{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.7389493483810291,
  "eval_steps": 500,
  "global_step": 27500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00013435442697836894,
      "grad_norm": 11.01321029663086,
      "learning_rate": 5.000000000000001e-07,
      "loss": 9.7918,
      "step": 5
    },
    {
      "epoch": 0.0002687088539567379,
      "grad_norm": 11.176947593688965,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 9.6385,
      "step": 10
    },
    {
      "epoch": 0.00040306328093510683,
      "grad_norm": 10.01496410369873,
      "learning_rate": 1.5e-06,
      "loss": 9.5219,
      "step": 15
    },
    {
      "epoch": 0.0005374177079134758,
      "grad_norm": 7.92938756942749,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 9.2768,
      "step": 20
    },
    {
      "epoch": 0.0006717721348918447,
      "grad_norm": 6.977506637573242,
      "learning_rate": 2.5e-06,
      "loss": 9.076,
      "step": 25
    },
    {
      "epoch": 0.0008061265618702137,
      "grad_norm": 5.584914684295654,
      "learning_rate": 3e-06,
      "loss": 8.8641,
      "step": 30
    },
    {
      "epoch": 0.0009404809888485826,
      "grad_norm": 4.832436561584473,
      "learning_rate": 3.5000000000000004e-06,
      "loss": 8.7415,
      "step": 35
    },
    {
      "epoch": 0.0010748354158269515,
      "grad_norm": 4.32351016998291,
      "learning_rate": 4.000000000000001e-06,
      "loss": 8.4831,
      "step": 40
    },
    {
      "epoch": 0.0012091898428053204,
      "grad_norm": 3.6954407691955566,
      "learning_rate": 4.5e-06,
      "loss": 8.343,
      "step": 45
    },
    {
      "epoch": 0.0013435442697836894,
      "grad_norm": 3.2270238399505615,
      "learning_rate": 5e-06,
      "loss": 8.1544,
      "step": 50
    },
    {
      "epoch": 0.0014778986967620583,
      "grad_norm": 2.855537176132202,
      "learning_rate": 5.500000000000001e-06,
      "loss": 8.0265,
      "step": 55
    },
    {
      "epoch": 0.0016122531237404273,
      "grad_norm": 2.827409505844116,
      "learning_rate": 6e-06,
      "loss": 7.9643,
      "step": 60
    },
    {
      "epoch": 0.0017466075507187962,
      "grad_norm": 2.896108627319336,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 7.9864,
      "step": 65
    },
    {
      "epoch": 0.0018809619776971652,
      "grad_norm": 2.5346200466156006,
      "learning_rate": 7.000000000000001e-06,
      "loss": 7.7691,
      "step": 70
    },
    {
      "epoch": 0.002015316404675534,
      "grad_norm": 2.589860200881958,
      "learning_rate": 7.5e-06,
      "loss": 7.8421,
      "step": 75
    },
    {
      "epoch": 0.002149670831653903,
      "grad_norm": 2.595316171646118,
      "learning_rate": 8.000000000000001e-06,
      "loss": 7.8043,
      "step": 80
    },
    {
      "epoch": 0.002284025258632272,
      "grad_norm": 2.718700885772705,
      "learning_rate": 8.500000000000002e-06,
      "loss": 7.7434,
      "step": 85
    },
    {
      "epoch": 0.0024183796856106408,
      "grad_norm": 3.426335096359253,
      "learning_rate": 9e-06,
      "loss": 7.6604,
      "step": 90
    },
    {
      "epoch": 0.00255273411258901,
      "grad_norm": 2.714006185531616,
      "learning_rate": 9.5e-06,
      "loss": 7.4985,
      "step": 95
    },
    {
      "epoch": 0.002687088539567379,
      "grad_norm": 2.5804998874664307,
      "learning_rate": 1e-05,
      "loss": 7.5679,
      "step": 100
    },
    {
      "epoch": 0.0028214429665457475,
      "grad_norm": 2.799619436264038,
      "learning_rate": 1.05e-05,
      "loss": 7.4543,
      "step": 105
    },
    {
      "epoch": 0.0029557973935241165,
      "grad_norm": 2.6874120235443115,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 7.5767,
      "step": 110
    },
    {
      "epoch": 0.0030901518205024856,
      "grad_norm": 2.5367376804351807,
      "learning_rate": 1.1500000000000002e-05,
      "loss": 7.4019,
      "step": 115
    },
    {
      "epoch": 0.0032245062474808546,
      "grad_norm": 2.844397783279419,
      "learning_rate": 1.2e-05,
      "loss": 7.4102,
      "step": 120
    },
    {
      "epoch": 0.0033588606744592233,
      "grad_norm": 3.0461487770080566,
      "learning_rate": 1.25e-05,
      "loss": 7.371,
      "step": 125
    },
    {
      "epoch": 0.0034932151014375923,
      "grad_norm": 2.188748359680176,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 7.2679,
      "step": 130
    },
    {
      "epoch": 0.0036275695284159614,
      "grad_norm": 2.494464874267578,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 7.2771,
      "step": 135
    },
    {
      "epoch": 0.0037619239553943304,
      "grad_norm": 2.836573362350464,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 7.2517,
      "step": 140
    },
    {
      "epoch": 0.003896278382372699,
      "grad_norm": 3.0456364154815674,
      "learning_rate": 1.45e-05,
      "loss": 7.1899,
      "step": 145
    },
    {
      "epoch": 0.004030632809351068,
      "grad_norm": 2.731494903564453,
      "learning_rate": 1.5e-05,
      "loss": 7.1298,
      "step": 150
    },
    {
      "epoch": 0.004164987236329437,
      "grad_norm": 2.6526591777801514,
      "learning_rate": 1.55e-05,
      "loss": 7.1122,
      "step": 155
    },
    {
      "epoch": 0.004299341663307806,
      "grad_norm": 2.3740646839141846,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 7.2048,
      "step": 160
    },
    {
      "epoch": 0.004433696090286175,
      "grad_norm": 2.8663558959960938,
      "learning_rate": 1.65e-05,
      "loss": 7.0526,
      "step": 165
    },
    {
      "epoch": 0.004568050517264544,
      "grad_norm": 2.5205307006835938,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 7.0873,
      "step": 170
    },
    {
      "epoch": 0.0047024049442429125,
      "grad_norm": 2.7289395332336426,
      "learning_rate": 1.75e-05,
      "loss": 6.8913,
      "step": 175
    },
    {
      "epoch": 0.0048367593712212815,
      "grad_norm": 2.7674458026885986,
      "learning_rate": 1.8e-05,
      "loss": 7.0275,
      "step": 180
    },
    {
      "epoch": 0.004971113798199651,
      "grad_norm": 2.6979026794433594,
      "learning_rate": 1.85e-05,
      "loss": 7.0515,
      "step": 185
    },
    {
      "epoch": 0.00510546822517802,
      "grad_norm": 2.633976936340332,
      "learning_rate": 1.9e-05,
      "loss": 6.9048,
      "step": 190
    },
    {
      "epoch": 0.005239822652156389,
      "grad_norm": 2.978367567062378,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 6.8549,
      "step": 195
    },
    {
      "epoch": 0.005374177079134758,
      "grad_norm": 2.6202895641326904,
      "learning_rate": 2e-05,
      "loss": 6.8476,
      "step": 200
    },
    {
      "epoch": 0.005508531506113127,
      "grad_norm": 3.0189406871795654,
      "learning_rate": 2.05e-05,
      "loss": 6.7656,
      "step": 205
    },
    {
      "epoch": 0.005642885933091495,
      "grad_norm": 2.660745143890381,
      "learning_rate": 2.1e-05,
      "loss": 6.7942,
      "step": 210
    },
    {
      "epoch": 0.005777240360069864,
      "grad_norm": 2.7300496101379395,
      "learning_rate": 2.15e-05,
      "loss": 6.8475,
      "step": 215
    },
    {
      "epoch": 0.005911594787048233,
      "grad_norm": 2.794863700866699,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 6.72,
      "step": 220
    },
    {
      "epoch": 0.006045949214026602,
      "grad_norm": 3.3329789638519287,
      "learning_rate": 2.25e-05,
      "loss": 6.6472,
      "step": 225
    },
    {
      "epoch": 0.006180303641004971,
      "grad_norm": 3.2194576263427734,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 6.7812,
      "step": 230
    },
    {
      "epoch": 0.00631465806798334,
      "grad_norm": 2.758150100708008,
      "learning_rate": 2.35e-05,
      "loss": 6.7237,
      "step": 235
    },
    {
      "epoch": 0.006449012494961709,
      "grad_norm": 2.5040268898010254,
      "learning_rate": 2.4e-05,
      "loss": 6.6982,
      "step": 240
    },
    {
      "epoch": 0.006583366921940078,
      "grad_norm": 2.948148727416992,
      "learning_rate": 2.45e-05,
      "loss": 6.501,
      "step": 245
    },
    {
      "epoch": 0.0067177213489184465,
      "grad_norm": 2.9348082542419434,
      "learning_rate": 2.5e-05,
      "loss": 6.7335,
      "step": 250
    },
    {
      "epoch": 0.0068520757758968156,
      "grad_norm": 2.4086520671844482,
      "learning_rate": 2.5500000000000003e-05,
      "loss": 6.6263,
      "step": 255
    },
    {
      "epoch": 0.006986430202875185,
      "grad_norm": 3.0014560222625732,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 6.621,
      "step": 260
    },
    {
      "epoch": 0.007120784629853554,
      "grad_norm": 2.662444591522217,
      "learning_rate": 2.6500000000000004e-05,
      "loss": 6.5413,
      "step": 265
    },
    {
      "epoch": 0.007255139056831923,
      "grad_norm": 3.1769156455993652,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 6.6365,
      "step": 270
    },
    {
      "epoch": 0.007389493483810292,
      "grad_norm": 2.5750327110290527,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 6.5601,
      "step": 275
    },
    {
      "epoch": 0.007523847910788661,
      "grad_norm": 2.687174081802368,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 6.7257,
      "step": 280
    },
    {
      "epoch": 0.007658202337767029,
      "grad_norm": 2.814142942428589,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 6.6035,
      "step": 285
    },
    {
      "epoch": 0.007792556764745398,
      "grad_norm": 2.5843849182128906,
      "learning_rate": 2.9e-05,
      "loss": 6.5071,
      "step": 290
    },
    {
      "epoch": 0.007926911191723767,
      "grad_norm": 2.9305198192596436,
      "learning_rate": 2.95e-05,
      "loss": 6.632,
      "step": 295
    },
    {
      "epoch": 0.008061265618702136,
      "grad_norm": 2.8853163719177246,
      "learning_rate": 3e-05,
      "loss": 6.5922,
      "step": 300
    },
    {
      "epoch": 0.008195620045680505,
      "grad_norm": 2.5634713172912598,
      "learning_rate": 3.05e-05,
      "loss": 6.4593,
      "step": 305
    },
    {
      "epoch": 0.008329974472658874,
      "grad_norm": 2.38299298286438,
      "learning_rate": 3.1e-05,
      "loss": 6.5774,
      "step": 310
    },
    {
      "epoch": 0.008464328899637243,
      "grad_norm": 2.8081862926483154,
      "learning_rate": 3.15e-05,
      "loss": 6.4521,
      "step": 315
    },
    {
      "epoch": 0.008598683326615612,
      "grad_norm": 2.3867132663726807,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 6.463,
      "step": 320
    },
    {
      "epoch": 0.008733037753593981,
      "grad_norm": 2.355497360229492,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 6.453,
      "step": 325
    },
    {
      "epoch": 0.00886739218057235,
      "grad_norm": 2.6186928749084473,
      "learning_rate": 3.3e-05,
      "loss": 6.3419,
      "step": 330
    },
    {
      "epoch": 0.00900174660755072,
      "grad_norm": 2.305065631866455,
      "learning_rate": 3.35e-05,
      "loss": 6.437,
      "step": 335
    },
    {
      "epoch": 0.009136101034529089,
      "grad_norm": 2.9329586029052734,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 6.4604,
      "step": 340
    },
    {
      "epoch": 0.009270455461507456,
      "grad_norm": 2.7121264934539795,
      "learning_rate": 3.45e-05,
      "loss": 6.1996,
      "step": 345
    },
    {
      "epoch": 0.009404809888485825,
      "grad_norm": 2.518139123916626,
      "learning_rate": 3.5e-05,
      "loss": 6.3747,
      "step": 350
    },
    {
      "epoch": 0.009539164315464194,
      "grad_norm": 2.612273931503296,
      "learning_rate": 3.55e-05,
      "loss": 6.4051,
      "step": 355
    },
    {
      "epoch": 0.009673518742442563,
      "grad_norm": 2.6475555896759033,
      "learning_rate": 3.6e-05,
      "loss": 6.3917,
      "step": 360
    },
    {
      "epoch": 0.009807873169420932,
      "grad_norm": 2.4955079555511475,
      "learning_rate": 3.65e-05,
      "loss": 6.3295,
      "step": 365
    },
    {
      "epoch": 0.009942227596399301,
      "grad_norm": 2.4744882583618164,
      "learning_rate": 3.7e-05,
      "loss": 6.2833,
      "step": 370
    },
    {
      "epoch": 0.01007658202337767,
      "grad_norm": 2.5392186641693115,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 6.2492,
      "step": 375
    },
    {
      "epoch": 0.01021093645035604,
      "grad_norm": 2.6065213680267334,
      "learning_rate": 3.8e-05,
      "loss": 6.3539,
      "step": 380
    },
    {
      "epoch": 0.010345290877334408,
      "grad_norm": 2.6622867584228516,
      "learning_rate": 3.85e-05,
      "loss": 6.2336,
      "step": 385
    },
    {
      "epoch": 0.010479645304312777,
      "grad_norm": 2.3949472904205322,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 6.2787,
      "step": 390
    },
    {
      "epoch": 0.010613999731291146,
      "grad_norm": 2.6841931343078613,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 6.3568,
      "step": 395
    },
    {
      "epoch": 0.010748354158269515,
      "grad_norm": 2.763218402862549,
      "learning_rate": 4e-05,
      "loss": 6.1288,
      "step": 400
    },
    {
      "epoch": 0.010882708585247884,
      "grad_norm": 2.549846649169922,
      "learning_rate": 4.05e-05,
      "loss": 6.2834,
      "step": 405
    },
    {
      "epoch": 0.011017063012226254,
      "grad_norm": 2.6757774353027344,
      "learning_rate": 4.1e-05,
      "loss": 6.2089,
      "step": 410
    },
    {
      "epoch": 0.011151417439204623,
      "grad_norm": 2.6386969089508057,
      "learning_rate": 4.15e-05,
      "loss": 6.2232,
      "step": 415
    },
    {
      "epoch": 0.01128577186618299,
      "grad_norm": 2.8432271480560303,
      "learning_rate": 4.2e-05,
      "loss": 6.1445,
      "step": 420
    },
    {
      "epoch": 0.011420126293161359,
      "grad_norm": 2.8131535053253174,
      "learning_rate": 4.25e-05,
      "loss": 6.1136,
      "step": 425
    },
    {
      "epoch": 0.011554480720139728,
      "grad_norm": 2.476287841796875,
      "learning_rate": 4.3e-05,
      "loss": 6.1919,
      "step": 430
    },
    {
      "epoch": 0.011688835147118097,
      "grad_norm": 2.672457695007324,
      "learning_rate": 4.35e-05,
      "loss": 6.1827,
      "step": 435
    },
    {
      "epoch": 0.011823189574096466,
      "grad_norm": 2.5320873260498047,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 6.1224,
      "step": 440
    },
    {
      "epoch": 0.011957544001074835,
      "grad_norm": 2.4476144313812256,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 6.1666,
      "step": 445
    },
    {
      "epoch": 0.012091898428053204,
      "grad_norm": 2.737291097640991,
      "learning_rate": 4.5e-05,
      "loss": 6.2006,
      "step": 450
    },
    {
      "epoch": 0.012226252855031573,
      "grad_norm": 2.5629758834838867,
      "learning_rate": 4.55e-05,
      "loss": 6.0941,
      "step": 455
    },
    {
      "epoch": 0.012360607282009942,
      "grad_norm": 2.4126691818237305,
      "learning_rate": 4.600000000000001e-05,
      "loss": 6.0879,
      "step": 460
    },
    {
      "epoch": 0.012494961708988311,
      "grad_norm": 2.3889412879943848,
      "learning_rate": 4.6500000000000005e-05,
      "loss": 6.0713,
      "step": 465
    },
    {
      "epoch": 0.01262931613596668,
      "grad_norm": 2.744819402694702,
      "learning_rate": 4.7e-05,
      "loss": 6.1562,
      "step": 470
    },
    {
      "epoch": 0.01276367056294505,
      "grad_norm": 2.4293222427368164,
      "learning_rate": 4.75e-05,
      "loss": 6.0595,
      "step": 475
    },
    {
      "epoch": 0.012898024989923419,
      "grad_norm": 2.5619595050811768,
      "learning_rate": 4.8e-05,
      "loss": 6.1103,
      "step": 480
    },
    {
      "epoch": 0.013032379416901788,
      "grad_norm": 2.2517168521881104,
      "learning_rate": 4.85e-05,
      "loss": 6.1078,
      "step": 485
    },
    {
      "epoch": 0.013166733843880157,
      "grad_norm": 2.696000814437866,
      "learning_rate": 4.9e-05,
      "loss": 5.9199,
      "step": 490
    },
    {
      "epoch": 0.013301088270858524,
      "grad_norm": 2.3833043575286865,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 6.0985,
      "step": 495
    },
    {
      "epoch": 0.013435442697836893,
      "grad_norm": 2.5483415126800537,
      "learning_rate": 5e-05,
      "loss": 6.0529,
      "step": 500
    },
    {
      "epoch": 0.013569797124815262,
      "grad_norm": 2.545701503753662,
      "learning_rate": 4.9993190793953425e-05,
      "loss": 6.0793,
      "step": 505
    },
    {
      "epoch": 0.013704151551793631,
      "grad_norm": 2.6781485080718994,
      "learning_rate": 4.998638158790685e-05,
      "loss": 6.0248,
      "step": 510
    },
    {
      "epoch": 0.013838505978772,
      "grad_norm": 2.700716972351074,
      "learning_rate": 4.9979572381860276e-05,
      "loss": 6.1312,
      "step": 515
    },
    {
      "epoch": 0.01397286040575037,
      "grad_norm": 2.3939549922943115,
      "learning_rate": 4.9972763175813705e-05,
      "loss": 6.0544,
      "step": 520
    },
    {
      "epoch": 0.014107214832728738,
      "grad_norm": 2.771691083908081,
      "learning_rate": 4.996595396976713e-05,
      "loss": 5.9878,
      "step": 525
    },
    {
      "epoch": 0.014241569259707107,
      "grad_norm": 2.369417428970337,
      "learning_rate": 4.995914476372055e-05,
      "loss": 6.0719,
      "step": 530
    },
    {
      "epoch": 0.014375923686685476,
      "grad_norm": 2.6986169815063477,
      "learning_rate": 4.995233555767398e-05,
      "loss": 6.1208,
      "step": 535
    },
    {
      "epoch": 0.014510278113663845,
      "grad_norm": 2.3913047313690186,
      "learning_rate": 4.99455263516274e-05,
      "loss": 6.0036,
      "step": 540
    },
    {
      "epoch": 0.014644632540642214,
      "grad_norm": 2.569361925125122,
      "learning_rate": 4.993871714558083e-05,
      "loss": 6.0675,
      "step": 545
    },
    {
      "epoch": 0.014778986967620584,
      "grad_norm": 2.287783622741699,
      "learning_rate": 4.993190793953425e-05,
      "loss": 5.9546,
      "step": 550
    },
    {
      "epoch": 0.014913341394598953,
      "grad_norm": 2.8044824600219727,
      "learning_rate": 4.992509873348768e-05,
      "loss": 6.0564,
      "step": 555
    },
    {
      "epoch": 0.015047695821577322,
      "grad_norm": 2.370957612991333,
      "learning_rate": 4.9918289527441104e-05,
      "loss": 5.9391,
      "step": 560
    },
    {
      "epoch": 0.01518205024855569,
      "grad_norm": 2.4755353927612305,
      "learning_rate": 4.9911480321394526e-05,
      "loss": 5.9702,
      "step": 565
    },
    {
      "epoch": 0.015316404675534058,
      "grad_norm": 2.473414421081543,
      "learning_rate": 4.990467111534795e-05,
      "loss": 6.0557,
      "step": 570
    },
    {
      "epoch": 0.015450759102512427,
      "grad_norm": 2.2430734634399414,
      "learning_rate": 4.989786190930138e-05,
      "loss": 6.0534,
      "step": 575
    },
    {
      "epoch": 0.015585113529490796,
      "grad_norm": 2.430976390838623,
      "learning_rate": 4.9891052703254806e-05,
      "loss": 5.7805,
      "step": 580
    },
    {
      "epoch": 0.015719467956469165,
      "grad_norm": 2.832655906677246,
      "learning_rate": 4.988424349720823e-05,
      "loss": 6.0228,
      "step": 585
    },
    {
      "epoch": 0.015853822383447534,
      "grad_norm": 2.3685193061828613,
      "learning_rate": 4.987743429116165e-05,
      "loss": 5.7415,
      "step": 590
    },
    {
      "epoch": 0.015988176810425903,
      "grad_norm": 2.479611396789551,
      "learning_rate": 4.987062508511508e-05,
      "loss": 5.9124,
      "step": 595
    },
    {
      "epoch": 0.016122531237404272,
      "grad_norm": 2.4757654666900635,
      "learning_rate": 4.98638158790685e-05,
      "loss": 5.9235,
      "step": 600
    },
    {
      "epoch": 0.01625688566438264,
      "grad_norm": 2.5986921787261963,
      "learning_rate": 4.9857006673021924e-05,
      "loss": 5.9294,
      "step": 605
    },
    {
      "epoch": 0.01639124009136101,
      "grad_norm": 2.630502939224243,
      "learning_rate": 4.985019746697535e-05,
      "loss": 5.7693,
      "step": 610
    },
    {
      "epoch": 0.01652559451833938,
      "grad_norm": 2.4950783252716064,
      "learning_rate": 4.984338826092878e-05,
      "loss": 5.9103,
      "step": 615
    },
    {
      "epoch": 0.01665994894531775,
      "grad_norm": 2.2728936672210693,
      "learning_rate": 4.9836579054882205e-05,
      "loss": 5.9684,
      "step": 620
    },
    {
      "epoch": 0.016794303372296118,
      "grad_norm": 2.619060516357422,
      "learning_rate": 4.982976984883563e-05,
      "loss": 5.8626,
      "step": 625
    },
    {
      "epoch": 0.016928657799274487,
      "grad_norm": 2.5404937267303467,
      "learning_rate": 4.982296064278905e-05,
      "loss": 5.7829,
      "step": 630
    },
    {
      "epoch": 0.017063012226252856,
      "grad_norm": 3.9172961711883545,
      "learning_rate": 4.981615143674248e-05,
      "loss": 5.9202,
      "step": 635
    },
    {
      "epoch": 0.017197366653231225,
      "grad_norm": 2.43959641456604,
      "learning_rate": 4.980934223069591e-05,
      "loss": 5.8023,
      "step": 640
    },
    {
      "epoch": 0.017331721080209594,
      "grad_norm": 2.476043224334717,
      "learning_rate": 4.980253302464933e-05,
      "loss": 5.9157,
      "step": 645
    },
    {
      "epoch": 0.017466075507187963,
      "grad_norm": 2.211029291152954,
      "learning_rate": 4.979572381860275e-05,
      "loss": 5.7362,
      "step": 650
    },
    {
      "epoch": 0.017600429934166332,
      "grad_norm": 2.5167977809906006,
      "learning_rate": 4.9788914612556174e-05,
      "loss": 5.9221,
      "step": 655
    },
    {
      "epoch": 0.0177347843611447,
      "grad_norm": 2.3679065704345703,
      "learning_rate": 4.97821054065096e-05,
      "loss": 5.736,
      "step": 660
    },
    {
      "epoch": 0.01786913878812307,
      "grad_norm": 2.331702947616577,
      "learning_rate": 4.9775296200463025e-05,
      "loss": 5.9375,
      "step": 665
    },
    {
      "epoch": 0.01800349321510144,
      "grad_norm": 2.4434773921966553,
      "learning_rate": 4.9768486994416454e-05,
      "loss": 5.9481,
      "step": 670
    },
    {
      "epoch": 0.018137847642079808,
      "grad_norm": 2.8634090423583984,
      "learning_rate": 4.976167778836988e-05,
      "loss": 5.8105,
      "step": 675
    },
    {
      "epoch": 0.018272202069058177,
      "grad_norm": 2.486222743988037,
      "learning_rate": 4.9754868582323306e-05,
      "loss": 5.8775,
      "step": 680
    },
    {
      "epoch": 0.018406556496036543,
      "grad_norm": 2.2074220180511475,
      "learning_rate": 4.974805937627673e-05,
      "loss": 5.9407,
      "step": 685
    },
    {
      "epoch": 0.018540910923014912,
      "grad_norm": 2.3490865230560303,
      "learning_rate": 4.974125017023015e-05,
      "loss": 5.7635,
      "step": 690
    },
    {
      "epoch": 0.01867526534999328,
      "grad_norm": 2.230475902557373,
      "learning_rate": 4.973444096418358e-05,
      "loss": 5.773,
      "step": 695
    },
    {
      "epoch": 0.01880961977697165,
      "grad_norm": 2.3701012134552,
      "learning_rate": 4.972763175813701e-05,
      "loss": 5.8768,
      "step": 700
    },
    {
      "epoch": 0.01894397420395002,
      "grad_norm": 2.5682291984558105,
      "learning_rate": 4.972082255209043e-05,
      "loss": 5.8418,
      "step": 705
    },
    {
      "epoch": 0.019078328630928388,
      "grad_norm": 2.3219289779663086,
      "learning_rate": 4.971401334604385e-05,
      "loss": 5.812,
      "step": 710
    },
    {
      "epoch": 0.019212683057906757,
      "grad_norm": 2.6463623046875,
      "learning_rate": 4.9707204139997275e-05,
      "loss": 5.7989,
      "step": 715
    },
    {
      "epoch": 0.019347037484885126,
      "grad_norm": 2.8326807022094727,
      "learning_rate": 4.9700394933950704e-05,
      "loss": 5.7261,
      "step": 720
    },
    {
      "epoch": 0.019481391911863495,
      "grad_norm": 2.723139524459839,
      "learning_rate": 4.9693585727904126e-05,
      "loss": 5.5954,
      "step": 725
    },
    {
      "epoch": 0.019615746338841864,
      "grad_norm": 2.45609712600708,
      "learning_rate": 4.9686776521857556e-05,
      "loss": 5.9225,
      "step": 730
    },
    {
      "epoch": 0.019750100765820233,
      "grad_norm": 1.9900792837142944,
      "learning_rate": 4.967996731581098e-05,
      "loss": 5.7652,
      "step": 735
    },
    {
      "epoch": 0.019884455192798602,
      "grad_norm": 2.4097843170166016,
      "learning_rate": 4.967315810976441e-05,
      "loss": 5.7977,
      "step": 740
    },
    {
      "epoch": 0.02001880961977697,
      "grad_norm": 2.3949177265167236,
      "learning_rate": 4.966634890371783e-05,
      "loss": 5.8127,
      "step": 745
    },
    {
      "epoch": 0.02015316404675534,
      "grad_norm": 2.2827463150024414,
      "learning_rate": 4.965953969767125e-05,
      "loss": 5.6738,
      "step": 750
    },
    {
      "epoch": 0.02028751847373371,
      "grad_norm": 2.2521233558654785,
      "learning_rate": 4.965273049162468e-05,
      "loss": 5.8771,
      "step": 755
    },
    {
      "epoch": 0.02042187290071208,
      "grad_norm": 2.4549942016601562,
      "learning_rate": 4.964592128557811e-05,
      "loss": 5.6877,
      "step": 760
    },
    {
      "epoch": 0.020556227327690448,
      "grad_norm": 2.5722155570983887,
      "learning_rate": 4.963911207953153e-05,
      "loss": 5.7352,
      "step": 765
    },
    {
      "epoch": 0.020690581754668817,
      "grad_norm": 2.1596367359161377,
      "learning_rate": 4.9632302873484954e-05,
      "loss": 5.7694,
      "step": 770
    },
    {
      "epoch": 0.020824936181647186,
      "grad_norm": 2.4517874717712402,
      "learning_rate": 4.9625493667438376e-05,
      "loss": 5.7355,
      "step": 775
    },
    {
      "epoch": 0.020959290608625555,
      "grad_norm": 2.0883655548095703,
      "learning_rate": 4.9618684461391805e-05,
      "loss": 5.673,
      "step": 780
    },
    {
      "epoch": 0.021093645035603924,
      "grad_norm": 2.5619287490844727,
      "learning_rate": 4.961187525534523e-05,
      "loss": 5.7157,
      "step": 785
    },
    {
      "epoch": 0.021227999462582293,
      "grad_norm": 2.2422034740448,
      "learning_rate": 4.960506604929866e-05,
      "loss": 5.7318,
      "step": 790
    },
    {
      "epoch": 0.021362353889560662,
      "grad_norm": 2.5861570835113525,
      "learning_rate": 4.959825684325208e-05,
      "loss": 5.6878,
      "step": 795
    },
    {
      "epoch": 0.02149670831653903,
      "grad_norm": 2.492201089859009,
      "learning_rate": 4.95914476372055e-05,
      "loss": 5.6216,
      "step": 800
    },
    {
      "epoch": 0.0216310627435174,
      "grad_norm": 2.8234431743621826,
      "learning_rate": 4.958463843115893e-05,
      "loss": 5.7494,
      "step": 805
    },
    {
      "epoch": 0.02176541717049577,
      "grad_norm": 2.094003915786743,
      "learning_rate": 4.957782922511235e-05,
      "loss": 5.7582,
      "step": 810
    },
    {
      "epoch": 0.021899771597474138,
      "grad_norm": 2.393399715423584,
      "learning_rate": 4.9571020019065775e-05,
      "loss": 5.7385,
      "step": 815
    },
    {
      "epoch": 0.022034126024452507,
      "grad_norm": 2.3262112140655518,
      "learning_rate": 4.9564210813019204e-05,
      "loss": 5.7897,
      "step": 820
    },
    {
      "epoch": 0.022168480451430876,
      "grad_norm": 2.2483649253845215,
      "learning_rate": 4.955740160697263e-05,
      "loss": 5.7137,
      "step": 825
    },
    {
      "epoch": 0.022302834878409245,
      "grad_norm": 2.1380503177642822,
      "learning_rate": 4.9550592400926055e-05,
      "loss": 5.5507,
      "step": 830
    },
    {
      "epoch": 0.022437189305387614,
      "grad_norm": 2.464599132537842,
      "learning_rate": 4.954378319487948e-05,
      "loss": 5.6652,
      "step": 835
    },
    {
      "epoch": 0.02257154373236598,
      "grad_norm": 2.2616629600524902,
      "learning_rate": 4.95369739888329e-05,
      "loss": 5.7686,
      "step": 840
    },
    {
      "epoch": 0.02270589815934435,
      "grad_norm": 2.13108491897583,
      "learning_rate": 4.953016478278633e-05,
      "loss": 5.6843,
      "step": 845
    },
    {
      "epoch": 0.022840252586322718,
      "grad_norm": 2.5671236515045166,
      "learning_rate": 4.952335557673976e-05,
      "loss": 5.8,
      "step": 850
    },
    {
      "epoch": 0.022974607013301087,
      "grad_norm": 2.473129987716675,
      "learning_rate": 4.951654637069318e-05,
      "loss": 5.6215,
      "step": 855
    },
    {
      "epoch": 0.023108961440279456,
      "grad_norm": 2.5920064449310303,
      "learning_rate": 4.95097371646466e-05,
      "loss": 5.7515,
      "step": 860
    },
    {
      "epoch": 0.023243315867257825,
      "grad_norm": 2.3798274993896484,
      "learning_rate": 4.950292795860003e-05,
      "loss": 5.6959,
      "step": 865
    },
    {
      "epoch": 0.023377670294236194,
      "grad_norm": 2.2634427547454834,
      "learning_rate": 4.9496118752553454e-05,
      "loss": 5.646,
      "step": 870
    },
    {
      "epoch": 0.023512024721214563,
      "grad_norm": 2.3553104400634766,
      "learning_rate": 4.9489309546506876e-05,
      "loss": 5.555,
      "step": 875
    },
    {
      "epoch": 0.023646379148192932,
      "grad_norm": 2.5519611835479736,
      "learning_rate": 4.9482500340460305e-05,
      "loss": 5.632,
      "step": 880
    },
    {
      "epoch": 0.0237807335751713,
      "grad_norm": 2.3024661540985107,
      "learning_rate": 4.9475691134413734e-05,
      "loss": 5.6788,
      "step": 885
    },
    {
      "epoch": 0.02391508800214967,
      "grad_norm": 2.5067193508148193,
      "learning_rate": 4.9468881928367156e-05,
      "loss": 5.6975,
      "step": 890
    },
    {
      "epoch": 0.02404944242912804,
      "grad_norm": 2.397663116455078,
      "learning_rate": 4.946207272232058e-05,
      "loss": 5.7011,
      "step": 895
    },
    {
      "epoch": 0.02418379685610641,
      "grad_norm": 2.439544200897217,
      "learning_rate": 4.9455263516274e-05,
      "loss": 5.5824,
      "step": 900
    },
    {
      "epoch": 0.024318151283084777,
      "grad_norm": 2.6345417499542236,
      "learning_rate": 4.944845431022743e-05,
      "loss": 5.739,
      "step": 905
    },
    {
      "epoch": 0.024452505710063147,
      "grad_norm": 2.163940668106079,
      "learning_rate": 4.944164510418086e-05,
      "loss": 5.5987,
      "step": 910
    },
    {
      "epoch": 0.024586860137041516,
      "grad_norm": 2.3982248306274414,
      "learning_rate": 4.943483589813428e-05,
      "loss": 5.714,
      "step": 915
    },
    {
      "epoch": 0.024721214564019885,
      "grad_norm": 2.4334447383880615,
      "learning_rate": 4.9428026692087703e-05,
      "loss": 5.7488,
      "step": 920
    },
    {
      "epoch": 0.024855568990998254,
      "grad_norm": 2.416102409362793,
      "learning_rate": 4.942121748604113e-05,
      "loss": 5.5881,
      "step": 925
    },
    {
      "epoch": 0.024989923417976623,
      "grad_norm": 2.588106632232666,
      "learning_rate": 4.9414408279994555e-05,
      "loss": 5.5839,
      "step": 930
    },
    {
      "epoch": 0.025124277844954992,
      "grad_norm": 2.413837432861328,
      "learning_rate": 4.940759907394798e-05,
      "loss": 5.769,
      "step": 935
    },
    {
      "epoch": 0.02525863227193336,
      "grad_norm": 2.239147186279297,
      "learning_rate": 4.9400789867901406e-05,
      "loss": 5.7587,
      "step": 940
    },
    {
      "epoch": 0.02539298669891173,
      "grad_norm": 2.5343501567840576,
      "learning_rate": 4.9393980661854835e-05,
      "loss": 5.5109,
      "step": 945
    },
    {
      "epoch": 0.0255273411258901,
      "grad_norm": 2.4583237171173096,
      "learning_rate": 4.938717145580826e-05,
      "loss": 5.6971,
      "step": 950
    },
    {
      "epoch": 0.025661695552868468,
      "grad_norm": 2.4051103591918945,
      "learning_rate": 4.938036224976168e-05,
      "loss": 5.6497,
      "step": 955
    },
    {
      "epoch": 0.025796049979846837,
      "grad_norm": 2.489861488342285,
      "learning_rate": 4.93735530437151e-05,
      "loss": 5.7268,
      "step": 960
    },
    {
      "epoch": 0.025930404406825206,
      "grad_norm": 2.2497615814208984,
      "learning_rate": 4.936674383766853e-05,
      "loss": 5.5657,
      "step": 965
    },
    {
      "epoch": 0.026064758833803575,
      "grad_norm": 2.4166667461395264,
      "learning_rate": 4.935993463162196e-05,
      "loss": 5.5962,
      "step": 970
    },
    {
      "epoch": 0.026199113260781944,
      "grad_norm": 2.4965009689331055,
      "learning_rate": 4.935312542557538e-05,
      "loss": 5.5071,
      "step": 975
    },
    {
      "epoch": 0.026333467687760313,
      "grad_norm": 2.6134660243988037,
      "learning_rate": 4.9346316219528805e-05,
      "loss": 5.6466,
      "step": 980
    },
    {
      "epoch": 0.026467822114738682,
      "grad_norm": 2.2586779594421387,
      "learning_rate": 4.933950701348223e-05,
      "loss": 5.5967,
      "step": 985
    },
    {
      "epoch": 0.026602176541717048,
      "grad_norm": 2.5898633003234863,
      "learning_rate": 4.9332697807435656e-05,
      "loss": 5.3913,
      "step": 990
    },
    {
      "epoch": 0.026736530968695417,
      "grad_norm": 2.391298294067383,
      "learning_rate": 4.932588860138908e-05,
      "loss": 5.5848,
      "step": 995
    },
    {
      "epoch": 0.026870885395673786,
      "grad_norm": 2.41550874710083,
      "learning_rate": 4.931907939534251e-05,
      "loss": 5.5264,
      "step": 1000
    },
    {
      "epoch": 0.027005239822652155,
      "grad_norm": 2.501523971557617,
      "learning_rate": 4.931227018929593e-05,
      "loss": 5.4967,
      "step": 1005
    },
    {
      "epoch": 0.027139594249630524,
      "grad_norm": 2.3912205696105957,
      "learning_rate": 4.930546098324936e-05,
      "loss": 5.5306,
      "step": 1010
    },
    {
      "epoch": 0.027273948676608893,
      "grad_norm": 2.2942121028900146,
      "learning_rate": 4.929865177720278e-05,
      "loss": 5.6278,
      "step": 1015
    },
    {
      "epoch": 0.027408303103587262,
      "grad_norm": 2.4391286373138428,
      "learning_rate": 4.92918425711562e-05,
      "loss": 5.4362,
      "step": 1020
    },
    {
      "epoch": 0.02754265753056563,
      "grad_norm": 2.450418472290039,
      "learning_rate": 4.9285033365109625e-05,
      "loss": 5.4614,
      "step": 1025
    },
    {
      "epoch": 0.027677011957544,
      "grad_norm": 2.4992129802703857,
      "learning_rate": 4.927822415906306e-05,
      "loss": 5.4093,
      "step": 1030
    },
    {
      "epoch": 0.02781136638452237,
      "grad_norm": 2.753678560256958,
      "learning_rate": 4.927141495301648e-05,
      "loss": 5.5318,
      "step": 1035
    },
    {
      "epoch": 0.02794572081150074,
      "grad_norm": 2.431758165359497,
      "learning_rate": 4.9264605746969906e-05,
      "loss": 5.562,
      "step": 1040
    },
    {
      "epoch": 0.028080075238479107,
      "grad_norm": 2.2790017127990723,
      "learning_rate": 4.925779654092333e-05,
      "loss": 5.5048,
      "step": 1045
    },
    {
      "epoch": 0.028214429665457477,
      "grad_norm": 2.3663244247436523,
      "learning_rate": 4.925098733487676e-05,
      "loss": 5.5602,
      "step": 1050
    },
    {
      "epoch": 0.028348784092435846,
      "grad_norm": 2.6443898677825928,
      "learning_rate": 4.924417812883018e-05,
      "loss": 5.5215,
      "step": 1055
    },
    {
      "epoch": 0.028483138519414215,
      "grad_norm": 2.1262621879577637,
      "learning_rate": 4.923736892278361e-05,
      "loss": 5.6381,
      "step": 1060
    },
    {
      "epoch": 0.028617492946392584,
      "grad_norm": 2.5652754306793213,
      "learning_rate": 4.923055971673703e-05,
      "loss": 5.4511,
      "step": 1065
    },
    {
      "epoch": 0.028751847373370953,
      "grad_norm": 2.287781000137329,
      "learning_rate": 4.922375051069046e-05,
      "loss": 5.4695,
      "step": 1070
    },
    {
      "epoch": 0.028886201800349322,
      "grad_norm": 2.26455020904541,
      "learning_rate": 4.921694130464388e-05,
      "loss": 5.586,
      "step": 1075
    },
    {
      "epoch": 0.02902055622732769,
      "grad_norm": 2.4085493087768555,
      "learning_rate": 4.9210132098597304e-05,
      "loss": 5.4816,
      "step": 1080
    },
    {
      "epoch": 0.02915491065430606,
      "grad_norm": 2.4961626529693604,
      "learning_rate": 4.9203322892550726e-05,
      "loss": 5.4164,
      "step": 1085
    },
    {
      "epoch": 0.02928926508128443,
      "grad_norm": 2.430562973022461,
      "learning_rate": 4.9196513686504155e-05,
      "loss": 5.4428,
      "step": 1090
    },
    {
      "epoch": 0.029423619508262798,
      "grad_norm": 2.3071069717407227,
      "learning_rate": 4.9189704480457584e-05,
      "loss": 5.4628,
      "step": 1095
    },
    {
      "epoch": 0.029557973935241167,
      "grad_norm": 2.728600263595581,
      "learning_rate": 4.918289527441101e-05,
      "loss": 5.4558,
      "step": 1100
    },
    {
      "epoch": 0.029692328362219536,
      "grad_norm": 2.7294223308563232,
      "learning_rate": 4.917608606836443e-05,
      "loss": 5.6166,
      "step": 1105
    },
    {
      "epoch": 0.029826682789197905,
      "grad_norm": 2.3402366638183594,
      "learning_rate": 4.916927686231785e-05,
      "loss": 5.4957,
      "step": 1110
    },
    {
      "epoch": 0.029961037216176274,
      "grad_norm": 2.441723585128784,
      "learning_rate": 4.916246765627128e-05,
      "loss": 5.5721,
      "step": 1115
    },
    {
      "epoch": 0.030095391643154643,
      "grad_norm": 2.482081651687622,
      "learning_rate": 4.915565845022471e-05,
      "loss": 5.3273,
      "step": 1120
    },
    {
      "epoch": 0.030229746070133012,
      "grad_norm": 2.434582233428955,
      "learning_rate": 4.914884924417813e-05,
      "loss": 5.493,
      "step": 1125
    },
    {
      "epoch": 0.03036410049711138,
      "grad_norm": 2.1452295780181885,
      "learning_rate": 4.9142040038131554e-05,
      "loss": 5.5232,
      "step": 1130
    },
    {
      "epoch": 0.03049845492408975,
      "grad_norm": 2.335707902908325,
      "learning_rate": 4.913523083208498e-05,
      "loss": 5.4953,
      "step": 1135
    },
    {
      "epoch": 0.030632809351068116,
      "grad_norm": 2.5285251140594482,
      "learning_rate": 4.9128421626038405e-05,
      "loss": 5.5154,
      "step": 1140
    },
    {
      "epoch": 0.030767163778046485,
      "grad_norm": 2.4380319118499756,
      "learning_rate": 4.912161241999183e-05,
      "loss": 5.3961,
      "step": 1145
    },
    {
      "epoch": 0.030901518205024854,
      "grad_norm": 2.517160177230835,
      "learning_rate": 4.9114803213945257e-05,
      "loss": 5.4653,
      "step": 1150
    },
    {
      "epoch": 0.031035872632003223,
      "grad_norm": 2.326714277267456,
      "learning_rate": 4.9107994007898686e-05,
      "loss": 5.3881,
      "step": 1155
    },
    {
      "epoch": 0.031170227058981592,
      "grad_norm": 2.2102363109588623,
      "learning_rate": 4.910118480185211e-05,
      "loss": 5.3753,
      "step": 1160
    },
    {
      "epoch": 0.031304581485959965,
      "grad_norm": 2.2875664234161377,
      "learning_rate": 4.909437559580553e-05,
      "loss": 5.3884,
      "step": 1165
    },
    {
      "epoch": 0.03143893591293833,
      "grad_norm": 2.52983021736145,
      "learning_rate": 4.908756638975895e-05,
      "loss": 5.5333,
      "step": 1170
    },
    {
      "epoch": 0.0315732903399167,
      "grad_norm": 2.466771125793457,
      "learning_rate": 4.908075718371238e-05,
      "loss": 5.4845,
      "step": 1175
    },
    {
      "epoch": 0.03170764476689507,
      "grad_norm": 2.3116300106048584,
      "learning_rate": 4.907394797766581e-05,
      "loss": 5.5213,
      "step": 1180
    },
    {
      "epoch": 0.03184199919387344,
      "grad_norm": 2.422750234603882,
      "learning_rate": 4.906713877161923e-05,
      "loss": 5.3867,
      "step": 1185
    },
    {
      "epoch": 0.031976353620851806,
      "grad_norm": 2.3109638690948486,
      "learning_rate": 4.9060329565572655e-05,
      "loss": 5.539,
      "step": 1190
    },
    {
      "epoch": 0.03211070804783018,
      "grad_norm": 2.3123369216918945,
      "learning_rate": 4.9053520359526084e-05,
      "loss": 5.4303,
      "step": 1195
    },
    {
      "epoch": 0.032245062474808545,
      "grad_norm": 2.449985980987549,
      "learning_rate": 4.9046711153479506e-05,
      "loss": 5.3668,
      "step": 1200
    },
    {
      "epoch": 0.03237941690178692,
      "grad_norm": 2.5965611934661865,
      "learning_rate": 4.903990194743293e-05,
      "loss": 5.4613,
      "step": 1205
    },
    {
      "epoch": 0.03251377132876528,
      "grad_norm": 2.5709943771362305,
      "learning_rate": 4.903309274138636e-05,
      "loss": 5.4945,
      "step": 1210
    },
    {
      "epoch": 0.032648125755743655,
      "grad_norm": 2.292269229888916,
      "learning_rate": 4.902628353533979e-05,
      "loss": 5.3906,
      "step": 1215
    },
    {
      "epoch": 0.03278248018272202,
      "grad_norm": 2.5061070919036865,
      "learning_rate": 4.901947432929321e-05,
      "loss": 5.3073,
      "step": 1220
    },
    {
      "epoch": 0.032916834609700386,
      "grad_norm": 2.5885732173919678,
      "learning_rate": 4.901266512324663e-05,
      "loss": 5.3256,
      "step": 1225
    },
    {
      "epoch": 0.03305118903667876,
      "grad_norm": 2.6703076362609863,
      "learning_rate": 4.9005855917200054e-05,
      "loss": 5.4299,
      "step": 1230
    },
    {
      "epoch": 0.033185543463657124,
      "grad_norm": 2.6485342979431152,
      "learning_rate": 4.899904671115348e-05,
      "loss": 5.4164,
      "step": 1235
    },
    {
      "epoch": 0.0333198978906355,
      "grad_norm": 2.5048696994781494,
      "learning_rate": 4.899223750510691e-05,
      "loss": 5.521,
      "step": 1240
    },
    {
      "epoch": 0.03345425231761386,
      "grad_norm": 2.3647947311401367,
      "learning_rate": 4.8985428299060334e-05,
      "loss": 5.4279,
      "step": 1245
    },
    {
      "epoch": 0.033588606744592235,
      "grad_norm": 2.5917186737060547,
      "learning_rate": 4.8978619093013756e-05,
      "loss": 5.2821,
      "step": 1250
    },
    {
      "epoch": 0.0337229611715706,
      "grad_norm": 2.7712130546569824,
      "learning_rate": 4.897180988696718e-05,
      "loss": 5.3969,
      "step": 1255
    },
    {
      "epoch": 0.03385731559854897,
      "grad_norm": 2.368302345275879,
      "learning_rate": 4.896500068092061e-05,
      "loss": 5.3883,
      "step": 1260
    },
    {
      "epoch": 0.03399167002552734,
      "grad_norm": 2.539823055267334,
      "learning_rate": 4.895819147487403e-05,
      "loss": 5.5243,
      "step": 1265
    },
    {
      "epoch": 0.03412602445250571,
      "grad_norm": 2.261368989944458,
      "learning_rate": 4.895138226882746e-05,
      "loss": 5.4575,
      "step": 1270
    },
    {
      "epoch": 0.03426037887948408,
      "grad_norm": 2.6786978244781494,
      "learning_rate": 4.894457306278088e-05,
      "loss": 5.4849,
      "step": 1275
    },
    {
      "epoch": 0.03439473330646245,
      "grad_norm": 2.6710543632507324,
      "learning_rate": 4.893776385673431e-05,
      "loss": 5.4024,
      "step": 1280
    },
    {
      "epoch": 0.034529087733440815,
      "grad_norm": 2.3116440773010254,
      "learning_rate": 4.893095465068773e-05,
      "loss": 5.4615,
      "step": 1285
    },
    {
      "epoch": 0.03466344216041919,
      "grad_norm": 2.3754544258117676,
      "learning_rate": 4.8924145444641155e-05,
      "loss": 5.4348,
      "step": 1290
    },
    {
      "epoch": 0.03479779658739755,
      "grad_norm": 2.5682246685028076,
      "learning_rate": 4.891733623859458e-05,
      "loss": 5.4357,
      "step": 1295
    },
    {
      "epoch": 0.034932151014375926,
      "grad_norm": 2.48640513420105,
      "learning_rate": 4.8910527032548006e-05,
      "loss": 5.4485,
      "step": 1300
    },
    {
      "epoch": 0.03506650544135429,
      "grad_norm": 2.571028470993042,
      "learning_rate": 4.8903717826501435e-05,
      "loss": 5.3691,
      "step": 1305
    },
    {
      "epoch": 0.035200859868332664,
      "grad_norm": 2.1914620399475098,
      "learning_rate": 4.889690862045486e-05,
      "loss": 5.4309,
      "step": 1310
    },
    {
      "epoch": 0.03533521429531103,
      "grad_norm": 2.3349716663360596,
      "learning_rate": 4.889009941440828e-05,
      "loss": 5.2666,
      "step": 1315
    },
    {
      "epoch": 0.0354695687222894,
      "grad_norm": 2.440905809402466,
      "learning_rate": 4.888329020836171e-05,
      "loss": 5.4088,
      "step": 1320
    },
    {
      "epoch": 0.03560392314926777,
      "grad_norm": 2.3953843116760254,
      "learning_rate": 4.887648100231513e-05,
      "loss": 5.3627,
      "step": 1325
    },
    {
      "epoch": 0.03573827757624614,
      "grad_norm": 2.581817865371704,
      "learning_rate": 4.886967179626856e-05,
      "loss": 5.4535,
      "step": 1330
    },
    {
      "epoch": 0.035872632003224506,
      "grad_norm": 2.3343987464904785,
      "learning_rate": 4.886286259022198e-05,
      "loss": 5.3891,
      "step": 1335
    },
    {
      "epoch": 0.03600698643020288,
      "grad_norm": 2.195507287979126,
      "learning_rate": 4.885605338417541e-05,
      "loss": 5.3466,
      "step": 1340
    },
    {
      "epoch": 0.036141340857181244,
      "grad_norm": 2.514629602432251,
      "learning_rate": 4.8849244178128833e-05,
      "loss": 5.4418,
      "step": 1345
    },
    {
      "epoch": 0.036275695284159616,
      "grad_norm": 2.5501441955566406,
      "learning_rate": 4.8842434972082256e-05,
      "loss": 5.37,
      "step": 1350
    },
    {
      "epoch": 0.03641004971113798,
      "grad_norm": 2.294858455657959,
      "learning_rate": 4.883562576603568e-05,
      "loss": 5.4486,
      "step": 1355
    },
    {
      "epoch": 0.036544404138116354,
      "grad_norm": 2.5770301818847656,
      "learning_rate": 4.882881655998911e-05,
      "loss": 5.3974,
      "step": 1360
    },
    {
      "epoch": 0.03667875856509472,
      "grad_norm": 2.296680212020874,
      "learning_rate": 4.8822007353942536e-05,
      "loss": 5.3902,
      "step": 1365
    },
    {
      "epoch": 0.036813112992073085,
      "grad_norm": 2.0749173164367676,
      "learning_rate": 4.881519814789596e-05,
      "loss": 5.4706,
      "step": 1370
    },
    {
      "epoch": 0.03694746741905146,
      "grad_norm": 2.461259126663208,
      "learning_rate": 4.880838894184938e-05,
      "loss": 5.3508,
      "step": 1375
    },
    {
      "epoch": 0.037081821846029824,
      "grad_norm": 2.320645332336426,
      "learning_rate": 4.880157973580281e-05,
      "loss": 5.3702,
      "step": 1380
    },
    {
      "epoch": 0.037216176273008196,
      "grad_norm": 2.358922243118286,
      "learning_rate": 4.879477052975623e-05,
      "loss": 5.3265,
      "step": 1385
    },
    {
      "epoch": 0.03735053069998656,
      "grad_norm": 2.890631914138794,
      "learning_rate": 4.878796132370966e-05,
      "loss": 5.375,
      "step": 1390
    },
    {
      "epoch": 0.037484885126964934,
      "grad_norm": 2.5326790809631348,
      "learning_rate": 4.878115211766308e-05,
      "loss": 5.4022,
      "step": 1395
    },
    {
      "epoch": 0.0376192395539433,
      "grad_norm": 2.805542230606079,
      "learning_rate": 4.8774342911616506e-05,
      "loss": 5.3828,
      "step": 1400
    },
    {
      "epoch": 0.03775359398092167,
      "grad_norm": 2.2423665523529053,
      "learning_rate": 4.8767533705569935e-05,
      "loss": 5.3923,
      "step": 1405
    },
    {
      "epoch": 0.03788794840790004,
      "grad_norm": 2.5512473583221436,
      "learning_rate": 4.876072449952336e-05,
      "loss": 5.4147,
      "step": 1410
    },
    {
      "epoch": 0.03802230283487841,
      "grad_norm": 2.252861499786377,
      "learning_rate": 4.875391529347678e-05,
      "loss": 5.418,
      "step": 1415
    },
    {
      "epoch": 0.038156657261856776,
      "grad_norm": 2.629993200302124,
      "learning_rate": 4.874710608743021e-05,
      "loss": 5.3635,
      "step": 1420
    },
    {
      "epoch": 0.03829101168883515,
      "grad_norm": 2.439270257949829,
      "learning_rate": 4.874029688138364e-05,
      "loss": 5.3613,
      "step": 1425
    },
    {
      "epoch": 0.038425366115813514,
      "grad_norm": 2.4896345138549805,
      "learning_rate": 4.873348767533706e-05,
      "loss": 5.352,
      "step": 1430
    },
    {
      "epoch": 0.03855972054279189,
      "grad_norm": 2.5188028812408447,
      "learning_rate": 4.872667846929048e-05,
      "loss": 5.4657,
      "step": 1435
    },
    {
      "epoch": 0.03869407496977025,
      "grad_norm": 2.387194871902466,
      "learning_rate": 4.8719869263243904e-05,
      "loss": 5.4058,
      "step": 1440
    },
    {
      "epoch": 0.038828429396748625,
      "grad_norm": 2.380246162414551,
      "learning_rate": 4.871306005719733e-05,
      "loss": 5.3044,
      "step": 1445
    },
    {
      "epoch": 0.03896278382372699,
      "grad_norm": 2.4011290073394775,
      "learning_rate": 4.8706250851150755e-05,
      "loss": 5.3825,
      "step": 1450
    },
    {
      "epoch": 0.03909713825070536,
      "grad_norm": 2.509749174118042,
      "learning_rate": 4.8699441645104184e-05,
      "loss": 5.4325,
      "step": 1455
    },
    {
      "epoch": 0.03923149267768373,
      "grad_norm": 2.646247148513794,
      "learning_rate": 4.869263243905761e-05,
      "loss": 5.4015,
      "step": 1460
    },
    {
      "epoch": 0.0393658471046621,
      "grad_norm": 2.2970354557037354,
      "learning_rate": 4.8685823233011036e-05,
      "loss": 5.4615,
      "step": 1465
    },
    {
      "epoch": 0.039500201531640466,
      "grad_norm": 2.2596287727355957,
      "learning_rate": 4.867901402696446e-05,
      "loss": 5.3141,
      "step": 1470
    },
    {
      "epoch": 0.03963455595861884,
      "grad_norm": 2.2906055450439453,
      "learning_rate": 4.867220482091788e-05,
      "loss": 5.3758,
      "step": 1475
    },
    {
      "epoch": 0.039768910385597205,
      "grad_norm": 2.752013683319092,
      "learning_rate": 4.866539561487131e-05,
      "loss": 5.2831,
      "step": 1480
    },
    {
      "epoch": 0.03990326481257558,
      "grad_norm": 2.4344067573547363,
      "learning_rate": 4.865858640882474e-05,
      "loss": 5.3224,
      "step": 1485
    },
    {
      "epoch": 0.04003761923955394,
      "grad_norm": 2.2967071533203125,
      "learning_rate": 4.865177720277816e-05,
      "loss": 5.3147,
      "step": 1490
    },
    {
      "epoch": 0.040171973666532315,
      "grad_norm": 2.4886233806610107,
      "learning_rate": 4.864496799673158e-05,
      "loss": 5.3791,
      "step": 1495
    },
    {
      "epoch": 0.04030632809351068,
      "grad_norm": 2.310999631881714,
      "learning_rate": 4.8638158790685005e-05,
      "loss": 5.3549,
      "step": 1500
    },
    {
      "epoch": 0.04044068252048905,
      "grad_norm": 2.321484088897705,
      "learning_rate": 4.8631349584638434e-05,
      "loss": 5.323,
      "step": 1505
    },
    {
      "epoch": 0.04057503694746742,
      "grad_norm": 2.517014503479004,
      "learning_rate": 4.8624540378591856e-05,
      "loss": 5.255,
      "step": 1510
    },
    {
      "epoch": 0.04070939137444579,
      "grad_norm": 2.575798511505127,
      "learning_rate": 4.8617731172545285e-05,
      "loss": 5.379,
      "step": 1515
    },
    {
      "epoch": 0.04084374580142416,
      "grad_norm": 2.2519052028656006,
      "learning_rate": 4.861092196649871e-05,
      "loss": 5.3484,
      "step": 1520
    },
    {
      "epoch": 0.04097810022840252,
      "grad_norm": 2.4548873901367188,
      "learning_rate": 4.860411276045214e-05,
      "loss": 5.3235,
      "step": 1525
    },
    {
      "epoch": 0.041112454655380895,
      "grad_norm": 2.3341224193573,
      "learning_rate": 4.859730355440556e-05,
      "loss": 5.1738,
      "step": 1530
    },
    {
      "epoch": 0.04124680908235926,
      "grad_norm": 2.8830623626708984,
      "learning_rate": 4.859049434835898e-05,
      "loss": 5.3159,
      "step": 1535
    },
    {
      "epoch": 0.04138116350933763,
      "grad_norm": 2.2873809337615967,
      "learning_rate": 4.858368514231241e-05,
      "loss": 5.4937,
      "step": 1540
    },
    {
      "epoch": 0.041515517936316,
      "grad_norm": 2.40313458442688,
      "learning_rate": 4.857687593626583e-05,
      "loss": 5.2487,
      "step": 1545
    },
    {
      "epoch": 0.04164987236329437,
      "grad_norm": 2.66757869720459,
      "learning_rate": 4.857006673021926e-05,
      "loss": 5.2708,
      "step": 1550
    },
    {
      "epoch": 0.04178422679027274,
      "grad_norm": 2.556011438369751,
      "learning_rate": 4.8563257524172684e-05,
      "loss": 5.2721,
      "step": 1555
    },
    {
      "epoch": 0.04191858121725111,
      "grad_norm": 2.6276113986968994,
      "learning_rate": 4.8556448318126106e-05,
      "loss": 5.3838,
      "step": 1560
    },
    {
      "epoch": 0.042052935644229475,
      "grad_norm": 2.3283538818359375,
      "learning_rate": 4.854963911207953e-05,
      "loss": 5.4259,
      "step": 1565
    },
    {
      "epoch": 0.04218729007120785,
      "grad_norm": 2.340780258178711,
      "learning_rate": 4.854282990603296e-05,
      "loss": 5.3155,
      "step": 1570
    },
    {
      "epoch": 0.04232164449818621,
      "grad_norm": 2.322408676147461,
      "learning_rate": 4.8536020699986387e-05,
      "loss": 5.242,
      "step": 1575
    },
    {
      "epoch": 0.042455998925164586,
      "grad_norm": 2.438852310180664,
      "learning_rate": 4.852921149393981e-05,
      "loss": 5.1866,
      "step": 1580
    },
    {
      "epoch": 0.04259035335214295,
      "grad_norm": 2.482048273086548,
      "learning_rate": 4.852240228789323e-05,
      "loss": 5.2217,
      "step": 1585
    },
    {
      "epoch": 0.042724707779121324,
      "grad_norm": 2.81852126121521,
      "learning_rate": 4.851559308184666e-05,
      "loss": 5.3781,
      "step": 1590
    },
    {
      "epoch": 0.04285906220609969,
      "grad_norm": 2.2842531204223633,
      "learning_rate": 4.850878387580008e-05,
      "loss": 5.2835,
      "step": 1595
    },
    {
      "epoch": 0.04299341663307806,
      "grad_norm": 2.5050878524780273,
      "learning_rate": 4.850197466975351e-05,
      "loss": 5.3429,
      "step": 1600
    },
    {
      "epoch": 0.04312777106005643,
      "grad_norm": 2.304842472076416,
      "learning_rate": 4.8495165463706934e-05,
      "loss": 5.2884,
      "step": 1605
    },
    {
      "epoch": 0.0432621254870348,
      "grad_norm": 2.4210867881774902,
      "learning_rate": 4.848835625766036e-05,
      "loss": 5.3353,
      "step": 1610
    },
    {
      "epoch": 0.043396479914013165,
      "grad_norm": 2.386096477508545,
      "learning_rate": 4.8481547051613785e-05,
      "loss": 5.4406,
      "step": 1615
    },
    {
      "epoch": 0.04353083434099154,
      "grad_norm": 2.1601107120513916,
      "learning_rate": 4.847473784556721e-05,
      "loss": 5.207,
      "step": 1620
    },
    {
      "epoch": 0.043665188767969904,
      "grad_norm": 2.3382086753845215,
      "learning_rate": 4.846792863952063e-05,
      "loss": 5.2645,
      "step": 1625
    },
    {
      "epoch": 0.043799543194948276,
      "grad_norm": 2.4124510288238525,
      "learning_rate": 4.846111943347406e-05,
      "loss": 5.2726,
      "step": 1630
    },
    {
      "epoch": 0.04393389762192664,
      "grad_norm": 2.1609771251678467,
      "learning_rate": 4.845431022742749e-05,
      "loss": 5.4364,
      "step": 1635
    },
    {
      "epoch": 0.044068252048905014,
      "grad_norm": 2.2471845149993896,
      "learning_rate": 4.844750102138091e-05,
      "loss": 5.1903,
      "step": 1640
    },
    {
      "epoch": 0.04420260647588338,
      "grad_norm": 2.5327093601226807,
      "learning_rate": 4.844069181533433e-05,
      "loss": 5.3189,
      "step": 1645
    },
    {
      "epoch": 0.04433696090286175,
      "grad_norm": 2.3454365730285645,
      "learning_rate": 4.843388260928776e-05,
      "loss": 5.2266,
      "step": 1650
    },
    {
      "epoch": 0.04447131532984012,
      "grad_norm": 2.4912145137786865,
      "learning_rate": 4.8427073403241184e-05,
      "loss": 5.2098,
      "step": 1655
    },
    {
      "epoch": 0.04460566975681849,
      "grad_norm": 2.2063989639282227,
      "learning_rate": 4.8420264197194606e-05,
      "loss": 5.293,
      "step": 1660
    },
    {
      "epoch": 0.044740024183796856,
      "grad_norm": 2.6512610912323,
      "learning_rate": 4.8413454991148035e-05,
      "loss": 5.4295,
      "step": 1665
    },
    {
      "epoch": 0.04487437861077523,
      "grad_norm": 2.7759170532226562,
      "learning_rate": 4.8406645785101464e-05,
      "loss": 5.3546,
      "step": 1670
    },
    {
      "epoch": 0.045008733037753594,
      "grad_norm": 2.256654739379883,
      "learning_rate": 4.8399836579054886e-05,
      "loss": 5.3061,
      "step": 1675
    },
    {
      "epoch": 0.04514308746473196,
      "grad_norm": 2.45420503616333,
      "learning_rate": 4.839302737300831e-05,
      "loss": 5.3713,
      "step": 1680
    },
    {
      "epoch": 0.04527744189171033,
      "grad_norm": 2.358295202255249,
      "learning_rate": 4.838621816696173e-05,
      "loss": 5.2431,
      "step": 1685
    },
    {
      "epoch": 0.0454117963186887,
      "grad_norm": 2.406562328338623,
      "learning_rate": 4.837940896091516e-05,
      "loss": 5.0867,
      "step": 1690
    },
    {
      "epoch": 0.04554615074566707,
      "grad_norm": 2.2944185733795166,
      "learning_rate": 4.837259975486859e-05,
      "loss": 5.2294,
      "step": 1695
    },
    {
      "epoch": 0.045680505172645436,
      "grad_norm": 2.407620906829834,
      "learning_rate": 4.836579054882201e-05,
      "loss": 5.1957,
      "step": 1700
    },
    {
      "epoch": 0.04581485959962381,
      "grad_norm": 2.550377130508423,
      "learning_rate": 4.835898134277543e-05,
      "loss": 5.2069,
      "step": 1705
    },
    {
      "epoch": 0.045949214026602174,
      "grad_norm": 2.2927417755126953,
      "learning_rate": 4.8352172136728856e-05,
      "loss": 5.2309,
      "step": 1710
    },
    {
      "epoch": 0.046083568453580546,
      "grad_norm": 2.30096697807312,
      "learning_rate": 4.8345362930682285e-05,
      "loss": 5.2355,
      "step": 1715
    },
    {
      "epoch": 0.04621792288055891,
      "grad_norm": 2.2296783924102783,
      "learning_rate": 4.833855372463571e-05,
      "loss": 5.3139,
      "step": 1720
    },
    {
      "epoch": 0.046352277307537285,
      "grad_norm": 2.24027681350708,
      "learning_rate": 4.8331744518589136e-05,
      "loss": 5.0176,
      "step": 1725
    },
    {
      "epoch": 0.04648663173451565,
      "grad_norm": 2.72025465965271,
      "learning_rate": 4.832493531254256e-05,
      "loss": 5.3124,
      "step": 1730
    },
    {
      "epoch": 0.04662098616149402,
      "grad_norm": 2.337449073791504,
      "learning_rate": 4.831812610649599e-05,
      "loss": 5.1725,
      "step": 1735
    },
    {
      "epoch": 0.04675534058847239,
      "grad_norm": 2.5324695110321045,
      "learning_rate": 4.831131690044941e-05,
      "loss": 5.1719,
      "step": 1740
    },
    {
      "epoch": 0.04688969501545076,
      "grad_norm": 2.5533106327056885,
      "learning_rate": 4.830450769440283e-05,
      "loss": 5.2212,
      "step": 1745
    },
    {
      "epoch": 0.047024049442429126,
      "grad_norm": 2.3930368423461914,
      "learning_rate": 4.829769848835626e-05,
      "loss": 5.2615,
      "step": 1750
    },
    {
      "epoch": 0.0471584038694075,
      "grad_norm": 2.4374313354492188,
      "learning_rate": 4.829088928230969e-05,
      "loss": 5.4474,
      "step": 1755
    },
    {
      "epoch": 0.047292758296385864,
      "grad_norm": 2.4467618465423584,
      "learning_rate": 4.828408007626311e-05,
      "loss": 5.1729,
      "step": 1760
    },
    {
      "epoch": 0.04742711272336424,
      "grad_norm": 2.3877153396606445,
      "learning_rate": 4.8277270870216534e-05,
      "loss": 5.2172,
      "step": 1765
    },
    {
      "epoch": 0.0475614671503426,
      "grad_norm": 2.432770252227783,
      "learning_rate": 4.827046166416996e-05,
      "loss": 5.1381,
      "step": 1770
    },
    {
      "epoch": 0.047695821577320975,
      "grad_norm": 2.4823038578033447,
      "learning_rate": 4.8263652458123386e-05,
      "loss": 5.2353,
      "step": 1775
    },
    {
      "epoch": 0.04783017600429934,
      "grad_norm": 2.2864232063293457,
      "learning_rate": 4.825684325207681e-05,
      "loss": 5.2289,
      "step": 1780
    },
    {
      "epoch": 0.04796453043127771,
      "grad_norm": 2.4113411903381348,
      "learning_rate": 4.825003404603024e-05,
      "loss": 5.1762,
      "step": 1785
    },
    {
      "epoch": 0.04809888485825608,
      "grad_norm": 2.7030029296875,
      "learning_rate": 4.824322483998366e-05,
      "loss": 5.295,
      "step": 1790
    },
    {
      "epoch": 0.04823323928523445,
      "grad_norm": 2.3226568698883057,
      "learning_rate": 4.823641563393709e-05,
      "loss": 5.2723,
      "step": 1795
    },
    {
      "epoch": 0.04836759371221282,
      "grad_norm": 2.478559970855713,
      "learning_rate": 4.822960642789051e-05,
      "loss": 5.278,
      "step": 1800
    },
    {
      "epoch": 0.04850194813919119,
      "grad_norm": 2.6240153312683105,
      "learning_rate": 4.822279722184393e-05,
      "loss": 5.2883,
      "step": 1805
    },
    {
      "epoch": 0.048636302566169555,
      "grad_norm": 2.363565444946289,
      "learning_rate": 4.821598801579736e-05,
      "loss": 5.4129,
      "step": 1810
    },
    {
      "epoch": 0.04877065699314793,
      "grad_norm": 2.653879165649414,
      "learning_rate": 4.820917880975079e-05,
      "loss": 5.3435,
      "step": 1815
    },
    {
      "epoch": 0.04890501142012629,
      "grad_norm": 2.356649875640869,
      "learning_rate": 4.820236960370421e-05,
      "loss": 5.166,
      "step": 1820
    },
    {
      "epoch": 0.04903936584710466,
      "grad_norm": 2.430574893951416,
      "learning_rate": 4.8195560397657636e-05,
      "loss": 5.3627,
      "step": 1825
    },
    {
      "epoch": 0.04917372027408303,
      "grad_norm": 2.2518692016601562,
      "learning_rate": 4.818875119161106e-05,
      "loss": 5.3862,
      "step": 1830
    },
    {
      "epoch": 0.0493080747010614,
      "grad_norm": 2.4937832355499268,
      "learning_rate": 4.818194198556449e-05,
      "loss": 5.2607,
      "step": 1835
    },
    {
      "epoch": 0.04944242912803977,
      "grad_norm": 2.4203927516937256,
      "learning_rate": 4.817513277951791e-05,
      "loss": 5.1208,
      "step": 1840
    },
    {
      "epoch": 0.049576783555018135,
      "grad_norm": 2.1159417629241943,
      "learning_rate": 4.816832357347134e-05,
      "loss": 5.3285,
      "step": 1845
    },
    {
      "epoch": 0.04971113798199651,
      "grad_norm": 2.1734254360198975,
      "learning_rate": 4.816151436742476e-05,
      "loss": 5.2623,
      "step": 1850
    },
    {
      "epoch": 0.04984549240897487,
      "grad_norm": 2.313054323196411,
      "learning_rate": 4.815470516137818e-05,
      "loss": 5.3001,
      "step": 1855
    },
    {
      "epoch": 0.049979846835953246,
      "grad_norm": 2.3189878463745117,
      "learning_rate": 4.814789595533161e-05,
      "loss": 5.3274,
      "step": 1860
    },
    {
      "epoch": 0.05011420126293161,
      "grad_norm": 2.7325174808502197,
      "learning_rate": 4.8141086749285034e-05,
      "loss": 5.1856,
      "step": 1865
    },
    {
      "epoch": 0.050248555689909984,
      "grad_norm": 2.5446767807006836,
      "learning_rate": 4.8134277543238456e-05,
      "loss": 5.3078,
      "step": 1870
    },
    {
      "epoch": 0.05038291011688835,
      "grad_norm": 2.719104766845703,
      "learning_rate": 4.8127468337191885e-05,
      "loss": 5.3787,
      "step": 1875
    },
    {
      "epoch": 0.05051726454386672,
      "grad_norm": 2.8305141925811768,
      "learning_rate": 4.8120659131145314e-05,
      "loss": 5.2352,
      "step": 1880
    },
    {
      "epoch": 0.05065161897084509,
      "grad_norm": 2.584324359893799,
      "learning_rate": 4.811384992509874e-05,
      "loss": 5.2162,
      "step": 1885
    },
    {
      "epoch": 0.05078597339782346,
      "grad_norm": 2.4216935634613037,
      "learning_rate": 4.810704071905216e-05,
      "loss": 5.3332,
      "step": 1890
    },
    {
      "epoch": 0.050920327824801825,
      "grad_norm": 2.4615275859832764,
      "learning_rate": 4.810023151300558e-05,
      "loss": 5.2437,
      "step": 1895
    },
    {
      "epoch": 0.0510546822517802,
      "grad_norm": 2.5613584518432617,
      "learning_rate": 4.809342230695901e-05,
      "loss": 5.2118,
      "step": 1900
    },
    {
      "epoch": 0.051189036678758564,
      "grad_norm": 2.37742018699646,
      "learning_rate": 4.808661310091244e-05,
      "loss": 5.0876,
      "step": 1905
    },
    {
      "epoch": 0.051323391105736936,
      "grad_norm": 2.275043249130249,
      "learning_rate": 4.807980389486586e-05,
      "loss": 5.1752,
      "step": 1910
    },
    {
      "epoch": 0.0514577455327153,
      "grad_norm": 2.3270816802978516,
      "learning_rate": 4.8072994688819284e-05,
      "loss": 5.2395,
      "step": 1915
    },
    {
      "epoch": 0.051592099959693674,
      "grad_norm": 2.2508606910705566,
      "learning_rate": 4.806618548277271e-05,
      "loss": 5.3465,
      "step": 1920
    },
    {
      "epoch": 0.05172645438667204,
      "grad_norm": 2.3401224613189697,
      "learning_rate": 4.8059376276726135e-05,
      "loss": 5.1628,
      "step": 1925
    },
    {
      "epoch": 0.05186080881365041,
      "grad_norm": 2.3724825382232666,
      "learning_rate": 4.805256707067956e-05,
      "loss": 5.2503,
      "step": 1930
    },
    {
      "epoch": 0.05199516324062878,
      "grad_norm": 2.3743178844451904,
      "learning_rate": 4.8045757864632986e-05,
      "loss": 5.2006,
      "step": 1935
    },
    {
      "epoch": 0.05212951766760715,
      "grad_norm": 2.4351425170898438,
      "learning_rate": 4.8038948658586416e-05,
      "loss": 5.2431,
      "step": 1940
    },
    {
      "epoch": 0.052263872094585516,
      "grad_norm": 2.3778090476989746,
      "learning_rate": 4.803213945253984e-05,
      "loss": 5.2104,
      "step": 1945
    },
    {
      "epoch": 0.05239822652156389,
      "grad_norm": 2.2631049156188965,
      "learning_rate": 4.802533024649326e-05,
      "loss": 5.1693,
      "step": 1950
    },
    {
      "epoch": 0.052532580948542254,
      "grad_norm": 2.2894062995910645,
      "learning_rate": 4.801852104044668e-05,
      "loss": 5.2593,
      "step": 1955
    },
    {
      "epoch": 0.05266693537552063,
      "grad_norm": 2.3397817611694336,
      "learning_rate": 4.801171183440011e-05,
      "loss": 5.1952,
      "step": 1960
    },
    {
      "epoch": 0.05280128980249899,
      "grad_norm": 2.0636684894561768,
      "learning_rate": 4.800490262835354e-05,
      "loss": 5.2578,
      "step": 1965
    },
    {
      "epoch": 0.052935644229477365,
      "grad_norm": 2.359027862548828,
      "learning_rate": 4.799809342230696e-05,
      "loss": 5.1703,
      "step": 1970
    },
    {
      "epoch": 0.05306999865645573,
      "grad_norm": 2.3653738498687744,
      "learning_rate": 4.7991284216260385e-05,
      "loss": 5.2127,
      "step": 1975
    },
    {
      "epoch": 0.053204353083434096,
      "grad_norm": 2.5569891929626465,
      "learning_rate": 4.7984475010213814e-05,
      "loss": 5.0935,
      "step": 1980
    },
    {
      "epoch": 0.05333870751041247,
      "grad_norm": 2.4698874950408936,
      "learning_rate": 4.7977665804167236e-05,
      "loss": 5.1307,
      "step": 1985
    },
    {
      "epoch": 0.053473061937390834,
      "grad_norm": 2.2079992294311523,
      "learning_rate": 4.797085659812066e-05,
      "loss": 5.1872,
      "step": 1990
    },
    {
      "epoch": 0.053607416364369206,
      "grad_norm": 2.270055055618286,
      "learning_rate": 4.796404739207409e-05,
      "loss": 5.2207,
      "step": 1995
    },
    {
      "epoch": 0.05374177079134757,
      "grad_norm": 2.425661563873291,
      "learning_rate": 4.795723818602751e-05,
      "loss": 5.2196,
      "step": 2000
    },
    {
      "epoch": 0.053876125218325945,
      "grad_norm": 2.0815727710723877,
      "learning_rate": 4.795042897998094e-05,
      "loss": 5.1668,
      "step": 2005
    },
    {
      "epoch": 0.05401047964530431,
      "grad_norm": 2.305485725402832,
      "learning_rate": 4.794361977393436e-05,
      "loss": 5.3338,
      "step": 2010
    },
    {
      "epoch": 0.05414483407228268,
      "grad_norm": 2.193415880203247,
      "learning_rate": 4.7936810567887783e-05,
      "loss": 5.215,
      "step": 2015
    },
    {
      "epoch": 0.05427918849926105,
      "grad_norm": 2.2659268379211426,
      "learning_rate": 4.793000136184121e-05,
      "loss": 5.1185,
      "step": 2020
    },
    {
      "epoch": 0.05441354292623942,
      "grad_norm": 2.536661386489868,
      "learning_rate": 4.792319215579464e-05,
      "loss": 5.1193,
      "step": 2025
    },
    {
      "epoch": 0.054547897353217786,
      "grad_norm": 2.5080933570861816,
      "learning_rate": 4.7916382949748064e-05,
      "loss": 5.1321,
      "step": 2030
    },
    {
      "epoch": 0.05468225178019616,
      "grad_norm": 2.3799896240234375,
      "learning_rate": 4.7909573743701486e-05,
      "loss": 5.1633,
      "step": 2035
    },
    {
      "epoch": 0.054816606207174524,
      "grad_norm": 2.2105393409729004,
      "learning_rate": 4.790276453765491e-05,
      "loss": 5.1659,
      "step": 2040
    },
    {
      "epoch": 0.0549509606341529,
      "grad_norm": 2.211395025253296,
      "learning_rate": 4.789595533160834e-05,
      "loss": 5.0961,
      "step": 2045
    },
    {
      "epoch": 0.05508531506113126,
      "grad_norm": 2.490356206893921,
      "learning_rate": 4.788914612556176e-05,
      "loss": 5.0887,
      "step": 2050
    },
    {
      "epoch": 0.055219669488109635,
      "grad_norm": 2.2620551586151123,
      "learning_rate": 4.788233691951519e-05,
      "loss": 5.2394,
      "step": 2055
    },
    {
      "epoch": 0.055354023915088,
      "grad_norm": 2.4422574043273926,
      "learning_rate": 4.787552771346861e-05,
      "loss": 5.0351,
      "step": 2060
    },
    {
      "epoch": 0.05548837834206637,
      "grad_norm": 2.757983922958374,
      "learning_rate": 4.786871850742204e-05,
      "loss": 5.2689,
      "step": 2065
    },
    {
      "epoch": 0.05562273276904474,
      "grad_norm": 2.3653347492218018,
      "learning_rate": 4.786190930137546e-05,
      "loss": 5.2375,
      "step": 2070
    },
    {
      "epoch": 0.05575708719602311,
      "grad_norm": 2.1607768535614014,
      "learning_rate": 4.7855100095328885e-05,
      "loss": 5.2352,
      "step": 2075
    },
    {
      "epoch": 0.05589144162300148,
      "grad_norm": 2.5361104011535645,
      "learning_rate": 4.784829088928231e-05,
      "loss": 5.1385,
      "step": 2080
    },
    {
      "epoch": 0.05602579604997985,
      "grad_norm": 2.4672648906707764,
      "learning_rate": 4.784148168323574e-05,
      "loss": 5.0615,
      "step": 2085
    },
    {
      "epoch": 0.056160150476958215,
      "grad_norm": 2.412135601043701,
      "learning_rate": 4.7834672477189165e-05,
      "loss": 5.1649,
      "step": 2090
    },
    {
      "epoch": 0.05629450490393659,
      "grad_norm": 2.3622710704803467,
      "learning_rate": 4.782786327114259e-05,
      "loss": 5.049,
      "step": 2095
    },
    {
      "epoch": 0.05642885933091495,
      "grad_norm": 2.3928167819976807,
      "learning_rate": 4.782105406509601e-05,
      "loss": 5.0205,
      "step": 2100
    },
    {
      "epoch": 0.056563213757893326,
      "grad_norm": 2.446167230606079,
      "learning_rate": 4.781424485904944e-05,
      "loss": 5.1282,
      "step": 2105
    },
    {
      "epoch": 0.05669756818487169,
      "grad_norm": 2.2952096462249756,
      "learning_rate": 4.780743565300286e-05,
      "loss": 5.0892,
      "step": 2110
    },
    {
      "epoch": 0.056831922611850064,
      "grad_norm": 2.3746469020843506,
      "learning_rate": 4.780062644695629e-05,
      "loss": 5.2202,
      "step": 2115
    },
    {
      "epoch": 0.05696627703882843,
      "grad_norm": 2.298170328140259,
      "learning_rate": 4.779381724090971e-05,
      "loss": 5.2387,
      "step": 2120
    },
    {
      "epoch": 0.0571006314658068,
      "grad_norm": 2.349518299102783,
      "learning_rate": 4.778700803486314e-05,
      "loss": 5.2094,
      "step": 2125
    },
    {
      "epoch": 0.05723498589278517,
      "grad_norm": 2.262782335281372,
      "learning_rate": 4.7780198828816563e-05,
      "loss": 5.0571,
      "step": 2130
    },
    {
      "epoch": 0.05736934031976353,
      "grad_norm": 2.4524972438812256,
      "learning_rate": 4.7773389622769986e-05,
      "loss": 5.1114,
      "step": 2135
    },
    {
      "epoch": 0.057503694746741905,
      "grad_norm": 2.3404948711395264,
      "learning_rate": 4.776658041672341e-05,
      "loss": 5.1455,
      "step": 2140
    },
    {
      "epoch": 0.05763804917372027,
      "grad_norm": 2.2864677906036377,
      "learning_rate": 4.775977121067684e-05,
      "loss": 5.1923,
      "step": 2145
    },
    {
      "epoch": 0.057772403600698644,
      "grad_norm": 2.7468621730804443,
      "learning_rate": 4.7752962004630266e-05,
      "loss": 5.1457,
      "step": 2150
    },
    {
      "epoch": 0.05790675802767701,
      "grad_norm": 2.581831455230713,
      "learning_rate": 4.774615279858369e-05,
      "loss": 5.1021,
      "step": 2155
    },
    {
      "epoch": 0.05804111245465538,
      "grad_norm": 2.2192766666412354,
      "learning_rate": 4.773934359253711e-05,
      "loss": 5.1822,
      "step": 2160
    },
    {
      "epoch": 0.05817546688163375,
      "grad_norm": 2.285222053527832,
      "learning_rate": 4.773253438649053e-05,
      "loss": 5.1948,
      "step": 2165
    },
    {
      "epoch": 0.05830982130861212,
      "grad_norm": 2.528031587600708,
      "learning_rate": 4.772572518044396e-05,
      "loss": 5.1731,
      "step": 2170
    },
    {
      "epoch": 0.058444175735590485,
      "grad_norm": 2.3036327362060547,
      "learning_rate": 4.771891597439739e-05,
      "loss": 5.2433,
      "step": 2175
    },
    {
      "epoch": 0.05857853016256886,
      "grad_norm": 2.3499045372009277,
      "learning_rate": 4.771210676835081e-05,
      "loss": 5.1839,
      "step": 2180
    },
    {
      "epoch": 0.05871288458954722,
      "grad_norm": 3.9805495738983154,
      "learning_rate": 4.7705297562304235e-05,
      "loss": 5.0259,
      "step": 2185
    },
    {
      "epoch": 0.058847239016525596,
      "grad_norm": 2.440889835357666,
      "learning_rate": 4.7698488356257665e-05,
      "loss": 5.1003,
      "step": 2190
    },
    {
      "epoch": 0.05898159344350396,
      "grad_norm": 2.622908115386963,
      "learning_rate": 4.769167915021109e-05,
      "loss": 5.1748,
      "step": 2195
    },
    {
      "epoch": 0.059115947870482334,
      "grad_norm": 2.197916030883789,
      "learning_rate": 4.768486994416451e-05,
      "loss": 5.2149,
      "step": 2200
    },
    {
      "epoch": 0.0592503022974607,
      "grad_norm": 2.259551763534546,
      "learning_rate": 4.767806073811794e-05,
      "loss": 5.0572,
      "step": 2205
    },
    {
      "epoch": 0.05938465672443907,
      "grad_norm": 2.4556822776794434,
      "learning_rate": 4.767125153207137e-05,
      "loss": 5.2114,
      "step": 2210
    },
    {
      "epoch": 0.05951901115141744,
      "grad_norm": 2.4652628898620605,
      "learning_rate": 4.766444232602479e-05,
      "loss": 5.1193,
      "step": 2215
    },
    {
      "epoch": 0.05965336557839581,
      "grad_norm": 2.231334924697876,
      "learning_rate": 4.765763311997821e-05,
      "loss": 5.0828,
      "step": 2220
    },
    {
      "epoch": 0.059787720005374176,
      "grad_norm": 2.242621660232544,
      "learning_rate": 4.7650823913931634e-05,
      "loss": 5.1689,
      "step": 2225
    },
    {
      "epoch": 0.05992207443235255,
      "grad_norm": 2.6371543407440186,
      "learning_rate": 4.764401470788506e-05,
      "loss": 5.1462,
      "step": 2230
    },
    {
      "epoch": 0.060056428859330914,
      "grad_norm": 2.414228677749634,
      "learning_rate": 4.763720550183849e-05,
      "loss": 5.1458,
      "step": 2235
    },
    {
      "epoch": 0.060190783286309286,
      "grad_norm": 2.395897626876831,
      "learning_rate": 4.7630396295791914e-05,
      "loss": 5.0641,
      "step": 2240
    },
    {
      "epoch": 0.06032513771328765,
      "grad_norm": 2.5467405319213867,
      "learning_rate": 4.7623587089745337e-05,
      "loss": 5.1559,
      "step": 2245
    },
    {
      "epoch": 0.060459492140266025,
      "grad_norm": 2.7299716472625732,
      "learning_rate": 4.7616777883698766e-05,
      "loss": 5.088,
      "step": 2250
    },
    {
      "epoch": 0.06059384656724439,
      "grad_norm": 2.2927069664001465,
      "learning_rate": 4.760996867765219e-05,
      "loss": 5.1461,
      "step": 2255
    },
    {
      "epoch": 0.06072820099422276,
      "grad_norm": 2.3553717136383057,
      "learning_rate": 4.760315947160561e-05,
      "loss": 5.0812,
      "step": 2260
    },
    {
      "epoch": 0.06086255542120113,
      "grad_norm": 2.4433770179748535,
      "learning_rate": 4.759635026555904e-05,
      "loss": 5.1092,
      "step": 2265
    },
    {
      "epoch": 0.0609969098481795,
      "grad_norm": 2.2795238494873047,
      "learning_rate": 4.758954105951247e-05,
      "loss": 4.9923,
      "step": 2270
    },
    {
      "epoch": 0.061131264275157866,
      "grad_norm": 2.5695433616638184,
      "learning_rate": 4.758273185346589e-05,
      "loss": 5.1081,
      "step": 2275
    },
    {
      "epoch": 0.06126561870213623,
      "grad_norm": 2.391657829284668,
      "learning_rate": 4.757592264741931e-05,
      "loss": 5.0966,
      "step": 2280
    },
    {
      "epoch": 0.061399973129114604,
      "grad_norm": 2.594651699066162,
      "learning_rate": 4.7569113441372735e-05,
      "loss": 4.9291,
      "step": 2285
    },
    {
      "epoch": 0.06153432755609297,
      "grad_norm": 2.371537446975708,
      "learning_rate": 4.756230423532616e-05,
      "loss": 5.1881,
      "step": 2290
    },
    {
      "epoch": 0.06166868198307134,
      "grad_norm": 2.6550984382629395,
      "learning_rate": 4.7555495029279586e-05,
      "loss": 5.0681,
      "step": 2295
    },
    {
      "epoch": 0.06180303641004971,
      "grad_norm": 2.2198615074157715,
      "learning_rate": 4.7548685823233015e-05,
      "loss": 5.2027,
      "step": 2300
    },
    {
      "epoch": 0.06193739083702808,
      "grad_norm": 2.3532638549804688,
      "learning_rate": 4.754187661718644e-05,
      "loss": 5.1195,
      "step": 2305
    },
    {
      "epoch": 0.062071745264006446,
      "grad_norm": 2.3932807445526123,
      "learning_rate": 4.753506741113986e-05,
      "loss": 5.1162,
      "step": 2310
    },
    {
      "epoch": 0.06220609969098482,
      "grad_norm": 2.3105628490448,
      "learning_rate": 4.752825820509329e-05,
      "loss": 5.1738,
      "step": 2315
    },
    {
      "epoch": 0.062340454117963184,
      "grad_norm": 2.4400136470794678,
      "learning_rate": 4.752144899904671e-05,
      "loss": 5.146,
      "step": 2320
    },
    {
      "epoch": 0.06247480854494156,
      "grad_norm": 2.5597667694091797,
      "learning_rate": 4.751463979300014e-05,
      "loss": 5.1395,
      "step": 2325
    },
    {
      "epoch": 0.06260916297191993,
      "grad_norm": 2.564850330352783,
      "learning_rate": 4.750783058695356e-05,
      "loss": 5.1092,
      "step": 2330
    },
    {
      "epoch": 0.06274351739889829,
      "grad_norm": 2.501206398010254,
      "learning_rate": 4.750102138090699e-05,
      "loss": 5.1121,
      "step": 2335
    },
    {
      "epoch": 0.06287787182587666,
      "grad_norm": 2.3284380435943604,
      "learning_rate": 4.7494212174860414e-05,
      "loss": 5.1128,
      "step": 2340
    },
    {
      "epoch": 0.06301222625285503,
      "grad_norm": 2.2180492877960205,
      "learning_rate": 4.7487402968813836e-05,
      "loss": 5.0992,
      "step": 2345
    },
    {
      "epoch": 0.0631465806798334,
      "grad_norm": 2.617906332015991,
      "learning_rate": 4.748059376276726e-05,
      "loss": 5.0616,
      "step": 2350
    },
    {
      "epoch": 0.06328093510681176,
      "grad_norm": 2.268599033355713,
      "learning_rate": 4.747378455672069e-05,
      "loss": 5.012,
      "step": 2355
    },
    {
      "epoch": 0.06341528953379014,
      "grad_norm": 2.400578260421753,
      "learning_rate": 4.7466975350674117e-05,
      "loss": 5.1167,
      "step": 2360
    },
    {
      "epoch": 0.06354964396076851,
      "grad_norm": 2.1784756183624268,
      "learning_rate": 4.746016614462754e-05,
      "loss": 5.2294,
      "step": 2365
    },
    {
      "epoch": 0.06368399838774688,
      "grad_norm": 2.2773759365081787,
      "learning_rate": 4.745335693858096e-05,
      "loss": 5.0885,
      "step": 2370
    },
    {
      "epoch": 0.06381835281472524,
      "grad_norm": 2.4615862369537354,
      "learning_rate": 4.744654773253439e-05,
      "loss": 5.077,
      "step": 2375
    },
    {
      "epoch": 0.06395270724170361,
      "grad_norm": 2.4551687240600586,
      "learning_rate": 4.743973852648781e-05,
      "loss": 5.0032,
      "step": 2380
    },
    {
      "epoch": 0.06408706166868199,
      "grad_norm": 2.506337881088257,
      "learning_rate": 4.743292932044124e-05,
      "loss": 5.1934,
      "step": 2385
    },
    {
      "epoch": 0.06422141609566036,
      "grad_norm": 2.1597728729248047,
      "learning_rate": 4.7426120114394664e-05,
      "loss": 5.2099,
      "step": 2390
    },
    {
      "epoch": 0.06435577052263872,
      "grad_norm": 2.6743435859680176,
      "learning_rate": 4.741931090834809e-05,
      "loss": 5.0705,
      "step": 2395
    },
    {
      "epoch": 0.06449012494961709,
      "grad_norm": 2.736037254333496,
      "learning_rate": 4.7412501702301515e-05,
      "loss": 5.1356,
      "step": 2400
    },
    {
      "epoch": 0.06462447937659546,
      "grad_norm": 2.264522075653076,
      "learning_rate": 4.740569249625494e-05,
      "loss": 4.9938,
      "step": 2405
    },
    {
      "epoch": 0.06475883380357383,
      "grad_norm": 2.451328754425049,
      "learning_rate": 4.739888329020836e-05,
      "loss": 5.0513,
      "step": 2410
    },
    {
      "epoch": 0.06489318823055219,
      "grad_norm": 2.328263998031616,
      "learning_rate": 4.739207408416179e-05,
      "loss": 4.9759,
      "step": 2415
    },
    {
      "epoch": 0.06502754265753057,
      "grad_norm": 2.426508903503418,
      "learning_rate": 4.738526487811522e-05,
      "loss": 5.0611,
      "step": 2420
    },
    {
      "epoch": 0.06516189708450894,
      "grad_norm": 2.6437110900878906,
      "learning_rate": 4.737845567206864e-05,
      "loss": 5.0155,
      "step": 2425
    },
    {
      "epoch": 0.06529625151148731,
      "grad_norm": 2.3646633625030518,
      "learning_rate": 4.737164646602206e-05,
      "loss": 4.9942,
      "step": 2430
    },
    {
      "epoch": 0.06543060593846567,
      "grad_norm": 2.51942777633667,
      "learning_rate": 4.7364837259975484e-05,
      "loss": 4.986,
      "step": 2435
    },
    {
      "epoch": 0.06556496036544404,
      "grad_norm": 2.5781991481781006,
      "learning_rate": 4.7358028053928914e-05,
      "loss": 4.9825,
      "step": 2440
    },
    {
      "epoch": 0.06569931479242241,
      "grad_norm": 2.2016520500183105,
      "learning_rate": 4.735121884788234e-05,
      "loss": 5.0746,
      "step": 2445
    },
    {
      "epoch": 0.06583366921940077,
      "grad_norm": 2.1721441745758057,
      "learning_rate": 4.7344409641835765e-05,
      "loss": 5.1294,
      "step": 2450
    },
    {
      "epoch": 0.06596802364637915,
      "grad_norm": 2.2609691619873047,
      "learning_rate": 4.733760043578919e-05,
      "loss": 4.9707,
      "step": 2455
    },
    {
      "epoch": 0.06610237807335752,
      "grad_norm": 2.3571248054504395,
      "learning_rate": 4.7330791229742616e-05,
      "loss": 4.9806,
      "step": 2460
    },
    {
      "epoch": 0.06623673250033589,
      "grad_norm": 2.3186182975769043,
      "learning_rate": 4.732398202369604e-05,
      "loss": 5.164,
      "step": 2465
    },
    {
      "epoch": 0.06637108692731425,
      "grad_norm": 2.4597530364990234,
      "learning_rate": 4.731717281764946e-05,
      "loss": 5.088,
      "step": 2470
    },
    {
      "epoch": 0.06650544135429262,
      "grad_norm": 2.6319687366485596,
      "learning_rate": 4.731036361160289e-05,
      "loss": 4.954,
      "step": 2475
    },
    {
      "epoch": 0.066639795781271,
      "grad_norm": 2.141265869140625,
      "learning_rate": 4.730355440555632e-05,
      "loss": 5.1503,
      "step": 2480
    },
    {
      "epoch": 0.06677415020824937,
      "grad_norm": 2.9543850421905518,
      "learning_rate": 4.729674519950974e-05,
      "loss": 5.1751,
      "step": 2485
    },
    {
      "epoch": 0.06690850463522773,
      "grad_norm": 2.47092342376709,
      "learning_rate": 4.728993599346316e-05,
      "loss": 5.0543,
      "step": 2490
    },
    {
      "epoch": 0.0670428590622061,
      "grad_norm": 2.329517126083374,
      "learning_rate": 4.7283126787416586e-05,
      "loss": 5.1407,
      "step": 2495
    },
    {
      "epoch": 0.06717721348918447,
      "grad_norm": 2.660526752471924,
      "learning_rate": 4.7276317581370015e-05,
      "loss": 4.9308,
      "step": 2500
    },
    {
      "epoch": 0.06731156791616284,
      "grad_norm": 2.4411728382110596,
      "learning_rate": 4.726950837532344e-05,
      "loss": 5.1104,
      "step": 2505
    },
    {
      "epoch": 0.0674459223431412,
      "grad_norm": 2.4146604537963867,
      "learning_rate": 4.7262699169276866e-05,
      "loss": 5.0502,
      "step": 2510
    },
    {
      "epoch": 0.06758027677011957,
      "grad_norm": 2.4033758640289307,
      "learning_rate": 4.725588996323029e-05,
      "loss": 4.9998,
      "step": 2515
    },
    {
      "epoch": 0.06771463119709795,
      "grad_norm": 2.470900058746338,
      "learning_rate": 4.724908075718372e-05,
      "loss": 5.0287,
      "step": 2520
    },
    {
      "epoch": 0.06784898562407632,
      "grad_norm": 2.6848537921905518,
      "learning_rate": 4.724227155113714e-05,
      "loss": 4.9814,
      "step": 2525
    },
    {
      "epoch": 0.06798334005105468,
      "grad_norm": 2.4615607261657715,
      "learning_rate": 4.723546234509056e-05,
      "loss": 5.1187,
      "step": 2530
    },
    {
      "epoch": 0.06811769447803305,
      "grad_norm": 2.095226764678955,
      "learning_rate": 4.722865313904399e-05,
      "loss": 4.9794,
      "step": 2535
    },
    {
      "epoch": 0.06825204890501142,
      "grad_norm": 2.9362363815307617,
      "learning_rate": 4.722184393299742e-05,
      "loss": 5.027,
      "step": 2540
    },
    {
      "epoch": 0.0683864033319898,
      "grad_norm": 2.225034713745117,
      "learning_rate": 4.721503472695084e-05,
      "loss": 4.9875,
      "step": 2545
    },
    {
      "epoch": 0.06852075775896815,
      "grad_norm": 2.2556567192077637,
      "learning_rate": 4.7208225520904264e-05,
      "loss": 5.1891,
      "step": 2550
    },
    {
      "epoch": 0.06865511218594653,
      "grad_norm": 2.3483803272247314,
      "learning_rate": 4.720141631485769e-05,
      "loss": 5.0629,
      "step": 2555
    },
    {
      "epoch": 0.0687894666129249,
      "grad_norm": 2.343886613845825,
      "learning_rate": 4.7194607108811116e-05,
      "loss": 5.081,
      "step": 2560
    },
    {
      "epoch": 0.06892382103990327,
      "grad_norm": 2.187832832336426,
      "learning_rate": 4.718779790276454e-05,
      "loss": 5.0394,
      "step": 2565
    },
    {
      "epoch": 0.06905817546688163,
      "grad_norm": 2.317049026489258,
      "learning_rate": 4.718098869671797e-05,
      "loss": 5.0632,
      "step": 2570
    },
    {
      "epoch": 0.06919252989386,
      "grad_norm": 2.4061245918273926,
      "learning_rate": 4.717417949067139e-05,
      "loss": 5.1499,
      "step": 2575
    },
    {
      "epoch": 0.06932688432083838,
      "grad_norm": 2.3993353843688965,
      "learning_rate": 4.716737028462482e-05,
      "loss": 5.0663,
      "step": 2580
    },
    {
      "epoch": 0.06946123874781673,
      "grad_norm": 2.314722776412964,
      "learning_rate": 4.716056107857824e-05,
      "loss": 5.1535,
      "step": 2585
    },
    {
      "epoch": 0.0695955931747951,
      "grad_norm": 2.522005319595337,
      "learning_rate": 4.715375187253166e-05,
      "loss": 5.0434,
      "step": 2590
    },
    {
      "epoch": 0.06972994760177348,
      "grad_norm": 2.3243606090545654,
      "learning_rate": 4.714694266648509e-05,
      "loss": 5.0773,
      "step": 2595
    },
    {
      "epoch": 0.06986430202875185,
      "grad_norm": 2.281684160232544,
      "learning_rate": 4.7140133460438514e-05,
      "loss": 5.1993,
      "step": 2600
    },
    {
      "epoch": 0.06999865645573021,
      "grad_norm": 2.241349220275879,
      "learning_rate": 4.713332425439194e-05,
      "loss": 5.1532,
      "step": 2605
    },
    {
      "epoch": 0.07013301088270858,
      "grad_norm": 2.460827112197876,
      "learning_rate": 4.7126515048345366e-05,
      "loss": 5.1667,
      "step": 2610
    },
    {
      "epoch": 0.07026736530968695,
      "grad_norm": 2.3938283920288086,
      "learning_rate": 4.711970584229879e-05,
      "loss": 5.0523,
      "step": 2615
    },
    {
      "epoch": 0.07040171973666533,
      "grad_norm": 2.3147292137145996,
      "learning_rate": 4.711289663625221e-05,
      "loss": 4.9852,
      "step": 2620
    },
    {
      "epoch": 0.07053607416364369,
      "grad_norm": 2.4808735847473145,
      "learning_rate": 4.710608743020564e-05,
      "loss": 5.1447,
      "step": 2625
    },
    {
      "epoch": 0.07067042859062206,
      "grad_norm": 2.366708517074585,
      "learning_rate": 4.709927822415907e-05,
      "loss": 4.9905,
      "step": 2630
    },
    {
      "epoch": 0.07080478301760043,
      "grad_norm": 2.293250322341919,
      "learning_rate": 4.709246901811249e-05,
      "loss": 5.1174,
      "step": 2635
    },
    {
      "epoch": 0.0709391374445788,
      "grad_norm": 2.2710607051849365,
      "learning_rate": 4.708565981206591e-05,
      "loss": 4.9623,
      "step": 2640
    },
    {
      "epoch": 0.07107349187155716,
      "grad_norm": 2.392634868621826,
      "learning_rate": 4.707885060601934e-05,
      "loss": 5.072,
      "step": 2645
    },
    {
      "epoch": 0.07120784629853553,
      "grad_norm": 2.289745807647705,
      "learning_rate": 4.7072041399972764e-05,
      "loss": 5.0861,
      "step": 2650
    },
    {
      "epoch": 0.07134220072551391,
      "grad_norm": 2.070662021636963,
      "learning_rate": 4.706523219392619e-05,
      "loss": 5.0058,
      "step": 2655
    },
    {
      "epoch": 0.07147655515249228,
      "grad_norm": 2.2136707305908203,
      "learning_rate": 4.7058422987879615e-05,
      "loss": 5.0072,
      "step": 2660
    },
    {
      "epoch": 0.07161090957947064,
      "grad_norm": 2.3967554569244385,
      "learning_rate": 4.7051613781833044e-05,
      "loss": 5.028,
      "step": 2665
    },
    {
      "epoch": 0.07174526400644901,
      "grad_norm": 3.260460615158081,
      "learning_rate": 4.704480457578647e-05,
      "loss": 5.1053,
      "step": 2670
    },
    {
      "epoch": 0.07187961843342738,
      "grad_norm": 2.705599784851074,
      "learning_rate": 4.703799536973989e-05,
      "loss": 4.956,
      "step": 2675
    },
    {
      "epoch": 0.07201397286040576,
      "grad_norm": 2.2675018310546875,
      "learning_rate": 4.703118616369331e-05,
      "loss": 5.0302,
      "step": 2680
    },
    {
      "epoch": 0.07214832728738411,
      "grad_norm": 2.222308397293091,
      "learning_rate": 4.702437695764674e-05,
      "loss": 5.0564,
      "step": 2685
    },
    {
      "epoch": 0.07228268171436249,
      "grad_norm": 2.364720344543457,
      "learning_rate": 4.701756775160017e-05,
      "loss": 5.0174,
      "step": 2690
    },
    {
      "epoch": 0.07241703614134086,
      "grad_norm": 2.5279181003570557,
      "learning_rate": 4.701075854555359e-05,
      "loss": 5.0167,
      "step": 2695
    },
    {
      "epoch": 0.07255139056831923,
      "grad_norm": 2.3830904960632324,
      "learning_rate": 4.7003949339507014e-05,
      "loss": 4.9756,
      "step": 2700
    },
    {
      "epoch": 0.07268574499529759,
      "grad_norm": 2.4851126670837402,
      "learning_rate": 4.699714013346044e-05,
      "loss": 5.0366,
      "step": 2705
    },
    {
      "epoch": 0.07282009942227596,
      "grad_norm": 2.32207989692688,
      "learning_rate": 4.6990330927413865e-05,
      "loss": 5.0195,
      "step": 2710
    },
    {
      "epoch": 0.07295445384925434,
      "grad_norm": 2.4050586223602295,
      "learning_rate": 4.698352172136729e-05,
      "loss": 5.1234,
      "step": 2715
    },
    {
      "epoch": 0.07308880827623271,
      "grad_norm": 2.2521891593933105,
      "learning_rate": 4.6976712515320716e-05,
      "loss": 5.0699,
      "step": 2720
    },
    {
      "epoch": 0.07322316270321107,
      "grad_norm": 2.549636125564575,
      "learning_rate": 4.6969903309274145e-05,
      "loss": 4.9025,
      "step": 2725
    },
    {
      "epoch": 0.07335751713018944,
      "grad_norm": 2.7519614696502686,
      "learning_rate": 4.696309410322757e-05,
      "loss": 4.9843,
      "step": 2730
    },
    {
      "epoch": 0.07349187155716781,
      "grad_norm": 2.649751901626587,
      "learning_rate": 4.695628489718099e-05,
      "loss": 5.0242,
      "step": 2735
    },
    {
      "epoch": 0.07362622598414617,
      "grad_norm": 2.270838499069214,
      "learning_rate": 4.694947569113441e-05,
      "loss": 4.9189,
      "step": 2740
    },
    {
      "epoch": 0.07376058041112454,
      "grad_norm": 2.2538695335388184,
      "learning_rate": 4.694266648508784e-05,
      "loss": 5.0254,
      "step": 2745
    },
    {
      "epoch": 0.07389493483810292,
      "grad_norm": 2.377070426940918,
      "learning_rate": 4.693585727904127e-05,
      "loss": 5.1025,
      "step": 2750
    },
    {
      "epoch": 0.07402928926508129,
      "grad_norm": 2.4636082649230957,
      "learning_rate": 4.692904807299469e-05,
      "loss": 5.0375,
      "step": 2755
    },
    {
      "epoch": 0.07416364369205965,
      "grad_norm": 2.2239975929260254,
      "learning_rate": 4.6922238866948115e-05,
      "loss": 4.8865,
      "step": 2760
    },
    {
      "epoch": 0.07429799811903802,
      "grad_norm": 2.287213087081909,
      "learning_rate": 4.691542966090154e-05,
      "loss": 5.1238,
      "step": 2765
    },
    {
      "epoch": 0.07443235254601639,
      "grad_norm": 2.3553996086120605,
      "learning_rate": 4.6908620454854966e-05,
      "loss": 4.9435,
      "step": 2770
    },
    {
      "epoch": 0.07456670697299476,
      "grad_norm": 2.5199694633483887,
      "learning_rate": 4.690181124880839e-05,
      "loss": 4.9821,
      "step": 2775
    },
    {
      "epoch": 0.07470106139997312,
      "grad_norm": 2.2390990257263184,
      "learning_rate": 4.689500204276182e-05,
      "loss": 5.0951,
      "step": 2780
    },
    {
      "epoch": 0.0748354158269515,
      "grad_norm": 2.681732177734375,
      "learning_rate": 4.688819283671524e-05,
      "loss": 5.0516,
      "step": 2785
    },
    {
      "epoch": 0.07496977025392987,
      "grad_norm": 2.4223461151123047,
      "learning_rate": 4.688138363066867e-05,
      "loss": 5.0633,
      "step": 2790
    },
    {
      "epoch": 0.07510412468090824,
      "grad_norm": 2.39823579788208,
      "learning_rate": 4.687457442462209e-05,
      "loss": 5.0584,
      "step": 2795
    },
    {
      "epoch": 0.0752384791078866,
      "grad_norm": 2.076220989227295,
      "learning_rate": 4.6867765218575513e-05,
      "loss": 4.9805,
      "step": 2800
    },
    {
      "epoch": 0.07537283353486497,
      "grad_norm": 2.4038548469543457,
      "learning_rate": 4.686095601252894e-05,
      "loss": 5.0362,
      "step": 2805
    },
    {
      "epoch": 0.07550718796184334,
      "grad_norm": 2.4277186393737793,
      "learning_rate": 4.685414680648237e-05,
      "loss": 5.029,
      "step": 2810
    },
    {
      "epoch": 0.07564154238882172,
      "grad_norm": 2.360374689102173,
      "learning_rate": 4.6847337600435794e-05,
      "loss": 4.9493,
      "step": 2815
    },
    {
      "epoch": 0.07577589681580008,
      "grad_norm": 2.3976619243621826,
      "learning_rate": 4.6840528394389216e-05,
      "loss": 4.8935,
      "step": 2820
    },
    {
      "epoch": 0.07591025124277845,
      "grad_norm": 2.2577006816864014,
      "learning_rate": 4.683371918834264e-05,
      "loss": 4.955,
      "step": 2825
    },
    {
      "epoch": 0.07604460566975682,
      "grad_norm": 2.4220681190490723,
      "learning_rate": 4.682690998229607e-05,
      "loss": 4.9848,
      "step": 2830
    },
    {
      "epoch": 0.0761789600967352,
      "grad_norm": 2.5361251831054688,
      "learning_rate": 4.682010077624949e-05,
      "loss": 4.8206,
      "step": 2835
    },
    {
      "epoch": 0.07631331452371355,
      "grad_norm": 2.34853196144104,
      "learning_rate": 4.681329157020292e-05,
      "loss": 4.9062,
      "step": 2840
    },
    {
      "epoch": 0.07644766895069192,
      "grad_norm": 2.45651912689209,
      "learning_rate": 4.680648236415634e-05,
      "loss": 4.8791,
      "step": 2845
    },
    {
      "epoch": 0.0765820233776703,
      "grad_norm": 2.2464826107025146,
      "learning_rate": 4.679967315810977e-05,
      "loss": 5.0624,
      "step": 2850
    },
    {
      "epoch": 0.07671637780464867,
      "grad_norm": 2.316802978515625,
      "learning_rate": 4.679286395206319e-05,
      "loss": 4.9207,
      "step": 2855
    },
    {
      "epoch": 0.07685073223162703,
      "grad_norm": 2.4910387992858887,
      "learning_rate": 4.6786054746016615e-05,
      "loss": 5.0292,
      "step": 2860
    },
    {
      "epoch": 0.0769850866586054,
      "grad_norm": 2.3384172916412354,
      "learning_rate": 4.6779245539970044e-05,
      "loss": 4.9664,
      "step": 2865
    },
    {
      "epoch": 0.07711944108558377,
      "grad_norm": 2.515455484390259,
      "learning_rate": 4.677243633392347e-05,
      "loss": 5.0691,
      "step": 2870
    },
    {
      "epoch": 0.07725379551256215,
      "grad_norm": 2.5678048133850098,
      "learning_rate": 4.6765627127876895e-05,
      "loss": 4.9324,
      "step": 2875
    },
    {
      "epoch": 0.0773881499395405,
      "grad_norm": 2.3092541694641113,
      "learning_rate": 4.675881792183032e-05,
      "loss": 5.0426,
      "step": 2880
    },
    {
      "epoch": 0.07752250436651888,
      "grad_norm": 2.2976062297821045,
      "learning_rate": 4.675200871578374e-05,
      "loss": 4.8977,
      "step": 2885
    },
    {
      "epoch": 0.07765685879349725,
      "grad_norm": 2.2484400272369385,
      "learning_rate": 4.674519950973716e-05,
      "loss": 5.1509,
      "step": 2890
    },
    {
      "epoch": 0.07779121322047561,
      "grad_norm": 2.293313980102539,
      "learning_rate": 4.673839030369059e-05,
      "loss": 5.057,
      "step": 2895
    },
    {
      "epoch": 0.07792556764745398,
      "grad_norm": 2.252776861190796,
      "learning_rate": 4.673158109764402e-05,
      "loss": 5.0953,
      "step": 2900
    },
    {
      "epoch": 0.07805992207443235,
      "grad_norm": 2.192119598388672,
      "learning_rate": 4.672477189159744e-05,
      "loss": 4.996,
      "step": 2905
    },
    {
      "epoch": 0.07819427650141073,
      "grad_norm": 2.5377445220947266,
      "learning_rate": 4.6717962685550864e-05,
      "loss": 4.8901,
      "step": 2910
    },
    {
      "epoch": 0.07832863092838908,
      "grad_norm": 2.3738508224487305,
      "learning_rate": 4.671115347950429e-05,
      "loss": 4.9915,
      "step": 2915
    },
    {
      "epoch": 0.07846298535536746,
      "grad_norm": 2.434407949447632,
      "learning_rate": 4.6704344273457716e-05,
      "loss": 4.9025,
      "step": 2920
    },
    {
      "epoch": 0.07859733978234583,
      "grad_norm": 2.2632462978363037,
      "learning_rate": 4.669753506741114e-05,
      "loss": 4.9604,
      "step": 2925
    },
    {
      "epoch": 0.0787316942093242,
      "grad_norm": 2.3816940784454346,
      "learning_rate": 4.669072586136457e-05,
      "loss": 4.9577,
      "step": 2930
    },
    {
      "epoch": 0.07886604863630256,
      "grad_norm": 2.3685219287872314,
      "learning_rate": 4.6683916655317996e-05,
      "loss": 4.9602,
      "step": 2935
    },
    {
      "epoch": 0.07900040306328093,
      "grad_norm": 2.2433366775512695,
      "learning_rate": 4.667710744927142e-05,
      "loss": 5.0234,
      "step": 2940
    },
    {
      "epoch": 0.0791347574902593,
      "grad_norm": 2.35847806930542,
      "learning_rate": 4.667029824322484e-05,
      "loss": 5.0336,
      "step": 2945
    },
    {
      "epoch": 0.07926911191723768,
      "grad_norm": 2.6191012859344482,
      "learning_rate": 4.666348903717826e-05,
      "loss": 4.8677,
      "step": 2950
    },
    {
      "epoch": 0.07940346634421604,
      "grad_norm": 2.3648300170898438,
      "learning_rate": 4.665667983113169e-05,
      "loss": 4.9328,
      "step": 2955
    },
    {
      "epoch": 0.07953782077119441,
      "grad_norm": 2.472687005996704,
      "learning_rate": 4.664987062508512e-05,
      "loss": 5.0775,
      "step": 2960
    },
    {
      "epoch": 0.07967217519817278,
      "grad_norm": 2.536130428314209,
      "learning_rate": 4.664306141903854e-05,
      "loss": 4.9376,
      "step": 2965
    },
    {
      "epoch": 0.07980652962515115,
      "grad_norm": 2.417445659637451,
      "learning_rate": 4.6636252212991965e-05,
      "loss": 5.0188,
      "step": 2970
    },
    {
      "epoch": 0.07994088405212951,
      "grad_norm": 2.951399087905884,
      "learning_rate": 4.6629443006945394e-05,
      "loss": 4.9488,
      "step": 2975
    },
    {
      "epoch": 0.08007523847910789,
      "grad_norm": 2.51804256439209,
      "learning_rate": 4.662263380089882e-05,
      "loss": 4.9984,
      "step": 2980
    },
    {
      "epoch": 0.08020959290608626,
      "grad_norm": 2.422224998474121,
      "learning_rate": 4.661582459485224e-05,
      "loss": 4.9569,
      "step": 2985
    },
    {
      "epoch": 0.08034394733306463,
      "grad_norm": 2.5573370456695557,
      "learning_rate": 4.660901538880567e-05,
      "loss": 4.9572,
      "step": 2990
    },
    {
      "epoch": 0.08047830176004299,
      "grad_norm": 2.5019991397857666,
      "learning_rate": 4.66022061827591e-05,
      "loss": 4.9161,
      "step": 2995
    },
    {
      "epoch": 0.08061265618702136,
      "grad_norm": 2.424339532852173,
      "learning_rate": 4.659539697671252e-05,
      "loss": 4.9821,
      "step": 3000
    },
    {
      "epoch": 0.08074701061399973,
      "grad_norm": 2.5886359214782715,
      "learning_rate": 4.658858777066594e-05,
      "loss": 4.9406,
      "step": 3005
    },
    {
      "epoch": 0.0808813650409781,
      "grad_norm": 2.5271315574645996,
      "learning_rate": 4.6581778564619364e-05,
      "loss": 4.9047,
      "step": 3010
    },
    {
      "epoch": 0.08101571946795647,
      "grad_norm": 2.215725898742676,
      "learning_rate": 4.657496935857279e-05,
      "loss": 4.9045,
      "step": 3015
    },
    {
      "epoch": 0.08115007389493484,
      "grad_norm": 2.2105448246002197,
      "learning_rate": 4.656816015252622e-05,
      "loss": 5.0101,
      "step": 3020
    },
    {
      "epoch": 0.08128442832191321,
      "grad_norm": 2.5475404262542725,
      "learning_rate": 4.6561350946479644e-05,
      "loss": 5.0141,
      "step": 3025
    },
    {
      "epoch": 0.08141878274889158,
      "grad_norm": 2.4291939735412598,
      "learning_rate": 4.6554541740433067e-05,
      "loss": 5.0046,
      "step": 3030
    },
    {
      "epoch": 0.08155313717586994,
      "grad_norm": 2.349987506866455,
      "learning_rate": 4.654773253438649e-05,
      "loss": 4.9757,
      "step": 3035
    },
    {
      "epoch": 0.08168749160284831,
      "grad_norm": 2.3299314975738525,
      "learning_rate": 4.654092332833992e-05,
      "loss": 5.002,
      "step": 3040
    },
    {
      "epoch": 0.08182184602982669,
      "grad_norm": 2.4213287830352783,
      "learning_rate": 4.653411412229334e-05,
      "loss": 5.0308,
      "step": 3045
    },
    {
      "epoch": 0.08195620045680505,
      "grad_norm": 2.607639789581299,
      "learning_rate": 4.652730491624677e-05,
      "loss": 4.8631,
      "step": 3050
    },
    {
      "epoch": 0.08209055488378342,
      "grad_norm": 2.5304009914398193,
      "learning_rate": 4.652049571020019e-05,
      "loss": 4.8938,
      "step": 3055
    },
    {
      "epoch": 0.08222490931076179,
      "grad_norm": 2.324573040008545,
      "learning_rate": 4.651368650415362e-05,
      "loss": 4.9556,
      "step": 3060
    },
    {
      "epoch": 0.08235926373774016,
      "grad_norm": 2.276792049407959,
      "learning_rate": 4.650687729810704e-05,
      "loss": 4.9162,
      "step": 3065
    },
    {
      "epoch": 0.08249361816471852,
      "grad_norm": 2.319880485534668,
      "learning_rate": 4.6500068092060465e-05,
      "loss": 4.9036,
      "step": 3070
    },
    {
      "epoch": 0.0826279725916969,
      "grad_norm": 2.460873603820801,
      "learning_rate": 4.6493258886013894e-05,
      "loss": 4.838,
      "step": 3075
    },
    {
      "epoch": 0.08276232701867527,
      "grad_norm": 2.4000260829925537,
      "learning_rate": 4.648644967996732e-05,
      "loss": 4.9028,
      "step": 3080
    },
    {
      "epoch": 0.08289668144565364,
      "grad_norm": 2.5372064113616943,
      "learning_rate": 4.6479640473920745e-05,
      "loss": 5.0519,
      "step": 3085
    },
    {
      "epoch": 0.083031035872632,
      "grad_norm": 2.3044536113739014,
      "learning_rate": 4.647283126787417e-05,
      "loss": 5.0919,
      "step": 3090
    },
    {
      "epoch": 0.08316539029961037,
      "grad_norm": 2.452742099761963,
      "learning_rate": 4.646602206182759e-05,
      "loss": 4.9606,
      "step": 3095
    },
    {
      "epoch": 0.08329974472658874,
      "grad_norm": 2.202953338623047,
      "learning_rate": 4.645921285578102e-05,
      "loss": 4.91,
      "step": 3100
    },
    {
      "epoch": 0.08343409915356712,
      "grad_norm": 2.39689564704895,
      "learning_rate": 4.645240364973444e-05,
      "loss": 4.8595,
      "step": 3105
    },
    {
      "epoch": 0.08356845358054547,
      "grad_norm": 2.6198947429656982,
      "learning_rate": 4.644559444368787e-05,
      "loss": 4.985,
      "step": 3110
    },
    {
      "epoch": 0.08370280800752385,
      "grad_norm": 2.3617098331451416,
      "learning_rate": 4.643878523764129e-05,
      "loss": 4.9485,
      "step": 3115
    },
    {
      "epoch": 0.08383716243450222,
      "grad_norm": 2.5182414054870605,
      "learning_rate": 4.643197603159472e-05,
      "loss": 4.904,
      "step": 3120
    },
    {
      "epoch": 0.08397151686148059,
      "grad_norm": 2.5316312313079834,
      "learning_rate": 4.6425166825548144e-05,
      "loss": 5.0195,
      "step": 3125
    },
    {
      "epoch": 0.08410587128845895,
      "grad_norm": 2.136096715927124,
      "learning_rate": 4.6418357619501566e-05,
      "loss": 5.0062,
      "step": 3130
    },
    {
      "epoch": 0.08424022571543732,
      "grad_norm": 2.3298838138580322,
      "learning_rate": 4.641154841345499e-05,
      "loss": 4.9261,
      "step": 3135
    },
    {
      "epoch": 0.0843745801424157,
      "grad_norm": 2.1770806312561035,
      "learning_rate": 4.6404739207408424e-05,
      "loss": 5.1226,
      "step": 3140
    },
    {
      "epoch": 0.08450893456939407,
      "grad_norm": 2.1904795169830322,
      "learning_rate": 4.6397930001361846e-05,
      "loss": 4.8955,
      "step": 3145
    },
    {
      "epoch": 0.08464328899637243,
      "grad_norm": 2.2624268531799316,
      "learning_rate": 4.639112079531527e-05,
      "loss": 5.0873,
      "step": 3150
    },
    {
      "epoch": 0.0847776434233508,
      "grad_norm": 2.3995871543884277,
      "learning_rate": 4.638431158926869e-05,
      "loss": 5.1072,
      "step": 3155
    },
    {
      "epoch": 0.08491199785032917,
      "grad_norm": 2.1977829933166504,
      "learning_rate": 4.637750238322212e-05,
      "loss": 5.0039,
      "step": 3160
    },
    {
      "epoch": 0.08504635227730754,
      "grad_norm": 2.339897871017456,
      "learning_rate": 4.637069317717554e-05,
      "loss": 5.0036,
      "step": 3165
    },
    {
      "epoch": 0.0851807067042859,
      "grad_norm": 2.371072769165039,
      "learning_rate": 4.636388397112897e-05,
      "loss": 4.8441,
      "step": 3170
    },
    {
      "epoch": 0.08531506113126427,
      "grad_norm": 2.293200969696045,
      "learning_rate": 4.6357074765082394e-05,
      "loss": 4.9125,
      "step": 3175
    },
    {
      "epoch": 0.08544941555824265,
      "grad_norm": 2.3724539279937744,
      "learning_rate": 4.6350265559035816e-05,
      "loss": 4.9601,
      "step": 3180
    },
    {
      "epoch": 0.08558376998522102,
      "grad_norm": 2.492265224456787,
      "learning_rate": 4.6343456352989245e-05,
      "loss": 4.9638,
      "step": 3185
    },
    {
      "epoch": 0.08571812441219938,
      "grad_norm": 2.412221670150757,
      "learning_rate": 4.633664714694267e-05,
      "loss": 4.9121,
      "step": 3190
    },
    {
      "epoch": 0.08585247883917775,
      "grad_norm": 2.345449447631836,
      "learning_rate": 4.632983794089609e-05,
      "loss": 4.826,
      "step": 3195
    },
    {
      "epoch": 0.08598683326615612,
      "grad_norm": 2.3856143951416016,
      "learning_rate": 4.632302873484952e-05,
      "loss": 4.9671,
      "step": 3200
    },
    {
      "epoch": 0.08612118769313448,
      "grad_norm": 2.2164435386657715,
      "learning_rate": 4.631621952880295e-05,
      "loss": 4.8246,
      "step": 3205
    },
    {
      "epoch": 0.08625554212011285,
      "grad_norm": 2.436462640762329,
      "learning_rate": 4.630941032275637e-05,
      "loss": 5.0598,
      "step": 3210
    },
    {
      "epoch": 0.08638989654709123,
      "grad_norm": 2.385547399520874,
      "learning_rate": 4.630260111670979e-05,
      "loss": 4.8489,
      "step": 3215
    },
    {
      "epoch": 0.0865242509740696,
      "grad_norm": 2.5401670932769775,
      "learning_rate": 4.6295791910663214e-05,
      "loss": 4.8629,
      "step": 3220
    },
    {
      "epoch": 0.08665860540104796,
      "grad_norm": 2.3023781776428223,
      "learning_rate": 4.6288982704616643e-05,
      "loss": 4.9337,
      "step": 3225
    },
    {
      "epoch": 0.08679295982802633,
      "grad_norm": 2.53151798248291,
      "learning_rate": 4.628217349857007e-05,
      "loss": 4.9401,
      "step": 3230
    },
    {
      "epoch": 0.0869273142550047,
      "grad_norm": 2.152036190032959,
      "learning_rate": 4.6275364292523495e-05,
      "loss": 4.9722,
      "step": 3235
    },
    {
      "epoch": 0.08706166868198308,
      "grad_norm": 2.1881864070892334,
      "learning_rate": 4.626855508647692e-05,
      "loss": 4.9625,
      "step": 3240
    },
    {
      "epoch": 0.08719602310896143,
      "grad_norm": 2.2732608318328857,
      "learning_rate": 4.6261745880430346e-05,
      "loss": 4.8329,
      "step": 3245
    },
    {
      "epoch": 0.08733037753593981,
      "grad_norm": 2.151388168334961,
      "learning_rate": 4.625493667438377e-05,
      "loss": 5.0471,
      "step": 3250
    },
    {
      "epoch": 0.08746473196291818,
      "grad_norm": 2.393622636795044,
      "learning_rate": 4.624812746833719e-05,
      "loss": 4.9852,
      "step": 3255
    },
    {
      "epoch": 0.08759908638989655,
      "grad_norm": 2.534522533416748,
      "learning_rate": 4.624131826229062e-05,
      "loss": 4.8568,
      "step": 3260
    },
    {
      "epoch": 0.08773344081687491,
      "grad_norm": 2.4262566566467285,
      "learning_rate": 4.623450905624405e-05,
      "loss": 5.0406,
      "step": 3265
    },
    {
      "epoch": 0.08786779524385328,
      "grad_norm": 2.323293924331665,
      "learning_rate": 4.622769985019747e-05,
      "loss": 5.0479,
      "step": 3270
    },
    {
      "epoch": 0.08800214967083166,
      "grad_norm": 2.4120004177093506,
      "learning_rate": 4.622089064415089e-05,
      "loss": 5.0464,
      "step": 3275
    },
    {
      "epoch": 0.08813650409781003,
      "grad_norm": 2.489932060241699,
      "learning_rate": 4.6214081438104316e-05,
      "loss": 4.9575,
      "step": 3280
    },
    {
      "epoch": 0.08827085852478839,
      "grad_norm": 2.620137929916382,
      "learning_rate": 4.6207272232057745e-05,
      "loss": 4.9206,
      "step": 3285
    },
    {
      "epoch": 0.08840521295176676,
      "grad_norm": 2.4803669452667236,
      "learning_rate": 4.6200463026011174e-05,
      "loss": 5.0104,
      "step": 3290
    },
    {
      "epoch": 0.08853956737874513,
      "grad_norm": 2.33331298828125,
      "learning_rate": 4.6193653819964596e-05,
      "loss": 4.9044,
      "step": 3295
    },
    {
      "epoch": 0.0886739218057235,
      "grad_norm": 2.533698558807373,
      "learning_rate": 4.618684461391802e-05,
      "loss": 5.0146,
      "step": 3300
    },
    {
      "epoch": 0.08880827623270186,
      "grad_norm": 2.2677273750305176,
      "learning_rate": 4.618003540787145e-05,
      "loss": 5.0116,
      "step": 3305
    },
    {
      "epoch": 0.08894263065968024,
      "grad_norm": 2.6705358028411865,
      "learning_rate": 4.617322620182487e-05,
      "loss": 4.9719,
      "step": 3310
    },
    {
      "epoch": 0.08907698508665861,
      "grad_norm": 2.181166648864746,
      "learning_rate": 4.616641699577829e-05,
      "loss": 5.0249,
      "step": 3315
    },
    {
      "epoch": 0.08921133951363698,
      "grad_norm": 2.366442918777466,
      "learning_rate": 4.615960778973172e-05,
      "loss": 4.9317,
      "step": 3320
    },
    {
      "epoch": 0.08934569394061534,
      "grad_norm": 2.1919569969177246,
      "learning_rate": 4.615279858368514e-05,
      "loss": 4.9039,
      "step": 3325
    },
    {
      "epoch": 0.08948004836759371,
      "grad_norm": 2.3448522090911865,
      "learning_rate": 4.614598937763857e-05,
      "loss": 4.8646,
      "step": 3330
    },
    {
      "epoch": 0.08961440279457208,
      "grad_norm": 2.3348026275634766,
      "learning_rate": 4.6139180171591994e-05,
      "loss": 4.9396,
      "step": 3335
    },
    {
      "epoch": 0.08974875722155046,
      "grad_norm": 2.298412799835205,
      "learning_rate": 4.613237096554542e-05,
      "loss": 4.9486,
      "step": 3340
    },
    {
      "epoch": 0.08988311164852882,
      "grad_norm": 2.1365394592285156,
      "learning_rate": 4.612556175949884e-05,
      "loss": 4.9046,
      "step": 3345
    },
    {
      "epoch": 0.09001746607550719,
      "grad_norm": 2.5887131690979004,
      "learning_rate": 4.611875255345227e-05,
      "loss": 4.9578,
      "step": 3350
    },
    {
      "epoch": 0.09015182050248556,
      "grad_norm": 2.4161834716796875,
      "learning_rate": 4.61119433474057e-05,
      "loss": 4.9373,
      "step": 3355
    },
    {
      "epoch": 0.09028617492946392,
      "grad_norm": 2.466649293899536,
      "learning_rate": 4.610513414135912e-05,
      "loss": 4.8397,
      "step": 3360
    },
    {
      "epoch": 0.09042052935644229,
      "grad_norm": 2.365720748901367,
      "learning_rate": 4.609832493531254e-05,
      "loss": 4.8461,
      "step": 3365
    },
    {
      "epoch": 0.09055488378342066,
      "grad_norm": 2.2871487140655518,
      "learning_rate": 4.609151572926597e-05,
      "loss": 4.971,
      "step": 3370
    },
    {
      "epoch": 0.09068923821039904,
      "grad_norm": 2.8257601261138916,
      "learning_rate": 4.608470652321939e-05,
      "loss": 4.9835,
      "step": 3375
    },
    {
      "epoch": 0.0908235926373774,
      "grad_norm": 2.2513794898986816,
      "learning_rate": 4.607789731717282e-05,
      "loss": 5.0145,
      "step": 3380
    },
    {
      "epoch": 0.09095794706435577,
      "grad_norm": 2.658571481704712,
      "learning_rate": 4.6071088111126244e-05,
      "loss": 4.9941,
      "step": 3385
    },
    {
      "epoch": 0.09109230149133414,
      "grad_norm": 2.5282750129699707,
      "learning_rate": 4.606427890507967e-05,
      "loss": 4.9412,
      "step": 3390
    },
    {
      "epoch": 0.09122665591831251,
      "grad_norm": 2.568573236465454,
      "learning_rate": 4.6057469699033095e-05,
      "loss": 4.949,
      "step": 3395
    },
    {
      "epoch": 0.09136101034529087,
      "grad_norm": 2.4815893173217773,
      "learning_rate": 4.605066049298652e-05,
      "loss": 5.0554,
      "step": 3400
    },
    {
      "epoch": 0.09149536477226924,
      "grad_norm": 2.4829440116882324,
      "learning_rate": 4.604385128693994e-05,
      "loss": 4.8589,
      "step": 3405
    },
    {
      "epoch": 0.09162971919924762,
      "grad_norm": 2.2588183879852295,
      "learning_rate": 4.603704208089337e-05,
      "loss": 4.9571,
      "step": 3410
    },
    {
      "epoch": 0.09176407362622599,
      "grad_norm": 2.1706416606903076,
      "learning_rate": 4.60302328748468e-05,
      "loss": 4.8513,
      "step": 3415
    },
    {
      "epoch": 0.09189842805320435,
      "grad_norm": 2.259800910949707,
      "learning_rate": 4.602342366880022e-05,
      "loss": 4.9065,
      "step": 3420
    },
    {
      "epoch": 0.09203278248018272,
      "grad_norm": 2.5378379821777344,
      "learning_rate": 4.601661446275364e-05,
      "loss": 5.0089,
      "step": 3425
    },
    {
      "epoch": 0.09216713690716109,
      "grad_norm": 2.327890634536743,
      "learning_rate": 4.600980525670707e-05,
      "loss": 4.9966,
      "step": 3430
    },
    {
      "epoch": 0.09230149133413947,
      "grad_norm": 2.3996288776397705,
      "learning_rate": 4.6002996050660494e-05,
      "loss": 4.9848,
      "step": 3435
    },
    {
      "epoch": 0.09243584576111782,
      "grad_norm": 2.450749158859253,
      "learning_rate": 4.599618684461392e-05,
      "loss": 4.8816,
      "step": 3440
    },
    {
      "epoch": 0.0925702001880962,
      "grad_norm": 2.3224682807922363,
      "learning_rate": 4.5989377638567345e-05,
      "loss": 5.0016,
      "step": 3445
    },
    {
      "epoch": 0.09270455461507457,
      "grad_norm": 2.397367238998413,
      "learning_rate": 4.5982568432520774e-05,
      "loss": 4.8428,
      "step": 3450
    },
    {
      "epoch": 0.09283890904205294,
      "grad_norm": 2.6210317611694336,
      "learning_rate": 4.5975759226474197e-05,
      "loss": 4.9057,
      "step": 3455
    },
    {
      "epoch": 0.0929732634690313,
      "grad_norm": 2.44633150100708,
      "learning_rate": 4.596895002042762e-05,
      "loss": 4.8284,
      "step": 3460
    },
    {
      "epoch": 0.09310761789600967,
      "grad_norm": 2.477090358734131,
      "learning_rate": 4.596214081438104e-05,
