{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999922761433239,
  "eval_steps": 500,
  "global_step": 32367,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003089542670446207,
      "grad_norm": 1.4342206015296557,
      "learning_rate": 1.852995676343422e-07,
      "loss": 0.459,
      "step": 10
    },
    {
      "epoch": 0.0006179085340892414,
      "grad_norm": 1.2541838529642675,
      "learning_rate": 3.705991352686844e-07,
      "loss": 0.4524,
      "step": 20
    },
    {
      "epoch": 0.0009268628011338621,
      "grad_norm": 0.9828360076982188,
      "learning_rate": 5.558987029030266e-07,
      "loss": 0.4481,
      "step": 30
    },
    {
      "epoch": 0.0012358170681784828,
      "grad_norm": 0.6974302325411567,
      "learning_rate": 7.411982705373688e-07,
      "loss": 0.4281,
      "step": 40
    },
    {
      "epoch": 0.0015447713352231035,
      "grad_norm": 0.547411288843025,
      "learning_rate": 9.26497838171711e-07,
      "loss": 0.3942,
      "step": 50
    },
    {
      "epoch": 0.0018537256022677243,
      "grad_norm": 0.6249425214319002,
      "learning_rate": 1.1117974058060532e-06,
      "loss": 0.3598,
      "step": 60
    },
    {
      "epoch": 0.002162679869312345,
      "grad_norm": 0.4046247574364963,
      "learning_rate": 1.2970969734403955e-06,
      "loss": 0.3279,
      "step": 70
    },
    {
      "epoch": 0.0024716341363569655,
      "grad_norm": 0.3667300122619454,
      "learning_rate": 1.4823965410747376e-06,
      "loss": 0.2969,
      "step": 80
    },
    {
      "epoch": 0.0027805884034015865,
      "grad_norm": 0.3371062673025034,
      "learning_rate": 1.6676961087090797e-06,
      "loss": 0.2878,
      "step": 90
    },
    {
      "epoch": 0.003089542670446207,
      "grad_norm": 0.3260702974723146,
      "learning_rate": 1.852995676343422e-06,
      "loss": 0.2784,
      "step": 100
    },
    {
      "epoch": 0.003398496937490828,
      "grad_norm": 0.34444353567760644,
      "learning_rate": 2.0382952439777643e-06,
      "loss": 0.2615,
      "step": 110
    },
    {
      "epoch": 0.0037074512045354485,
      "grad_norm": 0.3850956468437575,
      "learning_rate": 2.2235948116121064e-06,
      "loss": 0.2676,
      "step": 120
    },
    {
      "epoch": 0.004016405471580069,
      "grad_norm": 0.3649231621770138,
      "learning_rate": 2.4088943792464484e-06,
      "loss": 0.2661,
      "step": 130
    },
    {
      "epoch": 0.00432535973862469,
      "grad_norm": 0.33008261225178304,
      "learning_rate": 2.594193946880791e-06,
      "loss": 0.2578,
      "step": 140
    },
    {
      "epoch": 0.004634314005669311,
      "grad_norm": 0.37275477440355054,
      "learning_rate": 2.779493514515133e-06,
      "loss": 0.258,
      "step": 150
    },
    {
      "epoch": 0.004943268272713931,
      "grad_norm": 0.33558939418979133,
      "learning_rate": 2.964793082149475e-06,
      "loss": 0.2509,
      "step": 160
    },
    {
      "epoch": 0.005252222539758552,
      "grad_norm": 0.33590394685537367,
      "learning_rate": 3.1500926497838172e-06,
      "loss": 0.2441,
      "step": 170
    },
    {
      "epoch": 0.005561176806803173,
      "grad_norm": 0.39338675651403776,
      "learning_rate": 3.3353922174181593e-06,
      "loss": 0.2455,
      "step": 180
    },
    {
      "epoch": 0.005870131073847794,
      "grad_norm": 0.34625762130866167,
      "learning_rate": 3.5206917850525014e-06,
      "loss": 0.2394,
      "step": 190
    },
    {
      "epoch": 0.006179085340892414,
      "grad_norm": 0.3556251662900565,
      "learning_rate": 3.705991352686844e-06,
      "loss": 0.2385,
      "step": 200
    },
    {
      "epoch": 0.006488039607937035,
      "grad_norm": 0.3260564080707061,
      "learning_rate": 3.891290920321186e-06,
      "loss": 0.2322,
      "step": 210
    },
    {
      "epoch": 0.006796993874981656,
      "grad_norm": 0.3676995014349691,
      "learning_rate": 4.0765904879555285e-06,
      "loss": 0.2331,
      "step": 220
    },
    {
      "epoch": 0.007105948142026277,
      "grad_norm": 0.35446900070732784,
      "learning_rate": 4.26189005558987e-06,
      "loss": 0.2339,
      "step": 230
    },
    {
      "epoch": 0.007414902409070897,
      "grad_norm": 0.3819527072560428,
      "learning_rate": 4.447189623224213e-06,
      "loss": 0.2357,
      "step": 240
    },
    {
      "epoch": 0.007723856676115518,
      "grad_norm": 0.3480466322290397,
      "learning_rate": 4.632489190858555e-06,
      "loss": 0.2254,
      "step": 250
    },
    {
      "epoch": 0.008032810943160138,
      "grad_norm": 0.3194903733187964,
      "learning_rate": 4.817788758492897e-06,
      "loss": 0.2377,
      "step": 260
    },
    {
      "epoch": 0.008341765210204759,
      "grad_norm": 0.29604030153949573,
      "learning_rate": 5.003088326127239e-06,
      "loss": 0.2225,
      "step": 270
    },
    {
      "epoch": 0.00865071947724938,
      "grad_norm": 0.39212973620577857,
      "learning_rate": 5.188387893761582e-06,
      "loss": 0.2459,
      "step": 280
    },
    {
      "epoch": 0.008959673744294001,
      "grad_norm": 0.30359592816187564,
      "learning_rate": 5.3736874613959236e-06,
      "loss": 0.2261,
      "step": 290
    },
    {
      "epoch": 0.009268628011338622,
      "grad_norm": 0.3349637043598593,
      "learning_rate": 5.558987029030266e-06,
      "loss": 0.2211,
      "step": 300
    },
    {
      "epoch": 0.009577582278383243,
      "grad_norm": 0.3942177663321336,
      "learning_rate": 5.744286596664609e-06,
      "loss": 0.2318,
      "step": 310
    },
    {
      "epoch": 0.009886536545427862,
      "grad_norm": 0.349281010677901,
      "learning_rate": 5.92958616429895e-06,
      "loss": 0.2257,
      "step": 320
    },
    {
      "epoch": 0.010195490812472483,
      "grad_norm": 0.36174185858414665,
      "learning_rate": 6.114885731933293e-06,
      "loss": 0.2258,
      "step": 330
    },
    {
      "epoch": 0.010504445079517104,
      "grad_norm": 0.33791988292092384,
      "learning_rate": 6.3001852995676344e-06,
      "loss": 0.2244,
      "step": 340
    },
    {
      "epoch": 0.010813399346561725,
      "grad_norm": 0.36238831779965897,
      "learning_rate": 6.485484867201976e-06,
      "loss": 0.217,
      "step": 350
    },
    {
      "epoch": 0.011122353613606346,
      "grad_norm": 0.4074399279408642,
      "learning_rate": 6.670784434836319e-06,
      "loss": 0.2179,
      "step": 360
    },
    {
      "epoch": 0.011431307880650967,
      "grad_norm": 0.4061444760562626,
      "learning_rate": 6.856084002470661e-06,
      "loss": 0.2207,
      "step": 370
    },
    {
      "epoch": 0.011740262147695588,
      "grad_norm": 0.2998368254360446,
      "learning_rate": 7.041383570105003e-06,
      "loss": 0.2242,
      "step": 380
    },
    {
      "epoch": 0.012049216414740209,
      "grad_norm": 0.3792762067269328,
      "learning_rate": 7.226683137739345e-06,
      "loss": 0.221,
      "step": 390
    },
    {
      "epoch": 0.012358170681784828,
      "grad_norm": 0.38562436397551597,
      "learning_rate": 7.411982705373688e-06,
      "loss": 0.2233,
      "step": 400
    },
    {
      "epoch": 0.012667124948829449,
      "grad_norm": 0.37949063216436724,
      "learning_rate": 7.59728227300803e-06,
      "loss": 0.2272,
      "step": 410
    },
    {
      "epoch": 0.01297607921587407,
      "grad_norm": 0.4070170847933438,
      "learning_rate": 7.782581840642372e-06,
      "loss": 0.2201,
      "step": 420
    },
    {
      "epoch": 0.013285033482918691,
      "grad_norm": 0.3017443172169592,
      "learning_rate": 7.967881408276715e-06,
      "loss": 0.2205,
      "step": 430
    },
    {
      "epoch": 0.013593987749963312,
      "grad_norm": 0.3201901762844856,
      "learning_rate": 8.153180975911057e-06,
      "loss": 0.2173,
      "step": 440
    },
    {
      "epoch": 0.013902942017007933,
      "grad_norm": 0.3253472072116366,
      "learning_rate": 8.338480543545398e-06,
      "loss": 0.221,
      "step": 450
    },
    {
      "epoch": 0.014211896284052554,
      "grad_norm": 0.33183918202637014,
      "learning_rate": 8.52378011117974e-06,
      "loss": 0.2151,
      "step": 460
    },
    {
      "epoch": 0.014520850551097173,
      "grad_norm": 0.3243332278148931,
      "learning_rate": 8.709079678814083e-06,
      "loss": 0.2174,
      "step": 470
    },
    {
      "epoch": 0.014829804818141794,
      "grad_norm": 0.36798410428072925,
      "learning_rate": 8.894379246448425e-06,
      "loss": 0.2172,
      "step": 480
    },
    {
      "epoch": 0.015138759085186415,
      "grad_norm": 0.3261138949645182,
      "learning_rate": 9.079678814082766e-06,
      "loss": 0.21,
      "step": 490
    },
    {
      "epoch": 0.015447713352231036,
      "grad_norm": 0.2632848279909067,
      "learning_rate": 9.26497838171711e-06,
      "loss": 0.2133,
      "step": 500
    },
    {
      "epoch": 0.015756667619275657,
      "grad_norm": 0.3446013399586666,
      "learning_rate": 9.450277949351451e-06,
      "loss": 0.2152,
      "step": 510
    },
    {
      "epoch": 0.016065621886320276,
      "grad_norm": 0.29629460796616464,
      "learning_rate": 9.635577516985794e-06,
      "loss": 0.2184,
      "step": 520
    },
    {
      "epoch": 0.0163745761533649,
      "grad_norm": 0.3014262990944555,
      "learning_rate": 9.820877084620136e-06,
      "loss": 0.2171,
      "step": 530
    },
    {
      "epoch": 0.016683530420409518,
      "grad_norm": 0.32096296401164126,
      "learning_rate": 1.0006176652254479e-05,
      "loss": 0.2099,
      "step": 540
    },
    {
      "epoch": 0.01699248468745414,
      "grad_norm": 0.320451613139758,
      "learning_rate": 1.019147621988882e-05,
      "loss": 0.2139,
      "step": 550
    },
    {
      "epoch": 0.01730143895449876,
      "grad_norm": 0.35135365320921647,
      "learning_rate": 1.0376775787523164e-05,
      "loss": 0.2134,
      "step": 560
    },
    {
      "epoch": 0.01761039322154338,
      "grad_norm": 0.33802984085653676,
      "learning_rate": 1.0562075355157505e-05,
      "loss": 0.2099,
      "step": 570
    },
    {
      "epoch": 0.017919347488588002,
      "grad_norm": 0.30465329323172424,
      "learning_rate": 1.0747374922791847e-05,
      "loss": 0.207,
      "step": 580
    },
    {
      "epoch": 0.01822830175563262,
      "grad_norm": 0.39913719353068217,
      "learning_rate": 1.093267449042619e-05,
      "loss": 0.2162,
      "step": 590
    },
    {
      "epoch": 0.018537256022677244,
      "grad_norm": 0.39279649552480556,
      "learning_rate": 1.1117974058060532e-05,
      "loss": 0.2048,
      "step": 600
    },
    {
      "epoch": 0.018846210289721863,
      "grad_norm": 0.3392053906234814,
      "learning_rate": 1.1303273625694873e-05,
      "loss": 0.2108,
      "step": 610
    },
    {
      "epoch": 0.019155164556766486,
      "grad_norm": 0.2888812954242707,
      "learning_rate": 1.1488573193329217e-05,
      "loss": 0.2097,
      "step": 620
    },
    {
      "epoch": 0.019464118823811105,
      "grad_norm": 0.3372761067600913,
      "learning_rate": 1.1673872760963558e-05,
      "loss": 0.2119,
      "step": 630
    },
    {
      "epoch": 0.019773073090855724,
      "grad_norm": 0.3712557625995704,
      "learning_rate": 1.18591723285979e-05,
      "loss": 0.2226,
      "step": 640
    },
    {
      "epoch": 0.020082027357900347,
      "grad_norm": 0.3197200919411092,
      "learning_rate": 1.2044471896232243e-05,
      "loss": 0.2099,
      "step": 650
    },
    {
      "epoch": 0.020390981624944966,
      "grad_norm": 0.28518041257422966,
      "learning_rate": 1.2229771463866586e-05,
      "loss": 0.2092,
      "step": 660
    },
    {
      "epoch": 0.02069993589198959,
      "grad_norm": 0.28640971911900237,
      "learning_rate": 1.2415071031500926e-05,
      "loss": 0.2093,
      "step": 670
    },
    {
      "epoch": 0.021008890159034208,
      "grad_norm": 0.31681783495811866,
      "learning_rate": 1.2600370599135269e-05,
      "loss": 0.2175,
      "step": 680
    },
    {
      "epoch": 0.02131784442607883,
      "grad_norm": 0.3086403635983181,
      "learning_rate": 1.2785670166769611e-05,
      "loss": 0.2145,
      "step": 690
    },
    {
      "epoch": 0.02162679869312345,
      "grad_norm": 0.279274615260603,
      "learning_rate": 1.2970969734403952e-05,
      "loss": 0.2097,
      "step": 700
    },
    {
      "epoch": 0.021935752960168073,
      "grad_norm": 0.28404939336317203,
      "learning_rate": 1.3156269302038296e-05,
      "loss": 0.215,
      "step": 710
    },
    {
      "epoch": 0.022244707227212692,
      "grad_norm": 0.36193233481706244,
      "learning_rate": 1.3341568869672637e-05,
      "loss": 0.2147,
      "step": 720
    },
    {
      "epoch": 0.02255366149425731,
      "grad_norm": 0.2906216615512196,
      "learning_rate": 1.352686843730698e-05,
      "loss": 0.2063,
      "step": 730
    },
    {
      "epoch": 0.022862615761301934,
      "grad_norm": 0.3710603016488123,
      "learning_rate": 1.3712168004941322e-05,
      "loss": 0.2077,
      "step": 740
    },
    {
      "epoch": 0.023171570028346553,
      "grad_norm": 0.2886159866780826,
      "learning_rate": 1.3897467572575665e-05,
      "loss": 0.2093,
      "step": 750
    },
    {
      "epoch": 0.023480524295391176,
      "grad_norm": 0.3112205868629234,
      "learning_rate": 1.4082767140210006e-05,
      "loss": 0.2073,
      "step": 760
    },
    {
      "epoch": 0.023789478562435795,
      "grad_norm": 0.26528993566514497,
      "learning_rate": 1.426806670784435e-05,
      "loss": 0.2107,
      "step": 770
    },
    {
      "epoch": 0.024098432829480418,
      "grad_norm": 0.29394783645223177,
      "learning_rate": 1.445336627547869e-05,
      "loss": 0.2185,
      "step": 780
    },
    {
      "epoch": 0.024407387096525037,
      "grad_norm": 0.26589042931900475,
      "learning_rate": 1.4638665843113033e-05,
      "loss": 0.2132,
      "step": 790
    },
    {
      "epoch": 0.024716341363569656,
      "grad_norm": 0.266047120102079,
      "learning_rate": 1.4823965410747376e-05,
      "loss": 0.2164,
      "step": 800
    },
    {
      "epoch": 0.02502529563061428,
      "grad_norm": 0.28935549812532074,
      "learning_rate": 1.5009264978381716e-05,
      "loss": 0.2111,
      "step": 810
    },
    {
      "epoch": 0.025334249897658898,
      "grad_norm": 0.3268318837541725,
      "learning_rate": 1.519456454601606e-05,
      "loss": 0.2076,
      "step": 820
    },
    {
      "epoch": 0.02564320416470352,
      "grad_norm": 0.26553688783929824,
      "learning_rate": 1.53798641136504e-05,
      "loss": 0.2096,
      "step": 830
    },
    {
      "epoch": 0.02595215843174814,
      "grad_norm": 0.3230918273181566,
      "learning_rate": 1.5565163681284744e-05,
      "loss": 0.2058,
      "step": 840
    },
    {
      "epoch": 0.026261112698792763,
      "grad_norm": 0.3158040319486062,
      "learning_rate": 1.5750463248919087e-05,
      "loss": 0.2144,
      "step": 850
    },
    {
      "epoch": 0.026570066965837382,
      "grad_norm": 0.42005876152094646,
      "learning_rate": 1.593576281655343e-05,
      "loss": 0.2094,
      "step": 860
    },
    {
      "epoch": 0.026879021232882,
      "grad_norm": 0.29990865839111325,
      "learning_rate": 1.6121062384187768e-05,
      "loss": 0.2111,
      "step": 870
    },
    {
      "epoch": 0.027187975499926624,
      "grad_norm": 0.30161102177218907,
      "learning_rate": 1.6306361951822114e-05,
      "loss": 0.2068,
      "step": 880
    },
    {
      "epoch": 0.027496929766971243,
      "grad_norm": 0.37056174507591677,
      "learning_rate": 1.6491661519456457e-05,
      "loss": 0.2107,
      "step": 890
    },
    {
      "epoch": 0.027805884034015866,
      "grad_norm": 0.3655205446506857,
      "learning_rate": 1.6676961087090796e-05,
      "loss": 0.209,
      "step": 900
    },
    {
      "epoch": 0.028114838301060485,
      "grad_norm": 0.33860684112890227,
      "learning_rate": 1.6862260654725138e-05,
      "loss": 0.2075,
      "step": 910
    },
    {
      "epoch": 0.028423792568105108,
      "grad_norm": 0.25129594692224955,
      "learning_rate": 1.704756022235948e-05,
      "loss": 0.2067,
      "step": 920
    },
    {
      "epoch": 0.028732746835149727,
      "grad_norm": 0.3251631455751991,
      "learning_rate": 1.7232859789993823e-05,
      "loss": 0.2052,
      "step": 930
    },
    {
      "epoch": 0.029041701102194346,
      "grad_norm": 0.363155931575057,
      "learning_rate": 1.7418159357628166e-05,
      "loss": 0.2032,
      "step": 940
    },
    {
      "epoch": 0.02935065536923897,
      "grad_norm": 0.30100529535994425,
      "learning_rate": 1.7603458925262508e-05,
      "loss": 0.2031,
      "step": 950
    },
    {
      "epoch": 0.02965960963628359,
      "grad_norm": 0.3141907193101939,
      "learning_rate": 1.778875849289685e-05,
      "loss": 0.2032,
      "step": 960
    },
    {
      "epoch": 0.02996856390332821,
      "grad_norm": 0.298253799467931,
      "learning_rate": 1.7974058060531193e-05,
      "loss": 0.2062,
      "step": 970
    },
    {
      "epoch": 0.03027751817037283,
      "grad_norm": 0.2913001451283101,
      "learning_rate": 1.8159357628165532e-05,
      "loss": 0.2042,
      "step": 980
    },
    {
      "epoch": 0.030586472437417453,
      "grad_norm": 0.29004478573458653,
      "learning_rate": 1.8344657195799875e-05,
      "loss": 0.2115,
      "step": 990
    },
    {
      "epoch": 0.030895426704462072,
      "grad_norm": 0.3039673938016494,
      "learning_rate": 1.852995676343422e-05,
      "loss": 0.2078,
      "step": 1000
    },
    {
      "epoch": 0.03120438097150669,
      "grad_norm": 0.26992191709125957,
      "learning_rate": 1.8715256331068563e-05,
      "loss": 0.2144,
      "step": 1010
    },
    {
      "epoch": 0.031513335238551314,
      "grad_norm": 0.29224055505601537,
      "learning_rate": 1.8900555898702902e-05,
      "loss": 0.2051,
      "step": 1020
    },
    {
      "epoch": 0.03182228950559594,
      "grad_norm": 0.2813736747873565,
      "learning_rate": 1.9085855466337245e-05,
      "loss": 0.2046,
      "step": 1030
    },
    {
      "epoch": 0.03213124377264055,
      "grad_norm": 0.3617377978103828,
      "learning_rate": 1.9271155033971588e-05,
      "loss": 0.2077,
      "step": 1040
    },
    {
      "epoch": 0.032440198039685175,
      "grad_norm": 0.426903690417414,
      "learning_rate": 1.945645460160593e-05,
      "loss": 0.2057,
      "step": 1050
    },
    {
      "epoch": 0.0327491523067298,
      "grad_norm": 0.26884197701522644,
      "learning_rate": 1.9641754169240273e-05,
      "loss": 0.2021,
      "step": 1060
    },
    {
      "epoch": 0.033058106573774414,
      "grad_norm": 0.3138092113639732,
      "learning_rate": 1.9827053736874615e-05,
      "loss": 0.2041,
      "step": 1070
    },
    {
      "epoch": 0.033367060840819036,
      "grad_norm": 0.26273787982665225,
      "learning_rate": 2.0012353304508958e-05,
      "loss": 0.2046,
      "step": 1080
    },
    {
      "epoch": 0.03367601510786366,
      "grad_norm": 0.3305991889210736,
      "learning_rate": 2.01976528721433e-05,
      "loss": 0.2108,
      "step": 1090
    },
    {
      "epoch": 0.03398496937490828,
      "grad_norm": 0.28132184754515793,
      "learning_rate": 2.038295243977764e-05,
      "loss": 0.2024,
      "step": 1100
    },
    {
      "epoch": 0.0342939236419529,
      "grad_norm": 0.3310218166064453,
      "learning_rate": 2.0568252007411982e-05,
      "loss": 0.2077,
      "step": 1110
    },
    {
      "epoch": 0.03460287790899752,
      "grad_norm": 0.24084016233984354,
      "learning_rate": 2.0753551575046328e-05,
      "loss": 0.2304,
      "step": 1120
    },
    {
      "epoch": 0.03491183217604214,
      "grad_norm": 0.24689565123316048,
      "learning_rate": 2.0938851142680667e-05,
      "loss": 0.2084,
      "step": 1130
    },
    {
      "epoch": 0.03522078644308676,
      "grad_norm": 0.3202170546704081,
      "learning_rate": 2.112415071031501e-05,
      "loss": 0.2085,
      "step": 1140
    },
    {
      "epoch": 0.03552974071013138,
      "grad_norm": 0.220583117467106,
      "learning_rate": 2.1309450277949352e-05,
      "loss": 0.2069,
      "step": 1150
    },
    {
      "epoch": 0.035838694977176004,
      "grad_norm": 0.2080170682936219,
      "learning_rate": 2.1494749845583694e-05,
      "loss": 0.2069,
      "step": 1160
    },
    {
      "epoch": 0.03614764924422063,
      "grad_norm": 0.2627342483578785,
      "learning_rate": 2.1680049413218033e-05,
      "loss": 0.2123,
      "step": 1170
    },
    {
      "epoch": 0.03645660351126524,
      "grad_norm": 0.3358338758591602,
      "learning_rate": 2.186534898085238e-05,
      "loss": 0.2178,
      "step": 1180
    },
    {
      "epoch": 0.036765557778309865,
      "grad_norm": 0.38073592070845597,
      "learning_rate": 2.2050648548486722e-05,
      "loss": 0.2054,
      "step": 1190
    },
    {
      "epoch": 0.03707451204535449,
      "grad_norm": 0.28383460905604635,
      "learning_rate": 2.2235948116121064e-05,
      "loss": 0.215,
      "step": 1200
    },
    {
      "epoch": 0.037383466312399104,
      "grad_norm": 0.32201223136969137,
      "learning_rate": 2.2421247683755403e-05,
      "loss": 0.2049,
      "step": 1210
    },
    {
      "epoch": 0.037692420579443726,
      "grad_norm": 0.2330892975043787,
      "learning_rate": 2.2606547251389746e-05,
      "loss": 0.2085,
      "step": 1220
    },
    {
      "epoch": 0.03800137484648835,
      "grad_norm": 0.26739199739680014,
      "learning_rate": 2.279184681902409e-05,
      "loss": 0.207,
      "step": 1230
    },
    {
      "epoch": 0.03831032911353297,
      "grad_norm": 0.29859132653166187,
      "learning_rate": 2.2977146386658434e-05,
      "loss": 0.2101,
      "step": 1240
    },
    {
      "epoch": 0.03861928338057759,
      "grad_norm": 0.30552113069212283,
      "learning_rate": 2.3162445954292774e-05,
      "loss": 0.2095,
      "step": 1250
    },
    {
      "epoch": 0.03892823764762221,
      "grad_norm": 0.30270640732504084,
      "learning_rate": 2.3347745521927116e-05,
      "loss": 0.2125,
      "step": 1260
    },
    {
      "epoch": 0.03923719191466683,
      "grad_norm": 0.26397966225828484,
      "learning_rate": 2.353304508956146e-05,
      "loss": 0.2104,
      "step": 1270
    },
    {
      "epoch": 0.03954614618171145,
      "grad_norm": 0.28745917243179153,
      "learning_rate": 2.37183446571958e-05,
      "loss": 0.208,
      "step": 1280
    },
    {
      "epoch": 0.03985510044875607,
      "grad_norm": 0.27010837808728094,
      "learning_rate": 2.390364422483014e-05,
      "loss": 0.2044,
      "step": 1290
    },
    {
      "epoch": 0.040164054715800694,
      "grad_norm": 0.2792843380104636,
      "learning_rate": 2.4088943792464486e-05,
      "loss": 0.2005,
      "step": 1300
    },
    {
      "epoch": 0.04047300898284532,
      "grad_norm": 0.22371555206758015,
      "learning_rate": 2.427424336009883e-05,
      "loss": 0.2147,
      "step": 1310
    },
    {
      "epoch": 0.04078196324988993,
      "grad_norm": 0.29757581055889437,
      "learning_rate": 2.445954292773317e-05,
      "loss": 0.2028,
      "step": 1320
    },
    {
      "epoch": 0.041090917516934555,
      "grad_norm": 0.25619509674648794,
      "learning_rate": 2.464484249536751e-05,
      "loss": 0.2058,
      "step": 1330
    },
    {
      "epoch": 0.04139987178397918,
      "grad_norm": 0.2828499187719199,
      "learning_rate": 2.4830142063001853e-05,
      "loss": 0.2054,
      "step": 1340
    },
    {
      "epoch": 0.041708826051023794,
      "grad_norm": 0.23957860940910639,
      "learning_rate": 2.5015441630636195e-05,
      "loss": 0.2059,
      "step": 1350
    },
    {
      "epoch": 0.042017780318068416,
      "grad_norm": 0.2347743056596026,
      "learning_rate": 2.5200741198270538e-05,
      "loss": 0.2071,
      "step": 1360
    },
    {
      "epoch": 0.04232673458511304,
      "grad_norm": 0.2354164254244114,
      "learning_rate": 2.538604076590488e-05,
      "loss": 0.2036,
      "step": 1370
    },
    {
      "epoch": 0.04263568885215766,
      "grad_norm": 0.28253508725613863,
      "learning_rate": 2.5571340333539223e-05,
      "loss": 0.2067,
      "step": 1380
    },
    {
      "epoch": 0.04294464311920228,
      "grad_norm": 0.3359662573400643,
      "learning_rate": 2.5756639901173565e-05,
      "loss": 0.2068,
      "step": 1390
    },
    {
      "epoch": 0.0432535973862469,
      "grad_norm": 0.23342104156491575,
      "learning_rate": 2.5941939468807904e-05,
      "loss": 0.2048,
      "step": 1400
    },
    {
      "epoch": 0.04356255165329152,
      "grad_norm": 0.3288703789832579,
      "learning_rate": 2.6127239036442247e-05,
      "loss": 0.2106,
      "step": 1410
    },
    {
      "epoch": 0.043871505920336146,
      "grad_norm": 0.2311175530684186,
      "learning_rate": 2.6312538604076593e-05,
      "loss": 0.2035,
      "step": 1420
    },
    {
      "epoch": 0.04418046018738076,
      "grad_norm": 0.24902940547184585,
      "learning_rate": 2.6497838171710935e-05,
      "loss": 0.2101,
      "step": 1430
    },
    {
      "epoch": 0.044489414454425384,
      "grad_norm": 0.30318744715773366,
      "learning_rate": 2.6683137739345275e-05,
      "loss": 0.2121,
      "step": 1440
    },
    {
      "epoch": 0.04479836872147001,
      "grad_norm": 0.22206577360841506,
      "learning_rate": 2.6868437306979617e-05,
      "loss": 0.2103,
      "step": 1450
    },
    {
      "epoch": 0.04510732298851462,
      "grad_norm": 0.20158462025933055,
      "learning_rate": 2.705373687461396e-05,
      "loss": 0.1993,
      "step": 1460
    },
    {
      "epoch": 0.045416277255559245,
      "grad_norm": 0.2878798477872172,
      "learning_rate": 2.7239036442248302e-05,
      "loss": 0.1977,
      "step": 1470
    },
    {
      "epoch": 0.04572523152260387,
      "grad_norm": 0.2066911603227598,
      "learning_rate": 2.7424336009882645e-05,
      "loss": 0.2153,
      "step": 1480
    },
    {
      "epoch": 0.04603418578964849,
      "grad_norm": 0.18272356845350932,
      "learning_rate": 2.7609635577516987e-05,
      "loss": 0.2029,
      "step": 1490
    },
    {
      "epoch": 0.046343140056693106,
      "grad_norm": 0.28595402585297963,
      "learning_rate": 2.779493514515133e-05,
      "loss": 0.2136,
      "step": 1500
    },
    {
      "epoch": 0.04665209432373773,
      "grad_norm": 0.20644642694772639,
      "learning_rate": 2.7980234712785672e-05,
      "loss": 0.2042,
      "step": 1510
    },
    {
      "epoch": 0.04696104859078235,
      "grad_norm": 0.277579026973121,
      "learning_rate": 2.816553428042001e-05,
      "loss": 0.2034,
      "step": 1520
    },
    {
      "epoch": 0.04727000285782697,
      "grad_norm": 0.26497694489417567,
      "learning_rate": 2.8350833848054354e-05,
      "loss": 0.2031,
      "step": 1530
    },
    {
      "epoch": 0.04757895712487159,
      "grad_norm": 0.22929988466716691,
      "learning_rate": 2.85361334156887e-05,
      "loss": 0.2074,
      "step": 1540
    },
    {
      "epoch": 0.04788791139191621,
      "grad_norm": 0.26775708641135415,
      "learning_rate": 2.872143298332304e-05,
      "loss": 0.2031,
      "step": 1550
    },
    {
      "epoch": 0.048196865658960836,
      "grad_norm": 0.2239949588049736,
      "learning_rate": 2.890673255095738e-05,
      "loss": 0.2064,
      "step": 1560
    },
    {
      "epoch": 0.04850581992600545,
      "grad_norm": 0.20883167310182313,
      "learning_rate": 2.9092032118591724e-05,
      "loss": 0.206,
      "step": 1570
    },
    {
      "epoch": 0.048814774193050074,
      "grad_norm": 0.19285251737246453,
      "learning_rate": 2.9277331686226066e-05,
      "loss": 0.1983,
      "step": 1580
    },
    {
      "epoch": 0.0491237284600947,
      "grad_norm": 0.2956628718116073,
      "learning_rate": 2.9462631253860405e-05,
      "loss": 0.1988,
      "step": 1590
    },
    {
      "epoch": 0.04943268272713931,
      "grad_norm": 0.250995206935813,
      "learning_rate": 2.964793082149475e-05,
      "loss": 0.2004,
      "step": 1600
    },
    {
      "epoch": 0.049741636994183935,
      "grad_norm": 0.243205872744263,
      "learning_rate": 2.9833230389129094e-05,
      "loss": 0.1982,
      "step": 1610
    },
    {
      "epoch": 0.05005059126122856,
      "grad_norm": 0.25558818024233726,
      "learning_rate": 2.999999992170622e-05,
      "loss": 0.2036,
      "step": 1620
    },
    {
      "epoch": 0.05035954552827318,
      "grad_norm": 0.23149680065562725,
      "learning_rate": 2.9999990526453555e-05,
      "loss": 0.2027,
      "step": 1630
    },
    {
      "epoch": 0.050668499795317797,
      "grad_norm": 0.24011462992841165,
      "learning_rate": 2.9999965472456043e-05,
      "loss": 0.203,
      "step": 1640
    },
    {
      "epoch": 0.05097745406236242,
      "grad_norm": 0.24085468590450437,
      "learning_rate": 2.9999924759739835e-05,
      "loss": 0.2064,
      "step": 1650
    },
    {
      "epoch": 0.05128640832940704,
      "grad_norm": 0.24906142359876013,
      "learning_rate": 2.999986838834743e-05,
      "loss": 0.2047,
      "step": 1660
    },
    {
      "epoch": 0.05159536259645166,
      "grad_norm": 0.42395661256821077,
      "learning_rate": 2.999979635833768e-05,
      "loss": 0.2096,
      "step": 1670
    },
    {
      "epoch": 0.05190431686349628,
      "grad_norm": 0.20504456024115916,
      "learning_rate": 2.999970866978578e-05,
      "loss": 0.2003,
      "step": 1680
    },
    {
      "epoch": 0.0522132711305409,
      "grad_norm": 0.25305112383752665,
      "learning_rate": 2.999960532278326e-05,
      "loss": 0.2114,
      "step": 1690
    },
    {
      "epoch": 0.052522225397585526,
      "grad_norm": 0.2470280003459118,
      "learning_rate": 2.9999486317438012e-05,
      "loss": 0.2054,
      "step": 1700
    },
    {
      "epoch": 0.05283117966463014,
      "grad_norm": 0.17619984944090902,
      "learning_rate": 2.999935165387427e-05,
      "loss": 0.1978,
      "step": 1710
    },
    {
      "epoch": 0.053140133931674764,
      "grad_norm": 0.25629485039850225,
      "learning_rate": 2.9999201332232602e-05,
      "loss": 0.2056,
      "step": 1720
    },
    {
      "epoch": 0.05344908819871939,
      "grad_norm": 0.22095393558240067,
      "learning_rate": 2.999903535266994e-05,
      "loss": 0.2036,
      "step": 1730
    },
    {
      "epoch": 0.053758042465764,
      "grad_norm": 0.3143405292517971,
      "learning_rate": 2.9998853715359554e-05,
      "loss": 0.2093,
      "step": 1740
    },
    {
      "epoch": 0.054066996732808625,
      "grad_norm": 0.2309885475456474,
      "learning_rate": 2.9998656420491052e-05,
      "loss": 0.217,
      "step": 1750
    },
    {
      "epoch": 0.05437595099985325,
      "grad_norm": 0.20760194513380262,
      "learning_rate": 2.99984434682704e-05,
      "loss": 0.2166,
      "step": 1760
    },
    {
      "epoch": 0.05468490526689787,
      "grad_norm": 0.24300116908428407,
      "learning_rate": 2.9998214858919898e-05,
      "loss": 0.2033,
      "step": 1770
    },
    {
      "epoch": 0.05499385953394249,
      "grad_norm": 0.2219233975267644,
      "learning_rate": 2.9997970592678194e-05,
      "loss": 0.2021,
      "step": 1780
    },
    {
      "epoch": 0.05530281380098711,
      "grad_norm": 0.30710149760971917,
      "learning_rate": 2.999771066980028e-05,
      "loss": 0.2177,
      "step": 1790
    },
    {
      "epoch": 0.05561176806803173,
      "grad_norm": 0.16031344634197983,
      "learning_rate": 2.9997435090557503e-05,
      "loss": 0.2015,
      "step": 1800
    },
    {
      "epoch": 0.05592072233507635,
      "grad_norm": 0.32697590864254217,
      "learning_rate": 2.999714385523754e-05,
      "loss": 0.2177,
      "step": 1810
    },
    {
      "epoch": 0.05622967660212097,
      "grad_norm": 0.17668176381250358,
      "learning_rate": 2.9996836964144416e-05,
      "loss": 0.1981,
      "step": 1820
    },
    {
      "epoch": 0.05653863086916559,
      "grad_norm": 0.17168676181872455,
      "learning_rate": 2.9996514417598497e-05,
      "loss": 0.2106,
      "step": 1830
    },
    {
      "epoch": 0.056847585136210216,
      "grad_norm": 0.18135718361194356,
      "learning_rate": 2.9996176215936498e-05,
      "loss": 0.1999,
      "step": 1840
    },
    {
      "epoch": 0.05715653940325483,
      "grad_norm": 0.3034185998566844,
      "learning_rate": 2.9995822359511474e-05,
      "loss": 0.2145,
      "step": 1850
    },
    {
      "epoch": 0.057465493670299454,
      "grad_norm": 0.22175858229853335,
      "learning_rate": 2.999545284869282e-05,
      "loss": 0.2067,
      "step": 1860
    },
    {
      "epoch": 0.05777444793734408,
      "grad_norm": 0.23366649948214785,
      "learning_rate": 2.9995067683866277e-05,
      "loss": 0.2005,
      "step": 1870
    },
    {
      "epoch": 0.05808340220438869,
      "grad_norm": 0.18657800025148788,
      "learning_rate": 2.999466686543392e-05,
      "loss": 0.2143,
      "step": 1880
    },
    {
      "epoch": 0.058392356471433315,
      "grad_norm": 0.19587658784940068,
      "learning_rate": 2.9994250393814177e-05,
      "loss": 0.2015,
      "step": 1890
    },
    {
      "epoch": 0.05870131073847794,
      "grad_norm": 0.1988143078568588,
      "learning_rate": 2.9993818269441807e-05,
      "loss": 0.1985,
      "step": 1900
    },
    {
      "epoch": 0.05901026500552256,
      "grad_norm": 0.31463751837253334,
      "learning_rate": 2.999337049276791e-05,
      "loss": 0.2002,
      "step": 1910
    },
    {
      "epoch": 0.05931921927256718,
      "grad_norm": 0.29579355828477577,
      "learning_rate": 2.9992907064259932e-05,
      "loss": 0.2023,
      "step": 1920
    },
    {
      "epoch": 0.0596281735396118,
      "grad_norm": 0.2299056861396528,
      "learning_rate": 2.9992427984401645e-05,
      "loss": 0.2122,
      "step": 1930
    },
    {
      "epoch": 0.05993712780665642,
      "grad_norm": 0.18186606292999008,
      "learning_rate": 2.999193325369318e-05,
      "loss": 0.2,
      "step": 1940
    },
    {
      "epoch": 0.06024608207370104,
      "grad_norm": 0.21384750204974323,
      "learning_rate": 2.9991422872650983e-05,
      "loss": 0.2116,
      "step": 1950
    },
    {
      "epoch": 0.06055503634074566,
      "grad_norm": 1.6716992559379211,
      "learning_rate": 2.9990896841807862e-05,
      "loss": 0.39,
      "step": 1960
    },
    {
      "epoch": 0.06086399060779028,
      "grad_norm": 38.03714494100377,
      "learning_rate": 2.9990355161712942e-05,
      "loss": 0.3536,
      "step": 1970
    },
    {
      "epoch": 0.061172944874834906,
      "grad_norm": 3.9014531815641784,
      "learning_rate": 2.9989797832931692e-05,
      "loss": 0.2543,
      "step": 1980
    },
    {
      "epoch": 0.06148189914187952,
      "grad_norm": 7.019711079073617,
      "learning_rate": 2.998922485604592e-05,
      "loss": 0.302,
      "step": 1990
    },
    {
      "epoch": 0.061790853408924144,
      "grad_norm": 0.28537204731383636,
      "learning_rate": 2.9988636231653765e-05,
      "loss": 0.2307,
      "step": 2000
    },
    {
      "epoch": 0.06209980767596877,
      "grad_norm": 0.19569138798894245,
      "learning_rate": 2.99880319603697e-05,
      "loss": 0.2078,
      "step": 2010
    },
    {
      "epoch": 0.06240876194301338,
      "grad_norm": 0.20010523604026445,
      "learning_rate": 2.9987412042824535e-05,
      "loss": 0.2045,
      "step": 2020
    },
    {
      "epoch": 0.062717716210058,
      "grad_norm": 0.18477737555488893,
      "learning_rate": 2.9986776479665414e-05,
      "loss": 0.2062,
      "step": 2030
    },
    {
      "epoch": 0.06302667047710263,
      "grad_norm": 0.18012953650966962,
      "learning_rate": 2.9986125271555814e-05,
      "loss": 0.2035,
      "step": 2040
    },
    {
      "epoch": 0.06333562474414725,
      "grad_norm": 0.20312463120985744,
      "learning_rate": 2.998545841917554e-05,
      "loss": 0.2039,
      "step": 2050
    },
    {
      "epoch": 0.06364457901119187,
      "grad_norm": 0.2067516871530281,
      "learning_rate": 2.9984775923220725e-05,
      "loss": 0.2038,
      "step": 2060
    },
    {
      "epoch": 0.06395353327823648,
      "grad_norm": 0.23804247346031,
      "learning_rate": 2.9984077784403846e-05,
      "loss": 0.2002,
      "step": 2070
    },
    {
      "epoch": 0.0642624875452811,
      "grad_norm": 0.17368067775150903,
      "learning_rate": 2.9983364003453702e-05,
      "loss": 0.2027,
      "step": 2080
    },
    {
      "epoch": 0.06457144181232573,
      "grad_norm": 0.16626358367903837,
      "learning_rate": 2.998263458111541e-05,
      "loss": 0.2015,
      "step": 2090
    },
    {
      "epoch": 0.06488039607937035,
      "grad_norm": 0.27621519975999975,
      "learning_rate": 2.9981889518150445e-05,
      "loss": 0.2067,
      "step": 2100
    },
    {
      "epoch": 0.06518935034641497,
      "grad_norm": 0.21024444701085984,
      "learning_rate": 2.9981128815336574e-05,
      "loss": 0.1996,
      "step": 2110
    },
    {
      "epoch": 0.0654983046134596,
      "grad_norm": 0.24426130276969282,
      "learning_rate": 2.9980352473467915e-05,
      "loss": 0.2037,
      "step": 2120
    },
    {
      "epoch": 0.06580725888050422,
      "grad_norm": 0.1477033633417111,
      "learning_rate": 2.9979560493354907e-05,
      "loss": 0.2004,
      "step": 2130
    },
    {
      "epoch": 0.06611621314754883,
      "grad_norm": 0.24107343107321108,
      "learning_rate": 2.9978752875824302e-05,
      "loss": 0.2036,
      "step": 2140
    },
    {
      "epoch": 0.06642516741459345,
      "grad_norm": 0.2187941774851456,
      "learning_rate": 2.9977929621719197e-05,
      "loss": 0.2047,
      "step": 2150
    },
    {
      "epoch": 0.06673412168163807,
      "grad_norm": 0.1992578320035458,
      "learning_rate": 2.9977090731898995e-05,
      "loss": 0.1994,
      "step": 2160
    },
    {
      "epoch": 0.0670430759486827,
      "grad_norm": 0.19806105265419735,
      "learning_rate": 2.9976236207239426e-05,
      "loss": 0.1976,
      "step": 2170
    },
    {
      "epoch": 0.06735203021572732,
      "grad_norm": 0.17567955418535217,
      "learning_rate": 2.9975366048632544e-05,
      "loss": 0.2,
      "step": 2180
    },
    {
      "epoch": 0.06766098448277194,
      "grad_norm": 0.19530939639808778,
      "learning_rate": 2.9974480256986724e-05,
      "loss": 0.2001,
      "step": 2190
    },
    {
      "epoch": 0.06796993874981656,
      "grad_norm": 0.2044638958382385,
      "learning_rate": 2.997357883322666e-05,
      "loss": 0.2106,
      "step": 2200
    },
    {
      "epoch": 0.06827889301686117,
      "grad_norm": 0.18894945695103943,
      "learning_rate": 2.9972661778293362e-05,
      "loss": 0.2018,
      "step": 2210
    },
    {
      "epoch": 0.0685878472839058,
      "grad_norm": 0.1728822949445438,
      "learning_rate": 2.9971729093144157e-05,
      "loss": 0.1968,
      "step": 2220
    },
    {
      "epoch": 0.06889680155095042,
      "grad_norm": 0.1580606167173133,
      "learning_rate": 2.997078077875269e-05,
      "loss": 0.2067,
      "step": 2230
    },
    {
      "epoch": 0.06920575581799504,
      "grad_norm": 0.5254308545963697,
      "learning_rate": 2.996981683610893e-05,
      "loss": 0.195,
      "step": 2240
    },
    {
      "epoch": 0.06951471008503966,
      "grad_norm": 0.17131874970846256,
      "learning_rate": 2.9968837266219146e-05,
      "loss": 0.2162,
      "step": 2250
    },
    {
      "epoch": 0.06982366435208429,
      "grad_norm": 0.16078626988371186,
      "learning_rate": 2.9967842070105928e-05,
      "loss": 0.1984,
      "step": 2260
    },
    {
      "epoch": 0.07013261861912891,
      "grad_norm": 0.1736114792541168,
      "learning_rate": 2.9966831248808184e-05,
      "loss": 0.2044,
      "step": 2270
    },
    {
      "epoch": 0.07044157288617352,
      "grad_norm": 0.18838468255956162,
      "learning_rate": 2.9965804803381123e-05,
      "loss": 0.1997,
      "step": 2280
    },
    {
      "epoch": 0.07075052715321814,
      "grad_norm": 0.1707422151814035,
      "learning_rate": 2.996476273489627e-05,
      "loss": 0.1988,
      "step": 2290
    },
    {
      "epoch": 0.07105948142026276,
      "grad_norm": 0.17611613049009683,
      "learning_rate": 2.9963705044441453e-05,
      "loss": 0.2102,
      "step": 2300
    },
    {
      "epoch": 0.07136843568730739,
      "grad_norm": 0.1620119857718042,
      "learning_rate": 2.996263173312082e-05,
      "loss": 0.2023,
      "step": 2310
    },
    {
      "epoch": 0.07167738995435201,
      "grad_norm": 0.14850403653592215,
      "learning_rate": 2.9961542802054816e-05,
      "loss": 0.1987,
      "step": 2320
    },
    {
      "epoch": 0.07198634422139663,
      "grad_norm": 0.1567391362200369,
      "learning_rate": 2.9960438252380198e-05,
      "loss": 0.1988,
      "step": 2330
    },
    {
      "epoch": 0.07229529848844125,
      "grad_norm": 0.28124291829683645,
      "learning_rate": 2.995931808525002e-05,
      "loss": 0.2044,
      "step": 2340
    },
    {
      "epoch": 0.07260425275548586,
      "grad_norm": 0.19068209274883757,
      "learning_rate": 2.995818230183364e-05,
      "loss": 0.198,
      "step": 2350
    },
    {
      "epoch": 0.07291320702253049,
      "grad_norm": 0.17692328802750804,
      "learning_rate": 2.995703090331673e-05,
      "loss": 0.1976,
      "step": 2360
    },
    {
      "epoch": 0.07322216128957511,
      "grad_norm": 0.21242379957382648,
      "learning_rate": 2.9955863890901252e-05,
      "loss": 0.1988,
      "step": 2370
    },
    {
      "epoch": 0.07353111555661973,
      "grad_norm": 0.173614714476889,
      "learning_rate": 2.9954681265805465e-05,
      "loss": 0.1986,
      "step": 2380
    },
    {
      "epoch": 0.07384006982366435,
      "grad_norm": 0.1861446361803506,
      "learning_rate": 2.995348302926394e-05,
      "loss": 0.1985,
      "step": 2390
    },
    {
      "epoch": 0.07414902409070898,
      "grad_norm": 0.176519042598455,
      "learning_rate": 2.9952269182527526e-05,
      "loss": 0.2078,
      "step": 2400
    },
    {
      "epoch": 0.0744579783577536,
      "grad_norm": 0.2405258805951056,
      "learning_rate": 2.9951039726863385e-05,
      "loss": 0.2055,
      "step": 2410
    },
    {
      "epoch": 0.07476693262479821,
      "grad_norm": 0.17213732654915595,
      "learning_rate": 2.9949794663554968e-05,
      "loss": 0.1953,
      "step": 2420
    },
    {
      "epoch": 0.07507588689184283,
      "grad_norm": 0.21174539141948398,
      "learning_rate": 2.9948533993902016e-05,
      "loss": 0.2091,
      "step": 2430
    },
    {
      "epoch": 0.07538484115888745,
      "grad_norm": 0.17914977361908388,
      "learning_rate": 2.994725771922056e-05,
      "loss": 0.1972,
      "step": 2440
    },
    {
      "epoch": 0.07569379542593208,
      "grad_norm": 0.18681317713452364,
      "learning_rate": 2.994596584084293e-05,
      "loss": 0.198,
      "step": 2450
    },
    {
      "epoch": 0.0760027496929767,
      "grad_norm": 0.1702246734917539,
      "learning_rate": 2.9944658360117735e-05,
      "loss": 0.1982,
      "step": 2460
    },
    {
      "epoch": 0.07631170396002132,
      "grad_norm": 0.20079090661429203,
      "learning_rate": 2.994333527840988e-05,
      "loss": 0.2015,
      "step": 2470
    },
    {
      "epoch": 0.07662065822706594,
      "grad_norm": 0.16681783867686925,
      "learning_rate": 2.9941996597100553e-05,
      "loss": 0.2036,
      "step": 2480
    },
    {
      "epoch": 0.07692961249411055,
      "grad_norm": 0.23888069273818058,
      "learning_rate": 2.9940642317587222e-05,
      "loss": 0.2121,
      "step": 2490
    },
    {
      "epoch": 0.07723856676115518,
      "grad_norm": 0.15612939807020818,
      "learning_rate": 2.9939272441283643e-05,
      "loss": 0.1957,
      "step": 2500
    },
    {
      "epoch": 0.0775475210281998,
      "grad_norm": 0.15896537672170197,
      "learning_rate": 2.993788696961986e-05,
      "loss": 0.1983,
      "step": 2510
    },
    {
      "epoch": 0.07785647529524442,
      "grad_norm": 0.19879597737685326,
      "learning_rate": 2.993648590404218e-05,
      "loss": 0.2001,
      "step": 2520
    },
    {
      "epoch": 0.07816542956228904,
      "grad_norm": 0.213895580920291,
      "learning_rate": 2.9935069246013204e-05,
      "loss": 0.1956,
      "step": 2530
    },
    {
      "epoch": 0.07847438382933367,
      "grad_norm": 0.18008643894889034,
      "learning_rate": 2.9933636997011812e-05,
      "loss": 0.1942,
      "step": 2540
    },
    {
      "epoch": 0.07878333809637829,
      "grad_norm": 0.17288127839945946,
      "learning_rate": 2.993218915853314e-05,
      "loss": 0.1944,
      "step": 2550
    },
    {
      "epoch": 0.0790922923634229,
      "grad_norm": 0.19874886444946327,
      "learning_rate": 2.993072573208862e-05,
      "loss": 0.1977,
      "step": 2560
    },
    {
      "epoch": 0.07940124663046752,
      "grad_norm": 0.20028131905271443,
      "learning_rate": 2.992924671920595e-05,
      "loss": 0.1969,
      "step": 2570
    },
    {
      "epoch": 0.07971020089751214,
      "grad_norm": 0.22741761017119985,
      "learning_rate": 2.9927752121429085e-05,
      "loss": 0.1986,
      "step": 2580
    },
    {
      "epoch": 0.08001915516455677,
      "grad_norm": 0.14892790240624842,
      "learning_rate": 2.9926241940318275e-05,
      "loss": 0.1989,
      "step": 2590
    },
    {
      "epoch": 0.08032810943160139,
      "grad_norm": 0.1812090530263701,
      "learning_rate": 2.9924716177450013e-05,
      "loss": 0.2007,
      "step": 2600
    },
    {
      "epoch": 0.08063706369864601,
      "grad_norm": 0.21981022506889003,
      "learning_rate": 2.9923174834417073e-05,
      "loss": 0.1992,
      "step": 2610
    },
    {
      "epoch": 0.08094601796569063,
      "grad_norm": 0.1806725892016268,
      "learning_rate": 2.992161791282849e-05,
      "loss": 0.1985,
      "step": 2620
    },
    {
      "epoch": 0.08125497223273524,
      "grad_norm": 0.152449578467758,
      "learning_rate": 2.9920045414309562e-05,
      "loss": 0.1997,
      "step": 2630
    },
    {
      "epoch": 0.08156392649977987,
      "grad_norm": 0.16967064167562734,
      "learning_rate": 2.991845734050184e-05,
      "loss": 0.2003,
      "step": 2640
    },
    {
      "epoch": 0.08187288076682449,
      "grad_norm": 0.15273183917648353,
      "learning_rate": 2.991685369306315e-05,
      "loss": 0.199,
      "step": 2650
    },
    {
      "epoch": 0.08218183503386911,
      "grad_norm": 0.1785418164488158,
      "learning_rate": 2.991523447366756e-05,
      "loss": 0.2065,
      "step": 2660
    },
    {
      "epoch": 0.08249078930091373,
      "grad_norm": 0.19119076782838293,
      "learning_rate": 2.991359968400541e-05,
      "loss": 0.2043,
      "step": 2670
    },
    {
      "epoch": 0.08279974356795836,
      "grad_norm": 0.12963083951582324,
      "learning_rate": 2.991194932578327e-05,
      "loss": 0.193,
      "step": 2680
    },
    {
      "epoch": 0.08310869783500298,
      "grad_norm": 0.23260798475158634,
      "learning_rate": 2.9910283400723987e-05,
      "loss": 0.1975,
      "step": 2690
    },
    {
      "epoch": 0.08341765210204759,
      "grad_norm": 0.1453149378730413,
      "learning_rate": 2.9908601910566645e-05,
      "loss": 0.196,
      "step": 2700
    },
    {
      "epoch": 0.08372660636909221,
      "grad_norm": 0.19381474156061762,
      "learning_rate": 2.9906904857066585e-05,
      "loss": 0.2085,
      "step": 2710
    },
    {
      "epoch": 0.08403556063613683,
      "grad_norm": 0.15613430326411726,
      "learning_rate": 2.9905192241995386e-05,
      "loss": 0.192,
      "step": 2720
    },
    {
      "epoch": 0.08434451490318146,
      "grad_norm": 0.15517037231213784,
      "learning_rate": 2.990346406714088e-05,
      "loss": 0.1962,
      "step": 2730
    },
    {
      "epoch": 0.08465346917022608,
      "grad_norm": 0.19071627886237663,
      "learning_rate": 2.9901720334307126e-05,
      "loss": 0.1993,
      "step": 2740
    },
    {
      "epoch": 0.0849624234372707,
      "grad_norm": 0.18480484319556262,
      "learning_rate": 2.9899961045314455e-05,
      "loss": 0.2028,
      "step": 2750
    },
    {
      "epoch": 0.08527137770431532,
      "grad_norm": 0.18673748288484698,
      "learning_rate": 2.9898186201999404e-05,
      "loss": 0.194,
      "step": 2760
    },
    {
      "epoch": 0.08558033197135993,
      "grad_norm": 0.19721048721852688,
      "learning_rate": 2.9896395806214765e-05,
      "loss": 0.1937,
      "step": 2770
    },
    {
      "epoch": 0.08588928623840456,
      "grad_norm": 0.16917472375765735,
      "learning_rate": 2.9894589859829565e-05,
      "loss": 0.2035,
      "step": 2780
    },
    {
      "epoch": 0.08619824050544918,
      "grad_norm": 0.15929564098033613,
      "learning_rate": 2.9892768364729062e-05,
      "loss": 0.1959,
      "step": 2790
    },
    {
      "epoch": 0.0865071947724938,
      "grad_norm": 0.1564281325302,
      "learning_rate": 2.9890931322814746e-05,
      "loss": 0.1985,
      "step": 2800
    },
    {
      "epoch": 0.08681614903953842,
      "grad_norm": 0.18083462652656238,
      "learning_rate": 2.9889078736004337e-05,
      "loss": 0.2054,
      "step": 2810
    },
    {
      "epoch": 0.08712510330658305,
      "grad_norm": 0.15347293774038326,
      "learning_rate": 2.9887210606231775e-05,
      "loss": 0.1976,
      "step": 2820
    },
    {
      "epoch": 0.08743405757362767,
      "grad_norm": 0.14369548444314065,
      "learning_rate": 2.988532693544724e-05,
      "loss": 0.1965,
      "step": 2830
    },
    {
      "epoch": 0.08774301184067229,
      "grad_norm": 0.16755897503768843,
      "learning_rate": 2.9883427725617127e-05,
      "loss": 0.1923,
      "step": 2840
    },
    {
      "epoch": 0.0880519661077169,
      "grad_norm": 0.1766933017480405,
      "learning_rate": 2.9881512978724054e-05,
      "loss": 0.1999,
      "step": 2850
    },
    {
      "epoch": 0.08836092037476152,
      "grad_norm": 0.16335478491515748,
      "learning_rate": 2.987958269676685e-05,
      "loss": 0.196,
      "step": 2860
    },
    {
      "epoch": 0.08866987464180615,
      "grad_norm": 0.1412388792891497,
      "learning_rate": 2.9877636881760578e-05,
      "loss": 0.1968,
      "step": 2870
    },
    {
      "epoch": 0.08897882890885077,
      "grad_norm": 0.28855040940828436,
      "learning_rate": 2.9875675535736505e-05,
      "loss": 0.2192,
      "step": 2880
    },
    {
      "epoch": 0.08928778317589539,
      "grad_norm": 0.19016910033969708,
      "learning_rate": 2.9873698660742107e-05,
      "loss": 0.1973,
      "step": 2890
    },
    {
      "epoch": 0.08959673744294001,
      "grad_norm": 0.19178278954952716,
      "learning_rate": 2.987170625884109e-05,
      "loss": 0.2064,
      "step": 2900
    },
    {
      "epoch": 0.08990569170998464,
      "grad_norm": 0.15224503527765604,
      "learning_rate": 2.9869698332113346e-05,
      "loss": 0.1946,
      "step": 2910
    },
    {
      "epoch": 0.09021464597702925,
      "grad_norm": 0.13514359465615594,
      "learning_rate": 2.9867674882654983e-05,
      "loss": 0.193,
      "step": 2920
    },
    {
      "epoch": 0.09052360024407387,
      "grad_norm": 0.15927437454216875,
      "learning_rate": 2.9865635912578325e-05,
      "loss": 0.1944,
      "step": 2930
    },
    {
      "epoch": 0.09083255451111849,
      "grad_norm": 0.1412858450776157,
      "learning_rate": 2.9863581424011874e-05,
      "loss": 0.1954,
      "step": 2940
    },
    {
      "epoch": 0.09114150877816311,
      "grad_norm": 0.1692424504947583,
      "learning_rate": 2.9861511419100358e-05,
      "loss": 0.197,
      "step": 2950
    },
    {
      "epoch": 0.09145046304520774,
      "grad_norm": 0.15945249012427518,
      "learning_rate": 2.985942590000468e-05,
      "loss": 0.1987,
      "step": 2960
    },
    {
      "epoch": 0.09175941731225236,
      "grad_norm": 0.1509003969152127,
      "learning_rate": 2.9857324868901958e-05,
      "loss": 0.1961,
      "step": 2970
    },
    {
      "epoch": 0.09206837157929698,
      "grad_norm": 0.16215765059346146,
      "learning_rate": 2.9855208327985486e-05,
      "loss": 0.1997,
      "step": 2980
    },
    {
      "epoch": 0.09237732584634159,
      "grad_norm": 0.21205919686914562,
      "learning_rate": 2.985307627946476e-05,
      "loss": 0.2031,
      "step": 2990
    },
    {
      "epoch": 0.09268628011338621,
      "grad_norm": 0.18692914385469056,
      "learning_rate": 2.9850928725565464e-05,
      "loss": 0.1932,
      "step": 3000
    },
    {
      "epoch": 0.09299523438043084,
      "grad_norm": 0.1383747544451328,
      "learning_rate": 2.9848765668529463e-05,
      "loss": 0.1947,
      "step": 3010
    },
    {
      "epoch": 0.09330418864747546,
      "grad_norm": 0.17240403540350424,
      "learning_rate": 2.984658711061481e-05,
      "loss": 0.1981,
      "step": 3020
    },
    {
      "epoch": 0.09361314291452008,
      "grad_norm": 0.13524872029986645,
      "learning_rate": 2.9844393054095738e-05,
      "loss": 0.195,
      "step": 3030
    },
    {
      "epoch": 0.0939220971815647,
      "grad_norm": 0.16721476843562097,
      "learning_rate": 2.9842183501262664e-05,
      "loss": 0.1993,
      "step": 3040
    },
    {
      "epoch": 0.09423105144860933,
      "grad_norm": 0.1781358051524646,
      "learning_rate": 2.9839958454422165e-05,
      "loss": 0.1933,
      "step": 3050
    },
    {
      "epoch": 0.09454000571565394,
      "grad_norm": 0.15964728077303253,
      "learning_rate": 2.983771791589702e-05,
      "loss": 0.1956,
      "step": 3060
    },
    {
      "epoch": 0.09484895998269856,
      "grad_norm": 0.14892102028856208,
      "learning_rate": 2.9835461888026163e-05,
      "loss": 0.1952,
      "step": 3070
    },
    {
      "epoch": 0.09515791424974318,
      "grad_norm": 0.16357065961104184,
      "learning_rate": 2.9833190373164692e-05,
      "loss": 0.1942,
      "step": 3080
    },
    {
      "epoch": 0.0954668685167878,
      "grad_norm": 0.22398475313678176,
      "learning_rate": 2.9830903373683883e-05,
      "loss": 0.2065,
      "step": 3090
    },
    {
      "epoch": 0.09577582278383243,
      "grad_norm": 0.14766190514738772,
      "learning_rate": 2.982860089197118e-05,
      "loss": 0.2053,
      "step": 3100
    },
    {
      "epoch": 0.09608477705087705,
      "grad_norm": 0.3349719693911538,
      "learning_rate": 2.9826282930430175e-05,
      "loss": 0.2007,
      "step": 3110
    },
    {
      "epoch": 0.09639373131792167,
      "grad_norm": 0.1644366645964322,
      "learning_rate": 2.982394949148063e-05,
      "loss": 0.1938,
      "step": 3120
    },
    {
      "epoch": 0.09670268558496628,
      "grad_norm": 0.14493671521828133,
      "learning_rate": 2.9821600577558457e-05,
      "loss": 0.2013,
      "step": 3130
    },
    {
      "epoch": 0.0970116398520109,
      "grad_norm": 0.11524631469635575,
      "learning_rate": 2.9819236191115738e-05,
      "loss": 0.1967,
      "step": 3140
    },
    {
      "epoch": 0.09732059411905553,
      "grad_norm": 0.16223379075585082,
      "learning_rate": 2.9816856334620686e-05,
      "loss": 0.2079,
      "step": 3150
    },
    {
      "epoch": 0.09762954838610015,
      "grad_norm": 0.15651559888502994,
      "learning_rate": 2.9814461010557683e-05,
      "loss": 0.2084,
      "step": 3160
    },
    {
      "epoch": 0.09793850265314477,
      "grad_norm": 0.16731616566868066,
      "learning_rate": 2.981205022142724e-05,
      "loss": 0.2032,
      "step": 3170
    },
    {
      "epoch": 0.0982474569201894,
      "grad_norm": 0.161127910827826,
      "learning_rate": 2.9809623969746022e-05,
      "loss": 0.1942,
      "step": 3180
    },
    {
      "epoch": 0.09855641118723402,
      "grad_norm": 0.14089901940734195,
      "learning_rate": 2.9807182258046838e-05,
      "loss": 0.1959,
      "step": 3190
    },
    {
      "epoch": 0.09886536545427863,
      "grad_norm": 0.13991812367182896,
      "learning_rate": 2.980472508887863e-05,
      "loss": 0.1938,
      "step": 3200
    },
    {
      "epoch": 0.09917431972132325,
      "grad_norm": 0.16151724527912023,
      "learning_rate": 2.9802252464806477e-05,
      "loss": 0.2024,
      "step": 3210
    },
    {
      "epoch": 0.09948327398836787,
      "grad_norm": 0.17232581708425088,
      "learning_rate": 2.9799764388411597e-05,
      "loss": 0.2027,
      "step": 3220
    },
    {
      "epoch": 0.0997922282554125,
      "grad_norm": 0.15221857389640467,
      "learning_rate": 2.9797260862291332e-05,
      "loss": 0.2027,
      "step": 3230
    },
    {
      "epoch": 0.10010118252245712,
      "grad_norm": 0.17076789497687467,
      "learning_rate": 2.9794741889059154e-05,
      "loss": 0.1972,
      "step": 3240
    },
    {
      "epoch": 0.10041013678950174,
      "grad_norm": 0.14404182419615677,
      "learning_rate": 2.9792207471344675e-05,
      "loss": 0.1919,
      "step": 3250
    },
    {
      "epoch": 0.10071909105654636,
      "grad_norm": 0.16945885121142765,
      "learning_rate": 2.9789657611793598e-05,
      "loss": 0.2098,
      "step": 3260
    },
    {
      "epoch": 0.10102804532359097,
      "grad_norm": 0.16013235477830412,
      "learning_rate": 2.9787092313067775e-05,
      "loss": 0.196,
      "step": 3270
    },
    {
      "epoch": 0.10133699959063559,
      "grad_norm": 0.14813863429455534,
      "learning_rate": 2.9784511577845164e-05,
      "loss": 0.2013,
      "step": 3280
    },
    {
      "epoch": 0.10164595385768022,
      "grad_norm": 0.1590236946020729,
      "learning_rate": 2.9781915408819833e-05,
      "loss": 0.1977,
      "step": 3290
    },
    {
      "epoch": 0.10195490812472484,
      "grad_norm": 0.16008756310142963,
      "learning_rate": 2.9779303808701975e-05,
      "loss": 0.202,
      "step": 3300
    },
    {
      "epoch": 0.10226386239176946,
      "grad_norm": 0.14373767651911806,
      "learning_rate": 2.977667678021788e-05,
      "loss": 0.1959,
      "step": 3310
    },
    {
      "epoch": 0.10257281665881408,
      "grad_norm": 0.1410824571705092,
      "learning_rate": 2.9774034326109948e-05,
      "loss": 0.1916,
      "step": 3320
    },
    {
      "epoch": 0.1028817709258587,
      "grad_norm": 0.136187184514505,
      "learning_rate": 2.977137644913668e-05,
      "loss": 0.2026,
      "step": 3330
    },
    {
      "epoch": 0.10319072519290332,
      "grad_norm": 0.1800698776121755,
      "learning_rate": 2.9768703152072684e-05,
      "loss": 0.1949,
      "step": 3340
    },
    {
      "epoch": 0.10349967945994794,
      "grad_norm": 0.1538791403546913,
      "learning_rate": 2.9766014437708654e-05,
      "loss": 0.2025,
      "step": 3350
    },
    {
      "epoch": 0.10380863372699256,
      "grad_norm": 0.1568382831737684,
      "learning_rate": 2.9763310308851388e-05,
      "loss": 0.1953,
      "step": 3360
    },
    {
      "epoch": 0.10411758799403718,
      "grad_norm": 0.13122346416217745,
      "learning_rate": 2.9760590768323766e-05,
      "loss": 0.2014,
      "step": 3370
    },
    {
      "epoch": 0.1044265422610818,
      "grad_norm": 0.1806485020747406,
      "learning_rate": 2.9757855818964775e-05,
      "loss": 0.1971,
      "step": 3380
    },
    {
      "epoch": 0.10473549652812643,
      "grad_norm": 0.13537590653822115,
      "learning_rate": 2.9755105463629463e-05,
      "loss": 0.2065,
      "step": 3390
    },
    {
      "epoch": 0.10504445079517105,
      "grad_norm": 0.13595042959879605,
      "learning_rate": 2.975233970518898e-05,
      "loss": 0.1992,
      "step": 3400
    },
    {
      "epoch": 0.10535340506221566,
      "grad_norm": 0.1509154648140987,
      "learning_rate": 2.9749558546530542e-05,
      "loss": 0.1979,
      "step": 3410
    },
    {
      "epoch": 0.10566235932926028,
      "grad_norm": 0.15428441505969218,
      "learning_rate": 2.9746761990557453e-05,
      "loss": 0.2083,
      "step": 3420
    },
    {
      "epoch": 0.1059713135963049,
      "grad_norm": 0.1397814921074192,
      "learning_rate": 2.9743950040189086e-05,
      "loss": 0.2036,
      "step": 3430
    },
    {
      "epoch": 0.10628026786334953,
      "grad_norm": 0.14912878243829816,
      "learning_rate": 2.974112269836088e-05,
      "loss": 0.1956,
      "step": 3440
    },
    {
      "epoch": 0.10658922213039415,
      "grad_norm": 0.16469373458469788,
      "learning_rate": 2.9738279968024345e-05,
      "loss": 0.1947,
      "step": 3450
    },
    {
      "epoch": 0.10689817639743877,
      "grad_norm": 0.14096722380444265,
      "learning_rate": 2.9735421852147062e-05,
      "loss": 0.1955,
      "step": 3460
    },
    {
      "epoch": 0.1072071306644834,
      "grad_norm": 0.1712139882570602,
      "learning_rate": 2.9732548353712653e-05,
      "loss": 0.1977,
      "step": 3470
    },
    {
      "epoch": 0.107516084931528,
      "grad_norm": 0.182036434354717,
      "learning_rate": 2.972965947572083e-05,
      "loss": 0.1984,
      "step": 3480
    },
    {
      "epoch": 0.10782503919857263,
      "grad_norm": 0.12685205273511865,
      "learning_rate": 2.9726755221187325e-05,
      "loss": 0.194,
      "step": 3490
    },
    {
      "epoch": 0.10813399346561725,
      "grad_norm": 0.19036499223183592,
      "learning_rate": 2.972383559314395e-05,
      "loss": 0.1929,
      "step": 3500
    },
    {
      "epoch": 0.10844294773266187,
      "grad_norm": 0.18080925676285772,
      "learning_rate": 2.972090059463855e-05,
      "loss": 0.1979,
      "step": 3510
    },
    {
      "epoch": 0.1087519019997065,
      "grad_norm": 0.1607497216955787,
      "learning_rate": 2.971795022873502e-05,
      "loss": 0.2002,
      "step": 3520
    },
    {
      "epoch": 0.10906085626675112,
      "grad_norm": 0.14846714634409408,
      "learning_rate": 2.9714984498513298e-05,
      "loss": 0.2,
      "step": 3530
    },
    {
      "epoch": 0.10936981053379574,
      "grad_norm": 0.22370896372238852,
      "learning_rate": 2.9712003407069355e-05,
      "loss": 0.2133,
      "step": 3540
    },
    {
      "epoch": 0.10967876480084035,
      "grad_norm": 0.14800105867230223,
      "learning_rate": 2.970900695751521e-05,
      "loss": 0.2006,
      "step": 3550
    },
    {
      "epoch": 0.10998771906788497,
      "grad_norm": 0.15815792317396923,
      "learning_rate": 2.9705995152978907e-05,
      "loss": 0.1983,
      "step": 3560
    },
    {
      "epoch": 0.1102966733349296,
      "grad_norm": 0.18564579888879634,
      "learning_rate": 2.9702967996604512e-05,
      "loss": 0.1941,
      "step": 3570
    },
    {
      "epoch": 0.11060562760197422,
      "grad_norm": 0.9104975551943203,
      "learning_rate": 2.9699925491552137e-05,
      "loss": 0.2026,
      "step": 3580
    },
    {
      "epoch": 0.11091458186901884,
      "grad_norm": 0.17333001775135767,
      "learning_rate": 2.9696867640997895e-05,
      "loss": 0.4843,
      "step": 3590
    },
    {
      "epoch": 0.11122353613606346,
      "grad_norm": 0.15860053694459797,
      "learning_rate": 2.9693794448133928e-05,
      "loss": 0.1914,
      "step": 3600
    },
    {
      "epoch": 0.11153249040310809,
      "grad_norm": 0.1649188831618067,
      "learning_rate": 2.96907059161684e-05,
      "loss": 0.1986,
      "step": 3610
    },
    {
      "epoch": 0.1118414446701527,
      "grad_norm": 0.17291101908747689,
      "learning_rate": 2.9687602048325485e-05,
      "loss": 0.1991,
      "step": 3620
    },
    {
      "epoch": 0.11215039893719732,
      "grad_norm": 0.14504924285922557,
      "learning_rate": 2.9684482847845354e-05,
      "loss": 0.1965,
      "step": 3630
    },
    {
      "epoch": 0.11245935320424194,
      "grad_norm": 0.15988131838749894,
      "learning_rate": 2.96813483179842e-05,
      "loss": 0.198,
      "step": 3640
    },
    {
      "epoch": 0.11276830747128656,
      "grad_norm": 0.24942612461926558,
      "learning_rate": 2.9678198462014207e-05,
      "loss": 0.2018,
      "step": 3650
    },
    {
      "epoch": 0.11307726173833119,
      "grad_norm": 0.15620629317438656,
      "learning_rate": 2.9675033283223565e-05,
      "loss": 0.2005,
      "step": 3660
    },
    {
      "epoch": 0.11338621600537581,
      "grad_norm": 0.17244272644487615,
      "learning_rate": 2.9671852784916455e-05,
      "loss": 0.1994,
      "step": 3670
    },
    {
      "epoch": 0.11369517027242043,
      "grad_norm": 0.156712293447336,
      "learning_rate": 2.9668656970413063e-05,
      "loss": 0.1956,
      "step": 3680
    },
    {
      "epoch": 0.11400412453946504,
      "grad_norm": 0.13623076388298133,
      "learning_rate": 2.9665445843049543e-05,
      "loss": 0.1938,
      "step": 3690
    },
    {
      "epoch": 0.11431307880650966,
      "grad_norm": 0.1521454012963752,
      "learning_rate": 2.9662219406178055e-05,
      "loss": 0.1957,
      "step": 3700
    },
    {
      "epoch": 0.11462203307355429,
      "grad_norm": 0.17226312492776039,
      "learning_rate": 2.965897766316672e-05,
      "loss": 0.195,
      "step": 3710
    },
    {
      "epoch": 0.11493098734059891,
      "grad_norm": 0.17103571729681852,
      "learning_rate": 2.965572061739966e-05,
      "loss": 0.1917,
      "step": 3720
    },
    {
      "epoch": 0.11523994160764353,
      "grad_norm": 0.14787997443801582,
      "learning_rate": 2.965244827227695e-05,
      "loss": 0.1988,
      "step": 3730
    },
    {
      "epoch": 0.11554889587468815,
      "grad_norm": 0.16499617818184958,
      "learning_rate": 2.9649160631214656e-05,
      "loss": 0.1934,
      "step": 3740
    },
    {
      "epoch": 0.11585785014173278,
      "grad_norm": 0.13859267641190945,
      "learning_rate": 2.96458576976448e-05,
      "loss": 0.1992,
      "step": 3750
    },
    {
      "epoch": 0.11616680440877739,
      "grad_norm": 0.13925810169222835,
      "learning_rate": 2.9642539475015368e-05,
      "loss": 0.1974,
      "step": 3760
    },
    {
      "epoch": 0.11647575867582201,
      "grad_norm": 0.14590521946593527,
      "learning_rate": 2.9639205966790304e-05,
      "loss": 0.1952,
      "step": 3770
    },
    {
      "epoch": 0.11678471294286663,
      "grad_norm": 0.15328983346122252,
      "learning_rate": 2.963585717644952e-05,
      "loss": 0.1937,
      "step": 3780
    },
    {
      "epoch": 0.11709366720991125,
      "grad_norm": 0.1704191032635659,
      "learning_rate": 2.963249310748888e-05,
      "loss": 0.207,
      "step": 3790
    },
    {
      "epoch": 0.11740262147695588,
      "grad_norm": 0.16383535566426977,
      "learning_rate": 2.9629113763420187e-05,
      "loss": 0.1938,
      "step": 3800
    },
    {
      "epoch": 0.1177115757440005,
      "grad_norm": 0.14359121960553015,
      "learning_rate": 2.9625719147771188e-05,
      "loss": 0.201,
      "step": 3810
    },
    {
      "epoch": 0.11802053001104512,
      "grad_norm": 0.14131405219953524,
      "learning_rate": 2.9622309264085595e-05,
      "loss": 0.1927,
      "step": 3820
    },
    {
      "epoch": 0.11832948427808973,
      "grad_norm": 0.16335041544288204,
      "learning_rate": 2.9618884115923033e-05,
      "loss": 0.1952,
      "step": 3830
    },
    {
      "epoch": 0.11863843854513435,
      "grad_norm": 0.2502348189666573,
      "learning_rate": 2.961544370685908e-05,
      "loss": 0.2034,
      "step": 3840
    },
    {
      "epoch": 0.11894739281217898,
      "grad_norm": 0.17031687373001325,
      "learning_rate": 2.961198804048523e-05,
      "loss": 0.2008,
      "step": 3850
    },
    {
      "epoch": 0.1192563470792236,
      "grad_norm": 0.14658450456524985,
      "learning_rate": 2.9608517120408918e-05,
      "loss": 0.2015,
      "step": 3860
    },
    {
      "epoch": 0.11956530134626822,
      "grad_norm": 0.14591308196539304,
      "learning_rate": 2.960503095025349e-05,
      "loss": 0.1904,
      "step": 3870
    },
    {
      "epoch": 0.11987425561331284,
      "grad_norm": 0.13427534668219537,
      "learning_rate": 2.960152953365822e-05,
      "loss": 0.2046,
      "step": 3880
    },
    {
      "epoch": 0.12018320988035747,
      "grad_norm": 0.15168502466304842,
      "learning_rate": 2.95980128742783e-05,
      "loss": 0.1979,
      "step": 3890
    },
    {
      "epoch": 0.12049216414740208,
      "grad_norm": 0.14442317575764707,
      "learning_rate": 2.9594480975784833e-05,
      "loss": 0.2129,
      "step": 3900
    },
    {
      "epoch": 0.1208011184144467,
      "grad_norm": 0.1777783043137642,
      "learning_rate": 2.9590933841864815e-05,
      "loss": 0.2052,
      "step": 3910
    },
    {
      "epoch": 0.12111007268149132,
      "grad_norm": 0.14281818732894197,
      "learning_rate": 2.958737147622117e-05,
      "loss": 0.2001,
      "step": 3920
    },
    {
      "epoch": 0.12141902694853594,
      "grad_norm": 0.1354700247916489,
      "learning_rate": 2.9583793882572715e-05,
      "loss": 0.1987,
      "step": 3930
    },
    {
      "epoch": 0.12172798121558057,
      "grad_norm": 0.17757910288830733,
      "learning_rate": 2.9580201064654148e-05,
      "loss": 0.1993,
      "step": 3940
    },
    {
      "epoch": 0.12203693548262519,
      "grad_norm": 0.14055138458099284,
      "learning_rate": 2.957659302621608e-05,
      "loss": 0.1961,
      "step": 3950
    },
    {
      "epoch": 0.12234588974966981,
      "grad_norm": 0.1419351708319278,
      "learning_rate": 2.957296977102501e-05,
      "loss": 0.1925,
      "step": 3960
    },
    {
      "epoch": 0.12265484401671442,
      "grad_norm": 0.1296616328956848,
      "learning_rate": 2.9569331302863305e-05,
      "loss": 0.1991,
      "step": 3970
    },
    {
      "epoch": 0.12296379828375904,
      "grad_norm": 0.14738144874164277,
      "learning_rate": 2.9565677625529225e-05,
      "loss": 0.1906,
      "step": 3980
    },
    {
      "epoch": 0.12327275255080367,
      "grad_norm": 0.13274251994537717,
      "learning_rate": 2.956200874283691e-05,
      "loss": 0.2065,
      "step": 3990
    },
    {
      "epoch": 0.12358170681784829,
      "grad_norm": 0.18168042419561298,
      "learning_rate": 2.955832465861637e-05,
      "loss": 0.1918,
      "step": 4000
    },
    {
      "epoch": 0.12389066108489291,
      "grad_norm": 0.2078814909170653,
      "learning_rate": 2.9554625376713477e-05,
      "loss": 0.203,
      "step": 4010
    },
    {
      "epoch": 0.12419961535193753,
      "grad_norm": 0.15436707123570834,
      "learning_rate": 2.9550910900989982e-05,
      "loss": 0.1918,
      "step": 4020
    },
    {
      "epoch": 0.12450856961898216,
      "grad_norm": 0.15422646051605707,
      "learning_rate": 2.9547181235323485e-05,
      "loss": 0.194,
      "step": 4030
    },
    {
      "epoch": 0.12481752388602677,
      "grad_norm": 0.13549992108306505,
      "learning_rate": 2.9543436383607447e-05,
      "loss": 0.1919,
      "step": 4040
    },
    {
      "epoch": 0.1251264781530714,
      "grad_norm": 0.14308198767398161,
