{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 28389,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010568450533046223,
      "grad_norm": 1.7654616832733154,
      "learning_rate": 2.25e-05,
      "loss": 1.7475404739379883,
      "step": 10
    },
    {
      "epoch": 0.0021136901066092445,
      "grad_norm": 1.2546119689941406,
      "learning_rate": 4.75e-05,
      "loss": 1.6405866622924805,
      "step": 20
    },
    {
      "epoch": 0.0031705351599138672,
      "grad_norm": 1.2028111219406128,
      "learning_rate": 4.999998758328575e-05,
      "loss": 1.5736541748046875,
      "step": 30
    },
    {
      "epoch": 0.004227380213218489,
      "grad_norm": 0.9781141877174377,
      "learning_rate": 4.999994466132639e-05,
      "loss": 1.4142214775085449,
      "step": 40
    },
    {
      "epoch": 0.005284225266523112,
      "grad_norm": 1.3441731929779053,
      "learning_rate": 4.999987108088179e-05,
      "loss": 1.3679624557495118,
      "step": 50
    },
    {
      "epoch": 0.0063410703198277345,
      "grad_norm": 1.2224844694137573,
      "learning_rate": 4.999976684204217e-05,
      "loss": 1.465764045715332,
      "step": 60
    },
    {
      "epoch": 0.007397915373132356,
      "grad_norm": 1.4091647863388062,
      "learning_rate": 4.999963194493537e-05,
      "loss": 1.2119648933410645,
      "step": 70
    },
    {
      "epoch": 0.008454760426436978,
      "grad_norm": 1.4000672101974487,
      "learning_rate": 4.999946638972683e-05,
      "loss": 1.3579352378845215,
      "step": 80
    },
    {
      "epoch": 0.009511605479741602,
      "grad_norm": 1.5409306287765503,
      "learning_rate": 4.999927017661955e-05,
      "loss": 1.514436912536621,
      "step": 90
    },
    {
      "epoch": 0.010568450533046224,
      "grad_norm": 1.3000372648239136,
      "learning_rate": 4.999904330585417e-05,
      "loss": 1.3766579627990723,
      "step": 100
    },
    {
      "epoch": 0.011625295586350845,
      "grad_norm": 1.6676361560821533,
      "learning_rate": 4.999878577770892e-05,
      "loss": 1.4116031646728515,
      "step": 110
    },
    {
      "epoch": 0.012682140639655469,
      "grad_norm": 1.0672098398208618,
      "learning_rate": 4.9998497592499605e-05,
      "loss": 1.3582697868347169,
      "step": 120
    },
    {
      "epoch": 0.01373898569296009,
      "grad_norm": 1.0912415981292725,
      "learning_rate": 4.999817875057964e-05,
      "loss": 1.3272669792175293,
      "step": 130
    },
    {
      "epoch": 0.014795830746264713,
      "grad_norm": 1.016467809677124,
      "learning_rate": 4.999782925234004e-05,
      "loss": 1.446809673309326,
      "step": 140
    },
    {
      "epoch": 0.015852675799569334,
      "grad_norm": 1.393985390663147,
      "learning_rate": 4.999744909820941e-05,
      "loss": 1.3340075492858887,
      "step": 150
    },
    {
      "epoch": 0.016909520852873956,
      "grad_norm": 1.1194339990615845,
      "learning_rate": 4.999703828865394e-05,
      "loss": 1.4149459838867187,
      "step": 160
    },
    {
      "epoch": 0.01796636590617858,
      "grad_norm": 0.95782470703125,
      "learning_rate": 4.9996596824177434e-05,
      "loss": 1.2714218139648437,
      "step": 170
    },
    {
      "epoch": 0.019023210959483203,
      "grad_norm": 1.3998883962631226,
      "learning_rate": 4.9996124705321265e-05,
      "loss": 1.281935977935791,
      "step": 180
    },
    {
      "epoch": 0.020080056012787825,
      "grad_norm": 1.029639720916748,
      "learning_rate": 4.999562193266443e-05,
      "loss": 1.3220934867858887,
      "step": 190
    },
    {
      "epoch": 0.021136901066092447,
      "grad_norm": 0.9435111284255981,
      "learning_rate": 4.999508850682349e-05,
      "loss": 1.3129037857055663,
      "step": 200
    },
    {
      "epoch": 0.02219374611939707,
      "grad_norm": 1.2938188314437866,
      "learning_rate": 4.9994524428452605e-05,
      "loss": 1.385861587524414,
      "step": 210
    },
    {
      "epoch": 0.02325059117270169,
      "grad_norm": 0.9571965932846069,
      "learning_rate": 4.999392969824353e-05,
      "loss": 1.413961410522461,
      "step": 220
    },
    {
      "epoch": 0.024307436226006316,
      "grad_norm": 1.133141279220581,
      "learning_rate": 4.9993304316925616e-05,
      "loss": 1.3622368812561034,
      "step": 230
    },
    {
      "epoch": 0.025364281279310938,
      "grad_norm": 1.1292999982833862,
      "learning_rate": 4.999264828526577e-05,
      "loss": 1.3130494117736817,
      "step": 240
    },
    {
      "epoch": 0.02642112633261556,
      "grad_norm": 1.829329013824463,
      "learning_rate": 4.999196160406855e-05,
      "loss": 1.3417440414428712,
      "step": 250
    },
    {
      "epoch": 0.02747797138592018,
      "grad_norm": 1.002306580543518,
      "learning_rate": 4.999124427417603e-05,
      "loss": 1.3035873413085937,
      "step": 260
    },
    {
      "epoch": 0.028534816439224803,
      "grad_norm": 1.365328073501587,
      "learning_rate": 4.999049629646791e-05,
      "loss": 1.3701492309570313,
      "step": 270
    },
    {
      "epoch": 0.029591661492529425,
      "grad_norm": 1.608991026878357,
      "learning_rate": 4.998971767186148e-05,
      "loss": 1.2301697731018066,
      "step": 280
    },
    {
      "epoch": 0.03064850654583405,
      "grad_norm": 1.2118799686431885,
      "learning_rate": 4.998890840131158e-05,
      "loss": 1.2982401847839355,
      "step": 290
    },
    {
      "epoch": 0.03170535159913867,
      "grad_norm": 1.1417475938796997,
      "learning_rate": 4.998806848581067e-05,
      "loss": 1.3410799980163575,
      "step": 300
    },
    {
      "epoch": 0.032762196652443294,
      "grad_norm": 5.897799015045166,
      "learning_rate": 4.9987197926388753e-05,
      "loss": 1.3490518569946288,
      "step": 310
    },
    {
      "epoch": 0.03381904170574791,
      "grad_norm": 0.9379751682281494,
      "learning_rate": 4.998629672411346e-05,
      "loss": 1.4124285697937011,
      "step": 320
    },
    {
      "epoch": 0.03487588675905254,
      "grad_norm": 1.1347147226333618,
      "learning_rate": 4.9985364880089945e-05,
      "loss": 1.3287044525146485,
      "step": 330
    },
    {
      "epoch": 0.03593273181235716,
      "grad_norm": 1.8065316677093506,
      "learning_rate": 4.998440239546099e-05,
      "loss": 1.2255881309509278,
      "step": 340
    },
    {
      "epoch": 0.03698957686566178,
      "grad_norm": 1.0114781856536865,
      "learning_rate": 4.9983409271406925e-05,
      "loss": 1.1949782371520996,
      "step": 350
    },
    {
      "epoch": 0.03804642191896641,
      "grad_norm": 1.2204539775848389,
      "learning_rate": 4.9982385509145647e-05,
      "loss": 1.3540396690368652,
      "step": 360
    },
    {
      "epoch": 0.039103266972271025,
      "grad_norm": 1.480690360069275,
      "learning_rate": 4.998133110993265e-05,
      "loss": 1.3325285911560059,
      "step": 370
    },
    {
      "epoch": 0.04016011202557565,
      "grad_norm": 0.9284271597862244,
      "learning_rate": 4.998024607506099e-05,
      "loss": 1.371429443359375,
      "step": 380
    },
    {
      "epoch": 0.041216957078880276,
      "grad_norm": 1.3816322088241577,
      "learning_rate": 4.9979130405861284e-05,
      "loss": 1.3178449630737306,
      "step": 390
    },
    {
      "epoch": 0.042273802132184894,
      "grad_norm": 0.913960874080658,
      "learning_rate": 4.997798410370173e-05,
      "loss": 1.3649686813354491,
      "step": 400
    },
    {
      "epoch": 0.04333064718548952,
      "grad_norm": 0.9822927713394165,
      "learning_rate": 4.9976807169988086e-05,
      "loss": 1.4719487190246583,
      "step": 410
    },
    {
      "epoch": 0.04438749223879414,
      "grad_norm": 1.3878390789031982,
      "learning_rate": 4.997559960616367e-05,
      "loss": 1.2280084609985351,
      "step": 420
    },
    {
      "epoch": 0.04544433729209876,
      "grad_norm": 1.32184898853302,
      "learning_rate": 4.997436141370938e-05,
      "loss": 1.312828826904297,
      "step": 430
    },
    {
      "epoch": 0.04650118234540338,
      "grad_norm": 1.111492395401001,
      "learning_rate": 4.997309259414365e-05,
      "loss": 1.3730249404907227,
      "step": 440
    },
    {
      "epoch": 0.04755802739870801,
      "grad_norm": 1.0347301959991455,
      "learning_rate": 4.997179314902249e-05,
      "loss": 1.352734088897705,
      "step": 450
    },
    {
      "epoch": 0.04861487245201263,
      "grad_norm": 1.1026866436004639,
      "learning_rate": 4.997046307993947e-05,
      "loss": 1.3316572189331055,
      "step": 460
    },
    {
      "epoch": 0.04967171750531725,
      "grad_norm": 0.9227033257484436,
      "learning_rate": 4.9969102388525706e-05,
      "loss": 1.278825283050537,
      "step": 470
    },
    {
      "epoch": 0.050728562558621876,
      "grad_norm": 1.357617735862732,
      "learning_rate": 4.9967711076449876e-05,
      "loss": 1.2980709075927734,
      "step": 480
    },
    {
      "epoch": 0.051785407611926494,
      "grad_norm": 1.0833156108856201,
      "learning_rate": 4.9966289145418196e-05,
      "loss": 1.1954431533813477,
      "step": 490
    },
    {
      "epoch": 0.05284225266523112,
      "grad_norm": 0.7908650040626526,
      "learning_rate": 4.996483659717444e-05,
      "loss": 1.3347193717956543,
      "step": 500
    },
    {
      "epoch": 0.05284225266523112,
      "eval_loss": 0.6353857517242432,
      "eval_runtime": 3327.535,
      "eval_samples_per_second": 2.676,
      "eval_steps_per_second": 2.676,
      "step": 500
    },
    {
      "epoch": 0.05389909771853574,
      "grad_norm": 1.2030062675476074,
      "learning_rate": 4.996335343349994e-05,
      "loss": 1.369196605682373,
      "step": 510
    },
    {
      "epoch": 0.05495594277184036,
      "grad_norm": 1.775600552558899,
      "learning_rate": 4.9961839656213555e-05,
      "loss": 1.2971681594848632,
      "step": 520
    },
    {
      "epoch": 0.05601278782514499,
      "grad_norm": 1.265978217124939,
      "learning_rate": 4.996029526717169e-05,
      "loss": 1.3606646537780762,
      "step": 530
    },
    {
      "epoch": 0.05706963287844961,
      "grad_norm": 1.1371171474456787,
      "learning_rate": 4.99587202682683e-05,
      "loss": 1.2797603607177734,
      "step": 540
    },
    {
      "epoch": 0.05812647793175423,
      "grad_norm": 0.9954401254653931,
      "learning_rate": 4.995711466143488e-05,
      "loss": 1.2673745155334473,
      "step": 550
    },
    {
      "epoch": 0.05918332298505885,
      "grad_norm": 1.0505359172821045,
      "learning_rate": 4.995547844864043e-05,
      "loss": 1.3346149444580078,
      "step": 560
    },
    {
      "epoch": 0.060240168038363476,
      "grad_norm": 0.9504349231719971,
      "learning_rate": 4.995381163189153e-05,
      "loss": 1.2310306549072265,
      "step": 570
    },
    {
      "epoch": 0.0612970130916681,
      "grad_norm": 0.9347895383834839,
      "learning_rate": 4.995211421323226e-05,
      "loss": 1.2528314590454102,
      "step": 580
    },
    {
      "epoch": 0.06235385814497272,
      "grad_norm": 0.9463417530059814,
      "learning_rate": 4.995038619474423e-05,
      "loss": 1.287467098236084,
      "step": 590
    },
    {
      "epoch": 0.06341070319827734,
      "grad_norm": 1.202889323234558,
      "learning_rate": 4.994862757854659e-05,
      "loss": 1.165782642364502,
      "step": 600
    },
    {
      "epoch": 0.06446754825158196,
      "grad_norm": 0.8427596688270569,
      "learning_rate": 4.9946838366796e-05,
      "loss": 1.266599178314209,
      "step": 610
    },
    {
      "epoch": 0.06552439330488659,
      "grad_norm": 0.98178631067276,
      "learning_rate": 4.994501856168665e-05,
      "loss": 1.2369131088256835,
      "step": 620
    },
    {
      "epoch": 0.06658123835819121,
      "grad_norm": 0.7207362055778503,
      "learning_rate": 4.994316816545023e-05,
      "loss": 1.2621751785278321,
      "step": 630
    },
    {
      "epoch": 0.06763808341149583,
      "grad_norm": 0.8623623847961426,
      "learning_rate": 4.994128718035598e-05,
      "loss": 1.324796772003174,
      "step": 640
    },
    {
      "epoch": 0.06869492846480045,
      "grad_norm": 1.3600107431411743,
      "learning_rate": 4.993937560871063e-05,
      "loss": 1.130089282989502,
      "step": 650
    },
    {
      "epoch": 0.06975177351810508,
      "grad_norm": 1.3267031908035278,
      "learning_rate": 4.99374334528584e-05,
      "loss": 1.1540260314941406,
      "step": 660
    },
    {
      "epoch": 0.0708086185714097,
      "grad_norm": 0.8460550904273987,
      "learning_rate": 4.993546071518105e-05,
      "loss": 1.2651094436645507,
      "step": 670
    },
    {
      "epoch": 0.07186546362471433,
      "grad_norm": 1.0716408491134644,
      "learning_rate": 4.993345739809783e-05,
      "loss": 1.2272722244262695,
      "step": 680
    },
    {
      "epoch": 0.07292230867801894,
      "grad_norm": 1.6539790630340576,
      "learning_rate": 4.9931423504065496e-05,
      "loss": 1.3102322578430177,
      "step": 690
    },
    {
      "epoch": 0.07397915373132356,
      "grad_norm": 0.847364068031311,
      "learning_rate": 4.9929359035578295e-05,
      "loss": 1.2594082832336426,
      "step": 700
    },
    {
      "epoch": 0.07503599878462819,
      "grad_norm": 0.8033567070960999,
      "learning_rate": 4.992726399516797e-05,
      "loss": 1.116991901397705,
      "step": 710
    },
    {
      "epoch": 0.07609284383793281,
      "grad_norm": 1.7280633449554443,
      "learning_rate": 4.992513838540376e-05,
      "loss": 1.362265682220459,
      "step": 720
    },
    {
      "epoch": 0.07714968889123744,
      "grad_norm": 0.9090099334716797,
      "learning_rate": 4.9922982208892386e-05,
      "loss": 1.269299602508545,
      "step": 730
    },
    {
      "epoch": 0.07820653394454205,
      "grad_norm": 1.0359073877334595,
      "learning_rate": 4.992079546827806e-05,
      "loss": 1.2931873321533203,
      "step": 740
    },
    {
      "epoch": 0.07926337899784668,
      "grad_norm": 0.5516844391822815,
      "learning_rate": 4.991857816624248e-05,
      "loss": 1.1876733779907227,
      "step": 750
    },
    {
      "epoch": 0.0803202240511513,
      "grad_norm": 0.8674123287200928,
      "learning_rate": 4.991633030550481e-05,
      "loss": 1.1414837837219238,
      "step": 760
    },
    {
      "epoch": 0.08137706910445593,
      "grad_norm": 1.4054392576217651,
      "learning_rate": 4.99140518888217e-05,
      "loss": 1.3033767700195313,
      "step": 770
    },
    {
      "epoch": 0.08243391415776055,
      "grad_norm": 0.9793580174446106,
      "learning_rate": 4.991174291898727e-05,
      "loss": 1.2552642822265625,
      "step": 780
    },
    {
      "epoch": 0.08349075921106516,
      "grad_norm": 1.023809552192688,
      "learning_rate": 4.99094033988331e-05,
      "loss": 1.2558234214782715,
      "step": 790
    },
    {
      "epoch": 0.08454760426436979,
      "grad_norm": 0.8998658657073975,
      "learning_rate": 4.9907033331228235e-05,
      "loss": 1.2494474411010743,
      "step": 800
    },
    {
      "epoch": 0.08560444931767441,
      "grad_norm": 0.9202524423599243,
      "learning_rate": 4.990463271907921e-05,
      "loss": 1.3264954566955567,
      "step": 810
    },
    {
      "epoch": 0.08666129437097904,
      "grad_norm": 1.007606029510498,
      "learning_rate": 4.990220156532999e-05,
      "loss": 1.3033018112182617,
      "step": 820
    },
    {
      "epoch": 0.08771813942428365,
      "grad_norm": 0.9622219204902649,
      "learning_rate": 4.9899739872961994e-05,
      "loss": 1.2520596504211425,
      "step": 830
    },
    {
      "epoch": 0.08877498447758828,
      "grad_norm": 1.7036700248718262,
      "learning_rate": 4.989724764499409e-05,
      "loss": 1.318963623046875,
      "step": 840
    },
    {
      "epoch": 0.0898318295308929,
      "grad_norm": 0.9336941838264465,
      "learning_rate": 4.989472488448263e-05,
      "loss": 1.1108043670654297,
      "step": 850
    },
    {
      "epoch": 0.09088867458419753,
      "grad_norm": 0.9317730069160461,
      "learning_rate": 4.9892171594521364e-05,
      "loss": 1.2733315467834472,
      "step": 860
    },
    {
      "epoch": 0.09194551963750215,
      "grad_norm": 0.8659337759017944,
      "learning_rate": 4.98895877782415e-05,
      "loss": 1.2289716720581054,
      "step": 870
    },
    {
      "epoch": 0.09300236469080676,
      "grad_norm": 0.9475213885307312,
      "learning_rate": 4.988697343881168e-05,
      "loss": 1.2194920539855958,
      "step": 880
    },
    {
      "epoch": 0.09405920974411139,
      "grad_norm": 0.9211010932922363,
      "learning_rate": 4.9884328579437976e-05,
      "loss": 1.4439587593078613,
      "step": 890
    },
    {
      "epoch": 0.09511605479741601,
      "grad_norm": 0.776827335357666,
      "learning_rate": 4.98816532033639e-05,
      "loss": 1.0826983451843262,
      "step": 900
    },
    {
      "epoch": 0.09617289985072064,
      "grad_norm": 1.0454965829849243,
      "learning_rate": 4.987894731387037e-05,
      "loss": 1.151451873779297,
      "step": 910
    },
    {
      "epoch": 0.09722974490402526,
      "grad_norm": 1.0449923276901245,
      "learning_rate": 4.9876210914275745e-05,
      "loss": 1.2453297615051269,
      "step": 920
    },
    {
      "epoch": 0.09828658995732988,
      "grad_norm": 0.8443463444709778,
      "learning_rate": 4.9873444007935767e-05,
      "loss": 1.387489128112793,
      "step": 930
    },
    {
      "epoch": 0.0993434350106345,
      "grad_norm": 0.8886481523513794,
      "learning_rate": 4.9870646598243625e-05,
      "loss": 1.34130916595459,
      "step": 940
    },
    {
      "epoch": 0.10040028006393913,
      "grad_norm": 0.7917105555534363,
      "learning_rate": 4.98678186886299e-05,
      "loss": 1.2625954627990723,
      "step": 950
    },
    {
      "epoch": 0.10145712511724375,
      "grad_norm": 0.9016518592834473,
      "learning_rate": 4.986496028256257e-05,
      "loss": 1.2145988464355468,
      "step": 960
    },
    {
      "epoch": 0.10251397017054838,
      "grad_norm": 0.8979685306549072,
      "learning_rate": 4.9862071383547014e-05,
      "loss": 1.237309169769287,
      "step": 970
    },
    {
      "epoch": 0.10357081522385299,
      "grad_norm": 0.8628866672515869,
      "learning_rate": 4.985915199512602e-05,
      "loss": 1.330723190307617,
      "step": 980
    },
    {
      "epoch": 0.10462766027715761,
      "grad_norm": 0.8702941536903381,
      "learning_rate": 4.985620212087976e-05,
      "loss": 1.1970738410949706,
      "step": 990
    },
    {
      "epoch": 0.10568450533046224,
      "grad_norm": 0.843571662902832,
      "learning_rate": 4.9853221764425784e-05,
      "loss": 1.319435977935791,
      "step": 1000
    },
    {
      "epoch": 0.10568450533046224,
      "eval_loss": 0.6109004020690918,
      "eval_runtime": 3326.2181,
      "eval_samples_per_second": 2.677,
      "eval_steps_per_second": 2.677,
      "step": 1000
    },
    {
      "epoch": 0.10674135038376686,
      "grad_norm": 1.1977611780166626,
      "learning_rate": 4.9850210929419024e-05,
      "loss": 1.201138401031494,
      "step": 1010
    },
    {
      "epoch": 0.10779819543707148,
      "grad_norm": 1.4552743434906006,
      "learning_rate": 4.98471696195518e-05,
      "loss": 1.3288004875183106,
      "step": 1020
    },
    {
      "epoch": 0.1088550404903761,
      "grad_norm": 0.8560193777084351,
      "learning_rate": 4.984409783855379e-05,
      "loss": 1.359072780609131,
      "step": 1030
    },
    {
      "epoch": 0.10991188554368073,
      "grad_norm": 0.8196094036102295,
      "learning_rate": 4.984099559019207e-05,
      "loss": 1.2987495422363282,
      "step": 1040
    },
    {
      "epoch": 0.11096873059698535,
      "grad_norm": 1.0730071067810059,
      "learning_rate": 4.9837862878271037e-05,
      "loss": 0.9890561103820801,
      "step": 1050
    },
    {
      "epoch": 0.11202557565028998,
      "grad_norm": 1.5064266920089722,
      "learning_rate": 4.983469970663247e-05,
      "loss": 1.296586036682129,
      "step": 1060
    },
    {
      "epoch": 0.11308242070359459,
      "grad_norm": 0.9763716459274292,
      "learning_rate": 4.98315060791555e-05,
      "loss": 1.3068012237548827,
      "step": 1070
    },
    {
      "epoch": 0.11413926575689921,
      "grad_norm": 0.9289076328277588,
      "learning_rate": 4.982828199975662e-05,
      "loss": 1.3146883010864259,
      "step": 1080
    },
    {
      "epoch": 0.11519611081020384,
      "grad_norm": 1.173413872718811,
      "learning_rate": 4.982502747238964e-05,
      "loss": 1.3376843452453613,
      "step": 1090
    },
    {
      "epoch": 0.11625295586350846,
      "grad_norm": 1.2227745056152344,
      "learning_rate": 4.9821742501045724e-05,
      "loss": 1.3748673439025878,
      "step": 1100
    },
    {
      "epoch": 0.11730980091681309,
      "grad_norm": 1.4726097583770752,
      "learning_rate": 4.9818427089753384e-05,
      "loss": 1.1811174392700194,
      "step": 1110
    },
    {
      "epoch": 0.1183666459701177,
      "grad_norm": 1.2403807640075684,
      "learning_rate": 4.981508124257843e-05,
      "loss": 1.1157828330993653,
      "step": 1120
    },
    {
      "epoch": 0.11942349102342233,
      "grad_norm": 1.0265569686889648,
      "learning_rate": 4.981170496362403e-05,
      "loss": 1.1202966690063476,
      "step": 1130
    },
    {
      "epoch": 0.12048033607672695,
      "grad_norm": 1.0560208559036255,
      "learning_rate": 4.9808298257030654e-05,
      "loss": 1.214322280883789,
      "step": 1140
    },
    {
      "epoch": 0.12153718113003158,
      "grad_norm": 0.9267060160636902,
      "learning_rate": 4.980486112697609e-05,
      "loss": 1.2062045097351075,
      "step": 1150
    },
    {
      "epoch": 0.1225940261833362,
      "grad_norm": 1.5877197980880737,
      "learning_rate": 4.980139357767543e-05,
      "loss": 1.2259860992431642,
      "step": 1160
    },
    {
      "epoch": 0.12365087123664081,
      "grad_norm": 1.4388036727905273,
      "learning_rate": 4.9797895613381077e-05,
      "loss": 1.2811634063720703,
      "step": 1170
    },
    {
      "epoch": 0.12470771628994544,
      "grad_norm": 0.9840176105499268,
      "learning_rate": 4.9794367238382734e-05,
      "loss": 1.3244059562683106,
      "step": 1180
    },
    {
      "epoch": 0.12576456134325006,
      "grad_norm": 1.1563103199005127,
      "learning_rate": 4.97908084570074e-05,
      "loss": 1.1569110870361328,
      "step": 1190
    },
    {
      "epoch": 0.12682140639655468,
      "grad_norm": 1.3868827819824219,
      "learning_rate": 4.978721927361935e-05,
      "loss": 1.1699362754821778,
      "step": 1200
    },
    {
      "epoch": 0.12787825144985931,
      "grad_norm": 0.7807841897010803,
      "learning_rate": 4.978359969262016e-05,
      "loss": 1.1010542869567872,
      "step": 1210
    },
    {
      "epoch": 0.12893509650316393,
      "grad_norm": 1.5077811479568481,
      "learning_rate": 4.977994971844867e-05,
      "loss": 1.1900751113891601,
      "step": 1220
    },
    {
      "epoch": 0.12999194155646857,
      "grad_norm": 1.1352185010910034,
      "learning_rate": 4.9776269355581e-05,
      "loss": 1.3362363815307616,
      "step": 1230
    },
    {
      "epoch": 0.13104878660977318,
      "grad_norm": 1.0839064121246338,
      "learning_rate": 4.9772558608530516e-05,
      "loss": 1.31423978805542,
      "step": 1240
    },
    {
      "epoch": 0.1321056316630778,
      "grad_norm": 1.2738856077194214,
      "learning_rate": 4.976881748184789e-05,
      "loss": 1.1725255012512208,
      "step": 1250
    },
    {
      "epoch": 0.13316247671638243,
      "grad_norm": 1.0360294580459595,
      "learning_rate": 4.976504598012101e-05,
      "loss": 1.1044158935546875,
      "step": 1260
    },
    {
      "epoch": 0.13421932176968704,
      "grad_norm": 0.8492978811264038,
      "learning_rate": 4.976124410797503e-05,
      "loss": 1.1926722526550293,
      "step": 1270
    },
    {
      "epoch": 0.13527616682299165,
      "grad_norm": 1.3604161739349365,
      "learning_rate": 4.975741187007235e-05,
      "loss": 1.015962791442871,
      "step": 1280
    },
    {
      "epoch": 0.1363330118762963,
      "grad_norm": 0.7425869107246399,
      "learning_rate": 4.9753549271112585e-05,
      "loss": 1.2769925117492675,
      "step": 1290
    },
    {
      "epoch": 0.1373898569296009,
      "grad_norm": 0.8644911050796509,
      "learning_rate": 4.974965631583263e-05,
      "loss": 1.2593586921691895,
      "step": 1300
    },
    {
      "epoch": 0.13844670198290554,
      "grad_norm": 1.004532814025879,
      "learning_rate": 4.9745733009006566e-05,
      "loss": 1.2348335266113282,
      "step": 1310
    },
    {
      "epoch": 0.13950354703621015,
      "grad_norm": 0.9020411372184753,
      "learning_rate": 4.974177935544571e-05,
      "loss": 1.2660881996154785,
      "step": 1320
    },
    {
      "epoch": 0.14056039208951476,
      "grad_norm": 1.0645228624343872,
      "learning_rate": 4.9737795359998594e-05,
      "loss": 1.2413216590881349,
      "step": 1330
    },
    {
      "epoch": 0.1416172371428194,
      "grad_norm": 0.8682624101638794,
      "learning_rate": 4.973378102755096e-05,
      "loss": 1.2792637825012207,
      "step": 1340
    },
    {
      "epoch": 0.142674082196124,
      "grad_norm": 0.9943913221359253,
      "learning_rate": 4.9729736363025756e-05,
      "loss": 1.0542287826538086,
      "step": 1350
    },
    {
      "epoch": 0.14373092724942865,
      "grad_norm": 1.0856118202209473,
      "learning_rate": 4.9725661371383124e-05,
      "loss": 1.2900238037109375,
      "step": 1360
    },
    {
      "epoch": 0.14478777230273326,
      "grad_norm": 0.9085684418678284,
      "learning_rate": 4.972155605762039e-05,
      "loss": 1.3020459175109864,
      "step": 1370
    },
    {
      "epoch": 0.14584461735603788,
      "grad_norm": 1.1672264337539673,
      "learning_rate": 4.971742042677208e-05,
      "loss": 1.235660457611084,
      "step": 1380
    },
    {
      "epoch": 0.14690146240934251,
      "grad_norm": 0.9675402045249939,
      "learning_rate": 4.97132544839099e-05,
      "loss": 1.312955093383789,
      "step": 1390
    },
    {
      "epoch": 0.14795830746264713,
      "grad_norm": 1.0820773839950562,
      "learning_rate": 4.97090582341427e-05,
      "loss": 1.3633664131164551,
      "step": 1400
    },
    {
      "epoch": 0.14901515251595177,
      "grad_norm": 0.803695797920227,
      "learning_rate": 4.970483168261654e-05,
      "loss": 1.0210362434387208,
      "step": 1410
    },
    {
      "epoch": 0.15007199756925638,
      "grad_norm": 0.8423833250999451,
      "learning_rate": 4.97005748345146e-05,
      "loss": 1.2016429901123047,
      "step": 1420
    },
    {
      "epoch": 0.151128842622561,
      "grad_norm": 0.8605319261550903,
      "learning_rate": 4.969628769505725e-05,
      "loss": 1.214029312133789,
      "step": 1430
    },
    {
      "epoch": 0.15218568767586563,
      "grad_norm": 0.860576331615448,
      "learning_rate": 4.969197026950197e-05,
      "loss": 1.1323326110839844,
      "step": 1440
    },
    {
      "epoch": 0.15324253272917024,
      "grad_norm": 0.8505045175552368,
      "learning_rate": 4.9687622563143426e-05,
      "loss": 1.166429615020752,
      "step": 1450
    },
    {
      "epoch": 0.15429937778247488,
      "grad_norm": 1.6121652126312256,
      "learning_rate": 4.968324458131337e-05,
      "loss": 1.1942463874816895,
      "step": 1460
    },
    {
      "epoch": 0.1553562228357795,
      "grad_norm": 0.8112436532974243,
      "learning_rate": 4.9678836329380716e-05,
      "loss": 1.1984683990478515,
      "step": 1470
    },
    {
      "epoch": 0.1564130678890841,
      "grad_norm": 0.941604733467102,
      "learning_rate": 4.967439781275149e-05,
      "loss": 1.3117964744567872,
      "step": 1480
    },
    {
      "epoch": 0.15746991294238874,
      "grad_norm": 1.110789179801941,
      "learning_rate": 4.9669929036868826e-05,
      "loss": 1.338887596130371,
      "step": 1490
    },
    {
      "epoch": 0.15852675799569335,
      "grad_norm": 1.0409533977508545,
      "learning_rate": 4.9665430007212976e-05,
      "loss": 1.2026755332946777,
      "step": 1500
    },
    {
      "epoch": 0.15852675799569335,
      "eval_loss": 0.5906086564064026,
      "eval_runtime": 3058.4056,
      "eval_samples_per_second": 2.912,
      "eval_steps_per_second": 2.912,
      "step": 1500
    },
    {
      "epoch": 0.159583603048998,
      "grad_norm": 0.8225815892219543,
      "learning_rate": 4.966090072930129e-05,
      "loss": 1.2641072273254395,
      "step": 1510
    },
    {
      "epoch": 0.1606404481023026,
      "grad_norm": 1.1553276777267456,
      "learning_rate": 4.96563412086882e-05,
      "loss": 1.267894172668457,
      "step": 1520
    },
    {
      "epoch": 0.1616972931556072,
      "grad_norm": 0.7743565440177917,
      "learning_rate": 4.9651751450965265e-05,
      "loss": 1.1669331550598145,
      "step": 1530
    },
    {
      "epoch": 0.16275413820891185,
      "grad_norm": 0.8377006649971008,
      "learning_rate": 4.9647131461761074e-05,
      "loss": 1.3365713119506837,
      "step": 1540
    },
    {
      "epoch": 0.16381098326221646,
      "grad_norm": 1.1922225952148438,
      "learning_rate": 4.964248124674133e-05,
      "loss": 1.1858696937561035,
      "step": 1550
    },
    {
      "epoch": 0.1648678283155211,
      "grad_norm": 0.8595351576805115,
      "learning_rate": 4.963780081160878e-05,
      "loss": 1.2868152618408204,
      "step": 1560
    },
    {
      "epoch": 0.16592467336882571,
      "grad_norm": 0.911601722240448,
      "learning_rate": 4.963309016210324e-05,
      "loss": 1.2611271858215332,
      "step": 1570
    },
    {
      "epoch": 0.16698151842213033,
      "grad_norm": 0.9030196666717529,
      "learning_rate": 4.962834930400156e-05,
      "loss": 1.2623150825500489,
      "step": 1580
    },
    {
      "epoch": 0.16803836347543497,
      "grad_norm": 1.3613134622573853,
      "learning_rate": 4.962357824311769e-05,
      "loss": 1.287333869934082,
      "step": 1590
    },
    {
      "epoch": 0.16909520852873958,
      "grad_norm": 1.099332571029663,
      "learning_rate": 4.961877698530256e-05,
      "loss": 1.1825684547424316,
      "step": 1600
    },
    {
      "epoch": 0.17015205358204422,
      "grad_norm": 0.8475016951560974,
      "learning_rate": 4.961394553644416e-05,
      "loss": 1.0798020362854004,
      "step": 1610
    },
    {
      "epoch": 0.17120889863534883,
      "grad_norm": 1.004656195640564,
      "learning_rate": 4.96090839024675e-05,
      "loss": 1.2272758483886719,
      "step": 1620
    },
    {
      "epoch": 0.17226574368865344,
      "grad_norm": 0.9492601156234741,
      "learning_rate": 4.96041920893346e-05,
      "loss": 1.2727628707885743,
      "step": 1630
    },
    {
      "epoch": 0.17332258874195808,
      "grad_norm": 1.2830818891525269,
      "learning_rate": 4.95992701030445e-05,
      "loss": 1.168790340423584,
      "step": 1640
    },
    {
      "epoch": 0.1743794337952627,
      "grad_norm": 0.8927351832389832,
      "learning_rate": 4.959431794963324e-05,
      "loss": 1.1952330589294433,
      "step": 1650
    },
    {
      "epoch": 0.1754362788485673,
      "grad_norm": 1.1559181213378906,
      "learning_rate": 4.958933563517386e-05,
      "loss": 1.0206067085266113,
      "step": 1660
    },
    {
      "epoch": 0.17649312390187194,
      "grad_norm": 0.9883691668510437,
      "learning_rate": 4.958432316577638e-05,
      "loss": 1.268683910369873,
      "step": 1670
    },
    {
      "epoch": 0.17754996895517655,
      "grad_norm": 0.7860012054443359,
      "learning_rate": 4.957928054758779e-05,
      "loss": 1.3326000213623046,
      "step": 1680
    },
    {
      "epoch": 0.1786068140084812,
      "grad_norm": 0.8263247013092041,
      "learning_rate": 4.957420778679208e-05,
      "loss": 1.169423770904541,
      "step": 1690
    },
    {
      "epoch": 0.1796636590617858,
      "grad_norm": 1.0402946472167969,
      "learning_rate": 4.956910488961019e-05,
      "loss": 1.1556696891784668,
      "step": 1700
    },
    {
      "epoch": 0.1807205041150904,
      "grad_norm": 1.3067480325698853,
      "learning_rate": 4.9563971862299994e-05,
      "loss": 1.091926383972168,
      "step": 1710
    },
    {
      "epoch": 0.18177734916839505,
      "grad_norm": 1.1822742223739624,
      "learning_rate": 4.955880871115637e-05,
      "loss": 1.2627655029296876,
      "step": 1720
    },
    {
      "epoch": 0.18283419422169966,
      "grad_norm": 1.082460641860962,
      "learning_rate": 4.95536154425111e-05,
      "loss": 1.2526678085327148,
      "step": 1730
    },
    {
      "epoch": 0.1838910392750043,
      "grad_norm": 0.9203020930290222,
      "learning_rate": 4.954839206273289e-05,
      "loss": 1.2140596389770508,
      "step": 1740
    },
    {
      "epoch": 0.18494788432830891,
      "grad_norm": 1.0754363536834717,
      "learning_rate": 4.954313857822742e-05,
      "loss": 1.2126280784606933,
      "step": 1750
    },
    {
      "epoch": 0.18600472938161353,
      "grad_norm": 1.18026864528656,
      "learning_rate": 4.953785499543724e-05,
      "loss": 1.2314389228820801,
      "step": 1760
    },
    {
      "epoch": 0.18706157443491817,
      "grad_norm": 1.1377744674682617,
      "learning_rate": 4.953254132084183e-05,
      "loss": 1.106722640991211,
      "step": 1770
    },
    {
      "epoch": 0.18811841948822278,
      "grad_norm": 0.7564743161201477,
      "learning_rate": 4.952719756095758e-05,
      "loss": 1.2234151840209961,
      "step": 1780
    },
    {
      "epoch": 0.18917526454152742,
      "grad_norm": 0.8391842246055603,
      "learning_rate": 4.9521823722337765e-05,
      "loss": 1.274850559234619,
      "step": 1790
    },
    {
      "epoch": 0.19023210959483203,
      "grad_norm": 1.3820524215698242,
      "learning_rate": 4.951641981157255e-05,
      "loss": 1.1575454711914062,
      "step": 1800
    },
    {
      "epoch": 0.19128895464813664,
      "grad_norm": 0.9176732897758484,
      "learning_rate": 4.951098583528897e-05,
      "loss": 1.217789363861084,
      "step": 1810
    },
    {
      "epoch": 0.19234579970144128,
      "grad_norm": 1.01221764087677,
      "learning_rate": 4.9505521800150954e-05,
      "loss": 1.3363012313842773,
      "step": 1820
    },
    {
      "epoch": 0.1934026447547459,
      "grad_norm": 1.4353400468826294,
      "learning_rate": 4.9500027712859274e-05,
      "loss": 1.2144241333007812,
      "step": 1830
    },
    {
      "epoch": 0.19445948980805053,
      "grad_norm": 1.649135947227478,
      "learning_rate": 4.9494503580151555e-05,
      "loss": 1.2646055221557617,
      "step": 1840
    },
    {
      "epoch": 0.19551633486135514,
      "grad_norm": 0.893892228603363,
      "learning_rate": 4.948894940880227e-05,
      "loss": 1.1833094596862792,
      "step": 1850
    },
    {
      "epoch": 0.19657317991465975,
      "grad_norm": 1.0424269437789917,
      "learning_rate": 4.948336520562274e-05,
      "loss": 1.193265724182129,
      "step": 1860
    },
    {
      "epoch": 0.1976300249679644,
      "grad_norm": 0.9494489431381226,
      "learning_rate": 4.9477750977461126e-05,
      "loss": 1.2495148658752442,
      "step": 1870
    },
    {
      "epoch": 0.198686870021269,
      "grad_norm": 1.0949044227600098,
      "learning_rate": 4.9472106731202364e-05,
      "loss": 1.1997599601745605,
      "step": 1880
    },
    {
      "epoch": 0.19974371507457364,
      "grad_norm": 1.0748324394226074,
      "learning_rate": 4.9466432473768233e-05,
      "loss": 1.0886706352233886,
      "step": 1890
    },
    {
      "epoch": 0.20080056012787825,
      "grad_norm": 1.038313388824463,
      "learning_rate": 4.9460728212117334e-05,
      "loss": 1.1909976959228517,
      "step": 1900
    },
    {
      "epoch": 0.20185740518118286,
      "grad_norm": 1.116932988166809,
      "learning_rate": 4.945499395324503e-05,
      "loss": 1.1579012870788574,
      "step": 1910
    },
    {
      "epoch": 0.2029142502344875,
      "grad_norm": 0.7205058336257935,
      "learning_rate": 4.9449229704183484e-05,
      "loss": 1.2349910736083984,
      "step": 1920
    },
    {
      "epoch": 0.20397109528779211,
      "grad_norm": 1.1264851093292236,
      "learning_rate": 4.944343547200164e-05,
      "loss": 1.1743742942810058,
      "step": 1930
    },
    {
      "epoch": 0.20502794034109675,
      "grad_norm": 0.970874547958374,
      "learning_rate": 4.943761126380522e-05,
      "loss": 1.2700891494750977,
      "step": 1940
    },
    {
      "epoch": 0.20608478539440137,
      "grad_norm": 1.054227352142334,
      "learning_rate": 4.9431757086736674e-05,
      "loss": 1.1778717041015625,
      "step": 1950
    },
    {
      "epoch": 0.20714163044770598,
      "grad_norm": 1.2120506763458252,
      "learning_rate": 4.942587294797524e-05,
      "loss": 1.1417316436767577,
      "step": 1960
    },
    {
      "epoch": 0.20819847550101062,
      "grad_norm": 1.1574993133544922,
      "learning_rate": 4.941995885473688e-05,
      "loss": 1.0780052185058593,
      "step": 1970
    },
    {
      "epoch": 0.20925532055431523,
      "grad_norm": 1.1821141242980957,
      "learning_rate": 4.94140148142743e-05,
      "loss": 1.2846132278442384,
      "step": 1980
    },
    {
      "epoch": 0.21031216560761987,
      "grad_norm": 1.0167672634124756,
      "learning_rate": 4.9408040833876914e-05,
      "loss": 1.0536267280578613,
      "step": 1990
    },
    {
      "epoch": 0.21136901066092448,
      "grad_norm": 0.985373854637146,
      "learning_rate": 4.940203692087089e-05,
      "loss": 1.2360157012939452,
      "step": 2000
    },
    {
      "epoch": 0.21136901066092448,
      "eval_loss": 0.5834566950798035,
      "eval_runtime": 3060.8672,
      "eval_samples_per_second": 2.909,
      "eval_steps_per_second": 2.909,
      "step": 2000
    },
    {
      "epoch": 0.2124258557142291,
      "grad_norm": 0.8605251312255859,
      "learning_rate": 4.9396003082619066e-05,
      "loss": 1.0755422592163086,
      "step": 2010
    },
    {
      "epoch": 0.21348270076753373,
      "grad_norm": 1.2644226551055908,
      "learning_rate": 4.938993932652098e-05,
      "loss": 1.046428394317627,
      "step": 2020
    },
    {
      "epoch": 0.21453954582083834,
      "grad_norm": 0.9121807813644409,
      "learning_rate": 4.938384566001289e-05,
      "loss": 1.2153115272521973,
      "step": 2030
    },
    {
      "epoch": 0.21559639087414295,
      "grad_norm": 0.9428514242172241,
      "learning_rate": 4.937772209056772e-05,
      "loss": 1.0430753707885743,
      "step": 2040
    },
    {
      "epoch": 0.2166532359274476,
      "grad_norm": 1.0280274152755737,
      "learning_rate": 4.9371568625695045e-05,
      "loss": 1.0850043296813965,
      "step": 2050
    },
    {
      "epoch": 0.2177100809807522,
      "grad_norm": 0.8647369742393494,
      "learning_rate": 4.936538527294111e-05,
      "loss": 1.2649728775024414,
      "step": 2060
    },
    {
      "epoch": 0.21876692603405684,
      "grad_norm": 0.8536009192466736,
      "learning_rate": 4.935917203988885e-05,
      "loss": 1.2419397354125976,
      "step": 2070
    },
    {
      "epoch": 0.21982377108736145,
      "grad_norm": 1.1126015186309814,
      "learning_rate": 4.93529289341578e-05,
      "loss": 1.3811374664306642,
      "step": 2080
    },
    {
      "epoch": 0.22088061614066606,
      "grad_norm": 1.251502275466919,
      "learning_rate": 4.934665596340413e-05,
      "loss": 1.2229784965515136,
      "step": 2090
    },
    {
      "epoch": 0.2219374611939707,
      "grad_norm": 0.7914467453956604,
      "learning_rate": 4.934035313532067e-05,
      "loss": 1.1137020111083984,
      "step": 2100
    },
    {
      "epoch": 0.22299430624727531,
      "grad_norm": 0.8534243702888489,
      "learning_rate": 4.933402045763684e-05,
      "loss": 1.166792869567871,
      "step": 2110
    },
    {
      "epoch": 0.22405115130057995,
      "grad_norm": 0.936590850353241,
      "learning_rate": 4.932765793811866e-05,
      "loss": 1.3075145721435546,
      "step": 2120
    },
    {
      "epoch": 0.22510799635388457,
      "grad_norm": 0.9266261458396912,
      "learning_rate": 4.932126558456875e-05,
      "loss": 1.07562198638916,
      "step": 2130
    },
    {
      "epoch": 0.22616484140718918,
      "grad_norm": 1.1018176078796387,
      "learning_rate": 4.931484340482634e-05,
      "loss": 1.1802120208740234,
      "step": 2140
    },
    {
      "epoch": 0.22722168646049382,
      "grad_norm": 0.7998193502426147,
      "learning_rate": 4.9308391406767216e-05,
      "loss": 1.259147548675537,
      "step": 2150
    },
    {
      "epoch": 0.22827853151379843,
      "grad_norm": 0.8526458740234375,
      "learning_rate": 4.930190959830373e-05,
      "loss": 1.2277114868164063,
      "step": 2160
    },
    {
      "epoch": 0.22933537656710307,
      "grad_norm": 0.7910148501396179,
      "learning_rate": 4.9295397987384796e-05,
      "loss": 1.065322208404541,
      "step": 2170
    },
    {
      "epoch": 0.23039222162040768,
      "grad_norm": 1.1438912153244019,
      "learning_rate": 4.9288856581995865e-05,
      "loss": 1.3605783462524415,
      "step": 2180
    },
    {
      "epoch": 0.2314490666737123,
      "grad_norm": 1.5051559209823608,
      "learning_rate": 4.928228539015897e-05,
      "loss": 1.155210590362549,
      "step": 2190
    },
    {
      "epoch": 0.23250591172701693,
      "grad_norm": 1.0379410982131958,
      "learning_rate": 4.927568441993261e-05,
      "loss": 1.115980339050293,
      "step": 2200
    },
    {
      "epoch": 0.23356275678032154,
      "grad_norm": 1.0642622709274292,
      "learning_rate": 4.926905367941185e-05,
      "loss": 1.374610710144043,
      "step": 2210
    },
    {
      "epoch": 0.23461960183362618,
      "grad_norm": 0.6736540198326111,
      "learning_rate": 4.9262393176728246e-05,
      "loss": 1.2229202270507813,
      "step": 2220
    },
    {
      "epoch": 0.2356764468869308,
      "grad_norm": 0.9766952991485596,
      "learning_rate": 4.925570292004984e-05,
      "loss": 1.1859923362731934,
      "step": 2230
    },
    {
      "epoch": 0.2367332919402354,
      "grad_norm": 0.8515557050704956,
      "learning_rate": 4.924898291758119e-05,
      "loss": 1.3080815315246581,
      "step": 2240
    },
    {
      "epoch": 0.23779013699354004,
      "grad_norm": 0.7691314220428467,
      "learning_rate": 4.9242233177563305e-05,
      "loss": 1.1754939079284668,
      "step": 2250
    },
    {
      "epoch": 0.23884698204684465,
      "grad_norm": 0.8896235823631287,
      "learning_rate": 4.923545370827368e-05,
      "loss": 1.268226718902588,
      "step": 2260
    },
    {
      "epoch": 0.2399038271001493,
      "grad_norm": 0.9270082116127014,
      "learning_rate": 4.9228644518026274e-05,
      "loss": 1.1410390853881835,
      "step": 2270
    },
    {
      "epoch": 0.2409606721534539,
      "grad_norm": 0.9441484808921814,
      "learning_rate": 4.9221805615171465e-05,
      "loss": 1.1842496871948243,
      "step": 2280
    },
    {
      "epoch": 0.24201751720675851,
      "grad_norm": 0.9829817414283752,
      "learning_rate": 4.9214937008096114e-05,
      "loss": 1.25244083404541,
      "step": 2290
    },
    {
      "epoch": 0.24307436226006315,
      "grad_norm": 1.05659019947052,
      "learning_rate": 4.920803870522346e-05,
      "loss": 1.1846921920776368,
      "step": 2300
    },
    {
      "epoch": 0.24413120731336777,
      "grad_norm": 1.055021047592163,
      "learning_rate": 4.9201110715013186e-05,
      "loss": 1.2360570907592774,
      "step": 2310
    },
    {
      "epoch": 0.2451880523666724,
      "grad_norm": 0.8008033633232117,
      "learning_rate": 4.9194153045961385e-05,
      "loss": 1.168649959564209,
      "step": 2320
    },
    {
      "epoch": 0.24624489741997702,
      "grad_norm": 0.7553218603134155,
      "learning_rate": 4.9187165706600536e-05,
      "loss": 1.259486198425293,
      "step": 2330
    },
    {
      "epoch": 0.24730174247328163,
      "grad_norm": 0.8217469453811646,
      "learning_rate": 4.918014870549951e-05,
      "loss": 1.2099687576293945,
      "step": 2340
    },
    {
      "epoch": 0.24835858752658627,
      "grad_norm": 0.9581583142280579,
      "learning_rate": 4.9173102051263556e-05,
      "loss": 1.0941384315490723,
      "step": 2350
    },
    {
      "epoch": 0.24941543257989088,
      "grad_norm": 0.9553148150444031,
      "learning_rate": 4.916602575253427e-05,
      "loss": 1.266836452484131,
      "step": 2360
    },
    {
      "epoch": 0.2504722776331955,
      "grad_norm": 0.794839084148407,
      "learning_rate": 4.915891981798962e-05,
      "loss": 1.2640167236328126,
      "step": 2370
    },
    {
      "epoch": 0.25152912268650013,
      "grad_norm": 0.9496021866798401,
      "learning_rate": 4.915178425634391e-05,
      "loss": 1.1956175804138183,
      "step": 2380
    },
    {
      "epoch": 0.25258596773980474,
      "grad_norm": 1.1745727062225342,
      "learning_rate": 4.91446190763478e-05,
      "loss": 1.2333488464355469,
      "step": 2390
    },
    {
      "epoch": 0.25364281279310935,
      "grad_norm": 1.2743357419967651,
      "learning_rate": 4.9137424286788226e-05,
      "loss": 1.1847744941711427,
      "step": 2400
    },
    {
      "epoch": 0.254699657846414,
      "grad_norm": 1.4879859685897827,
      "learning_rate": 4.9130199896488474e-05,
      "loss": 1.2007080078125,
      "step": 2410
    },
    {
      "epoch": 0.25575650289971863,
      "grad_norm": 0.8005043864250183,
      "learning_rate": 4.912294591430811e-05,
      "loss": 1.149638557434082,
      "step": 2420
    },
    {
      "epoch": 0.25681334795302324,
      "grad_norm": 1.224259853363037,
      "learning_rate": 4.911566234914301e-05,
      "loss": 1.1342608451843261,
      "step": 2430
    },
    {
      "epoch": 0.25787019300632785,
      "grad_norm": 0.752318799495697,
      "learning_rate": 4.910834920992531e-05,
      "loss": 1.214425754547119,
      "step": 2440
    },
    {
      "epoch": 0.25892703805963246,
      "grad_norm": 1.2372101545333862,
      "learning_rate": 4.910100650562343e-05,
      "loss": 1.1838050842285157,
      "step": 2450
    },
    {
      "epoch": 0.25998388311293713,
      "grad_norm": 0.9092959761619568,
      "learning_rate": 4.909363424524201e-05,
      "loss": 1.157148838043213,
      "step": 2460
    },
    {
      "epoch": 0.26104072816624174,
      "grad_norm": 0.8812724351882935,
      "learning_rate": 4.908623243782199e-05,
      "loss": 1.178938102722168,
      "step": 2470
    },
    {
      "epoch": 0.26209757321954635,
      "grad_norm": 1.1365686655044556,
      "learning_rate": 4.907880109244052e-05,
      "loss": 1.0511191368103028,
      "step": 2480
    },
    {
      "epoch": 0.26315441827285097,
      "grad_norm": 0.8594139218330383,
      "learning_rate": 4.907134021821096e-05,
      "loss": 1.0847901344299316,
      "step": 2490
    },
    {
      "epoch": 0.2642112633261556,
      "grad_norm": 0.9383670687675476,
      "learning_rate": 4.906384982428289e-05,
      "loss": 1.2671892166137695,
      "step": 2500
    },
    {
      "epoch": 0.2642112633261556,
      "eval_loss": 0.5752180218696594,
      "eval_runtime": 3063.8851,
      "eval_samples_per_second": 2.906,
      "eval_steps_per_second": 2.906,
      "step": 2500
    },
    {
      "epoch": 0.26526810837946024,
      "grad_norm": 0.9530830979347229,
      "learning_rate": 4.90563299198421e-05,
      "loss": 1.18833589553833,
      "step": 2510
    },
    {
      "epoch": 0.26632495343276485,
      "grad_norm": 0.8559793829917908,
      "learning_rate": 4.904878051411058e-05,
      "loss": 1.1216675758361816,
      "step": 2520
    },
    {
      "epoch": 0.26738179848606947,
      "grad_norm": 0.778265118598938,
      "learning_rate": 4.904120161634647e-05,
      "loss": 1.19404296875,
      "step": 2530
    },
    {
      "epoch": 0.2684386435393741,
      "grad_norm": 0.9033781290054321,
      "learning_rate": 4.903359323584409e-05,
      "loss": 1.0375430107116699,
      "step": 2540
    },
    {
      "epoch": 0.2694954885926787,
      "grad_norm": 1.059201955795288,
      "learning_rate": 4.902595538193392e-05,
      "loss": 1.19734525680542,
      "step": 2550
    },
    {
      "epoch": 0.2705523336459833,
      "grad_norm": 1.0376242399215698,
      "learning_rate": 4.9018288063982584e-05,
      "loss": 1.1965457916259765,
      "step": 2560
    },
    {
      "epoch": 0.27160917869928797,
      "grad_norm": 0.9238499999046326,
      "learning_rate": 4.9010591291392836e-05,
      "loss": 1.1794482231140138,
      "step": 2570
    },
    {
      "epoch": 0.2726660237525926,
      "grad_norm": 0.7612287998199463,
      "learning_rate": 4.900286507360356e-05,
      "loss": 1.224607276916504,
      "step": 2580
    },
    {
      "epoch": 0.2737228688058972,
      "grad_norm": 0.7783312797546387,
      "learning_rate": 4.899510942008973e-05,
      "loss": 0.9918021202087403,
      "step": 2590
    },
    {
      "epoch": 0.2747797138592018,
      "grad_norm": 1.0226808786392212,
      "learning_rate": 4.898732434036244e-05,
      "loss": 1.1529096603393554,
      "step": 2600
    },
    {
      "epoch": 0.2758365589125064,
      "grad_norm": 0.9804856777191162,
      "learning_rate": 4.897950984396885e-05,
      "loss": 1.1476472854614257,
      "step": 2610
    },
    {
      "epoch": 0.2768934039658111,
      "grad_norm": 1.2189862728118896,
      "learning_rate": 4.897166594049222e-05,
      "loss": 1.1955551147460937,
      "step": 2620
    },
    {
      "epoch": 0.2779502490191157,
      "grad_norm": 1.654741883277893,
      "learning_rate": 4.8963792639551846e-05,
      "loss": 1.1972854614257813,
      "step": 2630
    },
    {
      "epoch": 0.2790070940724203,
      "grad_norm": 1.2485692501068115,
      "learning_rate": 4.89558899508031e-05,
      "loss": 1.280005645751953,
      "step": 2640
    },
    {
      "epoch": 0.2800639391257249,
      "grad_norm": 0.8232464790344238,
      "learning_rate": 4.894795788393738e-05,
      "loss": 1.156368350982666,
      "step": 2650
    },
    {
      "epoch": 0.2811207841790295,
      "grad_norm": 1.5244137048721313,
      "learning_rate": 4.89399964486821e-05,
      "loss": 1.259807300567627,
      "step": 2660
    },
    {
      "epoch": 0.2821776292323342,
      "grad_norm": 0.8922842144966125,
      "learning_rate": 4.893200565480073e-05,
      "loss": 1.2252914428710937,
      "step": 2670
    },
    {
      "epoch": 0.2832344742856388,
      "grad_norm": 1.0507988929748535,
      "learning_rate": 4.892398551209269e-05,
      "loss": 1.1477043151855468,
      "step": 2680
    },
    {
      "epoch": 0.2842913193389434,
      "grad_norm": 0.9536561965942383,
      "learning_rate": 4.891593603039343e-05,
      "loss": 1.1225727081298829,
      "step": 2690
    },
    {
      "epoch": 0.285348164392248,
      "grad_norm": 0.9035396575927734,
      "learning_rate": 4.8907857219574374e-05,
      "loss": 1.2055601119995116,
      "step": 2700
    },
    {
      "epoch": 0.28640500944555264,
      "grad_norm": 0.7615309357643127,
      "learning_rate": 4.88997490895429e-05,
      "loss": 1.2678559303283692,
      "step": 2710
    },
    {
      "epoch": 0.2874618544988573,
      "grad_norm": 1.1582328081130981,
      "learning_rate": 4.889161165024235e-05,
      "loss": 1.183156681060791,
      "step": 2720
    },
    {
      "epoch": 0.2885186995521619,
      "grad_norm": 1.561128854751587,
      "learning_rate": 4.8883444911652014e-05,
      "loss": 1.0841514587402343,
      "step": 2730
    },
    {
      "epoch": 0.28957554460546653,
      "grad_norm": 0.8774929642677307,
      "learning_rate": 4.88752488837871e-05,
      "loss": 1.1005674362182618,
      "step": 2740
    },
    {
      "epoch": 0.29063238965877114,
      "grad_norm": 0.8889621496200562,
      "learning_rate": 4.886702357669876e-05,
      "loss": 1.2401761054992675,
      "step": 2750
    },
    {
      "epoch": 0.29168923471207575,
      "grad_norm": 0.8391106724739075,
      "learning_rate": 4.8858769000474016e-05,
      "loss": 1.0482848167419434,
      "step": 2760
    },
    {
      "epoch": 0.2927460797653804,
      "grad_norm": 0.9510761499404907,
      "learning_rate": 4.88504851652358e-05,
      "loss": 1.138117790222168,
      "step": 2770
    },
    {
      "epoch": 0.29380292481868503,
      "grad_norm": 0.7536053657531738,
      "learning_rate": 4.884217208114294e-05,
      "loss": 1.2072933197021485,
      "step": 2780
    },
    {
      "epoch": 0.29485976987198964,
      "grad_norm": 0.740530252456665,
      "learning_rate": 4.8833829758390115e-05,
      "loss": 1.2056413650512696,
      "step": 2790
    },
    {
      "epoch": 0.29591661492529425,
      "grad_norm": 0.7609956860542297,
      "learning_rate": 4.8825458207207875e-05,
      "loss": 1.0717817306518556,
      "step": 2800
    },
    {
      "epoch": 0.29697345997859886,
      "grad_norm": 1.1526219844818115,
      "learning_rate": 4.881705743786259e-05,
      "loss": 1.265073013305664,
      "step": 2810
    },
    {
      "epoch": 0.29803030503190353,
      "grad_norm": 1.2439080476760864,
      "learning_rate": 4.8808627460656496e-05,
      "loss": 1.2710920333862306,
      "step": 2820
    },
    {
      "epoch": 0.29908715008520814,
      "grad_norm": 0.9535239934921265,
      "learning_rate": 4.880016828592762e-05,
      "loss": 1.0768875122070312,
      "step": 2830
    },
    {
      "epoch": 0.30014399513851275,
      "grad_norm": 0.9726995825767517,
      "learning_rate": 4.8791679924049796e-05,
      "loss": 1.1090085983276368,
      "step": 2840
    },
    {
      "epoch": 0.30120084019181737,
      "grad_norm": 0.8544860482215881,
      "learning_rate": 4.878316238543268e-05,
      "loss": 1.207863998413086,
      "step": 2850
    },
    {
      "epoch": 0.302257685245122,
      "grad_norm": 1.485451340675354,
      "learning_rate": 4.8774615680521665e-05,
      "loss": 1.172201919555664,
      "step": 2860
    },
    {
      "epoch": 0.30331453029842664,
      "grad_norm": 1.2799742221832275,
      "learning_rate": 4.876603981979795e-05,
      "loss": 1.1095293045043946,
      "step": 2870
    },
    {
      "epoch": 0.30437137535173125,
      "grad_norm": 0.9787461757659912,
      "learning_rate": 4.875743481377848e-05,
      "loss": 1.2827129364013672,
      "step": 2880
    },
    {
      "epoch": 0.30542822040503587,
      "grad_norm": 1.3931654691696167,
      "learning_rate": 4.874880067301592e-05,
      "loss": 1.242313766479492,
      "step": 2890
    },
    {
      "epoch": 0.3064850654583405,
      "grad_norm": 0.9191779494285583,
      "learning_rate": 4.874013740809869e-05,
      "loss": 1.1866607666015625,
      "step": 2900
    },
    {
      "epoch": 0.3075419105116451,
      "grad_norm": 0.69975346326828,
      "learning_rate": 4.873144502965091e-05,
      "loss": 1.0572191238403321,
      "step": 2910
    },
    {
      "epoch": 0.30859875556494976,
      "grad_norm": 1.066375970840454,
      "learning_rate": 4.872272354833242e-05,
      "loss": 1.2174957275390625,
      "step": 2920
    },
    {
      "epoch": 0.30965560061825437,
      "grad_norm": 0.6801135540008545,
      "learning_rate": 4.871397297483874e-05,
      "loss": 1.0045628547668457,
      "step": 2930
    },
    {
      "epoch": 0.310712445671559,
      "grad_norm": 1.2734187841415405,
      "learning_rate": 4.870519331990106e-05,
      "loss": 1.2693100929260255,
      "step": 2940
    },
    {
      "epoch": 0.3117692907248636,
      "grad_norm": 1.298256754875183,
      "learning_rate": 4.8696384594286223e-05,
      "loss": 1.0992612838745117,
      "step": 2950
    },
    {
      "epoch": 0.3128261357781682,
      "grad_norm": 1.0307362079620361,
      "learning_rate": 4.868754680879678e-05,
      "loss": 1.1889406204223634,
      "step": 2960
    },
    {
      "epoch": 0.31388298083147287,
      "grad_norm": 0.8127456903457642,
      "learning_rate": 4.867867997427085e-05,
      "loss": 1.1352449417114259,
      "step": 2970
    },
    {
      "epoch": 0.3149398258847775,
      "grad_norm": 0.9852636456489563,
      "learning_rate": 4.866978410158222e-05,
      "loss": 1.1954586029052734,
      "step": 2980
    },
    {
      "epoch": 0.3159966709380821,
      "grad_norm": 0.9863714575767517,
      "learning_rate": 4.866085920164026e-05,
      "loss": 1.1643128395080566,
      "step": 2990
    },
    {
      "epoch": 0.3170535159913867,
      "grad_norm": 0.9643257260322571,
      "learning_rate": 4.865190528538997e-05,
      "loss": 1.1665685653686524,
      "step": 3000
    },
    {
      "epoch": 0.3170535159913867,
      "eval_loss": 0.5579926371574402,
      "eval_runtime": 3063.1475,
      "eval_samples_per_second": 2.907,
      "eval_steps_per_second": 2.907,
      "step": 3000
    },
    {
      "epoch": 0.3181103610446913,
      "grad_norm": 0.8266704678535461,
      "learning_rate": 4.8642922363811894e-05,
      "loss": 1.1443010330200196,
      "step": 3010
    },
    {
      "epoch": 0.319167206097996,
      "grad_norm": 0.9833044409751892,
      "learning_rate": 4.863391044792218e-05,
      "loss": 1.275413703918457,
      "step": 3020
    },
    {
      "epoch": 0.3202240511513006,
      "grad_norm": 1.0374956130981445,
      "learning_rate": 4.8624869548772526e-05,
      "loss": 1.214223861694336,
      "step": 3030
    },
    {
      "epoch": 0.3212808962046052,
      "grad_norm": 0.933085024356842,
      "learning_rate": 4.861579967745015e-05,
      "loss": 1.170733642578125,
      "step": 3040
    },
    {
      "epoch": 0.3223377412579098,
      "grad_norm": 0.9936915040016174,
      "learning_rate": 4.860670084507782e-05,
      "loss": 1.2559396743774414,
      "step": 3050
    },
    {
      "epoch": 0.3233945863112144,
      "grad_norm": 0.8090853095054626,
      "learning_rate": 4.8597573062813825e-05,
      "loss": 1.2647581100463867,
      "step": 3060
    },
    {
      "epoch": 0.3244514313645191,
      "grad_norm": 0.8301842212677002,
      "learning_rate": 4.8588416341851955e-05,
      "loss": 1.1916956901550293,
      "step": 3070
    },
    {
      "epoch": 0.3255082764178237,
      "grad_norm": 1.4450124502182007,
      "learning_rate": 4.8579230693421475e-05,
      "loss": 1.2007198333740234,
      "step": 3080
    },
    {
      "epoch": 0.3265651214711283,
      "grad_norm": 0.8468474745750427,
      "learning_rate": 4.857001612878712e-05,
      "loss": 1.012873935699463,
      "step": 3090
    },
    {
      "epoch": 0.32762196652443293,
      "grad_norm": 0.7920951843261719,
      "learning_rate": 4.856077265924912e-05,
      "loss": 1.162527847290039,
      "step": 3100
    },
    {
      "epoch": 0.32867881157773754,
      "grad_norm": 0.9596951603889465,
      "learning_rate": 4.855150029614312e-05,
      "loss": 1.1453103065490722,
      "step": 3110
    },
    {
      "epoch": 0.3297356566310422,
      "grad_norm": 0.8677332997322083,
      "learning_rate": 4.8542199050840206e-05,
      "loss": 1.1582250595092773,
      "step": 3120
    },
    {
      "epoch": 0.3307925016843468,
      "grad_norm": 1.2981420755386353,
      "learning_rate": 4.8532868934746905e-05,
      "loss": 1.198538589477539,
      "step": 3130
    },
    {
      "epoch": 0.33184934673765143,
      "grad_norm": 1.019105076789856,
      "learning_rate": 4.852350995930511e-05,
      "loss": 1.082587718963623,
      "step": 3140
    },
    {
      "epoch": 0.33290619179095604,
      "grad_norm": 0.8941572904586792,
      "learning_rate": 4.851412213599213e-05,
      "loss": 1.2751832008361816,
      "step": 3150
    },
    {
      "epoch": 0.33396303684426065,
      "grad_norm": 1.4174563884735107,
      "learning_rate": 4.850470547632067e-05,
      "loss": 1.1817429542541504,
      "step": 3160
    },
    {
      "epoch": 0.3350198818975653,
      "grad_norm": 1.0290331840515137,
      "learning_rate": 4.849525999183875e-05,
      "loss": 1.1542068481445313,
      "step": 3170
    },
    {
      "epoch": 0.33607672695086993,
      "grad_norm": 0.8813462853431702,
      "learning_rate": 4.848578569412977e-05,
      "loss": 1.2674723625183106,
      "step": 3180
    },
    {
      "epoch": 0.33713357200417454,
      "grad_norm": 1.3243863582611084,
      "learning_rate": 4.847628259481248e-05,
      "loss": 1.1134199142456054,
      "step": 3190
    },
    {
      "epoch": 0.33819041705747915,
      "grad_norm": 0.7880538702011108,
      "learning_rate": 4.846675070554091e-05,
      "loss": 1.1357905387878418,
      "step": 3200
    },
    {
      "epoch": 0.33924726211078376,
      "grad_norm": 0.9974196553230286,
      "learning_rate": 4.8457190038004424e-05,
      "loss": 1.0668041229248046,
      "step": 3210
    },
    {
      "epoch": 0.34030410716408843,
      "grad_norm": 0.9568934440612793,
      "learning_rate": 4.8447600603927675e-05,
      "loss": 1.2080458641052245,
      "step": 3220
    },
    {
      "epoch": 0.34136095221739304,
      "grad_norm": 0.9942179322242737,
      "learning_rate": 4.843798241507058e-05,
      "loss": 1.1382678031921387,
      "step": 3230
    },
    {
      "epoch": 0.34241779727069765,
      "grad_norm": 0.8791975378990173,
      "learning_rate": 4.842833548322834e-05,
      "loss": 1.0070273399353027,
      "step": 3240
    },
    {
      "epoch": 0.34347464232400227,
      "grad_norm": 1.0540672540664673,
      "learning_rate": 4.8418659820231394e-05,
      "loss": 1.1470075607299806,
      "step": 3250
    },
    {
      "epoch": 0.3445314873773069,
      "grad_norm": 0.7428416609764099,
      "learning_rate": 4.84089554379454e-05,
      "loss": 1.2172751426696777,
      "step": 3260
    },
    {
      "epoch": 0.34558833243061154,
      "grad_norm": 0.8998673558235168,
      "learning_rate": 4.839922234827127e-05,
      "loss": 1.1432336807250976,
      "step": 3270
    },
    {
      "epoch": 0.34664517748391616,
      "grad_norm": 1.2393102645874023,
      "learning_rate": 4.83894605631451e-05,
      "loss": 1.1659416198730468,
      "step": 3280
    },
    {
      "epoch": 0.34770202253722077,
      "grad_norm": 1.2480427026748657,
      "learning_rate": 4.837967009453815e-05,
      "loss": 1.109553909301758,
      "step": 3290
    },
    {
      "epoch": 0.3487588675905254,
      "grad_norm": 0.9325940608978271,
      "learning_rate": 4.836985095445693e-05,
      "loss": 1.0595672607421875,
      "step": 3300
    },
    {
      "epoch": 0.34981571264383,
      "grad_norm": 0.8387763500213623,
      "learning_rate": 4.836000315494303e-05,
      "loss": 1.153205108642578,
      "step": 3310
    },
    {
      "epoch": 0.3508725576971346,
      "grad_norm": 1.5751017332077026,
      "learning_rate": 4.835012670807323e-05,
      "loss": 1.2370134353637696,
      "step": 3320
    },
    {
      "epoch": 0.35192940275043927,
      "grad_norm": 1.2510019540786743,
      "learning_rate": 4.8340221625959434e-05,
      "loss": 1.0579489707946776,
      "step": 3330
    },
    {
      "epoch": 0.3529862478037439,
      "grad_norm": 1.0222229957580566,
      "learning_rate": 4.833028792074867e-05,
      "loss": 1.1644598960876464,
      "step": 3340
    },
    {
      "epoch": 0.3540430928570485,
      "grad_norm": 0.7654703855514526,
      "learning_rate": 4.832032560462305e-05,
      "loss": 1.1702071189880372,
      "step": 3350
    },
    {
      "epoch": 0.3550999379103531,
      "grad_norm": 1.123583197593689,
      "learning_rate": 4.8310334689799783e-05,
      "loss": 1.271115779876709,
      "step": 3360
    },
    {
      "epoch": 0.3561567829636577,
      "grad_norm": 0.9085693359375,
      "learning_rate": 4.8300315188531153e-05,
      "loss": 1.2847204208374023,
      "step": 3370
    },
    {
      "epoch": 0.3572136280169624,
      "grad_norm": 0.8476790189743042,
      "learning_rate": 4.829026711310449e-05,
      "loss": 1.1056520462036132,
      "step": 3380
    },
    {
      "epoch": 0.358270473070267,
      "grad_norm": 0.8685449957847595,
      "learning_rate": 4.828019047584217e-05,
      "loss": 1.1500364303588868,
      "step": 3390
    },
    {
      "epoch": 0.3593273181235716,
      "grad_norm": 0.8599591255187988,
      "learning_rate": 4.8270085289101604e-05,
      "loss": 1.1476428985595704,
      "step": 3400
    },
    {
      "epoch": 0.3603841631768762,
      "grad_norm": 0.8890820145606995,
      "learning_rate": 4.825995156527521e-05,
      "loss": 1.2072113037109375,
      "step": 3410
    },
    {
      "epoch": 0.3614410082301808,
      "grad_norm": 1.0020203590393066,
      "learning_rate": 4.82497893167904e-05,
      "loss": 1.158389949798584,
      "step": 3420
    },
    {
      "epoch": 0.3624978532834855,
      "grad_norm": 0.8341196179389954,
      "learning_rate": 4.823959855610957e-05,
      "loss": 1.1784236907958985,
      "step": 3430
    },
    {
      "epoch": 0.3635546983367901,
      "grad_norm": 0.8681564927101135,
      "learning_rate": 4.822937929573006e-05,
      "loss": 1.0764629364013671,
      "step": 3440
    },
    {
      "epoch": 0.3646115433900947,
      "grad_norm": 1.547301173210144,
      "learning_rate": 4.82191315481842e-05,
      "loss": 1.2008303642272948,
      "step": 3450
    },
    {
      "epoch": 0.36566838844339933,
      "grad_norm": 0.861606776714325,
      "learning_rate": 4.8208855326039226e-05,
      "loss": 1.2699466705322267,
      "step": 3460
    },
    {
      "epoch": 0.36672523349670394,
      "grad_norm": 0.9358724355697632,
      "learning_rate": 4.819855064189731e-05,
      "loss": 1.15127534866333,
      "step": 3470
    },
    {
      "epoch": 0.3677820785500086,
      "grad_norm": 1.097021222114563,
      "learning_rate": 4.8188217508395504e-05,
      "loss": 1.1433833122253418,
      "step": 3480
    },
    {
      "epoch": 0.3688389236033132,
      "grad_norm": 1.1291453838348389,
      "learning_rate": 4.817785593820579e-05,
      "loss": 1.0388105392456055,
      "step": 3490
    },
    {
      "epoch": 0.36989576865661783,
      "grad_norm": 0.8982201814651489,
      "learning_rate": 4.816746594403497e-05,
      "loss": 1.1543403625488282,
      "step": 3500
    },
    {
      "epoch": 0.36989576865661783,
      "eval_loss": 0.5537382960319519,
      "eval_runtime": 3059.984,
      "eval_samples_per_second": 2.91,
      "eval_steps_per_second": 2.91,
      "step": 3500
    },
    {
      "epoch": 0.37095261370992244,
      "grad_norm": 0.8768877983093262,
      "learning_rate": 4.815704753862474e-05,
      "loss": 1.215390968322754,
      "step": 3510
    },
    {
      "epoch": 0.37200945876322705,
      "grad_norm": 1.0001225471496582,
      "learning_rate": 4.814660073475165e-05,
      "loss": 1.1505086898803711,
      "step": 3520
    },
    {
      "epoch": 0.3730663038165317,
      "grad_norm": 0.8758547902107239,
      "learning_rate": 4.8136125545227034e-05,
      "loss": 1.267770767211914,
      "step": 3530
    },
    {
      "epoch": 0.37412314886983633,
      "grad_norm": 0.8792843818664551,
      "learning_rate": 4.812562198289706e-05,
      "loss": 1.2173368453979492,
      "step": 3540
    },
    {
      "epoch": 0.37517999392314094,
      "grad_norm": 0.9152326583862305,
      "learning_rate": 4.8115090060642696e-05,
      "loss": 1.0126462936401368,
      "step": 3550
    },
    {
      "epoch": 0.37623683897644555,
