_norm": 0.31582343578338623,
      "learning_rate": 4.100479168473024e-06,
      "loss": 0.625,
      "step": 2545
    },
    {
      "epoch": 0.861299052774019,
      "grad_norm": 0.30717524886131287,
      "learning_rate": 4.099790932968335e-06,
      "loss": 0.6169,
      "step": 2546
    },
    {
      "epoch": 0.8616373477672531,
      "grad_norm": 0.30204007029533386,
      "learning_rate": 4.099102492081877e-06,
      "loss": 0.5991,
      "step": 2547
    },
    {
      "epoch": 0.8619756427604871,
      "grad_norm": 0.30182498693466187,
      "learning_rate": 4.098413845902033e-06,
      "loss": 0.6316,
      "step": 2548
    },
    {
      "epoch": 0.8623139377537212,
      "grad_norm": 0.2960222661495209,
      "learning_rate": 4.097724994517212e-06,
      "loss": 0.5938,
      "step": 2549
    },
    {
      "epoch": 0.8626522327469553,
      "grad_norm": 0.31207287311553955,
      "learning_rate": 4.0970359380158485e-06,
      "loss": 0.6287,
      "step": 2550
    },
    {
      "epoch": 0.8629905277401895,
      "grad_norm": 0.3209856152534485,
      "learning_rate": 4.096346676486403e-06,
      "loss": 0.6123,
      "step": 2551
    },
    {
      "epoch": 0.8633288227334236,
      "grad_norm": 0.3065491318702698,
      "learning_rate": 4.095657210017364e-06,
      "loss": 0.6063,
      "step": 2552
    },
    {
      "epoch": 0.8636671177266576,
      "grad_norm": 0.3043796718120575,
      "learning_rate": 4.094967538697245e-06,
      "loss": 0.616,
      "step": 2553
    },
    {
      "epoch": 0.8640054127198917,
      "grad_norm": 0.31421199440956116,
      "learning_rate": 4.094277662614586e-06,
      "loss": 0.6148,
      "step": 2554
    },
    {
      "epoch": 0.8643437077131259,
      "grad_norm": 0.30866771936416626,
      "learning_rate": 4.093587581857955e-06,
      "loss": 0.6283,
      "step": 2555
    },
    {
      "epoch": 0.86468200270636,
      "grad_norm": 0.3206896185874939,
      "learning_rate": 4.0928972965159445e-06,
      "loss": 0.6191,
      "step": 2556
    },
    {
      "epoch": 0.8650202976995941,
      "grad_norm": 0.318266898393631,
      "learning_rate": 4.092206806677172e-06,
      "loss": 0.5723,
      "step": 2557
    },
    {
      "epoch": 0.8653585926928281,
      "grad_norm": 0.3173162043094635,
      "learning_rate": 4.091516112430284e-06,
      "loss": 0.6346,
      "step": 2558
    },
    {
      "epoch": 0.8656968876860622,
      "grad_norm": 0.3083566427230835,
      "learning_rate": 4.090825213863953e-06,
      "loss": 0.6299,
      "step": 2559
    },
    {
      "epoch": 0.8660351826792964,
      "grad_norm": 0.2947406768798828,
      "learning_rate": 4.090134111066874e-06,
      "loss": 0.5935,
      "step": 2560
    },
    {
      "epoch": 0.8663734776725305,
      "grad_norm": 0.31538644433021545,
      "learning_rate": 4.089442804127773e-06,
      "loss": 0.6028,
      "step": 2561
    },
    {
      "epoch": 0.8667117726657646,
      "grad_norm": 0.32597365975379944,
      "learning_rate": 4.0887512931354016e-06,
      "loss": 0.5859,
      "step": 2562
    },
    {
      "epoch": 0.8670500676589986,
      "grad_norm": 0.30730724334716797,
      "learning_rate": 4.088059578178534e-06,
      "loss": 0.5936,
      "step": 2563
    },
    {
      "epoch": 0.8673883626522327,
      "grad_norm": 0.323289692401886,
      "learning_rate": 4.087367659345973e-06,
      "loss": 0.5876,
      "step": 2564
    },
    {
      "epoch": 0.8677266576454669,
      "grad_norm": 0.3257518708705902,
      "learning_rate": 4.086675536726548e-06,
      "loss": 0.6105,
      "step": 2565
    },
    {
      "epoch": 0.868064952638701,
      "grad_norm": 0.30332931876182556,
      "learning_rate": 4.085983210409114e-06,
      "loss": 0.6527,
      "step": 2566
    },
    {
      "epoch": 0.868403247631935,
      "grad_norm": 0.32552096247673035,
      "learning_rate": 4.085290680482552e-06,
      "loss": 0.6237,
      "step": 2567
    },
    {
      "epoch": 0.8687415426251691,
      "grad_norm": 0.31306159496307373,
      "learning_rate": 4.08459794703577e-06,
      "loss": 0.6318,
      "step": 2568
    },
    {
      "epoch": 0.8690798376184032,
      "grad_norm": 0.3016589283943176,
      "learning_rate": 4.0839050101577e-06,
      "loss": 0.5864,
      "step": 2569
    },
    {
      "epoch": 0.8694181326116374,
      "grad_norm": 0.3021901845932007,
      "learning_rate": 4.083211869937302e-06,
      "loss": 0.6214,
      "step": 2570
    },
    {
      "epoch": 0.8697564276048715,
      "grad_norm": 0.3040446639060974,
      "learning_rate": 4.082518526463562e-06,
      "loss": 0.6179,
      "step": 2571
    },
    {
      "epoch": 0.8700947225981055,
      "grad_norm": 0.31397125124931335,
      "learning_rate": 4.081824979825492e-06,
      "loss": 0.6121,
      "step": 2572
    },
    {
      "epoch": 0.8704330175913396,
      "grad_norm": 0.30847495794296265,
      "learning_rate": 4.081131230112128e-06,
      "loss": 0.6103,
      "step": 2573
    },
    {
      "epoch": 0.8707713125845737,
      "grad_norm": 0.2960346043109894,
      "learning_rate": 4.080437277412536e-06,
      "loss": 0.5899,
      "step": 2574
    },
    {
      "epoch": 0.8711096075778079,
      "grad_norm": 0.30658602714538574,
      "learning_rate": 4.079743121815805e-06,
      "loss": 0.5767,
      "step": 2575
    },
    {
      "epoch": 0.871447902571042,
      "grad_norm": 0.32288169860839844,
      "learning_rate": 4.07904876341105e-06,
      "loss": 0.6234,
      "step": 2576
    },
    {
      "epoch": 0.871786197564276,
      "grad_norm": 0.3071858584880829,
      "learning_rate": 4.0783542022874155e-06,
      "loss": 0.6106,
      "step": 2577
    },
    {
      "epoch": 0.8721244925575101,
      "grad_norm": 0.3182103931903839,
      "learning_rate": 4.0776594385340665e-06,
      "loss": 0.6347,
      "step": 2578
    },
    {
      "epoch": 0.8724627875507442,
      "grad_norm": 0.29850947856903076,
      "learning_rate": 4.0769644722402e-06,
      "loss": 0.5841,
      "step": 2579
    },
    {
      "epoch": 0.8728010825439784,
      "grad_norm": 0.33759528398513794,
      "learning_rate": 4.076269303495033e-06,
      "loss": 0.5971,
      "step": 2580
    },
    {
      "epoch": 0.8731393775372125,
      "grad_norm": 0.3135136067867279,
      "learning_rate": 4.075573932387814e-06,
      "loss": 0.6045,
      "step": 2581
    },
    {
      "epoch": 0.8734776725304465,
      "grad_norm": 0.30932480096817017,
      "learning_rate": 4.0748783590078136e-06,
      "loss": 0.6337,
      "step": 2582
    },
    {
      "epoch": 0.8738159675236806,
      "grad_norm": 0.3216150104999542,
      "learning_rate": 4.074182583444331e-06,
      "loss": 0.6145,
      "step": 2583
    },
    {
      "epoch": 0.8741542625169147,
      "grad_norm": 0.30273857712745667,
      "learning_rate": 4.073486605786689e-06,
      "loss": 0.5818,
      "step": 2584
    },
    {
      "epoch": 0.8744925575101489,
      "grad_norm": 0.31567350029945374,
      "learning_rate": 4.072790426124238e-06,
      "loss": 0.6204,
      "step": 2585
    },
    {
      "epoch": 0.874830852503383,
      "grad_norm": 0.31259241700172424,
      "learning_rate": 4.0720940445463536e-06,
      "loss": 0.6309,
      "step": 2586
    },
    {
      "epoch": 0.875169147496617,
      "grad_norm": 0.3071554899215698,
      "learning_rate": 4.071397461142437e-06,
      "loss": 0.6272,
      "step": 2587
    },
    {
      "epoch": 0.8755074424898511,
      "grad_norm": 0.31412243843078613,
      "learning_rate": 4.070700676001917e-06,
      "loss": 0.597,
      "step": 2588
    },
    {
      "epoch": 0.8758457374830853,
      "grad_norm": 0.3067740499973297,
      "learning_rate": 4.070003689214247e-06,
      "loss": 0.6275,
      "step": 2589
    },
    {
      "epoch": 0.8761840324763194,
      "grad_norm": 0.3077264130115509,
      "learning_rate": 4.0693065008689055e-06,
      "loss": 0.6126,
      "step": 2590
    },
    {
      "epoch": 0.8765223274695535,
      "grad_norm": 0.3074173033237457,
      "learning_rate": 4.0686091110553985e-06,
      "loss": 0.6069,
      "step": 2591
    },
    {
      "epoch": 0.8768606224627875,
      "grad_norm": 0.31100180745124817,
      "learning_rate": 4.067911519863257e-06,
      "loss": 0.6216,
      "step": 2592
    },
    {
      "epoch": 0.8771989174560216,
      "grad_norm": 0.3137838542461395,
      "learning_rate": 4.067213727382037e-06,
      "loss": 0.6112,
      "step": 2593
    },
    {
      "epoch": 0.8775372124492558,
      "grad_norm": 0.30964088439941406,
      "learning_rate": 4.066515733701325e-06,
      "loss": 0.6117,
      "step": 2594
    },
    {
      "epoch": 0.8778755074424899,
      "grad_norm": 0.31450891494750977,
      "learning_rate": 4.065817538910725e-06,
      "loss": 0.604,
      "step": 2595
    },
    {
      "epoch": 0.878213802435724,
      "grad_norm": 0.32473263144493103,
      "learning_rate": 4.065119143099874e-06,
      "loss": 0.6211,
      "step": 2596