      "learning_rate": 2.953967634975118e-05,
      "loss": 0.1994,
      "step": 4050
    },
    {
      "epoch": 0.125435432420116,
      "grad_norm": 0.20628750095306617,
      "learning_rate": 2.9535901137679848e-05,
      "loss": 0.2066,
      "step": 4060
    },
    {
      "epoch": 0.12574438668716062,
      "grad_norm": 0.16292002289397242,
      "learning_rate": 2.953211075133447e-05,
      "loss": 0.2044,
      "step": 4070
    },
    {
      "epoch": 0.12605334095420526,
      "grad_norm": 0.16507262758091465,
      "learning_rate": 2.9528305194671878e-05,
      "loss": 0.1915,
      "step": 4080
    },
    {
      "epoch": 0.12636229522124987,
      "grad_norm": 0.13034214436998967,
      "learning_rate": 2.9524484471664765e-05,
      "loss": 0.2021,
      "step": 4090
    },
    {
      "epoch": 0.1266712494882945,
      "grad_norm": 0.14593293004873178,
      "learning_rate": 2.9520648586301652e-05,
      "loss": 0.2032,
      "step": 4100
    },
    {
      "epoch": 0.1269802037553391,
      "grad_norm": 0.12500323412729464,
      "learning_rate": 2.951679754258688e-05,
      "loss": 0.1936,
      "step": 4110
    },
    {
      "epoch": 0.12728915802238375,
      "grad_norm": 0.1400278042626436,
      "learning_rate": 2.9512931344540617e-05,
      "loss": 0.1976,
      "step": 4120
    },
    {
      "epoch": 0.12759811228942836,
      "grad_norm": 0.16049356004715562,
      "learning_rate": 2.9509049996198863e-05,
      "loss": 0.1941,
      "step": 4130
    },
    {
      "epoch": 0.12790706655647296,
      "grad_norm": 0.1484062143138434,
      "learning_rate": 2.9505153501613407e-05,
      "loss": 0.1953,
      "step": 4140
    },
    {
      "epoch": 0.1282160208235176,
      "grad_norm": 0.14015163193632005,
      "learning_rate": 2.950124186485188e-05,
      "loss": 0.193,
      "step": 4150
    },
    {
      "epoch": 0.1285249750905622,
      "grad_norm": 0.1231425362581919,
      "learning_rate": 2.9497315089997702e-05,
      "loss": 0.2006,
      "step": 4160
    },
    {
      "epoch": 0.12883392935760685,
      "grad_norm": 0.12528784539325993,
      "learning_rate": 2.9493373181150106e-05,
      "loss": 0.1948,
      "step": 4170
    },
    {
      "epoch": 0.12914288362465146,
      "grad_norm": 0.11966360889407414,
      "learning_rate": 2.9489416142424104e-05,
      "loss": 0.2001,
      "step": 4180
    },
    {
      "epoch": 0.1294518378916961,
      "grad_norm": 0.14438982807807982,
      "learning_rate": 2.948544397795052e-05,
      "loss": 0.1942,
      "step": 4190
    },
    {
      "epoch": 0.1297607921587407,
      "grad_norm": 0.13507913856323525,
      "learning_rate": 2.948145669187597e-05,
      "loss": 0.1988,
      "step": 4200
    },
    {
      "epoch": 0.1300697464257853,
      "grad_norm": 0.13072933322754932,
      "learning_rate": 2.9477454288362847e-05,
      "loss": 0.2029,
      "step": 4210
    },
    {
      "epoch": 0.13037870069282995,
      "grad_norm": 0.1346665945344133,
      "learning_rate": 2.947343677158933e-05,
      "loss": 0.1977,
      "step": 4220
    },
    {
      "epoch": 0.13068765495987456,
      "grad_norm": 0.14419031028708462,
      "learning_rate": 2.9469404145749366e-05,
      "loss": 0.195,
      "step": 4230
    },
    {
      "epoch": 0.1309966092269192,
      "grad_norm": 0.15289759763473756,
      "learning_rate": 2.9465356415052687e-05,
      "loss": 0.1933,
      "step": 4240
    },
    {
      "epoch": 0.1313055634939638,
      "grad_norm": 0.14879122180002222,
      "learning_rate": 2.946129358372479e-05,
      "loss": 0.1928,
      "step": 4250
    },
    {
      "epoch": 0.13161451776100844,
      "grad_norm": 0.1233182967420781,
      "learning_rate": 2.9457215656006927e-05,
      "loss": 0.1913,
      "step": 4260
    },
    {
      "epoch": 0.13192347202805305,
      "grad_norm": 0.13905927812327482,
      "learning_rate": 2.945312263615612e-05,
      "loss": 0.2008,
      "step": 4270
    },
    {
      "epoch": 0.13223242629509765,
      "grad_norm": 0.1480347048625918,
      "learning_rate": 2.944901452844514e-05,
      "loss": 0.1922,
      "step": 4280
    },
    {
      "epoch": 0.1325413805621423,
      "grad_norm": 0.1759649098608902,
      "learning_rate": 2.9444891337162517e-05,
      "loss": 0.1962,
      "step": 4290
    },
    {
      "epoch": 0.1328503348291869,
      "grad_norm": 0.15444515511292223,
      "learning_rate": 2.944075306661251e-05,
      "loss": 0.202,
      "step": 4300
    },
    {
      "epoch": 0.13315928909623154,
      "grad_norm": 0.17237129521975675,
      "learning_rate": 2.9436599721115143e-05,
      "loss": 0.1935,
      "step": 4310
    },
    {
      "epoch": 0.13346824336327615,
      "grad_norm": 0.12984196981989796,
      "learning_rate": 2.9432431305006147e-05,
      "loss": 0.194,
      "step": 4320
    },
    {
      "epoch": 0.13377719763032078,
      "grad_norm": 0.13269341815635174,
      "learning_rate": 2.9428247822637015e-05,
      "loss": 0.203,
      "step": 4330
    },
    {
      "epoch": 0.1340861518973654,
      "grad_norm": 0.15149059382961613,
      "learning_rate": 2.942404927837495e-05,
      "loss": 0.191,
      "step": 4340
    },
    {
      "epoch": 0.13439510616441,
      "grad_norm": 0.1148468815620595,
      "learning_rate": 2.9419835676602893e-05,
      "loss": 0.1937,
      "step": 4350
    },
    {
      "epoch": 0.13470406043145464,
      "grad_norm": 0.13918237186519886,
      "learning_rate": 2.9415607021719482e-05,
      "loss": 0.2023,
      "step": 4360
    },
    {
      "epoch": 0.13501301469849925,
      "grad_norm": 0.1335764168362304,
      "learning_rate": 2.9411363318139083e-05,
      "loss": 0.2014,
      "step": 4370
    },
    {
      "epoch": 0.13532196896554388,
      "grad_norm": 0.1228777211146217,
      "learning_rate": 2.940710457029178e-05,
      "loss": 0.2,
      "step": 4380
    },
    {
      "epoch": 0.1356309232325885,
      "grad_norm": 0.13309593839884942,
      "learning_rate": 2.940283078262334e-05,
      "loss": 0.1916,
      "step": 4390
    },
    {
      "epoch": 0.13593987749963313,
      "grad_norm": 0.12294257730806753,
      "learning_rate": 2.939854195959525e-05,
      "loss": 0.199,
      "step": 4400
    },
    {
      "epoch": 0.13624883176667774,
      "grad_norm": 0.14335373300838533,
      "learning_rate": 2.939423810568468e-05,
      "loss": 0.1966,
      "step": 4410
    },
    {
      "epoch": 0.13655778603372234,
      "grad_norm": 0.12813785744442807,
      "learning_rate": 2.938991922538451e-05,
      "loss": 0.1926,
      "step": 4420
    },
    {
      "epoch": 0.13686674030076698,
      "grad_norm": 0.13376844823268128,
      "learning_rate": 2.9385585323203274e-05,
      "loss": 0.1934,
      "step": 4430
    },
    {
      "epoch": 0.1371756945678116,
      "grad_norm": 0.13510587816156738,
      "learning_rate": 2.938123640366522e-05,
      "loss": 0.1995,
      "step": 4440
    },
    {
      "epoch": 0.13748464883485623,
      "grad_norm": 0.14932950286836474,
      "learning_rate": 2.9376872471310247e-05,
      "loss": 0.1961,
      "step": 4450
    },
    {
      "epoch": 0.13779360310190084,
      "grad_norm": 0.12340399413892836,
      "learning_rate": 2.937249353069395e-05,
      "loss": 0.1902,
      "step": 4460
    },
    {
      "epoch": 0.13810255736894547,
      "grad_norm": 0.12742383096621718,
      "learning_rate": 2.936809958638757e-05,
      "loss": 0.1936,
      "step": 4470
    },
    {
      "epoch": 0.13841151163599008,
      "grad_norm": 0.13843377169803955,
      "learning_rate": 2.9363690642978025e-05,
      "loss": 0.1919,
      "step": 4480
    },
    {
      "epoch": 0.1387204659030347,
      "grad_norm": 0.12772195973728676,
      "learning_rate": 2.935926670506789e-05,
      "loss": 0.1933,
      "step": 4490
    },
    {
      "epoch": 0.13902942017007933,
      "grad_norm": 0.14589049855962488,
      "learning_rate": 2.9354827777275375e-05,
      "loss": 0.1933,
      "step": 4500
    },
    {
      "epoch": 0.13933837443712394,
      "grad_norm": 0.11227430428451665,
      "learning_rate": 2.9350373864234366e-05,
      "loss": 0.1909,
      "step": 4510
    },
    {
      "epoch": 0.13964732870416857,
      "grad_norm": 0.13853166961303737,
      "learning_rate": 2.9345904970594374e-05,
      "loss": 0.2041,
      "step": 4520
    },
    {
      "epoch": 0.13995628297121318,
      "grad_norm": 0.12350643169399671,
      "learning_rate": 2.9341421101020552e-05,
      "loss": 0.1996,
      "step": 4530
    },
    {
      "epoch": 0.14026523723825782,
      "grad_norm": 0.14290175255995127,
      "learning_rate": 2.933692226019369e-05,
      "loss": 0.1924,
      "step": 4540
    },
    {
      "epoch": 0.14057419150530243,
      "grad_norm": 0.12829786899010034,
      "learning_rate": 2.9332408452810196e-05,
      "loss": 0.1995,
      "step": 4550
    },
    {
      "epoch": 0.14088314577234703,
      "grad_norm": 0.14530621971898633,
      "learning_rate": 2.932787968358212e-05,
      "loss": 0.193,
      "step": 4560
    },
    {
      "epoch": 0.14119210003939167,
      "grad_norm": 0.12771847498386088,
      "learning_rate": 2.9323335957237118e-05,
      "loss": 0.202,
      "step": 4570
    },
    {
      "epoch": 0.14150105430643628,
      "grad_norm": 0.11736054576554526,
      "learning_rate": 2.931877727851846e-05,
      "loss": 0.1918,
      "step": 4580
    },
    {
      "epoch": 0.14181000857348092,
      "grad_norm": 0.13362073917143064,
      "learning_rate": 2.931420365218503e-05,
      "loss": 0.1989,
      "step": 4590
    },
    {
      "epoch": 0.14211896284052553,
      "grad_norm": 0.16854876515549833,
      "learning_rate": 2.9309615083011322e-05,
      "loss": 0.2035,
      "step": 4600
    },
    {
      "epoch": 0.14242791710757016,
      "grad_norm": 0.12606958463728377,
      "learning_rate": 2.9305011575787412e-05,
      "loss": 0.1947,
      "step": 4610
    },
    {
      "epoch": 0.14273687137461477,
      "grad_norm": 0.11958865479873768,
      "learning_rate": 2.9300393135318977e-05,
      "loss": 0.1971,
      "step": 4620
    },
    {
      "epoch": 0.14304582564165938,
      "grad_norm": 0.129237378103935,
      "learning_rate": 2.9295759766427296e-05,
      "loss": 0.1941,
      "step": 4630
    },
    {
      "epoch": 0.14335477990870402,
      "grad_norm": 0.13016163572306816,
      "learning_rate": 2.929111147394921e-05,
      "loss": 0.1982,
      "step": 4640
    },
    {
      "epoch": 0.14366373417574863,
      "grad_norm": 0.15129061732266857,
      "learning_rate": 2.928644826273716e-05,
      "loss": 0.2065,
      "step": 4650
    },
    {
      "epoch": 0.14397268844279326,
      "grad_norm": 0.1680228209888944,
      "learning_rate": 2.9281770137659147e-05,
      "loss": 0.1965,
      "step": 4660
    },
    {
      "epoch": 0.14428164270983787,
      "grad_norm": 0.1347795092156423,
      "learning_rate": 2.9277077103598745e-05,
      "loss": 0.2065,
      "step": 4670
    },
    {
      "epoch": 0.1445905969768825,
      "grad_norm": 0.13761617881117794,
      "learning_rate": 2.9272369165455092e-05,
      "loss": 0.1952,
      "step": 4680
    },
    {
      "epoch": 0.14489955124392712,
      "grad_norm": 0.11483309292959085,
      "learning_rate": 2.9267646328142888e-05,
      "loss": 0.2062,
      "step": 4690
    },
    {
      "epoch": 0.14520850551097172,
      "grad_norm": 0.14810852616214162,
      "learning_rate": 2.926290859659238e-05,
      "loss": 0.1935,
      "step": 4700
    },
    {
      "epoch": 0.14551745977801636,
      "grad_norm": 0.12469037615780773,
      "learning_rate": 2.9258155975749362e-05,
      "loss": 0.1918,
      "step": 4710
    },
    {
      "epoch": 0.14582641404506097,
      "grad_norm": 0.1408670467987656,
      "learning_rate": 2.9253388470575185e-05,
      "loss": 0.1927,
      "step": 4720
    },
    {
      "epoch": 0.1461353683121056,
      "grad_norm": 0.1363354513340469,
      "learning_rate": 2.9248606086046722e-05,
      "loss": 0.1995,
      "step": 4730
    },
    {
      "epoch": 0.14644432257915022,
      "grad_norm": 0.130509753951675,
      "learning_rate": 2.9243808827156385e-05,
      "loss": 0.1864,
      "step": 4740
    },
    {
      "epoch": 0.14675327684619485,
      "grad_norm": 0.11391094065019194,
      "learning_rate": 2.923899669891212e-05,
      "loss": 0.1981,
      "step": 4750
    },
    {
      "epoch": 0.14706223111323946,
      "grad_norm": 0.11044888884253695,
      "learning_rate": 2.9234169706337384e-05,
      "loss": 0.1908,
      "step": 4760
    },
    {
      "epoch": 0.14737118538028407,
      "grad_norm": 0.12922382038899408,
      "learning_rate": 2.922932785447116e-05,
      "loss": 0.194,
      "step": 4770
    },
    {
      "epoch": 0.1476801396473287,
      "grad_norm": 0.127886291516086,
      "learning_rate": 2.9224471148367936e-05,
      "loss": 0.1961,
      "step": 4780
    },
    {
      "epoch": 0.14798909391437332,
      "grad_norm": 0.12355832270568426,
      "learning_rate": 2.9219599593097712e-05,
      "loss": 0.1924,
      "step": 4790
    },
    {
      "epoch": 0.14829804818141795,
      "grad_norm": 0.1326979789376945,
      "learning_rate": 2.921471319374599e-05,
      "loss": 0.1983,
      "step": 4800
    },
    {
      "epoch": 0.14860700244846256,
      "grad_norm": 0.11511939455576024,
      "learning_rate": 2.920981195541376e-05,
      "loss": 0.1974,
      "step": 4810
    },
    {
      "epoch": 0.1489159567155072,
      "grad_norm": 0.1342465951058469,
      "learning_rate": 2.9204895883217505e-05,
      "loss": 0.2072,
      "step": 4820
    },
    {
      "epoch": 0.1492249109825518,
      "grad_norm": 0.12069929909097779,
      "learning_rate": 2.9199964982289204e-05,
      "loss": 0.1938,
      "step": 4830
    },
    {
      "epoch": 0.14953386524959641,
      "grad_norm": 0.16688076554041342,
      "learning_rate": 2.919501925777631e-05,
      "loss": 0.192,
      "step": 4840
    },
    {
      "epoch": 0.14984281951664105,
      "grad_norm": 0.11696584922616346,
      "learning_rate": 2.9190058714841742e-05,
      "loss": 0.1915,
      "step": 4850
    },
    {
      "epoch": 0.15015177378368566,
      "grad_norm": 0.1552950204583272,
      "learning_rate": 2.9185083358663897e-05,
      "loss": 0.1963,
      "step": 4860
    },
    {
      "epoch": 0.1504607280507303,
      "grad_norm": 0.12285118837217292,
      "learning_rate": 2.9180093194436633e-05,
      "loss": 0.2041,
      "step": 4870
    },
    {
      "epoch": 0.1507696823177749,
      "grad_norm": 0.1354730699847804,
      "learning_rate": 2.9175088227369273e-05,
      "loss": 0.1959,
      "step": 4880
    },
    {
      "epoch": 0.15107863658481954,
      "grad_norm": 0.1066152050049423,
      "learning_rate": 2.917006846268658e-05,
      "loss": 0.1941,
      "step": 4890
    },
    {
      "epoch": 0.15138759085186415,
      "grad_norm": 0.1465495271752134,
      "learning_rate": 2.9165033905628772e-05,
      "loss": 0.1939,
      "step": 4900
    },
    {
      "epoch": 0.15169654511890876,
      "grad_norm": 0.15692246190599393,
      "learning_rate": 2.9159984561451515e-05,
      "loss": 0.1893,
      "step": 4910
    },
    {
      "epoch": 0.1520054993859534,
      "grad_norm": 0.11807608098513886,
      "learning_rate": 2.9154920435425895e-05,
      "loss": 0.1918,
      "step": 4920
    },
    {
      "epoch": 0.152314453652998,
      "grad_norm": 0.13409966056661382,
      "learning_rate": 2.9149841532838448e-05,
      "loss": 0.1912,
      "step": 4930
    },
    {
      "epoch": 0.15262340792004264,
      "grad_norm": 0.18083372810012097,
      "learning_rate": 2.9144747858991122e-05,
      "loss": 0.1954,
      "step": 4940
    },
    {
      "epoch": 0.15293236218708725,
      "grad_norm": 0.11964200044359867,
      "learning_rate": 2.9139639419201296e-05,
      "loss": 0.1945,
      "step": 4950
    },
    {
      "epoch": 0.1532413164541319,
      "grad_norm": 0.13386397016898569,
      "learning_rate": 2.9134516218801747e-05,
      "loss": 0.1958,
      "step": 4960
    },
    {
      "epoch": 0.1535502707211765,
      "grad_norm": 0.15176520812335212,
      "learning_rate": 2.912937826314068e-05,
      "loss": 0.1939,
      "step": 4970
    },
    {
      "epoch": 0.1538592249882211,
      "grad_norm": 0.13802838419756916,
      "learning_rate": 2.912422555758169e-05,
      "loss": 0.1964,
      "step": 4980
    },
    {
      "epoch": 0.15416817925526574,
      "grad_norm": 0.22709981781596195,
      "learning_rate": 2.9119058107503774e-05,
      "loss": 0.2017,
      "step": 4990
    },
    {
      "epoch": 0.15447713352231035,
      "grad_norm": 0.30226292741230987,
      "learning_rate": 2.911387591830132e-05,
      "loss": 0.1966,
      "step": 5000
    },
    {
      "epoch": 0.154786087789355,
      "grad_norm": 0.1627564205425172,
      "learning_rate": 2.910867899538411e-05,
      "loss": 0.1989,
      "step": 5010
    },
    {
      "epoch": 0.1550950420563996,
      "grad_norm": 0.1317254809615595,
      "learning_rate": 2.910346734417729e-05,
      "loss": 0.1999,
      "step": 5020
    },
    {
      "epoch": 0.15540399632344423,
      "grad_norm": 0.13719263854058972,
      "learning_rate": 2.9098240970121395e-05,
      "loss": 0.1899,
      "step": 5030
    },
    {
      "epoch": 0.15571295059048884,
      "grad_norm": 0.13267611025762924,
      "learning_rate": 2.9092999878672326e-05,
      "loss": 0.1971,
      "step": 5040
    },
    {
      "epoch": 0.15602190485753345,
      "grad_norm": 0.14416703153906263,
      "learning_rate": 2.9087744075301347e-05,
      "loss": 0.1927,
      "step": 5050
    },
    {
      "epoch": 0.1563308591245781,
      "grad_norm": 0.12210235810758614,
      "learning_rate": 2.9082473565495087e-05,
      "loss": 0.1923,
      "step": 5060
    },
    {
      "epoch": 0.1566398133916227,
      "grad_norm": 0.11791810636207076,
      "learning_rate": 2.907718835475551e-05,
      "loss": 0.195,
      "step": 5070
    },
    {
      "epoch": 0.15694876765866733,
      "grad_norm": 0.12095560677835107,
      "learning_rate": 2.9071888448599947e-05,
      "loss": 0.1941,
      "step": 5080
    },
    {
      "epoch": 0.15725772192571194,
      "grad_norm": 0.12912300332818374,
      "learning_rate": 2.9066573852561047e-05,
      "loss": 0.1915,
      "step": 5090
    },
    {
      "epoch": 0.15756667619275658,
      "grad_norm": 0.13215524078407134,
      "learning_rate": 2.906124457218682e-05,
      "loss": 0.1907,
      "step": 5100
    },
    {
      "epoch": 0.1578756304598012,
      "grad_norm": 0.17097794726193138,
      "learning_rate": 2.9055900613040588e-05,
      "loss": 0.1978,
      "step": 5110
    },
    {
      "epoch": 0.1581845847268458,
      "grad_norm": 0.12947782531420673,
      "learning_rate": 2.9050541980701002e-05,
      "loss": 0.2019,
      "step": 5120
    },
    {
      "epoch": 0.15849353899389043,
      "grad_norm": 0.13748652286867397,
      "learning_rate": 2.9045168680762023e-05,
      "loss": 0.1963,
      "step": 5130
    },
    {
      "epoch": 0.15880249326093504,
      "grad_norm": 0.13779167566436834,
      "learning_rate": 2.903978071883294e-05,
      "loss": 0.1994,
      "step": 5140
    },
    {
      "epoch": 0.15911144752797968,
      "grad_norm": 0.13634464708954122,
      "learning_rate": 2.903437810053833e-05,
      "loss": 0.1934,
      "step": 5150
    },
    {
      "epoch": 0.15942040179502429,
      "grad_norm": 0.13717458947995578,
      "learning_rate": 2.9028960831518075e-05,
      "loss": 0.1944,
      "step": 5160
    },
    {
      "epoch": 0.15972935606206892,
      "grad_norm": 0.12646264871030233,
      "learning_rate": 2.902352891742737e-05,
      "loss": 0.1911,
      "step": 5170
    },
    {
      "epoch": 0.16003831032911353,
      "grad_norm": 0.13801332842129482,
      "learning_rate": 2.9018082363936667e-05,
      "loss": 0.1917,
      "step": 5180
    },
    {
      "epoch": 0.16034726459615814,
      "grad_norm": 0.1094078051116294,
      "learning_rate": 2.901262117673172e-05,
      "loss": 0.1911,
      "step": 5190
    },
    {
      "epoch": 0.16065621886320278,
      "grad_norm": 0.13287386784850339,
      "learning_rate": 2.9007145361513555e-05,
      "loss": 0.1952,
      "step": 5200
    },
    {
      "epoch": 0.16096517313024739,
      "grad_norm": 0.1053642665870786,
      "learning_rate": 2.9001654923998472e-05,
      "loss": 0.1949,
      "step": 5210
    },
    {
      "epoch": 0.16127412739729202,
      "grad_norm": 0.13283196358546498,
      "learning_rate": 2.899614986991803e-05,
      "loss": 0.1928,
      "step": 5220
    },
    {
      "epoch": 0.16158308166433663,
      "grad_norm": 0.10594792098557188,
      "learning_rate": 2.899063020501904e-05,
      "loss": 0.196,
      "step": 5230
    },
    {
      "epoch": 0.16189203593138127,
      "grad_norm": 0.12056290103985914,
      "learning_rate": 2.8985095935063592e-05,
      "loss": 0.1863,
      "step": 5240
    },
    {
      "epoch": 0.16220099019842588,
      "grad_norm": 0.13276276324563877,
      "learning_rate": 2.8979547065828986e-05,
      "loss": 0.1939,
      "step": 5250
    },
    {
      "epoch": 0.16250994446547048,
      "grad_norm": 0.11880304982202072,
      "learning_rate": 2.8973983603107795e-05,
      "loss": 0.1882,
      "step": 5260
    },
    {
      "epoch": 0.16281889873251512,
      "grad_norm": 0.13647860297794587,
      "learning_rate": 2.8968405552707805e-05,
      "loss": 0.1899,
      "step": 5270
    },
    {
      "epoch": 0.16312785299955973,
      "grad_norm": 0.13737173617789297,
      "learning_rate": 2.8962812920452035e-05,
      "loss": 0.1976,
      "step": 5280
    },
    {
      "epoch": 0.16343680726660437,
      "grad_norm": 0.14560561251250137,
      "learning_rate": 2.8957205712178734e-05,
      "loss": 0.1922,
      "step": 5290
    },
    {
      "epoch": 0.16374576153364898,
      "grad_norm": 0.12816948689448399,
      "learning_rate": 2.8951583933741365e-05,
      "loss": 0.1952,
      "step": 5300
    },
    {
      "epoch": 0.1640547158006936,
      "grad_norm": 0.13875960915862645,
      "learning_rate": 2.8945947591008584e-05,
      "loss": 0.1914,
      "step": 5310
    },
    {
      "epoch": 0.16436367006773822,
      "grad_norm": 0.12147448891632738,
      "learning_rate": 2.894029668986428e-05,
      "loss": 0.2001,
      "step": 5320
    },
    {
      "epoch": 0.16467262433478283,
      "grad_norm": 0.14172848510381716,
      "learning_rate": 2.8934631236207518e-05,
      "loss": 0.2005,
      "step": 5330
    },
    {
      "epoch": 0.16498157860182747,
      "grad_norm": 0.11965777792428922,
      "learning_rate": 2.8928951235952567e-05,
      "loss": 0.202,
      "step": 5340
    },
    {
      "epoch": 0.16529053286887208,
      "grad_norm": 0.12556181014437173,
      "learning_rate": 2.892325669502886e-05,
      "loss": 0.1943,
      "step": 5350
    },
    {
      "epoch": 0.1655994871359167,
      "grad_norm": 0.12988258572700673,
      "learning_rate": 2.8917547619381046e-05,
      "loss": 0.1929,
      "step": 5360
    },
    {
      "epoch": 0.16590844140296132,
      "grad_norm": 0.1591002578274483,
      "learning_rate": 2.8911824014968914e-05,
      "loss": 0.1923,
      "step": 5370
    },
    {
      "epoch": 0.16621739567000596,
      "grad_norm": 0.12842753220093361,
      "learning_rate": 2.890608588776743e-05,
      "loss": 0.1912,
      "step": 5380
    },
    {
      "epoch": 0.16652634993705057,
      "grad_norm": 0.16237973177755397,
      "learning_rate": 2.890033324376673e-05,
      "loss": 0.1982,
      "step": 5390
    },
    {
      "epoch": 0.16683530420409518,
      "grad_norm": 0.13252696464562266,
      "learning_rate": 2.8894566088972086e-05,
      "loss": 0.1941,
      "step": 5400
    },
    {
      "epoch": 0.1671442584711398,
      "grad_norm": 0.1562459648242363,
      "learning_rate": 2.8888784429403942e-05,
      "loss": 0.1951,
      "step": 5410
    },
    {
      "epoch": 0.16745321273818442,
      "grad_norm": 0.19214897406108242,
      "learning_rate": 2.8882988271097867e-05,
      "loss": 0.1944,
      "step": 5420
    },
    {
      "epoch": 0.16776216700522906,
      "grad_norm": 0.11645411992655547,
      "learning_rate": 2.8877177620104563e-05,
      "loss": 0.1976,
      "step": 5430
    },
    {
      "epoch": 0.16807112127227367,
      "grad_norm": 0.14830738696497464,
      "learning_rate": 2.8871352482489867e-05,
      "loss": 0.1943,
      "step": 5440
    },
    {
      "epoch": 0.1683800755393183,
      "grad_norm": 0.1476613773833638,
      "learning_rate": 2.886551286433475e-05,
      "loss": 0.1988,
      "step": 5450
    },
    {
      "epoch": 0.1686890298063629,
      "grad_norm": 0.363023628817013,
      "learning_rate": 2.8859658771735275e-05,
      "loss": 0.2041,
      "step": 5460
    },
    {
      "epoch": 0.16899798407340752,
      "grad_norm": 0.12877823250892356,
      "learning_rate": 2.885379021080264e-05,
      "loss": 0.1883,
      "step": 5470
    },
    {
      "epoch": 0.16930693834045216,
      "grad_norm": 0.16072479476005033,
      "learning_rate": 2.884790718766313e-05,
      "loss": 0.1957,
      "step": 5480
    },
    {
      "epoch": 0.16961589260749677,
      "grad_norm": 0.10660898088117149,
      "learning_rate": 2.8842009708458134e-05,
      "loss": 0.1879,
      "step": 5490
    },
    {
      "epoch": 0.1699248468745414,
      "grad_norm": 0.11287940246473271,
      "learning_rate": 2.8836097779344124e-05,
      "loss": 0.1967,
      "step": 5500
    },
    {
      "epoch": 0.170233801141586,
      "grad_norm": 0.1392226076869493,
      "learning_rate": 2.8830171406492675e-05,
      "loss": 0.1961,
      "step": 5510
    },
    {
      "epoch": 0.17054275540863065,
      "grad_norm": 0.14186426823834117,
      "learning_rate": 2.882423059609042e-05,
      "loss": 0.1909,
      "step": 5520
    },
    {
      "epoch": 0.17085170967567526,
      "grad_norm": 0.11802002523931364,
      "learning_rate": 2.8818275354339076e-05,
      "loss": 0.189,
      "step": 5530
    },
    {
      "epoch": 0.17116066394271987,
      "grad_norm": 0.1544264170886084,
      "learning_rate": 2.881230568745542e-05,
      "loss": 0.201,
      "step": 5540
    },
    {
      "epoch": 0.1714696182097645,
      "grad_norm": 0.1390375900467053,
      "learning_rate": 2.8806321601671283e-05,
      "loss": 0.2041,
      "step": 5550
    },
    {
      "epoch": 0.1717785724768091,
      "grad_norm": 0.13144810014781408,
      "learning_rate": 2.8800323103233562e-05,
      "loss": 0.1902,
      "step": 5560
    },
    {
      "epoch": 0.17208752674385375,
      "grad_norm": 0.4660580390716817,
      "learning_rate": 2.8794310198404187e-05,
      "loss": 0.1906,
      "step": 5570
    },
    {
      "epoch": 0.17239648101089836,
      "grad_norm": 0.12437154731352344,
      "learning_rate": 2.8788282893460132e-05,
      "loss": 0.2117,
      "step": 5580
    },
    {
      "epoch": 0.172705435277943,
      "grad_norm": 0.12213570462157555,
      "learning_rate": 2.8782241194693405e-05,
      "loss": 0.1932,
      "step": 5590
    },
    {
      "epoch": 0.1730143895449876,
      "grad_norm": 0.11938574410310657,
      "learning_rate": 2.877618510841104e-05,
      "loss": 0.1844,
      "step": 5600
    },
    {
      "epoch": 0.1733233438120322,
      "grad_norm": 0.18758761384669917,
      "learning_rate": 2.8770114640935082e-05,
      "loss": 0.1992,
      "step": 5610
    },
    {
      "epoch": 0.17363229807907685,
      "grad_norm": 0.1323353027887345,
      "learning_rate": 2.8764029798602595e-05,
      "loss": 0.1934,
      "step": 5620
    },
    {
      "epoch": 0.17394125234612146,
      "grad_norm": 0.13036359195874733,
      "learning_rate": 2.8757930587765656e-05,
      "loss": 0.1968,
      "step": 5630
    },
    {
      "epoch": 0.1742502066131661,
      "grad_norm": 0.1437468757002543,
      "learning_rate": 2.8751817014791328e-05,
      "loss": 0.1937,
      "step": 5640
    },
    {
      "epoch": 0.1745591608802107,
      "grad_norm": 0.11855881753153365,
      "learning_rate": 2.8745689086061676e-05,
      "loss": 0.1906,
      "step": 5650
    },
    {
      "epoch": 0.17486811514725534,
      "grad_norm": 0.14122479545822594,
      "learning_rate": 2.8739546807973755e-05,
      "loss": 0.1881,
      "step": 5660
    },
    {
      "epoch": 0.17517706941429995,
      "grad_norm": 0.15249348073207336,
      "learning_rate": 2.8733390186939583e-05,
      "loss": 0.1969,
      "step": 5670
    },
    {
      "epoch": 0.17548602368134458,
      "grad_norm": 0.1281228020957653,
      "learning_rate": 2.8727219229386165e-05,
      "loss": 0.1902,
      "step": 5680
    },
    {
      "epoch": 0.1757949779483892,
      "grad_norm": 0.13912156186757996,
      "learning_rate": 2.8721033941755475e-05,
      "loss": 0.1917,
      "step": 5690
    },
    {
      "epoch": 0.1761039322154338,
      "grad_norm": 0.13363292138326746,
      "learning_rate": 2.871483433050443e-05,
      "loss": 0.2003,
      "step": 5700
    },
    {
      "epoch": 0.17641288648247844,
      "grad_norm": 0.12385133385686929,
      "learning_rate": 2.8708620402104917e-05,
      "loss": 0.1877,
      "step": 5710
    },
    {
      "epoch": 0.17672184074952305,
      "grad_norm": 0.15347590707493236,
      "learning_rate": 2.8702392163043755e-05,
      "loss": 0.202,
      "step": 5720
    },
    {
      "epoch": 0.17703079501656768,
      "grad_norm": 0.14209395882176734,
      "learning_rate": 2.8696149619822714e-05,
      "loss": 0.1899,
      "step": 5730
    },
    {
      "epoch": 0.1773397492836123,
      "grad_norm": 0.11663542532440394,
      "learning_rate": 2.8689892778958487e-05,
      "loss": 0.1931,
      "step": 5740
    },
    {
      "epoch": 0.17764870355065693,
      "grad_norm": 0.1103800020632343,
      "learning_rate": 2.8683621646982696e-05,
      "loss": 0.1907,
      "step": 5750
    },
    {
      "epoch": 0.17795765781770154,
      "grad_norm": 0.12819159634700894,
      "learning_rate": 2.867733623044189e-05,
      "loss": 0.2143,
      "step": 5760
    },
    {
      "epoch": 0.17826661208474615,
      "grad_norm": 0.8223795049786881,
      "learning_rate": 2.867103653589751e-05,
      "loss": 0.2072,
      "step": 5770
    },
    {
      "epoch": 0.17857556635179078,
      "grad_norm": 4.599271520472337,
      "learning_rate": 2.8664722569925926e-05,
      "loss": 0.2243,
      "step": 5780
    },
    {
      "epoch": 0.1788845206188354,
      "grad_norm": 0.331101029184519,
      "learning_rate": 2.8658394339118385e-05,
      "loss": 0.2326,
      "step": 5790
    },
    {
      "epoch": 0.17919347488588003,
      "grad_norm": 0.16303299024889983,
      "learning_rate": 2.8652051850081038e-05,
      "loss": 0.1965,
      "step": 5800
    },
    {
      "epoch": 0.17950242915292464,
      "grad_norm": 0.1212517752245839,
      "learning_rate": 2.864569510943492e-05,
      "loss": 0.1912,
      "step": 5810
    },
    {
      "epoch": 0.17981138341996927,
      "grad_norm": 0.15802784822663518,
      "learning_rate": 2.8639324123815937e-05,
      "loss": 0.2154,
      "step": 5820
    },
    {
      "epoch": 0.18012033768701388,
      "grad_norm": 0.14573693043932212,
      "learning_rate": 2.8632938899874873e-05,
      "loss": 0.1971,
      "step": 5830
    },
    {
      "epoch": 0.1804292919540585,
      "grad_norm": 0.1144883581455652,
      "learning_rate": 2.862653944427736e-05,
      "loss": 0.1925,
      "step": 5840
    },
    {
      "epoch": 0.18073824622110313,
      "grad_norm": 0.12229559324577813,
      "learning_rate": 2.8620125763703915e-05,
      "loss": 0.192,
      "step": 5850
    },
    {
      "epoch": 0.18104720048814774,
      "grad_norm": 0.12451915938153357,
      "learning_rate": 2.8613697864849874e-05,
      "loss": 0.1998,
      "step": 5860
    },
    {
      "epoch": 0.18135615475519237,
      "grad_norm": 0.12089904409190706,
      "learning_rate": 2.860725575442544e-05,
      "loss": 0.2047,
      "step": 5870
    },
    {
      "epoch": 0.18166510902223698,
      "grad_norm": 0.13070519068349834,
      "learning_rate": 2.8600799439155636e-05,
      "loss": 0.2187,
      "step": 5880
    },
    {
      "epoch": 0.18197406328928162,
      "grad_norm": 0.140904960792321,
      "learning_rate": 2.859432892578032e-05,
      "loss": 0.1955,
      "step": 5890
    },
    {
      "epoch": 0.18228301755632623,
      "grad_norm": 0.11511321564596642,
      "learning_rate": 2.8587844221054174e-05,
      "loss": 0.2023,
      "step": 5900
    },
    {
      "epoch": 0.18259197182337084,
      "grad_norm": 0.14044793424844815,
      "learning_rate": 2.8581345331746685e-05,
      "loss": 0.1936,
      "step": 5910
    },
    {
      "epoch": 0.18290092609041547,
      "grad_norm": 0.12197453877103906,
      "learning_rate": 2.857483226464216e-05,
      "loss": 0.2058,
      "step": 5920
    },
    {
      "epoch": 0.18320988035746008,
      "grad_norm": 0.10550839853911653,
      "learning_rate": 2.85683050265397e-05,
      "loss": 0.1888,
      "step": 5930
    },
    {
      "epoch": 0.18351883462450472,
      "grad_norm": 0.3544110519251212,
      "learning_rate": 2.8561763624253193e-05,
      "loss": 0.2015,
      "step": 5940
    },
    {
      "epoch": 0.18382778889154933,
      "grad_norm": 0.12390373192178894,
      "learning_rate": 2.8555208064611327e-05,
      "loss": 0.1935,
      "step": 5950
    },
    {
      "epoch": 0.18413674315859396,
      "grad_norm": 0.10157626065289536,
      "learning_rate": 2.854863835445756e-05,
      "loss": 0.2014,
      "step": 5960
    },
    {
      "epoch": 0.18444569742563857,
      "grad_norm": 0.1213651122339615,
      "learning_rate": 2.854205450065012e-05,
      "loss": 0.1951,
      "step": 5970
    },
    {
      "epoch": 0.18475465169268318,
      "grad_norm": 0.110018669063968,
      "learning_rate": 2.8535456510062012e-05,
      "loss": 0.1937,
      "step": 5980
    },
    {
      "epoch": 0.18506360595972782,
      "grad_norm": 0.12020488555211208,
      "learning_rate": 2.8528844389580987e-05,
      "loss": 0.1933,
      "step": 5990
    },
    {
      "epoch": 0.18537256022677243,
      "grad_norm": 0.10212491511734467,
      "learning_rate": 2.852221814610954e-05,
      "loss": 0.1934,
      "step": 6000
    },
    {
      "epoch": 0.18568151449381706,
      "grad_norm": 0.11461418479211462,
      "learning_rate": 2.8515577786564938e-05,
      "loss": 0.1962,
      "step": 6010
    },
    {
      "epoch": 0.18599046876086167,
      "grad_norm": 0.12701914196924702,
      "learning_rate": 2.8508923317879146e-05,
      "loss": 0.1933,
      "step": 6020
    },
    {
      "epoch": 0.1862994230279063,
      "grad_norm": 0.10620878119551466,
      "learning_rate": 2.8502254746998892e-05,
      "loss": 0.1913,
      "step": 6030
    },
    {
      "epoch": 0.18660837729495092,
      "grad_norm": 0.12540398396874758,
      "learning_rate": 2.8495572080885604e-05,
      "loss": 0.2027,
      "step": 6040
    },
    {
      "epoch": 0.18691733156199553,
      "grad_norm": 0.11185891031649897,
      "learning_rate": 2.848887532651543e-05,
      "loss": 0.1914,
      "step": 6050
    },
    {
      "epoch": 0.18722628582904016,
      "grad_norm": 0.10471835154321292,
      "learning_rate": 2.8482164490879227e-05,
      "loss": 0.1889,
      "step": 6060
    },
    {
      "epoch": 0.18753524009608477,
      "grad_norm": 0.11272716282275089,
      "learning_rate": 2.8475439580982543e-05,
      "loss": 0.1933,
      "step": 6070
    },
    {
      "epoch": 0.1878441943631294,
      "grad_norm": 0.11032134196256359,
      "learning_rate": 2.8468700603845642e-05,
      "loss": 0.1956,
      "step": 6080
    },
    {
      "epoch": 0.18815314863017402,
      "grad_norm": 0.13815453925637802,
      "learning_rate": 2.846194756650344e-05,
      "loss": 0.1919,
      "step": 6090
    },
    {
      "epoch": 0.18846210289721865,
      "grad_norm": 0.11573234574605717,
      "learning_rate": 2.8455180476005557e-05,
      "loss": 0.2012,
      "step": 6100
    },
    {
      "epoch": 0.18877105716426326,
      "grad_norm": 0.11196139186060282,
      "learning_rate": 2.8448399339416274e-05,
      "loss": 0.1929,
      "step": 6110
    },
    {
      "epoch": 0.18908001143130787,
      "grad_norm": 0.10739449957825083,
      "learning_rate": 2.8441604163814525e-05,
      "loss": 0.1916,
      "step": 6120
    },
    {
      "epoch": 0.1893889656983525,
      "grad_norm": 0.1161372376814427,
      "learning_rate": 2.843479495629392e-05,
      "loss": 0.1863,
      "step": 6130
    },
    {
      "epoch": 0.18969791996539712,
      "grad_norm": 0.14113036096515547,
      "learning_rate": 2.8427971723962702e-05,
      "loss": 0.2056,
      "step": 6140
    },
    {
      "epoch": 0.19000687423244175,
      "grad_norm": 0.10329701247144614,
      "learning_rate": 2.842113447394376e-05,
      "loss": 0.1958,
      "step": 6150
    },
    {
      "epoch": 0.19031582849948636,
      "grad_norm": 0.11459855113810817,
      "learning_rate": 2.8414283213374613e-05,
      "loss": 0.1955,
      "step": 6160
    },
    {
      "epoch": 0.190624782766531,
      "grad_norm": 0.11189483533384283,
      "learning_rate": 2.8407417949407414e-05,
      "loss": 0.1933,
      "step": 6170
    },
    {
      "epoch": 0.1909337370335756,
      "grad_norm": 0.10358808296525418,
      "learning_rate": 2.8400538689208924e-05,
      "loss": 0.1917,
      "step": 6180
    },
    {
      "epoch": 0.19124269130062022,
      "grad_norm": 0.12269036980168484,
      "learning_rate": 2.8393645439960516e-05,
      "loss": 0.1963,
      "step": 6190
    },
    {
      "epoch": 0.19155164556766485,
      "grad_norm": 0.10795805628499067,
      "learning_rate": 2.8386738208858177e-05,
      "loss": 0.1907,
      "step": 6200
    },
    {
      "epoch": 0.19186059983470946,
      "grad_norm": 0.10460423251159999,
      "learning_rate": 2.8379817003112482e-05,
      "loss": 0.1903,
      "step": 6210
    },
    {
      "epoch": 0.1921695541017541,
      "grad_norm": 0.11496650985336623,
      "learning_rate": 2.837288182994859e-05,
      "loss": 0.1919,
      "step": 6220
    },
    {
      "epoch": 0.1924785083687987,
      "grad_norm": 0.14483110291964724,
      "learning_rate": 2.836593269660625e-05,
      "loss": 0.1936,
      "step": 6230
    },
    {
      "epoch": 0.19278746263584334,
      "grad_norm": 0.09787707275628683,
      "learning_rate": 2.835896961033978e-05,
      "loss": 0.189,
      "step": 6240
    },
    {
      "epoch": 0.19309641690288795,
      "grad_norm": 0.1399375909350434,
      "learning_rate": 2.8351992578418068e-05,
      "loss": 0.1922,
      "step": 6250
    },
    {
      "epoch": 0.19340537116993256,
      "grad_norm": 0.12969252303227272,
      "learning_rate": 2.8345001608124543e-05,
      "loss": 0.1915,
      "step": 6260
    },
    {
      "epoch": 0.1937143254369772,
      "grad_norm": 0.1145848580359697,
      "learning_rate": 2.833799670675721e-05,
      "loss": 0.2005,
      "step": 6270
    },
    {
      "epoch": 0.1940232797040218,
      "grad_norm": 0.12357801812493495,
      "learning_rate": 2.8330977881628603e-05,
      "loss": 0.1898,
      "step": 6280
    },
    {
      "epoch": 0.19433223397106644,
      "grad_norm": 0.10802118534444101,
      "learning_rate": 2.832394514006579e-05,
      "loss": 0.1903,
      "step": 6290
    },
    {
      "epoch": 0.19464118823811105,
      "grad_norm": 0.7038205985344811,
      "learning_rate": 2.8316898489410373e-05,
      "loss": 0.2084,
      "step": 6300
    },
    {
      "epoch": 0.1949501425051557,
      "grad_norm": 0.852860941813526,
      "learning_rate": 2.8309837937018464e-05,
      "loss": 0.2115,
      "step": 6310
    },
    {
      "epoch": 0.1952590967722003,
      "grad_norm": 0.11807169443069974,
      "learning_rate": 2.8302763490260696e-05,
      "loss": 0.2002,
      "step": 6320
    },
    {
      "epoch": 0.1955680510392449,
      "grad_norm": 0.12318553183844727,
      "learning_rate": 2.8295675156522204e-05,
      "loss": 0.1987,
      "step": 6330
    },
    {
      "epoch": 0.19587700530628954,
      "grad_norm": 0.11085622907333874,
      "learning_rate": 2.8288572943202624e-05,
      "loss": 0.1925,
      "step": 6340
    },
    {
      "epoch": 0.19618595957333415,
      "grad_norm": 0.16905478398175539,
      "learning_rate": 2.8281456857716076e-05,
      "loss": 0.2044,
      "step": 6350
    },
    {
      "epoch": 0.1964949138403788,
      "grad_norm": 0.10671870520994445,
      "learning_rate": 2.8274326907491156e-05,
      "loss": 0.1911,
      "step": 6360
    },
    {
      "epoch": 0.1968038681074234,
      "grad_norm": 0.10623440103784029,
      "learning_rate": 2.8267183099970945e-05,
      "loss": 0.19,
      "step": 6370
    },
    {
      "epoch": 0.19711282237446803,
      "grad_norm": 0.15928224004029165,
      "learning_rate": 2.8260025442612983e-05,
      "loss": 0.1949,
      "step": 6380
    },
    {
      "epoch": 0.19742177664151264,
      "grad_norm": 0.11034131216285688,
      "learning_rate": 2.8252853942889274e-05,
      "loss": 0.1918,
      "step": 6390
    },
    {
      "epoch": 0.19773073090855725,
      "grad_norm": 0.1140506297139604,
      "learning_rate": 2.824566860828627e-05,
      "loss": 0.1928,
      "step": 6400
    },
    {
      "epoch": 0.1980396851756019,
      "grad_norm": 0.11926635456622503,
      "learning_rate": 2.8238469446304854e-05,
      "loss": 0.1912,
      "step": 6410
    },
    {
      "epoch": 0.1983486394426465,
      "grad_norm": 0.12481884844878596,
      "learning_rate": 2.823125646446036e-05,
      "loss": 0.1973,
      "step": 6420
    },
    {
      "epoch": 0.19865759370969113,
      "grad_norm": 0.11023358164463454,
      "learning_rate": 2.8224029670282544e-05,
      "loss": 0.1912,
      "step": 6430
    },
    {
      "epoch": 0.19896654797673574,
      "grad_norm": 0.157612940617181,
      "learning_rate": 2.8216789071315577e-05,
      "loss": 0.1904,
      "step": 6440
    },
    {
      "epoch": 0.19927550224378038,
      "grad_norm": 0.17797632091205678,
      "learning_rate": 2.8209534675118037e-05,
      "loss": 0.198,
      "step": 6450
    },
    {
      "epoch": 0.199584456510825,
      "grad_norm": 0.12752648901399943,
      "learning_rate": 2.8202266489262927e-05,
      "loss": 0.1992,
      "step": 6460
    },
    {
      "epoch": 0.1998934107778696,
      "grad_norm": 0.10197005313284616,
      "learning_rate": 2.8194984521337615e-05,
      "loss": 0.1991,
      "step": 6470
    },
    {
      "epoch": 0.20020236504491423,
      "grad_norm": 0.11157616962657688,
      "learning_rate": 2.8187688778943873e-05,
      "loss": 0.1955,
      "step": 6480
    },
    {
      "epoch": 0.20051131931195884,
      "grad_norm": 0.12325180760299378,
      "learning_rate": 2.818037926969786e-05,
      "loss": 0.1936,
      "step": 6490
    },
    {
      "epoch": 0.20082027357900348,
      "grad_norm": 0.12801150415023763,
      "learning_rate": 2.8173056001230084e-05,
      "loss": 0.1881,
      "step": 6500
    },
    {
      "epoch": 0.2011292278460481,
      "grad_norm": 0.10416101082645696,
      "learning_rate": 2.816571898118544e-05,
      "loss": 0.1922,
      "step": 6510
    },
    {
      "epoch": 0.20143818211309272,
      "grad_norm": 0.10950422981342788,
      "learning_rate": 2.815836821722316e-05,
      "loss": 0.1882,
      "step": 6520
    },
    {
      "epoch": 0.20174713638013733,
      "grad_norm": 0.09763003963050149,
      "learning_rate": 2.815100371701684e-05,
      "loss": 0.19,
      "step": 6530
    },
    {
      "epoch": 0.20205609064718194,
      "grad_norm": 0.10277512232078516,
      "learning_rate": 2.8143625488254397e-05,
      "loss": 0.1899,
      "step": 6540
    },
    {
      "epoch": 0.20236504491422658,
      "grad_norm": 0.11553245013740743,
      "learning_rate": 2.8136233538638094e-05,
      "loss": 0.1947,
      "step": 6550
    },
    {
      "epoch": 0.20267399918127119,
      "grad_norm": 0.11943120745573048,
      "learning_rate": 2.8128827875884514e-05,
      "loss": 0.191,
      "step": 6560
    },
    {
      "epoch": 0.20298295344831582,
      "grad_norm": 0.12929683588195923,
      "learning_rate": 2.812140850772455e-05,
      "loss": 0.1917,
      "step": 6570
    },
    {
      "epoch": 0.20329190771536043,
      "grad_norm": 0.11462916010589808,
      "learning_rate": 2.8113975441903408e-05,
      "loss": 0.1892,
      "step": 6580
    },
    {
      "epoch": 0.20360086198240507,
      "grad_norm": 0.10741753547213757,
      "learning_rate": 2.8106528686180597e-05,
      "loss": 0.1909,
      "step": 6590
    },
    {
      "epoch": 0.20390981624944968,
      "grad_norm": 0.12371768694612524,
      "learning_rate": 2.8099068248329903e-05,
      "loss": 0.1918,
      "step": 6600
    },
    {
      "epoch": 0.20421877051649429,
      "grad_norm": 0.17782467422845236,
      "learning_rate": 2.8091594136139415e-05,
      "loss": 0.2014,
      "step": 6610
    },
    {
      "epoch": 0.20452772478353892,
      "grad_norm": 0.13069920151152928,
      "learning_rate": 2.8084106357411475e-05,
      "loss": 0.1873,
      "step": 6620
    },
    {
      "epoch": 0.20483667905058353,
      "grad_norm": 0.1055617718054131,
      "learning_rate": 2.8076604919962707e-05,
      "loss": 0.1904,
      "step": 6630
    },
    {
      "epoch": 0.20514563331762817,
      "grad_norm": 0.1159829518595496,
      "learning_rate": 2.8069089831623987e-05,
      "loss": 0.1963,
      "step": 6640
    },
    {
      "epoch": 0.20545458758467278,
      "grad_norm": 0.10335240155650839,
      "learning_rate": 2.8061561100240446e-05,
      "loss": 0.194,
      "step": 6650
    },
    {
      "epoch": 0.2057635418517174,
      "grad_norm": 0.10692461701538676,
      "learning_rate": 2.8054018733671454e-05,
      "loss": 0.1879,
      "step": 6660
    },
    {
      "epoch": 0.20607249611876202,
      "grad_norm": 0.11119981470419012,
      "learning_rate": 2.804646273979061e-05,
      "loss": 0.1985,
      "step": 6670
    },
    {
      "epoch": 0.20638145038580663,
      "grad_norm": 0.1350643179785673,
      "learning_rate": 2.8038893126485754e-05,
      "loss": 0.1936,
      "step": 6680
    },
    {
      "epoch": 0.20669040465285127,
      "grad_norm": 0.17007718558751797,
      "learning_rate": 2.8031309901658927e-05,
      "loss": 0.1916,
      "step": 6690
    },
    {
      "epoch": 0.20699935891989588,
      "grad_norm": 0.10746164982745886,
      "learning_rate": 2.802371307322639e-05,
      "loss": 0.1887,
      "step": 6700
    },
    {
      "epoch": 0.2073083131869405,
      "grad_norm": 0.10296855946725689,
      "learning_rate": 2.8016102649118596e-05,
      "loss": 0.1968,
      "step": 6710
    },
    {
      "epoch": 0.20761726745398512,
      "grad_norm": 0.11822232947799508,
      "learning_rate": 2.80084786372802e-05,
      "loss": 0.1948,
      "step": 6720
    },
    {
      "epoch": 0.20792622172102976,
      "grad_norm": 0.1049014841435582,
      "learning_rate": 2.8000841045670038e-05,
      "loss": 0.186,
      "step": 6730
    },
    {
      "epoch": 0.20823517598807437,
      "grad_norm": 0.12518456734681194,
      "learning_rate": 2.7993189882261124e-05,
      "loss": 0.1891,
      "step": 6740
    },
    {
      "epoch": 0.20854413025511898,
      "grad_norm": 0.11229137823017848,
      "learning_rate": 2.7985525155040626e-05,
      "loss": 0.1932,
      "step": 6750
    },
    {
      "epoch": 0.2088530845221636,
      "grad_norm": 0.10920721402430587,
      "learning_rate": 2.7977846872009897e-05,
      "loss": 0.1948,
      "step": 6760
    },
    {
      "epoch": 0.20916203878920822,
      "grad_norm": 0.14095829376780517,
      "learning_rate": 2.7970155041184414e-05,
      "loss": 0.1952,
      "step": 6770
    },
    {
      "epoch": 0.20947099305625286,
      "grad_norm": 0.13626041179383216,
      "learning_rate": 2.7962449670593822e-05,
      "loss": 0.2017,
      "step": 6780
    },
    {
      "epoch": 0.20977994732329747,
      "grad_norm": 0.10590137512169466,
      "learning_rate": 2.795473076828188e-05,
      "loss": 0.1933,
      "step": 6790
    },
    {
      "epoch": 0.2100889015903421,
      "grad_norm": 0.12858921928245667,
      "learning_rate": 2.7946998342306485e-05,
      "loss": 0.2017,
      "step": 6800
    },
    {
      "epoch": 0.2103978558573867,
      "grad_norm": 0.12263130724996488,
      "learning_rate": 2.793925240073965e-05,
      "loss": 0.1887,
      "step": 6810
    },
    {
      "epoch": 0.21070681012443132,
      "grad_norm": 0.11411065395288106,
      "learning_rate": 2.7931492951667486e-05,
      "loss": 0.2068,
      "step": 6820
    },
    {
      "epoch": 0.21101576439147596,
      "grad_norm": 0.14052334966829683,
      "learning_rate": 2.792372000319022e-05,
      "loss": 0.1903,
      "step": 6830
    },
    {
      "epoch": 0.21132471865852057,
      "grad_norm": 0.10184578448643603,
      "learning_rate": 2.7915933563422168e-05,
      "loss": 0.1957,
      "step": 6840
    },
    {
      "epoch": 0.2116336729255652,
      "grad_norm": 0.15600508192428814,
      "learning_rate": 2.7908133640491724e-05,
      "loss": 0.193,
      "step": 6850
    },
    {
      "epoch": 0.2119426271926098,
      "grad_norm": 0.1155550515031426,
      "learning_rate": 2.7900320242541355e-05,
      "loss": 0.1891,
      "step": 6860
    },
    {
      "epoch": 0.21225158145965445,
      "grad_norm": 0.13793849430052257,
      "learning_rate": 2.7892493377727602e-05,
      "loss": 0.1894,
      "step": 6870
    },
    {
      "epoch": 0.21256053572669906,
      "grad_norm": 0.10042125672168209,
      "learning_rate": 2.7884653054221068e-05,
      "loss": 0.1921,
      "step": 6880
    },
    {
      "epoch": 0.21286948999374367,
      "grad_norm": 0.13407576631381074,
      "learning_rate": 2.7876799280206394e-05,
      "loss": 0.201,
      "step": 6890
    },
    {
      "epoch": 0.2131784442607883,
      "grad_norm": 0.10873064021666524,
      "learning_rate": 2.786893206388227e-05,
      "loss": 0.1896,
      "step": 6900
    },
    {
      "epoch": 0.2134873985278329,
      "grad_norm": 0.11818610369536223,
      "learning_rate": 2.786105141346142e-05,
      "loss": 0.1966,
      "step": 6910
    },
    {
      "epoch": 0.21379635279487755,
      "grad_norm": 0.12481232088656549,
      "learning_rate": 2.785315733717058e-05,
      "loss": 0.1971,
      "step": 6920
    },
    {
      "epoch": 0.21410530706192216,
      "grad_norm": 0.14966870833411877,
      "learning_rate": 2.7845249843250518e-05,
      "loss": 0.2037,
      "step": 6930
    },
    {
      "epoch": 0.2144142613289668,
      "grad_norm": 0.1251945341593738,
      "learning_rate": 2.7837328939955997e-05,
      "loss": 0.1916,
      "step": 6940
    },
    {
      "epoch": 0.2147232155960114,
      "grad_norm": 0.12207705977247393,
      "learning_rate": 2.782939463555579e-05,
      "loss": 0.1974,
      "step": 6950
    },
    {
      "epoch": 0.215032169863056,
      "grad_norm": 0.12038734457901437,
      "learning_rate": 2.7821446938332637e-05,
      "loss": 0.197,
      "step": 6960
    },
    {
      "epoch": 0.21534112413010065,
      "grad_norm": 0.09818291730818668,
      "learning_rate": 2.781348585658329e-05,
      "loss": 0.1907,
      "step": 6970
    },
    {
      "epoch": 0.21565007839714526,
      "grad_norm": 0.09848192293933164,
      "learning_rate": 2.780551139861845e-05,
      "loss": 0.1878,
      "step": 6980
    },
    {
      "epoch": 0.2159590326641899,
      "grad_norm": 0.16289460630025418,
      "learning_rate": 2.779752357276279e-05,
      "loss": 0.1873,
      "step": 6990
    },
    {
      "epoch": 0.2162679869312345,
      "grad_norm": 0.10641772126448294,
      "learning_rate": 2.778952238735494e-05,
      "loss": 0.1844,
      "step": 7000
    },
    {
      "epoch": 0.21657694119827914,
      "grad_norm": 0.12320916475177035,
      "learning_rate": 2.7781507850747465e-05,
      "loss": 0.1905,
      "step": 7010
    },
    {
      "epoch": 0.21688589546532375,
      "grad_norm": 0.1206318889827832,
      "learning_rate": 2.777347997130688e-05,
      "loss": 0.193,
      "step": 7020
    },
    {
      "epoch": 0.21719484973236836,
      "grad_norm": 0.12415769443532793,
      "learning_rate": 2.776543875741363e-05,
      "loss": 0.1973,
      "step": 7030
    },
    {
      "epoch": 0.217503803999413,
      "grad_norm": 0.10105900635222193,
      "learning_rate": 2.7757384217462074e-05,
      "loss": 0.1855,
      "step": 7040
    },
    {
      "epoch": 0.2178127582664576,
      "grad_norm": 0.12183775230754369,
      "learning_rate": 2.7749316359860473e-05,
      "loss": 0.1974,
      "step": 7050
    },
    {
      "epoch": 0.21812171253350224,
      "grad_norm": 0.11627326751129516,
      "learning_rate": 2.7741235193031013e-05,
      "loss": 0.1901,
      "step": 7060
    },
    {
      "epoch": 0.21843066680054685,
      "grad_norm": 0.11888956708480215,
      "learning_rate": 2.7733140725409753e-05,
      "loss": 0.1919,
      "step": 7070
    },
    {
      "epoch": 0.21873962106759148,
      "grad_norm": 0.10836367743252319,
      "learning_rate": 2.772503296544665e-05,
      "loss": 0.1974,
      "step": 7080
    },
    {
      "epoch": 0.2190485753346361,
      "grad_norm": 0.09845270251147463,
      "learning_rate": 2.7716911921605527e-05,
      "loss": 0.192,
      "step": 7090
    },
    {
      "epoch": 0.2193575296016807,
      "grad_norm": 0.09759744848301183,
      "learning_rate": 2.7708777602364084e-05,
      "loss": 0.189,
      "step": 7100
    },
    {
      "epoch": 0.21966648386872534,
      "grad_norm": 0.16511348874597986,
      "learning_rate": 2.7700630016213873e-05,
      "loss": 0.1994,
      "step": 7110
    },
    {
      "epoch": 0.21997543813576995,
      "grad_norm": 0.10122062268642094,
      "learning_rate": 2.7692469171660298e-05,
      "loss": 0.1897,
      "step": 7120
    },
    {
      "epoch": 0.22028439240281458,
      "grad_norm": 0.1690160311789051,
      "learning_rate": 2.7684295077222606e-05,
      "loss": 0.1946,
      "step": 7130
    },
    {
      "epoch": 0.2205933466698592,
      "grad_norm": 0.1037572984921032,
      "learning_rate": 2.7676107741433868e-05,
      "loss": 0.196,
      "step": 7140
    },
    {
      "epoch": 0.22090230093690383,
      "grad_norm": 0.10057158452622221,
      "learning_rate": 2.7667907172840985e-05,
      "loss": 0.1923,
      "step": 7150
    },
    {
      "epoch": 0.22121125520394844,
      "grad_norm": 0.11703235990845123,
      "learning_rate": 2.7659693380004672e-05,
      "loss": 0.1901,
      "step": 7160
    },
    {
      "epoch": 0.22152020947099305,
      "grad_norm": 0.12712275759726363,
      "learning_rate": 2.7651466371499447e-05,
      "loss": 0.1934,
      "step": 7170
    },
    {
      "epoch": 0.22182916373803768,
      "grad_norm": 0.10892070832647131,
      "learning_rate": 2.764322615591362e-05,
      "loss": 0.1894,
      "step": 7180
    },
    {
      "epoch": 0.2221381180050823,
      "grad_norm": 0.12384774155062112,
      "learning_rate": 2.7634972741849293e-05,
      "loss": 0.2006,
      "step": 7190
    },
    {
      "epoch": 0.22244707227212693,
      "grad_norm": 0.1443436358378938,
      "learning_rate": 2.762670613792235e-05,
      "loss": 0.2042,
      "step": 7200
    },
    {
      "epoch": 0.22275602653917154,
      "grad_norm": 0.0902185286773544,
      "learning_rate": 2.7618426352762427e-05,
      "loss": 0.1925,
      "step": 7210
    },
    {
      "epoch": 0.22306498080621617,
      "grad_norm": 0.1058137828159615,
      "learning_rate": 2.7610133395012946e-05,
      "loss": 0.1901,
      "step": 7220
    },
    {
      "epoch": 0.22337393507326078,
      "grad_norm": 0.11373454837980301,
      "learning_rate": 2.7601827273331065e-05,
      "loss": 0.1941,
      "step": 7230
    },
    {
      "epoch": 0.2236828893403054,
      "grad_norm": 0.09076918742921525,
      "learning_rate": 2.7593507996387673e-05,
      "loss": 0.1964,
      "step": 7240
    },
    {
      "epoch": 0.22399184360735003,
      "grad_norm": 0.1950118432774784,
      "learning_rate": 2.758517557286742e-05,
      "loss": 0.1972,
      "step": 7250
    },
    {
      "epoch": 0.22430079787439464,
      "grad_norm": 0.10617294180572731,
      "learning_rate": 2.7576830011468652e-05,
      "loss": 0.195,
      "step": 7260
    },
    {
      "epoch": 0.22460975214143927,
      "grad_norm": 0.14309770436023087,
      "learning_rate": 2.7568471320903453e-05,
      "loss": 0.1935,
      "step": 7270
    },
    {
      "epoch": 0.22491870640848388,
      "grad_norm": 0.09947273293633722,
      "learning_rate": 2.7560099509897593e-05,
      "loss": 0.2082,
      "step": 7280
    },
    {
      "epoch": 0.22522766067552852,
      "grad_norm": 0.10477746443337488,
      "learning_rate": 2.755171458719055e-05,
      "loss": 0.1893,
      "step": 7290
    },
    {
      "epoch": 0.22553661494257313,
      "grad_norm": 0.11140754503234265,
      "learning_rate": 2.7543316561535493e-05,
      "loss": 0.1911,
      "step": 7300
    },
    {
      "epoch": 0.22584556920961774,
      "grad_norm": 0.10230070448574866,
      "learning_rate": 2.7534905441699255e-05,
      "loss": 0.1921,
      "step": 7310
    },
    {
      "epoch": 0.22615452347666237,
      "grad_norm": 0.10504554261460508,
      "learning_rate": 2.7526481236462355e-05,
      "loss": 0.1832,
      "step": 7320
    },
    {
      "epoch": 0.22646347774370698,
      "grad_norm": 0.09942788623431183,
      "learning_rate": 2.751804395461896e-05,
      "loss": 0.1901,
      "step": 7330
    },
    {
      "epoch": 0.22677243201075162,
      "grad_norm": 0.12870863618383507,
      "learning_rate": 2.7509593604976887e-05,
      "loss": 0.1894,
      "step": 7340
    },
    {
      "epoch": 0.22708138627779623,
      "grad_norm": 0.0906429526198156,
      "learning_rate": 2.750113019635761e-05,
      "loss": 0.1928,
      "step": 7350
    },
    {
      "epoch": 0.22739034054484086,
      "grad_norm": 0.11427297980006616,
      "learning_rate": 2.7492653737596216e-05,
      "loss": 0.2092,
      "step": 7360
    },
    {
      "epoch": 0.22769929481188547,
      "grad_norm": 0.1112130919921356,
      "learning_rate": 2.748416423754143e-05,
      "loss": 0.2022,
      "step": 7370
    },
    {
      "epoch": 0.22800824907893008,
      "grad_norm": 0.14104233187512377,
      "learning_rate": 2.747566170505558e-05,
      "loss": 0.2024,
      "step": 7380
    },
    {
      "epoch": 0.22831720334597472,
      "grad_norm": 0.12499517138025472,
      "learning_rate": 2.746714614901461e-05,
      "loss": 0.1911,
      "step": 7390
    },
    {
      "epoch": 0.22862615761301933,
      "grad_norm": 0.11490983239759463,
      "learning_rate": 2.7458617578308046e-05,
      "loss": 0.1887,
      "step": 7400
    },
    {
      "epoch": 0.22893511188006396,
      "grad_norm": 0.10861871017164962,
      "learning_rate": 2.7450076001839018e-05,
      "loss": 0.1975,
      "step": 7410
    },
    {
      "epoch": 0.22924406614710857,
      "grad_norm": 0.10381821105467194,
      "learning_rate": 2.7441521428524216e-05,
      "loss": 0.1942,
      "step": 7420
    },
    {
      "epoch": 0.2295530204141532,
      "grad_norm": 0.15815281263510628,
      "learning_rate": 2.74329538672939e-05,
      "loss": 0.1975,
      "step": 7430
    },
    {
      "epoch": 0.22986197468119782,
      "grad_norm": 0.13564816041505678,
      "learning_rate": 2.7424373327091897e-05,
      "loss": 0.1886,
      "step": 7440
    },
    {
      "epoch": 0.23017092894824243,
      "grad_norm": 0.13467168641030136,
      "learning_rate": 2.7415779816875585e-05,
      "loss": 0.198,
      "step": 7450
    },
    {
      "epoch": 0.23047988321528706,
      "grad_norm": 0.12859509587750653,
      "learning_rate": 2.7407173345615867e-05,
      "loss": 0.2001,
      "step": 7460
    },
    {
      "epoch": 0.23078883748233167,
      "grad_norm": 0.10624362646867024,
      "learning_rate": 2.7398553922297192e-05,
      "loss": 0.1908,
      "step": 7470
    },
    {
      "epoch": 0.2310977917493763,
      "grad_norm": 0.11668205761717289,
      "learning_rate": 2.7389921555917513e-05,
      "loss": 0.1864,
      "step": 7480
    },
    {
      "epoch": 0.23140674601642092,
      "grad_norm": 0.10783332997141962,
      "learning_rate": 2.738127625548831e-05,
      "loss": 0.1904,
      "step": 7490
    },
    {
      "epoch": 0.23171570028346555,
      "grad_norm": 0.11748499343151125,
      "learning_rate": 2.737261803003456e-05,
      "loss": 0.1864,
      "step": 7500
    },
    {
      "epoch": 0.23202465455051016,
      "grad_norm": 0.13975556960830696,
      "learning_rate": 2.7363946888594723e-05,
      "loss": 0.1934,
      "step": 7510
    },
    {
      "epoch": 0.23233360881755477,
      "grad_norm": 0.1114144356338949,
      "learning_rate": 2.7355262840220768e-05,
      "loss": 0.1898,
      "step": 7520
    },
    {
      "epoch": 0.2326425630845994,
      "grad_norm": 0.19218149019117217,
      "learning_rate": 2.73465658939781e-05,
      "loss": 0.1896,
      "step": 7530
    },
    {
      "epoch": 0.23295151735164402,
      "grad_norm": 0.11827272441099144,
      "learning_rate": 2.733785605894562e-05,
      "loss": 0.1896,
      "step": 7540
    },
    {
      "epoch": 0.23326047161868865,
      "grad_norm": 0.10231517550773084,
      "learning_rate": 2.732913334421567e-05,
      "loss": 0.1902,
      "step": 7550
    },
    {
      "epoch": 0.23356942588573326,
      "grad_norm": 0.13695325334624153,
      "learning_rate": 2.7320397758894045e-05,
      "loss": 0.1986,
      "step": 7560
    },
    {
      "epoch": 0.2338783801527779,
      "grad_norm": 0.09913754561076096,
      "learning_rate": 2.7311649312099964e-05,
      "loss": 0.1898,
      "step": 7570
    },
    {
      "epoch": 0.2341873344198225,
      "grad_norm": 0.10233966911193337,
      "learning_rate": 2.7302888012966085e-05,
      "loss": 0.1958,
      "step": 7580
    },
    {
      "epoch": 0.23449628868686712,
      "grad_norm": 0.11685434890734718,
      "learning_rate": 2.7294113870638474e-05,
      "loss": 0.1902,
      "step": 7590
    },
    {
      "epoch": 0.23480524295391175,
      "grad_norm": 0.10073743758366671,
      "learning_rate": 2.728532689427661e-05,
      "loss": 0.1922,
      "step": 7600
    },
    {
      "epoch": 0.23511419722095636,
      "grad_norm": 0.10684939012097934,
      "learning_rate": 2.727652709305336e-05,
      "loss": 0.1922,
      "step": 7610
    },
    {
      "epoch": 0.235423151488001,
      "grad_norm": 0.09699939241548962,
      "learning_rate": 2.726771447615499e-05,
      "loss": 0.1871,
      "step": 7620
    },
    {
      "epoch": 0.2357321057550456,
      "grad_norm": 0.15601884781782124,
      "learning_rate": 2.7258889052781138e-05,
      "loss": 0.1916,
      "step": 7630
    },
    {
      "epoch": 0.23604106002209024,
      "grad_norm": 0.11755291522610203,
      "learning_rate": 2.725005083214482e-05,
      "loss": 0.1926,
      "step": 7640
    },
    {
      "epoch": 0.23635001428913485,
      "grad_norm": 0.0957888696538469,
      "learning_rate": 2.72411998234724e-05,
      "loss": 0.1918,
      "step": 7650
    },
    {
      "epoch": 0.23665896855617946,
      "grad_norm": 0.10384235779595037,
      "learning_rate": 2.7232336036003595e-05,
      "loss": 0.2016,
      "step": 7660
    },
    {
      "epoch": 0.2369679228232241,
      "grad_norm": 0.10016558170903206,
      "learning_rate": 2.722345947899147e-05,
      "loss": 0.1927,
      "step": 7670
    },
    {
      "epoch": 0.2372768770902687,
      "grad_norm": 0.09514114947045538,
      "learning_rate": 2.7214570161702405e-05,
      "loss": 0.1871,
      "step": 7680
    },
    {
      "epoch": 0.23758583135731334,
      "grad_norm": 0.10949082524512631,
      "learning_rate": 2.720566809341611e-05,
      "loss": 0.1914,
      "step": 7690
    },
    {
      "epoch": 0.23789478562435795,
      "grad_norm": 0.10890151055771378,
      "learning_rate": 2.719675328342561e-05,
      "loss": 0.1908,
      "step": 7700
    },
    {
      "epoch": 0.2382037398914026,
      "grad_norm": 0.10465628872644676,
      "learning_rate": 2.7187825741037232e-05,
      "loss": 0.1882,
      "step": 7710
    },
    {
      "epoch": 0.2385126941584472,
      "grad_norm": 0.10402768513972506,
      "learning_rate": 2.7178885475570576e-05,
      "loss": 0.1926,
      "step": 7720
    },
    {
      "epoch": 0.2388216484254918,
      "grad_norm": 0.10955252072745258,
      "learning_rate": 2.7169932496358542e-05,
      "loss": 0.1911,
      "step": 7730
    },
    {
      "epoch": 0.23913060269253644,
      "grad_norm": 0.08596550377900035,
      "learning_rate": 2.71609668127473e-05,
      "loss": 0.1994,
      "step": 7740
    },
    {
      "epoch": 0.23943955695958105,
      "grad_norm": 0.0987463813479826,
      "learning_rate": 2.7151988434096275e-05,
      "loss": 0.1883,
      "step": 7750
    },
    {
      "epoch": 0.2397485112266257,
      "grad_norm": 0.1064028381898475,
      "learning_rate": 2.7142997369778155e-05,
      "loss": 0.2003,
      "step": 7760
    },
    {
      "epoch": 0.2400574654936703,
      "grad_norm": 0.11893651330649783,
      "learning_rate": 2.713399362917886e-05,
      "loss": 0.1929,
      "step": 7770
    },
    {
      "epoch": 0.24036641976071493,
      "grad_norm": 0.10998904702589943,
      "learning_rate": 2.7124977221697547e-05,
      "loss": 0.1872,
      "step": 7780
    },
    {
      "epoch": 0.24067537402775954,
      "grad_norm": 0.11354255934291793,
      "learning_rate": 2.7115948156746595e-05,
      "loss": 0.1883,
      "step": 7790
    },
    {
      "epoch": 0.24098432829480415,
      "grad_norm": 0.11292108116796575,
      "learning_rate": 2.710690644375161e-05,
      "loss": 0.2198,
      "step": 7800
    },
    {
      "epoch": 0.2412932825618488,
      "grad_norm": 0.10942923077248073,
      "learning_rate": 2.7097852092151376e-05,
      "loss": 0.1867,
      "step": 7810
    },
    {
      "epoch": 0.2416022368288934,
      "grad_norm": 0.11768641645499137,
      "learning_rate": 2.708878511139789e-05,
      "loss": 0.192,
      "step": 7820
    },
    {
      "epoch": 0.24191119109593803,
      "grad_norm": 0.11109350143861754,
      "learning_rate": 2.707970551095633e-05,
      "loss": 0.1913,
      "step": 7830
    },
    {
      "epoch": 0.24222014536298264,
      "grad_norm": 0.11858098291737695,
      "learning_rate": 2.707061330030504e-05,
      "loss": 0.1937,
      "step": 7840
    },
    {
      "epoch": 0.24252909963002728,
      "grad_norm": 0.14887299272500534,
      "learning_rate": 2.7061508488935535e-05,
      "loss": 0.1971,
      "step": 7850
    },
    {
      "epoch": 0.2428380538970719,
      "grad_norm": 0.1106518196749893,
      "learning_rate": 2.7052391086352486e-05,
      "loss": 0.191,
      "step": 7860
    },
    {
      "epoch": 0.2431470081641165,
      "grad_norm": 0.11219662644614184,
      "learning_rate": 2.7043261102073697e-05,
      "loss": 0.1871,
      "step": 7870
    },
    {
      "epoch": 0.24345596243116113,
      "grad_norm": 0.12963570048249912,
      "learning_rate": 2.703411854563012e-05,
      "loss": 0.191,
      "step": 7880
    },
    {
      "epoch": 0.24376491669820574,
      "grad_norm": 0.11378349978891705,
      "learning_rate": 2.702496342656582e-05,
      "loss": 0.1898,
      "step": 7890
    },
    {
      "epoch": 0.24407387096525038,
      "grad_norm": 0.11259272531566963,
      "learning_rate": 2.7015795754437985e-05,
      "loss": 0.1963,
      "step": 7900
    },
    {
      "epoch": 0.244382825232295,
      "grad_norm": 0.10641185316248417,
      "learning_rate": 2.7006615538816903e-05,
      "loss": 0.1863,
      "step": 7910
    },
    {
      "epoch": 0.24469177949933962,
      "grad_norm": 0.13819996798691975,
      "learning_rate": 2.6997422789285958e-05,
      "loss": 0.1986,
      "step": 7920
    },
    {
      "epoch": 0.24500073376638423,
      "grad_norm": 0.15848509465337973,
      "learning_rate": 2.6988217515441615e-05,
      "loss": 0.199,
      "step": 7930
    },
    {
      "epoch": 0.24530968803342884,
      "grad_norm": 0.1117568604112502,
      "learning_rate": 2.6978999726893417e-05,
      "loss": 0.1949,
      "step": 7940
    },
    {
      "epoch": 0.24561864230047348,
      "grad_norm": 0.1292327900706852,
      "learning_rate": 2.6969769433263965e-05,
      "loss": 0.1968,
      "step": 7950
    },
    {
      "epoch": 0.2459275965675181,
      "grad_norm": 0.10906225810666081,
      "learning_rate": 2.6960526644188927e-05,
      "loss": 0.1948,
      "step": 7960
    },
    {
      "epoch": 0.24623655083456272,
      "grad_norm": 0.09872685121132838,
      "learning_rate": 2.695127136931701e-05,
      "loss": 0.1878,
      "step": 7970
    },
    {
      "epoch": 0.24654550510160733,
      "grad_norm": 0.09211463702369019,
      "learning_rate": 2.6942003618309937e-05,
      "loss": 0.1911,
      "step": 7980
    },
    {
      "epoch": 0.24685445936865197,
      "grad_norm": 0.08386993835617362,
      "learning_rate": 2.6932723400842486e-05,
      "loss": 0.1883,
      "step": 7990
    },
    {
      "epoch": 0.24716341363569658,
      "grad_norm": 0.10076472491363259,
      "learning_rate": 2.6923430726602425e-05,
      "loss": 0.191,
      "step": 8000
    },
    {
      "epoch": 0.24747236790274119,
      "grad_norm": 0.10977878502585547,
      "learning_rate": 2.6914125605290542e-05,
      "loss": 0.1906,
      "step": 8010
    },
    {
      "epoch": 0.24778132216978582,
      "grad_norm": 0.10862240634766637,
      "learning_rate": 2.6904808046620604e-05,
      "loss": 0.1856,
      "step": 8020
    },
    {
      "epoch": 0.24809027643683043,
      "grad_norm": 0.11527181158680835,
      "learning_rate": 2.689547806031937e-05,
      "loss": 0.1887,
      "step": 8030
    },
    {
      "epoch": 0.24839923070387507,
      "grad_norm": 0.09557266955615018,
      "learning_rate": 2.6886135656126577e-05,
      "loss": 0.1867,
      "step": 8040
    },
    {
      "epoch": 0.24870818497091968,
      "grad_norm": 0.1021792289141178,
      "learning_rate": 2.6876780843794908e-05,
      "loss": 0.1915,
      "step": 8050
    },
    {
      "epoch": 0.2490171392379643,
      "grad_norm": 0.09753339540571344,
      "learning_rate": 2.6867413633090025e-05,
      "loss": 0.1989,
      "step": 8060
    },
    {
      "epoch": 0.24932609350500892,
      "grad_norm": 0.12230575386492942,
      "learning_rate": 2.6858034033790505e-05,
      "loss": 0.1902,
      "step": 8070
    },
    {
      "epoch": 0.24963504777205353,
      "grad_norm": 0.10760106835552179,
      "learning_rate": 2.684864205568788e-05,
      "loss": 0.1901,
      "step": 8080
    },
    {
      "epoch": 0.24994400203909817,
      "grad_norm": 0.12777365353219655,
      "learning_rate": 2.683923770858659e-05,
      "loss": 0.1861,
      "step": 8090
    },
    {
      "epoch": 0.2502529563061428,
      "grad_norm": 0.11362177095837529,
      "learning_rate": 2.6829821002303997e-05,