      "grad_norm": 0.9989923238754272,
      "learning_rate": 4.8104529791379684e-05,
      "loss": 1.1297735214233398,
      "step": 3560
    },
    {
      "epoch": 0.37729368402975016,
      "grad_norm": 0.9558157920837402,
      "learning_rate": 4.809394118805851e-05,
      "loss": 1.247045135498047,
      "step": 3570
    },
    {
      "epoch": 0.37835052908305483,
      "grad_norm": 1.4497350454330444,
      "learning_rate": 4.808332426366444e-05,
      "loss": 1.176620864868164,
      "step": 3580
    },
    {
      "epoch": 0.37940737413635944,
      "grad_norm": 1.0726107358932495,
      "learning_rate": 4.807267903121746e-05,
      "loss": 1.2145004272460938,
      "step": 3590
    },
    {
      "epoch": 0.38046421918966405,
      "grad_norm": 0.8552097678184509,
      "learning_rate": 4.806200550377226e-05,
      "loss": 1.1176979064941406,
      "step": 3600
    },
    {
      "epoch": 0.38152106424296867,
      "grad_norm": 0.7914222478866577,
      "learning_rate": 4.8051303694418224e-05,
      "loss": 1.1618346214294433,
      "step": 3610
    },
    {
      "epoch": 0.3825779092962733,
      "grad_norm": 1.2705551385879517,
      "learning_rate": 4.804057361627944e-05,
      "loss": 1.2143467903137206,
      "step": 3620
    },
    {
      "epoch": 0.38363475434957794,
      "grad_norm": 1.3545305728912354,
      "learning_rate": 4.8029815282514666e-05,
      "loss": 1.2387968063354493,
      "step": 3630
    },
    {
      "epoch": 0.38469159940288256,
      "grad_norm": 1.3950116634368896,
      "learning_rate": 4.8019028706317294e-05,
      "loss": 1.1618488311767579,
      "step": 3640
    },
    {
      "epoch": 0.38574844445618717,
      "grad_norm": 1.0610543489456177,
      "learning_rate": 4.800821390091534e-05,
      "loss": 1.135669994354248,
      "step": 3650
    },
    {
      "epoch": 0.3868052895094918,
      "grad_norm": 0.8629146218299866,
      "learning_rate": 4.799737087957149e-05,
      "loss": 1.0841713905334474,
      "step": 3660
    },
    {
      "epoch": 0.3878621345627964,
      "grad_norm": 1.1059176921844482,
      "learning_rate": 4.798649965558296e-05,
      "loss": 1.2752059936523437,
      "step": 3670
    },
    {
      "epoch": 0.38891897961610106,
      "grad_norm": 0.8461462259292603,
      "learning_rate": 4.797560024228161e-05,
      "loss": 1.2351527214050293,
      "step": 3680
    },
    {
      "epoch": 0.38997582466940567,
      "grad_norm": 1.2970623970031738,
      "learning_rate": 4.796467265303386e-05,
      "loss": 1.0935221672058106,
      "step": 3690
    },
    {
      "epoch": 0.3910326697227103,
      "grad_norm": 0.31666240096092224,
      "learning_rate": 4.7953716901240655e-05,
      "loss": 1.076282024383545,
      "step": 3700
    },
    {
      "epoch": 0.3920895147760149,
      "grad_norm": 1.049989938735962,
      "learning_rate": 4.794273300033749e-05,
      "loss": 1.1949956893920899,
      "step": 3710
    },
    {
      "epoch": 0.3931463598293195,
      "grad_norm": 1.0125386714935303,
      "learning_rate": 4.79317209637944e-05,
      "loss": 1.2482955932617188,
      "step": 3720
    },
    {
      "epoch": 0.39420320488262417,
      "grad_norm": 1.291111707687378,
      "learning_rate": 4.792068080511591e-05,
      "loss": 1.167121982574463,
      "step": 3730
    },
    {
      "epoch": 0.3952600499359288,
      "grad_norm": 1.067001223564148,
      "learning_rate": 4.790961253784102e-05,
      "loss": 1.3001273155212403,
      "step": 3740
    },
    {
      "epoch": 0.3963168949892334,
      "grad_norm": 1.057289719581604,
      "learning_rate": 4.789851617554322e-05,
      "loss": 1.2997982978820801,
      "step": 3750
    },
    {
      "epoch": 0.397373740042538,
      "grad_norm": 1.3210316896438599,
      "learning_rate": 4.788739173183044e-05,
      "loss": 1.2082348823547364,
      "step": 3760
    },
    {
      "epoch": 0.3984305850958426,
      "grad_norm": 1.204756736755371,
      "learning_rate": 4.787623922034508e-05,
      "loss": 1.1073071479797363,
      "step": 3770
    },
    {
      "epoch": 0.3994874301491473,
      "grad_norm": 1.0779764652252197,
      "learning_rate": 4.78650586547639e-05,
      "loss": 1.1844019889831543,
      "step": 3780
    },
    {
      "epoch": 0.4005442752024519,
      "grad_norm": 1.4453084468841553,
      "learning_rate": 4.78538500487981e-05,
      "loss": 1.1719649314880372,
      "step": 3790
    },
    {
      "epoch": 0.4016011202557565,
      "grad_norm": 1.047186017036438,
      "learning_rate": 4.7842613416193295e-05,
      "loss": 1.0990612983703614,
      "step": 3800
    },
    {
      "epoch": 0.4026579653090611,
      "grad_norm": 0.7978026270866394,
      "learning_rate": 4.783134877072941e-05,
      "loss": 1.1561716079711915,
      "step": 3810
    },
    {
      "epoch": 0.40371481036236573,
      "grad_norm": 0.9400219321250916,
      "learning_rate": 4.7820056126220783e-05,
      "loss": 1.1923961639404297,
      "step": 3820
    },
    {
      "epoch": 0.4047716554156704,
      "grad_norm": 1.41529381275177,
      "learning_rate": 4.7808735496516034e-05,
      "loss": 1.2170982360839844,
      "step": 3830
    },
    {
      "epoch": 0.405828500468975,
      "grad_norm": 0.9590666890144348,
      "learning_rate": 4.779738689549814e-05,
      "loss": 1.083742332458496,
      "step": 3840
    },
    {
      "epoch": 0.4068853455222796,
      "grad_norm": 1.041663646697998,
      "learning_rate": 4.778601033708438e-05,
      "loss": 1.1058547973632813,
      "step": 3850
    },
    {
      "epoch": 0.40794219057558423,
      "grad_norm": 0.9911213517189026,
      "learning_rate": 4.7774605835226286e-05,
      "loss": 1.1739565849304199,
      "step": 3860
    },
    {
      "epoch": 0.40899903562888884,
      "grad_norm": 0.9520569443702698,
      "learning_rate": 4.776317340390968e-05,
      "loss": 1.2985098838806153,
      "step": 3870
    },
    {
      "epoch": 0.4100558806821935,
      "grad_norm": 1.480704426765442,
      "learning_rate": 4.775171305715465e-05,
      "loss": 1.1813619613647461,
      "step": 3880
    },
    {
      "epoch": 0.4111127257354981,
      "grad_norm": 0.8852286338806152,
      "learning_rate": 4.774022480901548e-05,
      "loss": 1.0385025024414063,
      "step": 3890
    },
    {
      "epoch": 0.41216957078880273,
      "grad_norm": 0.9884142875671387,
      "learning_rate": 4.7728708673580716e-05,
      "loss": 1.2054457664489746,
      "step": 3900
    },
    {
      "epoch": 0.41322641584210734,
      "grad_norm": 0.9349421858787537,
      "learning_rate": 4.7717164664973074e-05,
      "loss": 1.176686668395996,
      "step": 3910
    },
    {
      "epoch": 0.41428326089541195,
      "grad_norm": 0.7188425660133362,
      "learning_rate": 4.7705592797349455e-05,
      "loss": 1.1372150421142577,
      "step": 3920
    },
    {
      "epoch": 0.4153401059487166,
      "grad_norm": 1.03571617603302,
      "learning_rate": 4.769399308490091e-05,
      "loss": 1.1410999298095703,
      "step": 3930
    },
    {
      "epoch": 0.41639695100202123,
      "grad_norm": 0.8717170357704163,
      "learning_rate": 4.768236554185269e-05,
      "loss": 1.2024547576904296,
      "step": 3940
    },
    {
      "epoch": 0.41745379605532584,
      "grad_norm": 0.904988706111908,
      "learning_rate": 4.7670710182464115e-05,
      "loss": 1.0487839698791503,
      "step": 3950
    },
    {
      "epoch": 0.41851064110863045,
      "grad_norm": 0.8945913910865784,
      "learning_rate": 4.765902702102864e-05,
      "loss": 1.1499239921569824,
      "step": 3960
    },
    {
      "epoch": 0.41956748616193507,
      "grad_norm": 0.7863814830780029,
      "learning_rate": 4.7647316071873846e-05,
      "loss": 1.0581459999084473,
      "step": 3970
    },
    {
      "epoch": 0.42062433121523973,
      "grad_norm": 1.0099494457244873,
      "learning_rate": 4.763557734936134e-05,
      "loss": 1.1828549385070801,
      "step": 3980
    },
    {
      "epoch": 0.42168117626854434,
      "grad_norm": 1.4351997375488281,
      "learning_rate": 4.762381086788682e-05,
      "loss": 1.112205696105957,
      "step": 3990
    },
    {
      "epoch": 0.42273802132184896,
      "grad_norm": 0.944873571395874,
      "learning_rate": 4.761201664188001e-05,
      "loss": 1.1933209419250488,
      "step": 4000
    },
    {
      "epoch": 0.42273802132184896,
      "eval_loss": 0.5499118566513062,
      "eval_runtime": 2996.9378,
      "eval_samples_per_second": 2.971,
      "eval_steps_per_second": 2.971,
      "step": 4000
    },
    {
      "epoch": 0.42379486637515357,
      "grad_norm": 1.0244677066802979,
      "learning_rate": 4.760019468580468e-05,
      "loss": 1.1696236610412598,
      "step": 4010
    },
    {
      "epoch": 0.4248517114284582,
      "grad_norm": 0.8759726881980896,
      "learning_rate": 4.758834501415857e-05,
      "loss": 1.0558359146118164,
      "step": 4020
    },
    {
      "epoch": 0.4259085564817628,
      "grad_norm": 1.1339246034622192,
      "learning_rate": 4.7576467641473464e-05,
      "loss": 1.1959860801696778,
      "step": 4030
    },
    {
      "epoch": 0.42696540153506746,
      "grad_norm": 1.6854870319366455,
      "learning_rate": 4.756456258231506e-05,
      "loss": 1.0786171913146974,
      "step": 4040
    },
    {
      "epoch": 0.42802224658837207,
      "grad_norm": 1.1308907270431519,
      "learning_rate": 4.7552629851283035e-05,
      "loss": 1.1679159164428712,
      "step": 4050
    },
    {
      "epoch": 0.4290790916416767,
      "grad_norm": 0.9798864126205444,
      "learning_rate": 4.754066946301101e-05,
      "loss": 1.1031585693359376,
      "step": 4060
    },
    {
      "epoch": 0.4301359366949813,
      "grad_norm": 0.9276269674301147,
      "learning_rate": 4.752868143216651e-05,
      "loss": 1.2838751792907714,
      "step": 4070
    },
    {
      "epoch": 0.4311927817482859,
      "grad_norm": 1.4536696672439575,
      "learning_rate": 4.7516665773450954e-05,
      "loss": 1.1679118156433106,
      "step": 4080
    },
    {
      "epoch": 0.43224962680159057,
      "grad_norm": 0.918262779712677,
      "learning_rate": 4.750462250159966e-05,
      "loss": 1.1335436820983886,
      "step": 4090
    },
    {
      "epoch": 0.4333064718548952,
      "grad_norm": 1.124735713005066,
      "learning_rate": 4.74925516313818e-05,
      "loss": 1.1431709289550782,
      "step": 4100
    },
    {
      "epoch": 0.4343633169081998,
      "grad_norm": 0.8246219754219055,
      "learning_rate": 4.748045317760038e-05,
      "loss": 1.3645395278930663,
      "step": 4110
    },
    {
      "epoch": 0.4354201619615044,
      "grad_norm": 1.0486845970153809,
      "learning_rate": 4.746832715509225e-05,
      "loss": 1.1693713188171386,
      "step": 4120
    },
    {
      "epoch": 0.436477007014809,
      "grad_norm": 0.973565936088562,
      "learning_rate": 4.7456173578728066e-05,
      "loss": 1.0508963584899902,
      "step": 4130
    },
    {
      "epoch": 0.4375338520681137,
      "grad_norm": 0.9348598718643188,
      "learning_rate": 4.744399246341227e-05,
      "loss": 1.2168437957763671,
      "step": 4140
    },
    {
      "epoch": 0.4385906971214183,
      "grad_norm": 0.7487098574638367,
      "learning_rate": 4.7431783824083066e-05,
      "loss": 1.1066270828247071,
      "step": 4150
    },
    {
      "epoch": 0.4396475421747229,
      "grad_norm": 1.0862822532653809,
      "learning_rate": 4.7419547675712433e-05,
      "loss": 1.1416950225830078,
      "step": 4160
    },
    {
      "epoch": 0.4407043872280275,
      "grad_norm": 1.3421623706817627,
      "learning_rate": 4.740728403330608e-05,
      "loss": 1.116429901123047,
      "step": 4170
    },
    {
      "epoch": 0.44176123228133213,
      "grad_norm": 0.888343334197998,
      "learning_rate": 4.7394992911903414e-05,
      "loss": 1.116634464263916,
      "step": 4180
    },
    {
      "epoch": 0.4428180773346368,
      "grad_norm": 0.9337888360023499,
      "learning_rate": 4.7382674326577556e-05,
      "loss": 1.2761999130249024,
      "step": 4190
    },
    {
      "epoch": 0.4438749223879414,
      "grad_norm": 1.0329607725143433,
      "learning_rate": 4.7370328292435326e-05,
      "loss": 1.1602212905883789,
      "step": 4200
    },
    {
      "epoch": 0.444931767441246,
      "grad_norm": 0.9183998703956604,
      "learning_rate": 4.735795482461716e-05,
      "loss": 1.17144775390625,
      "step": 4210
    },
    {
      "epoch": 0.44598861249455063,
      "grad_norm": 0.828599750995636,
      "learning_rate": 4.7345553938297184e-05,
      "loss": 1.1746791839599608,
      "step": 4220
    },
    {
      "epoch": 0.44704545754785524,
      "grad_norm": 0.8341064453125,
      "learning_rate": 4.733312564868311e-05,
      "loss": 1.1004858016967773,
      "step": 4230
    },
    {
      "epoch": 0.4481023026011599,
      "grad_norm": 1.16808021068573,
      "learning_rate": 4.732066997101629e-05,
      "loss": 1.1439414024353027,
      "step": 4240
    },
    {
      "epoch": 0.4491591476544645,
      "grad_norm": 0.7673822045326233,
      "learning_rate": 4.7308186920571627e-05,
      "loss": 1.1138094902038573,
      "step": 4250
    },
    {
      "epoch": 0.45021599270776913,
      "grad_norm": 1.1580101251602173,
      "learning_rate": 4.729567651265763e-05,
      "loss": 1.1672134399414062,
      "step": 4260
    },
    {
      "epoch": 0.45127283776107374,
      "grad_norm": 0.6843805909156799,
      "learning_rate": 4.728313876261633e-05,
      "loss": 1.2398715972900392,
      "step": 4270
    },
    {
      "epoch": 0.45232968281437835,
      "grad_norm": 1.0429879426956177,
      "learning_rate": 4.727057368582329e-05,
      "loss": 1.090687084197998,
      "step": 4280
    },
    {
      "epoch": 0.453386527867683,
      "grad_norm": 1.3027827739715576,
      "learning_rate": 4.725798129768762e-05,
      "loss": 1.2237675666809082,
      "step": 4290
    },
    {
      "epoch": 0.45444337292098763,
      "grad_norm": 0.8755503296852112,
      "learning_rate": 4.7245361613651863e-05,
      "loss": 1.049496841430664,
      "step": 4300
    },
    {
      "epoch": 0.45550021797429224,
      "grad_norm": 0.9413042664527893,
      "learning_rate": 4.723271464919209e-05,
      "loss": 1.313827896118164,
      "step": 4310
    },
    {
      "epoch": 0.45655706302759685,
      "grad_norm": 1.477206826210022,
      "learning_rate": 4.72200404198178e-05,
      "loss": 1.1051231384277345,
      "step": 4320
    },
    {
      "epoch": 0.45761390808090147,
      "grad_norm": 0.8855541348457336,
      "learning_rate": 4.720733894107193e-05,
      "loss": 1.1010584831237793,
      "step": 4330
    },
    {
      "epoch": 0.45867075313420613,
      "grad_norm": 0.8714260458946228,
      "learning_rate": 4.719461022853085e-05,
      "loss": 1.1482172966003419,
      "step": 4340
    },
    {
      "epoch": 0.45972759818751074,
      "grad_norm": 0.8883099555969238,
      "learning_rate": 4.7181854297804306e-05,
      "loss": 1.2059648513793946,
      "step": 4350
    },
    {
      "epoch": 0.46078444324081536,
      "grad_norm": 0.719153642654419,
      "learning_rate": 4.716907116453544e-05,
      "loss": 1.004715633392334,
      "step": 4360
    },
    {
      "epoch": 0.46184128829411997,
      "grad_norm": 1.057906985282898,
      "learning_rate": 4.715626084440075e-05,
      "loss": 1.095588493347168,
      "step": 4370
    },
    {
      "epoch": 0.4628981333474246,
      "grad_norm": 1.083505392074585,
      "learning_rate": 4.714342335311005e-05,
      "loss": 1.139674472808838,
      "step": 4380
    },
    {
      "epoch": 0.46395497840072925,
      "grad_norm": 1.0921616554260254,
      "learning_rate": 4.7130558706406515e-05,
      "loss": 1.137366485595703,
      "step": 4390
    },
    {
      "epoch": 0.46501182345403386,
      "grad_norm": 0.8708337545394897,
      "learning_rate": 4.7117666920066606e-05,
      "loss": 1.0842301368713378,
      "step": 4400
    },
    {
      "epoch": 0.46606866850733847,
      "grad_norm": 0.8267654776573181,
      "learning_rate": 4.7104748009900054e-05,
      "loss": 1.1224099159240724,
      "step": 4410
    },
    {
      "epoch": 0.4671255135606431,
      "grad_norm": 1.0303797721862793,
      "learning_rate": 4.709180199174986e-05,
      "loss": 0.9776995658874512,
      "step": 4420
    },
    {
      "epoch": 0.4681823586139477,
      "grad_norm": 0.9997529983520508,
      "learning_rate": 4.7078828881492284e-05,
      "loss": 1.1280205726623536,
      "step": 4430
    },
    {
      "epoch": 0.46923920366725236,
      "grad_norm": 0.8714220523834229,
      "learning_rate": 4.706582869503679e-05,
      "loss": 1.0819414138793946,
      "step": 4440
    },
    {
      "epoch": 0.47029604872055697,
      "grad_norm": 0.9913234710693359,
      "learning_rate": 4.7052801448326056e-05,
      "loss": 1.0859429359436035,
      "step": 4450
    },
    {
      "epoch": 0.4713528937738616,
      "grad_norm": 0.9478439688682556,
      "learning_rate": 4.703974715733594e-05,
      "loss": 1.223191738128662,
      "step": 4460
    },
    {
      "epoch": 0.4724097388271662,
      "grad_norm": 1.1998858451843262,
      "learning_rate": 4.702666583807547e-05,
      "loss": 1.2161993026733398,
      "step": 4470
    },
    {
      "epoch": 0.4734665838804708,
      "grad_norm": 0.8407456874847412,
      "learning_rate": 4.701355750658683e-05,
      "loss": 1.229953670501709,
      "step": 4480
    },
    {
      "epoch": 0.47452342893377547,
      "grad_norm": 1.101136565208435,
      "learning_rate": 4.70004221789453e-05,
      "loss": 1.0488434791564942,
      "step": 4490
    },
    {
      "epoch": 0.4755802739870801,
      "grad_norm": 1.1977648735046387,
      "learning_rate": 4.6987259871259293e-05,
      "loss": 1.2075418472290038,
      "step": 4500
    },
    {
      "epoch": 0.4755802739870801,
      "eval_loss": 0.54176926612854,
      "eval_runtime": 3000.0502,
      "eval_samples_per_second": 2.968,
      "eval_steps_per_second": 2.968,
      "step": 4500
    },
    {
      "epoch": 0.4766371190403847,
      "grad_norm": 1.0188548564910889,
      "learning_rate": 4.69740705996703e-05,
      "loss": 1.1928425788879395,
      "step": 4510
    },
    {
      "epoch": 0.4776939640936893,
      "grad_norm": 0.9434320330619812,
      "learning_rate": 4.6960854380352895e-05,
      "loss": 1.1588902473449707,
      "step": 4520
    },
    {
      "epoch": 0.4787508091469939,
      "grad_norm": 1.0094636678695679,
      "learning_rate": 4.694761122951467e-05,
      "loss": 1.1453936576843262,
      "step": 4530
    },
    {
      "epoch": 0.4798076542002986,
      "grad_norm": 1.171395182609558,
      "learning_rate": 4.693434116339626e-05,
      "loss": 1.2170626640319824,
      "step": 4540
    },
    {
      "epoch": 0.4808644992536032,
      "grad_norm": 1.5326592922210693,
      "learning_rate": 4.692104419827131e-05,
      "loss": 1.2162393569946288,
      "step": 4550
    },
    {
      "epoch": 0.4819213443069078,
      "grad_norm": 1.1375632286071777,
      "learning_rate": 4.690772035044645e-05,
      "loss": 1.1326173782348632,
      "step": 4560
    },
    {
      "epoch": 0.4829781893602124,
      "grad_norm": 1.0120803117752075,
      "learning_rate": 4.6894369636261265e-05,
      "loss": 1.2757190704345702,
      "step": 4570
    },
    {
      "epoch": 0.48403503441351703,
      "grad_norm": 0.8788567185401917,
      "learning_rate": 4.68809920720883e-05,
      "loss": 1.2282956123352051,
      "step": 4580
    },
    {
      "epoch": 0.4850918794668217,
      "grad_norm": 0.7742215394973755,
      "learning_rate": 4.686758767433305e-05,
      "loss": 1.2346811294555664,
      "step": 4590
    },
    {
      "epoch": 0.4861487245201263,
      "grad_norm": 0.8508361577987671,
      "learning_rate": 4.6854156459433865e-05,
      "loss": 1.0969385147094726,
      "step": 4600
    },
    {
      "epoch": 0.4872055695734309,
      "grad_norm": 0.08558930456638336,
      "learning_rate": 4.684069844386201e-05,
      "loss": 0.958951473236084,
      "step": 4610
    },
    {
      "epoch": 0.48826241462673553,
      "grad_norm": 1.042853832244873,
      "learning_rate": 4.6827213644121624e-05,
      "loss": 1.3046652793884277,
      "step": 4620
    },
    {
      "epoch": 0.48931925968004014,
      "grad_norm": 1.1812355518341064,
      "learning_rate": 4.68137020767497e-05,
      "loss": 1.16186466217041,
      "step": 4630
    },
    {
      "epoch": 0.4903761047333448,
      "grad_norm": 0.8571643829345703,
      "learning_rate": 4.6800163758316016e-05,
      "loss": 1.2297818183898925,
      "step": 4640
    },
    {
      "epoch": 0.4914329497866494,
      "grad_norm": 0.988733172416687,
      "learning_rate": 4.67865987054232e-05,
      "loss": 1.0920302391052246,
      "step": 4650
    },
    {
      "epoch": 0.49248979483995403,
      "grad_norm": 1.1338179111480713,
      "learning_rate": 4.677300693470664e-05,
      "loss": 1.1325450897216798,
      "step": 4660
    },
    {
      "epoch": 0.49354663989325864,
      "grad_norm": 0.7361093759536743,
      "learning_rate": 4.6759388462834495e-05,
      "loss": 1.2138700485229492,
      "step": 4670
    },
    {
      "epoch": 0.49460348494656325,
      "grad_norm": 0.7568942308425903,
      "learning_rate": 4.674574330650768e-05,
      "loss": 1.1726446151733398,
      "step": 4680
    },
    {
      "epoch": 0.4956603299998679,
      "grad_norm": 0.7813202142715454,
      "learning_rate": 4.6732071482459816e-05,
      "loss": 1.1246416091918945,
      "step": 4690
    },
    {
      "epoch": 0.49671717505317253,
      "grad_norm": 1.1250648498535156,
      "learning_rate": 4.671837300745724e-05,
      "loss": 1.125599479675293,
      "step": 4700
    },
    {
      "epoch": 0.49777402010647714,
      "grad_norm": 1.3319194316864014,
      "learning_rate": 4.670464789829897e-05,
      "loss": 1.2636144638061524,
      "step": 4710
    },
    {
      "epoch": 0.49883086515978176,
      "grad_norm": 1.2823762893676758,
      "learning_rate": 4.669089617181669e-05,
      "loss": 1.168724822998047,
      "step": 4720
    },
    {
      "epoch": 0.49988771021308637,
      "grad_norm": 1.150941252708435,
      "learning_rate": 4.667711784487471e-05,
      "loss": 1.1878308296203612,
      "step": 4730
    },
    {
      "epoch": 0.500944555266391,
      "grad_norm": 1.1700901985168457,
      "learning_rate": 4.6663312934369976e-05,
      "loss": 1.0694887161254882,
      "step": 4740
    },
    {
      "epoch": 0.5020014003196956,
      "grad_norm": 1.1769229173660278,
      "learning_rate": 4.664948145723203e-05,
      "loss": 1.1483065605163574,
      "step": 4750
    },
    {
      "epoch": 0.5030582453730003,
      "grad_norm": 1.1564068794250488,
      "learning_rate": 4.663562343042301e-05,
      "loss": 0.9774134635925293,
      "step": 4760
    },
    {
      "epoch": 0.5041150904263049,
      "grad_norm": 0.783980131149292,
      "learning_rate": 4.662173887093759e-05,
      "loss": 1.0641669273376464,
      "step": 4770
    },
    {
      "epoch": 0.5051719354796095,
      "grad_norm": 1.2440720796585083,
      "learning_rate": 4.6607827795802984e-05,
      "loss": 1.100968837738037,
      "step": 4780
    },
    {
      "epoch": 0.5062287805329141,
      "grad_norm": 0.8902670741081238,
      "learning_rate": 4.659389022207893e-05,
      "loss": 1.1236146926879882,
      "step": 4790
    },
    {
      "epoch": 0.5072856255862187,
      "grad_norm": 1.0380364656448364,
      "learning_rate": 4.6579926166857666e-05,
      "loss": 1.1546918869018554,
      "step": 4800
    },
    {
      "epoch": 0.5083424706395233,
      "grad_norm": 1.4376883506774902,
      "learning_rate": 4.6565935647263896e-05,
      "loss": 1.0818827629089356,
      "step": 4810
    },
    {
      "epoch": 0.509399315692828,
      "grad_norm": 1.2759841680526733,
      "learning_rate": 4.655191868045479e-05,
      "loss": 1.1487488746643066,
      "step": 4820
    },
    {
      "epoch": 0.5104561607461326,
      "grad_norm": 1.0565979480743408,
      "learning_rate": 4.653787528361995e-05,
      "loss": 1.201682949066162,
      "step": 4830
    },
    {
      "epoch": 0.5115130057994373,
      "grad_norm": 0.7899355888366699,
      "learning_rate": 4.6523805473981365e-05,
      "loss": 1.0663064002990723,
      "step": 4840
    },
    {
      "epoch": 0.5125698508527419,
      "grad_norm": 0.8458067774772644,
      "learning_rate": 4.6509709268793455e-05,
      "loss": 1.1666646957397462,
      "step": 4850
    },
    {
      "epoch": 0.5136266959060465,
      "grad_norm": 0.8895781636238098,
      "learning_rate": 4.649558668534298e-05,
      "loss": 1.2018442153930664,
      "step": 4860
    },
    {
      "epoch": 0.5146835409593511,
      "grad_norm": 0.8418881297111511,
      "learning_rate": 4.6481437740949063e-05,
      "loss": 1.130838966369629,
      "step": 4870
    },
    {
      "epoch": 0.5157403860126557,
      "grad_norm": 1.4033600091934204,
      "learning_rate": 4.6467262452963145e-05,
      "loss": 1.0859160423278809,
      "step": 4880
    },
    {
      "epoch": 0.5167972310659603,
      "grad_norm": 1.4001036882400513,
      "learning_rate": 4.6453060838768984e-05,
      "loss": 1.0864112854003907,
      "step": 4890
    },
    {
      "epoch": 0.5178540761192649,
      "grad_norm": 0.8056440949440002,
      "learning_rate": 4.643883291578262e-05,
      "loss": 0.9933370590209961,
      "step": 4900
    },
    {
      "epoch": 0.5189109211725695,
      "grad_norm": 0.9766755700111389,
      "learning_rate": 4.642457870145235e-05,
      "loss": 1.1287412643432617,
      "step": 4910
    },
    {
      "epoch": 0.5199677662258743,
      "grad_norm": 1.072831392288208,
      "learning_rate": 4.641029821325872e-05,
      "loss": 1.088283157348633,
      "step": 4920
    },
    {
      "epoch": 0.5210246112791789,
      "grad_norm": 1.244500756263733,
      "learning_rate": 4.6395991468714494e-05,
      "loss": 0.9080944061279297,
      "step": 4930
    },
    {
      "epoch": 0.5220814563324835,
      "grad_norm": 1.090822696685791,
      "learning_rate": 4.638165848536463e-05,
      "loss": 1.1479074478149414,
      "step": 4940
    },
    {
      "epoch": 0.5231383013857881,
      "grad_norm": 1.1869769096374512,
      "learning_rate": 4.6367299280786275e-05,
      "loss": 1.1481531143188477,
      "step": 4950
    },
    {
      "epoch": 0.5241951464390927,
      "grad_norm": 0.909321665763855,
      "learning_rate": 4.635291387258872e-05,
      "loss": 0.9884790420532227,
      "step": 4960
    },
    {
      "epoch": 0.5252519914923973,
      "grad_norm": 1.1562092304229736,
      "learning_rate": 4.633850227841341e-05,
      "loss": 1.1633078575134277,
      "step": 4970
    },
    {
      "epoch": 0.5263088365457019,
      "grad_norm": 1.1187078952789307,
      "learning_rate": 4.632406451593389e-05,
      "loss": 1.0225772857666016,
      "step": 4980
    },
    {
      "epoch": 0.5273656815990065,
      "grad_norm": 0.9759521484375,
      "learning_rate": 4.6309600602855766e-05,
      "loss": 1.0769665718078614,
      "step": 4990
    },
    {
      "epoch": 0.5284225266523112,
      "grad_norm": 0.7091618180274963,
      "learning_rate": 4.6295110556916774e-05,
      "loss": 1.068688201904297,
      "step": 5000
    },
    {
      "epoch": 0.5284225266523112,
      "eval_loss": 0.5442911982536316,
      "eval_runtime": 3000.9836,
      "eval_samples_per_second": 2.967,
      "eval_steps_per_second": 2.967,
      "step": 5000
    },
    {
      "epoch": 0.5294793717056158,
      "grad_norm": 1.557794451713562,
      "learning_rate": 4.628059439588667e-05,
      "loss": 1.0887509346008302,
      "step": 5010
    },
    {
      "epoch": 0.5305362167589205,
      "grad_norm": 0.9002912640571594,
      "learning_rate": 4.626605213756722e-05,
      "loss": 1.0685564041137696,
      "step": 5020
    },
    {
      "epoch": 0.5315930618122251,
      "grad_norm": 1.1652411222457886,
      "learning_rate": 4.625148379979221e-05,
      "loss": 1.0549007415771485,
      "step": 5030
    },
    {
      "epoch": 0.5326499068655297,
      "grad_norm": 0.6998447775840759,
      "learning_rate": 4.623688940042741e-05,
      "loss": 1.0208648681640624,
      "step": 5040
    },
    {
      "epoch": 0.5337067519188343,
      "grad_norm": 0.8890730738639832,
      "learning_rate": 4.6222268957370534e-05,
      "loss": 1.2226201057434083,
      "step": 5050
    },
    {
      "epoch": 0.5347635969721389,
      "grad_norm": 1.2933248281478882,
      "learning_rate": 4.6207622488551265e-05,
      "loss": 1.197833251953125,
      "step": 5060
    },
    {
      "epoch": 0.5358204420254435,
      "grad_norm": 1.575124979019165,
      "learning_rate": 4.619295001193118e-05,
      "loss": 1.1545106887817382,
      "step": 5070
    },
    {
      "epoch": 0.5368772870787482,
      "grad_norm": 1.306925654411316,
      "learning_rate": 4.617825154550375e-05,
      "loss": 1.1827224731445312,
      "step": 5080
    },
    {
      "epoch": 0.5379341321320528,
      "grad_norm": 0.9786710143089294,
      "learning_rate": 4.616352710729433e-05,
      "loss": 1.1237198829650878,
      "step": 5090
    },
    {
      "epoch": 0.5389909771853574,
      "grad_norm": 0.8946554064750671,
      "learning_rate": 4.614877671536011e-05,
      "loss": 1.1585297584533691,
      "step": 5100
    },
    {
      "epoch": 0.540047822238662,
      "grad_norm": 1.031363606452942,
      "learning_rate": 4.6134000387790124e-05,
      "loss": 1.123582363128662,
      "step": 5110
    },
    {
      "epoch": 0.5411046672919666,
      "grad_norm": 1.701038122177124,
      "learning_rate": 4.61191981427052e-05,
      "loss": 1.1748602867126465,
      "step": 5120
    },
    {
      "epoch": 0.5421615123452713,
      "grad_norm": 0.8009101748466492,
      "learning_rate": 4.610436999825797e-05,
      "loss": 1.1146693229675293,
      "step": 5130
    },
    {
      "epoch": 0.5432183573985759,
      "grad_norm": 1.4872727394104004,
      "learning_rate": 4.608951597263279e-05,
      "loss": 1.2213695526123047,
      "step": 5140
    },
    {
      "epoch": 0.5442752024518805,
      "grad_norm": 1.088947057723999,
      "learning_rate": 4.6074636084045787e-05,
      "loss": 1.1583603858947753,
      "step": 5150
    },
    {
      "epoch": 0.5453320475051852,
      "grad_norm": 0.857261061668396,
      "learning_rate": 4.6059730350744794e-05,
      "loss": 1.124300479888916,
      "step": 5160
    },
    {
      "epoch": 0.5463888925584898,
      "grad_norm": 0.9131742715835571,
      "learning_rate": 4.6044798791009346e-05,
      "loss": 1.1281977653503419,
      "step": 5170
    },
    {
      "epoch": 0.5474457376117944,
      "grad_norm": 0.9775580167770386,
      "learning_rate": 4.602984142315064e-05,
      "loss": 1.2213180541992188,
      "step": 5180
    },
    {
      "epoch": 0.548502582665099,
      "grad_norm": 0.7401923537254333,
      "learning_rate": 4.601485826551152e-05,
      "loss": 1.1453153610229492,
      "step": 5190
    },
    {
      "epoch": 0.5495594277184036,
      "grad_norm": 0.954428493976593,
      "learning_rate": 4.599984933646647e-05,
      "loss": 1.0825040817260743,
      "step": 5200
    },
    {
      "epoch": 0.5506162727717082,
      "grad_norm": 1.0791335105895996,
      "learning_rate": 4.598481465442157e-05,
      "loss": 1.1515878677368163,
      "step": 5210
    },
    {
      "epoch": 0.5516731178250128,
      "grad_norm": 0.9535002708435059,
      "learning_rate": 4.596975423781449e-05,
      "loss": 1.139879322052002,
      "step": 5220
    },
    {
      "epoch": 0.5527299628783175,
      "grad_norm": 0.8692033290863037,
      "learning_rate": 4.595466810511444e-05,
      "loss": 1.1290972709655762,
      "step": 5230
    },
    {
      "epoch": 0.5537868079316222,
      "grad_norm": 1.1980499029159546,
      "learning_rate": 4.593955627482219e-05,
      "loss": 1.1121634483337401,
      "step": 5240
    },
    {
      "epoch": 0.5548436529849268,
      "grad_norm": 0.9159504771232605,
      "learning_rate": 4.592441876547001e-05,
      "loss": 1.2194503784179687,
      "step": 5250
    },
    {
      "epoch": 0.5559004980382314,
      "grad_norm": 0.8827317953109741,
      "learning_rate": 4.590925559562167e-05,
      "loss": 1.0840855598449708,
      "step": 5260
    },
    {
      "epoch": 0.556957343091536,
      "grad_norm": 0.855391800403595,
      "learning_rate": 4.5894066783872396e-05,
      "loss": 0.8109023094177246,
      "step": 5270
    },
    {
      "epoch": 0.5580141881448406,
      "grad_norm": 1.15868079662323,
      "learning_rate": 4.587885234884888e-05,
      "loss": 1.229086685180664,
      "step": 5280
    },
    {
      "epoch": 0.5590710331981452,
      "grad_norm": 0.9313871264457703,
      "learning_rate": 4.586361230920921e-05,
      "loss": 1.078841209411621,
      "step": 5290
    },
    {
      "epoch": 0.5601278782514498,
      "grad_norm": 0.8777920007705688,
      "learning_rate": 4.58483466836429e-05,
      "loss": 1.132593536376953,
      "step": 5300
    },
    {
      "epoch": 0.5611847233047544,
      "grad_norm": 0.800299882888794,
      "learning_rate": 4.583305549087082e-05,
      "loss": 1.1072908401489259,
      "step": 5310
    },
    {
      "epoch": 0.562241568358059,
      "grad_norm": 1.2205872535705566,
      "learning_rate": 4.581773874964522e-05,
      "loss": 1.0527151107788086,
      "step": 5320
    },
    {
      "epoch": 0.5632984134113638,
      "grad_norm": 1.2091313600540161,
      "learning_rate": 4.5802396478749645e-05,
      "loss": 1.1779120445251465,
      "step": 5330
    },
    {
      "epoch": 0.5643552584646684,
      "grad_norm": 1.2267571687698364,
      "learning_rate": 4.578702869699898e-05,
      "loss": 1.155374240875244,
      "step": 5340
    },
    {
      "epoch": 0.565412103517973,
      "grad_norm": 1.3349448442459106,
      "learning_rate": 4.577163542323938e-05,
      "loss": 1.2411553382873535,
      "step": 5350
    },
    {
      "epoch": 0.5664689485712776,
      "grad_norm": 1.0918763875961304,
      "learning_rate": 4.575621667634827e-05,
      "loss": 1.1463045120239257,
      "step": 5360
    },
    {
      "epoch": 0.5675257936245822,
      "grad_norm": 0.993800699710846,
      "learning_rate": 4.57407724752343e-05,
      "loss": 1.1162293434143067,
      "step": 5370
    },
    {
      "epoch": 0.5685826386778868,
      "grad_norm": 0.8286049962043762,
      "learning_rate": 4.572530283883735e-05,
      "loss": 1.1724379539489747,
      "step": 5380
    },
    {
      "epoch": 0.5696394837311914,
      "grad_norm": 0.9208925366401672,
      "learning_rate": 4.5709807786128486e-05,
      "loss": 1.141054344177246,
      "step": 5390
    },
    {
      "epoch": 0.570696328784496,
      "grad_norm": 0.7895875573158264,
      "learning_rate": 4.569428733610994e-05,
      "loss": 0.9826958656311036,
      "step": 5400
    },
    {
      "epoch": 0.5717531738378007,
      "grad_norm": 1.0654233694076538,
      "learning_rate": 4.56787415078151e-05,
      "loss": 1.2282355308532715,
      "step": 5410
    },
    {
      "epoch": 0.5728100188911053,
      "grad_norm": 0.9458472728729248,
      "learning_rate": 4.566317032030847e-05,
      "loss": 1.1457016944885254,
      "step": 5420
    },
    {
      "epoch": 0.57386686394441,
      "grad_norm": 0.7015053629875183,
      "learning_rate": 4.564757379268565e-05,
      "loss": 1.2201309204101562,
      "step": 5430
    },
    {
      "epoch": 0.5749237089977146,
      "grad_norm": 0.976117730140686,
      "learning_rate": 4.5631951944073316e-05,
      "loss": 1.1378135681152344,
      "step": 5440
    },
    {
      "epoch": 0.5759805540510192,
      "grad_norm": 1.0348179340362549,
      "learning_rate": 4.56163047936292e-05,
      "loss": 1.2080279350280763,
      "step": 5450
    },
    {
      "epoch": 0.5770373991043238,
      "grad_norm": 0.8362106680870056,
      "learning_rate": 4.560063236054206e-05,
      "loss": 1.0412700653076172,
      "step": 5460
    },
    {
      "epoch": 0.5780942441576284,
      "grad_norm": 1.0793880224227905,
      "learning_rate": 4.558493466403167e-05,
      "loss": 1.1133831024169922,
      "step": 5470
    },
    {
      "epoch": 0.5791510892109331,
      "grad_norm": 1.1507678031921387,
      "learning_rate": 4.556921172334878e-05,
      "loss": 1.0750234603881836,
      "step": 5480
    },
    {
      "epoch": 0.5802079342642377,
      "grad_norm": 0.8763432502746582,
      "learning_rate": 4.555346355777508e-05,
      "loss": 1.1775161743164062,
      "step": 5490
    },
    {
      "epoch": 0.5812647793175423,
      "grad_norm": 1.4622514247894287,
      "learning_rate": 4.553769018662321e-05,
      "loss": 1.0309015274047852,
      "step": 5500
    },
    {
      "epoch": 0.5812647793175423,
      "eval_loss": 0.5370742082595825,
      "eval_runtime": 3000.9627,
      "eval_samples_per_second": 2.967,
      "eval_steps_per_second": 2.967,
      "step": 5500
    },
    {
      "epoch": 0.5823216243708469,
      "grad_norm": 0.8874773383140564,
      "learning_rate": 4.5521891629236725e-05,
      "loss": 1.05091552734375,
      "step": 5510
    },
    {
      "epoch": 0.5833784694241515,
      "grad_norm": 1.1255704164505005,
      "learning_rate": 4.550606790499006e-05,
      "loss": 1.2015826225280761,
      "step": 5520
    },
    {
      "epoch": 0.5844353144774562,
      "grad_norm": 1.2548490762710571,
      "learning_rate": 4.5490219033288516e-05,
      "loss": 1.1870062828063965,
      "step": 5530
    },
    {
      "epoch": 0.5854921595307608,
      "grad_norm": 1.3639283180236816,
      "learning_rate": 4.547434503356823e-05,
      "loss": 1.0649425506591796,
      "step": 5540
    },
    {
      "epoch": 0.5865490045840654,
      "grad_norm": 0.935068666934967,
      "learning_rate": 4.5458445925296164e-05,
      "loss": 1.1590648651123048,
      "step": 5550
    },
    {
      "epoch": 0.5876058496373701,
      "grad_norm": 1.3381494283676147,
      "learning_rate": 4.5442521727970054e-05,
      "loss": 0.9667620658874512,
      "step": 5560
    },
    {
      "epoch": 0.5886626946906747,
      "grad_norm": 1.0000063180923462,
      "learning_rate": 4.5426572461118414e-05,
      "loss": 1.161902904510498,
      "step": 5570
    },
    {
      "epoch": 0.5897195397439793,
      "grad_norm": 0.3359367847442627,
      "learning_rate": 4.541059814430051e-05,
      "loss": 0.9491029739379883,
      "step": 5580
    },
    {
      "epoch": 0.5907763847972839,
      "grad_norm": 1.2903763055801392,
      "learning_rate": 4.539459879710631e-05,
      "loss": 1.0871858596801758,
      "step": 5590
    },
    {
      "epoch": 0.5918332298505885,
      "grad_norm": 0.9106523990631104,
      "learning_rate": 4.53785744391565e-05,
      "loss": 1.166527557373047,
      "step": 5600
    },
    {
      "epoch": 0.5928900749038931,
      "grad_norm": 1.2166739702224731,
      "learning_rate": 4.536252509010241e-05,
      "loss": 1.150502872467041,
      "step": 5610
    },
    {
      "epoch": 0.5939469199571977,
      "grad_norm": 1.017216444015503,
      "learning_rate": 4.534645076962605e-05,
      "loss": 1.0088006973266601,
      "step": 5620
    },
    {
      "epoch": 0.5950037650105025,
      "grad_norm": 1.1611486673355103,
      "learning_rate": 4.533035149744002e-05,
      "loss": 1.1376809120178222,
      "step": 5630
    },
    {
      "epoch": 0.5960606100638071,
      "grad_norm": 0.8441755771636963,
      "learning_rate": 4.531422729328755e-05,
      "loss": 0.9843733787536622,
      "step": 5640
    },
    {
      "epoch": 0.5971174551171117,
      "grad_norm": 0.5741209983825684,
      "learning_rate": 4.529807817694242e-05,
      "loss": 0.9225249290466309,
      "step": 5650
    },
    {
      "epoch": 0.5981743001704163,
      "grad_norm": 1.179796814918518,
      "learning_rate": 4.528190416820897e-05,
      "loss": 1.2210892677307128,
      "step": 5660
    },
    {
      "epoch": 0.5992311452237209,
      "grad_norm": 1.0102235078811646,
      "learning_rate": 4.526570528692207e-05,
      "loss": 1.2831805229187012,
      "step": 5670
    },
    {
      "epoch": 0.6002879902770255,
      "grad_norm": 0.9570515751838684,
      "learning_rate": 4.52494815529471e-05,
      "loss": 1.1874283790588378,
      "step": 5680
    },
    {
      "epoch": 0.6013448353303301,
      "grad_norm": 1.0489470958709717,
      "learning_rate": 4.523323298617991e-05,
      "loss": 1.0246089935302733,
      "step": 5690
    },
    {
      "epoch": 0.6024016803836347,
      "grad_norm": 1.073297381401062,
      "learning_rate": 4.521695960654678e-05,
      "loss": 1.1010830879211426,
      "step": 5700
    },
    {
      "epoch": 0.6034585254369393,
      "grad_norm": 0.8267492651939392,
      "learning_rate": 4.5200661434004465e-05,
      "loss": 1.177913284301758,
      "step": 5710
    },
    {
      "epoch": 0.604515370490244,
      "grad_norm": 1.2806553840637207,
      "learning_rate": 4.518433848854009e-05,
      "loss": 1.2559059143066407,
      "step": 5720
    },
    {
      "epoch": 0.6055722155435487,
      "grad_norm": 1.148545503616333,
      "learning_rate": 4.516799079017116e-05,
      "loss": 1.2286364555358886,
      "step": 5730
    },
    {
      "epoch": 0.6066290605968533,
      "grad_norm": 0.717181384563446,
      "learning_rate": 4.5151618358945565e-05,
      "loss": 1.2152445793151856,
      "step": 5740
    },
    {
      "epoch": 0.6076859056501579,
      "grad_norm": 1.2763407230377197,
      "learning_rate": 4.51352212149415e-05,
      "loss": 1.1291117668151855,
      "step": 5750
    },
    {
      "epoch": 0.6087427507034625,
      "grad_norm": 0.9977017641067505,
      "learning_rate": 4.5118799378267475e-05,
      "loss": 1.15321044921875,
      "step": 5760
    },
    {
      "epoch": 0.6097995957567671,
      "grad_norm": 0.8340393900871277,
      "learning_rate": 4.510235286906227e-05,
      "loss": 1.1166489601135254,
      "step": 5770
    },
    {
      "epoch": 0.6108564408100717,
      "grad_norm": 1.11283278465271,
      "learning_rate": 4.5085881707494945e-05,
      "loss": 1.085316276550293,
      "step": 5780
    },
    {
      "epoch": 0.6119132858633763,
      "grad_norm": 0.968985378742218,
      "learning_rate": 4.5069385913764776e-05,
      "loss": 1.2764604568481446,
      "step": 5790
    },
    {
      "epoch": 0.612970130916681,
      "grad_norm": 1.005826473236084,
      "learning_rate": 4.505286550810125e-05,
      "loss": 1.1662323951721192,
      "step": 5800
    },
    {
      "epoch": 0.6140269759699856,
      "grad_norm": 0.8046248555183411,
      "learning_rate": 4.503632051076403e-05,
      "loss": 1.1204625129699708,
      "step": 5810
    },
    {
      "epoch": 0.6150838210232902,
      "grad_norm": 1.2128384113311768,
      "learning_rate": 4.501975094204295e-05,
      "loss": 1.1220998764038086,
      "step": 5820
    },
    {
      "epoch": 0.6161406660765949,
      "grad_norm": 1.0413150787353516,
      "learning_rate": 4.500315682225796e-05,
      "loss": 1.2045147895812989,
      "step": 5830
    },
    {
      "epoch": 0.6171975111298995,
      "grad_norm": 1.205271601676941,
      "learning_rate": 4.498653817175914e-05,
      "loss": 1.1731483459472656,
      "step": 5840
    },
    {
      "epoch": 0.6182543561832041,
      "grad_norm": 0.9491797685623169,
      "learning_rate": 4.496989501092664e-05,
      "loss": 1.1173004150390624,
      "step": 5850
    },
    {
      "epoch": 0.6193112012365087,
      "grad_norm": 0.7863172292709351,
      "learning_rate": 4.495322736017066e-05,
      "loss": 1.0782630920410157,
      "step": 5860
    },
    {
      "epoch": 0.6203680462898133,
      "grad_norm": 0.8407509326934814,
      "learning_rate": 4.4936535239931455e-05,
      "loss": 1.0582219123840333,
      "step": 5870
    },
    {
      "epoch": 0.621424891343118,
      "grad_norm": 1.052092432975769,
      "learning_rate": 4.491981867067927e-05,
      "loss": 1.1607937812805176,
      "step": 5880
    },
    {
      "epoch": 0.6224817363964226,
      "grad_norm": 1.5181868076324463,
      "learning_rate": 4.490307767291433e-05,
      "loss": 1.0175371170043945,
      "step": 5890
    },
    {
      "epoch": 0.6235385814497272,
      "grad_norm": 0.8994033336639404,
      "learning_rate": 4.488631226716683e-05,
      "loss": 1.0660499572753905,
      "step": 5900
    },
    {
      "epoch": 0.6245954265030318,
      "grad_norm": 0.8771438002586365,
      "learning_rate": 4.486952247399691e-05,
      "loss": 1.0841008186340333,
      "step": 5910
    },
    {
      "epoch": 0.6256522715563364,
      "grad_norm": 1.1296882629394531,
      "learning_rate": 4.485270831399457e-05,
      "loss": 1.1017311096191407,
      "step": 5920
    },
    {
      "epoch": 0.626709116609641,
      "grad_norm": 0.7634866237640381,
      "learning_rate": 4.4835869807779744e-05,
      "loss": 1.1355999946594237,
      "step": 5930
    },
    {
      "epoch": 0.6277659616629457,
      "grad_norm": 1.2123538255691528,
      "learning_rate": 4.48190069760022e-05,
      "loss": 1.1335145950317382,
      "step": 5940
    },
    {
      "epoch": 0.6288228067162503,
      "grad_norm": 0.9872637391090393,
      "learning_rate": 4.480211983934153e-05,
      "loss": 1.1260403633117675,
      "step": 5950
    },
    {
      "epoch": 0.629879651769555,
      "grad_norm": 1.1882764101028442,
      "learning_rate": 4.478520841850715e-05,
      "loss": 1.1198975563049316,
      "step": 5960
    },
    {
      "epoch": 0.6309364968228596,
      "grad_norm": 0.9587523937225342,
      "learning_rate": 4.4768272734238245e-05,
      "loss": 1.1514243125915526,
      "step": 5970
    },
    {
      "epoch": 0.6319933418761642,
      "grad_norm": 0.8491719365119934,
      "learning_rate": 4.475131280730376e-05,
      "loss": 1.178119945526123,
      "step": 5980
    },
    {
      "epoch": 0.6330501869294688,
      "grad_norm": 0.9650416374206543,
      "learning_rate": 4.473432865850236e-05,
      "loss": 1.1947060585021974,
      "step": 5990
    },
    {
      "epoch": 0.6341070319827734,
      "grad_norm": 1.4546890258789062,
      "learning_rate": 4.4717320308662425e-05,
      "loss": 0.9415807723999023,
      "step": 6000
    },
    {
      "epoch": 0.6341070319827734,
      "eval_loss": 0.5338805317878723,
      "eval_runtime": 3430.3494,
      "eval_samples_per_second": 2.596,
      "eval_steps_per_second": 2.596,
      "step": 6000
    },
    {
      "epoch": 0.635163877036078,
      "grad_norm": 0.8711633086204529,
      "learning_rate": 4.470028777864202e-05,
      "loss": 1.097114658355713,
      "step": 6010
    },
    {
      "epoch": 0.6362207220893826,
      "grad_norm": 0.8503980040550232,
      "learning_rate": 4.4683231089328844e-05,
      "loss": 1.2322132110595703,
      "step": 6020
    },
    {
      "epoch": 0.6372775671426872,
      "grad_norm": 1.0085301399230957,
      "learning_rate": 4.466615026164024e-05,
      "loss": 1.1725217819213867,
      "step": 6030
    },
    {
      "epoch": 0.638334412195992,
      "grad_norm": 0.7401624917984009,
      "learning_rate": 4.4649045316523136e-05,
      "loss": 1.1364385604858398,
      "step": 6040
    },
    {
      "epoch": 0.6393912572492966,
      "grad_norm": 0.7629167437553406,
      "learning_rate": 4.4631916274954056e-05,
      "loss": 1.110755729675293,
      "step": 6050
    },
    {
      "epoch": 0.6404481023026012,
      "grad_norm": 1.1209931373596191,
      "learning_rate": 4.461476315793907e-05,
      "loss": 1.0200496673583985,
      "step": 6060
    },
    {
      "epoch": 0.6415049473559058,
      "grad_norm": 1.0301475524902344,
      "learning_rate": 4.4597585986513755e-05,
      "loss": 1.0359652519226075,
      "step": 6070
    },
    {
      "epoch": 0.6425617924092104,
      "grad_norm": 0.8953731656074524,
      "learning_rate": 4.45803847817432e-05,
      "loss": 1.0919235229492188,
      "step": 6080
    },
    {
      "epoch": 0.643618637462515,
      "grad_norm": 1.2993005514144897,
      "learning_rate": 4.4563159564721984e-05,
      "loss": 1.0049720764160157,
      "step": 6090
    },
    {
      "epoch": 0.6446754825158196,
      "grad_norm": 1.2030848264694214,
      "learning_rate": 4.45459103565741e-05,
      "loss": 1.243735122680664,
      "step": 6100
    },
    {
      "epoch": 0.6457323275691242,
      "grad_norm": 1.2289901971817017,
      "learning_rate": 4.452863717845299e-05,
      "loss": 1.1300676345825196,
      "step": 6110
    },
    {
      "epoch": 0.6467891726224289,
      "grad_norm": 1.112920880317688,
      "learning_rate": 4.451134005154146e-05,
      "loss": 1.0580347061157227,
      "step": 6120
    },
    {
      "epoch": 0.6478460176757335,
      "grad_norm": 0.7784405946731567,
      "learning_rate": 4.449401899705173e-05,
      "loss": 1.1944661140441895,
      "step": 6130
    },
    {
      "epoch": 0.6489028627290382,
      "grad_norm": 0.9042595028877258,
      "learning_rate": 4.447667403622533e-05,
      "loss": 1.2696884155273438,
      "step": 6140
    },
    {
      "epoch": 0.6499597077823428,
      "grad_norm": 0.8447825908660889,
      "learning_rate": 4.445930519033311e-05,
      "loss": 1.0895686149597168,
      "step": 6150
    },
    {
      "epoch": 0.6510165528356474,
      "grad_norm": 1.1558383703231812,
      "learning_rate": 4.444191248067522e-05,
      "loss": 1.075807762145996,
      "step": 6160
    },
    {
      "epoch": 0.652073397888952,
      "grad_norm": 0.691000759601593,
      "learning_rate": 4.442449592858108e-05,
      "loss": 1.107386875152588,
      "step": 6170
    },
    {
      "epoch": 0.6531302429422566,
      "grad_norm": 0.8304535746574402,
      "learning_rate": 4.440705555540933e-05,
      "loss": 1.106486988067627,
      "step": 6180
    },
    {
      "epoch": 0.6541870879955612,
      "grad_norm": 0.8903964161872864,
      "learning_rate": 4.4389591382547836e-05,
      "loss": 1.2043279647827148,
      "step": 6190
    },
    {
      "epoch": 0.6552439330488659,
      "grad_norm": 1.1067222356796265,
      "learning_rate": 4.4372103431413656e-05,
      "loss": 1.1944497108459473,
      "step": 6200
    },
    {
      "epoch": 0.6563007781021705,
      "grad_norm": 0.8262659311294556,
      "learning_rate": 4.4354591723453e-05,
      "loss": 1.1423243522644042,
      "step": 6210
    },
    {
      "epoch": 0.6573576231554751,
      "grad_norm": 0.8807287216186523,
      "learning_rate": 4.433705628014121e-05,
      "loss": 1.0372736930847168,
      "step": 6220
    },
    {
      "epoch": 0.6584144682087797,
      "grad_norm": 0.8673675060272217,
      "learning_rate": 4.431949712298273e-05,
      "loss": 1.0648754119873047,
      "step": 6230
    },
    {
      "epoch": 0.6594713132620844,
      "grad_norm": 1.0473541021347046,
      "learning_rate": 4.430191427351112e-05,
      "loss": 1.1144136428833007,
      "step": 6240
    },
    {
      "epoch": 0.660528158315389,
      "grad_norm": 1.1446458101272583,
      "learning_rate": 4.428430775328894e-05,
      "loss": 1.2294601440429687,
      "step": 6250
    },
    {
      "epoch": 0.6615850033686936,
      "grad_norm": 0.7924918532371521,
      "learning_rate": 4.426667758390783e-05,
      "loss": 1.1074429512023927,
      "step": 6260
    },
    {
      "epoch": 0.6626418484219982,
      "grad_norm": 0.8084589838981628,
      "learning_rate": 4.42490237869884e-05,
      "loss": 1.1318479537963868,
      "step": 6270
    },
    {
      "epoch": 0.6636986934753029,
      "grad_norm": 0.7664782404899597,
      "learning_rate": 4.423134638418024e-05,
      "loss": 1.1694219589233399,
      "step": 6280
    },
    {
      "epoch": 0.6647555385286075,
      "grad_norm": 1.0755641460418701,
      "learning_rate": 4.421364539716189e-05,
      "loss": 1.0428519248962402,
      "step": 6290
    },
    {
      "epoch": 0.6658123835819121,
      "grad_norm": 1.0269646644592285,
      "learning_rate": 4.4195920847640835e-05,
      "loss": 1.141301155090332,
      "step": 6300
    },
    {
      "epoch": 0.6668692286352167,
      "grad_norm": 0.898543119430542,
      "learning_rate": 4.417817275735342e-05,
      "loss": 0.9985285758972168,
      "step": 6310
    },
    {
      "epoch": 0.6679260736885213,
      "grad_norm": 1.0463346242904663,
      "learning_rate": 4.416040114806489e-05,
      "loss": 1.0666017532348633,
      "step": 6320
    },
    {
      "epoch": 0.6689829187418259,
      "grad_norm": 1.628617763519287,
      "learning_rate": 4.4142606041569304e-05,
      "loss": 1.1072714805603028,
      "step": 6330
    },
    {
      "epoch": 0.6700397637951306,
      "grad_norm": 0.8976156711578369,
      "learning_rate": 4.412478745968956e-05,
      "loss": 1.133560848236084,
      "step": 6340
    },
    {
      "epoch": 0.6710966088484352,
      "grad_norm": 1.1304771900177002,
      "learning_rate": 4.4106945424277325e-05,
      "loss": 1.1185452461242675,
      "step": 6350
    },
    {
      "epoch": 0.6721534539017399,
      "grad_norm": 0.9053941965103149,
      "learning_rate": 4.408907995721304e-05,
      "loss": 1.1405296325683594,
      "step": 6360
    },
    {
      "epoch": 0.6732102989550445,
      "grad_norm": 0.8668646216392517,
      "learning_rate": 4.40711910804059e-05,
      "loss": 1.1561245918273926,
      "step": 6370
    },
    {
      "epoch": 0.6742671440083491,
      "grad_norm": 1.106961727142334,
      "learning_rate": 4.405327881579376e-05,
      "loss": 1.1716703414916991,
      "step": 6380
    },
    {
      "epoch": 0.6753239890616537,
      "grad_norm": 1.4131066799163818,
      "learning_rate": 4.4035343185343216e-05,
      "loss": 1.0259252548217774,
      "step": 6390
    },
    {
      "epoch": 0.6763808341149583,
      "grad_norm": 0.8054501414299011,
      "learning_rate": 4.401738421104945e-05,
      "loss": 0.9930323600769043,
      "step": 6400
    },
    {
      "epoch": 0.6774376791682629,
      "grad_norm": 0.9600028395652771,
      "learning_rate": 4.399940191493634e-05,
      "loss": 1.2869458198547363,
      "step": 6410
    },
    {
      "epoch": 0.6784945242215675,
      "grad_norm": 1.2845534086227417,
      "learning_rate": 4.398139631905632e-05,
      "loss": 0.8795278549194336,
      "step": 6420
    },
    {
      "epoch": 0.6795513692748721,
      "grad_norm": 0.9547559022903442,
      "learning_rate": 4.396336744549041e-05,
      "loss": 1.0312155723571776,
      "step": 6430
    },
    {
      "epoch": 0.6806082143281769,
      "grad_norm": 1.0302668809890747,
      "learning_rate": 4.394531531634818e-05,
      "loss": 1.0363924980163575,
      "step": 6440
    },
    {
      "epoch": 0.6816650593814815,
      "grad_norm": 1.1465940475463867,
      "learning_rate": 4.3927239953767706e-05,
      "loss": 1.1217042922973632,
      "step": 6450
    },
    {
      "epoch": 0.6827219044347861,
      "grad_norm": 1.0454508066177368,
      "learning_rate": 4.390914137991558e-05,
      "loss": 1.138959789276123,
      "step": 6460
    },
    {
      "epoch": 0.6837787494880907,
      "grad_norm": 0.920134961605072,
      "learning_rate": 4.389101961698686e-05,
      "loss": 1.1324640274047852,
      "step": 6470
    },
    {
      "epoch": 0.6848355945413953,
      "grad_norm": 0.8516987562179565,
      "learning_rate": 4.387287468720499e-05,
      "loss": 1.1263035774230956,
      "step": 6480
    },
    {
      "epoch": 0.6858924395946999,
      "grad_norm": 1.1365128755569458,
      "learning_rate": 4.385470661282188e-05,
      "loss": 1.2055530548095703,
      "step": 6490
    },
    {
      "epoch": 0.6869492846480045,
      "grad_norm": 1.017565131187439,
      "learning_rate": 4.383651541611782e-05,
      "loss": 1.1704683303833008,
      "step": 6500
    },
    {
      "epoch": 0.6869492846480045,
      "eval_loss": 0.5247741937637329,
      "eval_runtime": 3436.2643,
      "eval_samples_per_second": 2.591,
      "eval_steps_per_second": 2.591,
      "step": 6500
    },
    {
      "epoch": 0.6880061297013091,
      "grad_norm": 0.9438244700431824,
      "learning_rate": 4.381830111940141e-05,
      "loss": 1.1812341690063477,
      "step": 6510
    },
    {
      "epoch": 0.6890629747546138,
      "grad_norm": 1.2594578266143799,
      "learning_rate": 4.380006374500963e-05,
      "loss": 1.1841961860656738,
      "step": 6520
    },
    {
      "epoch": 0.6901198198079184,
      "grad_norm": 0.7943575978279114,
      "learning_rate": 4.378180331530775e-05,
      "loss": 1.0180980682373046,
      "step": 6530
    },
    {
      "epoch": 0.6911766648612231,
      "grad_norm": 0.5472072958946228,
      "learning_rate": 4.376351985268927e-05,
      "loss": 1.0292016983032226,
      "step": 6540
    },
    {
      "epoch": 0.6922335099145277,
      "grad_norm": 0.9752936363220215,
      "learning_rate": 4.3745213379576e-05,
      "loss": 1.0895339012145997,
      "step": 6550
    },
    {
      "epoch": 0.6932903549678323,
      "grad_norm": 1.0425853729248047,
      "learning_rate": 4.3726883918417926e-05,
      "loss": 1.140283203125,
      "step": 6560
    },
    {
      "epoch": 0.6943472000211369,
      "grad_norm": 0.8568474650382996,
      "learning_rate": 4.370853149169325e-05,
      "loss": 1.0779850959777832,
      "step": 6570
    },
    {
      "epoch": 0.6954040450744415,
      "grad_norm": 0.8064638376235962,
      "learning_rate": 4.369015612190831e-05,
      "loss": 1.2570948600769043,
      "step": 6580
    },
    {
      "epoch": 0.6964608901277461,
      "grad_norm": 0.8313029408454895,
      "learning_rate": 4.36717578315976e-05,
      "loss": 1.0632568359375,
      "step": 6590
    },
    {
      "epoch": 0.6975177351810508,
      "grad_norm": 1.1827223300933838,
      "learning_rate": 4.365333664332374e-05,
      "loss": 1.0646686553955078,
      "step": 6600
    },
    {
      "epoch": 0.6985745802343554,
      "grad_norm": 0.8570693135261536,
      "learning_rate": 4.3634892579677375e-05,
      "loss": 1.2249155044555664,
      "step": 6610
    },
    {
      "epoch": 0.69963142528766,
      "grad_norm": 0.8176441192626953,
      "learning_rate": 4.361642566327726e-05,
      "loss": 1.1988866806030274,
      "step": 6620
    },
    {
      "epoch": 0.7006882703409646,
      "grad_norm": 0.7914356589317322,
      "learning_rate": 4.359793591677015e-05,
      "loss": 1.065091609954834,
      "step": 6630
    },
    {
      "epoch": 0.7017451153942692,
      "grad_norm": 0.9804995656013489,
      "learning_rate": 4.357942336283079e-05,
      "loss": 1.2422232627868652,
      "step": 6640
    },
    {
      "epoch": 0.7028019604475739,
      "grad_norm": 0.8100864887237549,
      "learning_rate": 4.356088802416192e-05,
      "loss": 1.0071266174316407,
      "step": 6650
    },
    {
      "epoch": 0.7038588055008785,
      "grad_norm": 0.7567852735519409,
      "learning_rate": 4.354232992349418e-05,
      "loss": 1.0189359664916993,
      "step": 6660
    },
    {
      "epoch": 0.7049156505541831,
      "grad_norm": 1.285056710243225,
      "learning_rate": 4.3523749083586186e-05,
      "loss": 1.1893707275390626,
      "step": 6670
    },
    {
      "epoch": 0.7059724956074878,
      "grad_norm": 0.8412622809410095,
      "learning_rate": 4.3505145527224377e-05,
      "loss": 1.1057232856750487,
      "step": 6680
    },
    {
      "epoch": 0.7070293406607924,
      "grad_norm": 0.8966756463050842,
      "learning_rate": 4.348651927722309e-05,
      "loss": 1.0770642280578613,
      "step": 6690
    },
    {
      "epoch": 0.708086185714097,
      "grad_norm": 1.0799423456192017,
      "learning_rate": 4.346787035642449e-05,
      "loss": 1.0263534545898438,
      "step": 6700
    },
    {
      "epoch": 0.7091430307674016,
      "grad_norm": 1.0979242324829102,
      "learning_rate": 4.344919878769851e-05,
      "loss": 1.1741092681884766,
      "step": 6710
    },
    {
      "epoch": 0.7101998758207062,
      "grad_norm": 1.375991702079773,
      "learning_rate": 4.343050459394289e-05,
      "loss": 1.1386422157287597,
      "step": 6720
    },
    {
      "epoch": 0.7112567208740108,
      "grad_norm": 1.015297532081604,
      "learning_rate": 4.341178779808312e-05,
      "loss": 0.9870831489562988,
      "step": 6730
    },
    {
      "epoch": 0.7123135659273154,
      "grad_norm": 0.8563395738601685,
      "learning_rate": 4.339304842307237e-05,
      "loss": 1.186475658416748,
      "step": 6740
    },
    {
      "epoch": 0.7133704109806202,
      "grad_norm": 0.9056344628334045,
      "learning_rate": 4.337428649189156e-05,
      "loss": 1.1514577865600586,
      "step": 6750
    },
    {
      "epoch": 0.7144272560339248,
      "grad_norm": 0.8676119446754456,
      "learning_rate": 4.3355502027549205e-05,
      "loss": 1.2056139945983886,
      "step": 6760
    },
    {
      "epoch": 0.7154841010872294,
      "grad_norm": 1.2787514925003052,
      "learning_rate": 4.33366950530815e-05,
      "loss": 1.0436160087585449,
      "step": 6770
    },
    {
      "epoch": 0.716540946140534,
      "grad_norm": 0.8991152048110962,
      "learning_rate": 4.331786559155222e-05,
      "loss": 1.2597204208374024,
      "step": 6780
    },
    {
      "epoch": 0.7175977911938386,
      "grad_norm": 1.0079143047332764,
      "learning_rate": 4.3299013666052746e-05,
      "loss": 1.1362088203430176,
      "step": 6790
    },
    {
      "epoch": 0.7186546362471432,
      "grad_norm": 0.8794000148773193,
      "learning_rate": 4.3280139299701974e-05,
      "loss": 1.1042369842529296,
      "step": 6800
    },
    {
      "epoch": 0.7197114813004478,
      "grad_norm": 0.9752110242843628,
      "learning_rate": 4.3261242515646336e-05,
      "loss": 1.1472264289855958,
      "step": 6810
    },
    {
      "epoch": 0.7207683263537524,
      "grad_norm": 0.8998944163322449,
      "learning_rate": 4.3242323337059754e-05,
      "loss": 1.1496851921081543,
      "step": 6820
    },
    {
      "epoch": 0.721825171407057,
      "grad_norm": 0.906976580619812,
      "learning_rate": 4.3223381787143616e-05,
      "loss": 1.1562995910644531,
      "step": 6830
    },
    {
      "epoch": 0.7228820164603617,
      "grad_norm": 0.8969152569770813,
      "learning_rate": 4.320441788912673e-05,
      "loss": 1.2247733116149901,
      "step": 6840
    },
    {
      "epoch": 0.7239388615136664,
      "grad_norm": 1.0110164880752563,
      "learning_rate": 4.3185431666265355e-05,
      "loss": 1.1216233253479004,
      "step": 6850
    },
    {
      "epoch": 0.724995706566971,
      "grad_norm": 1.2717304229736328,
      "learning_rate": 4.316642314184306e-05,
      "loss": 1.1373701095581055,
      "step": 6860
    },
    {
      "epoch": 0.7260525516202756,
      "grad_norm": 0.7993730902671814,
      "learning_rate": 4.314739233917082e-05,
      "loss": 1.1532154083251953,
      "step": 6870
    },
    {
      "epoch": 0.7271093966735802,
      "grad_norm": 0.8261692523956299,
      "learning_rate": 4.3128339281586906e-05,
      "loss": 1.2290363311767578,
      "step": 6880
    },
    {
      "epoch": 0.7281662417268848,
      "grad_norm": 0.6552752256393433,
      "learning_rate": 4.3109263992456886e-05,
      "loss": 1.0327116966247558,
      "step": 6890
    },
    {
      "epoch": 0.7292230867801894,
      "grad_norm": 1.040673017501831,
      "learning_rate": 4.309016649517358e-05,
      "loss": 1.0663652420043945,
      "step": 6900
    },
    {
      "epoch": 0.730279931833494,
      "grad_norm": 0.996379017829895,
      "learning_rate": 4.3071046813157065e-05,
      "loss": 1.2130916595458985,
      "step": 6910
    },
    {
      "epoch": 0.7313367768867987,
      "grad_norm": 1.015352725982666,
      "learning_rate": 4.3051904969854606e-05,
      "loss": 1.0224145889282226,
      "step": 6920
    },
    {
      "epoch": 0.7323936219401033,
      "grad_norm": 0.7139058113098145,
      "learning_rate": 4.303274098874066e-05,
      "loss": 1.070046615600586,
      "step": 6930
    },
    {
      "epoch": 0.7334504669934079,
      "grad_norm": 1.2343186140060425,
      "learning_rate": 4.301355489331682e-05,
      "loss": 1.105808639526367,
      "step": 6940
    },
    {
      "epoch": 0.7345073120467126,
      "grad_norm": 0.6168025135993958,
      "learning_rate": 4.29943467071118e-05,
      "loss": 0.9937611579895019,
      "step": 6950
    },
    {
      "epoch": 0.7355641571000172,
      "grad_norm": 0.8480864763259888,
      "learning_rate": 4.297511645368141e-05,
      "loss": 1.0430447578430175,
      "step": 6960
    },
    {
      "epoch": 0.7366210021533218,
      "grad_norm": 0.8377276062965393,
      "learning_rate": 4.295586415660852e-05,
      "loss": 1.0254338264465332,
      "step": 6970
    },
    {
      "epoch": 0.7376778472066264,
      "grad_norm": 1.0675474405288696,
      "learning_rate": 4.293658983950303e-05,
      "loss": 1.1308331489562988,
      "step": 6980
    },
    {
      "epoch": 0.738734692259931,
      "grad_norm": 1.3924247026443481,
      "learning_rate": 4.291729352600185e-05,
      "loss": 1.1824389457702638,
      "step": 6990
    },
    {
      "epoch": 0.7397915373132357,
      "grad_norm": 1.1285814046859741,
      "learning_rate": 4.2897975239768864e-05,
      "loss": 0.9056249618530273,
      "step": 7000
    },
    {
      "epoch": 0.7397915373132357,
      "eval_loss": 0.5214832425117493,
      "eval_runtime": 3432.0331,
      "eval_samples_per_second": 2.595,
      "eval_steps_per_second": 2.595,
      "step": 7000
    },
    {
      "epoch": 0.7408483823665403,
      "grad_norm": 1.0226309299468994,
      "learning_rate": 4.2878635004494894e-05,
      "loss": 1.079299259185791,
      "step": 7010
    },
    {
      "epoch": 0.7419052274198449,
      "grad_norm": 0.12381237000226974,
      "learning_rate": 4.2859272843897694e-05,
      "loss": 1.0043484687805175,
      "step": 7020
    },
    {
      "epoch": 0.7429620724731495,
      "grad_norm": 0.8023759126663208,
      "learning_rate": 4.283988878172189e-05,
      "loss": 1.0914041519165039,
      "step": 7030
    },
    {
      "epoch": 0.7440189175264541,
      "grad_norm": 1.179512858390808,
      "learning_rate": 4.2820482841738976e-05,
      "loss": 1.1604860305786133,
      "step": 7040
    },
    {
      "epoch": 0.7450757625797588,
      "grad_norm": 0.9011765122413635,
      "learning_rate": 4.280105504774728e-05,
      "loss": 1.0293027877807617,
      "step": 7050
    },
    {
      "epoch": 0.7461326076330634,
      "grad_norm": 1.3321304321289062,
      "learning_rate": 4.2781605423571914e-05,
      "loss": 1.1074341773986816,
      "step": 7060
    },
    {
      "epoch": 0.747189452686368,
      "grad_norm": 0.951073944568634,
      "learning_rate": 4.2762133993064774e-05,
      "loss": 1.126558208465576,
      "step": 7070
    },
    {
      "epoch": 0.7482462977396727,
      "grad_norm": 1.5734994411468506,
      "learning_rate": 4.27426407801045e-05,
      "loss": 1.1356448173522948,
      "step": 7080
    },
    {
      "epoch": 0.7493031427929773,
      "grad_norm": 0.8921034336090088,
      "learning_rate": 4.2723125808596435e-05,
      "loss": 1.1105058670043946,
      "step": 7090
    },
    {
      "epoch": 0.7503599878462819,
      "grad_norm": 0.838878870010376,
      "learning_rate": 4.2703589102472615e-05,
      "loss": 1.1888006210327149,
      "step": 7100
    },
    {
      "epoch": 0.7514168328995865,
      "grad_norm": 0.8467165231704712,
      "learning_rate": 4.268403068569172e-05,
      "loss": 1.0850698471069335,
      "step": 7110
    },
    {
      "epoch": 0.7524736779528911,
      "grad_norm": 1.6273999214172363,