    },
    {
      "epoch": 0.878552097428958,
      "grad_norm": 0.3195638358592987,
      "learning_rate": 4.0644205463584315e-06,
      "loss": 0.6167,
      "step": 2597
    },
    {
      "epoch": 0.8788903924221921,
      "grad_norm": 0.2934994399547577,
      "learning_rate": 4.063721748776085e-06,
      "loss": 0.5956,
      "step": 2598
    },
    {
      "epoch": 0.8792286874154263,
      "grad_norm": 0.32306793332099915,
      "learning_rate": 4.063022750442545e-06,
      "loss": 0.5955,
      "step": 2599
    },
    {
      "epoch": 0.8795669824086604,
      "grad_norm": 0.30271050333976746,
      "learning_rate": 4.062323551447549e-06,
      "loss": 0.6352,
      "step": 2600
    },
    {
      "epoch": 0.8799052774018945,
      "grad_norm": 0.30306926369667053,
      "learning_rate": 4.061624151880862e-06,
      "loss": 0.6152,
      "step": 2601
    },
    {
      "epoch": 0.8802435723951285,
      "grad_norm": 0.307574987411499,
      "learning_rate": 4.060924551832272e-06,
      "loss": 0.6023,
      "step": 2602
    },
    {
      "epoch": 0.8805818673883626,
      "grad_norm": 0.3127698600292206,
      "learning_rate": 4.060224751391593e-06,
      "loss": 0.6265,
      "step": 2603
    },
    {
      "epoch": 0.8809201623815968,
      "grad_norm": 0.29466134309768677,
      "learning_rate": 4.059524750648668e-06,
      "loss": 0.6319,
      "step": 2604
    },
    {
      "epoch": 0.8812584573748309,
      "grad_norm": 0.31846871972084045,
      "learning_rate": 4.058824549693362e-06,
      "loss": 0.6084,
      "step": 2605
    },
    {
      "epoch": 0.881596752368065,
      "grad_norm": 0.30549389123916626,
      "learning_rate": 4.058124148615567e-06,
      "loss": 0.6154,
      "step": 2606
    },
    {
      "epoch": 0.881935047361299,
      "grad_norm": 0.3262697756290436,
      "learning_rate": 4.0574235475052e-06,
      "loss": 0.6217,
      "step": 2607
    },
    {
      "epoch": 0.8822733423545331,
      "grad_norm": 0.33191269636154175,
      "learning_rate": 4.056722746452208e-06,
      "loss": 0.6125,
      "step": 2608
    },
    {
      "epoch": 0.8826116373477673,
      "grad_norm": 0.29111433029174805,
      "learning_rate": 4.056021745546555e-06,
      "loss": 0.6128,
      "step": 2609
    },
    {
      "epoch": 0.8829499323410014,
      "grad_norm": 0.3281172513961792,
      "learning_rate": 4.055320544878239e-06,
      "loss": 0.6013,
      "step": 2610
    },
    {
      "epoch": 0.8832882273342354,
      "grad_norm": 0.3282298147678375,
      "learning_rate": 4.054619144537281e-06,
      "loss": 0.592,
      "step": 2611
    },
    {
      "epoch": 0.8836265223274695,
      "grad_norm": 0.30221420526504517,
      "learning_rate": 4.053917544613723e-06,
      "loss": 0.6152,
      "step": 2612
    },
    {
      "epoch": 0.8839648173207036,
      "grad_norm": 0.3031254708766937,
      "learning_rate": 4.053215745197642e-06,
      "loss": 0.5999,
      "step": 2613
    },
    {
      "epoch": 0.8843031123139378,
      "grad_norm": 0.3269789516925812,
      "learning_rate": 4.0525137463791305e-06,
      "loss": 0.5801,
      "step": 2614
    },
    {
      "epoch": 0.8846414073071719,
      "grad_norm": 0.3483508229255676,
      "learning_rate": 4.051811548248315e-06,
      "loss": 0.6048,
      "step": 2615
    },
    {
      "epoch": 0.8849797023004059,
      "grad_norm": 0.32144540548324585,
      "learning_rate": 4.051109150895343e-06,
      "loss": 0.6069,
      "step": 2616
    },
    {
      "epoch": 0.88531799729364,
      "grad_norm": 0.32789748907089233,
      "learning_rate": 4.050406554410387e-06,
      "loss": 0.6015,
      "step": 2617
    },
    {
      "epoch": 0.8856562922868741,
      "grad_norm": 0.3455045819282532,
      "learning_rate": 4.049703758883648e-06,
      "loss": 0.5856,
      "step": 2618
    },
    {
      "epoch": 0.8859945872801083,
      "grad_norm": 0.3017064034938812,
      "learning_rate": 4.0490007644053505e-06,
      "loss": 0.6102,
      "step": 2619
    },
    {
      "epoch": 0.8863328822733424,
      "grad_norm": 0.3187443017959595,
      "learning_rate": 4.0482975710657455e-06,
      "loss": 0.619,
      "step": 2620
    },
    {
      "epoch": 0.8866711772665764,
      "grad_norm": 0.34462404251098633,
      "learning_rate": 4.04759417895511e-06,
      "loss": 0.5936,
      "step": 2621
    },
    {
      "epoch": 0.8870094722598105,
      "grad_norm": 0.31052395701408386,
      "learning_rate": 4.0468905881637445e-06,
      "loss": 0.6181,
      "step": 2622
    },
    {
      "epoch": 0.8873477672530447,
      "grad_norm": 0.31216633319854736,
      "learning_rate": 4.046186798781977e-06,
      "loss": 0.5945,
      "step": 2623
    },
    {
      "epoch": 0.8876860622462788,
      "grad_norm": 0.3139075040817261,
      "learning_rate": 4.045482810900159e-06,
      "loss": 0.6165,
      "step": 2624
    },
    {
      "epoch": 0.8880243572395129,
      "grad_norm": 0.32648202776908875,
      "learning_rate": 4.0447786246086704e-06,
      "loss": 0.6108,
      "step": 2625
    },
    {
      "epoch": 0.8883626522327469,
      "grad_norm": 0.3133030831813812,
      "learning_rate": 4.0440742399979134e-06,
      "loss": 0.6022,
      "step": 2626
    },
    {
      "epoch": 0.888700947225981,
      "grad_norm": 0.323303759098053,
      "learning_rate": 4.043369657158319e-06,
      "loss": 0.6191,
      "step": 2627
    },
    {
      "epoch": 0.8890392422192152,
      "grad_norm": 0.30271631479263306,
      "learning_rate": 4.042664876180341e-06,
      "loss": 0.6155,
      "step": 2628
    },
    {
      "epoch": 0.8893775372124493,
      "grad_norm": 0.3351958990097046,
      "learning_rate": 4.041959897154459e-06,
      "loss": 0.6146,
      "step": 2629
    },
    {
      "epoch": 0.8897158322056834,
      "grad_norm": 0.3067910373210907,
      "learning_rate": 4.041254720171178e-06,
      "loss": 0.6344,
      "step": 2630
    },
    {
      "epoch": 0.8900541271989174,
      "grad_norm": 0.3026675283908844,
      "learning_rate": 4.04054934532103e-06,
      "loss": 0.6075,
      "step": 2631
    },
    {
      "epoch": 0.8903924221921515,
      "grad_norm": 0.3328397572040558,
      "learning_rate": 4.039843772694572e-06,
      "loss": 0.5929,
      "step": 2632
    },
    {
      "epoch": 0.8907307171853857,
      "grad_norm": 0.30871671438217163,
      "learning_rate": 4.039138002382383e-06,
      "loss": 0.612,
      "step": 2633
    },
    {
      "epoch": 0.8910690121786198,
      "grad_norm": 0.3169931173324585,
      "learning_rate": 4.038432034475072e-06,
      "loss": 0.6241,
      "step": 2634
    },
    {
      "epoch": 0.8914073071718539,
      "grad_norm": 0.3067953586578369,
      "learning_rate": 4.037725869063272e-06,
      "loss": 0.5942,
      "step": 2635
    },
    {
      "epoch": 0.8917456021650879,
      "grad_norm": 0.30025818943977356,
      "learning_rate": 4.037019506237638e-06,
      "loss": 0.5773,
      "step": 2636
    },
    {
      "epoch": 0.892083897158322,
      "grad_norm": 0.312900573015213,
      "learning_rate": 4.036312946088857e-06,
      "loss": 0.6087,
      "step": 2637
    },
    {
      "epoch": 0.8924221921515562,
      "grad_norm": 0.2944320738315582,
      "learning_rate": 4.035606188707635e-06,
      "loss": 0.6095,
      "step": 2638
    },
    {
      "epoch": 0.8927604871447903,
      "grad_norm": 0.3260572552680969,
      "learning_rate": 4.0348992341847045e-06,
      "loss": 0.6476,
      "step": 2639
    },
    {
      "epoch": 0.8930987821380244,
      "grad_norm": 0.3164714276790619,
      "learning_rate": 4.034192082610829e-06,
      "loss": 0.6117,
      "step": 2640
    },
    {
      "epoch": 0.8934370771312584,
      "grad_norm": 0.30086609721183777,
      "learning_rate": 4.033484734076788e-06,
      "loss": 0.6281,
      "step": 2641
    },
    {
      "epoch": 0.8937753721244925,
      "grad_norm": 0.30885812640190125,
      "learning_rate": 4.032777188673395e-06,
      "loss": 0.6157,
      "step": 2642
    },
    {
      "epoch": 0.8941136671177267,
      "grad_norm": 0.31484881043434143,
      "learning_rate": 4.032069446491483e-06,
      "loss": 0.6009,
      "step": 2643
    },
    {
      "epoch": 0.8944519621109608,
      "grad_norm": 0.3178473711013794,
      "learning_rate": 4.0313615076219105e-06,
      "loss": 0.6391,
      "step": 2644
    },
    {
      "epoch": 0.8947902571041949,
      "grad_norm": 0.3143467605113983,
      "learning_rate": 4.030653372155567e-06,
      "loss": 0.6442,
      "step": 2645
    },
    {
      "epoch": 0.8951285520974289,
      "grad_norm": 0.31792086362838745,
      "learning_rate": 4.029945040183361e-06,
      "loss": 0.6117,
      "step": 2646
    },
    {
      "epoch": 0.895466847090663,
      "grad_norm": 0.30739086866378784,
      "learning_rate": 4.029236511796228e-06,
      "loss": 0.6323,
      "step": 2647
    },
    {
      "epoch": 0.8958051420838972,
      "grad_norm": 0.3313511312007904,