      "loss": 4.8567,
      "step": 3465
    },
    {
      "epoch": 0.09324197232298805,
      "grad_norm": 2.488525390625,
      "learning_rate": 4.595533160833447e-05,
      "loss": 4.9228,
      "step": 3470
    },
    {
      "epoch": 0.09337632674996642,
      "grad_norm": 2.200896978378296,
      "learning_rate": 4.59485224022879e-05,
      "loss": 4.9107,
      "step": 3475
    },
    {
      "epoch": 0.09351068117694478,
      "grad_norm": 2.488555669784546,
      "learning_rate": 4.594171319624132e-05,
      "loss": 4.9039,
      "step": 3480
    },
    {
      "epoch": 0.09364503560392315,
      "grad_norm": 2.311699390411377,
      "learning_rate": 4.5934903990194744e-05,
      "loss": 4.8956,
      "step": 3485
    },
    {
      "epoch": 0.09377939003090152,
      "grad_norm": 2.2264766693115234,
      "learning_rate": 4.5928094784148166e-05,
      "loss": 4.9072,
      "step": 3490
    },
    {
      "epoch": 0.09391374445787988,
      "grad_norm": 2.348525047302246,
      "learning_rate": 4.5921285578101595e-05,
      "loss": 5.0134,
      "step": 3495
    },
    {
      "epoch": 0.09404809888485825,
      "grad_norm": 2.2881531715393066,
      "learning_rate": 4.5914476372055024e-05,
      "loss": 4.9881,
      "step": 3500
    },
    {
      "epoch": 0.09418245331183663,
      "grad_norm": 2.463690757751465,
      "learning_rate": 4.5907667166008446e-05,
      "loss": 4.9656,
      "step": 3505
    },
    {
      "epoch": 0.094316807738815,
      "grad_norm": 2.177170991897583,
      "learning_rate": 4.590085795996187e-05,
      "loss": 4.8708,
      "step": 3510
    },
    {
      "epoch": 0.09445116216579336,
      "grad_norm": 2.52181339263916,
      "learning_rate": 4.58940487539153e-05,
      "loss": 4.7624,
      "step": 3515
    },
    {
      "epoch": 0.09458551659277173,
      "grad_norm": 2.4385335445404053,
      "learning_rate": 4.588723954786872e-05,
      "loss": 4.8535,
      "step": 3520
    },
    {
      "epoch": 0.0947198710197501,
      "grad_norm": 2.4057161808013916,
      "learning_rate": 4.588043034182214e-05,
      "loss": 4.9165,
      "step": 3525
    },
    {
      "epoch": 0.09485422544672847,
      "grad_norm": 2.464326858520508,
      "learning_rate": 4.587362113577557e-05,
      "loss": 4.7491,
      "step": 3530
    },
    {
      "epoch": 0.09498857987370683,
      "grad_norm": 2.37044620513916,
      "learning_rate": 4.5866811929729e-05,
      "loss": 4.9403,
      "step": 3535
    },
    {
      "epoch": 0.0951229343006852,
      "grad_norm": 2.5464742183685303,
      "learning_rate": 4.586000272368242e-05,
      "loss": 4.8521,
      "step": 3540
    },
    {
      "epoch": 0.09525728872766358,
      "grad_norm": 2.3816168308258057,
      "learning_rate": 4.5853193517635845e-05,
      "loss": 5.0172,
      "step": 3545
    },
    {
      "epoch": 0.09539164315464195,
      "grad_norm": 2.3183231353759766,
      "learning_rate": 4.584638431158927e-05,
      "loss": 4.8062,
      "step": 3550
    },
    {
      "epoch": 0.09552599758162031,
      "grad_norm": 2.490726947784424,
      "learning_rate": 4.5839575105542696e-05,
      "loss": 5.0705,
      "step": 3555
    },
    {
      "epoch": 0.09566035200859868,
      "grad_norm": 2.7080209255218506,
      "learning_rate": 4.583276589949612e-05,
      "loss": 5.0083,
      "step": 3560
    },
    {
      "epoch": 0.09579470643557705,
      "grad_norm": 2.382826328277588,
      "learning_rate": 4.582595669344955e-05,
      "loss": 4.7448,
      "step": 3565
    },
    {
      "epoch": 0.09592906086255543,
      "grad_norm": 2.6137123107910156,
      "learning_rate": 4.581914748740297e-05,
      "loss": 4.8083,
      "step": 3570
    },
    {
      "epoch": 0.09606341528953379,
      "grad_norm": 2.500060796737671,
      "learning_rate": 4.58123382813564e-05,
      "loss": 4.9944,
      "step": 3575
    },
    {
      "epoch": 0.09619776971651216,
      "grad_norm": 2.218621015548706,
      "learning_rate": 4.580552907530982e-05,
      "loss": 4.9006,
      "step": 3580
    },
    {
      "epoch": 0.09633212414349053,
      "grad_norm": 2.533972978591919,
      "learning_rate": 4.579871986926324e-05,
      "loss": 4.808,
      "step": 3585
    },
    {
      "epoch": 0.0964664785704689,
      "grad_norm": 2.1555912494659424,
      "learning_rate": 4.579191066321667e-05,
      "loss": 4.8323,
      "step": 3590
    },
    {
      "epoch": 0.09660083299744726,
      "grad_norm": 2.311100721359253,
      "learning_rate": 4.57851014571701e-05,
      "loss": 4.8312,
      "step": 3595
    },
    {
      "epoch": 0.09673518742442563,
      "grad_norm": 2.2688076496124268,
      "learning_rate": 4.5778292251123524e-05,
      "loss": 4.8179,
      "step": 3600
    },
    {
      "epoch": 0.096869541851404,
      "grad_norm": 2.2784695625305176,
      "learning_rate": 4.5771483045076946e-05,
      "loss": 4.9883,
      "step": 3605
    },
    {
      "epoch": 0.09700389627838238,
      "grad_norm": 2.4393088817596436,
      "learning_rate": 4.576467383903037e-05,
      "loss": 4.8732,
      "step": 3610
    },
    {
      "epoch": 0.09713825070536074,
      "grad_norm": 2.2697343826293945,
      "learning_rate": 4.57578646329838e-05,
      "loss": 4.9243,
      "step": 3615
    },
    {
      "epoch": 0.09727260513233911,
      "grad_norm": 2.426361322402954,
      "learning_rate": 4.575105542693722e-05,
      "loss": 4.8421,
      "step": 3620
    },
    {
      "epoch": 0.09740695955931748,
      "grad_norm": 2.343950033187866,
      "learning_rate": 4.574424622089065e-05,
      "loss": 4.8981,
      "step": 3625
    },
    {
      "epoch": 0.09754131398629586,
      "grad_norm": 2.503652811050415,
      "learning_rate": 4.573743701484407e-05,
      "loss": 4.9175,
      "step": 3630
    },
    {
      "epoch": 0.09767566841327421,
      "grad_norm": 2.2795350551605225,
      "learning_rate": 4.573062780879749e-05,
      "loss": 4.8883,
      "step": 3635
    },
    {
      "epoch": 0.09781002284025259,
      "grad_norm": 2.2998039722442627,
      "learning_rate": 4.572381860275092e-05,
      "loss": 4.918,
      "step": 3640
    },
    {
      "epoch": 0.09794437726723096,
      "grad_norm": 2.231745481491089,
      "learning_rate": 4.5717009396704344e-05,
      "loss": 4.9396,
      "step": 3645
    },
    {
      "epoch": 0.09807873169420932,
      "grad_norm": 2.2835354804992676,
      "learning_rate": 4.5710200190657774e-05,
      "loss": 5.0009,
      "step": 3650
    },
    {
      "epoch": 0.09821308612118769,
      "grad_norm": 2.394118309020996,
      "learning_rate": 4.5703390984611196e-05,
      "loss": 4.9711,
      "step": 3655
    },
    {
      "epoch": 0.09834744054816606,
      "grad_norm": 2.459158182144165,
      "learning_rate": 4.5696581778564625e-05,
      "loss": 4.9808,
      "step": 3660
    },
    {
      "epoch": 0.09848179497514443,
      "grad_norm": 2.3846614360809326,
      "learning_rate": 4.568977257251805e-05,
      "loss": 4.8306,
      "step": 3665
    },
    {
      "epoch": 0.0986161494021228,
      "grad_norm": 2.3341453075408936,
      "learning_rate": 4.568296336647147e-05,
      "loss": 4.9436,
      "step": 3670
    },
    {
      "epoch": 0.09875050382910117,
      "grad_norm": 2.4033708572387695,
      "learning_rate": 4.567615416042489e-05,
      "loss": 4.9908,
      "step": 3675
    },
    {
      "epoch": 0.09888485825607954,
      "grad_norm": 2.154600143432617,
      "learning_rate": 4.566934495437832e-05,
      "loss": 4.6785,
      "step": 3680
    },
    {
      "epoch": 0.09901921268305791,
      "grad_norm": 2.5777153968811035,
      "learning_rate": 4.566253574833175e-05,
      "loss": 4.8747,
      "step": 3685
    },
    {
      "epoch": 0.09915356711003627,
      "grad_norm": 2.3845765590667725,
      "learning_rate": 4.565572654228517e-05,
      "loss": 4.9459,
      "step": 3690
    },
    {
      "epoch": 0.09928792153701464,
      "grad_norm": 2.464461326599121,
      "learning_rate": 4.5648917336238594e-05,
      "loss": 4.7985,
      "step": 3695
    },
    {
      "epoch": 0.09942227596399301,
      "grad_norm": 2.0905392169952393,
      "learning_rate": 4.564210813019202e-05,
      "loss": 4.9871,
      "step": 3700
    },
    {
      "epoch": 0.09955663039097139,
      "grad_norm": 2.4455697536468506,
      "learning_rate": 4.5635298924145446e-05,
      "loss": 4.9849,
      "step": 3705
    },
    {
      "epoch": 0.09969098481794975,
      "grad_norm": 2.5106849670410156,
      "learning_rate": 4.5628489718098875e-05,
      "loss": 4.8517,
      "step": 3710
    },
    {
      "epoch": 0.09982533924492812,
      "grad_norm": 2.6265969276428223,
      "learning_rate": 4.56216805120523e-05,
      "loss": 4.81,
      "step": 3715
    },
    {
      "epoch": 0.09995969367190649,
      "grad_norm": 2.0911262035369873,
      "learning_rate": 4.5614871306005726e-05,
      "loss": 4.8222,
      "step": 3720
    },
    {
      "epoch": 0.10009404809888486,
      "grad_norm": 2.6245334148406982,
      "learning_rate": 4.560806209995915e-05,
      "loss": 4.8013,
      "step": 3725
    },
    {
      "epoch": 0.10022840252586322,
      "grad_norm": 2.5066633224487305,
      "learning_rate": 4.560125289391257e-05,
      "loss": 4.8526,
      "step": 3730
    },
    {
      "epoch": 0.1003627569528416,
      "grad_norm": 2.2056989669799805,
      "learning_rate": 4.559444368786599e-05,
      "loss": 4.9282,
      "step": 3735
    },
    {
      "epoch": 0.10049711137981997,
      "grad_norm": 2.328695297241211,
      "learning_rate": 4.558763448181942e-05,
      "loss": 4.8001,
      "step": 3740
    },
    {
      "epoch": 0.10063146580679834,
      "grad_norm": 2.262921094894409,
      "learning_rate": 4.558082527577285e-05,
      "loss": 4.7917,
      "step": 3745
    },
    {
      "epoch": 0.1007658202337767,
      "grad_norm": 2.4201860427856445,
      "learning_rate": 4.557401606972627e-05,
      "loss": 4.7839,
      "step": 3750
    },
    {
      "epoch": 0.10090017466075507,
      "grad_norm": 2.142925977706909,
      "learning_rate": 4.5567206863679695e-05,
      "loss": 4.7718,
      "step": 3755
    },
    {
      "epoch": 0.10103452908773344,
      "grad_norm": 2.5661582946777344,
      "learning_rate": 4.5560397657633124e-05,
      "loss": 4.8784,
      "step": 3760
    },
    {
      "epoch": 0.10116888351471182,
      "grad_norm": 2.518620491027832,
      "learning_rate": 4.555358845158655e-05,
      "loss": 5.0041,
      "step": 3765
    },
    {
      "epoch": 0.10130323794169017,
      "grad_norm": 2.145602226257324,
      "learning_rate": 4.554677924553997e-05,
      "loss": 4.8942,
      "step": 3770
    },
    {
      "epoch": 0.10143759236866855,
      "grad_norm": 2.5080015659332275,
      "learning_rate": 4.55399700394934e-05,
      "loss": 4.9117,
      "step": 3775
    },
    {
      "epoch": 0.10157194679564692,
      "grad_norm": 2.5845413208007812,
      "learning_rate": 4.553316083344682e-05,
      "loss": 4.8026,
      "step": 3780
    },
    {
      "epoch": 0.10170630122262529,
      "grad_norm": 2.4090616703033447,
      "learning_rate": 4.552635162740025e-05,
      "loss": 4.8893,
      "step": 3785
    },
    {
      "epoch": 0.10184065564960365,
      "grad_norm": 2.2573182582855225,
      "learning_rate": 4.551954242135367e-05,
      "loss": 4.9695,
      "step": 3790
    },
    {
      "epoch": 0.10197501007658202,
      "grad_norm": 2.3865768909454346,
      "learning_rate": 4.5512733215307094e-05,
      "loss": 4.868,
      "step": 3795
    },
    {
      "epoch": 0.1021093645035604,
      "grad_norm": 2.2999753952026367,
      "learning_rate": 4.550592400926052e-05,
      "loss": 4.8374,
      "step": 3800
    },
    {
      "epoch": 0.10224371893053875,
      "grad_norm": 2.3399412631988525,
      "learning_rate": 4.549911480321395e-05,
      "loss": 4.8755,
      "step": 3805
    },
    {
      "epoch": 0.10237807335751713,
      "grad_norm": 2.1934211254119873,
      "learning_rate": 4.5492305597167374e-05,
      "loss": 4.8114,
      "step": 3810
    },
    {
      "epoch": 0.1025124277844955,
      "grad_norm": 2.4351773262023926,
      "learning_rate": 4.5485496391120796e-05,
      "loss": 4.8878,
      "step": 3815
    },
    {
      "epoch": 0.10264678221147387,
      "grad_norm": 2.1701009273529053,
      "learning_rate": 4.547868718507422e-05,
      "loss": 4.8687,
      "step": 3820
    },
    {
      "epoch": 0.10278113663845223,
      "grad_norm": 2.4852359294891357,
      "learning_rate": 4.547187797902765e-05,
      "loss": 4.9959,
      "step": 3825
    },
    {
      "epoch": 0.1029154910654306,
      "grad_norm": 2.4492363929748535,
      "learning_rate": 4.546506877298107e-05,
      "loss": 4.9215,
      "step": 3830
    },
    {
      "epoch": 0.10304984549240898,
      "grad_norm": 2.2819671630859375,
      "learning_rate": 4.54582595669345e-05,
      "loss": 4.9463,
      "step": 3835
    },
    {
      "epoch": 0.10318419991938735,
      "grad_norm": 2.2214770317077637,
      "learning_rate": 4.545145036088792e-05,
      "loss": 4.8956,
      "step": 3840
    },
    {
      "epoch": 0.1033185543463657,
      "grad_norm": 2.302542209625244,
      "learning_rate": 4.544464115484135e-05,
      "loss": 4.7953,
      "step": 3845
    },
    {
      "epoch": 0.10345290877334408,
      "grad_norm": 2.407921075820923,
      "learning_rate": 4.543783194879477e-05,
      "loss": 4.806,
      "step": 3850
    },
    {
      "epoch": 0.10358726320032245,
      "grad_norm": 2.5172712802886963,
      "learning_rate": 4.5431022742748195e-05,
      "loss": 4.8893,
      "step": 3855
    },
    {
      "epoch": 0.10372161762730082,
      "grad_norm": 2.548431158065796,
      "learning_rate": 4.5424213536701624e-05,
      "loss": 4.6952,
      "step": 3860
    },
    {
      "epoch": 0.10385597205427918,
      "grad_norm": 2.088240146636963,
      "learning_rate": 4.541740433065505e-05,
      "loss": 4.7694,
      "step": 3865
    },
    {
      "epoch": 0.10399032648125756,
      "grad_norm": 2.1839518547058105,
      "learning_rate": 4.5410595124608475e-05,
      "loss": 4.8916,
      "step": 3870
    },
    {
      "epoch": 0.10412468090823593,
      "grad_norm": 2.1736936569213867,
      "learning_rate": 4.54037859185619e-05,
      "loss": 4.73,
      "step": 3875
    },
    {
      "epoch": 0.1042590353352143,
      "grad_norm": 2.3348512649536133,
      "learning_rate": 4.539697671251532e-05,
      "loss": 4.8301,
      "step": 3880
    },
    {
      "epoch": 0.10439338976219266,
      "grad_norm": 2.2691328525543213,
      "learning_rate": 4.539016750646875e-05,
      "loss": 4.8908,
      "step": 3885
    },
    {
      "epoch": 0.10452774418917103,
      "grad_norm": 2.3222343921661377,
      "learning_rate": 4.538335830042217e-05,
      "loss": 4.7588,
      "step": 3890
    },
    {
      "epoch": 0.1046620986161494,
      "grad_norm": 2.2783796787261963,
      "learning_rate": 4.53765490943756e-05,
      "loss": 4.8291,
      "step": 3895
    },
    {
      "epoch": 0.10479645304312778,
      "grad_norm": 2.2894301414489746,
      "learning_rate": 4.536973988832902e-05,
      "loss": 4.9146,
      "step": 3900
    },
    {
      "epoch": 0.10493080747010614,
      "grad_norm": 2.2368619441986084,
      "learning_rate": 4.536293068228245e-05,
      "loss": 4.7488,
      "step": 3905
    },
    {
      "epoch": 0.10506516189708451,
      "grad_norm": 2.4413812160491943,
      "learning_rate": 4.5356121476235874e-05,
      "loss": 4.9104,
      "step": 3910
    },
    {
      "epoch": 0.10519951632406288,
      "grad_norm": 2.445951223373413,
      "learning_rate": 4.5349312270189296e-05,
      "loss": 4.7909,
      "step": 3915
    },
    {
      "epoch": 0.10533387075104125,
      "grad_norm": 2.5580689907073975,
      "learning_rate": 4.5342503064142725e-05,
      "loss": 4.9032,
      "step": 3920
    },
    {
      "epoch": 0.10546822517801961,
      "grad_norm": 2.387916326522827,
      "learning_rate": 4.533569385809615e-05,
      "loss": 4.992,
      "step": 3925
    },
    {
      "epoch": 0.10560257960499798,
      "grad_norm": 2.3785805702209473,
      "learning_rate": 4.5328884652049576e-05,
      "loss": 4.8345,
      "step": 3930
    },
    {
      "epoch": 0.10573693403197636,
      "grad_norm": 2.3815479278564453,
      "learning_rate": 4.5322075446003e-05,
      "loss": 4.8454,
      "step": 3935
    },
    {
      "epoch": 0.10587128845895473,
      "grad_norm": 2.535463333129883,
      "learning_rate": 4.531526623995642e-05,
      "loss": 4.9197,
      "step": 3940
    },
    {
      "epoch": 0.10600564288593309,
      "grad_norm": 2.25718092918396,
      "learning_rate": 4.530845703390984e-05,
      "loss": 4.7668,
      "step": 3945
    },
    {
      "epoch": 0.10613999731291146,
      "grad_norm": 2.453803777694702,
      "learning_rate": 4.530164782786327e-05,
      "loss": 4.9305,
      "step": 3950
    },
    {
      "epoch": 0.10627435173988983,
      "grad_norm": 2.783095598220825,
      "learning_rate": 4.52948386218167e-05,
      "loss": 4.8045,
      "step": 3955
    },
    {
      "epoch": 0.10640870616686819,
      "grad_norm": 2.3370118141174316,
      "learning_rate": 4.5288029415770124e-05,
      "loss": 4.7908,
      "step": 3960
    },
    {
      "epoch": 0.10654306059384656,
      "grad_norm": 2.4311699867248535,
      "learning_rate": 4.5281220209723546e-05,
      "loss": 5.0291,
      "step": 3965
    },
    {
      "epoch": 0.10667741502082494,
      "grad_norm": 2.5788986682891846,
      "learning_rate": 4.5274411003676975e-05,
      "loss": 4.8295,
      "step": 3970
    },
    {
      "epoch": 0.10681176944780331,
      "grad_norm": 2.515810251235962,
      "learning_rate": 4.52676017976304e-05,
      "loss": 4.9363,
      "step": 3975
    },
    {
      "epoch": 0.10694612387478167,
      "grad_norm": 2.2663097381591797,
      "learning_rate": 4.526079259158382e-05,
      "loss": 4.7726,
      "step": 3980
    },
    {
      "epoch": 0.10708047830176004,
      "grad_norm": 2.4607527256011963,
      "learning_rate": 4.525398338553725e-05,
      "loss": 4.7594,
      "step": 3985
    },
    {
      "epoch": 0.10721483272873841,
      "grad_norm": 2.100602865219116,
      "learning_rate": 4.524717417949068e-05,
      "loss": 4.8253,
      "step": 3990
    },
    {
      "epoch": 0.10734918715571679,
      "grad_norm": 2.069812536239624,
      "learning_rate": 4.52403649734441e-05,
      "loss": 4.9251,
      "step": 3995
    },
    {
      "epoch": 0.10748354158269514,
      "grad_norm": 2.5128419399261475,
      "learning_rate": 4.523355576739752e-05,
      "loss": 4.8984,
      "step": 4000
    },
    {
      "epoch": 0.10761789600967352,
      "grad_norm": 2.25811767578125,
      "learning_rate": 4.5226746561350944e-05,
      "loss": 4.7767,
      "step": 4005
    },
    {
      "epoch": 0.10775225043665189,
      "grad_norm": 2.6393041610717773,
      "learning_rate": 4.5219937355304373e-05,
      "loss": 4.8818,
      "step": 4010
    },
    {
      "epoch": 0.10788660486363026,
      "grad_norm": 2.3550100326538086,
      "learning_rate": 4.52131281492578e-05,
      "loss": 4.8485,
      "step": 4015
    },
    {
      "epoch": 0.10802095929060862,
      "grad_norm": 2.305943250656128,
      "learning_rate": 4.5206318943211225e-05,
      "loss": 4.9345,
      "step": 4020
    },
    {
      "epoch": 0.10815531371758699,
      "grad_norm": 2.2776548862457275,
      "learning_rate": 4.519950973716465e-05,
      "loss": 4.7306,
      "step": 4025
    },
    {
      "epoch": 0.10828966814456537,
      "grad_norm": 2.351945161819458,
      "learning_rate": 4.5192700531118076e-05,
      "loss": 4.7719,
      "step": 4030
    },
    {
      "epoch": 0.10842402257154374,
      "grad_norm": 2.304865837097168,
      "learning_rate": 4.51858913250715e-05,
      "loss": 4.7545,
      "step": 4035
    },
    {
      "epoch": 0.1085583769985221,
      "grad_norm": 2.3425676822662354,
      "learning_rate": 4.517908211902492e-05,
      "loss": 4.7759,
      "step": 4040
    },
    {
      "epoch": 0.10869273142550047,
      "grad_norm": 2.3676183223724365,
      "learning_rate": 4.517227291297835e-05,
      "loss": 4.9045,
      "step": 4045
    },
    {
      "epoch": 0.10882708585247884,
      "grad_norm": 2.3830666542053223,
      "learning_rate": 4.516546370693178e-05,
      "loss": 4.9098,
      "step": 4050
    },
    {
      "epoch": 0.10896144027945721,
      "grad_norm": 2.481192111968994,
      "learning_rate": 4.51586545008852e-05,
      "loss": 4.7721,
      "step": 4055
    },
    {
      "epoch": 0.10909579470643557,
      "grad_norm": 2.399505138397217,
      "learning_rate": 4.515184529483862e-05,
      "loss": 4.7054,
      "step": 4060
    },
    {
      "epoch": 0.10923014913341395,
      "grad_norm": 2.340904951095581,
      "learning_rate": 4.5145036088792045e-05,
      "loss": 4.9621,
      "step": 4065
    },
    {
      "epoch": 0.10936450356039232,
      "grad_norm": 2.3511340618133545,
      "learning_rate": 4.5138226882745475e-05,
      "loss": 4.771,
      "step": 4070
    },
    {
      "epoch": 0.10949885798737069,
      "grad_norm": 2.4745922088623047,
      "learning_rate": 4.5131417676698904e-05,
      "loss": 4.7935,
      "step": 4075
    },
    {
      "epoch": 0.10963321241434905,
      "grad_norm": 2.42933988571167,
      "learning_rate": 4.5124608470652326e-05,
      "loss": 4.9054,
      "step": 4080
    },
    {
      "epoch": 0.10976756684132742,
      "grad_norm": 2.2317187786102295,
      "learning_rate": 4.511779926460575e-05,
      "loss": 4.7962,
      "step": 4085
    },
    {
      "epoch": 0.1099019212683058,
      "grad_norm": 2.3949458599090576,
      "learning_rate": 4.511099005855917e-05,
      "loss": 4.7295,
      "step": 4090
    },
    {
      "epoch": 0.11003627569528417,
      "grad_norm": 2.3402559757232666,
      "learning_rate": 4.51041808525126e-05,
      "loss": 4.831,
      "step": 4095
    },
    {
      "epoch": 0.11017063012226253,
      "grad_norm": 2.512261390686035,
      "learning_rate": 4.509737164646602e-05,
      "loss": 4.847,
      "step": 4100
    },
    {
      "epoch": 0.1103049845492409,
      "grad_norm": 2.3977127075195312,
      "learning_rate": 4.509056244041945e-05,
      "loss": 4.9364,
      "step": 4105
    },
    {
      "epoch": 0.11043933897621927,
      "grad_norm": 2.0969760417938232,
      "learning_rate": 4.508375323437287e-05,
      "loss": 4.7632,
      "step": 4110
    },
    {
      "epoch": 0.11057369340319763,
      "grad_norm": 2.3379251956939697,
      "learning_rate": 4.50769440283263e-05,
      "loss": 4.8934,
      "step": 4115
    },
    {
      "epoch": 0.110708047830176,
      "grad_norm": 2.295227289199829,
      "learning_rate": 4.5070134822279724e-05,
      "loss": 4.8004,
      "step": 4120
    },
    {
      "epoch": 0.11084240225715437,
      "grad_norm": 2.161325454711914,
      "learning_rate": 4.5063325616233147e-05,
      "loss": 4.8489,
      "step": 4125
    },
    {
      "epoch": 0.11097675668413275,
      "grad_norm": 2.4868967533111572,
      "learning_rate": 4.505651641018657e-05,
      "loss": 4.6796,
      "step": 4130
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 2.30232310295105,
      "learning_rate": 4.5049707204140005e-05,
      "loss": 4.7677,
      "step": 4135
    },
    {
      "epoch": 0.11124546553808948,
      "grad_norm": 2.3230373859405518,
      "learning_rate": 4.504289799809343e-05,
      "loss": 4.8115,
      "step": 4140
    },
    {
      "epoch": 0.11137981996506785,
      "grad_norm": 2.38686466217041,
      "learning_rate": 4.503608879204685e-05,
      "loss": 4.7916,
      "step": 4145
    },
    {
      "epoch": 0.11151417439204622,
      "grad_norm": 2.492284059524536,
      "learning_rate": 4.502927958600027e-05,
      "loss": 4.7954,
      "step": 4150
    },
    {
      "epoch": 0.11164852881902458,
      "grad_norm": 2.453956127166748,
      "learning_rate": 4.50224703799537e-05,
      "loss": 4.7882,
      "step": 4155
    },
    {
      "epoch": 0.11178288324600295,
      "grad_norm": 2.505645275115967,
      "learning_rate": 4.501566117390712e-05,
      "loss": 4.8118,
      "step": 4160
    },
    {
      "epoch": 0.11191723767298133,
      "grad_norm": 2.40903639793396,
      "learning_rate": 4.500885196786055e-05,
      "loss": 4.8932,
      "step": 4165
    },
    {
      "epoch": 0.1120515920999597,
      "grad_norm": 2.4800972938537598,
      "learning_rate": 4.5002042761813974e-05,
      "loss": 4.7828,
      "step": 4170
    },
    {
      "epoch": 0.11218594652693806,
      "grad_norm": 2.289245843887329,
      "learning_rate": 4.49952335557674e-05,
      "loss": 4.9556,
      "step": 4175
    },
    {
      "epoch": 0.11232030095391643,
      "grad_norm": 2.3668859004974365,
      "learning_rate": 4.4988424349720825e-05,
      "loss": 4.8433,
      "step": 4180
    },
    {
      "epoch": 0.1124546553808948,
      "grad_norm": 2.5421972274780273,
      "learning_rate": 4.498161514367425e-05,
      "loss": 4.7374,
      "step": 4185
    },
    {
      "epoch": 0.11258900980787317,
      "grad_norm": 2.3949403762817383,
      "learning_rate": 4.497480593762767e-05,
      "loss": 4.8768,
      "step": 4190
    },
    {
      "epoch": 0.11272336423485153,
      "grad_norm": 2.4453353881835938,
      "learning_rate": 4.49679967315811e-05,
      "loss": 4.7807,
      "step": 4195
    },
    {
      "epoch": 0.1128577186618299,
      "grad_norm": 2.5284366607666016,
      "learning_rate": 4.496118752553453e-05,
      "loss": 4.8374,
      "step": 4200
    },
    {
      "epoch": 0.11299207308880828,
      "grad_norm": 2.2510130405426025,
      "learning_rate": 4.495437831948795e-05,
      "loss": 4.8367,
      "step": 4205
    },
    {
      "epoch": 0.11312642751578665,
      "grad_norm": 2.456310749053955,
      "learning_rate": 4.494756911344137e-05,
      "loss": 4.9486,
      "step": 4210
    },
    {
      "epoch": 0.11326078194276501,
      "grad_norm": 2.4718942642211914,
      "learning_rate": 4.49407599073948e-05,
      "loss": 4.7642,
      "step": 4215
    },
    {
      "epoch": 0.11339513636974338,
      "grad_norm": 2.5681488513946533,
      "learning_rate": 4.4933950701348224e-05,
      "loss": 4.7475,
      "step": 4220
    },
    {
      "epoch": 0.11352949079672175,
      "grad_norm": 2.713550090789795,
      "learning_rate": 4.492714149530165e-05,
      "loss": 4.9712,
      "step": 4225
    },
    {
      "epoch": 0.11366384522370013,
      "grad_norm": 2.3059887886047363,
      "learning_rate": 4.4920332289255075e-05,
      "loss": 4.8545,
      "step": 4230
    },
    {
      "epoch": 0.11379819965067849,
      "grad_norm": 2.271151304244995,
      "learning_rate": 4.49135230832085e-05,
      "loss": 4.9656,
      "step": 4235
    },
    {
      "epoch": 0.11393255407765686,
      "grad_norm": 2.380279064178467,
      "learning_rate": 4.4906713877161927e-05,
      "loss": 4.8414,
      "step": 4240
    },
    {
      "epoch": 0.11406690850463523,
      "grad_norm": 2.6072587966918945,
      "learning_rate": 4.489990467111535e-05,
      "loss": 4.7965,
      "step": 4245
    },
    {
      "epoch": 0.1142012629316136,
      "grad_norm": 2.4508564472198486,
      "learning_rate": 4.489309546506877e-05,
      "loss": 4.8604,
      "step": 4250
    },
    {
      "epoch": 0.11433561735859196,
      "grad_norm": 2.3962242603302,
      "learning_rate": 4.48862862590222e-05,
      "loss": 4.7706,
      "step": 4255
    },
    {
      "epoch": 0.11446997178557033,
      "grad_norm": 2.6643543243408203,
      "learning_rate": 4.487947705297563e-05,
      "loss": 4.8796,
      "step": 4260
    },
    {
      "epoch": 0.11460432621254871,
      "grad_norm": 2.2862443923950195,
      "learning_rate": 4.487266784692905e-05,
      "loss": 4.8573,
      "step": 4265
    },
    {
      "epoch": 0.11473868063952707,
      "grad_norm": 2.2548675537109375,
      "learning_rate": 4.4865858640882474e-05,
      "loss": 4.7709,
      "step": 4270
    },
    {
      "epoch": 0.11487303506650544,
      "grad_norm": 2.393411874771118,
      "learning_rate": 4.4859049434835896e-05,
      "loss": 4.8663,
      "step": 4275
    },
    {
      "epoch": 0.11500738949348381,
      "grad_norm": 2.3601062297821045,
      "learning_rate": 4.4852240228789325e-05,
      "loss": 4.8592,
      "step": 4280
    },
    {
      "epoch": 0.11514174392046218,
      "grad_norm": 2.2856903076171875,
      "learning_rate": 4.4845431022742754e-05,
      "loss": 4.8743,
      "step": 4285
    },
    {
      "epoch": 0.11527609834744054,
      "grad_norm": 2.2420032024383545,
      "learning_rate": 4.4838621816696176e-05,
      "loss": 4.8173,
      "step": 4290
    },
    {
      "epoch": 0.11541045277441891,
      "grad_norm": 2.34596848487854,
      "learning_rate": 4.48318126106496e-05,
      "loss": 4.9407,
      "step": 4295
    },
    {
      "epoch": 0.11554480720139729,
      "grad_norm": 2.249833583831787,
      "learning_rate": 4.482500340460303e-05,
      "loss": 4.7955,
      "step": 4300
    },
    {
      "epoch": 0.11567916162837566,
      "grad_norm": 2.2568256855010986,
      "learning_rate": 4.481819419855645e-05,
      "loss": 4.7546,
      "step": 4305
    },
    {
      "epoch": 0.11581351605535402,
      "grad_norm": 2.5375142097473145,
      "learning_rate": 4.481138499250987e-05,
      "loss": 4.8389,
      "step": 4310
    },
    {
      "epoch": 0.11594787048233239,
      "grad_norm": 2.4763433933258057,
      "learning_rate": 4.48045757864633e-05,
      "loss": 4.8254,
      "step": 4315
    },
    {
      "epoch": 0.11608222490931076,
      "grad_norm": 2.442808151245117,
      "learning_rate": 4.479776658041673e-05,
      "loss": 4.7394,
      "step": 4320
    },
    {
      "epoch": 0.11621657933628914,
      "grad_norm": 2.290862798690796,
      "learning_rate": 4.479095737437015e-05,
      "loss": 4.819,
      "step": 4325
    },
    {
      "epoch": 0.1163509337632675,
      "grad_norm": 2.581564426422119,
      "learning_rate": 4.4784148168323575e-05,
      "loss": 4.7278,
      "step": 4330
    },
    {
      "epoch": 0.11648528819024587,
      "grad_norm": 2.193546772003174,
      "learning_rate": 4.4777338962277e-05,
      "loss": 4.7554,
      "step": 4335
    },
    {
      "epoch": 0.11661964261722424,
      "grad_norm": 2.334792375564575,
      "learning_rate": 4.4770529756230426e-05,
      "loss": 4.8878,
      "step": 4340
    },
    {
      "epoch": 0.11675399704420261,
      "grad_norm": 2.1534948348999023,
      "learning_rate": 4.4763720550183855e-05,
      "loss": 4.8333,
      "step": 4345
    },
    {
      "epoch": 0.11688835147118097,
      "grad_norm": 2.372164011001587,
      "learning_rate": 4.475691134413728e-05,
      "loss": 4.8058,
      "step": 4350
    },
    {
      "epoch": 0.11702270589815934,
      "grad_norm": 2.6233463287353516,
      "learning_rate": 4.47501021380907e-05,
      "loss": 4.8134,
      "step": 4355
    },
    {
      "epoch": 0.11715706032513772,
      "grad_norm": 2.5761067867279053,
      "learning_rate": 4.474329293204413e-05,
      "loss": 4.7723,
      "step": 4360
    },
    {
      "epoch": 0.11729141475211609,
      "grad_norm": 2.2746036052703857,
      "learning_rate": 4.473648372599755e-05,
      "loss": 4.8873,
      "step": 4365
    },
    {
      "epoch": 0.11742576917909445,
      "grad_norm": 2.214812994003296,
      "learning_rate": 4.472967451995097e-05,
      "loss": 4.8907,
      "step": 4370
    },
    {
      "epoch": 0.11756012360607282,
      "grad_norm": 2.3953659534454346,
      "learning_rate": 4.47228653139044e-05,
      "loss": 4.6522,
      "step": 4375
    },
    {
      "epoch": 0.11769447803305119,
      "grad_norm": 2.6663784980773926,
      "learning_rate": 4.4716056107857825e-05,
      "loss": 4.8405,
      "step": 4380
    },
    {
      "epoch": 0.11782883246002956,
      "grad_norm": 2.4421095848083496,
      "learning_rate": 4.4709246901811254e-05,
      "loss": 4.7374,
      "step": 4385
    },
    {
      "epoch": 0.11796318688700792,
      "grad_norm": 2.476816415786743,
      "learning_rate": 4.4702437695764676e-05,
      "loss": 4.6978,
      "step": 4390
    },
    {
      "epoch": 0.1180975413139863,
      "grad_norm": 2.278705358505249,
      "learning_rate": 4.46956284897181e-05,
      "loss": 4.8287,
      "step": 4395
    },
    {
      "epoch": 0.11823189574096467,
      "grad_norm": 2.3131260871887207,
      "learning_rate": 4.468881928367152e-05,
      "loss": 4.6929,
      "step": 4400
    },
    {
      "epoch": 0.11836625016794304,
      "grad_norm": 2.562412977218628,
      "learning_rate": 4.468201007762495e-05,
      "loss": 4.7091,
      "step": 4405
    },
    {
      "epoch": 0.1185006045949214,
      "grad_norm": 2.254849910736084,
      "learning_rate": 4.467520087157838e-05,
      "loss": 4.7875,
      "step": 4410
    },
    {
      "epoch": 0.11863495902189977,
      "grad_norm": 2.204624891281128,
      "learning_rate": 4.46683916655318e-05,
      "loss": 4.692,
      "step": 4415
    },
    {
      "epoch": 0.11876931344887814,
      "grad_norm": 2.2432289123535156,
      "learning_rate": 4.466158245948522e-05,
      "loss": 4.8274,
      "step": 4420
    },
    {
      "epoch": 0.1189036678758565,
      "grad_norm": 2.5432636737823486,
      "learning_rate": 4.465477325343865e-05,
      "loss": 4.9673,
      "step": 4425
    },
    {
      "epoch": 0.11903802230283488,
      "grad_norm": 2.230104923248291,
      "learning_rate": 4.4647964047392074e-05,
      "loss": 4.7719,
      "step": 4430
    },
    {
      "epoch": 0.11917237672981325,
      "grad_norm": 2.272420644760132,
      "learning_rate": 4.4641154841345503e-05,
      "loss": 4.7614,
      "step": 4435
    },
    {
      "epoch": 0.11930673115679162,
      "grad_norm": 2.0710220336914062,
      "learning_rate": 4.4634345635298926e-05,
      "loss": 4.7047,
      "step": 4440
    },
    {
      "epoch": 0.11944108558376998,
      "grad_norm": 2.415905237197876,
      "learning_rate": 4.4627536429252355e-05,
      "loss": 4.9128,
      "step": 4445
    },
    {
      "epoch": 0.11957544001074835,
      "grad_norm": 2.3551747798919678,
      "learning_rate": 4.462072722320578e-05,
      "loss": 4.7327,
      "step": 4450
    },
    {
      "epoch": 0.11970979443772672,
      "grad_norm": 2.4013113975524902,
      "learning_rate": 4.46139180171592e-05,
      "loss": 4.7552,
      "step": 4455
    },
    {
      "epoch": 0.1198441488647051,
      "grad_norm": 2.282001495361328,
      "learning_rate": 4.460710881111262e-05,
      "loss": 4.7584,
      "step": 4460
    },
    {
      "epoch": 0.11997850329168346,
      "grad_norm": 2.4643735885620117,
      "learning_rate": 4.460029960506605e-05,
      "loss": 4.6315,
      "step": 4465
    },
    {
      "epoch": 0.12011285771866183,
      "grad_norm": 2.58450984954834,
      "learning_rate": 4.459349039901948e-05,
      "loss": 4.7834,
      "step": 4470
    },
    {
      "epoch": 0.1202472121456402,
      "grad_norm": 2.2559621334075928,
      "learning_rate": 4.45866811929729e-05,
      "loss": 4.8369,
      "step": 4475
    },
    {
      "epoch": 0.12038156657261857,
      "grad_norm": 2.3033199310302734,
      "learning_rate": 4.4579871986926324e-05,
      "loss": 4.6834,
      "step": 4480
    },
    {
      "epoch": 0.12051592099959693,
      "grad_norm": 2.2106590270996094,
      "learning_rate": 4.457306278087975e-05,
      "loss": 4.9238,
      "step": 4485
    },
    {
      "epoch": 0.1206502754265753,
      "grad_norm": 2.456451892852783,
      "learning_rate": 4.4566253574833176e-05,
      "loss": 4.832,
      "step": 4490
    },
    {
      "epoch": 0.12078462985355368,
      "grad_norm": 2.2797577381134033,
      "learning_rate": 4.4559444368786605e-05,
      "loss": 4.7856,
      "step": 4495
    },
    {
      "epoch": 0.12091898428053205,
      "grad_norm": 2.378916025161743,
      "learning_rate": 4.455263516274003e-05,
      "loss": 4.8582,
      "step": 4500
    },
    {
      "epoch": 0.12105333870751041,
      "grad_norm": 2.538640260696411,
      "learning_rate": 4.4545825956693456e-05,
      "loss": 4.7919,
      "step": 4505
    },
    {
      "epoch": 0.12118769313448878,
      "grad_norm": 2.3927602767944336,
      "learning_rate": 4.453901675064688e-05,
      "loss": 4.8186,
      "step": 4510
    },
    {
      "epoch": 0.12132204756146715,
      "grad_norm": 2.429544448852539,
      "learning_rate": 4.45322075446003e-05,
      "loss": 4.8529,
      "step": 4515
    },
    {
      "epoch": 0.12145640198844553,
      "grad_norm": 2.1288082599639893,
      "learning_rate": 4.452539833855372e-05,
      "loss": 4.7527,
      "step": 4520
    },
    {
      "epoch": 0.12159075641542388,
      "grad_norm": 2.492410659790039,
      "learning_rate": 4.451858913250715e-05,
      "loss": 4.8259,
      "step": 4525
    },
    {
      "epoch": 0.12172511084240226,
      "grad_norm": 2.3878021240234375,
      "learning_rate": 4.451177992646058e-05,
      "loss": 4.7131,
      "step": 4530
    },
    {
      "epoch": 0.12185946526938063,
      "grad_norm": 2.3997464179992676,
      "learning_rate": 4.4504970720414e-05,
      "loss": 4.8381,
      "step": 4535
    },
    {
      "epoch": 0.121993819696359,
      "grad_norm": 2.2738037109375,
      "learning_rate": 4.4498161514367425e-05,
      "loss": 4.6911,
      "step": 4540
    },
    {
      "epoch": 0.12212817412333736,
      "grad_norm": 2.4462459087371826,
      "learning_rate": 4.449135230832085e-05,
      "loss": 4.6052,
      "step": 4545
    },
    {
      "epoch": 0.12226252855031573,
      "grad_norm": 2.23532772064209,
      "learning_rate": 4.448454310227428e-05,
      "loss": 4.7455,
      "step": 4550
    },
    {
      "epoch": 0.1223968829772941,
      "grad_norm": 2.6852872371673584,
      "learning_rate": 4.4477733896227706e-05,
      "loss": 4.7792,
      "step": 4555
    },
    {
      "epoch": 0.12253123740427246,
      "grad_norm": 2.169888734817505,
      "learning_rate": 4.447092469018113e-05,
      "loss": 4.7512,
      "step": 4560
    },
    {
      "epoch": 0.12266559183125084,
      "grad_norm": 2.096015214920044,
      "learning_rate": 4.446411548413455e-05,
      "loss": 4.7795,
      "step": 4565
    },
    {
      "epoch": 0.12279994625822921,
      "grad_norm": 2.611541509628296,
      "learning_rate": 4.445730627808798e-05,
      "loss": 4.6357,
      "step": 4570
    },
    {
      "epoch": 0.12293430068520758,
      "grad_norm": 2.2759056091308594,
      "learning_rate": 4.44504970720414e-05,
      "loss": 4.6561,
      "step": 4575
    },
    {
      "epoch": 0.12306865511218594,
      "grad_norm": 2.3145864009857178,
      "learning_rate": 4.4443687865994824e-05,
      "loss": 4.8076,
      "step": 4580
    },
    {
      "epoch": 0.12320300953916431,
      "grad_norm": 2.4229812622070312,
      "learning_rate": 4.443687865994825e-05,
      "loss": 4.8465,
      "step": 4585
    },
    {
      "epoch": 0.12333736396614269,
      "grad_norm": 2.2861647605895996,
      "learning_rate": 4.443006945390168e-05,
      "loss": 4.8535,
      "step": 4590
    },
    {
      "epoch": 0.12347171839312106,
      "grad_norm": 2.146202564239502,
      "learning_rate": 4.4423260247855104e-05,
      "loss": 4.749,
      "step": 4595
    },
    {
      "epoch": 0.12360607282009942,
      "grad_norm": 2.3462462425231934,
      "learning_rate": 4.4416451041808526e-05,
      "loss": 4.784,
      "step": 4600
    },
    {
      "epoch": 0.12374042724707779,
      "grad_norm": 2.518326997756958,
      "learning_rate": 4.440964183576195e-05,
      "loss": 4.8874,
      "step": 4605
    },
    {
      "epoch": 0.12387478167405616,
      "grad_norm": 2.1689605712890625,
      "learning_rate": 4.440283262971538e-05,
      "loss": 4.7882,
      "step": 4610
    },
    {
      "epoch": 0.12400913610103453,
      "grad_norm": 2.4828147888183594,
      "learning_rate": 4.43960234236688e-05,
      "loss": 4.6951,
      "step": 4615
    },
    {
      "epoch": 0.12414349052801289,
      "grad_norm": 2.474313497543335,
      "learning_rate": 4.438921421762223e-05,
      "loss": 4.7948,
      "step": 4620
    },
    {
      "epoch": 0.12427784495499127,
      "grad_norm": 2.473074436187744,
      "learning_rate": 4.438240501157565e-05,
      "loss": 4.8089,
      "step": 4625
    },
    {
      "epoch": 0.12441219938196964,
      "grad_norm": 2.515138864517212,
      "learning_rate": 4.437559580552908e-05,
      "loss": 4.6175,
      "step": 4630
    },
    {
      "epoch": 0.12454655380894801,
      "grad_norm": 2.4840569496154785,
      "learning_rate": 4.43687865994825e-05,
      "loss": 4.7728,
      "step": 4635
    },
    {
      "epoch": 0.12468090823592637,
      "grad_norm": 2.4376375675201416,
      "learning_rate": 4.4361977393435925e-05,
      "loss": 4.7834,
      "step": 4640
    },
    {
      "epoch": 0.12481526266290474,
      "grad_norm": 2.5587005615234375,
      "learning_rate": 4.4355168187389354e-05,
      "loss": 4.7969,
      "step": 4645
    },
    {
      "epoch": 0.12494961708988311,
      "grad_norm": 2.200929880142212,
      "learning_rate": 4.434835898134278e-05,
      "loss": 4.7567,
      "step": 4650
    },
    {
      "epoch": 0.1250839715168615,
      "grad_norm": 2.485139846801758,
      "learning_rate": 4.4341549775296205e-05,
      "loss": 4.718,
      "step": 4655
    },
    {
      "epoch": 0.12521832594383986,
      "grad_norm": 2.618278980255127,
      "learning_rate": 4.433474056924963e-05,
      "loss": 4.7871,
      "step": 4660
    },
    {
      "epoch": 0.12535268037081823,
      "grad_norm": 2.3308727741241455,
      "learning_rate": 4.432793136320305e-05,
      "loss": 4.7209,
      "step": 4665
    },
    {
      "epoch": 0.12548703479779658,
      "grad_norm": 2.2121143341064453,
      "learning_rate": 4.432112215715647e-05,
      "loss": 4.7732,
      "step": 4670
    },
    {
      "epoch": 0.12562138922477495,
      "grad_norm": 2.4613654613494873,
      "learning_rate": 4.43143129511099e-05,
      "loss": 4.653,
      "step": 4675
    },
    {
      "epoch": 0.12575574365175332,
      "grad_norm": 2.215388298034668,
      "learning_rate": 4.430750374506333e-05,
      "loss": 4.7284,
      "step": 4680
    },
    {
      "epoch": 0.1258900980787317,
      "grad_norm": 2.3992061614990234,
      "learning_rate": 4.430069453901675e-05,
      "loss": 4.7304,
      "step": 4685
    },
    {
      "epoch": 0.12602445250571007,
      "grad_norm": 2.317227363586426,
      "learning_rate": 4.4293885332970175e-05,
      "loss": 4.7673,
      "step": 4690
    },
    {
      "epoch": 0.12615880693268844,
      "grad_norm": 2.5580008029937744,
      "learning_rate": 4.4287076126923604e-05,
      "loss": 4.7657,
      "step": 4695
    },
    {
      "epoch": 0.1262931613596668,
      "grad_norm": 2.4706919193267822,
      "learning_rate": 4.4280266920877026e-05,
      "loss": 4.8699,
      "step": 4700
    },
    {
      "epoch": 0.12642751578664518,
      "grad_norm": 2.168724775314331,
      "learning_rate": 4.4273457714830455e-05,
      "loss": 4.6182,
      "step": 4705
    },
    {
      "epoch": 0.12656187021362353,
      "grad_norm": 2.3658969402313232,
      "learning_rate": 4.426664850878388e-05,
      "loss": 4.7864,
      "step": 4710
    },
    {
      "epoch": 0.1266962246406019,
      "grad_norm": 2.393610715866089,
      "learning_rate": 4.4259839302737306e-05,
      "loss": 4.7453,
      "step": 4715
    },
    {
      "epoch": 0.12683057906758027,
      "grad_norm": 2.362715244293213,
      "learning_rate": 4.425303009669073e-05,
      "loss": 4.737,
      "step": 4720
    },
    {
      "epoch": 0.12696493349455865,
      "grad_norm": 2.5027658939361572,
      "learning_rate": 4.424622089064415e-05,
      "loss": 4.7266,
      "step": 4725
    },
    {
      "epoch": 0.12709928792153702,
      "grad_norm": 2.51792573928833,
      "learning_rate": 4.423941168459757e-05,
      "loss": 4.8414,
      "step": 4730
    },
    {
      "epoch": 0.1272336423485154,
      "grad_norm": 2.5710716247558594,
      "learning_rate": 4.4232602478551e-05,
      "loss": 4.6183,
      "step": 4735
    },
    {
      "epoch": 0.12736799677549376,
      "grad_norm": 2.19071888923645,
      "learning_rate": 4.422579327250443e-05,
      "loss": 4.7927,
      "step": 4740
    },
    {
      "epoch": 0.1275023512024721,
      "grad_norm": 2.284208297729492,
      "learning_rate": 4.4218984066457854e-05,
      "loss": 4.6194,
      "step": 4745
    },
    {
      "epoch": 0.12763670562945048,
      "grad_norm": 2.2364134788513184,
      "learning_rate": 4.4212174860411276e-05,
      "loss": 4.7763,
      "step": 4750
    },
    {
      "epoch": 0.12777106005642885,
      "grad_norm": 2.4097795486450195,
      "learning_rate": 4.4205365654364705e-05,
      "loss": 4.6742,
      "step": 4755
    },
    {
      "epoch": 0.12790541448340723,
      "grad_norm": 2.063389778137207,
      "learning_rate": 4.419855644831813e-05,
      "loss": 4.6319,
      "step": 4760
    },
    {
      "epoch": 0.1280397689103856,
      "grad_norm": 2.4098453521728516,
      "learning_rate": 4.4191747242271556e-05,
      "loss": 4.7207,
      "step": 4765
    },
    {
      "epoch": 0.12817412333736397,
      "grad_norm": 2.6554555892944336,
      "learning_rate": 4.418493803622498e-05,
      "loss": 4.7537,
      "step": 4770
    },
    {
      "epoch": 0.12830847776434234,
      "grad_norm": 2.4152040481567383,
      "learning_rate": 4.417812883017841e-05,
      "loss": 4.8198,
      "step": 4775
    },
    {
      "epoch": 0.12844283219132072,
      "grad_norm": 2.3129639625549316,
      "learning_rate": 4.417131962413183e-05,
      "loss": 4.7426,
      "step": 4780
    },
    {
      "epoch": 0.12857718661829906,
      "grad_norm": 2.2476649284362793,
      "learning_rate": 4.416451041808525e-05,
      "loss": 4.7686,
      "step": 4785
    },
    {
      "epoch": 0.12871154104527743,
      "grad_norm": 2.3596677780151367,
      "learning_rate": 4.4157701212038674e-05,
      "loss": 4.7868,
      "step": 4790
    },
    {
      "epoch": 0.1288458954722558,
      "grad_norm": 2.683835506439209,
      "learning_rate": 4.41508920059921e-05,
      "loss": 4.6513,
      "step": 4795
    },
    {
      "epoch": 0.12898024989923418,
      "grad_norm": 2.1998159885406494,
      "learning_rate": 4.414408279994553e-05,
      "loss": 4.6741,
      "step": 4800
    },
    {
      "epoch": 0.12911460432621255,
      "grad_norm": 2.3731305599212646,
      "learning_rate": 4.4137273593898955e-05,
      "loss": 4.694,
      "step": 4805
    },
    {
      "epoch": 0.12924895875319092,
      "grad_norm": 2.496506929397583,
      "learning_rate": 4.413046438785238e-05,
      "loss": 4.7821,
      "step": 4810
    },
    {
      "epoch": 0.1293833131801693,
      "grad_norm": 2.4163739681243896,
      "learning_rate": 4.41236551818058e-05,
      "loss": 4.6394,
      "step": 4815
    },
    {
      "epoch": 0.12951766760714767,
      "grad_norm": 2.572873592376709,
      "learning_rate": 4.411684597575923e-05,
      "loss": 4.6573,
      "step": 4820
    },
    {
      "epoch": 0.129652022034126,
      "grad_norm": 2.257214307785034,
      "learning_rate": 4.411003676971265e-05,
      "loss": 4.7253,
      "step": 4825
    },
    {
      "epoch": 0.12978637646110439,
      "grad_norm": 2.7237305641174316,
      "learning_rate": 4.410322756366608e-05,
      "loss": 4.4476,
      "step": 4830
    },
    {
      "epoch": 0.12992073088808276,
      "grad_norm": 2.396909475326538,
      "learning_rate": 4.40964183576195e-05,
      "loss": 4.7351,
      "step": 4835
    },
    {
      "epoch": 0.13005508531506113,
      "grad_norm": 2.5134711265563965,
      "learning_rate": 4.408960915157293e-05,
      "loss": 4.8653,
      "step": 4840
    },
    {
      "epoch": 0.1301894397420395,
      "grad_norm": 2.438359498977661,
      "learning_rate": 4.408279994552635e-05,
      "loss": 4.7611,
      "step": 4845
    },
    {
      "epoch": 0.13032379416901788,
      "grad_norm": 2.2780253887176514,
      "learning_rate": 4.4075990739479775e-05,
      "loss": 4.9026,
      "step": 4850
    },
    {
      "epoch": 0.13045814859599625,
      "grad_norm": 2.368100643157959,
      "learning_rate": 4.4069181533433204e-05,
      "loss": 4.7376,
      "step": 4855
    },
    {
      "epoch": 0.13059250302297462,
      "grad_norm": 2.1737263202667236,
      "learning_rate": 4.4062372327386634e-05,
      "loss": 4.8204,
      "step": 4860
    },
    {
      "epoch": 0.13072685744995297,
      "grad_norm": 2.2839977741241455,
      "learning_rate": 4.4055563121340056e-05,
      "loss": 4.7728,
      "step": 4865
    },
    {
      "epoch": 0.13086121187693134,
      "grad_norm": 2.3215525150299072,
      "learning_rate": 4.404875391529348e-05,
      "loss": 4.7073,
      "step": 4870
    },
    {
      "epoch": 0.1309955663039097,
      "grad_norm": 2.2425525188446045,
      "learning_rate": 4.40419447092469e-05,
      "loss": 4.7774,
      "step": 4875
    },
    {
      "epoch": 0.13112992073088808,
      "grad_norm": 2.4383742809295654,
      "learning_rate": 4.403513550320033e-05,
      "loss": 4.7138,
      "step": 4880
    },
    {
      "epoch": 0.13126427515786646,
      "grad_norm": 2.2931289672851562,
      "learning_rate": 4.402832629715375e-05,
      "loss": 4.7834,
      "step": 4885
    },
    {
      "epoch": 0.13139862958484483,
      "grad_norm": 2.4571781158447266,
      "learning_rate": 4.402151709110718e-05,
      "loss": 4.6783,
      "step": 4890
    },
    {
      "epoch": 0.1315329840118232,
      "grad_norm": 2.2795464992523193,
      "learning_rate": 4.40147078850606e-05,
      "loss": 4.6735,
      "step": 4895
    },
    {
      "epoch": 0.13166733843880155,
      "grad_norm": 2.322563648223877,
      "learning_rate": 4.400789867901403e-05,
      "loss": 4.7541,
      "step": 4900
    },
    {
      "epoch": 0.13180169286577992,
      "grad_norm": 2.354914665222168,
      "learning_rate": 4.4001089472967454e-05,
      "loss": 4.7263,
      "step": 4905
    },
    {
      "epoch": 0.1319360472927583,
      "grad_norm": 2.393778085708618,
      "learning_rate": 4.3994280266920877e-05,
      "loss": 4.6778,
      "step": 4910
    },
    {
      "epoch": 0.13207040171973666,
      "grad_norm": 2.352766752243042,
      "learning_rate": 4.3987471060874306e-05,
      "loss": 4.7806,
      "step": 4915
    },
    {
      "epoch": 0.13220475614671504,
      "grad_norm": 2.304657459259033,
      "learning_rate": 4.3980661854827735e-05,
      "loss": 4.6589,
      "step": 4920
    },
    {
      "epoch": 0.1323391105736934,
      "grad_norm": 2.1979665756225586,
      "learning_rate": 4.397385264878116e-05,
      "loss": 4.8172,
      "step": 4925
    },
    {
      "epoch": 0.13247346500067178,
      "grad_norm": 2.532045602798462,
      "learning_rate": 4.396704344273458e-05,
      "loss": 4.7536,
      "step": 4930
    },
    {
      "epoch": 0.13260781942765015,
      "grad_norm": 2.3392767906188965,
      "learning_rate": 4.3960234236688e-05,
      "loss": 4.8324,
      "step": 4935
    },
    {
      "epoch": 0.1327421738546285,
      "grad_norm": 2.3727712631225586,
      "learning_rate": 4.395342503064143e-05,
      "loss": 4.7825,
      "step": 4940
    },
    {
      "epoch": 0.13287652828160687,
      "grad_norm": 2.3137311935424805,
      "learning_rate": 4.394661582459485e-05,
      "loss": 4.7353,
      "step": 4945
    },
    {
      "epoch": 0.13301088270858524,
      "grad_norm": 2.325050115585327,
      "learning_rate": 4.393980661854828e-05,
      "loss": 4.6556,
      "step": 4950
    },
    {
      "epoch": 0.13314523713556362,
      "grad_norm": 2.3303592205047607,
      "learning_rate": 4.3932997412501704e-05,
      "loss": 4.7295,
      "step": 4955
    },
    {
      "epoch": 0.133279591562542,
      "grad_norm": 2.3345165252685547,
      "learning_rate": 4.3926188206455126e-05,
      "loss": 4.665,
      "step": 4960
    },
    {
      "epoch": 0.13341394598952036,
      "grad_norm": 2.2794764041900635,
      "learning_rate": 4.3919379000408555e-05,
      "loss": 4.7288,
      "step": 4965
    },
    {
      "epoch": 0.13354830041649873,
      "grad_norm": 2.4537861347198486,
      "learning_rate": 4.391256979436198e-05,
      "loss": 4.7142,
      "step": 4970
    },
    {
      "epoch": 0.1336826548434771,
      "grad_norm": 2.697866916656494,
      "learning_rate": 4.39057605883154e-05,
      "loss": 4.8093,
      "step": 4975
    },
    {
      "epoch": 0.13381700927045545,
      "grad_norm": 2.7227869033813477,
      "learning_rate": 4.389895138226883e-05,
      "loss": 4.7776,
      "step": 4980
    },
    {
      "epoch": 0.13395136369743382,
      "grad_norm": 2.3009612560272217,
      "learning_rate": 4.389214217622226e-05,
      "loss": 4.708,
      "step": 4985
    },
    {
      "epoch": 0.1340857181244122,
      "grad_norm": 2.4681789875030518,
      "learning_rate": 4.388533297017568e-05,
      "loss": 4.8216,
      "step": 4990
    },
    {
      "epoch": 0.13422007255139057,
      "grad_norm": 2.1975107192993164,
      "learning_rate": 4.38785237641291e-05,
      "loss": 4.799,
      "step": 4995
    },
    {
      "epoch": 0.13435442697836894,
      "grad_norm": 2.4353811740875244,
      "learning_rate": 4.3871714558082525e-05,
      "loss": 4.7116,
      "step": 5000
    },
    {
      "epoch": 0.1344887814053473,
      "grad_norm": 2.7893145084381104,
      "learning_rate": 4.3864905352035954e-05,
      "loss": 4.7318,
      "step": 5005
    },
    {
      "epoch": 0.13462313583232569,
      "grad_norm": 2.3651912212371826,
      "learning_rate": 4.385809614598938e-05,
      "loss": 4.881,
      "step": 5010
    },
    {
      "epoch": 0.13475749025930403,
      "grad_norm": 2.261274814605713,
      "learning_rate": 4.3851286939942805e-05,
      "loss": 4.8236,
      "step": 5015
    },
    {
      "epoch": 0.1348918446862824,
      "grad_norm": 2.3519933223724365,
      "learning_rate": 4.384447773389623e-05,
      "loss": 4.761,
      "step": 5020
    },
    {
      "epoch": 0.13502619911326078,
      "grad_norm": 2.2974727153778076,
      "learning_rate": 4.3837668527849656e-05,
      "loss": 4.738,
      "step": 5025
    },
    {
      "epoch": 0.13516055354023915,
      "grad_norm": 2.246166467666626,
      "learning_rate": 4.383085932180308e-05,
      "loss": 4.858,
      "step": 5030
    },
    {
      "epoch": 0.13529490796721752,
      "grad_norm": 2.214803457260132,
      "learning_rate": 4.38240501157565e-05,
      "loss": 4.7445,
      "step": 5035
    },
    {
      "epoch": 0.1354292623941959,
      "grad_norm": 2.1694109439849854,
      "learning_rate": 4.381724090970993e-05,
      "loss": 4.7438,
      "step": 5040
    },
    {
      "epoch": 0.13556361682117427,
      "grad_norm": 2.5251026153564453,
      "learning_rate": 4.381043170366336e-05,
      "loss": 4.6166,
      "step": 5045
    },
    {
      "epoch": 0.13569797124815264,
      "grad_norm": 2.5384957790374756,
      "learning_rate": 4.380362249761678e-05,
      "loss": 4.6988,
      "step": 5050
    },
    {
      "epoch": 0.13583232567513098,
      "grad_norm": 2.5966992378234863,
      "learning_rate": 4.3796813291570204e-05,
      "loss": 4.8781,
      "step": 5055
    },
    {
      "epoch": 0.13596668010210936,
      "grad_norm": 2.3633997440338135,
      "learning_rate": 4.3790004085523626e-05,
      "loss": 4.7447,
      "step": 5060
    },
    {
      "epoch": 0.13610103452908773,
      "grad_norm": 2.391948699951172,
      "learning_rate": 4.3783194879477055e-05,
      "loss": 4.7047,
      "step": 5065
    },
    {
      "epoch": 0.1362353889560661,
      "grad_norm": 2.0203118324279785,
      "learning_rate": 4.3776385673430484e-05,
      "loss": 4.7053,
      "step": 5070
    },
    {
      "epoch": 0.13636974338304447,
      "grad_norm": 2.174410820007324,
      "learning_rate": 4.3769576467383906e-05,
      "loss": 4.7354,
      "step": 5075
    },
    {
      "epoch": 0.13650409781002285,
      "grad_norm": 2.3941686153411865,
      "learning_rate": 4.376276726133733e-05,
      "loss": 4.7285,
      "step": 5080
    },
    {
      "epoch": 0.13663845223700122,
      "grad_norm": 2.2955944538116455,
      "learning_rate": 4.375595805529076e-05,
      "loss": 4.6808,
      "step": 5085
    },
    {
      "epoch": 0.1367728066639796,
      "grad_norm": 2.482304573059082,
      "learning_rate": 4.374914884924418e-05,
      "loss": 4.7423,
      "step": 5090
    },
    {
      "epoch": 0.13690716109095794,
      "grad_norm": 2.3817191123962402,
      "learning_rate": 4.37423396431976e-05,
      "loss": 4.6605,
      "step": 5095
    },
    {
      "epoch": 0.1370415155179363,
      "grad_norm": 2.444042205810547,
      "learning_rate": 4.373553043715103e-05,
      "loss": 4.7439,
      "step": 5100
    },
    {
      "epoch": 0.13717586994491468,
      "grad_norm": 2.229215145111084,
      "learning_rate": 4.372872123110446e-05,
      "loss": 4.7792,
      "step": 5105
    },
    {
      "epoch": 0.13731022437189305,
      "grad_norm": 2.2401981353759766,
      "learning_rate": 4.372191202505788e-05,
      "loss": 4.7043,
      "step": 5110
    },
    {
      "epoch": 0.13744457879887143,
      "grad_norm": 2.2581069469451904,
      "learning_rate": 4.3715102819011305e-05,
      "loss": 4.7434,
      "step": 5115
    },
    {
      "epoch": 0.1375789332258498,
      "grad_norm": 2.096940755844116,
      "learning_rate": 4.370829361296473e-05,
      "loss": 4.6921,
      "step": 5120
    },
    {
      "epoch": 0.13771328765282817,
      "grad_norm": 2.375547409057617,
      "learning_rate": 4.3701484406918156e-05,
      "loss": 4.8029,
      "step": 5125
    },
    {
      "epoch": 0.13784764207980654,
      "grad_norm": 2.2480034828186035,
      "learning_rate": 4.3694675200871585e-05,
      "loss": 4.5713,
      "step": 5130
    },
    {
      "epoch": 0.1379819965067849,
      "grad_norm": 2.2785820960998535,
      "learning_rate": 4.368786599482501e-05,
      "loss": 4.7943,
      "step": 5135
    },
    {
      "epoch": 0.13811635093376326,
      "grad_norm": 2.2028582096099854,
      "learning_rate": 4.368105678877843e-05,
      "loss": 4.7145,
      "step": 5140
    },
    {
      "epoch": 0.13825070536074163,
      "grad_norm": 2.2140886783599854,
      "learning_rate": 4.367424758273185e-05,
      "loss": 4.8231,
      "step": 5145
    },
    {
      "epoch": 0.13838505978772,
      "grad_norm": 2.371229648590088,
      "learning_rate": 4.366743837668528e-05,
      "loss": 4.6638,
      "step": 5150
    },
    {
      "epoch": 0.13851941421469838,
      "grad_norm": 2.5016679763793945,
      "learning_rate": 4.36606291706387e-05,
      "loss": 4.7156,
      "step": 5155
    },
    {
      "epoch": 0.13865376864167675,
      "grad_norm": 2.2718706130981445,
      "learning_rate": 4.365381996459213e-05,
      "loss": 4.645,
      "step": 5160
    },
    {
      "epoch": 0.13878812306865512,
      "grad_norm": 2.4279985427856445,
      "learning_rate": 4.3647010758545555e-05,
      "loss": 4.8226,
      "step": 5165
    },
    {
      "epoch": 0.13892247749563347,
      "grad_norm": 2.6278507709503174,
      "learning_rate": 4.3640201552498984e-05,
      "loss": 4.8118,
      "step": 5170
    },
    {
      "epoch": 0.13905683192261184,
      "grad_norm": 2.070530414581299,
      "learning_rate": 4.3633392346452406e-05,
      "loss": 4.6142,
      "step": 5175
    },
    {
      "epoch": 0.1391911863495902,
      "grad_norm": 2.2454535961151123,
      "learning_rate": 4.362658314040583e-05,
      "loss": 4.6687,
      "step": 5180
    },
    {
      "epoch": 0.13932554077656858,
      "grad_norm": 2.251532554626465,
      "learning_rate": 4.361977393435925e-05,
      "loss": 4.6862,
      "step": 5185
    },
    {
      "epoch": 0.13945989520354696,
      "grad_norm": 2.5029964447021484,
      "learning_rate": 4.3612964728312686e-05,
      "loss": 4.6623,
      "step": 5190
    },
    {
      "epoch": 0.13959424963052533,
      "grad_norm": 2.2702221870422363,
      "learning_rate": 4.360615552226611e-05,
      "loss": 4.619,
      "step": 5195
    },
    {
      "epoch": 0.1397286040575037,
      "grad_norm": 2.638491630554199,
      "learning_rate": 4.359934631621953e-05,
      "loss": 4.7212,
      "step": 5200
    },
    {
      "epoch": 0.13986295848448208,
      "grad_norm": 2.2260608673095703,
      "learning_rate": 4.359253711017295e-05,
      "loss": 4.6825,
      "step": 5205
    },
    {
      "epoch": 0.13999731291146042,
      "grad_norm": 2.407391309738159,
      "learning_rate": 4.358572790412638e-05,
      "loss": 4.6911,
      "step": 5210
    },
    {
      "epoch": 0.1401316673384388,
      "grad_norm": 2.4948818683624268,
      "learning_rate": 4.3578918698079804e-05,
      "loss": 4.7699,
      "step": 5215
    },
    {
      "epoch": 0.14026602176541716,
      "grad_norm": 2.351940155029297,
      "learning_rate": 4.3572109492033233e-05,
      "loss": 4.6927,
      "step": 5220
    },
    {
      "epoch": 0.14040037619239554,
      "grad_norm": 2.388455390930176,
      "learning_rate": 4.3565300285986656e-05,
      "loss": 4.8632,
      "step": 5225
    },
    {
      "epoch": 0.1405347306193739,
      "grad_norm": 2.1543734073638916,
      "learning_rate": 4.3558491079940085e-05,
      "loss": 4.7038,
      "step": 5230
    },
    {
      "epoch": 0.14066908504635228,
      "grad_norm": 2.450618267059326,
      "learning_rate": 4.355168187389351e-05,
      "loss": 4.7647,
      "step": 5235
    },
    {
      "epoch": 0.14080343947333065,
      "grad_norm": 2.4088172912597656,
      "learning_rate": 4.354487266784693e-05,
      "loss": 4.7614,
      "step": 5240
    },
    {
      "epoch": 0.14093779390030903,
      "grad_norm": 2.2573354244232178,
      "learning_rate": 4.353806346180035e-05,
      "loss": 4.71,
      "step": 5245
    },
    {
      "epoch": 0.14107214832728737,
      "grad_norm": 2.1550099849700928,
      "learning_rate": 4.353125425575378e-05,
      "loss": 4.6734,
      "step": 5250
    },
    {
      "epoch": 0.14120650275426574,
      "grad_norm": 2.2590415477752686,
      "learning_rate": 4.352444504970721e-05,
      "loss": 4.6657,
      "step": 5255
    },
    {
      "epoch": 0.14134085718124412,
      "grad_norm": 2.579711675643921,
      "learning_rate": 4.351763584366063e-05,
      "loss": 4.8048,
      "step": 5260
    },
    {
      "epoch": 0.1414752116082225,
      "grad_norm": 2.5554847717285156,
      "learning_rate": 4.3510826637614054e-05,
      "loss": 4.6923,
      "step": 5265
    },
    {
      "epoch": 0.14160956603520086,
      "grad_norm": 2.264326810836792,
      "learning_rate": 4.3504017431567476e-05,
      "loss": 4.6755,
      "step": 5270
    },
    {
      "epoch": 0.14174392046217923,
      "grad_norm": 2.633161783218384,
      "learning_rate": 4.3497208225520905e-05,
      "loss": 4.6751,
      "step": 5275
    },
    {
      "epoch": 0.1418782748891576,
      "grad_norm": 3.0556602478027344,
      "learning_rate": 4.3490399019474335e-05,
      "loss": 4.6202,
      "step": 5280
    },
    {
      "epoch": 0.14201262931613598,
      "grad_norm": 2.360241174697876,
      "learning_rate": 4.348358981342776e-05,
      "loss": 4.7366,
      "step": 5285
    },
    {
      "epoch": 0.14214698374311432,
      "grad_norm": 2.3771448135375977,
      "learning_rate": 4.347678060738118e-05,
      "loss": 4.6491,
      "step": 5290
    },
    {
      "epoch": 0.1422813381700927,
      "grad_norm": 2.277683734893799,
      "learning_rate": 4.346997140133461e-05,
      "loss": 4.6599,
      "step": 5295
    },
    {
      "epoch": 0.14241569259707107,
      "grad_norm": 2.3080601692199707,
      "learning_rate": 4.346316219528803e-05,
      "loss": 4.7805,
      "step": 5300
    },
    {
      "epoch": 0.14255004702404944,
      "grad_norm": 2.375811815261841,
      "learning_rate": 4.345635298924145e-05,
      "loss": 4.6954,
      "step": 5305
    },
    {
      "epoch": 0.14268440145102781,
      "grad_norm": 2.5283961296081543,
      "learning_rate": 4.344954378319488e-05,
      "loss": 4.5886,
      "step": 5310
    },
    {
      "epoch": 0.1428187558780062,
      "grad_norm": 2.4433255195617676,
      "learning_rate": 4.344273457714831e-05,
      "loss": 4.7019,
      "step": 5315
    },
    {
      "epoch": 0.14295311030498456,
      "grad_norm": 2.1628475189208984,
      "learning_rate": 4.343592537110173e-05,
      "loss": 4.6834,
      "step": 5320
    },
    {
      "epoch": 0.1430874647319629,
      "grad_norm": 2.1188840866088867,
      "learning_rate": 4.3429116165055155e-05,
      "loss": 4.7292,
      "step": 5325
    },
    {
      "epoch": 0.14322181915894128,
      "grad_norm": 2.0829310417175293,
      "learning_rate": 4.342230695900858e-05,
      "loss": 4.7359,
      "step": 5330
    },
    {
      "epoch": 0.14335617358591965,
      "grad_norm": 2.472219228744507,
      "learning_rate": 4.3415497752962007e-05,
      "loss": 4.7155,
      "step": 5335
    },
    {
      "epoch": 0.14349052801289802,
      "grad_norm": 2.3763554096221924,
      "learning_rate": 4.3408688546915436e-05,
      "loss": 4.704,
      "step": 5340
    },
    {
      "epoch": 0.1436248824398764,
      "grad_norm": 2.222320318222046,
      "learning_rate": 4.340187934086886e-05,
      "loss": 4.7033,
      "step": 5345
    },
    {
      "epoch": 0.14375923686685477,
      "grad_norm": 2.209571599960327,
      "learning_rate": 4.339507013482228e-05,
      "loss": 4.6626,
      "step": 5350
    },
    {
      "epoch": 0.14389359129383314,
      "grad_norm": 2.4574036598205566,
      "learning_rate": 4.338826092877571e-05,
      "loss": 4.6549,
      "step": 5355
    },
    {
      "epoch": 0.1440279457208115,
      "grad_norm": 2.506190538406372,
      "learning_rate": 4.338145172272913e-05,
      "loss": 4.6334,
      "step": 5360
    },
    {
      "epoch": 0.14416230014778986,
      "grad_norm": 2.230236291885376,
      "learning_rate": 4.3374642516682554e-05,
      "loss": 4.7253,
      "step": 5365
    },
    {
      "epoch": 0.14429665457476823,
      "grad_norm": 2.390226364135742,
      "learning_rate": 4.336783331063598e-05,
      "loss": 4.6153,
      "step": 5370
    },
    {
      "epoch": 0.1444310090017466,
      "grad_norm": 2.410153388977051,
      "learning_rate": 4.336102410458941e-05,
      "loss": 4.6611,
      "step": 5375
    },
    {
      "epoch": 0.14456536342872497,
      "grad_norm": 2.4580087661743164,
      "learning_rate": 4.3354214898542834e-05,
      "loss": 4.7531,
      "step": 5380
    },
    {
      "epoch": 0.14469971785570335,
      "grad_norm": 2.527559757232666,
      "learning_rate": 4.3347405692496256e-05,
      "loss": 4.5698,
      "step": 5385
    },
    {
      "epoch": 0.14483407228268172,
      "grad_norm": 2.5995399951934814,
      "learning_rate": 4.334059648644968e-05,
      "loss": 4.5992,
      "step": 5390
    },
    {
      "epoch": 0.1449684267096601,
      "grad_norm": 2.2917354106903076,
      "learning_rate": 4.333378728040311e-05,
      "loss": 4.7568,
      "step": 5395
    },
    {
      "epoch": 0.14510278113663846,
      "grad_norm": 2.3101868629455566,
      "learning_rate": 4.332697807435654e-05,
      "loss": 4.6168,
      "step": 5400
    },
    {
      "epoch": 0.1452371355636168,
      "grad_norm": 2.313093900680542,
      "learning_rate": 4.332016886830996e-05,
      "loss": 4.8072,
      "step": 5405
    },
    {
      "epoch": 0.14537148999059518,
      "grad_norm": 2.5541889667510986,
      "learning_rate": 4.331335966226338e-05,
      "loss": 4.7087,
      "step": 5410
    },
    {
      "epoch": 0.14550584441757355,
      "grad_norm": 2.508989095687866,
      "learning_rate": 4.3306550456216804e-05,
      "loss": 4.6275,
      "step": 5415
    },
    {
      "epoch": 0.14564019884455193,
      "grad_norm": 2.3997724056243896,
      "learning_rate": 4.329974125017023e-05,
      "loss": 4.6216,
      "step": 5420
    },
    {
      "epoch": 0.1457745532715303,
      "grad_norm": 2.179547071456909,
      "learning_rate": 4.3292932044123655e-05,
      "loss": 4.6877,
      "step": 5425
    },
    {
      "epoch": 0.14590890769850867,
      "grad_norm": 2.3510310649871826,
      "learning_rate": 4.3286122838077084e-05,
      "loss": 4.7389,
      "step": 5430
    },
    {
      "epoch": 0.14604326212548704,
      "grad_norm": 2.5767464637756348,
      "learning_rate": 4.3279313632030506e-05,
      "loss": 4.556,
      "step": 5435
    },
    {
      "epoch": 0.14617761655246542,
      "grad_norm": 2.4311118125915527,
      "learning_rate": 4.3272504425983935e-05,
      "loss": 4.7749,
      "step": 5440
    },
    {
      "epoch": 0.14631197097944376,
      "grad_norm": 2.3742549419403076,
      "learning_rate": 4.326569521993736e-05,
      "loss": 4.6938,
      "step": 5445
    },
    {
      "epoch": 0.14644632540642213,
      "grad_norm": 2.472548484802246,
      "learning_rate": 4.325888601389078e-05,
      "loss": 4.75,
      "step": 5450
    },
    {
      "epoch": 0.1465806798334005,
      "grad_norm": 2.5138728618621826,
      "learning_rate": 4.32520768078442e-05,
      "loss": 4.6166,
      "step": 5455
    },
    {
      "epoch": 0.14671503426037888,
      "grad_norm": 2.5273678302764893,
      "learning_rate": 4.324526760179763e-05,
      "loss": 4.7135,
      "step": 5460
    },
    {
      "epoch": 0.14684938868735725,
      "grad_norm": 2.2863588333129883,
      "learning_rate": 4.323845839575106e-05,
      "loss": 4.8035,
      "step": 5465
    },
    {
      "epoch": 0.14698374311433562,
      "grad_norm": 2.5088889598846436,
      "learning_rate": 4.323164918970448e-05,
      "loss": 4.5936,
      "step": 5470
    },
    {
      "epoch": 0.147118097541314,
      "grad_norm": 2.1304173469543457,
      "learning_rate": 4.3224839983657905e-05,
      "loss": 4.748,
      "step": 5475
    },
    {
      "epoch": 0.14725245196829234,
      "grad_norm": 2.287482261657715,
      "learning_rate": 4.3218030777611334e-05,
      "loss": 4.7375,
      "step": 5480
    },
    {
      "epoch": 0.14738680639527071,
      "grad_norm": 2.2737138271331787,
      "learning_rate": 4.3211221571564756e-05,
      "loss": 4.7367,
      "step": 5485
    },
    {
      "epoch": 0.1475211608222491,
      "grad_norm": 2.5050153732299805,
      "learning_rate": 4.3204412365518185e-05,
      "loss": 4.7159,
      "step": 5490
    },
    {
      "epoch": 0.14765551524922746,
      "grad_norm": 2.3213109970092773,
      "learning_rate": 4.319760315947161e-05,
      "loss": 4.5497,
      "step": 5495
    },
    {
      "epoch": 0.14778986967620583,
      "grad_norm": 2.4361581802368164,
      "learning_rate": 4.3190793953425036e-05,
      "loss": 4.7025,
      "step": 5500
    },
    {
      "epoch": 0.1479242241031842,
      "grad_norm": 2.4409122467041016,
      "learning_rate": 4.318398474737846e-05,
      "loss": 4.693,
      "step": 5505
    },
    {
      "epoch": 0.14805857853016258,
      "grad_norm": 2.3179500102996826,
      "learning_rate": 4.317717554133188e-05,
      "loss": 4.6942,
      "step": 5510
    },
    {
      "epoch": 0.14819293295714095,
      "grad_norm": 2.516125440597534,
      "learning_rate": 4.31703663352853e-05,
      "loss": 4.6148,
      "step": 5515
    },
    {
      "epoch": 0.1483272873841193,
      "grad_norm": 2.203394889831543,
      "learning_rate": 4.316355712923873e-05,
      "loss": 4.7775,
      "step": 5520
    },
    {
      "epoch": 0.14846164181109767,
      "grad_norm": 2.5464253425598145,
      "learning_rate": 4.315674792319216e-05,
      "loss": 4.6528,
      "step": 5525
    },
    {
      "epoch": 0.14859599623807604,
      "grad_norm": 2.224968671798706,
      "learning_rate": 4.3149938717145584e-05,
      "loss": 4.5648,
      "step": 5530
    },
    {
      "epoch": 0.1487303506650544,
      "grad_norm": 2.447449207305908,
      "learning_rate": 4.3143129511099006e-05,
      "loss": 4.6464,
      "step": 5535
    },
    {
      "epoch": 0.14886470509203278,
      "grad_norm": 2.341639280319214,
      "learning_rate": 4.3136320305052435e-05,
      "loss": 4.6618,
      "step": 5540
    },
    {
      "epoch": 0.14899905951901116,
      "grad_norm": 2.156888961791992,
      "learning_rate": 4.312951109900586e-05,
      "loss": 4.7622,
      "step": 5545
    },
    {
      "epoch": 0.14913341394598953,
      "grad_norm": 2.1902127265930176,
      "learning_rate": 4.3122701892959286e-05,
      "loss": 4.6198,
      "step": 5550
    },
    {
      "epoch": 0.1492677683729679,
      "grad_norm": 2.3121960163116455,
      "learning_rate": 4.311589268691271e-05,
      "loss": 4.6535,
      "step": 5555
    },
    {
      "epoch": 0.14940212279994625,
      "grad_norm": 2.141181468963623,
      "learning_rate": 4.310908348086613e-05,
      "loss": 4.7069,
      "step": 5560
    },
    {
      "epoch": 0.14953647722692462,
      "grad_norm": 2.5161216259002686,
      "learning_rate": 4.310227427481956e-05,
      "loss": 4.8493,
      "step": 5565
    },
    {
      "epoch": 0.149670831653903,
      "grad_norm": 2.417941093444824,
      "learning_rate": 4.309546506877298e-05,
      "loss": 4.5883,
      "step": 5570
    },
    {
      "epoch": 0.14980518608088136,
      "grad_norm": 2.4670495986938477,
      "learning_rate": 4.3088655862726404e-05,
      "loss": 4.7349,
      "step": 5575
    },
    {
      "epoch": 0.14993954050785974,
      "grad_norm": 2.4075560569763184,
      "learning_rate": 4.308184665667983e-05,
      "loss": 4.686,
      "step": 5580
    },
    {
      "epoch": 0.1500738949348381,
      "grad_norm": 2.2981793880462646,
      "learning_rate": 4.307503745063326e-05,
      "loss": 4.5618,
      "step": 5585
    },
    {
      "epoch": 0.15020824936181648,
      "grad_norm": 2.3917644023895264,
      "learning_rate": 4.3068228244586685e-05,
      "loss": 4.6336,
      "step": 5590
    },
    {
      "epoch": 0.15034260378879485,
      "grad_norm": 2.308039903640747,
      "learning_rate": 4.306141903854011e-05,
      "loss": 4.6664,
      "step": 5595
    },
    {
      "epoch": 0.1504769582157732,
      "grad_norm": 2.4636881351470947,
      "learning_rate": 4.305460983249353e-05,
      "loss": 4.6192,
      "step": 5600
    },
    {
      "epoch": 0.15061131264275157,
      "grad_norm": 2.4126856327056885,
      "learning_rate": 4.304780062644696e-05,
      "loss": 4.7288,
      "step": 5605
    },
    {
      "epoch": 0.15074566706972994,
      "grad_norm": 2.3596444129943848,
      "learning_rate": 4.304099142040039e-05,
      "loss": 4.7448,
      "step": 5610
    },
    {
      "epoch": 0.15088002149670832,
      "grad_norm": 2.4230477809906006,
      "learning_rate": 4.303418221435381e-05,
      "loss": 4.5681,
      "step": 5615
    },
    {
      "epoch": 0.1510143759236867,
      "grad_norm": 2.239949941635132,
      "learning_rate": 4.302737300830723e-05,
      "loss": 4.6868,
      "step": 5620
    },
    {
      "epoch": 0.15114873035066506,
      "grad_norm": 2.2889106273651123,
      "learning_rate": 4.302056380226066e-05,
      "loss": 4.7044,
      "step": 5625
    },
    {
      "epoch": 0.15128308477764343,
      "grad_norm": 2.311051368713379,
      "learning_rate": 4.301375459621408e-05,
      "loss": 4.631,
      "step": 5630
    },
    {
      "epoch": 0.15141743920462178,
      "grad_norm": 2.24658465385437,
      "learning_rate": 4.3006945390167505e-05,
      "loss": 4.6437,
      "step": 5635
    },
    {
      "epoch": 0.15155179363160015,
      "grad_norm": 2.239083766937256,
      "learning_rate": 4.3000136184120934e-05,
      "loss": 4.7385,
      "step": 5640
    },
    {
      "epoch": 0.15168614805857852,
      "grad_norm": 2.26863694190979,
      "learning_rate": 4.2993326978074363e-05,
      "loss": 4.6555,
      "step": 5645
    },
    {
      "epoch": 0.1518205024855569,
      "grad_norm": 2.4318182468414307,
      "learning_rate": 4.2986517772027786e-05,
      "loss": 4.645,
      "step": 5650
    },
    {
      "epoch": 0.15195485691253527,
      "grad_norm": 2.3421406745910645,
      "learning_rate": 4.297970856598121e-05,
      "loss": 4.6459,
      "step": 5655
    },
    {
      "epoch": 0.15208921133951364,
      "grad_norm": 2.5247678756713867,
      "learning_rate": 4.297289935993463e-05,
      "loss": 4.7227,
      "step": 5660
    },
    {
      "epoch": 0.152223565766492,
      "grad_norm": 2.3438174724578857,
      "learning_rate": 4.296609015388806e-05,
      "loss": 4.7852,
      "step": 5665
    },
    {
      "epoch": 0.1523579201934704,
      "grad_norm": 2.39847731590271,
      "learning_rate": 4.295928094784148e-05,
      "loss": 4.6858,
      "step": 5670
    },
    {
      "epoch": 0.15249227462044873,
      "grad_norm": 2.4760515689849854,
      "learning_rate": 4.295247174179491e-05,
      "loss": 4.6744,
      "step": 5675
    },
    {
      "epoch": 0.1526266290474271,
      "grad_norm": 2.105417490005493,
      "learning_rate": 4.294566253574833e-05,
      "loss": 4.6975,
      "step": 5680
    },
    {
      "epoch": 0.15276098347440548,
      "grad_norm": 2.1435978412628174,
      "learning_rate": 4.293885332970176e-05,
      "loss": 4.6832,
      "step": 5685
    },
    {
      "epoch": 0.15289533790138385,
      "grad_norm": 2.5127382278442383,
      "learning_rate": 4.2932044123655184e-05,
      "loss": 4.6844,
      "step": 5690
    },
    {
      "epoch": 0.15302969232836222,
      "grad_norm": 2.443847417831421,
      "learning_rate": 4.2925234917608606e-05,
      "loss": 4.5977,
      "step": 5695
    },
    {
      "epoch": 0.1531640467553406,
      "grad_norm": 2.358274221420288,
      "learning_rate": 4.2918425711562036e-05,
      "loss": 4.7252,
      "step": 5700
    },
    {
      "epoch": 0.15329840118231897,
      "grad_norm": 2.2438042163848877,
      "learning_rate": 4.291161650551546e-05,
      "loss": 4.7179,
      "step": 5705
    },
    {
      "epoch": 0.15343275560929734,
      "grad_norm": 2.232947826385498,
      "learning_rate": 4.290480729946889e-05,
      "loss": 4.6442,
      "step": 5710
    },
    {
      "epoch": 0.15356711003627568,
      "grad_norm": 2.2285115718841553,
      "learning_rate": 4.289799809342231e-05,
      "loss": 4.7839,
      "step": 5715
    },
    {
      "epoch": 0.15370146446325406,
      "grad_norm": 2.3586935997009277,
      "learning_rate": 4.289118888737573e-05,
      "loss": 4.7326,
      "step": 5720
    },
    {
      "epoch": 0.15383581889023243,
      "grad_norm": 2.3652799129486084,
      "learning_rate": 4.2884379681329154e-05,
      "loss": 4.6875,
      "step": 5725
    },
    {
      "epoch": 0.1539701733172108,
      "grad_norm": 2.1931145191192627,
      "learning_rate": 4.287757047528258e-05,
      "loss": 4.6205,
      "step": 5730
    },
    {
      "epoch": 0.15410452774418917,
      "grad_norm": 2.5673062801361084,
      "learning_rate": 4.287076126923601e-05,
      "loss": 4.7581,
      "step": 5735
    },
    {
      "epoch": 0.15423888217116755,
      "grad_norm": 2.4709861278533936,
      "learning_rate": 4.2863952063189434e-05,
      "loss": 4.6863,
      "step": 5740
    },
    {
      "epoch": 0.15437323659814592,
      "grad_norm": 2.0905582904815674,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 4.7595,
      "step": 5745
    },
    {
      "epoch": 0.1545075910251243,
      "grad_norm": 2.549795150756836,
      "learning_rate": 4.2850333651096285e-05,
      "loss": 4.8004,
      "step": 5750
    },
    {
      "epoch": 0.15464194545210264,
      "grad_norm": 2.912210464477539,
      "learning_rate": 4.284352444504971e-05,
      "loss": 4.4553,
      "step": 5755
    },
    {
      "epoch": 0.154776299879081,
      "grad_norm": 2.2438201904296875,
      "learning_rate": 4.283671523900314e-05,
      "loss": 4.6412,
      "step": 5760
    },
    {
      "epoch": 0.15491065430605938,
      "grad_norm": 2.1749048233032227,
      "learning_rate": 4.282990603295656e-05,
      "loss": 4.6736,
      "step": 5765
    },
    {
      "epoch": 0.15504500873303775,
      "grad_norm": 2.099368095397949,
      "learning_rate": 4.282309682690999e-05,
      "loss": 4.6474,
      "step": 5770
    },
    {
      "epoch": 0.15517936316001613,
      "grad_norm": 2.3514087200164795,
      "learning_rate": 4.281628762086341e-05,
      "loss": 4.5752,
      "step": 5775
    },
    {
      "epoch": 0.1553137175869945,
      "grad_norm": 2.193413734436035,
      "learning_rate": 4.280947841481683e-05,
      "loss": 4.5385,
      "step": 5780
    },
    {
      "epoch": 0.15544807201397287,
      "grad_norm": 2.358552932739258,
      "learning_rate": 4.2802669208770255e-05,
      "loss": 4.6124,
      "step": 5785
    },
    {
      "epoch": 0.15558242644095122,
      "grad_norm": 2.308378219604492,
      "learning_rate": 4.2795860002723684e-05,
      "loss": 4.6325,
      "step": 5790
    },
    {
      "epoch": 0.1557167808679296,
      "grad_norm": 2.1170716285705566,
      "learning_rate": 4.278905079667711e-05,
      "loss": 4.6792,
      "step": 5795
    },
    {
      "epoch": 0.15585113529490796,
      "grad_norm": 2.5537612438201904,
      "learning_rate": 4.2782241590630535e-05,
      "loss": 4.7151,
      "step": 5800
    },
    {
      "epoch": 0.15598548972188633,
      "grad_norm": 2.303457736968994,
      "learning_rate": 4.277543238458396e-05,
      "loss": 4.5988,
      "step": 5805
    },
    {
      "epoch": 0.1561198441488647,
      "grad_norm": 2.3445920944213867,
      "learning_rate": 4.2768623178537386e-05,
      "loss": 4.5099,
      "step": 5810
    },
    {
      "epoch": 0.15625419857584308,
      "grad_norm": 2.38425874710083,
      "learning_rate": 4.276181397249081e-05,
      "loss": 4.7772,
      "step": 5815
    },
    {
      "epoch": 0.15638855300282145,
      "grad_norm": 2.7910056114196777,
      "learning_rate": 4.275500476644423e-05,
      "loss": 4.6721,
      "step": 5820
    },
    {
      "epoch": 0.15652290742979982,
      "grad_norm": 2.4527087211608887,
      "learning_rate": 4.274819556039766e-05,
      "loss": 4.6279,
      "step": 5825
    },
    {
      "epoch": 0.15665726185677817,
      "grad_norm": 2.4822309017181396,
      "learning_rate": 4.274138635435109e-05,
      "loss": 4.6326,
      "step": 5830
    },
    {
      "epoch": 0.15679161628375654,
      "grad_norm": 2.3668267726898193,
      "learning_rate": 4.273457714830451e-05,
      "loss": 4.6652,
      "step": 5835
    },
    {
      "epoch": 0.1569259707107349,
      "grad_norm": 2.3012495040893555,
      "learning_rate": 4.2727767942257934e-05,
      "loss": 4.5839,
      "step": 5840
    },
    {
      "epoch": 0.15706032513771329,
      "grad_norm": 2.3303887844085693,
      "learning_rate": 4.2720958736211356e-05,
      "loss": 4.7545,
      "step": 5845
    },
    {
      "epoch": 0.15719467956469166,
      "grad_norm": 2.26655650138855,
      "learning_rate": 4.2714149530164785e-05,
      "loss": 4.5606,
      "step": 5850
    },
    {
      "epoch": 0.15732903399167003,
      "grad_norm": 2.1791794300079346,
      "learning_rate": 4.2707340324118214e-05,
      "loss": 4.6482,
      "step": 5855
    },
    {
      "epoch": 0.1574633884186484,
      "grad_norm": 2.3366024494171143,
      "learning_rate": 4.2700531118071636e-05,
      "loss": 4.6179,
      "step": 5860
    },
    {
      "epoch": 0.15759774284562678,
      "grad_norm": 2.2457427978515625,
      "learning_rate": 4.269372191202506e-05,
      "loss": 4.5906,
      "step": 5865
    },
    {
      "epoch": 0.15773209727260512,
      "grad_norm": 2.2892916202545166,
      "learning_rate": 4.268691270597848e-05,
      "loss": 4.6369,
      "step": 5870
    },
    {
      "epoch": 0.1578664516995835,
      "grad_norm": 2.3485372066497803,
      "learning_rate": 4.268010349993191e-05,
      "loss": 4.7087,
      "step": 5875
    },
    {
      "epoch": 0.15800080612656187,
      "grad_norm": 2.357516288757324,
      "learning_rate": 4.267329429388533e-05,
      "loss": 4.7383,
      "step": 5880
    },
    {
      "epoch": 0.15813516055354024,
      "grad_norm": 2.543384313583374,
      "learning_rate": 4.266648508783876e-05,
      "loss": 4.5478,
      "step": 5885
    },
    {
      "epoch": 0.1582695149805186,
      "grad_norm": 2.70294189453125,
      "learning_rate": 4.2659675881792183e-05,
      "loss": 4.6505,
      "step": 5890
    },
    {
      "epoch": 0.15840386940749698,
      "grad_norm": 2.223876714706421,
      "learning_rate": 4.265286667574561e-05,
      "loss": 4.6407,
      "step": 5895
    },
    {
      "epoch": 0.15853822383447536,
      "grad_norm": 2.4264795780181885,
      "learning_rate": 4.2646057469699035e-05,
      "loss": 4.8008,
      "step": 5900
    },
    {
      "epoch": 0.15867257826145373,
      "grad_norm": 2.2152278423309326,
      "learning_rate": 4.263924826365246e-05,
      "loss": 4.7099,
      "step": 5905
    },
    {
      "epoch": 0.15880693268843207,
      "grad_norm": 2.322824239730835,
      "learning_rate": 4.2632439057605886e-05,
      "loss": 4.7685,
      "step": 5910
    },
    {
      "epoch": 0.15894128711541045,
      "grad_norm": 2.4484431743621826,
      "learning_rate": 4.2625629851559315e-05,
      "loss": 4.6356,
      "step": 5915
    },
    {
      "epoch": 0.15907564154238882,
      "grad_norm": 2.285881280899048,
      "learning_rate": 4.261882064551274e-05,
      "loss": 4.6492,
      "step": 5920
    },
    {
      "epoch": 0.1592099959693672,
      "grad_norm": 2.434908151626587,
      "learning_rate": 4.261201143946616e-05,
      "loss": 4.6614,
      "step": 5925
    },
    {
      "epoch": 0.15934435039634556,
      "grad_norm": 2.502322196960449,
      "learning_rate": 4.260520223341958e-05,
      "loss": 4.7753,
      "step": 5930
    },
    {
      "epoch": 0.15947870482332394,
      "grad_norm": 2.623091697692871,
      "learning_rate": 4.259839302737301e-05,
      "loss": 4.8025,
      "step": 5935
    },
    {
      "epoch": 0.1596130592503023,
      "grad_norm": 2.3283679485321045,
      "learning_rate": 4.259158382132643e-05,
      "loss": 4.5877,
      "step": 5940
    },
    {
      "epoch": 0.15974741367728065,
      "grad_norm": 2.4143717288970947,
      "learning_rate": 4.258477461527986e-05,
      "loss": 4.6031,
      "step": 5945
    },
    {
      "epoch": 0.15988176810425903,
      "grad_norm": 2.351607084274292,
      "learning_rate": 4.2577965409233285e-05,
      "loss": 4.5914,
      "step": 5950
    },
    {
      "epoch": 0.1600161225312374,
      "grad_norm": 2.4132354259490967,
      "learning_rate": 4.2571156203186714e-05,
      "loss": 4.5631,
      "step": 5955
    },
    {
      "epoch": 0.16015047695821577,
      "grad_norm": 2.398253917694092,
      "learning_rate": 4.2564346997140136e-05,
      "loss": 4.6229,
      "step": 5960
    },
    {
      "epoch": 0.16028483138519414,
      "grad_norm": 2.437215566635132,
      "learning_rate": 4.255753779109356e-05,
      "loss": 4.7297,
      "step": 5965
    },
    {
      "epoch": 0.16041918581217252,
      "grad_norm": 2.2300703525543213,
      "learning_rate": 4.255072858504699e-05,
      "loss": 4.5491,
      "step": 5970
    },
    {
      "epoch": 0.1605535402391509,
      "grad_norm": 2.309344530105591,
      "learning_rate": 4.2543919379000416e-05,
      "loss": 4.7037,
      "step": 5975
    },
    {
      "epoch": 0.16068789466612926,
      "grad_norm": 2.2457475662231445,
      "learning_rate": 4.253711017295384e-05,
      "loss": 4.7216,
      "step": 5980
    },
    {
      "epoch": 0.1608222490931076,
      "grad_norm": 2.2409584522247314,
      "learning_rate": 4.253030096690726e-05,
      "loss": 4.6473,
      "step": 5985
    },
    {
      "epoch": 0.16095660352008598,
      "grad_norm": 2.3405046463012695,
      "learning_rate": 4.252349176086068e-05,
      "loss": 4.6641,
      "step": 5990
    },
    {
      "epoch": 0.16109095794706435,
      "grad_norm": 2.4117116928100586,
      "learning_rate": 4.251668255481411e-05,
      "loss": 4.5869,
      "step": 5995
    },
    {
      "epoch": 0.16122531237404272,
      "grad_norm": 2.4327499866485596,
      "learning_rate": 4.2509873348767534e-05,
      "loss": 4.6004,
      "step": 6000
    },
    {
      "epoch": 0.1613596668010211,
      "grad_norm": 2.4647257328033447,
      "learning_rate": 4.250306414272096e-05,
      "loss": 4.6908,
      "step": 6005
    },
    {
      "epoch": 0.16149402122799947,
      "grad_norm": 2.4930577278137207,
      "learning_rate": 4.2496254936674386e-05,
      "loss": 4.7004,
      "step": 6010
    },
    {
      "epoch": 0.16162837565497784,
      "grad_norm": 2.3261513710021973,
      "learning_rate": 4.248944573062781e-05,
      "loss": 4.6225,
      "step": 6015
    },
    {
      "epoch": 0.1617627300819562,
      "grad_norm": 2.5002598762512207,
      "learning_rate": 4.248263652458124e-05,
      "loss": 4.6243,
      "step": 6020
    },
    {
      "epoch": 0.16189708450893456,
      "grad_norm": 2.171928882598877,
      "learning_rate": 4.247582731853466e-05,
      "loss": 4.7203,
      "step": 6025
    },
    {
      "epoch": 0.16203143893591293,
      "grad_norm": 2.506356716156006,
      "learning_rate": 4.246901811248808e-05,
      "loss": 4.6799,
      "step": 6030
    },
    {
      "epoch": 0.1621657933628913,
      "grad_norm": 2.3867156505584717,
      "learning_rate": 4.246220890644151e-05,
      "loss": 4.6373,
      "step": 6035
    },
    {
      "epoch": 0.16230014778986968,
      "grad_norm": 2.478376626968384,
      "learning_rate": 4.245539970039494e-05,
      "loss": 4.7229,
      "step": 6040
    },
    {
      "epoch": 0.16243450221684805,
      "grad_norm": 2.2173712253570557,
      "learning_rate": 4.244859049434836e-05,
      "loss": 4.6154,
      "step": 6045
    },
    {
      "epoch": 0.16256885664382642,
      "grad_norm": 2.357283592224121,
      "learning_rate": 4.2441781288301784e-05,
      "loss": 4.5737,
      "step": 6050
    },
    {
      "epoch": 0.1627032110708048,
      "grad_norm": 2.4080758094787598,
      "learning_rate": 4.2434972082255206e-05,
      "loss": 4.5552,
      "step": 6055
    },
    {
      "epoch": 0.16283756549778317,
      "grad_norm": 2.4244191646575928,
      "learning_rate": 4.2428162876208635e-05,
      "loss": 4.7059,
      "step": 6060
    },
    {
      "epoch": 0.1629719199247615,
      "grad_norm": 2.342439889907837,
      "learning_rate": 4.2421353670162064e-05,
      "loss": 4.5859,
      "step": 6065
    },
    {
      "epoch": 0.16310627435173988,
      "grad_norm": 2.381415605545044,
      "learning_rate": 4.241454446411549e-05,
      "loss": 4.715,
      "step": 6070
    },
    {
      "epoch": 0.16324062877871826,
      "grad_norm": 2.187832832336426,
      "learning_rate": 4.240773525806891e-05,
      "loss": 4.6729,
      "step": 6075
    },
    {
      "epoch": 0.16337498320569663,
      "grad_norm": 2.2844033241271973,
      "learning_rate": 4.240092605202234e-05,
      "loss": 4.6235,
      "step": 6080
    },
    {
      "epoch": 0.163509337632675,
      "grad_norm": 2.1520915031433105,
      "learning_rate": 4.239411684597576e-05,
      "loss": 4.6282,
      "step": 6085
    },
    {
      "epoch": 0.16364369205965337,
      "grad_norm": 2.274156332015991,
      "learning_rate": 4.238730763992918e-05,
      "loss": 4.5433,
      "step": 6090
    },
    {
      "epoch": 0.16377804648663175,
      "grad_norm": 2.2229771614074707,
      "learning_rate": 4.238049843388261e-05,
      "loss": 4.7427,
      "step": 6095
    },
    {
      "epoch": 0.1639124009136101,
      "grad_norm": 2.12465238571167,
      "learning_rate": 4.237368922783604e-05,
      "loss": 4.6112,
      "step": 6100
    },
    {
      "epoch": 0.16404675534058846,
      "grad_norm": 2.288067579269409,
      "learning_rate": 4.236688002178946e-05,
      "loss": 4.5895,
      "step": 6105
    },
    {
      "epoch": 0.16418110976756684,
      "grad_norm": 2.296560525894165,
      "learning_rate": 4.2360070815742885e-05,
      "loss": 4.5356,
      "step": 6110
    },
    {
      "epoch": 0.1643154641945452,
      "grad_norm": 2.351196050643921,
      "learning_rate": 4.235326160969631e-05,
      "loss": 4.6556,
      "step": 6115
    },
    {
      "epoch": 0.16444981862152358,
      "grad_norm": 2.287306785583496,
      "learning_rate": 4.2346452403649737e-05,
      "loss": 4.5939,
      "step": 6120
    },
    {
      "epoch": 0.16458417304850195,
      "grad_norm": 2.4288675785064697,
      "learning_rate": 4.2339643197603166e-05,
      "loss": 4.5915,
      "step": 6125
    },
    {
      "epoch": 0.16471852747548033,
      "grad_norm": 2.171098470687866,
      "learning_rate": 4.233283399155659e-05,
      "loss": 4.5222,
      "step": 6130
    },
    {
      "epoch": 0.1648528819024587,
      "grad_norm": 2.141940116882324,
      "learning_rate": 4.232602478551001e-05,
      "loss": 4.6666,
      "step": 6135
    },
    {
      "epoch": 0.16498723632943704,
      "grad_norm": 2.3677592277526855,
      "learning_rate": 4.231921557946344e-05,
      "loss": 4.6978,
      "step": 6140
    },
    {
      "epoch": 0.16512159075641542,
      "grad_norm": 2.283712863922119,
      "learning_rate": 4.231240637341686e-05,
      "loss": 4.7622,
      "step": 6145
    },
    {
      "epoch": 0.1652559451833938,
      "grad_norm": 2.208111047744751,
      "learning_rate": 4.2305597167370284e-05,
      "loss": 4.697,
      "step": 6150
    },
    {
      "epoch": 0.16539029961037216,
      "grad_norm": 2.327491521835327,
      "learning_rate": 4.229878796132371e-05,
      "loss": 4.5452,
      "step": 6155
    },
    {
      "epoch": 0.16552465403735053,
      "grad_norm": 2.6018028259277344,
      "learning_rate": 4.2291978755277135e-05,
      "loss": 4.655,
      "step": 6160
    },
    {
      "epoch": 0.1656590084643289,
      "grad_norm": 2.3526546955108643,
      "learning_rate": 4.2285169549230564e-05,
      "loss": 4.5395,
      "step": 6165
    },
    {
      "epoch": 0.16579336289130728,
      "grad_norm": 2.5430386066436768,
      "learning_rate": 4.2278360343183986e-05,
      "loss": 4.5327,
      "step": 6170
    },
    {
      "epoch": 0.16592771731828565,
      "grad_norm": 2.2715277671813965,
      "learning_rate": 4.227155113713741e-05,
      "loss": 4.5757,
      "step": 6175
    },
    {
      "epoch": 0.166062071745264,
      "grad_norm": 2.0955557823181152,
      "learning_rate": 4.226474193109084e-05,
      "loss": 4.6361,
      "step": 6180
    },
    {
      "epoch": 0.16619642617224237,
      "grad_norm": 2.400583267211914,
      "learning_rate": 4.225793272504427e-05,
      "loss": 4.6402,
      "step": 6185
    },
    {
      "epoch": 0.16633078059922074,
      "grad_norm": 2.170935869216919,
      "learning_rate": 4.225112351899769e-05,
      "loss": 4.757,
      "step": 6190
    },
    {
      "epoch": 0.1664651350261991,
      "grad_norm": 2.524139404296875,
      "learning_rate": 4.224431431295111e-05,
      "loss": 4.621,
      "step": 6195
    },
    {
      "epoch": 0.16659948945317749,
      "grad_norm": 2.2586677074432373,
      "learning_rate": 4.2237505106904534e-05,
      "loss": 4.5916,
      "step": 6200
    },
    {
      "epoch": 0.16673384388015586,
      "grad_norm": 2.3482825756073,
      "learning_rate": 4.223069590085796e-05,
      "loss": 4.6336,
      "step": 6205
    },
    {
      "epoch": 0.16686819830713423,
      "grad_norm": 2.035543918609619,
      "learning_rate": 4.2223886694811385e-05,
      "loss": 4.6313,
      "step": 6210
    },
    {
      "epoch": 0.1670025527341126,
      "grad_norm": 2.2824630737304688,
      "learning_rate": 4.2217077488764814e-05,
      "loss": 4.5794,
      "step": 6215
    },
    {
      "epoch": 0.16713690716109095,
      "grad_norm": 2.4770102500915527,
      "learning_rate": 4.2210268282718236e-05,
      "loss": 4.7118,
      "step": 6220
    },
    {
      "epoch": 0.16727126158806932,
      "grad_norm": 2.48583722114563,
      "learning_rate": 4.2203459076671665e-05,
      "loss": 4.6572,
      "step": 6225
    },
    {
      "epoch": 0.1674056160150477,
      "grad_norm": 2.1995291709899902,
      "learning_rate": 4.219664987062509e-05,
      "loss": 4.6766,
      "step": 6230
    },
    {
      "epoch": 0.16753997044202606,
      "grad_norm": 2.4620847702026367,
      "learning_rate": 4.218984066457851e-05,
      "loss": 4.6487,
      "step": 6235
    },
    {
      "epoch": 0.16767432486900444,
      "grad_norm": 2.292832374572754,
      "learning_rate": 4.218303145853193e-05,
      "loss": 4.7578,
      "step": 6240
    },
    {
      "epoch": 0.1678086792959828,
      "grad_norm": 2.2099673748016357,
      "learning_rate": 4.217622225248537e-05,
      "loss": 4.6665,
      "step": 6245
    },
    {
      "epoch": 0.16794303372296118,
      "grad_norm": 2.0643885135650635,
      "learning_rate": 4.216941304643879e-05,
      "loss": 4.7198,
      "step": 6250
    },
    {
      "epoch": 0.16807738814993953,
      "grad_norm": 2.1985578536987305,
      "learning_rate": 4.216260384039221e-05,
      "loss": 4.6381,
      "step": 6255
    },
    {
      "epoch": 0.1682117425769179,
      "grad_norm": 2.3217663764953613,
      "learning_rate": 4.2155794634345635e-05,
      "loss": 4.6219,
      "step": 6260
    },
    {
      "epoch": 0.16834609700389627,
      "grad_norm": 2.3149125576019287,
      "learning_rate": 4.2148985428299064e-05,
      "loss": 4.6082,
      "step": 6265
    },
    {
      "epoch": 0.16848045143087464,
      "grad_norm": 2.496859073638916,
      "learning_rate": 4.2142176222252486e-05,
      "loss": 4.7181,
      "step": 6270
    },
    {
      "epoch": 0.16861480585785302,
      "grad_norm": 2.0511131286621094,
      "learning_rate": 4.2135367016205915e-05,
      "loss": 4.7112,
      "step": 6275
    },
    {
      "epoch": 0.1687491602848314,
      "grad_norm": 2.260448455810547,
      "learning_rate": 4.212855781015934e-05,
      "loss": 4.5699,
      "step": 6280
    },
    {
      "epoch": 0.16888351471180976,
      "grad_norm": 2.306929349899292,
      "learning_rate": 4.2121748604112766e-05,
      "loss": 4.6419,
      "step": 6285
    },
    {
      "epoch": 0.16901786913878813,
      "grad_norm": 2.1921920776367188,
      "learning_rate": 4.211493939806619e-05,
      "loss": 4.7025,
      "step": 6290
    },
    {
      "epoch": 0.16915222356576648,
      "grad_norm": 2.4268269538879395,
      "learning_rate": 4.210813019201961e-05,
      "loss": 4.5908,
      "step": 6295
    },
    {
      "epoch": 0.16928657799274485,
      "grad_norm": 2.3178722858428955,
      "learning_rate": 4.210132098597303e-05,
      "loss": 4.6871,
      "step": 6300
    },
    {
      "epoch": 0.16942093241972322,
      "grad_norm": 2.2124123573303223,
      "learning_rate": 4.209451177992646e-05,
      "loss": 4.7646,
      "step": 6305
    },
    {
      "epoch": 0.1695552868467016,
      "grad_norm": 2.371185779571533,
      "learning_rate": 4.208770257387989e-05,
      "loss": 4.5945,
      "step": 6310
    },
    {
      "epoch": 0.16968964127367997,
      "grad_norm": 2.21028208732605,
      "learning_rate": 4.2080893367833313e-05,
      "loss": 4.6358,
      "step": 6315
    },
    {
      "epoch": 0.16982399570065834,
      "grad_norm": 2.199416160583496,
      "learning_rate": 4.2074084161786736e-05,
      "loss": 4.7773,
      "step": 6320
    },
    {
      "epoch": 0.16995835012763671,
      "grad_norm": 2.350001096725464,
      "learning_rate": 4.206727495574016e-05,
      "loss": 4.5694,
      "step": 6325
    },
    {
      "epoch": 0.1700927045546151,
      "grad_norm": 2.2529332637786865,
      "learning_rate": 4.206046574969359e-05,
      "loss": 4.7092,
      "step": 6330
    },
    {
      "epoch": 0.17022705898159343,
      "grad_norm": 2.588395357131958,
      "learning_rate": 4.2053656543647016e-05,
      "loss": 4.6163,
      "step": 6335
    },
    {
      "epoch": 0.1703614134085718,
      "grad_norm": 2.208899974822998,
      "learning_rate": 4.204684733760044e-05,
      "loss": 4.5044,
      "step": 6340
    },
    {
      "epoch": 0.17049576783555018,
      "grad_norm": 2.4447290897369385,
      "learning_rate": 4.204003813155386e-05,
      "loss": 4.6403,
      "step": 6345
    },
    {
      "epoch": 0.17063012226252855,
      "grad_norm": 2.3725602626800537,
      "learning_rate": 4.203322892550729e-05,
      "loss": 4.6863,
      "step": 6350
    },
    {
      "epoch": 0.17076447668950692,
      "grad_norm": 2.3884928226470947,
      "learning_rate": 4.202641971946071e-05,
      "loss": 4.6021,
      "step": 6355
    },
    {
      "epoch": 0.1708988311164853,
      "grad_norm": 2.3537652492523193,
      "learning_rate": 4.2019610513414134e-05,
      "loss": 4.6261,
      "step": 6360
    },
    {
      "epoch": 0.17103318554346367,
      "grad_norm": 2.136301040649414,
      "learning_rate": 4.201280130736756e-05,
      "loss": 4.649,
      "step": 6365
    },
    {
      "epoch": 0.17116753997044204,
      "grad_norm": 2.3084094524383545,
      "learning_rate": 4.200599210132099e-05,
      "loss": 4.5735,
      "step": 6370
    },
    {
      "epoch": 0.17130189439742038,
      "grad_norm": 2.2184677124023438,
      "learning_rate": 4.1999182895274415e-05,
      "loss": 4.4994,
      "step": 6375
    },
    {
      "epoch": 0.17143624882439876,
      "grad_norm": 2.2734274864196777,
      "learning_rate": 4.199237368922784e-05,
      "loss": 4.4762,
      "step": 6380
    },
    {
      "epoch": 0.17157060325137713,
      "grad_norm": 2.3786418437957764,
      "learning_rate": 4.198556448318126e-05,
      "loss": 4.5552,
      "step": 6385
    },
    {
      "epoch": 0.1717049576783555,
      "grad_norm": 2.3438937664031982,
      "learning_rate": 4.197875527713469e-05,
      "loss": 4.782,
      "step": 6390
    },
    {
      "epoch": 0.17183931210533387,
      "grad_norm": 2.4480435848236084,
      "learning_rate": 4.197194607108812e-05,
      "loss": 4.7028,
      "step": 6395
    },
    {
      "epoch": 0.17197366653231225,
      "grad_norm": 2.3457813262939453,
      "learning_rate": 4.196513686504154e-05,
      "loss": 4.6757,
      "step": 6400
    },
    {
      "epoch": 0.17210802095929062,
      "grad_norm": 2.4116549491882324,
      "learning_rate": 4.195832765899496e-05,
      "loss": 4.5225,
      "step": 6405
    },
    {
      "epoch": 0.17224237538626896,
      "grad_norm": 2.4665629863739014,
      "learning_rate": 4.195151845294839e-05,
      "loss": 4.6547,
      "step": 6410
    },
    {
      "epoch": 0.17237672981324734,
      "grad_norm": 2.513230562210083,
      "learning_rate": 4.194470924690181e-05,
      "loss": 4.5844,
      "step": 6415
    },
    {
      "epoch": 0.1725110842402257,
      "grad_norm": 2.404723644256592,
      "learning_rate": 4.1937900040855235e-05,
      "loss": 4.6189,
      "step": 6420
    },
    {
      "epoch": 0.17264543866720408,
      "grad_norm": 2.268045425415039,
      "learning_rate": 4.1931090834808664e-05,
      "loss": 4.5955,
      "step": 6425
    },
    {
      "epoch": 0.17277979309418245,
      "grad_norm": 2.355091094970703,
      "learning_rate": 4.1924281628762093e-05,
      "loss": 4.634,
      "step": 6430
    },
    {
      "epoch": 0.17291414752116083,
      "grad_norm": 2.2507715225219727,
      "learning_rate": 4.1917472422715516e-05,
      "loss": 4.6452,
      "step": 6435
    },
    {
      "epoch": 0.1730485019481392,
      "grad_norm": 2.3553433418273926,
      "learning_rate": 4.191066321666894e-05,
      "loss": 4.7423,
      "step": 6440
    },
    {
      "epoch": 0.17318285637511757,
      "grad_norm": 2.272806406021118,
      "learning_rate": 4.190385401062236e-05,
      "loss": 4.567,
      "step": 6445
    },
    {
      "epoch": 0.17331721080209592,
      "grad_norm": 2.4044992923736572,
      "learning_rate": 4.189704480457578e-05,
      "loss": 4.5237,
      "step": 6450
    },
    {
      "epoch": 0.1734515652290743,
      "grad_norm": 2.1048424243927,
      "learning_rate": 4.189023559852922e-05,
      "loss": 4.6141,
      "step": 6455
    },
    {
      "epoch": 0.17358591965605266,
      "grad_norm": 2.1680407524108887,
      "learning_rate": 4.188342639248264e-05,
      "loss": 4.6585,
      "step": 6460
    },
    {
      "epoch": 0.17372027408303103,
      "grad_norm": 2.3715052604675293,
      "learning_rate": 4.187661718643606e-05,
      "loss": 4.7898,
      "step": 6465
    },
    {
      "epoch": 0.1738546285100094,
      "grad_norm": 2.68088960647583,
      "learning_rate": 4.1869807980389485e-05,
      "loss": 4.493,
      "step": 6470
    },
    {
      "epoch": 0.17398898293698778,
      "grad_norm": 2.558058023452759,
      "learning_rate": 4.1862998774342914e-05,
      "loss": 4.6277,
      "step": 6475
    },
    {
      "epoch": 0.17412333736396615,
      "grad_norm": 2.171090841293335,
      "learning_rate": 4.1856189568296336e-05,
      "loss": 4.6525,
      "step": 6480
    },
    {
      "epoch": 0.17425769179094452,
      "grad_norm": 2.1461739540100098,
      "learning_rate": 4.1849380362249765e-05,
      "loss": 4.5093,
      "step": 6485
    },
    {
      "epoch": 0.17439204621792287,
      "grad_norm": 2.1991193294525146,
      "learning_rate": 4.184257115620319e-05,
      "loss": 4.7081,
      "step": 6490
    },
    {
      "epoch": 0.17452640064490124,
      "grad_norm": 2.47517728805542,
      "learning_rate": 4.183576195015662e-05,
      "loss": 4.5958,
      "step": 6495
    },
    {
      "epoch": 0.17466075507187961,
      "grad_norm": 2.183870792388916,
      "learning_rate": 4.182895274411004e-05,
      "loss": 4.6201,
      "step": 6500
    },
    {
      "epoch": 0.174795109498858,
      "grad_norm": 2.359966993331909,
      "learning_rate": 4.182214353806346e-05,
      "loss": 4.6841,
      "step": 6505
    },
    {
      "epoch": 0.17492946392583636,
      "grad_norm": 2.2989532947540283,
      "learning_rate": 4.1815334332016884e-05,
      "loss": 4.6966,
      "step": 6510
    },
    {
      "epoch": 0.17506381835281473,
      "grad_norm": 2.254702091217041,
      "learning_rate": 4.180852512597031e-05,
      "loss": 4.6757,
      "step": 6515
    },
    {
      "epoch": 0.1751981727797931,
      "grad_norm": 2.3578250408172607,
      "learning_rate": 4.180171591992374e-05,
      "loss": 4.6516,
      "step": 6520
    },
    {
      "epoch": 0.17533252720677148,
      "grad_norm": 2.4930083751678467,
      "learning_rate": 4.1794906713877164e-05,
      "loss": 4.731,
      "step": 6525
    },
    {
      "epoch": 0.17546688163374982,
      "grad_norm": 2.198848009109497,
      "learning_rate": 4.1788097507830586e-05,
      "loss": 4.7227,
      "step": 6530
    },
    {
      "epoch": 0.1756012360607282,
      "grad_norm": 2.4210662841796875,
      "learning_rate": 4.1781288301784015e-05,
      "loss": 4.5918,
      "step": 6535
    },
    {
      "epoch": 0.17573559048770657,
      "grad_norm": 2.58354115486145,
      "learning_rate": 4.177447909573744e-05,
      "loss": 4.6029,
      "step": 6540
    },
    {
      "epoch": 0.17586994491468494,
      "grad_norm": 2.2262752056121826,
      "learning_rate": 4.1767669889690867e-05,
      "loss": 4.485,
      "step": 6545
    },
    {
      "epoch": 0.1760042993416633,
      "grad_norm": 2.6898539066314697,
      "learning_rate": 4.176086068364429e-05,
      "loss": 4.6133,
      "step": 6550
    },
    {
      "epoch": 0.17613865376864168,
      "grad_norm": 2.4759645462036133,
      "learning_rate": 4.175405147759772e-05,
      "loss": 4.7214,
      "step": 6555
    },
    {
      "epoch": 0.17627300819562006,
      "grad_norm": 2.338465929031372,
      "learning_rate": 4.174724227155114e-05,
      "loss": 4.6771,
      "step": 6560
    },
    {
      "epoch": 0.1764073626225984,
      "grad_norm": 2.3396828174591064,
      "learning_rate": 4.174043306550456e-05,
      "loss": 4.5749,
      "step": 6565
    },
    {
      "epoch": 0.17654171704957677,
      "grad_norm": 2.555844783782959,
      "learning_rate": 4.1733623859457985e-05,
      "loss": 4.4251,
      "step": 6570
    },
    {
      "epoch": 0.17667607147655515,
      "grad_norm": 2.3667538166046143,
      "learning_rate": 4.1726814653411414e-05,
      "loss": 4.6276,
      "step": 6575
    },
    {
      "epoch": 0.17681042590353352,
      "grad_norm": 2.3648812770843506,
      "learning_rate": 4.172000544736484e-05,
      "loss": 4.5702,
      "step": 6580
    },
    {
      "epoch": 0.1769447803305119,
      "grad_norm": 2.0452256202697754,
      "learning_rate": 4.1713196241318265e-05,
      "loss": 4.5153,
      "step": 6585
    },
    {
      "epoch": 0.17707913475749026,
      "grad_norm": 2.5574960708618164,
      "learning_rate": 4.170638703527169e-05,
      "loss": 4.5806,
      "step": 6590
    },
    {
      "epoch": 0.17721348918446864,
      "grad_norm": 2.6324303150177,
      "learning_rate": 4.169957782922511e-05,
      "loss": 4.5472,
      "step": 6595
    },
    {
      "epoch": 0.177347843611447,
      "grad_norm": 2.534785747528076,
      "learning_rate": 4.169276862317854e-05,
      "loss": 4.594,
      "step": 6600
    },
    {
      "epoch": 0.17748219803842535,
      "grad_norm": 2.409205913543701,
      "learning_rate": 4.168595941713197e-05,
      "loss": 4.6887,
      "step": 6605
    },
    {
      "epoch": 0.17761655246540373,
      "grad_norm": 2.389843463897705,
      "learning_rate": 4.167915021108539e-05,
      "loss": 4.5704,
      "step": 6610
    },
    {
      "epoch": 0.1777509068923821,
      "grad_norm": 2.202401638031006,
      "learning_rate": 4.167234100503881e-05,
      "loss": 4.5835,
      "step": 6615
    },
    {
      "epoch": 0.17788526131936047,
      "grad_norm": 2.4570469856262207,
      "learning_rate": 4.166553179899224e-05,
      "loss": 4.6597,
      "step": 6620
    },
    {
      "epoch": 0.17801961574633884,
      "grad_norm": 2.570322036743164,
      "learning_rate": 4.1658722592945664e-05,
      "loss": 4.4766,
      "step": 6625
    },
    {
      "epoch": 0.17815397017331722,
      "grad_norm": 2.451951026916504,
      "learning_rate": 4.1651913386899086e-05,
      "loss": 4.6803,
      "step": 6630
    },
    {
      "epoch": 0.1782883246002956,
      "grad_norm": 2.2731688022613525,
      "learning_rate": 4.1645104180852515e-05,
      "loss": 4.61,
      "step": 6635
    },
    {
      "epoch": 0.17842267902727396,
      "grad_norm": 2.4533755779266357,
      "learning_rate": 4.1638294974805944e-05,
      "loss": 4.6184,
      "step": 6640
    },
    {
      "epoch": 0.1785570334542523,
      "grad_norm": 2.3204565048217773,
      "learning_rate": 4.1631485768759366e-05,
      "loss": 4.5813,
      "step": 6645
    },
    {
      "epoch": 0.17869138788123068,
      "grad_norm": 2.2887136936187744,
      "learning_rate": 4.162467656271279e-05,
      "loss": 4.5791,
      "step": 6650
    },
    {
      "epoch": 0.17882574230820905,
      "grad_norm": 2.4397172927856445,
      "learning_rate": 4.161786735666621e-05,
      "loss": 4.43,
      "step": 6655
    },
    {
      "epoch": 0.17896009673518742,
      "grad_norm": 2.2695908546447754,
      "learning_rate": 4.161105815061964e-05,
      "loss": 4.4514,
      "step": 6660
    },
    {
      "epoch": 0.1790944511621658,
      "grad_norm": 2.322587013244629,
      "learning_rate": 4.160424894457306e-05,
      "loss": 4.6821,
      "step": 6665
    },
    {
      "epoch": 0.17922880558914417,
      "grad_norm": 2.3344407081604004,
      "learning_rate": 4.159743973852649e-05,
      "loss": 4.7318,
      "step": 6670
    },
    {
      "epoch": 0.17936316001612254,
      "grad_norm": 2.229515790939331,
      "learning_rate": 4.159063053247991e-05,
      "loss": 4.6057,
      "step": 6675
    },
    {
      "epoch": 0.17949751444310091,
      "grad_norm": 2.249580144882202,
      "learning_rate": 4.158382132643334e-05,
      "loss": 4.5167,
      "step": 6680
    },
    {
      "epoch": 0.17963186887007926,
      "grad_norm": 2.192352056503296,
      "learning_rate": 4.1577012120386765e-05,
      "loss": 4.6519,
      "step": 6685
    },
    {
      "epoch": 0.17976622329705763,
      "grad_norm": 2.2247118949890137,
      "learning_rate": 4.157020291434019e-05,
      "loss": 4.5879,
      "step": 6690
    },
    {
      "epoch": 0.179900577724036,
      "grad_norm": 2.4266550540924072,
      "learning_rate": 4.1563393708293616e-05,
      "loss": 4.776,
      "step": 6695
    },
    {
      "epoch": 0.18003493215101438,
      "grad_norm": 2.397467613220215,
      "learning_rate": 4.1556584502247045e-05,
      "loss": 4.6221,
      "step": 6700
    },
    {
      "epoch": 0.18016928657799275,
      "grad_norm": 2.3627591133117676,
      "learning_rate": 4.154977529620047e-05,
      "loss": 4.6666,
      "step": 6705
    },
    {
      "epoch": 0.18030364100497112,
      "grad_norm": 2.3746418952941895,
      "learning_rate": 4.154296609015389e-05,
      "loss": 4.5638,
      "step": 6710
    },
    {
      "epoch": 0.1804379954319495,
      "grad_norm": 2.334683895111084,
      "learning_rate": 4.153615688410731e-05,
      "loss": 4.4996,
      "step": 6715
    },
    {
      "epoch": 0.18057234985892784,
      "grad_norm": 2.2431259155273438,
      "learning_rate": 4.152934767806074e-05,
      "loss": 4.6149,
      "step": 6720
    },
    {
      "epoch": 0.1807067042859062,
      "grad_norm": 2.3584394454956055,
      "learning_rate": 4.152253847201416e-05,
      "loss": 4.5275,
      "step": 6725
    },
    {
      "epoch": 0.18084105871288458,
      "grad_norm": 2.269402265548706,
      "learning_rate": 4.151572926596759e-05,
      "loss": 4.4714,
      "step": 6730
    },
    {
      "epoch": 0.18097541313986296,
      "grad_norm": 2.4162747859954834,
      "learning_rate": 4.1508920059921014e-05,
      "loss": 4.6992,
      "step": 6735
    },
    {
      "epoch": 0.18110976756684133,
      "grad_norm": 2.2333920001983643,
      "learning_rate": 4.1502110853874444e-05,
      "loss": 4.69,
      "step": 6740
    },
    {
      "epoch": 0.1812441219938197,
      "grad_norm": 2.289665699005127,
      "learning_rate": 4.1495301647827866e-05,
      "loss": 4.6365,
      "step": 6745
    },
    {
      "epoch": 0.18137847642079807,
      "grad_norm": 2.664175033569336,
      "learning_rate": 4.148849244178129e-05,
      "loss": 4.5822,
      "step": 6750
    },
    {
      "epoch": 0.18151283084777645,
      "grad_norm": 2.292074680328369,
      "learning_rate": 4.148168323573472e-05,
      "loss": 4.4943,
      "step": 6755
    },
    {
      "epoch": 0.1816471852747548,
      "grad_norm": 2.5145933628082275,
      "learning_rate": 4.147487402968814e-05,
      "loss": 4.6949,
      "step": 6760
    },
    {
      "epoch": 0.18178153970173316,
      "grad_norm": 2.436025619506836,
      "learning_rate": 4.146806482364157e-05,
      "loss": 4.5713,
      "step": 6765
    },
    {
      "epoch": 0.18191589412871154,
      "grad_norm": 2.1950080394744873,
      "learning_rate": 4.146125561759499e-05,
      "loss": 4.577,
      "step": 6770
    },
    {
      "epoch": 0.1820502485556899,
      "grad_norm": 2.2889373302459717,
      "learning_rate": 4.145444641154841e-05,
      "loss": 4.5313,
      "step": 6775
    },
    {
      "epoch": 0.18218460298266828,
      "grad_norm": 2.360623836517334,
      "learning_rate": 4.1447637205501835e-05,
      "loss": 4.5529,
      "step": 6780
    },
    {
      "epoch": 0.18231895740964665,
      "grad_norm": 2.2756640911102295,
      "learning_rate": 4.1440827999455264e-05,
      "loss": 4.6463,
      "step": 6785
    },
    {
      "epoch": 0.18245331183662503,
      "grad_norm": 2.3278796672821045,
      "learning_rate": 4.143401879340869e-05,
      "loss": 4.6218,
      "step": 6790
    },
    {
      "epoch": 0.1825876662636034,
      "grad_norm": 2.2654950618743896,
      "learning_rate": 4.1427209587362116e-05,
      "loss": 4.6451,
      "step": 6795
    },
    {
      "epoch": 0.18272202069058174,
      "grad_norm": 2.2796127796173096,
      "learning_rate": 4.142040038131554e-05,
      "loss": 4.5603,
      "step": 6800
    },
    {
      "epoch": 0.18285637511756012,
      "grad_norm": 2.37497878074646,
      "learning_rate": 4.141359117526897e-05,
      "loss": 4.5774,
      "step": 6805
    },
    {
      "epoch": 0.1829907295445385,
      "grad_norm": 2.328749895095825,
      "learning_rate": 4.140678196922239e-05,
      "loss": 4.5676,
      "step": 6810
    },
    {
      "epoch": 0.18312508397151686,
      "grad_norm": 2.256537437438965,
      "learning_rate": 4.139997276317582e-05,
      "loss": 4.6365,
      "step": 6815
    },
    {
      "epoch": 0.18325943839849523,
      "grad_norm": 2.4028425216674805,
      "learning_rate": 4.139316355712924e-05,
      "loss": 4.4725,
      "step": 6820
    },
    {
      "epoch": 0.1833937928254736,
      "grad_norm": 2.1181318759918213,
      "learning_rate": 4.138635435108267e-05,
      "loss": 4.6774,
      "step": 6825
    },
    {
      "epoch": 0.18352814725245198,
      "grad_norm": 2.459195852279663,
      "learning_rate": 4.137954514503609e-05,
      "loss": 4.7415,
      "step": 6830
    },
    {
      "epoch": 0.18366250167943035,
      "grad_norm": 2.4519217014312744,
      "learning_rate": 4.1372735938989514e-05,
      "loss": 4.5747,
      "step": 6835
    },
    {
      "epoch": 0.1837968561064087,
      "grad_norm": 2.4589717388153076,
      "learning_rate": 4.1365926732942936e-05,
      "loss": 4.5586,
      "step": 6840
    },
    {
      "epoch": 0.18393121053338707,
      "grad_norm": 2.2202322483062744,
      "learning_rate": 4.1359117526896365e-05,
      "loss": 4.5939,
      "step": 6845
    },
    {
      "epoch": 0.18406556496036544,
      "grad_norm": 2.31744647026062,
      "learning_rate": 4.1352308320849794e-05,
      "loss": 4.5411,
      "step": 6850
    },
    {
      "epoch": 0.1841999193873438,
      "grad_norm": 2.469985008239746,
      "learning_rate": 4.134549911480322e-05,
      "loss": 4.686,
      "step": 6855
    },
    {
      "epoch": 0.18433427381432219,
      "grad_norm": 2.34891414642334,
      "learning_rate": 4.133868990875664e-05,
      "loss": 4.5964,
      "step": 6860
    },
    {
      "epoch": 0.18446862824130056,
      "grad_norm": 2.565234661102295,
      "learning_rate": 4.133188070271007e-05,
      "loss": 4.354,
      "step": 6865
    },
    {
      "epoch": 0.18460298266827893,
      "grad_norm": 2.820161819458008,
      "learning_rate": 4.132507149666349e-05,
      "loss": 4.5853,
      "step": 6870
    },
    {
      "epoch": 0.18473733709525728,
      "grad_norm": 2.431040048599243,
      "learning_rate": 4.131826229061691e-05,
      "loss": 4.5148,
      "step": 6875
    },
    {
      "epoch": 0.18487169152223565,
      "grad_norm": 2.17570161819458,
      "learning_rate": 4.131145308457034e-05,
      "loss": 4.5879,
      "step": 6880
    },
    {
      "epoch": 0.18500604594921402,
      "grad_norm": 2.269704580307007,
      "learning_rate": 4.130464387852377e-05,
      "loss": 4.6893,
      "step": 6885
    },
    {
      "epoch": 0.1851404003761924,
      "grad_norm": 2.3647964000701904,
      "learning_rate": 4.129783467247719e-05,
      "loss": 4.5912,
      "step": 6890
    },
    {
      "epoch": 0.18527475480317077,
      "grad_norm": 2.4728124141693115,
      "learning_rate": 4.1291025466430615e-05,
      "loss": 4.5582,
      "step": 6895
    },
    {
      "epoch": 0.18540910923014914,
      "grad_norm": 2.4384915828704834,
      "learning_rate": 4.128421626038404e-05,
      "loss": 4.6313,
      "step": 6900
    },
    {
      "epoch": 0.1855434636571275,
      "grad_norm": 2.3080649375915527,
      "learning_rate": 4.1277407054337467e-05,
      "loss": 4.51,
      "step": 6905
    },
    {
      "epoch": 0.18567781808410588,
      "grad_norm": 2.2000319957733154,