      "learning_rate": 4.266445058223907e-05,
      "loss": 1.047002124786377,
      "step": 7120
    },
    {
      "epoch": 0.7535305230061957,
      "grad_norm": 0.9308730959892273,
      "learning_rate": 4.264484881612656e-05,
      "loss": 1.0627096176147461,
      "step": 7130
    },
    {
      "epoch": 0.7545873680595003,
      "grad_norm": 1.2751259803771973,
      "learning_rate": 4.262522541139267e-05,
      "loss": 1.055751132965088,
      "step": 7140
    },
    {
      "epoch": 0.755644213112805,
      "grad_norm": 1.2570147514343262,
      "learning_rate": 4.2605580392102396e-05,
      "loss": 1.205671787261963,
      "step": 7150
    },
    {
      "epoch": 0.7567010581661097,
      "grad_norm": 0.9577274918556213,
      "learning_rate": 4.258591378234727e-05,
      "loss": 0.9497488021850586,
      "step": 7160
    },
    {
      "epoch": 0.7577579032194143,
      "grad_norm": 1.2533161640167236,
      "learning_rate": 4.2566225606245266e-05,
      "loss": 1.1991342544555663,
      "step": 7170
    },
    {
      "epoch": 0.7588147482727189,
      "grad_norm": 0.8398496508598328,
      "learning_rate": 4.254651588794084e-05,
      "loss": 0.9820039749145508,
      "step": 7180
    },
    {
      "epoch": 0.7598715933260235,
      "grad_norm": 1.2236303091049194,
      "learning_rate": 4.252678465160484e-05,
      "loss": 1.010343360900879,
      "step": 7190
    },
    {
      "epoch": 0.7609284383793281,
      "grad_norm": 0.9582582712173462,
      "learning_rate": 4.2507031921434515e-05,
      "loss": 1.0799099922180175,
      "step": 7200
    },
    {
      "epoch": 0.7619852834326327,
      "grad_norm": 1.571415901184082,
      "learning_rate": 4.248725772165347e-05,
      "loss": 1.066250991821289,
      "step": 7210
    },
    {
      "epoch": 0.7630421284859373,
      "grad_norm": 0.8605715036392212,
      "learning_rate": 4.2467462076511635e-05,
      "loss": 1.0658190727233887,
      "step": 7220
    },
    {
      "epoch": 0.7640989735392419,
      "grad_norm": 0.8661869764328003,
      "learning_rate": 4.244764501028525e-05,
      "loss": 1.014609432220459,
      "step": 7230
    },
    {
      "epoch": 0.7651558185925466,
      "grad_norm": 1.2807822227478027,
      "learning_rate": 4.2427806547276814e-05,
      "loss": 0.9443817138671875,
      "step": 7240
    },
    {
      "epoch": 0.7662126636458513,
      "grad_norm": 0.8982623219490051,
      "learning_rate": 4.240794671181507e-05,
      "loss": 1.0235004425048828,
      "step": 7250
    },
    {
      "epoch": 0.7672695086991559,
      "grad_norm": 1.0451029539108276,
      "learning_rate": 4.238806552825498e-05,
      "loss": 1.127985382080078,
      "step": 7260
    },
    {
      "epoch": 0.7683263537524605,
      "grad_norm": 1.122089147567749,
      "learning_rate": 4.2368163020977656e-05,
      "loss": 1.0352426528930665,
      "step": 7270
    },
    {
      "epoch": 0.7693831988057651,
      "grad_norm": 1.2411754131317139,
      "learning_rate": 4.2348239214390404e-05,
      "loss": 1.089774227142334,
      "step": 7280
    },
    {
      "epoch": 0.7704400438590697,
      "grad_norm": 0.9467443823814392,
      "learning_rate": 4.232829413292662e-05,
      "loss": 1.104805850982666,
      "step": 7290
    },
    {
      "epoch": 0.7714968889123743,
      "grad_norm": 0.664135754108429,
      "learning_rate": 4.230832780104579e-05,
      "loss": 1.1570183753967285,
      "step": 7300
    },
    {
      "epoch": 0.772553733965679,
      "grad_norm": 0.7806284427642822,
      "learning_rate": 4.228834024323347e-05,
      "loss": 1.1445112228393555,
      "step": 7310
    },
    {
      "epoch": 0.7736105790189836,
      "grad_norm": 0.8855127692222595,
      "learning_rate": 4.226833148400125e-05,
      "loss": 1.0501784324645995,
      "step": 7320
    },
    {
      "epoch": 0.7746674240722882,
      "grad_norm": 0.8988419771194458,
      "learning_rate": 4.224830154788671e-05,
      "loss": 1.086318302154541,
      "step": 7330
    },
    {
      "epoch": 0.7757242691255928,
      "grad_norm": 1.4195624589920044,
      "learning_rate": 4.2228250459453405e-05,
      "loss": 1.1767480850219727,
      "step": 7340
    },
    {
      "epoch": 0.7767811141788974,
      "grad_norm": 0.9933862686157227,
      "learning_rate": 4.220817824329083e-05,
      "loss": 1.0876965522766113,
      "step": 7350
    },
    {
      "epoch": 0.7778379592322021,
      "grad_norm": 0.9280545711517334,
      "learning_rate": 4.218808492401439e-05,
      "loss": 1.1307670593261718,
      "step": 7360
    },
    {
      "epoch": 0.7788948042855067,
      "grad_norm": 1.1481149196624756,
      "learning_rate": 4.216797052626537e-05,
      "loss": 1.1006409645080566,
      "step": 7370
    },
    {
      "epoch": 0.7799516493388113,
      "grad_norm": 0.7118038535118103,
      "learning_rate": 4.21478350747109e-05,
      "loss": 1.0785511016845704,
      "step": 7380
    },
    {
      "epoch": 0.781008494392116,
      "grad_norm": 1.5307883024215698,
      "learning_rate": 4.212767859404393e-05,
      "loss": 1.0442439079284669,
      "step": 7390
    },
    {
      "epoch": 0.7820653394454206,
      "grad_norm": 0.8034446835517883,
      "learning_rate": 4.2107501108983215e-05,
      "loss": 1.1154428482055665,
      "step": 7400
    },
    {
      "epoch": 0.7831221844987252,
      "grad_norm": 0.8714629411697388,
      "learning_rate": 4.2087302644273234e-05,
      "loss": 1.0755393981933594,
      "step": 7410
    },
    {
      "epoch": 0.7841790295520298,
      "grad_norm": 1.5238633155822754,
      "learning_rate": 4.206708322468423e-05,
      "loss": 0.988521671295166,
      "step": 7420
    },
    {
      "epoch": 0.7852358746053344,
      "grad_norm": 1.36735999584198,
      "learning_rate": 4.2046842875012126e-05,
      "loss": 0.9650484085083008,
      "step": 7430
    },
    {
      "epoch": 0.786292719658639,
      "grad_norm": 1.1565594673156738,
      "learning_rate": 4.202658162007852e-05,
      "loss": 1.0494613647460938,
      "step": 7440
    },
    {
      "epoch": 0.7873495647119436,
      "grad_norm": 1.1131751537322998,
      "learning_rate": 4.2006299484730625e-05,
      "loss": 1.0382685661315918,
      "step": 7450
    },
    {
      "epoch": 0.7884064097652483,
      "grad_norm": 0.7280011773109436,
      "learning_rate": 4.198599649384129e-05,
      "loss": 1.141117000579834,
      "step": 7460
    },
    {
      "epoch": 0.789463254818553,
      "grad_norm": 1.357185959815979,
      "learning_rate": 4.196567267230894e-05,
      "loss": 1.1405117988586426,
      "step": 7470
    },
    {
      "epoch": 0.7905200998718576,
      "grad_norm": 0.9919775724411011,
      "learning_rate": 4.194532804505752e-05,
      "loss": 1.0713905334472655,
      "step": 7480
    },
    {
      "epoch": 0.7915769449251622,
      "grad_norm": 1.167651891708374,
      "learning_rate": 4.19249626370365e-05,
      "loss": 1.1758601188659668,
      "step": 7490
    },
    {
      "epoch": 0.7926337899784668,
      "grad_norm": 0.7991864085197449,
      "learning_rate": 4.1904576473220834e-05,
      "loss": 1.0790603637695313,
      "step": 7500
    },
    {
      "epoch": 0.7926337899784668,
      "eval_loss": 0.5181804895401001,
      "eval_runtime": 3432.7568,
      "eval_samples_per_second": 2.594,
      "eval_steps_per_second": 2.594,
      "step": 7500
    },
    {
      "epoch": 0.7936906350317714,
      "grad_norm": 1.0892444849014282,
      "learning_rate": 4.1884169578610954e-05,
      "loss": 1.016141986846924,
      "step": 7510
    },
    {
      "epoch": 0.794747480085076,
      "grad_norm": 0.8035133481025696,
      "learning_rate": 4.186374197823268e-05,
      "loss": 1.0943964004516602,
      "step": 7520
    },
    {
      "epoch": 0.7958043251383806,
      "grad_norm": 0.9514089822769165,
      "learning_rate": 4.184329369713724e-05,
      "loss": 1.194356632232666,
      "step": 7530
    },
    {
      "epoch": 0.7968611701916852,
      "grad_norm": 0.8046138286590576,
      "learning_rate": 4.1822824760401224e-05,
      "loss": 1.0784614562988282,
      "step": 7540
    },
    {
      "epoch": 0.7979180152449898,
      "grad_norm": 1.1638352870941162,
      "learning_rate": 4.1802335193126555e-05,
      "loss": 1.1141912460327148,
      "step": 7550
    },
    {
      "epoch": 0.7989748602982946,
      "grad_norm": 0.8349569439888,
      "learning_rate": 4.178182502044043e-05,
      "loss": 1.0444686889648438,
      "step": 7560
    },
    {
      "epoch": 0.8000317053515992,
      "grad_norm": 0.8563461899757385,
      "learning_rate": 4.1761294267495376e-05,
      "loss": 0.9330798149108886,
      "step": 7570
    },
    {
      "epoch": 0.8010885504049038,
      "grad_norm": 1.4030108451843262,
      "learning_rate": 4.1740742959469103e-05,
      "loss": 0.991219425201416,
      "step": 7580
    },
    {
      "epoch": 0.8021453954582084,
      "grad_norm": 1.0926196575164795,
      "learning_rate": 4.172017112156453e-05,
      "loss": 1.0880531311035155,
      "step": 7590
    },
    {
      "epoch": 0.803202240511513,
      "grad_norm": 1.0161170959472656,
      "learning_rate": 4.169957877900979e-05,
      "loss": 1.1059306144714356,
      "step": 7600
    },
    {
      "epoch": 0.8042590855648176,
      "grad_norm": 0.8941255807876587,
      "learning_rate": 4.1678965957058125e-05,
      "loss": 1.094955062866211,
      "step": 7610
    },
    {
      "epoch": 0.8053159306181222,
      "grad_norm": 1.0602444410324097,
      "learning_rate": 4.165833268098792e-05,
      "loss": 1.02705659866333,
      "step": 7620
    },
    {
      "epoch": 0.8063727756714268,
      "grad_norm": 1.092079520225525,
      "learning_rate": 4.1637678976102624e-05,
      "loss": 1.1772047996520996,
      "step": 7630
    },
    {
      "epoch": 0.8074296207247315,
      "grad_norm": 1.3852494955062866,
      "learning_rate": 4.1617004867730735e-05,
      "loss": 1.0798553466796874,
      "step": 7640
    },
    {
      "epoch": 0.8084864657780361,
      "grad_norm": 0.7665895819664001,
      "learning_rate": 4.159631038122581e-05,
      "loss": 1.078364944458008,
      "step": 7650
    },
    {
      "epoch": 0.8095433108313408,
      "grad_norm": 0.9126380681991577,
      "learning_rate": 4.157559554196634e-05,
      "loss": 0.9516583442687988,
      "step": 7660
    },
    {
      "epoch": 0.8106001558846454,
      "grad_norm": 1.0580495595932007,
      "learning_rate": 4.155486037535582e-05,
      "loss": 1.1213530540466308,
      "step": 7670
    },
    {
      "epoch": 0.81165700093795,
      "grad_norm": 0.8816622495651245,
      "learning_rate": 4.1534104906822655e-05,
      "loss": 1.0890243530273438,
      "step": 7680
    },
    {
      "epoch": 0.8127138459912546,
      "grad_norm": 1.2742843627929688,
      "learning_rate": 4.1513329161820156e-05,
      "loss": 1.0329085350036622,
      "step": 7690
    },
    {
      "epoch": 0.8137706910445592,
      "grad_norm": 0.8025449514389038,
      "learning_rate": 4.14925331658265e-05,
      "loss": 1.023833656311035,
      "step": 7700
    },
    {
      "epoch": 0.8148275360978638,
      "grad_norm": 1.2994377613067627,
      "learning_rate": 4.147171694434466e-05,
      "loss": 1.0763998985290528,
      "step": 7710
    },
    {
      "epoch": 0.8158843811511685,
      "grad_norm": 1.0011236667633057,
      "learning_rate": 4.145088052290249e-05,
      "loss": 1.0904871940612793,
      "step": 7720
    },
    {
      "epoch": 0.8169412262044731,
      "grad_norm": 1.0946338176727295,
      "learning_rate": 4.1430023927052534e-05,
      "loss": 1.0079373359680175,
      "step": 7730
    },
    {
      "epoch": 0.8179980712577777,
      "grad_norm": 0.8024388551712036,
      "learning_rate": 4.1409147182372135e-05,
      "loss": 1.1211822509765625,
      "step": 7740
    },
    {
      "epoch": 0.8190549163110823,
      "grad_norm": 1.397788166999817,
      "learning_rate": 4.138825031446332e-05,
      "loss": 1.1016294479370117,
      "step": 7750
    },
    {
      "epoch": 0.820111761364387,
      "grad_norm": 1.0234079360961914,
      "learning_rate": 4.13673333489528e-05,
      "loss": 1.1285269737243653,
      "step": 7760
    },
    {
      "epoch": 0.8211686064176916,
      "grad_norm": 0.9909486174583435,
      "learning_rate": 4.1346396311491925e-05,
      "loss": 1.2218446731567383,
      "step": 7770
    },
    {
      "epoch": 0.8222254514709962,
      "grad_norm": 1.1322871446609497,
      "learning_rate": 4.1325439227756666e-05,
      "loss": 1.0101380348205566,
      "step": 7780
    },
    {
      "epoch": 0.8232822965243008,
      "grad_norm": 0.8919448256492615,
      "learning_rate": 4.1304462123447585e-05,
      "loss": 1.0671889305114746,
      "step": 7790
    },
    {
      "epoch": 0.8243391415776055,
      "grad_norm": 1.5752556324005127,
      "learning_rate": 4.1283465024289784e-05,
      "loss": 1.12045316696167,
      "step": 7800
    },
    {
      "epoch": 0.8253959866309101,
      "grad_norm": 0.7265978455543518,
      "learning_rate": 4.126244795603289e-05,
      "loss": 1.1254706382751465,
      "step": 7810
    },
    {
      "epoch": 0.8264528316842147,
      "grad_norm": 1.4501997232437134,
      "learning_rate": 4.124141094445102e-05,
      "loss": 1.1314648628234862,
      "step": 7820
    },
    {
      "epoch": 0.8275096767375193,
      "grad_norm": 1.2828516960144043,
      "learning_rate": 4.122035401534275e-05,
      "loss": 0.9155103683471679,
      "step": 7830
    },
    {
      "epoch": 0.8285665217908239,
      "grad_norm": 1.0397356748580933,
      "learning_rate": 4.119927719453107e-05,
      "loss": 1.2216338157653808,
      "step": 7840
    },
    {
      "epoch": 0.8296233668441285,
      "grad_norm": 0.9135522246360779,
      "learning_rate": 4.117818050786338e-05,
      "loss": 1.1568754196166993,
      "step": 7850
    },
    {
      "epoch": 0.8306802118974332,
      "grad_norm": 1.005700707435608,
      "learning_rate": 4.115706398121145e-05,
      "loss": 1.0420746803283691,
      "step": 7860
    },
    {
      "epoch": 0.8317370569507379,
      "grad_norm": 1.1724299192428589,
      "learning_rate": 4.113592764047134e-05,
      "loss": 1.2019783973693847,
      "step": 7870
    },
    {
      "epoch": 0.8327939020040425,
      "grad_norm": 1.2931668758392334,
      "learning_rate": 4.111477151156344e-05,
      "loss": 1.0428693771362305,
      "step": 7880
    },
    {
      "epoch": 0.8338507470573471,
      "grad_norm": 1.2621325254440308,
      "learning_rate": 4.1093595620432425e-05,
      "loss": 0.9249512672424316,
      "step": 7890
    },
    {
      "epoch": 0.8349075921106517,
      "grad_norm": 0.9824300408363342,
      "learning_rate": 4.107239999304715e-05,
      "loss": 1.2201066970825196,
      "step": 7900
    },
    {
      "epoch": 0.8359644371639563,
      "grad_norm": 1.045862078666687,
      "learning_rate": 4.105118465540073e-05,
      "loss": 1.108299446105957,
      "step": 7910
    },
    {
      "epoch": 0.8370212822172609,
      "grad_norm": 0.8786892294883728,
      "learning_rate": 4.1029949633510414e-05,
      "loss": 1.1539855003356934,
      "step": 7920
    },
    {
      "epoch": 0.8380781272705655,
      "grad_norm": 0.8637773394584656,
      "learning_rate": 4.100869495341761e-05,
      "loss": 1.0508352279663087,
      "step": 7930
    },
    {
      "epoch": 0.8391349723238701,
      "grad_norm": 1.0928964614868164,
      "learning_rate": 4.0987420641187835e-05,
      "loss": 0.994393253326416,
      "step": 7940
    },
    {
      "epoch": 0.8401918173771747,
      "grad_norm": 1.3324658870697021,
      "learning_rate": 4.0966126722910655e-05,
      "loss": 1.0687718391418457,
      "step": 7950
    },
    {
      "epoch": 0.8412486624304795,
      "grad_norm": 1.595292329788208,
      "learning_rate": 4.094481322469971e-05,
      "loss": 1.1032757759094238,
      "step": 7960
    },
    {
      "epoch": 0.8423055074837841,
      "grad_norm": 0.9195526242256165,
      "learning_rate": 4.0923480172692644e-05,
      "loss": 1.090679931640625,
      "step": 7970
    },
    {
      "epoch": 0.8433623525370887,
      "grad_norm": 1.0295969247817993,
      "learning_rate": 4.090212759305108e-05,
      "loss": 1.1312494277954102,
      "step": 7980
    },
    {
      "epoch": 0.8444191975903933,
      "grad_norm": 0.9847151041030884,
      "learning_rate": 4.0880755511960564e-05,
      "loss": 0.9561223983764648,
      "step": 7990
    },
    {
      "epoch": 0.8454760426436979,
      "grad_norm": 1.1761631965637207,
      "learning_rate": 4.0859363955630604e-05,
      "loss": 1.0081912994384765,
      "step": 8000
    },
    {
      "epoch": 0.8454760426436979,
      "eval_loss": 0.5137072801589966,
      "eval_runtime": 2985.9396,
      "eval_samples_per_second": 2.982,
      "eval_steps_per_second": 2.982,
      "step": 8000
    },
    {
      "epoch": 0.8465328876970025,
      "grad_norm": 0.9484139680862427,
      "learning_rate": 4.083795295029456e-05,
      "loss": 1.1009661674499511,
      "step": 8010
    },
    {
      "epoch": 0.8475897327503071,
      "grad_norm": 1.1517213582992554,
      "learning_rate": 4.081652252220964e-05,
      "loss": 1.0229705810546874,
      "step": 8020
    },
    {
      "epoch": 0.8486465778036117,
      "grad_norm": 1.1358071565628052,
      "learning_rate": 4.0795072697656886e-05,
      "loss": 1.1382295608520507,
      "step": 8030
    },
    {
      "epoch": 0.8497034228569164,
      "grad_norm": 0.9800382852554321,
      "learning_rate": 4.077360350294113e-05,
      "loss": 0.9064143180847168,
      "step": 8040
    },
    {
      "epoch": 0.850760267910221,
      "grad_norm": 1.2614620923995972,
      "learning_rate": 4.0752114964390945e-05,
      "loss": 1.080003833770752,
      "step": 8050
    },
    {
      "epoch": 0.8518171129635256,
      "grad_norm": 0.6888795495033264,
      "learning_rate": 4.073060710835862e-05,
      "loss": 0.9053874969482422,
      "step": 8060
    },
    {
      "epoch": 0.8528739580168303,
      "grad_norm": 1.295384407043457,
      "learning_rate": 4.070907996122016e-05,
      "loss": 1.12921199798584,
      "step": 8070
    },
    {
      "epoch": 0.8539308030701349,
      "grad_norm": 0.8068138360977173,
      "learning_rate": 4.06875335493752e-05,
      "loss": 1.0980145454406738,
      "step": 8080
    },
    {
      "epoch": 0.8549876481234395,
      "grad_norm": 1.0707600116729736,
      "learning_rate": 4.066596789924703e-05,
      "loss": 1.1173582077026367,
      "step": 8090
    },
    {
      "epoch": 0.8560444931767441,
      "grad_norm": 0.8965756297111511,
      "learning_rate": 4.0644383037282494e-05,
      "loss": 1.087883186340332,
      "step": 8100
    },
    {
      "epoch": 0.8571013382300487,
      "grad_norm": 1.1393240690231323,
      "learning_rate": 4.0622778989952024e-05,
      "loss": 1.1153321266174316,
      "step": 8110
    },
    {
      "epoch": 0.8581581832833534,
      "grad_norm": 1.286033272743225,
      "learning_rate": 4.060115578374959e-05,
      "loss": 1.1476593017578125,
      "step": 8120
    },
    {
      "epoch": 0.859215028336658,
      "grad_norm": 1.1083030700683594,
      "learning_rate": 4.0579513445192625e-05,
      "loss": 1.0944173812866211,
      "step": 8130
    },
    {
      "epoch": 0.8602718733899626,
      "grad_norm": 0.8625525832176208,
      "learning_rate": 4.055785200082204e-05,
      "loss": 1.099620532989502,
      "step": 8140
    },
    {
      "epoch": 0.8613287184432672,
      "grad_norm": 1.162726640701294,
      "learning_rate": 4.053617147720218e-05,
      "loss": 1.0889933586120606,
      "step": 8150
    },
    {
      "epoch": 0.8623855634965718,
      "grad_norm": 0.8006690740585327,
      "learning_rate": 4.0514471900920784e-05,
      "loss": 1.1881935119628906,
      "step": 8160
    },
    {
      "epoch": 0.8634424085498765,
      "grad_norm": 1.0894590616226196,
      "learning_rate": 4.049275329858895e-05,
      "loss": 1.020105743408203,
      "step": 8170
    },
    {
      "epoch": 0.8644992536031811,
      "grad_norm": 0.8492417931556702,
      "learning_rate": 4.0471015696841124e-05,
      "loss": 1.036696434020996,
      "step": 8180
    },
    {
      "epoch": 0.8655560986564858,
      "grad_norm": 0.9585519433021545,
      "learning_rate": 4.0449259122335044e-05,
      "loss": 1.0572170257568358,
      "step": 8190
    },
    {
      "epoch": 0.8666129437097904,
      "grad_norm": 0.8007840514183044,
      "learning_rate": 4.042748360175171e-05,
      "loss": 1.0378215789794922,
      "step": 8200
    },
    {
      "epoch": 0.867669788763095,
      "grad_norm": 1.0411627292633057,
      "learning_rate": 4.0405689161795356e-05,
      "loss": 1.087393856048584,
      "step": 8210
    },
    {
      "epoch": 0.8687266338163996,
      "grad_norm": 0.9710835814476013,
      "learning_rate": 4.038387582919343e-05,
      "loss": 1.1860026359558105,
      "step": 8220
    },
    {
      "epoch": 0.8697834788697042,
      "grad_norm": 0.863376259803772,
      "learning_rate": 4.036204363069653e-05,
      "loss": 1.2691640853881836,
      "step": 8230
    },
    {
      "epoch": 0.8708403239230088,
      "grad_norm": 0.8055329322814941,
      "learning_rate": 4.034019259307841e-05,
      "loss": 1.1820324897766112,
      "step": 8240
    },
    {
      "epoch": 0.8718971689763134,
      "grad_norm": 0.7863889932632446,
      "learning_rate": 4.0318322743135916e-05,
      "loss": 1.0700841903686524,
      "step": 8250
    },
    {
      "epoch": 0.872954014029618,
      "grad_norm": 0.9515518546104431,
      "learning_rate": 4.0296434107688966e-05,
      "loss": 0.977074146270752,
      "step": 8260
    },
    {
      "epoch": 0.8740108590829228,
      "grad_norm": 0.8248909711837769,
      "learning_rate": 4.0274526713580514e-05,
      "loss": 0.9980775833129882,
      "step": 8270
    },
    {
      "epoch": 0.8750677041362274,
      "grad_norm": 0.8874614834785461,
      "learning_rate": 4.025260058767652e-05,
      "loss": 1.1334486961364747,
      "step": 8280
    },
    {
      "epoch": 0.876124549189532,
      "grad_norm": 1.0823845863342285,
      "learning_rate": 4.023065575686592e-05,
      "loss": 1.0578557968139648,
      "step": 8290
    },
    {
      "epoch": 0.8771813942428366,
      "grad_norm": 1.2580658197402954,
      "learning_rate": 4.0208692248060576e-05,
      "loss": 0.9791107177734375,
      "step": 8300
    },
    {
      "epoch": 0.8782382392961412,
      "grad_norm": 0.9751501083374023,
      "learning_rate": 4.018671008819528e-05,
      "loss": 1.1559246063232422,
      "step": 8310
    },
    {
      "epoch": 0.8792950843494458,
      "grad_norm": 1.1333757638931274,
      "learning_rate": 4.016470930422766e-05,
      "loss": 1.1454938888549804,
      "step": 8320
    },
    {
      "epoch": 0.8803519294027504,
      "grad_norm": 0.8927004337310791,
      "learning_rate": 4.014268992313822e-05,
      "loss": 1.0712488174438477,
      "step": 8330
    },
    {
      "epoch": 0.881408774456055,
      "grad_norm": 1.0514440536499023,
      "learning_rate": 4.012065197193025e-05,
      "loss": 1.062297534942627,
      "step": 8340
    },
    {
      "epoch": 0.8824656195093596,
      "grad_norm": 1.2384426593780518,
      "learning_rate": 4.0098595477629806e-05,
      "loss": 0.979213809967041,
      "step": 8350
    },
    {
      "epoch": 0.8835224645626643,
      "grad_norm": 1.1735602617263794,
      "learning_rate": 4.007652046728572e-05,
      "loss": 1.1187929153442382,
      "step": 8360
    },
    {
      "epoch": 0.884579309615969,
      "grad_norm": 0.8846344351768494,
      "learning_rate": 4.0054426967969484e-05,
      "loss": 0.9032733917236329,
      "step": 8370
    },
    {
      "epoch": 0.8856361546692736,
      "grad_norm": 1.162982702255249,
      "learning_rate": 4.00323150067753e-05,
      "loss": 0.9907724380493164,
      "step": 8380
    },
    {
      "epoch": 0.8866929997225782,
      "grad_norm": 0.8608705401420593,
      "learning_rate": 4.0010184610819997e-05,
      "loss": 1.0868520736694336,
      "step": 8390
    },
    {
      "epoch": 0.8877498447758828,
      "grad_norm": 1.1657034158706665,
      "learning_rate": 3.9988035807243005e-05,
      "loss": 1.1400092124938965,
      "step": 8400
    },
    {
      "epoch": 0.8888066898291874,
      "grad_norm": 0.8754146695137024,
      "learning_rate": 3.996586862320634e-05,
      "loss": 1.0789137840270997,
      "step": 8410
    },
    {
      "epoch": 0.889863534882492,
      "grad_norm": 0.8743398189544678,
      "learning_rate": 3.994368308589456e-05,
      "loss": 1.1714950561523438,
      "step": 8420
    },
    {
      "epoch": 0.8909203799357966,
      "grad_norm": 1.1960371732711792,
      "learning_rate": 3.9921479222514705e-05,
      "loss": 1.0954792022705078,
      "step": 8430
    },
    {
      "epoch": 0.8919772249891013,
      "grad_norm": 1.0881860256195068,
      "learning_rate": 3.989925706029633e-05,
      "loss": 1.0817713737487793,
      "step": 8440
    },
    {
      "epoch": 0.8930340700424059,
      "grad_norm": 1.3883861303329468,
      "learning_rate": 3.9877016626491394e-05,
      "loss": 0.9834948539733886,
      "step": 8450
    },
    {
      "epoch": 0.8940909150957105,
      "grad_norm": 0.5159029364585876,
      "learning_rate": 3.985475794837428e-05,
      "loss": 0.9249711990356445,
      "step": 8460
    },
    {
      "epoch": 0.8951477601490152,
      "grad_norm": 1.1592384576797485,
      "learning_rate": 3.983248105324174e-05,
      "loss": 1.0571823120117188,
      "step": 8470
    },
    {
      "epoch": 0.8962046052023198,
      "grad_norm": 0.9482786059379578,
      "learning_rate": 3.981018596841288e-05,
      "loss": 1.0076674461364745,
      "step": 8480
    },
    {
      "epoch": 0.8972614502556244,
      "grad_norm": 1.0211185216903687,
      "learning_rate": 3.978787272122909e-05,
      "loss": 1.1418996810913087,
      "step": 8490
    },
    {
      "epoch": 0.898318295308929,
      "grad_norm": 1.1098864078521729,
      "learning_rate": 3.976554133905406e-05,
      "loss": 1.1300287246704102,
      "step": 8500
    },
    {
      "epoch": 0.898318295308929,
      "eval_loss": 0.512334942817688,
      "eval_runtime": 2987.7646,
      "eval_samples_per_second": 2.98,
      "eval_steps_per_second": 2.98,
      "step": 8500
    },
    {
      "epoch": 0.8993751403622336,
      "grad_norm": 0.9195661544799805,
      "learning_rate": 3.9743191849273686e-05,
      "loss": 0.9967521667480469,
      "step": 8510
    },
    {
      "epoch": 0.9004319854155383,
      "grad_norm": 0.7842029929161072,
      "learning_rate": 3.97208242792961e-05,
      "loss": 1.1253826141357421,
      "step": 8520
    },
    {
      "epoch": 0.9014888304688429,
      "grad_norm": 1.0699045658111572,
      "learning_rate": 3.96984386565516e-05,
      "loss": 1.078743839263916,
      "step": 8530
    },
    {
      "epoch": 0.9025456755221475,
      "grad_norm": 1.0729608535766602,
      "learning_rate": 3.9676035008492596e-05,
      "loss": 1.0769296646118165,
      "step": 8540
    },
    {
      "epoch": 0.9036025205754521,
      "grad_norm": 1.1123836040496826,
      "learning_rate": 3.965361336259365e-05,
      "loss": 1.098536491394043,
      "step": 8550
    },
    {
      "epoch": 0.9046593656287567,
      "grad_norm": 1.1240301132202148,
      "learning_rate": 3.963117374635136e-05,
      "loss": 1.0382189750671387,
      "step": 8560
    },
    {
      "epoch": 0.9057162106820614,
      "grad_norm": 0.8350022435188293,
      "learning_rate": 3.960871618728437e-05,
      "loss": 1.1752694129943848,
      "step": 8570
    },
    {
      "epoch": 0.906773055735366,
      "grad_norm": 0.9101041555404663,
      "learning_rate": 3.958624071293334e-05,
      "loss": 0.9887772560119629,
      "step": 8580
    },
    {
      "epoch": 0.9078299007886707,
      "grad_norm": 0.9899210333824158,
      "learning_rate": 3.956374735086088e-05,
      "loss": 1.002309513092041,
      "step": 8590
    },
    {
      "epoch": 0.9088867458419753,
      "grad_norm": 0.9482787251472473,
      "learning_rate": 3.954123612865156e-05,
      "loss": 1.115668487548828,
      "step": 8600
    },
    {
      "epoch": 0.9099435908952799,
      "grad_norm": 1.2325087785720825,
      "learning_rate": 3.951870707391183e-05,
      "loss": 1.1084575653076172,
      "step": 8610
    },
    {
      "epoch": 0.9110004359485845,
      "grad_norm": 1.1645845174789429,
      "learning_rate": 3.9496160214270015e-05,
      "loss": 0.9624773979187011,
      "step": 8620
    },
    {
      "epoch": 0.9120572810018891,
      "grad_norm": 0.9986498355865479,
      "learning_rate": 3.9473595577376296e-05,
      "loss": 1.1178292274475097,
      "step": 8630
    },
    {
      "epoch": 0.9131141260551937,
      "grad_norm": 0.978849470615387,
      "learning_rate": 3.945101319090262e-05,
      "loss": 1.1829820632934571,
      "step": 8640
    },
    {
      "epoch": 0.9141709711084983,
      "grad_norm": 0.9912046194076538,
      "learning_rate": 3.942841308254273e-05,
      "loss": 0.9736811637878418,
      "step": 8650
    },
    {
      "epoch": 0.9152278161618029,
      "grad_norm": 1.0539956092834473,
      "learning_rate": 3.9405795280012085e-05,
      "loss": 1.009325122833252,
      "step": 8660
    },
    {
      "epoch": 0.9162846612151077,
      "grad_norm": 1.341686487197876,
      "learning_rate": 3.938315981104785e-05,
      "loss": 1.2018801689147949,
      "step": 8670
    },
    {
      "epoch": 0.9173415062684123,
      "grad_norm": 0.2022629678249359,
      "learning_rate": 3.936050670340886e-05,
      "loss": 1.038057804107666,
      "step": 8680
    },
    {
      "epoch": 0.9183983513217169,
      "grad_norm": 1.0729854106903076,
      "learning_rate": 3.933783598487557e-05,
      "loss": 1.0424334526062011,
      "step": 8690
    },
    {
      "epoch": 0.9194551963750215,
      "grad_norm": 1.2410304546356201,
      "learning_rate": 3.9315147683250034e-05,
      "loss": 1.1475072860717774,
      "step": 8700
    },
    {
      "epoch": 0.9205120414283261,
      "grad_norm": 1.3946861028671265,
      "learning_rate": 3.929244182635587e-05,
      "loss": 1.0967617988586427,
      "step": 8710
    },
    {
      "epoch": 0.9215688864816307,
      "grad_norm": 1.3649346828460693,
      "learning_rate": 3.926971844203825e-05,
      "loss": 1.085676097869873,
      "step": 8720
    },
    {
      "epoch": 0.9226257315349353,
      "grad_norm": 1.028193712234497,
      "learning_rate": 3.924697755816379e-05,
      "loss": 1.044766902923584,
      "step": 8730
    },
    {
      "epoch": 0.9236825765882399,
      "grad_norm": 0.9187294840812683,
      "learning_rate": 3.9224219202620596e-05,
      "loss": 1.1421188354492187,
      "step": 8740
    },
    {
      "epoch": 0.9247394216415445,
      "grad_norm": 0.8156089782714844,
      "learning_rate": 3.920144340331821e-05,
      "loss": 1.041112232208252,
      "step": 8750
    },
    {
      "epoch": 0.9257962666948492,
      "grad_norm": 1.3749723434448242,
      "learning_rate": 3.917865018818755e-05,
      "loss": 0.9761978149414062,
      "step": 8760
    },
    {
      "epoch": 0.9268531117481538,
      "grad_norm": 1.0774803161621094,
      "learning_rate": 3.91558395851809e-05,
      "loss": 1.0752731323242188,
      "step": 8770
    },
    {
      "epoch": 0.9279099568014585,
      "grad_norm": 0.8063593506813049,
      "learning_rate": 3.9133011622271856e-05,
      "loss": 1.1302793502807618,
      "step": 8780
    },
    {
      "epoch": 0.9289668018547631,
      "grad_norm": 1.0337375402450562,
      "learning_rate": 3.911016632745531e-05,
      "loss": 1.0675479888916015,
      "step": 8790
    },
    {
      "epoch": 0.9300236469080677,
      "grad_norm": 0.9437271356582642,
      "learning_rate": 3.9087303728747414e-05,
      "loss": 1.0258854866027831,
      "step": 8800
    },
    {
      "epoch": 0.9310804919613723,
      "grad_norm": 0.9543331861495972,
      "learning_rate": 3.906442385418554e-05,
      "loss": 1.1828482627868653,
      "step": 8810
    },
    {
      "epoch": 0.9321373370146769,
      "grad_norm": 1.1384161710739136,
      "learning_rate": 3.904152673182823e-05,
      "loss": 1.1456493377685546,
      "step": 8820
    },
    {
      "epoch": 0.9331941820679815,
      "grad_norm": 0.9374061226844788,
      "learning_rate": 3.90186123897552e-05,
      "loss": 1.0885502815246582,
      "step": 8830
    },
    {
      "epoch": 0.9342510271212862,
      "grad_norm": 0.7394371032714844,
      "learning_rate": 3.8995680856067274e-05,
      "loss": 0.8799606323242187,
      "step": 8840
    },
    {
      "epoch": 0.9353078721745908,
      "grad_norm": 0.9894667267799377,
      "learning_rate": 3.897273215888635e-05,
      "loss": 1.0161099433898926,
      "step": 8850
    },
    {
      "epoch": 0.9363647172278954,
      "grad_norm": 1.2397270202636719,
      "learning_rate": 3.8949766326355375e-05,
      "loss": 1.0680577278137207,
      "step": 8860
    },
    {
      "epoch": 0.9374215622812,
      "grad_norm": 1.0823280811309814,
      "learning_rate": 3.892678338663834e-05,
      "loss": 0.9951482772827148,
      "step": 8870
    },
    {
      "epoch": 0.9384784073345047,
      "grad_norm": 1.138116717338562,
      "learning_rate": 3.890378336792017e-05,
      "loss": 0.9961891174316406,
      "step": 8880
    },
    {
      "epoch": 0.9395352523878093,
      "grad_norm": 0.8195068836212158,
      "learning_rate": 3.888076629840677e-05,
      "loss": 1.0693440437316895,
      "step": 8890
    },
    {
      "epoch": 0.9405920974411139,
      "grad_norm": 1.107131838798523,
      "learning_rate": 3.885773220632494e-05,
      "loss": 1.0631993293762207,
      "step": 8900
    },
    {
      "epoch": 0.9416489424944186,
      "grad_norm": 1.3992575407028198,
      "learning_rate": 3.883468111992235e-05,
      "loss": 1.1644817352294923,
      "step": 8910
    },
    {
      "epoch": 0.9427057875477232,
      "grad_norm": 1.591500163078308,
      "learning_rate": 3.8811613067467525e-05,
      "loss": 1.1591188430786132,
      "step": 8920
    },
    {
      "epoch": 0.9437626326010278,
      "grad_norm": 0.9797013998031616,
      "learning_rate": 3.878852807724979e-05,
      "loss": 1.145860767364502,
      "step": 8930
    },
    {
      "epoch": 0.9448194776543324,
      "grad_norm": 1.1045056581497192,
      "learning_rate": 3.876542617757924e-05,
      "loss": 1.1011402130126953,
      "step": 8940
    },
    {
      "epoch": 0.945876322707637,
      "grad_norm": 0.9365732073783875,
      "learning_rate": 3.8742307396786705e-05,
      "loss": 1.0173688888549806,
      "step": 8950
    },
    {
      "epoch": 0.9469331677609416,
      "grad_norm": 1.1072068214416504,
      "learning_rate": 3.8719171763223714e-05,
      "loss": 1.1174254417419434,
      "step": 8960
    },
    {
      "epoch": 0.9479900128142462,
      "grad_norm": 1.0571832656860352,
      "learning_rate": 3.869601930526249e-05,
      "loss": 1.0286152839660645,
      "step": 8970
    },
    {
      "epoch": 0.9490468578675509,
      "grad_norm": 0.9439742565155029,
      "learning_rate": 3.8672850051295845e-05,
      "loss": 1.0052661895751953,
      "step": 8980
    },
    {
      "epoch": 0.9501037029208556,
      "grad_norm": 1.0582489967346191,
      "learning_rate": 3.864966402973722e-05,
      "loss": 1.1112709045410156,
      "step": 8990
    },
    {
      "epoch": 0.9511605479741602,
      "grad_norm": 0.8677789568901062,
      "learning_rate": 3.862646126902061e-05,
      "loss": 1.080394172668457,
      "step": 9000
    },
    {
      "epoch": 0.9511605479741602,
      "eval_loss": 0.5095364451408386,
      "eval_runtime": 2987.9111,
      "eval_samples_per_second": 2.98,
      "eval_steps_per_second": 2.98,
      "step": 9000
    },
    {
      "epoch": 0.9522173930274648,
      "grad_norm": 1.0276281833648682,
      "learning_rate": 3.860324179760054e-05,
      "loss": 1.06030216217041,
      "step": 9010
    },
    {
      "epoch": 0.9532742380807694,
      "grad_norm": 0.9236569404602051,
      "learning_rate": 3.858000564395203e-05,
      "loss": 1.0085678100585938,
      "step": 9020
    },
    {
      "epoch": 0.954331083134074,
      "grad_norm": 1.195087194442749,
      "learning_rate": 3.8556752836570544e-05,
      "loss": 1.0834206581115722,
      "step": 9030
    },
    {
      "epoch": 0.9553879281873786,
      "grad_norm": 0.9809810519218445,
      "learning_rate": 3.8533483403971974e-05,
      "loss": 0.961732006072998,
      "step": 9040
    },
    {
      "epoch": 0.9564447732406832,
      "grad_norm": 0.8672991991043091,
      "learning_rate": 3.851019737469262e-05,
      "loss": 1.0726253509521484,
      "step": 9050
    },
    {
      "epoch": 0.9575016182939878,
      "grad_norm": 1.1261621713638306,
      "learning_rate": 3.848689477728912e-05,
      "loss": 1.0856582641601562,
      "step": 9060
    },
    {
      "epoch": 0.9585584633472924,
      "grad_norm": 0.965445876121521,
      "learning_rate": 3.846357564033842e-05,
      "loss": 0.9216752052307129,
      "step": 9070
    },
    {
      "epoch": 0.9596153084005972,
      "grad_norm": 1.0720208883285522,
      "learning_rate": 3.8440239992437766e-05,
      "loss": 1.059298038482666,
      "step": 9080
    },
    {
      "epoch": 0.9606721534539018,
      "grad_norm": 0.9337862133979797,
      "learning_rate": 3.841688786220465e-05,
      "loss": 1.1361495018005372,
      "step": 9090
    },
    {
      "epoch": 0.9617289985072064,
      "grad_norm": 1.6025053262710571,
      "learning_rate": 3.839351927827677e-05,
      "loss": 1.238121223449707,
      "step": 9100
    },
    {
      "epoch": 0.962785843560511,
      "grad_norm": 0.9100377559661865,
      "learning_rate": 3.837013426931199e-05,
      "loss": 1.0386764526367187,
      "step": 9110
    },
    {
      "epoch": 0.9638426886138156,
      "grad_norm": 1.401560664176941,
      "learning_rate": 3.8346732863988356e-05,
      "loss": 1.0504328727722168,
      "step": 9120
    },
    {
      "epoch": 0.9648995336671202,
      "grad_norm": 0.8137794733047485,
      "learning_rate": 3.832331509100399e-05,
      "loss": 1.198045253753662,
      "step": 9130
    },
    {
      "epoch": 0.9659563787204248,
      "grad_norm": 0.8146438002586365,
      "learning_rate": 3.8299880979077084e-05,
      "loss": 1.0876679420471191,
      "step": 9140
    },
    {
      "epoch": 0.9670132237737294,
      "grad_norm": 1.3340327739715576,
      "learning_rate": 3.8276430556945874e-05,
      "loss": 0.9491246223449707,
      "step": 9150
    },
    {
      "epoch": 0.9680700688270341,
      "grad_norm": 1.0964562892913818,
      "learning_rate": 3.825296385336863e-05,
      "loss": 0.9562935829162598,
      "step": 9160
    },
    {
      "epoch": 0.9691269138803387,
      "grad_norm": 1.1088480949401855,
      "learning_rate": 3.822948089712352e-05,
      "loss": 1.1593246459960938,
      "step": 9170
    },
    {
      "epoch": 0.9701837589336434,
      "grad_norm": 0.8632457852363586,
      "learning_rate": 3.820598171700871e-05,
      "loss": 1.0248550415039062,
      "step": 9180
    },
    {
      "epoch": 0.971240603986948,
      "grad_norm": 1.2048380374908447,
      "learning_rate": 3.818246634184222e-05,
      "loss": 1.1579178810119628,
      "step": 9190
    },
    {
      "epoch": 0.9722974490402526,
      "grad_norm": 1.0638751983642578,
      "learning_rate": 3.815893480046195e-05,
      "loss": 1.0616660118103027,
      "step": 9200
    },
    {
      "epoch": 0.9733542940935572,
      "grad_norm": 1.2749416828155518,
      "learning_rate": 3.813538712172562e-05,
      "loss": 1.0530613899230956,
      "step": 9210
    },
    {
      "epoch": 0.9744111391468618,
      "grad_norm": 1.0472627878189087,
      "learning_rate": 3.8111823334510746e-05,
      "loss": 1.065870761871338,
      "step": 9220
    },
    {
      "epoch": 0.9754679842001664,
      "grad_norm": 0.8699654340744019,
      "learning_rate": 3.808824346771458e-05,
      "loss": 0.8995500564575195,
      "step": 9230
    },
    {
      "epoch": 0.9765248292534711,
      "grad_norm": 1.1017961502075195,
      "learning_rate": 3.806464755025413e-05,
      "loss": 1.0475339889526367,
      "step": 9240
    },
    {
      "epoch": 0.9775816743067757,
      "grad_norm": 0.9898455739021301,
      "learning_rate": 3.804103561106603e-05,
      "loss": 1.1402777671813964,
      "step": 9250
    },
    {
      "epoch": 0.9786385193600803,
      "grad_norm": 0.889009416103363,
      "learning_rate": 3.801740767910662e-05,
      "loss": 1.1249762535095216,
      "step": 9260
    },
    {
      "epoch": 0.9796953644133849,
      "grad_norm": 1.0249567031860352,
      "learning_rate": 3.799376378335182e-05,
      "loss": 1.0641023635864257,
      "step": 9270
    },
    {
      "epoch": 0.9807522094666896,
      "grad_norm": 0.9896396398544312,
      "learning_rate": 3.797010395279713e-05,
      "loss": 1.0644102096557617,
      "step": 9280
    },
    {
      "epoch": 0.9818090545199942,
      "grad_norm": 0.9756669402122498,
      "learning_rate": 3.7946428216457606e-05,
      "loss": 1.0321008682250976,
      "step": 9290
    },
    {
      "epoch": 0.9828658995732988,
      "grad_norm": 1.7557940483093262,
      "learning_rate": 3.7922736603367806e-05,
      "loss": 1.2038480758666992,
      "step": 9300
    },
    {
      "epoch": 0.9839227446266035,
      "grad_norm": 1.5625696182250977,
      "learning_rate": 3.7899029142581744e-05,
      "loss": 1.0666563987731934,
      "step": 9310
    },
    {
      "epoch": 0.9849795896799081,
      "grad_norm": 1.3309168815612793,
      "learning_rate": 3.7875305863172876e-05,
      "loss": 0.9857009887695313,
      "step": 9320
    },
    {
      "epoch": 0.9860364347332127,
      "grad_norm": 0.873198926448822,
      "learning_rate": 3.785156679423406e-05,
      "loss": 1.0638373374938965,
      "step": 9330
    },
    {
      "epoch": 0.9870932797865173,
      "grad_norm": 1.05149507522583,
      "learning_rate": 3.7827811964877525e-05,
      "loss": 1.080675220489502,
      "step": 9340
    },
    {
      "epoch": 0.9881501248398219,
      "grad_norm": 0.9998891353607178,
      "learning_rate": 3.78040414042348e-05,
      "loss": 1.0450897216796875,
      "step": 9350
    },
    {
      "epoch": 0.9892069698931265,
      "grad_norm": 0.9954738020896912,
      "learning_rate": 3.778025514145676e-05,
      "loss": 0.9320241928100585,
      "step": 9360
    },
    {
      "epoch": 0.9902638149464311,
      "grad_norm": 1.185196042060852,
      "learning_rate": 3.775645320571345e-05,
      "loss": 1.0744903564453125,
      "step": 9370
    },
    {
      "epoch": 0.9913206599997358,
      "grad_norm": 0.9072500467300415,
      "learning_rate": 3.7732635626194234e-05,
      "loss": 1.0737714767456055,
      "step": 9380
    },
    {
      "epoch": 0.9923775050530405,
      "grad_norm": 1.1410707235336304,
      "learning_rate": 3.770880243210758e-05,
      "loss": 1.107829761505127,
      "step": 9390
    },
    {
      "epoch": 0.9934343501063451,
      "grad_norm": 1.0401357412338257,
      "learning_rate": 3.7684953652681164e-05,
      "loss": 1.1012465476989746,
      "step": 9400
    },
    {
      "epoch": 0.9944911951596497,
      "grad_norm": 1.5529963970184326,
      "learning_rate": 3.7661089317161723e-05,
      "loss": 1.1889557838439941,
      "step": 9410
    },
    {
      "epoch": 0.9955480402129543,
      "grad_norm": 1.0409170389175415,
      "learning_rate": 3.7637209454815124e-05,
      "loss": 1.1331655502319335,
      "step": 9420
    },
    {
      "epoch": 0.9966048852662589,
      "grad_norm": 1.05330491065979,
      "learning_rate": 3.761331409492622e-05,
      "loss": 1.0663962364196777,
      "step": 9430
    },
    {
      "epoch": 0.9976617303195635,
      "grad_norm": 0.8549003601074219,
      "learning_rate": 3.758940326679893e-05,
      "loss": 1.0907514572143555,
      "step": 9440
    },
    {
      "epoch": 0.9987185753728681,
      "grad_norm": 0.8117828965187073,
      "learning_rate": 3.756547699975608e-05,
      "loss": 1.087899875640869,
      "step": 9450
    },
    {
      "epoch": 0.9997754204261727,
      "grad_norm": 0.8432764410972595,
      "learning_rate": 3.754153532313947e-05,
      "loss": 1.0888066291809082,
      "step": 9460
    },
    {
      "epoch": 1.0007397915373133,
      "grad_norm": 1.0258173942565918,
      "learning_rate": 3.751757826630979e-05,
      "loss": 0.9602935791015625,
      "step": 9470
    },
    {
      "epoch": 1.001796636590618,
      "grad_norm": 1.0536227226257324,
      "learning_rate": 3.7493605858646585e-05,
      "loss": 1.0546730041503907,
      "step": 9480
    },
    {
      "epoch": 1.0028534816439225,
      "grad_norm": 1.2590547800064087,
      "learning_rate": 3.746961812954822e-05,
      "loss": 0.9594717979431152,
      "step": 9490
    },
    {
      "epoch": 1.0039103266972271,
      "grad_norm": 0.9837392568588257,
      "learning_rate": 3.744561510843188e-05,
      "loss": 0.9294740676879882,
      "step": 9500
    },
    {
      "epoch": 1.0039103266972271,
      "eval_loss": 0.5072765350341797,
      "eval_runtime": 2987.4494,
      "eval_samples_per_second": 2.981,
      "eval_steps_per_second": 2.981,
      "step": 9500
    },
    {
      "epoch": 1.0049671717505317,
      "grad_norm": 0.8590106964111328,
      "learning_rate": 3.742159682473346e-05,
      "loss": 1.0078791618347167,
      "step": 9510
    },
    {
      "epoch": 1.0060240168038364,
      "grad_norm": 1.074690580368042,
      "learning_rate": 3.7397563307907604e-05,
      "loss": 1.0454870223999024,
      "step": 9520
    },
    {
      "epoch": 1.007080861857141,
      "grad_norm": 0.8658530712127686,
      "learning_rate": 3.737351458742762e-05,
      "loss": 0.9593657493591309,
      "step": 9530
    },
    {
      "epoch": 1.0081377069104456,
      "grad_norm": 1.2475427389144897,
      "learning_rate": 3.734945069278547e-05,
      "loss": 0.9837224960327149,
      "step": 9540
    },
    {
      "epoch": 1.0091945519637502,
      "grad_norm": 0.8479542136192322,
      "learning_rate": 3.732537165349171e-05,
      "loss": 1.0366480827331543,
      "step": 9550
    },
    {
      "epoch": 1.0102513970170548,
      "grad_norm": 1.1570239067077637,
      "learning_rate": 3.730127749907551e-05,
      "loss": 1.0528388977050782,
      "step": 9560
    },
    {
      "epoch": 1.0113082420703594,
      "grad_norm": 0.9179108738899231,
      "learning_rate": 3.727716825908452e-05,
      "loss": 0.8828094482421875,
      "step": 9570
    },
    {
      "epoch": 1.012365087123664,
      "grad_norm": 1.3105874061584473,
      "learning_rate": 3.725304396308492e-05,
      "loss": 0.9408868789672852,
      "step": 9580
    },
    {
      "epoch": 1.0134219321769686,
      "grad_norm": 1.1715439558029175,
      "learning_rate": 3.7228904640661355e-05,
      "loss": 1.0274457931518555,
      "step": 9590
    },
    {
      "epoch": 1.0144787772302732,
      "grad_norm": 0.8546901941299438,
      "learning_rate": 3.72047503214169e-05,
      "loss": 1.078207302093506,
      "step": 9600
    },
    {
      "epoch": 1.0155356222835779,
      "grad_norm": 0.9035372138023376,
      "learning_rate": 3.7180581034973e-05,
      "loss": 1.0901476860046386,
      "step": 9610
    },
    {
      "epoch": 1.0165924673368825,
      "grad_norm": 0.9713095426559448,
      "learning_rate": 3.7156396810969474e-05,
      "loss": 0.8290703773498536,
      "step": 9620
    },
    {
      "epoch": 1.0176493123901873,
      "grad_norm": 1.0144702196121216,
      "learning_rate": 3.713219767906446e-05,
      "loss": 0.9668977737426758,
      "step": 9630
    },
    {
      "epoch": 1.018706157443492,
      "grad_norm": 0.9240712523460388,
      "learning_rate": 3.710798366893437e-05,
      "loss": 1.0244888305664062,
      "step": 9640
    },
    {
      "epoch": 1.0197630024967965,
      "grad_norm": 0.9347766041755676,
      "learning_rate": 3.7083754810273865e-05,
      "loss": 0.9687684059143067,
      "step": 9650
    },
    {
      "epoch": 1.0208198475501011,
      "grad_norm": 0.9049181342124939,
      "learning_rate": 3.705951113279581e-05,
      "loss": 0.9276289939880371,
      "step": 9660
    },
    {
      "epoch": 1.0218766926034057,
      "grad_norm": 1.057160496711731,
      "learning_rate": 3.7035252666231254e-05,
      "loss": 0.9800550460815429,
      "step": 9670
    },
    {
      "epoch": 1.0229335376567104,
      "grad_norm": 0.9341446757316589,
      "learning_rate": 3.701097944032938e-05,
      "loss": 1.0175441741943358,
      "step": 9680
    },
    {
      "epoch": 1.023990382710015,
      "grad_norm": 1.4323673248291016,
      "learning_rate": 3.6986691484857446e-05,
      "loss": 0.9825938224792481,
      "step": 9690
    },
    {
      "epoch": 1.0250472277633196,
      "grad_norm": 0.781416654586792,
      "learning_rate": 3.6962388829600824e-05,
      "loss": 1.2242677688598633,
      "step": 9700
    },
    {
      "epoch": 1.0261040728166242,
      "grad_norm": 1.1862646341323853,
      "learning_rate": 3.693807150436287e-05,
      "loss": 1.0485265731811524,
      "step": 9710
    },
    {
      "epoch": 1.0271609178699288,
      "grad_norm": 0.9284079670906067,
      "learning_rate": 3.6913739538964935e-05,
      "loss": 1.0121156692504882,
      "step": 9720
    },
    {
      "epoch": 1.0282177629232334,
      "grad_norm": 1.3418508768081665,
      "learning_rate": 3.6889392963246346e-05,
      "loss": 1.0273515701293945,
      "step": 9730
    },
    {
      "epoch": 1.029274607976538,
      "grad_norm": 1.5600773096084595,
      "learning_rate": 3.686503180706434e-05,
      "loss": 1.0140334129333497,
      "step": 9740
    },
    {
      "epoch": 1.0303314530298426,
      "grad_norm": 1.0218265056610107,
      "learning_rate": 3.6840656100294014e-05,
      "loss": 0.9696846008300781,
      "step": 9750
    },
    {
      "epoch": 1.0313882980831472,
      "grad_norm": 1.2220377922058105,
      "learning_rate": 3.681626587282833e-05,
      "loss": 1.0265470504760743,
      "step": 9760
    },
    {
      "epoch": 1.0324451431364519,
      "grad_norm": 0.8506842255592346,
      "learning_rate": 3.6791861154578054e-05,
      "loss": 0.9741664886474609,
      "step": 9770
    },
    {
      "epoch": 1.0335019881897565,
      "grad_norm": 1.1883351802825928,
      "learning_rate": 3.676744197547172e-05,
      "loss": 0.9969266891479492,
      "step": 9780
    },
    {
      "epoch": 1.034558833243061,
      "grad_norm": 0.969569206237793,
      "learning_rate": 3.674300836545559e-05,
      "loss": 0.8457400321960449,
      "step": 9790
    },
    {
      "epoch": 1.0356156782963657,
      "grad_norm": 1.2164390087127686,
      "learning_rate": 3.671856035449364e-05,
      "loss": 0.819858455657959,
      "step": 9800
    },
    {
      "epoch": 1.0366725233496703,
      "grad_norm": 1.015383005142212,
      "learning_rate": 3.6694097972567475e-05,
      "loss": 0.9455970764160156,
      "step": 9810
    },
    {
      "epoch": 1.037729368402975,
      "grad_norm": 0.8865264058113098,
      "learning_rate": 3.666962124967637e-05,
      "loss": 1.0069087028503418,
      "step": 9820
    },
    {
      "epoch": 1.0387862134562798,
      "grad_norm": 1.0854216814041138,
      "learning_rate": 3.664513021583714e-05,
      "loss": 0.9870096206665039,
      "step": 9830
    },
    {
      "epoch": 1.0398430585095844,
      "grad_norm": 1.1880682706832886,
      "learning_rate": 3.662062490108419e-05,
      "loss": 1.0078414916992187,
      "step": 9840
    },
    {
      "epoch": 1.040899903562889,
      "grad_norm": 1.1020530462265015,
      "learning_rate": 3.659610533546941e-05,
      "loss": 0.9827878952026368,
      "step": 9850
    },
    {
      "epoch": 1.0419567486161936,
      "grad_norm": 1.4285173416137695,
      "learning_rate": 3.657157154906217e-05,
      "loss": 1.024649429321289,
      "step": 9860
    },
    {
      "epoch": 1.0430135936694982,
      "grad_norm": 1.0874494314193726,
      "learning_rate": 3.65470235719493e-05,
      "loss": 0.9580174446105957,
      "step": 9870
    },
    {
      "epoch": 1.0440704387228028,
      "grad_norm": 1.1810954809188843,
      "learning_rate": 3.6522461434235e-05,
      "loss": 1.0391739845275878,
      "step": 9880
    },
    {
      "epoch": 1.0451272837761074,
      "grad_norm": 1.0377846956253052,
      "learning_rate": 3.649788516604088e-05,
      "loss": 1.0625799179077149,
      "step": 9890
    },
    {
      "epoch": 1.046184128829412,
      "grad_norm": 0.8741491436958313,
      "learning_rate": 3.6473294797505825e-05,
      "loss": 0.970775032043457,
      "step": 9900
    },
    {
      "epoch": 1.0472409738827166,
      "grad_norm": 0.9691803455352783,
      "learning_rate": 3.644869035878606e-05,
      "loss": 1.0845745086669922,
      "step": 9910
    },
    {
      "epoch": 1.0482978189360213,
      "grad_norm": 1.051673173904419,
      "learning_rate": 3.642407188005503e-05,
      "loss": 1.0454957962036133,
      "step": 9920
    },
    {
      "epoch": 1.0493546639893259,
      "grad_norm": 1.0918179750442505,
      "learning_rate": 3.639943939150343e-05,
      "loss": 1.188261127471924,
      "step": 9930
    },
    {
      "epoch": 1.0504115090426305,
      "grad_norm": 0.9021835923194885,
      "learning_rate": 3.6374792923339104e-05,
      "loss": 0.9768836975097657,
      "step": 9940
    },
    {
      "epoch": 1.051468354095935,
      "grad_norm": 0.9083480834960938,
      "learning_rate": 3.635013250578706e-05,
      "loss": 0.9843744277954102,
      "step": 9950
    },
    {
      "epoch": 1.0525251991492397,
      "grad_norm": 0.9820945858955383,
      "learning_rate": 3.63254581690894e-05,
      "loss": 0.9024385452270508,
      "step": 9960
    },
    {
      "epoch": 1.0535820442025443,
      "grad_norm": 1.0076723098754883,
      "learning_rate": 3.630076994350532e-05,
      "loss": 0.840729808807373,
      "step": 9970
    },
    {
      "epoch": 1.054638889255849,
      "grad_norm": 0.9208500385284424,
      "learning_rate": 3.6276067859311e-05,
      "loss": 1.0479769706726074,
      "step": 9980
    },
    {
      "epoch": 1.0556957343091535,
      "grad_norm": 0.9983416795730591,
      "learning_rate": 3.625135194679968e-05,
      "loss": 1.0025667190551757,
      "step": 9990
    },
    {
      "epoch": 1.0567525793624581,
      "grad_norm": 1.3875505924224854,
      "learning_rate": 3.622662223628151e-05,
      "loss": 0.9995326042175293,
      "step": 10000
    },
    {
      "epoch": 1.0567525793624581,
      "eval_loss": 0.5064773559570312,
      "eval_runtime": 2976.7554,
      "eval_samples_per_second": 2.992,
      "eval_steps_per_second": 2.992,
      "step": 10000
    },
    {
      "epoch": 1.0578094244157628,
      "grad_norm": 0.960098385810852,
      "learning_rate": 3.6201878758083574e-05,
      "loss": 1.0396389961242676,
      "step": 10010
    },
    {
      "epoch": 1.0588662694690674,
      "grad_norm": 1.0883495807647705,
      "learning_rate": 3.617712154254984e-05,
      "loss": 1.0639999389648438,
      "step": 10020
    },
    {
      "epoch": 1.059923114522372,
      "grad_norm": 1.2339361906051636,
      "learning_rate": 3.6152350620041134e-05,
      "loss": 0.9322826385498046,
      "step": 10030
    },
    {
      "epoch": 1.0609799595756768,
      "grad_norm": 1.081188440322876,
      "learning_rate": 3.612756602093506e-05,
      "loss": 1.0282750129699707,
      "step": 10040
    },
    {
      "epoch": 1.0620368046289814,
      "grad_norm": 1.0544086694717407,
      "learning_rate": 3.610276777562604e-05,
      "loss": 0.9352409362792968,
      "step": 10050
    },
    {
      "epoch": 1.063093649682286,
      "grad_norm": 1.3904283046722412,
      "learning_rate": 3.60779559145252e-05,
      "loss": 0.9763026237487793,
      "step": 10060
    },
    {
      "epoch": 1.0641504947355906,
      "grad_norm": 1.1958587169647217,
      "learning_rate": 3.6053130468060375e-05,
      "loss": 1.0520586967468262,
      "step": 10070
    },
    {
      "epoch": 1.0652073397888953,
      "grad_norm": 1.4271001815795898,
      "learning_rate": 3.602829146667604e-05,
      "loss": 1.0832015037536622,
      "step": 10080
    },
    {
      "epoch": 1.0662641848421999,
      "grad_norm": 0.8596353530883789,
      "learning_rate": 3.600343894083332e-05,
      "loss": 0.8907459259033204,
      "step": 10090
    },
    {
      "epoch": 1.0673210298955045,
      "grad_norm": 1.0677231550216675,
      "learning_rate": 3.597857292100991e-05,
      "loss": 0.9511052131652832,
      "step": 10100
    },
    {
      "epoch": 1.068377874948809,
      "grad_norm": 1.211313247680664,
      "learning_rate": 3.595369343770006e-05,
      "loss": 0.958437156677246,
      "step": 10110
    },
    {
      "epoch": 1.0694347200021137,
      "grad_norm": 1.4189872741699219,
      "learning_rate": 3.5928800521414537e-05,
      "loss": 1.0193836212158203,
      "step": 10120
    },
    {
      "epoch": 1.0704915650554183,
      "grad_norm": 1.0202900171279907,
      "learning_rate": 3.590389420268056e-05,
      "loss": 0.997863483428955,
      "step": 10130
    },
    {
      "epoch": 1.071548410108723,
      "grad_norm": 1.2425025701522827,
      "learning_rate": 3.587897451204179e-05,
      "loss": 1.040816307067871,
      "step": 10140
    },
    {
      "epoch": 1.0726052551620275,
      "grad_norm": 0.884363055229187,
      "learning_rate": 3.585404148005832e-05,
      "loss": 1.0276570320129395,
      "step": 10150
    },
    {
      "epoch": 1.0736621002153321,
      "grad_norm": 1.0173248052597046,
      "learning_rate": 3.582909513730655e-05,
      "loss": 0.9309420585632324,
      "step": 10160
    },
    {
      "epoch": 1.0747189452686368,
      "grad_norm": 1.1544816493988037,
      "learning_rate": 3.580413551437926e-05,
      "loss": 1.104496383666992,
      "step": 10170
    },
    {
      "epoch": 1.0757757903219414,
      "grad_norm": 1.1566483974456787,
      "learning_rate": 3.577916264188546e-05,
      "loss": 1.0383170127868653,
      "step": 10180
    },
    {
      "epoch": 1.076832635375246,
      "grad_norm": 0.8593904376029968,
      "learning_rate": 3.5754176550450444e-05,
      "loss": 0.9391170501708984,
      "step": 10190
    },
    {
      "epoch": 1.0778894804285506,
      "grad_norm": 0.8974155783653259,
      "learning_rate": 3.57291772707157e-05,
      "loss": 0.9734575271606445,
      "step": 10200
    },
    {
      "epoch": 1.0789463254818552,
      "grad_norm": 1.315087914466858,
      "learning_rate": 3.570416483333892e-05,
      "loss": 0.9753241539001465,
      "step": 10210
    },
    {
      "epoch": 1.0800031705351598,
      "grad_norm": 0.9139646887779236,
      "learning_rate": 3.5679139268993895e-05,
      "loss": 1.040658187866211,
      "step": 10220
    },
    {
      "epoch": 1.0810600155884647,
      "grad_norm": 0.9399287700653076,
      "learning_rate": 3.565410060837053e-05,
      "loss": 1.1236308097839356,
      "step": 10230
    },
    {
      "epoch": 1.0821168606417693,
      "grad_norm": 0.9876143932342529,
      "learning_rate": 3.5629048882174784e-05,
      "loss": 0.9247361183166504,
      "step": 10240
    },
    {
      "epoch": 1.0831737056950739,
      "grad_norm": 1.106675386428833,
      "learning_rate": 3.560398412112866e-05,
      "loss": 1.0239107131958007,
      "step": 10250
    },
    {
      "epoch": 1.0842305507483785,
      "grad_norm": 1.3117635250091553,
      "learning_rate": 3.557890635597012e-05,
      "loss": 1.0428512573242188,
      "step": 10260
    },
    {
      "epoch": 1.085287395801683,
      "grad_norm": 1.0871673822402954,
      "learning_rate": 3.555381561745308e-05,
      "loss": 1.0077105522155763,
      "step": 10270
    },
    {
      "epoch": 1.0863442408549877,
      "grad_norm": 1.0015088319778442,
      "learning_rate": 3.5528711936347375e-05,
      "loss": 0.9590434074401856,
      "step": 10280
    },
    {
      "epoch": 1.0874010859082923,
      "grad_norm": 1.0588245391845703,
      "learning_rate": 3.5503595343438714e-05,
      "loss": 0.8923368453979492,
      "step": 10290
    },
    {
      "epoch": 1.088457930961597,
      "grad_norm": 1.224919319152832,
      "learning_rate": 3.547846586952861e-05,
      "loss": 1.0265795707702636,
      "step": 10300
    },
    {
      "epoch": 1.0895147760149015,
      "grad_norm": 1.1033817529678345,
      "learning_rate": 3.5453323545434416e-05,
      "loss": 0.9855875968933105,
      "step": 10310
    },
    {
      "epoch": 1.0905716210682062,
      "grad_norm": 1.1828932762145996,
      "learning_rate": 3.542816840198922e-05,
      "loss": 0.9499850273132324,
      "step": 10320
    },
    {
      "epoch": 1.0916284661215108,
      "grad_norm": 1.2937909364700317,
      "learning_rate": 3.540300047004183e-05,
      "loss": 1.0121013641357421,
      "step": 10330
    },
    {
      "epoch": 1.0926853111748154,
      "grad_norm": 1.0738261938095093,
      "learning_rate": 3.537781978045674e-05,
      "loss": 0.8974007606506348,
      "step": 10340
    },
    {
      "epoch": 1.09374215622812,
      "grad_norm": 0.9981940984725952,
      "learning_rate": 3.53526263641141e-05,
      "loss": 1.0058355331420898,
      "step": 10350
    },
    {
      "epoch": 1.0947990012814246,
      "grad_norm": 1.3721922636032104,
      "learning_rate": 3.532742025190964e-05,
      "loss": 1.0321287155151366,
      "step": 10360
    },
    {
      "epoch": 1.0958558463347292,
      "grad_norm": 0.9843775629997253,
      "learning_rate": 3.53022014747547e-05,
      "loss": 0.9847609519958496,
      "step": 10370
    },
    {
      "epoch": 1.0969126913880338,
      "grad_norm": 1.6965134143829346,
      "learning_rate": 3.5276970063576103e-05,
      "loss": 0.9981295585632324,
      "step": 10380
    },
    {
      "epoch": 1.0979695364413384,
      "grad_norm": 1.2805424928665161,
      "learning_rate": 3.5251726049316214e-05,
      "loss": 0.9802008628845215,
      "step": 10390
    },
    {
      "epoch": 1.099026381494643,
      "grad_norm": 1.4362438917160034,
      "learning_rate": 3.5226469462932814e-05,
      "loss": 0.9879981994628906,
      "step": 10400
    },
    {
      "epoch": 1.1000832265479477,
      "grad_norm": 1.1449130773544312,
      "learning_rate": 3.520120033539914e-05,
      "loss": 0.8934724807739258,
      "step": 10410
    },
    {
      "epoch": 1.1011400716012523,
      "grad_norm": 1.2525060176849365,
      "learning_rate": 3.5175918697703755e-05,
      "loss": 0.9800069808959961,
      "step": 10420
    },
    {
      "epoch": 1.1021969166545569,
      "grad_norm": 1.1016544103622437,
      "learning_rate": 3.5150624580850635e-05,
      "loss": 0.9752134323120117,
      "step": 10430
    },
    {
      "epoch": 1.1032537617078617,
      "grad_norm": 1.0100404024124146,
      "learning_rate": 3.512531801585898e-05,
      "loss": 1.0574529647827149,
      "step": 10440
    },
    {
      "epoch": 1.1043106067611663,
      "grad_norm": 1.0399388074874878,
      "learning_rate": 3.509999903376334e-05,
      "loss": 0.9747302055358886,
      "step": 10450
    },
    {
      "epoch": 1.105367451814471,
      "grad_norm": 1.2418626546859741,
      "learning_rate": 3.5074667665613406e-05,
      "loss": 1.0113841056823731,
      "step": 10460
    },
    {
      "epoch": 1.1064242968677755,
      "grad_norm": 1.124110221862793,
      "learning_rate": 3.5049323942474146e-05,
      "loss": 0.8662083625793457,
      "step": 10470
    },
    {
      "epoch": 1.1074811419210802,
      "grad_norm": 1.5243151187896729,
      "learning_rate": 3.502396789542559e-05,
      "loss": 1.053980541229248,
      "step": 10480
    },
    {
      "epoch": 1.1085379869743848,
      "grad_norm": 1.2846652269363403,
      "learning_rate": 3.4998599555562965e-05,
      "loss": 0.9840180397033691,
      "step": 10490
    },
    {
      "epoch": 1.1095948320276894,
      "grad_norm": 0.9215627312660217,
      "learning_rate": 3.4973218953996514e-05,
      "loss": 1.0429940223693848,
      "step": 10500
    },
    {
      "epoch": 1.1095948320276894,
      "eval_loss": 0.5049917101860046,
      "eval_runtime": 2978.9917,
      "eval_samples_per_second": 2.989,
      "eval_steps_per_second": 2.989,
      "step": 10500
    },
    {
      "epoch": 1.110651677080994,
      "grad_norm": 1.0097287893295288,
      "learning_rate": 3.4947826121851546e-05,
      "loss": 0.9286057472229003,
      "step": 10510
    },
    {
      "epoch": 1.1117085221342986,
      "grad_norm": 0.9569366574287415,
      "learning_rate": 3.492242109026835e-05,
      "loss": 1.0177912712097168,
      "step": 10520
    },
    {
      "epoch": 1.1127653671876032,
      "grad_norm": 1.3863794803619385,
      "learning_rate": 3.4897003890402196e-05,
      "loss": 0.956661605834961,
      "step": 10530
    },
    {
      "epoch": 1.1138222122409078,
      "grad_norm": 0.9264512062072754,
      "learning_rate": 3.487157455342327e-05,
      "loss": 1.021488857269287,
      "step": 10540
    },
    {
      "epoch": 1.1148790572942124,
      "grad_norm": 0.9683257341384888,
      "learning_rate": 3.484613311051663e-05,
      "loss": 0.9438602447509765,
      "step": 10550
    },
    {
      "epoch": 1.115935902347517,
      "grad_norm": 1.2292051315307617,
      "learning_rate": 3.482067959288219e-05,
      "loss": 1.1107909202575683,
      "step": 10560
    },
    {
      "epoch": 1.1169927474008217,
      "grad_norm": 1.100102424621582,
      "learning_rate": 3.479521403173468e-05,
      "loss": 1.018934726715088,
      "step": 10570
    },
    {
      "epoch": 1.1180495924541263,
      "grad_norm": 0.9553000330924988,
      "learning_rate": 3.476973645830358e-05,
      "loss": 1.0619715690612792,
      "step": 10580
    },
    {
      "epoch": 1.1191064375074309,
      "grad_norm": 1.197253704071045,
      "learning_rate": 3.4744246903833124e-05,
      "loss": 1.0397119522094727,
      "step": 10590
    },
    {
      "epoch": 1.1201632825607355,
      "grad_norm": 0.9221305251121521,
      "learning_rate": 3.47187453995822e-05,
      "loss": 0.9063426971435546,
      "step": 10600
    },
    {
      "epoch": 1.12122012761404,
      "grad_norm": 1.1021281480789185,
      "learning_rate": 3.4693231976824416e-05,
      "loss": 1.044843864440918,
      "step": 10610
    },
    {
      "epoch": 1.1222769726673447,
      "grad_norm": 0.875792920589447,
      "learning_rate": 3.466770666684792e-05,
      "loss": 0.9562784194946289,
      "step": 10620
    },
    {
      "epoch": 1.1233338177206493,
      "grad_norm": 0.11101555079221725,
      "learning_rate": 3.46421695009555e-05,
      "loss": 0.9251822471618653,
      "step": 10630
    },
    {
      "epoch": 1.124390662773954,
      "grad_norm": 1.033149003982544,
      "learning_rate": 3.461662051046445e-05,
      "loss": 0.9189902305603027,
      "step": 10640
    },
    {
      "epoch": 1.1254475078272588,
      "grad_norm": 1.2588694095611572,
      "learning_rate": 3.4591059726706573e-05,
      "loss": 0.9638330459594726,
      "step": 10650
    },
    {
      "epoch": 1.1265043528805634,
      "grad_norm": 1.1429047584533691,
      "learning_rate": 3.456548718102813e-05,
      "loss": 0.9745611190795899,
      "step": 10660
    },
    {
      "epoch": 1.127561197933868,
      "grad_norm": 1.0573793649673462,
      "learning_rate": 3.4539902904789817e-05,
      "loss": 0.9007460594177246,
      "step": 10670
    },
    {
      "epoch": 1.1286180429871726,