      "learning_rate": 4.02852778708513e-06,
      "loss": 0.6114,
      "step": 2648
    },
    {
      "epoch": 0.8961434370771313,
      "grad_norm": 0.3093686103820801,
      "learning_rate": 4.027818866141054e-06,
      "loss": 0.5859,
      "step": 2649
    },
    {
      "epoch": 0.8964817320703654,
      "grad_norm": 0.3045222759246826,
      "learning_rate": 4.02710974905501e-06,
      "loss": 0.6055,
      "step": 2650
    },
    {
      "epoch": 0.8968200270635994,
      "grad_norm": 0.3013499081134796,
      "learning_rate": 4.026400435918036e-06,
      "loss": 0.6149,
      "step": 2651
    },
    {
      "epoch": 0.8971583220568335,
      "grad_norm": 0.32199162244796753,
      "learning_rate": 4.025690926821192e-06,
      "loss": 0.6149,
      "step": 2652
    },
    {
      "epoch": 0.8974966170500677,
      "grad_norm": 0.3080962300300598,
      "learning_rate": 4.024981221855567e-06,
      "loss": 0.5905,
      "step": 2653
    },
    {
      "epoch": 0.8978349120433018,
      "grad_norm": 0.3041344881057739,
      "learning_rate": 4.024271321112273e-06,
      "loss": 0.5939,
      "step": 2654
    },
    {
      "epoch": 0.8981732070365359,
      "grad_norm": 0.3215409815311432,
      "learning_rate": 4.023561224682447e-06,
      "loss": 0.5873,
      "step": 2655
    },
    {
      "epoch": 0.8985115020297699,
      "grad_norm": 0.30556267499923706,
      "learning_rate": 4.02285093265725e-06,
      "loss": 0.5818,
      "step": 2656
    },
    {
      "epoch": 0.898849797023004,
      "grad_norm": 0.30866739153862,
      "learning_rate": 4.0221404451278716e-06,
      "loss": 0.6209,
      "step": 2657
    },
    {
      "epoch": 0.8991880920162382,
      "grad_norm": 0.3049587905406952,
      "learning_rate": 4.0214297621855226e-06,
      "loss": 0.5831,
      "step": 2658
    },
    {
      "epoch": 0.8995263870094723,
      "grad_norm": 0.3200123906135559,
      "learning_rate": 4.0207188839214426e-06,
      "loss": 0.5902,
      "step": 2659
    },
    {
      "epoch": 0.8998646820027063,
      "grad_norm": 0.3214031457901001,
      "learning_rate": 4.020007810426895e-06,
      "loss": 0.6054,
      "step": 2660
    },
    {
      "epoch": 0.9002029769959404,
      "grad_norm": 0.3044063150882721,
      "learning_rate": 4.019296541793165e-06,
      "loss": 0.599,
      "step": 2661
    },
    {
      "epoch": 0.9005412719891746,
      "grad_norm": 0.3177679777145386,
      "learning_rate": 4.018585078111567e-06,
      "loss": 0.6274,
      "step": 2662
    },
    {
      "epoch": 0.9008795669824087,
      "grad_norm": 0.31022119522094727,
      "learning_rate": 4.017873419473439e-06,
      "loss": 0.6046,
      "step": 2663
    },
    {
      "epoch": 0.9012178619756428,
      "grad_norm": 0.30733996629714966,
      "learning_rate": 4.017161565970144e-06,
      "loss": 0.6033,
      "step": 2664
    },
    {
      "epoch": 0.9015561569688768,
      "grad_norm": 0.2972017526626587,
      "learning_rate": 4.0164495176930705e-06,
      "loss": 0.5987,
      "step": 2665
    },
    {
      "epoch": 0.9018944519621109,
      "grad_norm": 0.30705738067626953,
      "learning_rate": 4.0157372747336295e-06,
      "loss": 0.6355,
      "step": 2666
    },
    {
      "epoch": 0.902232746955345,
      "grad_norm": 0.30549269914627075,
      "learning_rate": 4.015024837183261e-06,
      "loss": 0.5899,
      "step": 2667
    },
    {
      "epoch": 0.9025710419485792,
      "grad_norm": 0.31101226806640625,
      "learning_rate": 4.014312205133428e-06,
      "loss": 0.5998,
      "step": 2668
    },
    {
      "epoch": 0.9029093369418133,
      "grad_norm": 0.30988049507141113,
      "learning_rate": 4.013599378675618e-06,
      "loss": 0.6124,
      "step": 2669
    },
    {
      "epoch": 0.9032476319350473,
      "grad_norm": 0.30311518907546997,
      "learning_rate": 4.0128863579013435e-06,
      "loss": 0.5839,
      "step": 2670
    },
    {
      "epoch": 0.9035859269282814,
      "grad_norm": 0.32499635219573975,
      "learning_rate": 4.012173142902142e-06,
      "loss": 0.6102,
      "step": 2671
    },
    {
      "epoch": 0.9039242219215156,
      "grad_norm": 0.3209855854511261,
      "learning_rate": 4.0114597337695785e-06,
      "loss": 0.6189,
      "step": 2672
    },
    {
      "epoch": 0.9042625169147497,
      "grad_norm": 0.3145082890987396,
      "learning_rate": 4.010746130595239e-06,
      "loss": 0.5888,
      "step": 2673
    },
    {
      "epoch": 0.9046008119079838,
      "grad_norm": 0.30791082978248596,
      "learning_rate": 4.010032333470736e-06,
      "loss": 0.5968,
      "step": 2674
    },
    {
      "epoch": 0.9049391069012178,
      "grad_norm": 0.2905557453632355,
      "learning_rate": 4.009318342487708e-06,
      "loss": 0.5784,
      "step": 2675
    },
    {
      "epoch": 0.9052774018944519,
      "grad_norm": 0.298450231552124,
      "learning_rate": 4.008604157737816e-06,
      "loss": 0.6372,
      "step": 2676
    },
    {
      "epoch": 0.9056156968876861,
      "grad_norm": 0.3029123842716217,
      "learning_rate": 4.007889779312749e-06,
      "loss": 0.609,
      "step": 2677
    },
    {
      "epoch": 0.9059539918809202,
      "grad_norm": 0.317172110080719,
      "learning_rate": 4.007175207304218e-06,
      "loss": 0.627,
      "step": 2678
    },
    {
      "epoch": 0.9062922868741543,
      "grad_norm": 0.29541000723838806,
      "learning_rate": 4.006460441803961e-06,
      "loss": 0.5996,
      "step": 2679
    },
    {
      "epoch": 0.9066305818673883,
      "grad_norm": 0.3268113136291504,
      "learning_rate": 4.005745482903739e-06,
      "loss": 0.5988,
      "step": 2680
    },
    {
      "epoch": 0.9069688768606224,
      "grad_norm": 0.29055312275886536,
      "learning_rate": 4.005030330695339e-06,
      "loss": 0.6216,
      "step": 2681
    },
    {
      "epoch": 0.9073071718538566,
      "grad_norm": 0.3054754137992859,
      "learning_rate": 4.004314985270572e-06,
      "loss": 0.5825,
      "step": 2682
    },
    {
      "epoch": 0.9076454668470907,
      "grad_norm": 0.29997196793556213,
      "learning_rate": 4.003599446721276e-06,
      "loss": 0.6122,
      "step": 2683
    },
    {
      "epoch": 0.9079837618403248,
      "grad_norm": 0.31519967317581177,
      "learning_rate": 4.0028837151393095e-06,
      "loss": 0.609,
      "step": 2684
    },
    {
      "epoch": 0.9083220568335588,
      "grad_norm": 0.2943481504917145,
      "learning_rate": 4.002167790616561e-06,
      "loss": 0.5932,
      "step": 2685
    },
    {
      "epoch": 0.908660351826793,
      "grad_norm": 0.30532655119895935,
      "learning_rate": 4.001451673244939e-06,
      "loss": 0.5858,
      "step": 2686
    },
    {
      "epoch": 0.9089986468200271,
      "grad_norm": 0.3183039426803589,
      "learning_rate": 4.000735363116382e-06,
      "loss": 0.6036,
      "step": 2687
    },
    {
      "epoch": 0.9093369418132612,
      "grad_norm": 0.31465283036231995,
      "learning_rate": 4.000018860322845e-06,
      "loss": 0.6308,
      "step": 2688
    },
    {
      "epoch": 0.9096752368064953,
      "grad_norm": 0.3152543008327484,
      "learning_rate": 3.999302164956318e-06,
      "loss": 0.6124,
      "step": 2689
    },
    {
      "epoch": 0.9100135317997293,
      "grad_norm": 0.3163517415523529,
      "learning_rate": 3.998585277108808e-06,
      "loss": 0.5924,
      "step": 2690
    },
    {
      "epoch": 0.9103518267929634,
      "grad_norm": 0.3063450753688812,
      "learning_rate": 3.9978681968723505e-06,
      "loss": 0.6141,
      "step": 2691
    },
    {
      "epoch": 0.9106901217861976,
      "grad_norm": 0.3280673027038574,
      "learning_rate": 3.997150924339003e-06,
      "loss": 0.6099,
      "step": 2692
    },
    {
      "epoch": 0.9110284167794317,
      "grad_norm": 0.32003989815711975,
      "learning_rate": 3.996433459600851e-06,
      "loss": 0.6075,
      "step": 2693
    },
    {
      "epoch": 0.9113667117726658,
      "grad_norm": 0.3295360505580902,
      "learning_rate": 3.995715802750002e-06,
      "loss": 0.5993,
      "step": 2694
    },
    {
      "epoch": 0.9117050067658998,
      "grad_norm": 0.29726871848106384,
      "learning_rate": 3.994997953878589e-06,
      "loss": 0.6215,
      "step": 2695
    },
    {
      "epoch": 0.912043301759134,
      "grad_norm": 0.3211500346660614,
      "learning_rate": 3.99427991307877e-06,
      "loss": 0.5925,
      "step": 2696
    },
    {
      "epoch": 0.9123815967523681,
      "grad_norm": 0.30280858278274536,
      "learning_rate": 3.993561680442727e-06,
      "loss": 0.6047,
      "step": 2697
    },
    {
      "epoch": 0.9127198917456022,
      "grad_norm": 0.3184455931186676,
      "learning_rate": 3.992843256062667e-06,
      "loss": 0.6285,
      "step": 2698
    },
    {
      "epoch": 0.9130581867388363,
      "grad_norm": 0.3031151294708252,
      "learning_rate": 3.9921246400308226e-06,
      "loss": 0.5943,
      "step": 2699
    },
    {