      "learning_rate": 4.1270597848290896e-05,
      "loss": 4.5785,
      "step": 6910
    },
    {
      "epoch": 0.18581217251108423,
      "grad_norm": 2.1962084770202637,
      "learning_rate": 4.126378864224432e-05,
      "loss": 4.4365,
      "step": 6915
    },
    {
      "epoch": 0.1859465269380626,
      "grad_norm": 2.268049478530884,
      "learning_rate": 4.125697943619774e-05,
      "loss": 4.6267,
      "step": 6920
    },
    {
      "epoch": 0.18608088136504097,
      "grad_norm": 2.5571939945220947,
      "learning_rate": 4.125017023015116e-05,
      "loss": 4.7028,
      "step": 6925
    },
    {
      "epoch": 0.18621523579201935,
      "grad_norm": 2.339184284210205,
      "learning_rate": 4.124336102410459e-05,
      "loss": 4.5349,
      "step": 6930
    },
    {
      "epoch": 0.18634959021899772,
      "grad_norm": 2.3793368339538574,
      "learning_rate": 4.1236551818058014e-05,
      "loss": 4.6581,
      "step": 6935
    },
    {
      "epoch": 0.1864839446459761,
      "grad_norm": 2.4376611709594727,
      "learning_rate": 4.122974261201144e-05,
      "loss": 4.653,
      "step": 6940
    },
    {
      "epoch": 0.18661829907295446,
      "grad_norm": 2.4305341243743896,
      "learning_rate": 4.1222933405964865e-05,
      "loss": 4.6353,
      "step": 6945
    },
    {
      "epoch": 0.18675265349993284,
      "grad_norm": 2.409497022628784,
      "learning_rate": 4.1216124199918294e-05,
      "loss": 4.5536,
      "step": 6950
    },
    {
      "epoch": 0.18688700792691118,
      "grad_norm": 2.2604665756225586,
      "learning_rate": 4.1209314993871716e-05,
      "loss": 4.6089,
      "step": 6955
    },
    {
      "epoch": 0.18702136235388955,
      "grad_norm": 2.374154806137085,
      "learning_rate": 4.120250578782514e-05,
      "loss": 4.635,
      "step": 6960
    },
    {
      "epoch": 0.18715571678086793,
      "grad_norm": 2.243194341659546,
      "learning_rate": 4.119569658177857e-05,
      "loss": 4.726,
      "step": 6965
    },
    {
      "epoch": 0.1872900712078463,
      "grad_norm": 2.1959033012390137,
      "learning_rate": 4.1188887375732e-05,
      "loss": 4.6781,
      "step": 6970
    },
    {
      "epoch": 0.18742442563482467,
      "grad_norm": 2.6171998977661133,
      "learning_rate": 4.118207816968542e-05,
      "loss": 4.6393,
      "step": 6975
    },
    {
      "epoch": 0.18755878006180304,
      "grad_norm": 2.568899393081665,
      "learning_rate": 4.117526896363884e-05,
      "loss": 4.6268,
      "step": 6980
    },
    {
      "epoch": 0.18769313448878142,
      "grad_norm": 2.537750244140625,
      "learning_rate": 4.1168459757592263e-05,
      "loss": 4.6418,
      "step": 6985
    },
    {
      "epoch": 0.18782748891575976,
      "grad_norm": 2.4498329162597656,
      "learning_rate": 4.116165055154569e-05,
      "loss": 4.527,
      "step": 6990
    },
    {
      "epoch": 0.18796184334273813,
      "grad_norm": 2.211693286895752,
      "learning_rate": 4.1154841345499115e-05,
      "loss": 4.4841,
      "step": 6995
    },
    {
      "epoch": 0.1880961977697165,
      "grad_norm": 2.204399347305298,
      "learning_rate": 4.1148032139452544e-05,
      "loss": 4.5687,
      "step": 7000
    },
    {
      "epoch": 0.18823055219669488,
      "grad_norm": 2.229062795639038,
      "learning_rate": 4.1141222933405966e-05,
      "loss": 4.6291,
      "step": 7005
    },
    {
      "epoch": 0.18836490662367325,
      "grad_norm": 2.236020565032959,
      "learning_rate": 4.1134413727359395e-05,
      "loss": 4.5222,
      "step": 7010
    },
    {
      "epoch": 0.18849926105065162,
      "grad_norm": 2.2904069423675537,
      "learning_rate": 4.112760452131282e-05,
      "loss": 4.7043,
      "step": 7015
    },
    {
      "epoch": 0.18863361547763,
      "grad_norm": 2.5037553310394287,
      "learning_rate": 4.112079531526624e-05,
      "loss": 4.4712,
      "step": 7020
    },
    {
      "epoch": 0.18876796990460837,
      "grad_norm": 2.514737606048584,
      "learning_rate": 4.111398610921967e-05,
      "loss": 4.561,
      "step": 7025
    },
    {
      "epoch": 0.1889023243315867,
      "grad_norm": 2.462205171585083,
      "learning_rate": 4.11071769031731e-05,
      "loss": 4.4063,
      "step": 7030
    },
    {
      "epoch": 0.18903667875856509,
      "grad_norm": 2.276282548904419,
      "learning_rate": 4.110036769712652e-05,
      "loss": 4.5442,
      "step": 7035
    },
    {
      "epoch": 0.18917103318554346,
      "grad_norm": 2.3674912452697754,
      "learning_rate": 4.109355849107994e-05,
      "loss": 4.5329,
      "step": 7040
    },
    {
      "epoch": 0.18930538761252183,
      "grad_norm": 2.249842643737793,
      "learning_rate": 4.1086749285033365e-05,
      "loss": 4.5903,
      "step": 7045
    },
    {
      "epoch": 0.1894397420395002,
      "grad_norm": 2.415100574493408,
      "learning_rate": 4.107994007898679e-05,
      "loss": 4.5262,
      "step": 7050
    },
    {
      "epoch": 0.18957409646647858,
      "grad_norm": 2.2215616703033447,
      "learning_rate": 4.1073130872940216e-05,
      "loss": 4.5712,
      "step": 7055
    },
    {
      "epoch": 0.18970845089345695,
      "grad_norm": 2.207418918609619,
      "learning_rate": 4.1066321666893645e-05,
      "loss": 4.7247,
      "step": 7060
    },
    {
      "epoch": 0.18984280532043532,
      "grad_norm": 2.1361162662506104,
      "learning_rate": 4.105951246084707e-05,
      "loss": 4.6041,
      "step": 7065
    },
    {
      "epoch": 0.18997715974741367,
      "grad_norm": 2.2110280990600586,
      "learning_rate": 4.105270325480049e-05,
      "loss": 4.4982,
      "step": 7070
    },
    {
      "epoch": 0.19011151417439204,
      "grad_norm": 2.2744858264923096,
      "learning_rate": 4.104589404875392e-05,
      "loss": 4.5524,
      "step": 7075
    },
    {
      "epoch": 0.1902458686013704,
      "grad_norm": 2.5295419692993164,
      "learning_rate": 4.103908484270734e-05,
      "loss": 4.4965,
      "step": 7080
    },
    {
      "epoch": 0.19038022302834878,
      "grad_norm": 2.436302423477173,
      "learning_rate": 4.103227563666076e-05,
      "loss": 4.5,
      "step": 7085
    },
    {
      "epoch": 0.19051457745532716,
      "grad_norm": 2.296764373779297,
      "learning_rate": 4.102546643061419e-05,
      "loss": 4.5614,
      "step": 7090
    },
    {
      "epoch": 0.19064893188230553,
      "grad_norm": 2.4771785736083984,
      "learning_rate": 4.101865722456762e-05,
      "loss": 4.5643,
      "step": 7095
    },
    {
      "epoch": 0.1907832863092839,
      "grad_norm": 2.495366334915161,
      "learning_rate": 4.1011848018521043e-05,
      "loss": 4.4974,
      "step": 7100
    },
    {
      "epoch": 0.19091764073626227,
      "grad_norm": 2.2070071697235107,
      "learning_rate": 4.1005038812474466e-05,
      "loss": 4.6395,
      "step": 7105
    },
    {
      "epoch": 0.19105199516324062,
      "grad_norm": 2.448603630065918,
      "learning_rate": 4.099822960642789e-05,
      "loss": 4.5929,
      "step": 7110
    },
    {
      "epoch": 0.191186349590219,
      "grad_norm": 2.2084875106811523,
      "learning_rate": 4.099142040038132e-05,
      "loss": 4.6274,
      "step": 7115
    },
    {
      "epoch": 0.19132070401719736,
      "grad_norm": 2.280116319656372,
      "learning_rate": 4.0984611194334746e-05,
      "loss": 4.5419,
      "step": 7120
    },
    {
      "epoch": 0.19145505844417574,
      "grad_norm": 2.4503848552703857,
      "learning_rate": 4.097780198828817e-05,
      "loss": 4.6398,
      "step": 7125
    },
    {
      "epoch": 0.1915894128711541,
      "grad_norm": 2.2648165225982666,
      "learning_rate": 4.097099278224159e-05,
      "loss": 4.5188,
      "step": 7130
    },
    {
      "epoch": 0.19172376729813248,
      "grad_norm": 2.187908411026001,
      "learning_rate": 4.096418357619502e-05,
      "loss": 4.3556,
      "step": 7135
    },
    {
      "epoch": 0.19185812172511085,
      "grad_norm": 2.270920753479004,
      "learning_rate": 4.095737437014844e-05,
      "loss": 4.5919,
      "step": 7140
    },
    {
      "epoch": 0.1919924761520892,
      "grad_norm": 2.424598217010498,
      "learning_rate": 4.0950565164101864e-05,
      "loss": 4.5214,
      "step": 7145
    },
    {
      "epoch": 0.19212683057906757,
      "grad_norm": 2.10286021232605,
      "learning_rate": 4.094375595805529e-05,
      "loss": 4.5211,
      "step": 7150
    },
    {
      "epoch": 0.19226118500604594,
      "grad_norm": 2.268779754638672,
      "learning_rate": 4.093694675200872e-05,
      "loss": 4.5586,
      "step": 7155
    },
    {
      "epoch": 0.19239553943302432,
      "grad_norm": 2.3719546794891357,
      "learning_rate": 4.0930137545962145e-05,
      "loss": 4.5612,
      "step": 7160
    },
    {
      "epoch": 0.1925298938600027,
      "grad_norm": 2.2428412437438965,
      "learning_rate": 4.092332833991557e-05,
      "loss": 4.5811,
      "step": 7165
    },
    {
      "epoch": 0.19266424828698106,
      "grad_norm": 2.2108829021453857,
      "learning_rate": 4.091651913386899e-05,
      "loss": 4.4952,
      "step": 7170
    },
    {
      "epoch": 0.19279860271395943,
      "grad_norm": 2.4711620807647705,
      "learning_rate": 4.090970992782242e-05,
      "loss": 4.4856,
      "step": 7175
    },
    {
      "epoch": 0.1929329571409378,
      "grad_norm": 2.248721122741699,
      "learning_rate": 4.090290072177585e-05,
      "loss": 4.4079,
      "step": 7180
    },
    {
      "epoch": 0.19306731156791615,
      "grad_norm": 2.2443575859069824,
      "learning_rate": 4.089609151572927e-05,
      "loss": 4.5448,
      "step": 7185
    },
    {
      "epoch": 0.19320166599489452,
      "grad_norm": 2.6840131282806396,
      "learning_rate": 4.088928230968269e-05,
      "loss": 4.5768,
      "step": 7190
    },
    {
      "epoch": 0.1933360204218729,
      "grad_norm": 2.3117644786834717,
      "learning_rate": 4.0882473103636114e-05,
      "loss": 4.5582,
      "step": 7195
    },
    {
      "epoch": 0.19347037484885127,
      "grad_norm": 2.3625028133392334,
      "learning_rate": 4.087566389758954e-05,
      "loss": 4.5589,
      "step": 7200
    },
    {
      "epoch": 0.19360472927582964,
      "grad_norm": 2.1942670345306396,
      "learning_rate": 4.0868854691542965e-05,
      "loss": 4.5218,
      "step": 7205
    },
    {
      "epoch": 0.193739083702808,
      "grad_norm": 2.1714208126068115,
      "learning_rate": 4.0862045485496394e-05,
      "loss": 4.5744,
      "step": 7210
    },
    {
      "epoch": 0.19387343812978639,
      "grad_norm": 2.3121001720428467,
      "learning_rate": 4.0855236279449817e-05,
      "loss": 4.4298,
      "step": 7215
    },
    {
      "epoch": 0.19400779255676476,
      "grad_norm": 2.083580493927002,
      "learning_rate": 4.0848427073403246e-05,
      "loss": 4.5238,
      "step": 7220
    },
    {
      "epoch": 0.1941421469837431,
      "grad_norm": 2.2846081256866455,
      "learning_rate": 4.084161786735667e-05,
      "loss": 4.6132,
      "step": 7225
    },
    {
      "epoch": 0.19427650141072147,
      "grad_norm": 2.436431884765625,
      "learning_rate": 4.083480866131009e-05,
      "loss": 4.6542,
      "step": 7230
    },
    {
      "epoch": 0.19441085583769985,
      "grad_norm": 2.298214912414551,
      "learning_rate": 4.082799945526352e-05,
      "loss": 4.4369,
      "step": 7235
    },
    {
      "epoch": 0.19454521026467822,
      "grad_norm": 2.5868639945983887,
      "learning_rate": 4.082119024921695e-05,
      "loss": 4.5831,
      "step": 7240
    },
    {
      "epoch": 0.1946795646916566,
      "grad_norm": 2.4647202491760254,
      "learning_rate": 4.081438104317037e-05,
      "loss": 4.6139,
      "step": 7245
    },
    {
      "epoch": 0.19481391911863497,
      "grad_norm": 2.2689716815948486,
      "learning_rate": 4.080757183712379e-05,
      "loss": 4.4812,
      "step": 7250
    },
    {
      "epoch": 0.19494827354561334,
      "grad_norm": 2.316291570663452,
      "learning_rate": 4.0800762631077215e-05,
      "loss": 4.5507,
      "step": 7255
    },
    {
      "epoch": 0.1950826279725917,
      "grad_norm": 2.335848569869995,
      "learning_rate": 4.0793953425030644e-05,
      "loss": 4.502,
      "step": 7260
    },
    {
      "epoch": 0.19521698239957005,
      "grad_norm": 2.3392393589019775,
      "learning_rate": 4.0787144218984066e-05,
      "loss": 4.524,
      "step": 7265
    },
    {
      "epoch": 0.19535133682654843,
      "grad_norm": 2.3226287364959717,
      "learning_rate": 4.0780335012937495e-05,
      "loss": 4.6766,
      "step": 7270
    },
    {
      "epoch": 0.1954856912535268,
      "grad_norm": 2.3067502975463867,
      "learning_rate": 4.077352580689092e-05,
      "loss": 4.5135,
      "step": 7275
    },
    {
      "epoch": 0.19562004568050517,
      "grad_norm": 2.4793460369110107,
      "learning_rate": 4.076671660084435e-05,
      "loss": 4.5692,
      "step": 7280
    },
    {
      "epoch": 0.19575440010748354,
      "grad_norm": 2.058846950531006,
      "learning_rate": 4.075990739479777e-05,
      "loss": 4.503,
      "step": 7285
    },
    {
      "epoch": 0.19588875453446192,
      "grad_norm": 2.3449316024780273,
      "learning_rate": 4.075309818875119e-05,
      "loss": 4.6599,
      "step": 7290
    },
    {
      "epoch": 0.1960231089614403,
      "grad_norm": 2.3653762340545654,
      "learning_rate": 4.0746288982704614e-05,
      "loss": 4.4842,
      "step": 7295
    },
    {
      "epoch": 0.19615746338841863,
      "grad_norm": 2.206763744354248,
      "learning_rate": 4.073947977665805e-05,
      "loss": 4.5657,
      "step": 7300
    },
    {
      "epoch": 0.196291817815397,
      "grad_norm": 2.232729196548462,
      "learning_rate": 4.073267057061147e-05,
      "loss": 4.4914,
      "step": 7305
    },
    {
      "epoch": 0.19642617224237538,
      "grad_norm": 2.45845103263855,
      "learning_rate": 4.0725861364564894e-05,
      "loss": 4.4808,
      "step": 7310
    },
    {
      "epoch": 0.19656052666935375,
      "grad_norm": 2.508060932159424,
      "learning_rate": 4.0719052158518316e-05,
      "loss": 4.5533,
      "step": 7315
    },
    {
      "epoch": 0.19669488109633212,
      "grad_norm": 2.391829013824463,
      "learning_rate": 4.0712242952471745e-05,
      "loss": 4.4748,
      "step": 7320
    },
    {
      "epoch": 0.1968292355233105,
      "grad_norm": 2.2996342182159424,
      "learning_rate": 4.070543374642517e-05,
      "loss": 4.5077,
      "step": 7325
    },
    {
      "epoch": 0.19696358995028887,
      "grad_norm": 2.506173610687256,
      "learning_rate": 4.0698624540378597e-05,
      "loss": 4.4025,
      "step": 7330
    },
    {
      "epoch": 0.19709794437726724,
      "grad_norm": 2.3830065727233887,
      "learning_rate": 4.069181533433202e-05,
      "loss": 4.5319,
      "step": 7335
    },
    {
      "epoch": 0.1972322988042456,
      "grad_norm": 2.4049127101898193,
      "learning_rate": 4.068500612828544e-05,
      "loss": 4.5387,
      "step": 7340
    },
    {
      "epoch": 0.19736665323122396,
      "grad_norm": 2.340664863586426,
      "learning_rate": 4.067819692223887e-05,
      "loss": 4.6736,
      "step": 7345
    },
    {
      "epoch": 0.19750100765820233,
      "grad_norm": 2.5061004161834717,
      "learning_rate": 4.067138771619229e-05,
      "loss": 4.6041,
      "step": 7350
    },
    {
      "epoch": 0.1976353620851807,
      "grad_norm": 2.234525680541992,
      "learning_rate": 4.0664578510145715e-05,
      "loss": 4.5771,
      "step": 7355
    },
    {
      "epoch": 0.19776971651215908,
      "grad_norm": 2.337937116622925,
      "learning_rate": 4.0657769304099144e-05,
      "loss": 4.52,
      "step": 7360
    },
    {
      "epoch": 0.19790407093913745,
      "grad_norm": 2.4057869911193848,
      "learning_rate": 4.065096009805257e-05,
      "loss": 4.5733,
      "step": 7365
    },
    {
      "epoch": 0.19803842536611582,
      "grad_norm": 2.2221574783325195,
      "learning_rate": 4.0644150892005995e-05,
      "loss": 4.5917,
      "step": 7370
    },
    {
      "epoch": 0.1981727797930942,
      "grad_norm": 2.2738208770751953,
      "learning_rate": 4.063734168595942e-05,
      "loss": 4.5989,
      "step": 7375
    },
    {
      "epoch": 0.19830713422007254,
      "grad_norm": 2.4379289150238037,
      "learning_rate": 4.063053247991284e-05,
      "loss": 4.4874,
      "step": 7380
    },
    {
      "epoch": 0.1984414886470509,
      "grad_norm": 2.1695709228515625,
      "learning_rate": 4.062372327386627e-05,
      "loss": 4.5976,
      "step": 7385
    },
    {
      "epoch": 0.19857584307402928,
      "grad_norm": 2.3560855388641357,
      "learning_rate": 4.06169140678197e-05,
      "loss": 4.5486,
      "step": 7390
    },
    {
      "epoch": 0.19871019750100766,
      "grad_norm": 2.196437358856201,
      "learning_rate": 4.061010486177312e-05,
      "loss": 4.5423,
      "step": 7395
    },
    {
      "epoch": 0.19884455192798603,
      "grad_norm": 2.367628812789917,
      "learning_rate": 4.060329565572654e-05,
      "loss": 4.5537,
      "step": 7400
    },
    {
      "epoch": 0.1989789063549644,
      "grad_norm": 2.389923572540283,
      "learning_rate": 4.059648644967997e-05,
      "loss": 4.5587,
      "step": 7405
    },
    {
      "epoch": 0.19911326078194277,
      "grad_norm": 2.2588648796081543,
      "learning_rate": 4.0589677243633394e-05,
      "loss": 4.5655,
      "step": 7410
    },
    {
      "epoch": 0.19924761520892115,
      "grad_norm": 2.048673629760742,
      "learning_rate": 4.0582868037586816e-05,
      "loss": 4.4984,
      "step": 7415
    },
    {
      "epoch": 0.1993819696358995,
      "grad_norm": 2.3108294010162354,
      "learning_rate": 4.0576058831540245e-05,
      "loss": 4.5398,
      "step": 7420
    },
    {
      "epoch": 0.19951632406287786,
      "grad_norm": 2.3519372940063477,
      "learning_rate": 4.0569249625493674e-05,
      "loss": 4.5037,
      "step": 7425
    },
    {
      "epoch": 0.19965067848985624,
      "grad_norm": 2.3422770500183105,
      "learning_rate": 4.0562440419447096e-05,
      "loss": 4.5414,
      "step": 7430
    },
    {
      "epoch": 0.1997850329168346,
      "grad_norm": 3.085261344909668,
      "learning_rate": 4.055563121340052e-05,
      "loss": 4.5873,
      "step": 7435
    },
    {
      "epoch": 0.19991938734381298,
      "grad_norm": 2.5949223041534424,
      "learning_rate": 4.054882200735394e-05,
      "loss": 4.6393,
      "step": 7440
    },
    {
      "epoch": 0.20005374177079135,
      "grad_norm": 2.243067502975464,
      "learning_rate": 4.054201280130737e-05,
      "loss": 4.5188,
      "step": 7445
    },
    {
      "epoch": 0.20018809619776973,
      "grad_norm": 2.3928871154785156,
      "learning_rate": 4.05352035952608e-05,
      "loss": 4.4724,
      "step": 7450
    },
    {
      "epoch": 0.20032245062474807,
      "grad_norm": 2.176637887954712,
      "learning_rate": 4.052839438921422e-05,
      "loss": 4.504,
      "step": 7455
    },
    {
      "epoch": 0.20045680505172644,
      "grad_norm": 2.3238308429718018,
      "learning_rate": 4.052158518316764e-05,
      "loss": 4.6474,
      "step": 7460
    },
    {
      "epoch": 0.20059115947870482,
      "grad_norm": 2.492518186569214,
      "learning_rate": 4.051477597712107e-05,
      "loss": 4.6531,
      "step": 7465
    },
    {
      "epoch": 0.2007255139056832,
      "grad_norm": 2.3349030017852783,
      "learning_rate": 4.0507966771074495e-05,
      "loss": 4.5371,
      "step": 7470
    },
    {
      "epoch": 0.20085986833266156,
      "grad_norm": 2.2879385948181152,
      "learning_rate": 4.050115756502792e-05,
      "loss": 4.566,
      "step": 7475
    },
    {
      "epoch": 0.20099422275963993,
      "grad_norm": 2.3234596252441406,
      "learning_rate": 4.0494348358981346e-05,
      "loss": 4.4785,
      "step": 7480
    },
    {
      "epoch": 0.2011285771866183,
      "grad_norm": 2.399777412414551,
      "learning_rate": 4.0487539152934775e-05,
      "loss": 4.6115,
      "step": 7485
    },
    {
      "epoch": 0.20126293161359668,
      "grad_norm": 2.2911088466644287,
      "learning_rate": 4.04807299468882e-05,
      "loss": 4.5406,
      "step": 7490
    },
    {
      "epoch": 0.20139728604057502,
      "grad_norm": 2.217829704284668,
      "learning_rate": 4.047392074084162e-05,
      "loss": 4.6243,
      "step": 7495
    },
    {
      "epoch": 0.2015316404675534,
      "grad_norm": 2.7082183361053467,
      "learning_rate": 4.046711153479504e-05,
      "loss": 4.5976,
      "step": 7500
    },
    {
      "epoch": 0.20166599489453177,
      "grad_norm": 2.781736373901367,
      "learning_rate": 4.0460302328748464e-05,
      "loss": 4.4752,
      "step": 7505
    },
    {
      "epoch": 0.20180034932151014,
      "grad_norm": 2.045954704284668,
      "learning_rate": 4.04534931227019e-05,
      "loss": 4.5119,
      "step": 7510
    },
    {
      "epoch": 0.20193470374848851,
      "grad_norm": 2.3848822116851807,
      "learning_rate": 4.044668391665532e-05,
      "loss": 4.5257,
      "step": 7515
    },
    {
      "epoch": 0.2020690581754669,
      "grad_norm": 2.1805412769317627,
      "learning_rate": 4.0439874710608744e-05,
      "loss": 4.5548,
      "step": 7520
    },
    {
      "epoch": 0.20220341260244526,
      "grad_norm": 2.234635829925537,
      "learning_rate": 4.043306550456217e-05,
      "loss": 4.6258,
      "step": 7525
    },
    {
      "epoch": 0.20233776702942363,
      "grad_norm": 2.2407567501068115,
      "learning_rate": 4.0426256298515596e-05,
      "loss": 4.4839,
      "step": 7530
    },
    {
      "epoch": 0.20247212145640198,
      "grad_norm": 2.2742607593536377,
      "learning_rate": 4.041944709246902e-05,
      "loss": 4.534,
      "step": 7535
    },
    {
      "epoch": 0.20260647588338035,
      "grad_norm": 2.6407322883605957,
      "learning_rate": 4.041263788642245e-05,
      "loss": 4.4093,
      "step": 7540
    },
    {
      "epoch": 0.20274083031035872,
      "grad_norm": 2.2201759815216064,
      "learning_rate": 4.040582868037587e-05,
      "loss": 4.3755,
      "step": 7545
    },
    {
      "epoch": 0.2028751847373371,
      "grad_norm": 2.5236337184906006,
      "learning_rate": 4.03990194743293e-05,
      "loss": 4.5846,
      "step": 7550
    },
    {
      "epoch": 0.20300953916431547,
      "grad_norm": 2.612666368484497,
      "learning_rate": 4.039221026828272e-05,
      "loss": 4.4706,
      "step": 7555
    },
    {
      "epoch": 0.20314389359129384,
      "grad_norm": 2.3540444374084473,
      "learning_rate": 4.038540106223614e-05,
      "loss": 4.462,
      "step": 7560
    },
    {
      "epoch": 0.2032782480182722,
      "grad_norm": 2.1694231033325195,
      "learning_rate": 4.0378591856189565e-05,
      "loss": 4.5415,
      "step": 7565
    },
    {
      "epoch": 0.20341260244525058,
      "grad_norm": 2.3195013999938965,
      "learning_rate": 4.0371782650142994e-05,
      "loss": 4.6151,
      "step": 7570
    },
    {
      "epoch": 0.20354695687222893,
      "grad_norm": 2.4916484355926514,
      "learning_rate": 4.036497344409642e-05,
      "loss": 4.4776,
      "step": 7575
    },
    {
      "epoch": 0.2036813112992073,
      "grad_norm": 2.20165753364563,
      "learning_rate": 4.0358164238049846e-05,
      "loss": 4.6301,
      "step": 7580
    },
    {
      "epoch": 0.20381566572618567,
      "grad_norm": 2.3772010803222656,
      "learning_rate": 4.035135503200327e-05,
      "loss": 4.4669,
      "step": 7585
    },
    {
      "epoch": 0.20395002015316405,
      "grad_norm": 2.5744669437408447,
      "learning_rate": 4.03445458259567e-05,
      "loss": 4.4857,
      "step": 7590
    },
    {
      "epoch": 0.20408437458014242,
      "grad_norm": 2.0861997604370117,
      "learning_rate": 4.033773661991012e-05,
      "loss": 4.5762,
      "step": 7595
    },
    {
      "epoch": 0.2042187290071208,
      "grad_norm": 2.240452527999878,
      "learning_rate": 4.033092741386355e-05,
      "loss": 4.4403,
      "step": 7600
    },
    {
      "epoch": 0.20435308343409916,
      "grad_norm": 2.383476495742798,
      "learning_rate": 4.032411820781697e-05,
      "loss": 4.483,
      "step": 7605
    },
    {
      "epoch": 0.2044874378610775,
      "grad_norm": 2.3453216552734375,
      "learning_rate": 4.03173090017704e-05,
      "loss": 4.5329,
      "step": 7610
    },
    {
      "epoch": 0.20462179228805588,
      "grad_norm": 2.2961390018463135,
      "learning_rate": 4.031049979572382e-05,
      "loss": 4.4499,
      "step": 7615
    },
    {
      "epoch": 0.20475614671503425,
      "grad_norm": 2.4231982231140137,
      "learning_rate": 4.0303690589677244e-05,
      "loss": 4.4969,
      "step": 7620
    },
    {
      "epoch": 0.20489050114201263,
      "grad_norm": 2.271404266357422,
      "learning_rate": 4.0296881383630666e-05,
      "loss": 4.5771,
      "step": 7625
    },
    {
      "epoch": 0.205024855568991,
      "grad_norm": 2.2238411903381348,
      "learning_rate": 4.0290072177584095e-05,
      "loss": 4.6415,
      "step": 7630
    },
    {
      "epoch": 0.20515920999596937,
      "grad_norm": 2.163578987121582,
      "learning_rate": 4.0283262971537524e-05,
      "loss": 4.6844,
      "step": 7635
    },
    {
      "epoch": 0.20529356442294774,
      "grad_norm": 2.4209017753601074,
      "learning_rate": 4.027645376549095e-05,
      "loss": 4.5357,
      "step": 7640
    },
    {
      "epoch": 0.20542791884992612,
      "grad_norm": 2.2471983432769775,
      "learning_rate": 4.026964455944437e-05,
      "loss": 4.664,
      "step": 7645
    },
    {
      "epoch": 0.20556227327690446,
      "grad_norm": 2.440857410430908,
      "learning_rate": 4.026283535339779e-05,
      "loss": 4.4413,
      "step": 7650
    },
    {
      "epoch": 0.20569662770388283,
      "grad_norm": 2.307403564453125,
      "learning_rate": 4.025602614735122e-05,
      "loss": 4.5154,
      "step": 7655
    },
    {
      "epoch": 0.2058309821308612,
      "grad_norm": 2.431759834289551,
      "learning_rate": 4.024921694130465e-05,
      "loss": 4.4362,
      "step": 7660
    },
    {
      "epoch": 0.20596533655783958,
      "grad_norm": 2.3965609073638916,
      "learning_rate": 4.024240773525807e-05,
      "loss": 4.4735,
      "step": 7665
    },
    {
      "epoch": 0.20609969098481795,
      "grad_norm": 2.2584922313690186,
      "learning_rate": 4.0235598529211494e-05,
      "loss": 4.5538,
      "step": 7670
    },
    {
      "epoch": 0.20623404541179632,
      "grad_norm": 2.4941208362579346,
      "learning_rate": 4.022878932316492e-05,
      "loss": 4.5343,
      "step": 7675
    },
    {
      "epoch": 0.2063683998387747,
      "grad_norm": 2.255657196044922,
      "learning_rate": 4.0221980117118345e-05,
      "loss": 4.431,
      "step": 7680
    },
    {
      "epoch": 0.20650275426575307,
      "grad_norm": 2.3047399520874023,
      "learning_rate": 4.021517091107177e-05,
      "loss": 4.6374,
      "step": 7685
    },
    {
      "epoch": 0.2066371086927314,
      "grad_norm": 2.51051664352417,
      "learning_rate": 4.0208361705025196e-05,
      "loss": 4.6606,
      "step": 7690
    },
    {
      "epoch": 0.2067714631197098,
      "grad_norm": 2.3574345111846924,
      "learning_rate": 4.0201552498978625e-05,
      "loss": 4.4689,
      "step": 7695
    },
    {
      "epoch": 0.20690581754668816,
      "grad_norm": 2.5600733757019043,
      "learning_rate": 4.019474329293205e-05,
      "loss": 4.3813,
      "step": 7700
    },
    {
      "epoch": 0.20704017197366653,
      "grad_norm": 2.2496025562286377,
      "learning_rate": 4.018793408688547e-05,
      "loss": 4.514,
      "step": 7705
    },
    {
      "epoch": 0.2071745264006449,
      "grad_norm": 2.3794093132019043,
      "learning_rate": 4.018112488083889e-05,
      "loss": 4.5071,
      "step": 7710
    },
    {
      "epoch": 0.20730888082762328,
      "grad_norm": 2.2429542541503906,
      "learning_rate": 4.017431567479232e-05,
      "loss": 4.5975,
      "step": 7715
    },
    {
      "epoch": 0.20744323525460165,
      "grad_norm": 2.5032238960266113,
      "learning_rate": 4.0167506468745744e-05,
      "loss": 4.6123,
      "step": 7720
    },
    {
      "epoch": 0.20757758968158002,
      "grad_norm": 2.3977210521698,
      "learning_rate": 4.016069726269917e-05,
      "loss": 4.4546,
      "step": 7725
    },
    {
      "epoch": 0.20771194410855837,
      "grad_norm": 2.219390869140625,
      "learning_rate": 4.0153888056652595e-05,
      "loss": 4.4911,
      "step": 7730
    },
    {
      "epoch": 0.20784629853553674,
      "grad_norm": 2.246828317642212,
      "learning_rate": 4.0147078850606024e-05,
      "loss": 4.5641,
      "step": 7735
    },
    {
      "epoch": 0.2079806529625151,
      "grad_norm": 2.7244341373443604,
      "learning_rate": 4.0140269644559446e-05,
      "loss": 4.6314,
      "step": 7740
    },
    {
      "epoch": 0.20811500738949348,
      "grad_norm": 2.2800581455230713,
      "learning_rate": 4.013346043851287e-05,
      "loss": 4.6105,
      "step": 7745
    },
    {
      "epoch": 0.20824936181647186,
      "grad_norm": 2.5239195823669434,
      "learning_rate": 4.01266512324663e-05,
      "loss": 4.4694,
      "step": 7750
    },
    {
      "epoch": 0.20838371624345023,
      "grad_norm": 2.491713523864746,
      "learning_rate": 4.0119842026419727e-05,
      "loss": 4.6261,
      "step": 7755
    },
    {
      "epoch": 0.2085180706704286,
      "grad_norm": 2.5509541034698486,
      "learning_rate": 4.011303282037315e-05,
      "loss": 4.6265,
      "step": 7760
    },
    {
      "epoch": 0.20865242509740695,
      "grad_norm": 2.385878324508667,
      "learning_rate": 4.010622361432657e-05,
      "loss": 4.5261,
      "step": 7765
    },
    {
      "epoch": 0.20878677952438532,
      "grad_norm": 2.5552380084991455,
      "learning_rate": 4.0099414408279993e-05,
      "loss": 4.6553,
      "step": 7770
    },
    {
      "epoch": 0.2089211339513637,
      "grad_norm": 2.1308703422546387,
      "learning_rate": 4.009260520223342e-05,
      "loss": 4.4779,
      "step": 7775
    },
    {
      "epoch": 0.20905548837834206,
      "grad_norm": 2.249119997024536,
      "learning_rate": 4.0085795996186845e-05,
      "loss": 4.5108,
      "step": 7780
    },
    {
      "epoch": 0.20918984280532044,
      "grad_norm": 2.231544017791748,
      "learning_rate": 4.0078986790140274e-05,
      "loss": 4.4367,
      "step": 7785
    },
    {
      "epoch": 0.2093241972322988,
      "grad_norm": 2.6178367137908936,
      "learning_rate": 4.0072177584093696e-05,
      "loss": 4.6185,
      "step": 7790
    },
    {
      "epoch": 0.20945855165927718,
      "grad_norm": 2.042090892791748,
      "learning_rate": 4.006536837804712e-05,
      "loss": 4.6021,
      "step": 7795
    },
    {
      "epoch": 0.20959290608625555,
      "grad_norm": 2.411900520324707,
      "learning_rate": 4.005855917200055e-05,
      "loss": 4.3821,
      "step": 7800
    },
    {
      "epoch": 0.2097272605132339,
      "grad_norm": 2.383136749267578,
      "learning_rate": 4.005174996595397e-05,
      "loss": 4.6023,
      "step": 7805
    },
    {
      "epoch": 0.20986161494021227,
      "grad_norm": 2.4700050354003906,
      "learning_rate": 4.00449407599074e-05,
      "loss": 4.5928,
      "step": 7810
    },
    {
      "epoch": 0.20999596936719064,
      "grad_norm": 2.20003342628479,
      "learning_rate": 4.003813155386082e-05,
      "loss": 4.5535,
      "step": 7815
    },
    {
      "epoch": 0.21013032379416902,
      "grad_norm": 2.229978322982788,
      "learning_rate": 4.003132234781425e-05,
      "loss": 4.6606,
      "step": 7820
    },
    {
      "epoch": 0.2102646782211474,
      "grad_norm": 2.4298512935638428,
      "learning_rate": 4.002451314176767e-05,
      "loss": 4.4771,
      "step": 7825
    },
    {
      "epoch": 0.21039903264812576,
      "grad_norm": 2.2240586280822754,
      "learning_rate": 4.0017703935721095e-05,
      "loss": 4.636,
      "step": 7830
    },
    {
      "epoch": 0.21053338707510413,
      "grad_norm": 2.243889808654785,
      "learning_rate": 4.001089472967452e-05,
      "loss": 4.4971,
      "step": 7835
    },
    {
      "epoch": 0.2106677415020825,
      "grad_norm": 2.212472915649414,
      "learning_rate": 4.0004085523627946e-05,
      "loss": 4.5702,
      "step": 7840
    },
    {
      "epoch": 0.21080209592906085,
      "grad_norm": 2.100231170654297,
      "learning_rate": 3.9997276317581375e-05,
      "loss": 4.4953,
      "step": 7845
    },
    {
      "epoch": 0.21093645035603922,
      "grad_norm": 2.4913434982299805,
      "learning_rate": 3.99904671115348e-05,
      "loss": 4.6602,
      "step": 7850
    },
    {
      "epoch": 0.2110708047830176,
      "grad_norm": 2.481914520263672,
      "learning_rate": 3.998365790548822e-05,
      "loss": 4.5769,
      "step": 7855
    },
    {
      "epoch": 0.21120515920999597,
      "grad_norm": 2.200019359588623,
      "learning_rate": 3.997684869944165e-05,
      "loss": 4.5763,
      "step": 7860
    },
    {
      "epoch": 0.21133951363697434,
      "grad_norm": 2.381456136703491,
      "learning_rate": 3.997003949339507e-05,
      "loss": 4.5627,
      "step": 7865
    },
    {
      "epoch": 0.2114738680639527,
      "grad_norm": 2.30440616607666,
      "learning_rate": 3.99632302873485e-05,
      "loss": 4.4959,
      "step": 7870
    },
    {
      "epoch": 0.21160822249093109,
      "grad_norm": 2.580388307571411,
      "learning_rate": 3.995642108130192e-05,
      "loss": 4.5989,
      "step": 7875
    },
    {
      "epoch": 0.21174257691790946,
      "grad_norm": 2.316093683242798,
      "learning_rate": 3.994961187525535e-05,
      "loss": 4.5402,
      "step": 7880
    },
    {
      "epoch": 0.2118769313448878,
      "grad_norm": 2.3373405933380127,
      "learning_rate": 3.994280266920877e-05,
      "loss": 4.4452,
      "step": 7885
    },
    {
      "epoch": 0.21201128577186618,
      "grad_norm": 2.2177529335021973,
      "learning_rate": 3.9935993463162196e-05,
      "loss": 4.6338,
      "step": 7890
    },
    {
      "epoch": 0.21214564019884455,
      "grad_norm": 2.2226641178131104,
      "learning_rate": 3.992918425711562e-05,
      "loss": 4.3612,
      "step": 7895
    },
    {
      "epoch": 0.21227999462582292,
      "grad_norm": 2.4098312854766846,
      "learning_rate": 3.992237505106905e-05,
      "loss": 4.5884,
      "step": 7900
    },
    {
      "epoch": 0.2124143490528013,
      "grad_norm": 2.2080001831054688,
      "learning_rate": 3.9915565845022476e-05,
      "loss": 4.465,
      "step": 7905
    },
    {
      "epoch": 0.21254870347977967,
      "grad_norm": 2.432154893875122,
      "learning_rate": 3.99087566389759e-05,
      "loss": 4.5543,
      "step": 7910
    },
    {
      "epoch": 0.21268305790675804,
      "grad_norm": 2.4245781898498535,
      "learning_rate": 3.990194743292932e-05,
      "loss": 4.5096,
      "step": 7915
    },
    {
      "epoch": 0.21281741233373638,
      "grad_norm": 2.126307249069214,
      "learning_rate": 3.989513822688275e-05,
      "loss": 4.5626,
      "step": 7920
    },
    {
      "epoch": 0.21295176676071476,
      "grad_norm": 2.131974697113037,
      "learning_rate": 3.988832902083617e-05,
      "loss": 4.4761,
      "step": 7925
    },
    {
      "epoch": 0.21308612118769313,
      "grad_norm": 2.3548102378845215,
      "learning_rate": 3.9881519814789594e-05,
      "loss": 4.4925,
      "step": 7930
    },
    {
      "epoch": 0.2132204756146715,
      "grad_norm": 2.6243896484375,
      "learning_rate": 3.987471060874302e-05,
      "loss": 4.5733,
      "step": 7935
    },
    {
      "epoch": 0.21335483004164987,
      "grad_norm": 2.3429207801818848,
      "learning_rate": 3.9867901402696445e-05,
      "loss": 4.5669,
      "step": 7940
    },
    {
      "epoch": 0.21348918446862825,
      "grad_norm": 2.5600531101226807,
      "learning_rate": 3.9861092196649874e-05,
      "loss": 4.5028,
      "step": 7945
    },
    {
      "epoch": 0.21362353889560662,
      "grad_norm": 2.3005564212799072,
      "learning_rate": 3.98542829906033e-05,
      "loss": 4.7203,
      "step": 7950
    },
    {
      "epoch": 0.213757893322585,
      "grad_norm": 2.3356740474700928,
      "learning_rate": 3.984747378455672e-05,
      "loss": 4.433,
      "step": 7955
    },
    {
      "epoch": 0.21389224774956334,
      "grad_norm": 2.6492056846618652,
      "learning_rate": 3.984066457851015e-05,
      "loss": 4.4739,
      "step": 7960
    },
    {
      "epoch": 0.2140266021765417,
      "grad_norm": 2.2943952083587646,
      "learning_rate": 3.983385537246358e-05,
      "loss": 4.5779,
      "step": 7965
    },
    {
      "epoch": 0.21416095660352008,
      "grad_norm": 2.26895809173584,
      "learning_rate": 3.9827046166417e-05,
      "loss": 4.6247,
      "step": 7970
    },
    {
      "epoch": 0.21429531103049845,
      "grad_norm": 2.3796446323394775,
      "learning_rate": 3.982023696037042e-05,
      "loss": 4.4667,
      "step": 7975
    },
    {
      "epoch": 0.21442966545747683,
      "grad_norm": 2.086341381072998,
      "learning_rate": 3.9813427754323844e-05,
      "loss": 4.5953,
      "step": 7980
    },
    {
      "epoch": 0.2145640198844552,
      "grad_norm": 2.2315192222595215,
      "learning_rate": 3.980661854827727e-05,
      "loss": 4.411,
      "step": 7985
    },
    {
      "epoch": 0.21469837431143357,
      "grad_norm": 2.3698182106018066,
      "learning_rate": 3.9799809342230695e-05,
      "loss": 4.476,
      "step": 7990
    },
    {
      "epoch": 0.21483272873841194,
      "grad_norm": 2.4338014125823975,
      "learning_rate": 3.9793000136184124e-05,
      "loss": 4.573,
      "step": 7995
    },
    {
      "epoch": 0.2149670831653903,
      "grad_norm": 2.3947324752807617,
      "learning_rate": 3.9786190930137547e-05,
      "loss": 4.5989,
      "step": 8000
    },
    {
      "epoch": 0.21510143759236866,
      "grad_norm": 2.3885648250579834,
      "learning_rate": 3.9779381724090976e-05,
      "loss": 4.623,
      "step": 8005
    },
    {
      "epoch": 0.21523579201934703,
      "grad_norm": 2.234739303588867,
      "learning_rate": 3.97725725180444e-05,
      "loss": 4.6016,
      "step": 8010
    },
    {
      "epoch": 0.2153701464463254,
      "grad_norm": 2.397773265838623,
      "learning_rate": 3.976576331199782e-05,
      "loss": 4.6686,
      "step": 8015
    },
    {
      "epoch": 0.21550450087330378,
      "grad_norm": 2.322126865386963,
      "learning_rate": 3.975895410595125e-05,
      "loss": 4.4855,
      "step": 8020
    },
    {
      "epoch": 0.21563885530028215,
      "grad_norm": 2.217851400375366,
      "learning_rate": 3.975214489990468e-05,
      "loss": 4.6483,
      "step": 8025
    },
    {
      "epoch": 0.21577320972726052,
      "grad_norm": 2.4423587322235107,
      "learning_rate": 3.97453356938581e-05,
      "loss": 4.4213,
      "step": 8030
    },
    {
      "epoch": 0.2159075641542389,
      "grad_norm": 2.4239814281463623,
      "learning_rate": 3.973852648781152e-05,
      "loss": 4.5082,
      "step": 8035
    },
    {
      "epoch": 0.21604191858121724,
      "grad_norm": 2.3297367095947266,
      "learning_rate": 3.9731717281764945e-05,
      "loss": 4.5876,
      "step": 8040
    },
    {
      "epoch": 0.2161762730081956,
      "grad_norm": 2.1813454627990723,
      "learning_rate": 3.9724908075718374e-05,
      "loss": 4.5188,
      "step": 8045
    },
    {
      "epoch": 0.21631062743517399,
      "grad_norm": 2.172741651535034,
      "learning_rate": 3.9718098869671796e-05,
      "loss": 4.5169,
      "step": 8050
    },
    {
      "epoch": 0.21644498186215236,
      "grad_norm": 2.2912702560424805,
      "learning_rate": 3.9711289663625225e-05,
      "loss": 4.4824,
      "step": 8055
    },
    {
      "epoch": 0.21657933628913073,
      "grad_norm": 2.3370800018310547,
      "learning_rate": 3.970448045757865e-05,
      "loss": 4.5895,
      "step": 8060
    },
    {
      "epoch": 0.2167136907161091,
      "grad_norm": 2.283795118331909,
      "learning_rate": 3.969767125153208e-05,
      "loss": 4.4954,
      "step": 8065
    },
    {
      "epoch": 0.21684804514308748,
      "grad_norm": 2.253535032272339,
      "learning_rate": 3.96908620454855e-05,
      "loss": 4.4669,
      "step": 8070
    },
    {
      "epoch": 0.21698239957006582,
      "grad_norm": 2.4070489406585693,
      "learning_rate": 3.968405283943892e-05,
      "loss": 4.4722,
      "step": 8075
    },
    {
      "epoch": 0.2171167539970442,
      "grad_norm": 2.3137452602386475,
      "learning_rate": 3.967724363339235e-05,
      "loss": 4.3965,
      "step": 8080
    },
    {
      "epoch": 0.21725110842402257,
      "grad_norm": 2.4391114711761475,
      "learning_rate": 3.967043442734577e-05,
      "loss": 4.5998,
      "step": 8085
    },
    {
      "epoch": 0.21738546285100094,
      "grad_norm": 2.4084930419921875,
      "learning_rate": 3.96636252212992e-05,
      "loss": 4.5332,
      "step": 8090
    },
    {
      "epoch": 0.2175198172779793,
      "grad_norm": 2.2953662872314453,
      "learning_rate": 3.9656816015252624e-05,
      "loss": 4.5608,
      "step": 8095
    },
    {
      "epoch": 0.21765417170495768,
      "grad_norm": 2.214895009994507,
      "learning_rate": 3.9650006809206046e-05,
      "loss": 4.5307,
      "step": 8100
    },
    {
      "epoch": 0.21778852613193606,
      "grad_norm": 2.494039297103882,
      "learning_rate": 3.964319760315947e-05,
      "loss": 4.489,
      "step": 8105
    },
    {
      "epoch": 0.21792288055891443,
      "grad_norm": 2.4110639095306396,
      "learning_rate": 3.96363883971129e-05,
      "loss": 4.4221,
      "step": 8110
    },
    {
      "epoch": 0.21805723498589277,
      "grad_norm": 2.4896717071533203,
      "learning_rate": 3.9629579191066327e-05,
      "loss": 4.6174,
      "step": 8115
    },
    {
      "epoch": 0.21819158941287115,
      "grad_norm": 2.271714925765991,
      "learning_rate": 3.962276998501975e-05,
      "loss": 4.6011,
      "step": 8120
    },
    {
      "epoch": 0.21832594383984952,
      "grad_norm": 2.264744758605957,
      "learning_rate": 3.961596077897317e-05,
      "loss": 4.343,
      "step": 8125
    },
    {
      "epoch": 0.2184602982668279,
      "grad_norm": 2.3860461711883545,
      "learning_rate": 3.96091515729266e-05,
      "loss": 4.6053,
      "step": 8130
    },
    {
      "epoch": 0.21859465269380626,
      "grad_norm": 2.51029896736145,
      "learning_rate": 3.960234236688002e-05,
      "loss": 4.5263,
      "step": 8135
    },
    {
      "epoch": 0.21872900712078464,
      "grad_norm": 2.1984598636627197,
      "learning_rate": 3.9595533160833445e-05,
      "loss": 4.5182,
      "step": 8140
    },
    {
      "epoch": 0.218863361547763,
      "grad_norm": 2.2205660343170166,
      "learning_rate": 3.9588723954786874e-05,
      "loss": 4.4588,
      "step": 8145
    },
    {
      "epoch": 0.21899771597474138,
      "grad_norm": 2.320554256439209,
      "learning_rate": 3.95819147487403e-05,
      "loss": 4.4623,
      "step": 8150
    },
    {
      "epoch": 0.21913207040171973,
      "grad_norm": 2.437398910522461,
      "learning_rate": 3.9575105542693725e-05,
      "loss": 4.5282,
      "step": 8155
    },
    {
      "epoch": 0.2192664248286981,
      "grad_norm": 2.2741754055023193,
      "learning_rate": 3.956829633664715e-05,
      "loss": 4.4615,
      "step": 8160
    },
    {
      "epoch": 0.21940077925567647,
      "grad_norm": 2.379453659057617,
      "learning_rate": 3.956148713060057e-05,
      "loss": 4.552,
      "step": 8165
    },
    {
      "epoch": 0.21953513368265484,
      "grad_norm": 2.46022891998291,
      "learning_rate": 3.9554677924554e-05,
      "loss": 4.5353,
      "step": 8170
    },
    {
      "epoch": 0.21966948810963322,
      "grad_norm": 2.3031411170959473,
      "learning_rate": 3.954786871850743e-05,
      "loss": 4.5261,
      "step": 8175
    },
    {
      "epoch": 0.2198038425366116,
      "grad_norm": 2.389411687850952,
      "learning_rate": 3.954105951246085e-05,
      "loss": 4.4525,
      "step": 8180
    },
    {
      "epoch": 0.21993819696358996,
      "grad_norm": 2.386228084564209,
      "learning_rate": 3.953425030641427e-05,
      "loss": 4.5909,
      "step": 8185
    },
    {
      "epoch": 0.22007255139056833,
      "grad_norm": 2.4298057556152344,
      "learning_rate": 3.95274411003677e-05,
      "loss": 4.4921,
      "step": 8190
    },
    {
      "epoch": 0.22020690581754668,
      "grad_norm": 2.4861178398132324,
      "learning_rate": 3.9520631894321123e-05,
      "loss": 4.5671,
      "step": 8195
    },
    {
      "epoch": 0.22034126024452505,
      "grad_norm": 2.6088831424713135,
      "learning_rate": 3.9513822688274546e-05,
      "loss": 4.5072,
      "step": 8200
    },
    {
      "epoch": 0.22047561467150342,
      "grad_norm": 2.4835610389709473,
      "learning_rate": 3.9507013482227975e-05,
      "loss": 4.4169,
      "step": 8205
    },
    {
      "epoch": 0.2206099690984818,
      "grad_norm": 2.3210246562957764,
      "learning_rate": 3.9500204276181404e-05,
      "loss": 4.4646,
      "step": 8210
    },
    {
      "epoch": 0.22074432352546017,
      "grad_norm": 2.2627928256988525,
      "learning_rate": 3.9493395070134826e-05,
      "loss": 4.4458,
      "step": 8215
    },
    {
      "epoch": 0.22087867795243854,
      "grad_norm": 2.3606464862823486,
      "learning_rate": 3.948658586408825e-05,
      "loss": 4.5374,
      "step": 8220
    },
    {
      "epoch": 0.2210130323794169,
      "grad_norm": 2.201375961303711,
      "learning_rate": 3.947977665804167e-05,
      "loss": 4.5093,
      "step": 8225
    },
    {
      "epoch": 0.22114738680639526,
      "grad_norm": 2.2698469161987305,
      "learning_rate": 3.94729674519951e-05,
      "loss": 4.5452,
      "step": 8230
    },
    {
      "epoch": 0.22128174123337363,
      "grad_norm": 2.178785800933838,
      "learning_rate": 3.946615824594853e-05,
      "loss": 4.5701,
      "step": 8235
    },
    {
      "epoch": 0.221416095660352,
      "grad_norm": 2.27838397026062,
      "learning_rate": 3.945934903990195e-05,
      "loss": 4.552,
      "step": 8240
    },
    {
      "epoch": 0.22155045008733038,
      "grad_norm": 2.4617958068847656,
      "learning_rate": 3.945253983385537e-05,
      "loss": 4.4574,
      "step": 8245
    },
    {
      "epoch": 0.22168480451430875,
      "grad_norm": 2.309865713119507,
      "learning_rate": 3.9445730627808796e-05,
      "loss": 4.4584,
      "step": 8250
    },
    {
      "epoch": 0.22181915894128712,
      "grad_norm": 2.5133237838745117,
      "learning_rate": 3.9438921421762225e-05,
      "loss": 4.5018,
      "step": 8255
    },
    {
      "epoch": 0.2219535133682655,
      "grad_norm": 2.528859853744507,
      "learning_rate": 3.943211221571565e-05,
      "loss": 4.5619,
      "step": 8260
    },
    {
      "epoch": 0.22208786779524387,
      "grad_norm": 2.5606112480163574,
      "learning_rate": 3.9425303009669076e-05,
      "loss": 4.5617,
      "step": 8265
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 2.261852979660034,
      "learning_rate": 3.94184938036225e-05,
      "loss": 4.6211,
      "step": 8270
    },
    {
      "epoch": 0.22235657664920058,
      "grad_norm": 2.1548967361450195,
      "learning_rate": 3.941168459757593e-05,
      "loss": 4.5193,
      "step": 8275
    },
    {
      "epoch": 0.22249093107617895,
      "grad_norm": 2.4585540294647217,
      "learning_rate": 3.940487539152935e-05,
      "loss": 4.5097,
      "step": 8280
    },
    {
      "epoch": 0.22262528550315733,
      "grad_norm": 2.482358932495117,
      "learning_rate": 3.939806618548277e-05,
      "loss": 4.5414,
      "step": 8285
    },
    {
      "epoch": 0.2227596399301357,
      "grad_norm": 2.2209229469299316,
      "learning_rate": 3.93912569794362e-05,
      "loss": 4.5977,
      "step": 8290
    },
    {
      "epoch": 0.22289399435711407,
      "grad_norm": 2.4583818912506104,
      "learning_rate": 3.938444777338963e-05,
      "loss": 4.5733,
      "step": 8295
    },
    {
      "epoch": 0.22302834878409245,
      "grad_norm": 2.567460536956787,
      "learning_rate": 3.937763856734305e-05,
      "loss": 4.4606,
      "step": 8300
    },
    {
      "epoch": 0.22316270321107082,
      "grad_norm": 2.2279093265533447,
      "learning_rate": 3.9370829361296474e-05,
      "loss": 4.6824,
      "step": 8305
    },
    {
      "epoch": 0.22329705763804916,
      "grad_norm": 2.3980607986450195,
      "learning_rate": 3.93640201552499e-05,
      "loss": 4.5479,
      "step": 8310
    },
    {
      "epoch": 0.22343141206502753,
      "grad_norm": 2.3374099731445312,
      "learning_rate": 3.9357210949203326e-05,
      "loss": 4.4721,
      "step": 8315
    },
    {
      "epoch": 0.2235657664920059,
      "grad_norm": 2.3053574562072754,
      "learning_rate": 3.935040174315675e-05,
      "loss": 4.4035,
      "step": 8320
    },
    {
      "epoch": 0.22370012091898428,
      "grad_norm": 2.326688289642334,
      "learning_rate": 3.934359253711018e-05,
      "loss": 4.5543,
      "step": 8325
    },
    {
      "epoch": 0.22383447534596265,
      "grad_norm": 2.328249931335449,
      "learning_rate": 3.93367833310636e-05,
      "loss": 4.5431,
      "step": 8330
    },
    {
      "epoch": 0.22396882977294102,
      "grad_norm": 2.3143770694732666,
      "learning_rate": 3.932997412501703e-05,
      "loss": 4.4885,
      "step": 8335
    },
    {
      "epoch": 0.2241031841999194,
      "grad_norm": 2.438821792602539,
      "learning_rate": 3.932316491897045e-05,
      "loss": 4.5825,
      "step": 8340
    },
    {
      "epoch": 0.22423753862689777,
      "grad_norm": 2.4263737201690674,
      "learning_rate": 3.931635571292387e-05,
      "loss": 4.5807,
      "step": 8345
    },
    {
      "epoch": 0.22437189305387611,
      "grad_norm": 2.2539799213409424,
      "learning_rate": 3.9309546506877295e-05,
      "loss": 4.5062,
      "step": 8350
    },
    {
      "epoch": 0.2245062474808545,
      "grad_norm": 2.293640375137329,
      "learning_rate": 3.930273730083073e-05,
      "loss": 4.4612,
      "step": 8355
    },
    {
      "epoch": 0.22464060190783286,
      "grad_norm": 2.491832733154297,
      "learning_rate": 3.929592809478415e-05,
      "loss": 4.5282,
      "step": 8360
    },
    {
      "epoch": 0.22477495633481123,
      "grad_norm": 2.5787856578826904,
      "learning_rate": 3.9289118888737575e-05,
      "loss": 4.5486,
      "step": 8365
    },
    {
      "epoch": 0.2249093107617896,
      "grad_norm": 2.106255054473877,
      "learning_rate": 3.9282309682691e-05,
      "loss": 4.5548,
      "step": 8370
    },
    {
      "epoch": 0.22504366518876798,
      "grad_norm": 2.4151759147644043,
      "learning_rate": 3.927550047664443e-05,
      "loss": 4.4303,
      "step": 8375
    },
    {
      "epoch": 0.22517801961574635,
      "grad_norm": 2.3859143257141113,
      "learning_rate": 3.926869127059785e-05,
      "loss": 4.4398,
      "step": 8380
    },
    {
      "epoch": 0.2253123740427247,
      "grad_norm": 2.494396686553955,
      "learning_rate": 3.926188206455128e-05,
      "loss": 4.3921,
      "step": 8385
    },
    {
      "epoch": 0.22544672846970307,
      "grad_norm": 3.0563199520111084,
      "learning_rate": 3.92550728585047e-05,
      "loss": 4.4749,
      "step": 8390
    },
    {
      "epoch": 0.22558108289668144,
      "grad_norm": 2.2936222553253174,
      "learning_rate": 3.924826365245812e-05,
      "loss": 4.458,
      "step": 8395
    },
    {
      "epoch": 0.2257154373236598,
      "grad_norm": 2.1167876720428467,
      "learning_rate": 3.924145444641155e-05,
      "loss": 4.506,
      "step": 8400
    },
    {
      "epoch": 0.22584979175063818,
      "grad_norm": 2.3850529193878174,
      "learning_rate": 3.9234645240364974e-05,
      "loss": 4.4521,
      "step": 8405
    },
    {
      "epoch": 0.22598414617761656,
      "grad_norm": 2.3167271614074707,
      "learning_rate": 3.9227836034318396e-05,
      "loss": 4.4543,
      "step": 8410
    },
    {
      "epoch": 0.22611850060459493,
      "grad_norm": 2.427035093307495,
      "learning_rate": 3.9221026828271825e-05,
      "loss": 4.5651,
      "step": 8415
    },
    {
      "epoch": 0.2262528550315733,
      "grad_norm": 2.547551155090332,
      "learning_rate": 3.9214217622225254e-05,
      "loss": 4.3067,
      "step": 8420
    },
    {
      "epoch": 0.22638720945855165,
      "grad_norm": 2.3174264430999756,
      "learning_rate": 3.9207408416178677e-05,
      "loss": 4.5975,
      "step": 8425
    },
    {
      "epoch": 0.22652156388553002,
      "grad_norm": 2.3896431922912598,
      "learning_rate": 3.92005992101321e-05,
      "loss": 4.3266,
      "step": 8430
    },
    {
      "epoch": 0.2266559183125084,
      "grad_norm": 2.4772348403930664,
      "learning_rate": 3.919379000408552e-05,
      "loss": 4.4856,
      "step": 8435
    },
    {
      "epoch": 0.22679027273948676,
      "grad_norm": 2.077833890914917,
      "learning_rate": 3.918698079803895e-05,
      "loss": 4.4156,
      "step": 8440
    },
    {
      "epoch": 0.22692462716646514,
      "grad_norm": 2.3736624717712402,
      "learning_rate": 3.918017159199238e-05,
      "loss": 4.4443,
      "step": 8445
    },
    {
      "epoch": 0.2270589815934435,
      "grad_norm": 2.409799337387085,
      "learning_rate": 3.91733623859458e-05,
      "loss": 4.5994,
      "step": 8450
    },
    {
      "epoch": 0.22719333602042188,
      "grad_norm": 2.3161847591400146,
      "learning_rate": 3.9166553179899224e-05,
      "loss": 4.5108,
      "step": 8455
    },
    {
      "epoch": 0.22732769044740025,
      "grad_norm": 2.3115291595458984,
      "learning_rate": 3.915974397385265e-05,
      "loss": 4.5899,
      "step": 8460
    },
    {
      "epoch": 0.2274620448743786,
      "grad_norm": 2.250058650970459,
      "learning_rate": 3.9152934767806075e-05,
      "loss": 4.5587,
      "step": 8465
    },
    {
      "epoch": 0.22759639930135697,
      "grad_norm": 2.448368787765503,
      "learning_rate": 3.91461255617595e-05,
      "loss": 4.519,
      "step": 8470
    },
    {
      "epoch": 0.22773075372833534,
      "grad_norm": 2.4634785652160645,
      "learning_rate": 3.9139316355712926e-05,
      "loss": 4.4725,
      "step": 8475
    },
    {
      "epoch": 0.22786510815531372,
      "grad_norm": 2.447174310684204,
      "learning_rate": 3.9132507149666355e-05,
      "loss": 4.5119,
      "step": 8480
    },
    {
      "epoch": 0.2279994625822921,
      "grad_norm": 2.297747850418091,
      "learning_rate": 3.912569794361978e-05,
      "loss": 4.47,
      "step": 8485
    },
    {
      "epoch": 0.22813381700927046,
      "grad_norm": 2.327514410018921,
      "learning_rate": 3.91188887375732e-05,
      "loss": 4.5262,
      "step": 8490
    },
    {
      "epoch": 0.22826817143624883,
      "grad_norm": 2.523096799850464,
      "learning_rate": 3.911207953152662e-05,
      "loss": 4.5991,
      "step": 8495
    },
    {
      "epoch": 0.2284025258632272,
      "grad_norm": 2.069838285446167,
      "learning_rate": 3.910527032548005e-05,
      "loss": 4.4604,
      "step": 8500
    },
    {
      "epoch": 0.22853688029020555,
      "grad_norm": 2.2846767902374268,
      "learning_rate": 3.909846111943348e-05,
      "loss": 4.3883,
      "step": 8505
    },
    {
      "epoch": 0.22867123471718392,
      "grad_norm": 2.458731174468994,
      "learning_rate": 3.90916519133869e-05,
      "loss": 4.4619,
      "step": 8510
    },
    {
      "epoch": 0.2288055891441623,
      "grad_norm": 2.4549267292022705,
      "learning_rate": 3.9084842707340325e-05,
      "loss": 4.5344,
      "step": 8515
    },
    {
      "epoch": 0.22893994357114067,
      "grad_norm": 2.4725306034088135,
      "learning_rate": 3.9078033501293754e-05,
      "loss": 4.4581,
      "step": 8520
    },
    {
      "epoch": 0.22907429799811904,
      "grad_norm": 2.3027560710906982,
      "learning_rate": 3.9071224295247176e-05,
      "loss": 4.4477,
      "step": 8525
    },
    {
      "epoch": 0.22920865242509741,
      "grad_norm": 2.1911067962646484,
      "learning_rate": 3.90644150892006e-05,
      "loss": 4.4504,
      "step": 8530
    },
    {
      "epoch": 0.2293430068520758,
      "grad_norm": 2.380733013153076,
      "learning_rate": 3.905760588315403e-05,
      "loss": 4.3704,
      "step": 8535
    },
    {
      "epoch": 0.22947736127905413,
      "grad_norm": 2.0949010848999023,
      "learning_rate": 3.905079667710745e-05,
      "loss": 4.4612,
      "step": 8540
    },
    {
      "epoch": 0.2296117157060325,
      "grad_norm": 2.3797502517700195,
      "learning_rate": 3.904398747106088e-05,
      "loss": 4.4963,
      "step": 8545
    },
    {
      "epoch": 0.22974607013301088,
      "grad_norm": 2.5473086833953857,
      "learning_rate": 3.90371782650143e-05,
      "loss": 4.4186,
      "step": 8550
    },
    {
      "epoch": 0.22988042455998925,
      "grad_norm": 2.38205885887146,
      "learning_rate": 3.903036905896772e-05,
      "loss": 4.5383,
      "step": 8555
    },
    {
      "epoch": 0.23001477898696762,
      "grad_norm": 2.4715940952301025,
      "learning_rate": 3.9023559852921146e-05,
      "loss": 4.4418,
      "step": 8560
    },
    {
      "epoch": 0.230149133413946,
      "grad_norm": 2.22713041305542,
      "learning_rate": 3.9016750646874575e-05,
      "loss": 4.5647,
      "step": 8565
    },
    {
      "epoch": 0.23028348784092437,
      "grad_norm": 2.3160481452941895,
      "learning_rate": 3.9009941440828004e-05,
      "loss": 4.5537,
      "step": 8570
    },
    {
      "epoch": 0.23041784226790274,
      "grad_norm": 2.2823052406311035,
      "learning_rate": 3.9003132234781426e-05,
      "loss": 4.56,
      "step": 8575
    },
    {
      "epoch": 0.23055219669488108,
      "grad_norm": 2.2835984230041504,
      "learning_rate": 3.899632302873485e-05,
      "loss": 4.4945,
      "step": 8580
    },
    {
      "epoch": 0.23068655112185946,
      "grad_norm": 2.418950319290161,
      "learning_rate": 3.898951382268828e-05,
      "loss": 4.4775,
      "step": 8585
    },
    {
      "epoch": 0.23082090554883783,
      "grad_norm": 2.2225818634033203,
      "learning_rate": 3.89827046166417e-05,
      "loss": 4.5085,
      "step": 8590
    },
    {
      "epoch": 0.2309552599758162,
      "grad_norm": 2.150221347808838,
      "learning_rate": 3.897589541059513e-05,
      "loss": 4.4651,
      "step": 8595
    },
    {
      "epoch": 0.23108961440279457,
      "grad_norm": 2.4287526607513428,
      "learning_rate": 3.896908620454855e-05,
      "loss": 4.616,
      "step": 8600
    },
    {
      "epoch": 0.23122396882977295,
      "grad_norm": 2.370195150375366,
      "learning_rate": 3.896227699850198e-05,
      "loss": 4.5388,
      "step": 8605
    },
    {
      "epoch": 0.23135832325675132,
      "grad_norm": 2.3494715690612793,
      "learning_rate": 3.89554677924554e-05,
      "loss": 4.4648,
      "step": 8610
    },
    {
      "epoch": 0.2314926776837297,
      "grad_norm": 2.2470271587371826,
      "learning_rate": 3.8948658586408824e-05,
      "loss": 4.4926,
      "step": 8615
    },
    {
      "epoch": 0.23162703211070804,
      "grad_norm": 2.0862324237823486,
      "learning_rate": 3.894184938036225e-05,
      "loss": 4.5883,
      "step": 8620
    },
    {
      "epoch": 0.2317613865376864,
      "grad_norm": 2.2382023334503174,
      "learning_rate": 3.8935040174315676e-05,
      "loss": 4.4649,
      "step": 8625
    },
    {
      "epoch": 0.23189574096466478,
      "grad_norm": 2.4097044467926025,
      "learning_rate": 3.8928230968269105e-05,
      "loss": 4.5404,
      "step": 8630
    },
    {
      "epoch": 0.23203009539164315,
      "grad_norm": 2.0999863147735596,
      "learning_rate": 3.892142176222253e-05,
      "loss": 4.53,
      "step": 8635
    },
    {
      "epoch": 0.23216444981862153,
      "grad_norm": 2.3411190509796143,
      "learning_rate": 3.891461255617595e-05,
      "loss": 4.5707,
      "step": 8640
    },
    {
      "epoch": 0.2322988042455999,
      "grad_norm": 2.1405792236328125,
      "learning_rate": 3.890780335012938e-05,
      "loss": 4.3955,
      "step": 8645
    },
    {
      "epoch": 0.23243315867257827,
      "grad_norm": 2.4790520668029785,
      "learning_rate": 3.89009941440828e-05,
      "loss": 4.4506,
      "step": 8650
    },
    {
      "epoch": 0.23256751309955664,
      "grad_norm": 2.6332178115844727,
      "learning_rate": 3.889418493803623e-05,
      "loss": 4.5147,
      "step": 8655
    },
    {
      "epoch": 0.232701867526535,
      "grad_norm": 2.314347505569458,
      "learning_rate": 3.888737573198965e-05,
      "loss": 4.4196,
      "step": 8660
    },
    {
      "epoch": 0.23283622195351336,
      "grad_norm": 2.422787666320801,
      "learning_rate": 3.888056652594308e-05,
      "loss": 4.4499,
      "step": 8665
    },
    {
      "epoch": 0.23297057638049173,
      "grad_norm": 2.620337724685669,
      "learning_rate": 3.88737573198965e-05,
      "loss": 4.5771,
      "step": 8670
    },
    {
      "epoch": 0.2331049308074701,
      "grad_norm": 2.2113358974456787,
      "learning_rate": 3.8866948113849926e-05,
      "loss": 4.5354,
      "step": 8675
    },
    {
      "epoch": 0.23323928523444848,
      "grad_norm": 2.293414354324341,
      "learning_rate": 3.886013890780335e-05,
      "loss": 4.5862,
      "step": 8680
    },
    {
      "epoch": 0.23337363966142685,
      "grad_norm": 2.2656936645507812,
      "learning_rate": 3.885332970175678e-05,
      "loss": 4.4915,
      "step": 8685
    },
    {
      "epoch": 0.23350799408840522,
      "grad_norm": 2.1302578449249268,
      "learning_rate": 3.8846520495710206e-05,
      "loss": 4.5319,
      "step": 8690
    },
    {
      "epoch": 0.23364234851538357,
      "grad_norm": 2.1620612144470215,
      "learning_rate": 3.883971128966363e-05,
      "loss": 4.5304,
      "step": 8695
    },
    {
      "epoch": 0.23377670294236194,
      "grad_norm": 2.3192574977874756,
      "learning_rate": 3.883290208361705e-05,
      "loss": 4.4739,
      "step": 8700
    },
    {
      "epoch": 0.23391105736934031,
      "grad_norm": 2.243558645248413,
      "learning_rate": 3.882609287757047e-05,
      "loss": 4.5051,
      "step": 8705
    },
    {
      "epoch": 0.2340454117963187,
      "grad_norm": 2.341973066329956,
      "learning_rate": 3.88192836715239e-05,
      "loss": 4.3077,
      "step": 8710
    },
    {
      "epoch": 0.23417976622329706,
      "grad_norm": 2.272465705871582,
      "learning_rate": 3.881247446547733e-05,
      "loss": 4.4325,
      "step": 8715
    },
    {
      "epoch": 0.23431412065027543,
      "grad_norm": 2.4624316692352295,
      "learning_rate": 3.880566525943075e-05,
      "loss": 4.5258,
      "step": 8720
    },
    {
      "epoch": 0.2344484750772538,
      "grad_norm": 2.387444019317627,
      "learning_rate": 3.8798856053384175e-05,
      "loss": 4.4817,
      "step": 8725
    },
    {
      "epoch": 0.23458282950423218,
      "grad_norm": 2.8348617553710938,
      "learning_rate": 3.8792046847337604e-05,
      "loss": 4.4936,
      "step": 8730
    },
    {
      "epoch": 0.23471718393121052,
      "grad_norm": 2.4759609699249268,
      "learning_rate": 3.878523764129103e-05,
      "loss": 4.4255,
      "step": 8735
    },
    {
      "epoch": 0.2348515383581889,
      "grad_norm": 2.4378538131713867,
      "learning_rate": 3.877842843524445e-05,
      "loss": 4.5988,
      "step": 8740
    },
    {
      "epoch": 0.23498589278516727,
      "grad_norm": 2.287172555923462,
      "learning_rate": 3.877161922919788e-05,
      "loss": 4.5726,
      "step": 8745
    },
    {
      "epoch": 0.23512024721214564,
      "grad_norm": 2.2528538703918457,
      "learning_rate": 3.876481002315131e-05,
      "loss": 4.5549,
      "step": 8750
    },
    {
      "epoch": 0.235254601639124,
      "grad_norm": 2.2703864574432373,
      "learning_rate": 3.875800081710473e-05,
      "loss": 4.4715,
      "step": 8755
    },
    {
      "epoch": 0.23538895606610238,
      "grad_norm": 2.429065704345703,
      "learning_rate": 3.875119161105815e-05,
      "loss": 4.4493,
      "step": 8760
    },
    {
      "epoch": 0.23552331049308076,
      "grad_norm": 2.346167802810669,
      "learning_rate": 3.8744382405011574e-05,
      "loss": 4.5248,
      "step": 8765
    },
    {
      "epoch": 0.23565766492005913,
      "grad_norm": 2.167602062225342,
      "learning_rate": 3.8737573198965e-05,
      "loss": 4.466,
      "step": 8770
    },
    {
      "epoch": 0.23579201934703747,
      "grad_norm": 2.6131200790405273,
      "learning_rate": 3.8730763992918425e-05,
      "loss": 4.5559,
      "step": 8775
    },
    {
      "epoch": 0.23592637377401585,
      "grad_norm": 2.3826704025268555,
      "learning_rate": 3.8723954786871854e-05,
      "loss": 4.4304,
      "step": 8780
    },
    {
      "epoch": 0.23606072820099422,
      "grad_norm": 2.5408835411071777,
      "learning_rate": 3.8717145580825277e-05,
      "loss": 4.5271,
      "step": 8785
    },
    {
      "epoch": 0.2361950826279726,
      "grad_norm": 2.528095245361328,
      "learning_rate": 3.8710336374778706e-05,
      "loss": 4.3296,
      "step": 8790
    },
    {
      "epoch": 0.23632943705495096,
      "grad_norm": 2.3806045055389404,
      "learning_rate": 3.870352716873213e-05,
      "loss": 4.3957,
      "step": 8795
    },
    {
      "epoch": 0.23646379148192934,
      "grad_norm": 2.254794120788574,
      "learning_rate": 3.869671796268555e-05,
      "loss": 4.4552,
      "step": 8800
    },
    {
      "epoch": 0.2365981459089077,
      "grad_norm": 2.223202705383301,
      "learning_rate": 3.868990875663898e-05,
      "loss": 4.4416,
      "step": 8805
    },
    {
      "epoch": 0.23673250033588608,
      "grad_norm": 2.321711301803589,
      "learning_rate": 3.868309955059241e-05,
      "loss": 4.5761,
      "step": 8810
    },
    {
      "epoch": 0.23686685476286443,
      "grad_norm": 2.138576030731201,
      "learning_rate": 3.867629034454583e-05,
      "loss": 4.3813,
      "step": 8815
    },
    {
      "epoch": 0.2370012091898428,
      "grad_norm": 2.2315521240234375,
      "learning_rate": 3.866948113849925e-05,
      "loss": 4.3917,
      "step": 8820
    },
    {
      "epoch": 0.23713556361682117,
      "grad_norm": 2.1541736125946045,
      "learning_rate": 3.8662671932452675e-05,
      "loss": 4.4306,
      "step": 8825
    },
    {
      "epoch": 0.23726991804379954,
      "grad_norm": 2.490097761154175,
      "learning_rate": 3.86558627264061e-05,
      "loss": 4.5851,
      "step": 8830
    },
    {
      "epoch": 0.23740427247077792,
      "grad_norm": 2.5991132259368896,
      "learning_rate": 3.8649053520359526e-05,
      "loss": 4.4972,
      "step": 8835
    },
    {
      "epoch": 0.2375386268977563,
      "grad_norm": 2.345184803009033,
      "learning_rate": 3.8642244314312955e-05,
      "loss": 4.4133,
      "step": 8840
    },
    {
      "epoch": 0.23767298132473466,
      "grad_norm": 2.3713176250457764,
      "learning_rate": 3.863543510826638e-05,
      "loss": 4.5444,
      "step": 8845
    },
    {
      "epoch": 0.237807335751713,
      "grad_norm": 2.3636770248413086,
      "learning_rate": 3.86286259022198e-05,
      "loss": 4.4242,
      "step": 8850
    },
    {
      "epoch": 0.23794169017869138,
      "grad_norm": 2.3559272289276123,
      "learning_rate": 3.862181669617323e-05,
      "loss": 4.528,
      "step": 8855
    },
    {
      "epoch": 0.23807604460566975,
      "grad_norm": 2.328324556350708,
      "learning_rate": 3.861500749012665e-05,
      "loss": 4.4464,
      "step": 8860
    },
    {
      "epoch": 0.23821039903264812,
      "grad_norm": 2.1875815391540527,
      "learning_rate": 3.860819828408008e-05,
      "loss": 4.3888,
      "step": 8865
    },
    {
      "epoch": 0.2383447534596265,
      "grad_norm": 2.2936065196990967,
      "learning_rate": 3.86013890780335e-05,
      "loss": 4.4511,
      "step": 8870
    },
    {
      "epoch": 0.23847910788660487,
      "grad_norm": 2.5145628452301025,
      "learning_rate": 3.859457987198693e-05,
      "loss": 4.4446,
      "step": 8875
    },
    {
      "epoch": 0.23861346231358324,
      "grad_norm": 2.300708293914795,
      "learning_rate": 3.8587770665940354e-05,
      "loss": 4.4348,
      "step": 8880
    },
    {
      "epoch": 0.2387478167405616,
      "grad_norm": 2.4654135704040527,
      "learning_rate": 3.8580961459893776e-05,
      "loss": 4.3634,
      "step": 8885
    },
    {
      "epoch": 0.23888217116753996,
      "grad_norm": 2.2008254528045654,
      "learning_rate": 3.85741522538472e-05,
      "loss": 4.5456,
      "step": 8890
    },
    {
      "epoch": 0.23901652559451833,
      "grad_norm": 2.32795786857605,
      "learning_rate": 3.856734304780063e-05,
      "loss": 4.3412,
      "step": 8895
    },
    {
      "epoch": 0.2391508800214967,
      "grad_norm": 2.6038146018981934,
      "learning_rate": 3.8560533841754056e-05,
      "loss": 4.5066,
      "step": 8900
    },
    {
      "epoch": 0.23928523444847508,
      "grad_norm": 2.289285898208618,
      "learning_rate": 3.855372463570748e-05,
      "loss": 4.4866,
      "step": 8905
    },
    {
      "epoch": 0.23941958887545345,
      "grad_norm": 2.2476844787597656,
      "learning_rate": 3.85469154296609e-05,
      "loss": 4.3769,
      "step": 8910
    },
    {
      "epoch": 0.23955394330243182,
      "grad_norm": 2.1925363540649414,
      "learning_rate": 3.854010622361433e-05,
      "loss": 4.4785,
      "step": 8915
    },
    {
      "epoch": 0.2396882977294102,
      "grad_norm": 2.213681697845459,
      "learning_rate": 3.853329701756775e-05,
      "loss": 4.3922,
      "step": 8920
    },
    {
      "epoch": 0.23982265215638857,
      "grad_norm": 2.39823055267334,
      "learning_rate": 3.852648781152118e-05,
      "loss": 4.4942,
      "step": 8925
    },
    {
      "epoch": 0.2399570065833669,
      "grad_norm": 2.709286689758301,
      "learning_rate": 3.8519678605474604e-05,
      "loss": 4.4324,
      "step": 8930
    },
    {
      "epoch": 0.24009136101034528,
      "grad_norm": 2.4521644115448,
      "learning_rate": 3.851286939942803e-05,
      "loss": 4.3661,
      "step": 8935
    },
    {
      "epoch": 0.24022571543732366,
      "grad_norm": 2.222060203552246,
      "learning_rate": 3.8506060193381455e-05,
      "loss": 4.5599,
      "step": 8940
    },
    {
      "epoch": 0.24036006986430203,
      "grad_norm": 2.293104648590088,
      "learning_rate": 3.849925098733488e-05,
      "loss": 4.5562,
      "step": 8945
    },
    {
      "epoch": 0.2404944242912804,
      "grad_norm": 2.38869571685791,
      "learning_rate": 3.84924417812883e-05,
      "loss": 4.5339,
      "step": 8950
    },
    {
      "epoch": 0.24062877871825877,
      "grad_norm": 2.325330972671509,
      "learning_rate": 3.848563257524173e-05,
      "loss": 4.5861,
      "step": 8955
    },
    {
      "epoch": 0.24076313314523715,
      "grad_norm": 2.5010159015655518,
      "learning_rate": 3.847882336919516e-05,
      "loss": 4.3589,
      "step": 8960
    },
    {
      "epoch": 0.2408974875722155,
      "grad_norm": 2.554168224334717,
      "learning_rate": 3.847201416314858e-05,
      "loss": 4.4188,
      "step": 8965
    },
    {
      "epoch": 0.24103184199919386,
      "grad_norm": 2.268612861633301,
      "learning_rate": 3.8465204957102e-05,
      "loss": 4.4852,
      "step": 8970
    },
    {
      "epoch": 0.24116619642617224,
      "grad_norm": 2.3356919288635254,
      "learning_rate": 3.8458395751055424e-05,
      "loss": 4.4218,
      "step": 8975
    },
    {
      "epoch": 0.2413005508531506,
      "grad_norm": 2.5257294178009033,
      "learning_rate": 3.8451586545008853e-05,
      "loss": 4.5004,
      "step": 8980
    },
    {
      "epoch": 0.24143490528012898,
      "grad_norm": 2.127638578414917,
      "learning_rate": 3.8444777338962276e-05,
      "loss": 4.3875,
      "step": 8985
    },
    {
      "epoch": 0.24156925970710735,
      "grad_norm": 2.363490581512451,
      "learning_rate": 3.8437968132915705e-05,
      "loss": 4.4058,
      "step": 8990
    },
    {
      "epoch": 0.24170361413408573,
      "grad_norm": 2.069204092025757,
      "learning_rate": 3.843115892686913e-05,
      "loss": 4.4866,
      "step": 8995
    },
    {
      "epoch": 0.2418379685610641,
      "grad_norm": 2.451849937438965,
      "learning_rate": 3.8424349720822556e-05,
      "loss": 4.2889,
      "step": 9000
    },
    {
      "epoch": 0.24197232298804244,
      "grad_norm": 2.3634588718414307,
      "learning_rate": 3.841754051477598e-05,
      "loss": 4.4167,
      "step": 9005
    },
    {
      "epoch": 0.24210667741502082,
      "grad_norm": 2.740828275680542,
      "learning_rate": 3.84107313087294e-05,
      "loss": 4.5071,
      "step": 9010
    },
    {
      "epoch": 0.2422410318419992,
      "grad_norm": 2.3479790687561035,
      "learning_rate": 3.840392210268283e-05,
      "loss": 4.4122,
      "step": 9015
    },
    {
      "epoch": 0.24237538626897756,
      "grad_norm": 2.4052951335906982,
      "learning_rate": 3.839711289663626e-05,
      "loss": 4.4962,
      "step": 9020
    },
    {
      "epoch": 0.24250974069595593,
      "grad_norm": 2.1615025997161865,
      "learning_rate": 3.839030369058968e-05,
      "loss": 4.4436,
      "step": 9025
    },
    {
      "epoch": 0.2426440951229343,
      "grad_norm": 2.497177839279175,
      "learning_rate": 3.83834944845431e-05,
      "loss": 4.4895,
      "step": 9030
    },
    {
      "epoch": 0.24277844954991268,
      "grad_norm": 2.334991216659546,
      "learning_rate": 3.8376685278496525e-05,
      "loss": 4.4673,
      "step": 9035
    },
    {
      "epoch": 0.24291280397689105,
      "grad_norm": 2.167146921157837,
      "learning_rate": 3.8369876072449955e-05,
      "loss": 4.4415,
      "step": 9040
    },
    {
      "epoch": 0.2430471584038694,
      "grad_norm": 2.415437936782837,
      "learning_rate": 3.836306686640338e-05,
      "loss": 4.4382,
      "step": 9045
    },
    {
      "epoch": 0.24318151283084777,
      "grad_norm": 2.3628439903259277,
      "learning_rate": 3.8356257660356806e-05,
      "loss": 4.3692,
      "step": 9050
    },
    {
      "epoch": 0.24331586725782614,
      "grad_norm": 2.343013286590576,
      "learning_rate": 3.834944845431023e-05,
      "loss": 4.4133,
      "step": 9055
    },
    {
      "epoch": 0.2434502216848045,
      "grad_norm": 2.317664384841919,
      "learning_rate": 3.834263924826366e-05,
      "loss": 4.4779,
      "step": 9060
    },
    {
      "epoch": 0.24358457611178289,
      "grad_norm": 2.4156370162963867,
      "learning_rate": 3.833583004221708e-05,
      "loss": 4.3418,
      "step": 9065
    },
    {
      "epoch": 0.24371893053876126,
      "grad_norm": 2.3349876403808594,
      "learning_rate": 3.83290208361705e-05,
      "loss": 4.3203,
      "step": 9070
    },
    {
      "epoch": 0.24385328496573963,
      "grad_norm": 2.193326234817505,
      "learning_rate": 3.832221163012393e-05,
      "loss": 4.525,
      "step": 9075
    },
    {
      "epoch": 0.243987639392718,
      "grad_norm": 2.5219979286193848,
      "learning_rate": 3.831540242407736e-05,
      "loss": 4.6186,
      "step": 9080
    },
    {
      "epoch": 0.24412199381969635,
      "grad_norm": 2.4387123584747314,
      "learning_rate": 3.830859321803078e-05,
      "loss": 4.4607,
      "step": 9085
    },
    {
      "epoch": 0.24425634824667472,
      "grad_norm": 2.266798973083496,
      "learning_rate": 3.8301784011984204e-05,
      "loss": 4.5249,
      "step": 9090
    },
    {
      "epoch": 0.2443907026736531,
      "grad_norm": 2.3985793590545654,
      "learning_rate": 3.8294974805937627e-05,
      "loss": 4.509,
      "step": 9095
    },
    {
      "epoch": 0.24452505710063147,
      "grad_norm": 2.538137912750244,
      "learning_rate": 3.8288165599891056e-05,
      "loss": 4.5016,
      "step": 9100
    },
    {
      "epoch": 0.24465941152760984,
      "grad_norm": 2.610562562942505,
      "learning_rate": 3.828135639384448e-05,
      "loss": 4.6595,
      "step": 9105
    },
    {
      "epoch": 0.2447937659545882,
      "grad_norm": 2.4562532901763916,
      "learning_rate": 3.827454718779791e-05,
      "loss": 4.4739,
      "step": 9110
    },
    {
      "epoch": 0.24492812038156658,
      "grad_norm": 2.3436148166656494,
      "learning_rate": 3.826773798175133e-05,
      "loss": 4.5131,
      "step": 9115
    },
    {
      "epoch": 0.24506247480854493,
      "grad_norm": 2.5559749603271484,
      "learning_rate": 3.826092877570476e-05,
      "loss": 4.37,
      "step": 9120
    },
    {
      "epoch": 0.2451968292355233,
      "grad_norm": 2.1195874214172363,
      "learning_rate": 3.825411956965818e-05,
      "loss": 4.5165,
      "step": 9125
    },
    {
      "epoch": 0.24533118366250167,
      "grad_norm": 2.151759147644043,
      "learning_rate": 3.82473103636116e-05,
      "loss": 4.5568,
      "step": 9130
    },
    {
      "epoch": 0.24546553808948005,
      "grad_norm": 2.2366557121276855,
      "learning_rate": 3.824050115756503e-05,
      "loss": 4.5393,
      "step": 9135
    },
    {
      "epoch": 0.24559989251645842,
      "grad_norm": 2.3647451400756836,
      "learning_rate": 3.8233691951518454e-05,
      "loss": 4.4958,
      "step": 9140
    },
    {
      "epoch": 0.2457342469434368,
      "grad_norm": 2.345599889755249,
      "learning_rate": 3.822688274547188e-05,
      "loss": 4.5471,
      "step": 9145
    },
    {
      "epoch": 0.24586860137041516,
      "grad_norm": 2.45613169670105,
      "learning_rate": 3.8220073539425305e-05,
      "loss": 4.439,
      "step": 9150
    },
    {
      "epoch": 0.24600295579739354,
      "grad_norm": 2.329835891723633,
      "learning_rate": 3.821326433337873e-05,
      "loss": 4.6392,
      "step": 9155
    },
    {
      "epoch": 0.24613731022437188,
      "grad_norm": 2.1473677158355713,
      "learning_rate": 3.820645512733215e-05,
      "loss": 4.3028,
      "step": 9160
    },
    {
      "epoch": 0.24627166465135025,
      "grad_norm": 2.5760326385498047,
      "learning_rate": 3.819964592128558e-05,
      "loss": 4.4658,
      "step": 9165
    },
    {
      "epoch": 0.24640601907832863,
      "grad_norm": 2.2796411514282227,
      "learning_rate": 3.819283671523901e-05,
      "loss": 4.544,
      "step": 9170
    },
    {
      "epoch": 0.246540373505307,
      "grad_norm": 2.3019275665283203,
      "learning_rate": 3.818602750919243e-05,
      "loss": 4.4505,
      "step": 9175
    },
    {
      "epoch": 0.24667472793228537,
      "grad_norm": 2.1995794773101807,
      "learning_rate": 3.817921830314585e-05,
      "loss": 4.4372,
      "step": 9180
    },
    {
      "epoch": 0.24680908235926374,
      "grad_norm": 2.2534918785095215,
      "learning_rate": 3.817240909709928e-05,
      "loss": 4.5352,
      "step": 9185
    },
    {
      "epoch": 0.24694343678624212,
      "grad_norm": 2.312190532684326,
      "learning_rate": 3.8165599891052704e-05,
      "loss": 4.4169,
      "step": 9190
    },
    {
      "epoch": 0.2470777912132205,
      "grad_norm": 2.111131191253662,
      "learning_rate": 3.8158790685006126e-05,
      "loss": 4.5093,
      "step": 9195
    },
    {
      "epoch": 0.24721214564019883,
      "grad_norm": 2.5359935760498047,
      "learning_rate": 3.8151981478959555e-05,
      "loss": 4.4724,
      "step": 9200
    },
    {
      "epoch": 0.2473465000671772,
      "grad_norm": 2.1045002937316895,
      "learning_rate": 3.8145172272912984e-05,
      "loss": 4.5626,
      "step": 9205
    },
    {
      "epoch": 0.24748085449415558,
      "grad_norm": 2.2645952701568604,
      "learning_rate": 3.8138363066866407e-05,
      "loss": 4.3804,
      "step": 9210
    },
    {
      "epoch": 0.24761520892113395,
      "grad_norm": 2.149087905883789,
      "learning_rate": 3.813155386081983e-05,
      "loss": 4.4206,
      "step": 9215
    },
    {
      "epoch": 0.24774956334811232,
      "grad_norm": 2.368006467819214,
      "learning_rate": 3.812474465477325e-05,
      "loss": 4.4235,
      "step": 9220
    },
    {
      "epoch": 0.2478839177750907,
      "grad_norm": 2.2952077388763428,
      "learning_rate": 3.811793544872668e-05,
      "loss": 4.4773,
      "step": 9225
    },
    {
      "epoch": 0.24801827220206907,
      "grad_norm": 2.197417736053467,
      "learning_rate": 3.811112624268011e-05,
      "loss": 4.4541,
      "step": 9230
    },
    {
      "epoch": 0.24815262662904744,
      "grad_norm": 2.4365744590759277,
      "learning_rate": 3.810431703663353e-05,
      "loss": 4.3973,
      "step": 9235
    },
    {
      "epoch": 0.24828698105602579,
      "grad_norm": 2.6713833808898926,
      "learning_rate": 3.8097507830586954e-05,
      "loss": 4.5353,
      "step": 9240
    },
    {
      "epoch": 0.24842133548300416,
      "grad_norm": 2.548422336578369,
      "learning_rate": 3.809069862454038e-05,
      "loss": 4.4258,
      "step": 9245
    },
    {
      "epoch": 0.24855568990998253,
      "grad_norm": 2.3496651649475098,
      "learning_rate": 3.8083889418493805e-05,
      "loss": 4.6104,
      "step": 9250
    },
    {
      "epoch": 0.2486900443369609,
      "grad_norm": 2.1973977088928223,
      "learning_rate": 3.807708021244723e-05,
      "loss": 4.4787,
      "step": 9255
    },
    {
      "epoch": 0.24882439876393928,
      "grad_norm": 2.3225276470184326,
      "learning_rate": 3.8070271006400656e-05,
      "loss": 4.4483,
      "step": 9260
    },
    {
      "epoch": 0.24895875319091765,
      "grad_norm": 2.1749799251556396,
      "learning_rate": 3.8063461800354085e-05,
      "loss": 4.5217,
      "step": 9265
    },
    {
      "epoch": 0.24909310761789602,
      "grad_norm": 2.273516893386841,
      "learning_rate": 3.805665259430751e-05,
      "loss": 4.4474,
      "step": 9270
    },
    {
      "epoch": 0.24922746204487436,
      "grad_norm": 2.4002926349639893,
      "learning_rate": 3.804984338826093e-05,
      "loss": 4.5451,
      "step": 9275
    },
    {
      "epoch": 0.24936181647185274,
      "grad_norm": 2.4176535606384277,
      "learning_rate": 3.804303418221435e-05,
      "loss": 4.3978,
      "step": 9280
    },
    {
      "epoch": 0.2494961708988311,
      "grad_norm": 2.40787410736084,
      "learning_rate": 3.803622497616778e-05,
      "loss": 4.474,
      "step": 9285
    },
    {
      "epoch": 0.24963052532580948,
      "grad_norm": 2.287524461746216,
      "learning_rate": 3.802941577012121e-05,
      "loss": 4.5904,
      "step": 9290
    },
    {
      "epoch": 0.24976487975278786,
      "grad_norm": 2.2771718502044678,
      "learning_rate": 3.802260656407463e-05,
      "loss": 4.507,
      "step": 9295
    },
    {
      "epoch": 0.24989923417976623,
      "grad_norm": 2.3056728839874268,
      "learning_rate": 3.8015797358028055e-05,
      "loss": 4.4315,
      "step": 9300
    },
    {
      "epoch": 0.2500335886067446,
      "grad_norm": 2.362565755844116,
      "learning_rate": 3.800898815198148e-05,
      "loss": 4.4862,
      "step": 9305
    },
    {
      "epoch": 0.250167943033723,
      "grad_norm": 2.3187263011932373,
      "learning_rate": 3.8002178945934906e-05,
      "loss": 4.4179,
      "step": 9310
    },
    {
      "epoch": 0.2503022974607013,
      "grad_norm": 2.4111485481262207,
      "learning_rate": 3.799536973988833e-05,
      "loss": 4.4404,
      "step": 9315
    },
    {
      "epoch": 0.2504366518876797,
      "grad_norm": 2.3228800296783447,
      "learning_rate": 3.798856053384176e-05,
      "loss": 4.4668,
      "step": 9320
    },
    {
      "epoch": 0.25057100631465806,
      "grad_norm": 2.4349305629730225,
      "learning_rate": 3.798175132779518e-05,
      "loss": 4.3094,
      "step": 9325
    },
    {
      "epoch": 0.25070536074163646,
      "grad_norm": 2.2491023540496826,
      "learning_rate": 3.797494212174861e-05,
      "loss": 4.5083,
      "step": 9330
    },
    {
      "epoch": 0.2508397151686148,
      "grad_norm": 2.369739532470703,
      "learning_rate": 3.796813291570203e-05,
      "loss": 4.4218,
      "step": 9335
    },
    {
      "epoch": 0.25097406959559315,
      "grad_norm": 2.3309056758880615,
      "learning_rate": 3.796132370965545e-05,
      "loss": 4.4126,
      "step": 9340
    },
    {
      "epoch": 0.25110842402257155,
      "grad_norm": 2.1860311031341553,
      "learning_rate": 3.7954514503608876e-05,
      "loss": 4.4451,
      "step": 9345
    },
    {
      "epoch": 0.2512427784495499,
      "grad_norm": 2.3741095066070557,
      "learning_rate": 3.794770529756231e-05,
      "loss": 4.415,
      "step": 9350
    },
    {
      "epoch": 0.2513771328765283,
      "grad_norm": 2.463059425354004,
      "learning_rate": 3.7940896091515734e-05,
      "loss": 4.4605,
      "step": 9355
    },
    {
      "epoch": 0.25151148730350664,
      "grad_norm": 2.4005463123321533,
      "learning_rate": 3.7934086885469156e-05,
      "loss": 4.5199,
      "step": 9360
    },
    {
      "epoch": 0.25164584173048504,
      "grad_norm": 2.7171578407287598,
      "learning_rate": 3.792727767942258e-05,
      "loss": 4.4095,
      "step": 9365
    },
    {
      "epoch": 0.2517801961574634,
      "grad_norm": 2.1840732097625732,
      "learning_rate": 3.792046847337601e-05,
      "loss": 4.6307,
      "step": 9370
    },
    {
      "epoch": 0.25191455058444173,
      "grad_norm": 2.632274866104126,
      "learning_rate": 3.791365926732943e-05,
      "loss": 4.3953,
      "step": 9375
    },
    {
      "epoch": 0.25204890501142013,
      "grad_norm": 2.2114365100860596,
      "learning_rate": 3.790685006128286e-05,
      "loss": 4.4049,
      "step": 9380
    },
    {
      "epoch": 0.2521832594383985,
      "grad_norm": 2.2385265827178955,
      "learning_rate": 3.790004085523628e-05,
      "loss": 4.4441,
      "step": 9385
    },
    {
      "epoch": 0.2523176138653769,
      "grad_norm": 2.204002618789673,
      "learning_rate": 3.789323164918971e-05,
      "loss": 4.3315,
      "step": 9390
    },
    {
      "epoch": 0.2524519682923552,
      "grad_norm": 2.837421178817749,
      "learning_rate": 3.788642244314313e-05,
      "loss": 4.4636,
      "step": 9395
    },
    {
      "epoch": 0.2525863227193336,
      "grad_norm": 2.271209716796875,
      "learning_rate": 3.7879613237096554e-05,
      "loss": 4.4716,
      "step": 9400
    },
    {
      "epoch": 0.25272067714631197,
      "grad_norm": 2.3757760524749756,
      "learning_rate": 3.787280403104998e-05,
      "loss": 4.2638,
      "step": 9405
    },
    {
      "epoch": 0.25285503157329037,
      "grad_norm": 2.3093769550323486,
      "learning_rate": 3.7865994825003406e-05,
      "loss": 4.3698,
      "step": 9410
    },
    {
      "epoch": 0.2529893860002687,
      "grad_norm": 2.305525064468384,
      "learning_rate": 3.7859185618956835e-05,
      "loss": 4.546,
      "step": 9415
    },
    {
      "epoch": 0.25312374042724706,
      "grad_norm": 2.623656749725342,
      "learning_rate": 3.785237641291026e-05,
      "loss": 4.5084,
      "step": 9420
    },
    {
      "epoch": 0.25325809485422546,
      "grad_norm": 2.4141461849212646,
      "learning_rate": 3.784556720686368e-05,
      "loss": 4.5286,
      "step": 9425
    },
    {
      "epoch": 0.2533924492812038,
      "grad_norm": 2.3922817707061768,
      "learning_rate": 3.78387580008171e-05,
      "loss": 4.3568,
      "step": 9430
    },
    {
      "epoch": 0.2535268037081822,
      "grad_norm": 2.359942674636841,
      "learning_rate": 3.783194879477053e-05,
      "loss": 4.4546,
      "step": 9435
    },
    {
      "epoch": 0.25366115813516055,
      "grad_norm": 2.224031925201416,
      "learning_rate": 3.782513958872396e-05,
      "loss": 4.5532,
      "step": 9440
    },
    {
      "epoch": 0.25379551256213895,
      "grad_norm": 2.1719138622283936,
      "learning_rate": 3.781833038267738e-05,
      "loss": 4.3671,
      "step": 9445
    },
    {
      "epoch": 0.2539298669891173,
      "grad_norm": 2.376143217086792,
      "learning_rate": 3.7811521176630804e-05,
      "loss": 4.4805,
      "step": 9450
    },
    {
      "epoch": 0.25406422141609564,
      "grad_norm": 2.81508731842041,
      "learning_rate": 3.780471197058423e-05,
      "loss": 4.533,
      "step": 9455
    },
    {
      "epoch": 0.25419857584307404,
      "grad_norm": 2.3779349327087402,
      "learning_rate": 3.7797902764537656e-05,
      "loss": 4.4187,
      "step": 9460
    },
    {
      "epoch": 0.2543329302700524,
      "grad_norm": 2.4163334369659424,
      "learning_rate": 3.779109355849108e-05,
      "loss": 4.3931,
      "step": 9465
    },
    {
      "epoch": 0.2544672846970308,
      "grad_norm": 2.3255393505096436,
      "learning_rate": 3.778428435244451e-05,
      "loss": 4.3101,
      "step": 9470
    },
    {
      "epoch": 0.2546016391240091,
      "grad_norm": 2.352632999420166,
      "learning_rate": 3.7777475146397936e-05,
      "loss": 4.4209,
      "step": 9475
    },
    {
      "epoch": 0.2547359935509875,
      "grad_norm": 2.4712870121002197,
      "learning_rate": 3.777066594035136e-05,
      "loss": 4.4008,
      "step": 9480
    },
    {
      "epoch": 0.25487034797796587,
      "grad_norm": 2.5995113849639893,
      "learning_rate": 3.776385673430478e-05,
      "loss": 4.5172,
      "step": 9485
    },
    {
      "epoch": 0.2550047024049442,
      "grad_norm": 2.4173362255096436,
      "learning_rate": 3.77570475282582e-05,
      "loss": 4.3905,
      "step": 9490
    },
    {
      "epoch": 0.2551390568319226,
      "grad_norm": 2.448518991470337,
      "learning_rate": 3.775023832221163e-05,
      "loss": 4.5255,
      "step": 9495
    },
    {
      "epoch": 0.25527341125890096,
      "grad_norm": 2.4114794731140137,
      "learning_rate": 3.774342911616506e-05,
      "loss": 4.4253,
      "step": 9500
    },
    {
      "epoch": 0.25540776568587936,
      "grad_norm": 2.410508394241333,
      "learning_rate": 3.773661991011848e-05,
      "loss": 4.4523,
      "step": 9505
    },
    {
      "epoch": 0.2555421201128577,
      "grad_norm": 2.3750996589660645,
      "learning_rate": 3.7729810704071905e-05,
      "loss": 4.5379,
      "step": 9510
    },
    {
      "epoch": 0.2556764745398361,
      "grad_norm": 2.520965576171875,
      "learning_rate": 3.7723001498025334e-05,
      "loss": 4.4393,
      "step": 9515
    },
    {
      "epoch": 0.25581082896681445,
      "grad_norm": 2.3848297595977783,
      "learning_rate": 3.771619229197876e-05,
      "loss": 4.4583,
      "step": 9520
    },
    {
      "epoch": 0.25594518339379285,
      "grad_norm": 2.315286874771118,
      "learning_rate": 3.770938308593218e-05,
      "loss": 4.5366,
      "step": 9525
    },
    {
      "epoch": 0.2560795378207712,
      "grad_norm": 2.2986245155334473,
      "learning_rate": 3.770257387988561e-05,
      "loss": 4.4422,
      "step": 9530
    },
    {
      "epoch": 0.25621389224774954,
      "grad_norm": 2.204393148422241,
      "learning_rate": 3.769576467383904e-05,
      "loss": 4.4746,
      "step": 9535
    },
    {
      "epoch": 0.25634824667472794,
      "grad_norm": 2.602311134338379,
      "learning_rate": 3.768895546779246e-05,
      "loss": 4.407,
      "step": 9540
    },
    {
      "epoch": 0.2564826011017063,
      "grad_norm": 2.519596576690674,
      "learning_rate": 3.768214626174588e-05,
      "loss": 4.404,
      "step": 9545
    },
    {
      "epoch": 0.2566169555286847,
      "grad_norm": 2.361661672592163,
      "learning_rate": 3.7675337055699304e-05,
      "loss": 4.3837,
      "step": 9550
    },
    {
      "epoch": 0.25675130995566303,
      "grad_norm": 2.2274861335754395,
      "learning_rate": 3.766852784965273e-05,
      "loss": 4.4144,
      "step": 9555
    },
    {
      "epoch": 0.25688566438264143,
      "grad_norm": 2.3781468868255615,
      "learning_rate": 3.766171864360616e-05,
      "loss": 4.4335,
      "step": 9560
    },
    {
      "epoch": 0.2570200188096198,
      "grad_norm": 2.120197296142578,
      "learning_rate": 3.7654909437559584e-05,
      "loss": 4.5055,
      "step": 9565
    },
    {
      "epoch": 0.2571543732365981,
      "grad_norm": 2.295009136199951,
      "learning_rate": 3.7648100231513006e-05,
      "loss": 4.3532,
      "step": 9570
    },
    {
      "epoch": 0.2572887276635765,
      "grad_norm": 2.354764461517334,
      "learning_rate": 3.764129102546643e-05,
      "loss": 4.4306,
      "step": 9575
    },
    {
      "epoch": 0.25742308209055487,
      "grad_norm": 2.46093487739563,
      "learning_rate": 3.763448181941986e-05,
      "loss": 4.4204,
      "step": 9580
    },
    {
      "epoch": 0.25755743651753327,
      "grad_norm": 2.1906797885894775,
      "learning_rate": 3.762767261337328e-05,
      "loss": 4.5042,
      "step": 9585
    },
    {
      "epoch": 0.2576917909445116,
      "grad_norm": 2.438903570175171,
      "learning_rate": 3.762086340732671e-05,
      "loss": 4.49,
      "step": 9590
    },
    {
      "epoch": 0.25782614537149,
      "grad_norm": 2.0955286026000977,
      "learning_rate": 3.761405420128013e-05,
      "loss": 4.418,
      "step": 9595
    },
    {
      "epoch": 0.25796049979846836,
      "grad_norm": 2.417271852493286,
      "learning_rate": 3.760724499523356e-05,
      "loss": 4.4343,
      "step": 9600
    },
    {
      "epoch": 0.2580948542254467,
      "grad_norm": 2.311737298965454,
      "learning_rate": 3.760043578918698e-05,
      "loss": 4.3654,
      "step": 9605
    },
    {
      "epoch": 0.2582292086524251,
      "grad_norm": 2.292494535446167,
      "learning_rate": 3.7593626583140405e-05,
      "loss": 4.5129,
      "step": 9610
    },
    {
      "epoch": 0.25836356307940345,
      "grad_norm": 2.616590738296509,
      "learning_rate": 3.758681737709383e-05,
      "loss": 4.4328,
      "step": 9615
    },
    {
      "epoch": 0.25849791750638185,
      "grad_norm": 2.3497307300567627,
      "learning_rate": 3.7580008171047256e-05,
      "loss": 4.5104,
      "step": 9620
    },
    {
      "epoch": 0.2586322719333602,
      "grad_norm": 2.2566702365875244,
      "learning_rate": 3.7573198965000685e-05,
      "loss": 4.4515,
      "step": 9625
    },
    {
      "epoch": 0.2587666263603386,
      "grad_norm": 2.3225743770599365,
      "learning_rate": 3.756638975895411e-05,
      "loss": 4.4815,
      "step": 9630
    },
    {
      "epoch": 0.25890098078731694,
      "grad_norm": 1.9311445951461792,
      "learning_rate": 3.755958055290753e-05,
      "loss": 4.4402,
      "step": 9635
    },
    {
      "epoch": 0.25903533521429534,
      "grad_norm": 2.3557026386260986,
      "learning_rate": 3.755277134686096e-05,
      "loss": 4.4655,
      "step": 9640
    },
    {
      "epoch": 0.2591696896412737,
      "grad_norm": 2.4471280574798584,
      "learning_rate": 3.754596214081438e-05,
      "loss": 4.4554,
      "step": 9645
    },
    {
      "epoch": 0.259304044068252,
      "grad_norm": 2.389927864074707,
      "learning_rate": 3.753915293476781e-05,
      "loss": 4.4767,
      "step": 9650
    },
    {
      "epoch": 0.2594383984952304,
      "grad_norm": 2.284169912338257,
      "learning_rate": 3.753234372872123e-05,
      "loss": 4.3781,
      "step": 9655
    },
    {
      "epoch": 0.25957275292220877,
      "grad_norm": 2.5234127044677734,
      "learning_rate": 3.752553452267466e-05,
      "loss": 4.4721,
      "step": 9660
    },
    {
      "epoch": 0.25970710734918717,
      "grad_norm": 2.4971258640289307,
      "learning_rate": 3.7518725316628084e-05,
      "loss": 4.4469,
      "step": 9665
    },
    {
      "epoch": 0.2598414617761655,
      "grad_norm": 2.0405125617980957,
      "learning_rate": 3.7511916110581506e-05,
      "loss": 4.3943,
      "step": 9670
    },
    {
      "epoch": 0.2599758162031439,
      "grad_norm": 2.242678165435791,
      "learning_rate": 3.750510690453493e-05,
      "loss": 4.4172,
      "step": 9675
    },
    {
      "epoch": 0.26011017063012226,
      "grad_norm": 2.285696506500244,
      "learning_rate": 3.749829769848836e-05,
      "loss": 4.3642,
      "step": 9680
    },
    {
      "epoch": 0.2602445250571006,
      "grad_norm": 2.272531509399414,
      "learning_rate": 3.7491488492441786e-05,
      "loss": 4.4344,
      "step": 9685
    },
    {
      "epoch": 0.260378879484079,
      "grad_norm": 2.4874634742736816,
      "learning_rate": 3.748467928639521e-05,
      "loss": 4.5439,
      "step": 9690
    },
    {
      "epoch": 0.26051323391105735,
      "grad_norm": 2.549420118331909,
      "learning_rate": 3.747787008034863e-05,
      "loss": 4.5267,
      "step": 9695
    },
    {
      "epoch": 0.26064758833803575,
      "grad_norm": 2.2837493419647217,
      "learning_rate": 3.747106087430206e-05,
      "loss": 4.319,
      "step": 9700
    },
    {
      "epoch": 0.2607819427650141,
      "grad_norm": 2.5838429927825928,
      "learning_rate": 3.746425166825548e-05,
      "loss": 4.4437,
      "step": 9705
    },
    {
      "epoch": 0.2609162971919925,
      "grad_norm": 2.4651379585266113,
      "learning_rate": 3.745744246220891e-05,
      "loss": 4.4631,
      "step": 9710
    },
    {
      "epoch": 0.26105065161897084,
      "grad_norm": 2.6954050064086914,
      "learning_rate": 3.7450633256162334e-05,
      "loss": 4.5162,
      "step": 9715
    },
    {
      "epoch": 0.26118500604594924,
      "grad_norm": 2.3330328464508057,
      "learning_rate": 3.7443824050115756e-05,
      "loss": 4.4874,
      "step": 9720
    },
    {
      "epoch": 0.2613193604729276,
      "grad_norm": 2.405008554458618,
      "learning_rate": 3.7437014844069185e-05,
      "loss": 4.483,
      "step": 9725
    },
    {
      "epoch": 0.26145371489990593,
      "grad_norm": 2.2013437747955322,
      "learning_rate": 3.743020563802261e-05,
      "loss": 4.4779,
      "step": 9730
    },
    {
      "epoch": 0.26158806932688433,
      "grad_norm": 2.2661209106445312,
      "learning_rate": 3.742339643197603e-05,
      "loss": 4.4922,
      "step": 9735
    },
    {
      "epoch": 0.2617224237538627,
      "grad_norm": 2.4545578956604004,
      "learning_rate": 3.741658722592946e-05,
      "loss": 4.5286,
      "step": 9740
    },
    {
      "epoch": 0.2618567781808411,
      "grad_norm": 2.7013602256774902,
      "learning_rate": 3.740977801988289e-05,
      "loss": 4.491,
      "step": 9745
    },
    {
      "epoch": 0.2619911326078194,
      "grad_norm": 2.254404067993164,
      "learning_rate": 3.740296881383631e-05,
      "loss": 4.4938,
      "step": 9750
    },
    {
      "epoch": 0.2621254870347978,
      "grad_norm": 2.3829667568206787,
      "learning_rate": 3.739615960778973e-05,
      "loss": 4.4186,
      "step": 9755
    },
    {
      "epoch": 0.26225984146177617,
      "grad_norm": 2.2982051372528076,
      "learning_rate": 3.7389350401743154e-05,
      "loss": 4.5197,
      "step": 9760
    },
    {
      "epoch": 0.2623941958887545,
      "grad_norm": 2.291553497314453,
      "learning_rate": 3.738254119569658e-05,
      "loss": 4.3647,
      "step": 9765
    },
    {
      "epoch": 0.2625285503157329,
      "grad_norm": 2.597904682159424,
      "learning_rate": 3.737573198965001e-05,
      "loss": 4.4534,
      "step": 9770
    },
    {
      "epoch": 0.26266290474271126,
      "grad_norm": 2.16689133644104,
      "learning_rate": 3.7368922783603435e-05,
      "loss": 4.4744,
      "step": 9775
    },
    {
      "epoch": 0.26279725916968966,
      "grad_norm": 2.276440382003784,
      "learning_rate": 3.736211357755686e-05,
      "loss": 4.47,
      "step": 9780
    },
    {
      "epoch": 0.262931613596668,
      "grad_norm": 2.2739968299865723,
      "learning_rate": 3.7355304371510286e-05,
      "loss": 4.4552,
      "step": 9785
    },
    {
      "epoch": 0.2630659680236464,
      "grad_norm": 2.153620481491089,
      "learning_rate": 3.734849516546371e-05,
      "loss": 4.6209,
      "step": 9790
    },
    {
      "epoch": 0.26320032245062475,
      "grad_norm": 2.309554100036621,
      "learning_rate": 3.734168595941713e-05,
      "loss": 4.4911,
      "step": 9795
    },
    {
      "epoch": 0.2633346768776031,
      "grad_norm": 2.1431453227996826,
      "learning_rate": 3.733487675337056e-05,
      "loss": 4.38,
      "step": 9800
    },
    {
      "epoch": 0.2634690313045815,
      "grad_norm": 2.358398914337158,
      "learning_rate": 3.732806754732399e-05,
      "loss": 4.5943,
      "step": 9805
    },
    {
      "epoch": 0.26360338573155984,
      "grad_norm": 2.278433322906494,
      "learning_rate": 3.732125834127741e-05,
      "loss": 4.4525,
      "step": 9810
    },
    {
      "epoch": 0.26373774015853824,
      "grad_norm": 2.511446475982666,
      "learning_rate": 3.731444913523083e-05,
      "loss": 4.3777,
      "step": 9815
    },
    {
      "epoch": 0.2638720945855166,
      "grad_norm": 2.6643190383911133,
      "learning_rate": 3.7307639929184255e-05,
      "loss": 4.644,
      "step": 9820
    },
    {
      "epoch": 0.264006449012495,
      "grad_norm": 2.4004883766174316,
      "learning_rate": 3.7300830723137684e-05,
      "loss": 4.4165,
      "step": 9825
    },
    {
      "epoch": 0.2641408034394733,
      "grad_norm": 2.514808416366577,
      "learning_rate": 3.729402151709111e-05,
      "loss": 4.436,
      "step": 9830
    },
    {
      "epoch": 0.2642751578664517,
      "grad_norm": 2.4849557876586914,
      "learning_rate": 3.7287212311044536e-05,
      "loss": 4.5066,
      "step": 9835
    },
    {
      "epoch": 0.26440951229343007,
      "grad_norm": 2.407912492752075,
      "learning_rate": 3.728040310499796e-05,
      "loss": 4.4581,
      "step": 9840
    },
    {
      "epoch": 0.2645438667204084,
      "grad_norm": 2.3319485187530518,
      "learning_rate": 3.727359389895139e-05,
      "loss": 4.5831,
      "step": 9845
    },
    {
      "epoch": 0.2646782211473868,
      "grad_norm": 2.386629819869995,
      "learning_rate": 3.726678469290481e-05,
      "loss": 4.4346,
      "step": 9850
    },
    {
      "epoch": 0.26481257557436516,
      "grad_norm": 2.3634583950042725,
      "learning_rate": 3.725997548685823e-05,
      "loss": 4.3664,
      "step": 9855
    },
    {
      "epoch": 0.26494693000134356,
      "grad_norm": 2.2978546619415283,
      "learning_rate": 3.725316628081166e-05,
      "loss": 4.3566,
      "step": 9860
    },
    {
      "epoch": 0.2650812844283219,
      "grad_norm": 2.196061134338379,
      "learning_rate": 3.724635707476508e-05,
      "loss": 4.2927,
      "step": 9865
    },
    {
      "epoch": 0.2652156388553003,
      "grad_norm": 2.523641347885132,
      "learning_rate": 3.723954786871851e-05,
      "loss": 4.3428,
      "step": 9870
    },
    {
      "epoch": 0.26534999328227865,
      "grad_norm": 2.3317599296569824,
      "learning_rate": 3.7232738662671934e-05,
      "loss": 4.4409,
      "step": 9875
    },
    {
      "epoch": 0.265484347709257,
      "grad_norm": 2.573763132095337,
      "learning_rate": 3.7225929456625357e-05,
      "loss": 4.2988,
      "step": 9880
    },
    {
      "epoch": 0.2656187021362354,
      "grad_norm": 2.4619436264038086,
      "learning_rate": 3.721912025057878e-05,
      "loss": 4.284,
      "step": 9885
    },
    {
      "epoch": 0.26575305656321374,
      "grad_norm": 2.335601568222046,
      "learning_rate": 3.721231104453221e-05,
      "loss": 4.5262,
      "step": 9890
    },
    {
      "epoch": 0.26588741099019214,
      "grad_norm": 2.5007078647613525,
      "learning_rate": 3.720550183848564e-05,
      "loss": 4.4662,
      "step": 9895
    },
    {
      "epoch": 0.2660217654171705,
      "grad_norm": 2.466279983520508,
      "learning_rate": 3.719869263243906e-05,
      "loss": 4.4318,
      "step": 9900
    },
    {
      "epoch": 0.2661561198441489,
      "grad_norm": 2.2760276794433594,
      "learning_rate": 3.719188342639248e-05,
      "loss": 4.4646,
      "step": 9905
    },
    {
      "epoch": 0.26629047427112723,
      "grad_norm": 2.3255913257598877,
      "learning_rate": 3.718507422034591e-05,
      "loss": 4.4533,
      "step": 9910
    },
    {
      "epoch": 0.2664248286981056,
      "grad_norm": 2.167842149734497,
      "learning_rate": 3.717826501429933e-05,
      "loss": 4.3864,
      "step": 9915
    },
    {
      "epoch": 0.266559183125084,
      "grad_norm": 2.4722235202789307,
      "learning_rate": 3.717145580825276e-05,
      "loss": 4.3986,
      "step": 9920
    },
    {
      "epoch": 0.2666935375520623,
      "grad_norm": 2.270608425140381,
      "learning_rate": 3.7164646602206184e-05,
      "loss": 4.4628,
      "step": 9925
    },
    {
      "epoch": 0.2668278919790407,
      "grad_norm": 2.0205249786376953,
      "learning_rate": 3.715783739615961e-05,
      "loss": 4.3178,
      "step": 9930
    },
    {
      "epoch": 0.26696224640601907,
      "grad_norm": 2.2083804607391357,
      "learning_rate": 3.7151028190113035e-05,
      "loss": 4.3497,
      "step": 9935
    },
    {
      "epoch": 0.26709660083299747,
      "grad_norm": 2.111360549926758,
      "learning_rate": 3.714421898406646e-05,
      "loss": 4.5364,
      "step": 9940
    },
    {
      "epoch": 0.2672309552599758,
      "grad_norm": 2.542346954345703,
      "learning_rate": 3.713740977801988e-05,
      "loss": 4.5795,
      "step": 9945
    },
    {
      "epoch": 0.2673653096869542,
      "grad_norm": 2.4858274459838867,
      "learning_rate": 3.713060057197331e-05,
      "loss": 4.4883,
      "step": 9950
    },
    {
      "epoch": 0.26749966411393256,
      "grad_norm": 2.094383955001831,
      "learning_rate": 3.712379136592674e-05,
      "loss": 4.4766,
      "step": 9955
    },
    {
      "epoch": 0.2676340185409109,
      "grad_norm": 2.312220811843872,
      "learning_rate": 3.711698215988016e-05,
      "loss": 4.4426,
      "step": 9960
    },
    {
      "epoch": 0.2677683729678893,
      "grad_norm": 2.50146746635437,
      "learning_rate": 3.711017295383358e-05,
      "loss": 4.4527,
      "step": 9965
    },
    {
      "epoch": 0.26790272739486765,
      "grad_norm": 2.3963401317596436,
      "learning_rate": 3.710336374778701e-05,
      "loss": 4.4038,
      "step": 9970
    },
    {
      "epoch": 0.26803708182184605,
      "grad_norm": 2.8256218433380127,
      "learning_rate": 3.7096554541740434e-05,
      "loss": 4.4912,
      "step": 9975
    },
    {
      "epoch": 0.2681714362488244,
      "grad_norm": 2.066718101501465,
      "learning_rate": 3.708974533569386e-05,
      "loss": 4.3603,
      "step": 9980
    },
    {
      "epoch": 0.2683057906758028,
      "grad_norm": 2.409860610961914,
      "learning_rate": 3.7082936129647285e-05,
      "loss": 4.5533,
      "step": 9985
    },
    {
      "epoch": 0.26844014510278114,
      "grad_norm": 2.360959768295288,
      "learning_rate": 3.7076126923600714e-05,
      "loss": 4.498,
      "step": 9990
    },
    {
      "epoch": 0.2685744995297595,
      "grad_norm": 2.37809681892395,
      "learning_rate": 3.7069317717554137e-05,
      "loss": 4.4151,
      "step": 9995
    },
    {
      "epoch": 0.2687088539567379,
      "grad_norm": 2.400447130203247,
      "learning_rate": 3.706250851150756e-05,
      "loss": 4.3155,
      "step": 10000
    },
    {
      "epoch": 0.2688432083837162,
      "grad_norm": 2.4247653484344482,
      "learning_rate": 3.705569930546098e-05,
      "loss": 4.3683,
      "step": 10005
    },
    {
      "epoch": 0.2689775628106946,
      "grad_norm": 2.2654831409454346,
      "learning_rate": 3.704889009941441e-05,
      "loss": 4.4586,
      "step": 10010
    },
    {
      "epoch": 0.26911191723767297,
      "grad_norm": 2.1549532413482666,
      "learning_rate": 3.704208089336784e-05,
      "loss": 4.4431,
      "step": 10015
    },
    {
      "epoch": 0.26924627166465137,
      "grad_norm": 2.375157117843628,
      "learning_rate": 3.703527168732126e-05,
      "loss": 4.4134,
      "step": 10020
    },
    {
      "epoch": 0.2693806260916297,
      "grad_norm": 2.561429977416992,
      "learning_rate": 3.7028462481274684e-05,
      "loss": 4.4236,
      "step": 10025
    },
    {
      "epoch": 0.26951498051860806,
      "grad_norm": 2.3778979778289795,
      "learning_rate": 3.7021653275228106e-05,
      "loss": 4.5557,
      "step": 10030
    },
    {
      "epoch": 0.26964933494558646,
      "grad_norm": 2.3131585121154785,
      "learning_rate": 3.7014844069181535e-05,
      "loss": 4.5653,
      "step": 10035
    },
    {
      "epoch": 0.2697836893725648,
      "grad_norm": 2.4765501022338867,
      "learning_rate": 3.700803486313496e-05,
      "loss": 4.4802,
      "step": 10040
    },
    {
      "epoch": 0.2699180437995432,
      "grad_norm": 2.3807547092437744,
      "learning_rate": 3.7001225657088386e-05,
      "loss": 4.4466,
      "step": 10045
    },
    {
      "epoch": 0.27005239822652155,
      "grad_norm": 2.347968816757202,
      "learning_rate": 3.699441645104181e-05,
      "loss": 4.3955,
      "step": 10050
    },
    {
      "epoch": 0.27018675265349995,
      "grad_norm": 2.708430528640747,
      "learning_rate": 3.698760724499524e-05,
      "loss": 4.5,
      "step": 10055
    },
    {
      "epoch": 0.2703211070804783,
      "grad_norm": 2.138028621673584,
      "learning_rate": 3.698079803894866e-05,
      "loss": 4.2832,
      "step": 10060
    },
    {
      "epoch": 0.2704554615074567,
      "grad_norm": 2.2226572036743164,
      "learning_rate": 3.697398883290208e-05,
      "loss": 4.4274,
      "step": 10065
    },
    {
      "epoch": 0.27058981593443504,
      "grad_norm": 2.441422939300537,
      "learning_rate": 3.696717962685551e-05,
      "loss": 4.5429,
      "step": 10070
    },
    {
      "epoch": 0.2707241703614134,
      "grad_norm": 2.2961292266845703,
      "learning_rate": 3.696037042080894e-05,
      "loss": 4.4615,
      "step": 10075
    },
    {
      "epoch": 0.2708585247883918,
      "grad_norm": 2.362414598464966,
      "learning_rate": 3.695356121476236e-05,
      "loss": 4.3932,
      "step": 10080
    },
    {
      "epoch": 0.27099287921537013,
      "grad_norm": 2.345384120941162,
      "learning_rate": 3.6946752008715785e-05,
      "loss": 4.4961,
      "step": 10085
    },
    {
      "epoch": 0.27112723364234853,
      "grad_norm": 2.411900758743286,
      "learning_rate": 3.693994280266921e-05,
      "loss": 4.4627,
      "step": 10090
    },
    {
      "epoch": 0.2712615880693269,
      "grad_norm": 2.3504819869995117,
      "learning_rate": 3.6933133596622636e-05,
      "loss": 4.3472,
      "step": 10095
    },
    {
      "epoch": 0.2713959424963053,
      "grad_norm": 2.5856263637542725,
      "learning_rate": 3.692632439057606e-05,
      "loss": 4.3207,
      "step": 10100
    },
    {
      "epoch": 0.2715302969232836,
      "grad_norm": 2.3424105644226074,
      "learning_rate": 3.691951518452949e-05,
      "loss": 4.5254,
      "step": 10105
    },
    {
      "epoch": 0.27166465135026197,
      "grad_norm": 2.2347137928009033,
      "learning_rate": 3.691270597848291e-05,
      "loss": 4.2565,
      "step": 10110
    },
    {
      "epoch": 0.27179900577724037,
      "grad_norm": 2.503782033920288,
      "learning_rate": 3.690589677243634e-05,
      "loss": 4.5038,
      "step": 10115
    },
    {
      "epoch": 0.2719333602042187,
      "grad_norm": 2.5684449672698975,
      "learning_rate": 3.689908756638976e-05,
      "loss": 4.3585,
      "step": 10120
    },
    {
      "epoch": 0.2720677146311971,
      "grad_norm": 2.4398772716522217,
      "learning_rate": 3.689227836034318e-05,
      "loss": 4.4603,
      "step": 10125
    },
    {
      "epoch": 0.27220206905817546,
      "grad_norm": 2.327562093734741,
      "learning_rate": 3.688546915429661e-05,
      "loss": 4.3809,
      "step": 10130
    },
    {
      "epoch": 0.27233642348515386,
      "grad_norm": 2.2435462474823,
      "learning_rate": 3.687865994825004e-05,
      "loss": 4.509,
      "step": 10135
    },
    {
      "epoch": 0.2724707779121322,
      "grad_norm": 2.418958902359009,
      "learning_rate": 3.6871850742203464e-05,
      "loss": 4.4447,
      "step": 10140
    },
    {
      "epoch": 0.2726051323391106,
      "grad_norm": 2.468705177307129,
      "learning_rate": 3.6865041536156886e-05,
      "loss": 4.5599,
      "step": 10145
    },
    {
      "epoch": 0.27273948676608895,
      "grad_norm": 2.3407859802246094,
      "learning_rate": 3.685823233011031e-05,
      "loss": 4.3735,
      "step": 10150
    },
    {
      "epoch": 0.2728738411930673,
      "grad_norm": 2.11291241645813,
      "learning_rate": 3.685142312406374e-05,
      "loss": 4.3764,
      "step": 10155
    },
    {
      "epoch": 0.2730081956200457,
      "grad_norm": 2.2554802894592285,
      "learning_rate": 3.684461391801716e-05,
      "loss": 4.5011,
      "step": 10160
    },
    {
      "epoch": 0.27314255004702404,
      "grad_norm": 2.3598947525024414,
      "learning_rate": 3.683780471197059e-05,
      "loss": 4.3916,
      "step": 10165
    },
    {
      "epoch": 0.27327690447400244,
      "grad_norm": 2.562696695327759,
      "learning_rate": 3.683099550592401e-05,
      "loss": 4.49,
      "step": 10170
    },
    {
      "epoch": 0.2734112589009808,
      "grad_norm": 2.5074269771575928,
      "learning_rate": 3.682418629987743e-05,
      "loss": 4.4121,
      "step": 10175
    },
    {
      "epoch": 0.2735456133279592,
      "grad_norm": 2.2123053073883057,
      "learning_rate": 3.681737709383086e-05,
      "loss": 4.5272,
      "step": 10180
    },
    {
      "epoch": 0.2736799677549375,
      "grad_norm": 2.396061658859253,
      "learning_rate": 3.6810567887784284e-05,
      "loss": 4.477,
      "step": 10185
    },
    {
      "epoch": 0.27381432218191587,
      "grad_norm": 2.271411418914795,
      "learning_rate": 3.680375868173771e-05,
      "loss": 4.4562,
      "step": 10190
    },
    {
      "epoch": 0.27394867660889427,
      "grad_norm": 2.235460042953491,
      "learning_rate": 3.6796949475691136e-05,
      "loss": 4.3722,
      "step": 10195
    },
    {
      "epoch": 0.2740830310358726,
      "grad_norm": 2.320960760116577,
      "learning_rate": 3.6790140269644565e-05,
      "loss": 4.4411,
      "step": 10200
    },
    {
      "epoch": 0.274217385462851,
      "grad_norm": 2.3031654357910156,
      "learning_rate": 3.678333106359799e-05,
      "loss": 4.3898,
      "step": 10205
    },
    {
      "epoch": 0.27435173988982936,
      "grad_norm": 2.1843700408935547,
      "learning_rate": 3.677652185755141e-05,
      "loss": 4.4716,
      "step": 10210
    },
    {
      "epoch": 0.27448609431680776,
      "grad_norm": 2.8004093170166016,
      "learning_rate": 3.676971265150483e-05,
      "loss": 4.4639,
      "step": 10215
    },
    {
      "epoch": 0.2746204487437861,
      "grad_norm": 2.3705034255981445,
      "learning_rate": 3.676290344545826e-05,
      "loss": 4.4949,
      "step": 10220
    },
    {
      "epoch": 0.27475480317076445,
      "grad_norm": 2.4161531925201416,
      "learning_rate": 3.675609423941169e-05,
      "loss": 4.4946,
      "step": 10225
    },
    {
      "epoch": 0.27488915759774285,
      "grad_norm": 2.491004228591919,
      "learning_rate": 3.674928503336511e-05,
      "loss": 4.5377,
      "step": 10230
    },
    {
      "epoch": 0.2750235120247212,
      "grad_norm": 2.0976486206054688,
      "learning_rate": 3.6742475827318534e-05,
      "loss": 4.4516,
      "step": 10235
    },
    {
      "epoch": 0.2751578664516996,
      "grad_norm": 2.4337034225463867,
      "learning_rate": 3.673566662127196e-05,
      "loss": 4.6469,
      "step": 10240
    },
    {
      "epoch": 0.27529222087867794,
      "grad_norm": 2.5322659015655518,
      "learning_rate": 3.6728857415225385e-05,
      "loss": 4.3459,
      "step": 10245
    },
    {
      "epoch": 0.27542657530565634,
      "grad_norm": 2.3156890869140625,
      "learning_rate": 3.672204820917881e-05,
      "loss": 4.4699,
      "step": 10250
    },
    {
      "epoch": 0.2755609297326347,
      "grad_norm": 2.657050609588623,
      "learning_rate": 3.671523900313224e-05,
      "loss": 4.4103,
      "step": 10255
    },
    {
      "epoch": 0.2756952841596131,
      "grad_norm": 2.39984130859375,
      "learning_rate": 3.6708429797085666e-05,
      "loss": 4.3438,
      "step": 10260
    },
    {
      "epoch": 0.27582963858659143,
      "grad_norm": 2.0272397994995117,
      "learning_rate": 3.670162059103909e-05,
      "loss": 4.3583,
      "step": 10265
    },
    {
      "epoch": 0.2759639930135698,
      "grad_norm": 2.1731221675872803,
      "learning_rate": 3.669481138499251e-05,
      "loss": 4.4528,
      "step": 10270
    },
    {
      "epoch": 0.2760983474405482,
      "grad_norm": 2.176909923553467,
      "learning_rate": 3.668800217894593e-05,
      "loss": 4.531,
      "step": 10275
    },
    {
      "epoch": 0.2762327018675265,
      "grad_norm": 2.547961473464966,
      "learning_rate": 3.668119297289936e-05,
      "loss": 4.346,
      "step": 10280
    },
    {
      "epoch": 0.2763670562945049,
      "grad_norm": 2.530648708343506,
      "learning_rate": 3.667438376685279e-05,
      "loss": 4.4228,
      "step": 10285
    },
    {
      "epoch": 0.27650141072148327,
      "grad_norm": 2.6035149097442627,
      "learning_rate": 3.666757456080621e-05,
      "loss": 4.3907,
      "step": 10290
    },
    {
      "epoch": 0.27663576514846167,
      "grad_norm": 2.4263744354248047,
      "learning_rate": 3.6660765354759635e-05,
      "loss": 4.403,
      "step": 10295
    },
    {
      "epoch": 0.27677011957544,
      "grad_norm": 2.1926891803741455,
      "learning_rate": 3.6653956148713064e-05,
      "loss": 4.5117,
      "step": 10300
    },
    {
      "epoch": 0.27690447400241835,
      "grad_norm": 2.2880146503448486,
      "learning_rate": 3.6647146942666487e-05,
      "loss": 4.4092,
      "step": 10305
    },
    {
      "epoch": 0.27703882842939676,
      "grad_norm": 2.328742742538452,
      "learning_rate": 3.664033773661991e-05,
      "loss": 4.4167,
      "step": 10310
    },
    {
      "epoch": 0.2771731828563751,
      "grad_norm": 2.5610458850860596,
      "learning_rate": 3.663352853057334e-05,
      "loss": 4.3274,
      "step": 10315
    },
    {
      "epoch": 0.2773075372833535,
      "grad_norm": 2.256901741027832,
      "learning_rate": 3.662671932452676e-05,
      "loss": 4.4294,
      "step": 10320
    },
    {
      "epoch": 0.27744189171033184,
      "grad_norm": 2.2029168605804443,
      "learning_rate": 3.661991011848019e-05,
      "loss": 4.3384,
      "step": 10325
    },
    {
      "epoch": 0.27757624613731025,
      "grad_norm": 2.049301862716675,
      "learning_rate": 3.661310091243361e-05,
      "loss": 4.4614,
      "step": 10330
    },
    {
      "epoch": 0.2777106005642886,
      "grad_norm": 2.651298999786377,
      "learning_rate": 3.6606291706387034e-05,
      "loss": 4.5494,
      "step": 10335
    },
    {
      "epoch": 0.27784495499126693,
      "grad_norm": 2.457087993621826,
      "learning_rate": 3.659948250034046e-05,
      "loss": 4.4691,
      "step": 10340
    },
    {
      "epoch": 0.27797930941824534,
      "grad_norm": 2.158074378967285,
      "learning_rate": 3.659267329429389e-05,
      "loss": 4.5279,
      "step": 10345
    },
    {
      "epoch": 0.2781136638452237,
      "grad_norm": 2.334291458129883,
      "learning_rate": 3.6585864088247314e-05,
      "loss": 4.4637,
      "step": 10350
    },
    {
      "epoch": 0.2782480182722021,
      "grad_norm": 2.2675631046295166,