      "grad_norm": 0.9109353423118591,
      "learning_rate": 3.4514306929366704e-05,
      "loss": 0.865353012084961,
      "step": 10680
    },
    {
      "epoch": 1.1296748880404772,
      "grad_norm": 1.112802267074585,
      "learning_rate": 3.4488699286148226e-05,
      "loss": 0.9748274803161621,
      "step": 10690
    },
    {
      "epoch": 1.1307317330937818,
      "grad_norm": 0.7796987891197205,
      "learning_rate": 3.44630800065381e-05,
      "loss": 0.9566667556762696,
      "step": 10700
    },
    {
      "epoch": 1.1317885781470864,
      "grad_norm": 1.084863543510437,
      "learning_rate": 3.443744912195434e-05,
      "loss": 1.0116111755371093,
      "step": 10710
    },
    {
      "epoch": 1.132845423200391,
      "grad_norm": 1.4855684041976929,
      "learning_rate": 3.4411806663829165e-05,
      "loss": 1.005078125,
      "step": 10720
    },
    {
      "epoch": 1.1339022682536957,
      "grad_norm": 1.0013991594314575,
      "learning_rate": 3.438615266360902e-05,
      "loss": 1.01129093170166,
      "step": 10730
    },
    {
      "epoch": 1.1349591133070003,
      "grad_norm": 0.8792324662208557,
      "learning_rate": 3.436048715275447e-05,
      "loss": 1.0450743675231933,
      "step": 10740
    },
    {
      "epoch": 1.136015958360305,
      "grad_norm": 1.5673162937164307,
      "learning_rate": 3.4334810162740226e-05,
      "loss": 0.949436092376709,
      "step": 10750
    },
    {
      "epoch": 1.1370728034136095,
      "grad_norm": 1.478444218635559,
      "learning_rate": 3.430912172505505e-05,
      "loss": 0.9472408294677734,
      "step": 10760
    },
    {
      "epoch": 1.1381296484669141,
      "grad_norm": 1.3334705829620361,
      "learning_rate": 3.428342187120176e-05,
      "loss": 1.0349833488464355,
      "step": 10770
    },
    {
      "epoch": 1.1391864935202187,
      "grad_norm": 1.1347382068634033,
      "learning_rate": 3.425771063269717e-05,
      "loss": 0.9633730888366699,
      "step": 10780
    },
    {
      "epoch": 1.1402433385735233,
      "grad_norm": 0.901841938495636,
      "learning_rate": 3.423198804107206e-05,
      "loss": 1.032583999633789,
      "step": 10790
    },
    {
      "epoch": 1.141300183626828,
      "grad_norm": 0.980816125869751,
      "learning_rate": 3.4206254127871115e-05,
      "loss": 1.0904212951660157,
      "step": 10800
    },
    {
      "epoch": 1.1423570286801326,
      "grad_norm": 1.107750415802002,
      "learning_rate": 3.418050892465292e-05,
      "loss": 0.9388874053955079,
      "step": 10810
    },
    {
      "epoch": 1.1434138737334372,
      "grad_norm": 1.0829613208770752,
      "learning_rate": 3.415475246298991e-05,
      "loss": 0.9979594230651856,
      "step": 10820
    },
    {
      "epoch": 1.1444707187867418,
      "grad_norm": 1.2724562883377075,
      "learning_rate": 3.412898477446833e-05,
      "loss": 0.9290336608886719,
      "step": 10830
    },
    {
      "epoch": 1.1455275638400466,
      "grad_norm": 1.3701120615005493,
      "learning_rate": 3.410320589068815e-05,
      "loss": 1.0164386749267578,
      "step": 10840
    },
    {
      "epoch": 1.146584408893351,
      "grad_norm": 1.1156965494155884,
      "learning_rate": 3.407741584326313e-05,
      "loss": 0.9006812095642089,
      "step": 10850
    },
    {
      "epoch": 1.1476412539466558,
      "grad_norm": 0.8329572677612305,
      "learning_rate": 3.4051614663820665e-05,
      "loss": 1.075608730316162,
      "step": 10860
    },
    {
      "epoch": 1.1486980989999604,
      "grad_norm": 1.2491638660430908,
      "learning_rate": 3.402580238400185e-05,
      "loss": 1.0135152816772461,
      "step": 10870
    },
    {
      "epoch": 1.149754944053265,
      "grad_norm": 1.3607935905456543,
      "learning_rate": 3.399997903546137e-05,
      "loss": 0.9211840629577637,
      "step": 10880
    },
    {
      "epoch": 1.1508117891065697,
      "grad_norm": 1.0567494630813599,
      "learning_rate": 3.397414464986747e-05,
      "loss": 0.9834133148193359,
      "step": 10890
    },
    {
      "epoch": 1.1518686341598743,
      "grad_norm": 0.9664644598960876,
      "learning_rate": 3.394829925890196e-05,
      "loss": 1.0323697090148927,
      "step": 10900
    },
    {
      "epoch": 1.152925479213179,
      "grad_norm": 1.0836256742477417,
      "learning_rate": 3.3922442894260124e-05,
      "loss": 1.035941982269287,
      "step": 10910
    },
    {
      "epoch": 1.1539823242664835,
      "grad_norm": 0.7941169738769531,
      "learning_rate": 3.389657558765072e-05,
      "loss": 0.9320117950439453,
      "step": 10920
    },
    {
      "epoch": 1.1550391693197881,
      "grad_norm": 1.501661777496338,
      "learning_rate": 3.38706973707959e-05,
      "loss": 0.9003686904907227,
      "step": 10930
    },
    {
      "epoch": 1.1560960143730927,
      "grad_norm": 1.407162070274353,
      "learning_rate": 3.384480827543123e-05,
      "loss": 1.0633143424987792,
      "step": 10940
    },
    {
      "epoch": 1.1571528594263973,
      "grad_norm": 1.6598080396652222,
      "learning_rate": 3.381890833330558e-05,
      "loss": 1.0424124717712402,
      "step": 10950
    },
    {
      "epoch": 1.158209704479702,
      "grad_norm": 1.0970258712768555,
      "learning_rate": 3.379299757618115e-05,
      "loss": 0.9395333290100097,
      "step": 10960
    },
    {
      "epoch": 1.1592665495330066,
      "grad_norm": 1.009871244430542,
      "learning_rate": 3.376707603583339e-05,
      "loss": 1.1170613288879394,
      "step": 10970
    },
    {
      "epoch": 1.1603233945863112,
      "grad_norm": 1.2280417680740356,
      "learning_rate": 3.374114374405098e-05,
      "loss": 0.9225652694702149,
      "step": 10980
    },
    {
      "epoch": 1.1613802396396158,
      "grad_norm": 0.8802953362464905,
      "learning_rate": 3.371520073263578e-05,
      "loss": 0.9138586044311523,
      "step": 10990
    },
    {
      "epoch": 1.1624370846929204,
      "grad_norm": 1.1920279264450073,
      "learning_rate": 3.3689247033402796e-05,
      "loss": 1.0754246711730957,
      "step": 11000
    },
    {
      "epoch": 1.1624370846929204,
      "eval_loss": 0.50250643491745,
      "eval_runtime": 2978.3253,
      "eval_samples_per_second": 2.99,
      "eval_steps_per_second": 2.99,
      "step": 11000
    },
    {
      "epoch": 1.163493929746225,
      "grad_norm": 1.8046077489852905,
      "learning_rate": 3.366328267818015e-05,
      "loss": 0.958592414855957,
      "step": 11010
    },
    {
      "epoch": 1.1645507747995296,
      "grad_norm": 1.0012996196746826,
      "learning_rate": 3.3637307698809015e-05,
      "loss": 1.1365827560424804,
      "step": 11020
    },
    {
      "epoch": 1.1656076198528342,
      "grad_norm": 1.1787137985229492,
      "learning_rate": 3.3611322127143604e-05,
      "loss": 1.0598387718200684,
      "step": 11030
    },
    {
      "epoch": 1.1666644649061388,
      "grad_norm": 0.9612213373184204,
      "learning_rate": 3.358532599505113e-05,
      "loss": 1.0291789054870606,
      "step": 11040
    },
    {
      "epoch": 1.1677213099594437,
      "grad_norm": 1.2199721336364746,
      "learning_rate": 3.355931933441173e-05,
      "loss": 1.0636857986450194,
      "step": 11050
    },
    {
      "epoch": 1.1687781550127483,
      "grad_norm": 0.9942165017127991,
      "learning_rate": 3.353330217711848e-05,
      "loss": 0.9055363655090332,
      "step": 11060
    },
    {
      "epoch": 1.169835000066053,
      "grad_norm": 0.8913378119468689,
      "learning_rate": 3.35072745550773e-05,
      "loss": 0.9619452476501464,
      "step": 11070
    },
    {
      "epoch": 1.1708918451193575,
      "grad_norm": 1.0257455110549927,
      "learning_rate": 3.348123650020697e-05,
      "loss": 0.9741548538208008,
      "step": 11080
    },
    {
      "epoch": 1.1719486901726621,
      "grad_norm": 1.1168900728225708,
      "learning_rate": 3.345518804443905e-05,
      "loss": 0.964204978942871,
      "step": 11090
    },
    {
      "epoch": 1.1730055352259667,
      "grad_norm": 1.7098729610443115,
      "learning_rate": 3.342912921971787e-05,
      "loss": 1.066093349456787,
      "step": 11100
    },
    {
      "epoch": 1.1740623802792713,
      "grad_norm": 1.2739917039871216,
      "learning_rate": 3.340306005800045e-05,
      "loss": 0.8376961708068847,
      "step": 11110
    },
    {
      "epoch": 1.175119225332576,
      "grad_norm": 1.915449619293213,
      "learning_rate": 3.33769805912565e-05,
      "loss": 1.0273649215698242,
      "step": 11120
    },
    {
      "epoch": 1.1761760703858806,
      "grad_norm": 1.049568772315979,
      "learning_rate": 3.335089085146838e-05,
      "loss": 0.8861307144165039,
      "step": 11130
    },
    {
      "epoch": 1.1772329154391852,
      "grad_norm": 1.018538236618042,
      "learning_rate": 3.332479087063104e-05,
      "loss": 0.8627653121948242,
      "step": 11140
    },
    {
      "epoch": 1.1782897604924898,
      "grad_norm": 0.948227047920227,
      "learning_rate": 3.329868068075196e-05,
      "loss": 0.9736764907836915,
      "step": 11150
    },
    {
      "epoch": 1.1793466055457944,
      "grad_norm": 0.9588805437088013,
      "learning_rate": 3.3272560313851194e-05,
      "loss": 0.9640019416809082,
      "step": 11160
    },
    {
      "epoch": 1.180403450599099,
      "grad_norm": 1.0656278133392334,
      "learning_rate": 3.324642980196123e-05,
      "loss": 1.020623016357422,
      "step": 11170
    },
    {
      "epoch": 1.1814602956524036,
      "grad_norm": 1.286974549293518,
      "learning_rate": 3.322028917712703e-05,
      "loss": 0.9494619369506836,
      "step": 11180
    },
    {
      "epoch": 1.1825171407057082,
      "grad_norm": 1.151004672050476,
      "learning_rate": 3.319413847140593e-05,
      "loss": 1.0114944458007813,
      "step": 11190
    },
    {
      "epoch": 1.1835739857590128,
      "grad_norm": 0.9209614396095276,
      "learning_rate": 3.316797771686766e-05,
      "loss": 1.0235469818115235,
      "step": 11200
    },
    {
      "epoch": 1.1846308308123175,
      "grad_norm": 1.8705112934112549,
      "learning_rate": 3.3141806945594234e-05,
      "loss": 1.0668043136596679,
      "step": 11210
    },
    {
      "epoch": 1.185687675865622,
      "grad_norm": 0.8731357455253601,
      "learning_rate": 3.311562618967999e-05,
      "loss": 0.9664183616638183,
      "step": 11220
    },
    {
      "epoch": 1.1867445209189267,
      "grad_norm": 1.1739320755004883,
      "learning_rate": 3.3089435481231474e-05,
      "loss": 1.029689407348633,
      "step": 11230
    },
    {
      "epoch": 1.1878013659722315,
      "grad_norm": 1.0118616819381714,
      "learning_rate": 3.306323485236747e-05,
      "loss": 0.9983315467834473,
      "step": 11240
    },
    {
      "epoch": 1.188858211025536,
      "grad_norm": 1.0141278505325317,
      "learning_rate": 3.30370243352189e-05,
      "loss": 0.9942756652832031,
      "step": 11250
    },
    {
      "epoch": 1.1899150560788407,
      "grad_norm": 1.0541365146636963,
      "learning_rate": 3.301080396192884e-05,
      "loss": 0.9655875205993653,
      "step": 11260
    },
    {
      "epoch": 1.1909719011321453,
      "grad_norm": 1.372337818145752,
      "learning_rate": 3.298457376465242e-05,
      "loss": 1.0287919998168946,
      "step": 11270
    },
    {
      "epoch": 1.19202874618545,
      "grad_norm": 1.2811415195465088,
      "learning_rate": 3.2958333775556854e-05,
      "loss": 0.8143916130065918,
      "step": 11280
    },
    {
      "epoch": 1.1930855912387546,
      "grad_norm": 1.20811128616333,
      "learning_rate": 3.293208402682134e-05,
      "loss": 1.032754611968994,
      "step": 11290
    },
    {
      "epoch": 1.1941424362920592,
      "grad_norm": 0.874935507774353,
      "learning_rate": 3.290582455063704e-05,
      "loss": 0.8877401351928711,
      "step": 11300
    },
    {
      "epoch": 1.1951992813453638,
      "grad_norm": 0.9666517972946167,
      "learning_rate": 3.2879555379207075e-05,
      "loss": 0.956077766418457,
      "step": 11310
    },
    {
      "epoch": 1.1962561263986684,
      "grad_norm": 0.8753248453140259,
      "learning_rate": 3.285327654474642e-05,
      "loss": 0.9089556694030761,
      "step": 11320
    },
    {
      "epoch": 1.197312971451973,
      "grad_norm": 0.9536595344543457,
      "learning_rate": 3.282698807948192e-05,
      "loss": 0.8900654792785645,
      "step": 11330
    },
    {
      "epoch": 1.1983698165052776,
      "grad_norm": 1.18777334690094,
      "learning_rate": 3.280069001565224e-05,
      "loss": 1.027297306060791,
      "step": 11340
    },
    {
      "epoch": 1.1994266615585822,
      "grad_norm": 1.2462939023971558,
      "learning_rate": 3.277438238550779e-05,
      "loss": 0.8474176406860352,
      "step": 11350
    },
    {
      "epoch": 1.2004835066118869,
      "grad_norm": 1.6496219635009766,
      "learning_rate": 3.274806522131072e-05,
      "loss": 1.0438383102416993,
      "step": 11360
    },
    {
      "epoch": 1.2015403516651915,
      "grad_norm": 1.5829248428344727,
      "learning_rate": 3.272173855533489e-05,
      "loss": 0.9806596755981445,
      "step": 11370
    },
    {
      "epoch": 1.202597196718496,
      "grad_norm": 1.0794823169708252,
      "learning_rate": 3.269540241986579e-05,
      "loss": 0.8882059097290039,
      "step": 11380
    },
    {
      "epoch": 1.2036540417718007,
      "grad_norm": 0.9498823285102844,
      "learning_rate": 3.266905684720055e-05,
      "loss": 1.0201994895935058,
      "step": 11390
    },
    {
      "epoch": 1.2047108868251053,
      "grad_norm": 0.9658240079879761,
      "learning_rate": 3.2642701869647834e-05,
      "loss": 1.049924373626709,
      "step": 11400
    },
    {
      "epoch": 1.20576773187841,
      "grad_norm": 1.2706503868103027,
      "learning_rate": 3.261633751952788e-05,
      "loss": 1.0616437911987304,
      "step": 11410
    },
    {
      "epoch": 1.2068245769317145,
      "grad_norm": 1.206912636756897,
      "learning_rate": 3.258996382917238e-05,
      "loss": 0.9900564193725586,
      "step": 11420
    },
    {
      "epoch": 1.2078814219850191,
      "grad_norm": 1.1858651638031006,
      "learning_rate": 3.256358083092452e-05,
      "loss": 0.9495376586914063,
      "step": 11430
    },
    {
      "epoch": 1.2089382670383237,
      "grad_norm": 1.0544397830963135,
      "learning_rate": 3.2537188557138873e-05,
      "loss": 0.9665173530578614,
      "step": 11440
    },
    {
      "epoch": 1.2099951120916286,
      "grad_norm": 0.8116346001625061,
      "learning_rate": 3.25107870401814e-05,
      "loss": 1.004383373260498,
      "step": 11450
    },
    {
      "epoch": 1.211051957144933,
      "grad_norm": 0.922507107257843,
      "learning_rate": 3.248437631242939e-05,
      "loss": 0.9940532684326172,
      "step": 11460
    },
    {
      "epoch": 1.2121088021982378,
      "grad_norm": 1.3066717386245728,
      "learning_rate": 3.245795640627143e-05,
      "loss": 0.967806339263916,
      "step": 11470
    },
    {
      "epoch": 1.2131656472515424,
      "grad_norm": 1.045356035232544,
      "learning_rate": 3.2431527354107365e-05,
      "loss": 0.996211051940918,
      "step": 11480
    },
    {
      "epoch": 1.214222492304847,
      "grad_norm": 1.174093246459961,
      "learning_rate": 3.240508918834826e-05,
      "loss": 0.9614363670349121,
      "step": 11490
    },
    {
      "epoch": 1.2152793373581516,
      "grad_norm": 1.211547613143921,
      "learning_rate": 3.2378641941416345e-05,
      "loss": 1.02583589553833,
      "step": 11500
    },
    {
      "epoch": 1.2152793373581516,
      "eval_loss": 0.5010405778884888,
      "eval_runtime": 2979.9917,
      "eval_samples_per_second": 2.988,
      "eval_steps_per_second": 2.988,
      "step": 11500
    },
    {
      "epoch": 1.2163361824114562,
      "grad_norm": 0.9992912411689758,
      "learning_rate": 3.235218564574499e-05,
      "loss": 0.9828355789184571,
      "step": 11510
    },
    {
      "epoch": 1.2173930274647609,
      "grad_norm": 0.9713661074638367,
      "learning_rate": 3.232572033377869e-05,
      "loss": 1.0149167060852051,
      "step": 11520
    },
    {
      "epoch": 1.2184498725180655,
      "grad_norm": 0.9441234469413757,
      "learning_rate": 3.229924603797293e-05,
      "loss": 0.9938035011291504,
      "step": 11530
    },
    {
      "epoch": 1.21950671757137,
      "grad_norm": 1.17690110206604,
      "learning_rate": 3.2272762790794286e-05,
      "loss": 0.9554742813110352,
      "step": 11540
    },
    {
      "epoch": 1.2205635626246747,
      "grad_norm": 2.0366289615631104,
      "learning_rate": 3.224627062472027e-05,
      "loss": 0.970057487487793,
      "step": 11550
    },
    {
      "epoch": 1.2216204076779793,
      "grad_norm": 1.2443503141403198,
      "learning_rate": 3.2219769572239347e-05,
      "loss": 1.0242173194885253,
      "step": 11560
    },
    {
      "epoch": 1.222677252731284,
      "grad_norm": 0.8777517676353455,
      "learning_rate": 3.2193259665850855e-05,
      "loss": 1.0647059440612794,
      "step": 11570
    },
    {
      "epoch": 1.2237340977845885,
      "grad_norm": 0.8020780682563782,
      "learning_rate": 3.216674093806503e-05,
      "loss": 0.9664150238037109,
      "step": 11580
    },
    {
      "epoch": 1.2247909428378931,
      "grad_norm": 1.0495637655258179,
      "learning_rate": 3.21402134214029e-05,
      "loss": 1.0083501815795899,
      "step": 11590
    },
    {
      "epoch": 1.2258477878911977,
      "grad_norm": 1.725145936012268,
      "learning_rate": 3.211367714839627e-05,
      "loss": 1.0444514274597168,
      "step": 11600
    },
    {
      "epoch": 1.2269046329445024,
      "grad_norm": 1.0575902462005615,
      "learning_rate": 3.208713215158769e-05,
      "loss": 0.9313383102416992,
      "step": 11610
    },
    {
      "epoch": 1.227961477997807,
      "grad_norm": 0.8725184202194214,
      "learning_rate": 3.2060578463530414e-05,
      "loss": 0.9940560340881348,
      "step": 11620
    },
    {
      "epoch": 1.2290183230511116,
      "grad_norm": 1.0746281147003174,
      "learning_rate": 3.203401611678834e-05,
      "loss": 0.9424043655395508,
      "step": 11630
    },
    {
      "epoch": 1.2300751681044162,
      "grad_norm": 1.467373251914978,
      "learning_rate": 3.200744514393601e-05,
      "loss": 0.9897848129272461,
      "step": 11640
    },
    {
      "epoch": 1.2311320131577208,
      "grad_norm": 0.9283172488212585,
      "learning_rate": 3.1980865577558504e-05,
      "loss": 1.055818748474121,
      "step": 11650
    },
    {
      "epoch": 1.2321888582110256,
      "grad_norm": 1.0851019620895386,
      "learning_rate": 3.195427745025148e-05,
      "loss": 1.037994956970215,
      "step": 11660
    },
    {
      "epoch": 1.2332457032643303,
      "grad_norm": 1.6464558839797974,
      "learning_rate": 3.192768079462107e-05,
      "loss": 1.0262088775634766,
      "step": 11670
    },
    {
      "epoch": 1.2343025483176349,
      "grad_norm": 1.0572965145111084,
      "learning_rate": 3.190107564328387e-05,
      "loss": 0.9359432220458984,
      "step": 11680
    },
    {
      "epoch": 1.2353593933709395,
      "grad_norm": 1.010490894317627,
      "learning_rate": 3.1874462028866914e-05,
      "loss": 1.0198223114013671,
      "step": 11690
    },
    {
      "epoch": 1.236416238424244,
      "grad_norm": 1.2359657287597656,
      "learning_rate": 3.184783998400759e-05,
      "loss": 0.9819694519042969,
      "step": 11700
    },
    {
      "epoch": 1.2374730834775487,
      "grad_norm": 0.8971331715583801,
      "learning_rate": 3.182120954135363e-05,
      "loss": 0.9137939453125,
      "step": 11710
    },
    {
      "epoch": 1.2385299285308533,
      "grad_norm": 1.0317646265029907,
      "learning_rate": 3.179457073356306e-05,
      "loss": 1.0403459548950196,
      "step": 11720
    },
    {
      "epoch": 1.239586773584158,
      "grad_norm": 1.1422234773635864,
      "learning_rate": 3.176792359330419e-05,
      "loss": 1.011024570465088,
      "step": 11730
    },
    {
      "epoch": 1.2406436186374625,
      "grad_norm": 1.1537139415740967,
      "learning_rate": 3.174126815325552e-05,
      "loss": 0.9205211639404297,
      "step": 11740
    },
    {
      "epoch": 1.2417004636907671,
      "grad_norm": 1.126369595527649,
      "learning_rate": 3.171460444610574e-05,
      "loss": 0.9044073104858399,
      "step": 11750
    },
    {
      "epoch": 1.2427573087440718,
      "grad_norm": 1.0465960502624512,
      "learning_rate": 3.168793250455369e-05,
      "loss": 0.9856606483459472,
      "step": 11760
    },
    {
      "epoch": 1.2438141537973764,
      "grad_norm": 0.9188291430473328,
      "learning_rate": 3.166125236130826e-05,
      "loss": 1.000680637359619,
      "step": 11770
    },
    {
      "epoch": 1.244870998850681,
      "grad_norm": 0.8994320631027222,
      "learning_rate": 3.163456404908848e-05,
      "loss": 0.9464153289794922,
      "step": 11780
    },
    {
      "epoch": 1.2459278439039856,
      "grad_norm": 0.9460260272026062,
      "learning_rate": 3.160786760062331e-05,
      "loss": 0.9891168594360351,
      "step": 11790
    },
    {
      "epoch": 1.2469846889572902,
      "grad_norm": 0.8047497272491455,
      "learning_rate": 3.1581163048651764e-05,
      "loss": 0.9742300987243653,
      "step": 11800
    },
    {
      "epoch": 1.2480415340105948,
      "grad_norm": 1.384482979774475,
      "learning_rate": 3.155445042592273e-05,
      "loss": 0.9539072990417481,
      "step": 11810
    },
    {
      "epoch": 1.2490983790638994,
      "grad_norm": 1.2500333786010742,
      "learning_rate": 3.152772976519505e-05,
      "loss": 0.9756543159484863,
      "step": 11820
    },
    {
      "epoch": 1.250155224117204,
      "grad_norm": 1.0395954847335815,
      "learning_rate": 3.150100109923737e-05,
      "loss": 0.9918712615966797,
      "step": 11830
    },
    {
      "epoch": 1.2512120691705086,
      "grad_norm": 0.9675731062889099,
      "learning_rate": 3.14742644608282e-05,
      "loss": 1.0383563041687012,
      "step": 11840
    },
    {
      "epoch": 1.2522689142238135,
      "grad_norm": 1.239517331123352,
      "learning_rate": 3.144751988275579e-05,
      "loss": 1.0370558738708495,
      "step": 11850
    },
    {
      "epoch": 1.2533257592771179,
      "grad_norm": 1.0913103818893433,
      "learning_rate": 3.142076739781815e-05,
      "loss": 0.9229194641113281,
      "step": 11860
    },
    {
      "epoch": 1.2543826043304227,
      "grad_norm": 0.8303108811378479,
      "learning_rate": 3.139400703882298e-05,
      "loss": 1.034541416168213,
      "step": 11870
    },
    {
      "epoch": 1.255439449383727,
      "grad_norm": 1.6862454414367676,
      "learning_rate": 3.136723883858763e-05,
      "loss": 0.9420527458190918,
      "step": 11880
    },
    {
      "epoch": 1.256496294437032,
      "grad_norm": 1.7413160800933838,
      "learning_rate": 3.1340462829939076e-05,
      "loss": 0.9840489387512207,
      "step": 11890
    },
    {
      "epoch": 1.2575531394903365,
      "grad_norm": 0.963922917842865,
      "learning_rate": 3.1313679045713874e-05,
      "loss": 0.9955129623413086,
      "step": 11900
    },
    {
      "epoch": 1.2586099845436411,
      "grad_norm": 0.9272423386573792,
      "learning_rate": 3.128688751875808e-05,
      "loss": 1.031690788269043,
      "step": 11910
    },
    {
      "epoch": 1.2596668295969458,
      "grad_norm": 0.8585636019706726,
      "learning_rate": 3.126008828192731e-05,
      "loss": 0.8772694587707519,
      "step": 11920
    },
    {
      "epoch": 1.2607236746502504,
      "grad_norm": 0.9889110326766968,
      "learning_rate": 3.123328136808656e-05,
      "loss": 0.960939598083496,
      "step": 11930
    },
    {
      "epoch": 1.261780519703555,
      "grad_norm": 1.3652293682098389,
      "learning_rate": 3.120646681011031e-05,
      "loss": 0.888133430480957,
      "step": 11940
    },
    {
      "epoch": 1.2628373647568596,
      "grad_norm": 1.1434203386306763,
      "learning_rate": 3.1179644640882356e-05,
      "loss": 0.9827966690063477,
      "step": 11950
    },
    {
      "epoch": 1.2638942098101642,
      "grad_norm": 1.0773773193359375,
      "learning_rate": 3.115281489329587e-05,
      "loss": 0.9343376159667969,
      "step": 11960
    },
    {
      "epoch": 1.2649510548634688,
      "grad_norm": 1.378223180770874,
      "learning_rate": 3.112597760025329e-05,
      "loss": 1.0460149765014648,
      "step": 11970
    },
    {
      "epoch": 1.2660078999167734,
      "grad_norm": 1.1242518424987793,
      "learning_rate": 3.1099132794666324e-05,
      "loss": 0.8921507835388184,
      "step": 11980
    },
    {
      "epoch": 1.267064744970078,
      "grad_norm": 1.7446564435958862,
      "learning_rate": 3.1072280509455883e-05,
      "loss": 1.0380349159240723,
      "step": 11990
    },
    {
      "epoch": 1.2681215900233826,
      "grad_norm": 1.8467313051223755,
      "learning_rate": 3.104542077755208e-05,
      "loss": 1.0719620704650878,
      "step": 12000
    },
    {
      "epoch": 1.2681215900233826,
      "eval_loss": 0.4989999830722809,
      "eval_runtime": 2973.473,
      "eval_samples_per_second": 2.995,
      "eval_steps_per_second": 2.995,
      "step": 12000
    },
    {
      "epoch": 1.2691784350766873,
      "grad_norm": 1.0043445825576782,
      "learning_rate": 3.101855363189411e-05,
      "loss": 0.9449195861816406,
      "step": 12010
    },
    {
      "epoch": 1.2702352801299919,
      "grad_norm": 1.0101310014724731,
      "learning_rate": 3.099167910543029e-05,
      "loss": 0.9580136299133301,
      "step": 12020
    },
    {
      "epoch": 1.2712921251832965,
      "grad_norm": 0.004622503649443388,
      "learning_rate": 3.0964797231117985e-05,
      "loss": 0.9546586036682129,
      "step": 12030
    },
    {
      "epoch": 1.2723489702366013,
      "grad_norm": 0.7505362033843994,
      "learning_rate": 3.093790804192358e-05,
      "loss": 0.9492700576782227,
      "step": 12040
    },
    {
      "epoch": 1.2734058152899057,
      "grad_norm": 1.4765963554382324,
      "learning_rate": 3.09110115708224e-05,
      "loss": 1.0406278610229491,
      "step": 12050
    },
    {
      "epoch": 1.2744626603432105,
      "grad_norm": 0.9391977190971375,
      "learning_rate": 3.088410785079874e-05,
      "loss": 0.9970436096191406,
      "step": 12060
    },
    {
      "epoch": 1.275519505396515,
      "grad_norm": 1.1520859003067017,
      "learning_rate": 3.085719691484575e-05,
      "loss": 1.022579288482666,
      "step": 12070
    },
    {
      "epoch": 1.2765763504498198,
      "grad_norm": 1.0914376974105835,
      "learning_rate": 3.083027879596546e-05,
      "loss": 1.0286978721618651,
      "step": 12080
    },
    {
      "epoch": 1.2776331955031244,
      "grad_norm": 1.0536584854125977,
      "learning_rate": 3.080335352716868e-05,
      "loss": 0.9635965347290039,
      "step": 12090
    },
    {
      "epoch": 1.278690040556429,
      "grad_norm": 1.0531929731369019,
      "learning_rate": 3.077642114147501e-05,
      "loss": 1.061149501800537,
      "step": 12100
    },
    {
      "epoch": 1.2797468856097336,
      "grad_norm": 1.200936198234558,
      "learning_rate": 3.074948167191276e-05,
      "loss": 0.9005508422851562,
      "step": 12110
    },
    {
      "epoch": 1.2808037306630382,
      "grad_norm": 1.000806450843811,
      "learning_rate": 3.0722535151518946e-05,
      "loss": 1.0646526336669921,
      "step": 12120
    },
    {
      "epoch": 1.2818605757163428,
      "grad_norm": 0.9455043077468872,
      "learning_rate": 3.0695581613339206e-05,
      "loss": 1.0011459350585938,
      "step": 12130
    },
    {
      "epoch": 1.2829174207696474,
      "grad_norm": 0.8770301342010498,
      "learning_rate": 3.0668621090427825e-05,
      "loss": 1.0637545585632324,
      "step": 12140
    },
    {
      "epoch": 1.283974265822952,
      "grad_norm": 0.9280707240104675,
      "learning_rate": 3.06416536158476e-05,
      "loss": 0.9697756767272949,
      "step": 12150
    },
    {
      "epoch": 1.2850311108762567,
      "grad_norm": 1.1433444023132324,
      "learning_rate": 3.0614679222669904e-05,
      "loss": 0.9156632423400879,
      "step": 12160
    },
    {
      "epoch": 1.2860879559295613,
      "grad_norm": 1.7080293893814087,
      "learning_rate": 3.058769794397456e-05,
      "loss": 0.8229113578796386,
      "step": 12170
    },
    {
      "epoch": 1.2871448009828659,
      "grad_norm": 0.9211262464523315,
      "learning_rate": 3.056070981284985e-05,
      "loss": 0.9508152008056641,
      "step": 12180
    },
    {
      "epoch": 1.2882016460361705,
      "grad_norm": 1.5991737842559814,
      "learning_rate": 3.053371486239245e-05,
      "loss": 1.0233508110046388,
      "step": 12190
    },
    {
      "epoch": 1.289258491089475,
      "grad_norm": 1.0919846296310425,
      "learning_rate": 3.050671312570741e-05,
      "loss": 0.8718077659606933,
      "step": 12200
    },
    {
      "epoch": 1.2903153361427797,
      "grad_norm": 0.8710769414901733,
      "learning_rate": 3.0479704635908098e-05,
      "loss": 0.820377254486084,
      "step": 12210
    },
    {
      "epoch": 1.2913721811960843,
      "grad_norm": 0.9522961378097534,
      "learning_rate": 3.045268942611616e-05,
      "loss": 1.0752711296081543,
      "step": 12220
    },
    {
      "epoch": 1.292429026249389,
      "grad_norm": 1.286372423171997,
      "learning_rate": 3.0425667529461483e-05,
      "loss": 0.9705039024353027,
      "step": 12230
    },
    {
      "epoch": 1.2934858713026935,
      "grad_norm": 1.7977975606918335,
      "learning_rate": 3.0398638979082166e-05,
      "loss": 0.991518211364746,
      "step": 12240
    },
    {
      "epoch": 1.2945427163559984,
      "grad_norm": 1.0053274631500244,
      "learning_rate": 3.037160380812445e-05,
      "loss": 0.9685774803161621,
      "step": 12250
    },
    {
      "epoch": 1.2955995614093028,
      "grad_norm": 1.6505308151245117,
      "learning_rate": 3.0344562049742715e-05,
      "loss": 1.0017837524414062,
      "step": 12260
    },
    {
      "epoch": 1.2966564064626076,
      "grad_norm": 1.048444390296936,
      "learning_rate": 3.0317513737099392e-05,
      "loss": 1.0016792297363282,
      "step": 12270
    },
    {
      "epoch": 1.297713251515912,
      "grad_norm": NaN,
      "learning_rate": 3.029045890336498e-05,
      "loss": 0.9765256881713867,
      "step": 12280
    },
    {
      "epoch": 1.2987700965692168,
      "grad_norm": 1.1300727128982544,
      "learning_rate": 3.0263397581717956e-05,
      "loss": 1.0391545295715332,
      "step": 12290
    },
    {
      "epoch": 1.2998269416225214,
      "grad_norm": 1.1742908954620361,
      "learning_rate": 3.0236329805344764e-05,
      "loss": 1.0786176681518556,
      "step": 12300
    },
    {
      "epoch": 1.300883786675826,
      "grad_norm": 1.1682708263397217,
      "learning_rate": 3.020925560743975e-05,
      "loss": 1.1216656684875488,
      "step": 12310
    },
    {
      "epoch": 1.3019406317291307,
      "grad_norm": 0.9234867691993713,
      "learning_rate": 3.018217502120515e-05,
      "loss": 1.051300621032715,
      "step": 12320
    },
    {
      "epoch": 1.3029974767824353,
      "grad_norm": 1.0599313974380493,
      "learning_rate": 3.0155088079851018e-05,
      "loss": 0.9394580841064453,
      "step": 12330
    },
    {
      "epoch": 1.3040543218357399,
      "grad_norm": 0.8995166420936584,
      "learning_rate": 3.012799481659523e-05,
      "loss": 1.0040600776672364,
      "step": 12340
    },
    {
      "epoch": 1.3051111668890445,
      "grad_norm": 1.2684423923492432,
      "learning_rate": 3.010089526466337e-05,
      "loss": 0.9638954162597656,
      "step": 12350
    },
    {
      "epoch": 1.306168011942349,
      "grad_norm": 1.7809553146362305,
      "learning_rate": 3.007378945728879e-05,
      "loss": 0.9877130508422851,
      "step": 12360
    },
    {
      "epoch": 1.3072248569956537,
      "grad_norm": 1.868577241897583,
      "learning_rate": 3.004667742771245e-05,
      "loss": 0.9525904655456543,
      "step": 12370
    },
    {
      "epoch": 1.3082817020489583,
      "grad_norm": 1.4532079696655273,
      "learning_rate": 3.0019559209183007e-05,
      "loss": 1.0894453048706054,
      "step": 12380
    },
    {
      "epoch": 1.309338547102263,
      "grad_norm": 1.1547967195510864,
      "learning_rate": 2.9992434834956647e-05,
      "loss": 0.9385774612426758,
      "step": 12390
    },
    {
      "epoch": 1.3103953921555676,
      "grad_norm": 1.07379949092865,
      "learning_rate": 2.996530433829715e-05,
      "loss": 1.0429876327514649,
      "step": 12400
    },
    {
      "epoch": 1.3114522372088722,
      "grad_norm": 1.0044845342636108,
      "learning_rate": 2.9938167752475783e-05,
      "loss": 1.0041521072387696,
      "step": 12410
    },
    {
      "epoch": 1.3125090822621768,
      "grad_norm": 1.419339656829834,
      "learning_rate": 2.9911025110771283e-05,
      "loss": 0.9595985412597656,
      "step": 12420
    },
    {
      "epoch": 1.3135659273154814,
      "grad_norm": 1.2143231630325317,
      "learning_rate": 2.9883876446469818e-05,
      "loss": 0.9343497276306152,
      "step": 12430
    },
    {
      "epoch": 1.3146227723687862,
      "grad_norm": 1.1605095863342285,
      "learning_rate": 2.9856721792864932e-05,
      "loss": 0.9635190963745117,
      "step": 12440
    },
    {
      "epoch": 1.3156796174220906,
      "grad_norm": 1.0209376811981201,
      "learning_rate": 2.9829561183257526e-05,
      "loss": 0.9634235382080079,
      "step": 12450
    },
    {
      "epoch": 1.3167364624753954,
      "grad_norm": 1.039394736289978,
      "learning_rate": 2.98023946509558e-05,
      "loss": 0.9907837867736816,
      "step": 12460
    },
    {
      "epoch": 1.3177933075286998,
      "grad_norm": 1.3528907299041748,
      "learning_rate": 2.9775222229275206e-05,
      "loss": 0.978056526184082,
      "step": 12470
    },
    {
      "epoch": 1.3188501525820047,
      "grad_norm": 1.325399398803711,
      "learning_rate": 2.974804395153845e-05,
      "loss": 0.9576947212219238,
      "step": 12480
    },
    {
      "epoch": 1.3199069976353093,
      "grad_norm": 0.9909560680389404,
      "learning_rate": 2.9720859851075378e-05,
      "loss": 0.9806527137756348,
      "step": 12490
    },
    {
      "epoch": 1.3209638426886139,
      "grad_norm": 1.2511221170425415,
      "learning_rate": 2.9693669961223013e-05,
      "loss": 1.0141329765319824,
      "step": 12500
    },
    {
      "epoch": 1.3209638426886139,
      "eval_loss": 0.4976560175418854,
      "eval_runtime": 3112.8287,
      "eval_samples_per_second": 2.861,
      "eval_steps_per_second": 2.861,
      "step": 12500
    },
    {
      "epoch": 1.3220206877419185,
      "grad_norm": 1.5076767206192017,
      "learning_rate": 2.9666474315325454e-05,
      "loss": 0.837836742401123,
      "step": 12510
    },
    {
      "epoch": 1.323077532795223,
      "grad_norm": 1.0528464317321777,
      "learning_rate": 2.963927294673387e-05,
      "loss": 0.9530182838439941,
      "step": 12520
    },
    {
      "epoch": 1.3241343778485277,
      "grad_norm": 0.9362907409667969,
      "learning_rate": 2.961206588880644e-05,
      "loss": 1.0613908767700195,
      "step": 12530
    },
    {
      "epoch": 1.3251912229018323,
      "grad_norm": 1.29265558719635,
      "learning_rate": 2.9584853174908333e-05,
      "loss": 0.9411553382873535,
      "step": 12540
    },
    {
      "epoch": 1.326248067955137,
      "grad_norm": 1.5129948854446411,
      "learning_rate": 2.9557634838411642e-05,
      "loss": 0.992014217376709,
      "step": 12550
    },
    {
      "epoch": 1.3273049130084416,
      "grad_norm": 1.6924667358398438,
      "learning_rate": 2.953041091269536e-05,
      "loss": 1.127364158630371,
      "step": 12560
    },
    {
      "epoch": 1.3283617580617462,
      "grad_norm": 1.3447213172912598,
      "learning_rate": 2.9503181431145344e-05,
      "loss": 1.0521799087524415,
      "step": 12570
    },
    {
      "epoch": 1.3294186031150508,
      "grad_norm": 0.8406025767326355,
      "learning_rate": 2.9475946427154232e-05,
      "loss": 0.9372285842895508,
      "step": 12580
    },
    {
      "epoch": 1.3304754481683554,
      "grad_norm": 0.9193592071533203,
      "learning_rate": 2.944870593412147e-05,
      "loss": 1.0496639251708983,
      "step": 12590
    },
    {
      "epoch": 1.33153229322166,
      "grad_norm": 0.009681480936706066,
      "learning_rate": 2.9421459985453226e-05,
      "loss": 0.8669826507568359,
      "step": 12600
    },
    {
      "epoch": 1.3325891382749646,
      "grad_norm": 1.0286856889724731,
      "learning_rate": 2.9394208614562352e-05,
      "loss": 0.9148401260375977,
      "step": 12610
    },
    {
      "epoch": 1.3336459833282692,
      "grad_norm": 1.3698532581329346,
      "learning_rate": 2.9366951854868345e-05,
      "loss": 0.9503305435180665,
      "step": 12620
    },
    {
      "epoch": 1.3347028283815738,
      "grad_norm": 1.3220117092132568,
      "learning_rate": 2.9339689739797325e-05,
      "loss": 0.9640748023986816,
      "step": 12630
    },
    {
      "epoch": 1.3357596734348784,
      "grad_norm": 1.0110541582107544,
      "learning_rate": 2.9312422302781974e-05,
      "loss": 1.0287860870361327,
      "step": 12640
    },
    {
      "epoch": 1.3368165184881833,
      "grad_norm": 0.9686492085456848,
      "learning_rate": 2.928514957726149e-05,
      "loss": 1.00208740234375,
      "step": 12650
    },
    {
      "epoch": 1.3378733635414877,
      "grad_norm": 1.021423101425171,
      "learning_rate": 2.9257871596681562e-05,
      "loss": 1.0256035804748536,
      "step": 12660
    },
    {
      "epoch": 1.3389302085947925,
      "grad_norm": 1.520031213760376,
      "learning_rate": 2.9230588394494345e-05,
      "loss": 1.0932630538940429,
      "step": 12670
    },
    {
      "epoch": 1.339987053648097,
      "grad_norm": 1.3187953233718872,
      "learning_rate": 2.920330000415836e-05,
      "loss": 1.0003758430480958,
      "step": 12680
    },
    {
      "epoch": 1.3410438987014017,
      "grad_norm": 1.3186508417129517,
      "learning_rate": 2.917600645913853e-05,
      "loss": 0.8638765335083007,
      "step": 12690
    },
    {
      "epoch": 1.3421007437547063,
      "grad_norm": 1.0406534671783447,
      "learning_rate": 2.914870779290606e-05,
      "loss": 0.9357864379882812,
      "step": 12700
    },
    {
      "epoch": 1.343157588808011,
      "grad_norm": 0.837749183177948,
      "learning_rate": 2.912140403893846e-05,
      "loss": 0.8501616477966308,
      "step": 12710
    },
    {
      "epoch": 1.3442144338613156,
      "grad_norm": 0.8668898940086365,
      "learning_rate": 2.9094095230719475e-05,
      "loss": 1.0323986053466796,
      "step": 12720
    },
    {
      "epoch": 1.3452712789146202,
      "grad_norm": 1.1722092628479004,
      "learning_rate": 2.9066781401739052e-05,
      "loss": 0.9820652961730957,
      "step": 12730
    },
    {
      "epoch": 1.3463281239679248,
      "grad_norm": 1.1448673009872437,
      "learning_rate": 2.9039462585493278e-05,
      "loss": 0.9758406639099121,
      "step": 12740
    },
    {
      "epoch": 1.3473849690212294,
      "grad_norm": 1.4592416286468506,
      "learning_rate": 2.901213881548438e-05,
      "loss": 1.0474366188049316,
      "step": 12750
    },
    {
      "epoch": 1.348441814074534,
      "grad_norm": 1.3101283311843872,
      "learning_rate": 2.898481012522064e-05,
      "loss": 0.985218620300293,
      "step": 12760
    },
    {
      "epoch": 1.3494986591278386,
      "grad_norm": 0.8609622120857239,
      "learning_rate": 2.895747654821639e-05,
      "loss": 0.9079687118530273,
      "step": 12770
    },
    {
      "epoch": 1.3505555041811432,
      "grad_norm": 1.3105452060699463,
      "learning_rate": 2.8930138117991933e-05,
      "loss": 0.9606691360473633,
      "step": 12780
    },
    {
      "epoch": 1.3516123492344478,
      "grad_norm": 1.1105401515960693,
      "learning_rate": 2.8902794868073558e-05,
      "loss": 1.0168540000915527,
      "step": 12790
    },
    {
      "epoch": 1.3526691942877525,
      "grad_norm": 1.032023310661316,
      "learning_rate": 2.887544683199342e-05,
      "loss": 0.9778794288635254,
      "step": 12800
    },
    {
      "epoch": 1.353726039341057,
      "grad_norm": 0.9796382784843445,
      "learning_rate": 2.8848094043289593e-05,
      "loss": 0.9430364608764649,
      "step": 12810
    },
    {
      "epoch": 1.3547828843943617,
      "grad_norm": 1.0539964437484741,
      "learning_rate": 2.8820736535505934e-05,
      "loss": 1.063535499572754,
      "step": 12820
    },
    {
      "epoch": 1.3558397294476663,
      "grad_norm": 0.9274889230728149,
      "learning_rate": 2.879337434219212e-05,
      "loss": 0.9502180099487305,
      "step": 12830
    },
    {
      "epoch": 1.356896574500971,
      "grad_norm": 0.9445379376411438,
      "learning_rate": 2.876600749690355e-05,
      "loss": 0.6779466152191163,
      "step": 12840
    },
    {
      "epoch": 1.3579534195542755,
      "grad_norm": 0.7578057646751404,
      "learning_rate": 2.873863603320135e-05,
      "loss": 0.9470339775085449,
      "step": 12850
    },
    {
      "epoch": 1.3590102646075803,
      "grad_norm": 1.2284084558486938,
      "learning_rate": 2.8711259984652293e-05,
      "loss": 0.9933061599731445,
      "step": 12860
    },
    {
      "epoch": 1.3600671096608847,
      "grad_norm": 1.2221896648406982,
      "learning_rate": 2.8683879384828783e-05,
      "loss": 0.9873261451721191,
      "step": 12870
    },
    {
      "epoch": 1.3611239547141896,
      "grad_norm": 0.9363458752632141,
      "learning_rate": 2.8656494267308798e-05,
      "loss": 1.0279909133911134,
      "step": 12880
    },
    {
      "epoch": 1.362180799767494,
      "grad_norm": 0.949638307094574,
      "learning_rate": 2.8629104665675872e-05,
      "loss": 1.1210532188415527,
      "step": 12890
    },
    {
      "epoch": 1.3632376448207988,
      "grad_norm": 1.0090981721878052,
      "learning_rate": 2.8601710613519013e-05,
      "loss": 0.9879040718078613,
      "step": 12900
    },
    {
      "epoch": 1.3642944898741034,
      "grad_norm": 1.2961561679840088,
      "learning_rate": 2.8574312144432712e-05,
      "loss": 1.0221685409545898,
      "step": 12910
    },
    {
      "epoch": 1.365351334927408,
      "grad_norm": 1.2903283834457397,
      "learning_rate": 2.854690929201686e-05,
      "loss": 0.8803214073181153,
      "step": 12920
    },
    {
      "epoch": 1.3664081799807126,
      "grad_norm": 1.7603809833526611,
      "learning_rate": 2.8519502089876737e-05,
      "loss": 0.9808445930480957,
      "step": 12930
    },
    {
      "epoch": 1.3674650250340172,
      "grad_norm": 1.7098790407180786,
      "learning_rate": 2.8492090571622936e-05,
      "loss": 0.8994461059570312,
      "step": 12940
    },
    {
      "epoch": 1.3685218700873218,
      "grad_norm": 1.4874464273452759,
      "learning_rate": 2.8464674770871368e-05,
      "loss": 0.9862920761108398,
      "step": 12950
    },
    {
      "epoch": 1.3695787151406265,
      "grad_norm": 1.7140120267868042,
      "learning_rate": 2.8437254721243178e-05,
      "loss": 1.053544044494629,
      "step": 12960
    },
    {
      "epoch": 1.370635560193931,
      "grad_norm": 1.3831058740615845,
      "learning_rate": 2.840983045636474e-05,
      "loss": 1.0090444564819336,
      "step": 12970
    },
    {
      "epoch": 1.3716924052472357,
      "grad_norm": 0.9055913686752319,
      "learning_rate": 2.8382402009867565e-05,
      "loss": 0.9606307983398438,
      "step": 12980
    },
    {
      "epoch": 1.3727492503005403,
      "grad_norm": 1.1635277271270752,
      "learning_rate": 2.8354969415388333e-05,
      "loss": 1.0462361335754395,
      "step": 12990
    },
    {
      "epoch": 1.373806095353845,
      "grad_norm": 1.6345858573913574,
      "learning_rate": 2.8327532706568773e-05,
      "loss": 0.9102317810058593,
      "step": 13000
    },
    {
      "epoch": 1.373806095353845,
      "eval_loss": 0.49603381752967834,
      "eval_runtime": 3112.383,
      "eval_samples_per_second": 2.861,
      "eval_steps_per_second": 2.861,
      "step": 13000
    },
    {
      "epoch": 1.3748629404071495,
      "grad_norm": 1.0874820947647095,
      "learning_rate": 2.8300091917055686e-05,
      "loss": 0.9850322723388671,
      "step": 13010
    },
    {
      "epoch": 1.3759197854604541,
      "grad_norm": 0.9994888305664062,
      "learning_rate": 2.8272647080500862e-05,
      "loss": 0.8016494750976563,
      "step": 13020
    },
    {
      "epoch": 1.3769766305137587,
      "grad_norm": 1.083375096321106,
      "learning_rate": 2.8245198230561077e-05,
      "loss": 0.9589960098266601,
      "step": 13030
    },
    {
      "epoch": 1.3780334755670633,
      "grad_norm": 1.1943925619125366,
      "learning_rate": 2.821774540089799e-05,
      "loss": 0.9983835220336914,
      "step": 13040
    },
    {
      "epoch": 1.3790903206203682,
      "grad_norm": 0.9452547430992126,
      "learning_rate": 2.8190288625178173e-05,
      "loss": 0.9672202110290528,
      "step": 13050
    },
    {
      "epoch": 1.3801471656736726,
      "grad_norm": 1.2200729846954346,
      "learning_rate": 2.8162827937073022e-05,
      "loss": 1.032448959350586,
      "step": 13060
    },
    {
      "epoch": 1.3812040107269774,
      "grad_norm": 0.9510646462440491,
      "learning_rate": 2.813536337025875e-05,
      "loss": 0.9761914253234864,
      "step": 13070
    },
    {
      "epoch": 1.3822608557802818,
      "grad_norm": 1.1493899822235107,
      "learning_rate": 2.810789495841629e-05,
      "loss": 1.040796661376953,
      "step": 13080
    },
    {
      "epoch": 1.3833177008335866,
      "grad_norm": 1.1624962091445923,
      "learning_rate": 2.8080422735231334e-05,
      "loss": 0.9178686141967773,
      "step": 13090
    },
    {
      "epoch": 1.3843745458868912,
      "grad_norm": 1.1089247465133667,
      "learning_rate": 2.8052946734394204e-05,
      "loss": 0.8613591194152832,
      "step": 13100
    },
    {
      "epoch": 1.3854313909401959,
      "grad_norm": 1.1791961193084717,
      "learning_rate": 2.8025466989599903e-05,
      "loss": 0.914364242553711,
      "step": 13110
    },
    {
      "epoch": 1.3864882359935005,
      "grad_norm": 1.7469711303710938,
      "learning_rate": 2.7997983534547977e-05,
      "loss": 0.9777909278869629,
      "step": 13120
    },
    {
      "epoch": 1.387545081046805,
      "grad_norm": 1.472291350364685,
      "learning_rate": 2.7970496402942563e-05,
      "loss": 0.9117387771606446,
      "step": 13130
    },
    {
      "epoch": 1.3886019261001097,
      "grad_norm": 1.2028942108154297,
      "learning_rate": 2.7943005628492274e-05,
      "loss": 1.0544869422912597,
      "step": 13140
    },
    {
      "epoch": 1.3896587711534143,
      "grad_norm": 0.8681358695030212,
      "learning_rate": 2.7915511244910213e-05,
      "loss": 0.928654956817627,
      "step": 13150
    },
    {
      "epoch": 1.390715616206719,
      "grad_norm": 1.2102211713790894,
      "learning_rate": 2.7888013285913888e-05,
      "loss": 0.8509919166564941,
      "step": 13160
    },
    {
      "epoch": 1.3917724612600235,
      "grad_norm": 1.1187399625778198,
      "learning_rate": 2.786051178522523e-05,
      "loss": 0.9375728607177735,
      "step": 13170
    },
    {
      "epoch": 1.3928293063133281,
      "grad_norm": 1.2524112462997437,
      "learning_rate": 2.7833006776570457e-05,
      "loss": 0.9738173484802246,
      "step": 13180
    },
    {
      "epoch": 1.3938861513666327,
      "grad_norm": 1.7928383350372314,
      "learning_rate": 2.7805498293680144e-05,
      "loss": 0.9367192268371582,
      "step": 13190
    },
    {
      "epoch": 1.3949429964199374,
      "grad_norm": 1.11674964427948,
      "learning_rate": 2.7777986370289088e-05,
      "loss": 1.0152222633361816,
      "step": 13200
    },
    {
      "epoch": 1.395999841473242,
      "grad_norm": 1.0085351467132568,
      "learning_rate": 2.7750471040136326e-05,
      "loss": 0.8898183822631835,
      "step": 13210
    },
    {
      "epoch": 1.3970566865265466,
      "grad_norm": 1.10589599609375,
      "learning_rate": 2.772295233696507e-05,
      "loss": 0.9593291282653809,
      "step": 13220
    },
    {
      "epoch": 1.3981135315798512,
      "grad_norm": 1.094484567642212,
      "learning_rate": 2.7695430294522657e-05,
      "loss": 1.0539199829101562,
      "step": 13230
    },
    {
      "epoch": 1.3991703766331558,
      "grad_norm": 0.9191016554832458,
      "learning_rate": 2.7667904946560526e-05,
      "loss": 1.0392595291137696,
      "step": 13240
    },
    {
      "epoch": 1.4002272216864604,
      "grad_norm": 1.125380277633667,
      "learning_rate": 2.7640376326834183e-05,
      "loss": 1.006352996826172,
      "step": 13250
    },
    {
      "epoch": 1.4012840667397652,
      "grad_norm": 1.2463370561599731,
      "learning_rate": 2.7612844469103117e-05,
      "loss": 0.9859341621398926,
      "step": 13260
    },
    {
      "epoch": 1.4023409117930696,
      "grad_norm": 1.2030493021011353,
      "learning_rate": 2.758530940713082e-05,
      "loss": 1.0505681037902832,
      "step": 13270
    },
    {
      "epoch": 1.4033977568463745,
      "grad_norm": 0.8734951019287109,
      "learning_rate": 2.755777117468469e-05,
      "loss": 0.859010124206543,
      "step": 13280
    },
    {
      "epoch": 1.4044546018996789,
      "grad_norm": 1.3471601009368896,
      "learning_rate": 2.7530229805536023e-05,
      "loss": 0.9805170059204101,
      "step": 13290
    },
    {
      "epoch": 1.4055114469529837,
      "grad_norm": 1.5719457864761353,
      "learning_rate": 2.750268533345996e-05,
      "loss": 0.8775164604187011,
      "step": 13300
    },
    {
      "epoch": 1.4065682920062883,
      "grad_norm": 1.0684454441070557,
      "learning_rate": 2.7475137792235457e-05,
      "loss": 1.0217350006103516,
      "step": 13310
    },
    {
      "epoch": 1.407625137059593,
      "grad_norm": 0.9529332518577576,
      "learning_rate": 2.7447587215645206e-05,
      "loss": 0.9078181266784668,
      "step": 13320
    },
    {
      "epoch": 1.4086819821128975,
      "grad_norm": 1.2451733350753784,
      "learning_rate": 2.742003363747565e-05,
      "loss": 1.0116129875183106,
      "step": 13330
    },
    {
      "epoch": 1.4097388271662021,
      "grad_norm": 0.9186391830444336,
      "learning_rate": 2.73924770915169e-05,
      "loss": 0.9860638618469239,
      "step": 13340
    },
    {
      "epoch": 1.4107956722195067,
      "grad_norm": 0.9930596351623535,
      "learning_rate": 2.7364917611562717e-05,
      "loss": 1.026687240600586,
      "step": 13350
    },
    {
      "epoch": 1.4118525172728114,
      "grad_norm": 0.8694455027580261,
      "learning_rate": 2.733735523141044e-05,
      "loss": 1.0380741119384767,
      "step": 13360
    },
    {
      "epoch": 1.412909362326116,
      "grad_norm": 1.1545166969299316,
      "learning_rate": 2.730978998486099e-05,
      "loss": 1.073397445678711,
      "step": 13370
    },
    {
      "epoch": 1.4139662073794206,
      "grad_norm": 1.3864213228225708,
      "learning_rate": 2.7282221905718773e-05,
      "loss": 0.9033360481262207,
      "step": 13380
    },
    {
      "epoch": 1.4150230524327252,
      "grad_norm": 0.8507371544837952,
      "learning_rate": 2.7254651027791704e-05,
      "loss": 0.8829507827758789,
      "step": 13390
    },
    {
      "epoch": 1.4160798974860298,
      "grad_norm": 0.9771236777305603,
      "learning_rate": 2.7227077384891096e-05,
      "loss": 1.0462319374084472,
      "step": 13400
    },
    {
      "epoch": 1.4171367425393344,
      "grad_norm": 0.9625804424285889,
      "learning_rate": 2.7199501010831687e-05,
      "loss": 1.0753205299377442,
      "step": 13410
    },
    {
      "epoch": 1.418193587592639,
      "grad_norm": 1.5133566856384277,
      "learning_rate": 2.717192193943152e-05,
      "loss": 0.924092960357666,
      "step": 13420
    },
    {
      "epoch": 1.4192504326459436,
      "grad_norm": 0.9601010084152222,
      "learning_rate": 2.7144340204512002e-05,
      "loss": 0.9999936103820801,
      "step": 13430
    },
    {
      "epoch": 1.4203072776992482,
      "grad_norm": 0.9849125146865845,
      "learning_rate": 2.7116755839897756e-05,
      "loss": 0.9956757545471191,
      "step": 13440
    },
    {
      "epoch": 1.4213641227525529,
      "grad_norm": 1.0696643590927124,
      "learning_rate": 2.708916887941666e-05,
      "loss": 1.0057406425476074,
      "step": 13450
    },
    {
      "epoch": 1.4224209678058575,
      "grad_norm": 1.1482635736465454,
      "learning_rate": 2.706157935689976e-05,
      "loss": 0.9493826866149903,
      "step": 13460
    },
    {
      "epoch": 1.4234778128591623,
      "grad_norm": 1.3135806322097778,
      "learning_rate": 2.7033987306181245e-05,
      "loss": 0.9762866020202636,
      "step": 13470
    },
    {
      "epoch": 1.4245346579124667,
      "grad_norm": 0.8799996376037598,
      "learning_rate": 2.700639276109842e-05,
      "loss": 0.9753519058227539,
      "step": 13480
    },
    {
      "epoch": 1.4255915029657715,
      "grad_norm": 1.0511465072631836,
      "learning_rate": 2.6978795755491636e-05,
      "loss": 0.9018738746643067,
      "step": 13490
    },
    {
      "epoch": 1.426648348019076,
      "grad_norm": 1.477697491645813,
      "learning_rate": 2.6951196323204265e-05,
      "loss": 1.0300671577453613,
      "step": 13500
    },
    {
      "epoch": 1.426648348019076,
      "eval_loss": 0.49512147903442383,
      "eval_runtime": 3119.375,
      "eval_samples_per_second": 2.855,
      "eval_steps_per_second": 2.855,
      "step": 13500
    },
    {
      "epoch": 1.4277051930723808,
      "grad_norm": 0.898289680480957,
      "learning_rate": 2.6923594498082645e-05,
      "loss": 0.9204854011535645,
      "step": 13510
    },
    {
      "epoch": 1.4287620381256854,
      "grad_norm": 1.053495168685913,
      "learning_rate": 2.6895990313976056e-05,
      "loss": 1.023977565765381,
      "step": 13520
    },
    {
      "epoch": 1.42981888317899,
      "grad_norm": 1.0286568403244019,
      "learning_rate": 2.6868383804736684e-05,
      "loss": 1.0242836952209473,
      "step": 13530
    },
    {
      "epoch": 1.4308757282322946,
      "grad_norm": 1.3398751020431519,
      "learning_rate": 2.684077500421955e-05,
      "loss": 1.0319798469543457,
      "step": 13540
    },
    {
      "epoch": 1.4319325732855992,
      "grad_norm": 1.2082726955413818,
      "learning_rate": 2.6813163946282483e-05,
      "loss": 1.0767157554626465,
      "step": 13550
    },
    {
      "epoch": 1.4329894183389038,
      "grad_norm": 1.5876312255859375,
      "learning_rate": 2.6785550664786087e-05,
      "loss": 0.982172966003418,
      "step": 13560
    },
    {
      "epoch": 1.4340462633922084,
      "grad_norm": 0.885707676410675,
      "learning_rate": 2.6757935193593696e-05,
      "loss": 1.1352093696594239,
      "step": 13570
    },
    {
      "epoch": 1.435103108445513,
      "grad_norm": 0.9432156682014465,
      "learning_rate": 2.6730317566571326e-05,
      "loss": 1.0344559669494628,
      "step": 13580
    },
    {
      "epoch": 1.4361599534988176,
      "grad_norm": 1.5523972511291504,
      "learning_rate": 2.6702697817587642e-05,
      "loss": 0.9791482925415039,
      "step": 13590
    },
    {
      "epoch": 1.4372167985521223,
      "grad_norm": 1.251367449760437,
      "learning_rate": 2.667507598051389e-05,
      "loss": 1.0065048217773438,
      "step": 13600
    },
    {
      "epoch": 1.4382736436054269,
      "grad_norm": 1.8895840644836426,
      "learning_rate": 2.6647452089223907e-05,
      "loss": 0.7605237483978271,
      "step": 13610
    },
    {
      "epoch": 1.4393304886587315,
      "grad_norm": 1.138325810432434,
      "learning_rate": 2.6619826177594027e-05,
      "loss": 1.0083083152770995,
      "step": 13620
    },
    {
      "epoch": 1.440387333712036,
      "grad_norm": 1.3049342632293701,
      "learning_rate": 2.6592198279503072e-05,
      "loss": 0.9150720596313476,
      "step": 13630
    },
    {
      "epoch": 1.4414441787653407,
      "grad_norm": 0.9175580739974976,
      "learning_rate": 2.6564568428832292e-05,
      "loss": 0.9515934944152832,
      "step": 13640
    },
    {
      "epoch": 1.4425010238186453,
      "grad_norm": 1.656027913093567,
      "learning_rate": 2.6536936659465343e-05,
      "loss": 0.9798171043395996,
      "step": 13650
    },
    {
      "epoch": 1.4435578688719501,
      "grad_norm": 0.9426175951957703,
      "learning_rate": 2.650930300528823e-05,
      "loss": 1.033842182159424,
      "step": 13660
    },
    {
      "epoch": 1.4446147139252545,
      "grad_norm": 1.6465877294540405,
      "learning_rate": 2.6481667500189256e-05,
      "loss": 0.9370835304260254,
      "step": 13670
    },
    {
      "epoch": 1.4456715589785594,
      "grad_norm": 0.9934461116790771,
      "learning_rate": 2.645403017805902e-05,
      "loss": 1.020310878753662,
      "step": 13680
    },
    {
      "epoch": 1.4467284040318638,
      "grad_norm": 0.9277760982513428,
      "learning_rate": 2.6426391072790326e-05,
      "loss": 0.997783088684082,
      "step": 13690
    },
    {
      "epoch": 1.4477852490851686,
      "grad_norm": 0.8519477248191833,
      "learning_rate": 2.6398750218278166e-05,
      "loss": 0.7061667919158936,
      "step": 13700
    },
    {
      "epoch": 1.4488420941384732,
      "grad_norm": 1.5990608930587769,
      "learning_rate": 2.6371107648419698e-05,
      "loss": 0.9040376663208007,
      "step": 13710
    },
    {
      "epoch": 1.4498989391917778,
      "grad_norm": 1.0683704614639282,
      "learning_rate": 2.634346339711416e-05,
      "loss": 0.864776611328125,
      "step": 13720
    },
    {
      "epoch": 1.4509557842450824,
      "grad_norm": 1.1445708274841309,
      "learning_rate": 2.631581749826287e-05,
      "loss": 1.0713841438293457,
      "step": 13730
    },
    {
      "epoch": 1.452012629298387,
      "grad_norm": 1.1435880661010742,
      "learning_rate": 2.6288169985769147e-05,
      "loss": 1.0360491752624512,
      "step": 13740
    },
    {
      "epoch": 1.4530694743516916,
      "grad_norm": 1.35905122756958,
      "learning_rate": 2.626052089353831e-05,
      "loss": 0.9811629295349121,
      "step": 13750
    },
    {
      "epoch": 1.4541263194049963,
      "grad_norm": 1.1251474618911743,
      "learning_rate": 2.6232870255477597e-05,
      "loss": 1.0189936637878418,
      "step": 13760
    },
    {
      "epoch": 1.4551831644583009,
      "grad_norm": 1.0899848937988281,
      "learning_rate": 2.6205218105496165e-05,
      "loss": 0.9147820472717285,
      "step": 13770
    },
    {
      "epoch": 1.4562400095116055,
      "grad_norm": 1.0635974407196045,
      "learning_rate": 2.6177564477504995e-05,
      "loss": 1.072320556640625,
      "step": 13780
    },
    {
      "epoch": 1.45729685456491,
      "grad_norm": 1.0795798301696777,
      "learning_rate": 2.6149909405416907e-05,
      "loss": 0.7479659080505371,
      "step": 13790
    },
    {
      "epoch": 1.4583536996182147,
      "grad_norm": 1.3274283409118652,
      "learning_rate": 2.6122252923146468e-05,
      "loss": 0.9929688453674317,
      "step": 13800
    },
    {
      "epoch": 1.4594105446715193,
      "grad_norm": 0.937099039554596,
      "learning_rate": 2.609459506461e-05,
      "loss": 0.9468090057373046,
      "step": 13810
    },
    {
      "epoch": 1.460467389724824,
      "grad_norm": 1.1674046516418457,
      "learning_rate": 2.6066935863725495e-05,
      "loss": 0.972877025604248,
      "step": 13820
    },
    {
      "epoch": 1.4615242347781285,
      "grad_norm": 1.0470142364501953,
      "learning_rate": 2.6039275354412597e-05,
      "loss": 0.9971712112426758,
      "step": 13830
    },
    {
      "epoch": 1.4625810798314332,
      "grad_norm": 1.1691310405731201,
      "learning_rate": 2.6011613570592546e-05,
      "loss": 0.9961535453796386,
      "step": 13840
    },
    {
      "epoch": 1.4636379248847378,
      "grad_norm": 1.1947191953659058,
      "learning_rate": 2.598395054618817e-05,
      "loss": 1.010035228729248,
      "step": 13850
    },
    {
      "epoch": 1.4646947699380424,
      "grad_norm": 0.8338949084281921,
      "learning_rate": 2.5956286315123784e-05,
      "loss": 1.0333810806274415,
      "step": 13860
    },
    {
      "epoch": 1.4657516149913472,
      "grad_norm": 1.3628466129302979,
      "learning_rate": 2.5928620911325214e-05,
      "loss": 0.9948605537414551,
      "step": 13870
    },
    {
      "epoch": 1.4668084600446516,
      "grad_norm": 0.8907881379127502,
      "learning_rate": 2.5900954368719692e-05,
      "loss": 0.9734779357910156,
      "step": 13880
    },
    {
      "epoch": 1.4678653050979564,
      "grad_norm": 1.1075876951217651,
      "learning_rate": 2.5873286721235883e-05,
      "loss": 0.939908504486084,
      "step": 13890
    },
    {
      "epoch": 1.4689221501512608,
      "grad_norm": 1.3091192245483398,
      "learning_rate": 2.5845618002803772e-05,
      "loss": 0.9021394729614258,
      "step": 13900
    },
    {
      "epoch": 1.4699789952045657,
      "grad_norm": 1.001982569694519,
      "learning_rate": 2.5817948247354695e-05,
      "loss": 0.918525218963623,
      "step": 13910
    },
    {
      "epoch": 1.4710358402578703,
      "grad_norm": 1.0917272567749023,
      "learning_rate": 2.579027748882122e-05,
      "loss": 0.9614727020263671,
      "step": 13920
    },
    {
      "epoch": 1.4720926853111749,
      "grad_norm": 1.0405288934707642,
      "learning_rate": 2.5762605761137176e-05,
      "loss": 0.9456025123596191,
      "step": 13930
    },
    {
      "epoch": 1.4731495303644795,
      "grad_norm": 0.8758184313774109,
      "learning_rate": 2.573493309823755e-05,
      "loss": 0.9341780662536621,
      "step": 13940
    },
    {
      "epoch": 1.474206375417784,
      "grad_norm": 1.321871280670166,
      "learning_rate": 2.570725953405852e-05,
      "loss": 1.0446890830993651,
      "step": 13950
    },
    {
      "epoch": 1.4752632204710887,
      "grad_norm": 0.7686223983764648,
      "learning_rate": 2.5679585102537322e-05,
      "loss": 0.9144983291625977,
      "step": 13960
    },
    {
      "epoch": 1.4763200655243933,
      "grad_norm": 1.1220104694366455,
      "learning_rate": 2.5651909837612293e-05,
      "loss": 1.0028388977050782,
      "step": 13970
    },
    {
      "epoch": 1.477376910577698,
      "grad_norm": 1.030059814453125,
      "learning_rate": 2.5624233773222762e-05,
      "loss": 1.0220446586608887,
      "step": 13980
    },
    {
      "epoch": 1.4784337556310025,
      "grad_norm": 1.704398512840271,
      "learning_rate": 2.5596556943309068e-05,
      "loss": 0.8279314994812011,
      "step": 13990
    },
    {
      "epoch": 1.4794906006843072,
      "grad_norm": 0.9655267000198364,
      "learning_rate": 2.5568879381812455e-05,
      "loss": 0.8989787101745605,
      "step": 14000
    },
    {
      "epoch": 1.4794906006843072,
      "eval_loss": 0.4934263229370117,
      "eval_runtime": 3112.5765,
      "eval_samples_per_second": 2.861,
      "eval_steps_per_second": 2.861,
      "step": 14000
    },
    {
      "epoch": 1.4805474457376118,
      "grad_norm": 1.1239063739776611,
      "learning_rate": 2.5541201122675102e-05,
      "loss": 0.8699859619140625,
      "step": 14010
    },
    {
      "epoch": 1.4816042907909164,
      "grad_norm": 1.2694112062454224,
      "learning_rate": 2.551352219984e-05,
      "loss": 0.9880992889404296,
      "step": 14020
    },
    {
      "epoch": 1.482661135844221,
      "grad_norm": 1.0786352157592773,
      "learning_rate": 2.5485842647251008e-05,
      "loss": 0.9780531883239746,
      "step": 14030
    },
    {
      "epoch": 1.4837179808975256,
      "grad_norm": 1.1692396402359009,
      "learning_rate": 2.54581624988527e-05,
      "loss": 0.8421342849731446,
      "step": 14040
    },
    {
      "epoch": 1.4847748259508302,
      "grad_norm": 1.1927731037139893,
      "learning_rate": 2.5430481788590426e-05,
      "loss": 0.9964051246643066,
      "step": 14050
    },
    {
      "epoch": 1.4858316710041348,
      "grad_norm": 1.2258387804031372,
      "learning_rate": 2.54028005504102e-05,
      "loss": 0.9121529579162597,
      "step": 14060
    },
    {
      "epoch": 1.4868885160574394,
      "grad_norm": 1.4006210565567017,
      "learning_rate": 2.5375118818258692e-05,
      "loss": 1.0169410705566406,
      "step": 14070
    },
    {
      "epoch": 1.4879453611107443,
      "grad_norm": 1.1763745546340942,
      "learning_rate": 2.5347436626083172e-05,
      "loss": 1.0245244026184082,
      "step": 14080
    },
    {
      "epoch": 1.4890022061640487,
      "grad_norm": 1.15553617477417,
      "learning_rate": 2.5319754007831486e-05,
      "loss": 0.9901868820190429,
      "step": 14090
    },
    {
      "epoch": 1.4900590512173535,
      "grad_norm": 1.5441700220108032,
      "learning_rate": 2.5292070997451988e-05,
      "loss": 0.9352835655212403,
      "step": 14100
    },
    {
      "epoch": 1.4911158962706579,
      "grad_norm": 1.716496467590332,
      "learning_rate": 2.5264387628893526e-05,
      "loss": 0.9851072311401368,
      "step": 14110
    },
    {
      "epoch": 1.4921727413239627,
      "grad_norm": 1.0329769849777222,
      "learning_rate": 2.523670393610537e-05,
      "loss": 0.9595724105834961,
      "step": 14120
    },
    {
      "epoch": 1.4932295863772673,
      "grad_norm": 1.305996298789978,
      "learning_rate": 2.5209019953037217e-05,
      "loss": 1.0069084167480469,
      "step": 14130
    },
    {
      "epoch": 1.494286431430572,
      "grad_norm": 1.0824334621429443,
      "learning_rate": 2.5181335713639083e-05,
      "loss": 1.04288330078125,
      "step": 14140
    },
    {
      "epoch": 1.4953432764838765,
      "grad_norm": 0.9529021978378296,
      "learning_rate": 2.5153651251861325e-05,
      "loss": 1.044207763671875,
      "step": 14150
    },
    {
      "epoch": 1.4964001215371812,
      "grad_norm": 1.5099083185195923,
      "learning_rate": 2.5125966601654567e-05,
      "loss": 0.8884766578674317,
      "step": 14160
    },
    {
      "epoch": 1.4974569665904858,
      "grad_norm": 1.5406105518341064,
      "learning_rate": 2.509828179696966e-05,
      "loss": 0.8623110771179199,
      "step": 14170
    },
    {
      "epoch": 1.4985138116437904,
      "grad_norm": 1.2146086692810059,
      "learning_rate": 2.507059687175764e-05,
      "loss": 0.9416168212890625,
      "step": 14180
    },
    {
      "epoch": 1.499570656697095,
      "grad_norm": 1.6035418510437012,
      "learning_rate": 2.504291185996971e-05,
      "loss": 0.9790706634521484,
      "step": 14190
    },
    {
      "epoch": 1.5006275017503996,
      "grad_norm": 1.2008086442947388,
      "learning_rate": 2.501522679555715e-05,
      "loss": 1.005690574645996,
      "step": 14200
    },
    {
      "epoch": 1.5016843468037042,
      "grad_norm": 1.098465919494629,
      "learning_rate": 2.4987541712471335e-05,
      "loss": 0.9864922523498535,
      "step": 14210
    },
    {
      "epoch": 1.5027411918570088,
      "grad_norm": 1.326710820198059,
      "learning_rate": 2.4959856644663645e-05,
      "loss": 0.9995342254638672,