      "epoch": 0.9133964817320703,
      "grad_norm": 0.3140672445297241,
      "learning_rate": 3.991405832439449e-06,
      "loss": 0.6112,
      "step": 2700
    },
    {
      "epoch": 0.9137347767253045,
      "grad_norm": 0.33230122923851013,
      "learning_rate": 3.990686833380827e-06,
      "loss": 0.5796,
      "step": 2701
    },
    {
      "epoch": 0.9140730717185386,
      "grad_norm": 0.30419784784317017,
      "learning_rate": 3.989967642947261e-06,
      "loss": 0.6077,
      "step": 2702
    },
    {
      "epoch": 0.9144113667117727,
      "grad_norm": 0.32081732153892517,
      "learning_rate": 3.989248261231084e-06,
      "loss": 0.5807,
      "step": 2703
    },
    {
      "epoch": 0.9147496617050067,
      "grad_norm": 0.35001420974731445,
      "learning_rate": 3.988528688324647e-06,
      "loss": 0.5917,
      "step": 2704
    },
    {
      "epoch": 0.9150879566982408,
      "grad_norm": 0.3250044584274292,
      "learning_rate": 3.9878089243203325e-06,
      "loss": 0.6045,
      "step": 2705
    },
    {
      "epoch": 0.915426251691475,
      "grad_norm": 0.32416415214538574,
      "learning_rate": 3.9870889693105405e-06,
      "loss": 0.6124,
      "step": 2706
    },
    {
      "epoch": 0.9157645466847091,
      "grad_norm": 0.3099220395088196,
      "learning_rate": 3.986368823387701e-06,
      "loss": 0.6043,
      "step": 2707
    },
    {
      "epoch": 0.9161028416779432,
      "grad_norm": 0.30382490158081055,
      "learning_rate": 3.985648486644267e-06,
      "loss": 0.621,
      "step": 2708
    },
    {
      "epoch": 0.9164411366711772,
      "grad_norm": 0.30252471566200256,
      "learning_rate": 3.984927959172715e-06,
      "loss": 0.635,
      "step": 2709
    },
    {
      "epoch": 0.9167794316644113,
      "grad_norm": 0.30487069487571716,
      "learning_rate": 3.984207241065546e-06,
      "loss": 0.604,
      "step": 2710
    },
    {
      "epoch": 0.9171177266576455,
      "grad_norm": 0.3196687400341034,
      "learning_rate": 3.983486332415288e-06,
      "loss": 0.6021,
      "step": 2711
    },
    {
      "epoch": 0.9174560216508796,
      "grad_norm": 0.30728936195373535,
      "learning_rate": 3.982765233314489e-06,
      "loss": 0.6035,
      "step": 2712
    },
    {
      "epoch": 0.9177943166441137,
      "grad_norm": 0.30180492997169495,
      "learning_rate": 3.982043943855726e-06,
      "loss": 0.613,
      "step": 2713
    },
    {
      "epoch": 0.9181326116373477,
      "grad_norm": 0.32103562355041504,
      "learning_rate": 3.981322464131597e-06,
      "loss": 0.6288,
      "step": 2714
    },
    {
      "epoch": 0.9184709066305818,
      "grad_norm": 0.3367891013622284,
      "learning_rate": 3.980600794234726e-06,
      "loss": 0.6103,
      "step": 2715
    },
    {
      "epoch": 0.918809201623816,
      "grad_norm": 0.3209404945373535,
      "learning_rate": 3.979878934257763e-06,
      "loss": 0.6361,
      "step": 2716
    },
    {
      "epoch": 0.9191474966170501,
      "grad_norm": 0.3047719895839691,
      "learning_rate": 3.979156884293378e-06,
      "loss": 0.6108,
      "step": 2717
    },
    {
      "epoch": 0.9194857916102842,
      "grad_norm": 0.3115750551223755,
      "learning_rate": 3.97843464443427e-06,
      "loss": 0.617,
      "step": 2718
    },
    {
      "epoch": 0.9198240866035182,
      "grad_norm": 0.3256355822086334,
      "learning_rate": 3.97771221477316e-06,
      "loss": 0.6153,
      "step": 2719
    },
    {
      "epoch": 0.9201623815967523,
      "grad_norm": 0.3168798089027405,
      "learning_rate": 3.976989595402794e-06,
      "loss": 0.6352,
      "step": 2720
    },
    {
      "epoch": 0.9205006765899865,
      "grad_norm": 0.31463319063186646,
      "learning_rate": 3.976266786415941e-06,
      "loss": 0.6181,
      "step": 2721
    },
    {
      "epoch": 0.9208389715832206,
      "grad_norm": 0.3075326383113861,
      "learning_rate": 3.975543787905397e-06,
      "loss": 0.6109,
      "step": 2722
    },
    {
      "epoch": 0.9211772665764547,
      "grad_norm": 0.3231752812862396,
      "learning_rate": 3.97482059996398e-06,
      "loss": 0.6216,
      "step": 2723
    },
    {
      "epoch": 0.9215155615696887,
      "grad_norm": 0.3100549578666687,
      "learning_rate": 3.974097222684532e-06,
      "loss": 0.6194,
      "step": 2724
    },
    {
      "epoch": 0.9218538565629228,
      "grad_norm": 0.3243764042854309,
      "learning_rate": 3.973373656159923e-06,
      "loss": 0.6092,
      "step": 2725
    },
    {
      "epoch": 0.922192151556157,
      "grad_norm": 0.40476149320602417,
      "learning_rate": 3.9726499004830435e-06,
      "loss": 0.6016,
      "step": 2726
    },
    {
      "epoch": 0.9225304465493911,
      "grad_norm": 0.3071513772010803,
      "learning_rate": 3.971925955746808e-06,
      "loss": 0.6241,
      "step": 2727
    },
    {
      "epoch": 0.9228687415426252,
      "grad_norm": 0.3396247327327728,
      "learning_rate": 3.9712018220441595e-06,
      "loss": 0.6003,
      "step": 2728
    },
    {
      "epoch": 0.9232070365358592,
      "grad_norm": 0.32305842638015747,
      "learning_rate": 3.97047749946806e-06,
      "loss": 0.5911,
      "step": 2729
    },
    {
      "epoch": 0.9235453315290933,
      "grad_norm": 0.3153322637081146,
      "learning_rate": 3.9697529881115015e-06,
      "loss": 0.6289,
      "step": 2730
    },
    {
      "epoch": 0.9238836265223275,
      "grad_norm": 0.29865762591362,
      "learning_rate": 3.969028288067495e-06,
      "loss": 0.6164,
      "step": 2731
    },
    {
      "epoch": 0.9242219215155616,
      "grad_norm": 0.31667447090148926,
      "learning_rate": 3.968303399429077e-06,
      "loss": 0.6183,
      "step": 2732
    },
    {
      "epoch": 0.9245602165087957,
      "grad_norm": 0.37699103355407715,
      "learning_rate": 3.967578322289311e-06,
      "loss": 0.6141,
      "step": 2733
    },
    {
      "epoch": 0.9248985115020297,
      "grad_norm": 0.30763012170791626,
      "learning_rate": 3.96685305674128e-06,
      "loss": 0.6054,
      "step": 2734
    },
    {
      "epoch": 0.9252368064952639,
      "grad_norm": 0.31958621740341187,
      "learning_rate": 3.966127602878096e-06,
      "loss": 0.6359,
      "step": 2735
    },
    {
      "epoch": 0.925575101488498,
      "grad_norm": 0.33669278025627136,
      "learning_rate": 3.965401960792894e-06,
      "loss": 0.6073,
      "step": 2736
    },
    {
      "epoch": 0.9259133964817321,
      "grad_norm": 0.31136423349380493,
      "learning_rate": 3.964676130578829e-06,
      "loss": 0.6406,
      "step": 2737
    },
    {
      "epoch": 0.9262516914749662,
      "grad_norm": 0.31435713171958923,
      "learning_rate": 3.963950112329086e-06,
      "loss": 0.6045,
      "step": 2738
    },
    {
      "epoch": 0.9265899864682002,
      "grad_norm": 0.3053027093410492,
      "learning_rate": 3.963223906136869e-06,
      "loss": 0.6029,
      "step": 2739
    },
    {
      "epoch": 0.9269282814614344,
      "grad_norm": 0.36422377824783325,
      "learning_rate": 3.962497512095412e-06,
      "loss": 0.601,
      "step": 2740
    },
    {
      "epoch": 0.9272665764546685,
      "grad_norm": 0.3055732250213623,
      "learning_rate": 3.961770930297968e-06,
      "loss": 0.6066,
      "step": 2741
    },
    {
      "epoch": 0.9276048714479026,
      "grad_norm": 0.3042244017124176,
      "learning_rate": 3.961044160837815e-06,
      "loss": 0.5942,
      "step": 2742
    },
    {
      "epoch": 0.9279431664411367,
      "grad_norm": 0.3053058087825775,
      "learning_rate": 3.960317203808257e-06,
      "loss": 0.5995,
      "step": 2743
    },
    {
      "epoch": 0.9282814614343707,
      "grad_norm": 0.2914317846298218,
      "learning_rate": 3.95959005930262e-06,
      "loss": 0.6176,
      "step": 2744
    },
    {
      "epoch": 0.9286197564276049,
      "grad_norm": 0.31416183710098267,
      "learning_rate": 3.958862727414256e-06,
      "loss": 0.6035,
      "step": 2745
    },
    {
      "epoch": 0.928958051420839,
      "grad_norm": 0.2997271418571472,
      "learning_rate": 3.958135208236541e-06,
      "loss": 0.6159,
      "step": 2746
    },
    {
      "epoch": 0.9292963464140731,
      "grad_norm": 0.3054497241973877,
      "learning_rate": 3.957407501862873e-06,
      "loss": 0.632,
      "step": 2747
    },
    {
      "epoch": 0.9296346414073072,
      "grad_norm": 0.2939889132976532,
      "learning_rate": 3.956679608386675e-06,
      "loss": 0.6327,
      "step": 2748
    },
    {
      "epoch": 0.9299729364005412,
      "grad_norm": 0.3000098764896393,
      "learning_rate": 3.955951527901395e-06,
      "loss": 0.6365,
      "step": 2749
    },
    {
      "epoch": 0.9303112313937754,
      "grad_norm": 0.3275192379951477,
      "learning_rate": 3.955223260500503e-06,
      "loss": 0.5955,
      "step": 2750
    },
    {
      "epoch": 0.9306495263870095,
      "grad_norm": 0.32291221618652344,
      "learning_rate": 3.954494806277496e-06,