      "loss": 0.1914,
      "step": 8100
    },
    {
      "epoch": 0.2505619105731874,
      "grad_norm": 0.16071911627078958,
      "learning_rate": 2.682039194667036e-05,
      "loss": 0.2075,
      "step": 8110
    },
    {
      "epoch": 0.250870864840232,
      "grad_norm": 0.0994911056032934,
      "learning_rate": 2.681095055152883e-05,
      "loss": 0.185,
      "step": 8120
    },
    {
      "epoch": 0.25117981910727666,
      "grad_norm": 0.10651352049455586,
      "learning_rate": 2.6801496826735432e-05,
      "loss": 0.1934,
      "step": 8130
    },
    {
      "epoch": 0.25148877337432124,
      "grad_norm": 0.11171894213037324,
      "learning_rate": 2.6792030782159082e-05,
      "loss": 0.1872,
      "step": 8140
    },
    {
      "epoch": 0.2517977276413659,
      "grad_norm": 0.12181299260141097,
      "learning_rate": 2.6782552427681533e-05,
      "loss": 0.1909,
      "step": 8150
    },
    {
      "epoch": 0.2521066819084105,
      "grad_norm": 0.11060480769569164,
      "learning_rate": 2.6773061773197413e-05,
      "loss": 0.1866,
      "step": 8160
    },
    {
      "epoch": 0.25241563617545515,
      "grad_norm": 0.10113486964251968,
      "learning_rate": 2.6763558828614166e-05,
      "loss": 0.1889,
      "step": 8170
    },
    {
      "epoch": 0.25272459044249973,
      "grad_norm": 0.11633152832284499,
      "learning_rate": 2.6754043603852088e-05,
      "loss": 0.1892,
      "step": 8180
    },
    {
      "epoch": 0.25303354470954437,
      "grad_norm": 0.1129318134656178,
      "learning_rate": 2.6744516108844274e-05,
      "loss": 0.2029,
      "step": 8190
    },
    {
      "epoch": 0.253342498976589,
      "grad_norm": 0.08594486928553212,
      "learning_rate": 2.673497635353664e-05,
      "loss": 0.1853,
      "step": 8200
    },
    {
      "epoch": 0.2536514532436336,
      "grad_norm": 0.10438205925856926,
      "learning_rate": 2.6725424347887903e-05,
      "loss": 0.1888,
      "step": 8210
    },
    {
      "epoch": 0.2539604075106782,
      "grad_norm": 0.0830780927378656,
      "learning_rate": 2.671586010186956e-05,
      "loss": 0.1933,
      "step": 8220
    },
    {
      "epoch": 0.25426936177772286,
      "grad_norm": 0.1654442951133926,
      "learning_rate": 2.670628362546589e-05,
      "loss": 0.2002,
      "step": 8230
    },
    {
      "epoch": 0.2545783160447675,
      "grad_norm": 0.1531186496346408,
      "learning_rate": 2.669669492867394e-05,
      "loss": 0.1959,
      "step": 8240
    },
    {
      "epoch": 0.2548872703118121,
      "grad_norm": 0.10758917016702108,
      "learning_rate": 2.6687094021503515e-05,
      "loss": 0.1877,
      "step": 8250
    },
    {
      "epoch": 0.2551962245788567,
      "grad_norm": 0.0996373769963642,
      "learning_rate": 2.667748091397716e-05,
      "loss": 0.1872,
      "step": 8260
    },
    {
      "epoch": 0.25550517884590135,
      "grad_norm": 0.10931966816599097,
      "learning_rate": 2.666785561613017e-05,
      "loss": 0.1907,
      "step": 8270
    },
    {
      "epoch": 0.25581413311294593,
      "grad_norm": 0.14853693797824316,
      "learning_rate": 2.6658218138010552e-05,
      "loss": 0.1882,
      "step": 8280
    },
    {
      "epoch": 0.25612308737999057,
      "grad_norm": 0.1121180081710152,
      "learning_rate": 2.6648568489679027e-05,
      "loss": 0.1954,
      "step": 8290
    },
    {
      "epoch": 0.2564320416470352,
      "grad_norm": 0.10488948376769945,
      "learning_rate": 2.6638906681209036e-05,
      "loss": 0.1879,
      "step": 8300
    },
    {
      "epoch": 0.25674099591407984,
      "grad_norm": 1.0799296386711583,
      "learning_rate": 2.66292327226867e-05,
      "loss": 0.1971,
      "step": 8310
    },
    {
      "epoch": 0.2570499501811244,
      "grad_norm": 0.08770130701395235,
      "learning_rate": 2.661954662421083e-05,
      "loss": 0.1869,
      "step": 8320
    },
    {
      "epoch": 0.25735890444816906,
      "grad_norm": 0.10937918136719363,
      "learning_rate": 2.6609848395892912e-05,
      "loss": 0.19,
      "step": 8330
    },
    {
      "epoch": 0.2576678587152137,
      "grad_norm": 0.11924512956831727,
      "learning_rate": 2.6600138047857087e-05,
      "loss": 0.1902,
      "step": 8340
    },
    {
      "epoch": 0.2579768129822583,
      "grad_norm": 0.1510606292166939,
      "learning_rate": 2.659041559024015e-05,
      "loss": 0.1999,
      "step": 8350
    },
    {
      "epoch": 0.2582857672493029,
      "grad_norm": 0.14802474236102292,
      "learning_rate": 2.6580681033191548e-05,
      "loss": 0.1944,
      "step": 8360
    },
    {
      "epoch": 0.25859472151634755,
      "grad_norm": 0.09347449504038548,
      "learning_rate": 2.6570934386873343e-05,
      "loss": 0.1885,
      "step": 8370
    },
    {
      "epoch": 0.2589036757833922,
      "grad_norm": 0.09351642486398651,
      "learning_rate": 2.6561175661460226e-05,
      "loss": 0.1861,
      "step": 8380
    },
    {
      "epoch": 0.25921263005043677,
      "grad_norm": 0.10833451271862764,
      "learning_rate": 2.6551404867139503e-05,
      "loss": 0.1863,
      "step": 8390
    },
    {
      "epoch": 0.2595215843174814,
      "grad_norm": 0.20552700992879117,
      "learning_rate": 2.6541622014111058e-05,
      "loss": 0.1948,
      "step": 8400
    },
    {
      "epoch": 0.25983053858452604,
      "grad_norm": 0.11559804825891065,
      "learning_rate": 2.653182711258739e-05,
      "loss": 0.1888,
      "step": 8410
    },
    {
      "epoch": 0.2601394928515706,
      "grad_norm": 0.09789129860045476,
      "learning_rate": 2.6522020172793555e-05,
      "loss": 0.1891,
      "step": 8420
    },
    {
      "epoch": 0.26044844711861526,
      "grad_norm": 0.1238643847309866,
      "learning_rate": 2.6512201204967194e-05,
      "loss": 0.2003,
      "step": 8430
    },
    {
      "epoch": 0.2607574013856599,
      "grad_norm": 0.12050982567076174,
      "learning_rate": 2.6502370219358484e-05,
      "loss": 0.1865,
      "step": 8440
    },
    {
      "epoch": 0.26106635565270453,
      "grad_norm": 0.11492470628163776,
      "learning_rate": 2.6492527226230162e-05,
      "loss": 0.1905,
      "step": 8450
    },
    {
      "epoch": 0.2613753099197491,
      "grad_norm": 0.09696514200967198,
      "learning_rate": 2.6482672235857496e-05,
      "loss": 0.1997,
      "step": 8460
    },
    {
      "epoch": 0.26168426418679375,
      "grad_norm": 0.10480432447934097,
      "learning_rate": 2.6472805258528274e-05,
      "loss": 0.1917,
      "step": 8470
    },
    {
      "epoch": 0.2619932184538384,
      "grad_norm": 0.16726727929445964,
      "learning_rate": 2.6462926304542807e-05,
      "loss": 0.1946,
      "step": 8480
    },
    {
      "epoch": 0.26230217272088296,
      "grad_norm": 0.1228501564007655,
      "learning_rate": 2.6453035384213902e-05,
      "loss": 0.1888,
      "step": 8490
    },
    {
      "epoch": 0.2626111269879276,
      "grad_norm": 0.10757566489326317,
      "learning_rate": 2.644313250786686e-05,
      "loss": 0.1879,
      "step": 8500
    },
    {
      "epoch": 0.26292008125497224,
      "grad_norm": 0.10570638616014276,
      "learning_rate": 2.643321768583945e-05,
      "loss": 0.19,
      "step": 8510
    },
    {
      "epoch": 0.2632290355220169,
      "grad_norm": 0.13579051567335834,
      "learning_rate": 2.6423290928481936e-05,
      "loss": 0.1917,
      "step": 8520
    },
    {
      "epoch": 0.26353798978906146,
      "grad_norm": 0.11825966431712678,
      "learning_rate": 2.6413352246157026e-05,
      "loss": 0.1887,
      "step": 8530
    },
    {
      "epoch": 0.2638469440561061,
      "grad_norm": 0.09504572464949657,
      "learning_rate": 2.6403401649239876e-05,
      "loss": 0.1903,
      "step": 8540
    },
    {
      "epoch": 0.26415589832315073,
      "grad_norm": 0.10563766950029646,
      "learning_rate": 2.6393439148118087e-05,
      "loss": 0.1839,
      "step": 8550
    },
    {
      "epoch": 0.2644648525901953,
      "grad_norm": 0.1189064346208586,
      "learning_rate": 2.6383464753191682e-05,
      "loss": 0.1976,
      "step": 8560
    },
    {
      "epoch": 0.26477380685723995,
      "grad_norm": 0.10491874329682777,
      "learning_rate": 2.6373478474873095e-05,
      "loss": 0.1869,
      "step": 8570
    },
    {
      "epoch": 0.2650827611242846,
      "grad_norm": 0.13054816684170087,
      "learning_rate": 2.6363480323587184e-05,
      "loss": 0.2002,
      "step": 8580
    },
    {
      "epoch": 0.2653917153913292,
      "grad_norm": 0.10884577615051691,
      "learning_rate": 2.6353470309771176e-05,
      "loss": 0.1871,
      "step": 8590
    },
    {
      "epoch": 0.2657006696583738,
      "grad_norm": 0.33459869289623057,
      "learning_rate": 2.63434484438747e-05,
      "loss": 0.2186,
      "step": 8600
    },
    {
      "epoch": 0.26600962392541844,
      "grad_norm": 0.11021312725841806,
      "learning_rate": 2.6333414736359757e-05,
      "loss": 0.1889,
      "step": 8610
    },
    {
      "epoch": 0.2663185781924631,
      "grad_norm": 0.6727696788407207,
      "learning_rate": 2.6323369197700696e-05,
      "loss": 0.1947,
      "step": 8620
    },
    {
      "epoch": 0.26662753245950765,
      "grad_norm": 0.1090738335756151,
      "learning_rate": 2.6313311838384227e-05,
      "loss": 0.1951,
      "step": 8630
    },
    {
      "epoch": 0.2669364867265523,
      "grad_norm": 0.10213255707094637,
      "learning_rate": 2.6303242668909404e-05,
      "loss": 0.1979,
      "step": 8640
    },
    {
      "epoch": 0.2672454409935969,
      "grad_norm": 0.12960536832691863,
      "learning_rate": 2.6293161699787604e-05,
      "loss": 0.1884,
      "step": 8650
    },
    {
      "epoch": 0.26755439526064156,
      "grad_norm": 0.19761273648447258,
      "learning_rate": 2.6283068941542513e-05,
      "loss": 0.1952,
      "step": 8660
    },
    {
      "epoch": 0.26786334952768615,
      "grad_norm": 0.10896656958532387,
      "learning_rate": 2.6272964404710135e-05,
      "loss": 0.188,
      "step": 8670
    },
    {
      "epoch": 0.2681723037947308,
      "grad_norm": 0.13125066011172226,
      "learning_rate": 2.6262848099838777e-05,
      "loss": 0.1992,
      "step": 8680
    },
    {
      "epoch": 0.2684812580617754,
      "grad_norm": 0.13997887897511796,
      "learning_rate": 2.6252720037489012e-05,
      "loss": 0.2042,
      "step": 8690
    },
    {
      "epoch": 0.26879021232882,
      "grad_norm": 0.09697033217541794,
      "learning_rate": 2.6242580228233702e-05,
      "loss": 0.1879,
      "step": 8700
    },
    {
      "epoch": 0.26909916659586464,
      "grad_norm": 0.11741103151081521,
      "learning_rate": 2.6232428682657964e-05,
      "loss": 0.1874,
      "step": 8710
    },
    {
      "epoch": 0.2694081208629093,
      "grad_norm": 0.08964007187317638,
      "learning_rate": 2.6222265411359168e-05,
      "loss": 0.1966,
      "step": 8720
    },
    {
      "epoch": 0.2697170751299539,
      "grad_norm": 0.08888805133889255,
      "learning_rate": 2.621209042494692e-05,
      "loss": 0.188,
      "step": 8730
    },
    {
      "epoch": 0.2700260293969985,
      "grad_norm": 0.09658041854013423,
      "learning_rate": 2.6201903734043074e-05,
      "loss": 0.1888,
      "step": 8740
    },
    {
      "epoch": 0.2703349836640431,
      "grad_norm": 0.11319723780103118,
      "learning_rate": 2.6191705349281675e-05,
      "loss": 0.1868,
      "step": 8750
    },
    {
      "epoch": 0.27064393793108776,
      "grad_norm": 0.10743229763229148,
      "learning_rate": 2.6181495281309003e-05,
      "loss": 0.2103,
      "step": 8760
    },
    {
      "epoch": 0.27095289219813234,
      "grad_norm": 0.14559623069747357,
      "learning_rate": 2.6171273540783514e-05,
      "loss": 0.198,
      "step": 8770
    },
    {
      "epoch": 0.271261846465177,
      "grad_norm": 0.23249937801724405,
      "learning_rate": 2.616104013837586e-05,
      "loss": 0.1895,
      "step": 8780
    },
    {
      "epoch": 0.2715708007322216,
      "grad_norm": 0.09492811965368149,
      "learning_rate": 2.6150795084768862e-05,
      "loss": 0.1892,
      "step": 8790
    },
    {
      "epoch": 0.27187975499926625,
      "grad_norm": 0.09734065538760589,
      "learning_rate": 2.6140538390657506e-05,
      "loss": 0.1999,
      "step": 8800
    },
    {
      "epoch": 0.27218870926631084,
      "grad_norm": 0.0965002759362586,
      "learning_rate": 2.6130270066748924e-05,
      "loss": 0.187,
      "step": 8810
    },
    {
      "epoch": 0.27249766353335547,
      "grad_norm": 0.11472850759947459,
      "learning_rate": 2.6119990123762406e-05,
      "loss": 0.1872,
      "step": 8820
    },
    {
      "epoch": 0.2728066178004001,
      "grad_norm": 0.11213872064955528,
      "learning_rate": 2.6109698572429344e-05,
      "loss": 0.1939,
      "step": 8830
    },
    {
      "epoch": 0.2731155720674447,
      "grad_norm": 0.10841550853694587,
      "learning_rate": 2.6099395423493278e-05,
      "loss": 0.1853,
      "step": 8840
    },
    {
      "epoch": 0.2734245263344893,
      "grad_norm": 0.09602089207539032,
      "learning_rate": 2.6089080687709827e-05,
      "loss": 0.1925,
      "step": 8850
    },
    {
      "epoch": 0.27373348060153396,
      "grad_norm": 0.0872433254648909,
      "learning_rate": 2.6078754375846726e-05,
      "loss": 0.1997,
      "step": 8860
    },
    {
      "epoch": 0.2740424348685786,
      "grad_norm": 0.11114739795765963,
      "learning_rate": 2.6068416498683786e-05,
      "loss": 0.187,
      "step": 8870
    },
    {
      "epoch": 0.2743513891356232,
      "grad_norm": 0.09779538164019706,
      "learning_rate": 2.6058067067012893e-05,
      "loss": 0.1887,
      "step": 8880
    },
    {
      "epoch": 0.2746603434026678,
      "grad_norm": 0.10694174384574226,
      "learning_rate": 2.6047706091637995e-05,
      "loss": 0.1864,
      "step": 8890
    },
    {
      "epoch": 0.27496929766971245,
      "grad_norm": 0.11458632793553461,
      "learning_rate": 2.6037333583375087e-05,
      "loss": 0.1903,
      "step": 8900
    },
    {
      "epoch": 0.27527825193675703,
      "grad_norm": 0.13835046571924234,
      "learning_rate": 2.602694955305221e-05,
      "loss": 0.1941,
      "step": 8910
    },
    {
      "epoch": 0.27558720620380167,
      "grad_norm": 0.09842039219454346,
      "learning_rate": 2.6016554011509428e-05,
      "loss": 0.1849,
      "step": 8920
    },
    {
      "epoch": 0.2758961604708463,
      "grad_norm": 0.11019182951426115,
      "learning_rate": 2.6006146969598822e-05,
      "loss": 0.1879,
      "step": 8930
    },
    {
      "epoch": 0.27620511473789094,
      "grad_norm": 0.10991414892083742,
      "learning_rate": 2.5995728438184485e-05,
      "loss": 0.1904,
      "step": 8940
    },
    {
      "epoch": 0.2765140690049355,
      "grad_norm": 0.0901963469107233,
      "learning_rate": 2.598529842814249e-05,
      "loss": 0.1844,
      "step": 8950
    },
    {
      "epoch": 0.27682302327198016,
      "grad_norm": 0.09683160975761855,
      "learning_rate": 2.597485695036091e-05,
      "loss": 0.1895,
      "step": 8960
    },
    {
      "epoch": 0.2771319775390248,
      "grad_norm": 0.10166995909330437,
      "learning_rate": 2.596440401573978e-05,
      "loss": 0.1857,
      "step": 8970
    },
    {
      "epoch": 0.2774409318060694,
      "grad_norm": 0.10826540384259199,
      "learning_rate": 2.5953939635191087e-05,
      "loss": 0.189,
      "step": 8980
    },
    {
      "epoch": 0.277749886073114,
      "grad_norm": 0.082213056193883,
      "learning_rate": 2.5943463819638792e-05,
      "loss": 0.1893,
      "step": 8990
    },
    {
      "epoch": 0.27805884034015865,
      "grad_norm": 0.10518238040177312,
      "learning_rate": 2.593297658001876e-05,
      "loss": 0.1896,
      "step": 9000
    },
    {
      "epoch": 0.2783677946072033,
      "grad_norm": 0.09821841783566572,
      "learning_rate": 2.592247792727881e-05,
      "loss": 0.1882,
      "step": 9010
    },
    {
      "epoch": 0.27867674887424787,
      "grad_norm": 0.10344017418225036,
      "learning_rate": 2.5911967872378655e-05,
      "loss": 0.1911,
      "step": 9020
    },
    {
      "epoch": 0.2789857031412925,
      "grad_norm": 0.10640760406134377,
      "learning_rate": 2.590144642628993e-05,
      "loss": 0.187,
      "step": 9030
    },
    {
      "epoch": 0.27929465740833714,
      "grad_norm": 0.11580836430543065,
      "learning_rate": 2.5890913599996143e-05,
      "loss": 0.1891,
      "step": 9040
    },
    {
      "epoch": 0.2796036116753817,
      "grad_norm": 0.09905157455059976,
      "learning_rate": 2.588036940449269e-05,
      "loss": 0.1852,
      "step": 9050
    },
    {
      "epoch": 0.27991256594242636,
      "grad_norm": 0.11576126917189689,
      "learning_rate": 2.586981385078684e-05,
      "loss": 0.1933,
      "step": 9060
    },
    {
      "epoch": 0.280221520209471,
      "grad_norm": 0.11856880648441576,
      "learning_rate": 2.5859246949897713e-05,
      "loss": 0.1974,
      "step": 9070
    },
    {
      "epoch": 0.28053047447651563,
      "grad_norm": 0.11182800280519953,
      "learning_rate": 2.5848668712856282e-05,
      "loss": 0.1899,
      "step": 9080
    },
    {
      "epoch": 0.2808394287435602,
      "grad_norm": 0.11527572973477145,
      "learning_rate": 2.5838079150705337e-05,
      "loss": 0.1984,
      "step": 9090
    },
    {
      "epoch": 0.28114838301060485,
      "grad_norm": 0.10445552575840167,
      "learning_rate": 2.582747827449951e-05,
      "loss": 0.1891,
      "step": 9100
    },
    {
      "epoch": 0.2814573372776495,
      "grad_norm": 0.14182003203057386,
      "learning_rate": 2.5816866095305237e-05,
      "loss": 0.1893,
      "step": 9110
    },
    {
      "epoch": 0.28176629154469407,
      "grad_norm": 0.11571260491824123,
      "learning_rate": 2.580624262420074e-05,
      "loss": 0.1916,
      "step": 9120
    },
    {
      "epoch": 0.2820752458117387,
      "grad_norm": 0.10869138832778663,
      "learning_rate": 2.5795607872276057e-05,
      "loss": 0.1869,
      "step": 9130
    },
    {
      "epoch": 0.28238420007878334,
      "grad_norm": 0.10892045289021521,
      "learning_rate": 2.5784961850632977e-05,
      "loss": 0.1864,
      "step": 9140
    },
    {
      "epoch": 0.282693154345828,
      "grad_norm": 0.10479307702793711,
      "learning_rate": 2.577430457038506e-05,
      "loss": 0.2088,
      "step": 9150
    },
    {
      "epoch": 0.28300210861287256,
      "grad_norm": 0.09978126119728896,
      "learning_rate": 2.5763636042657633e-05,
      "loss": 0.1929,
      "step": 9160
    },
    {
      "epoch": 0.2833110628799172,
      "grad_norm": 0.10169798878249496,
      "learning_rate": 2.5752956278587746e-05,
      "loss": 0.1876,
      "step": 9170
    },
    {
      "epoch": 0.28362001714696183,
      "grad_norm": 0.10708803279437269,
      "learning_rate": 2.5742265289324178e-05,
      "loss": 0.1902,
      "step": 9180
    },
    {
      "epoch": 0.2839289714140064,
      "grad_norm": 0.11891559449824145,
      "learning_rate": 2.573156308602745e-05,
      "loss": 0.1947,
      "step": 9190
    },
    {
      "epoch": 0.28423792568105105,
      "grad_norm": 0.1405744328768296,
      "learning_rate": 2.5720849679869765e-05,
      "loss": 0.1908,
      "step": 9200
    },
    {
      "epoch": 0.2845468799480957,
      "grad_norm": 0.10049760138123942,
      "learning_rate": 2.5710125082035034e-05,
      "loss": 0.1881,
      "step": 9210
    },
    {
      "epoch": 0.2848558342151403,
      "grad_norm": 0.09016115351030712,
      "learning_rate": 2.569938930371884e-05,
      "loss": 0.2003,
      "step": 9220
    },
    {
      "epoch": 0.2851647884821849,
      "grad_norm": 0.09967127985671986,
      "learning_rate": 2.568864235612845e-05,
      "loss": 0.1929,
      "step": 9230
    },
    {
      "epoch": 0.28547374274922954,
      "grad_norm": 0.11403371889988403,
      "learning_rate": 2.5677884250482784e-05,
      "loss": 0.1889,
      "step": 9240
    },
    {
      "epoch": 0.2857826970162742,
      "grad_norm": 0.09433735452091101,
      "learning_rate": 2.566711499801241e-05,
      "loss": 0.1842,
      "step": 9250
    },
    {
      "epoch": 0.28609165128331876,
      "grad_norm": 0.10939804567546585,
      "learning_rate": 2.565633460995953e-05,
      "loss": 0.19,
      "step": 9260
    },
    {
      "epoch": 0.2864006055503634,
      "grad_norm": 0.10746949395383892,
      "learning_rate": 2.5645543097577987e-05,
      "loss": 0.2016,
      "step": 9270
    },
    {
      "epoch": 0.28670955981740803,
      "grad_norm": 0.10198863160123904,
      "learning_rate": 2.563474047213321e-05,
      "loss": 0.1863,
      "step": 9280
    },
    {
      "epoch": 0.28701851408445267,
      "grad_norm": 0.10077440810770365,
      "learning_rate": 2.562392674490225e-05,
      "loss": 0.1905,
      "step": 9290
    },
    {
      "epoch": 0.28732746835149725,
      "grad_norm": 0.10855984916578419,
      "learning_rate": 2.561310192717374e-05,
      "loss": 0.1887,
      "step": 9300
    },
    {
      "epoch": 0.2876364226185419,
      "grad_norm": 0.0845171337600054,
      "learning_rate": 2.5602266030247886e-05,
      "loss": 0.1943,
      "step": 9310
    },
    {
      "epoch": 0.2879453768855865,
      "grad_norm": 0.10092603226413406,
      "learning_rate": 2.5591419065436472e-05,
      "loss": 0.187,
      "step": 9320
    },
    {
      "epoch": 0.2882543311526311,
      "grad_norm": 0.1435878134945247,
      "learning_rate": 2.5580561044062827e-05,
      "loss": 0.1936,
      "step": 9330
    },
    {
      "epoch": 0.28856328541967574,
      "grad_norm": 0.09253494052059409,
      "learning_rate": 2.556969197746182e-05,
      "loss": 0.1859,
      "step": 9340
    },
    {
      "epoch": 0.2888722396867204,
      "grad_norm": 8.814361473086858,
      "learning_rate": 2.5558811876979855e-05,
      "loss": 0.2474,
      "step": 9350
    },
    {
      "epoch": 0.289181193953765,
      "grad_norm": 0.15148451030285348,
      "learning_rate": 2.5547920753974863e-05,
      "loss": 0.2022,
      "step": 9360
    },
    {
      "epoch": 0.2894901482208096,
      "grad_norm": 0.09255025901365488,
      "learning_rate": 2.553701861981626e-05,
      "loss": 0.1864,
      "step": 9370
    },
    {
      "epoch": 0.28979910248785423,
      "grad_norm": 0.104459976300163,
      "learning_rate": 2.5526105485884973e-05,
      "loss": 0.191,
      "step": 9380
    },
    {
      "epoch": 0.29010805675489887,
      "grad_norm": 0.10441613403727636,
      "learning_rate": 2.5515181363573406e-05,
      "loss": 0.188,
      "step": 9390
    },
    {
      "epoch": 0.29041701102194345,
      "grad_norm": 0.11765425543594707,
      "learning_rate": 2.5504246264285445e-05,
      "loss": 0.1977,
      "step": 9400
    },
    {
      "epoch": 0.2907259652889881,
      "grad_norm": 0.1306047238430537,
      "learning_rate": 2.5493300199436413e-05,
      "loss": 0.1888,
      "step": 9410
    },
    {
      "epoch": 0.2910349195560327,
      "grad_norm": 0.09658760288385525,
      "learning_rate": 2.5482343180453104e-05,
      "loss": 0.2029,
      "step": 9420
    },
    {
      "epoch": 0.29134387382307736,
      "grad_norm": 0.11503788646311412,
      "learning_rate": 2.5471375218773733e-05,
      "loss": 0.1906,
      "step": 9430
    },
    {
      "epoch": 0.29165282809012194,
      "grad_norm": 0.10380154574814283,
      "learning_rate": 2.5460396325847934e-05,
      "loss": 0.1815,
      "step": 9440
    },
    {
      "epoch": 0.2919617823571666,
      "grad_norm": 0.12772727064183623,
      "learning_rate": 2.5449406513136768e-05,
      "loss": 0.1937,
      "step": 9450
    },
    {
      "epoch": 0.2922707366242112,
      "grad_norm": 0.12065907520857644,
      "learning_rate": 2.543840579211269e-05,
      "loss": 0.1933,
      "step": 9460
    },
    {
      "epoch": 0.2925796908912558,
      "grad_norm": 0.11855525364711275,
      "learning_rate": 2.5427394174259523e-05,
      "loss": 0.1875,
      "step": 9470
    },
    {
      "epoch": 0.29288864515830043,
      "grad_norm": 0.08503805054916591,
      "learning_rate": 2.5416371671072505e-05,
      "loss": 0.1947,
      "step": 9480
    },
    {
      "epoch": 0.29319759942534507,
      "grad_norm": 0.11370146401105172,
      "learning_rate": 2.5405338294058194e-05,
      "loss": 0.2035,
      "step": 9490
    },
    {
      "epoch": 0.2935065536923897,
      "grad_norm": 0.10353773038207591,
      "learning_rate": 2.539429405473453e-05,
      "loss": 0.1928,
      "step": 9500
    },
    {
      "epoch": 0.2938155079594343,
      "grad_norm": 0.09346658558384335,
      "learning_rate": 2.5383238964630782e-05,
      "loss": 0.1857,
      "step": 9510
    },
    {
      "epoch": 0.2941244622264789,
      "grad_norm": 0.09188718153400204,
      "learning_rate": 2.5372173035287542e-05,
      "loss": 0.2027,
      "step": 9520
    },
    {
      "epoch": 0.29443341649352356,
      "grad_norm": 0.09914190766514895,
      "learning_rate": 2.5361096278256726e-05,
      "loss": 0.1871,
      "step": 9530
    },
    {
      "epoch": 0.29474237076056814,
      "grad_norm": 0.07930791004599667,
      "learning_rate": 2.5350008705101547e-05,
      "loss": 0.1863,
      "step": 9540
    },
    {
      "epoch": 0.2950513250276128,
      "grad_norm": 0.11050563294669064,
      "learning_rate": 2.5338910327396513e-05,
      "loss": 0.1868,
      "step": 9550
    },
    {
      "epoch": 0.2953602792946574,
      "grad_norm": 0.09858219263880891,
      "learning_rate": 2.5327801156727402e-05,
      "loss": 0.1878,
      "step": 9560
    },
    {
      "epoch": 0.29566923356170205,
      "grad_norm": 0.18479412983473878,
      "learning_rate": 2.531668120469127e-05,
      "loss": 0.1973,
      "step": 9570
    },
    {
      "epoch": 0.29597818782874663,
      "grad_norm": 0.08603486027474677,
      "learning_rate": 2.5305550482896428e-05,
      "loss": 0.1924,
      "step": 9580
    },
    {
      "epoch": 0.29628714209579127,
      "grad_norm": 0.09563990873301001,
      "learning_rate": 2.529440900296242e-05,
      "loss": 0.1834,
      "step": 9590
    },
    {
      "epoch": 0.2965960963628359,
      "grad_norm": 0.09781713993143305,
      "learning_rate": 2.5283256776520028e-05,
      "loss": 0.1889,
      "step": 9600
    },
    {
      "epoch": 0.2969050506298805,
      "grad_norm": 0.11485623459415951,
      "learning_rate": 2.527209381521125e-05,
      "loss": 0.1954,
      "step": 9610
    },
    {
      "epoch": 0.2972140048969251,
      "grad_norm": 0.09526265527710998,
      "learning_rate": 2.526092013068929e-05,
      "loss": 0.1872,
      "step": 9620
    },
    {
      "epoch": 0.29752295916396976,
      "grad_norm": 0.09604807066414829,
      "learning_rate": 2.5249735734618552e-05,
      "loss": 0.1915,
      "step": 9630
    },
    {
      "epoch": 0.2978319134310144,
      "grad_norm": 0.10545061928010044,
      "learning_rate": 2.5238540638674614e-05,
      "loss": 0.1893,
      "step": 9640
    },
    {
      "epoch": 0.298140867698059,
      "grad_norm": 0.09842714903768751,
      "learning_rate": 2.5227334854544227e-05,
      "loss": 0.1878,
      "step": 9650
    },
    {
      "epoch": 0.2984498219651036,
      "grad_norm": 0.09554273140667502,
      "learning_rate": 2.5216118393925296e-05,
      "loss": 0.1947,
      "step": 9660
    },
    {
      "epoch": 0.29875877623214825,
      "grad_norm": 0.12366791756895854,
      "learning_rate": 2.5204891268526878e-05,
      "loss": 0.1907,
      "step": 9670
    },
    {
      "epoch": 0.29906773049919283,
      "grad_norm": 0.10651055652542657,
      "learning_rate": 2.5193653490069154e-05,
      "loss": 0.1889,
      "step": 9680
    },
    {
      "epoch": 0.29937668476623747,
      "grad_norm": 0.09103700810970648,
      "learning_rate": 2.518240507028344e-05,
      "loss": 0.1884,
      "step": 9690
    },
    {
      "epoch": 0.2996856390332821,
      "grad_norm": 0.11344449962662458,
      "learning_rate": 2.5171146020912155e-05,
      "loss": 0.188,
      "step": 9700
    },
    {
      "epoch": 0.29999459330032674,
      "grad_norm": 0.1018815606614934,
      "learning_rate": 2.5159876353708803e-05,
      "loss": 0.2021,
      "step": 9710
    },
    {
      "epoch": 0.3003035475673713,
      "grad_norm": 0.11178044146446087,
      "learning_rate": 2.5148596080437976e-05,
      "loss": 0.1959,
      "step": 9720
    },
    {
      "epoch": 0.30061250183441596,
      "grad_norm": 0.11026769938443516,
      "learning_rate": 2.5137305212875363e-05,
      "loss": 0.1903,
      "step": 9730
    },
    {
      "epoch": 0.3009214561014606,
      "grad_norm": 0.10047849455876935,
      "learning_rate": 2.5126003762807675e-05,
      "loss": 0.1973,
      "step": 9740
    },
    {
      "epoch": 0.3012304103685052,
      "grad_norm": 0.10931511029328492,
      "learning_rate": 2.5114691742032694e-05,
      "loss": 0.1879,
      "step": 9750
    },
    {
      "epoch": 0.3015393646355498,
      "grad_norm": 0.11494970064041439,
      "learning_rate": 2.5103369162359227e-05,
      "loss": 0.1988,
      "step": 9760
    },
    {
      "epoch": 0.30184831890259445,
      "grad_norm": 0.09125287216445102,
      "learning_rate": 2.5092036035607112e-05,
      "loss": 0.1867,
      "step": 9770
    },
    {
      "epoch": 0.3021572731696391,
      "grad_norm": 0.09819220491579592,
      "learning_rate": 2.5080692373607192e-05,
      "loss": 0.1855,
      "step": 9780
    },
    {
      "epoch": 0.30246622743668367,
      "grad_norm": 0.0976611633634228,
      "learning_rate": 2.5069338188201307e-05,
      "loss": 0.2054,
      "step": 9790
    },
    {
      "epoch": 0.3027751817037283,
      "grad_norm": 0.1160725370149606,
      "learning_rate": 2.505797349124228e-05,
      "loss": 0.1914,
      "step": 9800
    },
    {
      "epoch": 0.30308413597077294,
      "grad_norm": 0.13094878962368744,
      "learning_rate": 2.5046598294593915e-05,
      "loss": 0.1975,
      "step": 9810
    },
    {
      "epoch": 0.3033930902378175,
      "grad_norm": 0.11198671576007047,
      "learning_rate": 2.5035212610130974e-05,
      "loss": 0.1871,
      "step": 9820
    },
    {
      "epoch": 0.30370204450486216,
      "grad_norm": 0.10673072784365434,
      "learning_rate": 2.502381644973917e-05,
      "loss": 0.1835,
      "step": 9830
    },
    {
      "epoch": 0.3040109987719068,
      "grad_norm": 0.09530537911686453,
      "learning_rate": 2.5012409825315137e-05,
      "loss": 0.1864,
      "step": 9840
    },
    {
      "epoch": 0.30431995303895143,
      "grad_norm": 0.11900423902554536,
      "learning_rate": 2.5000992748766454e-05,
      "loss": 0.1866,
      "step": 9850
    },
    {
      "epoch": 0.304628907305996,
      "grad_norm": 0.09929015203337174,
      "learning_rate": 2.4989565232011594e-05,
      "loss": 0.1849,
      "step": 9860
    },
    {
      "epoch": 0.30493786157304065,
      "grad_norm": 0.11285518932683963,
      "learning_rate": 2.4978127286979944e-05,
      "loss": 0.1816,
      "step": 9870
    },
    {
      "epoch": 0.3052468158400853,
      "grad_norm": 0.09456894976012721,
      "learning_rate": 2.4966678925611766e-05,
      "loss": 0.186,
      "step": 9880
    },
    {
      "epoch": 0.30555577010712986,
      "grad_norm": 0.10323981633268041,
      "learning_rate": 2.4955220159858193e-05,
      "loss": 0.1875,
      "step": 9890
    },
    {
      "epoch": 0.3058647243741745,
      "grad_norm": 0.09901307303137544,
      "learning_rate": 2.4943751001681234e-05,
      "loss": 0.1879,
      "step": 9900
    },
    {
      "epoch": 0.30617367864121914,
      "grad_norm": 0.10046839593159085,
      "learning_rate": 2.4932271463053728e-05,
      "loss": 0.1916,
      "step": 9910
    },
    {
      "epoch": 0.3064826329082638,
      "grad_norm": 0.13907399457393424,
      "learning_rate": 2.492078155595937e-05,
      "loss": 0.1915,
      "step": 9920
    },
    {
      "epoch": 0.30679158717530836,
      "grad_norm": 0.0964486904973029,
      "learning_rate": 2.490928129239266e-05,
      "loss": 0.2045,
      "step": 9930
    },
    {
      "epoch": 0.307100541442353,
      "grad_norm": 0.1718161375236824,
      "learning_rate": 2.4897770684358927e-05,
      "loss": 0.2027,
      "step": 9940
    },
    {
      "epoch": 0.30740949570939763,
      "grad_norm": 0.11382680564698643,
      "learning_rate": 2.488624974387429e-05,
      "loss": 0.1887,
      "step": 9950
    },
    {
      "epoch": 0.3077184499764422,
      "grad_norm": 0.09241856142612667,
      "learning_rate": 2.4874718482965647e-05,
      "loss": 0.1868,
      "step": 9960
    },
    {
      "epoch": 0.30802740424348685,
      "grad_norm": 0.10912190096338768,
      "learning_rate": 2.486317691367068e-05,
      "loss": 0.1895,
      "step": 9970
    },
    {
      "epoch": 0.3083363585105315,
      "grad_norm": 0.08454020533402198,
      "learning_rate": 2.4851625048037833e-05,
      "loss": 0.1918,
      "step": 9980
    },
    {
      "epoch": 0.3086453127775761,
      "grad_norm": 0.09952497416826764,
      "learning_rate": 2.484006289812629e-05,
      "loss": 0.1856,
      "step": 9990
    },
    {
      "epoch": 0.3089542670446207,
      "grad_norm": 0.11209923667424684,
      "learning_rate": 2.482849047600598e-05,
      "loss": 0.1901,
      "step": 10000
    },
    {
      "epoch": 0.30926322131166534,
      "grad_norm": 0.09340912080530649,
      "learning_rate": 2.481690779375755e-05,
      "loss": 0.1936,
      "step": 10010
    },
    {
      "epoch": 0.30957217557871,
      "grad_norm": 0.09136718288585276,
      "learning_rate": 2.4805314863472355e-05,
      "loss": 0.1934,
      "step": 10020
    },
    {
      "epoch": 0.30988112984575455,
      "grad_norm": 0.09725966361145463,
      "learning_rate": 2.4793711697252453e-05,
      "loss": 0.1977,
      "step": 10030
    },
    {
      "epoch": 0.3101900841127992,
      "grad_norm": 0.14851332633478187,
      "learning_rate": 2.478209830721059e-05,
      "loss": 0.1948,
      "step": 10040
    },
    {
      "epoch": 0.31049903837984383,
      "grad_norm": 0.09331412732773223,
      "learning_rate": 2.4770474705470176e-05,
      "loss": 0.1876,
      "step": 10050
    },
    {
      "epoch": 0.31080799264688846,
      "grad_norm": 0.0985111396819373,
      "learning_rate": 2.4758840904165293e-05,
      "loss": 0.1889,
      "step": 10060
    },
    {
      "epoch": 0.31111694691393305,
      "grad_norm": 0.1074244405660999,
      "learning_rate": 2.4747196915440654e-05,
      "loss": 0.1878,
      "step": 10070
    },
    {
      "epoch": 0.3114259011809777,
      "grad_norm": 0.11270406523031921,
      "learning_rate": 2.4735542751451624e-05,
      "loss": 0.1903,
      "step": 10080
    },
    {
      "epoch": 0.3117348554480223,
      "grad_norm": 0.14444786045852906,
      "learning_rate": 2.4723878424364183e-05,
      "loss": 0.1926,
      "step": 10090
    },
    {
      "epoch": 0.3120438097150669,
      "grad_norm": 0.13199514765736367,
      "learning_rate": 2.4712203946354918e-05,
      "loss": 0.1916,
      "step": 10100
    },
    {
      "epoch": 0.31235276398211154,
      "grad_norm": 0.1102216519163516,
      "learning_rate": 2.4700519329611013e-05,
      "loss": 0.1875,
      "step": 10110
    },
    {
      "epoch": 0.3126617182491562,
      "grad_norm": 0.08172713783176928,
      "learning_rate": 2.4688824586330242e-05,
      "loss": 0.2014,
      "step": 10120
    },
    {
      "epoch": 0.3129706725162008,
      "grad_norm": 0.08766853575869235,
      "learning_rate": 2.4677119728720944e-05,
      "loss": 0.1948,
      "step": 10130
    },
    {
      "epoch": 0.3132796267832454,
      "grad_norm": 0.0864321078467092,
      "learning_rate": 2.466540476900202e-05,
      "loss": 0.1911,
      "step": 10140
    },
    {
      "epoch": 0.31358858105029,
      "grad_norm": 0.13369147149520363,
      "learning_rate": 2.4653679719402907e-05,
      "loss": 0.1879,
      "step": 10150
    },
    {
      "epoch": 0.31389753531733466,
      "grad_norm": 0.10138950969987723,
      "learning_rate": 2.4641944592163595e-05,
      "loss": 0.1926,
      "step": 10160
    },
    {
      "epoch": 0.31420648958437924,
      "grad_norm": 0.1131289750323316,
      "learning_rate": 2.4630199399534576e-05,
      "loss": 0.189,
      "step": 10170
    },
    {
      "epoch": 0.3145154438514239,
      "grad_norm": 0.08786637618867353,
      "learning_rate": 2.461844415377686e-05,
      "loss": 0.1958,
      "step": 10180
    },
    {
      "epoch": 0.3148243981184685,
      "grad_norm": 0.09390997282261213,
      "learning_rate": 2.4606678867161946e-05,
      "loss": 0.1945,
      "step": 10190
    },
    {
      "epoch": 0.31513335238551315,
      "grad_norm": 0.1067984321824277,
      "learning_rate": 2.4594903551971814e-05,
      "loss": 0.1861,
      "step": 10200
    },
    {
      "epoch": 0.31544230665255774,
      "grad_norm": 0.10431041990990578,
      "learning_rate": 2.458311822049892e-05,
      "loss": 0.1897,
      "step": 10210
    },
    {
      "epoch": 0.3157512609196024,
      "grad_norm": 0.08883021936643981,
      "learning_rate": 2.4571322885046166e-05,
      "loss": 0.1864,
      "step": 10220
    },
    {
      "epoch": 0.316060215186647,
      "grad_norm": 0.08876563960489203,
      "learning_rate": 2.4559517557926908e-05,
      "loss": 0.2012,
      "step": 10230
    },
    {
      "epoch": 0.3163691694536916,
      "grad_norm": 0.13279127653095946,
      "learning_rate": 2.4547702251464926e-05,
      "loss": 0.1951,
      "step": 10240
    },
    {
      "epoch": 0.3166781237207362,
      "grad_norm": 0.08223666907358407,
      "learning_rate": 2.453587697799442e-05,
      "loss": 0.1928,
      "step": 10250
    },
    {
      "epoch": 0.31698707798778086,
      "grad_norm": 0.1462141674424809,
      "learning_rate": 2.4524041749859992e-05,
      "loss": 0.2007,
      "step": 10260
    },
    {
      "epoch": 0.3172960322548255,
      "grad_norm": 0.08504556283374634,
      "learning_rate": 2.4512196579416636e-05,
      "loss": 0.1899,
      "step": 10270
    },
    {
      "epoch": 0.3176049865218701,
      "grad_norm": 0.08145333178233713,
      "learning_rate": 2.4500341479029734e-05,
      "loss": 0.1936,
      "step": 10280
    },
    {
      "epoch": 0.3179139407889147,
      "grad_norm": 0.11105143484390026,
      "learning_rate": 2.4488476461075014e-05,
      "loss": 0.187,
      "step": 10290
    },
    {
      "epoch": 0.31822289505595935,
      "grad_norm": 0.10120758320452351,
      "learning_rate": 2.4476601537938585e-05,
      "loss": 0.1976,
      "step": 10300
    },
    {
      "epoch": 0.31853184932300393,
      "grad_norm": 0.09868681939796119,
      "learning_rate": 2.4464716722016868e-05,
      "loss": 0.187,
      "step": 10310
    },
    {
      "epoch": 0.31884080359004857,
      "grad_norm": 0.11260503884674848,
      "learning_rate": 2.4452822025716633e-05,
      "loss": 0.1894,
      "step": 10320
    },
    {
      "epoch": 0.3191497578570932,
      "grad_norm": 0.11147937208027198,
      "learning_rate": 2.444091746145495e-05,
      "loss": 0.187,
      "step": 10330
    },
    {
      "epoch": 0.31945871212413784,
      "grad_norm": 0.10292959866660813,
      "learning_rate": 2.44290030416592e-05,
      "loss": 0.1834,
      "step": 10340
    },
    {
      "epoch": 0.3197676663911824,
      "grad_norm": 0.10594382495485939,
      "learning_rate": 2.4417078778767042e-05,
      "loss": 0.1876,
      "step": 10350
    },
    {
      "epoch": 0.32007662065822706,
      "grad_norm": 0.10032254310574651,
      "learning_rate": 2.4405144685226423e-05,
      "loss": 0.1916,
      "step": 10360
    },
    {
      "epoch": 0.3203855749252717,
      "grad_norm": 0.10627717057008569,
      "learning_rate": 2.439320077349554e-05,
      "loss": 0.1861,
      "step": 10370
    },
    {
      "epoch": 0.3206945291923163,
      "grad_norm": 0.13099542633637604,
      "learning_rate": 2.438124705604285e-05,
      "loss": 0.1884,
      "step": 10380
    },
    {
      "epoch": 0.3210034834593609,
      "grad_norm": 0.10802554360054083,
      "learning_rate": 2.436928354534704e-05,
      "loss": 0.2024,
      "step": 10390
    },
    {
      "epoch": 0.32131243772640555,
      "grad_norm": 0.08637551242366232,
      "learning_rate": 2.4357310253897024e-05,
      "loss": 0.1915,
      "step": 10400
    },
    {
      "epoch": 0.3216213919934502,
      "grad_norm": 0.0898888594607148,
      "learning_rate": 2.434532719419192e-05,
      "loss": 0.1884,
      "step": 10410
    },
    {
      "epoch": 0.32193034626049477,
      "grad_norm": 0.08570850184237087,
      "learning_rate": 2.4333334378741054e-05,
      "loss": 0.1911,
      "step": 10420
    },
    {
      "epoch": 0.3222393005275394,
      "grad_norm": 0.10249856361831708,
      "learning_rate": 2.4321331820063918e-05,
      "loss": 0.1982,
      "step": 10430
    },
    {
      "epoch": 0.32254825479458404,
      "grad_norm": 0.10257845052867674,
      "learning_rate": 2.43093195306902e-05,
      "loss": 0.1951,
      "step": 10440
    },
    {
      "epoch": 0.3228572090616286,
      "grad_norm": 0.10229604671643684,
      "learning_rate": 2.429729752315973e-05,
      "loss": 0.1887,
      "step": 10450
    },
    {
      "epoch": 0.32316616332867326,
      "grad_norm": 0.10314819818016309,
      "learning_rate": 2.428526581002248e-05,
      "loss": 0.1854,
      "step": 10460
    },
    {
      "epoch": 0.3234751175957179,
      "grad_norm": 0.11144197702516244,
      "learning_rate": 2.4273224403838562e-05,
      "loss": 0.1837,
      "step": 10470
    },
    {
      "epoch": 0.32378407186276253,
      "grad_norm": 0.08803908428000846,
      "learning_rate": 2.4261173317178206e-05,
      "loss": 0.1862,
      "step": 10480
    },
    {
      "epoch": 0.3240930261298071,
      "grad_norm": 0.09657899131923031,
      "learning_rate": 2.4249112562621744e-05,
      "loss": 0.1846,
      "step": 10490
    },
    {
      "epoch": 0.32440198039685175,
      "grad_norm": 0.08073177587693156,
      "learning_rate": 2.4237042152759604e-05,
      "loss": 0.1835,
      "step": 10500
    },
    {
      "epoch": 0.3247109346638964,
      "grad_norm": 0.09804505805837463,
      "learning_rate": 2.4224962100192297e-05,
      "loss": 0.1988,
      "step": 10510
    },
    {
      "epoch": 0.32501988893094097,
      "grad_norm": 0.09316989255139811,
      "learning_rate": 2.4212872417530387e-05,
      "loss": 0.1875,
      "step": 10520
    },
    {
      "epoch": 0.3253288431979856,
      "grad_norm": 0.10100730513827896,
      "learning_rate": 2.42007731173945e-05,
      "loss": 0.1871,
      "step": 10530
    },
    {
      "epoch": 0.32563779746503024,
      "grad_norm": 0.11426894363762669,
      "learning_rate": 2.4188664212415303e-05,
      "loss": 0.1867,
      "step": 10540
    },
    {
      "epoch": 0.3259467517320749,
      "grad_norm": 0.08657873556350776,
      "learning_rate": 2.4176545715233487e-05,
      "loss": 0.1949,
      "step": 10550
    },
    {
      "epoch": 0.32625570599911946,
      "grad_norm": 0.11819434882118317,
      "learning_rate": 2.416441763849976e-05,
      "loss": 0.1944,
      "step": 10560
    },
    {
      "epoch": 0.3265646602661641,
      "grad_norm": 0.1075874849306456,
      "learning_rate": 2.4152279994874825e-05,
      "loss": 0.1959,
      "step": 10570
    },
    {
      "epoch": 0.32687361453320873,
      "grad_norm": 0.09047038439088448,
      "learning_rate": 2.4140132797029374e-05,
      "loss": 0.2071,
      "step": 10580
    },
    {
      "epoch": 0.3271825688002533,
      "grad_norm": 0.12922438663777203,
      "learning_rate": 2.4127976057644066e-05,
      "loss": 0.1866,
      "step": 10590
    },
    {
      "epoch": 0.32749152306729795,
      "grad_norm": 0.09071465290826776,
      "learning_rate": 2.4115809789409538e-05,
      "loss": 0.1968,
      "step": 10600
    },
    {
      "epoch": 0.3278004773343426,
      "grad_norm": 0.09615055209291347,
      "learning_rate": 2.4103634005026362e-05,
      "loss": 0.1942,
      "step": 10610
    },
    {
      "epoch": 0.3281094316013872,
      "grad_norm": 0.11297404179659212,
      "learning_rate": 2.4091448717205045e-05,
      "loss": 0.1982,
      "step": 10620
    },
    {
      "epoch": 0.3284183858684318,
      "grad_norm": 0.08103250825755323,
      "learning_rate": 2.4079253938666015e-05,
      "loss": 0.1883,
      "step": 10630
    },
    {
      "epoch": 0.32872734013547644,
      "grad_norm": 0.10027791187893295,
      "learning_rate": 2.406704968213961e-05,
      "loss": 0.1934,
      "step": 10640
    },
    {
      "epoch": 0.3290362944025211,
      "grad_norm": 0.10728333469606902,
      "learning_rate": 2.4054835960366055e-05,
      "loss": 0.1956,
      "step": 10650
    },
    {
      "epoch": 0.32934524866956566,
      "grad_norm": 0.10145867895118323,
      "learning_rate": 2.404261278609547e-05,
      "loss": 0.1842,
      "step": 10660
    },
    {
      "epoch": 0.3296542029366103,
      "grad_norm": 0.08577620774947799,
      "learning_rate": 2.403038017208783e-05,
      "loss": 0.19,
      "step": 10670
    },
    {
      "epoch": 0.32996315720365493,
      "grad_norm": 0.11295170789310942,
      "learning_rate": 2.4018138131112972e-05,
      "loss": 0.198,
      "step": 10680
    },
    {
      "epoch": 0.33027211147069957,
      "grad_norm": 0.09213130233065495,
      "learning_rate": 2.4005886675950562e-05,
      "loss": 0.1901,
      "step": 10690
    },
    {
      "epoch": 0.33058106573774415,
      "grad_norm": 0.09202019556291367,
      "learning_rate": 2.3993625819390114e-05,
      "loss": 0.1883,
      "step": 10700
    },
    {
      "epoch": 0.3308900200047888,
      "grad_norm": 0.0873260575047738,
      "learning_rate": 2.398135557423093e-05,
      "loss": 0.1859,
      "step": 10710
    },
    {
      "epoch": 0.3311989742718334,
      "grad_norm": 0.1953741414952931,
      "learning_rate": 2.3969075953282143e-05,
      "loss": 0.1905,
      "step": 10720
    },
    {
      "epoch": 0.331507928538878,
      "grad_norm": 0.08764013609672623,
      "learning_rate": 2.3956786969362647e-05,
      "loss": 0.1857,
      "step": 10730
    },
    {
      "epoch": 0.33181688280592264,
      "grad_norm": 0.08642787068756606,
      "learning_rate": 2.3944488635301122e-05,
      "loss": 0.1875,
      "step": 10740
    },
    {
      "epoch": 0.3321258370729673,
      "grad_norm": 0.0940671293624528,
      "learning_rate": 2.3932180963936013e-05,
      "loss": 0.1833,
      "step": 10750
    },
    {
      "epoch": 0.3324347913400119,
      "grad_norm": 0.1489525719605523,
      "learning_rate": 2.39198639681155e-05,
      "loss": 0.1946,
      "step": 10760
    },
    {
      "epoch": 0.3327437456070565,
      "grad_norm": 0.08370747837677327,
      "learning_rate": 2.3907537660697512e-05,
      "loss": 0.1844,
      "step": 10770
    },
    {
      "epoch": 0.33305269987410113,
      "grad_norm": 0.08082000819454536,
      "learning_rate": 2.3895202054549684e-05,
      "loss": 0.1946,
      "step": 10780
    },
    {
      "epoch": 0.33336165414114577,
      "grad_norm": 0.08554354991880381,
      "learning_rate": 2.3882857162549374e-05,
      "loss": 0.1899,
      "step": 10790
    },
    {
      "epoch": 0.33367060840819035,
      "grad_norm": 0.10646003301479967,
      "learning_rate": 2.3870502997583614e-05,
      "loss": 0.1873,
      "step": 10800
    },
    {
      "epoch": 0.333979562675235,
      "grad_norm": 0.11072901291957145,
      "learning_rate": 2.385813957254913e-05,
      "loss": 0.2096,
      "step": 10810
    },
    {
      "epoch": 0.3342885169422796,
      "grad_norm": 0.10355008643861074,
      "learning_rate": 2.384576690035231e-05,
      "loss": 0.1874,
      "step": 10820
    },
    {
      "epoch": 0.33459747120932426,
      "grad_norm": 0.08504136163217271,
      "learning_rate": 2.38333849939092e-05,
      "loss": 0.1892,
      "step": 10830
    },
    {
      "epoch": 0.33490642547636884,
      "grad_norm": 0.11696581059408875,
      "learning_rate": 2.3820993866145485e-05,
      "loss": 0.1831,
      "step": 10840
    },
    {
      "epoch": 0.3352153797434135,
      "grad_norm": 0.12789603723509432,
      "learning_rate": 2.3808593529996462e-05,
      "loss": 0.187,
      "step": 10850
    },
    {
      "epoch": 0.3355243340104581,
      "grad_norm": 0.09406362406435059,
      "learning_rate": 2.3796183998407067e-05,
      "loss": 0.1956,
      "step": 10860
    },
    {
      "epoch": 0.3358332882775027,
      "grad_norm": 0.0963012641045247,
      "learning_rate": 2.378376528433181e-05,
      "loss": 0.1893,
      "step": 10870
    },
    {
      "epoch": 0.33614224254454733,
      "grad_norm": 0.09137989792286952,
      "learning_rate": 2.3771337400734803e-05,
      "loss": 0.1945,
      "step": 10880
    },
    {
      "epoch": 0.33645119681159197,
      "grad_norm": 0.10276076002740937,
      "learning_rate": 2.375890036058972e-05,
      "loss": 0.1935,
      "step": 10890
    },
    {
      "epoch": 0.3367601510786366,
      "grad_norm": 0.10542029603064709,
      "learning_rate": 2.374645417687981e-05,
      "loss": 0.183,
      "step": 10900
    },
    {
      "epoch": 0.3370691053456812,
      "grad_norm": 0.15502207119275505,
      "learning_rate": 2.373399886259784e-05,
      "loss": 0.1916,
      "step": 10910
    },
    {
      "epoch": 0.3373780596127258,
      "grad_norm": 0.0902706320845401,
      "learning_rate": 2.372153443074613e-05,
      "loss": 0.1868,
      "step": 10920
    },
    {
      "epoch": 0.33768701387977046,
      "grad_norm": 0.10491192336946742,
      "learning_rate": 2.3709060894336516e-05,
      "loss": 0.1839,
      "step": 10930
    },
    {
      "epoch": 0.33799596814681504,
      "grad_norm": 0.09407652570377974,
      "learning_rate": 2.3696578266390333e-05,
      "loss": 0.1951,
      "step": 10940
    },
    {
      "epoch": 0.3383049224138597,
      "grad_norm": 0.10131817320848324,
      "learning_rate": 2.368408655993841e-05,
      "loss": 0.185,
      "step": 10950
    },
    {
      "epoch": 0.3386138766809043,
      "grad_norm": 0.09312775430656653,
      "learning_rate": 2.367158578802104e-05,
      "loss": 0.1804,
      "step": 10960
    },
    {
      "epoch": 0.33892283094794895,
      "grad_norm": 0.08868384788858018,
      "learning_rate": 2.3659075963688005e-05,
      "loss": 0.1879,
      "step": 10970
    },
    {
      "epoch": 0.33923178521499353,
      "grad_norm": 0.12383857112167239,
      "learning_rate": 2.364655709999852e-05,
      "loss": 0.1915,
      "step": 10980
    },
    {
      "epoch": 0.33954073948203817,
      "grad_norm": 0.09606980612192445,
      "learning_rate": 2.3634029210021234e-05,
      "loss": 0.1831,
      "step": 10990
    },
    {
      "epoch": 0.3398496937490828,
      "grad_norm": 0.09531477336107455,
      "learning_rate": 2.3621492306834228e-05,
      "loss": 0.1931,
      "step": 11000
    },
    {
      "epoch": 0.3401586480161274,
      "grad_norm": 0.09670879312647006,
      "learning_rate": 2.3608946403524988e-05,
      "loss": 0.1943,
      "step": 11010
    },
    {
      "epoch": 0.340467602283172,
      "grad_norm": 0.11263160779466505,
      "learning_rate": 2.3596391513190393e-05,
      "loss": 0.1977,
      "step": 11020
    },
    {
      "epoch": 0.34077655655021666,
      "grad_norm": 0.09297761654801473,
      "learning_rate": 2.3583827648936715e-05,
      "loss": 0.1889,
      "step": 11030
    },
    {
      "epoch": 0.3410855108172613,
      "grad_norm": 0.13797057326064396,
      "learning_rate": 2.3571254823879573e-05,
      "loss": 0.1962,
      "step": 11040
    },
    {
      "epoch": 0.3413944650843059,
      "grad_norm": 0.08912817358442929,
      "learning_rate": 2.355867305114396e-05,
      "loss": 0.193,
      "step": 11050
    },
    {
      "epoch": 0.3417034193513505,
      "grad_norm": 0.09595013583838197,
      "learning_rate": 2.3546082343864198e-05,
      "loss": 0.1848,
      "step": 11060
    },
    {
      "epoch": 0.34201237361839515,
      "grad_norm": 0.10821307526466348,
      "learning_rate": 2.3533482715183946e-05,
      "loss": 0.1895,
      "step": 11070
    },
    {
      "epoch": 0.34232132788543973,
      "grad_norm": 0.11873312543475707,
      "learning_rate": 2.3520874178256165e-05,
      "loss": 0.1867,
      "step": 11080
    },
    {
      "epoch": 0.34263028215248437,
      "grad_norm": 0.11502975717030715,
      "learning_rate": 2.350825674624312e-05,
      "loss": 0.1893,
      "step": 11090
    },
    {
      "epoch": 0.342939236419529,
      "grad_norm": 0.09755080794306602,
      "learning_rate": 2.3495630432316368e-05,
      "loss": 0.1897,
      "step": 11100
    },
    {
      "epoch": 0.34324819068657364,
      "grad_norm": 0.08872588077759884,
      "learning_rate": 2.3482995249656725e-05,
      "loss": 0.1986,
      "step": 11110
    },
    {
      "epoch": 0.3435571449536182,
      "grad_norm": 0.09838896910229696,
      "learning_rate": 2.3470351211454285e-05,
      "loss": 0.1908,
      "step": 11120
    },
    {
      "epoch": 0.34386609922066286,
      "grad_norm": 0.0923233551081439,
      "learning_rate": 2.3457698330908364e-05,
      "loss": 0.2002,
      "step": 11130
    },
    {
      "epoch": 0.3441750534877075,
      "grad_norm": 0.08737552496078425,
      "learning_rate": 2.344503662122752e-05,
      "loss": 0.1895,
      "step": 11140
    },
    {
      "epoch": 0.3444840077547521,
      "grad_norm": 0.10668889181697425,
      "learning_rate": 2.3432366095629528e-05,
      "loss": 0.1888,
      "step": 11150
    },
    {
      "epoch": 0.3447929620217967,
      "grad_norm": 0.08291984012290848,
      "learning_rate": 2.3419686767341367e-05,
      "loss": 0.1987,
      "step": 11160
    },
    {
      "epoch": 0.34510191628884135,
      "grad_norm": 0.10764632441930712,
      "learning_rate": 2.3406998649599202e-05,
      "loss": 0.1832,
      "step": 11170
    },
    {
      "epoch": 0.345410870555886,
      "grad_norm": 0.09017312726891885,
      "learning_rate": 2.3394301755648375e-05,
      "loss": 0.192,
      "step": 11180
    },
    {
      "epoch": 0.34571982482293057,
      "grad_norm": 0.1324785929205862,
      "learning_rate": 2.3381596098743387e-05,
      "loss": 0.192,
      "step": 11190
    },
    {
      "epoch": 0.3460287790899752,
      "grad_norm": 0.10070251954726174,
      "learning_rate": 2.336888169214789e-05,
      "loss": 0.1869,
      "step": 11200
    },
    {
      "epoch": 0.34633773335701984,
      "grad_norm": 0.08935782758742528,
      "learning_rate": 2.335615854913467e-05,
      "loss": 0.1928,
      "step": 11210
    },
    {
      "epoch": 0.3466466876240644,
      "grad_norm": 0.0855348666768417,
      "learning_rate": 2.3343426682985625e-05,
      "loss": 0.1872,
      "step": 11220
    },
    {
      "epoch": 0.34695564189110906,
      "grad_norm": 0.09737930135217468,
      "learning_rate": 2.333068610699178e-05,
      "loss": 0.1869,
      "step": 11230
    },
    {
      "epoch": 0.3472645961581537,
      "grad_norm": 0.11069764429002865,
      "learning_rate": 2.3317936834453228e-05,
      "loss": 0.1829,
      "step": 11240
    },
    {
      "epoch": 0.34757355042519833,
      "grad_norm": 0.0924798319053801,
      "learning_rate": 2.3305178878679153e-05,
      "loss": 0.1981,
      "step": 11250
    },
    {
      "epoch": 0.3478825046922429,
      "grad_norm": 0.10323547298579855,
      "learning_rate": 2.3292412252987806e-05,
      "loss": 0.1915,
      "step": 11260
    },
    {
      "epoch": 0.34819145895928755,
      "grad_norm": 0.08749789524208305,
      "learning_rate": 2.327963697070648e-05,
      "loss": 0.1927,
      "step": 11270
    },
    {
      "epoch": 0.3485004132263322,
      "grad_norm": 0.09480047005422201,
      "learning_rate": 2.326685304517151e-05,
      "loss": 0.1872,
      "step": 11280
    },
    {
      "epoch": 0.34880936749337677,
      "grad_norm": 0.10359229819355556,
      "learning_rate": 2.325406048972826e-05,
      "loss": 0.1959,
      "step": 11290
    },
    {
      "epoch": 0.3491183217604214,
      "grad_norm": 0.1041559744501966,
      "learning_rate": 2.324125931773108e-05,
      "loss": 0.1916,
      "step": 11300
    },
    {
      "epoch": 0.34942727602746604,
      "grad_norm": 0.08958972823331576,
      "learning_rate": 2.3228449542543343e-05,
      "loss": 0.1824,
      "step": 11310
    },
    {
      "epoch": 0.3497362302945107,
      "grad_norm": 0.11017769684432532,
      "learning_rate": 2.321563117753739e-05,
      "loss": 0.1901,
      "step": 11320
    },
    {
      "epoch": 0.35004518456155526,
      "grad_norm": 0.09160695776194668,
      "learning_rate": 2.3202804236094525e-05,
      "loss": 0.1944,
      "step": 11330
    },
    {
      "epoch": 0.3503541388285999,
      "grad_norm": 0.08626306454489814,
      "learning_rate": 2.3189968731605015e-05,
      "loss": 0.1876,
      "step": 11340
    },
    {
      "epoch": 0.35066309309564453,
      "grad_norm": 0.0886446415178499,
      "learning_rate": 2.317712467746806e-05,
      "loss": 0.1877,
      "step": 11350
    },
    {
      "epoch": 0.35097204736268917,
      "grad_norm": 0.10801126161955527,
      "learning_rate": 2.3164272087091783e-05,
      "loss": 0.1937,
      "step": 11360
    },
    {
      "epoch": 0.35128100162973375,
      "grad_norm": 0.09010158779533652,
      "learning_rate": 2.3151410973893225e-05,
      "loss": 0.1921,
      "step": 11370
    },
    {
      "epoch": 0.3515899558967784,
      "grad_norm": 0.10195229727148444,
      "learning_rate": 2.313854135129832e-05,
      "loss": 0.1874,
      "step": 11380
    },
    {
      "epoch": 0.351898910163823,
      "grad_norm": 0.10378462738337658,
      "learning_rate": 2.3125663232741882e-05,
      "loss": 0.19,
      "step": 11390
    },
    {
      "epoch": 0.3522078644308676,
      "grad_norm": 0.09872375304401278,
      "learning_rate": 2.31127766316676e-05,
      "loss": 0.1974,
      "step": 11400
    },
    {
      "epoch": 0.35251681869791224,
      "grad_norm": 0.09793706394169703,
      "learning_rate": 2.3099881561528018e-05,
      "loss": 0.1858,
      "step": 11410
    },
    {
      "epoch": 0.3528257729649569,
      "grad_norm": 0.148466539275305,
      "learning_rate": 2.3086978035784517e-05,
      "loss": 0.1995,
      "step": 11420
    },
    {
      "epoch": 0.3531347272320015,
      "grad_norm": 0.0811008808934049,
      "learning_rate": 2.307406606790731e-05,
      "loss": 0.1937,
      "step": 11430
    },
    {
      "epoch": 0.3534436814990461,
      "grad_norm": 0.12512398588345597,
      "learning_rate": 2.3061145671375416e-05,
      "loss": 0.1855,
      "step": 11440
    },
    {
      "epoch": 0.35375263576609073,
      "grad_norm": 0.17576087917517788,
      "learning_rate": 2.304821685967665e-05,
      "loss": 0.1924,
      "step": 11450
    },
    {
      "epoch": 0.35406159003313536,
      "grad_norm": 0.10825937957450911,
      "learning_rate": 2.3035279646307625e-05,
      "loss": 0.1908,
      "step": 11460
    },
    {
      "epoch": 0.35437054430017995,
      "grad_norm": 0.11619912660972567,
      "learning_rate": 2.302233404477372e-05,
      "loss": 0.1864,
      "step": 11470
    },
    {
      "epoch": 0.3546794985672246,
      "grad_norm": 0.10569052440962179,
      "learning_rate": 2.3009380068589064e-05,
      "loss": 0.1871,
      "step": 11480
    },
    {
      "epoch": 0.3549884528342692,
      "grad_norm": 0.14648434410348218,
      "learning_rate": 2.299641773127654e-05,
      "loss": 0.1994,
      "step": 11490
    },
    {
      "epoch": 0.35529740710131386,
      "grad_norm": 0.123038590761249,
      "learning_rate": 2.2983447046367742e-05,
      "loss": 0.1985,
      "step": 11500
    },
    {
      "epoch": 0.35560636136835844,
      "grad_norm": 0.12576949179128505,
      "learning_rate": 2.297046802740299e-05,
      "loss": 0.1877,
      "step": 11510
    },
    {
      "epoch": 0.3559153156354031,
      "grad_norm": 0.11812104987432641,
      "learning_rate": 2.295748068793131e-05,
      "loss": 0.1883,
      "step": 11520
    },
    {
      "epoch": 0.3562242699024477,
      "grad_norm": 0.09611722879304281,
      "learning_rate": 2.2944485041510394e-05,
      "loss": 0.187,
      "step": 11530
    },
    {
      "epoch": 0.3565332241694923,
      "grad_norm": 0.09523547540301808,
      "learning_rate": 2.2931481101706634e-05,
      "loss": 0.1989,
      "step": 11540
    },
    {
      "epoch": 0.3568421784365369,
      "grad_norm": 0.1107326367612505,
      "learning_rate": 2.2918468882095046e-05,
      "loss": 0.1875,
      "step": 11550
    },
    {
      "epoch": 0.35715113270358156,
      "grad_norm": 0.07632770384439395,
      "learning_rate": 2.2905448396259327e-05,
      "loss": 0.1883,
      "step": 11560
    },
    {
      "epoch": 0.3574600869706262,
      "grad_norm": 0.10984237859706728,
      "learning_rate": 2.289241965779176e-05,
      "loss": 0.1848,
      "step": 11570
    },
    {
      "epoch": 0.3577690412376708,
      "grad_norm": 0.102672162754776,
      "learning_rate": 2.2879382680293286e-05,
      "loss": 0.1875,
      "step": 11580
    },
    {
      "epoch": 0.3580779955047154,
      "grad_norm": 0.16716939251737217,
      "learning_rate": 2.2866337477373413e-05,
      "loss": 0.1925,
      "step": 11590
    },
    {
      "epoch": 0.35838694977176005,
      "grad_norm": 0.10693715887367708,
      "learning_rate": 2.2853284062650263e-05,
      "loss": 0.1893,
      "step": 11600
    },
    {
      "epoch": 0.35869590403880464,
      "grad_norm": 0.0929253979596654,
      "learning_rate": 2.284022244975051e-05,
      "loss": 0.1974,
      "step": 11610
    },
    {
      "epoch": 0.3590048583058493,
      "grad_norm": 0.08777840184290545,
      "learning_rate": 2.282715265230939e-05,
      "loss": 0.1876,
      "step": 11620
    },
    {
      "epoch": 0.3593138125728939,
      "grad_norm": 0.0762413574768266,
      "learning_rate": 2.2814074683970693e-05,
      "loss": 0.1917,
      "step": 11630
    },
    {
      "epoch": 0.35962276683993855,
      "grad_norm": 0.09454809705985567,
      "learning_rate": 2.280098855838673e-05,
      "loss": 0.1835,
      "step": 11640
    },
    {
      "epoch": 0.3599317211069831,
      "grad_norm": 0.09051571813041435,
      "learning_rate": 2.2787894289218332e-05,
      "loss": 0.1891,
      "step": 11650
    },
    {
      "epoch": 0.36024067537402776,
      "grad_norm": 0.0894254880628476,
      "learning_rate": 2.2774791890134825e-05,
      "loss": 0.1891,
      "step": 11660
    },
    {
      "epoch": 0.3605496296410724,
      "grad_norm": 0.09466970727294428,
      "learning_rate": 2.2761681374814028e-05,
      "loss": 0.204,
      "step": 11670
    },
    {
      "epoch": 0.360858583908117,
      "grad_norm": 0.08531259730374659,
      "learning_rate": 2.2748562756942232e-05,
      "loss": 0.1861,
      "step": 11680
    },
    {
      "epoch": 0.3611675381751616,
      "grad_norm": 0.12284005507850895,
      "learning_rate": 2.2735436050214178e-05,
      "loss": 0.1951,
      "step": 11690
    },
    {
      "epoch": 0.36147649244220625,
      "grad_norm": 0.11919850377776255,
      "learning_rate": 2.272230126833307e-05,
      "loss": 0.1887,
      "step": 11700
    },
    {
      "epoch": 0.3617854467092509,
      "grad_norm": 0.10155349442380728,
      "learning_rate": 2.270915842501052e-05,
      "loss": 0.1881,
      "step": 11710
    },
    {
      "epoch": 0.36209440097629547,
      "grad_norm": 0.0873831466599122,
      "learning_rate": 2.2696007533966568e-05,
      "loss": 0.1873,
      "step": 11720
    },
    {
      "epoch": 0.3624033552433401,
      "grad_norm": 0.0793548762125726,
      "learning_rate": 2.2682848608929655e-05,
      "loss": 0.1851,
      "step": 11730
    },
    {
      "epoch": 0.36271230951038474,
      "grad_norm": 0.11524622716827497,
      "learning_rate": 2.2669681663636598e-05,
      "loss": 0.1862,
      "step": 11740
    },
    {
      "epoch": 0.3630212637774293,
      "grad_norm": 0.10495555651573284,
      "learning_rate": 2.265650671183261e-05,
      "loss": 0.187,
      "step": 11750
    },
    {
      "epoch": 0.36333021804447396,
      "grad_norm": 0.10260436277766406,
      "learning_rate": 2.2643323767271233e-05,
      "loss": 0.184,
      "step": 11760
    },
    {
      "epoch": 0.3636391723115186,
      "grad_norm": 0.0932148539544743,
      "learning_rate": 2.2630132843714373e-05,
      "loss": 0.1807,
      "step": 11770
    },
    {
      "epoch": 0.36394812657856324,
      "grad_norm": 0.12898993106004603,
      "learning_rate": 2.261693395493226e-05,
      "loss": 0.1897,
      "step": 11780
    },
    {
      "epoch": 0.3642570808456078,
      "grad_norm": 0.09356212327035456,
      "learning_rate": 2.2603727114703432e-05,
      "loss": 0.1942,
      "step": 11790
    },
    {
      "epoch": 0.36456603511265245,
      "grad_norm": 0.11527288431873035,
      "learning_rate": 2.259051233681474e-05,
      "loss": 0.1886,
      "step": 11800
    },
    {
      "epoch": 0.3648749893796971,
      "grad_norm": 0.10182329082139296,
      "learning_rate": 2.2577289635061314e-05,
      "loss": 0.1952,
      "step": 11810
    },
    {
      "epoch": 0.36518394364674167,
      "grad_norm": 0.10126866107020173,
      "learning_rate": 2.2564059023246552e-05,
      "loss": 0.2,
      "step": 11820
    },
    {
      "epoch": 0.3654928979137863,
      "grad_norm": 0.11290011810118633,
      "learning_rate": 2.255082051518212e-05,
      "loss": 0.1907,
      "step": 11830
    },
    {
      "epoch": 0.36580185218083094,
      "grad_norm": 0.10399887523638827,
      "learning_rate": 2.253757412468792e-05,
      "loss": 0.1846,
      "step": 11840
    },
    {
      "epoch": 0.3661108064478756,
      "grad_norm": 0.08962367793400801,
      "learning_rate": 2.2524319865592085e-05,
      "loss": 0.1896,
      "step": 11850
    },
    {
      "epoch": 0.36641976071492016,
      "grad_norm": 0.0971662300096469,
      "learning_rate": 2.2511057751730957e-05,
      "loss": 0.1848,
      "step": 11860
    },
    {
      "epoch": 0.3667287149819648,
      "grad_norm": 0.08926981262442568,
      "learning_rate": 2.2497787796949086e-05,
      "loss": 0.1891,
      "step": 11870
    },
    {
      "epoch": 0.36703766924900944,
      "grad_norm": 0.0997820021140413,
      "learning_rate": 2.24845100150992e-05,
      "loss": 0.1883,
      "step": 11880
    },
    {
      "epoch": 0.367346623516054,
      "grad_norm": 0.11019032194902358,
      "learning_rate": 2.247122442004221e-05,
      "loss": 0.214,
      "step": 11890
    },
    {
      "epoch": 0.36765557778309865,
      "grad_norm": 0.08843270186981546,
      "learning_rate": 2.2457931025647164e-05,
      "loss": 0.1948,
      "step": 11900
    },
    {
      "epoch": 0.3679645320501433,
      "grad_norm": 0.08081469587518253,
      "learning_rate": 2.2444629845791263e-05,
      "loss": 0.1864,
      "step": 11910
    },
    {
      "epoch": 0.3682734863171879,
      "grad_norm": 0.08643601308806263,
      "learning_rate": 2.2431320894359837e-05,
      "loss": 0.189,
      "step": 11920
    },
    {
      "epoch": 0.3685824405842325,
      "grad_norm": 0.1043042633827644,
      "learning_rate": 2.2418004185246325e-05,
      "loss": 0.1863,
      "step": 11930
    },
    {
      "epoch": 0.36889139485127714,
      "grad_norm": 0.10909555625908413,
      "learning_rate": 2.240467973235227e-05,
      "loss": 0.1853,
      "step": 11940
    },
    {
      "epoch": 0.3692003491183218,
      "grad_norm": 0.10081295667943212,
      "learning_rate": 2.2391347549587288e-05,
      "loss": 0.1835,
      "step": 11950
    },
    {
      "epoch": 0.36950930338536636,
      "grad_norm": 0.10853639856568237,
      "learning_rate": 2.237800765086908e-05,
      "loss": 0.1869,
      "step": 11960
    },
    {
      "epoch": 0.369818257652411,
      "grad_norm": 0.10906644104912402,
      "learning_rate": 2.236466005012338e-05,
      "loss": 0.1841,
      "step": 11970
    },
    {
      "epoch": 0.37012721191945563,
      "grad_norm": 0.09492831998510312,
      "learning_rate": 2.2351304761283985e-05,
      "loss": 0.1873,
      "step": 11980
    },
    {
      "epoch": 0.37043616618650027,
      "grad_norm": 0.08404413315269131,
      "learning_rate": 2.233794179829271e-05,
      "loss": 0.1834,
      "step": 11990
    },
    {
      "epoch": 0.37074512045354485,
      "grad_norm": 0.11624722382524719,
      "learning_rate": 2.232457117509937e-05,
      "loss": 0.1999,
      "step": 12000
    },
    {
      "epoch": 0.3710540747205895,
      "grad_norm": 0.10604518786522354,
      "learning_rate": 2.2311192905661795e-05,
      "loss": 0.1876,
      "step": 12010
    },
    {
      "epoch": 0.3713630289876341,
      "grad_norm": 0.12277243084657756,
      "learning_rate": 2.229780700394578e-05,
      "loss": 0.1858,
      "step": 12020
    },
    {
      "epoch": 0.3716719832546787,
      "grad_norm": 0.09428607361303694,
      "learning_rate": 2.2284413483925097e-05,
      "loss": 0.1858,
      "step": 12030
    },
    {
      "epoch": 0.37198093752172334,
      "grad_norm": 0.10303962243287938,
      "learning_rate": 2.2271012359581468e-05,
      "loss": 0.1854,
      "step": 12040
    },
    {
      "epoch": 0.372289891788768,
      "grad_norm": 0.08662878274471632,
      "learning_rate": 2.2257603644904557e-05,
      "loss": 0.1833,
      "step": 12050
    },
    {
      "epoch": 0.3725988460558126,
      "grad_norm": 0.08877536584269159,
      "learning_rate": 2.224418735389195e-05,
      "loss": 0.1961,
      "step": 12060
    },
    {
      "epoch": 0.3729078003228572,
      "grad_norm": 0.09638428066654024,
      "learning_rate": 2.2230763500549136e-05,
      "loss": 0.19,
      "step": 12070
    },
    {
      "epoch": 0.37321675458990183,
      "grad_norm": 0.11273893075014302,
      "learning_rate": 2.2217332098889507e-05,
      "loss": 0.2024,
      "step": 12080
    },
    {
      "epoch": 0.37352570885694647,
      "grad_norm": 0.08659890915115652,
      "learning_rate": 2.2203893162934333e-05,
      "loss": 0.1927,
      "step": 12090
    },
    {
      "epoch": 0.37383466312399105,
      "grad_norm": 0.09580197460355058,
      "learning_rate": 2.2190446706712743e-05,
      "loss": 0.1927,
      "step": 12100
    },
    {
      "epoch": 0.3741436173910357,
      "grad_norm": 0.0861993079505064,
      "learning_rate": 2.2176992744261728e-05,
      "loss": 0.1848,
      "step": 12110
    },
    {
      "epoch": 0.3744525716580803,
      "grad_norm": 0.09478748886063199,
      "learning_rate": 2.21635312896261e-05,
      "loss": 0.1868,
      "step": 12120
    },
    {
      "epoch": 0.37476152592512496,
      "grad_norm": 0.10030432908753674,
      "learning_rate": 2.2150062356858507e-05,
      "loss": 0.1878,
      "step": 12130
    },
    {
      "epoch": 0.37507048019216954,
      "grad_norm": 0.1017851961915696,
      "learning_rate": 2.2136585960019387e-05,
      "loss": 0.1993,
      "step": 12140
    },
    {
      "epoch": 0.3753794344592142,