      "step": 14220
    },
    {
      "epoch": 1.5037980369103134,
      "grad_norm": 1.0296269655227661,
      "learning_rate": 2.4932171626085435e-05,
      "loss": 0.9565994262695312,
      "step": 14230
    },
    {
      "epoch": 1.504854881963618,
      "grad_norm": 1.1887726783752441,
      "learning_rate": 2.4904486690688026e-05,
      "loss": 1.0055840492248536,
      "step": 14240
    },
    {
      "epoch": 1.5059117270169229,
      "grad_norm": 1.3150161504745483,
      "learning_rate": 2.4876801872422602e-05,
      "loss": 0.993474292755127,
      "step": 14250
    },
    {
      "epoch": 1.5069685720702273,
      "grad_norm": 0.994532585144043,
      "learning_rate": 2.484911720524024e-05,
      "loss": 0.8586830139160156,
      "step": 14260
    },
    {
      "epoch": 1.508025417123532,
      "grad_norm": 1.0039764642715454,
      "learning_rate": 2.4821432723091797e-05,
      "loss": 0.9698620796203613,
      "step": 14270
    },
    {
      "epoch": 1.5090822621768365,
      "grad_norm": 1.57463538646698,
      "learning_rate": 2.4793748459927934e-05,
      "loss": 0.884907341003418,
      "step": 14280
    },
    {
      "epoch": 1.5101391072301413,
      "grad_norm": 1.3926845788955688,
      "learning_rate": 2.476606444969902e-05,
      "loss": 0.88040132522583,
      "step": 14290
    },
    {
      "epoch": 1.5111959522834457,
      "grad_norm": 1.2808918952941895,
      "learning_rate": 2.473838072635513e-05,
      "loss": 1.0188097953796387,
      "step": 14300
    },
    {
      "epoch": 1.5122527973367506,
      "grad_norm": 0.9576188921928406,
      "learning_rate": 2.4710697323845967e-05,
      "loss": 0.9462577819824218,
      "step": 14310
    },
    {
      "epoch": 1.513309642390055,
      "grad_norm": 1.2590616941452026,
      "learning_rate": 2.4683014276120868e-05,
      "loss": 0.9118215560913085,
      "step": 14320
    },
    {
      "epoch": 1.5143664874433598,
      "grad_norm": 1.7673619985580444,
      "learning_rate": 2.4655331617128716e-05,
      "loss": 0.932167911529541,
      "step": 14330
    },
    {
      "epoch": 1.5154233324966644,
      "grad_norm": 1.2882215976715088,
      "learning_rate": 2.462764938081792e-05,
      "loss": 0.9241623878479004,
      "step": 14340
    },
    {
      "epoch": 1.516480177549969,
      "grad_norm": 0.9530185461044312,
      "learning_rate": 2.4599967601136374e-05,
      "loss": 0.975899600982666,
      "step": 14350
    },
    {
      "epoch": 1.5175370226032736,
      "grad_norm": 1.125974416732788,
      "learning_rate": 2.457228631203141e-05,
      "loss": 0.9446175575256348,
      "step": 14360
    },
    {
      "epoch": 1.5185938676565782,
      "grad_norm": 0.8658575415611267,
      "learning_rate": 2.4544605547449753e-05,
      "loss": 0.9801442146301269,
      "step": 14370
    },
    {
      "epoch": 1.5196507127098828,
      "grad_norm": 1.015976071357727,
      "learning_rate": 2.4516925341337504e-05,
      "loss": 0.988077449798584,
      "step": 14380
    },
    {
      "epoch": 1.5207075577631874,
      "grad_norm": 0.9415561556816101,
      "learning_rate": 2.4489245727640054e-05,
      "loss": 0.9707791328430175,
      "step": 14390
    },
    {
      "epoch": 1.521764402816492,
      "grad_norm": 1.1535972356796265,
      "learning_rate": 2.446156674030209e-05,
      "loss": 1.0617791175842286,
      "step": 14400
    },
    {
      "epoch": 1.5228212478697967,
      "grad_norm": 1.1551810503005981,
      "learning_rate": 2.4433888413267504e-05,
      "loss": 1.004469108581543,
      "step": 14410
    },
    {
      "epoch": 1.5238780929231013,
      "grad_norm": 1.1744545698165894,
      "learning_rate": 2.440621078047941e-05,
      "loss": 0.7939053535461426,
      "step": 14420
    },
    {
      "epoch": 1.524934937976406,
      "grad_norm": 1.0094341039657593,
      "learning_rate": 2.4378533875880044e-05,
      "loss": 0.7863187313079834,
      "step": 14430
    },
    {
      "epoch": 1.5259917830297105,
      "grad_norm": 1.2507113218307495,
      "learning_rate": 2.4350857733410772e-05,
      "loss": 0.9274443626403809,
      "step": 14440
    },
    {
      "epoch": 1.5270486280830151,
      "grad_norm": 1.018841028213501,
      "learning_rate": 2.4323182387012e-05,
      "loss": 1.1057291984558106,
      "step": 14450
    },
    {
      "epoch": 1.52810547313632,
      "grad_norm": 0.9743399024009705,
      "learning_rate": 2.429550787062318e-05,
      "loss": 0.9572390556335449,
      "step": 14460
    },
    {
      "epoch": 1.5291623181896243,
      "grad_norm": 0.9750691056251526,
      "learning_rate": 2.4267834218182736e-05,
      "loss": 0.9624860763549805,
      "step": 14470
    },
    {
      "epoch": 1.5302191632429292,
      "grad_norm": 1.0624510049819946,
      "learning_rate": 2.4240161463628033e-05,
      "loss": 0.8067160606384277,
      "step": 14480
    },
    {
      "epoch": 1.5312760082962336,
      "grad_norm": 1.3519049882888794,
      "learning_rate": 2.4212489640895332e-05,
      "loss": 0.9275388717651367,
      "step": 14490
    },
    {
      "epoch": 1.5323328533495384,
      "grad_norm": 1.243648886680603,
      "learning_rate": 2.4184818783919762e-05,
      "loss": 1.0046199798583983,
      "step": 14500
    },
    {
      "epoch": 1.5323328533495384,
      "eval_loss": 0.49223580956459045,
      "eval_runtime": 3053.0681,
      "eval_samples_per_second": 2.917,
      "eval_steps_per_second": 2.917,
      "step": 14500
    },
    {
      "epoch": 1.5333896984028428,
      "grad_norm": 1.4370043277740479,
      "learning_rate": 2.4157148926635248e-05,
      "loss": 0.9035457611083985,
      "step": 14510
    },
    {
      "epoch": 1.5344465434561476,
      "grad_norm": 1.6177852153778076,
      "learning_rate": 2.412948010297452e-05,
      "loss": 0.9266929626464844,
      "step": 14520
    },
    {
      "epoch": 1.535503388509452,
      "grad_norm": 1.0297484397888184,
      "learning_rate": 2.410181234686901e-05,
      "loss": 0.792657470703125,
      "step": 14530
    },
    {
      "epoch": 1.5365602335627568,
      "grad_norm": 1.0418305397033691,
      "learning_rate": 2.4074145692248847e-05,
      "loss": 0.9746110916137696,
      "step": 14540
    },
    {
      "epoch": 1.5376170786160615,
      "grad_norm": 0.9904856085777283,
      "learning_rate": 2.404648017304283e-05,
      "loss": 0.9777215957641602,
      "step": 14550
    },
    {
      "epoch": 1.538673923669366,
      "grad_norm": 1.2570757865905762,
      "learning_rate": 2.4018815823178343e-05,
      "loss": 0.8564304351806641,
      "step": 14560
    },
    {
      "epoch": 1.5397307687226707,
      "grad_norm": 1.0025792121887207,
      "learning_rate": 2.3991152676581343e-05,
      "loss": 0.9419879913330078,
      "step": 14570
    },
    {
      "epoch": 1.5407876137759753,
      "grad_norm": 1.1992599964141846,
      "learning_rate": 2.396349076717631e-05,
      "loss": 1.0001663208007812,
      "step": 14580
    },
    {
      "epoch": 1.54184445882928,
      "grad_norm": 1.4536199569702148,
      "learning_rate": 2.3935830128886218e-05,
      "loss": 0.9819825172424317,
      "step": 14590
    },
    {
      "epoch": 1.5429013038825845,
      "grad_norm": 0.9757686257362366,
      "learning_rate": 2.3908170795632458e-05,
      "loss": 1.0715882301330566,
      "step": 14600
    },
    {
      "epoch": 1.5439581489358891,
      "grad_norm": 1.170846939086914,
      "learning_rate": 2.388051280133485e-05,
      "loss": 0.882880687713623,
      "step": 14610
    },
    {
      "epoch": 1.5450149939891937,
      "grad_norm": 1.1465717554092407,
      "learning_rate": 2.385285617991155e-05,
      "loss": 0.9470605850219727,
      "step": 14620
    },
    {
      "epoch": 1.5460718390424983,
      "grad_norm": 1.4650954008102417,
      "learning_rate": 2.382520096527904e-05,
      "loss": 0.9640069007873535,
      "step": 14630
    },
    {
      "epoch": 1.547128684095803,
      "grad_norm": 1.1343480348587036,
      "learning_rate": 2.379754719135207e-05,
      "loss": 0.8617137908935547,
      "step": 14640
    },
    {
      "epoch": 1.5481855291491078,
      "grad_norm": 1.1373974084854126,
      "learning_rate": 2.3769894892043635e-05,
      "loss": 0.8291407585144043,
      "step": 14650
    },
    {
      "epoch": 1.5492423742024122,
      "grad_norm": 1.1479990482330322,
      "learning_rate": 2.374224410126491e-05,
      "loss": 0.9569210052490235,
      "step": 14660
    },
    {
      "epoch": 1.550299219255717,
      "grad_norm": 1.3664257526397705,
      "learning_rate": 2.3714594852925232e-05,
      "loss": 1.0602656364440919,
      "step": 14670
    },
    {
      "epoch": 1.5513560643090214,
      "grad_norm": 1.243080496788025,
      "learning_rate": 2.3686947180932025e-05,
      "loss": 0.9241955757141114,
      "step": 14680
    },
    {
      "epoch": 1.5524129093623262,
      "grad_norm": 1.0449622869491577,
      "learning_rate": 2.365930111919081e-05,
      "loss": 0.9484186172485352,
      "step": 14690
    },
    {
      "epoch": 1.5534697544156306,
      "grad_norm": 1.6673952341079712,
      "learning_rate": 2.36316567016051e-05,
      "loss": 0.934824275970459,
      "step": 14700
    },
    {
      "epoch": 1.5545265994689355,
      "grad_norm": 1.4593589305877686,
      "learning_rate": 2.3604013962076425e-05,
      "loss": 0.942042064666748,
      "step": 14710
    },
    {
      "epoch": 1.5555834445222398,
      "grad_norm": 1.3035603761672974,
      "learning_rate": 2.3576372934504227e-05,
      "loss": 1.0142813682556153,
      "step": 14720
    },
    {
      "epoch": 1.5566402895755447,
      "grad_norm": 1.2585322856903076,
      "learning_rate": 2.3548733652785877e-05,
      "loss": 0.9118974685668946,
      "step": 14730
    },
    {
      "epoch": 1.557697134628849,
      "grad_norm": 1.182294249534607,
      "learning_rate": 2.3521096150816578e-05,
      "loss": 0.9695131301879882,
      "step": 14740
    },
    {
      "epoch": 1.558753979682154,
      "grad_norm": 1.0862663984298706,
      "learning_rate": 2.3493460462489366e-05,
      "loss": 1.0133913040161133,
      "step": 14750
    },
    {
      "epoch": 1.5598108247354585,
      "grad_norm": 1.153545618057251,
      "learning_rate": 2.346582662169505e-05,
      "loss": 0.9311132431030273,
      "step": 14760
    },
    {
      "epoch": 1.5608676697887631,
      "grad_norm": 1.0027930736541748,
      "learning_rate": 2.343819466232218e-05,
      "loss": 1.042623233795166,
      "step": 14770
    },
    {
      "epoch": 1.5619245148420677,
      "grad_norm": 1.4574506282806396,
      "learning_rate": 2.341056461825698e-05,
      "loss": 0.906948184967041,
      "step": 14780
    },
    {
      "epoch": 1.5629813598953723,
      "grad_norm": 1.457961082458496,
      "learning_rate": 2.3382936523383347e-05,
      "loss": 1.0996670722961426,
      "step": 14790
    },
    {
      "epoch": 1.564038204948677,
      "grad_norm": 0.9610999822616577,
      "learning_rate": 2.335531041158277e-05,
      "loss": 0.8165182113647461,
      "step": 14800
    },
    {
      "epoch": 1.5650950500019816,
      "grad_norm": 1.1106327772140503,
      "learning_rate": 2.3327686316734322e-05,
      "loss": 0.9831228256225586,
      "step": 14810
    },
    {
      "epoch": 1.5661518950552862,
      "grad_norm": 1.0796129703521729,
      "learning_rate": 2.3300064272714582e-05,
      "loss": 0.891743278503418,
      "step": 14820
    },
    {
      "epoch": 1.5672087401085908,
      "grad_norm": 1.5491268634796143,
      "learning_rate": 2.327244431339764e-05,
      "loss": 0.9339691162109375,
      "step": 14830
    },
    {
      "epoch": 1.5682655851618954,
      "grad_norm": 1.3412647247314453,
      "learning_rate": 2.3244826472655006e-05,
      "loss": 0.825522518157959,
      "step": 14840
    },
    {
      "epoch": 1.5693224302152,
      "grad_norm": 0.9619156718254089,
      "learning_rate": 2.3217210784355605e-05,
      "loss": 0.9456214904785156,
      "step": 14850
    },
    {
      "epoch": 1.5703792752685048,
      "grad_norm": 1.3019708395004272,
      "learning_rate": 2.3189597282365715e-05,
      "loss": 0.9104877471923828,
      "step": 14860
    },
    {
      "epoch": 1.5714361203218092,
      "grad_norm": 1.0473719835281372,
      "learning_rate": 2.3161986000548946e-05,
      "loss": 0.8444808006286622,
      "step": 14870
    },
    {
      "epoch": 1.572492965375114,
      "grad_norm": 1.6697397232055664,
      "learning_rate": 2.3134376972766163e-05,
      "loss": 0.85548095703125,
      "step": 14880
    },
    {
      "epoch": 1.5735498104284185,
      "grad_norm": 1.0900558233261108,
      "learning_rate": 2.3106770232875495e-05,
      "loss": 1.007416343688965,
      "step": 14890
    },
    {
      "epoch": 1.5746066554817233,
      "grad_norm": 1.3618656396865845,
      "learning_rate": 2.3079165814732232e-05,
      "loss": 0.7454955577850342,
      "step": 14900
    },
    {
      "epoch": 1.5756635005350277,
      "grad_norm": 1.1298524141311646,
      "learning_rate": 2.3051563752188852e-05,
      "loss": 0.8755311965942383,
      "step": 14910
    },
    {
      "epoch": 1.5767203455883325,
      "grad_norm": 1.0251624584197998,
      "learning_rate": 2.3023964079094913e-05,
      "loss": 0.9421425819396972,
      "step": 14920
    },
    {
      "epoch": 1.577777190641637,
      "grad_norm": 1.0953928232192993,
      "learning_rate": 2.2996366829297073e-05,
      "loss": 1.0154877662658692,
      "step": 14930
    },
    {
      "epoch": 1.5788340356949417,
      "grad_norm": 0.9726709127426147,
      "learning_rate": 2.2968772036638986e-05,
      "loss": 0.9537530899047851,
      "step": 14940
    },
    {
      "epoch": 1.5798908807482464,
      "grad_norm": 1.185957908630371,
      "learning_rate": 2.294117973496132e-05,
      "loss": 0.9403071403503418,
      "step": 14950
    },
    {
      "epoch": 1.580947725801551,
      "grad_norm": 0.8696361780166626,
      "learning_rate": 2.291358995810166e-05,
      "loss": 0.9159477233886719,
      "step": 14960
    },
    {
      "epoch": 1.5820045708548556,
      "grad_norm": 1.5825350284576416,
      "learning_rate": 2.2886002739894534e-05,
      "loss": 0.9750908851623535,
      "step": 14970
    },
    {
      "epoch": 1.5830614159081602,
      "grad_norm": 1.1101535558700562,
      "learning_rate": 2.2858418114171297e-05,
      "loss": 0.9421724319458008,
      "step": 14980
    },
    {
      "epoch": 1.5841182609614648,
      "grad_norm": 0.7684609293937683,
      "learning_rate": 2.283083611476014e-05,
      "loss": 1.053288173675537,
      "step": 14990
    },
    {
      "epoch": 1.5851751060147694,
      "grad_norm": 1.149171233177185,
      "learning_rate": 2.2803256775486024e-05,
      "loss": 0.8760776519775391,
      "step": 15000
    },
    {
      "epoch": 1.5851751060147694,
      "eval_loss": 0.4909125864505768,
      "eval_runtime": 3059.2719,
      "eval_samples_per_second": 2.911,
      "eval_steps_per_second": 2.911,
      "step": 15000
    },
    {
      "epoch": 1.586231951068074,
      "grad_norm": 1.5201232433319092,
      "learning_rate": 2.2775680130170673e-05,
      "loss": 0.909596061706543,
      "step": 15010
    },
    {
      "epoch": 1.5872887961213786,
      "grad_norm": 0.9772037267684937,
      "learning_rate": 2.2748106212632466e-05,
      "loss": 0.9660028457641602,
      "step": 15020
    },
    {
      "epoch": 1.5883456411746832,
      "grad_norm": 0.9582693576812744,
      "learning_rate": 2.272053505668648e-05,
      "loss": 0.970238208770752,
      "step": 15030
    },
    {
      "epoch": 1.5894024862279879,
      "grad_norm": 1.7418413162231445,
      "learning_rate": 2.2692966696144366e-05,
      "loss": 0.8086073875427247,
      "step": 15040
    },
    {
      "epoch": 1.5904593312812927,
      "grad_norm": 0.9805629253387451,
      "learning_rate": 2.266540116481438e-05,
      "loss": 0.9041144371032714,
      "step": 15050
    },
    {
      "epoch": 1.591516176334597,
      "grad_norm": 1.4411168098449707,
      "learning_rate": 2.2637838496501287e-05,
      "loss": 0.9971260070800781,
      "step": 15060
    },
    {
      "epoch": 1.592573021387902,
      "grad_norm": 1.8252668380737305,
      "learning_rate": 2.2610278725006354e-05,
      "loss": 0.9893118858337402,
      "step": 15070
    },
    {
      "epoch": 1.5936298664412063,
      "grad_norm": 1.2312268018722534,
      "learning_rate": 2.2582721884127295e-05,
      "loss": 1.0505160331726073,
      "step": 15080
    },
    {
      "epoch": 1.5946867114945111,
      "grad_norm": 1.248013973236084,
      "learning_rate": 2.2555168007658194e-05,
      "loss": 1.0151949882507325,
      "step": 15090
    },
    {
      "epoch": 1.5957435565478155,
      "grad_norm": 1.0283071994781494,
      "learning_rate": 2.2527617129389576e-05,
      "loss": 0.990869140625,
      "step": 15100
    },
    {
      "epoch": 1.5968004016011204,
      "grad_norm": 0.8925349712371826,
      "learning_rate": 2.25000692831082e-05,
      "loss": 1.0431639671325683,
      "step": 15110
    },
    {
      "epoch": 1.5978572466544247,
      "grad_norm": 1.0664665699005127,
      "learning_rate": 2.2472524502597185e-05,
      "loss": 0.9856441497802735,
      "step": 15120
    },
    {
      "epoch": 1.5989140917077296,
      "grad_norm": 0.905748724937439,
      "learning_rate": 2.2444982821635827e-05,
      "loss": 0.936825180053711,
      "step": 15130
    },
    {
      "epoch": 1.599970936761034,
      "grad_norm": 1.4228750467300415,
      "learning_rate": 2.2417444273999678e-05,
      "loss": 0.9539938926696777,
      "step": 15140
    },
    {
      "epoch": 1.6010277818143388,
      "grad_norm": 1.0083496570587158,
      "learning_rate": 2.2389908893460394e-05,
      "loss": 1.0487512588500976,
      "step": 15150
    },
    {
      "epoch": 1.6020846268676434,
      "grad_norm": 0.9503002762794495,
      "learning_rate": 2.2362376713785793e-05,
      "loss": 0.9467906951904297,
      "step": 15160
    },
    {
      "epoch": 1.603141471920948,
      "grad_norm": 2.063251495361328,
      "learning_rate": 2.2334847768739724e-05,
      "loss": 0.9028276443481446,
      "step": 15170
    },
    {
      "epoch": 1.6041983169742526,
      "grad_norm": 1.059080719947815,
      "learning_rate": 2.2307322092082116e-05,
      "loss": 1.044071865081787,
      "step": 15180
    },
    {
      "epoch": 1.6052551620275572,
      "grad_norm": 1.0284655094146729,
      "learning_rate": 2.227979971756884e-05,
      "loss": 0.9510218620300293,
      "step": 15190
    },
    {
      "epoch": 1.6063120070808619,
      "grad_norm": 1.14297616481781,
      "learning_rate": 2.2252280678951777e-05,
      "loss": 0.983592414855957,
      "step": 15200
    },
    {
      "epoch": 1.6073688521341665,
      "grad_norm": 1.4292491674423218,
      "learning_rate": 2.2224765009978646e-05,
      "loss": 1.0177149772644043,
      "step": 15210
    },
    {
      "epoch": 1.608425697187471,
      "grad_norm": 0.8662945628166199,
      "learning_rate": 2.2197252744393103e-05,
      "loss": 1.0127575874328614,
      "step": 15220
    },
    {
      "epoch": 1.6094825422407757,
      "grad_norm": 1.1826201677322388,
      "learning_rate": 2.2169743915934568e-05,
      "loss": 0.9332745552062989,
      "step": 15230
    },
    {
      "epoch": 1.6105393872940803,
      "grad_norm": 1.0026191473007202,
      "learning_rate": 2.214223855833831e-05,
      "loss": 0.8983642578125,
      "step": 15240
    },
    {
      "epoch": 1.611596232347385,
      "grad_norm": 1.308311104774475,
      "learning_rate": 2.2114736705335274e-05,
      "loss": 0.9662628173828125,
      "step": 15250
    },
    {
      "epoch": 1.6126530774006897,
      "grad_norm": 0.8262240290641785,
      "learning_rate": 2.208723839065217e-05,
      "loss": 1.0383686065673827,
      "step": 15260
    },
    {
      "epoch": 1.6137099224539941,
      "grad_norm": 1.2308176755905151,
      "learning_rate": 2.205974364801131e-05,
      "loss": 0.8952901840209961,
      "step": 15270
    },
    {
      "epoch": 1.614766767507299,
      "grad_norm": 1.1960350275039673,
      "learning_rate": 2.203225251113069e-05,
      "loss": 0.9311738967895508,
      "step": 15280
    },
    {
      "epoch": 1.6158236125606034,
      "grad_norm": 1.266990065574646,
      "learning_rate": 2.200476501372381e-05,
      "loss": 0.954487419128418,
      "step": 15290
    },
    {
      "epoch": 1.6168804576139082,
      "grad_norm": 0.9228602051734924,
      "learning_rate": 2.1977281189499788e-05,
      "loss": 0.9471879959106445,
      "step": 15300
    },
    {
      "epoch": 1.6179373026672126,
      "grad_norm": 1.2539465427398682,
      "learning_rate": 2.1949801072163155e-05,
      "loss": 0.9079504013061523,
      "step": 15310
    },
    {
      "epoch": 1.6189941477205174,
      "grad_norm": 0.9965641498565674,
      "learning_rate": 2.1922324695413973e-05,
      "loss": 1.014242458343506,
      "step": 15320
    },
    {
      "epoch": 1.6200509927738218,
      "grad_norm": 1.160041332244873,
      "learning_rate": 2.1894852092947647e-05,
      "loss": 0.951849365234375,
      "step": 15330
    },
    {
      "epoch": 1.6211078378271266,
      "grad_norm": 0.9412044286727905,
      "learning_rate": 2.1867383298455017e-05,
      "loss": 0.8339347839355469,
      "step": 15340
    },
    {
      "epoch": 1.622164682880431,
      "grad_norm": 1.3474202156066895,
      "learning_rate": 2.183991834562219e-05,
      "loss": 0.8694920539855957,
      "step": 15350
    },
    {
      "epoch": 1.6232215279337359,
      "grad_norm": 0.9191882014274597,
      "learning_rate": 2.181245726813063e-05,
      "loss": 1.0029685020446777,
      "step": 15360
    },
    {
      "epoch": 1.6242783729870405,
      "grad_norm": 1.0446580648422241,
      "learning_rate": 2.178500009965698e-05,
      "loss": 1.0428786277770996,
      "step": 15370
    },
    {
      "epoch": 1.625335218040345,
      "grad_norm": 1.0893741846084595,
      "learning_rate": 2.1757546873873157e-05,
      "loss": 0.9086989402770996,
      "step": 15380
    },
    {
      "epoch": 1.6263920630936497,
      "grad_norm": 1.2540227174758911,
      "learning_rate": 2.173009762444617e-05,
      "loss": 1.0068238258361817,
      "step": 15390
    },
    {
      "epoch": 1.6274489081469543,
      "grad_norm": 0.9434010982513428,
      "learning_rate": 2.1702652385038223e-05,
      "loss": 1.0189486503601075,
      "step": 15400
    },
    {
      "epoch": 1.628505753200259,
      "grad_norm": 1.0710039138793945,
      "learning_rate": 2.1675211189306543e-05,
      "loss": 1.036940097808838,
      "step": 15410
    },
    {
      "epoch": 1.6295625982535635,
      "grad_norm": 1.4863547086715698,
      "learning_rate": 2.1647774070903453e-05,
      "loss": 0.9753593444824219,
      "step": 15420
    },
    {
      "epoch": 1.6306194433068681,
      "grad_norm": 1.1370391845703125,
      "learning_rate": 2.162034106347621e-05,
      "loss": 0.9852591514587402,
      "step": 15430
    },
    {
      "epoch": 1.6316762883601728,
      "grad_norm": 1.0257749557495117,
      "learning_rate": 2.159291220066711e-05,
      "loss": 0.8920008659362793,
      "step": 15440
    },
    {
      "epoch": 1.6327331334134774,
      "grad_norm": 1.0444551706314087,
      "learning_rate": 2.1565487516113282e-05,
      "loss": 0.8473061561584473,
      "step": 15450
    },
    {
      "epoch": 1.633789978466782,
      "grad_norm": 1.4183458089828491,
      "learning_rate": 2.1538067043446803e-05,
      "loss": 1.0219858169555665,
      "step": 15460
    },
    {
      "epoch": 1.6348468235200868,
      "grad_norm": 0.9618796110153198,
      "learning_rate": 2.151065081629453e-05,
      "loss": 0.9708030700683594,
      "step": 15470
    },
    {
      "epoch": 1.6359036685733912,
      "grad_norm": 1.143337368965149,
      "learning_rate": 2.1483238868278162e-05,
      "loss": 0.8943169593811036,
      "step": 15480
    },
    {
      "epoch": 1.636960513626696,
      "grad_norm": 1.169110894203186,
      "learning_rate": 2.145583123301409e-05,
      "loss": 0.945102310180664,
      "step": 15490
    },
    {
      "epoch": 1.6380173586800004,
      "grad_norm": 1.2800606489181519,
      "learning_rate": 2.1428427944113498e-05,
      "loss": 1.043473243713379,
      "step": 15500
    },
    {
      "epoch": 1.6380173586800004,
      "eval_loss": 0.4897018373012543,
      "eval_runtime": 3057.7975,
      "eval_samples_per_second": 2.912,
      "eval_steps_per_second": 2.912,
      "step": 15500
    },
    {
      "epoch": 1.6390742037333053,
      "grad_norm": 1.4023667573928833,
      "learning_rate": 2.1401029035182147e-05,
      "loss": 1.042397689819336,
      "step": 15510
    },
    {
      "epoch": 1.6401310487866096,
      "grad_norm": 1.127929925918579,
      "learning_rate": 2.1373634539820516e-05,
      "loss": 0.9989598274230957,
      "step": 15520
    },
    {
      "epoch": 1.6411878938399145,
      "grad_norm": 0.9940389394760132,
      "learning_rate": 2.1346244491623584e-05,
      "loss": 1.006131362915039,
      "step": 15530
    },
    {
      "epoch": 1.6422447388932189,
      "grad_norm": 1.1468394994735718,
      "learning_rate": 2.1318858924180968e-05,
      "loss": 0.9758601188659668,
      "step": 15540
    },
    {
      "epoch": 1.6433015839465237,
      "grad_norm": 1.1256625652313232,
      "learning_rate": 2.1291477871076697e-05,
      "loss": 1.0234405517578125,
      "step": 15550
    },
    {
      "epoch": 1.6443584289998283,
      "grad_norm": 1.1484577655792236,
      "learning_rate": 2.1264101365889354e-05,
      "loss": 0.9304427146911621,
      "step": 15560
    },
    {
      "epoch": 1.645415274053133,
      "grad_norm": 1.0025906562805176,
      "learning_rate": 2.123672944219186e-05,
      "loss": 1.0083551406860352,
      "step": 15570
    },
    {
      "epoch": 1.6464721191064375,
      "grad_norm": 1.142071008682251,
      "learning_rate": 2.1209362133551597e-05,
      "loss": 0.9500226020812989,
      "step": 15580
    },
    {
      "epoch": 1.6475289641597421,
      "grad_norm": 1.2195106744766235,
      "learning_rate": 2.118199947353021e-05,
      "loss": 1.0455318450927735,
      "step": 15590
    },
    {
      "epoch": 1.6485858092130468,
      "grad_norm": 1.012463092803955,
      "learning_rate": 2.1154641495683712e-05,
      "loss": 0.996275520324707,
      "step": 15600
    },
    {
      "epoch": 1.6496426542663514,
      "grad_norm": 1.1109004020690918,
      "learning_rate": 2.112728823356232e-05,
      "loss": 0.9795701026916503,
      "step": 15610
    },
    {
      "epoch": 1.650699499319656,
      "grad_norm": 1.2298986911773682,
      "learning_rate": 2.109993972071052e-05,
      "loss": 0.9488958358764649,
      "step": 15620
    },
    {
      "epoch": 1.6517563443729606,
      "grad_norm": 1.117938756942749,
      "learning_rate": 2.107259599066692e-05,
      "loss": 0.8748033523559571,
      "step": 15630
    },
    {
      "epoch": 1.6528131894262652,
      "grad_norm": 1.1766976118087769,
      "learning_rate": 2.1045257076964296e-05,
      "loss": 1.0114870071411133,
      "step": 15640
    },
    {
      "epoch": 1.6538700344795698,
      "grad_norm": 1.2001243829727173,
      "learning_rate": 2.1017923013129513e-05,
      "loss": 1.0272076606750489,
      "step": 15650
    },
    {
      "epoch": 1.6549268795328747,
      "grad_norm": 1.071753740310669,
      "learning_rate": 2.0990593832683473e-05,
      "loss": 1.0470212936401366,
      "step": 15660
    },
    {
      "epoch": 1.655983724586179,
      "grad_norm": 0.9853981137275696,
      "learning_rate": 2.096326956914112e-05,
      "loss": 0.9654929161071777,
      "step": 15670
    },
    {
      "epoch": 1.6570405696394839,
      "grad_norm": 1.113097906112671,
      "learning_rate": 2.093595025601133e-05,
      "loss": 0.9804095268249512,
      "step": 15680
    },
    {
      "epoch": 1.6580974146927883,
      "grad_norm": 1.035984992980957,
      "learning_rate": 2.090863592679695e-05,
      "loss": 0.9574061393737793,
      "step": 15690
    },
    {
      "epoch": 1.659154259746093,
      "grad_norm": 0.8454504609107971,
      "learning_rate": 2.088132661499467e-05,
      "loss": 0.9917940139770508,
      "step": 15700
    },
    {
      "epoch": 1.6602111047993975,
      "grad_norm": 1.7564895153045654,
      "learning_rate": 2.0854022354095063e-05,
      "loss": 0.8316843032836914,
      "step": 15710
    },
    {
      "epoch": 1.6612679498527023,
      "grad_norm": 1.084649920463562,
      "learning_rate": 2.0826723177582485e-05,
      "loss": 0.9198145866394043,
      "step": 15720
    },
    {
      "epoch": 1.6623247949060067,
      "grad_norm": 0.9473570585250854,
      "learning_rate": 2.079942911893508e-05,
      "loss": 0.8318938255310059,
      "step": 15730
    },
    {
      "epoch": 1.6633816399593115,
      "grad_norm": 1.2215439081192017,
      "learning_rate": 2.0772140211624687e-05,
      "loss": 1.0007565498352051,
      "step": 15740
    },
    {
      "epoch": 1.664438485012616,
      "grad_norm": 1.2325350046157837,
      "learning_rate": 2.0744856489116853e-05,
      "loss": 0.9506117820739746,
      "step": 15750
    },
    {
      "epoch": 1.6654953300659208,
      "grad_norm": 0.9199979901313782,
      "learning_rate": 2.0717577984870748e-05,
      "loss": 1.0144503593444825,
      "step": 15760
    },
    {
      "epoch": 1.6665521751192254,
      "grad_norm": 0.9982006549835205,
      "learning_rate": 2.0690304732339162e-05,
      "loss": 1.0006795883178712,
      "step": 15770
    },
    {
      "epoch": 1.66760902017253,
      "grad_norm": 1.7438162565231323,
      "learning_rate": 2.0663036764968418e-05,
      "loss": 0.9910264015197754,
      "step": 15780
    },
    {
      "epoch": 1.6686658652258346,
      "grad_norm": 1.2460399866104126,
      "learning_rate": 2.0635774116198387e-05,
      "loss": 0.8913528442382812,
      "step": 15790
    },
    {
      "epoch": 1.6697227102791392,
      "grad_norm": 1.5714384317398071,
      "learning_rate": 2.0608516819462396e-05,
      "loss": 0.943422794342041,
      "step": 15800
    },
    {
      "epoch": 1.6707795553324438,
      "grad_norm": 1.270508885383606,
      "learning_rate": 2.058126490818722e-05,
      "loss": 0.997713279724121,
      "step": 15810
    },
    {
      "epoch": 1.6718364003857484,
      "grad_norm": 1.1486079692840576,
      "learning_rate": 2.055401841579302e-05,
      "loss": 1.046572780609131,
      "step": 15820
    },
    {
      "epoch": 1.672893245439053,
      "grad_norm": 1.0291783809661865,
      "learning_rate": 2.052677737569333e-05,
      "loss": 1.1071664810180664,
      "step": 15830
    },
    {
      "epoch": 1.6739500904923577,
      "grad_norm": 1.1676501035690308,
      "learning_rate": 2.0499541821294965e-05,
      "loss": 0.9514002799987793,
      "step": 15840
    },
    {
      "epoch": 1.6750069355456623,
      "grad_norm": 1.1145401000976562,
      "learning_rate": 2.0472311785998054e-05,
      "loss": 0.9818894386291503,
      "step": 15850
    },
    {
      "epoch": 1.6760637805989669,
      "grad_norm": 0.9500560164451599,
      "learning_rate": 2.044508730319592e-05,
      "loss": 0.9481066703796387,
      "step": 15860
    },
    {
      "epoch": 1.6771206256522717,
      "grad_norm": 1.8700789213180542,
      "learning_rate": 2.0417868406275107e-05,
      "loss": 1.013495635986328,
      "step": 15870
    },
    {
      "epoch": 1.678177470705576,
      "grad_norm": 1.3341236114501953,
      "learning_rate": 2.039065512861528e-05,
      "loss": 0.8866471290588379,
      "step": 15880
    },
    {
      "epoch": 1.679234315758881,
      "grad_norm": 1.5769577026367188,
      "learning_rate": 2.036344750358924e-05,
      "loss": 0.8301256179809571,
      "step": 15890
    },
    {
      "epoch": 1.6802911608121853,
      "grad_norm": 1.0305225849151611,
      "learning_rate": 2.0336245564562835e-05,
      "loss": 1.061864185333252,
      "step": 15900
    },
    {
      "epoch": 1.6813480058654902,
      "grad_norm": 1.5124924182891846,
      "learning_rate": 2.0309049344894956e-05,
      "loss": 0.9499337196350097,
      "step": 15910
    },
    {
      "epoch": 1.6824048509187945,
      "grad_norm": 1.2007012367248535,
      "learning_rate": 2.0281858877937464e-05,
      "loss": 0.8820402145385742,
      "step": 15920
    },
    {
      "epoch": 1.6834616959720994,
      "grad_norm": 1.721504807472229,
      "learning_rate": 2.0254674197035188e-05,
      "loss": 0.8545654296875,
      "step": 15930
    },
    {
      "epoch": 1.6845185410254038,
      "grad_norm": 1.4302971363067627,
      "learning_rate": 2.0227495335525833e-05,
      "loss": 0.9610498428344727,
      "step": 15940
    },
    {
      "epoch": 1.6855753860787086,
      "grad_norm": 0.9706851243972778,
      "learning_rate": 2.020032232673999e-05,
      "loss": 0.9970791816711426,
      "step": 15950
    },
    {
      "epoch": 1.686632231132013,
      "grad_norm": 1.3307527303695679,
      "learning_rate": 2.0173155204001054e-05,
      "loss": 0.9410307884216309,
      "step": 15960
    },
    {
      "epoch": 1.6876890761853178,
      "grad_norm": 1.0494908094406128,
      "learning_rate": 2.014599400062523e-05,
      "loss": 1.038601016998291,
      "step": 15970
    },
    {
      "epoch": 1.6887459212386224,
      "grad_norm": 1.3198423385620117,
      "learning_rate": 2.0118838749921428e-05,
      "loss": 1.0488903999328614,
      "step": 15980
    },
    {
      "epoch": 1.689802766291927,
      "grad_norm": 1.2443501949310303,
      "learning_rate": 2.009168948519129e-05,
      "loss": 0.8924653053283691,
      "step": 15990
    },
    {
      "epoch": 1.6908596113452317,
      "grad_norm": 1.237757682800293,
      "learning_rate": 2.0064546239729094e-05,
      "loss": 0.970335865020752,
      "step": 16000
    },
    {
      "epoch": 1.6908596113452317,
      "eval_loss": 0.48750653862953186,
      "eval_runtime": 3041.1419,
      "eval_samples_per_second": 2.928,
      "eval_steps_per_second": 2.928,
      "step": 16000
    },
    {
      "epoch": 1.6919164563985363,
      "grad_norm": 1.6246154308319092,
      "learning_rate": 2.003740904682175e-05,
      "loss": 1.0975980758666992,
      "step": 16010
    },
    {
      "epoch": 1.6929733014518409,
      "grad_norm": 1.2005983591079712,
      "learning_rate": 2.001027793974873e-05,
      "loss": 0.9837453842163086,
      "step": 16020
    },
    {
      "epoch": 1.6940301465051455,
      "grad_norm": 1.1049518585205078,
      "learning_rate": 1.9983152951782074e-05,
      "loss": 0.9141158103942871,
      "step": 16030
    },
    {
      "epoch": 1.69508699155845,
      "grad_norm": 0.9200447201728821,
      "learning_rate": 1.9956034116186272e-05,
      "loss": 1.026270866394043,
      "step": 16040
    },
    {
      "epoch": 1.6961438366117547,
      "grad_norm": 1.157163143157959,
      "learning_rate": 1.992892146621832e-05,
      "loss": 0.9300186157226562,
      "step": 16050
    },
    {
      "epoch": 1.6972006816650593,
      "grad_norm": 0.8886343240737915,
      "learning_rate": 1.9901815035127582e-05,
      "loss": 0.8931681632995605,
      "step": 16060
    },
    {
      "epoch": 1.698257526718364,
      "grad_norm": 1.4814364910125732,
      "learning_rate": 1.9874714856155834e-05,
      "loss": 0.9867460250854492,
      "step": 16070
    },
    {
      "epoch": 1.6993143717716688,
      "grad_norm": 1.440529227256775,
      "learning_rate": 1.984762096253715e-05,
      "loss": 0.9464458465576172,
      "step": 16080
    },
    {
      "epoch": 1.7003712168249732,
      "grad_norm": 1.3577831983566284,
      "learning_rate": 1.9820533387497924e-05,
      "loss": 1.007056427001953,
      "step": 16090
    },
    {
      "epoch": 1.701428061878278,
      "grad_norm": 1.1183768510818481,
      "learning_rate": 1.9793452164256787e-05,
      "loss": 0.9177047729492187,
      "step": 16100
    },
    {
      "epoch": 1.7024849069315824,
      "grad_norm": 0.8974764943122864,
      "learning_rate": 1.976637732602459e-05,
      "loss": 0.9991282463073731,
      "step": 16110
    },
    {
      "epoch": 1.7035417519848872,
      "grad_norm": 1.1656705141067505,
      "learning_rate": 1.973930890600433e-05,
      "loss": 0.9225249290466309,
      "step": 16120
    },
    {
      "epoch": 1.7045985970381916,
      "grad_norm": 1.3560844659805298,
      "learning_rate": 1.9712246937391173e-05,
      "loss": 0.9752987861633301,
      "step": 16130
    },
    {
      "epoch": 1.7056554420914964,
      "grad_norm": 1.0366978645324707,
      "learning_rate": 1.9685191453372333e-05,
      "loss": 1.0417604446411133,
      "step": 16140
    },
    {
      "epoch": 1.7067122871448008,
      "grad_norm": 1.3147910833358765,
      "learning_rate": 1.9658142487127105e-05,
      "loss": 0.9578855514526368,
      "step": 16150
    },
    {
      "epoch": 1.7077691321981057,
      "grad_norm": 1.232405662536621,
      "learning_rate": 1.9631100071826763e-05,
      "loss": 0.9088631629943847,
      "step": 16160
    },
    {
      "epoch": 1.7088259772514103,
      "grad_norm": 1.0008997917175293,
      "learning_rate": 1.9604064240634568e-05,
      "loss": 0.9593038558959961,
      "step": 16170
    },
    {
      "epoch": 1.7098828223047149,
      "grad_norm": 1.1990646123886108,
      "learning_rate": 1.9577035026705697e-05,
      "loss": 0.9211307525634765,
      "step": 16180
    },
    {
      "epoch": 1.7109396673580195,
      "grad_norm": 1.1719162464141846,
      "learning_rate": 1.9550012463187208e-05,
      "loss": 0.8593046188354492,
      "step": 16190
    },
    {
      "epoch": 1.7119965124113241,
      "grad_norm": 1.082959532737732,
      "learning_rate": 1.952299658321802e-05,
      "loss": 0.9516560554504394,
      "step": 16200
    },
    {
      "epoch": 1.7130533574646287,
      "grad_norm": 1.2788375616073608,
      "learning_rate": 1.949598741992883e-05,
      "loss": 0.9744342803955078,
      "step": 16210
    },
    {
      "epoch": 1.7141102025179333,
      "grad_norm": 1.9034672975540161,
      "learning_rate": 1.9468985006442124e-05,
      "loss": 1.029246997833252,
      "step": 16220
    },
    {
      "epoch": 1.715167047571238,
      "grad_norm": 2.1154751777648926,
      "learning_rate": 1.9441989375872098e-05,
      "loss": 1.0807592391967773,
      "step": 16230
    },
    {
      "epoch": 1.7162238926245426,
      "grad_norm": 1.1081537008285522,
      "learning_rate": 1.9415000561324627e-05,
      "loss": 1.0384562492370606,
      "step": 16240
    },
    {
      "epoch": 1.7172807376778472,
      "grad_norm": 1.3060991764068604,
      "learning_rate": 1.938801859589723e-05,
      "loss": 0.9800905227661133,
      "step": 16250
    },
    {
      "epoch": 1.7183375827311518,
      "grad_norm": 1.2061007022857666,
      "learning_rate": 1.936104351267904e-05,
      "loss": 0.8132293701171875,
      "step": 16260
    },
    {
      "epoch": 1.7193944277844566,
      "grad_norm": 1.0858416557312012,
      "learning_rate": 1.9334075344750717e-05,
      "loss": 0.9868954658508301,
      "step": 16270
    },
    {
      "epoch": 1.720451272837761,
      "grad_norm": 0.8633347749710083,
      "learning_rate": 1.9307114125184482e-05,
      "loss": 0.8176230430603028,
      "step": 16280
    },
    {
      "epoch": 1.7215081178910658,
      "grad_norm": 0.9935333728790283,
      "learning_rate": 1.9280159887044e-05,
      "loss": 1.0078252792358398,
      "step": 16290
    },
    {
      "epoch": 1.7225649629443702,
      "grad_norm": 1.4650909900665283,
      "learning_rate": 1.92532126633844e-05,
      "loss": 0.9453054428100586,
      "step": 16300
    },
    {
      "epoch": 1.723621807997675,
      "grad_norm": 1.455993413925171,
      "learning_rate": 1.9226272487252185e-05,
      "loss": 0.9755555152893066,
      "step": 16310
    },
    {
      "epoch": 1.7246786530509794,
      "grad_norm": 1.2221165895462036,
      "learning_rate": 1.9199339391685237e-05,
      "loss": 0.9090837478637696,
      "step": 16320
    },
    {
      "epoch": 1.7257354981042843,
      "grad_norm": 1.8252278566360474,
      "learning_rate": 1.917241340971274e-05,
      "loss": 0.9535693168640137,
      "step": 16330
    },
    {
      "epoch": 1.7267923431575887,
      "grad_norm": 1.3256672620773315,
      "learning_rate": 1.9145494574355168e-05,
      "loss": 0.9649052619934082,
      "step": 16340
    },
    {
      "epoch": 1.7278491882108935,
      "grad_norm": 1.4124003648757935,
      "learning_rate": 1.911858291862421e-05,
      "loss": 0.888246726989746,
      "step": 16350
    },
    {
      "epoch": 1.728906033264198,
      "grad_norm": 1.025699257850647,
      "learning_rate": 1.909167847552278e-05,
      "loss": 1.002976894378662,
      "step": 16360
    },
    {
      "epoch": 1.7299628783175027,
      "grad_norm": 1.4224721193313599,
      "learning_rate": 1.9064781278044913e-05,
      "loss": 0.9178348541259765,
      "step": 16370
    },
    {
      "epoch": 1.7310197233708073,
      "grad_norm": 0.9423581957817078,
      "learning_rate": 1.903789135917579e-05,
      "loss": 0.8939762115478516,
      "step": 16380
    },
    {
      "epoch": 1.732076568424112,
      "grad_norm": 1.1359885931015015,
      "learning_rate": 1.9011008751891642e-05,
      "loss": 1.0512442588806152,
      "step": 16390
    },
    {
      "epoch": 1.7331334134774166,
      "grad_norm": 1.1545013189315796,
      "learning_rate": 1.8984133489159745e-05,
      "loss": 0.967808723449707,
      "step": 16400
    },
    {
      "epoch": 1.7341902585307212,
      "grad_norm": 1.0292633771896362,
      "learning_rate": 1.895726560393836e-05,
      "loss": 0.8976780891418457,
      "step": 16410
    },
    {
      "epoch": 1.7352471035840258,
      "grad_norm": 0.8994625210762024,
      "learning_rate": 1.8930405129176717e-05,
      "loss": 1.0991263389587402,
      "step": 16420
    },
    {
      "epoch": 1.7363039486373304,
      "grad_norm": 1.0030157566070557,
      "learning_rate": 1.8903552097814943e-05,
      "loss": 0.9734399795532227,
      "step": 16430
    },
    {
      "epoch": 1.737360793690635,
      "grad_norm": 0.949092447757721,
      "learning_rate": 1.887670654278404e-05,
      "loss": 0.9934487342834473,
      "step": 16440
    },
    {
      "epoch": 1.7384176387439396,
      "grad_norm": 1.2949284315109253,
      "learning_rate": 1.8849868497005845e-05,
      "loss": 0.824336051940918,
      "step": 16450
    },
    {
      "epoch": 1.7394744837972442,
      "grad_norm": 0.9957984089851379,
      "learning_rate": 1.8823037993392984e-05,
      "loss": 0.9529772758483886,
      "step": 16460
    },
    {
      "epoch": 1.7405313288505488,
      "grad_norm": 1.1070786714553833,
      "learning_rate": 1.8796215064848828e-05,
      "loss": 1.0027576446533204,
      "step": 16470
    },
    {
      "epoch": 1.7415881739038537,
      "grad_norm": 1.212620496749878,
      "learning_rate": 1.8769399744267475e-05,
      "loss": 0.9541872024536133,
      "step": 16480
    },
    {
      "epoch": 1.742645018957158,
      "grad_norm": 1.2480942010879517,
      "learning_rate": 1.8742592064533676e-05,
      "loss": 0.9191214561462402,
      "step": 16490
    },
    {
      "epoch": 1.743701864010463,
      "grad_norm": 1.1040199995040894,
      "learning_rate": 1.8715792058522817e-05,
      "loss": 0.8900608062744141,
      "step": 16500
    },
    {
      "epoch": 1.743701864010463,
      "eval_loss": 0.4857068657875061,
      "eval_runtime": 3043.3919,
      "eval_samples_per_second": 2.926,
      "eval_steps_per_second": 2.926,
      "step": 16500
    },
    {
      "epoch": 1.7447587090637673,
      "grad_norm": 1.0166687965393066,
      "learning_rate": 1.8688999759100875e-05,
      "loss": 0.9593010902404785,
      "step": 16510
    },
    {
      "epoch": 1.7458155541170721,
      "grad_norm": 1.1431307792663574,
      "learning_rate": 1.866221519912438e-05,
      "loss": 0.975101089477539,
      "step": 16520
    },
    {
      "epoch": 1.7468723991703765,
      "grad_norm": 1.0124188661575317,
      "learning_rate": 1.863543841144036e-05,
      "loss": 0.9151116371154785,
      "step": 16530
    },
    {
      "epoch": 1.7479292442236813,
      "grad_norm": 1.0851496458053589,
      "learning_rate": 1.8608669428886322e-05,
      "loss": 1.0264881134033204,
      "step": 16540
    },
    {
      "epoch": 1.7489860892769857,
      "grad_norm": 1.2364506721496582,
      "learning_rate": 1.858190828429019e-05,
      "loss": 0.9725762367248535,
      "step": 16550
    },
    {
      "epoch": 1.7500429343302906,
      "grad_norm": 1.0631964206695557,
      "learning_rate": 1.8555155010470295e-05,
      "loss": 0.9212296485900879,
      "step": 16560
    },
    {
      "epoch": 1.751099779383595,
      "grad_norm": 1.2657827138900757,
      "learning_rate": 1.852840964023529e-05,
      "loss": 0.9146785736083984,
      "step": 16570
    },
    {
      "epoch": 1.7521566244368998,
      "grad_norm": 1.16664719581604,
      "learning_rate": 1.850167220638416e-05,
      "loss": 0.8493237495422363,
      "step": 16580
    },
    {
      "epoch": 1.7532134694902044,
      "grad_norm": 1.3615429401397705,
      "learning_rate": 1.8474942741706135e-05,
      "loss": 0.9358036994934082,
      "step": 16590
    },
    {
      "epoch": 1.754270314543509,
      "grad_norm": 1.1035430431365967,
      "learning_rate": 1.8448221278980704e-05,
      "loss": 1.0355838775634765,
      "step": 16600
    },
    {
      "epoch": 1.7553271595968136,
      "grad_norm": 1.0050832033157349,
      "learning_rate": 1.8421507850977497e-05,
      "loss": 0.9752060890197753,
      "step": 16610
    },
    {
      "epoch": 1.7563840046501182,
      "grad_norm": 1.2342474460601807,
      "learning_rate": 1.8394802490456336e-05,
      "loss": 0.9991334915161133,
      "step": 16620
    },
    {
      "epoch": 1.7574408497034228,
      "grad_norm": 1.0519160032272339,
      "learning_rate": 1.8368105230167114e-05,
      "loss": 0.9203132629394531,
      "step": 16630
    },
    {
      "epoch": 1.7584976947567275,
      "grad_norm": 1.1150805950164795,
      "learning_rate": 1.834141610284982e-05,
      "loss": 0.987761402130127,
      "step": 16640
    },
    {
      "epoch": 1.759554539810032,
      "grad_norm": 1.4006835222244263,
      "learning_rate": 1.8314735141234445e-05,
      "loss": 0.8921330451965332,
      "step": 16650
    },
    {
      "epoch": 1.7606113848633367,
      "grad_norm": 0.9083954095840454,
      "learning_rate": 1.8288062378040984e-05,
      "loss": 0.9279559135437012,
      "step": 16660
    },
    {
      "epoch": 1.7616682299166413,
      "grad_norm": 1.2708629369735718,
      "learning_rate": 1.826139784597936e-05,
      "loss": 0.9566614151000976,
      "step": 16670
    },
    {
      "epoch": 1.762725074969946,
      "grad_norm": 1.095931887626648,
      "learning_rate": 1.8234741577749422e-05,
      "loss": 0.9766691207885743,
      "step": 16680
    },
    {
      "epoch": 1.7637819200232507,
      "grad_norm": 1.7623662948608398,
      "learning_rate": 1.820809360604087e-05,
      "loss": 0.93494873046875,
      "step": 16690
    },
    {
      "epoch": 1.7648387650765551,
      "grad_norm": 1.2194961309432983,
      "learning_rate": 1.8181453963533235e-05,
      "loss": 0.9367061614990234,
      "step": 16700
    },
    {
      "epoch": 1.76589561012986,
      "grad_norm": 1.3538312911987305,
      "learning_rate": 1.815482268289583e-05,
      "loss": 0.9636688232421875,
      "step": 16710
    },
    {
      "epoch": 1.7669524551831643,
      "grad_norm": 1.3224133253097534,
      "learning_rate": 1.812819979678772e-05,
      "loss": 0.9113118171691894,
      "step": 16720
    },
    {
      "epoch": 1.7680093002364692,
      "grad_norm": 1.0886527299880981,
      "learning_rate": 1.8101585337857667e-05,
      "loss": 0.920682430267334,
      "step": 16730
    },
    {
      "epoch": 1.7690661452897736,
      "grad_norm": 0.9373747706413269,
      "learning_rate": 1.8074979338744113e-05,
      "loss": 0.7965409278869628,
      "step": 16740
    },
    {
      "epoch": 1.7701229903430784,
      "grad_norm": 1.083687424659729,
      "learning_rate": 1.8048381832075117e-05,
      "loss": 0.8952385902404785,
      "step": 16750
    },
    {
      "epoch": 1.7711798353963828,
      "grad_norm": 1.1592779159545898,
      "learning_rate": 1.8021792850468295e-05,
      "loss": 0.9936323165893555,
      "step": 16760
    },
    {
      "epoch": 1.7722366804496876,
      "grad_norm": 0.9977505207061768,
      "learning_rate": 1.7995212426530867e-05,
      "loss": 0.9088336944580078,
      "step": 16770
    },
    {
      "epoch": 1.7732935255029922,
      "grad_norm": 1.0375714302062988,
      "learning_rate": 1.79686405928595e-05,
      "loss": 0.9886848449707031,
      "step": 16780
    },
    {
      "epoch": 1.7743503705562969,
      "grad_norm": 0.990752637386322,
      "learning_rate": 1.7942077382040383e-05,
      "loss": 0.9729957580566406,
      "step": 16790
    },
    {
      "epoch": 1.7754072156096015,
      "grad_norm": 1.4082671403884888,
      "learning_rate": 1.791552282664907e-05,
      "loss": 0.8913538932800293,
      "step": 16800
    },
    {
      "epoch": 1.776464060662906,
      "grad_norm": 1.014238953590393,
      "learning_rate": 1.7888976959250558e-05,
      "loss": 0.9575340270996093,
      "step": 16810
    },
    {
      "epoch": 1.7775209057162107,
      "grad_norm": 0.928726077079773,
      "learning_rate": 1.7862439812399136e-05,
      "loss": 0.897059440612793,
      "step": 16820
    },
    {
      "epoch": 1.7785777507695153,
      "grad_norm": 1.0944421291351318,
      "learning_rate": 1.783591141863845e-05,
      "loss": 0.988118839263916,
      "step": 16830
    },
    {
      "epoch": 1.77963459582282,
      "grad_norm": 1.2754912376403809,
      "learning_rate": 1.780939181050136e-05,
      "loss": 0.9385346412658692,
      "step": 16840
    },
    {
      "epoch": 1.7806914408761245,
      "grad_norm": 1.2028642892837524,
      "learning_rate": 1.7782881020510012e-05,
      "loss": 1.0269272804260254,
      "step": 16850
    },
    {
      "epoch": 1.7817482859294291,
      "grad_norm": 1.522111415863037,
      "learning_rate": 1.775637908117567e-05,
      "loss": 0.998070240020752,
      "step": 16860
    },
    {
      "epoch": 1.7828051309827337,
      "grad_norm": 1.0815643072128296,
      "learning_rate": 1.7729886024998808e-05,
      "loss": 0.9459043502807617,
      "step": 16870
    },
    {
      "epoch": 1.7838619760360386,
      "grad_norm": 1.5040841102600098,
      "learning_rate": 1.770340188446895e-05,
      "loss": 0.8215196609497071,
      "step": 16880
    },
    {
      "epoch": 1.784918821089343,
      "grad_norm": 0.9555218815803528,
      "learning_rate": 1.7676926692064744e-05,
      "loss": 0.9528297424316406,
      "step": 16890
    },
    {
      "epoch": 1.7859756661426478,
      "grad_norm": 1.8246442079544067,
      "learning_rate": 1.7650460480253797e-05,
      "loss": 1.0405531883239747,
      "step": 16900
    },
    {
      "epoch": 1.7870325111959522,
      "grad_norm": 0.9257091283798218,
      "learning_rate": 1.7624003281492773e-05,
      "loss": 0.9948619842529297,
      "step": 16910
    },
    {
      "epoch": 1.788089356249257,
      "grad_norm": 0.9648146033287048,
      "learning_rate": 1.7597555128227225e-05,
      "loss": 0.9135502815246582,
      "step": 16920
    },
    {
      "epoch": 1.7891462013025614,
      "grad_norm": 1.1536953449249268,
      "learning_rate": 1.7571116052891664e-05,
      "loss": 1.034188938140869,
      "step": 16930
    },
    {
      "epoch": 1.7902030463558662,
      "grad_norm": 1.6734118461608887,
      "learning_rate": 1.7544686087909402e-05,
      "loss": 0.9807829856872559,
      "step": 16940
    },
    {
      "epoch": 1.7912598914091706,
      "grad_norm": 1.235108494758606,
      "learning_rate": 1.7518265265692662e-05,
      "loss": 0.9491137504577637,
      "step": 16950
    },
    {
      "epoch": 1.7923167364624755,
      "grad_norm": 1.1582393646240234,
      "learning_rate": 1.7491853618642377e-05,
      "loss": 0.9547677993774414,
      "step": 16960
    },
    {
      "epoch": 1.7933735815157799,
      "grad_norm": 1.5912330150604248,
      "learning_rate": 1.7465451179148288e-05,
      "loss": 0.9783409118652344,
      "step": 16970
    },
    {
      "epoch": 1.7944304265690847,
      "grad_norm": 1.0394923686981201,
      "learning_rate": 1.7439057979588795e-05,
      "loss": 0.9290766716003418,
      "step": 16980
    },
    {
      "epoch": 1.7954872716223893,
      "grad_norm": 1.0731515884399414,
      "learning_rate": 1.741267405233102e-05,
      "loss": 0.964657211303711,
      "step": 16990
    },
    {
      "epoch": 1.796544116675694,
      "grad_norm": 1.102556586265564,
      "learning_rate": 1.7386299429730647e-05,
      "loss": 0.9523024559020996,
      "step": 17000
    },
    {
      "epoch": 1.796544116675694,
      "eval_loss": 0.48546352982521057,
      "eval_runtime": 3040.9175,
      "eval_samples_per_second": 2.928,
      "eval_steps_per_second": 2.928,
      "step": 17000
    },
    {
      "epoch": 1.7976009617289985,
      "grad_norm": 1.0496493577957153,
      "learning_rate": 1.7359934144132023e-05,
      "loss": 0.8735774040222168,
      "step": 17010
    },
    {
      "epoch": 1.7986578067823031,
      "grad_norm": 1.2210373878479004,
      "learning_rate": 1.733357822786798e-05,
      "loss": 1.0372821807861328,
      "step": 17020
    },
    {
      "epoch": 1.7997146518356077,
      "grad_norm": 1.1708858013153076,
      "learning_rate": 1.7307231713259913e-05,
      "loss": 1.046485424041748,
      "step": 17030
    },
    {
      "epoch": 1.8007714968889124,
      "grad_norm": 1.4242784976959229,
      "learning_rate": 1.7280894632617632e-05,
      "loss": 0.9750542640686035,
      "step": 17040
    },
    {
      "epoch": 1.801828341942217,
      "grad_norm": 0.9717543125152588,
      "learning_rate": 1.7254567018239447e-05,
      "loss": 1.006408977508545,
      "step": 17050
    },
    {
      "epoch": 1.8028851869955216,
      "grad_norm": 1.3366122245788574,
      "learning_rate": 1.7228248902411985e-05,
      "loss": 0.9424902915954589,
      "step": 17060
    },
    {
      "epoch": 1.8039420320488262,
      "grad_norm": 1.226166844367981,
      "learning_rate": 1.7201940317410288e-05,
      "loss": 1.0407800674438477,
      "step": 17070
    },
    {
      "epoch": 1.8049988771021308,
      "grad_norm": 0.9872190356254578,
      "learning_rate": 1.7175641295497653e-05,
      "loss": 0.9309253692626953,
      "step": 17080
    },
    {
      "epoch": 1.8060557221554356,
      "grad_norm": 2.066547393798828,
      "learning_rate": 1.714935186892571e-05,
      "loss": 1.0164306640625,
      "step": 17090
    },
    {
      "epoch": 1.80711256720874,
      "grad_norm": 1.6657413244247437,
      "learning_rate": 1.712307206993426e-05,
      "loss": 0.8152523040771484,
      "step": 17100
    },
    {
      "epoch": 1.8081694122620449,
      "grad_norm": 1.173179268836975,
      "learning_rate": 1.709680193075136e-05,
      "loss": 1.0912933349609375,
      "step": 17110
    },
    {
      "epoch": 1.8092262573153493,
      "grad_norm": 0.9084728956222534,
      "learning_rate": 1.7070541483593157e-05,
      "loss": 0.9357505798339844,
      "step": 17120
    },
    {
      "epoch": 1.810283102368654,
      "grad_norm": 1.0670437812805176,
      "learning_rate": 1.7044290760663967e-05,
      "loss": 0.9077163696289062,
      "step": 17130
    },
    {
      "epoch": 1.8113399474219585,
      "grad_norm": 1.8896702527999878,
      "learning_rate": 1.7018049794156132e-05,
      "loss": 1.0126530647277832,
      "step": 17140
    },
    {
      "epoch": 1.8123967924752633,
      "grad_norm": 1.1447577476501465,
      "learning_rate": 1.6991818616250088e-05,
      "loss": 1.027334213256836,
      "step": 17150
    },
    {
      "epoch": 1.8134536375285677,
      "grad_norm": 1.2356138229370117,
      "learning_rate": 1.6965597259114203e-05,
      "loss": 0.9770386695861817,
      "step": 17160
    },
    {
      "epoch": 1.8145104825818725,
      "grad_norm": 1.3413753509521484,
      "learning_rate": 1.6939385754904858e-05,
      "loss": 0.8935284614562988,
      "step": 17170
    },
    {
      "epoch": 1.8155673276351771,
      "grad_norm": 1.5614864826202393,
      "learning_rate": 1.6913184135766292e-05,
      "loss": 0.8167881011962891,
      "step": 17180
    },
    {
      "epoch": 1.8166241726884818,
      "grad_norm": 1.1042428016662598,
      "learning_rate": 1.6886992433830693e-05,
      "loss": 0.8608649253845215,
      "step": 17190
    },
    {
      "epoch": 1.8176810177417864,
      "grad_norm": 1.1336779594421387,
      "learning_rate": 1.686081068121801e-05,
      "loss": 0.9864889144897461,
      "step": 17200
    },
    {
      "epoch": 1.818737862795091,
      "grad_norm": 1.2629783153533936,
      "learning_rate": 1.6834638910036065e-05,
      "loss": 0.8690094947814941,
      "step": 17210
    },
    {
      "epoch": 1.8197947078483956,
      "grad_norm": 1.2294223308563232,
      "learning_rate": 1.6808477152380374e-05,
      "loss": 0.9985342025756836,
      "step": 17220
    },
    {
      "epoch": 1.8208515529017002,
      "grad_norm": 1.1546047925949097,
      "learning_rate": 1.678232544033423e-05,
      "loss": 1.0338425636291504,
      "step": 17230
    },
    {
      "epoch": 1.8219083979550048,
      "grad_norm": 1.9059128761291504,
      "learning_rate": 1.6756183805968543e-05,
      "loss": 0.928558349609375,
      "step": 17240
    },
    {
      "epoch": 1.8229652430083094,
      "grad_norm": 1.271100640296936,
      "learning_rate": 1.673005228134194e-05,
      "loss": 0.9989326477050782,
      "step": 17250
    },
    {
      "epoch": 1.824022088061614,
      "grad_norm": 1.5284476280212402,
      "learning_rate": 1.6703930898500576e-05,
      "loss": 0.9634498596191406,
      "step": 17260
    },
    {
      "epoch": 1.8250789331149186,
      "grad_norm": 1.094624638557434,
      "learning_rate": 1.6677819689478236e-05,
      "loss": 0.9509388923645019,
      "step": 17270
    },
    {
      "epoch": 1.8261357781682233,
      "grad_norm": 1.2367336750030518,
      "learning_rate": 1.6651718686296165e-05,
      "loss": 0.945991325378418,
      "step": 17280
    },
    {
      "epoch": 1.8271926232215279,
      "grad_norm": 1.6015136241912842,
      "learning_rate": 1.6625627920963148e-05,
      "loss": 0.9954188346862793,
      "step": 17290
    },
    {
      "epoch": 1.8282494682748327,
      "grad_norm": 1.1669470071792603,
      "learning_rate": 1.6599547425475377e-05,
      "loss": 0.9762043952941895,
      "step": 17300
    },
    {
      "epoch": 1.829306313328137,
      "grad_norm": 0.9764379262924194,
      "learning_rate": 1.6573477231816455e-05,
      "loss": 0.9192074775695801,
      "step": 17310
    },
    {
      "epoch": 1.830363158381442,
      "grad_norm": 1.0812150239944458,
      "learning_rate": 1.6547417371957368e-05,
      "loss": 0.9784024238586426,
      "step": 17320
    },
    {
      "epoch": 1.8314200034347463,
      "grad_norm": 1.235129952430725,
      "learning_rate": 1.6521367877856416e-05,
      "loss": 0.8718074798583985,
      "step": 17330
    },
    {
      "epoch": 1.8324768484880511,
      "grad_norm": 1.3876386880874634,
      "learning_rate": 1.64953287814592e-05,
      "loss": 1.004378890991211,
      "step": 17340
    },
    {
      "epoch": 1.8335336935413555,
      "grad_norm": 1.0377479791641235,
      "learning_rate": 1.646930011469855e-05,
      "loss": 1.0211670875549317,
      "step": 17350
    },
    {
      "epoch": 1.8345905385946604,
      "grad_norm": 0.8227308988571167,
      "learning_rate": 1.6443281909494526e-05,
      "loss": 0.9382139205932617,
      "step": 17360
    },
    {
      "epoch": 1.8356473836479648,
      "grad_norm": 1.2808406352996826,
      "learning_rate": 1.6417274197754338e-05,
      "loss": 0.9920742034912109,
      "step": 17370
    },
    {
      "epoch": 1.8367042287012696,
      "grad_norm": 1.015110731124878,
      "learning_rate": 1.639127701137235e-05,
      "loss": 1.029364776611328,
      "step": 17380
    },
    {
      "epoch": 1.8377610737545742,
      "grad_norm": 1.2698861360549927,
      "learning_rate": 1.6365290382229994e-05,
      "loss": 0.9941561698913575,
      "step": 17390
    },
    {
      "epoch": 1.8388179188078788,
      "grad_norm": 0.862926721572876,
      "learning_rate": 1.6339314342195784e-05,
      "loss": 0.8721780776977539,
      "step": 17400
    },
    {
      "epoch": 1.8398747638611834,
      "grad_norm": 1.3410500288009644,
      "learning_rate": 1.6313348923125217e-05,
      "loss": 0.9076086997985839,
      "step": 17410
    },
    {
      "epoch": 1.840931608914488,
      "grad_norm": 0.9307976961135864,
      "learning_rate": 1.6287394156860788e-05,
      "loss": 1.0047386169433594,
      "step": 17420
    },
    {
      "epoch": 1.8419884539677926,
      "grad_norm": 1.090110182762146,
      "learning_rate": 1.6261450075231915e-05,
      "loss": 0.9007748603820801,
      "step": 17430
    },
    {
      "epoch": 1.8430452990210973,
      "grad_norm": 1.0016855001449585,
      "learning_rate": 1.623551671005493e-05,
      "loss": 0.9457592010498047,
      "step": 17440
    },
    {
      "epoch": 1.8441021440744019,
      "grad_norm": 1.1468185186386108,
      "learning_rate": 1.6209594093132995e-05,
      "loss": 1.016793918609619,
      "step": 17450
    },
    {
      "epoch": 1.8451589891277065,
      "grad_norm": 1.1226716041564941,
      "learning_rate": 1.6183682256256123e-05,
      "loss": 1.0123113632202148,
      "step": 17460
    },
    {
      "epoch": 1.846215834181011,
      "grad_norm": 1.0778731107711792,
      "learning_rate": 1.6157781231201076e-05,
      "loss": 1.0921284675598144,
      "step": 17470
    },
    {
      "epoch": 1.8472726792343157,
      "grad_norm": 1.1274769306182861,
      "learning_rate": 1.6131891049731384e-05,
      "loss": 0.9863181114196777,
      "step": 17480
    },
    {
      "epoch": 1.8483295242876205,
      "grad_norm": 1.747165322303772,
      "learning_rate": 1.610601174359726e-05,
      "loss": 0.9949548721313477,
      "step": 17490
    },
    {
      "epoch": 1.849386369340925,
      "grad_norm": 1.1273820400238037,
      "learning_rate": 1.608014334453559e-05,
      "loss": 0.9662731170654297,
      "step": 17500
    },
    {
      "epoch": 1.849386369340925,
      "eval_loss": 0.4837694466114044,
      "eval_runtime": 3044.1869,
      "eval_samples_per_second": 2.925,
      "eval_steps_per_second": 2.925,
      "step": 17500
    },
    {
      "epoch": 1.8504432143942298,
      "grad_norm": 0.06518067419528961,
      "learning_rate": 1.6054285884269883e-05,
      "loss": 0.8749103546142578,
      "step": 17510
    },
    {
      "epoch": 1.8515000594475342,
      "grad_norm": 1.290062427520752,
      "learning_rate": 1.602843939451023e-05,
      "loss": 1.01004056930542,
      "step": 17520
    },
    {
      "epoch": 1.852556904500839,
      "grad_norm": 0.9690845012664795,
      "learning_rate": 1.6002603906953264e-05,
      "loss": 0.9632661819458008,
      "step": 17530
    },
    {
      "epoch": 1.8536137495541434,
      "grad_norm": 1.2537935972213745,
      "learning_rate": 1.597677945328215e-05,
      "loss": 0.9716744422912598,
      "step": 17540
    },
    {
      "epoch": 1.8546705946074482,
      "grad_norm": 1.3297415971755981,
      "learning_rate": 1.5950966065166477e-05,
      "loss": 0.9754163742065429,
      "step": 17550
    },
    {
      "epoch": 1.8557274396607526,
      "grad_norm": 0.9993301630020142,
      "learning_rate": 1.5925163774262314e-05,
      "loss": 0.8256655693054199,
      "step": 17560
    },
    {
      "epoch": 1.8567842847140574,
      "grad_norm": 1.2453118562698364,
      "learning_rate": 1.5899372612212083e-05,
      "loss": 0.9114871978759765,
      "step": 17570
    },
    {
      "epoch": 1.8578411297673618,
      "grad_norm": 1.0350525379180908,
      "learning_rate": 1.587359261064458e-05,
      "loss": 0.9124351501464844,
      "step": 17580
    },
    {
      "epoch": 1.8588979748206667,
      "grad_norm": 1.0898957252502441,
      "learning_rate": 1.58478238011749e-05,
      "loss": 0.973476505279541,
      "step": 17590
    },
    {
      "epoch": 1.8599548198739713,
      "grad_norm": 1.082497477531433,
      "learning_rate": 1.5822066215404428e-05,
      "loss": 0.868715763092041,
      "step": 17600
    },
    {
      "epoch": 1.8610116649272759,
      "grad_norm": 1.9386441707611084,
      "learning_rate": 1.5796319884920767e-05,
      "loss": 0.9856988906860351,
      "step": 17610
    },
    {
      "epoch": 1.8620685099805805,
      "grad_norm": 1.3330377340316772,
      "learning_rate": 1.5770584841297742e-05,
      "loss": 0.9397290229797364,
      "step": 17620
    },
    {
      "epoch": 1.863125355033885,
      "grad_norm": 1.2687551975250244,
      "learning_rate": 1.57448611160953e-05,
      "loss": 0.9091946601867675,
      "step": 17630
    },
    {
      "epoch": 1.8641822000871897,
      "grad_norm": 1.2734438180923462,
      "learning_rate": 1.5719148740859547e-05,
      "loss": 0.8300318717956543,
      "step": 17640
    },
    {
      "epoch": 1.8652390451404943,
      "grad_norm": 1.5192053318023682,
      "learning_rate": 1.5693447747122636e-05,
      "loss": 0.9630986213684082,
      "step": 17650
    },
    {
      "epoch": 1.866295890193799,
      "grad_norm": 1.1217422485351562,
      "learning_rate": 1.5667758166402795e-05,
      "loss": 0.9419375419616699,
      "step": 17660
    },
    {
      "epoch": 1.8673527352471035,
      "grad_norm": 1.218955397605896,
      "learning_rate": 1.5642080030204225e-05,
      "loss": 0.9753809928894043,
      "step": 17670
    },
    {
      "epoch": 1.8684095803004082,
      "grad_norm": 0.9898252487182617,
      "learning_rate": 1.5616413370017118e-05,
      "loss": 0.9253408432006835,
      "step": 17680
    },
    {
      "epoch": 1.8694664253537128,
      "grad_norm": 1.057386875152588,
      "learning_rate": 1.5590758217317563e-05,
      "loss": 0.942049217224121,
      "step": 17690
    },
    {
      "epoch": 1.8705232704070176,
      "grad_norm": 1.1009392738342285,
      "learning_rate": 1.5565114603567576e-05,
      "loss": 0.9803997993469238,
      "step": 17700
    },
    {
      "epoch": 1.871580115460322,
      "grad_norm": 1.1835993528366089,
      "learning_rate": 1.5539482560214984e-05,
      "loss": 0.8219508171081543,
      "step": 17710
    },
    {
      "epoch": 1.8726369605136268,
      "grad_norm": 1.2400360107421875,
      "learning_rate": 1.5513862118693456e-05,
      "loss": 0.9645069122314454,
      "step": 17720
    },
    {
      "epoch": 1.8736938055669312,
      "grad_norm": 1.765958547592163,
      "learning_rate": 1.5488253310422403e-05,
      "loss": 0.9085442543029785,
      "step": 17730
    },
    {
      "epoch": 1.874750650620236,
      "grad_norm": 1.3000437021255493,
      "learning_rate": 1.5462656166806998e-05,
      "loss": 0.8692684173583984,
      "step": 17740
    },
    {
      "epoch": 1.8758074956735404,
      "grad_norm": 0.9950700998306274,
      "learning_rate": 1.5437070719238088e-05,
      "loss": 0.9573290824890137,
      "step": 17750
    },
    {
      "epoch": 1.8768643407268453,
      "grad_norm": 1.321108102798462,
      "learning_rate": 1.5411496999092196e-05,
      "loss": 0.962657356262207,