      "loss": 0.6173,
      "step": 2751
    },
    {
      "epoch": 0.9309878213802436,
      "grad_norm": 0.30584001541137695,
      "learning_rate": 3.953766165325892e-06,
      "loss": 0.6249,
      "step": 2752
    },
    {
      "epoch": 0.9313261163734776,
      "grad_norm": 0.3476294279098511,
      "learning_rate": 3.953037337739235e-06,
      "loss": 0.5937,
      "step": 2753
    },
    {
      "epoch": 0.9316644113667117,
      "grad_norm": 0.30108344554901123,
      "learning_rate": 3.952308323611091e-06,
      "loss": 0.6359,
      "step": 2754
    },
    {
      "epoch": 0.9320027063599459,
      "grad_norm": 0.3059139549732208,
      "learning_rate": 3.951579123035053e-06,
      "loss": 0.6388,
      "step": 2755
    },
    {
      "epoch": 0.93234100135318,
      "grad_norm": 0.309368371963501,
      "learning_rate": 3.9508497361047334e-06,
      "loss": 0.6241,
      "step": 2756
    },
    {
      "epoch": 0.9326792963464141,
      "grad_norm": 0.3595433533191681,
      "learning_rate": 3.950120162913773e-06,
      "loss": 0.5929,
      "step": 2757
    },
    {
      "epoch": 0.9330175913396481,
      "grad_norm": 0.32127654552459717,
      "learning_rate": 3.9493904035558356e-06,
      "loss": 0.6169,
      "step": 2758
    },
    {
      "epoch": 0.9333558863328822,
      "grad_norm": 0.2973577678203583,
      "learning_rate": 3.948660458124606e-06,
      "loss": 0.6124,
      "step": 2759
    },
    {
      "epoch": 0.9336941813261164,
      "grad_norm": 0.30579710006713867,
      "learning_rate": 3.947930326713795e-06,
      "loss": 0.5929,
      "step": 2760
    },
    {
      "epoch": 0.9340324763193505,
      "grad_norm": 0.32044917345046997,
      "learning_rate": 3.947200009417138e-06,
      "loss": 0.59,
      "step": 2761
    },
    {
      "epoch": 0.9343707713125846,
      "grad_norm": 0.30661648511886597,
      "learning_rate": 3.946469506328393e-06,
      "loss": 0.6504,
      "step": 2762
    },
    {
      "epoch": 0.9347090663058186,
      "grad_norm": 0.3152296543121338,
      "learning_rate": 3.945738817541343e-06,
      "loss": 0.6298,
      "step": 2763
    },
    {
      "epoch": 0.9350473612990527,
      "grad_norm": 0.30985817313194275,
      "learning_rate": 3.945007943149794e-06,
      "loss": 0.6477,
      "step": 2764
    },
    {
      "epoch": 0.9353856562922869,
      "grad_norm": 0.3215924799442291,
      "learning_rate": 3.944276883247575e-06,
      "loss": 0.6092,
      "step": 2765
    },
    {
      "epoch": 0.935723951285521,
      "grad_norm": 0.3165227174758911,
      "learning_rate": 3.943545637928539e-06,
      "loss": 0.6021,
      "step": 2766
    },
    {
      "epoch": 0.9360622462787551,
      "grad_norm": 0.323017954826355,
      "learning_rate": 3.942814207286566e-06,
      "loss": 0.6183,
      "step": 2767
    },
    {
      "epoch": 0.9364005412719891,
      "grad_norm": 0.2983512282371521,
      "learning_rate": 3.942082591415556e-06,
      "loss": 0.5848,
      "step": 2768
    },
    {
      "epoch": 0.9367388362652233,
      "grad_norm": 0.32698339223861694,
      "learning_rate": 3.941350790409434e-06,
      "loss": 0.5975,
      "step": 2769
    },
    {
      "epoch": 0.9370771312584574,
      "grad_norm": 0.3031372129917145,
      "learning_rate": 3.94061880436215e-06,
      "loss": 0.583,
      "step": 2770
    },
    {
      "epoch": 0.9374154262516915,
      "grad_norm": 0.2965066432952881,
      "learning_rate": 3.939886633367674e-06,
      "loss": 0.6025,
      "step": 2771
    },
    {
      "epoch": 0.9377537212449256,
      "grad_norm": 0.3047961890697479,
      "learning_rate": 3.939154277520006e-06,
      "loss": 0.6077,
      "step": 2772
    },
    {
      "epoch": 0.9380920162381596,
      "grad_norm": 0.3104235529899597,
      "learning_rate": 3.938421736913163e-06,
      "loss": 0.5801,
      "step": 2773
    },
    {
      "epoch": 0.9384303112313938,
      "grad_norm": 0.30181562900543213,
      "learning_rate": 3.93768901164119e-06,
      "loss": 0.6032,
      "step": 2774
    },
    {
      "epoch": 0.9387686062246279,
      "grad_norm": 0.30882957577705383,
      "learning_rate": 3.936956101798156e-06,
      "loss": 0.6071,
      "step": 2775
    },
    {
      "epoch": 0.939106901217862,
      "grad_norm": 0.3213382661342621,
      "learning_rate": 3.936223007478151e-06,
      "loss": 0.6529,
      "step": 2776
    },
    {
      "epoch": 0.9394451962110961,
      "grad_norm": 0.30974048376083374,
      "learning_rate": 3.935489728775288e-06,
      "loss": 0.6231,
      "step": 2777
    },
    {
      "epoch": 0.9397834912043301,
      "grad_norm": 0.31085005402565,
      "learning_rate": 3.934756265783709e-06,
      "loss": 0.5694,
      "step": 2778
    },
    {
      "epoch": 0.9401217861975643,
      "grad_norm": 0.29983943700790405,
      "learning_rate": 3.934022618597574e-06,
      "loss": 0.6022,
      "step": 2779
    },
    {
      "epoch": 0.9404600811907984,
      "grad_norm": 0.3146032691001892,
      "learning_rate": 3.93328878731107e-06,
      "loss": 0.6145,
      "step": 2780
    },
    {
      "epoch": 0.9407983761840325,
      "grad_norm": 0.2978469729423523,
      "learning_rate": 3.932554772018405e-06,
      "loss": 0.6127,
      "step": 2781
    },
    {
      "epoch": 0.9411366711772666,
      "grad_norm": 0.2859659194946289,
      "learning_rate": 3.931820572813815e-06,
      "loss": 0.6107,
      "step": 2782
    },
    {
      "epoch": 0.9414749661705006,
      "grad_norm": 0.3153398931026459,
      "learning_rate": 3.931086189791554e-06,
      "loss": 0.6268,
      "step": 2783
    },
    {
      "epoch": 0.9418132611637348,
      "grad_norm": 0.2963147759437561,
      "learning_rate": 3.930351623045904e-06,
      "loss": 0.6248,
      "step": 2784
    },
    {
      "epoch": 0.9421515561569689,
      "grad_norm": 0.3015451431274414,
      "learning_rate": 3.929616872671168e-06,
      "loss": 0.6161,
      "step": 2785
    },
    {
      "epoch": 0.942489851150203,
      "grad_norm": 0.2986791133880615,
      "learning_rate": 3.928881938761674e-06,
      "loss": 0.6027,
      "step": 2786
    },
    {
      "epoch": 0.9428281461434371,
      "grad_norm": 0.3088212013244629,
      "learning_rate": 3.928146821411773e-06,
      "loss": 0.6087,
      "step": 2787
    },
    {
      "epoch": 0.9431664411366711,
      "grad_norm": 0.3106054365634918,
      "learning_rate": 3.92741152071584e-06,
      "loss": 0.6186,
      "step": 2788
    },
    {
      "epoch": 0.9435047361299053,
      "grad_norm": 0.314456969499588,
      "learning_rate": 3.926676036768273e-06,
      "loss": 0.6141,
      "step": 2789
    },
    {
      "epoch": 0.9438430311231394,
      "grad_norm": 0.3026145398616791,
      "learning_rate": 3.925940369663494e-06,
      "loss": 0.6058,
      "step": 2790
    },
    {
      "epoch": 0.9441813261163735,
      "grad_norm": 0.3158716559410095,
      "learning_rate": 3.925204519495946e-06,
      "loss": 0.6106,
      "step": 2791
    },
    {
      "epoch": 0.9445196211096076,
      "grad_norm": 0.31110692024230957,
      "learning_rate": 3.924468486360101e-06,
      "loss": 0.6032,
      "step": 2792
    },
    {
      "epoch": 0.9448579161028416,
      "grad_norm": 0.3241216242313385,
      "learning_rate": 3.9237322703504495e-06,
      "loss": 0.593,
      "step": 2793
    },
    {
      "epoch": 0.9451962110960758,
      "grad_norm": 0.31992748379707336,
      "learning_rate": 3.922995871561508e-06,
      "loss": 0.6314,
      "step": 2794
    },
    {
      "epoch": 0.9455345060893099,
      "grad_norm": 0.30860215425491333,
      "learning_rate": 3.922259290087814e-06,
      "loss": 0.6047,
      "step": 2795
    },
    {
      "epoch": 0.945872801082544,
      "grad_norm": 0.3147035241127014,
      "learning_rate": 3.9215225260239315e-06,
      "loss": 0.6355,
      "step": 2796
    },
    {
      "epoch": 0.946211096075778,
      "grad_norm": 0.3141655921936035,
      "learning_rate": 3.920785579464447e-06,
      "loss": 0.6038,
      "step": 2797
    },
    {
      "epoch": 0.9465493910690121,
      "grad_norm": 0.3091292381286621,
      "learning_rate": 3.9200484505039685e-06,
      "loss": 0.645,
      "step": 2798
    },
    {
      "epoch": 0.9468876860622463,
      "grad_norm": 0.29969021677970886,
      "learning_rate": 3.919311139237129e-06,
      "loss": 0.6036,
      "step": 2799
    },
    {
      "epoch": 0.9472259810554804,
      "grad_norm": 0.3013696074485779,
      "learning_rate": 3.918573645758586e-06,
      "loss": 0.6194,
      "step": 2800
    },
    {
      "epoch": 0.9475642760487145,
      "grad_norm": 0.30952438712120056,
      "learning_rate": 3.917835970163019e-06,
      "loss": 0.6001,
      "step": 2801
    },
    {
      "epoch": 0.9479025710419485,
      "grad_norm": 0.313272088766098,
      "learning_rate": 3.917098112545131e-06,
      "loss": 0.5993,
      "step": 2802
    },
    {
      "epoch": 0.9482408660351827,
      "grad_norm": 0.31668946146965027,