      "learning_rate": 3.6579054882200736e-05,
      "loss": 4.4725,
      "step": 10355
    },
    {
      "epoch": 0.2783823726991804,
      "grad_norm": 3.0601279735565186,
      "learning_rate": 3.657224567615416e-05,
      "loss": 4.265,
      "step": 10360
    },
    {
      "epoch": 0.2785167271261588,
      "grad_norm": 2.5093634128570557,
      "learning_rate": 3.656543647010759e-05,
      "loss": 4.3841,
      "step": 10365
    },
    {
      "epoch": 0.27865108155313717,
      "grad_norm": 2.434187650680542,
      "learning_rate": 3.655862726406101e-05,
      "loss": 4.4422,
      "step": 10370
    },
    {
      "epoch": 0.27878543598011557,
      "grad_norm": 3.026028871536255,
      "learning_rate": 3.655181805801444e-05,
      "loss": 4.3557,
      "step": 10375
    },
    {
      "epoch": 0.2789197904070939,
      "grad_norm": 2.350429058074951,
      "learning_rate": 3.654500885196786e-05,
      "loss": 4.401,
      "step": 10380
    },
    {
      "epoch": 0.27905414483407226,
      "grad_norm": 2.2864737510681152,
      "learning_rate": 3.653819964592129e-05,
      "loss": 4.5293,
      "step": 10385
    },
    {
      "epoch": 0.27918849926105066,
      "grad_norm": 2.4145596027374268,
      "learning_rate": 3.653139043987471e-05,
      "loss": 4.4275,
      "step": 10390
    },
    {
      "epoch": 0.279322853688029,
      "grad_norm": 2.5522289276123047,
      "learning_rate": 3.6524581233828135e-05,
      "loss": 4.5022,
      "step": 10395
    },
    {
      "epoch": 0.2794572081150074,
      "grad_norm": 2.3700194358825684,
      "learning_rate": 3.651777202778156e-05,
      "loss": 4.3757,
      "step": 10400
    },
    {
      "epoch": 0.27959156254198575,
      "grad_norm": 2.554453134536743,
      "learning_rate": 3.651096282173499e-05,
      "loss": 4.3841,
      "step": 10405
    },
    {
      "epoch": 0.27972591696896415,
      "grad_norm": 2.4297685623168945,
      "learning_rate": 3.6504153615688415e-05,
      "loss": 4.4564,
      "step": 10410
    },
    {
      "epoch": 0.2798602713959425,
      "grad_norm": 2.2345170974731445,
      "learning_rate": 3.649734440964184e-05,
      "loss": 4.5281,
      "step": 10415
    },
    {
      "epoch": 0.27999462582292084,
      "grad_norm": 2.3470749855041504,
      "learning_rate": 3.649053520359526e-05,
      "loss": 4.4515,
      "step": 10420
    },
    {
      "epoch": 0.28012898024989924,
      "grad_norm": 2.357478618621826,
      "learning_rate": 3.648372599754869e-05,
      "loss": 4.4801,
      "step": 10425
    },
    {
      "epoch": 0.2802633346768776,
      "grad_norm": 2.428985118865967,
      "learning_rate": 3.647691679150211e-05,
      "loss": 4.4497,
      "step": 10430
    },
    {
      "epoch": 0.280397689103856,
      "grad_norm": 2.398602247238159,
      "learning_rate": 3.647010758545554e-05,
      "loss": 4.3434,
      "step": 10435
    },
    {
      "epoch": 0.28053204353083433,
      "grad_norm": 2.3062753677368164,
      "learning_rate": 3.646329837940896e-05,
      "loss": 4.4495,
      "step": 10440
    },
    {
      "epoch": 0.28066639795781273,
      "grad_norm": 2.2553212642669678,
      "learning_rate": 3.645648917336239e-05,
      "loss": 4.4459,
      "step": 10445
    },
    {
      "epoch": 0.2808007523847911,
      "grad_norm": 2.2880945205688477,
      "learning_rate": 3.6449679967315814e-05,
      "loss": 4.4541,
      "step": 10450
    },
    {
      "epoch": 0.2809351068117695,
      "grad_norm": 2.3593807220458984,
      "learning_rate": 3.6442870761269236e-05,
      "loss": 4.4576,
      "step": 10455
    },
    {
      "epoch": 0.2810694612387478,
      "grad_norm": 2.469602108001709,
      "learning_rate": 3.643606155522266e-05,
      "loss": 4.4915,
      "step": 10460
    },
    {
      "epoch": 0.28120381566572616,
      "grad_norm": 2.403080701828003,
      "learning_rate": 3.642925234917609e-05,
      "loss": 4.5021,
      "step": 10465
    },
    {
      "epoch": 0.28133817009270456,
      "grad_norm": 2.3165926933288574,
      "learning_rate": 3.6422443143129516e-05,
      "loss": 4.4755,
      "step": 10470
    },
    {
      "epoch": 0.2814725245196829,
      "grad_norm": 2.2108871936798096,
      "learning_rate": 3.641563393708294e-05,
      "loss": 4.4574,
      "step": 10475
    },
    {
      "epoch": 0.2816068789466613,
      "grad_norm": 2.4958202838897705,
      "learning_rate": 3.640882473103636e-05,
      "loss": 4.565,
      "step": 10480
    },
    {
      "epoch": 0.28174123337363965,
      "grad_norm": 2.2051994800567627,
      "learning_rate": 3.640201552498978e-05,
      "loss": 4.3889,
      "step": 10485
    },
    {
      "epoch": 0.28187558780061805,
      "grad_norm": 2.220006227493286,
      "learning_rate": 3.639520631894321e-05,
      "loss": 4.3849,
      "step": 10490
    },
    {
      "epoch": 0.2820099422275964,
      "grad_norm": 2.4894070625305176,
      "learning_rate": 3.638839711289664e-05,
      "loss": 4.4398,
      "step": 10495
    },
    {
      "epoch": 0.28214429665457474,
      "grad_norm": 2.513720750808716,
      "learning_rate": 3.6381587906850064e-05,
      "loss": 4.4799,
      "step": 10500
    },
    {
      "epoch": 0.28227865108155314,
      "grad_norm": 2.4000284671783447,
      "learning_rate": 3.6374778700803486e-05,
      "loss": 4.5191,
      "step": 10505
    },
    {
      "epoch": 0.2824130055085315,
      "grad_norm": 2.2847397327423096,
      "learning_rate": 3.6367969494756915e-05,
      "loss": 4.4295,
      "step": 10510
    },
    {
      "epoch": 0.2825473599355099,
      "grad_norm": 2.3887035846710205,
      "learning_rate": 3.636116028871034e-05,
      "loss": 4.4475,
      "step": 10515
    },
    {
      "epoch": 0.28268171436248823,
      "grad_norm": 2.2457971572875977,
      "learning_rate": 3.635435108266376e-05,
      "loss": 4.4302,
      "step": 10520
    },
    {
      "epoch": 0.28281606878946663,
      "grad_norm": 2.4918901920318604,
      "learning_rate": 3.634754187661719e-05,
      "loss": 4.4958,
      "step": 10525
    },
    {
      "epoch": 0.282950423216445,
      "grad_norm": 2.3126444816589355,
      "learning_rate": 3.634073267057062e-05,
      "loss": 4.4957,
      "step": 10530
    },
    {
      "epoch": 0.2830847776434233,
      "grad_norm": 2.216681957244873,
      "learning_rate": 3.633392346452404e-05,
      "loss": 4.484,
      "step": 10535
    },
    {
      "epoch": 0.2832191320704017,
      "grad_norm": 2.374052047729492,
      "learning_rate": 3.632711425847746e-05,
      "loss": 4.3977,
      "step": 10540
    },
    {
      "epoch": 0.28335348649738007,
      "grad_norm": 2.356959342956543,
      "learning_rate": 3.6320305052430884e-05,
      "loss": 4.4048,
      "step": 10545
    },
    {
      "epoch": 0.28348784092435847,
      "grad_norm": 2.2532029151916504,
      "learning_rate": 3.631349584638431e-05,
      "loss": 4.3279,
      "step": 10550
    },
    {
      "epoch": 0.2836221953513368,
      "grad_norm": 2.3313169479370117,
      "learning_rate": 3.630668664033774e-05,
      "loss": 4.4452,
      "step": 10555
    },
    {
      "epoch": 0.2837565497783152,
      "grad_norm": 2.250466823577881,
      "learning_rate": 3.6299877434291165e-05,
      "loss": 4.4416,
      "step": 10560
    },
    {
      "epoch": 0.28389090420529356,
      "grad_norm": 2.3865766525268555,
      "learning_rate": 3.629306822824459e-05,
      "loss": 4.357,
      "step": 10565
    },
    {
      "epoch": 0.28402525863227196,
      "grad_norm": 2.56585955619812,
      "learning_rate": 3.6286259022198016e-05,
      "loss": 4.4071,
      "step": 10570
    },
    {
      "epoch": 0.2841596130592503,
      "grad_norm": 2.4112820625305176,
      "learning_rate": 3.627944981615144e-05,
      "loss": 4.496,
      "step": 10575
    },
    {
      "epoch": 0.28429396748622865,
      "grad_norm": 2.4157190322875977,
      "learning_rate": 3.627264061010486e-05,
      "loss": 4.4601,
      "step": 10580
    },
    {
      "epoch": 0.28442832191320705,
      "grad_norm": 2.5503697395324707,
      "learning_rate": 3.626583140405829e-05,
      "loss": 4.3415,
      "step": 10585
    },
    {
      "epoch": 0.2845626763401854,
      "grad_norm": 2.1146814823150635,
      "learning_rate": 3.625902219801172e-05,
      "loss": 4.3361,
      "step": 10590
    },
    {
      "epoch": 0.2846970307671638,
      "grad_norm": 2.23808217048645,
      "learning_rate": 3.625221299196514e-05,
      "loss": 4.4614,
      "step": 10595
    },
    {
      "epoch": 0.28483138519414214,
      "grad_norm": 2.391080379486084,
      "learning_rate": 3.624540378591856e-05,
      "loss": 4.3429,
      "step": 10600
    },
    {
      "epoch": 0.28496573962112054,
      "grad_norm": 2.510503053665161,
      "learning_rate": 3.6238594579871985e-05,
      "loss": 4.4833,
      "step": 10605
    },
    {
      "epoch": 0.2851000940480989,
      "grad_norm": 2.296050786972046,
      "learning_rate": 3.623178537382541e-05,
      "loss": 4.4417,
      "step": 10610
    },
    {
      "epoch": 0.28523444847507723,
      "grad_norm": 2.3357465267181396,
      "learning_rate": 3.6224976167778843e-05,
      "loss": 4.4396,
      "step": 10615
    },
    {
      "epoch": 0.28536880290205563,
      "grad_norm": 2.3391647338867188,
      "learning_rate": 3.6218166961732266e-05,
      "loss": 4.4079,
      "step": 10620
    },
    {
      "epoch": 0.285503157329034,
      "grad_norm": 2.3151395320892334,
      "learning_rate": 3.621135775568569e-05,
      "loss": 4.4692,
      "step": 10625
    },
    {
      "epoch": 0.2856375117560124,
      "grad_norm": 2.2852585315704346,
      "learning_rate": 3.620454854963911e-05,
      "loss": 4.4648,
      "step": 10630
    },
    {
      "epoch": 0.2857718661829907,
      "grad_norm": 2.4205870628356934,
      "learning_rate": 3.619773934359254e-05,
      "loss": 4.3853,
      "step": 10635
    },
    {
      "epoch": 0.2859062206099691,
      "grad_norm": 2.181964635848999,
      "learning_rate": 3.619093013754596e-05,
      "loss": 4.4435,
      "step": 10640
    },
    {
      "epoch": 0.28604057503694746,
      "grad_norm": 2.2425522804260254,
      "learning_rate": 3.618412093149939e-05,
      "loss": 4.3077,
      "step": 10645
    },
    {
      "epoch": 0.2861749294639258,
      "grad_norm": 2.562311887741089,
      "learning_rate": 3.617731172545281e-05,
      "loss": 4.3823,
      "step": 10650
    },
    {
      "epoch": 0.2863092838909042,
      "grad_norm": 2.3054237365722656,
      "learning_rate": 3.617050251940624e-05,
      "loss": 4.3243,
      "step": 10655
    },
    {
      "epoch": 0.28644363831788255,
      "grad_norm": 2.220569133758545,
      "learning_rate": 3.6163693313359664e-05,
      "loss": 4.3406,
      "step": 10660
    },
    {
      "epoch": 0.28657799274486095,
      "grad_norm": 2.321383476257324,
      "learning_rate": 3.6156884107313087e-05,
      "loss": 4.3767,
      "step": 10665
    },
    {
      "epoch": 0.2867123471718393,
      "grad_norm": 2.3288893699645996,
      "learning_rate": 3.615007490126651e-05,
      "loss": 4.4126,
      "step": 10670
    },
    {
      "epoch": 0.2868467015988177,
      "grad_norm": 2.47627329826355,
      "learning_rate": 3.614326569521994e-05,
      "loss": 4.3013,
      "step": 10675
    },
    {
      "epoch": 0.28698105602579604,
      "grad_norm": 2.422590732574463,
      "learning_rate": 3.613645648917337e-05,
      "loss": 4.4255,
      "step": 10680
    },
    {
      "epoch": 0.28711541045277444,
      "grad_norm": 2.144662857055664,
      "learning_rate": 3.612964728312679e-05,
      "loss": 4.4167,
      "step": 10685
    },
    {
      "epoch": 0.2872497648797528,
      "grad_norm": 2.419856309890747,
      "learning_rate": 3.612283807708021e-05,
      "loss": 4.3795,
      "step": 10690
    },
    {
      "epoch": 0.28738411930673113,
      "grad_norm": 2.5130300521850586,
      "learning_rate": 3.611602887103364e-05,
      "loss": 4.3167,
      "step": 10695
    },
    {
      "epoch": 0.28751847373370953,
      "grad_norm": 2.4698777198791504,
      "learning_rate": 3.610921966498706e-05,
      "loss": 4.3989,
      "step": 10700
    },
    {
      "epoch": 0.2876528281606879,
      "grad_norm": 2.261470079421997,
      "learning_rate": 3.610241045894049e-05,
      "loss": 4.4147,
      "step": 10705
    },
    {
      "epoch": 0.2877871825876663,
      "grad_norm": 2.4074389934539795,
      "learning_rate": 3.6095601252893914e-05,
      "loss": 4.3915,
      "step": 10710
    },
    {
      "epoch": 0.2879215370146446,
      "grad_norm": 2.2203352451324463,
      "learning_rate": 3.608879204684734e-05,
      "loss": 4.4411,
      "step": 10715
    },
    {
      "epoch": 0.288055891441623,
      "grad_norm": 2.2533769607543945,
      "learning_rate": 3.6081982840800765e-05,
      "loss": 4.4036,
      "step": 10720
    },
    {
      "epoch": 0.28819024586860137,
      "grad_norm": 2.5158045291900635,
      "learning_rate": 3.607517363475419e-05,
      "loss": 4.3132,
      "step": 10725
    },
    {
      "epoch": 0.2883246002955797,
      "grad_norm": 2.354360818862915,
      "learning_rate": 3.606836442870761e-05,
      "loss": 4.4459,
      "step": 10730
    },
    {
      "epoch": 0.2884589547225581,
      "grad_norm": 2.3289177417755127,
      "learning_rate": 3.606155522266104e-05,
      "loss": 4.4517,
      "step": 10735
    },
    {
      "epoch": 0.28859330914953646,
      "grad_norm": 2.3871381282806396,
      "learning_rate": 3.605474601661447e-05,
      "loss": 4.3293,
      "step": 10740
    },
    {
      "epoch": 0.28872766357651486,
      "grad_norm": 2.4907991886138916,
      "learning_rate": 3.604793681056789e-05,
      "loss": 4.4504,
      "step": 10745
    },
    {
      "epoch": 0.2888620180034932,
      "grad_norm": 2.368387222290039,
      "learning_rate": 3.604112760452131e-05,
      "loss": 4.4293,
      "step": 10750
    },
    {
      "epoch": 0.2889963724304716,
      "grad_norm": 2.35514497756958,
      "learning_rate": 3.603431839847474e-05,
      "loss": 4.4665,
      "step": 10755
    },
    {
      "epoch": 0.28913072685744995,
      "grad_norm": 2.514583110809326,
      "learning_rate": 3.6027509192428164e-05,
      "loss": 4.2985,
      "step": 10760
    },
    {
      "epoch": 0.28926508128442835,
      "grad_norm": 2.2475037574768066,
      "learning_rate": 3.602069998638159e-05,
      "loss": 4.4275,
      "step": 10765
    },
    {
      "epoch": 0.2893994357114067,
      "grad_norm": 2.2607076168060303,
      "learning_rate": 3.6013890780335015e-05,
      "loss": 4.4705,
      "step": 10770
    },
    {
      "epoch": 0.28953379013838504,
      "grad_norm": 2.3649258613586426,
      "learning_rate": 3.600708157428844e-05,
      "loss": 4.4056,
      "step": 10775
    },
    {
      "epoch": 0.28966814456536344,
      "grad_norm": 2.1760125160217285,
      "learning_rate": 3.6000272368241866e-05,
      "loss": 4.4091,
      "step": 10780
    },
    {
      "epoch": 0.2898024989923418,
      "grad_norm": 2.48974347114563,
      "learning_rate": 3.599346316219529e-05,
      "loss": 4.4789,
      "step": 10785
    },
    {
      "epoch": 0.2899368534193202,
      "grad_norm": 2.404629945755005,
      "learning_rate": 3.598665395614871e-05,
      "loss": 4.4131,
      "step": 10790
    },
    {
      "epoch": 0.29007120784629853,
      "grad_norm": 2.3304474353790283,
      "learning_rate": 3.597984475010214e-05,
      "loss": 4.4051,
      "step": 10795
    },
    {
      "epoch": 0.29020556227327693,
      "grad_norm": 2.4542365074157715,
      "learning_rate": 3.597303554405557e-05,
      "loss": 4.3341,
      "step": 10800
    },
    {
      "epoch": 0.2903399167002553,
      "grad_norm": 2.328561782836914,
      "learning_rate": 3.596622633800899e-05,
      "loss": 4.3188,
      "step": 10805
    },
    {
      "epoch": 0.2904742711272336,
      "grad_norm": 2.4523963928222656,
      "learning_rate": 3.5959417131962414e-05,
      "loss": 4.3761,
      "step": 10810
    },
    {
      "epoch": 0.290608625554212,
      "grad_norm": 2.400569200515747,
      "learning_rate": 3.5952607925915836e-05,
      "loss": 4.5373,
      "step": 10815
    },
    {
      "epoch": 0.29074297998119036,
      "grad_norm": 2.5078623294830322,
      "learning_rate": 3.5945798719869265e-05,
      "loss": 4.4075,
      "step": 10820
    },
    {
      "epoch": 0.29087733440816876,
      "grad_norm": 2.3044204711914062,
      "learning_rate": 3.5938989513822694e-05,
      "loss": 4.367,
      "step": 10825
    },
    {
      "epoch": 0.2910116888351471,
      "grad_norm": 2.2870519161224365,
      "learning_rate": 3.5932180307776116e-05,
      "loss": 4.4056,
      "step": 10830
    },
    {
      "epoch": 0.2911460432621255,
      "grad_norm": 2.4397590160369873,
      "learning_rate": 3.592537110172954e-05,
      "loss": 4.3579,
      "step": 10835
    },
    {
      "epoch": 0.29128039768910385,
      "grad_norm": 2.2334470748901367,
      "learning_rate": 3.591856189568297e-05,
      "loss": 4.4265,
      "step": 10840
    },
    {
      "epoch": 0.2914147521160822,
      "grad_norm": 2.2962162494659424,
      "learning_rate": 3.591175268963639e-05,
      "loss": 4.3412,
      "step": 10845
    },
    {
      "epoch": 0.2915491065430606,
      "grad_norm": 2.281691551208496,
      "learning_rate": 3.590494348358981e-05,
      "loss": 4.4389,
      "step": 10850
    },
    {
      "epoch": 0.29168346097003894,
      "grad_norm": 2.365532398223877,
      "learning_rate": 3.589813427754324e-05,
      "loss": 4.4951,
      "step": 10855
    },
    {
      "epoch": 0.29181781539701734,
      "grad_norm": 2.326634168624878,
      "learning_rate": 3.589132507149667e-05,
      "loss": 4.3434,
      "step": 10860
    },
    {
      "epoch": 0.2919521698239957,
      "grad_norm": 2.6413421630859375,
      "learning_rate": 3.588451586545009e-05,
      "loss": 4.3872,
      "step": 10865
    },
    {
      "epoch": 0.2920865242509741,
      "grad_norm": 2.3470871448516846,
      "learning_rate": 3.5877706659403515e-05,
      "loss": 4.3722,
      "step": 10870
    },
    {
      "epoch": 0.29222087867795243,
      "grad_norm": 2.4437406063079834,
      "learning_rate": 3.587089745335694e-05,
      "loss": 4.4265,
      "step": 10875
    },
    {
      "epoch": 0.29235523310493083,
      "grad_norm": 2.169276475906372,
      "learning_rate": 3.5864088247310366e-05,
      "loss": 4.3829,
      "step": 10880
    },
    {
      "epoch": 0.2924895875319092,
      "grad_norm": 2.2414743900299072,
      "learning_rate": 3.585727904126379e-05,
      "loss": 4.3935,
      "step": 10885
    },
    {
      "epoch": 0.2926239419588875,
      "grad_norm": 2.283107042312622,
      "learning_rate": 3.585046983521722e-05,
      "loss": 4.3185,
      "step": 10890
    },
    {
      "epoch": 0.2927582963858659,
      "grad_norm": 2.3491806983947754,
      "learning_rate": 3.584366062917064e-05,
      "loss": 4.4095,
      "step": 10895
    },
    {
      "epoch": 0.29289265081284427,
      "grad_norm": 2.252155065536499,
      "learning_rate": 3.583685142312407e-05,
      "loss": 4.456,
      "step": 10900
    },
    {
      "epoch": 0.29302700523982267,
      "grad_norm": 2.1191248893737793,
      "learning_rate": 3.583004221707749e-05,
      "loss": 4.4297,
      "step": 10905
    },
    {
      "epoch": 0.293161359666801,
      "grad_norm": 2.2293434143066406,
      "learning_rate": 3.582323301103091e-05,
      "loss": 4.4096,
      "step": 10910
    },
    {
      "epoch": 0.2932957140937794,
      "grad_norm": 2.3526315689086914,
      "learning_rate": 3.581642380498434e-05,
      "loss": 4.3077,
      "step": 10915
    },
    {
      "epoch": 0.29343006852075776,
      "grad_norm": 2.2286932468414307,
      "learning_rate": 3.5809614598937765e-05,
      "loss": 4.3041,
      "step": 10920
    },
    {
      "epoch": 0.2935644229477361,
      "grad_norm": 2.38466739654541,
      "learning_rate": 3.5802805392891194e-05,
      "loss": 4.4351,
      "step": 10925
    },
    {
      "epoch": 0.2936987773747145,
      "grad_norm": 2.3098201751708984,
      "learning_rate": 3.5795996186844616e-05,
      "loss": 4.4392,
      "step": 10930
    },
    {
      "epoch": 0.29383313180169285,
      "grad_norm": 2.259516954421997,
      "learning_rate": 3.578918698079804e-05,
      "loss": 4.4694,
      "step": 10935
    },
    {
      "epoch": 0.29396748622867125,
      "grad_norm": 2.221287965774536,
      "learning_rate": 3.578237777475146e-05,
      "loss": 4.2827,
      "step": 10940
    },
    {
      "epoch": 0.2941018406556496,
      "grad_norm": 2.5505993366241455,
      "learning_rate": 3.577556856870489e-05,
      "loss": 4.4528,
      "step": 10945
    },
    {
      "epoch": 0.294236195082628,
      "grad_norm": 2.2073144912719727,
      "learning_rate": 3.576875936265832e-05,
      "loss": 4.4195,
      "step": 10950
    },
    {
      "epoch": 0.29437054950960634,
      "grad_norm": 2.5149948596954346,
      "learning_rate": 3.576195015661174e-05,
      "loss": 4.4233,
      "step": 10955
    },
    {
      "epoch": 0.2945049039365847,
      "grad_norm": 2.5190043449401855,
      "learning_rate": 3.575514095056516e-05,
      "loss": 4.3336,
      "step": 10960
    },
    {
      "epoch": 0.2946392583635631,
      "grad_norm": 2.2740297317504883,
      "learning_rate": 3.574833174451859e-05,
      "loss": 4.434,
      "step": 10965
    },
    {
      "epoch": 0.29477361279054143,
      "grad_norm": 2.2321646213531494,
      "learning_rate": 3.5741522538472014e-05,
      "loss": 4.5357,
      "step": 10970
    },
    {
      "epoch": 0.29490796721751983,
      "grad_norm": 2.4944801330566406,
      "learning_rate": 3.573471333242544e-05,
      "loss": 4.4857,
      "step": 10975
    },
    {
      "epoch": 0.2950423216444982,
      "grad_norm": 2.4806230068206787,
      "learning_rate": 3.5727904126378866e-05,
      "loss": 4.422,
      "step": 10980
    },
    {
      "epoch": 0.2951766760714766,
      "grad_norm": 2.392289400100708,
      "learning_rate": 3.5721094920332295e-05,
      "loss": 4.3546,
      "step": 10985
    },
    {
      "epoch": 0.2953110304984549,
      "grad_norm": 2.4640696048736572,
      "learning_rate": 3.571428571428572e-05,
      "loss": 4.5044,
      "step": 10990
    },
    {
      "epoch": 0.2954453849254333,
      "grad_norm": 2.2156856060028076,
      "learning_rate": 3.570747650823914e-05,
      "loss": 4.4228,
      "step": 10995
    },
    {
      "epoch": 0.29557973935241166,
      "grad_norm": 2.3098814487457275,
      "learning_rate": 3.570066730219256e-05,
      "loss": 4.3766,
      "step": 11000
    },
    {
      "epoch": 0.29571409377939,
      "grad_norm": 2.0859768390655518,
      "learning_rate": 3.569385809614599e-05,
      "loss": 4.4271,
      "step": 11005
    },
    {
      "epoch": 0.2958484482063684,
      "grad_norm": 2.573568820953369,
      "learning_rate": 3.568704889009942e-05,
      "loss": 4.43,
      "step": 11010
    },
    {
      "epoch": 0.29598280263334675,
      "grad_norm": 2.2669425010681152,
      "learning_rate": 3.568023968405284e-05,
      "loss": 4.2922,
      "step": 11015
    },
    {
      "epoch": 0.29611715706032515,
      "grad_norm": 2.443179130554199,
      "learning_rate": 3.5673430478006264e-05,
      "loss": 4.3523,
      "step": 11020
    },
    {
      "epoch": 0.2962515114873035,
      "grad_norm": 2.3064255714416504,
      "learning_rate": 3.566662127195969e-05,
      "loss": 4.4146,
      "step": 11025
    },
    {
      "epoch": 0.2963858659142819,
      "grad_norm": 2.366471767425537,
      "learning_rate": 3.5659812065913115e-05,
      "loss": 4.4128,
      "step": 11030
    },
    {
      "epoch": 0.29652022034126024,
      "grad_norm": 2.3048744201660156,
      "learning_rate": 3.565300285986654e-05,
      "loss": 4.4668,
      "step": 11035
    },
    {
      "epoch": 0.2966545747682386,
      "grad_norm": 2.431608200073242,
      "learning_rate": 3.564619365381997e-05,
      "loss": 4.3948,
      "step": 11040
    },
    {
      "epoch": 0.296788929195217,
      "grad_norm": 2.1769654750823975,
      "learning_rate": 3.5639384447773396e-05,
      "loss": 4.4429,
      "step": 11045
    },
    {
      "epoch": 0.29692328362219533,
      "grad_norm": 2.2304766178131104,
      "learning_rate": 3.563257524172682e-05,
      "loss": 4.4558,
      "step": 11050
    },
    {
      "epoch": 0.29705763804917373,
      "grad_norm": 2.6495771408081055,
      "learning_rate": 3.562576603568024e-05,
      "loss": 4.3541,
      "step": 11055
    },
    {
      "epoch": 0.2971919924761521,
      "grad_norm": 2.193000078201294,
      "learning_rate": 3.561895682963366e-05,
      "loss": 4.398,
      "step": 11060
    },
    {
      "epoch": 0.2973263469031305,
      "grad_norm": 2.5126960277557373,
      "learning_rate": 3.561214762358709e-05,
      "loss": 4.3713,
      "step": 11065
    },
    {
      "epoch": 0.2974607013301088,
      "grad_norm": 2.3089957237243652,
      "learning_rate": 3.560533841754052e-05,
      "loss": 4.3482,
      "step": 11070
    },
    {
      "epoch": 0.2975950557570872,
      "grad_norm": 2.3385934829711914,
      "learning_rate": 3.559852921149394e-05,
      "loss": 4.34,
      "step": 11075
    },
    {
      "epoch": 0.29772941018406557,
      "grad_norm": 2.3425333499908447,
      "learning_rate": 3.5591720005447365e-05,
      "loss": 4.3819,
      "step": 11080
    },
    {
      "epoch": 0.2978637646110439,
      "grad_norm": 2.419217824935913,
      "learning_rate": 3.558491079940079e-05,
      "loss": 4.3309,
      "step": 11085
    },
    {
      "epoch": 0.2979981190380223,
      "grad_norm": 2.4940073490142822,
      "learning_rate": 3.5578101593354217e-05,
      "loss": 4.4694,
      "step": 11090
    },
    {
      "epoch": 0.29813247346500066,
      "grad_norm": 2.1517605781555176,
      "learning_rate": 3.557129238730764e-05,
      "loss": 4.4293,
      "step": 11095
    },
    {
      "epoch": 0.29826682789197906,
      "grad_norm": 2.553548574447632,
      "learning_rate": 3.556448318126107e-05,
      "loss": 4.3153,
      "step": 11100
    },
    {
      "epoch": 0.2984011823189574,
      "grad_norm": 2.276975154876709,
      "learning_rate": 3.555767397521449e-05,
      "loss": 4.2643,
      "step": 11105
    },
    {
      "epoch": 0.2985355367459358,
      "grad_norm": 2.2542800903320312,
      "learning_rate": 3.555086476916792e-05,
      "loss": 4.5045,
      "step": 11110
    },
    {
      "epoch": 0.29866989117291415,
      "grad_norm": 2.1629130840301514,
      "learning_rate": 3.554405556312134e-05,
      "loss": 4.3954,
      "step": 11115
    },
    {
      "epoch": 0.2988042455998925,
      "grad_norm": 2.6322877407073975,
      "learning_rate": 3.5537246357074764e-05,
      "loss": 4.2769,
      "step": 11120
    },
    {
      "epoch": 0.2989386000268709,
      "grad_norm": 2.171420097351074,
      "learning_rate": 3.553043715102819e-05,
      "loss": 4.4171,
      "step": 11125
    },
    {
      "epoch": 0.29907295445384924,
      "grad_norm": 2.386714458465576,
      "learning_rate": 3.552362794498162e-05,
      "loss": 4.469,
      "step": 11130
    },
    {
      "epoch": 0.29920730888082764,
      "grad_norm": 2.2246439456939697,
      "learning_rate": 3.5516818738935044e-05,
      "loss": 4.3756,
      "step": 11135
    },
    {
      "epoch": 0.299341663307806,
      "grad_norm": 2.513181447982788,
      "learning_rate": 3.5510009532888466e-05,
      "loss": 4.4257,
      "step": 11140
    },
    {
      "epoch": 0.2994760177347844,
      "grad_norm": 2.3215301036834717,
      "learning_rate": 3.550320032684189e-05,
      "loss": 4.4151,
      "step": 11145
    },
    {
      "epoch": 0.29961037216176273,
      "grad_norm": 2.5489184856414795,
      "learning_rate": 3.549639112079532e-05,
      "loss": 4.4168,
      "step": 11150
    },
    {
      "epoch": 0.2997447265887411,
      "grad_norm": 2.383864164352417,
      "learning_rate": 3.548958191474874e-05,
      "loss": 4.3205,
      "step": 11155
    },
    {
      "epoch": 0.2998790810157195,
      "grad_norm": 2.7077414989471436,
      "learning_rate": 3.548277270870217e-05,
      "loss": 4.3645,
      "step": 11160
    },
    {
      "epoch": 0.3000134354426978,
      "grad_norm": 2.3905186653137207,
      "learning_rate": 3.547596350265559e-05,
      "loss": 4.5806,
      "step": 11165
    },
    {
      "epoch": 0.3001477898696762,
      "grad_norm": 2.443939208984375,
      "learning_rate": 3.546915429660902e-05,
      "loss": 4.3603,
      "step": 11170
    },
    {
      "epoch": 0.30028214429665456,
      "grad_norm": 2.447462558746338,
      "learning_rate": 3.546234509056244e-05,
      "loss": 4.2862,
      "step": 11175
    },
    {
      "epoch": 0.30041649872363296,
      "grad_norm": 2.6068060398101807,
      "learning_rate": 3.5455535884515865e-05,
      "loss": 4.3966,
      "step": 11180
    },
    {
      "epoch": 0.3005508531506113,
      "grad_norm": 2.393958568572998,
      "learning_rate": 3.5448726678469294e-05,
      "loss": 4.3843,
      "step": 11185
    },
    {
      "epoch": 0.3006852075775897,
      "grad_norm": 2.32204008102417,
      "learning_rate": 3.544191747242272e-05,
      "loss": 4.3656,
      "step": 11190
    },
    {
      "epoch": 0.30081956200456805,
      "grad_norm": 2.2890074253082275,
      "learning_rate": 3.5435108266376145e-05,
      "loss": 4.2652,
      "step": 11195
    },
    {
      "epoch": 0.3009539164315464,
      "grad_norm": 2.2149224281311035,
      "learning_rate": 3.542829906032957e-05,
      "loss": 4.4849,
      "step": 11200
    },
    {
      "epoch": 0.3010882708585248,
      "grad_norm": 2.5467464923858643,
      "learning_rate": 3.542148985428299e-05,
      "loss": 4.508,
      "step": 11205
    },
    {
      "epoch": 0.30122262528550314,
      "grad_norm": 2.1042325496673584,
      "learning_rate": 3.541468064823641e-05,
      "loss": 4.3913,
      "step": 11210
    },
    {
      "epoch": 0.30135697971248154,
      "grad_norm": 2.135972738265991,
      "learning_rate": 3.540787144218984e-05,
      "loss": 4.5184,
      "step": 11215
    },
    {
      "epoch": 0.3014913341394599,
      "grad_norm": 2.2988927364349365,
      "learning_rate": 3.540106223614327e-05,
      "loss": 4.4383,
      "step": 11220
    },
    {
      "epoch": 0.3016256885664383,
      "grad_norm": 2.4023237228393555,
      "learning_rate": 3.539425303009669e-05,
      "loss": 4.4747,
      "step": 11225
    },
    {
      "epoch": 0.30176004299341663,
      "grad_norm": 2.3424360752105713,
      "learning_rate": 3.5387443824050115e-05,
      "loss": 4.326,
      "step": 11230
    },
    {
      "epoch": 0.301894397420395,
      "grad_norm": 2.3838813304901123,
      "learning_rate": 3.5380634618003544e-05,
      "loss": 4.4709,
      "step": 11235
    },
    {
      "epoch": 0.3020287518473734,
      "grad_norm": 2.0433359146118164,
      "learning_rate": 3.5373825411956966e-05,
      "loss": 4.2943,
      "step": 11240
    },
    {
      "epoch": 0.3021631062743517,
      "grad_norm": 2.2191832065582275,
      "learning_rate": 3.536701620591039e-05,
      "loss": 4.4336,
      "step": 11245
    },
    {
      "epoch": 0.3022974607013301,
      "grad_norm": 2.4385364055633545,
      "learning_rate": 3.536020699986382e-05,
      "loss": 4.3519,
      "step": 11250
    },
    {
      "epoch": 0.30243181512830847,
      "grad_norm": 2.5782687664031982,
      "learning_rate": 3.5353397793817246e-05,
      "loss": 4.4066,
      "step": 11255
    },
    {
      "epoch": 0.30256616955528687,
      "grad_norm": 2.1897218227386475,
      "learning_rate": 3.534658858777067e-05,
      "loss": 4.4081,
      "step": 11260
    },
    {
      "epoch": 0.3027005239822652,
      "grad_norm": 2.364387273788452,
      "learning_rate": 3.533977938172409e-05,
      "loss": 4.4163,
      "step": 11265
    },
    {
      "epoch": 0.30283487840924356,
      "grad_norm": 2.163578748703003,
      "learning_rate": 3.533297017567751e-05,
      "loss": 4.4617,
      "step": 11270
    },
    {
      "epoch": 0.30296923283622196,
      "grad_norm": 2.3879506587982178,
      "learning_rate": 3.532616096963094e-05,
      "loss": 4.4299,
      "step": 11275
    },
    {
      "epoch": 0.3031035872632003,
      "grad_norm": 2.589524269104004,
      "learning_rate": 3.531935176358437e-05,
      "loss": 4.3473,
      "step": 11280
    },
    {
      "epoch": 0.3032379416901787,
      "grad_norm": 2.3256139755249023,
      "learning_rate": 3.5312542557537793e-05,
      "loss": 4.4461,
      "step": 11285
    },
    {
      "epoch": 0.30337229611715705,
      "grad_norm": 2.5100226402282715,
      "learning_rate": 3.5305733351491216e-05,
      "loss": 4.4542,
      "step": 11290
    },
    {
      "epoch": 0.30350665054413545,
      "grad_norm": 2.4867424964904785,
      "learning_rate": 3.5298924145444645e-05,
      "loss": 4.479,
      "step": 11295
    },
    {
      "epoch": 0.3036410049711138,
      "grad_norm": 2.80302095413208,
      "learning_rate": 3.529211493939807e-05,
      "loss": 4.4085,
      "step": 11300
    },
    {
      "epoch": 0.3037753593980922,
      "grad_norm": 2.315727949142456,
      "learning_rate": 3.528530573335149e-05,
      "loss": 4.4746,
      "step": 11305
    },
    {
      "epoch": 0.30390971382507054,
      "grad_norm": 2.3622002601623535,
      "learning_rate": 3.527849652730492e-05,
      "loss": 4.3331,
      "step": 11310
    },
    {
      "epoch": 0.3040440682520489,
      "grad_norm": 2.4788386821746826,
      "learning_rate": 3.527168732125835e-05,
      "loss": 4.3672,
      "step": 11315
    },
    {
      "epoch": 0.3041784226790273,
      "grad_norm": 2.3090922832489014,
      "learning_rate": 3.526487811521177e-05,
      "loss": 4.391,
      "step": 11320
    },
    {
      "epoch": 0.3043127771060056,
      "grad_norm": 2.557281255722046,
      "learning_rate": 3.525806890916519e-05,
      "loss": 4.4872,
      "step": 11325
    },
    {
      "epoch": 0.304447131532984,
      "grad_norm": 2.4267678260803223,
      "learning_rate": 3.5251259703118614e-05,
      "loss": 4.3654,
      "step": 11330
    },
    {
      "epoch": 0.3045814859599624,
      "grad_norm": 2.4860310554504395,
      "learning_rate": 3.524445049707204e-05,
      "loss": 4.3791,
      "step": 11335
    },
    {
      "epoch": 0.3047158403869408,
      "grad_norm": 2.445606231689453,
      "learning_rate": 3.523764129102547e-05,
      "loss": 4.3557,
      "step": 11340
    },
    {
      "epoch": 0.3048501948139191,
      "grad_norm": 2.7110044956207275,
      "learning_rate": 3.5230832084978895e-05,
      "loss": 4.3676,
      "step": 11345
    },
    {
      "epoch": 0.30498454924089746,
      "grad_norm": 2.3757598400115967,
      "learning_rate": 3.522402287893232e-05,
      "loss": 4.2883,
      "step": 11350
    },
    {
      "epoch": 0.30511890366787586,
      "grad_norm": 2.3647594451904297,
      "learning_rate": 3.521721367288574e-05,
      "loss": 4.2846,
      "step": 11355
    },
    {
      "epoch": 0.3052532580948542,
      "grad_norm": 2.4244589805603027,
      "learning_rate": 3.521040446683917e-05,
      "loss": 4.5249,
      "step": 11360
    },
    {
      "epoch": 0.3053876125218326,
      "grad_norm": 2.1759564876556396,
      "learning_rate": 3.520359526079259e-05,
      "loss": 4.3643,
      "step": 11365
    },
    {
      "epoch": 0.30552196694881095,
      "grad_norm": 2.3249053955078125,
      "learning_rate": 3.519678605474602e-05,
      "loss": 4.3989,
      "step": 11370
    },
    {
      "epoch": 0.30565632137578935,
      "grad_norm": 2.5302419662475586,
      "learning_rate": 3.518997684869944e-05,
      "loss": 4.3542,
      "step": 11375
    },
    {
      "epoch": 0.3057906758027677,
      "grad_norm": 2.640538215637207,
      "learning_rate": 3.518316764265287e-05,
      "loss": 4.4415,
      "step": 11380
    },
    {
      "epoch": 0.3059250302297461,
      "grad_norm": 2.354644536972046,
      "learning_rate": 3.517635843660629e-05,
      "loss": 4.3588,
      "step": 11385
    },
    {
      "epoch": 0.30605938465672444,
      "grad_norm": 2.2276113033294678,
      "learning_rate": 3.5169549230559715e-05,
      "loss": 4.3923,
      "step": 11390
    },
    {
      "epoch": 0.3061937390837028,
      "grad_norm": 2.4160144329071045,
      "learning_rate": 3.5162740024513144e-05,
      "loss": 4.3419,
      "step": 11395
    },
    {
      "epoch": 0.3063280935106812,
      "grad_norm": 2.4271352291107178,
      "learning_rate": 3.5155930818466573e-05,
      "loss": 4.4491,
      "step": 11400
    },
    {
      "epoch": 0.30646244793765953,
      "grad_norm": 2.392716646194458,
      "learning_rate": 3.5149121612419996e-05,
      "loss": 4.4226,
      "step": 11405
    },
    {
      "epoch": 0.30659680236463793,
      "grad_norm": 2.318420886993408,
      "learning_rate": 3.514231240637342e-05,
      "loss": 4.4825,
      "step": 11410
    },
    {
      "epoch": 0.3067311567916163,
      "grad_norm": 2.277195930480957,
      "learning_rate": 3.513550320032684e-05,
      "loss": 4.429,
      "step": 11415
    },
    {
      "epoch": 0.3068655112185947,
      "grad_norm": 2.4826502799987793,
      "learning_rate": 3.512869399428027e-05,
      "loss": 4.3727,
      "step": 11420
    },
    {
      "epoch": 0.306999865645573,
      "grad_norm": 2.258009672164917,
      "learning_rate": 3.512188478823369e-05,
      "loss": 4.3508,
      "step": 11425
    },
    {
      "epoch": 0.30713422007255137,
      "grad_norm": 2.3815462589263916,
      "learning_rate": 3.511507558218712e-05,
      "loss": 4.4836,
      "step": 11430
    },
    {
      "epoch": 0.30726857449952977,
      "grad_norm": 2.2452762126922607,
      "learning_rate": 3.510826637614054e-05,
      "loss": 4.472,
      "step": 11435
    },
    {
      "epoch": 0.3074029289265081,
      "grad_norm": 2.2901456356048584,
      "learning_rate": 3.510145717009397e-05,
      "loss": 4.3503,
      "step": 11440
    },
    {
      "epoch": 0.3075372833534865,
      "grad_norm": 2.1576340198516846,
      "learning_rate": 3.5094647964047394e-05,
      "loss": 4.2907,
      "step": 11445
    },
    {
      "epoch": 0.30767163778046486,
      "grad_norm": 2.279529571533203,
      "learning_rate": 3.5087838758000816e-05,
      "loss": 4.341,
      "step": 11450
    },
    {
      "epoch": 0.30780599220744326,
      "grad_norm": 2.130805492401123,
      "learning_rate": 3.508102955195424e-05,
      "loss": 4.35,
      "step": 11455
    },
    {
      "epoch": 0.3079403466344216,
      "grad_norm": 2.3189432621002197,
      "learning_rate": 3.5074220345907675e-05,
      "loss": 4.3288,
      "step": 11460
    },
    {
      "epoch": 0.30807470106139995,
      "grad_norm": 2.7252275943756104,
      "learning_rate": 3.50674111398611e-05,
      "loss": 4.4134,
      "step": 11465
    },
    {
      "epoch": 0.30820905548837835,
      "grad_norm": 2.7170603275299072,
      "learning_rate": 3.506060193381452e-05,
      "loss": 4.3571,
      "step": 11470
    },
    {
      "epoch": 0.3083434099153567,
      "grad_norm": 2.2511637210845947,
      "learning_rate": 3.505379272776794e-05,
      "loss": 4.399,
      "step": 11475
    },
    {
      "epoch": 0.3084777643423351,
      "grad_norm": 2.2870779037475586,
      "learning_rate": 3.504698352172137e-05,
      "loss": 4.2764,
      "step": 11480
    },
    {
      "epoch": 0.30861211876931344,
      "grad_norm": 2.5355606079101562,
      "learning_rate": 3.504017431567479e-05,
      "loss": 4.3383,
      "step": 11485
    },
    {
      "epoch": 0.30874647319629184,
      "grad_norm": 2.5333197116851807,
      "learning_rate": 3.503336510962822e-05,
      "loss": 4.3589,
      "step": 11490
    },
    {
      "epoch": 0.3088808276232702,
      "grad_norm": 2.444096565246582,
      "learning_rate": 3.5026555903581644e-05,
      "loss": 4.4517,
      "step": 11495
    },
    {
      "epoch": 0.3090151820502486,
      "grad_norm": 2.6205103397369385,
      "learning_rate": 3.5019746697535066e-05,
      "loss": 4.2829,
      "step": 11500
    },
    {
      "epoch": 0.3091495364772269,
      "grad_norm": 2.2746925354003906,
      "learning_rate": 3.5012937491488495e-05,
      "loss": 4.4552,
      "step": 11505
    },
    {
      "epoch": 0.30928389090420527,
      "grad_norm": 2.1340034008026123,
      "learning_rate": 3.500612828544192e-05,
      "loss": 4.3261,
      "step": 11510
    },
    {
      "epoch": 0.30941824533118367,
      "grad_norm": 2.2832868099212646,
      "learning_rate": 3.499931907939534e-05,
      "loss": 4.3634,
      "step": 11515
    },
    {
      "epoch": 0.309552599758162,
      "grad_norm": 2.423352003097534,
      "learning_rate": 3.499250987334877e-05,
      "loss": 4.4669,
      "step": 11520
    },
    {
      "epoch": 0.3096869541851404,
      "grad_norm": 2.549466133117676,
      "learning_rate": 3.49857006673022e-05,
      "loss": 4.4056,
      "step": 11525
    },
    {
      "epoch": 0.30982130861211876,
      "grad_norm": 2.4861085414886475,
      "learning_rate": 3.497889146125562e-05,
      "loss": 4.3539,
      "step": 11530
    },
    {
      "epoch": 0.30995566303909716,
      "grad_norm": 2.405954360961914,
      "learning_rate": 3.497208225520904e-05,
      "loss": 4.4271,
      "step": 11535
    },
    {
      "epoch": 0.3100900174660755,
      "grad_norm": 2.41330623626709,
      "learning_rate": 3.4965273049162465e-05,
      "loss": 4.336,
      "step": 11540
    },
    {
      "epoch": 0.31022437189305385,
      "grad_norm": 2.473904848098755,
      "learning_rate": 3.4958463843115894e-05,
      "loss": 4.322,
      "step": 11545
    },
    {
      "epoch": 0.31035872632003225,
      "grad_norm": 2.2048988342285156,
      "learning_rate": 3.495165463706932e-05,
      "loss": 4.3809,
      "step": 11550
    },
    {
      "epoch": 0.3104930807470106,
      "grad_norm": 2.625628709793091,
      "learning_rate": 3.4944845431022745e-05,
      "loss": 4.3046,
      "step": 11555
    },
    {
      "epoch": 0.310627435173989,
      "grad_norm": 2.1766016483306885,
      "learning_rate": 3.493803622497617e-05,
      "loss": 4.4934,
      "step": 11560
    },
    {
      "epoch": 0.31076178960096734,
      "grad_norm": 2.361522674560547,
      "learning_rate": 3.4931227018929596e-05,
      "loss": 4.3739,
      "step": 11565
    },
    {
      "epoch": 0.31089614402794574,
      "grad_norm": 2.4544107913970947,
      "learning_rate": 3.492441781288302e-05,
      "loss": 4.2935,
      "step": 11570
    },
    {
      "epoch": 0.3110304984549241,
      "grad_norm": 2.2605831623077393,
      "learning_rate": 3.491760860683644e-05,
      "loss": 4.3944,
      "step": 11575
    },
    {
      "epoch": 0.31116485288190243,
      "grad_norm": 2.448289155960083,
      "learning_rate": 3.491079940078987e-05,
      "loss": 4.413,
      "step": 11580
    },
    {
      "epoch": 0.31129920730888083,
      "grad_norm": 2.324584484100342,
      "learning_rate": 3.49039901947433e-05,
      "loss": 4.3521,
      "step": 11585
    },
    {
      "epoch": 0.3114335617358592,
      "grad_norm": 2.1733994483947754,
      "learning_rate": 3.489718098869672e-05,
      "loss": 4.2927,
      "step": 11590
    },
    {
      "epoch": 0.3115679161628376,
      "grad_norm": 2.2224202156066895,
      "learning_rate": 3.4890371782650144e-05,
      "loss": 4.3877,
      "step": 11595
    },
    {
      "epoch": 0.3117022705898159,
      "grad_norm": 2.3119382858276367,
      "learning_rate": 3.4883562576603566e-05,
      "loss": 4.4716,
      "step": 11600
    },
    {
      "epoch": 0.3118366250167943,
      "grad_norm": 2.497523307800293,
      "learning_rate": 3.4876753370556995e-05,
      "loss": 4.3796,
      "step": 11605
    },
    {
      "epoch": 0.31197097944377267,
      "grad_norm": 2.5951461791992188,
      "learning_rate": 3.4869944164510424e-05,
      "loss": 4.4597,
      "step": 11610
    },
    {
      "epoch": 0.31210533387075107,
      "grad_norm": 2.310804605484009,
      "learning_rate": 3.4863134958463846e-05,
      "loss": 4.3686,
      "step": 11615
    },
    {
      "epoch": 0.3122396882977294,
      "grad_norm": 2.299835205078125,
      "learning_rate": 3.485632575241727e-05,
      "loss": 4.3243,
      "step": 11620
    },
    {
      "epoch": 0.31237404272470776,
      "grad_norm": 2.4097862243652344,
      "learning_rate": 3.48495165463707e-05,
      "loss": 4.2957,
      "step": 11625
    },
    {
      "epoch": 0.31250839715168616,
      "grad_norm": 2.3484880924224854,
      "learning_rate": 3.484270734032412e-05,
      "loss": 4.4187,
      "step": 11630
    },
    {
      "epoch": 0.3126427515786645,
      "grad_norm": 2.4458000659942627,
      "learning_rate": 3.483589813427754e-05,
      "loss": 4.4133,
      "step": 11635
    },
    {
      "epoch": 0.3127771060056429,
      "grad_norm": 2.198395252227783,
      "learning_rate": 3.482908892823097e-05,
      "loss": 4.3864,
      "step": 11640
    },
    {
      "epoch": 0.31291146043262125,
      "grad_norm": 2.3579471111297607,
      "learning_rate": 3.48222797221844e-05,
      "loss": 4.364,
      "step": 11645
    },
    {
      "epoch": 0.31304581485959965,
      "grad_norm": 2.3208608627319336,
      "learning_rate": 3.481547051613782e-05,
      "loss": 4.3395,
      "step": 11650
    },
    {
      "epoch": 0.313180169286578,
      "grad_norm": 2.4523866176605225,
      "learning_rate": 3.4808661310091245e-05,
      "loss": 4.3132,
      "step": 11655
    },
    {
      "epoch": 0.31331452371355634,
      "grad_norm": 2.200252056121826,
      "learning_rate": 3.480185210404467e-05,
      "loss": 4.4443,
      "step": 11660
    },
    {
      "epoch": 0.31344887814053474,
      "grad_norm": 2.851299524307251,
      "learning_rate": 3.479504289799809e-05,
      "loss": 4.3955,
      "step": 11665
    },
    {
      "epoch": 0.3135832325675131,
      "grad_norm": 2.353484630584717,
      "learning_rate": 3.4788233691951525e-05,
      "loss": 4.4971,
      "step": 11670
    },
    {
      "epoch": 0.3137175869944915,
      "grad_norm": 2.378593921661377,
      "learning_rate": 3.478142448590495e-05,
      "loss": 4.3491,
      "step": 11675
    },
    {
      "epoch": 0.3138519414214698,
      "grad_norm": 2.4765584468841553,
      "learning_rate": 3.477461527985837e-05,
      "loss": 4.2711,
      "step": 11680
    },
    {
      "epoch": 0.3139862958484482,
      "grad_norm": 2.45198655128479,
      "learning_rate": 3.476780607381179e-05,
      "loss": 4.5146,
      "step": 11685
    },
    {
      "epoch": 0.31412065027542657,
      "grad_norm": 2.3231348991394043,
      "learning_rate": 3.476099686776522e-05,
      "loss": 4.4728,
      "step": 11690
    },
    {
      "epoch": 0.31425500470240497,
      "grad_norm": 2.1213226318359375,
      "learning_rate": 3.475418766171864e-05,
      "loss": 4.3763,
      "step": 11695
    },
    {
      "epoch": 0.3143893591293833,
      "grad_norm": 2.402273416519165,
      "learning_rate": 3.474737845567207e-05,
      "loss": 4.333,
      "step": 11700
    },
    {
      "epoch": 0.31452371355636166,
      "grad_norm": 2.523179531097412,
      "learning_rate": 3.4740569249625494e-05,
      "loss": 4.4392,
      "step": 11705
    },
    {
      "epoch": 0.31465806798334006,
      "grad_norm": 2.1239614486694336,
      "learning_rate": 3.4733760043578924e-05,
      "loss": 4.2906,
      "step": 11710
    },
    {
      "epoch": 0.3147924224103184,
      "grad_norm": 2.474504232406616,
      "learning_rate": 3.4726950837532346e-05,
      "loss": 4.3859,
      "step": 11715
    },
    {
      "epoch": 0.3149267768372968,
      "grad_norm": 2.3247148990631104,
      "learning_rate": 3.472014163148577e-05,
      "loss": 4.3754,
      "step": 11720
    },
    {
      "epoch": 0.31506113126427515,
      "grad_norm": 2.348541021347046,
      "learning_rate": 3.471333242543919e-05,
      "loss": 4.4282,
      "step": 11725
    },
    {
      "epoch": 0.31519548569125355,
      "grad_norm": 2.3271644115448,
      "learning_rate": 3.470652321939262e-05,
      "loss": 4.3513,
      "step": 11730
    },
    {
      "epoch": 0.3153298401182319,
      "grad_norm": 2.366567850112915,
      "learning_rate": 3.469971401334605e-05,
      "loss": 4.3926,
      "step": 11735
    },
    {
      "epoch": 0.31546419454521024,
      "grad_norm": 2.4325003623962402,
      "learning_rate": 3.469290480729947e-05,
      "loss": 4.2803,
      "step": 11740
    },
    {
      "epoch": 0.31559854897218864,
      "grad_norm": 2.245438814163208,
      "learning_rate": 3.468609560125289e-05,
      "loss": 4.3194,
      "step": 11745
    },
    {
      "epoch": 0.315732903399167,
      "grad_norm": 2.609093427658081,
      "learning_rate": 3.467928639520632e-05,
      "loss": 4.4137,
      "step": 11750
    },
    {
      "epoch": 0.3158672578261454,
      "grad_norm": 2.327552556991577,
      "learning_rate": 3.4672477189159744e-05,
      "loss": 4.2121,
      "step": 11755
    },
    {
      "epoch": 0.31600161225312373,
      "grad_norm": 2.444973945617676,
      "learning_rate": 3.466566798311317e-05,
      "loss": 4.5011,
      "step": 11760
    },
    {
      "epoch": 0.31613596668010213,
      "grad_norm": 2.243318796157837,
      "learning_rate": 3.4658858777066596e-05,
      "loss": 4.5054,
      "step": 11765
    },
    {
      "epoch": 0.3162703211070805,
      "grad_norm": 2.444439649581909,
      "learning_rate": 3.4652049571020025e-05,
      "loss": 4.429,
      "step": 11770
    },
    {
      "epoch": 0.3164046755340588,
      "grad_norm": 2.259615182876587,
      "learning_rate": 3.464524036497345e-05,
      "loss": 4.3728,
      "step": 11775
    },
    {
      "epoch": 0.3165390299610372,
      "grad_norm": 2.425847053527832,
      "learning_rate": 3.463843115892687e-05,
      "loss": 4.4365,
      "step": 11780
    },
    {
      "epoch": 0.31667338438801557,
      "grad_norm": 2.1365108489990234,
      "learning_rate": 3.463162195288029e-05,
      "loss": 4.2796,
      "step": 11785
    },
    {
      "epoch": 0.31680773881499397,
      "grad_norm": 2.366035223007202,
      "learning_rate": 3.462481274683372e-05,
      "loss": 4.3081,
      "step": 11790
    },
    {
      "epoch": 0.3169420932419723,
      "grad_norm": 2.532768964767456,
      "learning_rate": 3.461800354078715e-05,
      "loss": 4.4481,
      "step": 11795
    },
    {
      "epoch": 0.3170764476689507,
      "grad_norm": 2.1273610591888428,
      "learning_rate": 3.461119433474057e-05,
      "loss": 4.2419,
      "step": 11800
    },
    {
      "epoch": 0.31721080209592906,
      "grad_norm": 2.3638885021209717,
      "learning_rate": 3.4604385128693994e-05,
      "loss": 4.3347,
      "step": 11805
    },
    {
      "epoch": 0.31734515652290746,
      "grad_norm": 2.5616016387939453,
      "learning_rate": 3.4597575922647416e-05,
      "loss": 4.3735,
      "step": 11810
    },
    {
      "epoch": 0.3174795109498858,
      "grad_norm": 2.497040033340454,
      "learning_rate": 3.4590766716600845e-05,
      "loss": 4.3,
      "step": 11815
    },
    {
      "epoch": 0.31761386537686415,
      "grad_norm": 2.1773123741149902,
      "learning_rate": 3.4583957510554274e-05,
      "loss": 4.3572,
      "step": 11820
    },
    {
      "epoch": 0.31774821980384255,
      "grad_norm": 2.2486603260040283,
      "learning_rate": 3.45771483045077e-05,
      "loss": 4.3517,
      "step": 11825
    },
    {
      "epoch": 0.3178825742308209,
      "grad_norm": 2.387951374053955,
      "learning_rate": 3.457033909846112e-05,
      "loss": 4.3378,
      "step": 11830
    },
    {
      "epoch": 0.3180169286577993,
      "grad_norm": 2.4602630138397217,
      "learning_rate": 3.456352989241455e-05,
      "loss": 4.2555,
      "step": 11835
    },
    {
      "epoch": 0.31815128308477764,
      "grad_norm": 2.4527013301849365,
      "learning_rate": 3.455672068636797e-05,
      "loss": 4.3989,
      "step": 11840
    },
    {
      "epoch": 0.31828563751175604,
      "grad_norm": 2.3894920349121094,
      "learning_rate": 3.454991148032139e-05,
      "loss": 4.3874,
      "step": 11845
    },
    {
      "epoch": 0.3184199919387344,
      "grad_norm": 2.2721242904663086,
      "learning_rate": 3.454310227427482e-05,
      "loss": 4.3077,
      "step": 11850
    },
    {
      "epoch": 0.3185543463657127,
      "grad_norm": 2.3930463790893555,
      "learning_rate": 3.453629306822825e-05,
      "loss": 4.1812,
      "step": 11855
    },
    {
      "epoch": 0.3186887007926911,
      "grad_norm": 2.3637478351593018,
      "learning_rate": 3.452948386218167e-05,
      "loss": 4.4238,
      "step": 11860
    },
    {
      "epoch": 0.31882305521966947,
      "grad_norm": 2.376488208770752,
      "learning_rate": 3.4522674656135095e-05,
      "loss": 4.3669,
      "step": 11865
    },
    {
      "epoch": 0.31895740964664787,
      "grad_norm": 2.1426193714141846,
      "learning_rate": 3.451586545008852e-05,
      "loss": 4.2486,
      "step": 11870
    },
    {
      "epoch": 0.3190917640736262,
      "grad_norm": 2.336909055709839,
      "learning_rate": 3.4509056244041947e-05,
      "loss": 4.2129,
      "step": 11875
    },
    {
      "epoch": 0.3192261185006046,
      "grad_norm": 2.4798941612243652,
      "learning_rate": 3.450224703799537e-05,
      "loss": 4.2543,
      "step": 11880
    },
    {
      "epoch": 0.31936047292758296,
      "grad_norm": 2.344862699508667,
      "learning_rate": 3.44954378319488e-05,
      "loss": 4.3187,
      "step": 11885
    },
    {
      "epoch": 0.3194948273545613,
      "grad_norm": 2.3275115489959717,
      "learning_rate": 3.448862862590222e-05,
      "loss": 4.429,
      "step": 11890
    },
    {
      "epoch": 0.3196291817815397,
      "grad_norm": 2.7610864639282227,
      "learning_rate": 3.448181941985565e-05,
      "loss": 4.3432,
      "step": 11895
    },
    {
      "epoch": 0.31976353620851805,
      "grad_norm": 2.594911813735962,
      "learning_rate": 3.447501021380907e-05,
      "loss": 4.3449,
      "step": 11900
    },
    {
      "epoch": 0.31989789063549645,
      "grad_norm": 2.357860565185547,
      "learning_rate": 3.4468201007762494e-05,
      "loss": 4.3676,
      "step": 11905
    },
    {
      "epoch": 0.3200322450624748,
      "grad_norm": 2.3438656330108643,
      "learning_rate": 3.446139180171592e-05,
      "loss": 4.4071,
      "step": 11910
    },
    {
      "epoch": 0.3201665994894532,
      "grad_norm": 2.2888834476470947,
      "learning_rate": 3.445458259566935e-05,
      "loss": 4.3398,
      "step": 11915
    },
    {
      "epoch": 0.32030095391643154,
      "grad_norm": 2.390756368637085,
      "learning_rate": 3.4447773389622774e-05,
      "loss": 4.4164,
      "step": 11920
    },
    {
      "epoch": 0.32043530834340994,
      "grad_norm": 2.4769511222839355,
      "learning_rate": 3.4440964183576196e-05,
      "loss": 4.3921,
      "step": 11925
    },
    {
      "epoch": 0.3205696627703883,
      "grad_norm": 2.3073854446411133,
      "learning_rate": 3.443415497752962e-05,
      "loss": 4.4784,
      "step": 11930
    },
    {
      "epoch": 0.32070401719736663,
      "grad_norm": 2.395254135131836,
      "learning_rate": 3.442734577148305e-05,
      "loss": 4.3011,
      "step": 11935
    },
    {
      "epoch": 0.32083837162434503,
      "grad_norm": 2.1752660274505615,
      "learning_rate": 3.442053656543647e-05,
      "loss": 4.403,
      "step": 11940
    },
    {
      "epoch": 0.3209727260513234,
      "grad_norm": 2.0925235748291016,
      "learning_rate": 3.44137273593899e-05,
      "loss": 4.3205,
      "step": 11945
    },
    {
      "epoch": 0.3211070804783018,
      "grad_norm": 2.3411688804626465,
      "learning_rate": 3.440691815334332e-05,
      "loss": 4.2918,
      "step": 11950
    },
    {
      "epoch": 0.3212414349052801,
      "grad_norm": 2.327207326889038,
      "learning_rate": 3.4400108947296743e-05,
      "loss": 4.4743,
      "step": 11955
    },
    {
      "epoch": 0.3213757893322585,
      "grad_norm": 2.190216541290283,
      "learning_rate": 3.439329974125017e-05,
      "loss": 4.3268,
      "step": 11960
    },
    {
      "epoch": 0.32151014375923687,
      "grad_norm": 2.3104262351989746,
      "learning_rate": 3.4386490535203595e-05,
      "loss": 4.3758,
      "step": 11965
    },
    {
      "epoch": 0.3216444981862152,
      "grad_norm": 2.736016035079956,
      "learning_rate": 3.4379681329157024e-05,
      "loss": 4.3665,
      "step": 11970
    },
    {
      "epoch": 0.3217788526131936,
      "grad_norm": 2.324835777282715,
      "learning_rate": 3.4372872123110446e-05,
      "loss": 4.4447,
      "step": 11975
    },
    {
      "epoch": 0.32191320704017196,
      "grad_norm": 2.1983792781829834,
      "learning_rate": 3.4366062917063875e-05,
      "loss": 4.3595,
      "step": 11980
    },
    {
      "epoch": 0.32204756146715036,
      "grad_norm": 2.416890859603882,
      "learning_rate": 3.43592537110173e-05,
      "loss": 4.2341,
      "step": 11985
    },
    {
      "epoch": 0.3221819158941287,
      "grad_norm": 2.243213653564453,
      "learning_rate": 3.435244450497072e-05,
      "loss": 4.3102,
      "step": 11990
    },
    {
      "epoch": 0.3223162703211071,
      "grad_norm": 2.4709558486938477,
      "learning_rate": 3.434563529892414e-05,
      "loss": 4.3355,
      "step": 11995
    },
    {
      "epoch": 0.32245062474808545,
      "grad_norm": 2.502027750015259,
      "learning_rate": 3.433882609287757e-05,
      "loss": 4.3304,
      "step": 12000
    },
    {
      "epoch": 0.3225849791750638,
      "grad_norm": 2.3686423301696777,
      "learning_rate": 3.4332016886831e-05,
      "loss": 4.3382,
      "step": 12005
    },
    {
      "epoch": 0.3227193336020422,
      "grad_norm": 2.3989109992980957,
      "learning_rate": 3.432520768078442e-05,
      "loss": 4.4672,
      "step": 12010
    },
    {
      "epoch": 0.32285368802902054,
      "grad_norm": 2.2610034942626953,
      "learning_rate": 3.4318398474737845e-05,
      "loss": 4.4217,
      "step": 12015
    },
    {
      "epoch": 0.32298804245599894,
      "grad_norm": 2.316645860671997,
      "learning_rate": 3.4311589268691274e-05,
      "loss": 4.2564,
      "step": 12020
    },
    {
      "epoch": 0.3231223968829773,
      "grad_norm": 2.1691601276397705,
      "learning_rate": 3.4304780062644696e-05,
      "loss": 4.3506,
      "step": 12025
    },
    {
      "epoch": 0.3232567513099557,
      "grad_norm": 2.4404349327087402,
      "learning_rate": 3.4297970856598125e-05,
      "loss": 4.4703,
      "step": 12030
    },
    {
      "epoch": 0.323391105736934,
      "grad_norm": 2.2524938583374023,
      "learning_rate": 3.429116165055155e-05,
      "loss": 4.418,
      "step": 12035
    },
    {
      "epoch": 0.3235254601639124,
      "grad_norm": 2.566941976547241,
      "learning_rate": 3.4284352444504976e-05,
      "loss": 4.2888,
      "step": 12040
    },
    {
      "epoch": 0.32365981459089077,
      "grad_norm": 2.5094141960144043,
      "learning_rate": 3.42775432384584e-05,
      "loss": 4.2487,
      "step": 12045
    },
    {
      "epoch": 0.3237941690178691,
      "grad_norm": 2.543335199356079,
      "learning_rate": 3.427073403241182e-05,
      "loss": 4.4021,
      "step": 12050
    },
    {
      "epoch": 0.3239285234448475,
      "grad_norm": 2.3157460689544678,
      "learning_rate": 3.426392482636524e-05,
      "loss": 4.2615,
      "step": 12055
    },
    {
      "epoch": 0.32406287787182586,
      "grad_norm": 2.25266170501709,
      "learning_rate": 3.425711562031867e-05,
      "loss": 4.4013,
      "step": 12060
    },
    {
      "epoch": 0.32419723229880426,
      "grad_norm": 2.391850709915161,
      "learning_rate": 3.42503064142721e-05,
      "loss": 4.4549,
      "step": 12065
    },
    {
      "epoch": 0.3243315867257826,
      "grad_norm": 2.4575178623199463,
      "learning_rate": 3.4243497208225523e-05,
      "loss": 4.3986,
      "step": 12070
    },
    {
      "epoch": 0.324465941152761,
      "grad_norm": 2.5197598934173584,
      "learning_rate": 3.4236688002178946e-05,
      "loss": 4.1211,
      "step": 12075
    },
    {
      "epoch": 0.32460029557973935,
      "grad_norm": 2.5386927127838135,
      "learning_rate": 3.4229878796132375e-05,
      "loss": 4.5548,
      "step": 12080
    },
    {
      "epoch": 0.3247346500067177,
      "grad_norm": 2.444425582885742,
      "learning_rate": 3.42230695900858e-05,
      "loss": 4.3745,
      "step": 12085
    },
    {
      "epoch": 0.3248690044336961,
      "grad_norm": 2.4903130531311035,
      "learning_rate": 3.421626038403922e-05,
      "loss": 4.2791,
      "step": 12090
    },
    {
      "epoch": 0.32500335886067444,
      "grad_norm": 2.2152304649353027,
      "learning_rate": 3.420945117799265e-05,
      "loss": 4.3719,
      "step": 12095
    },
    {
      "epoch": 0.32513771328765284,
      "grad_norm": 2.3170926570892334,
      "learning_rate": 3.420264197194607e-05,
      "loss": 4.2828,
      "step": 12100
    },
    {
      "epoch": 0.3252720677146312,
      "grad_norm": 2.5880281925201416,
      "learning_rate": 3.41958327658995e-05,
      "loss": 4.3803,
      "step": 12105
    },
    {
      "epoch": 0.3254064221416096,
      "grad_norm": 2.205782175064087,
      "learning_rate": 3.418902355985292e-05,
      "loss": 4.3611,
      "step": 12110
    },
    {
      "epoch": 0.32554077656858793,
      "grad_norm": 2.1587607860565186,
      "learning_rate": 3.4182214353806344e-05,
      "loss": 4.4166,
      "step": 12115
    },
    {
      "epoch": 0.32567513099556633,
      "grad_norm": 2.41005539894104,
      "learning_rate": 3.417540514775977e-05,
      "loss": 4.4173,
      "step": 12120
    },
    {
      "epoch": 0.3258094854225447,
      "grad_norm": 2.3019816875457764,
      "learning_rate": 3.41685959417132e-05,
      "loss": 4.251,
      "step": 12125
    },
    {
      "epoch": 0.325943839849523,
      "grad_norm": 2.3289191722869873,
      "learning_rate": 3.4161786735666625e-05,
      "loss": 4.2742,
      "step": 12130
    },
    {
      "epoch": 0.3260781942765014,
      "grad_norm": 2.285503625869751,
      "learning_rate": 3.415497752962005e-05,
      "loss": 4.4542,
      "step": 12135
    },
    {
      "epoch": 0.32621254870347977,
      "grad_norm": 2.4369699954986572,
      "learning_rate": 3.414816832357347e-05,
      "loss": 4.4588,
      "step": 12140
    },
    {
      "epoch": 0.32634690313045817,
      "grad_norm": 2.3111672401428223,
      "learning_rate": 3.41413591175269e-05,
      "loss": 4.3306,
      "step": 12145
    },
    {
      "epoch": 0.3264812575574365,
      "grad_norm": 2.472026824951172,
      "learning_rate": 3.413454991148032e-05,
      "loss": 4.3568,
      "step": 12150
    },
    {
      "epoch": 0.3266156119844149,
      "grad_norm": 2.2053983211517334,
      "learning_rate": 3.412774070543375e-05,
      "loss": 4.4839,
      "step": 12155
    },
    {
      "epoch": 0.32674996641139326,
      "grad_norm": 2.419576406478882,
      "learning_rate": 3.412093149938717e-05,
      "loss": 4.3624,
      "step": 12160
    },
    {
      "epoch": 0.3268843208383716,
      "grad_norm": 2.6585793495178223,
      "learning_rate": 3.41141222933406e-05,
      "loss": 4.4714,
      "step": 12165
    },
    {
      "epoch": 0.32701867526535,
      "grad_norm": 2.3292903900146484,
      "learning_rate": 3.410731308729402e-05,
      "loss": 4.4811,
      "step": 12170
    },
    {
      "epoch": 0.32715302969232835,
      "grad_norm": 2.4216301441192627,
      "learning_rate": 3.4100503881247445e-05,
      "loss": 4.3285,
      "step": 12175
    },
    {
      "epoch": 0.32728738411930675,
      "grad_norm": 2.373793363571167,
      "learning_rate": 3.4093694675200874e-05,
      "loss": 4.3334,
      "step": 12180
    },
    {
      "epoch": 0.3274217385462851,
      "grad_norm": 2.598907709121704,
      "learning_rate": 3.40868854691543e-05,
      "loss": 4.4658,
      "step": 12185
    },
    {
      "epoch": 0.3275560929732635,
      "grad_norm": 2.5631680488586426,
      "learning_rate": 3.4080076263107726e-05,
      "loss": 4.3555,
      "step": 12190
    },
    {
      "epoch": 0.32769044740024184,
      "grad_norm": 2.4126274585723877,
      "learning_rate": 3.407326705706115e-05,
      "loss": 4.3303,
      "step": 12195
    },
    {
      "epoch": 0.3278248018272202,
      "grad_norm": 2.44565486907959,
      "learning_rate": 3.406645785101457e-05,
      "loss": 4.3847,
      "step": 12200
    },
    {
      "epoch": 0.3279591562541986,
      "grad_norm": 2.249321699142456,
      "learning_rate": 3.4059648644968e-05,
      "loss": 4.3886,
      "step": 12205
    },
    {
      "epoch": 0.3280935106811769,
      "grad_norm": 2.511990785598755,
      "learning_rate": 3.405283943892142e-05,
      "loss": 4.3684,
      "step": 12210
    },
    {
      "epoch": 0.3282278651081553,
      "grad_norm": 2.346261978149414,
      "learning_rate": 3.404603023287485e-05,
      "loss": 4.3515,
      "step": 12215
    },
    {
      "epoch": 0.32836221953513367,
      "grad_norm": 2.244824171066284,
      "learning_rate": 3.403922102682827e-05,
      "loss": 4.4045,
      "step": 12220
    },
    {
      "epoch": 0.32849657396211207,
      "grad_norm": 2.2844960689544678,
      "learning_rate": 3.40324118207817e-05,
      "loss": 4.2933,
      "step": 12225
    },
    {
      "epoch": 0.3286309283890904,
      "grad_norm": 2.235485553741455,
      "learning_rate": 3.4025602614735124e-05,
      "loss": 4.3726,
      "step": 12230
    },
    {
      "epoch": 0.3287652828160688,
      "grad_norm": 2.3331940174102783,
      "learning_rate": 3.4018793408688546e-05,
      "loss": 4.2503,
      "step": 12235
    },
    {
      "epoch": 0.32889963724304716,
      "grad_norm": 2.458235025405884,
      "learning_rate": 3.4011984202641975e-05,
      "loss": 4.5644,
      "step": 12240
    },
    {
      "epoch": 0.3290339916700255,
      "grad_norm": 2.502300262451172,
      "learning_rate": 3.40051749965954e-05,
      "loss": 4.4537,
      "step": 12245
    },
    {
      "epoch": 0.3291683460970039,
      "grad_norm": 2.342432737350464,
      "learning_rate": 3.399836579054883e-05,
      "loss": 4.2628,
      "step": 12250
    },
    {
      "epoch": 0.32930270052398225,
      "grad_norm": 2.379096746444702,
      "learning_rate": 3.399155658450225e-05,
      "loss": 4.3394,
      "step": 12255
    },
    {
      "epoch": 0.32943705495096065,
      "grad_norm": 2.4168946743011475,
      "learning_rate": 3.398474737845567e-05,
      "loss": 4.3453,
      "step": 12260
    },
    {
      "epoch": 0.329571409377939,
      "grad_norm": 2.3565354347229004,
      "learning_rate": 3.3977938172409094e-05,
      "loss": 4.351,
      "step": 12265
    },
    {
      "epoch": 0.3297057638049174,
      "grad_norm": 2.2729830741882324,
      "learning_rate": 3.397112896636252e-05,
      "loss": 4.2997,
      "step": 12270
    },
    {
      "epoch": 0.32984011823189574,
      "grad_norm": 2.5583789348602295,
      "learning_rate": 3.396431976031595e-05,
      "loss": 4.3637,
      "step": 12275
    },
    {
      "epoch": 0.3299744726588741,
      "grad_norm": 2.4167089462280273,
      "learning_rate": 3.3957510554269374e-05,
      "loss": 4.3675,
      "step": 12280
    },
    {
      "epoch": 0.3301088270858525,
      "grad_norm": 2.198096513748169,
      "learning_rate": 3.3950701348222796e-05,
      "loss": 4.2933,
      "step": 12285
    },
    {
      "epoch": 0.33024318151283083,
      "grad_norm": 2.355056047439575,
      "learning_rate": 3.3943892142176225e-05,
      "loss": 4.3765,
      "step": 12290
    },
    {
      "epoch": 0.33037753593980923,
      "grad_norm": 2.283850908279419,
      "learning_rate": 3.393708293612965e-05,
      "loss": 4.3287,
      "step": 12295
    },
    {
      "epoch": 0.3305118903667876,
      "grad_norm": 2.4181854724884033,
      "learning_rate": 3.393027373008307e-05,
      "loss": 4.5121,
      "step": 12300
    },
    {
      "epoch": 0.330646244793766,
      "grad_norm": 2.1839256286621094,
      "learning_rate": 3.39234645240365e-05,
      "loss": 4.2679,
      "step": 12305
    },
    {
      "epoch": 0.3307805992207443,
      "grad_norm": 2.5157525539398193,
      "learning_rate": 3.391665531798993e-05,
      "loss": 4.4237,
      "step": 12310
    },
    {
      "epoch": 0.33091495364772266,
      "grad_norm": 2.530059814453125,
      "learning_rate": 3.390984611194335e-05,
      "loss": 4.4203,
      "step": 12315
    },
    {
      "epoch": 0.33104930807470107,
      "grad_norm": 2.445878267288208,
      "learning_rate": 3.390303690589677e-05,
      "loss": 4.2602,
      "step": 12320
    },
    {
      "epoch": 0.3311836625016794,
      "grad_norm": 2.3912410736083984,
      "learning_rate": 3.3896227699850195e-05,
      "loss": 4.4294,
      "step": 12325
    },
    {
      "epoch": 0.3313180169286578,
      "grad_norm": 2.694054126739502,
      "learning_rate": 3.3889418493803624e-05,
      "loss": 4.3965,
      "step": 12330
    },
    {
      "epoch": 0.33145237135563616,
      "grad_norm": 2.2209055423736572,
      "learning_rate": 3.388260928775705e-05,
      "loss": 4.3231,
      "step": 12335
    },
    {
      "epoch": 0.33158672578261456,
      "grad_norm": 2.3513989448547363,
      "learning_rate": 3.3875800081710475e-05,
      "loss": 4.2444,
      "step": 12340
    },
    {
      "epoch": 0.3317210802095929,
      "grad_norm": 2.4620614051818848,
      "learning_rate": 3.38689908756639e-05,
      "loss": 4.4604,
      "step": 12345
    },
    {
      "epoch": 0.3318554346365713,
      "grad_norm": 2.3156909942626953,
      "learning_rate": 3.3862181669617326e-05,
      "loss": 4.3777,
      "step": 12350
    },
    {
      "epoch": 0.33198978906354965,
      "grad_norm": 2.2513797283172607,
      "learning_rate": 3.385537246357075e-05,
      "loss": 4.3868,
      "step": 12355
    },
    {
      "epoch": 0.332124143490528,
      "grad_norm": 2.136876106262207,
      "learning_rate": 3.384856325752417e-05,
      "loss": 4.3392,
      "step": 12360
    },
    {
      "epoch": 0.3322584979175064,
      "grad_norm": 2.2603936195373535,
      "learning_rate": 3.38417540514776e-05,
      "loss": 4.2622,
      "step": 12365
    },
    {
      "epoch": 0.33239285234448473,
      "grad_norm": 2.3585004806518555,
      "learning_rate": 3.383494484543103e-05,
      "loss": 4.4019,
      "step": 12370
    },
    {
      "epoch": 0.33252720677146314,
      "grad_norm": 2.2861709594726562,
      "learning_rate": 3.382813563938445e-05,
      "loss": 4.341,
      "step": 12375
    },
    {
      "epoch": 0.3326615611984415,
      "grad_norm": 2.525719165802002,
      "learning_rate": 3.3821326433337874e-05,
      "loss": 4.3403,
      "step": 12380
    },
    {
      "epoch": 0.3327959156254199,
      "grad_norm": 2.331448793411255,
      "learning_rate": 3.3814517227291296e-05,
      "loss": 4.3904,
      "step": 12385
    },
    {
      "epoch": 0.3329302700523982,
      "grad_norm": 2.2829172611236572,
      "learning_rate": 3.3807708021244725e-05,
      "loss": 4.3851,
      "step": 12390
    },
    {
      "epoch": 0.33306462447937657,
      "grad_norm": 2.4504599571228027,
      "learning_rate": 3.3800898815198154e-05,
      "loss": 4.4037,
      "step": 12395
    },
    {
      "epoch": 0.33319897890635497,
      "grad_norm": 2.14096999168396,
      "learning_rate": 3.3794089609151576e-05,
      "loss": 4.2153,
      "step": 12400
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 2.502445697784424,
      "learning_rate": 3.3787280403105e-05,
      "loss": 4.3228,
      "step": 12405
    },
    {
      "epoch": 0.3334676877603117,
      "grad_norm": 2.3934991359710693,
      "learning_rate": 3.378047119705842e-05,
      "loss": 4.3316,
      "step": 12410
    },
    {
      "epoch": 0.33360204218729006,
      "grad_norm": 2.35772705078125,
      "learning_rate": 3.377366199101185e-05,
      "loss": 4.3413,
      "step": 12415
    },
    {
      "epoch": 0.33373639661426846,
      "grad_norm": 2.1789119243621826,
      "learning_rate": 3.376685278496527e-05,
      "loss": 4.4542,
      "step": 12420
    },
    {
      "epoch": 0.3338707510412468,
      "grad_norm": 2.1237103939056396,
      "learning_rate": 3.37600435789187e-05,
      "loss": 4.1719,
      "step": 12425
    },
    {
      "epoch": 0.3340051054682252,
      "grad_norm": 2.3214142322540283,
      "learning_rate": 3.375323437287212e-05,
      "loss": 4.3039,
      "step": 12430
    },
    {
      "epoch": 0.33413945989520355,
      "grad_norm": 2.4996509552001953,
      "learning_rate": 3.374642516682555e-05,
      "loss": 4.4156,
      "step": 12435
    },
    {
      "epoch": 0.3342738143221819,
      "grad_norm": 2.3055732250213623,
      "learning_rate": 3.3739615960778975e-05,
      "loss": 4.3311,
      "step": 12440
    },
    {
      "epoch": 0.3344081687491603,
      "grad_norm": 2.5904111862182617,
      "learning_rate": 3.37328067547324e-05,
      "loss": 4.2898,
      "step": 12445
    },
    {
      "epoch": 0.33454252317613864,
      "grad_norm": 2.240269422531128,
      "learning_rate": 3.3725997548685826e-05,
      "loss": 4.3548,
      "step": 12450
    },
    {
      "epoch": 0.33467687760311704,
      "grad_norm": 2.281118869781494,
      "learning_rate": 3.3719188342639255e-05,
      "loss": 4.4745,
      "step": 12455
    },
    {
      "epoch": 0.3348112320300954,
      "grad_norm": 2.4933812618255615,
      "learning_rate": 3.371237913659268e-05,
      "loss": 4.4069,
      "step": 12460
    },
    {
      "epoch": 0.3349455864570738,
      "grad_norm": 2.5674359798431396,
      "learning_rate": 3.37055699305461e-05,
      "loss": 4.2655,
      "step": 12465
    },
    {
      "epoch": 0.33507994088405213,
      "grad_norm": 2.341557025909424,
      "learning_rate": 3.369876072449952e-05,
      "loss": 4.3043,
      "step": 12470
    },
    {
      "epoch": 0.3352142953110305,
      "grad_norm": 2.2489495277404785,
      "learning_rate": 3.369195151845295e-05,
      "loss": 4.3071,
      "step": 12475
    },
    {
      "epoch": 0.3353486497380089,
      "grad_norm": 2.3209519386291504,
      "learning_rate": 3.368514231240637e-05,
      "loss": 4.4017,
      "step": 12480
    },
    {
      "epoch": 0.3354830041649872,
      "grad_norm": 2.4912731647491455,
      "learning_rate": 3.36783331063598e-05,
      "loss": 4.1886,
      "step": 12485
    },
    {
      "epoch": 0.3356173585919656,
      "grad_norm": 2.1422030925750732,
      "learning_rate": 3.3671523900313224e-05,
      "loss": 4.2687,
      "step": 12490
    },
    {
      "epoch": 0.33575171301894396,
      "grad_norm": 2.25996732711792,
      "learning_rate": 3.3664714694266653e-05,
      "loss": 4.3919,
      "step": 12495
    },
    {
      "epoch": 0.33588606744592236,
      "grad_norm": 2.4497909545898438,
      "learning_rate": 3.3657905488220076e-05,
      "loss": 4.3657,
      "step": 12500
    },
    {
      "epoch": 0.3360204218729007,
      "grad_norm": 2.573125123977661,
      "learning_rate": 3.36510962821735e-05,
      "loss": 4.514,
      "step": 12505
    },
    {
      "epoch": 0.33615477629987905,
      "grad_norm": 2.1329214572906494,
      "learning_rate": 3.364428707612692e-05,
      "loss": 4.2861,
      "step": 12510
    },
    {
      "epoch": 0.33628913072685745,
      "grad_norm": 2.522408962249756,
      "learning_rate": 3.3637477870080356e-05,
      "loss": 4.4201,
      "step": 12515
    },
    {
      "epoch": 0.3364234851538358,
      "grad_norm": 2.377274990081787,
      "learning_rate": 3.363066866403378e-05,
      "loss": 4.4027,
      "step": 12520
    },
    {
      "epoch": 0.3365578395808142,
      "grad_norm": 2.287916660308838,
      "learning_rate": 3.36238594579872e-05,
      "loss": 4.3578,
      "step": 12525
    },
    {
      "epoch": 0.33669219400779254,
      "grad_norm": 2.212514638900757,
      "learning_rate": 3.361705025194062e-05,
      "loss": 4.3872,
      "step": 12530
    },
    {
      "epoch": 0.33682654843477094,
      "grad_norm": 2.34199857711792,
      "learning_rate": 3.361024104589405e-05,
      "loss": 4.3825,
      "step": 12535
    },
    {
      "epoch": 0.3369609028617493,
      "grad_norm": 2.276085376739502,
      "learning_rate": 3.3603431839847474e-05,
      "loss": 4.2627,
      "step": 12540
    },
    {
      "epoch": 0.3370952572887277,
      "grad_norm": 2.4497063159942627,
      "learning_rate": 3.35966226338009e-05,
      "loss": 4.3662,
      "step": 12545
    },
    {
      "epoch": 0.33722961171570603,
      "grad_norm": 2.360079050064087,
      "learning_rate": 3.3589813427754326e-05,
      "loss": 4.3821,
      "step": 12550
    },
    {
      "epoch": 0.3373639661426844,
      "grad_norm": 2.5171477794647217,
      "learning_rate": 3.358300422170775e-05,
      "loss": 4.3701,
      "step": 12555
    },
    {
      "epoch": 0.3374983205696628,
      "grad_norm": 2.528137445449829,
      "learning_rate": 3.357619501566118e-05,
      "loss": 4.4686,
      "step": 12560
    },
    {
      "epoch": 0.3376326749966411,
      "grad_norm": 2.028557300567627,
      "learning_rate": 3.35693858096146e-05,
      "loss": 4.4379,
      "step": 12565
    },
    {
      "epoch": 0.3377670294236195,
      "grad_norm": 2.183546543121338,
      "learning_rate": 3.356257660356802e-05,
      "loss": 4.2262,
      "step": 12570
    },
    {
      "epoch": 0.33790138385059787,
      "grad_norm": 2.586716890335083,
      "learning_rate": 3.355576739752145e-05,
      "loss": 4.2206,
      "step": 12575
    },
    {
      "epoch": 0.33803573827757627,
      "grad_norm": 2.306382656097412,
      "learning_rate": 3.354895819147488e-05,
      "loss": 4.3338,
      "step": 12580
    },
    {
      "epoch": 0.3381700927045546,
      "grad_norm": 2.430414915084839,
      "learning_rate": 3.35421489854283e-05,
      "loss": 4.4614,
      "step": 12585
    },
    {
      "epoch": 0.33830444713153296,
      "grad_norm": 2.2400527000427246,
      "learning_rate": 3.3535339779381724e-05,
      "loss": 4.4942,
      "step": 12590
    },
    {
      "epoch": 0.33843880155851136,
      "grad_norm": 2.146414279937744,
      "learning_rate": 3.3528530573335146e-05,
      "loss": 4.3551,
      "step": 12595
    },
    {
      "epoch": 0.3385731559854897,
      "grad_norm": 2.246049404144287,
      "learning_rate": 3.3521721367288575e-05,
      "loss": 4.3162,
      "step": 12600
    },
    {
      "epoch": 0.3387075104124681,
      "grad_norm": 2.318774700164795,
      "learning_rate": 3.3514912161242004e-05,
      "loss": 4.3223,
      "step": 12605
    },
    {
      "epoch": 0.33884186483944645,
      "grad_norm": 2.380481004714966,
      "learning_rate": 3.350810295519543e-05,
      "loss": 4.4088,
      "step": 12610
    },
    {
      "epoch": 0.33897621926642485,
      "grad_norm": 2.5404300689697266,
      "learning_rate": 3.350129374914885e-05,
      "loss": 4.3706,
      "step": 12615
    },
    {
      "epoch": 0.3391105736934032,
      "grad_norm": 2.256519079208374,
      "learning_rate": 3.349448454310228e-05,
      "loss": 4.2743,
      "step": 12620
    },
    {
      "epoch": 0.33924492812038154,
      "grad_norm": 2.2663304805755615,
      "learning_rate": 3.34876753370557e-05,
      "loss": 4.2736,
      "step": 12625
    },
    {
      "epoch": 0.33937928254735994,
      "grad_norm": 2.308063268661499,
      "learning_rate": 3.348086613100912e-05,
      "loss": 4.4443,
      "step": 12630
    },
    {
      "epoch": 0.3395136369743383,
      "grad_norm": 2.628347635269165,
      "learning_rate": 3.347405692496255e-05,
      "loss": 4.3288,
      "step": 12635
    },
    {
      "epoch": 0.3396479914013167,
      "grad_norm": 2.6036171913146973,
      "learning_rate": 3.346724771891598e-05,
      "loss": 4.373,
      "step": 12640
    },
    {
      "epoch": 0.33978234582829503,
      "grad_norm": 2.3996665477752686,
      "learning_rate": 3.34604385128694e-05,
      "loss": 4.291,
      "step": 12645
    },
    {
      "epoch": 0.33991670025527343,
      "grad_norm": 2.418679714202881,
      "learning_rate": 3.3453629306822825e-05,
      "loss": 4.3245,
      "step": 12650
    },
    {
      "epoch": 0.3400510546822518,
      "grad_norm": 2.441160202026367,
      "learning_rate": 3.344682010077625e-05,
      "loss": 4.2828,
      "step": 12655
    },
    {
      "epoch": 0.3401854091092302,
      "grad_norm": 2.4043376445770264,
      "learning_rate": 3.3440010894729676e-05,
      "loss": 4.3061,
      "step": 12660
    },
    {
      "epoch": 0.3403197635362085,
      "grad_norm": 2.0237512588500977,
      "learning_rate": 3.3433201688683106e-05,
      "loss": 4.2617,
      "step": 12665
    },
    {
      "epoch": 0.34045411796318686,
      "grad_norm": 2.437177896499634,
      "learning_rate": 3.342639248263653e-05,
      "loss": 4.327,
      "step": 12670
    },
    {
      "epoch": 0.34058847239016526,
      "grad_norm": 2.318439483642578,
      "learning_rate": 3.341958327658995e-05,
      "loss": 4.3084,
      "step": 12675
    },
    {
      "epoch": 0.3407228268171436,
      "grad_norm": 2.2808539867401123,
      "learning_rate": 3.341277407054338e-05,
      "loss": 4.3637,
      "step": 12680
    },
    {
      "epoch": 0.340857181244122,
      "grad_norm": 2.3590006828308105,
      "learning_rate": 3.34059648644968e-05,
      "loss": 4.3297,
      "step": 12685
    },
    {
      "epoch": 0.34099153567110035,
      "grad_norm": 2.226768732070923,
      "learning_rate": 3.3399155658450224e-05,
      "loss": 4.2035,
      "step": 12690
    },
    {
      "epoch": 0.34112589009807875,
      "grad_norm": 2.8055317401885986,
      "learning_rate": 3.339234645240365e-05,
      "loss": 4.3896,
      "step": 12695
    },
    {
      "epoch": 0.3412602445250571,
      "grad_norm": 2.292235851287842,
      "learning_rate": 3.3385537246357075e-05,
      "loss": 4.413,
      "step": 12700
    },
    {
      "epoch": 0.34139459895203544,
      "grad_norm": 2.29434871673584,
      "learning_rate": 3.3378728040310504e-05,
      "loss": 4.3033,
      "step": 12705
    },
    {
      "epoch": 0.34152895337901384,
      "grad_norm": 2.349175214767456,
      "learning_rate": 3.3371918834263926e-05,
      "loss": 4.4828,
      "step": 12710
    },
    {
      "epoch": 0.3416633078059922,
      "grad_norm": 2.490933656692505,
      "learning_rate": 3.336510962821735e-05,
      "loss": 4.2965,
      "step": 12715
    },
    {
      "epoch": 0.3417976622329706,
      "grad_norm": 2.208439588546753,
      "learning_rate": 3.335830042217077e-05,
      "loss": 4.2578,
      "step": 12720
    },
    {
      "epoch": 0.34193201665994893,
      "grad_norm": 2.362440586090088,
      "learning_rate": 3.335149121612421e-05,
      "loss": 4.2685,
      "step": 12725
    },
    {
      "epoch": 0.34206637108692733,
      "grad_norm": 2.432011127471924,
      "learning_rate": 3.334468201007763e-05,
      "loss": 4.3871,
      "step": 12730
    },
    {
      "epoch": 0.3422007255139057,
      "grad_norm": 2.506363868713379,
      "learning_rate": 3.333787280403105e-05,
      "loss": 4.2423,
      "step": 12735
    },
    {
      "epoch": 0.3423350799408841,
      "grad_norm": 2.4747965335845947,
      "learning_rate": 3.3331063597984473e-05,
      "loss": 4.3385,
      "step": 12740
    },
    {
      "epoch": 0.3424694343678624,
      "grad_norm": 2.325584650039673,
      "learning_rate": 3.33242543919379e-05,
      "loss": 4.3641,
      "step": 12745
    },
    {
      "epoch": 0.34260378879484077,
      "grad_norm": 2.3004395961761475,
      "learning_rate": 3.3317445185891325e-05,
      "loss": 4.2633,
      "step": 12750
    },
    {
      "epoch": 0.34273814322181917,
      "grad_norm": 2.0634853839874268,
      "learning_rate": 3.3310635979844754e-05,
      "loss": 4.3731,
      "step": 12755
    },
    {
      "epoch": 0.3428724976487975,
      "grad_norm": 2.610539436340332,
      "learning_rate": 3.3303826773798176e-05,
      "loss": 4.3725,
      "step": 12760
    },
    {
      "epoch": 0.3430068520757759,
      "grad_norm": 2.309396266937256,
      "learning_rate": 3.3297017567751605e-05,
      "loss": 4.3448,
      "step": 12765
    },
    {
      "epoch": 0.34314120650275426,
      "grad_norm": 2.2891199588775635,
      "learning_rate": 3.329020836170503e-05,
      "loss": 4.3961,
      "step": 12770
    },
    {
      "epoch": 0.34327556092973266,
      "grad_norm": 2.1714365482330322,
      "learning_rate": 3.328339915565845e-05,
      "loss": 4.2788,
      "step": 12775
    },
    {
      "epoch": 0.343409915356711,
      "grad_norm": 2.311539888381958,
      "learning_rate": 3.327658994961187e-05,
      "loss": 4.2987,
      "step": 12780
    },
    {
      "epoch": 0.34354426978368935,
      "grad_norm": 2.3169705867767334,
      "learning_rate": 3.32697807435653e-05,
      "loss": 4.2607,
      "step": 12785
    },
    {
      "epoch": 0.34367862421066775,
      "grad_norm": 2.112238645553589,
      "learning_rate": 3.326297153751873e-05,
      "loss": 4.2827,
      "step": 12790
    },
    {
      "epoch": 0.3438129786376461,
      "grad_norm": 2.3358473777770996,
      "learning_rate": 3.325616233147215e-05,
      "loss": 4.323,
      "step": 12795
    },
    {
      "epoch": 0.3439473330646245,
      "grad_norm": 2.499443292617798,
      "learning_rate": 3.3249353125425575e-05,
      "loss": 4.2559,
      "step": 12800
    },
    {
      "epoch": 0.34408168749160284,
      "grad_norm": 2.4832966327667236,
      "learning_rate": 3.3242543919379004e-05,
      "loss": 4.41,
      "step": 12805
    },
    {
      "epoch": 0.34421604191858124,
      "grad_norm": 2.3134751319885254,
      "learning_rate": 3.3235734713332426e-05,
      "loss": 4.3617,
      "step": 12810
    },
    {
      "epoch": 0.3443503963455596,
      "grad_norm": 2.6137137413024902,
      "learning_rate": 3.3228925507285855e-05,
      "loss": 4.1836,
      "step": 12815
    },
    {
      "epoch": 0.34448475077253793,
      "grad_norm": 2.100457191467285,
      "learning_rate": 3.322211630123928e-05,
      "loss": 4.2716,
      "step": 12820
    },
    {
      "epoch": 0.34461910519951633,
      "grad_norm": 2.5705742835998535,
      "learning_rate": 3.3215307095192706e-05,
      "loss": 4.3527,
      "step": 12825
    },
    {
      "epoch": 0.3447534596264947,
      "grad_norm": 2.684929132461548,
      "learning_rate": 3.320849788914613e-05,
      "loss": 4.2143,
      "step": 12830
    },
    {
      "epoch": 0.3448878140534731,
      "grad_norm": 2.2381880283355713,
      "learning_rate": 3.320168868309955e-05,
      "loss": 4.4224,
      "step": 12835
    },
    {
      "epoch": 0.3450221684804514,
      "grad_norm": 2.333702325820923,
      "learning_rate": 3.319487947705297e-05,
      "loss": 4.2755,
      "step": 12840
    },
    {
      "epoch": 0.3451565229074298,
      "grad_norm": 2.3880534172058105,
      "learning_rate": 3.31880702710064e-05,
      "loss": 4.2388,
      "step": 12845
    },
    {
      "epoch": 0.34529087733440816,
      "grad_norm": 2.6067099571228027,
      "learning_rate": 3.318126106495983e-05,
      "loss": 4.3602,
      "step": 12850
    },
    {
      "epoch": 0.34542523176138656,
      "grad_norm": 2.1584560871124268,
      "learning_rate": 3.317445185891325e-05,
      "loss": 4.3264,
      "step": 12855
    },
    {
      "epoch": 0.3455595861883649,
      "grad_norm": 2.2544124126434326,
      "learning_rate": 3.3167642652866676e-05,
      "loss": 4.3131,
      "step": 12860
    },
    {
      "epoch": 0.34569394061534325,
      "grad_norm": 2.511993885040283,
      "learning_rate": 3.31608334468201e-05,
      "loss": 4.2717,
      "step": 12865
    },
    {
      "epoch": 0.34582829504232165,
      "grad_norm": 2.2463836669921875,
      "learning_rate": 3.315402424077353e-05,
      "loss": 4.3676,
      "step": 12870
    },
    {
      "epoch": 0.3459626494693,
      "grad_norm": 2.109424114227295,
      "learning_rate": 3.3147215034726956e-05,
      "loss": 4.3602,
      "step": 12875
    },
    {
      "epoch": 0.3460970038962784,
      "grad_norm": 2.3144948482513428,
      "learning_rate": 3.314040582868038e-05,
      "loss": 4.4366,
      "step": 12880
    },
    {
      "epoch": 0.34623135832325674,
      "grad_norm": 2.4145798683166504,
      "learning_rate": 3.31335966226338e-05,
      "loss": 4.2313,
      "step": 12885
    },
    {
      "epoch": 0.34636571275023514,
      "grad_norm": 2.1766817569732666,
      "learning_rate": 3.312678741658723e-05,
      "loss": 4.4385,
      "step": 12890
    },
    {
      "epoch": 0.3465000671772135,
      "grad_norm": 2.543679714202881,
      "learning_rate": 3.311997821054065e-05,
      "loss": 4.3023,
      "step": 12895
    },
    {
      "epoch": 0.34663442160419183,
      "grad_norm": 2.249096632003784,
      "learning_rate": 3.3113169004494074e-05,
      "loss": 4.2675,
      "step": 12900
    },
    {
      "epoch": 0.34676877603117023,
      "grad_norm": 2.3533482551574707,
      "learning_rate": 3.31063597984475e-05,
      "loss": 4.3446,
      "step": 12905
    },
    {
      "epoch": 0.3469031304581486,
      "grad_norm": 2.6078922748565674,
      "learning_rate": 3.309955059240093e-05,
      "loss": 4.4185,
      "step": 12910
    },
    {
      "epoch": 0.347037484885127,
      "grad_norm": 2.3863863945007324,
      "learning_rate": 3.3092741386354354e-05,
      "loss": 4.3962,
      "step": 12915
    },
    {
      "epoch": 0.3471718393121053,
      "grad_norm": 2.592764377593994,
      "learning_rate": 3.308593218030778e-05,
      "loss": 4.3918,
      "step": 12920
    },
    {
      "epoch": 0.3473061937390837,
      "grad_norm": 2.2816545963287354,
      "learning_rate": 3.30791229742612e-05,
      "loss": 4.4112,
      "step": 12925
    },
    {
      "epoch": 0.34744054816606207,
      "grad_norm": 2.369417190551758,
      "learning_rate": 3.307231376821463e-05,
      "loss": 4.2738,
      "step": 12930
    },
    {
      "epoch": 0.3475749025930404,
      "grad_norm": 2.3674509525299072,
      "learning_rate": 3.306550456216805e-05,
      "loss": 4.3954,
      "step": 12935
    },
    {
      "epoch": 0.3477092570200188,
      "grad_norm": 2.3611881732940674,
      "learning_rate": 3.305869535612148e-05,
      "loss": 4.3391,
      "step": 12940
    },
    {
      "epoch": 0.34784361144699716,
      "grad_norm": 2.3024861812591553,
      "learning_rate": 3.30518861500749e-05,
      "loss": 4.4389,
      "step": 12945
    },
    {
      "epoch": 0.34797796587397556,
      "grad_norm": 2.271338701248169,
      "learning_rate": 3.304507694402833e-05,
      "loss": 4.3156,
      "step": 12950
    },
    {
      "epoch": 0.3481123203009539,
      "grad_norm": 2.339171886444092,
      "learning_rate": 3.303826773798175e-05,
      "loss": 4.4184,
      "step": 12955
    },
    {
      "epoch": 0.3482466747279323,
      "grad_norm": 2.6277692317962646,
      "learning_rate": 3.3031458531935175e-05,
      "loss": 4.3438,
      "step": 12960
    },
    {
      "epoch": 0.34838102915491065,
      "grad_norm": 2.1784210205078125,
      "learning_rate": 3.3024649325888604e-05,
      "loss": 4.3129,
      "step": 12965
    },
    {
      "epoch": 0.34851538358188905,
      "grad_norm": 2.840534210205078,
      "learning_rate": 3.301784011984203e-05,
      "loss": 4.3633,
      "step": 12970
    },
    {
      "epoch": 0.3486497380088674,
      "grad_norm": 2.3672585487365723,
      "learning_rate": 3.3011030913795456e-05,
      "loss": 4.2966,
      "step": 12975
    },
    {
      "epoch": 0.34878409243584574,
      "grad_norm": 2.3058366775512695,
      "learning_rate": 3.300422170774888e-05,
      "loss": 4.2895,
      "step": 12980
    },
    {
      "epoch": 0.34891844686282414,
      "grad_norm": 2.5000271797180176,
      "learning_rate": 3.29974125017023e-05,
      "loss": 4.325,
      "step": 12985
    },
    {
      "epoch": 0.3490528012898025,
      "grad_norm": 2.2673470973968506,
      "learning_rate": 3.299060329565572e-05,
      "loss": 4.3071,
      "step": 12990
    },
    {
      "epoch": 0.3491871557167809,
      "grad_norm": 2.4056758880615234,
      "learning_rate": 3.298379408960915e-05,
      "loss": 4.2686,
      "step": 12995
    },
    {
      "epoch": 0.34932151014375923,
      "grad_norm": 2.2597715854644775,
      "learning_rate": 3.297698488356258e-05,
      "loss": 4.2819,
      "step": 13000
    },
    {
      "epoch": 0.34945586457073763,
      "grad_norm": 2.7082595825195312,
      "learning_rate": 3.2970175677516e-05,
      "loss": 4.2946,
      "step": 13005
    },
    {
      "epoch": 0.349590218997716,
      "grad_norm": 2.4204490184783936,
      "learning_rate": 3.2963366471469425e-05,
      "loss": 4.359,
      "step": 13010
    },
    {
      "epoch": 0.3497245734246943,
      "grad_norm": 2.5691988468170166,
      "learning_rate": 3.2956557265422854e-05,
      "loss": 4.3456,
      "step": 13015
    },
    {
      "epoch": 0.3498589278516727,
      "grad_norm": 2.4026260375976562,
      "learning_rate": 3.2949748059376276e-05,
      "loss": 4.3767,
      "step": 13020
    },
    {
      "epoch": 0.34999328227865106,
      "grad_norm": 2.2824840545654297,
      "learning_rate": 3.2942938853329705e-05,
      "loss": 4.3572,
      "step": 13025
    },
    {
      "epoch": 0.35012763670562946,
      "grad_norm": 2.3221051692962646,
      "learning_rate": 3.293612964728313e-05,
      "loss": 4.3843,
      "step": 13030
    },
    {
      "epoch": 0.3502619911326078,
      "grad_norm": 2.3892171382904053,
      "learning_rate": 3.292932044123656e-05,
      "loss": 4.2276,
      "step": 13035
    },
    {
      "epoch": 0.3503963455595862,
      "grad_norm": 2.467667579650879,
      "learning_rate": 3.292251123518998e-05,
      "loss": 4.3588,
      "step": 13040
    },
    {
      "epoch": 0.35053069998656455,
      "grad_norm": 2.259399175643921,
      "learning_rate": 3.29157020291434e-05,
      "loss": 4.2737,
      "step": 13045
    },
    {
      "epoch": 0.35066505441354295,
      "grad_norm": 2.447152853012085,
      "learning_rate": 3.2908892823096824e-05,
      "loss": 4.3613,
      "step": 13050
    },
    {
      "epoch": 0.3507994088405213,
      "grad_norm": 2.2995541095733643,
      "learning_rate": 3.290208361705025e-05,
      "loss": 4.2341,
      "step": 13055
    },
    {
      "epoch": 0.35093376326749964,
      "grad_norm": 2.4681267738342285,
      "learning_rate": 3.289527441100368e-05,
      "loss": 4.3615,
      "step": 13060
    },
    {
      "epoch": 0.35106811769447804,
      "grad_norm": 2.305631160736084,
      "learning_rate": 3.2888465204957104e-05,
      "loss": 4.2977,
      "step": 13065
    },
    {
      "epoch": 0.3512024721214564,
      "grad_norm": 2.4446351528167725,
      "learning_rate": 3.2881655998910526e-05,
      "loss": 4.271,
      "step": 13070
    },
    {
      "epoch": 0.3513368265484348,
      "grad_norm": 2.445784330368042,
      "learning_rate": 3.2874846792863955e-05,
      "loss": 4.3712,
      "step": 13075
    },
    {
      "epoch": 0.35147118097541313,
      "grad_norm": 2.290306806564331,
      "learning_rate": 3.286803758681738e-05,
      "loss": 4.3611,
      "step": 13080
    },
    {
      "epoch": 0.35160553540239153,
      "grad_norm": 2.186196804046631,
      "learning_rate": 3.2861228380770807e-05,
      "loss": 4.3724,
      "step": 13085
    },
    {
      "epoch": 0.3517398898293699,
      "grad_norm": 2.5417280197143555,
      "learning_rate": 3.285441917472423e-05,
      "loss": 4.3885,
      "step": 13090
    },
    {
      "epoch": 0.3518742442563482,
      "grad_norm": 2.3365049362182617,
      "learning_rate": 3.284760996867766e-05,
      "loss": 4.4129,
      "step": 13095
    },
    {
      "epoch": 0.3520085986833266,
      "grad_norm": 2.5200552940368652,
      "learning_rate": 3.284080076263108e-05,
      "loss": 4.3778,
      "step": 13100
    },
    {
      "epoch": 0.35214295311030497,
      "grad_norm": 2.069540023803711,
      "learning_rate": 3.28339915565845e-05,
      "loss": 4.2942,
      "step": 13105
    },
    {
      "epoch": 0.35227730753728337,
      "grad_norm": 2.589008331298828,
      "learning_rate": 3.2827182350537925e-05,
      "loss": 4.2585,
      "step": 13110
    },
    {
      "epoch": 0.3524116619642617,
      "grad_norm": 2.5103719234466553,
      "learning_rate": 3.2820373144491354e-05,
      "loss": 4.3366,
      "step": 13115
    },
    {
      "epoch": 0.3525460163912401,
      "grad_norm": 2.3338003158569336,
      "learning_rate": 3.281356393844478e-05,
      "loss": 4.2475,
      "step": 13120
    },
    {
      "epoch": 0.35268037081821846,
      "grad_norm": 2.2297842502593994,
      "learning_rate": 3.2806754732398205e-05,
      "loss": 4.3412,
      "step": 13125
    },
    {
      "epoch": 0.3528147252451968,
      "grad_norm": 2.464806318283081,
      "learning_rate": 3.279994552635163e-05,
      "loss": 4.3853,
      "step": 13130
    },
    {
      "epoch": 0.3529490796721752,
      "grad_norm": 2.184504270553589,
      "learning_rate": 3.279313632030505e-05,
      "loss": 4.3079,
      "step": 13135
    },
    {
      "epoch": 0.35308343409915355,
      "grad_norm": 2.5422070026397705,
      "learning_rate": 3.278632711425848e-05,
      "loss": 4.47,
      "step": 13140
    },
    {
      "epoch": 0.35321778852613195,
      "grad_norm": 2.4231884479522705,
      "learning_rate": 3.27795179082119e-05,
      "loss": 4.3651,
      "step": 13145
    },
    {
      "epoch": 0.3533521429531103,
      "grad_norm": 2.534411668777466,
      "learning_rate": 3.277270870216533e-05,
      "loss": 4.3518,
      "step": 13150
    },
    {
      "epoch": 0.3534864973800887,
      "grad_norm": 2.259965419769287,
      "learning_rate": 3.276589949611875e-05,
      "loss": 4.2198,
      "step": 13155
    },
    {
      "epoch": 0.35362085180706704,
      "grad_norm": 2.5303735733032227,
      "learning_rate": 3.275909029007218e-05,
      "loss": 4.2907,
      "step": 13160
    },
    {
      "epoch": 0.35375520623404544,
      "grad_norm": 2.239152431488037,
      "learning_rate": 3.2752281084025603e-05,
      "loss": 4.2998,
      "step": 13165
    },
    {
      "epoch": 0.3538895606610238,
      "grad_norm": 2.473051071166992,
      "learning_rate": 3.2745471877979026e-05,
      "loss": 4.2962,
      "step": 13170
    },
    {
      "epoch": 0.35402391508800213,
      "grad_norm": 2.3368377685546875,
      "learning_rate": 3.2738662671932455e-05,
      "loss": 4.1866,
      "step": 13175
    },
    {
      "epoch": 0.35415826951498053,
      "grad_norm": 2.282496929168701,
      "learning_rate": 3.2731853465885884e-05,
      "loss": 4.367,
      "step": 13180
    },
    {
      "epoch": 0.3542926239419589,
      "grad_norm": 2.385413646697998,
      "learning_rate": 3.2725044259839306e-05,
      "loss": 4.343,
      "step": 13185
    },
    {
      "epoch": 0.3544269783689373,
      "grad_norm": 2.589935064315796,
      "learning_rate": 3.271823505379273e-05,
      "loss": 4.4793,
      "step": 13190
    },
    {
      "epoch": 0.3545613327959156,
      "grad_norm": 2.264702558517456,
      "learning_rate": 3.271142584774615e-05,
      "loss": 4.4538,
      "step": 13195
    },
    {
      "epoch": 0.354695687222894,
      "grad_norm": 2.366159439086914,
      "learning_rate": 3.270461664169958e-05,
      "loss": 4.3138,
      "step": 13200
    },
    {
      "epoch": 0.35483004164987236,
      "grad_norm": 2.3087618350982666,
      "learning_rate": 3.2697807435653e-05,
      "loss": 4.3036,
      "step": 13205
    },
    {
      "epoch": 0.3549643960768507,
      "grad_norm": 2.1213228702545166,
      "learning_rate": 3.269099822960643e-05,
      "loss": 4.2427,
      "step": 13210
    },
    {
      "epoch": 0.3550987505038291,
      "grad_norm": 2.2862708568573,
      "learning_rate": 3.268418902355985e-05,
      "loss": 4.2808,
      "step": 13215
    },
    {
      "epoch": 0.35523310493080745,
      "grad_norm": 2.5117104053497314,
      "learning_rate": 3.267737981751328e-05,
      "loss": 4.4748,
      "step": 13220
    },
    {
      "epoch": 0.35536745935778585,
      "grad_norm": 2.301981210708618,
      "learning_rate": 3.2670570611466705e-05,
      "loss": 4.2165,
      "step": 13225
    },
    {
      "epoch": 0.3555018137847642,
      "grad_norm": 2.513745069503784,
      "learning_rate": 3.266376140542013e-05,
      "loss": 4.1783,
      "step": 13230
    },
    {
      "epoch": 0.3556361682117426,
      "grad_norm": 2.4079809188842773,
      "learning_rate": 3.2656952199373556e-05,
      "loss": 4.2204,
      "step": 13235
    },
    {
      "epoch": 0.35577052263872094,
      "grad_norm": 2.3042685985565186,
      "learning_rate": 3.2650142993326985e-05,
      "loss": 4.3002,
      "step": 13240
    },
    {
      "epoch": 0.3559048770656993,
      "grad_norm": 2.2028629779815674,
      "learning_rate": 3.264333378728041e-05,
      "loss": 4.441,
      "step": 13245
    },
    {
      "epoch": 0.3560392314926777,
      "grad_norm": 2.6395745277404785,
      "learning_rate": 3.263652458123383e-05,
      "loss": 4.3542,
      "step": 13250
    },
    {
      "epoch": 0.35617358591965603,
      "grad_norm": 2.1632800102233887,
      "learning_rate": 3.262971537518725e-05,
      "loss": 4.4031,
      "step": 13255
    },
    {
      "epoch": 0.35630794034663443,
      "grad_norm": 2.2891623973846436,
      "learning_rate": 3.262290616914068e-05,
      "loss": 4.3825,
      "step": 13260
    },
    {
      "epoch": 0.3564422947736128,
      "grad_norm": 2.374030113220215,
      "learning_rate": 3.26160969630941e-05,
      "loss": 4.3131,
      "step": 13265
    },
    {
      "epoch": 0.3565766492005912,
      "grad_norm": 2.316019058227539,
      "learning_rate": 3.260928775704753e-05,
      "loss": 4.3345,
      "step": 13270
    },
    {
      "epoch": 0.3567110036275695,
      "grad_norm": 2.604022979736328,
      "learning_rate": 3.2602478551000954e-05,
      "loss": 4.3721,
      "step": 13275
    },
    {
      "epoch": 0.3568453580545479,
      "grad_norm": 2.4381356239318848,
      "learning_rate": 3.2595669344954383e-05,
      "loss": 4.312,
      "step": 13280
    },
    {
      "epoch": 0.35697971248152627,
      "grad_norm": 2.338482618331909,
      "learning_rate": 3.2588860138907806e-05,
      "loss": 4.3368,
      "step": 13285
    },
    {
      "epoch": 0.3571140669085046,
      "grad_norm": 2.236905813217163,
      "learning_rate": 3.258205093286123e-05,
      "loss": 4.2244,
      "step": 13290
    },
    {
      "epoch": 0.357248421335483,
      "grad_norm": 2.308786153793335,
      "learning_rate": 3.257524172681466e-05,
      "loss": 4.2951,
      "step": 13295
    },
    {
      "epoch": 0.35738277576246136,
      "grad_norm": 2.677859306335449,
      "learning_rate": 3.256843252076808e-05,
      "loss": 4.2441,
      "step": 13300
    },
    {
      "epoch": 0.35751713018943976,
      "grad_norm": 2.5888803005218506,
      "learning_rate": 3.256162331472151e-05,
      "loss": 4.2934,
      "step": 13305
    },
    {
      "epoch": 0.3576514846164181,
      "grad_norm": 2.2555806636810303,
      "learning_rate": 3.255481410867493e-05,
      "loss": 4.3666,
      "step": 13310
    },
    {
      "epoch": 0.3577858390433965,
      "grad_norm": 2.345254898071289,
      "learning_rate": 3.254800490262835e-05,
      "loss": 4.478,
      "step": 13315
    },
    {
      "epoch": 0.35792019347037485,
      "grad_norm": 2.3655991554260254,
      "learning_rate": 3.2541195696581775e-05,
      "loss": 4.3758,
      "step": 13320
    },
    {
      "epoch": 0.3580545478973532,
      "grad_norm": 2.3388137817382812,
      "learning_rate": 3.2534386490535204e-05,
      "loss": 4.3717,
      "step": 13325
    },
    {
      "epoch": 0.3581889023243316,
      "grad_norm": 2.4523465633392334,
      "learning_rate": 3.252757728448863e-05,
      "loss": 4.3566,
      "step": 13330
    },
    {
      "epoch": 0.35832325675130994,
      "grad_norm": 2.712498903274536,
      "learning_rate": 3.2520768078442056e-05,
      "loss": 4.3991,
      "step": 13335
    },
    {
      "epoch": 0.35845761117828834,
      "grad_norm": 2.667022943496704,
      "learning_rate": 3.251395887239548e-05,
      "loss": 4.4253,
      "step": 13340
    },
    {
      "epoch": 0.3585919656052667,
      "grad_norm": 2.307260751724243,
      "learning_rate": 3.250714966634891e-05,
      "loss": 4.319,
      "step": 13345
    },
    {
      "epoch": 0.3587263200322451,
      "grad_norm": 2.3503916263580322,
      "learning_rate": 3.250034046030233e-05,
      "loss": 4.4879,
      "step": 13350
    },
    {
      "epoch": 0.3588606744592234,
      "grad_norm": 2.235161781311035,
      "learning_rate": 3.249353125425575e-05,
      "loss": 4.3408,
      "step": 13355
    },
    {
      "epoch": 0.35899502888620183,
      "grad_norm": 2.282172918319702,
      "learning_rate": 3.248672204820918e-05,
      "loss": 4.3556,
      "step": 13360
    },
    {
      "epoch": 0.3591293833131802,
      "grad_norm": 2.2733702659606934,
      "learning_rate": 3.247991284216261e-05,
      "loss": 4.2609,
      "step": 13365
    },
    {
      "epoch": 0.3592637377401585,
      "grad_norm": 2.234506607055664,
      "learning_rate": 3.247310363611603e-05,
      "loss": 4.3787,
      "step": 13370
    },
    {
      "epoch": 0.3593980921671369,
      "grad_norm": 2.5725324153900146,
      "learning_rate": 3.2466294430069454e-05,
      "loss": 4.2591,
      "step": 13375
    },
    {
      "epoch": 0.35953244659411526,
      "grad_norm": 2.399301767349243,
      "learning_rate": 3.2459485224022876e-05,
      "loss": 4.3134,
      "step": 13380
    },
    {
      "epoch": 0.35966680102109366,
      "grad_norm": 2.3300259113311768,
      "learning_rate": 3.2452676017976305e-05,
      "loss": 4.4212,
      "step": 13385
    },
    {
      "epoch": 0.359801155448072,
      "grad_norm": 2.4857661724090576,
      "learning_rate": 3.2445866811929734e-05,
      "loss": 4.3523,
      "step": 13390
    },
    {
      "epoch": 0.3599355098750504,
      "grad_norm": 2.1681809425354004,
      "learning_rate": 3.243905760588316e-05,
      "loss": 4.2649,
      "step": 13395
    },
    {
      "epoch": 0.36006986430202875,
      "grad_norm": 2.599060535430908,
      "learning_rate": 3.243224839983658e-05,
      "loss": 4.2693,
      "step": 13400
    },
    {
      "epoch": 0.3602042187290071,
      "grad_norm": 2.467041254043579,
      "learning_rate": 3.242543919379001e-05,
      "loss": 4.3746,
      "step": 13405
    },
    {
      "epoch": 0.3603385731559855,
      "grad_norm": 2.2224814891815186,
      "learning_rate": 3.241862998774343e-05,
      "loss": 4.2686,
      "step": 13410
    },
    {
      "epoch": 0.36047292758296384,
      "grad_norm": 2.422572374343872,
      "learning_rate": 3.241182078169685e-05,
      "loss": 4.3886,
      "step": 13415
    },
    {
      "epoch": 0.36060728200994224,
      "grad_norm": 2.1943535804748535,
      "learning_rate": 3.240501157565028e-05,
      "loss": 4.3362,
      "step": 13420
    },
    {
      "epoch": 0.3607416364369206,
      "grad_norm": 2.3149867057800293,
      "learning_rate": 3.239820236960371e-05,
      "loss": 4.4284,
      "step": 13425
    },
    {
      "epoch": 0.360875990863899,
      "grad_norm": 2.480860948562622,
      "learning_rate": 3.239139316355713e-05,
      "loss": 4.3047,
      "step": 13430
    },
    {
      "epoch": 0.36101034529087733,
      "grad_norm": 2.400604248046875,
      "learning_rate": 3.2384583957510555e-05,
      "loss": 4.3188,
      "step": 13435
    },
    {
      "epoch": 0.3611446997178557,
      "grad_norm": 2.2765445709228516,
      "learning_rate": 3.237777475146398e-05,
      "loss": 4.4669,
      "step": 13440
    },
    {
      "epoch": 0.3612790541448341,
      "grad_norm": 2.3425376415252686,
      "learning_rate": 3.2370965545417406e-05,
      "loss": 4.3459,
      "step": 13445
    },
    {
      "epoch": 0.3614134085718124,
      "grad_norm": 2.1203501224517822,
      "learning_rate": 3.2364156339370835e-05,
      "loss": 4.2457,
      "step": 13450
    },
    {
      "epoch": 0.3615477629987908,
      "grad_norm": 2.41072678565979,
      "learning_rate": 3.235734713332426e-05,
      "loss": 4.2384,
      "step": 13455
    },
    {
      "epoch": 0.36168211742576917,
      "grad_norm": 2.6223561763763428,
      "learning_rate": 3.235053792727768e-05,
      "loss": 4.4556,
      "step": 13460
    },
    {
      "epoch": 0.36181647185274757,
      "grad_norm": 2.5104267597198486,
      "learning_rate": 3.23437287212311e-05,
      "loss": 4.3188,
      "step": 13465
    },
    {
      "epoch": 0.3619508262797259,
      "grad_norm": 2.3373522758483887,
      "learning_rate": 3.233691951518453e-05,
      "loss": 4.3959,
      "step": 13470
    },
    {
      "epoch": 0.3620851807067043,
      "grad_norm": 2.3905205726623535,
      "learning_rate": 3.2330110309137954e-05,
      "loss": 4.2754,
      "step": 13475
    },
    {
      "epoch": 0.36221953513368266,
      "grad_norm": 2.557098150253296,
      "learning_rate": 3.232330110309138e-05,
      "loss": 4.3411,
      "step": 13480
    },
    {
      "epoch": 0.362353889560661,
      "grad_norm": 2.4500479698181152,
      "learning_rate": 3.2316491897044805e-05,
      "loss": 4.234,
      "step": 13485
    },
    {
      "epoch": 0.3624882439876394,
      "grad_norm": 2.1909146308898926,
      "learning_rate": 3.2309682690998234e-05,
      "loss": 4.4129,
      "step": 13490
    },
    {
      "epoch": 0.36262259841461775,
      "grad_norm": 2.441007614135742,
      "learning_rate": 3.2302873484951656e-05,
      "loss": 4.3136,
      "step": 13495
    },
    {
      "epoch": 0.36275695284159615,
      "grad_norm": 2.4471328258514404,
      "learning_rate": 3.229606427890508e-05,
      "loss": 4.3156,
      "step": 13500
    },
    {
      "epoch": 0.3628913072685745,
      "grad_norm": 2.4198410511016846,
      "learning_rate": 3.228925507285851e-05,
      "loss": 4.3298,
      "step": 13505
    },
    {
      "epoch": 0.3630256616955529,
      "grad_norm": 2.2383103370666504,
      "learning_rate": 3.2282445866811937e-05,
      "loss": 4.4694,
      "step": 13510
    },
    {
      "epoch": 0.36316001612253124,
      "grad_norm": 2.175715684890747,
      "learning_rate": 3.227563666076536e-05,
      "loss": 4.3111,
      "step": 13515
    },
    {
      "epoch": 0.3632943705495096,
      "grad_norm": 2.485316514968872,
      "learning_rate": 3.226882745471878e-05,
      "loss": 4.2738,
      "step": 13520
    },
    {
      "epoch": 0.363428724976488,
      "grad_norm": 2.4540674686431885,
      "learning_rate": 3.22620182486722e-05,
      "loss": 4.175,
      "step": 13525
    },
    {
      "epoch": 0.3635630794034663,
      "grad_norm": 2.538102388381958,
      "learning_rate": 3.225520904262563e-05,
      "loss": 4.3979,
      "step": 13530
    },
    {
      "epoch": 0.3636974338304447,
      "grad_norm": 2.215350389480591,
      "learning_rate": 3.2248399836579055e-05,
      "loss": 4.2962,
      "step": 13535
    },
    {
      "epoch": 0.36383178825742307,
      "grad_norm": 2.4535844326019287,
      "learning_rate": 3.2241590630532484e-05,
      "loss": 4.2784,
      "step": 13540
    },
    {
      "epoch": 0.3639661426844015,
      "grad_norm": 2.6061220169067383,
      "learning_rate": 3.2234781424485906e-05,
      "loss": 4.4522,
      "step": 13545
    },
    {
      "epoch": 0.3641004971113798,
      "grad_norm": 2.3602547645568848,
      "learning_rate": 3.2227972218439335e-05,
      "loss": 4.2543,
      "step": 13550
    },
    {
      "epoch": 0.36423485153835816,
      "grad_norm": 2.4458539485931396,
      "learning_rate": 3.222116301239276e-05,
      "loss": 4.3424,
      "step": 13555
    },
    {
      "epoch": 0.36436920596533656,
      "grad_norm": 2.3571009635925293,
      "learning_rate": 3.221435380634618e-05,
      "loss": 4.277,
      "step": 13560
    },
    {
      "epoch": 0.3645035603923149,
      "grad_norm": 2.3767616748809814,
      "learning_rate": 3.22075446002996e-05,
      "loss": 4.2392,
      "step": 13565
    },
    {
      "epoch": 0.3646379148192933,
      "grad_norm": 2.26343035697937,
      "learning_rate": 3.220073539425304e-05,
      "loss": 4.3238,
      "step": 13570
    },
    {
      "epoch": 0.36477226924627165,
      "grad_norm": 2.4008126258850098,
      "learning_rate": 3.219392618820646e-05,
      "loss": 4.3178,
      "step": 13575
    },
    {
      "epoch": 0.36490662367325005,
      "grad_norm": 2.539283037185669,
      "learning_rate": 3.218711698215988e-05,
      "loss": 4.2648,
      "step": 13580
    },
    {
      "epoch": 0.3650409781002284,
      "grad_norm": 2.628666400909424,
      "learning_rate": 3.2180307776113304e-05,
      "loss": 4.2871,
      "step": 13585
    },
    {
      "epoch": 0.3651753325272068,
      "grad_norm": 2.194849729537964,
      "learning_rate": 3.217349857006673e-05,
      "loss": 4.256,
      "step": 13590
    },
    {
      "epoch": 0.36530968695418514,
      "grad_norm": 2.3149709701538086,
      "learning_rate": 3.2166689364020156e-05,
      "loss": 4.2944,
      "step": 13595
    },
    {
      "epoch": 0.3654440413811635,
      "grad_norm": 2.3692774772644043,
      "learning_rate": 3.2159880157973585e-05,
      "loss": 4.3548,
      "step": 13600
    },
    {
      "epoch": 0.3655783958081419,
      "grad_norm": 2.481020212173462,
      "learning_rate": 3.215307095192701e-05,
      "loss": 4.2844,
      "step": 13605
    },
    {
      "epoch": 0.36571275023512023,
      "grad_norm": 2.3730828762054443,
      "learning_rate": 3.214626174588043e-05,
      "loss": 4.3499,
      "step": 13610
    },
    {
      "epoch": 0.36584710466209863,
      "grad_norm": 2.391204595565796,
      "learning_rate": 3.213945253983386e-05,
      "loss": 4.323,
      "step": 13615
    },
    {
      "epoch": 0.365981459089077,
      "grad_norm": 3.4085335731506348,
      "learning_rate": 3.213264333378728e-05,
      "loss": 4.3378,
      "step": 13620
    },
    {
      "epoch": 0.3661158135160554,
      "grad_norm": 2.3898353576660156,
      "learning_rate": 3.21258341277407e-05,
      "loss": 4.2884,
      "step": 13625
    },
    {
      "epoch": 0.3662501679430337,
      "grad_norm": 2.4828672409057617,
      "learning_rate": 3.211902492169413e-05,
      "loss": 4.2994,
      "step": 13630
    },
    {
      "epoch": 0.36638452237001207,
      "grad_norm": 2.3456521034240723,
      "learning_rate": 3.211221571564756e-05,
      "loss": 4.2502,
      "step": 13635
    },
    {
      "epoch": 0.36651887679699047,
      "grad_norm": 2.24570894241333,
      "learning_rate": 3.210540650960098e-05,
      "loss": 4.2802,
      "step": 13640
    },
    {
      "epoch": 0.3666532312239688,
      "grad_norm": 2.4074454307556152,
      "learning_rate": 3.2098597303554406e-05,
      "loss": 4.3179,
      "step": 13645
    },
    {
      "epoch": 0.3667875856509472,
      "grad_norm": 2.4296061992645264,
      "learning_rate": 3.209178809750783e-05,
      "loss": 4.3309,
      "step": 13650
    },
    {
      "epoch": 0.36692194007792556,
      "grad_norm": 2.282538890838623,
      "learning_rate": 3.208497889146126e-05,
      "loss": 4.3102,
      "step": 13655
    },
    {
      "epoch": 0.36705629450490396,
      "grad_norm": 2.2850332260131836,
      "learning_rate": 3.2078169685414686e-05,
      "loss": 4.3872,
      "step": 13660
    },
    {
      "epoch": 0.3671906489318823,
      "grad_norm": 2.5425775051116943,
      "learning_rate": 3.207136047936811e-05,
      "loss": 4.3569,
      "step": 13665
    },
    {
      "epoch": 0.3673250033588607,
      "grad_norm": 2.4117627143859863,
      "learning_rate": 3.206455127332153e-05,
      "loss": 4.2418,
      "step": 13670
    },
    {
      "epoch": 0.36745935778583905,
      "grad_norm": 2.239330291748047,
      "learning_rate": 3.205774206727496e-05,
      "loss": 4.4028,
      "step": 13675
    },
    {
      "epoch": 0.3675937122128174,
      "grad_norm": 2.353691816329956,
      "learning_rate": 3.205093286122838e-05,
      "loss": 4.3272,
      "step": 13680
    },
    {
      "epoch": 0.3677280666397958,
      "grad_norm": 2.436859369277954,
      "learning_rate": 3.2044123655181804e-05,
      "loss": 4.2692,
      "step": 13685
    },
    {
      "epoch": 0.36786242106677414,
      "grad_norm": 2.4121267795562744,
      "learning_rate": 3.203731444913523e-05,
      "loss": 4.2616,
      "step": 13690
    },
    {
      "epoch": 0.36799677549375254,
      "grad_norm": 2.618720293045044,
      "learning_rate": 3.203050524308866e-05,
      "loss": 4.3374,
      "step": 13695
    },
    {
      "epoch": 0.3681311299207309,
      "grad_norm": 2.3474345207214355,
      "learning_rate": 3.2023696037042084e-05,
      "loss": 4.2059,
      "step": 13700
    },
    {
      "epoch": 0.3682654843477093,
      "grad_norm": 2.301142692565918,
      "learning_rate": 3.201688683099551e-05,
      "loss": 4.1871,
      "step": 13705
    },
    {
      "epoch": 0.3683998387746876,
      "grad_norm": 2.4006447792053223,
      "learning_rate": 3.201007762494893e-05,
      "loss": 4.3669,
      "step": 13710
    },
    {
      "epoch": 0.36853419320166597,
      "grad_norm": 2.057865858078003,
      "learning_rate": 3.200326841890236e-05,
      "loss": 4.2507,
      "step": 13715
    },
    {
      "epoch": 0.36866854762864437,
      "grad_norm": 2.4721176624298096,
      "learning_rate": 3.199645921285579e-05,
      "loss": 4.2579,
      "step": 13720
    },
    {
      "epoch": 0.3688029020556227,
      "grad_norm": 2.228379011154175,
      "learning_rate": 3.198965000680921e-05,
      "loss": 4.3001,
      "step": 13725
    },
    {
      "epoch": 0.3689372564826011,
      "grad_norm": 2.6983954906463623,
      "learning_rate": 3.198284080076263e-05,
      "loss": 4.2866,
      "step": 13730
    },
    {
      "epoch": 0.36907161090957946,
      "grad_norm": 2.450631856918335,
      "learning_rate": 3.1976031594716054e-05,
      "loss": 4.2417,
      "step": 13735
    },
    {
      "epoch": 0.36920596533655786,
      "grad_norm": 2.277873992919922,
      "learning_rate": 3.196922238866948e-05,
      "loss": 4.3859,
      "step": 13740
    },
    {
      "epoch": 0.3693403197635362,
      "grad_norm": 2.459268808364868,
      "learning_rate": 3.1962413182622905e-05,
      "loss": 4.4825,
      "step": 13745
    },
    {
      "epoch": 0.36947467419051455,
      "grad_norm": 2.222341299057007,
      "learning_rate": 3.1955603976576334e-05,
      "loss": 4.4071,
      "step": 13750
    },
    {
      "epoch": 0.36960902861749295,
      "grad_norm": 2.232779026031494,
      "learning_rate": 3.1948794770529757e-05,
      "loss": 4.3603,
      "step": 13755
    },
    {
      "epoch": 0.3697433830444713,
      "grad_norm": 2.4482953548431396,
      "learning_rate": 3.1941985564483186e-05,
      "loss": 4.4021,
      "step": 13760
    },
    {
      "epoch": 0.3698777374714497,
      "grad_norm": 2.452336549758911,
      "learning_rate": 3.193517635843661e-05,
      "loss": 4.3711,
      "step": 13765
    },
    {
      "epoch": 0.37001209189842804,
      "grad_norm": 2.470099449157715,
      "learning_rate": 3.192836715239003e-05,
      "loss": 4.4044,
      "step": 13770
    },
    {
      "epoch": 0.37014644632540644,
      "grad_norm": 2.5558035373687744,
      "learning_rate": 3.192155794634345e-05,
      "loss": 4.3643,
      "step": 13775
    },
    {
      "epoch": 0.3702808007523848,
      "grad_norm": 2.4871790409088135,
      "learning_rate": 3.191474874029688e-05,
      "loss": 4.2087,
      "step": 13780
    },
    {
      "epoch": 0.3704151551793632,
      "grad_norm": 2.2882003784179688,