      "step": 17760
    },
    {
      "epoch": 1.8779211857801497,
      "grad_norm": 1.261818528175354,
      "learning_rate": 1.5385935037731435e-05,
      "loss": 0.9742824554443359,
      "step": 17770
    },
    {
      "epoch": 1.8789780308334545,
      "grad_norm": 1.0316439867019653,
      "learning_rate": 1.5360384866503533e-05,
      "loss": 0.9861832618713379,
      "step": 17780
    },
    {
      "epoch": 1.880034875886759,
      "grad_norm": 1.1645760536193848,
      "learning_rate": 1.5334846516741726e-05,
      "loss": 0.881318187713623,
      "step": 17790
    },
    {
      "epoch": 1.8810917209400637,
      "grad_norm": 0.9543135762214661,
      "learning_rate": 1.5309320019764783e-05,
      "loss": 0.8560358047485351,
      "step": 17800
    },
    {
      "epoch": 1.8821485659933683,
      "grad_norm": 0.9353059530258179,
      "learning_rate": 1.528380540687691e-05,
      "loss": 0.9967783927917481,
      "step": 17810
    },
    {
      "epoch": 1.883205411046673,
      "grad_norm": 1.653192162513733,
      "learning_rate": 1.5258302709367758e-05,
      "loss": 0.9072349548339844,
      "step": 17820
    },
    {
      "epoch": 1.8842622560999776,
      "grad_norm": 1.3922604322433472,
      "learning_rate": 1.5232811958512355e-05,
      "loss": 1.0420916557312012,
      "step": 17830
    },
    {
      "epoch": 1.8853191011532822,
      "grad_norm": 1.8204625844955444,
      "learning_rate": 1.5207333185571091e-05,
      "loss": 0.8548428535461425,
      "step": 17840
    },
    {
      "epoch": 1.8863759462065868,
      "grad_norm": 1.2377101182937622,
      "learning_rate": 1.5181866421789653e-05,
      "loss": 0.842220401763916,
      "step": 17850
    },
    {
      "epoch": 1.8874327912598914,
      "grad_norm": 1.4535653591156006,
      "learning_rate": 1.5156411698398998e-05,
      "loss": 0.9752124786376953,
      "step": 17860
    },
    {
      "epoch": 1.888489636313196,
      "grad_norm": 1.7208360433578491,
      "learning_rate": 1.513096904661534e-05,
      "loss": 0.9729183197021485,
      "step": 17870
    },
    {
      "epoch": 1.8895464813665006,
      "grad_norm": 1.0490894317626953,
      "learning_rate": 1.5105538497640065e-05,
      "loss": 1.0051823616027833,
      "step": 17880
    },
    {
      "epoch": 1.8906033264198054,
      "grad_norm": 1.3072855472564697,
      "learning_rate": 1.5080120082659732e-05,
      "loss": 0.8698942184448242,
      "step": 17890
    },
    {
      "epoch": 1.8916601714731098,
      "grad_norm": 1.3858368396759033,
      "learning_rate": 1.505471383284601e-05,
      "loss": 1.001598072052002,
      "step": 17900
    },
    {
      "epoch": 1.8927170165264147,
      "grad_norm": 1.119990587234497,
      "learning_rate": 1.5029319779355657e-05,
      "loss": 0.9451250076293946,
      "step": 17910
    },
    {
      "epoch": 1.893773861579719,
      "grad_norm": 1.294818639755249,
      "learning_rate": 1.5003937953330466e-05,
      "loss": 0.9376727104187011,
      "step": 17920
    },
    {
      "epoch": 1.8948307066330239,
      "grad_norm": 0.8632761240005493,
      "learning_rate": 1.4978568385897245e-05,
      "loss": 0.9278356552124023,
      "step": 17930
    },
    {
      "epoch": 1.8958875516863283,
      "grad_norm": 0.990278959274292,
      "learning_rate": 1.4953211108167759e-05,
      "loss": 1.0064745903015138,
      "step": 17940
    },
    {
      "epoch": 1.896944396739633,
      "grad_norm": 1.424938440322876,
      "learning_rate": 1.4927866151238707e-05,
      "loss": 0.9627618789672852,
      "step": 17950
    },
    {
      "epoch": 1.8980012417929375,
      "grad_norm": 1.247955560684204,
      "learning_rate": 1.4902533546191675e-05,
      "loss": 0.9008722305297852,
      "step": 17960
    },
    {
      "epoch": 1.8990580868462423,
      "grad_norm": 1.1273075342178345,
      "learning_rate": 1.487721332409311e-05,
      "loss": 0.9771639823913574,
      "step": 17970
    },
    {
      "epoch": 1.9001149318995467,
      "grad_norm": 0.762605607509613,
      "learning_rate": 1.4851905515994258e-05,
      "loss": 0.9498398780822754,
      "step": 17980
    },
    {
      "epoch": 1.9011717769528516,
      "grad_norm": 2.0653393268585205,
      "learning_rate": 1.4826610152931164e-05,
      "loss": 0.8865217208862305,
      "step": 17990
    },
    {
      "epoch": 1.9022286220061562,
      "grad_norm": 1.2799004316329956,
      "learning_rate": 1.480132726592458e-05,
      "loss": 0.9741304397583008,
      "step": 18000
    },
    {
      "epoch": 1.9022286220061562,
      "eval_loss": 0.4830879271030426,
      "eval_runtime": 3040.2143,
      "eval_samples_per_second": 2.929,
      "eval_steps_per_second": 2.929,
      "step": 18000
    },
    {
      "epoch": 1.9032854670594608,
      "grad_norm": 1.1796945333480835,
      "learning_rate": 1.4776056885979988e-05,
      "loss": 0.9963871955871582,
      "step": 18010
    },
    {
      "epoch": 1.9043423121127654,
      "grad_norm": 0.9647114872932434,
      "learning_rate": 1.4750799044087513e-05,
      "loss": 1.006678581237793,
      "step": 18020
    },
    {
      "epoch": 1.90539915716607,
      "grad_norm": 0.8601939678192139,
      "learning_rate": 1.4725553771221918e-05,
      "loss": 0.9009536743164063,
      "step": 18030
    },
    {
      "epoch": 1.9064560022193746,
      "grad_norm": 1.5132955312728882,
      "learning_rate": 1.4700321098342532e-05,
      "loss": 0.8412443161010742,
      "step": 18040
    },
    {
      "epoch": 1.9075128472726792,
      "grad_norm": 0.8298953175544739,
      "learning_rate": 1.4675101056393259e-05,
      "loss": 0.9263592720031738,
      "step": 18050
    },
    {
      "epoch": 1.9085696923259838,
      "grad_norm": 1.447222352027893,
      "learning_rate": 1.4649893676302484e-05,
      "loss": 0.9911669731140137,
      "step": 18060
    },
    {
      "epoch": 1.9096265373792884,
      "grad_norm": 1.3189270496368408,
      "learning_rate": 1.4624698988983093e-05,
      "loss": 0.9665737152099609,
      "step": 18070
    },
    {
      "epoch": 1.910683382432593,
      "grad_norm": 0.9956954121589661,
      "learning_rate": 1.4599517025332382e-05,
      "loss": 0.9582039833068847,
      "step": 18080
    },
    {
      "epoch": 1.9117402274858977,
      "grad_norm": 1.7711209058761597,
      "learning_rate": 1.4574347816232065e-05,
      "loss": 0.9661527633666992,
      "step": 18090
    },
    {
      "epoch": 1.9127970725392025,
      "grad_norm": 1.2257004976272583,
      "learning_rate": 1.4549191392548183e-05,
      "loss": 0.9142245292663574,
      "step": 18100
    },
    {
      "epoch": 1.913853917592507,
      "grad_norm": 0.9886197447776794,
      "learning_rate": 1.4524047785131153e-05,
      "loss": 0.9936203956604004,
      "step": 18110
    },
    {
      "epoch": 1.9149107626458117,
      "grad_norm": 1.3881423473358154,
      "learning_rate": 1.44989170248156e-05,
      "loss": 0.8859989166259765,
      "step": 18120
    },
    {
      "epoch": 1.9159676076991161,
      "grad_norm": 1.1195544004440308,
      "learning_rate": 1.4473799142420461e-05,
      "loss": 0.7937178134918212,
      "step": 18130
    },
    {
      "epoch": 1.917024452752421,
      "grad_norm": 1.0878417491912842,
      "learning_rate": 1.4448694168748832e-05,
      "loss": 0.9974874496459961,
      "step": 18140
    },
    {
      "epoch": 1.9180812978057253,
      "grad_norm": 1.1011698246002197,
      "learning_rate": 1.4423602134588037e-05,
      "loss": 0.9105242729187012,
      "step": 18150
    },
    {
      "epoch": 1.9191381428590302,
      "grad_norm": 1.4028321504592896,
      "learning_rate": 1.4398523070709447e-05,
      "loss": 0.9755278587341308,
      "step": 18160
    },
    {
      "epoch": 1.9201949879123346,
      "grad_norm": 1.2676739692687988,
      "learning_rate": 1.437345700786861e-05,
      "loss": 0.9609894752502441,
      "step": 18170
    },
    {
      "epoch": 1.9212518329656394,
      "grad_norm": 1.4375134706497192,
      "learning_rate": 1.4348403976805069e-05,
      "loss": 0.9367487907409668,
      "step": 18180
    },
    {
      "epoch": 1.9223086780189438,
      "grad_norm": 0.8992677330970764,
      "learning_rate": 1.4323364008242446e-05,
      "loss": 0.9339933395385742,
      "step": 18190
    },
    {
      "epoch": 1.9233655230722486,
      "grad_norm": 1.4033421277999878,
      "learning_rate": 1.4298337132888268e-05,
      "loss": 1.0861198425292968,
      "step": 18200
    },
    {
      "epoch": 1.9244223681255532,
      "grad_norm": 1.5736299753189087,
      "learning_rate": 1.4273323381434078e-05,
      "loss": 0.8618276596069336,
      "step": 18210
    },
    {
      "epoch": 1.9254792131788578,
      "grad_norm": 1.0226414203643799,
      "learning_rate": 1.4248322784555267e-05,
      "loss": 0.9526123046875,
      "step": 18220
    },
    {
      "epoch": 1.9265360582321625,
      "grad_norm": 1.0963102579116821,
      "learning_rate": 1.4223335372911158e-05,
      "loss": 0.8445890426635743,
      "step": 18230
    },
    {
      "epoch": 1.927592903285467,
      "grad_norm": 1.1241637468338013,
      "learning_rate": 1.4198361177144823e-05,
      "loss": 1.0155341148376464,
      "step": 18240
    },
    {
      "epoch": 1.9286497483387717,
      "grad_norm": 0.7999724745750427,
      "learning_rate": 1.4173400227883198e-05,
      "loss": 0.8864949226379395,
      "step": 18250
    },
    {
      "epoch": 1.9297065933920763,
      "grad_norm": 1.011460781097412,
      "learning_rate": 1.414845255573693e-05,
      "loss": 0.9812465667724609,
      "step": 18260
    },
    {
      "epoch": 1.930763438445381,
      "grad_norm": 0.9159804582595825,
      "learning_rate": 1.4123518191300428e-05,
      "loss": 0.8984330177307129,
      "step": 18270
    },
    {
      "epoch": 1.9318202834986855,
      "grad_norm": 1.0511664152145386,
      "learning_rate": 1.4098597165151717e-05,
      "loss": 0.9282947540283203,
      "step": 18280
    },
    {
      "epoch": 1.9328771285519901,
      "grad_norm": 1.3192466497421265,
      "learning_rate": 1.4073689507852527e-05,
      "loss": 1.0243149757385255,
      "step": 18290
    },
    {
      "epoch": 1.9339339736052947,
      "grad_norm": 1.1828497648239136,
      "learning_rate": 1.4048795249948148e-05,
      "loss": 0.9151687622070312,
      "step": 18300
    },
    {
      "epoch": 1.9349908186585996,
      "grad_norm": 1.1846466064453125,
      "learning_rate": 1.4023914421967487e-05,
      "loss": 0.9660979270935058,
      "step": 18310
    },
    {
      "epoch": 1.936047663711904,
      "grad_norm": 1.395215630531311,
      "learning_rate": 1.3999047054422915e-05,
      "loss": 0.9342232704162597,
      "step": 18320
    },
    {
      "epoch": 1.9371045087652088,
      "grad_norm": 1.2342299222946167,
      "learning_rate": 1.397419317781036e-05,
      "loss": 0.92709321975708,
      "step": 18330
    },
    {
      "epoch": 1.9381613538185132,
      "grad_norm": 1.1838783025741577,
      "learning_rate": 1.3949352822609158e-05,
      "loss": 0.9767477989196778,
      "step": 18340
    },
    {
      "epoch": 1.939218198871818,
      "grad_norm": 1.0344548225402832,
      "learning_rate": 1.3924526019282114e-05,
      "loss": 0.952883243560791,
      "step": 18350
    },
    {
      "epoch": 1.9402750439251224,
      "grad_norm": 1.2441740036010742,
      "learning_rate": 1.3899712798275338e-05,
      "loss": 0.9649330139160156,
      "step": 18360
    },
    {
      "epoch": 1.9413318889784272,
      "grad_norm": 1.2067946195602417,
      "learning_rate": 1.3874913190018363e-05,
      "loss": 1.1383502960205079,
      "step": 18370
    },
    {
      "epoch": 1.9423887340317316,
      "grad_norm": 1.0177148580551147,
      "learning_rate": 1.385012722492397e-05,
      "loss": 0.9787127494812011,
      "step": 18380
    },
    {
      "epoch": 1.9434455790850365,
      "grad_norm": 0.9557192325592041,
      "learning_rate": 1.3825354933388269e-05,
      "loss": 0.897181510925293,
      "step": 18390
    },
    {
      "epoch": 1.944502424138341,
      "grad_norm": 1.5747487545013428,
      "learning_rate": 1.3800596345790523e-05,
      "loss": 0.9553994178771973,
      "step": 18400
    },
    {
      "epoch": 1.9455592691916457,
      "grad_norm": 1.0466935634613037,
      "learning_rate": 1.3775851492493242e-05,
      "loss": 0.8787156105041504,
      "step": 18410
    },
    {
      "epoch": 1.9466161142449503,
      "grad_norm": 1.2227885723114014,
      "learning_rate": 1.37511204038421e-05,
      "loss": 0.946174144744873,
      "step": 18420
    },
    {
      "epoch": 1.947672959298255,
      "grad_norm": 1.0443825721740723,
      "learning_rate": 1.3726403110165853e-05,
      "loss": 1.0222416877746583,
      "step": 18430
    },
    {
      "epoch": 1.9487298043515595,
      "grad_norm": 1.4991309642791748,
      "learning_rate": 1.3701699641776373e-05,
      "loss": 0.8525497436523437,
      "step": 18440
    },
    {
      "epoch": 1.9497866494048641,
      "grad_norm": 1.3419233560562134,
      "learning_rate": 1.3677010028968546e-05,
      "loss": 0.9364970207214356,
      "step": 18450
    },
    {
      "epoch": 1.9508434944581687,
      "grad_norm": 1.253547191619873,
      "learning_rate": 1.3652334302020303e-05,
      "loss": 0.9723645210266113,
      "step": 18460
    },
    {
      "epoch": 1.9519003395114733,
      "grad_norm": 1.5022222995758057,
      "learning_rate": 1.3627672491192522e-05,
      "loss": 0.9641678810119629,
      "step": 18470
    },
    {
      "epoch": 1.952957184564778,
      "grad_norm": 1.2753891944885254,
      "learning_rate": 1.3603024626729021e-05,
      "loss": 0.9783175468444825,
      "step": 18480
    },
    {
      "epoch": 1.9540140296180826,
      "grad_norm": 1.638152837753296,
      "learning_rate": 1.35783907388565e-05,
      "loss": 0.956610107421875,
      "step": 18490
    },
    {
      "epoch": 1.9550708746713874,
      "grad_norm": 1.3578416109085083,
      "learning_rate": 1.3553770857784553e-05,
      "loss": 0.9685772895812989,
      "step": 18500
    },
    {
      "epoch": 1.9550708746713874,
      "eval_loss": 0.4817064702510834,
      "eval_runtime": 3413.9531,
      "eval_samples_per_second": 2.608,
      "eval_steps_per_second": 2.608,
      "step": 18500
    },
    {
      "epoch": 1.9561277197246918,
      "grad_norm": 1.360504150390625,
      "learning_rate": 1.3529165013705574e-05,
      "loss": 0.911770248413086,
      "step": 18510
    },
    {
      "epoch": 1.9571845647779966,
      "grad_norm": 1.4865000247955322,
      "learning_rate": 1.3504573236794742e-05,
      "loss": 0.8905097961425781,
      "step": 18520
    },
    {
      "epoch": 1.958241409831301,
      "grad_norm": 1.1035640239715576,
      "learning_rate": 1.3479995557209978e-05,
      "loss": 0.9842826843261718,
      "step": 18530
    },
    {
      "epoch": 1.9592982548846058,
      "grad_norm": 1.4912103414535522,
      "learning_rate": 1.3455432005091944e-05,
      "loss": 0.9721610069274902,
      "step": 18540
    },
    {
      "epoch": 1.9603550999379102,
      "grad_norm": 1.5399175882339478,
      "learning_rate": 1.3430882610563955e-05,
      "loss": 0.9186005592346191,
      "step": 18550
    },
    {
      "epoch": 1.961411944991215,
      "grad_norm": 1.2922111749649048,
      "learning_rate": 1.3406347403731964e-05,
      "loss": 0.8527621269226074,
      "step": 18560
    },
    {
      "epoch": 1.9624687900445195,
      "grad_norm": 1.017149806022644,
      "learning_rate": 1.3381826414684518e-05,
      "loss": 0.8733506202697754,
      "step": 18570
    },
    {
      "epoch": 1.9635256350978243,
      "grad_norm": 0.9372082948684692,
      "learning_rate": 1.3357319673492761e-05,
      "loss": 0.8834692001342773,
      "step": 18580
    },
    {
      "epoch": 1.9645824801511287,
      "grad_norm": 1.6242564916610718,
      "learning_rate": 1.3332827210210327e-05,
      "loss": 0.9070075035095215,
      "step": 18590
    },
    {
      "epoch": 1.9656393252044335,
      "grad_norm": 1.448710322380066,
      "learning_rate": 1.3308349054873363e-05,
      "loss": 0.971399974822998,
      "step": 18600
    },
    {
      "epoch": 1.9666961702577381,
      "grad_norm": 1.3200303316116333,
      "learning_rate": 1.3283885237500446e-05,
      "loss": 0.8751208305358886,
      "step": 18610
    },
    {
      "epoch": 1.9677530153110427,
      "grad_norm": 0.9964113235473633,
      "learning_rate": 1.3259435788092606e-05,
      "loss": 0.9694125175476074,
      "step": 18620
    },
    {
      "epoch": 1.9688098603643474,
      "grad_norm": 1.347999095916748,
      "learning_rate": 1.3235000736633227e-05,
      "loss": 1.0132827758789062,
      "step": 18630
    },
    {
      "epoch": 1.969866705417652,
      "grad_norm": 1.0527310371398926,
      "learning_rate": 1.3210580113088038e-05,
      "loss": 1.0048596382141113,
      "step": 18640
    },
    {
      "epoch": 1.9709235504709566,
      "grad_norm": 1.059240460395813,
      "learning_rate": 1.3186173947405073e-05,
      "loss": 0.8814691543579102,
      "step": 18650
    },
    {
      "epoch": 1.9719803955242612,
      "grad_norm": 1.114592432975769,
      "learning_rate": 1.3161782269514655e-05,
      "loss": 0.9698432922363281,
      "step": 18660
    },
    {
      "epoch": 1.9730372405775658,
      "grad_norm": 1.0070087909698486,
      "learning_rate": 1.313740510932932e-05,
      "loss": 0.8863020896911621,
      "step": 18670
    },
    {
      "epoch": 1.9740940856308704,
      "grad_norm": 1.2769526243209839,
      "learning_rate": 1.3113042496743805e-05,
      "loss": 0.9521626472473145,
      "step": 18680
    },
    {
      "epoch": 1.975150930684175,
      "grad_norm": 1.1237651109695435,
      "learning_rate": 1.3088694461634998e-05,
      "loss": 0.8230314254760742,
      "step": 18690
    },
    {
      "epoch": 1.9762077757374796,
      "grad_norm": 1.28242826461792,
      "learning_rate": 1.3064361033861944e-05,
      "loss": 0.8378606796264648,
      "step": 18700
    },
    {
      "epoch": 1.9772646207907845,
      "grad_norm": 1.3443553447723389,
      "learning_rate": 1.3040042243265732e-05,
      "loss": 0.9253696441650391,
      "step": 18710
    },
    {
      "epoch": 1.9783214658440889,
      "grad_norm": 1.0326238870620728,
      "learning_rate": 1.3015738119669527e-05,
      "loss": 0.7609794616699219,
      "step": 18720
    },
    {
      "epoch": 1.9793783108973937,
      "grad_norm": 0.9612667560577393,
      "learning_rate": 1.2991448692878489e-05,
      "loss": 0.8756623268127441,
      "step": 18730
    },
    {
      "epoch": 1.980435155950698,
      "grad_norm": 1.248964548110962,
      "learning_rate": 1.296717399267978e-05,
      "loss": 0.9373766899108886,
      "step": 18740
    },
    {
      "epoch": 1.981492001004003,
      "grad_norm": 1.1252400875091553,
      "learning_rate": 1.294291404884248e-05,
      "loss": 0.9439261436462403,
      "step": 18750
    },
    {
      "epoch": 1.9825488460573073,
      "grad_norm": 1.1698882579803467,
      "learning_rate": 1.2918668891117583e-05,
      "loss": 0.9446320533752441,
      "step": 18760
    },
    {
      "epoch": 1.9836056911106121,
      "grad_norm": 1.0439155101776123,
      "learning_rate": 1.2894438549237935e-05,
      "loss": 0.9832762718200684,
      "step": 18770
    },
    {
      "epoch": 1.9846625361639165,
      "grad_norm": 1.0949651002883911,
      "learning_rate": 1.2870223052918245e-05,
      "loss": 0.9804287910461426,
      "step": 18780
    },
    {
      "epoch": 1.9857193812172214,
      "grad_norm": 0.9793189764022827,
      "learning_rate": 1.2846022431854985e-05,
      "loss": 0.9093479156494141,
      "step": 18790
    },
    {
      "epoch": 1.9867762262705257,
      "grad_norm": 1.385335087776184,
      "learning_rate": 1.2821836715726398e-05,
      "loss": 0.8733051300048829,
      "step": 18800
    },
    {
      "epoch": 1.9878330713238306,
      "grad_norm": 0.9694952964782715,
      "learning_rate": 1.2797665934192435e-05,
      "loss": 0.9776005744934082,
      "step": 18810
    },
    {
      "epoch": 1.9888899163771352,
      "grad_norm": 1.3093479871749878,
      "learning_rate": 1.2773510116894766e-05,
      "loss": 0.8875185012817383,
      "step": 18820
    },
    {
      "epoch": 1.9899467614304398,
      "grad_norm": 0.9388787150382996,
      "learning_rate": 1.2749369293456672e-05,
      "loss": 0.9124444961547852,
      "step": 18830
    },
    {
      "epoch": 1.9910036064837444,
      "grad_norm": 1.6103373765945435,
      "learning_rate": 1.272524349348307e-05,
      "loss": 0.9413721084594726,
      "step": 18840
    },
    {
      "epoch": 1.992060451537049,
      "grad_norm": 0.9924553632736206,
      "learning_rate": 1.2701132746560423e-05,
      "loss": 0.9896018981933594,
      "step": 18850
    },
    {
      "epoch": 1.9931172965903536,
      "grad_norm": 1.401532769203186,
      "learning_rate": 1.2677037082256788e-05,
      "loss": 0.9367332458496094,
      "step": 18860
    },
    {
      "epoch": 1.9941741416436582,
      "grad_norm": 1.1648004055023193,
      "learning_rate": 1.2652956530121679e-05,
      "loss": 0.8520607948303223,
      "step": 18870
    },
    {
      "epoch": 1.9952309866969629,
      "grad_norm": 1.1731551885604858,
      "learning_rate": 1.2628891119686087e-05,
      "loss": 0.9384458541870118,
      "step": 18880
    },
    {
      "epoch": 1.9962878317502675,
      "grad_norm": 1.0429105758666992,
      "learning_rate": 1.2604840880462432e-05,
      "loss": 0.9205928802490234,
      "step": 18890
    },
    {
      "epoch": 1.997344676803572,
      "grad_norm": 1.1114870309829712,
      "learning_rate": 1.2580805841944556e-05,
      "loss": 0.9092113494873046,
      "step": 18900
    },
    {
      "epoch": 1.9984015218568767,
      "grad_norm": 1.0932546854019165,
      "learning_rate": 1.255678603360763e-05,
      "loss": 0.9752060890197753,
      "step": 18910
    },
    {
      "epoch": 1.9994583669101815,
      "grad_norm": 1.155789852142334,
      "learning_rate": 1.2532781484908152e-05,
      "loss": 0.9143109321594238,
      "step": 18920
    },
    {
      "epoch": 2.0004227380213218,
      "grad_norm": 1.3319833278656006,
      "learning_rate": 1.2508792225283907e-05,
      "loss": 0.77767014503479,
      "step": 18930
    },
    {
      "epoch": 2.0014795830746266,
      "grad_norm": 0.9162007570266724,
      "learning_rate": 1.2484818284153952e-05,
      "loss": 0.8607779502868652,
      "step": 18940
    },
    {
      "epoch": 2.002536428127931,
      "grad_norm": 1.0594452619552612,
      "learning_rate": 1.246085969091853e-05,
      "loss": 1.0024656295776366,
      "step": 18950
    },
    {
      "epoch": 2.003593273181236,
      "grad_norm": 0.9745901823043823,
      "learning_rate": 1.2436916474959076e-05,
      "loss": 0.8703121185302735,
      "step": 18960
    },
    {
      "epoch": 2.00465011823454,
      "grad_norm": 1.0322976112365723,
      "learning_rate": 1.2412988665638162e-05,
      "loss": 0.8727169036865234,
      "step": 18970
    },
    {
      "epoch": 2.005706963287845,
      "grad_norm": 0.9298354983329773,
      "learning_rate": 1.2389076292299456e-05,
      "loss": 0.9847866058349609,
      "step": 18980
    },
    {
      "epoch": 2.0067638083411494,
      "grad_norm": 1.7697001695632935,
      "learning_rate": 1.2365179384267736e-05,
      "loss": 0.8769575119018554,
      "step": 18990
    },
    {
      "epoch": 2.0078206533944543,
      "grad_norm": 1.231011152267456,
      "learning_rate": 1.234129797084877e-05,
      "loss": 0.8207708358764648,
      "step": 19000
    },
    {
      "epoch": 2.0078206533944543,
      "eval_loss": 0.4859752655029297,
      "eval_runtime": 3415.2816,
      "eval_samples_per_second": 2.607,
      "eval_steps_per_second": 2.607,
      "step": 19000
    },
    {
      "epoch": 2.0088774984477586,
      "grad_norm": 1.5658514499664307,
      "learning_rate": 1.2317432081329345e-05,
      "loss": 0.8333698272705078,
      "step": 19010
    },
    {
      "epoch": 2.0099343435010635,
      "grad_norm": 1.0395725965499878,
      "learning_rate": 1.2293581744977195e-05,
      "loss": 0.9051530838012696,
      "step": 19020
    },
    {
      "epoch": 2.010991188554368,
      "grad_norm": 1.334608793258667,
      "learning_rate": 1.2269746991041011e-05,
      "loss": 0.9400140762329101,
      "step": 19030
    },
    {
      "epoch": 2.0120480336076727,
      "grad_norm": 1.3669302463531494,
      "learning_rate": 1.2245927848750352e-05,
      "loss": 0.9139886856079101,
      "step": 19040
    },
    {
      "epoch": 2.0131048786609775,
      "grad_norm": 1.0966477394104004,
      "learning_rate": 1.2222124347315628e-05,
      "loss": 0.7775619029998779,
      "step": 19050
    },
    {
      "epoch": 2.014161723714282,
      "grad_norm": 1.2230108976364136,
      "learning_rate": 1.219833651592807e-05,
      "loss": 0.839134407043457,
      "step": 19060
    },
    {
      "epoch": 2.0152185687675868,
      "grad_norm": 1.476204752922058,
      "learning_rate": 1.2174564383759715e-05,
      "loss": 0.9009620666503906,
      "step": 19070
    },
    {
      "epoch": 2.016275413820891,
      "grad_norm": 1.0673693418502808,
      "learning_rate": 1.2150807979963321e-05,
      "loss": 0.917660903930664,
      "step": 19080
    },
    {
      "epoch": 2.017332258874196,
      "grad_norm": 2.316054105758667,
      "learning_rate": 1.212706733367237e-05,
      "loss": 0.919400405883789,
      "step": 19090
    },
    {
      "epoch": 2.0183891039275004,
      "grad_norm": 1.47184157371521,
      "learning_rate": 1.2103342474001e-05,
      "loss": 0.9117672920227051,
      "step": 19100
    },
    {
      "epoch": 2.019445948980805,
      "grad_norm": 1.2421603202819824,
      "learning_rate": 1.2079633430044032e-05,
      "loss": 0.9053256988525391,
      "step": 19110
    },
    {
      "epoch": 2.0205027940341096,
      "grad_norm": 1.4530417919158936,
      "learning_rate": 1.2055940230876853e-05,
      "loss": 0.8508424758911133,
      "step": 19120
    },
    {
      "epoch": 2.0215596390874144,
      "grad_norm": 1.3169317245483398,
      "learning_rate": 1.203226290555543e-05,
      "loss": 0.8872398376464844,
      "step": 19130
    },
    {
      "epoch": 2.022616484140719,
      "grad_norm": 1.8090072870254517,
      "learning_rate": 1.2008601483116255e-05,
      "loss": 0.8384584426879883,
      "step": 19140
    },
    {
      "epoch": 2.0236733291940237,
      "grad_norm": 1.4061241149902344,
      "learning_rate": 1.198495599257635e-05,
      "loss": 0.8707221031188965,
      "step": 19150
    },
    {
      "epoch": 2.024730174247328,
      "grad_norm": 1.6427223682403564,
      "learning_rate": 1.1961326462933162e-05,
      "loss": 0.8639749526977539,
      "step": 19160
    },
    {
      "epoch": 2.025787019300633,
      "grad_norm": 1.1009454727172852,
      "learning_rate": 1.1937712923164579e-05,
      "loss": 0.8300027847290039,
      "step": 19170
    },
    {
      "epoch": 2.0268438643539373,
      "grad_norm": 1.7804609537124634,
      "learning_rate": 1.1914115402228873e-05,
      "loss": 0.8962211608886719,
      "step": 19180
    },
    {
      "epoch": 2.027900709407242,
      "grad_norm": 0.8853987455368042,
      "learning_rate": 1.1890533929064693e-05,
      "loss": 0.7668407440185547,
      "step": 19190
    },
    {
      "epoch": 2.0289575544605465,
      "grad_norm": 1.2284526824951172,
      "learning_rate": 1.1866968532590989e-05,
      "loss": 0.8714509010314941,
      "step": 19200
    },
    {
      "epoch": 2.0300143995138513,
      "grad_norm": 1.24924635887146,
      "learning_rate": 1.1843419241706994e-05,
      "loss": 0.9602168083190918,
      "step": 19210
    },
    {
      "epoch": 2.0310712445671557,
      "grad_norm": 1.003387451171875,
      "learning_rate": 1.1819886085292186e-05,
      "loss": 0.790290117263794,
      "step": 19220
    },
    {
      "epoch": 2.0321280896204605,
      "grad_norm": 1.5093246698379517,
      "learning_rate": 1.1796369092206288e-05,
      "loss": 0.9144254684448242,
      "step": 19230
    },
    {
      "epoch": 2.033184934673765,
      "grad_norm": 1.6490952968597412,
      "learning_rate": 1.1772868291289172e-05,
      "loss": 0.8733613014221191,
      "step": 19240
    },
    {
      "epoch": 2.0342417797270698,
      "grad_norm": 1.2598282098770142,
      "learning_rate": 1.1749383711360854e-05,
      "loss": 0.9332692146301269,
      "step": 19250
    },
    {
      "epoch": 2.0352986247803746,
      "grad_norm": 1.2390943765640259,
      "learning_rate": 1.1725915381221459e-05,
      "loss": 0.8605259895324707,
      "step": 19260
    },
    {
      "epoch": 2.036355469833679,
      "grad_norm": 1.1741539239883423,
      "learning_rate": 1.1702463329651204e-05,
      "loss": 0.9551490783691406,
      "step": 19270
    },
    {
      "epoch": 2.037412314886984,
      "grad_norm": 1.3198668956756592,
      "learning_rate": 1.1679027585410318e-05,
      "loss": 0.7725104808807373,
      "step": 19280
    },
    {
      "epoch": 2.038469159940288,
      "grad_norm": 0.9841579794883728,
      "learning_rate": 1.1655608177239046e-05,
      "loss": 0.9345494270324707,
      "step": 19290
    },
    {
      "epoch": 2.039526004993593,
      "grad_norm": 1.408856987953186,
      "learning_rate": 1.1632205133857583e-05,
      "loss": 0.8708246231079102,
      "step": 19300
    },
    {
      "epoch": 2.0405828500468974,
      "grad_norm": 0.9056268930435181,
      "learning_rate": 1.1608818483966085e-05,
      "loss": 0.9127551078796386,
      "step": 19310
    },
    {
      "epoch": 2.0416396951002023,
      "grad_norm": 1.2289749383926392,
      "learning_rate": 1.1585448256244577e-05,
      "loss": 0.8922224998474121,
      "step": 19320
    },
    {
      "epoch": 2.0426965401535067,
      "grad_norm": 1.1148377656936646,
      "learning_rate": 1.1562094479352952e-05,
      "loss": 0.8635320663452148,
      "step": 19330
    },
    {
      "epoch": 2.0437533852068115,
      "grad_norm": 1.1285805702209473,
      "learning_rate": 1.1538757181930926e-05,
      "loss": 0.8017943382263184,
      "step": 19340
    },
    {
      "epoch": 2.044810230260116,
      "grad_norm": 1.5179861783981323,
      "learning_rate": 1.1515436392598025e-05,
      "loss": 0.9235546112060546,
      "step": 19350
    },
    {
      "epoch": 2.0458670753134207,
      "grad_norm": 1.690525770187378,
      "learning_rate": 1.1492132139953504e-05,
      "loss": 0.8365184783935546,
      "step": 19360
    },
    {
      "epoch": 2.046923920366725,
      "grad_norm": 1.0142967700958252,
      "learning_rate": 1.1468844452576354e-05,
      "loss": 0.7580907344818115,
      "step": 19370
    },
    {
      "epoch": 2.04798076542003,
      "grad_norm": 1.1472538709640503,
      "learning_rate": 1.1445573359025235e-05,
      "loss": 0.8970897674560547,
      "step": 19380
    },
    {
      "epoch": 2.0490376104733343,
      "grad_norm": 1.1511974334716797,
      "learning_rate": 1.142231888783849e-05,
      "loss": 0.8467702865600586,
      "step": 19390
    },
    {
      "epoch": 2.050094455526639,
      "grad_norm": 1.6069294214248657,
      "learning_rate": 1.1399081067534046e-05,
      "loss": 0.8725573539733886,
      "step": 19400
    },
    {
      "epoch": 2.0511513005799435,
      "grad_norm": 1.0612343549728394,
      "learning_rate": 1.137585992660942e-05,
      "loss": 0.8278934478759765,
      "step": 19410
    },
    {
      "epoch": 2.0522081456332484,
      "grad_norm": 1.208942174911499,
      "learning_rate": 1.1352655493541667e-05,
      "loss": 0.8727795600891113,
      "step": 19420
    },
    {
      "epoch": 2.0532649906865528,
      "grad_norm": 1.2241053581237793,
      "learning_rate": 1.1329467796787377e-05,
      "loss": 0.8124959945678711,
      "step": 19430
    },
    {
      "epoch": 2.0543218357398576,
      "grad_norm": 1.2027287483215332,
      "learning_rate": 1.1306296864782595e-05,
      "loss": 0.7925668239593506,
      "step": 19440
    },
    {
      "epoch": 2.055378680793162,
      "grad_norm": 1.4054142236709595,
      "learning_rate": 1.1283142725942808e-05,
      "loss": 0.98065185546875,
      "step": 19450
    },
    {
      "epoch": 2.056435525846467,
      "grad_norm": 1.1359169483184814,
      "learning_rate": 1.1260005408662901e-05,
      "loss": 0.9188995361328125,
      "step": 19460
    },
    {
      "epoch": 2.0574923708997717,
      "grad_norm": 1.4521484375,
      "learning_rate": 1.1236884941317164e-05,
      "loss": 0.8247655868530274,
      "step": 19470
    },
    {
      "epoch": 2.058549215953076,
      "grad_norm": 1.2090479135513306,
      "learning_rate": 1.1213781352259187e-05,
      "loss": 0.809943675994873,
      "step": 19480
    },
    {
      "epoch": 2.059606061006381,
      "grad_norm": 1.0523725748062134,
      "learning_rate": 1.1190694669821874e-05,
      "loss": 0.9451144218444825,
      "step": 19490
    },
    {
      "epoch": 2.0606629060596853,
      "grad_norm": 1.0714218616485596,
      "learning_rate": 1.116762492231739e-05,
      "loss": 0.9067349433898926,
      "step": 19500
    },
    {
      "epoch": 2.0606629060596853,
      "eval_loss": 0.48665887117385864,
      "eval_runtime": 3406.4117,
      "eval_samples_per_second": 2.614,
      "eval_steps_per_second": 2.614,
      "step": 19500
    },
    {
      "epoch": 2.06171975111299,
      "grad_norm": 1.659161925315857,
      "learning_rate": 1.1144572138037155e-05,
      "loss": 0.8405235290527344,
      "step": 19510
    },
    {
      "epoch": 2.0627765961662945,
      "grad_norm": 1.0032196044921875,
      "learning_rate": 1.1121536345251768e-05,
      "loss": 0.8048068046569824,
      "step": 19520
    },
    {
      "epoch": 2.0638334412195993,
      "grad_norm": 1.3926972150802612,
      "learning_rate": 1.1098517572210967e-05,
      "loss": 0.8898597717285156,
      "step": 19530
    },
    {
      "epoch": 2.0648902862729037,
      "grad_norm": 1.2451847791671753,
      "learning_rate": 1.107551584714367e-05,
      "loss": 0.8828273773193359,
      "step": 19540
    },
    {
      "epoch": 2.0659471313262086,
      "grad_norm": 2.326307773590088,
      "learning_rate": 1.105253119825784e-05,
      "loss": 0.7767197132110596,
      "step": 19550
    },
    {
      "epoch": 2.067003976379513,
      "grad_norm": 1.0408319234848022,
      "learning_rate": 1.1029563653740551e-05,
      "loss": 0.9306610107421875,
      "step": 19560
    },
    {
      "epoch": 2.0680608214328178,
      "grad_norm": 1.2963956594467163,
      "learning_rate": 1.1006613241757834e-05,
      "loss": 0.8577400207519531,
      "step": 19570
    },
    {
      "epoch": 2.069117666486122,
      "grad_norm": 1.4078346490859985,
      "learning_rate": 1.0983679990454774e-05,
      "loss": 0.8690587043762207,
      "step": 19580
    },
    {
      "epoch": 2.070174511539427,
      "grad_norm": 1.3232961893081665,
      "learning_rate": 1.0960763927955362e-05,
      "loss": 0.8590095520019532,
      "step": 19590
    },
    {
      "epoch": 2.0712313565927314,
      "grad_norm": 0.9246929883956909,
      "learning_rate": 1.0937865082362558e-05,
      "loss": 0.813138484954834,
      "step": 19600
    },
    {
      "epoch": 2.072288201646036,
      "grad_norm": 1.1355174779891968,
      "learning_rate": 1.0914983481758146e-05,
      "loss": 0.8583235740661621,
      "step": 19610
    },
    {
      "epoch": 2.0733450466993406,
      "grad_norm": 1.521921992301941,
      "learning_rate": 1.0892119154202823e-05,
      "loss": 0.8740495681762696,
      "step": 19620
    },
    {
      "epoch": 2.0744018917526454,
      "grad_norm": 1.374373197555542,
      "learning_rate": 1.0869272127736063e-05,
      "loss": 0.9954909324645996,
      "step": 19630
    },
    {
      "epoch": 2.07545873680595,
      "grad_norm": 1.3481348752975464,
      "learning_rate": 1.0846442430376158e-05,
      "loss": 0.8521230697631836,
      "step": 19640
    },
    {
      "epoch": 2.0765155818592547,
      "grad_norm": 1.236142873764038,
      "learning_rate": 1.0823630090120095e-05,
      "loss": 0.9641447067260742,
      "step": 19650
    },
    {
      "epoch": 2.0775724269125595,
      "grad_norm": 1.0774903297424316,
      "learning_rate": 1.0800835134943635e-05,
      "loss": 0.8203763008117676,
      "step": 19660
    },
    {
      "epoch": 2.078629271965864,
      "grad_norm": 0.9625611305236816,
      "learning_rate": 1.0778057592801172e-05,
      "loss": 0.8196265220642089,
      "step": 19670
    },
    {
      "epoch": 2.0796861170191687,
      "grad_norm": 1.648374319076538,
      "learning_rate": 1.0755297491625794e-05,
      "loss": 0.8760377883911132,
      "step": 19680
    },
    {
      "epoch": 2.080742962072473,
      "grad_norm": 0.9929314851760864,
      "learning_rate": 1.0732554859329133e-05,
      "loss": 0.8794764518737793,
      "step": 19690
    },
    {
      "epoch": 2.081799807125778,
      "grad_norm": 1.5838043689727783,
      "learning_rate": 1.0709829723801468e-05,
      "loss": 0.8895299911499024,
      "step": 19700
    },
    {
      "epoch": 2.0828566521790823,
      "grad_norm": 1.2397404909133911,
      "learning_rate": 1.0687122112911569e-05,
      "loss": 0.8958721160888672,
      "step": 19710
    },
    {
      "epoch": 2.083913497232387,
      "grad_norm": 1.2126717567443848,
      "learning_rate": 1.0664432054506765e-05,
      "loss": 0.8675051689147949,
      "step": 19720
    },
    {
      "epoch": 2.0849703422856916,
      "grad_norm": 1.0189484357833862,
      "learning_rate": 1.0641759576412793e-05,
      "loss": 0.8706711769104004,
      "step": 19730
    },
    {
      "epoch": 2.0860271873389964,
      "grad_norm": 0.9279252886772156,
      "learning_rate": 1.0619104706433896e-05,
      "loss": 0.8167675018310547,
      "step": 19740
    },
    {
      "epoch": 2.087084032392301,
      "grad_norm": 1.2094358205795288,
      "learning_rate": 1.0596467472352676e-05,
      "loss": 0.9141324996948242,
      "step": 19750
    },
    {
      "epoch": 2.0881408774456056,
      "grad_norm": 1.06949782371521,
      "learning_rate": 1.0573847901930153e-05,
      "loss": 0.8340410232543946,
      "step": 19760
    },
    {
      "epoch": 2.08919772249891,
      "grad_norm": 1.16788649559021,
      "learning_rate": 1.0551246022905625e-05,
      "loss": 0.7494045734405518,
      "step": 19770
    },
    {
      "epoch": 2.090254567552215,
      "grad_norm": 1.427208423614502,
      "learning_rate": 1.0528661862996755e-05,
      "loss": 0.7682887077331543,
      "step": 19780
    },
    {
      "epoch": 2.0913114126055192,
      "grad_norm": 1.1094392538070679,
      "learning_rate": 1.0506095449899433e-05,
      "loss": 0.8626992225646972,
      "step": 19790
    },
    {
      "epoch": 2.092368257658824,
      "grad_norm": 1.1520034074783325,
      "learning_rate": 1.0483546811287828e-05,
      "loss": 0.7974586486816406,
      "step": 19800
    },
    {
      "epoch": 2.0934251027121284,
      "grad_norm": 1.177968978881836,
      "learning_rate": 1.0461015974814245e-05,
      "loss": 0.886676025390625,
      "step": 19810
    },
    {
      "epoch": 2.0944819477654333,
      "grad_norm": 1.5676970481872559,
      "learning_rate": 1.0438502968109232e-05,
      "loss": 0.883698558807373,
      "step": 19820
    },
    {
      "epoch": 2.0955387928187377,
      "grad_norm": 1.0705815553665161,
      "learning_rate": 1.0416007818781414e-05,
      "loss": 0.8416871070861817,
      "step": 19830
    },
    {
      "epoch": 2.0965956378720425,
      "grad_norm": 1.028084635734558,
      "learning_rate": 1.0393530554417572e-05,
      "loss": 0.8227169036865234,
      "step": 19840
    },
    {
      "epoch": 2.097652482925347,
      "grad_norm": 1.2211451530456543,
      "learning_rate": 1.037107120258248e-05,
      "loss": 0.859865665435791,
      "step": 19850
    },
    {
      "epoch": 2.0987093279786517,
      "grad_norm": 1.5461682081222534,
      "learning_rate": 1.034862979081902e-05,
      "loss": 0.8222611427307129,
      "step": 19860
    },
    {
      "epoch": 2.0997661730319566,
      "grad_norm": 1.0266878604888916,
      "learning_rate": 1.0326206346648021e-05,
      "loss": 0.8730506896972656,
      "step": 19870
    },
    {
      "epoch": 2.100823018085261,
      "grad_norm": 1.1718858480453491,
      "learning_rate": 1.0303800897568322e-05,
      "loss": 0.9231551170349122,
      "step": 19880
    },
    {
      "epoch": 2.101879863138566,
      "grad_norm": 1.2648252248764038,
      "learning_rate": 1.0281413471056633e-05,
      "loss": 0.9020763397216797,
      "step": 19890
    },
    {
      "epoch": 2.10293670819187,
      "grad_norm": 1.0451613664627075,
      "learning_rate": 1.025904409456763e-05,
      "loss": 0.7737362861633301,
      "step": 19900
    },
    {
      "epoch": 2.103993553245175,
      "grad_norm": 1.4927723407745361,
      "learning_rate": 1.0236692795533798e-05,
      "loss": 0.836970329284668,
      "step": 19910
    },
    {
      "epoch": 2.1050503982984794,
      "grad_norm": 1.2865327596664429,
      "learning_rate": 1.0214359601365506e-05,
      "loss": 0.9219345092773438,
      "step": 19920
    },
    {
      "epoch": 2.1061072433517842,
      "grad_norm": 1.3077359199523926,
      "learning_rate": 1.0192044539450856e-05,
      "loss": 0.8815442085266113,
      "step": 19930
    },
    {
      "epoch": 2.1071640884050886,
      "grad_norm": 0.9819065928459167,
      "learning_rate": 1.0169747637155775e-05,
      "loss": 0.894477367401123,
      "step": 19940
    },
    {
      "epoch": 2.1082209334583935,
      "grad_norm": 1.4828518629074097,
      "learning_rate": 1.0147468921823871e-05,
      "loss": 1.0255541801452637,
      "step": 19950
    },
    {
      "epoch": 2.109277778511698,
      "grad_norm": 1.567456841468811,
      "learning_rate": 1.0125208420776504e-05,
      "loss": 0.795464038848877,
      "step": 19960
    },
    {
      "epoch": 2.1103346235650027,
      "grad_norm": 0.9253625869750977,
      "learning_rate": 1.0102966161312627e-05,
      "loss": 0.8447951316833496,
      "step": 19970
    },
    {
      "epoch": 2.111391468618307,
      "grad_norm": 1.5132482051849365,
      "learning_rate": 1.0080742170708884e-05,
      "loss": 0.8636096954345703,
      "step": 19980
    },
    {
      "epoch": 2.112448313671612,
      "grad_norm": 1.6259821653366089,
      "learning_rate": 1.0058536476219474e-05,
      "loss": 0.8283861160278321,
      "step": 19990
    },
    {
      "epoch": 2.1135051587249163,
      "grad_norm": 1.4928781986236572,
      "learning_rate": 1.0036349105076204e-05,
      "loss": 0.8943181037902832,
      "step": 20000
    },
    {
      "epoch": 2.1135051587249163,
      "eval_loss": 0.48731473088264465,
      "eval_runtime": 3276.9029,
      "eval_samples_per_second": 2.718,
      "eval_steps_per_second": 2.718,
      "step": 20000
    },
    {
      "epoch": 2.114562003778221,
      "grad_norm": 0.9563535451889038,
      "learning_rate": 1.0014180084488345e-05,
      "loss": 0.7777707576751709,
      "step": 20010
    },
    {
      "epoch": 2.1156188488315255,
      "grad_norm": 1.2591712474822998,
      "learning_rate": 9.992029441642728e-06,
      "loss": 0.8797025680541992,
      "step": 20020
    },
    {
      "epoch": 2.1166756938848303,
      "grad_norm": 1.1730784177780151,
      "learning_rate": 9.969897203703602e-06,
      "loss": 0.9072016716003418,
      "step": 20030
    },
    {
      "epoch": 2.1177325389381347,
      "grad_norm": 1.8978923559188843,
      "learning_rate": 9.947783397812685e-06,
      "loss": 0.8863043785095215,
      "step": 20040
    },
    {
      "epoch": 2.1187893839914396,
      "grad_norm": 1.2955552339553833,
      "learning_rate": 9.92568805108904e-06,
      "loss": 0.9036584854125976,
      "step": 20050
    },
    {
      "epoch": 2.119846229044744,
      "grad_norm": 1.629235863685608,
      "learning_rate": 9.903611190629145e-06,
      "loss": 0.7058406352996827,
      "step": 20060
    },
    {
      "epoch": 2.120903074098049,
      "grad_norm": 1.4561424255371094,
      "learning_rate": 9.881552843506772e-06,
      "loss": 0.8088406562805176,
      "step": 20070
    },
    {
      "epoch": 2.1219599191513536,
      "grad_norm": 1.0040875673294067,
      "learning_rate": 9.859513036773007e-06,
      "loss": 0.8048751831054688,
      "step": 20080
    },
    {
      "epoch": 2.123016764204658,
      "grad_norm": 1.5896157026290894,
      "learning_rate": 9.837491797456189e-06,
      "loss": 0.7703395366668702,
      "step": 20090
    },
    {
      "epoch": 2.124073609257963,
      "grad_norm": 1.8215241432189941,
      "learning_rate": 9.815489152561885e-06,
      "loss": 0.9717734336853028,
      "step": 20100
    },
    {
      "epoch": 2.1251304543112672,
      "grad_norm": 0.9666393995285034,
      "learning_rate": 9.793505129072884e-06,
      "loss": 0.8387363433837891,
      "step": 20110
    },
    {
      "epoch": 2.126187299364572,
      "grad_norm": 1.0454455614089966,
      "learning_rate": 9.771539753949105e-06,
      "loss": 0.8208915710449218,
      "step": 20120
    },
    {
      "epoch": 2.1272441444178765,
      "grad_norm": 1.314156413078308,
      "learning_rate": 9.749593054127621e-06,
      "loss": 0.7010626792907715,
      "step": 20130
    },
    {
      "epoch": 2.1283009894711813,
      "grad_norm": 1.2660032510757446,
      "learning_rate": 9.727665056522577e-06,
      "loss": 0.8174416542053222,
      "step": 20140
    },
    {
      "epoch": 2.1293578345244857,
      "grad_norm": 1.3227639198303223,
      "learning_rate": 9.705755788025228e-06,
      "loss": 0.8532349586486816,
      "step": 20150
    },
    {
      "epoch": 2.1304146795777905,
      "grad_norm": 0.8495364189147949,
      "learning_rate": 9.683865275503817e-06,
      "loss": 0.7597885131835938,
      "step": 20160
    },
    {
      "epoch": 2.131471524631095,
      "grad_norm": 1.1927849054336548,
      "learning_rate": 9.661993545803602e-06,
      "loss": 0.9093083381652832,
      "step": 20170
    },
    {
      "epoch": 2.1325283696843997,
      "grad_norm": 1.396560549736023,
      "learning_rate": 9.6401406257468e-06,
      "loss": 0.9485726356506348,
      "step": 20180
    },
    {
      "epoch": 2.133585214737704,
      "grad_norm": 1.5068135261535645,
      "learning_rate": 9.618306542132583e-06,
      "loss": 0.7994839668273925,
      "step": 20190
    },
    {
      "epoch": 2.134642059791009,
      "grad_norm": 1.2415924072265625,
      "learning_rate": 9.596491321737e-06,
      "loss": 0.901005744934082,
      "step": 20200
    },
    {
      "epoch": 2.1356989048443134,
      "grad_norm": 1.360911250114441,
      "learning_rate": 9.574694991312977e-06,
      "loss": 0.9780127525329589,
      "step": 20210
    },
    {
      "epoch": 2.136755749897618,
      "grad_norm": 1.5826635360717773,
      "learning_rate": 9.552917577590257e-06,
      "loss": 0.8111052513122559,
      "step": 20220
    },
    {
      "epoch": 2.1378125949509226,
      "grad_norm": 1.7501533031463623,
      "learning_rate": 9.531159107275422e-06,
      "loss": 0.7468459129333496,
      "step": 20230
    },
    {
      "epoch": 2.1388694400042274,
      "grad_norm": 1.2226203680038452,
      "learning_rate": 9.509419607051792e-06,
      "loss": 0.8548343658447266,
      "step": 20240
    },
    {
      "epoch": 2.139926285057532,
      "grad_norm": 1.2593287229537964,
      "learning_rate": 9.48769910357943e-06,
      "loss": 0.8602652549743652,
      "step": 20250
    },
    {
      "epoch": 2.1409831301108366,
      "grad_norm": 1.5624585151672363,
      "learning_rate": 9.465997623495098e-06,
      "loss": 0.9549854278564454,
      "step": 20260
    },
    {
      "epoch": 2.142039975164141,
      "grad_norm": 1.1964505910873413,
      "learning_rate": 9.444315193412252e-06,
      "loss": 0.9200518608093262,
      "step": 20270
    },
    {
      "epoch": 2.143096820217446,
      "grad_norm": 1.00190007686615,
      "learning_rate": 9.42265183992096e-06,
      "loss": 0.9042006492614746,
      "step": 20280
    },
    {
      "epoch": 2.1441536652707507,
      "grad_norm": 1.6136075258255005,
      "learning_rate": 9.4010075895879e-06,
      "loss": 0.741997766494751,
      "step": 20290
    },
    {
      "epoch": 2.145210510324055,
      "grad_norm": 1.4245508909225464,
      "learning_rate": 9.37938246895633e-06,
      "loss": 0.8433687210083007,
      "step": 20300
    },
    {
      "epoch": 2.14626735537736,
      "grad_norm": 1.1085989475250244,
      "learning_rate": 9.357776504546054e-06,
      "loss": 0.8139483451843261,
      "step": 20310
    },
    {
      "epoch": 2.1473242004306643,
      "grad_norm": 1.2043304443359375,
      "learning_rate": 9.336189722853376e-06,
      "loss": 0.9100760459899903,
      "step": 20320
    },
    {
      "epoch": 2.148381045483969,
      "grad_norm": 1.3986866474151611,
      "learning_rate": 9.314622150351066e-06,
      "loss": 0.8731118202209472,
      "step": 20330
    },
    {
      "epoch": 2.1494378905372735,
      "grad_norm": 1.2441811561584473,
      "learning_rate": 9.293073813488346e-06,
      "loss": 0.7809085845947266,
      "step": 20340
    },
    {
      "epoch": 2.1504947355905784,
      "grad_norm": 1.355765461921692,
      "learning_rate": 9.271544738690863e-06,
      "loss": 0.8749888420104981,
      "step": 20350
    },
    {
      "epoch": 2.1515515806438827,
      "grad_norm": 1.072855830192566,
      "learning_rate": 9.25003495236062e-06,
      "loss": 0.8202900886535645,
      "step": 20360
    },
    {
      "epoch": 2.1526084256971876,
      "grad_norm": 1.4892768859863281,
      "learning_rate": 9.228544480875975e-06,
      "loss": 0.8801441192626953,
      "step": 20370
    },
    {
      "epoch": 2.153665270750492,
      "grad_norm": 1.2602416276931763,
      "learning_rate": 9.207073350591589e-06,
      "loss": 0.8223505973815918,
      "step": 20380
    },
    {
      "epoch": 2.154722115803797,
      "grad_norm": 1.2211744785308838,
      "learning_rate": 9.18562158783843e-06,
      "loss": 0.9198463439941407,
      "step": 20390
    },
    {
      "epoch": 2.155778960857101,
      "grad_norm": 1.5020973682403564,
      "learning_rate": 9.164189218923694e-06,
      "loss": 0.8651206970214844,
      "step": 20400
    },
    {
      "epoch": 2.156835805910406,
      "grad_norm": 1.3377459049224854,
      "learning_rate": 9.142776270130793e-06,
      "loss": 0.8239879608154297,
      "step": 20410
    },
    {
      "epoch": 2.1578926509637104,
      "grad_norm": 1.5689959526062012,
      "learning_rate": 9.12138276771932e-06,
      "loss": 0.8948896408081055,
      "step": 20420
    },
    {
      "epoch": 2.1589494960170152,
      "grad_norm": 1.1158559322357178,
      "learning_rate": 9.100008737925047e-06,
      "loss": 0.7634482383728027,
      "step": 20430
    },
    {
      "epoch": 2.1600063410703196,
      "grad_norm": 1.0613447427749634,
      "learning_rate": 9.078654206959838e-06,
      "loss": 0.8469998359680175,
      "step": 20440
    },
    {
      "epoch": 2.1610631861236245,
      "grad_norm": 1.4286909103393555,
      "learning_rate": 9.057319201011655e-06,
      "loss": 0.7755132675170898,
      "step": 20450
    },
    {
      "epoch": 2.1621200311769293,
      "grad_norm": 1.38727867603302,
      "learning_rate": 9.036003746244503e-06,
      "loss": 0.8403165817260743,
      "step": 20460
    },
    {
      "epoch": 2.1631768762302337,
      "grad_norm": 1.9889625310897827,
      "learning_rate": 9.01470786879844e-06,
      "loss": 0.893074893951416,
      "step": 20470
    },
    {
      "epoch": 2.1642337212835385,
      "grad_norm": 1.331406593322754,
      "learning_rate": 8.993431594789494e-06,
      "loss": 0.8329508781433106,
      "step": 20480
    },
    {
      "epoch": 2.165290566336843,
      "grad_norm": 1.3607966899871826,
      "learning_rate": 8.972174950309651e-06,
      "loss": 0.8310437202453613,
      "step": 20490
    },
    {
      "epoch": 2.1663474113901477,
      "grad_norm": 1.157660722732544,
      "learning_rate": 8.950937961426819e-06,
      "loss": 0.9204286575317383,
      "step": 20500
    },
    {
      "epoch": 2.1663474113901477,
      "eval_loss": 0.4862632155418396,
      "eval_runtime": 3305.9378,
      "eval_samples_per_second": 2.694,
      "eval_steps_per_second": 2.694,
      "step": 20500
    },
    {
      "epoch": 2.167404256443452,
      "grad_norm": 1.111608862876892,
      "learning_rate": 8.929720654184836e-06,
      "loss": 0.8862844467163086,
      "step": 20510
    },
    {
      "epoch": 2.168461101496757,
      "grad_norm": 1.357601523399353,
      "learning_rate": 8.908523054603365e-06,
      "loss": 0.7965086460113525,
      "step": 20520
    },
    {
      "epoch": 2.1695179465500614,
      "grad_norm": 1.1828148365020752,
      "learning_rate": 8.88734518867792e-06,
      "loss": 0.900972843170166,
      "step": 20530
    },
    {
      "epoch": 2.170574791603366,
      "grad_norm": 1.159686803817749,
      "learning_rate": 8.866187082379798e-06,
      "loss": 0.868040943145752,
      "step": 20540
    },
    {
      "epoch": 2.1716316366566706,
      "grad_norm": 1.1369285583496094,
      "learning_rate": 8.845048761656094e-06,
      "loss": 0.8159896850585937,
      "step": 20550
    },
    {
      "epoch": 2.1726884817099754,
      "grad_norm": 1.2116305828094482,
      "learning_rate": 8.823930252429615e-06,
      "loss": 0.911412239074707,
      "step": 20560
    },
    {
      "epoch": 2.17374532676328,
      "grad_norm": 1.6507627964019775,
      "learning_rate": 8.802831580598876e-06,
      "loss": 0.7966672420501709,
      "step": 20570
    },
    {
      "epoch": 2.1748021718165846,
      "grad_norm": 1.328449010848999,
      "learning_rate": 8.781752772038058e-06,
      "loss": 0.7506781101226807,
      "step": 20580
    },
    {
      "epoch": 2.175859016869889,
      "grad_norm": 1.1520620584487915,
      "learning_rate": 8.760693852597007e-06,
      "loss": 0.8050737380981445,
      "step": 20590
    },
    {
      "epoch": 2.176915861923194,
      "grad_norm": 1.2641127109527588,
      "learning_rate": 8.739654848101162e-06,
      "loss": 0.8606757164001465,
      "step": 20600
    },
    {
      "epoch": 2.1779727069764983,
      "grad_norm": 1.0801430940628052,
      "learning_rate": 8.718635784351534e-06,
      "loss": 0.8459340095520019,
      "step": 20610
    },
    {
      "epoch": 2.179029552029803,
      "grad_norm": 0.9044227004051208,
      "learning_rate": 8.697636687124688e-06,
      "loss": 0.8079410552978515,
      "step": 20620
    },
    {
      "epoch": 2.1800863970831075,
      "grad_norm": 1.8413033485412598,
      "learning_rate": 8.676657582172692e-06,
      "loss": 0.8475300788879394,
      "step": 20630
    },
    {
      "epoch": 2.1811432421364123,
      "grad_norm": 1.1152527332305908,
      "learning_rate": 8.655698495223127e-06,
      "loss": 0.8519956588745117,
      "step": 20640
    },
    {
      "epoch": 2.1822000871897167,
      "grad_norm": 1.6309038400650024,
      "learning_rate": 8.634759451978992e-06,
      "loss": 0.8535550117492676,
      "step": 20650
    },
    {
      "epoch": 2.1832569322430215,
      "grad_norm": 1.4777299165725708,
      "learning_rate": 8.613840478118718e-06,
      "loss": 0.8756319046020508,
      "step": 20660
    },
    {
      "epoch": 2.1843137772963264,
      "grad_norm": 1.373591661453247,
      "learning_rate": 8.592941599296114e-06,
      "loss": 0.8630623817443848,
      "step": 20670
    },
    {
      "epoch": 2.1853706223496308,
      "grad_norm": 1.3738982677459717,
      "learning_rate": 8.572062841140378e-06,
      "loss": 0.9193549156188965,
      "step": 20680
    },
    {
      "epoch": 2.1864274674029356,
      "grad_norm": 1.0653462409973145,
      "learning_rate": 8.551204229255999e-06,
      "loss": 0.9219851493835449,
      "step": 20690
    },
    {
      "epoch": 2.18748431245624,
      "grad_norm": 1.198649287223816,
      "learning_rate": 8.530365789222769e-06,
      "loss": 0.9033349990844727,
      "step": 20700
    },
    {
      "epoch": 2.188541157509545,
      "grad_norm": 1.9451066255569458,
      "learning_rate": 8.509547546595739e-06,
      "loss": 0.7488329410552979,
      "step": 20710
    },
    {
      "epoch": 2.189598002562849,
      "grad_norm": 1.5392249822616577,
      "learning_rate": 8.488749526905213e-06,
      "loss": 0.9561814308166504,
      "step": 20720
    },
    {
      "epoch": 2.190654847616154,
      "grad_norm": 1.4859110116958618,
      "learning_rate": 8.467971755656673e-06,
      "loss": 0.8451541900634766,
      "step": 20730
    },
    {
      "epoch": 2.1917116926694584,
      "grad_norm": 1.4351060390472412,
      "learning_rate": 8.447214258330769e-06,
      "loss": 0.9541596412658692,
      "step": 20740
    },
    {
      "epoch": 2.1927685377227633,
      "grad_norm": 1.6806269884109497,
      "learning_rate": 8.426477060383292e-06,
      "loss": 0.8074274063110352,
      "step": 20750
    },
    {
      "epoch": 2.1938253827760676,
      "grad_norm": 0.9549673199653625,
      "learning_rate": 8.405760187245155e-06,
      "loss": 0.9108348846435547,
      "step": 20760
    },
    {
      "epoch": 2.1948822278293725,
      "grad_norm": 1.323884129524231,
      "learning_rate": 8.385063664322324e-06,
      "loss": 0.8780827522277832,
      "step": 20770
    },
    {
      "epoch": 2.195939072882677,
      "grad_norm": 1.4830173254013062,
      "learning_rate": 8.364387516995818e-06,
      "loss": 0.8244524955749511,
      "step": 20780
    },
    {
      "epoch": 2.1969959179359817,
      "grad_norm": 0.9951741099357605,
      "learning_rate": 8.34373177062166e-06,
      "loss": 0.8389791488647461,
      "step": 20790
    },
    {
      "epoch": 2.198052762989286,
      "grad_norm": 1.1441352367401123,
      "learning_rate": 8.323096450530874e-06,
      "loss": 0.8416620254516601,
      "step": 20800
    },
    {
      "epoch": 2.199109608042591,
      "grad_norm": 1.3743914365768433,
      "learning_rate": 8.30248158202942e-06,
      "loss": 0.9651534080505371,
      "step": 20810
    },
    {
      "epoch": 2.2001664530958953,
      "grad_norm": 1.0174548625946045,
      "learning_rate": 8.281887190398174e-06,
      "loss": 0.7778762340545654,
      "step": 20820
    },
    {
      "epoch": 2.2012232981492,
      "grad_norm": 1.282898187637329,
      "learning_rate": 8.261313300892904e-06,
      "loss": 0.878294849395752,
      "step": 20830
    },
    {
      "epoch": 2.2022801432025045,
      "grad_norm": 1.7894026041030884,
      "learning_rate": 8.240759938744247e-06,
      "loss": 0.9645803451538086,
      "step": 20840
    },
    {
      "epoch": 2.2033369882558094,
      "grad_norm": 1.385988712310791,
      "learning_rate": 8.220227129157654e-06,
      "loss": 0.8176486015319824,
      "step": 20850
    },
    {
      "epoch": 2.2043938333091138,
      "grad_norm": 1.504180669784546,
      "learning_rate": 8.19971489731337e-06,
      "loss": 0.8722283363342285,
      "step": 20860
    },
    {
      "epoch": 2.2054506783624186,
      "grad_norm": 1.336534023284912,
      "learning_rate": 8.179223268366407e-06,
      "loss": 0.8746189117431641,
      "step": 20870
    },
    {
      "epoch": 2.2065075234157234,
      "grad_norm": 1.2868709564208984,
      "learning_rate": 8.15875226744652e-06,
      "loss": 0.9707441329956055,
      "step": 20880
    },
    {
      "epoch": 2.207564368469028,
      "grad_norm": 1.5498573780059814,
      "learning_rate": 8.138301919658161e-06,
      "loss": 0.8586291313171387,
      "step": 20890
    },
    {
      "epoch": 2.2086212135223326,
      "grad_norm": 1.3689712285995483,
      "learning_rate": 8.117872250080446e-06,
      "loss": 0.9241810798645019,
      "step": 20900
    },
    {
      "epoch": 2.209678058575637,
      "grad_norm": 1.7055948972702026,
      "learning_rate": 8.097463283767137e-06,
      "loss": 0.8744813919067382,
      "step": 20910
    },
    {
      "epoch": 2.210734903628942,
      "grad_norm": 1.5548864603042603,
      "learning_rate": 8.077075045746619e-06,
      "loss": 0.9733401298522949,
      "step": 20920
    },
    {
      "epoch": 2.2117917486822463,
      "grad_norm": 1.1938731670379639,
      "learning_rate": 8.056707561021845e-06,
      "loss": 0.9661501884460449,
      "step": 20930
    },
    {
      "epoch": 2.212848593735551,
      "grad_norm": 1.1824127435684204,
      "learning_rate": 8.036360854570316e-06,
      "loss": 0.8302247047424316,
      "step": 20940
    },
    {
      "epoch": 2.2139054387888555,
      "grad_norm": 0.9413944482803345,
      "learning_rate": 8.016034951344046e-06,
      "loss": 0.8593485832214356,
      "step": 20950
    },
    {
      "epoch": 2.2149622838421603,
      "grad_norm": 1.0882384777069092,
      "learning_rate": 7.995729876269565e-06,
      "loss": 0.9179246902465821,
      "step": 20960
    },
    {
      "epoch": 2.2160191288954647,
      "grad_norm": 1.1921666860580444,
      "learning_rate": 7.975445654247835e-06,
      "loss": 0.7205573081970215,
      "step": 20970
    },
    {
      "epoch": 2.2170759739487695,
      "grad_norm": 1.1850742101669312,
      "learning_rate": 7.95518231015425e-06,
      "loss": 0.8172308921813964,
      "step": 20980
    },
    {
      "epoch": 2.218132819002074,
      "grad_norm": 1.2005894184112549,
      "learning_rate": 7.934939868838592e-06,
      "loss": 0.829469871520996,
      "step": 20990
    },
    {
      "epoch": 2.2191896640553788,
      "grad_norm": 1.24151611328125,
      "learning_rate": 7.914718355125043e-06,
      "loss": 0.8342708587646485,
      "step": 21000
    },
    {
      "epoch": 2.2191896640553788,
      "eval_loss": 0.48631376028060913,
      "eval_runtime": 3295.1272,
      "eval_samples_per_second": 2.702,
      "eval_steps_per_second": 2.702,
      "step": 21000
    },
    {
      "epoch": 2.220246509108683,
      "grad_norm": 1.6857826709747314,
      "learning_rate": 7.894517793812078e-06,
      "loss": 0.8496136665344238,
      "step": 21010
    },
    {
      "epoch": 2.221303354161988,
      "grad_norm": 1.4029793739318848,
      "learning_rate": 7.874338209672507e-06,
      "loss": 0.8891298294067382,
      "step": 21020
    },
    {
      "epoch": 2.2223601992152924,
      "grad_norm": 1.215582251548767,
      "learning_rate": 7.854179627453389e-06,
      "loss": 0.8832895278930664,
      "step": 21030
    },
    {
      "epoch": 2.223417044268597,
      "grad_norm": 1.3796038627624512,
      "learning_rate": 7.834042071876063e-06,
      "loss": 0.8879108428955078,
      "step": 21040
    },
    {
      "epoch": 2.2244738893219016,
      "grad_norm": 1.357569694519043,
      "learning_rate": 7.813925567636051e-06,
      "loss": 0.8945812225341797,
      "step": 21050
    },
    {
      "epoch": 2.2255307343752064,
      "grad_norm": 1.4801578521728516,
      "learning_rate": 7.793830139403072e-06,
      "loss": 0.9161118507385254,
      "step": 21060
    },
    {
      "epoch": 2.226587579428511,
      "grad_norm": 1.2665520906448364,
      "learning_rate": 7.773755811820985e-06,
      "loss": 0.8399243354797363,
      "step": 21070
    },
    {
      "epoch": 2.2276444244818157,
      "grad_norm": 1.1792436838150024,
      "learning_rate": 7.753702609507801e-06,
      "loss": 0.8259225845336914,
      "step": 21080
    },
    {
      "epoch": 2.2287012695351205,
      "grad_norm": 1.274572730064392,
      "learning_rate": 7.7336705570556e-06,
      "loss": 0.8148788452148438,
      "step": 21090
    },
    {
      "epoch": 2.229758114588425,
      "grad_norm": 1.6488457918167114,
      "learning_rate": 7.713659679030527e-06,
      "loss": 0.8844100952148437,
      "step": 21100
    },
    {
      "epoch": 2.2308149596417297,
      "grad_norm": 1.4412405490875244,
      "learning_rate": 7.693669999972763e-06,
      "loss": 0.8640574455261231,
      "step": 21110
    },
    {
      "epoch": 2.231871804695034,
      "grad_norm": 1.5173754692077637,
      "learning_rate": 7.673701544396508e-06,
      "loss": 0.8501100540161133,
      "step": 21120
    },
    {
      "epoch": 2.232928649748339,
      "grad_norm": 1.2773516178131104,
      "learning_rate": 7.653754336789909e-06,
      "loss": 0.900639533996582,
      "step": 21130
    },
    {
      "epoch": 2.2339854948016433,
      "grad_norm": 1.5317981243133545,
      "learning_rate": 7.633828401615071e-06,
      "loss": 0.9276094436645508,
      "step": 21140
    },
    {
      "epoch": 2.235042339854948,
      "grad_norm": 1.3817973136901855,
      "learning_rate": 7.6139237633079956e-06,
      "loss": 0.800080394744873,
      "step": 21150
    },
    {
      "epoch": 2.2360991849082525,
      "grad_norm": 1.4337248802185059,
      "learning_rate": 7.594040446278597e-06,
      "loss": 0.961793041229248,
      "step": 21160
    },
    {
      "epoch": 2.2371560299615574,
      "grad_norm": 1.4675263166427612,
      "learning_rate": 7.5741784749106145e-06,
      "loss": 0.8298232078552246,
      "step": 21170
    },
    {
      "epoch": 2.2382128750148618,
      "grad_norm": 1.1299012899398804,
      "learning_rate": 7.554337873561626e-06,
      "loss": 0.9074227333068847,
      "step": 21180
    },
    {
      "epoch": 2.2392697200681666,
      "grad_norm": 1.122652292251587,
      "learning_rate": 7.5345186665629925e-06,
      "loss": 0.9688516616821289,
      "step": 21190
    },
    {
      "epoch": 2.240326565121471,
      "grad_norm": 1.3946301937103271,
      "learning_rate": 7.514720878219833e-06,
      "loss": 0.8214277267456055,
      "step": 21200
    },
    {
      "epoch": 2.241383410174776,
      "grad_norm": 2.1314175128936768,
      "learning_rate": 7.494944532811038e-06,
      "loss": 0.9431739807128906,
      "step": 21210
    },
    {
      "epoch": 2.24244025522808,
      "grad_norm": 1.3471664190292358,
      "learning_rate": 7.475189654589138e-06,
      "loss": 0.7802850246429444,
      "step": 21220
    },
    {
      "epoch": 2.243497100281385,
      "grad_norm": 1.7846940755844116,
      "learning_rate": 7.455456267780395e-06,
      "loss": 0.8822522163391113,
      "step": 21230
    },
    {
      "epoch": 2.2445539453346894,
      "grad_norm": 1.3426530361175537,
      "learning_rate": 7.435744396584679e-06,
      "loss": 0.8295673370361328,
      "step": 21240
    },
    {
      "epoch": 2.2456107903879943,
      "grad_norm": 1.3698786497116089,
      "learning_rate": 7.416054065175513e-06,
      "loss": 0.815796184539795,
      "step": 21250
    },
    {
      "epoch": 2.2466676354412987,
      "grad_norm": 1.4198206663131714,
      "learning_rate": 7.396385297699943e-06,
      "loss": 0.9136198043823243,
      "step": 21260
    },
    {
      "epoch": 2.2477244804946035,
      "grad_norm": 0.9479147791862488,
      "learning_rate": 7.376738118278634e-06,
      "loss": 0.9125507354736329,
      "step": 21270
    },
    {
      "epoch": 2.248781325547908,
      "grad_norm": 1.5470222234725952,
      "learning_rate": 7.357112551005735e-06,
      "loss": 0.8383374214172363,
      "step": 21280
    },
    {
      "epoch": 2.2498381706012127,
      "grad_norm": 1.3305402994155884,
      "learning_rate": 7.33750861994893e-06,
      "loss": 0.8419397354125977,
      "step": 21290
    },
    {
      "epoch": 2.2508950156545176,
      "grad_norm": 1.2185657024383545,
      "learning_rate": 7.3179263491493095e-06,
      "loss": 0.8711481094360352,
      "step": 21300
    },
    {
      "epoch": 2.251951860707822,