      "learning_rate": 3.9163600729996474e-06,
      "loss": 0.6078,
      "step": 2803
    },
    {
      "epoch": 0.9485791610284168,
      "grad_norm": 0.30897411704063416,
      "learning_rate": 3.915621851621318e-06,
      "loss": 0.5887,
      "step": 2804
    },
    {
      "epoch": 0.9489174560216509,
      "grad_norm": 0.3138033449649811,
      "learning_rate": 3.914883448504918e-06,
      "loss": 0.615,
      "step": 2805
    },
    {
      "epoch": 0.949255751014885,
      "grad_norm": 0.3061548173427582,
      "learning_rate": 3.914144863745243e-06,
      "loss": 0.6038,
      "step": 2806
    },
    {
      "epoch": 0.949594046008119,
      "grad_norm": 0.310561865568161,
      "learning_rate": 3.913406097437111e-06,
      "loss": 0.6094,
      "step": 2807
    },
    {
      "epoch": 0.9499323410013532,
      "grad_norm": 0.3132334351539612,
      "learning_rate": 3.9126671496753665e-06,
      "loss": 0.6124,
      "step": 2808
    },
    {
      "epoch": 0.9502706359945873,
      "grad_norm": 0.32545071840286255,
      "learning_rate": 3.911928020554877e-06,
      "loss": 0.6089,
      "step": 2809
    },
    {
      "epoch": 0.9506089309878214,
      "grad_norm": 0.30556997656822205,
      "learning_rate": 3.91118871017053e-06,
      "loss": 0.5793,
      "step": 2810
    },
    {
      "epoch": 0.9509472259810555,
      "grad_norm": 0.32106563448905945,
      "learning_rate": 3.910449218617239e-06,
      "loss": 0.598,
      "step": 2811
    },
    {
      "epoch": 0.9512855209742895,
      "grad_norm": 0.311365008354187,
      "learning_rate": 3.909709545989942e-06,
      "loss": 0.6421,
      "step": 2812
    },
    {
      "epoch": 0.9516238159675237,
      "grad_norm": 0.3102574348449707,
      "learning_rate": 3.908969692383597e-06,
      "loss": 0.6311,
      "step": 2813
    },
    {
      "epoch": 0.9519621109607578,
      "grad_norm": 0.3076390326023102,
      "learning_rate": 3.908229657893186e-06,
      "loss": 0.6083,
      "step": 2814
    },
    {
      "epoch": 0.9523004059539919,
      "grad_norm": 0.3074639141559601,
      "learning_rate": 3.907489442613716e-06,
      "loss": 0.6183,
      "step": 2815
    },
    {
      "epoch": 0.952638700947226,
      "grad_norm": 0.31547728180885315,
      "learning_rate": 3.906749046640216e-06,
      "loss": 0.6267,
      "step": 2816
    },
    {
      "epoch": 0.95297699594046,
      "grad_norm": 0.3008100688457489,
      "learning_rate": 3.906008470067738e-06,
      "loss": 0.6207,
      "step": 2817
    },
    {
      "epoch": 0.9533152909336942,
      "grad_norm": 0.307170569896698,
      "learning_rate": 3.9052677129913575e-06,
      "loss": 0.601,
      "step": 2818
    },
    {
      "epoch": 0.9536535859269283,
      "grad_norm": 0.2974233627319336,
      "learning_rate": 3.904526775506174e-06,
      "loss": 0.6406,
      "step": 2819
    },
    {
      "epoch": 0.9539918809201624,
      "grad_norm": 0.3185262978076935,
      "learning_rate": 3.903785657707308e-06,
      "loss": 0.6407,
      "step": 2820
    },
    {
      "epoch": 0.9543301759133965,
      "grad_norm": 0.3027080297470093,
      "learning_rate": 3.903044359689905e-06,
      "loss": 0.6286,
      "step": 2821
    },
    {
      "epoch": 0.9546684709066305,
      "grad_norm": 0.3234482705593109,
      "learning_rate": 3.902302881549134e-06,
      "loss": 0.6347,
      "step": 2822
    },
    {
      "epoch": 0.9550067658998647,
      "grad_norm": 0.3211921453475952,
      "learning_rate": 3.9015612233801845e-06,
      "loss": 0.6094,
      "step": 2823
    },
    {
      "epoch": 0.9553450608930988,
      "grad_norm": 0.2871362864971161,
      "learning_rate": 3.9008193852782735e-06,
      "loss": 0.6051,
      "step": 2824
    },
    {
      "epoch": 0.9556833558863329,
      "grad_norm": 0.3141300082206726,
      "learning_rate": 3.900077367338635e-06,
      "loss": 0.6121,
      "step": 2825
    },
    {
      "epoch": 0.956021650879567,
      "grad_norm": 0.31430670619010925,
      "learning_rate": 3.899335169656533e-06,
      "loss": 0.6163,
      "step": 2826
    },
    {
      "epoch": 0.956359945872801,
      "grad_norm": 0.29906728863716125,
      "learning_rate": 3.898592792327248e-06,
      "loss": 0.6102,
      "step": 2827
    },
    {
      "epoch": 0.9566982408660352,
      "grad_norm": 0.30080512166023254,
      "learning_rate": 3.897850235446089e-06,
      "loss": 0.5982,
      "step": 2828
    },
    {
      "epoch": 0.9570365358592693,
      "grad_norm": 0.29242387413978577,
      "learning_rate": 3.897107499108385e-06,
      "loss": 0.6152,
      "step": 2829
    },
    {
      "epoch": 0.9573748308525034,
      "grad_norm": 0.2824954390525818,
      "learning_rate": 3.896364583409488e-06,
      "loss": 0.6023,
      "step": 2830
    },
    {
      "epoch": 0.9577131258457375,
      "grad_norm": 0.3062789738178253,
      "learning_rate": 3.895621488444777e-06,
      "loss": 0.6267,
      "step": 2831
    },
    {
      "epoch": 0.9580514208389715,
      "grad_norm": 0.3123863637447357,
      "learning_rate": 3.894878214309645e-06,
      "loss": 0.6391,
      "step": 2832
    },
    {
      "epoch": 0.9583897158322057,
      "grad_norm": 0.29575687646865845,
      "learning_rate": 3.894134761099519e-06,
      "loss": 0.6311,
      "step": 2833
    },
    {
      "epoch": 0.9587280108254398,
      "grad_norm": 0.2981586158275604,
      "learning_rate": 3.893391128909841e-06,
      "loss": 0.6131,
      "step": 2834
    },
    {
      "epoch": 0.9590663058186739,
      "grad_norm": 0.3023635149002075,
      "learning_rate": 3.89264731783608e-06,
      "loss": 0.5943,
      "step": 2835
    },
    {
      "epoch": 0.959404600811908,
      "grad_norm": 0.3178964853286743,
      "learning_rate": 3.891903327973727e-06,
      "loss": 0.6031,
      "step": 2836
    },
    {
      "epoch": 0.959742895805142,
      "grad_norm": 0.30327779054641724,
      "learning_rate": 3.891159159418294e-06,
      "loss": 0.613,
      "step": 2837
    },
    {
      "epoch": 0.9600811907983762,
      "grad_norm": 0.30893656611442566,
      "learning_rate": 3.89041481226532e-06,
      "loss": 0.6023,
      "step": 2838
    },
    {
      "epoch": 0.9604194857916103,
      "grad_norm": 0.2962033450603485,
      "learning_rate": 3.889670286610363e-06,
      "loss": 0.649,
      "step": 2839
    },
    {
      "epoch": 0.9607577807848444,
      "grad_norm": 0.330939382314682,
      "learning_rate": 3.888925582549006e-06,
      "loss": 0.595,
      "step": 2840
    },
    {
      "epoch": 0.9610960757780784,
      "grad_norm": 0.29611915349960327,
      "learning_rate": 3.888180700176854e-06,
      "loss": 0.5919,
      "step": 2841
    },
    {
      "epoch": 0.9614343707713126,
      "grad_norm": 0.29588085412979126,
      "learning_rate": 3.887435639589535e-06,
      "loss": 0.6002,
      "step": 2842
    },
    {
      "epoch": 0.9617726657645467,
      "grad_norm": 0.3030603528022766,
      "learning_rate": 3.886690400882702e-06,
      "loss": 0.5704,
      "step": 2843
    },
    {
      "epoch": 0.9621109607577808,
      "grad_norm": 0.3341445028781891,
      "learning_rate": 3.885944984152027e-06,
      "loss": 0.6159,
      "step": 2844
    },
    {
      "epoch": 0.9624492557510149,
      "grad_norm": 0.307426393032074,
      "learning_rate": 3.885199389493207e-06,
      "loss": 0.6229,
      "step": 2845
    },
    {
      "epoch": 0.9627875507442489,
      "grad_norm": 0.3231599032878876,
      "learning_rate": 3.884453617001964e-06,
      "loss": 0.612,
      "step": 2846
    },
    {
      "epoch": 0.963125845737483,
      "grad_norm": 0.3122326731681824,
      "learning_rate": 3.883707666774037e-06,
      "loss": 0.6397,
      "step": 2847
    },
    {
      "epoch": 0.9634641407307172,
      "grad_norm": 0.29532092809677124,
      "learning_rate": 3.882961538905194e-06,
      "loss": 0.5833,
      "step": 2848
    },
    {
      "epoch": 0.9638024357239513,
      "grad_norm": 0.31446200609207153,
      "learning_rate": 3.882215233491224e-06,
      "loss": 0.6094,
      "step": 2849
    },
    {
      "epoch": 0.9641407307171854,
      "grad_norm": 0.3175473213195801,
      "learning_rate": 3.881468750627934e-06,
      "loss": 0.6163,
      "step": 2850
    },
    {
      "epoch": 0.9644790257104194,
      "grad_norm": 0.3167509138584137,
      "learning_rate": 3.880722090411161e-06,
      "loss": 0.6268,
      "step": 2851
    },
    {
      "epoch": 0.9648173207036536,
      "grad_norm": 0.3074794411659241,
      "learning_rate": 3.879975252936761e-06,
      "loss": 0.5633,
      "step": 2852
    },
    {
      "epoch": 0.9651556156968877,
      "grad_norm": 0.3008774220943451,
      "learning_rate": 3.879228238300613e-06,
      "loss": 0.6086,
      "step": 2853
    },
    {
      "epoch": 0.9654939106901218,
      "grad_norm": 0.30035844445228577,
      "learning_rate": 3.878481046598619e-06,
      "loss": 0.5901,
      "step": 2854
    },
    {
      "epoch": 0.9658322056833559,