      "grad_norm": 0.10067172635604223,
      "learning_rate": 2.212310211317699e-05,
      "loss": 0.1934,
      "step": 12150
    },
    {
      "epoch": 0.3756883887262588,
      "grad_norm": 0.08778870365569208,
      "learning_rate": 2.2109610830407327e-05,
      "loss": 0.1862,
      "step": 12160
    },
    {
      "epoch": 0.3759973429933034,
      "grad_norm": 0.09508651264190547,
      "learning_rate": 2.2096112125794174e-05,
      "loss": 0.1972,
      "step": 12170
    },
    {
      "epoch": 0.37630629726034803,
      "grad_norm": 0.08730341557296374,
      "learning_rate": 2.208260601342907e-05,
      "loss": 0.1919,
      "step": 12180
    },
    {
      "epoch": 0.37661525152739267,
      "grad_norm": 0.10788799527763362,
      "learning_rate": 2.2069092507411265e-05,
      "loss": 0.195,
      "step": 12190
    },
    {
      "epoch": 0.3769242057944373,
      "grad_norm": 0.11811812837511006,
      "learning_rate": 2.2055571621847737e-05,
      "loss": 0.1864,
      "step": 12200
    },
    {
      "epoch": 0.3772331600614819,
      "grad_norm": 0.0878050403220875,
      "learning_rate": 2.2042043370853174e-05,
      "loss": 0.1855,
      "step": 12210
    },
    {
      "epoch": 0.3775421143285265,
      "grad_norm": 0.08853991233275472,
      "learning_rate": 2.2028507768549944e-05,
      "loss": 0.1885,
      "step": 12220
    },
    {
      "epoch": 0.37785106859557116,
      "grad_norm": 0.08452375786593688,
      "learning_rate": 2.2014964829068087e-05,
      "loss": 0.1925,
      "step": 12230
    },
    {
      "epoch": 0.37816002286261574,
      "grad_norm": 0.09903481689275309,
      "learning_rate": 2.2001414566545323e-05,
      "loss": 0.1859,
      "step": 12240
    },
    {
      "epoch": 0.3784689771296604,
      "grad_norm": 0.0902044230703504,
      "learning_rate": 2.198785699512698e-05,
      "loss": 0.1898,
      "step": 12250
    },
    {
      "epoch": 0.378777931396705,
      "grad_norm": 0.09789504524374835,
      "learning_rate": 2.1974292128966054e-05,
      "loss": 0.193,
      "step": 12260
    },
    {
      "epoch": 0.37908688566374965,
      "grad_norm": 0.0761749796310836,
      "learning_rate": 2.196071998222313e-05,
      "loss": 0.186,
      "step": 12270
    },
    {
      "epoch": 0.37939583993079423,
      "grad_norm": 0.09601634226279154,
      "learning_rate": 2.1947140569066405e-05,
      "loss": 0.1813,
      "step": 12280
    },
    {
      "epoch": 0.37970479419783887,
      "grad_norm": 0.10033439403921206,
      "learning_rate": 2.193355390367166e-05,
      "loss": 0.1877,
      "step": 12290
    },
    {
      "epoch": 0.3800137484648835,
      "grad_norm": 0.09969839405797135,
      "learning_rate": 2.1919960000222245e-05,
      "loss": 0.1894,
      "step": 12300
    },
    {
      "epoch": 0.3803227027319281,
      "grad_norm": 0.1060276229550653,
      "learning_rate": 2.1906358872909068e-05,
      "loss": 0.1877,
      "step": 12310
    },
    {
      "epoch": 0.3806316569989727,
      "grad_norm": 0.10895069946809241,
      "learning_rate": 2.1892750535930575e-05,
      "loss": 0.1882,
      "step": 12320
    },
    {
      "epoch": 0.38094061126601736,
      "grad_norm": 0.09820742351990522,
      "learning_rate": 2.187913500349274e-05,
      "loss": 0.1963,
      "step": 12330
    },
    {
      "epoch": 0.381249565533062,
      "grad_norm": 0.09472323522818168,
      "learning_rate": 2.1865512289809052e-05,
      "loss": 0.1857,
      "step": 12340
    },
    {
      "epoch": 0.3815585198001066,
      "grad_norm": 0.09988799948578274,
      "learning_rate": 2.185188240910049e-05,
      "loss": 0.1868,
      "step": 12350
    },
    {
      "epoch": 0.3818674740671512,
      "grad_norm": 0.12750044657653234,
      "learning_rate": 2.1838245375595517e-05,
      "loss": 0.1921,
      "step": 12360
    },
    {
      "epoch": 0.38217642833419585,
      "grad_norm": 0.09912066288832452,
      "learning_rate": 2.182460120353006e-05,
      "loss": 0.1874,
      "step": 12370
    },
    {
      "epoch": 0.38248538260124043,
      "grad_norm": 0.08189537759998616,
      "learning_rate": 2.1810949907147518e-05,
      "loss": 0.1783,
      "step": 12380
    },
    {
      "epoch": 0.38279433686828507,
      "grad_norm": 0.09036618563864209,
      "learning_rate": 2.179729150069869e-05,
      "loss": 0.1866,
      "step": 12390
    },
    {
      "epoch": 0.3831032911353297,
      "grad_norm": 0.09923347261178064,
      "learning_rate": 2.1783625998441832e-05,
      "loss": 0.1861,
      "step": 12400
    },
    {
      "epoch": 0.38341224540237434,
      "grad_norm": 0.0970615990744683,
      "learning_rate": 2.1769953414642595e-05,
      "loss": 0.187,
      "step": 12410
    },
    {
      "epoch": 0.3837211996694189,
      "grad_norm": 0.0971170760607018,
      "learning_rate": 2.1756273763574015e-05,
      "loss": 0.1914,
      "step": 12420
    },
    {
      "epoch": 0.38403015393646356,
      "grad_norm": 0.14313686063009398,
      "learning_rate": 2.1742587059516515e-05,
      "loss": 0.195,
      "step": 12430
    },
    {
      "epoch": 0.3843391082035082,
      "grad_norm": 0.10942143582420122,
      "learning_rate": 2.1728893316757873e-05,
      "loss": 0.1865,
      "step": 12440
    },
    {
      "epoch": 0.3846480624705528,
      "grad_norm": 0.09674273343079418,
      "learning_rate": 2.1715192549593225e-05,
      "loss": 0.1846,
      "step": 12450
    },
    {
      "epoch": 0.3849570167375974,
      "grad_norm": 0.11089546161791718,
      "learning_rate": 2.1701484772325033e-05,
      "loss": 0.1863,
      "step": 12460
    },
    {
      "epoch": 0.38526597100464205,
      "grad_norm": 0.08988728264341979,
      "learning_rate": 2.1687769999263082e-05,
      "loss": 0.1838,
      "step": 12470
    },
    {
      "epoch": 0.3855749252716867,
      "grad_norm": 0.10942846423576381,
      "learning_rate": 2.1674048244724448e-05,
      "loss": 0.1918,
      "step": 12480
    },
    {
      "epoch": 0.38588387953873127,
      "grad_norm": 0.0960690443747848,
      "learning_rate": 2.1660319523033513e-05,
      "loss": 0.1858,
      "step": 12490
    },
    {
      "epoch": 0.3861928338057759,
      "grad_norm": 0.088145278690529,
      "learning_rate": 2.164658384852191e-05,
      "loss": 0.1898,
      "step": 12500
    },
    {
      "epoch": 0.38650178807282054,
      "grad_norm": 0.09096558863983556,
      "learning_rate": 2.163284123552856e-05,
      "loss": 0.1896,
      "step": 12510
    },
    {
      "epoch": 0.3868107423398651,
      "grad_norm": 0.1047677160787241,
      "learning_rate": 2.1619091698399597e-05,
      "loss": 0.1884,
      "step": 12520
    },
    {
      "epoch": 0.38711969660690976,
      "grad_norm": 0.10141960460779591,
      "learning_rate": 2.1605335251488402e-05,
      "loss": 0.1857,
      "step": 12530
    },
    {
      "epoch": 0.3874286508739544,
      "grad_norm": 0.07862972805395212,
      "learning_rate": 2.1591571909155562e-05,
      "loss": 0.1885,
      "step": 12540
    },
    {
      "epoch": 0.38773760514099903,
      "grad_norm": 0.08583249061967742,
      "learning_rate": 2.1577801685768863e-05,
      "loss": 0.1848,
      "step": 12550
    },
    {
      "epoch": 0.3880465594080436,
      "grad_norm": 0.09202201423974528,
      "learning_rate": 2.1564024595703277e-05,
      "loss": 0.19,
      "step": 12560
    },
    {
      "epoch": 0.38835551367508825,
      "grad_norm": 0.1406654518530953,
      "learning_rate": 2.1550240653340943e-05,
      "loss": 0.1903,
      "step": 12570
    },
    {
      "epoch": 0.3886644679421329,
      "grad_norm": 0.09258293574371818,
      "learning_rate": 2.153644987307115e-05,
      "loss": 0.1884,
      "step": 12580
    },
    {
      "epoch": 0.38897342220917747,
      "grad_norm": 0.11507710882309509,
      "learning_rate": 2.152265226929033e-05,
      "loss": 0.194,
      "step": 12590
    },
    {
      "epoch": 0.3892823764762221,
      "grad_norm": 0.07649515273569724,
      "learning_rate": 2.150884785640203e-05,
      "loss": 0.1961,
      "step": 12600
    },
    {
      "epoch": 0.38959133074326674,
      "grad_norm": 0.08426992115426334,
      "learning_rate": 2.1495036648816916e-05,
      "loss": 0.1829,
      "step": 12610
    },
    {
      "epoch": 0.3899002850103114,
      "grad_norm": 0.09726246348335965,
      "learning_rate": 2.1481218660952744e-05,
      "loss": 0.1933,
      "step": 12620
    },
    {
      "epoch": 0.39020923927735596,
      "grad_norm": 0.5326000685879965,
      "learning_rate": 2.146739390723434e-05,
      "loss": 0.189,
      "step": 12630
    },
    {
      "epoch": 0.3905181935444006,
      "grad_norm": 0.08204811848597375,
      "learning_rate": 2.145356240209361e-05,
      "loss": 0.1931,
      "step": 12640
    },
    {
      "epoch": 0.39082714781144523,
      "grad_norm": 0.07572955421670016,
      "learning_rate": 2.1439724159969487e-05,
      "loss": 0.1913,
      "step": 12650
    },
    {
      "epoch": 0.3911361020784898,
      "grad_norm": 0.09723300209292356,
      "learning_rate": 2.142587919530795e-05,
      "loss": 0.1848,
      "step": 12660
    },
    {
      "epoch": 0.39144505634553445,
      "grad_norm": 0.087439750920017,
      "learning_rate": 2.1412027522561993e-05,
      "loss": 0.1869,
      "step": 12670
    },
    {
      "epoch": 0.3917540106125791,
      "grad_norm": 0.09070355508572686,
      "learning_rate": 2.1398169156191608e-05,
      "loss": 0.1994,
      "step": 12680
    },
    {
      "epoch": 0.3920629648796237,
      "grad_norm": 0.08435644984154654,
      "learning_rate": 2.1384304110663793e-05,
      "loss": 0.1885,
      "step": 12690
    },
    {
      "epoch": 0.3923719191466683,
      "grad_norm": 0.1021550545994983,
      "learning_rate": 2.137043240045249e-05,
      "loss": 0.1891,
      "step": 12700
    },
    {
      "epoch": 0.39268087341371294,
      "grad_norm": 0.08711715789476,
      "learning_rate": 2.1356554040038613e-05,
      "loss": 0.1836,
      "step": 12710
    },
    {
      "epoch": 0.3929898276807576,
      "grad_norm": 0.0981042539086943,
      "learning_rate": 2.134266904391003e-05,
      "loss": 0.1853,
      "step": 12720
    },
    {
      "epoch": 0.39329878194780216,
      "grad_norm": 0.11903424237869814,
      "learning_rate": 2.1328777426561508e-05,
      "loss": 0.1924,
      "step": 12730
    },
    {
      "epoch": 0.3936077362148468,
      "grad_norm": 0.09828569735706146,
      "learning_rate": 2.131487920249476e-05,
      "loss": 0.1878,
      "step": 12740
    },
    {
      "epoch": 0.39391669048189143,
      "grad_norm": 0.10736634595032453,
      "learning_rate": 2.1300974386218364e-05,
      "loss": 0.1865,
      "step": 12750
    },
    {
      "epoch": 0.39422564474893607,
      "grad_norm": 0.1078794855524927,
      "learning_rate": 2.12870629922478e-05,
      "loss": 0.1907,
      "step": 12760
    },
    {
      "epoch": 0.39453459901598065,
      "grad_norm": 0.09504763593375629,
      "learning_rate": 2.1273145035105407e-05,
      "loss": 0.1859,
      "step": 12770
    },
    {
      "epoch": 0.3948435532830253,
      "grad_norm": 0.10680741045769311,
      "learning_rate": 2.125922052932038e-05,
      "loss": 0.1941,
      "step": 12780
    },
    {
      "epoch": 0.3951525075500699,
      "grad_norm": 0.10606809238189722,
      "learning_rate": 2.1245289489428747e-05,
      "loss": 0.1853,
      "step": 12790
    },
    {
      "epoch": 0.3954614618171145,
      "grad_norm": 0.08659377105273015,
      "learning_rate": 2.1231351929973354e-05,
      "loss": 0.1922,
      "step": 12800
    },
    {
      "epoch": 0.39577041608415914,
      "grad_norm": 0.11632369655608324,
      "learning_rate": 2.1217407865503853e-05,
      "loss": 0.1865,
      "step": 12810
    },
    {
      "epoch": 0.3960793703512038,
      "grad_norm": 0.11702368925564903,
      "learning_rate": 2.1203457310576692e-05,
      "loss": 0.1836,
      "step": 12820
    },
    {
      "epoch": 0.3963883246182484,
      "grad_norm": 0.09962331373161037,
      "learning_rate": 2.11895002797551e-05,
      "loss": 0.1895,
      "step": 12830
    },
    {
      "epoch": 0.396697278885293,
      "grad_norm": 0.07817913257160077,
      "learning_rate": 2.117553678760905e-05,
      "loss": 0.1849,
      "step": 12840
    },
    {
      "epoch": 0.39700623315233763,
      "grad_norm": 0.07912360578462507,
      "learning_rate": 2.1161566848715264e-05,
      "loss": 0.1875,
      "step": 12850
    },
    {
      "epoch": 0.39731518741938227,
      "grad_norm": 0.09695989534077361,
      "learning_rate": 2.1147590477657214e-05,
      "loss": 0.1921,
      "step": 12860
    },
    {
      "epoch": 0.39762414168642685,
      "grad_norm": 0.10486524951927254,
      "learning_rate": 2.113360768902506e-05,
      "loss": 0.1955,
      "step": 12870
    },
    {
      "epoch": 0.3979330959534715,
      "grad_norm": 0.09637892490561646,
      "learning_rate": 2.1119618497415673e-05,
      "loss": 0.182,
      "step": 12880
    },
    {
      "epoch": 0.3982420502205161,
      "grad_norm": 0.08949872499108744,
      "learning_rate": 2.110562291743261e-05,
      "loss": 0.1828,
      "step": 12890
    },
    {
      "epoch": 0.39855100448756076,
      "grad_norm": 0.09086233140354587,
      "learning_rate": 2.1091620963686092e-05,
      "loss": 0.1908,
      "step": 12900
    },
    {
      "epoch": 0.39885995875460534,
      "grad_norm": 0.10980143053783954,
      "learning_rate": 2.1077612650793004e-05,
      "loss": 0.188,
      "step": 12910
    },
    {
      "epoch": 0.39916891302165,
      "grad_norm": 0.13226727670311497,
      "learning_rate": 2.1063597993376857e-05,
      "loss": 0.1938,
      "step": 12920
    },
    {
      "epoch": 0.3994778672886946,
      "grad_norm": 0.08670786506261809,
      "learning_rate": 2.104957700606778e-05,
      "loss": 0.1866,
      "step": 12930
    },
    {
      "epoch": 0.3997868215557392,
      "grad_norm": 0.10261197420659128,
      "learning_rate": 2.1035549703502533e-05,
      "loss": 0.1918,
      "step": 12940
    },
    {
      "epoch": 0.40009577582278383,
      "grad_norm": 0.09071894437343855,
      "learning_rate": 2.1021516100324455e-05,
      "loss": 0.192,
      "step": 12950
    },
    {
      "epoch": 0.40040473008982846,
      "grad_norm": 0.08839448769964206,
      "learning_rate": 2.1007476211183453e-05,
      "loss": 0.1894,
      "step": 12960
    },
    {
      "epoch": 0.4007136843568731,
      "grad_norm": 0.22092952503263738,
      "learning_rate": 2.0993430050736017e-05,
      "loss": 0.1942,
      "step": 12970
    },
    {
      "epoch": 0.4010226386239177,
      "grad_norm": 0.14461578183039348,
      "learning_rate": 2.0979377633645163e-05,
      "loss": 0.1883,
      "step": 12980
    },
    {
      "epoch": 0.4013315928909623,
      "grad_norm": 0.09938517723185097,
      "learning_rate": 2.0965318974580452e-05,
      "loss": 0.1838,
      "step": 12990
    },
    {
      "epoch": 0.40164054715800696,
      "grad_norm": 0.1013939273930599,
      "learning_rate": 2.0951254088217958e-05,
      "loss": 0.1901,
      "step": 13000
    },
    {
      "epoch": 0.40194950142505154,
      "grad_norm": 0.08510276310095609,
      "learning_rate": 2.093718298924025e-05,
      "loss": 0.1872,
      "step": 13010
    },
    {
      "epoch": 0.4022584556920962,
      "grad_norm": 0.09845304927890826,
      "learning_rate": 2.0923105692336396e-05,
      "loss": 0.1892,
      "step": 13020
    },
    {
      "epoch": 0.4025674099591408,
      "grad_norm": 0.08600848269514792,
      "learning_rate": 2.090902221220192e-05,
      "loss": 0.1861,
      "step": 13030
    },
    {
      "epoch": 0.40287636422618545,
      "grad_norm": 0.08941561354144911,
      "learning_rate": 2.0894932563538802e-05,
      "loss": 0.1853,
      "step": 13040
    },
    {
      "epoch": 0.40318531849323,
      "grad_norm": 0.09258580376709308,
      "learning_rate": 2.088083676105547e-05,
      "loss": 0.1924,
      "step": 13050
    },
    {
      "epoch": 0.40349427276027466,
      "grad_norm": 0.09935786665947385,
      "learning_rate": 2.0866734819466776e-05,
      "loss": 0.1883,
      "step": 13060
    },
    {
      "epoch": 0.4038032270273193,
      "grad_norm": 0.10612466525117271,
      "learning_rate": 2.0852626753493966e-05,
      "loss": 0.1869,
      "step": 13070
    },
    {
      "epoch": 0.4041121812943639,
      "grad_norm": 0.08229392848397417,
      "learning_rate": 2.0838512577864698e-05,
      "loss": 0.1882,
      "step": 13080
    },
    {
      "epoch": 0.4044211355614085,
      "grad_norm": 0.08769203289903779,
      "learning_rate": 2.0824392307312998e-05,
      "loss": 0.1837,
      "step": 13090
    },
    {
      "epoch": 0.40473008982845315,
      "grad_norm": 0.08902737795021919,
      "learning_rate": 2.0810265956579255e-05,
      "loss": 0.1851,
      "step": 13100
    },
    {
      "epoch": 0.4050390440954978,
      "grad_norm": 0.10275579899241351,
      "learning_rate": 2.0796133540410204e-05,
      "loss": 0.1852,
      "step": 13110
    },
    {
      "epoch": 0.40534799836254237,
      "grad_norm": 0.10587801739043626,
      "learning_rate": 2.0781995073558918e-05,
      "loss": 0.1874,
      "step": 13120
    },
    {
      "epoch": 0.405656952629587,
      "grad_norm": 0.1246824199375735,
      "learning_rate": 2.0767850570784784e-05,
      "loss": 0.1931,
      "step": 13130
    },
    {
      "epoch": 0.40596590689663165,
      "grad_norm": 0.09930681277470046,
      "learning_rate": 2.0753700046853483e-05,
      "loss": 0.1913,
      "step": 13140
    },
    {
      "epoch": 0.4062748611636762,
      "grad_norm": 0.08695439519398754,
      "learning_rate": 2.0739543516536993e-05,
      "loss": 0.1883,
      "step": 13150
    },
    {
      "epoch": 0.40658381543072086,
      "grad_norm": 0.09422358361375846,
      "learning_rate": 2.0725380994613546e-05,
      "loss": 0.1869,
      "step": 13160
    },
    {
      "epoch": 0.4068927696977655,
      "grad_norm": 0.0968254311671267,
      "learning_rate": 2.071121249586765e-05,
      "loss": 0.1929,
      "step": 13170
    },
    {
      "epoch": 0.40720172396481014,
      "grad_norm": 0.10575057340327401,
      "learning_rate": 2.0697038035090043e-05,
      "loss": 0.2018,
      "step": 13180
    },
    {
      "epoch": 0.4075106782318547,
      "grad_norm": 0.09948013686687163,
      "learning_rate": 2.068285762707768e-05,
      "loss": 0.1863,
      "step": 13190
    },
    {
      "epoch": 0.40781963249889935,
      "grad_norm": 0.09195181849704964,
      "learning_rate": 2.0668671286633725e-05,
      "loss": 0.1841,
      "step": 13200
    },
    {
      "epoch": 0.408128586765944,
      "grad_norm": 0.09366078386563499,
      "learning_rate": 2.065447902856755e-05,
      "loss": 0.1937,
      "step": 13210
    },
    {
      "epoch": 0.40843754103298857,
      "grad_norm": 0.08190487731690498,
      "learning_rate": 2.0640280867694688e-05,
      "loss": 0.1865,
      "step": 13220
    },
    {
      "epoch": 0.4087464953000332,
      "grad_norm": 0.09882543862828375,
      "learning_rate": 2.062607681883684e-05,
      "loss": 0.1911,
      "step": 13230
    },
    {
      "epoch": 0.40905544956707784,
      "grad_norm": 0.08367448640349641,
      "learning_rate": 2.061186689682186e-05,
      "loss": 0.1948,
      "step": 13240
    },
    {
      "epoch": 0.4093644038341225,
      "grad_norm": 0.11294440587485939,
      "learning_rate": 2.0597651116483717e-05,
      "loss": 0.1942,
      "step": 13250
    },
    {
      "epoch": 0.40967335810116706,
      "grad_norm": 0.08153232269152608,
      "learning_rate": 2.0583429492662514e-05,
      "loss": 0.1906,
      "step": 13260
    },
    {
      "epoch": 0.4099823123682117,
      "grad_norm": 0.09853181618840731,
      "learning_rate": 2.056920204020444e-05,
      "loss": 0.1927,
      "step": 13270
    },
    {
      "epoch": 0.41029126663525634,
      "grad_norm": 0.0790572919252489,
      "learning_rate": 2.0554968773961778e-05,
      "loss": 0.1914,
      "step": 13280
    },
    {
      "epoch": 0.4106002209023009,
      "grad_norm": 0.09194298049947046,
      "learning_rate": 2.0540729708792875e-05,
      "loss": 0.1946,
      "step": 13290
    },
    {
      "epoch": 0.41090917516934555,
      "grad_norm": 0.09819303095831762,
      "learning_rate": 2.052648485956213e-05,
      "loss": 0.1871,
      "step": 13300
    },
    {
      "epoch": 0.4112181294363902,
      "grad_norm": 0.09155683839623317,
      "learning_rate": 2.051223424113999e-05,
      "loss": 0.1889,
      "step": 13310
    },
    {
      "epoch": 0.4115270837034348,
      "grad_norm": 0.08773435869517471,
      "learning_rate": 2.0497977868402914e-05,
      "loss": 0.1853,
      "step": 13320
    },
    {
      "epoch": 0.4118360379704794,
      "grad_norm": 0.08760743257171492,
      "learning_rate": 2.0483715756233373e-05,
      "loss": 0.1883,
      "step": 13330
    },
    {
      "epoch": 0.41214499223752404,
      "grad_norm": 0.10691711717069,
      "learning_rate": 2.046944791951982e-05,
      "loss": 0.186,
      "step": 13340
    },
    {
      "epoch": 0.4124539465045687,
      "grad_norm": 0.08634815961598084,
      "learning_rate": 2.04551743731567e-05,
      "loss": 0.1947,
      "step": 13350
    },
    {
      "epoch": 0.41276290077161326,
      "grad_norm": 0.1468646772202215,
      "learning_rate": 2.0440895132044412e-05,
      "loss": 0.1953,
      "step": 13360
    },
    {
      "epoch": 0.4130718550386579,
      "grad_norm": 0.0829618595581519,
      "learning_rate": 2.0426610211089293e-05,
      "loss": 0.1873,
      "step": 13370
    },
    {
      "epoch": 0.41338080930570253,
      "grad_norm": 0.09593809763101668,
      "learning_rate": 2.0412319625203617e-05,
      "loss": 0.1868,
      "step": 13380
    },
    {
      "epoch": 0.41368976357274717,
      "grad_norm": 0.10604014121023575,
      "learning_rate": 2.039802338930557e-05,
      "loss": 0.1937,
      "step": 13390
    },
    {
      "epoch": 0.41399871783979175,
      "grad_norm": 0.0835303284590433,
      "learning_rate": 2.038372151831923e-05,
      "loss": 0.1888,
      "step": 13400
    },
    {
      "epoch": 0.4143076721068364,
      "grad_norm": 0.09142589495290121,
      "learning_rate": 2.0369414027174572e-05,
      "loss": 0.1882,
      "step": 13410
    },
    {
      "epoch": 0.414616626373881,
      "grad_norm": 0.11497398429958629,
      "learning_rate": 2.035510093080742e-05,
      "loss": 0.1886,
      "step": 13420
    },
    {
      "epoch": 0.4149255806409256,
      "grad_norm": 0.09442063761529558,
      "learning_rate": 2.034078224415947e-05,
      "loss": 0.1801,
      "step": 13430
    },
    {
      "epoch": 0.41523453490797024,
      "grad_norm": 0.13278925914404902,
      "learning_rate": 2.0326457982178228e-05,
      "loss": 0.2021,
      "step": 13440
    },
    {
      "epoch": 0.4155434891750149,
      "grad_norm": 0.09970126251068312,
      "learning_rate": 2.0312128159817043e-05,
      "loss": 0.1863,
      "step": 13450
    },
    {
      "epoch": 0.4158524434420595,
      "grad_norm": 0.10704701412369369,
      "learning_rate": 2.029779279203506e-05,
      "loss": 0.1844,
      "step": 13460
    },
    {
      "epoch": 0.4161613977091041,
      "grad_norm": 0.09866383422506134,
      "learning_rate": 2.0283451893797212e-05,
      "loss": 0.1952,
      "step": 13470
    },
    {
      "epoch": 0.41647035197614873,
      "grad_norm": 0.08402645934569061,
      "learning_rate": 2.0269105480074204e-05,
      "loss": 0.1877,
      "step": 13480
    },
    {
      "epoch": 0.41677930624319337,
      "grad_norm": 0.08989225505028141,
      "learning_rate": 2.02547535658425e-05,
      "loss": 0.1884,
      "step": 13490
    },
    {
      "epoch": 0.41708826051023795,
      "grad_norm": 0.10578288777862112,
      "learning_rate": 2.0240396166084315e-05,
      "loss": 0.1877,
      "step": 13500
    },
    {
      "epoch": 0.4173972147772826,
      "grad_norm": 0.07430768141184434,
      "learning_rate": 2.022603329578757e-05,
      "loss": 0.1876,
      "step": 13510
    },
    {
      "epoch": 0.4177061690443272,
      "grad_norm": 0.08544179582130298,
      "learning_rate": 2.021166496994592e-05,
      "loss": 0.1855,
      "step": 13520
    },
    {
      "epoch": 0.41801512331137186,
      "grad_norm": 0.08505474087708828,
      "learning_rate": 2.0197291203558696e-05,
      "loss": 0.1916,
      "step": 13530
    },
    {
      "epoch": 0.41832407757841644,
      "grad_norm": 0.07972136122383491,
      "learning_rate": 2.0182912011630923e-05,
      "loss": 0.1912,
      "step": 13540
    },
    {
      "epoch": 0.4186330318454611,
      "grad_norm": 0.08981097110612854,
      "learning_rate": 2.0168527409173284e-05,
      "loss": 0.1827,
      "step": 13550
    },
    {
      "epoch": 0.4189419861125057,
      "grad_norm": 0.08533861833995386,
      "learning_rate": 2.015413741120211e-05,
      "loss": 0.185,
      "step": 13560
    },
    {
      "epoch": 0.4192509403795503,
      "grad_norm": 0.09544531938911857,
      "learning_rate": 2.0139742032739356e-05,
      "loss": 0.1898,
      "step": 13570
    },
    {
      "epoch": 0.41955989464659493,
      "grad_norm": 0.07626397128140812,
      "learning_rate": 2.012534128881261e-05,
      "loss": 0.1903,
      "step": 13580
    },
    {
      "epoch": 0.41986884891363957,
      "grad_norm": 0.10049543968215914,
      "learning_rate": 2.0110935194455054e-05,
      "loss": 0.1839,
      "step": 13590
    },
    {
      "epoch": 0.4201778031806842,
      "grad_norm": 0.09195127165080279,
      "learning_rate": 2.0096523764705447e-05,
      "loss": 0.1845,
      "step": 13600
    },
    {
      "epoch": 0.4204867574477288,
      "grad_norm": 0.08843111699709341,
      "learning_rate": 2.0082107014608128e-05,
      "loss": 0.1839,
      "step": 13610
    },
    {
      "epoch": 0.4207957117147734,
      "grad_norm": 0.09131074992090338,
      "learning_rate": 2.0067684959212995e-05,
      "loss": 0.1866,
      "step": 13620
    },
    {
      "epoch": 0.42110466598181806,
      "grad_norm": 0.082216641269758,
      "learning_rate": 2.0053257613575467e-05,
      "loss": 0.183,
      "step": 13630
    },
    {
      "epoch": 0.42141362024886264,
      "grad_norm": 0.08831693579553958,
      "learning_rate": 2.0038824992756496e-05,
      "loss": 0.1841,
      "step": 13640
    },
    {
      "epoch": 0.4217225745159073,
      "grad_norm": 0.09112703304305547,
      "learning_rate": 2.0024387111822547e-05,
      "loss": 0.182,
      "step": 13650
    },
    {
      "epoch": 0.4220315287829519,
      "grad_norm": 0.094754745798283,
      "learning_rate": 2.0009943985845564e-05,
      "loss": 0.1869,
      "step": 13660
    },
    {
      "epoch": 0.42234048304999655,
      "grad_norm": 0.09299186425263704,
      "learning_rate": 1.999549562990297e-05,
      "loss": 0.1947,
      "step": 13670
    },
    {
      "epoch": 0.42264943731704113,
      "grad_norm": 0.09819702932619447,
      "learning_rate": 1.9981042059077657e-05,
      "loss": 0.1854,
      "step": 13680
    },
    {
      "epoch": 0.42295839158408577,
      "grad_norm": 0.09652302540077827,
      "learning_rate": 1.9966583288457944e-05,
      "loss": 0.1836,
      "step": 13690
    },
    {
      "epoch": 0.4232673458511304,
      "grad_norm": 0.09629148997305158,
      "learning_rate": 1.9952119333137594e-05,
      "loss": 0.1881,
      "step": 13700
    },
    {
      "epoch": 0.423576300118175,
      "grad_norm": 0.08819754976061543,
      "learning_rate": 1.9937650208215766e-05,
      "loss": 0.1902,
      "step": 13710
    },
    {
      "epoch": 0.4238852543852196,
      "grad_norm": 0.09763247427356099,
      "learning_rate": 1.9923175928797035e-05,
      "loss": 0.185,
      "step": 13720
    },
    {
      "epoch": 0.42419420865226426,
      "grad_norm": 0.09401799831882597,
      "learning_rate": 1.990869650999134e-05,
      "loss": 0.1854,
      "step": 13730
    },
    {
      "epoch": 0.4245031629193089,
      "grad_norm": 0.08850753614982744,
      "learning_rate": 1.9894211966913998e-05,
      "loss": 0.1911,
      "step": 13740
    },
    {
      "epoch": 0.4248121171863535,
      "grad_norm": 0.0963336941553544,
      "learning_rate": 1.987972231468567e-05,
      "loss": 0.1847,
      "step": 13750
    },
    {
      "epoch": 0.4251210714533981,
      "grad_norm": 0.10177981975001972,
      "learning_rate": 1.9865227568432343e-05,
      "loss": 0.1844,
      "step": 13760
    },
    {
      "epoch": 0.42543002572044275,
      "grad_norm": 0.09698420187479102,
      "learning_rate": 1.9850727743285336e-05,
      "loss": 0.1869,
      "step": 13770
    },
    {
      "epoch": 0.42573897998748733,
      "grad_norm": 0.1035133011371867,
      "learning_rate": 1.9836222854381258e-05,
      "loss": 0.1883,
      "step": 13780
    },
    {
      "epoch": 0.42604793425453197,
      "grad_norm": 0.10466490582533701,
      "learning_rate": 1.9821712916862015e-05,
      "loss": 0.187,
      "step": 13790
    },
    {
      "epoch": 0.4263568885215766,
      "grad_norm": 0.07449034686477328,
      "learning_rate": 1.9807197945874768e-05,
      "loss": 0.1925,
      "step": 13800
    },
    {
      "epoch": 0.42666584278862124,
      "grad_norm": 0.0773908754113331,
      "learning_rate": 1.979267795657195e-05,
      "loss": 0.1995,
      "step": 13810
    },
    {
      "epoch": 0.4269747970556658,
      "grad_norm": 0.08664857002889029,
      "learning_rate": 1.9778152964111222e-05,
      "loss": 0.1939,
      "step": 13820
    },
    {
      "epoch": 0.42728375132271046,
      "grad_norm": 0.08680359746443131,
      "learning_rate": 1.9763622983655464e-05,
      "loss": 0.1925,
      "step": 13830
    },
    {
      "epoch": 0.4275927055897551,
      "grad_norm": 0.08025786602482227,
      "learning_rate": 1.974908803037278e-05,
      "loss": 0.1852,
      "step": 13840
    },
    {
      "epoch": 0.4279016598567997,
      "grad_norm": 0.08289192691947796,
      "learning_rate": 1.9734548119436443e-05,
      "loss": 0.1995,
      "step": 13850
    },
    {
      "epoch": 0.4282106141238443,
      "grad_norm": 0.08901264043929087,
      "learning_rate": 1.9720003266024923e-05,
      "loss": 0.1826,
      "step": 13860
    },
    {
      "epoch": 0.42851956839088895,
      "grad_norm": 0.09935947498977617,
      "learning_rate": 1.9705453485321836e-05,
      "loss": 0.1857,
      "step": 13870
    },
    {
      "epoch": 0.4288285226579336,
      "grad_norm": 0.09254167682803593,
      "learning_rate": 1.9690898792515944e-05,
      "loss": 0.1882,
      "step": 13880
    },
    {
      "epoch": 0.42913747692497817,
      "grad_norm": 0.07926805991861635,
      "learning_rate": 1.9676339202801138e-05,
      "loss": 0.1851,
      "step": 13890
    },
    {
      "epoch": 0.4294464311920228,
      "grad_norm": 0.10148392340442433,
      "learning_rate": 1.9661774731376423e-05,
      "loss": 0.1828,
      "step": 13900
    },
    {
      "epoch": 0.42975538545906744,
      "grad_norm": 0.08943261878553953,
      "learning_rate": 1.9647205393445893e-05,
      "loss": 0.1865,
      "step": 13910
    },
    {
      "epoch": 0.430064339726112,
      "grad_norm": 0.08878554964023881,
      "learning_rate": 1.963263120421874e-05,
      "loss": 0.1875,
      "step": 13920
    },
    {
      "epoch": 0.43037329399315666,
      "grad_norm": 0.12975106655918497,
      "learning_rate": 1.961805217890919e-05,
      "loss": 0.1893,
      "step": 13930
    },
    {
      "epoch": 0.4306822482602013,
      "grad_norm": 0.08926495385588933,
      "learning_rate": 1.960346833273655e-05,
      "loss": 0.1891,
      "step": 13940
    },
    {
      "epoch": 0.43099120252724593,
      "grad_norm": 0.08942323160507486,
      "learning_rate": 1.9588879680925135e-05,
      "loss": 0.1891,
      "step": 13950
    },
    {
      "epoch": 0.4313001567942905,
      "grad_norm": 0.09477219895998903,
      "learning_rate": 1.9574286238704288e-05,
      "loss": 0.1868,
      "step": 13960
    },
    {
      "epoch": 0.43160911106133515,
      "grad_norm": 0.1056608039747648,
      "learning_rate": 1.9559688021308356e-05,
      "loss": 0.1848,
      "step": 13970
    },
    {
      "epoch": 0.4319180653283798,
      "grad_norm": 0.09683030360269723,
      "learning_rate": 1.9545085043976662e-05,
      "loss": 0.1884,
      "step": 13980
    },
    {
      "epoch": 0.43222701959542437,
      "grad_norm": 0.07194826476901177,
      "learning_rate": 1.9530477321953506e-05,
      "loss": 0.1984,
      "step": 13990
    },
    {
      "epoch": 0.432535973862469,
      "grad_norm": 0.08807504764938884,
      "learning_rate": 1.9515864870488132e-05,
      "loss": 0.1956,
      "step": 14000
    },
    {
      "epoch": 0.43284492812951364,
      "grad_norm": 0.09157827705197799,
      "learning_rate": 1.950124770483473e-05,
      "loss": 0.1853,
      "step": 14010
    },
    {
      "epoch": 0.4331538823965583,
      "grad_norm": 0.09126112475800932,
      "learning_rate": 1.9486625840252407e-05,
      "loss": 0.1875,
      "step": 14020
    },
    {
      "epoch": 0.43346283666360286,
      "grad_norm": 0.08139981920700338,
      "learning_rate": 1.9471999292005172e-05,
      "loss": 0.1856,
      "step": 14030
    },
    {
      "epoch": 0.4337717909306475,
      "grad_norm": 0.09353478022404632,
      "learning_rate": 1.945736807536193e-05,
      "loss": 0.1858,
      "step": 14040
    },
    {
      "epoch": 0.43408074519769213,
      "grad_norm": 0.08798355421745707,
      "learning_rate": 1.9442732205596462e-05,
      "loss": 0.1901,
      "step": 14050
    },
    {
      "epoch": 0.4343896994647367,
      "grad_norm": 0.09421803651890609,
      "learning_rate": 1.942809169798739e-05,
      "loss": 0.1886,
      "step": 14060
    },
    {
      "epoch": 0.43469865373178135,
      "grad_norm": 0.10686880282959817,
      "learning_rate": 1.941344656781819e-05,
      "loss": 0.1865,
      "step": 14070
    },
    {
      "epoch": 0.435007607998826,
      "grad_norm": 0.10075315487419693,
      "learning_rate": 1.9398796830377173e-05,
      "loss": 0.1824,
      "step": 14080
    },
    {
      "epoch": 0.4353165622658706,
      "grad_norm": 0.08157116161631962,
      "learning_rate": 1.9384142500957438e-05,
      "loss": 0.1899,
      "step": 14090
    },
    {
      "epoch": 0.4356255165329152,
      "grad_norm": 0.10215857572524423,
      "learning_rate": 1.9369483594856895e-05,
      "loss": 0.1936,
      "step": 14100
    },
    {
      "epoch": 0.43593447079995984,
      "grad_norm": 0.09463285224034536,
      "learning_rate": 1.935482012737822e-05,
      "loss": 0.1883,
      "step": 14110
    },
    {
      "epoch": 0.4362434250670045,
      "grad_norm": 0.1133192243836287,
      "learning_rate": 1.934015211382886e-05,
      "loss": 0.193,
      "step": 14120
    },
    {
      "epoch": 0.43655237933404906,
      "grad_norm": 0.085173513733126,
      "learning_rate": 1.9325479569521e-05,
      "loss": 0.1893,
      "step": 14130
    },
    {
      "epoch": 0.4368613336010937,
      "grad_norm": 0.08015246872997993,
      "learning_rate": 1.931080250977156e-05,
      "loss": 0.1976,
      "step": 14140
    },
    {
      "epoch": 0.43717028786813833,
      "grad_norm": 0.10004318093720328,
      "learning_rate": 1.929612094990217e-05,
      "loss": 0.1968,
      "step": 14150
    },
    {
      "epoch": 0.43747924213518297,
      "grad_norm": 0.10879740445860499,
      "learning_rate": 1.9281434905239162e-05,
      "loss": 0.1903,
      "step": 14160
    },
    {
      "epoch": 0.43778819640222755,
      "grad_norm": 0.08899220468383998,
      "learning_rate": 1.9266744391113547e-05,
      "loss": 0.1851,
      "step": 14170
    },
    {
      "epoch": 0.4380971506692722,
      "grad_norm": 0.07604691874249775,
      "learning_rate": 1.9252049422860998e-05,
      "loss": 0.1868,
      "step": 14180
    },
    {
      "epoch": 0.4384061049363168,
      "grad_norm": 0.09320862750343213,
      "learning_rate": 1.9237350015821847e-05,
      "loss": 0.1894,
      "step": 14190
    },
    {
      "epoch": 0.4387150592033614,
      "grad_norm": 0.09481159082107075,
      "learning_rate": 1.922264618534106e-05,
      "loss": 0.1869,
      "step": 14200
    },
    {
      "epoch": 0.43902401347040604,
      "grad_norm": 0.11383050740508886,
      "learning_rate": 1.9207937946768198e-05,
      "loss": 0.188,
      "step": 14210
    },
    {
      "epoch": 0.4393329677374507,
      "grad_norm": 0.08376420694724454,
      "learning_rate": 1.9193225315457458e-05,
      "loss": 0.1822,
      "step": 14220
    },
    {
      "epoch": 0.4396419220044953,
      "grad_norm": 0.09349902209276498,
      "learning_rate": 1.9178508306767596e-05,
      "loss": 0.1886,
      "step": 14230
    },
    {
      "epoch": 0.4399508762715399,
      "grad_norm": 0.10217687349968801,
      "learning_rate": 1.9163786936061954e-05,
      "loss": 0.189,
      "step": 14240
    },
    {
      "epoch": 0.44025983053858453,
      "grad_norm": 0.07900799082968471,
      "learning_rate": 1.914906121870842e-05,
      "loss": 0.1993,
      "step": 14250
    },
    {
      "epoch": 0.44056878480562917,
      "grad_norm": 0.07887609581075224,
      "learning_rate": 1.913433117007941e-05,
      "loss": 0.1884,
      "step": 14260
    },
    {
      "epoch": 0.44087773907267375,
      "grad_norm": 0.10239182083915561,
      "learning_rate": 1.9119596805551892e-05,
      "loss": 0.1876,
      "step": 14270
    },
    {
      "epoch": 0.4411866933397184,
      "grad_norm": 0.09126597517046119,
      "learning_rate": 1.9104858140507295e-05,
      "loss": 0.1921,
      "step": 14280
    },
    {
      "epoch": 0.441495647606763,
      "grad_norm": 0.09596889892113322,
      "learning_rate": 1.909011519033158e-05,
      "loss": 0.1849,
      "step": 14290
    },
    {
      "epoch": 0.44180460187380766,
      "grad_norm": 0.08987253107795462,
      "learning_rate": 1.9075367970415155e-05,
      "loss": 0.1844,
      "step": 14300
    },
    {
      "epoch": 0.44211355614085224,
      "grad_norm": 0.10479803884031098,
      "learning_rate": 1.9060616496152895e-05,
      "loss": 0.1873,
      "step": 14310
    },
    {
      "epoch": 0.4424225104078969,
      "grad_norm": 0.08682188064868668,
      "learning_rate": 1.9045860782944114e-05,
      "loss": 0.1873,
      "step": 14320
    },
    {
      "epoch": 0.4427314646749415,
      "grad_norm": 0.08269565963791928,
      "learning_rate": 1.9031100846192553e-05,
      "loss": 0.1937,
      "step": 14330
    },
    {
      "epoch": 0.4430404189419861,
      "grad_norm": 0.09119937691065266,
      "learning_rate": 1.901633670130636e-05,
      "loss": 0.1969,
      "step": 14340
    },
    {
      "epoch": 0.44334937320903073,
      "grad_norm": 0.10316256903724103,
      "learning_rate": 1.900156836369808e-05,
      "loss": 0.1862,
      "step": 14350
    },
    {
      "epoch": 0.44365832747607536,
      "grad_norm": 0.08498107350359378,
      "learning_rate": 1.8986795848784624e-05,
      "loss": 0.1955,
      "step": 14360
    },
    {
      "epoch": 0.44396728174312,
      "grad_norm": 0.09596006383160492,
      "learning_rate": 1.897201917198728e-05,
      "loss": 0.1854,
      "step": 14370
    },
    {
      "epoch": 0.4442762360101646,
      "grad_norm": 0.10427921561793586,
      "learning_rate": 1.8957238348731663e-05,
      "loss": 0.1895,
      "step": 14380
    },
    {
      "epoch": 0.4445851902772092,
      "grad_norm": 0.10313493002397604,
      "learning_rate": 1.894245339444773e-05,
      "loss": 0.1856,
      "step": 14390
    },
    {
      "epoch": 0.44489414454425386,
      "grad_norm": 0.10719812868284746,
      "learning_rate": 1.8927664324569747e-05,
      "loss": 0.1878,
      "step": 14400
    },
    {
      "epoch": 0.44520309881129844,
      "grad_norm": 0.07234710279240211,
      "learning_rate": 1.891287115453627e-05,
      "loss": 0.1882,
      "step": 14410
    },
    {
      "epoch": 0.4455120530783431,
      "grad_norm": 0.08478092043839999,
      "learning_rate": 1.8898073899790148e-05,
      "loss": 0.1949,
      "step": 14420
    },
    {
      "epoch": 0.4458210073453877,
      "grad_norm": 0.07844482376710984,
      "learning_rate": 1.8883272575778482e-05,
      "loss": 0.1909,
      "step": 14430
    },
    {
      "epoch": 0.44612996161243235,
      "grad_norm": 0.12773304235287128,
      "learning_rate": 1.8868467197952627e-05,
      "loss": 0.1904,
      "step": 14440
    },
    {
      "epoch": 0.4464389158794769,
      "grad_norm": 0.09154512340021774,
      "learning_rate": 1.8853657781768162e-05,
      "loss": 0.1868,
      "step": 14450
    },
    {
      "epoch": 0.44674787014652156,
      "grad_norm": 0.08640986151865691,
      "learning_rate": 1.8838844342684896e-05,
      "loss": 0.1878,
      "step": 14460
    },
    {
      "epoch": 0.4470568244135662,
      "grad_norm": 0.08791857572167769,
      "learning_rate": 1.8824026896166834e-05,
      "loss": 0.187,
      "step": 14470
    },
    {
      "epoch": 0.4473657786806108,
      "grad_norm": 0.09113953242817693,
      "learning_rate": 1.8809205457682144e-05,
      "loss": 0.191,
      "step": 14480
    },
    {
      "epoch": 0.4476747329476554,
      "grad_norm": 0.0872648088127917,
      "learning_rate": 1.8794380042703193e-05,
      "loss": 0.1865,
      "step": 14490
    },
    {
      "epoch": 0.44798368721470005,
      "grad_norm": 0.09349549807896199,
      "learning_rate": 1.8779550666706476e-05,
      "loss": 0.1847,
      "step": 14500
    },
    {
      "epoch": 0.4482926414817447,
      "grad_norm": 0.09931289139986942,
      "learning_rate": 1.8764717345172628e-05,
      "loss": 0.1947,
      "step": 14510
    },
    {
      "epoch": 0.4486015957487893,
      "grad_norm": 0.1094366859375454,
      "learning_rate": 1.874988009358641e-05,
      "loss": 0.1923,
      "step": 14520
    },
    {
      "epoch": 0.4489105500158339,
      "grad_norm": 0.10188243848239363,
      "learning_rate": 1.873503892743668e-05,
      "loss": 0.1855,
      "step": 14530
    },
    {
      "epoch": 0.44921950428287855,
      "grad_norm": 0.09185140886105803,
      "learning_rate": 1.872019386221638e-05,
      "loss": 0.1889,
      "step": 14540
    },
    {
      "epoch": 0.4495284585499231,
      "grad_norm": 0.1104889080374331,
      "learning_rate": 1.870534491342253e-05,
      "loss": 0.188,
      "step": 14550
    },
    {
      "epoch": 0.44983741281696776,
      "grad_norm": 0.09094614020062022,
      "learning_rate": 1.86904920965562e-05,
      "loss": 0.1853,
      "step": 14560
    },
    {
      "epoch": 0.4501463670840124,
      "grad_norm": 0.07721404526513222,
      "learning_rate": 1.8675635427122496e-05,
      "loss": 0.1923,
      "step": 14570
    },
    {
      "epoch": 0.45045532135105704,
      "grad_norm": 0.0905134023141106,
      "learning_rate": 1.8660774920630547e-05,
      "loss": 0.1824,
      "step": 14580
    },
    {
      "epoch": 0.4507642756181016,
      "grad_norm": 0.10448425942826392,
      "learning_rate": 1.864591059259349e-05,
      "loss": 0.1902,
      "step": 14590
    },
    {
      "epoch": 0.45107322988514625,
      "grad_norm": 0.07024958158034736,
      "learning_rate": 1.863104245852845e-05,
      "loss": 0.1837,
      "step": 14600
    },
    {
      "epoch": 0.4513821841521909,
      "grad_norm": 0.07809229034165306,
      "learning_rate": 1.861617053395652e-05,
      "loss": 0.1899,
      "step": 14610
    },
    {
      "epoch": 0.45169113841923547,
      "grad_norm": 0.08308319492796883,
      "learning_rate": 1.860129483440276e-05,
      "loss": 0.1865,
      "step": 14620
    },
    {
      "epoch": 0.4520000926862801,
      "grad_norm": 0.09811794640335583,
      "learning_rate": 1.8586415375396162e-05,
      "loss": 0.1852,
      "step": 14630
    },
    {
      "epoch": 0.45230904695332474,
      "grad_norm": 0.10580092058921427,
      "learning_rate": 1.8571532172469644e-05,
      "loss": 0.1809,
      "step": 14640
    },
    {
      "epoch": 0.4526180012203694,
      "grad_norm": 0.07895169754725709,
      "learning_rate": 1.8556645241160042e-05,
      "loss": 0.188,
      "step": 14650
    },
    {
      "epoch": 0.45292695548741396,
      "grad_norm": 0.09332059811801918,
      "learning_rate": 1.8541754597008066e-05,
      "loss": 0.1871,
      "step": 14660
    },
    {
      "epoch": 0.4532359097544586,
      "grad_norm": 0.10443773639876203,
      "learning_rate": 1.852686025555832e-05,
      "loss": 0.1889,
      "step": 14670
    },
    {
      "epoch": 0.45354486402150324,
      "grad_norm": 0.10221171996264637,
      "learning_rate": 1.851196223235926e-05,
      "loss": 0.1873,
      "step": 14680
    },
    {
      "epoch": 0.4538538182885478,
      "grad_norm": 0.08900271888338614,
      "learning_rate": 1.8497060542963183e-05,
      "loss": 0.1861,
      "step": 14690
    },
    {
      "epoch": 0.45416277255559245,
      "grad_norm": 0.10763607054704716,
      "learning_rate": 1.848215520292622e-05,
      "loss": 0.187,
      "step": 14700
    },
    {
      "epoch": 0.4544717268226371,
      "grad_norm": 0.09251662789119114,
      "learning_rate": 1.84672462278083e-05,
      "loss": 0.1867,
      "step": 14710
    },
    {
      "epoch": 0.4547806810896817,
      "grad_norm": 0.11126069465565208,
      "learning_rate": 1.845233363317316e-05,
      "loss": 0.1858,
      "step": 14720
    },
    {
      "epoch": 0.4550896353567263,
      "grad_norm": 0.10369781754615437,
      "learning_rate": 1.843741743458831e-05,
      "loss": 0.1934,
      "step": 14730
    },
    {
      "epoch": 0.45539858962377094,
      "grad_norm": 0.09510775173864588,
      "learning_rate": 1.8422497647625024e-05,
      "loss": 0.19,
      "step": 14740
    },
    {
      "epoch": 0.4557075438908156,
      "grad_norm": 0.0749622110991937,
      "learning_rate": 1.840757428785832e-05,
      "loss": 0.1876,
      "step": 14750
    },
    {
      "epoch": 0.45601649815786016,
      "grad_norm": 0.09607320796450156,
      "learning_rate": 1.839264737086695e-05,
      "loss": 0.1852,
      "step": 14760
    },
    {
      "epoch": 0.4563254524249048,
      "grad_norm": 0.08572065356867109,
      "learning_rate": 1.8377716912233372e-05,
      "loss": 0.1859,
      "step": 14770
    },
    {
      "epoch": 0.45663440669194943,
      "grad_norm": 0.09045929059827176,
      "learning_rate": 1.8362782927543748e-05,
      "loss": 0.1852,
      "step": 14780
    },
    {
      "epoch": 0.45694336095899407,
      "grad_norm": 0.08558816792747759,
      "learning_rate": 1.8347845432387914e-05,
      "loss": 0.1869,
      "step": 14790
    },
    {
      "epoch": 0.45725231522603865,
      "grad_norm": 0.09781406993689364,
      "learning_rate": 1.8332904442359382e-05,
      "loss": 0.1879,
      "step": 14800
    },
    {
      "epoch": 0.4575612694930833,
      "grad_norm": 0.09270766692079555,
      "learning_rate": 1.83179599730553e-05,
      "loss": 0.1848,
      "step": 14810
    },
    {
      "epoch": 0.4578702237601279,
      "grad_norm": 0.0921124967651755,
      "learning_rate": 1.8303012040076456e-05,
      "loss": 0.1949,
      "step": 14820
    },
    {
      "epoch": 0.4581791780271725,
      "grad_norm": 0.09234506599274829,
      "learning_rate": 1.8288060659027245e-05,
      "loss": 0.1847,
      "step": 14830
    },
    {
      "epoch": 0.45848813229421714,
      "grad_norm": 0.09590755864665845,
      "learning_rate": 1.8273105845515677e-05,
      "loss": 0.184,
      "step": 14840
    },
    {
      "epoch": 0.4587970865612618,
      "grad_norm": 0.09562393573248085,
      "learning_rate": 1.825814761515333e-05,
      "loss": 0.1914,
      "step": 14850
    },
    {
      "epoch": 0.4591060408283064,
      "grad_norm": 0.11029514353162492,
      "learning_rate": 1.8243185983555354e-05,
      "loss": 0.1882,
      "step": 14860
    },
    {
      "epoch": 0.459414995095351,
      "grad_norm": 0.0966739344651011,
      "learning_rate": 1.822822096634045e-05,
      "loss": 0.187,
      "step": 14870
    },
    {
      "epoch": 0.45972394936239563,
      "grad_norm": 0.10608787716057655,
      "learning_rate": 1.821325257913086e-05,
      "loss": 0.1881,
      "step": 14880
    },
    {
      "epoch": 0.46003290362944027,
      "grad_norm": 0.09042449727808383,
      "learning_rate": 1.819828083755233e-05,
      "loss": 0.1864,
      "step": 14890
    },
    {
      "epoch": 0.46034185789648485,
      "grad_norm": 0.0901560389505401,
      "learning_rate": 1.8183305757234122e-05,
      "loss": 0.1819,
      "step": 14900
    },
    {
      "epoch": 0.4606508121635295,
      "grad_norm": 0.26809753696792066,
      "learning_rate": 1.8168327353808966e-05,
      "loss": 0.1891,
      "step": 14910
    },
    {
      "epoch": 0.4609597664305741,
      "grad_norm": 0.14454486069777536,
      "learning_rate": 1.8153345642913092e-05,
      "loss": 0.1838,
      "step": 14920
    },
    {
      "epoch": 0.46126872069761876,
      "grad_norm": 0.09401613139941153,
      "learning_rate": 1.813836064018614e-05,
      "loss": 0.1902,
      "step": 14930
    },
    {
      "epoch": 0.46157767496466334,
      "grad_norm": 0.09662306497757542,
      "learning_rate": 1.812337236127122e-05,
      "loss": 0.1886,
      "step": 14940
    },
    {
      "epoch": 0.461886629231708,
      "grad_norm": 0.08011186199733386,
      "learning_rate": 1.810838082181485e-05,
      "loss": 0.1836,
      "step": 14950
    },
    {
      "epoch": 0.4621955834987526,
      "grad_norm": 0.0959391996675847,
      "learning_rate": 1.8093386037466948e-05,
      "loss": 0.1934,
      "step": 14960
    },
    {
      "epoch": 0.4625045377657972,
      "grad_norm": 0.0984128991567507,
      "learning_rate": 1.807838802388083e-05,
      "loss": 0.1841,
      "step": 14970
    },
    {
      "epoch": 0.46281349203284183,
      "grad_norm": 0.09960467420393028,
      "learning_rate": 1.8063386796713175e-05,
      "loss": 0.1868,
      "step": 14980
    },
    {
      "epoch": 0.46312244629988647,
      "grad_norm": 0.08738018195659383,
      "learning_rate": 1.8048382371624022e-05,
      "loss": 0.1841,
      "step": 14990
    },
    {
      "epoch": 0.4634314005669311,
      "grad_norm": 0.0858575462808963,
      "learning_rate": 1.803337476427674e-05,
      "loss": 0.1846,
      "step": 15000
    },
    {
      "epoch": 0.4637403548339757,
      "grad_norm": 0.11008539341333988,
      "learning_rate": 1.8018363990338022e-05,
      "loss": 0.1876,
      "step": 15010
    },
    {
      "epoch": 0.4640493091010203,
      "grad_norm": 0.08855632244478165,
      "learning_rate": 1.8003350065477877e-05,
      "loss": 0.1905,
      "step": 15020
    },
    {
      "epoch": 0.46435826336806496,
      "grad_norm": 0.09543370847758748,
      "learning_rate": 1.7988333005369595e-05,
      "loss": 0.1856,
      "step": 15030
    },
    {
      "epoch": 0.46466721763510954,
      "grad_norm": 0.11468696699357876,
      "learning_rate": 1.797331282568974e-05,
      "loss": 0.185,
      "step": 15040
    },
    {
      "epoch": 0.4649761719021542,
      "grad_norm": 0.0875863282134661,
      "learning_rate": 1.7958289542118132e-05,
      "loss": 0.1825,
      "step": 15050
    },
    {
      "epoch": 0.4652851261691988,
      "grad_norm": 0.09530661495635731,
      "learning_rate": 1.794326317033783e-05,
      "loss": 0.185,
      "step": 15060
    },
    {
      "epoch": 0.46559408043624345,
      "grad_norm": 0.10196903942408467,
      "learning_rate": 1.792823372603512e-05,
      "loss": 0.1836,
      "step": 15070
    },
    {
      "epoch": 0.46590303470328803,
      "grad_norm": 0.0807715602315761,
      "learning_rate": 1.7913201224899495e-05,
      "loss": 0.183,
      "step": 15080
    },
    {
      "epoch": 0.46621198897033267,
      "grad_norm": 0.09615660065111356,
      "learning_rate": 1.7898165682623637e-05,
      "loss": 0.1919,
      "step": 15090
    },
    {
      "epoch": 0.4665209432373773,
      "grad_norm": 0.10904360246970066,
      "learning_rate": 1.7883127114903406e-05,
      "loss": 0.1852,
      "step": 15100
    },
    {
      "epoch": 0.4668298975044219,
      "grad_norm": 0.12369054276626484,
      "learning_rate": 1.7868085537437812e-05,
      "loss": 0.1893,
      "step": 15110
    },
    {
      "epoch": 0.4671388517714665,
      "grad_norm": 0.08634182660069728,
      "learning_rate": 1.7853040965929023e-05,
      "loss": 0.1933,
      "step": 15120
    },
    {
      "epoch": 0.46744780603851116,
      "grad_norm": 0.08031336460288044,
      "learning_rate": 1.7837993416082318e-05,
      "loss": 0.1867,
      "step": 15130
    },
    {
      "epoch": 0.4677567603055558,
      "grad_norm": 0.09689839372096767,
      "learning_rate": 1.7822942903606088e-05,
      "loss": 0.1866,
      "step": 15140
    },
    {
      "epoch": 0.4680657145726004,
      "grad_norm": 0.10264211622153026,
      "learning_rate": 1.780788944421182e-05,
      "loss": 0.191,
      "step": 15150
    },
    {
      "epoch": 0.468374668839645,
      "grad_norm": 0.09872002146016506,
      "learning_rate": 1.7792833053614076e-05,
      "loss": 0.1876,
      "step": 15160
    },
    {
      "epoch": 0.46868362310668965,
      "grad_norm": 0.09667286155920068,
      "learning_rate": 1.777777374753048e-05,
      "loss": 0.1964,
      "step": 15170
    },
    {
      "epoch": 0.46899257737373423,
      "grad_norm": 0.09691054345384882,
      "learning_rate": 1.7762711541681695e-05,
      "loss": 0.1932,
      "step": 15180
    },
    {
      "epoch": 0.46930153164077887,
      "grad_norm": 0.09086575672944316,
      "learning_rate": 1.7747646451791408e-05,
      "loss": 0.1977,
      "step": 15190
    },
    {
      "epoch": 0.4696104859078235,
      "grad_norm": 0.09227703776636187,
      "learning_rate": 1.773257849358633e-05,
      "loss": 0.1828,
      "step": 15200
    },
    {
      "epoch": 0.46991944017486814,
      "grad_norm": 0.09505838352684795,
      "learning_rate": 1.7717507682796156e-05,
      "loss": 0.1864,
      "step": 15210
    },
    {
      "epoch": 0.4702283944419127,
      "grad_norm": 0.14433938921800793,
      "learning_rate": 1.7702434035153563e-05,
      "loss": 0.1914,
      "step": 15220
    },
    {
      "epoch": 0.47053734870895736,
      "grad_norm": 0.1092071177720221,
      "learning_rate": 1.7687357566394186e-05,
      "loss": 0.1868,
      "step": 15230
    },
    {
      "epoch": 0.470846302976002,
      "grad_norm": 0.08638758400992423,
      "learning_rate": 1.767227829225661e-05,
      "loss": 0.1851,
      "step": 15240
    },
    {
      "epoch": 0.4711552572430466,
      "grad_norm": 0.09367265333682509,
      "learning_rate": 1.7657196228482337e-05,
      "loss": 0.1883,
      "step": 15250
    },
    {
      "epoch": 0.4714642115100912,
      "grad_norm": 0.07692000952585477,
      "learning_rate": 1.7642111390815796e-05,
      "loss": 0.187,
      "step": 15260
    },
    {
      "epoch": 0.47177316577713585,
      "grad_norm": 0.08526520780711176,
      "learning_rate": 1.762702379500431e-05,
      "loss": 0.1856,
      "step": 15270
    },
    {
      "epoch": 0.4720821200441805,
      "grad_norm": 0.08646246182945996,
      "learning_rate": 1.761193345679807e-05,
      "loss": 0.1812,
      "step": 15280
    },
    {
      "epoch": 0.47239107431122507,
      "grad_norm": 0.07694228733864118,
      "learning_rate": 1.759684039195013e-05,
      "loss": 0.1904,
      "step": 15290
    },
    {
      "epoch": 0.4727000285782697,
      "grad_norm": 0.11305783364625918,
      "learning_rate": 1.7581744616216407e-05,
      "loss": 0.1921,
      "step": 15300
    },
    {
      "epoch": 0.47300898284531434,
      "grad_norm": 0.09757586204286521,
      "learning_rate": 1.7566646145355636e-05,
      "loss": 0.1841,
      "step": 15310
    },
    {
      "epoch": 0.4733179371123589,
      "grad_norm": 0.09147307713691727,
      "learning_rate": 1.755154499512936e-05,
      "loss": 0.1846,
      "step": 15320
    },
    {
      "epoch": 0.47362689137940356,
      "grad_norm": 0.09257491179768478,
      "learning_rate": 1.753644118130194e-05,
      "loss": 0.1828,
      "step": 15330
    },
    {
      "epoch": 0.4739358456464482,
      "grad_norm": 0.08393761776522794,
      "learning_rate": 1.7521334719640487e-05,
      "loss": 0.1861,
      "step": 15340
    },
    {
      "epoch": 0.47424479991349283,
      "grad_norm": 0.08964750078537113,
      "learning_rate": 1.7506225625914905e-05,
      "loss": 0.1816,
      "step": 15350
    },
    {
      "epoch": 0.4745537541805374,
      "grad_norm": 0.0848320103314238,
      "learning_rate": 1.7491113915897832e-05,
      "loss": 0.1827,
      "step": 15360
    },
    {
      "epoch": 0.47486270844758205,
      "grad_norm": 0.08596297431686581,
      "learning_rate": 1.7475999605364633e-05,
      "loss": 0.1821,
      "step": 15370
    },
    {
      "epoch": 0.4751716627146267,
      "grad_norm": 0.09322592009027474,
      "learning_rate": 1.7460882710093403e-05,
      "loss": 0.1918,
      "step": 15380
    },
    {
      "epoch": 0.47548061698167127,
      "grad_norm": 0.1071344134291526,
      "learning_rate": 1.7445763245864917e-05,
      "loss": 0.1842,
      "step": 15390
    },
    {
      "epoch": 0.4757895712487159,
      "grad_norm": 0.0939611714636112,
      "learning_rate": 1.7430641228462642e-05,
      "loss": 0.1832,
      "step": 15400
    },
    {
      "epoch": 0.47609852551576054,
      "grad_norm": 0.10232536322383694,
      "learning_rate": 1.7415516673672713e-05,
      "loss": 0.1848,
      "step": 15410
    },
    {
      "epoch": 0.4764074797828052,
      "grad_norm": 0.0782286356173196,
      "learning_rate": 1.7400389597283904e-05,
      "loss": 0.1805,
      "step": 15420
    },
    {
      "epoch": 0.47671643404984976,
      "grad_norm": 0.09208469805587671,
      "learning_rate": 1.7385260015087627e-05,
      "loss": 0.1832,
      "step": 15430
    },
    {
      "epoch": 0.4770253883168944,
      "grad_norm": 0.10857360895510182,
      "learning_rate": 1.7370127942877916e-05,
      "loss": 0.1925,
      "step": 15440
    },
    {
      "epoch": 0.47733434258393903,
      "grad_norm": 0.09873037066071629,
      "learning_rate": 1.7354993396451392e-05,
      "loss": 0.1814,
      "step": 15450
    },
    {
      "epoch": 0.4776432968509836,
      "grad_norm": 0.08573551565865423,
      "learning_rate": 1.7339856391607268e-05,
      "loss": 0.1865,
      "step": 15460
    },
    {
      "epoch": 0.47795225111802825,
      "grad_norm": 0.08016237916706054,
      "learning_rate": 1.7324716944147322e-05,
      "loss": 0.1862,
      "step": 15470
    },
    {
      "epoch": 0.4782612053850729,
      "grad_norm": 0.11483734956349438,
      "learning_rate": 1.7309575069875874e-05,
      "loss": 0.1893,
      "step": 15480
    },
    {
      "epoch": 0.4785701596521175,
      "grad_norm": 0.08225215622883346,
      "learning_rate": 1.729443078459979e-05,
      "loss": 0.1967,
      "step": 15490
    },
    {
      "epoch": 0.4788791139191621,
      "grad_norm": 0.09516950564167469,
      "learning_rate": 1.727928410412845e-05,
      "loss": 0.1968,
      "step": 15500
    },
    {
      "epoch": 0.47918806818620674,
      "grad_norm": 0.13169828452168447,
      "learning_rate": 1.726413504427372e-05,
      "loss": 0.1944,
      "step": 15510
    },
    {
      "epoch": 0.4794970224532514,
      "grad_norm": 0.0841868815675506,
      "learning_rate": 1.7248983620849967e-05,
      "loss": 0.1878,
      "step": 15520
    },
    {
      "epoch": 0.47980597672029596,
      "grad_norm": 0.06872492837280653,
      "learning_rate": 1.723382984967402e-05,
      "loss": 0.1856,
      "step": 15530
    },
    {
      "epoch": 0.4801149309873406,
      "grad_norm": 0.0724457164126773,
      "learning_rate": 1.7218673746565154e-05,
      "loss": 0.186,
      "step": 15540
    },
    {
      "epoch": 0.48042388525438523,
      "grad_norm": 0.09733579992189363,
      "learning_rate": 1.7203515327345085e-05,
      "loss": 0.1905,
      "step": 15550
    },
    {
      "epoch": 0.48073283952142987,
      "grad_norm": 0.08527111330048416,
      "learning_rate": 1.7188354607837945e-05,
      "loss": 0.1909,
      "step": 15560
    },
    {
      "epoch": 0.48104179378847445,
      "grad_norm": 0.12445708330679682,
      "learning_rate": 1.7173191603870268e-05,
      "loss": 0.1934,
      "step": 15570
    },
    {
      "epoch": 0.4813507480555191,
      "grad_norm": 0.08252945797593983,
      "learning_rate": 1.7158026331270966e-05,
      "loss": 0.1924,
      "step": 15580
    },
    {
      "epoch": 0.4816597023225637,
      "grad_norm": 0.09498089069156578,
      "learning_rate": 1.714285880587133e-05,
      "loss": 0.1914,
      "step": 15590
    },
    {
      "epoch": 0.4819686565896083,
      "grad_norm": 0.08935502259651591,
      "learning_rate": 1.7127689043505e-05,
      "loss": 0.1948,
      "step": 15600
    },
    {
      "epoch": 0.48227761085665294,
      "grad_norm": 0.08517155396689621,
      "learning_rate": 1.7112517060007936e-05,
      "loss": 0.1821,
      "step": 15610
    },
    {
      "epoch": 0.4825865651236976,
      "grad_norm": 0.08451627463555345,
      "learning_rate": 1.7097342871218443e-05,
      "loss": 0.1893,
      "step": 15620
    },
    {
      "epoch": 0.4828955193907422,
      "grad_norm": 0.08643284441209946,
      "learning_rate": 1.7082166492977106e-05,
      "loss": 0.181,
      "step": 15630
    },
    {
      "epoch": 0.4832044736577868,
      "grad_norm": 0.17393290533310926,
      "learning_rate": 1.7066987941126807e-05,
      "loss": 0.1901,
      "step": 15640
    },
    {
      "epoch": 0.48351342792483143,
      "grad_norm": 0.0795347019991599,
      "learning_rate": 1.7051807231512693e-05,
      "loss": 0.1951,
      "step": 15650
    },
    {
      "epoch": 0.48382238219187607,
      "grad_norm": 0.09238073290615917,
      "learning_rate": 1.703662437998216e-05,
      "loss": 0.1826,
      "step": 15660
    },
    {
      "epoch": 0.48413133645892065,
      "grad_norm": 0.08937813081599442,
      "learning_rate": 1.7021439402384854e-05,
      "loss": 0.1878,
      "step": 15670
    },
    {
      "epoch": 0.4844402907259653,
      "grad_norm": 0.0991162805047522,
      "learning_rate": 1.7006252314572626e-05,
      "loss": 0.1871,
      "step": 15680
    },
    {
      "epoch": 0.4847492449930099,
      "grad_norm": 0.08471012686275417,
      "learning_rate": 1.6991063132399536e-05,
      "loss": 0.1893,
      "step": 15690
    },
    {
      "epoch": 0.48505819926005456,
      "grad_norm": 0.10987187352916403,
      "learning_rate": 1.697587187172183e-05,
      "loss": 0.1839,
      "step": 15700
    },
    {
      "epoch": 0.48536715352709914,
      "grad_norm": 0.09193744545349228,
      "learning_rate": 1.6960678548397917e-05,
      "loss": 0.1808,
      "step": 15710
    },
    {
      "epoch": 0.4856761077941438,
      "grad_norm": 0.09837930663245117,
      "learning_rate": 1.6945483178288373e-05,
      "loss": 0.1871,
      "step": 15720
    },
    {
      "epoch": 0.4859850620611884,
      "grad_norm": 0.09894497268456612,
      "learning_rate": 1.6930285777255903e-05,
      "loss": 0.189,
      "step": 15730
    },
    {
      "epoch": 0.486294016328233,
      "grad_norm": 0.07747379122039307,
      "learning_rate": 1.6915086361165326e-05,
      "loss": 0.1828,
      "step": 15740
    },
    {
      "epoch": 0.48660297059527763,
      "grad_norm": 0.10507113461789214,
      "learning_rate": 1.689988494588357e-05,
      "loss": 0.183,
      "step": 15750
    },
    {
      "epoch": 0.48691192486232227,
      "grad_norm": 0.09465091321443915,
      "learning_rate": 1.688468154727966e-05,
      "loss": 0.1837,
      "step": 15760
    },
    {
      "epoch": 0.4872208791293669,
      "grad_norm": 0.10640222169454058,
      "learning_rate": 1.6869476181224674e-05,
      "loss": 0.1894,
      "step": 15770
    },
    {
      "epoch": 0.4875298333964115,
      "grad_norm": 0.0918403510265723,
      "learning_rate": 1.6854268863591748e-05,
      "loss": 0.1831,
      "step": 15780
    },
    {
      "epoch": 0.4878387876634561,
      "grad_norm": 0.09166627082137223,
      "learning_rate": 1.6839059610256072e-05,
      "loss": 0.1837,
      "step": 15790
    },
    {
      "epoch": 0.48814774193050076,
      "grad_norm": 0.10530109258748743,
      "learning_rate": 1.6823848437094833e-05,
      "loss": 0.201,
      "step": 15800
    },
    {
      "epoch": 0.48845669619754534,
      "grad_norm": 0.08801482707933495,
      "learning_rate": 1.680863535998724e-05,
      "loss": 0.1891,
      "step": 15810
    },
    {
      "epoch": 0.48876565046459,
      "grad_norm": 0.07688953322096566,
      "learning_rate": 1.6793420394814476e-05,
      "loss": 0.1843,
      "step": 15820
    },
    {
      "epoch": 0.4890746047316346,
      "grad_norm": 0.0995223853600139,
      "learning_rate": 1.677820355745971e-05,
      "loss": 0.1896,
      "step": 15830
    },
    {
      "epoch": 0.48938355899867925,
      "grad_norm": 0.07868309536869768,
      "learning_rate": 1.676298486380805e-05,
      "loss": 0.1868,
      "step": 15840
    },
    {
      "epoch": 0.48969251326572383,
      "grad_norm": 0.11595632274024195,
      "learning_rate": 1.674776432974655e-05,
      "loss": 0.1834,
      "step": 15850
    },
    {
      "epoch": 0.49000146753276846,
      "grad_norm": 0.08790260634307571,
      "learning_rate": 1.6732541971164184e-05,
      "loss": 0.1883,
      "step": 15860
    },
    {
      "epoch": 0.4903104217998131,
      "grad_norm": 0.10067202460327114,
      "learning_rate": 1.6717317803951828e-05,
      "loss": 0.1866,
      "step": 15870
    },
    {
      "epoch": 0.4906193760668577,
      "grad_norm": 0.0897278223733937,
      "learning_rate": 1.6702091844002256e-05,
      "loss": 0.1882,
      "step": 15880
    },
    {
      "epoch": 0.4909283303339023,
      "grad_norm": 0.10306827915240233,
      "learning_rate": 1.66868641072101e-05,
      "loss": 0.1826,
      "step": 15890
    },
    {
      "epoch": 0.49123728460094696,
      "grad_norm": 0.08544730376805611,
      "learning_rate": 1.667163460947186e-05,
      "loss": 0.192,
      "step": 15900
    },
    {
      "epoch": 0.4915462388679916,
      "grad_norm": 0.09608200618848285,
      "learning_rate": 1.665640336668586e-05,
      "loss": 0.1845,
      "step": 15910
    },
    {
      "epoch": 0.4918551931350362,
      "grad_norm": 0.09511371865163808,
      "learning_rate": 1.664117039475225e-05,
      "loss": 0.1952,
      "step": 15920
    },
    {
      "epoch": 0.4921641474020808,
      "grad_norm": 0.10923160317234618,
      "learning_rate": 1.6625935709572997e-05,
      "loss": 0.1878,
      "step": 15930
    },
    {
      "epoch": 0.49247310166912545,
      "grad_norm": 0.07187305100740843,
      "learning_rate": 1.6610699327051842e-05,
      "loss": 0.1842,
      "step": 15940
    },
    {
      "epoch": 0.49278205593617,
      "grad_norm": 0.08941512923731464,
      "learning_rate": 1.6595461263094316e-05,
      "loss": 0.1849,
      "step": 15950
    },
    {
      "epoch": 0.49309101020321466,
      "grad_norm": 0.09825501997247763,
      "learning_rate": 1.6580221533607672e-05,
      "loss": 0.1831,
      "step": 15960
    },
    {
      "epoch": 0.4933999644702593,
      "grad_norm": 0.08596116657977662,
      "learning_rate": 1.6564980154500937e-05,
      "loss": 0.1839,
      "step": 15970
    },
    {
      "epoch": 0.49370891873730394,
      "grad_norm": 0.09269597813122786,
      "learning_rate": 1.6549737141684837e-05,
      "loss": 0.1885,
      "step": 15980
    },
    {
      "epoch": 0.4940178730043485,
      "grad_norm": 0.0799256163556364,
      "learning_rate": 1.653449251107182e-05,
      "loss": 0.1846,
      "step": 15990
    },
    {
      "epoch": 0.49432682727139315,
      "grad_norm": 0.09250933308177511,
      "learning_rate": 1.651924627857601e-05,
      "loss": 0.1843,
      "step": 16000
    },
    {
      "epoch": 0.4946357815384378,
      "grad_norm": 0.10959023050865643,
      "learning_rate": 1.650399846011321e-05,
      "loss": 0.1834,
      "step": 16010
    },
    {
      "epoch": 0.49494473580548237,
      "grad_norm": 0.07759338064387165,
      "learning_rate": 1.6488749071600875e-05,
      "loss": 0.1796,
      "step": 16020
    },
    {
      "epoch": 0.495253690072527,
      "grad_norm": 0.11275098432585824,
      "learning_rate": 1.64734981289581e-05,
      "loss": 0.1835,
      "step": 16030
    },
    {
      "epoch": 0.49556264433957165,
      "grad_norm": 0.09234924044507742,
      "learning_rate": 1.6458245648105604e-05,
      "loss": 0.184,
      "step": 16040
    },
    {
      "epoch": 0.4958715986066163,
      "grad_norm": 0.0933133747622666,
      "learning_rate": 1.6442991644965716e-05,
      "loss": 0.1839,
      "step": 16050
    },
    {
      "epoch": 0.49618055287366086,
      "grad_norm": 0.09976661795148338,
      "learning_rate": 1.642773613546234e-05,
      "loss": 0.1838,
      "step": 16060
    },
    {
      "epoch": 0.4964895071407055,
      "grad_norm": 0.08267966670246835,
      "learning_rate": 1.6412479135520967e-05,
      "loss": 0.1842,
      "step": 16070
    },
    {
      "epoch": 0.49679846140775014,
      "grad_norm": 0.10400517024831424,
      "learning_rate": 1.6397220661068637e-05,
      "loss": 0.1878,
      "step": 16080
    },
    {
      "epoch": 0.4971074156747947,
      "grad_norm": 0.09110354071661138,
      "learning_rate": 1.638196072803393e-05,
      "loss": 0.1845,
      "step": 16090
    },
    {
      "epoch": 0.49741636994183935,
      "grad_norm": 0.09954264527195751,
      "learning_rate": 1.636669935234695e-05,
      "loss": 0.1869,
      "step": 16100
    },
    {
      "epoch": 0.497725324208884,
      "grad_norm": 0.09282185253342333,
      "learning_rate": 1.6351436549939305e-05,
      "loss": 0.1844,
      "step": 16110
    },
    {
      "epoch": 0.4980342784759286,
      "grad_norm": 0.08516110638246024,
      "learning_rate": 1.6336172336744096e-05,
      "loss": 0.1892,
      "step": 16120
    },
    {
      "epoch": 0.4983432327429732,
      "grad_norm": 0.07979420834822085,
      "learning_rate": 1.6320906728695886e-05,
      "loss": 0.1896,
      "step": 16130
    },
    {
      "epoch": 0.49865218701001784,
      "grad_norm": 0.0960932375296174,
      "learning_rate": 1.6305639741730718e-05,
      "loss": 0.1989,
      "step": 16140
    },
    {
      "epoch": 0.4989611412770625,
      "grad_norm": 0.09548736710256746,
      "learning_rate": 1.6290371391786047e-05,
      "loss": 0.1916,
      "step": 16150
    },
    {
      "epoch": 0.49927009554410706,
      "grad_norm": 0.07591727105674324,
      "learning_rate": 1.6275101694800766e-05,
      "loss": 0.1875,
      "step": 16160
    },
    {
      "epoch": 0.4995790498111517,
      "grad_norm": 0.08142558024410718,
      "learning_rate": 1.6259830666715173e-05,
      "loss": 0.1829,
      "step": 16170
    },
    {
      "epoch": 0.49988800407819634,
      "grad_norm": 0.08927629097026177,
      "learning_rate": 1.624455832347095e-05,