      "learning_rate": 3.190793953425031e-05,
      "loss": 4.3122,
      "step": 13785
    },
    {
      "epoch": 0.37054950960634153,
      "grad_norm": 2.282665967941284,
      "learning_rate": 3.190113032820373e-05,
      "loss": 4.3829,
      "step": 13790
    },
    {
      "epoch": 0.3706838640333199,
      "grad_norm": 2.388610363006592,
      "learning_rate": 3.1894321122157155e-05,
      "loss": 4.3001,
      "step": 13795
    },
    {
      "epoch": 0.3708182184602983,
      "grad_norm": 2.1769843101501465,
      "learning_rate": 3.1887511916110584e-05,
      "loss": 4.4585,
      "step": 13800
    },
    {
      "epoch": 0.3709525728872766,
      "grad_norm": 2.489595413208008,
      "learning_rate": 3.1880702710064006e-05,
      "loss": 4.2977,
      "step": 13805
    },
    {
      "epoch": 0.371086927314255,
      "grad_norm": 2.352616548538208,
      "learning_rate": 3.1873893504017435e-05,
      "loss": 4.2531,
      "step": 13810
    },
    {
      "epoch": 0.37122128174123337,
      "grad_norm": 2.5033323764801025,
      "learning_rate": 3.186708429797086e-05,
      "loss": 4.4416,
      "step": 13815
    },
    {
      "epoch": 0.37135563616821177,
      "grad_norm": 2.4697649478912354,
      "learning_rate": 3.186027509192429e-05,
      "loss": 4.3765,
      "step": 13820
    },
    {
      "epoch": 0.3714899905951901,
      "grad_norm": 2.4438958168029785,
      "learning_rate": 3.185346588587771e-05,
      "loss": 4.334,
      "step": 13825
    },
    {
      "epoch": 0.37162434502216846,
      "grad_norm": 2.189882278442383,
      "learning_rate": 3.184665667983113e-05,
      "loss": 4.2742,
      "step": 13830
    },
    {
      "epoch": 0.37175869944914686,
      "grad_norm": 2.1358470916748047,
      "learning_rate": 3.1839847473784553e-05,
      "loss": 4.3132,
      "step": 13835
    },
    {
      "epoch": 0.3718930538761252,
      "grad_norm": 2.26397967338562,
      "learning_rate": 3.183303826773798e-05,
      "loss": 4.2145,
      "step": 13840
    },
    {
      "epoch": 0.3720274083031036,
      "grad_norm": 2.44718074798584,
      "learning_rate": 3.182622906169141e-05,
      "loss": 4.3329,
      "step": 13845
    },
    {
      "epoch": 0.37216176273008195,
      "grad_norm": 2.214589834213257,
      "learning_rate": 3.1819419855644834e-05,
      "loss": 4.2368,
      "step": 13850
    },
    {
      "epoch": 0.37229611715706035,
      "grad_norm": 2.757277727127075,
      "learning_rate": 3.1812610649598256e-05,
      "loss": 4.2521,
      "step": 13855
    },
    {
      "epoch": 0.3724304715840387,
      "grad_norm": 2.1392934322357178,
      "learning_rate": 3.1805801443551685e-05,
      "loss": 4.3144,
      "step": 13860
    },
    {
      "epoch": 0.37256482601101704,
      "grad_norm": 2.5721726417541504,
      "learning_rate": 3.179899223750511e-05,
      "loss": 4.2698,
      "step": 13865
    },
    {
      "epoch": 0.37269918043799544,
      "grad_norm": 2.332751512527466,
      "learning_rate": 3.1792183031458536e-05,
      "loss": 4.2705,
      "step": 13870
    },
    {
      "epoch": 0.3728335348649738,
      "grad_norm": 2.238003969192505,
      "learning_rate": 3.178537382541196e-05,
      "loss": 4.3019,
      "step": 13875
    },
    {
      "epoch": 0.3729678892919522,
      "grad_norm": 2.2340259552001953,
      "learning_rate": 3.177856461936538e-05,
      "loss": 4.3363,
      "step": 13880
    },
    {
      "epoch": 0.3731022437189305,
      "grad_norm": 2.435854911804199,
      "learning_rate": 3.177175541331881e-05,
      "loss": 4.3889,
      "step": 13885
    },
    {
      "epoch": 0.3732365981459089,
      "grad_norm": 2.3172571659088135,
      "learning_rate": 3.176494620727223e-05,
      "loss": 4.3017,
      "step": 13890
    },
    {
      "epoch": 0.37337095257288727,
      "grad_norm": 2.4550118446350098,
      "learning_rate": 3.1758137001225655e-05,
      "loss": 4.3483,
      "step": 13895
    },
    {
      "epoch": 0.37350530699986567,
      "grad_norm": 2.4662134647369385,
      "learning_rate": 3.1751327795179084e-05,
      "loss": 4.1626,
      "step": 13900
    },
    {
      "epoch": 0.373639661426844,
      "grad_norm": 2.231564998626709,
      "learning_rate": 3.174451858913251e-05,
      "loss": 4.3447,
      "step": 13905
    },
    {
      "epoch": 0.37377401585382236,
      "grad_norm": 2.454305648803711,
      "learning_rate": 3.1737709383085935e-05,
      "loss": 4.3161,
      "step": 13910
    },
    {
      "epoch": 0.37390837028080076,
      "grad_norm": 2.379092216491699,
      "learning_rate": 3.173090017703936e-05,
      "loss": 4.2613,
      "step": 13915
    },
    {
      "epoch": 0.3740427247077791,
      "grad_norm": 2.5741708278656006,
      "learning_rate": 3.172409097099278e-05,
      "loss": 4.2513,
      "step": 13920
    },
    {
      "epoch": 0.3741770791347575,
      "grad_norm": 2.3459253311157227,
      "learning_rate": 3.171728176494621e-05,
      "loss": 4.2775,
      "step": 13925
    },
    {
      "epoch": 0.37431143356173585,
      "grad_norm": 2.249350070953369,
      "learning_rate": 3.171047255889964e-05,
      "loss": 4.2756,
      "step": 13930
    },
    {
      "epoch": 0.37444578798871425,
      "grad_norm": 2.5505788326263428,
      "learning_rate": 3.170366335285306e-05,
      "loss": 4.2877,
      "step": 13935
    },
    {
      "epoch": 0.3745801424156926,
      "grad_norm": 2.5322370529174805,
      "learning_rate": 3.169685414680648e-05,
      "loss": 4.3698,
      "step": 13940
    },
    {
      "epoch": 0.37471449684267094,
      "grad_norm": 2.253079891204834,
      "learning_rate": 3.169004494075991e-05,
      "loss": 4.2765,
      "step": 13945
    },
    {
      "epoch": 0.37484885126964934,
      "grad_norm": 2.479083776473999,
      "learning_rate": 3.1683235734713333e-05,
      "loss": 4.3739,
      "step": 13950
    },
    {
      "epoch": 0.3749832056966277,
      "grad_norm": 2.4715497493743896,
      "learning_rate": 3.1676426528666756e-05,
      "loss": 4.3076,
      "step": 13955
    },
    {
      "epoch": 0.3751175601236061,
      "grad_norm": 2.3707046508789062,
      "learning_rate": 3.1669617322620185e-05,
      "loss": 4.2982,
      "step": 13960
    },
    {
      "epoch": 0.37525191455058443,
      "grad_norm": 2.500237226486206,
      "learning_rate": 3.1662808116573614e-05,
      "loss": 4.2391,
      "step": 13965
    },
    {
      "epoch": 0.37538626897756283,
      "grad_norm": 2.3745200634002686,
      "learning_rate": 3.1655998910527036e-05,
      "loss": 4.3926,
      "step": 13970
    },
    {
      "epoch": 0.3755206234045412,
      "grad_norm": 2.675363302230835,
      "learning_rate": 3.164918970448046e-05,
      "loss": 4.336,
      "step": 13975
    },
    {
      "epoch": 0.3756549778315195,
      "grad_norm": 2.282299280166626,
      "learning_rate": 3.164238049843388e-05,
      "loss": 4.2732,
      "step": 13980
    },
    {
      "epoch": 0.3757893322584979,
      "grad_norm": 2.0888710021972656,
      "learning_rate": 3.163557129238731e-05,
      "loss": 4.283,
      "step": 13985
    },
    {
      "epoch": 0.37592368668547627,
      "grad_norm": 2.2703559398651123,
      "learning_rate": 3.162876208634073e-05,
      "loss": 4.3679,
      "step": 13990
    },
    {
      "epoch": 0.37605804111245467,
      "grad_norm": 2.280846118927002,
      "learning_rate": 3.162195288029416e-05,
      "loss": 4.3027,
      "step": 13995
    },
    {
      "epoch": 0.376192395539433,
      "grad_norm": 2.2215635776519775,
      "learning_rate": 3.161514367424758e-05,
      "loss": 4.1959,
      "step": 14000
    },
    {
      "epoch": 0.3763267499664114,
      "grad_norm": 2.3503646850585938,
      "learning_rate": 3.160833446820101e-05,
      "loss": 4.2136,
      "step": 14005
    },
    {
      "epoch": 0.37646110439338976,
      "grad_norm": 2.187411069869995,
      "learning_rate": 3.1601525262154435e-05,
      "loss": 4.2578,
      "step": 14010
    },
    {
      "epoch": 0.37659545882036816,
      "grad_norm": 2.5558669567108154,
      "learning_rate": 3.159471605610786e-05,
      "loss": 4.304,
      "step": 14015
    },
    {
      "epoch": 0.3767298132473465,
      "grad_norm": 2.2085471153259277,
      "learning_rate": 3.1587906850061286e-05,
      "loss": 4.3951,
      "step": 14020
    },
    {
      "epoch": 0.37686416767432485,
      "grad_norm": 2.408627510070801,
      "learning_rate": 3.1581097644014715e-05,
      "loss": 4.3445,
      "step": 14025
    },
    {
      "epoch": 0.37699852210130325,
      "grad_norm": 2.470081329345703,
      "learning_rate": 3.157428843796814e-05,
      "loss": 4.2864,
      "step": 14030
    },
    {
      "epoch": 0.3771328765282816,
      "grad_norm": 2.3218884468078613,
      "learning_rate": 3.156747923192156e-05,
      "loss": 4.274,
      "step": 14035
    },
    {
      "epoch": 0.37726723095526,
      "grad_norm": 2.378514051437378,
      "learning_rate": 3.156067002587498e-05,
      "loss": 4.3995,
      "step": 14040
    },
    {
      "epoch": 0.37740158538223834,
      "grad_norm": 2.340200424194336,
      "learning_rate": 3.1553860819828404e-05,
      "loss": 4.3735,
      "step": 14045
    },
    {
      "epoch": 0.37753593980921674,
      "grad_norm": 2.374241590499878,
      "learning_rate": 3.154705161378183e-05,
      "loss": 4.2402,
      "step": 14050
    },
    {
      "epoch": 0.3776702942361951,
      "grad_norm": 2.232448101043701,
      "learning_rate": 3.154024240773526e-05,
      "loss": 4.3424,
      "step": 14055
    },
    {
      "epoch": 0.3778046486631734,
      "grad_norm": 2.2524218559265137,
      "learning_rate": 3.1533433201688684e-05,
      "loss": 4.2701,
      "step": 14060
    },
    {
      "epoch": 0.3779390030901518,
      "grad_norm": 2.14957594871521,
      "learning_rate": 3.152662399564211e-05,
      "loss": 4.1266,
      "step": 14065
    },
    {
      "epoch": 0.37807335751713017,
      "grad_norm": 2.4026060104370117,
      "learning_rate": 3.1519814789595536e-05,
      "loss": 4.407,
      "step": 14070
    },
    {
      "epoch": 0.37820771194410857,
      "grad_norm": 2.387247323989868,
      "learning_rate": 3.151300558354896e-05,
      "loss": 4.2958,
      "step": 14075
    },
    {
      "epoch": 0.3783420663710869,
      "grad_norm": 2.4510486125946045,
      "learning_rate": 3.150619637750239e-05,
      "loss": 4.3604,
      "step": 14080
    },
    {
      "epoch": 0.3784764207980653,
      "grad_norm": 2.552234649658203,
      "learning_rate": 3.149938717145581e-05,
      "loss": 4.3157,
      "step": 14085
    },
    {
      "epoch": 0.37861077522504366,
      "grad_norm": 2.2374775409698486,
      "learning_rate": 3.149257796540924e-05,
      "loss": 4.2936,
      "step": 14090
    },
    {
      "epoch": 0.37874512965202206,
      "grad_norm": 2.369304895401001,
      "learning_rate": 3.148576875936266e-05,
      "loss": 4.3154,
      "step": 14095
    },
    {
      "epoch": 0.3788794840790004,
      "grad_norm": 2.432016134262085,
      "learning_rate": 3.147895955331608e-05,
      "loss": 4.1877,
      "step": 14100
    },
    {
      "epoch": 0.37901383850597875,
      "grad_norm": 2.4371907711029053,
      "learning_rate": 3.1472150347269505e-05,
      "loss": 4.1549,
      "step": 14105
    },
    {
      "epoch": 0.37914819293295715,
      "grad_norm": 2.350525379180908,
      "learning_rate": 3.1465341141222934e-05,
      "loss": 4.4245,
      "step": 14110
    },
    {
      "epoch": 0.3792825473599355,
      "grad_norm": 2.5332536697387695,
      "learning_rate": 3.145853193517636e-05,
      "loss": 4.2288,
      "step": 14115
    },
    {
      "epoch": 0.3794169017869139,
      "grad_norm": 2.175351619720459,
      "learning_rate": 3.1451722729129785e-05,
      "loss": 4.3708,
      "step": 14120
    },
    {
      "epoch": 0.37955125621389224,
      "grad_norm": 2.519005537033081,
      "learning_rate": 3.144491352308321e-05,
      "loss": 4.2499,
      "step": 14125
    },
    {
      "epoch": 0.37968561064087064,
      "grad_norm": 2.4850316047668457,
      "learning_rate": 3.143810431703664e-05,
      "loss": 4.3916,
      "step": 14130
    },
    {
      "epoch": 0.379819965067849,
      "grad_norm": 2.2136423587799072,
      "learning_rate": 3.143129511099006e-05,
      "loss": 4.1762,
      "step": 14135
    },
    {
      "epoch": 0.37995431949482733,
      "grad_norm": 2.305082321166992,
      "learning_rate": 3.142448590494349e-05,
      "loss": 4.2126,
      "step": 14140
    },
    {
      "epoch": 0.38008867392180573,
      "grad_norm": 2.4834587574005127,
      "learning_rate": 3.141767669889691e-05,
      "loss": 4.2659,
      "step": 14145
    },
    {
      "epoch": 0.3802230283487841,
      "grad_norm": 2.492098569869995,
      "learning_rate": 3.141086749285034e-05,
      "loss": 4.3247,
      "step": 14150
    },
    {
      "epoch": 0.3803573827757625,
      "grad_norm": 2.2131361961364746,
      "learning_rate": 3.140405828680376e-05,
      "loss": 4.1219,
      "step": 14155
    },
    {
      "epoch": 0.3804917372027408,
      "grad_norm": 2.1240522861480713,
      "learning_rate": 3.1397249080757184e-05,
      "loss": 4.2959,
      "step": 14160
    },
    {
      "epoch": 0.3806260916297192,
      "grad_norm": 2.4772067070007324,
      "learning_rate": 3.1390439874710606e-05,
      "loss": 4.2243,
      "step": 14165
    },
    {
      "epoch": 0.38076044605669757,
      "grad_norm": 2.3326475620269775,
      "learning_rate": 3.1383630668664035e-05,
      "loss": 4.3017,
      "step": 14170
    },
    {
      "epoch": 0.3808948004836759,
      "grad_norm": 2.458195686340332,
      "learning_rate": 3.1376821462617464e-05,
      "loss": 4.2366,
      "step": 14175
    },
    {
      "epoch": 0.3810291549106543,
      "grad_norm": 2.4461262226104736,
      "learning_rate": 3.1370012256570887e-05,
      "loss": 4.3435,
      "step": 14180
    },
    {
      "epoch": 0.38116350933763266,
      "grad_norm": 2.315472364425659,
      "learning_rate": 3.136320305052431e-05,
      "loss": 4.349,
      "step": 14185
    },
    {
      "epoch": 0.38129786376461106,
      "grad_norm": 2.5450448989868164,
      "learning_rate": 3.135639384447773e-05,
      "loss": 4.3052,
      "step": 14190
    },
    {
      "epoch": 0.3814322181915894,
      "grad_norm": 2.1828768253326416,
      "learning_rate": 3.134958463843116e-05,
      "loss": 4.2564,
      "step": 14195
    },
    {
      "epoch": 0.3815665726185678,
      "grad_norm": 2.2555363178253174,
      "learning_rate": 3.134277543238458e-05,
      "loss": 4.2913,
      "step": 14200
    },
    {
      "epoch": 0.38170092704554615,
      "grad_norm": 2.1589887142181396,
      "learning_rate": 3.133596622633801e-05,
      "loss": 4.2235,
      "step": 14205
    },
    {
      "epoch": 0.38183528147252455,
      "grad_norm": 2.4267561435699463,
      "learning_rate": 3.1329157020291434e-05,
      "loss": 4.3017,
      "step": 14210
    },
    {
      "epoch": 0.3819696358995029,
      "grad_norm": 2.5112481117248535,
      "learning_rate": 3.132234781424486e-05,
      "loss": 4.0831,
      "step": 14215
    },
    {
      "epoch": 0.38210399032648124,
      "grad_norm": 2.3390679359436035,
      "learning_rate": 3.1315538608198285e-05,
      "loss": 4.2947,
      "step": 14220
    },
    {
      "epoch": 0.38223834475345964,
      "grad_norm": 2.448854923248291,
      "learning_rate": 3.130872940215171e-05,
      "loss": 4.122,
      "step": 14225
    },
    {
      "epoch": 0.382372699180438,
      "grad_norm": 2.604612112045288,
      "learning_rate": 3.1301920196105136e-05,
      "loss": 4.258,
      "step": 14230
    },
    {
      "epoch": 0.3825070536074164,
      "grad_norm": 2.5440499782562256,
      "learning_rate": 3.1295110990058565e-05,
      "loss": 4.3373,
      "step": 14235
    },
    {
      "epoch": 0.3826414080343947,
      "grad_norm": 2.497462749481201,
      "learning_rate": 3.128830178401199e-05,
      "loss": 4.2936,
      "step": 14240
    },
    {
      "epoch": 0.3827757624613731,
      "grad_norm": 2.372606039047241,
      "learning_rate": 3.128149257796541e-05,
      "loss": 4.2315,
      "step": 14245
    },
    {
      "epoch": 0.38291011688835147,
      "grad_norm": 2.5296621322631836,
      "learning_rate": 3.127468337191883e-05,
      "loss": 4.2571,
      "step": 14250
    },
    {
      "epoch": 0.3830444713153298,
      "grad_norm": 2.2454700469970703,
      "learning_rate": 3.126787416587226e-05,
      "loss": 4.3954,
      "step": 14255
    },
    {
      "epoch": 0.3831788257423082,
      "grad_norm": 2.3120620250701904,
      "learning_rate": 3.1261064959825684e-05,
      "loss": 4.2695,
      "step": 14260
    },
    {
      "epoch": 0.38331318016928656,
      "grad_norm": 2.09098482131958,
      "learning_rate": 3.125425575377911e-05,
      "loss": 4.3242,
      "step": 14265
    },
    {
      "epoch": 0.38344753459626496,
      "grad_norm": 2.4609668254852295,
      "learning_rate": 3.1247446547732535e-05,
      "loss": 4.1983,
      "step": 14270
    },
    {
      "epoch": 0.3835818890232433,
      "grad_norm": 2.365065336227417,
      "learning_rate": 3.1240637341685964e-05,
      "loss": 4.2711,
      "step": 14275
    },
    {
      "epoch": 0.3837162434502217,
      "grad_norm": 2.483854055404663,
      "learning_rate": 3.1233828135639386e-05,
      "loss": 4.3801,
      "step": 14280
    },
    {
      "epoch": 0.38385059787720005,
      "grad_norm": 2.4485063552856445,
      "learning_rate": 3.122701892959281e-05,
      "loss": 4.3037,
      "step": 14285
    },
    {
      "epoch": 0.3839849523041784,
      "grad_norm": 2.3685925006866455,
      "learning_rate": 3.122020972354624e-05,
      "loss": 4.2255,
      "step": 14290
    },
    {
      "epoch": 0.3841193067311568,
      "grad_norm": 2.5178205966949463,
      "learning_rate": 3.1213400517499667e-05,
      "loss": 4.3208,
      "step": 14295
    },
    {
      "epoch": 0.38425366115813514,
      "grad_norm": 2.51224422454834,
      "learning_rate": 3.120659131145309e-05,
      "loss": 4.3695,
      "step": 14300
    },
    {
      "epoch": 0.38438801558511354,
      "grad_norm": 2.542829990386963,
      "learning_rate": 3.119978210540651e-05,
      "loss": 4.2582,
      "step": 14305
    },
    {
      "epoch": 0.3845223700120919,
      "grad_norm": 2.5083277225494385,
      "learning_rate": 3.119297289935993e-05,
      "loss": 4.3223,
      "step": 14310
    },
    {
      "epoch": 0.3846567244390703,
      "grad_norm": 2.367457151412964,
      "learning_rate": 3.118616369331336e-05,
      "loss": 4.351,
      "step": 14315
    },
    {
      "epoch": 0.38479107886604863,
      "grad_norm": 2.4680230617523193,
      "learning_rate": 3.1179354487266785e-05,
      "loss": 4.2761,
      "step": 14320
    },
    {
      "epoch": 0.38492543329302703,
      "grad_norm": 2.4924192428588867,
      "learning_rate": 3.1172545281220214e-05,
      "loss": 4.3583,
      "step": 14325
    },
    {
      "epoch": 0.3850597877200054,
      "grad_norm": 2.4994428157806396,
      "learning_rate": 3.1165736075173636e-05,
      "loss": 4.3397,
      "step": 14330
    },
    {
      "epoch": 0.3851941421469837,
      "grad_norm": 2.4816064834594727,
      "learning_rate": 3.115892686912706e-05,
      "loss": 4.1394,
      "step": 14335
    },
    {
      "epoch": 0.3853284965739621,
      "grad_norm": 2.2779366970062256,
      "learning_rate": 3.115211766308049e-05,
      "loss": 4.4211,
      "step": 14340
    },
    {
      "epoch": 0.38546285100094047,
      "grad_norm": 2.1995530128479004,
      "learning_rate": 3.114530845703391e-05,
      "loss": 4.299,
      "step": 14345
    },
    {
      "epoch": 0.38559720542791887,
      "grad_norm": 2.7301137447357178,
      "learning_rate": 3.113849925098734e-05,
      "loss": 4.3092,
      "step": 14350
    },
    {
      "epoch": 0.3857315598548972,
      "grad_norm": 2.349949836730957,
      "learning_rate": 3.113169004494076e-05,
      "loss": 4.2077,
      "step": 14355
    },
    {
      "epoch": 0.3858659142818756,
      "grad_norm": 2.3989739418029785,
      "learning_rate": 3.112488083889419e-05,
      "loss": 4.0892,
      "step": 14360
    },
    {
      "epoch": 0.38600026870885396,
      "grad_norm": 2.335411310195923,
      "learning_rate": 3.111807163284761e-05,
      "loss": 4.3051,
      "step": 14365
    },
    {
      "epoch": 0.3861346231358323,
      "grad_norm": 2.5236847400665283,
      "learning_rate": 3.1111262426801034e-05,
      "loss": 4.1836,
      "step": 14370
    },
    {
      "epoch": 0.3862689775628107,
      "grad_norm": 2.4553840160369873,
      "learning_rate": 3.110445322075446e-05,
      "loss": 4.2342,
      "step": 14375
    },
    {
      "epoch": 0.38640333198978905,
      "grad_norm": 2.4479806423187256,
      "learning_rate": 3.1097644014707886e-05,
      "loss": 4.1587,
      "step": 14380
    },
    {
      "epoch": 0.38653768641676745,
      "grad_norm": 2.241891384124756,
      "learning_rate": 3.1090834808661315e-05,
      "loss": 4.2843,
      "step": 14385
    },
    {
      "epoch": 0.3866720408437458,
      "grad_norm": 2.3287885189056396,
      "learning_rate": 3.108402560261474e-05,
      "loss": 4.2179,
      "step": 14390
    },
    {
      "epoch": 0.3868063952707242,
      "grad_norm": 2.5230860710144043,
      "learning_rate": 3.107721639656816e-05,
      "loss": 4.335,
      "step": 14395
    },
    {
      "epoch": 0.38694074969770254,
      "grad_norm": 2.3073525428771973,
      "learning_rate": 3.107040719052159e-05,
      "loss": 4.2206,
      "step": 14400
    },
    {
      "epoch": 0.38707510412468094,
      "grad_norm": 2.0871682167053223,
      "learning_rate": 3.106359798447501e-05,
      "loss": 4.3839,
      "step": 14405
    },
    {
      "epoch": 0.3872094585516593,
      "grad_norm": 2.3388237953186035,
      "learning_rate": 3.105678877842843e-05,
      "loss": 4.29,
      "step": 14410
    },
    {
      "epoch": 0.3873438129786376,
      "grad_norm": 2.275890588760376,
      "learning_rate": 3.104997957238186e-05,
      "loss": 4.2855,
      "step": 14415
    },
    {
      "epoch": 0.387478167405616,
      "grad_norm": 2.410067319869995,
      "learning_rate": 3.104317036633529e-05,
      "loss": 4.2585,
      "step": 14420
    },
    {
      "epoch": 0.38761252183259437,
      "grad_norm": 2.5437259674072266,
      "learning_rate": 3.103636116028871e-05,
      "loss": 4.2653,
      "step": 14425
    },
    {
      "epoch": 0.38774687625957277,
      "grad_norm": 2.3848185539245605,
      "learning_rate": 3.1029551954242136e-05,
      "loss": 4.3734,
      "step": 14430
    },
    {
      "epoch": 0.3878812306865511,
      "grad_norm": 2.6258974075317383,
      "learning_rate": 3.102274274819556e-05,
      "loss": 4.2421,
      "step": 14435
    },
    {
      "epoch": 0.3880155851135295,
      "grad_norm": 2.394111394882202,
      "learning_rate": 3.101593354214899e-05,
      "loss": 4.2514,
      "step": 14440
    },
    {
      "epoch": 0.38814993954050786,
      "grad_norm": 2.386808156967163,
      "learning_rate": 3.1009124336102416e-05,
      "loss": 4.1963,
      "step": 14445
    },
    {
      "epoch": 0.3882842939674862,
      "grad_norm": 2.3925538063049316,
      "learning_rate": 3.100231513005584e-05,
      "loss": 4.2571,
      "step": 14450
    },
    {
      "epoch": 0.3884186483944646,
      "grad_norm": 2.5595006942749023,
      "learning_rate": 3.099550592400926e-05,
      "loss": 4.3216,
      "step": 14455
    },
    {
      "epoch": 0.38855300282144295,
      "grad_norm": 2.333843469619751,
      "learning_rate": 3.098869671796269e-05,
      "loss": 4.3402,
      "step": 14460
    },
    {
      "epoch": 0.38868735724842135,
      "grad_norm": 2.466310977935791,
      "learning_rate": 3.098188751191611e-05,
      "loss": 4.2875,
      "step": 14465
    },
    {
      "epoch": 0.3888217116753997,
      "grad_norm": 2.2852773666381836,
      "learning_rate": 3.0975078305869534e-05,
      "loss": 4.3023,
      "step": 14470
    },
    {
      "epoch": 0.3889560661023781,
      "grad_norm": 2.5199766159057617,
      "learning_rate": 3.096826909982296e-05,
      "loss": 4.3194,
      "step": 14475
    },
    {
      "epoch": 0.38909042052935644,
      "grad_norm": 2.1740705966949463,
      "learning_rate": 3.0961459893776385e-05,
      "loss": 4.4294,
      "step": 14480
    },
    {
      "epoch": 0.3892247749563348,
      "grad_norm": 2.358670234680176,
      "learning_rate": 3.0954650687729814e-05,
      "loss": 4.3519,
      "step": 14485
    },
    {
      "epoch": 0.3893591293833132,
      "grad_norm": 2.6331212520599365,
      "learning_rate": 3.094784148168324e-05,
      "loss": 4.099,
      "step": 14490
    },
    {
      "epoch": 0.38949348381029153,
      "grad_norm": 2.4660329818725586,
      "learning_rate": 3.094103227563666e-05,
      "loss": 4.3083,
      "step": 14495
    },
    {
      "epoch": 0.38962783823726993,
      "grad_norm": 2.4906811714172363,
      "learning_rate": 3.093422306959009e-05,
      "loss": 4.3835,
      "step": 14500
    },
    {
      "epoch": 0.3897621926642483,
      "grad_norm": 2.560702323913574,
      "learning_rate": 3.092741386354352e-05,
      "loss": 4.3658,
      "step": 14505
    },
    {
      "epoch": 0.3898965470912267,
      "grad_norm": 2.466280698776245,
      "learning_rate": 3.092060465749694e-05,
      "loss": 4.3304,
      "step": 14510
    },
    {
      "epoch": 0.390030901518205,
      "grad_norm": 2.138554096221924,
      "learning_rate": 3.091379545145036e-05,
      "loss": 4.3575,
      "step": 14515
    },
    {
      "epoch": 0.3901652559451834,
      "grad_norm": 2.5083069801330566,
      "learning_rate": 3.0906986245403784e-05,
      "loss": 4.3453,
      "step": 14520
    },
    {
      "epoch": 0.39029961037216176,
      "grad_norm": 2.420567750930786,
      "learning_rate": 3.090017703935721e-05,
      "loss": 4.2272,
      "step": 14525
    },
    {
      "epoch": 0.3904339647991401,
      "grad_norm": 2.2206883430480957,
      "learning_rate": 3.0893367833310635e-05,
      "loss": 4.1968,
      "step": 14530
    },
    {
      "epoch": 0.3905683192261185,
      "grad_norm": 2.1825122833251953,
      "learning_rate": 3.0886558627264064e-05,
      "loss": 4.2142,
      "step": 14535
    },
    {
      "epoch": 0.39070267365309685,
      "grad_norm": 2.488612413406372,
      "learning_rate": 3.0879749421217486e-05,
      "loss": 4.1741,
      "step": 14540
    },
    {
      "epoch": 0.39083702808007525,
      "grad_norm": 2.635913372039795,
      "learning_rate": 3.0872940215170916e-05,
      "loss": 4.2384,
      "step": 14545
    },
    {
      "epoch": 0.3909713825070536,
      "grad_norm": 2.381077766418457,
      "learning_rate": 3.086613100912434e-05,
      "loss": 4.3086,
      "step": 14550
    },
    {
      "epoch": 0.391105736934032,
      "grad_norm": 2.255950689315796,
      "learning_rate": 3.085932180307776e-05,
      "loss": 4.2243,
      "step": 14555
    },
    {
      "epoch": 0.39124009136101034,
      "grad_norm": 2.60119891166687,
      "learning_rate": 3.085251259703118e-05,
      "loss": 4.3109,
      "step": 14560
    },
    {
      "epoch": 0.3913744457879887,
      "grad_norm": 2.424812078475952,
      "learning_rate": 3.084570339098462e-05,
      "loss": 4.2584,
      "step": 14565
    },
    {
      "epoch": 0.3915088002149671,
      "grad_norm": 2.320416212081909,
      "learning_rate": 3.083889418493804e-05,
      "loss": 4.1829,
      "step": 14570
    },
    {
      "epoch": 0.39164315464194543,
      "grad_norm": 2.7615835666656494,
      "learning_rate": 3.083208497889146e-05,
      "loss": 4.3671,
      "step": 14575
    },
    {
      "epoch": 0.39177750906892383,
      "grad_norm": 2.309784173965454,
      "learning_rate": 3.0825275772844885e-05,
      "loss": 4.3134,
      "step": 14580
    },
    {
      "epoch": 0.3919118634959022,
      "grad_norm": 2.4076573848724365,
      "learning_rate": 3.0818466566798314e-05,
      "loss": 4.2382,
      "step": 14585
    },
    {
      "epoch": 0.3920462179228806,
      "grad_norm": 2.3375957012176514,
      "learning_rate": 3.0811657360751736e-05,
      "loss": 4.3127,
      "step": 14590
    },
    {
      "epoch": 0.3921805723498589,
      "grad_norm": 2.443911552429199,
      "learning_rate": 3.0804848154705165e-05,
      "loss": 4.2665,
      "step": 14595
    },
    {
      "epoch": 0.39231492677683727,
      "grad_norm": 2.487468957901001,
      "learning_rate": 3.079803894865859e-05,
      "loss": 4.2395,
      "step": 14600
    },
    {
      "epoch": 0.39244928120381567,
      "grad_norm": 2.3549230098724365,
      "learning_rate": 3.079122974261202e-05,
      "loss": 4.3414,
      "step": 14605
    },
    {
      "epoch": 0.392583635630794,
      "grad_norm": 2.296757221221924,
      "learning_rate": 3.078442053656544e-05,
      "loss": 4.3853,
      "step": 14610
    },
    {
      "epoch": 0.3927179900577724,
      "grad_norm": 2.708432674407959,
      "learning_rate": 3.077761133051886e-05,
      "loss": 4.3558,
      "step": 14615
    },
    {
      "epoch": 0.39285234448475076,
      "grad_norm": 2.183584213256836,
      "learning_rate": 3.0770802124472283e-05,
      "loss": 4.3252,
      "step": 14620
    },
    {
      "epoch": 0.39298669891172916,
      "grad_norm": 2.314103364944458,
      "learning_rate": 3.076399291842571e-05,
      "loss": 4.2631,
      "step": 14625
    },
    {
      "epoch": 0.3931210533387075,
      "grad_norm": 2.1969337463378906,
      "learning_rate": 3.075718371237914e-05,
      "loss": 4.1227,
      "step": 14630
    },
    {
      "epoch": 0.3932554077656859,
      "grad_norm": 2.2576987743377686,
      "learning_rate": 3.0750374506332564e-05,
      "loss": 4.3324,
      "step": 14635
    },
    {
      "epoch": 0.39338976219266425,
      "grad_norm": 2.5025899410247803,
      "learning_rate": 3.0743565300285986e-05,
      "loss": 4.3568,
      "step": 14640
    },
    {
      "epoch": 0.3935241166196426,
      "grad_norm": 2.507436752319336,
      "learning_rate": 3.073675609423941e-05,
      "loss": 4.3309,
      "step": 14645
    },
    {
      "epoch": 0.393658471046621,
      "grad_norm": 2.682708978652954,
      "learning_rate": 3.072994688819284e-05,
      "loss": 4.3191,
      "step": 14650
    },
    {
      "epoch": 0.39379282547359934,
      "grad_norm": 2.4889323711395264,
      "learning_rate": 3.0723137682146266e-05,
      "loss": 4.2975,
      "step": 14655
    },
    {
      "epoch": 0.39392717990057774,
      "grad_norm": 2.4475367069244385,
      "learning_rate": 3.071632847609969e-05,
      "loss": 4.3637,
      "step": 14660
    },
    {
      "epoch": 0.3940615343275561,
      "grad_norm": 2.423344135284424,
      "learning_rate": 3.070951927005311e-05,
      "loss": 4.3865,
      "step": 14665
    },
    {
      "epoch": 0.3941958887545345,
      "grad_norm": 2.3686540126800537,
      "learning_rate": 3.070271006400654e-05,
      "loss": 4.2723,
      "step": 14670
    },
    {
      "epoch": 0.39433024318151283,
      "grad_norm": 2.49894118309021,
      "learning_rate": 3.069590085795996e-05,
      "loss": 4.2414,
      "step": 14675
    },
    {
      "epoch": 0.3944645976084912,
      "grad_norm": 2.3771181106567383,
      "learning_rate": 3.0689091651913385e-05,
      "loss": 4.3081,
      "step": 14680
    },
    {
      "epoch": 0.3945989520354696,
      "grad_norm": 2.3944599628448486,
      "learning_rate": 3.0682282445866814e-05,
      "loss": 4.3964,
      "step": 14685
    },
    {
      "epoch": 0.3947333064624479,
      "grad_norm": 2.4124321937561035,
      "learning_rate": 3.067547323982024e-05,
      "loss": 4.1933,
      "step": 14690
    },
    {
      "epoch": 0.3948676608894263,
      "grad_norm": 2.43300724029541,
      "learning_rate": 3.0668664033773665e-05,
      "loss": 4.3368,
      "step": 14695
    },
    {
      "epoch": 0.39500201531640466,
      "grad_norm": 2.250229597091675,
      "learning_rate": 3.066185482772709e-05,
      "loss": 4.376,
      "step": 14700
    },
    {
      "epoch": 0.39513636974338306,
      "grad_norm": 2.409778594970703,
      "learning_rate": 3.065504562168051e-05,
      "loss": 4.3241,
      "step": 14705
    },
    {
      "epoch": 0.3952707241703614,
      "grad_norm": 2.294586420059204,
      "learning_rate": 3.064823641563394e-05,
      "loss": 4.2036,
      "step": 14710
    },
    {
      "epoch": 0.3954050785973398,
      "grad_norm": 2.2948389053344727,
      "learning_rate": 3.064142720958737e-05,
      "loss": 4.2007,
      "step": 14715
    },
    {
      "epoch": 0.39553943302431815,
      "grad_norm": 2.7229578495025635,
      "learning_rate": 3.063461800354079e-05,
      "loss": 4.3387,
      "step": 14720
    },
    {
      "epoch": 0.3956737874512965,
      "grad_norm": 2.3358476161956787,
      "learning_rate": 3.062780879749421e-05,
      "loss": 4.324,
      "step": 14725
    },
    {
      "epoch": 0.3958081418782749,
      "grad_norm": 2.2669789791107178,
      "learning_rate": 3.062099959144764e-05,
      "loss": 4.2444,
      "step": 14730
    },
    {
      "epoch": 0.39594249630525324,
      "grad_norm": 2.391108512878418,
      "learning_rate": 3.061419038540106e-05,
      "loss": 4.2416,
      "step": 14735
    },
    {
      "epoch": 0.39607685073223164,
      "grad_norm": 2.172057628631592,
      "learning_rate": 3.0607381179354486e-05,
      "loss": 4.2266,
      "step": 14740
    },
    {
      "epoch": 0.39621120515921,
      "grad_norm": 2.415519952774048,
      "learning_rate": 3.0600571973307915e-05,
      "loss": 4.2517,
      "step": 14745
    },
    {
      "epoch": 0.3963455595861884,
      "grad_norm": 2.3594329357147217,
      "learning_rate": 3.0593762767261344e-05,
      "loss": 4.2522,
      "step": 14750
    },
    {
      "epoch": 0.39647991401316673,
      "grad_norm": 2.3897271156311035,
      "learning_rate": 3.0586953561214766e-05,
      "loss": 4.3044,
      "step": 14755
    },
    {
      "epoch": 0.3966142684401451,
      "grad_norm": 2.520068407058716,
      "learning_rate": 3.058014435516819e-05,
      "loss": 4.3266,
      "step": 14760
    },
    {
      "epoch": 0.3967486228671235,
      "grad_norm": 2.099815845489502,
      "learning_rate": 3.057333514912161e-05,
      "loss": 4.3265,
      "step": 14765
    },
    {
      "epoch": 0.3968829772941018,
      "grad_norm": 2.750869035720825,
      "learning_rate": 3.056652594307503e-05,
      "loss": 4.322,
      "step": 14770
    },
    {
      "epoch": 0.3970173317210802,
      "grad_norm": 2.6960182189941406,
      "learning_rate": 3.055971673702847e-05,
      "loss": 4.2079,
      "step": 14775
    },
    {
      "epoch": 0.39715168614805857,
      "grad_norm": 2.3361809253692627,
      "learning_rate": 3.055290753098189e-05,
      "loss": 4.2933,
      "step": 14780
    },
    {
      "epoch": 0.39728604057503697,
      "grad_norm": 2.231854200363159,
      "learning_rate": 3.054609832493531e-05,
      "loss": 4.2878,
      "step": 14785
    },
    {
      "epoch": 0.3974203950020153,
      "grad_norm": 2.33640456199646,
      "learning_rate": 3.0539289118888735e-05,
      "loss": 4.2761,
      "step": 14790
    },
    {
      "epoch": 0.39755474942899366,
      "grad_norm": 2.4944357872009277,
      "learning_rate": 3.0532479912842164e-05,
      "loss": 4.2499,
      "step": 14795
    },
    {
      "epoch": 0.39768910385597206,
      "grad_norm": 2.3984992504119873,
      "learning_rate": 3.052567070679559e-05,
      "loss": 4.1777,
      "step": 14800
    },
    {
      "epoch": 0.3978234582829504,
      "grad_norm": 2.363731622695923,
      "learning_rate": 3.0518861500749016e-05,
      "loss": 4.3235,
      "step": 14805
    },
    {
      "epoch": 0.3979578127099288,
      "grad_norm": 2.428506374359131,
      "learning_rate": 3.0512052294702438e-05,
      "loss": 4.2432,
      "step": 14810
    },
    {
      "epoch": 0.39809216713690715,
      "grad_norm": 2.2954978942871094,
      "learning_rate": 3.0505243088655867e-05,
      "loss": 4.3462,
      "step": 14815
    },
    {
      "epoch": 0.39822652156388555,
      "grad_norm": 2.387347936630249,
      "learning_rate": 3.049843388260929e-05,
      "loss": 4.2013,
      "step": 14820
    },
    {
      "epoch": 0.3983608759908639,
      "grad_norm": 2.5069823265075684,
      "learning_rate": 3.0491624676562715e-05,
      "loss": 4.2306,
      "step": 14825
    },
    {
      "epoch": 0.3984952304178423,
      "grad_norm": 2.564117670059204,
      "learning_rate": 3.0484815470516137e-05,
      "loss": 4.2126,
      "step": 14830
    },
    {
      "epoch": 0.39862958484482064,
      "grad_norm": 2.3405795097351074,
      "learning_rate": 3.0478006264469566e-05,
      "loss": 4.3728,
      "step": 14835
    },
    {
      "epoch": 0.398763939271799,
      "grad_norm": 2.156245470046997,
      "learning_rate": 3.047119705842299e-05,
      "loss": 4.2642,
      "step": 14840
    },
    {
      "epoch": 0.3988982936987774,
      "grad_norm": 2.3241055011749268,
      "learning_rate": 3.0464387852376414e-05,
      "loss": 4.2607,
      "step": 14845
    },
    {
      "epoch": 0.39903264812575573,
      "grad_norm": 2.4331064224243164,
      "learning_rate": 3.0457578646329837e-05,
      "loss": 4.2721,
      "step": 14850
    },
    {
      "epoch": 0.39916700255273413,
      "grad_norm": 2.3952794075012207,
      "learning_rate": 3.0450769440283266e-05,
      "loss": 4.266,
      "step": 14855
    },
    {
      "epoch": 0.3993013569797125,
      "grad_norm": 2.445282220840454,
      "learning_rate": 3.044396023423669e-05,
      "loss": 4.2446,
      "step": 14860
    },
    {
      "epoch": 0.3994357114066909,
      "grad_norm": 2.4237923622131348,
      "learning_rate": 3.0437151028190114e-05,
      "loss": 4.1766,
      "step": 14865
    },
    {
      "epoch": 0.3995700658336692,
      "grad_norm": 2.685258388519287,
      "learning_rate": 3.043034182214354e-05,
      "loss": 4.3549,
      "step": 14870
    },
    {
      "epoch": 0.39970442026064756,
      "grad_norm": 2.370173215866089,
      "learning_rate": 3.0423532616096968e-05,
      "loss": 4.3332,
      "step": 14875
    },
    {
      "epoch": 0.39983877468762596,
      "grad_norm": 2.275838613510132,
      "learning_rate": 3.041672341005039e-05,
      "loss": 4.3136,
      "step": 14880
    },
    {
      "epoch": 0.3999731291146043,
      "grad_norm": 2.8002326488494873,
      "learning_rate": 3.0409914204003813e-05,
      "loss": 4.3134,
      "step": 14885
    },
    {
      "epoch": 0.4001074835415827,
      "grad_norm": 2.200334072113037,
      "learning_rate": 3.040310499795724e-05,
      "loss": 4.4386,
      "step": 14890
    },
    {
      "epoch": 0.40024183796856105,
      "grad_norm": 2.3915858268737793,
      "learning_rate": 3.0396295791910667e-05,
      "loss": 4.289,
      "step": 14895
    },
    {
      "epoch": 0.40037619239553945,
      "grad_norm": 2.397400379180908,
      "learning_rate": 3.038948658586409e-05,
      "loss": 4.2631,
      "step": 14900
    },
    {
      "epoch": 0.4005105468225178,
      "grad_norm": 2.279953718185425,
      "learning_rate": 3.0382677379817515e-05,
      "loss": 4.2948,
      "step": 14905
    },
    {
      "epoch": 0.40064490124949614,
      "grad_norm": 2.3294737339019775,
      "learning_rate": 3.0375868173770938e-05,
      "loss": 4.2607,
      "step": 14910
    },
    {
      "epoch": 0.40077925567647454,
      "grad_norm": 2.601576328277588,
      "learning_rate": 3.0369058967724367e-05,
      "loss": 4.2404,
      "step": 14915
    },
    {
      "epoch": 0.4009136101034529,
      "grad_norm": 2.3031506538391113,
      "learning_rate": 3.0362249761677792e-05,
      "loss": 4.2878,
      "step": 14920
    },
    {
      "epoch": 0.4010479645304313,
      "grad_norm": 2.8369812965393066,
      "learning_rate": 3.0355440555631215e-05,
      "loss": 4.3155,
      "step": 14925
    },
    {
      "epoch": 0.40118231895740963,
      "grad_norm": 2.2411205768585205,
      "learning_rate": 3.0348631349584637e-05,
      "loss": 4.3156,
      "step": 14930
    },
    {
      "epoch": 0.40131667338438803,
      "grad_norm": 2.224097728729248,
      "learning_rate": 3.0341822143538063e-05,
      "loss": 4.1881,
      "step": 14935
    },
    {
      "epoch": 0.4014510278113664,
      "grad_norm": 2.1425905227661133,
      "learning_rate": 3.033501293749149e-05,
      "loss": 4.216,
      "step": 14940
    },
    {
      "epoch": 0.4015853822383448,
      "grad_norm": 2.3168671131134033,
      "learning_rate": 3.0328203731444914e-05,
      "loss": 4.2863,
      "step": 14945
    },
    {
      "epoch": 0.4017197366653231,
      "grad_norm": 2.4455814361572266,
      "learning_rate": 3.032139452539834e-05,
      "loss": 4.1961,
      "step": 14950
    },
    {
      "epoch": 0.40185409109230147,
      "grad_norm": 2.408461570739746,
      "learning_rate": 3.0314585319351762e-05,
      "loss": 4.2671,
      "step": 14955
    },
    {
      "epoch": 0.40198844551927987,
      "grad_norm": 2.2294387817382812,
      "learning_rate": 3.030777611330519e-05,
      "loss": 4.3296,
      "step": 14960
    },
    {
      "epoch": 0.4021227999462582,
      "grad_norm": 2.4524431228637695,
      "learning_rate": 3.0300966907258617e-05,
      "loss": 4.3318,
      "step": 14965
    },
    {
      "epoch": 0.4022571543732366,
      "grad_norm": 2.2154600620269775,
      "learning_rate": 3.029415770121204e-05,
      "loss": 4.3565,
      "step": 14970
    },
    {
      "epoch": 0.40239150880021496,
      "grad_norm": 2.629197359085083,
      "learning_rate": 3.0287348495165464e-05,
      "loss": 4.2913,
      "step": 14975
    },
    {
      "epoch": 0.40252586322719336,
      "grad_norm": 2.20185923576355,
      "learning_rate": 3.0280539289118893e-05,
      "loss": 4.3749,
      "step": 14980
    },
    {
      "epoch": 0.4026602176541717,
      "grad_norm": 2.567887306213379,
      "learning_rate": 3.0273730083072316e-05,
      "loss": 4.2403,
      "step": 14985
    },
    {
      "epoch": 0.40279457208115005,
      "grad_norm": 2.561450719833374,
      "learning_rate": 3.0266920877025738e-05,
      "loss": 4.2801,
      "step": 14990
    },
    {
      "epoch": 0.40292892650812845,
      "grad_norm": 2.7562851905822754,
      "learning_rate": 3.0260111670979164e-05,
      "loss": 4.1768,
      "step": 14995
    },
    {
      "epoch": 0.4030632809351068,
      "grad_norm": 2.4966940879821777,
      "learning_rate": 3.0253302464932593e-05,
      "loss": 4.364,
      "step": 15000
    },
    {
      "epoch": 0.4031976353620852,
      "grad_norm": 2.5060904026031494,
      "learning_rate": 3.0246493258886015e-05,
      "loss": 4.308,
      "step": 15005
    },
    {
      "epoch": 0.40333198978906354,
      "grad_norm": 2.2610812187194824,
      "learning_rate": 3.023968405283944e-05,
      "loss": 4.2393,
      "step": 15010
    },
    {
      "epoch": 0.40346634421604194,
      "grad_norm": 2.411501884460449,
      "learning_rate": 3.0232874846792863e-05,
      "loss": 4.254,
      "step": 15015
    },
    {
      "epoch": 0.4036006986430203,
      "grad_norm": 2.28826642036438,
      "learning_rate": 3.0226065640746292e-05,
      "loss": 4.3133,
      "step": 15020
    },
    {
      "epoch": 0.4037350530699987,
      "grad_norm": 2.2608704566955566,
      "learning_rate": 3.0219256434699718e-05,
      "loss": 4.1253,
      "step": 15025
    },
    {
      "epoch": 0.40386940749697703,
      "grad_norm": 2.1428303718566895,
      "learning_rate": 3.021244722865314e-05,
      "loss": 4.162,
      "step": 15030
    },
    {
      "epoch": 0.4040037619239554,
      "grad_norm": 2.4092249870300293,
      "learning_rate": 3.0205638022606562e-05,
      "loss": 4.1827,
      "step": 15035
    },
    {
      "epoch": 0.4041381163509338,
      "grad_norm": 2.355943202972412,
      "learning_rate": 3.0198828816559995e-05,
      "loss": 4.2591,
      "step": 15040
    },
    {
      "epoch": 0.4042724707779121,
      "grad_norm": 2.637709379196167,
      "learning_rate": 3.0192019610513417e-05,
      "loss": 4.305,
      "step": 15045
    },
    {
      "epoch": 0.4044068252048905,
      "grad_norm": 2.2113492488861084,
      "learning_rate": 3.018521040446684e-05,
      "loss": 4.3268,
      "step": 15050
    },
    {
      "epoch": 0.40454117963186886,
      "grad_norm": 2.410249948501587,
      "learning_rate": 3.0178401198420265e-05,
      "loss": 4.3034,
      "step": 15055
    },
    {
      "epoch": 0.40467553405884726,
      "grad_norm": 2.1468658447265625,
      "learning_rate": 3.0171591992373694e-05,
      "loss": 4.2177,
      "step": 15060
    },
    {
      "epoch": 0.4048098884858256,
      "grad_norm": 2.650271415710449,
      "learning_rate": 3.0164782786327116e-05,
      "loss": 4.224,
      "step": 15065
    },
    {
      "epoch": 0.40494424291280395,
      "grad_norm": 2.519599676132202,
      "learning_rate": 3.0157973580280542e-05,
      "loss": 4.2121,
      "step": 15070
    },
    {
      "epoch": 0.40507859733978235,
      "grad_norm": 2.3711276054382324,
      "learning_rate": 3.0151164374233964e-05,
      "loss": 4.315,
      "step": 15075
    },
    {
      "epoch": 0.4052129517667607,
      "grad_norm": 2.265425205230713,
      "learning_rate": 3.014435516818739e-05,
      "loss": 4.1935,
      "step": 15080
    },
    {
      "epoch": 0.4053473061937391,
      "grad_norm": 2.3750174045562744,
      "learning_rate": 3.013754596214082e-05,
      "loss": 4.2176,
      "step": 15085
    },
    {
      "epoch": 0.40548166062071744,
      "grad_norm": 2.3182108402252197,
      "learning_rate": 3.013073675609424e-05,
      "loss": 4.265,
      "step": 15090
    },
    {
      "epoch": 0.40561601504769584,
      "grad_norm": 2.5446548461914062,
      "learning_rate": 3.0123927550047663e-05,
      "loss": 4.1888,
      "step": 15095
    },
    {
      "epoch": 0.4057503694746742,
      "grad_norm": 2.3328607082366943,
      "learning_rate": 3.011711834400109e-05,
      "loss": 4.2458,
      "step": 15100
    },
    {
      "epoch": 0.40588472390165253,
      "grad_norm": 2.4188597202301025,
      "learning_rate": 3.0110309137954518e-05,
      "loss": 4.3838,
      "step": 15105
    },
    {
      "epoch": 0.40601907832863093,
      "grad_norm": 2.267055034637451,
      "learning_rate": 3.010349993190794e-05,
      "loss": 4.3703,
      "step": 15110
    },
    {
      "epoch": 0.4061534327556093,
      "grad_norm": 2.4511544704437256,
      "learning_rate": 3.0096690725861366e-05,
      "loss": 4.2743,
      "step": 15115
    },
    {
      "epoch": 0.4062877871825877,
      "grad_norm": 2.320572853088379,
      "learning_rate": 3.0089881519814788e-05,
      "loss": 4.0622,
      "step": 15120
    },
    {
      "epoch": 0.406422141609566,
      "grad_norm": 2.3764612674713135,
      "learning_rate": 3.0083072313768217e-05,
      "loss": 4.3091,
      "step": 15125
    },
    {
      "epoch": 0.4065564960365444,
      "grad_norm": 2.329913377761841,
      "learning_rate": 3.0076263107721643e-05,
      "loss": 4.2826,
      "step": 15130
    },
    {
      "epoch": 0.40669085046352277,
      "grad_norm": 2.2480175495147705,
      "learning_rate": 3.0069453901675065e-05,
      "loss": 4.3427,
      "step": 15135
    },
    {
      "epoch": 0.40682520489050117,
      "grad_norm": 2.583582878112793,
      "learning_rate": 3.0062644695628487e-05,
      "loss": 4.1923,
      "step": 15140
    },
    {
      "epoch": 0.4069595593174795,
      "grad_norm": 2.2606606483459473,
      "learning_rate": 3.005583548958192e-05,
      "loss": 4.2712,
      "step": 15145
    },
    {
      "epoch": 0.40709391374445786,
      "grad_norm": 2.2507436275482178,
      "learning_rate": 3.0049026283535342e-05,
      "loss": 4.1473,
      "step": 15150
    },
    {
      "epoch": 0.40722826817143626,
      "grad_norm": 2.137530565261841,
      "learning_rate": 3.0042217077488764e-05,
      "loss": 4.2804,
      "step": 15155
    },
    {
      "epoch": 0.4073626225984146,
      "grad_norm": 2.306863307952881,
      "learning_rate": 3.003540787144219e-05,
      "loss": 4.2638,
      "step": 15160
    },
    {
      "epoch": 0.407496977025393,
      "grad_norm": 2.3660683631896973,
      "learning_rate": 3.002859866539562e-05,
      "loss": 4.405,
      "step": 15165
    },
    {
      "epoch": 0.40763133145237135,
      "grad_norm": 2.4482462406158447,
      "learning_rate": 3.002178945934904e-05,
      "loss": 4.288,
      "step": 15170
    },
    {
      "epoch": 0.40776568587934975,
      "grad_norm": 2.33117938041687,
      "learning_rate": 3.0014980253302467e-05,
      "loss": 4.3375,
      "step": 15175
    },
    {
      "epoch": 0.4079000403063281,
      "grad_norm": 2.2881529331207275,
      "learning_rate": 3.000817104725589e-05,
      "loss": 4.2885,
      "step": 15180
    },
    {
      "epoch": 0.40803439473330644,
      "grad_norm": 2.23380446434021,
      "learning_rate": 3.000136184120932e-05,
      "loss": 4.0958,
      "step": 15185
    },
    {
      "epoch": 0.40816874916028484,
      "grad_norm": 2.4789693355560303,
      "learning_rate": 2.9994552635162744e-05,
      "loss": 4.3061,
      "step": 15190
    },
    {
      "epoch": 0.4083031035872632,
      "grad_norm": 2.3406922817230225,
      "learning_rate": 2.9987743429116166e-05,
      "loss": 4.3921,
      "step": 15195
    },
    {
      "epoch": 0.4084374580142416,
      "grad_norm": 2.4371888637542725,
      "learning_rate": 2.998093422306959e-05,
      "loss": 4.2097,
      "step": 15200
    },
    {
      "epoch": 0.40857181244121993,
      "grad_norm": 2.0882675647735596,
      "learning_rate": 2.9974125017023018e-05,
      "loss": 4.1242,
      "step": 15205
    },
    {
      "epoch": 0.40870616686819833,
      "grad_norm": 2.570497989654541,
      "learning_rate": 2.9967315810976443e-05,
      "loss": 4.3416,
      "step": 15210
    },
    {
      "epoch": 0.4088405212951767,
      "grad_norm": 2.3471245765686035,
      "learning_rate": 2.9960506604929866e-05,
      "loss": 4.3881,
      "step": 15215
    },
    {
      "epoch": 0.408974875722155,
      "grad_norm": 2.3258862495422363,
      "learning_rate": 2.995369739888329e-05,
      "loss": 4.3062,
      "step": 15220
    },
    {
      "epoch": 0.4091092301491334,
      "grad_norm": 2.3135623931884766,
      "learning_rate": 2.9946888192836713e-05,
      "loss": 4.2543,
      "step": 15225
    },
    {
      "epoch": 0.40924358457611176,
      "grad_norm": 2.445925712585449,
      "learning_rate": 2.9940078986790142e-05,
      "loss": 4.2006,
      "step": 15230
    },
    {
      "epoch": 0.40937793900309016,
      "grad_norm": 2.381124973297119,
      "learning_rate": 2.9933269780743568e-05,
      "loss": 4.2782,
      "step": 15235
    },
    {
      "epoch": 0.4095122934300685,
      "grad_norm": 2.4999194145202637,
      "learning_rate": 2.992646057469699e-05,
      "loss": 4.2521,
      "step": 15240
    },
    {
      "epoch": 0.4096466478570469,
      "grad_norm": 2.268766164779663,
      "learning_rate": 2.9919651368650413e-05,
      "loss": 4.1848,
      "step": 15245
    },
    {
      "epoch": 0.40978100228402525,
      "grad_norm": 2.3853793144226074,
      "learning_rate": 2.9912842162603845e-05,
      "loss": 4.2844,
      "step": 15250
    },
    {
      "epoch": 0.40991535671100365,
      "grad_norm": 2.2645647525787354,
      "learning_rate": 2.9906032956557267e-05,
      "loss": 4.2929,
      "step": 15255
    },
    {
      "epoch": 0.410049711137982,
      "grad_norm": 2.361607074737549,
      "learning_rate": 2.989922375051069e-05,
      "loss": 4.3022,
      "step": 15260
    },
    {
      "epoch": 0.41018406556496034,
      "grad_norm": 2.481165885925293,
      "learning_rate": 2.9892414544464115e-05,
      "loss": 4.2606,
      "step": 15265
    },
    {
      "epoch": 0.41031841999193874,
      "grad_norm": 2.334106206893921,
      "learning_rate": 2.9885605338417544e-05,
      "loss": 4.3242,
      "step": 15270
    },
    {
      "epoch": 0.4104527744189171,
      "grad_norm": 2.417473554611206,
      "learning_rate": 2.9878796132370967e-05,
      "loss": 4.1249,
      "step": 15275
    },
    {
      "epoch": 0.4105871288458955,
      "grad_norm": 2.33376145362854,
      "learning_rate": 2.9871986926324392e-05,
      "loss": 4.3304,
      "step": 15280
    },
    {
      "epoch": 0.41072148327287383,
      "grad_norm": 2.1903340816497803,
      "learning_rate": 2.9865177720277815e-05,
      "loss": 4.2321,
      "step": 15285
    },
    {
      "epoch": 0.41085583769985223,
      "grad_norm": 2.3100905418395996,
      "learning_rate": 2.9858368514231244e-05,
      "loss": 4.3124,
      "step": 15290
    },
    {
      "epoch": 0.4109901921268306,
      "grad_norm": 2.26353120803833,
      "learning_rate": 2.985155930818467e-05,
      "loss": 4.2926,
      "step": 15295
    },
    {
      "epoch": 0.4111245465538089,
      "grad_norm": 2.370575428009033,
      "learning_rate": 2.984475010213809e-05,
      "loss": 4.2766,
      "step": 15300
    },
    {
      "epoch": 0.4112589009807873,
      "grad_norm": 2.46513032913208,
      "learning_rate": 2.9837940896091514e-05,
      "loss": 4.29,
      "step": 15305
    },
    {
      "epoch": 0.41139325540776567,
      "grad_norm": 2.3344225883483887,
      "learning_rate": 2.9831131690044943e-05,
      "loss": 4.2037,
      "step": 15310
    },
    {
      "epoch": 0.41152760983474407,
      "grad_norm": 2.2919318675994873,
      "learning_rate": 2.982432248399837e-05,
      "loss": 4.3345,
      "step": 15315
    },
    {
      "epoch": 0.4116619642617224,
      "grad_norm": 2.172579526901245,
      "learning_rate": 2.981751327795179e-05,
      "loss": 4.1602,
      "step": 15320
    },
    {
      "epoch": 0.4117963186887008,
      "grad_norm": 2.434432029724121,
      "learning_rate": 2.9810704071905216e-05,
      "loss": 4.2995,
      "step": 15325
    },
    {
      "epoch": 0.41193067311567916,
      "grad_norm": 2.534857988357544,
      "learning_rate": 2.9803894865858645e-05,
      "loss": 4.313,
      "step": 15330
    },
    {
      "epoch": 0.41206502754265756,
      "grad_norm": 2.4175307750701904,
      "learning_rate": 2.9797085659812068e-05,
      "loss": 4.3759,
      "step": 15335
    },
    {
      "epoch": 0.4121993819696359,
      "grad_norm": 2.5274789333343506,
      "learning_rate": 2.9790276453765493e-05,
      "loss": 4.3469,
      "step": 15340
    },
    {
      "epoch": 0.41233373639661425,
      "grad_norm": 2.3748111724853516,
      "learning_rate": 2.9783467247718916e-05,
      "loss": 4.179,
      "step": 15345
    },
    {
      "epoch": 0.41246809082359265,
      "grad_norm": 2.3223702907562256,
      "learning_rate": 2.9776658041672345e-05,
      "loss": 4.1739,
      "step": 15350
    },
    {
      "epoch": 0.412602445250571,
      "grad_norm": 2.3797850608825684,
      "learning_rate": 2.976984883562577e-05,
      "loss": 4.178,
      "step": 15355
    },
    {
      "epoch": 0.4127367996775494,
      "grad_norm": 2.349032402038574,
      "learning_rate": 2.9763039629579193e-05,
      "loss": 4.3435,
      "step": 15360
    },
    {
      "epoch": 0.41287115410452774,
      "grad_norm": 2.6877405643463135,
      "learning_rate": 2.9756230423532615e-05,
      "loss": 4.2759,
      "step": 15365
    },
    {
      "epoch": 0.41300550853150614,
      "grad_norm": 2.4556851387023926,
      "learning_rate": 2.974942121748604e-05,
      "loss": 4.1202,
      "step": 15370
    },
    {
      "epoch": 0.4131398629584845,
      "grad_norm": 2.4657938480377197,
      "learning_rate": 2.974261201143947e-05,
      "loss": 4.3019,
      "step": 15375
    },
    {
      "epoch": 0.4132742173854628,
      "grad_norm": 2.379432439804077,
      "learning_rate": 2.9735802805392892e-05,
      "loss": 4.2152,
      "step": 15380
    },
    {
      "epoch": 0.41340857181244123,
      "grad_norm": 2.2791435718536377,
      "learning_rate": 2.9728993599346318e-05,
      "loss": 4.2521,
      "step": 15385
    },
    {
      "epoch": 0.4135429262394196,
      "grad_norm": 2.5904486179351807,
      "learning_rate": 2.972218439329974e-05,
      "loss": 4.0848,
      "step": 15390
    },
    {
      "epoch": 0.413677280666398,
      "grad_norm": 2.5664472579956055,
      "learning_rate": 2.971537518725317e-05,
      "loss": 4.2924,
      "step": 15395
    },
    {
      "epoch": 0.4138116350933763,
      "grad_norm": 2.36448335647583,
      "learning_rate": 2.9708565981206595e-05,
      "loss": 4.3416,
      "step": 15400
    },
    {
      "epoch": 0.4139459895203547,
      "grad_norm": 2.3223648071289062,
      "learning_rate": 2.9701756775160017e-05,
      "loss": 4.3747,
      "step": 15405
    },
    {
      "epoch": 0.41408034394733306,
      "grad_norm": 2.1586551666259766,
      "learning_rate": 2.969494756911344e-05,
      "loss": 4.2933,
      "step": 15410
    },
    {
      "epoch": 0.4142146983743114,
      "grad_norm": 2.236051559448242,
      "learning_rate": 2.9688138363066868e-05,
      "loss": 4.3789,
      "step": 15415
    },
    {
      "epoch": 0.4143490528012898,
      "grad_norm": 2.394160509109497,
      "learning_rate": 2.9681329157020294e-05,
      "loss": 4.2693,
      "step": 15420
    },
    {
      "epoch": 0.41448340722826815,
      "grad_norm": 2.3833975791931152,
      "learning_rate": 2.9674519950973716e-05,
      "loss": 4.2282,
      "step": 15425
    },
    {
      "epoch": 0.41461776165524655,
      "grad_norm": 2.242784023284912,
      "learning_rate": 2.966771074492714e-05,
      "loss": 4.2533,
      "step": 15430
    },
    {
      "epoch": 0.4147521160822249,
      "grad_norm": 2.4441890716552734,
      "learning_rate": 2.966090153888057e-05,
      "loss": 4.2694,
      "step": 15435
    },
    {
      "epoch": 0.4148864705092033,
      "grad_norm": 2.5111019611358643,
      "learning_rate": 2.9654092332833993e-05,
      "loss": 4.2356,
      "step": 15440
    },
    {
      "epoch": 0.41502082493618164,
      "grad_norm": 2.2310760021209717,
      "learning_rate": 2.964728312678742e-05,
      "loss": 4.2776,
      "step": 15445
    },
    {
      "epoch": 0.41515517936316004,
      "grad_norm": 2.354158878326416,
      "learning_rate": 2.964047392074084e-05,
      "loss": 4.2804,
      "step": 15450
    },
    {
      "epoch": 0.4152895337901384,
      "grad_norm": 2.1626672744750977,
      "learning_rate": 2.963366471469427e-05,
      "loss": 4.2974,
      "step": 15455
    },
    {
      "epoch": 0.41542388821711673,
      "grad_norm": 2.473813056945801,
      "learning_rate": 2.9626855508647696e-05,
      "loss": 4.2955,
      "step": 15460
    },
    {
      "epoch": 0.41555824264409513,
      "grad_norm": 2.185945510864258,
      "learning_rate": 2.9620046302601118e-05,
      "loss": 4.1219,
      "step": 15465
    },
    {
      "epoch": 0.4156925970710735,
      "grad_norm": 2.626720428466797,
      "learning_rate": 2.961323709655454e-05,
      "loss": 4.1857,
      "step": 15470
    },
    {
      "epoch": 0.4158269514980519,
      "grad_norm": 2.3311195373535156,
      "learning_rate": 2.960642789050797e-05,
      "loss": 4.2611,
      "step": 15475
    },
    {
      "epoch": 0.4159613059250302,
      "grad_norm": 2.170579195022583,
      "learning_rate": 2.9599618684461395e-05,
      "loss": 4.3102,
      "step": 15480
    },
    {
      "epoch": 0.4160956603520086,
      "grad_norm": 2.331987142562866,
      "learning_rate": 2.9592809478414817e-05,
      "loss": 4.1262,
      "step": 15485
    },
    {
      "epoch": 0.41623001477898697,
      "grad_norm": 2.2168030738830566,
      "learning_rate": 2.9586000272368243e-05,
      "loss": 4.1843,
      "step": 15490
    },
    {
      "epoch": 0.4163643692059653,
      "grad_norm": 2.3450028896331787,
      "learning_rate": 2.9579191066321672e-05,
      "loss": 4.2369,
      "step": 15495
    },
    {
      "epoch": 0.4164987236329437,
      "grad_norm": 2.378307580947876,
      "learning_rate": 2.9572381860275094e-05,
      "loss": 4.223,
      "step": 15500
    },
    {
      "epoch": 0.41663307805992206,
      "grad_norm": 2.3677380084991455,
      "learning_rate": 2.956557265422852e-05,
      "loss": 4.1944,
      "step": 15505
    },
    {
      "epoch": 0.41676743248690046,
      "grad_norm": 2.76792573928833,
      "learning_rate": 2.9558763448181942e-05,
      "loss": 4.3932,
      "step": 15510
    },
    {
      "epoch": 0.4169017869138788,
      "grad_norm": 2.4063220024108887,
      "learning_rate": 2.9551954242135364e-05,
      "loss": 4.2682,
      "step": 15515
    },
    {
      "epoch": 0.4170361413408572,
      "grad_norm": 2.0933823585510254,
      "learning_rate": 2.9545145036088793e-05,
      "loss": 4.2798,
      "step": 15520
    },
    {
      "epoch": 0.41717049576783555,
      "grad_norm": 2.150895357131958,
      "learning_rate": 2.953833583004222e-05,
      "loss": 4.3057,
      "step": 15525
    },
    {
      "epoch": 0.4173048501948139,
      "grad_norm": 2.9398670196533203,
      "learning_rate": 2.953152662399564e-05,
      "loss": 4.2759,
      "step": 15530
    },
    {
      "epoch": 0.4174392046217923,
      "grad_norm": 2.4601643085479736,
      "learning_rate": 2.9524717417949067e-05,
      "loss": 4.0938,
      "step": 15535
    },
    {
      "epoch": 0.41757355904877064,
      "grad_norm": 2.6426565647125244,
      "learning_rate": 2.9517908211902496e-05,
      "loss": 4.2446,
      "step": 15540
    },
    {
      "epoch": 0.41770791347574904,
      "grad_norm": 2.5562407970428467,
      "learning_rate": 2.9511099005855918e-05,
      "loss": 4.2428,
      "step": 15545
    },
    {
      "epoch": 0.4178422679027274,
      "grad_norm": 2.6013388633728027,
      "learning_rate": 2.9504289799809344e-05,
      "loss": 4.2926,
      "step": 15550
    },
    {
      "epoch": 0.4179766223297058,
      "grad_norm": 2.298737049102783,
      "learning_rate": 2.9497480593762766e-05,
      "loss": 4.2539,
      "step": 15555
    },
    {
      "epoch": 0.4181109767566841,
      "grad_norm": 2.445235013961792,
      "learning_rate": 2.9490671387716195e-05,
      "loss": 4.2892,
      "step": 15560
    },
    {
      "epoch": 0.4182453311836625,
      "grad_norm": 2.329190492630005,
      "learning_rate": 2.948386218166962e-05,
      "loss": 4.2953,
      "step": 15565
    },
    {
      "epoch": 0.4183796856106409,
      "grad_norm": 2.494016170501709,
      "learning_rate": 2.9477052975623043e-05,
      "loss": 4.1849,
      "step": 15570
    },
    {
      "epoch": 0.4185140400376192,
      "grad_norm": 2.3938581943511963,
      "learning_rate": 2.9470243769576465e-05,
      "loss": 4.202,
      "step": 15575
    },
    {
      "epoch": 0.4186483944645976,
      "grad_norm": 2.4662394523620605,
      "learning_rate": 2.9463434563529894e-05,
      "loss": 4.2463,
      "step": 15580
    },
    {
      "epoch": 0.41878274889157596,
      "grad_norm": 2.4669694900512695,
      "learning_rate": 2.945662535748332e-05,
      "loss": 4.2475,
      "step": 15585
    },
    {
      "epoch": 0.41891710331855436,
      "grad_norm": 2.4376564025878906,
      "learning_rate": 2.9449816151436742e-05,
      "loss": 4.3352,
      "step": 15590
    },
    {
      "epoch": 0.4190514577455327,
      "grad_norm": 2.1778059005737305,
      "learning_rate": 2.9443006945390168e-05,
      "loss": 4.164,
      "step": 15595
    },
    {
      "epoch": 0.4191858121725111,
      "grad_norm": 2.559877395629883,
      "learning_rate": 2.9436197739343597e-05,
      "loss": 4.3062,
      "step": 15600
    },
    {
      "epoch": 0.41932016659948945,
      "grad_norm": 2.3988101482391357,
      "learning_rate": 2.942938853329702e-05,
      "loss": 4.3022,
      "step": 15605
    },
    {
      "epoch": 0.4194545210264678,
      "grad_norm": 2.220656394958496,
      "learning_rate": 2.9422579327250445e-05,
      "loss": 4.3416,
      "step": 15610
    },
    {
      "epoch": 0.4195888754534462,
      "grad_norm": 2.5570366382598877,
      "learning_rate": 2.9415770121203867e-05,
      "loss": 4.1994,
      "step": 15615
    },
    {
      "epoch": 0.41972322988042454,
      "grad_norm": 2.2814064025878906,
      "learning_rate": 2.9408960915157296e-05,
      "loss": 4.1622,
      "step": 15620
    },
    {
      "epoch": 0.41985758430740294,
      "grad_norm": 2.3180532455444336,
      "learning_rate": 2.940215170911072e-05,
      "loss": 4.2521,
      "step": 15625
    },
    {
      "epoch": 0.4199919387343813,
      "grad_norm": 2.4440035820007324,
      "learning_rate": 2.9395342503064144e-05,
      "loss": 4.2457,
      "step": 15630
    },
    {
      "epoch": 0.4201262931613597,
      "grad_norm": 2.3329436779022217,
      "learning_rate": 2.9388533297017567e-05,
      "loss": 4.2131,
      "step": 15635
    },
    {
      "epoch": 0.42026064758833803,
      "grad_norm": 2.373483896255493,
      "learning_rate": 2.9381724090970996e-05,
      "loss": 4.348,
      "step": 15640
    },
    {
      "epoch": 0.42039500201531643,
      "grad_norm": 2.350590944290161,
      "learning_rate": 2.937491488492442e-05,
      "loss": 4.2993,
      "step": 15645
    },
    {
      "epoch": 0.4205293564422948,
      "grad_norm": 2.516461133956909,
      "learning_rate": 2.9368105678877843e-05,
      "loss": 4.2517,
      "step": 15650
    },
    {
      "epoch": 0.4206637108692731,
      "grad_norm": 2.2071080207824707,
      "learning_rate": 2.936129647283127e-05,
      "loss": 4.2219,
      "step": 15655
    },
    {
      "epoch": 0.4207980652962515,
      "grad_norm": 2.14993953704834,
      "learning_rate": 2.9354487266784698e-05,
      "loss": 4.2534,
      "step": 15660
    },
    {
      "epoch": 0.42093241972322987,
      "grad_norm": 2.2577908039093018,
      "learning_rate": 2.934767806073812e-05,
      "loss": 4.2327,
      "step": 15665
    },
    {
      "epoch": 0.42106677415020827,
      "grad_norm": 2.4337992668151855,
      "learning_rate": 2.9340868854691546e-05,
      "loss": 4.1449,
      "step": 15670
    },
    {
      "epoch": 0.4212011285771866,
      "grad_norm": 2.216754913330078,
      "learning_rate": 2.933405964864497e-05,
      "loss": 4.3407,
      "step": 15675
    },
    {
      "epoch": 0.421335483004165,
      "grad_norm": 2.3895459175109863,
      "learning_rate": 2.932725044259839e-05,
      "loss": 4.3221,
      "step": 15680
    },
    {
      "epoch": 0.42146983743114336,
      "grad_norm": 2.391270399093628,
      "learning_rate": 2.932044123655182e-05,
      "loss": 4.2125,
      "step": 15685
    },
    {
      "epoch": 0.4216041918581217,
      "grad_norm": 2.336050271987915,
      "learning_rate": 2.9313632030505245e-05,
      "loss": 4.2911,
      "step": 15690
    },
    {
      "epoch": 0.4217385462851001,
      "grad_norm": 2.289990186691284,
      "learning_rate": 2.9306822824458668e-05,
      "loss": 4.2434,
      "step": 15695
    },
    {
      "epoch": 0.42187290071207845,
      "grad_norm": 2.1537325382232666,
      "learning_rate": 2.9300013618412093e-05,
      "loss": 4.2653,
      "step": 15700
    },
    {
      "epoch": 0.42200725513905685,
      "grad_norm": 2.3626420497894287,
      "learning_rate": 2.9293204412365522e-05,
      "loss": 4.2741,
      "step": 15705
    },
    {
      "epoch": 0.4221416095660352,
      "grad_norm": 2.5895607471466064,
      "learning_rate": 2.9286395206318945e-05,
      "loss": 4.2704,
      "step": 15710
    },
    {
      "epoch": 0.4222759639930136,
      "grad_norm": 2.508665084838867,
      "learning_rate": 2.927958600027237e-05,
      "loss": 4.3302,
      "step": 15715
    },
    {
      "epoch": 0.42241031841999194,
      "grad_norm": 2.4382503032684326,
      "learning_rate": 2.9272776794225793e-05,
      "loss": 4.3504,
      "step": 15720
    },
    {
      "epoch": 0.4225446728469703,
      "grad_norm": 2.6313557624816895,
      "learning_rate": 2.926596758817922e-05,
      "loss": 4.2784,
      "step": 15725
    },
    {
      "epoch": 0.4226790272739487,
      "grad_norm": 2.280480146408081,
      "learning_rate": 2.9259158382132644e-05,
      "loss": 4.3419,
      "step": 15730
    },
    {
      "epoch": 0.422813381700927,
      "grad_norm": 2.2819299697875977,
      "learning_rate": 2.925234917608607e-05,
      "loss": 4.3885,
      "step": 15735
    },
    {
      "epoch": 0.4229477361279054,
      "grad_norm": 2.3130624294281006,
      "learning_rate": 2.9245539970039492e-05,
      "loss": 4.2851,
      "step": 15740
    },
    {
      "epoch": 0.42308209055488377,
      "grad_norm": 2.536611795425415,
      "learning_rate": 2.923873076399292e-05,
      "loss": 4.147,
      "step": 15745
    },
    {
      "epoch": 0.42321644498186217,
      "grad_norm": 2.3833565711975098,
      "learning_rate": 2.9231921557946346e-05,
      "loss": 4.1647,
      "step": 15750
    },
    {
      "epoch": 0.4233507994088405,
      "grad_norm": 2.502312660217285,
      "learning_rate": 2.922511235189977e-05,
      "loss": 4.3536,
      "step": 15755
    },
    {
      "epoch": 0.4234851538358189,
      "grad_norm": 2.4071357250213623,
      "learning_rate": 2.9218303145853194e-05,
      "loss": 4.2648,
      "step": 15760
    },
    {
      "epoch": 0.42361950826279726,
      "grad_norm": 2.3981125354766846,
      "learning_rate": 2.9211493939806623e-05,
      "loss": 4.2304,
      "step": 15765
    },
    {
      "epoch": 0.4237538626897756,
      "grad_norm": 2.381065607070923,
      "learning_rate": 2.9204684733760046e-05,
      "loss": 4.3488,
      "step": 15770
    },
    {
      "epoch": 0.423888217116754,
      "grad_norm": 2.742440938949585,
      "learning_rate": 2.919787552771347e-05,
      "loss": 4.2627,
      "step": 15775
    },
    {
      "epoch": 0.42402257154373235,
      "grad_norm": 2.414639949798584,
      "learning_rate": 2.9191066321666894e-05,
      "loss": 4.1813,
      "step": 15780
    },
    {
      "epoch": 0.42415692597071075,
      "grad_norm": 2.519242286682129,
      "learning_rate": 2.9184257115620323e-05,
      "loss": 4.2614,
      "step": 15785
    },
    {
      "epoch": 0.4242912803976891,
      "grad_norm": 2.365382194519043,
      "learning_rate": 2.9177447909573745e-05,
      "loss": 4.2572,
      "step": 15790
    },
    {
      "epoch": 0.4244256348246675,
      "grad_norm": 2.306090831756592,
      "learning_rate": 2.917063870352717e-05,
      "loss": 4.2328,
      "step": 15795
    },
    {
      "epoch": 0.42455998925164584,
      "grad_norm": 2.2574098110198975,
      "learning_rate": 2.9163829497480593e-05,
      "loss": 4.2036,
      "step": 15800
    },
    {
      "epoch": 0.4246943436786242,
      "grad_norm": 2.481963634490967,
      "learning_rate": 2.9157020291434022e-05,
      "loss": 4.3622,
      "step": 15805
    },
    {
      "epoch": 0.4248286981056026,
      "grad_norm": 2.133134365081787,
      "learning_rate": 2.9150211085387448e-05,
      "loss": 4.2293,
      "step": 15810
    },
    {
      "epoch": 0.42496305253258093,
      "grad_norm": 2.2951953411102295,
      "learning_rate": 2.914340187934087e-05,
      "loss": 4.2183,
      "step": 15815
    },
    {
      "epoch": 0.42509740695955933,
      "grad_norm": 2.374598264694214,
      "learning_rate": 2.9136592673294296e-05,
      "loss": 4.1375,
      "step": 15820
    },
    {
      "epoch": 0.4252317613865377,
      "grad_norm": 2.261354684829712,
      "learning_rate": 2.9129783467247718e-05,
      "loss": 4.3236,
      "step": 15825
    },
    {
      "epoch": 0.4253661158135161,
      "grad_norm": 2.373946189880371,
      "learning_rate": 2.9122974261201147e-05,
      "loss": 4.1737,
      "step": 15830
    },
    {
      "epoch": 0.4255004702404944,
      "grad_norm": 2.2643213272094727,
      "learning_rate": 2.911616505515457e-05,
      "loss": 4.3299,
      "step": 15835
    },
    {
      "epoch": 0.42563482466747277,
      "grad_norm": 2.3239800930023193,
      "learning_rate": 2.9109355849107995e-05,
      "loss": 4.3079,
      "step": 15840
    },
    {
      "epoch": 0.42576917909445117,
      "grad_norm": 2.3317983150482178,
      "learning_rate": 2.9102546643061417e-05,
      "loss": 4.3979,
      "step": 15845
    },
    {
      "epoch": 0.4259035335214295,
      "grad_norm": 2.6195383071899414,
      "learning_rate": 2.9095737437014846e-05,
      "loss": 4.249,
      "step": 15850
    },
    {
      "epoch": 0.4260378879484079,
      "grad_norm": 2.2538535594940186,
      "learning_rate": 2.9088928230968272e-05,
      "loss": 4.2613,
      "step": 15855
    },
    {
      "epoch": 0.42617224237538626,
      "grad_norm": 2.4361369609832764,
      "learning_rate": 2.9082119024921694e-05,
      "loss": 4.0982,
      "step": 15860
    },
    {
      "epoch": 0.42630659680236466,
      "grad_norm": 2.3043746948242188,
      "learning_rate": 2.907530981887512e-05,
      "loss": 4.2159,
      "step": 15865
    },
    {
      "epoch": 0.426440951229343,
      "grad_norm": 2.6173923015594482,
      "learning_rate": 2.906850061282855e-05,
      "loss": 4.1797,
      "step": 15870
    },
    {
      "epoch": 0.4265753056563214,
      "grad_norm": 2.2595863342285156,
      "learning_rate": 2.906169140678197e-05,
      "loss": 4.2729,
      "step": 15875
    },
    {
      "epoch": 0.42670966008329975,
      "grad_norm": 2.6921050548553467,
      "learning_rate": 2.9054882200735393e-05,
      "loss": 4.1995,
      "step": 15880
    },
    {
      "epoch": 0.4268440145102781,
      "grad_norm": 2.3651561737060547,
      "learning_rate": 2.904807299468882e-05,
      "loss": 4.196,
      "step": 15885
    },
    {
      "epoch": 0.4269783689372565,
      "grad_norm": 2.7142176628112793,
      "learning_rate": 2.9041263788642248e-05,
      "loss": 4.1929,
      "step": 15890
    },
    {
      "epoch": 0.42711272336423484,
      "grad_norm": 2.3585288524627686,
      "learning_rate": 2.903445458259567e-05,
      "loss": 4.2118,
      "step": 15895
    },
    {
      "epoch": 0.42724707779121324,
      "grad_norm": 2.525831699371338,
      "learning_rate": 2.9027645376549096e-05,
      "loss": 4.2202,
      "step": 15900
    },
    {
      "epoch": 0.4273814322181916,
      "grad_norm": 2.475548505783081,
      "learning_rate": 2.9020836170502518e-05,
      "loss": 4.2349,
      "step": 15905
    },
    {
      "epoch": 0.42751578664517,
      "grad_norm": 2.3495709896087646,
      "learning_rate": 2.9014026964455947e-05,
      "loss": 4.3478,
      "step": 15910
    },
    {
      "epoch": 0.4276501410721483,
      "grad_norm": 2.495866298675537,
      "learning_rate": 2.9007217758409373e-05,
      "loss": 4.2731,
      "step": 15915
    },
    {
      "epoch": 0.42778449549912667,
      "grad_norm": 2.3341360092163086,
      "learning_rate": 2.9000408552362795e-05,
      "loss": 4.1796,
      "step": 15920
    },
    {
      "epoch": 0.42791884992610507,
      "grad_norm": 2.4594175815582275,
      "learning_rate": 2.899359934631622e-05,
      "loss": 4.2713,
      "step": 15925
    },
    {
      "epoch": 0.4280532043530834,
      "grad_norm": 2.548840284347534,
      "learning_rate": 2.898679014026965e-05,
      "loss": 4.2566,
      "step": 15930
    },
    {
      "epoch": 0.4281875587800618,
      "grad_norm": 2.6312308311462402,
      "learning_rate": 2.8979980934223072e-05,
      "loss": 4.2493,
      "step": 15935
    },
    {
      "epoch": 0.42832191320704016,
      "grad_norm": 2.3427841663360596,
      "learning_rate": 2.8973171728176494e-05,
      "loss": 4.1521,
      "step": 15940
    },
    {
      "epoch": 0.42845626763401856,
      "grad_norm": 2.225707769393921,
      "learning_rate": 2.896636252212992e-05,
      "loss": 4.2175,
      "step": 15945
    },
    {
      "epoch": 0.4285906220609969,
      "grad_norm": 2.355645179748535,
      "learning_rate": 2.895955331608335e-05,
      "loss": 4.2413,
      "step": 15950
    },
    {
      "epoch": 0.42872497648797525,
      "grad_norm": 2.465419054031372,
      "learning_rate": 2.895274411003677e-05,
      "loss": 4.3133,
      "step": 15955
    },
    {
      "epoch": 0.42885933091495365,
      "grad_norm": 2.4609131813049316,
      "learning_rate": 2.8945934903990197e-05,
      "loss": 4.1369,
      "step": 15960
    },
    {
      "epoch": 0.428993685341932,
      "grad_norm": 2.462458848953247,
      "learning_rate": 2.893912569794362e-05,
      "loss": 4.2616,
      "step": 15965
    },
    {
      "epoch": 0.4291280397689104,
      "grad_norm": 2.5317609310150146,
      "learning_rate": 2.8932316491897045e-05,
      "loss": 4.1,
      "step": 15970
    },
    {
      "epoch": 0.42926239419588874,
      "grad_norm": 2.3024344444274902,
      "learning_rate": 2.8925507285850474e-05,
      "loss": 4.3086,
      "step": 15975
    },
    {
      "epoch": 0.42939674862286714,
      "grad_norm": 2.417215585708618,
      "learning_rate": 2.8918698079803896e-05,
      "loss": 4.2283,
      "step": 15980
    },
    {
      "epoch": 0.4295311030498455,
      "grad_norm": 2.3241524696350098,
      "learning_rate": 2.891188887375732e-05,
      "loss": 4.2974,
      "step": 15985
    },
    {
      "epoch": 0.4296654574768239,
      "grad_norm": 2.4867942333221436,
      "learning_rate": 2.8905079667710744e-05,
      "loss": 4.1405,
      "step": 15990
    },
    {
      "epoch": 0.42979981190380223,
      "grad_norm": 2.3965344429016113,
      "learning_rate": 2.8898270461664173e-05,
      "loss": 4.2104,
      "step": 15995
    },
    {
      "epoch": 0.4299341663307806,
      "grad_norm": 2.107988119125366,
      "learning_rate": 2.8891461255617595e-05,
      "loss": 4.1407,
      "step": 16000
    },
    {
      "epoch": 0.430068520757759,
      "grad_norm": 2.2528016567230225,
      "learning_rate": 2.888465204957102e-05,
      "loss": 4.1995,
      "step": 16005
    },
    {
      "epoch": 0.4302028751847373,
      "grad_norm": 2.3667545318603516,
      "learning_rate": 2.8877842843524443e-05,
      "loss": 4.3516,
      "step": 16010
    },
    {
      "epoch": 0.4303372296117157,
      "grad_norm": 2.5429117679595947,
      "learning_rate": 2.8871033637477872e-05,
      "loss": 4.285,
      "step": 16015
    },
    {
      "epoch": 0.43047158403869407,
      "grad_norm": 2.8680622577667236,
      "learning_rate": 2.8864224431431298e-05,
      "loss": 4.2952,
      "step": 16020
    },
    {
      "epoch": 0.43060593846567247,
      "grad_norm": 2.4054863452911377,
      "learning_rate": 2.885741522538472e-05,
      "loss": 4.194,
      "step": 16025
    },
    {
      "epoch": 0.4307402928926508,
      "grad_norm": 2.44966459274292,
      "learning_rate": 2.8850606019338146e-05,
      "loss": 4.27,
      "step": 16030
    },
    {
      "epoch": 0.43087464731962916,
      "grad_norm": 2.4636518955230713,
      "learning_rate": 2.8843796813291575e-05,
      "loss": 4.3184,
      "step": 16035
    },
    {
      "epoch": 0.43100900174660756,
      "grad_norm": 2.4376299381256104,
      "learning_rate": 2.8836987607244997e-05,
      "loss": 4.1826,
      "step": 16040
    },
    {
      "epoch": 0.4311433561735859,
      "grad_norm": 2.610029458999634,
      "learning_rate": 2.883017840119842e-05,
      "loss": 4.1749,
      "step": 16045
    },
    {
      "epoch": 0.4312777106005643,
      "grad_norm": 2.3537282943725586,
      "learning_rate": 2.8823369195151845e-05,
      "loss": 4.1886,
      "step": 16050
    },
    {
      "epoch": 0.43141206502754265,
      "grad_norm": 2.2158799171447754,
      "learning_rate": 2.8816559989105274e-05,
      "loss": 4.1719,
      "step": 16055
    },
    {
      "epoch": 0.43154641945452105,
      "grad_norm": 2.2209975719451904,
      "learning_rate": 2.8809750783058697e-05,
      "loss": 4.2062,
      "step": 16060
    },
    {
      "epoch": 0.4316807738814994,
      "grad_norm": 2.4708924293518066,
      "learning_rate": 2.8802941577012122e-05,
      "loss": 4.2911,
      "step": 16065
    },
    {
      "epoch": 0.4318151283084778,
      "grad_norm": 2.4639086723327637,
      "learning_rate": 2.8796132370965544e-05,
      "loss": 4.2626,
      "step": 16070
    },
    {
      "epoch": 0.43194948273545614,
      "grad_norm": 2.3329756259918213,
      "learning_rate": 2.8789323164918974e-05,
      "loss": 4.1889,
      "step": 16075
    },
    {
      "epoch": 0.4320838371624345,
      "grad_norm": 2.6666438579559326,
      "learning_rate": 2.87825139588724e-05,
      "loss": 4.204,
      "step": 16080
    },
    {
      "epoch": 0.4322181915894129,
      "grad_norm": 2.5674540996551514,
      "learning_rate": 2.877570475282582e-05,
      "loss": 4.2896,
      "step": 16085
    },
    {
      "epoch": 0.4323525460163912,
      "grad_norm": 2.536684513092041,
      "learning_rate": 2.8768895546779244e-05,
      "loss": 4.2739,
      "step": 16090
    },
    {
      "epoch": 0.4324869004433696,
      "grad_norm": 2.595012903213501,
      "learning_rate": 2.8762086340732676e-05,
      "loss": 4.2995,
      "step": 16095
    },
    {
      "epoch": 0.43262125487034797,
      "grad_norm": 2.2966651916503906,
      "learning_rate": 2.87552771346861e-05,
      "loss": 4.2315,
      "step": 16100
    },
    {
      "epoch": 0.43275560929732637,
      "grad_norm": 2.257228374481201,
      "learning_rate": 2.874846792863952e-05,
      "loss": 4.2044,
      "step": 16105
    },
    {
      "epoch": 0.4328899637243047,
      "grad_norm": 2.4460296630859375,
      "learning_rate": 2.8741658722592946e-05,
      "loss": 4.2559,
      "step": 16110
    },
    {
      "epoch": 0.43302431815128306,
      "grad_norm": 2.3575544357299805,
      "learning_rate": 2.873484951654637e-05,
      "loss": 4.2435,
      "step": 16115
    },
    {
      "epoch": 0.43315867257826146,
      "grad_norm": 2.2947487831115723,
      "learning_rate": 2.8728040310499798e-05,
      "loss": 4.254,
      "step": 16120
    },
    {
      "epoch": 0.4332930270052398,
      "grad_norm": 2.544032096862793,
      "learning_rate": 2.8721231104453223e-05,
      "loss": 4.2012,
      "step": 16125
    },
    {
      "epoch": 0.4334273814322182,
      "grad_norm": 2.6002767086029053,
      "learning_rate": 2.8714421898406646e-05,
      "loss": 4.3138,
      "step": 16130
    },
    {
      "epoch": 0.43356173585919655,
      "grad_norm": 2.2768044471740723,
      "learning_rate": 2.870761269236007e-05,
      "loss": 4.2789,
      "step": 16135
    },
    {
      "epoch": 0.43369609028617495,
      "grad_norm": 2.584963798522949,
      "learning_rate": 2.87008034863135e-05,
      "loss": 4.4201,
      "step": 16140
    },
    {
      "epoch": 0.4338304447131533,
      "grad_norm": 2.327904224395752,
      "learning_rate": 2.8693994280266923e-05,
      "loss": 4.2532,
      "step": 16145
    },
    {
      "epoch": 0.43396479914013164,
      "grad_norm": 2.3140289783477783,
      "learning_rate": 2.8687185074220345e-05,
      "loss": 4.2682,
      "step": 16150
    },
    {
      "epoch": 0.43409915356711004,
      "grad_norm": 2.5426039695739746,
      "learning_rate": 2.868037586817377e-05,
      "loss": 4.2457,
      "step": 16155
    },
    {
      "epoch": 0.4342335079940884,
      "grad_norm": 2.2707035541534424,
      "learning_rate": 2.86735666621272e-05,
      "loss": 4.3181,
      "step": 16160
    },
    {
      "epoch": 0.4343678624210668,
      "grad_norm": 2.5963802337646484,
      "learning_rate": 2.8666757456080622e-05,
      "loss": 4.1863,
      "step": 16165
    },
    {
      "epoch": 0.43450221684804513,
      "grad_norm": 2.124098539352417,
      "learning_rate": 2.8659948250034047e-05,
      "loss": 4.2961,
      "step": 16170
    },
    {
      "epoch": 0.43463657127502353,
      "grad_norm": 2.386662483215332,
      "learning_rate": 2.865313904398747e-05,
      "loss": 4.165,
      "step": 16175
    },
    {
      "epoch": 0.4347709257020019,
      "grad_norm": 2.4544901847839355,
      "learning_rate": 2.86463298379409e-05,
      "loss": 4.1838,
      "step": 16180
    },
    {
      "epoch": 0.4349052801289803,
      "grad_norm": 2.22908878326416,
      "learning_rate": 2.8639520631894324e-05,
      "loss": 4.1704,
      "step": 16185
    },
    {
      "epoch": 0.4350396345559586,
      "grad_norm": 2.346069574356079,
      "learning_rate": 2.8632711425847747e-05,
      "loss": 4.2814,
      "step": 16190
    },
    {
      "epoch": 0.43517398898293697,
      "grad_norm": 2.26861310005188,
      "learning_rate": 2.862590221980117e-05,
      "loss": 4.2697,
      "step": 16195
    },
    {
      "epoch": 0.43530834340991537,
      "grad_norm": 2.4290709495544434,
      "learning_rate": 2.86190930137546e-05,
      "loss": 4.2381,
      "step": 16200
    },
    {
      "epoch": 0.4354426978368937,
      "grad_norm": 2.4715416431427,
      "learning_rate": 2.8612283807708024e-05,
      "loss": 4.2549,
      "step": 16205
    },
    {
      "epoch": 0.4355770522638721,
      "grad_norm": 2.6182286739349365,
      "learning_rate": 2.8605474601661446e-05,
      "loss": 4.3408,
      "step": 16210
    },
    {
      "epoch": 0.43571140669085046,
      "grad_norm": 2.336998701095581,
      "learning_rate": 2.859866539561487e-05,
      "loss": 4.248,
      "step": 16215
    },
    {
      "epoch": 0.43584576111782886,
      "grad_norm": 2.3119163513183594,
      "learning_rate": 2.85918561895683e-05,
      "loss": 4.3169,
      "step": 16220
    },
    {
      "epoch": 0.4359801155448072,
      "grad_norm": 2.2941818237304688,
      "learning_rate": 2.8585046983521723e-05,
      "loss": 4.1995,
      "step": 16225
    },
    {
      "epoch": 0.43611446997178555,
      "grad_norm": 2.4523491859436035,
      "learning_rate": 2.857823777747515e-05,
      "loss": 4.2168,
      "step": 16230
    },
    {
      "epoch": 0.43624882439876395,
      "grad_norm": 2.5257530212402344,
      "learning_rate": 2.857142857142857e-05,
      "loss": 4.236,
      "step": 16235
    },
    {
      "epoch": 0.4363831788257423,
      "grad_norm": 2.489412307739258,
      "learning_rate": 2.8564619365382e-05,
      "loss": 4.1993,
      "step": 16240
    },
    {
      "epoch": 0.4365175332527207,
      "grad_norm": 2.306957483291626,
      "learning_rate": 2.8557810159335426e-05,
      "loss": 4.1267,
      "step": 16245
    },
    {
      "epoch": 0.43665188767969904,
      "grad_norm": 2.3173604011535645,
      "learning_rate": 2.8551000953288848e-05,
      "loss": 4.29,
      "step": 16250
    },
    {
      "epoch": 0.43678624210667744,
      "grad_norm": 2.4237756729125977,
      "learning_rate": 2.854419174724227e-05,
      "loss": 4.2242,
      "step": 16255
    },
    {
      "epoch": 0.4369205965336558,
      "grad_norm": 2.4559576511383057,
      "learning_rate": 2.8537382541195696e-05,
      "loss": 4.1845,
      "step": 16260
    },
    {
      "epoch": 0.4370549509606341,
      "grad_norm": 2.5824778079986572,
      "learning_rate": 2.8530573335149125e-05,
      "loss": 4.1047,
      "step": 16265
    },
    {
      "epoch": 0.4371893053876125,
      "grad_norm": 2.972888231277466,
      "learning_rate": 2.8523764129102547e-05,
      "loss": 4.1979,
      "step": 16270
    },
    {
      "epoch": 0.43732365981459087,
      "grad_norm": 2.3336901664733887,
      "learning_rate": 2.8516954923055973e-05,
      "loss": 4.2603,
      "step": 16275
    },
    {
      "epoch": 0.43745801424156927,
      "grad_norm": 2.27392315864563,
      "learning_rate": 2.8510145717009395e-05,
      "loss": 4.2711,
      "step": 16280
    },
    {
      "epoch": 0.4375923686685476,
      "grad_norm": 2.2228550910949707,
      "learning_rate": 2.8503336510962824e-05,
      "loss": 4.1704,
      "step": 16285
    },
    {
      "epoch": 0.437726723095526,
      "grad_norm": 2.3461356163024902,
      "learning_rate": 2.849652730491625e-05,
      "loss": 4.1636,
      "step": 16290
    },
    {
      "epoch": 0.43786107752250436,
      "grad_norm": 2.279881715774536,
      "learning_rate": 2.8489718098869672e-05,
      "loss": 4.15,
      "step": 16295
    },
    {
      "epoch": 0.43799543194948276,
      "grad_norm": 2.360457181930542,
      "learning_rate": 2.8482908892823094e-05,
      "loss": 4.1969,
      "step": 16300
    },
    {
      "epoch": 0.4381297863764611,
      "grad_norm": 2.5320329666137695,
      "learning_rate": 2.8476099686776527e-05,
      "loss": 4.146,
      "step": 16305
    },
    {
      "epoch": 0.43826414080343945,
      "grad_norm": 2.245178699493408,
      "learning_rate": 2.846929048072995e-05,
      "loss": 4.3665,
      "step": 16310
    },
    {
      "epoch": 0.43839849523041785,
      "grad_norm": 2.297659397125244,
      "learning_rate": 2.846248127468337e-05,
      "loss": 4.2898,
      "step": 16315
    },
    {
      "epoch": 0.4385328496573962,
      "grad_norm": 2.2972676753997803,
      "learning_rate": 2.8455672068636797e-05,
      "loss": 4.2515,
      "step": 16320
    },
    {
      "epoch": 0.4386672040843746,
      "grad_norm": 2.63649582862854,
      "learning_rate": 2.8448862862590226e-05,
      "loss": 4.3153,
      "step": 16325
    },
    {
      "epoch": 0.43880155851135294,
      "grad_norm": 2.322937488555908,
      "learning_rate": 2.8442053656543648e-05,
      "loss": 4.1691,
      "step": 16330
    },
    {
      "epoch": 0.43893591293833134,
      "grad_norm": 2.166642904281616,
      "learning_rate": 2.8435244450497074e-05,
      "loss": 4.2791,
      "step": 16335
    },
    {
      "epoch": 0.4390702673653097,
      "grad_norm": 2.330350875854492,
      "learning_rate": 2.8428435244450496e-05,
      "loss": 4.3459,
      "step": 16340
    },
    {
      "epoch": 0.43920462179228803,
      "grad_norm": 2.468033790588379,
      "learning_rate": 2.8421626038403925e-05,
      "loss": 4.1909,
      "step": 16345
    },
    {
      "epoch": 0.43933897621926643,
      "grad_norm": 2.2964963912963867,
      "learning_rate": 2.841481683235735e-05,
      "loss": 4.2369,
      "step": 16350
    },
    {
      "epoch": 0.4394733306462448,
      "grad_norm": 2.519709825515747,
      "learning_rate": 2.8408007626310773e-05,
      "loss": 4.2757,
      "step": 16355
    },
    {
      "epoch": 0.4396076850732232,
      "grad_norm": 2.3391635417938232,
      "learning_rate": 2.8401198420264195e-05,
      "loss": 4.1615,
      "step": 16360
    },
    {
      "epoch": 0.4397420395002015,
      "grad_norm": 2.4118688106536865,
      "learning_rate": 2.8394389214217624e-05,
      "loss": 4.1488,
      "step": 16365
    },
    {
      "epoch": 0.4398763939271799,
      "grad_norm": 2.360112428665161,
      "learning_rate": 2.838758000817105e-05,
      "loss": 4.3131,
      "step": 16370
    },
    {
      "epoch": 0.44001074835415827,
      "grad_norm": 2.4447109699249268,
      "learning_rate": 2.8380770802124472e-05,
      "loss": 4.1939,
      "step": 16375
    },
    {
      "epoch": 0.44014510278113667,
      "grad_norm": 2.4136271476745605,
      "learning_rate": 2.8373961596077898e-05,
      "loss": 4.2756,
      "step": 16380
    },
    {
      "epoch": 0.440279457208115,
      "grad_norm": 2.199481964111328,
      "learning_rate": 2.8367152390031327e-05,
      "loss": 4.2668,
      "step": 16385
    },
    {
      "epoch": 0.44041381163509336,
      "grad_norm": 2.580078601837158,
      "learning_rate": 2.836034318398475e-05,
      "loss": 4.3542,
      "step": 16390
    },
    {
      "epoch": 0.44054816606207176,
      "grad_norm": 2.3920347690582275,
      "learning_rate": 2.8353533977938175e-05,
      "loss": 4.2211,
      "step": 16395
    },
    {
      "epoch": 0.4406825204890501,
      "grad_norm": 2.3929126262664795,
      "learning_rate": 2.8346724771891597e-05,
      "loss": 4.2629,
      "step": 16400
    },
    {
      "epoch": 0.4408168749160285,
      "grad_norm": 2.3883540630340576,
      "learning_rate": 2.833991556584502e-05,
      "loss": 4.1322,
      "step": 16405
    },
    {
      "epoch": 0.44095122934300685,
      "grad_norm": 2.2770512104034424,
      "learning_rate": 2.8333106359798452e-05,
      "loss": 4.0467,
      "step": 16410
    },
    {
      "epoch": 0.44108558376998525,
      "grad_norm": 2.3421852588653564,
      "learning_rate": 2.8326297153751874e-05,
      "loss": 4.2041,
      "step": 16415
    },
    {
      "epoch": 0.4412199381969636,
      "grad_norm": 2.461590528488159,
      "learning_rate": 2.8319487947705296e-05,
      "loss": 4.2816,
      "step": 16420
    },
    {
      "epoch": 0.44135429262394194,
      "grad_norm": 2.4068994522094727,
      "learning_rate": 2.8312678741658722e-05,
      "loss": 4.1461,
      "step": 16425
    },
    {
      "epoch": 0.44148864705092034,
      "grad_norm": 2.1470868587493896,
      "learning_rate": 2.830586953561215e-05,
      "loss": 4.1923,
      "step": 16430
    },
    {
      "epoch": 0.4416230014778987,
      "grad_norm": 2.259752035140991,
      "learning_rate": 2.8299060329565573e-05,
      "loss": 4.0845,
      "step": 16435
    },
    {
      "epoch": 0.4417573559048771,
      "grad_norm": 2.2395246028900146,
      "learning_rate": 2.8292251123519e-05,
      "loss": 4.2308,
      "step": 16440
    },
    {
      "epoch": 0.4418917103318554,
      "grad_norm": 2.5562379360198975,
      "learning_rate": 2.828544191747242e-05,
      "loss": 4.2582,
      "step": 16445
    },
    {
      "epoch": 0.4420260647588338,
      "grad_norm": 2.1990792751312256,
      "learning_rate": 2.827863271142585e-05,
      "loss": 4.0991,
      "step": 16450
    },
    {
      "epoch": 0.44216041918581217,
      "grad_norm": 2.6462621688842773,
      "learning_rate": 2.8271823505379276e-05,
      "loss": 4.219,
      "step": 16455
    },
    {
      "epoch": 0.4422947736127905,
      "grad_norm": 2.408582925796509,
      "learning_rate": 2.82650142993327e-05,
      "loss": 4.151,
      "step": 16460
    },
    {
      "epoch": 0.4424291280397689,
      "grad_norm": 2.3861005306243896,
      "learning_rate": 2.825820509328612e-05,
      "loss": 4.2187,
      "step": 16465
    },
    {
      "epoch": 0.44256348246674726,
      "grad_norm": 3.3405652046203613,
      "learning_rate": 2.825139588723955e-05,
      "loss": 4.2529,
      "step": 16470
    },
    {
      "epoch": 0.44269783689372566,
      "grad_norm": 2.3474910259246826,
      "learning_rate": 2.8244586681192975e-05,
      "loss": 4.1749,
      "step": 16475
    },
    {
      "epoch": 0.442832191320704,
      "grad_norm": 2.4719419479370117,
      "learning_rate": 2.8237777475146398e-05,
      "loss": 4.09,
      "step": 16480
    },
    {
      "epoch": 0.4429665457476824,
      "grad_norm": 2.3105177879333496,
      "learning_rate": 2.8230968269099823e-05,
      "loss": 4.2202,
      "step": 16485
    },
    {
      "epoch": 0.44310090017466075,
      "grad_norm": 2.320051431655884,
      "learning_rate": 2.8224159063053252e-05,
      "loss": 4.1952,
      "step": 16490
    },
    {
      "epoch": 0.44323525460163915,
      "grad_norm": 2.18603253364563,
      "learning_rate": 2.8217349857006675e-05,
      "loss": 4.2338,
      "step": 16495
    },
    {
      "epoch": 0.4433696090286175,
      "grad_norm": 2.512141704559326,
      "learning_rate": 2.82105406509601e-05,
      "loss": 4.2908,
      "step": 16500
    },
    {
      "epoch": 0.44350396345559584,
      "grad_norm": 2.251185178756714,
      "learning_rate": 2.8203731444913522e-05,
      "loss": 4.3063,
      "step": 16505
    },
    {
      "epoch": 0.44363831788257424,
      "grad_norm": 2.357961893081665,
      "learning_rate": 2.819692223886695e-05,
      "loss": 4.1761,
      "step": 16510
    },
    {
      "epoch": 0.4437726723095526,
      "grad_norm": 2.427928924560547,
      "learning_rate": 2.8190113032820377e-05,
      "loss": 4.2287,
      "step": 16515
    },
    {
      "epoch": 0.443907026736531,
      "grad_norm": 2.463430643081665,
      "learning_rate": 2.81833038267738e-05,
      "loss": 4.2661,
      "step": 16520
    },
    {
      "epoch": 0.44404138116350933,
      "grad_norm": 2.1994524002075195,
      "learning_rate": 2.8176494620727222e-05,
      "loss": 4.2353,
      "step": 16525
    },
    {
      "epoch": 0.44417573559048773,
      "grad_norm": 2.338409662246704,
      "learning_rate": 2.816968541468065e-05,
      "loss": 4.3711,
      "step": 16530
    },
    {
      "epoch": 0.4443100900174661,
      "grad_norm": 2.452298641204834,
      "learning_rate": 2.8162876208634076e-05,
      "loss": 4.1041,
      "step": 16535
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 2.5668511390686035,
      "learning_rate": 2.81560670025875e-05,
      "loss": 4.1343,
      "step": 16540
    },
    {
      "epoch": 0.4445787988714228,
      "grad_norm": 2.501283645629883,
      "learning_rate": 2.8149257796540924e-05,
      "loss": 4.252,
      "step": 16545
    },
    {
      "epoch": 0.44471315329840116,
      "grad_norm": 2.3662846088409424,
      "learning_rate": 2.8142448590494353e-05,
      "loss": 4.252,
      "step": 16550
    },
    {
      "epoch": 0.44484750772537957,
      "grad_norm": 2.4809865951538086,
      "learning_rate": 2.8135639384447776e-05,
      "loss": 4.1486,
      "step": 16555
    },
    {
      "epoch": 0.4449818621523579,
      "grad_norm": 2.655667543411255,
      "learning_rate": 2.81288301784012e-05,
      "loss": 4.159,
      "step": 16560
    },
    {
      "epoch": 0.4451162165793363,
      "grad_norm": 2.2696967124938965,
      "learning_rate": 2.8122020972354624e-05,
      "loss": 4.3239,
      "step": 16565
    },
    {
      "epoch": 0.44525057100631465,
      "grad_norm": 2.820187568664551,
      "learning_rate": 2.8115211766308046e-05,
      "loss": 4.369,
      "step": 16570
    },
    {
      "epoch": 0.445384925433293,
      "grad_norm": 2.1930160522460938,
      "learning_rate": 2.8108402560261475e-05,
      "loss": 4.1778,
      "step": 16575
    },
    {
      "epoch": 0.4455192798602714,
      "grad_norm": 2.3937129974365234,
      "learning_rate": 2.81015933542149e-05,
      "loss": 4.3096,
      "step": 16580
    },
    {
      "epoch": 0.44565363428724974,
      "grad_norm": 2.46380877494812,
      "learning_rate": 2.8094784148168323e-05,
      "loss": 4.2947,
      "step": 16585
    },
    {
      "epoch": 0.44578798871422814,
      "grad_norm": 2.473848581314087,
      "learning_rate": 2.808797494212175e-05,
      "loss": 4.1486,
      "step": 16590
    },
    {
      "epoch": 0.4459223431412065,
      "grad_norm": 2.1642918586730957,
      "learning_rate": 2.8081165736075178e-05,
      "loss": 4.3182,
      "step": 16595
    },
    {
      "epoch": 0.4460566975681849,
      "grad_norm": 2.41929030418396,
      "learning_rate": 2.80743565300286e-05,
      "loss": 4.2923,
      "step": 16600
    },
    {
      "epoch": 0.44619105199516323,
      "grad_norm": 2.5490310192108154,
      "learning_rate": 2.8067547323982025e-05,
      "loss": 4.2793,
      "step": 16605
    },
    {
      "epoch": 0.44632540642214164,
      "grad_norm": 2.5455315113067627,
      "learning_rate": 2.8060738117935448e-05,
      "loss": 4.1978,
      "step": 16610
    },
    {
      "epoch": 0.44645976084912,
      "grad_norm": 2.274678945541382,
      "learning_rate": 2.8053928911888877e-05,
      "loss": 4.2547,
      "step": 16615
    },
    {
      "epoch": 0.4465941152760983,
      "grad_norm": 2.721982955932617,
      "learning_rate": 2.8047119705842302e-05,
      "loss": 4.119,
      "step": 16620
    },
    {
      "epoch": 0.4467284697030767,
      "grad_norm": 2.5358617305755615,
      "learning_rate": 2.8040310499795725e-05,
      "loss": 4.2037,
      "step": 16625
    },
    {
      "epoch": 0.44686282413005507,
      "grad_norm": 2.323357582092285,
      "learning_rate": 2.8033501293749147e-05,
      "loss": 4.2736,
      "step": 16630
    },
    {
      "epoch": 0.44699717855703347,
      "grad_norm": 2.411027193069458,
      "learning_rate": 2.8026692087702576e-05,
      "loss": 4.1978,
      "step": 16635
    },
    {
      "epoch": 0.4471315329840118,
      "grad_norm": 2.6535913944244385,
      "learning_rate": 2.8019882881656e-05,
      "loss": 4.2786,
      "step": 16640
    },
    {
      "epoch": 0.4472658874109902,
      "grad_norm": 2.3581197261810303,
      "learning_rate": 2.8013073675609424e-05,
      "loss": 4.2367,
      "step": 16645
    },
    {
      "epoch": 0.44740024183796856,
      "grad_norm": 2.307725429534912,
      "learning_rate": 2.800626446956285e-05,
      "loss": 4.2442,
      "step": 16650
    },
    {
      "epoch": 0.4475345962649469,
      "grad_norm": 2.2293784618377686,
      "learning_rate": 2.799945526351628e-05,
      "loss": 4.2545,
      "step": 16655
    },
    {
      "epoch": 0.4476689506919253,
      "grad_norm": 2.4522855281829834,
      "learning_rate": 2.79926460574697e-05,
      "loss": 4.2684,
      "step": 16660
    },
    {
      "epoch": 0.44780330511890365,
      "grad_norm": 2.384861469268799,
      "learning_rate": 2.7985836851423127e-05,
      "loss": 4.2494,
      "step": 16665
    },
    {
      "epoch": 0.44793765954588205,
      "grad_norm": 2.419438123703003,
      "learning_rate": 2.797902764537655e-05,
      "loss": 4.2345,
      "step": 16670
    },
    {
      "epoch": 0.4480720139728604,
      "grad_norm": 2.312316417694092,
      "learning_rate": 2.7972218439329978e-05,
      "loss": 4.2121,
      "step": 16675
    },
    {
      "epoch": 0.4482063683998388,
      "grad_norm": 2.3352880477905273,
      "learning_rate": 2.79654092332834e-05,
      "loss": 4.1864,
      "step": 16680
    },
    {
      "epoch": 0.44834072282681714,
      "grad_norm": 2.4615979194641113,
      "learning_rate": 2.7958600027236826e-05,
      "loss": 4.2601,
      "step": 16685
    },
    {
      "epoch": 0.44847507725379554,
      "grad_norm": 2.1711671352386475,
      "learning_rate": 2.7951790821190248e-05,
      "loss": 4.2544,
      "step": 16690
    },
    {
      "epoch": 0.4486094316807739,
      "grad_norm": 2.54040789604187,
      "learning_rate": 2.7944981615143677e-05,
      "loss": 4.1784,
      "step": 16695
    },
    {
      "epoch": 0.44874378610775223,
      "grad_norm": 2.468158721923828,
      "learning_rate": 2.7938172409097103e-05,
      "loss": 4.2734,
      "step": 16700
    },
    {
      "epoch": 0.44887814053473063,
      "grad_norm": 2.4374277591705322,
      "learning_rate": 2.7931363203050525e-05,
      "loss": 4.1635,
      "step": 16705
    },
    {
      "epoch": 0.449012494961709,
      "grad_norm": 2.549758195877075,
      "learning_rate": 2.792455399700395e-05,
      "loss": 4.2302,
      "step": 16710
    },
    {
      "epoch": 0.4491468493886874,
      "grad_norm": 2.3844382762908936,
      "learning_rate": 2.7917744790957373e-05,
      "loss": 4.3163,
      "step": 16715
    },
    {
      "epoch": 0.4492812038156657,
      "grad_norm": 2.311689615249634,
      "learning_rate": 2.7910935584910802e-05,
      "loss": 4.226,
      "step": 16720
    },
    {
      "epoch": 0.4494155582426441,
      "grad_norm": 2.5290637016296387,
      "learning_rate": 2.7904126378864224e-05,
      "loss": 4.3294,
      "step": 16725
    },
    {
      "epoch": 0.44954991266962246,
      "grad_norm": 2.617156982421875,
      "learning_rate": 2.789731717281765e-05,
      "loss": 4.1743,
      "step": 16730
    },
    {
      "epoch": 0.4496842670966008,
      "grad_norm": 2.384774923324585,
      "learning_rate": 2.7890507966771072e-05,
      "loss": 4.2958,
      "step": 16735
    },
    {
      "epoch": 0.4498186215235792,
      "grad_norm": 2.4706966876983643,
      "learning_rate": 2.78836987607245e-05,
      "loss": 4.2683,
      "step": 16740
    },
    {
      "epoch": 0.44995297595055755,
      "grad_norm": 2.587946653366089,
      "learning_rate": 2.7876889554677927e-05,
      "loss": 4.3468,
      "step": 16745
    },
    {
      "epoch": 0.45008733037753595,
      "grad_norm": 2.2501325607299805,
      "learning_rate": 2.787008034863135e-05,
      "loss": 4.1242,
      "step": 16750
    },
    {
      "epoch": 0.4502216848045143,
      "grad_norm": 2.2695369720458984,
      "learning_rate": 2.7863271142584775e-05,
      "loss": 4.2609,
      "step": 16755
    },
    {
      "epoch": 0.4503560392314927,
      "grad_norm": 2.4646825790405273,
      "learning_rate": 2.7856461936538204e-05,
      "loss": 4.2798,
      "step": 16760
    },
    {
      "epoch": 0.45049039365847104,
      "grad_norm": 2.6789493560791016,
      "learning_rate": 2.7849652730491626e-05,
      "loss": 4.1272,
      "step": 16765
    },
    {
      "epoch": 0.4506247480854494,
      "grad_norm": 2.4618709087371826,
      "learning_rate": 2.7842843524445052e-05,
      "loss": 4.146,
      "step": 16770
    },
    {
      "epoch": 0.4507591025124278,
      "grad_norm": 2.187411069869995,
      "learning_rate": 2.7836034318398474e-05,
      "loss": 4.108,
      "step": 16775
    },
    {
      "epoch": 0.45089345693940613,
      "grad_norm": 2.58382248878479,
      "learning_rate": 2.7829225112351903e-05,
      "loss": 4.2531,
      "step": 16780
    },
    {
      "epoch": 0.45102781136638453,
      "grad_norm": 2.456743001937866,
      "learning_rate": 2.7822415906305325e-05,
      "loss": 4.3009,
      "step": 16785
    },
    {
      "epoch": 0.4511621657933629,
      "grad_norm": 2.5162689685821533,
      "learning_rate": 2.781560670025875e-05,
      "loss": 4.1872,
      "step": 16790
    },
    {
      "epoch": 0.4512965202203413,
      "grad_norm": 2.3018100261688232,
      "learning_rate": 2.7808797494212173e-05,
      "loss": 4.283,
      "step": 16795
    },
    {
      "epoch": 0.4514308746473196,
      "grad_norm": 2.4532432556152344,
      "learning_rate": 2.7801988288165602e-05,
      "loss": 4.2603,
      "step": 16800
    },
    {
      "epoch": 0.451565229074298,
      "grad_norm": 2.2642500400543213,
      "learning_rate": 2.7795179082119028e-05,
      "loss": 4.205,
      "step": 16805
    },
    {
      "epoch": 0.45169958350127637,
      "grad_norm": 2.441150426864624,
      "learning_rate": 2.778836987607245e-05,
      "loss": 4.1176,
      "step": 16810
    },
    {
      "epoch": 0.4518339379282547,
      "grad_norm": 2.5511395931243896,
      "learning_rate": 2.7781560670025876e-05,
      "loss": 4.3164,
      "step": 16815
    },
    {
      "epoch": 0.4519682923552331,
      "grad_norm": 2.7007107734680176,
      "learning_rate": 2.7774751463979305e-05,
      "loss": 4.1382,
      "step": 16820
    },
    {
      "epoch": 0.45210264678221146,
      "grad_norm": 2.3082969188690186,
      "learning_rate": 2.7767942257932727e-05,
      "loss": 4.1662,
      "step": 16825
    },
    {
      "epoch": 0.45223700120918986,
      "grad_norm": 2.1427457332611084,
      "learning_rate": 2.776113305188615e-05,
      "loss": 4.0684,
      "step": 16830
    },
    {
      "epoch": 0.4523713556361682,
      "grad_norm": 2.4540045261383057,
      "learning_rate": 2.7754323845839575e-05,
      "loss": 4.1479,
      "step": 16835
    },
    {
      "epoch": 0.4525057100631466,
      "grad_norm": 2.539919137954712,
      "learning_rate": 2.7747514639793004e-05,
      "loss": 4.309,
      "step": 16840
    },
    {
      "epoch": 0.45264006449012495,
      "grad_norm": 2.131465196609497,
      "learning_rate": 2.7740705433746427e-05,
      "loss": 4.2961,
      "step": 16845
    },
    {
      "epoch": 0.4527744189171033,
      "grad_norm": 2.4124083518981934,
      "learning_rate": 2.7733896227699852e-05,
      "loss": 4.1728,
      "step": 16850
    },
    {
      "epoch": 0.4529087733440817,
      "grad_norm": 2.4689693450927734,
      "learning_rate": 2.7727087021653274e-05,
      "loss": 4.2876,
      "step": 16855
    },
    {
      "epoch": 0.45304312777106004,
      "grad_norm": 2.43318510055542,
      "learning_rate": 2.77202778156067e-05,
      "loss": 4.214,
      "step": 16860
    },
    {
      "epoch": 0.45317748219803844,
      "grad_norm": 2.3449084758758545,
      "learning_rate": 2.771346860956013e-05,
      "loss": 4.277,
      "step": 16865
    },
    {
      "epoch": 0.4533118366250168,
      "grad_norm": 2.183117628097534,
      "learning_rate": 2.770665940351355e-05,
      "loss": 4.3097,
      "step": 16870
    },
    {
      "epoch": 0.4534461910519952,
      "grad_norm": 2.4677445888519287,
      "learning_rate": 2.7699850197466977e-05,
      "loss": 4.2961,
      "step": 16875
    },
    {
      "epoch": 0.45358054547897353,
      "grad_norm": 2.373764753341675,
      "learning_rate": 2.76930409914204e-05,
      "loss": 4.3081,
      "step": 16880
    },
    {
      "epoch": 0.4537148999059519,
      "grad_norm": 2.315349578857422,
      "learning_rate": 2.768623178537383e-05,
      "loss": 4.4613,
      "step": 16885
    },
    {
      "epoch": 0.4538492543329303,
      "grad_norm": 2.489366292953491,
      "learning_rate": 2.767942257932725e-05,
      "loss": 4.24,
      "step": 16890
    },
    {
      "epoch": 0.4539836087599086,
      "grad_norm": 2.3375909328460693,
      "learning_rate": 2.7672613373280676e-05,
      "loss": 4.3812,
      "step": 16895
    },
    {
      "epoch": 0.454117963186887,
      "grad_norm": 2.539177417755127,
      "learning_rate": 2.76658041672341e-05,
      "loss": 4.2191,
      "step": 16900
    },
    {
      "epoch": 0.45425231761386536,
      "grad_norm": 2.5202620029449463,
      "learning_rate": 2.7658994961187528e-05,
      "loss": 4.1013,
      "step": 16905
    },
    {
      "epoch": 0.45438667204084376,
      "grad_norm": 2.3986546993255615,
      "learning_rate": 2.7652185755140953e-05,
      "loss": 4.2712,
      "step": 16910
    },
    {
      "epoch": 0.4545210264678221,
      "grad_norm": 2.3960037231445312,
      "learning_rate": 2.7645376549094376e-05,
      "loss": 4.1948,
      "step": 16915
    },
    {
      "epoch": 0.4546553808948005,
      "grad_norm": 2.4452643394470215,
      "learning_rate": 2.76385673430478e-05,
      "loss": 4.2279,
      "step": 16920
    },
    {
      "epoch": 0.45478973532177885,
      "grad_norm": 2.552942991256714,
      "learning_rate": 2.763175813700123e-05,
      "loss": 4.2151,
      "step": 16925
    },
    {
      "epoch": 0.4549240897487572,
      "grad_norm": 2.4289443492889404,
      "learning_rate": 2.7624948930954653e-05,
      "loss": 4.1386,
      "step": 16930
    },
    {
      "epoch": 0.4550584441757356,
      "grad_norm": 2.380992889404297,
      "learning_rate": 2.7618139724908075e-05,
      "loss": 4.0512,
      "step": 16935
    },
    {
      "epoch": 0.45519279860271394,
      "grad_norm": 2.267054557800293,
      "learning_rate": 2.76113305188615e-05,
      "loss": 4.2075,
      "step": 16940
    },
    {
      "epoch": 0.45532715302969234,
      "grad_norm": 2.4519221782684326,
      "learning_rate": 2.760452131281493e-05,
      "loss": 4.0911,
      "step": 16945
    },
    {
      "epoch": 0.4554615074566707,
      "grad_norm": 2.26326060295105,
      "learning_rate": 2.7597712106768352e-05,
      "loss": 4.1349,
      "step": 16950
    },
    {
      "epoch": 0.4555958618836491,
      "grad_norm": 2.2596428394317627,
      "learning_rate": 2.7590902900721777e-05,
      "loss": 4.1746,
      "step": 16955
    },
    {
      "epoch": 0.45573021631062743,
      "grad_norm": 2.338775873184204,
      "learning_rate": 2.75840936946752e-05,
      "loss": 4.1893,
      "step": 16960
    },
    {
      "epoch": 0.4558645707376058,
      "grad_norm": 2.572000503540039,
      "learning_rate": 2.757728448862863e-05,
      "loss": 4.2582,
      "step": 16965
    },
    {
      "epoch": 0.4559989251645842,
      "grad_norm": 2.2636094093322754,
      "learning_rate": 2.7570475282582054e-05,
      "loss": 4.1843,
      "step": 16970
    },
    {
      "epoch": 0.4561332795915625,
      "grad_norm": 2.6545205116271973,
      "learning_rate": 2.7563666076535477e-05,
      "loss": 4.0825,
      "step": 16975
    },
    {
      "epoch": 0.4562676340185409,
      "grad_norm": 2.1999871730804443,
      "learning_rate": 2.7556856870488902e-05,
      "loss": 4.1855,
      "step": 16980
    },
    {
      "epoch": 0.45640198844551927,
      "grad_norm": 2.7572648525238037,
      "learning_rate": 2.755004766444233e-05,
      "loss": 4.3447,
      "step": 16985
    },
    {
      "epoch": 0.45653634287249767,
      "grad_norm": 2.615355968475342,
      "learning_rate": 2.7543238458395754e-05,
      "loss": 4.3807,
      "step": 16990
    },
    {
      "epoch": 0.456670697299476,
      "grad_norm": 2.6059889793395996,
      "learning_rate": 2.7536429252349176e-05,
      "loss": 4.3114,
      "step": 16995
    },
    {
      "epoch": 0.4568050517264544,
      "grad_norm": 2.7207038402557373,
      "learning_rate": 2.75296200463026e-05,
      "loss": 4.251,
      "step": 17000
    },
    {
      "epoch": 0.45693940615343276,
      "grad_norm": 2.539463996887207,
      "learning_rate": 2.7522810840256024e-05,
      "loss": 4.1167,
      "step": 17005
    },
    {
      "epoch": 0.4570737605804111,
      "grad_norm": 2.6501057147979736,
      "learning_rate": 2.7516001634209453e-05,
      "loss": 4.2057,
      "step": 17010
    },
    {
      "epoch": 0.4572081150073895,
      "grad_norm": 2.492187023162842,
      "learning_rate": 2.750919242816288e-05,
      "loss": 4.2154,
      "step": 17015
    },
    {
      "epoch": 0.45734246943436785,
      "grad_norm": 2.1854114532470703,
      "learning_rate": 2.75023832221163e-05,
      "loss": 4.2126,
      "step": 17020
    },
    {
      "epoch": 0.45747682386134625,
      "grad_norm": 2.345491409301758,
      "learning_rate": 2.7495574016069726e-05,
      "loss": 4.2608,
      "step": 17025
    },
    {
      "epoch": 0.4576111782883246,
      "grad_norm": 2.3118348121643066,
      "learning_rate": 2.7488764810023156e-05,
      "loss": 4.1381,
      "step": 17030
    },
    {
      "epoch": 0.457745532715303,
      "grad_norm": 2.4007415771484375,
      "learning_rate": 2.7481955603976578e-05,
      "loss": 4.193,
      "step": 17035
    },
    {
      "epoch": 0.45787988714228134,
      "grad_norm": 2.3045997619628906,
      "learning_rate": 2.747514639793e-05,
      "loss": 4.1246,
      "step": 17040
    },
    {
      "epoch": 0.4580142415692597,
      "grad_norm": 2.3691446781158447,
      "learning_rate": 2.7468337191883426e-05,
      "loss": 4.138,
      "step": 17045
    },
    {
      "epoch": 0.4581485959962381,
      "grad_norm": 2.2577271461486816,
      "learning_rate": 2.7461527985836855e-05,
      "loss": 4.2171,
      "step": 17050
    },
    {
      "epoch": 0.45828295042321643,
      "grad_norm": 2.394845962524414,
      "learning_rate": 2.7454718779790277e-05,
      "loss": 4.2636,
      "step": 17055
    },
    {
      "epoch": 0.45841730485019483,
      "grad_norm": 2.302999258041382,
      "learning_rate": 2.7447909573743703e-05,
      "loss": 4.2129,
      "step": 17060
    },
    {
      "epoch": 0.4585516592771732,
      "grad_norm": 2.6222727298736572,
      "learning_rate": 2.7441100367697125e-05,
      "loss": 4.1679,
      "step": 17065
    },
    {
      "epoch": 0.4586860137041516,
      "grad_norm": 2.294062852859497,
      "learning_rate": 2.7434291161650554e-05,
      "loss": 4.1649,
      "step": 17070
    },
    {
      "epoch": 0.4588203681311299,
      "grad_norm": 2.4459240436553955,
      "learning_rate": 2.742748195560398e-05,
      "loss": 4.171,
      "step": 17075
    },
    {
      "epoch": 0.45895472255810826,
      "grad_norm": 2.1085197925567627,
      "learning_rate": 2.7420672749557402e-05,
      "loss": 4.1963,
      "step": 17080
    },
    {
      "epoch": 0.45908907698508666,
      "grad_norm": 2.6606881618499756,
      "learning_rate": 2.7413863543510828e-05,
      "loss": 4.251,
      "step": 17085
    },
    {
      "epoch": 0.459223431412065,
      "grad_norm": 2.4329206943511963,
      "learning_rate": 2.7407054337464257e-05,
      "loss": 4.1332,
      "step": 17090
    },
    {
      "epoch": 0.4593577858390434,
      "grad_norm": 2.483705997467041,
      "learning_rate": 2.740024513141768e-05,
      "loss": 4.1251,
      "step": 17095
    },
    {
      "epoch": 0.45949214026602175,
      "grad_norm": 2.378059148788452,
      "learning_rate": 2.73934359253711e-05,
      "loss": 4.1439,
      "step": 17100
    },
    {
      "epoch": 0.45962649469300015,
      "grad_norm": 2.274507522583008,
      "learning_rate": 2.7386626719324527e-05,
      "loss": 4.2173,
      "step": 17105
    },
    {
      "epoch": 0.4597608491199785,
      "grad_norm": 2.444655179977417,
      "learning_rate": 2.7379817513277956e-05,
      "loss": 4.2779,
      "step": 17110
    },
    {
      "epoch": 0.4598952035469569,
      "grad_norm": 2.2635371685028076,
      "learning_rate": 2.7373008307231378e-05,
      "loss": 4.2555,
      "step": 17115
    },
    {
      "epoch": 0.46002955797393524,
      "grad_norm": 2.609412908554077,
      "learning_rate": 2.7366199101184804e-05,
      "loss": 4.4117,
      "step": 17120
    },
    {
      "epoch": 0.4601639124009136,
      "grad_norm": 2.393961191177368,
      "learning_rate": 2.7359389895138226e-05,
      "loss": 4.327,
      "step": 17125
    },
    {
      "epoch": 0.460298266827892,
      "grad_norm": 2.4917209148406982,
      "learning_rate": 2.7352580689091655e-05,
      "loss": 4.2426,
      "step": 17130
    },
    {
      "epoch": 0.46043262125487033,
      "grad_norm": 2.5080511569976807,
      "learning_rate": 2.734577148304508e-05,
      "loss": 4.1369,
      "step": 17135
    },
    {
      "epoch": 0.46056697568184873,
      "grad_norm": 2.3354811668395996,
      "learning_rate": 2.7338962276998503e-05,
      "loss": 4.166,
      "step": 17140
    },
    {
      "epoch": 0.4607013301088271,
      "grad_norm": 2.3345820903778076,
      "learning_rate": 2.7332153070951925e-05,
      "loss": 4.2162,
      "step": 17145
    },
    {
      "epoch": 0.4608356845358055,
      "grad_norm": 2.4131319522857666,
      "learning_rate": 2.732534386490535e-05,
      "loss": 4.2319,
      "step": 17150
    },
    {
      "epoch": 0.4609700389627838,
      "grad_norm": 2.32200026512146,
      "learning_rate": 2.731853465885878e-05,
      "loss": 4.2485,
      "step": 17155
    },
    {
      "epoch": 0.46110439338976217,
      "grad_norm": 2.574925422668457,
      "learning_rate": 2.7311725452812202e-05,
      "loss": 4.2692,
      "step": 17160
    },
    {
      "epoch": 0.46123874781674057,
      "grad_norm": 2.3081295490264893,
      "learning_rate": 2.7304916246765628e-05,
      "loss": 4.349,
      "step": 17165
    },
    {
      "epoch": 0.4613731022437189,
      "grad_norm": 2.494278907775879,
      "learning_rate": 2.729810704071905e-05,
      "loss": 4.2664,
      "step": 17170
    },
    {
      "epoch": 0.4615074566706973,
      "grad_norm": 2.282325029373169,
      "learning_rate": 2.729129783467248e-05,
      "loss": 4.2515,
      "step": 17175
    },
    {
      "epoch": 0.46164181109767566,
      "grad_norm": 2.6024065017700195,
      "learning_rate": 2.7284488628625905e-05,
      "loss": 4.1557,
      "step": 17180
    },
    {
      "epoch": 0.46177616552465406,
      "grad_norm": 2.3965132236480713,
      "learning_rate": 2.7277679422579327e-05,
      "loss": 4.3227,
      "step": 17185
    },
    {
      "epoch": 0.4619105199516324,
      "grad_norm": 2.4947686195373535,
      "learning_rate": 2.7270870216532753e-05,
      "loss": 4.2534,
      "step": 17190
    },
    {
      "epoch": 0.46204487437861075,
      "grad_norm": 2.54958438873291,
      "learning_rate": 2.7264061010486182e-05,
      "loss": 4.2254,
      "step": 17195
    },
    {
      "epoch": 0.46217922880558915,
      "grad_norm": 2.666404962539673,
      "learning_rate": 2.7257251804439604e-05,
      "loss": 4.0959,
      "step": 17200
    },
    {
      "epoch": 0.4623135832325675,
      "grad_norm": 2.3037772178649902,
      "learning_rate": 2.7250442598393026e-05,
      "loss": 4.0639,
      "step": 17205
    },
    {
      "epoch": 0.4624479376595459,
      "grad_norm": 2.432457685470581,
      "learning_rate": 2.7243633392346452e-05,