      "grad_norm": 1.182152271270752,
      "learning_rate": 7.298365762621464e-06,
      "loss": 0.8621840476989746,
      "step": 21310
    },
    {
      "epoch": 2.2530087057611268,
      "grad_norm": 1.4320274591445923,
      "learning_rate": 7.278826884353351e-06,
      "loss": 0.8451634407043457,
      "step": 21320
    },
    {
      "epoch": 2.254065550814431,
      "grad_norm": 1.0900665521621704,
      "learning_rate": 7.2593097383063384e-06,
      "loss": 0.8664793968200684,
      "step": 21330
    },
    {
      "epoch": 2.255122395867736,
      "grad_norm": 0.867480456829071,
      "learning_rate": 7.2398143484150975e-06,
      "loss": 0.8150432586669922,
      "step": 21340
    },
    {
      "epoch": 2.2561792409210404,
      "grad_norm": 1.2458523511886597,
      "learning_rate": 7.220340738587669e-06,
      "loss": 0.9331179618835449,
      "step": 21350
    },
    {
      "epoch": 2.257236085974345,
      "grad_norm": 1.2626800537109375,
      "learning_rate": 7.200888932705344e-06,
      "loss": 0.8389029502868652,
      "step": 21360
    },
    {
      "epoch": 2.2582929310276496,
      "grad_norm": 1.0448240041732788,
      "learning_rate": 7.181458954622714e-06,
      "loss": 0.934419059753418,
      "step": 21370
    },
    {
      "epoch": 2.2593497760809544,
      "grad_norm": 1.4713551998138428,
      "learning_rate": 7.162050828167549e-06,
      "loss": 0.871403980255127,
      "step": 21380
    },
    {
      "epoch": 2.260406621134259,
      "grad_norm": 1.7052724361419678,
      "learning_rate": 7.142664577140876e-06,
      "loss": 0.9119677543640137,
      "step": 21390
    },
    {
      "epoch": 2.2614634661875637,
      "grad_norm": 1.777775764465332,
      "learning_rate": 7.123300225316859e-06,
      "loss": 0.9340780258178711,
      "step": 21400
    },
    {
      "epoch": 2.262520311240868,
      "grad_norm": 1.014190673828125,
      "learning_rate": 7.103957796442834e-06,
      "loss": 0.7185611248016357,
      "step": 21410
    },
    {
      "epoch": 2.263577156294173,
      "grad_norm": 1.2191739082336426,
      "learning_rate": 7.084637314239212e-06,
      "loss": 0.9621832847595215,
      "step": 21420
    },
    {
      "epoch": 2.2646340013474773,
      "grad_norm": 1.7386350631713867,
      "learning_rate": 7.065338802399538e-06,
      "loss": 0.8671950340270996,
      "step": 21430
    },
    {
      "epoch": 2.265690846400782,
      "grad_norm": 0.9992101192474365,
      "learning_rate": 7.0460622845903686e-06,
      "loss": 0.811929988861084,
      "step": 21440
    },
    {
      "epoch": 2.2667476914540865,
      "grad_norm": 1.4000431299209595,
      "learning_rate": 7.0268077844513335e-06,
      "loss": 0.8797445297241211,
      "step": 21450
    },
    {
      "epoch": 2.2678045365073913,
      "grad_norm": 1.0444756746292114,
      "learning_rate": 7.007575325595012e-06,
      "loss": 0.840949535369873,
      "step": 21460
    },
    {
      "epoch": 2.268861381560696,
      "grad_norm": 1.2659364938735962,
      "learning_rate": 6.988364931607e-06,
      "loss": 0.878659439086914,
      "step": 21470
    },
    {
      "epoch": 2.2699182266140006,
      "grad_norm": 1.2796216011047363,
      "learning_rate": 6.969176626045792e-06,
      "loss": 0.8297262191772461,
      "step": 21480
    },
    {
      "epoch": 2.270975071667305,
      "grad_norm": 0.9651225805282593,
      "learning_rate": 6.950010432442844e-06,
      "loss": 0.7489469051361084,
      "step": 21490
    },
    {
      "epoch": 2.27203191672061,
      "grad_norm": 1.1623140573501587,
      "learning_rate": 6.930866374302436e-06,
      "loss": 0.8541905403137207,
      "step": 21500
    },
    {
      "epoch": 2.27203191672061,
      "eval_loss": 0.4870993494987488,
      "eval_runtime": 3511.8057,
      "eval_samples_per_second": 2.536,
      "eval_steps_per_second": 2.536,
      "step": 21500
    },
    {
      "epoch": 2.2730887617739146,
      "grad_norm": 1.2774441242218018,
      "learning_rate": 6.911744475101753e-06,
      "loss": 0.7981479644775391,
      "step": 21510
    },
    {
      "epoch": 2.274145606827219,
      "grad_norm": 1.3004357814788818,
      "learning_rate": 6.89264475829077e-06,
      "loss": 0.8721466064453125,
      "step": 21520
    },
    {
      "epoch": 2.275202451880524,
      "grad_norm": 1.1777870655059814,
      "learning_rate": 6.873567247292301e-06,
      "loss": 0.8742006301879883,
      "step": 21530
    },
    {
      "epoch": 2.2762592969338282,
      "grad_norm": 1.8485766649246216,
      "learning_rate": 6.854511965501867e-06,
      "loss": 0.8359010696411133,
      "step": 21540
    },
    {
      "epoch": 2.277316141987133,
      "grad_norm": 1.1853916645050049,
      "learning_rate": 6.835478936287787e-06,
      "loss": 0.9021844863891602,
      "step": 21550
    },
    {
      "epoch": 2.2783729870404374,
      "grad_norm": 1.023695945739746,
      "learning_rate": 6.816468182991054e-06,
      "loss": 0.8478774070739746,
      "step": 21560
    },
    {
      "epoch": 2.2794298320937423,
      "grad_norm": 1.2798343896865845,
      "learning_rate": 6.797479728925374e-06,
      "loss": 0.8600960731506347,
      "step": 21570
    },
    {
      "epoch": 2.2804866771470467,
      "grad_norm": 1.6018675565719604,
      "learning_rate": 6.778513597377056e-06,
      "loss": 0.8204137802124023,
      "step": 21580
    },
    {
      "epoch": 2.2815435222003515,
      "grad_norm": 1.224265217781067,
      "learning_rate": 6.759569811605096e-06,
      "loss": 0.8136817932128906,
      "step": 21590
    },
    {
      "epoch": 2.282600367253656,
      "grad_norm": 1.1798418760299683,
      "learning_rate": 6.740648394841037e-06,
      "loss": 0.8771353721618652,
      "step": 21600
    },
    {
      "epoch": 2.2836572123069607,
      "grad_norm": 1.2579540014266968,
      "learning_rate": 6.721749370289035e-06,
      "loss": 0.782154655456543,
      "step": 21610
    },
    {
      "epoch": 2.284714057360265,
      "grad_norm": 1.1669632196426392,
      "learning_rate": 6.7028727611257255e-06,
      "loss": 0.7530049800872802,
      "step": 21620
    },
    {
      "epoch": 2.28577090241357,
      "grad_norm": 1.356535792350769,
      "learning_rate": 6.6840185905003125e-06,
      "loss": 0.9189202308654785,
      "step": 21630
    },
    {
      "epoch": 2.2868277474668743,
      "grad_norm": 1.0248284339904785,
      "learning_rate": 6.665186881534447e-06,
      "loss": 0.8952820777893067,
      "step": 21640
    },
    {
      "epoch": 2.287884592520179,
      "grad_norm": 1.2640215158462524,
      "learning_rate": 6.64637765732227e-06,
      "loss": 0.8403078079223633,
      "step": 21650
    },
    {
      "epoch": 2.2889414375734836,
      "grad_norm": 1.3134421110153198,
      "learning_rate": 6.6275909409302925e-06,
      "loss": 0.8424525260925293,
      "step": 21660
    },
    {
      "epoch": 2.2899982826267884,
      "grad_norm": 1.6390712261199951,
      "learning_rate": 6.608826755397482e-06,
      "loss": 0.7606714248657227,
      "step": 21670
    },
    {
      "epoch": 2.2910551276800932,
      "grad_norm": 1.2190744876861572,
      "learning_rate": 6.590085123735132e-06,
      "loss": 0.8777219772338867,
      "step": 21680
    },
    {
      "epoch": 2.2921119727333976,
      "grad_norm": 1.4501478672027588,
      "learning_rate": 6.571366068926918e-06,
      "loss": 0.9744075775146485,
      "step": 21690
    },
    {
      "epoch": 2.293168817786702,
      "grad_norm": 1.2468026876449585,
      "learning_rate": 6.552669613928777e-06,
      "loss": 0.8104689598083497,
      "step": 21700
    },
    {
      "epoch": 2.294225662840007,
      "grad_norm": 1.5447064638137817,
      "learning_rate": 6.533995781668981e-06,
      "loss": 0.8271440505981446,
      "step": 21710
    },
    {
      "epoch": 2.2952825078933117,
      "grad_norm": 1.213416337966919,
      "learning_rate": 6.51534459504802e-06,
      "loss": 0.8573740959167481,
      "step": 21720
    },
    {
      "epoch": 2.296339352946616,
      "grad_norm": 1.7305474281311035,
      "learning_rate": 6.496716076938655e-06,
      "loss": 0.7943529605865478,
      "step": 21730
    },
    {
      "epoch": 2.297396197999921,
      "grad_norm": 1.9436882734298706,
      "learning_rate": 6.478110250185801e-06,
      "loss": 0.7765072822570801,
      "step": 21740
    },
    {
      "epoch": 2.2984530430532253,
      "grad_norm": 1.260341763496399,
      "learning_rate": 6.459527137606564e-06,
      "loss": 0.8352829933166503,
      "step": 21750
    },
    {
      "epoch": 2.29950988810653,
      "grad_norm": 1.3751847743988037,
      "learning_rate": 6.4409667619902136e-06,
      "loss": 0.909604835510254,
      "step": 21760
    },
    {
      "epoch": 2.3005667331598345,
      "grad_norm": 1.5838974714279175,
      "learning_rate": 6.42242914609811e-06,
      "loss": 0.9560894966125488,
      "step": 21770
    },
    {
      "epoch": 2.3016235782131393,
      "grad_norm": 1.2319731712341309,
      "learning_rate": 6.40391431266372e-06,
      "loss": 0.9408965110778809,
      "step": 21780
    },
    {
      "epoch": 2.3026804232664437,
      "grad_norm": 1.23921799659729,
      "learning_rate": 6.3854222843925505e-06,
      "loss": 0.8203904151916503,
      "step": 21790
    },
    {
      "epoch": 2.3037372683197486,
      "grad_norm": 1.2413350343704224,
      "learning_rate": 6.366953083962171e-06,
      "loss": 0.9035897254943848,
      "step": 21800
    },
    {
      "epoch": 2.304794113373053,
      "grad_norm": 1.1476987600326538,
      "learning_rate": 6.348506734022133e-06,
      "loss": 0.8532477378845215,
      "step": 21810
    },
    {
      "epoch": 2.305850958426358,
      "grad_norm": 1.2279222011566162,
      "learning_rate": 6.3300832571939726e-06,
      "loss": 0.70803861618042,
      "step": 21820
    },
    {
      "epoch": 2.306907803479662,
      "grad_norm": 1.2416325807571411,
      "learning_rate": 6.311682676071165e-06,
      "loss": 0.9474679946899414,
      "step": 21830
    },
    {
      "epoch": 2.307964648532967,
      "grad_norm": 1.818001627922058,
      "learning_rate": 6.293305013219139e-06,
      "loss": 0.79036545753479,
      "step": 21840
    },
    {
      "epoch": 2.3090214935862714,
      "grad_norm": 1.5271974802017212,
      "learning_rate": 6.274950291175183e-06,
      "loss": 0.9784195899963379,
      "step": 21850
    },
    {
      "epoch": 2.3100783386395762,
      "grad_norm": 2.2225849628448486,
      "learning_rate": 6.256618532448469e-06,
      "loss": 0.8486021995544434,
      "step": 21860
    },
    {
      "epoch": 2.3111351836928806,
      "grad_norm": 1.3552395105361938,
      "learning_rate": 6.238309759519997e-06,
      "loss": 0.9231218338012696,
      "step": 21870
    },
    {
      "epoch": 2.3121920287461855,
      "grad_norm": 1.9233365058898926,
      "learning_rate": 6.220023994842602e-06,
      "loss": 0.8213667869567871,
      "step": 21880
    },
    {
      "epoch": 2.3132488737994903,
      "grad_norm": 2.1537973880767822,
      "learning_rate": 6.201761260840877e-06,
      "loss": 0.8258940696716308,
      "step": 21890
    },
    {
      "epoch": 2.3143057188527947,
      "grad_norm": 1.969558596611023,
      "learning_rate": 6.183521579911189e-06,
      "loss": 0.859220314025879,
      "step": 21900
    },
    {
      "epoch": 2.315362563906099,
      "grad_norm": 1.3265037536621094,
      "learning_rate": 6.165304974421612e-06,
      "loss": 0.8516258239746094,
      "step": 21910
    },
    {
      "epoch": 2.316419408959404,
      "grad_norm": 1.2713056802749634,
      "learning_rate": 6.147111466711955e-06,
      "loss": 0.8140851020812988,
      "step": 21920
    },
    {
      "epoch": 2.3174762540127087,
      "grad_norm": 1.049216866493225,
      "learning_rate": 6.128941079093675e-06,
      "loss": 0.8771238327026367,
      "step": 21930
    },
    {
      "epoch": 2.318533099066013,
      "grad_norm": 1.178865909576416,
      "learning_rate": 6.1107938338498815e-06,
      "loss": 0.8282979965209961,
      "step": 21940
    },
    {
      "epoch": 2.319589944119318,
      "grad_norm": 1.343706727027893,
      "learning_rate": 6.092669753235306e-06,
      "loss": 0.9009005546569824,
      "step": 21950
    },
    {
      "epoch": 2.3206467891726223,
      "grad_norm": 1.4885045289993286,
      "learning_rate": 6.0745688594762765e-06,
      "loss": 0.8584798812866211,
      "step": 21960
    },
    {
      "epoch": 2.321703634225927,
      "grad_norm": 1.502579689025879,
      "learning_rate": 6.056491174770681e-06,
      "loss": 0.8620627403259278,
      "step": 21970
    },
    {
      "epoch": 2.3227604792792316,
      "grad_norm": 1.4320780038833618,
      "learning_rate": 6.038436721287943e-06,
      "loss": 0.682434368133545,
      "step": 21980
    },
    {
      "epoch": 2.3238173243325364,
      "grad_norm": 1.0771578550338745,
      "learning_rate": 6.020405521168995e-06,
      "loss": 0.8723282814025879,
      "step": 21990
    },
    {
      "epoch": 2.324874169385841,
      "grad_norm": 1.1333130598068237,
      "learning_rate": 6.002397596526269e-06,
      "loss": 0.9198422431945801,
      "step": 22000
    },
    {
      "epoch": 2.324874169385841,
      "eval_loss": 0.48629045486450195,
      "eval_runtime": 3509.5817,
      "eval_samples_per_second": 2.537,
      "eval_steps_per_second": 2.537,
      "step": 22000
    },
    {
      "epoch": 2.3259310144391456,
      "grad_norm": 1.3871567249298096,
      "learning_rate": 5.984412969443634e-06,
      "loss": 0.9254296302795411,
      "step": 22010
    },
    {
      "epoch": 2.32698785949245,
      "grad_norm": 1.268003225326538,
      "learning_rate": 5.9664516619764e-06,
      "loss": 0.8642647743225098,
      "step": 22020
    },
    {
      "epoch": 2.328044704545755,
      "grad_norm": 1.1451398134231567,
      "learning_rate": 5.948513696151262e-06,
      "loss": 0.8746049880981446,
      "step": 22030
    },
    {
      "epoch": 2.3291015495990592,
      "grad_norm": 1.4766108989715576,
      "learning_rate": 5.9305990939663185e-06,
      "loss": 0.8248427391052247,
      "step": 22040
    },
    {
      "epoch": 2.330158394652364,
      "grad_norm": 2.6106605529785156,
      "learning_rate": 5.912707877390997e-06,
      "loss": 0.7980892181396484,
      "step": 22050
    },
    {
      "epoch": 2.3312152397056685,
      "grad_norm": 1.9352383613586426,
      "learning_rate": 5.894840068366048e-06,
      "loss": 0.8715050697326661,
      "step": 22060
    },
    {
      "epoch": 2.3322720847589733,
      "grad_norm": 1.1761614084243774,
      "learning_rate": 5.8769956888035135e-06,
      "loss": 0.9654042243957519,
      "step": 22070
    },
    {
      "epoch": 2.3333289298122777,
      "grad_norm": 1.5245227813720703,
      "learning_rate": 5.8591747605867165e-06,
      "loss": 0.8835679054260254,
      "step": 22080
    },
    {
      "epoch": 2.3343857748655825,
      "grad_norm": 1.6027209758758545,
      "learning_rate": 5.841377305570211e-06,
      "loss": 0.9289098739624023,
      "step": 22090
    },
    {
      "epoch": 2.3354426199188874,
      "grad_norm": 1.0510950088500977,
      "learning_rate": 5.823603345579765e-06,
      "loss": 0.8589025497436523,
      "step": 22100
    },
    {
      "epoch": 2.3364994649721917,
      "grad_norm": 1.0408364534378052,
      "learning_rate": 5.805852902412321e-06,
      "loss": 0.8148789405822754,
      "step": 22110
    },
    {
      "epoch": 2.3375563100254966,
      "grad_norm": 1.3413861989974976,
      "learning_rate": 5.788125997836016e-06,
      "loss": 0.8739341735839844,
      "step": 22120
    },
    {
      "epoch": 2.338613155078801,
      "grad_norm": 1.1350109577178955,
      "learning_rate": 5.77042265359009e-06,
      "loss": 0.8852316856384277,
      "step": 22130
    },
    {
      "epoch": 2.339670000132106,
      "grad_norm": 0.9434782266616821,
      "learning_rate": 5.752742891384902e-06,
      "loss": 0.7698347568511963,
      "step": 22140
    },
    {
      "epoch": 2.34072684518541,
      "grad_norm": 1.7187564373016357,
      "learning_rate": 5.735086732901882e-06,
      "loss": 0.8575002670288085,
      "step": 22150
    },
    {
      "epoch": 2.341783690238715,
      "grad_norm": 1.3506795167922974,
      "learning_rate": 5.717454199793529e-06,
      "loss": 0.8601404190063476,
      "step": 22160
    },
    {
      "epoch": 2.3428405352920194,
      "grad_norm": 1.5778638124465942,
      "learning_rate": 5.69984531368336e-06,
      "loss": 1.0068857192993164,
      "step": 22170
    },
    {
      "epoch": 2.3438973803453242,
      "grad_norm": 1.472481369972229,
      "learning_rate": 5.682260096165895e-06,
      "loss": 0.8584482192993164,
      "step": 22180
    },
    {
      "epoch": 2.3449542253986286,
      "grad_norm": 1.3226392269134521,
      "learning_rate": 5.664698568806614e-06,
      "loss": 0.8959616661071778,
      "step": 22190
    },
    {
      "epoch": 2.3460110704519335,
      "grad_norm": 1.4932729005813599,
      "learning_rate": 5.647160753141978e-06,
      "loss": 0.8586052894592285,
      "step": 22200
    },
    {
      "epoch": 2.347067915505238,
      "grad_norm": 1.417768955230713,
      "learning_rate": 5.629646670679337e-06,
      "loss": 0.9558155059814453,
      "step": 22210
    },
    {
      "epoch": 2.3481247605585427,
      "grad_norm": 1.1663753986358643,
      "learning_rate": 5.612156342896952e-06,
      "loss": 0.8841658592224121,
      "step": 22220
    },
    {
      "epoch": 2.349181605611847,
      "grad_norm": 1.153314232826233,
      "learning_rate": 5.59468979124394e-06,
      "loss": 0.8298846244812011,
      "step": 22230
    },
    {
      "epoch": 2.350238450665152,
      "grad_norm": 1.3455487489700317,
      "learning_rate": 5.577247037140282e-06,
      "loss": 1.0092418670654297,
      "step": 22240
    },
    {
      "epoch": 2.3512952957184563,
      "grad_norm": 1.0661485195159912,
      "learning_rate": 5.559828101976758e-06,
      "loss": 0.8286648750305176,
      "step": 22250
    },
    {
      "epoch": 2.352352140771761,
      "grad_norm": 0.9532726407051086,
      "learning_rate": 5.542433007114939e-06,
      "loss": 0.8310384750366211,
      "step": 22260
    },
    {
      "epoch": 2.353408985825066,
      "grad_norm": 1.601749062538147,
      "learning_rate": 5.525061773887155e-06,
      "loss": 0.8679019927978515,
      "step": 22270
    },
    {
      "epoch": 2.3544658308783704,
      "grad_norm": 1.5838093757629395,
      "learning_rate": 5.507714423596499e-06,
      "loss": 0.8853231430053711,
      "step": 22280
    },
    {
      "epoch": 2.3555226759316747,
      "grad_norm": 1.3495243787765503,
      "learning_rate": 5.490390977516749e-06,
      "loss": 0.823605728149414,
      "step": 22290
    },
    {
      "epoch": 2.3565795209849796,
      "grad_norm": 0.32371050119400024,
      "learning_rate": 5.473091456892376e-06,
      "loss": 0.8216146469116211,
      "step": 22300
    },
    {
      "epoch": 2.3576363660382844,
      "grad_norm": 1.3388196229934692,
      "learning_rate": 5.455815882938508e-06,
      "loss": 0.7355672359466553,
      "step": 22310
    },
    {
      "epoch": 2.358693211091589,
      "grad_norm": 1.1211774349212646,
      "learning_rate": 5.4385642768409075e-06,
      "loss": 0.7688534736633301,
      "step": 22320
    },
    {
      "epoch": 2.3597500561448936,
      "grad_norm": 1.4913078546524048,
      "learning_rate": 5.421336659755958e-06,
      "loss": 0.9290646553039551,
      "step": 22330
    },
    {
      "epoch": 2.360806901198198,
      "grad_norm": 1.9259843826293945,
      "learning_rate": 5.4041330528106034e-06,
      "loss": 0.9917824745178223,
      "step": 22340
    },
    {
      "epoch": 2.361863746251503,
      "grad_norm": 1.5015066862106323,
      "learning_rate": 5.386953477102352e-06,
      "loss": 0.7632968425750732,
      "step": 22350
    },
    {
      "epoch": 2.3629205913048072,
      "grad_norm": 1.4189671277999878,
      "learning_rate": 5.3697979536992345e-06,
      "loss": 0.8901939392089844,
      "step": 22360
    },
    {
      "epoch": 2.363977436358112,
      "grad_norm": 1.7591795921325684,
      "learning_rate": 5.352666503639805e-06,
      "loss": 0.892976951599121,
      "step": 22370
    },
    {
      "epoch": 2.3650342814114165,
      "grad_norm": 1.4250272512435913,
      "learning_rate": 5.335559147933078e-06,
      "loss": 0.9519137382507324,
      "step": 22380
    },
    {
      "epoch": 2.3660911264647213,
      "grad_norm": 1.2947829961776733,
      "learning_rate": 5.318475907558523e-06,
      "loss": 0.8492285728454589,
      "step": 22390
    },
    {
      "epoch": 2.3671479715180257,
      "grad_norm": 1.3048641681671143,
      "learning_rate": 5.301416803466028e-06,
      "loss": 0.8885746002197266,
      "step": 22400
    },
    {
      "epoch": 2.3682048165713305,
      "grad_norm": 1.2906007766723633,
      "learning_rate": 5.2843818565759096e-06,
      "loss": 0.8271060943603515,
      "step": 22410
    },
    {
      "epoch": 2.369261661624635,
      "grad_norm": 1.82673180103302,
      "learning_rate": 5.26737108777883e-06,
      "loss": 0.8994010925292969,
      "step": 22420
    },
    {
      "epoch": 2.3703185066779398,
      "grad_norm": 1.5160003900527954,
      "learning_rate": 5.250384517935811e-06,
      "loss": 0.8481312751770019,
      "step": 22430
    },
    {
      "epoch": 2.371375351731244,
      "grad_norm": 1.2504756450653076,
      "learning_rate": 5.233422167878196e-06,
      "loss": 1.0348443031311034,
      "step": 22440
    },
    {
      "epoch": 2.372432196784549,
      "grad_norm": 1.6158336400985718,
      "learning_rate": 5.216484058407642e-06,
      "loss": 0.8417998313903808,
      "step": 22450
    },
    {
      "epoch": 2.3734890418378534,
      "grad_norm": 1.5011577606201172,
      "learning_rate": 5.1995702102960536e-06,
      "loss": 0.8816354751586915,
      "step": 22460
    },
    {
      "epoch": 2.374545886891158,
      "grad_norm": 0.006966839544475079,
      "learning_rate": 5.1826806442856035e-06,
      "loss": 0.8297355651855469,
      "step": 22470
    },
    {
      "epoch": 2.375602731944463,
      "grad_norm": 1.9482330083847046,
      "learning_rate": 5.165815381088668e-06,
      "loss": 0.7662561893463135,
      "step": 22480
    },
    {
      "epoch": 2.3766595769977674,
      "grad_norm": 1.1779075860977173,
      "learning_rate": 5.148974441387838e-06,
      "loss": 0.8810214996337891,
      "step": 22490
    },
    {
      "epoch": 2.377716422051072,
      "grad_norm": 1.3055111169815063,
      "learning_rate": 5.132157845835869e-06,
      "loss": 0.8754467010498047,
      "step": 22500
    },
    {
      "epoch": 2.377716422051072,
      "eval_loss": 0.48592498898506165,
      "eval_runtime": 3515.4853,
      "eval_samples_per_second": 2.533,
      "eval_steps_per_second": 2.533,
      "step": 22500
    },
    {
      "epoch": 2.3787732671043766,
      "grad_norm": 1.2479827404022217,
      "learning_rate": 5.1153656150556565e-06,
      "loss": 0.949852180480957,
      "step": 22510
    },
    {
      "epoch": 2.3798301121576815,
      "grad_norm": 1.5962028503417969,
      "learning_rate": 5.098597769640215e-06,
      "loss": 0.8807212829589843,
      "step": 22520
    },
    {
      "epoch": 2.380886957210986,
      "grad_norm": 1.484584927558899,
      "learning_rate": 5.081854330152669e-06,
      "loss": 0.8568793296813965,
      "step": 22530
    },
    {
      "epoch": 2.3819438022642907,
      "grad_norm": 1.4548224210739136,
      "learning_rate": 5.065135317126202e-06,
      "loss": 0.746189546585083,
      "step": 22540
    },
    {
      "epoch": 2.383000647317595,
      "grad_norm": 2.1082072257995605,
      "learning_rate": 5.048440751064043e-06,
      "loss": 0.9457200050354004,
      "step": 22550
    },
    {
      "epoch": 2.3840574923709,
      "grad_norm": 1.0722838640213013,
      "learning_rate": 5.031770652439433e-06,
      "loss": 0.9006097793579102,
      "step": 22560
    },
    {
      "epoch": 2.3851143374242043,
      "grad_norm": 1.38092041015625,
      "learning_rate": 5.0151250416956285e-06,
      "loss": 0.8838394165039063,
      "step": 22570
    },
    {
      "epoch": 2.386171182477509,
      "grad_norm": 1.0486894845962524,
      "learning_rate": 4.998503939245841e-06,
      "loss": 0.7889172077178955,
      "step": 22580
    },
    {
      "epoch": 2.3872280275308135,
      "grad_norm": 1.0623983144760132,
      "learning_rate": 4.981907365473224e-06,
      "loss": 0.8435184478759765,
      "step": 22590
    },
    {
      "epoch": 2.3882848725841184,
      "grad_norm": 1.6209259033203125,
      "learning_rate": 4.965335340730851e-06,
      "loss": 0.7904808044433593,
      "step": 22600
    },
    {
      "epoch": 2.3893417176374228,
      "grad_norm": 2.127464532852173,
      "learning_rate": 4.9487878853417074e-06,
      "loss": 0.8574931144714355,
      "step": 22610
    },
    {
      "epoch": 2.3903985626907276,
      "grad_norm": 1.1802486181259155,
      "learning_rate": 4.932265019598628e-06,
      "loss": 0.8761675834655762,
      "step": 22620
    },
    {
      "epoch": 2.391455407744032,
      "grad_norm": 1.41487455368042,
      "learning_rate": 4.915766763764298e-06,
      "loss": 0.8601420402526856,
      "step": 22630
    },
    {
      "epoch": 2.392512252797337,
      "grad_norm": 1.493059754371643,
      "learning_rate": 4.899293138071218e-06,
      "loss": 0.7219619750976562,
      "step": 22640
    },
    {
      "epoch": 2.393569097850641,
      "grad_norm": 1.1608288288116455,
      "learning_rate": 4.8828441627217e-06,
      "loss": 0.7906303405761719,
      "step": 22650
    },
    {
      "epoch": 2.394625942903946,
      "grad_norm": 1.369144320487976,
      "learning_rate": 4.866419857887811e-06,
      "loss": 0.9535908699035645,
      "step": 22660
    },
    {
      "epoch": 2.3956827879572504,
      "grad_norm": 1.1065051555633545,
      "learning_rate": 4.850020243711367e-06,
      "loss": 0.7766180515289307,
      "step": 22670
    },
    {
      "epoch": 2.3967396330105553,
      "grad_norm": 1.7036246061325073,
      "learning_rate": 4.833645340303902e-06,
      "loss": 0.9261297225952149,
      "step": 22680
    },
    {
      "epoch": 2.39779647806386,
      "grad_norm": 1.2312992811203003,
      "learning_rate": 4.817295167746657e-06,
      "loss": 0.8721088409423828,
      "step": 22690
    },
    {
      "epoch": 2.3988533231171645,
      "grad_norm": 0.9364454746246338,
      "learning_rate": 4.800969746090539e-06,
      "loss": 0.8372042655944825,
      "step": 22700
    },
    {
      "epoch": 2.399910168170469,
      "grad_norm": 1.8071925640106201,
      "learning_rate": 4.784669095356092e-06,
      "loss": 0.8741565704345703,
      "step": 22710
    },
    {
      "epoch": 2.4009670132237737,
      "grad_norm": 1.3230243921279907,
      "learning_rate": 4.7683932355334885e-06,
      "loss": 0.8548223495483398,
      "step": 22720
    },
    {
      "epoch": 2.4020238582770785,
      "grad_norm": 1.8477859497070312,
      "learning_rate": 4.752142186582514e-06,
      "loss": 0.8865860939025879,
      "step": 22730
    },
    {
      "epoch": 2.403080703330383,
      "grad_norm": 1.4208992719650269,
      "learning_rate": 4.735915968432508e-06,
      "loss": 0.8563644409179687,
      "step": 22740
    },
    {
      "epoch": 2.4041375483836878,
      "grad_norm": 1.088090419769287,
      "learning_rate": 4.719714600982364e-06,
      "loss": 0.8926313400268555,
      "step": 22750
    },
    {
      "epoch": 2.405194393436992,
      "grad_norm": 1.0687013864517212,
      "learning_rate": 4.703538104100494e-06,
      "loss": 0.7474089622497558,
      "step": 22760
    },
    {
      "epoch": 2.406251238490297,
      "grad_norm": 2.3841500282287598,
      "learning_rate": 4.687386497624832e-06,
      "loss": 0.9630848884582519,
      "step": 22770
    },
    {
      "epoch": 2.4073080835436014,
      "grad_norm": 1.2973947525024414,
      "learning_rate": 4.671259801362768e-06,
      "loss": 0.8640506744384766,
      "step": 22780
    },
    {
      "epoch": 2.408364928596906,
      "grad_norm": 1.5217310190200806,
      "learning_rate": 4.655158035091148e-06,
      "loss": 0.6639967441558838,
      "step": 22790
    },
    {
      "epoch": 2.4094217736502106,
      "grad_norm": 1.4702610969543457,
      "learning_rate": 4.639081218556238e-06,
      "loss": 0.8056562423706055,
      "step": 22800
    },
    {
      "epoch": 2.4104786187035154,
      "grad_norm": 1.429703712463379,
      "learning_rate": 4.623029371473731e-06,
      "loss": 0.8481059074401855,
      "step": 22810
    },
    {
      "epoch": 2.41153546375682,
      "grad_norm": 1.1393393278121948,
      "learning_rate": 4.607002513528674e-06,
      "loss": 0.8754949569702148,
      "step": 22820
    },
    {
      "epoch": 2.4125923088101247,
      "grad_norm": 0.977617084980011,
      "learning_rate": 4.591000664375483e-06,
      "loss": 0.8954818725585938,
      "step": 22830
    },
    {
      "epoch": 2.413649153863429,
      "grad_norm": 1.3991941213607788,
      "learning_rate": 4.575023843637893e-06,
      "loss": 1.0231868743896484,
      "step": 22840
    },
    {
      "epoch": 2.414705998916734,
      "grad_norm": 1.3157215118408203,
      "learning_rate": 4.559072070908951e-06,
      "loss": 0.849634838104248,
      "step": 22850
    },
    {
      "epoch": 2.4157628439700383,
      "grad_norm": 1.1848855018615723,
      "learning_rate": 4.5431453657510056e-06,
      "loss": 0.7842686176300049,
      "step": 22860
    },
    {
      "epoch": 2.416819689023343,
      "grad_norm": 1.2610275745391846,
      "learning_rate": 4.52724374769562e-06,
      "loss": 0.860294246673584,
      "step": 22870
    },
    {
      "epoch": 2.4178765340766475,
      "grad_norm": 1.1803334951400757,
      "learning_rate": 4.511367236243638e-06,
      "loss": 0.9370849609375,
      "step": 22880
    },
    {
      "epoch": 2.4189333791299523,
      "grad_norm": 1.5955299139022827,
      "learning_rate": 4.495515850865082e-06,
      "loss": 0.8480146408081055,
      "step": 22890
    },
    {
      "epoch": 2.419990224183257,
      "grad_norm": 1.2943682670593262,
      "learning_rate": 4.479689610999191e-06,
      "loss": 0.733238410949707,
      "step": 22900
    },
    {
      "epoch": 2.4210470692365615,
      "grad_norm": 1.5199047327041626,
      "learning_rate": 4.463888536054325e-06,
      "loss": 0.9120878219604492,
      "step": 22910
    },
    {
      "epoch": 2.422103914289866,
      "grad_norm": 1.403417944908142,
      "learning_rate": 4.448112645408026e-06,
      "loss": 0.9125898361206055,
      "step": 22920
    },
    {
      "epoch": 2.4231607593431708,
      "grad_norm": 1.4127452373504639,
      "learning_rate": 4.432361958406925e-06,
      "loss": 0.8227740287780761,
      "step": 22930
    },
    {
      "epoch": 2.4242176043964756,
      "grad_norm": 1.0714178085327148,
      "learning_rate": 4.416636494366766e-06,
      "loss": 0.8884865760803222,
      "step": 22940
    },
    {
      "epoch": 2.42527444944978,
      "grad_norm": 1.336416244506836,
      "learning_rate": 4.400936272572326e-06,
      "loss": 0.9314665794372559,
      "step": 22950
    },
    {
      "epoch": 2.426331294503085,
      "grad_norm": 1.4064407348632812,
      "learning_rate": 4.38526131227747e-06,
      "loss": 0.9632927894592285,
      "step": 22960
    },
    {
      "epoch": 2.427388139556389,
      "grad_norm": 1.204851508140564,
      "learning_rate": 4.369611632705043e-06,
      "loss": 0.8029598236083985,
      "step": 22970
    },
    {
      "epoch": 2.428444984609694,
      "grad_norm": 1.3317060470581055,
      "learning_rate": 4.3539872530469286e-06,
      "loss": 0.8091843605041504,
      "step": 22980
    },
    {
      "epoch": 2.4295018296629984,
      "grad_norm": 1.1220862865447998,
      "learning_rate": 4.338388192463935e-06,
      "loss": 0.9447767257690429,
      "step": 22990
    },
    {
      "epoch": 2.4305586747163033,
      "grad_norm": 1.038973331451416,
      "learning_rate": 4.32281447008587e-06,
      "loss": 0.8644864082336425,
      "step": 23000
    },
    {
      "epoch": 2.4305586747163033,
      "eval_loss": 0.4852326214313507,
      "eval_runtime": 3520.2948,
      "eval_samples_per_second": 2.53,
      "eval_steps_per_second": 2.53,
      "step": 23000
    },
    {
      "epoch": 2.4316155197696077,
      "grad_norm": 1.0539884567260742,
      "learning_rate": 4.307266105011429e-06,
      "loss": 0.8683359146118164,
      "step": 23010
    },
    {
      "epoch": 2.4326723648229125,
      "grad_norm": 1.1603624820709229,
      "learning_rate": 4.291743116308247e-06,
      "loss": 0.8582318305969239,
      "step": 23020
    },
    {
      "epoch": 2.433729209876217,
      "grad_norm": 1.5810259580612183,
      "learning_rate": 4.276245523012795e-06,
      "loss": 0.8695478439331055,
      "step": 23030
    },
    {
      "epoch": 2.4347860549295217,
      "grad_norm": 1.0569771528244019,
      "learning_rate": 4.2607733441304446e-06,
      "loss": 0.8560164451599122,
      "step": 23040
    },
    {
      "epoch": 2.435842899982826,
      "grad_norm": 1.2011107206344604,
      "learning_rate": 4.245326598635363e-06,
      "loss": 0.8591733932495117,
      "step": 23050
    },
    {
      "epoch": 2.436899745036131,
      "grad_norm": 1.4565941095352173,
      "learning_rate": 4.229905305470566e-06,
      "loss": 0.9155927658081054,
      "step": 23060
    },
    {
      "epoch": 2.4379565900894353,
      "grad_norm": 1.5875345468521118,
      "learning_rate": 4.2145094835478125e-06,
      "loss": 0.8588168144226074,
      "step": 23070
    },
    {
      "epoch": 2.43901343514274,
      "grad_norm": 1.05668044090271,
      "learning_rate": 4.199139151747664e-06,
      "loss": 0.8813247680664062,
      "step": 23080
    },
    {
      "epoch": 2.4400702801960445,
      "grad_norm": 1.632947325706482,
      "learning_rate": 4.183794328919391e-06,
      "loss": 0.8570672035217285,
      "step": 23090
    },
    {
      "epoch": 2.4411271252493494,
      "grad_norm": 1.0372904539108276,
      "learning_rate": 4.16847503388102e-06,
      "loss": 0.9538976669311523,
      "step": 23100
    },
    {
      "epoch": 2.442183970302654,
      "grad_norm": 0.9510229229927063,
      "learning_rate": 4.1531812854192196e-06,
      "loss": 0.9025184631347656,
      "step": 23110
    },
    {
      "epoch": 2.4432408153559586,
      "grad_norm": 1.178378701210022,
      "learning_rate": 4.137913102289376e-06,
      "loss": 0.8457724571228027,
      "step": 23120
    },
    {
      "epoch": 2.4442976604092634,
      "grad_norm": 1.28530752658844,
      "learning_rate": 4.122670503215495e-06,
      "loss": 0.8655507087707519,
      "step": 23130
    },
    {
      "epoch": 2.445354505462568,
      "grad_norm": 1.372251272201538,
      "learning_rate": 4.107453506890236e-06,
      "loss": 0.7378663539886474,
      "step": 23140
    },
    {
      "epoch": 2.4464113505158727,
      "grad_norm": 1.4338390827178955,
      "learning_rate": 4.0922621319748135e-06,
      "loss": 0.8116654396057129,
      "step": 23150
    },
    {
      "epoch": 2.447468195569177,
      "grad_norm": 1.4513016939163208,
      "learning_rate": 4.077096397099078e-06,
      "loss": 0.9057016372680664,
      "step": 23160
    },
    {
      "epoch": 2.448525040622482,
      "grad_norm": 1.3275303840637207,
      "learning_rate": 4.06195632086139e-06,
      "loss": 0.7659556865692139,
      "step": 23170
    },
    {
      "epoch": 2.4495818856757863,
      "grad_norm": 1.1128970384597778,
      "learning_rate": 4.046841921828681e-06,
      "loss": 0.7971051692962646,
      "step": 23180
    },
    {
      "epoch": 2.450638730729091,
      "grad_norm": 1.180591106414795,
      "learning_rate": 4.031753218536355e-06,
      "loss": 0.7383405685424804,
      "step": 23190
    },
    {
      "epoch": 2.4516955757823955,
      "grad_norm": 1.1126251220703125,
      "learning_rate": 4.01669022948834e-06,
      "loss": 0.8407711029052735,
      "step": 23200
    },
    {
      "epoch": 2.4527524208357003,
      "grad_norm": 1.125542163848877,
      "learning_rate": 4.001652973157002e-06,
      "loss": 0.8256967544555665,
      "step": 23210
    },
    {
      "epoch": 2.4538092658890047,
      "grad_norm": 1.332027792930603,
      "learning_rate": 3.986641467983179e-06,
      "loss": 0.9701010704040527,
      "step": 23220
    },
    {
      "epoch": 2.4548661109423096,
      "grad_norm": 2.1409993171691895,
      "learning_rate": 3.97165573237609e-06,
      "loss": 0.7986093044281006,
      "step": 23230
    },
    {
      "epoch": 2.455922955995614,
      "grad_norm": 1.382464051246643,
      "learning_rate": 3.95669578471339e-06,
      "loss": 0.852566146850586,
      "step": 23240
    },
    {
      "epoch": 2.4569798010489188,
      "grad_norm": 1.0755739212036133,
      "learning_rate": 3.941761643341086e-06,
      "loss": 0.840428352355957,
      "step": 23250
    },
    {
      "epoch": 2.458036646102223,
      "grad_norm": 1.2600350379943848,
      "learning_rate": 3.926853326573554e-06,
      "loss": 0.8778380393981934,
      "step": 23260
    },
    {
      "epoch": 2.459093491155528,
      "grad_norm": 0.9600110650062561,
      "learning_rate": 3.911970852693472e-06,
      "loss": 0.8465672492980957,
      "step": 23270
    },
    {
      "epoch": 2.4601503362088324,
      "grad_norm": 1.3003545999526978,
      "learning_rate": 3.897114239951866e-06,
      "loss": 0.9392643928527832,
      "step": 23280
    },
    {
      "epoch": 2.4612071812621372,
      "grad_norm": 1.2168279886245728,
      "learning_rate": 3.882283506568008e-06,
      "loss": 0.8372897148132324,
      "step": 23290
    },
    {
      "epoch": 2.4622640263154416,
      "grad_norm": 1.267059326171875,
      "learning_rate": 3.867478670729471e-06,
      "loss": 0.8479422569274903,
      "step": 23300
    },
    {
      "epoch": 2.4633208713687464,
      "grad_norm": 1.5321176052093506,
      "learning_rate": 3.852699750592026e-06,
      "loss": 0.8081297874450684,
      "step": 23310
    },
    {
      "epoch": 2.4643777164220513,
      "grad_norm": 1.3649382591247559,
      "learning_rate": 3.837946764279701e-06,
      "loss": 0.7799179553985596,
      "step": 23320
    },
    {
      "epoch": 2.4654345614753557,
      "grad_norm": 1.0977129936218262,
      "learning_rate": 3.8232197298846945e-06,
      "loss": 0.9287117004394532,
      "step": 23330
    },
    {
      "epoch": 2.4664914065286605,
      "grad_norm": 0.9643582701683044,
      "learning_rate": 3.8085186654674088e-06,
      "loss": 0.8689750671386719,
      "step": 23340
    },
    {
      "epoch": 2.467548251581965,
      "grad_norm": 1.5677587985992432,
      "learning_rate": 3.7938435890563506e-06,
      "loss": 0.8598687171936035,
      "step": 23350
    },
    {
      "epoch": 2.4686050966352697,
      "grad_norm": 1.2845227718353271,
      "learning_rate": 3.7791945186482046e-06,
      "loss": 0.7723999977111816,
      "step": 23360
    },
    {
      "epoch": 2.469661941688574,
      "grad_norm": 1.3694114685058594,
      "learning_rate": 3.7645714722077287e-06,
      "loss": 0.8388255119323731,
      "step": 23370
    },
    {
      "epoch": 2.470718786741879,
      "grad_norm": 1.5750685930252075,
      "learning_rate": 3.7499744676678024e-06,
      "loss": 0.8922327041625977,
      "step": 23380
    },
    {
      "epoch": 2.4717756317951833,
      "grad_norm": 1.4409635066986084,
      "learning_rate": 3.7354035229293206e-06,
      "loss": 1.0040422439575196,
      "step": 23390
    },
    {
      "epoch": 2.472832476848488,
      "grad_norm": 1.0123897790908813,
      "learning_rate": 3.7208586558612647e-06,
      "loss": 0.7927358150482178,
      "step": 23400
    },
    {
      "epoch": 2.4738893219017926,
      "grad_norm": 1.260427713394165,
      "learning_rate": 3.7063398843006135e-06,
      "loss": 0.8372198104858398,
      "step": 23410
    },
    {
      "epoch": 2.4749461669550974,
      "grad_norm": 1.3015683889389038,
      "learning_rate": 3.6918472260523474e-06,
      "loss": 0.8817134857177734,
      "step": 23420
    },
    {
      "epoch": 2.476003012008402,
      "grad_norm": 1.120436668395996,
      "learning_rate": 3.6773806988894238e-06,
      "loss": 0.968020248413086,
      "step": 23430
    },
    {
      "epoch": 2.4770598570617066,
      "grad_norm": 1.3525846004486084,
      "learning_rate": 3.6629403205527464e-06,
      "loss": 0.8205045700073242,
      "step": 23440
    },
    {
      "epoch": 2.478116702115011,
      "grad_norm": 1.798214077949524,
      "learning_rate": 3.648526108751174e-06,
      "loss": 1.0139251708984376,
      "step": 23450
    },
    {
      "epoch": 2.479173547168316,
      "grad_norm": 1.311659336090088,
      "learning_rate": 3.634138081161456e-06,
      "loss": 0.8796846389770507,
      "step": 23460
    },
    {
      "epoch": 2.4802303922216202,
      "grad_norm": 1.5185130834579468,
      "learning_rate": 3.619776255428234e-06,
      "loss": 0.9015286445617676,
      "step": 23470
    },
    {
      "epoch": 2.481287237274925,
      "grad_norm": 2.0761685371398926,
      "learning_rate": 3.605440649164016e-06,
      "loss": 0.8657657623291015,
      "step": 23480
    },
    {
      "epoch": 2.48234408232823,
      "grad_norm": 1.1563559770584106,
      "learning_rate": 3.5911312799491697e-06,
      "loss": 0.9102652549743653,
      "step": 23490
    },
    {
      "epoch": 2.4834009273815343,
      "grad_norm": 0.982991635799408,
      "learning_rate": 3.5768481653318755e-06,
      "loss": 0.8500093460083008,
      "step": 23500
    },
    {
      "epoch": 2.4834009273815343,
      "eval_loss": 0.4841485023498535,
      "eval_runtime": 3080.2606,
      "eval_samples_per_second": 2.891,
      "eval_steps_per_second": 2.891,
      "step": 23500
    },
    {
      "epoch": 2.4844577724348387,
      "grad_norm": 1.9158236980438232,
      "learning_rate": 3.562591322828113e-06,
      "loss": 0.7097861289978027,
      "step": 23510
    },
    {
      "epoch": 2.4855146174881435,
      "grad_norm": 1.2013194561004639,
      "learning_rate": 3.5483607699216465e-06,
      "loss": 0.8366837501525879,
      "step": 23520
    },
    {
      "epoch": 2.4865714625414483,
      "grad_norm": 0.9812779426574707,
      "learning_rate": 3.534156524064017e-06,
      "loss": 0.8759609222412109,
      "step": 23530
    },
    {
      "epoch": 2.4876283075947527,
      "grad_norm": 1.659834623336792,
      "learning_rate": 3.5199786026744757e-06,
      "loss": 0.870569896697998,
      "step": 23540
    },
    {
      "epoch": 2.4886851526480576,
      "grad_norm": 1.4421653747558594,
      "learning_rate": 3.5058270231400127e-06,
      "loss": 0.9578547477722168,
      "step": 23550
    },
    {
      "epoch": 2.489741997701362,
      "grad_norm": 1.1133043766021729,
      "learning_rate": 3.491701802815295e-06,
      "loss": 0.9503613471984863,
      "step": 23560
    },
    {
      "epoch": 2.490798842754667,
      "grad_norm": 1.2955139875411987,
      "learning_rate": 3.4776029590226884e-06,
      "loss": 0.8230159759521485,
      "step": 23570
    },
    {
      "epoch": 2.491855687807971,
      "grad_norm": 1.4397048950195312,
      "learning_rate": 3.463530509052193e-06,
      "loss": 0.8947482109069824,
      "step": 23580
    },
    {
      "epoch": 2.492912532861276,
      "grad_norm": 1.369035005569458,
      "learning_rate": 3.4494844701614487e-06,
      "loss": 0.823580265045166,
      "step": 23590
    },
    {
      "epoch": 2.4939693779145804,
      "grad_norm": 1.1338411569595337,
      "learning_rate": 3.4354648595756987e-06,
      "loss": 0.8480932235717773,
      "step": 23600
    },
    {
      "epoch": 2.4950262229678852,
      "grad_norm": 1.273505449295044,
      "learning_rate": 3.4214716944877904e-06,
      "loss": 0.8858339309692382,
      "step": 23610
    },
    {
      "epoch": 2.4960830680211896,
      "grad_norm": 0.9313640594482422,
      "learning_rate": 3.4075049920581305e-06,
      "loss": 0.8909448623657227,
      "step": 23620
    },
    {
      "epoch": 2.4971399130744945,
      "grad_norm": 1.2421674728393555,
      "learning_rate": 3.3935647694146726e-06,
      "loss": 0.8180133819580078,
      "step": 23630
    },
    {
      "epoch": 2.498196758127799,
      "grad_norm": 1.4765287637710571,
      "learning_rate": 3.37965104365289e-06,
      "loss": 0.7960214614868164,
      "step": 23640
    },
    {
      "epoch": 2.4992536031811037,
      "grad_norm": 1.4904160499572754,
      "learning_rate": 3.365763831835786e-06,
      "loss": 0.7986099243164062,
      "step": 23650
    },
    {
      "epoch": 2.500310448234408,
      "grad_norm": 1.1891931295394897,
      "learning_rate": 3.3519031509938265e-06,
      "loss": 0.8491926193237305,
      "step": 23660
    },
    {
      "epoch": 2.501367293287713,
      "grad_norm": 1.6332101821899414,
      "learning_rate": 3.338069018124948e-06,
      "loss": 0.9066998481750488,
      "step": 23670
    },
    {
      "epoch": 2.5024241383410173,
      "grad_norm": 1.2750554084777832,
      "learning_rate": 3.324261450194527e-06,
      "loss": 0.6907270431518555,
      "step": 23680
    },
    {
      "epoch": 2.503480983394322,
      "grad_norm": 1.5005512237548828,
      "learning_rate": 3.310480464135374e-06,
      "loss": 0.7863218784332275,
      "step": 23690
    },
    {
      "epoch": 2.504537828447627,
      "grad_norm": 1.5355650186538696,
      "learning_rate": 3.2967260768476867e-06,
      "loss": 0.9214285850524903,
      "step": 23700
    },
    {
      "epoch": 2.5055946735009313,
      "grad_norm": 2.1079940795898438,
      "learning_rate": 3.282998305199053e-06,
      "loss": 0.8672579765319824,
      "step": 23710
    },
    {
      "epoch": 2.5066515185542357,
      "grad_norm": 1.6121920347213745,
      "learning_rate": 3.2692971660244097e-06,
      "loss": 0.7592312812805175,
      "step": 23720
    },
    {
      "epoch": 2.5077083636075406,
      "grad_norm": 1.1751784086227417,
      "learning_rate": 3.2556226761260524e-06,
      "loss": 0.9205509185791015,
      "step": 23730
    },
    {
      "epoch": 2.5087652086608454,
      "grad_norm": 1.0793232917785645,
      "learning_rate": 3.24197485227358e-06,
      "loss": 0.82030029296875,
      "step": 23740
    },
    {
      "epoch": 2.50982205371415,
      "grad_norm": 1.5924186706542969,
      "learning_rate": 3.2283537112038908e-06,
      "loss": 0.8374110221862793,
      "step": 23750
    },
    {
      "epoch": 2.510878898767454,
      "grad_norm": 1.0167285203933716,
      "learning_rate": 3.2147592696211594e-06,
      "loss": 0.9616507530212403,
      "step": 23760
    },
    {
      "epoch": 2.511935743820759,
      "grad_norm": 1.58920419216156,
      "learning_rate": 3.2011915441968354e-06,
      "loss": 0.8993430137634277,
      "step": 23770
    },
    {
      "epoch": 2.512992588874064,
      "grad_norm": 1.3038891553878784,
      "learning_rate": 3.187650551569582e-06,
      "loss": 0.896573829650879,
      "step": 23780
    },
    {
      "epoch": 2.5140494339273682,
      "grad_norm": 1.2767966985702515,
      "learning_rate": 3.1741363083452936e-06,
      "loss": 0.7777239322662354,
      "step": 23790
    },
    {
      "epoch": 2.515106278980673,
      "grad_norm": 1.7405807971954346,
      "learning_rate": 3.160648831097049e-06,
      "loss": 0.8474624633789063,
      "step": 23800
    },
    {
      "epoch": 2.5161631240339775,
      "grad_norm": 1.740067720413208,
      "learning_rate": 3.14718813636512e-06,
      "loss": 0.8864502906799316,
      "step": 23810
    },
    {
      "epoch": 2.5172199690872823,
      "grad_norm": 1.4359275102615356,
      "learning_rate": 3.1337542406569194e-06,
      "loss": 0.8431582450866699,
      "step": 23820
    },
    {
      "epoch": 2.5182768141405867,
      "grad_norm": 1.7237999439239502,
      "learning_rate": 3.1203471604469958e-06,
      "loss": 0.9572831153869629,
      "step": 23830
    },
    {
      "epoch": 2.5193336591938915,
      "grad_norm": 1.599173665046692,
      "learning_rate": 3.1069669121770145e-06,
      "loss": 0.8382308006286621,
      "step": 23840
    },
    {
      "epoch": 2.520390504247196,
      "grad_norm": 1.1189723014831543,
      "learning_rate": 3.0936135122557495e-06,
      "loss": 0.8675414085388183,
      "step": 23850
    },
    {
      "epoch": 2.5214473493005007,
      "grad_norm": 1.8953349590301514,
      "learning_rate": 3.080286977059027e-06,
      "loss": 0.9011523246765136,
      "step": 23860
    },
    {
      "epoch": 2.5225041943538056,
      "grad_norm": 1.5016778707504272,
      "learning_rate": 3.066987322929743e-06,
      "loss": 0.882810115814209,
      "step": 23870
    },
    {
      "epoch": 2.52356103940711,
      "grad_norm": 1.9474232196807861,
      "learning_rate": 3.053714566177818e-06,
      "loss": 0.8711090087890625,
      "step": 23880
    },
    {
      "epoch": 2.5246178844604144,
      "grad_norm": 1.2045246362686157,
      "learning_rate": 3.040468723080203e-06,
      "loss": 0.8886404991149902,
      "step": 23890
    },
    {
      "epoch": 2.525674729513719,
      "grad_norm": 1.9790517091751099,
      "learning_rate": 3.027249809880825e-06,
      "loss": 0.9234767913818359,
      "step": 23900
    },
    {
      "epoch": 2.526731574567024,
      "grad_norm": 1.4411211013793945,
      "learning_rate": 3.014057842790599e-06,
      "loss": 0.8274395942687989,
      "step": 23910
    },
    {
      "epoch": 2.5277884196203284,
      "grad_norm": 1.463689923286438,
      "learning_rate": 3.000892837987379e-06,
      "loss": 0.9284679412841796,
      "step": 23920
    },
    {
      "epoch": 2.528845264673633,
      "grad_norm": 1.623043417930603,
      "learning_rate": 2.987754811615981e-06,
      "loss": 0.8507776260375977,
      "step": 23930
    },
    {
      "epoch": 2.5299021097269376,
      "grad_norm": 1.5722020864486694,
      "learning_rate": 2.974643779788111e-06,
      "loss": 0.8275420188903808,
      "step": 23940
    },
    {
      "epoch": 2.5309589547802425,
      "grad_norm": 1.5392374992370605,
      "learning_rate": 2.961559758582383e-06,
      "loss": 0.9033407211303711,
      "step": 23950
    },
    {
      "epoch": 2.532015799833547,
      "grad_norm": 1.255174160003662,
      "learning_rate": 2.948502764044281e-06,
      "loss": 0.9223237037658691,
      "step": 23960
    },
    {
      "epoch": 2.5330726448868517,
      "grad_norm": 1.157383680343628,
      "learning_rate": 2.9354728121861407e-06,
      "loss": 0.8584443092346191,
      "step": 23970
    },
    {
      "epoch": 2.534129489940156,
      "grad_norm": 1.154889464378357,
      "learning_rate": 2.922469918987156e-06,
      "loss": 0.8240571975708008,
      "step": 23980
    },
    {
      "epoch": 2.535186334993461,
      "grad_norm": 1.5352853536605835,
      "learning_rate": 2.9094941003933117e-06,
      "loss": 0.8902150154113769,
      "step": 23990
    },
    {
      "epoch": 2.5362431800467653,
      "grad_norm": 0.9624236822128296,
      "learning_rate": 2.8965453723174078e-06,
      "loss": 0.8698938369750977,
      "step": 24000
    },
    {
      "epoch": 2.5362431800467653,
      "eval_loss": 0.4842025637626648,
      "eval_runtime": 3086.0173,
      "eval_samples_per_second": 2.886,
      "eval_steps_per_second": 2.886,
      "step": 24000
    },
    {
      "epoch": 2.53730002510007,
      "grad_norm": 1.978960394859314,
      "learning_rate": 2.883623750639003e-06,
      "loss": 0.8591943740844726,
      "step": 24010
    },
    {
      "epoch": 2.5383568701533745,
      "grad_norm": 1.096195936203003,
      "learning_rate": 2.8707292512044405e-06,
      "loss": 0.7295454978942871,
      "step": 24020
    },
    {
      "epoch": 2.5394137152066794,
      "grad_norm": 1.5400471687316895,
      "learning_rate": 2.8578618898267824e-06,
      "loss": 0.7907662868499756,
      "step": 24030
    },
    {
      "epoch": 2.5404705602599837,
      "grad_norm": 1.217390775680542,
      "learning_rate": 2.8450216822858143e-06,
      "loss": 0.9901625633239746,
      "step": 24040
    },
    {
      "epoch": 2.5415274053132886,
      "grad_norm": 1.4197161197662354,
      "learning_rate": 2.8322086443280217e-06,
      "loss": 0.8910771369934082,
      "step": 24050
    },
    {
      "epoch": 2.542584250366593,
      "grad_norm": 1.083265781402588,
      "learning_rate": 2.8194227916665788e-06,
      "loss": 0.804392147064209,
      "step": 24060
    },
    {
      "epoch": 2.543641095419898,
      "grad_norm": 1.29573392868042,
      "learning_rate": 2.8066641399813133e-06,
      "loss": 0.8007929801940918,
      "step": 24070
    },
    {
      "epoch": 2.5446979404732026,
      "grad_norm": 1.2519774436950684,
      "learning_rate": 2.7939327049186946e-06,
      "loss": 0.7949441432952881,
      "step": 24080
    },
    {
      "epoch": 2.545754785526507,
      "grad_norm": 1.0886164903640747,
      "learning_rate": 2.781228502091818e-06,
      "loss": 0.7969067096710205,
      "step": 24090
    },
    {
      "epoch": 2.5468116305798114,
      "grad_norm": 1.3556891679763794,
      "learning_rate": 2.768551547080386e-06,
      "loss": 0.9263777732849121,
      "step": 24100
    },
    {
      "epoch": 2.5478684756331162,
      "grad_norm": 1.542667269706726,
      "learning_rate": 2.7559018554306813e-06,
      "loss": 0.8447686195373535,
      "step": 24110
    },
    {
      "epoch": 2.548925320686421,
      "grad_norm": 1.4266626834869385,
      "learning_rate": 2.7432794426555555e-06,
      "loss": 0.9068968772888184,
      "step": 24120
    },
    {
      "epoch": 2.5499821657397255,
      "grad_norm": 0.9692893028259277,
      "learning_rate": 2.730684324234395e-06,
      "loss": 0.9383960723876953,
      "step": 24130
    },
    {
      "epoch": 2.55103901079303,
      "grad_norm": 1.062852382659912,
      "learning_rate": 2.718116515613137e-06,
      "loss": 0.8981532096862793,
      "step": 24140
    },
    {
      "epoch": 2.5520958558463347,
      "grad_norm": 1.4505829811096191,
      "learning_rate": 2.70557603220421e-06,
      "loss": 0.9334762573242188,
      "step": 24150
    },
    {
      "epoch": 2.5531527008996395,
      "grad_norm": 1.017637848854065,
      "learning_rate": 2.6930628893865355e-06,
      "loss": 0.8380067825317383,
      "step": 24160
    },
    {
      "epoch": 2.554209545952944,
      "grad_norm": 0.9763848185539246,
      "learning_rate": 2.680577102505505e-06,
      "loss": 0.9251277923583985,
      "step": 24170
    },
    {
      "epoch": 2.5552663910062487,
      "grad_norm": 1.4261773824691772,
      "learning_rate": 2.668118686872972e-06,
      "loss": 0.8821608543395996,
      "step": 24180
    },
    {
      "epoch": 2.556323236059553,
      "grad_norm": 1.8000972270965576,
      "learning_rate": 2.655687657767214e-06,
      "loss": 0.7949012279510498,
      "step": 24190
    },
    {
      "epoch": 2.557380081112858,
      "grad_norm": 1.0616176128387451,
      "learning_rate": 2.643284030432924e-06,
      "loss": 0.8193815231323243,
      "step": 24200
    },
    {
      "epoch": 2.5584369261661624,
      "grad_norm": 1.903865098953247,
      "learning_rate": 2.630907820081188e-06,
      "loss": 0.7629337310791016,
      "step": 24210
    },
    {
      "epoch": 2.559493771219467,
      "grad_norm": 1.1355401277542114,
      "learning_rate": 2.6185590418894847e-06,
      "loss": 0.8050679206848145,
      "step": 24220
    },
    {
      "epoch": 2.5605506162727716,
      "grad_norm": 1.3910589218139648,
      "learning_rate": 2.6062377110016354e-06,
      "loss": 0.8422387123107911,
      "step": 24230
    },
    {
      "epoch": 2.5616074613260764,
      "grad_norm": 1.4090571403503418,
      "learning_rate": 2.5939438425278073e-06,
      "loss": 0.910915756225586,
      "step": 24240
    },
    {
      "epoch": 2.562664306379381,
      "grad_norm": 1.177096962928772,
      "learning_rate": 2.581677451544484e-06,
      "loss": 0.7913078308105469,
      "step": 24250
    },
    {
      "epoch": 2.5637211514326856,
      "grad_norm": 1.266347050666809,
      "learning_rate": 2.5694385530944665e-06,
      "loss": 0.8032961845397949,
      "step": 24260
    },
    {
      "epoch": 2.56477799648599,
      "grad_norm": 1.1477205753326416,
      "learning_rate": 2.5572271621868287e-06,
      "loss": 0.826166820526123,
      "step": 24270
    },
    {
      "epoch": 2.565834841539295,
      "grad_norm": 1.3714931011199951,
      "learning_rate": 2.5450432937969144e-06,
      "loss": 0.7762300491333007,
      "step": 24280
    },
    {
      "epoch": 2.5668916865925997,
      "grad_norm": 1.5450215339660645,
      "learning_rate": 2.532886962866307e-06,
      "loss": 0.8306907653808594,
      "step": 24290
    },
    {
      "epoch": 2.567948531645904,
      "grad_norm": 0.9924198389053345,
      "learning_rate": 2.5207581843028376e-06,
      "loss": 0.7906289100646973,
      "step": 24300
    },
    {
      "epoch": 2.5690053766992085,
      "grad_norm": 1.0977224111557007,
      "learning_rate": 2.508656972980533e-06,
      "loss": 0.8363621711730957,
      "step": 24310
    },
    {
      "epoch": 2.5700622217525133,
      "grad_norm": 1.1323976516723633,
      "learning_rate": 2.4965833437396196e-06,
      "loss": 0.8123137474060058,
      "step": 24320
    },
    {
      "epoch": 2.571119066805818,
      "grad_norm": 1.4290274381637573,
      "learning_rate": 2.4845373113864952e-06,
      "loss": 0.8087060928344727,
      "step": 24330
    },
    {
      "epoch": 2.5721759118591225,
      "grad_norm": 1.274782419204712,
      "learning_rate": 2.47251889069372e-06,
      "loss": 0.8340163230895996,
      "step": 24340
    },
    {
      "epoch": 2.573232756912427,
      "grad_norm": 1.2931162118911743,
      "learning_rate": 2.4605280963999887e-06,
      "loss": 0.9275397300720215,
      "step": 24350
    },
    {
      "epoch": 2.5742896019657318,
      "grad_norm": 1.2593858242034912,
      "learning_rate": 2.448564943210119e-06,
      "loss": 0.8701230049133301,
      "step": 24360
    },
    {
      "epoch": 2.5753464470190366,
      "grad_norm": 1.105502963066101,
      "learning_rate": 2.436629445795022e-06,
      "loss": 0.8823878288269043,
      "step": 24370
    },
    {
      "epoch": 2.576403292072341,
      "grad_norm": 1.299667477607727,
      "learning_rate": 2.4247216187917087e-06,
      "loss": 0.8870471000671387,
      "step": 24380
    },
    {
      "epoch": 2.577460137125646,
      "grad_norm": 1.3929492235183716,
      "learning_rate": 2.4128414768032516e-06,
      "loss": 0.9132563591003418,
      "step": 24390
    },
    {
      "epoch": 2.57851698217895,
      "grad_norm": 1.2267035245895386,
      "learning_rate": 2.4009890343987596e-06,
      "loss": 0.7918870925903321,
      "step": 24400
    },
    {
      "epoch": 2.579573827232255,
      "grad_norm": 1.4647573232650757,
      "learning_rate": 2.389164306113384e-06,
      "loss": 0.8331314086914062,
      "step": 24410
    },
    {
      "epoch": 2.5806306722855594,
      "grad_norm": 1.518871545791626,
      "learning_rate": 2.377367306448297e-06,
      "loss": 0.8936835289001465,
      "step": 24420
    },
    {
      "epoch": 2.5816875173388643,
      "grad_norm": 1.0814834833145142,
      "learning_rate": 2.365598049870646e-06,
      "loss": 0.8019541740417481,
      "step": 24430
    },
    {
      "epoch": 2.5827443623921686,
      "grad_norm": 1.2545267343521118,
      "learning_rate": 2.353856550813571e-06,
      "loss": 0.9050207138061523,
      "step": 24440
    },
    {
      "epoch": 2.5838012074454735,
      "grad_norm": 1.1497942209243774,
      "learning_rate": 2.3421428236761635e-06,
      "loss": 0.8207810401916504,
      "step": 24450
    },
    {
      "epoch": 2.584858052498778,
      "grad_norm": 1.23008394241333,
      "learning_rate": 2.3304568828234647e-06,
      "loss": 0.741386079788208,
      "step": 24460
    },
    {
      "epoch": 2.5859148975520827,
      "grad_norm": 1.4558444023132324,
      "learning_rate": 2.318798742586434e-06,
      "loss": 0.8894210815429687,
      "step": 24470
    },
    {
      "epoch": 2.586971742605387,
      "grad_norm": 1.5478399991989136,
      "learning_rate": 2.3071684172619375e-06,
      "loss": 0.8352909088134766,
      "step": 24480
    },
    {
      "epoch": 2.588028587658692,
      "grad_norm": 1.3600914478302002,
      "learning_rate": 2.295565921112733e-06,
      "loss": 0.8790791511535645,
      "step": 24490
    },
    {
      "epoch": 2.5890854327119968,
      "grad_norm": 0.9909800887107849,
      "learning_rate": 2.283991268367458e-06,
      "loss": 0.84494047164917,
      "step": 24500
    },
    {
      "epoch": 2.5890854327119968,
      "eval_loss": 0.48412781953811646,
      "eval_runtime": 3585.4875,
      "eval_samples_per_second": 2.484,
      "eval_steps_per_second": 2.484,
      "step": 24500
    },
    {
      "epoch": 2.590142277765301,
      "grad_norm": 1.7249137163162231,
      "learning_rate": 2.272444473220595e-06,
      "loss": 0.8941320419311524,
      "step": 24510
    },
    {
      "epoch": 2.5911991228186055,
      "grad_norm": 2.0033507347106934,
      "learning_rate": 2.2609255498324518e-06,
      "loss": 0.879764461517334,
      "step": 24520
    },
    {
      "epoch": 2.5922559678719104,
      "grad_norm": 1.3314927816390991,
      "learning_rate": 2.249434512329185e-06,
      "loss": 0.9309842109680175,
      "step": 24530
    },
    {
      "epoch": 2.593312812925215,
      "grad_norm": 1.602346420288086,
      "learning_rate": 2.237971374802725e-06,
      "loss": 0.7171894073486328,
      "step": 24540
    },
    {
      "epoch": 2.5943696579785196,
      "grad_norm": 1.2939352989196777,
      "learning_rate": 2.2265361513108206e-06,
      "loss": 0.9081261634826661,
      "step": 24550
    },
    {
      "epoch": 2.595426503031824,
      "grad_norm": 1.5555986166000366,
      "learning_rate": 2.215128855876941e-06,
      "loss": 0.85440034866333,
      "step": 24560
    },
    {
      "epoch": 2.596483348085129,
      "grad_norm": 1.4534785747528076,
      "learning_rate": 2.203749502490354e-06,
      "loss": 0.854392147064209,
      "step": 24570
    },
    {
      "epoch": 2.5975401931384337,
      "grad_norm": 1.130503535270691,
      "learning_rate": 2.1923981051060305e-06,
      "loss": 0.8787725448608399,
      "step": 24580
    },
    {
      "epoch": 2.598597038191738,
      "grad_norm": 0.5307117104530334,
      "learning_rate": 2.181074677644676e-06,
      "loss": 0.6440413951873779,
      "step": 24590
    },
    {
      "epoch": 2.599653883245043,
      "grad_norm": 1.0976006984710693,
      "learning_rate": 2.169779233992672e-06,
      "loss": 0.7589848518371582,
      "step": 24600
    },
    {
      "epoch": 2.6007107282983473,
      "grad_norm": 1.4910426139831543,
      "learning_rate": 2.15851178800211e-06,
      "loss": 0.9413095474243164,
      "step": 24610
    },
    {
      "epoch": 2.601767573351652,
      "grad_norm": 1.3141834735870361,
      "learning_rate": 2.147272353490723e-06,
      "loss": 0.8296442985534668,
      "step": 24620
    },
    {
      "epoch": 2.6028244184049565,
      "grad_norm": 1.185226321220398,
      "learning_rate": 2.136060944241919e-06,
      "loss": 0.8214632987976074,
      "step": 24630
    },
    {
      "epoch": 2.6038812634582613,
      "grad_norm": 1.2331238985061646,
      "learning_rate": 2.1248775740047022e-06,
      "loss": 0.8318029403686523,
      "step": 24640
    },
    {
      "epoch": 2.6049381085115657,
      "grad_norm": 1.777467966079712,
      "learning_rate": 2.113722256493725e-06,
      "loss": 0.8425689697265625,
      "step": 24650
    },
    {
      "epoch": 2.6059949535648705,
      "grad_norm": 1.643373727798462,
      "learning_rate": 2.102595005389213e-06,
      "loss": 0.8838297843933105,
      "step": 24660
    },
    {
      "epoch": 2.607051798618175,
      "grad_norm": 1.9544092416763306,
      "learning_rate": 2.0914958343369956e-06,
      "loss": 0.9341596603393555,
      "step": 24670
    },
    {
      "epoch": 2.6081086436714798,
      "grad_norm": 1.04557204246521,
      "learning_rate": 2.0804247569484377e-06,
      "loss": 0.8369438171386718,
      "step": 24680
    },
    {
      "epoch": 2.609165488724784,
      "grad_norm": 1.6333458423614502,
      "learning_rate": 2.069381786800481e-06,
      "loss": 0.8260891914367676,
      "step": 24690
    },
    {
      "epoch": 2.610222333778089,
      "grad_norm": 1.8241239786148071,
      "learning_rate": 2.058366937435577e-06,
      "loss": 0.8719179153442382,
      "step": 24700
    },
    {
      "epoch": 2.611279178831394,
      "grad_norm": 1.0632768869400024,
      "learning_rate": 2.047380222361711e-06,
      "loss": 0.90439453125,
      "step": 24710
    },
    {
      "epoch": 2.612336023884698,
      "grad_norm": 1.5980945825576782,
      "learning_rate": 2.036421655052334e-06,
      "loss": 0.7425655364990235,
      "step": 24720
    },
    {
      "epoch": 2.6133928689380026,
      "grad_norm": 1.3114948272705078,
      "learning_rate": 2.025491248946418e-06,
      "loss": 0.9221746444702148,
      "step": 24730
    },
    {
      "epoch": 2.6144497139913074,
      "grad_norm": 1.0135310888290405,
      "learning_rate": 2.0145890174483706e-06,
      "loss": 0.778179407119751,
      "step": 24740
    },
    {
      "epoch": 2.6155065590446123,
      "grad_norm": 1.6120519638061523,
      "learning_rate": 2.003714973928067e-06,
      "loss": 0.7764588832855225,
      "step": 24750
    },
    {
      "epoch": 2.6165634040979167,
      "grad_norm": 1.6988710165023804,
      "learning_rate": 1.99286913172079e-06,
      "loss": 0.9033034324645997,
      "step": 24760
    },
    {
      "epoch": 2.617620249151221,
      "grad_norm": 1.2013013362884521,
      "learning_rate": 1.9820515041272636e-06,
      "loss": 0.870674991607666,
      "step": 24770
    },
    {
      "epoch": 2.618677094204526,
      "grad_norm": 1.7562565803527832,
      "learning_rate": 1.971262104413596e-06,
      "loss": 0.6981424331665039,
      "step": 24780
    },
    {
      "epoch": 2.6197339392578307,
      "grad_norm": 1.5893019437789917,
      "learning_rate": 1.960500945811289e-06,
      "loss": 0.8781041145324707,
      "step": 24790
    },
    {
      "epoch": 2.620790784311135,
      "grad_norm": 1.2281033992767334,
      "learning_rate": 1.9497680415171945e-06,
      "loss": 0.8040219306945801,
      "step": 24800
    },
    {
      "epoch": 2.62184762936444,
      "grad_norm": 1.2383966445922852,
      "learning_rate": 1.9390634046935337e-06,
      "loss": 0.8896523475646972,
      "step": 24810
    },
    {
      "epoch": 2.6229044744177443,
      "grad_norm": 1.0631343126296997,
      "learning_rate": 1.9283870484678447e-06,
      "loss": 0.8266668319702148,
      "step": 24820
    },
    {
      "epoch": 2.623961319471049,
      "grad_norm": 1.3666954040527344,
      "learning_rate": 1.917738985933007e-06,
      "loss": 0.94783935546875,
      "step": 24830
    },
    {
      "epoch": 2.6250181645243535,
      "grad_norm": 1.952837347984314,
      "learning_rate": 1.9071192301471724e-06,
      "loss": 0.7584238052368164,
      "step": 24840
    },
    {
      "epoch": 2.6260750095776584,
      "grad_norm": 2.0475823879241943,
      "learning_rate": 1.8965277941338066e-06,
      "loss": 0.7772571086883545,
      "step": 24850
    },
    {
      "epoch": 2.6271318546309628,
      "grad_norm": 0.9829609990119934,