      "loss": 0.1904,
      "step": 16180
    },
    {
      "epoch": 0.5001969583452409,
      "grad_norm": 0.09574685711476999,
      "learning_rate": 1.6229284681011163e-05,
      "loss": 0.1888,
      "step": 16190
    },
    {
      "epoch": 0.5005059126122856,
      "grad_norm": 0.14845410169283285,
      "learning_rate": 1.6214009755280216e-05,
      "loss": 0.1892,
      "step": 16200
    },
    {
      "epoch": 0.5008148668793302,
      "grad_norm": 0.11082846390690437,
      "learning_rate": 1.6198733562223867e-05,
      "loss": 0.1937,
      "step": 16210
    },
    {
      "epoch": 0.5011238211463748,
      "grad_norm": 0.08003900253173751,
      "learning_rate": 1.61834561177892e-05,
      "loss": 0.1853,
      "step": 16220
    },
    {
      "epoch": 0.5014327754134195,
      "grad_norm": 0.08739416279208018,
      "learning_rate": 1.6168177437924588e-05,
      "loss": 0.1981,
      "step": 16230
    },
    {
      "epoch": 0.501741729680464,
      "grad_norm": 0.11047366649303564,
      "learning_rate": 1.615289753857971e-05,
      "loss": 0.1879,
      "step": 16240
    },
    {
      "epoch": 0.5020506839475086,
      "grad_norm": 0.1056427332737057,
      "learning_rate": 1.6137616435705515e-05,
      "loss": 0.1889,
      "step": 16250
    },
    {
      "epoch": 0.5023596382145533,
      "grad_norm": 0.10615334695738451,
      "learning_rate": 1.6122334145254203e-05,
      "loss": 0.1863,
      "step": 16260
    },
    {
      "epoch": 0.5026685924815979,
      "grad_norm": 0.08721635048425189,
      "learning_rate": 1.6107050683179215e-05,
      "loss": 0.1846,
      "step": 16270
    },
    {
      "epoch": 0.5029775467486425,
      "grad_norm": 0.1018767876951892,
      "learning_rate": 1.609176606543522e-05,
      "loss": 0.1891,
      "step": 16280
    },
    {
      "epoch": 0.5032865010156872,
      "grad_norm": 0.10092129533153009,
      "learning_rate": 1.607648030797809e-05,
      "loss": 0.1901,
      "step": 16290
    },
    {
      "epoch": 0.5035954552827318,
      "grad_norm": 0.1040256373196848,
      "learning_rate": 1.6061193426764878e-05,
      "loss": 0.1855,
      "step": 16300
    },
    {
      "epoch": 0.5039044095497764,
      "grad_norm": 0.09425086408387737,
      "learning_rate": 1.6045905437753836e-05,
      "loss": 0.1861,
      "step": 16310
    },
    {
      "epoch": 0.504213363816821,
      "grad_norm": 0.09241402729188,
      "learning_rate": 1.6030616356904338e-05,
      "loss": 0.1817,
      "step": 16320
    },
    {
      "epoch": 0.5045223180838656,
      "grad_norm": 0.10000972176400486,
      "learning_rate": 1.6015326200176925e-05,
      "loss": 0.1836,
      "step": 16330
    },
    {
      "epoch": 0.5048312723509103,
      "grad_norm": 0.0902206852113459,
      "learning_rate": 1.6000034983533248e-05,
      "loss": 0.1828,
      "step": 16340
    },
    {
      "epoch": 0.5051402266179549,
      "grad_norm": 0.0829357861710911,
      "learning_rate": 1.598474272293607e-05,
      "loss": 0.1915,
      "step": 16350
    },
    {
      "epoch": 0.5054491808849995,
      "grad_norm": 0.08269650438296887,
      "learning_rate": 1.5969449434349236e-05,
      "loss": 0.1924,
      "step": 16360
    },
    {
      "epoch": 0.5057581351520442,
      "grad_norm": 0.07297811207458096,
      "learning_rate": 1.5954155133737674e-05,
      "loss": 0.1986,
      "step": 16370
    },
    {
      "epoch": 0.5060670894190887,
      "grad_norm": 0.0785437829061446,
      "learning_rate": 1.5938859837067366e-05,
      "loss": 0.1887,
      "step": 16380
    },
    {
      "epoch": 0.5063760436861333,
      "grad_norm": 0.07713942412917614,
      "learning_rate": 1.5923563560305332e-05,
      "loss": 0.1866,
      "step": 16390
    },
    {
      "epoch": 0.506684997953178,
      "grad_norm": 0.13864850216880983,
      "learning_rate": 1.5908266319419613e-05,
      "loss": 0.1839,
      "step": 16400
    },
    {
      "epoch": 0.5069939522202226,
      "grad_norm": 0.09274446071187946,
      "learning_rate": 1.5892968130379256e-05,
      "loss": 0.1917,
      "step": 16410
    },
    {
      "epoch": 0.5073029064872672,
      "grad_norm": 0.08491881663539312,
      "learning_rate": 1.5877669009154302e-05,
      "loss": 0.1896,
      "step": 16420
    },
    {
      "epoch": 0.5076118607543119,
      "grad_norm": 0.0888411644518085,
      "learning_rate": 1.5862368971715765e-05,
      "loss": 0.1846,
      "step": 16430
    },
    {
      "epoch": 0.5079208150213564,
      "grad_norm": 0.08537009899854889,
      "learning_rate": 1.5847068034035612e-05,
      "loss": 0.193,
      "step": 16440
    },
    {
      "epoch": 0.5082297692884011,
      "grad_norm": 0.08129500642041267,
      "learning_rate": 1.5831766212086754e-05,
      "loss": 0.1918,
      "step": 16450
    },
    {
      "epoch": 0.5085387235554457,
      "grad_norm": 0.0933932025729624,
      "learning_rate": 1.581646352184302e-05,
      "loss": 0.1947,
      "step": 16460
    },
    {
      "epoch": 0.5088476778224903,
      "grad_norm": 0.09391013949854107,
      "learning_rate": 1.5801159979279153e-05,
      "loss": 0.1912,
      "step": 16470
    },
    {
      "epoch": 0.509156632089535,
      "grad_norm": 0.07950096960636836,
      "learning_rate": 1.5785855600370777e-05,
      "loss": 0.1905,
      "step": 16480
    },
    {
      "epoch": 0.5094655863565796,
      "grad_norm": 0.09767106159810406,
      "learning_rate": 1.5770550401094394e-05,
      "loss": 0.1831,
      "step": 16490
    },
    {
      "epoch": 0.5097745406236242,
      "grad_norm": 0.09136753456551168,
      "learning_rate": 1.5755244397427362e-05,
      "loss": 0.1931,
      "step": 16500
    },
    {
      "epoch": 0.5100834948906688,
      "grad_norm": 0.07932070523632545,
      "learning_rate": 1.5739937605347877e-05,
      "loss": 0.1879,
      "step": 16510
    },
    {
      "epoch": 0.5103924491577134,
      "grad_norm": 0.08653253033055035,
      "learning_rate": 1.5724630040834958e-05,
      "loss": 0.1905,
      "step": 16520
    },
    {
      "epoch": 0.510701403424758,
      "grad_norm": 0.09940768000206279,
      "learning_rate": 1.5709321719868436e-05,
      "loss": 0.1954,
      "step": 16530
    },
    {
      "epoch": 0.5110103576918027,
      "grad_norm": 0.07755635560688848,
      "learning_rate": 1.5694012658428924e-05,
      "loss": 0.187,
      "step": 16540
    },
    {
      "epoch": 0.5113193119588473,
      "grad_norm": 0.10369840113350164,
      "learning_rate": 1.567870287249781e-05,
      "loss": 0.199,
      "step": 16550
    },
    {
      "epoch": 0.5116282662258919,
      "grad_norm": 0.0858597212972308,
      "learning_rate": 1.566339237805724e-05,
      "loss": 0.1968,
      "step": 16560
    },
    {
      "epoch": 0.5119372204929366,
      "grad_norm": 0.08688322729501115,
      "learning_rate": 1.5648081191090103e-05,
      "loss": 0.1933,
      "step": 16570
    },
    {
      "epoch": 0.5122461747599811,
      "grad_norm": 0.08073024960628804,
      "learning_rate": 1.5632769327580003e-05,
      "loss": 0.1821,
      "step": 16580
    },
    {
      "epoch": 0.5125551290270258,
      "grad_norm": 0.08278731832124943,
      "learning_rate": 1.5617456803511256e-05,
      "loss": 0.1807,
      "step": 16590
    },
    {
      "epoch": 0.5128640832940704,
      "grad_norm": 0.07728647706716552,
      "learning_rate": 1.5602143634868865e-05,
      "loss": 0.1837,
      "step": 16600
    },
    {
      "epoch": 0.513173037561115,
      "grad_norm": 0.07327476351709011,
      "learning_rate": 1.5586829837638512e-05,
      "loss": 0.1889,
      "step": 16610
    },
    {
      "epoch": 0.5134819918281597,
      "grad_norm": 0.08987695047246985,
      "learning_rate": 1.557151542780652e-05,
      "loss": 0.1863,
      "step": 16620
    },
    {
      "epoch": 0.5137909460952043,
      "grad_norm": 0.08638883049527403,
      "learning_rate": 1.5556200421359875e-05,
      "loss": 0.1883,
      "step": 16630
    },
    {
      "epoch": 0.5140999003622488,
      "grad_norm": 0.08866244801088069,
      "learning_rate": 1.5540884834286157e-05,
      "loss": 0.1858,
      "step": 16640
    },
    {
      "epoch": 0.5144088546292935,
      "grad_norm": 0.09021040523108524,
      "learning_rate": 1.5525568682573574e-05,
      "loss": 0.1883,
      "step": 16650
    },
    {
      "epoch": 0.5147178088963381,
      "grad_norm": 0.09170773703557734,
      "learning_rate": 1.551025198221092e-05,
      "loss": 0.1897,
      "step": 16660
    },
    {
      "epoch": 0.5150267631633827,
      "grad_norm": 0.08009894945135204,
      "learning_rate": 1.549493474918756e-05,
      "loss": 0.1866,
      "step": 16670
    },
    {
      "epoch": 0.5153357174304274,
      "grad_norm": 0.07829703404777481,
      "learning_rate": 1.5479616999493403e-05,
      "loss": 0.1868,
      "step": 16680
    },
    {
      "epoch": 0.515644671697472,
      "grad_norm": 0.0892501958293745,
      "learning_rate": 1.546429874911892e-05,
      "loss": 0.1859,
      "step": 16690
    },
    {
      "epoch": 0.5159536259645165,
      "grad_norm": 0.07684214678739372,
      "learning_rate": 1.5448980014055088e-05,
      "loss": 0.1902,
      "step": 16700
    },
    {
      "epoch": 0.5162625802315612,
      "grad_norm": 0.11422270533105577,
      "learning_rate": 1.5433660810293393e-05,
      "loss": 0.1839,
      "step": 16710
    },
    {
      "epoch": 0.5165715344986058,
      "grad_norm": 0.09663701450094417,
      "learning_rate": 1.5418341153825815e-05,
      "loss": 0.1903,
      "step": 16720
    },
    {
      "epoch": 0.5168804887656505,
      "grad_norm": 0.1188369578916549,
      "learning_rate": 1.54030210606448e-05,
      "loss": 0.1826,
      "step": 16730
    },
    {
      "epoch": 0.5171894430326951,
      "grad_norm": 0.1048352198299216,
      "learning_rate": 1.538770054674326e-05,
      "loss": 0.1835,
      "step": 16740
    },
    {
      "epoch": 0.5174983972997397,
      "grad_norm": 0.10024792665403055,
      "learning_rate": 1.537237962811453e-05,
      "loss": 0.181,
      "step": 16750
    },
    {
      "epoch": 0.5178073515667844,
      "grad_norm": 0.08165890938174093,
      "learning_rate": 1.535705832075238e-05,
      "loss": 0.2021,
      "step": 16760
    },
    {
      "epoch": 0.518116305833829,
      "grad_norm": 0.08070950685872727,
      "learning_rate": 1.5341736640650996e-05,
      "loss": 0.187,
      "step": 16770
    },
    {
      "epoch": 0.5184252601008735,
      "grad_norm": 0.08983897419723173,
      "learning_rate": 1.532641460380492e-05,
      "loss": 0.1854,
      "step": 16780
    },
    {
      "epoch": 0.5187342143679182,
      "grad_norm": 0.08956672505699013,
      "learning_rate": 1.5311092226209095e-05,
      "loss": 0.1915,
      "step": 16790
    },
    {
      "epoch": 0.5190431686349628,
      "grad_norm": 0.10233131738545687,
      "learning_rate": 1.5295769523858814e-05,
      "loss": 0.1926,
      "step": 16800
    },
    {
      "epoch": 0.5193521229020074,
      "grad_norm": 0.09231501611292307,
      "learning_rate": 1.5280446512749707e-05,
      "loss": 0.1875,
      "step": 16810
    },
    {
      "epoch": 0.5196610771690521,
      "grad_norm": 0.09091437555359963,
      "learning_rate": 1.5265123208877722e-05,
      "loss": 0.1834,
      "step": 16820
    },
    {
      "epoch": 0.5199700314360967,
      "grad_norm": 0.08961319032273445,
      "learning_rate": 1.5249799628239112e-05,
      "loss": 0.1886,
      "step": 16830
    },
    {
      "epoch": 0.5202789857031412,
      "grad_norm": 0.08716056858195882,
      "learning_rate": 1.5234475786830432e-05,
      "loss": 0.1854,
      "step": 16840
    },
    {
      "epoch": 0.5205879399701859,
      "grad_norm": 0.0775112634053927,
      "learning_rate": 1.5219151700648499e-05,
      "loss": 0.1908,
      "step": 16850
    },
    {
      "epoch": 0.5208968942372305,
      "grad_norm": 0.07994281723800888,
      "learning_rate": 1.5203827385690378e-05,
      "loss": 0.1842,
      "step": 16860
    },
    {
      "epoch": 0.5212058485042752,
      "grad_norm": 0.11038659043417026,
      "learning_rate": 1.518850285795339e-05,
      "loss": 0.1848,
      "step": 16870
    },
    {
      "epoch": 0.5215148027713198,
      "grad_norm": 0.09098567725841361,
      "learning_rate": 1.5173178133435067e-05,
      "loss": 0.183,
      "step": 16880
    },
    {
      "epoch": 0.5218237570383644,
      "grad_norm": 0.08853172387115174,
      "learning_rate": 1.515785322813315e-05,
      "loss": 0.1869,
      "step": 16890
    },
    {
      "epoch": 0.5221327113054091,
      "grad_norm": 0.0919017222927224,
      "learning_rate": 1.5142528158045563e-05,
      "loss": 0.1951,
      "step": 16900
    },
    {
      "epoch": 0.5224416655724536,
      "grad_norm": 0.09828075180749284,
      "learning_rate": 1.5127202939170418e-05,
      "loss": 0.1826,
      "step": 16910
    },
    {
      "epoch": 0.5227506198394982,
      "grad_norm": 0.1090552992323409,
      "learning_rate": 1.511187758750596e-05,
      "loss": 0.1856,
      "step": 16920
    },
    {
      "epoch": 0.5230595741065429,
      "grad_norm": 0.08778878304175707,
      "learning_rate": 1.5096552119050588e-05,
      "loss": 0.1877,
      "step": 16930
    },
    {
      "epoch": 0.5233685283735875,
      "grad_norm": 0.08228476322473942,
      "learning_rate": 1.5081226549802823e-05,
      "loss": 0.185,
      "step": 16940
    },
    {
      "epoch": 0.5236774826406321,
      "grad_norm": 0.08286146340483935,
      "learning_rate": 1.5065900895761281e-05,
      "loss": 0.1867,
      "step": 16950
    },
    {
      "epoch": 0.5239864369076768,
      "grad_norm": 0.08397935696767478,
      "learning_rate": 1.5050575172924676e-05,
      "loss": 0.1885,
      "step": 16960
    },
    {
      "epoch": 0.5242953911747213,
      "grad_norm": 0.11852410287882821,
      "learning_rate": 1.5035249397291792e-05,
      "loss": 0.192,
      "step": 16970
    },
    {
      "epoch": 0.5246043454417659,
      "grad_norm": 0.09165341600222472,
      "learning_rate": 1.5019923584861459e-05,
      "loss": 0.1811,
      "step": 16980
    },
    {
      "epoch": 0.5249132997088106,
      "grad_norm": 0.09532246851600214,
      "learning_rate": 1.5004597751632555e-05,
      "loss": 0.1913,
      "step": 16990
    },
    {
      "epoch": 0.5252222539758552,
      "grad_norm": 0.10521407904617962,
      "learning_rate": 1.4989271913603985e-05,
      "loss": 0.1855,
      "step": 17000
    },
    {
      "epoch": 0.5255312082428999,
      "grad_norm": 0.08637722581336506,
      "learning_rate": 1.4973946086774648e-05,
      "loss": 0.1805,
      "step": 17010
    },
    {
      "epoch": 0.5258401625099445,
      "grad_norm": 0.0928785275306785,
      "learning_rate": 1.4958620287143426e-05,
      "loss": 0.1895,
      "step": 17020
    },
    {
      "epoch": 0.5261491167769891,
      "grad_norm": 0.09415359707936143,
      "learning_rate": 1.4943294530709196e-05,
      "loss": 0.1818,
      "step": 17030
    },
    {
      "epoch": 0.5264580710440337,
      "grad_norm": 0.09163024023045135,
      "learning_rate": 1.4927968833470763e-05,
      "loss": 0.1866,
      "step": 17040
    },
    {
      "epoch": 0.5267670253110783,
      "grad_norm": 0.10236905094453971,
      "learning_rate": 1.4912643211426894e-05,
      "loss": 0.1843,
      "step": 17050
    },
    {
      "epoch": 0.5270759795781229,
      "grad_norm": 0.10445863766997315,
      "learning_rate": 1.4897317680576255e-05,
      "loss": 0.191,
      "step": 17060
    },
    {
      "epoch": 0.5273849338451676,
      "grad_norm": 0.0897013205674567,
      "learning_rate": 1.4881992256917435e-05,
      "loss": 0.179,
      "step": 17070
    },
    {
      "epoch": 0.5276938881122122,
      "grad_norm": 0.08051453323056623,
      "learning_rate": 1.4866666956448902e-05,
      "loss": 0.1845,
      "step": 17080
    },
    {
      "epoch": 0.5280028423792568,
      "grad_norm": 0.0960572970135507,
      "learning_rate": 1.4851341795169001e-05,
      "loss": 0.1828,
      "step": 17090
    },
    {
      "epoch": 0.5283117966463015,
      "grad_norm": 0.09099435178945896,
      "learning_rate": 1.4836016789075918e-05,
      "loss": 0.1929,
      "step": 17100
    },
    {
      "epoch": 0.528620750913346,
      "grad_norm": 0.08925916899274329,
      "learning_rate": 1.48206919541677e-05,
      "loss": 0.1881,
      "step": 17110
    },
    {
      "epoch": 0.5289297051803906,
      "grad_norm": 0.09593747889871479,
      "learning_rate": 1.4805367306442196e-05,
      "loss": 0.1888,
      "step": 17120
    },
    {
      "epoch": 0.5292386594474353,
      "grad_norm": 0.07405438597813077,
      "learning_rate": 1.4790042861897058e-05,
      "loss": 0.192,
      "step": 17130
    },
    {
      "epoch": 0.5295476137144799,
      "grad_norm": 0.09421400762818595,
      "learning_rate": 1.4774718636529751e-05,
      "loss": 0.1884,
      "step": 17140
    },
    {
      "epoch": 0.5298565679815246,
      "grad_norm": 0.09931827531418859,
      "learning_rate": 1.4759394646337477e-05,
      "loss": 0.187,
      "step": 17150
    },
    {
      "epoch": 0.5301655222485692,
      "grad_norm": 0.09243735280179094,
      "learning_rate": 1.4744070907317227e-05,
      "loss": 0.1852,
      "step": 17160
    },
    {
      "epoch": 0.5304744765156137,
      "grad_norm": 0.08297586570125884,
      "learning_rate": 1.4728747435465693e-05,
      "loss": 0.1862,
      "step": 17170
    },
    {
      "epoch": 0.5307834307826584,
      "grad_norm": 0.09711190720355078,
      "learning_rate": 1.4713424246779327e-05,
      "loss": 0.1838,
      "step": 17180
    },
    {
      "epoch": 0.531092385049703,
      "grad_norm": 0.09893471219976875,
      "learning_rate": 1.4698101357254254e-05,
      "loss": 0.1889,
      "step": 17190
    },
    {
      "epoch": 0.5314013393167476,
      "grad_norm": 0.1009962099184244,
      "learning_rate": 1.4682778782886301e-05,
      "loss": 0.1835,
      "step": 17200
    },
    {
      "epoch": 0.5317102935837923,
      "grad_norm": 0.08314577919257489,
      "learning_rate": 1.466745653967096e-05,
      "loss": 0.1815,
      "step": 17210
    },
    {
      "epoch": 0.5320192478508369,
      "grad_norm": 0.07981247358516776,
      "learning_rate": 1.465213464360339e-05,
      "loss": 0.1866,
      "step": 17220
    },
    {
      "epoch": 0.5323282021178815,
      "grad_norm": 0.0948002124448693,
      "learning_rate": 1.4636813110678367e-05,
      "loss": 0.182,
      "step": 17230
    },
    {
      "epoch": 0.5326371563849261,
      "grad_norm": 0.0823123813200111,
      "learning_rate": 1.4621491956890313e-05,
      "loss": 0.187,
      "step": 17240
    },
    {
      "epoch": 0.5329461106519707,
      "grad_norm": 0.10001350183120634,
      "learning_rate": 1.4606171198233225e-05,
      "loss": 0.1824,
      "step": 17250
    },
    {
      "epoch": 0.5332550649190153,
      "grad_norm": 0.0799570894990248,
      "learning_rate": 1.459085085070072e-05,
      "loss": 0.1854,
      "step": 17260
    },
    {
      "epoch": 0.53356401918606,
      "grad_norm": 0.08894190948983392,
      "learning_rate": 1.4575530930285952e-05,
      "loss": 0.1944,
      "step": 17270
    },
    {
      "epoch": 0.5338729734531046,
      "grad_norm": 0.07546407103998903,
      "learning_rate": 1.4560211452981662e-05,
      "loss": 0.2028,
      "step": 17280
    },
    {
      "epoch": 0.5341819277201493,
      "grad_norm": 0.10185925884645297,
      "learning_rate": 1.4544892434780101e-05,
      "loss": 0.1829,
      "step": 17290
    },
    {
      "epoch": 0.5344908819871939,
      "grad_norm": 0.10005081945027251,
      "learning_rate": 1.4529573891673057e-05,
      "loss": 0.1793,
      "step": 17300
    },
    {
      "epoch": 0.5347998362542384,
      "grad_norm": 0.08457714031441274,
      "learning_rate": 1.4514255839651815e-05,
      "loss": 0.1797,
      "step": 17310
    },
    {
      "epoch": 0.5351087905212831,
      "grad_norm": 0.08018724714628987,
      "learning_rate": 1.4498938294707157e-05,
      "loss": 0.1938,
      "step": 17320
    },
    {
      "epoch": 0.5354177447883277,
      "grad_norm": 0.07996280965923389,
      "learning_rate": 1.4483621272829314e-05,
      "loss": 0.1912,
      "step": 17330
    },
    {
      "epoch": 0.5357266990553723,
      "grad_norm": 0.09015109495118928,
      "learning_rate": 1.4468304790008002e-05,
      "loss": 0.1853,
      "step": 17340
    },
    {
      "epoch": 0.536035653322417,
      "grad_norm": 0.0802749764196976,
      "learning_rate": 1.4452988862232348e-05,
      "loss": 0.1818,
      "step": 17350
    },
    {
      "epoch": 0.5363446075894616,
      "grad_norm": 0.08152069249656302,
      "learning_rate": 1.4437673505490904e-05,
      "loss": 0.1805,
      "step": 17360
    },
    {
      "epoch": 0.5366535618565061,
      "grad_norm": 0.08491495530511571,
      "learning_rate": 1.4422358735771644e-05,
      "loss": 0.1848,
      "step": 17370
    },
    {
      "epoch": 0.5369625161235508,
      "grad_norm": 0.09352929532539385,
      "learning_rate": 1.4407044569061904e-05,
      "loss": 0.1807,
      "step": 17380
    },
    {
      "epoch": 0.5372714703905954,
      "grad_norm": 0.0712002076927719,
      "learning_rate": 1.4391731021348411e-05,
      "loss": 0.1836,
      "step": 17390
    },
    {
      "epoch": 0.53758042465764,
      "grad_norm": 0.07654844660693336,
      "learning_rate": 1.4376418108617233e-05,
      "loss": 0.1922,
      "step": 17400
    },
    {
      "epoch": 0.5378893789246847,
      "grad_norm": 0.10044903255580094,
      "learning_rate": 1.4361105846853785e-05,
      "loss": 0.188,
      "step": 17410
    },
    {
      "epoch": 0.5381983331917293,
      "grad_norm": 0.09616695490663842,
      "learning_rate": 1.4345794252042793e-05,
      "loss": 0.1837,
      "step": 17420
    },
    {
      "epoch": 0.538507287458774,
      "grad_norm": 0.09121957303038127,
      "learning_rate": 1.4330483340168294e-05,
      "loss": 0.1818,
      "step": 17430
    },
    {
      "epoch": 0.5388162417258185,
      "grad_norm": 0.08569007567192438,
      "learning_rate": 1.4315173127213597e-05,
      "loss": 0.1849,
      "step": 17440
    },
    {
      "epoch": 0.5391251959928631,
      "grad_norm": 0.12991475621721707,
      "learning_rate": 1.4299863629161312e-05,
      "loss": 0.1864,
      "step": 17450
    },
    {
      "epoch": 0.5394341502599078,
      "grad_norm": 0.09185756381582284,
      "learning_rate": 1.4284554861993266e-05,
      "loss": 0.1836,
      "step": 17460
    },
    {
      "epoch": 0.5397431045269524,
      "grad_norm": 0.08448691611438457,
      "learning_rate": 1.426924684169056e-05,
      "loss": 0.1825,
      "step": 17470
    },
    {
      "epoch": 0.540052058793997,
      "grad_norm": 0.08242456970627907,
      "learning_rate": 1.4253939584233475e-05,
      "loss": 0.1817,
      "step": 17480
    },
    {
      "epoch": 0.5403610130610417,
      "grad_norm": 0.0988976302001369,
      "learning_rate": 1.4238633105601537e-05,
      "loss": 0.1852,
      "step": 17490
    },
    {
      "epoch": 0.5406699673280863,
      "grad_norm": 0.07820452586212755,
      "learning_rate": 1.4223327421773422e-05,
      "loss": 0.1888,
      "step": 17500
    },
    {
      "epoch": 0.5409789215951308,
      "grad_norm": 0.1046724761210211,
      "learning_rate": 1.420802254872701e-05,
      "loss": 0.1833,
      "step": 17510
    },
    {
      "epoch": 0.5412878758621755,
      "grad_norm": 0.10141302621041458,
      "learning_rate": 1.4192718502439305e-05,
      "loss": 0.1858,
      "step": 17520
    },
    {
      "epoch": 0.5415968301292201,
      "grad_norm": 0.0887367435383199,
      "learning_rate": 1.4177415298886468e-05,
      "loss": 0.1831,
      "step": 17530
    },
    {
      "epoch": 0.5419057843962647,
      "grad_norm": 0.09632885655951946,
      "learning_rate": 1.4162112954043763e-05,
      "loss": 0.1844,
      "step": 17540
    },
    {
      "epoch": 0.5422147386633094,
      "grad_norm": 0.07623364132872182,
      "learning_rate": 1.4146811483885582e-05,
      "loss": 0.1924,
      "step": 17550
    },
    {
      "epoch": 0.542523692930354,
      "grad_norm": 0.09477616654210215,
      "learning_rate": 1.4131510904385376e-05,
      "loss": 0.1926,
      "step": 17560
    },
    {
      "epoch": 0.5428326471973987,
      "grad_norm": 0.08307270244024169,
      "learning_rate": 1.4116211231515693e-05,
      "loss": 0.191,
      "step": 17570
    },
    {
      "epoch": 0.5431416014644432,
      "grad_norm": 0.0879987139114687,
      "learning_rate": 1.4100912481248114e-05,
      "loss": 0.1846,
      "step": 17580
    },
    {
      "epoch": 0.5434505557314878,
      "grad_norm": 0.32627772183764964,
      "learning_rate": 1.4085614669553257e-05,
      "loss": 0.1856,
      "step": 17590
    },
    {
      "epoch": 0.5437595099985325,
      "grad_norm": 0.08654146767886002,
      "learning_rate": 1.4070317812400783e-05,
      "loss": 0.1847,
      "step": 17600
    },
    {
      "epoch": 0.5440684642655771,
      "grad_norm": 0.09291518300851469,
      "learning_rate": 1.4055021925759326e-05,
      "loss": 0.1897,
      "step": 17610
    },
    {
      "epoch": 0.5443774185326217,
      "grad_norm": 0.0830066940923024,
      "learning_rate": 1.4039727025596537e-05,
      "loss": 0.1841,
      "step": 17620
    },
    {
      "epoch": 0.5446863727996664,
      "grad_norm": 0.37472320467808995,
      "learning_rate": 1.402443312787901e-05,
      "loss": 0.1833,
      "step": 17630
    },
    {
      "epoch": 0.5449953270667109,
      "grad_norm": 0.17567192167695006,
      "learning_rate": 1.4009140248572307e-05,
      "loss": 0.1909,
      "step": 17640
    },
    {
      "epoch": 0.5453042813337555,
      "grad_norm": 0.0792282290952821,
      "learning_rate": 1.3993848403640926e-05,
      "loss": 0.196,
      "step": 17650
    },
    {
      "epoch": 0.5456132356008002,
      "grad_norm": 0.08045439612250929,
      "learning_rate": 1.3978557609048288e-05,
      "loss": 0.1924,
      "step": 17660
    },
    {
      "epoch": 0.5459221898678448,
      "grad_norm": 0.08474439759336545,
      "learning_rate": 1.3963267880756702e-05,
      "loss": 0.1899,
      "step": 17670
    },
    {
      "epoch": 0.5462311441348894,
      "grad_norm": 0.08076183615781608,
      "learning_rate": 1.3947979234727389e-05,
      "loss": 0.1857,
      "step": 17680
    },
    {
      "epoch": 0.5465400984019341,
      "grad_norm": 0.10006441860898663,
      "learning_rate": 1.393269168692041e-05,
      "loss": 0.1825,
      "step": 17690
    },
    {
      "epoch": 0.5468490526689787,
      "grad_norm": 0.09497652449343258,
      "learning_rate": 1.3917405253294712e-05,
      "loss": 0.186,
      "step": 17700
    },
    {
      "epoch": 0.5471580069360233,
      "grad_norm": 0.07872838283830029,
      "learning_rate": 1.390211994980805e-05,
      "loss": 0.1817,
      "step": 17710
    },
    {
      "epoch": 0.5474669612030679,
      "grad_norm": 0.0854526101511236,
      "learning_rate": 1.3886835792417021e-05,
      "loss": 0.1871,
      "step": 17720
    },
    {
      "epoch": 0.5477759154701125,
      "grad_norm": 0.08515358675952887,
      "learning_rate": 1.387155279707701e-05,
      "loss": 0.1804,
      "step": 17730
    },
    {
      "epoch": 0.5480848697371572,
      "grad_norm": 0.0848473404926865,
      "learning_rate": 1.3856270979742203e-05,
      "loss": 0.1902,
      "step": 17740
    },
    {
      "epoch": 0.5483938240042018,
      "grad_norm": 0.08507319549854123,
      "learning_rate": 1.384099035636554e-05,
      "loss": 0.1905,
      "step": 17750
    },
    {
      "epoch": 0.5487027782712464,
      "grad_norm": 0.09140757552102367,
      "learning_rate": 1.382571094289873e-05,
      "loss": 0.1845,
      "step": 17760
    },
    {
      "epoch": 0.549011732538291,
      "grad_norm": 0.09854616195628617,
      "learning_rate": 1.3810432755292205e-05,
      "loss": 0.1864,
      "step": 17770
    },
    {
      "epoch": 0.5493206868053356,
      "grad_norm": 0.07821144682589172,
      "learning_rate": 1.3795155809495132e-05,
      "loss": 0.1796,
      "step": 17780
    },
    {
      "epoch": 0.5496296410723802,
      "grad_norm": 0.09636875779609737,
      "learning_rate": 1.3779880121455367e-05,
      "loss": 0.1874,
      "step": 17790
    },
    {
      "epoch": 0.5499385953394249,
      "grad_norm": 0.10398916307274209,
      "learning_rate": 1.376460570711947e-05,
      "loss": 0.1904,
      "step": 17800
    },
    {
      "epoch": 0.5502475496064695,
      "grad_norm": 0.0775535615915769,
      "learning_rate": 1.3749332582432653e-05,
      "loss": 0.1904,
      "step": 17810
    },
    {
      "epoch": 0.5505565038735141,
      "grad_norm": 0.09082598461858564,
      "learning_rate": 1.3734060763338789e-05,
      "loss": 0.195,
      "step": 17820
    },
    {
      "epoch": 0.5508654581405588,
      "grad_norm": 0.07848013156671774,
      "learning_rate": 1.3718790265780398e-05,
      "loss": 0.1854,
      "step": 17830
    },
    {
      "epoch": 0.5511744124076033,
      "grad_norm": 0.08192677840117994,
      "learning_rate": 1.3703521105698601e-05,
      "loss": 0.1837,
      "step": 17840
    },
    {
      "epoch": 0.551483366674648,
      "grad_norm": 0.08848765115082168,
      "learning_rate": 1.3688253299033146e-05,
      "loss": 0.1805,
      "step": 17850
    },
    {
      "epoch": 0.5517923209416926,
      "grad_norm": 0.09920079662348166,
      "learning_rate": 1.3672986861722348e-05,
      "loss": 0.1811,
      "step": 17860
    },
    {
      "epoch": 0.5521012752087372,
      "grad_norm": 0.10624153800518651,
      "learning_rate": 1.3657721809703098e-05,
      "loss": 0.1893,
      "step": 17870
    },
    {
      "epoch": 0.5524102294757819,
      "grad_norm": 0.08227130144677,
      "learning_rate": 1.3642458158910847e-05,
      "loss": 0.1835,
      "step": 17880
    },
    {
      "epoch": 0.5527191837428265,
      "grad_norm": 0.08570427994284543,
      "learning_rate": 1.3627195925279581e-05,
      "loss": 0.1847,
      "step": 17890
    },
    {
      "epoch": 0.553028138009871,
      "grad_norm": 0.10391660517592205,
      "learning_rate": 1.3611935124741798e-05,
      "loss": 0.1865,
      "step": 17900
    },
    {
      "epoch": 0.5533370922769157,
      "grad_norm": 0.09700018352924356,
      "learning_rate": 1.3596675773228515e-05,
      "loss": 0.1831,
      "step": 17910
    },
    {
      "epoch": 0.5536460465439603,
      "grad_norm": 0.09473579301432845,
      "learning_rate": 1.358141788666922e-05,
      "loss": 0.1841,
      "step": 17920
    },
    {
      "epoch": 0.5539550008110049,
      "grad_norm": 0.08051417888890819,
      "learning_rate": 1.3566161480991892e-05,
      "loss": 0.1867,
      "step": 17930
    },
    {
      "epoch": 0.5542639550780496,
      "grad_norm": 0.08367151773433916,
      "learning_rate": 1.3550906572122935e-05,
      "loss": 0.1817,
      "step": 17940
    },
    {
      "epoch": 0.5545729093450942,
      "grad_norm": 0.09016611338068892,
      "learning_rate": 1.3535653175987224e-05,
      "loss": 0.1899,
      "step": 17950
    },
    {
      "epoch": 0.5548818636121388,
      "grad_norm": 0.07994393203403673,
      "learning_rate": 1.3520401308508029e-05,
      "loss": 0.1848,
      "step": 17960
    },
    {
      "epoch": 0.5551908178791835,
      "grad_norm": 0.09898322055199392,
      "learning_rate": 1.3505150985607035e-05,
      "loss": 0.1882,
      "step": 17970
    },
    {
      "epoch": 0.555499772146228,
      "grad_norm": 0.0832540759484844,
      "learning_rate": 1.3489902223204313e-05,
      "loss": 0.1824,
      "step": 17980
    },
    {
      "epoch": 0.5558087264132727,
      "grad_norm": 0.07501050463966243,
      "learning_rate": 1.347465503721831e-05,
      "loss": 0.1873,
      "step": 17990
    },
    {
      "epoch": 0.5561176806803173,
      "grad_norm": 0.084183164690891,
      "learning_rate": 1.345940944356581e-05,
      "loss": 0.1882,
      "step": 18000
    },
    {
      "epoch": 0.5564266349473619,
      "grad_norm": 0.09927738611382306,
      "learning_rate": 1.3444165458161961e-05,
      "loss": 0.1875,
      "step": 18010
    },
    {
      "epoch": 0.5567355892144066,
      "grad_norm": 0.10254120092547032,
      "learning_rate": 1.3428923096920207e-05,
      "loss": 0.1865,
      "step": 18020
    },
    {
      "epoch": 0.5570445434814512,
      "grad_norm": 0.08292525689385516,
      "learning_rate": 1.341368237575232e-05,
      "loss": 0.1881,
      "step": 18030
    },
    {
      "epoch": 0.5573534977484957,
      "grad_norm": 0.08145256947523416,
      "learning_rate": 1.339844331056834e-05,
      "loss": 0.1809,
      "step": 18040
    },
    {
      "epoch": 0.5576624520155404,
      "grad_norm": 0.12036097904889985,
      "learning_rate": 1.3383205917276581e-05,
      "loss": 0.1853,
      "step": 18050
    },
    {
      "epoch": 0.557971406282585,
      "grad_norm": 0.0974876488142113,
      "learning_rate": 1.3367970211783629e-05,
      "loss": 0.196,
      "step": 18060
    },
    {
      "epoch": 0.5582803605496296,
      "grad_norm": 0.08900797031577047,
      "learning_rate": 1.335273620999428e-05,
      "loss": 0.1859,
      "step": 18070
    },
    {
      "epoch": 0.5585893148166743,
      "grad_norm": 0.08376806481567103,
      "learning_rate": 1.3337503927811586e-05,
      "loss": 0.1876,
      "step": 18080
    },
    {
      "epoch": 0.5588982690837189,
      "grad_norm": 0.09983652945209968,
      "learning_rate": 1.332227338113677e-05,
      "loss": 0.1997,
      "step": 18090
    },
    {
      "epoch": 0.5592072233507634,
      "grad_norm": 0.08272073156927462,
      "learning_rate": 1.3307044585869264e-05,
      "loss": 0.1836,
      "step": 18100
    },
    {
      "epoch": 0.5595161776178081,
      "grad_norm": 0.08019848440170073,
      "learning_rate": 1.3291817557906665e-05,
      "loss": 0.1902,
      "step": 18110
    },
    {
      "epoch": 0.5598251318848527,
      "grad_norm": 0.07660768377815788,
      "learning_rate": 1.3276592313144732e-05,
      "loss": 0.1864,
      "step": 18120
    },
    {
      "epoch": 0.5601340861518974,
      "grad_norm": 0.08427816260084622,
      "learning_rate": 1.3261368867477343e-05,
      "loss": 0.192,
      "step": 18130
    },
    {
      "epoch": 0.560443040418942,
      "grad_norm": 0.09165925543699757,
      "learning_rate": 1.3246147236796528e-05,
      "loss": 0.187,
      "step": 18140
    },
    {
      "epoch": 0.5607519946859866,
      "grad_norm": 0.08533068321333423,
      "learning_rate": 1.323092743699239e-05,
      "loss": 0.1832,
      "step": 18150
    },
    {
      "epoch": 0.5610609489530313,
      "grad_norm": 0.08163197125245189,
      "learning_rate": 1.321570948395315e-05,
      "loss": 0.1824,
      "step": 18160
    },
    {
      "epoch": 0.5613699032200758,
      "grad_norm": 0.08632504950505282,
      "learning_rate": 1.3200493393565074e-05,
      "loss": 0.1879,
      "step": 18170
    },
    {
      "epoch": 0.5616788574871204,
      "grad_norm": 0.07476858547105282,
      "learning_rate": 1.3185279181712513e-05,
      "loss": 0.1834,
      "step": 18180
    },
    {
      "epoch": 0.5619878117541651,
      "grad_norm": 0.07541841695829907,
      "learning_rate": 1.317006686427783e-05,
      "loss": 0.181,
      "step": 18190
    },
    {
      "epoch": 0.5622967660212097,
      "grad_norm": 0.1077618205576265,
      "learning_rate": 1.3154856457141423e-05,
      "loss": 0.1922,
      "step": 18200
    },
    {
      "epoch": 0.5626057202882543,
      "grad_norm": 0.0818609664964366,
      "learning_rate": 1.3139647976181696e-05,
      "loss": 0.1807,
      "step": 18210
    },
    {
      "epoch": 0.562914674555299,
      "grad_norm": 0.0908286567884653,
      "learning_rate": 1.3124441437275045e-05,
      "loss": 0.1835,
      "step": 18220
    },
    {
      "epoch": 0.5632236288223436,
      "grad_norm": 0.09575946100216455,
      "learning_rate": 1.310923685629582e-05,
      "loss": 0.1871,
      "step": 18230
    },
    {
      "epoch": 0.5635325830893881,
      "grad_norm": 0.08080778197175675,
      "learning_rate": 1.309403424911636e-05,
      "loss": 0.1866,
      "step": 18240
    },
    {
      "epoch": 0.5638415373564328,
      "grad_norm": 0.07256286544884057,
      "learning_rate": 1.3078833631606908e-05,
      "loss": 0.1837,
      "step": 18250
    },
    {
      "epoch": 0.5641504916234774,
      "grad_norm": 0.08937246492506239,
      "learning_rate": 1.3063635019635662e-05,
      "loss": 0.1841,
      "step": 18260
    },
    {
      "epoch": 0.5644594458905221,
      "grad_norm": 0.09723740333864475,
      "learning_rate": 1.3048438429068703e-05,
      "loss": 0.1826,
      "step": 18270
    },
    {
      "epoch": 0.5647684001575667,
      "grad_norm": 0.12124340141480298,
      "learning_rate": 1.3033243875770002e-05,
      "loss": 0.1963,
      "step": 18280
    },
    {
      "epoch": 0.5650773544246113,
      "grad_norm": 0.08216877458298519,
      "learning_rate": 1.3018051375601428e-05,
      "loss": 0.1852,
      "step": 18290
    },
    {
      "epoch": 0.565386308691656,
      "grad_norm": 0.08488433886810821,
      "learning_rate": 1.3002860944422676e-05,
      "loss": 0.179,
      "step": 18300
    },
    {
      "epoch": 0.5656952629587005,
      "grad_norm": 0.09152647645936096,
      "learning_rate": 1.2987672598091307e-05,
      "loss": 0.1844,
      "step": 18310
    },
    {
      "epoch": 0.5660042172257451,
      "grad_norm": 0.0902168511301893,
      "learning_rate": 1.2972486352462681e-05,
      "loss": 0.1856,
      "step": 18320
    },
    {
      "epoch": 0.5663131714927898,
      "grad_norm": 0.10166543212631995,
      "learning_rate": 1.295730222338999e-05,
      "loss": 0.1888,
      "step": 18330
    },
    {
      "epoch": 0.5666221257598344,
      "grad_norm": 0.08335037816166078,
      "learning_rate": 1.2942120226724193e-05,
      "loss": 0.1923,
      "step": 18340
    },
    {
      "epoch": 0.566931080026879,
      "grad_norm": 0.07675305074037349,
      "learning_rate": 1.2926940378314045e-05,
      "loss": 0.1846,
      "step": 18350
    },
    {
      "epoch": 0.5672400342939237,
      "grad_norm": 0.08900068087168743,
      "learning_rate": 1.2911762694006035e-05,
      "loss": 0.1833,
      "step": 18360
    },
    {
      "epoch": 0.5675489885609682,
      "grad_norm": 0.13945464616694303,
      "learning_rate": 1.289658718964442e-05,
      "loss": 0.1873,
      "step": 18370
    },
    {
      "epoch": 0.5678579428280128,
      "grad_norm": 0.08486324017069379,
      "learning_rate": 1.2881413881071151e-05,
      "loss": 0.1851,
      "step": 18380
    },
    {
      "epoch": 0.5681668970950575,
      "grad_norm": 0.09162914427517899,
      "learning_rate": 1.2866242784125916e-05,
      "loss": 0.1882,
      "step": 18390
    },
    {
      "epoch": 0.5684758513621021,
      "grad_norm": 0.07513189668788745,
      "learning_rate": 1.2851073914646072e-05,
      "loss": 0.1877,
      "step": 18400
    },
    {
      "epoch": 0.5687848056291468,
      "grad_norm": 0.10635056481968717,
      "learning_rate": 1.2835907288466668e-05,
      "loss": 0.1859,
      "step": 18410
    },
    {
      "epoch": 0.5690937598961914,
      "grad_norm": 0.1084024767487765,
      "learning_rate": 1.2820742921420394e-05,
      "loss": 0.1869,
      "step": 18420
    },
    {
      "epoch": 0.569402714163236,
      "grad_norm": 0.09261344954353902,
      "learning_rate": 1.2805580829337596e-05,
      "loss": 0.1911,
      "step": 18430
    },
    {
      "epoch": 0.5697116684302806,
      "grad_norm": 0.10332341591055433,
      "learning_rate": 1.2790421028046234e-05,
      "loss": 0.1928,
      "step": 18440
    },
    {
      "epoch": 0.5700206226973252,
      "grad_norm": 0.09147366175987018,
      "learning_rate": 1.277526353337189e-05,
      "loss": 0.1815,
      "step": 18450
    },
    {
      "epoch": 0.5703295769643698,
      "grad_norm": 0.12406447856940347,
      "learning_rate": 1.276010836113772e-05,
      "loss": 0.2033,
      "step": 18460
    },
    {
      "epoch": 0.5706385312314145,
      "grad_norm": 0.08312351723744336,
      "learning_rate": 1.2744955527164474e-05,
      "loss": 0.1846,
      "step": 18470
    },
    {
      "epoch": 0.5709474854984591,
      "grad_norm": 0.09805344865224558,
      "learning_rate": 1.2729805047270447e-05,
      "loss": 0.1856,
      "step": 18480
    },
    {
      "epoch": 0.5712564397655037,
      "grad_norm": 0.09322922472078514,
      "learning_rate": 1.271465693727149e-05,
      "loss": 0.1864,
      "step": 18490
    },
    {
      "epoch": 0.5715653940325484,
      "grad_norm": 0.07423211075291146,
      "learning_rate": 1.2699511212980967e-05,
      "loss": 0.184,
      "step": 18500
    },
    {
      "epoch": 0.5718743482995929,
      "grad_norm": 0.073608149716638,
      "learning_rate": 1.2684367890209752e-05,
      "loss": 0.1891,
      "step": 18510
    },
    {
      "epoch": 0.5721833025666375,
      "grad_norm": 0.10314568671849617,
      "learning_rate": 1.2669226984766227e-05,
      "loss": 0.1849,
      "step": 18520
    },
    {
      "epoch": 0.5724922568336822,
      "grad_norm": 0.08692076340230744,
      "learning_rate": 1.2654088512456228e-05,
      "loss": 0.1836,
      "step": 18530
    },
    {
      "epoch": 0.5728012111007268,
      "grad_norm": 0.07738189014599374,
      "learning_rate": 1.2638952489083077e-05,
      "loss": 0.1829,
      "step": 18540
    },
    {
      "epoch": 0.5731101653677715,
      "grad_norm": 0.09949104059203269,
      "learning_rate": 1.2623818930447517e-05,
      "loss": 0.1824,
      "step": 18550
    },
    {
      "epoch": 0.5734191196348161,
      "grad_norm": 0.09055018055054191,
      "learning_rate": 1.2608687852347733e-05,
      "loss": 0.1852,
      "step": 18560
    },
    {
      "epoch": 0.5737280739018606,
      "grad_norm": 0.11175457125740802,
      "learning_rate": 1.2593559270579303e-05,
      "loss": 0.1844,
      "step": 18570
    },
    {
      "epoch": 0.5740370281689053,
      "grad_norm": 0.13301535459360475,
      "learning_rate": 1.2578433200935226e-05,
      "loss": 0.1881,
      "step": 18580
    },
    {
      "epoch": 0.5743459824359499,
      "grad_norm": 0.0899955346323926,
      "learning_rate": 1.2563309659205848e-05,
      "loss": 0.1831,
      "step": 18590
    },
    {
      "epoch": 0.5746549367029945,
      "grad_norm": 0.08306036513954007,
      "learning_rate": 1.2548188661178903e-05,
      "loss": 0.1781,
      "step": 18600
    },
    {
      "epoch": 0.5749638909700392,
      "grad_norm": 0.08011029389984016,
      "learning_rate": 1.2533070222639449e-05,
      "loss": 0.1897,
      "step": 18610
    },
    {
      "epoch": 0.5752728452370838,
      "grad_norm": 0.0846103975665748,
      "learning_rate": 1.2517954359369888e-05,
      "loss": 0.1932,
      "step": 18620
    },
    {
      "epoch": 0.5755817995041284,
      "grad_norm": 0.09252490726584991,
      "learning_rate": 1.250284108714992e-05,
      "loss": 0.1863,
      "step": 18630
    },
    {
      "epoch": 0.575890753771173,
      "grad_norm": 0.0829174294001005,
      "learning_rate": 1.2487730421756552e-05,
      "loss": 0.1902,
      "step": 18640
    },
    {
      "epoch": 0.5761997080382176,
      "grad_norm": 0.08778215776406699,
      "learning_rate": 1.2472622378964062e-05,
      "loss": 0.179,
      "step": 18650
    },
    {
      "epoch": 0.5765086623052622,
      "grad_norm": 0.09976640750723119,
      "learning_rate": 1.2457516974543992e-05,
      "loss": 0.1817,
      "step": 18660
    },
    {
      "epoch": 0.5768176165723069,
      "grad_norm": 0.38975443046706637,
      "learning_rate": 1.2442414224265125e-05,
      "loss": 0.1812,
      "step": 18670
    },
    {
      "epoch": 0.5771265708393515,
      "grad_norm": 0.08529668233447649,
      "learning_rate": 1.2427314143893492e-05,
      "loss": 0.1832,
      "step": 18680
    },
    {
      "epoch": 0.5774355251063962,
      "grad_norm": 0.19581723497176823,
      "learning_rate": 1.2412216749192307e-05,
      "loss": 0.1877,
      "step": 18690
    },
    {
      "epoch": 0.5777444793734408,
      "grad_norm": 0.09003964574893565,
      "learning_rate": 1.239712205592201e-05,
      "loss": 0.1873,
      "step": 18700
    },
    {
      "epoch": 0.5780534336404853,
      "grad_norm": 0.0758620885951063,
      "learning_rate": 1.2382030079840197e-05,
      "loss": 0.1846,
      "step": 18710
    },
    {
      "epoch": 0.57836238790753,
      "grad_norm": 0.08928353947856521,
      "learning_rate": 1.2366940836701652e-05,
      "loss": 0.1873,
      "step": 18720
    },
    {
      "epoch": 0.5786713421745746,
      "grad_norm": 0.08620986156199108,
      "learning_rate": 1.235185434225828e-05,
      "loss": 0.1905,
      "step": 18730
    },
    {
      "epoch": 0.5789802964416192,
      "grad_norm": 0.08774801161377836,
      "learning_rate": 1.233677061225913e-05,
      "loss": 0.1926,
      "step": 18740
    },
    {
      "epoch": 0.5792892507086639,
      "grad_norm": 0.0845975530435648,
      "learning_rate": 1.2321689662450371e-05,
      "loss": 0.1927,
      "step": 18750
    },
    {
      "epoch": 0.5795982049757085,
      "grad_norm": 0.09635276684974903,
      "learning_rate": 1.2306611508575256e-05,
      "loss": 0.1837,
      "step": 18760
    },
    {
      "epoch": 0.579907159242753,
      "grad_norm": 0.08282334929380135,
      "learning_rate": 1.2291536166374126e-05,
      "loss": 0.1836,
      "step": 18770
    },
    {
      "epoch": 0.5802161135097977,
      "grad_norm": 0.08588169105054738,
      "learning_rate": 1.227646365158439e-05,
      "loss": 0.184,
      "step": 18780
    },
    {
      "epoch": 0.5805250677768423,
      "grad_norm": 0.08938532218341083,
      "learning_rate": 1.2261393979940504e-05,
      "loss": 0.19,
      "step": 18790
    },
    {
      "epoch": 0.5808340220438869,
      "grad_norm": 0.08034961012220436,
      "learning_rate": 1.2246327167173943e-05,
      "loss": 0.1842,
      "step": 18800
    },
    {
      "epoch": 0.5811429763109316,
      "grad_norm": 0.09202410296299979,
      "learning_rate": 1.2231263229013226e-05,
      "loss": 0.1819,
      "step": 18810
    },
    {
      "epoch": 0.5814519305779762,
      "grad_norm": 0.09497765274832858,
      "learning_rate": 1.2216202181183836e-05,
      "loss": 0.1857,
      "step": 18820
    },
    {
      "epoch": 0.5817608848450209,
      "grad_norm": 0.08245146407877549,
      "learning_rate": 1.2201144039408273e-05,
      "loss": 0.1812,
      "step": 18830
    },
    {
      "epoch": 0.5820698391120654,
      "grad_norm": 0.09185748333333948,
      "learning_rate": 1.218608881940597e-05,
      "loss": 0.1958,
      "step": 18840
    },
    {
      "epoch": 0.58237879337911,
      "grad_norm": 0.08503410922279848,
      "learning_rate": 1.2171036536893344e-05,
      "loss": 0.1827,
      "step": 18850
    },
    {
      "epoch": 0.5826877476461547,
      "grad_norm": 0.08937965076157045,
      "learning_rate": 1.2155987207583718e-05,
      "loss": 0.1808,
      "step": 18860
    },
    {
      "epoch": 0.5829967019131993,
      "grad_norm": 0.07385702718310716,
      "learning_rate": 1.2140940847187345e-05,
      "loss": 0.1936,
      "step": 18870
    },
    {
      "epoch": 0.5833056561802439,
      "grad_norm": 0.08983367724418617,
      "learning_rate": 1.2125897471411374e-05,
      "loss": 0.1878,
      "step": 18880
    },
    {
      "epoch": 0.5836146104472886,
      "grad_norm": 0.12091610173052615,
      "learning_rate": 1.2110857095959848e-05,
      "loss": 0.1843,
      "step": 18890
    },
    {
      "epoch": 0.5839235647143332,
      "grad_norm": 0.08723914659503822,
      "learning_rate": 1.2095819736533656e-05,
      "loss": 0.1826,
      "step": 18900
    },
    {
      "epoch": 0.5842325189813777,
      "grad_norm": 0.09361335930489353,
      "learning_rate": 1.2080785408830568e-05,
      "loss": 0.1888,
      "step": 18910
    },
    {
      "epoch": 0.5845414732484224,
      "grad_norm": 0.1117420757778778,
      "learning_rate": 1.2065754128545162e-05,
      "loss": 0.1848,
      "step": 18920
    },
    {
      "epoch": 0.584850427515467,
      "grad_norm": 0.10537147119926492,
      "learning_rate": 1.205072591136886e-05,
      "loss": 0.1862,
      "step": 18930
    },
    {
      "epoch": 0.5851593817825116,
      "grad_norm": 0.0899076765838517,
      "learning_rate": 1.2035700772989858e-05,
      "loss": 0.1842,
      "step": 18940
    },
    {
      "epoch": 0.5854683360495563,
      "grad_norm": 0.13167115985599936,
      "learning_rate": 1.2020678729093165e-05,
      "loss": 0.1875,
      "step": 18950
    },
    {
      "epoch": 0.5857772903166009,
      "grad_norm": 0.0940759886393431,
      "learning_rate": 1.2005659795360548e-05,
      "loss": 0.1952,
      "step": 18960
    },
    {
      "epoch": 0.5860862445836456,
      "grad_norm": 0.08439561408458067,
      "learning_rate": 1.1990643987470513e-05,
      "loss": 0.1814,
      "step": 18970
    },
    {
      "epoch": 0.5863951988506901,
      "grad_norm": 0.09277608571555924,
      "learning_rate": 1.1975631321098331e-05,
      "loss": 0.1844,
      "step": 18980
    },
    {
      "epoch": 0.5867041531177347,
      "grad_norm": 0.09229614294737931,
      "learning_rate": 1.1960621811915972e-05,
      "loss": 0.1842,
      "step": 18990
    },
    {
      "epoch": 0.5870131073847794,
      "grad_norm": 0.08444775309449502,
      "learning_rate": 1.1945615475592123e-05,
      "loss": 0.1846,
      "step": 19000
    },
    {
      "epoch": 0.587322061651824,
      "grad_norm": 0.07699398247589585,
      "learning_rate": 1.1930612327792149e-05,
      "loss": 0.1848,
      "step": 19010
    },
    {
      "epoch": 0.5876310159188686,
      "grad_norm": 0.0811198704798943,
      "learning_rate": 1.1915612384178095e-05,
      "loss": 0.1891,
      "step": 19020
    },
    {
      "epoch": 0.5879399701859133,
      "grad_norm": 0.08420393526672411,
      "learning_rate": 1.1900615660408646e-05,
      "loss": 0.1889,
      "step": 19030
    },
    {
      "epoch": 0.5882489244529578,
      "grad_norm": 0.09157326851832695,
      "learning_rate": 1.1885622172139154e-05,
      "loss": 0.1869,
      "step": 19040
    },
    {
      "epoch": 0.5885578787200024,
      "grad_norm": 0.07432839476221717,
      "learning_rate": 1.1870631935021558e-05,
      "loss": 0.1914,
      "step": 19050
    },
    {
      "epoch": 0.5888668329870471,
      "grad_norm": 0.10557693446278404,
      "learning_rate": 1.1855644964704437e-05,
      "loss": 0.1827,
      "step": 19060
    },
    {
      "epoch": 0.5891757872540917,
      "grad_norm": 0.09831711453984064,
      "learning_rate": 1.1840661276832932e-05,
      "loss": 0.1869,
      "step": 19070
    },
    {
      "epoch": 0.5894847415211363,
      "grad_norm": 0.09770759844682415,
      "learning_rate": 1.1825680887048779e-05,
      "loss": 0.1822,
      "step": 19080
    },
    {
      "epoch": 0.589793695788181,
      "grad_norm": 0.10264452372425766,
      "learning_rate": 1.1810703810990254e-05,
      "loss": 0.1959,
      "step": 19090
    },
    {
      "epoch": 0.5901026500552256,
      "grad_norm": 0.07565463215403447,
      "learning_rate": 1.179573006429218e-05,
      "loss": 0.1871,
      "step": 19100
    },
    {
      "epoch": 0.5904116043222702,
      "grad_norm": 0.07221104152275891,
      "learning_rate": 1.178075966258591e-05,
      "loss": 0.1891,
      "step": 19110
    },
    {
      "epoch": 0.5907205585893148,
      "grad_norm": 0.1329860173858145,
      "learning_rate": 1.1765792621499305e-05,
      "loss": 0.1904,
      "step": 19120
    },
    {
      "epoch": 0.5910295128563594,
      "grad_norm": 0.09125889728006076,
      "learning_rate": 1.17508289566567e-05,
      "loss": 0.1833,
      "step": 19130
    },
    {
      "epoch": 0.5913384671234041,
      "grad_norm": 0.07184380342756125,
      "learning_rate": 1.1735868683678933e-05,
      "loss": 0.181,
      "step": 19140
    },
    {
      "epoch": 0.5916474213904487,
      "grad_norm": 0.07817020505022121,
      "learning_rate": 1.1720911818183273e-05,
      "loss": 0.182,
      "step": 19150
    },
    {
      "epoch": 0.5919563756574933,
      "grad_norm": 0.08513551887533943,
      "learning_rate": 1.1705958375783461e-05,
      "loss": 0.1927,
      "step": 19160
    },
    {
      "epoch": 0.592265329924538,
      "grad_norm": 0.09529922445323878,
      "learning_rate": 1.1691008372089635e-05,
      "loss": 0.1805,
      "step": 19170
    },
    {
      "epoch": 0.5925742841915825,
      "grad_norm": 0.08981535353884065,
      "learning_rate": 1.1676061822708372e-05,
      "loss": 0.1885,
      "step": 19180
    },
    {
      "epoch": 0.5928832384586271,
      "grad_norm": 0.08430387130843447,
      "learning_rate": 1.166111874324262e-05,
      "loss": 0.1844,
      "step": 19190
    },
    {
      "epoch": 0.5931921927256718,
      "grad_norm": 0.08695512811368436,
      "learning_rate": 1.164617914929171e-05,
      "loss": 0.1835,
      "step": 19200
    },
    {
      "epoch": 0.5935011469927164,
      "grad_norm": 0.07833563244574043,
      "learning_rate": 1.1631243056451347e-05,
      "loss": 0.1916,
      "step": 19210
    },
    {
      "epoch": 0.593810101259761,
      "grad_norm": 0.09216158564512843,
      "learning_rate": 1.1616310480313565e-05,
      "loss": 0.187,
      "step": 19220
    },
    {
      "epoch": 0.5941190555268057,
      "grad_norm": 0.10356040039665805,
      "learning_rate": 1.1601381436466738e-05,
      "loss": 0.1799,
      "step": 19230
    },
    {
      "epoch": 0.5944280097938502,
      "grad_norm": 0.0785531722095939,
      "learning_rate": 1.1586455940495547e-05,
      "loss": 0.1853,
      "step": 19240
    },
    {
      "epoch": 0.5947369640608949,
      "grad_norm": 0.11098212443906112,
      "learning_rate": 1.1571534007980972e-05,
      "loss": 0.1844,
      "step": 19250
    },
    {
      "epoch": 0.5950459183279395,
      "grad_norm": 0.09027492882496103,
      "learning_rate": 1.1556615654500264e-05,
      "loss": 0.1883,
      "step": 19260
    },
    {
      "epoch": 0.5953548725949841,
      "grad_norm": 0.10120630638431274,
      "learning_rate": 1.1541700895626958e-05,
      "loss": 0.191,
      "step": 19270
    },
    {
      "epoch": 0.5956638268620288,
      "grad_norm": 0.12492870297484539,
      "learning_rate": 1.1526789746930815e-05,
      "loss": 0.1943,
      "step": 19280
    },
    {
      "epoch": 0.5959727811290734,
      "grad_norm": 0.09996879456988371,
      "learning_rate": 1.1511882223977844e-05,
      "loss": 0.1901,
      "step": 19290
    },
    {
      "epoch": 0.596281735396118,
      "grad_norm": 0.08282819545882,
      "learning_rate": 1.149697834233025e-05,
      "loss": 0.184,
      "step": 19300
    },
    {
      "epoch": 0.5965906896631626,
      "grad_norm": 0.0844753368413714,
      "learning_rate": 1.148207811754646e-05,
      "loss": 0.1977,
      "step": 19310
    },
    {
      "epoch": 0.5968996439302072,
      "grad_norm": 0.08188342055077452,
      "learning_rate": 1.1467181565181063e-05,
      "loss": 0.1823,
      "step": 19320
    },
    {
      "epoch": 0.5972085981972518,
      "grad_norm": 0.07948457256262524,
      "learning_rate": 1.1452288700784827e-05,
      "loss": 0.1837,
      "step": 19330
    },
    {
      "epoch": 0.5975175524642965,
      "grad_norm": 0.0982865574621929,
      "learning_rate": 1.1437399539904668e-05,
      "loss": 0.1874,
      "step": 19340
    },
    {
      "epoch": 0.5978265067313411,
      "grad_norm": 0.08748222062334479,
      "learning_rate": 1.1422514098083631e-05,
      "loss": 0.1844,
      "step": 19350
    },
    {
      "epoch": 0.5981354609983857,
      "grad_norm": 0.0849517310534258,
      "learning_rate": 1.140763239086088e-05,
      "loss": 0.1843,
      "step": 19360
    },
    {
      "epoch": 0.5984444152654304,
      "grad_norm": 0.07401851701885574,
      "learning_rate": 1.1392754433771687e-05,
      "loss": 0.1831,
      "step": 19370
    },
    {
      "epoch": 0.5987533695324749,
      "grad_norm": 0.09235152080487286,
      "learning_rate": 1.1377880242347397e-05,
      "loss": 0.1868,
      "step": 19380
    },
    {
      "epoch": 0.5990623237995196,
      "grad_norm": 0.0788790577912453,
      "learning_rate": 1.1363009832115439e-05,
      "loss": 0.1958,
      "step": 19390
    },
    {
      "epoch": 0.5993712780665642,
      "grad_norm": 0.08257094376240925,
      "learning_rate": 1.1348143218599278e-05,
      "loss": 0.179,
      "step": 19400
    },
    {
      "epoch": 0.5996802323336088,
      "grad_norm": 0.08331281503892017,
      "learning_rate": 1.1333280417318434e-05,
      "loss": 0.1779,
      "step": 19410
    },
    {
      "epoch": 0.5999891866006535,
      "grad_norm": 0.07932051293633373,
      "learning_rate": 1.1318421443788436e-05,
      "loss": 0.1859,
      "step": 19420
    },
    {
      "epoch": 0.6002981408676981,
      "grad_norm": 0.08526455481169266,
      "learning_rate": 1.1303566313520806e-05,
      "loss": 0.1842,
      "step": 19430
    },
    {
      "epoch": 0.6006070951347426,
      "grad_norm": 0.09674375852523574,
      "learning_rate": 1.1288715042023085e-05,
      "loss": 0.1881,
      "step": 19440
    },
    {
      "epoch": 0.6009160494017873,
      "grad_norm": 0.09569181761747295,
      "learning_rate": 1.1273867644798754e-05,
      "loss": 0.1857,
      "step": 19450
    },
    {
      "epoch": 0.6012250036688319,
      "grad_norm": 0.07962622292524987,
      "learning_rate": 1.1259024137347272e-05,
      "loss": 0.186,
      "step": 19460
    },
    {
      "epoch": 0.6015339579358765,
      "grad_norm": 0.08719966337395878,
      "learning_rate": 1.1244184535164018e-05,
      "loss": 0.1856,
      "step": 19470
    },
    {
      "epoch": 0.6018429122029212,
      "grad_norm": 0.08897484019878672,
      "learning_rate": 1.1229348853740317e-05,
      "loss": 0.1859,
      "step": 19480
    },
    {
      "epoch": 0.6021518664699658,
      "grad_norm": 0.07763418090346706,
      "learning_rate": 1.1214517108563377e-05,
      "loss": 0.1821,
      "step": 19490
    },
    {
      "epoch": 0.6024608207370103,
      "grad_norm": 0.08294157293505787,
      "learning_rate": 1.1199689315116323e-05,
      "loss": 0.1857,
      "step": 19500
    },
    {
      "epoch": 0.602769775004055,
      "grad_norm": 0.09126750085376728,
      "learning_rate": 1.1184865488878123e-05,
      "loss": 0.1841,
      "step": 19510
    },
    {
      "epoch": 0.6030787292710996,
      "grad_norm": 0.08780804555609552,
      "learning_rate": 1.1170045645323638e-05,
      "loss": 0.1819,
      "step": 19520
    },
    {
      "epoch": 0.6033876835381443,
      "grad_norm": 0.07561012286175459,
      "learning_rate": 1.115522979992354e-05,
      "loss": 0.1845,
      "step": 19530
    },
    {
      "epoch": 0.6036966378051889,
      "grad_norm": 0.08305953026102175,
      "learning_rate": 1.1140417968144355e-05,
      "loss": 0.1824,
      "step": 19540
    },
    {
      "epoch": 0.6040055920722335,
      "grad_norm": 0.09252131059633029,
      "learning_rate": 1.1125610165448394e-05,
      "loss": 0.1895,
      "step": 19550
    },
    {
      "epoch": 0.6043145463392782,
      "grad_norm": 0.08466140339181151,
      "learning_rate": 1.1110806407293784e-05,
      "loss": 0.1851,
      "step": 19560
    },
    {
      "epoch": 0.6046235006063228,
      "grad_norm": 0.0849766690814433,
      "learning_rate": 1.1096006709134404e-05,
      "loss": 0.1843,
      "step": 19570
    },
    {
      "epoch": 0.6049324548733673,
      "grad_norm": 0.08097195703181048,
      "learning_rate": 1.1081211086419929e-05,
      "loss": 0.1829,
      "step": 19580
    },
    {
      "epoch": 0.605241409140412,
      "grad_norm": 0.08667928640904893,
      "learning_rate": 1.1066419554595745e-05,
      "loss": 0.1914,
      "step": 19590
    },
    {
      "epoch": 0.6055503634074566,
      "grad_norm": 0.15393379599495152,
      "learning_rate": 1.1051632129102997e-05,
      "loss": 0.1945,
      "step": 19600
    },
    {
      "epoch": 0.6058593176745012,
      "grad_norm": 0.09093211482736684,
      "learning_rate": 1.103684882537852e-05,
      "loss": 0.1897,
      "step": 19610
    },
    {
      "epoch": 0.6061682719415459,
      "grad_norm": 0.08624215793715367,
      "learning_rate": 1.102206965885486e-05,
      "loss": 0.1865,
      "step": 19620
    },
    {
      "epoch": 0.6064772262085905,
      "grad_norm": 0.08456190559451564,
      "learning_rate": 1.1007294644960235e-05,
      "loss": 0.1859,
      "step": 19630
    },
    {
      "epoch": 0.606786180475635,
      "grad_norm": 0.08966066955579385,
      "learning_rate": 1.0992523799118543e-05,
      "loss": 0.1859,
      "step": 19640
    },
    {
      "epoch": 0.6070951347426797,
      "grad_norm": 0.08657352283109619,
      "learning_rate": 1.0977757136749313e-05,
      "loss": 0.1902,
      "step": 19650
    },
    {
      "epoch": 0.6074040890097243,
      "grad_norm": 0.07975700433016425,
      "learning_rate": 1.0962994673267711e-05,
      "loss": 0.1835,
      "step": 19660
    },
    {
      "epoch": 0.607713043276769,
      "grad_norm": 0.08009979147722592,
      "learning_rate": 1.0948236424084534e-05,
      "loss": 0.1803,
      "step": 19670
    },
    {
      "epoch": 0.6080219975438136,
      "grad_norm": 0.08506789440825369,
      "learning_rate": 1.0933482404606164e-05,
      "loss": 0.1797,
      "step": 19680
    },
    {
      "epoch": 0.6083309518108582,
      "grad_norm": 0.10157857136421491,
      "learning_rate": 1.091873263023457e-05,
      "loss": 0.191,
      "step": 19690
    },
    {
      "epoch": 0.6086399060779029,
      "grad_norm": 0.07557224513227645,
      "learning_rate": 1.0903987116367289e-05,
      "loss": 0.1857,
      "step": 19700
    },
    {
      "epoch": 0.6089488603449474,
      "grad_norm": 0.11128795404350794,
      "learning_rate": 1.0889245878397426e-05,
      "loss": 0.1831,
      "step": 19710
    },
    {
      "epoch": 0.609257814611992,
      "grad_norm": 0.10292068874463854,
      "learning_rate": 1.0874508931713593e-05,
      "loss": 0.1852,
      "step": 19720
    },
    {
      "epoch": 0.6095667688790367,
      "grad_norm": 0.08079840064866879,
      "learning_rate": 1.0859776291699956e-05,
      "loss": 0.178,
      "step": 19730
    },
    {
      "epoch": 0.6098757231460813,
      "grad_norm": 0.08623791196191595,
      "learning_rate": 1.0845047973736153e-05,
      "loss": 0.1924,
      "step": 19740
    },
    {
      "epoch": 0.6101846774131259,
      "grad_norm": 0.12145228572606707,
      "learning_rate": 1.0830323993197339e-05,
      "loss": 0.1915,
      "step": 19750
    },
    {
      "epoch": 0.6104936316801706,
      "grad_norm": 0.08729285672960486,
      "learning_rate": 1.0815604365454113e-05,
      "loss": 0.184,
      "step": 19760
    },
    {
      "epoch": 0.6108025859472151,
      "grad_norm": 0.08460342384525342,
      "learning_rate": 1.0800889105872559e-05,
      "loss": 0.1912,
      "step": 19770
    },
    {
      "epoch": 0.6111115402142597,
      "grad_norm": 0.0871823576044211,
      "learning_rate": 1.0786178229814176e-05,
      "loss": 0.1871,
      "step": 19780
    },
    {
      "epoch": 0.6114204944813044,
      "grad_norm": 0.07870026373917863,
      "learning_rate": 1.0771471752635903e-05,
      "loss": 0.1864,
      "step": 19790
    },
    {
      "epoch": 0.611729448748349,
      "grad_norm": 0.08073327956470966,
      "learning_rate": 1.0756769689690074e-05,
      "loss": 0.183,
      "step": 19800
    },
    {
      "epoch": 0.6120384030153937,
      "grad_norm": 0.08087990409818094,
      "learning_rate": 1.0742072056324433e-05,
      "loss": 0.1842,
      "step": 19810
    },
    {
      "epoch": 0.6123473572824383,
      "grad_norm": 0.07840518720141709,
      "learning_rate": 1.0727378867882075e-05,
      "loss": 0.1844,
      "step": 19820
    },
    {
      "epoch": 0.6126563115494829,
      "grad_norm": 0.0857951081060033,
      "learning_rate": 1.0712690139701487e-05,
      "loss": 0.1807,
      "step": 19830
    },
    {
      "epoch": 0.6129652658165275,
      "grad_norm": 0.1230016763922673,
      "learning_rate": 1.0698005887116463e-05,
      "loss": 0.1949,
      "step": 19840
    },
    {
      "epoch": 0.6132742200835721,
      "grad_norm": 0.08977077281458193,
      "learning_rate": 1.0683326125456157e-05,
      "loss": 0.1828,
      "step": 19850
    },
    {
      "epoch": 0.6135831743506167,
      "grad_norm": 0.07661706950343256,
      "learning_rate": 1.0668650870045013e-05,
      "loss": 0.1805,
      "step": 19860
    },
    {
      "epoch": 0.6138921286176614,
      "grad_norm": 0.08009017554123325,
      "learning_rate": 1.0653980136202788e-05,
      "loss": 0.1868,
      "step": 19870
    },
    {
      "epoch": 0.614201082884706,
      "grad_norm": 0.08070096082327056,
      "learning_rate": 1.0639313939244509e-05,
      "loss": 0.1804,
      "step": 19880
    },
    {
      "epoch": 0.6145100371517506,
      "grad_norm": 0.07529212229842093,
      "learning_rate": 1.062465229448046e-05,
      "loss": 0.184,
      "step": 19890
    },
    {
      "epoch": 0.6148189914187953,
      "grad_norm": 0.0785703936539135,
      "learning_rate": 1.0609995217216183e-05,
      "loss": 0.1833,
      "step": 19900
    },
    {
      "epoch": 0.6151279456858398,
      "grad_norm": 0.08097943876546655,
      "learning_rate": 1.0595342722752455e-05,
      "loss": 0.1978,
      "step": 19910
    },
    {
      "epoch": 0.6154368999528844,
      "grad_norm": 0.08474437129651521,
      "learning_rate": 1.0580694826385264e-05,
      "loss": 0.1836,
      "step": 19920
    },
    {
      "epoch": 0.6157458542199291,
      "grad_norm": 0.08672105010327077,
      "learning_rate": 1.056605154340579e-05,
      "loss": 0.1876,
      "step": 19930
    },
    {
      "epoch": 0.6160548084869737,
      "grad_norm": 0.07380386784963713,
      "learning_rate": 1.0551412889100413e-05,
      "loss": 0.1846,
      "step": 19940
    },
    {
      "epoch": 0.6163637627540184,
      "grad_norm": 0.09422829500226426,
      "learning_rate": 1.0536778878750665e-05,
      "loss": 0.1825,
      "step": 19950
    },
    {
      "epoch": 0.616672717021063,
      "grad_norm": 0.08064736933954149,
      "learning_rate": 1.052214952763325e-05,
      "loss": 0.1851,
      "step": 19960
    },
    {
      "epoch": 0.6169816712881075,
      "grad_norm": 0.08299793921296503,
      "learning_rate": 1.050752485101998e-05,
      "loss": 0.189,
      "step": 19970
    },
    {
      "epoch": 0.6172906255551522,
      "grad_norm": 0.20022126832501388,
      "learning_rate": 1.049290486417782e-05,
      "loss": 0.189,
      "step": 19980
    },
    {
      "epoch": 0.6175995798221968,
      "grad_norm": 0.08780516760895216,
      "learning_rate": 1.047828958236881e-05,
      "loss": 0.1945,
      "step": 19990
    },
    {
      "epoch": 0.6179085340892414,
      "grad_norm": 0.08728220336156364,
      "learning_rate": 1.0463679020850098e-05,
      "loss": 0.1849,
      "step": 20000
    },
    {
      "epoch": 0.6182174883562861,
      "grad_norm": 0.07646045519503682,
      "learning_rate": 1.0449073194873893e-05,
      "loss": 0.1827,
      "step": 20010
    },
    {
      "epoch": 0.6185264426233307,
      "grad_norm": 0.0848586359033337,
      "learning_rate": 1.0434472119687468e-05,
      "loss": 0.1984,
      "step": 20020
    },
    {
      "epoch": 0.6188353968903753,
      "grad_norm": 0.07910424422713831,
      "learning_rate": 1.0419875810533129e-05,
      "loss": 0.1889,
      "step": 20030
    },
    {
      "epoch": 0.61914435115742,
      "grad_norm": 0.07327911861754191,
      "learning_rate": 1.0405284282648216e-05,
      "loss": 0.1898,
      "step": 20040
    },
    {
      "epoch": 0.6194533054244645,
      "grad_norm": 0.09398610431686306,
      "learning_rate": 1.0390697551265068e-05,
      "loss": 0.1869,
      "step": 20050
    },
    {
      "epoch": 0.6197622596915091,
      "grad_norm": 0.07253967693690853,
      "learning_rate": 1.0376115631611029e-05,
      "loss": 0.181,
      "step": 20060
    },
    {
      "epoch": 0.6200712139585538,
      "grad_norm": 0.09232348245256568,
      "learning_rate": 1.03615385389084e-05,
      "loss": 0.1879,
      "step": 20070
    },
    {
      "epoch": 0.6203801682255984,
      "grad_norm": 0.0923122510204038,
      "learning_rate": 1.0346966288374472e-05,
      "loss": 0.1829,
      "step": 20080
    },
    {
      "epoch": 0.6206891224926431,
      "grad_norm": 0.07936874657939968,
      "learning_rate": 1.0332398895221449e-05,
      "loss": 0.1818,
      "step": 20090
    },
    {
      "epoch": 0.6209980767596877,
      "grad_norm": 0.08894380415302283,
      "learning_rate": 1.0317836374656494e-05,
      "loss": 0.1847,
      "step": 20100
    },
    {
      "epoch": 0.6213070310267322,
      "grad_norm": 0.07863944507287474,
      "learning_rate": 1.030327874188166e-05,
      "loss": 0.1845,
      "step": 20110
    },
    {
      "epoch": 0.6216159852937769,
      "grad_norm": 0.08140127687302948,
      "learning_rate": 1.0288726012093909e-05,
      "loss": 0.1804,
      "step": 20120
    },
    {
      "epoch": 0.6219249395608215,
      "grad_norm": 0.08163237136633517,
      "learning_rate": 1.0274178200485086e-05,
      "loss": 0.1872,
      "step": 20130
    },
    {
      "epoch": 0.6222338938278661,
      "grad_norm": 0.16992938890110298,
      "learning_rate": 1.0259635322241894e-05,
      "loss": 0.193,
      "step": 20140
    },
    {
      "epoch": 0.6225428480949108,
      "grad_norm": 0.12292714328551844,
      "learning_rate": 1.0245097392545896e-05,
      "loss": 0.1919,
      "step": 20150
    },
    {
      "epoch": 0.6228518023619554,
      "grad_norm": 0.0719191327526604,
      "learning_rate": 1.0230564426573476e-05,
      "loss": 0.1879,
      "step": 20160
    },
    {
      "epoch": 0.623160756629,
      "grad_norm": 0.09503697141926552,
      "learning_rate": 1.0216036439495852e-05,
      "loss": 0.1797,
      "step": 20170
    },
    {
      "epoch": 0.6234697108960446,
      "grad_norm": 0.08358688600952284,
      "learning_rate": 1.0201513446479028e-05,
      "loss": 0.1838,
      "step": 20180
    },
    {
      "epoch": 0.6237786651630892,
      "grad_norm": 0.09438394393783162,
      "learning_rate": 1.0186995462683815e-05,
      "loss": 0.1828,
      "step": 20190
    },
    {
      "epoch": 0.6240876194301338,
      "grad_norm": 0.11716791528411023,
      "learning_rate": 1.0172482503265768e-05,
      "loss": 0.1935,
      "step": 20200
    },
    {
      "epoch": 0.6243965736971785,
      "grad_norm": 0.09361063759492731,
      "learning_rate": 1.0157974583375225e-05,
      "loss": 0.1818,
      "step": 20210
    },
    {
      "epoch": 0.6247055279642231,
      "grad_norm": 0.08105502479633177,
      "learning_rate": 1.0143471718157245e-05,
      "loss": 0.1887,
      "step": 20220