      "loss": 4.3129,
      "step": 17210
    },
    {
      "epoch": 0.46258229208652424,
      "grad_norm": 2.548823833465576,
      "learning_rate": 2.723682418629988e-05,
      "loss": 4.2745,
      "step": 17215
    },
    {
      "epoch": 0.46271664651350264,
      "grad_norm": 2.524909734725952,
      "learning_rate": 2.7230014980253303e-05,
      "loss": 4.2576,
      "step": 17220
    },
    {
      "epoch": 0.462851000940481,
      "grad_norm": 2.4421544075012207,
      "learning_rate": 2.722320577420673e-05,
      "loss": 4.2409,
      "step": 17225
    },
    {
      "epoch": 0.4629853553674594,
      "grad_norm": 2.226337432861328,
      "learning_rate": 2.721639656816015e-05,
      "loss": 4.1865,
      "step": 17230
    },
    {
      "epoch": 0.46311970979443773,
      "grad_norm": 2.6972339153289795,
      "learning_rate": 2.720958736211358e-05,
      "loss": 4.2183,
      "step": 17235
    },
    {
      "epoch": 0.4632540642214161,
      "grad_norm": 2.2679033279418945,
      "learning_rate": 2.7202778156067006e-05,
      "loss": 4.2592,
      "step": 17240
    },
    {
      "epoch": 0.4633884186483945,
      "grad_norm": 2.443800210952759,
      "learning_rate": 2.7195968950020428e-05,
      "loss": 4.2271,
      "step": 17245
    },
    {
      "epoch": 0.4635227730753728,
      "grad_norm": 2.4577763080596924,
      "learning_rate": 2.718915974397385e-05,
      "loss": 4.2984,
      "step": 17250
    },
    {
      "epoch": 0.4636571275023512,
      "grad_norm": 2.6686177253723145,
      "learning_rate": 2.7182350537927283e-05,
      "loss": 4.2698,
      "step": 17255
    },
    {
      "epoch": 0.46379148192932956,
      "grad_norm": 2.4595165252685547,
      "learning_rate": 2.7175541331880705e-05,
      "loss": 4.1804,
      "step": 17260
    },
    {
      "epoch": 0.46392583635630796,
      "grad_norm": 2.492065668106079,
      "learning_rate": 2.7168732125834128e-05,
      "loss": 4.1254,
      "step": 17265
    },
    {
      "epoch": 0.4640601907832863,
      "grad_norm": 2.230637311935425,
      "learning_rate": 2.7161922919787553e-05,
      "loss": 4.3272,
      "step": 17270
    },
    {
      "epoch": 0.46419454521026465,
      "grad_norm": 2.575864553451538,
      "learning_rate": 2.7155113713740982e-05,
      "loss": 4.0523,
      "step": 17275
    },
    {
      "epoch": 0.46432889963724305,
      "grad_norm": 2.1774120330810547,
      "learning_rate": 2.7148304507694405e-05,
      "loss": 4.2356,
      "step": 17280
    },
    {
      "epoch": 0.4644632540642214,
      "grad_norm": 2.5381484031677246,
      "learning_rate": 2.714149530164783e-05,
      "loss": 4.3091,
      "step": 17285
    },
    {
      "epoch": 0.4645976084911998,
      "grad_norm": 2.4964675903320312,
      "learning_rate": 2.7134686095601252e-05,
      "loss": 4.2254,
      "step": 17290
    },
    {
      "epoch": 0.46473196291817814,
      "grad_norm": 2.316100835800171,
      "learning_rate": 2.712787688955468e-05,
      "loss": 4.1697,
      "step": 17295
    },
    {
      "epoch": 0.46486631734515654,
      "grad_norm": 2.543409585952759,
      "learning_rate": 2.7121067683508107e-05,
      "loss": 4.4318,
      "step": 17300
    },
    {
      "epoch": 0.4650006717721349,
      "grad_norm": 2.597555637359619,
      "learning_rate": 2.711425847746153e-05,
      "loss": 4.2439,
      "step": 17305
    },
    {
      "epoch": 0.4651350261991133,
      "grad_norm": 2.7488832473754883,
      "learning_rate": 2.710744927141495e-05,
      "loss": 4.313,
      "step": 17310
    },
    {
      "epoch": 0.46526938062609163,
      "grad_norm": 2.3650803565979004,
      "learning_rate": 2.7100640065368377e-05,
      "loss": 4.1937,
      "step": 17315
    },
    {
      "epoch": 0.46540373505307,
      "grad_norm": 2.3278439044952393,
      "learning_rate": 2.7093830859321806e-05,
      "loss": 4.2019,
      "step": 17320
    },
    {
      "epoch": 0.4655380894800484,
      "grad_norm": 2.516564130783081,
      "learning_rate": 2.708702165327523e-05,
      "loss": 4.0634,
      "step": 17325
    },
    {
      "epoch": 0.4656724439070267,
      "grad_norm": 2.408878803253174,
      "learning_rate": 2.7080212447228654e-05,
      "loss": 4.1538,
      "step": 17330
    },
    {
      "epoch": 0.4658067983340051,
      "grad_norm": 2.5517773628234863,
      "learning_rate": 2.7073403241182077e-05,
      "loss": 4.1877,
      "step": 17335
    },
    {
      "epoch": 0.46594115276098347,
      "grad_norm": 2.3522565364837646,
      "learning_rate": 2.7066594035135506e-05,
      "loss": 4.215,
      "step": 17340
    },
    {
      "epoch": 0.46607550718796187,
      "grad_norm": 2.408596992492676,
      "learning_rate": 2.705978482908893e-05,
      "loss": 4.2429,
      "step": 17345
    },
    {
      "epoch": 0.4662098616149402,
      "grad_norm": 2.480827808380127,
      "learning_rate": 2.7052975623042354e-05,
      "loss": 4.2463,
      "step": 17350
    },
    {
      "epoch": 0.46634421604191856,
      "grad_norm": 2.461484432220459,
      "learning_rate": 2.7046166416995776e-05,
      "loss": 4.2163,
      "step": 17355
    },
    {
      "epoch": 0.46647857046889696,
      "grad_norm": 2.4052164554595947,
      "learning_rate": 2.7039357210949208e-05,
      "loss": 4.2415,
      "step": 17360
    },
    {
      "epoch": 0.4666129248958753,
      "grad_norm": 2.864283323287964,
      "learning_rate": 2.703254800490263e-05,
      "loss": 4.2927,
      "step": 17365
    },
    {
      "epoch": 0.4667472793228537,
      "grad_norm": 2.6016738414764404,
      "learning_rate": 2.7025738798856053e-05,
      "loss": 4.2187,
      "step": 17370
    },
    {
      "epoch": 0.46688163374983205,
      "grad_norm": 2.4745779037475586,
      "learning_rate": 2.701892959280948e-05,
      "loss": 4.2827,
      "step": 17375
    },
    {
      "epoch": 0.46701598817681045,
      "grad_norm": 2.3813042640686035,
      "learning_rate": 2.7012120386762907e-05,
      "loss": 4.2336,
      "step": 17380
    },
    {
      "epoch": 0.4671503426037888,
      "grad_norm": 2.422985792160034,
      "learning_rate": 2.700531118071633e-05,
      "loss": 4.2419,
      "step": 17385
    },
    {
      "epoch": 0.46728469703076714,
      "grad_norm": 2.4753432273864746,
      "learning_rate": 2.6998501974669755e-05,
      "loss": 4.0826,
      "step": 17390
    },
    {
      "epoch": 0.46741905145774554,
      "grad_norm": 2.1041274070739746,
      "learning_rate": 2.6991692768623178e-05,
      "loss": 4.191,
      "step": 17395
    },
    {
      "epoch": 0.4675534058847239,
      "grad_norm": 2.290698528289795,
      "learning_rate": 2.6984883562576607e-05,
      "loss": 4.1772,
      "step": 17400
    },
    {
      "epoch": 0.4676877603117023,
      "grad_norm": 2.7109265327453613,
      "learning_rate": 2.6978074356530032e-05,
      "loss": 4.1148,
      "step": 17405
    },
    {
      "epoch": 0.46782211473868063,
      "grad_norm": 2.367698907852173,
      "learning_rate": 2.6971265150483455e-05,
      "loss": 4.0795,
      "step": 17410
    },
    {
      "epoch": 0.46795646916565903,
      "grad_norm": 2.3059141635894775,
      "learning_rate": 2.6964455944436877e-05,
      "loss": 4.1232,
      "step": 17415
    },
    {
      "epoch": 0.4680908235926374,
      "grad_norm": 2.4251043796539307,
      "learning_rate": 2.6957646738390306e-05,
      "loss": 4.1203,
      "step": 17420
    },
    {
      "epoch": 0.4682251780196158,
      "grad_norm": 2.641751766204834,
      "learning_rate": 2.695083753234373e-05,
      "loss": 4.3078,
      "step": 17425
    },
    {
      "epoch": 0.4683595324465941,
      "grad_norm": 2.50394344329834,
      "learning_rate": 2.6944028326297154e-05,
      "loss": 4.1737,
      "step": 17430
    },
    {
      "epoch": 0.46849388687357246,
      "grad_norm": 2.2386152744293213,
      "learning_rate": 2.693721912025058e-05,
      "loss": 4.2443,
      "step": 17435
    },
    {
      "epoch": 0.46862824130055086,
      "grad_norm": 2.272341251373291,
      "learning_rate": 2.693040991420401e-05,
      "loss": 4.2314,
      "step": 17440
    },
    {
      "epoch": 0.4687625957275292,
      "grad_norm": 2.4722914695739746,
      "learning_rate": 2.692360070815743e-05,
      "loss": 4.2652,
      "step": 17445
    },
    {
      "epoch": 0.4688969501545076,
      "grad_norm": 2.5747063159942627,
      "learning_rate": 2.6916791502110857e-05,
      "loss": 4.2788,
      "step": 17450
    },
    {
      "epoch": 0.46903130458148595,
      "grad_norm": 2.4937143325805664,
      "learning_rate": 2.690998229606428e-05,
      "loss": 4.2571,
      "step": 17455
    },
    {
      "epoch": 0.46916565900846435,
      "grad_norm": 2.3794162273406982,
      "learning_rate": 2.69031730900177e-05,
      "loss": 4.2946,
      "step": 17460
    },
    {
      "epoch": 0.4693000134354427,
      "grad_norm": 2.4419944286346436,
      "learning_rate": 2.6896363883971133e-05,
      "loss": 4.1737,
      "step": 17465
    },
    {
      "epoch": 0.46943436786242104,
      "grad_norm": 2.308689594268799,
      "learning_rate": 2.6889554677924556e-05,
      "loss": 4.144,
      "step": 17470
    },
    {
      "epoch": 0.46956872228939944,
      "grad_norm": 2.5754618644714355,
      "learning_rate": 2.6882745471877978e-05,
      "loss": 4.1665,
      "step": 17475
    },
    {
      "epoch": 0.4697030767163778,
      "grad_norm": 2.61972975730896,
      "learning_rate": 2.6875936265831404e-05,
      "loss": 4.1958,
      "step": 17480
    },
    {
      "epoch": 0.4698374311433562,
      "grad_norm": 2.571610450744629,
      "learning_rate": 2.6869127059784833e-05,
      "loss": 4.2231,
      "step": 17485
    },
    {
      "epoch": 0.46997178557033453,
      "grad_norm": 2.424769163131714,
      "learning_rate": 2.6862317853738255e-05,
      "loss": 4.3764,
      "step": 17490
    },
    {
      "epoch": 0.47010613999731293,
      "grad_norm": 2.2183611392974854,
      "learning_rate": 2.685550864769168e-05,
      "loss": 4.2899,
      "step": 17495
    },
    {
      "epoch": 0.4702404944242913,
      "grad_norm": 2.213998317718506,
      "learning_rate": 2.6848699441645103e-05,
      "loss": 4.1956,
      "step": 17500
    },
    {
      "epoch": 0.4703748488512696,
      "grad_norm": 2.266749858856201,
      "learning_rate": 2.6841890235598532e-05,
      "loss": 4.2263,
      "step": 17505
    },
    {
      "epoch": 0.470509203278248,
      "grad_norm": 2.4420711994171143,
      "learning_rate": 2.6835081029551958e-05,
      "loss": 4.2332,
      "step": 17510
    },
    {
      "epoch": 0.47064355770522637,
      "grad_norm": 2.535036325454712,
      "learning_rate": 2.682827182350538e-05,
      "loss": 4.2994,
      "step": 17515
    },
    {
      "epoch": 0.47077791213220477,
      "grad_norm": 2.3297159671783447,
      "learning_rate": 2.6821462617458802e-05,
      "loss": 4.1712,
      "step": 17520
    },
    {
      "epoch": 0.4709122665591831,
      "grad_norm": 2.4511682987213135,
      "learning_rate": 2.681465341141223e-05,
      "loss": 4.2028,
      "step": 17525
    },
    {
      "epoch": 0.4710466209861615,
      "grad_norm": 2.225996732711792,
      "learning_rate": 2.6807844205365657e-05,
      "loss": 4.145,
      "step": 17530
    },
    {
      "epoch": 0.47118097541313986,
      "grad_norm": 2.2444803714752197,
      "learning_rate": 2.680103499931908e-05,
      "loss": 4.1656,
      "step": 17535
    },
    {
      "epoch": 0.47131532984011826,
      "grad_norm": 2.502364158630371,
      "learning_rate": 2.6794225793272505e-05,
      "loss": 4.1595,
      "step": 17540
    },
    {
      "epoch": 0.4714496842670966,
      "grad_norm": 2.463336229324341,
      "learning_rate": 2.6787416587225934e-05,
      "loss": 4.2751,
      "step": 17545
    },
    {
      "epoch": 0.47158403869407495,
      "grad_norm": 2.412121295928955,
      "learning_rate": 2.6780607381179356e-05,
      "loss": 4.181,
      "step": 17550
    },
    {
      "epoch": 0.47171839312105335,
      "grad_norm": 2.612711191177368,
      "learning_rate": 2.6773798175132782e-05,
      "loss": 4.1914,
      "step": 17555
    },
    {
      "epoch": 0.4718527475480317,
      "grad_norm": 2.5185494422912598,
      "learning_rate": 2.6766988969086204e-05,
      "loss": 4.1948,
      "step": 17560
    },
    {
      "epoch": 0.4719871019750101,
      "grad_norm": 2.2122249603271484,
      "learning_rate": 2.6760179763039633e-05,
      "loss": 4.2405,
      "step": 17565
    },
    {
      "epoch": 0.47212145640198844,
      "grad_norm": 2.373488187789917,
      "learning_rate": 2.6753370556993055e-05,
      "loss": 4.1455,
      "step": 17570
    },
    {
      "epoch": 0.47225581082896684,
      "grad_norm": 2.4546902179718018,
      "learning_rate": 2.674656135094648e-05,
      "loss": 4.1075,
      "step": 17575
    },
    {
      "epoch": 0.4723901652559452,
      "grad_norm": 2.311859130859375,
      "learning_rate": 2.6739752144899903e-05,
      "loss": 4.1556,
      "step": 17580
    },
    {
      "epoch": 0.4725245196829235,
      "grad_norm": 2.518477201461792,
      "learning_rate": 2.6732942938853332e-05,
      "loss": 4.0893,
      "step": 17585
    },
    {
      "epoch": 0.4726588741099019,
      "grad_norm": 2.4475951194763184,
      "learning_rate": 2.6726133732806758e-05,
      "loss": 4.2612,
      "step": 17590
    },
    {
      "epoch": 0.4727932285368803,
      "grad_norm": 2.544950246810913,
      "learning_rate": 2.671932452676018e-05,
      "loss": 4.1614,
      "step": 17595
    },
    {
      "epoch": 0.4729275829638587,
      "grad_norm": 2.3136157989501953,
      "learning_rate": 2.6712515320713606e-05,
      "loss": 4.2227,
      "step": 17600
    },
    {
      "epoch": 0.473061937390837,
      "grad_norm": 2.471916437149048,
      "learning_rate": 2.6705706114667028e-05,
      "loss": 4.2485,
      "step": 17605
    },
    {
      "epoch": 0.4731962918178154,
      "grad_norm": 2.614469289779663,
      "learning_rate": 2.6698896908620457e-05,
      "loss": 4.1375,
      "step": 17610
    },
    {
      "epoch": 0.47333064624479376,
      "grad_norm": 2.3315632343292236,
      "learning_rate": 2.6692087702573883e-05,
      "loss": 4.2119,
      "step": 17615
    },
    {
      "epoch": 0.47346500067177216,
      "grad_norm": 2.298759937286377,
      "learning_rate": 2.6685278496527305e-05,
      "loss": 4.2169,
      "step": 17620
    },
    {
      "epoch": 0.4735993550987505,
      "grad_norm": 2.42462158203125,
      "learning_rate": 2.6678469290480727e-05,
      "loss": 4.1998,
      "step": 17625
    },
    {
      "epoch": 0.47373370952572885,
      "grad_norm": 2.3904147148132324,
      "learning_rate": 2.6671660084434156e-05,
      "loss": 4.1437,
      "step": 17630
    },
    {
      "epoch": 0.47386806395270725,
      "grad_norm": 2.4099156856536865,
      "learning_rate": 2.6664850878387582e-05,
      "loss": 4.234,
      "step": 17635
    },
    {
      "epoch": 0.4740024183796856,
      "grad_norm": 2.592886447906494,
      "learning_rate": 2.6658041672341004e-05,
      "loss": 4.1939,
      "step": 17640
    },
    {
      "epoch": 0.474136772806664,
      "grad_norm": 2.1890766620635986,
      "learning_rate": 2.665123246629443e-05,
      "loss": 4.3098,
      "step": 17645
    },
    {
      "epoch": 0.47427112723364234,
      "grad_norm": 2.3893442153930664,
      "learning_rate": 2.664442326024786e-05,
      "loss": 4.2674,
      "step": 17650
    },
    {
      "epoch": 0.47440548166062074,
      "grad_norm": 2.4617128372192383,
      "learning_rate": 2.663761405420128e-05,
      "loss": 4.1032,
      "step": 17655
    },
    {
      "epoch": 0.4745398360875991,
      "grad_norm": 2.385512590408325,
      "learning_rate": 2.6630804848154707e-05,
      "loss": 4.2033,
      "step": 17660
    },
    {
      "epoch": 0.47467419051457743,
      "grad_norm": 2.579432487487793,
      "learning_rate": 2.662399564210813e-05,
      "loss": 4.1823,
      "step": 17665
    },
    {
      "epoch": 0.47480854494155583,
      "grad_norm": 2.691681385040283,
      "learning_rate": 2.661718643606156e-05,
      "loss": 4.2908,
      "step": 17670
    },
    {
      "epoch": 0.4749428993685342,
      "grad_norm": 2.675417423248291,
      "learning_rate": 2.661037723001498e-05,
      "loss": 4.3303,
      "step": 17675
    },
    {
      "epoch": 0.4750772537955126,
      "grad_norm": 2.2550888061523438,
      "learning_rate": 2.6603568023968406e-05,
      "loss": 4.2557,
      "step": 17680
    },
    {
      "epoch": 0.4752116082224909,
      "grad_norm": 2.2960779666900635,
      "learning_rate": 2.659675881792183e-05,
      "loss": 4.2482,
      "step": 17685
    },
    {
      "epoch": 0.4753459626494693,
      "grad_norm": 2.44120192527771,
      "learning_rate": 2.6589949611875258e-05,
      "loss": 4.3297,
      "step": 17690
    },
    {
      "epoch": 0.47548031707644767,
      "grad_norm": 2.276158332824707,
      "learning_rate": 2.6583140405828683e-05,
      "loss": 4.1763,
      "step": 17695
    },
    {
      "epoch": 0.475614671503426,
      "grad_norm": 2.441298007965088,
      "learning_rate": 2.6576331199782106e-05,
      "loss": 4.288,
      "step": 17700
    },
    {
      "epoch": 0.4757490259304044,
      "grad_norm": 2.1532068252563477,
      "learning_rate": 2.656952199373553e-05,
      "loss": 4.2026,
      "step": 17705
    },
    {
      "epoch": 0.47588338035738276,
      "grad_norm": 2.3710954189300537,
      "learning_rate": 2.656271278768896e-05,
      "loss": 4.235,
      "step": 17710
    },
    {
      "epoch": 0.47601773478436116,
      "grad_norm": 2.4574921131134033,
      "learning_rate": 2.6555903581642382e-05,
      "loss": 4.1355,
      "step": 17715
    },
    {
      "epoch": 0.4761520892113395,
      "grad_norm": 2.499556303024292,
      "learning_rate": 2.6549094375595808e-05,
      "loss": 4.2417,
      "step": 17720
    },
    {
      "epoch": 0.4762864436383179,
      "grad_norm": 2.3928616046905518,
      "learning_rate": 2.654228516954923e-05,
      "loss": 4.1453,
      "step": 17725
    },
    {
      "epoch": 0.47642079806529625,
      "grad_norm": 2.384659767150879,
      "learning_rate": 2.653547596350266e-05,
      "loss": 4.2524,
      "step": 17730
    },
    {
      "epoch": 0.47655515249227465,
      "grad_norm": 2.28247332572937,
      "learning_rate": 2.6528666757456082e-05,
      "loss": 4.0807,
      "step": 17735
    },
    {
      "epoch": 0.476689506919253,
      "grad_norm": 2.531473398208618,
      "learning_rate": 2.6521857551409507e-05,
      "loss": 4.2648,
      "step": 17740
    },
    {
      "epoch": 0.47682386134623134,
      "grad_norm": 2.4178764820098877,
      "learning_rate": 2.651504834536293e-05,
      "loss": 4.2422,
      "step": 17745
    },
    {
      "epoch": 0.47695821577320974,
      "grad_norm": 2.4847092628479004,
      "learning_rate": 2.6508239139316355e-05,
      "loss": 4.1256,
      "step": 17750
    },
    {
      "epoch": 0.4770925702001881,
      "grad_norm": 2.4517786502838135,
      "learning_rate": 2.6501429933269784e-05,
      "loss": 4.218,
      "step": 17755
    },
    {
      "epoch": 0.4772269246271665,
      "grad_norm": 2.4654595851898193,
      "learning_rate": 2.6494620727223207e-05,
      "loss": 4.1453,
      "step": 17760
    },
    {
      "epoch": 0.4773612790541448,
      "grad_norm": 2.3986763954162598,
      "learning_rate": 2.6487811521176632e-05,
      "loss": 4.1833,
      "step": 17765
    },
    {
      "epoch": 0.4774956334811232,
      "grad_norm": 2.4019174575805664,
      "learning_rate": 2.6481002315130055e-05,
      "loss": 4.2353,
      "step": 17770
    },
    {
      "epoch": 0.47762998790810157,
      "grad_norm": 2.45977783203125,
      "learning_rate": 2.6474193109083484e-05,
      "loss": 4.126,
      "step": 17775
    },
    {
      "epoch": 0.4777643423350799,
      "grad_norm": 2.6667075157165527,
      "learning_rate": 2.6467383903036906e-05,
      "loss": 4.1043,
      "step": 17780
    },
    {
      "epoch": 0.4778986967620583,
      "grad_norm": 2.2129361629486084,
      "learning_rate": 2.646057469699033e-05,
      "loss": 4.186,
      "step": 17785
    },
    {
      "epoch": 0.47803305118903666,
      "grad_norm": 2.371061325073242,
      "learning_rate": 2.6453765490943754e-05,
      "loss": 4.3292,
      "step": 17790
    },
    {
      "epoch": 0.47816740561601506,
      "grad_norm": 2.7223236560821533,
      "learning_rate": 2.6446956284897183e-05,
      "loss": 4.1094,
      "step": 17795
    },
    {
      "epoch": 0.4783017600429934,
      "grad_norm": 2.4327235221862793,
      "learning_rate": 2.644014707885061e-05,
      "loss": 4.1131,
      "step": 17800
    },
    {
      "epoch": 0.4784361144699718,
      "grad_norm": 2.59607195854187,
      "learning_rate": 2.643333787280403e-05,
      "loss": 4.1903,
      "step": 17805
    },
    {
      "epoch": 0.47857046889695015,
      "grad_norm": 2.2929842472076416,
      "learning_rate": 2.6426528666757456e-05,
      "loss": 4.1537,
      "step": 17810
    },
    {
      "epoch": 0.4787048233239285,
      "grad_norm": 2.3058764934539795,
      "learning_rate": 2.6419719460710885e-05,
      "loss": 4.1598,
      "step": 17815
    },
    {
      "epoch": 0.4788391777509069,
      "grad_norm": 2.4818291664123535,
      "learning_rate": 2.6412910254664308e-05,
      "loss": 4.3039,
      "step": 17820
    },
    {
      "epoch": 0.47897353217788524,
      "grad_norm": 2.560182809829712,
      "learning_rate": 2.6406101048617733e-05,
      "loss": 4.2255,
      "step": 17825
    },
    {
      "epoch": 0.47910788660486364,
      "grad_norm": 2.327904462814331,
      "learning_rate": 2.6399291842571156e-05,
      "loss": 4.2429,
      "step": 17830
    },
    {
      "epoch": 0.479242241031842,
      "grad_norm": 2.2578577995300293,
      "learning_rate": 2.6392482636524585e-05,
      "loss": 4.3041,
      "step": 17835
    },
    {
      "epoch": 0.4793765954588204,
      "grad_norm": 2.7132349014282227,
      "learning_rate": 2.6385673430478007e-05,
      "loss": 4.1051,
      "step": 17840
    },
    {
      "epoch": 0.47951094988579873,
      "grad_norm": 2.4492204189300537,
      "learning_rate": 2.6378864224431433e-05,
      "loss": 4.2183,
      "step": 17845
    },
    {
      "epoch": 0.47964530431277713,
      "grad_norm": 2.81508207321167,
      "learning_rate": 2.6372055018384855e-05,
      "loss": 4.1398,
      "step": 17850
    },
    {
      "epoch": 0.4797796587397555,
      "grad_norm": 2.25669002532959,
      "learning_rate": 2.6365245812338284e-05,
      "loss": 4.2405,
      "step": 17855
    },
    {
      "epoch": 0.4799140131667338,
      "grad_norm": 2.654933452606201,
      "learning_rate": 2.635843660629171e-05,
      "loss": 4.2503,
      "step": 17860
    },
    {
      "epoch": 0.4800483675937122,
      "grad_norm": 2.435884714126587,
      "learning_rate": 2.6351627400245132e-05,
      "loss": 4.1012,
      "step": 17865
    },
    {
      "epoch": 0.48018272202069057,
      "grad_norm": 2.656956672668457,
      "learning_rate": 2.6344818194198558e-05,
      "loss": 4.2217,
      "step": 17870
    },
    {
      "epoch": 0.48031707644766897,
      "grad_norm": 2.3096468448638916,
      "learning_rate": 2.6338008988151987e-05,
      "loss": 4.1609,
      "step": 17875
    },
    {
      "epoch": 0.4804514308746473,
      "grad_norm": 2.297157049179077,
      "learning_rate": 2.633119978210541e-05,
      "loss": 4.117,
      "step": 17880
    },
    {
      "epoch": 0.4805857853016257,
      "grad_norm": 2.4415862560272217,
      "learning_rate": 2.632439057605883e-05,
      "loss": 4.2962,
      "step": 17885
    },
    {
      "epoch": 0.48072013972860406,
      "grad_norm": 2.620678186416626,
      "learning_rate": 2.6317581370012257e-05,
      "loss": 4.2288,
      "step": 17890
    },
    {
      "epoch": 0.4808544941555824,
      "grad_norm": 2.480482578277588,
      "learning_rate": 2.631077216396568e-05,
      "loss": 4.1715,
      "step": 17895
    },
    {
      "epoch": 0.4809888485825608,
      "grad_norm": 2.3759844303131104,
      "learning_rate": 2.6303962957919108e-05,
      "loss": 4.2503,
      "step": 17900
    },
    {
      "epoch": 0.48112320300953915,
      "grad_norm": 2.4821155071258545,
      "learning_rate": 2.6297153751872534e-05,
      "loss": 4.1572,
      "step": 17905
    },
    {
      "epoch": 0.48125755743651755,
      "grad_norm": 2.5098817348480225,
      "learning_rate": 2.6290344545825956e-05,
      "loss": 4.1967,
      "step": 17910
    },
    {
      "epoch": 0.4813919118634959,
      "grad_norm": 2.459284782409668,
      "learning_rate": 2.628353533977938e-05,
      "loss": 4.1701,
      "step": 17915
    },
    {
      "epoch": 0.4815262662904743,
      "grad_norm": 2.27101469039917,
      "learning_rate": 2.627672613373281e-05,
      "loss": 4.1538,
      "step": 17920
    },
    {
      "epoch": 0.48166062071745264,
      "grad_norm": 2.3389089107513428,
      "learning_rate": 2.6269916927686233e-05,
      "loss": 4.1438,
      "step": 17925
    },
    {
      "epoch": 0.481794975144431,
      "grad_norm": 2.3425369262695312,
      "learning_rate": 2.626310772163966e-05,
      "loss": 4.1803,
      "step": 17930
    },
    {
      "epoch": 0.4819293295714094,
      "grad_norm": 2.4553651809692383,
      "learning_rate": 2.625629851559308e-05,
      "loss": 4.233,
      "step": 17935
    },
    {
      "epoch": 0.4820636839983877,
      "grad_norm": 2.6110517978668213,
      "learning_rate": 2.624948930954651e-05,
      "loss": 4.2309,
      "step": 17940
    },
    {
      "epoch": 0.4821980384253661,
      "grad_norm": 2.1823177337646484,
      "learning_rate": 2.6242680103499932e-05,
      "loss": 4.1027,
      "step": 17945
    },
    {
      "epoch": 0.48233239285234447,
      "grad_norm": 2.272648811340332,
      "learning_rate": 2.6235870897453358e-05,
      "loss": 4.2132,
      "step": 17950
    },
    {
      "epoch": 0.48246674727932287,
      "grad_norm": 2.4722824096679688,
      "learning_rate": 2.622906169140678e-05,
      "loss": 4.0944,
      "step": 17955
    },
    {
      "epoch": 0.4826011017063012,
      "grad_norm": 2.638125419616699,
      "learning_rate": 2.622225248536021e-05,
      "loss": 4.0836,
      "step": 17960
    },
    {
      "epoch": 0.4827354561332796,
      "grad_norm": 2.454495668411255,
      "learning_rate": 2.6215443279313635e-05,
      "loss": 4.2388,
      "step": 17965
    },
    {
      "epoch": 0.48286981056025796,
      "grad_norm": 2.429551362991333,
      "learning_rate": 2.6208634073267057e-05,
      "loss": 4.1451,
      "step": 17970
    },
    {
      "epoch": 0.4830041649872363,
      "grad_norm": 2.261261224746704,
      "learning_rate": 2.6201824867220483e-05,
      "loss": 4.0884,
      "step": 17975
    },
    {
      "epoch": 0.4831385194142147,
      "grad_norm": 2.5089619159698486,
      "learning_rate": 2.6195015661173912e-05,
      "loss": 4.2319,
      "step": 17980
    },
    {
      "epoch": 0.48327287384119305,
      "grad_norm": 2.2266359329223633,
      "learning_rate": 2.6188206455127334e-05,
      "loss": 4.1895,
      "step": 17985
    },
    {
      "epoch": 0.48340722826817145,
      "grad_norm": 2.4228975772857666,
      "learning_rate": 2.6181397249080756e-05,
      "loss": 4.302,
      "step": 17990
    },
    {
      "epoch": 0.4835415826951498,
      "grad_norm": 2.444434881210327,
      "learning_rate": 2.6174588043034182e-05,
      "loss": 4.132,
      "step": 17995
    },
    {
      "epoch": 0.4836759371221282,
      "grad_norm": 2.5518808364868164,
      "learning_rate": 2.616777883698761e-05,
      "loss": 4.3463,
      "step": 18000
    },
    {
      "epoch": 0.48381029154910654,
      "grad_norm": 2.2340524196624756,
      "learning_rate": 2.6160969630941033e-05,
      "loss": 4.1674,
      "step": 18005
    },
    {
      "epoch": 0.4839446459760849,
      "grad_norm": 2.3275156021118164,
      "learning_rate": 2.615416042489446e-05,
      "loss": 4.0952,
      "step": 18010
    },
    {
      "epoch": 0.4840790004030633,
      "grad_norm": 2.6701903343200684,
      "learning_rate": 2.614735121884788e-05,
      "loss": 4.1777,
      "step": 18015
    },
    {
      "epoch": 0.48421335483004163,
      "grad_norm": 2.301490545272827,
      "learning_rate": 2.614054201280131e-05,
      "loss": 4.23,
      "step": 18020
    },
    {
      "epoch": 0.48434770925702003,
      "grad_norm": 2.430833339691162,
      "learning_rate": 2.6133732806754736e-05,
      "loss": 4.28,
      "step": 18025
    },
    {
      "epoch": 0.4844820636839984,
      "grad_norm": 2.329726219177246,
      "learning_rate": 2.6126923600708158e-05,
      "loss": 4.1946,
      "step": 18030
    },
    {
      "epoch": 0.4846164181109768,
      "grad_norm": 2.191117525100708,
      "learning_rate": 2.6120114394661584e-05,
      "loss": 4.2334,
      "step": 18035
    },
    {
      "epoch": 0.4847507725379551,
      "grad_norm": 2.4942314624786377,
      "learning_rate": 2.6113305188615006e-05,
      "loss": 4.2172,
      "step": 18040
    },
    {
      "epoch": 0.4848851269649335,
      "grad_norm": 2.5436015129089355,
      "learning_rate": 2.6106495982568435e-05,
      "loss": 4.2046,
      "step": 18045
    },
    {
      "epoch": 0.48501948139191187,
      "grad_norm": 2.5554635524749756,
      "learning_rate": 2.6099686776521857e-05,
      "loss": 4.1776,
      "step": 18050
    },
    {
      "epoch": 0.4851538358188902,
      "grad_norm": 2.2278881072998047,
      "learning_rate": 2.6092877570475283e-05,
      "loss": 4.2295,
      "step": 18055
    },
    {
      "epoch": 0.4852881902458686,
      "grad_norm": 2.4933924674987793,
      "learning_rate": 2.6086068364428705e-05,
      "loss": 4.1942,
      "step": 18060
    },
    {
      "epoch": 0.48542254467284696,
      "grad_norm": 2.309387683868408,
      "learning_rate": 2.6079259158382134e-05,
      "loss": 4.2468,
      "step": 18065
    },
    {
      "epoch": 0.48555689909982536,
      "grad_norm": 2.7151975631713867,
      "learning_rate": 2.607244995233556e-05,
      "loss": 4.0826,
      "step": 18070
    },
    {
      "epoch": 0.4856912535268037,
      "grad_norm": 2.4385697841644287,
      "learning_rate": 2.6065640746288982e-05,
      "loss": 4.2234,
      "step": 18075
    },
    {
      "epoch": 0.4858256079537821,
      "grad_norm": 2.408919095993042,
      "learning_rate": 2.6058831540242408e-05,
      "loss": 4.2883,
      "step": 18080
    },
    {
      "epoch": 0.48595996238076045,
      "grad_norm": 2.4114458560943604,
      "learning_rate": 2.6052022334195837e-05,
      "loss": 4.2383,
      "step": 18085
    },
    {
      "epoch": 0.4860943168077388,
      "grad_norm": 2.50626802444458,
      "learning_rate": 2.604521312814926e-05,
      "loss": 4.1921,
      "step": 18090
    },
    {
      "epoch": 0.4862286712347172,
      "grad_norm": 2.3679516315460205,
      "learning_rate": 2.603840392210268e-05,
      "loss": 4.1849,
      "step": 18095
    },
    {
      "epoch": 0.48636302566169554,
      "grad_norm": 2.5321896076202393,
      "learning_rate": 2.6031594716056107e-05,
      "loss": 4.2015,
      "step": 18100
    },
    {
      "epoch": 0.48649738008867394,
      "grad_norm": 2.3566226959228516,
      "learning_rate": 2.6024785510009536e-05,
      "loss": 4.3387,
      "step": 18105
    },
    {
      "epoch": 0.4866317345156523,
      "grad_norm": 2.444742441177368,
      "learning_rate": 2.601797630396296e-05,
      "loss": 4.2679,
      "step": 18110
    },
    {
      "epoch": 0.4867660889426307,
      "grad_norm": 2.2895700931549072,
      "learning_rate": 2.6011167097916384e-05,
      "loss": 4.2327,
      "step": 18115
    },
    {
      "epoch": 0.486900443369609,
      "grad_norm": 2.533116340637207,
      "learning_rate": 2.6004357891869807e-05,
      "loss": 4.1546,
      "step": 18120
    },
    {
      "epoch": 0.48703479779658737,
      "grad_norm": 2.1641852855682373,
      "learning_rate": 2.5997548685823236e-05,
      "loss": 4.299,
      "step": 18125
    },
    {
      "epoch": 0.48716915222356577,
      "grad_norm": 2.540921688079834,
      "learning_rate": 2.599073947977666e-05,
      "loss": 4.2227,
      "step": 18130
    },
    {
      "epoch": 0.4873035066505441,
      "grad_norm": 2.612032651901245,
      "learning_rate": 2.5983930273730083e-05,
      "loss": 4.2153,
      "step": 18135
    },
    {
      "epoch": 0.4874378610775225,
      "grad_norm": 2.280641555786133,
      "learning_rate": 2.597712106768351e-05,
      "loss": 4.2552,
      "step": 18140
    },
    {
      "epoch": 0.48757221550450086,
      "grad_norm": 2.409398078918457,
      "learning_rate": 2.5970311861636938e-05,
      "loss": 4.2235,
      "step": 18145
    },
    {
      "epoch": 0.48770656993147926,
      "grad_norm": 2.3794333934783936,
      "learning_rate": 2.596350265559036e-05,
      "loss": 4.0942,
      "step": 18150
    },
    {
      "epoch": 0.4878409243584576,
      "grad_norm": 2.6261370182037354,
      "learning_rate": 2.5956693449543783e-05,
      "loss": 4.1852,
      "step": 18155
    },
    {
      "epoch": 0.487975278785436,
      "grad_norm": 2.2404489517211914,
      "learning_rate": 2.594988424349721e-05,
      "loss": 4.2908,
      "step": 18160
    },
    {
      "epoch": 0.48810963321241435,
      "grad_norm": 2.5504374504089355,
      "learning_rate": 2.5943075037450637e-05,
      "loss": 4.2369,
      "step": 18165
    },
    {
      "epoch": 0.4882439876393927,
      "grad_norm": 2.4524526596069336,
      "learning_rate": 2.593626583140406e-05,
      "loss": 4.269,
      "step": 18170
    },
    {
      "epoch": 0.4883783420663711,
      "grad_norm": 2.5140442848205566,
      "learning_rate": 2.5929456625357485e-05,
      "loss": 4.1981,
      "step": 18175
    },
    {
      "epoch": 0.48851269649334944,
      "grad_norm": 2.481140613555908,
      "learning_rate": 2.5922647419310908e-05,
      "loss": 4.248,
      "step": 18180
    },
    {
      "epoch": 0.48864705092032784,
      "grad_norm": 2.419713020324707,
      "learning_rate": 2.5915838213264337e-05,
      "loss": 4.111,
      "step": 18185
    },
    {
      "epoch": 0.4887814053473062,
      "grad_norm": 2.344277858734131,
      "learning_rate": 2.5909029007217762e-05,
      "loss": 4.2457,
      "step": 18190
    },
    {
      "epoch": 0.4889157597742846,
      "grad_norm": 2.5366199016571045,
      "learning_rate": 2.5902219801171185e-05,
      "loss": 4.171,
      "step": 18195
    },
    {
      "epoch": 0.48905011420126293,
      "grad_norm": 2.462343692779541,
      "learning_rate": 2.5895410595124607e-05,
      "loss": 4.1888,
      "step": 18200
    },
    {
      "epoch": 0.4891844686282413,
      "grad_norm": 2.4411468505859375,
      "learning_rate": 2.5888601389078033e-05,
      "loss": 4.2335,
      "step": 18205
    },
    {
      "epoch": 0.4893188230552197,
      "grad_norm": 2.454390048980713,
      "learning_rate": 2.588179218303146e-05,
      "loss": 4.1776,
      "step": 18210
    },
    {
      "epoch": 0.489453177482198,
      "grad_norm": 2.463078022003174,
      "learning_rate": 2.5874982976984884e-05,
      "loss": 4.1615,
      "step": 18215
    },
    {
      "epoch": 0.4895875319091764,
      "grad_norm": 2.545379161834717,
      "learning_rate": 2.586817377093831e-05,
      "loss": 4.1149,
      "step": 18220
    },
    {
      "epoch": 0.48972188633615477,
      "grad_norm": 2.4900882244110107,
      "learning_rate": 2.5861364564891732e-05,
      "loss": 4.1357,
      "step": 18225
    },
    {
      "epoch": 0.48985624076313317,
      "grad_norm": 2.1092634201049805,
      "learning_rate": 2.585455535884516e-05,
      "loss": 4.2677,
      "step": 18230
    },
    {
      "epoch": 0.4899905951901115,
      "grad_norm": 2.2814910411834717,
      "learning_rate": 2.5847746152798586e-05,
      "loss": 4.1796,
      "step": 18235
    },
    {
      "epoch": 0.49012494961708986,
      "grad_norm": 2.356308937072754,
      "learning_rate": 2.584093694675201e-05,
      "loss": 4.2341,
      "step": 18240
    },
    {
      "epoch": 0.49025930404406826,
      "grad_norm": 2.57517409324646,
      "learning_rate": 2.5834127740705434e-05,
      "loss": 4.1961,
      "step": 18245
    },
    {
      "epoch": 0.4903936584710466,
      "grad_norm": 2.4986705780029297,
      "learning_rate": 2.5827318534658863e-05,
      "loss": 4.0699,
      "step": 18250
    },
    {
      "epoch": 0.490528012898025,
      "grad_norm": 2.3548741340637207,
      "learning_rate": 2.5820509328612286e-05,
      "loss": 4.1076,
      "step": 18255
    },
    {
      "epoch": 0.49066236732500335,
      "grad_norm": 2.4799814224243164,
      "learning_rate": 2.5813700122565708e-05,
      "loss": 4.2076,
      "step": 18260
    },
    {
      "epoch": 0.49079672175198175,
      "grad_norm": 2.2077834606170654,
      "learning_rate": 2.5806890916519134e-05,
      "loss": 4.281,
      "step": 18265
    },
    {
      "epoch": 0.4909310761789601,
      "grad_norm": 2.4273972511291504,
      "learning_rate": 2.5800081710472563e-05,
      "loss": 4.0183,
      "step": 18270
    },
    {
      "epoch": 0.4910654306059385,
      "grad_norm": 2.2849979400634766,
      "learning_rate": 2.5793272504425985e-05,
      "loss": 4.3167,
      "step": 18275
    },
    {
      "epoch": 0.49119978503291684,
      "grad_norm": 2.0790834426879883,
      "learning_rate": 2.578646329837941e-05,
      "loss": 4.048,
      "step": 18280
    },
    {
      "epoch": 0.4913341394598952,
      "grad_norm": 2.6466872692108154,
      "learning_rate": 2.5779654092332833e-05,
      "loss": 4.228,
      "step": 18285
    },
    {
      "epoch": 0.4914684938868736,
      "grad_norm": 2.357388734817505,
      "learning_rate": 2.5772844886286262e-05,
      "loss": 4.1201,
      "step": 18290
    },
    {
      "epoch": 0.4916028483138519,
      "grad_norm": 2.4313485622406006,
      "learning_rate": 2.5766035680239688e-05,
      "loss": 4.035,
      "step": 18295
    },
    {
      "epoch": 0.4917372027408303,
      "grad_norm": 2.523571014404297,
      "learning_rate": 2.575922647419311e-05,
      "loss": 4.2451,
      "step": 18300
    },
    {
      "epoch": 0.49187155716780867,
      "grad_norm": 2.2592031955718994,
      "learning_rate": 2.5752417268146532e-05,
      "loss": 4.2008,
      "step": 18305
    },
    {
      "epoch": 0.49200591159478707,
      "grad_norm": 2.4525113105773926,
      "learning_rate": 2.5745608062099965e-05,
      "loss": 4.288,
      "step": 18310
    },
    {
      "epoch": 0.4921402660217654,
      "grad_norm": 2.269056558609009,
      "learning_rate": 2.5738798856053387e-05,
      "loss": 4.2353,
      "step": 18315
    },
    {
      "epoch": 0.49227462044874376,
      "grad_norm": 2.251535654067993,
      "learning_rate": 2.573198965000681e-05,
      "loss": 4.3064,
      "step": 18320
    },
    {
      "epoch": 0.49240897487572216,
      "grad_norm": 2.343065023422241,
      "learning_rate": 2.5725180443960235e-05,
      "loss": 4.1948,
      "step": 18325
    },
    {
      "epoch": 0.4925433293027005,
      "grad_norm": 2.4005606174468994,
      "learning_rate": 2.5718371237913664e-05,
      "loss": 4.0632,
      "step": 18330
    },
    {
      "epoch": 0.4926776837296789,
      "grad_norm": 2.308690309524536,
      "learning_rate": 2.5711562031867086e-05,
      "loss": 4.1645,
      "step": 18335
    },
    {
      "epoch": 0.49281203815665725,
      "grad_norm": 2.5850331783294678,
      "learning_rate": 2.5704752825820512e-05,
      "loss": 4.2443,
      "step": 18340
    },
    {
      "epoch": 0.49294639258363565,
      "grad_norm": 2.6231513023376465,
      "learning_rate": 2.5697943619773934e-05,
      "loss": 4.2387,
      "step": 18345
    },
    {
      "epoch": 0.493080747010614,
      "grad_norm": 2.2974085807800293,
      "learning_rate": 2.569113441372736e-05,
      "loss": 4.2754,
      "step": 18350
    },
    {
      "epoch": 0.4932151014375924,
      "grad_norm": 2.4447906017303467,
      "learning_rate": 2.568432520768079e-05,
      "loss": 4.1533,
      "step": 18355
    },
    {
      "epoch": 0.49334945586457074,
      "grad_norm": 2.3164422512054443,
      "learning_rate": 2.567751600163421e-05,
      "loss": 4.1415,
      "step": 18360
    },
    {
      "epoch": 0.4934838102915491,
      "grad_norm": 2.508737087249756,
      "learning_rate": 2.5670706795587633e-05,
      "loss": 4.0885,
      "step": 18365
    },
    {
      "epoch": 0.4936181647185275,
      "grad_norm": 2.281087875366211,
      "learning_rate": 2.566389758954106e-05,
      "loss": 4.2904,
      "step": 18370
    },
    {
      "epoch": 0.49375251914550583,
      "grad_norm": 2.570521831512451,
      "learning_rate": 2.5657088383494488e-05,
      "loss": 4.2599,
      "step": 18375
    },
    {
      "epoch": 0.49388687357248423,
      "grad_norm": 2.4338788986206055,
      "learning_rate": 2.565027917744791e-05,
      "loss": 4.1128,
      "step": 18380
    },
    {
      "epoch": 0.4940212279994626,
      "grad_norm": 2.37184739112854,
      "learning_rate": 2.5643469971401336e-05,
      "loss": 4.2131,
      "step": 18385
    },
    {
      "epoch": 0.494155582426441,
      "grad_norm": 2.2635748386383057,
      "learning_rate": 2.5636660765354758e-05,
      "loss": 4.3039,
      "step": 18390
    },
    {
      "epoch": 0.4942899368534193,
      "grad_norm": 2.269552707672119,
      "learning_rate": 2.5629851559308187e-05,
      "loss": 4.2044,
      "step": 18395
    },
    {
      "epoch": 0.49442429128039767,
      "grad_norm": 2.489553451538086,
      "learning_rate": 2.5623042353261613e-05,
      "loss": 4.1601,
      "step": 18400
    },
    {
      "epoch": 0.49455864570737607,
      "grad_norm": 2.3646981716156006,
      "learning_rate": 2.5616233147215035e-05,
      "loss": 4.111,
      "step": 18405
    },
    {
      "epoch": 0.4946930001343544,
      "grad_norm": 2.3927223682403564,
      "learning_rate": 2.5609423941168457e-05,
      "loss": 4.0381,
      "step": 18410
    },
    {
      "epoch": 0.4948273545613328,
      "grad_norm": 2.395575761795044,
      "learning_rate": 2.560261473512189e-05,
      "loss": 4.253,
      "step": 18415
    },
    {
      "epoch": 0.49496170898831116,
      "grad_norm": 2.612895965576172,
      "learning_rate": 2.5595805529075312e-05,
      "loss": 4.2215,
      "step": 18420
    },
    {
      "epoch": 0.49509606341528956,
      "grad_norm": 2.5365970134735107,
      "learning_rate": 2.5588996323028734e-05,
      "loss": 4.1465,
      "step": 18425
    },
    {
      "epoch": 0.4952304178422679,
      "grad_norm": 2.2703559398651123,
      "learning_rate": 2.558218711698216e-05,
      "loss": 4.2273,
      "step": 18430
    },
    {
      "epoch": 0.49536477226924625,
      "grad_norm": 2.5235514640808105,
      "learning_rate": 2.557537791093559e-05,
      "loss": 4.2287,
      "step": 18435
    },
    {
      "epoch": 0.49549912669622465,
      "grad_norm": 2.7003815174102783,
      "learning_rate": 2.556856870488901e-05,
      "loss": 4.2661,
      "step": 18440
    },
    {
      "epoch": 0.495633481123203,
      "grad_norm": 2.1436195373535156,
      "learning_rate": 2.5561759498842437e-05,
      "loss": 4.1953,
      "step": 18445
    },
    {
      "epoch": 0.4957678355501814,
      "grad_norm": 2.4063124656677246,
      "learning_rate": 2.555495029279586e-05,
      "loss": 4.1223,
      "step": 18450
    },
    {
      "epoch": 0.49590218997715974,
      "grad_norm": 2.580117702484131,
      "learning_rate": 2.5548141086749288e-05,
      "loss": 4.1311,
      "step": 18455
    },
    {
      "epoch": 0.49603654440413814,
      "grad_norm": 2.4798243045806885,
      "learning_rate": 2.5541331880702714e-05,
      "loss": 4.338,
      "step": 18460
    },
    {
      "epoch": 0.4961708988311165,
      "grad_norm": 2.1716935634613037,
      "learning_rate": 2.5534522674656136e-05,
      "loss": 4.1286,
      "step": 18465
    },
    {
      "epoch": 0.4963052532580949,
      "grad_norm": 2.449183464050293,
      "learning_rate": 2.552771346860956e-05,
      "loss": 4.1662,
      "step": 18470
    },
    {
      "epoch": 0.4964396076850732,
      "grad_norm": 2.6888210773468018,
      "learning_rate": 2.5520904262562988e-05,
      "loss": 4.1947,
      "step": 18475
    },
    {
      "epoch": 0.49657396211205157,
      "grad_norm": 2.248119831085205,
      "learning_rate": 2.5514095056516413e-05,
      "loss": 4.0818,
      "step": 18480
    },
    {
      "epoch": 0.49670831653902997,
      "grad_norm": 2.6109721660614014,
      "learning_rate": 2.5507285850469835e-05,
      "loss": 4.0692,
      "step": 18485
    },
    {
      "epoch": 0.4968426709660083,
      "grad_norm": 2.658236503601074,
      "learning_rate": 2.550047664442326e-05,
      "loss": 4.2491,
      "step": 18490
    },
    {
      "epoch": 0.4969770253929867,
      "grad_norm": 2.3914830684661865,
      "learning_rate": 2.5493667438376683e-05,
      "loss": 4.1663,
      "step": 18495
    },
    {
      "epoch": 0.49711137981996506,
      "grad_norm": 2.273757219314575,
      "learning_rate": 2.5486858232330112e-05,
      "loss": 4.2471,
      "step": 18500
    },
    {
      "epoch": 0.49724573424694346,
      "grad_norm": 2.586331844329834,
      "learning_rate": 2.5480049026283538e-05,
      "loss": 4.0336,
      "step": 18505
    },
    {
      "epoch": 0.4973800886739218,
      "grad_norm": 2.536170244216919,
      "learning_rate": 2.547323982023696e-05,
      "loss": 4.2369,
      "step": 18510
    },
    {
      "epoch": 0.49751444310090015,
      "grad_norm": 2.3158907890319824,
      "learning_rate": 2.5466430614190383e-05,
      "loss": 4.2748,
      "step": 18515
    },
    {
      "epoch": 0.49764879752787855,
      "grad_norm": 2.5497050285339355,
      "learning_rate": 2.545962140814381e-05,
      "loss": 4.0997,
      "step": 18520
    },
    {
      "epoch": 0.4977831519548569,
      "grad_norm": 2.5585503578186035,
      "learning_rate": 2.5452812202097237e-05,
      "loss": 4.2408,
      "step": 18525
    },
    {
      "epoch": 0.4979175063818353,
      "grad_norm": 2.3359177112579346,
      "learning_rate": 2.544600299605066e-05,
      "loss": 4.1353,
      "step": 18530
    },
    {
      "epoch": 0.49805186080881364,
      "grad_norm": 2.433411121368408,
      "learning_rate": 2.5439193790004085e-05,
      "loss": 4.2434,
      "step": 18535
    },
    {
      "epoch": 0.49818621523579204,
      "grad_norm": 2.677987575531006,
      "learning_rate": 2.5432384583957514e-05,
      "loss": 4.2788,
      "step": 18540
    },
    {
      "epoch": 0.4983205696627704,
      "grad_norm": 2.7603280544281006,
      "learning_rate": 2.5425575377910937e-05,
      "loss": 4.1544,
      "step": 18545
    },
    {
      "epoch": 0.49845492408974873,
      "grad_norm": 2.78523588180542,
      "learning_rate": 2.5418766171864362e-05,
      "loss": 4.2633,
      "step": 18550
    },
    {
      "epoch": 0.49858927851672713,
      "grad_norm": 2.3146603107452393,
      "learning_rate": 2.5411956965817785e-05,
      "loss": 4.1985,
      "step": 18555
    },
    {
      "epoch": 0.4987236329437055,
      "grad_norm": 2.600654125213623,
      "learning_rate": 2.5405147759771214e-05,
      "loss": 4.2182,
      "step": 18560
    },
    {
      "epoch": 0.4988579873706839,
      "grad_norm": 2.6447434425354004,
      "learning_rate": 2.539833855372464e-05,
      "loss": 4.2255,
      "step": 18565
    },
    {
      "epoch": 0.4989923417976622,
      "grad_norm": 2.43923020362854,
      "learning_rate": 2.539152934767806e-05,
      "loss": 4.2077,
      "step": 18570
    },
    {
      "epoch": 0.4991266962246406,
      "grad_norm": 2.505922794342041,
      "learning_rate": 2.5384720141631484e-05,
      "loss": 4.2201,
      "step": 18575
    },
    {
      "epoch": 0.49926105065161897,
      "grad_norm": 2.757676362991333,
      "learning_rate": 2.5377910935584913e-05,
      "loss": 4.1282,
      "step": 18580
    },
    {
      "epoch": 0.49939540507859737,
      "grad_norm": 2.5847113132476807,
      "learning_rate": 2.537110172953834e-05,
      "loss": 4.2183,
      "step": 18585
    },
    {
      "epoch": 0.4995297595055757,
      "grad_norm": 2.4257121086120605,
      "learning_rate": 2.536429252349176e-05,
      "loss": 4.2264,
      "step": 18590
    },
    {
      "epoch": 0.49966411393255405,
      "grad_norm": 2.2203452587127686,
      "learning_rate": 2.5357483317445186e-05,
      "loss": 4.174,
      "step": 18595
    },
    {
      "epoch": 0.49979846835953246,
      "grad_norm": 2.455432415008545,
      "learning_rate": 2.5350674111398615e-05,
      "loss": 4.2194,
      "step": 18600
    },
    {
      "epoch": 0.4999328227865108,
      "grad_norm": 2.7966620922088623,
      "learning_rate": 2.5343864905352038e-05,
      "loss": 4.3535,
      "step": 18605
    },
    {
      "epoch": 0.5000671772134891,
      "grad_norm": 2.2308266162872314,
      "learning_rate": 2.5337055699305463e-05,
      "loss": 4.1993,
      "step": 18610
    },
    {
      "epoch": 0.5002015316404675,
      "grad_norm": 2.414926052093506,
      "learning_rate": 2.5330246493258886e-05,
      "loss": 4.2803,
      "step": 18615
    },
    {
      "epoch": 0.500335886067446,
      "grad_norm": 2.612109899520874,
      "learning_rate": 2.5323437287212315e-05,
      "loss": 4.1166,
      "step": 18620
    },
    {
      "epoch": 0.5004702404944243,
      "grad_norm": 2.3976144790649414,
      "learning_rate": 2.5316628081165737e-05,
      "loss": 4.1845,
      "step": 18625
    },
    {
      "epoch": 0.5006045949214026,
      "grad_norm": 2.6069955825805664,
      "learning_rate": 2.5309818875119163e-05,
      "loss": 4.3571,
      "step": 18630
    },
    {
      "epoch": 0.500738949348381,
      "grad_norm": 2.329179525375366,
      "learning_rate": 2.5303009669072585e-05,
      "loss": 4.3201,
      "step": 18635
    },
    {
      "epoch": 0.5008733037753594,
      "grad_norm": 2.615586042404175,
      "learning_rate": 2.529620046302601e-05,
      "loss": 4.2336,
      "step": 18640
    },
    {
      "epoch": 0.5010076582023377,
      "grad_norm": 2.347893714904785,
      "learning_rate": 2.528939125697944e-05,
      "loss": 4.0614,
      "step": 18645
    },
    {
      "epoch": 0.5011420126293161,
      "grad_norm": 2.32383131980896,
      "learning_rate": 2.5282582050932862e-05,
      "loss": 4.1241,
      "step": 18650
    },
    {
      "epoch": 0.5012763670562945,
      "grad_norm": 2.35555362701416,
      "learning_rate": 2.5275772844886287e-05,
      "loss": 4.3148,
      "step": 18655
    },
    {
      "epoch": 0.5014107214832729,
      "grad_norm": 2.240527391433716,
      "learning_rate": 2.526896363883971e-05,
      "loss": 4.1477,
      "step": 18660
    },
    {
      "epoch": 0.5015450759102512,
      "grad_norm": 2.566396951675415,
      "learning_rate": 2.526215443279314e-05,
      "loss": 4.2485,
      "step": 18665
    },
    {
      "epoch": 0.5016794303372296,
      "grad_norm": 2.304514169692993,
      "learning_rate": 2.5255345226746564e-05,
      "loss": 4.3448,
      "step": 18670
    },
    {
      "epoch": 0.501813784764208,
      "grad_norm": 2.4402577877044678,
      "learning_rate": 2.5248536020699987e-05,
      "loss": 4.1485,
      "step": 18675
    },
    {
      "epoch": 0.5019481391911863,
      "grad_norm": 2.29703688621521,
      "learning_rate": 2.524172681465341e-05,
      "loss": 4.2909,
      "step": 18680
    },
    {
      "epoch": 0.5020824936181647,
      "grad_norm": 2.25199294090271,
      "learning_rate": 2.5234917608606838e-05,
      "loss": 4.1789,
      "step": 18685
    },
    {
      "epoch": 0.5022168480451431,
      "grad_norm": 2.413407802581787,
      "learning_rate": 2.5228108402560264e-05,
      "loss": 4.2036,
      "step": 18690
    },
    {
      "epoch": 0.5023512024721215,
      "grad_norm": 2.395397901535034,
      "learning_rate": 2.5221299196513686e-05,
      "loss": 4.169,
      "step": 18695
    },
    {
      "epoch": 0.5024855568990998,
      "grad_norm": 2.479898691177368,
      "learning_rate": 2.521448999046711e-05,
      "loss": 4.175,
      "step": 18700
    },
    {
      "epoch": 0.5026199113260782,
      "grad_norm": 2.5617711544036865,
      "learning_rate": 2.520768078442054e-05,
      "loss": 4.2006,
      "step": 18705
    },
    {
      "epoch": 0.5027542657530566,
      "grad_norm": 2.299938440322876,
      "learning_rate": 2.5200871578373963e-05,
      "loss": 4.2693,
      "step": 18710
    },
    {
      "epoch": 0.5028886201800349,
      "grad_norm": 2.3865394592285156,
      "learning_rate": 2.519406237232739e-05,
      "loss": 4.026,
      "step": 18715
    },
    {
      "epoch": 0.5030229746070133,
      "grad_norm": 2.438220262527466,
      "learning_rate": 2.518725316628081e-05,
      "loss": 4.3007,
      "step": 18720
    },
    {
      "epoch": 0.5031573290339917,
      "grad_norm": 2.450645685195923,
      "learning_rate": 2.518044396023424e-05,
      "loss": 4.2856,
      "step": 18725
    },
    {
      "epoch": 0.5032916834609701,
      "grad_norm": 2.4057185649871826,
      "learning_rate": 2.5173634754187662e-05,
      "loss": 4.2396,
      "step": 18730
    },
    {
      "epoch": 0.5034260378879484,
      "grad_norm": 2.6239986419677734,
      "learning_rate": 2.5166825548141088e-05,
      "loss": 4.1156,
      "step": 18735
    },
    {
      "epoch": 0.5035603923149268,
      "grad_norm": 2.380948781967163,
      "learning_rate": 2.516001634209451e-05,
      "loss": 4.2586,
      "step": 18740
    },
    {
      "epoch": 0.5036947467419052,
      "grad_norm": 2.5175178050994873,
      "learning_rate": 2.515320713604794e-05,
      "loss": 4.2468,
      "step": 18745
    },
    {
      "epoch": 0.5038291011688835,
      "grad_norm": 2.7464826107025146,
      "learning_rate": 2.5146397930001365e-05,
      "loss": 4.1266,
      "step": 18750
    },
    {
      "epoch": 0.5039634555958619,
      "grad_norm": 2.3505687713623047,
      "learning_rate": 2.5139588723954787e-05,
      "loss": 4.2056,
      "step": 18755
    },
    {
      "epoch": 0.5040978100228403,
      "grad_norm": 2.288691759109497,
      "learning_rate": 2.5132779517908213e-05,
      "loss": 4.2492,
      "step": 18760
    },
    {
      "epoch": 0.5042321644498187,
      "grad_norm": 2.4170656204223633,
      "learning_rate": 2.5125970311861642e-05,
      "loss": 4.2391,
      "step": 18765
    },
    {
      "epoch": 0.504366518876797,
      "grad_norm": 2.63655686378479,
      "learning_rate": 2.5119161105815064e-05,
      "loss": 4.2176,
      "step": 18770
    },
    {
      "epoch": 0.5045008733037754,
      "grad_norm": 2.398902177810669,
      "learning_rate": 2.511235189976849e-05,
      "loss": 4.1487,
      "step": 18775
    },
    {
      "epoch": 0.5046352277307538,
      "grad_norm": 2.8057374954223633,
      "learning_rate": 2.5105542693721912e-05,
      "loss": 4.2147,
      "step": 18780
    },
    {
      "epoch": 0.504769582157732,
      "grad_norm": 2.477985143661499,
      "learning_rate": 2.5098733487675334e-05,
      "loss": 4.1204,
      "step": 18785
    },
    {
      "epoch": 0.5049039365847104,
      "grad_norm": 2.428417205810547,
      "learning_rate": 2.5091924281628763e-05,
      "loss": 4.2479,
      "step": 18790
    },
    {
      "epoch": 0.5050382910116888,
      "grad_norm": 2.491898536682129,
      "learning_rate": 2.508511507558219e-05,
      "loss": 4.1498,
      "step": 18795
    },
    {
      "epoch": 0.5051726454386672,
      "grad_norm": 2.533215045928955,
      "learning_rate": 2.507830586953561e-05,
      "loss": 4.0924,
      "step": 18800
    },
    {
      "epoch": 0.5053069998656455,
      "grad_norm": 2.603253126144409,
      "learning_rate": 2.5071496663489037e-05,
      "loss": 4.148,
      "step": 18805
    },
    {
      "epoch": 0.5054413542926239,
      "grad_norm": 2.3862900733947754,
      "learning_rate": 2.5064687457442466e-05,
      "loss": 4.1716,
      "step": 18810
    },
    {
      "epoch": 0.5055757087196023,
      "grad_norm": 2.451064109802246,
      "learning_rate": 2.5057878251395888e-05,
      "loss": 4.2225,
      "step": 18815
    },
    {
      "epoch": 0.5057100631465807,
      "grad_norm": 2.4568440914154053,
      "learning_rate": 2.5051069045349314e-05,
      "loss": 4.1678,
      "step": 18820
    },
    {
      "epoch": 0.505844417573559,
      "grad_norm": 2.6371846199035645,
      "learning_rate": 2.5044259839302736e-05,
      "loss": 4.121,
      "step": 18825
    },
    {
      "epoch": 0.5059787720005374,
      "grad_norm": 2.446119546890259,
      "learning_rate": 2.5037450633256165e-05,
      "loss": 4.2146,
      "step": 18830
    },
    {
      "epoch": 0.5061131264275158,
      "grad_norm": 2.512579917907715,
      "learning_rate": 2.5030641427209587e-05,
      "loss": 4.071,
      "step": 18835
    },
    {
      "epoch": 0.5062474808544941,
      "grad_norm": 2.5288357734680176,
      "learning_rate": 2.5023832221163013e-05,
      "loss": 4.2804,
      "step": 18840
    },
    {
      "epoch": 0.5063818352814725,
      "grad_norm": 2.5771141052246094,
      "learning_rate": 2.5017023015116435e-05,
      "loss": 4.1704,
      "step": 18845
    },
    {
      "epoch": 0.5065161897084509,
      "grad_norm": 2.3051512241363525,
      "learning_rate": 2.5010213809069864e-05,
      "loss": 4.1872,
      "step": 18850
    },
    {
      "epoch": 0.5066505441354293,
      "grad_norm": 2.387204885482788,
      "learning_rate": 2.500340460302329e-05,
      "loss": 4.1235,
      "step": 18855
    },
    {
      "epoch": 0.5067848985624076,
      "grad_norm": 2.467358112335205,
      "learning_rate": 2.4996595396976712e-05,
      "loss": 4.2603,
      "step": 18860
    },
    {
      "epoch": 0.506919252989386,
      "grad_norm": 2.3400490283966064,
      "learning_rate": 2.4989786190930138e-05,
      "loss": 4.0972,
      "step": 18865
    },
    {
      "epoch": 0.5070536074163644,
      "grad_norm": 2.454592227935791,
      "learning_rate": 2.4982976984883564e-05,
      "loss": 4.3898,
      "step": 18870
    },
    {
      "epoch": 0.5071879618433427,
      "grad_norm": 2.404630422592163,
      "learning_rate": 2.497616777883699e-05,
      "loss": 4.205,
      "step": 18875
    },
    {
      "epoch": 0.5073223162703211,
      "grad_norm": 2.2266414165496826,
      "learning_rate": 2.4969358572790415e-05,
      "loss": 4.2474,
      "step": 18880
    },
    {
      "epoch": 0.5074566706972995,
      "grad_norm": 2.4259536266326904,
      "learning_rate": 2.496254936674384e-05,
      "loss": 4.1152,
      "step": 18885
    },
    {
      "epoch": 0.5075910251242779,
      "grad_norm": 2.758829116821289,
      "learning_rate": 2.4955740160697263e-05,
      "loss": 4.2244,
      "step": 18890
    },
    {
      "epoch": 0.5077253795512562,
      "grad_norm": 2.3042891025543213,
      "learning_rate": 2.494893095465069e-05,
      "loss": 4.1847,
      "step": 18895
    },
    {
      "epoch": 0.5078597339782346,
      "grad_norm": 2.343984603881836,
      "learning_rate": 2.4942121748604114e-05,
      "loss": 4.084,
      "step": 18900
    },
    {
      "epoch": 0.507994088405213,
      "grad_norm": 2.493533134460449,
      "learning_rate": 2.493531254255754e-05,
      "loss": 4.1229,
      "step": 18905
    },
    {
      "epoch": 0.5081284428321913,
      "grad_norm": 2.4453001022338867,
      "learning_rate": 2.4928503336510962e-05,
      "loss": 4.2356,
      "step": 18910
    },
    {
      "epoch": 0.5082627972591697,
      "grad_norm": 2.4494309425354004,
      "learning_rate": 2.492169413046439e-05,
      "loss": 4.1724,
      "step": 18915
    },
    {
      "epoch": 0.5083971516861481,
      "grad_norm": 2.5660715103149414,
      "learning_rate": 2.4914884924417813e-05,
      "loss": 4.1469,
      "step": 18920
    },
    {
      "epoch": 0.5085315061131265,
      "grad_norm": 2.352365255355835,
      "learning_rate": 2.490807571837124e-05,
      "loss": 4.1633,
      "step": 18925
    },
    {
      "epoch": 0.5086658605401048,
      "grad_norm": 2.2898879051208496,
      "learning_rate": 2.4901266512324665e-05,
      "loss": 4.2138,
      "step": 18930
    },
    {
      "epoch": 0.5088002149670832,
      "grad_norm": 2.643263101577759,
      "learning_rate": 2.4894457306278087e-05,
      "loss": 4.1248,
      "step": 18935
    },
    {
      "epoch": 0.5089345693940616,
      "grad_norm": 2.4025118350982666,
      "learning_rate": 2.4887648100231513e-05,
      "loss": 4.2399,
      "step": 18940
    },
    {
      "epoch": 0.5090689238210399,
      "grad_norm": 2.0745902061462402,
      "learning_rate": 2.488083889418494e-05,
      "loss": 4.257,
      "step": 18945
    },
    {
      "epoch": 0.5092032782480183,
      "grad_norm": 2.488585948944092,
      "learning_rate": 2.4874029688138364e-05,
      "loss": 4.2229,
      "step": 18950
    },
    {
      "epoch": 0.5093376326749967,
      "grad_norm": 2.2216603755950928,
      "learning_rate": 2.486722048209179e-05,
      "loss": 4.2164,
      "step": 18955
    },
    {
      "epoch": 0.509471987101975,
      "grad_norm": 2.3368899822235107,
      "learning_rate": 2.4860411276045215e-05,
      "loss": 4.2855,
      "step": 18960
    },
    {
      "epoch": 0.5096063415289533,
      "grad_norm": 2.5822532176971436,
      "learning_rate": 2.4853602069998638e-05,
      "loss": 4.1867,
      "step": 18965
    },
    {
      "epoch": 0.5097406959559317,
      "grad_norm": 2.4659416675567627,
      "learning_rate": 2.4846792863952063e-05,
      "loss": 4.2239,
      "step": 18970
    },
    {
      "epoch": 0.5098750503829101,
      "grad_norm": 2.497880458831787,
      "learning_rate": 2.483998365790549e-05,
      "loss": 4.1616,
      "step": 18975
    },
    {
      "epoch": 0.5100094048098884,
      "grad_norm": 2.593658447265625,
      "learning_rate": 2.4833174451858915e-05,
      "loss": 4.1589,
      "step": 18980
    },
    {
      "epoch": 0.5101437592368668,
      "grad_norm": 2.472313404083252,
      "learning_rate": 2.482636524581234e-05,
      "loss": 4.2971,
      "step": 18985
    },
    {
      "epoch": 0.5102781136638452,
      "grad_norm": 2.56112003326416,
      "learning_rate": 2.4819556039765766e-05,
      "loss": 4.1221,
      "step": 18990
    },
    {
      "epoch": 0.5104124680908236,
      "grad_norm": 2.4737329483032227,
      "learning_rate": 2.4812746833719188e-05,
      "loss": 4.2118,
      "step": 18995
    },
    {
      "epoch": 0.5105468225178019,
      "grad_norm": 2.2520573139190674,
      "learning_rate": 2.4805937627672614e-05,
      "loss": 4.1486,
      "step": 19000
    },
    {
      "epoch": 0.5106811769447803,
      "grad_norm": 2.6526594161987305,
      "learning_rate": 2.479912842162604e-05,
      "loss": 4.1834,
      "step": 19005
    },
    {
      "epoch": 0.5108155313717587,
      "grad_norm": 2.4082844257354736,
      "learning_rate": 2.4792319215579465e-05,
      "loss": 4.2263,
      "step": 19010
    },
    {
      "epoch": 0.510949885798737,
      "grad_norm": 2.490719795227051,
      "learning_rate": 2.4785510009532887e-05,
      "loss": 4.1295,
      "step": 19015
    },
    {
      "epoch": 0.5110842402257154,
      "grad_norm": 2.6129188537597656,
      "learning_rate": 2.4778700803486316e-05,
      "loss": 4.1738,
      "step": 19020
    },
    {
      "epoch": 0.5112185946526938,
      "grad_norm": 2.576753854751587,
      "learning_rate": 2.477189159743974e-05,
      "loss": 4.1368,
      "step": 19025
    },
    {
      "epoch": 0.5113529490796722,
      "grad_norm": 2.5014922618865967,
      "learning_rate": 2.4765082391393164e-05,
      "loss": 4.2416,
      "step": 19030
    },
    {
      "epoch": 0.5114873035066505,
      "grad_norm": 2.293637275695801,
      "learning_rate": 2.475827318534659e-05,
      "loss": 4.2046,
      "step": 19035
    },
    {
      "epoch": 0.5116216579336289,
      "grad_norm": 2.6398375034332275,
      "learning_rate": 2.4751463979300016e-05,
      "loss": 4.1795,
      "step": 19040
    },
    {
      "epoch": 0.5117560123606073,
      "grad_norm": 2.442321538925171,
      "learning_rate": 2.4744654773253438e-05,
      "loss": 4.304,
      "step": 19045
    },
    {
      "epoch": 0.5118903667875857,
      "grad_norm": 2.187950849533081,
      "learning_rate": 2.4737845567206867e-05,
      "loss": 4.2276,
      "step": 19050
    },
    {
      "epoch": 0.512024721214564,
      "grad_norm": 2.5990514755249023,
      "learning_rate": 2.473103636116029e-05,
      "loss": 4.3029,
      "step": 19055
    },
    {
      "epoch": 0.5121590756415424,
      "grad_norm": 2.5766282081604004,
      "learning_rate": 2.4724227155113715e-05,
      "loss": 4.1229,
      "step": 19060
    },
    {
      "epoch": 0.5122934300685208,
      "grad_norm": 2.4270172119140625,
      "learning_rate": 2.471741794906714e-05,
      "loss": 4.154,
      "step": 19065
    },
    {
      "epoch": 0.5124277844954991,
      "grad_norm": 2.4832682609558105,
      "learning_rate": 2.4710608743020566e-05,
      "loss": 4.2012,
      "step": 19070
    },
    {
      "epoch": 0.5125621389224775,
      "grad_norm": 2.432749032974243,
      "learning_rate": 2.470379953697399e-05,
      "loss": 4.2145,
      "step": 19075
    },
    {
      "epoch": 0.5126964933494559,
      "grad_norm": 2.4161741733551025,
      "learning_rate": 2.4696990330927418e-05,
      "loss": 4.2185,
      "step": 19080
    },
    {
      "epoch": 0.5128308477764343,
      "grad_norm": 2.3733983039855957,
      "learning_rate": 2.469018112488084e-05,
      "loss": 4.1963,
      "step": 19085
    },
    {
      "epoch": 0.5129652022034126,
      "grad_norm": 2.3789846897125244,
      "learning_rate": 2.4683371918834265e-05,
      "loss": 4.1089,
      "step": 19090
    },
    {
      "epoch": 0.513099556630391,
      "grad_norm": 2.6909961700439453,
      "learning_rate": 2.467656271278769e-05,
      "loss": 4.2658,
      "step": 19095
    },
    {
      "epoch": 0.5132339110573694,
      "grad_norm": 2.537989854812622,
      "learning_rate": 2.4669753506741113e-05,
      "loss": 4.1815,
      "step": 19100
    },
    {
      "epoch": 0.5133682654843477,
      "grad_norm": 2.4563934803009033,
      "learning_rate": 2.466294430069454e-05,
      "loss": 4.1817,
      "step": 19105
    },
    {
      "epoch": 0.5135026199113261,
      "grad_norm": 2.511692762374878,
      "learning_rate": 2.4656135094647965e-05,
      "loss": 4.2143,
      "step": 19110
    },
    {
      "epoch": 0.5136369743383045,
      "grad_norm": 2.3812544345855713,
      "learning_rate": 2.464932588860139e-05,
      "loss": 4.1468,
      "step": 19115
    },
    {
      "epoch": 0.5137713287652829,
      "grad_norm": 2.327343702316284,
      "learning_rate": 2.4642516682554813e-05,
      "loss": 4.2647,
      "step": 19120
    },
    {
      "epoch": 0.5139056831922612,
      "grad_norm": 2.407703399658203,
      "learning_rate": 2.463570747650824e-05,
      "loss": 4.0866,
      "step": 19125
    },
    {
      "epoch": 0.5140400376192396,
      "grad_norm": 2.5500471591949463,
      "learning_rate": 2.4628898270461664e-05,
      "loss": 4.1409,
      "step": 19130
    },
    {
      "epoch": 0.514174392046218,
      "grad_norm": 2.407898187637329,
      "learning_rate": 2.462208906441509e-05,
      "loss": 4.2394,
      "step": 19135
    },
    {
      "epoch": 0.5143087464731962,
      "grad_norm": 2.4540274143218994,
      "learning_rate": 2.4615279858368515e-05,
      "loss": 4.2458,
      "step": 19140
    },
    {
      "epoch": 0.5144431009001746,
      "grad_norm": 2.365863800048828,
      "learning_rate": 2.460847065232194e-05,
      "loss": 4.1183,
      "step": 19145
    },
    {
      "epoch": 0.514577455327153,
      "grad_norm": 2.2912471294403076,
      "learning_rate": 2.4601661446275363e-05,
      "loss": 4.2602,
      "step": 19150
    },
    {
      "epoch": 0.5147118097541314,
      "grad_norm": 2.261556386947632,
      "learning_rate": 2.4594852240228792e-05,
      "loss": 4.226,
      "step": 19155
    },
    {
      "epoch": 0.5148461641811097,
      "grad_norm": 2.3950469493865967,
      "learning_rate": 2.4588043034182215e-05,
      "loss": 4.2662,
      "step": 19160
    },
    {
      "epoch": 0.5149805186080881,
      "grad_norm": 2.227627992630005,
      "learning_rate": 2.458123382813564e-05,
      "loss": 4.0405,
      "step": 19165
    },
    {
      "epoch": 0.5151148730350665,
      "grad_norm": 2.258918523788452,
      "learning_rate": 2.4574424622089066e-05,
      "loss": 4.1317,
      "step": 19170
    },
    {
      "epoch": 0.5152492274620448,
      "grad_norm": 2.4608702659606934,
      "learning_rate": 2.456761541604249e-05,
      "loss": 4.128,
      "step": 19175
    },
    {
      "epoch": 0.5153835818890232,
      "grad_norm": 2.6952149868011475,
      "learning_rate": 2.4560806209995914e-05,
      "loss": 4.1355,
      "step": 19180
    },
    {
      "epoch": 0.5155179363160016,
      "grad_norm": 2.3763391971588135,
      "learning_rate": 2.4553997003949343e-05,
      "loss": 4.1548,
      "step": 19185
    },
    {
      "epoch": 0.51565229074298,
      "grad_norm": 2.4071786403656006,
      "learning_rate": 2.4547187797902765e-05,
      "loss": 4.0729,
      "step": 19190
    },
    {
      "epoch": 0.5157866451699583,
      "grad_norm": 2.690171003341675,
      "learning_rate": 2.454037859185619e-05,
      "loss": 4.1408,
      "step": 19195
    },
    {
      "epoch": 0.5159209995969367,
      "grad_norm": 2.3048369884490967,
      "learning_rate": 2.4533569385809616e-05,
      "loss": 4.1109,
      "step": 19200
    },
    {
      "epoch": 0.5160553540239151,
      "grad_norm": 2.2279212474823,
      "learning_rate": 2.4526760179763042e-05,
      "loss": 4.2529,
      "step": 19205
    },
    {
      "epoch": 0.5161897084508934,
      "grad_norm": 2.5764353275299072,
      "learning_rate": 2.4519950973716464e-05,
      "loss": 4.2532,
      "step": 19210
    },
    {
      "epoch": 0.5163240628778718,
      "grad_norm": 2.549027442932129,
      "learning_rate": 2.4513141767669893e-05,
      "loss": 4.0419,
      "step": 19215
    },
    {
      "epoch": 0.5164584173048502,
      "grad_norm": 2.3325693607330322,
      "learning_rate": 2.4506332561623316e-05,
      "loss": 4.2411,
      "step": 19220
    },
    {
      "epoch": 0.5165927717318286,
      "grad_norm": 2.813455820083618,
      "learning_rate": 2.449952335557674e-05,
      "loss": 4.1887,
      "step": 19225
    },
    {
      "epoch": 0.5167271261588069,
      "grad_norm": 2.2589588165283203,
      "learning_rate": 2.4492714149530167e-05,
      "loss": 4.1948,
      "step": 19230
    },
    {
      "epoch": 0.5168614805857853,
      "grad_norm": 2.4710195064544678,
      "learning_rate": 2.448590494348359e-05,
      "loss": 4.1208,
      "step": 19235
    },
    {
      "epoch": 0.5169958350127637,
      "grad_norm": 2.4131791591644287,
      "learning_rate": 2.4479095737437015e-05,
      "loss": 3.9861,
      "step": 19240
    },
    {
      "epoch": 0.5171301894397421,
      "grad_norm": 2.5135104656219482,
      "learning_rate": 2.447228653139044e-05,
      "loss": 4.1288,
      "step": 19245
    },
    {
      "epoch": 0.5172645438667204,
      "grad_norm": 2.712293863296509,
      "learning_rate": 2.4465477325343866e-05,
      "loss": 4.2116,
      "step": 19250
    },
    {
      "epoch": 0.5173988982936988,
      "grad_norm": 2.577848434448242,
      "learning_rate": 2.445866811929729e-05,
      "loss": 4.2711,
      "step": 19255
    },
    {
      "epoch": 0.5175332527206772,
      "grad_norm": 2.5128767490386963,
      "learning_rate": 2.4451858913250717e-05,
      "loss": 4.2365,
      "step": 19260
    },
    {
      "epoch": 0.5176676071476555,
      "grad_norm": 2.6061699390411377,
      "learning_rate": 2.444504970720414e-05,
      "loss": 4.1478,
      "step": 19265
    },
    {
      "epoch": 0.5178019615746339,
      "grad_norm": 2.9753358364105225,
      "learning_rate": 2.4438240501157565e-05,
      "loss": 4.1677,
      "step": 19270
    },
    {
      "epoch": 0.5179363160016123,
      "grad_norm": 2.3526861667633057,
      "learning_rate": 2.443143129511099e-05,
      "loss": 4.1839,
      "step": 19275
    },
    {
      "epoch": 0.5180706704285907,
      "grad_norm": 2.3954553604125977,
      "learning_rate": 2.4424622089064417e-05,
      "loss": 4.2778,
      "step": 19280
    },
    {
      "epoch": 0.518205024855569,
      "grad_norm": 2.2088117599487305,
      "learning_rate": 2.441781288301784e-05,
      "loss": 4.1926,
      "step": 19285
    },
    {
      "epoch": 0.5183393792825474,
      "grad_norm": 2.542858600616455,
      "learning_rate": 2.4411003676971268e-05,
      "loss": 4.2399,
      "step": 19290
    },
    {
      "epoch": 0.5184737337095258,
      "grad_norm": 2.441488265991211,
      "learning_rate": 2.440419447092469e-05,
      "loss": 4.2247,
      "step": 19295
    },
    {
      "epoch": 0.518608088136504,
      "grad_norm": 2.5491998195648193,
      "learning_rate": 2.4397385264878116e-05,
      "loss": 4.3001,
      "step": 19300
    },
    {
      "epoch": 0.5187424425634825,
      "grad_norm": 2.3421573638916016,
      "learning_rate": 2.439057605883154e-05,
      "loss": 4.0951,
      "step": 19305
    },
    {
      "epoch": 0.5188767969904609,
      "grad_norm": 2.2438547611236572,
      "learning_rate": 2.4383766852784967e-05,
      "loss": 4.1919,
      "step": 19310
    },
    {
      "epoch": 0.5190111514174393,
      "grad_norm": 2.3466169834136963,
      "learning_rate": 2.437695764673839e-05,
      "loss": 4.1804,
      "step": 19315
    },
    {
      "epoch": 0.5191455058444175,
      "grad_norm": 2.467222213745117,
      "learning_rate": 2.437014844069182e-05,
      "loss": 4.1472,
      "step": 19320
    },
    {
      "epoch": 0.5192798602713959,
      "grad_norm": 2.591048002243042,
      "learning_rate": 2.436333923464524e-05,
      "loss": 4.0082,
      "step": 19325
    },
    {
      "epoch": 0.5194142146983743,
      "grad_norm": 2.5330843925476074,
      "learning_rate": 2.4356530028598667e-05,
      "loss": 4.1657,
      "step": 19330
    },
    {
      "epoch": 0.5195485691253526,
      "grad_norm": 2.4015862941741943,
      "learning_rate": 2.4349720822552092e-05,
      "loss": 4.2106,
      "step": 19335
    },
    {
      "epoch": 0.519682923552331,
      "grad_norm": 2.364415407180786,
      "learning_rate": 2.4342911616505518e-05,
      "loss": 4.1525,
      "step": 19340
    },
    {
      "epoch": 0.5198172779793094,
      "grad_norm": 2.3948943614959717,
      "learning_rate": 2.433610241045894e-05,
      "loss": 4.254,
      "step": 19345
    },
    {
      "epoch": 0.5199516324062878,
      "grad_norm": 2.387593984603882,
      "learning_rate": 2.432929320441237e-05,
      "loss": 4.2196,
      "step": 19350
    },
    {
      "epoch": 0.5200859868332661,
      "grad_norm": 2.355522632598877,
      "learning_rate": 2.432248399836579e-05,
      "loss": 4.118,
      "step": 19355
    },
    {
      "epoch": 0.5202203412602445,
      "grad_norm": 2.610097646713257,
      "learning_rate": 2.4315674792319217e-05,
      "loss": 4.2102,
      "step": 19360
    },
    {
      "epoch": 0.5203546956872229,
      "grad_norm": 2.479344367980957,
      "learning_rate": 2.4308865586272643e-05,
      "loss": 4.1395,
      "step": 19365
    },
    {
      "epoch": 0.5204890501142012,
      "grad_norm": 2.5003228187561035,
      "learning_rate": 2.430205638022607e-05,
      "loss": 4.1125,
      "step": 19370
    },
    {
      "epoch": 0.5206234045411796,
      "grad_norm": 2.2299089431762695,
      "learning_rate": 2.429524717417949e-05,
      "loss": 4.1119,
      "step": 19375
    },
    {
      "epoch": 0.520757758968158,
      "grad_norm": 2.498774528503418,
      "learning_rate": 2.4288437968132916e-05,
      "loss": 4.155,
      "step": 19380
    },
    {
      "epoch": 0.5208921133951364,
      "grad_norm": 2.692408800125122,
      "learning_rate": 2.4281628762086342e-05,
      "loss": 4.0259,
      "step": 19385
    },
    {
      "epoch": 0.5210264678221147,
      "grad_norm": 2.306915760040283,
      "learning_rate": 2.4274819556039764e-05,
      "loss": 4.1222,
      "step": 19390
    },
    {
      "epoch": 0.5211608222490931,
      "grad_norm": 2.2933619022369385,
      "learning_rate": 2.4268010349993193e-05,
      "loss": 4.1561,
      "step": 19395
    },
    {
      "epoch": 0.5212951766760715,
      "grad_norm": 2.6175570487976074,
      "learning_rate": 2.4261201143946616e-05,
      "loss": 4.152,
      "step": 19400
    },
    {
      "epoch": 0.5214295311030498,
      "grad_norm": 2.5819554328918457,
      "learning_rate": 2.425439193790004e-05,
      "loss": 4.2009,
      "step": 19405
    },
    {
      "epoch": 0.5215638855300282,
      "grad_norm": 2.2931675910949707,
      "learning_rate": 2.4247582731853467e-05,
      "loss": 4.1642,
      "step": 19410
    },
    {
      "epoch": 0.5216982399570066,
      "grad_norm": 2.3909666538238525,
      "learning_rate": 2.4240773525806893e-05,
      "loss": 4.189,
      "step": 19415
    },
    {
      "epoch": 0.521832594383985,
      "grad_norm": 2.3716351985931396,
      "learning_rate": 2.4233964319760315e-05,
      "loss": 4.2543,
      "step": 19420
    },
    {
      "epoch": 0.5219669488109633,
      "grad_norm": 2.3029158115386963,
      "learning_rate": 2.4227155113713744e-05,
      "loss": 4.2417,
      "step": 19425
    },
    {
      "epoch": 0.5221013032379417,
      "grad_norm": 2.3708932399749756,
      "learning_rate": 2.4220345907667166e-05,
      "loss": 4.2177,
      "step": 19430
    },
    {
      "epoch": 0.5222356576649201,
      "grad_norm": 2.4914543628692627,
      "learning_rate": 2.4213536701620592e-05,
      "loss": 4.0838,
      "step": 19435
    },
    {
      "epoch": 0.5223700120918985,
      "grad_norm": 2.5556857585906982,
      "learning_rate": 2.4206727495574017e-05,
      "loss": 4.1052,
      "step": 19440
    },
    {
      "epoch": 0.5225043665188768,
      "grad_norm": 2.434607744216919,
      "learning_rate": 2.4199918289527443e-05,
      "loss": 4.1205,
      "step": 19445
    },
    {
      "epoch": 0.5226387209458552,
      "grad_norm": 2.6894402503967285,
      "learning_rate": 2.4193109083480865e-05,
      "loss": 4.1588,
      "step": 19450
    },
    {
      "epoch": 0.5227730753728336,
      "grad_norm": 2.4294652938842773,
      "learning_rate": 2.4186299877434294e-05,
      "loss": 4.2142,
      "step": 19455
    },
    {
      "epoch": 0.5229074297998119,
      "grad_norm": 2.4246666431427,
      "learning_rate": 2.4179490671387717e-05,
      "loss": 4.223,
      "step": 19460
    },
    {
      "epoch": 0.5230417842267903,
      "grad_norm": 2.605745792388916,
      "learning_rate": 2.4172681465341142e-05,
      "loss": 4.1628,
      "step": 19465
    },
    {
      "epoch": 0.5231761386537687,
      "grad_norm": 2.401439905166626,
      "learning_rate": 2.4165872259294568e-05,
      "loss": 4.1364,
      "step": 19470
    },
    {
      "epoch": 0.5233104930807471,
      "grad_norm": 2.5756683349609375,
      "learning_rate": 2.4159063053247994e-05,
      "loss": 4.1874,
      "step": 19475
    },
    {
      "epoch": 0.5234448475077254,
      "grad_norm": 2.373011589050293,
      "learning_rate": 2.4152253847201416e-05,
      "loss": 4.1208,
      "step": 19480
    },
    {
      "epoch": 0.5235792019347038,
      "grad_norm": 2.520259380340576,
      "learning_rate": 2.4145444641154845e-05,
      "loss": 4.1642,
      "step": 19485
    },
    {
      "epoch": 0.5237135563616822,
      "grad_norm": 2.32365345954895,
      "learning_rate": 2.4138635435108267e-05,
      "loss": 4.1664,
      "step": 19490
    },
    {
      "epoch": 0.5238479107886604,
      "grad_norm": 2.2045445442199707,
      "learning_rate": 2.4131826229061693e-05,
      "loss": 4.0915,
      "step": 19495
    },
    {
      "epoch": 0.5239822652156388,
      "grad_norm": 2.259805679321289,
      "learning_rate": 2.412501702301512e-05,
      "loss": 4.1102,
      "step": 19500
    },
    {
      "epoch": 0.5241166196426172,
      "grad_norm": 2.488907814025879,
      "learning_rate": 2.4118207816968544e-05,
      "loss": 4.0722,
      "step": 19505
    },
    {
      "epoch": 0.5242509740695956,
      "grad_norm": 2.419842481613159,
      "learning_rate": 2.4111398610921966e-05,
      "loss": 4.1472,
      "step": 19510
    },
    {
      "epoch": 0.5243853284965739,
      "grad_norm": 2.644954204559326,
      "learning_rate": 2.4104589404875396e-05,
      "loss": 4.1447,
      "step": 19515
    },
    {
      "epoch": 0.5245196829235523,
      "grad_norm": 2.2284414768218994,
      "learning_rate": 2.4097780198828818e-05,
      "loss": 4.1575,
      "step": 19520
    },
    {
      "epoch": 0.5246540373505307,
      "grad_norm": 2.508227825164795,
      "learning_rate": 2.4090970992782243e-05,
      "loss": 4.1627,
      "step": 19525
    },
    {
      "epoch": 0.524788391777509,
      "grad_norm": 2.3828859329223633,
      "learning_rate": 2.408416178673567e-05,
      "loss": 4.2278,
      "step": 19530
    },
    {
      "epoch": 0.5249227462044874,
      "grad_norm": 2.3429534435272217,
      "learning_rate": 2.407735258068909e-05,
      "loss": 4.1168,
      "step": 19535
    },
    {
      "epoch": 0.5250571006314658,
      "grad_norm": 2.59507417678833,
      "learning_rate": 2.4070543374642517e-05,
      "loss": 4.2073,
      "step": 19540
    },
    {
      "epoch": 0.5251914550584442,
      "grad_norm": 2.602620840072632,
      "learning_rate": 2.4063734168595943e-05,
      "loss": 4.2225,
      "step": 19545
    },
    {
      "epoch": 0.5253258094854225,
      "grad_norm": 2.5905463695526123,
      "learning_rate": 2.405692496254937e-05,
      "loss": 4.2048,
      "step": 19550
    },
    {
      "epoch": 0.5254601639124009,
      "grad_norm": 2.5372214317321777,
      "learning_rate": 2.405011575650279e-05,
      "loss": 4.1774,
      "step": 19555
    },
    {
      "epoch": 0.5255945183393793,
      "grad_norm": 2.317359447479248,
      "learning_rate": 2.404330655045622e-05,
      "loss": 4.1496,
      "step": 19560
    },
    {
      "epoch": 0.5257288727663576,
      "grad_norm": 2.5819942951202393,
      "learning_rate": 2.4036497344409642e-05,
      "loss": 4.0662,
      "step": 19565
    },
    {
      "epoch": 0.525863227193336,
      "grad_norm": 2.517676830291748,
      "learning_rate": 2.4029688138363068e-05,
      "loss": 4.1158,
      "step": 19570
    },
    {
      "epoch": 0.5259975816203144,
      "grad_norm": 2.4072768688201904,
      "learning_rate": 2.4022878932316493e-05,
      "loss": 4.1718,
      "step": 19575
    },
    {
      "epoch": 0.5261319360472928,
      "grad_norm": 2.226813793182373,
      "learning_rate": 2.401606972626992e-05,
      "loss": 4.2765,
      "step": 19580
    },
    {
      "epoch": 0.5262662904742711,
      "grad_norm": 2.7242965698242188,
      "learning_rate": 2.400926052022334e-05,
      "loss": 4.187,
      "step": 19585
    },
    {
      "epoch": 0.5264006449012495,
      "grad_norm": 2.587934970855713,
      "learning_rate": 2.400245131417677e-05,
      "loss": 4.1164,
      "step": 19590
    },
    {
      "epoch": 0.5265349993282279,
      "grad_norm": 2.553853750228882,
      "learning_rate": 2.3995642108130192e-05,
      "loss": 4.1806,
      "step": 19595
    },
    {
      "epoch": 0.5266693537552062,
      "grad_norm": 2.2665791511535645,
      "learning_rate": 2.3988832902083618e-05,
      "loss": 4.1183,
      "step": 19600
    },
    {
      "epoch": 0.5268037081821846,
      "grad_norm": 2.559373617172241,
      "learning_rate": 2.3982023696037044e-05,
      "loss": 4.2135,
      "step": 19605
    },
    {
      "epoch": 0.526938062609163,
      "grad_norm": 2.413282632827759,
      "learning_rate": 2.397521448999047e-05,
      "loss": 4.1154,
      "step": 19610
    },
    {
      "epoch": 0.5270724170361414,
      "grad_norm": 2.336554765701294,
      "learning_rate": 2.3968405283943892e-05,
      "loss": 4.209,
      "step": 19615
    },
    {
      "epoch": 0.5272067714631197,
      "grad_norm": 2.4473876953125,
      "learning_rate": 2.396159607789732e-05,
      "loss": 4.2153,
      "step": 19620
    },
    {
      "epoch": 0.5273411258900981,
      "grad_norm": 2.460981607437134,
      "learning_rate": 2.3954786871850743e-05,
      "loss": 4.1904,
      "step": 19625
    },
    {
      "epoch": 0.5274754803170765,
      "grad_norm": 2.3463668823242188,
      "learning_rate": 2.394797766580417e-05,
      "loss": 4.2273,
      "step": 19630
    },
    {
      "epoch": 0.5276098347440548,
      "grad_norm": 2.411773681640625,
      "learning_rate": 2.3941168459757594e-05,
      "loss": 4.0936,
      "step": 19635
    },
    {
      "epoch": 0.5277441891710332,
      "grad_norm": 2.409456253051758,
      "learning_rate": 2.393435925371102e-05,
      "loss": 4.0884,
      "step": 19640
    },
    {
      "epoch": 0.5278785435980116,
      "grad_norm": 2.3401482105255127,
      "learning_rate": 2.3927550047664442e-05,
      "loss": 4.0523,
      "step": 19645
    },
    {
      "epoch": 0.52801289802499,
      "grad_norm": 2.3576653003692627,
      "learning_rate": 2.392074084161787e-05,
      "loss": 4.1577,
      "step": 19650
    },
    {
      "epoch": 0.5281472524519683,
      "grad_norm": 2.2831578254699707,
      "learning_rate": 2.3913931635571294e-05,
      "loss": 4.0053,
      "step": 19655
    },
    {
      "epoch": 0.5282816068789467,
      "grad_norm": 2.438647985458374,
      "learning_rate": 2.390712242952472e-05,
      "loss": 4.1787,
      "step": 19660
    },
    {
      "epoch": 0.528415961305925,
      "grad_norm": 2.5329291820526123,
      "learning_rate": 2.3900313223478145e-05,
      "loss": 4.1798,
      "step": 19665
    },
    {
      "epoch": 0.5285503157329035,
      "grad_norm": 2.2505908012390137,
      "learning_rate": 2.389350401743157e-05,
      "loss": 4.1231,
      "step": 19670
    },
    {
      "epoch": 0.5286846701598817,
      "grad_norm": 2.3395848274230957,
      "learning_rate": 2.3886694811384993e-05,
      "loss": 4.0006,
      "step": 19675
    },
    {
      "epoch": 0.5288190245868601,
      "grad_norm": 2.4946913719177246,
      "learning_rate": 2.387988560533842e-05,
      "loss": 4.0995,
      "step": 19680
    },
    {
      "epoch": 0.5289533790138385,
      "grad_norm": 2.4786462783813477,
      "learning_rate": 2.3873076399291844e-05,
      "loss": 4.0761,
      "step": 19685
    },
    {
      "epoch": 0.5290877334408168,
      "grad_norm": 2.446791410446167,
      "learning_rate": 2.3866267193245266e-05,
      "loss": 4.2139,
      "step": 19690
    },
    {
      "epoch": 0.5292220878677952,
      "grad_norm": 2.624887228012085,
      "learning_rate": 2.3859457987198695e-05,
      "loss": 4.1317,
      "step": 19695
    },
    {
      "epoch": 0.5293564422947736,
      "grad_norm": 2.192580461502075,
      "learning_rate": 2.3852648781152118e-05,
      "loss": 4.2087,
      "step": 19700
    },
    {
      "epoch": 0.529490796721752,
      "grad_norm": 2.3577182292938232,
      "learning_rate": 2.3845839575105543e-05,
      "loss": 4.187,
      "step": 19705
    },
    {
      "epoch": 0.5296251511487303,
      "grad_norm": 2.6276166439056396,
      "learning_rate": 2.383903036905897e-05,
      "loss": 4.3099,
      "step": 19710
    },
    {
      "epoch": 0.5297595055757087,
      "grad_norm": 2.3528847694396973,
      "learning_rate": 2.3832221163012395e-05,
      "loss": 4.3097,
      "step": 19715
    },
    {
      "epoch": 0.5298938600026871,
      "grad_norm": 2.4681408405303955,
      "learning_rate": 2.3825411956965817e-05,
      "loss": 4.2341,
      "step": 19720
    },
    {
      "epoch": 0.5300282144296654,
      "grad_norm": 2.308509111404419,
      "learning_rate": 2.3818602750919246e-05,
      "loss": 4.2227,
      "step": 19725
    },
    {
      "epoch": 0.5301625688566438,
      "grad_norm": 2.575002670288086,
      "learning_rate": 2.3811793544872668e-05,
      "loss": 4.0359,
      "step": 19730
    },
    {
      "epoch": 0.5302969232836222,
      "grad_norm": 2.4957404136657715,
      "learning_rate": 2.3804984338826094e-05,
      "loss": 4.1513,
      "step": 19735
    },
    {
      "epoch": 0.5304312777106006,
      "grad_norm": 2.232130527496338,
      "learning_rate": 2.379817513277952e-05,
      "loss": 4.1318,
      "step": 19740
    },
    {
      "epoch": 0.5305656321375789,
      "grad_norm": 2.394183397293091,
      "learning_rate": 2.3791365926732945e-05,
      "loss": 4.1482,
      "step": 19745
    },
    {
      "epoch": 0.5306999865645573,
      "grad_norm": 2.6029720306396484,
      "learning_rate": 2.3784556720686368e-05,
      "loss": 4.2619,
      "step": 19750
    },
    {
      "epoch": 0.5308343409915357,
      "grad_norm": 2.2995316982269287,
      "learning_rate": 2.3777747514639793e-05,
      "loss": 4.2579,
      "step": 19755
    },
    {
      "epoch": 0.530968695418514,
      "grad_norm": 2.455338954925537,
      "learning_rate": 2.377093830859322e-05,
      "loss": 4.1183,
      "step": 19760
    },
    {
      "epoch": 0.5311030498454924,
      "grad_norm": 2.4550743103027344,
      "learning_rate": 2.3764129102546645e-05,
      "loss": 4.1018,
      "step": 19765
    },
    {
      "epoch": 0.5312374042724708,
      "grad_norm": 2.2408323287963867,
      "learning_rate": 2.375731989650007e-05,
      "loss": 4.1946,
      "step": 19770
    },
    {
      "epoch": 0.5313717586994492,
      "grad_norm": 2.5385196208953857,
      "learning_rate": 2.3750510690453496e-05,
      "loss": 4.1251,
      "step": 19775
    },
    {
      "epoch": 0.5315061131264275,
      "grad_norm": 2.33528208732605,
      "learning_rate": 2.3743701484406918e-05,
      "loss": 4.157,
      "step": 19780
    },
    {
      "epoch": 0.5316404675534059,
      "grad_norm": 2.416646718978882,
      "learning_rate": 2.3736892278360344e-05,
      "loss": 4.047,
      "step": 19785
    },
    {
      "epoch": 0.5317748219803843,
      "grad_norm": 2.622180223464966,
      "learning_rate": 2.373008307231377e-05,
      "loss": 4.2253,
      "step": 19790
    },
    {
      "epoch": 0.5319091764073626,
      "grad_norm": 2.2783236503601074,
      "learning_rate": 2.3723273866267195e-05,
      "loss": 4.1883,
      "step": 19795
    },
    {
      "epoch": 0.532043530834341,
      "grad_norm": 2.503242254257202,
      "learning_rate": 2.371646466022062e-05,
      "loss": 4.2299,
      "step": 19800
    },
    {
      "epoch": 0.5321778852613194,
      "grad_norm": 2.4925386905670166,
      "learning_rate": 2.3709655454174046e-05,
      "loss": 4.1227,
      "step": 19805
    },
    {
      "epoch": 0.5323122396882978,
      "grad_norm": 2.6601483821868896,
      "learning_rate": 2.370284624812747e-05,
      "loss": 4.1279,
      "step": 19810
    },
    {
      "epoch": 0.5324465941152761,
      "grad_norm": 2.529906749725342,
      "learning_rate": 2.3696037042080894e-05,
      "loss": 4.1269,
      "step": 19815
    },
    {
      "epoch": 0.5325809485422545,
      "grad_norm": 2.488236665725708,
      "learning_rate": 2.368922783603432e-05,
      "loss": 4.0333,
      "step": 19820
    },
    {
      "epoch": 0.5327153029692329,
      "grad_norm": 2.481128454208374,
      "learning_rate": 2.3682418629987742e-05,
      "loss": 4.0917,
      "step": 19825
    },
    {
      "epoch": 0.5328496573962112,
      "grad_norm": 2.5271215438842773,
      "learning_rate": 2.367560942394117e-05,
      "loss": 4.0599,
      "step": 19830
    },
    {
      "epoch": 0.5329840118231896,
      "grad_norm": 2.4280197620391846,
      "learning_rate": 2.3668800217894594e-05,
      "loss": 4.1872,
      "step": 19835
    },
    {
      "epoch": 0.533118366250168,
      "grad_norm": 2.4330058097839355,
      "learning_rate": 2.366199101184802e-05,
      "loss": 4.221,
      "step": 19840
    },
    {
      "epoch": 0.5332527206771464,
      "grad_norm": 2.532562017440796,
      "learning_rate": 2.3655181805801445e-05,
      "loss": 4.0234,
      "step": 19845
    },
    {
      "epoch": 0.5333870751041246,
      "grad_norm": 2.502290725708008,
      "learning_rate": 2.364837259975487e-05,
      "loss": 4.1461,
      "step": 19850
    },
    {
      "epoch": 0.533521429531103,
      "grad_norm": 2.227602005004883,
      "learning_rate": 2.3641563393708293e-05,
      "loss": 4.1852,
      "step": 19855
    },
    {
      "epoch": 0.5336557839580814,
      "grad_norm": 2.3136398792266846,
      "learning_rate": 2.363475418766172e-05,
      "loss": 4.1688,
      "step": 19860
    },
    {
      "epoch": 0.5337901383850598,
      "grad_norm": 2.6027657985687256,
      "learning_rate": 2.3627944981615144e-05,
      "loss": 4.0856,
      "step": 19865
    },
    {
      "epoch": 0.5339244928120381,
      "grad_norm": 2.760267972946167,
      "learning_rate": 2.362113577556857e-05,
      "loss": 4.1458,
      "step": 19870
    },
    {
      "epoch": 0.5340588472390165,
      "grad_norm": 2.481855630874634,
      "learning_rate": 2.3614326569521995e-05,
      "loss": 4.2034,
      "step": 19875
    },
    {
      "epoch": 0.5341932016659949,
      "grad_norm": 2.400932550430298,
      "learning_rate": 2.360751736347542e-05,
      "loss": 4.1213,
      "step": 19880
    },
    {
      "epoch": 0.5343275560929732,
      "grad_norm": 2.3457772731781006,
      "learning_rate": 2.3600708157428843e-05,
      "loss": 4.1471,
      "step": 19885
    },
    {
      "epoch": 0.5344619105199516,
      "grad_norm": 2.9542276859283447,
      "learning_rate": 2.359389895138227e-05,
      "loss": 4.1602,
      "step": 19890
    },
    {
      "epoch": 0.53459626494693,
      "grad_norm": 2.4567620754241943,
      "learning_rate": 2.3587089745335695e-05,
      "loss": 4.1741,
      "step": 19895
    },
    {
      "epoch": 0.5347306193739084,
      "grad_norm": 2.6443021297454834,
      "learning_rate": 2.358028053928912e-05,
      "loss": 4.2107,
      "step": 19900
    },
    {
      "epoch": 0.5348649738008867,
      "grad_norm": 2.4837722778320312,
      "learning_rate": 2.3573471333242546e-05,
      "loss": 4.1673,
      "step": 19905
    },
    {
      "epoch": 0.5349993282278651,
      "grad_norm": 2.377662181854248,
      "learning_rate": 2.356666212719597e-05,
      "loss": 4.1024,
      "step": 19910
    },
    {
      "epoch": 0.5351336826548435,
      "grad_norm": 2.6272366046905518,
      "learning_rate": 2.3559852921149394e-05,
      "loss": 4.2115,
      "step": 19915
    },
    {
      "epoch": 0.5352680370818218,
      "grad_norm": 3.010737895965576,
      "learning_rate": 2.355304371510282e-05,
      "loss": 4.2124,
      "step": 19920
    },
    {
      "epoch": 0.5354023915088002,
      "grad_norm": 2.346762180328369,
      "learning_rate": 2.3546234509056245e-05,
      "loss": 4.1673,
      "step": 19925
    },
    {
      "epoch": 0.5355367459357786,
      "grad_norm": 2.6834068298339844,
      "learning_rate": 2.353942530300967e-05,
      "loss": 4.2344,
      "step": 19930
    },
    {
      "epoch": 0.535671100362757,
      "grad_norm": 2.4434335231781006,
      "learning_rate": 2.3532616096963097e-05,
      "loss": 4.2325,
      "step": 19935
    },
    {
      "epoch": 0.5358054547897353,
      "grad_norm": 2.7561428546905518,
      "learning_rate": 2.3525806890916522e-05,
      "loss": 4.205,
      "step": 19940
    },
    {
      "epoch": 0.5359398092167137,
      "grad_norm": 2.382530927658081,
      "learning_rate": 2.3518997684869944e-05,
      "loss": 4.2802,
      "step": 19945
    },
    {
      "epoch": 0.5360741636436921,
      "grad_norm": 2.6401822566986084,
      "learning_rate": 2.351218847882337e-05,
      "loss": 4.2178,
      "step": 19950
    },
    {
      "epoch": 0.5362085180706704,
      "grad_norm": 2.2574403285980225,
      "learning_rate": 2.3505379272776796e-05,
      "loss": 4.2785,
      "step": 19955
    },
    {
      "epoch": 0.5363428724976488,
      "grad_norm": 2.3282604217529297,
      "learning_rate": 2.349857006673022e-05,
      "loss": 4.1487,
      "step": 19960
    },
    {
      "epoch": 0.5364772269246272,
      "grad_norm": 2.201552391052246,
      "learning_rate": 2.3491760860683644e-05,
      "loss": 4.2107,
      "step": 19965
    },
    {
      "epoch": 0.5366115813516056,
      "grad_norm": 2.5604617595672607,
      "learning_rate": 2.3484951654637073e-05,
      "loss": 4.0575,
      "step": 19970
    },
    {
      "epoch": 0.5367459357785839,
      "grad_norm": 2.2823944091796875,
      "learning_rate": 2.3478142448590495e-05,
      "loss": 4.1463,
      "step": 19975
    },
    {
      "epoch": 0.5368802902055623,
      "grad_norm": 2.5601329803466797,
      "learning_rate": 2.347133324254392e-05,
      "loss": 4.1585,
      "step": 19980
    },
    {
      "epoch": 0.5370146446325407,
      "grad_norm": 2.380958318710327,
      "learning_rate": 2.3464524036497346e-05,
      "loss": 4.3339,
      "step": 19985
    },
    {
      "epoch": 0.537148999059519,
      "grad_norm": 2.4014508724212646,
      "learning_rate": 2.345771483045077e-05,
      "loss": 4.1118,
      "step": 19990
    },
    {
      "epoch": 0.5372833534864974,
      "grad_norm": 2.4354753494262695,
      "learning_rate": 2.3450905624404194e-05,
      "loss": 4.1837,
      "step": 19995
    },
    {
      "epoch": 0.5374177079134758,
      "grad_norm": 2.4942574501037598,
      "learning_rate": 2.344409641835762e-05,
      "loss": 4.2234,
      "step": 20000
    },
    {
      "epoch": 0.5375520623404542,
      "grad_norm": 2.3543336391448975,
      "learning_rate": 2.3437287212311046e-05,
      "loss": 4.1467,
      "step": 20005
    },
    {
      "epoch": 0.5376864167674325,
      "grad_norm": 2.5599191188812256,
      "learning_rate": 2.343047800626447e-05,
      "loss": 4.1367,
      "step": 20010
    },
    {
      "epoch": 0.5378207711944109,
      "grad_norm": 2.486945390701294,
      "learning_rate": 2.3423668800217897e-05,
      "loss": 4.0903,
      "step": 20015
    },
    {
      "epoch": 0.5379551256213893,
      "grad_norm": 2.5304880142211914,
      "learning_rate": 2.341685959417132e-05,
      "loss": 4.2478,
      "step": 20020
    },
    {
      "epoch": 0.5380894800483675,
      "grad_norm": 2.5263166427612305,
      "learning_rate": 2.3410050388124745e-05,
      "loss": 4.127,
      "step": 20025
    },
    {
      "epoch": 0.5382238344753459,
      "grad_norm": 2.4813787937164307,
      "learning_rate": 2.340324118207817e-05,
      "loss": 4.1907,
      "step": 20030
    },
    {
      "epoch": 0.5383581889023243,
      "grad_norm": 2.518040418624878,
      "learning_rate": 2.3396431976031596e-05,
      "loss": 4.1916,
      "step": 20035
    },
    {
      "epoch": 0.5384925433293027,
      "grad_norm": 2.369868516921997,
      "learning_rate": 2.3389622769985022e-05,
      "loss": 4.2327,
      "step": 20040
    },
    {
      "epoch": 0.538626897756281,
      "grad_norm": 2.5046117305755615,
      "learning_rate": 2.3382813563938447e-05,
      "loss": 4.149,
      "step": 20045
    },
    {
      "epoch": 0.5387612521832594,
      "grad_norm": 2.341040849685669,
      "learning_rate": 2.337600435789187e-05,
      "loss": 4.1431,
      "step": 20050
    },
    {
      "epoch": 0.5388956066102378,
      "grad_norm": 2.3161439895629883,
      "learning_rate": 2.3369195151845295e-05,
      "loss": 4.2159,
      "step": 20055
    },
    {
      "epoch": 0.5390299610372161,
      "grad_norm": 2.426114559173584,
      "learning_rate": 2.336238594579872e-05,
      "loss": 4.0949,
      "step": 20060
    },
    {
      "epoch": 0.5391643154641945,
      "grad_norm": 2.4937424659729004,
      "learning_rate": 2.3355576739752147e-05,
      "loss": 4.1788,
      "step": 20065
    },
    {
      "epoch": 0.5392986698911729,
      "grad_norm": 2.465135335922241,
      "learning_rate": 2.334876753370557e-05,
      "loss": 4.2443,
      "step": 20070
    },
    {
      "epoch": 0.5394330243181513,
      "grad_norm": 2.316635847091675,
      "learning_rate": 2.3341958327658998e-05,
      "loss": 4.0954,
      "step": 20075
    },
    {
      "epoch": 0.5395673787451296,
      "grad_norm": 2.580977201461792,
      "learning_rate": 2.333514912161242e-05,
      "loss": 4.2978,
      "step": 20080
    },
    {
      "epoch": 0.539701733172108,
      "grad_norm": 2.176295757293701,
      "learning_rate": 2.3328339915565846e-05,
      "loss": 4.1919,
      "step": 20085
    },
    {
      "epoch": 0.5398360875990864,
      "grad_norm": 2.401799440383911,
      "learning_rate": 2.332153070951927e-05,
      "loss": 4.2586,
      "step": 20090
    },
    {
      "epoch": 0.5399704420260648,
      "grad_norm": 2.4833359718322754,
      "learning_rate": 2.3314721503472697e-05,
      "loss": 4.16,
      "step": 20095
    },
    {
      "epoch": 0.5401047964530431,
      "grad_norm": 2.337731122970581,
      "learning_rate": 2.330791229742612e-05,
      "loss": 4.0634,
      "step": 20100
    },
    {
      "epoch": 0.5402391508800215,
      "grad_norm": 2.3508505821228027,
      "learning_rate": 2.330110309137955e-05,
      "loss": 4.169,
      "step": 20105
    },
    {
      "epoch": 0.5403735053069999,
      "grad_norm": 2.610184907913208,
      "learning_rate": 2.329429388533297e-05,
      "loss": 4.2276,
      "step": 20110
    },
    {
      "epoch": 0.5405078597339782,
      "grad_norm": 2.68168044090271,
      "learning_rate": 2.3287484679286396e-05,
      "loss": 4.2143,
      "step": 20115
    },
    {
      "epoch": 0.5406422141609566,
      "grad_norm": 2.3111865520477295,
      "learning_rate": 2.3280675473239822e-05,
      "loss": 4.169,
      "step": 20120
    },
    {
      "epoch": 0.540776568587935,
      "grad_norm": 2.2827577590942383,
      "learning_rate": 2.3273866267193244e-05,
      "loss": 4.1083,
      "step": 20125
    },
    {
      "epoch": 0.5409109230149134,
      "grad_norm": 2.402019500732422,
      "learning_rate": 2.326705706114667e-05,
      "loss": 4.2195,
      "step": 20130
    },
    {
      "epoch": 0.5410452774418917,
      "grad_norm": 2.4111995697021484,
      "learning_rate": 2.3260247855100096e-05,
      "loss": 4.2482,
      "step": 20135
    },
    {
      "epoch": 0.5411796318688701,
      "grad_norm": 2.2512338161468506,
      "learning_rate": 2.325343864905352e-05,
      "loss": 3.9686,
      "step": 20140
    },
    {
      "epoch": 0.5413139862958485,
      "grad_norm": 2.5896291732788086,
      "learning_rate": 2.3246629443006947e-05,
      "loss": 4.263,
      "step": 20145
    },
    {
      "epoch": 0.5414483407228268,
      "grad_norm": 2.5307767391204834,
      "learning_rate": 2.3239820236960373e-05,
      "loss": 4.1731,
      "step": 20150
    },
    {
      "epoch": 0.5415826951498052,
      "grad_norm": 2.5398061275482178,
      "learning_rate": 2.3233011030913795e-05,
      "loss": 4.2506,
      "step": 20155
    },
    {
      "epoch": 0.5417170495767836,
      "grad_norm": 2.365388870239258,
      "learning_rate": 2.322620182486722e-05,
      "loss": 4.2236,
      "step": 20160
    },
    {
      "epoch": 0.541851404003762,
      "grad_norm": 2.2030937671661377,
      "learning_rate": 2.3219392618820646e-05,
      "loss": 4.2017,
      "step": 20165
    },
    {
      "epoch": 0.5419857584307403,
      "grad_norm": 2.5959572792053223,
      "learning_rate": 2.3212583412774072e-05,
      "loss": 4.1611,
      "step": 20170
    },
    {
      "epoch": 0.5421201128577187,
      "grad_norm": 2.372133255004883,
      "learning_rate": 2.3205774206727494e-05,
      "loss": 4.0889,
      "step": 20175
    },
    {
      "epoch": 0.5422544672846971,
      "grad_norm": 2.4216184616088867,
      "learning_rate": 2.3198965000680923e-05,
      "loss": 4.1612,
      "step": 20180
    },
    {
      "epoch": 0.5423888217116754,
      "grad_norm": 2.2745449542999268,
      "learning_rate": 2.3192155794634346e-05,
      "loss": 4.1637,
      "step": 20185
    },
    {
      "epoch": 0.5425231761386538,
      "grad_norm": 2.36331844329834,
      "learning_rate": 2.318534658858777e-05,
      "loss": 4.2324,
      "step": 20190
    },
    {
      "epoch": 0.5426575305656322,
      "grad_norm": 2.3993477821350098,
      "learning_rate": 2.3178537382541197e-05,
      "loss": 4.0434,
      "step": 20195
    },
    {
      "epoch": 0.5427918849926106,
      "grad_norm": 2.526160717010498,
      "learning_rate": 2.3171728176494622e-05,
      "loss": 4.2902,
      "step": 20200
    },
    {
      "epoch": 0.5429262394195888,
      "grad_norm": 2.3316924571990967,
      "learning_rate": 2.3164918970448045e-05,
      "loss": 4.2886,
      "step": 20205
    },
    {
      "epoch": 0.5430605938465672,
      "grad_norm": 2.6067090034484863,
      "learning_rate": 2.3158109764401474e-05,
      "loss": 4.016,
      "step": 20210
    },
    {
      "epoch": 0.5431949482735456,
      "grad_norm": 2.7873034477233887,
      "learning_rate": 2.3151300558354896e-05,
      "loss": 4.2331,
      "step": 20215
    },
    {
      "epoch": 0.5433293027005239,
      "grad_norm": 2.609191417694092,
      "learning_rate": 2.3144491352308322e-05,
      "loss": 4.0894,
      "step": 20220
    },
    {
      "epoch": 0.5434636571275023,
      "grad_norm": 2.416948080062866,
      "learning_rate": 2.3137682146261747e-05,
      "loss": 4.2224,
      "step": 20225
    },
    {
      "epoch": 0.5435980115544807,
      "grad_norm": 2.4412498474121094,
      "learning_rate": 2.3130872940215173e-05,
      "loss": 4.2322,
      "step": 20230
    },
    {
      "epoch": 0.5437323659814591,
      "grad_norm": 2.6170599460601807,
      "learning_rate": 2.3124063734168595e-05,
      "loss": 4.2063,
      "step": 20235
    },
    {
      "epoch": 0.5438667204084374,
      "grad_norm": 2.6210646629333496,
      "learning_rate": 2.3117254528122024e-05,
      "loss": 4.2434,
      "step": 20240
    },
    {
      "epoch": 0.5440010748354158,
      "grad_norm": 2.3776369094848633,
      "learning_rate": 2.3110445322075447e-05,
      "loss": 4.0863,
      "step": 20245
    },
    {
      "epoch": 0.5441354292623942,
      "grad_norm": 2.761976957321167,
      "learning_rate": 2.3103636116028872e-05,
      "loss": 4.2419,
      "step": 20250
    },
    {
      "epoch": 0.5442697836893725,
      "grad_norm": 2.256260633468628,
      "learning_rate": 2.3096826909982298e-05,
      "loss": 4.1232,
      "step": 20255
    },
    {
      "epoch": 0.5444041381163509,
      "grad_norm": 2.4522786140441895,
      "learning_rate": 2.3090017703935724e-05,
      "loss": 4.1971,
      "step": 20260
    },
    {
      "epoch": 0.5445384925433293,
      "grad_norm": 2.531339168548584,
      "learning_rate": 2.3083208497889146e-05,
      "loss": 4.357,
      "step": 20265
    },
    {
      "epoch": 0.5446728469703077,
      "grad_norm": 2.354278087615967,
      "learning_rate": 2.307639929184257e-05,
      "loss": 4.1372,
      "step": 20270
    },
    {
      "epoch": 0.544807201397286,
      "grad_norm": 2.4236373901367188,
      "learning_rate": 2.3069590085795997e-05,
      "loss": 4.1985,
      "step": 20275
    },
    {
      "epoch": 0.5449415558242644,
      "grad_norm": 2.2392358779907227,
      "learning_rate": 2.306278087974942e-05,
      "loss": 4.1053,
      "step": 20280
    },
    {
      "epoch": 0.5450759102512428,
      "grad_norm": 2.3322393894195557,
      "learning_rate": 2.305597167370285e-05,
      "loss": 4.0994,
      "step": 20285
    },
    {
      "epoch": 0.5452102646782212,
      "grad_norm": 2.534773588180542,
      "learning_rate": 2.304916246765627e-05,
      "loss": 4.2588,
      "step": 20290
    },
    {
      "epoch": 0.5453446191051995,
      "grad_norm": 2.353281259536743,
      "learning_rate": 2.3042353261609696e-05,
      "loss": 4.0934,
      "step": 20295
    },
    {
      "epoch": 0.5454789735321779,
      "grad_norm": 2.494717597961426,
      "learning_rate": 2.3035544055563122e-05,
      "loss": 4.003,
      "step": 20300
    },
    {
      "epoch": 0.5456133279591563,
      "grad_norm": 2.321584463119507,
      "learning_rate": 2.3028734849516548e-05,
      "loss": 4.1906,
      "step": 20305
    },
    {
      "epoch": 0.5457476823861346,
      "grad_norm": 2.26697039604187,
      "learning_rate": 2.302192564346997e-05,
      "loss": 4.2283,
      "step": 20310
    },
    {
      "epoch": 0.545882036813113,
      "grad_norm": 2.3440897464752197,
      "learning_rate": 2.30151164374234e-05,
      "loss": 4.1798,
      "step": 20315
    },
    {
      "epoch": 0.5460163912400914,
      "grad_norm": 2.5112297534942627,
      "learning_rate": 2.300830723137682e-05,
      "loss": 4.2309,
      "step": 20320
    },
    {
      "epoch": 0.5461507456670698,
      "grad_norm": 2.456052303314209,
      "learning_rate": 2.3001498025330247e-05,
      "loss": 4.1208,
      "step": 20325
    },
    {
      "epoch": 0.5462851000940481,
      "grad_norm": 2.561737298965454,
      "learning_rate": 2.2994688819283673e-05,
      "loss": 4.1376,
      "step": 20330
    },
    {
      "epoch": 0.5464194545210265,
      "grad_norm": 2.1906509399414062,
      "learning_rate": 2.2987879613237098e-05,
      "loss": 4.1033,
      "step": 20335
    },
    {
      "epoch": 0.5465538089480049,
      "grad_norm": 2.2807629108428955,
      "learning_rate": 2.298107040719052e-05,
      "loss": 4.0566,
      "step": 20340
    },
    {
      "epoch": 0.5466881633749832,
      "grad_norm": 2.4163055419921875,
      "learning_rate": 2.297426120114395e-05,
      "loss": 4.1012,
      "step": 20345
    },
    {
      "epoch": 0.5468225178019616,
      "grad_norm": 2.231922149658203,
      "learning_rate": 2.2967451995097372e-05,
      "loss": 4.0911,
      "step": 20350
    },
    {
      "epoch": 0.54695687222894,
      "grad_norm": 2.2921791076660156,
      "learning_rate": 2.2960642789050798e-05,
      "loss": 4.1309,
      "step": 20355
    },
    {
      "epoch": 0.5470912266559184,
      "grad_norm": 2.6215689182281494,
      "learning_rate": 2.2953833583004223e-05,
      "loss": 4.1754,
      "step": 20360
    },
    {
      "epoch": 0.5472255810828967,
      "grad_norm": 2.6653294563293457,
      "learning_rate": 2.294702437695765e-05,
      "loss": 4.018,
      "step": 20365
    },
    {
      "epoch": 0.547359935509875,
      "grad_norm": 2.459768295288086,
      "learning_rate": 2.294021517091107e-05,
      "loss": 4.2054,
      "step": 20370
    },
    {
      "epoch": 0.5474942899368535,
      "grad_norm": 2.4907214641571045,
      "learning_rate": 2.29334059648645e-05,
      "loss": 4.2036,
      "step": 20375
    },
    {
      "epoch": 0.5476286443638317,
      "grad_norm": 2.1773276329040527,
      "learning_rate": 2.2926596758817922e-05,
      "loss": 4.1489,
      "step": 20380
    },
    {
      "epoch": 0.5477629987908101,
      "grad_norm": 2.4334163665771484,
      "learning_rate": 2.2919787552771348e-05,
      "loss": 4.1849,
      "step": 20385
    },
    {
      "epoch": 0.5478973532177885,
      "grad_norm": 2.4763669967651367,
      "learning_rate": 2.2912978346724774e-05,
      "loss": 4.2323,
      "step": 20390
    },
    {
      "epoch": 0.5480317076447669,
      "grad_norm": 2.412395715713501,
      "learning_rate": 2.29061691406782e-05,
      "loss": 4.1651,
      "step": 20395
    },
    {
      "epoch": 0.5481660620717452,
      "grad_norm": 2.480158805847168,
      "learning_rate": 2.289935993463162e-05,
      "loss": 4.2449,
      "step": 20400
    },
    {
      "epoch": 0.5483004164987236,
      "grad_norm": 2.4406051635742188,
      "learning_rate": 2.289255072858505e-05,
      "loss": 4.1259,
      "step": 20405
    },
    {
      "epoch": 0.548434770925702,
      "grad_norm": 2.395028829574585,
      "learning_rate": 2.2885741522538473e-05,
      "loss": 4.049,
      "step": 20410
    },
    {
      "epoch": 0.5485691253526803,
      "grad_norm": 2.440051794052124,
      "learning_rate": 2.28789323164919e-05,
      "loss": 4.1243,
      "step": 20415
    },
    {
      "epoch": 0.5487034797796587,
      "grad_norm": 2.453970193862915,
      "learning_rate": 2.2872123110445324e-05,
      "loss": 4.2168,
      "step": 20420
    },
    {
      "epoch": 0.5488378342066371,
      "grad_norm": 2.291794538497925,
      "learning_rate": 2.2865313904398747e-05,
      "loss": 4.0539,
      "step": 20425
    },
    {
      "epoch": 0.5489721886336155,
      "grad_norm": 2.299351692199707,
      "learning_rate": 2.2858504698352172e-05,
      "loss": 4.1295,
      "step": 20430
    },
    {
      "epoch": 0.5491065430605938,
      "grad_norm": 2.4439659118652344,
      "learning_rate": 2.2851695492305598e-05,
      "loss": 4.2213,
      "step": 20435
    },
    {
      "epoch": 0.5492408974875722,
      "grad_norm": 2.3744044303894043,
      "learning_rate": 2.2844886286259024e-05,
      "loss": 4.2061,
      "step": 20440
    },
    {
      "epoch": 0.5493752519145506,
      "grad_norm": 2.4056203365325928,
      "learning_rate": 2.2838077080212446e-05,
      "loss": 4.3279,
      "step": 20445
    },
    {
      "epoch": 0.5495096063415289,
      "grad_norm": 2.2639098167419434,
      "learning_rate": 2.2831267874165875e-05,
      "loss": 4.1631,
      "step": 20450
    },
    {
      "epoch": 0.5496439607685073,
      "grad_norm": 2.667806625366211,
      "learning_rate": 2.2824458668119297e-05,
      "loss": 4.2885,
      "step": 20455
    },
    {
      "epoch": 0.5497783151954857,
      "grad_norm": 2.5146920680999756,
      "learning_rate": 2.2817649462072723e-05,
      "loss": 4.2166,
      "step": 20460
    },
    {
      "epoch": 0.5499126696224641,
      "grad_norm": 2.5367133617401123,
      "learning_rate": 2.281084025602615e-05,
      "loss": 4.1183,
      "step": 20465
    },
    {
      "epoch": 0.5500470240494424,
      "grad_norm": 2.7820565700531006,
      "learning_rate": 2.2804031049979574e-05,
      "loss": 4.312,
      "step": 20470
    },
    {
      "epoch": 0.5501813784764208,
      "grad_norm": 2.483693838119507,
      "learning_rate": 2.2797221843932996e-05,
      "loss": 4.0924,
      "step": 20475
    },
    {
      "epoch": 0.5503157329033992,
      "grad_norm": 2.498946189880371,
      "learning_rate": 2.2790412637886425e-05,
      "loss": 4.0793,
      "step": 20480
    },
    {
      "epoch": 0.5504500873303776,
      "grad_norm": 2.5973119735717773,
      "learning_rate": 2.2783603431839848e-05,
      "loss": 4.1266,
      "step": 20485
    },
    {
      "epoch": 0.5505844417573559,
      "grad_norm": 2.579442024230957,
      "learning_rate": 2.2776794225793273e-05,
      "loss": 4.139,
      "step": 20490
    },
    {
      "epoch": 0.5507187961843343,
      "grad_norm": 2.523515224456787,
      "learning_rate": 2.27699850197467e-05,
      "loss": 4.0474,
      "step": 20495
    },
    {
      "epoch": 0.5508531506113127,
      "grad_norm": 2.3674824237823486,
      "learning_rate": 2.2763175813700125e-05,
      "loss": 4.1653,
      "step": 20500
    },
    {
      "epoch": 0.550987505038291,
      "grad_norm": 2.652372360229492,
      "learning_rate": 2.2756366607653547e-05,
      "loss": 4.1785,
      "step": 20505
    },
    {
      "epoch": 0.5511218594652694,
      "grad_norm": 2.6511383056640625,
      "learning_rate": 2.2749557401606976e-05,
      "loss": 4.2264,
      "step": 20510
    },
    {
      "epoch": 0.5512562138922478,
      "grad_norm": 2.3554482460021973,
      "learning_rate": 2.2742748195560398e-05,
      "loss": 3.9877,
      "step": 20515
    },
    {
      "epoch": 0.5513905683192262,
      "grad_norm": 2.494297504425049,
      "learning_rate": 2.2735938989513824e-05,
      "loss": 4.1563,
      "step": 20520
    },
    {
      "epoch": 0.5515249227462045,
      "grad_norm": 2.714960813522339,
      "learning_rate": 2.272912978346725e-05,
      "loss": 4.2001,
      "step": 20525
    },
    {
      "epoch": 0.5516592771731829,
      "grad_norm": 2.6100189685821533,
      "learning_rate": 2.2722320577420675e-05,
      "loss": 4.2379,
      "step": 20530
    },
    {
      "epoch": 0.5517936316001613,
      "grad_norm": 2.5327558517456055,
      "learning_rate": 2.2715511371374097e-05,
      "loss": 4.2828,
      "step": 20535
    },
    {
      "epoch": 0.5519279860271395,
      "grad_norm": 2.411360740661621,
      "learning_rate": 2.2708702165327527e-05,
      "loss": 4.1847,
      "step": 20540
    },
    {
      "epoch": 0.552062340454118,
      "grad_norm": 2.309910297393799,
      "learning_rate": 2.270189295928095e-05,
      "loss": 4.0581,
      "step": 20545
    },
    {
      "epoch": 0.5521966948810964,
      "grad_norm": 2.232516050338745,
      "learning_rate": 2.2695083753234374e-05,
      "loss": 4.1728,
      "step": 20550
    },
    {
      "epoch": 0.5523310493080748,
      "grad_norm": 2.402663230895996,
      "learning_rate": 2.26882745471878e-05,
      "loss": 4.0971,
      "step": 20555
    },
    {
      "epoch": 0.552465403735053,
      "grad_norm": 2.5849103927612305,
      "learning_rate": 2.2681465341141226e-05,
      "loss": 4.2701,
      "step": 20560
    },
    {
      "epoch": 0.5525997581620314,
      "grad_norm": 2.499608278274536,
      "learning_rate": 2.2674656135094648e-05,
      "loss": 4.1197,
      "step": 20565
    },
    {
      "epoch": 0.5527341125890098,
      "grad_norm": 2.7025771141052246,
      "learning_rate": 2.2667846929048074e-05,
      "loss": 4.1194,
      "step": 20570
    },
    {
      "epoch": 0.5528684670159881,
      "grad_norm": 2.3658978939056396,
      "learning_rate": 2.26610377230015e-05,
      "loss": 4.2911,
      "step": 20575
    },
    {
      "epoch": 0.5530028214429665,
      "grad_norm": 2.522081136703491,
      "learning_rate": 2.265422851695492e-05,
      "loss": 4.0739,
      "step": 20580
    },
    {
      "epoch": 0.5531371758699449,
      "grad_norm": 2.35756516456604,
      "learning_rate": 2.264741931090835e-05,
      "loss": 4.1436,
      "step": 20585
    },
    {
      "epoch": 0.5532715302969233,
      "grad_norm": 2.6146721839904785,
      "learning_rate": 2.2640610104861773e-05,
      "loss": 4.1043,
      "step": 20590
    },
    {
      "epoch": 0.5534058847239016,
      "grad_norm": 2.3395774364471436,
      "learning_rate": 2.26338008988152e-05,
      "loss": 4.1637,
      "step": 20595
    },
    {
      "epoch": 0.55354023915088,
      "grad_norm": 2.4828941822052,
      "learning_rate": 2.2626991692768624e-05,
      "loss": 4.034,
      "step": 20600
    },
    {
      "epoch": 0.5536745935778584,
      "grad_norm": 2.860846519470215,
      "learning_rate": 2.262018248672205e-05,
      "loss": 4.2536,
      "step": 20605
    },
    {
      "epoch": 0.5538089480048367,
      "grad_norm": 2.4971113204956055,
      "learning_rate": 2.2613373280675472e-05,
      "loss": 4.1792,
      "step": 20610
    },
    {
      "epoch": 0.5539433024318151,
      "grad_norm": 2.5421531200408936,
      "learning_rate": 2.26065640746289e-05,
      "loss": 4.2742,
      "step": 20615
    },
    {
      "epoch": 0.5540776568587935,
      "grad_norm": 2.6608033180236816,
      "learning_rate": 2.2599754868582323e-05,
      "loss": 4.1716,
      "step": 20620
    },
    {
      "epoch": 0.5542120112857719,
      "grad_norm": 2.200185775756836,
      "learning_rate": 2.259294566253575e-05,
      "loss": 4.2262,
      "step": 20625
    },
    {
      "epoch": 0.5543463657127502,
      "grad_norm": 2.2931649684906006,
      "learning_rate": 2.2586136456489175e-05,
      "loss": 4.1396,
      "step": 20630
    },
    {
      "epoch": 0.5544807201397286,
      "grad_norm": 2.406660318374634,
      "learning_rate": 2.25793272504426e-05,
      "loss": 4.1707,
      "step": 20635
    },
    {
      "epoch": 0.554615074566707,
      "grad_norm": 2.3451435565948486,
      "learning_rate": 2.2572518044396023e-05,