      "learning_rate": 1.8859646908816254e-06,
      "loss": 0.9099949836730957,
      "step": 24860
    },
    {
      "epoch": 2.6281886996842676,
      "grad_norm": 0.9788087606430054,
      "learning_rate": 1.875429933344619e-06,
      "loss": 0.8030402183532714,
      "step": 24870
    },
    {
      "epoch": 2.6292455447375724,
      "grad_norm": 1.480823278427124,
      "learning_rate": 1.8649235344419896e-06,
      "loss": 0.8793014526367188,
      "step": 24880
    },
    {
      "epoch": 2.630302389790877,
      "grad_norm": 1.5198723077774048,
      "learning_rate": 1.8544455070581863e-06,
      "loss": 0.7999726295471191,
      "step": 24890
    },
    {
      "epoch": 2.631359234844181,
      "grad_norm": 1.3899685144424438,
      "learning_rate": 1.8439958640428468e-06,
      "loss": 0.8004817962646484,
      "step": 24900
    },
    {
      "epoch": 2.632416079897486,
      "grad_norm": 1.2706265449523926,
      "learning_rate": 1.833574618210826e-06,
      "loss": 0.822208309173584,
      "step": 24910
    },
    {
      "epoch": 2.633472924950791,
      "grad_norm": 1.2377679347991943,
      "learning_rate": 1.8231817823421149e-06,
      "loss": 0.9131343841552735,
      "step": 24920
    },
    {
      "epoch": 2.6345297700040953,
      "grad_norm": 1.3104361295700073,
      "learning_rate": 1.8128173691819016e-06,
      "loss": 0.7778268337249756,
      "step": 24930
    },
    {
      "epoch": 2.6355866150573997,
      "grad_norm": 2.4507977962493896,
      "learning_rate": 1.8024813914404942e-06,
      "loss": 0.847867488861084,
      "step": 24940
    },
    {
      "epoch": 2.6366434601107045,
      "grad_norm": 1.328380823135376,
      "learning_rate": 1.7921738617933447e-06,
      "loss": 0.8171755790710449,
      "step": 24950
    },
    {
      "epoch": 2.6377003051640093,
      "grad_norm": 1.917490839958191,
      "learning_rate": 1.7818947928810032e-06,
      "loss": 0.8682936668395996,
      "step": 24960
    },
    {
      "epoch": 2.6387571502173137,
      "grad_norm": 2.447152853012085,
      "learning_rate": 1.7716441973091302e-06,
      "loss": 0.9056599617004395,
      "step": 24970
    },
    {
      "epoch": 2.6398139952706186,
      "grad_norm": 1.197642207145691,
      "learning_rate": 1.7614220876484566e-06,
      "loss": 0.9065261840820312,
      "step": 24980
    },
    {
      "epoch": 2.640870840323923,
      "grad_norm": 0.9973809719085693,
      "learning_rate": 1.7512284764347992e-06,
      "loss": 0.8309863090515137,
      "step": 24990
    },
    {
      "epoch": 2.6419276853772278,
      "grad_norm": 2.2911322116851807,
      "learning_rate": 1.7410633761689971e-06,
      "loss": 0.8593295097351075,
      "step": 25000
    },
    {
      "epoch": 2.6419276853772278,
      "eval_loss": 0.4841359853744507,
      "eval_runtime": 3592.5944,
      "eval_samples_per_second": 2.479,
      "eval_steps_per_second": 2.479,
      "step": 25000
    },
    {
      "epoch": 2.642984530430532,
      "grad_norm": 1.0897414684295654,
      "learning_rate": 1.730926799316951e-06,
      "loss": 0.898221492767334,
      "step": 25010
    },
    {
      "epoch": 2.644041375483837,
      "grad_norm": 1.1797876358032227,
      "learning_rate": 1.720818758309567e-06,
      "loss": 0.8269160270690918,
      "step": 25020
    },
    {
      "epoch": 2.6450982205371414,
      "grad_norm": 0.008216723799705505,
      "learning_rate": 1.7107392655427683e-06,
      "loss": 0.7647016048431396,
      "step": 25030
    },
    {
      "epoch": 2.646155065590446,
      "grad_norm": 0.9912148118019104,
      "learning_rate": 1.700688333377451e-06,
      "loss": 0.8777134895324707,
      "step": 25040
    },
    {
      "epoch": 2.6472119106437506,
      "grad_norm": 1.3423762321472168,
      "learning_rate": 1.6906659741395076e-06,
      "loss": 0.8300362586975097,
      "step": 25050
    },
    {
      "epoch": 2.6482687556970554,
      "grad_norm": 1.3518489599227905,
      "learning_rate": 1.6806722001197734e-06,
      "loss": 0.8546700477600098,
      "step": 25060
    },
    {
      "epoch": 2.64932560075036,
      "grad_norm": 1.2961088418960571,
      "learning_rate": 1.6707070235740363e-06,
      "loss": 0.8832725524902344,
      "step": 25070
    },
    {
      "epoch": 2.6503824458036647,
      "grad_norm": 1.1620392799377441,
      "learning_rate": 1.6607704567230098e-06,
      "loss": 0.8782902717590332,
      "step": 25080
    },
    {
      "epoch": 2.6514392908569695,
      "grad_norm": 1.118697166442871,
      "learning_rate": 1.6508625117523213e-06,
      "loss": 0.7436760425567627,
      "step": 25090
    },
    {
      "epoch": 2.652496135910274,
      "grad_norm": 1.3026891946792603,
      "learning_rate": 1.6409832008125126e-06,
      "loss": 0.8787477493286133,
      "step": 25100
    },
    {
      "epoch": 2.6535529809635783,
      "grad_norm": 1.5950932502746582,
      "learning_rate": 1.6311325360189894e-06,
      "loss": 0.8501700401306153,
      "step": 25110
    },
    {
      "epoch": 2.654609826016883,
      "grad_norm": 1.5678941011428833,
      "learning_rate": 1.6213105294520386e-06,
      "loss": 0.8728120803833008,
      "step": 25120
    },
    {
      "epoch": 2.655666671070188,
      "grad_norm": 2.2090353965759277,
      "learning_rate": 1.6115171931568001e-06,
      "loss": 0.94146728515625,
      "step": 25130
    },
    {
      "epoch": 2.6567235161234923,
      "grad_norm": 1.1723214387893677,
      "learning_rate": 1.6017525391432586e-06,
      "loss": 0.6845496654510498,
      "step": 25140
    },
    {
      "epoch": 2.6577803611767967,
      "grad_norm": 1.4690494537353516,
      "learning_rate": 1.5920165793862158e-06,
      "loss": 0.8264222145080566,
      "step": 25150
    },
    {
      "epoch": 2.6588372062301016,
      "grad_norm": 1.3477823734283447,
      "learning_rate": 1.5823093258252959e-06,
      "loss": 0.788766622543335,
      "step": 25160
    },
    {
      "epoch": 2.6598940512834064,
      "grad_norm": 1.1822670698165894,
      "learning_rate": 1.5726307903649012e-06,
      "loss": 0.8294227600097657,
      "step": 25170
    },
    {
      "epoch": 2.660950896336711,
      "grad_norm": 1.3546762466430664,
      "learning_rate": 1.5629809848742404e-06,
      "loss": 0.7739646434783936,
      "step": 25180
    },
    {
      "epoch": 2.6620077413900156,
      "grad_norm": 2.009603261947632,
      "learning_rate": 1.5533599211872697e-06,
      "loss": 0.91397066116333,
      "step": 25190
    },
    {
      "epoch": 2.66306458644332,
      "grad_norm": 1.3820005655288696,
      "learning_rate": 1.543767611102706e-06,
      "loss": 0.8534990310668945,
      "step": 25200
    },
    {
      "epoch": 2.664121431496625,
      "grad_norm": 1.4135783910751343,
      "learning_rate": 1.5342040663839986e-06,
      "loss": 0.8213818550109864,
      "step": 25210
    },
    {
      "epoch": 2.6651782765499292,
      "grad_norm": 1.2411288022994995,
      "learning_rate": 1.5246692987593348e-06,
      "loss": 0.8296313285827637,
      "step": 25220
    },
    {
      "epoch": 2.666235121603234,
      "grad_norm": 1.2234959602355957,
      "learning_rate": 1.5151633199215976e-06,
      "loss": 0.7509736537933349,
      "step": 25230
    },
    {
      "epoch": 2.6672919666565384,
      "grad_norm": 1.082650899887085,
      "learning_rate": 1.5056861415283674e-06,
      "loss": 1.0148805618286132,
      "step": 25240
    },
    {
      "epoch": 2.6683488117098433,
      "grad_norm": 1.8947820663452148,
      "learning_rate": 1.496237775201903e-06,
      "loss": 0.8151909828186035,
      "step": 25250
    },
    {
      "epoch": 2.6694056567631477,
      "grad_norm": 1.132002830505371,
      "learning_rate": 1.4868182325291442e-06,
      "loss": 0.895200252532959,
      "step": 25260
    },
    {
      "epoch": 2.6704625018164525,
      "grad_norm": 1.756451964378357,
      "learning_rate": 1.4774275250616698e-06,
      "loss": 0.8201100349426269,
      "step": 25270
    },
    {
      "epoch": 2.671519346869757,
      "grad_norm": 1.1977587938308716,
      "learning_rate": 1.4680656643156959e-06,
      "loss": 0.8326722145080566,
      "step": 25280
    },
    {
      "epoch": 2.6725761919230617,
      "grad_norm": 1.3164470195770264,
      "learning_rate": 1.4587326617720664e-06,
      "loss": 0.9184953689575195,
      "step": 25290
    },
    {
      "epoch": 2.6736330369763666,
      "grad_norm": 2.071582078933716,
      "learning_rate": 1.4494285288762427e-06,
      "loss": 0.8763561248779297,
      "step": 25300
    },
    {
      "epoch": 2.674689882029671,
      "grad_norm": 1.4667444229125977,
      "learning_rate": 1.4401532770382698e-06,
      "loss": 0.8820364952087403,
      "step": 25310
    },
    {
      "epoch": 2.6757467270829753,
      "grad_norm": 1.1434218883514404,
      "learning_rate": 1.4309069176327795e-06,
      "loss": 0.8576408386230469,
      "step": 25320
    },
    {
      "epoch": 2.67680357213628,
      "grad_norm": 1.5335693359375,
      "learning_rate": 1.4216894619989679e-06,
      "loss": 0.9435688972473144,
      "step": 25330
    },
    {
      "epoch": 2.677860417189585,
      "grad_norm": 1.9208725690841675,
      "learning_rate": 1.4125009214405987e-06,
      "loss": 0.8178473472595215,
      "step": 25340
    },
    {
      "epoch": 2.6789172622428894,
      "grad_norm": 1.5155963897705078,
      "learning_rate": 1.403341307225961e-06,
      "loss": 0.8865723609924316,
      "step": 25350
    },
    {
      "epoch": 2.679974107296194,
      "grad_norm": 1.8876574039459229,
      "learning_rate": 1.3942106305878777e-06,
      "loss": 0.8492396354675293,
      "step": 25360
    },
    {
      "epoch": 2.6810309523494986,
      "grad_norm": 1.3866446018218994,
      "learning_rate": 1.3851089027236753e-06,
      "loss": 0.7161221981048584,
      "step": 25370
    },
    {
      "epoch": 2.6820877974028035,
      "grad_norm": 1.2901755571365356,
      "learning_rate": 1.3760361347951944e-06,
      "loss": 0.8619924545288086,
      "step": 25380
    },
    {
      "epoch": 2.683144642456108,
      "grad_norm": 1.577934741973877,
      "learning_rate": 1.3669923379287492e-06,
      "loss": 0.8645180702209473,
      "step": 25390
    },
    {
      "epoch": 2.6842014875094127,
      "grad_norm": 1.1323283910751343,
      "learning_rate": 1.3579775232151286e-06,
      "loss": 0.7833078384399415,
      "step": 25400
    },
    {
      "epoch": 2.685258332562717,
      "grad_norm": 1.5796769857406616,
      "learning_rate": 1.3489917017095788e-06,
      "loss": 0.8954771995544434,
      "step": 25410
    },
    {
      "epoch": 2.686315177616022,
      "grad_norm": 2.365306854248047,
      "learning_rate": 1.3400348844317928e-06,
      "loss": 0.8733098983764649,
      "step": 25420
    },
    {
      "epoch": 2.6873720226693263,
      "grad_norm": 1.5984480381011963,
      "learning_rate": 1.3311070823658955e-06,
      "loss": 0.764401626586914,
      "step": 25430
    },
    {
      "epoch": 2.688428867722631,
      "grad_norm": 1.1971217393875122,
      "learning_rate": 1.3222083064604207e-06,
      "loss": 0.9436601638793946,
      "step": 25440
    },
    {
      "epoch": 2.6894857127759355,
      "grad_norm": 1.1460214853286743,
      "learning_rate": 1.3133385676283138e-06,
      "loss": 0.6938650608062744,
      "step": 25450
    },
    {
      "epoch": 2.6905425578292403,
      "grad_norm": 1.364591360092163,
      "learning_rate": 1.304497876746913e-06,
      "loss": 0.8469770431518555,
      "step": 25460
    },
    {
      "epoch": 2.6915994028825447,
      "grad_norm": 1.849131464958191,
      "learning_rate": 1.295686244657926e-06,
      "loss": 0.8292087554931641,
      "step": 25470
    },
    {
      "epoch": 2.6926562479358496,
      "grad_norm": 1.6298426389694214,
      "learning_rate": 1.2869036821674312e-06,
      "loss": 0.8807653427124024,
      "step": 25480
    },
    {
      "epoch": 2.693713092989154,
      "grad_norm": 1.275016188621521,
      "learning_rate": 1.2781502000458496e-06,
      "loss": 0.8281562805175782,
      "step": 25490
    },
    {
      "epoch": 2.694769938042459,
      "grad_norm": 1.4999579191207886,
      "learning_rate": 1.269425809027952e-06,
      "loss": 0.7918920516967773,
      "step": 25500
    },
    {
      "epoch": 2.694769938042459,
      "eval_loss": 0.4836723804473877,
      "eval_runtime": 3581.2336,
      "eval_samples_per_second": 2.487,
      "eval_steps_per_second": 2.487,
      "step": 25500
    },
    {
      "epoch": 2.6958267830957636,
      "grad_norm": 1.5187034606933594,
      "learning_rate": 1.260730519812825e-06,
      "loss": 0.7277170181274414,
      "step": 25510
    },
    {
      "epoch": 2.696883628149068,
      "grad_norm": 1.1514390707015991,
      "learning_rate": 1.2520643430638663e-06,
      "loss": 0.8673351287841797,
      "step": 25520
    },
    {
      "epoch": 2.6979404732023724,
      "grad_norm": 1.3913224935531616,
      "learning_rate": 1.243427289408769e-06,
      "loss": 0.6728174686431885,
      "step": 25530
    },
    {
      "epoch": 2.6989973182556772,
      "grad_norm": 1.4514811038970947,
      "learning_rate": 1.2348193694395272e-06,
      "loss": 0.7472646713256836,
      "step": 25540
    },
    {
      "epoch": 2.700054163308982,
      "grad_norm": 1.4204553365707397,
      "learning_rate": 1.2262405937123856e-06,
      "loss": 0.8817256927490235,
      "step": 25550
    },
    {
      "epoch": 2.7011110083622865,
      "grad_norm": 1.458237886428833,
      "learning_rate": 1.2176909727478647e-06,
      "loss": 0.9650115013122559,
      "step": 25560
    },
    {
      "epoch": 2.702167853415591,
      "grad_norm": 1.2677947282791138,
      "learning_rate": 1.2091705170307167e-06,
      "loss": 0.6903873920440674,
      "step": 25570
    },
    {
      "epoch": 2.7032246984688957,
      "grad_norm": 1.3746811151504517,
      "learning_rate": 1.200679237009944e-06,
      "loss": 0.7687429904937744,
      "step": 25580
    },
    {
      "epoch": 2.7042815435222005,
      "grad_norm": 1.0828635692596436,
      "learning_rate": 1.1922171430987588e-06,
      "loss": 0.797096586227417,
      "step": 25590
    },
    {
      "epoch": 2.705338388575505,
      "grad_norm": 1.3409805297851562,
      "learning_rate": 1.1837842456745822e-06,
      "loss": 0.8471038818359375,
      "step": 25600
    },
    {
      "epoch": 2.7063952336288097,
      "grad_norm": 1.459222435951233,
      "learning_rate": 1.175380555079028e-06,
      "loss": 0.8134431838989258,
      "step": 25610
    },
    {
      "epoch": 2.707452078682114,
      "grad_norm": 1.6120543479919434,
      "learning_rate": 1.1670060816179024e-06,
      "loss": 0.8672909736633301,
      "step": 25620
    },
    {
      "epoch": 2.708508923735419,
      "grad_norm": 1.094730019569397,
      "learning_rate": 1.1586608355611705e-06,
      "loss": 0.9326777458190918,
      "step": 25630
    },
    {
      "epoch": 2.7095657687887234,
      "grad_norm": 1.2813743352890015,
      "learning_rate": 1.150344827142963e-06,
      "loss": 0.8530865669250488,
      "step": 25640
    },
    {
      "epoch": 2.710622613842028,
      "grad_norm": 1.7042064666748047,
      "learning_rate": 1.1420580665615494e-06,
      "loss": 0.8707626342773438,
      "step": 25650
    },
    {
      "epoch": 2.7116794588953326,
      "grad_norm": 1.5477614402770996,
      "learning_rate": 1.1338005639793286e-06,
      "loss": 0.7898700714111329,
      "step": 25660
    },
    {
      "epoch": 2.7127363039486374,
      "grad_norm": 1.2240617275238037,
      "learning_rate": 1.125572329522831e-06,
      "loss": 0.7861843109130859,
      "step": 25670
    },
    {
      "epoch": 2.713793149001942,
      "grad_norm": 1.8002835512161255,
      "learning_rate": 1.1173733732826847e-06,
      "loss": 0.8167343139648438,
      "step": 25680
    },
    {
      "epoch": 2.7148499940552466,
      "grad_norm": 1.3717800378799438,
      "learning_rate": 1.1092037053136162e-06,
      "loss": 0.8364693641662597,
      "step": 25690
    },
    {
      "epoch": 2.715906839108551,
      "grad_norm": 1.4798004627227783,
      "learning_rate": 1.1010633356344252e-06,
      "loss": 0.9404829025268555,
      "step": 25700
    },
    {
      "epoch": 2.716963684161856,
      "grad_norm": 1.3681268692016602,
      "learning_rate": 1.0929522742280042e-06,
      "loss": 0.772843885421753,
      "step": 25710
    },
    {
      "epoch": 2.7180205292151607,
      "grad_norm": 1.4095805883407593,
      "learning_rate": 1.0848705310412794e-06,
      "loss": 0.89883394241333,
      "step": 25720
    },
    {
      "epoch": 2.719077374268465,
      "grad_norm": 1.120995044708252,
      "learning_rate": 1.0768181159852342e-06,
      "loss": 0.9047643661499023,
      "step": 25730
    },
    {
      "epoch": 2.7201342193217695,
      "grad_norm": 1.0260404348373413,
      "learning_rate": 1.0687950389348832e-06,
      "loss": 0.821286678314209,
      "step": 25740
    },
    {
      "epoch": 2.7211910643750743,
      "grad_norm": 1.0351752042770386,
      "learning_rate": 1.060801309729273e-06,
      "loss": 0.8785189628601074,
      "step": 25750
    },
    {
      "epoch": 2.722247909428379,
      "grad_norm": 1.53580641746521,
      "learning_rate": 1.0528369381714393e-06,
      "loss": 0.8252410888671875,
      "step": 25760
    },
    {
      "epoch": 2.7233047544816835,
      "grad_norm": 1.3767670392990112,
      "learning_rate": 1.0449019340284334e-06,
      "loss": 0.9103215217590332,
      "step": 25770
    },
    {
      "epoch": 2.724361599534988,
      "grad_norm": 1.1384886503219604,
      "learning_rate": 1.0369963070312765e-06,
      "loss": 0.897468090057373,
      "step": 25780
    },
    {
      "epoch": 2.7254184445882927,
      "grad_norm": 1.5794233083724976,
      "learning_rate": 1.0291200668749833e-06,
      "loss": 0.8892041206359863,
      "step": 25790
    },
    {
      "epoch": 2.7264752896415976,
      "grad_norm": 1.2753510475158691,
      "learning_rate": 1.0212732232185163e-06,
      "loss": 0.8361375808715821,
      "step": 25800
    },
    {
      "epoch": 2.727532134694902,
      "grad_norm": 1.517862319946289,
      "learning_rate": 1.013455785684783e-06,
      "loss": 0.7951413154602051,
      "step": 25810
    },
    {
      "epoch": 2.728588979748207,
      "grad_norm": 1.265634536743164,
      "learning_rate": 1.0056677638606403e-06,
      "loss": 0.7505497932434082,
      "step": 25820
    },
    {
      "epoch": 2.729645824801511,
      "grad_norm": 1.0019102096557617,
      "learning_rate": 9.979091672968703e-07,
      "loss": 0.8875041961669922,
      "step": 25830
    },
    {
      "epoch": 2.730702669854816,
      "grad_norm": 1.7696545124053955,
      "learning_rate": 9.901800055081677e-07,
      "loss": 0.9681895256042481,
      "step": 25840
    },
    {
      "epoch": 2.7317595149081204,
      "grad_norm": 1.455264925956726,
      "learning_rate": 9.824802879731226e-07,
      "loss": 0.8223307609558106,
      "step": 25850
    },
    {
      "epoch": 2.7328163599614252,
      "grad_norm": 1.1740044355392456,
      "learning_rate": 9.748100241342263e-07,
      "loss": 0.82203369140625,
      "step": 25860
    },
    {
      "epoch": 2.7338732050147296,
      "grad_norm": 1.7094998359680176,
      "learning_rate": 9.671692233978485e-07,
      "loss": 0.9379940032958984,
      "step": 25870
    },
    {
      "epoch": 2.7349300500680345,
      "grad_norm": 1.394016146659851,
      "learning_rate": 9.59557895134222e-07,
      "loss": 0.8494709014892579,
      "step": 25880
    },
    {
      "epoch": 2.735986895121339,
      "grad_norm": 2.176985263824463,
      "learning_rate": 9.519760486774415e-07,
      "loss": 0.8280457496643067,
      "step": 25890
    },
    {
      "epoch": 2.7370437401746437,
      "grad_norm": 1.2098026275634766,
      "learning_rate": 9.444236933254414e-07,
      "loss": 0.7964268684387207,
      "step": 25900
    },
    {
      "epoch": 2.738100585227948,
      "grad_norm": 1.49423348903656,
      "learning_rate": 9.369008383399991e-07,
      "loss": 0.9303435325622559,
      "step": 25910
    },
    {
      "epoch": 2.739157430281253,
      "grad_norm": 1.2591193914413452,
      "learning_rate": 9.294074929467045e-07,
      "loss": 0.844270133972168,
      "step": 25920
    },
    {
      "epoch": 2.7402142753345577,
      "grad_norm": 1.46416437625885,
      "learning_rate": 9.219436663349623e-07,
      "loss": 0.8647645950317383,
      "step": 25930
    },
    {
      "epoch": 2.741271120387862,
      "grad_norm": 1.1967055797576904,
      "learning_rate": 9.145093676579758e-07,
      "loss": 0.7509352207183838,
      "step": 25940
    },
    {
      "epoch": 2.7423279654411665,
      "grad_norm": 1.2810783386230469,
      "learning_rate": 9.071046060327437e-07,
      "loss": 0.8655160903930664,
      "step": 25950
    },
    {
      "epoch": 2.7433848104944714,
      "grad_norm": 1.1912168264389038,
      "learning_rate": 8.997293905400355e-07,
      "loss": 0.7269118785858154,
      "step": 25960
    },
    {
      "epoch": 2.744441655547776,
      "grad_norm": 1.2226120233535767,
      "learning_rate": 8.923837302243859e-07,
      "loss": 0.7469388484954834,
      "step": 25970
    },
    {
      "epoch": 2.7454985006010806,
      "grad_norm": 1.2012122869491577,
      "learning_rate": 8.850676340940889e-07,
      "loss": 0.8466925621032715,
      "step": 25980
    },
    {
      "epoch": 2.746555345654385,
      "grad_norm": 1.5948420763015747,
      "learning_rate": 8.777811111211842e-07,
      "loss": 0.7647356510162353,
      "step": 25990
    },
    {
      "epoch": 2.74761219070769,
      "grad_norm": 1.1651471853256226,
      "learning_rate": 8.705241702414379e-07,
      "loss": 0.7707324504852295,
      "step": 26000
    },
    {
      "epoch": 2.74761219070769,
      "eval_loss": 0.4841079115867615,
      "eval_runtime": 3247.2941,
      "eval_samples_per_second": 2.742,
      "eval_steps_per_second": 2.742,
      "step": 26000
    },
    {
      "epoch": 2.7486690357609946,
      "grad_norm": 1.2559800148010254,
      "learning_rate": 8.632968203543479e-07,
      "loss": 0.8697671890258789,
      "step": 26010
    },
    {
      "epoch": 2.749725880814299,
      "grad_norm": 1.5562371015548706,
      "learning_rate": 8.560990703231076e-07,
      "loss": 0.7838995456695557,
      "step": 26020
    },
    {
      "epoch": 2.750782725867604,
      "grad_norm": 1.4812114238739014,
      "learning_rate": 8.489309289746312e-07,
      "loss": 0.8386811256408692,
      "step": 26030
    },
    {
      "epoch": 2.7518395709209083,
      "grad_norm": 1.5466045141220093,
      "learning_rate": 8.417924050995091e-07,
      "loss": 0.9416836738586426,
      "step": 26040
    },
    {
      "epoch": 2.752896415974213,
      "grad_norm": 1.0009251832962036,
      "learning_rate": 8.34683507452011e-07,
      "loss": 0.803339672088623,
      "step": 26050
    },
    {
      "epoch": 2.7539532610275175,
      "grad_norm": 0.8774063587188721,
      "learning_rate": 8.276042447500798e-07,
      "loss": 0.7977450847625732,
      "step": 26060
    },
    {
      "epoch": 2.7550101060808223,
      "grad_norm": 1.9335358142852783,
      "learning_rate": 8.205546256753155e-07,
      "loss": 0.811455249786377,
      "step": 26070
    },
    {
      "epoch": 2.7560669511341267,
      "grad_norm": 1.1358672380447388,
      "learning_rate": 8.135346588729609e-07,
      "loss": 0.7964441776275635,
      "step": 26080
    },
    {
      "epoch": 2.7571237961874315,
      "grad_norm": 1.051363468170166,
      "learning_rate": 8.065443529518962e-07,
      "loss": 0.7896131992340087,
      "step": 26090
    },
    {
      "epoch": 2.7581806412407364,
      "grad_norm": 1.323849081993103,
      "learning_rate": 7.995837164846254e-07,
      "loss": 0.8578215599060058,
      "step": 26100
    },
    {
      "epoch": 2.7592374862940408,
      "grad_norm": 1.8221696615219116,
      "learning_rate": 7.926527580072784e-07,
      "loss": 0.8593016624450683,
      "step": 26110
    },
    {
      "epoch": 2.760294331347345,
      "grad_norm": 1.1230549812316895,
      "learning_rate": 7.857514860195758e-07,
      "loss": 0.9132707595825196,
      "step": 26120
    },
    {
      "epoch": 2.76135117640065,
      "grad_norm": 1.4039736986160278,
      "learning_rate": 7.78879908984842e-07,
      "loss": 0.8549971580505371,
      "step": 26130
    },
    {
      "epoch": 2.762408021453955,
      "grad_norm": 1.1945620775222778,
      "learning_rate": 7.720380353299723e-07,
      "loss": 0.8757041931152344,
      "step": 26140
    },
    {
      "epoch": 2.763464866507259,
      "grad_norm": 1.540859341621399,
      "learning_rate": 7.652258734454548e-07,
      "loss": 0.8529929161071778,
      "step": 26150
    },
    {
      "epoch": 2.7645217115605636,
      "grad_norm": 1.0105916261672974,
      "learning_rate": 7.584434316853267e-07,
      "loss": 0.8029880523681641,
      "step": 26160
    },
    {
      "epoch": 2.7655785566138684,
      "grad_norm": 1.257565975189209,
      "learning_rate": 7.516907183671845e-07,
      "loss": 0.8466609001159668,
      "step": 26170
    },
    {
      "epoch": 2.7666354016671733,
      "grad_norm": 1.6623660326004028,
      "learning_rate": 7.449677417721595e-07,
      "loss": 0.8461898803710938,
      "step": 26180
    },
    {
      "epoch": 2.7676922467204776,
      "grad_norm": 1.4042961597442627,
      "learning_rate": 7.382745101449206e-07,
      "loss": 0.8191940307617187,
      "step": 26190
    },
    {
      "epoch": 2.7687490917737825,
      "grad_norm": 1.3385872840881348,
      "learning_rate": 7.316110316936686e-07,
      "loss": 0.9188255310058594,
      "step": 26200
    },
    {
      "epoch": 2.769805936827087,
      "grad_norm": 1.4837671518325806,
      "learning_rate": 7.249773145901001e-07,
      "loss": 0.9334070205688476,
      "step": 26210
    },
    {
      "epoch": 2.7708627818803917,
      "grad_norm": 1.5961406230926514,
      "learning_rate": 7.18373366969427e-07,
      "loss": 0.9512948036193848,
      "step": 26220
    },
    {
      "epoch": 2.771919626933696,
      "grad_norm": 1.5439788103103638,
      "learning_rate": 7.117991969303406e-07,
      "loss": 0.9036375999450683,
      "step": 26230
    },
    {
      "epoch": 2.772976471987001,
      "grad_norm": 1.284685730934143,
      "learning_rate": 7.052548125350306e-07,
      "loss": 0.9399537086486817,
      "step": 26240
    },
    {
      "epoch": 2.7740333170403053,
      "grad_norm": 1.124746322631836,
      "learning_rate": 6.987402218091494e-07,
      "loss": 0.8188632965087891,
      "step": 26250
    },
    {
      "epoch": 2.77509016209361,
      "grad_norm": 1.1138720512390137,
      "learning_rate": 6.922554327418118e-07,
      "loss": 0.849545669555664,
      "step": 26260
    },
    {
      "epoch": 2.7761470071469145,
      "grad_norm": 1.3012336492538452,
      "learning_rate": 6.858004532855871e-07,
      "loss": 0.7704813957214356,
      "step": 26270
    },
    {
      "epoch": 2.7772038522002194,
      "grad_norm": 1.4133007526397705,
      "learning_rate": 6.793752913564932e-07,
      "loss": 0.755006217956543,
      "step": 26280
    },
    {
      "epoch": 2.7782606972535238,
      "grad_norm": 1.5382380485534668,
      "learning_rate": 6.729799548339716e-07,
      "loss": 0.9188412666320801,
      "step": 26290
    },
    {
      "epoch": 2.7793175423068286,
      "grad_norm": 1.118983268737793,
      "learning_rate": 6.666144515608957e-07,
      "loss": 0.8384716033935546,
      "step": 26300
    },
    {
      "epoch": 2.7803743873601334,
      "grad_norm": 0.9546668529510498,
      "learning_rate": 6.602787893435464e-07,
      "loss": 0.9510625839233399,
      "step": 26310
    },
    {
      "epoch": 2.781431232413438,
      "grad_norm": 1.231308937072754,
      "learning_rate": 6.53972975951625e-07,
      "loss": 0.8607993125915527,
      "step": 26320
    },
    {
      "epoch": 2.782488077466742,
      "grad_norm": 1.1837126016616821,
      "learning_rate": 6.476970191181986e-07,
      "loss": 0.794313907623291,
      "step": 26330
    },
    {
      "epoch": 2.783544922520047,
      "grad_norm": 1.2471506595611572,
      "learning_rate": 6.414509265397523e-07,
      "loss": 0.7897680759429931,
      "step": 26340
    },
    {
      "epoch": 2.784601767573352,
      "grad_norm": 1.4369264841079712,
      "learning_rate": 6.352347058761227e-07,
      "loss": 0.8809552192687988,
      "step": 26350
    },
    {
      "epoch": 2.7856586126266563,
      "grad_norm": 1.2720290422439575,
      "learning_rate": 6.290483647505313e-07,
      "loss": 0.9152257919311524,
      "step": 26360
    },
    {
      "epoch": 2.7867154576799607,
      "grad_norm": 1.2662700414657593,
      "learning_rate": 6.228919107495429e-07,
      "loss": 0.8154282569885254,
      "step": 26370
    },
    {
      "epoch": 2.7877723027332655,
      "grad_norm": 1.074960708618164,
      "learning_rate": 6.167653514230765e-07,
      "loss": 0.7931228160858155,
      "step": 26380
    },
    {
      "epoch": 2.7888291477865703,
      "grad_norm": 1.234781265258789,
      "learning_rate": 6.106686942843914e-07,
      "loss": 0.8610293388366699,
      "step": 26390
    },
    {
      "epoch": 2.7898859928398747,
      "grad_norm": 1.6662968397140503,
      "learning_rate": 6.046019468100794e-07,
      "loss": 0.836701488494873,
      "step": 26400
    },
    {
      "epoch": 2.7909428378931795,
      "grad_norm": 1.1285278797149658,
      "learning_rate": 5.985651164400391e-07,
      "loss": 0.8368837356567382,
      "step": 26410
    },
    {
      "epoch": 2.791999682946484,
      "grad_norm": 1.3944694995880127,
      "learning_rate": 5.925582105774985e-07,
      "loss": 0.8267829895019532,
      "step": 26420
    },
    {
      "epoch": 2.7930565279997888,
      "grad_norm": 1.6194689273834229,
      "learning_rate": 5.865812365889734e-07,
      "loss": 0.8430244445800781,
      "step": 26430
    },
    {
      "epoch": 2.794113373053093,
      "grad_norm": 1.343076467514038,
      "learning_rate": 5.806342018042837e-07,
      "loss": 0.8900135040283204,
      "step": 26440
    },
    {
      "epoch": 2.795170218106398,
      "grad_norm": 1.3233532905578613,
      "learning_rate": 5.747171135165236e-07,
      "loss": 0.828860855102539,
      "step": 26450
    },
    {
      "epoch": 2.7962270631597024,
      "grad_norm": 1.3770543336868286,
      "learning_rate": 5.688299789820717e-07,
      "loss": 0.8319100379943848,
      "step": 26460
    },
    {
      "epoch": 2.797283908213007,
      "grad_norm": 1.2158243656158447,
      "learning_rate": 5.629728054205641e-07,
      "loss": 0.7684806823730469,
      "step": 26470
    },
    {
      "epoch": 2.7983407532663116,
      "grad_norm": 1.9458143711090088,
      "learning_rate": 5.571456000149106e-07,
      "loss": 0.8570037841796875,
      "step": 26480
    },
    {
      "epoch": 2.7993975983196164,
      "grad_norm": 1.318337082862854,
      "learning_rate": 5.513483699112449e-07,
      "loss": 0.6957658767700196,
      "step": 26490
    },
    {
      "epoch": 2.800454443372921,
      "grad_norm": 1.9897847175598145,
      "learning_rate": 5.455811222189605e-07,
      "loss": 0.8306305885314942,
      "step": 26500
    },
    {
      "epoch": 2.800454443372921,
      "eval_loss": 0.48385825753211975,
      "eval_runtime": 3248.323,
      "eval_samples_per_second": 2.741,
      "eval_steps_per_second": 2.741,
      "step": 26500
    },
    {
      "epoch": 2.8015112884262257,
      "grad_norm": 1.9299246072769165,
      "learning_rate": 5.398438640106779e-07,
      "loss": 0.8144389152526855,
      "step": 26510
    },
    {
      "epoch": 2.8025681334795305,
      "grad_norm": 0.9104908108711243,
      "learning_rate": 5.341366023222382e-07,
      "loss": 0.821961784362793,
      "step": 26520
    },
    {
      "epoch": 2.803624978532835,
      "grad_norm": 1.5640869140625,
      "learning_rate": 5.28459344152693e-07,
      "loss": 0.9149079322814941,
      "step": 26530
    },
    {
      "epoch": 2.8046818235861393,
      "grad_norm": 1.244314193725586,
      "learning_rate": 5.228120964643091e-07,
      "loss": 0.9188257217407226,
      "step": 26540
    },
    {
      "epoch": 2.805738668639444,
      "grad_norm": 1.081162929534912,
      "learning_rate": 5.171948661825383e-07,
      "loss": 0.9935465812683105,
      "step": 26550
    },
    {
      "epoch": 2.806795513692749,
      "grad_norm": 1.2323633432388306,
      "learning_rate": 5.116076601960312e-07,
      "loss": 0.6649397373199463,
      "step": 26560
    },
    {
      "epoch": 2.8078523587460533,
      "grad_norm": 1.327189326286316,
      "learning_rate": 5.0605048535661e-07,
      "loss": 0.9716681480407715,
      "step": 26570
    },
    {
      "epoch": 2.8089092037993577,
      "grad_norm": 2.3878941535949707,
      "learning_rate": 5.005233484792732e-07,
      "loss": 0.8050069808959961,
      "step": 26580
    },
    {
      "epoch": 2.8099660488526625,
      "grad_norm": 1.3986676931381226,
      "learning_rate": 4.950262563421848e-07,
      "loss": 0.8436168670654297,
      "step": 26590
    },
    {
      "epoch": 2.8110228939059674,
      "grad_norm": 1.4687979221343994,
      "learning_rate": 4.89559215686658e-07,
      "loss": 0.9139415740966796,
      "step": 26600
    },
    {
      "epoch": 2.8120797389592718,
      "grad_norm": 1.5567797422409058,
      "learning_rate": 4.841222332171547e-07,
      "loss": 0.8087290763854981,
      "step": 26610
    },
    {
      "epoch": 2.8131365840125766,
      "grad_norm": 1.3808571100234985,
      "learning_rate": 4.787153156012775e-07,
      "loss": 0.8186630249023438,
      "step": 26620
    },
    {
      "epoch": 2.814193429065881,
      "grad_norm": 1.1831523180007935,
      "learning_rate": 4.733384694697529e-07,
      "loss": 0.8694831848144531,
      "step": 26630
    },
    {
      "epoch": 2.815250274119186,
      "grad_norm": 1.1402710676193237,
      "learning_rate": 4.6799170141643975e-07,
      "loss": 0.9373096466064453,
      "step": 26640
    },
    {
      "epoch": 2.81630711917249,
      "grad_norm": 1.1445986032485962,
      "learning_rate": 4.6267501799830137e-07,
      "loss": 0.7676166534423828,
      "step": 26650
    },
    {
      "epoch": 2.817363964225795,
      "grad_norm": 1.0112861394882202,
      "learning_rate": 4.5738842573541105e-07,
      "loss": 0.8648663520812988,
      "step": 26660
    },
    {
      "epoch": 2.8184208092790994,
      "grad_norm": 1.0685625076293945,
      "learning_rate": 4.5213193111093567e-07,
      "loss": 0.8525484085083008,
      "step": 26670
    },
    {
      "epoch": 2.8194776543324043,
      "grad_norm": 1.2633132934570312,
      "learning_rate": 4.469055405711464e-07,
      "loss": 0.8944845199584961,
      "step": 26680
    },
    {
      "epoch": 2.8205344993857087,
      "grad_norm": 1.3676868677139282,
      "learning_rate": 4.4170926052537755e-07,
      "loss": 0.8062040328979492,
      "step": 26690
    },
    {
      "epoch": 2.8215913444390135,
      "grad_norm": 1.1630644798278809,
      "learning_rate": 4.3654309734604824e-07,
      "loss": 0.692777442932129,
      "step": 26700
    },
    {
      "epoch": 2.822648189492318,
      "grad_norm": 1.457903265953064,
      "learning_rate": 4.3140705736864074e-07,
      "loss": 0.8080066680908203,
      "step": 26710
    },
    {
      "epoch": 2.8237050345456227,
      "grad_norm": 1.1904067993164062,
      "learning_rate": 4.2630114689170274e-07,
      "loss": 0.7961954593658447,
      "step": 26720
    },
    {
      "epoch": 2.8247618795989275,
      "grad_norm": 1.1864668130874634,
      "learning_rate": 4.2122537217682e-07,
      "loss": 0.9311445236206055,
      "step": 26730
    },
    {
      "epoch": 2.825818724652232,
      "grad_norm": 0.8986813426017761,
      "learning_rate": 4.1617973944863277e-07,
      "loss": 0.9414053916931152,
      "step": 26740
    },
    {
      "epoch": 2.8268755697055363,
      "grad_norm": 0.9542723894119263,
      "learning_rate": 4.111642548948108e-07,
      "loss": 0.8287828445434571,
      "step": 26750
    },
    {
      "epoch": 2.827932414758841,
      "grad_norm": 1.1576122045516968,
      "learning_rate": 4.0617892466605913e-07,
      "loss": 0.8872566223144531,
      "step": 26760
    },
    {
      "epoch": 2.828989259812146,
      "grad_norm": 1.0072605609893799,
      "learning_rate": 4.0122375487609274e-07,
      "loss": 0.8073710441589356,
      "step": 26770
    },
    {
      "epoch": 2.8300461048654504,
      "grad_norm": 1.126381516456604,
      "learning_rate": 3.962987516016453e-07,
      "loss": 0.8986640930175781,
      "step": 26780
    },
    {
      "epoch": 2.8311029499187548,
      "grad_norm": 1.7424330711364746,
      "learning_rate": 3.914039208824605e-07,
      "loss": 0.8590470314025879,
      "step": 26790
    },
    {
      "epoch": 2.8321597949720596,
      "grad_norm": 1.4568023681640625,
      "learning_rate": 3.8653926872127265e-07,
      "loss": 0.9155697822570801,
      "step": 26800
    },
    {
      "epoch": 2.8332166400253644,
      "grad_norm": 0.9550079703330994,
      "learning_rate": 3.817048010838098e-07,
      "loss": 0.8732338905334472,
      "step": 26810
    },
    {
      "epoch": 2.834273485078669,
      "grad_norm": 1.2651375532150269,
      "learning_rate": 3.769005238987794e-07,
      "loss": 0.8492950439453125,
      "step": 26820
    },
    {
      "epoch": 2.8353303301319737,
      "grad_norm": 0.9784697890281677,
      "learning_rate": 3.721264430578769e-07,
      "loss": 0.8389402389526367,
      "step": 26830
    },
    {
      "epoch": 2.836387175185278,
      "grad_norm": 1.4436160326004028,
      "learning_rate": 3.673825644157525e-07,
      "loss": 0.9255143165588379,
      "step": 26840
    },
    {
      "epoch": 2.837444020238583,
      "grad_norm": 1.7619984149932861,
      "learning_rate": 3.626688937900247e-07,
      "loss": 0.8366087913513184,
      "step": 26850
    },
    {
      "epoch": 2.8385008652918873,
      "grad_norm": 1.0855940580368042,
      "learning_rate": 3.5798543696126953e-07,
      "loss": 0.9173385620117187,
      "step": 26860
    },
    {
      "epoch": 2.839557710345192,
      "grad_norm": 1.3698971271514893,
      "learning_rate": 3.5333219967300657e-07,
      "loss": 0.8469768524169922,
      "step": 26870
    },
    {
      "epoch": 2.8406145553984965,
      "grad_norm": 0.9370644092559814,
      "learning_rate": 3.487091876316961e-07,
      "loss": 0.8652812004089355,
      "step": 26880
    },
    {
      "epoch": 2.8416714004518013,
      "grad_norm": 1.0339382886886597,
      "learning_rate": 3.441164065067309e-07,
      "loss": 0.8822210311889649,
      "step": 26890
    },
    {
      "epoch": 2.8427282455051057,
      "grad_norm": 1.6216886043548584,
      "learning_rate": 3.3955386193043337e-07,
      "loss": 0.9235361099243165,
      "step": 26900
    },
    {
      "epoch": 2.8437850905584106,
      "grad_norm": 1.6126363277435303,
      "learning_rate": 3.3502155949804434e-07,
      "loss": 0.8712035179138183,
      "step": 26910
    },
    {
      "epoch": 2.844841935611715,
      "grad_norm": 1.4139511585235596,
      "learning_rate": 3.3051950476771787e-07,
      "loss": 0.9199986457824707,
      "step": 26920
    },
    {
      "epoch": 2.84589878066502,
      "grad_norm": 1.9093321561813354,
      "learning_rate": 3.2604770326051246e-07,
      "loss": 0.8430895805358887,
      "step": 26930
    },
    {
      "epoch": 2.8469556257183246,
      "grad_norm": 1.7255467176437378,
      "learning_rate": 3.216061604603832e-07,
      "loss": 0.8337729454040528,
      "step": 26940
    },
    {
      "epoch": 2.848012470771629,
      "grad_norm": 1.0879219770431519,
      "learning_rate": 3.1719488181418953e-07,
      "loss": 0.8216140747070313,
      "step": 26950
    },
    {
      "epoch": 2.8490693158249334,
      "grad_norm": 1.7883909940719604,
      "learning_rate": 3.1281387273165983e-07,
      "loss": 0.8646152496337891,
      "step": 26960
    },
    {
      "epoch": 2.8501261608782382,
      "grad_norm": 1.4252965450286865,
      "learning_rate": 3.0846313858541586e-07,
      "loss": 0.8458906173706054,
      "step": 26970
    },
    {
      "epoch": 2.851183005931543,
      "grad_norm": 1.3380956649780273,
      "learning_rate": 3.041426847109424e-07,
      "loss": 0.9189237594604492,
      "step": 26980
    },
    {
      "epoch": 2.8522398509848474,
      "grad_norm": 1.1667988300323486,
      "learning_rate": 2.998525164065985e-07,
      "loss": 0.7479388236999511,
      "step": 26990
    },
    {
      "epoch": 2.853296696038152,
      "grad_norm": 1.483039140701294,
      "learning_rate": 2.9559263893359225e-07,
      "loss": 0.7968756198883057,
      "step": 27000
    },
    {
      "epoch": 2.853296696038152,
      "eval_loss": 0.4837360978126526,
      "eval_runtime": 3235.4917,
      "eval_samples_per_second": 2.752,
      "eval_steps_per_second": 2.752,
      "step": 27000
    },
    {
      "epoch": 2.8543535410914567,
      "grad_norm": 1.3329942226409912,
      "learning_rate": 2.913630575159948e-07,
      "loss": 0.8142178535461426,
      "step": 27010
    },
    {
      "epoch": 2.8554103861447615,
      "grad_norm": 1.702086091041565,
      "learning_rate": 2.871637773407182e-07,
      "loss": 0.9280745506286621,
      "step": 27020
    },
    {
      "epoch": 2.856467231198066,
      "grad_norm": 1.8790521621704102,
      "learning_rate": 2.8299480355751807e-07,
      "loss": 0.7913022041320801,
      "step": 27030
    },
    {
      "epoch": 2.8575240762513707,
      "grad_norm": 1.5209611654281616,
      "learning_rate": 2.788561412789853e-07,
      "loss": 0.8468524932861328,
      "step": 27040
    },
    {
      "epoch": 2.858580921304675,
      "grad_norm": 1.3454110622406006,
      "learning_rate": 2.747477955805294e-07,
      "loss": 0.8319315910339355,
      "step": 27050
    },
    {
      "epoch": 2.85963776635798,
      "grad_norm": 1.8111493587493896,
      "learning_rate": 2.7066977150038973e-07,
      "loss": 0.8330341339111328,
      "step": 27060
    },
    {
      "epoch": 2.8606946114112843,
      "grad_norm": 2.1242687702178955,
      "learning_rate": 2.6662207403962134e-07,
      "loss": 0.835017204284668,
      "step": 27070
    },
    {
      "epoch": 2.861751456464589,
      "grad_norm": 1.2925316095352173,
      "learning_rate": 2.626047081620869e-07,
      "loss": 0.8465988159179687,
      "step": 27080
    },
    {
      "epoch": 2.8628083015178936,
      "grad_norm": 1.213625192642212,
      "learning_rate": 2.586176787944483e-07,
      "loss": 0.6700841903686523,
      "step": 27090
    },
    {
      "epoch": 2.8638651465711984,
      "grad_norm": 1.2060260772705078,
      "learning_rate": 2.546609908261666e-07,
      "loss": 0.7241594314575195,
      "step": 27100
    },
    {
      "epoch": 2.8649219916245032,
      "grad_norm": 1.536970853805542,
      "learning_rate": 2.507346491094992e-07,
      "loss": 0.7766438484191894,
      "step": 27110
    },
    {
      "epoch": 2.8659788366778076,
      "grad_norm": 1.3087584972381592,
      "learning_rate": 2.468386584594834e-07,
      "loss": 0.8395292282104492,
      "step": 27120
    },
    {
      "epoch": 2.867035681731112,
      "grad_norm": 1.3019362688064575,
      "learning_rate": 2.429730236539335e-07,
      "loss": 0.859778881072998,
      "step": 27130
    },
    {
      "epoch": 2.868092526784417,
      "grad_norm": 1.2664761543273926,
      "learning_rate": 2.391377494334407e-07,
      "loss": 0.8038371086120606,
      "step": 27140
    },
    {
      "epoch": 2.8691493718377217,
      "grad_norm": 1.9160573482513428,
      "learning_rate": 2.3533284050136773e-07,
      "loss": 0.8620212554931641,
      "step": 27150
    },
    {
      "epoch": 2.870206216891026,
      "grad_norm": 1.4003230333328247,
      "learning_rate": 2.3155830152383206e-07,
      "loss": 0.7990819454193115,
      "step": 27160
    },
    {
      "epoch": 2.8712630619443305,
      "grad_norm": 1.1039137840270996,
      "learning_rate": 2.2781413712970877e-07,
      "loss": 0.8349677085876465,
      "step": 27170
    },
    {
      "epoch": 2.8723199069976353,
      "grad_norm": 1.06977379322052,
      "learning_rate": 2.2410035191062496e-07,
      "loss": 0.6936860084533691,
      "step": 27180
    },
    {
      "epoch": 2.87337675205094,
      "grad_norm": 1.4056994915008545,
      "learning_rate": 2.204169504209541e-07,
      "loss": 0.9277229309082031,
      "step": 27190
    },
    {
      "epoch": 2.8744335971042445,
      "grad_norm": 1.009191632270813,
      "learning_rate": 2.167639371778024e-07,
      "loss": 0.8260046005249023,
      "step": 27200
    },
    {
      "epoch": 2.8754904421575493,
      "grad_norm": 1.1400233507156372,
      "learning_rate": 2.1314131666101967e-07,
      "loss": 0.7836112976074219,
      "step": 27210
    },
    {
      "epoch": 2.8765472872108537,
      "grad_norm": 1.4883966445922852,
      "learning_rate": 2.0954909331316885e-07,
      "loss": 0.866888427734375,
      "step": 27220
    },
    {
      "epoch": 2.8776041322641586,
      "grad_norm": 1.1166480779647827,
      "learning_rate": 2.0598727153955388e-07,
      "loss": 0.9715520858764648,
      "step": 27230
    },
    {
      "epoch": 2.878660977317463,
      "grad_norm": 1.5349382162094116,
      "learning_rate": 2.024558557081835e-07,
      "loss": 0.8285865783691406,
      "step": 27240
    },
    {
      "epoch": 2.879717822370768,
      "grad_norm": 1.2943569421768188,
      "learning_rate": 1.9895485014978233e-07,
      "loss": 0.8026994705200196,
      "step": 27250
    },
    {
      "epoch": 2.880774667424072,
      "grad_norm": 1.3469159603118896,
      "learning_rate": 1.9548425915777992e-07,
      "loss": 0.8699614524841308,
      "step": 27260
    },
    {
      "epoch": 2.881831512477377,
      "grad_norm": 1.4328128099441528,
      "learning_rate": 1.920440869883078e-07,
      "loss": 0.8536157608032227,
      "step": 27270
    },
    {
      "epoch": 2.8828883575306814,
      "grad_norm": 1.2541406154632568,
      "learning_rate": 1.88634337860194e-07,
      "loss": 0.8342042922973633,
      "step": 27280
    },
    {
      "epoch": 2.8839452025839862,
      "grad_norm": 2.249098777770996,
      "learning_rate": 1.8525501595496043e-07,
      "loss": 0.845174503326416,
      "step": 27290
    },
    {
      "epoch": 2.8850020476372906,
      "grad_norm": 1.5102159976959229,
      "learning_rate": 1.8190612541680863e-07,
      "loss": 0.8163376808166504,
      "step": 27300
    },
    {
      "epoch": 2.8860588926905955,
      "grad_norm": 1.7284940481185913,
      "learning_rate": 1.785876703526257e-07,
      "loss": 0.7949755191802979,
      "step": 27310
    },
    {
      "epoch": 2.8871157377439003,
      "grad_norm": 1.8445193767547607,
      "learning_rate": 1.752996548319702e-07,
      "loss": 0.7026355743408204,
      "step": 27320
    },
    {
      "epoch": 2.8881725827972047,
      "grad_norm": 1.1528066396713257,
      "learning_rate": 1.7204208288708057e-07,
      "loss": 0.8528539657592773,
      "step": 27330
    },
    {
      "epoch": 2.889229427850509,
      "grad_norm": 0.02888985350728035,
      "learning_rate": 1.6881495851284735e-07,
      "loss": 0.74155592918396,
      "step": 27340
    },
    {
      "epoch": 2.890286272903814,
      "grad_norm": 1.249377965927124,
      "learning_rate": 1.6561828566683259e-07,
      "loss": 0.7978826522827148,
      "step": 27350
    },
    {
      "epoch": 2.8913431179571187,
      "grad_norm": 1.1799516677856445,
      "learning_rate": 1.6245206826925043e-07,
      "loss": 0.8972041130065918,
      "step": 27360
    },
    {
      "epoch": 2.892399963010423,
      "grad_norm": 1.556994080543518,
      "learning_rate": 1.5931631020296435e-07,
      "loss": 0.7983786582946777,
      "step": 27370
    },
    {
      "epoch": 2.8934568080637275,
      "grad_norm": 1.4587657451629639,
      "learning_rate": 1.562110153134927e-07,
      "loss": 0.8749009132385254,
      "step": 27380
    },
    {
      "epoch": 2.8945136531170323,
      "grad_norm": 1.393249273300171,
      "learning_rate": 1.5313618740898096e-07,
      "loss": 0.7992466926574707,
      "step": 27390
    },
    {
      "epoch": 2.895570498170337,
      "grad_norm": 1.7130308151245117,
      "learning_rate": 1.5009183026022666e-07,
      "loss": 0.908242130279541,
      "step": 27400
    },
    {
      "epoch": 2.8966273432236416,
      "grad_norm": 1.693440318107605,
      "learning_rate": 1.4707794760065174e-07,
      "loss": 0.8443113327026367,
      "step": 27410
    },
    {
      "epoch": 2.8976841882769464,
      "grad_norm": 1.2884533405303955,
      "learning_rate": 1.44094543126308e-07,
      "loss": 0.8517096519470215,
      "step": 27420
    },
    {
      "epoch": 2.898741033330251,
      "grad_norm": 1.479288935661316,
      "learning_rate": 1.4114162049586598e-07,
      "loss": 0.8652219772338867,
      "step": 27430
    },
    {
      "epoch": 2.8997978783835556,
      "grad_norm": 1.2793476581573486,
      "learning_rate": 1.3821918333062346e-07,
      "loss": 0.7785836219787597,
      "step": 27440
    },
    {
      "epoch": 2.90085472343686,
      "grad_norm": 1.384728193283081,
      "learning_rate": 1.3532723521448575e-07,
      "loss": 0.7392659664154053,
      "step": 27450
    },
    {
      "epoch": 2.901911568490165,
      "grad_norm": 1.2038296461105347,
      "learning_rate": 1.3246577969397156e-07,
      "loss": 0.8446949005126954,
      "step": 27460
    },
    {
      "epoch": 2.9029684135434692,
      "grad_norm": 1.13865065574646,
      "learning_rate": 1.296348202782044e-07,
      "loss": 0.8144206047058106,
      "step": 27470
    },
    {
      "epoch": 2.904025258596774,
      "grad_norm": 1.2718274593353271,
      "learning_rate": 1.2683436043891272e-07,
      "loss": 0.9359066963195801,
      "step": 27480
    },
    {
      "epoch": 2.9050821036500785,
      "grad_norm": 1.6730624437332153,
      "learning_rate": 1.240644036104105e-07,
      "loss": 0.8088506698608399,
      "step": 27490
    },
    {
      "epoch": 2.9061389487033833,
      "grad_norm": 1.367834210395813,
      "learning_rate": 1.2132495318961933e-07,
      "loss": 0.9231465339660645,
      "step": 27500
    },
    {
      "epoch": 2.9061389487033833,
      "eval_loss": 0.4836849272251129,
      "eval_runtime": 3260.7118,
      "eval_samples_per_second": 2.731,
      "eval_steps_per_second": 2.731,
      "step": 27500
    },
    {
      "epoch": 2.9071957937566877,
      "grad_norm": 1.1255334615707397,
      "learning_rate": 1.1861601253603805e-07,
      "loss": 0.8801615715026856,
      "step": 27510
    },
    {
      "epoch": 2.9082526388099925,
      "grad_norm": 1.3776588439941406,
      "learning_rate": 1.1593758497175645e-07,
      "loss": 0.9475391387939454,
      "step": 27520
    },
    {
      "epoch": 2.9093094838632974,
      "grad_norm": 1.2158108949661255,
      "learning_rate": 1.1328967378144428e-07,
      "loss": 0.8257597923278809,
      "step": 27530
    },
    {
      "epoch": 2.9103663289166017,
      "grad_norm": 1.6508708000183105,
      "learning_rate": 1.1067228221234571e-07,
      "loss": 0.8203993797302246,
      "step": 27540
    },
    {
      "epoch": 2.911423173969906,
      "grad_norm": 1.1121104955673218,
      "learning_rate": 1.0808541347427924e-07,
      "loss": 0.9220774650573731,
      "step": 27550
    },
    {
      "epoch": 2.912480019023211,
      "grad_norm": 1.137789011001587,
      "learning_rate": 1.055290707396267e-07,
      "loss": 0.9362641334533691,
      "step": 27560
    },
    {
      "epoch": 2.913536864076516,
      "grad_norm": 1.502265453338623,
      "learning_rate": 1.0300325714334702e-07,
      "loss": 0.755186653137207,
      "step": 27570
    },
    {
      "epoch": 2.91459370912982,
      "grad_norm": 1.4068892002105713,
      "learning_rate": 1.0050797578294302e-07,
      "loss": 0.7925820827484131,
      "step": 27580
    },
    {
      "epoch": 2.9156505541831246,
      "grad_norm": 1.291330337524414,
      "learning_rate": 9.804322971848634e-08,
      "loss": 0.9629791259765625,
      "step": 27590
    },
    {
      "epoch": 2.9167073992364294,
      "grad_norm": 2.191682815551758,
      "learning_rate": 9.560902197260357e-08,
      "loss": 0.8413119316101074,
      "step": 27600
    },
    {
      "epoch": 2.9177642442897342,
      "grad_norm": 2.0278501510620117,
      "learning_rate": 9.320535553045962e-08,
      "loss": 0.873691177368164,
      "step": 27610
    },
    {
      "epoch": 2.9188210893430386,
      "grad_norm": 1.5349745750427246,
      "learning_rate": 9.083223333977709e-08,
      "loss": 0.868031120300293,
      "step": 27620
    },
    {
      "epoch": 2.9198779343963435,
      "grad_norm": 1.0735278129577637,
      "learning_rate": 8.848965831081135e-08,
      "loss": 0.7167880535125732,
      "step": 27630
    },
    {
      "epoch": 2.920934779449648,
      "grad_norm": 1.2061855792999268,
      "learning_rate": 8.617763331636164e-08,
      "loss": 0.8321484565734864,
      "step": 27640
    },
    {
      "epoch": 2.9219916245029527,
      "grad_norm": 1.3995698690414429,
      "learning_rate": 8.389616119176547e-08,
      "loss": 0.9530677795410156,
      "step": 27650
    },
    {
      "epoch": 2.923048469556257,
      "grad_norm": 1.59152090549469,
      "learning_rate": 8.164524473488478e-08,
      "loss": 0.7641724109649658,
      "step": 27660
    },
    {
      "epoch": 2.924105314609562,
      "grad_norm": 1.7498530149459839,
      "learning_rate": 7.942488670611148e-08,
      "loss": 0.8666961669921875,
      "step": 27670
    },
    {
      "epoch": 2.9251621596628663,
      "grad_norm": 1.3722951412200928,
      "learning_rate": 7.723508982837024e-08,
      "loss": 0.8071813583374023,
      "step": 27680
    },
    {
      "epoch": 2.926219004716171,
      "grad_norm": 1.2983567714691162,
      "learning_rate": 7.507585678709627e-08,
      "loss": 0.7690010070800781,
      "step": 27690
    },
    {
      "epoch": 2.9272758497694755,
      "grad_norm": 1.382222056388855,
      "learning_rate": 7.29471902302492e-08,
      "loss": 0.7965548515319825,
      "step": 27700
    },
    {
      "epoch": 2.9283326948227804,
      "grad_norm": 1.0959749221801758,
      "learning_rate": 7.084909276830198e-08,
      "loss": 0.8796310424804688,
      "step": 27710
    },
    {
      "epoch": 2.9293895398760847,
      "grad_norm": 1.099875807762146,
      "learning_rate": 6.878156697424087e-08,
      "loss": 0.755972146987915,
      "step": 27720
    },
    {
      "epoch": 2.9304463849293896,
      "grad_norm": 1.2165768146514893,
      "learning_rate": 6.674461538355992e-08,
      "loss": 0.7147684097290039,
      "step": 27730
    },
    {
      "epoch": 2.9315032299826944,
      "grad_norm": 1.5964301824569702,
      "learning_rate": 6.473824049425814e-08,
      "loss": 0.7858142852783203,
      "step": 27740
    },
    {
      "epoch": 2.932560075035999,
      "grad_norm": 1.1737765073776245,
      "learning_rate": 6.276244476683679e-08,
      "loss": 0.7688723564147949,
      "step": 27750
    },
    {
      "epoch": 2.933616920089303,
      "grad_norm": 1.6675068140029907,
      "learning_rate": 6.081723062429934e-08,
      "loss": 0.8898821830749511,
      "step": 27760
    },
    {
      "epoch": 2.934673765142608,
      "grad_norm": 1.2582188844680786,
      "learning_rate": 5.890260045214313e-08,
      "loss": 0.8989505767822266,
      "step": 27770
    },
    {
      "epoch": 2.935730610195913,
      "grad_norm": 1.2190916538238525,
      "learning_rate": 5.701855659835942e-08,
      "loss": 0.8934839248657227,
      "step": 27780
    },
    {
      "epoch": 2.9367874552492172,
      "grad_norm": 1.023996114730835,
      "learning_rate": 5.516510137342778e-08,
      "loss": 0.8898865699768066,
      "step": 27790
    },
    {
      "epoch": 2.9378443003025216,
      "grad_norm": 1.2424720525741577,
      "learning_rate": 5.3342237050324486e-08,
      "loss": 0.8652112007141113,
      "step": 27800
    },
    {
      "epoch": 2.9389011453558265,
      "grad_norm": 1.8073620796203613,
      "learning_rate": 5.154996586450023e-08,
      "loss": 0.8635285377502442,
      "step": 27810
    },
    {
      "epoch": 2.9399579904091313,
      "grad_norm": 0.9342585206031799,
      "learning_rate": 4.978829001389684e-08,
      "loss": 0.8965847969055176,
      "step": 27820
    },
    {
      "epoch": 2.9410148354624357,
      "grad_norm": 1.3139944076538086,
      "learning_rate": 4.805721165892507e-08,
      "loss": 0.9550253868103027,
      "step": 27830
    },
    {
      "epoch": 2.9420716805157405,
      "grad_norm": 1.133662462234497,
      "learning_rate": 4.635673292248399e-08,
      "loss": 0.8050590515136719,
      "step": 27840
    },
    {
      "epoch": 2.943128525569045,
      "grad_norm": 1.0443992614746094,
      "learning_rate": 4.468685588994159e-08,
      "loss": 0.8798318862915039,
      "step": 27850
    },
    {
      "epoch": 2.9441853706223498,
      "grad_norm": 1.761603832244873,
      "learning_rate": 4.3047582609140325e-08,
      "loss": 0.8785643577575684,
      "step": 27860
    },
    {
      "epoch": 2.945242215675654,
      "grad_norm": 1.4768801927566528,
      "learning_rate": 4.1438915090386e-08,
      "loss": 0.9248263359069824,
      "step": 27870
    },
    {
      "epoch": 2.946299060728959,
      "grad_norm": 1.2776798009872437,
      "learning_rate": 3.9860855306456115e-08,
      "loss": 0.8557572364807129,
      "step": 27880
    },
    {
      "epoch": 2.9473559057822634,
      "grad_norm": 0.98610520362854,
      "learning_rate": 3.8313405192594295e-08,
      "loss": 0.9024643898010254,
      "step": 27890
    },
    {
      "epoch": 2.948412750835568,
      "grad_norm": 1.4780617952346802,
      "learning_rate": 3.6796566646501996e-08,
      "loss": 0.8767160415649414,
      "step": 27900
    },
    {
      "epoch": 2.9494695958888726,
      "grad_norm": 1.3243608474731445,
      "learning_rate": 3.5310341528344003e-08,
      "loss": 0.8812615394592285,
      "step": 27910
    },
    {
      "epoch": 2.9505264409421774,
      "grad_norm": 1.6667804718017578,
      "learning_rate": 3.385473166073738e-08,
      "loss": 0.8832919120788574,
      "step": 27920
    },
    {
      "epoch": 2.951583285995482,
      "grad_norm": 1.35738205909729,
      "learning_rate": 3.242973882876255e-08,
      "loss": 0.9232030868530273,
      "step": 27930
    },
    {
      "epoch": 2.9526401310487866,
      "grad_norm": 1.2711490392684937,
      "learning_rate": 3.1035364779943845e-08,
      "loss": 0.843757438659668,
      "step": 27940
    },
    {
      "epoch": 2.9536969761020915,
      "grad_norm": 1.3193193674087524,
      "learning_rate": 2.967161122426343e-08,
      "loss": 0.8704364776611329,
      "step": 27950
    },
    {
      "epoch": 2.954753821155396,
      "grad_norm": 1.2254630327224731,
      "learning_rate": 2.833847983414739e-08,
      "loss": 0.7721566677093505,
      "step": 27960
    },
    {
      "epoch": 2.9558106662087003,
      "grad_norm": 0.9613962173461914,
      "learning_rate": 2.7035972244474073e-08,
      "loss": 0.7524127006530762,
      "step": 27970
    },
    {
      "epoch": 2.956867511262005,
      "grad_norm": 1.301660418510437,
      "learning_rate": 2.5764090052560198e-08,
      "loss": 0.9322748184204102,
      "step": 27980
    },
    {
      "epoch": 2.95792435631531,
      "grad_norm": 1.4124857187271118,
      "learning_rate": 2.4522834818171968e-08,
      "loss": 0.8287073135375976,
      "step": 27990
    },
    {
      "epoch": 2.9589812013686143,
      "grad_norm": 1.1561635732650757,
      "learning_rate": 2.33122080635112e-08,
      "loss": 0.9108712196350097,
      "step": 28000
    },
    {
      "epoch": 2.9589812013686143,
      "eval_loss": 0.483615905046463,
      "eval_runtime": 3237.6456,
      "eval_samples_per_second": 2.75,
      "eval_steps_per_second": 2.75,
      "step": 28000
    },
    {
      "epoch": 2.9600380464219187,
      "grad_norm": 1.5094038248062134,
      "learning_rate": 2.2132211273220847e-08,
      "loss": 0.8209236145019532,
      "step": 28010
    },
    {
      "epoch": 2.9610948914752235,
      "grad_norm": 1.6858611106872559,
      "learning_rate": 2.0982845894376713e-08,
      "loss": 0.8495326995849609,
      "step": 28020
    },
    {
      "epoch": 2.9621517365285284,
      "grad_norm": 1.41942298412323,
      "learning_rate": 1.986411333650129e-08,
      "loss": 0.8880720138549805,
      "step": 28030
    },
    {
      "epoch": 2.9632085815818328,
      "grad_norm": 1.4312872886657715,
      "learning_rate": 1.8776014971538803e-08,
      "loss": 0.9381152153015136,
      "step": 28040
    },
    {
      "epoch": 2.9642654266351376,
      "grad_norm": 1.4956817626953125,
      "learning_rate": 1.771855213386908e-08,
      "loss": 0.8022502899169922,
      "step": 28050
    },
    {
      "epoch": 2.965322271688442,
      "grad_norm": 1.7753782272338867,
      "learning_rate": 1.669172612030756e-08,
      "loss": 0.7477090358734131,
      "step": 28060
    },
    {
      "epoch": 2.966379116741747,
      "grad_norm": 1.0993744134902954,
      "learning_rate": 1.5695538190091397e-08,
      "loss": 0.8584689140319824,
      "step": 28070
    },
    {
      "epoch": 2.967435961795051,
      "grad_norm": 2.141275644302368,
      "learning_rate": 1.4729989564887803e-08,
      "loss": 0.8031422615051269,
      "step": 28080
    },
    {
      "epoch": 2.968492806848356,
      "grad_norm": 1.3441087007522583,
      "learning_rate": 1.3795081428788493e-08,
      "loss": 0.8645223617553711,
      "step": 28090
    },
    {
      "epoch": 2.9695496519016604,
      "grad_norm": 1.0135029554367065,
      "learning_rate": 1.2890814928315231e-08,
      "loss": 0.8085169792175293,
      "step": 28100
    },
    {
      "epoch": 2.9706064969549653,
      "grad_norm": 1.4154417514801025,
      "learning_rate": 1.2017191172405962e-08,
      "loss": 0.8781013488769531,
      "step": 28110
    },
    {
      "epoch": 2.9716633420082696,
      "grad_norm": 1.3878732919692993,
      "learning_rate": 1.1174211232417575e-08,
      "loss": 0.8837569236755372,
      "step": 28120
    },
    {
      "epoch": 2.9727201870615745,
      "grad_norm": 1.2050855159759521,
      "learning_rate": 1.0361876142139792e-08,
      "loss": 0.8309374809265136,
      "step": 28130
    },
    {
      "epoch": 2.973777032114879,
      "grad_norm": 0.9934705495834351,
      "learning_rate": 9.580186897767407e-09,
      "loss": 0.8521280288696289,
      "step": 28140
    },
    {
      "epoch": 2.9748338771681837,
      "grad_norm": 1.339559555053711,
      "learning_rate": 8.829144457922489e-09,
      "loss": 0.7942620754241944,
      "step": 28150
    },
    {
      "epoch": 2.9758907222214885,
      "grad_norm": 1.5272369384765625,
      "learning_rate": 8.108749743637734e-09,
      "loss": 0.8327934265136718,
      "step": 28160
    },
    {
      "epoch": 2.976947567274793,
      "grad_norm": 1.5825679302215576,
      "learning_rate": 7.419003638364785e-09,
      "loss": 0.8163703918457031,
      "step": 28170
    },
    {
      "epoch": 2.9780044123280973,
      "grad_norm": 1.743831992149353,
      "learning_rate": 6.759906987968689e-09,
      "loss": 0.8478512763977051,
      "step": 28180
    },
    {
      "epoch": 2.979061257381402,
      "grad_norm": 1.4517738819122314,
      "learning_rate": 6.131460600725114e-09,
      "loss": 0.8876804351806641,
      "step": 28190
    },
    {
      "epoch": 2.980118102434707,
      "grad_norm": 1.115539789199829,
      "learning_rate": 5.533665247325903e-09,
      "loss": 0.9241631507873536,
      "step": 28200
    },
    {
      "epoch": 2.9811749474880114,
      "grad_norm": 1.241997480392456,
      "learning_rate": 4.9665216608735245e-09,
      "loss": 0.8331340789794922,
      "step": 28210
    },
    {
      "epoch": 2.9822317925413158,
      "grad_norm": 1.9293731451034546,
      "learning_rate": 4.430030536878294e-09,
      "loss": 0.7604406833648681,
      "step": 28220
    },
    {
      "epoch": 2.9832886375946206,
      "grad_norm": 1.9037610292434692,
      "learning_rate": 3.924192533263926e-09,
      "loss": 0.755042839050293,
      "step": 28230
    },
    {
      "epoch": 2.9843454826479254,
      "grad_norm": 1.1857556104660034,
      "learning_rate": 3.449008270359211e-09,
      "loss": 0.8992856979370117,
      "step": 28240
    },
    {
      "epoch": 2.98540232770123,
      "grad_norm": 1.5789319276809692,
      "learning_rate": 3.004478330900784e-09,
      "loss": 0.9083795547485352,
      "step": 28250
    },
    {
      "epoch": 2.9864591727545347,
      "grad_norm": 1.7061728239059448,
      "learning_rate": 2.590603260038682e-09,
      "loss": 0.9431068420410156,
      "step": 28260
    },
    {
      "epoch": 2.987516017807839,
      "grad_norm": 1.4992685317993164,
      "learning_rate": 2.2073835653252383e-09,
      "loss": 0.8583915710449219,
      "step": 28270
    },
    {
      "epoch": 2.988572862861144,
      "grad_norm": 1.2664291858673096,
      "learning_rate": 1.8548197167150838e-09,
      "loss": 0.8603200912475586,
      "step": 28280
    },
    {
      "epoch": 2.9896297079144483,
      "grad_norm": 1.313591718673706,
      "learning_rate": 1.532912146573473e-09,
      "loss": 0.8772127151489257,
      "step": 28290
    },
    {
      "epoch": 2.990686552967753,
      "grad_norm": 1.2386982440948486,
      "learning_rate": 1.2416612496707335e-09,
      "loss": 0.8571273803710937,
      "step": 28300
    },
    {
      "epoch": 2.9917433980210575,
      "grad_norm": 1.1998481750488281,
      "learning_rate": 9.810673831794903e-10,
      "loss": 0.8896389961242676,
      "step": 28310
    },
    {
      "epoch": 2.9928002430743623,
      "grad_norm": 1.4557340145111084,
      "learning_rate": 7.511308666774408e-10,
      "loss": 0.9232834815979004,
      "step": 28320
    },
    {
      "epoch": 2.993857088127667,
      "grad_norm": 1.3959667682647705,
      "learning_rate": 5.5185198214458e-10,
      "loss": 0.9544175148010254,
      "step": 28330
    },
    {
      "epoch": 2.9949139331809715,
      "grad_norm": 1.6930980682373047,
      "learning_rate": 3.8323097396320006e-10,
      "loss": 0.9191384315490723,
      "step": 28340
    },
    {
      "epoch": 2.995970778234276,
      "grad_norm": 1.811262607574463,
      "learning_rate": 2.4526804892344157e-10,
      "loss": 0.8569545745849609,
      "step": 28350
    },
    {
      "epoch": 2.9970276232875808,
      "grad_norm": 1.187766194343567,
      "learning_rate": 1.379633762149668e-10,
      "loss": 1.032188892364502,
      "step": 28360
    },
    {
      "epoch": 2.9980844683408856,
      "grad_norm": 1.189721941947937,
      "learning_rate": 6.131708742695974e-11,
      "loss": 0.8739087104797363,
      "step": 28370
    },
    {
      "epoch": 2.99914131339419,
      "grad_norm": 1.2776782512664795,
      "learning_rate": 1.5329276556452776e-11,
      "loss": 0.8936100006103516,
      "step": 28380
    },
    {
      "epoch": 3.0,
      "step": 28389,
      "total_flos": 1.5323224005185126e+18,
      "train_loss": 0.011790263345844626,
      "train_runtime": 3504.1342,
      "train_samples_per_second": 64.807,
      "train_steps_per_second": 8.102
    }
  ],
  "logging_steps": 10,
  "max_steps": 28389,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5323224005185126e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}