    },
    {
      "epoch": 0.6250144822312678,
      "grad_norm": 0.09295395575622786,
      "learning_rate": 1.0128973922751613e-05,
      "loss": 0.1898,
      "step": 20230
    },
    {
      "epoch": 0.6253234364983123,
      "grad_norm": 0.08766105008046463,
      "learning_rate": 1.0114481212292826e-05,
      "loss": 0.1809,
      "step": 20240
    },
    {
      "epoch": 0.6256323907653569,
      "grad_norm": 0.08985001838450198,
      "learning_rate": 1.0099993601910073e-05,
      "loss": 0.1775,
      "step": 20250
    },
    {
      "epoch": 0.6259413450324016,
      "grad_norm": 0.08344503889344561,
      "learning_rate": 1.0085511106727207e-05,
      "loss": 0.1794,
      "step": 20260
    },
    {
      "epoch": 0.6262502992994462,
      "grad_norm": 0.09504270442450875,
      "learning_rate": 1.0071033741862764e-05,
      "loss": 0.1835,
      "step": 20270
    },
    {
      "epoch": 0.6265592535664908,
      "grad_norm": 0.0857553897863654,
      "learning_rate": 1.00565615224299e-05,
      "loss": 0.1861,
      "step": 20280
    },
    {
      "epoch": 0.6268682078335355,
      "grad_norm": 0.08574806320199437,
      "learning_rate": 1.0042094463536419e-05,
      "loss": 0.1892,
      "step": 20290
    },
    {
      "epoch": 0.62717716210058,
      "grad_norm": 0.07834983542251327,
      "learning_rate": 1.0027632580284723e-05,
      "loss": 0.1866,
      "step": 20300
    },
    {
      "epoch": 0.6274861163676246,
      "grad_norm": 0.08745193791786973,
      "learning_rate": 1.0013175887771824e-05,
      "loss": 0.1887,
      "step": 20310
    },
    {
      "epoch": 0.6277950706346693,
      "grad_norm": 0.0788487355313793,
      "learning_rate": 9.998724401089301e-06,
      "loss": 0.1854,
      "step": 20320
    },
    {
      "epoch": 0.6281040249017139,
      "grad_norm": 0.08009857430142646,
      "learning_rate": 9.984278135323312e-06,
      "loss": 0.181,
      "step": 20330
    },
    {
      "epoch": 0.6284129791687585,
      "grad_norm": 0.0926032013008688,
      "learning_rate": 9.96983710555456e-06,
      "loss": 0.1895,
      "step": 20340
    },
    {
      "epoch": 0.6287219334358032,
      "grad_norm": 0.09862161472165971,
      "learning_rate": 9.955401326858275e-06,
      "loss": 0.1796,
      "step": 20350
    },
    {
      "epoch": 0.6290308877028478,
      "grad_norm": 0.07656621564990976,
      "learning_rate": 9.940970814304217e-06,
      "loss": 0.1814,
      "step": 20360
    },
    {
      "epoch": 0.6293398419698925,
      "grad_norm": 0.09098582721427066,
      "learning_rate": 9.92654558295664e-06,
      "loss": 0.1829,
      "step": 20370
    },
    {
      "epoch": 0.629648796236937,
      "grad_norm": 0.09097843459638964,
      "learning_rate": 9.912125647874293e-06,
      "loss": 0.1788,
      "step": 20380
    },
    {
      "epoch": 0.6299577505039816,
      "grad_norm": 0.10792002657150579,
      "learning_rate": 9.897711024110381e-06,
      "loss": 0.1826,
      "step": 20390
    },
    {
      "epoch": 0.6302667047710263,
      "grad_norm": 0.07447835947155786,
      "learning_rate": 9.88330172671259e-06,
      "loss": 0.1821,
      "step": 20400
    },
    {
      "epoch": 0.6305756590380709,
      "grad_norm": 0.09406839755830929,
      "learning_rate": 9.868897770723012e-06,
      "loss": 0.1908,
      "step": 20410
    },
    {
      "epoch": 0.6308846133051155,
      "grad_norm": 0.1285771773616567,
      "learning_rate": 9.854499171178197e-06,
      "loss": 0.1885,
      "step": 20420
    },
    {
      "epoch": 0.6311935675721602,
      "grad_norm": 0.09292480698582771,
      "learning_rate": 9.840105943109077e-06,
      "loss": 0.1889,
      "step": 20430
    },
    {
      "epoch": 0.6315025218392047,
      "grad_norm": 0.07294220435329125,
      "learning_rate": 9.825718101540994e-06,
      "loss": 0.1817,
      "step": 20440
    },
    {
      "epoch": 0.6318114761062493,
      "grad_norm": 0.0895501299879043,
      "learning_rate": 9.81133566149366e-06,
      "loss": 0.1797,
      "step": 20450
    },
    {
      "epoch": 0.632120430373294,
      "grad_norm": 0.08606985323050444,
      "learning_rate": 9.79695863798114e-06,
      "loss": 0.1895,
      "step": 20460
    },
    {
      "epoch": 0.6324293846403386,
      "grad_norm": 0.07606324384658682,
      "learning_rate": 9.782587046011863e-06,
      "loss": 0.1823,
      "step": 20470
    },
    {
      "epoch": 0.6327383389073832,
      "grad_norm": 0.08687194167340641,
      "learning_rate": 9.768220900588574e-06,
      "loss": 0.1899,
      "step": 20480
    },
    {
      "epoch": 0.6330472931744279,
      "grad_norm": 0.0936679680166201,
      "learning_rate": 9.753860216708333e-06,
      "loss": 0.1848,
      "step": 20490
    },
    {
      "epoch": 0.6333562474414725,
      "grad_norm": 0.07860288394024362,
      "learning_rate": 9.73950500936251e-06,
      "loss": 0.1835,
      "step": 20500
    },
    {
      "epoch": 0.6336652017085171,
      "grad_norm": 0.09208404294384408,
      "learning_rate": 9.725155293536739e-06,
      "loss": 0.1926,
      "step": 20510
    },
    {
      "epoch": 0.6339741559755617,
      "grad_norm": 0.08090329111864401,
      "learning_rate": 9.71081108421094e-06,
      "loss": 0.1795,
      "step": 20520
    },
    {
      "epoch": 0.6342831102426063,
      "grad_norm": 0.09430035972466351,
      "learning_rate": 9.696472396359274e-06,
      "loss": 0.1807,
      "step": 20530
    },
    {
      "epoch": 0.634592064509651,
      "grad_norm": 0.09330412143446754,
      "learning_rate": 9.682139244950143e-06,
      "loss": 0.1836,
      "step": 20540
    },
    {
      "epoch": 0.6349010187766956,
      "grad_norm": 0.07667001788217906,
      "learning_rate": 9.667811644946167e-06,
      "loss": 0.1806,
      "step": 20550
    },
    {
      "epoch": 0.6352099730437402,
      "grad_norm": 0.09372099288123097,
      "learning_rate": 9.653489611304164e-06,
      "loss": 0.1816,
      "step": 20560
    },
    {
      "epoch": 0.6355189273107849,
      "grad_norm": 0.09008663985017154,
      "learning_rate": 9.63917315897516e-06,
      "loss": 0.1809,
      "step": 20570
    },
    {
      "epoch": 0.6358278815778294,
      "grad_norm": 0.12952734707392144,
      "learning_rate": 9.624862302904331e-06,
      "loss": 0.1905,
      "step": 20580
    },
    {
      "epoch": 0.636136835844874,
      "grad_norm": 0.09219556248390978,
      "learning_rate": 9.610557058031032e-06,
      "loss": 0.1867,
      "step": 20590
    },
    {
      "epoch": 0.6364457901119187,
      "grad_norm": 0.09330127555515816,
      "learning_rate": 9.59625743928874e-06,
      "loss": 0.1828,
      "step": 20600
    },
    {
      "epoch": 0.6367547443789633,
      "grad_norm": 0.08658418890523488,
      "learning_rate": 9.581963461605083e-06,
      "loss": 0.1784,
      "step": 20610
    },
    {
      "epoch": 0.6370636986460079,
      "grad_norm": 0.08868316701971833,
      "learning_rate": 9.567675139901774e-06,
      "loss": 0.1907,
      "step": 20620
    },
    {
      "epoch": 0.6373726529130526,
      "grad_norm": 0.0856353186910355,
      "learning_rate": 9.55339248909465e-06,
      "loss": 0.1948,
      "step": 20630
    },
    {
      "epoch": 0.6376816071800971,
      "grad_norm": 0.10250504390189848,
      "learning_rate": 9.539115524093597e-06,
      "loss": 0.1891,
      "step": 20640
    },
    {
      "epoch": 0.6379905614471418,
      "grad_norm": 0.07868317571192376,
      "learning_rate": 9.524844259802594e-06,
      "loss": 0.1914,
      "step": 20650
    },
    {
      "epoch": 0.6382995157141864,
      "grad_norm": 0.09384218652468425,
      "learning_rate": 9.510578711119645e-06,
      "loss": 0.1874,
      "step": 20660
    },
    {
      "epoch": 0.638608469981231,
      "grad_norm": 0.11899893173518487,
      "learning_rate": 9.496318892936809e-06,
      "loss": 0.1829,
      "step": 20670
    },
    {
      "epoch": 0.6389174242482757,
      "grad_norm": 0.0816515123513843,
      "learning_rate": 9.482064820140146e-06,
      "loss": 0.1808,
      "step": 20680
    },
    {
      "epoch": 0.6392263785153203,
      "grad_norm": 0.0773146102183417,
      "learning_rate": 9.467816507609728e-06,
      "loss": 0.189,
      "step": 20690
    },
    {
      "epoch": 0.6395353327823649,
      "grad_norm": 0.08757599292315726,
      "learning_rate": 9.453573970219603e-06,
      "loss": 0.1884,
      "step": 20700
    },
    {
      "epoch": 0.6398442870494095,
      "grad_norm": 0.10298427025583606,
      "learning_rate": 9.439337222837808e-06,
      "loss": 0.1826,
      "step": 20710
    },
    {
      "epoch": 0.6401532413164541,
      "grad_norm": 0.09188953546962604,
      "learning_rate": 9.425106280326318e-06,
      "loss": 0.1906,
      "step": 20720
    },
    {
      "epoch": 0.6404621955834987,
      "grad_norm": 0.08883895755213242,
      "learning_rate": 9.410881157541062e-06,
      "loss": 0.1816,
      "step": 20730
    },
    {
      "epoch": 0.6407711498505434,
      "grad_norm": 0.09682712583338604,
      "learning_rate": 9.396661869331877e-06,
      "loss": 0.1884,
      "step": 20740
    },
    {
      "epoch": 0.641080104117588,
      "grad_norm": 0.09008449684419392,
      "learning_rate": 9.382448430542532e-06,
      "loss": 0.1849,
      "step": 20750
    },
    {
      "epoch": 0.6413890583846326,
      "grad_norm": 0.09547832620192426,
      "learning_rate": 9.368240856010667e-06,
      "loss": 0.181,
      "step": 20760
    },
    {
      "epoch": 0.6416980126516773,
      "grad_norm": 0.08531693574310624,
      "learning_rate": 9.354039160567818e-06,
      "loss": 0.1811,
      "step": 20770
    },
    {
      "epoch": 0.6420069669187218,
      "grad_norm": 0.0941283910320591,
      "learning_rate": 9.339843359039373e-06,
      "loss": 0.1873,
      "step": 20780
    },
    {
      "epoch": 0.6423159211857665,
      "grad_norm": 0.08622755761927506,
      "learning_rate": 9.325653466244565e-06,
      "loss": 0.1817,
      "step": 20790
    },
    {
      "epoch": 0.6426248754528111,
      "grad_norm": 0.10957996664407295,
      "learning_rate": 9.311469496996472e-06,
      "loss": 0.1851,
      "step": 20800
    },
    {
      "epoch": 0.6429338297198557,
      "grad_norm": 0.09196273218754084,
      "learning_rate": 9.297291466101975e-06,
      "loss": 0.1839,
      "step": 20810
    },
    {
      "epoch": 0.6432427839869004,
      "grad_norm": 0.09089175446246295,
      "learning_rate": 9.283119388361768e-06,
      "loss": 0.1838,
      "step": 20820
    },
    {
      "epoch": 0.643551738253945,
      "grad_norm": 0.11048996293256481,
      "learning_rate": 9.268953278570314e-06,
      "loss": 0.1834,
      "step": 20830
    },
    {
      "epoch": 0.6438606925209895,
      "grad_norm": 0.08942417078026148,
      "learning_rate": 9.254793151515864e-06,
      "loss": 0.1878,
      "step": 20840
    },
    {
      "epoch": 0.6441696467880342,
      "grad_norm": 0.09466700710009077,
      "learning_rate": 9.240639021980406e-06,
      "loss": 0.1834,
      "step": 20850
    },
    {
      "epoch": 0.6444786010550788,
      "grad_norm": 0.09873195487288632,
      "learning_rate": 9.226490904739689e-06,
      "loss": 0.188,
      "step": 20860
    },
    {
      "epoch": 0.6447875553221234,
      "grad_norm": 0.09173169657963821,
      "learning_rate": 9.212348814563158e-06,
      "loss": 0.1816,
      "step": 20870
    },
    {
      "epoch": 0.6450965095891681,
      "grad_norm": 0.09500502487193062,
      "learning_rate": 9.198212766213997e-06,
      "loss": 0.1839,
      "step": 20880
    },
    {
      "epoch": 0.6454054638562127,
      "grad_norm": 0.09436950076336052,
      "learning_rate": 9.18408277444905e-06,
      "loss": 0.1808,
      "step": 20890
    },
    {
      "epoch": 0.6457144181232572,
      "grad_norm": 0.10117431682472817,
      "learning_rate": 9.169958854018874e-06,
      "loss": 0.1812,
      "step": 20900
    },
    {
      "epoch": 0.6460233723903019,
      "grad_norm": 0.09242615454166188,
      "learning_rate": 9.155841019667655e-06,
      "loss": 0.1811,
      "step": 20910
    },
    {
      "epoch": 0.6463323266573465,
      "grad_norm": 0.08406771936945807,
      "learning_rate": 9.141729286133246e-06,
      "loss": 0.1823,
      "step": 20920
    },
    {
      "epoch": 0.6466412809243912,
      "grad_norm": 0.09330851113984145,
      "learning_rate": 9.12762366814712e-06,
      "loss": 0.1774,
      "step": 20930
    },
    {
      "epoch": 0.6469502351914358,
      "grad_norm": 0.09296950290615348,
      "learning_rate": 9.11352418043438e-06,
      "loss": 0.187,
      "step": 20940
    },
    {
      "epoch": 0.6472591894584804,
      "grad_norm": 0.08599265520687546,
      "learning_rate": 9.09943083771371e-06,
      "loss": 0.1803,
      "step": 20950
    },
    {
      "epoch": 0.6475681437255251,
      "grad_norm": 0.09119409994774319,
      "learning_rate": 9.085343654697401e-06,
      "loss": 0.1928,
      "step": 20960
    },
    {
      "epoch": 0.6478770979925697,
      "grad_norm": 0.08082505933686907,
      "learning_rate": 9.07126264609129e-06,
      "loss": 0.1888,
      "step": 20970
    },
    {
      "epoch": 0.6481860522596142,
      "grad_norm": 0.09407633132491798,
      "learning_rate": 9.057187826594796e-06,
      "loss": 0.1836,
      "step": 20980
    },
    {
      "epoch": 0.6484950065266589,
      "grad_norm": 0.09247638767842638,
      "learning_rate": 9.04311921090085e-06,
      "loss": 0.1821,
      "step": 20990
    },
    {
      "epoch": 0.6488039607937035,
      "grad_norm": 0.09549021328910906,
      "learning_rate": 9.029056813695927e-06,
      "loss": 0.1904,
      "step": 21000
    },
    {
      "epoch": 0.6491129150607481,
      "grad_norm": 0.14298300777568937,
      "learning_rate": 9.015000649659997e-06,
      "loss": 0.1823,
      "step": 21010
    },
    {
      "epoch": 0.6494218693277928,
      "grad_norm": 0.10420840897491478,
      "learning_rate": 9.00095073346653e-06,
      "loss": 0.1856,
      "step": 21020
    },
    {
      "epoch": 0.6497308235948374,
      "grad_norm": 0.08341315756376709,
      "learning_rate": 8.98690707978247e-06,
      "loss": 0.1779,
      "step": 21030
    },
    {
      "epoch": 0.6500397778618819,
      "grad_norm": 0.12059619479918031,
      "learning_rate": 8.972869703268226e-06,
      "loss": 0.1803,
      "step": 21040
    },
    {
      "epoch": 0.6503487321289266,
      "grad_norm": 0.09904239652535235,
      "learning_rate": 8.958838618577658e-06,
      "loss": 0.1848,
      "step": 21050
    },
    {
      "epoch": 0.6506576863959712,
      "grad_norm": 0.09769625000327982,
      "learning_rate": 8.944813840358046e-06,
      "loss": 0.184,
      "step": 21060
    },
    {
      "epoch": 0.6509666406630159,
      "grad_norm": 0.0915300797061274,
      "learning_rate": 8.930795383250105e-06,
      "loss": 0.1828,
      "step": 21070
    },
    {
      "epoch": 0.6512755949300605,
      "grad_norm": 0.12815218888121502,
      "learning_rate": 8.91678326188793e-06,
      "loss": 0.1857,
      "step": 21080
    },
    {
      "epoch": 0.6515845491971051,
      "grad_norm": 0.10658342707078763,
      "learning_rate": 8.902777490899021e-06,
      "loss": 0.1851,
      "step": 21090
    },
    {
      "epoch": 0.6518935034641498,
      "grad_norm": 0.09753244504022564,
      "learning_rate": 8.88877808490423e-06,
      "loss": 0.1871,
      "step": 21100
    },
    {
      "epoch": 0.6522024577311943,
      "grad_norm": 0.08982957125297955,
      "learning_rate": 8.874785058517785e-06,
      "loss": 0.1998,
      "step": 21110
    },
    {
      "epoch": 0.6525114119982389,
      "grad_norm": 0.09968162051710036,
      "learning_rate": 8.860798426347239e-06,
      "loss": 0.1823,
      "step": 21120
    },
    {
      "epoch": 0.6528203662652836,
      "grad_norm": 0.09544534951209893,
      "learning_rate": 8.846818202993474e-06,
      "loss": 0.1848,
      "step": 21130
    },
    {
      "epoch": 0.6531293205323282,
      "grad_norm": 0.08866250879940554,
      "learning_rate": 8.832844403050685e-06,
      "loss": 0.1845,
      "step": 21140
    },
    {
      "epoch": 0.6534382747993728,
      "grad_norm": 0.09520192185065951,
      "learning_rate": 8.818877041106362e-06,
      "loss": 0.1828,
      "step": 21150
    },
    {
      "epoch": 0.6537472290664175,
      "grad_norm": 0.10104510627259106,
      "learning_rate": 8.804916131741258e-06,
      "loss": 0.1856,
      "step": 21160
    },
    {
      "epoch": 0.654056183333462,
      "grad_norm": 0.08513641643427416,
      "learning_rate": 8.790961689529421e-06,
      "loss": 0.1889,
      "step": 21170
    },
    {
      "epoch": 0.6543651376005066,
      "grad_norm": 0.09525142227450835,
      "learning_rate": 8.777013729038117e-06,
      "loss": 0.1893,
      "step": 21180
    },
    {
      "epoch": 0.6546740918675513,
      "grad_norm": 0.0934711455148004,
      "learning_rate": 8.763072264827866e-06,
      "loss": 0.1844,
      "step": 21190
    },
    {
      "epoch": 0.6549830461345959,
      "grad_norm": 0.11279023065894526,
      "learning_rate": 8.749137311452397e-06,
      "loss": 0.1857,
      "step": 21200
    },
    {
      "epoch": 0.6552920004016406,
      "grad_norm": 0.09846335887645649,
      "learning_rate": 8.735208883458648e-06,
      "loss": 0.1903,
      "step": 21210
    },
    {
      "epoch": 0.6556009546686852,
      "grad_norm": 0.09183429171380303,
      "learning_rate": 8.721286995386732e-06,
      "loss": 0.1829,
      "step": 21220
    },
    {
      "epoch": 0.6559099089357298,
      "grad_norm": 0.0903583720516295,
      "learning_rate": 8.707371661769953e-06,
      "loss": 0.1886,
      "step": 21230
    },
    {
      "epoch": 0.6562188632027744,
      "grad_norm": 0.10429231801413788,
      "learning_rate": 8.693462897134765e-06,
      "loss": 0.1859,
      "step": 21240
    },
    {
      "epoch": 0.656527817469819,
      "grad_norm": 0.08954233179434426,
      "learning_rate": 8.679560716000753e-06,
      "loss": 0.1826,
      "step": 21250
    },
    {
      "epoch": 0.6568367717368636,
      "grad_norm": 0.09577332953752825,
      "learning_rate": 8.665665132880652e-06,
      "loss": 0.1813,
      "step": 21260
    },
    {
      "epoch": 0.6571457260039083,
      "grad_norm": 0.08073918206578158,
      "learning_rate": 8.651776162280288e-06,
      "loss": 0.1864,
      "step": 21270
    },
    {
      "epoch": 0.6574546802709529,
      "grad_norm": 0.11017800688296119,
      "learning_rate": 8.637893818698602e-06,
      "loss": 0.185,
      "step": 21280
    },
    {
      "epoch": 0.6577636345379975,
      "grad_norm": 0.08123014275827171,
      "learning_rate": 8.624018116627597e-06,
      "loss": 0.1879,
      "step": 21290
    },
    {
      "epoch": 0.6580725888050422,
      "grad_norm": 0.12589748507751666,
      "learning_rate": 8.61014907055237e-06,
      "loss": 0.1825,
      "step": 21300
    },
    {
      "epoch": 0.6583815430720867,
      "grad_norm": 0.10644301918790272,
      "learning_rate": 8.596286694951037e-06,
      "loss": 0.186,
      "step": 21310
    },
    {
      "epoch": 0.6586904973391313,
      "grad_norm": 0.10167555722509153,
      "learning_rate": 8.582431004294781e-06,
      "loss": 0.1896,
      "step": 21320
    },
    {
      "epoch": 0.658999451606176,
      "grad_norm": 0.09356937316529804,
      "learning_rate": 8.56858201304778e-06,
      "loss": 0.1831,
      "step": 21330
    },
    {
      "epoch": 0.6593084058732206,
      "grad_norm": 0.10472028241752443,
      "learning_rate": 8.554739735667245e-06,
      "loss": 0.1885,
      "step": 21340
    },
    {
      "epoch": 0.6596173601402653,
      "grad_norm": 0.10025067675591194,
      "learning_rate": 8.540904186603351e-06,
      "loss": 0.1911,
      "step": 21350
    },
    {
      "epoch": 0.6599263144073099,
      "grad_norm": 0.11666191500386813,
      "learning_rate": 8.527075380299272e-06,
      "loss": 0.1843,
      "step": 21360
    },
    {
      "epoch": 0.6602352686743544,
      "grad_norm": 0.0813420857349242,
      "learning_rate": 8.51325333119113e-06,
      "loss": 0.1885,
      "step": 21370
    },
    {
      "epoch": 0.6605442229413991,
      "grad_norm": 0.1038112741059007,
      "learning_rate": 8.499438053707999e-06,
      "loss": 0.1859,
      "step": 21380
    },
    {
      "epoch": 0.6608531772084437,
      "grad_norm": 0.08261320957489784,
      "learning_rate": 8.485629562271876e-06,
      "loss": 0.1895,
      "step": 21390
    },
    {
      "epoch": 0.6611621314754883,
      "grad_norm": 0.11290315441437623,
      "learning_rate": 8.471827871297688e-06,
      "loss": 0.1842,
      "step": 21400
    },
    {
      "epoch": 0.661471085742533,
      "grad_norm": 0.09672724467882982,
      "learning_rate": 8.458032995193252e-06,
      "loss": 0.1814,
      "step": 21410
    },
    {
      "epoch": 0.6617800400095776,
      "grad_norm": 0.11404081140399208,
      "learning_rate": 8.444244948359272e-06,
      "loss": 0.1818,
      "step": 21420
    },
    {
      "epoch": 0.6620889942766222,
      "grad_norm": 0.09391262554098156,
      "learning_rate": 8.430463745189321e-06,
      "loss": 0.1843,
      "step": 21430
    },
    {
      "epoch": 0.6623979485436668,
      "grad_norm": 0.11187205517468783,
      "learning_rate": 8.416689400069836e-06,
      "loss": 0.1832,
      "step": 21440
    },
    {
      "epoch": 0.6627069028107114,
      "grad_norm": 0.09598670858020575,
      "learning_rate": 8.402921927380083e-06,
      "loss": 0.1821,
      "step": 21450
    },
    {
      "epoch": 0.663015857077756,
      "grad_norm": 0.09915378313178032,
      "learning_rate": 8.389161341492172e-06,
      "loss": 0.1827,
      "step": 21460
    },
    {
      "epoch": 0.6633248113448007,
      "grad_norm": 0.08341507724445604,
      "learning_rate": 8.375407656771008e-06,
      "loss": 0.1853,
      "step": 21470
    },
    {
      "epoch": 0.6636337656118453,
      "grad_norm": 0.09587779095111101,
      "learning_rate": 8.361660887574286e-06,
      "loss": 0.1841,
      "step": 21480
    },
    {
      "epoch": 0.66394271987889,
      "grad_norm": 0.12766745149581377,
      "learning_rate": 8.347921048252503e-06,
      "loss": 0.1828,
      "step": 21490
    },
    {
      "epoch": 0.6642516741459346,
      "grad_norm": 0.1031721423190585,
      "learning_rate": 8.334188153148904e-06,
      "loss": 0.1818,
      "step": 21500
    },
    {
      "epoch": 0.6645606284129791,
      "grad_norm": 0.09195452025706184,
      "learning_rate": 8.320462216599501e-06,
      "loss": 0.1983,
      "step": 21510
    },
    {
      "epoch": 0.6648695826800238,
      "grad_norm": 0.09793006585130341,
      "learning_rate": 8.30674325293302e-06,
      "loss": 0.1797,
      "step": 21520
    },
    {
      "epoch": 0.6651785369470684,
      "grad_norm": 0.10820003579466461,
      "learning_rate": 8.293031276470932e-06,
      "loss": 0.1862,
      "step": 21530
    },
    {
      "epoch": 0.665487491214113,
      "grad_norm": 0.08924766978867837,
      "learning_rate": 8.27932630152739e-06,
      "loss": 0.1867,
      "step": 21540
    },
    {
      "epoch": 0.6657964454811577,
      "grad_norm": 0.0864161019197335,
      "learning_rate": 8.265628342409263e-06,
      "loss": 0.1955,
      "step": 21550
    },
    {
      "epoch": 0.6661053997482023,
      "grad_norm": 0.0928457064161821,
      "learning_rate": 8.251937413416071e-06,
      "loss": 0.1829,
      "step": 21560
    },
    {
      "epoch": 0.6664143540152468,
      "grad_norm": 0.12068934710591628,
      "learning_rate": 8.238253528840018e-06,
      "loss": 0.1775,
      "step": 21570
    },
    {
      "epoch": 0.6667233082822915,
      "grad_norm": 0.08515900181313076,
      "learning_rate": 8.224576702965931e-06,
      "loss": 0.1827,
      "step": 21580
    },
    {
      "epoch": 0.6670322625493361,
      "grad_norm": 0.0972574825682533,
      "learning_rate": 8.210906950071296e-06,
      "loss": 0.1816,
      "step": 21590
    },
    {
      "epoch": 0.6673412168163807,
      "grad_norm": 0.11425152744846681,
      "learning_rate": 8.197244284426185e-06,
      "loss": 0.1801,
      "step": 21600
    },
    {
      "epoch": 0.6676501710834254,
      "grad_norm": 0.10850147262928118,
      "learning_rate": 8.183588720293301e-06,
      "loss": 0.1831,
      "step": 21610
    },
    {
      "epoch": 0.66795912535047,
      "grad_norm": 0.09469469601554154,
      "learning_rate": 8.169940271927906e-06,
      "loss": 0.1895,
      "step": 21620
    },
    {
      "epoch": 0.6682680796175147,
      "grad_norm": 0.096275588799572,
      "learning_rate": 8.156298953577859e-06,
      "loss": 0.182,
      "step": 21630
    },
    {
      "epoch": 0.6685770338845592,
      "grad_norm": 0.09430258563893375,
      "learning_rate": 8.142664779483558e-06,
      "loss": 0.1865,
      "step": 21640
    },
    {
      "epoch": 0.6688859881516038,
      "grad_norm": 0.09105192280482428,
      "learning_rate": 8.129037763877948e-06,
      "loss": 0.1829,
      "step": 21650
    },
    {
      "epoch": 0.6691949424186485,
      "grad_norm": 0.09771768495441543,
      "learning_rate": 8.1154179209865e-06,
      "loss": 0.1837,
      "step": 21660
    },
    {
      "epoch": 0.6695038966856931,
      "grad_norm": 0.11779791734450315,
      "learning_rate": 8.101805265027207e-06,
      "loss": 0.1856,
      "step": 21670
    },
    {
      "epoch": 0.6698128509527377,
      "grad_norm": 0.0927670959552032,
      "learning_rate": 8.088199810210542e-06,
      "loss": 0.1849,
      "step": 21680
    },
    {
      "epoch": 0.6701218052197824,
      "grad_norm": 0.10318928461551366,
      "learning_rate": 8.07460157073948e-06,
      "loss": 0.1844,
      "step": 21690
    },
    {
      "epoch": 0.670430759486827,
      "grad_norm": 0.10207803569915344,
      "learning_rate": 8.061010560809452e-06,
      "loss": 0.1814,
      "step": 21700
    },
    {
      "epoch": 0.6707397137538715,
      "grad_norm": 0.10859725615607474,
      "learning_rate": 8.047426794608333e-06,
      "loss": 0.1827,
      "step": 21710
    },
    {
      "epoch": 0.6710486680209162,
      "grad_norm": 0.11216570380804705,
      "learning_rate": 8.033850286316465e-06,
      "loss": 0.1838,
      "step": 21720
    },
    {
      "epoch": 0.6713576222879608,
      "grad_norm": 0.08581601704543383,
      "learning_rate": 8.020281050106578e-06,
      "loss": 0.1848,
      "step": 21730
    },
    {
      "epoch": 0.6716665765550054,
      "grad_norm": 0.1132973714821087,
      "learning_rate": 8.006719100143842e-06,
      "loss": 0.1885,
      "step": 21740
    },
    {
      "epoch": 0.6719755308220501,
      "grad_norm": 0.09055021607729273,
      "learning_rate": 7.993164450585797e-06,
      "loss": 0.1817,
      "step": 21750
    },
    {
      "epoch": 0.6722844850890947,
      "grad_norm": 0.11510073927318624,
      "learning_rate": 7.979617115582381e-06,
      "loss": 0.1852,
      "step": 21760
    },
    {
      "epoch": 0.6725934393561394,
      "grad_norm": 0.09656673805707162,
      "learning_rate": 7.966077109275877e-06,
      "loss": 0.1824,
      "step": 21770
    },
    {
      "epoch": 0.6729023936231839,
      "grad_norm": 0.09612752042169377,
      "learning_rate": 7.95254444580094e-06,
      "loss": 0.1829,
      "step": 21780
    },
    {
      "epoch": 0.6732113478902285,
      "grad_norm": 0.09804574768999724,
      "learning_rate": 7.939019139284533e-06,
      "loss": 0.1903,
      "step": 21790
    },
    {
      "epoch": 0.6735203021572732,
      "grad_norm": 0.1015705438595519,
      "learning_rate": 7.925501203845967e-06,
      "loss": 0.1852,
      "step": 21800
    },
    {
      "epoch": 0.6738292564243178,
      "grad_norm": 0.10521792775158305,
      "learning_rate": 7.91199065359683e-06,
      "loss": 0.1854,
      "step": 21810
    },
    {
      "epoch": 0.6741382106913624,
      "grad_norm": 0.09240116148105681,
      "learning_rate": 7.898487502641032e-06,
      "loss": 0.184,
      "step": 21820
    },
    {
      "epoch": 0.6744471649584071,
      "grad_norm": 0.09456845391420433,
      "learning_rate": 7.884991765074722e-06,
      "loss": 0.1818,
      "step": 21830
    },
    {
      "epoch": 0.6747561192254516,
      "grad_norm": 0.08608038992547533,
      "learning_rate": 7.87150345498635e-06,
      "loss": 0.1832,
      "step": 21840
    },
    {
      "epoch": 0.6750650734924962,
      "grad_norm": 0.09980873652774314,
      "learning_rate": 7.85802258645658e-06,
      "loss": 0.1819,
      "step": 21850
    },
    {
      "epoch": 0.6753740277595409,
      "grad_norm": 0.10895283523915741,
      "learning_rate": 7.844549173558315e-06,
      "loss": 0.1871,
      "step": 21860
    },
    {
      "epoch": 0.6756829820265855,
      "grad_norm": 0.09950732048600308,
      "learning_rate": 7.83108323035669e-06,
      "loss": 0.1804,
      "step": 21870
    },
    {
      "epoch": 0.6759919362936301,
      "grad_norm": 0.1192050036517003,
      "learning_rate": 7.81762477090903e-06,
      "loss": 0.1811,
      "step": 21880
    },
    {
      "epoch": 0.6763008905606748,
      "grad_norm": 0.10769897254606019,
      "learning_rate": 7.804173809264841e-06,
      "loss": 0.1817,
      "step": 21890
    },
    {
      "epoch": 0.6766098448277194,
      "grad_norm": 0.10617910560379765,
      "learning_rate": 7.790730359465823e-06,
      "loss": 0.1803,
      "step": 21900
    },
    {
      "epoch": 0.676918799094764,
      "grad_norm": 0.08724439637965004,
      "learning_rate": 7.777294435545808e-06,
      "loss": 0.1789,
      "step": 21910
    },
    {
      "epoch": 0.6772277533618086,
      "grad_norm": 0.11793380282645627,
      "learning_rate": 7.763866051530798e-06,
      "loss": 0.184,
      "step": 21920
    },
    {
      "epoch": 0.6775367076288532,
      "grad_norm": 0.09792740654299632,
      "learning_rate": 7.750445221438906e-06,
      "loss": 0.1849,
      "step": 21930
    },
    {
      "epoch": 0.6778456618958979,
      "grad_norm": 0.09669760023692602,
      "learning_rate": 7.737031959280356e-06,
      "loss": 0.1815,
      "step": 21940
    },
    {
      "epoch": 0.6781546161629425,
      "grad_norm": 0.09117546342880395,
      "learning_rate": 7.723626279057496e-06,
      "loss": 0.1802,
      "step": 21950
    },
    {
      "epoch": 0.6784635704299871,
      "grad_norm": 0.11057930039309655,
      "learning_rate": 7.710228194764727e-06,
      "loss": 0.1938,
      "step": 21960
    },
    {
      "epoch": 0.6787725246970318,
      "grad_norm": 0.103987368803885,
      "learning_rate": 7.696837720388554e-06,
      "loss": 0.1796,
      "step": 21970
    },
    {
      "epoch": 0.6790814789640763,
      "grad_norm": 0.08877131667329387,
      "learning_rate": 7.6834548699075e-06,
      "loss": 0.184,
      "step": 21980
    },
    {
      "epoch": 0.6793904332311209,
      "grad_norm": 0.08602755979863845,
      "learning_rate": 7.67007965729217e-06,
      "loss": 0.192,
      "step": 21990
    },
    {
      "epoch": 0.6796993874981656,
      "grad_norm": 0.08188259413918099,
      "learning_rate": 7.656712096505164e-06,
      "loss": 0.1907,
      "step": 22000
    },
    {
      "epoch": 0.6800083417652102,
      "grad_norm": 0.08245297029447374,
      "learning_rate": 7.64335220150111e-06,
      "loss": 0.1842,
      "step": 22010
    },
    {
      "epoch": 0.6803172960322548,
      "grad_norm": 0.08354299187463678,
      "learning_rate": 7.629999986226626e-06,
      "loss": 0.1811,
      "step": 22020
    },
    {
      "epoch": 0.6806262502992995,
      "grad_norm": 0.14769271108737972,
      "learning_rate": 7.616655464620321e-06,
      "loss": 0.1895,
      "step": 22030
    },
    {
      "epoch": 0.680935204566344,
      "grad_norm": 0.09456168217757971,
      "learning_rate": 7.60331865061276e-06,
      "loss": 0.1796,
      "step": 22040
    },
    {
      "epoch": 0.6812441588333887,
      "grad_norm": 0.0878522299972071,
      "learning_rate": 7.589989558126481e-06,
      "loss": 0.1834,
      "step": 22050
    },
    {
      "epoch": 0.6815531131004333,
      "grad_norm": 0.10993209787891756,
      "learning_rate": 7.576668201075936e-06,
      "loss": 0.1896,
      "step": 22060
    },
    {
      "epoch": 0.6818620673674779,
      "grad_norm": 0.09166496995708409,
      "learning_rate": 7.56335459336753e-06,
      "loss": 0.181,
      "step": 22070
    },
    {
      "epoch": 0.6821710216345226,
      "grad_norm": 0.09855574804314984,
      "learning_rate": 7.550048748899558e-06,
      "loss": 0.1783,
      "step": 22080
    },
    {
      "epoch": 0.6824799759015672,
      "grad_norm": 0.09649936095243887,
      "learning_rate": 7.536750681562211e-06,
      "loss": 0.1858,
      "step": 22090
    },
    {
      "epoch": 0.6827889301686118,
      "grad_norm": 0.09165491513749559,
      "learning_rate": 7.523460405237579e-06,
      "loss": 0.1875,
      "step": 22100
    },
    {
      "epoch": 0.6830978844356564,
      "grad_norm": 0.07787733586066546,
      "learning_rate": 7.5101779337996e-06,
      "loss": 0.1814,
      "step": 22110
    },
    {
      "epoch": 0.683406838702701,
      "grad_norm": 0.16490718623260733,
      "learning_rate": 7.496903281114067e-06,
      "loss": 0.2018,
      "step": 22120
    },
    {
      "epoch": 0.6837157929697456,
      "grad_norm": 0.09403627623821387,
      "learning_rate": 7.483636461038628e-06,
      "loss": 0.1811,
      "step": 22130
    },
    {
      "epoch": 0.6840247472367903,
      "grad_norm": 0.09960587837256916,
      "learning_rate": 7.470377487422727e-06,
      "loss": 0.1811,
      "step": 22140
    },
    {
      "epoch": 0.6843337015038349,
      "grad_norm": 0.10772946164623394,
      "learning_rate": 7.457126374107646e-06,
      "loss": 0.1824,
      "step": 22150
    },
    {
      "epoch": 0.6846426557708795,
      "grad_norm": 0.09580612300921638,
      "learning_rate": 7.4438831349264415e-06,
      "loss": 0.1792,
      "step": 22160
    },
    {
      "epoch": 0.6849516100379242,
      "grad_norm": 0.09591036425310501,
      "learning_rate": 7.430647783703949e-06,
      "loss": 0.185,
      "step": 22170
    },
    {
      "epoch": 0.6852605643049687,
      "grad_norm": 0.08932953414207497,
      "learning_rate": 7.417420334256787e-06,
      "loss": 0.1834,
      "step": 22180
    },
    {
      "epoch": 0.6855695185720134,
      "grad_norm": 0.0992893464161745,
      "learning_rate": 7.404200800393305e-06,
      "loss": 0.1832,
      "step": 22190
    },
    {
      "epoch": 0.685878472839058,
      "grad_norm": 0.10052745564077986,
      "learning_rate": 7.39098919591361e-06,
      "loss": 0.1877,
      "step": 22200
    },
    {
      "epoch": 0.6861874271061026,
      "grad_norm": 0.1131068532052622,
      "learning_rate": 7.377785534609509e-06,
      "loss": 0.181,
      "step": 22210
    },
    {
      "epoch": 0.6864963813731473,
      "grad_norm": 0.10094189745848517,
      "learning_rate": 7.364589830264537e-06,
      "loss": 0.1816,
      "step": 22220
    },
    {
      "epoch": 0.6868053356401919,
      "grad_norm": 0.09653480119282912,
      "learning_rate": 7.3514020966539065e-06,
      "loss": 0.1815,
      "step": 22230
    },
    {
      "epoch": 0.6871142899072364,
      "grad_norm": 0.0993360220318092,
      "learning_rate": 7.338222347544526e-06,
      "loss": 0.1823,
      "step": 22240
    },
    {
      "epoch": 0.6874232441742811,
      "grad_norm": 0.08422552040154874,
      "learning_rate": 7.325050596694947e-06,
      "loss": 0.1866,
      "step": 22250
    },
    {
      "epoch": 0.6877321984413257,
      "grad_norm": 0.10108479744433081,
      "learning_rate": 7.311886857855398e-06,
      "loss": 0.1833,
      "step": 22260
    },
    {
      "epoch": 0.6880411527083703,
      "grad_norm": 0.1026968384760547,
      "learning_rate": 7.2987311447677155e-06,
      "loss": 0.1849,
      "step": 22270
    },
    {
      "epoch": 0.688350106975415,
      "grad_norm": 0.10249702981191872,
      "learning_rate": 7.285583471165385e-06,
      "loss": 0.187,
      "step": 22280
    },
    {
      "epoch": 0.6886590612424596,
      "grad_norm": 0.11137966018067741,
      "learning_rate": 7.272443850773474e-06,
      "loss": 0.1865,
      "step": 22290
    },
    {
      "epoch": 0.6889680155095042,
      "grad_norm": 0.10304373007231557,
      "learning_rate": 7.259312297308665e-06,
      "loss": 0.1763,
      "step": 22300
    },
    {
      "epoch": 0.6892769697765488,
      "grad_norm": 0.09274915547043855,
      "learning_rate": 7.2461888244792075e-06,
      "loss": 0.1797,
      "step": 22310
    },
    {
      "epoch": 0.6895859240435934,
      "grad_norm": 0.10089718809079859,
      "learning_rate": 7.233073445984917e-06,
      "loss": 0.1882,
      "step": 22320
    },
    {
      "epoch": 0.6898948783106381,
      "grad_norm": 0.10824528144194742,
      "learning_rate": 7.219966175517151e-06,
      "loss": 0.1842,
      "step": 22330
    },
    {
      "epoch": 0.6902038325776827,
      "grad_norm": 0.10557538951085883,
      "learning_rate": 7.206867026758828e-06,
      "loss": 0.1861,
      "step": 22340
    },
    {
      "epoch": 0.6905127868447273,
      "grad_norm": 0.12675439247659584,
      "learning_rate": 7.193776013384355e-06,
      "loss": 0.1815,
      "step": 22350
    },
    {
      "epoch": 0.690821741111772,
      "grad_norm": 0.093012673616093,
      "learning_rate": 7.180693149059679e-06,
      "loss": 0.1811,
      "step": 22360
    },
    {
      "epoch": 0.6911306953788166,
      "grad_norm": 0.11022055746496426,
      "learning_rate": 7.167618447442212e-06,
      "loss": 0.1836,
      "step": 22370
    },
    {
      "epoch": 0.6914396496458611,
      "grad_norm": 0.10088552228914249,
      "learning_rate": 7.154551922180864e-06,
      "loss": 0.1831,
      "step": 22380
    },
    {
      "epoch": 0.6917486039129058,
      "grad_norm": 0.07829621733751456,
      "learning_rate": 7.141493586916004e-06,
      "loss": 0.1826,
      "step": 22390
    },
    {
      "epoch": 0.6920575581799504,
      "grad_norm": 0.0889419944283404,
      "learning_rate": 7.12844345527944e-06,
      "loss": 0.1835,
      "step": 22400
    },
    {
      "epoch": 0.692366512446995,
      "grad_norm": 0.10599624941114881,
      "learning_rate": 7.115401540894437e-06,
      "loss": 0.1912,
      "step": 22410
    },
    {
      "epoch": 0.6926754667140397,
      "grad_norm": 0.10729721552355075,
      "learning_rate": 7.102367857375663e-06,
      "loss": 0.208,
      "step": 22420
    },
    {
      "epoch": 0.6929844209810843,
      "grad_norm": 0.1002851903157931,
      "learning_rate": 7.08934241832921e-06,
      "loss": 0.1844,
      "step": 22430
    },
    {
      "epoch": 0.6932933752481288,
      "grad_norm": 0.10067248745974343,
      "learning_rate": 7.076325237352543e-06,
      "loss": 0.1783,
      "step": 22440
    },
    {
      "epoch": 0.6936023295151735,
      "grad_norm": 0.09227533535580679,
      "learning_rate": 7.063316328034533e-06,
      "loss": 0.1823,
      "step": 22450
    },
    {
      "epoch": 0.6939112837822181,
      "grad_norm": 0.13208989858189216,
      "learning_rate": 7.050315703955386e-06,
      "loss": 0.1803,
      "step": 22460
    },
    {
      "epoch": 0.6942202380492628,
      "grad_norm": 0.11043560756916097,
      "learning_rate": 7.03732337868669e-06,
      "loss": 0.1814,
      "step": 22470
    },
    {
      "epoch": 0.6945291923163074,
      "grad_norm": 0.11219602108245105,
      "learning_rate": 7.024339365791338e-06,
      "loss": 0.1871,
      "step": 22480
    },
    {
      "epoch": 0.694838146583352,
      "grad_norm": 0.0909153194407292,
      "learning_rate": 7.011363678823577e-06,
      "loss": 0.1898,
      "step": 22490
    },
    {
      "epoch": 0.6951471008503967,
      "grad_norm": 0.0930940860533655,
      "learning_rate": 6.998396331328933e-06,
      "loss": 0.1844,
      "step": 22500
    },
    {
      "epoch": 0.6954560551174412,
      "grad_norm": 0.09377328331032149,
      "learning_rate": 6.985437336844253e-06,
      "loss": 0.1813,
      "step": 22510
    },
    {
      "epoch": 0.6957650093844858,
      "grad_norm": 0.10876888713001147,
      "learning_rate": 6.972486708897638e-06,
      "loss": 0.1797,
      "step": 22520
    },
    {
      "epoch": 0.6960739636515305,
      "grad_norm": 0.09387171624159234,
      "learning_rate": 6.959544461008483e-06,
      "loss": 0.1814,
      "step": 22530
    },
    {
      "epoch": 0.6963829179185751,
      "grad_norm": 0.13612121256321463,
      "learning_rate": 6.9466106066874115e-06,
      "loss": 0.2073,
      "step": 22540
    },
    {
      "epoch": 0.6966918721856197,
      "grad_norm": 0.09290309901674741,
      "learning_rate": 6.933685159436296e-06,
      "loss": 0.183,
      "step": 22550
    },
    {
      "epoch": 0.6970008264526644,
      "grad_norm": 0.0951727569438628,
      "learning_rate": 6.920768132748224e-06,
      "loss": 0.1825,
      "step": 22560
    },
    {
      "epoch": 0.697309780719709,
      "grad_norm": 0.09333447460953703,
      "learning_rate": 6.90785954010751e-06,
      "loss": 0.1785,
      "step": 22570
    },
    {
      "epoch": 0.6976187349867535,
      "grad_norm": 0.09795160090038436,
      "learning_rate": 6.894959394989643e-06,
      "loss": 0.1797,
      "step": 22580
    },
    {
      "epoch": 0.6979276892537982,
      "grad_norm": 0.1005629358636702,
      "learning_rate": 6.882067710861313e-06,
      "loss": 0.1782,
      "step": 22590
    },
    {
      "epoch": 0.6982366435208428,
      "grad_norm": 0.10495606872393157,
      "learning_rate": 6.8691845011803585e-06,
      "loss": 0.181,
      "step": 22600
    },
    {
      "epoch": 0.6985455977878875,
      "grad_norm": 0.09891331134426463,
      "learning_rate": 6.85630977939579e-06,
      "loss": 0.1837,
      "step": 22610
    },
    {
      "epoch": 0.6988545520549321,
      "grad_norm": 0.09608712038311525,
      "learning_rate": 6.843443558947744e-06,
      "loss": 0.1826,
      "step": 22620
    },
    {
      "epoch": 0.6991635063219767,
      "grad_norm": 0.10823888714310015,
      "learning_rate": 6.8305858532674825e-06,
      "loss": 0.1812,
      "step": 22630
    },
    {
      "epoch": 0.6994724605890213,
      "grad_norm": 0.11483176949334299,
      "learning_rate": 6.81773667577739e-06,
      "loss": 0.1863,
      "step": 22640
    },
    {
      "epoch": 0.6997814148560659,
      "grad_norm": 0.12046064900719501,
      "learning_rate": 6.804896039890933e-06,
      "loss": 0.1841,
      "step": 22650
    },
    {
      "epoch": 0.7000903691231105,
      "grad_norm": 0.10013959191086173,
      "learning_rate": 6.792063959012681e-06,
      "loss": 0.1832,
      "step": 22660
    },
    {
      "epoch": 0.7003993233901552,
      "grad_norm": 0.09003294217810345,
      "learning_rate": 6.779240446538248e-06,
      "loss": 0.1863,
      "step": 22670
    },
    {
      "epoch": 0.7007082776571998,
      "grad_norm": 0.12499293314608835,
      "learning_rate": 6.766425515854328e-06,
      "loss": 0.1812,
      "step": 22680
    },
    {
      "epoch": 0.7010172319242444,
      "grad_norm": 0.14008647305253966,
      "learning_rate": 6.753619180338634e-06,
      "loss": 0.1767,
      "step": 22690
    },
    {
      "epoch": 0.7013261861912891,
      "grad_norm": 0.10941872997400424,
      "learning_rate": 6.740821453359928e-06,
      "loss": 0.1826,
      "step": 22700
    },
    {
      "epoch": 0.7016351404583336,
      "grad_norm": 0.10831878593425659,
      "learning_rate": 6.728032348277963e-06,
      "loss": 0.1889,
      "step": 22710
    },
    {
      "epoch": 0.7019440947253783,
      "grad_norm": 0.0946684603744815,
      "learning_rate": 6.7152518784435175e-06,
      "loss": 0.1849,
      "step": 22720
    },
    {
      "epoch": 0.7022530489924229,
      "grad_norm": 0.10259548829572136,
      "learning_rate": 6.702480057198325e-06,
      "loss": 0.1898,
      "step": 22730
    },
    {
      "epoch": 0.7025620032594675,
      "grad_norm": 0.10224576886689879,
      "learning_rate": 6.68971689787512e-06,
      "loss": 0.1845,
      "step": 22740
    },
    {
      "epoch": 0.7028709575265122,
      "grad_norm": 0.09881312799382255,
      "learning_rate": 6.676962413797575e-06,
      "loss": 0.1811,
      "step": 22750
    },
    {
      "epoch": 0.7031799117935568,
      "grad_norm": 0.0934974541588881,
      "learning_rate": 6.664216618280309e-06,
      "loss": 0.1873,
      "step": 22760
    },
    {
      "epoch": 0.7034888660606013,
      "grad_norm": 0.09846852501265885,
      "learning_rate": 6.6514795246288834e-06,
      "loss": 0.1858,
      "step": 22770
    },
    {
      "epoch": 0.703797820327646,
      "grad_norm": 0.0810910375668225,
      "learning_rate": 6.638751146139762e-06,
      "loss": 0.1865,
      "step": 22780
    },
    {
      "epoch": 0.7041067745946906,
      "grad_norm": 0.10243833744847768,
      "learning_rate": 6.626031496100308e-06,
      "loss": 0.1903,
      "step": 22790
    },
    {
      "epoch": 0.7044157288617352,
      "grad_norm": 0.10522763872225695,
      "learning_rate": 6.613320587788794e-06,
      "loss": 0.1799,
      "step": 22800
    },
    {
      "epoch": 0.7047246831287799,
      "grad_norm": 0.08690977005698929,
      "learning_rate": 6.600618434474339e-06,
      "loss": 0.1789,
      "step": 22810
    },
    {
      "epoch": 0.7050336373958245,
      "grad_norm": 0.11267023576935571,
      "learning_rate": 6.587925049416951e-06,
      "loss": 0.1788,
      "step": 22820
    },
    {
      "epoch": 0.7053425916628691,
      "grad_norm": 0.1401753366454194,
      "learning_rate": 6.575240445867455e-06,
      "loss": 0.194,
      "step": 22830
    },
    {
      "epoch": 0.7056515459299137,
      "grad_norm": 0.11646166361342897,
      "learning_rate": 6.56256463706754e-06,
      "loss": 0.1848,
      "step": 22840
    },
    {
      "epoch": 0.7059605001969583,
      "grad_norm": 0.08855345720613042,
      "learning_rate": 6.5498976362496884e-06,
      "loss": 0.1844,
      "step": 22850
    },
    {
      "epoch": 0.706269454464003,
      "grad_norm": 0.09725128694436247,
      "learning_rate": 6.537239456637193e-06,
      "loss": 0.1878,
      "step": 22860
    },
    {
      "epoch": 0.7065784087310476,
      "grad_norm": 0.08664039348579067,
      "learning_rate": 6.524590111444156e-06,
      "loss": 0.1843,
      "step": 22870
    },
    {
      "epoch": 0.7068873629980922,
      "grad_norm": 0.11022183963189743,
      "learning_rate": 6.5119496138754295e-06,
      "loss": 0.1834,
      "step": 22880
    },
    {
      "epoch": 0.7071963172651369,
      "grad_norm": 0.0818328358931418,
      "learning_rate": 6.499317977126656e-06,
      "loss": 0.1792,
      "step": 22890
    },
    {
      "epoch": 0.7075052715321815,
      "grad_norm": 0.09874177970560724,
      "learning_rate": 6.486695214384205e-06,
      "loss": 0.1843,
      "step": 22900
    },
    {
      "epoch": 0.707814225799226,
      "grad_norm": 0.11297624589105548,
      "learning_rate": 6.474081338825202e-06,
      "loss": 0.1844,
      "step": 22910
    },
    {
      "epoch": 0.7081231800662707,
      "grad_norm": 0.10243376858465415,
      "learning_rate": 6.461476363617477e-06,
      "loss": 0.1882,
      "step": 22920
    },
    {
      "epoch": 0.7084321343333153,
      "grad_norm": 0.0840623172311264,
      "learning_rate": 6.448880301919589e-06,
      "loss": 0.1814,
      "step": 22930
    },
    {
      "epoch": 0.7087410886003599,
      "grad_norm": 0.10230810556633248,
      "learning_rate": 6.436293166880767e-06,
      "loss": 0.1894,
      "step": 22940
    },
    {
      "epoch": 0.7090500428674046,
      "grad_norm": 0.09772364321195781,
      "learning_rate": 6.4237149716409475e-06,
      "loss": 0.1839,
      "step": 22950
    },
    {
      "epoch": 0.7093589971344492,
      "grad_norm": 0.11072981532651247,
      "learning_rate": 6.411145729330711e-06,
      "loss": 0.1836,
      "step": 22960
    },
    {
      "epoch": 0.7096679514014937,
      "grad_norm": 0.09011194193047975,
      "learning_rate": 6.398585453071316e-06,
      "loss": 0.179,
      "step": 22970
    },
    {
      "epoch": 0.7099769056685384,
      "grad_norm": 0.10363318374429248,
      "learning_rate": 6.38603415597464e-06,
      "loss": 0.1818,
      "step": 22980
    },
    {
      "epoch": 0.710285859935583,
      "grad_norm": 0.08146492866877648,
      "learning_rate": 6.37349185114319e-06,
      "loss": 0.1767,
      "step": 22990
    },
    {
      "epoch": 0.7105948142026277,
      "grad_norm": 0.09431933489269974,
      "learning_rate": 6.360958551670104e-06,
      "loss": 0.1907,
      "step": 23000
    },
    {
      "epoch": 0.7109037684696723,
      "grad_norm": 0.11048345189042445,
      "learning_rate": 6.348434270639098e-06,
      "loss": 0.1845,
      "step": 23010
    },
    {
      "epoch": 0.7112127227367169,
      "grad_norm": 0.13106656928455201,
      "learning_rate": 6.335919021124478e-06,
      "loss": 0.1916,
      "step": 23020
    },
    {
      "epoch": 0.7115216770037616,
      "grad_norm": 0.0980751235847233,
      "learning_rate": 6.3234128161911374e-06,
      "loss": 0.1833,
      "step": 23030
    },
    {
      "epoch": 0.7118306312708061,
      "grad_norm": 0.10255968816851609,
      "learning_rate": 6.310915668894501e-06,
      "loss": 0.1824,
      "step": 23040
    },
    {
      "epoch": 0.7121395855378507,
      "grad_norm": 0.08728828922083484,
      "learning_rate": 6.298427592280569e-06,
      "loss": 0.1814,
      "step": 23050
    },
    {
      "epoch": 0.7124485398048954,
      "grad_norm": 0.10542018314338772,
      "learning_rate": 6.285948599385842e-06,
      "loss": 0.1835,
      "step": 23060
    },
    {
      "epoch": 0.71275749407194,
      "grad_norm": 0.08436415370702952,
      "learning_rate": 6.273478703237365e-06,
      "loss": 0.1858,
      "step": 23070
    },
    {
      "epoch": 0.7130664483389846,
      "grad_norm": 0.1022259595350936,
      "learning_rate": 6.261017916852668e-06,
      "loss": 0.1762,
      "step": 23080
    },
    {
      "epoch": 0.7133754026060293,
      "grad_norm": 0.11046728828954396,
      "learning_rate": 6.248566253239774e-06,
      "loss": 0.1834,
      "step": 23090
    },
    {
      "epoch": 0.7136843568730739,
      "grad_norm": 0.08697901231904889,
      "learning_rate": 6.236123725397196e-06,
      "loss": 0.1835,
      "step": 23100
    },
    {
      "epoch": 0.7139933111401184,
      "grad_norm": 0.09245674162671554,
      "learning_rate": 6.223690346313888e-06,
      "loss": 0.1809,
      "step": 23110
    },
    {
      "epoch": 0.7143022654071631,
      "grad_norm": 0.1312160942879207,
      "learning_rate": 6.211266128969278e-06,
      "loss": 0.1826,
      "step": 23120
    },
    {
      "epoch": 0.7146112196742077,
      "grad_norm": 0.0926064562348625,
      "learning_rate": 6.198851086333206e-06,
      "loss": 0.1834,
      "step": 23130
    },
    {
      "epoch": 0.7149201739412524,
      "grad_norm": 0.10756064976094376,
      "learning_rate": 6.186445231365957e-06,
      "loss": 0.1824,
      "step": 23140
    },
    {
      "epoch": 0.715229128208297,
      "grad_norm": 0.11918399401915028,
      "learning_rate": 6.174048577018203e-06,
      "loss": 0.1857,
      "step": 23150
    },
    {
      "epoch": 0.7155380824753416,
      "grad_norm": 0.10214871420915306,
      "learning_rate": 6.161661136231031e-06,
      "loss": 0.1808,
      "step": 23160
    },
    {
      "epoch": 0.7158470367423863,
      "grad_norm": 0.09120118719203635,
      "learning_rate": 6.149282921935893e-06,
      "loss": 0.1791,
      "step": 23170
    },
    {
      "epoch": 0.7161559910094308,
      "grad_norm": 0.10105557237813377,
      "learning_rate": 6.136913947054626e-06,
      "loss": 0.1818,
      "step": 23180
    },
    {
      "epoch": 0.7164649452764754,
      "grad_norm": 0.08502730000820673,
      "learning_rate": 6.124554224499403e-06,
      "loss": 0.1851,
      "step": 23190
    },
    {
      "epoch": 0.7167738995435201,
      "grad_norm": 0.10244948066698438,
      "learning_rate": 6.112203767172758e-06,
      "loss": 0.1851,
      "step": 23200
    },
    {
      "epoch": 0.7170828538105647,
      "grad_norm": 0.09565504810996643,
      "learning_rate": 6.099862587967538e-06,
      "loss": 0.1877,
      "step": 23210
    },
    {
      "epoch": 0.7173918080776093,
      "grad_norm": 0.09854345398680876,
      "learning_rate": 6.087530699766906e-06,
      "loss": 0.1788,
      "step": 23220
    },
    {
      "epoch": 0.717700762344654,
      "grad_norm": 0.11502196801484005,
      "learning_rate": 6.075208115444337e-06,
      "loss": 0.1885,
      "step": 23230
    },
    {
      "epoch": 0.7180097166116985,
      "grad_norm": 0.09624064877420897,
      "learning_rate": 6.062894847863584e-06,
      "loss": 0.1927,
      "step": 23240
    },
    {
      "epoch": 0.7183186708787431,
      "grad_norm": 0.11049030334103088,
      "learning_rate": 6.050590909878668e-06,
      "loss": 0.1872,
      "step": 23250
    },
    {
      "epoch": 0.7186276251457878,
      "grad_norm": 0.08633503283061199,
      "learning_rate": 6.038296314333891e-06,
      "loss": 0.1787,
      "step": 23260
    },
    {
      "epoch": 0.7189365794128324,
      "grad_norm": 0.09875871867168223,
      "learning_rate": 6.0260110740637804e-06,
      "loss": 0.185,
      "step": 23270
    },
    {
      "epoch": 0.7192455336798771,
      "grad_norm": 0.08476965760872901,
      "learning_rate": 6.013735201893116e-06,
      "loss": 0.1803,
      "step": 23280
    },
    {
      "epoch": 0.7195544879469217,
      "grad_norm": 0.10772440899999361,
      "learning_rate": 6.001468710636878e-06,
      "loss": 0.1864,
      "step": 23290
    },
    {
      "epoch": 0.7198634422139663,
      "grad_norm": 0.09776798378035491,
      "learning_rate": 5.989211613100279e-06,
      "loss": 0.1822,
      "step": 23300
    },
    {
      "epoch": 0.720172396481011,
      "grad_norm": 0.09050535034739404,
      "learning_rate": 5.976963922078703e-06,
      "loss": 0.1835,
      "step": 23310
    },
    {
      "epoch": 0.7204813507480555,
      "grad_norm": 0.10821419772111825,
      "learning_rate": 5.964725650357717e-06,
      "loss": 0.1899,
      "step": 23320
    },
    {
      "epoch": 0.7207903050151001,
      "grad_norm": 0.0870279225799357,
      "learning_rate": 5.952496810713073e-06,
      "loss": 0.1801,
      "step": 23330
    },
    {
      "epoch": 0.7210992592821448,
      "grad_norm": 0.09399342395698655,
      "learning_rate": 5.940277415910654e-06,
      "loss": 0.1838,
      "step": 23340
    },
    {
      "epoch": 0.7214082135491894,
      "grad_norm": 0.09643234650439193,
      "learning_rate": 5.928067478706503e-06,
      "loss": 0.1809,
      "step": 23350
    },
    {
      "epoch": 0.721717167816234,
      "grad_norm": 0.09629637524672571,
      "learning_rate": 5.91586701184677e-06,
      "loss": 0.18,
      "step": 23360
    },
    {
      "epoch": 0.7220261220832787,
      "grad_norm": 0.09717067559422377,
      "learning_rate": 5.9036760280677395e-06,
      "loss": 0.1828,
      "step": 23370
    },
    {
      "epoch": 0.7223350763503232,
      "grad_norm": 0.09057261004876756,
      "learning_rate": 5.89149454009578e-06,
      "loss": 0.1916,
      "step": 23380
    },
    {
      "epoch": 0.7226440306173678,
      "grad_norm": 0.1301195912918232,
      "learning_rate": 5.879322560647361e-06,
      "loss": 0.1836,
      "step": 23390
    },
    {
      "epoch": 0.7229529848844125,
      "grad_norm": 0.09215293888878717,
      "learning_rate": 5.867160102429008e-06,
      "loss": 0.1855,
      "step": 23400
    },
    {
      "epoch": 0.7232619391514571,
      "grad_norm": 0.11605752935290935,
      "learning_rate": 5.85500717813733e-06,
      "loss": 0.1834,
      "step": 23410
    },
    {
      "epoch": 0.7235708934185018,
      "grad_norm": 0.09248674661172235,
      "learning_rate": 5.8428638004589576e-06,
      "loss": 0.1767,
      "step": 23420
    },
    {
      "epoch": 0.7238798476855464,
      "grad_norm": 0.1236332915964266,
      "learning_rate": 5.830729982070582e-06,
      "loss": 0.1817,
      "step": 23430
    },
    {
      "epoch": 0.7241888019525909,
      "grad_norm": 0.08963978029874643,
      "learning_rate": 5.818605735638895e-06,
      "loss": 0.1822,
      "step": 23440
    },
    {
      "epoch": 0.7244977562196356,
      "grad_norm": 0.08748669932058856,
      "learning_rate": 5.806491073820604e-06,
      "loss": 0.1778,
      "step": 23450
    },
    {
      "epoch": 0.7248067104866802,
      "grad_norm": 0.10141517797103793,
      "learning_rate": 5.794386009262408e-06,
      "loss": 0.1901,
      "step": 23460
    },
    {
      "epoch": 0.7251156647537248,
      "grad_norm": 0.09028198862404607,
      "learning_rate": 5.782290554600997e-06,
      "loss": 0.1842,
      "step": 23470
    },
    {
      "epoch": 0.7254246190207695,
      "grad_norm": 0.107960057694832,
      "learning_rate": 5.770204722463011e-06,
      "loss": 0.1853,
      "step": 23480
    },
    {
      "epoch": 0.7257335732878141,
      "grad_norm": 0.10440752222594185,
      "learning_rate": 5.758128525465066e-06,
      "loss": 0.1805,
      "step": 23490
    },
    {
      "epoch": 0.7260425275548587,
      "grad_norm": 0.08911869025863042,
      "learning_rate": 5.7460619762137e-06,
      "loss": 0.1835,
      "step": 23500
    },
    {
      "epoch": 0.7263514818219033,
      "grad_norm": 0.10135969852159145,
      "learning_rate": 5.734005087305397e-06,
      "loss": 0.1817,
      "step": 23510
    },
    {
      "epoch": 0.7266604360889479,
      "grad_norm": 0.10504601399610826,
      "learning_rate": 5.7219578713265394e-06,
      "loss": 0.1852,
      "step": 23520
    },
    {
      "epoch": 0.7269693903559925,
      "grad_norm": 0.10418614054488268,
      "learning_rate": 5.70992034085343e-06,
      "loss": 0.1818,
      "step": 23530
    },
    {
      "epoch": 0.7272783446230372,
      "grad_norm": 0.10266555162815832,
      "learning_rate": 5.6978925084522474e-06,
      "loss": 0.182,
      "step": 23540
    },
    {
      "epoch": 0.7275872988900818,
      "grad_norm": 0.11505602967127468,
      "learning_rate": 5.6858743866790436e-06,
      "loss": 0.1808,
      "step": 23550
    },
    {
      "epoch": 0.7278962531571265,
      "grad_norm": 0.09420727352399028,
      "learning_rate": 5.673865988079752e-06,
      "loss": 0.1819,
      "step": 23560
    },
    {
      "epoch": 0.728205207424171,
      "grad_norm": 0.12686431057703784,
      "learning_rate": 5.661867325190131e-06,
      "loss": 0.1822,
      "step": 23570
    },
    {
      "epoch": 0.7285141616912156,
      "grad_norm": 0.10894260390245611,
      "learning_rate": 5.6498784105358e-06,
      "loss": 0.1993,
      "step": 23580
    },
    {
      "epoch": 0.7288231159582603,
      "grad_norm": 0.10968589067160978,
      "learning_rate": 5.637899256632181e-06,
      "loss": 0.1889,
      "step": 23590
    },
    {
      "epoch": 0.7291320702253049,
      "grad_norm": 0.08247539447465888,
      "learning_rate": 5.625929875984526e-06,
      "loss": 0.1799,
      "step": 23600
    },
    {
      "epoch": 0.7294410244923495,
      "grad_norm": 0.1053181577897392,
      "learning_rate": 5.613970281087865e-06,
      "loss": 0.1809,
      "step": 23610
    },
    {
      "epoch": 0.7297499787593942,
      "grad_norm": 0.1057329141337531,
      "learning_rate": 5.602020484427031e-06,
      "loss": 0.186,
      "step": 23620
    },
    {
      "epoch": 0.7300589330264388,
      "grad_norm": 0.10549607851658224,
      "learning_rate": 5.5900804984766105e-06,
      "loss": 0.1825,
      "step": 23630
    },
    {
      "epoch": 0.7303678872934833,
      "grad_norm": 0.1304769295554578,
      "learning_rate": 5.578150335700968e-06,
      "loss": 0.1824,
      "step": 23640
    },
    {
      "epoch": 0.730676841560528,
      "grad_norm": 0.09441833755987829,
      "learning_rate": 5.566230008554193e-06,
      "loss": 0.179,
      "step": 23650
    },
    {
      "epoch": 0.7309857958275726,
      "grad_norm": 0.0873047333241518,
      "learning_rate": 5.5543195294801266e-06,
      "loss": 0.1834,
      "step": 23660
    },
    {
      "epoch": 0.7312947500946172,
      "grad_norm": 0.0978571906611976,
      "learning_rate": 5.5424189109123145e-06,
      "loss": 0.1837,
      "step": 23670
    },
    {
      "epoch": 0.7316037043616619,
      "grad_norm": 0.09212224170428467,
      "learning_rate": 5.530528165274017e-06,
      "loss": 0.1853,
      "step": 23680
    },
    {
      "epoch": 0.7319126586287065,
      "grad_norm": 0.09675995488973269,
      "learning_rate": 5.518647304978179e-06,
      "loss": 0.1793,
      "step": 23690
    },
    {
      "epoch": 0.7322216128957512,
      "grad_norm": 0.09594312676716647,
      "learning_rate": 5.506776342427443e-06,
      "loss": 0.1905,
      "step": 23700
    },
    {
      "epoch": 0.7325305671627957,
      "grad_norm": 0.08795201298524304,
      "learning_rate": 5.494915290014095e-06,
      "loss": 0.1767,
      "step": 23710
    },
    {
      "epoch": 0.7328395214298403,
      "grad_norm": 0.08975525669259507,
      "learning_rate": 5.483064160120104e-06,
      "loss": 0.1785,
      "step": 23720
    },
    {
      "epoch": 0.733148475696885,
      "grad_norm": 0.09798393327238356,
      "learning_rate": 5.471222965117053e-06,
      "loss": 0.1844,
      "step": 23730
    },
    {
      "epoch": 0.7334574299639296,
      "grad_norm": 0.08887985967333485,
      "learning_rate": 5.459391717366179e-06,
      "loss": 0.1848,
      "step": 23740
    },
    {
      "epoch": 0.7337663842309742,
      "grad_norm": 0.09556267467208845,
      "learning_rate": 5.447570429218309e-06,
      "loss": 0.1892,
      "step": 23750
    },
    {
      "epoch": 0.7340753384980189,
      "grad_norm": 0.09954155983837434,
      "learning_rate": 5.435759113013898e-06,
      "loss": 0.1829,
      "step": 23760
    },
    {
      "epoch": 0.7343842927650635,
      "grad_norm": 0.09170270036599845,
      "learning_rate": 5.423957781082975e-06,
      "loss": 0.1817,
      "step": 23770
    },
    {
      "epoch": 0.734693247032108,
      "grad_norm": 0.10152918256092645,
      "learning_rate": 5.412166445745146e-06,
      "loss": 0.1784,
      "step": 23780
    },
    {
      "epoch": 0.7350022012991527,
      "grad_norm": 0.10255613442004073,
      "learning_rate": 5.400385119309595e-06,
      "loss": 0.1861,
      "step": 23790
    },
    {
      "epoch": 0.7353111555661973,
      "grad_norm": 0.08830756639633765,
      "learning_rate": 5.388613814075038e-06,
      "loss": 0.1815,
      "step": 23800
    },
    {
      "epoch": 0.7356201098332419,
      "grad_norm": 0.0935694713904321,
      "learning_rate": 5.3768525423297535e-06,
      "loss": 0.193,
      "step": 23810
    },
    {
      "epoch": 0.7359290641002866,
      "grad_norm": 0.1028806529008077,
      "learning_rate": 5.36510131635152e-06,
      "loss": 0.1816,
      "step": 23820
    },
    {
      "epoch": 0.7362380183673312,
      "grad_norm": 0.0877809893886209,
      "learning_rate": 5.353360148407655e-06,
      "loss": 0.1856,
      "step": 23830
    },
    {
      "epoch": 0.7365469726343759,
      "grad_norm": 0.11192160145740775,
      "learning_rate": 5.34162905075495e-06,
      "loss": 0.1843,
      "step": 23840
    },
    {
      "epoch": 0.7368559269014204,
      "grad_norm": 0.1148928991099739,
      "learning_rate": 5.329908035639707e-06,
      "loss": 0.1818,
      "step": 23850
    },
    {
      "epoch": 0.737164881168465,
      "grad_norm": 0.10522764460026213,
      "learning_rate": 5.318197115297686e-06,
      "loss": 0.1818,
      "step": 23860
    },
    {
      "epoch": 0.7374738354355097,
      "grad_norm": 0.12476438434685744,
      "learning_rate": 5.3064963019541224e-06,
      "loss": 0.1851,
      "step": 23870
    },
    {
      "epoch": 0.7377827897025543,
      "grad_norm": 0.10183551431313007,
      "learning_rate": 5.294805607823691e-06,
      "loss": 0.1775,
      "step": 23880
    },
    {
      "epoch": 0.7380917439695989,
      "grad_norm": 0.10808010473544906,
      "learning_rate": 5.283125045110499e-06,
      "loss": 0.1833,
      "step": 23890
    },
    {
      "epoch": 0.7384006982366436,
      "grad_norm": 0.09474897606017593,
      "learning_rate": 5.271454626008097e-06,
      "loss": 0.1799,
      "step": 23900
    },
    {
      "epoch": 0.7387096525036881,
      "grad_norm": 0.08593317067663166,
      "learning_rate": 5.2597943626994244e-06,
      "loss": 0.1816,
      "step": 23910
    },
    {
      "epoch": 0.7390186067707327,
      "grad_norm": 0.09640191337609255,
      "learning_rate": 5.248144267356827e-06,
      "loss": 0.1804,
      "step": 23920
    },
    {
      "epoch": 0.7393275610377774,
      "grad_norm": 0.08851606579777704,
      "learning_rate": 5.236504352142046e-06,
      "loss": 0.182,
      "step": 23930
    },
    {
      "epoch": 0.739636515304822,
      "grad_norm": 0.09498005045178236,
      "learning_rate": 5.224874629206176e-06,
      "loss": 0.185,
      "step": 23940
    },
    {
      "epoch": 0.7399454695718666,
      "grad_norm": 0.09617212108109258,
      "learning_rate": 5.213255110689694e-06,
      "loss": 0.1837,
      "step": 23950
    },
    {
      "epoch": 0.7402544238389113,
      "grad_norm": 0.09896703733829941,
      "learning_rate": 5.201645808722399e-06,
      "loss": 0.1826,
      "step": 23960
    },
    {
      "epoch": 0.7405633781059558,
      "grad_norm": 0.0959834731669881,
      "learning_rate": 5.190046735423453e-06,
      "loss": 0.1821,
      "step": 23970
    },
    {
      "epoch": 0.7408723323730005,
      "grad_norm": 0.09230008342047381,
      "learning_rate": 5.178457902901312e-06,
      "loss": 0.1801,
      "step": 23980
    },
    {
      "epoch": 0.7411812866400451,
      "grad_norm": 0.10128022750500347,
      "learning_rate": 5.166879323253766e-06,
      "loss": 0.1843,
      "step": 23990
    },
    {
      "epoch": 0.7414902409070897,
      "grad_norm": 0.0995053570730854,
      "learning_rate": 5.155311008567888e-06,
      "loss": 0.1793,
      "step": 24000
    },
    {
      "epoch": 0.7417991951741344,
      "grad_norm": 0.11384768264514729,
      "learning_rate": 5.143752970920031e-06,
      "loss": 0.1848,
      "step": 24010
    },
    {
      "epoch": 0.742108149441179,
      "grad_norm": 0.09487273464577489,
      "learning_rate": 5.132205222375837e-06,
      "loss": 0.1822,
      "step": 24020
    },
    {
      "epoch": 0.7424171037082236,
      "grad_norm": 0.09016893581053735,
      "learning_rate": 5.1206677749901884e-06,
      "loss": 0.1819,
      "step": 24030
    },
    {
      "epoch": 0.7427260579752683,
      "grad_norm": 0.11529766024246506,
      "learning_rate": 5.109140640807231e-06,
      "loss": 0.1867,
      "step": 24040
    },
    {
      "epoch": 0.7430350122423128,
      "grad_norm": 0.07877091685317325,
      "learning_rate": 5.097623831860326e-06,
      "loss": 0.1813,
      "step": 24050
    },
    {
      "epoch": 0.7433439665093574,
      "grad_norm": 0.09752304307877216,
      "learning_rate": 5.086117360172075e-06,
      "loss": 0.1811,
      "step": 24060
    },
    {
      "epoch": 0.7436529207764021,
      "grad_norm": 0.0952612245690614,
      "learning_rate": 5.074621237754271e-06,
      "loss": 0.1835,
      "step": 24070
    },
    {
      "epoch": 0.7439618750434467,
      "grad_norm": 0.08884004176471252,
      "learning_rate": 5.063135476607921e-06,
      "loss": 0.181,
      "step": 24080
    },
    {
      "epoch": 0.7442708293104913,
      "grad_norm": 0.11584434045708043,
      "learning_rate": 5.051660088723196e-06,
      "loss": 0.1889,
      "step": 24090
    },
    {
      "epoch": 0.744579783577536,
      "grad_norm": 0.10743692729312139,
      "learning_rate": 5.040195086079457e-06,
      "loss": 0.1877,
      "step": 24100
    },
    {
      "epoch": 0.7448887378445805,
      "grad_norm": 0.1090829668196744,
      "learning_rate": 5.028740480645211e-06,
      "loss": 0.1843,
      "step": 24110
    },
    {
      "epoch": 0.7451976921116252,
      "grad_norm": 0.09412843117885072,
      "learning_rate": 5.017296284378112e-06,
      "loss": 0.1844,
      "step": 24120
    },
    {
      "epoch": 0.7455066463786698,
      "grad_norm": 0.098847194500982,
      "learning_rate": 5.0058625092249585e-06,
      "loss": 0.1817,
      "step": 24130
    },
    {
      "epoch": 0.7458156006457144,
      "grad_norm": 0.09326719430806776,
      "learning_rate": 4.994439167121656e-06,
      "loss": 0.1826,
      "step": 24140
    },
    {
      "epoch": 0.7461245549127591,
      "grad_norm": 0.11397199967200321,
      "learning_rate": 4.983026269993225e-06,
      "loss": 0.1856,
      "step": 24150
    },
    {
      "epoch": 0.7464335091798037,
      "grad_norm": 5.363108996119458,
      "learning_rate": 4.971623829753788e-06,
      "loss": 0.2017,
      "step": 24160
    },
    {
      "epoch": 0.7467424634468482,
      "grad_norm": 0.09358995210467394,
      "learning_rate": 4.960231858306539e-06,
      "loss": 0.178,
      "step": 24170
    },
    {
      "epoch": 0.7470514177138929,
      "grad_norm": 0.09567347404863191,
      "learning_rate": 4.948850367543761e-06,
      "loss": 0.1817,
      "step": 24180
    },
    {
      "epoch": 0.7473603719809375,
      "grad_norm": 0.09588836437827075,
      "learning_rate": 4.937479369346772e-06,
      "loss": 0.1846,
      "step": 24190
    },
    {
      "epoch": 0.7476693262479821,
      "grad_norm": 0.1070859782676116,
      "learning_rate": 4.926118875585962e-06,
      "loss": 0.1789,
      "step": 24200
    },
    {
      "epoch": 0.7479782805150268,
      "grad_norm": 0.0972454782099022,
      "learning_rate": 4.914768898120735e-06,
      "loss": 0.1863,
      "step": 24210
    },
    {
      "epoch": 0.7482872347820714,
      "grad_norm": 0.08375075180811167,
      "learning_rate": 4.903429448799533e-06,
      "loss": 0.1815,
      "step": 24220
    },
    {
      "epoch": 0.748596189049116,
      "grad_norm": 0.09041052362777277,
      "learning_rate": 4.892100539459795e-06,
      "loss": 0.1808,
      "step": 24230
    },
    {
      "epoch": 0.7489051433161606,
      "grad_norm": 0.14079957159817338,
      "learning_rate": 4.880782181927958e-06,
      "loss": 0.1889,
      "step": 24240
    },
    {
      "epoch": 0.7492140975832052,
      "grad_norm": 0.08887254059664551,
      "learning_rate": 4.869474388019457e-06,
      "loss": 0.1898,
      "step": 24250
    },
    {
      "epoch": 0.7495230518502499,
      "grad_norm": 0.10819632457395752,
      "learning_rate": 4.85817716953868e-06,
      "loss": 0.1824,
      "step": 24260
    },
    {
      "epoch": 0.7498320061172945,
      "grad_norm": 0.09821873041202388,
      "learning_rate": 4.846890538278995e-06,
      "loss": 0.1862,
      "step": 24270
    },
    {
      "epoch": 0.7501409603843391,
      "grad_norm": 0.10241921776075516,
      "learning_rate": 4.835614506022698e-06,
      "loss": 0.1861,
      "step": 24280
    },
    {
      "epoch": 0.7504499146513838,