      "grad_norm": 0.31396716833114624,
      "learning_rate": 3.877733677926704e-06,
      "loss": 0.6236,
      "step": 2855
    },
    {
      "epoch": 0.9661705006765899,
      "grad_norm": 0.31159961223602295,
      "learning_rate": 3.8769861323808146e-06,
      "loss": 0.5898,
      "step": 2856
    },
    {
      "epoch": 0.9665087956698241,
      "grad_norm": 0.30015599727630615,
      "learning_rate": 3.876238410056922e-06,
      "loss": 0.6235,
      "step": 2857
    },
    {
      "epoch": 0.9668470906630582,
      "grad_norm": 0.30023249983787537,
      "learning_rate": 3.875490511051018e-06,
      "loss": 0.5967,
      "step": 2858
    },
    {
      "epoch": 0.9671853856562923,
      "grad_norm": 0.3081527352333069,
      "learning_rate": 3.874742435459119e-06,
      "loss": 0.6186,
      "step": 2859
    },
    {
      "epoch": 0.9675236806495264,
      "grad_norm": 0.29001158475875854,
      "learning_rate": 3.8739941833772645e-06,
      "loss": 0.6069,
      "step": 2860
    },
    {
      "epoch": 0.9678619756427604,
      "grad_norm": 0.31207314133644104,
      "learning_rate": 3.873245754901513e-06,
      "loss": 0.6173,
      "step": 2861
    },
    {
      "epoch": 0.9682002706359946,
      "grad_norm": 0.3128235936164856,
      "learning_rate": 3.872497150127948e-06,
      "loss": 0.5948,
      "step": 2862
    },
    {
      "epoch": 0.9685385656292287,
      "grad_norm": 0.3001565933227539,
      "learning_rate": 3.871748369152678e-06,
      "loss": 0.6256,
      "step": 2863
    },
    {
      "epoch": 0.9688768606224628,
      "grad_norm": 0.3021819591522217,
      "learning_rate": 3.87099941207183e-06,
      "loss": 0.6186,
      "step": 2864
    },
    {
      "epoch": 0.9692151556156969,
      "grad_norm": 0.31208890676498413,
      "learning_rate": 3.870250278981556e-06,
      "loss": 0.5921,
      "step": 2865
    },
    {
      "epoch": 0.969553450608931,
      "grad_norm": 0.3111448585987091,
      "learning_rate": 3.8695009699780295e-06,
      "loss": 0.588,
      "step": 2866
    },
    {
      "epoch": 0.9698917456021651,
      "grad_norm": 0.29972732067108154,
      "learning_rate": 3.868751485157448e-06,
      "loss": 0.6179,
      "step": 2867
    },
    {
      "epoch": 0.9702300405953992,
      "grad_norm": 0.3013609051704407,
      "learning_rate": 3.868001824616029e-06,
      "loss": 0.5683,
      "step": 2868
    },
    {
      "epoch": 0.9705683355886333,
      "grad_norm": 0.32852277159690857,
      "learning_rate": 3.867251988450017e-06,
      "loss": 0.6204,
      "step": 2869
    },
    {
      "epoch": 0.9709066305818674,
      "grad_norm": 0.3015483021736145,
      "learning_rate": 3.866501976755673e-06,
      "loss": 0.608,
      "step": 2870
    },
    {
      "epoch": 0.9712449255751014,
      "grad_norm": 0.3175235390663147,
      "learning_rate": 3.865751789629285e-06,
      "loss": 0.6,
      "step": 2871
    },
    {
      "epoch": 0.9715832205683356,
      "grad_norm": 0.30726557970046997,
      "learning_rate": 3.865001427167164e-06,
      "loss": 0.5984,
      "step": 2872
    },
    {
      "epoch": 0.9719215155615697,
      "grad_norm": 0.2998640537261963,
      "learning_rate": 3.86425088946564e-06,
      "loss": 0.5782,
      "step": 2873
    },
    {
      "epoch": 0.9722598105548038,
      "grad_norm": 0.31798794865608215,
      "learning_rate": 3.863500176621068e-06,
      "loss": 0.5884,
      "step": 2874
    },
    {
      "epoch": 0.9725981055480379,
      "grad_norm": 0.3241869807243347,
      "learning_rate": 3.8627492887298234e-06,
      "loss": 0.5987,
      "step": 2875
    },
    {
      "epoch": 0.972936400541272,
      "grad_norm": 0.309407502412796,
      "learning_rate": 3.8619982258883074e-06,
      "loss": 0.6201,
      "step": 2876
    },
    {
      "epoch": 0.9732746955345061,
      "grad_norm": 0.29837316274642944,
      "learning_rate": 3.861246988192941e-06,
      "loss": 0.6208,
      "step": 2877
    },
    {
      "epoch": 0.9736129905277402,
      "grad_norm": 0.3008418083190918,
      "learning_rate": 3.860495575740167e-06,
      "loss": 0.6403,
      "step": 2878
    },
    {
      "epoch": 0.9739512855209743,
      "grad_norm": 0.3007146418094635,
      "learning_rate": 3.859743988626456e-06,
      "loss": 0.6002,
      "step": 2879
    },
    {
      "epoch": 0.9742895805142084,
      "grad_norm": 0.30299657583236694,
      "learning_rate": 3.8589922269482925e-06,
      "loss": 0.5899,
      "step": 2880
    },
    {
      "epoch": 0.9746278755074425,
      "grad_norm": 0.3261502683162689,
      "learning_rate": 3.85824029080219e-06,
      "loss": 0.617,
      "step": 2881
    },
    {
      "epoch": 0.9749661705006766,
      "grad_norm": 0.3027297854423523,
      "learning_rate": 3.857488180284682e-06,
      "loss": 0.6109,
      "step": 2882
    },
    {
      "epoch": 0.9753044654939107,
      "grad_norm": 0.3169786036014557,
      "learning_rate": 3.856735895492325e-06,
      "loss": 0.5901,
      "step": 2883
    },
    {
      "epoch": 0.9756427604871448,
      "grad_norm": 0.3069666624069214,
      "learning_rate": 3.855983436521698e-06,
      "loss": 0.6233,
      "step": 2884
    },
    {
      "epoch": 0.975981055480379,
      "grad_norm": 0.2997592091560364,
      "learning_rate": 3.855230803469402e-06,
      "loss": 0.6072,
      "step": 2885
    },
    {
      "epoch": 0.976319350473613,
      "grad_norm": 0.3085188567638397,
      "learning_rate": 3.854477996432059e-06,
      "loss": 0.6224,
      "step": 2886
    },
    {
      "epoch": 0.9766576454668471,
      "grad_norm": 0.3022494912147522,
      "learning_rate": 3.8537250155063155e-06,
      "loss": 0.5949,
      "step": 2887
    },
    {
      "epoch": 0.9769959404600812,
      "grad_norm": 0.3045576512813568,
      "learning_rate": 3.8529718607888394e-06,
      "loss": 0.5966,
      "step": 2888
    },
    {
      "epoch": 0.9773342354533153,
      "grad_norm": 0.31438544392585754,
      "learning_rate": 3.8522185323763215e-06,
      "loss": 0.6054,
      "step": 2889
    },
    {
      "epoch": 0.9776725304465493,
      "grad_norm": 0.30769291520118713,
      "learning_rate": 3.851465030365473e-06,
      "loss": 0.6155,
      "step": 2890
    },
    {
      "epoch": 0.9780108254397835,
      "grad_norm": 0.31114768981933594,
      "learning_rate": 3.850711354853031e-06,
      "loss": 0.6059,
      "step": 2891
    },
    {
      "epoch": 0.9783491204330176,
      "grad_norm": 0.3039344251155853,
      "learning_rate": 3.849957505935751e-06,
      "loss": 0.5995,
      "step": 2892
    },
    {
      "epoch": 0.9786874154262517,
      "grad_norm": 0.309730589389801,
      "learning_rate": 3.849203483710412e-06,
      "loss": 0.6121,
      "step": 2893
    },
    {
      "epoch": 0.9790257104194858,
      "grad_norm": 0.3096121847629547,
      "learning_rate": 3.8484492882738165e-06,
      "loss": 0.6118,
      "step": 2894
    },
    {
      "epoch": 0.9793640054127198,
      "grad_norm": 0.30355656147003174,
      "learning_rate": 3.847694919722789e-06,
      "loss": 0.6314,
      "step": 2895
    },
    {
      "epoch": 0.979702300405954,
      "grad_norm": 0.3071487545967102,
      "learning_rate": 3.8469403781541745e-06,
      "loss": 0.6022,
      "step": 2896
    },
    {
      "epoch": 0.9800405953991881,
      "grad_norm": 0.3176629841327667,
      "learning_rate": 3.846185663664842e-06,
      "loss": 0.606,
      "step": 2897
    },
    {
      "epoch": 0.9803788903924222,
      "grad_norm": 0.30271345376968384,
      "learning_rate": 3.845430776351682e-06,
      "loss": 0.6023,
      "step": 2898
    },
    {
      "epoch": 0.9807171853856563,
      "grad_norm": 0.29765617847442627,
      "learning_rate": 3.844675716311607e-06,
      "loss": 0.5989,
      "step": 2899
    },
    {
      "epoch": 0.9810554803788903,
      "grad_norm": 0.2919057309627533,
      "learning_rate": 3.843920483641551e-06,
      "loss": 0.6009,
      "step": 2900
    },
    {
      "epoch": 0.9813937753721245,
      "grad_norm": 0.31493720412254333,
      "learning_rate": 3.843165078438473e-06,
      "loss": 0.601,
      "step": 2901
    },
    {
      "epoch": 0.9817320703653586,
      "grad_norm": 0.3035096526145935,
      "learning_rate": 3.842409500799351e-06,
      "loss": 0.6052,
      "step": 2902
    },
    {
      "epoch": 0.9820703653585927,
      "grad_norm": 0.30932676792144775,
      "learning_rate": 3.841653750821185e-06,
      "loss": 0.5919,
      "step": 2903
    },
    {
      "epoch": 0.9824086603518268,
      "grad_norm": 0.3056243062019348,
      "learning_rate": 3.840897828601002e-06,
      "loss": 0.6193,
      "step": 2904
    },
    {
      "epoch": 0.9827469553450608,
      "grad_norm": 0.3046973645687103,
      "learning_rate": 3.840141734235845e-06,
      "loss": 0.6029,
      "step": 2905
    },
    {
      "epoch": 0.983085250338295,
      "grad_norm": 0.30299368500709534,
      "learning_rate": 3.839385467822782e-06,