      "loss": 4.1712,
      "step": 20640
    },
    {
      "epoch": 0.5547494289936853,
      "grad_norm": 2.387694835662842,
      "learning_rate": 2.2565708838349452e-05,
      "loss": 4.2478,
      "step": 20645
    },
    {
      "epoch": 0.5548837834206637,
      "grad_norm": 2.3927502632141113,
      "learning_rate": 2.2558899632302874e-05,
      "loss": 4.1262,
      "step": 20650
    },
    {
      "epoch": 0.5550181378476421,
      "grad_norm": 2.4236385822296143,
      "learning_rate": 2.25520904262563e-05,
      "loss": 4.1831,
      "step": 20655
    },
    {
      "epoch": 0.5551524922746205,
      "grad_norm": 2.401318311691284,
      "learning_rate": 2.2545281220209725e-05,
      "loss": 4.0151,
      "step": 20660
    },
    {
      "epoch": 0.5552868467015988,
      "grad_norm": 2.52939510345459,
      "learning_rate": 2.253847201416315e-05,
      "loss": 4.151,
      "step": 20665
    },
    {
      "epoch": 0.5554212011285772,
      "grad_norm": 2.6336584091186523,
      "learning_rate": 2.2531662808116573e-05,
      "loss": 4.1574,
      "step": 20670
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 2.317826271057129,
      "learning_rate": 2.2524853602070002e-05,
      "loss": 4.1583,
      "step": 20675
    },
    {
      "epoch": 0.5556899099825339,
      "grad_norm": 2.409836530685425,
      "learning_rate": 2.2518044396023425e-05,
      "loss": 4.2743,
      "step": 20680
    },
    {
      "epoch": 0.5558242644095123,
      "grad_norm": 2.292531728744507,
      "learning_rate": 2.251123518997685e-05,
      "loss": 4.1753,
      "step": 20685
    },
    {
      "epoch": 0.5559586188364907,
      "grad_norm": 2.4504222869873047,
      "learning_rate": 2.2504425983930276e-05,
      "loss": 4.1164,
      "step": 20690
    },
    {
      "epoch": 0.5560929732634691,
      "grad_norm": 2.6997151374816895,
      "learning_rate": 2.24976167778837e-05,
      "loss": 4.1654,
      "step": 20695
    },
    {
      "epoch": 0.5562273276904474,
      "grad_norm": 2.6537537574768066,
      "learning_rate": 2.2490807571837124e-05,
      "loss": 4.1708,
      "step": 20700
    },
    {
      "epoch": 0.5563616821174258,
      "grad_norm": 2.4333767890930176,
      "learning_rate": 2.248399836579055e-05,
      "loss": 4.0793,
      "step": 20705
    },
    {
      "epoch": 0.5564960365444042,
      "grad_norm": 2.375190496444702,
      "learning_rate": 2.2477189159743975e-05,
      "loss": 4.1256,
      "step": 20710
    },
    {
      "epoch": 0.5566303909713826,
      "grad_norm": 2.3570146560668945,
      "learning_rate": 2.24703799536974e-05,
      "loss": 4.1686,
      "step": 20715
    },
    {
      "epoch": 0.5567647453983608,
      "grad_norm": 2.606595754623413,
      "learning_rate": 2.2463570747650826e-05,
      "loss": 4.2137,
      "step": 20720
    },
    {
      "epoch": 0.5568990998253392,
      "grad_norm": 2.5160701274871826,
      "learning_rate": 2.245676154160425e-05,
      "loss": 4.111,
      "step": 20725
    },
    {
      "epoch": 0.5570334542523177,
      "grad_norm": 2.5048515796661377,
      "learning_rate": 2.2449952335557674e-05,
      "loss": 4.0975,
      "step": 20730
    },
    {
      "epoch": 0.5571678086792959,
      "grad_norm": 2.754521369934082,
      "learning_rate": 2.24431431295111e-05,
      "loss": 4.0853,
      "step": 20735
    },
    {
      "epoch": 0.5573021631062743,
      "grad_norm": 2.6441848278045654,
      "learning_rate": 2.2436333923464526e-05,
      "loss": 4.1549,
      "step": 20740
    },
    {
      "epoch": 0.5574365175332527,
      "grad_norm": 2.5024492740631104,
      "learning_rate": 2.2429524717417948e-05,
      "loss": 4.1744,
      "step": 20745
    },
    {
      "epoch": 0.5575708719602311,
      "grad_norm": 2.540158987045288,
      "learning_rate": 2.2422715511371377e-05,
      "loss": 4.2186,
      "step": 20750
    },
    {
      "epoch": 0.5577052263872094,
      "grad_norm": 2.559013843536377,
      "learning_rate": 2.24159063053248e-05,
      "loss": 4.2427,
      "step": 20755
    },
    {
      "epoch": 0.5578395808141878,
      "grad_norm": 2.4841814041137695,
      "learning_rate": 2.2409097099278225e-05,
      "loss": 4.2206,
      "step": 20760
    },
    {
      "epoch": 0.5579739352411662,
      "grad_norm": 2.331387519836426,
      "learning_rate": 2.240228789323165e-05,
      "loss": 4.1377,
      "step": 20765
    },
    {
      "epoch": 0.5581082896681445,
      "grad_norm": 2.410034656524658,
      "learning_rate": 2.2395478687185076e-05,
      "loss": 4.1734,
      "step": 20770
    },
    {
      "epoch": 0.5582426440951229,
      "grad_norm": 2.638117551803589,
      "learning_rate": 2.23886694811385e-05,
      "loss": 4.2508,
      "step": 20775
    },
    {
      "epoch": 0.5583769985221013,
      "grad_norm": 2.524665355682373,
      "learning_rate": 2.2381860275091928e-05,
      "loss": 4.0133,
      "step": 20780
    },
    {
      "epoch": 0.5585113529490797,
      "grad_norm": 2.251084327697754,
      "learning_rate": 2.237505106904535e-05,
      "loss": 4.1248,
      "step": 20785
    },
    {
      "epoch": 0.558645707376058,
      "grad_norm": 2.410855293273926,
      "learning_rate": 2.2368241862998776e-05,
      "loss": 4.1253,
      "step": 20790
    },
    {
      "epoch": 0.5587800618030364,
      "grad_norm": 2.6031699180603027,
      "learning_rate": 2.23614326569522e-05,
      "loss": 4.0467,
      "step": 20795
    },
    {
      "epoch": 0.5589144162300148,
      "grad_norm": 2.643695831298828,
      "learning_rate": 2.2354623450905627e-05,
      "loss": 4.2206,
      "step": 20800
    },
    {
      "epoch": 0.5590487706569931,
      "grad_norm": 2.3928866386413574,
      "learning_rate": 2.234781424485905e-05,
      "loss": 4.1707,
      "step": 20805
    },
    {
      "epoch": 0.5591831250839715,
      "grad_norm": 2.354193925857544,
      "learning_rate": 2.2341005038812475e-05,
      "loss": 4.1936,
      "step": 20810
    },
    {
      "epoch": 0.5593174795109499,
      "grad_norm": 2.660057544708252,
      "learning_rate": 2.23341958327659e-05,
      "loss": 4.2323,
      "step": 20815
    },
    {
      "epoch": 0.5594518339379283,
      "grad_norm": 2.5935213565826416,
      "learning_rate": 2.2327386626719326e-05,
      "loss": 4.1615,
      "step": 20820
    },
    {
      "epoch": 0.5595861883649066,
      "grad_norm": 2.332077741622925,
      "learning_rate": 2.2320577420672752e-05,
      "loss": 4.2397,
      "step": 20825
    },
    {
      "epoch": 0.559720542791885,
      "grad_norm": 2.9610326290130615,
      "learning_rate": 2.2313768214626177e-05,
      "loss": 4.174,
      "step": 20830
    },
    {
      "epoch": 0.5598548972188634,
      "grad_norm": 2.5245041847229004,
      "learning_rate": 2.23069590085796e-05,
      "loss": 4.1049,
      "step": 20835
    },
    {
      "epoch": 0.5599892516458417,
      "grad_norm": 2.5319952964782715,
      "learning_rate": 2.2300149802533025e-05,
      "loss": 4.2179,
      "step": 20840
    },
    {
      "epoch": 0.5601236060728201,
      "grad_norm": 2.4502251148223877,
      "learning_rate": 2.229334059648645e-05,
      "loss": 4.2045,
      "step": 20845
    },
    {
      "epoch": 0.5602579604997985,
      "grad_norm": 2.5761771202087402,
      "learning_rate": 2.2286531390439877e-05,
      "loss": 4.1342,
      "step": 20850
    },
    {
      "epoch": 0.5603923149267769,
      "grad_norm": 2.6918392181396484,
      "learning_rate": 2.2279722184393302e-05,
      "loss": 4.3097,
      "step": 20855
    },
    {
      "epoch": 0.5605266693537552,
      "grad_norm": 2.4351468086242676,
      "learning_rate": 2.2272912978346728e-05,
      "loss": 4.1196,
      "step": 20860
    },
    {
      "epoch": 0.5606610237807336,
      "grad_norm": 2.493361711502075,
      "learning_rate": 2.226610377230015e-05,
      "loss": 4.3482,
      "step": 20865
    },
    {
      "epoch": 0.560795378207712,
      "grad_norm": 2.2544288635253906,
      "learning_rate": 2.2259294566253576e-05,
      "loss": 4.1034,
      "step": 20870
    },
    {
      "epoch": 0.5609297326346903,
      "grad_norm": 2.2137484550476074,
      "learning_rate": 2.2252485360207e-05,
      "loss": 4.1159,
      "step": 20875
    },
    {
      "epoch": 0.5610640870616687,
      "grad_norm": 2.253361225128174,
      "learning_rate": 2.2245676154160424e-05,
      "loss": 4.1537,
      "step": 20880
    },
    {
      "epoch": 0.5611984414886471,
      "grad_norm": 2.312361478805542,
      "learning_rate": 2.2238866948113853e-05,
      "loss": 4.1809,
      "step": 20885
    },
    {
      "epoch": 0.5613327959156255,
      "grad_norm": 2.561953067779541,
      "learning_rate": 2.2232057742067275e-05,
      "loss": 4.1555,
      "step": 20890
    },
    {
      "epoch": 0.5614671503426037,
      "grad_norm": 2.6642825603485107,
      "learning_rate": 2.22252485360207e-05,
      "loss": 4.2089,
      "step": 20895
    },
    {
      "epoch": 0.5616015047695821,
      "grad_norm": 2.364079475402832,
      "learning_rate": 2.2218439329974126e-05,
      "loss": 4.1105,
      "step": 20900
    },
    {
      "epoch": 0.5617358591965605,
      "grad_norm": 2.522148609161377,
      "learning_rate": 2.2211630123927552e-05,
      "loss": 4.153,
      "step": 20905
    },
    {
      "epoch": 0.561870213623539,
      "grad_norm": 2.6086487770080566,
      "learning_rate": 2.2204820917880974e-05,
      "loss": 4.1529,
      "step": 20910
    },
    {
      "epoch": 0.5620045680505172,
      "grad_norm": 2.3919057846069336,
      "learning_rate": 2.21980117118344e-05,
      "loss": 4.0552,
      "step": 20915
    },
    {
      "epoch": 0.5621389224774956,
      "grad_norm": 2.891925811767578,
      "learning_rate": 2.2191202505787826e-05,
      "loss": 4.2503,
      "step": 20920
    },
    {
      "epoch": 0.562273276904474,
      "grad_norm": 2.4272701740264893,
      "learning_rate": 2.218439329974125e-05,
      "loss": 4.0537,
      "step": 20925
    },
    {
      "epoch": 0.5624076313314523,
      "grad_norm": 2.4619650840759277,
      "learning_rate": 2.2177584093694677e-05,
      "loss": 4.1305,
      "step": 20930
    },
    {
      "epoch": 0.5625419857584307,
      "grad_norm": 2.7332234382629395,
      "learning_rate": 2.2170774887648103e-05,
      "loss": 4.0587,
      "step": 20935
    },
    {
      "epoch": 0.5626763401854091,
      "grad_norm": 2.4257700443267822,
      "learning_rate": 2.2163965681601525e-05,
      "loss": 4.1599,
      "step": 20940
    },
    {
      "epoch": 0.5628106946123875,
      "grad_norm": 2.581960916519165,
      "learning_rate": 2.215715647555495e-05,
      "loss": 4.1741,
      "step": 20945
    },
    {
      "epoch": 0.5629450490393658,
      "grad_norm": 2.698566198348999,
      "learning_rate": 2.2150347269508376e-05,
      "loss": 4.0987,
      "step": 20950
    },
    {
      "epoch": 0.5630794034663442,
      "grad_norm": 2.712111711502075,
      "learning_rate": 2.2143538063461802e-05,
      "loss": 4.1611,
      "step": 20955
    },
    {
      "epoch": 0.5632137578933226,
      "grad_norm": 2.419236183166504,
      "learning_rate": 2.2136728857415228e-05,
      "loss": 4.141,
      "step": 20960
    },
    {
      "epoch": 0.5633481123203009,
      "grad_norm": 3.135730504989624,
      "learning_rate": 2.2129919651368653e-05,
      "loss": 4.0541,
      "step": 20965
    },
    {
      "epoch": 0.5634824667472793,
      "grad_norm": 2.21832537651062,
      "learning_rate": 2.2123110445322075e-05,
      "loss": 4.1552,
      "step": 20970
    },
    {
      "epoch": 0.5636168211742577,
      "grad_norm": 2.5601906776428223,
      "learning_rate": 2.21163012392755e-05,
      "loss": 3.9808,
      "step": 20975
    },
    {
      "epoch": 0.5637511756012361,
      "grad_norm": 2.39929461479187,
      "learning_rate": 2.2109492033228927e-05,
      "loss": 4.1359,
      "step": 20980
    },
    {
      "epoch": 0.5638855300282144,
      "grad_norm": 2.49480938911438,
      "learning_rate": 2.2102682827182352e-05,
      "loss": 4.1742,
      "step": 20985
    },
    {
      "epoch": 0.5640198844551928,
      "grad_norm": 2.5368010997772217,
      "learning_rate": 2.2095873621135778e-05,
      "loss": 4.2245,
      "step": 20990
    },
    {
      "epoch": 0.5641542388821712,
      "grad_norm": 2.842209815979004,
      "learning_rate": 2.2089064415089204e-05,
      "loss": 4.1006,
      "step": 20995
    },
    {
      "epoch": 0.5642885933091495,
      "grad_norm": 2.4134883880615234,
      "learning_rate": 2.2082255209042626e-05,
      "loss": 4.1595,
      "step": 21000
    },
    {
      "epoch": 0.5644229477361279,
      "grad_norm": 2.6063039302825928,
      "learning_rate": 2.207544600299605e-05,
      "loss": 4.0428,
      "step": 21005
    },
    {
      "epoch": 0.5645573021631063,
      "grad_norm": 2.3195695877075195,
      "learning_rate": 2.2068636796949477e-05,
      "loss": 4.0183,
      "step": 21010
    },
    {
      "epoch": 0.5646916565900847,
      "grad_norm": 2.7520480155944824,
      "learning_rate": 2.20618275909029e-05,
      "loss": 4.101,
      "step": 21015
    },
    {
      "epoch": 0.564826011017063,
      "grad_norm": 2.5005245208740234,
      "learning_rate": 2.2055018384856325e-05,
      "loss": 4.1274,
      "step": 21020
    },
    {
      "epoch": 0.5649603654440414,
      "grad_norm": 2.555042028427124,
      "learning_rate": 2.204820917880975e-05,
      "loss": 4.1382,
      "step": 21025
    },
    {
      "epoch": 0.5650947198710198,
      "grad_norm": 2.6153619289398193,
      "learning_rate": 2.2041399972763177e-05,
      "loss": 4.2286,
      "step": 21030
    },
    {
      "epoch": 0.5652290742979981,
      "grad_norm": 2.40244197845459,
      "learning_rate": 2.2034590766716602e-05,
      "loss": 4.1739,
      "step": 21035
    },
    {
      "epoch": 0.5653634287249765,
      "grad_norm": 2.350398063659668,
      "learning_rate": 2.2027781560670028e-05,
      "loss": 4.2636,
      "step": 21040
    },
    {
      "epoch": 0.5654977831519549,
      "grad_norm": 2.4911649227142334,
      "learning_rate": 2.202097235462345e-05,
      "loss": 4.1035,
      "step": 21045
    },
    {
      "epoch": 0.5656321375789333,
      "grad_norm": 2.304189443588257,
      "learning_rate": 2.2014163148576876e-05,
      "loss": 4.1321,
      "step": 21050
    },
    {
      "epoch": 0.5657664920059116,
      "grad_norm": 2.6927993297576904,
      "learning_rate": 2.20073539425303e-05,
      "loss": 4.1029,
      "step": 21055
    },
    {
      "epoch": 0.56590084643289,
      "grad_norm": 2.429262638092041,
      "learning_rate": 2.2000544736483727e-05,
      "loss": 4.1829,
      "step": 21060
    },
    {
      "epoch": 0.5660352008598684,
      "grad_norm": 2.62589955329895,
      "learning_rate": 2.1993735530437153e-05,
      "loss": 4.1843,
      "step": 21065
    },
    {
      "epoch": 0.5661695552868466,
      "grad_norm": 2.417908191680908,
      "learning_rate": 2.198692632439058e-05,
      "loss": 4.2379,
      "step": 21070
    },
    {
      "epoch": 0.566303909713825,
      "grad_norm": 2.3710758686065674,
      "learning_rate": 2.1980117118344e-05,
      "loss": 4.1463,
      "step": 21075
    },
    {
      "epoch": 0.5664382641408034,
      "grad_norm": 2.3683390617370605,
      "learning_rate": 2.1973307912297426e-05,
      "loss": 4.1374,
      "step": 21080
    },
    {
      "epoch": 0.5665726185677818,
      "grad_norm": 2.4248833656311035,
      "learning_rate": 2.1966498706250852e-05,
      "loss": 4.2737,
      "step": 21085
    },
    {
      "epoch": 0.5667069729947601,
      "grad_norm": 2.4917361736297607,
      "learning_rate": 2.1959689500204278e-05,
      "loss": 4.1595,
      "step": 21090
    },
    {
      "epoch": 0.5668413274217385,
      "grad_norm": 2.4692296981811523,
      "learning_rate": 2.19528802941577e-05,
      "loss": 4.0705,
      "step": 21095
    },
    {
      "epoch": 0.5669756818487169,
      "grad_norm": 2.393446922302246,
      "learning_rate": 2.194607108811113e-05,
      "loss": 4.1911,
      "step": 21100
    },
    {
      "epoch": 0.5671100362756953,
      "grad_norm": 2.706038475036621,
      "learning_rate": 2.193926188206455e-05,
      "loss": 4.1224,
      "step": 21105
    },
    {
      "epoch": 0.5672443907026736,
      "grad_norm": 2.3723690509796143,
      "learning_rate": 2.1932452676017977e-05,
      "loss": 4.0568,
      "step": 21110
    },
    {
      "epoch": 0.567378745129652,
      "grad_norm": 2.3602070808410645,
      "learning_rate": 2.1925643469971403e-05,
      "loss": 4.0901,
      "step": 21115
    },
    {
      "epoch": 0.5675130995566304,
      "grad_norm": 2.5478780269622803,
      "learning_rate": 2.1918834263924828e-05,
      "loss": 4.1488,
      "step": 21120
    },
    {
      "epoch": 0.5676474539836087,
      "grad_norm": 2.8540353775024414,
      "learning_rate": 2.191202505787825e-05,
      "loss": 4.1776,
      "step": 21125
    },
    {
      "epoch": 0.5677818084105871,
      "grad_norm": 2.3678102493286133,
      "learning_rate": 2.190521585183168e-05,
      "loss": 4.2047,
      "step": 21130
    },
    {
      "epoch": 0.5679161628375655,
      "grad_norm": 2.411693572998047,
      "learning_rate": 2.1898406645785102e-05,
      "loss": 4.1799,
      "step": 21135
    },
    {
      "epoch": 0.5680505172645439,
      "grad_norm": 2.441145181655884,
      "learning_rate": 2.1891597439738527e-05,
      "loss": 3.9997,
      "step": 21140
    },
    {
      "epoch": 0.5681848716915222,
      "grad_norm": 2.348498582839966,
      "learning_rate": 2.1884788233691953e-05,
      "loss": 4.1109,
      "step": 21145
    },
    {
      "epoch": 0.5683192261185006,
      "grad_norm": 2.433619499206543,
      "learning_rate": 2.187797902764538e-05,
      "loss": 4.0969,
      "step": 21150
    },
    {
      "epoch": 0.568453580545479,
      "grad_norm": 2.2475550174713135,
      "learning_rate": 2.18711698215988e-05,
      "loss": 4.0951,
      "step": 21155
    },
    {
      "epoch": 0.5685879349724573,
      "grad_norm": 2.471480131149292,
      "learning_rate": 2.186436061555223e-05,
      "loss": 4.086,
      "step": 21160
    },
    {
      "epoch": 0.5687222893994357,
      "grad_norm": 2.2577619552612305,
      "learning_rate": 2.1857551409505652e-05,
      "loss": 4.1726,
      "step": 21165
    },
    {
      "epoch": 0.5688566438264141,
      "grad_norm": 2.572582483291626,
      "learning_rate": 2.1850742203459078e-05,
      "loss": 4.2507,
      "step": 21170
    },
    {
      "epoch": 0.5689909982533925,
      "grad_norm": 2.219261884689331,
      "learning_rate": 2.1843932997412504e-05,
      "loss": 4.1053,
      "step": 21175
    },
    {
      "epoch": 0.5691253526803708,
      "grad_norm": 2.6091339588165283,
      "learning_rate": 2.1837123791365926e-05,
      "loss": 4.2243,
      "step": 21180
    },
    {
      "epoch": 0.5692597071073492,
      "grad_norm": 2.366713047027588,
      "learning_rate": 2.183031458531935e-05,
      "loss": 4.348,
      "step": 21185
    },
    {
      "epoch": 0.5693940615343276,
      "grad_norm": 2.5276148319244385,
      "learning_rate": 2.1823505379272777e-05,
      "loss": 4.1545,
      "step": 21190
    },
    {
      "epoch": 0.5695284159613059,
      "grad_norm": 2.3242762088775635,
      "learning_rate": 2.1816696173226203e-05,
      "loss": 4.1163,
      "step": 21195
    },
    {
      "epoch": 0.5696627703882843,
      "grad_norm": 2.544060468673706,
      "learning_rate": 2.1809886967179625e-05,
      "loss": 4.1426,
      "step": 21200
    },
    {
      "epoch": 0.5697971248152627,
      "grad_norm": 2.455115795135498,
      "learning_rate": 2.1803077761133054e-05,
      "loss": 3.987,
      "step": 21205
    },
    {
      "epoch": 0.5699314792422411,
      "grad_norm": 2.310176134109497,
      "learning_rate": 2.1796268555086477e-05,
      "loss": 4.1611,
      "step": 21210
    },
    {
      "epoch": 0.5700658336692194,
      "grad_norm": 2.234252691268921,
      "learning_rate": 2.1789459349039902e-05,
      "loss": 4.2473,
      "step": 21215
    },
    {
      "epoch": 0.5702001880961978,
      "grad_norm": 2.49185848236084,
      "learning_rate": 2.1782650142993328e-05,
      "loss": 4.1715,
      "step": 21220
    },
    {
      "epoch": 0.5703345425231762,
      "grad_norm": 2.4109110832214355,
      "learning_rate": 2.1775840936946754e-05,
      "loss": 4.224,
      "step": 21225
    },
    {
      "epoch": 0.5704688969501545,
      "grad_norm": 2.588263988494873,
      "learning_rate": 2.1769031730900176e-05,
      "loss": 4.201,
      "step": 21230
    },
    {
      "epoch": 0.5706032513771329,
      "grad_norm": 2.3941123485565186,
      "learning_rate": 2.1762222524853605e-05,
      "loss": 4.1412,
      "step": 21235
    },
    {
      "epoch": 0.5707376058041113,
      "grad_norm": 2.6348400115966797,
      "learning_rate": 2.1755413318807027e-05,
      "loss": 4.0545,
      "step": 21240
    },
    {
      "epoch": 0.5708719602310897,
      "grad_norm": 2.5924599170684814,
      "learning_rate": 2.1748604112760453e-05,
      "loss": 4.1349,
      "step": 21245
    },
    {
      "epoch": 0.571006314658068,
      "grad_norm": 2.41534686088562,
      "learning_rate": 2.174179490671388e-05,
      "loss": 4.1472,
      "step": 21250
    },
    {
      "epoch": 0.5711406690850463,
      "grad_norm": 2.1966664791107178,
      "learning_rate": 2.1734985700667304e-05,
      "loss": 4.1537,
      "step": 21255
    },
    {
      "epoch": 0.5712750235120247,
      "grad_norm": 2.8363473415374756,
      "learning_rate": 2.1728176494620726e-05,
      "loss": 3.9646,
      "step": 21260
    },
    {
      "epoch": 0.571409377939003,
      "grad_norm": 2.4740140438079834,
      "learning_rate": 2.1721367288574155e-05,
      "loss": 4.1995,
      "step": 21265
    },
    {
      "epoch": 0.5715437323659814,
      "grad_norm": 2.234431743621826,
      "learning_rate": 2.1714558082527578e-05,
      "loss": 4.183,
      "step": 21270
    },
    {
      "epoch": 0.5716780867929598,
      "grad_norm": 2.312194347381592,
      "learning_rate": 2.1707748876481003e-05,
      "loss": 4.1702,
      "step": 21275
    },
    {
      "epoch": 0.5718124412199382,
      "grad_norm": 2.1881096363067627,
      "learning_rate": 2.170093967043443e-05,
      "loss": 4.1251,
      "step": 21280
    },
    {
      "epoch": 0.5719467956469165,
      "grad_norm": 2.6155736446380615,
      "learning_rate": 2.1694130464387855e-05,
      "loss": 4.1716,
      "step": 21285
    },
    {
      "epoch": 0.5720811500738949,
      "grad_norm": 2.5879318714141846,
      "learning_rate": 2.1687321258341277e-05,
      "loss": 4.1118,
      "step": 21290
    },
    {
      "epoch": 0.5722155045008733,
      "grad_norm": 2.5080370903015137,
      "learning_rate": 2.1680512052294706e-05,
      "loss": 4.2132,
      "step": 21295
    },
    {
      "epoch": 0.5723498589278516,
      "grad_norm": 2.4968149662017822,
      "learning_rate": 2.1673702846248128e-05,
      "loss": 4.1445,
      "step": 21300
    },
    {
      "epoch": 0.57248421335483,
      "grad_norm": 2.4467763900756836,
      "learning_rate": 2.1666893640201554e-05,
      "loss": 4.1272,
      "step": 21305
    },
    {
      "epoch": 0.5726185677818084,
      "grad_norm": 2.5330915451049805,
      "learning_rate": 2.166008443415498e-05,
      "loss": 4.0701,
      "step": 21310
    },
    {
      "epoch": 0.5727529222087868,
      "grad_norm": 2.406627655029297,
      "learning_rate": 2.1653275228108402e-05,
      "loss": 4.0937,
      "step": 21315
    },
    {
      "epoch": 0.5728872766357651,
      "grad_norm": 2.6535279750823975,
      "learning_rate": 2.1646466022061827e-05,
      "loss": 4.0587,
      "step": 21320
    },
    {
      "epoch": 0.5730216310627435,
      "grad_norm": 2.5053555965423584,
      "learning_rate": 2.1639656816015253e-05,
      "loss": 4.2499,
      "step": 21325
    },
    {
      "epoch": 0.5731559854897219,
      "grad_norm": 2.459890127182007,
      "learning_rate": 2.163284760996868e-05,
      "loss": 4.3215,
      "step": 21330
    },
    {
      "epoch": 0.5732903399167003,
      "grad_norm": 2.328902244567871,
      "learning_rate": 2.16260384039221e-05,
      "loss": 4.0986,
      "step": 21335
    },
    {
      "epoch": 0.5734246943436786,
      "grad_norm": 2.621497631072998,
      "learning_rate": 2.161922919787553e-05,
      "loss": 4.1851,
      "step": 21340
    },
    {
      "epoch": 0.573559048770657,
      "grad_norm": 2.6003329753875732,
      "learning_rate": 2.1612419991828952e-05,
      "loss": 4.2634,
      "step": 21345
    },
    {
      "epoch": 0.5736934031976354,
      "grad_norm": 2.21806263923645,
      "learning_rate": 2.1605610785782378e-05,
      "loss": 4.1281,
      "step": 21350
    },
    {
      "epoch": 0.5738277576246137,
      "grad_norm": 2.319343328475952,
      "learning_rate": 2.1598801579735804e-05,
      "loss": 4.1196,
      "step": 21355
    },
    {
      "epoch": 0.5739621120515921,
      "grad_norm": 2.5140366554260254,
      "learning_rate": 2.159199237368923e-05,
      "loss": 4.2326,
      "step": 21360
    },
    {
      "epoch": 0.5740964664785705,
      "grad_norm": 2.5040814876556396,
      "learning_rate": 2.158518316764265e-05,
      "loss": 4.1909,
      "step": 21365
    },
    {
      "epoch": 0.5742308209055489,
      "grad_norm": 2.4275403022766113,
      "learning_rate": 2.157837396159608e-05,
      "loss": 4.1706,
      "step": 21370
    },
    {
      "epoch": 0.5743651753325272,
      "grad_norm": 2.3612277507781982,
      "learning_rate": 2.1571564755549503e-05,
      "loss": 4.1657,
      "step": 21375
    },
    {
      "epoch": 0.5744995297595056,
      "grad_norm": 2.4991815090179443,
      "learning_rate": 2.156475554950293e-05,
      "loss": 4.1876,
      "step": 21380
    },
    {
      "epoch": 0.574633884186484,
      "grad_norm": 2.522857904434204,
      "learning_rate": 2.1557946343456354e-05,
      "loss": 4.0154,
      "step": 21385
    },
    {
      "epoch": 0.5747682386134623,
      "grad_norm": 2.778536081314087,
      "learning_rate": 2.155113713740978e-05,
      "loss": 4.127,
      "step": 21390
    },
    {
      "epoch": 0.5749025930404407,
      "grad_norm": 2.424699544906616,
      "learning_rate": 2.1544327931363202e-05,
      "loss": 4.079,
      "step": 21395
    },
    {
      "epoch": 0.5750369474674191,
      "grad_norm": 2.8420650959014893,
      "learning_rate": 2.153751872531663e-05,
      "loss": 4.1867,
      "step": 21400
    },
    {
      "epoch": 0.5751713018943975,
      "grad_norm": 2.427560567855835,
      "learning_rate": 2.1530709519270053e-05,
      "loss": 4.1642,
      "step": 21405
    },
    {
      "epoch": 0.5753056563213758,
      "grad_norm": 2.263148784637451,
      "learning_rate": 2.152390031322348e-05,
      "loss": 4.0444,
      "step": 21410
    },
    {
      "epoch": 0.5754400107483542,
      "grad_norm": 2.7196755409240723,
      "learning_rate": 2.1517091107176905e-05,
      "loss": 4.0274,
      "step": 21415
    },
    {
      "epoch": 0.5755743651753326,
      "grad_norm": 2.6666336059570312,
      "learning_rate": 2.151028190113033e-05,
      "loss": 4.2947,
      "step": 21420
    },
    {
      "epoch": 0.5757087196023108,
      "grad_norm": 2.3433547019958496,
      "learning_rate": 2.1503472695083753e-05,
      "loss": 4.1682,
      "step": 21425
    },
    {
      "epoch": 0.5758430740292892,
      "grad_norm": 2.3195765018463135,
      "learning_rate": 2.1496663489037182e-05,
      "loss": 4.1648,
      "step": 21430
    },
    {
      "epoch": 0.5759774284562676,
      "grad_norm": 2.3919289112091064,
      "learning_rate": 2.1489854282990604e-05,
      "loss": 4.141,
      "step": 21435
    },
    {
      "epoch": 0.576111782883246,
      "grad_norm": 2.2470927238464355,
      "learning_rate": 2.148304507694403e-05,
      "loss": 4.232,
      "step": 21440
    },
    {
      "epoch": 0.5762461373102243,
      "grad_norm": 2.4836103916168213,
      "learning_rate": 2.1476235870897455e-05,
      "loss": 4.0608,
      "step": 21445
    },
    {
      "epoch": 0.5763804917372027,
      "grad_norm": 2.297057867050171,
      "learning_rate": 2.146942666485088e-05,
      "loss": 4.0939,
      "step": 21450
    },
    {
      "epoch": 0.5765148461641811,
      "grad_norm": 2.409677028656006,
      "learning_rate": 2.1462617458804303e-05,
      "loss": 4.2906,
      "step": 21455
    },
    {
      "epoch": 0.5766492005911594,
      "grad_norm": 2.8025062084198,
      "learning_rate": 2.145580825275773e-05,
      "loss": 4.1385,
      "step": 21460
    },
    {
      "epoch": 0.5767835550181378,
      "grad_norm": 2.5799732208251953,
      "learning_rate": 2.1448999046711155e-05,
      "loss": 4.247,
      "step": 21465
    },
    {
      "epoch": 0.5769179094451162,
      "grad_norm": 2.610898494720459,
      "learning_rate": 2.1442189840664577e-05,
      "loss": 4.1302,
      "step": 21470
    },
    {
      "epoch": 0.5770522638720946,
      "grad_norm": 2.526172399520874,
      "learning_rate": 2.1435380634618006e-05,
      "loss": 4.1503,
      "step": 21475
    },
    {
      "epoch": 0.5771866182990729,
      "grad_norm": 2.4924721717834473,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 4.1262,
      "step": 21480
    },
    {
      "epoch": 0.5773209727260513,
      "grad_norm": 2.412170171737671,
      "learning_rate": 2.1421762222524854e-05,
      "loss": 4.1073,
      "step": 21485
    },
    {
      "epoch": 0.5774553271530297,
      "grad_norm": 2.571408271789551,
      "learning_rate": 2.141495301647828e-05,
      "loss": 4.154,
      "step": 21490
    },
    {
      "epoch": 0.577589681580008,
      "grad_norm": 2.2729480266571045,
      "learning_rate": 2.1408143810431705e-05,
      "loss": 4.1942,
      "step": 21495
    },
    {
      "epoch": 0.5777240360069864,
      "grad_norm": 2.198112964630127,
      "learning_rate": 2.1401334604385127e-05,
      "loss": 4.1168,
      "step": 21500
    },
    {
      "epoch": 0.5778583904339648,
      "grad_norm": 2.3941686153411865,
      "learning_rate": 2.1394525398338556e-05,
      "loss": 4.0659,
      "step": 21505
    },
    {
      "epoch": 0.5779927448609432,
      "grad_norm": 2.497990846633911,
      "learning_rate": 2.138771619229198e-05,
      "loss": 4.1463,
      "step": 21510
    },
    {
      "epoch": 0.5781270992879215,
      "grad_norm": 2.365633249282837,
      "learning_rate": 2.1380906986245404e-05,
      "loss": 4.1456,
      "step": 21515
    },
    {
      "epoch": 0.5782614537148999,
      "grad_norm": 2.716989755630493,
      "learning_rate": 2.137409778019883e-05,
      "loss": 4.1191,
      "step": 21520
    },
    {
      "epoch": 0.5783958081418783,
      "grad_norm": 2.3200552463531494,
      "learning_rate": 2.1367288574152256e-05,
      "loss": 4.1264,
      "step": 21525
    },
    {
      "epoch": 0.5785301625688567,
      "grad_norm": 2.610867738723755,
      "learning_rate": 2.1360479368105678e-05,
      "loss": 4.1158,
      "step": 21530
    },
    {
      "epoch": 0.578664516995835,
      "grad_norm": 2.740169048309326,
      "learning_rate": 2.1353670162059107e-05,
      "loss": 4.1211,
      "step": 21535
    },
    {
      "epoch": 0.5787988714228134,
      "grad_norm": 2.4921469688415527,
      "learning_rate": 2.134686095601253e-05,
      "loss": 4.06,
      "step": 21540
    },
    {
      "epoch": 0.5789332258497918,
      "grad_norm": 2.4875035285949707,
      "learning_rate": 2.1340051749965955e-05,
      "loss": 4.1232,
      "step": 21545
    },
    {
      "epoch": 0.5790675802767701,
      "grad_norm": 2.601644992828369,
      "learning_rate": 2.133324254391938e-05,
      "loss": 4.1255,
      "step": 21550
    },
    {
      "epoch": 0.5792019347037485,
      "grad_norm": 2.507610559463501,
      "learning_rate": 2.1326433337872806e-05,
      "loss": 4.1206,
      "step": 21555
    },
    {
      "epoch": 0.5793362891307269,
      "grad_norm": 2.703639268875122,
      "learning_rate": 2.131962413182623e-05,
      "loss": 4.1207,
      "step": 21560
    },
    {
      "epoch": 0.5794706435577053,
      "grad_norm": 2.456883430480957,
      "learning_rate": 2.1312814925779658e-05,
      "loss": 4.0974,
      "step": 21565
    },
    {
      "epoch": 0.5796049979846836,
      "grad_norm": 2.650303840637207,
      "learning_rate": 2.130600571973308e-05,
      "loss": 4.167,
      "step": 21570
    },
    {
      "epoch": 0.579739352411662,
      "grad_norm": 2.395326614379883,
      "learning_rate": 2.1299196513686505e-05,
      "loss": 4.2058,
      "step": 21575
    },
    {
      "epoch": 0.5798737068386404,
      "grad_norm": 2.5663039684295654,
      "learning_rate": 2.129238730763993e-05,
      "loss": 4.0195,
      "step": 21580
    },
    {
      "epoch": 0.5800080612656187,
      "grad_norm": 2.563399076461792,
      "learning_rate": 2.1285578101593357e-05,
      "loss": 4.2077,
      "step": 21585
    },
    {
      "epoch": 0.5801424156925971,
      "grad_norm": 2.460015058517456,
      "learning_rate": 2.127876889554678e-05,
      "loss": 4.1429,
      "step": 21590
    },
    {
      "epoch": 0.5802767701195755,
      "grad_norm": 2.565241813659668,
      "learning_rate": 2.1271959689500208e-05,
      "loss": 4.2122,
      "step": 21595
    },
    {
      "epoch": 0.5804111245465539,
      "grad_norm": 2.3245444297790527,
      "learning_rate": 2.126515048345363e-05,
      "loss": 4.1782,
      "step": 21600
    },
    {
      "epoch": 0.5805454789735321,
      "grad_norm": 2.3632869720458984,
      "learning_rate": 2.1258341277407056e-05,
      "loss": 4.2287,
      "step": 21605
    },
    {
      "epoch": 0.5806798334005105,
      "grad_norm": 2.4487085342407227,
      "learning_rate": 2.125153207136048e-05,
      "loss": 4.0185,
      "step": 21610
    },
    {
      "epoch": 0.580814187827489,
      "grad_norm": 2.6842966079711914,
      "learning_rate": 2.1244722865313904e-05,
      "loss": 4.0466,
      "step": 21615
    },
    {
      "epoch": 0.5809485422544672,
      "grad_norm": 2.2872655391693115,
      "learning_rate": 2.123791365926733e-05,
      "loss": 4.1465,
      "step": 21620
    },
    {
      "epoch": 0.5810828966814456,
      "grad_norm": 2.5575578212738037,
      "learning_rate": 2.1231104453220755e-05,
      "loss": 4.0063,
      "step": 21625
    },
    {
      "epoch": 0.581217251108424,
      "grad_norm": 2.6015141010284424,
      "learning_rate": 2.122429524717418e-05,
      "loss": 4.0867,
      "step": 21630
    },
    {
      "epoch": 0.5813516055354024,
      "grad_norm": 2.2065632343292236,
      "learning_rate": 2.1217486041127603e-05,
      "loss": 4.0896,
      "step": 21635
    },
    {
      "epoch": 0.5814859599623807,
      "grad_norm": 2.5754806995391846,
      "learning_rate": 2.1210676835081032e-05,
      "loss": 4.2525,
      "step": 21640
    },
    {
      "epoch": 0.5816203143893591,
      "grad_norm": 2.376110792160034,
      "learning_rate": 2.1203867629034455e-05,
      "loss": 4.0846,
      "step": 21645
    },
    {
      "epoch": 0.5817546688163375,
      "grad_norm": 2.5564475059509277,
      "learning_rate": 2.119705842298788e-05,
      "loss": 4.1456,
      "step": 21650
    },
    {
      "epoch": 0.5818890232433158,
      "grad_norm": 2.4386701583862305,
      "learning_rate": 2.1190249216941306e-05,
      "loss": 4.0591,
      "step": 21655
    },
    {
      "epoch": 0.5820233776702942,
      "grad_norm": 2.3748061656951904,
      "learning_rate": 2.118344001089473e-05,
      "loss": 4.0772,
      "step": 21660
    },
    {
      "epoch": 0.5821577320972726,
      "grad_norm": 2.557924747467041,
      "learning_rate": 2.1176630804848154e-05,
      "loss": 4.0266,
      "step": 21665
    },
    {
      "epoch": 0.582292086524251,
      "grad_norm": 2.337669849395752,
      "learning_rate": 2.1169821598801583e-05,
      "loss": 4.1596,
      "step": 21670
    },
    {
      "epoch": 0.5824264409512293,
      "grad_norm": 2.524343729019165,
      "learning_rate": 2.1163012392755005e-05,
      "loss": 4.2227,
      "step": 21675
    },
    {
      "epoch": 0.5825607953782077,
      "grad_norm": 2.3988471031188965,
      "learning_rate": 2.115620318670843e-05,
      "loss": 4.0089,
      "step": 21680
    },
    {
      "epoch": 0.5826951498051861,
      "grad_norm": 2.673433780670166,
      "learning_rate": 2.1149393980661856e-05,
      "loss": 4.21,
      "step": 21685
    },
    {
      "epoch": 0.5828295042321644,
      "grad_norm": 2.5309226512908936,
      "learning_rate": 2.1142584774615282e-05,
      "loss": 4.2518,
      "step": 21690
    },
    {
      "epoch": 0.5829638586591428,
      "grad_norm": 2.3771448135375977,
      "learning_rate": 2.1135775568568704e-05,
      "loss": 4.2681,
      "step": 21695
    },
    {
      "epoch": 0.5830982130861212,
      "grad_norm": 2.400330066680908,
      "learning_rate": 2.1128966362522133e-05,
      "loss": 4.0531,
      "step": 21700
    },
    {
      "epoch": 0.5832325675130996,
      "grad_norm": 2.4196760654449463,
      "learning_rate": 2.1122157156475556e-05,
      "loss": 4.1098,
      "step": 21705
    },
    {
      "epoch": 0.5833669219400779,
      "grad_norm": 2.4828715324401855,
      "learning_rate": 2.111534795042898e-05,
      "loss": 4.1702,
      "step": 21710
    },
    {
      "epoch": 0.5835012763670563,
      "grad_norm": 2.8380351066589355,
      "learning_rate": 2.1108538744382407e-05,
      "loss": 4.1871,
      "step": 21715
    },
    {
      "epoch": 0.5836356307940347,
      "grad_norm": 2.6905953884124756,
      "learning_rate": 2.1101729538335833e-05,
      "loss": 4.1665,
      "step": 21720
    },
    {
      "epoch": 0.583769985221013,
      "grad_norm": 2.534278392791748,
      "learning_rate": 2.1094920332289255e-05,
      "loss": 4.1575,
      "step": 21725
    },
    {
      "epoch": 0.5839043396479914,
      "grad_norm": 2.3490195274353027,
      "learning_rate": 2.1088111126242684e-05,
      "loss": 4.0772,
      "step": 21730
    },
    {
      "epoch": 0.5840386940749698,
      "grad_norm": 2.5073323249816895,
      "learning_rate": 2.1081301920196106e-05,
      "loss": 4.078,
      "step": 21735
    },
    {
      "epoch": 0.5841730485019482,
      "grad_norm": 2.3051095008850098,
      "learning_rate": 2.1074492714149532e-05,
      "loss": 4.2001,
      "step": 21740
    },
    {
      "epoch": 0.5843074029289265,
      "grad_norm": 2.3367397785186768,
      "learning_rate": 2.1067683508102957e-05,
      "loss": 4.1001,
      "step": 21745
    },
    {
      "epoch": 0.5844417573559049,
      "grad_norm": 2.6593167781829834,
      "learning_rate": 2.1060874302056383e-05,
      "loss": 4.2346,
      "step": 21750
    },
    {
      "epoch": 0.5845761117828833,
      "grad_norm": 2.541367530822754,
      "learning_rate": 2.1054065096009805e-05,
      "loss": 4.2642,
      "step": 21755
    },
    {
      "epoch": 0.5847104662098617,
      "grad_norm": 2.3647971153259277,
      "learning_rate": 2.104725588996323e-05,
      "loss": 4.1797,
      "step": 21760
    },
    {
      "epoch": 0.58484482063684,
      "grad_norm": 2.5706794261932373,
      "learning_rate": 2.1040446683916657e-05,
      "loss": 4.1423,
      "step": 21765
    },
    {
      "epoch": 0.5849791750638184,
      "grad_norm": 2.45180082321167,
      "learning_rate": 2.103363747787008e-05,
      "loss": 4.1894,
      "step": 21770
    },
    {
      "epoch": 0.5851135294907968,
      "grad_norm": 2.325651168823242,
      "learning_rate": 2.1026828271823508e-05,
      "loss": 4.1712,
      "step": 21775
    },
    {
      "epoch": 0.585247883917775,
      "grad_norm": 2.274667501449585,
      "learning_rate": 2.102001906577693e-05,
      "loss": 4.1197,
      "step": 21780
    },
    {
      "epoch": 0.5853822383447534,
      "grad_norm": 2.317607879638672,
      "learning_rate": 2.1013209859730356e-05,
      "loss": 4.0482,
      "step": 21785
    },
    {
      "epoch": 0.5855165927717318,
      "grad_norm": 2.290598154067993,
      "learning_rate": 2.100640065368378e-05,
      "loss": 4.1337,
      "step": 21790
    },
    {
      "epoch": 0.5856509471987102,
      "grad_norm": 2.2927801609039307,
      "learning_rate": 2.0999591447637207e-05,
      "loss": 4.1949,
      "step": 21795
    },
    {
      "epoch": 0.5857853016256885,
      "grad_norm": 2.4697580337524414,
      "learning_rate": 2.099278224159063e-05,
      "loss": 4.2268,
      "step": 21800
    },
    {
      "epoch": 0.5859196560526669,
      "grad_norm": 2.3395369052886963,
      "learning_rate": 2.098597303554406e-05,
      "loss": 4.131,
      "step": 21805
    },
    {
      "epoch": 0.5860540104796453,
      "grad_norm": 2.3884780406951904,
      "learning_rate": 2.097916382949748e-05,
      "loss": 4.1601,
      "step": 21810
    },
    {
      "epoch": 0.5861883649066236,
      "grad_norm": 2.6849544048309326,
      "learning_rate": 2.0972354623450907e-05,
      "loss": 4.2637,
      "step": 21815
    },
    {
      "epoch": 0.586322719333602,
      "grad_norm": 2.737492322921753,
      "learning_rate": 2.0965545417404332e-05,
      "loss": 4.1959,
      "step": 21820
    },
    {
      "epoch": 0.5864570737605804,
      "grad_norm": 2.4521985054016113,
      "learning_rate": 2.0958736211357758e-05,
      "loss": 4.1551,
      "step": 21825
    },
    {
      "epoch": 0.5865914281875588,
      "grad_norm": 2.733161211013794,
      "learning_rate": 2.095192700531118e-05,
      "loss": 4.194,
      "step": 21830
    },
    {
      "epoch": 0.5867257826145371,
      "grad_norm": 2.5179874897003174,
      "learning_rate": 2.094511779926461e-05,
      "loss": 4.1532,
      "step": 21835
    },
    {
      "epoch": 0.5868601370415155,
      "grad_norm": 2.318056344985962,
      "learning_rate": 2.093830859321803e-05,
      "loss": 4.2084,
      "step": 21840
    },
    {
      "epoch": 0.5869944914684939,
      "grad_norm": 2.405965566635132,
      "learning_rate": 2.0931499387171457e-05,
      "loss": 4.1152,
      "step": 21845
    },
    {
      "epoch": 0.5871288458954722,
      "grad_norm": 2.3581771850585938,
      "learning_rate": 2.0924690181124883e-05,
      "loss": 4.0698,
      "step": 21850
    },
    {
      "epoch": 0.5872632003224506,
      "grad_norm": 2.7119951248168945,
      "learning_rate": 2.091788097507831e-05,
      "loss": 4.2815,
      "step": 21855
    },
    {
      "epoch": 0.587397554749429,
      "grad_norm": 2.4672129154205322,
      "learning_rate": 2.091107176903173e-05,
      "loss": 4.0579,
      "step": 21860
    },
    {
      "epoch": 0.5875319091764074,
      "grad_norm": 2.7595431804656982,
      "learning_rate": 2.0904262562985156e-05,
      "loss": 4.1299,
      "step": 21865
    },
    {
      "epoch": 0.5876662636033857,
      "grad_norm": 2.457031488418579,
      "learning_rate": 2.0897453356938582e-05,
      "loss": 4.0784,
      "step": 21870
    },
    {
      "epoch": 0.5878006180303641,
      "grad_norm": 2.6421079635620117,
      "learning_rate": 2.0890644150892008e-05,
      "loss": 4.15,
      "step": 21875
    },
    {
      "epoch": 0.5879349724573425,
      "grad_norm": 2.50795578956604,
      "learning_rate": 2.0883834944845433e-05,
      "loss": 4.1966,
      "step": 21880
    },
    {
      "epoch": 0.5880693268843208,
      "grad_norm": 2.4230599403381348,
      "learning_rate": 2.087702573879886e-05,
      "loss": 4.0272,
      "step": 21885
    },
    {
      "epoch": 0.5882036813112992,
      "grad_norm": 2.385113000869751,
      "learning_rate": 2.087021653275228e-05,
      "loss": 3.9862,
      "step": 21890
    },
    {
      "epoch": 0.5883380357382776,
      "grad_norm": 2.7116150856018066,
      "learning_rate": 2.0863407326705707e-05,
      "loss": 4.1324,
      "step": 21895
    },
    {
      "epoch": 0.588472390165256,
      "grad_norm": 2.4621129035949707,
      "learning_rate": 2.0856598120659133e-05,
      "loss": 4.2575,
      "step": 21900
    },
    {
      "epoch": 0.5886067445922343,
      "grad_norm": 2.600630521774292,
      "learning_rate": 2.0849788914612555e-05,
      "loss": 4.1075,
      "step": 21905
    },
    {
      "epoch": 0.5887410990192127,
      "grad_norm": 2.488813638687134,
      "learning_rate": 2.0842979708565984e-05,
      "loss": 4.1968,
      "step": 21910
    },
    {
      "epoch": 0.5888754534461911,
      "grad_norm": 2.484891653060913,
      "learning_rate": 2.0836170502519406e-05,
      "loss": 4.1315,
      "step": 21915
    },
    {
      "epoch": 0.5890098078731694,
      "grad_norm": 2.4340641498565674,
      "learning_rate": 2.0829361296472832e-05,
      "loss": 4.2214,
      "step": 21920
    },
    {
      "epoch": 0.5891441623001478,
      "grad_norm": 2.7433059215545654,
      "learning_rate": 2.0822552090426257e-05,
      "loss": 4.1408,
      "step": 21925
    },
    {
      "epoch": 0.5892785167271262,
      "grad_norm": 2.3965468406677246,
      "learning_rate": 2.0815742884379683e-05,
      "loss": 4.2074,
      "step": 21930
    },
    {
      "epoch": 0.5894128711541046,
      "grad_norm": 2.3174972534179688,
      "learning_rate": 2.0808933678333105e-05,
      "loss": 4.2261,
      "step": 21935
    },
    {
      "epoch": 0.5895472255810829,
      "grad_norm": 2.507331371307373,
      "learning_rate": 2.080212447228653e-05,
      "loss": 4.1473,
      "step": 21940
    },
    {
      "epoch": 0.5896815800080613,
      "grad_norm": 2.5130231380462646,
      "learning_rate": 2.0795315266239957e-05,
      "loss": 4.1083,
      "step": 21945
    },
    {
      "epoch": 0.5898159344350397,
      "grad_norm": 2.4337618350982666,
      "learning_rate": 2.0788506060193382e-05,
      "loss": 4.1723,
      "step": 21950
    },
    {
      "epoch": 0.5899502888620181,
      "grad_norm": 2.610342502593994,
      "learning_rate": 2.0781696854146808e-05,
      "loss": 4.1378,
      "step": 21955
    },
    {
      "epoch": 0.5900846432889963,
      "grad_norm": 2.3479819297790527,
      "learning_rate": 2.0774887648100234e-05,
      "loss": 4.1549,
      "step": 21960
    },
    {
      "epoch": 0.5902189977159747,
      "grad_norm": 2.491716146469116,
      "learning_rate": 2.0768078442053656e-05,
      "loss": 3.9447,
      "step": 21965
    },
    {
      "epoch": 0.5903533521429531,
      "grad_norm": 2.4012537002563477,
      "learning_rate": 2.076126923600708e-05,
      "loss": 4.1278,
      "step": 21970
    },
    {
      "epoch": 0.5904877065699314,
      "grad_norm": 2.4862544536590576,
      "learning_rate": 2.0754460029960507e-05,
      "loss": 4.2167,
      "step": 21975
    },
    {
      "epoch": 0.5906220609969098,
      "grad_norm": 2.5078492164611816,
      "learning_rate": 2.0747650823913933e-05,
      "loss": 4.2002,
      "step": 21980
    },
    {
      "epoch": 0.5907564154238882,
      "grad_norm": 2.36895489692688,
      "learning_rate": 2.074084161786736e-05,
      "loss": 4.2097,
      "step": 21985
    },
    {
      "epoch": 0.5908907698508666,
      "grad_norm": 2.631854295730591,
      "learning_rate": 2.0734032411820784e-05,
      "loss": 4.1482,
      "step": 21990
    },
    {
      "epoch": 0.5910251242778449,
      "grad_norm": 2.496811628341675,
      "learning_rate": 2.0727223205774206e-05,
      "loss": 4.2098,
      "step": 21995
    },
    {
      "epoch": 0.5911594787048233,
      "grad_norm": 2.411099910736084,
      "learning_rate": 2.0720413999727632e-05,
      "loss": 4.2255,
      "step": 22000
    },
    {
      "epoch": 0.5912938331318017,
      "grad_norm": 2.4848039150238037,
      "learning_rate": 2.0713604793681058e-05,
      "loss": 3.9915,
      "step": 22005
    },
    {
      "epoch": 0.59142818755878,
      "grad_norm": 2.4488344192504883,
      "learning_rate": 2.0706795587634483e-05,
      "loss": 4.0866,
      "step": 22010
    },
    {
      "epoch": 0.5915625419857584,
      "grad_norm": 2.357998847961426,
      "learning_rate": 2.069998638158791e-05,
      "loss": 4.129,
      "step": 22015
    },
    {
      "epoch": 0.5916968964127368,
      "grad_norm": 2.5680506229400635,
      "learning_rate": 2.0693177175541335e-05,
      "loss": 4.0841,
      "step": 22020
    },
    {
      "epoch": 0.5918312508397152,
      "grad_norm": 2.5710983276367188,
      "learning_rate": 2.0686367969494757e-05,
      "loss": 4.2075,
      "step": 22025
    },
    {
      "epoch": 0.5919656052666935,
      "grad_norm": 2.276674270629883,
      "learning_rate": 2.0679558763448183e-05,
      "loss": 4.1095,
      "step": 22030
    },
    {
      "epoch": 0.5920999596936719,
      "grad_norm": 2.4630093574523926,
      "learning_rate": 2.067274955740161e-05,
      "loss": 4.1682,
      "step": 22035
    },
    {
      "epoch": 0.5922343141206503,
      "grad_norm": 2.5769236087799072,
      "learning_rate": 2.0665940351355034e-05,
      "loss": 4.1505,
      "step": 22040
    },
    {
      "epoch": 0.5923686685476286,
      "grad_norm": 2.5042190551757812,
      "learning_rate": 2.0659131145308456e-05,
      "loss": 4.0999,
      "step": 22045
    },
    {
      "epoch": 0.592503022974607,
      "grad_norm": 2.5071041584014893,
      "learning_rate": 2.0652321939261885e-05,
      "loss": 4.1222,
      "step": 22050
    },
    {
      "epoch": 0.5926373774015854,
      "grad_norm": 2.7088139057159424,
      "learning_rate": 2.0645512733215308e-05,
      "loss": 4.2416,
      "step": 22055
    },
    {
      "epoch": 0.5927717318285638,
      "grad_norm": 2.429441213607788,
      "learning_rate": 2.0638703527168733e-05,
      "loss": 4.2183,
      "step": 22060
    },
    {
      "epoch": 0.5929060862555421,
      "grad_norm": 2.394956588745117,
      "learning_rate": 2.063189432112216e-05,
      "loss": 4.0863,
      "step": 22065
    },
    {
      "epoch": 0.5930404406825205,
      "grad_norm": 2.723422050476074,
      "learning_rate": 2.062508511507558e-05,
      "loss": 4.11,
      "step": 22070
    },
    {
      "epoch": 0.5931747951094989,
      "grad_norm": 2.4403798580169678,
      "learning_rate": 2.0618275909029007e-05,
      "loss": 4.1297,
      "step": 22075
    },
    {
      "epoch": 0.5933091495364772,
      "grad_norm": 2.635958433151245,
      "learning_rate": 2.0611466702982432e-05,
      "loss": 4.0512,
      "step": 22080
    },
    {
      "epoch": 0.5934435039634556,
      "grad_norm": 2.4640920162200928,
      "learning_rate": 2.0604657496935858e-05,
      "loss": 4.16,
      "step": 22085
    },
    {
      "epoch": 0.593577858390434,
      "grad_norm": 2.5697410106658936,
      "learning_rate": 2.0597848290889284e-05,
      "loss": 4.1867,
      "step": 22090
    },
    {
      "epoch": 0.5937122128174124,
      "grad_norm": 2.4925036430358887,
      "learning_rate": 2.059103908484271e-05,
      "loss": 4.1497,
      "step": 22095
    },
    {
      "epoch": 0.5938465672443907,
      "grad_norm": 2.5219483375549316,
      "learning_rate": 2.0584229878796132e-05,
      "loss": 4.1575,
      "step": 22100
    },
    {
      "epoch": 0.5939809216713691,
      "grad_norm": 2.61572003364563,
      "learning_rate": 2.0577420672749557e-05,
      "loss": 4.0744,
      "step": 22105
    },
    {
      "epoch": 0.5941152760983475,
      "grad_norm": 2.399329900741577,
      "learning_rate": 2.0570611466702983e-05,
      "loss": 4.1379,
      "step": 22110
    },
    {
      "epoch": 0.5942496305253258,
      "grad_norm": 2.555159091949463,
      "learning_rate": 2.056380226065641e-05,
      "loss": 4.2174,
      "step": 22115
    },
    {
      "epoch": 0.5943839849523042,
      "grad_norm": 2.8729875087738037,
      "learning_rate": 2.0556993054609834e-05,
      "loss": 4.2415,
      "step": 22120
    },
    {
      "epoch": 0.5945183393792826,
      "grad_norm": 2.3400285243988037,
      "learning_rate": 2.055018384856326e-05,
      "loss": 4.1075,
      "step": 22125
    },
    {
      "epoch": 0.594652693806261,
      "grad_norm": 2.435699939727783,
      "learning_rate": 2.0543374642516682e-05,
      "loss": 4.024,
      "step": 22130
    },
    {
      "epoch": 0.5947870482332392,
      "grad_norm": 2.5164711475372314,
      "learning_rate": 2.0536565436470108e-05,
      "loss": 4.0939,
      "step": 22135
    },
    {
      "epoch": 0.5949214026602176,
      "grad_norm": 2.3962244987487793,
      "learning_rate": 2.0529756230423534e-05,
      "loss": 4.1665,
      "step": 22140
    },
    {
      "epoch": 0.595055757087196,
      "grad_norm": 2.6777961254119873,
      "learning_rate": 2.052294702437696e-05,
      "loss": 4.2075,
      "step": 22145
    },
    {
      "epoch": 0.5951901115141744,
      "grad_norm": 2.7548539638519287,
      "learning_rate": 2.051613781833038e-05,
      "loss": 4.1348,
      "step": 22150
    },
    {
      "epoch": 0.5953244659411527,
      "grad_norm": 2.589216470718384,
      "learning_rate": 2.050932861228381e-05,
      "loss": 4.1641,
      "step": 22155
    },
    {
      "epoch": 0.5954588203681311,
      "grad_norm": 2.634667158126831,
      "learning_rate": 2.0502519406237233e-05,
      "loss": 4.1732,
      "step": 22160
    },
    {
      "epoch": 0.5955931747951095,
      "grad_norm": 2.416656494140625,
      "learning_rate": 2.049571020019066e-05,
      "loss": 4.0596,
      "step": 22165
    },
    {
      "epoch": 0.5957275292220878,
      "grad_norm": 2.495785713195801,
      "learning_rate": 2.0488900994144084e-05,
      "loss": 4.1884,
      "step": 22170
    },
    {
      "epoch": 0.5958618836490662,
      "grad_norm": 2.389918088912964,
      "learning_rate": 2.048209178809751e-05,
      "loss": 4.1949,
      "step": 22175
    },
    {
      "epoch": 0.5959962380760446,
      "grad_norm": 2.7620787620544434,
      "learning_rate": 2.0475282582050932e-05,
      "loss": 4.1606,
      "step": 22180
    },
    {
      "epoch": 0.596130592503023,
      "grad_norm": 2.427457809448242,
      "learning_rate": 2.046847337600436e-05,
      "loss": 4.1459,
      "step": 22185
    },
    {
      "epoch": 0.5962649469300013,
      "grad_norm": 2.3366200923919678,
      "learning_rate": 2.0461664169957783e-05,
      "loss": 4.1681,
      "step": 22190
    },
    {
      "epoch": 0.5963993013569797,
      "grad_norm": 2.52113938331604,
      "learning_rate": 2.045485496391121e-05,
      "loss": 4.0902,
      "step": 22195
    },
    {
      "epoch": 0.5965336557839581,
      "grad_norm": 2.553727149963379,
      "learning_rate": 2.0448045757864635e-05,
      "loss": 3.9956,
      "step": 22200
    },
    {
      "epoch": 0.5966680102109364,
      "grad_norm": 2.5613043308258057,
      "learning_rate": 2.0441236551818057e-05,
      "loss": 3.9883,
      "step": 22205
    },
    {
      "epoch": 0.5968023646379148,
      "grad_norm": 2.2177212238311768,
      "learning_rate": 2.0434427345771483e-05,
      "loss": 4.1396,
      "step": 22210
    },
    {
      "epoch": 0.5969367190648932,
      "grad_norm": 2.3658573627471924,
      "learning_rate": 2.0427618139724908e-05,
      "loss": 4.0434,
      "step": 22215
    },
    {
      "epoch": 0.5970710734918716,
      "grad_norm": 2.5011370182037354,
      "learning_rate": 2.0420808933678334e-05,
      "loss": 4.0113,
      "step": 22220
    },
    {
      "epoch": 0.5972054279188499,
      "grad_norm": 2.560046672821045,
      "learning_rate": 2.041399972763176e-05,
      "loss": 4.1362,
      "step": 22225
    },
    {
      "epoch": 0.5973397823458283,
      "grad_norm": 2.45143723487854,
      "learning_rate": 2.0407190521585185e-05,
      "loss": 4.1644,
      "step": 22230
    },
    {
      "epoch": 0.5974741367728067,
      "grad_norm": 2.7357852458953857,
      "learning_rate": 2.0400381315538608e-05,
      "loss": 4.145,
      "step": 22235
    },
    {
      "epoch": 0.597608491199785,
      "grad_norm": 2.4763541221618652,
      "learning_rate": 2.0393572109492033e-05,
      "loss": 4.1228,
      "step": 22240
    },
    {
      "epoch": 0.5977428456267634,
      "grad_norm": 2.6149709224700928,
      "learning_rate": 2.038676290344546e-05,
      "loss": 4.0492,
      "step": 22245
    },
    {
      "epoch": 0.5978772000537418,
      "grad_norm": 2.40150785446167,
      "learning_rate": 2.0379953697398885e-05,
      "loss": 4.0577,
      "step": 22250
    },
    {
      "epoch": 0.5980115544807202,
      "grad_norm": 2.3508381843566895,
      "learning_rate": 2.0373144491352307e-05,
      "loss": 4.1088,
      "step": 22255
    },
    {
      "epoch": 0.5981459089076985,
      "grad_norm": 2.3051350116729736,
      "learning_rate": 2.0366335285305736e-05,
      "loss": 4.1648,
      "step": 22260
    },
    {
      "epoch": 0.5982802633346769,
      "grad_norm": 2.347996711730957,
      "learning_rate": 2.0359526079259158e-05,
      "loss": 4.2044,
      "step": 22265
    },
    {
      "epoch": 0.5984146177616553,
      "grad_norm": 2.5053303241729736,
      "learning_rate": 2.0352716873212584e-05,
      "loss": 4.1205,
      "step": 22270
    },
    {
      "epoch": 0.5985489721886336,
      "grad_norm": 2.537123918533325,
      "learning_rate": 2.034590766716601e-05,
      "loss": 4.1183,
      "step": 22275
    },
    {
      "epoch": 0.598683326615612,
      "grad_norm": 2.2641921043395996,
      "learning_rate": 2.0339098461119435e-05,
      "loss": 4.2135,
      "step": 22280
    },
    {
      "epoch": 0.5988176810425904,
      "grad_norm": 2.5343079566955566,
      "learning_rate": 2.0332289255072857e-05,
      "loss": 4.0324,
      "step": 22285
    },
    {
      "epoch": 0.5989520354695688,
      "grad_norm": 2.4034557342529297,
      "learning_rate": 2.0325480049026286e-05,
      "loss": 4.032,
      "step": 22290
    },
    {
      "epoch": 0.5990863898965471,
      "grad_norm": 2.7022643089294434,
      "learning_rate": 2.031867084297971e-05,
      "loss": 4.0532,
      "step": 22295
    },
    {
      "epoch": 0.5992207443235255,
      "grad_norm": 2.4151153564453125,
      "learning_rate": 2.0311861636933134e-05,
      "loss": 4.2941,
      "step": 22300
    },
    {
      "epoch": 0.5993550987505039,
      "grad_norm": 2.5753273963928223,
      "learning_rate": 2.030505243088656e-05,
      "loss": 4.0646,
      "step": 22305
    },
    {
      "epoch": 0.5994894531774821,
      "grad_norm": 2.689833164215088,
      "learning_rate": 2.0298243224839986e-05,
      "loss": 4.0747,
      "step": 22310
    },
    {
      "epoch": 0.5996238076044605,
      "grad_norm": 2.3037264347076416,
      "learning_rate": 2.0291434018793408e-05,
      "loss": 4.1806,
      "step": 22315
    },
    {
      "epoch": 0.599758162031439,
      "grad_norm": 2.5877652168273926,
      "learning_rate": 2.0284624812746837e-05,
      "loss": 4.146,
      "step": 22320
    },
    {
      "epoch": 0.5998925164584173,
      "grad_norm": 2.3690359592437744,
      "learning_rate": 2.027781560670026e-05,
      "loss": 4.2882,
      "step": 22325
    },
    {
      "epoch": 0.6000268708853956,
      "grad_norm": 2.466937303543091,
      "learning_rate": 2.0271006400653685e-05,
      "loss": 4.1442,
      "step": 22330
    },
    {
      "epoch": 0.600161225312374,
      "grad_norm": 2.5335566997528076,
      "learning_rate": 2.026419719460711e-05,
      "loss": 4.215,
      "step": 22335
    },
    {
      "epoch": 0.6002955797393524,
      "grad_norm": 2.398975372314453,
      "learning_rate": 2.0257387988560536e-05,
      "loss": 4.0675,
      "step": 22340
    },
    {
      "epoch": 0.6004299341663307,
      "grad_norm": 2.4398951530456543,
      "learning_rate": 2.025057878251396e-05,
      "loss": 4.1741,
      "step": 22345
    },
    {
      "epoch": 0.6005642885933091,
      "grad_norm": 2.5973434448242188,
      "learning_rate": 2.0243769576467387e-05,
      "loss": 4.0094,
      "step": 22350
    },
    {
      "epoch": 0.6006986430202875,
      "grad_norm": 2.7983126640319824,
      "learning_rate": 2.023696037042081e-05,
      "loss": 4.1926,
      "step": 22355
    },
    {
      "epoch": 0.6008329974472659,
      "grad_norm": 2.5550599098205566,
      "learning_rate": 2.0230151164374232e-05,
      "loss": 4.2696,
      "step": 22360
    },
    {
      "epoch": 0.6009673518742442,
      "grad_norm": 2.4594802856445312,
      "learning_rate": 2.022334195832766e-05,
      "loss": 4.1812,
      "step": 22365
    },
    {
      "epoch": 0.6011017063012226,
      "grad_norm": 2.3103930950164795,
      "learning_rate": 2.0216532752281083e-05,
      "loss": 4.1217,
      "step": 22370
    },
    {
      "epoch": 0.601236060728201,
      "grad_norm": 2.6952009201049805,
      "learning_rate": 2.020972354623451e-05,
      "loss": 4.2248,
      "step": 22375
    },
    {
      "epoch": 0.6013704151551794,
      "grad_norm": 2.544992446899414,
      "learning_rate": 2.0202914340187935e-05,
      "loss": 4.1474,
      "step": 22380
    },
    {
      "epoch": 0.6015047695821577,
      "grad_norm": 2.479602575302124,
      "learning_rate": 2.019610513414136e-05,
      "loss": 4.1407,
      "step": 22385
    },
    {
      "epoch": 0.6016391240091361,
      "grad_norm": 2.51560378074646,
      "learning_rate": 2.0189295928094783e-05,
      "loss": 4.0434,
      "step": 22390
    },
    {
      "epoch": 0.6017734784361145,
      "grad_norm": 2.2815964221954346,
      "learning_rate": 2.018248672204821e-05,
      "loss": 4.0485,
      "step": 22395
    },
    {
      "epoch": 0.6019078328630928,
      "grad_norm": 2.4722535610198975,
      "learning_rate": 2.0175677516001634e-05,
      "loss": 4.1385,
      "step": 22400
    },
    {
      "epoch": 0.6020421872900712,
      "grad_norm": 2.462843418121338,
      "learning_rate": 2.016886830995506e-05,
      "loss": 4.1007,
      "step": 22405
    },
    {
      "epoch": 0.6021765417170496,
      "grad_norm": 2.6341021060943604,
      "learning_rate": 2.0162059103908485e-05,
      "loss": 4.1208,
      "step": 22410
    },
    {
      "epoch": 0.602310896144028,
      "grad_norm": 2.4814608097076416,
      "learning_rate": 2.015524989786191e-05,
      "loss": 4.2284,
      "step": 22415
    },
    {
      "epoch": 0.6024452505710063,
      "grad_norm": 2.462836742401123,
      "learning_rate": 2.0148440691815333e-05,
      "loss": 4.1433,
      "step": 22420
    },
    {
      "epoch": 0.6025796049979847,
      "grad_norm": 2.3886430263519287,
      "learning_rate": 2.0141631485768762e-05,
      "loss": 4.1581,
      "step": 22425
    },
    {
      "epoch": 0.6027139594249631,
      "grad_norm": 2.4767515659332275,
      "learning_rate": 2.0134822279722184e-05,
      "loss": 4.145,
      "step": 22430
    },
    {
      "epoch": 0.6028483138519414,
      "grad_norm": 2.5555667877197266,
      "learning_rate": 2.012801307367561e-05,
      "loss": 4.0701,
      "step": 22435
    },
    {
      "epoch": 0.6029826682789198,
      "grad_norm": 2.3435330390930176,
      "learning_rate": 2.0121203867629036e-05,
      "loss": 4.2202,
      "step": 22440
    },
    {
      "epoch": 0.6031170227058982,
      "grad_norm": 2.296830415725708,
      "learning_rate": 2.011439466158246e-05,
      "loss": 4.1289,
      "step": 22445
    },
    {
      "epoch": 0.6032513771328766,
      "grad_norm": 2.3952317237854004,
      "learning_rate": 2.0107585455535884e-05,
      "loss": 4.0736,
      "step": 22450
    },
    {
      "epoch": 0.6033857315598549,
      "grad_norm": 2.4349145889282227,
      "learning_rate": 2.0100776249489313e-05,
      "loss": 4.2261,
      "step": 22455
    },
    {
      "epoch": 0.6035200859868333,
      "grad_norm": 2.370131492614746,
      "learning_rate": 2.0093967043442735e-05,
      "loss": 4.1734,
      "step": 22460
    },
    {
      "epoch": 0.6036544404138117,
      "grad_norm": 2.5898501873016357,
      "learning_rate": 2.008715783739616e-05,
      "loss": 4.1112,
      "step": 22465
    },
    {
      "epoch": 0.60378879484079,
      "grad_norm": 2.5350003242492676,
      "learning_rate": 2.0080348631349586e-05,
      "loss": 4.2075,
      "step": 22470
    },
    {
      "epoch": 0.6039231492677684,
      "grad_norm": 2.880460500717163,
      "learning_rate": 2.0073539425303012e-05,
      "loss": 4.1533,
      "step": 22475
    },
    {
      "epoch": 0.6040575036947468,
      "grad_norm": 2.5978922843933105,
      "learning_rate": 2.0066730219256434e-05,
      "loss": 4.0797,
      "step": 22480
    },
    {
      "epoch": 0.6041918581217252,
      "grad_norm": 2.5833215713500977,
      "learning_rate": 2.0059921013209863e-05,
      "loss": 4.1007,
      "step": 22485
    },
    {
      "epoch": 0.6043262125487034,
      "grad_norm": 2.2015109062194824,
      "learning_rate": 2.0053111807163286e-05,
      "loss": 4.0737,
      "step": 22490
    },
    {
      "epoch": 0.6044605669756818,
      "grad_norm": 2.531384229660034,
      "learning_rate": 2.004630260111671e-05,
      "loss": 4.2376,
      "step": 22495
    },
    {
      "epoch": 0.6045949214026602,
      "grad_norm": 2.487185478210449,
      "learning_rate": 2.0039493395070137e-05,
      "loss": 4.1552,
      "step": 22500
    },
    {
      "epoch": 0.6047292758296385,
      "grad_norm": 2.480384111404419,
      "learning_rate": 2.003268418902356e-05,
      "loss": 4.065,
      "step": 22505
    },
    {
      "epoch": 0.6048636302566169,
      "grad_norm": 2.320406675338745,
      "learning_rate": 2.0025874982976985e-05,
      "loss": 3.9812,
      "step": 22510
    },
    {
      "epoch": 0.6049979846835953,
      "grad_norm": 2.50247859954834,
      "learning_rate": 2.001906577693041e-05,
      "loss": 4.1094,
      "step": 22515
    },
    {
      "epoch": 0.6051323391105737,
      "grad_norm": 2.3134708404541016,
      "learning_rate": 2.0012256570883836e-05,
      "loss": 4.103,
      "step": 22520
    },
    {
      "epoch": 0.605266693537552,
      "grad_norm": 2.3216352462768555,
      "learning_rate": 2.000544736483726e-05,
      "loss": 4.0664,
      "step": 22525
    },
    {
      "epoch": 0.6054010479645304,
      "grad_norm": 2.5877232551574707,
      "learning_rate": 1.9998638158790687e-05,
      "loss": 4.0288,
      "step": 22530
    },
    {
      "epoch": 0.6055354023915088,
      "grad_norm": 2.344456672668457,
      "learning_rate": 1.999182895274411e-05,
      "loss": 4.1939,
      "step": 22535
    },
    {
      "epoch": 0.6056697568184871,
      "grad_norm": 2.8099780082702637,
      "learning_rate": 1.9985019746697535e-05,
      "loss": 4.2484,
      "step": 22540
    },
    {
      "epoch": 0.6058041112454655,
      "grad_norm": 2.29130220413208,
      "learning_rate": 1.997821054065096e-05,
      "loss": 4.121,
      "step": 22545
    },
    {
      "epoch": 0.6059384656724439,
      "grad_norm": 2.745087146759033,
      "learning_rate": 1.9971401334604387e-05,
      "loss": 4.045,
      "step": 22550
    },
    {
      "epoch": 0.6060728200994223,
      "grad_norm": 2.53539776802063,
      "learning_rate": 1.996459212855781e-05,
      "loss": 4.1005,
      "step": 22555
    },
    {
      "epoch": 0.6062071745264006,
      "grad_norm": 2.525324821472168,
      "learning_rate": 1.9957782922511238e-05,
      "loss": 4.1507,
      "step": 22560
    },
    {
      "epoch": 0.606341528953379,
      "grad_norm": 2.255467176437378,
      "learning_rate": 1.995097371646466e-05,
      "loss": 4.1514,
      "step": 22565
    },
    {
      "epoch": 0.6064758833803574,
      "grad_norm": 2.606435775756836,
      "learning_rate": 1.9944164510418086e-05,
      "loss": 4.074,
      "step": 22570
    },
    {
      "epoch": 0.6066102378073358,
      "grad_norm": 2.4113659858703613,
      "learning_rate": 1.993735530437151e-05,
      "loss": 4.1237,
      "step": 22575
    },
    {
      "epoch": 0.6067445922343141,
      "grad_norm": 2.4721245765686035,
      "learning_rate": 1.9930546098324937e-05,
      "loss": 4.2263,
      "step": 22580
    },
    {
      "epoch": 0.6068789466612925,
      "grad_norm": 2.4250636100769043,
      "learning_rate": 1.992373689227836e-05,
      "loss": 3.9658,
      "step": 22585
    },
    {
      "epoch": 0.6070133010882709,
      "grad_norm": 2.2815535068511963,
      "learning_rate": 1.991692768623179e-05,
      "loss": 4.087,
      "step": 22590
    },
    {
      "epoch": 0.6071476555152492,
      "grad_norm": 2.380229949951172,
      "learning_rate": 1.991011848018521e-05,
      "loss": 4.0158,
      "step": 22595
    },
    {
      "epoch": 0.6072820099422276,
      "grad_norm": 2.5534026622772217,
      "learning_rate": 1.9903309274138636e-05,
      "loss": 4.1271,
      "step": 22600
    },
    {
      "epoch": 0.607416364369206,
      "grad_norm": 2.5666069984436035,
      "learning_rate": 1.9896500068092062e-05,
      "loss": 4.161,
      "step": 22605
    },
    {
      "epoch": 0.6075507187961844,
      "grad_norm": 2.3857409954071045,
      "learning_rate": 1.9889690862045488e-05,
      "loss": 4.0488,
      "step": 22610
    },
    {
      "epoch": 0.6076850732231627,
      "grad_norm": 2.5152769088745117,
      "learning_rate": 1.988288165599891e-05,
      "loss": 4.1582,
      "step": 22615
    },
    {
      "epoch": 0.6078194276501411,
      "grad_norm": 2.4552152156829834,
      "learning_rate": 1.987607244995234e-05,
      "loss": 4.1694,
      "step": 22620
    },
    {
      "epoch": 0.6079537820771195,
      "grad_norm": 2.242696523666382,
      "learning_rate": 1.986926324390576e-05,
      "loss": 4.104,
      "step": 22625
    },
    {
      "epoch": 0.6080881365040978,
      "grad_norm": 2.6513919830322266,
      "learning_rate": 1.9862454037859187e-05,
      "loss": 4.016,
      "step": 22630
    },
    {
      "epoch": 0.6082224909310762,
      "grad_norm": 2.298769235610962,
      "learning_rate": 1.9855644831812613e-05,
      "loss": 4.0393,
      "step": 22635
    },
    {
      "epoch": 0.6083568453580546,
      "grad_norm": 2.584549903869629,
      "learning_rate": 1.984883562576604e-05,
      "loss": 4.1942,
      "step": 22640
    },
    {
      "epoch": 0.608491199785033,
      "grad_norm": 2.8168532848358154,
      "learning_rate": 1.984202641971946e-05,
      "loss": 4.2382,
      "step": 22645
    },
    {
      "epoch": 0.6086255542120113,
      "grad_norm": 2.5264101028442383,
      "learning_rate": 1.9835217213672886e-05,
      "loss": 4.1248,
      "step": 22650
    },
    {
      "epoch": 0.6087599086389897,
      "grad_norm": 2.3816030025482178,
      "learning_rate": 1.9828408007626312e-05,
      "loss": 4.0828,
      "step": 22655
    },
    {
      "epoch": 0.608894263065968,
      "grad_norm": 2.3510773181915283,
      "learning_rate": 1.9821598801579734e-05,
      "loss": 4.2254,
      "step": 22660
    },
    {
      "epoch": 0.6090286174929463,
      "grad_norm": 2.527829885482788,
      "learning_rate": 1.9814789595533163e-05,
      "loss": 4.1724,
      "step": 22665
    },
    {
      "epoch": 0.6091629719199247,
      "grad_norm": 2.591336250305176,
      "learning_rate": 1.9807980389486586e-05,
      "loss": 4.1205,
      "step": 22670
    },
    {
      "epoch": 0.6092973263469031,
      "grad_norm": 2.543318271636963,
      "learning_rate": 1.980117118344001e-05,
      "loss": 4.1714,
      "step": 22675
    },
    {
      "epoch": 0.6094316807738815,
      "grad_norm": 2.7578721046447754,
      "learning_rate": 1.9794361977393437e-05,
      "loss": 4.1432,
      "step": 22680
    },
    {
      "epoch": 0.6095660352008598,
      "grad_norm": 2.45051908493042,
      "learning_rate": 1.9787552771346862e-05,
      "loss": 4.1863,
      "step": 22685
    },
    {
      "epoch": 0.6097003896278382,
      "grad_norm": 2.3178536891937256,
      "learning_rate": 1.9780743565300285e-05,
      "loss": 4.1344,
      "step": 22690
    },
    {
      "epoch": 0.6098347440548166,
      "grad_norm": 2.548139810562134,
      "learning_rate": 1.9773934359253714e-05,
      "loss": 4.154,
      "step": 22695
    },
    {
      "epoch": 0.6099690984817949,
      "grad_norm": 2.522615909576416,
      "learning_rate": 1.9767125153207136e-05,
      "loss": 4.1093,
      "step": 22700
    },
    {
      "epoch": 0.6101034529087733,
      "grad_norm": 2.471193552017212,
      "learning_rate": 1.9760315947160562e-05,
      "loss": 4.0663,
      "step": 22705
    },
    {
      "epoch": 0.6102378073357517,
      "grad_norm": 2.573371171951294,
      "learning_rate": 1.9753506741113987e-05,
      "loss": 4.2157,
      "step": 22710
    },
    {
      "epoch": 0.6103721617627301,
      "grad_norm": 2.499436378479004,
      "learning_rate": 1.9746697535067413e-05,
      "loss": 4.0427,
      "step": 22715
    },
    {
      "epoch": 0.6105065161897084,
      "grad_norm": 2.2254090309143066,
      "learning_rate": 1.9739888329020835e-05,
      "loss": 4.1383,
      "step": 22720
    },
    {
      "epoch": 0.6106408706166868,
      "grad_norm": 2.4136171340942383,
      "learning_rate": 1.9733079122974264e-05,
      "loss": 4.1263,
      "step": 22725
    },
    {
      "epoch": 0.6107752250436652,
      "grad_norm": 2.8414485454559326,
      "learning_rate": 1.9726269916927687e-05,
      "loss": 4.2372,
      "step": 22730
    },
    {
      "epoch": 0.6109095794706435,
      "grad_norm": 2.499788761138916,
      "learning_rate": 1.9719460710881112e-05,
      "loss": 4.2404,
      "step": 22735
    },
    {
      "epoch": 0.6110439338976219,
      "grad_norm": 2.7592554092407227,
      "learning_rate": 1.9712651504834538e-05,
      "loss": 4.0928,
      "step": 22740
    },
    {
      "epoch": 0.6111782883246003,
      "grad_norm": 2.6579251289367676,
      "learning_rate": 1.9705842298787964e-05,
      "loss": 4.08,
      "step": 22745
    },
    {
      "epoch": 0.6113126427515787,
      "grad_norm": 2.562711477279663,
      "learning_rate": 1.9699033092741386e-05,
      "loss": 4.1041,
      "step": 22750
    },
    {
      "epoch": 0.611446997178557,
      "grad_norm": 2.5202126502990723,
      "learning_rate": 1.9692223886694815e-05,
      "loss": 4.1395,
      "step": 22755
    },
    {
      "epoch": 0.6115813516055354,
      "grad_norm": 2.36799693107605,
      "learning_rate": 1.9685414680648237e-05,
      "loss": 4.1119,
      "step": 22760
    },
    {
      "epoch": 0.6117157060325138,
      "grad_norm": 2.37776780128479,
      "learning_rate": 1.9678605474601663e-05,
      "loss": 4.0161,
      "step": 22765
    },
    {
      "epoch": 0.6118500604594922,
      "grad_norm": 2.215059280395508,
      "learning_rate": 1.967179626855509e-05,
      "loss": 4.2113,
      "step": 22770
    },
    {
      "epoch": 0.6119844148864705,
      "grad_norm": 2.1743268966674805,
      "learning_rate": 1.9664987062508514e-05,
      "loss": 4.0858,
      "step": 22775
    },
    {
      "epoch": 0.6121187693134489,
      "grad_norm": 2.383988380432129,
      "learning_rate": 1.9658177856461936e-05,
      "loss": 4.1286,
      "step": 22780
    },
    {
      "epoch": 0.6122531237404273,
      "grad_norm": 2.6955838203430176,
      "learning_rate": 1.9651368650415365e-05,
      "loss": 3.9857,
      "step": 22785
    },
    {
      "epoch": 0.6123874781674056,
      "grad_norm": 2.483969211578369,
      "learning_rate": 1.9644559444368788e-05,
      "loss": 4.0749,
      "step": 22790
    },
    {
      "epoch": 0.612521832594384,
      "grad_norm": 2.5466063022613525,
      "learning_rate": 1.9637750238322213e-05,
      "loss": 4.1663,
      "step": 22795
    },
    {
      "epoch": 0.6126561870213624,
      "grad_norm": 2.290076494216919,
      "learning_rate": 1.963094103227564e-05,
      "loss": 4.0823,
      "step": 22800
    },
    {
      "epoch": 0.6127905414483408,
      "grad_norm": 2.249882459640503,
      "learning_rate": 1.962413182622906e-05,
      "loss": 3.9947,
      "step": 22805
    },
    {
      "epoch": 0.6129248958753191,
      "grad_norm": 2.435056686401367,
      "learning_rate": 1.9617322620182487e-05,
      "loss": 3.9806,
      "step": 22810
    },
    {
      "epoch": 0.6130592503022975,
      "grad_norm": 2.6419472694396973,
      "learning_rate": 1.9610513414135913e-05,
      "loss": 4.1311,
      "step": 22815
    },
    {
      "epoch": 0.6131936047292759,
      "grad_norm": 2.5757863521575928,
      "learning_rate": 1.9603704208089338e-05,
      "loss": 4.1397,
      "step": 22820
    },
    {
      "epoch": 0.6133279591562542,
      "grad_norm": 2.287513017654419,
      "learning_rate": 1.959689500204276e-05,
      "loss": 4.1116,
      "step": 22825
    },
    {
      "epoch": 0.6134623135832326,
      "grad_norm": 2.4869492053985596,
      "learning_rate": 1.959008579599619e-05,
      "loss": 4.1547,
      "step": 22830
    },
    {
      "epoch": 0.613596668010211,
      "grad_norm": 2.454310894012451,
      "learning_rate": 1.9583276589949612e-05,
      "loss": 4.0231,
      "step": 22835
    },
    {
      "epoch": 0.6137310224371894,
      "grad_norm": 2.587329864501953,
      "learning_rate": 1.9576467383903038e-05,
      "loss": 4.0321,
      "step": 22840
    },
    {
      "epoch": 0.6138653768641676,
      "grad_norm": 2.681021213531494,
      "learning_rate": 1.9569658177856463e-05,
      "loss": 4.0545,
      "step": 22845
    },
    {
      "epoch": 0.613999731291146,
      "grad_norm": 2.6363871097564697,
      "learning_rate": 1.956284897180989e-05,
      "loss": 4.0448,
      "step": 22850
    },
    {
      "epoch": 0.6141340857181244,
      "grad_norm": 2.472973585128784,
      "learning_rate": 1.955603976576331e-05,
      "loss": 4.1081,
      "step": 22855
    },
    {
      "epoch": 0.6142684401451027,
      "grad_norm": 2.3673269748687744,
      "learning_rate": 1.954923055971674e-05,
      "loss": 4.1646,
      "step": 22860
    },
    {
      "epoch": 0.6144027945720811,
      "grad_norm": 2.4687883853912354,
      "learning_rate": 1.9542421353670162e-05,
      "loss": 4.2417,
      "step": 22865
    },
    {
      "epoch": 0.6145371489990595,
      "grad_norm": 2.32926869392395,
      "learning_rate": 1.9535612147623588e-05,
      "loss": 4.1214,
      "step": 22870
    },
    {
      "epoch": 0.6146715034260379,
      "grad_norm": 2.5463759899139404,
      "learning_rate": 1.9528802941577014e-05,
      "loss": 4.2027,
      "step": 22875
    },
    {
      "epoch": 0.6148058578530162,
      "grad_norm": 2.2446656227111816,
      "learning_rate": 1.952199373553044e-05,
      "loss": 4.0666,
      "step": 22880
    },
    {
      "epoch": 0.6149402122799946,
      "grad_norm": 2.4559826850891113,
      "learning_rate": 1.951518452948386e-05,
      "loss": 4.1547,
      "step": 22885
    },
    {
      "epoch": 0.615074566706973,
      "grad_norm": 2.57631516456604,
      "learning_rate": 1.9508375323437287e-05,
      "loss": 4.0874,
      "step": 22890
    },
    {
      "epoch": 0.6152089211339513,
      "grad_norm": 2.479353427886963,
      "learning_rate": 1.9501566117390713e-05,
      "loss": 4.1256,
      "step": 22895
    },
    {
      "epoch": 0.6153432755609297,
      "grad_norm": 2.6702775955200195,
      "learning_rate": 1.949475691134414e-05,
      "loss": 4.1814,
      "step": 22900
    },
    {
      "epoch": 0.6154776299879081,
      "grad_norm": 2.3950088024139404,
      "learning_rate": 1.9487947705297564e-05,
      "loss": 3.8723,
      "step": 22905
    },
    {
      "epoch": 0.6156119844148865,
      "grad_norm": 2.299225091934204,
      "learning_rate": 1.948113849925099e-05,
      "loss": 3.9452,
      "step": 22910
    },
    {
      "epoch": 0.6157463388418648,
      "grad_norm": 2.4507553577423096,
      "learning_rate": 1.9474329293204412e-05,
      "loss": 4.1586,
      "step": 22915
    },
    {
      "epoch": 0.6158806932688432,
      "grad_norm": 3.0516319274902344,
      "learning_rate": 1.9467520087157838e-05,
      "loss": 4.2188,
      "step": 22920
    },
    {
      "epoch": 0.6160150476958216,
      "grad_norm": 2.723109722137451,
      "learning_rate": 1.9460710881111264e-05,
      "loss": 4.08,
      "step": 22925
    },
    {
      "epoch": 0.6161494021227999,
      "grad_norm": 2.381190538406372,
      "learning_rate": 1.945390167506469e-05,
      "loss": 4.1565,
      "step": 22930
    },
    {
      "epoch": 0.6162837565497783,
      "grad_norm": 2.8123371601104736,
      "learning_rate": 1.9447092469018115e-05,
      "loss": 4.2339,
      "step": 22935
    },
    {
      "epoch": 0.6164181109767567,
      "grad_norm": 2.401191473007202,
      "learning_rate": 1.944028326297154e-05,
      "loss": 4.1952,
      "step": 22940
    },
    {
      "epoch": 0.6165524654037351,
      "grad_norm": 2.8417563438415527,
      "learning_rate": 1.9433474056924963e-05,
      "loss": 4.1601,
      "step": 22945
    },
    {
      "epoch": 0.6166868198307134,
      "grad_norm": 2.0689034461975098,
      "learning_rate": 1.942666485087839e-05,
      "loss": 4.1328,
      "step": 22950
    },
    {
      "epoch": 0.6168211742576918,
      "grad_norm": 2.408452272415161,
      "learning_rate": 1.9419855644831814e-05,
      "loss": 4.0412,
      "step": 22955
    },
    {
      "epoch": 0.6169555286846702,
      "grad_norm": 2.252880334854126,
      "learning_rate": 1.9413046438785236e-05,
      "loss": 4.0882,
      "step": 22960
    },
    {
      "epoch": 0.6170898831116485,
      "grad_norm": 2.4818756580352783,
      "learning_rate": 1.9406237232738665e-05,
      "loss": 4.2262,
      "step": 22965
    },
    {
      "epoch": 0.6172242375386269,
      "grad_norm": 2.6965458393096924,
      "learning_rate": 1.9399428026692088e-05,
      "loss": 4.2184,
      "step": 22970
    },
    {
      "epoch": 0.6173585919656053,
      "grad_norm": 2.53266978263855,
      "learning_rate": 1.9392618820645513e-05,
      "loss": 4.1018,
      "step": 22975
    },
    {
      "epoch": 0.6174929463925837,
      "grad_norm": 2.3349545001983643,
      "learning_rate": 1.938580961459894e-05,
      "loss": 4.1257,
      "step": 22980
    },
    {
      "epoch": 0.617627300819562,
      "grad_norm": 2.656087875366211,
      "learning_rate": 1.9379000408552365e-05,
      "loss": 4.2083,
      "step": 22985
    },
    {
      "epoch": 0.6177616552465404,
      "grad_norm": 2.5874805450439453,
      "learning_rate": 1.9372191202505787e-05,
      "loss": 3.9964,
      "step": 22990
    },
    {
      "epoch": 0.6178960096735188,
      "grad_norm": 2.5172836780548096,
      "learning_rate": 1.9365381996459213e-05,
      "loss": 4.1564,
      "step": 22995
    },
    {
      "epoch": 0.6180303641004972,
      "grad_norm": 2.54607892036438,
      "learning_rate": 1.9358572790412638e-05,
      "loss": 4.031,
      "step": 23000
    },
    {
      "epoch": 0.6181647185274755,
      "grad_norm": 2.553339719772339,
      "learning_rate": 1.9351763584366064e-05,
      "loss": 4.1521,
      "step": 23005
    },
    {
      "epoch": 0.6182990729544539,
      "grad_norm": 2.3109114170074463,
      "learning_rate": 1.934495437831949e-05,
      "loss": 3.9983,
      "step": 23010
    },
    {
      "epoch": 0.6184334273814323,
      "grad_norm": 2.8013088703155518,
      "learning_rate": 1.9338145172272915e-05,
      "loss": 4.1422,
      "step": 23015
    },
    {
      "epoch": 0.6185677818084105,
      "grad_norm": 2.4424445629119873,
      "learning_rate": 1.9331335966226337e-05,
      "loss": 4.088,
      "step": 23020
    },
    {
      "epoch": 0.6187021362353889,
      "grad_norm": 2.584385633468628,
      "learning_rate": 1.9324526760179763e-05,
      "loss": 4.0885,
      "step": 23025
    },
    {
      "epoch": 0.6188364906623673,
      "grad_norm": 2.710015296936035,
      "learning_rate": 1.931771755413319e-05,
      "loss": 4.0941,
      "step": 23030
    },
    {
      "epoch": 0.6189708450893457,
      "grad_norm": 2.3924832344055176,
      "learning_rate": 1.9310908348086614e-05,
      "loss": 4.0324,
      "step": 23035
    },
    {
      "epoch": 0.619105199516324,
      "grad_norm": 2.335460662841797,
      "learning_rate": 1.930409914204004e-05,
      "loss": 4.0464,
      "step": 23040
    },
    {
      "epoch": 0.6192395539433024,
      "grad_norm": 2.639539957046509,
      "learning_rate": 1.9297289935993466e-05,
      "loss": 4.1846,
      "step": 23045
    },
    {
      "epoch": 0.6193739083702808,
      "grad_norm": 2.5236294269561768,
      "learning_rate": 1.9290480729946888e-05,
      "loss": 4.054,
      "step": 23050
    },
    {
      "epoch": 0.6195082627972591,
      "grad_norm": 2.551213264465332,
      "learning_rate": 1.9283671523900314e-05,
      "loss": 4.0919,
      "step": 23055
    },
    {
      "epoch": 0.6196426172242375,
      "grad_norm": 2.537510395050049,
      "learning_rate": 1.927686231785374e-05,
      "loss": 4.1182,
      "step": 23060
    },
    {
      "epoch": 0.6197769716512159,
      "grad_norm": 2.229283332824707,
      "learning_rate": 1.9270053111807165e-05,
      "loss": 4.0587,
      "step": 23065
    },
    {
      "epoch": 0.6199113260781943,
      "grad_norm": 2.4474892616271973,
      "learning_rate": 1.926324390576059e-05,
      "loss": 4.158,
      "step": 23070
    },
    {
      "epoch": 0.6200456805051726,
      "grad_norm": 2.518063545227051,
      "learning_rate": 1.9256434699714016e-05,
      "loss": 4.0565,
      "step": 23075
    },
    {
      "epoch": 0.620180034932151,
      "grad_norm": 2.9664876461029053,
      "learning_rate": 1.924962549366744e-05,
      "loss": 4.037,
      "step": 23080
    },
    {
      "epoch": 0.6203143893591294,
      "grad_norm": 2.599026679992676,
      "learning_rate": 1.9242816287620864e-05,
      "loss": 4.1896,
      "step": 23085
    },
    {
      "epoch": 0.6204487437861077,
      "grad_norm": 2.598365068435669,
      "learning_rate": 1.923600708157429e-05,
      "loss": 4.1031,
      "step": 23090
    },
    {
      "epoch": 0.6205830982130861,
      "grad_norm": 2.4832563400268555,
      "learning_rate": 1.9229197875527712e-05,
      "loss": 4.132,
      "step": 23095
    },
    {
      "epoch": 0.6207174526400645,
      "grad_norm": 2.8324997425079346,
      "learning_rate": 1.9222388669481138e-05,
      "loss": 4.1435,
      "step": 23100
    },
    {
      "epoch": 0.6208518070670429,
      "grad_norm": 2.4392940998077393,
      "learning_rate": 1.9215579463434564e-05,
      "loss": 4.0553,
      "step": 23105
    },
    {
      "epoch": 0.6209861614940212,
      "grad_norm": 2.378605365753174,
      "learning_rate": 1.920877025738799e-05,
      "loss": 4.1543,
      "step": 23110
    },
    {
      "epoch": 0.6211205159209996,
      "grad_norm": 2.645852565765381,
      "learning_rate": 1.9201961051341415e-05,
      "loss": 4.1299,
      "step": 23115
    },
    {
      "epoch": 0.621254870347978,
      "grad_norm": 2.9687037467956543,
      "learning_rate": 1.919515184529484e-05,
      "loss": 4.0111,
      "step": 23120
    },
    {
      "epoch": 0.6213892247749563,
      "grad_norm": 2.52667498588562,
      "learning_rate": 1.9188342639248263e-05,
      "loss": 4.2092,
      "step": 23125
    },
    {
      "epoch": 0.6215235792019347,
      "grad_norm": 2.6552600860595703,
      "learning_rate": 1.918153343320169e-05,
      "loss": 4.1042,
      "step": 23130
    },
    {
      "epoch": 0.6216579336289131,
      "grad_norm": 2.560513734817505,
      "learning_rate": 1.9174724227155114e-05,
      "loss": 4.1721,
      "step": 23135
    },
    {
      "epoch": 0.6217922880558915,
      "grad_norm": 2.587587594985962,
      "learning_rate": 1.916791502110854e-05,
      "loss": 4.0285,
      "step": 23140
    },
    {
      "epoch": 0.6219266424828698,
      "grad_norm": 2.562006711959839,
      "learning_rate": 1.9161105815061965e-05,
      "loss": 4.1151,
      "step": 23145
    },
    {
      "epoch": 0.6220609969098482,
      "grad_norm": 2.562177896499634,
      "learning_rate": 1.915429660901539e-05,
      "loss": 4.1395,
      "step": 23150
    },
    {
      "epoch": 0.6221953513368266,
      "grad_norm": 2.449181318283081,
      "learning_rate": 1.9147487402968813e-05,
      "loss": 4.1576,
      "step": 23155
    },
    {
      "epoch": 0.6223297057638049,
      "grad_norm": 2.398756265640259,
      "learning_rate": 1.914067819692224e-05,
      "loss": 4.1432,
      "step": 23160
    },
    {
      "epoch": 0.6224640601907833,
      "grad_norm": 2.3193199634552,
      "learning_rate": 1.9133868990875665e-05,
      "loss": 4.2457,
      "step": 23165
    },
    {
      "epoch": 0.6225984146177617,
      "grad_norm": 2.4474058151245117,
      "learning_rate": 1.912705978482909e-05,
      "loss": 4.1154,
      "step": 23170
    },
    {
      "epoch": 0.6227327690447401,
      "grad_norm": 2.473879814147949,
      "learning_rate": 1.9120250578782516e-05,
      "loss": 4.2144,
      "step": 23175
    },
    {
      "epoch": 0.6228671234717184,
      "grad_norm": 2.364337205886841,
      "learning_rate": 1.911344137273594e-05,
      "loss": 4.1412,
      "step": 23180
    },
    {
      "epoch": 0.6230014778986968,
      "grad_norm": 2.6174914836883545,
      "learning_rate": 1.9106632166689364e-05,
      "loss": 4.2325,
      "step": 23185
    },
    {
      "epoch": 0.6231358323256752,
      "grad_norm": 2.6878976821899414,
      "learning_rate": 1.909982296064279e-05,
      "loss": 4.1644,
      "step": 23190
    },
    {
      "epoch": 0.6232701867526536,
      "grad_norm": 2.5180115699768066,
      "learning_rate": 1.9093013754596215e-05,
      "loss": 4.0946,
      "step": 23195
    },
    {
      "epoch": 0.6234045411796318,
      "grad_norm": 2.4423635005950928,
      "learning_rate": 1.908620454854964e-05,
      "loss": 4.083,
      "step": 23200
    },
    {
      "epoch": 0.6235388956066102,
      "grad_norm": 2.2712295055389404,
      "learning_rate": 1.9079395342503063e-05,
      "loss": 4.1627,
      "step": 23205
    },
    {
      "epoch": 0.6236732500335886,
      "grad_norm": 2.4675514698028564,
      "learning_rate": 1.9072586136456492e-05,
      "loss": 4.0117,
      "step": 23210
    },
    {
      "epoch": 0.6238076044605669,
      "grad_norm": 2.3903191089630127,
      "learning_rate": 1.9065776930409914e-05,
      "loss": 4.0991,
      "step": 23215
    },
    {
      "epoch": 0.6239419588875453,
      "grad_norm": 2.2487778663635254,
      "learning_rate": 1.905896772436334e-05,
      "loss": 4.049,
      "step": 23220
    },
    {
      "epoch": 0.6240763133145237,
      "grad_norm": 2.2209014892578125,
      "learning_rate": 1.9052158518316766e-05,
      "loss": 4.1354,
      "step": 23225
    },
    {
      "epoch": 0.6242106677415021,
      "grad_norm": 2.3601720333099365,
      "learning_rate": 1.904534931227019e-05,
      "loss": 4.1254,
      "step": 23230
    },
    {
      "epoch": 0.6243450221684804,
      "grad_norm": 2.379819393157959,
      "learning_rate": 1.9038540106223614e-05,
      "loss": 4.0795,
      "step": 23235
    },
    {
      "epoch": 0.6244793765954588,
      "grad_norm": 2.4841928482055664,
      "learning_rate": 1.9031730900177043e-05,
      "loss": 3.9729,
      "step": 23240
    },
    {
      "epoch": 0.6246137310224372,
      "grad_norm": 2.404937982559204,
      "learning_rate": 1.9024921694130465e-05,
      "loss": 4.0806,
      "step": 23245
    },
    {
      "epoch": 0.6247480854494155,
      "grad_norm": 2.404106378555298,
      "learning_rate": 1.901811248808389e-05,
      "loss": 4.0698,
      "step": 23250
    },
    {
      "epoch": 0.6248824398763939,
      "grad_norm": 2.6380505561828613,
      "learning_rate": 1.9011303282037316e-05,
      "loss": 4.1091,
      "step": 23255
    },
    {
      "epoch": 0.6250167943033723,
      "grad_norm": 2.383516788482666,
      "learning_rate": 1.900449407599074e-05,
      "loss": 4.1119,
      "step": 23260
    },
    {
      "epoch": 0.6251511487303507,
      "grad_norm": 2.487488269805908,
      "learning_rate": 1.8997684869944164e-05,
      "loss": 4.0706,
      "step": 23265
    },
    {
      "epoch": 0.625285503157329,
      "grad_norm": 2.4726924896240234,
      "learning_rate": 1.899087566389759e-05,
      "loss": 4.1028,
      "step": 23270
    },
    {
      "epoch": 0.6254198575843074,
      "grad_norm": 2.603287696838379,
      "learning_rate": 1.8984066457851016e-05,
      "loss": 4.1014,
      "step": 23275
    },
    {
      "epoch": 0.6255542120112858,
      "grad_norm": 2.420926332473755,
      "learning_rate": 1.8977257251804438e-05,
      "loss": 4.0684,
      "step": 23280
    },
    {
      "epoch": 0.6256885664382641,
      "grad_norm": 2.2720630168914795,
      "learning_rate": 1.8970448045757867e-05,
      "loss": 4.1713,
      "step": 23285
    },
    {
      "epoch": 0.6258229208652425,
      "grad_norm": 2.4648218154907227,
      "learning_rate": 1.896363883971129e-05,
      "loss": 4.2537,
      "step": 23290
    },
    {
      "epoch": 0.6259572752922209,
      "grad_norm": 2.4751076698303223,
      "learning_rate": 1.8956829633664715e-05,
      "loss": 3.9498,
      "step": 23295
    },
    {
      "epoch": 0.6260916297191993,
      "grad_norm": 2.5886898040771484,
      "learning_rate": 1.895002042761814e-05,
      "loss": 4.1581,
      "step": 23300
    },
    {
      "epoch": 0.6262259841461776,
      "grad_norm": 2.632793664932251,
      "learning_rate": 1.8943211221571566e-05,
      "loss": 4.1309,
      "step": 23305
    },
    {
      "epoch": 0.626360338573156,
      "grad_norm": 2.313241481781006,
      "learning_rate": 1.893640201552499e-05,
      "loss": 4.0782,
      "step": 23310
    },
    {
      "epoch": 0.6264946930001344,
      "grad_norm": 2.5592658519744873,
      "learning_rate": 1.8929592809478417e-05,
      "loss": 4.1223,
      "step": 23315
    },
    {
      "epoch": 0.6266290474271127,
      "grad_norm": 2.6594362258911133,
      "learning_rate": 1.892278360343184e-05,
      "loss": 4.0878,
      "step": 23320
    },
    {
      "epoch": 0.6267634018540911,
      "grad_norm": 2.3843441009521484,
      "learning_rate": 1.8915974397385265e-05,
      "loss": 4.1715,
      "step": 23325
    },
    {
      "epoch": 0.6268977562810695,
      "grad_norm": 2.774778127670288,
      "learning_rate": 1.890916519133869e-05,
      "loss": 4.1485,
      "step": 23330
    },
    {
      "epoch": 0.6270321107080479,
      "grad_norm": 2.4670071601867676,
      "learning_rate": 1.8902355985292117e-05,
      "loss": 4.033,
      "step": 23335
    },
    {
      "epoch": 0.6271664651350262,
      "grad_norm": 2.425337076187134,
      "learning_rate": 1.889554677924554e-05,
      "loss": 4.1102,
      "step": 23340
    },
    {
      "epoch": 0.6273008195620046,
      "grad_norm": 2.4518814086914062,
      "learning_rate": 1.8888737573198968e-05,
      "loss": 4.0707,
      "step": 23345
    },
    {
      "epoch": 0.627435173988983,
      "grad_norm": 2.4463744163513184,
      "learning_rate": 1.888192836715239e-05,
      "loss": 4.0472,
      "step": 23350
    },
    {
      "epoch": 0.6275695284159613,
      "grad_norm": 2.4812541007995605,
      "learning_rate": 1.8875119161105816e-05,
      "loss": 4.1445,
      "step": 23355
    },
    {
      "epoch": 0.6277038828429397,
      "grad_norm": 2.60377836227417,
      "learning_rate": 1.886830995505924e-05,
      "loss": 4.1688,
      "step": 23360
    },
    {
      "epoch": 0.627838237269918,
      "grad_norm": 2.58404278755188,
      "learning_rate": 1.8861500749012667e-05,
      "loss": 4.0139,
      "step": 23365
    },
    {
      "epoch": 0.6279725916968965,
      "grad_norm": 2.515747547149658,
      "learning_rate": 1.885469154296609e-05,
      "loss": 4.2004,
      "step": 23370
    },
    {
      "epoch": 0.6281069461238747,
      "grad_norm": 2.4702308177948,
      "learning_rate": 1.884788233691952e-05,
      "loss": 4.0859,
      "step": 23375
    },
    {
      "epoch": 0.6282413005508531,
      "grad_norm": 2.334240674972534,
      "learning_rate": 1.884107313087294e-05,
      "loss": 4.1229,
      "step": 23380
    },
    {
      "epoch": 0.6283756549778315,
      "grad_norm": 2.655916929244995,
      "learning_rate": 1.8834263924826366e-05,
      "loss": 4.1102,
      "step": 23385
    },
    {
      "epoch": 0.6285100094048099,
      "grad_norm": 2.7336225509643555,
      "learning_rate": 1.8827454718779792e-05,
      "loss": 4.2423,
      "step": 23390
    },
    {
      "epoch": 0.6286443638317882,
      "grad_norm": 2.4753386974334717,
      "learning_rate": 1.8820645512733214e-05,
      "loss": 4.1041,
      "step": 23395
    },
    {
      "epoch": 0.6287787182587666,
      "grad_norm": 2.513751745223999,
      "learning_rate": 1.881383630668664e-05,
      "loss": 4.1182,
      "step": 23400
    },
    {
      "epoch": 0.628913072685745,
      "grad_norm": 2.4351415634155273,
      "learning_rate": 1.8807027100640066e-05,
      "loss": 4.1418,
      "step": 23405
    },
    {
      "epoch": 0.6290474271127233,
      "grad_norm": 2.57922101020813,
      "learning_rate": 1.880021789459349e-05,
      "loss": 4.0837,
      "step": 23410
    },
    {
      "epoch": 0.6291817815397017,
      "grad_norm": 2.4887595176696777,
      "learning_rate": 1.8793408688546914e-05,
      "loss": 3.9817,
      "step": 23415
    },
    {
      "epoch": 0.6293161359666801,
      "grad_norm": 2.7448618412017822,
      "learning_rate": 1.8786599482500343e-05,
      "loss": 4.2374,
      "step": 23420
    },
    {
      "epoch": 0.6294504903936585,
      "grad_norm": 2.3277223110198975,
      "learning_rate": 1.8779790276453765e-05,
      "loss": 4.2015,
      "step": 23425
    },
    {
      "epoch": 0.6295848448206368,
      "grad_norm": 2.533170700073242,
      "learning_rate": 1.877298107040719e-05,
      "loss": 4.0211,
      "step": 23430
    },
    {
      "epoch": 0.6297191992476152,
      "grad_norm": 2.5229806900024414,
      "learning_rate": 1.8766171864360616e-05,
      "loss": 4.2786,
      "step": 23435
    },
    {
      "epoch": 0.6298535536745936,
      "grad_norm": 2.602210760116577,
      "learning_rate": 1.8759362658314042e-05,
      "loss": 4.1016,
      "step": 23440
    },
    {
      "epoch": 0.6299879081015719,
      "grad_norm": 2.4182891845703125,
      "learning_rate": 1.8752553452267464e-05,
      "loss": 4.1262,
      "step": 23445
    },
    {
      "epoch": 0.6301222625285503,
      "grad_norm": 2.6210291385650635,
      "learning_rate": 1.8745744246220893e-05,
      "loss": 4.2641,
      "step": 23450
    },
    {
      "epoch": 0.6302566169555287,
      "grad_norm": 2.6051249504089355,
      "learning_rate": 1.8738935040174315e-05,
      "loss": 4.0891,
      "step": 23455
    },
    {
      "epoch": 0.6303909713825071,
      "grad_norm": 2.6324195861816406,
      "learning_rate": 1.873212583412774e-05,
      "loss": 3.988,
      "step": 23460
    },
    {
      "epoch": 0.6305253258094854,
      "grad_norm": 2.5638813972473145,
      "learning_rate": 1.8725316628081167e-05,
      "loss": 4.1243,
      "step": 23465
    },
    {
      "epoch": 0.6306596802364638,
      "grad_norm": 2.108503580093384,
      "learning_rate": 1.8718507422034592e-05,
      "loss": 3.9905,
      "step": 23470
    },
    {
      "epoch": 0.6307940346634422,
      "grad_norm": 2.7509284019470215,
      "learning_rate": 1.8711698215988015e-05,
      "loss": 4.1112,
      "step": 23475
    },
    {
      "epoch": 0.6309283890904205,
      "grad_norm": 2.510148048400879,
      "learning_rate": 1.8704889009941444e-05,
      "loss": 4.2239,
      "step": 23480
    },
    {
      "epoch": 0.6310627435173989,
      "grad_norm": 2.4859464168548584,
      "learning_rate": 1.8698079803894866e-05,
      "loss": 4.0365,
      "step": 23485
    },
    {
      "epoch": 0.6311970979443773,
      "grad_norm": 2.697118043899536,
      "learning_rate": 1.869127059784829e-05,
      "loss": 4.1497,
      "step": 23490
    },
    {
      "epoch": 0.6313314523713557,
      "grad_norm": 2.3508384227752686,
      "learning_rate": 1.8684461391801717e-05,
      "loss": 4.0723,
      "step": 23495
    },
    {
      "epoch": 0.631465806798334,
      "grad_norm": 2.4708471298217773,
      "learning_rate": 1.8677652185755143e-05,
      "loss": 4.1264,
      "step": 23500
    },
    {
      "epoch": 0.6316001612253124,
      "grad_norm": 2.430537462234497,
      "learning_rate": 1.8670842979708565e-05,
      "loss": 4.0167,
      "step": 23505
    },
    {
      "epoch": 0.6317345156522908,
      "grad_norm": 2.691636562347412,
      "learning_rate": 1.8664033773661994e-05,
      "loss": 4.0456,
      "step": 23510
    },
    {
      "epoch": 0.6318688700792691,
      "grad_norm": 2.747800350189209,
      "learning_rate": 1.8657224567615417e-05,
      "loss": 4.0016,
      "step": 23515
    },
    {
      "epoch": 0.6320032245062475,
      "grad_norm": 2.418670415878296,
      "learning_rate": 1.8650415361568842e-05,
      "loss": 4.1811,
      "step": 23520
    },
    {
      "epoch": 0.6321375789332259,
      "grad_norm": 2.52363920211792,
      "learning_rate": 1.8643606155522268e-05,
      "loss": 4.124,
      "step": 23525
    },
    {
      "epoch": 0.6322719333602043,
      "grad_norm": 2.4816198348999023,
      "learning_rate": 1.8636796949475694e-05,
      "loss": 4.16,
      "step": 23530
    },
    {
      "epoch": 0.6324062877871826,
      "grad_norm": 2.465169668197632,
      "learning_rate": 1.8629987743429116e-05,
      "loss": 4.0824,
      "step": 23535
    },
    {
      "epoch": 0.632540642214161,
      "grad_norm": 2.6134915351867676,
      "learning_rate": 1.862317853738254e-05,
      "loss": 4.1515,
      "step": 23540
    },
    {
      "epoch": 0.6326749966411394,
      "grad_norm": 2.3546528816223145,
      "learning_rate": 1.8616369331335967e-05,
      "loss": 4.1687,
      "step": 23545
    },
    {
      "epoch": 0.6328093510681176,
      "grad_norm": 2.637240409851074,
      "learning_rate": 1.860956012528939e-05,
      "loss": 4.1086,
      "step": 23550
    },
    {
      "epoch": 0.632943705495096,
      "grad_norm": 2.5345969200134277,
      "learning_rate": 1.860275091924282e-05,
      "loss": 4.2036,
      "step": 23555
    },
    {
      "epoch": 0.6330780599220744,
      "grad_norm": 2.5617806911468506,
      "learning_rate": 1.859594171319624e-05,
      "loss": 4.1482,
      "step": 23560
    },
    {
      "epoch": 0.6332124143490528,
      "grad_norm": 2.3352832794189453,
      "learning_rate": 1.8589132507149666e-05,
      "loss": 4.0621,
      "step": 23565
    },
    {
      "epoch": 0.6333467687760311,
      "grad_norm": 2.714040517807007,
      "learning_rate": 1.8582323301103092e-05,
      "loss": 4.0689,
      "step": 23570
    },
    {
      "epoch": 0.6334811232030095,
      "grad_norm": 2.555560350418091,
      "learning_rate": 1.8575514095056518e-05,
      "loss": 4.2055,
      "step": 23575
    },
    {
      "epoch": 0.6336154776299879,
      "grad_norm": 2.444643020629883,
      "learning_rate": 1.856870488900994e-05,
      "loss": 4.1872,
      "step": 23580
    },
    {
      "epoch": 0.6337498320569662,
      "grad_norm": 2.4564497470855713,
      "learning_rate": 1.856189568296337e-05,
      "loss": 4.0709,
      "step": 23585
    },
    {
      "epoch": 0.6338841864839446,
      "grad_norm": 2.537414073944092,
      "learning_rate": 1.855508647691679e-05,
      "loss": 4.1854,
      "step": 23590
    },
    {
      "epoch": 0.634018540910923,
      "grad_norm": 2.719449520111084,
      "learning_rate": 1.8548277270870217e-05,
      "loss": 4.0443,
      "step": 23595
    },
    {
      "epoch": 0.6341528953379014,
      "grad_norm": 2.7071638107299805,
      "learning_rate": 1.8541468064823643e-05,
      "loss": 4.2322,
      "step": 23600
    },
    {
      "epoch": 0.6342872497648797,
      "grad_norm": 2.389280080795288,
      "learning_rate": 1.8534658858777068e-05,
      "loss": 4.1885,
      "step": 23605
    },
    {
      "epoch": 0.6344216041918581,
      "grad_norm": 2.6367533206939697,
      "learning_rate": 1.852784965273049e-05,
      "loss": 4.145,
      "step": 23610
    },
    {
      "epoch": 0.6345559586188365,
      "grad_norm": 2.486896276473999,
      "learning_rate": 1.852104044668392e-05,
      "loss": 4.0314,
      "step": 23615
    },
    {
      "epoch": 0.6346903130458149,
      "grad_norm": 2.4305179119110107,
      "learning_rate": 1.8514231240637342e-05,
      "loss": 4.1263,
      "step": 23620
    },
    {
      "epoch": 0.6348246674727932,
      "grad_norm": 2.4880106449127197,
      "learning_rate": 1.8507422034590767e-05,
      "loss": 4.0739,
      "step": 23625
    },
    {
      "epoch": 0.6349590218997716,
      "grad_norm": 2.849853038787842,
      "learning_rate": 1.8500612828544193e-05,
      "loss": 4.2104,
      "step": 23630
    },
    {
      "epoch": 0.63509337632675,
      "grad_norm": 2.3692846298217773,
      "learning_rate": 1.849380362249762e-05,
      "loss": 4.1339,
      "step": 23635
    },
    {
      "epoch": 0.6352277307537283,
      "grad_norm": 2.431380271911621,
      "learning_rate": 1.848699441645104e-05,
      "loss": 4.2618,
      "step": 23640
    },
    {
      "epoch": 0.6353620851807067,
      "grad_norm": 2.4687514305114746,
      "learning_rate": 1.848018521040447e-05,
      "loss": 4.0515,
      "step": 23645
    },
    {
      "epoch": 0.6354964396076851,
      "grad_norm": 2.6786999702453613,
      "learning_rate": 1.8473376004357892e-05,
      "loss": 4.0787,
      "step": 23650
    },
    {
      "epoch": 0.6356307940346635,
      "grad_norm": 2.6440019607543945,
      "learning_rate": 1.8466566798311318e-05,
      "loss": 4.0959,
      "step": 23655
    },
    {
      "epoch": 0.6357651484616418,
      "grad_norm": 2.59122371673584,
      "learning_rate": 1.8459757592264744e-05,
      "loss": 4.1463,
      "step": 23660
    },
    {
      "epoch": 0.6358995028886202,
      "grad_norm": 2.390857696533203,
      "learning_rate": 1.845294838621817e-05,
      "loss": 4.1967,
      "step": 23665
    },
    {
      "epoch": 0.6360338573155986,
      "grad_norm": 2.3467447757720947,
      "learning_rate": 1.844613918017159e-05,
      "loss": 4.1183,
      "step": 23670
    },
    {
      "epoch": 0.6361682117425769,
      "grad_norm": 2.643886089324951,
      "learning_rate": 1.843932997412502e-05,
      "loss": 4.1481,
      "step": 23675
    },
    {
      "epoch": 0.6363025661695553,
      "grad_norm": 2.538806438446045,
      "learning_rate": 1.8432520768078443e-05,
      "loss": 4.0686,
      "step": 23680
    },
    {
      "epoch": 0.6364369205965337,
      "grad_norm": 2.4303650856018066,
      "learning_rate": 1.842571156203187e-05,
      "loss": 4.154,
      "step": 23685
    },
    {
      "epoch": 0.6365712750235121,
      "grad_norm": 2.5481116771698,
      "learning_rate": 1.8418902355985294e-05,
      "loss": 4.0916,
      "step": 23690
    },
    {
      "epoch": 0.6367056294504904,
      "grad_norm": 2.4088287353515625,
      "learning_rate": 1.8412093149938717e-05,
      "loss": 4.1122,
      "step": 23695
    },
    {
      "epoch": 0.6368399838774688,
      "grad_norm": 2.5548882484436035,
      "learning_rate": 1.8405283943892142e-05,
      "loss": 4.2029,
      "step": 23700
    },
    {
      "epoch": 0.6369743383044472,
      "grad_norm": 2.6370370388031006,
      "learning_rate": 1.8398474737845568e-05,
      "loss": 4.1001,
      "step": 23705
    },
    {
      "epoch": 0.6371086927314255,
      "grad_norm": 2.501089096069336,
      "learning_rate": 1.8391665531798994e-05,
      "loss": 4.0841,
      "step": 23710
    },
    {
      "epoch": 0.6372430471584039,
      "grad_norm": 2.562368869781494,
      "learning_rate": 1.8384856325752416e-05,
      "loss": 4.1725,
      "step": 23715
    },
    {
      "epoch": 0.6373774015853823,
      "grad_norm": 2.37835955619812,
      "learning_rate": 1.8378047119705845e-05,
      "loss": 4.1658,
      "step": 23720
    },
    {
      "epoch": 0.6375117560123607,
      "grad_norm": 2.4549942016601562,
      "learning_rate": 1.8371237913659267e-05,
      "loss": 4.2019,
      "step": 23725
    },
    {
      "epoch": 0.6376461104393389,
      "grad_norm": 2.302461624145508,
      "learning_rate": 1.8364428707612693e-05,
      "loss": 4.178,
      "step": 23730
    },
    {
      "epoch": 0.6377804648663173,
      "grad_norm": 2.283285140991211,
      "learning_rate": 1.835761950156612e-05,
      "loss": 4.1144,
      "step": 23735
    },
    {
      "epoch": 0.6379148192932957,
      "grad_norm": 2.2567577362060547,
      "learning_rate": 1.8350810295519544e-05,
      "loss": 4.1405,
      "step": 23740
    },
    {
      "epoch": 0.638049173720274,
      "grad_norm": 2.4493722915649414,
      "learning_rate": 1.8344001089472966e-05,
      "loss": 4.1162,
      "step": 23745
    },
    {
      "epoch": 0.6381835281472524,
      "grad_norm": 2.5306286811828613,
      "learning_rate": 1.8337191883426395e-05,
      "loss": 4.1075,
      "step": 23750
    },
    {
      "epoch": 0.6383178825742308,
      "grad_norm": 2.7744383811950684,
      "learning_rate": 1.8330382677379818e-05,
      "loss": 4.1201,
      "step": 23755
    },
    {
      "epoch": 0.6384522370012092,
      "grad_norm": 2.6288442611694336,
      "learning_rate": 1.8323573471333243e-05,
      "loss": 4.0181,
      "step": 23760
    },
    {
      "epoch": 0.6385865914281875,
      "grad_norm": 2.571013927459717,
      "learning_rate": 1.831676426528667e-05,
      "loss": 4.0778,
      "step": 23765
    },
    {
      "epoch": 0.6387209458551659,
      "grad_norm": 2.5074269771575928,
      "learning_rate": 1.8309955059240095e-05,
      "loss": 3.9797,
      "step": 23770
    },
    {
      "epoch": 0.6388553002821443,
      "grad_norm": 2.270559310913086,
      "learning_rate": 1.8303145853193517e-05,
      "loss": 4.1654,
      "step": 23775
    },
    {
      "epoch": 0.6389896547091226,
      "grad_norm": 2.520716428756714,
      "learning_rate": 1.8296336647146946e-05,
      "loss": 4.1029,
      "step": 23780
    },
    {
      "epoch": 0.639124009136101,
      "grad_norm": 2.38871431350708,
      "learning_rate": 1.8289527441100368e-05,
      "loss": 4.1179,
      "step": 23785
    },
    {
      "epoch": 0.6392583635630794,
      "grad_norm": 2.7593979835510254,
      "learning_rate": 1.8282718235053794e-05,
      "loss": 4.1032,
      "step": 23790
    },
    {
      "epoch": 0.6393927179900578,
      "grad_norm": 2.3903448581695557,
      "learning_rate": 1.827590902900722e-05,
      "loss": 4.0687,
      "step": 23795
    },
    {
      "epoch": 0.6395270724170361,
      "grad_norm": 2.509956121444702,
      "learning_rate": 1.8269099822960645e-05,
      "loss": 4.0281,
      "step": 23800
    },
    {
      "epoch": 0.6396614268440145,
      "grad_norm": 2.368028402328491,
      "learning_rate": 1.8262290616914067e-05,
      "loss": 4.1365,
      "step": 23805
    },
    {
      "epoch": 0.6397957812709929,
      "grad_norm": 2.5107011795043945,
      "learning_rate": 1.8255481410867496e-05,
      "loss": 4.1545,
      "step": 23810
    },
    {
      "epoch": 0.6399301356979713,
      "grad_norm": 2.436769962310791,
      "learning_rate": 1.824867220482092e-05,
      "loss": 4.1481,
      "step": 23815
    },
    {
      "epoch": 0.6400644901249496,
      "grad_norm": 2.3996737003326416,
      "learning_rate": 1.8241862998774344e-05,
      "loss": 4.1041,
      "step": 23820
    },
    {
      "epoch": 0.640198844551928,
      "grad_norm": 2.498392343521118,
      "learning_rate": 1.823505379272777e-05,
      "loss": 4.0279,
      "step": 23825
    },
    {
      "epoch": 0.6403331989789064,
      "grad_norm": 2.4884490966796875,
      "learning_rate": 1.8228244586681196e-05,
      "loss": 4.0626,
      "step": 23830
    },
    {
      "epoch": 0.6404675534058847,
      "grad_norm": 2.7227866649627686,
      "learning_rate": 1.8221435380634618e-05,
      "loss": 4.1442,
      "step": 23835
    },
    {
      "epoch": 0.6406019078328631,
      "grad_norm": 2.567849636077881,
      "learning_rate": 1.8214626174588044e-05,
      "loss": 4.1008,
      "step": 23840
    },
    {
      "epoch": 0.6407362622598415,
      "grad_norm": 2.673973798751831,
      "learning_rate": 1.820781696854147e-05,
      "loss": 4.0957,
      "step": 23845
    },
    {
      "epoch": 0.6408706166868199,
      "grad_norm": 2.6582841873168945,
      "learning_rate": 1.820100776249489e-05,
      "loss": 4.1615,
      "step": 23850
    },
    {
      "epoch": 0.6410049711137982,
      "grad_norm": 2.7365658283233643,
      "learning_rate": 1.819419855644832e-05,
      "loss": 4.0928,
      "step": 23855
    },
    {
      "epoch": 0.6411393255407766,
      "grad_norm": 2.4883408546447754,
      "learning_rate": 1.8187389350401743e-05,
      "loss": 4.0893,
      "step": 23860
    },
    {
      "epoch": 0.641273679967755,
      "grad_norm": 2.628448247909546,
      "learning_rate": 1.818058014435517e-05,
      "loss": 4.137,
      "step": 23865
    },
    {
      "epoch": 0.6414080343947333,
      "grad_norm": 2.6400766372680664,
      "learning_rate": 1.8173770938308594e-05,
      "loss": 4.0436,
      "step": 23870
    },
    {
      "epoch": 0.6415423888217117,
      "grad_norm": 2.8825888633728027,
      "learning_rate": 1.816696173226202e-05,
      "loss": 4.0475,
      "step": 23875
    },
    {
      "epoch": 0.6416767432486901,
      "grad_norm": 2.4864284992218018,
      "learning_rate": 1.8160152526215442e-05,
      "loss": 3.9497,
      "step": 23880
    },
    {
      "epoch": 0.6418110976756685,
      "grad_norm": 2.3812966346740723,
      "learning_rate": 1.815334332016887e-05,
      "loss": 4.1262,
      "step": 23885
    },
    {
      "epoch": 0.6419454521026468,
      "grad_norm": 2.404181957244873,
      "learning_rate": 1.8146534114122293e-05,
      "loss": 4.1794,
      "step": 23890
    },
    {
      "epoch": 0.6420798065296252,
      "grad_norm": 2.670949935913086,
      "learning_rate": 1.813972490807572e-05,
      "loss": 4.1559,
      "step": 23895
    },
    {
      "epoch": 0.6422141609566036,
      "grad_norm": 2.530696392059326,
      "learning_rate": 1.8132915702029145e-05,
      "loss": 3.9382,
      "step": 23900
    },
    {
      "epoch": 0.6423485153835818,
      "grad_norm": 2.38751220703125,
      "learning_rate": 1.812610649598257e-05,
      "loss": 4.1264,
      "step": 23905
    },
    {
      "epoch": 0.6424828698105602,
      "grad_norm": 2.4727799892425537,
      "learning_rate": 1.8119297289935993e-05,
      "loss": 4.0848,
      "step": 23910
    },
    {
      "epoch": 0.6426172242375386,
      "grad_norm": 2.458280324935913,
      "learning_rate": 1.8112488083889422e-05,
      "loss": 4.106,
      "step": 23915
    },
    {
      "epoch": 0.642751578664517,
      "grad_norm": 2.6680119037628174,
      "learning_rate": 1.8105678877842844e-05,
      "loss": 4.1118,
      "step": 23920
    },
    {
      "epoch": 0.6428859330914953,
      "grad_norm": 2.6694414615631104,
      "learning_rate": 1.809886967179627e-05,
      "loss": 4.0348,
      "step": 23925
    },
    {
      "epoch": 0.6430202875184737,
      "grad_norm": 2.6987314224243164,
      "learning_rate": 1.8092060465749695e-05,
      "loss": 4.056,
      "step": 23930
    },
    {
      "epoch": 0.6431546419454521,
      "grad_norm": 2.3725807666778564,
      "learning_rate": 1.808525125970312e-05,
      "loss": 4.1045,
      "step": 23935
    },
    {
      "epoch": 0.6432889963724304,
      "grad_norm": 2.5497794151306152,
      "learning_rate": 1.8078442053656543e-05,
      "loss": 4.1706,
      "step": 23940
    },
    {
      "epoch": 0.6434233507994088,
      "grad_norm": 2.352515459060669,
      "learning_rate": 1.807163284760997e-05,
      "loss": 3.9439,
      "step": 23945
    },
    {
      "epoch": 0.6435577052263872,
      "grad_norm": 2.4654340744018555,
      "learning_rate": 1.8064823641563395e-05,
      "loss": 4.1169,
      "step": 23950
    },
    {
      "epoch": 0.6436920596533656,
      "grad_norm": 2.6187195777893066,
      "learning_rate": 1.805801443551682e-05,
      "loss": 4.087,
      "step": 23955
    },
    {
      "epoch": 0.6438264140803439,
      "grad_norm": 2.564617872238159,
      "learning_rate": 1.8051205229470246e-05,
      "loss": 4.0549,
      "step": 23960
    },
    {
      "epoch": 0.6439607685073223,
      "grad_norm": 2.4929139614105225,
      "learning_rate": 1.804439602342367e-05,
      "loss": 4.1109,
      "step": 23965
    },
    {
      "epoch": 0.6440951229343007,
      "grad_norm": 2.246645927429199,
      "learning_rate": 1.8037586817377094e-05,
      "loss": 4.2088,
      "step": 23970
    },
    {
      "epoch": 0.644229477361279,
      "grad_norm": 2.221834421157837,
      "learning_rate": 1.803077761133052e-05,
      "loss": 4.1836,
      "step": 23975
    },
    {
      "epoch": 0.6443638317882574,
      "grad_norm": 2.3892807960510254,
      "learning_rate": 1.8023968405283945e-05,
      "loss": 4.144,
      "step": 23980
    },
    {
      "epoch": 0.6444981862152358,
      "grad_norm": 2.687943696975708,
      "learning_rate": 1.801715919923737e-05,
      "loss": 4.1796,
      "step": 23985
    },
    {
      "epoch": 0.6446325406422142,
      "grad_norm": 2.4180691242218018,
      "learning_rate": 1.8010349993190796e-05,
      "loss": 4.1278,
      "step": 23990
    },
    {
      "epoch": 0.6447668950691925,
      "grad_norm": 2.6881208419799805,
      "learning_rate": 1.800354078714422e-05,
      "loss": 4.1401,
      "step": 23995
    },
    {
      "epoch": 0.6449012494961709,
      "grad_norm": 2.2429795265197754,
      "learning_rate": 1.7996731581097644e-05,
      "loss": 4.075,
      "step": 24000
    },
    {
      "epoch": 0.6450356039231493,
      "grad_norm": 2.478213310241699,
      "learning_rate": 1.798992237505107e-05,
      "loss": 4.1086,
      "step": 24005
    },
    {
      "epoch": 0.6451699583501276,
      "grad_norm": 2.5723352432250977,
      "learning_rate": 1.7983113169004496e-05,
      "loss": 4.17,
      "step": 24010
    },
    {
      "epoch": 0.645304312777106,
      "grad_norm": 2.44465970993042,
      "learning_rate": 1.7976303962957918e-05,
      "loss": 4.2546,
      "step": 24015
    },
    {
      "epoch": 0.6454386672040844,
      "grad_norm": 2.5174989700317383,
      "learning_rate": 1.7969494756911347e-05,
      "loss": 4.2404,
      "step": 24020
    },
    {
      "epoch": 0.6455730216310628,
      "grad_norm": 2.5027527809143066,
      "learning_rate": 1.796268555086477e-05,
      "loss": 4.0736,
      "step": 24025
    },
    {
      "epoch": 0.6457073760580411,
      "grad_norm": 2.6018006801605225,
      "learning_rate": 1.7955876344818195e-05,
      "loss": 4.0513,
      "step": 24030
    },
    {
      "epoch": 0.6458417304850195,
      "grad_norm": 2.2839057445526123,
      "learning_rate": 1.794906713877162e-05,
      "loss": 4.0872,
      "step": 24035
    },
    {
      "epoch": 0.6459760849119979,
      "grad_norm": 2.470343589782715,
      "learning_rate": 1.7942257932725046e-05,
      "loss": 4.1103,
      "step": 24040
    },
    {
      "epoch": 0.6461104393389763,
      "grad_norm": 2.6035096645355225,
      "learning_rate": 1.793544872667847e-05,
      "loss": 3.9309,
      "step": 24045
    },
    {
      "epoch": 0.6462447937659546,
      "grad_norm": 2.5959882736206055,
      "learning_rate": 1.7928639520631894e-05,
      "loss": 4.2064,
      "step": 24050
    },
    {
      "epoch": 0.646379148192933,
      "grad_norm": 2.7281436920166016,
      "learning_rate": 1.792183031458532e-05,
      "loss": 4.1966,
      "step": 24055
    },
    {
      "epoch": 0.6465135026199114,
      "grad_norm": 2.656104326248169,
      "learning_rate": 1.7915021108538745e-05,
      "loss": 4.1146,
      "step": 24060
    },
    {
      "epoch": 0.6466478570468897,
      "grad_norm": 2.3168928623199463,
      "learning_rate": 1.790821190249217e-05,
      "loss": 3.9576,
      "step": 24065
    },
    {
      "epoch": 0.646782211473868,
      "grad_norm": 2.439548969268799,
      "learning_rate": 1.7901402696445597e-05,
      "loss": 4.0711,
      "step": 24070
    },
    {
      "epoch": 0.6469165659008465,
      "grad_norm": 2.649630546569824,