      "grad_norm": 0.12094620874012614,
      "learning_rate": 4.82434908454104e-06,
      "loss": 0.186,
      "step": 24290
    },
    {
      "epoch": 0.7507588689184284,
      "grad_norm": 0.10047937217461048,
      "learning_rate": 4.813094285594178e-06,
      "loss": 0.1822,
      "step": 24300
    },
    {
      "epoch": 0.7510678231854729,
      "grad_norm": 0.11834769443848707,
      "learning_rate": 4.801850120931198e-06,
      "loss": 0.1853,
      "step": 24310
    },
    {
      "epoch": 0.7513767774525176,
      "grad_norm": 0.09511623380065012,
      "learning_rate": 4.790616602290061e-06,
      "loss": 0.1815,
      "step": 24320
    },
    {
      "epoch": 0.7516857317195622,
      "grad_norm": 0.14707564861017536,
      "learning_rate": 4.77939374139764e-06,
      "loss": 0.1874,
      "step": 24330
    },
    {
      "epoch": 0.7519946859866068,
      "grad_norm": 0.09398758028656463,
      "learning_rate": 4.768181549969664e-06,
      "loss": 0.1823,
      "step": 24340
    },
    {
      "epoch": 0.7523036402536515,
      "grad_norm": 0.11256436160827804,
      "learning_rate": 4.756980039710728e-06,
      "loss": 0.1821,
      "step": 24350
    },
    {
      "epoch": 0.7526125945206961,
      "grad_norm": 0.09078154595153126,
      "learning_rate": 4.745789222314285e-06,
      "loss": 0.1926,
      "step": 24360
    },
    {
      "epoch": 0.7529215487877406,
      "grad_norm": 0.10197504872218038,
      "learning_rate": 4.734609109462617e-06,
      "loss": 0.1885,
      "step": 24370
    },
    {
      "epoch": 0.7532305030547853,
      "grad_norm": 0.09246076799905716,
      "learning_rate": 4.723439712826827e-06,
      "loss": 0.1841,
      "step": 24380
    },
    {
      "epoch": 0.7535394573218299,
      "grad_norm": 0.10161241559797542,
      "learning_rate": 4.7122810440668496e-06,
      "loss": 0.1768,
      "step": 24390
    },
    {
      "epoch": 0.7538484115888746,
      "grad_norm": 0.08090771359825617,
      "learning_rate": 4.701133114831396e-06,
      "loss": 0.1797,
      "step": 24400
    },
    {
      "epoch": 0.7541573658559192,
      "grad_norm": 0.09316268411410783,
      "learning_rate": 4.689995936757992e-06,
      "loss": 0.1798,
      "step": 24410
    },
    {
      "epoch": 0.7544663201229638,
      "grad_norm": 0.09417929681523575,
      "learning_rate": 4.678869521472913e-06,
      "loss": 0.1851,
      "step": 24420
    },
    {
      "epoch": 0.7547752743900085,
      "grad_norm": 0.11206295383355991,
      "learning_rate": 4.667753880591227e-06,
      "loss": 0.1799,
      "step": 24430
    },
    {
      "epoch": 0.755084228657053,
      "grad_norm": 0.11473768654406073,
      "learning_rate": 4.656649025716729e-06,
      "loss": 0.1855,
      "step": 24440
    },
    {
      "epoch": 0.7553931829240976,
      "grad_norm": 0.09069677363493305,
      "learning_rate": 4.645554968441974e-06,
      "loss": 0.1834,
      "step": 24450
    },
    {
      "epoch": 0.7557021371911423,
      "grad_norm": 0.10102245084622918,
      "learning_rate": 4.634471720348232e-06,
      "loss": 0.1868,
      "step": 24460
    },
    {
      "epoch": 0.7560110914581869,
      "grad_norm": 0.09449436609248286,
      "learning_rate": 4.623399293005492e-06,
      "loss": 0.1894,
      "step": 24470
    },
    {
      "epoch": 0.7563200457252315,
      "grad_norm": 0.10801735260821065,
      "learning_rate": 4.612337697972455e-06,
      "loss": 0.1879,
      "step": 24480
    },
    {
      "epoch": 0.7566289999922762,
      "grad_norm": 0.11635466347841847,
      "learning_rate": 4.601286946796502e-06,
      "loss": 0.188,
      "step": 24490
    },
    {
      "epoch": 0.7569379542593208,
      "grad_norm": 0.09888514655538418,
      "learning_rate": 4.590247051013706e-06,
      "loss": 0.1862,
      "step": 24500
    },
    {
      "epoch": 0.7572469085263653,
      "grad_norm": 0.10107275520320858,
      "learning_rate": 4.579218022148792e-06,
      "loss": 0.1897,
      "step": 24510
    },
    {
      "epoch": 0.75755586279341,
      "grad_norm": 0.10341155313562964,
      "learning_rate": 4.5681998717151616e-06,
      "loss": 0.1837,
      "step": 24520
    },
    {
      "epoch": 0.7578648170604546,
      "grad_norm": 0.10502438563128154,
      "learning_rate": 4.557192611214842e-06,
      "loss": 0.1799,
      "step": 24530
    },
    {
      "epoch": 0.7581737713274993,
      "grad_norm": 0.09835721540275669,
      "learning_rate": 4.5461962521385044e-06,
      "loss": 0.1828,
      "step": 24540
    },
    {
      "epoch": 0.7584827255945439,
      "grad_norm": 0.08791103137599914,
      "learning_rate": 4.535210805965427e-06,
      "loss": 0.1812,
      "step": 24550
    },
    {
      "epoch": 0.7587916798615885,
      "grad_norm": 0.10404060454522732,
      "learning_rate": 4.524236284163515e-06,
      "loss": 0.1818,
      "step": 24560
    },
    {
      "epoch": 0.7591006341286332,
      "grad_norm": 0.09153629747520635,
      "learning_rate": 4.513272698189251e-06,
      "loss": 0.1883,
      "step": 24570
    },
    {
      "epoch": 0.7594095883956777,
      "grad_norm": 0.11948459711738549,
      "learning_rate": 4.502320059487708e-06,
      "loss": 0.1828,
      "step": 24580
    },
    {
      "epoch": 0.7597185426627223,
      "grad_norm": 0.0945741265252532,
      "learning_rate": 4.49137837949253e-06,
      "loss": 0.1773,
      "step": 24590
    },
    {
      "epoch": 0.760027496929767,
      "grad_norm": 0.09549838705180028,
      "learning_rate": 4.480447669625929e-06,
      "loss": 0.1825,
      "step": 24600
    },
    {
      "epoch": 0.7603364511968116,
      "grad_norm": 0.11788533556362385,
      "learning_rate": 4.469527941298651e-06,
      "loss": 0.1817,
      "step": 24610
    },
    {
      "epoch": 0.7606454054638562,
      "grad_norm": 0.08925962238880984,
      "learning_rate": 4.458619205909993e-06,
      "loss": 0.1799,
      "step": 24620
    },
    {
      "epoch": 0.7609543597309009,
      "grad_norm": 0.10693628034532461,
      "learning_rate": 4.447721474847763e-06,
      "loss": 0.1868,
      "step": 24630
    },
    {
      "epoch": 0.7612633139979454,
      "grad_norm": 0.104860623770395,
      "learning_rate": 4.436834759488294e-06,
      "loss": 0.1862,
      "step": 24640
    },
    {
      "epoch": 0.76157226826499,
      "grad_norm": 0.10664557214643158,
      "learning_rate": 4.425959071196406e-06,
      "loss": 0.1785,
      "step": 24650
    },
    {
      "epoch": 0.7618812225320347,
      "grad_norm": 0.08596747890700626,
      "learning_rate": 4.4150944213254236e-06,
      "loss": 0.1823,
      "step": 24660
    },
    {
      "epoch": 0.7621901767990793,
      "grad_norm": 0.10739148452238738,
      "learning_rate": 4.4042408212171274e-06,
      "loss": 0.1792,
      "step": 24670
    },
    {
      "epoch": 0.762499131066124,
      "grad_norm": 0.09046668376876144,
      "learning_rate": 4.393398282201788e-06,
      "loss": 0.1864,
      "step": 24680
    },
    {
      "epoch": 0.7628080853331686,
      "grad_norm": 0.132524535699756,
      "learning_rate": 4.38256681559811e-06,
      "loss": 0.1887,
      "step": 24690
    },
    {
      "epoch": 0.7631170396002132,
      "grad_norm": 0.1341344024501761,
      "learning_rate": 4.371746432713239e-06,
      "loss": 0.1836,
      "step": 24700
    },
    {
      "epoch": 0.7634259938672578,
      "grad_norm": 0.10341488293705148,
      "learning_rate": 4.360937144842766e-06,
      "loss": 0.1795,
      "step": 24710
    },
    {
      "epoch": 0.7637349481343024,
      "grad_norm": 0.09795578826884338,
      "learning_rate": 4.350138963270683e-06,
      "loss": 0.1819,
      "step": 24720
    },
    {
      "epoch": 0.764043902401347,
      "grad_norm": 0.1039672908776288,
      "learning_rate": 4.339351899269403e-06,
      "loss": 0.1833,
      "step": 24730
    },
    {
      "epoch": 0.7643528566683917,
      "grad_norm": 0.10484384868946324,
      "learning_rate": 4.328575964099717e-06,
      "loss": 0.185,
      "step": 24740
    },
    {
      "epoch": 0.7646618109354363,
      "grad_norm": 0.10653571158897689,
      "learning_rate": 4.317811169010814e-06,
      "loss": 0.186,
      "step": 24750
    },
    {
      "epoch": 0.7649707652024809,
      "grad_norm": 0.08368143444105516,
      "learning_rate": 4.30705752524024e-06,
      "loss": 0.1883,
      "step": 24760
    },
    {
      "epoch": 0.7652797194695256,
      "grad_norm": 0.0890081773640572,
      "learning_rate": 4.296315044013913e-06,
      "loss": 0.1845,
      "step": 24770
    },
    {
      "epoch": 0.7655886737365701,
      "grad_norm": 0.09470903053284774,
      "learning_rate": 4.285583736546083e-06,
      "loss": 0.1793,
      "step": 24780
    },
    {
      "epoch": 0.7658976280036147,
      "grad_norm": 0.10418292073091513,
      "learning_rate": 4.274863614039354e-06,
      "loss": 0.1837,
      "step": 24790
    },
    {
      "epoch": 0.7662065822706594,
      "grad_norm": 0.09603698538873698,
      "learning_rate": 4.26415468768464e-06,
      "loss": 0.1841,
      "step": 24800
    },
    {
      "epoch": 0.766515536537704,
      "grad_norm": 0.09274982431505632,
      "learning_rate": 4.253456968661169e-06,
      "loss": 0.1829,
      "step": 24810
    },
    {
      "epoch": 0.7668244908047487,
      "grad_norm": 0.10597247125711916,
      "learning_rate": 4.242770468136468e-06,
      "loss": 0.188,
      "step": 24820
    },
    {
      "epoch": 0.7671334450717933,
      "grad_norm": 0.09575521592452313,
      "learning_rate": 4.232095197266367e-06,
      "loss": 0.1825,
      "step": 24830
    },
    {
      "epoch": 0.7674423993388378,
      "grad_norm": 0.09090213374277013,
      "learning_rate": 4.221431167194951e-06,
      "loss": 0.1812,
      "step": 24840
    },
    {
      "epoch": 0.7677513536058825,
      "grad_norm": 0.08056447313594031,
      "learning_rate": 4.210778389054593e-06,
      "loss": 0.1873,
      "step": 24850
    },
    {
      "epoch": 0.7680603078729271,
      "grad_norm": 0.09664373027301218,
      "learning_rate": 4.200136873965899e-06,
      "loss": 0.1809,
      "step": 24860
    },
    {
      "epoch": 0.7683692621399717,
      "grad_norm": 0.08286401524412505,
      "learning_rate": 4.1895066330377354e-06,
      "loss": 0.1814,
      "step": 24870
    },
    {
      "epoch": 0.7686782164070164,
      "grad_norm": 0.09756381686685321,
      "learning_rate": 4.1788876773671846e-06,
      "loss": 0.1831,
      "step": 24880
    },
    {
      "epoch": 0.768987170674061,
      "grad_norm": 0.10250866255103959,
      "learning_rate": 4.168280018039561e-06,
      "loss": 0.1956,
      "step": 24890
    },
    {
      "epoch": 0.7692961249411056,
      "grad_norm": 0.10863430594661831,
      "learning_rate": 4.1576836661283745e-06,
      "loss": 0.1796,
      "step": 24900
    },
    {
      "epoch": 0.7696050792081502,
      "grad_norm": 0.10923429111387054,
      "learning_rate": 4.147098632695344e-06,
      "loss": 0.1795,
      "step": 24910
    },
    {
      "epoch": 0.7699140334751948,
      "grad_norm": 0.09035054050720479,
      "learning_rate": 4.136524928790361e-06,
      "loss": 0.1839,
      "step": 24920
    },
    {
      "epoch": 0.7702229877422394,
      "grad_norm": 0.10010626082870915,
      "learning_rate": 4.125962565451491e-06,
      "loss": 0.1823,
      "step": 24930
    },
    {
      "epoch": 0.7705319420092841,
      "grad_norm": 0.1098501348723468,
      "learning_rate": 4.115411553704974e-06,
      "loss": 0.1862,
      "step": 24940
    },
    {
      "epoch": 0.7708408962763287,
      "grad_norm": 0.08330661053469954,
      "learning_rate": 4.104871904565177e-06,
      "loss": 0.1833,
      "step": 24950
    },
    {
      "epoch": 0.7711498505433734,
      "grad_norm": 0.09553979714716174,
      "learning_rate": 4.094343629034632e-06,
      "loss": 0.1776,
      "step": 24960
    },
    {
      "epoch": 0.771458804810418,
      "grad_norm": 0.09578591656101941,
      "learning_rate": 4.083826738103976e-06,
      "loss": 0.1789,
      "step": 24970
    },
    {
      "epoch": 0.7717677590774625,
      "grad_norm": 0.09682309178619027,
      "learning_rate": 4.0733212427519774e-06,
      "loss": 0.1838,
      "step": 24980
    },
    {
      "epoch": 0.7720767133445072,
      "grad_norm": 0.09677859659507333,
      "learning_rate": 4.062827153945491e-06,
      "loss": 0.1868,
      "step": 24990
    },
    {
      "epoch": 0.7723856676115518,
      "grad_norm": 0.130969337160774,
      "learning_rate": 4.052344482639487e-06,
      "loss": 0.1834,
      "step": 25000
    },
    {
      "epoch": 0.7726946218785964,
      "grad_norm": 0.10810964448034079,
      "learning_rate": 4.041873239776996e-06,
      "loss": 0.1822,
      "step": 25010
    },
    {
      "epoch": 0.7730035761456411,
      "grad_norm": 0.12253053442932915,
      "learning_rate": 4.031413436289123e-06,
      "loss": 0.1836,
      "step": 25020
    },
    {
      "epoch": 0.7733125304126857,
      "grad_norm": 0.09211352300720242,
      "learning_rate": 4.0209650830950445e-06,
      "loss": 0.1826,
      "step": 25030
    },
    {
      "epoch": 0.7736214846797302,
      "grad_norm": 0.11529614778976754,
      "learning_rate": 4.010528191101966e-06,
      "loss": 0.1869,
      "step": 25040
    },
    {
      "epoch": 0.7739304389467749,
      "grad_norm": 0.11543787517700385,
      "learning_rate": 4.000102771205137e-06,
      "loss": 0.1898,
      "step": 25050
    },
    {
      "epoch": 0.7742393932138195,
      "grad_norm": 0.10341416860828288,
      "learning_rate": 3.989688834287835e-06,
      "loss": 0.1827,
      "step": 25060
    },
    {
      "epoch": 0.7745483474808641,
      "grad_norm": 0.10971970533837791,
      "learning_rate": 3.979286391221338e-06,
      "loss": 0.1794,
      "step": 25070
    },
    {
      "epoch": 0.7748573017479088,
      "grad_norm": 0.09906974833820673,
      "learning_rate": 3.968895452864942e-06,
      "loss": 0.183,
      "step": 25080
    },
    {
      "epoch": 0.7751662560149534,
      "grad_norm": 0.10414415075035399,
      "learning_rate": 3.9585160300659155e-06,
      "loss": 0.1923,
      "step": 25090
    },
    {
      "epoch": 0.7754752102819981,
      "grad_norm": 0.0939762528781646,
      "learning_rate": 3.948148133659524e-06,
      "loss": 0.1871,
      "step": 25100
    },
    {
      "epoch": 0.7757841645490426,
      "grad_norm": 0.09046857533809367,
      "learning_rate": 3.937791774468978e-06,
      "loss": 0.1848,
      "step": 25110
    },
    {
      "epoch": 0.7760931188160872,
      "grad_norm": 0.08905404711090185,
      "learning_rate": 3.927446963305469e-06,
      "loss": 0.1861,
      "step": 25120
    },
    {
      "epoch": 0.7764020730831319,
      "grad_norm": 0.09020382783293679,
      "learning_rate": 3.917113710968113e-06,
      "loss": 0.1819,
      "step": 25130
    },
    {
      "epoch": 0.7767110273501765,
      "grad_norm": 0.09496973877682766,
      "learning_rate": 3.906792028243974e-06,
      "loss": 0.1796,
      "step": 25140
    },
    {
      "epoch": 0.7770199816172211,
      "grad_norm": 0.09796008618843378,
      "learning_rate": 3.896481925908027e-06,
      "loss": 0.1821,
      "step": 25150
    },
    {
      "epoch": 0.7773289358842658,
      "grad_norm": 0.11406156153732327,
      "learning_rate": 3.8861834147231615e-06,
      "loss": 0.182,
      "step": 25160
    },
    {
      "epoch": 0.7776378901513104,
      "grad_norm": 0.10670777254215298,
      "learning_rate": 3.875896505440174e-06,
      "loss": 0.1874,
      "step": 25170
    },
    {
      "epoch": 0.7779468444183549,
      "grad_norm": 0.11516002557890306,
      "learning_rate": 3.865621208797738e-06,
      "loss": 0.1927,
      "step": 25180
    },
    {
      "epoch": 0.7782557986853996,
      "grad_norm": 0.11069305984302626,
      "learning_rate": 3.855357535522415e-06,
      "loss": 0.1884,
      "step": 25190
    },
    {
      "epoch": 0.7785647529524442,
      "grad_norm": 0.11518501842536932,
      "learning_rate": 3.8451054963286245e-06,
      "loss": 0.1813,
      "step": 25200
    },
    {
      "epoch": 0.7788737072194888,
      "grad_norm": 0.12654688882798518,
      "learning_rate": 3.834865101918647e-06,
      "loss": 0.1855,
      "step": 25210
    },
    {
      "epoch": 0.7791826614865335,
      "grad_norm": 0.10198879893025203,
      "learning_rate": 3.824636362982599e-06,
      "loss": 0.1798,
      "step": 25220
    },
    {
      "epoch": 0.7794916157535781,
      "grad_norm": 0.09364822371009877,
      "learning_rate": 3.8144192901984426e-06,
      "loss": 0.1904,
      "step": 25230
    },
    {
      "epoch": 0.7798005700206228,
      "grad_norm": 0.09277265898743273,
      "learning_rate": 3.8042138942319504e-06,
      "loss": 0.1784,
      "step": 25240
    },
    {
      "epoch": 0.7801095242876673,
      "grad_norm": 0.1260953472460793,
      "learning_rate": 3.7940201857367025e-06,
      "loss": 0.1872,
      "step": 25250
    },
    {
      "epoch": 0.7804184785547119,
      "grad_norm": 0.11903159460229215,
      "learning_rate": 3.7838381753540947e-06,
      "loss": 0.1849,
      "step": 25260
    },
    {
      "epoch": 0.7807274328217566,
      "grad_norm": 0.10265665865549166,
      "learning_rate": 3.7736678737132947e-06,
      "loss": 0.1811,
      "step": 25270
    },
    {
      "epoch": 0.7810363870888012,
      "grad_norm": 0.10642599001532839,
      "learning_rate": 3.76350929143125e-06,
      "loss": 0.1859,
      "step": 25280
    },
    {
      "epoch": 0.7813453413558458,
      "grad_norm": 0.12096619689749742,
      "learning_rate": 3.7533624391126835e-06,
      "loss": 0.185,
      "step": 25290
    },
    {
      "epoch": 0.7816542956228905,
      "grad_norm": 0.10298733681963296,
      "learning_rate": 3.7432273273500615e-06,
      "loss": 0.1871,
      "step": 25300
    },
    {
      "epoch": 0.781963249889935,
      "grad_norm": 0.13644706559300795,
      "learning_rate": 3.7331039667236055e-06,
      "loss": 0.1802,
      "step": 25310
    },
    {
      "epoch": 0.7822722041569796,
      "grad_norm": 0.11522219296539363,
      "learning_rate": 3.722992367801255e-06,
      "loss": 0.191,
      "step": 25320
    },
    {
      "epoch": 0.7825811584240243,
      "grad_norm": 0.10725527487457508,
      "learning_rate": 3.712892541138688e-06,
      "loss": 0.1817,
      "step": 25330
    },
    {
      "epoch": 0.7828901126910689,
      "grad_norm": 0.10883563343966805,
      "learning_rate": 3.7028044972792795e-06,
      "loss": 0.1822,
      "step": 25340
    },
    {
      "epoch": 0.7831990669581135,
      "grad_norm": 0.11503872745234577,
      "learning_rate": 3.692728246754115e-06,
      "loss": 0.1833,
      "step": 25350
    },
    {
      "epoch": 0.7835080212251582,
      "grad_norm": 0.10575277807950564,
      "learning_rate": 3.6826638000819575e-06,
      "loss": 0.1792,
      "step": 25360
    },
    {
      "epoch": 0.7838169754922027,
      "grad_norm": 0.08601618952674542,
      "learning_rate": 3.672611167769262e-06,
      "loss": 0.1778,
      "step": 25370
    },
    {
      "epoch": 0.7841259297592474,
      "grad_norm": 0.09357258961607313,
      "learning_rate": 3.6625703603101347e-06,
      "loss": 0.1857,
      "step": 25380
    },
    {
      "epoch": 0.784434884026292,
      "grad_norm": 0.10233492578376589,
      "learning_rate": 3.652541388186344e-06,
      "loss": 0.1799,
      "step": 25390
    },
    {
      "epoch": 0.7847438382933366,
      "grad_norm": 0.10152795001207753,
      "learning_rate": 3.6425242618673123e-06,
      "loss": 0.1794,
      "step": 25400
    },
    {
      "epoch": 0.7850527925603813,
      "grad_norm": 0.10989492354317137,
      "learning_rate": 3.6325189918100794e-06,
      "loss": 0.1801,
      "step": 25410
    },
    {
      "epoch": 0.7853617468274259,
      "grad_norm": 0.10837134213569849,
      "learning_rate": 3.622525588459324e-06,
      "loss": 0.1808,
      "step": 25420
    },
    {
      "epoch": 0.7856707010944705,
      "grad_norm": 0.12192784770304528,
      "learning_rate": 3.612544062247321e-06,
      "loss": 0.1839,
      "step": 25430
    },
    {
      "epoch": 0.7859796553615152,
      "grad_norm": 0.09966334737502888,
      "learning_rate": 3.6025744235939638e-06,
      "loss": 0.1806,
      "step": 25440
    },
    {
      "epoch": 0.7862886096285597,
      "grad_norm": 0.10109780873211506,
      "learning_rate": 3.5926166829067207e-06,
      "loss": 0.1814,
      "step": 25450
    },
    {
      "epoch": 0.7865975638956043,
      "grad_norm": 0.09813537785873704,
      "learning_rate": 3.5826708505806533e-06,
      "loss": 0.18,
      "step": 25460
    },
    {
      "epoch": 0.786906518162649,
      "grad_norm": 0.10611977069156357,
      "learning_rate": 3.5727369369983828e-06,
      "loss": 0.1859,
      "step": 25470
    },
    {
      "epoch": 0.7872154724296936,
      "grad_norm": 0.10485472316535571,
      "learning_rate": 3.5628149525300817e-06,
      "loss": 0.1806,
      "step": 25480
    },
    {
      "epoch": 0.7875244266967382,
      "grad_norm": 0.09427157604721462,
      "learning_rate": 3.5529049075334906e-06,
      "loss": 0.1817,
      "step": 25490
    },
    {
      "epoch": 0.7878333809637829,
      "grad_norm": 0.08384182792045067,
      "learning_rate": 3.5430068123538674e-06,
      "loss": 0.1828,
      "step": 25500
    },
    {
      "epoch": 0.7881423352308274,
      "grad_norm": 0.09922342997052187,
      "learning_rate": 3.5331206773239978e-06,
      "loss": 0.1809,
      "step": 25510
    },
    {
      "epoch": 0.7884512894978721,
      "grad_norm": 0.0910035117708188,
      "learning_rate": 3.523246512764195e-06,
      "loss": 0.1879,
      "step": 25520
    },
    {
      "epoch": 0.7887602437649167,
      "grad_norm": 0.12024080516660918,
      "learning_rate": 3.513384328982259e-06,
      "loss": 0.1808,
      "step": 25530
    },
    {
      "epoch": 0.7890691980319613,
      "grad_norm": 0.09952543576053921,
      "learning_rate": 3.503534136273497e-06,
      "loss": 0.1805,
      "step": 25540
    },
    {
      "epoch": 0.789378152299006,
      "grad_norm": 0.09815265413663787,
      "learning_rate": 3.4936959449206873e-06,
      "loss": 0.1862,
      "step": 25550
    },
    {
      "epoch": 0.7896871065660506,
      "grad_norm": 0.10812201758935382,
      "learning_rate": 3.483869765194093e-06,
      "loss": 0.1822,
      "step": 25560
    },
    {
      "epoch": 0.7899960608330951,
      "grad_norm": 0.10562731468441008,
      "learning_rate": 3.474055607351419e-06,
      "loss": 0.1835,
      "step": 25570
    },
    {
      "epoch": 0.7903050151001398,
      "grad_norm": 0.09156005852258871,
      "learning_rate": 3.4642534816378414e-06,
      "loss": 0.1794,
      "step": 25580
    },
    {
      "epoch": 0.7906139693671844,
      "grad_norm": 0.08828434729860438,
      "learning_rate": 3.454463398285958e-06,
      "loss": 0.182,
      "step": 25590
    },
    {
      "epoch": 0.790922923634229,
      "grad_norm": 0.0908660503922225,
      "learning_rate": 3.444685367515809e-06,
      "loss": 0.1888,
      "step": 25600
    },
    {
      "epoch": 0.7912318779012737,
      "grad_norm": 0.10320120559463324,
      "learning_rate": 3.434919399534846e-06,
      "loss": 0.1785,
      "step": 25610
    },
    {
      "epoch": 0.7915408321683183,
      "grad_norm": 0.0841973932355569,
      "learning_rate": 3.4251655045379206e-06,
      "loss": 0.1806,
      "step": 25620
    },
    {
      "epoch": 0.7918497864353629,
      "grad_norm": 0.09528283527352377,
      "learning_rate": 3.4154236927073025e-06,
      "loss": 0.1795,
      "step": 25630
    },
    {
      "epoch": 0.7921587407024075,
      "grad_norm": 0.08683950961777887,
      "learning_rate": 3.405693974212622e-06,
      "loss": 0.1829,
      "step": 25640
    },
    {
      "epoch": 0.7924676949694521,
      "grad_norm": 0.104330368206283,
      "learning_rate": 3.395976359210908e-06,
      "loss": 0.1828,
      "step": 25650
    },
    {
      "epoch": 0.7927766492364968,
      "grad_norm": 0.10091920942922289,
      "learning_rate": 3.386270857846535e-06,
      "loss": 0.1914,
      "step": 25660
    },
    {
      "epoch": 0.7930856035035414,
      "grad_norm": 0.10033476065215786,
      "learning_rate": 3.3765774802512495e-06,
      "loss": 0.1858,
      "step": 25670
    },
    {
      "epoch": 0.793394557770586,
      "grad_norm": 0.09334760840161517,
      "learning_rate": 3.366896236544123e-06,
      "loss": 0.1822,
      "step": 25680
    },
    {
      "epoch": 0.7937035120376307,
      "grad_norm": 0.09678851010727342,
      "learning_rate": 3.3572271368315814e-06,
      "loss": 0.1817,
      "step": 25690
    },
    {
      "epoch": 0.7940124663046753,
      "grad_norm": 0.09973566574864082,
      "learning_rate": 3.347570191207354e-06,
      "loss": 0.1839,
      "step": 25700
    },
    {
      "epoch": 0.7943214205717198,
      "grad_norm": 0.08610035124366384,
      "learning_rate": 3.337925409752495e-06,
      "loss": 0.1909,
      "step": 25710
    },
    {
      "epoch": 0.7946303748387645,
      "grad_norm": 0.10119454927901549,
      "learning_rate": 3.3282928025353476e-06,
      "loss": 0.1827,
      "step": 25720
    },
    {
      "epoch": 0.7949393291058091,
      "grad_norm": 0.13967062222556495,
      "learning_rate": 3.318672379611564e-06,
      "loss": 0.189,
      "step": 25730
    },
    {
      "epoch": 0.7952482833728537,
      "grad_norm": 0.11231914811684597,
      "learning_rate": 3.309064151024058e-06,
      "loss": 0.1876,
      "step": 25740
    },
    {
      "epoch": 0.7955572376398984,
      "grad_norm": 0.10296738133323749,
      "learning_rate": 3.299468126803033e-06,
      "loss": 0.1844,
      "step": 25750
    },
    {
      "epoch": 0.795866191906943,
      "grad_norm": 0.11107133897996203,
      "learning_rate": 3.2898843169659304e-06,
      "loss": 0.1814,
      "step": 25760
    },
    {
      "epoch": 0.7961751461739875,
      "grad_norm": 0.09354877349047017,
      "learning_rate": 3.2803127315174625e-06,
      "loss": 0.1862,
      "step": 25770
    },
    {
      "epoch": 0.7964841004410322,
      "grad_norm": 0.09370649145844268,
      "learning_rate": 3.2707533804495597e-06,
      "loss": 0.182,
      "step": 25780
    },
    {
      "epoch": 0.7967930547080768,
      "grad_norm": 0.09180617405999886,
      "learning_rate": 3.2612062737414016e-06,
      "loss": 0.1838,
      "step": 25790
    },
    {
      "epoch": 0.7971020089751215,
      "grad_norm": 0.1015999856063287,
      "learning_rate": 3.2516714213593657e-06,
      "loss": 0.1846,
      "step": 25800
    },
    {
      "epoch": 0.7974109632421661,
      "grad_norm": 0.10029948167309428,
      "learning_rate": 3.242148833257053e-06,
      "loss": 0.1829,
      "step": 25810
    },
    {
      "epoch": 0.7977199175092107,
      "grad_norm": 0.09299225626189431,
      "learning_rate": 3.232638519375249e-06,
      "loss": 0.1841,
      "step": 25820
    },
    {
      "epoch": 0.7980288717762554,
      "grad_norm": 0.09937895416502573,
      "learning_rate": 3.2231404896419382e-06,
      "loss": 0.1845,
      "step": 25830
    },
    {
      "epoch": 0.7983378260433,
      "grad_norm": 0.10619821626272419,
      "learning_rate": 3.2136547539722726e-06,
      "loss": 0.1776,
      "step": 25840
    },
    {
      "epoch": 0.7986467803103445,
      "grad_norm": 0.08838672358304156,
      "learning_rate": 3.2041813222685683e-06,
      "loss": 0.1801,
      "step": 25850
    },
    {
      "epoch": 0.7989557345773892,
      "grad_norm": 0.11498206831516014,
      "learning_rate": 3.1947202044203093e-06,
      "loss": 0.1785,
      "step": 25860
    },
    {
      "epoch": 0.7992646888444338,
      "grad_norm": 0.0890786518364633,
      "learning_rate": 3.1852714103041103e-06,
      "loss": 0.1907,
      "step": 25870
    },
    {
      "epoch": 0.7995736431114784,
      "grad_norm": 0.1026602108963983,
      "learning_rate": 3.175834949783734e-06,
      "loss": 0.1781,
      "step": 25880
    },
    {
      "epoch": 0.7998825973785231,
      "grad_norm": 0.09939878605795972,
      "learning_rate": 3.1664108327100544e-06,
      "loss": 0.1795,
      "step": 25890
    },
    {
      "epoch": 0.8001915516455677,
      "grad_norm": 0.09105069229392243,
      "learning_rate": 3.1569990689210764e-06,
      "loss": 0.1878,
      "step": 25900
    },
    {
      "epoch": 0.8005005059126122,
      "grad_norm": 0.11413370928292792,
      "learning_rate": 3.1475996682418912e-06,
      "loss": 0.1837,
      "step": 25910
    },
    {
      "epoch": 0.8008094601796569,
      "grad_norm": 0.1160283753139262,
      "learning_rate": 3.1382126404846993e-06,
      "loss": 0.1804,
      "step": 25920
    },
    {
      "epoch": 0.8011184144467015,
      "grad_norm": 0.11437225315421572,
      "learning_rate": 3.128837995448778e-06,
      "loss": 0.179,
      "step": 25930
    },
    {
      "epoch": 0.8014273687137462,
      "grad_norm": 0.12904363174166586,
      "learning_rate": 3.119475742920476e-06,
      "loss": 0.1819,
      "step": 25940
    },
    {
      "epoch": 0.8017363229807908,
      "grad_norm": 0.1027034226463252,
      "learning_rate": 3.1101258926732036e-06,
      "loss": 0.1847,
      "step": 25950
    },
    {
      "epoch": 0.8020452772478354,
      "grad_norm": 0.10556362274882444,
      "learning_rate": 3.100788454467439e-06,
      "loss": 0.1829,
      "step": 25960
    },
    {
      "epoch": 0.8023542315148801,
      "grad_norm": 0.10631526692793652,
      "learning_rate": 3.0914634380506794e-06,
      "loss": 0.1859,
      "step": 25970
    },
    {
      "epoch": 0.8026631857819246,
      "grad_norm": 0.10186607706552472,
      "learning_rate": 3.082150853157481e-06,
      "loss": 0.1801,
      "step": 25980
    },
    {
      "epoch": 0.8029721400489692,
      "grad_norm": 0.09672703905085761,
      "learning_rate": 3.0728507095093985e-06,
      "loss": 0.1782,
      "step": 25990
    },
    {
      "epoch": 0.8032810943160139,
      "grad_norm": 0.09494310970245266,
      "learning_rate": 3.063563016815017e-06,
      "loss": 0.1772,
      "step": 26000
    },
    {
      "epoch": 0.8035900485830585,
      "grad_norm": 0.1261578815363774,
      "learning_rate": 3.0542877847699113e-06,
      "loss": 0.1871,
      "step": 26010
    },
    {
      "epoch": 0.8038990028501031,
      "grad_norm": 0.09858750860348602,
      "learning_rate": 3.0450250230566596e-06,
      "loss": 0.1824,
      "step": 26020
    },
    {
      "epoch": 0.8042079571171478,
      "grad_norm": 0.12924161690654817,
      "learning_rate": 3.03577474134481e-06,
      "loss": 0.1798,
      "step": 26030
    },
    {
      "epoch": 0.8045169113841923,
      "grad_norm": 0.12313499376004705,
      "learning_rate": 3.026536949290895e-06,
      "loss": 0.1804,
      "step": 26040
    },
    {
      "epoch": 0.8048258656512369,
      "grad_norm": 0.10588291446113972,
      "learning_rate": 3.0173116565383956e-06,
      "loss": 0.1786,
      "step": 26050
    },
    {
      "epoch": 0.8051348199182816,
      "grad_norm": 0.09779036024356536,
      "learning_rate": 3.00809887271776e-06,
      "loss": 0.1823,
      "step": 26060
    },
    {
      "epoch": 0.8054437741853262,
      "grad_norm": 0.08743651968312417,
      "learning_rate": 2.998898607446363e-06,
      "loss": 0.1839,
      "step": 26070
    },
    {
      "epoch": 0.8057527284523709,
      "grad_norm": 0.09384200809989345,
      "learning_rate": 2.989710870328516e-06,
      "loss": 0.1872,
      "step": 26080
    },
    {
      "epoch": 0.8060616827194155,
      "grad_norm": 0.0998580179141942,
      "learning_rate": 2.9805356709554626e-06,
      "loss": 0.19,
      "step": 26090
    },
    {
      "epoch": 0.80637063698646,
      "grad_norm": 0.09007294558796745,
      "learning_rate": 2.97137301890534e-06,
      "loss": 0.1784,
      "step": 26100
    },
    {
      "epoch": 0.8066795912535047,
      "grad_norm": 0.09481269282717532,
      "learning_rate": 2.9622229237432037e-06,
      "loss": 0.1801,
      "step": 26110
    },
    {
      "epoch": 0.8069885455205493,
      "grad_norm": 0.09856570964423522,
      "learning_rate": 2.953085395020986e-06,
      "loss": 0.1823,
      "step": 26120
    },
    {
      "epoch": 0.8072974997875939,
      "grad_norm": 0.09033459745834758,
      "learning_rate": 2.9439604422775163e-06,
      "loss": 0.1803,
      "step": 26130
    },
    {
      "epoch": 0.8076064540546386,
      "grad_norm": 0.0954125577341479,
      "learning_rate": 2.934848075038484e-06,
      "loss": 0.1826,
      "step": 26140
    },
    {
      "epoch": 0.8079154083216832,
      "grad_norm": 0.08962536980220799,
      "learning_rate": 2.9257483028164405e-06,
      "loss": 0.1811,
      "step": 26150
    },
    {
      "epoch": 0.8082243625887278,
      "grad_norm": 0.12086052233069301,
      "learning_rate": 2.9166611351107996e-06,
      "loss": 0.183,
      "step": 26160
    },
    {
      "epoch": 0.8085333168557725,
      "grad_norm": 0.09011053516911019,
      "learning_rate": 2.9075865814078064e-06,
      "loss": 0.1858,
      "step": 26170
    },
    {
      "epoch": 0.808842271122817,
      "grad_norm": 0.09099461955071193,
      "learning_rate": 2.8985246511805375e-06,
      "loss": 0.1814,
      "step": 26180
    },
    {
      "epoch": 0.8091512253898616,
      "grad_norm": 0.10358805842616044,
      "learning_rate": 2.889475353888903e-06,
      "loss": 0.1798,
      "step": 26190
    },
    {
      "epoch": 0.8094601796569063,
      "grad_norm": 0.10379934010578504,
      "learning_rate": 2.880438698979611e-06,
      "loss": 0.1889,
      "step": 26200
    },
    {
      "epoch": 0.8097691339239509,
      "grad_norm": 0.13552284107458618,
      "learning_rate": 2.8714146958861877e-06,
      "loss": 0.1787,
      "step": 26210
    },
    {
      "epoch": 0.8100780881909956,
      "grad_norm": 0.11537436983097167,
      "learning_rate": 2.8624033540289334e-06,
      "loss": 0.1869,
      "step": 26220
    },
    {
      "epoch": 0.8103870424580402,
      "grad_norm": 0.10467640868514755,
      "learning_rate": 2.8534046828149506e-06,
      "loss": 0.1831,
      "step": 26230
    },
    {
      "epoch": 0.8106959967250847,
      "grad_norm": 0.10466198430777411,
      "learning_rate": 2.8444186916380966e-06,
      "loss": 0.1826,
      "step": 26240
    },
    {
      "epoch": 0.8110049509921294,
      "grad_norm": 0.08690735946650803,
      "learning_rate": 2.8354453898790074e-06,
      "loss": 0.1804,
      "step": 26250
    },
    {
      "epoch": 0.811313905259174,
      "grad_norm": 0.10398113807428562,
      "learning_rate": 2.8264847869050564e-06,
      "loss": 0.1802,
      "step": 26260
    },
    {
      "epoch": 0.8116228595262186,
      "grad_norm": 0.09015814681323908,
      "learning_rate": 2.8175368920703797e-06,
      "loss": 0.1831,
      "step": 26270
    },
    {
      "epoch": 0.8119318137932633,
      "grad_norm": 0.09636705603596216,
      "learning_rate": 2.8086017147158273e-06,
      "loss": 0.1846,
      "step": 26280
    },
    {
      "epoch": 0.8122407680603079,
      "grad_norm": 0.13561475956548416,
      "learning_rate": 2.799679264168989e-06,
      "loss": 0.1811,
      "step": 26290
    },
    {
      "epoch": 0.8125497223273525,
      "grad_norm": 0.08541995734046805,
      "learning_rate": 2.790769549744161e-06,
      "loss": 0.1832,
      "step": 26300
    },
    {
      "epoch": 0.8128586765943971,
      "grad_norm": 0.10269967585773117,
      "learning_rate": 2.781872580742342e-06,
      "loss": 0.184,
      "step": 26310
    },
    {
      "epoch": 0.8131676308614417,
      "grad_norm": 0.10694390579247381,
      "learning_rate": 2.7729883664512334e-06,
      "loss": 0.1844,
      "step": 26320
    },
    {
      "epoch": 0.8134765851284863,
      "grad_norm": 0.10484432211326815,
      "learning_rate": 2.7641169161452122e-06,
      "loss": 0.185,
      "step": 26330
    },
    {
      "epoch": 0.813785539395531,
      "grad_norm": 0.08086703065465609,
      "learning_rate": 2.7552582390853436e-06,
      "loss": 0.1783,
      "step": 26340
    },
    {
      "epoch": 0.8140944936625756,
      "grad_norm": 0.08832847223338984,
      "learning_rate": 2.7464123445193434e-06,
      "loss": 0.1841,
      "step": 26350
    },
    {
      "epoch": 0.8144034479296203,
      "grad_norm": 0.096732950352893,
      "learning_rate": 2.737579241681598e-06,
      "loss": 0.1798,
      "step": 26360
    },
    {
      "epoch": 0.8147124021966649,
      "grad_norm": 0.1258447057640801,
      "learning_rate": 2.72875893979313e-06,
      "loss": 0.1906,
      "step": 26370
    },
    {
      "epoch": 0.8150213564637094,
      "grad_norm": 0.10718431506263308,
      "learning_rate": 2.719951448061599e-06,
      "loss": 0.1803,
      "step": 26380
    },
    {
      "epoch": 0.8153303107307541,
      "grad_norm": 0.0849113598593184,
      "learning_rate": 2.7111567756813016e-06,
      "loss": 0.1787,
      "step": 26390
    },
    {
      "epoch": 0.8156392649977987,
      "grad_norm": 0.09298630797136359,
      "learning_rate": 2.7023749318331424e-06,
      "loss": 0.1799,
      "step": 26400
    },
    {
      "epoch": 0.8159482192648433,
      "grad_norm": 0.10733842023666557,
      "learning_rate": 2.6936059256846353e-06,
      "loss": 0.1813,
      "step": 26410
    },
    {
      "epoch": 0.816257173531888,
      "grad_norm": 0.12126086751845842,
      "learning_rate": 2.6848497663899003e-06,
      "loss": 0.1862,
      "step": 26420
    },
    {
      "epoch": 0.8165661277989326,
      "grad_norm": 0.10450695004230649,
      "learning_rate": 2.6761064630896325e-06,
      "loss": 0.1878,
      "step": 26430
    },
    {
      "epoch": 0.8168750820659771,
      "grad_norm": 0.11306162621207497,
      "learning_rate": 2.6673760249111245e-06,
      "loss": 0.1833,
      "step": 26440
    },
    {
      "epoch": 0.8171840363330218,
      "grad_norm": 0.08166325960287754,
      "learning_rate": 2.6586584609682206e-06,
      "loss": 0.1829,
      "step": 26450
    },
    {
      "epoch": 0.8174929906000664,
      "grad_norm": 0.11217080019997884,
      "learning_rate": 2.64995378036134e-06,
      "loss": 0.1861,
      "step": 26460
    },
    {
      "epoch": 0.817801944867111,
      "grad_norm": 0.09301487805870076,
      "learning_rate": 2.6412619921774418e-06,
      "loss": 0.1795,
      "step": 26470
    },
    {
      "epoch": 0.8181108991341557,
      "grad_norm": 0.10610936078817064,
      "learning_rate": 2.632583105490036e-06,
      "loss": 0.182,
      "step": 26480
    },
    {
      "epoch": 0.8184198534012003,
      "grad_norm": 0.10507038732506645,
      "learning_rate": 2.623917129359153e-06,
      "loss": 0.1911,
      "step": 26490
    },
    {
      "epoch": 0.818728807668245,
      "grad_norm": 0.09456845841819382,
      "learning_rate": 2.615264072831362e-06,
      "loss": 0.1856,
      "step": 26500
    },
    {
      "epoch": 0.8190377619352895,
      "grad_norm": 0.10497736812272437,
      "learning_rate": 2.606623944939726e-06,
      "loss": 0.1813,
      "step": 26510
    },
    {
      "epoch": 0.8193467162023341,
      "grad_norm": 0.09871780872343341,
      "learning_rate": 2.5979967547038304e-06,
      "loss": 0.1815,
      "step": 26520
    },
    {
      "epoch": 0.8196556704693788,
      "grad_norm": 0.10846989193582626,
      "learning_rate": 2.5893825111297387e-06,
      "loss": 0.1828,
      "step": 26530
    },
    {
      "epoch": 0.8199646247364234,
      "grad_norm": 0.08585119211990397,
      "learning_rate": 2.5807812232100062e-06,
      "loss": 0.178,
      "step": 26540
    },
    {
      "epoch": 0.820273579003468,
      "grad_norm": 0.10086585109572187,
      "learning_rate": 2.5721928999236694e-06,
      "loss": 0.177,
      "step": 26550
    },
    {
      "epoch": 0.8205825332705127,
      "grad_norm": 0.11676183264373326,
      "learning_rate": 2.5636175502362157e-06,
      "loss": 0.1815,
      "step": 26560
    },
    {
      "epoch": 0.8208914875375573,
      "grad_norm": 0.10019563588066743,
      "learning_rate": 2.555055183099608e-06,
      "loss": 0.181,
      "step": 26570
    },
    {
      "epoch": 0.8212004418046018,
      "grad_norm": 0.11278864270987979,
      "learning_rate": 2.5465058074522403e-06,
      "loss": 0.1818,
      "step": 26580
    },
    {
      "epoch": 0.8215093960716465,
      "grad_norm": 0.12229159159220361,
      "learning_rate": 2.537969432218954e-06,
      "loss": 0.1831,
      "step": 26590
    },
    {
      "epoch": 0.8218183503386911,
      "grad_norm": 0.08952713642378439,
      "learning_rate": 2.5294460663110164e-06,
      "loss": 0.1826,
      "step": 26600
    },
    {
      "epoch": 0.8221273046057357,
      "grad_norm": 3.9091861062561364,
      "learning_rate": 2.5209357186261086e-06,
      "loss": 0.1873,
      "step": 26610
    },
    {
      "epoch": 0.8224362588727804,
      "grad_norm": 0.11250191251129393,
      "learning_rate": 2.512438398048335e-06,
      "loss": 0.1837,
      "step": 26620
    },
    {
      "epoch": 0.822745213139825,
      "grad_norm": 0.09671601902846973,
      "learning_rate": 2.5039541134481892e-06,
      "loss": 0.18,
      "step": 26630
    },
    {
      "epoch": 0.8230541674068697,
      "grad_norm": 0.11477264808943827,
      "learning_rate": 2.4954828736825558e-06,
      "loss": 0.1905,
      "step": 26640
    },
    {
      "epoch": 0.8233631216739142,
      "grad_norm": 0.10524931662881779,
      "learning_rate": 2.487024687594713e-06,
      "loss": 0.1779,
      "step": 26650
    },
    {
      "epoch": 0.8236720759409588,
      "grad_norm": 0.1361755300026651,
      "learning_rate": 2.478579564014299e-06,
      "loss": 0.1778,
      "step": 26660
    },
    {
      "epoch": 0.8239810302080035,
      "grad_norm": 0.09541713607264742,
      "learning_rate": 2.470147511757327e-06,
      "loss": 0.187,
      "step": 26670
    },
    {
      "epoch": 0.8242899844750481,
      "grad_norm": 0.09585684365974927,
      "learning_rate": 2.4617285396261542e-06,
      "loss": 0.1817,
      "step": 26680
    },
    {
      "epoch": 0.8245989387420927,
      "grad_norm": 0.0896669875629135,
      "learning_rate": 2.453322656409495e-06,
      "loss": 0.1785,
      "step": 26690
    },
    {
      "epoch": 0.8249078930091374,
      "grad_norm": 0.09318421443377176,
      "learning_rate": 2.444929870882386e-06,
      "loss": 0.1806,
      "step": 26700
    },
    {
      "epoch": 0.8252168472761819,
      "grad_norm": 0.09788279781486932,
      "learning_rate": 2.436550191806206e-06,
      "loss": 0.1872,
      "step": 26710
    },
    {
      "epoch": 0.8255258015432265,
      "grad_norm": 0.12644756525281617,
      "learning_rate": 2.4281836279286385e-06,
      "loss": 0.1821,
      "step": 26720
    },
    {
      "epoch": 0.8258347558102712,
      "grad_norm": 0.11951582000276369,
      "learning_rate": 2.4198301879836865e-06,
      "loss": 0.1812,
      "step": 26730
    },
    {
      "epoch": 0.8261437100773158,
      "grad_norm": 0.12389603664861762,
      "learning_rate": 2.411489880691643e-06,
      "loss": 0.1813,
      "step": 26740
    },
    {
      "epoch": 0.8264526643443604,
      "grad_norm": 0.0962938853711917,
      "learning_rate": 2.4031627147591e-06,
      "loss": 0.1872,
      "step": 26750
    },
    {
      "epoch": 0.8267616186114051,
      "grad_norm": 0.0928120806627728,
      "learning_rate": 2.3948486988789277e-06,
      "loss": 0.1876,
      "step": 26760
    },
    {
      "epoch": 0.8270705728784497,
      "grad_norm": 0.09627266159755687,
      "learning_rate": 2.386547841730263e-06,
      "loss": 0.1881,
      "step": 26770
    },
    {
      "epoch": 0.8273795271454943,
      "grad_norm": 0.10702124431748927,
      "learning_rate": 2.3782601519785193e-06,
      "loss": 0.189,
      "step": 26780
    },
    {
      "epoch": 0.8276884814125389,
      "grad_norm": 0.09047835459394946,
      "learning_rate": 2.369985638275349e-06,
      "loss": 0.1856,
      "step": 26790
    },
    {
      "epoch": 0.8279974356795835,
      "grad_norm": 0.11013789930904871,
      "learning_rate": 2.3617243092586654e-06,
      "loss": 0.183,
      "step": 26800
    },
    {
      "epoch": 0.8283063899466282,
      "grad_norm": 0.11244225824158363,
      "learning_rate": 2.3534761735526037e-06,
      "loss": 0.1799,
      "step": 26810
    },
    {
      "epoch": 0.8286153442136728,
      "grad_norm": 0.10482758814454557,
      "learning_rate": 2.345241239767537e-06,
      "loss": 0.1811,
      "step": 26820
    },
    {
      "epoch": 0.8289242984807174,
      "grad_norm": 0.11023236711917002,
      "learning_rate": 2.3370195165000517e-06,
      "loss": 0.1767,
      "step": 26830
    },
    {
      "epoch": 0.829233252747762,
      "grad_norm": 0.08547987864910896,
      "learning_rate": 2.3288110123329443e-06,
      "loss": 0.1839,
      "step": 26840
    },
    {
      "epoch": 0.8295422070148066,
      "grad_norm": 0.08199369439749662,
      "learning_rate": 2.3206157358352065e-06,
      "loss": 0.1844,
      "step": 26850
    },
    {
      "epoch": 0.8298511612818512,
      "grad_norm": 0.09945310347435897,
      "learning_rate": 2.3124336955620335e-06,
      "loss": 0.177,
      "step": 26860
    },
    {
      "epoch": 0.8301601155488959,
      "grad_norm": 0.11886878897328611,
      "learning_rate": 2.304264900054791e-06,
      "loss": 0.1846,
      "step": 26870
    },
    {
      "epoch": 0.8304690698159405,
      "grad_norm": 0.1053010572103629,
      "learning_rate": 2.29610935784103e-06,
      "loss": 0.1777,
      "step": 26880
    },
    {
      "epoch": 0.8307780240829851,
      "grad_norm": 0.10941271941189175,
      "learning_rate": 2.2879670774344524e-06,
      "loss": 0.1813,
      "step": 26890
    },
    {
      "epoch": 0.8310869783500298,
      "grad_norm": 0.1029952541818937,
      "learning_rate": 2.2798380673349286e-06,
      "loss": 0.1818,
      "step": 26900
    },
    {
      "epoch": 0.8313959326170743,
      "grad_norm": 0.09693366522100368,
      "learning_rate": 2.2717223360284645e-06,
      "loss": 0.18,
      "step": 26910
    },
    {
      "epoch": 0.831704886884119,
      "grad_norm": 0.1034666131190422,
      "learning_rate": 2.263619891987217e-06,
      "loss": 0.185,
      "step": 26920
    },
    {
      "epoch": 0.8320138411511636,
      "grad_norm": 0.10629781939319172,
      "learning_rate": 2.255530743669457e-06,
      "loss": 0.1832,
      "step": 26930
    },
    {
      "epoch": 0.8323227954182082,
      "grad_norm": 0.10468632450440872,
      "learning_rate": 2.2474548995195904e-06,
      "loss": 0.1795,
      "step": 26940
    },
    {
      "epoch": 0.8326317496852529,
      "grad_norm": 0.13031490570482887,
      "learning_rate": 2.2393923679681244e-06,
      "loss": 0.1883,
      "step": 26950
    },
    {
      "epoch": 0.8329407039522975,
      "grad_norm": 0.11430169231322544,
      "learning_rate": 2.2313431574316762e-06,
      "loss": 0.1784,
      "step": 26960
    },
    {
      "epoch": 0.833249658219342,
      "grad_norm": 0.10471546126512164,
      "learning_rate": 2.2233072763129474e-06,
      "loss": 0.1793,
      "step": 26970
    },
    {
      "epoch": 0.8335586124863867,
      "grad_norm": 0.10835248995073464,
      "learning_rate": 2.215284733000738e-06,
      "loss": 0.1885,
      "step": 26980
    },
    {
      "epoch": 0.8338675667534313,
      "grad_norm": 0.08384096342789985,
      "learning_rate": 2.2072755358699137e-06,
      "loss": 0.1836,
      "step": 26990
    },
    {
      "epoch": 0.8341765210204759,
      "grad_norm": 0.08235840301835703,
      "learning_rate": 2.199279693281408e-06,
      "loss": 0.1787,
      "step": 27000
    },
    {
      "epoch": 0.8344854752875206,
      "grad_norm": 0.09118788705424001,
      "learning_rate": 2.191297213582224e-06,
      "loss": 0.1797,
      "step": 27010
    },
    {
      "epoch": 0.8347944295545652,
      "grad_norm": 0.11678461718831075,
      "learning_rate": 2.1833281051053994e-06,
      "loss": 0.1785,
      "step": 27020
    },
    {
      "epoch": 0.8351033838216098,
      "grad_norm": 0.09495668764902095,
      "learning_rate": 2.1753723761700293e-06,
      "loss": 0.1844,
      "step": 27030
    },
    {
      "epoch": 0.8354123380886544,
      "grad_norm": 0.11089841329393607,
      "learning_rate": 2.167430035081229e-06,
      "loss": 0.1816,
      "step": 27040
    },
    {
      "epoch": 0.835721292355699,
      "grad_norm": 0.09614891688568993,
      "learning_rate": 2.159501090130148e-06,
      "loss": 0.1902,
      "step": 27050
    },
    {
      "epoch": 0.8360302466227437,
      "grad_norm": 0.10128158337378926,
      "learning_rate": 2.1515855495939453e-06,
      "loss": 0.1844,
      "step": 27060
    },
    {
      "epoch": 0.8363392008897883,
      "grad_norm": 0.10813236952458656,
      "learning_rate": 2.143683421735787e-06,
      "loss": 0.1783,
      "step": 27070
    },
    {
      "epoch": 0.8366481551568329,
      "grad_norm": 0.1138112838456115,
      "learning_rate": 2.1357947148048334e-06,
      "loss": 0.1792,
      "step": 27080
    },
    {
      "epoch": 0.8369571094238776,
      "grad_norm": 0.09877559199607756,
      "learning_rate": 2.1279194370362506e-06,
      "loss": 0.1842,
      "step": 27090
    },
    {
      "epoch": 0.8372660636909222,
      "grad_norm": 0.12273893003811713,
      "learning_rate": 2.120057596651163e-06,
      "loss": 0.1808,
      "step": 27100
    },
    {
      "epoch": 0.8375750179579667,
      "grad_norm": 0.1268366219120988,
      "learning_rate": 2.11220920185669e-06,
      "loss": 0.1798,
      "step": 27110
    },
    {
      "epoch": 0.8378839722250114,
      "grad_norm": 0.0893402573141918,
      "learning_rate": 2.104374260845896e-06,
      "loss": 0.1819,
      "step": 27120
    },
    {
      "epoch": 0.838192926492056,
      "grad_norm": 0.10505320963658311,
      "learning_rate": 2.096552781797815e-06,
      "loss": 0.1844,
      "step": 27130
    },
    {
      "epoch": 0.8385018807591006,
      "grad_norm": 0.11043174167319414,
      "learning_rate": 2.0887447728774167e-06,
      "loss": 0.1843,
      "step": 27140
    },
    {
      "epoch": 0.8388108350261453,
      "grad_norm": 0.10512945708601514,
      "learning_rate": 2.08095024223562e-06,
      "loss": 0.1823,
      "step": 27150
    },
    {
      "epoch": 0.8391197892931899,
      "grad_norm": 0.09004760228100747,
      "learning_rate": 2.07316919800926e-06,
      "loss": 0.1842,
      "step": 27160
    },
    {
      "epoch": 0.8394287435602344,
      "grad_norm": 0.10913220744232488,
      "learning_rate": 2.0654016483211096e-06,
      "loss": 0.1827,
      "step": 27170
    },
    {
      "epoch": 0.8397376978272791,
      "grad_norm": 0.09261622097557005,
      "learning_rate": 2.0576476012798383e-06,
      "loss": 0.181,
      "step": 27180
    },
    {
      "epoch": 0.8400466520943237,
      "grad_norm": 0.10220607992457943,
      "learning_rate": 2.0499070649800343e-06,
      "loss": 0.1915,
      "step": 27190
    },
    {
      "epoch": 0.8403556063613684,
      "grad_norm": 0.1064276272435423,
      "learning_rate": 2.042180047502169e-06,
      "loss": 0.1844,
      "step": 27200
    },
    {
      "epoch": 0.840664560628413,
      "grad_norm": 0.10315979384687024,
      "learning_rate": 2.0344665569126115e-06,
      "loss": 0.1803,
      "step": 27210
    },
    {
      "epoch": 0.8409735148954576,
      "grad_norm": 0.1091343413173612,
      "learning_rate": 2.0267666012636056e-06,
      "loss": 0.1802,
      "step": 27220
    },
    {
      "epoch": 0.8412824691625023,
      "grad_norm": 0.10350326492163203,
      "learning_rate": 2.019080188593258e-06,
      "loss": 0.1798,
      "step": 27230
    },
    {
      "epoch": 0.8415914234295468,
      "grad_norm": 0.0979584507564618,
      "learning_rate": 2.011407326925557e-06,
      "loss": 0.1793,
      "step": 27240
    },
    {
      "epoch": 0.8419003776965914,
      "grad_norm": 0.08892882254327583,
      "learning_rate": 2.0037480242703226e-06,
      "loss": 0.1866,
      "step": 27250
    },
    {
      "epoch": 0.8422093319636361,
      "grad_norm": 0.09177020165375138,
      "learning_rate": 1.99610228862324e-06,
      "loss": 0.1907,
      "step": 27260
    },
    {
      "epoch": 0.8425182862306807,
      "grad_norm": 0.1329194927914558,
      "learning_rate": 1.988470127965818e-06,
      "loss": 0.1792,
      "step": 27270
    },
    {
      "epoch": 0.8428272404977253,
      "grad_norm": 0.10081756590933665,
      "learning_rate": 1.9808515502653968e-06,
      "loss": 0.1868,
      "step": 27280
    },
    {
      "epoch": 0.84313619476477,
      "grad_norm": 0.08951332764717372,
      "learning_rate": 1.9732465634751444e-06,
      "loss": 0.1798,
      "step": 27290
    },
    {
      "epoch": 0.8434451490318146,
      "grad_norm": 0.09778559595819382,
      "learning_rate": 1.9656551755340308e-06,
      "loss": 0.1884,
      "step": 27300
    },
    {
      "epoch": 0.8437541032988591,
      "grad_norm": 0.11482275963719252,
      "learning_rate": 1.9580773943668346e-06,
      "loss": 0.1801,
      "step": 27310
    },
    {
      "epoch": 0.8440630575659038,
      "grad_norm": 0.12016895786340931,
      "learning_rate": 1.9505132278841364e-06,
      "loss": 0.1786,
      "step": 27320
    },
    {
      "epoch": 0.8443720118329484,
      "grad_norm": 0.09166064739050628,
      "learning_rate": 1.9429626839822886e-06,
      "loss": 0.1845,
      "step": 27330
    },
    {
      "epoch": 0.8446809660999931,
      "grad_norm": 0.09642933193736093,
      "learning_rate": 1.9354257705434417e-06,
      "loss": 0.1818,
      "step": 27340
    },
    {
      "epoch": 0.8449899203670377,
      "grad_norm": 0.08630323863341553,
      "learning_rate": 1.9279024954355e-06,
      "loss": 0.188,
      "step": 27350
    },
    {
      "epoch": 0.8452988746340823,
      "grad_norm": 0.09861229733485333,
      "learning_rate": 1.920392866512146e-06,
      "loss": 0.1831,
      "step": 27360
    },
    {
      "epoch": 0.845607828901127,
      "grad_norm": 0.12747007442421499,
      "learning_rate": 1.9128968916127993e-06,
      "loss": 0.186,
      "step": 27370
    },
    {
      "epoch": 0.8459167831681715,
      "grad_norm": 0.10135488698645409,
      "learning_rate": 1.9054145785626448e-06,
      "loss": 0.1839,
      "step": 27380
    },
    {
      "epoch": 0.8462257374352161,
      "grad_norm": 0.12100329091546554,
      "learning_rate": 1.8979459351725893e-06,
      "loss": 0.1782,
      "step": 27390
    },
    {
      "epoch": 0.8465346917022608,
      "grad_norm": 0.11794669930626465,
      "learning_rate": 1.8904909692392808e-06,
      "loss": 0.177,
      "step": 27400
    },
    {
      "epoch": 0.8468436459693054,
      "grad_norm": 0.10304527310776555,
      "learning_rate": 1.883049688545081e-06,
      "loss": 0.1834,
      "step": 27410
    },
    {
      "epoch": 0.84715260023635,
      "grad_norm": 0.11079993470676724,
      "learning_rate": 1.8756221008580742e-06,
      "loss": 0.178,
      "step": 27420
    },
    {
      "epoch": 0.8474615545033947,
      "grad_norm": 0.10529550594569145,
      "learning_rate": 1.86820821393204e-06,
      "loss": 0.1868,
      "step": 27430
    },
    {
      "epoch": 0.8477705087704392,
      "grad_norm": 0.09865252971963585,
      "learning_rate": 1.8608080355064654e-06,
      "loss": 0.1805,
      "step": 27440
    },
    {
      "epoch": 0.8480794630374838,
      "grad_norm": 0.11146268569008994,
      "learning_rate": 1.8534215733065195e-06,
      "loss": 0.187,
      "step": 27450
    },
    {
      "epoch": 0.8483884173045285,
      "grad_norm": 0.09285684930021669,
      "learning_rate": 1.8460488350430543e-06,
      "loss": 0.1798,
      "step": 27460
    },
    {
      "epoch": 0.8486973715715731,
      "grad_norm": 0.09441379882019936,
      "learning_rate": 1.8386898284126014e-06,
      "loss": 0.1825,
      "step": 27470
    },
    {
      "epoch": 0.8490063258386178,
      "grad_norm": 0.09828897913817153,
      "learning_rate": 1.8313445610973473e-06,
      "loss": 0.1798,
      "step": 27480
    },
    {
      "epoch": 0.8493152801056624,
      "grad_norm": 0.09712604689226159,
      "learning_rate": 1.8240130407651468e-06,
      "loss": 0.1811,
      "step": 27490
    },
    {
      "epoch": 0.849624234372707,
      "grad_norm": 0.09650863253736786,
      "learning_rate": 1.816695275069496e-06,
      "loss": 0.1903,
      "step": 27500
    },
    {
      "epoch": 0.8499331886397516,
      "grad_norm": 0.10159140277852872,
      "learning_rate": 1.8093912716495309e-06,
      "loss": 0.1823,
      "step": 27510
    },
    {
      "epoch": 0.8502421429067962,
      "grad_norm": 0.10257575057222768,
      "learning_rate": 1.8021010381300306e-06,
      "loss": 0.1842,
      "step": 27520
    },
    {
      "epoch": 0.8505510971738408,
      "grad_norm": 0.11416021507753527,
      "learning_rate": 1.7948245821213927e-06,
      "loss": 0.1819,
      "step": 27530
    },
    {
      "epoch": 0.8508600514408855,
      "grad_norm": 0.1096439031539581,
      "learning_rate": 1.7875619112196272e-06,
      "loss": 0.1806,
      "step": 27540
    },
    {
      "epoch": 0.8511690057079301,
      "grad_norm": 0.12544886724105817,
      "learning_rate": 1.7803130330063683e-06,
      "loss": 0.185,
      "step": 27550
    },
    {
      "epoch": 0.8514779599749747,
      "grad_norm": 0.08653214946357521,
      "learning_rate": 1.7730779550488357e-06,
      "loss": 0.1809,
      "step": 27560
    },
    {
      "epoch": 0.8517869142420194,
      "grad_norm": 0.09734798167744035,
      "learning_rate": 1.7658566848998564e-06,
      "loss": 0.179,
      "step": 27570
    },
    {
      "epoch": 0.8520958685090639,
      "grad_norm": 0.11337981038383727,
      "learning_rate": 1.7586492300978302e-06,
      "loss": 0.1794,
      "step": 27580
    },
    {
      "epoch": 0.8524048227761085,
      "grad_norm": 0.12537202799438474,
      "learning_rate": 1.7514555981667502e-06,
      "loss": 0.1771,
      "step": 27590
    },
    {
      "epoch": 0.8527137770431532,
      "grad_norm": 0.09630437412885685,
      "learning_rate": 1.744275796616161e-06,
      "loss": 0.1803,
      "step": 27600
    },
    {
      "epoch": 0.8530227313101978,
      "grad_norm": 0.13605555272709954,
      "learning_rate": 1.7371098329411889e-06,
      "loss": 0.1886,
      "step": 27610
    },
    {
      "epoch": 0.8533316855772425,
      "grad_norm": 0.10014710385574904,
      "learning_rate": 1.7299577146224965e-06,
      "loss": 0.1879,
      "step": 27620
    },
    {
      "epoch": 0.8536406398442871,
      "grad_norm": 0.1153995981947399,
      "learning_rate": 1.7228194491263083e-06,
      "loss": 0.192,
      "step": 27630
    },
    {
      "epoch": 0.8539495941113316,
      "grad_norm": 0.10125870202744387,
      "learning_rate": 1.7156950439043767e-06,
      "loss": 0.1804,
      "step": 27640
    },
    {
      "epoch": 0.8542585483783763,
      "grad_norm": 0.10037355542983763,
      "learning_rate": 1.7085845063939925e-06,
      "loss": 0.1795,
      "step": 27650
    },
    {
      "epoch": 0.8545675026454209,
      "grad_norm": 0.10366857000831413,
      "learning_rate": 1.7014878440179631e-06,
      "loss": 0.1815,
      "step": 27660
    },
    {
      "epoch": 0.8548764569124655,
      "grad_norm": 0.151792790296212,
      "learning_rate": 1.6944050641846193e-06,
      "loss": 0.1844,
      "step": 27670
    },
    {
      "epoch": 0.8551854111795102,
      "grad_norm": 0.11647008610684406,
      "learning_rate": 1.687336174287793e-06,
      "loss": 0.1799,
      "step": 27680
    },
    {
      "epoch": 0.8554943654465548,
      "grad_norm": 0.11979148085812992,
      "learning_rate": 1.6802811817068147e-06,
      "loss": 0.1862,
      "step": 27690
    },
    {
      "epoch": 0.8558033197135994,
      "grad_norm": 0.09533420254047427,
      "learning_rate": 1.6732400938065168e-06,
      "loss": 0.1822,
      "step": 27700
    },
    {
      "epoch": 0.856112273980644,
      "grad_norm": 0.09103012165005812,
      "learning_rate": 1.666212917937206e-06,
      "loss": 0.182,
      "step": 27710
    },
    {
      "epoch": 0.8564212282476886,
      "grad_norm": 0.10631327289049497,
      "learning_rate": 1.6591996614346743e-06,
      "loss": 0.1873,
      "step": 27720
    },
    {
      "epoch": 0.8567301825147332,
      "grad_norm": 0.09131778998475253,
      "learning_rate": 1.652200331620179e-06,
      "loss": 0.1856,
      "step": 27730
    },
    {
      "epoch": 0.8570391367817779,
      "grad_norm": 0.09480692081231779,
      "learning_rate": 1.645214935800438e-06,
      "loss": 0.1804,
      "step": 27740
    },
    {
      "epoch": 0.8573480910488225,
      "grad_norm": 0.10350323929665772,
      "learning_rate": 1.6382434812676206e-06,
      "loss": 0.184,
      "step": 27750
    },
    {
      "epoch": 0.8576570453158672,
      "grad_norm": 0.09218235957061148,
      "learning_rate": 1.631285975299353e-06,
      "loss": 0.1806,
      "step": 27760
    },
    {
      "epoch": 0.8579659995829118,
      "grad_norm": 0.10087742572076434,
      "learning_rate": 1.6243424251586892e-06,
      "loss": 0.1798,
      "step": 27770
    },
    {
      "epoch": 0.8582749538499563,
      "grad_norm": 0.10238011832918574,
      "learning_rate": 1.617412838094124e-06,
      "loss": 0.1898,
      "step": 27780
    },
    {
      "epoch": 0.858583908117001,
      "grad_norm": 0.10578972574701347,
      "learning_rate": 1.6104972213395658e-06,
      "loss": 0.1821,
      "step": 27790
    },
    {
      "epoch": 0.8588928623840456,
      "grad_norm": 0.09645345818839185,
      "learning_rate": 1.60359558211435e-06,
      "loss": 0.1796,
      "step": 27800
    },
    {
      "epoch": 0.8592018166510902,
      "grad_norm": 0.12193528467144923,
      "learning_rate": 1.5967079276232083e-06,
      "loss": 0.1865,
      "step": 27810
    },
    {
      "epoch": 0.8595107709181349,
      "grad_norm": 0.12029400636170644,
      "learning_rate": 1.5898342650562876e-06,
      "loss": 0.1811,
      "step": 27820
    },
    {
      "epoch": 0.8598197251851795,
      "grad_norm": 0.09458211987564656,
      "learning_rate": 1.5829746015891127e-06,
      "loss": 0.1804,
      "step": 27830
    },
    {
      "epoch": 0.860128679452224,
      "grad_norm": 0.10716056428475798,
      "learning_rate": 1.576128944382611e-06,
      "loss": 0.191,
      "step": 27840
    },
    {
      "epoch": 0.8604376337192687,
      "grad_norm": 0.11321245368661662,
      "learning_rate": 1.569297300583074e-06,
      "loss": 0.1854,
      "step": 27850
    },
    {
      "epoch": 0.8607465879863133,
      "grad_norm": 0.09429588332951563,
      "learning_rate": 1.5624796773221755e-06,
      "loss": 0.1846,
      "step": 27860
    },
    {
      "epoch": 0.8610555422533579,
      "grad_norm": 0.09995022017383527,
      "learning_rate": 1.5556760817169424e-06,
      "loss": 0.1876,
      "step": 27870
    },
    {
      "epoch": 0.8613644965204026,
      "grad_norm": 0.10694670781507666,
      "learning_rate": 1.5488865208697717e-06,
      "loss": 0.1828,
      "step": 27880
    },
    {
      "epoch": 0.8616734507874472,
      "grad_norm": 0.10492273533689082,
      "learning_rate": 1.542111001868392e-06,
      "loss": 0.1778,
      "step": 27890
    },
    {
      "epoch": 0.8619824050544919,
      "grad_norm": 0.09622600235705367,
      "learning_rate": 1.5353495317858923e-06,
      "loss": 0.1824,
      "step": 27900
    },
    {
      "epoch": 0.8622913593215364,
      "grad_norm": 0.09557007187218863,
      "learning_rate": 1.5286021176806797e-06,
      "loss": 0.1892,
      "step": 27910
    },
    {
      "epoch": 0.862600313588581,
      "grad_norm": 0.10837043908556614,
      "learning_rate": 1.5218687665964948e-06,
      "loss": 0.1813,
      "step": 27920
    },
    {
      "epoch": 0.8629092678556257,
      "grad_norm": 0.10418243323508425,
      "learning_rate": 1.5151494855623998e-06,
      "loss": 0.1833,
      "step": 27930
    },
    {
      "epoch": 0.8632182221226703,
      "grad_norm": 0.10555062612200378,
      "learning_rate": 1.5084442815927624e-06,
      "loss": 0.1838,
      "step": 27940
    },
    {
      "epoch": 0.8635271763897149,
      "grad_norm": 0.09854942097035761,
      "learning_rate": 1.5017531616872633e-06,
      "loss": 0.1847,
      "step": 27950
    },
    {
      "epoch": 0.8638361306567596,
      "grad_norm": 0.11210731673036076,
      "learning_rate": 1.4950761328308732e-06,
      "loss": 0.1798,
      "step": 27960
    },
    {
      "epoch": 0.8641450849238042,
      "grad_norm": 0.09776934590645367,
      "learning_rate": 1.4884132019938584e-06,
      "loss": 0.1787,
      "step": 27970
    },
    {
      "epoch": 0.8644540391908487,
      "grad_norm": 0.09231585913918622,
      "learning_rate": 1.4817643761317618e-06,
      "loss": 0.187,
      "step": 27980
    },
    {
      "epoch": 0.8647629934578934,
      "grad_norm": 0.09509297143759726,
      "learning_rate": 1.4751296621854093e-06,
      "loss": 0.1904,
      "step": 27990
    },
    {
      "epoch": 0.865071947724938,
      "grad_norm": 0.09832590235087585,
      "learning_rate": 1.4685090670808876e-06,
      "loss": 0.1876,
      "step": 28000
    },
    {
      "epoch": 0.8653809019919826,
      "grad_norm": 0.0848024447055945,
      "learning_rate": 1.461902597729553e-06,
      "loss": 0.1795,
      "step": 28010
    },
    {
      "epoch": 0.8656898562590273,
      "grad_norm": 0.0978696028892343,
      "learning_rate": 1.4553102610280056e-06,
      "loss": 0.1858,
      "step": 28020
    },
    {
      "epoch": 0.8659988105260719,
      "grad_norm": 0.10249821211189425,
      "learning_rate": 1.4487320638581053e-06,
      "loss": 0.1803,
      "step": 28030
    },
    {
      "epoch": 0.8663077647931166,
      "grad_norm": 0.10341479468088537,
      "learning_rate": 1.4421680130869342e-06,
      "loss": 0.1804,
      "step": 28040
    },
    {
      "epoch": 0.8666167190601611,
      "grad_norm": 0.10514413389059125,
      "learning_rate": 1.4356181155668258e-06,
      "loss": 0.183,
      "step": 28050
    },
    {
      "epoch": 0.8669256733272057,
      "grad_norm": 0.14211967646179546,
      "learning_rate": 1.4290823781353224e-06,
      "loss": 0.1889,
      "step": 28060
    },
    {
      "epoch": 0.8672346275942504,
      "grad_norm": 0.09157833309032826,
      "learning_rate": 1.4225608076151964e-06,
      "loss": 0.1815,
      "step": 28070
    },
    {
      "epoch": 0.867543581861295,
      "grad_norm": 0.1257052164852312,
      "learning_rate": 1.4160534108144218e-06,
      "loss": 0.1883,
      "step": 28080
    },
    {
      "epoch": 0.8678525361283396,
      "grad_norm": 0.11028424163893609,
      "learning_rate": 1.409560194526185e-06,
      "loss": 0.1843,
      "step": 28090
    },
    {
      "epoch": 0.8681614903953843,
      "grad_norm": 0.09495208514787493,
      "learning_rate": 1.4030811655288583e-06,
      "loss": 0.1795,
      "step": 28100
    },
    {
      "epoch": 0.8684704446624288,
      "grad_norm": 0.09347743867915254,
      "learning_rate": 1.3966163305860168e-06,
      "loss": 0.182,
      "step": 28110
    },
    {
      "epoch": 0.8687793989294734,
      "grad_norm": 0.10122675508013164,
      "learning_rate": 1.390165696446407e-06,
      "loss": 0.184,
      "step": 28120
    },
    {
      "epoch": 0.8690883531965181,
      "grad_norm": 0.08562657360948904,
      "learning_rate": 1.383729269843959e-06,
      "loss": 0.1847,
      "step": 28130
    },
    {
      "epoch": 0.8693973074635627,
      "grad_norm": 0.13636203679532566,
      "learning_rate": 1.3773070574977663e-06,
      "loss": 0.1896,
      "step": 28140
    },
    {
      "epoch": 0.8697062617306073,
      "grad_norm": 0.09964190624577135,
      "learning_rate": 1.3708990661120812e-06,
      "loss": 0.1868,
      "step": 28150
    },
    {
      "epoch": 0.870015215997652,
      "grad_norm": 0.08693331717577828,
      "learning_rate": 1.3645053023763227e-06,
      "loss": 0.1878,
      "step": 28160
    },
    {
      "epoch": 0.8703241702646966,
      "grad_norm": 0.11100386596048474,
      "learning_rate": 1.3581257729650448e-06,
      "loss": 0.1832,
      "step": 28170
    },
    {
      "epoch": 0.8706331245317412,
      "grad_norm": 0.1752273090891548,
      "learning_rate": 1.3517604845379455e-06,
      "loss": 0.1897,
      "step": 28180
    },
    {
      "epoch": 0.8709420787987858,
      "grad_norm": 0.08455187605101577,
      "learning_rate": 1.3454094437398595e-06,
      "loss": 0.1792,
      "step": 28190
    },
    {
      "epoch": 0.8712510330658304,
      "grad_norm": 0.11473654980052184,
      "learning_rate": 1.3390726572007483e-06,
      "loss": 0.1891,
      "step": 28200
    },
    {
      "epoch": 0.8715599873328751,
      "grad_norm": 0.09969500923639464,
      "learning_rate": 1.3327501315356838e-06,
      "loss": 0.1801,
      "step": 28210
    },
    {
      "epoch": 0.8718689415999197,
      "grad_norm": 0.13141736816992747,
      "learning_rate": 1.3264418733448647e-06,
      "loss": 0.1806,
      "step": 28220
    },
    {
      "epoch": 0.8721778958669643,
      "grad_norm": 0.09542391208819641,
      "learning_rate": 1.320147889213585e-06,
      "loss": 0.1784,
      "step": 28230
    },
    {
      "epoch": 0.872486850134009,
      "grad_norm": 0.10039030875895556,
      "learning_rate": 1.313868185712247e-06,
      "loss": 0.1783,
      "step": 28240
    },
    {
      "epoch": 0.8727958044010535,
      "grad_norm": 0.10101882821079176,
      "learning_rate": 1.3076027693963322e-06,
      "loss": 0.1851,
      "step": 28250
    },
    {
      "epoch": 0.8731047586680981,
      "grad_norm": 0.10052606067948769,
      "learning_rate": 1.3013516468064218e-06,
      "loss": 0.1783,
      "step": 28260
    },
    {
      "epoch": 0.8734137129351428,
      "grad_norm": 0.10399175254378158,
      "learning_rate": 1.2951148244681626e-06,
      "loss": 0.18,
      "step": 28270
    },
    {
      "epoch": 0.8737226672021874,
      "grad_norm": 0.10732087087718088,
      "learning_rate": 1.2888923088922866e-06,
      "loss": 0.1875,
      "step": 28280
    },
    {
      "epoch": 0.874031621469232,
      "grad_norm": 0.12140957936172334,
      "learning_rate": 1.2826841065745759e-06,
      "loss": 0.1799,
      "step": 28290
    },
    {
      "epoch": 0.8743405757362767,
      "grad_norm": 0.13048644050569663,
      "learning_rate": 1.2764902239958848e-06,
      "loss": 0.1965,
      "step": 28300
    },
    {
      "epoch": 0.8746495300033212,
      "grad_norm": 0.08823311951562303,
      "learning_rate": 1.2703106676221078e-06,
      "loss": 0.1899,
      "step": 28310
    },
    {
      "epoch": 0.8749584842703659,
      "grad_norm": 0.09118016851784427,
      "learning_rate": 1.2641454439041944e-06,
      "loss": 0.1795,
      "step": 28320
    },
    {
      "epoch": 0.8752674385374105,
      "grad_norm": 0.09712283858601073,
      "learning_rate": 1.2579945592781183e-06,
      "loss": 0.1779,
      "step": 28330
    },
    {
      "epoch": 0.8755763928044551,