      "loss": 0.6203,
      "step": 2906
    },
    {
      "epoch": 0.9834235453315291,
      "grad_norm": 0.29637983441352844,
      "learning_rate": 3.838629029458903e-06,
      "loss": 0.605,
      "step": 2907
    },
    {
      "epoch": 0.9837618403247632,
      "grad_norm": 0.2971667945384979,
      "learning_rate": 3.837872419241321e-06,
      "loss": 0.6026,
      "step": 2908
    },
    {
      "epoch": 0.9841001353179973,
      "grad_norm": 0.3333316147327423,
      "learning_rate": 3.837115637267167e-06,
      "loss": 0.6092,
      "step": 2909
    },
    {
      "epoch": 0.9844384303112313,
      "grad_norm": 0.31869786977767944,
      "learning_rate": 3.8363586836336e-06,
      "loss": 0.6202,
      "step": 2910
    },
    {
      "epoch": 0.9847767253044655,
      "grad_norm": 0.33361250162124634,
      "learning_rate": 3.835601558437796e-06,
      "loss": 0.5869,
      "step": 2911
    },
    {
      "epoch": 0.9851150202976996,
      "grad_norm": 0.30500105023384094,
      "learning_rate": 3.834844261776957e-06,
      "loss": 0.628,
      "step": 2912
    },
    {
      "epoch": 0.9854533152909337,
      "grad_norm": 0.2937752902507782,
      "learning_rate": 3.834086793748302e-06,
      "loss": 0.6018,
      "step": 2913
    },
    {
      "epoch": 0.9857916102841678,
      "grad_norm": 0.3345789909362793,
      "learning_rate": 3.833329154449078e-06,
      "loss": 0.6189,
      "step": 2914
    },
    {
      "epoch": 0.9861299052774019,
      "grad_norm": 0.31569352746009827,
      "learning_rate": 3.83257134397655e-06,
      "loss": 0.6182,
      "step": 2915
    },
    {
      "epoch": 0.986468200270636,
      "grad_norm": 0.3059478998184204,
      "learning_rate": 3.831813362428005e-06,
      "loss": 0.615,
      "step": 2916
    },
    {
      "epoch": 0.9868064952638701,
      "grad_norm": 0.31113189458847046,
      "learning_rate": 3.831055209900754e-06,
      "loss": 0.608,
      "step": 2917
    },
    {
      "epoch": 0.9871447902571042,
      "grad_norm": 0.3061697781085968,
      "learning_rate": 3.830296886492129e-06,
      "loss": 0.6223,
      "step": 2918
    },
    {
      "epoch": 0.9874830852503383,
      "grad_norm": 0.3173771798610687,
      "learning_rate": 3.829538392299484e-06,
      "loss": 0.6239,
      "step": 2919
    },
    {
      "epoch": 0.9878213802435724,
      "grad_norm": 0.3158840537071228,
      "learning_rate": 3.828779727420193e-06,
      "loss": 0.614,
      "step": 2920
    },
    {
      "epoch": 0.9881596752368065,
      "grad_norm": 0.2991497218608856,
      "learning_rate": 3.828020891951656e-06,
      "loss": 0.5877,
      "step": 2921
    },
    {
      "epoch": 0.9884979702300406,
      "grad_norm": 0.3036811947822571,
      "learning_rate": 3.8272618859912914e-06,
      "loss": 0.6183,
      "step": 2922
    },
    {
      "epoch": 0.9888362652232747,
      "grad_norm": 0.3067082464694977,
      "learning_rate": 3.826502709636541e-06,
      "loss": 0.606,
      "step": 2923
    },
    {
      "epoch": 0.9891745602165088,
      "grad_norm": 0.3082413971424103,
      "learning_rate": 3.825743362984868e-06,
      "loss": 0.5925,
      "step": 2924
    },
    {
      "epoch": 0.9895128552097429,
      "grad_norm": 0.3056364059448242,
      "learning_rate": 3.824983846133758e-06,
      "loss": 0.5819,
      "step": 2925
    },
    {
      "epoch": 0.989851150202977,
      "grad_norm": 0.3220091164112091,
      "learning_rate": 3.824224159180717e-06,
      "loss": 0.6179,
      "step": 2926
    },
    {
      "epoch": 0.9901894451962111,
      "grad_norm": 0.31654736399650574,
      "learning_rate": 3.8234643022232756e-06,
      "loss": 0.5863,
      "step": 2927
    },
    {
      "epoch": 0.9905277401894452,
      "grad_norm": 0.31448447704315186,
      "learning_rate": 3.822704275358983e-06,
      "loss": 0.6154,
      "step": 2928
    },
    {
      "epoch": 0.9908660351826793,
      "grad_norm": 0.29887446761131287,
      "learning_rate": 3.821944078685411e-06,
      "loss": 0.6123,
      "step": 2929
    },
    {
      "epoch": 0.9912043301759134,
      "grad_norm": 0.3173508942127228,
      "learning_rate": 3.8211837123001564e-06,
      "loss": 0.6175,
      "step": 2930
    },
    {
      "epoch": 0.9915426251691475,
      "grad_norm": 0.3123912513256073,
      "learning_rate": 3.820423176300834e-06,
      "loss": 0.5858,
      "step": 2931
    },
    {
      "epoch": 0.9918809201623816,
      "grad_norm": 0.2958592176437378,
      "learning_rate": 3.819662470785082e-06,
      "loss": 0.6147,
      "step": 2932
    },
    {
      "epoch": 0.9922192151556157,
      "grad_norm": 0.30872708559036255,
      "learning_rate": 3.818901595850561e-06,
      "loss": 0.6271,
      "step": 2933
    },
    {
      "epoch": 0.9925575101488497,
      "grad_norm": 0.30018481612205505,
      "learning_rate": 3.81814055159495e-06,
      "loss": 0.6332,
      "step": 2934
    },
    {
      "epoch": 0.9928958051420839,
      "grad_norm": 0.3036275804042816,
      "learning_rate": 3.817379338115953e-06,
      "loss": 0.6368,
      "step": 2935
    },
    {
      "epoch": 0.993234100135318,
      "grad_norm": 0.3049243092536926,
      "learning_rate": 3.816617955511296e-06,
      "loss": 0.5834,
      "step": 2936
    },
    {
      "epoch": 0.9935723951285521,
      "grad_norm": 0.3022926449775696,
      "learning_rate": 3.815856403878724e-06,
      "loss": 0.5902,
      "step": 2937
    },
    {
      "epoch": 0.9939106901217862,
      "grad_norm": 0.31031396985054016,
      "learning_rate": 3.815094683316007e-06,
      "loss": 0.5986,
      "step": 2938
    },
    {
      "epoch": 0.9942489851150202,
      "grad_norm": 0.30683907866477966,
      "learning_rate": 3.814332793920935e-06,
      "loss": 0.5875,
      "step": 2939
    },
    {
      "epoch": 0.9945872801082544,
      "grad_norm": 0.3027743995189667,
      "learning_rate": 3.8135707357913177e-06,
      "loss": 0.5862,
      "step": 2940
    },
    {
      "epoch": 0.9949255751014885,
      "grad_norm": 0.2959025502204895,
      "learning_rate": 3.81280850902499e-06,
      "loss": 0.5967,
      "step": 2941
    },
    {
      "epoch": 0.9952638700947226,
      "grad_norm": 0.3143369257450104,
      "learning_rate": 3.812046113719806e-06,
      "loss": 0.6155,
      "step": 2942
    },
    {
      "epoch": 0.9956021650879567,
      "grad_norm": 0.3105371594429016,
      "learning_rate": 3.8112835499736434e-06,
      "loss": 0.6392,
      "step": 2943
    },
    {
      "epoch": 0.9959404600811907,
      "grad_norm": 0.3036314845085144,
      "learning_rate": 3.8105208178843988e-06,
      "loss": 0.5768,
      "step": 2944
    },
    {
      "epoch": 0.9962787550744249,
      "grad_norm": 0.30836254358291626,
      "learning_rate": 3.8097579175499942e-06,
      "loss": 0.625,
      "step": 2945
    },
    {
      "epoch": 0.996617050067659,
      "grad_norm": 0.29898902773857117,
      "learning_rate": 3.8089948490683692e-06,
      "loss": 0.6177,
      "step": 2946
    },
    {
      "epoch": 0.9969553450608931,
      "grad_norm": 0.3097745180130005,
      "learning_rate": 3.8082316125374874e-06,
      "loss": 0.5863,
      "step": 2947
    },
    {
      "epoch": 0.9972936400541272,
      "grad_norm": 0.3105376362800598,
      "learning_rate": 3.807468208055334e-06,
      "loss": 0.6305,
      "step": 2948
    },
    {
      "epoch": 0.9976319350473613,
      "grad_norm": 0.30030593276023865,
      "learning_rate": 3.8067046357199135e-06,
      "loss": 0.6263,
      "step": 2949
    },
    {
      "epoch": 0.9979702300405954,
      "grad_norm": 0.3033324182033539,
      "learning_rate": 3.8059408956292553e-06,
      "loss": 0.5988,
      "step": 2950
    },
    {
      "epoch": 0.9983085250338295,
      "grad_norm": 0.30815133452415466,
      "learning_rate": 3.805176987881408e-06,
      "loss": 0.5969,
      "step": 2951
    },
    {
      "epoch": 0.9986468200270636,
      "grad_norm": 0.3052274286746979,
      "learning_rate": 3.8044129125744423e-06,
      "loss": 0.6082,
      "step": 2952
    },
    {
      "epoch": 0.9989851150202977,
      "grad_norm": 0.3146717846393585,
      "learning_rate": 3.8036486698064513e-06,
      "loss": 0.5964,
      "step": 2953
    },
    {
      "epoch": 0.9993234100135318,
      "grad_norm": 0.33001822233200073,
      "learning_rate": 3.8028842596755467e-06,
      "loss": 0.6081,
      "step": 2954
    },
    {
      "epoch": 0.9996617050067659,
      "grad_norm": 0.30166327953338623,
      "learning_rate": 3.8021196822798656e-06,
      "loss": 0.5951,
      "step": 2955
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.32355165481567383,
      "learning_rate": 3.8013549377175653e-06,
      "loss": 0.6117,
      "step": 2956
    }
  ],
  "logging_steps": 1,
  "max_steps": 8868,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 2956,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.648689084357778e+19,
  "train_batch_size": 12,
  "trial_name": null,
  "trial_params": null
}