      "learning_rate": 1.789459349039902e-05,
      "loss": 4.0327,
      "step": 24075
    },
    {
      "epoch": 0.6470509203278249,
      "grad_norm": 2.47031569480896,
      "learning_rate": 1.7887784284352445e-05,
      "loss": 3.9958,
      "step": 24080
    },
    {
      "epoch": 0.6471852747548031,
      "grad_norm": 2.575955867767334,
      "learning_rate": 1.788097507830587e-05,
      "loss": 4.027,
      "step": 24085
    },
    {
      "epoch": 0.6473196291817815,
      "grad_norm": 2.2539994716644287,
      "learning_rate": 1.7874165872259296e-05,
      "loss": 3.9701,
      "step": 24090
    },
    {
      "epoch": 0.6474539836087599,
      "grad_norm": 2.60310697555542,
      "learning_rate": 1.786735666621272e-05,
      "loss": 4.09,
      "step": 24095
    },
    {
      "epoch": 0.6475883380357382,
      "grad_norm": 2.4930222034454346,
      "learning_rate": 1.7860547460166147e-05,
      "loss": 4.1123,
      "step": 24100
    },
    {
      "epoch": 0.6477226924627166,
      "grad_norm": 2.4262917041778564,
      "learning_rate": 1.785373825411957e-05,
      "loss": 4.2577,
      "step": 24105
    },
    {
      "epoch": 0.647857046889695,
      "grad_norm": 2.3126251697540283,
      "learning_rate": 1.7846929048072995e-05,
      "loss": 4.1058,
      "step": 24110
    },
    {
      "epoch": 0.6479914013166734,
      "grad_norm": 2.5512583255767822,
      "learning_rate": 1.784011984202642e-05,
      "loss": 4.1864,
      "step": 24115
    },
    {
      "epoch": 0.6481257557436517,
      "grad_norm": 2.454341173171997,
      "learning_rate": 1.7833310635979847e-05,
      "loss": 4.2259,
      "step": 24120
    },
    {
      "epoch": 0.6482601101706301,
      "grad_norm": 2.76254940032959,
      "learning_rate": 1.782650142993327e-05,
      "loss": 4.0664,
      "step": 24125
    },
    {
      "epoch": 0.6483944645976085,
      "grad_norm": 2.512204885482788,
      "learning_rate": 1.7819692223886698e-05,
      "loss": 4.1579,
      "step": 24130
    },
    {
      "epoch": 0.6485288190245868,
      "grad_norm": 2.597622871398926,
      "learning_rate": 1.781288301784012e-05,
      "loss": 4.1518,
      "step": 24135
    },
    {
      "epoch": 0.6486631734515652,
      "grad_norm": 2.5101253986358643,
      "learning_rate": 1.7806073811793546e-05,
      "loss": 3.9893,
      "step": 24140
    },
    {
      "epoch": 0.6487975278785436,
      "grad_norm": 2.403233051300049,
      "learning_rate": 1.779926460574697e-05,
      "loss": 4.1149,
      "step": 24145
    },
    {
      "epoch": 0.648931882305522,
      "grad_norm": 2.6212189197540283,
      "learning_rate": 1.7792455399700394e-05,
      "loss": 4.1019,
      "step": 24150
    },
    {
      "epoch": 0.6490662367325003,
      "grad_norm": 2.514524459838867,
      "learning_rate": 1.778564619365382e-05,
      "loss": 4.1005,
      "step": 24155
    },
    {
      "epoch": 0.6492005911594787,
      "grad_norm": 2.342414379119873,
      "learning_rate": 1.7778836987607245e-05,
      "loss": 4.1548,
      "step": 24160
    },
    {
      "epoch": 0.6493349455864571,
      "grad_norm": 2.471890687942505,
      "learning_rate": 1.777202778156067e-05,
      "loss": 4.1681,
      "step": 24165
    },
    {
      "epoch": 0.6494693000134354,
      "grad_norm": 2.527994394302368,
      "learning_rate": 1.7765218575514096e-05,
      "loss": 3.9966,
      "step": 24170
    },
    {
      "epoch": 0.6496036544404138,
      "grad_norm": 2.55450439453125,
      "learning_rate": 1.7758409369467522e-05,
      "loss": 4.1512,
      "step": 24175
    },
    {
      "epoch": 0.6497380088673922,
      "grad_norm": 2.499756097793579,
      "learning_rate": 1.7751600163420944e-05,
      "loss": 4.08,
      "step": 24180
    },
    {
      "epoch": 0.6498723632943706,
      "grad_norm": 2.152082920074463,
      "learning_rate": 1.774479095737437e-05,
      "loss": 4.0514,
      "step": 24185
    },
    {
      "epoch": 0.6500067177213489,
      "grad_norm": 2.4695308208465576,
      "learning_rate": 1.7737981751327796e-05,
      "loss": 4.1256,
      "step": 24190
    },
    {
      "epoch": 0.6501410721483273,
      "grad_norm": 2.271550178527832,
      "learning_rate": 1.773117254528122e-05,
      "loss": 4.1455,
      "step": 24195
    },
    {
      "epoch": 0.6502754265753057,
      "grad_norm": 2.519408702850342,
      "learning_rate": 1.7724363339234647e-05,
      "loss": 4.0881,
      "step": 24200
    },
    {
      "epoch": 0.650409781002284,
      "grad_norm": 2.741361379623413,
      "learning_rate": 1.7717554133188073e-05,
      "loss": 4.0979,
      "step": 24205
    },
    {
      "epoch": 0.6505441354292624,
      "grad_norm": 2.3821797370910645,
      "learning_rate": 1.7710744927141495e-05,
      "loss": 4.0628,
      "step": 24210
    },
    {
      "epoch": 0.6506784898562408,
      "grad_norm": 2.4963490962982178,
      "learning_rate": 1.770393572109492e-05,
      "loss": 4.093,
      "step": 24215
    },
    {
      "epoch": 0.6508128442832192,
      "grad_norm": 2.154620885848999,
      "learning_rate": 1.7697126515048346e-05,
      "loss": 4.0586,
      "step": 24220
    },
    {
      "epoch": 0.6509471987101975,
      "grad_norm": 2.65537166595459,
      "learning_rate": 1.7690317309001772e-05,
      "loss": 4.2049,
      "step": 24225
    },
    {
      "epoch": 0.6510815531371759,
      "grad_norm": 2.3128857612609863,
      "learning_rate": 1.7683508102955194e-05,
      "loss": 4.0714,
      "step": 24230
    },
    {
      "epoch": 0.6512159075641543,
      "grad_norm": 2.6045753955841064,
      "learning_rate": 1.7676698896908623e-05,
      "loss": 4.1728,
      "step": 24235
    },
    {
      "epoch": 0.6513502619911327,
      "grad_norm": 2.408907413482666,
      "learning_rate": 1.7669889690862045e-05,
      "loss": 4.1417,
      "step": 24240
    },
    {
      "epoch": 0.651484616418111,
      "grad_norm": 2.5807552337646484,
      "learning_rate": 1.766308048481547e-05,
      "loss": 4.1074,
      "step": 24245
    },
    {
      "epoch": 0.6516189708450894,
      "grad_norm": 2.642622470855713,
      "learning_rate": 1.7656271278768897e-05,
      "loss": 4.0736,
      "step": 24250
    },
    {
      "epoch": 0.6517533252720678,
      "grad_norm": 2.3323090076446533,
      "learning_rate": 1.7649462072722322e-05,
      "loss": 4.1872,
      "step": 24255
    },
    {
      "epoch": 0.651887679699046,
      "grad_norm": 2.597944974899292,
      "learning_rate": 1.7642652866675745e-05,
      "loss": 4.1838,
      "step": 24260
    },
    {
      "epoch": 0.6520220341260244,
      "grad_norm": 2.5142457485198975,
      "learning_rate": 1.7635843660629174e-05,
      "loss": 4.0876,
      "step": 24265
    },
    {
      "epoch": 0.6521563885530028,
      "grad_norm": 2.6136395931243896,
      "learning_rate": 1.7629034454582596e-05,
      "loss": 4.1168,
      "step": 24270
    },
    {
      "epoch": 0.6522907429799812,
      "grad_norm": 2.5781824588775635,
      "learning_rate": 1.762222524853602e-05,
      "loss": 3.9381,
      "step": 24275
    },
    {
      "epoch": 0.6524250974069595,
      "grad_norm": 2.4265387058258057,
      "learning_rate": 1.7615416042489447e-05,
      "loss": 4.1294,
      "step": 24280
    },
    {
      "epoch": 0.6525594518339379,
      "grad_norm": 2.331796407699585,
      "learning_rate": 1.760860683644287e-05,
      "loss": 3.9434,
      "step": 24285
    },
    {
      "epoch": 0.6526938062609163,
      "grad_norm": 2.469675064086914,
      "learning_rate": 1.7601797630396295e-05,
      "loss": 4.099,
      "step": 24290
    },
    {
      "epoch": 0.6528281606878946,
      "grad_norm": 2.712191343307495,
      "learning_rate": 1.759498842434972e-05,
      "loss": 4.0513,
      "step": 24295
    },
    {
      "epoch": 0.652962515114873,
      "grad_norm": 2.5762579441070557,
      "learning_rate": 1.7588179218303147e-05,
      "loss": 4.0775,
      "step": 24300
    },
    {
      "epoch": 0.6530968695418514,
      "grad_norm": 2.7294833660125732,
      "learning_rate": 1.7581370012256572e-05,
      "loss": 4.0909,
      "step": 24305
    },
    {
      "epoch": 0.6532312239688298,
      "grad_norm": 2.6736409664154053,
      "learning_rate": 1.7574560806209998e-05,
      "loss": 4.0962,
      "step": 24310
    },
    {
      "epoch": 0.6533655783958081,
      "grad_norm": 2.439054250717163,
      "learning_rate": 1.756775160016342e-05,
      "loss": 4.2383,
      "step": 24315
    },
    {
      "epoch": 0.6534999328227865,
      "grad_norm": 2.476149559020996,
      "learning_rate": 1.7560942394116846e-05,
      "loss": 4.0377,
      "step": 24320
    },
    {
      "epoch": 0.6536342872497649,
      "grad_norm": 2.71405291557312,
      "learning_rate": 1.755413318807027e-05,
      "loss": 4.3044,
      "step": 24325
    },
    {
      "epoch": 0.6537686416767432,
      "grad_norm": 2.6060214042663574,
      "learning_rate": 1.7547323982023697e-05,
      "loss": 4.0513,
      "step": 24330
    },
    {
      "epoch": 0.6539029961037216,
      "grad_norm": 2.4815680980682373,
      "learning_rate": 1.754051477597712e-05,
      "loss": 4.1232,
      "step": 24335
    },
    {
      "epoch": 0.6540373505307,
      "grad_norm": 2.401942014694214,
      "learning_rate": 1.753370556993055e-05,
      "loss": 4.0883,
      "step": 24340
    },
    {
      "epoch": 0.6541717049576784,
      "grad_norm": 2.4696788787841797,
      "learning_rate": 1.752689636388397e-05,
      "loss": 4.0169,
      "step": 24345
    },
    {
      "epoch": 0.6543060593846567,
      "grad_norm": 2.653489112854004,
      "learning_rate": 1.7520087157837396e-05,
      "loss": 4.0838,
      "step": 24350
    },
    {
      "epoch": 0.6544404138116351,
      "grad_norm": 2.515920877456665,
      "learning_rate": 1.7513277951790822e-05,
      "loss": 4.0425,
      "step": 24355
    },
    {
      "epoch": 0.6545747682386135,
      "grad_norm": 2.6152942180633545,
      "learning_rate": 1.7506468745744248e-05,
      "loss": 4.0437,
      "step": 24360
    },
    {
      "epoch": 0.6547091226655918,
      "grad_norm": 2.244910955429077,
      "learning_rate": 1.749965953969767e-05,
      "loss": 4.0964,
      "step": 24365
    },
    {
      "epoch": 0.6548434770925702,
      "grad_norm": 2.3278908729553223,
      "learning_rate": 1.74928503336511e-05,
      "loss": 4.1439,
      "step": 24370
    },
    {
      "epoch": 0.6549778315195486,
      "grad_norm": 2.4289348125457764,
      "learning_rate": 1.748604112760452e-05,
      "loss": 4.1834,
      "step": 24375
    },
    {
      "epoch": 0.655112185946527,
      "grad_norm": 2.6016147136688232,
      "learning_rate": 1.7479231921557947e-05,
      "loss": 3.963,
      "step": 24380
    },
    {
      "epoch": 0.6552465403735053,
      "grad_norm": 2.5902950763702393,
      "learning_rate": 1.7472422715511373e-05,
      "loss": 4.1394,
      "step": 24385
    },
    {
      "epoch": 0.6553808948004837,
      "grad_norm": 2.422241449356079,
      "learning_rate": 1.7465613509464798e-05,
      "loss": 4.1666,
      "step": 24390
    },
    {
      "epoch": 0.6555152492274621,
      "grad_norm": 2.633828639984131,
      "learning_rate": 1.745880430341822e-05,
      "loss": 4.1107,
      "step": 24395
    },
    {
      "epoch": 0.6556496036544404,
      "grad_norm": 2.316654682159424,
      "learning_rate": 1.745199509737165e-05,
      "loss": 4.0794,
      "step": 24400
    },
    {
      "epoch": 0.6557839580814188,
      "grad_norm": 2.8310577869415283,
      "learning_rate": 1.7445185891325072e-05,
      "loss": 4.0834,
      "step": 24405
    },
    {
      "epoch": 0.6559183125083972,
      "grad_norm": 2.440497636795044,
      "learning_rate": 1.7438376685278497e-05,
      "loss": 4.1115,
      "step": 24410
    },
    {
      "epoch": 0.6560526669353756,
      "grad_norm": 2.45755672454834,
      "learning_rate": 1.7431567479231923e-05,
      "loss": 4.1449,
      "step": 24415
    },
    {
      "epoch": 0.6561870213623539,
      "grad_norm": 2.4841885566711426,
      "learning_rate": 1.742475827318535e-05,
      "loss": 4.057,
      "step": 24420
    },
    {
      "epoch": 0.6563213757893323,
      "grad_norm": 2.41619873046875,
      "learning_rate": 1.741794906713877e-05,
      "loss": 4.0798,
      "step": 24425
    },
    {
      "epoch": 0.6564557302163107,
      "grad_norm": 2.747192621231079,
      "learning_rate": 1.74111398610922e-05,
      "loss": 4.0493,
      "step": 24430
    },
    {
      "epoch": 0.656590084643289,
      "grad_norm": 2.42177677154541,
      "learning_rate": 1.7404330655045622e-05,
      "loss": 4.1373,
      "step": 24435
    },
    {
      "epoch": 0.6567244390702673,
      "grad_norm": 2.3373184204101562,
      "learning_rate": 1.7397521448999045e-05,
      "loss": 4.0239,
      "step": 24440
    },
    {
      "epoch": 0.6568587934972457,
      "grad_norm": 2.5453245639801025,
      "learning_rate": 1.7390712242952474e-05,
      "loss": 4.0732,
      "step": 24445
    },
    {
      "epoch": 0.6569931479242241,
      "grad_norm": 2.41300106048584,
      "learning_rate": 1.7383903036905896e-05,
      "loss": 4.0503,
      "step": 24450
    },
    {
      "epoch": 0.6571275023512024,
      "grad_norm": 2.3283960819244385,
      "learning_rate": 1.737709383085932e-05,
      "loss": 4.1827,
      "step": 24455
    },
    {
      "epoch": 0.6572618567781808,
      "grad_norm": 2.498337984085083,
      "learning_rate": 1.7370284624812747e-05,
      "loss": 4.1566,
      "step": 24460
    },
    {
      "epoch": 0.6573962112051592,
      "grad_norm": 2.5517868995666504,
      "learning_rate": 1.7363475418766173e-05,
      "loss": 4.0128,
      "step": 24465
    },
    {
      "epoch": 0.6575305656321376,
      "grad_norm": 2.4267733097076416,
      "learning_rate": 1.7356666212719595e-05,
      "loss": 4.1411,
      "step": 24470
    },
    {
      "epoch": 0.6576649200591159,
      "grad_norm": 2.5806798934936523,
      "learning_rate": 1.7349857006673024e-05,
      "loss": 4.1616,
      "step": 24475
    },
    {
      "epoch": 0.6577992744860943,
      "grad_norm": 2.52950382232666,
      "learning_rate": 1.7343047800626446e-05,
      "loss": 4.1669,
      "step": 24480
    },
    {
      "epoch": 0.6579336289130727,
      "grad_norm": 2.48915696144104,
      "learning_rate": 1.7336238594579872e-05,
      "loss": 4.0702,
      "step": 24485
    },
    {
      "epoch": 0.658067983340051,
      "grad_norm": 2.544912576675415,
      "learning_rate": 1.7329429388533298e-05,
      "loss": 4.0026,
      "step": 24490
    },
    {
      "epoch": 0.6582023377670294,
      "grad_norm": 2.432206630706787,
      "learning_rate": 1.7322620182486723e-05,
      "loss": 4.1387,
      "step": 24495
    },
    {
      "epoch": 0.6583366921940078,
      "grad_norm": 2.521678924560547,
      "learning_rate": 1.7315810976440146e-05,
      "loss": 4.1354,
      "step": 24500
    },
    {
      "epoch": 0.6584710466209862,
      "grad_norm": 2.418189525604248,
      "learning_rate": 1.7309001770393575e-05,
      "loss": 4.1564,
      "step": 24505
    },
    {
      "epoch": 0.6586054010479645,
      "grad_norm": 2.4838197231292725,
      "learning_rate": 1.7302192564346997e-05,
      "loss": 4.0486,
      "step": 24510
    },
    {
      "epoch": 0.6587397554749429,
      "grad_norm": 2.4238879680633545,
      "learning_rate": 1.7295383358300423e-05,
      "loss": 4.1166,
      "step": 24515
    },
    {
      "epoch": 0.6588741099019213,
      "grad_norm": 2.8184475898742676,
      "learning_rate": 1.728857415225385e-05,
      "loss": 4.047,
      "step": 24520
    },
    {
      "epoch": 0.6590084643288996,
      "grad_norm": 2.291536331176758,
      "learning_rate": 1.7281764946207274e-05,
      "loss": 3.9852,
      "step": 24525
    },
    {
      "epoch": 0.659142818755878,
      "grad_norm": 2.384695529937744,
      "learning_rate": 1.7274955740160696e-05,
      "loss": 4.0376,
      "step": 24530
    },
    {
      "epoch": 0.6592771731828564,
      "grad_norm": 2.104480743408203,
      "learning_rate": 1.7268146534114125e-05,
      "loss": 4.2217,
      "step": 24535
    },
    {
      "epoch": 0.6594115276098348,
      "grad_norm": 2.3060505390167236,
      "learning_rate": 1.7261337328067548e-05,
      "loss": 4.1561,
      "step": 24540
    },
    {
      "epoch": 0.6595458820368131,
      "grad_norm": 2.5552971363067627,
      "learning_rate": 1.7254528122020973e-05,
      "loss": 4.0796,
      "step": 24545
    },
    {
      "epoch": 0.6596802364637915,
      "grad_norm": 2.4885447025299072,
      "learning_rate": 1.72477189159744e-05,
      "loss": 4.1043,
      "step": 24550
    },
    {
      "epoch": 0.6598145908907699,
      "grad_norm": 2.44622802734375,
      "learning_rate": 1.7240909709927825e-05,
      "loss": 4.1431,
      "step": 24555
    },
    {
      "epoch": 0.6599489453177482,
      "grad_norm": 2.3995563983917236,
      "learning_rate": 1.7234100503881247e-05,
      "loss": 4.1439,
      "step": 24560
    },
    {
      "epoch": 0.6600832997447266,
      "grad_norm": 2.566452741622925,
      "learning_rate": 1.7227291297834676e-05,
      "loss": 4.1962,
      "step": 24565
    },
    {
      "epoch": 0.660217654171705,
      "grad_norm": 2.6441328525543213,
      "learning_rate": 1.7220482091788098e-05,
      "loss": 4.0496,
      "step": 24570
    },
    {
      "epoch": 0.6603520085986834,
      "grad_norm": 2.536961078643799,
      "learning_rate": 1.7213672885741524e-05,
      "loss": 4.135,
      "step": 24575
    },
    {
      "epoch": 0.6604863630256617,
      "grad_norm": 2.541011333465576,
      "learning_rate": 1.720686367969495e-05,
      "loss": 4.1777,
      "step": 24580
    },
    {
      "epoch": 0.6606207174526401,
      "grad_norm": 2.3747928142547607,
      "learning_rate": 1.7200054473648372e-05,
      "loss": 4.0871,
      "step": 24585
    },
    {
      "epoch": 0.6607550718796185,
      "grad_norm": 3.007676362991333,
      "learning_rate": 1.7193245267601797e-05,
      "loss": 4.0789,
      "step": 24590
    },
    {
      "epoch": 0.6608894263065968,
      "grad_norm": 2.5727133750915527,
      "learning_rate": 1.7186436061555223e-05,
      "loss": 4.1358,
      "step": 24595
    },
    {
      "epoch": 0.6610237807335752,
      "grad_norm": 2.504718542098999,
      "learning_rate": 1.717962685550865e-05,
      "loss": 4.0425,
      "step": 24600
    },
    {
      "epoch": 0.6611581351605536,
      "grad_norm": 2.3066539764404297,
      "learning_rate": 1.717281764946207e-05,
      "loss": 4.1306,
      "step": 24605
    },
    {
      "epoch": 0.661292489587532,
      "grad_norm": 2.6594419479370117,
      "learning_rate": 1.71660084434155e-05,
      "loss": 4.1269,
      "step": 24610
    },
    {
      "epoch": 0.6614268440145102,
      "grad_norm": 2.69704008102417,
      "learning_rate": 1.7159199237368922e-05,
      "loss": 4.1212,
      "step": 24615
    },
    {
      "epoch": 0.6615611984414886,
      "grad_norm": 2.783600330352783,
      "learning_rate": 1.7152390031322348e-05,
      "loss": 4.1565,
      "step": 24620
    },
    {
      "epoch": 0.661695552868467,
      "grad_norm": 2.550464630126953,
      "learning_rate": 1.7145580825275774e-05,
      "loss": 4.1148,
      "step": 24625
    },
    {
      "epoch": 0.6618299072954453,
      "grad_norm": 2.5871169567108154,
      "learning_rate": 1.71387716192292e-05,
      "loss": 4.1259,
      "step": 24630
    },
    {
      "epoch": 0.6619642617224237,
      "grad_norm": 2.4237937927246094,
      "learning_rate": 1.713196241318262e-05,
      "loss": 4.0795,
      "step": 24635
    },
    {
      "epoch": 0.6620986161494021,
      "grad_norm": 2.2721176147460938,
      "learning_rate": 1.712515320713605e-05,
      "loss": 4.0973,
      "step": 24640
    },
    {
      "epoch": 0.6622329705763805,
      "grad_norm": 2.4347729682922363,
      "learning_rate": 1.7118344001089473e-05,
      "loss": 4.1864,
      "step": 24645
    },
    {
      "epoch": 0.6623673250033588,
      "grad_norm": 2.442473888397217,
      "learning_rate": 1.71115347950429e-05,
      "loss": 3.996,
      "step": 24650
    },
    {
      "epoch": 0.6625016794303372,
      "grad_norm": 2.4915847778320312,
      "learning_rate": 1.7104725588996324e-05,
      "loss": 4.1246,
      "step": 24655
    },
    {
      "epoch": 0.6626360338573156,
      "grad_norm": 2.4784507751464844,
      "learning_rate": 1.709791638294975e-05,
      "loss": 4.1265,
      "step": 24660
    },
    {
      "epoch": 0.662770388284294,
      "grad_norm": 2.3777551651000977,
      "learning_rate": 1.7091107176903172e-05,
      "loss": 4.0704,
      "step": 24665
    },
    {
      "epoch": 0.6629047427112723,
      "grad_norm": 2.485563278198242,
      "learning_rate": 1.70842979708566e-05,
      "loss": 4.1407,
      "step": 24670
    },
    {
      "epoch": 0.6630390971382507,
      "grad_norm": 2.5457584857940674,
      "learning_rate": 1.7077488764810023e-05,
      "loss": 4.099,
      "step": 24675
    },
    {
      "epoch": 0.6631734515652291,
      "grad_norm": 2.586071729660034,
      "learning_rate": 1.707067955876345e-05,
      "loss": 4.0926,
      "step": 24680
    },
    {
      "epoch": 0.6633078059922074,
      "grad_norm": 2.3716979026794434,
      "learning_rate": 1.7063870352716875e-05,
      "loss": 4.0058,
      "step": 24685
    },
    {
      "epoch": 0.6634421604191858,
      "grad_norm": 2.5290942192077637,
      "learning_rate": 1.70570611466703e-05,
      "loss": 4.0418,
      "step": 24690
    },
    {
      "epoch": 0.6635765148461642,
      "grad_norm": 2.341618537902832,
      "learning_rate": 1.7050251940623723e-05,
      "loss": 4.1547,
      "step": 24695
    },
    {
      "epoch": 0.6637108692731426,
      "grad_norm": 2.4289629459381104,
      "learning_rate": 1.704344273457715e-05,
      "loss": 4.0125,
      "step": 24700
    },
    {
      "epoch": 0.6638452237001209,
      "grad_norm": 2.4045491218566895,
      "learning_rate": 1.7036633528530574e-05,
      "loss": 4.1559,
      "step": 24705
    },
    {
      "epoch": 0.6639795781270993,
      "grad_norm": 2.6459834575653076,
      "learning_rate": 1.7029824322484e-05,
      "loss": 3.957,
      "step": 24710
    },
    {
      "epoch": 0.6641139325540777,
      "grad_norm": 2.489624500274658,
      "learning_rate": 1.7023015116437425e-05,
      "loss": 4.0732,
      "step": 24715
    },
    {
      "epoch": 0.664248286981056,
      "grad_norm": 2.6033644676208496,
      "learning_rate": 1.701620591039085e-05,
      "loss": 3.9813,
      "step": 24720
    },
    {
      "epoch": 0.6643826414080344,
      "grad_norm": 2.848665237426758,
      "learning_rate": 1.7009396704344273e-05,
      "loss": 4.1094,
      "step": 24725
    },
    {
      "epoch": 0.6645169958350128,
      "grad_norm": 2.356055974960327,
      "learning_rate": 1.70025874982977e-05,
      "loss": 4.1987,
      "step": 24730
    },
    {
      "epoch": 0.6646513502619912,
      "grad_norm": 2.2857542037963867,
      "learning_rate": 1.6995778292251125e-05,
      "loss": 3.851,
      "step": 24735
    },
    {
      "epoch": 0.6647857046889695,
      "grad_norm": 2.4776763916015625,
      "learning_rate": 1.6988969086204547e-05,
      "loss": 4.1136,
      "step": 24740
    },
    {
      "epoch": 0.6649200591159479,
      "grad_norm": 2.8978004455566406,
      "learning_rate": 1.6982159880157976e-05,
      "loss": 4.0751,
      "step": 24745
    },
    {
      "epoch": 0.6650544135429263,
      "grad_norm": 2.4352550506591797,
      "learning_rate": 1.6975350674111398e-05,
      "loss": 4.1022,
      "step": 24750
    },
    {
      "epoch": 0.6651887679699046,
      "grad_norm": 2.5188076496124268,
      "learning_rate": 1.6968541468064824e-05,
      "loss": 4.0651,
      "step": 24755
    },
    {
      "epoch": 0.665323122396883,
      "grad_norm": 2.5174953937530518,
      "learning_rate": 1.696173226201825e-05,
      "loss": 3.9803,
      "step": 24760
    },
    {
      "epoch": 0.6654574768238614,
      "grad_norm": 2.65630841255188,
      "learning_rate": 1.6954923055971675e-05,
      "loss": 4.1051,
      "step": 24765
    },
    {
      "epoch": 0.6655918312508398,
      "grad_norm": 2.598051071166992,
      "learning_rate": 1.6948113849925097e-05,
      "loss": 4.1021,
      "step": 24770
    },
    {
      "epoch": 0.665726185677818,
      "grad_norm": 2.4839766025543213,
      "learning_rate": 1.6941304643878526e-05,
      "loss": 3.9821,
      "step": 24775
    },
    {
      "epoch": 0.6658605401047965,
      "grad_norm": 2.4451494216918945,
      "learning_rate": 1.693449543783195e-05,
      "loss": 3.9787,
      "step": 24780
    },
    {
      "epoch": 0.6659948945317749,
      "grad_norm": 2.370553731918335,
      "learning_rate": 1.6927686231785374e-05,
      "loss": 4.0361,
      "step": 24785
    },
    {
      "epoch": 0.6661292489587531,
      "grad_norm": 2.4774012565612793,
      "learning_rate": 1.69208770257388e-05,
      "loss": 4.1408,
      "step": 24790
    },
    {
      "epoch": 0.6662636033857315,
      "grad_norm": 2.44213604927063,
      "learning_rate": 1.6914067819692226e-05,
      "loss": 4.0494,
      "step": 24795
    },
    {
      "epoch": 0.6663979578127099,
      "grad_norm": 2.6451785564422607,
      "learning_rate": 1.6907258613645648e-05,
      "loss": 4.2173,
      "step": 24800
    },
    {
      "epoch": 0.6665323122396883,
      "grad_norm": 2.5545859336853027,
      "learning_rate": 1.6900449407599077e-05,
      "loss": 4.0913,
      "step": 24805
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 2.8899617195129395,
      "learning_rate": 1.68936402015525e-05,
      "loss": 3.9716,
      "step": 24810
    },
    {
      "epoch": 0.666801021093645,
      "grad_norm": 2.3707151412963867,
      "learning_rate": 1.6886830995505925e-05,
      "loss": 4.0142,
      "step": 24815
    },
    {
      "epoch": 0.6669353755206234,
      "grad_norm": 2.4498772621154785,
      "learning_rate": 1.688002178945935e-05,
      "loss": 4.1129,
      "step": 24820
    },
    {
      "epoch": 0.6670697299476017,
      "grad_norm": 2.6980397701263428,
      "learning_rate": 1.6873212583412776e-05,
      "loss": 4.1241,
      "step": 24825
    },
    {
      "epoch": 0.6672040843745801,
      "grad_norm": 2.606616735458374,
      "learning_rate": 1.68664033773662e-05,
      "loss": 4.0066,
      "step": 24830
    },
    {
      "epoch": 0.6673384388015585,
      "grad_norm": 2.4780101776123047,
      "learning_rate": 1.6859594171319628e-05,
      "loss": 4.146,
      "step": 24835
    },
    {
      "epoch": 0.6674727932285369,
      "grad_norm": 2.627702236175537,
      "learning_rate": 1.685278496527305e-05,
      "loss": 4.0169,
      "step": 24840
    },
    {
      "epoch": 0.6676071476555152,
      "grad_norm": 2.54738712310791,
      "learning_rate": 1.6845975759226475e-05,
      "loss": 4.1185,
      "step": 24845
    },
    {
      "epoch": 0.6677415020824936,
      "grad_norm": 2.6914215087890625,
      "learning_rate": 1.68391665531799e-05,
      "loss": 4.1587,
      "step": 24850
    },
    {
      "epoch": 0.667875856509472,
      "grad_norm": 2.2947144508361816,
      "learning_rate": 1.6832357347133327e-05,
      "loss": 4.1164,
      "step": 24855
    },
    {
      "epoch": 0.6680102109364504,
      "grad_norm": 2.473254680633545,
      "learning_rate": 1.682554814108675e-05,
      "loss": 3.9639,
      "step": 24860
    },
    {
      "epoch": 0.6681445653634287,
      "grad_norm": 2.7933645248413086,
      "learning_rate": 1.6818738935040178e-05,
      "loss": 4.0223,
      "step": 24865
    },
    {
      "epoch": 0.6682789197904071,
      "grad_norm": 2.551804542541504,
      "learning_rate": 1.68119297289936e-05,
      "loss": 4.1079,
      "step": 24870
    },
    {
      "epoch": 0.6684132742173855,
      "grad_norm": 2.2804107666015625,
      "learning_rate": 1.6805120522947026e-05,
      "loss": 4.1846,
      "step": 24875
    },
    {
      "epoch": 0.6685476286443638,
      "grad_norm": 2.984001636505127,
      "learning_rate": 1.679831131690045e-05,
      "loss": 4.0941,
      "step": 24880
    },
    {
      "epoch": 0.6686819830713422,
      "grad_norm": 2.3935153484344482,
      "learning_rate": 1.6791502110853874e-05,
      "loss": 4.09,
      "step": 24885
    },
    {
      "epoch": 0.6688163374983206,
      "grad_norm": 2.5095157623291016,
      "learning_rate": 1.67846929048073e-05,
      "loss": 4.1348,
      "step": 24890
    },
    {
      "epoch": 0.668950691925299,
      "grad_norm": 2.4136135578155518,
      "learning_rate": 1.6777883698760725e-05,
      "loss": 4.1284,
      "step": 24895
    },
    {
      "epoch": 0.6690850463522773,
      "grad_norm": 2.368994951248169,
      "learning_rate": 1.677107449271415e-05,
      "loss": 4.1264,
      "step": 24900
    },
    {
      "epoch": 0.6692194007792557,
      "grad_norm": 2.667616367340088,
      "learning_rate": 1.6764265286667573e-05,
      "loss": 4.0379,
      "step": 24905
    },
    {
      "epoch": 0.6693537552062341,
      "grad_norm": 2.409773349761963,
      "learning_rate": 1.6757456080621002e-05,
      "loss": 3.9984,
      "step": 24910
    },
    {
      "epoch": 0.6694881096332124,
      "grad_norm": 2.5774850845336914,
      "learning_rate": 1.6750646874574424e-05,
      "loss": 4.0731,
      "step": 24915
    },
    {
      "epoch": 0.6696224640601908,
      "grad_norm": 2.4436726570129395,
      "learning_rate": 1.674383766852785e-05,
      "loss": 4.1422,
      "step": 24920
    },
    {
      "epoch": 0.6697568184871692,
      "grad_norm": 2.680602788925171,
      "learning_rate": 1.6737028462481276e-05,
      "loss": 4.2136,
      "step": 24925
    },
    {
      "epoch": 0.6698911729141476,
      "grad_norm": 2.725328207015991,
      "learning_rate": 1.67302192564347e-05,
      "loss": 4.1725,
      "step": 24930
    },
    {
      "epoch": 0.6700255273411259,
      "grad_norm": 2.410062789916992,
      "learning_rate": 1.6723410050388124e-05,
      "loss": 4.0956,
      "step": 24935
    },
    {
      "epoch": 0.6701598817681043,
      "grad_norm": 3.037128210067749,
      "learning_rate": 1.6716600844341553e-05,
      "loss": 4.1038,
      "step": 24940
    },
    {
      "epoch": 0.6702942361950827,
      "grad_norm": 2.517120361328125,
      "learning_rate": 1.6709791638294975e-05,
      "loss": 4.1687,
      "step": 24945
    },
    {
      "epoch": 0.670428590622061,
      "grad_norm": 2.439589262008667,
      "learning_rate": 1.67029824322484e-05,
      "loss": 4.1117,
      "step": 24950
    },
    {
      "epoch": 0.6705629450490393,
      "grad_norm": 2.5008647441864014,
      "learning_rate": 1.6696173226201826e-05,
      "loss": 4.1593,
      "step": 24955
    },
    {
      "epoch": 0.6706972994760177,
      "grad_norm": 2.4022274017333984,
      "learning_rate": 1.6689364020155252e-05,
      "loss": 4.1285,
      "step": 24960
    },
    {
      "epoch": 0.6708316539029962,
      "grad_norm": 2.344142436981201,
      "learning_rate": 1.6682554814108674e-05,
      "loss": 4.0911,
      "step": 24965
    },
    {
      "epoch": 0.6709660083299744,
      "grad_norm": 2.4908666610717773,
      "learning_rate": 1.6675745608062103e-05,
      "loss": 4.1673,
      "step": 24970
    },
    {
      "epoch": 0.6711003627569528,
      "grad_norm": 2.468057632446289,
      "learning_rate": 1.6668936402015526e-05,
      "loss": 4.1656,
      "step": 24975
    },
    {
      "epoch": 0.6712347171839312,
      "grad_norm": 2.608591318130493,
      "learning_rate": 1.666212719596895e-05,
      "loss": 4.1317,
      "step": 24980
    },
    {
      "epoch": 0.6713690716109095,
      "grad_norm": 2.7042131423950195,
      "learning_rate": 1.6655317989922377e-05,
      "loss": 4.0841,
      "step": 24985
    },
    {
      "epoch": 0.6715034260378879,
      "grad_norm": 2.3576300144195557,
      "learning_rate": 1.6648508783875803e-05,
      "loss": 3.9972,
      "step": 24990
    },
    {
      "epoch": 0.6716377804648663,
      "grad_norm": 2.3807365894317627,
      "learning_rate": 1.6641699577829225e-05,
      "loss": 4.0903,
      "step": 24995
    },
    {
      "epoch": 0.6717721348918447,
      "grad_norm": 2.4945669174194336,
      "learning_rate": 1.663489037178265e-05,
      "loss": 4.1182,
      "step": 25000
    },
    {
      "epoch": 0.671906489318823,
      "grad_norm": 2.506798028945923,
      "learning_rate": 1.6628081165736076e-05,
      "loss": 4.0514,
      "step": 25005
    },
    {
      "epoch": 0.6720408437458014,
      "grad_norm": 2.511164903640747,
      "learning_rate": 1.6621271959689502e-05,
      "loss": 4.1195,
      "step": 25010
    },
    {
      "epoch": 0.6721751981727798,
      "grad_norm": 2.6371006965637207,
      "learning_rate": 1.6614462753642927e-05,
      "loss": 4.1205,
      "step": 25015
    },
    {
      "epoch": 0.6723095525997581,
      "grad_norm": 2.484762191772461,
      "learning_rate": 1.6607653547596353e-05,
      "loss": 4.0628,
      "step": 25020
    },
    {
      "epoch": 0.6724439070267365,
      "grad_norm": 2.40940260887146,
      "learning_rate": 1.6600844341549775e-05,
      "loss": 4.053,
      "step": 25025
    },
    {
      "epoch": 0.6725782614537149,
      "grad_norm": 2.408735752105713,
      "learning_rate": 1.65940351355032e-05,
      "loss": 3.9809,
      "step": 25030
    },
    {
      "epoch": 0.6727126158806933,
      "grad_norm": 2.540616989135742,
      "learning_rate": 1.6587225929456627e-05,
      "loss": 4.1174,
      "step": 25035
    },
    {
      "epoch": 0.6728469703076716,
      "grad_norm": 2.550203800201416,
      "learning_rate": 1.658041672341005e-05,
      "loss": 4.0909,
      "step": 25040
    },
    {
      "epoch": 0.67298132473465,
      "grad_norm": 2.496736764907837,
      "learning_rate": 1.6573607517363478e-05,
      "loss": 4.1767,
      "step": 25045
    },
    {
      "epoch": 0.6731156791616284,
      "grad_norm": 2.5579285621643066,
      "learning_rate": 1.65667983113169e-05,
      "loss": 4.1306,
      "step": 25050
    },
    {
      "epoch": 0.6732500335886068,
      "grad_norm": 2.306535482406616,
      "learning_rate": 1.6559989105270326e-05,
      "loss": 4.0516,
      "step": 25055
    },
    {
      "epoch": 0.6733843880155851,
      "grad_norm": 2.364377498626709,
      "learning_rate": 1.655317989922375e-05,
      "loss": 4.1486,
      "step": 25060
    },
    {
      "epoch": 0.6735187424425635,
      "grad_norm": 2.5267958641052246,
      "learning_rate": 1.6546370693177177e-05,
      "loss": 4.0085,
      "step": 25065
    },
    {
      "epoch": 0.6736530968695419,
      "grad_norm": 2.5147600173950195,
      "learning_rate": 1.65395614871306e-05,
      "loss": 4.028,
      "step": 25070
    },
    {
      "epoch": 0.6737874512965202,
      "grad_norm": 2.49813175201416,
      "learning_rate": 1.6532752281084025e-05,
      "loss": 4.1508,
      "step": 25075
    },
    {
      "epoch": 0.6739218057234986,
      "grad_norm": 2.7407331466674805,
      "learning_rate": 1.652594307503745e-05,
      "loss": 4.0682,
      "step": 25080
    },
    {
      "epoch": 0.674056160150477,
      "grad_norm": 2.3560874462127686,
      "learning_rate": 1.6519133868990876e-05,
      "loss": 4.1007,
      "step": 25085
    },
    {
      "epoch": 0.6741905145774554,
      "grad_norm": 2.745978593826294,
      "learning_rate": 1.6512324662944302e-05,
      "loss": 4.1509,
      "step": 25090
    },
    {
      "epoch": 0.6743248690044337,
      "grad_norm": 2.433622360229492,
      "learning_rate": 1.6505515456897728e-05,
      "loss": 4.0509,
      "step": 25095
    },
    {
      "epoch": 0.6744592234314121,
      "grad_norm": 2.4271531105041504,
      "learning_rate": 1.649870625085115e-05,
      "loss": 4.1035,
      "step": 25100
    },
    {
      "epoch": 0.6745935778583905,
      "grad_norm": 2.503331422805786,
      "learning_rate": 1.6491897044804576e-05,
      "loss": 4.083,
      "step": 25105
    },
    {
      "epoch": 0.6747279322853688,
      "grad_norm": 2.6872217655181885,
      "learning_rate": 1.6485087838758e-05,
      "loss": 4.1697,
      "step": 25110
    },
    {
      "epoch": 0.6748622867123472,
      "grad_norm": 2.4616341590881348,
      "learning_rate": 1.6478278632711427e-05,
      "loss": 3.9935,
      "step": 25115
    },
    {
      "epoch": 0.6749966411393256,
      "grad_norm": 2.5138587951660156,
      "learning_rate": 1.6471469426664853e-05,
      "loss": 4.0369,
      "step": 25120
    },
    {
      "epoch": 0.675130995566304,
      "grad_norm": 2.6759262084960938,
      "learning_rate": 1.646466022061828e-05,
      "loss": 4.195,
      "step": 25125
    },
    {
      "epoch": 0.6752653499932822,
      "grad_norm": 2.4312796592712402,
      "learning_rate": 1.64578510145717e-05,
      "loss": 4.0867,
      "step": 25130
    },
    {
      "epoch": 0.6753997044202606,
      "grad_norm": 2.4523637294769287,
      "learning_rate": 1.6451041808525126e-05,
      "loss": 4.0974,
      "step": 25135
    },
    {
      "epoch": 0.675534058847239,
      "grad_norm": 2.490473508834839,
      "learning_rate": 1.6444232602478552e-05,
      "loss": 4.0327,
      "step": 25140
    },
    {
      "epoch": 0.6756684132742173,
      "grad_norm": 2.505502462387085,
      "learning_rate": 1.6437423396431978e-05,
      "loss": 4.1572,
      "step": 25145
    },
    {
      "epoch": 0.6758027677011957,
      "grad_norm": 2.736435651779175,
      "learning_rate": 1.6430614190385403e-05,
      "loss": 4.1907,
      "step": 25150
    },
    {
      "epoch": 0.6759371221281741,
      "grad_norm": 2.2636497020721436,
      "learning_rate": 1.642380498433883e-05,
      "loss": 4.0376,
      "step": 25155
    },
    {
      "epoch": 0.6760714765551525,
      "grad_norm": 2.6706228256225586,
      "learning_rate": 1.641699577829225e-05,
      "loss": 4.0184,
      "step": 25160
    },
    {
      "epoch": 0.6762058309821308,
      "grad_norm": 2.5945546627044678,
      "learning_rate": 1.6410186572245677e-05,
      "loss": 4.0317,
      "step": 25165
    },
    {
      "epoch": 0.6763401854091092,
      "grad_norm": 2.7675836086273193,
      "learning_rate": 1.6403377366199103e-05,
      "loss": 4.1769,
      "step": 25170
    },
    {
      "epoch": 0.6764745398360876,
      "grad_norm": 2.761752128601074,
      "learning_rate": 1.6396568160152525e-05,
      "loss": 4.1013,
      "step": 25175
    },
    {
      "epoch": 0.6766088942630659,
      "grad_norm": 2.3509011268615723,
      "learning_rate": 1.638975895410595e-05,
      "loss": 4.0968,
      "step": 25180
    },
    {
      "epoch": 0.6767432486900443,
      "grad_norm": 2.4771106243133545,
      "learning_rate": 1.6382949748059376e-05,
      "loss": 3.9593,
      "step": 25185
    },
    {
      "epoch": 0.6768776031170227,
      "grad_norm": 2.4847488403320312,
      "learning_rate": 1.6376140542012802e-05,
      "loss": 4.1489,
      "step": 25190
    },
    {
      "epoch": 0.6770119575440011,
      "grad_norm": 2.6947457790374756,
      "learning_rate": 1.6369331335966227e-05,
      "loss": 4.0419,
      "step": 25195
    },
    {
      "epoch": 0.6771463119709794,
      "grad_norm": 2.3962581157684326,
      "learning_rate": 1.6362522129919653e-05,
      "loss": 4.1903,
      "step": 25200
    },
    {
      "epoch": 0.6772806663979578,
      "grad_norm": 2.4352054595947266,
      "learning_rate": 1.6355712923873075e-05,
      "loss": 4.0537,
      "step": 25205
    },
    {
      "epoch": 0.6774150208249362,
      "grad_norm": 2.3407158851623535,
      "learning_rate": 1.63489037178265e-05,
      "loss": 4.081,
      "step": 25210
    },
    {
      "epoch": 0.6775493752519145,
      "grad_norm": 2.4292871952056885,
      "learning_rate": 1.6342094511779927e-05,
      "loss": 3.9978,
      "step": 25215
    },
    {
      "epoch": 0.6776837296788929,
      "grad_norm": 2.278900384902954,
      "learning_rate": 1.6335285305733352e-05,
      "loss": 4.1258,
      "step": 25220
    },
    {
      "epoch": 0.6778180841058713,
      "grad_norm": 2.290712833404541,
      "learning_rate": 1.6328476099686778e-05,
      "loss": 4.0273,
      "step": 25225
    },
    {
      "epoch": 0.6779524385328497,
      "grad_norm": 2.54319167137146,
      "learning_rate": 1.6321666893640204e-05,
      "loss": 3.9982,
      "step": 25230
    },
    {
      "epoch": 0.678086792959828,
      "grad_norm": 2.1533541679382324,
      "learning_rate": 1.6314857687593626e-05,
      "loss": 4.071,
      "step": 25235
    },
    {
      "epoch": 0.6782211473868064,
      "grad_norm": 2.38978910446167,
      "learning_rate": 1.630804848154705e-05,
      "loss": 3.9851,
      "step": 25240
    },
    {
      "epoch": 0.6783555018137848,
      "grad_norm": 2.516136884689331,
      "learning_rate": 1.6301239275500477e-05,
      "loss": 3.9925,
      "step": 25245
    },
    {
      "epoch": 0.6784898562407631,
      "grad_norm": 2.4939894676208496,
      "learning_rate": 1.6294430069453903e-05,
      "loss": 4.0702,
      "step": 25250
    },
    {
      "epoch": 0.6786242106677415,
      "grad_norm": 2.3189172744750977,
      "learning_rate": 1.628762086340733e-05,
      "loss": 3.9893,
      "step": 25255
    },
    {
      "epoch": 0.6787585650947199,
      "grad_norm": 2.460808753967285,
      "learning_rate": 1.6280811657360754e-05,
      "loss": 4.168,
      "step": 25260
    },
    {
      "epoch": 0.6788929195216983,
      "grad_norm": 2.590711832046509,
      "learning_rate": 1.6274002451314176e-05,
      "loss": 4.0801,
      "step": 25265
    },
    {
      "epoch": 0.6790272739486766,
      "grad_norm": 2.290926694869995,
      "learning_rate": 1.6267193245267602e-05,
      "loss": 4.1355,
      "step": 25270
    },
    {
      "epoch": 0.679161628375655,
      "grad_norm": 2.607848882675171,
      "learning_rate": 1.6260384039221028e-05,
      "loss": 4.1542,
      "step": 25275
    },
    {
      "epoch": 0.6792959828026334,
      "grad_norm": 2.6196579933166504,
      "learning_rate": 1.6253574833174453e-05,
      "loss": 4.2005,
      "step": 25280
    },
    {
      "epoch": 0.6794303372296118,
      "grad_norm": 2.4785892963409424,
      "learning_rate": 1.6246765627127876e-05,
      "loss": 4.1412,
      "step": 25285
    },
    {
      "epoch": 0.6795646916565901,
      "grad_norm": 2.637890338897705,
      "learning_rate": 1.6239956421081305e-05,
      "loss": 3.8874,
      "step": 25290
    },
    {
      "epoch": 0.6796990460835685,
      "grad_norm": 2.7890937328338623,
      "learning_rate": 1.6233147215034727e-05,
      "loss": 4.1866,
      "step": 25295
    },
    {
      "epoch": 0.6798334005105469,
      "grad_norm": 2.3304672241210938,
      "learning_rate": 1.6226338008988153e-05,
      "loss": 4.0755,
      "step": 25300
    },
    {
      "epoch": 0.6799677549375251,
      "grad_norm": 2.596991777420044,
      "learning_rate": 1.621952880294158e-05,
      "loss": 4.1018,
      "step": 25305
    },
    {
      "epoch": 0.6801021093645035,
      "grad_norm": 2.4190316200256348,
      "learning_rate": 1.6212719596895004e-05,
      "loss": 4.1101,
      "step": 25310
    },
    {
      "epoch": 0.680236463791482,
      "grad_norm": 2.4319753646850586,
      "learning_rate": 1.6205910390848426e-05,
      "loss": 4.1525,
      "step": 25315
    },
    {
      "epoch": 0.6803708182184603,
      "grad_norm": 2.641751766204834,
      "learning_rate": 1.6199101184801855e-05,
      "loss": 4.0055,
      "step": 25320
    },
    {
      "epoch": 0.6805051726454386,
      "grad_norm": 2.360935926437378,
      "learning_rate": 1.6192291978755278e-05,
      "loss": 4.2018,
      "step": 25325
    },
    {
      "epoch": 0.680639527072417,
      "grad_norm": 2.505279302597046,
      "learning_rate": 1.6185482772708703e-05,
      "loss": 4.0116,
      "step": 25330
    },
    {
      "epoch": 0.6807738814993954,
      "grad_norm": 2.4192252159118652,
      "learning_rate": 1.617867356666213e-05,
      "loss": 4.092,
      "step": 25335
    },
    {
      "epoch": 0.6809082359263737,
      "grad_norm": 2.542574167251587,
      "learning_rate": 1.617186436061555e-05,
      "loss": 4.0643,
      "step": 25340
    },
    {
      "epoch": 0.6810425903533521,
      "grad_norm": 2.599747896194458,
      "learning_rate": 1.6165055154568977e-05,
      "loss": 4.0905,
      "step": 25345
    },
    {
      "epoch": 0.6811769447803305,
      "grad_norm": 2.850055694580078,
      "learning_rate": 1.6158245948522402e-05,
      "loss": 4.0939,
      "step": 25350
    },
    {
      "epoch": 0.6813112992073089,
      "grad_norm": 2.4007153511047363,
      "learning_rate": 1.6151436742475828e-05,
      "loss": 4.1392,
      "step": 25355
    },
    {
      "epoch": 0.6814456536342872,
      "grad_norm": 2.3454623222351074,
      "learning_rate": 1.6144627536429254e-05,
      "loss": 4.0838,
      "step": 25360
    },
    {
      "epoch": 0.6815800080612656,
      "grad_norm": 2.3933870792388916,
      "learning_rate": 1.613781833038268e-05,
      "loss": 4.0893,
      "step": 25365
    },
    {
      "epoch": 0.681714362488244,
      "grad_norm": 2.5503878593444824,
      "learning_rate": 1.61310091243361e-05,
      "loss": 4.0006,
      "step": 25370
    },
    {
      "epoch": 0.6818487169152223,
      "grad_norm": 2.646012783050537,
      "learning_rate": 1.6124199918289527e-05,
      "loss": 4.0072,
      "step": 25375
    },
    {
      "epoch": 0.6819830713422007,
      "grad_norm": 2.5767366886138916,
      "learning_rate": 1.6117390712242953e-05,
      "loss": 4.085,
      "step": 25380
    },
    {
      "epoch": 0.6821174257691791,
      "grad_norm": 2.274829864501953,
      "learning_rate": 1.611058150619638e-05,
      "loss": 4.0885,
      "step": 25385
    },
    {
      "epoch": 0.6822517801961575,
      "grad_norm": 2.2980470657348633,
      "learning_rate": 1.61037723001498e-05,
      "loss": 3.99,
      "step": 25390
    },
    {
      "epoch": 0.6823861346231358,
      "grad_norm": 2.6445915699005127,
      "learning_rate": 1.609696309410323e-05,
      "loss": 4.1387,
      "step": 25395
    },
    {
      "epoch": 0.6825204890501142,
      "grad_norm": 2.5391974449157715,
      "learning_rate": 1.6090153888056652e-05,
      "loss": 4.084,
      "step": 25400
    },
    {
      "epoch": 0.6826548434770926,
      "grad_norm": 2.553640604019165,
      "learning_rate": 1.6083344682010078e-05,
      "loss": 4.1359,
      "step": 25405
    },
    {
      "epoch": 0.6827891979040709,
      "grad_norm": 2.5445332527160645,
      "learning_rate": 1.6076535475963504e-05,
      "loss": 4.1411,
      "step": 25410
    },
    {
      "epoch": 0.6829235523310493,
      "grad_norm": 4.114529609680176,
      "learning_rate": 1.606972626991693e-05,
      "loss": 4.0377,
      "step": 25415
    },
    {
      "epoch": 0.6830579067580277,
      "grad_norm": 2.8669817447662354,
      "learning_rate": 1.606291706387035e-05,
      "loss": 4.1125,
      "step": 25420
    },
    {
      "epoch": 0.6831922611850061,
      "grad_norm": 2.3865182399749756,
      "learning_rate": 1.605610785782378e-05,
      "loss": 4.095,
      "step": 25425
    },
    {
      "epoch": 0.6833266156119844,
      "grad_norm": 2.2028820514678955,
      "learning_rate": 1.6049298651777203e-05,
      "loss": 4.0908,
      "step": 25430
    },
    {
      "epoch": 0.6834609700389628,
      "grad_norm": 2.378664016723633,
      "learning_rate": 1.604248944573063e-05,
      "loss": 4.0043,
      "step": 25435
    },
    {
      "epoch": 0.6835953244659412,
      "grad_norm": 2.6211299896240234,
      "learning_rate": 1.6035680239684054e-05,
      "loss": 4.0018,
      "step": 25440
    },
    {
      "epoch": 0.6837296788929195,
      "grad_norm": 2.7178854942321777,
      "learning_rate": 1.602887103363748e-05,
      "loss": 4.21,
      "step": 25445
    },
    {
      "epoch": 0.6838640333198979,
      "grad_norm": 2.5928947925567627,
      "learning_rate": 1.6022061827590902e-05,
      "loss": 4.0779,
      "step": 25450
    },
    {
      "epoch": 0.6839983877468763,
      "grad_norm": 2.3193347454071045,
      "learning_rate": 1.601525262154433e-05,
      "loss": 4.0455,
      "step": 25455
    },
    {
      "epoch": 0.6841327421738547,
      "grad_norm": 2.448166847229004,
      "learning_rate": 1.6008443415497753e-05,
      "loss": 4.1464,
      "step": 25460
    },
    {
      "epoch": 0.684267096600833,
      "grad_norm": 2.6618316173553467,
      "learning_rate": 1.600163420945118e-05,
      "loss": 4.1221,
      "step": 25465
    },
    {
      "epoch": 0.6844014510278114,
      "grad_norm": 2.28771710395813,
      "learning_rate": 1.5994825003404605e-05,
      "loss": 3.9775,
      "step": 25470
    },
    {
      "epoch": 0.6845358054547898,
      "grad_norm": 2.210848093032837,
      "learning_rate": 1.5988015797358027e-05,
      "loss": 4.0862,
      "step": 25475
    },
    {
      "epoch": 0.6846701598817682,
      "grad_norm": 2.332528591156006,
      "learning_rate": 1.5981206591311453e-05,
      "loss": 4.241,
      "step": 25480
    },
    {
      "epoch": 0.6848045143087464,
      "grad_norm": 2.4872846603393555,
      "learning_rate": 1.5974397385264878e-05,
      "loss": 4.059,
      "step": 25485
    },
    {
      "epoch": 0.6849388687357248,
      "grad_norm": 2.25761079788208,
      "learning_rate": 1.5967588179218304e-05,
      "loss": 3.96,
      "step": 25490
    },
    {
      "epoch": 0.6850732231627032,
      "grad_norm": 2.8823843002319336,
      "learning_rate": 1.5960778973171726e-05,
      "loss": 4.1613,
      "step": 25495
    },
    {
      "epoch": 0.6852075775896815,
      "grad_norm": 2.3922770023345947,
      "learning_rate": 1.5953969767125155e-05,
      "loss": 4.1417,
      "step": 25500
    },
    {
      "epoch": 0.6853419320166599,
      "grad_norm": 2.8931331634521484,
      "learning_rate": 1.5947160561078577e-05,
      "loss": 4.0753,
      "step": 25505
    },
    {
      "epoch": 0.6854762864436383,
      "grad_norm": 2.5333609580993652,
      "learning_rate": 1.5940351355032003e-05,
      "loss": 4.2131,
      "step": 25510
    },
    {
      "epoch": 0.6856106408706167,
      "grad_norm": 2.664149045944214,
      "learning_rate": 1.593354214898543e-05,
      "loss": 4.0082,
      "step": 25515
    },
    {
      "epoch": 0.685744995297595,
      "grad_norm": 2.6572744846343994,
      "learning_rate": 1.5926732942938854e-05,
      "loss": 4.1089,
      "step": 25520
    },
    {
      "epoch": 0.6858793497245734,
      "grad_norm": 2.384441614151001,
      "learning_rate": 1.5919923736892277e-05,
      "loss": 3.9885,
      "step": 25525
    },
    {
      "epoch": 0.6860137041515518,
      "grad_norm": 2.392374038696289,
      "learning_rate": 1.5913114530845706e-05,
      "loss": 4.0874,
      "step": 25530
    },
    {
      "epoch": 0.6861480585785301,
      "grad_norm": 2.5323967933654785,
      "learning_rate": 1.5906305324799128e-05,
      "loss": 4.1106,
      "step": 25535
    },
    {
      "epoch": 0.6862824130055085,
      "grad_norm": 2.487945318222046,
      "learning_rate": 1.5899496118752554e-05,
      "loss": 4.0933,
      "step": 25540
    },
    {
      "epoch": 0.6864167674324869,
      "grad_norm": 2.9004316329956055,
      "learning_rate": 1.589268691270598e-05,
      "loss": 4.1185,
      "step": 25545
    },
    {
      "epoch": 0.6865511218594653,
      "grad_norm": 2.3116226196289062,
      "learning_rate": 1.5885877706659405e-05,
      "loss": 4.1599,
      "step": 25550
    },
    {
      "epoch": 0.6866854762864436,
      "grad_norm": 2.4261958599090576,
      "learning_rate": 1.5879068500612827e-05,
      "loss": 3.9264,
      "step": 25555
    },
    {
      "epoch": 0.686819830713422,
      "grad_norm": 2.582418918609619,
      "learning_rate": 1.5872259294566256e-05,
      "loss": 4.08,
      "step": 25560
    },
    {
      "epoch": 0.6869541851404004,
      "grad_norm": 2.6855051517486572,
      "learning_rate": 1.586545008851968e-05,
      "loss": 4.0962,
      "step": 25565
    },
    {
      "epoch": 0.6870885395673787,
      "grad_norm": 2.6909902095794678,
      "learning_rate": 1.5858640882473104e-05,
      "loss": 4.1101,
      "step": 25570
    },
    {
      "epoch": 0.6872228939943571,
      "grad_norm": 2.3775010108947754,
      "learning_rate": 1.585183167642653e-05,
      "loss": 4.1017,
      "step": 25575
    },
    {
      "epoch": 0.6873572484213355,
      "grad_norm": 2.6080777645111084,
      "learning_rate": 1.5845022470379956e-05,
      "loss": 4.2074,
      "step": 25580
    },
    {
      "epoch": 0.6874916028483139,
      "grad_norm": 2.846959114074707,
      "learning_rate": 1.5838213264333378e-05,
      "loss": 4.0835,
      "step": 25585
    },
    {
      "epoch": 0.6876259572752922,
      "grad_norm": 2.5025112628936768,
      "learning_rate": 1.5831404058286807e-05,
      "loss": 4.1205,
      "step": 25590
    },
    {
      "epoch": 0.6877603117022706,
      "grad_norm": 2.6334714889526367,
      "learning_rate": 1.582459485224023e-05,
      "loss": 4.2026,
      "step": 25595
    },
    {
      "epoch": 0.687894666129249,
      "grad_norm": 2.3208086490631104,
      "learning_rate": 1.5817785646193655e-05,
      "loss": 4.1398,
      "step": 25600
    },
    {
      "epoch": 0.6880290205562273,
      "grad_norm": 2.4433276653289795,
      "learning_rate": 1.581097644014708e-05,
      "loss": 4.0472,
      "step": 25605
    },
    {
      "epoch": 0.6881633749832057,
      "grad_norm": 2.306164503097534,
      "learning_rate": 1.5804167234100506e-05,
      "loss": 4.065,
      "step": 25610
    },
    {
      "epoch": 0.6882977294101841,
      "grad_norm": 2.7130324840545654,
      "learning_rate": 1.579735802805393e-05,
      "loss": 4.2022,
      "step": 25615
    },
    {
      "epoch": 0.6884320838371625,
      "grad_norm": 2.398503303527832,
      "learning_rate": 1.5790548822007357e-05,
      "loss": 4.102,
      "step": 25620
    },
    {
      "epoch": 0.6885664382641408,
      "grad_norm": 2.4207873344421387,
      "learning_rate": 1.578373961596078e-05,
      "loss": 4.0968,
      "step": 25625
    },
    {
      "epoch": 0.6887007926911192,
      "grad_norm": 2.4732120037078857,
      "learning_rate": 1.5776930409914202e-05,
      "loss": 4.1201,
      "step": 25630
    },
    {
      "epoch": 0.6888351471180976,
      "grad_norm": 2.6079118251800537,
      "learning_rate": 1.577012120386763e-05,
      "loss": 4.1456,
      "step": 25635
    },
    {
      "epoch": 0.6889695015450759,
      "grad_norm": 2.6444857120513916,
      "learning_rate": 1.5763311997821053e-05,
      "loss": 4.1871,
      "step": 25640
    },
    {
      "epoch": 0.6891038559720543,
      "grad_norm": 2.657609224319458,
      "learning_rate": 1.575650279177448e-05,
      "loss": 3.9731,
      "step": 25645
    },
    {
      "epoch": 0.6892382103990327,
      "grad_norm": 2.6098034381866455,
      "learning_rate": 1.5749693585727905e-05,
      "loss": 4.099,
      "step": 25650
    },
    {
      "epoch": 0.6893725648260111,
      "grad_norm": 2.384411096572876,
      "learning_rate": 1.574288437968133e-05,
      "loss": 4.0822,
      "step": 25655
    },
    {
      "epoch": 0.6895069192529893,
      "grad_norm": 2.4045567512512207,
      "learning_rate": 1.5736075173634753e-05,
      "loss": 3.9606,
      "step": 25660
    },
    {
      "epoch": 0.6896412736799677,
      "grad_norm": 2.6607918739318848,
      "learning_rate": 1.572926596758818e-05,
      "loss": 4.061,
      "step": 25665
    },
    {
      "epoch": 0.6897756281069461,
      "grad_norm": 2.9390389919281006,
      "learning_rate": 1.5722456761541604e-05,
      "loss": 4.0317,
      "step": 25670
    },
    {
      "epoch": 0.6899099825339244,
      "grad_norm": 2.4175853729248047,
      "learning_rate": 1.571564755549503e-05,
      "loss": 3.949,
      "step": 25675
    },
    {
      "epoch": 0.6900443369609028,
      "grad_norm": 2.511821746826172,
      "learning_rate": 1.5708838349448455e-05,
      "loss": 4.0159,
      "step": 25680
    },
    {
      "epoch": 0.6901786913878812,
      "grad_norm": 2.6313982009887695,
      "learning_rate": 1.570202914340188e-05,
      "loss": 3.9743,
      "step": 25685
    },
    {
      "epoch": 0.6903130458148596,
      "grad_norm": 2.465038537979126,
      "learning_rate": 1.5695219937355303e-05,
      "loss": 4.0921,
      "step": 25690
    },
    {
      "epoch": 0.6904474002418379,
      "grad_norm": 2.6531152725219727,
      "learning_rate": 1.5688410731308732e-05,
      "loss": 3.9834,
      "step": 25695
    },
    {
      "epoch": 0.6905817546688163,
      "grad_norm": 2.4204304218292236,
      "learning_rate": 1.5681601525262154e-05,
      "loss": 4.1214,
      "step": 25700
    },
    {
      "epoch": 0.6907161090957947,
      "grad_norm": 2.6551647186279297,
      "learning_rate": 1.567479231921558e-05,
      "loss": 4.1632,
      "step": 25705
    },
    {
      "epoch": 0.6908504635227731,
      "grad_norm": 2.617694139480591,
      "learning_rate": 1.5667983113169006e-05,
      "loss": 4.0901,
      "step": 25710
    },
    {
      "epoch": 0.6909848179497514,
      "grad_norm": 2.4258627891540527,
      "learning_rate": 1.566117390712243e-05,
      "loss": 4.1443,
      "step": 25715
    },
    {
      "epoch": 0.6911191723767298,
      "grad_norm": 2.5481674671173096,
      "learning_rate": 1.5654364701075854e-05,
      "loss": 4.0165,
      "step": 25720
    },
    {
      "epoch": 0.6912535268037082,
      "grad_norm": 2.423997640609741,
      "learning_rate": 1.5647555495029283e-05,
      "loss": 3.9961,
      "step": 25725
    },
    {
      "epoch": 0.6913878812306865,
      "grad_norm": 2.5215060710906982,
      "learning_rate": 1.5640746288982705e-05,
      "loss": 4.18,
      "step": 25730
    },
    {
      "epoch": 0.6915222356576649,
      "grad_norm": 2.3623666763305664,
      "learning_rate": 1.563393708293613e-05,
      "loss": 3.9764,
      "step": 25735
    },
    {
      "epoch": 0.6916565900846433,
      "grad_norm": 2.2654683589935303,
      "learning_rate": 1.5627127876889556e-05,
      "loss": 4.1225,
      "step": 25740
    },
    {
      "epoch": 0.6917909445116217,
      "grad_norm": 2.313547372817993,
      "learning_rate": 1.5620318670842982e-05,
      "loss": 4.0553,
      "step": 25745
    },
    {
      "epoch": 0.6919252989386,
      "grad_norm": 2.5809359550476074,
      "learning_rate": 1.5613509464796404e-05,
      "loss": 4.156,
      "step": 25750
    },
    {
      "epoch": 0.6920596533655784,
      "grad_norm": 2.3951072692871094,
      "learning_rate": 1.5606700258749833e-05,
      "loss": 4.035,
      "step": 25755
    },
    {
      "epoch": 0.6921940077925568,
      "grad_norm": 2.5846664905548096,
      "learning_rate": 1.5599891052703256e-05,
      "loss": 4.1759,
      "step": 25760
    },
    {
      "epoch": 0.6923283622195351,
      "grad_norm": 2.52017879486084,
      "learning_rate": 1.559308184665668e-05,
      "loss": 4.0831,
      "step": 25765
    },
    {
      "epoch": 0.6924627166465135,
      "grad_norm": 2.46498441696167,
      "learning_rate": 1.5586272640610107e-05,
      "loss": 4.1885,
      "step": 25770
    },
    {
      "epoch": 0.6925970710734919,
      "grad_norm": 2.6984992027282715,
      "learning_rate": 1.557946343456353e-05,
      "loss": 4.0509,
      "step": 25775
    },
    {
      "epoch": 0.6927314255004703,
      "grad_norm": 2.6313719749450684,
      "learning_rate": 1.5572654228516955e-05,
      "loss": 3.9799,
      "step": 25780
    },
    {
      "epoch": 0.6928657799274486,
      "grad_norm": 2.61991024017334,
      "learning_rate": 1.556584502247038e-05,
      "loss": 4.0061,
      "step": 25785
    },
    {
      "epoch": 0.693000134354427,
      "grad_norm": 2.458453893661499,
      "learning_rate": 1.5559035816423806e-05,
      "loss": 4.064,
      "step": 25790
    },
    {
      "epoch": 0.6931344887814054,
      "grad_norm": 2.4678404331207275,
      "learning_rate": 1.555222661037723e-05,
      "loss": 4.1361,
      "step": 25795
    },
    {
      "epoch": 0.6932688432083837,
      "grad_norm": 2.52002215385437,
      "learning_rate": 1.5545417404330657e-05,
      "loss": 4.1595,
      "step": 25800
    },
    {
      "epoch": 0.6934031976353621,
      "grad_norm": 2.4061975479125977,
      "learning_rate": 1.553860819828408e-05,
      "loss": 4.1731,
      "step": 25805
    },
    {
      "epoch": 0.6935375520623405,
      "grad_norm": 2.614473342895508,
      "learning_rate": 1.5531798992237505e-05,
      "loss": 4.0842,
      "step": 25810
    },
    {
      "epoch": 0.6936719064893189,
      "grad_norm": 2.4207589626312256,
      "learning_rate": 1.552498978619093e-05,
      "loss": 4.028,
      "step": 25815
    },
    {
      "epoch": 0.6938062609162972,
      "grad_norm": 2.4167206287384033,
      "learning_rate": 1.5518180580144357e-05,
      "loss": 4.0356,
      "step": 25820
    },
    {
      "epoch": 0.6939406153432756,
      "grad_norm": 2.510004997253418,
      "learning_rate": 1.551137137409778e-05,
      "loss": 4.0186,
      "step": 25825
    },
    {
      "epoch": 0.694074969770254,
      "grad_norm": 2.481091022491455,
      "learning_rate": 1.5504562168051208e-05,
      "loss": 4.1985,
      "step": 25830
    },
    {
      "epoch": 0.6942093241972322,
      "grad_norm": 2.5186636447906494,
      "learning_rate": 1.549775296200463e-05,
      "loss": 4.0809,
      "step": 25835
    },
    {
      "epoch": 0.6943436786242106,
      "grad_norm": 2.8444268703460693,
      "learning_rate": 1.5490943755958056e-05,
      "loss": 4.0639,
      "step": 25840
    },
    {
      "epoch": 0.694478033051189,
      "grad_norm": 2.498209238052368,
      "learning_rate": 1.548413454991148e-05,
      "loss": 4.1191,
      "step": 25845
    },
    {
      "epoch": 0.6946123874781674,
      "grad_norm": 2.4159672260284424,
      "learning_rate": 1.5477325343864907e-05,
      "loss": 4.1708,
      "step": 25850
    },
    {
      "epoch": 0.6947467419051457,
      "grad_norm": 2.6812493801116943,
      "learning_rate": 1.547051613781833e-05,
      "loss": 3.9502,
      "step": 25855
    },
    {
      "epoch": 0.6948810963321241,
      "grad_norm": 2.5849053859710693,
      "learning_rate": 1.546370693177176e-05,
      "loss": 4.1407,
      "step": 25860
    },
    {
      "epoch": 0.6950154507591025,
      "grad_norm": 2.4570093154907227,
      "learning_rate": 1.545689772572518e-05,
      "loss": 4.2139,
      "step": 25865
    },
    {
      "epoch": 0.6951498051860808,
      "grad_norm": 2.679844856262207,
      "learning_rate": 1.5450088519678606e-05,
      "loss": 4.1888,
      "step": 25870
    },
    {
      "epoch": 0.6952841596130592,
      "grad_norm": 2.3229005336761475,
      "learning_rate": 1.5443279313632032e-05,
      "loss": 4.0711,
      "step": 25875
    },
    {
      "epoch": 0.6954185140400376,
      "grad_norm": 2.542799949645996,
      "learning_rate": 1.5436470107585458e-05,
      "loss": 4.0377,
      "step": 25880
    },
    {
      "epoch": 0.695552868467016,
      "grad_norm": 2.6928157806396484,
      "learning_rate": 1.542966090153888e-05,
      "loss": 3.9875,
      "step": 25885
    },
    {
      "epoch": 0.6956872228939943,
      "grad_norm": 2.333193063735962,
      "learning_rate": 1.542285169549231e-05,
      "loss": 4.0253,
      "step": 25890
    },
    {
      "epoch": 0.6958215773209727,
      "grad_norm": 2.515695571899414,
      "learning_rate": 1.541604248944573e-05,
      "loss": 3.9663,
      "step": 25895
    },
    {
      "epoch": 0.6959559317479511,
      "grad_norm": 2.560884952545166,
      "learning_rate": 1.5409233283399157e-05,
      "loss": 4.1259,
      "step": 25900
    },
    {
      "epoch": 0.6960902861749295,
      "grad_norm": 2.3469066619873047,
      "learning_rate": 1.5402424077352583e-05,
      "loss": 4.0477,
      "step": 25905
    },
    {
      "epoch": 0.6962246406019078,
      "grad_norm": 2.4805779457092285,
      "learning_rate": 1.539561487130601e-05,
      "loss": 4.1603,
      "step": 25910
    },
    {
      "epoch": 0.6963589950288862,
      "grad_norm": 2.5253777503967285,
      "learning_rate": 1.538880566525943e-05,
      "loss": 4.1476,
      "step": 25915
    },
    {
      "epoch": 0.6964933494558646,
      "grad_norm": 2.321458101272583,
      "learning_rate": 1.5381996459212856e-05,
      "loss": 3.9253,
      "step": 25920
    },
    {
      "epoch": 0.6966277038828429,
      "grad_norm": 2.587799549102783,
      "learning_rate": 1.5375187253166282e-05,
      "loss": 4.2668,
      "step": 25925
    },
    {
      "epoch": 0.6967620583098213,
      "grad_norm": 2.582622528076172,
      "learning_rate": 1.5368378047119704e-05,
      "loss": 3.9654,
      "step": 25930
    },
    {
      "epoch": 0.6968964127367997,
      "grad_norm": 2.7423887252807617,
      "learning_rate": 1.5361568841073133e-05,
      "loss": 4.1667,
      "step": 25935
    },
    {
      "epoch": 0.6970307671637781,
      "grad_norm": 2.495927572250366,
      "learning_rate": 1.5354759635026555e-05,
      "loss": 4.0584,
      "step": 25940
    },
    {
      "epoch": 0.6971651215907564,
      "grad_norm": 2.783794403076172,
      "learning_rate": 1.534795042897998e-05,
      "loss": 3.925,
      "step": 25945
    },
    {
      "epoch": 0.6972994760177348,
      "grad_norm": 2.5604984760284424,
      "learning_rate": 1.5341141222933407e-05,
      "loss": 4.0635,
      "step": 25950
    },
    {
      "epoch": 0.6974338304447132,
      "grad_norm": 2.4534244537353516,
      "learning_rate": 1.5334332016886832e-05,
      "loss": 4.1049,
      "step": 25955
    },
    {
      "epoch": 0.6975681848716915,
      "grad_norm": 2.4351933002471924,
      "learning_rate": 1.5327522810840255e-05,
      "loss": 4.1289,
      "step": 25960
    },
    {
      "epoch": 0.6977025392986699,
      "grad_norm": 2.7799644470214844,
      "learning_rate": 1.5320713604793684e-05,
      "loss": 4.0219,
      "step": 25965
    },
    {
      "epoch": 0.6978368937256483,
      "grad_norm": 2.5682930946350098,
      "learning_rate": 1.5313904398747106e-05,
      "loss": 4.0897,
      "step": 25970
    },
    {
      "epoch": 0.6979712481526267,
      "grad_norm": 2.560134172439575,
      "learning_rate": 1.530709519270053e-05,
      "loss": 4.0724,
      "step": 25975
    },
    {
      "epoch": 0.698105602579605,
      "grad_norm": 2.387601613998413,
      "learning_rate": 1.5300285986653957e-05,
      "loss": 4.0859,
      "step": 25980
    },
    {
      "epoch": 0.6982399570065834,
      "grad_norm": 2.8333263397216797,
      "learning_rate": 1.5293476780607383e-05,
      "loss": 4.1522,
      "step": 25985
    },
    {
      "epoch": 0.6983743114335618,
      "grad_norm": 2.4672279357910156,
      "learning_rate": 1.5286667574560805e-05,
      "loss": 4.0748,
      "step": 25990
    },
    {
      "epoch": 0.6985086658605401,
      "grad_norm": 2.6123502254486084,
      "learning_rate": 1.5279858368514234e-05,
      "loss": 3.9964,
      "step": 25995
    },
    {
      "epoch": 0.6986430202875185,
      "grad_norm": 2.3748061656951904,
      "learning_rate": 1.5273049162467657e-05,
      "loss": 4.2095,
      "step": 26000
    },
    {
      "epoch": 0.6987773747144969,
      "grad_norm": 2.618396759033203,
      "learning_rate": 1.5266239956421082e-05,
      "loss": 3.9188,
      "step": 26005
    },
    {
      "epoch": 0.6989117291414753,
      "grad_norm": 2.712648868560791,
      "learning_rate": 1.5259430750374508e-05,
      "loss": 4.0466,
      "step": 26010
    },
    {
      "epoch": 0.6990460835684535,
      "grad_norm": 2.7643544673919678,
      "learning_rate": 1.5252621544327934e-05,
      "loss": 4.2019,
      "step": 26015
    },
    {
      "epoch": 0.699180437995432,
      "grad_norm": 2.5711889266967773,
      "learning_rate": 1.5245812338281358e-05,
      "loss": 4.0279,
      "step": 26020
    },
    {
      "epoch": 0.6993147924224103,
      "grad_norm": 2.510981321334839,
      "learning_rate": 1.5239003132234783e-05,
      "loss": 4.1401,
      "step": 26025
    },
    {
      "epoch": 0.6994491468493886,
      "grad_norm": 2.73475980758667,
      "learning_rate": 1.5232193926188207e-05,
      "loss": 4.0333,
      "step": 26030
    },
    {
      "epoch": 0.699583501276367,
      "grad_norm": 2.507845878601074,
      "learning_rate": 1.5225384720141633e-05,
      "loss": 4.0343,
      "step": 26035
    },
    {
      "epoch": 0.6997178557033454,
      "grad_norm": 2.4098801612854004,
      "learning_rate": 1.5218575514095057e-05,
      "loss": 4.0579,
      "step": 26040
    },
    {
      "epoch": 0.6998522101303238,
      "grad_norm": 2.5739855766296387,
      "learning_rate": 1.5211766308048484e-05,
      "loss": 4.0744,
      "step": 26045
    },
    {
      "epoch": 0.6999865645573021,
      "grad_norm": 2.4235591888427734,
      "learning_rate": 1.5204957102001906e-05,
      "loss": 4.2313,
      "step": 26050
    },
    {
      "epoch": 0.7001209189842805,
      "grad_norm": 2.592965602874756,
      "learning_rate": 1.5198147895955334e-05,
      "loss": 4.0123,
      "step": 26055
    },
    {
      "epoch": 0.7002552734112589,
      "grad_norm": 2.484647274017334,
      "learning_rate": 1.5191338689908758e-05,
      "loss": 4.1326,
      "step": 26060
    },
    {
      "epoch": 0.7003896278382372,
      "grad_norm": 2.39859676361084,
      "learning_rate": 1.5184529483862183e-05,
      "loss": 4.0571,
      "step": 26065
    },
    {
      "epoch": 0.7005239822652156,
      "grad_norm": 2.7295615673065186,
      "learning_rate": 1.5177720277815607e-05,
      "loss": 4.0313,
      "step": 26070
    },
    {
      "epoch": 0.700658336692194,
      "grad_norm": 2.836071252822876,
      "learning_rate": 1.5170911071769031e-05,
      "loss": 4.1449,
      "step": 26075
    },
    {
      "epoch": 0.7007926911191724,
      "grad_norm": 2.424800395965576,
      "learning_rate": 1.5164101865722457e-05,
      "loss": 4.1189,
      "step": 26080
    },
    {
      "epoch": 0.7009270455461507,
      "grad_norm": 2.645490884780884,
      "learning_rate": 1.5157292659675881e-05,
      "loss": 4.1088,
      "step": 26085
    },
    {
      "epoch": 0.7010613999731291,
      "grad_norm": 2.3608758449554443,
      "learning_rate": 1.5150483453629308e-05,
      "loss": 4.1022,
      "step": 26090
    },
    {
      "epoch": 0.7011957544001075,
      "grad_norm": 2.830937623977661,
      "learning_rate": 1.5143674247582732e-05,
      "loss": 4.1119,
      "step": 26095
    },
    {
      "epoch": 0.7013301088270859,
      "grad_norm": 2.8303327560424805,
      "learning_rate": 1.5136865041536158e-05,
      "loss": 4.1447,
      "step": 26100
    },
    {
      "epoch": 0.7014644632540642,
      "grad_norm": 2.8125314712524414,
      "learning_rate": 1.5130055835489582e-05,
      "loss": 4.0966,
      "step": 26105
    },
    {
      "epoch": 0.7015988176810426,
      "grad_norm": 2.4406838417053223,
      "learning_rate": 1.5123246629443008e-05,
      "loss": 3.9993,
      "step": 26110
    },
    {
      "epoch": 0.701733172108021,
      "grad_norm": 2.3273072242736816,
      "learning_rate": 1.5116437423396431e-05,
      "loss": 4.1481,
      "step": 26115
    },
    {
      "epoch": 0.7018675265349993,
      "grad_norm": 2.6003124713897705,
      "learning_rate": 1.5109628217349859e-05,
      "loss": 3.9864,
      "step": 26120
    },
    {
      "epoch": 0.7020018809619777,
      "grad_norm": 2.5560860633850098,
      "learning_rate": 1.5102819011303281e-05,
      "loss": 4.0363,
      "step": 26125
    },
    {
      "epoch": 0.7021362353889561,
      "grad_norm": 2.6664600372314453,
      "learning_rate": 1.5096009805256708e-05,
      "loss": 3.9703,
      "step": 26130
    },
    {
      "epoch": 0.7022705898159345,
      "grad_norm": 2.293596029281616,
      "learning_rate": 1.5089200599210132e-05,
      "loss": 4.125,
      "step": 26135
    },
    {
      "epoch": 0.7024049442429128,
      "grad_norm": 2.6780011653900146,
      "learning_rate": 1.5082391393163558e-05,
      "loss": 4.0173,
      "step": 26140
    },
    {
      "epoch": 0.7025392986698912,
      "grad_norm": 2.6006250381469727,
      "learning_rate": 1.5075582187116982e-05,
      "loss": 4.102,
      "step": 26145
    },
    {
      "epoch": 0.7026736530968696,
      "grad_norm": 2.3827452659606934,
      "learning_rate": 1.506877298107041e-05,
      "loss": 4.016,
      "step": 26150
    },
    {
      "epoch": 0.7028080075238479,
      "grad_norm": 2.819798469543457,
      "learning_rate": 1.5061963775023832e-05,
      "loss": 4.0897,
      "step": 26155
    },
    {
      "epoch": 0.7029423619508263,
      "grad_norm": 2.3865647315979004,
      "learning_rate": 1.5055154568977259e-05,
      "loss": 4.2287,
      "step": 26160
    },
    {
      "epoch": 0.7030767163778047,
      "grad_norm": 2.5091264247894287,
      "learning_rate": 1.5048345362930683e-05,
      "loss": 4.0695,
      "step": 26165
    },
    {
      "epoch": 0.7032110708047831,
      "grad_norm": 2.2062034606933594,
      "learning_rate": 1.5041536156884109e-05,
      "loss": 4.0841,
      "step": 26170
    },
    {
      "epoch": 0.7033454252317614,
      "grad_norm": 2.5209779739379883,
      "learning_rate": 1.5034726950837533e-05,
      "loss": 4.0818,
      "step": 26175
    },
    {
      "epoch": 0.7034797796587398,
      "grad_norm": 2.980149030685425,
      "learning_rate": 1.502791774479096e-05,
      "loss": 4.1858,
      "step": 26180
    },
    {
      "epoch": 0.7036141340857182,
      "grad_norm": 2.374828338623047,
      "learning_rate": 1.5021108538744382e-05,
      "loss": 3.852,
      "step": 26185
    },
    {
      "epoch": 0.7037484885126964,
      "grad_norm": 2.4109325408935547,
      "learning_rate": 1.501429933269781e-05,
      "loss": 4.0448,
      "step": 26190
    },
    {
      "epoch": 0.7038828429396748,
      "grad_norm": 2.5528323650360107,
      "learning_rate": 1.5007490126651234e-05,
      "loss": 4.0493,
      "step": 26195
    },
    {
      "epoch": 0.7040171973666532,
      "grad_norm": 2.489856243133545,
      "learning_rate": 1.500068092060466e-05,
      "loss": 4.1052,
      "step": 26200
    },
    {
      "epoch": 0.7041515517936316,
      "grad_norm": 2.6666648387908936,
      "learning_rate": 1.4993871714558083e-05,
      "loss": 3.9554,
      "step": 26205
    },
    {
      "epoch": 0.7042859062206099,
      "grad_norm": 2.6465401649475098,
      "learning_rate": 1.4987062508511509e-05,
      "loss": 4.1362,
      "step": 26210
    },
    {
      "epoch": 0.7044202606475883,
      "grad_norm": 2.885828971862793,
      "learning_rate": 1.4980253302464933e-05,
      "loss": 4.0758,
      "step": 26215
    },
    {
      "epoch": 0.7045546150745667,
      "grad_norm": 2.9625277519226074,
      "learning_rate": 1.4973444096418357e-05,
      "loss": 4.1831,
      "step": 26220
    },
    {
      "epoch": 0.704688969501545,
      "grad_norm": 2.434601068496704,
      "learning_rate": 1.4966634890371784e-05,
      "loss": 4.0336,
      "step": 26225
    },
    {
      "epoch": 0.7048233239285234,
      "grad_norm": 2.815753221511841,
      "learning_rate": 1.4959825684325206e-05,
      "loss": 4.0622,
      "step": 26230
    },
    {
      "epoch": 0.7049576783555018,
      "grad_norm": 2.665599822998047,
      "learning_rate": 1.4953016478278634e-05,
      "loss": 4.0688,
      "step": 26235
    },
    {
      "epoch": 0.7050920327824802,
      "grad_norm": 2.629580497741699,
      "learning_rate": 1.4946207272232058e-05,
      "loss": 4.0754,
      "step": 26240
    },
    {
      "epoch": 0.7052263872094585,
      "grad_norm": 2.5625669956207275,
      "learning_rate": 1.4939398066185483e-05,
      "loss": 4.1611,
      "step": 26245
    },
    {
      "epoch": 0.7053607416364369,
      "grad_norm": 2.6608223915100098,
      "learning_rate": 1.4932588860138907e-05,
      "loss": 4.095,
      "step": 26250
    },
    {
      "epoch": 0.7054950960634153,
      "grad_norm": 2.640458106994629,
      "learning_rate": 1.4925779654092335e-05,
      "loss": 4.1411,
      "step": 26255
    },
    {
      "epoch": 0.7056294504903936,
      "grad_norm": 2.2115797996520996,
      "learning_rate": 1.4918970448045757e-05,
      "loss": 4.1492,
      "step": 26260
    },
    {
      "epoch": 0.705763804917372,
      "grad_norm": 2.2571756839752197,
      "learning_rate": 1.4912161241999184e-05,
      "loss": 4.1457,
      "step": 26265
    },
    {
      "epoch": 0.7058981593443504,
      "grad_norm": 2.726890802383423,
      "learning_rate": 1.4905352035952608e-05,
      "loss": 4.1387,
      "step": 26270
    },
    {
      "epoch": 0.7060325137713288,
      "grad_norm": 2.4840593338012695,
      "learning_rate": 1.4898542829906034e-05,
      "loss": 4.1677,
      "step": 26275
    },
    {
      "epoch": 0.7061668681983071,
      "grad_norm": 2.315808057785034,
      "learning_rate": 1.4891733623859458e-05,
      "loss": 4.0852,
      "step": 26280
    },
    {
      "epoch": 0.7063012226252855,
      "grad_norm": 2.670393228530884,
      "learning_rate": 1.4884924417812885e-05,
      "loss": 4.0576,
      "step": 26285
    },
    {
      "epoch": 0.7064355770522639,
      "grad_norm": 2.3993144035339355,
      "learning_rate": 1.4878115211766307e-05,
      "loss": 4.0056,
      "step": 26290
    },
    {
      "epoch": 0.7065699314792422,
      "grad_norm": 2.2575385570526123,
      "learning_rate": 1.4871306005719735e-05,
      "loss": 4.0221,
      "step": 26295
    },
    {
      "epoch": 0.7067042859062206,
      "grad_norm": 2.685051202774048,
      "learning_rate": 1.4864496799673159e-05,
      "loss": 4.1057,
      "step": 26300
    },
    {
      "epoch": 0.706838640333199,
      "grad_norm": 2.6557161808013916,
      "learning_rate": 1.4857687593626584e-05,
      "loss": 4.0754,
      "step": 26305
    },
    {
      "epoch": 0.7069729947601774,
      "grad_norm": 2.5684964656829834,
      "learning_rate": 1.4850878387580008e-05,
      "loss": 3.9826,
      "step": 26310
    },
    {
      "epoch": 0.7071073491871557,
      "grad_norm": 2.6620099544525146,
      "learning_rate": 1.4844069181533434e-05,
      "loss": 4.0753,
      "step": 26315
    },
    {
      "epoch": 0.7072417036141341,
      "grad_norm": 2.7526049613952637,
      "learning_rate": 1.4837259975486858e-05,
      "loss": 3.9656,
      "step": 26320
    },
    {
      "epoch": 0.7073760580411125,
      "grad_norm": 2.5699222087860107,
      "learning_rate": 1.4830450769440285e-05,
      "loss": 4.0007,
      "step": 26325
    },
    {
      "epoch": 0.7075104124680909,
      "grad_norm": 2.7119340896606445,
      "learning_rate": 1.482364156339371e-05,
      "loss": 4.138,
      "step": 26330
    },
    {
      "epoch": 0.7076447668950692,
      "grad_norm": 2.333024024963379,
      "learning_rate": 1.4816832357347135e-05,
      "loss": 4.0829,
      "step": 26335
    },
    {
      "epoch": 0.7077791213220476,
      "grad_norm": 2.4139792919158936,
      "learning_rate": 1.4810023151300559e-05,
      "loss": 3.9464,
      "step": 26340
    },
    {
      "epoch": 0.707913475749026,
      "grad_norm": 2.683587074279785,
      "learning_rate": 1.4803213945253985e-05,
      "loss": 3.9938,
      "step": 26345
    },
    {
      "epoch": 0.7080478301760043,
      "grad_norm": 2.499746799468994,
      "learning_rate": 1.4796404739207409e-05,
      "loss": 4.1731,
      "step": 26350
    },
    {
      "epoch": 0.7081821846029827,
      "grad_norm": 2.4002585411071777,
      "learning_rate": 1.4789595533160836e-05,
      "loss": 4.0827,
      "step": 26355
    },
    {
      "epoch": 0.7083165390299611,
      "grad_norm": 2.405299186706543,
      "learning_rate": 1.478278632711426e-05,
      "loss": 4.0191,
      "step": 26360
    },
    {
      "epoch": 0.7084508934569395,
      "grad_norm": 2.2736575603485107,
      "learning_rate": 1.4775977121067682e-05,
      "loss": 3.9334,
      "step": 26365
    },
    {
      "epoch": 0.7085852478839177,
      "grad_norm": 2.3930933475494385,
      "learning_rate": 1.476916791502111e-05,
      "loss": 4.1344,
      "step": 26370
    },
    {
      "epoch": 0.7087196023108961,
      "grad_norm": 2.4402098655700684,
      "learning_rate": 1.4762358708974533e-05,
      "loss": 4.0797,
      "step": 26375
    },
    {
      "epoch": 0.7088539567378745,
      "grad_norm": 2.5580220222473145,
      "learning_rate": 1.4755549502927959e-05,
      "loss": 4.0266,
      "step": 26380
    },
    {
      "epoch": 0.7089883111648528,
      "grad_norm": 2.6297922134399414,
      "learning_rate": 1.4748740296881383e-05,
      "loss": 4.1007,
      "step": 26385
    },
    {
      "epoch": 0.7091226655918312,
      "grad_norm": 2.739976167678833,
      "learning_rate": 1.474193109083481e-05,
      "loss": 4.0292,
      "step": 26390
    },
    {
      "epoch": 0.7092570200188096,
      "grad_norm": 2.9241387844085693,
      "learning_rate": 1.4735121884788233e-05,
      "loss": 4.0865,
      "step": 26395
    },
    {
      "epoch": 0.709391374445788,
      "grad_norm": 2.477668285369873,
      "learning_rate": 1.472831267874166e-05,
      "loss": 4.0826,
      "step": 26400
    },
    {
      "epoch": 0.7095257288727663,
      "grad_norm": 2.4050114154815674,
      "learning_rate": 1.4721503472695084e-05,
      "loss": 4.2485,
      "step": 26405
    },
    {
      "epoch": 0.7096600832997447,
      "grad_norm": 2.5341122150421143,
      "learning_rate": 1.471469426664851e-05,
      "loss": 4.0752,
      "step": 26410
    },
    {
      "epoch": 0.7097944377267231,
      "grad_norm": 2.33636474609375,
      "learning_rate": 1.4707885060601934e-05,
      "loss": 4.0872,
      "step": 26415
    },
    {
      "epoch": 0.7099287921537014,
      "grad_norm": 2.5028128623962402,
      "learning_rate": 1.470107585455536e-05,
      "loss": 4.1374,
      "step": 26420
    },
    {
      "epoch": 0.7100631465806798,
      "grad_norm": 2.4000141620635986,
      "learning_rate": 1.4694266648508783e-05,
      "loss": 4.1175,
      "step": 26425
    },
    {
      "epoch": 0.7101975010076582,
      "grad_norm": 2.6969828605651855,
      "learning_rate": 1.468745744246221e-05,
      "loss": 4.0836,
      "step": 26430
    },
    {
      "epoch": 0.7103318554346366,
      "grad_norm": 2.5366203784942627,
      "learning_rate": 1.4680648236415635e-05,
      "loss": 4.1898,
      "step": 26435
    },
    {
      "epoch": 0.7104662098616149,
      "grad_norm": 2.7102112770080566,
      "learning_rate": 1.467383903036906e-05,
      "loss": 4.092,
      "step": 26440
    },
    {
      "epoch": 0.7106005642885933,
      "grad_norm": 2.5550217628479004,
      "learning_rate": 1.4667029824322484e-05,
      "loss": 4.1527,
      "step": 26445
    },
    {
      "epoch": 0.7107349187155717,
      "grad_norm": 2.520310640335083,
      "learning_rate": 1.466022061827591e-05,
      "loss": 4.0852,
      "step": 26450
    },
    {
      "epoch": 0.71086927314255,
      "grad_norm": 2.46159291267395,
      "learning_rate": 1.4653411412229334e-05,
      "loss": 4.0588,
      "step": 26455
    },
    {
      "epoch": 0.7110036275695284,
      "grad_norm": 2.423255205154419,
      "learning_rate": 1.4646602206182761e-05,
      "loss": 4.0273,
      "step": 26460
    },
    {
      "epoch": 0.7111379819965068,
      "grad_norm": 2.449784755706787,
      "learning_rate": 1.4639793000136185e-05,
      "loss": 4.0283,
      "step": 26465
    },
    {
      "epoch": 0.7112723364234852,
      "grad_norm": 2.699544906616211,
      "learning_rate": 1.463298379408961e-05,
      "loss": 4.0476,
      "step": 26470
    },
    {
      "epoch": 0.7114066908504635,
      "grad_norm": 2.4506659507751465,
      "learning_rate": 1.4626174588043035e-05,
      "loss": 4.0428,
      "step": 26475
    },
    {
      "epoch": 0.7115410452774419,
      "grad_norm": 2.6937453746795654,
      "learning_rate": 1.461936538199646e-05,
      "loss": 3.9574,
      "step": 26480
    },
    {
      "epoch": 0.7116753997044203,
      "grad_norm": 2.4880151748657227,
      "learning_rate": 1.4612556175949884e-05,
      "loss": 4.111,
      "step": 26485
    },
    {
      "epoch": 0.7118097541313986,
      "grad_norm": 2.4697813987731934,
      "learning_rate": 1.4605746969903312e-05,
      "loss": 4.0468,
      "step": 26490
    },
    {
      "epoch": 0.711944108558377,
      "grad_norm": 2.5098752975463867,
      "learning_rate": 1.4598937763856736e-05,
      "loss": 3.977,
      "step": 26495
    },
    {
      "epoch": 0.7120784629853554,
      "grad_norm": 2.6967036724090576,
      "learning_rate": 1.4592128557810161e-05,
      "loss": 4.1669,
      "step": 26500
    },
    {
      "epoch": 0.7122128174123338,
      "grad_norm": 2.4671475887298584,
      "learning_rate": 1.4585319351763585e-05,
      "loss": 4.051,
      "step": 26505
    },
    {
      "epoch": 0.7123471718393121,
      "grad_norm": 2.224522352218628,
      "learning_rate": 1.4578510145717011e-05,
      "loss": 4.1308,
      "step": 26510
    },
    {
      "epoch": 0.7124815262662905,
      "grad_norm": 2.4402294158935547,
      "learning_rate": 1.4571700939670435e-05,
      "loss": 4.0558,
      "step": 26515
    },
    {
      "epoch": 0.7126158806932689,
      "grad_norm": 2.3905768394470215,
      "learning_rate": 1.4564891733623859e-05,
      "loss": 4.1375,
      "step": 26520
    },
    {
      "epoch": 0.7127502351202473,
      "grad_norm": 2.386465311050415,
      "learning_rate": 1.4558082527577285e-05,
      "loss": 4.1022,
      "step": 26525
    },
    {
      "epoch": 0.7128845895472256,
      "grad_norm": 2.5660958290100098,
      "learning_rate": 1.4551273321530709e-05,
      "loss": 4.0988,
      "step": 26530
    },
    {
      "epoch": 0.713018943974204,
      "grad_norm": 2.4855527877807617,
      "learning_rate": 1.4544464115484136e-05,
      "loss": 4.0624,
      "step": 26535
    },
    {
      "epoch": 0.7131532984011824,
      "grad_norm": 2.485211133956909,
      "learning_rate": 1.453765490943756e-05,
      "loss": 4.1359,
      "step": 26540
    },
    {
      "epoch": 0.7132876528281606,
      "grad_norm": 2.645197868347168,
      "learning_rate": 1.4530845703390985e-05,
      "loss": 4.0461,
      "step": 26545
    },
    {
      "epoch": 0.713422007255139,
      "grad_norm": 2.536564826965332,
      "learning_rate": 1.452403649734441e-05,
      "loss": 4.0964,
      "step": 26550
    },
    {
      "epoch": 0.7135563616821174,
      "grad_norm": 2.5270562171936035,
      "learning_rate": 1.4517227291297835e-05,
      "loss": 4.0555,
      "step": 26555
    },
    {
      "epoch": 0.7136907161090958,
      "grad_norm": 2.8785104751586914,
      "learning_rate": 1.4510418085251259e-05,
      "loss": 4.0692,
      "step": 26560
    },
    {
      "epoch": 0.7138250705360741,
      "grad_norm": 2.7163147926330566,
      "learning_rate": 1.4503608879204686e-05,
      "loss": 4.0407,
      "step": 26565
    },
    {
      "epoch": 0.7139594249630525,
      "grad_norm": 2.8238093852996826,
      "learning_rate": 1.449679967315811e-05,
      "loss": 4.1241,
      "step": 26570
    },
    {
      "epoch": 0.7140937793900309,
      "grad_norm": 2.3929879665374756,
      "learning_rate": 1.4489990467111536e-05,
      "loss": 3.9612,
      "step": 26575
    },
    {
      "epoch": 0.7142281338170092,
      "grad_norm": 2.4446349143981934,
      "learning_rate": 1.448318126106496e-05,
      "loss": 3.9631,
      "step": 26580
    },
    {
      "epoch": 0.7143624882439876,
      "grad_norm": 2.558551073074341,
      "learning_rate": 1.4476372055018386e-05,
      "loss": 4.0756,
      "step": 26585
    },
    {
      "epoch": 0.714496842670966,
      "grad_norm": 2.3390660285949707,
      "learning_rate": 1.446956284897181e-05,
      "loss": 4.0166,
      "step": 26590
    },
    {
      "epoch": 0.7146311970979444,
      "grad_norm": 2.4889774322509766,
      "learning_rate": 1.4462753642925237e-05,
      "loss": 4.1378,
      "step": 26595
    },
    {
      "epoch": 0.7147655515249227,
      "grad_norm": 2.6086084842681885,
      "learning_rate": 1.445594443687866e-05,
      "loss": 4.0735,
      "step": 26600
    },
    {
      "epoch": 0.7148999059519011,
      "grad_norm": 2.5231826305389404,
      "learning_rate": 1.4449135230832087e-05,
      "loss": 3.9422,
      "step": 26605
    },
    {
      "epoch": 0.7150342603788795,
      "grad_norm": 2.4458184242248535,
      "learning_rate": 1.444232602478551e-05,
      "loss": 4.1077,
      "step": 26610
    },
    {
      "epoch": 0.7151686148058578,
      "grad_norm": 2.3456358909606934,
      "learning_rate": 1.4435516818738936e-05,
      "loss": 4.1583,
      "step": 26615
    },
    {
      "epoch": 0.7153029692328362,
      "grad_norm": 2.6229805946350098,
      "learning_rate": 1.442870761269236e-05,
      "loss": 3.9935,
      "step": 26620
    },
    {
      "epoch": 0.7154373236598146,
      "grad_norm": 2.239637613296509,
      "learning_rate": 1.4421898406645788e-05,
      "loss": 3.9939,
      "step": 26625
    },
    {
      "epoch": 0.715571678086793,
      "grad_norm": 2.418419122695923,
      "learning_rate": 1.441508920059921e-05,
      "loss": 4.1198,
      "step": 26630
    },
    {
      "epoch": 0.7157060325137713,
      "grad_norm": 2.3493337631225586,
      "learning_rate": 1.4408279994552637e-05,
      "loss": 4.1135,
      "step": 26635
    },
    {
      "epoch": 0.7158403869407497,
      "grad_norm": 2.7357707023620605,
      "learning_rate": 1.4401470788506061e-05,
      "loss": 4.1177,
      "step": 26640
    },
    {
      "epoch": 0.7159747413677281,
      "grad_norm": 2.4043729305267334,
      "learning_rate": 1.4394661582459487e-05,
      "loss": 4.1447,
      "step": 26645
    },
    {
      "epoch": 0.7161090957947064,
      "grad_norm": 2.467005491256714,
      "learning_rate": 1.438785237641291e-05,
      "loss": 4.0203,
      "step": 26650
    },
    {
      "epoch": 0.7162434502216848,
      "grad_norm": 2.7044520378112793,
      "learning_rate": 1.4381043170366338e-05,
      "loss": 4.0562,
      "step": 26655
    },
    {
      "epoch": 0.7163778046486632,
      "grad_norm": 2.4532554149627686,
      "learning_rate": 1.437423396431976e-05,
      "loss": 4.1097,
      "step": 26660
    },
    {
      "epoch": 0.7165121590756416,
      "grad_norm": 2.3541104793548584,
      "learning_rate": 1.4367424758273184e-05,
      "loss": 3.916,
      "step": 26665
    },
    {
      "epoch": 0.7166465135026199,
      "grad_norm": 2.4951586723327637,
      "learning_rate": 1.4360615552226612e-05,
      "loss": 4.1502,
      "step": 26670
    },
    {
      "epoch": 0.7167808679295983,
      "grad_norm": 2.4293692111968994,
      "learning_rate": 1.4353806346180036e-05,
      "loss": 4.1134,
      "step": 26675
    },
    {
      "epoch": 0.7169152223565767,
      "grad_norm": 2.527618169784546,
      "learning_rate": 1.4346997140133461e-05,
      "loss": 3.9361,
      "step": 26680
    },
    {
      "epoch": 0.717049576783555,
      "grad_norm": 2.3842506408691406,
      "learning_rate": 1.4340187934086885e-05,
      "loss": 4.0462,
      "step": 26685
    },
    {
      "epoch": 0.7171839312105334,
      "grad_norm": 2.7903149127960205,
      "learning_rate": 1.4333378728040311e-05,
      "loss": 4.1503,
      "step": 26690
    },
    {
      "epoch": 0.7173182856375118,
      "grad_norm": 2.3270418643951416,
      "learning_rate": 1.4326569521993735e-05,
      "loss": 3.9648,
      "step": 26695
    },
    {
      "epoch": 0.7174526400644902,
      "grad_norm": 2.5134284496307373,
      "learning_rate": 1.4319760315947162e-05,
      "loss": 4.0265,
      "step": 26700
    },
    {
      "epoch": 0.7175869944914685,
      "grad_norm": 2.501375675201416,
      "learning_rate": 1.4312951109900584e-05,
      "loss": 4.0328,
      "step": 26705
    },
    {
      "epoch": 0.7177213489184469,
      "grad_norm": 2.3473119735717773,
      "learning_rate": 1.4306141903854012e-05,
      "loss": 4.1044,
      "step": 26710
    },
    {
      "epoch": 0.7178557033454253,
      "grad_norm": 2.488936424255371,
      "learning_rate": 1.4299332697807436e-05,
      "loss": 4.1315,
      "step": 26715
    },
    {
      "epoch": 0.7179900577724037,
      "grad_norm": 2.3840649127960205,
      "learning_rate": 1.4292523491760861e-05,
      "loss": 4.0205,
      "step": 26720
    },
    {
      "epoch": 0.718124412199382,
      "grad_norm": 2.489004135131836,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 3.9739,
      "step": 26725
    },
    {
      "epoch": 0.7182587666263603,
      "grad_norm": 2.424358367919922,
      "learning_rate": 1.4278905079667713e-05,
      "loss": 4.0712,
      "step": 26730
    },
    {
      "epoch": 0.7183931210533387,
      "grad_norm": 2.6073362827301025,
      "learning_rate": 1.4272095873621135e-05,
      "loss": 4.1718,
      "step": 26735
    },
    {
      "epoch": 0.718527475480317,
      "grad_norm": 2.295586585998535,
      "learning_rate": 1.4265286667574562e-05,
      "loss": 4.2611,
      "step": 26740
    },
    {
      "epoch": 0.7186618299072954,
      "grad_norm": 2.448373794555664,
      "learning_rate": 1.4258477461527986e-05,
      "loss": 4.1206,
      "step": 26745
    },
    {
      "epoch": 0.7187961843342738,
      "grad_norm": 2.617144823074341,
      "learning_rate": 1.4251668255481412e-05,
      "loss": 4.0857,
      "step": 26750
    },
    {
      "epoch": 0.7189305387612522,
      "grad_norm": 2.4683449268341064,
      "learning_rate": 1.4244859049434836e-05,
      "loss": 4.1375,
      "step": 26755
    },
    {
      "epoch": 0.7190648931882305,
      "grad_norm": 2.7070183753967285,
      "learning_rate": 1.4238049843388263e-05,
      "loss": 4.0269,
      "step": 26760
    },
    {
      "epoch": 0.7191992476152089,
      "grad_norm": 2.5308358669281006,
      "learning_rate": 1.4231240637341686e-05,
      "loss": 3.9775,
      "step": 26765
    },
    {
      "epoch": 0.7193336020421873,
      "grad_norm": 2.491135597229004,
      "learning_rate": 1.4224431431295113e-05,
      "loss": 4.087,
      "step": 26770
    },
    {
      "epoch": 0.7194679564691656,
      "grad_norm": 2.4731311798095703,
      "learning_rate": 1.4217622225248537e-05,
      "loss": 4.1474,
      "step": 26775
    },
    {
      "epoch": 0.719602310896144,
      "grad_norm": 2.589618444442749,
      "learning_rate": 1.4210813019201963e-05,
      "loss": 4.0977,
      "step": 26780
    },
    {
      "epoch": 0.7197366653231224,
      "grad_norm": 2.7208192348480225,
      "learning_rate": 1.4204003813155387e-05,
      "loss": 4.1437,
      "step": 26785
    },
    {
      "epoch": 0.7198710197501008,
      "grad_norm": 2.7720112800598145,
      "learning_rate": 1.4197194607108812e-05,
      "loss": 4.0763,
      "step": 26790
    },
    {
      "epoch": 0.7200053741770791,
      "grad_norm": 2.3922579288482666,
      "learning_rate": 1.4190385401062236e-05,
      "loss": 4.101,
      "step": 26795
    },
    {
      "epoch": 0.7201397286040575,
      "grad_norm": 2.4025979042053223,
      "learning_rate": 1.4183576195015664e-05,
      "loss": 4.0523,
      "step": 26800
    },
    {
      "epoch": 0.7202740830310359,
      "grad_norm": 2.4548211097717285,
      "learning_rate": 1.4176766988969087e-05,
      "loss": 4.055,
      "step": 26805
    },
    {
      "epoch": 0.7204084374580142,
      "grad_norm": 2.406764030456543,
      "learning_rate": 1.416995778292251e-05,
      "loss": 3.9302,
      "step": 26810
    },
    {
      "epoch": 0.7205427918849926,
      "grad_norm": 2.4489247798919678,
      "learning_rate": 1.4163148576875937e-05,
      "loss": 3.9484,
      "step": 26815
    },
    {
      "epoch": 0.720677146311971,
      "grad_norm": 2.570671796798706,
      "learning_rate": 1.4156339370829361e-05,
      "loss": 4.1053,
      "step": 26820
    },
    {
      "epoch": 0.7208115007389494,
      "grad_norm": 2.5745930671691895,
      "learning_rate": 1.4149530164782787e-05,
      "loss": 4.1426,
      "step": 26825
    },
    {
      "epoch": 0.7209458551659277,
      "grad_norm": 2.4958407878875732,
      "learning_rate": 1.414272095873621e-05,
      "loss": 4.0809,
      "step": 26830
    },
    {
      "epoch": 0.7210802095929061,
      "grad_norm": 2.6374850273132324,
      "learning_rate": 1.4135911752689638e-05,
      "loss": 4.1116,
      "step": 26835
    },
    {
      "epoch": 0.7212145640198845,
      "grad_norm": 2.7088441848754883,
      "learning_rate": 1.412910254664306e-05,
      "loss": 4.0093,
      "step": 26840
    },
    {
      "epoch": 0.7213489184468628,
      "grad_norm": 2.5646162033081055,
      "learning_rate": 1.4122293340596488e-05,
      "loss": 4.0006,
      "step": 26845
    },
    {
      "epoch": 0.7214832728738412,
      "grad_norm": 2.7098500728607178,
      "learning_rate": 1.4115484134549912e-05,
      "loss": 3.9412,
      "step": 26850
    },
    {
      "epoch": 0.7216176273008196,
      "grad_norm": 2.699536085128784,
      "learning_rate": 1.4108674928503337e-05,
      "loss": 4.0701,
      "step": 26855
    },
    {
      "epoch": 0.721751981727798,
      "grad_norm": 2.53094220161438,
      "learning_rate": 1.4101865722456761e-05,
      "loss": 4.0556,
      "step": 26860
    },
    {
      "epoch": 0.7218863361547763,
      "grad_norm": 2.6329643726348877,
      "learning_rate": 1.4095056516410189e-05,
      "loss": 3.9674,
      "step": 26865
    },
    {
      "epoch": 0.7220206905817547,
      "grad_norm": 2.528913736343384,
      "learning_rate": 1.4088247310363611e-05,
      "loss": 4.1399,
      "step": 26870
    },
    {
      "epoch": 0.7221550450087331,
      "grad_norm": 2.6272060871124268,
      "learning_rate": 1.4081438104317038e-05,
      "loss": 4.1591,
      "step": 26875
    },
    {
      "epoch": 0.7222893994357114,
      "grad_norm": 2.2884609699249268,
      "learning_rate": 1.4074628898270462e-05,
      "loss": 4.1442,
      "step": 26880
    },
    {
      "epoch": 0.7224237538626898,
      "grad_norm": 2.301384210586548,
      "learning_rate": 1.4067819692223888e-05,
      "loss": 4.1532,
      "step": 26885
    },
    {
      "epoch": 0.7225581082896682,
      "grad_norm": 2.676708221435547,
      "learning_rate": 1.4061010486177312e-05,
      "loss": 4.1725,
      "step": 26890
    },
    {
      "epoch": 0.7226924627166466,
      "grad_norm": 2.662069797515869,
      "learning_rate": 1.4054201280130737e-05,
      "loss": 4.0778,
      "step": 26895
    },
    {
      "epoch": 0.7228268171436248,
      "grad_norm": 2.5326523780822754,
      "learning_rate": 1.4047392074084161e-05,
      "loss": 3.9873,
      "step": 26900
    },
    {
      "epoch": 0.7229611715706032,
      "grad_norm": 2.6270782947540283,
      "learning_rate": 1.4040582868037589e-05,
      "loss": 4.1274,
      "step": 26905
    },
    {
      "epoch": 0.7230955259975816,
      "grad_norm": 2.7088165283203125,
      "learning_rate": 1.4033773661991013e-05,
      "loss": 4.0239,
      "step": 26910
    },
    {
      "epoch": 0.7232298804245599,
      "grad_norm": 2.6652817726135254,
      "learning_rate": 1.4026964455944438e-05,
      "loss": 4.1503,
      "step": 26915
    },
    {
      "epoch": 0.7233642348515383,
      "grad_norm": 2.4388930797576904,
      "learning_rate": 1.4020155249897862e-05,
      "loss": 4.0661,
      "step": 26920
    },
    {
      "epoch": 0.7234985892785167,
      "grad_norm": 2.6126644611358643,
      "learning_rate": 1.4013346043851288e-05,
      "loss": 4.015,
      "step": 26925
    },
    {
      "epoch": 0.7236329437054951,
      "grad_norm": 2.4859893321990967,
      "learning_rate": 1.4006536837804712e-05,
      "loss": 3.9891,
      "step": 26930
    },
    {
      "epoch": 0.7237672981324734,
      "grad_norm": 2.994967222213745,
      "learning_rate": 1.399972763175814e-05,
      "loss": 4.0666,
      "step": 26935
    },
    {
      "epoch": 0.7239016525594518,
      "grad_norm": 2.969782829284668,
      "learning_rate": 1.3992918425711563e-05,
      "loss": 3.9794,
      "step": 26940
    },
    {
      "epoch": 0.7240360069864302,
      "grad_norm": 2.380441665649414,
      "learning_rate": 1.3986109219664989e-05,
      "loss": 4.1918,
      "step": 26945
    },
    {
      "epoch": 0.7241703614134086,
      "grad_norm": 2.4934520721435547,
      "learning_rate": 1.3979300013618413e-05,
      "loss": 4.0658,
      "step": 26950
    },
    {
      "epoch": 0.7243047158403869,
      "grad_norm": 2.27193546295166,
      "learning_rate": 1.3972490807571839e-05,
      "loss": 4.1259,
      "step": 26955
    },
    {
      "epoch": 0.7244390702673653,
      "grad_norm": 2.8975205421447754,
      "learning_rate": 1.3965681601525263e-05,
      "loss": 4.1436,
      "step": 26960
    },
    {
      "epoch": 0.7245734246943437,
      "grad_norm": 2.549117088317871,
      "learning_rate": 1.3958872395478686e-05,
      "loss": 4.0436,
      "step": 26965
    },
    {
      "epoch": 0.724707779121322,
      "grad_norm": 2.5144097805023193,
      "learning_rate": 1.3952063189432112e-05,
      "loss": 4.048,
      "step": 26970
    },
    {
      "epoch": 0.7248421335483004,
      "grad_norm": 2.318542718887329,
      "learning_rate": 1.3945253983385536e-05,
      "loss": 3.9959,
      "step": 26975
    },
    {
      "epoch": 0.7249764879752788,
      "grad_norm": 2.5673329830169678,
      "learning_rate": 1.3938444777338963e-05,
      "loss": 4.1235,
      "step": 26980
    },
    {
      "epoch": 0.7251108424022572,
      "grad_norm": 2.5433247089385986,
      "learning_rate": 1.3931635571292387e-05,
      "loss": 4.1209,
      "step": 26985
    },
    {
      "epoch": 0.7252451968292355,
      "grad_norm": 2.7180373668670654,
      "learning_rate": 1.3924826365245813e-05,
      "loss": 4.015,
      "step": 26990
    },
    {
      "epoch": 0.7253795512562139,
      "grad_norm": 2.5770130157470703,
      "learning_rate": 1.3918017159199237e-05,
      "loss": 4.0642,
      "step": 26995
    },
    {
      "epoch": 0.7255139056831923,
      "grad_norm": 2.6936371326446533,
      "learning_rate": 1.3911207953152663e-05,
      "loss": 4.1594,
      "step": 27000
    },
    {
      "epoch": 0.7256482601101706,
      "grad_norm": 2.5869860649108887,
      "learning_rate": 1.3904398747106087e-05,
      "loss": 4.211,
      "step": 27005
    },
    {
      "epoch": 0.725782614537149,
      "grad_norm": 2.4290127754211426,
      "learning_rate": 1.3897589541059514e-05,
      "loss": 4.0083,
      "step": 27010
    },
    {
      "epoch": 0.7259169689641274,
      "grad_norm": 2.606985092163086,
      "learning_rate": 1.3890780335012938e-05,
      "loss": 4.0914,
      "step": 27015
    },
    {
      "epoch": 0.7260513233911058,
      "grad_norm": 2.483344316482544,
      "learning_rate": 1.3883971128966364e-05,
      "loss": 4.1123,
      "step": 27020
    },
    {
      "epoch": 0.7261856778180841,
      "grad_norm": 2.521674633026123,
      "learning_rate": 1.3877161922919788e-05,
      "loss": 4.0989,
      "step": 27025
    },
    {
      "epoch": 0.7263200322450625,
      "grad_norm": 2.3827970027923584,
      "learning_rate": 1.3870352716873213e-05,
      "loss": 4.0836,
      "step": 27030
    },
    {
      "epoch": 0.7264543866720409,
      "grad_norm": 2.3737032413482666,
      "learning_rate": 1.3863543510826637e-05,
      "loss": 4.0838,
      "step": 27035
    },
    {
      "epoch": 0.7265887410990192,
      "grad_norm": 2.5359184741973877,
      "learning_rate": 1.3856734304780065e-05,
      "loss": 4.0516,
      "step": 27040
    },
    {
      "epoch": 0.7267230955259976,
      "grad_norm": 2.4275460243225098,
      "learning_rate": 1.3849925098733489e-05,
      "loss": 4.0862,
      "step": 27045
    },
    {
      "epoch": 0.726857449952976,
      "grad_norm": 2.7106385231018066,
      "learning_rate": 1.3843115892686914e-05,
      "loss": 4.016,
      "step": 27050
    },
    {
      "epoch": 0.7269918043799544,
      "grad_norm": 2.7803778648376465,
      "learning_rate": 1.3836306686640338e-05,
      "loss": 3.9879,
      "step": 27055
    },
    {
      "epoch": 0.7271261588069327,
      "grad_norm": 2.810943126678467,
      "learning_rate": 1.3829497480593764e-05,
      "loss": 4.0587,
      "step": 27060
    },
    {
      "epoch": 0.727260513233911,
      "grad_norm": 2.1625187397003174,
      "learning_rate": 1.3822688274547188e-05,
      "loss": 3.9987,
      "step": 27065
    },
    {
      "epoch": 0.7273948676608895,
      "grad_norm": 2.5665905475616455,
      "learning_rate": 1.3815879068500615e-05,
      "loss": 4.031,
      "step": 27070
    },
    {
      "epoch": 0.7275292220878677,
      "grad_norm": 2.3750298023223877,
      "learning_rate": 1.3809069862454037e-05,
      "loss": 3.9724,
      "step": 27075
    },
    {
      "epoch": 0.7276635765148461,
      "grad_norm": 2.4392318725585938,
      "learning_rate": 1.3802260656407465e-05,
      "loss": 3.9896,
      "step": 27080
    },
    {
      "epoch": 0.7277979309418245,
      "grad_norm": 2.6792218685150146,
      "learning_rate": 1.3795451450360889e-05,
      "loss": 4.0385,
      "step": 27085
    },
    {
      "epoch": 0.727932285368803,
      "grad_norm": 2.608372688293457,
      "learning_rate": 1.3788642244314314e-05,
      "loss": 4.005,
      "step": 27090
    },
    {
      "epoch": 0.7280666397957812,
      "grad_norm": 2.510970115661621,
      "learning_rate": 1.3781833038267738e-05,
      "loss": 4.0427,
      "step": 27095
    },
    {
      "epoch": 0.7282009942227596,
      "grad_norm": 2.5644888877868652,
      "learning_rate": 1.3775023832221166e-05,
      "loss": 4.1408,
      "step": 27100
    },
    {
      "epoch": 0.728335348649738,
      "grad_norm": 2.51778507232666,
      "learning_rate": 1.3768214626174588e-05,
      "loss": 3.9722,
      "step": 27105
    },
    {
      "epoch": 0.7284697030767163,
      "grad_norm": 2.6161813735961914,
      "learning_rate": 1.3761405420128012e-05,
      "loss": 3.9461,
      "step": 27110
    },
    {
      "epoch": 0.7286040575036947,
      "grad_norm": 2.4981157779693604,
      "learning_rate": 1.375459621408144e-05,
      "loss": 4.16,
      "step": 27115
    },
    {
      "epoch": 0.7287384119306731,
      "grad_norm": 2.6664180755615234,
      "learning_rate": 1.3747787008034863e-05,
      "loss": 4.1732,
      "step": 27120
    },
    {
      "epoch": 0.7288727663576515,
      "grad_norm": 2.487860679626465,
      "learning_rate": 1.3740977801988289e-05,
      "loss": 4.0698,
      "step": 27125
    },
    {
      "epoch": 0.7290071207846298,
      "grad_norm": 2.5451912879943848,
      "learning_rate": 1.3734168595941713e-05,
      "loss": 4.1232,
      "step": 27130
    },
    {
      "epoch": 0.7291414752116082,
      "grad_norm": 2.696127414703369,
      "learning_rate": 1.3727359389895139e-05,
      "loss": 3.983,
      "step": 27135
    },
    {
      "epoch": 0.7292758296385866,
      "grad_norm": 2.427971839904785,
      "learning_rate": 1.3720550183848562e-05,
      "loss": 4.0678,
      "step": 27140
    },
    {
      "epoch": 0.729410184065565,
      "grad_norm": 2.5199203491210938,
      "learning_rate": 1.371374097780199e-05,
      "loss": 4.0019,
      "step": 27145
    },
    {
      "epoch": 0.7295445384925433,
      "grad_norm": 2.5085549354553223,
      "learning_rate": 1.3706931771755414e-05,
      "loss": 4.042,
      "step": 27150
    },
    {
      "epoch": 0.7296788929195217,
      "grad_norm": 2.4763777256011963,
      "learning_rate": 1.370012256570884e-05,
      "loss": 4.0221,
      "step": 27155
    },
    {
      "epoch": 0.7298132473465001,
      "grad_norm": 2.6997640132904053,
      "learning_rate": 1.3693313359662263e-05,
      "loss": 4.045,
      "step": 27160
    },
    {
      "epoch": 0.7299476017734784,
      "grad_norm": 2.5226073265075684,
      "learning_rate": 1.3686504153615689e-05,
      "loss": 4.0735,
      "step": 27165
    },
    {
      "epoch": 0.7300819562004568,
      "grad_norm": 2.5416927337646484,
      "learning_rate": 1.3679694947569113e-05,
      "loss": 4.0002,
      "step": 27170
    },
    {
      "epoch": 0.7302163106274352,
      "grad_norm": 2.5626394748687744,
      "learning_rate": 1.367288574152254e-05,
      "loss": 4.1077,
      "step": 27175
    },
    {
      "epoch": 0.7303506650544136,
      "grad_norm": 2.611567735671997,
      "learning_rate": 1.3666076535475963e-05,
      "loss": 4.0402,
      "step": 27180
    },
    {
      "epoch": 0.7304850194813919,
      "grad_norm": 2.5577335357666016,
      "learning_rate": 1.365926732942939e-05,
      "loss": 4.1411,
      "step": 27185
    },
    {
      "epoch": 0.7306193739083703,
      "grad_norm": 2.5707876682281494,
      "learning_rate": 1.3652458123382814e-05,
      "loss": 4.0449,
      "step": 27190
    },
    {
      "epoch": 0.7307537283353487,
      "grad_norm": 2.495161771774292,
      "learning_rate": 1.364564891733624e-05,
      "loss": 4.1005,
      "step": 27195
    },
    {
      "epoch": 0.730888082762327,
      "grad_norm": 2.575080156326294,
      "learning_rate": 1.3638839711289664e-05,
      "loss": 4.1017,
      "step": 27200
    },
    {
      "epoch": 0.7310224371893054,
      "grad_norm": 2.792287826538086,
      "learning_rate": 1.3632030505243091e-05,
      "loss": 4.0119,
      "step": 27205
    },
    {
      "epoch": 0.7311567916162838,
      "grad_norm": 2.409008741378784,
      "learning_rate": 1.3625221299196513e-05,
      "loss": 4.015,
      "step": 27210
    },
    {
      "epoch": 0.7312911460432622,
      "grad_norm": 2.53548264503479,
      "learning_rate": 1.361841209314994e-05,
      "loss": 3.9637,
      "step": 27215
    },
    {
      "epoch": 0.7314255004702405,
      "grad_norm": 2.6357340812683105,
      "learning_rate": 1.3611602887103365e-05,
      "loss": 4.1447,
      "step": 27220
    },
    {
      "epoch": 0.7315598548972189,
      "grad_norm": 2.479964256286621,
      "learning_rate": 1.360479368105679e-05,
      "loss": 4.0892,
      "step": 27225
    },
    {
      "epoch": 0.7316942093241973,
      "grad_norm": 2.343773365020752,
      "learning_rate": 1.3597984475010214e-05,
      "loss": 3.9581,
      "step": 27230
    },
    {
      "epoch": 0.7318285637511756,
      "grad_norm": 2.45793080329895,
      "learning_rate": 1.3591175268963641e-05,
      "loss": 4.0773,
      "step": 27235
    },
    {
      "epoch": 0.731962918178154,
      "grad_norm": 2.63311505317688,
      "learning_rate": 1.3584366062917064e-05,
      "loss": 4.2286,
      "step": 27240
    },
    {
      "epoch": 0.7320972726051324,
      "grad_norm": 2.664856433868408,
      "learning_rate": 1.3577556856870491e-05,
      "loss": 3.9846,
      "step": 27245
    },
    {
      "epoch": 0.7322316270321108,
      "grad_norm": 2.6219139099121094,
      "learning_rate": 1.3570747650823915e-05,
      "loss": 4.1253,
      "step": 27250
    },
    {
      "epoch": 0.732365981459089,
      "grad_norm": 2.734297037124634,
      "learning_rate": 1.356393844477734e-05,
      "loss": 4.0539,
      "step": 27255
    },
    {
      "epoch": 0.7325003358860674,
      "grad_norm": 2.7208316326141357,
      "learning_rate": 1.3557129238730765e-05,
      "loss": 4.0312,
      "step": 27260
    },
    {
      "epoch": 0.7326346903130458,
      "grad_norm": 2.797855854034424,
      "learning_rate": 1.3550320032684189e-05,
      "loss": 4.2847,
      "step": 27265
    },
    {
      "epoch": 0.7327690447400241,
      "grad_norm": 2.7639501094818115,
      "learning_rate": 1.3543510826637614e-05,
      "loss": 4.0207,
      "step": 27270
    },
    {
      "epoch": 0.7329033991670025,
      "grad_norm": 2.6977858543395996,
      "learning_rate": 1.3536701620591038e-05,
      "loss": 3.9776,
      "step": 27275
    },
    {
      "epoch": 0.7330377535939809,
      "grad_norm": 2.822171449661255,
      "learning_rate": 1.3529892414544466e-05,
      "loss": 3.996,
      "step": 27280
    },
    {
      "epoch": 0.7331721080209593,
      "grad_norm": 2.54716157913208,
      "learning_rate": 1.3523083208497888e-05,
      "loss": 4.0811,
      "step": 27285
    },
    {
      "epoch": 0.7333064624479376,
      "grad_norm": 2.502875804901123,
      "learning_rate": 1.3516274002451315e-05,
      "loss": 4.2152,
      "step": 27290
    },
    {
      "epoch": 0.733440816874916,
      "grad_norm": 2.431347370147705,
      "learning_rate": 1.350946479640474e-05,
      "loss": 4.0868,
      "step": 27295
    },
    {
      "epoch": 0.7335751713018944,
      "grad_norm": 2.706016778945923,
      "learning_rate": 1.3502655590358165e-05,
      "loss": 4.0415,
      "step": 27300
    },
    {
      "epoch": 0.7337095257288727,
      "grad_norm": 2.4048750400543213,
      "learning_rate": 1.3495846384311589e-05,
      "loss": 4.0433,
      "step": 27305
    },
    {
      "epoch": 0.7338438801558511,
      "grad_norm": 2.784902334213257,
      "learning_rate": 1.3489037178265016e-05,
      "loss": 4.1096,
      "step": 27310
    },
    {
      "epoch": 0.7339782345828295,
      "grad_norm": 2.674161195755005,
      "learning_rate": 1.3482227972218438e-05,
      "loss": 3.9948,
      "step": 27315
    },
    {
      "epoch": 0.7341125890098079,
      "grad_norm": 2.5665650367736816,
      "learning_rate": 1.3475418766171866e-05,
      "loss": 4.0708,
      "step": 27320
    },
    {
      "epoch": 0.7342469434367862,
      "grad_norm": 2.6523518562316895,
      "learning_rate": 1.346860956012529e-05,
      "loss": 4.0739,
      "step": 27325
    },
    {
      "epoch": 0.7343812978637646,
      "grad_norm": 2.5107228755950928,
      "learning_rate": 1.3461800354078715e-05,
      "loss": 3.995,
      "step": 27330
    },
    {
      "epoch": 0.734515652290743,
      "grad_norm": 2.6661760807037354,
      "learning_rate": 1.345499114803214e-05,
      "loss": 3.9873,
      "step": 27335
    },
    {
      "epoch": 0.7346500067177214,
      "grad_norm": 2.4277145862579346,
      "learning_rate": 1.3448181941985567e-05,
      "loss": 4.1612,
      "step": 27340
    },
    {
      "epoch": 0.7347843611446997,
      "grad_norm": 2.481402635574341,
      "learning_rate": 1.3441372735938989e-05,
      "loss": 4.1054,
      "step": 27345
    },
    {
      "epoch": 0.7349187155716781,
      "grad_norm": 2.4681496620178223,
      "learning_rate": 1.3434563529892416e-05,
      "loss": 4.0591,
      "step": 27350
    },
    {
      "epoch": 0.7350530699986565,
      "grad_norm": 2.743870258331299,
      "learning_rate": 1.342775432384584e-05,
      "loss": 4.0498,
      "step": 27355
    },
    {
      "epoch": 0.7351874244256348,
      "grad_norm": 2.7099568843841553,
      "learning_rate": 1.3420945117799266e-05,
      "loss": 3.8206,
      "step": 27360
    },
    {
      "epoch": 0.7353217788526132,
      "grad_norm": 2.474907159805298,
      "learning_rate": 1.341413591175269e-05,
      "loss": 4.1962,
      "step": 27365
    },
    {
      "epoch": 0.7354561332795916,
      "grad_norm": 2.406263589859009,
      "learning_rate": 1.3407326705706116e-05,
      "loss": 4.1776,
      "step": 27370
    },
    {
      "epoch": 0.73559048770657,
      "grad_norm": 2.7678208351135254,
      "learning_rate": 1.340051749965954e-05,
      "loss": 4.1382,
      "step": 27375
    },
    {
      "epoch": 0.7357248421335483,
      "grad_norm": 2.481567859649658,
      "learning_rate": 1.3393708293612967e-05,
      "loss": 4.0872,
      "step": 27380
    },
    {
      "epoch": 0.7358591965605267,
      "grad_norm": 2.584256649017334,
      "learning_rate": 1.3386899087566391e-05,
      "loss": 4.0516,
      "step": 27385
    },
    {
      "epoch": 0.7359935509875051,
      "grad_norm": 2.579026460647583,
      "learning_rate": 1.3380089881519817e-05,
      "loss": 4.0055,
      "step": 27390
    },
    {
      "epoch": 0.7361279054144834,
      "grad_norm": 2.5526628494262695,
      "learning_rate": 1.337328067547324e-05,
      "loss": 4.0874,
      "step": 27395
    },
    {
      "epoch": 0.7362622598414618,
      "grad_norm": 2.367049217224121,
      "learning_rate": 1.3366471469426666e-05,
      "loss": 3.9098,
      "step": 27400
    },
    {
      "epoch": 0.7363966142684402,
      "grad_norm": 2.5932586193084717,
      "learning_rate": 1.335966226338009e-05,
      "loss": 4.0575,
      "step": 27405
    },
    {
      "epoch": 0.7365309686954186,
      "grad_norm": 2.5667803287506104,
      "learning_rate": 1.3352853057333514e-05,
      "loss": 4.0633,
      "step": 27410
    },
    {
      "epoch": 0.7366653231223969,
      "grad_norm": 2.404111385345459,
      "learning_rate": 1.3346043851286941e-05,
      "loss": 4.0838,
      "step": 27415
    },
    {
      "epoch": 0.7367996775493753,
      "grad_norm": 2.534876823425293,
      "learning_rate": 1.3339234645240364e-05,
      "loss": 4.1252,
      "step": 27420
    },
    {
      "epoch": 0.7369340319763537,
      "grad_norm": 2.597630262374878,
      "learning_rate": 1.3332425439193791e-05,
      "loss": 4.0765,
      "step": 27425
    },
    {
      "epoch": 0.7370683864033319,
      "grad_norm": 2.4507393836975098,
      "learning_rate": 1.3325616233147215e-05,
      "loss": 4.1558,
      "step": 27430
    },
    {
      "epoch": 0.7372027408303103,
      "grad_norm": 2.5739331245422363,
      "learning_rate": 1.331880702710064e-05,
      "loss": 4.0139,
      "step": 27435
    },
    {
      "epoch": 0.7373370952572887,
      "grad_norm": 2.312662124633789,
      "learning_rate": 1.3311997821054065e-05,
      "loss": 4.0445,
      "step": 27440
    },
    {
      "epoch": 0.7374714496842671,
      "grad_norm": 2.679407835006714,
      "learning_rate": 1.330518861500749e-05,
      "loss": 4.1218,
      "step": 27445
    },
    {
      "epoch": 0.7376058041112454,
      "grad_norm": 2.493534803390503,
      "learning_rate": 1.3298379408960914e-05,
      "loss": 4.0074,
      "step": 27450
    },
    {
      "epoch": 0.7377401585382238,
      "grad_norm": 2.6585192680358887,
      "learning_rate": 1.3291570202914342e-05,
      "loss": 4.0521,
      "step": 27455
    },
    {
      "epoch": 0.7378745129652022,
      "grad_norm": 2.637596845626831,
      "learning_rate": 1.3284760996867766e-05,
      "loss": 4.1324,
      "step": 27460
    },
    {
      "epoch": 0.7380088673921805,
      "grad_norm": 2.4800329208374023,
      "learning_rate": 1.3277951790821191e-05,
      "loss": 3.9652,
      "step": 27465
    },
    {
      "epoch": 0.7381432218191589,
      "grad_norm": 2.598944664001465,
      "learning_rate": 1.3271142584774615e-05,
      "loss": 4.026,
      "step": 27470
    },
    {
      "epoch": 0.7382775762461373,
      "grad_norm": 2.6920406818389893,
      "learning_rate": 1.3264333378728041e-05,
      "loss": 4.1017,
      "step": 27475
    },
    {
      "epoch": 0.7384119306731157,
      "grad_norm": 2.6152634620666504,
      "learning_rate": 1.3257524172681465e-05,
      "loss": 4.0182,
      "step": 27480
    },
    {
      "epoch": 0.738546285100094,
      "grad_norm": 2.587390899658203,
      "learning_rate": 1.3250714966634892e-05,
      "loss": 4.0262,
      "step": 27485
    },
    {
      "epoch": 0.7386806395270724,
      "grad_norm": 2.7638044357299805,
      "learning_rate": 1.3243905760588316e-05,
      "loss": 4.205,
      "step": 27490
    },
    {
      "epoch": 0.7388149939540508,
      "grad_norm": 2.6114258766174316,
      "learning_rate": 1.3237096554541742e-05,
      "loss": 4.028,
      "step": 27495
    },
    {
      "epoch": 0.7389493483810291,
      "grad_norm": 2.3654181957244873,
      "learning_rate": 1.3230287348495166e-05,
      "loss": 4.0319,
      "step": 27500
    }
  ],
  "logging_steps": 5,
  "max_steps": 37215,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0038542288754278e+17,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}