      "grad_norm": 0.09555532218485613,
      "learning_rate": 1.2518580201649e-06,
      "loss": 0.1854,
      "step": 28340
    },
    {
      "epoch": 0.8758853470714998,
      "grad_norm": 0.10978721326611847,
      "learning_rate": 1.2457358329705714e-06,
      "loss": 0.1919,
      "step": 28350
    },
    {
      "epoch": 0.8761943013385444,
      "grad_norm": 0.09877770628622538,
      "learning_rate": 1.2396280040861923e-06,
      "loss": 0.1793,
      "step": 28360
    },
    {
      "epoch": 0.876503255605589,
      "grad_norm": 0.09840061995672636,
      "learning_rate": 1.2335345398878256e-06,
      "loss": 0.1849,
      "step": 28370
    },
    {
      "epoch": 0.8768122098726336,
      "grad_norm": 0.10842132193982282,
      "learning_rate": 1.2274554467365429e-06,
      "loss": 0.1859,
      "step": 28380
    },
    {
      "epoch": 0.8771211641396782,
      "grad_norm": 0.09453393531545135,
      "learning_rate": 1.2213907309784144e-06,
      "loss": 0.1812,
      "step": 28390
    },
    {
      "epoch": 0.8774301184067228,
      "grad_norm": 0.09734470135064001,
      "learning_rate": 1.2153403989444977e-06,
      "loss": 0.184,
      "step": 28400
    },
    {
      "epoch": 0.8777390726737675,
      "grad_norm": 0.10300073893610687,
      "learning_rate": 1.2093044569508371e-06,
      "loss": 0.1869,
      "step": 28410
    },
    {
      "epoch": 0.8780480269408121,
      "grad_norm": 0.09515505870941551,
      "learning_rate": 1.2032829112984573e-06,
      "loss": 0.1837,
      "step": 28420
    },
    {
      "epoch": 0.8783569812078567,
      "grad_norm": 0.10274256747449963,
      "learning_rate": 1.1972757682733497e-06,
      "loss": 0.1797,
      "step": 28430
    },
    {
      "epoch": 0.8786659354749013,
      "grad_norm": 0.11855058908400182,
      "learning_rate": 1.191283034146472e-06,
      "loss": 0.1818,
      "step": 28440
    },
    {
      "epoch": 0.8789748897419459,
      "grad_norm": 0.09452027487374964,
      "learning_rate": 1.1853047151737461e-06,
      "loss": 0.1805,
      "step": 28450
    },
    {
      "epoch": 0.8792838440089906,
      "grad_norm": 0.09298885395420112,
      "learning_rate": 1.179340817596035e-06,
      "loss": 0.1805,
      "step": 28460
    },
    {
      "epoch": 0.8795927982760352,
      "grad_norm": 0.11865235122609638,
      "learning_rate": 1.1733913476391572e-06,
      "loss": 0.1766,
      "step": 28470
    },
    {
      "epoch": 0.8799017525430798,
      "grad_norm": 0.09375499954975781,
      "learning_rate": 1.1674563115138609e-06,
      "loss": 0.1826,
      "step": 28480
    },
    {
      "epoch": 0.8802107068101245,
      "grad_norm": 0.10916748571180004,
      "learning_rate": 1.1615357154158367e-06,
      "loss": 0.1806,
      "step": 28490
    },
    {
      "epoch": 0.8805196610771691,
      "grad_norm": 0.10295313678788805,
      "learning_rate": 1.1556295655256894e-06,
      "loss": 0.1838,
      "step": 28500
    },
    {
      "epoch": 0.8808286153442136,
      "grad_norm": 0.1330241211205987,
      "learning_rate": 1.149737868008955e-06,
      "loss": 0.1912,
      "step": 28510
    },
    {
      "epoch": 0.8811375696112583,
      "grad_norm": 0.10395038689526555,
      "learning_rate": 1.1438606290160703e-06,
      "loss": 0.1808,
      "step": 28520
    },
    {
      "epoch": 0.8814465238783029,
      "grad_norm": 0.10450541582801408,
      "learning_rate": 1.1379978546823911e-06,
      "loss": 0.1851,
      "step": 28530
    },
    {
      "epoch": 0.8817554781453475,
      "grad_norm": 0.11449832943291852,
      "learning_rate": 1.1321495511281616e-06,
      "loss": 0.191,
      "step": 28540
    },
    {
      "epoch": 0.8820644324123922,
      "grad_norm": 0.088844180214581,
      "learning_rate": 1.1263157244585315e-06,
      "loss": 0.1818,
      "step": 28550
    },
    {
      "epoch": 0.8823733866794368,
      "grad_norm": 0.1045786256839124,
      "learning_rate": 1.1204963807635243e-06,
      "loss": 0.1819,
      "step": 28560
    },
    {
      "epoch": 0.8826823409464813,
      "grad_norm": 0.0917028822218501,
      "learning_rate": 1.1146915261180574e-06,
      "loss": 0.1862,
      "step": 28570
    },
    {
      "epoch": 0.882991295213526,
      "grad_norm": 0.09509239739242586,
      "learning_rate": 1.1089011665819132e-06,
      "loss": 0.1798,
      "step": 28580
    },
    {
      "epoch": 0.8833002494805706,
      "grad_norm": 0.08936307446247876,
      "learning_rate": 1.1031253081997517e-06,
      "loss": 0.1862,
      "step": 28590
    },
    {
      "epoch": 0.8836092037476153,
      "grad_norm": 0.09163361020269918,
      "learning_rate": 1.0973639570010845e-06,
      "loss": 0.1796,
      "step": 28600
    },
    {
      "epoch": 0.8839181580146599,
      "grad_norm": 0.10171996435464667,
      "learning_rate": 1.0916171190002844e-06,
      "loss": 0.179,
      "step": 28610
    },
    {
      "epoch": 0.8842271122817045,
      "grad_norm": 0.10084838396440059,
      "learning_rate": 1.085884800196577e-06,
      "loss": 0.1874,
      "step": 28620
    },
    {
      "epoch": 0.8845360665487492,
      "grad_norm": 0.09798025936235355,
      "learning_rate": 1.0801670065740226e-06,
      "loss": 0.1763,
      "step": 28630
    },
    {
      "epoch": 0.8848450208157937,
      "grad_norm": 0.09644348296261553,
      "learning_rate": 1.0744637441015248e-06,
      "loss": 0.1812,
      "step": 28640
    },
    {
      "epoch": 0.8851539750828383,
      "grad_norm": 0.09165574629779512,
      "learning_rate": 1.0687750187328176e-06,
      "loss": 0.1763,
      "step": 28650
    },
    {
      "epoch": 0.885462929349883,
      "grad_norm": 0.09126176597133256,
      "learning_rate": 1.0631008364064575e-06,
      "loss": 0.1819,
      "step": 28660
    },
    {
      "epoch": 0.8857718836169276,
      "grad_norm": 0.10177329966346589,
      "learning_rate": 1.0574412030458163e-06,
      "loss": 0.1843,
      "step": 28670
    },
    {
      "epoch": 0.8860808378839722,
      "grad_norm": 0.09142026877755322,
      "learning_rate": 1.0517961245590862e-06,
      "loss": 0.1783,
      "step": 28680
    },
    {
      "epoch": 0.8863897921510169,
      "grad_norm": 0.10315681333619384,
      "learning_rate": 1.0461656068392568e-06,
      "loss": 0.1794,
      "step": 28690
    },
    {
      "epoch": 0.8866987464180615,
      "grad_norm": 0.10808907562664705,
      "learning_rate": 1.0405496557641241e-06,
      "loss": 0.1844,
      "step": 28700
    },
    {
      "epoch": 0.887007700685106,
      "grad_norm": 0.08912185514779751,
      "learning_rate": 1.0349482771962727e-06,
      "loss": 0.1796,
      "step": 28710
    },
    {
      "epoch": 0.8873166549521507,
      "grad_norm": 0.10877406337780003,
      "learning_rate": 1.0293614769830794e-06,
      "loss": 0.1893,
      "step": 28720
    },
    {
      "epoch": 0.8876256092191953,
      "grad_norm": 0.10967357711705748,
      "learning_rate": 1.0237892609566962e-06,
      "loss": 0.1807,
      "step": 28730
    },
    {
      "epoch": 0.88793456348624,
      "grad_norm": 0.08563485597898633,
      "learning_rate": 1.0182316349340565e-06,
      "loss": 0.1811,
      "step": 28740
    },
    {
      "epoch": 0.8882435177532846,
      "grad_norm": 0.10322718457396875,
      "learning_rate": 1.0126886047168592e-06,
      "loss": 0.1772,
      "step": 28750
    },
    {
      "epoch": 0.8885524720203292,
      "grad_norm": 0.0910924622392542,
      "learning_rate": 1.007160176091571e-06,
      "loss": 0.1842,
      "step": 28760
    },
    {
      "epoch": 0.8888614262873739,
      "grad_norm": 0.09560388790456413,
      "learning_rate": 1.001646354829407e-06,
      "loss": 0.1827,
      "step": 28770
    },
    {
      "epoch": 0.8891703805544184,
      "grad_norm": 0.1110580301572007,
      "learning_rate": 9.961471466863442e-07,
      "loss": 0.1808,
      "step": 28780
    },
    {
      "epoch": 0.889479334821463,
      "grad_norm": 0.09320975049039067,
      "learning_rate": 9.906625574030942e-07,
      "loss": 0.1839,
      "step": 28790
    },
    {
      "epoch": 0.8897882890885077,
      "grad_norm": 0.08565687704294897,
      "learning_rate": 9.85192592705117e-07,
      "loss": 0.1838,
      "step": 28800
    },
    {
      "epoch": 0.8900972433555523,
      "grad_norm": 0.0949658311326804,
      "learning_rate": 9.797372583026e-07,
      "loss": 0.1865,
      "step": 28810
    },
    {
      "epoch": 0.8904061976225969,
      "grad_norm": 0.10439513176418971,
      "learning_rate": 9.742965598904608e-07,
      "loss": 0.1819,
      "step": 28820
    },
    {
      "epoch": 0.8907151518896416,
      "grad_norm": 0.10962135114351249,
      "learning_rate": 9.688705031483347e-07,
      "loss": 0.1847,
      "step": 28830
    },
    {
      "epoch": 0.8910241061566861,
      "grad_norm": 0.08592384740259364,
      "learning_rate": 9.634590937405742e-07,
      "loss": 0.1875,
      "step": 28840
    },
    {
      "epoch": 0.8913330604237307,
      "grad_norm": 0.08823603395008275,
      "learning_rate": 9.58062337316245e-07,
      "loss": 0.1824,
      "step": 28850
    },
    {
      "epoch": 0.8916420146907754,
      "grad_norm": 0.09501198504753797,
      "learning_rate": 9.526802395091111e-07,
      "loss": 0.1779,
      "step": 28860
    },
    {
      "epoch": 0.89195096895782,
      "grad_norm": 0.10401951929700107,
      "learning_rate": 9.473128059376345e-07,
      "loss": 0.1849,
      "step": 28870
    },
    {
      "epoch": 0.8922599232248647,
      "grad_norm": 0.09717035872126756,
      "learning_rate": 9.419600422049685e-07,
      "loss": 0.1753,
      "step": 28880
    },
    {
      "epoch": 0.8925688774919093,
      "grad_norm": 0.10169162217251684,
      "learning_rate": 9.366219538989595e-07,
      "loss": 0.1814,
      "step": 28890
    },
    {
      "epoch": 0.8928778317589539,
      "grad_norm": 0.11756627345092277,
      "learning_rate": 9.312985465921237e-07,
      "loss": 0.187,
      "step": 28900
    },
    {
      "epoch": 0.8931867860259985,
      "grad_norm": 0.09539420893620165,
      "learning_rate": 9.259898258416588e-07,
      "loss": 0.1878,
      "step": 28910
    },
    {
      "epoch": 0.8934957402930431,
      "grad_norm": 0.09122934330826722,
      "learning_rate": 9.20695797189427e-07,
      "loss": 0.1862,
      "step": 28920
    },
    {
      "epoch": 0.8938046945600877,
      "grad_norm": 0.10825014330310029,
      "learning_rate": 9.15416466161959e-07,
      "loss": 0.1807,
      "step": 28930
    },
    {
      "epoch": 0.8941136488271324,
      "grad_norm": 0.07922736658973209,
      "learning_rate": 9.101518382704316e-07,
      "loss": 0.1873,
      "step": 28940
    },
    {
      "epoch": 0.894422603094177,
      "grad_norm": 0.11104154065294454,
      "learning_rate": 9.04901919010685e-07,
      "loss": 0.1877,
      "step": 28950
    },
    {
      "epoch": 0.8947315573612216,
      "grad_norm": 0.08437513465972055,
      "learning_rate": 8.996667138631937e-07,
      "loss": 0.1798,
      "step": 28960
    },
    {
      "epoch": 0.8950405116282663,
      "grad_norm": 0.09917149705694504,
      "learning_rate": 8.944462282930827e-07,
      "loss": 0.1835,
      "step": 28970
    },
    {
      "epoch": 0.8953494658953108,
      "grad_norm": 0.09337580339209801,
      "learning_rate": 8.89240467750101e-07,
      "loss": 0.1829,
      "step": 28980
    },
    {
      "epoch": 0.8956584201623554,
      "grad_norm": 0.10589561235209922,
      "learning_rate": 8.840494376686348e-07,
      "loss": 0.1795,
      "step": 28990
    },
    {
      "epoch": 0.8959673744294001,
      "grad_norm": 0.09909898805593338,
      "learning_rate": 8.788731434676827e-07,
      "loss": 0.1784,
      "step": 29000
    },
    {
      "epoch": 0.8962763286964447,
      "grad_norm": 0.1153548371364837,
      "learning_rate": 8.737115905508741e-07,
      "loss": 0.18,
      "step": 29010
    },
    {
      "epoch": 0.8965852829634894,
      "grad_norm": 0.10130362018181113,
      "learning_rate": 8.685647843064337e-07,
      "loss": 0.1808,
      "step": 29020
    },
    {
      "epoch": 0.896894237230534,
      "grad_norm": 0.10985169077075072,
      "learning_rate": 8.634327301072054e-07,
      "loss": 0.1801,
      "step": 29030
    },
    {
      "epoch": 0.8972031914975785,
      "grad_norm": 0.08066976093560242,
      "learning_rate": 8.583154333106269e-07,
      "loss": 0.1851,
      "step": 29040
    },
    {
      "epoch": 0.8975121457646232,
      "grad_norm": 0.10468372757573076,
      "learning_rate": 8.532128992587268e-07,
      "loss": 0.1785,
      "step": 29050
    },
    {
      "epoch": 0.8978211000316678,
      "grad_norm": 0.1102368775175334,
      "learning_rate": 8.481251332781343e-07,
      "loss": 0.1826,
      "step": 29060
    },
    {
      "epoch": 0.8981300542987124,
      "grad_norm": 0.12688113008564225,
      "learning_rate": 8.430521406800473e-07,
      "loss": 0.1791,
      "step": 29070
    },
    {
      "epoch": 0.8984390085657571,
      "grad_norm": 0.10385916131286953,
      "learning_rate": 8.379939267602549e-07,
      "loss": 0.185,
      "step": 29080
    },
    {
      "epoch": 0.8987479628328017,
      "grad_norm": 0.09292387950499816,
      "learning_rate": 8.329504967991114e-07,
      "loss": 0.1811,
      "step": 29090
    },
    {
      "epoch": 0.8990569170998463,
      "grad_norm": 0.0910318387040759,
      "learning_rate": 8.279218560615354e-07,
      "loss": 0.1804,
      "step": 29100
    },
    {
      "epoch": 0.899365871366891,
      "grad_norm": 0.10269314685387879,
      "learning_rate": 8.229080097970126e-07,
      "loss": 0.1807,
      "step": 29110
    },
    {
      "epoch": 0.8996748256339355,
      "grad_norm": 0.12226884148247592,
      "learning_rate": 8.179089632395831e-07,
      "loss": 0.1813,
      "step": 29120
    },
    {
      "epoch": 0.8999837799009801,
      "grad_norm": 0.10890893639008979,
      "learning_rate": 8.129247216078339e-07,
      "loss": 0.1803,
      "step": 29130
    },
    {
      "epoch": 0.9002927341680248,
      "grad_norm": 0.10686181822274243,
      "learning_rate": 8.079552901049031e-07,
      "loss": 0.1792,
      "step": 29140
    },
    {
      "epoch": 0.9006016884350694,
      "grad_norm": 0.11025449599331412,
      "learning_rate": 8.030006739184609e-07,
      "loss": 0.1856,
      "step": 29150
    },
    {
      "epoch": 0.9009106427021141,
      "grad_norm": 0.09973545902540376,
      "learning_rate": 7.9806087822072e-07,
      "loss": 0.1786,
      "step": 29160
    },
    {
      "epoch": 0.9012195969691587,
      "grad_norm": 0.08963783098207974,
      "learning_rate": 7.93135908168412e-07,
      "loss": 0.1813,
      "step": 29170
    },
    {
      "epoch": 0.9015285512362032,
      "grad_norm": 0.11163810463198195,
      "learning_rate": 7.882257689028011e-07,
      "loss": 0.1841,
      "step": 29180
    },
    {
      "epoch": 0.9018375055032479,
      "grad_norm": 0.10297976402767278,
      "learning_rate": 7.833304655496604e-07,
      "loss": 0.1788,
      "step": 29190
    },
    {
      "epoch": 0.9021464597702925,
      "grad_norm": 0.11550709958897573,
      "learning_rate": 7.784500032192853e-07,
      "loss": 0.1788,
      "step": 29200
    },
    {
      "epoch": 0.9024554140373371,
      "grad_norm": 0.10279863424340445,
      "learning_rate": 7.735843870064674e-07,
      "loss": 0.1833,
      "step": 29210
    },
    {
      "epoch": 0.9027643683043818,
      "grad_norm": 0.09129452739293219,
      "learning_rate": 7.687336219905134e-07,
      "loss": 0.1879,
      "step": 29220
    },
    {
      "epoch": 0.9030733225714264,
      "grad_norm": 0.10351253777475707,
      "learning_rate": 7.638977132352127e-07,
      "loss": 0.187,
      "step": 29230
    },
    {
      "epoch": 0.9033822768384709,
      "grad_norm": 0.11887945138669731,
      "learning_rate": 7.59076665788857e-07,
      "loss": 0.183,
      "step": 29240
    },
    {
      "epoch": 0.9036912311055156,
      "grad_norm": 0.09319794345234407,
      "learning_rate": 7.542704846842157e-07,
      "loss": 0.1803,
      "step": 29250
    },
    {
      "epoch": 0.9040001853725602,
      "grad_norm": 0.11671175457831218,
      "learning_rate": 7.494791749385472e-07,
      "loss": 0.1942,
      "step": 29260
    },
    {
      "epoch": 0.9043091396396048,
      "grad_norm": 0.09578553004986752,
      "learning_rate": 7.447027415535784e-07,
      "loss": 0.1829,
      "step": 29270
    },
    {
      "epoch": 0.9046180939066495,
      "grad_norm": 0.10359140237182186,
      "learning_rate": 7.399411895155095e-07,
      "loss": 0.1816,
      "step": 29280
    },
    {
      "epoch": 0.9049270481736941,
      "grad_norm": 0.10032944776746226,
      "learning_rate": 7.351945237950059e-07,
      "loss": 0.187,
      "step": 29290
    },
    {
      "epoch": 0.9052360024407388,
      "grad_norm": 0.10687990697775257,
      "learning_rate": 7.304627493471938e-07,
      "loss": 0.1812,
      "step": 29300
    },
    {
      "epoch": 0.9055449567077833,
      "grad_norm": 0.09628847482176447,
      "learning_rate": 7.257458711116488e-07,
      "loss": 0.1808,
      "step": 29310
    },
    {
      "epoch": 0.9058539109748279,
      "grad_norm": 0.10458719491106042,
      "learning_rate": 7.210438940124054e-07,
      "loss": 0.1781,
      "step": 29320
    },
    {
      "epoch": 0.9061628652418726,
      "grad_norm": 0.10001730084102851,
      "learning_rate": 7.163568229579337e-07,
      "loss": 0.1825,
      "step": 29330
    },
    {
      "epoch": 0.9064718195089172,
      "grad_norm": 0.0991383494556449,
      "learning_rate": 7.116846628411467e-07,
      "loss": 0.1896,
      "step": 29340
    },
    {
      "epoch": 0.9067807737759618,
      "grad_norm": 0.09150038024499577,
      "learning_rate": 7.07027418539396e-07,
      "loss": 0.1774,
      "step": 29350
    },
    {
      "epoch": 0.9070897280430065,
      "grad_norm": 0.09839956558796607,
      "learning_rate": 7.023850949144511e-07,
      "loss": 0.1801,
      "step": 29360
    },
    {
      "epoch": 0.907398682310051,
      "grad_norm": 0.09434902663828376,
      "learning_rate": 6.977576968125205e-07,
      "loss": 0.1785,
      "step": 29370
    },
    {
      "epoch": 0.9077076365770956,
      "grad_norm": 0.1058722958615921,
      "learning_rate": 6.931452290642149e-07,
      "loss": 0.1854,
      "step": 29380
    },
    {
      "epoch": 0.9080165908441403,
      "grad_norm": 0.09113919344971512,
      "learning_rate": 6.885476964845761e-07,
      "loss": 0.1837,
      "step": 29390
    },
    {
      "epoch": 0.9083255451111849,
      "grad_norm": 0.08606955551639808,
      "learning_rate": 6.839651038730399e-07,
      "loss": 0.18,
      "step": 29400
    },
    {
      "epoch": 0.9086344993782295,
      "grad_norm": 0.10582500713480847,
      "learning_rate": 6.793974560134581e-07,
      "loss": 0.1823,
      "step": 29410
    },
    {
      "epoch": 0.9089434536452742,
      "grad_norm": 0.10987654427791921,
      "learning_rate": 6.748447576740713e-07,
      "loss": 0.1832,
      "step": 29420
    },
    {
      "epoch": 0.9092524079123188,
      "grad_norm": 0.10692373351150367,
      "learning_rate": 6.703070136075212e-07,
      "loss": 0.1787,
      "step": 29430
    },
    {
      "epoch": 0.9095613621793635,
      "grad_norm": 0.1023497268432669,
      "learning_rate": 6.657842285508337e-07,
      "loss": 0.1782,
      "step": 29440
    },
    {
      "epoch": 0.909870316446408,
      "grad_norm": 0.09836289116065891,
      "learning_rate": 6.612764072254251e-07,
      "loss": 0.1788,
      "step": 29450
    },
    {
      "epoch": 0.9101792707134526,
      "grad_norm": 0.10072022627718054,
      "learning_rate": 6.567835543370815e-07,
      "loss": 0.1873,
      "step": 29460
    },
    {
      "epoch": 0.9104882249804973,
      "grad_norm": 0.10000037604708514,
      "learning_rate": 6.523056745759726e-07,
      "loss": 0.1796,
      "step": 29470
    },
    {
      "epoch": 0.9107971792475419,
      "grad_norm": 0.09392768576244617,
      "learning_rate": 6.478427726166275e-07,
      "loss": 0.1812,
      "step": 29480
    },
    {
      "epoch": 0.9111061335145865,
      "grad_norm": 0.10380911407942348,
      "learning_rate": 6.433948531179529e-07,
      "loss": 0.1831,
      "step": 29490
    },
    {
      "epoch": 0.9114150877816312,
      "grad_norm": 0.11039595676455562,
      "learning_rate": 6.389619207232028e-07,
      "loss": 0.1816,
      "step": 29500
    },
    {
      "epoch": 0.9117240420486757,
      "grad_norm": 0.09300630064449593,
      "learning_rate": 6.345439800599906e-07,
      "loss": 0.1796,
      "step": 29510
    },
    {
      "epoch": 0.9120329963157203,
      "grad_norm": 0.09061544982607218,
      "learning_rate": 6.30141035740282e-07,
      "loss": 0.1854,
      "step": 29520
    },
    {
      "epoch": 0.912341950582765,
      "grad_norm": 0.09678414878726044,
      "learning_rate": 6.257530923603838e-07,
      "loss": 0.1913,
      "step": 29530
    },
    {
      "epoch": 0.9126509048498096,
      "grad_norm": 0.10063326496934295,
      "learning_rate": 6.213801545009417e-07,
      "loss": 0.1803,
      "step": 29540
    },
    {
      "epoch": 0.9129598591168542,
      "grad_norm": 0.08755798029903349,
      "learning_rate": 6.170222267269459e-07,
      "loss": 0.1808,
      "step": 29550
    },
    {
      "epoch": 0.9132688133838989,
      "grad_norm": 0.09012231097949266,
      "learning_rate": 6.126793135877074e-07,
      "loss": 0.1865,
      "step": 29560
    },
    {
      "epoch": 0.9135777676509435,
      "grad_norm": 0.102220680898261,
      "learning_rate": 6.083514196168661e-07,
      "loss": 0.1858,
      "step": 29570
    },
    {
      "epoch": 0.9138867219179881,
      "grad_norm": 0.10101546876552676,
      "learning_rate": 6.04038549332388e-07,
      "loss": 0.1816,
      "step": 29580
    },
    {
      "epoch": 0.9141956761850327,
      "grad_norm": 0.09639490661184655,
      "learning_rate": 5.997407072365452e-07,
      "loss": 0.1849,
      "step": 29590
    },
    {
      "epoch": 0.9145046304520773,
      "grad_norm": 0.10361945677028739,
      "learning_rate": 5.954578978159353e-07,
      "loss": 0.1789,
      "step": 29600
    },
    {
      "epoch": 0.914813584719122,
      "grad_norm": 0.10471801022044404,
      "learning_rate": 5.911901255414504e-07,
      "loss": 0.1931,
      "step": 29610
    },
    {
      "epoch": 0.9151225389861666,
      "grad_norm": 0.10896051572266999,
      "learning_rate": 5.869373948682949e-07,
      "loss": 0.1875,
      "step": 29620
    },
    {
      "epoch": 0.9154314932532112,
      "grad_norm": 0.08887670788550045,
      "learning_rate": 5.826997102359627e-07,
      "loss": 0.1757,
      "step": 29630
    },
    {
      "epoch": 0.9157404475202559,
      "grad_norm": 0.10405189299446264,
      "learning_rate": 5.784770760682467e-07,
      "loss": 0.178,
      "step": 29640
    },
    {
      "epoch": 0.9160494017873004,
      "grad_norm": 0.10633378942465,
      "learning_rate": 5.742694967732259e-07,
      "loss": 0.1862,
      "step": 29650
    },
    {
      "epoch": 0.916358356054345,
      "grad_norm": 0.10989376626685819,
      "learning_rate": 5.700769767432634e-07,
      "loss": 0.185,
      "step": 29660
    },
    {
      "epoch": 0.9166673103213897,
      "grad_norm": 0.10611869518374586,
      "learning_rate": 5.658995203550033e-07,
      "loss": 0.1957,
      "step": 29670
    },
    {
      "epoch": 0.9169762645884343,
      "grad_norm": 0.09543215073081474,
      "learning_rate": 5.61737131969362e-07,
      "loss": 0.1836,
      "step": 29680
    },
    {
      "epoch": 0.9172852188554789,
      "grad_norm": 0.10380110789519864,
      "learning_rate": 5.575898159315273e-07,
      "loss": 0.1867,
      "step": 29690
    },
    {
      "epoch": 0.9175941731225236,
      "grad_norm": 0.10505856858325727,
      "learning_rate": 5.534575765709559e-07,
      "loss": 0.1761,
      "step": 29700
    },
    {
      "epoch": 0.9179031273895681,
      "grad_norm": 0.0918711821906746,
      "learning_rate": 5.493404182013573e-07,
      "loss": 0.178,
      "step": 29710
    },
    {
      "epoch": 0.9182120816566128,
      "grad_norm": 0.09006996867686663,
      "learning_rate": 5.452383451207083e-07,
      "loss": 0.185,
      "step": 29720
    },
    {
      "epoch": 0.9185210359236574,
      "grad_norm": 0.08282612653706493,
      "learning_rate": 5.411513616112302e-07,
      "loss": 0.1803,
      "step": 29730
    },
    {
      "epoch": 0.918829990190702,
      "grad_norm": 0.1029773562128662,
      "learning_rate": 5.370794719393934e-07,
      "loss": 0.1792,
      "step": 29740
    },
    {
      "epoch": 0.9191389444577467,
      "grad_norm": 0.10228421669832138,
      "learning_rate": 5.330226803559146e-07,
      "loss": 0.183,
      "step": 29750
    },
    {
      "epoch": 0.9194478987247913,
      "grad_norm": 0.09512145208943297,
      "learning_rate": 5.289809910957477e-07,
      "loss": 0.1794,
      "step": 29760
    },
    {
      "epoch": 0.9197568529918358,
      "grad_norm": 0.10968131771617376,
      "learning_rate": 5.249544083780761e-07,
      "loss": 0.1793,
      "step": 29770
    },
    {
      "epoch": 0.9200658072588805,
      "grad_norm": 0.0944946814003296,
      "learning_rate": 5.209429364063256e-07,
      "loss": 0.183,
      "step": 29780
    },
    {
      "epoch": 0.9203747615259251,
      "grad_norm": 0.08907065647558841,
      "learning_rate": 5.16946579368135e-07,
      "loss": 0.1853,
      "step": 29790
    },
    {
      "epoch": 0.9206837157929697,
      "grad_norm": 0.10808489485030408,
      "learning_rate": 5.129653414353685e-07,
      "loss": 0.1823,
      "step": 29800
    },
    {
      "epoch": 0.9209926700600144,
      "grad_norm": 0.09297344698286493,
      "learning_rate": 5.089992267641119e-07,
      "loss": 0.1771,
      "step": 29810
    },
    {
      "epoch": 0.921301624327059,
      "grad_norm": 0.10569189195410038,
      "learning_rate": 5.050482394946549e-07,
      "loss": 0.1818,
      "step": 29820
    },
    {
      "epoch": 0.9216105785941036,
      "grad_norm": 0.1031636284663917,
      "learning_rate": 5.011123837515047e-07,
      "loss": 0.1797,
      "step": 29830
    },
    {
      "epoch": 0.9219195328611482,
      "grad_norm": 0.09695635680227692,
      "learning_rate": 4.971916636433633e-07,
      "loss": 0.184,
      "step": 29840
    },
    {
      "epoch": 0.9222284871281928,
      "grad_norm": 0.10368983285980565,
      "learning_rate": 4.932860832631431e-07,
      "loss": 0.1808,
      "step": 29850
    },
    {
      "epoch": 0.9225374413952375,
      "grad_norm": 0.12075062802191754,
      "learning_rate": 4.893956466879395e-07,
      "loss": 0.1828,
      "step": 29860
    },
    {
      "epoch": 0.9228463956622821,
      "grad_norm": 0.10468553119957653,
      "learning_rate": 4.855203579790506e-07,
      "loss": 0.1881,
      "step": 29870
    },
    {
      "epoch": 0.9231553499293267,
      "grad_norm": 0.08886598344406338,
      "learning_rate": 4.816602211819504e-07,
      "loss": 0.1813,
      "step": 29880
    },
    {
      "epoch": 0.9234643041963714,
      "grad_norm": 0.09755390173940867,
      "learning_rate": 4.778152403263087e-07,
      "loss": 0.1839,
      "step": 29890
    },
    {
      "epoch": 0.923773258463416,
      "grad_norm": 0.11342180876900834,
      "learning_rate": 4.7398541942595797e-07,
      "loss": 0.1881,
      "step": 29900
    },
    {
      "epoch": 0.9240822127304605,
      "grad_norm": 0.10265418369914854,
      "learning_rate": 4.701707624789203e-07,
      "loss": 0.1812,
      "step": 29910
    },
    {
      "epoch": 0.9243911669975052,
      "grad_norm": 0.0860545327122374,
      "learning_rate": 4.66371273467377e-07,
      "loss": 0.1829,
      "step": 29920
    },
    {
      "epoch": 0.9247001212645498,
      "grad_norm": 0.0869079032605448,
      "learning_rate": 4.6258695635768013e-07,
      "loss": 0.1831,
      "step": 29930
    },
    {
      "epoch": 0.9250090755315944,
      "grad_norm": 0.08322258554903587,
      "learning_rate": 4.5881781510034306e-07,
      "loss": 0.18,
      "step": 29940
    },
    {
      "epoch": 0.9253180297986391,
      "grad_norm": 0.11584902980690737,
      "learning_rate": 4.5506385363003666e-07,
      "loss": 0.1811,
      "step": 29950
    },
    {
      "epoch": 0.9256269840656837,
      "grad_norm": 0.11449546928077053,
      "learning_rate": 4.513250758655846e-07,
      "loss": 0.179,
      "step": 29960
    },
    {
      "epoch": 0.9259359383327282,
      "grad_norm": 0.09511237831508776,
      "learning_rate": 4.476014857099581e-07,
      "loss": 0.1808,
      "step": 29970
    },
    {
      "epoch": 0.9262448925997729,
      "grad_norm": 0.09777927733821344,
      "learning_rate": 4.438930870502794e-07,
      "loss": 0.1815,
      "step": 29980
    },
    {
      "epoch": 0.9265538468668175,
      "grad_norm": 0.09791697691167553,
      "learning_rate": 4.401998837578086e-07,
      "loss": 0.1795,
      "step": 29990
    },
    {
      "epoch": 0.9268628011338622,
      "grad_norm": 0.09410905844978593,
      "learning_rate": 4.3652187968794166e-07,
      "loss": 0.1823,
      "step": 30000
    },
    {
      "epoch": 0.9271717554009068,
      "grad_norm": 0.11233822212239762,
      "learning_rate": 4.328590786802089e-07,
      "loss": 0.187,
      "step": 30010
    },
    {
      "epoch": 0.9274807096679514,
      "grad_norm": 0.10961403369670594,
      "learning_rate": 4.292114845582751e-07,
      "loss": 0.1819,
      "step": 30020
    },
    {
      "epoch": 0.9277896639349961,
      "grad_norm": 0.09096732247698736,
      "learning_rate": 4.255791011299193e-07,
      "loss": 0.1821,
      "step": 30030
    },
    {
      "epoch": 0.9280986182020406,
      "grad_norm": 0.10444246248378831,
      "learning_rate": 4.2196193218705327e-07,
      "loss": 0.1812,
      "step": 30040
    },
    {
      "epoch": 0.9284075724690852,
      "grad_norm": 0.09909974058953407,
      "learning_rate": 4.183599815056982e-07,
      "loss": 0.1809,
      "step": 30050
    },
    {
      "epoch": 0.9287165267361299,
      "grad_norm": 0.10396932966086093,
      "learning_rate": 4.14773252845993e-07,
      "loss": 0.1814,
      "step": 30060
    },
    {
      "epoch": 0.9290254810031745,
      "grad_norm": 0.10829532855810024,
      "learning_rate": 4.1120174995218417e-07,
      "loss": 0.1786,
      "step": 30070
    },
    {
      "epoch": 0.9293344352702191,
      "grad_norm": 0.10929377514996606,
      "learning_rate": 4.076454765526244e-07,
      "loss": 0.1787,
      "step": 30080
    },
    {
      "epoch": 0.9296433895372638,
      "grad_norm": 0.0870955571350648,
      "learning_rate": 4.0410443635976737e-07,
      "loss": 0.1767,
      "step": 30090
    },
    {
      "epoch": 0.9299523438043084,
      "grad_norm": 0.09428167257410036,
      "learning_rate": 4.0057863307016775e-07,
      "loss": 0.1802,
      "step": 30100
    },
    {
      "epoch": 0.9302612980713529,
      "grad_norm": 0.12259391774896332,
      "learning_rate": 3.9706807036446635e-07,
      "loss": 0.1816,
      "step": 30110
    },
    {
      "epoch": 0.9305702523383976,
      "grad_norm": 0.10314358065032986,
      "learning_rate": 3.9357275190740503e-07,
      "loss": 0.1848,
      "step": 30120
    },
    {
      "epoch": 0.9308792066054422,
      "grad_norm": 0.10591627551081069,
      "learning_rate": 3.900926813478001e-07,
      "loss": 0.185,
      "step": 30130
    },
    {
      "epoch": 0.9311881608724869,
      "grad_norm": 0.08618325405762615,
      "learning_rate": 3.8662786231856204e-07,
      "loss": 0.1819,
      "step": 30140
    },
    {
      "epoch": 0.9314971151395315,
      "grad_norm": 0.11218447029041394,
      "learning_rate": 3.831782984366694e-07,
      "loss": 0.1844,
      "step": 30150
    },
    {
      "epoch": 0.9318060694065761,
      "grad_norm": 0.08629681276531403,
      "learning_rate": 3.7974399330318487e-07,
      "loss": 0.1824,
      "step": 30160
    },
    {
      "epoch": 0.9321150236736208,
      "grad_norm": 0.09666582244614211,
      "learning_rate": 3.76324950503234e-07,
      "loss": 0.183,
      "step": 30170
    },
    {
      "epoch": 0.9324239779406653,
      "grad_norm": 0.11795066468193195,
      "learning_rate": 3.7292117360601676e-07,
      "loss": 0.1799,
      "step": 30180
    },
    {
      "epoch": 0.9327329322077099,
      "grad_norm": 0.0907756439699543,
      "learning_rate": 3.6953266616479077e-07,
      "loss": 0.1783,
      "step": 30190
    },
    {
      "epoch": 0.9330418864747546,
      "grad_norm": 0.10095984370387073,
      "learning_rate": 3.661594317168782e-07,
      "loss": 0.1811,
      "step": 30200
    },
    {
      "epoch": 0.9333508407417992,
      "grad_norm": 0.09059533748283437,
      "learning_rate": 3.6280147378365714e-07,
      "loss": 0.1856,
      "step": 30210
    },
    {
      "epoch": 0.9336597950088438,
      "grad_norm": 0.10715326895262456,
      "learning_rate": 3.594587958705553e-07,
      "loss": 0.1763,
      "step": 30220
    },
    {
      "epoch": 0.9339687492758885,
      "grad_norm": 0.0961058305704472,
      "learning_rate": 3.5613140146705137e-07,
      "loss": 0.1782,
      "step": 30230
    },
    {
      "epoch": 0.934277703542933,
      "grad_norm": 0.09863720445187962,
      "learning_rate": 3.5281929404666515e-07,
      "loss": 0.1974,
      "step": 30240
    },
    {
      "epoch": 0.9345866578099776,
      "grad_norm": 0.09458104062093899,
      "learning_rate": 3.495224770669675e-07,
      "loss": 0.1791,
      "step": 30250
    },
    {
      "epoch": 0.9348956120770223,
      "grad_norm": 0.11277168452021748,
      "learning_rate": 3.462409539695588e-07,
      "loss": 0.1813,
      "step": 30260
    },
    {
      "epoch": 0.9352045663440669,
      "grad_norm": 0.09949895573709748,
      "learning_rate": 3.4297472818007714e-07,
      "loss": 0.1834,
      "step": 30270
    },
    {
      "epoch": 0.9355135206111116,
      "grad_norm": 0.09233645489688903,
      "learning_rate": 3.3972380310819174e-07,
      "loss": 0.1809,
      "step": 30280
    },
    {
      "epoch": 0.9358224748781562,
      "grad_norm": 0.10647168986460545,
      "learning_rate": 3.364881821475979e-07,
      "loss": 0.1804,
      "step": 30290
    },
    {
      "epoch": 0.9361314291452008,
      "grad_norm": 0.09382266832164135,
      "learning_rate": 3.332678686760138e-07,
      "loss": 0.1854,
      "step": 30300
    },
    {
      "epoch": 0.9364403834122454,
      "grad_norm": 0.12441950134822441,
      "learning_rate": 3.3006286605518356e-07,
      "loss": 0.1828,
      "step": 30310
    },
    {
      "epoch": 0.93674933767929,
      "grad_norm": 0.12066314839556808,
      "learning_rate": 3.2687317763085923e-07,
      "loss": 0.1792,
      "step": 30320
    },
    {
      "epoch": 0.9370582919463346,
      "grad_norm": 0.09219619759749195,
      "learning_rate": 3.236988067328139e-07,
      "loss": 0.1855,
      "step": 30330
    },
    {
      "epoch": 0.9373672462133793,
      "grad_norm": 0.10156500316665396,
      "learning_rate": 3.205397566748236e-07,
      "loss": 0.1804,
      "step": 30340
    },
    {
      "epoch": 0.9376762004804239,
      "grad_norm": 0.1350016992242699,
      "learning_rate": 3.173960307546786e-07,
      "loss": 0.1839,
      "step": 30350
    },
    {
      "epoch": 0.9379851547474685,
      "grad_norm": 0.10143674237682757,
      "learning_rate": 3.142676322541654e-07,
      "loss": 0.1822,
      "step": 30360
    },
    {
      "epoch": 0.9382941090145132,
      "grad_norm": 0.09664601462339328,
      "learning_rate": 3.111545644390751e-07,
      "loss": 0.1867,
      "step": 30370
    },
    {
      "epoch": 0.9386030632815577,
      "grad_norm": 0.11019493789410766,
      "learning_rate": 3.0805683055918633e-07,
      "loss": 0.1833,
      "step": 30380
    },
    {
      "epoch": 0.9389120175486023,
      "grad_norm": 0.11548150338207855,
      "learning_rate": 3.0497443384828237e-07,
      "loss": 0.1846,
      "step": 30390
    },
    {
      "epoch": 0.939220971815647,
      "grad_norm": 0.09089515510412766,
      "learning_rate": 3.0190737752412434e-07,
      "loss": 0.1798,
      "step": 30400
    },
    {
      "epoch": 0.9395299260826916,
      "grad_norm": 0.09324919353782359,
      "learning_rate": 2.9885566478846937e-07,
      "loss": 0.1834,
      "step": 30410
    },
    {
      "epoch": 0.9398388803497363,
      "grad_norm": 0.09823795553559096,
      "learning_rate": 2.9581929882704917e-07,
      "loss": 0.1798,
      "step": 30420
    },
    {
      "epoch": 0.9401478346167809,
      "grad_norm": 0.10024622202052019,
      "learning_rate": 2.9279828280957656e-07,
      "loss": 0.1906,
      "step": 30430
    },
    {
      "epoch": 0.9404567888838254,
      "grad_norm": 0.09194187185121717,
      "learning_rate": 2.897926198897438e-07,
      "loss": 0.1852,
      "step": 30440
    },
    {
      "epoch": 0.9407657431508701,
      "grad_norm": 0.08630621901403171,
      "learning_rate": 2.8680231320521276e-07,
      "loss": 0.1801,
      "step": 30450
    },
    {
      "epoch": 0.9410746974179147,
      "grad_norm": 0.10337890614170019,
      "learning_rate": 2.838273658776164e-07,
      "loss": 0.1802,
      "step": 30460
    },
    {
      "epoch": 0.9413836516849593,
      "grad_norm": 0.11286791158076939,
      "learning_rate": 2.808677810125504e-07,
      "loss": 0.1885,
      "step": 30470
    },
    {
      "epoch": 0.941692605952004,
      "grad_norm": 0.10328023624849803,
      "learning_rate": 2.779235616995784e-07,
      "loss": 0.1791,
      "step": 30480
    },
    {
      "epoch": 0.9420015602190486,
      "grad_norm": 0.08821982637772095,
      "learning_rate": 2.7499471101221696e-07,
      "loss": 0.1813,
      "step": 30490
    },
    {
      "epoch": 0.9423105144860932,
      "grad_norm": 0.09749760321832934,
      "learning_rate": 2.7208123200794856e-07,
      "loss": 0.1833,
      "step": 30500
    },
    {
      "epoch": 0.9426194687531378,
      "grad_norm": 0.10999300650270014,
      "learning_rate": 2.6918312772819866e-07,
      "loss": 0.1868,
      "step": 30510
    },
    {
      "epoch": 0.9429284230201824,
      "grad_norm": 0.10523617642925231,
      "learning_rate": 2.663004011983505e-07,
      "loss": 0.1831,
      "step": 30520
    },
    {
      "epoch": 0.943237377287227,
      "grad_norm": 0.12299685114607545,
      "learning_rate": 2.6343305542772865e-07,
      "loss": 0.1765,
      "step": 30530
    },
    {
      "epoch": 0.9435463315542717,
      "grad_norm": 0.08514869164601965,
      "learning_rate": 2.60581093409607e-07,
      "loss": 0.1793,
      "step": 30540
    },
    {
      "epoch": 0.9438552858213163,
      "grad_norm": 0.09729866740408408,
      "learning_rate": 2.5774451812119406e-07,
      "loss": 0.1788,
      "step": 30550
    },
    {
      "epoch": 0.944164240088361,
      "grad_norm": 0.09401990369983891,
      "learning_rate": 2.549233325236411e-07,
      "loss": 0.1869,
      "step": 30560
    },
    {
      "epoch": 0.9444731943554056,
      "grad_norm": 0.10212319595435185,
      "learning_rate": 2.5211753956203064e-07,
      "loss": 0.1798,
      "step": 30570
    },
    {
      "epoch": 0.9447821486224501,
      "grad_norm": 0.11541113039392047,
      "learning_rate": 2.493271421653798e-07,
      "loss": 0.1806,
      "step": 30580
    },
    {
      "epoch": 0.9450911028894948,
      "grad_norm": 0.09586379427206575,
      "learning_rate": 2.4655214324662835e-07,
      "loss": 0.1831,
      "step": 30590
    },
    {
      "epoch": 0.9454000571565394,
      "grad_norm": 0.10640865997994073,
      "learning_rate": 2.437925457026491e-07,
      "loss": 0.178,
      "step": 30600
    },
    {
      "epoch": 0.945709011423584,
      "grad_norm": 0.09314700680012072,
      "learning_rate": 2.410483524142276e-07,
      "loss": 0.1824,
      "step": 30610
    },
    {
      "epoch": 0.9460179656906287,
      "grad_norm": 0.10211712835299988,
      "learning_rate": 2.3831956624607744e-07,
      "loss": 0.1766,
      "step": 30620
    },
    {
      "epoch": 0.9463269199576733,
      "grad_norm": 0.08845261052946496,
      "learning_rate": 2.3560619004682316e-07,
      "loss": 0.1874,
      "step": 30630
    },
    {
      "epoch": 0.9466358742247178,
      "grad_norm": 0.09570351981970587,
      "learning_rate": 2.3290822664900568e-07,
      "loss": 0.1794,
      "step": 30640
    },
    {
      "epoch": 0.9469448284917625,
      "grad_norm": 0.11036580762220033,
      "learning_rate": 2.3022567886907543e-07,
      "loss": 0.183,
      "step": 30650
    },
    {
      "epoch": 0.9472537827588071,
      "grad_norm": 0.09655892173293927,
      "learning_rate": 2.275585495073873e-07,
      "loss": 0.1857,
      "step": 30660
    },
    {
      "epoch": 0.9475627370258517,
      "grad_norm": 0.09270858918674574,
      "learning_rate": 2.249068413482025e-07,
      "loss": 0.1858,
      "step": 30670
    },
    {
      "epoch": 0.9478716912928964,
      "grad_norm": 0.0999655019877312,
      "learning_rate": 2.222705571596867e-07,
      "loss": 0.1865,
      "step": 30680
    },
    {
      "epoch": 0.948180645559941,
      "grad_norm": 0.10282030825716987,
      "learning_rate": 2.1964969969390014e-07,
      "loss": 0.1782,
      "step": 30690
    },
    {
      "epoch": 0.9484895998269857,
      "grad_norm": 0.11895223395832247,
      "learning_rate": 2.1704427168680097e-07,
      "loss": 0.183,
      "step": 30700
    },
    {
      "epoch": 0.9487985540940302,
      "grad_norm": 0.09845930480604034,
      "learning_rate": 2.1445427585824019e-07,
      "loss": 0.1802,
      "step": 30710
    },
    {
      "epoch": 0.9491075083610748,
      "grad_norm": 0.12006083456041604,
      "learning_rate": 2.1187971491195668e-07,
      "loss": 0.1838,
      "step": 30720
    },
    {
      "epoch": 0.9494164626281195,
      "grad_norm": 0.09735950439066168,
      "learning_rate": 2.093205915355806e-07,
      "loss": 0.1827,
      "step": 30730
    },
    {
      "epoch": 0.9497254168951641,
      "grad_norm": 0.09079479696483302,
      "learning_rate": 2.0677690840062158e-07,
      "loss": 0.1813,
      "step": 30740
    },
    {
      "epoch": 0.9500343711622087,
      "grad_norm": 0.1008015348084071,
      "learning_rate": 2.0424866816247723e-07,
      "loss": 0.1815,
      "step": 30750
    },
    {
      "epoch": 0.9503433254292534,
      "grad_norm": 0.1016307295892918,
      "learning_rate": 2.01735873460418e-07,
      "loss": 0.1791,
      "step": 30760
    },
    {
      "epoch": 0.950652279696298,
      "grad_norm": 0.10985734853998837,
      "learning_rate": 1.9923852691759392e-07,
      "loss": 0.1787,
      "step": 30770
    },
    {
      "epoch": 0.9509612339633425,
      "grad_norm": 0.09154176746505865,
      "learning_rate": 1.9675663114102628e-07,
      "loss": 0.1781,
      "step": 30780
    },
    {
      "epoch": 0.9512701882303872,
      "grad_norm": 0.08927489490479662,
      "learning_rate": 1.9429018872161087e-07,
      "loss": 0.18,
      "step": 30790
    },
    {
      "epoch": 0.9515791424974318,
      "grad_norm": 0.10352047287878095,
      "learning_rate": 1.9183920223410479e-07,
      "loss": 0.1819,
      "step": 30800
    },
    {
      "epoch": 0.9518880967644764,
      "grad_norm": 0.08637391387427489,
      "learning_rate": 1.89403674237138e-07,
      "loss": 0.1849,
      "step": 30810
    },
    {
      "epoch": 0.9521970510315211,
      "grad_norm": 0.09406334007056122,
      "learning_rate": 1.8698360727319675e-07,
      "loss": 0.181,
      "step": 30820
    },
    {
      "epoch": 0.9525060052985657,
      "grad_norm": 0.08298020608762798,
      "learning_rate": 1.845790038686318e-07,
      "loss": 0.1797,
      "step": 30830
    },
    {
      "epoch": 0.9528149595656104,
      "grad_norm": 0.10134935237751644,
      "learning_rate": 1.8218986653365022e-07,
      "loss": 0.1903,
      "step": 30840
    },
    {
      "epoch": 0.9531239138326549,
      "grad_norm": 0.1158332743556505,
      "learning_rate": 1.7981619776231194e-07,
      "loss": 0.1841,
      "step": 30850
    },
    {
      "epoch": 0.9534328680996995,
      "grad_norm": 0.09956372543851025,
      "learning_rate": 1.7745800003252988e-07,
      "loss": 0.1796,
      "step": 30860
    },
    {
      "epoch": 0.9537418223667442,
      "grad_norm": 0.10284171298465782,
      "learning_rate": 1.7511527580606478e-07,
      "loss": 0.1803,
      "step": 30870
    },
    {
      "epoch": 0.9540507766337888,
      "grad_norm": 0.11365079076638167,
      "learning_rate": 1.727880275285304e-07,
      "loss": 0.1846,
      "step": 30880
    },
    {
      "epoch": 0.9543597309008334,
      "grad_norm": 0.09891187498037164,
      "learning_rate": 1.7047625762938003e-07,
      "loss": 0.1895,
      "step": 30890
    },
    {
      "epoch": 0.9546686851678781,
      "grad_norm": 0.09859577223295982,
      "learning_rate": 1.6817996852190653e-07,
      "loss": 0.1818,
      "step": 30900
    },
    {
      "epoch": 0.9549776394349226,
      "grad_norm": 0.10919210441048655,
      "learning_rate": 1.6589916260324744e-07,
      "loss": 0.1826,
      "step": 30910
    },
    {
      "epoch": 0.9552865937019672,
      "grad_norm": 0.138930233256035,
      "learning_rate": 1.6363384225437316e-07,
      "loss": 0.1878,
      "step": 30920
    },
    {
      "epoch": 0.9555955479690119,
      "grad_norm": 0.10440889364143319,
      "learning_rate": 1.6138400984009206e-07,
      "loss": 0.18,
      "step": 30930
    },
    {
      "epoch": 0.9559045022360565,
      "grad_norm": 0.10886040349441196,
      "learning_rate": 1.5914966770904204e-07,
      "loss": 0.1834,
      "step": 30940
    },
    {
      "epoch": 0.9562134565031011,
      "grad_norm": 0.09283335419920878,
      "learning_rate": 1.5693081819368736e-07,
      "loss": 0.1841,
      "step": 30950
    },
    {
      "epoch": 0.9565224107701458,
      "grad_norm": 0.09958767730349954,
      "learning_rate": 1.5472746361032852e-07,
      "loss": 0.1821,
      "step": 30960
    },
    {
      "epoch": 0.9568313650371904,
      "grad_norm": 0.10424635733138844,
      "learning_rate": 1.5253960625908058e-07,
      "loss": 0.1844,
      "step": 30970
    },
    {
      "epoch": 0.957140319304235,
      "grad_norm": 0.09634602265514253,
      "learning_rate": 1.503672484238866e-07,
      "loss": 0.181,
      "step": 30980
    },
    {
      "epoch": 0.9574492735712796,
      "grad_norm": 0.09673533450203761,
      "learning_rate": 1.482103923725059e-07,
      "loss": 0.1819,
      "step": 30990
    },
    {
      "epoch": 0.9577582278383242,
      "grad_norm": 0.09281362169035717,
      "learning_rate": 1.460690403565207e-07,
      "loss": 0.1802,
      "step": 31000
    },
    {
      "epoch": 0.9580671821053689,
      "grad_norm": 0.09907673820780423,
      "learning_rate": 1.4394319461132122e-07,
      "loss": 0.1796,
      "step": 31010
    },
    {
      "epoch": 0.9583761363724135,
      "grad_norm": 0.10754740225687179,
      "learning_rate": 1.4183285735611728e-07,
      "loss": 0.1844,
      "step": 31020
    },
    {
      "epoch": 0.9586850906394581,
      "grad_norm": 0.10687190844780521,
      "learning_rate": 1.3973803079392166e-07,
      "loss": 0.1847,
      "step": 31030
    },
    {
      "epoch": 0.9589940449065028,
      "grad_norm": 0.09590553826888838,
      "learning_rate": 1.376587171115634e-07,
      "loss": 0.1874,
      "step": 31040
    },
    {
      "epoch": 0.9593029991735473,
      "grad_norm": 0.09899282104756171,
      "learning_rate": 1.3559491847966953e-07,
      "loss": 0.1792,
      "step": 31050
    },
    {
      "epoch": 0.9596119534405919,
      "grad_norm": 0.09017603490684323,
      "learning_rate": 1.3354663705267833e-07,
      "loss": 0.1795,
      "step": 31060
    },
    {
      "epoch": 0.9599209077076366,
      "grad_norm": 0.10258765398510361,
      "learning_rate": 1.3151387496882272e-07,
      "loss": 0.1843,
      "step": 31070
    },
    {
      "epoch": 0.9602298619746812,
      "grad_norm": 0.10480588957737341,
      "learning_rate": 1.2949663435014024e-07,
      "loss": 0.1836,
      "step": 31080
    },
    {
      "epoch": 0.9605388162417258,
      "grad_norm": 0.12923547949842573,
      "learning_rate": 1.2749491730246144e-07,
      "loss": 0.187,
      "step": 31090
    },
    {
      "epoch": 0.9608477705087705,
      "grad_norm": 0.1086299013748916,
      "learning_rate": 1.2550872591540973e-07,
      "loss": 0.1773,
      "step": 31100
    },
    {
      "epoch": 0.961156724775815,
      "grad_norm": 0.09940283741691684,
      "learning_rate": 1.2353806226240826e-07,
      "loss": 0.1829,
      "step": 31110
    },
    {
      "epoch": 0.9614656790428597,
      "grad_norm": 0.1296964227102436,
      "learning_rate": 1.2158292840066476e-07,
      "loss": 0.186,
      "step": 31120
    },
    {
      "epoch": 0.9617746333099043,
      "grad_norm": 0.10238375975253859,
      "learning_rate": 1.1964332637117659e-07,
      "loss": 0.1825,
      "step": 31130
    },
    {
      "epoch": 0.9620835875769489,
      "grad_norm": 0.09661343115266914,
      "learning_rate": 1.1771925819872575e-07,
      "loss": 0.1854,
      "step": 31140
    },
    {
      "epoch": 0.9623925418439936,
      "grad_norm": 0.1032271540916861,
      "learning_rate": 1.1581072589188224e-07,
      "loss": 0.1778,
      "step": 31150
    },
    {
      "epoch": 0.9627014961110382,
      "grad_norm": 0.10483871215715546,
      "learning_rate": 1.13917731442994e-07,
      "loss": 0.1858,
      "step": 31160
    },
    {
      "epoch": 0.9630104503780827,
      "grad_norm": 0.09993807670422895,
      "learning_rate": 1.1204027682819195e-07,
      "loss": 0.1789,
      "step": 31170
    },
    {
      "epoch": 0.9633194046451274,
      "grad_norm": 0.08679339716029881,
      "learning_rate": 1.1017836400737835e-07,
      "loss": 0.1752,
      "step": 31180
    },
    {
      "epoch": 0.963628358912172,
      "grad_norm": 0.08653556817925051,
      "learning_rate": 1.0833199492424006e-07,
      "loss": 0.1878,
      "step": 31190
    },
    {
      "epoch": 0.9639373131792166,
      "grad_norm": 0.10718704382019693,
      "learning_rate": 1.0650117150623195e-07,
      "loss": 0.183,
      "step": 31200
    },
    {
      "epoch": 0.9642462674462613,
      "grad_norm": 0.0829982620588724,
      "learning_rate": 1.046858956645802e-07,
      "loss": 0.1898,
      "step": 31210
    },
    {
      "epoch": 0.9645552217133059,
      "grad_norm": 0.0992572624909408,
      "learning_rate": 1.0288616929428396e-07,
      "loss": 0.1839,
      "step": 31220
    },
    {
      "epoch": 0.9648641759803505,
      "grad_norm": 0.09805582122223254,
      "learning_rate": 1.0110199427410704e-07,
      "loss": 0.1839,
      "step": 31230
    },
    {
      "epoch": 0.9651731302473952,
      "grad_norm": 0.10440858101052101,
      "learning_rate": 9.933337246658125e-08,
      "loss": 0.1811,
      "step": 31240
    },
    {
      "epoch": 0.9654820845144397,
      "grad_norm": 0.10158246278329792,
      "learning_rate": 9.758030571799969e-08,
      "loss": 0.1834,
      "step": 31250
    },
    {
      "epoch": 0.9657910387814844,
      "grad_norm": 0.09828361638263758,
      "learning_rate": 9.584279585841682e-08,
      "loss": 0.1834,
      "step": 31260
    },
    {
      "epoch": 0.966099993048529,
      "grad_norm": 0.10025441577548697,
      "learning_rate": 9.412084470165339e-08,
      "loss": 0.176,
      "step": 31270
    },
    {
      "epoch": 0.9664089473155736,
      "grad_norm": 0.09825165491218962,
      "learning_rate": 9.241445404527982e-08,
      "loss": 0.1831,
      "step": 31280
    },
    {
      "epoch": 0.9667179015826183,
      "grad_norm": 0.09542113090603005,
      "learning_rate": 9.072362567062787e-08,
      "loss": 0.1816,
      "step": 31290
    },
    {
      "epoch": 0.9670268558496629,
      "grad_norm": 0.08671810627123758,
      "learning_rate": 8.90483613427806e-08,
      "loss": 0.1855,
      "step": 31300
    },
    {
      "epoch": 0.9673358101167074,
      "grad_norm": 0.0981594118534967,
      "learning_rate": 8.738866281057578e-08,
      "loss": 0.1883,
      "step": 31310
    },
    {
      "epoch": 0.9676447643837521,
      "grad_norm": 0.12426012121630006,
      "learning_rate": 8.574453180660246e-08,
      "loss": 0.1869,
      "step": 31320
    },
    {
      "epoch": 0.9679537186507967,
      "grad_norm": 0.09324399655838103,
      "learning_rate": 8.411597004719274e-08,
      "loss": 0.1893,
      "step": 31330
    },
    {
      "epoch": 0.9682626729178413,
      "grad_norm": 0.10012993605343407,
      "learning_rate": 8.250297923243333e-08,
      "loss": 0.1822,
      "step": 31340
    },
    {
      "epoch": 0.968571627184886,
      "grad_norm": 0.1056940359745438,
      "learning_rate": 8.090556104615232e-08,
      "loss": 0.1937,
      "step": 31350
    },
    {
      "epoch": 0.9688805814519306,
      "grad_norm": 0.09725261456385074,
      "learning_rate": 7.932371715592246e-08,
      "loss": 0.1776,
      "step": 31360
    },
    {
      "epoch": 0.9691895357189751,
      "grad_norm": 0.10154256931540719,
      "learning_rate": 7.77574492130545e-08,
      "loss": 0.1848,
      "step": 31370
    },
    {
      "epoch": 0.9694984899860198,
      "grad_norm": 0.09149603023414711,
      "learning_rate": 7.620675885260387e-08,
      "loss": 0.1804,
      "step": 31380
    },
    {
      "epoch": 0.9698074442530644,
      "grad_norm": 0.1007167048314042,
      "learning_rate": 7.467164769336232e-08,
      "loss": 0.185,
      "step": 31390
    },
    {
      "epoch": 0.9701163985201091,
      "grad_norm": 0.09307577434497681,
      "learning_rate": 7.315211733785965e-08,
      "loss": 0.1778,
      "step": 31400
    },
    {
      "epoch": 0.9704253527871537,
      "grad_norm": 0.0955782683346041,
      "learning_rate": 7.164816937235697e-08,
      "loss": 0.1829,
      "step": 31410
    },
    {
      "epoch": 0.9707343070541983,
      "grad_norm": 0.10212685387837438,
      "learning_rate": 7.01598053668534e-08,
      "loss": 0.1826,
      "step": 31420
    },
    {
      "epoch": 0.971043261321243,
      "grad_norm": 0.09672828605485725,
      "learning_rate": 6.868702687507777e-08,
      "loss": 0.1789,
      "step": 31430
    },
    {
      "epoch": 0.9713522155882875,
      "grad_norm": 0.11117230987639781,
      "learning_rate": 6.722983543448524e-08,
      "loss": 0.1893,
      "step": 31440
    },
    {
      "epoch": 0.9716611698553321,
      "grad_norm": 0.10042907643952784,
      "learning_rate": 6.578823256626565e-08,
      "loss": 0.1796,
      "step": 31450
    },
    {
      "epoch": 0.9719701241223768,
      "grad_norm": 0.08935027307421677,
      "learning_rate": 6.436221977533352e-08,
      "loss": 0.1819,
      "step": 31460
    },
    {
      "epoch": 0.9722790783894214,
      "grad_norm": 0.08751376649380807,
      "learning_rate": 6.295179855032306e-08,
      "loss": 0.1828,
      "step": 31470
    },
    {
      "epoch": 0.972588032656466,
      "grad_norm": 0.11205014031190086,
      "learning_rate": 6.15569703636032e-08,
      "loss": 0.1816,
      "step": 31480
    },
    {
      "epoch": 0.9728969869235107,
      "grad_norm": 0.10445823831525844,
      "learning_rate": 6.017773667125415e-08,
      "loss": 0.1812,
      "step": 31490
    },
    {
      "epoch": 0.9732059411905553,
      "grad_norm": 0.09412638655296177,
      "learning_rate": 5.8814098913082536e-08,
      "loss": 0.1849,
      "step": 31500
    },
    {
      "epoch": 0.9735148954575998,
      "grad_norm": 0.09574035116215587,
      "learning_rate": 5.7466058512612996e-08,
      "loss": 0.1756,
      "step": 31510
    },
    {
      "epoch": 0.9738238497246445,
      "grad_norm": 0.08973661460284516,
      "learning_rate": 5.613361687708985e-08,
      "loss": 0.1849,
      "step": 31520
    },
    {
      "epoch": 0.9741328039916891,
      "grad_norm": 0.10257262602884924,
      "learning_rate": 5.48167753974671e-08,
      "loss": 0.1878,
      "step": 31530
    },
    {
      "epoch": 0.9744417582587338,
      "grad_norm": 0.10515722185838597,
      "learning_rate": 5.351553544842347e-08,
      "loss": 0.1805,
      "step": 31540
    },
    {
      "epoch": 0.9747507125257784,
      "grad_norm": 0.10687366098704164,
      "learning_rate": 5.222989838834236e-08,
      "loss": 0.1759,
      "step": 31550
    },
    {
      "epoch": 0.975059666792823,
      "grad_norm": 0.1285002417425426,
      "learning_rate": 5.0959865559320195e-08,
      "loss": 0.1873,
      "step": 31560
    },
    {
      "epoch": 0.9753686210598677,
      "grad_norm": 0.12217417870223621,
      "learning_rate": 4.970543828716978e-08,
      "loss": 0.1834,
      "step": 31570
    },
    {
      "epoch": 0.9756775753269122,
      "grad_norm": 0.11109395563424355,
      "learning_rate": 4.846661788140694e-08,
      "loss": 0.1845,
      "step": 31580
    },
    {
      "epoch": 0.9759865295939568,
      "grad_norm": 0.09556218542673506,
      "learning_rate": 4.7243405635257196e-08,
      "loss": 0.1804,
      "step": 31590
    },
    {
      "epoch": 0.9762954838610015,
      "grad_norm": 0.21101658083209426,
      "learning_rate": 4.603580282565412e-08,
      "loss": 0.1835,
      "step": 31600
    },
    {
      "epoch": 0.9766044381280461,
      "grad_norm": 0.12027403171534985,
      "learning_rate": 4.4843810713232625e-08,
      "loss": 0.1794,
      "step": 31610
    },
    {
      "epoch": 0.9769133923950907,
      "grad_norm": 0.10372268218321802,
      "learning_rate": 4.366743054233402e-08,
      "loss": 0.1888,
      "step": 31620
    },
    {
      "epoch": 0.9772223466621354,
      "grad_norm": 0.12108822656519833,
      "learning_rate": 4.25066635410043e-08,
      "loss": 0.1849,
      "step": 31630
    },
    {
      "epoch": 0.97753130092918,
      "grad_norm": 0.11970145616282128,
      "learning_rate": 4.136151092098417e-08,
      "loss": 0.1939,
      "step": 31640
    },
    {
      "epoch": 0.9778402551962245,
      "grad_norm": 0.091423837825491,
      "learning_rate": 4.023197387771904e-08,
      "loss": 0.179,
      "step": 31650
    },
    {
      "epoch": 0.9781492094632692,
      "grad_norm": 0.1073034529224064,
      "learning_rate": 3.9118053590352365e-08,
      "loss": 0.1806,
      "step": 31660
    },
    {
      "epoch": 0.9784581637303138,
      "grad_norm": 0.09482698429636137,
      "learning_rate": 3.8019751221725627e-08,
      "loss": 0.1876,
      "step": 31670
    },
    {
      "epoch": 0.9787671179973585,
      "grad_norm": 0.09545180835913966,
      "learning_rate": 3.693706791837337e-08,
      "loss": 0.1815,
      "step": 31680
    },
    {
      "epoch": 0.9790760722644031,
      "grad_norm": 0.10220994595932353,
      "learning_rate": 3.587000481052649e-08,
      "loss": 0.1845,
      "step": 31690
    },
    {
      "epoch": 0.9793850265314477,
      "grad_norm": 0.10371370734131442,
      "learning_rate": 3.481856301211228e-08,
      "loss": 0.1835,
      "step": 31700
    },
    {
      "epoch": 0.9796939807984923,
      "grad_norm": 0.09816401526168778,
      "learning_rate": 3.37827436207494e-08,
      "loss": 0.1816,
      "step": 31710
    },
    {
      "epoch": 0.9800029350655369,
      "grad_norm": 0.12582193678874096,
      "learning_rate": 3.2762547717742895e-08,
      "loss": 0.1875,
      "step": 31720
    },
    {
      "epoch": 0.9803118893325815,
      "grad_norm": 0.10248223971322745,
      "learning_rate": 3.175797636809752e-08,
      "loss": 0.185,
      "step": 31730
    },
    {
      "epoch": 0.9806208435996262,
      "grad_norm": 0.0991056835340578,
      "learning_rate": 3.0769030620499404e-08,
      "loss": 0.1801,
      "step": 31740
    },
    {
      "epoch": 0.9809297978666708,
      "grad_norm": 0.10545917331151045,
      "learning_rate": 2.9795711507327738e-08,
      "loss": 0.1821,
      "step": 31750
    },
    {
      "epoch": 0.9812387521337154,
      "grad_norm": 0.09048129469489209,
      "learning_rate": 2.8838020044644754e-08,
      "loss": 0.1802,
      "step": 31760
    },
    {
      "epoch": 0.9815477064007601,
      "grad_norm": 0.13834450805184254,
      "learning_rate": 2.789595723220406e-08,
      "loss": 0.1865,
      "step": 31770
    },
    {
      "epoch": 0.9818566606678046,
      "grad_norm": 0.09663800478828244,
      "learning_rate": 2.6969524053437333e-08,
      "loss": 0.1845,
      "step": 31780
    },
    {
      "epoch": 0.9821656149348492,
      "grad_norm": 0.09261691444251614,
      "learning_rate": 2.6058721475465953e-08,
      "loss": 0.1858,
      "step": 31790
    },
    {
      "epoch": 0.9824745692018939,
      "grad_norm": 0.09601610429596383,
      "learning_rate": 2.516355044909102e-08,
      "loss": 0.1784,
      "step": 31800
    },
    {
      "epoch": 0.9827835234689385,
      "grad_norm": 0.09133263106149017,
      "learning_rate": 2.4284011908796698e-08,
      "loss": 0.186,
      "step": 31810
    },
    {
      "epoch": 0.9830924777359832,
      "grad_norm": 0.08744538727191584,
      "learning_rate": 2.3420106772750194e-08,
      "loss": 0.184,
      "step": 31820
    },
    {
      "epoch": 0.9834014320030278,
      "grad_norm": 0.0956606415326059,
      "learning_rate": 2.257183594279677e-08,
      "loss": 0.1818,
      "step": 31830
    },
    {
      "epoch": 0.9837103862700723,
      "grad_norm": 0.10678364015992076,
      "learning_rate": 2.1739200304458083e-08,
      "loss": 0.1844,
      "step": 31840
    },
    {
      "epoch": 0.984019340537117,
      "grad_norm": 0.1046628681183441,
      "learning_rate": 2.092220072693718e-08,
      "loss": 0.1866,
      "step": 31850
    },
    {
      "epoch": 0.9843282948041616,
      "grad_norm": 0.08722371357166642,
      "learning_rate": 2.012083806311682e-08,
      "loss": 0.1809,
      "step": 31860
    },
    {
      "epoch": 0.9846372490712062,
      "grad_norm": 0.1120488600333755,
      "learning_rate": 1.93351131495495e-08,
      "loss": 0.1829,
      "step": 31870
    },
    {
      "epoch": 0.9849462033382509,
      "grad_norm": 0.10191755512640596,
      "learning_rate": 1.8565026806469098e-08,
      "loss": 0.1793,
      "step": 31880
    },
    {
      "epoch": 0.9852551576052955,
      "grad_norm": 0.11345238523008164,
      "learning_rate": 1.781057983777923e-08,
      "loss": 0.1786,
      "step": 31890
    },
    {
      "epoch": 0.98556411187234,
      "grad_norm": 0.09688113374479147,
      "learning_rate": 1.707177303106322e-08,
      "loss": 0.178,
      "step": 31900
    },
    {
      "epoch": 0.9858730661393847,
      "grad_norm": 0.12737817120201278,
      "learning_rate": 1.634860715757247e-08,
      "loss": 0.1908,
      "step": 31910
    },
    {
      "epoch": 0.9861820204064293,
      "grad_norm": 0.10680908672211273,
      "learning_rate": 1.5641082972231435e-08,
      "loss": 0.1816,
      "step": 31920
    },
    {
      "epoch": 0.9864909746734739,
      "grad_norm": 0.11016360323064711,
      "learning_rate": 1.4949201213637632e-08,
      "loss": 0.2067,
      "step": 31930
    },
    {
      "epoch": 0.9867999289405186,
      "grad_norm": 0.09699509535315748,
      "learning_rate": 1.4272962604058303e-08,
      "loss": 0.1767,
      "step": 31940
    },
    {
      "epoch": 0.9871088832075632,
      "grad_norm": 0.10111583516920525,
      "learning_rate": 1.3612367849428764e-08,
      "loss": 0.1832,
      "step": 31950
    },
    {
      "epoch": 0.9874178374746079,
      "grad_norm": 0.11807058180440379,
      "learning_rate": 1.2967417639357381e-08,
      "loss": 0.1878,
      "step": 31960
    },
    {
      "epoch": 0.9877267917416525,
      "grad_norm": 0.08964540220012428,
      "learning_rate": 1.233811264711726e-08,
      "loss": 0.1793,
      "step": 31970
    },
    {
      "epoch": 0.988035746008697,
      "grad_norm": 0.09161961602329624,
      "learning_rate": 1.1724453529651236e-08,
      "loss": 0.1818,
      "step": 31980
    },
    {
      "epoch": 0.9883447002757417,
      "grad_norm": 0.10667775010404798,
      "learning_rate": 1.1126440927568538e-08,
      "loss": 0.1861,
      "step": 31990
    },
    {
      "epoch": 0.9886536545427863,
      "grad_norm": 0.11851523497796096,
      "learning_rate": 1.0544075465143133e-08,
      "loss": 0.1827,
      "step": 32000
    },
    {
      "epoch": 0.9889626088098309,
      "grad_norm": 0.12087866982137467,
      "learning_rate": 9.977357750318716e-09,
      "loss": 0.1815,
      "step": 32010
    },
    {
      "epoch": 0.9892715630768756,
      "grad_norm": 0.08792807386076792,
      "learning_rate": 9.426288374698721e-09,
      "loss": 0.1846,
      "step": 32020
    },
    {
      "epoch": 0.9895805173439202,
      "grad_norm": 0.0977213762999282,
      "learning_rate": 8.890867913556312e-09,
      "loss": 0.1808,
      "step": 32030
    },
    {
      "epoch": 0.9898894716109647,
      "grad_norm": 0.11257705339231296,
      "learning_rate": 8.371096925824384e-09,
      "loss": 0.1848,
      "step": 32040
    },
    {
      "epoch": 0.9901984258780094,
      "grad_norm": 0.15921899585788993,
      "learning_rate": 7.866975954100574e-09,
      "loss": 0.1867,
      "step": 32050
    },
    {
      "epoch": 0.990507380145054,
      "grad_norm": 0.0977229273406542,
      "learning_rate": 7.37850552464725e-09,
      "loss": 0.1798,
      "step": 32060
    },
    {
      "epoch": 0.9908163344120986,
      "grad_norm": 0.08003348434572334,
      "learning_rate": 6.905686147384849e-09,
      "loss": 0.1818,
      "step": 32070
    },
    {
      "epoch": 0.9911252886791433,
      "grad_norm": 0.10666315233146619,
      "learning_rate": 6.4485183158985436e-09,
      "loss": 0.187,
      "step": 32080
    },
    {
      "epoch": 0.9914342429461879,
      "grad_norm": 0.08181195577030377,
      "learning_rate": 6.007002507434911e-09,
      "loss": 0.1825,
      "step": 32090
    },
    {
      "epoch": 0.9917431972132326,
      "grad_norm": 0.0841212971747019,
      "learning_rate": 5.581139182896933e-09,
      "loss": 0.1848,
      "step": 32100
    },
    {
      "epoch": 0.9920521514802771,
      "grad_norm": 0.0948863518499954,
      "learning_rate": 5.170928786852325e-09,
      "loss": 0.1835,
      "step": 32110
    },
    {
      "epoch": 0.9923611057473217,
      "grad_norm": 0.10020880135974115,
      "learning_rate": 4.776371747526875e-09,
      "loss": 0.1816,
      "step": 32120
    },
    {
      "epoch": 0.9926700600143664,
      "grad_norm": 0.08954383451493052,
      "learning_rate": 4.3974684768044445e-09,
      "loss": 0.1875,
      "step": 32130
    },
    {
      "epoch": 0.992979014281411,
      "grad_norm": 0.10052940338941503,
      "learning_rate": 4.034219370228631e-09,
      "loss": 0.1834,
      "step": 32140
    },
    {
      "epoch": 0.9932879685484556,
      "grad_norm": 0.08533222400980797,
      "learning_rate": 3.6866248070027697e-09,
      "loss": 0.179,
      "step": 32150
    },
    {
      "epoch": 0.9935969228155003,
      "grad_norm": 0.11467637551400281,
      "learning_rate": 3.3546851499849373e-09,
      "loss": 0.1856,
      "step": 32160
    },
    {
      "epoch": 0.9939058770825449,
      "grad_norm": 0.10168311919267087,
      "learning_rate": 3.038400745694614e-09,
      "loss": 0.1819,
      "step": 32170
    },
    {
      "epoch": 0.9942148313495894,
      "grad_norm": 0.09641899091001285,
      "learning_rate": 2.73777192430269e-09,
      "loss": 0.1872,
      "step": 32180
    },
    {
      "epoch": 0.9945237856166341,
      "grad_norm": 0.1073219913352132,
      "learning_rate": 2.452798999646455e-09,
      "loss": 0.1816,
      "step": 32190
    },
    {
      "epoch": 0.9948327398836787,
      "grad_norm": 0.09670467218894499,
      "learning_rate": 2.183482269207948e-09,
      "loss": 0.1804,
      "step": 32200
    },
    {
      "epoch": 0.9951416941507233,
      "grad_norm": 0.08275465208071395,
      "learning_rate": 1.9298220141356073e-09,
      "loss": 0.1765,
      "step": 32210
    },
    {
      "epoch": 0.995450648417768,
      "grad_norm": 0.09641862390453226,
      "learning_rate": 1.6918184992276153e-09,
      "loss": 0.1819,
      "step": 32220
    },
    {
      "epoch": 0.9957596026848126,
      "grad_norm": 0.10014425397261856,
      "learning_rate": 1.4694719729418937e-09,
      "loss": 0.1867,
      "step": 32230
    },
    {
      "epoch": 0.9960685569518573,
      "grad_norm": 0.10490746975916102,
      "learning_rate": 1.2627826673877741e-09,
      "loss": 0.1909,
      "step": 32240
    },
    {
      "epoch": 0.9963775112189018,
      "grad_norm": 0.09579448254766845,
      "learning_rate": 1.0717507983326603e-09,
      "loss": 0.1812,
      "step": 32250
    },
    {
      "epoch": 0.9966864654859464,
      "grad_norm": 0.08367775215507442,
      "learning_rate": 8.963765651970324e-10,
      "loss": 0.1838,
      "step": 32260
    },
    {
      "epoch": 0.9969954197529911,
      "grad_norm": 0.07440305022188369,
      "learning_rate": 7.366601510577775e-10,
      "loss": 0.179,
      "step": 32270
    },
    {
      "epoch": 0.9973043740200357,
      "grad_norm": 0.12597024720194774,
      "learning_rate": 5.926017226465242e-10,
      "loss": 0.1869,
      "step": 32280
    },
    {
      "epoch": 0.9976133282870803,
      "grad_norm": 0.09689654259661641,
      "learning_rate": 4.642014303463116e-10,
      "loss": 0.1792,
      "step": 32290
    },
    {
      "epoch": 0.997922282554125,
      "grad_norm": 0.10967883052574684,
      "learning_rate": 3.5145940819825137e-10,
      "loss": 0.1922,
      "step": 32300
    },
    {
      "epoch": 0.9982312368211695,
      "grad_norm": 0.10385219130875901,
      "learning_rate": 2.5437577389486597e-10,
      "loss": 0.1865,
      "step": 32310
    },
    {
      "epoch": 0.9985401910882141,
      "grad_norm": 0.10349690468541285,
      "learning_rate": 1.7295062878341928e-10,
      "loss": 0.1812,
      "step": 32320
    },
    {
      "epoch": 0.9988491453552588,
      "grad_norm": 0.10267401157150524,
      "learning_rate": 1.0718405786425134e-10,
      "loss": 0.1787,
      "step": 32330
    },
    {
      "epoch": 0.9991580996223034,
      "grad_norm": 0.10773225191306192,
      "learning_rate": 5.707612979244381e-11,
      "loss": 0.1795,
      "step": 32340
    },
    {
      "epoch": 0.999467053889348,
      "grad_norm": 0.09825659917396752,
      "learning_rate": 2.2626896877819826e-11,
      "loss": 0.1849,
      "step": 32350
    },
    {
      "epoch": 0.9997760081563927,
      "grad_norm": 0.09261055154357052,
      "learning_rate": 3.836395081613375e-12,
      "loss": 0.1789,
      "step": 32360
    },
    {
      "epoch": 0.9999922761433239,
      "step": 32367,
      "total_flos": 8484206288044032.0,
      "train_loss": 0.19035680645326206,
      "train_runtime": 188551.2044,
      "train_samples_per_second": 5.493,
      "train_steps_per_second": 0.172
    }
  ],
  "logging_steps": 10,
  "max_steps": 32367,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8484206288044032.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}